MODERN METHODS OF

GEOCHEMICAL ANALYSIS
Monographs in Geoscience
General Editor: Rhodes W. Fairhridge
Department of Geology, Columbia University, New York City

B. B. Zvyagin
Electron-Diffraction Analysis of Clay Mineral Structures-1967
E. I. Parkhomenko
Electrical Properties of Rocks -1967
L. M. Lebedev
Metacolloids in Endogenic Deposits-1967
A. I. Perel'man
The Geochemistry of Epigenesis-1967
s. J. Lefond
Handbook of World Salt Resources-1969
A. D. Danilov
Chemistry of the Ionosphere-1970
G. S. Gorshkov
Volcanism and the Upper Mantle: Investigations in the
Kurile Island Arc-1970
E. L. Krinitzsky
Radiography in the Earth Sciences and Soil Mechanics-1970
B. Persons
Laterite-Genesis, Location, Use-1970
D. Carroll
Rock Weathering-1970
E. I. Parkhomenko
Electrification Phenomena in Rocks -1971
R. E. Wainerdi and E. A. Uken
Modern Methods of Geochemical Analysis-1971

In preparation:
A. S. Povarennykh
Crystal Chemical Classification of Minerals
MODERN METHODS OF
GEOCHEMICAL ANALYSIS
Edited by
Richard E. Wainerdi
Associate Dean of Engineering, and
Head, Activation Analysis Research Laboratory
College of Engineering
Texas A&M University
College Station, Texas

and
Ernst A. Uken
Scientific Advisory Council
Pretoria, South Africa

With a Foreword by
Sir Edward Bullard, F. R. S.

g:> PLENUM PRESS NEW YORK-LONDON 1971

Library of Congress Catalog Card Number 75-157148

ISBN 978-1-4684-1832-3 ISBN 978-1-4684-1830-9 (eBook)

DOI 10.1007/978-1-4684-1830-9

@ 1971 Plenum Press, New York

Softcover reprint of the hardcover 1st edition 1971
A Division of Plenum Publishing Corporation
227 West 17th Street, New York, N.Y. 10011

United Kingdom edition published by Plenum Press, London

A Division of Plenum Publishing Company, Ltd.
Davis House (4th Floor) , 8 Scrubs Lane, Harlesden, NW1 0 6SE, England

All rights reserved

No part of this publication may be reproduced in any form

without written permission from the publisher
FOREWORD

The founders of geology at the beginning of the last century were

suspicious oflaboratories. Hutton's well-known dictum illustrates the point:
"There are also superficial reasoning men ... they judge of the great oper-
ations of the mineral kingdom from having kindled a fire, and looked
into the bottom of a little crucible." The idea was not unreasonable; the
earth is so large and its changes are so slow and so complicated that labo-
ratory tests and experiments were of little help. The earth had to be studied
in its own terms and geology grew up as a separate science and not as a
branch of physics or chemistry. Its practitioners were, for the most part,
experts in structure, stratigraphy, or paleontology, not in silicate chemistry
or mechanics.
The chemists broke into this closed circle before the physicists did.
The problems of the classification of rocks, particularly igneous rocks, and
of the nature and genesis of ores are obviously chemical and, by the mid-
19th century, chemistry was in a state where rocks could be effectively
analyzed, and a classification built up depending partly on chemistry and
partly on the optical study of thin specimens. Gradually the chemical
study of rocks became one of the central themes of earth science. It was,
however, a very tedious task; and not many able people were prepared to
devote their main efforts to the labyrinthine, and insufferably boring, prog-
ress of the wet chemical analysis of a set of rocks where each specimen
needs a week of careful work to determine its main constituents and the
determination of the many trace elements is hardly to be contemplated.
Efforts to get the work done on a large scale, as a business proposition were
also not entirely satisfactory owing to the risk of error in a routine operation.
There was clearly an enormous incentive to develop quick and reliable
methods. Because the earth is so large, but yet has processes on all scales
from the atomic to the global, the need is not for tens or hundreds of
analyses but for hundreds of thousands. There are about 90 elements in
the rocks, most of them in abundances measured in parts per million or

v
vi Foreword

parts per billion. How should we estimate them? How far can we believe
the results? These are the themes of this book.
There is a further question: why should we estimate them? What are
we trying to do? There are two kinds of answers; some things are important
because they are interesting and some are interesting because they are
practically important. The possibility of determining not only the main
constituents of a rock but also a large number of trace elements and their
distribution among the constituent minerals should characterize the rock
in such detail as to make it effectively unique. If this can be done it should
be possible to be much more certain than we are at present about the
displacement of the great transcurrent and transform faults of the con-
tinents and the oceans by comparing the displaced rocks on the two sides.
In this way we should be able to get additional confidence in the fits of
the continents and to work out the history of their movements. We should
also be able to assign the continental remnants stranded in mid ocean to
their former positions. Were the Seychelles once alongside Madagascar or
were they attached to the mainland of Africa or perhaps to Australia or
Antarctica? The success of purely geometrical methods of fitting the con-
tinents around the Atlantic should not blind us to the ambiguities in the
Indian Ocean and still more to the almost untouched subject of the arrange-
ment of the continents in pre-Mesozoic times. The record is so blurred by
time and later movements that we need all the help we can get from geo-
chemistry.
On the practical side, nothing is more certain than that we shall have
to mine and use poorer grades of ore and that it is undesirable to waste
our efforts in grinding and chemically processing lumps of barren rock.
There must be a great future for what a computer engineer would call
"real time" chemical analysis; such as the determination of one or a few
valuable elements in a lump of rock as it passes on a conveyor belt, for
example. Behind this is the larger question of ore genesis; we all know
that "ore is where you find it" but this is not a very fruitful rule for pros-
pecting. Can we, "by looking into a little crucible" learn enough about
rocks to understand why ore is where it is, and then direct our search to
places where the omnipresent trace elements have come together into work-
able ore? Why should a field in western England, and no field anywhere
else, have several feet of pure strontium sulfate just under the surface?
How can such a concentration of so rare an element occur and why just
in one field? Such questions are practically important and also intellectual
problems of the first class; it just happens that they are a little unfashion-
able at the moment.
Foreword vii

By the time this book is published we shall have the first cores from the
deep drilling in the oceans by project JOIDES and the first specimens from
the moon. No one can tell what enlightenment and what problems these
will bring. They cannot fail to tell us much about the history of the earth
and of the solar system. The proper chemical analysis of these materials,
gathered with so much effort, will be the culmination of two of the great
adventures of our age. This book is, indeed, timely.

EDWARD BULLARD, F.R.S.

PREFACE

The field of geochemical analysis is at once an ancient art, and a modern

science. It is becoming of importance in every country and, lately, in the
areas between countries, and in the heavens above them. The economic
pressure behind mineral exploration is only part of the reason geochemical
analyses are important. The basic urge to know and understand the proc-
esses of the geologic past requires that analytical chemistry, of a most
sophisticated sort, be applied to presently available materials in order to
attempt to provide basic data for speculation about what went on in past
eras.
Analytical geochemistry has been applied to moon samples, to undersea
and ocean samples, and to many terrestrial samples, but very few workers
have applied more than one of the methods described in this book to the
same sample. There are many methods which are mutually complementary,
and an important objective of this effort is to acquaint geochemical analysts
with recent developments in competitive and complementary methods so
that they can utilize them, where needed, to complete the analytical descrip-
tion of a given sample.
The Editors wish to thank all of the contributors to this book for their
efforts, and particularly to thank Sir Edward Bullard, F.R.S., for his signi-
ficant and thoughtful Foreword.
RICHARD E. WAINERDI
ERNST A. UKEN

viii
CONTENTS

Foreword, by Sir Edward Bullard v

Chapter 1
Introduction
by R. E. Robinson, W. R. Liebenberg, and S. A. Hiemstra

1. Classical Definition of Geochemistry

2. The Role of Geochemical Analysis 4
2.1. Geology. . . 4
2.2. Geochemistry 4
2.3. Mineralogy 4
2.4. Petrology . 5
2.5. Exploration 5
2.6. Mining . . 6
2.7. Ore Dressing 7
2.8. Chemical Processing 8
3. Summary . . . . . . . 9

Chapter 2
Statistics
by A. B. Calder

1. Introduction . . . . . . . . . . . . . . . . . . 11
1.1. The Nature and Scope of Statistics . . . . . 11
1.2. The Importance of a Statistical Approach in Analytical
Chemistry. 12
2. Basic Concepts . 14
2.1. Variation . 14

ix
x Contents

2.2. Populations and Samples: Parameters and Statistics 14

2.3. Types of Population and Variable 14
2.4. Probability and Random Selection 15
3. Measures of Location and Dispersion 15
3.1. Average: Arithmetic Mean . . . 16
3.2. Variability: Standard Deviation and Variance . 17
3.3. Degrees of Freedom . . . . . . . . 17
3.4. Calculation of the Standard Deviation 18
3.5. Coefficient of Variation. 18
4. Frequency Distributions. . . 19
4.1. Normal Distribution . . 19
4.2. Equation and Properties of the Normal Distribution 19
4.3. Justification for the Assumption of Normality in Practice 20
4.4. Binomial and Poisson Distributions 21
5. Standard Error. . . . . . 22
5.1. Sampling Distributions 22
5.2. Standard Error 22
5.3. Theoretical Derivations 23
5.4. The t Distribution 24
6. Tests of Significance . . 25
6.1. Introductory Note . 25
6.2. Normal Deviate Test 25
6.3. t Test. . . . . . . 25
6.4. Variance Ratio (F) Test 26
7. Regression. . . . . . . . . 26
7.1. Introduction . . . . . . 26
7.2. Method of Least Squares 27
7.3. Variance about Regression 28
7.4. Error of Estimation of X from Regression Line 28
8. Analysis of Variance . . 29
9. Sampling Methods . . . 31
9.1. General Comments. 31
9.2. Remarks on Analytical Procedure 32
9.3. Computation of Error Distribution 33
10. Statistical Tables . . . . . . 34
10.1. Values of t. . . . . . 34
10.2. Values of F (PI = 0.05) 35
Contents xi

11. Applications to Analytical Methods 35

11.1. Introduction . . . . . . . . 35
11.2. Reduction in Operational Time . 36
11.3. Sampling Errors 40
11.4. Calibration . . . . . . . . . . 43
11.5. Blank Variation . . . . . . . . 45
11.6. Multicomponent Methods of Analysis . 48
11. 7. Particle Counting 50
References 51

Chapter 3

Chemical Analysis and Sample Preparation

by V. C. O. SchUler

1. Introduction . . 53
2. Sample Preparation. . . 53
2.1. Sampling Procedures 54
2.2. Sample Dividing 55
2.3. Homogenizing . . . 56
2.4. Sample Storage 56
3. Dissolution of Geologic Samples. 56
3.1. Acid Digestion. . 57
3.2. Sample Fusions . . . 57
4. Methods of Separation . . 61
4.1. Precipitation Methods 61
4.2. Electrolytic Separations 63
4.3. Solvent Extraction . . 64
5. Methods of Determination 65
5.1. Gravimetric Analysis 65
5.2. Fire Assay 66
5.3. Volumetric Analysis 67
6. Advantages, Disadvantages, and Limitations of Wet Chemical
Analyses. . . . . . . . . 69
6.1. General Considerations 69
6.2. Standardization 69
References 70
xii Contents

Chapter 4
Ion-Exchange Chromatography
by H. F. Walton

1. Introduction . . . . . . 73
2. Ion-Exchanging Materials 74
2.1. Ion-Exchange Resins 74
2.2. Inorganic Exchangers 75
2.3. Ion Exchangers Based on Cellulose 76
3. Ion-Exchange Selectivity 76
3.1. Inherent Selectivity . . . . . . . 76
3.2. Complex Ion Formation . . . . 78
4. Anion-Exchange Separations of Metals 79
5. Mixed Solvents in Ion-Exchange Separations 83
6. Special Ion-Exchange Techniques 84
6.1. Ligand Exchange. . . . . 84
6.2. Isotopic Ion Exchange . . 85
6.3. Precipitation Ion Exchange 85
6.4. Difference Chromatography 86
6.5. Ion-Exchange Papers . . . 86
7. Applications to Geochemical Analysis 87
7.1. Concentration of Traces of Metals. 88
7.2. Removal of Interfering Ions. . . . 89
7.3. Systematic Analysis of Silicate Rocks. 90
7.4. Individual Elements in Silicate Rocks 92
References 94

Chapter 5
Colorimetry
by Gordon A. Parker, and D. F. Boltz

1. Introduction . . . 97
2. Theoretical Considerations 98
2.1. Measurement of Light Absorption 98
2.2. Colorimetric Techniques 101
3. Instrumentation 105
3.1. Filter Photometers . . . 105
Contents xiii

3.2. Spectrophotometers . 107

3.3. Instrument Selection 110
4. Sample Preparation. . . 112
4.1. Range of Optimum Concentration 112
4.2. Preparation of the Colored System 113
5. Photometric Measurements . 115
5.1. Accuracy and Precision . 115
5.2. Advantages 118
5.3. Limitations 119
6. Applications . . 121
6.1. Determination of Metals 121
6.2. Determination of Nonmetals 123
References 123

Chapter 6
Infrared Spectrometry
by W. M. Tuddenham and J. D. Stephens

1. Introduction . 127
1.1. Historical 127
1.2. Theory . 128
1.3. Instrumentation 128
2. Techniques of Sample Presentation. 129
2.1. Absorption . . . . . . . 129
2.2. Attenuated Total Reflection (ATR) 131
2.3. Reflection . . . . . . . . 131
2.4. Emission Spectroscopy . . 132
3. Quantitative Applications in Geochemistry 132
4. Mineral Identification with Infrared . . . 136
4.1. Minerals Containing H 20, OH-, and Hydrogen Bond. 137
4.2. Carbonate Minerals 138
4.3. Nitrate Minerals . . . . . . 139
4.4. Borate Minerals . . . . . . 140
4.5. The Sulfate Group Minerals 142
4.6. The Silicate Minerals . . . . 143
4.7. Phosphates, Vanadates, and Arsenates 151
4.8. Molybdates and Tungstates 155
4.9. Oxide Minerals . . . . . 157
xiv Contents

5. Crystal-Chemistry Studies 159

5.1. Solid-Solution Series 159
5.2. Polymorphous Series 162
5.3. Leaching Studies . 163
5.4. Firing Studies . . 163
5.5. Alteration Studies 165
References 166

Chapter 7

Optical Emission Spectroscopy

by Armin P. Langheinrich. and D. Blair Roberts

1. Introduction . 169
1.1. History . 169
1.2. Principles 170
1.3. Relation to Other Techniques 172
2. Equipment and Facilities . . . 173
2.1. Components and Functions 173
2.2. Laboratory Facilities . . . 176
2.3. Field Facilities . . . . . . 176
2.4. Special-Purpose Equipment 177
3. Techniques . . . . . . . . . 178
3.1. Qualitative Analyses . . . 180
3.2. Semiquantitative Analyses . 180
3.3. Quantitative Analyses. . . 191
4. Geochemical Applications . . . 193
4.1. Element Abundance and Distribution 194
4.2. Environmental Studies 194
4.3. Mineral Exploration 194
5. Literature . . . . . . . 196
6. Evaluation of the Method 197
7. Future Developments 198
References 200
Contents xv

Chapter 8
Atomic Absorption
by L. R. P. Butler and M. L. Kokot

1. Introduction . . . . . . 205
2. Principles of Atomic Absorption . 206
3. Apparatus . . . . . . 208
3.1. Source . . . . . 208
3.2. Absorption Flame 209
3.3. Spectrometer. . . 211
3.4. Detector. . . . . 212
3.5. Amplifier and Measuring System 212
4. Sensitivity, Accuracy, and Interferences. 213
4.1. Sensitivity . 213
4.2. Accuracy . . . 214
4.3. Interferences . . 214
5. Analysis of Geologic Samples by Atomic Absorption 217
5.1. Major Elements . . . . . . . . . . . . . . . 217
5.2. Minor and Trace Elements . . . . . . . . . . 229
5.3. Microtrace Elements after Chemical Enrichment 234
5.4. Water Analysis 235
6. Geochemical Prospecting . . . . 237
7. Recent and Future Developments 240
References 241

Chapter 9
X-Ray Techniques
by H. A. Liebhafsky and H. G. Pfeiffer

1. Introduction . . . . . . . 245
2. Theoretical Considerations . . . . 246
2.1. Nature and Excitation of X Rays 246
2.2. Interaction of X Rays with Matter 249
2.3. X Rays and Atomic Structure. 251
3. Instrumentation . . . . . . . . . 254
3.1. Analytical Systems . . . . . . 254
3.2. Measurement of X-Ray Intensities 255
xvi Contents

4. X-Ray Diffraction 258

4.1. Description . 258
4.2. Applications . 260
5. X-Ray AbsorptiOlpetry 261
6. X-Ray-Emission Spectrography 262
6.1. Discussion. . . . . 262
6.2. Applications . . . . . . . 264
6.3. Electron Microprobe . . . 266
6.4. Portable X-Ray-Emission Spectrograph 268
References 269

Chapter 10
Radiometric Technique
by L. Rybach

1. Introduction . . . . . . . . 271
2. Fundamentals . . . . . . . 272
2.1. Nuclear Data of Natural Radioisotopes 272
2.2. Background Radiation . . . . . . . 277
2.3. Interaction of Radiation with Matter 278
3. Radiation Detectors . . . . . . . 281
3.1. Beta and Gamma Measurement . . . 281
3.2. Alpha Measurement . . . . . . . . 285
3.3. Scaler and Ratemeter Circuits for Integral Counting, Pulse-
Height Analysis . . . . . . . . . . . . . . . . . 288
3.4. Autoradiographic Methods . . . . . . . . . . . . 291
3.5. Background Reduction for Laboratory Measurements 293
4. Radiometric Prospecting 294
4.1. Airborne Radiometric Survey 294
4.2. Carborne Prospection. . . . 296
4.3. Prospecting on Foot . . . . 297
5. Quantitative Determination of Natural Radioelements 298
5.1. Sample Preparation.. .......... . 298
5.2. Integral Counting for U and K Determinations . 298
5.3. Gamma-Ray Spectrometry for Simultaneous U, Th, and K
Determinations. . . . . . . . . . . . . . . . . . . . 301
Contents xvii

5.4. Beta-Gamma Counting for Equilibrium Studies 310

5.5. Alpha Counting, Alpha Spectrometry 311
5.6. Fission Track Counting . . . 313
6. Radiometric Age Determinations. 315
References 315

Chapter 11
Nuclear Activation Analysis
by L. E. Fite, E. A. Schweikert, R. E. Wainerdi, and E. A. Uken
1. Introduction . . . . . . . . . . 319
1.1. Irradiation. . . . . . . . . 320
1.2. Measurement of Induced Radioactivity 322
1.3. Gamma-Ray Interactions . . . . . . . 322
1.4. Data Interpretation. . . . . . . . . . 325
1.5. Automated Nuclear Activation-Analysis Systems 327
2. Reactor-Thermal-Neutron Activation Analysis 329
3. 14-MeV Neutron Activation Analysis 334
3.1. Cyclic Activation Analysis 337
4. Charged-Particle and Photon Activation Analyses 337
4.1. Introduction . . . . . . . . . . . 337
4.2. Photon Activation Analysis . . . . 337
4.3. Charged-Particle Activation Analysis 340
4.4. Analytical Applications . . . . . 342
5. Activation Analysis in the Geosciences 343
5.1. Mineral Prospecting 343
5.2. Logging Techniques 344
5.3. Ore Sorting 345
References 346

Chapter 12
Mass Spectrometry
by J. N. Weber and P. Deines
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . 351
2. Stable Carbon- and Oxygen-Isotope Mass Spectrometry. . . . 352
2.1. Principles Underlying the Construction of an Isotope Ratio
Recording Mass Spectrometer . . . . . . . . . . . . . 352
xviii Contents

2.2. Reporting of Carbon- and Oxygen-Isotope Data 356

2.3. Sample-Preparation Procedures . . . . . . . 359
2.4. Geochemical Applications . . . . . . . . . . 367
2.5. Future Developments of Stable-Isotope Mass Spectrometry 370
3. Trace-Element-Concentration Determination by Spark-Source
Mass Spectrometry . . . . . . . . . . . . . . . . 370
3.l. The Radiofrequency Spark-Source Mass Spectrograph 371
3.2. Analytical Procedure . . . . . . . . . . 380
3.3. Sample-Handling Techniques . . . . . . 384
3.4. Detection Limit, Precision, and Accuracy . 386
3.5. Geochemical Applications . 389
References 390
Chapter 1

INTRODUCTION

R. E. Robinson, W. R. liebenberg,
and S. A. Hiemstra
National Institute for Metallurgy
Johannesburg, South Africa

1. CLASSICAL DEFINITION OF GEOCHEMISTRY

Geochemistry, the study of the chemistry of the earth, is one of the original
boundary sciences, relying heavily on the techniques and principles of
chemistry and physics to study the distribution of elements in the earth
and to explain the processes that give rise to these distribution patterns.
Many of those same early scientists that contributed to the knowledge and
techniques of chemistry also contrjbuted notably to the knowledge of geo-
chemistry, simply because their subjects for study were natural products of
the earth, in many instances minerals and rocks. However, the ideas and
objectives of inorganic chemistry crystallized more quickly than those of
geochemistry. It was easy to gather data on the composition of natural
materials-but how could these be organized and classified into a formal,
logical science? What is meant by the term "Geochemistry"?
Among the early attempts to provide a formal definition for geo-
chemistry, the same elements can be noted in the attempts made by W. A.
Wernadski, A. E. Fersman, and V. M. Goldschmidt. The classical definition
given by Goldschmidt in his papers published in the interval from 1923 to
1937 can still be regarded as the most precise. Rankama and Sahama
translate this definition as follows:

1
2 R. E. Robinson. W. R. Liebenberg. and S. A. Hiemstra

(a) To establish the abundance relationships of elements and nuclides

in the earth.
(b) To account for the distribution of elements in the geochemical
spheres of the earth, e.g., in the minerals and rocks of the litho-
sphere and in natural products of various kinds.
(c) To detect the laws governing the abundance relationships and the
distribution of the elements.

The first stage of the science of geochemistry is analytical-the gather-

ing of data on the distribution of elements. This book on modern methods
of geochemical analysis thus deals with the techniques that can be used to
realize the first object of geochemistry-that of accumulating data on the
composition of natural materials.
This process started in an informal way when Bronze-Age man first
realized that certain ores contained the elements necessary to his trade.
Early in the 19th century, John Dalton, J. L. Gay-Lussac, and Amedeo
Avogadro made the necessary observations that led to the ideas that chem-
ical compounds had fixed compositions and that molecules and atoms exist.
The necessary foundations were laid for quantitative analyses to be carried
out, using the classical but laborious techniques of wet chemical analysis.
During the 19th century, a wealth of data on the composition of geo-
logic materials, viz., minerals and rocks, were accumulated. The Swedish
chemist Jons Jacob Berzelius made notable contributions both to the
science of geochemistry and to the science of chemistry, and is known for
his many analyses of minerals and his compilation of an accurate table of
atomic weights.
During the 19th century and the first half of the 20th century -this
process of accumulation of data continued, and, in the period 1908-1924,
Frank Wigglesworth Clarke's invaluable compilation The Data of Geo-
chemistry saw print.
The technique of wet chemical analysis is still, perhaps, the most
valuable technique in the analysis of natural materials, since many of the
modern physical methods are dependent on standards analyzed in the
classical way. It is a laborious method, however, and for many elements
the detection limits are poor. A need for more rapid and more sensitive
methods was felt long ago.
In the meantime, the science of optical spectrography based on the
classical experiments of Isaac Newton in 1664 and Joseph von Fraunhofer
in 1814 was slowly evolving. In 1860 and 1861 Robert Bunsen and Gustav
Introduction 3

Kirchhoff discovered caesium and rubidium by means of optical spectro-

graphy. In 1922 Assar Hadding demonstrated that optical spectrography is
a powerful geochemical tool. From about the third decade of the 20th
century it was used increasingly for gathering the facts of geochemistry,
particularly the distribution of minor and trace elements in rocks and
minerals. Today, optical spectrography is still used extensively for the
analysis of large numbers of samples of rocks, minerals, and soil, and the
instrumentation varies from the simple and dependable to the huge, so-
phisticated, direct-reading equipment capable of performing hundreds of
analyses a day. In addition to techniques based on optical emission spectro-
graphy, infrared spectrophotometry and atomic absorption are being in-
creasingly applied to geochemical problems. Although emission spectro-
graphy evolved into a rapid and sensitive tool with low detection limits for
many elements, high precision could usually be attained only with difficulty,
and the technique had limitations in the analysis of major components.
At the end of the 19th century, two epoch-making scientific discoveries
were made within a period of a few months: the detection of X rays by
Wilhelm Conrad Rontgen on November 8, 1895, and the discovery of
natural radioactivity by A. H. Becquere1 in February 1896. These discoveries
form the foundation stones of a number of very powerful analytical tech-
niques that became significant in the postwar years of the 20th century, and
from about 1960 they have climbed to a position that threatens to over-
shadow the classical techniques for the analysis of geologic materials.
During 1903 the Russian botanist M. S. Tswett discovered a technique
-which later became known as chromatography-for the separation of
leaf pigments. In 1931 R. Kuhn and his coworkers demonstrated the efficacy
of the method in separating vitamin-A fractions, and since that time the
technique of chromatography has grown rapidly to the eminent position it
occupies today.
Nowadays the number of instruments and techniques that can be used
to analyze geologic materials is increasing rapidly, and they are continuously
being refined.
The attainment of speed has always been antithetical to the achievement
of high accuracy, and it still is. However, today we are nearer the ideal in
both speed and accuracy in the analysis of geologic material than ever
before.
While the pages of this book explore in detail the different chemical
and physical methods for analyzing materials from the earth, it is also
advisable to turn our attention briefly to the use that is being made of these
analyses.
4 R. E. Robinson. W. R. Liebenberg. and S. A. Hiemstra

2. THE ROLE OF GEOCHEMICAL ANALYSIS

2.1. Geology
In geology very extensive use is made of the techniques and results of
geochemical analysis. Broadly speaking, one can distinguish four fields of
application: geochemistry, mineralogy, petrology, and exploration.

2.2. Geochemistry
Geochemical analysis provides the fundamental data from which the
distribution of every element and nuclide in the main rock types of the earth,
the lithosphere, the hydrosphere, the biosphere, and even the atmosphere
and outer space, is calculated or estimated. Because of the availability of
an increasing number of accurate analyses, old figures for these quantities
can be refined and made more and more accurate.
Geochemical analyses are also essential in the study of the various
processes in and on the earth that cause elements and nuclides to migrate,
disperse, fractionate, or concentrate. Sometimes these processes are well
defined and easy to detect with only a few analyses; other times the trends
are weak and the concentration values are open to much variation, so that
numerous chemical analyses and involved statistical calculations are nec-
essary to detect them.

2.3. Mineralogy
In mineralogy the chemical composition of the minerals studied is of
the utmost importance. No description of a mineral is complete without
its chemical composition.
Isomorphous series are present in many of the most important rock-
forming mineral families. The correlation of the physical properties and the
chemical compositions of these minerals form an important branch of
mineralogy. The laws that control the admittance, capture, or substitution
of minor or trace elements in the structure of these minerals are very impor-
tant because of the influence on differentiation trends and the enrichment
of certain rare elements in the residual products of differentiation.
In the past the study of rare minerals that were available in very small
amounts presented serious difficulties. Whereas the techniques for studying
the physical properties of minute amounts were available, it was often
extremely difficult or impossible to obtain high-quality chemical analyses
of these materials. The advent of certain physico analytical techniques,
Introduction 5

notably that of electron-microprobe analysis, can be regarded as a major

advance.

2.4. Petrology
In igneous petrology extensive use is being made of chemical analyses
in the comparison of rocks from different areas and in the study of dif-
ferentiation trends and petrographic provinces.
In metamorphic petrology the chemical compositions afford important
clues to the original nature of the rocks and the processes responsible for
their transformation, and also to whether metasomatic processes we~e
operative or not.
In sedimentary petrology the chemical composition of sediments and
their constitutive minerals may provide information on the provenances,
the chemical and physical conditions that prevailed during their formation,
and the weathering potential of these rocks.
One fundamental problem in petrology is that of obtaining samples
that are representative of the parent-rock body. One or a few samples
cannot reveal the amount of variation of rock bodies that may cover tens
or hundreds of square miles. The availability of methods whereby accurate
analyses of many samples can be made in a relatively short time promises
to revolutionize the study of petrology.

2.5. Exploration
In the technique known as geochemical prospecting, analyses of samples
of eluvial soil or plant material are made in the hope that areas in which
there are anomalous concentrations of certain elements will be found. If
such areas are found, they are studied in more detail, and this technique
sometimes leads to the discovery of ore deposits of a grade and size suitable
for exploitation. The technique is critically dependent on analytical methods
that are rapid enough to permit thousands of analyses to be carried out in a
short time, and sensitive enough to detect low concentrations of the elements
being sought.
More indirect techniques include the study of an element known to be
associated with the metals sought, for which a better analytical technique
is available. One example of this is that a mercury halo is associated with
certain ore deposits, and there exist extremely sensitive analytical techniques
for the determination of mercury.
Sometimes a geochemical study of certain geologic formations and rock
bodies may afford clues to the presence or absence of ore deposits in them.
6 R. E. Robinson. W. R. Liebenberg. and S. A. Hiemstra

2.6. Mining
Strictly speaking, the term "geochemical analysis" should not be related
to activities other than geochemistry and geochemical prospecting. But it is
such rigid adherence to formal and somewhat arbitrary definitions that has
been responsible for so much compartmentalization in science, and it is
hardly an encouragement for cross fertilization between different disciplines.
It will become clear that almost all of the techniques described in this book
will have application in many directions other than those conforming to the
strict definition of geochemistry.
Probably the first real instance of a geochemical method of analysis
being of direct use to the mining engineer was the application of radio-
metric methods of analysis in the deep-level mines both on the Witwaters-
rand and in the Orange Free State. The rather crude Geiger counters of the
period immediately following World War II proved to be invaluable to the
mining engineer by providing a farily good idea of the exact locality of the
gold- and uranium-bearing conglomerates and also of their content of
uranium. Many of the skills of low-cost, deep-level mining revolve round
the ability of the mines to operate on those mine sections that are "payable,"
and to leave untouched those sections that are "unpayable." Any delays
between the sampling of a newly exposed face and the receipt of the analyt-
ical results must inevitably bring about delays in mining and a reduced
output for the mine. The use of a Geiger counter at rock faces provided a
rough but nevertheless valuable assessment of the uranium content, which
could be related to both gold and uranium payability.
More recently the demand for rapid on-the-spot analysis in mining
has become even more apparent. The development of an efficient and
economical rock-cutting unit by the mining research laboratory of the
Chamber of Mines of South Africa threatens to revolutionize gold-mining
techniques in South Africa. This new technique has made it possible to cut
out a slice of rock containing the valuable constituents, rather than to blast
it out with explosives; and it promises to greatly diminish the amount of
waste rock associated with the valuable constituents, and thus considerably
reduce the mining costs and increase the concentration of the valuable con-
stituents in the feed to the metallurgical treatment plant. The maximum
benefit to be derived from rock-cutting depends on the ability to decide
beforehand where the valuable constituents are situated. In the gold mines
in South Africa this does not at the moment represent too serious a problem
since the gold and uranium are contained in fairly well-defined reefs that
can be observed quite easily. Nevertheless, it has been proved that the gold
Introduction 7

and uranium are not distributed evenly throughout the reefs, and the possi-
bility of being able to cut out only those portions of the reefs with payable
quantities of gold is very attractive.
At the moment no obvious solution to this problem is in sight. What
is required is a method of analysis that will give at least a semiquantitative
estimate of the gold content, in the region of a few parts per million, in a
relatively short time, and will be capable of being used underground at the
rock face. It appears that the techniques of neutron activation or x-ray
fluorescence cannot at present achieve the sensitivity required for them to
be of much value; but further developments in the direction of stronger
activation sources could well bring this very demanding problem into the
region of being solvable.
Another area in mining where geochemical methods of analysis could
play an important role in the future is in that of opencast mining, par-
ticularly for base metals such as copper and zinc. Many of the recently
discovered, large ore bodies of base metals not only are of low grade but
also are very complex in structure.
A typical case is the complex carbonatite ore body at Phalaborwa, in
the Transvaal, which is being mined and processed by the Palabora Mining
Company. This body consists of many zones of material-high-titanium
magnetite, low-titanium magnetite, phoscorite-rich areas, carbonatite-rich
areas-all of which contain copper in varying amounts. Certain areas of
the deposit can be considered to contain payable quantities of copper,
others not; and in the planning of the mining strategy it has been necessary
to obtain a fairly accurate three-dimensional model of the whole body.
The samples were obtained by drilling on a preestablished grid pattern,
but the examination of all the samples by chemical analysis proved to be
tedious and expensive. Even with the accurate three-dimensional model,
the mining engineer would find it invaluable to have a simple instrumental
method of analysis that could be used for checking the grade of each section
of the pit before blasting, so that its destiny-whether the treatment plant
or waste heap-can be established as early as possible. This concept is not
far-fetched science fiction. The latest instruments using radioactive materials
as sources of both y rays and neutrons (e.g., californium 252) are coming
close to achieving just this.

2.7. Ore Dressing

Where mining stops and ore dressing commences is perhaps a deba-
table point, but, if one accepts that the ore-dressing stage of the operation
8 R. E. Robinson, W. R. Liebenberg, and S. A. Hiemstra

begins with the mechanical crushing of the ore to be treated, it is fairly

obvious that there is scope for, and benefits to be derived from, rapid
methods of analysis. Crushing, grinding, and the treatment of waste material
represent inefficiency and wastage of money. The greater the degree of
selectivity between waste and valuable constituents that can be exercised
at each ore-dressing stage, the more efficient will be the operation.
The method that has been used the most successfully is that of radio-
metric analysis, which is now accepted as a standard procedure.
This method of analysis has been engineered to a point where either
completely full trucks of ore can be "analyzed" or individual rocks on a
moving belt can be separated on the basis of their content of radioactive
materials.
Obviously this method is limited to the ores that contain naturally
radioactive elements (which means essentially uranium and thorium).
However, the problems in imparting a short-lived radioactivity to other
ores, so as to enable separations to be carried out, are not really insur-
mountable.
Over the years the ore dresser has made use of just about every physical
property of mineral particles to concentrate valuable constituents from waste
materials. The use of properties such as specific gravity, size, and shape are
well known and have been used since ancient times. Some of the more recent
developments are of greater interest. Instruments have now been developed
that measure the light reflected from individual mineral particles and, de-
pending on the color, provide a signal that will allow one colored mineral
to be separated from another. These instruments work at very high speed.
If one considers the basic features, one realizes that this "color sorter" is
only another form of activation analysis-that is to say, the measurement
of photons of a certain wavelength produced by the excitation of the mineral
by photons of a range of wavelengths.
When it is appreciated that most of mineral dressing deals with major
concentrations of constituents and not with trace elements, it does not seem
unrealistic to imagine that all the other "activation" techniques of analysis
can be seriously considered as being sufficiently rapid to make mineral
separation feasible.

2.8. Chemical Processing

The role of chemical and instrumental analysis in chemical processing
is a major one. In the modern metallurgical processing plant, particular
attention is being paid to accurate metallurgical balances, to the control
Introduction 9

and automation of all stages of the process, and to the specification analysis
of the final product.
The specifications, imposed in the first instance by the nuclear-power
and atomic energy programs for metals of extremely high purity, have
caused repercussions in many other metallurgical practices.
The optical spectrograph is now becoming a standard instrument of
metallurgical plant control, and the degree of expense and sophistication
introduced into the latest automatic spectrographs is in itself an indication
of the importance of this type of analysis to the metallurgist.
Very recently the mass spectrograph was applied to the control and
final-product specification analysis of the metallurgical plant. The produc-
tion of titanium dioxide pigment from ilmenite is a case where a rigid control
over impurities such as iron, vanadium, and chromium is necessary at all
stages of the operation, and where the mass spectrometer is apparently
doing a very successful job.

3. SUMMARY

One can expect that in future years analytical methods will show many
profound improvements and will have greatly increased applications.
Geochemical requirements are among the most challenging. Speed and
sensitivity are probably the primary requisites, and for this reason the
methods developed will probably find equally valuable applications in other
areas, such as the mining, dressing, and chemical processing of minerals.
Thus, this book, which presents those methods of analysis of interest
to geochemists, must not be read and studied with a restrictive attitude.
Most of the methods described are standard techniques employed in a host
of analytical laboratories throughout the world. The wider the audience
that this book attracts, and the greater the degree of cross fertilization be-
tween the many analysts, the more all will benefit.
Chapter 2

STATISTICS

A. B. Calder
Newcastle upon Tyne Polytechnic
Newcastle, England

1. INTRODUCTION

1.1. The Nature and Scope of Statistics

Statistical methods are sometimes described as merely being techniques for

treating numerical data. It is necessary, however, to restrict both the nature
of the data and the reasons for studying them, before such methods can
rightfully be called statistical. Statistics is concerned essentially with data
that have been obtained by taking a set of observations, in the form of
measurements or counts, from a source of such observations. The ultimate
purpose of statistics is to infer, from the set of observations made, some-
thing about the source of the observations from which the set was taken
and, in particular, to make estimates of certain numerical characteristics
or parameters of the source of the observations.
Consider a granitic outcrop approximately 50 ft in diameter, of fairly
uniform appearance, structure, and composition. Now consider the case
of a chemist who is confronted with the task of assessing the cobalt content
of this particular outcrop. Clearly the whole outcrop cannot be removed,
crushed, and analyzed. Instead the chemist takes a number of samples, say
20 from different spots in the area. Each sample is crushed, thoroughly
mixed, and a suitable portion of, say, 2.0 g may be weighed out for spectro-
chemical analysis. The results obtained may be as follows:

11
12 A. B. Calder

Co (in ppm):

0.25 0.20 0.25 0.22 0.24 0.23 0.23 0.28 0.26 0.23
0.24 0.24 0.25 0.22 0.23 0.21 0.23 0.22 0.21 0.20

Having obtained these values, the chemist will report that the average
cobalt content for the granitic outcrop considered is (0.23 0.005) ppm,
where 0.005 is the standard error, discussed later. The statistical nature of
the chemist's results is evident. One may, of course, regard the entire out-
crop as a source of many millions of units or aliquots, each of 2.0 g. The
average cobalt content of the set of 20 granite samples is of no particular
interest in itself, but the 20 results may be used to make some inference
regarding the average cobalt content for the entire outcrop, i.e., in estimating
a numerical characteristic, the average.
The variation between the 20 results suggests that a similar variation
probably exists for the outcrop as a whole, so that if a further set of 20
specimens were taken, a different average would probably be obtained.
One may therefore ask whether such a determination i.e., the average of a
set, can provide reliable information about the true average for the entire
outcrop. It will be demonstrated that by statistical methods one can, in
fact, ascertain the limits between which the true average must lie.

1.2. The Importance of a Statistical Approach in Ana-

lytical Chemistry

Each observation in analytical chemistry, no less than in any other

branch of scientific investigation, is inaccurate in some degree. While the
"accurate" value for the concentration of some particular constituent in
the analysis material cannot be determined, it is reasonable to assume that
the "accurate" value exists and it is important to estimate the limits within
which this value lies.
The description of an analytical procedure is never complete without
reference to the error involved in the process. It is therefore imperative that
the chemist should be familiar with the basics of statistical methods, not
only from the point of view of consistency in the general presentation of
analytical results, but also in order to derive reliable estimates from the
observed data. This eliminates the risk of misinterpretation and permits
the interested analyst to assess correctly the application of a particular
analytical technique within his own field of study. For example, where large
2. Statistics 13

sampling errors arise it is not enough to suggest "replication at the sampling

stage" without further qualification. A quantitative estimate of the sampling
variation will avoid the effects of overrep1ication with possible sacrifice of
speed, or underrep1ication with loss in precision. It is also important to
bear in mind that sampling errors may vary for different types of the same
material. An example may be found in the variation in trace-element con-
tent of granites of different grain size. Again when the analytical treatment
involves several stages (multistage analysis) it,is desirable to indicate how
the total error of determination is distributed. Such presentation may indi-
cate how the time of operation can be cut down in any particular stage
without significantly affecting the precision of the final estimate. Such
considerations, of course, are not meant to imply that the analytical chemist
is not interested in the general "accuracy" of his results. Many a chemist
does, however, tend to interpret these results intuitively and, therefore,
often rather inefficiently.
It should be emphasized that the statistical approach is concerned with
the appraisal of experimental design and data only, whereas the analytical
approach is often confined to the analytical process. Statistical techniques
can neither detect nor evaluate constant errors or bias; the detection and
elimination of inaccuracies are analytical problems. Nevertheless, statistical
techniques can assist considerably in determining whether or not inaccuracies
exist, and also whether procedural modifications have effectively reduced
them. It would be wrong for the research worker to therefore assume that
mathematical calculations by themselves can obviate the need for common
sense and sound analytical technique in the chemica11aboratory. To quote
from Chambers1: " ... statistical methods are merely tools for a research
worker. They enable him to describe, relate and assess the value of his
observations. They cannot make amends for incorrect observation nor can
they of themselves provide a single fact of psychology, biology or any
other subject of research".
It is therefore appropriate to briefly outline the mathematical treatment
required for evaluating analytical data. This treatment is by no means
exhaustive, but it is hoped that it will already reveal to the reader the
statistical nature of certain problems encountered in analytical chemistry.
Statistical concepts are more fully discussed in a monograph by Pantony.2
An introduction to actual statistical calculations is provided by Chambers.1
For further background reading in the application of statistical methods to
chemical analyses Saunders and Fleming3 and Youden 7 may be consulted.
14 A. B. Calder

2. BASIC CONCEPTS

2.1. Variation
All analysts must have observed a certain amount of variability or
variation in their results (see Section 1.1). The science of statistics is to a
very large extent concerned with the study of variation, and statistical
methods are therefore applicable to all types of work where this phenomenon
is encountered. There remains, unfortunately, a tendency on the part of
the analyst to ascertain only average values and often the important char-
acteristic of variation is completely ignored. In this connection such loose
statements as "good duplicates" and "reasonably reliable data" are qualita-
tive statements only. They serve no useful purpose and are even liable to
misinterpretation. In chemical analyses the proper evaluation of data is an
intrinsic part of the procedure and calls for correct presentation. It will be
shown that variation is an important characteristic of a series of quantitative
data and that its assessment is essential for arriving at the objective type of
conclusion that each analyst should aim to achieve.

2.2. Populations and Samples: Parameters and

Statistics
A set of data that is taken from some source of observations for the
purpose of obtaining information about the source is called a sample,
whereas the source of these observations is called a population. Statistical
methods are concerned with the study of populations and their application
permits one to infer from observations made on a sample something about
the population from which the sample has been drawn. The distribution of
the total population can usually be expressed in a mathematical form by
using a small number of constants or parameters. These estimates are termed
statistics. Measures of parameters are often called descriptive statistics, since
they yield in a condensed form a description of a whole series of observations.

2.3. Types of Population and Variable

Previously (Section 1.1), a population (the granitic outcrop) was de-
scribed, with a variable or variate (the cobalt content) which may take any
value between certain limits. Here real (material) population and a con-
tinuous variable were considered. In statistical work other types of popula-
tion and variable may be dealt with. The population could be a hypothetical
one, for example, the possible results of an experiment; it could be afinite
population, for example, the number of granitic outcrops in the United
2. Statistics 15

Kingdom; or it could be an infinite population, for example, the possible

yields of a chemical reaction. The variable may be continuous, as above,
or discontinuous, for example, the number of defective items in a batch.
In this chapter the variables will be denoted by X.

2.4. Probability and Random Selection

In the throwing of an ordinary, symmetrical, cubical dice the total
number of possible events is six. A particular event, such as throwing a
four, constitutes one out of these six, and therefore we say that the probabil-
ity of throwing a four is t. Similarly the probability of throwing an even
number is i, i.e., t. We assume that all the events are likely or equally
probable.
In expressing limits of error, the probability P may be defined as the
probability that a quantity will lie between certain limits. In tests of sig-
nificance (see Section 6) the probability P' is the probability that a quantity
will lie outside the limits. For given limits the sum of the probabilities that
the quantity will be either within or outside these limits is 1 (i.e., the one
or the other of these hypotheses will certainly be correct):
P+P' = 1 or P' = I-P

The term random is used in connection with the selection or allocation

of one or more objects from a group. In the statistical sense a selection is
said to be random when it is made in such a way that every object has the
same probability of being chosen. Random sampling, as we shall see later,
is often difficult to achieve. The human mind has a tendency to exercise a
preference or bias, according to some physical characteristic which influences
the senses. To overcome this type of influence, complete mechanization of
the sampling procedure is desirable. Sometimes it may be necessary to select
a number of items, in which case an individual may have a personal bias
for particular numbers. This may be overcome by consulting a table of
random numbers prepared by mathematicians for this purpose. If one does
not have a set of random numbers available it may be more convenient to
draw cards from a pack.

3. MEASURES OF LOCATION AND DISPERSION

If a whole population is examined for a particular property, a plot of

values of the measurement of this property vs the number of times a value
occurs (i.e., its frequency) is termed thefrequency distribution of the variable
16 A. B. Calder

Value of measurement,X

Fig. 1. Histogram of measurements.

in question. It is usually impossible to draw a continuous curve through the

points of a frequency distribution because of the relatively small number of
measurements made. In such cases the values are grouped into increments
of equal size (i.e., class intervals) and the distribution then takes the form
of a histogram shown in Fig. 1. If, however, a very large number of observa-
tions are made enabling a smaller class interval to be used, i.e., if the points
along the base line are much more numerous, then the boundary of the
histogram becomes a smooth continuous line corresponding in shape to
the curve shown in Fig. 2.
The concept of a frequency distribution is basic to all statistical work
and we must therefore now consider the parameters defining a frequency
distribution and their numerical expression. The measures of primary inter-
est are the location of the distribution or average value of the variable, and
the spread of the distribution, i.e., the variability.

3.1. Average: Arithmetic Mean

The average value may be expressed in a number of forms, such as,
the arithmetic mean, the median, or the mode. The best known and most
useful form of average is the arithmetic mean, usually referred to simply

,.,
()
c
OJ
:J

~
U.

Distribution

Fig. 2. Frequency-distribution curve.

2. Statistics 17

as the "average" or "mean." For a series of n observations Xl' X 2 , , X n ,

the mean, written as X, is given by

where E represents the summation. Our purpose here is not merely to

calculate the mean of a series of sample values, but to estimate the mean
in the population from which the sample was drawn. The corresponding
population parameter is written as ft and it can be shown that the sample
mean, X, is in fact a satisfactory estimate of ft.

3.2. Variability: Standard Deviation and Variance

The only useful measure of scatter or dispersion suitable for critical
work is the standard deviation. For a'series of n observations Xl' X 2 , , Xn
the standard deviation is given by s = [E(X - ft)2jn]1I2; in practice ft is
not generally known, and we rely on the sample estimate X. The quantity
Sl = [E(X - X)2jn]1/2 is not, however, an unbiased estimate of the cor-
responding population parameter a and this effect is counterbalanced by
replacing n by n - 1 in the denominator. The general rule is to use Roman
letters for sample estimates (e.g., X for mean; s for standard deviation)
and Greek letters for population parameters (i.e., ft for mean; a for standard
deviation). Hence, the standard deviation is given by s = [E(X-X)2jn-l ]1/2
with the usual notation. The square of the standard deviation, i.e., S2, is
known as the mean square or variance v of the population. The interesting
property of variance is that it is linearly additive provided that the sources
of the variances are independent.

3.3. Degrees of Freedom

A concept of great importance in statistical analysis is that of degrees
of freedom. Consider again the series of n observations Xl , X 2 , , Xn and
their sum El(X). If the average (or sum) is fixed and n is also fixed, then
we can give n - 1 of the X's any values we please but the nth will be fixed.
Thus, for n = 6 and E(X) = 25 if we give to Xl up to X5 the values 7, - 2,
6, 8, -1 summing to 25, then X6 must be 7. We say there are five degrees
of freedom given by n( = 6), minus the number of constraints (= 1) that
have been placed on the system by fixing the average (or the sum). A small
sample (in the statistical sense) of observations tends to underestimate the
variance, and a better estimate is obtained by dividing the sum of squares
E(X - X)2 by the number of degrees of freedom n - 1, whence v(variance,
unbiased estimate) = [E(X - X)2jn - 1] (see above).
18 A. B. Calder

It must be pointed out that the number of degrees of freedom will

not always be simply n - 1, since in certain more advanced calculations
there may be r constraints r > I and therefore only n - r degrees of freedom.

3.4. Calculation of the Standard Deviation

The calculation of the standard deviation is not a difficult matter when
the observations consist of small integral values and X is also integral.
This is seldom the case in practice and to facilitate the arithmetic the follow-
ing procedure is adopted for obtaining the sum of squares. Subtracting
some quantity M from the X's and working with the quantities X - M
we have
r(x - M)/n = r(X)/n - M = X - M
and, therefore,
X = r(X - M)/n +M
Now
r(x - M)2 = r[(X - X) + (X - M)]2

r
= r(X - X)2 + 2r[(X - X)(X - M)] + r(X - M)2
= r(X - X)2 + 2(X - M) r(X - X) + n[ r(X --;; M)

= r(X _ X)2 + [reX - M)]2

n

since
r(x - X) =

Thus
r(x - X)2 = r(x __ M)2 _ [reX - M))2
n

If a calculating machine is available we simply put M = 0, giving

r(x - X)2 = r(X2) - [r(X))2/n

3.5. Coefficient of Variation

An error or variation is expressed as a coefficient of variation. A per-
centage coefficient of variation is derived from the standard deviation
(Section 3.2) according to the relation
coefficient of variation (V) = lO~s
X
2. Statistics 19

4. FREQUENCY DISTRIBUTION

4.1. Normal Distribution

The majority of frequency distributions encountered in analytical
practice are of much the same shape as that shown in Fig. 2, which depicts
the shape of a normal or Gaussian distribution. Such curves are virtually,
though not exactly, symmetrical. They may be made nearly symmetrical
by a suitable transformation of the variable, such as expressing the variable
in logarithms for example. Many of the methods of statistics depend on the
assumption that the variables under consideration are normally distributed,
and most of the methods cannot strictly be applied unless this assumption
is justified. Biological variables are, for example, often distributed in such a
manner, but it is not safe to assume that any particular distribution is
normal without prior examination.

4.2. Equation and Properties of the Normal Distribution

The normal curve has an exact mathematical formula and its equation

r
can be given in terms of the standard deviation:

loge ( :J = - ~ (X ~ Il
where y is the height of the curve above the horizontal axis and Yo is the
value of the ordinate corresponding to X = Il. The equation therefore
defines y for any value of X between 00.
This curve has important properties. It is a continuous curve and
applies to continuous variables, where the difference between one value
of the variable and the next can be infinitesimally small. It is symmetrical
about the mean Il and falls off on either side, tailing off asymptotically to
the X axis in both directions. Only the portion drawn in Fig. 2 is of practical
importance. The points of inflexion occur at X = Il (1. The area under
any part of the curve is proportional to the fraction of the population lying
between the corresponding values of X. Thus, if a single observation is
drawn at random from this population the probability that it will lie be-
tween X = p and X = q is proportional to the area between these limits.
The probability P' of main practical interest here is that of an observation
lying outside the limits Il d. The value will, of course, depend on (1, and
to calculate it we form first what is known as the normal deviate:
X-Il d
U=
(1 (1
20 A. B. Calder

The probability P corresponding to any value of u( = d/Cf) may be derived

mathematically. Some numerical values of pi for given values of u and vice
versa are given in Table I:

Table I
Probability of a Random Observation Lying outside the Limits ft d

u( = dla) P'(= 1 - P) P'(= 1 - P) u( = dla)

0.3173 0.10 1.64

2 0.0455 0.05 1. 96

3 0.0027 0.01 2.58

These figures mean that if a value X of the variable is selected at random

from a normal distribution, there is a 0.3173 (or 31.7%) probability that
its normal deviate will lie outside the limits 1, a 0.0455 (or 4.5%) proba-
bility that it will exceed 2, and so on. These are often referred to as the
68 and 95% confidence limits of a series of results. Such probabilities refer
to the plus and minus range, and are called double tailed; those referring
to the plus or minus range alone are called single tailed, for which the
probability is exactly one half of that of the double tailed data. Alternatively,
referring again to Table I and selecting round-figure probabilities, there is
a 0.10 (10%) probability that the normal deviate will exceed 1.64 and a
0.01 (1%) probability that it will exceed 2.58.

4.3. Justification for the Assumption of Normality in

Practice

The majority of continuous variables encountered in analytical chemical

practice are distributed in approximately normal fashion. A moderate dis-
placement from normality seldom gives rise to appreciable errors, since (a)
it can be shown that the distribution of the mean tends to normality with
increase in sample size, irrespective of the form of the parent distribution
(central limit theorem), and the measures of probability in significance
tests are computed with reference to the sample average, and (b) marked
2. Statistics 21

deviations from normality can be greatly reduced by transformation of the

data onto another scale, such as the logarithmic scale, for example.

4.4. Binomial and Poisson Distributions

In some cases the significance of observed results may be tested by the

use of the binomial distribution. This distribution consists of the expansion
of the expression

N(p + q)n
= N ( pn + npn-Iq + ... + n(n - 1) ... (n - r + 1) pn-rqr+ ... + qn
)
r!

Usually in probability problems, p is the probability of an event occurring

(i.e., a success) and q is the probability of an event not occurring (i.e., a
failure). If both are expressed as fractions, then p + q = 1. N is the number
of trials made and n is the number of events in each trial. It can be shown
that the mean of a binomial distribution (p + q)n is np and that the standard
deviation is (npq)1!2. The form of the binomial distribution varies consider-
ably depending on the values of p and n. When p is not small but n is large
enough, it can be verified that the binomial distribution approximates to the
normal error curve.
One important practical case is when p is very small, that is, the prob-
ability of the event happening is very small, but n is large, so large, in fact,
that np is not insignificant. It can be shown that if p is small while n is large,
such that np = m, the binomial expansion of (p + q)n approximates closely
to the series

This is known as the Poisson series and any distribution which corresponds
to the successive terms of this series is called a Poisson distribution. The
mean of the distribution is m and the standard deviation is ml / 2 Data con-
forming approximately to Poisson distributions occur widely in science.
They also arise in many industrial problems.
These distributions can be applied to the study of counting techniques
such as those used in dust-particle analysis and the quantitative estimation
of radioactivity. In these measurements only a small proportion of the total
number of particles present are counted by the device used and so the rare
event is that a particle happens to be counted. Compared with the applica-
tion of the Gaussian expression, however, their use is limited.
22 A. B. Calder

5. STANDARD ERROR

5.1. Sampling Distributions

In most instances we are more interested in the reliability of the sample

mean X than with that of an individual observation. When systematic errors
are eliminated the standard deviation of a set of repeated measurements
may be used to give an assessment of the random error of the mean. More
definite conclusions about the reliability of a mean can be obtained by con-
sidering the scatter of a series of means, each obtained from a number n
of repeated measurements, about the true value of the measurement. Con-
sider a population of single results divided into groups or samples, each
containing a set of n values. The distribution of the X's is known as a
sampling distribution and may be treated in exactly the same way as the
parent distribution.

5.2. Standard Error

Obviously, the mean of a sample will generally not be the same as the
mean of the whole population. If n is large then two means may not differ
very much, but if n is small they may. It is possible to express mathematically
the manner in which the means of different samples of given size are distrib-
uted. It can be shown that this distribution is itself normal and such that
its mean equals the true mean of the whole population, while its standard
deviation equals (J Iyn, where (J is the standard deviation of the whole
population. In Fig. 3 it is illustrated how the distribution of the means of
samples retains its normal form, but decreases in dispersion with increase
in sample size. The standard deviation of the sampling distribution of the
mean has been given a special term, the standard error and may be used for
probability calculations concerning the mean in exactly the same way as
the standard deviation is used in connection with single 'observations. Since

.... means of91tems

____ . means of 4 items

____ individual items

Fig. 3. Distribution of the means of samples.

X is the grand average for entire population.
2. Statistics 23

an experiment or analysis usually consists of a single sample only (using

sample in the statistical sense) the standard error cannot be estimated
directly and recourse is made instead to the relationship already referred
to above, namely, ax = ajyn. In practice (a being generally unknown),
an unbiased estimate of ax is obtained by calculating Sx = sjyn, where
Sx is an estimate of the standard deviation of the series of means that would
be obtained if further samples were drawn.

5.3. Theoretical Derivations

5.3.1. Variance of a Function of Variates

Let A be a function of X and Y, i.e., A = f(X, Y). If A, X, and Yare
the mean values of A, X, and Y, respectively, then .1'= f(X, Y). Consider
two values of X and Y which differ from their means by the small amounts
LlX and LI Y, respectively. The deviation LlA of the corresponding value
of A from its mean is given by

aA aA
LlA ~ ax LlX + ay LlY

If a large number n of pairs of values of X and Yare considered, the variance

of the resulting values of A about their mean will be I(LlA)2jn. But

(LlA)2 = ( :~ r (LlX)2 + 2( :~ )( :~ )(LlX)(LI Y) + ( :~ r (LI yp

Therefore

aA
aA 2 = ( ax )2 I(LlXpjn + 2 ( ax
aA) ( aA)
ay I[(LlX)(LI Y)J1n +
+ ( :~ rI(LI y)2jn

The term [I(LlX)(LI Y)]jn represents the covariance of X and Yand is defined
as the average value of the product (X - flx)(Y - fly). But if the X's and
Y's are independent, i.e., if the probability that anyone of them will assume
a certain value does not depend on the value assumed by any other, then
I[(LlX)(LI Y)] = O. Now, I(LlX)2jn = ax2 and I(LI y)2jn = ay2. Therefore,

2_ (
aA -
aA)2
ax ax
2 + (aA )2 2
ay ay
24 A. B. Calder

5.3.2. Variance of a Sample Mean

Consider a large distribution of values of the variable X. If X is the
mean of a sample of n of these values, then

As Xl' X 2 , , Xn are drawn at random from the distribution, they can be

regarded as independent of one another and, therefore,

From Section 5.3.1 we can deduce ax 2 thus:

Now ai-I' ai-., ... are each equal to the distribution variance a 2, and as
there are n terms in the expression for ax 2 , it follows that

As ax 2 is the variance of the sample mean about the population mean p"
ax is the standard deviation of X about p" i.e., the standard error of X;
thus,

5.4. The t distribution

For a variable X distributed normally about a mean p, with standard
deviation a, we can calculate the probability that a random observation
will lie within or outside certain limits by obtaining the normal deviate
(Section 4.2) as u = (X - p,)/a, and entering the appropriate probability
table. Similar calculations may be made in connection with a sample mean
X by taking u = (X - p,)/ag = (X - p,)y'ii/a. In practice a will generally
not be known and we have to depend on the estimate s. We then have
corresponding to the normal deviate the quantity t given by t = (X - p,)/sx
= (X - p,)y'ii/s. For large values of n the distribution is nearly the same
as that of the normal deviate, but for small values the discrepancy becomes
more marked. The values of t vary with the number of degrees of freedom
in the calculation of s and with the probability level chosen for the limits.
A table of values of t for the 0.95 probability level is given in Section 10.1.
2. Statistics 25

6. TESTS OF SIGNIFICANCE

6.1. Introductory Note

If two mean results differ from one another, two hypotheses can be
advanced to explain the discrepancy; either it is due to chance experimental
error, or it is due to a significant difference. The first hypothesis is known
as the null hypothesis, and the level of significance of this hypothesis can be
estimated by using tests of significance.

6.2. Normal Deviate Test

This test is used to determine whether a sample mean X differs sig-

nificantly from a known population mean ft, the population standard devia-
tion (] being known. If the difference between X and ft is due only to chance,
i.e., the null hypothesis is assumed, this means that X is a sample mean
drawn from a normal distribution of such means about ft. The probability
that a sample mean (chosen at random) will lie outside the limits given by
u is Pl( = 1 - P). If P' is small, the probability that a chosen sample mean
will have as large a value as u will be small. The probability, then, that the
sample mean has been drawn from the normal distribution is small and the
null hypothesis is likely to be incorrect. p' can therefore be regarded as a
measure of the level of significance of the null hypothesis. The exact level
of significance selected for acceptance or rejection of the null hypothesis
varies with the type of problem considered. Conventionally, but arbitrarily,
an effect is considered to be significant if the probability p' is less than 0.05
and highly significant if it is less than 0.01. In other words, if pI > (i.05 the
null hypothesis is accepted, and the difference X - ft is ascribed to chance.
If P' < 0.05, the null hypothesis is rejected, and the difference X - ft is
considered to be significant. Referring to the table in Section 4.2 we find
that the value of u corresponding to P' = 0.05 is 1.96. If u lies outside
these limits, X - ft is significant, whereas if u lies within these limits, X - ft
is not significant.

6.3. t Test
The t test is used to decide whether the difference between two sample
means is significant when the population mean ft and standard deviation (]
are unknown. The quantity t which has a theoretical limiting value for a
given probability level, is calculated from two sets of experimental data.
26 A. B. Calder

As in the normal deviate test, if the calculated value exceeds the theoretical
value at P' = 0.05, the null hypothesis is likely to be incorrect, and there
is a significant difference between the two sets of results (see Section 10.1).

6.4. Variance Ratio (F) Test

Another method by which two samples of values of a quantity can be
compared with one another is the "variance ratio" test. Here we have to
introduce the F distribution, F being the ratio of the two variances (always
the larger variance divided by the smaller variance). If we have two inde-
pendent estimates of variance from the same normal population, S1 2
(d.f. = n1 ) and S2 2 (d.f. = n2 ), the probability that the ratio F = S12/S22
exceeds a certain value is a function of n1 and n 2 only, and this function
can be referred to tables for various levels of probability (see Section 10.2).

7. REGRESSION

7.1. Introduction
The scatter diagram obtained when values of Y, a quantity liable to
random error, are plotted against X, a quantity with negligible error, is
shown in Fig. 4. The line of best fit through these points can be drawn by
eye but is dependent upon the subjective judgment of the person who draws
it. Consequently the position of the line will differ slightly from person to
person. A line of best fit independent of individual judgment must be drawn
mathematically. Such a line is called a regression line. Regression provides
the best-fitting relation between two quantities, one of which is subject to
chance error. For example, in instrumental analysis the calibration involves
plotting the instrumental "response" (Y, e.g., "wave height" in polaro-
graphy, "drum reading" in absorptiometric analysis) against concentration

x
2Q;
E
o x
-g
ra
a:
>~--------------------
X (Negligible error)

Fig. 4. Scatter diagram of Y against X.

2. Statistics 27

of standard (X). The experimental error in X is negligible but an appreciable

error may arise in Y. The simplest case, and the one we shall deal with here,
is when the relation between Yand X should theoretically be a straight line.

7.2. Method of Least Squares

When drawing a line of best fit, an attempt is made to minimize the
total divergence of the points from the line. In computing the line mathe-
matically, the same idea is followed, but it has been found that the best line
(the regression line) is one which minimises the sum of the squares of the
vertical deviations from the straight line. This approach is known as the
method of least squares.
If the equation of the regression line is

y=mx+a

and if the experimental points are (Xl' Yl ), (X2' Y2), . .. , (Xn' Yn ) and Yl is
the value of Y calculated from the equation of the regression line at x = Xl'
the vertical deviation Lll of the point (Xl' Yl ) from the line is given by

and
(Lll)2 = (Yl - mXl - a)2
= Y1 2 + m 2X 2 -
1 2mXl Yl - 2aYl + 2amXl + a2
If z represents the sum of the squares for all similar squared deviations,
then

z = .E(P) + m2.E(X2) - 2m.E(XY) - 2a.E(Y) + 2am.E(X) + na 2

where n is the number of points. For a given set of points all the sums .E(
in the above equation will be constants, and the value of z will vary with
the values given to m and a. The conditions for z to be a minimum are

( ~)
aa m =0
( ~)
am a =0
Now
(:: t = -2.E(Y) + 2m.E(X) + 2na = 0
28 A. B. Calder

i.e.,
E(Y) mE(X)
--=--+a
n n
or
Y= mX +a
The regression line therefore passes through the point (X, f). Again

( :~ ) a = 2mE(X2) - 2E(XY) + 2aE(X) = 0

Therefore,
E(XY) = mE(X2) + aE(X)
But a = Y - mX, so that

E(XY) = mE(X2) + (Y - mX)E(X)

Writing nX for E(X) gives

E(XY) - nXY
m= E(X2) - nX2

and therefore
E(XY) - E(X)E(Y)jn
m=
E(X2) - [E(X)]2jn

The regression line is then given by y - Y = m(x - X), a being eliminated

by the result that when y = Y, then x = X.

7.3. Variance about Regression

It can be shown that the variance of a set of points about their regression
line 3 is given by
E(P) - nY2 - m2[E(X2) - nX2]
S y2 = -''--'---------:;c----'---''--------=-
n-2

7.4. Error of Estimation of X from Regression Line

In instrumental methods of analysis, the instrument is calibrated with

substances of known concentration. A regression line is used as the calibra-
tion curve for determining the concentration of an unknown substance
from the instrumental "response." The error involved in predicting x values
2. Statistics 29

from this curve can be computed by recognized statistical methods 3 as

sxt, where t has (n - 2) degrees of freedom, and

where Sy2 represents the variance about regression (see Section 7.3), and r
denotes the appropriate replication.

8. ANALYSIS OF VARIANCE

The technique of analysis of variance is a powerful tool for determin-

ing the separate effects of different sources of variation in experimental data.
To illustrate the principles involved we shall examine its usefulness for
interpreting the results of biological assays.15
In bioassay processes, doses or diets of increasing magnitude are given
to groups of animals and the biological responses observed. The response
measured may be, for example, the gain in weight by a whole animal during
a period on a specific diet (vitamin assay) or the gain in weight of a partic-
ular organ (hormone assay), and so on. In an ideal situation the response
of each animal to a given dose would always be the same. In addition, the
responses at the different dose levels would be a linear function of the
logarithm of the dose. This situation is virtually never found in practice
and the object of the statistical technique is to compare the variation in
response due to increasing dosage with that due to other causes, particularly
individual animal variation within a dose group. It can then be assessed
whether the variation due to increasing dosage is significant when viewed
against the background of animal variability. If the latter contributes sig-
nificantly to the total variation the selected bioassay process is unsatisfactory
and the responses do not truly represent the effect of increasing dosage.
It is also possible from the statistical analysis to decide whether the response-
log dose relationship is linear within the limits imposed by animal
variation.
The calculations required for an analysis of variance in a problem of
this type with a total of n observations arranged in c columns each con-
taining n/c observations are shown in Table II. Y represents an individual
response and X is the logarithm of the dose for each group, i.e., the quantity
which is varied from one column to another, and m is the average response
for a dose group, i.e., the column mean.
The results may be interpreted as follows. The mean square (variance)
due to dose difference M2 is compared with that due to animal variability
30 A. B. Calder

Table II
Analysis of Variancea

Source of variation Sum of squares d.f. Mean square

Total Sl = EP - (EY)'/n n- 1
n
Between columns S. = - (Em') - (EY)'/n c - 1 M. = S./(c - 1)
c
Within columns Sa = Sl - S. n-c Ma = Sa/en - c)
About regression S, = S. - Ss c-2 M, = S./(c - 2)
_ n (EXm - EXEm/c)' )
Due to regression Ss - - Ms = Ss
c EX' - (EX)'/c

a Courtesy The Statistician. IS

Ma. If M2 is very much greater than Ma the response gives a reasonable

measure of dose level or concentration and can be used for assay purposes;
i.e., the bioassay process is sensitive. To confirm that the two variances are
significantly different, the ratio (larger mean square)/(smaller mean square)
is calculated and compared with the theoretical value of the variance ratio
F for pi = 0.05 and the appropriate degrees of freedom.
Additional information regarding the suitability of the process for
purposes of biological assaying is obtained by carrying out a regression
analysis on the data, assuming that the average group response m is a
linear function of log dose X. The sum of squares due to doses S2 is sub-
divided into two components: (a) S4' the sum of the squared deviations of
the (X, m) points from the regression line, and (b) S5' the sum of the
squared deviations from the mean of the theoretical responses calculated
from this line. S4' the sum of squares about regression, measures the scatter
of the (X, m) points about the line. S5, the sum of squares due to regression,
measures the squared deviations from the grand mean due to the fact that
there is an m/ X regression line with finite slope.
Although M4 may be greater than M a, it is reasonable to assume,
providing M4/Ma (the mean-square ratio about regression) <Ffor P = 0.95,
that the difference between variances "within columns" and "about re-
gression" is not significant, or that they are sample variances from the
same distribution. Therefore the scatter of the (X, m) points about their
regression is only what might be expected from animal variation and it is
thus reasonable to assume that there is a linear relationship between response
2. Statistics 31

and log dose. If M4/ Ma is greater than F, the scatter about regression is
greater than can be ascribed to animal variability thereby suggesting that
the true relationship is not linear.
In instrumental analytical techniques the treatment of known stan-
dard dilutions (in solid or solution form) is replicated according to some
recognized instrumental procedure and the instrumental response or read-
ing is noted. In such techniques the "within-columns" variation is always
negligible compared to the "between-columns" variation and this is fully
demonstrated in typical calibrations. In geochemical assaying procedures,
it may be quite considerable since, in general, the reproducibility of mea-
surements on geologic systems in usually poor. Nevertheless, in instrumen-
tal techniques the analyst often assumes the within-columns variation to
be less than it actually is. For example, in the colorimetric determination
of manganese in a specific rock sample to be analyzed, the pretreatment
involves oxidation of the manganese to permanganate with subsequent
measurement of the optical density of the colored solution. Usually known
standard solutions of KMn04 are made up covering the required concentra-
tion range and the colorimetric measurements are carried out in duplicate.
The closeness of the duplicates (or even replicates) merely indicates the
precision of the colorimetric procedure and yields a false comparison with
the "between-concentration" (i.e., "between columns") variation for the
analytical process as a whole. A more correct procedure of determining
the precision of the method, would involve preparing a known set of stan-
dards conforming both in general composition and physical form to the test
samples. The required concentration range with regard to manganese should
also be covered, of course. The standards should then be treated in every
respect in the same way as the test samples. A simplified treatment of
variance and its evaluation may be found elsewhere. 4

9. SAMPLING METHODS

9.1. General Comments

An analytical determination, whether carried out in the laboratory
or on factory scale, is only as useful as the sample on which the measurement
is made. No amount of replication at the analytical stage will increase the
accuracy of our information, or make up for deficiencies in the sampling
technique during the collection of the analysis material. If the analysis
material is homogeneous in character, i.e., of a single phase of solid or
liquid such as a bar of metal alloy or a tank of homogeneous liquid, for
32 A. B. Calder

example, little attention need be paid to the sampling process, since any
part of the material will be properly representative of the whole. In the
laboratory it is generally not particularly difficult to obtain a sample which
is truly representative of the whole, although errors may arise due to faulty
mixing in routine-control operations (Section 11). In dealing with hetero-
geneous material, involving a mixture of two or more phases, such as in
sludges or minerals, a small sample taken from anyone part of the aggregate
is liable to give a false picture of the aggregate as a whole. Furthermore,
with heterogeneous material no sampling process can ever give an exact
representation of the aggregate and a sampling error will therefore occur.
This error will depend upon the size of the sample, the variability of the
material, and the sampling scheme adopted. The sampling error must be
sufficiently low to meet the requirements for which the sampling is being
made. The requirements are, then, to obtain a sample that will provide an
unbiased estimate of the corresponding aggregate parameter, and the sam-
pling error showed be reduced to some adequately low level. An important
objective of statistical examination here should be to minimize expenditure
of time and effort on the sampling and analytical procedures, in order to
obtain the degree of precision which the situation demands.

9.2. Remarks on Analytical Procedure

The standard deviation has the same dimensions as the variable to
which it refers: but the variance S2 has the advantage of additivity. This
implies that if there are a number of independent causes of variation operat-
ing simultaneously on a system with variances S1 2 , S2 2 , , the resultmg total
variance ST 2 is given by

Since the coefficient of variation V is by definition related directly to the

standard deviation s, V2 ex: S2 and, hence,

In the analysis for a particular element in the sample to be analyzed, there

are two main factors which may contribute to the normal variation VI"
The first is the variation or error involved in the laboratory VL , i.e., the
error inherent in the analytical technique, and the second, the variation or
error involved in the collection of the analysis sample Vs , i.e., the sampling
error. From the above equation relating component variances to total
2. Statistics 33

variance, it follows that

If there was no error in sampling (i.e., Vs = 0), the total variation would
depend entirely on the analytical technique (i.e., VT = VL ) and could be
reduced by replication of analysis.
In multistage analytical techniques that often involve some form of
chemical pretreatment prior to an instrumental assay, it is desirable to
consider how the error is distributed within the analytical system. Such
considerations eliminate the possibility of confusion arising in the discussion
of the method and may suggest how the operational time may be reduced.
Suitable presentation of the error distribution also permits the interested
analyst to assess the application of that particular analytical procedure with-
in his own field of work. We can therefore write

For example, in the colorimetric determination of copper in ore samples,

the analytical method involves a chemical pretreatment to obtain the sample
in solution before the instrumental measurement may be made.

9.3. Computation of Error Distribution

How do we determine the magnitudes of these quantities for, say, a
granitic outcrop? VT 2 may be obtained by calculating ST 2 from the analytical
data for n rock samples, taken at random from the particular area concerned
(uniform within itself with regard to rock type). V L 2 is obtained by eliminat-
ing the effect of Vi and computing SL 2 from the data for several similar
portions of the same granite sample; i.e., the complete analysis is performed
on each of n similar portions of a carefully mixed ground-rock sample.
Similarly, Vfnstr. is derived via slnstr. which is calculated from the analytical
results obtained for n aliquots of one particular extract.
In the above example the total error may be reduced in practice, either
by (a) increasing the number of samples taken and analyzing each sample
individually, or (b) if Vs is large in comparison with VL , by making a
composite sample of n subsamples and analyzing the composite sample.
In (a) the error of the mean is given by

where n represents the number of samples taken. In (b) the error of estima-
34 A. B. Calder

tion V 2 is given by the relationship

Any replication contemplated in the sampling or analysis must necessarily

involve the following considerations: (a) the permissible error to be attached
to the final result, (b) the relative magnitudes of the analytical and sampling
errors, (c) the obvious necessity of keeping the sampling within practicable
limits of expenditure of time and labor, and (d) the futility, in many cases,
of replicating beyond the point where the major components have equal
magnitude in the final result.
In the above examples VI and V2 may be referred to as the errors of
determination of the method as opposed to the analytical error VL, and
that fraction of the error contributed by the analytical method is given by
V L2/(nVI 2) and V L2/V22, respectively. A useful account of sampling methods
is given by Laitinen. 5

10. STATISTICAL TABLES

10.1. Values of t
Table III

Value of t Value of t
Degrees of freedom Degrees of freedom
P' = 0.05 P' = 0.05

12.71 15 2.13
2 4.30 20 2.09
3 3.18 30 2.04
4 2.78 40 2.02
5 2.57 60 2.00
6 2.45 120 1.98
7 2.36 00 1.96
8 2.31
9 2.26
10 2.23
2. Statistics 35

10.2. Values of F (P' = 0.05)

Table IVa

n, b

n. 2 3 4 5 10 20 00

161 200 216 225 230 242 248 254

2 18.51 19.00 19.16 19.25 19.30 19.39 19.44 19.50
3 10.13 9.55 9.28 9.12 9.01 8.78 8.66 8.53
4 7.71 6.94 6.59 6.39 6.26 5.96 5.80 5.63
5 6.61 5.79 5.41 5.19 5.05 4.74 4.56 4.36
10 4.96 4.10 3.71 3.48 3.33 2.97 2.77 2.54
20 4.35 3.49 3.10 2.87 2.71 2.35 2.12 1.84
00 3.84 2.99 2.60 2.37 2.21 1.83 1.57 1.00

a Extracts from tables in Fisher and Yates: Statistical Tables for Biological, Agricultural, and
Medical Research, reproduced by kind permission of Oliver and Boyd, Ltd., Edinburgh.
b n, is the number of degrees of freedom for the greater estimate of variance.

11. APPLICATIONS TO ANALYTICAL METHODS

11.1. Introduction
The errors which affect an experimental result may be divided into two
types: (a) determinate, i.e., constant or systematic errors, and (b) indeter-
minate errors.
Determinate errors (a) are those which can be avoided or whose mag-
nitude can be determined and the measurements corrected. A determinate
error may have the same value throughout a series of observations and
may remain constant under a variety of conditions (a constant error) or it
may vary in magnitude and sign with the conditions (a systematic error).
Determinate errors are numerous and include instrumental errors and those
arising from reagents, errors associated with the manipulations of an anal-
ysis, personal errors, and errors which have their origin in the chemical or
physicochemical properties of the system.
Indeterminate errors (b) are revealed by the slight variations that occur
in measurements made by the same observer with the greatest care under
as nearly identical conditions as possible. They may be described as the
errors which remain even when all effects due to determinate errors have
been eliminated.
36 A. B. Calder

Important factors in the choice of an analytical method include sen-

sitivity, accuracy, precision, selectivity, concentration range, availability of
standards, sampling requirements, instrumental and equipment costs, and
time of analysis. Sensitivity, accuracy, and precision are of primary im-
portance.
The precision of a result is its reproducibility; the accuracy is its near-
ness or closeness to the truth. A determinate error causes a loss of accuracy
which mayor may not affect the precision depending on whether the error
is constant or variable. Indeterminate errors lower reproducibility, but by
making sufficient observations it is possible to reduce the scatter to within
certain limits so that the accuracy may not necessarily be affected. Sensitivity
in chemical analysis concerns the ability to detect the difference between
very small amounts of the analysis element, and is generally used to indicate
the smallest quantity of element that can be detected or what is better
defined as the lower limit of detection. In analysis we obtain a reading from
an instrument (the instrumental response) which, after correction for back-
ground effects (i.e., noise level, blank, etc.), is taken as a measure of the
amount of analysis-element present. It is thus really a difference in two
responses which determines the concentration of element present, and at
the lower detection limit these responses are similar in magnitude. Difficul-
ties of interpretation must obviously arise and it is therefore necessary to
have available an objective method for assessing when a real difference in
response exists. The requirement of objectivity involves the application of
statistical method. Accordingly, Kaiser and Specker9 have defined the min-
imum detectable difference in response as kV2s b , where Sb is the standard
deviation of the blank reading and the values of k are those associated with
the probability levels of the normal distribution, i.e., k = 1.00, 1.96, and
3.00 for 68, 95, and 99.7% confidence levels, respectively.

11.2. Reduction in Operational Time

In Sections 9.2 and 9.3 the distribution of errors is discussed with

special reference to spectrophotometric analytical data. Two examples based
on actual spectrographic results are given in Table V.
It is evident from Table V that for potassium the error inherent in the
spectrographic technique accounts for 25% of the total analytical error, and
about 2.5% of the error of determination. For manganese, the spectro-
graphic error accounts for some 50% of the total analytical error and about
2% of the error of determination.
2. Statistics 37

Table V
Distribution of Errors a

Analytical, V L 2
Error of
Sampling, Total,
Variate determina-
Spectro- Chemical Total Vi V L 2 + VB 2
tion, %
graphic pretreatment analytical

K (major) 9 27 36 364 400 20

Mn (trace) 25 24 49 1176 1225 35

a Variance as V2, based on single observations.

The spectrographic errors are therefore negligible and the analytical

errors relatively unimportant, or in other words, the normal variation is
almost wholly accounted for by the sampling variation. It is obvious that
effective reduction of the errors of determination can only be achieved by
replication at the sampling stage. Assuming a 25 X replication in the sam-
pling technique, the errors of determination for K and Mn 6 would become
vi 36 + 364/25 = 7.1% and vi 49 + 1l76/25 = 9.8%, respectively. Now,
in the spectrographic method employed, for example, it has often been
recommended that duplicate spectra should be recorded for each sample
to be examined. The idea behind this is to increase the reproducibility of
the instrumental performance. From the purely analytical point of view,
there is really no justification for such replication when we view the process
of determination as a whole. In Table V the spectrographic variances for K
and Mn are, respectively, 9 and 25, and these refer to results evaluated
from duplicate spectra. Suppose we were to record only one spectrum per
sample to be examined. This would be equivalent to doubling the spectro-
graphic variance. Hence, if we evaluate our final result from a single spec-
trum only, but still using a 25 X replication at the sampling stage, the total
variance for K would increase from (9 + 27 + 364/25) to [(9 X 2) + 27
+ 364/25] = 59.6. For Mn, the total variance would increase from
(25 + 24 + 1l76/25) to [(25 X 2) + 24 + 1176/25] = 121. The errors of
determination would therefore increase from 7.1 to 7.7% for K, and from
9.8 to 11.0% for Mn, respectively.
The differences in precision as between single and duplicate instrumental
analyses are therefore not sufficiently large to justify a preference for either
alternative. The operational time is, however, reduced to approximately
one-half when only single spectra are taken.
38 A. B. Calder

Let us now investigate more closely the instrumental stage of the

analysis. In flame-spectrographic methods we plot ()E/()B against concen-
tration, where ()E is the galvanometer deflection (i.e., the "blackening") for
the analysis line; ()B is the corresponding deflection for the background (i.e.,
the background spectrum of the carbon monoxide flame from an air-
acetylene source) measured at a convenient specified point in the immediate
neighborhood of the analysis line concerned. The background varies in
density throughout the length of the spectrum, but at any particular wave-
length it should be reproducible for a series of spectra. Its introduction
therefore compensates for fluctuations in the light source or small varia-
tions in total emission from one spectrum to the next. It serves, in fact, as
an internal standard. Each photographic emulsion carries in addition to the
sample spectra, a number of standard spectra from which the ()E/()B vs
concentration relationship is derived. In determining the elements K and
Mn it is necessary to obtain four readings: ()B(K), ()E(K), ()B(Mn), and
()E(Mn). Since the analysis lines for K and Mn, and hence their respective
backgrounds, lie very close together in the spectrum, the question arises
whether one background reading could serve for both lines? For example,
in the spectrographic determination of potassium and manganese, using
the K(4044 A) and Mn(4130 A) lines, respectively, the background read-
ings shown in Table VI were obtained,S from an emulsion carrying 30
spectra.
Table VI
Background Reading, (}B (cm)a

(}B(K) (}B(Mn) (jB(K) (}B(Mn) (}B(K) (}B(Mn)

26.5 26.8 24.8 24.7 25.9 26.3

26.4 26.9 24.7 24.9 26.5 27.3
27.2 26.6 25.8 25.9 26.5 26.4
27.2 27.4 26.2 26.0 26.8 26.7
27.2 27.3 26.9 25.8 26.6 26.5
26.6 25.9 26.2 26.1 27.0 27.3
27.4 26.0 26.0 25.8 26.2 27.7
27.3 27.3 26.3 26.7 26.2 26.2
26.8 26.4 26.6 26.2 26.4 26.7
26.4 26.1 26.0 25.8 26.1 26.8

a Courtesy, J. Appl. Statistics.

2. Statistics 39

The differences are small but not consistent; if we regard them as a

sample from the hypothetical population formed by the differences from
an infinite number of future assays, it may be determined whether the
observed values could have been drawn from a population where the mean
difference is zero. With the usual notation and writing the differences in
background as X, we have

n = 30 E(X2) = 8.92 S2 = 0.30754

E(X) = 0.2 (EX)2/n = 0.00133 S = 0.554

X = 0.0066 E(X - X)2 = 8.91867 Sg = 0.1011

The statistic t is calculated from

t = (X - ",)/Sg

where '" is the hypothetical population mean. On the null hypothesis,

'" = 0, and
t = (0.0066 - 0)/0.1011 = 0.065
Entering the t table with d.f. = 29, we have t = 2.045 (Pi = 0.05). The
mean is therefore not significantly different from zero. Further tests con-
firmed these results and on the basis of these the null hypothesis was ac-
cepted. Hence, there is no need to read both backgrounds and in this case
the background for the manganese line was recommended for use from the
point of view of operational suitability.
Let us now examine the combined effect of these procedural modifica-
tions at the instrumental stage of the analysis. Instead of recording two
spectra per sample, leading to four subsequent measurements per spectrum,
the proposed modifications only require the measurement of one spectrum
per unknown sample, and only three subsequent photometric observations
per spectrogram. The total exposure time has therefore been reduced by
one-half, and the time associated with photometric evaluation has been
reduced by five-eighths. Most important, is, however, the fact that no sig-
nificant sacrifice has been made in reproducibility. In actual practice, making
allowance for photographic processing, emulsion change over and so on,
the saving in operational time in the complete analytical process has been
found to be about 40%. If one is dealing with several hundred samples per
week, this saving in time would be invaluable. Further examples of how
simple statistical methods can be applied to the problem of standardization
in spectrochemical analysis, are given elsewhere. 8
40 A. B. Calder

11.3. Sampling Errors

If analytical results display a below-normal precision, then the cause
may lie in faulty sampling. The latter may be due to insufficient grinding
and/or mixing of the sample material.
In Table VII some analytical results are given for Fe20 a , as obtained
colorimetric ally from samples that were mixed for different periods of time.
It is apparent from Table VII that there is a substantial variation in
Fe content within each set, although the individual duplicates are in good
agreement with each other. These results, therefore, suggest a relatively
large sampling error, which may be due to the lack of homogeneity in the

Table VII
Analytical Results from Samples That Were Mixed for Different Periods of Time

Set 1, Set 2, Set 3, Set 4,

Time of mixing 1 min l!min 3 min 5 min

mg Fe.O. mg Fe,O. mg Fe,O. mg Fe,O.

0.180} 0 1800 0.193} 0 1940 0.183} 0 1865 0.170} 0 1715

0.180 . 0.195 . 0.190 . 0.173 .

0.187} 0 1870 0.193} 0 1900 0.197} 0 1920 0.173} 0 1715

0.187 . 0.187 . 0.187 . 0.170 .

0.183} 0 1825 0.193} 0 1915 0.170} 0 1715 0.173} 0 1730

0.182 . 0.190 . 0.173 . 0.173 .

0.201} 0 2035 0.183} 0 1815 0.173} 0 1835 0.173} 0 1715

0.206 . 0.180 . 0.194 . 0.170 .

0.167} 0 1645 0.185} 0 1835 0.173} 0 1775 0.173} 0 1730

0.162 . 0.182 . 0.182 . 0.173 .

0.197} 0 1985 0.170} 0 1685 0.181} 0 1790 0.170} 0 1715

0.200 . 0.167 . 0.177 . 0.173 .

0.195} 0 1960 0.170} 0 1700 0.195} 0 1860 0.175} 0 1725

0.197 . 0.170 . 0.177 . 0.170 .

0.198} 0 2000 0.183} 0.183} 0 1805 0.172} 0 1725

0.202 . 0.170 0.1765 0.178 . 0.173 .

Standard
deviation 0.01308 0.01062 0.00632 0.00069

Coefficient
of variation 6.9 5.7 3.5 0.4
2. Statistics 41

prepared sample. The calculated standard deviation and coefficient of varia-

tion clearly demonstrate the fact that longer periods of mixing systematically
improve the precision of the results.

11.3.1. Experimental Procedure

(a) Known solutions of purified 6N HCI containing Al and (varying)
Fe were treated to yield mixed 8-quinolinolate precipitates. 12 The pre-
cipitates were filtered, washed, dried in an oven, and finally ignited at 450C.
The mixed-oxide concentrates thereby obtained were ground in an agate
mortar, each one for a specified interval of time (in this case, four in all,
at 1, 1 to 3, and 5 min, respectively). Then from each concentrate a set of
eight 5-mg samples were weighed out and analyzed for Fe20 a by the colori-
metric method prescribed by ScottP The only modification was introduced
at the colorimetric stage proper, when two aliquots of the final analysis
solution were colored up and measured and the average Fe20 a content
evaluated by reference to a calibration curve.
(b) Ten 20-ml aliquots of a standard Fe solution (falling within the
concentration range of the prescribed method) were colored up in the usual
manner and the results noted.

11.3.2. Results
The results for (a) and (b) above are shown in full in Tables VIII
and IX. These results have been subjected to a statistical analysis and
Table X records the standard deviation and coefficient of variation for
each set. It should be noted that in the normal recognized method there
is no duplication at the colorimetric stage, although the effect of such dupli-
cation on the coefficient of variation is negligible.

11.3.3. Discussion
It is immediately apparent from the results in Table VIII that there is a
substantial variation in Fe content within each set, although the duplicates
are extremely close (cf. Table IX), suggesting a relatively large sampling
error (i.e., lac~ li)f homogeneity in the prepared concentrate). It is also
noticeable, as would be expected, that the variation within a set decreases
with the time of mixing and approximates eventually to that associated with
the precision of the colorimetric procedure.
The above results are specially significant in works-routine practice
where "unqualified" junior assistants may be given a set of instructions in
42 A. B. Calder

Table VIII
Analytical Data for Sets

Set 1, Set 2, Set 3, Set 4,

1 min mixing I! min mixing 3 min mixing 5 min mixing

mg FeS03 mg FeZ0 3 mg Fe S03 mg FesOs

0.180} 0 1800 0.193} 0 1940 0.183} 0 1865 0.170} 0 1715

0.180 . 0.195 . 0.190 . 0.173 .

0.187} 0 1870 0.193} 0 1900 0.197} 0 1920 0.173} 0 1715

0.187 . 0.187 . 0.187 . 0.170 .

0.183} 0 1825 0.193} 0 1915 0.170} 0 1715 O. 173} 0 1730

0.182 . 0.190 . 0.173 . 0.173 .

0.201} 0 2035 0.183} 0 1815 0.173} 0 1835 0.173} 0 1715

0.206 . 0.180 . 0.194 . 0.170 .

0.167} 0 1645 o.185} 0 1835 0.173} 0 1775 O. 173} 0 1730

0.162 . 0.182 . 0.182 . 0.173 .

~:;~} 0.1985 0.170} 0 1685

0.167 .
0.181} 0 1790
0.177 .
0.170} 0 1715
0.173 .

0.195} 0 1960
0.197 .
0.170} 0 1700 o0.177
.195} 0 1860
.
0.175} 0 1725
0.170 .
0.170 .

0.198} 0 2000 0.183} 0 1765 0.183} 0 1805 0.172} 0 1725

0.202 . 0.170 . 0.178 . 0.173 .

Table IX
Reproducibility Data for Colorimetric Procedure

mg FeS0 3 (20 ml aliquots) mg Fe Z0 3

0.201 0.200
0.201 0.201
0.200 0.200
0.200 0.202
0.201 0.200
2. Statistics 43

Table X
Statistical Analysis of Experimental Data

Set Standard deviation Coefficient of variation

0.01308 6.9
2 0.01062 5.7
3 0.00632 3.5
4 0.000694 0.4
Colorimeter 0.000494 0.2

which it is required to "thoroughly grind and mix the concentrate." In

such cases it may be assumed that 2, 3, or 4 min "thorough mixing" is all
that is required. Further, after mixing some dozen concentrates, the pro-
cedure becomes tedious for junior (and senior?) analysts and this fact must
be accepted realistically. It would, of course, be wrong to stipulate any
particular time of mixing, as the actual process of grinding and mixing will
always depend on the efficiency of the analyst concerned: at the same time,
assuming a high degree of efficiency in the above experiments, it would
appear that at least 5 min grinding and mixing is required for this par-
ticular stage in the analytical process. *

11.4. Calibration

In Section 7 we discussed the best-fitting relation between two quanti-

ties, one of which is subject to chance error. In experimental work, the
values for that quantity Yare usually measured for given values of the
other (X). Thus, in gravimetric analysis, for example, Y would represent
the weight of precipitate produced under certain specified conditions by a
concentration of the element represented by X. Let us, by way of illustra-
tion, consider data obtained during the development of a gravimetric method
of assay for the estimation of aluminium (as oxinate). A set of typical results
(n = 6) is shown in Table XI.
Referring to Section 7.2 and using the results in Table XI, we derive

This section is reproduced in its entirety by kind permission of the Council of the
Institute of Chemistry of Ireland.H
44 A. B. Calder

Table XI
Gravimetric Assay for Ala

Al (in mg) Al oxinate (in mg)

3.5 59.5
7.0 118.3
14.0 232.0
17.5 298.4
21.0 352.8
24.5 416.7

a Courtesy, Anal. Chern.!O

the equation relating weight of precipitate to concentration as

x = (y + 1.03)/16.96
and we can compute the error involved in predicting x values from this
expression by statistical methods referred to in Section 7.4.
Now, in teaching practice, for example, it is usual to draw up a series
of gravimetric exercises and ask the student to carry out each in duplicate.
It is doubtful if this procedure provides the student with anything other
than practice in manipulation, and certainly the information is extremely
limited. A more correct procedure would be to record additional data for
each exercise in the manner described above at the expense of the number
of exercises to be performed and to use the technique of regression for
dealing with the data in each case. These recommendations would, of
course, apply equally well to the analyst who is developing a new method
of gravimetric assay. The advantages in adopting such a procedure are as
follows: (a) more information is immediately available from the results.
A correct measure of the precision of the analysis is obtainable for the
full range of concentration investigated. If we put the regression (calibra-
tion) line equation in the form y = mx + a (see Section 7.2), the error in
I a I can be estimated and it can be determined whether I a I is significant,
both statistically and, therefore, chemically, in which case a "blank" effect
(constant error) is present. The standard deviation of an intercept is given
by Sa = sy{EX2/[nEX 2 - (EX)2]}l/2 and the value for t = I a I /sa is ex-
amined by means of the t table with n - 2 degrees of freedom. A value of t
in excess of the selected critical value is evidence that the data do not
2. Statistics 45

support the expectation that the calibration graph passes through the
origin. This should lead to further work in the laboratory to find the cause
for this departure. A positive blank here could indicate the presence of Al
impurity in the reagents and/or solutions employed. A negative blank,
although unusual, could indicate the presence of an impurity (e.g., citric
acid) which complexes with part of the aluminium, thereby rendering it
innocuous. (b) It is possible to determine if a significant difference exists
between the theoretical and observed values of m, the theoretical value in
this example being given by [AI(C9 H sON)a] -;- [AI]. The difference be-
tween m(theoretical) and m(observed) is tested by pooling the variance
estimates for each [zero for m(theoretical)]. Thus, the variance of the
difference is given by sy2[1/(EX2 - XEX)] and
t= 1 m(theor.) - m(obs.) 1

V Sy2 { EX2 ~ XEX}

with n - 2 degrees of freedom. A significant difference implies that the
reaction is not fully stoichiometric, but it should be noted that this by
itself does not justify rejection of the method. It is also important to note
that the above information would not be derived from an examination of
the data of replicate analyses carried out for only one point on the con-
centration scale. In other words, if we wish to examine the reliability of a
gravimetric (or, for that matter, instrumental) method of assay, it is essen-
tial to obtain estimates of m and 1a I. The extra time involved is completely
justified in view of the information to be obtained.
(c) The treatment outlined disposes with the need for graphical compu-
tation and eliminates errors associated with reading graphs.

11.5. Blan k Variation

In determining the concentration of a particular element X in the
analysis material, it is essential during any particular run of analyses to
make a blank determination, i.e., to obtain a measure of the amount of X
in the reagents and solutions employed. Any error arising due to the pre-
sence of X in the solutions, if it exists, will obviously be constant from one
standard dilution to the next; and when we balance test and reference
solutions in a photometer, we are, in effect, carrying out a simple subtrac-
tion, viz., X(analysis material + blank) - X(blank), which gives the value
of X referred to the analysis material. The same argument applies when we
are dealing with the addition of solid reagents (e.g., in fusions) in our
46 A. B. Calder

analytical scheme, provided that there is no variation in the X content of

the reagent. If the solid reagent is not homogeneous it would be futile to
carry out routine analyses using specific aliquots of reagent and to base
the estimate of the blank on the X content of only one such aliquot, since
the error concerned will add to the existing analytical errors. The possibility
of such errors arising is best eliminated by thorough mixing of the reagent( s)
at the commencement of the analysis. Reagent purification only adds to
the time of determination and is not always feasible or even necessary,
provided the average X content of the reagent is not sufficiently high to
cause a reduction in photometer sensitivity.
Such effects have been encountered for various batches of laboratory-
grade reagents. For example, the variation in Fe content of a particular
batch of KHS0 4 was found to be of the order of 20% (P = 0.95). Obviously
the use of such heterogeneous material for the fusion of analysis samples
in trace-Fe determinations could give rise to serious errors, depending on
the quantities of analysis and fusion materials used and the range of con-
centration being investigated.
The calibration curves shown in Fig. 5 demonstrate the effect. In the
first calibration 75 g KHS0 4 (in lots) were fused in a silica crucible, the
melts dissolved in 25% citric acid (200 ml altogether, to render Al innoc-
uous) and the extracts combined and made up with distilled water to
500 ml (A). A standard Fe solution (B) was then prepared in which 1 ml (B)
= 30 p,g Fe. Standard dilutions were obtained by adding to 50-ml aliquots
of (A) varying amounts of Fe, i.e., 2, 4, 6, 8, and 10 ml (B). To each standard
dilution in a 100-ml flask was added 1 ml of 25% thioglycollic acid followed
by concentrated ammonia drop by drop until a red coloration just appeared
due to the formation of the intensely colored ferrous thioglycollate ion
[Fe(SCH2COO)]2-' and a further two drops in excess. The solution was
cooled and made up to the mark with distilled water. The optical density

10i-----.---.--?"];J

O'51---719~==1l-:;----I

100 200 300

Ilg. Fe
Fig. 5. Blank variation due to iron in KHSO .
Courtesy of Spectrovision.
2. Statistics 47

was measured (against water) on a Unicam SP 500 Spectrophotometer at

Amax = 530 mfl using I-cm cells. In the second calibration the same general
procedure was adopted, except that the base solution (A) was not prepared
in bulk; each base solution aliquot was obtained by fusing a 7.5-g portion
of KHS0 4 selected from the reagent bottle, extracting the melt in 20 ml
of 25% citric acid and making up to 50 ml. The relatively large scatter of
points about the second calibration (regression) curve is due to variation in
the Fe content of the batch of KHS0 4 used; the error attaching to the blank
estimate may be evaluated from the regression analysis (see Section 7).
In certain cases it may not be possible to eliminate the effect of variation
in the blank. All that can then be done is to determine the analytical pre-
cision with the blank effect present; provided the error falls within the
desired limits, there is no need to remove the interfering element, bearing
in mind also the consequent increase in analysis time and analytical manip-
ulation, if the latter procedure is adopted. For example, small amounts
of Al in Ti02 (0.05--0.35%) can be determined by fusing a specific aliquot
of the titania with NaOH, and extracting the melt with water. To a specified
aliquot of the extract (containing 10-75 flg AI) in a degreased separating
funnel are added 45 ml of NH40H-NH4Cl buffer (to yield pH 9.0) and
10 ml of 1% wjv oxine in chloroform. The mixture is shaken for 3 min and
allowed to settle; the organic layer is then run off into a l-cm cell and its
optical density measured (against a reagent blank) at Amax = 390 mfl. A
series of standard dilutions of Al 20 3 in solution form (a) in extracts of
melts of specified aliquots of pure Ti0 2 , and (b) without the addition of
Ti0 2 , gave the calibration (regression) curves shown in Fig. 6a and b,
respectively. From the relatively large scatter associated with the former
curve and practically absent from the latter, the blank obtained in Fig. 6a
suggested possible interference due to the presence of Ti caused by in-
complete and nonreproducible hydrolysis of the titanate. This was con-

15.--.........---,-------.

10t---+-~~___r'---j

60

Fig. 6. Blank variation due to aluminum In

TiD . Courtesy of Spectrovision.
48 A. B. Calder

firmed by deriving the absorption spectrum of titanyl oxinate in chloroform

which indicated A. max (Ti) = 380 mt-t and partial absorption at A. max (AI)
= 390 mt-t. Future results may be read from the Al calibration curve and
the precision evaluated as in ordinary regression analysis. In the event of
the precision being found not to be acceptable, the Ti can be removed by
an initial precipitation with cupferron prior to extraction and color devel-
opment.*

11.6. Multicomponent Methods of Analyses

In spectrophotometric analysis we measure the amount oflight absorb-
ed A at a specific wavelength A. by a known length of solution b and relate
this to the concentration C of some particular constituent in the solution.
The fundamental relation between these quantities is given by
log (loll) = A = abc

where 10 is the intensity of the incident radiation, I is the intensity of the

transmitted radiation, A equals log (loll) which equals the absorbance, and
a is a constant characteristic of the absorbing material and wavelength
concerned.
In spectrophotometric methods for the quantitative analysis of mixtures,
the analysis depends upon the fact that the absorption spectrum of a mixture
is a linear superposition of the absorption spectra of the individual compo-
nents. The analysis therefore involves the setting up of a series of simul-
taneous linear equations. 15
The instrumental response (the absorbance in this case) due to the jth
component at the ith wavelength will be

where aij is the absorptivity, i.e., the absorption constant characteristic of

the absorbing species and the wavelength of the radiation; b is the cell
thickness and Cj is the concentration of the jth component. Assuming the
absorbances due to different components to be additive, the total observed
absorbance at the ith wavelength should be
Ai = L aijbcj = L kijcj
j j

This section is reproduced in its entirety by kind permission of Pye Unicam Ltd.,
Cambridge. 11
2. Statistics 49

For example, in the analysis of neodymium and praseodymium in acid

solution, the absorbance At, measured at A = 575 mp" is given by

and the absorbance A 2 , measured at A = 590 mp" is given by

Writing the linear equations in matrix form, we have

or
[Ai] = [k ij ] [Cj]

A =KC

Similarly in mass-spectrographic methods for the quantitative analysis of

complex petroleum fractions, the analysis depends upon the fact that the
mass spectrum of a mixture is a linear superposition of the mass spectra
of the individual components. To determine the contribution of each
component at a given mass, it is necessary to measure the mass spectrum
of each component in order to find the peak height for a given partial
pressure of the component. Since the peak heights are directly proportional
to the partial pressure of the gas, the coefficients determined for the pure
component can then be substituted into the series of simultaneous equations.
Thus, the peak height due to the jth component measured at mass m will be

where Xmj is the peak height at mass m with unit pressure of the jth com-
ponent in the inlet sample system; Pj is the partial pressure of the jth com-
ponent. Writing the equations in matrix form, we have

H=XP.

Let us return to our system of linear equations of the form A = KC.

Assuming the Ai and k ij have been measured, the problem is to determine
the best value of the Cj. If the number of elements in the matrix A is just
equal to the number in the matrix C, the equations have a unique solution.
Error in one of the Ai values produces an error in the C matrix. When
the problem is overdetermined (by measuring more Ai than the number of
50 A. B. Calder

Cj to be determined) the equations have no unique solution but a "best"

solution which is found from the least-squares criterion. It is often desirable

in multicomponent analysis to overdetermine the concentrations, thereby
minimizing the effects of random experimental errors.
Let LI = A - KC be the matrix of errors. Differentiating the product
LI* LI with regard to C to find the best value of C gives t

K*A = K*KC

Putting K*A = A' and K*K= K', the equation becomes

A' = K'C

A' and C have the same dimension which is the order of the square matrix,
K'. Since the matrix K' is symmetric, evaluation of the determinant or in-
verse of K' requires less effort than that required for a nonsymmetric matrix
of the same size.
The above methods have been applied with considerable advantage in
the evaluation of mass-spectrographic and absorption-spectroscopic ana-
lytical data, as, for example, in the simultaneous uv spectroscopic analysis
of molybdenum, titanium, and vanadium in solution. In mass-spectrographic
work the number of equations to be solved may be in the region of seven
or eight, but the analysis may be simplified initially by first fractionating
the mixture into several components or component mixtures by distillation.

11.7. Particle Counting

The Poisson distribution was discussed in Section 4.4. One of the best
instances of large n and small p is afforded by radioactivity. A small mass
of radium contains many millions of atoms. In a specified interval of time
a moderate number of atoms, constituting a very small proportion of the
number in the mass, will change with the emission of a or f3 particles. If we
treat the occurrence as one of pure chance, the Poisson distribution provides
a mathematical model, to which the variation in the number of particles
emitted in the specific time interval may be expected to conform. For
example, Rutherford and Geiger using the scintillation method counted the
number of a particles emitted by polonium per unit of time. Their results
are given below; f is the number of times n a particles were observed.

t Ll* represents transpose of Ll, i.e., the matrix whose rows are the columns of ,1.
2. Statistics 51

n o 2 3 4 5 6 7 8 9 10 11 12 13
f 57 203 383 525 532 408 273 139 45 27 10 4

We shall find the mean number of a particles emitted and the Poisson
distribution corresponding to this mean.
The mean number of particles is found to be 10105/2608 = 3.87. The
terms of the Poisson series with m = 3.87 are

r + 3.87 + 3.87 /2 + 3.87 /6 + ... )

3 . S7 (1 2 3

that is, 0.021 + 0.081 + 0.156 + 0.202 + 0.195 + 0.151 + 0.100 + 0.054
+ 0.026 + 0.011 + 0.004 + 0.002 + 0.001 + 0.000.
On multiplying by 2608 the successive terms become 55, 211, 407, 527,
509,417,261, 141,68,29, 10, 5, 3,0, which are in good agreement with
the values of the frequency given above.

REFERENCES

1. Chambers, E. G., Statistical Calculation for Beginners, University Press, Cambridge,

1958.
2. Pantony, D. A., A Chemist's Introduction to Statistics, Theory of Error and Design
of Experiment, Roy. Inst. Chern. London, Lecture Ser. 1961, No.2, 1961.
3. Saunders, L., and Fleming, R., Mathematics and Statistics for Use in Pharmacy,
Biology and Chemistry, Pharmaceutical Press, London, 1966.
4. Calder, A. B., The British Chemist 34, 60 (1967).
5. Laitinen, H. A., Chemical Analysis, McGraw-Hill, New York, 1960.
6. Calder, A. B., Evaluation and Presentation of Spectro-Analytical Results, Hilger &
Watts, London, 1959.
7. Youden, W. J., Statistical Methods for Chemists, John Wiley & Sons, New York,
1951.
8. Calder, A. B., J. Appl. Stat. 9, 170 (1961).
9. Kaiser, H., and Specker, H., Z. Anal. Chem. 149, 46 (1955).
10. Calder, A. B., Anal. Chem. 36, 27A (1964).
11. Calder, A. B., Spectrovision No. 13, 8 (1965).
12. Mitchell, R. L., The Spectrochemical Analysis of Soils, Plants and Related Materials,
Commonwealth Bureau of Soil Science, Tech. Comm. No. 44A, England, 1964.
13. Scott, R. 0., Analyst, 66, 142 (1941).
14. Calder, A. B., Operational Statistics in the Analytical Laboratory, Lecture delivered
to Conference on Statistics in Research and Quality Control (Institute of Chemistry of
Ireland) held in Dublin, April (1967).
15. Calder, A. B., The Statistician 16, 203 (1966).
Chapter 3

CHEMICAL ANALYSIS AND SAMPLE

PREPARATION

v. C. O. Schuler
Anglo American Corporation of South Africa, Ltd.
Johannesburg, South Africa

1. INTRODUCTION

Instrumental techniques usually require careful sample preparation. In many

cases some preliminary chemical treatment has to be performed before the
instrumental measurements can be made. It is therefore essential that the
user of modern analytical instruments has a knowledge of sample-prepara-
tion techniques, as well as an insight into some of the wet chemical proce-
dures that are applicable to the analyses of geologic samples.

2. SAMPLE PREPARATION

Any analytical result can only be as accurate as the sample that was
used for the determination of its constituents. It is therefore of paramount
importance that the portion of material used is completely representative
of the bulk sample submitted for analysis. The sample received by an
analytical laboratory must be assumed to be representative of the material
from which it was taken. The analyst often has little or no control over the
collection of samples. The importance of representative sampling, especially
in the field, cannot be overemphasized (see Chap. 2, Section 11.3).
The sampling procedure actually constitutes part of the analytical
process, and as such it will affect both the precision and the accuracy of

53
54 v. c. O. Schuler

the results. Precision may be defined as the concordance of a series of

measurements of the same quantity, while accuracy expresses the correct-
ness of the measurement. Good precision, therefore, does not necessarily
imply good accuracy.

2.1. Sampling Procedures

Geochemical samples submitted for an analysis could be in the liquid,

solid, or, possibly, gaseous phase.
(a) Solutions. If a sample of well or river water, for example, is sub-
mitted to the laboratory as a clear solution, it is comparatively
easy to extract a representative portion for analysis. A suitable
portion may be taken after thorough mixing. The latter may be
achieved by inverting or shaking in closed containers. It should
be remembered that liquid samples must be stored in closed con-
tainers in order to reduce concentration changes due to evapora-
tion of the solvent.
(b) Solids. In the case of ores and rocks the samples are usually ground
in stages. Successively smaller portions are taken for further grind-
ing, until a small amount finally remains. This portion is ground
fine enough so that a representative portion may be withdrawn
after mixing. The minimum quantity of sample that may be taken
at the various stages of grinding will depend on the nature of the
material and its particle size. These minimum quantities have been
determined empirically.1 Sample crushing can be done manually
by pounding on a steel plate with a hammerlike muller. The sample
may then be ground further in a pestle and mortar. Powered
mechanical devices are, however, normally used-especially if large
numbers of samples have to be dealt with. Coarse samples are first
broken down in a "jaw crusher" which consists of two steel plates.
Large sample lumps are broken up by the hammering motion of
one plate against the other. Crushing may be followed by milling
or pulverizing in laboratory machines which may be of a disc-,
ball-, rod-, or swing-mill type.

During these processes intermittent screening is often employed. The

oversize particles which do not pass through the particular sieve are re-
ground until the whole sample passes through it. This is followed by thor-
ough mixing so that the sample may be divided into smaller portions.
3. Chemical Analysis and Sample Preparation 55

2.2. Sample Dividing

The bulk sample must be reduced in such a manner that each portion
remains representative of the original. A simple means of achieving this
without the use of special equipment is by dumping the sample on a clean,
flat surface to form an inverted cone. It is then flattened and the resulting
circular layer of sample is divided into four equal segments. Two opposite
segments are discarded and the remainder again "coned and quartered,"
until the sample bulk has been reduced sufficiently for the analyst to proceed
with his determination.
The sample-splitting procedure may be carried out more readily by
mechanical devices. One of the better known sample dividers is the "riffle"
or the "Jones splitter." It consists of a hopper, the base of which is fitted
with a series of adjoining chutes arranged in such a manner that a sample
passing through the device will be directed in two directions by alternate
chutes and collected in one of two troughs standing on either side of the
splitter.
Another sampling device consists of a disc mounted horizontally on a
rotatable spindle. The sample is fed into a hopper above the disc. As the
disc is rotated the sample is deposited onto it in the form of a ring. The
hopper is mounted on a screwlike spindle which also rotates, lifting the
hopper as sample di ~ribution on the disc proceeds. By means of an adjust-
able sector any desire \ sample volume may be removed from the stationary
disc.
The weight of th, final portion to be retained for chemical analysis
will depend on the met lOd to be employed, but a sample weighing between
50 and 100 g is generally sufficient. It is of utmost importance that no sample
contamination should occur during these operations. In order to eliminate
abrasion of the grinding surfaces of the mills they must be made of materials
that are considerably harder than the sample to be ground. For most rock
crushers special steels are used in the manufacture of the grinding surfaces.
Agate vessels should be used if contamination by the metal crushers is to
be eliminated completely. It is important that the mechanical crushing and
milling devices should be designed in such a way that they retain a minimum
amount of material after the sample has passed through. This will simplify
cleaning procedures and the risk of contaminating subsequent samples. In
addition use may be made of inert materials such as quartz, with which the
crushers may be flushed, to avoid intersample contamination.
The fineness to which a sample has to be ground may be determined
by replicate analyses of the ground material. The "spread" of the analytical
56 v. C. O. Schuler

results will give an indication of the homogeneity of the material, although

these results will also indicate errors due to limitations of the analytical
procedure itself.
Rule-of-thumb sample-preparation procedures have in most cases been
developed for each particular type of geologic material. When preparing
material for chemical analyses it is not only essential to retain a homo-
geneous sample from which representative portions may be taken, but the
material must also be of sufficient fineness. The finer the material is ground,
the more readily it will succumb to chemical attack, such as acid digestion
or fusion with the various fluxes. As a general rule, geologic material which
is amenable to acid digestion should be ground fine enough to pass through
a 200-mesh sieve. Samples which are to be decomposed by fusion with
fluxes should, on the other hand, at least pass through a lOa-mesh sieve,
provided that the material is sufficiently uniform to yield representative
samples at this particle-size level.

2.3. Homogenizing

The final portion from which amounts will be taken for the analysis
must be as homogeneous as possible, since "dip" samples are usually taken
at this stage. After grinding the sample to a suitable fineness it is intimately
mixed. This can be done manually by rolling the sample on a clean sheet
of paper, plastic, or rubberized cloth. Mechanical devices for homogenizing
samples or "blenders" are available commercially. A popular type consists
of a hollow cube rotated about an offset axis. Another device consists of
two cylinders assembled in the form of a V. These are rotated about an axis
such that the apex describes a circle in the vertical plane.

2.4. Sample Storage

The finely crushed and homogenized samples should be carefully stored
in clean, dry containers with positive closures such as screw or snap-on caps.
All containers should be immediately labelled to avoid confusion and to
facilitate identification.

3. DISSOLUTION OF GEOLOGIC SAMPLES

Once a geologic sample has been reduced to sufficient fineness, it has

to be brought into solution before a wet chemical analysis can be per-
formed. There is no universal method for obtaining a solution of the sample,
3. Chemical Analysis and Sample Preparation 57

since the procedure to be adopted will depend on the nature of the material
to be analyzed. In the following sections the more general methods of obtain-
ing solutions of solid materials are outlined. It is advisable to test the
solubility of a small portion of the sample before subjecting the entire
sample to a specific treatment. If a particular sample is soluble in water,
then full use should be made of this fact. Very few geochemical samples
are, however, water soluble excepting some of the salts such as rock salt
(sodium chloride) and nitre (potassium nitrate).

3.1. Acid Digestion

Many geologic materials are soluble in acids or in mixtures thereof.
Oxidizing or, in some cases, reducing agents may be added to effect solution.
In many cases the sample will not dissolve completely, but the constituent
to be determined may dissolve out readily. For example, acid-soluble
oxides or sulfides may occur with insoluble silicates.
The following acids are frequently used to obtain solutions of the ele-
ments or minerals indicated. If the solubility of an unknown sample is to
be determined, the various acids should be tested in the sequence outlined
below.

3.1.1. Hydrochloric Acid

Carbonate and oxide ores of Ba, Ca, Fe, and Mg are soluble in hydro-
chloric acid.
Hydrochloric acid plus reducing agent. Hydrochloric acid
plus reducing agents such as stannous chloride (SnCI 2) or hydrazine hydro-
chloride will more readily dissolve Mn, U, Cd, and Zn oxide ores, than will
hydrochloric acid by itself.
Hydrochloric acid plus oxidizing agent. Hydrochloric acid
plus oxidizing agents such as nitric acid, bromine, potassium chlorate, and
perchloric acid may be used for dissolving oxide ores such as those of Cu,
Ce, Pb, Mo, U, and Zn.

3.1.2. Nitric Acid

HN0 3 in the dilute or concentrated form may be used to dissolve ores
of Cd, Co, Cu, Pb, Mn, and Ni. For sulfide ores it is often beneficial to
make the solvent more oxidizing by the addition of bromine. With the
addition of hydrochloric acid the strongly oxidizing solvent aqua regia is
58 v. C. O. Schuler

obtained. It consists of three parts hydrochloric acid to one part nitric acid.
Aqua regia is used for sulfide ores and may be made more oxidizing by the
addition of potassium chlorate or bromine.

3.1.3. Sulfuric Acid

H 2S04 may be used for ores of AI, Be, Mn, Th, Ti, U, and Pb. In the
latter case lead sulfate is formed which is comparatively insoluble in dilute
sulfuric acid and may be separated as such after dilution. It may then be
dissolved in a solvent such as a solution of ammonium acetate.

3.1.4. Hydrofluoric Acid

HF is an excellent solvent for many silicates forming volatile silicon
tetrafluoride, which may be completely expelled from the sample solution
by heating in the presence of a high-boiling-point acid such as sulfuric acid.
Since hydrofluoric acid attacks silicates, glass vessels cannot be used, and
platinum or plastic ware such as Teflon is a requirement.

3.1.5. Perch/oric Acid

HCI03 is a powerful oxidizing agent with a relatively high boiling point.
It is an excellent solvent for certain refractory materials such as chromite.
Perchloric acid should not be heated in contact with organic matter as it is
strongly oxidizing and may react with explosive violence.
Acid digestions are in most cases the preferred means of bringing
geologic materials into solution as they introduce a minimum of salts and
cations into the final sample solution. In cases where anionic constituents
are to be determined, care must be taken that the corresponding acids are
less volatile than the solvent. For instance, if fluorspar is heated with sul-
furic acid the hydrofluoric acid formed would be volatilized and lost when
heating the solution. A subsequent fluorine determination would therefore
be invalidated.

3.1.6. Pressure Leaching

The decomposition of geologic samples, especially silicates, may be
facilitated by conducting the dissolution under pressure. For this purpose
"bombs" constructed from, or lined with material inert to the decomposing
solution are used. Teflon, which is inert to hydrofluoric and other acids, is
suitable for lining such vessels. Bernas2 has described such a Teflon-lined
pressure-reaction vessel and a procedure for decomposing silicates with
3. Chemical Analysis and Sample Preparation 59

aqua regia and hydrofluoric acid. The time required to decompose a 50-
300-mg sample portion appears to be 30-40 min. After cooling and diluting,
boric acid is added to dissolve precipitated metal fluorides.

3.2. Sample Fusions

Geologic materials that do not dissolve in acids may be brought into
solution by fusing the sample with a suitable flux.

3.2.1. Types of Fluxes

Broadly speaking, there are four types of fluxes:
Alkaline fluxes. These include sodium carbonate, potassium car-
bonate, potassium hydroxide, sodium hydroxide, and sodium peroxide.
Sodium peroxide also has oxidizing properties. The alkaline fluxes may be
used individually or ~.s mixtures. Mixtures usually have a melting point
lower than the individual components, thus facilitating lower fusion tem-
peratures.
Acid fluxes. The better known acid fluxes are potassium and sodium
bisulfate, and their dried counterparts, the pyrosulfates.
Oxidizing flu xes. These contain sodium peroxide, potassium nitrate,
or sodium nitrate, usually in admixture with alkaline components.
Neutral fluxes. Examples of neutral fluxes include sodium fluoride,
borax, and lithium fluoride.

3.2.2. Choice of Crucible

The type of crucible to be used for fusions must be chosen such that
the attack on it is negligible. The material from which the crucible is made
should not contain any of the elements for which the sample is being
analyzed.
Platinum crucibles. These may be used for fusions with potassium
and sodium carbonate or with potassium pyrosulfate. However, the samples
should not contain metals which alloy with platinum such as lead, zinc,
tin, bismuth, silver, gold, or copper. Mixtures from which these elements
are reduced to metals from their components can also not be tolerated. In
addition, sulfur, selenium, tellurium, phosphorus, arsenic, and antimony
all combine readily with platinum on heating, and they should therefore
be absent.
60 v. c. O. Schuler

Due to the high cost of platinumware, samples of unknown composi-

tion should not be fused in platinumware, but should rather first be tested
in crucibles made of some other material.
Molten sodium peroxide attacks platinum metal and should not be
used, but alkaline fusions with sodium and potassium carbonate, alone or
admixed, can safely be undertaken. Fusions with bi- and pyrosulfates of
sodium and potassium, alone or mixed with sodium fluoride or lithium
fluoride, may be done in platinumware.
Iron, nickel, silver, or zirconium crucibles. These may be used
for alkaline and oxidizing fusions. In each case the crucibles are attacked
to some extent, and a crucible must therefore be chosen that is made of a
metal not sought in the sample. The crucible material should be such that
it can easily be separated in the subsequent analytical procedure, if necessary.
Zirconium crucibles, although expensive, are the most suitable for most
alkaline fusions, since they are attacked only slightly by alkaline fluxes.
Silica crucibles. Silica crucibles are excellent for fusions with acid
fluxes, such as sodium and potassium bi- or pyrosulfates, which are com-
monly used for samples low in silicates.

3.2.3. Choice of Flux

For samples of unknown composition, pilot fusions should first be
undertaken. If the sample is thought to consist mainly of refractory oxides
with a low silica content, fusion with potassium pyrosulfate in a silica
crucible is suggested. For siliceous material an alkaline fusion with a mixture
of equimolar amounts of the carbonates of sodium and potassium in plat-
inum crucibles may be attempted. The fusion may with advantage first
be tested in a nickel or iron crucible. In this case sodium peroxide may be
included in the flux to reduce the fusion temperature.
Silica-rich samples may also be fused in platinumware with a flux
consisting of potassium pyrosulfate and lithium or sodium fluoride to attack
the silica or silicates. In this case the cooled fusion mass may be digested
with sulfuric acid in the platinum vessel and heated to expel silicon as the
tetrafluoride. The procedure is not applicable where silicon has to be
estimated.

3.2.4. Fusion Procedure

Once the type of flux to be used has been decided upon, the most
convenient type of crucible is chosen. The sample and the flux are placed
into the crucible, mixed well by stirring, and gently heated. The ratio of
3. Chemical Analysis and Sample Preparation 61

sample weight to flux weight should be in the range of 1 :10-1 :20. Once
the mass starts to melt the crucible is gently swirled over the heat source,
until a clear melt is obtained. In the case of sodium peroxide-sodium
carbonate flux fusions this will require about 5-10 min, depending on the
burner temperature. Fusions with sodium carbonate, potassium carbonate,
and potassium pyrosulfate may require as long as 30 min, even at high
temperature, for satisfactory decomposition of the sample. Once the molten
fusion becomes clear the crucible and contents are cooled and the melt is
dissolved in water or acids, depending on the particular analysis.
Melts from potassium pyrosulfate fusions are usually dissolved in
dilute acids. Alkali-fusion melts are normally first treated with water, giving
rise to two phases-a solution and a hydroxide precipitate. The whole
mixture may be acidified to obtain complete solution of the sample. The
precipitate may also be filtered or centrifuged off and examined separately.
For example if chromite ore is fused with a mixture of sodium carbonate
and sodium peroxide, and on cooling leached with water, the solution will
contain the water-soluble sodium chromate and aluminate formed during
fusion. The precipitate, on the other hand, will contain ferric hydroxide
and other material insoluble in sodium hydroxide-sodium carbonate solu-
tions, such as nickel hydroxide which will result if a nickel crucible is used
for the fusion. By filtering the alkaline solution one thus achieves a separa-
tion of the iron and chromium present in the sample, facilitating their
estimation.

4. METHODS OF SEPARATION

Once the sample to be analyzed has been obtained in solution, the

constituent to be determined must often first be separated from the other
components. These may interfere with the estimation depending on what
particular technique is to be employed.
Only an outline of the more common separations and techniques can
be given in the scope of this chapter.

4.1. Precipitation Methods

4.1.1. Group Precipitation
The classical group-precipitation scheme of qualitative chemical analysis
consists essentially of separating the cations into groups of elements, accord-
ing to the solubilities of their salts.
62 v. C. O. Schuler

The sample solutions are treated successively with reagents which form
insoluble components with certain cations. Usually each reagent forms
insoluble components with several elements or groups of elements. The
precipitated elements are then isolated by filtering or by centrifuging. They
are then redissolved and separated further by the addition of more specific
reactants. Details of such procedures may be found in texts on qualitative
inorganic analysis. 3,4
Many of these reactions find application in quantitative analyses, where
a constituent may have to be isolated before it can be determined. In such
cases it is of importance that the precipitation reaction should be quantita-
tive, and preferably rapid and specific.

4.1.2. Controlled pH Precipitations

In addition to the above scheme, a number of separations may be

achieved by the precipitation of various hydroxides at controlled pH values.
Many cations are precipitated as hydroxides, provided the alkalinity or the
pH is above a certain value. The latter is characteristic of the metal itself,
as well as the valency state in which it is present in solution.
The pH value may be controlled by the addition of solutions of am-
monia or sodium hydroxide, together with suitable buffering agents. If
the pH values at which two or more elements hydrolyze are sufficiently
different, a separation may be achieved. Either ammonium or sodium
hydroxide may be used to adjust the pH value of the solution from which
the cations are to be hydrolyzed. Both of these may, however, be used in
certain instances to achieve greater selectivity.
In the case of ammonia, some metals like copper, nickel, and cobalt,
for example, form soluble ammine complexes in the presence of an excess
of ammonium radicals. Thus, if ammonium hydroxide solution is added
to an acid solution of cupric ions, a hydroxide precipitate will appear at
a pH value of approximately 5.4. If an excess of ammonium hydroxide or
an ammonium salt is added the cupric ammine complex Cu(NH3)~+ will
be formed, and the precipitate will redissolve. This property may be put
to use in separating the elements that form soluble am mine complexes
from those which do not, such as Fe(III), aluminium, titanium, etc.
Similarly, if sodium hydroxide is used to increase the pH value of a
solution, the cations will be hydrolyzed at those particular pH values at
which precipitation of the cations occur. Advantage may frequently be
taken of the amphoteric nature of some elements, in that they form soluble
3. Chemical Analysis and Sample Preparation 63

sodium salts in the presence of excess sodium hydroxide. For example,

AP+ + 30H- -+ AI(OH)a

(precipitate)
AI(OH)a + NaOH -+ NaAI0 + 2H 0
2 2

(soluble)

Thus, metals such as AI, Be, Zn, Sn, Pb, As, Sb, V, and Mo, which are
amphoteric in nature and therefore soluble in an excess of sodium hydroxide,
may be separated from Fe, Ti, Zr, Th, Se, Y, and the rare-earth elements.
The hydroxides of the latter are insoluble in sodium hydroxide.

4.2. Electrolytic Separations

4.2.1. Electrodeposition
This electrochemical procedure as applied to chemical analysis, is
usually confined to the isolation of metals from aqueous solutions of their
salts. When the potential of a direct current applied to two inert electrodes
immersed in an electrolyte of the metal saIt solution is increased from zero
upwards, a potential will be reached where electrolysis commences. At this
stage the metal ions will migrate towards the negative electrode, to be plated
onto the cathode. This potential is known as the decomposition potential
and is characteristic of the electrolyte in question. If the electrolyte consists
of salts of more than one metal, it is possible, by controlling the potential,
to separate these metals by plating them out one at a time. Their decomposi-
tion potentials must, however, differ by at least 0.25 V if a virtually com-
plete separation is to be achieved.

4.2.2. Mercury Cathode Cell

A number of useful electrolytic separations can be achieved by means
of this apparatus. It consists essentially of a glass vessel containing a mer-
cury pool, which is made the cathode of an electrolysis system. The solution
to be electrolyzed is poured into the vessel and an inert platinum electrode
is inserted into the solution to form the anode. A direct current having a
potential higher than the metals to be separated is applied to the electrodes.
The metals will be electrodeposited into the mercury to form an amalgam.
In this way metals such as Fe, Cu, Co, Ni, Zn, Cr, Cd, Mn, Pb, Bi, and Sn
may be separated from a solution containing elements that cannot normally
be plated from the solution, such as AI, Ti, U, the alkaline earths and rare
64 v. c. O. Schuler

earths. Dilute sulfuric acid solutions are usually employed for this type of
separation. An advantage of this separation procedure is that no reagents
have to be added, leaving the electrolyzed solution free of possible conta-
minants and high salt concentrations.

4.3. Solvent Extraction

If two immiscible solvents are intimately contacted by mechanical

mixing, any substance soluble in both will distribute itself between the
two solvents in a definite proportion. This ratio is termed the distribution
coefficient for the solute between the two solvents.
In analytical chemistry this principle of solvent extraction is frequently
applied. 5 6 Conditions have to be correctly controlled to transfer certain
substances more or less quantitatively from an aqueous solution to an or-
ganic solvent. The aqueous solution is shaken together with the organic
solvent in a stoppered separatory funnel and then allowed to stand to allow
the two phases to separate out. They may then be drawn off one at a time.
Common metal salts are not normally soluble in organic solvents.
This is to be expected because of their ionic nature and the tendency for
the metal ions to be solvated. Before such metals can be extracted into
organic solvents it appears that uncharged species must first be formed.
All the water molecules coordinated to the metal ion must also be displaced.
However, these requirements make it possible to achieve selective separa-
tions by solvent extraction. By forming extractable species only of the metal
to be separated, it may be isolated by solvent extraction from the other
metals in solution, provided these remain in the ionic state.
A neutral ion may be formed by coordination of anions with the metal
cation, thus excluding solvated water. Some charged coordination com-
plexes can also associate to form neutral extractable ions. Chelating agents
may be used to form neutral ions.
For solvent extraction to be successful the operating conditions have
to be closely controlled. Sufficient complexing agent must, for instance, be
present to ensure that all the metal ions are complexed. An excess of com-
plexing agent is therefore essential, since reversible reactions are usually
involved. In some cases the valency state of the metal ion influences the
complexing reaction. For example, Fe(III) forms complexes more readily
than Fe(II). The addition of a salt containing the same anion as the species
to be extracted may effectively increase the distribution coefficient. The
extraction of uranium, for example, as the nitrate complex from a solution
of nitric acid with ether is enhanced by the addition of either ammonium
3. Chemical Analysis and Sample Preparation 65

nitrate or aluminium nitrate. This effect may be due to the higher concen-
tration of complexing ions as well as the solvation tendencies of the added
salts to help remove the water of hydration from the extractable species.
In some cases the extraction of metals can be rendered more selective
by the use of so-called masking agents. These prevent selected metals from
being extracted into the organic phase. Two types of masking agents are
generally recognized: (a) those which form charged-ion complexes and thus
prevent extraction, and (b) masking agents which form complexes that are
stronger than those which are obtained with the complexing agent itself.
Thus, aluminium can be extracted in the presence of Fe(II) as the 8-hydroxy-
quinolate by the prior addition of an alkaline cyanide. The stable ferrocya-
nide ion will be formed, which does not react with 8-hydroxyquinoline.

5. METHODS OF DETERMINATION

In wet chemical analyses the final concentration of the element of inter-

est is usually determined by gravimetric or volumetric procedures.

5.1. Gravimetric Analysis

In gravimetric analyses the element to be determined must be isolated

and weighed in its pure form, either as the element itself or as a stoichio-
metric compound. For this purpose the separation procedures already
described may be used once a solution of the sample has been obtained.
If the other constituents of the sample are known, simplified separation
procedures may often be applied. Where possible, a specific reagent should
be used that will yield an insoluble compound of only that element which
is of interest. Various techniques may be used to make a reagent specific
for certain applications. For instance, it is sometimes possible to complex
the other ions present so that they do not react with the precipitant. Many
organic reagents can be used in this way. For example, nickel reacts with
dimethylglyoxime in an ammoniacal solution to give a quantitative precipi-
tate. However, in an ammoniacal solution, iron, if present, will also precipi-
tate out. This may be prevented by forming a soluble iron citrate or tartrate
complex by the addition of these reactants to the acidic solution.
Once the precipitate has been formed quantitatively it is usually sepa-
rated by filtering or centrifuging. The precipitate is then washed, dried, and
finally weighed. In the case of organic precipitants the precipitate may be
dried at about 110C and weighed as such, provided that a stoichiometric
66 v. C. O. Schuler
precipitate is formed. Otherwise the precipitate may be gently ignited in air
to form the oxide, which is then weighed. It is, of course, essential that all
transfer operations are done without any loss of the precipitate, as this
would obviously negate the results.
In some cases neat gravimetric procedures based on electrolytic sepa-
rations may be employed, provided the decomposition potentials of the
salts of any other metals in the solution differ sufficiently from that of the
element sought. For instance, copper may be plated quantitatively from
sulfuric acid solution onto preweighed platinum electrodes. Reweighing the
electrode after the copper has been deposited onto it affords a convenient
means of estimating the amount of copper present in the sample.

5.2. Fire Assay

In fire- or dry-assaying techniques, the metal or metals sought are

separated from the other components of the ore by heating the sample.
Suitable reagents, referred to as fluxes, are usually added and the separated
metal is then weighed directly. The method is used mainly for the estimation
of gold-, silver-, and platinum-group metals. 7 8
A weighed amount of ore, say 30-150 g, is mixed with a flux consisting
usually of sodium carbonate, lead oxide (litharge), and borax or sodium
fluoride. A reducing agent such as powdered charcoal is also added. The
mixture is fused in a fire-clay crucible at red heat. The lead oxide is reduced
by the charcoal to metal droplets, which will alloy the precious metals
present in the sample. The silica, alumina, and other sample constituents
form a slag with the remaining components of the flux. The molten mass is
poured into a conical steel mold, in which the lead containing the precious
metals will settle to form a "button" at the bottom. After cooling, the mold
is emptied and the lead button is broken away from the slag. The lead button
is now placed onto a flat roasting dish or cupel which is made of bone ash,
magnesium oxide, or other porous material. With further heating under
oxidizing conditions, the lead is oxidized and partly volatilized. It is also
partly absorbed by the cupel, leaving a bead of precious metal which can
then be weighed after cooling. If more than one precious metal is present
in the bead, chemical separations are employed to separate the metals by
selective dissolution.
The fire-assay method is exceedingly sensitive, and by weighing the
separated precious metal on a suitable microbalance, ores containing less
than 1 ppm of gold, say, may be evaluated quantitatively.
3. Chemical Analysis and Sample Preparation 67

5.3. Volumetric Analysis

Volumetric or titrimetric analyses consist of the determination of that

volume of a standard solution, or reagent of known concentration, which
is required to react quantitatively with the substance to be determined. In
titrimetric analysis the reactions must go to completion rapidly and stoichio-
metrically in dilute solutions. At the equivalence point or endpoint, there
must be a marked change in some detectable property of the test solution
so that the completeness of the reaction may be easily recognized. Broadly,
titrimetric procedures9 fall into the following categories:
(a) Acid-base reactions, which consist of the combination of hydrogen
and hydroxyl ions to form water.
(b) Oxidation-reduction reactions involve a valency change of the
reactants.
(c) Precipitation reactions are accompanied by the formation of an
insoluble precipitate.
(d) Complex-formation reactions cause a negligibly dissociated ion
or compound to be formed. The equivalence point of the reaction
may be determined either visually or by instrumental aids.

5.3.1. Visual Indicators

In certain cases where a highly colored solution constitutes one of the

reactants, the sudden disappearance of its color may indicate the equivalence
point. If the color of the reactant which is added is discharged by the re-
action, then the equivalence point will be at that value, beyond which
further reagent additions begin to color the test solution. In such cases a
correction will have to be made for the slight reagent excess that is required
to produce the color. A well-known example of this type of titration is the
reaction of reduced ions in acid solution, with a standard potassium per-
manganate solution. A large number of organic substances is available which
will intensify and therefore assist the detection of endpoints.
Acid-base indicators have the characteristic of changing color at dif-
ferent pH values or hydrogen ion concentrations. This color change may
either be associated with ionized and unionized forms of the substance or
it may be due to inherent structural changes which are pH sensitive.
Reduction-oxidation indicators or "redox" indicators are organic sub-
stances, such as dyestuffs, which change color at definite oxidation-reduc-
tion potentials. These may therefore be used to follow titrations which
68 v. C. O. Schuler

involve a change in oxidation potential. Some substances may be used as

indicators because they form a colored compound with a metal ion, provided
the latter is in a certain state of oxidation. Thiocyanate ions, for example,
react with ferric, but not with ferrous, ions to form a blood-red compound.
This phenomenon may be utilized for the determination of titanium. Ferric
ions are added to a solution of titanium initially present in the reduced
state. In the presence of thiocyanate, the endpoint will be indicated by a
pronounced red color, which will result from the presence of excess ferric
ions.
Several visual methods are available for the determination of endpoints
in precipitation and complex-formation reactions. They lend themselves,
however, to specific rather than general applications. For example, in the
titration of chloride ions with silver ions, sodium chromate may be used
as an indicator. When all the chloride ions have reacted with the silver
which is being added as titrant, any excess silver ions will react to form a
red silver chromate precipitate, thus indicating the endpoint.
Metal ion indicators are used extensively for the complexometric titra-
tions of metals.lO These indicators are dyestuffs which react with metal
ions to form strongly colored complexes. The indicator must form a fairly
stable colored compound with the metal to be titrated. The complex which
the metal forms with the titrant must, however, be more stable than the
indicator complex. After the addition of the indicator, a highly colored
complex will be formed. During the titration, the titrant will react with the
metal to form a more stable complex. At the endpoint the highly colored
metal-indicator complex will be destroyed due to the removal of the metal
by the titrant. At this point the color will suddenly disappear, indicating
that the equivalence point has been reached.

5.3.2. Instrumental Indicators

Most of the conventional volumetric methods rely on visual endpoint
detection, but instrumental procedures have by now also been well esta-
blished. Acid-base titrations are easily followed potentiometrically by taking
successive pH measurements. Oxidation-reduction titrations may also be
followed potentiometrically with the use of suitable electrodes. Titrations
that are accompanied by changes in the conductivity of the solution may
have their equivalence point determined by conductance measurements.
Precipitation reactions are an example of this application. An alternating
current may be used to reduce the polarization and other electrolytic effects
at the electrodes.
3. Chemical Analysis and Sample Preparation 69

6. ADVANTAGES, DISADVANTAGES, AND

LIMITATIONS OF WET CHEMICAL ANALYSES

6.1. General Considerations

The development of modern instrumentation has made the rapid,
virtually complete instrumental analysis of large numbers of samples pos-
sible. In most cases the conventional wet chemical procedures are more time
consuming. Further, except for the simple routine analyses, experienced
analysts are required to conduct delicate chemical manipulations. However,
many instrumental methods require some chemical treatment before the
samples can be presented for instrumental measurement. For example,
many instrumental techniques require the sample to be in the form of a
solution, and in some cases interfering elements must first be removed. It
is thus to the operator's advantage to have a good knowledge of the prin-
ciples of wet chemical analysis.
The comparatively modest capital outlay to equip a wet-chemical-
analysis laboratory may be considered an advantage, although this may only
be significant if small numbers of samples have to be analyzed.
Wet chemical methods are, in general, less suited for the estimation
of trace constituents than are many instrumental procedures, although
these may have only specific applications. For instance, where concentra-
tions of less than 10 ppm of a particular element have to be determined,
very few gravimetric procedures are satisfactory. A well-known exception
to this is the fire-assay procedure for precious metals which, however, is a
dry-assay technique.
A basic advantage of the so-called classical methods is that they offer
a direct means of analysis. The element sought may be isolated as the
element, or as one of its compounds, and subsequently weighed as such.
Alternatively, the concentration of a particular element may be deduced
from some well-established stoichiometric relationship. For these reasons
wet chemical methods are still used as reference methods in cases where
different instrumental techniques render different results. The major role
which wet chemical procedures play in the field of standardization warrants
a more detailed discussion.

6.2. Standardization
Most instrumental methods are comparative techniques and inde-
pendent standardization procedures are necessary. In many cases instru-
ments can be calibrated with so-called pure substances, such as reagent
70 v. C. O. Schuler

grade chemicals or their highly refined counterparts. With many instru-

mental procedures, especially if no prior chemical treatment is included,
the presence of elements or compounds other than those sought may have
significant effects on the results obtained. To compensate for such matrix
effects the instruments are calibrated with standard samples which are
similar in composition to those to be analyzed. Reference samples are
carefully prepared and analyzed for the constituents for which they will
serve. These analyses are usually performed by a number .of different
procedures. Preferably several analysts from different laboratories should
be involved. Wet chemical methods are used extensively for this purpose,
and there are no indications that they will, in the foreseeable future, be
replaced by other procedures.
A number of commercial firms and bureaus market such standard,
reference, or comparative samples, and for many applications such samples
are adequate. For the analysis of geologic samples it is, however, often
difficult to obtain samples similar to those to be analyzed. It then becomes
necessary for the laboratory to prepare and analyze its own reference
samples. A series of standards covering the whole range of concentrations
likely to be encountered should, if possible, be prepared so that they can
be used for the compilation of instrument calibration curves. Once these
curves have been established it is often only necessary to see whether two
standards yield results that still lie on the calibration curve. This procedure
will verify that no shift in instrument calibration has occurred since the
original calibration.
The analysis of such reference samples is usually carried out by wet
chemical procedures, where the criterion is accuracy rather than speed.
Once all the analytical results have been collected they should be analyzed
statistically to determine the validity of the results (see Chap. 2), and, if
possible, any analytical bias should be removed. Results that suggest an
analytical bias are obviously erroneous and should be discarded. The
weighted average of all acceptable values should be calculated and used to
allocate a standard value to the standard sample.

REFERENCES

1. Taggart, A. F., Handbook of Mineral Dressing, John Wiley & Sons, New York, 1953.
2. Bernas, B., Anal. Chem. 40, 1682 (1968).
3. Vogel, A. I., A Textbook of Macro and Semimicro Qualitative Inorganic Analysis,
Longmans, Green and Co., London, 1954.
4. Feigl, F., Qualitative Analysis by Spot Tests, Elsevier, New York, 1947.
3. Chemical Analysis and Sample Preparation 71

5. Morrison, G. H., and Freiser, H., Solvent Extraction in Analytical Chemistry, John
Wiley & Sons, New York, 1957.
6. Stary, J., The Solvent Extraction of Metal Chelates, H. Irving, ed., Pergamon Press,
London, 1964.
7. Beamish, F., The Analytical Chemistry of the Noble Metals, Pergamon Press, Oxford,
1966.
8. Dillon, V. S., Assay Practice on the Witwatersrand, Cape Times, Cape Town, 1955.
9. Vogel, A. I., A Textbook of Quantitative Inorganic Analysis, 3rd ed., Longman,
London, 1962.
10. Schwarzenbach, G., Die Komplexometrische Titration, Ferdinand Enke, Stuttgart,
1957.
Chapter 4

ION-EXCHANGE CHROMATOGRAPHY

H. F. Walton
University of Colorado
Boulder, Colorado

1. INTRODUCTION

Ion-exchange chromatography is a versatile method of separation and con-

centration which is applicable to many kinds of analysis, inorganic, organic,
and biochemical. The experimental technique is simple and in most cases
fairly rapid. In the great majority of inorganic applications, all that is needed
is a tube about 25 cm long and 1 cm i.d., constricted at one end, with a
glass wool plug or sintered disc to support a column of exchanger some
15 cm high. A typical ion-exchange column used for analytical purposes is
shown in Fig. 1. Obviously, the column can be scaled up or down according
to sample size. A common mistake is to use a column that is too large.
Gravity flow is generally sufficient with a convenient linear flow rate of
1-2 cm/min. There are a few simple details that should be noted for efficient
column operation. One of these is that the ion-exchanging material should
be packed as uniformly as possible, and that the column should not be
allowed to drain or to accumulate air bubbles during use. The column
should always be kept filled with liquid. If the granules of exchanger are
sufficiently fine (100-200 mesh or finer) this presents no problem, as surface
tension prevents the level of water from dropping more than a couple of
millimeters below the top of the bed. (The solvent may, of course, not be
water, as mentioned in Section 5.)

73
74 H. F. Walton

50ML FLASK

IO-12MM
INTERNAL
DIAMETER

COARSE
SINTERED
DISC

OUTLET TUBE
(OPTIONALl STOPCOCK

Fig. I. Ion-exchange column for analytical use.

2. ION-EXCHANGING MATERIALS

There are many kinds of solid ion-exchanging materials, organic and

inorganic. All are insoluble polyelectrolytes, that is, they consist chemically
of a continuous matrix of very high molecular weight which carries fixed
ionic groups. The charge on these groups may be negative or positive,
depending on their chemical nature. Associated with these, and having
opposite charges, are the small mobile "counterions" which can change
places with other small ions of similar charge in the surrounding solution.
One more condition is necessary. The structure of the matrix must be such
that it permits easy movement of counterions in and out.

2.1. Ion-Exchange Resins

Ion-exchange resins are used in at least fo of the analytical applications,
and the ion-exchange resins most commonly used have a matrix of cross-
linked polystyrene. Styrene is mixed with divinylbenzene in liquid form and
polymerized in droplets to give solid spherical beads whose size can be
controlled. The proportion of divinylbenzene determines the crosslinking
or tightness of the matrix. High crosslinking means a high density of ionic
charges, but at the same time it means slow diffusion and reduced efficiency
in column operation. For most analytical purposes a resin of 8% cross-
linking is preferred, that is to say, a resin prepared from 8% divinylbenzene
and 92% styrene.
4. Ion-Exchange Chromatography 75

Ionic groups are introduced into the polymer beads by appropriate

chemical means. Treatment with fuming sulfuric acid introduces sulfonic
acid groups, -S03-H+, giving a cation-exchange resin. By a more com-
plicated reaction the groups -CH2N(CH3)3+Cl- may be introduced, thus
forming an anion-exchange resin. These two types of resin are by far the
most common. Another functional group that may be introduced into
polystyrene is the iminodiacetate group, -CH2N(CH 2COOH)2. This forms
chelated compounds with metal ions and binds doubly and triply charged
ions very strongly, giving interesting selectivity sequences.!
Resins containing the carboxyl group -COOH are also available. The
polymer matrix in these resins is not polystyrene, but polyacrylic acid. A
great many polymers and condensation products have been made with
special functional groups, and a few are made commercially, such as a
phosphonic resin and one containing guanidine groups. The latter is highly
selective for the platinum metals and gold. 2 An anion-exchange resin se-
lective for borate has sorbitol molecules grafted on to a weak-base derivative
of polystyrene. 3 This resin is also available commercially.

2.2. Inorganic Exchangers

Inorganic exchangers of many types have been prepared. The first
ion-exchanging materials ever to be made artificially were aluminosilicates.
These are little used today because of their sensitivity to acids and alkalis,
but the crystalline aluminosilicates called "molecular sieves" may find some
use in ion-exchange chromatography. More important are the amorphous
or microcrystalline zirconium phosphate, zirconium tungstate, molybdate,
and related compounds. Ammonium molybdophosphate (AMP) is a repre-
sentative of the heteropoly acid group. Other types are silver ferrocyanide 4
and potassium hexacyanocobalt (II) ferrate (11).5 All these compounds are
cation exchangers. Hydrous zirconium oxide is a cation exchanger in alka-
line solution and an anion exchanger in acid solution.
The distinctive feature of inorganic exchangers is their very high se-
lectivity for certain ions. Everyone of the cation exchangers mentioned is
selective for cesium, and absorbs Cs+ much more strongly than Rb+ or K+.
They are also selective for radium over barium, and permit very effective
chromatographic separations of the heavier alkali metal ions and alkaline
earth metal ions. These cation exchangers also allow separation of the
alkaline earth ions as a group from the alkali metal ions. The great impetus
to their development has been the need to separate cesium from radioactive
wastes. Cation exchangers have been used to concentrate traces of cesium
76 H. F. Walton

in geochemical analysis. The basis of this extraordinary selectivity is not

well understood, and it is worth noting that the selectivity is greatest at low
pH values (where the exchange capacity is low). The selectivity is much
less in well-crystallized materials than in amorphous or microcrystalline
solids. 6 Zirconium oxide as an anion exchanger is particularly selective for
fluoride, phosphate, and polyvalent anions in general.

2.3. Ion Exchangers Based on Cellulose

Ion exchangers based on cellulose are useful in biochemistry, where
large ions and molecules must be absorbed, but they have little application
to geochemistry. Phosphorylated cellulose has, however, been used to absorb
traces of uranium from natural waters. 7 Carboxymethyl-cellulose combined
with dithizone absorbs the same metals that dithizone does, and takes up
zinc, lead, copper, and silver from sea water.s

3. ION-EXCHANGE SELECTIVITY

3.1. Inherent Selectivity

We have noted that certain ion-exchanging materials show specially
strong absorption for certain ions. In general, however, this is not the case.
Ion exchange is a competitive process, with one kind of ion displacing
another to give an equilibrium which, to a first approximation, follows the
law of mass action. Thus, the exchange of sodium and potassium ions be-
tween an aqueous solution and a cation-exchange resin, expressed as
K+ + NaRes:;;;:=: K Res + Na+
may be represented fairly accurately by an equilibrium constant:
EK _ [KRes][Na+]
(1)
Na - [NaRes][K+]

where the brackets indicate molar concentrations. For exchanges of ions

of unequal charge, for example, the sodium-calcium exchange, the con-
centrations must be raised to the appropriate power:
E Ca _ [CaRes][Na+]2
(2)
Na - [NaRes]2[Ca2+]

It will readily be seen that in the example given in Eq. (2), dilution of the
solution will force the ion of higher charge (Ca2+) into the exchanger. This
4. Ion-Exchange Chromatography 77

effect is put to good use in chromatographic separations. It should also be

noted that the quotients E are not true constants, because neither the ex-
changer nor the solution behaves ideally.
In a sulfonated polystyrene cation-exchange resin with 8% crosslinking
the values of E for the alkali metals increase by a factor of 4 as one goes
from Li+ to Cs+. For the Li+ to Ag+ exchange Eis about 10. The differences
in selectivity are thus not great. In anion exchange the range of E values is
larger. In going from F- to 1-, the E values increase by a factor of nearly
100. The effect is great enough that simple, practical procedures have been
worked out to separate fluoride from the products of alkali fusion of rocks
by retaining all other ions on a anion-exchange resin. 9 10 For the separation
of metal ions, however, the selectivity of the usual cation-exchange resins
is not enough, and it is essential to supplement it by another selective
process, the formation of complex ions in solution, as outlined in Section 3.2.
Before discussing this topic, however, we should note the "nonchro-
matographic" uses of ion-exchange columns, the most common of which
is the absorption of all cations in a solution and their replacement by an
equivalent amount of hydrogen ions. This is done by passing the solution
through a column of strongly acidic cation-exchange resin, such as sul-
fonated polystyrene, in its hydrogen form. The hydrogen ions that are
released may be titrated to find the total salt concentration of the original
solution. This is a common technique in water analysis, and is far simpler
and more informative than the old method of evaporating and weighing
the residue. Another use for the complete exchange of dissolved cations for
hydrogen ions is to permit the accurate determination of the anions that
pass through the column and emerge as their corresponding acids. Well-
known and established methods exist for the determination of sulfur in iron
pyrites by oxidation to sulfuric acid, and for phosphate in phosphate rock
by conversion to phosphoric acid. In each case the solution is freed from
interfering metal ions by passing through a cation-exchange resin; then the
acid of interest is determined by titration. Recently, this same procedure
has been used in conjunction with activation analysis to determine phos-
phorus in silicate rocks.u
Though separations like these are not properly "chromatographic,"
the question of selectivity still exists, for it is important to keep the acidity
of the influent solutions as low as possible. Ion exchange is a competitive
process, and if the initial concentration of hydrogen ions is too great, it
will prevent complete absorption of the metal ions by the column of cation-
exchange resin. Generally, an unnecessary excess of acid or salt should be
avoided in the preparation of samples for ion-exchange processing.
78 H. F. Walton

3.2. Complex Ion Formation

Complex ion formation is very widely used for separating metal ions
from each other. The classic case is the separation of the rare earths or
lanthanides by chromatography on columns of cation-exchange resin, using
as the eluent a solution of ammonium citrate of controlled pH. The element
forming the most stable citrate complex (lutecium) is eluted first, while the
one forming the least stable complex (lanthanum) comes out last. Selectivity
of the resin itself has little or nothing to do with this separation. The stabi-
lities of the dissolved complexes control the separation, and even though
the differences in these stabilities are small, effective separations are possible
by using columns that are sufficiently long and have sufficiently high "plate
numbers" The plate number is an experimental parameter which expresses
the resolving power of a chromatographic column or other multistage
separator. Recent publications on the ion-exchange separation of lanthani-
des cite the use of a-hydroxyisobutyric acid, rather than citric acid, and
quite rapid separations for techniques such as activation analysis have been
made with this reagent. 12
Magnesium, calcium, strontium, and barium ions are eluted in this
order from a cation-exchange resin by an ammonium lactate solution of
continuously increasing pH. This technique is called "gradient elution" and
the most stable complex is eluted first. The concentrations in the effluent
have been measured by flame photometryP This carefully executed research
illustrates how rapidly analytical processes become obsolete. Soon after it
was published, the atomic absorption technique came into general use.
Interelement effects are much less important in atomic absorption spectros-
copy than in flame emission, and the need for separating these ions has
now become much less. It is nevertheless quite easy to separate barium
from the other alkaline earths by cation exchange, since barium is held
very tightly by sulfonic acid resins.
Citrate, lactate, a-hydroxyisobutyrate, and EDTA complexes tend to
be more stable, especially for metal ions of high charge. Their complexes
are usually uncharged or negatively charged, and therefore not bound by a
cation-exchange resin. Metals can thus be separated on the basis of ionic
charge. A good illustration is the separation of radioactive Y-90 (charge
of +3) from Sr-90 (charge of +2) on a cation-exchange resin column.
Yttrium is eluted by citrate or EDTA while strontium remains behind. The
gamma emission ofY-90 is easily counted and serves to measure the amount
of the parent Sr-90.14
Chloride, sulfate, and nitrate ions form neutral or negatively charged
4. Ion-Exchange Chromatography 79

complexes with nearly all metals if the concentrations are high enough.
These complexes may therefore be used to pull metal ions off a cation-
exchange resin, one at a time, giving elegant separations. 1s This same effect,
the formation of negatively charged complex ions, will put metal ions on
to an anion-exchange resin. The absorption of metal ions from hydrochloric
acid solutions by strong-base anion-exchange resins is so important that it
merits a separate discussion.

4. ANION-EXCHANGE SEPARATIONS OF METALS

A dilute solution of Co (II) chloride is pink, due to the hydrated cation
Co(OH2)~+' If hydrochloric acid is added to this solution to make it 4M
or more, in acid the color changes to blue, indicating the presence of the
complex anion CoC1~-. If some strong-base anion-exchange resin is mixed
with the blue solution, the resin will acquire a blue color as it absorbs this
complex anion. The absorption increases with increasing hydrochloric acid
concentration up to a maximum value of about 100 ml/g, that is, the weight
of cobalt in I ml of solution is about 100 times that in 1 g of resin. This
maximum absorption is reached at roughly 9M HCI. At higher concentra-
tions the distribution ratio D drops slightly, as shown in Fig. 2.
Each element has its characteristic graph of distribution ratio against
hydrochloric acid concentration. Graphs for most of the metals in the

6~------~--------~------~

Fe(ml

o
0>
o
...J

[Hel] , M
Fig. 2. Adsorption of metals from hydrochloric
acid solutions. D is the distribution ratio
in ml/g.
80 H. F. Walton

Table I
Anion Exchange of Metals in Hydrochloric Acid a

Element Oxidation state [HCI]nax, M log D.nax

Ag I <1 3
As III 10
Au III <1 7
Hi III <I 4.5
Cd lib 2 3.5
Co II 9 1.7
Cr III sl. ads.
VIC str. ads.
Cu I <2 2
II 4 2
Fe II 12
III 10 4.5
Ga III 7 5
Ge IV 12 2
Hf IV 12 6
Hg lIe <1 5
In III 3
Ir III <1 2
IV <1 4
Mn II 11 0.3
Mo VI 4 2.5
Nb va 8 3
Ni II no ads.
Os III <1 4
Pb II 1 1.5
Pd I; <1 3
Po IV str. ads.
Pt IV <1 3.5
Rh III <1 1.5
IV str. ads.
Ru IV <2 3
Sb III 2 3
VO 10 5.5
Sc III sl. ads.
Se Iva >6 >1
Sn II <1 3
IV 6 4
4. Ion-Exchange Chromatography 81

Table I (Continued)

Element Oxidation state [HCI]max, M log Dmax

Ta va 12 2.5
Tc VII 4 2.5
Te IV str. ads.
Th IV no ads.
Ti III sl. ads.
IV 12 1.2
TI III <I 1.5
U IV 12 2.5
VI 12 3
V IV sl. ads.
V 12 3
W VI 9 1.5
Y III no ads.
Zn lib 2 3.2
Zr IV 12 3

a From L. Meites, Handbook of Analytical Chemistry, McGraw-Hill, New York, 1963. Used
by permission.
b Values of D for Cd and Zn differ significantly below 2M; in O.IM HCI, log D ~ 1.0 for Zn,
2.2 for Cd.
C Cr(VI), Hg(II), and Sb(V) attack the resin.

a Values for Nb(V), Se(IV), and Ta(V) are erratic because of hydrolysis.

periodic table are given in the original papers of K. A. Kraus,I6 and have
been reproduced in numerous publications. Representative curves are shown
in Fig. 2, and Table I summarizes the data. The distribution ratio, measured
by shaking the resin and the solution together until equilibrium is reached,
is related directly to the elution volume in a chromatographic column. It
is therefore easy to devise an almost unlimited number of separation schemes.
The separation of Fe(III), cobalt, nickel, and zinc will serve as an example:

Make sample solution 9M in HCI; place on anion-exchange resin

column previously washed with 9M HCl.
Pass more 9M HCI- Ni is eluted.
Pass 4M HCI - Co is eluted.
Pass 1M HCl- Fe is eluted.
Pass O.IM HCI or water - Zn is eluted.
82 H. F. Walton

The analytical literature is full of separations of this kind, and applications

to geochemistry are described in Section 7. Anion exchange in hydrochloric
acid is a convenient way to separate minor elements as a group from large
amounts of common elements, for the alkalis, alkaline earth metals, and
aluminum are not absorbed.
Hydrochloric acid is a convenient reagent for such separations, for it
can be obtained very pure with minimum trace element contamination.
Excess acid can easily be removed from the separated solutions by evapora-
tion. As a practical caution it should be noted that it is imperative for the
anion-exchange resin to be pure. Most commercially purchased resins con-
tain not only metallic impurities, but organic impurities and bypro ducts
from the manufacturing processes, that often ruin the separations. Resins
must be carefully washed before use. The experimenter may save himself
a great deal of trouble by buying specially purified grades of resin.
What can be done with hydrochloric acid can also be done with other
acids and salts, whose anions complex metal ions. Studies have been made
of anion-exchange separations in hydrobromic acid,I7 hydrofluoric acid,ls
nitric acid,19 sulfuric acid,20 and thiocyanate solutions. 21 Separations in
hydrofluoric acid are of interest, because the ions of titanium, zirconium,
hafnium, niobium, and tantalum hydrolyze strongly in water and are best
manipulated in fluoride solutions. Anion-exchange chromatography in hy-
drofluoric acid is perhaps the best way to separate niobium and tantalum,22
and is used for this purpose in geochemical analysis. 23 Zirconium and haf-
nium are jointly isolated by anion exchange in hydrofluoric acid. They
may be separated from each other by solvent extraction with thenoyitri-
fluoroacetone. 24
The strength of absorption of a metal complex by an anion-exchange
resin is determined by two factors that are independent of each other. The
first is the stability of the negatively charged complex in solution, which
can be measured by the usual techniques of coordination chemistry. The
second factor is the strength of binding of the complex by the resin. The
complex ion FeCI4- has a very low stability in water. There is, in fact,
some doubt whether it is formed at all, except in very concentrated hydro-
chloric acid. It is, however, absorbed very strongly indeed by the resin.
Singly charged complexes, such as FeCI4-, AuCI4-, and GaCI4-, appear to
be especially strongly absorbed. Theoretical grounds have been advanced
to explain this behavior.
4. Ion-Exchange Chromatography 83

5. MIXED SOLVENTS IN ION-EXCHANGE

SEPARATIONS

Ion-exchange selectivity can be modified radically by substituting

another solvent for water. This was shown by Fritz 25 ,26 who found that zinc
and cadmium ions could be stripped from a cation-exchange resin by hydro-
chloric acid much more effectively if ethyl alcohol or acetone were added to
the water used as solvent. Lowering the dielectric constant of the solvent
evidently stabilizes the complexes by increasing the electrostatic attractions
between the metal ions and Cl-. Absorption of many metal ions by anion-
exchange resins is increased for the same reason.27 In some cases, however,
the nonaqueous solvent decreases the absorption. Fe(1I1) is not absorbed
at all by an anion-exchange resin from 1M HCl in 80-90% acetone. 28 The
reason for this apparently anomalous behavior is that the low dielectric
constant stabilizes not only the complex ion FeCl4 -, but also the ion pair
H+FeCI 4 -. The latter, being uncharged, is not absorbed by the resin. The
iron-cobalt-nickel separation cited in Section 4 may now be modified as
follows: 29

To sample solution in 6M aq. HCl, add 4 vols. of acetone. Place

on anion-exchange resin column previously washed by 1 vol. 6M aq.
HCl: 4 vols. acetone. Wash with more of this solvent-Fe is eluted.
Pass 1 vol. 6M aq. HCl: 4 vols. acetone: 1 vol. water - Ni is eluted.
Pass 1M aq. HCl or water - Co is eluted.

The proportions of water and acetone are not critical, and in practice one
simply adds more water to get the next element out. This procedure has
certain advantages over that with aqueous hydrochloric acid. Firstly, iron
is in large excess in most analytical samples. As a rule small amounts of
cobalt and nickel must be determined in the presence of large amounts of
iron. If aqueous acetone is used, a small column will suffice to concentrate
the minor constituents. Secondly, the hydrochloric acid concentrations are
much lower, and thirdly, the viscosity of the solvent is lower and therefore
allows faster flow rates than with water.
This separation will also occur on a cation-exchange resin column. 29
Again, iron, which forms an uncharged species, is not absorbed from 80%
acetone or 80% tetrahydrofuran-20% 3M HCl. Cobalt is eluted before
nickel in this case, as it forms a stronger anionic complex. Aluminum, if
present, is absorbed by the cation exchanger, since its chloride complex
is very weak. 30 Mixed-solvent ion-exchange chromatography has been ex-
84 H. F. Walton

plored thoroughly by Korkisch and coworkers. They call their method

"combined ion exchange and solvent extraction. 29 " Fe(lII) is one of re-
latively few metals that forms uncharged ion pairs under these conditions.
Gallium, indium, and bismuth form stable uncharged ion pairs with chloride
ions and can be separated from aluminum and from divalent ions. 29 ,31
Another metal that forms uncharged ion pairs is U(VI). It forms a nitrate
complex, apparently SH+U02 (N0 3)3-' where S is an oxygen-containing
solvent, tetrahydrofuran or (less effectively) acetone. Thus, other elements
are absorbed on a resin column from 90% tetrahydrofuran-lO% 6M HN03
while uranium passes through.32 In sulfate or chloride solutions, however,
U(VI) forms an anionic species which is strongly absorbed. Traces of uran-
ium are recovered from sea water and marine sediments by making a
solution in 90% methyl glycol-lO% 6M Hel, and passing this through an
anion-exchange resin. Uranium is strongly and selectively absorbed. After
washing out accompanying elements with 80% methyl glycol-20% 6M Hel,
the uranium is eluted with 1M aqueous Hel and determined fiuorimetrically.
A cursory review of the available literature reveals that the mixed-
solvent ion-exchange technique has not been adequately exploited in geo-
chemical analysis.

6. SPECIAL ION-EXCHANGE TECHNIQUES

In this section a number of special ways of using ion-exchanging

materials in geochemical analysis are reviewed.

6.1. Ligand Exchange

This is primarily a technique for separating organic compounds, but a
geochemical application that has been explored is the concentration of
amino acids from sea water. 34 The absorbent is a chelating resin loaded
with cupric ions. In such an absorbent the cupric ion is coordinated with a
nitrogen atom and two oxygen atoms in the functional group of the exchang-
er, and its fourth coordination site is free to accept a water molecule, an
ammonia molecule, or another electron-donating ligand. 35 When sea water
is passed through a column of this resin the amino acids become absorbed,
displacing the coordinated water molecules. They are later displaced
from the column in concentrated form by passing aqueous ammonia.
Ligand-exchange chromatography is used primarily for compounds
having basic nitrogen atoms, but it can also separate amino acids and
polyhydroxy compounds.
4. Ion-Exchange Chromatography 85

6.2. Isotopic Ion Exchange

In activation analysis it often happens that the actlVlty of a minor

component is masked by an enormously greater activity of a major com-
ponent or matrix element. To measure the minor component it is not
always necessary to remove the matrix element. It is sufficient to replace
the radioactive matrix by a nonradioactive isotope of the same element.
This may be achieved by a simple ion-exchange procedure. 36 For example,
large amounts of radioactive Ba-133 are separated from trace amounts of
radioactive sodium, silver, zinc, and other elements by passing a solution
of the chlorides in O.IM HCl through a column of sulfonated polystyrene
cation-exchange resin, saturated with inactive barium ions. This resin has
a high affinity for barium ions, and other divalent ions are only slightly
absorbed and retarded as the solution flows through the column. Radio-
active barium ions, however, exchange easily with isotopic nonradioactive
barium ions in the resin and are retained. The column is rinsed with a
small volume of dilute, inactive, barium chloride solution, to wash out the
weakly absorbed ions of the minor constituents.
A variation of this method is used to separate strongly absorbed minor
constituents from a weakly absorbed matrix element. Again a column of
cation-exchange resin is used which is loaded with strongly absorbed cations,
for example, Ba2+. The solution containing the weakly absorbed matrix
ions-say Na-22-and strongly absorbed trace activities is passed. The
matrix passes on, and is washed out with, say, inactive NaCl solution, while
nearly all the trace elements are retained. These may be counted on the
column or eluted by appropriate means.
"Isotopic ion exchange" has been adapted to thin-layer chromato-
graphy. A mixture of Sr-90 and Y-90 is placed on a thin-layer plate coated
with strontium sulfate and developed with dilute sulfuric acid. The yttrium
moves ahead while the radioactive strontium exchanges with the absorbent
and stays behind. 37

6.3. Precipitation Ion Exchange

This is another method for concentrating the trace elements present

in a matrix of another element. 38 The salt mixture to be analyzed is dissolved
in a minimum amount of water and absorbed in a small excess of cation-
exchange resin. This is rinsed with a very little water, followed by dioxane.
The effluent, which contains some of the trace elements, is reserved. The
resin is now dried, and the resulting mixture of resin and precipitated salt
86 H. F. Walton

placed on top of a new column of resin previously washed with 70% diox-
ane-30% 12M HCI. The whole column is now washed with more of this
solvent. A small volume of wash liquid suffices to bring out nearly all the
trace elements, leaving most of the matrix behind. The method works for
elements whose chlorides are sparingly soluble in 12M HCl, that is, Na, K,
Ba, Sr, and also Ca, Mg, Li. By modifying the solvent other matrix elements
can be precipitated, including Ni, Cr, Mn, Pb, and AI.39

6.4. Difference Chromatography

This is a method for accurately measuring small differences in compo-

sition among very similar solutions. It was designed for studying variations
in the composition of sea water. 40 A "standard" sample of sea water was
passed through a long column of cation-exchange resin until equilibrium
was established between the resin and the standard solution. Then a quantity
of sea water of slightly different ionic composition was injected into the top
of the column. Each ionic constituent traveled down the column at its own
rate. This depends on the slope of its absorption isotherm, giving concen-
tration steps or fronts which may be detected by appropriate means at the
column outlet. Several very sensitive differential techniques are now avail-
able for measuring such fronts. They include potentiometric, conducto-
metric, refractometric, and thermometric (heat of absorption) techniques.
The latter has been developed since the paper of Mangelsdorf was published.
Possibly "difference chromatography" will find applications to industry as
well as oceanography.

6.5. Ion-Exchange Papers

Papers are now made from cellulose fibers and finely ground ion-
exchange resins in about equal proportions by weight. Four types of resin
are used: strong acid (sulfonic resin), weak acid (carboxylic resin), strong
base (quaternary ammonium resin), and weak base (polyamine resin with
secondary and tertiary amine groups). Another type of ion-exchanging
paper has the ionic functional groups attached to the cellulose itself. They
have been used analytically in two ways-for paper chromatography of
inorganic and organic ions, and as a "filter" to collect strongly absorbed
ions from a large volume of solution in preparation for their measurement
by such means as x-ray fluorescence or radioactivity.
Significant literature exists describing the use of resin-impregnated
papers for the chromatography of inorganic ions. 41 ,42 Broadly, the paper-
4. Ion-Exchange Chromatography 87

chromatography separations parallel the column separations discussed ear-

lier. In paper chromatography much smaller amounts of material are
handled than in columns, but the former are more rapid and give sharper
separations. The "number of theoretical plates" in a 20-cm strip of resin-
impregnated paper is much more than in a column 20 or 30 cm long. In the
references cited, Sherma shows that aqueous-nonaqueous solvent mixtures
can be used with both cation- and anion-exchanging papers to give selective
migration of certain ions. For example, using a cation-exchange resin paper
and 0.6N nitric acid in 90% tetrahydrofuran, gold moved with the solvent
front, U(VI) migrated at an appreciable rate (RF = 0.13), while other ele-
ments stayed close to the origin. Tests such as this will surely find use in
geochemical analysis.
One important practical detail must be noted by potential users of
this technique in the laboratory.43.44 The solvent is allowed to move a little
way up the paper strip, and then the sample spot is placed 1-2 cm behind
the solvent front. The paper is then replaced in the solvent chamber and
the solvent left to move up the paper. If the sample spot is placed on the
dry paper and the solvent allowed to overtake it in the usual way, severe
streaking may result. The reason is that the resin particles undergo a drastic
change in swelling and in ionic content when the solvent meets them.
Chromatography on resin-impregnated paper can be used as a quick
exploratory tool to predict the results of column operations, but it must
be used with caution as the cellulose matrix does on occasion influence the
migration of the ions. It is a nice way to measure the stability of complex
ions,44 especially complexes with low formation constants. Modified cellu-
lose papers, those in which the ionic groups are built into the cellulose
molecules, have also been used widely for inorganic paper chromatog-
raphy.45
Turning to the use of ion-exchanging paper as a filter for strongly
absorbed ions, the technique is to clamp the paper in a circular aperture
through which the solution is made to flow. Usually two or three paper
discs are used together. The paper circles loaded with metal ions are generally
dried and analyzed by x-ray fluorescence or other means without further
treatment. 46

7. APPLICATIONS TO GEOCHEMICAL ANALYSIS

In this section we shall note a few representative applications of ion-

exchange techniques to geochemical problems.
88 H. F. Walton

7.1. Concentration of Traces of Metals

Sea water has a high concentration of sodium and chloride ions, and
therefore an absorption process of high selectivity is needed to separate and
concentrate trace elements. Some ion-exchange procedures that are in use
have already been cited. 8,33 Resins with special functional groups, such as
carboxylate or iminodiacetate (chelating), are very selective for "heavy
metals" and have a relatively low affinity for sodium ions. Thus, traces of
copper, zinc, and bismuth have been absorbed from sea water by passing
it through a sodium-form carboxylic resin. 47 These elements are later stripped
off the exchanger by passing dilute hydrochloric acid through the column
and a spectrographic determination may subsequently be performed. Che-
lating resins have been used in a similar way with industrial waste waters.48
Strong-base quaternary anion-exchange resins can absorb those metals that
form strongly absorbed chloride complexes. These are primarily the triva-
lent metals that give singly charged complexes such as MCl~-. Because sea
water is already 0.5M in chloride ions it is not necessary to add much
hydrochloric acid. Thus, bismuth, whose concentration in sea water is
about 2 parts in 1011, is removed by adding 90 ml of concentrated hydro-
chloric acid to 10 liters of sea water and passing through a small column
of quaternary-base resin.49 It is eluted with 1M nitric acid.
Cesium is concentrated from sea water, river water, and marine sedi-
ments by using selective inorganic exchangers such as ammonium molyb-
dophosphate. 5o,51 In river waters the cesium concentration is about 2 parts
in 1011. Radioactive Cs-137 has been collected from sea water, milk, and
other environmental samples,52 with potassium hexacyanatocobalt(II) fer-
rate(II). Radium has been recovered from sea water by coprecipitation with
calcium carbonate followed by separation on a cation-exchange resin of the
sulfonated polystyrene type. 53
Schemes have been proposed for the separation and determination of
major cations in sea water,54-56 that is, Na+, K+, Mg2+, Ca 2+, Sr2+, and Ba2+.
The techniques depend on the relative strengths of absorption by a sul-
fonated polystyrene resin, as well as other selective reactions, such as the
precipitation of K + by sodium tetraphenylboron.
The analyst should never restrict himself to one method of separation,
and the method that most often complements ion exchange for the con-
centration and separation of metals is solvent extraction. Brooks57 ,58 has
used the Craig countercurrent extraction technique to separate trace metals
in geochemical analysis, and in one of his methods,57 sea water is passed
through a succession of extraction tubes containing dithizone in carbon
4. Ion-Exchange Chromatography 89

tetrachloride. The elements TI, Au, Cu, Pd, and Pt are retained in that
order. Another reagent used in the same way is 8-hydroxyquinoline in
chloroform. Here, gold is the first element to be retained, followed by tin.
Eight liters of sea water gave 20 mg of combined solid extract. After remov-
ing the solvent and ashing, a concentration ratio of 400,000 was achieved.
The final determination was by emission spectroscopy.
Countercurrent solvent extraction has this in common with ion-ex-
change chromatography, that it is a multistage process and allows small
differences in distribution ratios to be magnified into acceptable separations.

7.2. Removal of Interfering Ions

This is a straightforward application of an ion-exchange column, but
it has many practical uses. As noted in Section 3, anions can be separated
from all metallic cations and converted into their respective acids by passing
the solution through a cation-exchange resin in its hydrogen form. A some-
what more complicated process is the separation of fluoride ions from inter-
fering cations and anions. This is used in the analysis of ores and phosphate
rocks 59 as well as natural and industrial water. 60 ,61 Where separation from
other anions, such as phosphate, is required the fact is used that fluoride
ions are absorbed less strongly than other anions and are the first to be
released from an anion-exchange resin column by passing hydrochloric
acid or sodium hydroxide. By making use of sodium hydroxide, the fluo-
silicate ion, SiF~-, is broken down into silicate and fluoride, and the fluoride
is eluted first.
Of course, the practical need for separating fluoride from other ions
has been reduced by the development of the lanthanum fluoride membrane
electrode.
A somewhat analogous case to the separation of fluoride ions is the
separation of lithium ions from other cations. Sulfonated polystyrene resins
have a lower affinity for Li+ than for any other cation, and the difference
between Li2+ and K + is accentuated by adding alcohol. In analyzing water
from the Dead Sea, lithium chloride was separated from 1000 X its weight
of sodium chloride by absorbing the salts in a cation-exchange-resin column
and eluting with hydrochloric acid in aqueous methanol. The lithium
emerged first.62
Ion-exchange resins are sometimes used in batch form. They may be
used, for example, to release anions or cations from difficultly soluble salts.
Calcium sulfate and even barium sulfate63 is disproportionated by stirring
with a suspension of sulfonated polystyrene resin in the hydrogen form at
90 H. F. Walton

80-90 for some hours. Barium and calcium ions go into the resin and may
be recovered later if desired. The sulfate ions yield their equivalent in sul-
furic acid, which remains in solution and may be titrated as such. This
procedure has obvious use in mineral analysis. Another "batch" use of
resins is in the determination of traces of boron in rocks. 64 The rock is
fused with potassium carbonate and the cake from the fusion is broken
up with a slurry of sulfonated polystyrene resin in the hydrogen form, to
which mannitol is added. The mannitol keeps the boron in solution as the
borate-mannitol complex ion, while most of the cations from the rock and
the flux, which would otherwise interfere with the borate determination,
enter the resin. After some hours of contact the resin is filtered and washed,
while the boron may be determined spectrophotometrically in the filtrate.
Ten flg of added boron could be recovered from a 200-mg rock sample.
Traces of boron in sedimentary rocks serve as indicators of paleosalinity.

7.3. Systematic Analysis of Silicate Rocks

7.3.1. Ion Exchange and Solvent Extraction

The most comprehensive scheme for the analysis of silicate rocks,
based on separations by ion exchange and solvent extraction, is that of
Ahrens, Edge, and Brooks. 65 They point out that many minor elements
commonly occur in concentrations below the limit of spectrographic de-
tection, and for this reason alone, separation and preconcentration is
required. The separation of each individual element in pure form was not
necessary, since the final determination was usually made spectrographically.
However, some degree of separation was necessary to control matrix effects
and interelement interferences.
To dissolve the rocks they used hydrofluoric acid mixed with sulfuric,
nitric, hydrochloric, or perchloric acids. In this way they avoided introduc-
ing other cations that would compete with the cations sought in the ion-
exchange separations. The solid residues were dissolved in 2M hydro-
chloric acid, and the solutions divided into four aliquot portions, which
were treated according to the following scheme65 :
Portion No.1. Pass through strong-base anion-exchange resin, wash
with 2M HCI.
Retained on column: (a) Cd, Zn, Bi, n, Sn, and Ag; elute with O.25M
HNO a . (b) Platinum metals and gold; elute with O.25M HNO a followed
by IMNH a .
Passing: alkali and alkaline earth elements, lanthanides, AI, Fe, Ti,
4. Ion-Exchange Chromatography 91

Mn, Ni, Co, and V. This solution is further analyzed by cation exchange
(see Portion No.4).
Portion No.2. Treated by solvent extraction: (a) Make 6M in HCI,
extract with tributyl phosphate; Mo extracted, with some Fe and V.
Mo is then back-extracted from tributyl phosphate by shaking with
water. (b) Take another portion, make 1.7M in HCI and O.IM in HI,
pass S02 gas to reduce thallium to n(I). Extract with ethyl ether:
n and In are extracted as iodo-complexes.
Portion No.3. Make l1.3M in HCI, pass through strong-base anion-
exchange resin; wash with l1.3M HCI.
Retained on column: Ti, Zr, Co, Fe, and Ga. These are eluted succes-
sively in that order by decreasing concentrations of hydrochloric acid
(see Section 4). Ti and Zr are eluted with 8M, Co with 4M, and Fe
and Ga with 1M HCI.
Passing: alkali and alkaline earth elements, lanthanides, AI, Mn, Ni,
V, Pb, and Ag. This solution is further analyzed by cation exchange
(see Portion No.4).
Portion No.4. Pass through a column of sulfonated polystyrene cation-
exchange resin; wash with 2M HCI, increasing the HCI concentration
later to speed up the elution. Elements are eluted sequentially in the
following order: AI, Ti, Zr, Zn, Sn, and Pb; these elements form
negatively charged chloro-complexes and are easily eluted. Li, followed
by Na and V [and Fe(III) if this has not already been removed],
Mg, Mn, Ni, Co, K, Rb, Cs, Ga, AI, Ca, Sr, and much later Ba.

The original publication 65 gives details of the separation and deter-

mination of each element, and explains how the procedure may be modified
to suit different rock types.

7.3.2. Countercurrent Solvent Extraction

Boswell and Brooks58.66.67 have also developed comprehensive separa-
tion schemes for metallic elements in rocks that depend entirely on counter-
current solvent extraction. They use a I20-tube Craig extractor. The moving
solvent is methyl isobutyl ketone, and the stationary solvent is aqueous
hydrochloric acid which starts in the first 20 tubes at 7M. HCI is added to
the remaining tubes in progressively lower concentrations. The first ele-
ments to be extracted from 7M HCI are Fe(III), Ga, and Mo, and they
move along farthest with the solvent, while such elements as calcium and
92 H. F. Walton

sodium remain behind in the first few tubes. Roughly, the order of extrac-
tion parallels the order of binding by an anion-exchange resin in hydro-
chloric acid. The big drawbacks to countercurrent solvent extraction as an
analytical tool are the cumbersome equipment and the limited number of
"theoretical plates," which in this case is the actual number of tubes. The
advantages, compared to ion exchange, are that larger quantities of material
can be separated, that "tailing" is much less serious, and that the solvents
can be purified from trace contaminants more easily than in the case of a
resin.

7.3.3. Anion Exchange

Returning to ion exchange, a scheme has been published for silicate

rock analysis that uses differential complexation by sulfosalicylic acid. 68
This scheme separates primarily the more abundant elements, Ti, AI, P,
Fe, Mg, Mn, Ca, Na, and K, as well as Si. The metals Fe, AI, and Ti form
sulfosalicylate complexes and are bound by an anion-exchange resin. They
are further separated by anion exchange in hydrochloric acid, where Fe
and Ti are separated on a column of carboxylic cation-exchange resin and
are eluted in that order.
Anion exchange in hydrochloric acid solution has been used to collect
a group of trace elements from rocks and soils. 69 The rock is dissolved in
hydrofluoric, nitric, and sulfuric acids, the solution is evaporated, and the
metal salts dissolved in 1.5M HCl and passed through an anion-exchange
resin. The column is washed with 1.5M HCl until molybdenum just starts
to appear. Then the resin is removed and ashed at a controlled temperature,
and the ash is analyzed spectrographically. The elements thus collected are
Ag, Bi, Cd, Sb, Sn, Zn, Pb, and Mo.

7.4. Individual Elements in Silicate Rocks

The literature is rich in references to the use of ion exchange as one

part of a general separation scheme, where it is used to separate or isolate
certain elements. A good source of information is a lO-page table of ana-
lytical determinations of major and minor elements in the standard granite
and diabase samples, G-l and W-l, in a review by Fleischer. 70 Many sepa-
rations by ion exchange are cited, together with even more by solvent ex-
traction. Anion exchange in hydrochloric acid is the most popular ion-
exchange method.
For determining trace elements in rocks and meteorites activation
4. Ion-Exchange Chromatography 93

analysis is frequently used, and ion-exchange separations often follow the

irradiation step. The isolation of potassium, rubidium, and cesium from
chondrite meteorites may be cited.71 The irradiated sample is sintered with
sodium peroxide, then dissolved in hydrochloric acid. Inactive rubidium
and cesium carriers are added, and K, Rb, and Cs precipitated together
with sodium tetraphenylboron. The precipitate is oxidized with nitric and
perchloric acids, and after an intermediate scavenging step the three alkali
metals are separated by chromatography on ammonium molybdophosphate
(AMP).
A general separation of the alkali metals, alkaline earth metals, and
rare earths (lanthanides) from chondrites following neutron irradiation has
been reported, using a cation-exchange resin. 72 The three groups of cations
are eluted by successively increasing concentrations of hydrochloric acid.
Scandium 73 is separated from silicate rocks, following hydrofluoric-per-
chloric acid treatment, by a three-stage ion-exchange procedure, the inter-
esting part of which is that scandium is absorbed on an anion-exchange resin
in the sulfate form, then eluted (as its soluble sulfate complex) by an am-
monium sulfate-sulfuric acid mixture. The final purification steps are copre-
cipitation with hydrous ferric oxide, followed by dissolving in 6M hydro-
chloric acid and passing through an anion-exchange resin. Iron and other
associated trace elements are retained, while the scandium passes on.
Titanium, niobium, and tantalum are separated as their fluoride com-
plexes by anion exchange,22.23,74 and if they are present in large amounts
they can be separated as chloride complexes with sodium chloride eluents. 75
Zinc,76 antimony, tin, and mercury77 have been isolated from meteorites
by anion exchange, using hydrochloric acid eluents (zinc is retained very
strongly from hydrochloric acid) as well as fluoride (which elutes antimony)
and phosphoric acid (which elutes tin). The elements Se, As, Sb, Sn, and
Hg have been removed from the matrix after neutron irradiation by distilla-
tion as their volatile chlorides. 77 Distillation (or volatilization) is a useful
and "clean" method for separating posttransition elements from mixtures,
and it has been used frequently in geochemical studies.
Gold and the platinum metals have frequently been separated from
meteorites and silicate rocks with the help of ion-exchange techniques. The
authoritative work on separations of the platinum metals is that of Beam-
ish. 78 The application to meteorites is illustrated by a recent paper79 which
describes analysis by neutron irradiation followed by the absorption of
Pt, Pd, Ir, and Au on an anion-exchange resin which is mixed with cerium
dioxide, to ensure conversion of these elements to their highest oxidation
states. These four elements are then separated by selective elution with
94 H. F. Walton

thiourea and hydrochloric acid; Pb, Pd, and Au are eluted as their thiourea
complexes, while Ir remains on the column and is later removed by 6M He!.
Thiourea is a useful eluent for ion-exchange separations of precious metals
and has been used in the determination of these metals by activation analysis
in deep-sea manganese nodules. so The existence of a chelating resin which is
selective for precious metals was noted earlier (see Section 2.1), and now
that this resin can be purchased commercially, it may find uses in geo-
chemical analysis.
Many more examples could be cited, but these are representative, not
only of the elements that can be separated by ion exchange, but also of the
various ways of using ion exchange in conjunction with other separation
methods. A compilation of ion-exchange separations of individual elements
(many of them of geochemical interest) appears in alternate years in the
"Fundamental Reviews" section of Analytical Chemistry.S!

REFERENCES
1. Rosset, R., Bull. Soc. Chim. France 1966, 59.
2. Koster, G., and Schmuckler, G., Anal. Chim. Acta 38, 179 (1967).
3. Pinon, F., Deson, J., and Rosset, R., Bull. Soc. Chem. France 1968, 3454.
4. Gorenc, B., and Kostka, L., A. Anal. Chem. 223, 410 (1966).
5. Prout, W. E., Russell, E. R., and Groh, R. J., J. Inorg. Nucl. Chem. 27, 473 (1965).
6. Albertsson, J. Acta Chem. Scand. 20, 1689 (1966).
7. Bruce, T., and Ashley, R. W. At. Energy Canada Ltd., AECL-2648 (1966).
8. Bauman, A. J., Weetall, H. R., and Weliky, N., Anal. Chem. 39, 932 (1967).
9. Glasa, O. S., Anal. Chem. Acta 28, 543 (1963).
10. Evans, W. R., and Sargent, G. A., Analyst 92, 690 (1967).
11. Greenland, L. P., Professional Papers USGS 1967, No. 575-C, 137.
12. Rengan, K., and Meinke, W. W., Anal. Chem. 36, 157 (1964).
13. Pollard, F. H., Nickless, G., and Spincer, D., J. Chromatog. 10, 215 (1963).
14. Blake, W. E., Oldham, G., and Sumpter, D., Nature 203, 862 (1964).
15. Strelow, F. W. E., Rethemeyer, R., and Bothma, C. J. c., Anal. Chem. 37,106 (1965).
16. Kraus, K. A., and Nelson, F., Proc. 1st United Nations Con! on Peaceful Uses of
Atomic Energy 7, 113 (1955); The Structure of Electrolytic Solutions, W. H. Hamer,
ed., John Wiley & Sons, New York, 1959, Chap. 23.
17. Fritz, J. S., and Garralda, B. B., Anal. Chem. 34, 102 (1962).
18. Faris, J. P., Anal. Chem. 32, 521 (1960).
19. Faris, J. P., and Buchanan, R. F., Anal. Chem. 36, 1158 (1964).
20. Strelow, F. W. E., and Bothma, C. J. C., Anal. Chem. 39, 595 (1967).
21. Turner, J. B., Philip, R. H., and Day, R. A., Anal. Chim. Acta 26, 94 (1962).
22. Hague, J. L., and Machlan, L. A., J. Res. Nat. Bur. Standards 62, 53 (1959).
23. Dixon, E. J., and Headridge, J. B., Analyst 89, 185 (1964).
24. Setser, J. L., and Ehmann, W. D., Geochim. Cosmochim. Acta 28, 769 (1964).
25. Fritz, J. S., and Rettig, T. A., Anal. Chem. 34, 1562 (1962).
26. Fritz, J. S., and Abbink, J. E., Anal. Chem. 37, 1274 (1965).
4. Ion-Exchange Chromatography 95

27. Korkisch, J., and Hazan, I., Anal. Chern. 37, 707 (1965).
28. Hazan, I., and Korkisch, J., Anal. Chirn. Acta 32, 46 (1965).
29. Korkish, J., Separation Sci. 1, 159 (1966).
30. Korkisch, J., and Ahluwalia, S. S., Anal. Chirn. Acta 34, 308 (1966).
31. Korkisch, J., and Hazan, I., Anal. Chern. 36, 2308 (1964).
32. Korkisch, J., Mikrochirn. Acta 1967, 401.
33. Hazan, I., Korkisch, J., and Arrhenius, G., Z. Anal. Chern. 213, 182 (1965).
34. Siegel, A., and Degens, E. T., Science 151, 1098 (1966).
35. Shimomura, K., Dickson, L., and Walton, H. F., Anal. Chirn. Acta 37, 102 (1967).
36. Tera, F., and Morrison, G. H., Anal. Chern. 38, 959 (1966).
37. Kuroda, R., and Oguma, K., Anal. Chern. 39, 1003 (1967).
38. Tera, F., Ruch, R. R., and Morrison, G. H., Anal. Chern. 37, 358, 1565 (1965).
39. Morrison, G. H., in Trace Characterization, W. W. Meinke and B. F. Scribner, eds.
U.S. National Bureau of Standards, Monograph 100, 1967.
40. Mangelsdorf, P. c., Anal. Chern. 38, 1540 (1966).
41. Sherma, J., Separation Sci. 2, 177 (1967).
42. Sherma, J., Anal. Chern. 39, 1497 (1967).
43. Lederer, M., Ann. del. Inst. Superiore de San ita 2, 150 (1966).
44. Lederer, M., and Ossicini, L., J. Chrornatog. 22, 200 (1966).
45. Muzzarelli, R. A. A., Talanta 13, 193, 1908 (1966).
46. Campbell, W. J., Spano, E. F., and Green, T. E., Anal. Chern. 38, 987 (1966).
47. Brooks, R. R., Analyst 85, 745 (1960).
48. Biechler, D. G., Anal. Chern. 37, 1054 (1965).
49. Portman, J. E., and Riley, J. P., Anal. Chirn. Acta 34, 201 (1966).
50. Feldman, C., and Rains, T. C., Anal. Chern. 36, 405 (1964).
51. Sreekumaran, C., Pillai, K. C., and Folsom, T. R., Geochirn. Cosrnochirn. Acta 32,
1229 (1968).
52. Boni, A. L., Anal. Chern. 38, 89 (1966).
53. Sugimura, Y., and Tsubota, H., J. Marine Res. 21, 74 (1963).
54. Khristova, R., and Krushevska, A., Anal. Chirn. Acta 36, 392 (1966).
55. Greenhalgh, R., Riley, J. P., and Tongudai, M., Anal. Chirn. Acta 36, 439 (1966).
56. Szabo, B. J., and Joensun, 0., Envir. Sci. Technol. 1, 499 (1967).
57. Brooks, R. R., Talanta 12, 505, 511 (1965).
58. Brooks, R. R., Geochirn. Cosrnochirn. Acta 29, 1369 (1965).
59. Glasoe, O. S., Anal. Chirn. Acta 28, 543 (1963).
60. Jeffery, P. G., and Williams, D., Analyst 86, 590 (1961).
61. Kelso, F. S., Matthews, J. M., and Kramer, H. P., Anal. Chern. 36, 577 (1964).
62. Ratner, R., and Ludmer, Z., Israel J. Chern. 2, 21 (1964).
63. Osborn, G. H., Synthetic Ion-Exchangers, Macmillan, New York, 1956.
64. Fleet, M. E., Anal. Chern. 39, 253 (1967).
65. Ahrens, L. H., Edge, R. A., and Brooks, R. R., Anal. Chirn. Acta 28, 551 (1963).
66. Boswell, C. R., and Brooks, R. R., Talanta 14, 655 (1967).
67. Boswell, C. R., and Brooks, R. R., Mikrochirn. Ichnoanal. Acta 1965, 814.
68. Maynes, A. D., Anal. Chirn. Acta 32, 211 (1965).
69. LeRiche, H. H., Geochirn. Cosrnochirn. Acta 32, 791 (1968).
70. Fleischer, M., Geochirn. Cosrnochirn. Acta 29, 1263 (1965).
71. Smales, A. A., Hughes, T. C., Mapper, D., McInnes, C. A. J., and Webster, R. K.,
Geochirn. Cosrnochirn. Acta 28, 209 (1964).
96 H. F. Walton

72. Shima, M., and Honda, M., Geochim. Cosmochim. Acta 31, 1995 (1967).
73. Shimizu, T., Anal. Chim. Acta 37, 75 (1967).
74. Kallman, S., Oberthin, H., and Liu, R., Anal. Chem. 34, 609 (1962).
75. Smith, F. W., J. Chromatog. 22, 500 (1966).
76. Nishimura, M., and Sandell, E. B., Anal. Chim. Acta 26, 242 (1962).
77. Kiesl, W., Z. Anal. Chem. 227, 13 (1967).
78. Beamish, F. E., Talanta 14, 991, 1133 (1967).
79. Crocket, J. H., Keays, R. R., and Hsieh, S., Geochim. Cosmochim. Acta 31, 1615
(1967).
80. Harris, R. C., Crocket, J. H., and Stainton, M., Geochim. Cosmochim. Acta 32,1049
(1968).
81. Walton, H. F., Anal. Chem. 36, SIR (1964); 38, 79R (1966); 40, SIR (1968).
Chapter 5

COLORIMETRY

Gordon A. Parker
Toledo University
Toledo, Ohio

and
D. F. Boltz
Wayne State University
Detroit, Michigan

1. INTRODUCTION

This chapter consists of a discussion of the fundamental principles, tech-

niques, and methods involved in colorimetric analysis. For many years
chemical colorimetry has been a leading analytical method for the deter-
mination of small amounts of inorganic constituents, i.e., when the per-
centage of the desired constituent is in the 10-5-5% by weight concentration
range. The determination of elements in geological samples based on the
intensity of colored solutions obtained by appropriate chemical treatment
of these materials is a well-established quantitative method that is frequently
used in geochemical analysis. In addition to the necessity of developing a
suitable colored solution, one also has to find a suitable means of relating
color intensity to the concentration of a specific element. In the first one-
third of the 20th century it was common analytical practice to use the human
eye to compare the color intensities of solutions of known concentration
with those of similar solutions of unknown concentration. Visual color-
imetry is now used only to a very limited extent. Presently, modern photo-

97
98 Gordon A. Parker and D. F. Boltz

electric instrumentation has replaced the visual color comparators, with

their inherent limitations involving possible defective visual acuity and
human judgement, and provides a more reliable, discriminate, and sensitive
means of relating color intensity to concentration. In fact, modern instru-
mentation has permitted the extension of the spectrophotometric method to
the measurement of the absorptivity of solutions in regions of the electro-
magnetic spectrum not discernible to the human eye. Hence, UV spectro-
photometry and IR spectrophotometry (see Chapter 6) are adjunct methods
to colorimetry.

2. THEORETICAL CONSIDERATIONS
2.1. Measurement of Light Absorption
Light is but a small portion of the electromagnetic spectrum composed
at one extreme of x-ray radiations and at the other extreme of radio waves.
Electromagnetic waves are characterized by their wavelength, the distance
between corresponding points of a repeating cycle, and by their frequency,
the number of complete wave cycles that pass a reference point in a given
time. Using A to represent the wavelength and v the frequency, the relation
Av = C = const = speed of light (1)

holds for all electromagnetic radiant energy. The portion of the electro-
magnetic spectrum which is discernible to the human eye consists of the
wavelength region from approximately 380 to 750 mf-l. It varies somewhat
among different people and with the same person, depending upon age.
The neighboring region consisting of wavelengths shorter than this is the
UV region, and the neighboring region consisting of wavelengths longer
than this is the IR region. Both wavelength and frequency are commonly
expressed in several different units. In addition, the term wave number is
used in place of wavelength to represent the location of a point in an electro-
magnetic spectral region. The wave number, number of wave cycles per unit
length, is represented by ji and is defined as

(2)
where A is in centimeters.
For a colored solution, the relation of color intensity to concentration
of colored component may be obvious to a human observer and one speaks
of a colorimetric method of analysis or colorimetry. Analogous methods are
possible in the neighboring UV and near IR regions of the electromagnetic
5. Colorimetry 99

spectrum. The relationships are not obvious because the human eye does
not perceive radiant energy in these regions. The term colorimetry is often
replaced by the more general term spectrophotometry, or a more specific
term, light-absorption spectrometry. Principles of spectrophotometry apply
equally well to measurements in all three regions of the spectrum, UV,
visible, and near IR. The discussion in this chapter will be limited to visible
spectrophotometry but the relationships developed and the procedures
presented will, provided the proper instrumentation is utilized, apply also
to other spectral regions.
To state that a solution is colored means that of all wavelengths of white
light incident upon the solution only selected wavelengths are absorbed,
depending on the color of the solution; the remaining wavelengths are
transmitted. A red solution, for example, appears red because it absorbs the
shorter wavelengths of the visible region and transmits the longer wave-
lengths. Thus, color is attributed to the selective absorption of incident
radiant energy of certain wavelengths. The absorption of certain wavelengths
of light by colored solutions is due to energy transitions of the electrons
within the atoms of the absorbing component. At a given temperature each
electron resides in a definite lowest energy state, the ground state, and
transitions to higher energy states occur provided the exact amount of
energy, corresponding to the difference between the energy levels involved,
is available. Each wavelength of radiant energy corresponds to a different
amount of energy, that is,

E=hv=~ (3)
A

where E is the amount of energy corresponding to a given frequency of

radiant energy v. The quotient c / A may be written in place of v. The term fl,
Planck's constant, relates these quantities. If polychromatic light, radiant
energy of several wavelengths, is incident upon a solution, those wavelengths
(energies) corresponding exactly to electronic transitions to higher energy
states will be absorbed and those which do not correspond to electronic
transitions for the particular species in solution will be transmitted. Hence,
a red solution does not absorb the longer wavelengths of light because the
corresponding energy is unable to cause electronic transitions within the
atoms of the absorbing species. A colored solution absorbs radiant energy
at those wavelengths at which a solution of complementary color transmits
those wavelengths. A red solution absorbs radiant energy in the blue-
green region of the spectrum. Table I lists some absorbed and transmitted
colors.
100 Gordon A. Parker and D. F. Boltz

Table I
Relationship Between Selective Light Absorption and Color

Observed (transmitted) color Region of maximum Complementary (absorbed) color

(color of solution) transmittance (color of suitable glass filter)

Red 610-750 mil Blue-green

Orange 595-610 mil Green-blue
Yellow 570-600 mil Blue
Green 500-570 mil Purple
Green-blue 480-500 mil Red
Blue 440-480 mil Yellow
Violet 400-440 mil Yellow-green

If monochromatic radiant energy known to be absorbed by a particular

species in solution is allowed to irradiate this solution, the extent to which
it is absorbed depends both upon the depth of solution through which it
passes and the concentration of absorbing species in the solution. These
relationships between intensity, light path, and concentration were developed
by Bouguer, Lambert, and Beer. They are expressed mathematically as
follows 85 :
_P = e- kbc = lo-abc (4)
Po

where Po is the incident, monochromatic, radiant power, P is the transmitted,

monochromatic, radiant power, PjPo is the transmittance, the fraction of
radiant power transmitted, b is the path length, the distance traversed by
the light in passing through the sample, c is the concentration of the absorb-
ing species in solution, e is the base of natural logarithms, k is a constant
relating the various quantities, and a is a constant equal to kj2.3.
Taking the logarithm of both sides of Eq. (4) and inverting to change
sign gives
Po 1
A = log- = log- = -log T = abc (5)
P T

where A is the absorbance, the negative logarithm of the transmittance.

Beer's law is most often expressed in the form A = abc. The constant a is
the absorptivity. Its numerical value depends upon the choice of concentra-
tion units for c. Generally, b, the path length, is in centimeters, and if c is
in moles per liter, a becomes the molar absorptivity and is represented by
5. Colorimetry 101

the symbol B in units of liters per mole per cm. Absorbance is a dimension-
less quantity.
In the past various other symbols and names have been used to repre-
sent the terms of the Beer's law relation but these are not recommended. 2 ,62

2.2. Colorimetric Techniques

Beer's law serves as the basis for the quantitative colorimetric deter-
mination of many substances. It relates the absorbance of a solution directly
to the concentration of absorbing species. Several different techniques are
available for ascertaining concentration in colorimetry.

2.2.1. The Standard Series Technique

This technique requires the preparation of a series of solutions of
known concentration in respect to the desired constituent. The absorbance
of each solution is measured and a calibration plot of absorbance vs con-
centration is prepared. By measuring the absorbance of the unknown solu-
tion, its concentration may be read directly from the calibration curve. If the
system under investigation obeys Beer's law exactly over the entire con-
centration range studied, the calibration graph is linear. If Beer's law is
not obeyed and a nonlinear plot is obtained, the data are still valid and the
graph serves as a calibration curve for the quantitative determination of the
desired constituent.

07

06

05

.
'"~
'"
a:
04

.'"
~ 03
A
02

01

C
0~~~2--~~4~~~6~~78~~

CONCENTRATION OF DESIRED CONSTITUENT, mQ/1

Fig. 1. Beer's law graph: plot of absorbance

vs concentration. A') absorbance of unknown
solution; C') concentration of unknown
solution.
102 Gordon A. Parker and D. F. Boltz

Figure I shows a typical Beer's law plot expected when using the
standard series technique. This method is widely used and reliable results
are obtained provided instrumental settings and sample treatment are iden-
tical for all known and unknown solutions. A calibration curve is especial-
ly useful in repetitive analyses of similar samples. The time necessary to
prepare a calibration curve is a disadvantage of this technique, especially
if only one or two samples are to be analyzed.

2.2.2. The Absolute Technique

This technique, although most applicable to a single determination,
is normally not as reliable. The absorbance of a solution containing a known
concentration of the desired constituent in a cell of known path length is
measured and the absorptivity is calculated. This value is substituted into
the Beer's law equation and solved for the concentration of the unknown
solution based upon its observed absorbance. Deviations from Beer's law
by the system under study negates the accurate determination of unknown
concentrations by this method. Even if the known concentration is nearly
equal to the unknown concentration, the possibility of errors based upon a
single absorbance measurement may be appreciable.

2.2.3. Indirect Spectrophotometric Method

Some constituents are best determined by an indirect spectrophoto-
metric method. The decrease in absorbance of a colored solution is a measure
of the amount of constituent reacting, and may therefore be used to deter-
mine the concentration of the desired constituent present. 75

2.2.4. Simultaneous Determination

Colorimetric methods are not restricted to the determination of a single
constituent. The simultaneous determination of two or more constituents in
the same sample is feasible. The technique to be adopted will depend on the
nature of the particular chemical system. If for a two-component system
the absorption spectra of constituents I and II possess an absorbance
maximum at wavelengths where the other does not absorb, neither con-
stituent will interfere with the determination of the other-assuming, of
course, that no chemical interactions occur. Thus, constituent I can be
determined by measurements at the wavelength of its absorbance maximum
and constituent II can be determined by measurements at the wavelength of
its absorbance maximum. If the absorption spectra of I and II overlap, as
5. Colorimetry 103

14

12

10

WAVELENGTH. m p

Fig. 2. Absorption spectra of two-component

system.

is shown in Fig. 2, the simultaneous determination may still be possible.

In cases where the spectra of both constituents obey Beer's law at the
wavelengths of interest the absorbance values observed for the mixture
will be the sum of the absorbances of each constituent. The optimum wave-
length for measurement of each constituent is the wavelength at which the
ratio of the absorptivity of one constituent to that of the other is a maximum.
Then, the absorptivity of each constituent at all wavelengths selected for
measurement are determined. The absorbance of the mixture is measured
at these same wavelengths using cells of known path length. From these
data the concentration of each constituent may be calculated by solution
of a set of simultaneous equations. For constituents I and II and measure-
ment at wavelengths A(X) and A(y) the following equations apply:

at A(X)

at A(y)

where C1 is the concentration of I and CII is the concentration of II.

Equations (6) and (7) are solved simultaneously for C1 and Cn .
Detailed procedures for simultaneous spectrophotometric analysis of poly-
component systems are available. 12 If any or all of the constituents com-
prising the sample do not obey Beer's law, the above approach must be
modified and a more involved treatment is necessary.132
104 Gordon A. Parker and D. F. Boltz

2.2.5. Photometric Titrimetry

Another interesting aspect of chemical colorimetry is found in photo-
metric titrimetry. The desired constituent is titrated with a suitable reagent
and the equivalence point of the titration is detected by measurement of
a color change of the solution. Observed data are plotted to obtain
titration graphs similar to one of the types shown in Fig. 3. The graphs are
linear, except perhaps in the immediate vicinity of the endpoint region.
Generally, data are recorded at a few points prior to the endpoint and at
a few points beyond the endpoint. The linear portions of the curves based
on these points are extrapolated and the point of intersection is taken as the
endpoint of the titration. A photometric titration can be performed in a
titration flask by periodically removing aliquots for photometric measure-
ments, or a flow system is used whereby the solution in the titration flask
is circulated through the absorption cell in the spectrophotometer. The
solution is forced to circulate during the titration by means of a pump.
Another alternative technique is to perform the titration directly in a vessel
located inside the cell compartment of the instrument. Many instruments
have cell compartments large enough to permit this method to be used.
Photometric titration graphs differ depending upon whether the sample,
titrant, or both, absorb at the wavelength chosen to measure the absorbance
of the solution. Photometric titration methods for many elements appear
in the literature. 56

A DESIRED CONSTITUENT ABSORBS. TI TRANT

AND PRODUCTS DO NOT ABSORB

II~ FRACTION TITRATED

B TITRANT ABSORBS; DESIRED CONSTITUENT

AND PRODUCTS DO NOT ABSORB

II~
FRACTION TITRATED
C DESIRED CONSTITUENT AND TiTRANT ABSORB;
PRODUCTS DO NOT ABSORB

FRACTION TITRATED

Fig. 3. Photometric titration graphs.

5. Colorimetry 105

3. INSTRUMENTATION

Color comparators utilizing visual observation to match the color in-

tensity of known and unknown solutions are not as widely used at the present
time and have been replaced by photoelectric instruments which are more
convenient and reliable. Photoelectric instruments differ widely with respect
to their design and capabilities. A device which measures the radiant power
or relative radiant power of light beams is called a photometer. If in addi-
tion to the photometer a spectrometer or monochromator is employed for
isolating a particular spectral region, the instrument is termed a spectro-
photometer. Spectrophotometers are described in terms of five components:
(a) source of radiant energy; (b) monochromator, with its accompanying
entrance and exit slits; (c) absorption cell compartment; (d) detector, which
responds to the radiant power incident upon it; and (e) readout device.
Spectrophotometers are also classified as single-beam or double-beam in-
struments. Single-beam instruments have one optical path from the exit
slit of the monochromator through the sample cell to the detector. Double-
beam instruments have two optical paths, of which one passes through the
sample cell and the other through the reference cell. Double-beam instru-
ments may utilize two detectors, one for each light beam, or a single de-
tector, in which case the response from each light beam is alternately
allowed to impinge upon the single detector. Many spectrophotometers,
depending upon their optical and photometric systems, can be utilized for
absorbance measurements in the UV and near IR spectral regions as well
as in the visible region. Additional information pertaining to instrumenta-
tion is given elsewhere. 73,121 ,148

3.1. Filter Photometers

The source of continuous radiant energy in filter photometers and
spectrophotometers for the visible region is generally an incandescent
tungsten-filament lamp. It emitts radiant energy throughout the visible
region, although the intensity of this radiation is not uniform. The spectral
emittance of a tungsten lamp is higher near the red region of the visible
spectrum, and the relative spectral emittance increases with increasing tem-
perature of the lamp filament. The power source operating the lamp must
be stable in as much as fluctuations in source power will cause fluctuations
in filament temperature and result in variations in the emittance. Variations
in source intensity will cause variations in the detector response and lead
to erroneous readings. A battery-operated de power supply or a stabilized
106 Gordon A. Parker and D. F. Boltz

ac power supply is necessary to insure a steady lamp output. With double-

beam instruments, lamp fluctuations are compensated for because the light
beams passing through the sample and reference solutions are simultaneously
subjected to the same variations. With single-beam instruments, in which
a time interval elapses between the measurement of the reference
solution and sample solution, source fluctuations can be a serious source
of error.
Filter photometers employ a filter to isolate a desired portion of the
spectrum. The filter usually consists of a piece of colored glass, or dyed
gelatin film mounted between glass plates, in order to transmit radiant
energy of a limited spectral region. Generally, the width of the wavelength
band passed by a filter is about 50 mil. The location of the transmitted band
depends upon the color of the filter. These are available for isolation of
almost any desired spectral region. The manufacturer's literature should be
consulted for the appropriate filter suitable for a particular wavelength
region. A compilation of these data is available. 142 The filter chosen for a
colorimetric analysis should transmit radiant energy in the region of maxi-
mum absorbance for the colored species being measured. In general, the
appropriate filter will have a color complementary to that of the absorbing
species (see Table I). Interference filters which isolate a narrower wavelength
region are also available and a compilation pertaining to their transmittance
characteristics should be consulted for further information. 15
Glass absorption cells are employed for measurements in the visible
spectral region. Silica cells, necessary for measurements in the UV region,
may also be used. Round cells are satisfactory if precautions are taken to
insure their consistent positioning in the cell holders. Absorption cells with
flat, parallel surfaces perpendicular to the optical path are preferred. These
cells have a definite path length, a necessary value in Beer's law calcula-
tions. Cells should also be optically matched, that is, they should give the
same transmittance or absorbance reading when filled with aliquots of the
same solution. If unmatched cells are used care must be take to always use
the same cell for sample and reference solutions. Cells require careful han-
dling and must be scrupulously clean. Cells of 1-, 2-, and 5-cm path length are
common. Larger cells, semi micro cells, microcells, variable-path-Iength cells,
flow-through cells, thermostated cells, high- and low-temperature cells, and
numerous other cells for specialized applications are available.
The detector employed in photoelectric instruments consists of a device
capable of converting radiant power into an electrical response. Some photo-
meters employ photo voltaic cells, or barrier-layer cells. These cells consist
of a layer of conducting and of semiconducting material deposited upon a
5. Colorimetry 107

metallic base. Upon irradiation, a current flows between these layers and a
galvanometer attached to the barrier layer cell registers the magnitude of
this current. In a direct-reading photometer, the galvanometer scale is cali-
brated directly in transmittance or absorbance units. In a null-type instru-
ment, the detector output is balanced by an external circuit until no current
flows through a galvanometer. The transmittance or absorbance of the
sample solution relative to that of the reference solution is read from a scale
on the external compensating device. Barrier-layer cells are sensitive to
radiant energy over approximately the same region as the human eye and
are widely used in filter photometers. The sensitivity of this type of photo-
cell decreases near the IR and UV regions. The electrical current produced
by barrier-layer cells is a linear formation of irradiance, provided the exter-
nal circuit resistance is low. Photovoltaic cells are rugged and relatively
inexpensive but because of their low impedance the photocurrent is not
easily amplified electronically. Upon continuous, prolonged exposure to
radiant energy their response changes. This fatigue effect, the slow response
to intermittent irradiation, and the sensitivity to fluctuations in temperature,
have seriously limited the use of this type of photocell, except in filter
photometers.
Numerous filter photometers are commercially available. The manu-
facturer's literature should be consulted for the specific details pertaining
to a particular instrument. A comparison of some of the more popular
photometers is available.128 The merits of several newer instruments have
also been discussed. 92

3.2. Spectrophotometers
The light source of a spectrophotometer for use in the visible region
is also an incandescent tungsten-filament lamp.
A spectrophotometer employs a monochromator consisting of either a
prism or grating as a dispersive element. White light composed of all possible
wavelengths incident upon the face of a prism is dispersed upon passing
through the prism and emerges as a spectrum of rays. The prism material
must transmit radiant energy in the region of interest. Glass can be used
for the visible region and quartz for the visible, UV, and near IR regions.
The dispersion of a prism is not linear and decreases as one approaches
longer wavelengths. A grating consists of a series of finely ruled lines upon
the face of a transparent or reflecting surface. White light incident upon
these rulings is dispersed into its spectrum. The dispersion from a grating
is approximately linear with wavelength if the striking angle of the incident
108 Gordon A. Parker and D. F. Boltz

radiation is small. Gratings produce second-, third-, and higher order

spectra. In some instances a higher order spectrum overlaps another of a
lower order. If this occurs the emerging light from the exit slit will no longer
be monochromatic. Gratings can be used in regions other than the visible.
Replica gratings, reproductions molded from a master grating, are commonly
used and are relatively less expensive.
Not only the dispersing element, either prism or grating, but also the
entrance and exit slits of the monochromator system determine the ability
of the system to isolate a narrow spectral region. A narrower entrance slit
will decrease the total intensity of the radiant energy incident upon the
dispersing element but, at the same time, increase the spectral purity of the
dispersed rays. An even narrower exit slit will further decrease the intensity
of the light available to the detector and isolate a narrower wavelength
region. Many commercial spectrophotometers have entrance and exit slits
of equal width and give a triangular energy distribution about the nominal
wavelength. The nominal wavelength is the wavelength value read from the
monochromator setting. A spectral band of slightly higher and lower wave-
lengths is emitted from the exit slit. The spectral slit width, also called the
spectral region isolated, encompasses the wavelength region actually emitted
from the exit slit. Within this isolated spectral region a narrower spectral
region is defined as the effective slit width, which represents the wavelength
region within three-fourths of the total radiant energy of the emergent beam.
It is desirable that the effective slit width be approximately one-tenth the
natural band width, the width of the wavelength region at half the height
of the absorbance maximum. A species exhibiting a narrow absorbance
maximum in its absorption spectrum will require a narrow effective slit
width for optimum resolution, sensitivity, and more likely adherence to
Beer's law. Thus, better resolution is obtained by specifying a narrow
mechanical slit width for the spectrophotometer used in making the measure-
ments. Many spectrophotometers have provisions for manual or automatic
adjustment of the slit width to take into account the requirements for
optimum resolution and nonlinear dispersion of prism instruments. Several
high-resolution spectrophotometers employ a double-monochromator sys-
tem with two prisms or gratings or a prism-grating combination to achieve
more nearly monochromatic radiant energy.
Spectrophotometers generally employ a photoemissive cell for detection
of radiant energy. The detector has a photoemissive cathode which, when
exposed to a beam of radiant power, emits electrons. The photoelectrons
are attracted to a positive anode present with the cathode in an evacuated
glass envelope, thus causing the generation of a photocurrent. The mag-
5. Colorimetry 109

nitude of the current is proportional to the radiant power of the incident

radiant energy. The sensitivity of a photoemissive tube varies with the
wavelength of incident radiant energy, and the nature of the photocathode
surface. More than one phototube may be necessary in a spectrophotometer,
depending upon which spectral region is being utilized. An increased photo-
current may be achieved for a photoemissive tube if the electrons emitted
from the cathode are allowed to strike a second cathode within the evacuated
envelope. More photoelectrons are emitted from the second cathode than
from the first cathode. If these second-stage photoelectrons next impinge
upon a third photocathode, etc., an avalanche effect is achieved which
results in a gain in amplification within the tube. The acceleration of elec-
trons from one photocathode to the next photocathode (dynode) is achieved
by having each successive photocathode at a slightly more positive potential.
Such tubes, containing 10-15 dynodes, are known as photomultiplier tubes
and they produce a relatively larger output signal than does a simple photo-
emissive tube. A large amplification is necessary with high-resolution spec-
trophotometers because the narrow slits required for a high-resolution
spectrum attenuate the total radiant energy reaching the detector. Photo-
tubes and photomultiplier tubes have a high impedance and their current
output can be increased conveniently by electronic amplifiers. However,
higher output signal amplification also results in an increase in electronic
noise and this ultimately limits the extent to which meaningful amplification
is possible. Photoemissive detectors require a power supply and with the
amplifier components contribute to the complexity and cost of a spectro-
photometer.
The amplified signal from a photoemissive tube can be directed to an
appropriate meter, in which case the instrument is direct reading in either
absorbance or transmittance units, depending upon the calibration of the
meter scale. The detector signal can also be compensated for by an external
signal with a galvanometer null indicator. This compensation can be either
optical, perhaps a movable optical wedge in the reference beam, or electrical.
In place of a galvanometer null indicator, the imbalance between detector
signal and reference signal can be used to drive a recorder pen. If the record-
er is coupled to the wavelength drive of the spectrophotometer, the auto-
matic recording of spectra may be achieved. The detector signal can be
displayed on the face of a digital meter, adjusted to read directly the absor-
bance of the sample solution. With proper calibration digital readout of
component concentration is possible. The output signal can be recorded on
punch cards or magnetic tape and utilized for computer calculation of
desired quantities.
110 Gordon A. Parker and D. F. Boltz

Numerous spectrophotometers are available from various manu-

facturers. The specifications of some of these instruments have been
summarized. 73

3.3. Instrument Selection

Depending upon its intended use, one type of instrument may be more
advantageous than another. For routine analysis of only a few constituents,
each of which forms a well-characterized species suitable for colorimetric
measurement, a filter photometer may be entirely satisfactory. If the colored
system under investigation demands the use of nearly monochromatic
radiant energy, a spectrophotometer with adequate slit control and sensitive
detector response is required. Studies involving new and untried photometric
procedures should be made with a high-resolution sensitive spectrophoto-
meter to insure proper delineation of optimum operating conditions. For
photometric measurements at a single wavelength, as in the preparation of
Beer's law plots, a manual, single-beam instrument is suitable. If one wishes
to obtain a complete absorption spectrum, a recording, double-beam in-
strument is faster and more convenient. For systems showing conformity
to Beer's law, meter or digital readout provides the absorbance or trans-
mittance value associated with each measured concentration. When UV
or near IR measurements are to be made, an instrument with an exchange-
able source and detector system, which allow for measurements in the UV
and near IR regions, should be considered. Numerous accessories are
available for certain spectrophotometers and the acquisition of some of
these accessory units may be advantageous for a particular laboratory.
Automatic sampling equipment, for instance, enables one to make measure-
ments on numerous samples automatically. Thermostated cells, provisions
for plotting or direct readout in either absorbance or transmittance, and
scale expansion over a narrower optical region are possible options. Cell-
compartment size for accommodation of a variety of cells may be important.
Provisions for flame emission, fluorescence, internal reflection, rapid scan-
ning, dual wavelength, and photometric titration measurements are available
with many instruments. One should also consider the initial cost, availability
of competent operators, adequate laboratory space, and proper maintenance
procedures.
Deviations from Beer's law occur from instrumental causes as well as
from chemical changes involving the measured species. The proper choice
of measuring instrument minimizes these deviations. Thus, when too wide
a slit width is used, an apparent deviation from Beer's law results. A slit
5. Colorimetry 111

width larger than the natural band width of the absorbance maximum being
measured will result in a low absorbance value. Narrow slit-width settings
are especially necessary for adherence to Beer's law when the absorbance
maxima have narrow natural band widths. This finite slit-width effect is
the main cause for differences between observed and true absorbance
values.
A second instrumental source of error occurs because of reflections of
the optical beam from the surfaces of the absorption cell. Internal reflection
causes the incident beam to traverse the sample solution more than once,
leading to a multiple-reflection-path effect. The beam which finally emerges
and proceeds to the detector will produce a higher absorbance reading than
that which would have been observed from a single pass through the sample
solution. The multiple-reflection-path effect is most pronounced when meas-
uring very dilute solutions having low absorbance values. The use of longer-
path-length cells help to minimize this source of error.
Stray polychromatic radiant energy which has been reflected and scat-
tered by the various components within the interior of the instrument may
be superimposed on a monochromatic beam. In measuring solutions of high
absorbance very little energy reaches the detector and the stray radiant
energy can constitute an appreciable fraction of the detector signal. In this
case, a lower absorbance value would be observed in the absence of stray
radiant energy. Stray radiant energy is minimized by employing a double-
monochromator system, a filter, and proper optical design. Additional in-
formation pertaining to instrumental causes of error is available. 7 ,48,87
Any spectrophotometer, if it is to provide meaningful, reproducible
data, must be calibrated and the calibration checked at various intervals.
Both wavelength and photometric accuracy are necessary and various cali-
bration procedures have been recommended. 39 ,49 Comparison and adjust-
ment of the wavelength-scale reading taken when measuring the emission
line of a mercury lamp at 546.1 m,u serve to check the wavelength scale.
Other emission lines from a hydrogen lamp and the location of absorbance
maxima of certain rare-earth oxides are also used. Absorbance comparison
and adjustment are made using standard absorbing glasses, perforated
screens, or chemical solutions with highly reproducible photometric char-
acteristics. Acidic copper sulfate, acidic cobalt ammonium sulfate, and basic
potassium chromate are used for this purpose. The references cited above
should be consulted for detailed calibration procedures.
112 Gordon A. Parker and D. F. Boltz

4. SAMPLE PREPARATION

4.1. Range of Optimum Concentration

The exact concentration range over which a constituent can be deter-

mined colorimetrically may be limited, depending primarily upon the par-
ticular constituent and chromogenic reagent being used. In general, solu-
tions more concentrated than 0.01 mole of desired constituent per liter
cannot be conveniently determined. Solutions of this concentration usually
have too large an absorbance for accurate measurement.133 In ascertaining
a lower limit of colorimetric detection one must consider the sensitivity of
the measuring instrument to small changes in absorbance. Sensitive instru-
ments can measure accurately differences of 0.001 absorbance. The light-
absorbing capacity of the desired constituent also determines the sensitivity
of a colorimetric method, as does the particle size. A larger cross sectional
area of the absorbing species will have a greater probability of interaction
with incident radiant energy, and the probability that after interaction an
electronic transition occurs within the absorber limits the absorptivity of
a species. These limitations set an upper value for the molar absorptivity
of the absorbing species. Generally, molar absorptivities do not exceed
10,000 liters per cm per mole, although values larger than this have been
reported in favorable cases. Assuming a molar absorptivity of 10,000, a
minimum absorbance reading of 0.001, and the use of l-cm cells, the lower
concentration limit will be 10-8 M.17.145 This value can be further reduced
if cells of longer path length are employed.
Concentration units commonly used in colorimetric measurements are
generally moles per liter or weight per unit volume, such as g per liter, g per
100 ml, mg per 100 ml, or ppm. The latter is equivalent to mg per liter or
f-lg per ml. Absorptivity determines the sensitivity of a colorimetric method.
The larger the absorptivity the smaller the amount of desired constituent
that can be detected. Another measure of detection limit is the sensitivity
value. It is defined as the number of f-lg of desired constituent per cm path
length of absorbing solution required to produce an absorbance reading
of 0.001.116 The smaller the sensitivity the smaller the amount of desired
constituent that can be detected.
Within the extremes of limiting concentration over which colorimetric
measurements are possible, there is a much narrower concentration region
which yields the most meaningful dataY69.86.1l7 The limits of this region,
the optimum-concentration range, are best found graphically by the method
of Ringbom. 3 111 A typical Ringbom plot is shown in Fig. 4. It is a plot of
5. Colorimetry 113

100r--------------,

80

60

40

20

LOGARITHM OF CONCENTRATION

(concentration In ppm)

Fig. 4. Ringbom plot.

percent transmittance against the logarithm of the concentration. The linear

portion of the plot, having the steepest slope, represents the region of
optimum concentration for a spectrophotometric determination of that
constituent.

4.2. Preparation of the Colored System

The preliminary treatment of a sample often requires dissolution by
fusion or acid treatment. Although details of recommended reagents and
techniques are beyond the scope of this chapter, it is essential that every
effort be made to avoid contamination, since colorimetric methods are
especially sensitive to traces of contaminants. It is advisable to always
perform a blank determination so that any trace contaminants in reagents
can be detected and compensated for. Likewise, it is convenient to dilute
the sample solution to a specific volume. In analyzing for one or more
constituents by colorimetric methods, it is very helpful to be able to take
smaller or larger aliquots from a sample stock solution.
Depending on the composition of the matrix of the sample, it may be
necessary to either isolate the desired constituent prior to development" of a
colored system or to remove some interfering substance. In analyzing inor-
ganic systems, ion-exchange, extractions, and electrodeposition methods are
usually the most applicable. Distillation, precipitation, and various chroma-
tographic methods are occasionally used.
Another approach to eliminate the deleterious effect of a specific ion
may be to resort to "masking"-a method of suppressing an undesired
114 Gordon A. Parker and D. F. Boltz

reaction by adding a complexing agent which preferentially reacts with the

interfering ion. EDT A may be used, for example, to eliminate the inter-
ference of certain metal ions in the colorimetric determination of phosphates.
In all cases the preparative treatment must be consistent for all samples and
standards to ensure reproducible analytical results.
Suspended particles and foreign matter cannot be tolerated in the test
solution. Ion-pair formation, polymerization, solvent association, and other
undesirable reactions must also be avoided, since they may cause changes in
both the location and magnitude of the absorbance maximum.
The final chemical treatment consists of adding a reagent which under-
goes a reaction with the desired constituent, producing a color proportional
to the concentration of this specific constituent. Color formation should be
rapid and the reaction stoichiometric. The resulting color must be stable,
reproducible, and very sensitive to variation in concentration of the desired
constituent. For these reasons the direct measurement of naturally colored
species is seldom used. A single reagent such as a complex ant, reductant, or
oxidant is often sufficient to develop a colored solution. Some color-forming
reactions depend on a combination of reagents or a systematic sequence of
reactions. It is advantageous to use high-purity, colorless reagents or those
giving a small, reproducible blank. The amount of each reagent used should
be measured accurately.
Both the rate of color development and ultimate intensity of the color
depend on the temperature and the pH value of the solution. At low tem-
peratures the rate of color formation is too slow, and at high temperatures
chemical decomposition of the colored species may occur. Absorbance
values may also be temperature dependent because of changes in solution
density and refractive index. 36
The stability of the colored system must be well known. Some colored
systems are stable for long periods of time, while other systems will increase
gradually in intensity and then remain relatively constant. Some colored
systems begin to fade after a limited period of stability, while a few used
systems exhibit a gradual but continuous change in absorptivity. The time
elapsed between preparation of a colored solution and photometric measure-
ment may be critical. A thorough knowledge of the specificity of the colori-
metric method or, more significantly, an understanding of what substances
interfere and the nature of their interference is essential,u
5. Colorimetry 115

5. PHOTOMETRIC MEASUREMENTS

5.1. Accuracy and Precision

Assuming, for the moment, that other sources contributing to the error
of a photometric determination are negligible, consider the effect of trans-
mittance error from a conventional photoelectric instrument in which the
error in transmittance LIT is constant over the entire transmittance range.
This error will cause a corresponding error LI C in the concentration obtained
from the photometric reading. If Beer's law is obeyed for the system under
consideration, the relative concentration error LlC/C is found by differen-
tiating Beer's law with respect to concentration. 5 ,58
LlC 0.434 LIT
(8)
C TlogT
The magnitude of the relative concentration error is determined by assigning
a value to the transmittance error suitable for the photometric instrument
used and evaluating the right-hand side of Eq. (8). A plot of these data is
shown in Fig. 5. The uncertainty associated with a concentration value at
the extremes of photometric measurements is large. The position of most
accurate transmittance measurement, smallest relative concentration error,
is found by setting equal to zero the derivative of Eq. (8) taken with respect
to concentration. This point occurs at a percent transmittance of 36.8 (0.434
absorbance). At this point the relative concentration error per unit photo-
metric error is at a minimum

( LlCjC) = -2.72%
LIP max

14
LIT ~ 1%
12
L1C
C
10

% TRANSMITTANCE

Fig. 5. Relative concentration error.

116 Gordon A. Parker and D. F. Boltz

The relative concentration error does not increase appreciably above its
minimum value over the range 20-60% transmittance (0.7-0.2 absorbance).
Hence, this region is recommended for photometric studies. l6 Transmittance
readings outside this range are possible provided the corresponding increase
in relative concentration error is tolerable. Measurements near the limiting
extremes of 0 and 100% transmittance should be avoided.

5.1. 1. Conventional Method

In conventional photoelectric measurements the lower limit of the trans-
mittance scale is set by adjusting the photometer readout to 0% T with the
detector protected from all possible light sources; the upper limit of the
transmittance scale is set by adjusting the photometer to read 100% Twith
only solvent in the reference cell which is placed in the light path. The 0% T
setting is referred to as the dark-current setting. Commonly, in place of
solvent transmittance at 100% T, a blank or reference solution containing
all constituents present in the sample but the desired constituent is used to
set the 100% T value. This practice has the advantage of cancelling possible
interferences from species other than the desired constituent. These inter-
ferences, present in both reference and sample solution, are effectively
cancelled by adjustment of the instrument to 100% T (zero absorbance)
with the blank solution in the reference cell placed in the light path. By this
technique only the absorbance of the desired constituent should be obtained.
Thus, after the dark-current and 100% Tsettings have been made, the sample
solution is placed in the sample cell located in the light path of the instru-
ment and the percent transmittance or absorbance measured.
In an effort to improve the precision of photometric measurement, the
conventional method of measurement has been modified. Various modifica-
tions are possible and the choice of procedure depends upon whether the
sample of interest is more concentrated or more dilute than necessary to
give ordinary transmittance readings in the range of optimum photometric
accuracy. These methods, called precision or differential methods, allow
measurements outside the ordinary range of photometric reliability and with
a higher precision than is possible by the conventional method. lOo ,lo9,l36
Precision photometric measurements are accomplished by judiciously fixing
one or both ends of the transmittance-scale solutions containing specific
concentration of the desired constituent. If the special reference solution is
identical to the sample solution in all respects except the concentration of
the desired constituent, both the compensation effects of a blank solution
and an increase in precision will result.
5. Colorimetry 117

5.1.2. Transmittance-Ratio Method

The transmittance-ratio method of precision photometry is applicable
to concentrated solutions of the desired constituent. With the conventional
photometric method and most spectrophotometers the transmittance values
observed for concentrated solutions would be too low for reliable measure-
ment. With the transmittance-ratio method the dark current, 0% T, is set
as in the conventional method, the detector in total darkness. The 100% T
setting is made with a solution slightly less concentrated with respect to the
desired constituent than the sample to be analyzed. This method has the
effect of expanding the readout scale (see Fig. 6a).

Ordinary
0 50 100 Photometric
Scale

I0
~: 50 100
Transmittance
Ratio
Scale

Ordinary
0 50 100 Photometric
Scale

~I
0 50 100
Trace
Analysis
Scale

z":ss;::
Ordinary
0 50 100 Photometric
Scale

Ultimate
Precision
0 50 100 Scale

Fig. 6. PrecisIOn photometry. (a) Transmittance-

ratio method: TB1 is determined by a standard slightly
less concentrated than the unknown (reading T",) and
is used to fix one limit of the expanded scale. Fivefold
expansion. (b) Trace-analysis method: TB is deter-
mined by a standard slightly more concentrated than
the unknown (reading T.) and is used to fix one limit
of the expanded scale. Fivefold expansion. (c) Ulti-
mate-precision method: TS1 is determined by a
standard slightly less concentrated than the unknown
(reading To:) and is used to fix one limit of the
expanded scale. Ts! is determined by a standard
slightly more concentrated than the unknown (reading
T",) and is used to fix the other limit of the expanded
scale. Fivefold expansion.
118 Gordon A. Parker and D. F. Boltz

5.1.3. Trace-Analysis Method

The trace-analysis method of precision photometry is applicable to very
dilute solutions of the desired constituent. With the conventional photo-
metric method the transmittance values observed in these cases would be
too high for reliable measurement. In the trace-analysis method the 100% T
limit is set in the ordinary way with a solvent or a reagent-blank solution
in the reference cell. The 0% T setting is made with a solution slightly more
concentrated in respect to the desired constituent than the sample solution.
This method has the effect of further expanding the photometer scale
(see Fig. 6b).

5.1.4. Ultimate-Precision Method

The ultimate-precision method of differential photometry is applicable
to those solutions where measurements of the highest reliability are desired.
With this method the limits of the instrument scale are set using two solu-
tions, one slightly more concentrated than the unknown sample and the
other slightly less concentrated. The 0% T limit is set with the slightly more
concentrated standard and the 100% T limit is made with the slightly less
concentrated standard (see Fig. 6c).
Precision methods have been applied to indirect and multicomponent
photometric methods of analysis. lOO ,13G Precision methods, although appli-
cable at concentration extremes and yielding more accurate results than
ordinary photometric methods, are not without disadvantages. The validity
of Beer's law cannot be assumed. Calibration graphs with known concen-
trations of the desired constituent are not always linear. Prior knowledge of
the approximate amount of desired constituent is necessary before known
standards of the proper concentration can be prepared. The preparation of
these standards is time consuming and the spectrophotometric determina-
tion may take considerably longer than when the conventional photometric
method of measurement is employed. Small differences in the sample cells
and in the positioning of the cells in the optical path become increasingly
significant in the precision methods.

5.2. Advantages
Colorimetric methods for the quantitative determination of trace con-
stituents in a sample satisfy many of the desirable requirements to be con-
sidered in choosing an analytical procedure. The sensitivity is generally
good, falling in the ppm range. Spectrophotometric methods are accurate
and provide, within the optimum absorbance region and concentration
5. Colorimetrv 119

limits described above, a reliable and precise method of analysis. With

suitable, sensitive instrumentation ordinary photometric measurements are
capable of a relative concentration precision of less than 1%. Even more
precise measurements are possible if one uses a differential photometric
method. Minor interferences and the interferences from solvent and other
reagents are eliminated by the proper choice of a blank solution for setting
the transmittance (absorbance) limits of the photometric instrument. If
many samples of similar composition are to be analyzed repetitively for the
same constituent, a calibration graph will be very useful for subsequent
determinations, provided the same procedure and the same instrument are
used. If a constituent is not colored itself, colorimetric analysis is still
possible provided a suitable reagent can be found which when reacted with
the desired constituent will produce a colored product. Simultaneous spectro-
photometric determinations of two or more constituents are possible. Major
constituents can be analyzed by use of the transmittance-ratio method and
minor trace constituents too small to be measured by the conventional
photometric method can be determined by the trace-analysis method. In-
strumentation for colorimetric analysis is generally available in most chem-
ical laboratories together with specific procedures.

5.3. Limitations
The accuracy and precision of a colorimetric analysis diminishes when
measurements are made outside the optimum absorbance and concentration
limits. It is possible to dilute a concentrated sample exhibiting too high ab-
sorbance (i.e., too low transmittance). Extreme dilution is not recommended
for the dilution errors generally become larger than the photometric errors.
Very dilute sample solutions can be concentrated. An ion-exchange tech-
nique, for example, can be used to concentrate a dilute solution. A large
volume of the sample solution is passed through an appropriate ion-exchange
column. When a sufficient amount of the desired constituent has been col-
lected on the column, it is eluted with a small volume of appropriate eluting
solution. Extraction of a colored species in a relatively small volume of an
immiscible organic solvent is another method of concentrating colored
systems. Some colored systems are not suitable for colorimetric measure-
ments. They may be unstable, exhibit a large temperature effect on absor-
bance, or may not adhere to Beer's law. A color-forming reagent may
form colored species with several constituents in the sample. The absor-
bance of other species in the sample may also interfere with that of the
desired constituent.
120 Gordon A. Parker and D. F. Boltz

Table II
Typical Spectrophotometric Methods for the Determination of Metals

Metal Reagent, or method Reference

Ag Dithizone 41, 115

p-Dimethylaminobenzylidenerhodanine 22, 115, 118, 122
Al Eriochrome cyanine R 57
Alizarinsulfonic acid (Alizarin Red S) 6, 115
Ammonium aurintricarboxylate (Aluminon) 115
Xylenol orange 19, 38, 90, 138
As Heteropoly blue 13, 91, 135
12-Molybdoarsenic acid 141
Au p-Diethylaminobenzylidenerhodanine i 14
Bromoauric acid 80
Rhodamine B 83
2,2'-Dipyridyl ketoxime 59
Be Quinalizarin 42
p- Nitrobenzeneazo-orcinol 29, 140
Bi Iodide 74, 81, 146
Dithizone 115
Ammonium I-pyrrolidinecarbodithioate 67
Thiourea 74, 94
Co Sodium I-nitroso-2-hydroxynaphthalene-
3,6-disulfonate (Nitroso-R-Salt) 105
I-Nitroso-2-naphthol 79, 93
Ammonium I-pyrrolidinecarbodithioate 67
Cr Chromate 26, 115
Peroxy-2,2'-bipyridine 103
1,5-Diphenylcarbohydrazide 106
Cu 2,9-Dimethyl-I, 10-phenanthroline 45, 126
(Neocuproine)
Sodium diethylaminecarbodithioate 60
Fe 1,10-Phenanthroline 44, 119
Thiocyanate 149
Mg Sodium methylbenzothiazole-(1,3)-
4,4' -diazoaminobenzene-(2,2'-
disulfonate (Titan Yellow) 50, 76
0,0'-Dihydroxyazobenzene 37
1,2-Hydroxy-3,2,4-dimethylcarboxyanilide
naphth-l-ylazo-2-hydroxybenzene (Magon) 1, 123
Mn Permanganate 78,99, 115, 147
Mo Thiocyanate 30, 51, 63
Ammonium-l-pyrrolidinecarbodithioate 67
4-Methyl-l ,2-dimercaptobenzene 53
5. Colorimetry 121

Table II (continued)

Metal Reagent, or method Reference

Ni Dimethylglyoxime 27, 98
Pb Dithizone 129
Sodium diethyldithiocarbamate 68
4-(2-Pyridylazo )-resorcinol 34
Sb Rhodamine B 107
Brilliant green 18, 130
Iodide 40, 81
Sn 4-Methyl-l ,2-dimercaptobenzene 28
Catechol violet 33
Ti Hydrogen peroxide 144
Disodium-l ,2-dihydroxybenzene-3, 5-1
disulfonate (Tiron) 152
U Hydrogen peroxide 112
Thiocyanate 32, 96
1-(2-Pyridylazo )-2-naphthol 4
V Tungstovanadophosphoric acid 124, 150
Hydrogen peroxide 137, 151
Molybdovanadophosphoric acid 64
W Thiocyanate 31, 47, 108
4-Methyl-l ,2-dimercaptobenzene 23, 89
Peroxytungstic acid 104
Zn Dithizone 115, 139

Although suitable colorimetric methods may not be available for the

determination of certain constituents, analytical information resulting from
the absorption of electromagnetic radiant energy may still be possible. Many
chemical systems which exhibit no color have definite and distinct absor-
bance maxima in the UV or near IR spectral regions. For certain elements
spectrometric, fiurophotometric, or atomic absorption spectrometric meth-
ods can be utilized advantageously.

6. APPLICATIONS

6.1. Determination of Metals

The spectrophotometric method for the determination of metals is one
of the most extensively used when trace quantities are to be determined,
because the technique is sensitive, specific, rapid, and the instrumentation
122 Gordon A. Parker and D. F. Boltz

required is relatively inexpensive. The nature of the matrix of the sample

material, the specific metal to be determined, and/or the number of metals
in the sample to be analyzed are factors to be considered before selecting
a particular colorimetric method.
The excellent treatises of Sandell,115 Snell and Snell,127 and Charlot 24
contain numerous colorimetric methods for each metal, and procedures for
specific applications to particular materials including minerals, ores, and
water. In Table II typical colorimetric methods for 22 common metals are
summarized. New reagents, improved methods, and recent applications may
be found in reviews on this methodology.14

Table III
Typical Spectrophotometric Methods for the Determination of Nonmetals

Nonmetal Reagent, or method References

B Carminic acid 11,20,54, 113

Quinalizarin ll, 82
1,1 '-Dianthrimide 11, 35
Curcumin ll, 52, 55, 131
F Zirconium--eriochrome cyanine R ll, 88
Zirconium-alizarin ll,n
Zirconium-SPADNS 9
N(NH.) or (NH. +) Indirect: Hypobromite 61
Pyridine-pyrazolone 71
Phenol-hypochlorite 84, 143
Nessler II
(NO.-) Phenoldisulfonic acid II
Brucine 97
NO.-) Griess-II 5, 110
UV: 4-Aminobenzenesulfonic acid 102
P (PO.-') Heteropoly blue 1l,13,77
Molybdovanadophosphoric acid 70
S (S-2) Methylene blue 43, 65
(SO,-2) Indirect: Barium chloranilate 10, 120
Methylene blue 65
Se 3,3'-Diaminobenzidene 25
Si Heteropoly blue 13, 101, 134
Molybdosilicic acid 21, 125
Te Iodotellurite 46, 66
Thiourea 95
5. Colorimetry 123

6.2. Determination of Nonmetals

Spectrophotometry is a method of choice in the determination of non-
metals. Several reference worksll.24.127 contain information of specific me-
thods and applications. In Table III typical colorimetric methods for eight
common nonmetals are quoted.

REFERENCES

1. Apple, R. F., and White, J. c., Talanta 8, 419 (1961).

2. ASTM Committee Manual on Recommended Practices in Spectrophotometry, Amer-
ican Society for Testing and Materials, Philadelphia, Pa., 1966, p. 1.
3. Ayres, G. H., Anal. Chem. 21, 652 (1949).
4. Baltisberger, R. J., Anal. Chem. 36, 2369 (1964).
5. Barnes, H., and Folkard, A. R., Analyst 76, 599 (1951).
6. Barton, C. J., Anal. Chem. 20, 1068 (1948).
7. Bauman, R. P., Absorption Spectroscopy, John Wiley & Sons, New York, 1962,
p. 149.
8. Bauman, R. P., ibid., p. 378.
9. Bellack, E., and Schoubee, P. J., Anal. Chem. 30, 2032 (1958).
10. Bertolacini, R. J., and Barney, J. E., Anal. Chem. 30, 202 (1958).
11. Boltz, D. F., ed., Colorimetric Determination of Nonmetals, Interscience, New York,
1958.
12. Boltz, D. F., Selected Topics in Modern Instrumental Analysis, Prentice-Hall, New
York, 1952, p. 149.
13. Boltz, D. F., and Mellon, M. G., Anal. Chem. 19, 873 (1947).
14. Boltz, D. F., and Mellon, M. G., Anal. Chem. 36, 256R (1964); 38, 317R (1966);
40, 255R (1968); 42, 152R (1970).
15. Boltz, D. F., and Schenk, G. H., in Handbook of Analytical Chemistry, (L. Meites,
ed.), McGraw-Hill, New York, 1963, p. 6-9.
16. Boltz, D. F., and Schenk, G. H., ibid., p. 6-16.
17. Braude, E. A., in Determination of Organic Structures by Physical Methods, Volume I,
(E. A. Braude, and E. C. Nachod, eds.), Academic Press, New York, 1955, p. 131.
18. Burke, R. W., and Menis, 0., Anal. Chem. 38, 1719 (1966).
19. Budesinsky, B., Zh. Anal. Khim. 18, 1071 (1963).
20. Callicoat, D., and Wolszon, J. D., Anal. Chem. 31, 1434 (1959).
21. Case, O. P., Ind. Eng. Chem., Anal. Ed. 16, 309 (1944).
22. Cave, G. C. B., and Hume, D. N., Anal. Chem. 24, 1503 (1952).
23. Chan, K. M., and Riley, J. P., Anal. Chem. 36, 220 (1966).
24. Charlot, G., Colorimetric Determination of Elements. Principles and Methods, Else-
vier, New York, 1964.
25. Cheng, K. L., Anal. Chem. 28, 1738 (1956).
26. Christow, S., Z. Anal. Chem. 125, 278 (1943).
27. Claassen, A., and Bastings, L.. Analyst 91, 725 (1966).
28. Clark, R. E. D., Analyst 61, 242 (1936); 62, 661 (1937).
29. Covington, L. c., and Miles, M. J., Anal. Chem. 28, 1728 (1956).
124 Gordon A. Parker and D. F. Boltz

30. Crouthamel, C. E., and Johnson, C. E., Anal. Chern. 26, 1284 (1954).
31. Crouthamel, C. E., and Johnson, C. E., Anal. Chern. 26, 1288 (1954).
32. Currah, J. E., and Beamish, F. E., Anal. Chern. 19,609 (1947).
33. Dagnall, R. M., West, T. S., and Young, P., Analyst 92,27 (1967).
34. Dagnall, R. M., West, T. S., and Young, P., Talanta 12, 583, 589 (1965).
35. Danielsson, L., Talanta 3, 138, 203 (1959).
36. Delahay, P., Instrurnental Analysis, Macmillan, New York, 1957, p. 208.
37. Diehl, H., Olsen, R., Spielholtz, G. E., and Jensen, R., Anal. Chern. 35, 1144 (1963).
38. Dvorak, J., and Nyvltova, E., Mikrochirn. Acta 1966, 1082.
39. Edisbury, J. P., Practical Hints on Absorption Spectrornetry, Plenum Press, New
York, 1967.
40. Elkind, A., Gayer, K. H., and Boltz, D. F., Anal. Chern. 25, 1744 (1953).
41. Fischer, H., Leopoldi, G., and Vol Uslar, H., Z. Anal. Chern. 101, 1 (1935).
42. Fischer, W., and Wernet, J., Angew. Chern. A60, 729 (1948).
43. Fogo, J. K., and Popowsky, M., Anal. Chern. 21, 732 (1949).
44. Fortune, W. B., and Mellon, M. G., Ind. Eng. Chern., Anal. Ed. 10, 60 (1938).
45. Gahler, A. R., Anal. Chern. 26, 577 (1954).
46. Geiersberger, K., and Durst, A., Z. Anal. Chern. 135, 11 (1952).
47. Gentry, C. H. R., and Sherrington, L. G., Analyst 73, 57 (1948).
48. Gibson, K. S., in Analytical Absorption Spectroscopy, (M. G. Mellon, ed.), John
Wiley & Sons, New York, 1950, p. 244.
49. Gibson, K. S., ibid., p. 256.
50. Gillam, W. S., Ind. Eng. Chern., Anal. Ed., 13, 499 (1941).
51. Grimaldi, F. S., and Wells, R. C., Ind. Eng. Chern., Anal. Ed. 15, 315 (1943).
52. Grinstead, R. R., and Snider, S., Analyst 92, 532 (1967).
53. Hamence, J. H., Analyst 65, 152 (1940).
54. Hatcher, J. T., and Wilcox, L. V., Anal. Chern. 22, 567 (1950).
55. Hayes, M. R., and Metcalfe, J., Analyst 87, 956 (1962).
56. Headridge, J. B., Photornetric Titrations, Pergamon Press, New York, 1961.
57. Hill, U. T., Anal. Chern. 28, 1419 (1956).
58. Hiskey, C. T., Anal. Chern. 21, 1440 (1949).
59. Holland, W. J., and Bozic, J., Anal. Chern. 39, 109 (1967).
60. Howard, J. M., and Spauschus, H. 0., Anal. Chern. 35, 1016 (1963).
61. Howell, J. A., and Boltz, D. F., Anal. Chern. 36, 1799 (1964).
62. Hughes, H. K., Anal. Chern. 24, 1349 (1952).
63. Hurd, L. C., and Allen, H. 0., Ind. Eng. Chern., Anal. Ed. 7, 396 (1935).
64. Jakubiec, R., and Boltz, D. F., Anal. Chirn. Acta 43, 137 (1968).
65. Johnson, C. M., and Nishita, H., Anal. Chern. 24, 736 (1952).
66. Johnson, R. A., and Kwan, F. P., Anal. Chern. 23, 651 (1951).
67. Kalt, M. B., and Boltz, D. F., Anal. Chern. 40, 1086 (1968).
68. Keil, R., Z. Anal. Chern. 229, 117 (1967).
69. Kirkbright, G. F., Talanta 13, 1 (1966).
70. Kitson, R. E., and Mellon, M. G., Ind. Eng. Chern., Anal. Ed. 16, 379 (1944).
71. Kruse, J. M., and Mellon, M. G., Anal. Chern. 25, 1188 (1953).
72. Lamar, W. L., Ind. Eng. Chern., Anal. Ed. 17, 148 (1945).
73. Lewin, S. Z., J. Chern. Educ. 37, A401, A507, A637 (1960).
74. Lisicki, N. M., and Boltz, D. F., Anal. Chern. 27, 1722 (1955).
75. Lothe, J. J., Anal. Chern. 27, 1546 (1955).
5. Colorimetry 125

76. Ludwig, E. E., and Johnson, C. R., Ind. Eng. Chern., Anal. Ed. 14, 895 (1942).
77. Lueck, C. H., and Boltz, D. F., Anal. Chern. 28, 1168 (1956).
78. Luke, C. L., Anal. Chirn. Acta 34, 302 (1966).
79. Lundquist, R., Markle, G. E., and Boltz, D. F., Anal. Chern. 27, 1731 (1955).
SO. McBryde, W. A., and Yoe, J. H., Anal. Chern. 20, 1094 (1948).
S1. McChesney, E. W., Ind. Eng. Chern., Anal. Ed. 18, 146 (1946).
82. MacDougal, D., and Biggs, D. A., Anal. Chern. 24, 566 (1952).
83. MacNulty, B. J., and Woollard, L. D., Anal. Chirn. Acta 13, 154 (1955).
84. Mann, L. T., Jr., Anal. Chern. 35, 2179 (1963).
85. Meehan, E. J., in Treatise on Analytical Chernistry, (I. M. Kolthoff, and P. J. Elving,
eds.), Interscience, New York 1964, Part I, Vol. 5, p. 2755.
86. Meehan, E. J., ibid., p. 2758.
87. Meehan, E. J., ibid., p. 2765.
88. Megregian, S., Anal. Chern. 26, 1161 (1954).
89. Miller, C. C., Analyst 69, 109 (1944).
90. Miyajima, T., Bunseki Kagaku 13, 1042 (1964).
91. Morris, H. J., and Calvery, H. 0., Ind. Eng. Chern., Anal. Ed. 9, 447 (1937).
92. Muller, R. H., Anal. Chern. 40 (6), 109A (1968).
93. Nichol, W. E., Can. J. Chern. 31, 145 (1953).
94. Nielsch, W., and Boltz, G., Z. Anal. Chern. 142, 321 (1954).
95. Nielsch, W., and Giefer, L., Z. Anal. Chern. 145, 347 (1955); 155, 401 (1957).
96. Nietzel, O. A., and DeSesa, M. A., Anal. Chern. 29, 756 (1957).
97. Noll, c. A., Ind. Eng. Chern., Anal. Ed. 17, 426 (1945).
9S. Norwitz, G., and Gordon, H., Anal. Chern. 37, 417 (1965).
99. Nydahl, F., Anal. Chirn. Acta 3, 144 (1949).
100. O'Laughlin, J. W., and Banks, C. V., in The Encyclopedia of Spectroscopy, (G. L.
Clark, ed.), Reinhold, New York, 1960, pp. 19-33.
101. Pakalns, P., and Flynn, W. W., Anal. Chirn. Acta 38, 403 (1967).
102. Pappenhagen, J. M., and Mellon, M. G., Anal. Chern. 25, 341 (1953).
103. Parker, G. A., and Boltz, D. F., Anal. Chern. 30, 420 (196S).
104. Parker, G. A., and Boltz, D. F., Anal. Letters 1, 679 (1968).
105. Pascual, J. N., Shipman, W. H., and Simon, W., Anal. Chern. 25, 1830 (1953).
106. Pilkington, E. S., and Smith, P. R., Anal. Chirn. Acta 39, 321 (1967).
107. Ramette, R. W., and Sandell, E. B., Anal. Chirn. Acta 13, 455 (1955).
108. Reef, B., and Doge, H. G., Talanta 14, 967 (1967).
109. Reilley, C. N., and Crawford, C. M., Anal. Chern. 27, 716 (1955).
110. Rider, B. F., and Mellon, M. G., Ind. Eng. Chern., Anal. Ed., 18, 96 (1946).
111. Ringbom, A., Z. Anal. Chern. 115, 332 (1939).
112. Rodden, C. J., ed., Analytical Chern is try of Manhattan Project, Chap. 1 by C. J.
Rodden and J. C. Warf, McGraw-Hili, New York, 1950.
113. Ross, W. J., and White, J. c., Talanta 3, 311 (1960).
114. Sandell, E. B., Anal. Chern. 20, 253 (1948).
115. Sandell, E. B., Colorirnetric Deterrnination of Traces of Metals, Interscience,
New York, 1959.
116. Sandell, E. B., ibid., p. 80.
117. Sandell, E. B., ibid., p. 105.
118. Sandell, E. B., and Neumayer, J. J., Anal. Chern. 23, 1863 (1951).
119. Saywell, L. G., and Cunningham, B. B., Ind. Eng. Chern., Anal. Ed. 9, 67 (1937).
126 Gordon A. Parker and D. F. Boltz

120. Schafer, H. N. S., Anal. Chem. 39, 1719 (1967).

121. Schilt, A. A., and Jaselskis, B., in Treatise on Analytical ChemiMry, (I. M. Kolthoff,
and P. J. Elving, eds.), Interscience, New York, 1964, Part I, Vol. 5, Chap. 58.
122. Schoonover, I. c., J. Res. Natl. Bur. Standards 15, 377 (1935).
123. Shafran, I. G., and Solover, E. A., Tr. Vses. Nauch.-Issled. Inst. Khim. Reaktivov
Osobo Chist Khim Veshihestv. 28, 96 (1966).
124. Sherwood, R. M., and Chapman, F. W., Anal. Chem. 27, 88 (1955).
125. Shink, D. R., Anal. Chem. 37, 764 (1965).
126. Smith, G. F., and McCurdy, W. H., Anal. Chem. 24, 371 (1952).
127. Snell, F. D., and Snell, C. T., Colorimetric Methods of Analysis, 3rd Ed., Van No-
strand, New York, 1959, Vols. II and IIA.
128. Snell, F. D., and Snell, C. T., in Encyclopedia of Industrial Chemical Analysis, (F. D.
Snell, and C. L. Hilton, eds.), Interscience, New York, 1966, Vol. I, pp. 358-369.
129. Snyder, L. J., Anal. Chem. 19, 684 (1947).
130. Soldatova, L. A., Kilina, Z. G., and Kataev, G. A., Zh. Anal. Khim. 19, 1267 (1964).
131. Spicer, G. S., and Strickland, J. D. H., Anal. Chim. Acta 15, 231 (1958).
132. Stearns, E. I., in Analytical Absorption Spectroscopy, (M. G. Mellon, ed.), John
Wiley & Sons, New York, 1950, Chap. 7.
133. Strobel, H. A., Chemical Instrumentation, Addison-Wesley, Reading, Mass., 1960,
p. 153.
134. Sturton, J. M., Anal. Chim. Acta 32, 394 (1965).
135. Sutzaberger, J. A., Ind. Eng. Chem., Anal. Ed. 15, 408 (1943).
136. Svehla, G., Talanta 13, 641 (1966).
137. Telep, G., and Boltz, D. F., Anal. Chem. 23, 901 (1951).
138. Tikhonov, V. N., Zh. Anal. Khim. 20, 941 (1965).
139. Vallee, B. L., Anal. Chem. 26, 914 (1954).
140. Vinci, F. A., Anal. Chem. 25, 1580 (1953).
141. Wade1in, c., and Mellon, M. G., Analyst 77, 708 (1952).
142. Weast, R. C., ed., Handbook of Chemistry and Physics, 48th Ed. The Chemical
Rubber Co., Cleveland, Ohio, 1967, pp. E-165-E-181.
143. Weatherburn, M. W., Anal. Chem. 39, 971 (1967).
144. Weissler, A., Ind. Eng. Chem., Anal. Ed. 17, 695 (1945).
145. West, T. S., Analyst 91, 69 (1966).
146. Wiegand, C. S. W., Lann, G. H., and Kalich, F. V., Ind. Eng. Chem., Anal. Ed.
13, 912 (1941).
147. Willard, H. H., and Greathouse, L. R., J. Am. Chem. Soc. 39, 2366 (1917).
148. Willard, H. H., Merritt, L. L., and Dean, J. A., Instrumental Methods of Analysis,
Van Nostrand, Princeton, N. J., 1965, Chap. 3.
149. Woods, J. T., and Mellon, M. G., Ind. Eng. Chem., Anal. Ed. 13, 551 (1941).
150. Wright, E. R., and Mellon, M. G., Ind. Eng. Chem., Anal. Ed. 9, 251 (1937).
151. Wright, E. R., and Mellon, M. G., Ind. Eng. Chem., Anal. Ed. 9, 375 (1937).
152. Yoe, J. H., and Arinstrong, A. R., Anal. Chem. 19, 100 (1947).
Chapter 6

INFRARED SPECTROPHOTOMETRY

w. M. Tuddenham and J. D. Stephens

Kennecott Copper Corporation
Salt Lake City, Utah

1. INTRODUCTION

The wavelengths at which radiation is absorbed or emitted by minerals in

the IR region can be related to the interatomic vibrations in the molecules
or crystals. Infrared measurements, therefore, have definite theoretical sig-
nificance. The widest application to date, however, has been empirical.
The very fact that vibrations within molecules of a sample are related to
the frequencies absorbed results in no two minerals giving exactly the same
pattern when transmission of radiation is plotted against wavelength. Not
only does each member of a mixture of minerals give a distinctly character-
istic pattern, but the features of each mineral are seen in the absorption
curve of the mixture. Furthermore, the depth of an individual absorption
band can be related to the concentration of the material responsible for it.
The method is thus suited for quantitative work. By combining quantitative
and qualitative capabilities, IR spectrometry provides a powerful tool for
crystal chemistry studies.

1.1. Historical

Developments within the past 20 years have made possible the addition
of IR analysis to the list of instrumental techniques which may be used
routinely in the analysis of minerals. A review of the work of Coblentz!
published in 1906 emphasizes that the primary barrier to earlier recognition

127
128 W. M. Tuddenham and J. D. Stephens

of IR as an important geochemical tool was the result of inadequate instru-

mentation rather than lack of demonstrated capability of the technique as a
tool in studying minerals. Lack of a really acceptable sample preparation
technique further delayed wide application of the method. Coblentz utilized
thin sections in most of his work and was consequently faced with difficulty
in sample preparation and limited in the resolution available in the areas
of high absorption.
A review of all mineral-related IR studies from the time of Coblentz
to 1949 indicates an average ofless than one paper per year. A marked rise
in the number of such papers published commenced in 1950 and has con-
tinued through the present. A good analysis of these publication trends is
presented in a critical bibliography by Lyon 2 that covers published work
through 1962.
One of the most significant papers describing application of IR tech-
niques to the solution of geochemical and mineralogical problems was
published in 1950 by Hunt and coworkers.3 This paper was followed in
1952 by a catalog of IR spectra published by Miller and Wilkins' and a
significant paper on the IR absorption spectra of silicate minerals by
Launer. 5

1.2. Theory
A thorough coverage of the theoretical principles of IR absorption or
emission is beyond the scope of this study. For this reason, reference to the
literature is recommended to those interested in a theoretical background.
Williams 6 has presented the fundamentals of infrared in a readily under-
stood manner. A more sophisticated treatment is found in Herzberg's classic
work Infrared and Raman Spectra7 and Nakamot08 has given particular
attention to the principles as they apply to inorganic compounds specifically.

1.3. Instrumentation
Modern IR instrumentation has advanced substantially since the time
Coblentz made point-by-point measurements utilizing a rock-salt prism for
energy dispersion and a radiometer as a detector. The increased convenience
and reliability of commercially available instruments have been the critical
factors in developing IR as a major analytical tool. Herscher9 has given
good coverage of the subject of spectrometer development. He also com-
pares available systems and lists their individual advantages and disadvan-
tages. The reader is further referred to instrument manufacturers for informa-
tion on the most recent developments in commercially available instruments.
6. Infrared Spectrophotometry 129

Until recently the majority of instruments used rock-salt prisms which

had a wavelength range of about 2-15 fl (5000-650 em-I). This range was
extended in some instances by means of prisms of potassium bromide to
25 fl, cesium bromide to 38 fl, or thallous bromide-iodide (KRS-5) to 40 fl.
The energy dispersion was sometimes improved in the short-wavelength
regions by the use of lithium fluoride to 6 fl, or with calcium fluoride
to 9 fl.
With the availability of high-quality gratings, grating and prism-grating
instruments came into use, and at present these allow studies out to over
700 fl (14 em-I). The grating instrument has the inherent advantage over
the prism instrument of giving better resolution. This, however, is not con-
sidered to be too critical in the case of mineral studies because of the rather
different nature of the absorption bands of minerals as compared to those
of organic compounds.
A new approach, multiple-scan interferometry, uses a combination of
the Michaelson interferometer with a computer. This instrumentation prom-
ises further advances in the use of IR for mineralogical examinations.lO
This is especially true in situations where location and sampling are insoluble
by other means. The interferometer technique allows studies at wavelengths
normally considered impractical for conventional spectrometer.
For the geochemist or mineralogist, equipment cost must be taken into
consideration as well as versatility. Prism instruments are disappearing from
the scene, although a number of relatively inexpensive prism instruments
covering the 2.5-15 fl range are still available. Instrument costs cover a
wide range and are dependent upon the degree of sophistication. Limited-
spectral-range instruments sell for as little as $ 3000. Highly sophisticated
grating and prism-grating instruments are available at prices up to $ 30,000,
while a complete multiple-scan interferometer system having its own analog
computer could cost over $ 40,000. The equipment utilized to prepare the
IR curves in this chapter cost about $ 6400.

2. TECHNIQUES OF SAMPLE PRESENTATION

2.1. Absorption

The primary technique available for mineral studies involves absorp-

tion of IR radiation by the sample. Important techniques developed for the
preparation and presentation of samples for IR analysis by absorption mea-
surement include the following.
130 W. M. Tuddenham and J. D. Stephens

2.1.1. Thin Sections

The sample is ground and polished to a thin section following conven-
tional preparative techniquesY Because of the low IR transmission of most
minerals and rocks, sample preparation and control of thickness with this
technique is difficult. This method was used in early studies but is little
used at present because of the difficulties involved.

2.1.2. Oil Mulls

The finely ground sample is dispersed in mineral oil or other viscous
liquid with a high degree of transparency in the areas of interest. The
resulting slurry is sandwiched between two rock-salt or potassium bromide
windows. The technique is applied easily but suffers because of difficulty
in quantization. Interference from absorption bands present in the dispersing
liquids adds to the problems. Another inherent difficulty is related to non-
matching of refractive indices of the oil and the samples. This causes radia-
tion scattering particularly in the 2-4 f-l region.

2. 1.3. Deposited Powder

In this technique, described by Hunt et ai.,3 the ground sample is
dispersed in water and allowed to settle to separate coarse particles. The
supernatant slurry containing the fine material is removed and the water
is evaporated. The deposited powder is dispersed in a volatile, organic
liquid which is allowed to evaporate on an IR transparent plate. The IR
radiation is passed through the plate and the thin deposit. Quantization is
difficult and the technique primarily is restricted to qualitative measure-
ments. Mixtures of minerals may also undergo segregation during the sepa-
ration of the finely ground material.

2.1.4. Alkali Halide

Of all methods, the alkali halide disc technique has contributed the
most to the increased application oflR analyses to problems in geochemistry.
This technique was described independently by Stimson and O'Donnel,12
and by Schiedt and Reinwein,13 The technique involves mixing a small
amount of finely ground sample with high-purity potassium bromide or
other alkali halide, and then pressing the mixture in an evacuated die. Since
the potassium bromide has no absorption spectrum in the region of interest,
the resulting clear or translucent disc yields an absorption curve character-
istic of the sample. When using this technique one should be alert to possible
anion exchange between reactive minerals and the alkali halide.14 This effect
6. Infrared Spectrophotometry 131

is very evident with chalcanthite (CuS0 4 SH 20). The authors have ob-
served it in practice with very few other natural minerals. Details of the
recommended potassium bromide method are as follows. 15
Place in a mullite mortar 20 mg of sample prepared to pass through a
200-mesh screen. Add approximately 2 ml of ethyl alcohol and grind with
an automatic grinder for 15 min. Weigh out 2.5 mg of dry material after
scraping down the entire grinding area and add it to 1 g of high-purity
potassium bromide (IR quality). Blend the sample and the potassium bro-
mide thoroughly; this is conveniently accomplished by using a dentist's
amalgamator. The use of a plastic vial or ball in this operation may con-
taminate the sampleI6 and give extraneous absorptions. Prepare the sample
disc by weighing the appropriate amount of the blended mixture into a
vacuum die. The weight of material should be adjusted to yield a disc 1-
1.5 mm thick. Evacuate the die for 5 min and, while evacuated, apply a
pressure of 70 tons/in. 2 for 1 min. The sample thickness can be adjusted to
yield the best balance between details in areas of strongest absorption and
weaker absorption. Discs can be prepared without evacuation prior to
pressing, but these will not remain permanently clear.

2.2. Attenuated Total Reflection (ATR)

Attenuated total reflection, which was introduced in 1959 by Fahren-
fort,17 is a new approach to sampling which may in time prove to be of
considerable importance in the IR examination of minerals. A sample held
in close contact with a prism surface acts to absorb a portion of the radia-
tion striking the prism surface. The reflected spectrum is characteristic of
the sample. This technique has found considerable use in examining ma-
terials that normally are opaque to IR. Problems observed in utilizing this
approach are the difficulty of obtaining intimate contact of the sample with
the prism surface and an inability to get quantitative results with presently
available equipment. The use of a thin, silver chloride film between the
sample and prism has reportedly overcome some of the contact problems. Is
These problems will probably diminish with time and the literature should
be watched closely for developments in this field. HarrickI9 has described
the technique in his book Internal Reflection Spectroscopy.

2.3. Reflection
In contrast with A TR, reflection deals with specular or diffuse reflection
from polished or powdered samples. Specular reflection has been used to
some extent in studying polished surfaces,20-22 but generally is considered
132 W. M. Tuddenham and J. D. Stephens

to have limited usefulness in mineral examination. Low 23 made a qualitative

study of the IR reflection of small and rough specimens using a multiple-
scan interference spectrometer of the type discussed in Section 1.3. The
phenomenon studied in this case was not diffuse reflection but rather specular
reflection from very small areas. The success of the method, therefore, is
dependent upon the unique features of the interferometer, wherein multiple-
scan techniques can be used to strengthen the recorded spectrum.
The second technique which involves diffuse reflectance would require
the utilization of an integrating sphere or its equivalent. 24 Lyon and Burns 25
reported some success in obtaining spectra from powders and rough surfaces
using this approach. Equipment to perform this type of study has also been
suggested by manufacturers of interferometric apparatus. The technique
appears to be restricted to qualitative spectra.

2.4. Emission Spectroscopy

Emission as well as absorption of radiation by a molecule occurs when
there is a change in its dipole moment during a vibration period. To the
earthbound geochemist or mineralogist, emission IR spectroscopy is of
minor importance. The technique has been used extensively by astronomers
and its utilization in space exploration is of importance. A paper by Lyon
and Burns 25 covers emittance from rough and powdered samples.

3. QUANTITATIVE APPLICATIONS IN GEOCHEMISTRY

Approaches to quantitative mineralogy have been made in the past
utilizing optical and chemical techniques, but these techniques are slow
and yield incomplete results. Infrared spectrometry has added a new di-
mension to quantitative mineralogy. The results obtainable by this technique
are competitive with more laborious methods and generally the results are
better than any simple method devised thus far. Infrared analysis is con-
sidered to be complementary to x-ray diffraction, differential thermal ana-
lysis, and optical examination in the solution of mineralogical problems. 26
Some specific advantages of the IR technique may be summarized as
follows.
(a) Relatively inexpensive instrumentation is utilized.
(b) The technique is adaptable to the analyses of small samples.
(c) Both chemical and crystallographic data may be obtained.
(d) Infrared techniques have a high sensitivity for many economically
important minerals.
(e) Both qualitative and quantitative data may be obtained.
6. Infrared Spectrophotometry 133

Some disadvantages are as follows.

(a) Particle size and optical effects can influence spectra.
(b) Some important minerals are opaque in readily accessible regions
of the IR spectrum.
(c) Some important minerals are difficult to determine at low con-
centration levels.
(d) Special techniques must be used with reactive minerals.
Development of alkali halide disc sample preparation was a major step
in quantization of mineralogical analyses by IR. With halide discs, effective
sample thickness and concentration can be controlled so that correlation
of mineral concentration and strength of absorption bands is possible.
Accurate estimation of the contribution of a component to the total
absorbance at a diagnostic wavelength is critical in quantitative work. The
"baseline" technique is a useful approach for analyzing mineral mixtures.
A line is drawn tangentially to the curve and as close as possible to
points of minimum absorbance immediately adjacent to the diagnostic
wavelength.
The absorbance measured on this line at the diagnostic wavelength is
then subtracted from the total absorbance of the band being measured.
This method corrects for radiation scattering and attenuation from causes
other than absorption by the species being measured. These "baseline"
absorption values have been found to be additive in most instances. When
substantially all of the contribution to an absorption band is from a single
component, calibration curves can be prepared directly using baseline ab-
sorbancy values. When more than one component substantially contributes
to the wavelength being measured, the value may be related to absorption
values for bands characteristic of the contributing species to develop
simultaneous equations from which concentrations of the components can
be determined. The strengths and weaknesses of the method have been
discussed in some detail by Kendall et al. 27
Lyon and coworkers 28 demonstrated the capability of the technique
by analyzing a standard granite sample. They showed that by this method
the mineralogy of a sample could be determined quantitatively, regardless
of the alteration the sample had undergone. It was not possible to determine
rock texture and thereby deduce the full petrographic history of a sample.
The present composition of the rock could, however, be resolved, and the
method was shown to be sensitive to minerals formed by alteration of the
initial mineral constituents. The IR method is useful in determining the
carbonate content of phosphate samples15 and can be utilized in the deter-
134 W. M. Tuddenham and J. D. Stephens

mination of trace quantities of anions, such as sulfate in ground-water

systems. 29 ,30
The most critical problem encountered in quantitative analyses of
minerals by IR techniques is the particle size of the sample to be analyzed.
This is illustrated by studies with quartz and clay minerals. 15 Experimental
results with quartz relating absorption at 12.5, 12.8, and 14.4 fl to particle
size are shown in Fig. 1. Working with a quartz plate, Stein 3i found that the
transmission minimum from the ordinary ray occurs at 12.4 fl. The mini-
mum for the extraordinary ray occurs at 12.8 fl. In Fig. 1 it is shown that
the effect of the ordinary ray becomes more marked as particle size de-
creases.
Using a theoretical approach, Phillippi 3ia developed the proposition
that purely optical processes are responsible for a significant fraction of the
inconsistencies in the IR absorption spectra of powdered solid materials

w
u
z
c(
CD
0::
o
(f)
CD
c(

12 13 14 15
WAVE LENGTH,MICRONS
Fig. 1. Effect of particle size on IR absorption curve of quartz.
Reprinted by permission from Anal. Chern. 32 (12), 163 (1960).
6. Infrared Spectrophotometry 135

suspended in potassium bromide or Nujol. He showed that for certain

classes of materials scattering reflection losses at the surfaces of the particles
dominate the form of their powder spectra. Where such factors become
important, quantitative estimation and identification by group frequency,
as described in Section 4 would be compromised. Fortunately, these incon-
sistencies have done little to reduce the usefulness of the IR method in
mineralogical studies.
Grinding experiments with clay minerals have shown that the struc-
tures of some minerals can be disrupted enough to drastically alter the IR
curves. This structural breakdown can be avoided by grinding under ethyl
alcohol or some other suitable liquid. This is demonstrated for kaolinite
by the IR and x-ray diffraction curves in Fig. 2.15 In order to do repro-
ducible work with minerals it is necessary to standardize grinding proce-
dures. Automatic grinders allow the control of grinding pressure and time,
and are superior to hand grinding in most cases.
The sensitivity of minerals to IR analysis depends upon the strength
of their individual absorption bands and freedom from interference by

2 4 6 8 10 12 14
MICRONS

XRD-A

X RD-B-~"'--""----

---- 2 e- DEGREES

Fig. 2. Comparison of IR and X-ray diffraction curves of

wet (A) and dry (B) ground kaolinite. Reprinted by per-
mission from Anal. Chern. 32 (12), 163 (1960).
136 W. M. Tuddenham and J. D. Stephens

associated species. Sulfate and carbonate minerals can often be determined

to less than 1% in their natural environment. Clay minerals such as mont-
morillonite are more difficult to determine and would be limited to a lower
limit of 15% because of compositional variations and interference by other
minerals. The sulfides and many oxides cannot be determined in the region
from 2 to 15 p,. At longer wavelengths they show relatively low sensitivity
because of the broad, shallow absorption bands obtained. Precision of the
method is related to the above limitations as well as to the instrumental
problems involved.

4. MINERAL IDENTI FICATION WITH IN FRARED

Infrared absorption spectra have been characterized as "fingerprints"

of molecules and can be utilized in much the same manner for identification.
The necessity of compiling personal sets of standard spectra has probably
contributed to slow adoption of IR analysis for mineral studies. Perhaps
the most comprehensive set of mineral spectra published to date is that of
Moenke. 32 Compilation of spectral data for minerals in an easily used form,
such as is available for x-ray diffraction work, would be most useful.
The unique nature of the IR spectrum of any mineral species is of
prime value. The direct relation of the IR spectrum to the chemistry of
minerals is also important. Spectral structure correlations have been con-
sidered in some detail for inorganic compounds by Lawson 33 and Naka-
moto. s With the availability of high-quality grating instruments and Fourier
spectrometry, researchers have been able to extend inorganic spectra correla-
tion studies into the far IR region. 34
The compositions of principal complex anions or structural groups
generally relate to the major IR absorption bands. The ionic radius and
coordination number of the positive member of the complex anion and
the mass and charge density of the nearest neighbors to the anions or
structural groups directly affect the spectra. For anyone anion or group,
the wavelengths of the absorption bands are controlled by the mass of
associated cations. The wavelengths of absorptions due to different struc-
tural groups may overlap.
Spectra of selected minerals from important mineral groups are pre-
sented in Figs. 3-15. The mineral groups are ordered according to their
principal complex anions or structural groups. Spectral correlations in
the 2.5-25-p, region are discussed below as they relate to these selected
minerals.
6. Infrared Spectrophotometry 137

4.1. Minerals Containing H20, OH-, and Hydrogen Bond

These minerals generally have a first significant absorption band in the

wavelength region between 2.7 and 3.5 fl.. When bonded H 20 groups are
present, a second set of significant absorption bands occurs between 5.8
and 6.5 fl..
Figure 3 illustrates examples of some of these minerals. The spectrum
at the top of the figure was obtained with brucite, Mg(OH)2' The first
major absorption band in this spectrum occurs at 2.75 fl. and is related to
the presence of the OH- groups in the mineral. Other major absorption
bands occur in this spectrum at 17.6 and 22.7 fl.. The cause of these bands
has not been determined, but absorptions in this wavelength region are
generally considered to be related to crystal-lattice vibrational frequencies.
Minor absorption bands at 2.9 and 6.2 fl. are related to water absorbed in
the potassium bromide mounting medium.
The lower two curves in Fig. 3 show the isostructural minerals goethite,
FeO(OH), and diaspore, AIO(OH). Infrared spectra for these minerals and
many of the clay minerals were investigated by Adler et al. 35 The sharp

BRUCITE
Mg(OHl2

GOETHITE
FeD OH

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 3. Infrared spectra of typical minerals containing H.O, OH-, and hydrogen bonding.
138 W. M. Tuddenham and J. D. Stephens

bands at 2.85 and 2.95 p, are probably related to the O-H stretching vibra-
tions while the broad band at 3.4 p, is related to the hydrogen bonding.
Simple metal-oxide hydrates are uncommon as mineral species. Because of
this rarity they are not shown in Fig. 3. Some excellent examples of the
spectral effects caused by the presence of bonded H 20 in silicate mineral
lattices, however, may be seen in Fig. 11. The IR spectra obtained with the
two zeolite minerals, chabazite and scolecite, are shown in this figure. Both
of these minerals show strong absorption bands near 3 and 6 p, which are
related to the presence of bonded H 20. The spectra obtained with these
minerals will be discussed in detail in the section relating to the tektosilicate
minerals.

4.2. Carbonate Minerals

All carbonate minerals have one or more absorption bands at or near
7 p,. A study of the IR spectra of aragonite and calcite was made by Adler
and Kerr. 37 These same authors studied the spectra of other carbonate
minerals in subsequent publications. 38 .39 An earlier study of the IR spectra
of carbonate minerals was made by Huang and Kerr.40
In Fig. 4 a series of IR spectra are shown which are obtained with
three different carbonate minerals. The first spectrum was obtained with
calcite, CaC03. Besides calcium, the simple carbonates of magnesium, iron,
manganese, cobalt, zinc, and cadmium also form minerals with the calcite
structure. In many cases there are solid solution systems between the min-
erals of these species and, as will be discussed in the section on crystal
chemistry, IR spectrometry can be most useful in the study of such solid
solution series.
The second curve shown in Fig. 4 was obtained with cerussite, PbC03.
All the absorption bands of cerussite are displaced to longer wavelengths
than the equivalent bands of calcite. A major cause of this shift in wave-
lengths of the absorption bands is the increased mass of the lead ion in
cerussite as compared with the calcium ion in calcite.
The bottom spectrum in Fig. 4 was obtained with hydromagnesite,
M&(OHMC03)33H20. Hydromagnesite is structurally more complex than
cerussite or calcite and this is reflected in the complexity of its IR spectrum.
The presence of OH- or H 20 groups in the mineral is confirmed by the
presence of absorption bands near 3 p" and the presence of H 20 groups is
confirmed by the shoulder which occurs at 6 p, on the major CO~- absorp-
tion band. The presence of a strong absorption band at about 3.4 p, suggests
that there is hydrogen bonding in the mineral.
6. Infrared Spectrophotometry 139

CALCITE
CaCO.

lJJ
o
Z
<t
III
It:
o
(/)
III
<t

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 4. Infrared spectra of selected carbonate minerals.

As is the case with all carbonates, the strongest band of hydromagne-

site is near 7 fl. This band is quite complex, which implies that the CO~
vibrations in hydromagnesite are modified by attached atoms or atom
groups.

4.3. Nitrate Minerals

All nitrate minerals have one or more absorption bands at or near
7.2 fl. Menzies 41 studied the absorption spectrum of the nitrate ion and
pointed out the similarity and relations between the spectra of the carbonate
and nitrate ions. The crystal structures of the nitrate minerals are closely
related to those of the carbonates. This relation is apparent when the IR
spectra of the simple nitrates are compared with those of the simple car-
bonates.
The two most common nitrate minerals are soda niter, NaNO a , and
niter, KN0 3 The IR spectra of these two minerals are shown in Fig. 5.
Soda niter is isostructural with calcite, and there are obvious similarities
140 w. M. Tuddenham and J. D. Stephens

SODA NITER ...,---

V
NoNO,

r- :---v.
r--~ r /l
III
(>
z

m
0:
o
If)
NITER
KNO,
~
V
fJ

/
~
,- i

r
m

r-it""
~

,;

/
3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGT H IN MICRONS
Fig. 5. Infrared spectra of selected nitrate minerals.

between its spectrum and that of calcite. Similarities between the IR spectra
of soda niter and niter are also apparent, although these two minerals are
not isostructural. While the IR spectra of soda niter and niter are similar
they are not identical. The band which occurs at 11.90 fl in soda niter has
been shifted to 12.05 fl in niter. This shift seems to be related to the greater
mass of the potassium ion when compared with the sodium ion in soda
niter.

4.4. Borate Minerals

The first major IR absorption band of borate minerals generally occurs
in the wavelength region between 7 and 8 fl. The BO~- anion is similar to
the C05- and N0 3- anions, but there are few other areas of similarity
between the borate minerals and the carbonate or nitrate minerals. Christ42
studied the borate minerals and proposed a structural classification for them
relating to the composition and shape of the complex anions and polyions.
A further study of the borates and borosilicates was made by Tennyson 43
who classified all known species according to their known or assumed
structures, as based on the BO~- triangle, the BO~- tetrahedron, or polyions
formed by linking these groups in various ways.
In Fig. 6 tracings of the IR spectra obtained with a number of borate
minerals are shown. The top spectrum was obtained with ludwigite,
(Mg, Fe2+)2Fe3+B05. The IR spectrum of ludwigite is atypical of the bo-
rates in its simplicity, and it is suspected that the structure of ludwigite is
markedly different from that of the other borate minerals.
6. Infrared Spectrophotometry 141

The remaining three spectra in the illustration are typical of the borate
minerals in the complexity of their IR spectra. The second spectrum in Fig. 6
was obtained with boracite, Mg3 B70 13Cl. Neither ludwigite nor boracite
have strong absorption bands near 3 fl, indicating that they are anhydrous.
The weak absorptions at 3 and 6 fl in the boracite spectrum are due to water
absorbed in the KBr mounting medium.
The third spectrum in Fig. 6 was obtained with szaibelyite,
Mg(B02)(OH). In contrast with ludwigite and boracite, the szaibelyite
spectrum does have a strong O-H stretching absorption band at about
2.70 fl. There is no sign of a band near 6.2 fl, confirming that this mineral

BORACITE
MO,B 7 0 13 CI

w
<.)
z
<[
III
0::
o SZAIBELYITE
en M g (BO.)(OH)
III
<[

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 6. Infrared spectra of selected borate minerals.

142 W. M. Tuddenham and J. D. Stephens

does contain OH- groups rather than H 20 groups. An unusual aspect of

the szaibelyite spectrum is a major absorption band occurring at 6.90 /1.
This is a shorter wavelength than is normal for the borate group minerals
and overlaps absorptions of the carbonate and nitrate minerals.
The final spectrum in Fig. 6 was obtained with colemanite, Ca2B60 n
5H20. This spectrum shows complex O-H stretching bands near 2.7 and
2.8 /1, and also a broad band at about 3.1 /1. The colemanite spectrum also
has two bands near 6 /1. These bands, coupled with the bands between 2.7
and 3.1 /1, indicate that bonded H 20 groups are present in the colemanite
crystal structure.

4.5. The Sulfate Group Minerals

The IR spectra of the sulfate minerals are characterized by one or
more major absorption bands in the wavelength region between 8.5 and 9 /1.
These absorption bands are related to vibrations of the SO~- complex anion.
The IR spectra of many of the sulfate minerals were studied by Omori
and Kerr44 and additional sulfate species were investigated by Adler and
Kerr. 45 These latter investigators related the IR spectral configuration to
the crystal chemistry of the sulfate species, and the manner in which this
chemistry modifies the vibrational character of the SO~- group.
In Fig. 7 the IR spectra of a number of typical sulfate minerals are
shown. The top two spectra in this figure were obtained with anhydrite,
CaS04, and anglesite, PbS0 4, respectively. These two minerals are crystal-
lographically dissimilar but their IR spectra show some similar character-
istics which are typical of simple anhydrous sulfates.
The third spectrum was obtained with jarosite, KFe a(S04MOH)6' A
very strong O-H stretching band occurs near 3/1. This band, coupled with
the absence of major bands near 6 /1, confirms the presence of OH- groups
in the mineral. The position and to a lesser extent the configuration of
absorption bands in the wavelength region from 8 to 25/1 may vary with
different jarosite samples. This results from substitutions of other ions for
potassium and iron in the mineral's crystal structure. All of the jarosite
group minerals, however, yield IR spectra that are obviously related.
The fourth spectrum in Fig. 7 is of hanksite, Na22K(S04MCOa)2Cl.
The mineral is rather uncommon but its spectrum illustrates how chemical
information can be obtained from IR studies.
The absence of O-H stretching or bending bands near 3 and 6 /1 con-
firms the fact that hanksite is anhydrous. The bands which occur at 6.85,
7, and ll.l /1 are caused by the presence of the C05- anion in the hank site
6. Infrared Spectrophotometry 143

I
ANHYDRITE
CaSO.
~
V
I--
r ( --
V-~ IV' / ~I
r-1/

~~vPI
ANGLESITE

-:'
rv V Ihf
I-Ir
JAROSITE

I~ (
v
K Fo,(SO.).(OH).

V' N~
<I
(IJ !--
J..-

~~I( 1\rv
HANKSITE
I
N QUK{S04)9(C 03)2C1
V
1\V
r
-y vr ~I

I
I 1\ 10\ (1
V
\
3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 7. Infrared spectra of selected sulfate minerals.

structure. The bands occurring between 8.4 and 9 fl and those occurring
near 16 fl are caused by the presence of the SO~- anion in the hanksite
structure.

4.6. The Silicate Minerals

With the exception of very high-pressure phases, all silicates contain
silicon tetrahedrally coordinated with oxygen. The tetrahedral groups may
occur as isolated units or they may share their corners to form pairs, rings,
or infinitely extending chains, sheets, or three-dimensional frameworks.
These different units form the basis of several structural classifications of
silicate minerals. One of the better classifications is that of Berry and
Mason,46 as reproduced in Table I.
144 W. M. Tuddenham and J. D. Stephens

Table I
Structural Classification of the Silicatesa

Classification Structural arrangement Si:O Examples

Nesosilicates Independent tetrahedra 1 :4 Forsterite, MgzSiO,

Sorosilicates Two tetrahedra sharing one oxygen 2:7 Hemimorphite,
Zn,Si.O.(OH). HaO
Cyclosilicates Closed rings of tetrahedra each shar-
ing two oxygens
Inosilicates Continuous single chains of tetra-
hedra each sharing two oxygens 1 :3 Enstatite, MgSiO.
Continuous double chains of tetra-
hedra sharing alternately two and 4:11 Anthophyllite,
three oxygens Mg. (Si,Ou)z(OH)z
Phyllosilicates Continuous sheets of tetrahedra each Talc, Mg.Si,Olo(OH).
sharing three oxygens 2:5 Phlogopite,
KMg.(AISi.01o)(OH)z
Tektosilicates Continuous framework of tetrahedra
each sharing all four oxygens 1 :2 Quatz. SiO.
Nepheline, NaAISiO,

a From Berry, L. G., and Mason, Brian, Mineralogy: Concepts, Descriptions, Determinations,
W. H. Freeman, San Francisco, 1959.

Aluminum may also fit in tetrahedral coordination with oxygen, and

partial substitution of aluminum for silicon often occurs in the silicates.
Since aluminum ions have a 3+ valence charge as compared with a 4+
valence charge for silicon, this substitution effectively increases the negative
charge of the silicate structural group. These charge unbalances must be
compensated for elsewhere in the crystal lattices of the minerals. Substitu-
tions of this type give rise to the aluminosilicate minerals.
The first major IR absorption bands related to vibrations of the tetra-
hedral SiOl- groups or linked silicate groups occur between 8.5 and 12 f-l.

4.6.1. Nesosilicate and Sorosilicate Minerals

The nesosilicates are minerals in which the complex anion is the isolated
SiOl- group. Important minerals in this group include the olivines and
garnets. Aluminum generally does not substitute for silicon in the neso-
silicates. Most nesosilicates have divalent or trivalent elements in the cation
positions. Nesosilicates containing major amounts of alkali elements or
hydroxyl groups are rare.
6. Infrared Spectrophotometry 145

The sorosilicates are minerals in which the structural unit is the Si20~
group. This structural group is formed by linking two SiO~- tetrahedra
together at adjacent corners, with both tetrahedra sharing one oxygen
between them.
In Fig. 8 the IR spectra obtained with a number of nesosilicate and
sorosilicate minerals are shown. The first three spectra in this figure are
nesosilicates and the last spectrum is a sorosilicate. The top spectrum is that
of zircon, ZrSi0 4 This is one of the simplest of the silicate minerals since
the single zirconium cation balances the valence charge of the single SiO~
tetrahedral anion.
The second spectrum in Fig. 8 was obtained with forsterite, Mg2Si04
This is also a relatively simple silicate since the two magnesium cations
balance the negative charge of the single SiO~- anion. Forsterite is a member
of a much larger group of minerals, the olivine group. The configuration

ZIRCON

PYIIOPE
... ! AI2 1814'3

AKERMANITE
C"3 81 2 07

4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS
Fig. 8. Infrared spectra of selected nesosilicate and sorosilicate minerals.
146 W. M. Tuddenahm and J. D. Stephens

of the forsterite spectrum in the lO-l2-,u region is typical for olivine-group

minerals. Specific absorption bands of the different olivine species will
occur at slightly different wavelengths. The slight absorption band at 7 ,u
in the forsterite spectrum indicates the presence of a trace of carbonate
impurity in the mineral sample.
The third spectrum in Fig. 8 was obtained with pyrope, one of the
garnet group minerals. In this mineral the negative charge of three isolated
SiO~- groups is balanced by three magnesium and two aluminum cations.
The configuration of the pyrope spectrum between 8 and 12,u is typical for
garnet group minerals. It is difficult to differentiate members of the spessar-
tite-almandite-pyrope family by these bands only. Systematic wavelength
shifts of bands in the l6-25-,u region makes differentiation of them possible.
More detailed studies of solid solution relations in the garnet group min-
erals have been made by Hunt et al.3 and by Lyon. 47 The weak absorption
bands at 8.5 and 9.2,u in the pyrope spectrum probably represent a trace
of quartz impurity in the mineral sample.
The final spectrum in Fig. 8 was obtained with akermanite, Ca3 Si20 7
Minerals which contain only the Si20~- group are relatively rare, but species
which contain both SiOt- groups and Si20~- groups are more common.
Some examples of these minerals are the epidote group and vesuvianite.
The spectra of minerals containing both nesosilicate and sorosilicate groups
do not necessarily resemble those of either group and each species must be
interpreted individually.

4.6.2. Inosilicate and Cyclosilicate Minerals

In Fig. 9 the tracings of the IR spectra obtained with four inosilicate
and cyclosilicate minerals are shown. The top spectrum was obtained with
wollastonite, CaSi03 , and the second spectrum is that of diopside,
CaMg(Si0 3 )2' Both wollastonite and diopside have single-chain structures.
Diopside belongs to the pyroxene group minerals while wollastonite belongs
to the pyroxenoid group minerals. The diopside spectrum is typical for the
monoclinic pyroxenes. The spectra of augite, hedenbergite, fassaite, pigeo-
nite, and clinoenstatite are quite similar to it and can be easily recognized
as members of the same mineral family. The wollastonite spectrum is typical
for the pyroxenoid group minerals, and rhodonite, bustamite, and fowlerite
can be recognized as members of this same group by similarities in their
spectra. Pectolite, HNaCa(Si0 3)3, is a hydrated mineral considered to be
structurally related to wollastonite. Distinct differences appear in the spectra
of the two minerals, but there is enough similarity to support the theory of
structural relationship.
6. Infrared Spectrophotometry 147

WOLLASTONITE
COSiO,

3 4 5 6 7 8 10 12 14 16 18 20 2.2 24
WAVELENGTH IN MICRONS

Fig. 9. Infrared spectra of selected inosilicate and cyc10silicate minerals.

The third spectrum in Fig. 9 was obtained with tremolite,

Ca2Mg 5 (Si 40 11 MOH)2' Structural OH- is known to be present in tremolite
and it is interesting that the presence of this OH- group is not indicated
by an absorption band near 3 /-t. The fact that this absorption band is not
evident probably relates to the orientation of the OH- group in the mineral's
crystal structure. A phenomenon of this type was noted by Bassett48 in the
IR spectrum of phlogopite, the magnesium mica.
The final spectrum in Fig. 9 was obtained with beryl, Be3A12Si6018'
The IR spectra obtained with beryl have been studied separately by Wicker-
sheim and Buchanan49 and by Plyusina. 50 Although the presence of bonded
water is not indicated in the beryl formula, the IR spectrum clearly shows
148 w. M. Tuddenham and J. D. Stephens

that such bonded water is present in the mineral. There are several O-H
stretching bands near 3.7 fl and several H 20 bending bands near 6 fl. Water
groups along with other gas molecules and large alkali ions are known to
occupy tube or channellike holes in the Si60}~- ring structure. 51 The nature
of the IR bands indicates that the water is bonded in the crystal lattice and
is not simply absorbed in the holes.

4.6.3. Phyl/osilicate Minerals

The phyllosilicates are minerals in which the silicate structure forms
an indefinitely extending sheet which has an average composition of Si 4 0to
In Fig. 10 tracings of the IR spectra are shown which are obtained with

PYROPHYLLITE
AI2 s~ 010 (OHI2

KAOLINITE
AI4 Si4 010 (OHle 4H20

w
(,)
z

!D
0: MUSCOVITE
o
(J)
KAI2 (AI Si 3 010)(OHI 2
!D

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 10. Infrared spectra of selected phyllosilicate minerals.

6. Infrared Spectrophotometry 149

three phyllosilicate minerals. The first two spectra were obtained with pyro-
phyllite, AI 2Si 4 0 lO (OH)2' and kaolinite, AI 4 Si 4 0 lO (OH)s, respectively. Both
of these minerals have the basic Si4 0to sheet structure. The final spectrum
was obtained with muscovite. In muscovite one-fourth of the silicon in the
Si 4 0to sheet has been replaced by tetrahedrally coordinated aluminum. The
most obvious effect of the substitution of aluminum for silicon in the sheet
structure is a decrease in the intensity of most of the absorption bands.
While pyrophyllite, kaolinite, and muscovite vary considerably in their
crystal structures, comparison of the spectra in Fig. 10 clearly shows a
basic similarity in their IR patterns. Although the spectra differ in detail,
the similarities between them are such that one would intuitively feel that
the minerals are related. The IR spectra of all of the kaolin group minerals
are examined in more detail in Fig. 18 and discussed in Section 5.

4.6.4. Tektosilicate Minerals

The tektosilicates are very common and important mineral species.
Included in this group are the silica minerals, the feldspars, the feldspathoids,
and the zeolites. It is impossible to obtain framework silicate structures with
compositions other than Si02 without changing the silicon-to-oxygen ratio.
The most common modification of the tektosilicate structure is obtained
by substituting aluminum for silicon in some of the SiO~- tetrahedra. This
gives the framework structure a negative charge which must be balanced
by cations in other positions in the crystal lattice.
The IR spectra obtained with many of the tektosilicate minerals were
studied by Milkey.52 Although stilbite was misidentified as natrolite and
natrolite was identified incorrectly as mesolite in Milkey's work, this is
fundamentally a good source of information about IR analyses of the tekto-
silicates. Additional information about IR analysis in studying solid solu-
tion relationships in the plagioclase feldspar minerals is found in the work
of Thompson and Wadsworth. 53 The IR spectra of many of the silica group
minerals have been published by Lyon. 47
In Fig. 11 the IR spectra obtained with a number of the tektosilicate
minerals are shown. The top spectrum was obtained with alpha quartz,
Si02. Most members of the silica group minerals, including alpha quartz,
alpha cristobalite, alpha tridymite, and coesite, have spectral characteristics
somewhat similar to each other. An exception to this general similarity is
seen in stishovite, which has an IR pattern totally unlike those of other
members of the silica group. This dissimilarity of the stishovite spectrum
from those of the other members of the group is to be expected, since silicon
150 w. M. Tuddenham and J. D. Stephens

QUARTZ

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. II. Infrared spectra of selected tektosilicate minerals.

coordinates six oxygen atoms in stishovite as compared with the coordina-

tion of four oxygen atoms in all other silica minerals.
The second spectrum in Fig. 11 was obtained with anorthite,
CaAI 2Si20 s , the calcium end member of the plagioclase feldspar series.
This spectrum is typical of the spectra obtained by IR analysis of calcic
plagioclases through the composition range from 85 to 100% anorthite,
but because of crystal-structure modifications it is quite different from
spectra obtained with the sodic plagioclases. According to Deer, Howie,
and Zussman54 there are six structural modifications in the plagioclase feld-
spar series. Three of these can be clearly distinguished by IR analyses, as
was shown by Thompson and Wadsworth. 53
6. Infrared Spectrophotometry 151

The last two spectra in Fig. II were obtained with the two zeolite
minerals, chabazite, CaA12 Si 4 0 12 6H 20, and scolecite, CaAI 2 Si 30 lO 3H20.
Structural relationships of the zeolite minerals are complex and will not be
discussed in detail. Basically, however, the zeolite minerals are alumino-
silicate framework structures in which various percentages of the silicon
atoms have been replaced by aluminum. All zeolites have long tube- or
channellike holes in their structures, and the negative charge caused by
substitution of aluminum in the framework is balanced by alkaline ions or
alkali-earth ions, part of which occupy the holes. The cations occupying
the channellike holes may be exchanged by other cations from aqueous
solution, so long as the charge balance is maintained. Water or other gaseous
molecules may also occupy the holes in the zeolite structure.
The presence of water in chabazite and scolecite is shown clearly by
the absorption bands in their IR spectra near 3 and 6 fl. The great difference
in detail of the water bands of these two minerals, however, leads one to
suspect that there is a difference in the manner in which the water molecules
are bonded into their structures. This suspicion has been confirmed by
dehydration studies and differential thermal analysis of the two minerals,
as reported by Koizumi. 55

4.7. Phosphates, Vanadates, and Arsenates

Minerals containing phosphorus, vanadium, and arsenic often show

structural similarities which are related to similarities of their complex
anions. The general formula for all these complex anions is XO~-, where X
is pentavalent P, V, or As. Linking of the XO~- groups to form more
complex structural groups is well known in inorganic chemicals, but is
relatively rare in naturally occurring minerals.
In Fig. 12 the IR spectra obtained with a number of phosphate minerals
are shown. The first major absorption band related to the PO~- anion in
minerals generally occurs between 9 and 10 fl. The first two spectra in this
figure are those which were obtained with apatite, Ca5(P04)3F, and pyro-
morphite, Pb5(P04)3Cl. Although they differ considerably in composition,
these two minerals are isostructural and this relationship can be seen clearly
from similarities of their IR spectra.
Apatite varies in composition, and these variations affect the IR spectra.
Variations in the IR spectra of apatite samples and how these relate to com-
position have been investigated separately by Coles 56 and by Fischer and
Ring. 57 For example, hydroxyl groups or chloride ions may substitute for
152 W. M. Tuddenham and J. D. Stephens

APATITE
Ca.(PO.'.F

PYROMORPHITE
Pb.(PO.'.CI

LIBETHENIT E

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 12. Infrared spectra of selected phosphate minerals.

fluoride in the apatite structure. As would be expected, when hydroxyl

groups are substituted for the fluoride, a sharp O-H stretching band is
noted at 2.80 fl. When chloride ions are substituted into the structure the
bands near 17 fl are modified. In addition to these substitutions, carbonate
ions frequently substitute for part of the phosphate in the apatite structure.
When this occurs a sharp double absorption band is always present
near 7 fl.
The composition of pyromorphite also varies widely. Up to 8% calcium
may substitute for lead in this mineral, and both vanadate and arsenate
ions may substitute for the phosphate. The substitution of phosphate and
vanadate in this case is limited, but there is a complete solid solution series
6. Infrared Spectrophotometry 153

between pyromorphite and mimetite, Pb5(As04)aCl. The IR spectrum for

mimetite may be seen in Fig. 13. Intermediates in the pyromorphite-mime-
tite system have not been investigated, but their IR spectra are expected
to show intermediate characteristics between those of the end-member
minerals.
The third spectrum in Fig. 12 was obtained with libethenite,
CU2 (P04)OH. This mineral is isostructural with olivenite, Cu 2 (As0 4)OH,
whose spectrum may be seen in Fig. 13. The presence of OH- groups in
libethenite is demonstrated clearly by the presence of sharp bands at about
2.9 ft. While not indicated in the formula, the presence of bonded H 2 0 is
suggested by the presence of a broad band near 3 ft and by the presence of

VANADINITE

w
<.)
z

III
et:
o
en
III

DLIVENITE
CU.(A.O.IOH

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS
Fig. 13. Infrared spectra of selected vanadate and arsenate minerals.
154 W. M. Tuddenham and J. D. Stephens

a band near 5.85 fl and another at about 6.2 fl. In addition, a minor amount
of bonded CO~- in the mineral is suggested by the series of small but well-
defined absorption bands between 6.6 and 7.3 fl.
The final spectrum in Fig. 12 was obtained with the mineral wardite,
Na4CaAI12(P04MOH)93H20. The two sharp bands near 2.75 and 2.8 fl
give clear evidence of O-H stretching vibrations. In addition, the presence
of bands at 3.1 and 3.5 fl indicate the presence of hydrogen bonding. Cou-
pled with these bands, a series of bands centered about 6 fl indicate that
bonded H 20 groups are present.
In Fig. 13 the IR spectra which were obtained with a series of vanadate
and arsenate minerals are shown. The first major absorption band caused
by the presence of the VO~- anion in minerals occurs between 9 and 12 fl,
and the first major band related to the AsO~- anion generally occurs in
this same wavelength region. The first spectrum in Fig. 13 was obtained
with vanadinite, Pb5(V04)3Cl. Pyromorphite, vanadinite, and mimetite are
isostructural with each other and with apatite. While they are isostructural,
comparison of the IR spectra of the three minerals shows that the spectral
configuration of each is distinctly different. From this it is evident that the
complex anion in a mineral plays a stronger role than crystal structure in
controlling its spectral configuration.
The second spectrum in Fig. 13 was obtained with calciovolborthite,
CuCa(V0 4 )OH. The calciovolborthite spectrum shows a number of unusual
aspects. Although the presence of an OH- group is indicated in the mineral
formula, there is little indication of this OH- group in the form of an absorp-
tion band near 3 fl. The relatively broad band which occurs near 2.9 fl in this
spectrum resembles a band which is caused by absorbed water rather than
one which is caused by a functional OH- group. The presence of a band
near 6.1 fl also suggests that there may be H 20 in the sample or in the
mounting medium. Another unusual aspect of this spectrum is the series
of shallow but well-defined bands which occur between 3.4 and 5.5 fl. This
region is generally reserved for organic species and it is uncommon for
minerals to have absorption bands in this spectral region. The major absorp-
tion region between 9 and 15 fl for calciovolborthite is much more complex
than the relatively simple spectrum of vanadinite in this region. This dif-
ference in the major absorption bands of calciovolborthite and vanadinite
may be related to a linkage of the VO~- groups to form a more complex
structural unit in calciovolborthite.
The third spectrum in Fig. 13 was obtained with mimetite, Pb5(As04)3Cl.
This spectrum and its relation to vanadinite, pyromorphite, and apatite have
already been discussed.
6. Infrared Spectrophotometry 155

The fourth spectrum in Fig. 13 was obtained with conichalcite,

CuCa(As0 4 )OH. Although they are compositionally similar except for their
principal complex anions, calciovolborthite and conichalcite are not struc-
turally related. As was the case with calciovolborthite, however, there are
some unusual aspects to the conichalcite spectrum. At least two O-H stretch-
ing bands are noted in the conichalcite spectrum. The first of these is at 2.9 /1
and the second at 3.15/1. The long wavelength of the band at 3.15/1 sug-
gests that this band may be due to hydrogen bonding rather than to a
normal OH- group. There are no sharp, well-defined H 20 bands near 6/1,
but a broad band similar to those caused by absorbed water is present. A
series of small but well-defined bands near 7 /1 indicate that there is some
bonded CO~- in this conichalcite sample.
The final spectrum in Fig. 13 was obtained with olivenite, Cu 2 (As0 4 )OH.
This mineral is structurally related to libethenite, whose spectrum was
shown in Fig. 12. While the minerals are structurally related, there is little
similarity between their IR spectra. This indicates once again that the
complex anion or structural group is the major factor which controls the
configuration of the IR spectrum.

4.8. Molybdates and Tungstates

The molybdate and tungstate minerals contain complex anionic groups

with the molybdenum or tungsten cations at the center of a distorted tetra-
hedron of oxygen ions. Minerals containing molybdate or tungstate groups
are often structurally related, and extensive solid solutions exist between
members of certain molybdate and tungstate minerals. The first major
absorption band caused by the MoO~- or the WO~- anion in minerals
generally occurs between 11 and 13 /1.
In Fig. 14 the IR spectra are shown, which are obtained with a number
of molybdate and tungstate minerals. The first spectrum in this figure was
obtained with powellite, CaMo0 4 Powellite is isostructural with scheelite,
CaW0 4 , whose IR spectrum is also presented in the figure. A complete
solid solution system exists between powellite and scheelite. The powellite
end member of the series is relatively uncommon. Most material which is
called powellite in field identifications is actually a variety of scheelite which
contains a relatively minor amount of molybdenum. The material used to
obtain the powellite spectrum in Fig. 14, however, was essentially a pure
calcium molybdate which formed as the oxidation product of molybdenite,
MoS 2
The similarity of the IR spectra of powellite and scheelite is striking.
156 w. M. Tuddenham and J. D. Stephens

POWELLITE

lJ.J
U WULFENITE
Z
<t
aJ
a::
o(f)
aJ SCHEELITE
<t
CaWO. _-+-~

WOLFRAMITE
(Ft,Mn}W04

3 4 5 6 7 8 10 12 I 4 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 14. Infrared spectra of selected molybdate and tungstate minerals.

Examination of other isostructural species showed that changes in the

complex anion generally caused major changes in configuration of the IR
spectrum. The remarkably close similarity of the IR spectra of powellite
and scheelite suggests that the MoO~- and WO~- complex anions must also
have very similar bond and vibrational characteristics.
The second spectrum in Fig. 14 was obtained with wulfenite, PbMo0 4 o

Wulfenite is structurally related to powellite and scheelite, but is not iso-

structural with them since it lacks a mirror-plane crystal symmetry element
that is present in them. The structural similarity of wulfenite to powellite
and scheelite is suggested by similarities of their IR spectra. Extensive solid
solution relations exist between wulfenite and each of the other minerals.
Analytical data presented in Dana's System of Mineralogy58 shows that
wulfenite may contain at least 6.88% calcium, which would give it a com-
position intermediate between wulfenite and powellite. Other data from
this same source shows that at least half of the molybdenum in wulfenite
may be replaced by tungsten, giving a composition intermediate between
6. Infrared Spectrophotometry 157

wulfenite and its lead tungstate analog, stolzite. These intermediate mineral
species probably have IR spectra with characteristics intermediate between
their respective end members. The displacement of the major absorption
band of wulfenite as compared with the major bands of powellite and schee-
lite is almost certainly related to the increased mass of lead in wulfenite as
compared with calcium in powellite and scheelite.
The final spectrum in Fig. 14 was obtained with wolframite,
(Fe, Mn)W0 4 Wolframite is crystallographically dissimilar from the pre-
viously discussed molybdate and tungstate minerals, and this dissimilarity
is reflected by differences in its IR spectrum. Wolframite samples show a
wide range in composition, relating to the ratio of their iron and manganese
contents. In separate studies Grubb 59 and Chang60 showed that minor
amounts of calcium may substitute into wolframite, and that consequently
there is some degree of solid solution between wolframite and scheelite.
Solid solution of this type would presumably produce IR spectra inter-
mediate between those of scheelite and wolframite.

4.9. Oxide Minerals

The oxide minerals do not contain complex anions or other complex
structural groups. They therefore cannot be conveniently included in a
systematic classification that is based on the relations between complex
chemical or structural groups and the IR absorption bands that are char-
acteristic of them. Metal ions which form oxides vary widely in charge and
in ionic radius. The different cations therefore, coordinate different numbers
of oxygen ions in their oxide structures. As a result the structures and IR
spectra of oxide minerals also vary widely. The first major absorption band
for oxide minerals may occur anywhere in the wavelength region from 8.5
to 20.0 fl.
A study of the IR spectra of simple metal oxides was made by McDevitt
and Baun. 61 They showed that while the IR spectra of metal oxides were
generally relatively simple and lacking in sharp absorption bands, they were
characteristic and showed systematic relations with structurally related
compounds.
In Fig. 15 the IR absorption spectra are shown which were obtained
with three oxide minerals. There is a wide range in spectral complexity of
the patterns obtained with these three minerals. The first spectrum in Fig. 15
was obtained with rutile, Ti0 2 Other minerals having the rutile structure
have quite different lR spectra.
The second spectrum was obtained with corundum, AI 20 a . This pattern
158 W. M. Tuddenham and J. D. Stephens

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 15. Infrared spectra of selected oxide minerals.

is typical of those obtained with both natural and synthetic corundum

samples, but individual spectra vary slightly. The bands that occur between
8 and 10 fl vary in their intensities. The spectrum shown illustrates approxi-
mately the maximum intensity for these bands, but in some cases they are
so poorly developed that they are difficult to detect. The bands between 20
and 22 fl show variations in intensity and wavelength. These variations may
relate to the degree of ordering of the mineral's crystal lattice. Other poly-
morphous forms of Al20 3 have different patterns that are distinctive and
characteristic of their species.
The third spectrum in Fig. 15 was obtained with spinel, MgAI20 4
Extensive solid solution systems occur between members of spinel group
minerals. Divalent iron, zinc, and manganese may substitute for magnesium;
and trivalent iron, trivalent manganese, chromium, vanadium, and titanium
often substitute for aluminum in spinel. This mineral system has not been
studied in detail by the authors, but the IR spectra of the series may be
expected to vary systematically with changes in composition.
 6. Infrared Spectrophotometry 159

5. CRYSTAL CHEMISTRY STUDIES

5.1. Solid Solution Series

Infrared absorption spectroscopy offers a particularly useful technique
for the study of solid solution series. Studies have been made with chlorites,
feldspars, and olivines, among others. Tuddenham and Lyon62 found that
by relating the IR absorption spectra to chemical analyses of 21 chlorites,
both the degree of substitution of aluminum for silicon and the total iron
content of the mineral could be estimated. The structural type (7 or 14 A)
could also be deduced readily from the IR curve. Stubican and Roy63 used
IR and x-ray diffraction in a study of synthetic chlorites, but were unable to
note the effects of the AP+ for Si4+ substitutions. This suggests that the
synthetic minerals were in disequilibrium as compared to the natural min-
erals. Hayashi and Oinuma64 used IR to study the Si-O and OH- absorp-
tion bands of chlorites.
Thompson and Wadsworth 53 demonstrated that there is a systematic
variation in the IR spectra of plagioclase feldspars throughout the albite-

18.0 18.0

17.8 17.8
VJ
z 17.6 17.6
0
0:
U
~ 17.4 17.4
z
:r: 17.2 17.2
I-
(!)
Z
LJ.J
..J 17.0 17.0
LJ.J
>

~ 16.8 16.8

16.6 16.6

16.4 16.4
0 10 20 30 40 50 60 70 80 90 100
FAYALITE MOL PER CENT FORSTERITE
FORSTERITE
Fig. 16. Plot of wavelength vs composition for one absorption band from the forsterite-
fayalite solid solution series.'
160 W. M. Tuddenham and J. D. Stephens

anorthite series. They found definite correlation between the band positions,
number of bands and composition. At least two modifications of the plagio-
clase crystal structure were indicated by the IR data.
Duke and Stephens65 investigated the relations between the IR spectra
and compositions of members of the olivine-group minerals. They demon-
strated a continuous displacement of the IR absorption maxima of members
of the forsterite, Mg2Si0 4 , to fayalite, Fe 2Si0 4 , solid solution series toward
longer wavelengths as the iron contents of the samples increased. This
relation is shown in Fig. 16. Duke66 confirmed earlier conclusions that there
is a continuous solid solution system between forsterite and tephroite,
Mn 2Si04
A solid solution series extends from adamite, Zn 2 (OH)As0 4 , through
Cu Zn
ADAMITE % %
Zn,(OH) AsO.

0.3 46.7

14.0 35.2 I
1

3 4 5 6 7 8 10 12 14 16 18202224
WAVELENGTH iN MICRONS

Fig. 17. Infrared absorption spectra of members of the adamite-olivenite

series. Note systematic changes of spectral configuration in contrast with
the wavelength shifts exhibited by the forsterite-fayalite system (see Fig. 16)
6. Infrared Spectrophotometry 161

cuproadamate, Zn-olivenite, to olivenite, Cu 2 (OH)As0 4 Composition of

the samples chosen varied continuously with respect to their copper and
zinc contents.
Infrared spectra obtained with members of this series are shown in
Fig. 17. Unlike x-ray diffraction spectra, the IR spectra vary continuously
throughout the entire composition range. In contrast with the forsterite-
fayalite series, the systematic variations are not in wavelength but rather
in configuration of the spectra. The changes in spectral configuration are
probably associated with progressive distortion of the crystal lattice which
is caused by the isomorphous substitutions of copper and zinc. These distor-
tions were also indicated by the discontinuous variations of x-ray dif-
fraction spectra.

3 4 5 6 7 8 10 12 14 16 18 202224
WAVELENGTH N MICRONS

Fig. 18. Infrared spectra of kaolin group minerals. Arrows

point to absorptions which show significant variations in the
IR spectra. (Nacrite: U. S. National Museum, Cat. No. 3918.)
162 W. M. Tuddenahm and J. D. Stephens

5.2. Polymorphous Series

The kaolin group of minerals are polymorphous forms of AI 4Si 4 0 lo (OH)s
and consist of nacrite, dickite, kaolinite, and halloysite. Halloysite also con-
tains excess H 20. Arrangement of the spectra in logical sequence of decreas-
ing spectral complexity produces the series seen in Fig. 18. It seems more
than a coincidence that this is also the order into which these minerals fall
when arranged in the sequence of increasing complexity of crystal structure
and decreasing number of stacked and ordered kaolin sheets. 67
Changes in the crystal structure of the kaolin-group minerals are so
slight that differentiating them by x-ray diffraction analysis is difficult. Be-
cause of the compositional and structural similarities of the kaolin group

POLYMORPHS OF
A 12 S i 0 5

I I I
KYANITE

w
(,)
Z
<l
m
a:
o
(/)
m
<l

8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 19. Infrared spectra of aluminum silicate

trimorphs.
6. Infrared Spectrophotometry 163

minerals, it would be expected that their IR spectra would also be similar.

The spectra in Fig. 18 show that this is indeed the case, but small, repro-
ducible differences can be seen as indicated by arrows in the figure. The
differences, particularly in the O-H stretching region near 2.75 fl, could be
related to changes in nearest neighbors as the stacked kaolin sheets shift.
Another well-known series of this type is the trimorphous forms of
AI 2SiOs-kyanite, andalusite, and sillimanite. Tracings of the IR spectra
obtained with these minerals are shown in Fig. 19. Similarities between the
spectra of kyanite and andalusite are apparent. The relation between these
minerals and sillimanite, however, is not so clear. The IR spectra become
more complex proceeding through the series from kyanite to andalusite to
sillimanite. The complexity of the spectra is probably related to structural
changes in the crystal lattices. This conclusion is supported by studiess4
showing structural modifications of the three minerals to occur in the same
order as that established empirically from IR spectra.

5.3. Leaching Studies

In a study of loughlinite, Na203MgO6Si02 8H20, Fayhey, Ross,
and Axelrod68 showed that it was transformed to sepiolite, 4MgO6Si02
8H20, by leaching sodium from its structure with distilled water. This
decomposition was followed using IR analysis. Figure 20 shows that the
IR spectra of the 60-day leach product resembles that of a natural sepiolite,
but it lacks the sharp definition of a natural mineral. This lack of detail is
probably due to disordering caused by removal of sodium from the lough-
linite lattice.

5.4. Firing Studies

The IR study of fired serpentine minerals illustrated in Fig. 21 points
out the applicability of the technique to this type of study. In the chrysotile
series no reactions took place up to 500C. From 500 to 600C the mineral
decomposed with loss of structural hydroxyl, as indicated by the disappear-
ance of the OH- absorption bands at 2.7 fl. With dehydration the chrysotile
crystal lattice broke down leaving a material nearly amorphous to x-ray
diffraction studies. This structureless state persisted through the 700C
temperature interval. From 700 to 800C recrystallization occurred with the
formation of a mixture of forsterite and enstatite. This mixture remained
stable through 1000C with only a little additional enstatite forming at
higher temperatures.
164 w. M. Tuddenahm and J. D. Stephens

LEACHING OF LOUGHLINITE
I I I I
LOUGHLINITE

lJJ
U
Z

III
a::
o
(/)
III

3 4 5 6 7 8 10 12 14 16 18 20 22 24
WAVELENGTH IN MICRONS

Fig. 20. Infrared spectra illustrating effects of water leaching of loughlinite.

Antigorite showed a tendency to become disordered at a relatively low

temperature, as indicated by the broadening and loss of detail of its spec-
trum after firing at 400-500C. The structureless interval noted in the case
of chrysotile did not occur in the firing of antigorite. Recrystallization to
forsterite apparently occurred simultaneously with the destruction of the
antigorite crystal lattice. Some bonded hydroxyl remained in the 600C
spectrum as evidenced by absorption at 2.7 fl. Forsterite absorptions were
apparent in the same spectrum. Enstatite, which apparently formed simul-
taneously with forsterite in the thermal recrystallization of chrysotile, did
not form in the antigorite series until 900C. As with the chrysotile series,
the final reaction products were enstatite and forsterite. Similar techniques
have been utilized in the study of quartz, kaolin, and microcline. 69
6. Infrared Spectrophotometry 165

WAVELENGTH IN MICRONS

Fig. 21. Infrared spectra showing the pattern of thermal recrystallization

followed by the serpentine minerals, chrysotile and antigorite.

5.5. Alteration Studies

Finding new economic mineral deposits requires many approaches.
One approach is based on the study of mineral alteration in areas surround-
ing ore bodies. Bleached and altered areas are important clues to the pres-
ence of sulfide mineralization. Geologists often use these halos of altered
rock to determine drilling sites; but to make best use of the alteration halo,
they must map in detail the degree of alteration. These mineral changes are
often difficult to follow with conventional methods, so special techniques
166 w. M. Tuddenham and J. D. Stephens

10.1
o
z
'"o
m
II:
Ul
m
'"
!
Fig. 22. Alteration studies with IR. Absorption curve A shows presence of 20%
kaolinite (K), 25% quartz (Q), and 20% calcite (CO:-). Curve B, for a 3-mg
sample from a thin vein crossing rock A, shows 34% kaolinite (K) and 40% quartz
(Q); siderite (S), dolomite (D), and calcite (C) make up a total of about 25%
CO:- in the sample. [Reprinted by permission from Eng. Min. Journ. 161 (7),
93 (1960)].

must be used. The IR method is especially valuable in this respect. Figure 22

shows the IR curve of an altered rock in which the minerals were so fine
grained that accurate optical estimation of the relative concentrations was
impossible. 70 With IR it was easy to identify the important minerals and to
measure their amounts.

REFERENCES

1. Coblentz, W. W., Investigation of Infrared Spectra, Part III: 1906; IV: 1906; V: 1908,
VI: 1908; VII: 1905; Carnegie Institution of Washington, republished by the Coblentz
Society and the Perkin-Elmer Corp., 1962.
2. Lyon, R. J. P., Minerals in the Infrared, Stanford Research Institute, Menlo Park,
Calif., 1962.
3. Hunt, J. M., Wisherd, M. P., and Bonham, L. C., Anal. Chem. 22, 1478 (1950).
4. Miller, F., and Wilkins, C. H., Anal. Chem. 24, 1253 (1952).
5. Launer, P. J., Am. Min. 37, 764 (1952).
6. Williams, V. Z., Rev. Sci. Instr. 19, 135 (1948).
7. Herzberg, G., Molecular Spectra and Molecular Structure, II. Infrared and Raman
Spectra of Polyatomic Molecules, Van Nostrand, New York, 1945.
8. Nakamoto, K., Infrared Spectra of Inorganic and Coordination Compounds, John
Wiley & Sons, New York, 1963.
9. Herscher, L. W., Instrumentation, in Applied Infrared Spectroscopy, (D. N. Kendall,
ed.), Reinhold, New York, 1966, pp. 88-135.
6. Infrared Spectrophotometry 167

10. Low, M. J. D., Int. Science and Techn., February (1967).

11. Rogers, A. F., and Kerr, P. F., Optical Mineralogy, McGraw-Hill, New York, 1942,
pp. 3-8.
12. Stimson, M. M., and O'Donnel, M. J., J. Am. Chem. Soc. 74, 1805 (1952).
13. Schiedt, U., and Reinwein, H., Z. Naturforsch. 7b, 270 (1952).
14. Meloche, V. W., and Kalbus, G. E., J. Inorg. Nucl. Chem. 6, 104 (1958).
15. Tuddenham, W. M., and Lyon, R. J. P., Anal. Chem. 32, 1630 (1960).
16. Lyon, R. J. P., Am. Min. 48, 1170 (1963).
17. Fahrenfort, J., Spectrochim. Acta 17, 698 (1961).
18. Hirschfeld, T., Appl. Spectry. 21, 335 (1967).
19. Harrick, N. J., Internal Reflection Spectroscopy, Interscience, New York, 1967.
20. Lyon, R. J. P., and Burns, E. A., Econ. Geol. 58, 274 (1963).
21. Hovis, W. A., Applied Optics 5, 245 (1966).
22. Simon, J., and McMahon, H. 0., J. Chem. Phys. 21, 23 (1953).
23. Low, M. J. D., Applied Optics 6, 1503 (1967).
24. Kropotkin, M. A., and Kozyrev, B. P., Optics and Spect. 17, 136 (1964).
25. Lyon, R. J. P., and Burns, E. A., Infrared Emittance and Reflectance Spectra of
Rough and Powdered Rock Surfaces, Presented at Meeting of the Working Group
on Extraterrestrial Resources, Cape Kennedy, Florida, November, 1964.
26. Lyon, R. J. P., and Tuddenham, W. M., Min. Eng. 11, 1233 (1959).
27. Kendall, D. N., Survey of Practical Information, pp. 59-60; R. D. Moss and W. J.
Potts, Jr., Infrared on the Chemist's Bench, pp. 216-217; J. L. Koening, Application
of Infrared Spectroscopy to Polymers, pp. 262-263; in Applied Infrared Spectroscopy,
(D. N. Kendall, ed.), Reinhold, New York, 1966.
28. Lyon, R. J. P., Thompson, C. S., and Tuddenham, W. M., Econ. Geol. 54, 1047
(1954).
29. Tai, H., and Underwood, A. L., Anal. Chem. 29, 1430 (1957).
30. Citron, I., and Underwood, A. L., Anal. Chim. Acta 22, 338 (1960).
31. Stein, W., Annal. PhYJick. 36, 462 (1939).
31a. Phillippi, C. M., AFML-TR-67-437, Air Force Materials Laboratory, Wright Pat-
terson Air Force Base, Ohio, April 1968.
32. Moenke, H., Mineral Spectren, Deutsche Akademie der Wissenschaften Zu Berlin,
Akademie Verlag, Berlin, 1962.
33. Lawson, K. E., Infrared Absorption of Inorganic Substances, Reinhold, New York,
1961.
34. Ferraro, J. R., Anal. Chem. 40, 24A (1968).
35. Adler, H. H., et al., Am. Petro Inst., Project 49, 146 (1950).
36. Busing, W. R., and Levy, H. A., Acta Cryst. 11, 798 (1958).
37. Adler, H. H., and Kerr, P. F., Bull. Geol. Soc. Am. 72 (1961); Am. Min. 48, 700
(1962).
38. Adler, H. H., and Kerr, P. F., Am. Min. 48, 124 (1963).
39. Adler, H. H., and Kerr, P. F., Am. Min. 48, 839 (1963).
40. Huang, C. K., and Kerr, P. F., Am. Min. 45, 311 (1960).
41. Menzies, A. C., Proc. Roy. Soc. (London) A134, 265 (1931).
42. Christ, C. L., Am. Min. 45,334 (1960).
43. Tennyson, C., Fortschr. Mineral. 41, 64 (1963).
44. Omari, K., and Kerr, P. F., Bull. Geol. Soc. Am. 74, 709 (1963).
45. Adler, H. H., and Kerr, P. F., Am. Min. 50, 132 (1965).
168 W. M. Tuddenham and J. D. Stephens

46. Berry, L. G., and Mason, B., Mineralogy; Concepts, Descriptions, Determinations,
W. H. Freeman, San Francisco, 1959, p. 467.
47. Lyon, R. J. P., Evaluation of Infrared Spectrophotometry for Compositional Analysis
of Lunar and Planetary Soils, Stanford Research Institute Project No. PSU-3943,
1962.
48. Bassett, W. A., Bull. Geol. Soc. Am. 71, 449 (1959).
49. Wickersheim, K. A., and Buchanan, R. A., Am. Min. 44, 440 (1959).
50. Ply us ina, I. I., Geokhimiya 1, 31 (1964).
51. Deer, W. A., Howie, R. A., and Zussman, J., Rock Forming Minerals, Vol. 1. Ortho
and Ring Silicates, Longmans, London, 1963.
52. Milkey, R. G., Am. Min. 45, 990 (1960).
53. Thompson, C. S., and Wadsworth, M. E., Am. Min. 42, 334 (1957).
54. Deer, W. A., Howie, R. A., and Zussman, J., Rock Forming Minerals. Vol. 4. Frame-
work Silicates, Longmans, London, 1963, p. 96.
55. Koizumi, M., Miner. Jour. 1, 36 (1963).
56. Coles, J. L., A Study of Some Synthetic Apatites, Ph.D. dissertation, Dept. of Minera-
logy, University of Utah, 1963.
57. Fischer, R. B., and Ring, C. E., Anal. Chem. 29, 431 (1957).
58. Dana, J. D., and Dana, E. S., The System of Mineralogy, II, 7th Ed., as rewritten and
enlarged by C. Palache, H. Berman, and C. Frondel, John Wiley & Sons, New York,
1951.
59. Grubb, P. L. c., Am. Min. 52, 418 (1967).
60. Chang, L. L. Y., Am. Min. 52, 427 (1967).
61. McDevitt, N. L., and Baun, W. L., A Study of the Absorption Spectra of the Simple
Metal Oxides in the Infrared Region 700-240 cm-l, Technical Documentary Report
No. RTD-TDR-63-4172, A. F. Materials Laboratory Research and Technology
Division, Wright-Patterson Air Force Base, Ohio, Project # 7360, Task No. 736005,
1964.
62. Tuddenham, W. M., and Lyon, R. J. P., Anal. Chem. 31, 377 (1959).
63. Stubican, Y., and Roy, R., J. Am. Ceram. Soc. 44, 625 (1961).
64. Hayashi, H., and Oinuma, K., Am. Min. 52, 1210 (1967).
65. Duke, D. A., and Stephens, J. D., Am. Min. 49, 1388 (1964).
66. Duke, D. A., Infrared Investigation of the Crystal Chemistry of Olivine and Humite
Minerals, Ph.D. dissertation, Department of Mining and Geological Engineering,
University of Utah, 1962.
67. Deer, W. A., Howie, R. A., and Zussman, J., Rock Forming Minerals, Vol. 3. Sheet
Silicates, Longmans, London, 1963.
68. Fahey, J. J., Ross, M., and Axelrod, J. M., Loughlinite, Am. Min. 45, 270 (1960).
69. Stephens, J. D., and Tuddenham, W. M., Am. Ceram. Soc. Bul. 46, 725 (1967).
70. Tuddenham, W. M., and Zimmerly, S. R., Eng. and Min. J. 161 (7), 92 (1960).
Chapter 7

OPTICAL EMISSION SPECTROSCOPY

Armin P. Langheinrich and D. Blair Roberts*

Kennecott Copper Corporation
Salt Lake City, Utah

1. INTRODUCTION

Optical emission spectroscopy includes the fields of fiame-, arc-, and spark-
induced emission phenomena in the UV, visible, and near IR regions of
the electromagnetic spectrum. As a technique, it furnishes the analytical
investigator with qualitative and quantitative information on the elemental
composition of matter through simultaneous multielement determinations.
Detection limits in the low-ppm range are quite common.
Emission spectroscopy, often referred ,to as spectrochemistry, was the
first direct instrumental analytical technique to be widely used in geo-
chemical investigations. In spite of the development of new methods and
techniques and the ready availability of corresponding instrumentation, emis-
sion spectroscopy remains today an indispensable tool for the geochemist.

1.1. History
Approximately 1800 years passed between Seneca's observation of the
colors of the visible spectrum through a primitive prisml and the develop-
ment of a practical spectroscope. 2 In 1860 Bunsen and KirchhofP initiated
spectral analyses. By 1907, 13 elements had been discovered by spectroscopic
means. 2 Though the use of this instrument became widely accepted during

* Presently at Westinghouse Georesearch Laboratory Boulder, Colorado.

169
170 Armin P. Langheinrich and D. Blair Roberts

this period, its application to quantitative work remained limited until

Gerlach, 4 in 1925, reported the application of internal standards. The result
was a dramatic improvement in the precision and accuracy of the technique.
From 1929 to 1936 Goldschmidt 5- IO demonstrated the great potential of
spectrochemistry in geochemical work. He and his coworkers made element-
abundance studies, the results of which are still quite acceptable today.
Historical introductions are presented in most textbooks and treatises
on emission spectroscopy. An interesting chapter on the subject was written
by Weise.1 Recent presentations have included the history of flame emission
by Fassel,n of the spark discharge by Walters,12 and of the dc arc by Strock.13
Since L. W. Strock himself was a member of Goldschmidt's team, his paper
adds considerably to the knowledge of that era.

1.2. Principles
Detailed discussions of basic theory and the origin of atomic spectra
are beyond the scope of this chapter and can be found in textbooks on
modern physics and spectroscopy, such as Herzberg,14 Richtmyer, Kennard,
and LauritsenI5 and McNally.16 It is appropriate, however, to include the
following summary of the underlying principles.
Optical emission spectra may be observed when transitions of the outer
energy-level electrons occur within an atom. This phenomenon is illustrated
diagrammatically in Fig. 1, where ground-state conditions are compared
with excitation and emission steps. An electron in a given atom is defined
by a set of quantum numbers and is associated with a definite amount of
energy. When electrical, optical, or thermal energy is introduced into the
atom, electrons are transferred from low-energy to high-energy quantum
states, i.e., the "ground-state" atom is now in an "excited state." After
extremely short time intervals the energy-rich electrons undergo transitions
into states of lower energy. The resulting energy difference is emitted in the
form of electromagnetic radiation. From the law of conservation of energy
the energy of the emitted photon fzv must equal LI E, the difference in energy
content of the electron prior to and after transition, where fz is Planck's
constant (6.6 X 10-27 erg-sec) and v is the frequency in vibrations per sec.
According to the Bohr theory electrons are stable in definite quantum
states: Therefore, only specific energy differences and frequencies are possi-
ble. This explains the occurence of a definite line spectrum for each atom.
A considerable number of transitions are possible, and a large number of
spectral lines can be found for most elements. In the case of iron, for ex-
ample, nearly 5000 lines are listed in the M.I.T. wavelength tables,17 Since
7. Optical Emission Spectroscopy 171

IONIZATION

-------
ENERGY
~----- .. -----
E'"
E"

E'

E
v, ~ V
~ hv = AE

A B c

Fig. 1. Energy-level diagram showing electronic transitions. (A) Ground-state

conditions. (B) Excitation. (C) Emission.

the probabilities of these transitions vary, certain lines are found to domi-
nate a spectrum. In most cases these dominating lines are used to establish
the presence of a particular element.
Excitation takes place as soon as the "excitation potential" bf a specific
transition has been reached. The excitation potential is a numerical expres-
sion of the minimum energy required to cause excitation, and is usually
reported in volts or electron volts. When the exciting energy is increased
further, ionization may occur. In such a case an electron is removed com-
pletely from the influence of the parent nucleus of atomic number Z. The
resulting ion may also be excited, but its spectrum differs significantly from
the atom spectrum of element Z, and resembles the atom spectrum of
element Z - 1. Similarly, the ion spectrum of a doubly ionized element Z
resembles the atom spectrum of element Z - 2. and so on. In addition to
atoms and ions, certain molecules can be excited under spectrochemical
conditions. The resulting molecular band spectra often interfere with line
spectra and are of limited analytical value. Figure 2 represents a partial
energy-level diagram of the Grotrian type. IS It shows transitions for the
four most sensitive lines of the neutral sodium atom. Excitation and ioniza-
tion potentials (in volts) are taken from Ahrens. I9
In spectrochemical analyses samples are vaporized and excited by means
172 Armin P. Langheinrich and D. Blair Roberts

FIRST IONIZATION POTENTIAL 5. 14V

n =4

n =3

o n =3
Zs Zp Zp
1/2 1/2 3/2

Fig. 2. Energy-level diagram showing transitions for the

four most sensitive lines of the neutral sodium atom.

of flames, arcs, sparks, or lasers. The resulting atomic, ionic, and molecular
radiation is focused on a narrow slit, dispersed by gratings or prisms, and
observed visually or recorded photographically or electronically. The pro-
jected spectrum or photographic record consists of a series of lines where
each line represents an image of the slit and corresponds to a specific wave-
length of radiation. The determination of line position, and hence of wave-
length, allows qualitative identification. By measuring the intensity of the
emitted characteristic radiation, a quantitative result may be obtained.

1.3. Relation to Other Techniques

Spectrochemical analyses may be used independently of other analytical
techniques. However, it must be realized that no analytical instrument is
available that will meet all requirements. In many cases information obtained
from two or more complementary methods must be correlated to arrive at
a complete answer. This is, however, not often possible within the desired
limits of speed, accuracy, precision, and practicality. Spectrochemical pro-
cedures can contribute significantly to the solution of other instrumental and
7. Optical Emission Spectroscopy 173

chemical problems. The spectrograph may furnish details on impurity levels

in reagents and process intermediates, in "insoluble" residues and in pre-
cipitates. In many instances the spectrographic results provide significant
"survey" information prior to the selection of final geochemical or analytical
procedures.
Laboratories with emission spectrographic equipment often have other
analytical facilities available. The spectrochemist should utilize these. He
can benefit greatly from information obtained by other methods. For ex-
ample, IR spectra yield information on anions, on specific minerals, and on
mineral types. X-ray diffractometer scans provide information on major
concentrations of chemical compounds and minerals. X-ray emission deter-
minations furnish elemental compositions in a nondestructive way. Atomic
absorption, flame photometric, and colorimetric techniques assist where
external standards are being generated or verified. Such information will
enable the spectrochemist to make proper selection of sample size, internal
standards, external standards, dilution materials, operating conditions, etc.
Although the multimethod approach is certainly not necessary in every
case, its potential value to both spectrochemist and geochemist must not
be underrated. It may lead to significant shortcuts and it is of inestimable
value where the history of a sample is uncertain.

2. EQUIPMENT AND FACILITIES

In the following section specific instruments are not discussed, since

excellent advertising material from a number of firms adequately covers this
area. Lists of suppliers of complete analytical systems as well as of major
and auxiliary components have been published. 20 21 Manufacturers have
generally been found helpful in assisting in the solution of specific instru-
mental and even analytical problems. Their help, however, is no full sub-
stitute for experience and sound knowledge on the part of the spectroscopist.

2.1. Components and Functions

Basically, an emission spectrographic system consists of three compo-

nents: (a) the excitation source, (b) the dispersing unit, and (c) detection
equipment. The analytical sample, properly positioned between the elec-
trodes of an arc-spark stand, is excited. It emits radiation characteristic of
the elements present in the sample. The spectrograph separates and isolates
these radiant energies prior to detection and recording.
174 Armin P. Langheinrich and D. Blair Roberts

2.1.1. Excitation Sources

In addition to flame, several other excitation techniques have been
utilized. These include the dc arc, the ac arc, the high-voltage ac spark,
and various modified arc-spark sources, such as the interrupted or ignited
arc. Recently, laser vaporization with auxiliary arc excitation has been
added to this list. The dc arc is the popular excitation source for most types
of analyses encountered in geochemical and related work. With the dc arc
good sensitivity may be achieved, and numerous semiquantitative and quan-
titative determinations have been performed successfully. Ahrens and Tay-
10r,19 for example, based their well-known textbook on spectrochemical
analyses solely on the use of this source. The ac arc, most widely used
for some time, has been largely replaced by the high-voltage ac spark or
by various combinations of ac-spark sources. Good precision is obtain-
able with the spark source, but the corresponding sensitivity is generally
poorer than that obtainable with the dc arc. The spark source is widely
used for the quantitative analysis of metal and metal-alloy systems. The
laser may be used for the primary vaporization of small sample volumes
prior to arc excitation of the vaporized sample. Its usefulness in geochem-
istry, especially in mineralogical studies of small inclusions or small mineral
grains, has already been recognized.

2.1.2. Dispersing Units

Spectral dispersion is accomplished by prisms or diffraction gratings.
The use of prisms is declining, mainly as a result of improved technology
in the production of high-quality gratings. Whenever prisms are used in
geochemical work they should be of the quartz type since both the visible
and the UV regions of the spectrum are of interest. Prism instruments have
the advantage of large dispersions in the UV region. They produce no over-
lapping spectral orders. Grating instruments, on the other hand, provide
nearly constant dispersion at all wavelengths. Higher orders are produced
and are utilized when large dispersion is a necessity. Prism mounts (e.g.,
Littrow and Cornu) and grating mounts (e.g., Eagle, Ebert, Wadsworth,
Paschen-Runge) have been discussed adequately in the literature. 2 ,22 All
these types are currently in use.

2.1.3. Detection Equipment

The detection of dispersed line spectra can be accomplished visually,
by photographic means, or electronically with photomultiplier tubes. Photo-
7. Optical Emission Spectroscopy 175

__ - - __ "
FOCAL CURVE
A~ci!.~~~C

"
\ RATING

/
" ,,/

. . . . ~ENTRANcrM;-R;
' . -......;: ENTRANCE SLIT
........ ENTRANCE LENS
QUARTZ WINOOW
"'-~-':-:----NEUTRALIlING SHUTTER
.. --SOURCE

Fig. 3. Optics diagram of a direct reader. Courtesy

of Baird-Atomic, Inc.

graphic methods, when applied to quantitative work, necessitate some kind

of plate or film calibration to relate line intensities of the photographic
image to light intensities originating from the excited sample. The applica-
tion of direct readers, i.e., instrumentation with electronic detection devices,
is expanding rapidly in rock and mineral work. Commercial units are avail-
able that can detect about 60 elements simultaneously plus up to eight in
the vacuum region below 2000 A. A diagram of a direct-reader arrangement
is presented in Fig. 3. The dispersed radiation must pass through properly
positioned exit slits prior to reaching the photomultiplier tubes. These
convert light into electrical energy. The resulting signals are integrated
during the period of arcing and finally amplified and recorded. This ap-
proach is especially valuable where much routine work has to be done. It
has by no means replaced the photographic technique. Variations often
encountered in geochemical work can be met with maximum flexibility by
the latter. In this case permanent records of all elements present are pro-
duced. The photometric step is done on microdensitometers available com-
mercially in various sizes and complexities.

2.1.4. Auxiliary Equipment and Materials

Commercially available spectrochemical materials allow the operator

to concentrate on his analytical work load. Except in the case of unusual
applications, it is no longer necessary to spend valuable time in preparing
and shaping electrodes and other time-consuming preparatory steps. Items
to be considered in equipping a laboratory include the following: electrodes,
electrode shapers, milling and grinding units, sample presses, gas supplies
and atmosphere controllers, photographic film, plates and chemicals, and
calculating accessories.
176 Armin P. Langheinrich and D. Blair Roberts

2.2. Laboratory Facilities

Cost and space are the principal restrictions when purchasing spectro-
graphic equipment. Instrumentation and materials are available in a wide
range of complexity, price, and usefulness for specific applications. Simple
visual comparators are available for less than $ 2000. Highly sophisticated
direct-reading spectrometers for air and vacuum applications, coupled with
computer-controlled operation and determination, may cost up to $ 200,000.
Among read-out options one can choose strip-chart recorders, digital volt-
meters, punched paper tape, electric typewriters, and completely computer-
controlled spectrometer systems or interfaces to operate with most digital
computers. When photographic detection is used film or plate processing
facilities are needed. The laboratory should have controlled temperature
and humidity, and it should be kept scrupulously clean.

2.3. Field Facilities

The general instrumentation is not different from laboratory units,

although space restrictions are much more severe. Field instruments, by
necessity, have to be small to allow mounting in mobile laboratories. Day-
light developing equipment must be used if darkroom facilities are not
available. With respect to sample preparation, i.e., crushing, grinding, and
homogenizing, the operator has to depend on his own ability and apparatus.
All equipment must be able to withstand effects of vibration and changes in
temperature and humidity.
One manufacturer features a small-size spectrograph that can be
mounted in an area of less than 3 X 6 ft. Its reciprocal linear dispersion in
the second order is listed as 6.4 A/mm. Although equipment of this type is
somewhat limited in its usefulness, it can provide rapid on-the-spot informa-
tion of great value. An example of a field instrument is shown in Fig. 4.

Fig. 4. Table-top ARL spectrograph with reader. Courtesy of Applied Research Labo-
ratories.
7. Optical Emission Spectroscopy 177

I I

Fig. 5. Interior of mobile laboratory. Right, 1.5-m JACO Spectrograph. Left, ARL
microdensitometer. Courtesy of the Jarrell-Ash Co.

Mobile laboratories are gaining in popularity. They provide desirable close

liaison with geologic mapping or prospecting. Considerable experience in
field applications of emission spectrographs has been gained by the U. S.
Geological Survey. Their laboratories, mounted on 2.5-ton trucks, are
equipped with 1.5-m spectrographs. Each unit can handle up to 100 samples
per day if 30 elements are to be determined. 23 The interior of one of these
laboratories is shown in Fig. 5.

2.4. Special-Purpose Equipment

The experienced analyst will undoubtedly develop his own variations
of procedure and equipment. He may look beyond a minimum instrument
package. Special attachments are available which allow arcings under con-
trolled atmospheric conditions. 24 25 Automated sample presentation systems
have been developed. 26 Lasers, as shown in Fig. 6, are used in direct excita-
tion and in primary excitation to feed the normal arc. Automatic develop-
178 Armin P. Langheinrich and D. Blair Roberts

~"---"-"""""""""'-CJ
g= --
~~IIUII

c::::::J .-..

Fig. 6. Laser microprobe by JACO. Courtesy of the Jarrell-Ash Co.

ing equipment facilitates the photographic steps. Vacuum attachments

allow the analytical evaluation of the most sensitive lines of antimony,
arsenic, carbon, mercury, phosphorus, selenium, sulfur, and zinc. Combina-
tion spectrometer-spectrographs are available, which permit changeover
from the direct-reading to the photographic method, and vice versa, in less
than I min. Finally, computers, including small desk computers, can swiftly
perform time-consuming calibration and working-curve calculations.

3. TECHNIQUES

Spectrochemical techniques vary widely from laboratory to laboratory,

even from operator to operator. In spite of such variations certain basic
steps are common to all methods. The following discussion is divided into
qualitative, semiquantitative, and quantitative analyses, although clearly
defined divisions do not exist. Whether an analytical result is called quanti-
tative or semiquantitative depends on the particular requirements of the
project and on the type of laboratory available. By convention, however,
7. Optical Emission Spectroscopy 179

the term "semiquantitative" usually refers to results that lie within 1'-3 X
the amount actually present.
Since the dc arc is the common excitation source in geochemical work,
the text refers primarily to this source. Ordinarily the sample is used as the
anode or positive electrode. The positive ions from the discharge migrate
towards the cathode or negative electrode. In certain techniques, however,
the sample may be made the negative electrode. 27 A narrow layer of highly
concentrated ions and atoms (through the recombination of ions and elec-
trons) will then form near the cathode. When this region is focused on the
slit increased sensitivities may be obtained, especially for elements that
ionize readily in the discharge. This procedure, however, is quite sensitive
to operational variations.
A block diagram explaining the various stages of a spectrochemical
determination is given in Fig. 7. The steps involved include sample prepara-
tion, excitation, and data accumulation and treatment. Important aspects
of the diagram are discussed more fully in the text.

I. ROCKS

l
MINERALS

1.
THE "UNKNOWN 'SAMPLE I WATER

A.
I l. CRUSHIN"C
MILLING

I PRECON~'ENTRATION

SAMPLE SPUTTING
PREPARATION

I REDJ~TIONJ I 3. EVAPORATION
HEAVY LIQUID CONCENTRATION

l + 5.
BLENDING
4. ADDlTIVES
FIRE ASSA Y CONCENTRA TION
CHEMICAL CONCENTRATION
ASHING (VEGETATION)

I
I
4. INTERNAL STANDARDS
B. DlLUENTS, BUFFERS

J 6.
EXTERNAL
STANDARDS
FLUXES, BINDERS
CARRIERS
SPECTRO-
CHEMISTRY 5. WEIGHING (SAMPLES)
WEIGHING (ADDITIVES)
7.
f PELLETIZING

I EXCI~~ I
ELECTRICAL
AND PREHEATING
TlDN OPTICAL
AND PARAMETERS 7. ARC, SPARK, LASER
EMISSION CURRENT

C. t ELECTRODES

DATA
9. DISPERSION J FILTERS
STEP SECTORS

I
COLLECTION LENSES
AND 10. LINE SELECTION
EVALUATION 8. GAP
I TIME

! !
ATMOSPHERE

~
I
9. GRATING, PRISM
Il. DISPERSION
PHOTOGRAPHIC RESOLUTION
11.
VISUAL , 13.,
RECORDING ELECTRONIC ORnER
EVALUATlO~
AND RECORDING
11. DIRECT OBSERVATION
DENSITETR' ~ VISUAL EVALUATION OF PLATE,

1
FILM

PLATE, FILM, EMULSION

14. CALCULATIOr-.;s EMULSION CALlBRA TION
D.
IL 14.
THE ANALYTICAL RESULT
II COMPUTER

Fig. 7. Flow sheet for spectrochemical analysis.

180 Armin P. Langheinrich and D. Blair Roberts

3.1. Qualitative Analyses

An experienced spectrochemist should be able to detect and report the
presence of about 50 elements in less than 1 hr. This estimate is based on a
good instrument used independently of other methods. Standard reference
spectra are commercially available for comparison with unknown spectra.
In the simplest case identification is accomplished by visual inspection of
the emission spectrum projected onto a diffusion screen. In the more general
case the spectrum is retained on film or plate and is studied with the unaided
eye, with the help of hand magnifiers, or with projection equipment. For
purposes of direct comparison elements or compounds of elements of inter-
est are normally recorded on the same plate or film as is used for the un-
known(s). With purely electronic detection devices the data output is
dependent upon the number and positions of the direct-reading channels.
In actual practice a strictly qualitative analysis is seldom asked for.
Identification of elements without distinguishing between trace and major
constituents is normally of little value. Consequently, most analyses, even
those called qualitative, include a means of estimating concentrations by
simple classification, such as trace, minor, or major. The value of this
approach increases when the categories are defined numerically, or when
the above groupings are subdivided.

3.2. Semiquantitative Analyses

The upgrading and refining of qualitative work eventually leads to the
so-called semiquantitative analysis. In many geochemical applications this
approach is more important than either qualitative or quantitative work.
Visual comparisons in semiquantitative analyses are commonly re-
placed by densitometric measurements of spectral lines and, quite often, by
background measurements in the vicinity of these lines. Line-to-background
intensity ratios may improve the quality of final results through compen-
sation of uncontrolled variables. Widely practiced visual techniques have
also been reported. Among these are the three-step and six-step methods
reported by the U. S. Geological Survey.28.29 In these techniques stan-
dards are prepared of all elements to be determined, in three or six dif-
ferent concentrations, for several orders of magnitude. In the three-step
method, for example, concentrations of approximately 1, 3, and 7 ppm,
10, 30, and 70 ppm, 100, 300, and 700 ppm, etc., are selected. The
unknown spectra are compared visually with the standard spectra on a
projection microdensitometer. Concentration estimates are obtained by
bracketing. This procedure has given results that compare well with other
7. Optical Emission Spectroscopy 181

semiquantitative techniques. Emulsion calibrations are ordinarily omitted,

but comparison standards are recorded on each plate or film.
Semiquantitative work normally requires the addition of diluents to
the samples. Through this step the matrix and, hence, excitation conditions
are approximated for all samples and standards under investigation. The
effect of buffer materials on the dc arc has been studied by Decker and Eve. 30
Additives in general have been discussed briefly by Leys and Dehm. 31 In
powdered rock samples, powdered graphite in comixture with an alkali-
metal salt is often used. The addition of high-purity quartz is also fairly
common. Quartz not only serves as a diluent, but it is also important in
promoting volatilization of the sample. Quartz additions are generally per-
missible since many commonly analyzed samples contain major amounts of
silicon, so that the spectrochemical determination of this element is of little
interest.
Since the semiquantitative method is based on the comparison of
samples with standards of known concentration, any bias may be reduced
if the standards and samples are similar in composition. When samples
ranging from acidic or intermediate igneous rocks to limestones or dolo-
mites are analyzed with the same set of standards and working curves, the
resulting errors may be quite large. Matrix changes in the samples must,
therefore, be borne in mind when spectrochemical data are to be evaluated.
In normal procedures these errors rarely exceed a factor of two; but
in volatile-element techniques they may change the result by an order of
magnitude. This effect becomes especially apparent in the determination of
tungsten and tin. These elements can be reported in volatile-element analyses
of silicate rocks, but not in volatile-element analyses of limonite-pyrite-
magnetite samples, where the matrix effect is much more pronounced.
Spectrochemical techniques for the determination of low concentrations of
volatile elements have been developed by Adamson et af.32 They used IR
absorption methods for mineralogical identification of residual melt beads.
Interferences due to iron are minimized and volatile-element intensities are
enhanced when the sample composition and the additions to the sample
result in residue beads containing olivine- and pyroxene-group minerals.
The method is applicable to copper, to iron and manganese oxides, to copper
and iron sulfides, and to silica and silicate minerals in general. Recom-
mendations of sample sizes and additives are summarized in Table I.
Various methods exist for presenting geochemical materials to the
discharge. The original samples are preferably ground to - 325 mesh and
homogenized for uniformity. After blending with additives they are added
to electrodes made of graphite or carbon such as ASTM shapes S-4 and
182 Armin P. Langheinrich and D. Blair Roberts

Table I
Composition of Electrode Charge for Volatile-Element Determination a

Sample Additions, mg
Sample weight,
mg SiO. Fe.O. MgO MnO. K.CO.

Iron oxides, hydrates, or

sulfides 75 15 8 2

Manganese oxides or hy-

drates 75 15 8 2

Silica or silicates 65 35 2

Copper oxides or sulfides 40 50 40 8 2

Copper-iron sulfide,
Cu 5 FeS. 50 50 30 8 2

Copper-iron sulfide,
CuFeS. 80 50 5 8 2

a Reproduced from Ana/. Chern. 39 (1967).

S-12.33 These electrodes possess a cavity to receive the powdered sample.

Other types are large-size carrier-distillation electrodes with boiler caps for
obtaining improved sensitivities for volatile elements. Rotating disk, rotat-
ing horizontal platform, and vacuum-distillation electrodes are used in
solution work. Powder samples are sometim~s pressed into pellets prior to
presentation to the discharge. Controlled atmospheres have come into wide
spread use in the trace-element geochemical work of solid materials. For
solutions the so-called plasma-arc or plasma-jet modifications of the dc-
arc technique are of value. The sample solution is introduced into the arc
discharge by an atomizing gas. At the same time a tangentially flowing
auxiliary gas around the atomizer stabilizes and constricts the arc.
Certain elemental substances will not readily reduce in size on grinding.
Malleable native silver is an example. If this occurs, 5- or IO-mg analytical
samples, such as ordinarily used in dc-arc analysis, can produce erratic
results. Fortunately, spectrographic silver determinations on thousands of
rock pulps, soils, and other geologic materials in the writers' laboratories
have shown that these errors are often less serious than may be expected.
Errors in excess of the data in Table II are seldom encountered. In this
table assay values of silver and copper in soils are compared by atomic
7. Optical Emission Spectroscopy 183

Table II
Comparison of Semiquantitative Spectrochemical with Atomic Absorption Data(ppm)

Spectrochemistry Atomic absorption

Sample
Copper Silver Copper Silvera

40 0.3 45 0.5
2 30 0.3 33 0.7
3 400 4.0 475 2.8
4 40 0.6 56 1.0

a Preconcentration by solvent extraction.

absorption and by dc-arc spectrography. The spectrochemical data were

obtained without emulsion calibration or internal standards, but with con-
trolled atmosphere and densitometric measurements.
The problem is more serious in the case of gold or platinum. These
elements mostly occur in the metallic form as discrete particles of malleable
metal. Nevertheless, results obtained with 5-mg samples have proved useful
when fairly large numbers of samples from each presumed occurrence were
analyzed.
The accuracy of spectrochemical data depends greatly on reliable values
for available standards. The development of such standards can be time
consuming and costly. In many instances poor agreement is obtained where
several laboratories are involved. This fact is illustrated in Table III. It has
also been shown by cooperative investigations conducted by the U. S.
Geological Survey and others. 34 ,35 As a result of these studies, standards
such as Granite G-l, Diabase W-1, Sulfide Ore 1, Syenite Rock 1, Granite
G-2, Granodiorite GSP-I, Andesite AGV-l, Peridotite PCC-l, Dunite
DTS-l, Basalt BCR-l, and others have become available and recommended
concentration values have been reported for many elements. 36 - 38
A typical dc-arc spectrographic procedure for geochemical analyses
is shown in Table IV. It can be applied readily to other instrumentation.
In Table V data on wavelengths and detection limits for 51 elements are
presented. These elements are regularly determined at the writers' laborato-
ries. Excitation potentials are calculated mainly from wave numbers reported
by Meggers, Corliss, and Scribner. 39 First ionization potentials are from the
Handbook of Chemistry and Physics. 40 The most sensitive lines of seven ele-
 Table III ...
:
Development of Spectrochemical Standards; Nonspectrographic Assay Data (ppm)

Sample Number of Range Sample Number of Range

Element Element
No. laboratories reported No. laboratories reported

Antimony 5 2 0.3-0.5 Lead 1 6 2-6

12 2 0.6-1.1 15 5 10-52
11 2 1.5-2.6 16 5 96-235
6 5 597-788
Arsenic 4 2 1.5-2.3
6 3.0-12.0 Manganese 2 10-12 ):0
12 5 19.5-25.0 12 3 40-90
..2.
3 3 490-730 ~
11 4 100-161
:"II
Cadmium 5 2 <1.0-1.1 Molybdenum 15 4 <0.5-3.0 ....
DI
11 3 11.0-12.0 ~
Tin 5 2 2.0-3.0 IQ
12 2 5.0-6.0 :::r
Cobalt 2 0.5-2.0 !!.
~
7 3 1.0-3.0 Tungsten 5 3 0.5-3.0 ..n
5 3 17.0-25.0 4 3 1.0-5.0 :::r
DI
Copper 2 7 5-11 12 2 10-19 ~
Q.
6 7 13-26 Zinc 5 13-14 c
15 4 110-180 15 4 35-75 IJI
1 10 210-361 3 4 105-148 !:
12 10 710-952 8 5 712-1016 .
::II
16 4 3000-3300 16 4 3970-4600 0
c:T
14 8 4856-7424 11 4 8290-10700 CD
III
...
7. Optical Emission Spectroscopy 185

Table IV
Example of Semiquantitative dc-Arc Spectrochemical Procedure

Instrumentation

Excitation source Multisource with arc-spark stand

Atmosphere control Enclosed Stallwood jet with gas supply
Spectrograph Three-meter, concave grating, 30,000 lines/in.; reciprocal
dispersion 2.75 A/mm, first order
Recording equipment Plate holder with 4 x lO-in. photographic plates
Developing equipment Mechanical photo processor with timer, automatic agita-
tion, and temperature control
Microdensitometer Projection type, 15 x and 20 x magnification
Calculating equipment Programmable desk calculator

Procedure
Sample preparation Powders, -325 mesh; 5-mg samples, lO-mg additives
(87.5% graphite, 12.5% GeO.); natural rock samples
and synthetic materials as primary standards; second-
ary standards for most routine work; 10 sec pre-
roasting in burner flame
Electrode system Upper electrode (cathode), high-purity graphite, loin.
diameter, l200 included angle tip; lower (sample)
electrode (anode), high-purity graphite, shallow crater,
loin. dIameter (ASTM S-4); 3.5-mm analytical gap;
70% argon-30% oxygen atmosphere, flow rate 5 I/min
Excitation conditions dc-Arc discharge; 12 A, 230 V
Exposure conditions Spectral region 2200-3600 A, first order; slit-width 251';
three-step rotating sector (50, 12!, and 3l% trans-
missions); exposure time 90 sec, complete burn, no
prearc
Photographic processing Emulsion, Kodak SA-l plates; developing, Kodak D-19,
4 min at 71 0 0.3 0 F; short stop, 5 sec, 5% acetic
acid; fixing, 5 min; washing, 5 min, running water;
drying, warm air, about 2 min

ments listed at the end of Table V are ordinarily detected only with vacuum
equipment.
Molecular band spectra are of interest for the indirect determination
of nonmetals. Examples are the use of CaF bands for fluorine and CN
bands for carbon. The determination of water in minerals and rocks through
use of the 3064 A OH-band system has been reported by Quesada and
Dennen. 41
 Table V ~
!XI
Selected Data for Use in Geochemical Analyses en

Detection limits for

various elements, %
Excitation First ionization
Element Symbol Wavelength Spectrum
potential, V potential, V
Normal Special
work procedure

Aluminum Al 3082.2 I 0.001 O.O~.O 5.96

2378.4 I 0.0--5.2
Antimony Sb 2877.9 I 0.002 0.001 1.1-5.4 8.5 ...
3267.5 I 2.0--5.8 2.
:I
Arsenic As 2860.5 0.02 0.002 2.3-6.6 10.5
"lJ
2288.1 I 1.4--6.8
r
Barium Ba 4554.0 II 0.0005 0.0-2.7 5.19 \I)
:I
2335.3 II 0.7-6.0 IC
~
Beryllium Be 3130.4 II 0.0001 0.0--3.9 9.28 ~.
:I
2348.6 I 0.1-5.4 :::!.
Ia 0
Bismuth Bi 3067.7 0.0003 0.0001 O.~.O 8.0 ~
\I)
2898.0 I 1.3-5.6 :I
Boron B 2496.8 I 0.002 0.0~.9 8.26 c.
2497.7 I 0.~.9 c
Cadmium Cd 2288.0 I 0.003 0.0001 0.0-5.4 8.96 IJI
iii
3261.1 I 0.0-3.8 ...
Calcium Ca 4226.7 I 0.001 0.0-2.9 6.09 ::Il
0
3181.3 II 3.2-7.0 C"
Cerium Ce 3201.7 II 0.05 0.9~.7 6.54 ...r+CD
III
4222.6 II 0.4--3.3
Chromium Cr 2843.3 II 0.0005 1.5-5.9 6.74
4274.8 I 0.0-2.9 =-'
Cobalt Co 3453.5 I 0.0002 O.~.O 7.81 0
"C
r+
3449.2 I 0.6-4.2 n'
Copper Cu 3274.0 I 0.0005 0.0-3.8 7.68 !!.
2824.4 I 1.4-6.8 m
3
Gallium Ga 2943.6 I 0.0002 0.1-4.3 5.97 III
III
Germanium Ge 3039.1 I 0.001 0.00001 0.9-4.9 8.09 o
~
2651.2 I 0.1-4.8
en
Gold Au 2676.0 I 0.0003 0.00004 0.0-4.6 9.18 "C
CD
2428.0 I 0.0-5.1 n
Hafnium Hf 3134.7 II 0.01 0.~.3
....r+
0
III
2866.4 I 0.0-4.3 n
0
Indium In 3039.4 I 0.001 0.00002 0.0-4.1 5.76 "C
<
3256.1 I 0.3-3.0
Iron Fe 3020.6 I 0.001 0.0-4.1 7.83
2844.0 I 1.0-5.4
2599.4 II o 0-4.8
Lanthanum La 3337.5 II 0.003 0.~.1 5.6
Lead Pb 2833.1 I 0.0003 0.00005 0.0-4 4 7.38
2663.2 I 1.3-6 0
Lithium Li 4602.9 b I 0.02 1. 8-4. 5 5.36
3232.6 I 0.0-3.8
Magnesium Mg 2852.1 I 0.001 0.0-4.3 7.61
2776.7 I 2.7-7.2
Manganese Mn 2798.3 I 0.001 0.0-4.4 7.41
3070.3 I 2.2-6.2
Mercury Hg 2536.5 I 0.1 0.01 0.0-4.9 10.39
Molybdenum Mo 3170.4 I 0.0005 0.0-3.9 7.35
3194.0 I 0.0-3.9 ...
00

"
 Table V (continued) ....
00
00

Detection limits for

various elements, %
Excitation First ionization
Element Symbol Wavelength Spectrum
potential, V potential, V
Normal Special
work procedure

Nickel Ni 3050.8 I 0.0003 0.0-4.1 7.61

3414.8 I 0.0-3.6
Niobium Nb 3195.0 II 0.005 0.3-4.2
3094.2 II 0.5-4.5
Palladium Pd 3404.6 I 0.001 0.0001 0.8-4.4 8.3
..,l>
3
3421.2 I 1.0-4.6 ~

Phosphorus pc 2554.9 I 0.03 2.3-7.1 10.9 :u

Platinum Pt 3064.7 I 0.001 0.0-4.0 8.88 r-
III
2659.5 I 0.0-4.6 ~
(Q
Potassium K 4044.1 I 0.5 0.1 0.0-3.0 4.32 :T
3446.7 I -3.6 !!.
~
..,
Rhenium Re 3460.5 I 0.005 d O.OOOld 0.0-3.6 o:T
3464.7 I 0.0-3.6
III
Scandium Sc 3353.7 II 0.0005 0.3-4.0 6.7 ~
Q.
Silicon Si 2881.6 I 0.001 0.8-5.1 8.12 C
2568.6 I -6.7
OJ
2435.2 I -5.9
!:
..,
Silver Ag 3280.7" I 0.00002 0.0-3.8 7.54
I 0.0-3.6 :21
3382.9 0
Sodium Na 3302.3 I 0.01 0.0-3.8 5.12 cr
..,CD
PO
3303.0 0.0-3.8 1/1
Strontium Sr 4607.3 I 0.001 0.0-2.7 5.67 :"'"
3464.5 II 3.0-6.6 0
"tI
Tantalum Ta 3311.2 I 0.3 0.7-4.4 r+

II 0.5-5.1
o
2685.2 !.
Tellurium Te 2385.8 I 0.02 0.0006 0.6-5.8 8.96 m
Thallium T1 2767.9 I 0.01 0.0001 0.0-4.5 6.07 ~.
CIl

Thorium Th 2870.4 II 0.05 o 2-4.6 !!!.

0
:;,
2837.3 II
en
3325.1 II 0.5-4.2 "tI
CD
Tin Sn 3175.1 0.001 0.0001' 0.4-4.3 7.30 0
r+
2840.0 I 0.4-4.8
...0
CIl
3168.5 II 0.001 0.2-4.1 6.81 0
Titanium Ti 0
3372.8 II 0.0-3.7 "tI
<
3349.4 II 0.0-3.7
Tungsten W 2947.0 I 0.005 0.0001" 0.4-4.6 8.1
2896.4 0.4-4.6
Uranium U 4244.4 II 0.05 0.0-2.9
2860.5 h
2882.6
2882.7
2882.9 II
3270.1 II 0.0-3.8
Vanadium V 3184.0 0.001 0.0-3.9 6.71
3185.4 0.1-3.9
Ytterbium Yb 3464.4 0.001 0.0-3.3 7.1
3289.4 II 0.0-3.8
Yttrium Y 3327.9 II 0.002 0.4-4.1 6.5
3242.3 II 0.2-4.0 ..
co
CD
 ~
Table V (continued) CD
o
Detection limits for
various elements, %
Excitation First ionization
Element Symbol Wavelength Spectrum
potential, V potential, V
Normal Special
work procedure

Zinc Zn 3345.0 I 0.005 0.001 4.1-7.8 9.36

3345.6 I 4.1-7.8
Zirconium Zr 3273.1 II 0.002 0.2-4.0 6.92
3392.0 II 0.2-3.8

..3
The most sensitive lines of the following elements are below the range of wavelengths ordinarily used:
::l
As 1890.5
Hg 1849.7
:0
r-
P 1774.9 I
::l
CQ
S 1807.4 ~
Sb 2068.4 !!!.
::l
Se 1960.9 .
Zn 2138.6
o
~
I
::l
a Actually two lines of about equal intensity 0.09 A apart. c-
b Iron interference is serious on Li 4602.9. O
C Molecular bandheads at 3255.3 and 3270.5 can also be used. III
a Sensitivities in siliceous matrix only; not in MoS. matrix. iii
Mn 3280.8 may interfere. .
f Sensitivity in siliceous matrix only. :II
0
9 Sensitivity in siliceous matrix only; interelement effects are severe. co
" Arsenic interference. CD
Note: The term "special procedure" in most cases refers to volatile-element techniques; I refers to atom lines, II refers to ion lines. Ul
...
7. Optical Emission Spectroscopy 191

3.3. Quantitative Analyses

The transition from semiquantitative to quantitative spectrographic

techniques is accomplished through emulsion calibration and the addition
of one or more internal standards. Their purpose is the compensation of
variables that are beyond satisfactory control by the spectrochemist. Com-
pensation is accomplished through the use of an intensity ratio of the
analytical line to the internal standard line. Internal standards are added
directly to the sample material and are selected to have volatility, excitation
properties, and line positions similar to the elements to be determined. The
internal standard must be an element not of analytical interest. The choice
of internal standards has been well discussed by Ahrens and Taylor. 19 A list
of recommended internal standards as related to the volatilization character-
istics of many elements is given in Table VI.

Table VI
Volatilization Behavior of 51 Elements and Recommended Internal Standards

Volatile elements
For the best sensitivity volatile-element methods are required (unmixed powders,
low currents, recording of early arcing periods only)
Group IA - Li, Na, K Group VIA - Te
Group IlIA - Ga,a In, Tl Group IB - Cu,a Ag,a Au
Group IV A - Ge, Sn, Pb Group lIB - Zn, Cd, Hg
Group VA - P, As, Sb, Bi
Recommended internal standars b : Ge, In, Li (Group IA)

Medium-volatile and non-volatile elements

Conventional methods are satisfactory (usually total consumption, carbon admix-
tures, high currents)
Group IIA - Be, Mg, Ca, Sr, Ba Group VB - V, Nb, Ta
Group IlIA - B, Al Group VIB - Cr, Mo, we
Group IV A - Si Group VIIB - Mn, Ree
Group IIIB - Sc, Y, La Group VIII - Fe, Co, Ni, Pde, Pt
Group IVB - Ti, Zr, Hf Lanthanides - Ce, Yb
Actinides - Th, U
Recommended internal standardsd : Pd, Ge, Be (Group IV A)

a This is somewhat more sensitive by volatile-element techniques, but ordinarily determined

by nonvolatile element procedures.
b Others used occasionally are Na and K .
. Sensitivity in siliceous materials can be improved by use of volatile-element techniques.
d Others used occasionally are Ca, Fe, Sr, and Ti.
192 Armin P. Langheinrich and D. Blair Roberts

Most of the semiquantitative and quantitative dc-arc methods follow

the "total energy" technique reported by Slavin in 1938 in which samples
are burned to completion. 42 During this process elements volatilize selectiv-
ely. This has led to valuable modifications of the technique in which cer-
tain groups of elements are favored over others. For example, the detection
limits of volatile elements can be lowered by recording only the early excita-
tion periods and by omitting powdered graphite from mixtures with diluents.
Graphite suppresses selective volatilization.
When photographic methods are used in quantitative work, calibration
of the photographic emulsion becomes necessary. Many methods of calibra-
tion have been discussed in the literature. 43 ,44 Variation of the two-step
method is illustrated in Fig. 8. Steps of different densities are produced on
the photographic record through the use of filters or rotating step sectors
near the slit. The transmissions of two line steps of known relation are
measured. This is repeated for several lines of different densities within the
spectral region of interest. In this way data are accumulated for plotting
the so-called preliminary curve. This curve is utilized in the preparation of
the emulsion calibration curve. In a specific example successive readings are
taken by starting with a transmission value of one on the ordinate. The
corresponding abscissa value is 2.5. If 2.5 is now used as the new ordinate
value, a second abscissa value is obtained, and so on. This procedure results
in a sequence of numbers as shown in Table VII.
L A. EXPOSURE J IRON SPECTRI1M (fWO STEP<; ONLy 1

l R. PRELIMINARY CURVE I
ABSCISSA
PERCENT TRAN~OF LIGHT STEP

ORDINATE
PERCEN r TRANSMiSSiON OF DARK STEP

I C. EMULSION CALIBRATION CURVE I

ABSCISSA
RELA TiV'Ei'N'T'ENSITlES
(FROM KNOWN TWO_STEP RELATION)

ORDINATE
T RAN~MiSSiONVA LUES
(FROM PRELIMINARY CURVE)
LOG

L D. ANALYTICAL WORKING CURVE j

ABSCISSA
CONCENTR.A'TiONO"F ELEMENT
IN STANDARD SAMPLE

ORDINATE
INTE~UES
(FROM EMULSION CALIBRATION CURVE)
LOG

Fig. 8. Example of photographic calibration.

7. Optical Emission Spectroscopy 193

Table VII

Step No. Relative intensity Transmission, %

1.66 96.4
2 1.66" 89.0
3 1.663 69.3
4 1.66' 36.7
5 1.665 16 0
6 1.666 6.5
7 1.66' 2.5
8 1.668 1.0

These numbers represent transmissions that would be obtained if eight steps

were used on one line instead of two steps on a group of lines. Each number
is related to the others by the relative transmissions of the two steps. In
the table one step is 1.66 X darker than the previous step. When the cor-
responding relative intensity and transmission values are plotted logarithmi-
cally against each other, the emulsion-calibration curve results. It permits
the analyst to relate photographic response to the intensity of light emitted
by a sample. The so-called multistep method is similar, but calibration is
done on several steps of a single line. Such methods have the disadvantage
of requiring an optical alignment that will evenly illuminate the entire
length of the spectral line. Emulsion-calibration curves are influenced by
factors such as wavelength, emulsion characteristics and history, photo-
graphic processing, exposure time, slit width, and intermittency effects.
The homologous-line method uses pairs of lines of known intensity
ratios obtained from the literature. This method does not require spectral
lines to be uniformly illuminated throughout their vertical extent. It depends,
however, upon intensity values which are not always known correctly. A
recent improvement in the two-step method involves a beam splitter which
separates the two steps horizontally instead of vertically so that uniform
illumination is not necessary.45 Emulsion calibration has been studied by
Parodi and Burch. 46

4. GEOCHEMICAL APPLICATIONS
Geochemical applications of optical emission spectroscopy are as
diverse as the various fields of geochemistry. They can, however, be divided
into three principal categories: (a) element abundance and distribution,
(b) environmental studies, and (c) mineral exploration.
194 Armin P. Langheinrich and D. Blair Roberts

4.1. Element Abundance and Distribution

Spectrochemistry is well suited for studies of element abundance or
distribution and has been used for this purpose for several decades. Early
work by Clarke 47 and Clarke and Washington 48 was completed prior to
the full development of spectrochemical techniques. It preceded Gold-
schmidt's49 significant contributions to both geochemistry and spectro-
chemistry. More recently Green,50 Turekian and Wedepohl,51 and Vi no-
gradov 52 ,53 have added to the Goldschmidt data. Computer treatment of
spectrochemical results has come into widespread practice. Horn and
Adams 54 applied spectrochemical abundance data to theoretical studies of
the origin of the earth, its continents and sedimentary basins, and smaller
areas.

4.2. Environmental Studies

Spectroscopy finds application wherever it is necessary to develop data
on the chemical composition of materials. One example of environmental
analysis involves the study of trace elements in soils or waters at potential
industrial construction sites. Prior to construction it is desirable to estab-
lish base levels of environmental conditions so that subsequent changes
caused by the industrial activity can be determined. Such studies have legal
implications, especially in relation to pollution and health. For example,
the construction of a lead-zinc mill or smelter may be unwise in a populated
area with preexisting high natural levels of lead in soils and waters. A second
category of environmental studies involves the monitoring of trace elements
in food supplies, water supplies, and soils where food crops are grown.
The important elements in such studies are toxic elements such as arsenic,
cadmium, lead, and selenium. Others, such as cobalt, iron, manganese, and
zinc, are also of interest. Contributions by Warren et al. 55 to this field are
noteworthy.

4.3. Mineral Exploration

The use of spectrochemistry in mineral exploration is limited only by
the ability and imagination of the scientist conducting the survey. Rock-
chip samples are frequently taken as character samples for a prospect or for
the various rock types in the area under investigation. In detailed work
samples are typically taken 250 or 1000 ft apart.
Soil surveys are conducted in a similar way. In this case various horizons
are sampled. In a common system of identification these horizons or layers
7. Optical Emission Spectroscopy 195

of the well-developed soil profile are labelled A, B, and C. In simple terms,

the surface soil, including the organic-rich layer, is called the A horizon.
The B horizon is the intermediate layer just below A. The parent material
from which the soil was derived is the C horizon. These major designations
are further subdivided, sometimes in considerable detail. In general, the
organic-rich A horizon, if present, is high in trace elements. The resulting
data, however, are often erratic since they depend on localized concentra-
tions of organic materials. When it is well developed the B horizon is used
more frequently. Metal enrichment makes it highly suitable for geochemical
prospecting. The C horizon is also useful. In comparison with the other
horizons it is more uniform and more closely related to the composition
of the underlying rock. Its trace composition is frequently lower. This type
of work can extend from the collection of a few characteristic samples in
each area to detailed grid surveys. A pioneering text on soil and plant
analysis was published by Mitchell in 1948. It was most recently updated
in 1964. 56
Surface-water samples are frequently taken from streams or less fre-
quently from lakes or swamps in an effort to relate the trace-metal content
of these waters to the terrain from which they have been derived. Special-
purpose studies can be made of the runoff from an area during different
seasons or under changing conditions. Ground water and formation water
(connate water) can be analyzed to determine the source material for sedi-
ments in a given formation. Spectrochemical methods are well suited to
these applications, especially when information on several elements is
desired. If only one or two elements are to be studied simpler methods are
often more practical.
Stream-sediment sampling is an extremely important part of geo-
chemical prospecting and is probably used more than any other approach,
especially in the early reconnaissance stages of a survey. It is also applied
to the follow-up of anomalies found during the reconnaissance stage. When
the investigation proceeds to more detailed work, soil sampling or rock-
chip sampling would ordinarily be utilized. The proper interpretation of
anomalously high metal content in stream sediments is a complicated
subject and one that may receive increased emphasis during the next few
years. Emission spectroscopy can be used to provide more complete infor-
mation than is gained by the usual geochemical determination of only one
or two elements. A preliminary spectrochemical analysis will produce data
which can be followed up by more precise methods for selected elements.
Cruft57 has published on trace-element determinations in soils and stream
sediments by an internal-standard spectrographic procedure.
196 Armin P. Langheinrich and D. Blair Roberts

The analysis of drill cores or drill cuttings is very similar to work on

rock chips. The analyses may be on composite samples made up, for example,
of 50- or 100-ft intervals, or may be on samples selected on the basis of
their apparent composition.
In biogeochemical prospecting samples usually analyzed consist of
ashed plant material from deep-rooted plants such as mesquite or conif-
erous trees growing in the area. Recently, Bedrosian et af.5 have reported
on the direct-emission spectrographic analysis of trace elements in biological
substances. They were able to achieve detection limits of 1 ppm or less on
26 elements without ashing or preconcentration. Prior to the analysis the
samples were merely dried, mixed, and pelletized.
In an unusual application of spectrochemistry Hendricks et al. 59 have
applied spectrographic techniques in conjunction with wet chemical methods
to the investigation of metalrich oozes occurring in closed deeps in the
median trough of the Red Sea. Y oung60 has reported spectrographic data
on cores from the Pacific Ocean and the Gulf of Mexico. Degens et al. 61
have reported criteria for differentiating marine and fresh-water shales
based on spectrochemical data. Their results indicate that in the region
studied, marine and fresh-water shales can be differentiated by means of
quantitative spectrographic determinations of boron, rubidium, and gallium.
Joensuu 62 has discussed spectrochemical methods in geochemistry in general
with emphasis on the investigation of marine sediments. He reported inter-
esting and significant details related to technique, including buffers, internal
standards, automated sample presentation, direct readers, double-arc and
rotating-disk procedures, the cathode-layer method, etc.
The application of specialized apparatus and techniques to geochemical
problems has been reported in a number of interesting papers. Maxwe1l 63
and Snetsinger and Keil 64 discussed the use of lasers in mineral analysis,
Cruft and Giles 65 the use of the direct reader as a geochemical tool, and
Shaw 66 and Joensuu and Suhr67 the Stallwood jet in silicate and general
rock, mineral and related work, respectively.

5. LITERATURE
One or more textbooks and reference volumes are necessary in any
geochemical laboratory with optical emission equipment. Among the more
prominent pUblications are Spectrochemical Analysis by Ahrens and Taylor,19
Spectrochemical Procedures and Semiquantitative Spectrochemistry by Har-
vey,68.69 Principles and Practice of Spectrochemical Analysis by Nachtrieb,70
and, with respect to spectroscopy in general, Practical Spectroscopy by Har-
7. Optical Emission Spectroscopy 197

rison, Lord, and Loofbourow 2 Of equal importance are Harrison's M.I.T.

Wavelength Tables,17 Moore's71 publications of atomic energy levels, and
tables of spectrum lines by Zaider et al. 72 The theory of spectrochemical
excitation is discussed in depth by Boumans. 73
Among periodicals three distinct groups stand out. The first one is
analytically oriented; it includes Analytical Chemistry, Applied Spectroscopy,
and Spectrochimica Acta, Part B, Atomic Spectroscopy. A second group
covers geological and mineralogical publications. These journals occasional-
ly contain papers based on spectrochemical data or articles written to cover
specific techniques. Among these publications are Economic Geology, Can-
adian Mineralogist, and Geochimica et Cosmochimica Acta. Related to this
group are periodicals in the field of engineering, mining, and mineral in-
dustries. A third group is bibliographic in character. Analytical Chemistry,
for example, carries biennial analytical reviews. In its April 1968 review on
emission spectrometry 676 references were cited. In comparison, the April
1966 issue contained 351 references. Spectrochemical Abstracts has been
published by Hilger and Watts for the years 1933 to 1951. The American
Society for Testing Materials in its Index to the Literature on Spectrochemical
Analysis (four parts) covered the period 1920 to 1955. In the first year, 1920,
five articles were cited. By 1955 this number had risen to 579. Additional
volumes of this valuable publication cannot be expected in the very near
future, but the biennial review mentioned earlier should help to bridge the
gap from 1955 to the present. Chemical Titles, published semimonthly by
the American Chemical Society in a key word arrangement, allows rapid
selection of significant articles from the current literature. Chemical Ab-
stracts furnishes more detailed information on a delayed basis.
Among government publications related to spectrochemistry are mono-
graphs by the U.S. Bureau of Standards and bulletins by the U.S. Geological
Survey.74,75
Other valuable sources of information include programs of scientific
meetings and conferences which quite often contain abstracts of all papers
presented, and literature obtainable from instrument manufacturers.

6. EVALUATION OF THE METHOD

As evidenced by its widespread use and multiplicity of applications,
optical emission spectroscopy is a powerful analytical technique. In the
writers' laboratories the spectrograph was found nearly indispensable in a
particular comprehensive geochemical program. The technique has the
advantage of rapidity and simplicity when large numbers of determinations
198 Armin P. Langheinrich and D. Blair Roberts

are to be made on one sample, or when large numbers of samples must be

processed. When photographic methods are used, permanent records of
analyses become available for later checking or for the further determina-
tion of additional elements. Sample sizes are small. A reasonably complete
semiquantitative analysis for 3~O elements can be done rapidly on a I-mg
sample using routine methods. Even smaller samples can be handled by
special procedures. High absolute sensitivities and low detection limits can
be achieved by this technique.
One of the most significant advantages of optical emission work in
geochemical applications is the certainty with which elemental identifications
and estimates of abundance can be made. The possibility of spectral inter-
ference in elemental identification is often stated as a disadvantage of the
method relative to chemical methods or other techniques. There are many
possibilities of spectral interference, but in actual practice there is no danger
of incorrect identification at the levels normally encountered if the work is
performed carefully. For example, the identification of 10 ppm beryllium
or silver in a rock sample is a completely unambiguous determination. The
examination of several spectral lines of each element eliminates any possibility
that an identification may be incorrect because of the coincidence of lines.
On the negative side, spectrochemistry is a destructive process. The
sample is not available for further investigation. This, however, may not
be important since the amount of sample used is small, and often large
amounts of sample are kept in storage.
Since spectrochemical equipment is available throughout a wide price
range, the question of high-vs-moderate cost may arise. The main advan-
tage of the more expensive instrumentation lies in convenience and speed.
With regard to "good results" the skill and patience of the operator are
generally of greater importance than the price of the unit. The equipment
and techniques are complex and do require trained personnel, especially
with respect to method development and spectrum interpretation.
Examples of the analytical precision obtained in routine semiquanti-
tative work are presented in Table VIII. These data stem from intralabora-
tory analyses performed at minimal two month intervals and reflect a pro-
cedure such as outlined in Table IV.

7. FUTURE DEVELOPMENTS
In spite of the relative age of spectrochemistry in comparison with
other instrumental techniques, there is today considerable room and need
for progress. Some areas of potential development are pointed out below.
7. Optical Emission Spectroscopy 199

TABLE VIII
Examples of Long-Term Reproducibility by a Semiquantitative DC Arc Procedure
(in Percent)

Number of Range Standard Coe fficien t

Element Mean
analyses reported deviation of variation

Aluminum 6 5-8.6 6.3 1.40 22

8 5-9 6.6 1.3 20
Barium 6 0.08-0.20 0.14 0.042 30
Beryllium 6 0.0003-0.0004 0.00032 0.000041 13
Cobalt 6 0.001-0.004 0.0024 0.0012 50
Copper 8 0.06-0.10 0.079 0.013 16
Chromium 6 0.004-0.020 0.010 0.0057 57
Gallium 6 0.001-0.002 0.0018 0.0004 22
Lanthanum 4 0.010-0.015 0.011 0.0022 20
Lead 5 0.0028-0.0050 0.0038 0.0009 24
8 0.0007-0.0010 0.00086 0.00011 13
Lithium 3 0.0010-0.0025 0.0018 0.00076 42
Magnesium 6 1.0-3.0 1.8 0.72 40
Manganese 5 0.008-0.03 0.014 0.008 57
Molybdenum 5 0.02-0.04 0.029 0.007 24
9 0.0010-0.0020 0.0015 0.00025 17
Nickel 6 0.0023-0.008 0.0052 0.0022 42
Niobium 4 0.00J-0.002 0.0012 0.0005 42
Silicon 6 26-34 31 2.7 9
Silver 6 0.00023-0.0006 0.00043 0.00016 37
Sodium 6 1.5-2.0 1.8 0.19 11
Tin 6 0.0002-0.003 0.0012 0.0010 83
Titanium 6 0.21-0.60 0.38 0.13 34
Vanadium 5 0.009-0.020 0.014 0.0056 40
Yttrium 6 0.002-0.003 0.0023 0.0005 22
Zinc 8 0.04-0.07 0.056 0.013 23
Zirconium 6 0.009-0.03 0.018 0.0078 43

As was mentioned earlier, the value of mobile spectrochemicallabora-

tories has been established. The impetus for their use has come largely
from the programs of the U.S. Geological Survey and from foreign efforts,
notably in Australia, Canada, and the USSR. In the future such truck-
mounted or trailer-mounted facilities will become more common in mineral
200 Armin P. Langheinrich and D. Blair Roberts

exploration, also in environmental studies. Current trends are toward the

installation of direct-reading spectrometers. Eventually, spectrograph-
spectrometer combinations might be installed.
Direct readers interfaced with computers have been introduced into
permanent laboratory facilities. Future trends may see these installations
refined so that the data output would be in the form of contoured geo-
chemical maps. Matrix effects will be corrected through programmed inter-
element compensation and arc-temperature measurements. Thompson
et al. 76 and Decker and Eve 77 have reported on this topic.
A significant development in direct-reader instrumentation was reported
by Margoshes. 78 This particular spectrometer-computer system, still in the
product-development stage, reportedly can measure 2048 individual wave-
lengths simultaneously and will print out analytical results in numerical form.
Improvements are being made in the stability of the excitation discharge,
especially the dc arc, as reported by Gordon. 79 Servocontrolled discharges
combined with better focusing elements, perhaps interferometerically or
electronically controlled, can be expected to improve the performance of
emission equipment. Such advances may better the precision of the semi-
quantitative type of multielement analysis by as much as an order of mag-
nitude and could open up new areas of application.
There will still be a need for photographic recording to obtain maximum
sensitivity and reliability in the case of unusual element occurrences. The
photographic process could be speeded up through the use of self-developing
dry-process film, at least for qualitative determinations. An instrument for
producing a photographic record simultaneously with direct reading of
about 20 common elements would be of considerable interest. A system for
computer analysis of photographic optical emission spectra has been re-
ported by Helz et al. 80
Much work remains to be done on systems and techniques for atmos-
pheres other than air and pressures other than atmospheric. The use of
controlled static atmospheres as contrasted to flow systems offers a promise
of improved sensitivity.
We can expect further miniaturization of equipment, at least of elec-
tronic components, as advances continue in this area. The use of laser
vaporization in microprobe analysis has already been introduced. Its appli-
cations can be expected to expand rapidly in the years ahead.
Many of these anticipated advances can be achieved with existing
technology. The workers active in this field must provide the stimulus to
bring about these developments within the limits of the economics involved
in each application.
7. Optical Emission Spectroscopy 201

REFERENCES
1. Weise, E. K., History and Origin, in The Encyclopedia of Spectroscopy, G. L. Clark,
ed., Reinhold, New York, 1960.
2. Harrison, G. R., Lord, R. C., and Loofbourow, J. R., Practical Spectroscopy, Pren-
tice-Hall, Englewood Cliffs, N. J., 1963.
3. Kirchhoff, G., and Bunsen, R., Chemical Analysis by Meam of Spectral Observations
(in German), Pogg. Ann. Physik u. Chem. 110, 161 (1860).
4. Gerlach, W., "The Correct Execution and Interpretation of Quantitative Spectrum
Analysis" (in German), Z. Anorg. Chem. 142, 383 (1925).
5. Goldschmidt, V. M., and Peters, C. (in German), Nachr. Ges. Wiss. Gottingen,
Math. Phys. Klasse, 165, 257 (1931).
6. Goldschmidt, V. M., and Peters, C. (in German), Nachr. Ges. Wiss. Gottingen,
Math. Phys. Klasse, 2, 360, 528 (1931).
7. Goldschmidt, V. M., Berman, H., Hauptmann, H., and Peters, C. (in German),
Nach. Ges. Wi5S. Gottingen, Math. Phys. Klasse 3, 141, 235, 278, 371 (1933).
8. Goldschmidt, V. M., Hauptmann, H., and Peters, c., Rare Elements in Rock Analysis
(in German), Naturwiss. 21, 363 (1933).
9. Goldschmidt, V. M., and Peters, c., The Geochemistry of Arsenic (in German),
Nachr. Ges. Wiss. Gottingen, Math. Phys. Klasse, N. F. Fachgr. IV 1, II (1934).
10. Goldschmidt, V. M., Bauer, H., and Witte, H., Geochemistry of the Alkali Metals. II.
(in German), Nachr. Ges. Win. Gottingen, Math. Phys. Klasse, N. F. Fachgr. IV 1,
39 (1935).
II. Fassel, V., History of Flame Emission Spectrochemical Methods, paper presented at
the Seventh National Meeting, SAS, Chicago, Ill., May 14, 1968.
12. Walters, J. P., Thirty Years of Emission Spectrochemical Development. The Spark
Discharge, paper presented at the Seventh National Meeting, SAS, Chicago, Ill.,
May 14, 1968.
13. Strock, L. W., Thirty Years of Development of D. C. Arc Spectrochemical Analysis
Methods, paper presented at the Seventh National Meeting, SAS, Chicago, Ill.,
May 14, 1968.
14. Herzberg, G., Atomic Spectra and Atomic Structure, Dover, New York, 1944.
15. Richtmyer, F. K., Kennard, E. H., and Lauritsen, T., Introduction to Modern Physics,
McGraw-Hill, New York, 1955.
16. McNally, J. R., Atomic Spectra, in Handbook of Physics, E. U. Condon and H.
Odishaw, eds., McGraw-Hill, New York, 1967, pp. 7-38-53.
17. Harrison, G. R., M.l.T. Wavelength Tables, John Wiley & Sons, New York, 1939.
18. Grotrian, W., Graphical Presentation of the Spectra of Atoms and Ions with One,
Two, and Three Valence Electrons. I, II, J. Springer, Berlin, 1928.
19. Ahrens, L. H., and Taylor, S. R., Spectrochemical Analysis, 2nd Ed., Addison-Wesley,
Reading, Mass., 1961.
20. Guide to Scientific Instruments, Science, p. I58A, Nov. 28, 1967.
21. Buyers' Guide, Industrial Research, May 15, 1967.
22. Scribner, B. F., and Margoshes, M., Emission Spectroscopy, in Treatise on Analytical
Chemistry, Part I, I. M. Kolthoff and P. J. Elving, eds., Vol. 6, pp. 3372-3379.
23. Gold Rush, 1967-Style, Spectrum Scanner, Jarrell-Ash Company, 22, 2 (1967).
24. Stallwood, B. J., Air Cooled Electrodes for the Spectrochemical Analysis of Powders,
J. Opt. Soc. Am. 44, 171 (1954).
202 Armin P. Langheinrich and D. Blair Roberts

25. Boyd, B. R., and Goldblatt, A., Atmosphere Excitation in Non-Enclosed Spark Stands,
paper presented at the Pittsburgh Conference on Analytical Chemistry and Applied
Spectroscopy, March 1963.
26. Danielsson, A., Lundgren, F., and Sundkvist, G., The Tape Machine, I, II, and III,
Spectrochim. Acta 1959, 122.
27. Strock, L. W., Spectrum Analysis with the Carbon Arc Cathode Layer, A. Hilger,
London, 1936.
28. Barnett, P. R., An Evaluation of Whole-Order, ! Order, and 1 Order Reporting in
Semiquantitative Spectrochemical Analysis, U.S. Geol. Survey Bull. 1084-H (1961).
29. Grimes, D. J., and Marranzino, A. P., Six-Step Field Standards for Semiquantitative
Analysis of Rocks and Soils, u.S. Geol. Survey, Denver, Colo., paper presented at
the 1968 Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy,
March 1968.
30. Decker, R. J., and Eve, D. J., Appl. Spectros. 22, 13, 263 (1968).
31. Leys, J. A., and Dehm, R. L., Semiquantitative Industrial Chemical Analysis, in
The Encyclopedia of Spectroscopy, G. L. Clark, ed., Reinhold, New York, 1960,
pp. 285-286.
32. Adamson, D. L., Stephens, J. D., and Tuddenham, W. M., Application of Mineral-
ogical Principles and Infrared Spectra in Development of Spectrographic Techniques,
Anal. Chem. 39, 574 (1967).
33. Designation of Shapes and Sizes of Graphite Electrodes, in General Test Methods,
ASTM Standards Part 30, ASTM, Philadelphia, Pa. 1967, p. 334.
34. Stevens, R. E., et al., Second Report on a Cooperative Investigation of the Composi-
tion of Two Silicate Rocks, U.S. Geol. Survey Bull. 1113, 1 (1960).
35. Report of Nonmetallic Standards Committee of the Canadian Association for Applied
Spectroscopy, Appl. Spectros. 15, 159 (1961).
36. Webber, G. R., Second Report of Analytical Data for CAAS Syenite and Sulphide
Standards, Geochim. Cosmochim. Acta 29, 229 (1964).
37. Ingamells, C. 0., and Suhr, N. H., Chemical and Spectrochemical Analysis of Stand-
ard Silicate Samples, Geochim. Cosmo Chim. Acta 27, 897 (1963).
38. Flanagan, F. J., U.S. Geological Survey Silicate Rock Standards, Geochim. Cosmo-
chim. Acta 31, 289 (1966).
39. Meggers, W. F., Corliss, C. H., and Scribner, B. F., Tables of Spectral-Line Intensities,
NBS Mon. 32, Parts I and II, 1961.
40. Weast, R. C., ed., Handbook of Chemistry and Physics, 45th Ed., The Chemical
Rubber Co., Cleveland, Ohio, 1964, p. E-41.
41. Quesada, A., and Dennen, W. H., Spectrochemical Determination of Water in
Minerals and Rocks, Appl. Spectros. 21, 155 (1967).
42. Slavin, M., Quantitative Analysis Based on Spectral Energy, Ind. Eng. Chem. Anal.
Ed. 10, 407 (1938).
43. Churchill, J. R., Techniques of Quantitative Spectrographic Analysis, Ind. Eng.
Chem. Anal. Ed. 16, 653 (1944).
44. Photographic Photometry in Spectrochemical Analysis, in ASTM Methods Chemical
Analysis of Metals, ASTM, Philadelphia, Pa., 1956, pp. 570-593.
45. Feldman, c., A Beam-Splitter for Use in Calibrating Spectrographic Emulsions, paper
presented at the Seventh National Meeting, SAS, Chicago, Ill., May 16, 1968.
7. Optical Emission Spectroscopy 203

46. Parodi, J. A., and Burch, W. G., Jr., A Study of Photographic Emulsion Calibration
Techniques, G.E., Hanford Atomic Products Operation, Richland, Wash., HW-28803,
July 27, 1953.
47. Clarke, F. W., The Data of Geochemistry, U.S. Geol. Survey Bull., 770 (1924).
48. Clarke, F. W., and Washington, H. S., The Composition of the Earth's Crust, U.S.
Geol. Survey PP 127, 1924.
49. Goldschmidt, V. M., Geochemistry, Clarenden Press, Oxford, 1954.
50. Green, J., Geochemical Table of the Elements for 1959, Geol. Soc. Am. Bull. 70,
1127 (1959).
51. Turekian, K. K., and Wedepohl, K. H., Distribution of the Elements in Some Major
Units of the Earth's Crust, Geol. Soc. Am. Bull. 72, 175 (1961).
52. Vinogradov, A. P., The Regularity of Distribution of Chemical Elements in the Earth's
Crust, Geokyhimiya 1, 6 (1956) (Eng!. Trans!. by Atomic Energy Research Establish-
ment, Harwell, Berkshire, 1957).
53. Vinogradov, A. P., The Geochemistry of Rare and Dispersed Chemical Elements in
Soils, 2nd Ed. (Eng!. Trans!. from the Russian by Consultants Bureau, New York,
1959).
54. Horn, M. K., and Adams, J. A. S., Computer-Derived Geochemical Balances and
Element Abundances, Geochim. Cosmochim. Acta 30, 279 (1966).
55. Warren, H. V., and Delavault, R. E., A History of Biogeochemical Investigation in
British Columbia, Transactions of the Canadian Institute of Mining and Metallurgy
53, 236 (1950).
56. Mitchell, R. L., The Spectrographic Analysis of Soils, Plants, and Related Materials,
Commonwealth Bur. Soil Sci. Tec. Common. No. 44,1948; reprinted with addendum,
1964.
57. Cruft, E. F., Trace Element Determinations in Soils and Stream Sediments by an
Internal Standard Spectrographic Procedure, Econ. Geol. 59, 458 (1964).
58. Bedrosian, A. J., Skogerboe, R. K., and Morrison, G. H., Direct Emission Spectro-
graphic Method for Trace Elements in Biological Materials, Anal. Chem. 40, 854
(1968).
59. Hendricks, R. L., Reisbick, F. B., Mahaffey, E. G., Roberts, D. B., and Peterson,
M. N. A., Chemical Composition of Sediments and Intersticial Brines from the
Atlantis II Discovery and Chain Deeps, in Hot Brines and Recent Heavy Metal
Deposits in the Red Sea, E. T. Degens and D. A. Ross, eds., Springer Verlag, New
York, in press.
60. Young, E. J., Spectrographic Data on Cores from the Pacific Ocean and Gulf of
Mexico, Geochim. Cosmochim. Acta 32, 466 (1968).
61. Degens, E. T., Williams, E. G., and Keith, M. L., Environmental Studies of Carbon-
iferous Sediments, I. Geochemical Criteria for Differentiating Marine and Freshwater
Shales, Bull. Am. Assoc. Petrol. Geologists 41, 2427 (1957).
62. Joensuu, O. I., Spectrochemical Methods in Geochemistry, paper presented at the
Seventh National Meeting, SAS, Chicago, Ill., May 15, 1968.
63. Maxwell, J. A., The Laser as a Tool in Mineral Identification, Can. Mineral. 5, 727
(1963).
64. Snetsinger, K. G., and Keil, K., Microspectrochemical Analysis of Minerals with
the Laser Microprobe, Amer. Mineral. 52, 1842 (1967).
65. Cruft, E. F., and Giles, D. L., Direct Reading Emission Spectrometry as a Geo-
chemical Tool, Econ. Geol. 62, 406 (1967).
204 Armin P. Langheinrich and D. Blair Roberts

66. Shaw, D. M., Spectrochemical Analysis of Silicates Using the Stallwood Jet, Can.
Mineral. 6, 467 (1960).
67. Joensuu, O. I., and Suhr, N. A., Spectrochemical Analysis of Rocks, Minerals, and
Related Materials, Appl. Spectros. 14, 101 (1962).
6S. Harvey, C. E., Spectrochemical Procedures, Applied Research Laboratories, Glendale,
Calif., 1950.
69. Harvey, C. E., Semiquantitative Spectrochemistry, Applied Research Laboratories,
Glendale, Calif., 1964.
70. Nachtrieb, N. H., Principles and Practice of Spectrochemical Analysis, McGraw-Hili,
New York, 1950.
7l. Moore, C. E., Atomic Energy Levels, U. S. Bur. Standards Circ. 467, Vol. I, 1949,
Vol. II, 1952.
72. Zaidel, A. N., Prokof'ev, V. K, and Raiskll, S. M., Tables of Spectrum Lines, Per-
gamon Press, New York, 1961.
73. Boumans, P. W. J. M., Theory of Spectrochemical Excitation, Plenum Press, New
York, 1966.
74. Bastron, H., Barnett, P. R., and Murata, K K, Method for the Quantitative Spectro-
chemical Analysis of Rocks, Minerals, Ores, and Other Materials by a Powder dc
Arc Technique, U. S. Geol. Survey Bull. 1084-G (1960).
75. Myers, A. T., Havens, R. G., and Dunton, P. J., A Spectrochemical Method for the
Semiquantitative Analysis of Rocks, Minerals and Ores, U. S. Geol. Survey Bull.
1084-1 (1961).
76. Thompson, G., Paine, K., and Manheim, F., A Flexible Computer Program for
Evaluation of Emission Spectrometric Data, Appl. Spectros. 23, 264 (1969).
77. Decker, R. S., and Eve, D. J., dc Arc in Emission Spectrography, IV. Correction
for Matrix Effects, Appl. Spectros. 23, 497 (1969).
7S. Margoshes, M., The Digilab 204S-Channel TVS Spectrometer, paper presented at the
Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March
1970.
79. Gordon, W. A., Improvement of Analytical Precision Using a Servocontrolled dc Arc
with Current Feedback, paper presented at the Fifth National Meeting, SAS, Chicago,
Ill., June 1966.
SO. Helz, A. W., Walthall, F. G., and Berman, S., Computer Analysis of Photographed
Optical Emission Spectra, Appl. Spectros. 23, 50S (1969).
Chapter 8

ATOMIC ABSORPTION
L. R. P. Butler and M. L. Kokot
National Physical Research Laboratory
Pretoria, South Africa

1. INTRODUCTION

The relatively new method of atomic absorption spectrometryl has efficiently

filled a gap that has existed for many years, by providing a simple, inexpen-
sive technique whereby a wide range of elements in many different types of
materials may be determined accurately and precisely. The large number of
atomic absorption instruments now in use in geochemical and other labora-
tories bears testimony to the usefulness and popularity of the technique.
Many papers have been published describing applications of atomic
absorption in geochemistry. These include the analysis of rocks,2-4,22 ores,5
waters,6 brines,7 etc. Several books have also been published on atomic ab-
sorption,8-10 although only oneIl describes its use exclusively for the analysis
of geologic materials.
The purpose of this chapter is to describe the method and to give a
brief outline of its use for the analysis of silicates. It is not intended to give
a general literature survey. The reader is referred to the papers and books
mentioned for further and more detailed references. In particular the articles
and bibliographies which appear in the Atomic Absorption Newsletter
published by Perkin-Elmer (Norwalk, Conn., USA) provide an invaluable
source of information.
Because of the high sensitivity of atomic absorption spectrometry, it
has been applied in the majority of cases for the determination of minor
and trace elements, but it has been shown that the technique is also ade-
quately precise and accurate for the determination of major elements.

205
206 L. R. P. Butler and M. L. Kokot

The requirement that the sample must be brought into solution is

sometimes time consuming, but sample dissolution of silicate-base materials
is generally not considered difficult. Inherent interference effects present with
the determination of certain elements, notably some of the alkaline earths,
and other elements that form refractory oxides, may create difficulties. The
suggestion of using the nitrous oxide-acetylene flame l2 and a better under-
standing of the processes taking place in flames has enabled many of these
difficulties to be overcome.
Atomic absorption techniques today offer very distinct advantages in
respect of capital outlay, simplicity, accuracy, precision, speed, sensitivity,
and cost of analysis. Most of the instruments offered by manufacturers,
especially those in the medium and upper cost brackets, are reliable and
well suited for the application to geochemical problems.

2. PRINCIPLES OF ATOMIC ABSORPTION

Atomic absorption depends on the phenomenon whereby atoms of an

element are able to absorb electromagnetic radiation. This occurs when the
atoms are unionized and unbonded to other or similar atoms. The wave-
length of light at which atomic absorption proceeds is specific for the type
of atom and occurs at the resonant frequencies. These correspond to the
electronic transitions between the first allowable energy level and the ground
state of the atoms. Atomic absorption also occurs for some elements with
spectral lines other than those originating in the ground state, but the ab-
sorption probability, and thus the analytical sensitivity, is correspondingly
lower. The reader who is interested in the theoretical explanation of the
phenomena of atomic emission and absorption is referred to the literature
dealing with spectral theory.l.13.14
A schematic diagram of the essential components of an atomic absorp-
tion instrument is given in Fig. 1.
In practice the sample solution is vaporized by means of a nebulizer
and the aerosol is passed into a flame. The temperature of the flame is
sufficiently high to further reduce the droplets by dehydration and dissocia-
tion into an atomic form so that the analytical elements in the vapor exist
mostly, and preferably, as neutral unbound atoms in the ground state. If
the flame temperature is high atoms may be raised to excited or even to
ionized states. If this happens the population density of the ground state
is decreased and the lower limit of detection is accordingly decreased.
The atomic vapor is illuminated by a light source radiating the char-
8. Atomic Absorption 207

r ~j(&FL ~~~IN.2. - - - - ,..---, AMPLIFIER

: r------------ ~
~ 8
READOUT
I I S.upP~Y
I I SPECTROMETER
r - -1- t- - - - - - - - - - - - - --:
: lSOURCE :
I I

:Ill~__ ~~i~~~~~~
,
,
IIi:::::..-,
HOLLOW CATHcloE
~Ll'~ ~~L! ____ L_A.!A~ ___ _:

I ,
ATOMIZER
IL _____________________ J ~~~~ ,

Fig. 1. Schematic diagram of atomic absorption apparatus.

acteristic light of the analytical element. A spectrometer is usually used to

select the resonance spectral line of the element and to measure its intensity.
When a solution is sprayed the atoms of the analytical element present
absorb a portion of the light at the selected wavelength; the decrease in the
intensity of the analytical line is measured by the spectrometer.
The usual procedure for determining the relationship between absorp-
tion and atomic concentration in the solution is by measuring the absorb-
ance of a number of solutions containing known concentrations of the
analytical element and then to draw an analytical graph or working curve
by plotting absorbance against concentration. This graph is linear at low
concentrations but may deviate from linearity at higher concentrations.
It is preferable to measure absorbance rather than absorption. Atomic
absorption follows Beer's law namely,

/
- = 10gKCl
/0

where /0 and / are the intensities before and after passing through the flame,
respectively, K is the absorption coefficient, C is the concentration, and I is
the flame path length. If absorbance i.e., 10g(I//o), is plotted against con-
centration on linear paper, the graph is nearly linear, but if absorption 1/10
is plotted, the analytical graph is curved.
There are a number of different steps in the procedure, each with
variables which can contribute towards the final result. Whenever possible
these variables are kept constant to improve the precision and accuracy of
the method.
208 L. R. P. Butler and M. L. Kokot

3. APPARATUS

3.1. Source
The source should radiate the characteristic spectral lines of the ana-
lytical element with sufficient intensity to enable the lines to be measured
with ease on a relatively low-dispersion monochromator. Further require-
ments of the source are as follows.
(a) The emitted spectral lines should have narrow profiles, free from
self-absorption.
(b) The spectrum should be pure with no interfering lines from other
elements within the band pass of the spectrometer.
(c) The emission of light should be constant.
(d) The lifetime of the lamp should be long and its cost should be low.

The most successful source for atomic absorption has been the sealed-
off hollow-cathode lamp. These lamps, manufactured for a wide variety of
elements, are obtainable at reasonable prices. Other sources which have
been used with some success for atomic absorption purposes are: electrode-
less discharge tubes,I5.16.30 vapor-discharge lamps,17 continuous emission
sources,I8 fiames,19 and dc arcs. 20 Although a considerable amount of research
is being conducted to develop new and more efficient sources, it is likely
that the hollow-cathode lamp will continue to be used in atomic absorption
instruments for analytical purposes for some time to come. In many in-
stances lamps radiating the characteristic light of more than one element
may be obtained.
Hollow-cathode lamps should be used according to manufacturers'
instructions and the quoted maximum current should not be exceeded. The
lifetimes of these lamps have been extended considerably in recent years by
applying modern manufacturing techniques and it is not unusual to find
lamps operating for considerably longer than 1000 h. The main reason for
lamp failure is "cleanup" of the carrier gas. By this process sputtered
particles from the metallic cathode trap gas atoms and thus reduce the
pressure. Other reasons for lamp failure are out-gassing of metal compo-
nents in the lamp and diffusion of molecular gases into the tubes. Most
lamps are provided with "getters" and may be reconditioned by heating
the getters. If the getter has been placed on the anode it may be heated by
reversing the current through the lamp. This effectively extends the lifetime
of the lamp.
Where vapor-discharge lamps are used as sources it is important that
8. Atomic Absorption 209

the current through the lamp should be reduced to increase sensitivity. The
sources used for atomic absorption spectrometry have been exhaustively
discussed in several books and publications. 8 ,21,45

3.2. Absorption Flame

The sensitivity, precision, and accuracy of atomic absorption depends

to a large extent on the type of flame used. This is especially important to
the geochemist who is essentially working with solutions containing large
quantities of various metal salts, as interferences are therefore most likely
to be encountered.
The premixed acetylene-air flame is used for elements that are easily
atomized, while the acetylene-nitrous oxide flame is used for refractory
elements or those prone to interferences. By using a slit-type burner the
flame can be made long and thin, giving the long path length suitable for
atomic absorption. The sensitivity of the method depends on the number
of atoms in the flame as well as the path length through which the light
must pass. Most commercial instruments supply premixed, laminar-flow
burners. With the latter the premixed gases issue from the burner orifice in
a nonturbulent stream. The manner in which the flame burns plays a role
in the precision with which absorption measurements can be made. When
the ratio of fuel gas to oxydizing gas is high, the flame is termed fuel rich
and is "softer" and burns at a lower velocity. Flames of this type are more
easily affected by movement of the ambient air and may be less stable
than more oxydizing flames which are "stiffer" and burn at a higher
velocity.
Forced-diffusion burner-nebulizers, often called total consumption
burners, are sometimes used for viscous solutions. These burner-nebulizers
were actually developed for flame-emission spectrometry. Their short path
length and wide spectrum of particle sizes make them unsuited for atomic
absorption where high sensitivity and accuracy are required, as in the case
of geochemical analyses.
The fuel gases used for the premixed burner are very important as they
play an important role in controlling the temperature and the combustion
properties of the flame. The gas mixtures which have been found most
suitable are summarized in Table I, together with their maximum measured
temperatures. For geochemical analyses the most popular gas mixtures are
acetylene-air, acetylene-nitrous oxide, and in cases where selenium and
arsenic are to be determined, hydrogen-air or hydrogen-argon with diffused
air maintaining combustion. The latter flames are more transparent to UV
210 L. R. P. Butler and M. L. Kokot

Table I
Maximum Temperatures of Atomic Absorption Flamesa

Maximum Height above

Flame type temperature, burner edge,
C mm

1. Acetylene-air, fuel-rich 2060 8-12

Acetylene-air, stoichiometric 2150 2.0
Acetylene-air, lean 2145 0.5
2. Propane-butane-nitrous oxide, rich 2480 10-14
Propane-butane-nitrous oxide, stoichiometric 2550 4
Propane-butane-nitrous oxide, lean 2535 0.5
3. Acetylene-nitrous oxide, stoichiometric 2795 4
Acetylene-nitrous oxide, lean 2740 0.5

a As measured by Butler and Fulton."

light in the Schumann or vacuum spectral region of wavelengths shorter

than 2000 A.
An important part of the burner is the nebulizer. The nebulizer must
convert the sample solution into a cloud of small droplets, the finest of
which are conducted into the flame. There are many different types of
nebulizers, but the most popular is the concentric pneumatic type. It has
two anuli concentric about a common axis. Air, or the support gas, forced
through the outer anulus creates sufficient suction in the central tube to lift
a solution from the reservoir. The solution is drawn through the fine central
tube and is broken up by the action of the fast-flowing gas. It is important
that the droplets which are formed are small; particles less than 5 p, in
diameter are preferred. The size of the droplets reaching the flame significant-
ly influences the measurements. If a large spectrum of droplet sizes reaches
the flame the process of atomization may be erratic because of the time
required by the droplets to reach the final stages of dissociation. Not only
will this cause absorption readings to fluctuate, but chemical interference
effects (see Section 7.1) may be significantly greater. Larger droplets are
prevented from reaching the flame by using spray chambers which throw
them onto the walls, or by using spoiler tubes or glass bulbs to cause them
to coalesce. Droplets smaller than 15 p, do not coalesce easily and may
reach the flame without adhering to the walls of the spray chamber of the
burner.
8. Atomic Absorption 211

The greater the amount of solution that is fed into the flame, and the
larger the number of atoms present, the better the sensitivity will be. Un-
fortunately, there is a limit to the amount of aerosol that can be fed into
the flame, as too much aerosol may cool or even quench it.
A well-designed nebulizer will have a sample uptake of 1-3 ml per min
and an efficiency, i.e., percentage of droplets less than 15 fl, of between 8
and 12% using water at 20C as the test solution. The efficiency of a nebu-
lizer may be increased by spraying organic solvents. 24 Well-designed neb-
ulizer systems do not, however, give a very large increase in efficiency when
organic solvents are sprayed.

3.3. Spectrometer
The role of the spectrometer is to separate the resonant spectral line
from other lines being radiated by the lamp source. If as many elements a&
possible are to be determined the monochromator should transmit light
from about 1900 to 10,000 A. The monochromator's dispersion and resolu-
tion should be sufficient to separate the resonant line from other lines which
do not show atomic absorption characteristics. In some cases, such as nickel,
for example, where this is not possible with the low-dispersion monochro-
mators, the sensitivity may be limited. If light sources such as the high-
intensity lamp which boosts the intensity of resonance lines are used, this
difficulty may be overcome. 25
There are two dispersive media used in monochromators, namely, the
prism and the grating. The former gives approximately uniform intensity
across the spectrum, but the dispersion and resolving power change with
wavelength. The grating gives (nearly) linear dispersion of the spectrum,
but the intensity of reflectance varies with wavelength.
Prisms, with their poor dispersion at longer wavelengths, may give
poorer sensitivity for elements having their resonance lines in the visible
portions of the spectrum. The sodium doublet, for example, at 5890 and
5895 A should be resolved for maximum sensitivity.
Modern gratings have what is known as "broad-blaze" angles, indicat-
ing that their reflectance is more uniform over a wider spectral range. In-
struments with gratings may give better sensitivities for some elements with
resonance lines at longer wavelengths. In some cases, however, sensitivity
is lost, because higher lamp currents must be used to obtain sufficient
spectral intensity for measurement.
It is imperative that the monochromator has reproducible slit openings.
Because wavelength setting is critical for atomic absorption work, the
212 L. R. P. Butler and M. L. Kokot

wavelength adjustment should be precise and free from backlash. A coarse

and fine adjustment is desirable, the coarse setting being used for rapid
changeover from one element's wavelength to another and the fine setting
for adjusting the wavelength critically so that the peak of the resonance
line is measured. If only one setting is provided a compromise must be made.

3.4. Detector
It is desirable to use low source-lamp currents to reduce source self-
absorption for maximum analytical sensitivity. Although the resonance
lines are usually the most intense at currents as low as 5-15 mA, the actual
intensity of these lines is very low, and it is desirable to use highly sensitive
detectors. Photomultiplier tubes are extremely well suited for this purpose.
The sensitivity of most light detectors, including photomultipliers, depends
largely on wavelength. For the spectral range used in atomic absorption
spectroscopy a single photomultiplier that is selected to give low noise is
often sufficient. With a monochromator having a narrow-blazed grating it
may be necessary to have available a second photomultiplier tube, sensitive
to the wavelength region which is poorly reflected by the grating.

3.5. Amplifier and Measuring System

Photomultiph~rs seldom have sufficient photocurrent to enable direct
measurement; therefore an electronic amplifier is required. The system pro-
posed by Russell, Shelton, and Walsh17 makes use of an ac amplifier, able
to amplify only a very specific frequency. By introducing a mechanical
modulator between the source and the flame or by switching the lamp (see
Fig. 1) the light is modulated to the frequency to which the amplifier is
tuned. This system has the advantage that any light emitted by the flame,
at the wavelength of the resonance line, will be continuous, or at least not
at the frequency of the amplifier. This light will give rise to a dc signal in
the detector and will thus not be amplified. Many modern electronic systems
make use of "lock-in" amplifiers which have a very narrow frequency
acceptance and very low noise factor.
Some instruments make use of a double-beam system where the light
from the hollow-cathode lamp is cyclically deflected around and through
the flame. An advantage here is that fluctuations caused by the light source
are reduced. 26
Readout may be by a direct-indicating meter or galvanometer, or by
digital voltmeters. It is advantageous to have some means for damping
8. Atomic Absorption 213

meter fluctuations or averaging results, as with some types of analysis

fluctuations may be severe.
Scale expansion of the indicator meter is also a decided advantage,
especially if very low concentrations are to be determined.
Digital concentration readout systems have proved useful, especially
where rapid analyses are a requirement. 27 Small function generators are
used to "linearize" calibration curves. This readout is then calibrated to
read concentrations directly. This obviates the need for drawing a calibra-
tion graph. Function generators are more commonly used with automatic
systems, where samples and standards are automatically nebulized on a
conveyer system and the results typed out directly in units of concentra-
tion. 28

4. SENSITIVITY, ACCURACY, AND INTERFERENCES

These factors play an important role in an analytical method, and as

such must be discussed. As with many spectrochemical techniques a good
deal depends on the care with which the instruments are operated and the
analyses carried out.

4.1. Sensitivity
Gatehouse and Willis 29 suggested that a concentration glVlng 1.0%
absorption or 0.004 absorbance should be accepted as the limit of detection.
This is a very practical quantity, well suited to the measurement methods of
atomic absorption, although not according to the accepted practices in
spectrochemistry. A concentration giving a signal of intensity 2 X that of
the statistical fluctuation or standard deviation of the background signal
is usually defined as the detection limit. The latter definition is more correct
scientifically, but unless a recorder is used to enable the background and
signal fluctuations to be recorded it is difficult to determine in atomic
absorption under normal working conditions. With a very stable measuring
system of low noise, the latter definition will yield a somewhat lower de-
tection limit than what is really the case. Many atomic absorption spectro-
scopists, therefore, prefer the more practical detection-limit figures accord-
ing to the 0.004 absorbance criterion.
Sensitivity is defined as the slope of the analytical curve at any point.
Most atomic absorption working curves are linear up to absorbance values
of about 0.5. Gatehouse and Willis quoted concentrations giving 50%
absorption, which, together with their lower limit I % figures, lead to a
214 L. R. P. Butler and M. L. Kokot

good estimate of the sensitivity of atomic absorption for the various ele-
ments.
Analytical range is of equal importance to the analyst, although it may
vary considerably from one person to another, depending on the require-
ments. It is felt that the lower realistic analytical range should commence
where the signal is clearly distinguishable above background, and correspond
to an absorbance reading of 0.01-0.02 A without scale expansion, i.e., about
3 X the detection limit.
The upper analytical range figures given in Tables III-XI are based on
the criterion that the limit is selected where the graph curves to approxi-
mately half its normal linear sensitivity. This varies from element to element.
On an average the analytical range for an element by atomic absorption
spreads over two orders of magnitude, e.g., 1-100 ppm.

4.2. Accuracy
The accuracy depends largely on the reliability of the standards and
on the inherent interference effects. As with other comparative analytical
methods great care must be taken to ensure that freshly standardized refe-
rence materials and standards are used. Should interference effects be pre-
sent a standard material should be selected to ensure the same interferences
as are present in the sample.
The accuracy which may be achieved will also depend on the precision
with which measurements can be made. Atomic absorption results can
usually be repeated, within a 95% confidence level, to between 0.5 and 2.5%.
When the precision is poorer than this it is usually an indication of either
instrumental failure or faulty technique.

4.3. Interferences
Interference effects are the main causes of inaccurate atomic absorption
results. Several workers who in the past experienced the severe effects that
occur with some elements and element groups made use of analyzed rock
standards, which, hopefully, had the same interferences as the samples.
This procedure is very useful, providing such standards are available.
The advent of the acetylene-nitrous oxide flame with its high tempera-
ture and special chemical properties has changed the situation. Many ele-
ments that are prone to severe interferences, such as calcium, may now be
determined without difficulty. Interferences which have been recognized
may be divided into the following categories: chemical, molecular, and
spectral.
8. Atomic Absorption 215

4.3.1. Chemical Interferences

Complex chemical reactions occur in the various regions of the flame

with the result that a metallic species may appear in them from the aspi-
rated sample solutions.
When a pure aqueous solution containing the analytical element is
atomized some of the atoms will become excited, some ionized, and some
may become chemically involved with other species or remain bound as
molecules. If A is the total possible number of atoms, then

A = A' + XA + A* + [A]
where A' are ions, A* are excited atoms, XA are chemical combinations
involving A where X may be either anions or cations, and [A] are free atoms
available for atomic absorption.
If an element P is present in the solution, the equilibrium between the
various states may also lead to interference.
Enhancement occurs if [A] increases:
(i) P + XA -+ PX + [A] i.e., the interfering element P reacts
preferentially with X to release atoms
of element A.
(ii) +e
P -+ pi ionization enhancement, but the inter-
then A' + +e
pi -+ P' + [A] fering element has a low ionization
potential; equilibrium shifts and the
ionized atoms of the analytical element
A become neutralized.
(iii) P + A* -+ P* + [A] excitation enhancement. The same ef-
fect is achieved if the flame temperature
decreases as a result of the combustion
properties changing.

Depression will, on the other hand, occur if [A] decreases:

(iv) P + [A] -+ PA straight chemical combination.
(v) P + XA + [A] -+PA + XA chemical replacement.
(vi) P* + [A] -+ A* + P excitation transfer.
(vii) [A] -+ A' + e high flame temperature causes ioniza-
tion.
(viii) [A] -+ A* excitation depletion of ground-state
atoms.
216 L. R. P. Butler and M. L. Kokot

The following methods may be used for overcoming interference.

For depression:
(a) High-temperature flames may be employed to break down molec-
ular bonds. Although A' will be higher, [A] will be constant even
if P is present.
(b) Flames may be used with combustion properties that will prevent
the formation of those molecules which do not dissociate easily
such as refractory oxides, of which Al 20 3 is a good example.
(c) Add an excess of suppressing agent S which will preferentially
react with the interfering element P, i.e., where

P + [A] + S-+PS + [A]

and P and S have a higher affinity for each other than P and [A].

For enhancement:
(a) Add an excess of an easily ionized element R so that the ionization
potential of R is lower than S. This will effectively flood the flame
with free electrons to prevent ionization of the analytical element:

R-+R' +e
R' + e + A' -+ R' + [A]
(b) Combinations of ionization suppression and releasing agents.
The suggested methods for the different elements discussed in Section 5
have been found to be successful.

4.3.2. Molecular Interference

Angino and Billingsll have reported an interference caused by non-
atomic absorption effects. When a complex rock solution is measured against
aqueous standards absorption caused by bands of molecular species may
occur. This may lead to results that are too high. The effect is worse for
elements that have resonance lines at shorter wavelengths or in bandheads.

4.3.3. Spectral Interference

Spectral interference is caused by the light source radiating spectral
lines which are not resolved by the monochromator. The lines may be non-
resonant lines of the analytical element, such as nickel, or they may originate
from the carrier gas of the lamps. Neon and argon are usually used, and
8. Atomic Absorption 217

although they have relatively "clean" spectra there are cases where their
lines are nearly coincident with analytical lines, as in the case of argon
(2175 A) with lead (2170 A).8
When a highly luminous acetylene-nitrous oxide flame is used with a
spectrometer having a wide entrance slit the high-intensity continuous
radiation may saturate the photomultiplier detector, or cause the first
amplifier stage to be overloaded. This can lead to severe fluctuations or
erroneous readings. It is wise to use as narrow a slit as possible when a
highly luminous flame is used.
Provided the correct procedures as recommended by the instrument
manufacturers are followed, and, provided good lamps are used, spectral
interference is seldom encountered. It does not necessarily lead to inaccura-
cies, but rather to loss of sensitivity and poor precision.
Once proper precautions have been taken to prevent interference the
analyst may proceed to analyze samples with confidence. These measures
include the addition of the necessary buffer or releasing elements and the
use of the correct flame.

5. ANALYSIS OF GEOLOGIC SAMPLES BY ATOMIC

ABSORPTION
In Chap. 3 a number of methods are given by which rock samples may
be brought into solution. A scheme that has been found very useful for
atomic absorption is outlined in Fig. 2. The following three sections are
devoted to the complete analysis of rocks by atomic absorption.

5.1. Major Elements

The conditions listed below for the determinations of the various major
elements may differ for different instruments. Analytical limits quoted are
for the final sample solution, under normal conditions, such as the burner
at full length, etc. A reduction in sensitivity may be achieved by using less
sensitive lines or turning the burner through an angle to present a shorter
flame path. This method has been found very useful and is recommended.

5.1.1. Silicon
The development of the acetylene-nitrous oxide flame has enabled
silicon and other refractory elements to be determined. Amos and Willis 12
used a nitrogen-oxygen mixture with acetylene, but found the nitrous oxide
flame to be superior and safer.
218 L. R. P. Butler and M. L. Kokot

DRIED POWDERED SAMPLE

CONCENTRATIONS MINOR a TRACE CONCENTRATIONS
:ULTRAMAFIC ROCKS 200 mo ROCK +
AND MINERALS 10ml(H F: HC 104 '
:200 mil (or Ie.. ' 2 .I
+:IOml HF(+5mIHCI04 '
~-:--:~"="""'"=,.,

DILUTE 20-50X
ANALYSE

5,. AI.Co. (Fe, No,K'

Fig. 2. Scheme for the dissolution of silicate rocks.

Sample dissolution is best done by lithium metaborate or other fusions,

and provided the silicon is not too concentrated the solutions will keep for
several days. It is advisable, however, to spray solutions as soon as possible
after preparation. Standards should be prepared by the fusion of oven-dried
silicon dioxide with lithium metaborate, and then standardized chemically.
For low concentrations of silicon, high-pressure techniques (see Chap-
ter 3) may be employed to obtain more concentrated solutions. Standards
prepared from waterglass or by dissolving silicon dioxide in hydrofluoric
acid (see Table II) are satisfactory.
A fuel-rich acetylene-nitrous oxide flame gives the highest sensitivity.
The region of the flame where maximum absorbance occurs is in the "red-
feather" region. The authors have noticed that if the burner becomes very
hot, a "memory" effect appears, i.e., absorbance values keep changing.
The effect is aggrevated if the solution contains a high salt content. Cooling
of the burner improves the situation, since partial clogging of the burner
slot may result.
Interferences from other major elements in silicate rocks have not been
reported to date.
The most sensitive line for silicon is the 2516-A line. A normal hollow-
cathode lamp will provide sufficient intensity with most spectrometers,
even if only a narrow entrance slit is used to reduce flame radiation. Table III
lists the conditions for silicon. The detection limits and analytical ranges for
 Table II !'O
~
Preparation of Standards ...0
3
n'
Resultant Stock Stock value, ~
Element Stock Procedure C"
compound g/liter ppm 1/1
0
..
"C
Li Li.CO. Dry in oven at 110C; dissolve in dilute HCla LiCI 1.656 640
...o
::2
Li metal Dissolve in excess HCI; evaporate to dryness; LiCI 3.910 640
weigh out chloride; standardize (chloride)
Na NaClb Dissolve in freshly distilled water and store in NaCI 1.627 640
polyethylene containers
K KCI Dissolve in distilled water, after drying at 1l0C KCI 1.2210 640
KNO. Ditto KNO. 1.655 640
Rb RbCI Dissolve in water; standardization is recommended; RbCI 0.9055 640
pipette 25 ml to Pt dish; add 1 drop H.SO"
evaporate to dryness, and heat to constant weight
Mg Mg metal Dissolve metal in minimum amount appropriate MgCI. 0.640 640
acida (if HCI)
Ca CaCO. Dry at 1l0oC; dissolve in minimum amount dilute CaCl. 1.598 640
HCI
Sr SrCO. Dissolve 1.078 g SrCO. in 300 ml water plus 2 ml SrCI. 1.078 640
conc. HCla and dilute to volume
Ba BaCI. Dissolve 0.9747 g anhydrous BaCl. dried at 250C BaCI. 0.9747 640
for 2 h; dilute to volume
BaCO. Dry at 110C; dissolve in dilute HCla BaCl. 0.9196 640
Al Al metal Dissolve in HCI with one small drop of Mg added AICl a 0.640
as catalyst; filter off the mercury after the Al has (AI.Cl6 )
dissolved N
..
co
 Table II (continued)
N
N
0
Resultant Stock Stock value,
Element Stock Procedure
compound g/liter ppm

Si SiO. Dissolve in HF + HNO. without heating; use SiF 4 1.369 640

Teflon or polypropelene containers; standardize
SiO. Melt with 5 g of LiBO. at 1000C or Na.B.O, at SiF 4 1.369 640
5000 C in carbon crucible; pour redhot melt into
polypropelene beaker containing 3% HNO.
solution while stirring (cold); standardize
Na.SiO. Dissolve in cold water; filter and standardize Na.SiO. 2.7814 640
Cr K.Cr.O, Dissolve in water, dilute to volume K.Cr.O, 1.461l 640
Cr metal Dissolve in HCl, or dilute H.so 4a CrCI. 0.640 640
Mo (NH.).Mo,O ... Dissolve in water and add few drops ammonia (NH.).Mo,O .. 1.1777 640
4H.O solution to make alkaline and standardize by
gravimetric method
Mn MnO. Dissolve in concentrated HCla MnCl. 1.013 640
Fe Fe metal Dissolve fresh iron filings (or pure iron wire) in FeCI. 0.640 640 r-
IO ml 6N HCI and dilute to volume ::0
Ni NI metal Dissolve fresh turnings in minimum amount nitric Ni(NO.). 0.640 640 :c
acid and dilute to volume ID
I:
Pd Pd metal Dissolve fresh metal filings in aqua regia a PdCl. and 0.640 640 r+
;-
Pd(NO.). ...
III
Pt Pt metal Ditto PtCI. and 0.640 640 :I
Pt(NO.). c..
Cu CuSO., 5H.O Dissolve fresh crystals in water; standardize CUS04 2.523 640 s:
Cu metal Dissolve fresh filings in dilute HNo.a plus a few Cu(NO.). 0.640 640 r-
drops HCl ;:0:
0
~
0
a Use as little acid as possible. r+

b All salts should be dried.

8. Atomic Absorption 221

Table m
Conditions for the Determination of Silicon

Approx. relative
Wavelength, A
sensitivity

Analytical lines 2516 1.0

2507 0.36
2529 0.32
2217 0.26
2211 0.15
2208 0.06
Lamp Hollow-cathode (10-20 rnA), high-intensity
lamps are used with some instruments
Band pass 2-4 A (see text)
Flame C 2H 2-N 20 fuel rich
Burner 5 cm x 0.4 mm slot
Burner height Light should pass through "red feather" of
flame
Analytical range, ppm, 2-300
Detection limit, ppma 0.9

a The detection limit is usually defined as a signal of magnitude twice the standard deviation
of the background fluctuations.

this and all subsequent tables apply to the most sensitive conditions of the
line and flame settings. More precise readings may be obtained for high
concentrations by turning the burner.

5.1.2. Aluminum
Aluminum was the first element to be determined using the acetylene-
nitrous oxide flame. 30 Since then several papers have been published on the
determination of aluminum. Its determination in cement together with
titanium and silicon31 has been an important development. Aluminum has
been determined in soils32 and in high-silicon materials. 33 In all cases good
agreement with accepted chemical results is reported.
Absorbance is found to be strongly dependent on the amount of acety-
lene in the flame. This is at a maximum with a strongly luminous fuel-rich
flame. The region of the flame at which maximum absorbance occurs is
critical and occurs in the region at the base of the flame. For maximum
sensitivity it was found that by limiting the size of the cathode in the flame
higher absorbance readings may be obtained.
222 L. R. P. Butler and M. L. Kokot

Most elements cause enhancement, although calcium, zinc, copper,

lead, magnesium, sodium, phosphate, and sulfate ions may cause depression.
This is considerably reduced or eliminated by the addition of a releasing
agent such as lanthanum and an element with a low ionization potential
such as potassium or caesium.
For the rapid determination of aluminum in concentrations of over 5%,
the fusion method of rock dissolution may be used, but for lower concentra-
tions the acid-dissolution technique has been found to give more satisfactory
results. In Table IV conditions are summarized for the determination of
aluminum.

5.1.3. 'ron
Iron is a major element in most rocks and exists in a variety of forms,
namely, free metallic iron, primary oxides in several valency states, sulfide
minerals, primary silicate minerals, and secondary compounds. Atomic
absorption spectrometry provides a rapid and sensitive method for the

Table IV
Conditions for the Determination of Aluminum

Sensitivity
Wavelength, A
factor

Analytical lines 3092.7 1.0

3961.5 0.78
3082.2 0.70
3944.0 0.50
2373.4 0.30
2367.1 0.25
2575.4 0.12
Lamp Hollow cathode, 10 rnA
Band pass 2-4 A
Flame C 2H.-N20 fuel rich (luminous)
Burner 5 cm x 0.4 mm slot
Burner height (below optical axes) 20-25 mm, adjust for max. absorbance

With La
Without La
(1000 ppm)

Analytical range, ppm 3.0-500 1.0-200

Detection limit, ppm 0.90 0.6
8. Atomic Absorption 223

determination of this important element. Several authors describe the use

of atomic absorption in routine assays for iron, nickel, copper, and cobalt
in ores. Price and Ragland 34 have determined iron in pure quartz and
Galle 35 has determined iron, manganese, etc., in several rock types. Iron
has also been determined in sc1eractinian corals and other skeletal carbonate
material in recent sediments. 36 Silicate rocks have also been analyzed for
iron by several other analysts. 3 ,37-40
Interference with iron is mostly enhancement and it is usually small.
A fuel-rich acetylene-air flame shows less enhancement and gives a higher
absorbance signal than an acetylene-nitrous-oxide flame.
Both fusion and acid-digestion solution methods are suitable for the
determination of iron. For basic rocks, where concentrations are high,
considerable dilution is necessary, but for acid rocks, a dilution factor of
100 or 1000 usually suffices. Turning the burner and using the less sensitive
lines listed in Table V reduces the sensitivity and hence the need for high
dilution.
Table V also shows the conditions required for the determination of
iron in rocks.

Table V
Conditions for the Determination of Iron

Sensitivity
Wavelength, A
factor

Analytical lines 2483.3

2522.8 0.5
3020.6 0.2
3719.9 0.1
3589.9 0.05

Lamp Hollow cathode 10-15 rnA

Band pass 1-3 A
Flame C 2 H 2-air (slightly fuel rich)
Burner 10 cm x 0.5 mm slot
Burner height a IOmm
Analytical range, ppm 0.2-25
Detection limit, ppm 0.1

a Distance from cathode image to burner top.

224 L. R. P. Butler and M. L. Kokot

5. 1.4. Calcium
Calcium has become one of the elements most frequently determined
by atomic absorption spectrometry. Geochemically, calcium is considered
an important element and is included in most rock analyses. Its determina-
tion is of importance for rock classification as well as for geochemical
studies of minerals and rocks.
Several interference effects have been reported in the literature, and
many of these are associated with the types of flames and gases used. Figure 3
shows the interference of phosphorus on calcium in various flames-hence
a strong dependence on the flame type.
Aluminum also interferes with calcium, but this can be eliminated by
using the acetylene-nitrous oxide flame and a releasing agent. Figure 4
shows how various amounts of lanthanum will reduce and eliminate this
interference. The hotter acetylene-nitrous oxide flame requires much less
lanthanum and also gives higher sensitivity when lanthanum is added.
Calcium absorbance depends markedly on the region of the flame
through which the light passes as well as on the fuel flow.
If calcium concentrations are sufficiently high, sample fusions are more
rapid and can be used for rapid dissolution, provided lanthanum is added
to overcome interference from silicon. Acid digestion can also be used for

02

PROPANE - BUTANE I N20

UJ
~Ol

al
a::
o
(f)
al

PROPANE - BUTANE I AIR

100 200 300 400 500 600

CONC P (ppm)

Fig. 3. Diagram illustrating the interference

of phosphorus on calcium in various flames.
(No releasing agent is present.)
8. Atomic Absorption 225

5 ppm Co
2000 ppm La

50 8 1000 ppm La

--
45 75
---"!t---""'1I"- C-Hj!"2Jlair
- _ 2000 ppm La

4 7
"-<..... .....
.....
..... '1(

\
\
35 \
\
\ \
,
I&J
U
Z ~
\
~ 3 6 I, \
a:
o
II) H ' C2 H 2 laor
, r:===7
,, ,
IX)

\ \
25 55
, \
\ , \
200 ppm La

2
, , ....
.... , .... _ 200ppm La

15
--~ -----x

1 4~----~--~~----~------~~--~
100 200 300 400 500
CONe. AI (ppm)

Fig. 4. Influence of aluminum on calcium in various flames, with

various amounts of lanthanum added as releasing agent.

dissolution. The addition of caesium or potassium to standards is also

recommended. The conditions for the determination of calcium are shown
in Table VI.

5.1.5. Magnesium
The first published application of atomic absorption was the determina-
tion of magnesium by AllanY Magnesium is subject to interferences similar
to calcium but to a lesser degree, and analytical conditions are also similar.
 N
N
01
Table VI
Conditions for the Determination of Magnesium and Calcium in Silicate Rocks

Magnesium Calcium

Wavelength 2852 A 4220 A

Lamp Hollow cathode 5-10 rnA Hollow cathode 5-10 rnA

Band pass 2-4 A 1-3 A

Flame Max. freedom from Max. freedom from
Max. sensitivity Normal condition
interference interference

C.H.-air C.H.-N.O C.H.-air C.H.-N.O

Burner 10 cm x O. 5 mm slot 5 cm x 0.4 mm slot 10 cm x 0.5 mm slot 5 cm x 0.4 mm slot
add 200-1000 ppm La add 1000 ppm La :
?J
Flame setting Lean or Slightly fuel-rich Fuel-rich Fuel-rich bright
:0
stoichiometric (short red feather) luminous red feather III
s::r+
Burner height 5 mm lOmm 10mm 12 mm (light passes iD
..,
(below opt. axis) above red feather) III
:l
Q.
Analytical range 0.02-5.0 0.2-20 0.1-20 0.03-15
(in solution), ppm s:
:
Detection limit, ppm 0.005 0.04 0.03 0.01
0
~
'"
0
r+
8. Atomic Absorption 227

Belcher and Bray42 found interference from aluminum, but this was sup-
pressed by the addition of 1500-ppm strontium. Calcium and iron cause a
sharp enhancement in the air-acetylene flame up to a concentration of
100 ppm, after which the absorbance remains constant.
Unless releasing agents are used changes in the concentration of any
of the matrix elements will cause serious errors in the analysis, if aqueous
standards are used. Even if synthetic or natural rock standards are used,
the composition would have to match that of the samples very closely if
accurate results are to be obtained. It is for this reason that releasing agents
must be used for all alkaline earth determinations in rocks and minerals.
Dissolution methods for magnesium are the same as those for calcium.
Table VI gives the analytical conditions for the determination of magnesium.
The accuracy obtained with the atomic absorption method appears to
be good. While the precision may not be quite as good as x-ray fluorescence
and colorimetric methods, the precision is similar to that obtained by
emission spark methods. 43 ,44 The fact that the same precision is maintained
to a very low concentration of magnesium in rocks also makes the method
attractive to geochemists.

5.1.6. Sodium and Potassium

Sodium and potassium are considered to be major constituents in most
geologic matter. They occur relatively abundantly in large numbers of
silicate minerals typically associated with igneous rocks, as well as in a
large variety of dissolved or soluble salts in sea water and deposits. In
ultramafic rocks the concentrations of the alkali metals may be very low
so that a sensitive method is required for their determination.
The atomic absorption method has been used with considerable success
for the determination of sodium and potassium for all types of rocks and
minerals. Accurate results have been reported by Billings,38 Trent and
Slavin,37 and Belt. 40 Various methods of sample preparation were used,
and the results were similar regardless of whether fusion or acid-dissolution
techniques were used.
Earlier publications 29 listed a relatively cool flame, a town gas-air
flame, as the most suitable for the determination of the alkali metals, and
although this flame gives high sensitivity, it also gives rise to considerable
depressh'e interference. Figure 5a shows the interference of various major
elements on sodium, using propane-butane-air and in Fig. 5b the acetylene-
air flames. Although sensitivity is reduced with the hotter flame depressive
interference is less. Enhancement is increased, however.
228 L. R. P. Butler and M. L. Kokot

CONC K Mg. Co. ppm - -

500 1000
45
25 ppm No 5890 Ii

A
UJ 25 ppm No
U
045
Z + 500 M~ 0 31
<t + 1000 MQ026
'"~ 40
+2000 F, 029
+4000Fe 026
(J)
+ 1754 AI
'"<t
13
+ 3509 AI 13

38

2000 4000
36
CONC. Fe, AI, ppm

o
UJ
2'5 ppm No 5890 A
u
Z'09
(0) <t
~
K
_________ 0 ________________________________ 0

K, Mg, Co g r-o=,;-:=--:-=-::--::-::--=-=-=-=--====:-
~ 07
a
,-=====---=:::::====--'::
o 1000
Mg
Co

r----------~C~O~NC~K~,~M-g,~C-o,-p-p-m===~--------~

UJ 09
u
Z
(b) <t
Fe, AI '"0:0
(J)
07
'"<t
1000 2000 3000 4000
05~----------L-----~----L---------------------~
CONC. Fe, AI, ppm - - -

Fig. 5. Interference on sodium by other major elements in (a) propane-butane-

air, and (b) acetylene-air flames.

A useful means for overcoming the enhancement interference is to add

an excess of a low-ionization-potential element such as strontium or caesium
to both standards and samples, similar to the technique described by Butler
and Brink. 45
When the concentration of sodium and potassium is high the potassium
4044-A and sodium 3303-A lines may be used. 40 Although these lines are
less sensitive absorption is affected far less by the interfering elements.
This is due to the relatively higher element concentrations. 46 It should be
noted that when this is done the hollow-cathode lamps must be run at
higher currents to provide sufficient intensity of these higher energy lines.
8. Atomic Absorption 229

Table VII
Conditions for the Determination of Sodium and Potassium

Maximum sensitivity

Sodium Potassium

Wavelength 5890 A 7665 A

Analytical range, ppm 0.01-2.0 0.03-5.0
Burner Broad flame a Broad flame
Flame Prop.-but.-air Prop.-but.-air
Band pass 2-4 A 4-6 A

Maximum freedom from interference

Sodium Potassium

Wavelength 5890 A 3303 A 7665 A 4044 A

Analytical range, ppm 0.1-10.0 10-500 0.1-10.0 20-1000
Burner 10 cm x O. 5 mm slot 10 cm x 0.5 mm slot
Flame C.H.-air stoichiometric C.H.-air stoichiometric
Band pass 4-6 A 4-6 A

a From Butler, L. R. P., Atomic Absorption Newsletter 5, 99 (1966).

For the same reason sensitivity with the acetylene-air flame is greater for
these lines than with the propane-butane-air flame.
Table VII gives the most suitable conditions for the determination of
sodium and potassium.

5.2. Minor and Trace Elements

Of the many elements that may be determined by atomic absorption
only those elements considered to be of interest for geochemical prospecting
will be discussed. Standard textbooks on atomic absorption 8 9 give details
for the analysis of other elements.

5.2.1. Manganese
Manganese was determined by atomic absorption spectrometry by
Allan 47 and David. 48 The sensitivity is good and few interference effects are
encountered. It has been reported that silicon interferes with manganese. 6
However, when analyses of international rock standards were conducted by
230 L. R. P. Butler and M. L. Kokot

the authors using both fusion and acid-dissolution techniques, very little
difference in manganese absorbance was noted. Lanthanum, which is gen-
erally used as a releasing agent, may eliminate this interference. The acid-
dissolution technique has proved very successful for the determination of
manganese (see Table VIII).

5.2.2. Copper and Zinc

The limit of detection for the determination of copper and zinc is low,
and both elements are known to be free from interference effects:,,1l,49 The
atomic absorption method has been used for the determination of these
elements in a wide variety of materials, including geologic samples. 39 ,5o
Burrell51 reported a slight enhancement on 5 ppm of zinc by iron above
the IOO-ppm concentration. He also reports severe depression by magnesium
nitrate. Other authors 45 ,52,53,54 reported no interference from a "number of
elements on copper and zinc in an acetylene-air flame. Belt55 reports no
interference from other major .elements in rocks.
Slight interference is noted when using the propane-butane flame.
Although the sensitivity is poorer, the acetylene-air flame eliminates all
interference and the results are correspondingly more reliable.

Table VIII
Conditions for the Determination of Manganese

Sensivity
Wavelength, A
factor

Analytical lines 2795

4041 0.1

Lamp Hollow cathode 10 mA

Band pass 2--4 A

Flame C.H.-air prop.-but.-air

(for maximum (for maximum
freedom for sensitivity)
interference )

Burner 10 cm x O. 5 mm slot

Burner height 15 mm 25mm

Analytical range, ppm 0.1-10 0.05-5
Detection limit, ppm 0.05 0.02
8. Atomic Absorption 231

The highest sensitivity is required for the determination of copper and

zinc in common rocks, as their concentrations are low. For this reason acid
digestion is recommended. In cases where all the silicate minerals are to be
dissolved hydrofluoric-perchloric acid should be used. SIavin50 describes a
method whereby use is made of a hydrochloric-nitric acid mixture, followed
by sulfuric acid. Farrar 49 describes the use of sodium hyposulfate followed
by fusion. High-temperature fusion may, however, cause zinc to be lost
through evaporation.
Certain minerals and elemental copper are dissolved more easily by
dilute nitric acid. This characteristic is used by prospecting geochemists to
obtain an indication of the mineral type. 56 The conditions for the determina-
tion of copper and zinc in rocks are given in Table IX. Other elements that
require similar conditions to copper and zinc, are silver, gold, cadmium,
lead, antimony, and bismuth.

5.2.3. Chromium
A wide variety of materials has been analyzed for chromium by atomic
absorption spectrometry.6,57
Some interferences are experienced in an air-acetylene flame and iron
and nickel suppress absorption in a fuel-rich air-acetylene flame. 8 When
a hot oxidizing flame is used, however, the interference effects are reduced.
The nitrous oxide-air flame is better suited for the determination of chro-
mium in the various rock types, as interferences are reduced or even elimi-
nated. The addition of 2% ammonium chloride to samples and standards is
reported to have controlled interference effects of iron in chromium. 58 The
addition of a low ionization element such as potassium or caesium will
prevent ionization enhancement effects. Table X gives the conditions for
the determination of chromium.

5.2.4. Lithium and Its Isotopes

Lithium has been the subject of some interest in the field of geo-
chemistry. It may be determined comparatively easily by atomic absorption
spectrometry using the strongly absorbent 6708-A line. This technique has
been employed by a number of researchers. 59- 62
The best flame for maximum sensitivity has been found to be a highly
oxidizing fuel-lean acetylene-air flame with the light beam passing through
as much of the primary combustion zone as possible (see Fig. 6).
Lithium has very few interference effects although sodium and potas-
sium cause slight enhancement, and magnesium, calcium, and aluminum
 N
W
N
Table IX
Conditions for the Determination of Copper and Zinc in Silicate Rocks

Copper Zinc

Wavelength, A. Sensitivity factor Wavelength, A. Sensitivity factor

Analytical lines 3247 1.0 213S 1.0

3274 0.5 3076 0.001
2226 0.05 (approx.)
Lamp Hollow cathode 5-10 mA Hollow cathode 5-10 mA
Band pass 2-4 A. sA.
Flame Max. sensitivity Max. freedom from Max. sensitivity Max. freedom from
interference interference
!""
::a
prop.-but.-air C.H.-air prop.-but.-air C.H.-air
Burner Broad flame a 10cm x 0.5 mm slot Broad flame a 10cm x 0.5 mm slot :u
m
Flame setting stoichiometric stoichiometric stoichiometric slightly fuel rich c
~
(slightly fuel rich) ...CD
Burner height Smm 4mm Smm 10mm III
:l
Analytical range, ppm 0.02-4.0 0.5-10.0 0.005-2.0 0.05-4.0 Q.
Detection limit, ppm 0.005 0.1 0.001 0.01 3:
!""
a From Butler, L. R. P., and Mathews, P. M., Anal. Chim. Acta 36, 319 (1966).
~
'oo"
...
8. Atomic Absorption 233

Table X
Conditions for the Determination of Chromium in Silicate

Approx. relative
Wavelength, A
sensitivity

Analytical lines 3594 1.0

3579 0.73
3605 0.62
4254 0.40
4275 0.29
4290 0.22
Lamp Hollow cathode 10-20 mA
Band pass 2-4 A
Flame C.H.-air C.H.-N.O
(slightly fuel (for freedom from
rich) interference )
Burner IOcm x 0.5 mm 5 cm x O.4mm
slot slot
Burner height Light must pass through luminous zone above
primary reaction region
Analytical range, ppm 0.2-15 0.5-20.0
Detection limit, ppm 0.06 0.08

have a slight depressing effect. Iron alone has been found to cause depres-
sion. 63 However, a small amount of sodium present in the standards and
samples virtually eliminates all interferences.
Fusion techniques are unsuitable for lithium determinations, as most
alkali-fusion salts cannot be obtained in bulk in pure enough form. The
most satisfactory method for dissolution when lithium is to be determined
is a HF-HCI0 4 digestion. Conditions for the determination of lithium are
given in Table XI.
The isotopes of lithium were first determined by atomic absorption
spectrometry by Zaidel and Korennoi. 64 Manning and Slavin 65 used a flame
as source. When solutions of Li-6 and Li-7 were sprayed into this flame,
absorbance could be measured in a normal absorption flame. A demount-
able Schiiller-Golnow hollow-cathode lamp was used by others 66 as a means
for producing an atomic cloud. Instrumental methods for determining the
isotopic ratio of Li-6 to Li-7 have also been developed. 63 97
234 L. R. P. Butler and M. L. Kokot

0'13~---------------'

LITHIUM CONCENTRATION
012
10 ppm

~ 009
z
<
CD
a:
o
::l 008
<

0'07

006

-6 o 6 12 18 24
BURNER HEIGHT BELDW OPTICAL AXIS (m m.)

Fig. 6. The importance of flame-height setting and gas

composition for lithium absorbance.

5.3. Microtrace Elements after Chemical Enrichment

Special methods must be applied for microtrace-element analysis, i.e.,
elements with concentrations below the normal detection limit when brought
into solution. If interference effects are severe it may also be necessary to
apply separation techniques to isolate, and concentrate, the analytical
elements.
The most popular chemical method of extraction as applied to atomic
absorption spectrometry is the extraction of metals from an aqueous solu-
tion into an organic solvent. Slavin8 mentions four advantages when solvent
extraction methods are used for atomic absorption.
(a) There is a 300-500% enhancement in sensitivity.
(b) The metal can be concentrated in the organic phase.
8. Atomic Absorption 235

Table XI
Conditions for the Determination of Lithium

Analytical line 6708 A

Lamp Hollow cathode, 5 rnA (currents above 10 rnA result
in self-absorption and decrease in sensitivity)
Band pass 2-4 A
Flame C2H z-air (fuel lean)Q
Burner 10 cm x 0.5 mm slot
Analytical range 0.01-4.0 ppm
Detection limit 0.005 ppm

a The best sensitivity is obtained when the burner is adjusted without the flame burning to
cause 10% of the light to be cut off.

(c) The salt content of a solution used for extraction can be far greater
than that which can be handled in any burner.
(d) Interfering elements can often be left behind in the aqueous phase.
The most suitable organic solvents for use with atomic absorption
have been found to be esters and ketones. 53 The use of ammonium pyrroli-
dine dithio-carbamate (APDC) as a complexing agent extracted into methyl-
isobutyl ketone (MIBK) as a means for the concentration of many trace
elements is well known, and is described in the literature for geologic
applications.u,37,67 Zinc, chromium, nickel, lead, cobalt, cadmium and many
other metals can be extracted by this method. Wilson 52 determined copper,
zinc, lead, and molybdenum at levels below 100 ppm in rock samples by
precipitating the metals from acid digests with thioacetamide. Butler et al. 68
developed a completely automatic instrument to assay gold in cyanide
mining solutions. Extraction of gold was effected by MIBK. A useful tech-
nique using normal amyl-methyl ketone for determining molybdenum in
rocks was also developed. 69 Table XII shows extraction methods for various
elements.

5.4. Water Analysis

Atomic absorption spectrometry has been applied with considerable
success to water analyses. The main advantage of this method is that the
sample can be sprayed into the flame directly. In the event that the metals
sought are present below the instrumental detection limit they can easily be
concentrated by partial evaporation or by organic solvent or ion-exchange
extraction.
236 L. R. P. Butler and M. L. Kokot

Table XII
Extraction Reagents for Different Elements

Element Solvent References

Antimony APDC-MIBK 82, 83

Arsenic APDC-MIBK 83
Bismuth APDC-MIBK 82, 83
Cadmium APDC-MIBK 84
Chromium Diphenylthiocarbazone-MIBK APDC-MIBK 85, 67
Cobalt APDC-MIBK 84, 86
Copper APDC-ethyl amyl ketone (or MIBK) 53
Gallium APDC-MIBK 83
Gold MIBK 68
Indium APDC-MIBK 83, 87
Iron APDC-MIBK 83, 88, 89
Lead APDC-MIBK 82, 88, 90
Dithizone-MIBK 91
Magnesium 8-Hydroxyquinoline-MIBK 92
Manganese APDS-MIBK 67
Oxine in chloroform-methanol 67
8-Hydroxyquinoline-chloroform 93
Molybdenum Toluene-3,4-dithiol-MIBK 69
APDC-n-methlamyl ketone 85
Nickel APDC-MIBK 84
Palladium Pyridine thiocyanate-hexone 94
Selenium APDC-MIBK 83
Silver Di-n-butylammonium salicylate-MIBK 42
Dithiozonate complex-ethyl propionate 95
Vanadium Cupferron-MIBK 96
Zinc APDC-MIBK 53

Butler and Brink4s describe the application of atomic absorption

methods to the analysis of calcium, magnesium, iron, sodium, potassium,
and copper in river, borehole, and well waters. All the samples were filtered
to remove any suspended material. Formalin was added to prevent bac-
terial-induced changes in the pH from causing the precipitation of salts.
In some cases the concentrations of sodium and magnesium were too high
to be determined directly, and suitable dilutions had to be made. Some
interferences were noted, particularly for calcium and magnesium, but
these were suppressed by the addition of 1500-ppm strontium chloride.
A method for determining iron, chromium, nickel, cobalt, manganese,
and copper in water at the parts-per-thousand-million level has been de-
8. Atomic Absorption 237

scribed by Wheat.70 The samples were filtered and concentrated tenfold by

evaporation. BurreFl determined nickel and cobalt by preconcentration and
solvent extraction, and Platte and Marcy6 described the atomic absorption
method as a general tool for the water chemist. Price and Ragland 61 deter-
mined alkali metals and earths, and a few transition elements in ground
water. The samples were filtered and treated with EDTA. Fishman and
Downs 60 have analyzed natural waters by filtering the sample through
O.4S-.u micropore-membrane filters. The elements manganese, zinc, copper,
lithium, sodium, and potassium were determined by direct aspiration into
the flame.
Little, if any, prior treatment is required for the study of most trace
metals in the concentration range of 0.01-10 ppm in fresh water. Filtration
is advisable, however, to ensure that only the dissolved portion is being
examined.

6. GEOCHEMICAL PROSPECTING

One of the most successful applications of atomic absorption spectrom-

etry has been in the field of geochemical prospecting. The underlying
principles for the discovery of mineral deposits have been described in
several books and articles. 72- 74 Certain elements, such as copper, zinc, and
lead, are considered to be good tracer elements. Mercury has also received
considerable attention. 74 It is therefore not surprising that atomic absorp-
tion is widely practiced by many mining companies and geology laboratories
throughout the world. The reasons for this are the relatively low cost of
the equipment, the good precision and sensitivity obtainable, the versatility,
and the low cost of sample analyses. The limitation of determining only one
element at a time is offset by the advantages mentioned. When more than
one element is being determined in a large number of samples, it is not
unusual to find several atomic absorption spectrometers in use, each of
which is set for determining one particular element only. This system is
often less expensive than a large direct-reading spectrometer with its asso-
ciated accessories to determine the elements simultaneously.
As the accent is on rapidity of analysis with reasonable precision, soil
samples are not ground unless a large proportion of the sample is gravel.
The samples may be taken a few inches below the surface, according to a
grid pattern, or along chosen traverses or river beds. It is more usual to
collect the portion passing through a 80-1 SO-mesh sieve. After mixing, a por-
tion of this fraction is then weighed and transferred to the dissolution vessel.
238 L. R. P. Butler and M. L. Kokot

Dissolution may be carried out stepwise, i.e., a number of samples

placed together in a basket or tray and then heated on a steam or sand
bath, or they may be treated individually on a conveyor belt or continuous
feeding mechanism.
The most important requirement is that samples should be treated in
an identical way, as the absolute value of the element determined is oflesser
importance than deviations from a norm. An unusually high value is termed
an anomaly, and it is these anomalies which are followed. Figure 7 shows
how an underground ore body shows up as an anomaly.
With uniform soil types it may be possible to forego the time-consuming
weighing process with the use of a scoop of a constant volume. Surprisingly
precise weights of soil may be measured in this way.
Many laboratories have automated the dissolution process, as this is
the most time-consuming part of the analysis. Sample handling is kept to
a minimum, and it is usual for the sample to be kept in the same vessel for
the remainder of the procedure. Solutions are usually added with commer-
cial or selfmade automatic dispensing flasks. After dissolution and making

300

:>
U
E ABNORMAL
8: ZONE

BACKGROUND
ZONE
o .1000 2000 3000 4000 5000 6000

LEAKAGE HALO

DISTANCE fFEET)-

Fig. 7. Schematic diagram relating an anomaly of the concen-

tration of copper along a traverse, with an underground ore body.
Note the lower copper content above the impervious xenolith.
8. Atomic Absorption 239

CAPI LLARY - - --I

DISC (ATTAOtED - .....,.,"U"~

TO CAPILLARY)

DISSOLUTION VESSEL

UNDISSOL VED POWDER

Fig. 8. An example of a typical dissolution-centrifugation

vessel for the rapid dissolution of a prospecting sample.

up to volume, the sample vessel may be centrifuged or left to stand to enable

undissolved material to settle. The solution above the powder is then pre-
sented to the nebulizer. Precautions are usually taken to prevent the capillary
from inadvertently being dipped into the settled powder and becoming
blocked. Figure 8 shows a typical dissolution flask and capillary with guard.
Other types of samples that may also be analyzed by atomic absorption
include drill or water samples. Drill samples are usually taken where definite
indications of mineralization have been found to obtain a three-dimensional
picture of the deposit. In these cases, a higher degree of precision and
accuracy is necessary than with grid samples, and greater care must be
taken to ensure that these requirements are met.
Samples are dried, ground, and mixed, and several carefully weighed
aliquots are taken to improve precision. It is also necessary that the dissolu-
tion procedure used is adequate for dissolving the minerals quantitatively
(see Chap. 3, Sect. 3).
Atomic absorption is also well suited to the analysis of water samples.
The presence of elements in subsurface water in dry regions may be an indi-
cation that the water has percolated or passed over mineralized strata.
Many factors will influence the degree to which minerals will dissolve in
such water, e.g., pH of the water, contact time or flow rate, rainfall pattern,
temperature, etc. The analytical methods used must be very sensitive as
concentrations may be relatively low. Simple concentration by evaporation
or solvent extraction enables very low concentrations to be determined
accurately.
Plants growing in mineralized soils may also have tell-tale high con-
centrations of indicator elements present in their leaves. By analyzing plant
240 L. R. P. Butler and M. L. Kokot

material it may thus be possible to detect anomalies. While this method of

geochemical prospecting is relatively new, it may be very useful when used
in conjunction with other methods.

7. RECENT AND FUTURE DEVELOPMENTS

The advances made in the technique and instrumentation of atomic

absorption have been remarkable, and the question arises of whether this
progress can be maintained. A trend which will prove useful to the geo-
chemist will be the development of a satisfactory multielement analysis
system, aimed at certain applications such as geochemical prospecting and
routine water analysis. Multielement systems have been suggested. 75 ,76 Al-
though "custom-built" instruments for specific applications may be devel-
oped, general multielement atomic absorption instruments appear to have
inherent difficulties due to the problem of obtaining efficient multielement
sources as well as difficulties in determining several elements with a single
flame setting.
A selective modulator 77 with a solar-blind photomultiplier and inter-
ference-filter system for each element around a common flame was built
by Walsh's group for geochemical prospecting. Selective modulation makes
it possible to develop inexpensive instruments for each element and so do
away with the most expensive item of instrumentation-the spectrometer.
This modification may be useful for the geologist or geochemist with the
possible extension to field instruments.
The resonant-detector atomic absorption spectrometer, suggested by
Sullivan and Walsh,25,78 has found some useful applications. The resonance
fluorescence from a hollow-cathode or thermal-sputtering tube is measured
using a photomultiplier and electronic set. As no wavelength adjustments
are required, the spectrometer is insensitive to mechanical or thermal shocks
and changes. These features are ideal for rough usage, and an interesting
instrument for automatically measuring the gold content on gold-mine
cyanide solutions using a resonance spectrometer was developed. 98 There
are also several other applications of these spectrometers for geochemical
purposes.
Considerable interest has been shown recently in the development of
new atomization sources for converting solid, undissolved material into an
atomic cloud to enable atomic absorption measurements to be made. Ven-
ghiattis79 has suggested the use of solid-rocket-propellent powder mixed
with an ore or concentrate sample to provide an atomic vapor of the sample.
8. Atomic Absorption 241

This method may be very useful for geochemical prospecting. The recent
L'vov furnace has been followed by other nonflame atomizing devices. 8o ,8!
One of the drawbacks of nonflame atomizers is to efficiently vaporize the
elements without fractional distillation and separation occurring.

REFERENCES

1. Walsh, A., Spectrochim. Acta 7, 108 (1955).

2. Langmyhr, F. J., and Paus, P. E., Anal. Chim. Acta 43, 397 (1968).
3. Medlin, J. H., Suhr, N. H., and Bodkin, J. B., Atomic Absorption Newsletter 8, 25
(1969).
4. Yule, J. W., and Swanson, G. A., Atomic Absorption Newsletter 8, 30 (1969).
5. Strasheim, A., Strelow, F. W. E., and Butler, L. R. P., J. S. African Chem. Inst. 13,
73 (1960).
6. Platte, J. A., and Marcy, V. M., Atomic Absorption Newsletter 4, 289 (1965).
7. Billings, G. K., and Angino, E. E., Bull. Can. Petrol. Geol. 13, 529 (1965).
8. Slavin, W., Atomic Absorption Spectroscopy, Interscience, New York, 1968.
9. Ramirez-MufIOZ, J., Atomic Absorption Spectroscopy and Analysis by Atomic Absorp-
tion Flame Photometry, Elsevier, New York, 1968.
10. Atomic Absorption Spectroscopy, A.S.T.M. STP 443.
11. Angino, E. E., and Billings, G. K., Atomic Absorption Spectrometry in Geology,
Elsevier, Amsterdam, 1967.
12. Amos, M. P., and Willis, J. B., Spectrochim. Acta 22, 1325 (1966).
13. Mitchell, A. C. G., and Zemansky, M. W., Resonance Radiation and Excited Atoms,
Cambridge University Press, 1934, new Ed., London, 1961.
14. Alkemade, C. Th. J., Appl. Optics 7, 1261 (1968).
15. Dagnall, R. M., and West, T. S. Appl. Optics 7, 1287 (1968).
16. Winefordner, J. D., and Vickers, T. J., Anal. Chem. 36, 161 (1964).
17. Russel, B. J., Shelton, J. P., and Walsh, A., Spectrochim. Acta 8, 317 (1957).
18. Fassel, V. A., Mossotti, V. G., Grossman, W. E. L., and Kniseiey, R. N., Spectrochim.
Acta 22, 347 (1966).
19. Rann, C. S., Spectrochim. Acta 23B, 245 (1968).
20. Human, H. G. c., Butler, L. R. P., and Strasheim, A., Analyst 94, 81 (1968).
21. Mavrodineanu, R., and Boiteux, H., Flame Spectroscopy, John Wiley & Sons, New
York, 1965.
22. Butler, L. R. P. and Kokot, M. L., Geol. Soc. S. A. Special Publication No.2, 449
(1969).
23. Butler, L. R. P., and Fulton, H. A., Appl. Optics 7, 2131 (1968).
24. Robinson, J. W., Anal. Chim. Acta, 23, 479 (1960).
25. Sullivan, J. V., and Walsh, A., Spectrochim. Acta 21, 721 (1965).
26. Kahn, H. L. and Slavin, W., Appl. Optics, 2, 931 (1963).
27. Slavin, W., and Trent, D. J., Atomic Absorption Newsletter 4, 351 (1965).
28. Gaumer, M. W., Sprague, S., and Slavin, W., Atomic Absorption Newsletter 5, 58
(1966).
29. Gatehouse, B. M., and Willis, J. B., Spectrochim. Acta 17, 710 (1961).
30. Dagnall, R. M., Thompson, K. c., and West, T. S., Talanta, 14, 1511 (1967).
242 L. R. P. Butler and M. L. Kokot

31. Capacho-Delgade, L., and Manning, D. C., Analyst. 92, 553 (1967).
32. Laflamme, Y., Atomic Absorption Newsletter 6, 70 (1967).
33. Van Loon, J. c., Atomic Absorption Newsletter 7, 3 (1968).
34. Price, V., and Ragland, P. c., Southeastern Geol. 7, 93 (1966).
35. Galle, O. K, Appl. Spectros. 22, 404 (1968).
36. Harris, R. c., and Almy, C. c., Bull. Marine Sci. Gulf Caribbean 14, 418 (1964).
37. Trent, D. J., and Slavin, W., Atomic Absorption Newsletter 3, 17, 118 (1964).
38. Billings, G. K, Atomic Absorption Newsletter 4, 312 (1965).
39. Bowditch, D. c., Chalmers, A. H., and Powell, J. A., Tech. Memo. Sf3, A.M.D.E.L.,
Adelaide, Australia, 1966.
40. Belt, C. B., Jr., Anal. Chem. 39, 676 (1967).
41. Allan, J. E., Analyst 83, 466 (1958).
42. Belcher, C. B., and Bray, H. M., Anal. Chim. Acta 26, 322 (1963).
43. Suhr, N. H., and Ingamells, C. 0., Anal. Chem. 38, 730 (1966).
44. Govindaraju, K, Publication G.A.M.S. 3, 217 (1963).
45. Butler, L. R. P., and Brink, D., S. African Ind. Chemist 17, 152 (1963).
46. Hollander, T., Ph.D. thesis, University of Utrecht, Holland, 1964.
47. Allan, J. E., Spectrochim. Acta 10, 800 (1959).
48. David, D. J., Atomic Absorption Newsletter No.9 (1962).
49. Farrar, B., Atomic Absorption Newsletter 4, 325 (1965).
50. Slavin, W., Atomic Absorption Newsletter 4, 243 (1965).
51. Burrel, D. C., Norsk. Geol. Tidssler 45, 21 (1965).
52. Wilson, L., Austr. Dept. Supply, Report ARLfMET 17, 1963.
53. Allan, J. E., Spectrochim. Acta 17, 467 (1961).
54. David, D. J., Analyst 83, 655 (1958).
55. Belt, C. B., Jr., Econ. Geol. 59, 240 (1964).
56. Sampey, D., personal communication, 1968.
57. Kinson, K, Hodges, R. J., and Belcher, C. B., Anal. Chim. Acta 29, 134 (1963).
58. Giammarise, A., Atomic Absorption Newsletter 5, 113 (1966).
59. Angino, E. E., and Billings, G. K, Geochim. Cosmochim. Acta 30, 153 (1966).
60. Fishman, M. J., and Downs, S. C., U.S. Geol. Surv. Water Supply Paper, 154O-C,
U.S. Yort, Printing Office, Washington, 1966.
61. Price, V., and Ragland, P. C., Water Resources Paper No. 18, University N. Carolina,
1966, p. 8.
62. Slavin, W., Atomic Absorption Newsletter 7, 11 (1967).
63. Butler, L. R. P., Atomic Absorption Spectrometry and Its Application in Geochemistry,
Ph.D. thesis, University of Cape town, 1968.
64. Zaidel, A. N., and Korennoi, E. P., Opt. and Spectroscopy 10, 299 (1961).
65. Manning, D. c., and Slavin, W., Atomic Absorption Newsletter 1, 39 (1962).
66. Goleb, J. A., and Yoko Yuma, Y., Anal. Chim. Acta 30, 213 (1964).
67. Mansell, R. E., and Emmel, H. W., Atomic Absorption Newsletter 4, 365 (1965).
68. Butler, L. R. P., Strasheim, A., and Strelow, F. W. E., XII ColI. Spectr. Int., Exeter,
England, 1965.
69. Butler, L. R. P., and Mathews, P. M., Anal. Chim. Acta 36, 319 (1966).
70. Wheat, J. A., Report DP-879, U.S. Dept. Comm. Office Tech. Serv., also AEC Res.
and Dev. Report TID-4500, 1964.
71. Burrel, D. C., Atomic Absorption Newsletter 4, 328 (1965).
8. Atomic Absorption 243

72. Hawkes, H. E., a:ld Webb, J. S., Geochemistry in Mineral Exploration, Harper and
Row, New York, 1962.
73. Kvalheim, A., l.D.S.A. 37, 585 (1967).
74. Cameron, E. M., Ed. Proc. Symp. on Geochem. Prospecting, Ottawa, 1966.
75. Butler, L. R. P., and Strasheim, A., Spectrochim. Acta 21, 1207 (1965).
76. Dawson, J. B., and Ellis, D. J., XII Coli. Spectr. Int., Exeter, England, 1965.
77. Bowman, J. A., Sullivan, J. U., and Walsh, A., Spectrochim. Acta 22, 205 (1966).
78. Sullivan, J. V., and Walsh, A., Spectrochim. Acta 22, 1843 (1966).
79. Venghiattis, A. A., Atomic Absorption Newsletter 6, 19 (1967).
80. Massmann, H., Spectrochim. Acta 23B, 215.
81. West, T. S., S. African Minerals Science and Eng. 1 (1969).
82. Willis, J. B., Anal. Chem. 34, 614 (1962).
83. Mulford, E. C., Atomic Absorption Newsletter 6, 28 (1966).
84. Sprague, S., and Slavin, W., Atomic Absorption Newsletter 3, 37 (1964).
85. Delaughter, B., Atomic Absorption Newsletter 6, 66 (1965).
86. Burrel, D. C., Atomic Absorption Newsletter 4, 309 (1965).
87. Lakanen, E., Ann. Agr. Finniae 1, 109 (1962).
88. Trent, D. J., and Slavin, W., Atomic Absorption Newsletter 3, 118 (1964).
89. Menis, 0., and Rains, T. C., Anal. Chem. 32, 1837 (1960).
90. Berman, F., Atomic Absorption Newsletter 3, 111 (1964).
91. Elwell, W. T., and Gidley, J. A. F., Atomic Absorption Spectrophotometry, Pergamon
Press, London, 1966.
92. Suzuki, M., Yanagisawa, M., and Takeuchi, T., Talanta 12, 989 (1965).
93. Calkins, R. c., Appl. Spectros. 20, 146 (1966).
94. Erinc, G., and Magee, R. J., Anal. Chem. Acta 31, 197 (1964).
95. West, P. W., and Ramakrishna, T. V., Environ Sci. Technol. 1, 717 (1967).
96. Sachdev, S. L., Robinson, J. W., and West, P. W., Anal. Chim. Acta 37, 12 (1967).
97. Butler, L. R. P. and Schroeder, W. W., Int. Atomic Absorption Spectroscopy Con!,
Sheffield, England, 1969.
98. Rossouw, J., personal communication, 1968.
Chapter 9

X-RAY TECHNIQUES

H. A. Liebhafsky*
Texas A & M University
College Station, Texas

and
H. G. Pfeiffer*
General Electric Company
Schenectady, New York

1. INTRODUCTION

A chapter of this length can be at most a guide to x-ray methods in geo-

chemical analysis. To see why this is true, consider, for example, that the
1968 review in Analytical Chemistry of x-ray absorption and emission by
Campbell and Brown! lists 17 new books and 695 references, and that its
counterpart by Merritt and Streib 2 on x-ray-diffraction records over a third
as many items of each kind. These reviews, which appear every two years,
and several books 3 - 6 are the main sources of information for this chapter.
They are only a fraction of what should be consulted by anyone seriously
interested in using x-ray methods in geochemistry.
This chapter aims at giving enough general information about x rays
to convince the reader (if necessary) of their comparative simplicity and

* The authors are grateful for permission to use certain passages, tables, and figures
from Liebhafsky, H. A., Pfeiffer, H. G., Winslow, E. H., and Zemany, P. D., X-Ray
Absorption and Emission in Analytical Chemistry, John Wiley & Sons, New York, 1960.

245
246 H. A. Liebhafsky and H. G. Pfeiffer

outstanding usefulness in geochemistry, and to reinforce this general in-

formation by citing representative specific applications. The x-ray processes
involved are absorption, emission, and diffraction. These processes will
usually lead to determinations of elements in a sample or, in the case of
diffraction, to a determination of its crystal structure. Chemical analysis,
defined as the characterization and control of materials, 7 has a much
broader assignment that x-ray methods alone usually cannot accomplish.
The explosive increase in the use of x-ray methods in analytical chem-
istry since World War II did not result from new knowledge about the rays
themselves. It occurred primarily because methods of x-ray detection were
improved until it was a matter of simple routine to measure x-ray intensity
with high precision. Rapid progress in experimental nuclear physics accele-
rated this improvement, which soon led to advances, still continuing, in
x-ray equipment other than detectors. Now, the computer is beginning to
make itself felt, and the end is not in sight.
X rays are the most powerful radiant energy obtainable from an atom
without entering its nucleus. Their wavelengths range roughly from 0.1
("hard" x rays) to 100 A ("ultrasoft" x rays), which means that they lie
between y rays and the UV rays of shortest wavelength. The important
advantages ofx rays for chemical analysis are the simplicity of their spectra;
almost complete freedom (for most elements) from chemical influences;
predictability of behavior, which often includes predictability of deviations
from simple behavior; the opportunity of designing the desired precision
into determinations by x-ray methods; and the nondestructive nature of
these methods. Most of these weighty advantages are traceable ultimately
to the high energy of x rays.

2. THEORETICAL CONSIDERATIONS

2.1. Nature and Excitation of X Rays

As x rays originate outside the atomic nucleus, they must be excited;

that is, energy must be supplied so that atoms (or ions) can emit them.
Electrons and absorbed x rays, both of sufficient energy, are the two most
important means of excitation. The Coolidge tube is the most common
source ofx rays for analytical work. It is sealed and operates at high vacuum.
In the Coolidge tube electrons leave a coiled tungsten filament (cathode)
at high temperature (upper center of Fig. 1) to strike a target (anode),
from which the x rays leave the tube through a window, usually of beryllium,
9. X-Ray Techniques 247

High voltage
Filament - - - - - -......--,
current _ _ _ _ _ _- - ,

Cathode
focusing
cup

Target
metal

X-ray
wmdow

Evacuated
envelope

I Olut
In Cooling water

Fig. 1. Schematic diagram of the Coolidge (high-

vacuum) x-ray tube. Coolidge tubes are widely used
because they are stable and long-lived and because
they permit tube current and voltage to be controlled
independently.

chosen because this light metal does not absorb the rays unduly [Eq. (6)].
Tubes with interchangeable targets are available. Tungsten, molybdenum,
chromium, copper, silver, nickel, iron, and cobalt are common target ma-
terials. Tungsten (for "hard" x rays) and chromium (for "soft" x rays)
make a good pair of targets for a single tube.
Figure 2 shows the x-ray spectra observed to pass the window of a
tungsten-target tube when the voltages accelerating the electrons are 20 and
50 kV. These are continuous emitted x-ray spectra and they are polychromatic
beams. They have a short-wavelength limit Ao , at which they intersect the
abscissa. This shows that the excitation of x rays is a quantum phenomenon.
The relation of Ao to the maximum potential difference V (in kV) across the
tube is given by
AO = 12.393JV (1 )
248 H. A. Liebhafsky and H. G. Pfeiffer

Tungsten spectrum
1.0

0.8 -

04

0.2
B 20 kv
0.0 ~---:1:L--:--:-~~q;::t:::::",-~::::l::;j;;;;;;;t;;;;;i;:;;::i;;;;:!;;;;;~
0.0 02 04 0.6 0.8 I0 I 2 1.4 1.6 I8 2.0
Wavelength (A)

Fig. 2. The continuous x-ray spectrum. Note that the short-

wavelength limit [Equation (1) 1 is 0.248 A for 50 kV. (a) and
0.620 A for 20 kV. (b) Such spectra are useless for the iden-
tification of elements.

Had V been increased above 70 kV the spectra in Fig. 2 would have

undergone a startling change, which is shown in Fig. 3 for molybdenum
excited by electrons and by x rays. The spikes are characteristic lines, or
monochromatic beams, that can be used as analytical lines for the elements,
just as the far more numerous lines in the visible or the UV region have

K/3 Ka

Z Electron
excitation

X-ray eXCttatlon~

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Wavelength (A)

Fig. 3. An idealized drawing of the molybdenum spectrum

excited by 35-kV electrons and by the polychromatic beam
from a 35-kV x-ray tube. With x-ray excitation, most of the
energy appears in the characteristic lines. With electron
excitation, much of it is wasted in the continuous spectrum.
9. X-Ray Techniques 249

long been used in emission spectrography. Note that electron excitation of

the molybdenum lines in Fig. 3 could have been accomplished with a
molybdenum target in a Coolidge tube (Fig. 1), while x-ray excitation could
have resulted had molybdenum as element or as compound been exposed to
a suitable polychromatic beam generated by a Coolidge tube.

2.2. Interaction of X Rays with Matter

For our purposes the interaction of x rays with matter can be described
under the headings absorption, scattering, and diffraction. We have already
said that absorptive interaction (and electron bombardment of matter) can
result in x-ray emission.
In the main, x-ray absorption is a physical process governed by Beer's
law, which for a narrow, parallel, monochromatic x-ray beam incident per-
pendicularly upon an absorber of unit area and of uniform thickness and
composition takes the form

(2)

where II and 12 are intensities transmitted by two samples of mass m 2 and

mI of the same material, expressed in g/cm 2 of sample area for identical
x-ray beams of initial intensity 10 , The proportionality constant is the mass-
absorption coefficient fl.
The absorption of x rays principally results from photoelectric absorp-
tion and from scattering. In the simplest case of photoelectric absorption,
all the energy of the absorbed x ray goes into the kinetic energy of a photo-
electron and into the potential energy of an excited atom that can subse-
quently emit a characteristic line. The scattering of x rays is the basis of
all x-ray-diffraction phenomena. Scattering may be unmodified [no change
in wavelength; see Eq. (5)] or modified (wavelength increased; the Compton
effect). Formally
(3)

where T, the coefficient for photoelectric absorption, generally makes the

greater contribution to fl. Scattering increases in relative importance as
atomic number and wavelength decrease.
Owing to their high energies the wavelengths of x rays often are com-
parable to spacings in crystals. Consequently, a crystal can act as an x-ray
grating made up of equidistant parallel planes (Bragg planes) of atoms or
ions from which unmodified scattering of x rays may occur in such a fashion
250 H. A. Liebhafsky and H. G. Pfeiffer

Incident
rays

/I
A--------~----------~~--~--------A'

B--------------~~----~----~------B'

Fig. 4. Simplified "reflection" derivation of Bragg's law.

that the waves from different planes are in phase, and reinforce each other.
When this happens the x rays are said to undergo Bragg reflection by the
crystal, and a diffraction pattern results.
The conditions for Bragg reflection to occur are illustrated in Fig. 4,
where AA' and BB' are the traces of successive Bragg planes, d is their
distance apart, and (J is the glancing angle of the incident x-ray beam on
these planes. Bragg reflection occurs when a wave scattered at 0' can re-
inforce an identical wave scattered at 0, these being the points at which
the incident beam meets the Bragg planes.
Reinforcement will occur if the path length of a beam specularly re-
flected at 0' exceeds by an integral number (n) of wavelengths the path
length of a beam similarly reflected at O. This requires, therefore, that in
Fig. 4
(J = (J' (4)
and (Bragg's law)
nA = 2d sin (J (5)

Planes below BB' will contribute to the reflected beam to an exponentially

decreasing extent. The relationship between A and sin (J determines the useful
wavelength range of a particular crystal.
Equation (5) is obeyed so well that it is possible to use x-ray diffraction
from crystals for highly precise determinations either of d or of A. The
former type of determination is basic in establishing crystal structure,
whereas the latter is used in x-ray-emission spectrography. Both determina-
9. X-Ray Techniques 251

tions are highly precise ultimately, because d is remarkably constant for the
same Bragg planes in different crystals properly grown. When a polychro-
matic beam strikes such a crystal, only those wavelengths are detectable
above background for which Eqs. (4) and (5) are obeyed. By positioning
the detector to intercept a reflected beam that makes an angle of 2(J with
the beam incident on the crystal, it is possible to measure the combined
intensities of the wavelengths for which these equations are satisfied for a
given crystal. By varying 2(J, and, if necessary, by changing crystals, the
intensity-wavelength distribution of a polychromatic beam can be obtained,
as in the case of Fig. 2. In this way, also, a "monochromatic" beam of
desired wavelength can be selected from a polychromatic beam. If an even
purer beam is needed two crystals in series (double monochromator) may
be used. Clearly, the purer the beam, the lower will be its intensity.

2.3. X Rays and Atomic Structure

The short-wavelength limit (Fig. 2) and the appearance of character-

istic lines (Fig. 3) both show that the interaction of x rays with matter
involves quantum transitions, probably from one atomic energy level to
another. X-ray-absorption spectra also bear this out.
Figure 5 shows that one may regard the absorption spectra roughly as
parallel lines broken by the absorption edges, which are called K, L, M, . ..
edges in order of increasing wavelength. Each edge has associated with it
an emission spectrum, or a series containing relatively few lines, which are
identified by the same letter. The wavelength of each such line exceeds that
of the edge. Between edges the following approximation holds [see Eq. (3)]:

(6)

where C is a proportionality constant and Avogadro's number N divided

by the atomic weight A gives the number of atoms per g. The approxima-
tion 7: = fl becomes more nearly true as Z and A increase. In Fig. 5 note
that (a) aluminum shows no absorption edges, since these occur at wave-
lengths beyond loA, (b) that copper shows only the K edge, which occurs
at a longer wavelength than the K edge for lead and at a wavelength com-
parable to those for the three L edges of lead, and (c) that the decrease of
slope at short wavelengths, which is most noticeable for Al but undetectable
for Pb, reflects the effect of (1 on fl [see Eq. (3)].
Equation (6) shows why a polychromatic beam is strongly filtered as it
passes through a sample. Filtering implies the preferential removal of longer
252 H. A. Liebhafsky and H. G. Pfeiffer

Pb Cu+Al
/
500
(~)
II "
200
,/1, /
100
/i/
50

1/ l'
~ 20
u :' /
~ 10 ,'/ /
:' " //
,/ /
1.0
eu,'
:' /
/
0.5 " /Al
,
0.2 /' /
./"
Ol~~~~~~~~~~~~
0.05 0.1 0.2 10
Wavelength (A)

Fig. 5. Log-log plot showing mass-absorption

coefficient as a function of wavelength for
three common metals. Note that the discon-
tinuities locate the absorption edges. After
H. A. Liebhafsky, Ann. N. Y. A cad. Sci. 53,
997 (1951).

wavelengths, which occurs because T varies as A. 3 Another way to describe

this change is by saying that the effective wavelength of the polychromatic
beam has decreased.
Let us pass from absorption to emission. The K spectra-namely, those
that are associated with the K absorption edge-will be described, although
the others of longer wavelength are governed by the same fundamental
considerations. Only the K and L spectra are presently important in ana-
lytical chemistry.
The simplicity of x-ray spectra, which derives from their high energy,
is best illustrated by the fundamental significance of atomic number in
determining the frequency of the characteristic lines. Moseley's celebrated
relationship for the frequency 'J1 of the K~ line of the element with atomic
number Z is
'J1 = 0.248(1016 )(Z - 1)2 (7)
9. X-Ray Techniques 253

How is a K line generated? Every chemist knows that atomic nuclei

are surrounded by shells of electrons that, when completed, contain 2, 8, 8
18, ... electrons, this being the explanation of periodicity in chemical prop-
erties. For reasons that will become apparent later, these may be called
K, L, M, N,... electrons, respectively, on the basis of conclusive x-ray
evidence. Neither hydrogen nor helium has a K series, although each has K
electrons. The reason is that the K series is generated only when the
inner K shell contains a hole that is being filled by an electron from one of
the outer L, M, . .. shells. That is, the generation of the K series requires
(a) the absence of a K electron and (b) the presence of an outer-shell elec-
tron, whose transition to the K shell is permitted by the selection rules. This
explains why-no matter what the method of excitation-all K lines, for a
given element, have the same excitation threshold and appear together if
they appear at all.
An atom (or ion) with a single hole (from which an electron has been
removed) in the K shell is in the K state; the L, M, . .. states are similarly
defined. The generation of the K lines may be accordingly represented as
follows:
K state -+ L state + K~ and K state -+ M state + KfJ (8)

The K spectrum is thus a series of lines originating from a single initial

state, and the wavelengths of the characteristic lines in each series differ
because each line represents a transition to a different final state. Because
the energy of any M state is less than that of any L state, it follows that KfJ
will have a shorter wavelength than K~.

CIOENT x RAY
0" ELECTRON -==Tf1rtt==.;:ttI----.'
BEAM

Fig. 6. Schematic diagram showing x-ray or electron excitation

of analytical x-ray lines for any element with atomic number 29
(copper) or greater. The K spectrum results when x rays of wave-
lengths shorter than that of the K edge (Fig. 5) are absorbed to
eject a K electron, and the vacancy is filled by an electron from
an outer shell. Electrons may replace x rays to eject the electron.
L spectra are produced in a similar manner.
254 H. A. Liebhafskyand H. G. Pfeiffer

In Fig. 6 the characteristic lines of possible importance in analytical

chemistry are shown for a typical atom. The simplicity of x-ray emission is
attested to not only by the small number of lines, but even more eloquently
by the fact that such a simple drawing can represent the x-ray excitation of
analytical x-ray lines for any element with atomic number 29 (copper) or
greater. The L spectra are useful substitutes for the K spectra when, as with
very heavy atoms, the energies required to excite the K spectra are beyond
the capabilities of the equipment available.

3. INSTRUMENTATION

3.1. Analytical Systems

A photometer consists of a source, sample, and detector. The sample
acts as absorber to filter and attenuate a polychromatic beam. Let us replace
the sample with a crystal undergoing Bragg reflection. If the crystal is being
used to produce a monochromatic beam of known wavelength, the instru-
ment is a spectrometer. If there is in place of the crystal a sample whose
lattice spacings are being determined, we have a diffractometer. Let us add
a sample as fourth component to the spectrometer. If the sample undergoes
excitation by x rays (or electrons), we have an emission spectrograph, with
the crystal serving to analyze (resolve) the emitted spectrum. If the sample
primarily attenuates an x-ray beam, we have a spectrophotometer (see Fig. 7).
In the spectrophotometer the crystal may be placed between source and
sample, in which case it acts as a monochromator. Or, it may be inserted
between sample and detector, where it acts as an analyzer of the transmitted
beam.
This abbreviated summary is only an introduction, but no more is
possible. An x-ray-emission electron microprobe will later be described
more fully to give the reader a feeling for the complexities that are possible,
and sometimes indispensable.

3.2. Measurement of X-Ray Intensities

X rays are detected by observing an effect of their interaction with
matter. These effects often are translated into electric currents, pulsed or
continuous, for which complex electronic circuitry is usually needed. Among
the more important effects are (a) latent image formation on a photographic
plate (chemical effect), (b) ionization in a gas (electrical effect), and (c)
excitation of a phosphor to yield visible light (optical effect).
9. X-Ray Techniques 255

A X-ray spectrometer
(no sample in place)
or X-ray spectropho-
tometer (sample in /,) ('\
either position) ','~\'
,', 9 '\
B X-ray spectrograph 8q: (../ \,. ) ~\e
'17.01. 9~
e Crystal
Fig. 7. Schematic diagrams of instruments for x-ray spectral analysis with x-ray excita-
tion. To cover a spectral range, the crystal is moved with angular velocity ~, and the
detectors at twice this rate, or W.

An instantaneous detector measures directly the intensity I, which is the

number of x-ray quanta per sec that produce the effects mentioned above.
When I is low, an accumulative detector may be required. The x-ray intensity
is never constant, although the fluctuations are usually negligible in a good
spectrograph. An instantaneous detector will indicate these fluctuations,
whereas an accumulative (or integrating) detector will not. The absorption
of x-ray quanta on their way from the sample is often a vital matter that is
governed by Eq. (6). Absorption on the way to the detecting medium
reduces intensity and is generally undesirable. Absorption within the detect-
ing medium is desirable as it produces the effect on which the intensity
measurement rests.
In most detectors the quantum effects resulting from the x-ray quanta
will eventually yield electrons. Under the simplest conditions these electrons
appear as separate, well-defined pulses or bundles, one pulse for each x-ray
quantum. The pulses may then be counted individually, and each pulse,
hence each x-ray quantum, will be registered as a unit. As the intensity
increases it becomes increasingly difficult to maintain the individuality of
the pulses.
The principal functions performed by electronic circuitry in the mea-
surement of x-ray intensity are as follows:
(a) Stabilization of x-ray voltage and current.
(b) Stabilization of detector voltage.
(c) Amplification as required to give useful signal.
(d) Shaping of pulses, when required.
(e) Pulse-height selection. In some detectors the mean height of the
pulse (roughly, the size of the electron bundle just mentioned) can
256 H. A. Liebhafsky and H. G. Pfeiffer

be made proportional to the energy of the x-ray quantum. By

means of pulse-height selection, which is a sort of electronic filter-
ing process, it is possible to select pulses of certain heights for
counting and to reject others.
(f) Counting. Individual pulses are counted either by accumulating
them for a fixed counting interval or by measuring the time required
for a predetermined number of counts.
(g) Scaling. The number of counts often is so large that they cannot
be recorded individually, as by a mechanical counter alone. Under
this condition a fixed number of counts (say, 2 or 10) is accumulated
into a larger unit. These units are similarly accumulated into a still
larger unit, and so on. This process is known as scaling.
(h) Integration. A detector that yields discrete pulses can be used as
an instantaneous detector if the pulses can be averaged to form a
continuous electric current, as in a counting-rate meter.
The precision of x-ray-intensity measurements by counting techniques
is limited ultimately by errors inherent in the statistics of counting. In-
stability in the electronic circuitry, which tends to become more serious as
the counting circuits grow more complex, introduces errors in the intensity
measurements that will combine with, and may overshadow, the irreducible
statistical counting error.
Our main concern for the measurement of x-ray intensities is with the
photographic plate, the Geiger counter, the proportional counter, and the
scintillation counter.

3.2.1. Photographic Plate

The photographic plate, especially useful in x-ray diffraction, benefits
as detector because, for an x-ray beam, but not for ordinary light, the same
photographic density results from intensity A acting for time B as from
intensity B acting for time A.

3.2.2. Geiger Counter

In the Geiger counter the absorbed x-ray quantum initially ionizes a
gas, and the photoelectron thus produced is caused by an electric field to
ionize other molecules in the gas-filled counter. Because the field is high
the initial ionization is amplified enormously. The Geiger counter is trouble-
free, easy to use, and particularly valuable for x rays of long wavelength,
for which the energy produced in the initial ionization is low. The Geiger
counter cannot discriminate among x rays of different wavelengths.
9. X-Ray Techniques 257

For identical geometry proportional counters operate with internal

electric fields below those of Geiger counters; consequently, the mean pulse
height (or total energy) can be made proportional to the energy (inverse
wavelength) of the x-ray quantum. The detector can also handle intensities
that give counts approaching a million per sec without serious losses.
Neither advantage accrues to the Geiger counter. Proportional counters are
particularly valuable in x-ray-emission spectrography in the following three
ways: (a) as simple counters, in the manner of Geiger counters, especially
for high counting rates; (b) with a pulse-height analyzer, as a means of
resolving an x-ray beam when the wavelengths of interest have no close
neighbors; and (c) with a pulse-height selector or discriminator to obtain
reduced background.
It has always been difficult to do quantitative work with the character-
istic x-ray lines of elements below titanium in atomic number. These spectra
are not easy to obtain at high intensity, and the long wavelength of the lines
makes attenuation by absorption a serious problem. The use of helium in the
optical path has been very helpful. The design of special proportional
counters, called gas-flow proportional counters, has made further progress
possible. The windows in these gas-flow proportional counters are exceed-
ingly thin, fragile, and unavoidably leaky. Such a window is satisfactory
only if a steady flow of the filling gas is maintained at minimum pressure
differential against the helium atmosphere in the optical path-hence the
term gas-flow counter. The purging of impurities from the counter is an
incidental benefit derived from the gas flow.

3.2.3. Scintillation Counter

In the scintillation counters usual in analytical chemistry, individual
x-ray quanta can be absorbed by a single crystal which is highly transparent
to light (for example, an alkali halide crystal with thallium as activator).
The resultant visible scintillations can produce an output pulse of electrons
from a multiplier phototube. The multiplier phototube internally amplifies
up to a millionfold the photoelectric current produced when the light of
these scintillations strikes the photocathode of the tube. The electrical
pulses that result can be counted similarly to those from a proportional
counter.

3.2.4. Proportional Counter

Proportional counters and scintillation counters can both give electrical
pulses of a height, amplitude, or total energy proportional to that of the
258 H. A. Liebhafsky and H. G. Pfeiffer

x-ray quantum absorbed. Because this energy is inversely proportional to

the x-ray wavelength, pulse-height selection, which is done electronically,
offers the possibility of spectrum analysis. Though inferior in this respect to
Bragg reflection [Eq. (5)], pulse-height selection is useful for intensity
measurements on analytical lines in special cases and for suppressing back-
ground and eliminating unwanted characteristic lines of an order [n in
Eq. (5)] different from that of the analytical line. Pulse-height selection is
less effective with scintillation than with proportional counters. New solid-
state detectors based on germanium and silicon give promise of great future
usefulness because they can give extremely narrow peaks for x-ray lines
under favorable conditions.

4. X-RAY DIFFRACTION

4.1. Description

The invariant angles between crystal faces (or between faces obtained
by cleavage) can be explained only in terms of an internally periodic crystal
structure consisting of parallel planes that contain relatively large numbers
of atoms or molecules and that are separated by a uniform distance d. Any
crystal will have more than one kind of plane, as illustrated in Fig. 4, and
hence more than one value of d, the interplanar spacing or repeat distance.
It may therefore be assumed that

d (Bragg reflection) = dhkl (9)

for every spacing in every crystal. Here d, the Bragg reflection, is either an
experimental value obtained by x-ray diffraction or a theoretical value ob-
tained by a mathematical analysis of the diffraction process; dhk1 is based
on a geometrical analysis of crystal symmetry that makes use of the unit
cell, where the subscripts are the "Miller indices." This oversimplified
summary amounts to saying that crystals are useful three-dimensional dif-
fraction gratings, and that determinations of crystal structure by x-ray dif-
fraction may be relied upon in chemical analysis. For most applications of
this kind, the statement just made can be narrowed further by substituting
"determinations of interplanar spacings (d's) and of concomitant intensities".
Only diffraction from fine powders (Hull-Debye-Sherrer) need be con-
sidered here. A narrow beam of monochromatic x rays impinges upon the
sample, which is usually rotated so as to produce the effect of completely
random crystal orientation, with all possible values of () for the same weight.
9. X-Ray Techniques 259

We have then an experiment at constant A and variable e. Bragg reflection

will occur according to Eq. (5), which means for the simplest geometry (see
Fig. 8) these reflections will form cones about the incident x-ray beam. The
cones will form diffraction rings on a vertical photographic plate as shown.
The numbers are the hkl values for each reflection, in this case, for a face-
centered cubic lattice whose relation to the d spacings was indicated above.
One ring can obviously be formed by reflections from more than one plane.
In practice the diffraction measurements are done differently, since,
not only the spacings, but the intensities of the different reflections are
usually needed. The measurements may be done in a diffraction camera,
usually with a cylindrical photographic film around the x-ray beam as
detector. Alternately, a diffractometer may be used, in which case a trace
is obtained. With the photographic film as detector the entire diffraction
pattern is recorded at once, for which one pays the price of slow response,
need for developing film, and low precision in intensity measurements. The
recording diffractometer has the advantage as regards the last three items.
Complex qualitative determinations show photographic methods at their
best. With some cameras samples need weigh only 0.1 mg.
An x-ray diffraction determination for an unknown substance consists
of (a) sample preparation, (b) generation and interpretation of the dif-
fraction pattern, and (c) comparison of the pattern with that of known
substances. If no pattern is obtained, or if the pattern is so diffuse as to be
useless, the method has shown only that the sample is not well crystallized
and cannot be identified in this way. If the pattern matches a known pattern
for a single substance completely, identification is almost always satisfactory.
If the pattern is good but the match is incomplete, the sample is a crystalline

INC IONT ~
X-RAYS II
SOURCE P
L
E
"""=:=---

DETECTOR
(PiiOTOGRAPHIC PLATEI

Fig. 8. Schematic representation of Hull-Debye-Sherrer powder method for simple ge-

ometry. The x-ray beam transmitted by the sample strikes the photographic plate at T.
After W. L. Bragg, The Crystalline State.
260 H. A. Liebhafsky and H. G. Pfeiffer

mixture for which intensities must be known or estimated if quantitative or

semiquantitative results are desired.
In complex cases the matching of diffraction patterns for sample and
knowns can be a Herculean task. Such matching could proceed effectively
only after Hanawalt, Rinn, and Frevel published their compiled patterns
in 1938. The Joint Committee on Chemical Analysis (sic!) by Powder Dif-
fraction, which includes ASTM, has made available a file of 8000 patterns
and published several books. Frevel 8 devised a comprehensive computer
program to carry out the entire searching and matching process, and
describes the program in a paper that is an excellent critique of the dif-
fraction method. He has gone one step further 9 by using a fully digitized
precision microphotometer in the automated measurement of powder pat-
terns to obtain a printout of interplanar spacings (d's) and of concomitant
intensities.
We conclude by pointing out some relationships of x-ray diffraction
to other x-ray methods. (a) Because diffraction can at best establish crystal
structure, it must in most cases be supplemented by a method that identifies
at least some of the elements in the sample. For this x-ray-emission spectro-
graphy is almost ideal for the heavier elements and gaining in value for the
light elements as analytical lines of longer wavelength (e.g., K lines of F
or Na) become easier to use. (b) In some cases these lines are generated
when diffraction measurements are made, but they now disappear into the
background: simultaneous recording of diffracted and emitted lines for
analytical purposes may some day be possible. (c) Monochromatic K(1.lines
for diffraction work are obtained by filtering out the longer wavelength KfJ
lines according to Eq. (6); this relates diffractometry and absorptiometry.
(d) In diffractometry there is absorption by the sample of incident and
diffracted beams, or there is an absorption effect similar to those in x-ray-
emission spectrography. In both methods internal standards are used to
cope with these effects.
The interested reader will wish to consult Ref. 6 for a complete, and
Ref. 10 for a briefer, treatment of x-ray diffraction as an analytical tool.

4.2. Applications

The high incidence of crystallinity among minerals invites x-ray dif-

fraction studies, and there have been thousands. Two, both involving silica
in some form, have been selected for description because they show the
relation of x-ray-diffraction determinations to geochemical analysis.
Four forms of silica, namely, amorphous, quartz, cristobalite, and tri-
9. X-Ray Techniques 261

dymite, need to be considered and distinguished in the examination of

siliceous dusts that are silicosis hazards. Klug and Alexander (Ref. 6,
pp. 417-433) give an excellent discussion of x-ray-diffraction determinations
on siliceous samples. Noteworthy features are (a) emphasis on careful prep-
aration of samples, (b) use of an internal standard (fluorite), (c) careful
determination and analysis of the various errors, (d) test of method on
known mixtures, (e) comparison of photographic and scanning methods,
(f) early use of automatic recording, and (g) determinations of silica in
three crystalline forms. Needless to say, this is an application in which
conventional wet methods of analysis are at an enormous disadvantage.
The nature of a soil sample is largely fixed by its minerals, especially
by those in the clay fraction, in which primary and secondary silicate min-
erals are of first importance. After pointing this out Whittigl l claims the
following: "x-ray diffraction has contributed more to mineralogical char-
acterization of clay fractions of soils than has any other single method of
analysis." The great care needed in pretreating soil samples shows the com-
plexity of this kind of geochemical analysis. Kunze12 states that removal
during pretreatment of free iron oxide, of soluble salts, of carbonates, and
of organic matter from a soil sample is either desirable or necessary for each
of the constituents named if good x-ray-diffraction patterns are expected.
A monograph by Brindley13 is a guide to earlier work in this field.

5. X-RAY ABSORPTIOMETRY

Little need be said about absorptiometric methods because they are

the least important of the x-ray methods, although an understanding of
x-ray absorption is of prime importance. The methods use either poly-
chromatic or monochromatic beams. Absorptiometry with polychromatic
beams is closely related to radiography and is analogous to colorimetry
with white light. The method is not specific, but it can give a fast response
with simple equipment. An obvious application is, therefore, the control of
industrial processes, or the rapid examination of ores to indicate, for ex-
ample, the total content of the heavy metals. The most interesting use of
monochromatic beams is in measuring absorption on two sides of an absorp-
tion edge as a way of both identifying an element and estimating its amount.
Reference 5 describes both kinds of absorptiometry.
The advent of portable x-ray equipment (see Section 6.4) may make
x-ray absorptiometry more popular as a tool for geologic or mineralogical
exploration.
262 H. A. Liebhafsky and H. G. Pfeiffer

6. X-RAY-EMISSION SPECTROGRAPHY

6.1. Discussion

X-ray-emission spectrography consists of exciting, either with x rays or

with electrons, a characteristic analytical line for each element sought in a
sample. Each individual element is then identified by measuring the wave-
length of the analytical line. The amount of each element present is then
determined by measuring the intensity of the analytical line. From Fig. 9
the usefulness of the method may be deduced for identifying those elements
that give a characteristic K or L line within the range of the spectrograph.
The simplicity of the spectra, the speed with which determinations can be
made, the nondestructive character of the method, its applicability either
to an element or to most of its compounds-all these advantages recommend
the method to every laboratory that is large enough to support an x-ray
spectrograph, or that has x-ray-diffraction equipment that can be converted
into such a spectrograph. X-ray-emission spectrography is already the most
important of the x-ray methods under discussion. If progress continues
toward removing difficulties in the determination of light elements, the
method may become one of the most important for the determination of
elements above atomic number 7 (nitrogen).
Moseley began x-ray-emission spectrography in 1913 with a detector
in the form of a photographic plate, which has now been replaced mainly
by proportional and scintillation counters. The broad scope of the method
has led to a great diversity of equipment. The first spectrographs were
often converted diffractometers. Gradually, emission spectrographs with

.~---.
!'1ltEE
I ) IJ ~ I~

Fig. 9. Chart recording, made in minutes, of the emission spectrum from a

genuine bank note. The goniometer reading for a peak identifies the element
through Eq. (5) and the height of the peak is related to the amount of element
present.
9. X-Ray Techniques 263

air paths were developed. Because air strongly absorbs x rays of longer
wavelength, air came to be replaced by helium. Vacuum spectrographs are
now increasingly available. Curved crystals to focus characteristic lines from
small areas are becoming more popular. The range in size of spectrographs
is illustrated by two instruments at the U.S. Geological Survey. The curved-
crystal spectrograph of Adler and Axelrod 4 takes tiny crystals as samples
and can be placed on a desk top. The other spectrograph is a one-ton Gen-
eral Electric device to be used on lunar samples. Automated spectrographs
for industrial use are either multichannel or sequential. In the former, the
beam from a sample travels simultaneously along as many as 22 channels,
each with its own crystal and detector. In the sequential spectrograph, as
many as 24 analytical lines in an unknown sample can each be compared
at the proper goniometer setting with that of the same line in a standard.
The corresponding angles are traversed sequentially.
The intensity of an analytical line (a) may depend upon the thickness
of the sample, (b) will depend upon the amount present of the element
sought, and (c) will often depend upon the matrix, namely, the other ele-
ments in the sample.
The variation of line intensity with thickness provides an introduction
to absorption effects encountered whenever characteristic lines are emitted
from samples of finite thickness. Consider individual atoms of an element
deposited on a thin substrate highly transparent to x rays-say, atoms of
molybdenum upon paper. Let a characteristic line (say molybdenum K~) be
excited by a polychromatic beam, with the x-ray source and detector both
being located above the sample. As long as the number of molybdenum
atoms is small, they will not noticeably attenuate the incident beam, nor
will an x-ray quantum radiated by any molybdenum atom be absorbed by
any other. Under these conditions, the intensity of the characteristic line
will be proportional to the number of molybdenum atoms and hence to the
thickness of the molybdenum film.
As the film of molybdenum grows thicker the metal will begin to filter
and to attenuate the incident polychromatic beam, which will become
shorter in wavelength and weaker as it penetrates. The intensity of molyb-
denum K<y' will now increase with thickness at a continuously decreasing rate.
If the metal is thick enough, even the shortest wavelength x rays will fail
at a certain depth to excite K<y' quanta at a rate high enough to reinforce
measurably the emergent beam. Virtually all such "deep" quanta will be
absorbed by the molybdenum on their way to the detector. The depth at
which this first occurs is the critical depth for the experimental conditions.
Unless the thickness of a sample exceeds this critical depth, the intensity
264 H. A. LiebhafskY. and H. G. Pfeiffer

of the analytical line will be too low for the amount of the element present
in the sample.
To proceed, let us take for granted that it is logical to relate analytical-
line intensity to weight fraction. Suppose the weight fraction WES of ele-
ment E in sample S is to be determined by measuring the intensity lES of an
analytical line. In the simplest case, we may assume that

(10)

where lEE is the line intensity for the pure element at "infinite" thickness.
Deviations from Eq. (10), called deviations from proportionality, fall
into three principal classes: (a) absorption and enhancement effects, which
are placed together because they both involve absorption; (b) effects trace-
able to heterogeneity in the samples, principally surface effects and segrega-
tion; and (c) instability, including drifts and fluctuations, in the spectro-
graph and associated equipment.
Comparison with similar deviations elsewhere in chemical analysis
shows these to be easier to cope with, because they are to some extent pre-
dictable. Comparison of sample and standard use of an internal standard,
and dilution of the sample with a material such as water, cellulose, or alu-
mina, which are all transparent to x rays, may be very helpful. References
3, 4, and 5 may be consulted in this regard.
Moseley's law [Eq. (7)1 assures us that the K spectra of elements with
low atomic numbers will lie in or near the ultrasoft-x-ray region of, say,
11 A and up. Among these elements are several of great importance in
geochemistry, such as sodium and oxygen. Though work in this x-ray region
is difficult because of low line intensities and absorption effects it has proved
most rewarding.14.15.16

6.2. Applications

The reader is urged to consult Adler's book 4 for an overview of what

this method has already done in geochemistry. The determinations carried
out cover the entire range of the method, including trace to major con-
stituents of large samples, light to heavy elements, qualitative through semi-
quantitative to quantitative results.
Carl and Campbell17 point out that mineral analyses by x-ray-emission
spectrography may be conveniently classified as follows:
Group I. Element of interest variable in a virtually constant matrix.
Examples-Se in S, Th in monazite.
9. X-Ray Techniques 265

Group II. A pair of elements, each independently variable.

Examples-Nb and Ta, Hf and Zr.
Group III. Two or more elements of interest, each independently
variable in a virtually constant matrix.
Examples-Fe and Mn in domestic ores.
Group IV. One or more elements of interest in a variable matrix.
Examples-too diverse to list.

Group I, the simplest, usually requires only that the analytical line be
counted, that a background correction be applied if necessary, and that the
content of the element sought be obtained by simple proportion from counts
made on a suitable standard, which can usually be prepared by adding the
element to a synthetic matrix. The group-II problems listed are of particular
interest because they show the importance of achieving high resolution
without undue loss of intensity. In the determinations of niobium and tan-
talum in the same ore, this was accomplished by a happy marriage of
chemical and x-ray methods. The constant "matrix" in group III is con-
veniently defined as all the sample except the elements being determined,
but a word of caution is necessary. If these elements are present at appreci-
able concentration, then absorption and enhancement effects to which they
give rise must of course be considered, for these effects do not reside in the
"matrix" alone. Group IV is by all odds the most diverse and the most
interesting of the four categories. Because the matrix is variable and often
unknown, an internal standard is almost obligatory for all but the crudest
exploratory work, and Carl and Campbell!7 outline an interesting approach
to the quantitative determination of an element in a sample from this group.
Along with the determinations of silica by x-ray diffraction and of light
elements by x-ray emission, the quantitative analysis of rare-earth ores ranks
as an outstanding opportunity for x-ray methods to prove themselves in
analytical chemistry. Not only do these ores often contain quite a few
elements to be analyzed for, but these elements are generally so similar
that their determination by conventional methods is laborious and costly.
In the hands of Lytle, Botsford, and Heller x-ray-emission spectrography
has proved highly successful in giving quantitative results for a relatively
large number of rare-earth elements in a sample. Their Bureau of Mines
Report of Investigations 5378 repays close study. The analytical problems
faced by these investigators were so complex that they could be solved
satisfactorily only by comparing the unknowns with standards closely re-
sembling them in composition.
The general discussion of applications will have to be concluded by
266 H. A. Liebhafsky and H. G. Pfeiffer

citing only a few of the more important activities directly related to geo-
chemistry. The Annual Conferences on the Applications of X-Ray Analysis
have covered many of these activities, and the papers of the conferences since
1957 have been published. *
As the (present) climax of the x-ray-emission work at Pomona Col-
lege14 - 16 we cite a study18 based on results for 949 samples of the Rattles-
nake Mountain pluton-results obtained with a spectrograph that permitted
determinations of Na, Mg, AI, Si, P, S, K, Ca, Ti, Mn, and Fe on a single
preparation of a silicate rock. The work is noteworthy for, among other
things, the extensive use of computers to automate the x-ray determinations
and to expedite the extensive statistical treatment of the data. Reference 18
is a useful guide to related papers published from Pomona College.
The U.S. Geological Survey has published many applications of the
kind under consideration here. For the earlier work, see Ref. 4. Interesting
among the more recent work is the "solution-dilution" method, which
can cope with small samples and reduce deviations from proportionality,I9
the determinations of individual rare earths in complex minerals,20 and the
use of chemical and electrochemical methods in conjunction with x-ray-
emission spectrography.21 The U.S. Geological Survey has acquired from
the General Electric Co. a "I-ton, 6t-ft" x-ray-emission spectrograph that
can work in the ultras oft range. 22 To be used on lunar samples, the spectro-
graph will give information about elements as light as boron and it can
detect spectral shifts of 0.03 A.
Finally, a few words about the remarkable work being done by M. L.
Salmon and his associates in the Fluo-X-Spec Analytical Laboratory,
Denver, Colorado. A distinctive feature of x-ray-emission spectrography is
this: for many samples, especially samples of minerals, semiquantitative
results of a reliability sufficient for most purposes cost little more than
qualitative results. This laboratory wisely exploits this feature. In the period
1956-1959, it processed about 30,000 samples. A description of its earlier
activities 23 is supplemented in later references.4.24.25

6.3. Electron Microprobe

The modern electron microprobe may well be the most sophisticated

instrument used in analytical chemistry. We restrict ourselves here to its
main function, which is that of an x-ray-emission spectrograph with electron

* Volume 5 of this series, entitled Advances in X-Ray Analysis, has been cited,,,,'5 Vol-
ume 14 deals with the 1970 conference.
9. X-Ray Techniques 267

excitation of analytical lines from small parts of a sample. The diameter

of the region studied may be in the vicinity of 1 ft or 10-4 cm, and the surface
should be electrically and thermally conducting.
As compared with x-ray excitation, electron excitation will usually give
an analytical line of higher intensity with less absorption and enhancement
effects. Its comparative disadvantages are a greater background, a need for
high vacuum, and a greater risk of changing the sample. Apart from the
exploration of a small area, electron excitation has one other advantage,
and that is that the electron beams can be focused to make a point source.
This may give an analytical line from a small area at far higher intensity
than would be possible with x rays. Electrons are generally less penetrating
than x rays of comparable energy, and the region from which analytical
lines reach a detector will therefore be severely restricted in depth as well
as in area. Few samples will be homogeneous enough to give electron-

\26 (ANGULAR
VELOCITY)
ELECTR N SOURCE

ELECTROMAGNETIC
CONDENSER LENS

ELECTROMAGNETIC CURVED
OBJECTIVE LENS ~-c.---#-~ REFLECTING
CRYSTAL
(BENT TO RADIUS 2R)

Fig. 10. Schematic operating diagram (not to scale) for an electron micro-
probe. Only one spectrograph channel is shown. Sample and curved reflecting
(analyzing) crystal lie on the focusing (or Rowland) circle while the detector
is near enough to the circle to intercept the focused analytical line at full in-
tensity. Complex means are used to maintain good focusing when () is changed.
The curved crystal is needed because the x rays come from a point source.
268 H. A. Liebhafsky and H. G. Pfeiffer

microprobe results independent of location, and therefore both the strength

and the weakness of the probe are self-evident.
The bare essentials of an electron microprobe appear in Fig. 10. One
important missing member is an optical microscope to give visual informa-
tion for the area being explored and to help locate the electron beam.
Baird and Zenger26 emphasize the importance of this matter in geochemistry.
What geochemistry needs, and Baird and Zenger have provided, is a micro-
scope akin to petrographic microscopes operating on transmitted light.
It should therefore be apparent that the x-ray-emission spectrograph
for the determinations on all or much of a sample and the electron micro-
probe for the detailed examination of such a sample are really comple-
mentary instruments.

6.4. Portable X-Ray-Emission Spectrograph

The significant contribution x-ray-emission spectrography can make to

geochemistry would become even greater if it could be made in the field
as well as in the well-equipped laboratories that are needed today for
geochemical determinations by this method. Portable spectrographs, such
as that being developed on the basis of British work27 ,28 by the Texas Nuclear
Corp.,29 are badly needed and highly welcome. The problem of this portable
instrument reminds one of exploring for uranium with the Geiger counter,
but the problem is in reality much more difficult. X-rays must be excited
in the field, and more than one element will often have to be determined.
Taking a detector into the field is not enough.
The portable x-ray-emission spectrograph uses radioactive isotopes as
the ultimate source of the energy that excites the analytical lines in the
sample. The excitation process may be simple, as in the case where the
isotope Fe-55 excites the K lines of vanadium or chromium. Here, x-ray
excitation of the sample is accomplished by the K lines of manganese, which
are given off by the source mainly because its radioactive nuclei can capture
its K electrons. In Fig. 11 a "central-source" geometry that gives reasonable
analytical-line intensity even from weak radioactive sources is illustrated.
More complex excitation processes are needed for the effective excitation
of analytical lines for elements over a range of atomic numbers. If a "target
element" is chosen for bombardment by electrons, y rays, or a particles
from the radioactive source, then the target will emit characteristic lines that
effectively excite analytical lines in the sample.
Weak sources that yield intensities that are approximately 10-7 times
that of an x-ray tube will require no special shielding. A complete instrument
9. X-Ray Techniques 269

SAM PLE

x- RAY
~
FILTER _________ ~SOURCE I N
SEALED

HOLDER
F~===:;;;;;;;;;;

DETEC TOR
WINDOW

DETECTOR
SCALE

o 2 3em

Fig. 11. Simple "central source" arrangement of source, sample,

and detector in a portable x-ray emission spectrograph. The
source could be 'Fe and the sample, chromium. A scintillation
counter is the preferred detector. With such an arrangement,
the K lines of manganese would travel from source to sample,
but not from source to detector. X rays from the sample would
reach the detector after passing through two Ross balanced fil-
ters 30 operating sequentially.

weighing about 15 Ib will give rise to analytical lines at intensities that give
103-10 5 counts per sec from pure elements-or, intensities high enough so
that determinations can be made on minerals over counting times ranging
from 10-100 sec.
Some portable spectrographs may be used as is on a flat surface.
Samples of other kinds are placed above the source (see Fig. 11) and cov-
ered with a metal cap. Applications envisaged are the field assay of minerals,
exploration of dry holes, and determinations on slurries of cements and ores,
for example. A special version containing six filters has been used on board
ship for determinations of the principal constituents (Mn, Ni, Fe, Co) in
the manganese nodules present in enomous amounts on the floors of the
Pacific and Indian Oceans. 31 This instrument will not only expedite explora-
tion by enabling one to make determinations almost as fast as one can
travel, but it will help further by giving a basis for quick, on-the-spot
decisions.

REFERENCES
1. Campbell, W. J., and Brown, J. D., Anal. Chern. 40, 346R (1968).
2. Merritt, L. L., Jr., and Streib, W. E., Anal. Chern. 40, 429R (1968).
3. Jenkins, R., and De Vries, J. L., Practical X-Ray Spectrometry, Springer-Verlag,
New York, 1967.
270 H. A. Liebhafsky and H. G. Pfeiffer

4. Adler, I., X-Ray Emission Spectrography in Geology, Elsevier, New York, 1966.
5. Liebhafsky, H. A., Pfeiffer, H. G., Winslow, E. H., and Zemany, P. D., X-Ray
Absorption and Emission in Analytical Chemistry, John Wiley & Sons, New York,
1960.
6. Klug, H. P., and Alexander, L. E., X-Ray Diffraction Procedures, John Wiley & Sons,
New York, 1954.
7. Liebhafsky, H. A., Anal. Chem. 34, 23A (1962).
8. Frevel, L. K, Anal. Chem. 37, 471 (1965).
9. Frevel, L. K, Anal. Chem. 38, 1914 (1966).
10. Liebhafsky, H. A., Pfeiffer, H. G., and Winslow, E. H., in Treatise on Analytical
Chemistry, I. M. Kolthoff and P. J. Elving, eds., Interscience, New York, 1964,
Part I, Vol. 5, p. 3081.
11. Whittig, L. D., Agronomy 9(1), 671 (1965).
12. Kunze, G. W., Agronomy 9(1), 568 (1965).
13. Brindley, G. W., ed., X-Ray Identification and Crystal Structure of Clay Minerals,
The Mineralogical Society, London, 1951, 345 pp.
14. Henke, B. L., in: Advances in X-Ray Analysis, Plenum Press, New York, 1962, Vol. 5,
p. 285.
15. Baird, A. K., MacColl, R. S., and McIntyre, D. B., in: Advances in X-Ray Analysis,
Plenum Press, New York, 1962, Vol. 5, p. 412.
16. Baird, A. K, and Henke, B. L., Anal. Chem. 37, 727 (1965).
17. Carl, H. F., and Campbell, W. J., Am. Soc. Testing Materials Spec. Tech. Publ.,
No. 157, 1954, p. 63.
18. Baird, A. K., McIntyre, D. B., and Welday, E. E., Geol. Soc. America Bull. 78, 191
(1967).
19. Rose, H. J., Jr., Cuttitta, F., and Larson, R. R., U.S. Geol. Survey Prof Paper 525-B,
1965.
20. Rose, H. J., Jr., and Cuttitta, F., private communication, June 18, 1968.
21. Rose, H. J., Jr., and Cuttitta, F., in: Advances in X-Ray Analysis, Plenum Press,
New York, 1968, Vol. 11, p. 23.
22. U.S. Department of Interior, news release, May 11, 1968.
23. Liebhafsky, H. A., Winslow, E. H., and Pfeiffer, H. G., Anal. Chem. 32, 240R (1960).
24. Salmon, M. L., in: Advances in X-Ray Analysis, Plenum Press, New York, 1961,
Vol. 4, p. 433.
25. Salmon, M. L., in: Advances in X-Ray Analysis, Plenum Press, 1964, Vol. 7, p. 604.
26. Baird, A. K., and Zenger, D. H., in: Advances in X-Ray Analysis, Plenum Press,
New York, 1966, Vol. 9, p. 487.
27. Bowie, S. H. U., Darnley, A. G., and Rhodes, J. R., Trans. Instn. Min. Metall. 74,
361 (1964--65).
28. Rhodes, J. R., Analyst 91, 683 (1966).
29. Rhodes, J. R., and Furuta, T., in: Advances in X-Ray Analysis, Plenum Press, New
York, 1968, Vol. 11, p. 249.
30. Ross, P. A., J. Opt. Soc. Am. Rev. Sci. Instr. 16, 433 (1928).
31. Rhodes, J. R., and Furuta, T., Trans. Instn. Min. Metall., Nov. or Dec., 1968.
Chap"ter 10

RADIOMETRIC TECHNIQUES

L. Rybach
lnstitut fur Kristallographie und Petrographie
Eidgenossische Technische Hochschule
Zurich, Switzerland

1. INTRODUCTION
The discovery of natural radioactivity radically affected developments in
science and technology. The study of radioactive rocks and minerals yielded
a vast amount of information on decay processes, half-lives, isotopic abun-
dances, etc. Now that these physical constants have been evaluated, today's
investigations aim towards the determination of abundance and distribution
of the natural radioelements in the earth's crust, in its upper mantle, in the
atmosphere, and even in space.
The most important natural radioelements-uranium, thorium, and
potassium-are very inhomogeneously distributed; to establish their distri-
bution patterns a great number of data is indispensable. Simple and rapid
radiometric techniques have been developed to perform serial analyses.
Statistical analysis of the results (see Chap. 2) is necessary for solution of
geologic problems related to the origin of rock bodies and for both small-
and large-scale assessments of geochemical balance among different rock
units and other material in the earth's crust,!
In geologic processes such as magmatic differentiation, hydrothermal
alteration, or geochemical cycles, the natural radioelements act as tracers;
the Th/U ratio in particular is characteristic of the mechanism involved (see
the excellent review of the geochemistry of uranium and thorium by Adams
et al. 2 ). Radiometric methods are applied in sedimentology to trace the
ancient routes of sediments and to determine the rate of sedimentation. 3

271
272 L. Rybach

Geophysical investigations of the terrestrial heat flow are carried out

to obtain information about the earth's internal structure. The main heat
source is the radioactive heat generation in rocks; the generally accepted
heat production constants are: 0.73 cal/yr/g U, 0.20 cal/yr/g Th, and
27 flcal/yr/g K.4 From U, Th, and K abundances determined by radio-
metric methods, their heat production can be calculated; by combining
radioactive heat production and heat-flow data in a given area one can
construct possible geophysical-geological structure models. 5
In common rocks U, Th, and 4K occur only in trace amounts. To be
economically significant, for instance, uranium contents of several hundred
grams per ton must be present. After a temporary decline in activity during
the early 1960's, the fast-growing demand for nuclear fuel gave impetus to
the world-wide prospecting for uranium, and, to a certain extent, for
thorium ores. In the course of this activity more man-hours were spent
in the search for uranium than for all other metals together throughout
history. At current prices ($ 8/1b U 3 0 S ) exploitation of a uranium deposit
is economically feasible if more than 2000 tons of uranium metal are present
at a concentration of at least 1000 ppm. To evaluate these limits two kinds
of investigation must be carried out.
(a) Determination of the size of the ore body. Proceeding from the
grade distribution at the surface (anomaly map) the depth-structure
below radioactive outcrops is studied by exploratory trenches,
drifts and crosscuts, and/or by drilling.
(b) Determination of the uranium concentration distribution. Besides
semiquantitative determinations in situ this is done by analyzing
a large number of samples taken at the surface, from the drift
walls, in drill holes. Radiometric methods are applied in almost all
phases of exporation.
Age dating is an independent method of solving such geologic problems
as the origin of formations, metamorphic processes, volcanic activity, etc.
Though modern age dating is carried out by mass spectrometry preliminary
screening of accessory mineral concentrates by radiometric techniques sim-
plifies the dating procedure considerably.

2. FUNDAMENTALS

2.1. Nuclear Data of Natural Radioisotopes

All elements having an atomic number greater than 83 are radioactive.
They decay towards stable end products by emitting alpha and/or beta
10. Radiometric Techniques 273

Table I
Some Naturally Occurring Radioisotopes"

Isotopic Abundance in
Radioactive Type of Half-life, Stable end
abundance, lithosphere,
isotope disintegration yr product
% ppm

~~K p, EC 1.3 X 10' 0.0118 3.0 ,0Ca, .0A

:;Rb p 5.2 x 1010 27.85 75.0 "Sr

1:~Sm a 1.1 X 1011 14.97 1.0 143Nd

l;~Lu p, EC 2.2 X 1010 2.59 0.01 l'6Hf

l~~Re p 5.0 X 1010 62.93 0.001 18'OS

a From Chart of the Nuclides, 9th Ed., USAEC, 1966.

particles (the latter often accompanied by gamma radiation with character-

istic energy) or by electron capture (EC). Whereas naturally radioactive
isotopes of several lighter elements (some of them are listed in Table I)
reach their stable end products in most cases through a simple one-stage
decay process the long-lived heavy radioactive elements decay into daughter
products that are in themselves radioactive. They in turn decay and form
multistage chains (see Table II) which end when finally a stable daughter
isotope is formed. The uranium, thorium, and actinium series start with
23SU, 232Th, and 235U as parent members, the lead isotopes 206Pb, 2osPb,
and 207Pb, respectively, being the stable end products.
The decay of any type of radioactive nucleus can be described by the
exponential law. Given No radioactive atoms at a time t = 0, there are

N = No exp(-At) (1)

atoms at any given later time t. The decay constant A is the probability of
decay per nucleus per unit time and is related to the half-life Tl/2 by

T, _ 0.693
1/2 - --A- (2)

For N nuclei present, the activity, or number of decays per second, is AN.
The activity unit is I Ci which is defined as 3.7 X 1010 disintegrations/sec
274 L. Rybach

Table II
Half-Lives, Energy Levels, and Abundances for Members of the Naturally Radio-
active 238U and 232Th Seriesa

Isotope Half-life Radiation Energy, MeV, abundance (in parentheses), %

I::U 4.5 X 109 yr ex 4.18(77), 4.13(23)

2:~Th 24.1 day p 0.19(65), 0.10(35)
l' 0.09(15), 0.06(7), 0.03(7)
2:~Pa 1.18 min p 2.31(93), 1.45(6),0.55(1)
l' 1.01(2), 0.77(1), 0.04(3)
I::U 2.50 x 105 yr ex 4.77(72), 4.72(28)
l' 0.05(28)
2:gTh 8.0 X lOt yr ex 4.68(76), 4.62(24)
2::Ra 1622 yr ex 4.78(94), 4.59(6)
l' 0.19(4)
2::Rn 3.82 day ex 5.48(100)
2~:PO 3.05 min ex 6.00(100)
1~:Pb 26.8 min p 1.03(6),0.66(40),0.46(50), 0.40(4)
l' 0.35(44), 0.29(24),0.24(11), 0.05(2)
2~:Bi 19.7 min p 3.18(15),2.56(4), 1.79(8),
1.33(33),1.03(22),0.74(20)
l' 2.43(2), 2.20(6), 2.12(1),
1 .85(3), 1.76(26),
1.73(2), 1.51(3), 1.42(4),
1.38(7), 1.28(2), 1.24(7),
1.16(2), 1.12(20),
0.94(5), 0.81(2), 0.77(7),
0.61(45), 0.29(26), 0.24(12), 0.035(45)
2~:PO 160 x 10-6 sec ex 7.68(100)
2J~Pb 19.4 yr p 0.06(17), 0.02(83)
l' 0.05(4)
2~~Bi 5.0 day p 1.16(100)
~~Po 138.4 day ex 5.30(100)
2~:Pb Stable
1:~Th 1.41 x 1010 yr ex 4.01(76), 3.95(24)
l' 0.06(24)
10. Radiometric Techniques 275

Table II (continued)

Isotope Half-life Radiation Energy, MeV, abundance (in parentheses), %

2::Ra 6.7 yr f3 0.05(100)

2:ZAc 6.13 h f3 2.18(10), 1.85(9), 1.72(7),
1.13(53), 0.64(8), 0.45(13)
Y 1.64(13), 1.59(12), 1.10, 1.04,0.97(18),
0.91(25),0.46(3),0.41(2),0.34(11),
0.23, 0.18(3), 0.13(6), 0.11, 0.10, 0.08
2i~Th 1. 91 yr a 5.42(72), 5.34(28)
Y 0.08(2)
2::Ra 3.64 day a 5.68(95), 5.45(5)
Y 0.24(5)
2:~Rn 54.5 sec a 6.28(99+ )
2~:pO 0.158 sec a 6.78(100)
2~~Pb 1O.64h f3 0.58(14),0.34(80), 0.16(6)
Y 0.30(5), 0.24(82), 0.18(1), 0.12(2)
2~:Bi 60.5 min a 6.09(10), 6.04(25)
f3 2.25(56), 1.52(4),0.74(1),0.63(2)
Y 0.04(1), with a 2.20(2), 1.81(1), 1.61(3),
1.34(2),1.04(2),0.83(8), 0.73(10), with f3
'~!Po 0.30 X 10- 6 sec a 8.78(100)
'~~Ti 3.1 min f3 2.37(2), 1.79(47), 1.52, 1.25
Y 2.62(100), 0.86(14), 0.76(2), 0.58(83),
0.51(25), 0.28(9), 0.25(2)
'~~Pb stable

a From Ref. 58.

(dps). The fractional units mCi (10- 3 Ci) and ,uCi (10- 6 Ci) are more fre-
quently used. The activity of 1 kg 238U is about 0.3 mCi.
For a number of radioactive elements (A, B, C; T1I2A very long) in a
decay chain which are in conditions of natural disintegration in an isolated
system for prolonged periods, a state of radioactive equilibrium is reached
which is determined by the equation

(3)
276 L. Rybach

This holds especially true for the uranium and thorium series where the
parent half-life is very long as compared to the daughter half-lives (see
Table II). Thus, when a radioactive decay chain is in equilibrium the activity
of each radioactive daughter isotope, measured in Ci, is the same. That is,
in a rock sample the daughter isotopes are present in amounts which are
proportional to the respective half-lives (e.g., 0.3 mg Ra in 1 kg U).
Geologic processes such as erosion, leaching, etc., can separate parent
and daughter isotopes, since they are chemically different. The gas
members of the 238U, 235U, and 232Th series, 222Rn, 219Rn, and 220Rn,
could escape from the system which would seriously affect the degree of
equilibrium.
The presence or absence of radioactive equilibrium within a rock or
mineral is of fundamental importance in radiometric determinations. Ura-
nium, for example, is usually determined by measuring the gamma radia-
tion emitted by the radioisotope 214Bi, a member of the 238U decay chain.
If Eq. (3) were not fulfilled for a particular sample, the uranium value found
by this means would be erroneous. Section 5.4 describes methods to master
this situation.
Several modes of decay may be possible for a radioactive isotope so
that several a particles and y rays of discrete energies may be emitted (the
energy is usually given in MeV units, I MeV = 1.6 X 10-13 J). For a par-
ticular nucleus each mode of decay has a definite probability, so that the
energy distribution of the emitted particles is characteristic of that isotope.
Emission probabilities (or abundances) of the different members in the 238U
and 232Th series are given in Table II; beta energies are the maximum values,
since beta emission has a continuous energy distribution. The relatively
simple decay of 4K is shown in Fig. 1.

274j3/seC,9

E m ... 130Mev

Fig. 1. Decay scheme of .oK. Branching ratio (EC-gamma

emission to beta emission) = 0.123.
10. Radiometric Techniques 277

2.2. Background Radiation

Radioactivity measured by a detector in the absence of a sample is

called the background. This background arises from different sources:
(a) cosmic radiation, (b) radioactivity in the environment, and (c) inherent
radioactivity of the detector assembly.
(a) The primary flux of cosmic rays contains high-energy protons
(0.19 p/cm2. sec sr), alpha particles (0.03 a/cm 2sec.sr), and a
fraction of lighter nuclei (Li, Be); the incident energy being of the
order of 1.5 BeV/cm2sec.sr. Their reaction with nuclei of the
earth's atmosphere produces a complex mixture of mesons, muons,
hard and soft gamma rays, etc. (for more details see the excellent
review by May and Marinelli 6 ).
The intensity of this secondary radiation increases markedly with
barometric altitude (roughly by a factor of two between 6000 and
12,000 ft). A much slighter variation with the geomagnetic latitude
is also present. Considerable, but never total, absorption can be
achieved by lead, iron, or mercury shields (see, also, Section 3.5).
(b) The environmental radiation itself is a sum of different contribu-
tions. Trace amounts of radioactive elements are present in any
kind of material. The floor and walls surrounding a laboratory
detector, even the shielding, contain minute quantities of potas-
sium, uranium, and thorium. This contribution is constant, and
thus can easily be considered.
A rather variable portion (usually less than 5% of the total back-
ground) represents the atmospheric radiation. Its main component
is 222Rn from the 238U series diffusing from the earth's surface. The
mean flux density of the exhalated Rn atoms is about! X 2 X 10-10
,uCi/cm2. sec, depending on the soil porosity [7 The 222Rn con-
centration in the air is strongly dependent on meteorological factors
like barometric pressure, precipitation, etc.; the daily variation can
extend over more than one order of magnitude. 8
Fission products from nuclear-weapon tests may also contribute to
the atmospheric radioactivity. Washed out by fallout, these radio-
nuclides can add substantially to the normal soil radioactivity (up
to 30% of the total background9 ). The main product, 95Zr-95Nb
decays with a half-life of about 60 days.
(c) Usually, special care is taken to select pure materials for detector
construction. However, trace amounts of potassium, uranium,
thorium, radium, etc., cannot be completely eliminated even with
278 L. Rybach

complicated procedures. Radioactivity traces are present in

scintillation crystals, in the glass window of the multiplier photo-
tube* (mainly potassium), etc. These contributions are constant
for a given detector assembly. Besides these sources nonradio-
active effects can also contribute to the background: the dark
current of the phototube in a scintillation detector, and extremely
small amounts of light penetrating to the cathode of the phototube.
Some of the components of the total background radiation can
be reduced considerably. For detailed discussion see Section 3.5.

2.3. Interaction of Radiation with Matter

Interactions of charged particles and gamma rays make their detection
possible. Alpha particles interact with matter by removing the electrons
from the target atom shells (ionization). The range or depth of penetration
depends on the incident energy and the absorbing material. A I-MeV alpha
particle has a range in aluminum of about 0.0004 cm only. Electrons (beta
radiation) can penetrate deeper: for I-MeV electrons the range in aluminum
is 0.15 cm. Besides ionization, electrons may lose energy by radiation as
well (bremsstrahlung), depending on the atomic number Z ofthe absorber:
Radiation loss EpZ
(4)
Ionization loss = 800
where Ep is the incident energy. In geologic materials (solid rocks or pow-
dered samples) the maximum range of betas is a few millimeters only.
The production of secondary electrons gives the key to the detection
of gamma rays. Gamma rays can be scattered and/or absorbed in three
ways. In the first way, a gamma-ray photon strikes an orbital electron
which carries away a part of the initial energy Ey; the deflected gamma
photon goes on with diminished energy, depending on the angle of the
emerging electron. This type of interaction is called the Compton effect.
There is no direct relationship between secondary electron energy and
incident gamma energy. The electron energies are continuously distributed
between zero and a maximum value, given by
Ep
Emax = ---"---:- (5)
1 + moc2
2Ey
where moc2 is the rest energy of the electron (0.511 MeV).

* For low-background application it is desirable to remove the mica-filled plastic base lo

10. Radiometric Techniques 279

The second type of interaction is called the photoeffect. The whole

incident energy of the gamma quantum is used up to eject a photoelectron.
The binding energy Eb in a given electron shell is constant, thus a direct
relationship exists between secondary electron energy Ee and incident
gamma energy Ey:
(6)
or
(7)

The third kind of interaction is the transformation of a gamma ray

into a positron-electron pair. The positron immediately combines with an
orbital electron, annihilating both and releasing, in opposite directions, two
gamma-ray photons (the process is called, therefore, pair production) of
0.511 MeV each. Therefore, this type of interaction occurs only if Ey ;;:;; 1.022
MeV. The secondary electron energy is

Ey - 1.022
Ee= 2 (8)

The probabilities for Compton scattering, photoelectric absorption,

and pair production (cross sections T, (J, and ", respectively) depend on the
absorber material (atomic number Z) and incident gamma energy (Ey):

T =f(Ey , Z) ex: Z4
(J = f(Ey , Z) ex: Z (9)
" = f(Ey , Z) ex: Z2

Gamma rays do not have definite ranges but are attenuated exponen-
tially with increasing thickness of absorber. The intensity 1 after passing
through x cm of material is

1 = 10 exp(-f-lx) (10)

where f-l = T + (J + ". Since f-l is usually given in units of cm 2 /g (mass

absorption coefficient, a), in Eq. (10) the density e (g/cm 3 ) must be intro-
duced:
1 = 10 exp(-aex) (11 )

Figures 2a and 2b show the energy dependence of a, T, (J, and" in the most
frequently used shielding (Pb) and detector (NaI) materials. Photoelectric
280 L. Rybach

10

\"'/P
\
.1
~
01 "
~~---
\
,-,I
/ \ / .....
/"~I
l'
01 A
/,,-
_.1-__
~
~ \
-'I" KII'
/\
.....
. ",/ Tip I / 'I TI P \ I ,
/" 0-1' 't\.......... / Up \ I "
.... " \ I
I \..... \ I

.001
\
'-
' .001 '------'----~.,.,---'-----'
01.1 10 MeV Ey
01.1 10 MeV Ey
Fig. 2. Total mass-absorption coefficient (a = p/e), photoelectric absorption (T/e),
Compton scattering (a/e), and pair production (x/e) cross sections in (a) lead and (b) NaI.

absorption is predominant at low, Compton scattering at medium, and pair

production at high energies.
With increasing absorber thickness originally parallel gamma rays are
more and more scattered diffusely and a deviation from the exponential
law occurs (Fig. 3): scattered gammas can join the original ones and the
effective absorption decreases.l l To describe this phenomenon the buildup
factor B is introduced:
1= IoB(ax') exp(-ax') (12)

log( 1/101

Fig. 3. Deviation from the ideal exponential

law of gamma-ray absorption due to the
buildup effect.
10. Radiometric Techniques 281

where x' = X(! (g/cm 2 ). B itself is a function of the absorber material and
thickness and of the incident gamma energy. In most cases one has to deal
with this type of absorption:

3. RADIATION DETECTORS

3.1. Beta and Gamma Measurement*

The function of radiation detectors is to convert the energy of incident

particles of gamma photons into pulses of electrical energy that can then
be measured. The classical and still widely used radiation detector is the
Geiger (GM) tube. Though alpha particles and, to a certain extent, gamma
rays can be detected, its main application is the measurement of beta ra-
diation. For laboratory measurements the end-window type is commonly
used (the window consists of a very thin, <2 mg/cm 2 mica foil), while for
field prospecting thin wall (30 mg AI/cm 2 ) types are more convenient. Fig-
ure 4 shows a circuit for adapting a longer cable between the detector and
the pulse-registering counter.
The number of pulses produced in a GM tube (for a given radiation
intensity) is relatively independent of the high voltage applied (if operating
in the "Geiger-plateau" voltage range), which allows the use of simple,
un stabilized power-supply units. Disadvantages are numerous:
(a) After each registered pulse the counter is blocked for a certain
time. This time ("dead time") is in the order of 200 ,usec/count.
Especially at higher count rates (the usual units are cpm, counts
per minute, or cps, counts per second) a correction for this effect
must be made: the effective incident count rate Aeff. can be calcu-
lated by
A eff -_ A reg.
~ Areg.(1 + TAregJ (13)
. 1 - TAreg.

where A reg . is the count rate measured and T is the dead time.
T can easily be determined by the two radioactive-source method. 12

(b) The counting efficiency for gammas is very low (around 1%).
Though a certain improvement can be achieved by coupling several

For practical use a vast number of counting devices, electronic equipment, etc., exist
on the market. It is far beyond the scope of this article to select or to compare different
types and manufacturers. If reference is given to a particular type this is because the
writer has special experiences with that model or unit.
282 L. Rybach

,...---,.-- 12 V

lOM O 10,uF

l MO 150KO c.bleto
tf'll nsmU -
pulses

Fig. 4. Emitter-follower circuit with a low-impedance output

for a longer cable between GM tube and registering unit.

GM tubes in parallel, scintillation counters are definitely better

suited for gamma registration.
(c) There is no relation between the amplitude of the pulse generated
by the incident particle and particle energy (all pulses have the same
height). Thus, no spectral differentiation is possible.
The latter two shortcomings are eliminated with a scintillation detector,
which is the most commonly used instrument for gamma measurements.
Figure 5 shows the principle of the scintillation counter. The main parts,
the crystal (usually sodium iodide containing a trace of thallium) and the
multiplier phototube, are mounted together. Gamma rays entering the
crystal produce secondary electrons by the three processes mentioned in
Section 2.3. These electrons cause the crystal to scintillate. The scintillation
light is converted to electric pulse and amplified in the attached phototube.
The photocathode and dynodes of the multiplier phototube are made of
material that is capable of producing secondary electrons and are maintained
at successively higher positive potential.

Na I(TI) crystal multiplier pho totube

gla.s
~1i50V +450V

light . light oIuml nlum cop

.bsorbs mo.t of jl radlotlon

Fig. 5. Scintillation detector for gamma counting. (NaI is

hygroscopic and must be hermetically sealed.)
10. Radiometric Techniques 283

The gain V obtainable from a multiplier photo tube depends mainly

on the voltage Ed between each of the n dynodes:

(14)

where k is a constant, depending on the dynode material. Usually a resistor

string supplies the individual potentials to the dynodes. If there is a small
change LlEd in Ed' from Eq. (14) if follows that

V Ed
(15)
LlV = n LlEd

thus, stabilization of the supply voltage within close limits is very important.
Different gain-stabilizing equipment is now commercially available (see also
Section 5.3). The amplification of commercial phototubes is approximately
108 Even with such high amplification the output signal is relatively small
(as compared to a GM tube) and special additional electronics are necessary.
The dead time i is, on the other hand, about 103 X less than for a GM
counter.
The amount of scintillation light (and thus the pulse height at the
photomultiplier output) is proportional to the energy of the secondary
electron. Depending upon the process involved a unique pulse height (photo-
electric absorption) or a continuous distribution (Compton scattering) will
appear at lower and medium incident energies. At high energies a positron-
electron pair is produced. The positron is immediately annihilated, giving
rise to two O.51-MeV gamma rays. If both of these gammas are stopped in
the crystal the pulse height will correspond to the incident gamma energy E y
If one or both gammas can escape the pulse height will be less, correspond-

100r------:1.7.46:7.Mey:--...,

.
photOpeH

~
a 50

i
c Compton
continuum

0.5

Fig. 6. Gamma-ray spectrum of .oK measured

by scintillation detector.
284 L. Rybach

100~------------------r------'
1.52 Mev
pholopeak --

1.84 Mev
1.01 Mev sum peak
pa~peak I

0.5 1.0 1.5

Fig. 7. Gamma-ray spectrum of <OK showing

pair and sum peaks.

ing to Ey -0.51 MeV or Ey -1.02 MeV. Thus, for a simple radioisotope like
4K emitting monoenergetic gamma rays (see Fig. 1) a whole distribution
of pulse heights occur (Fig. 6). A similar spectrum is shown by the artificial
isotope 42K (Fig. 7). Besides photopeaks, Compton continua, and pair
production peaks, three additional features are visible in Figs. 6 and 7.
(a) The effect of instrumental resolution (a broad peak instead of a
single line appears). With the half-width L1 E the resolution r is
given by
r(%) = 100 ~ y
(16)

NaI(TI) crystals have a resolution* of about 7% in the region of

0.7 MeV.
(b) The bremsstrahlung increase. When beta particles are stopped in
matter a certain fraction of their energy is radiated as brems-
strahlung (photons). A continuous energy distribution of photons
results which predominates at lower energies. For radioisotopes
with high beta branching directly to the ground state (branching
ratio for 4K is 8 : 1, see Fig. 1) the contribution of the brems-
strahlung to the gamma-ray spectrum is considerable. The proba-
bility of bremsstrahlung emission increases with Z2 of the absorber.
Thus, for beta absorbers low Z material (Be, polyethylene) must
be used (see detailed discussion in Ref. 10, p. 19).

* Much better resolution can be achieved by solid-state detectors, e.g., Ge(U). However,
their sensitivity is insufficient for the activities encountered in normal radiometric
practice. Figure 8 shows a spectrum of a relatively strong source (~O.2 ,uCi).
10. Radiometric Techniques 285

Pb 21

Pb 214

U235

Pt 214

\ e 214 81 2'4

0.5 1.0
1.S

Fig. 8. Gamma-ray spectrum of the mineral brannerite (U, Th, Ca)(Ti, Fe).06' A I-g
_
r
substance at 6 cm distance from a 20-cm 3 Ge(Li) detector. Note the sharp lines of ,uBi.

(c) Summation peak (at 0.32 +

1.52 = 1.84 MeV) in the 42K spectrum
which occurs because of the finite resolving time of the scintilla-
tion detector. This effect is especially important at higher activities,
since the probability p for the overlap of decay events (they will
sum in amplitude) in time is given by
(17)

where A is the incident count rate and T the overall resolving time.
For further details see Ref. 13, p. 18.

3.2. Alpha Measurement

Of the different possibilities of alpha-particle detection the two most
popular detector types should be mentioned here. Zinc sulfide as a scintil-
lator has been utilized since the early days of nuclear research. ZnS in
contact with a multiplier phototube is still a very useful device for carrying
out integral alpha counting (see Section 5.5). Besides gas-flow GM counters
operating in the "proportional region" (see Section 3.1) and ionization
chambers for special applications,14 solid-state detectors are commonly used,
especially for alpha spectrometry.
286 L. Rybach

V. aluminium back
ele<:lrode

Fig. 9. Surface-barrier-type detector diode

(schematic). D, diameter; L, total thickness;
W, depth of the sensitive (depleted) region.

Solid-state detectors are usually of the surface-barrier type: these are

basically semiconductor (silicon) wafers between deposited electrodes (see
Fig. 9). An incident alpha particle releases ionization charge carriers which
are separated and collected by the applied electric field Vb. The detector
area Q = D2 n /4 determines the sensitivity (counting rate) of the device,
and therefore large-area detectors should be used. On the other hand, the
signal capacity of the diode, which mainly determines its noise, is

(18)

Thus, a compromise must be made in selecting an appropriate sensitive

area.
A simple equivalent circuit is shown in Fig. 10, where Rs is the diode
series resistance, RL corresponds to the reverse leakage in the diode, Rb is
the load resistor through which the bias voltage is applied, Cs represents
all of the stray capacity in cables, etc., and CA is the input capacitance of
the associated preamplifier.
Detector

r
R. Rb

Cd RL ..;. Vb

1
Fig. 10. Equivalent circuit for surface-barrier
detector used in alpha spectrometry.
10. Radiometric Techniques 287

1000",

10 '" L---~10-----'::!100-=-=='Mev
0< particle energy

Fig. 11. Range of alpha particles in silicon.

If the total energy of the incident particle is deposited in the sensitive

(depleted) region the output pulse height will be proportional to the incident
energy. Thus, the maximum range of the particles to be analyzed establishes
the minimum usable sensitive depth. Figure I I gives the ranges for alpha
particles with different energies in silicon.
Alpha sources emit particles of discrete energies ("line spectrum"). In
this type of detector, again, a broadening occurs; the measured spectrum
shows peaks, instead of lines (see Fig. 12). The instrumental resolution is
usually given as the peak width at the half-maximum height (FWHM).

4
cpn

2 4 8 8 Mev
ol pertlcle -.gy

Fig. 12. Alpha spectrum of the mineral thorite

(ThSiO.) measured by a I-em" solid-state
detector. Resolution ca. 5% FWHM (after
Cherry et aU3 ).
288 L. Rybach

Table III
Semiconductor Alpha-Detector Dates

Sensi ti ve area, Depletion depth, Resolution FWHM,

mm" fI keY

25 100 18-25 a
100 100 20-35
200 100 25-40
450 100 35-55
950 100 70-100

a This depends on the ORTEC model.

Table III lists typical FWHM values for different detector sizes measured
at the 5.48-MeV 24IAm peak. Since no secondary processes take place in
alpha absorption, measured alpha spectra show simple forms, making their
interpretation simple.
It should be mentioned here that the introduction of portable x-ray-
fluorescence analyzers to mineral exploration has been a success. I5 The
surface of the rock to be analyzed is irradiated with low-energetic gamma
rays from an artificial radioisotope source (55Fe, 238PU, I09Cd, 3H-Zr, and
147Pu-AI are commonly used). The excited characteristic x-rays (see Chap. 9)
are filtered and registered by a scintillation counter. Though only the outer-
most "skin" of the rock is analyzed satisfactory results have been obtained,
especially in the case of tin ores.

3.3. Scaler and Ratemeter Circuits for Integral Count-

ing, Pulse-Height Analysis

Detectors supply electric pulses with a frequency proportional to the

incident radiation intensity. The number of pulses can be counted by a
scaler or their time average can be indicated (or recorded) by a ratemeter.
Usually the pulses are discriminated, amplified, and shaped (PS) before
registration (Fig. 13). A threshold circuit (T) discriminates the pulses below
a given pulse height; low pulses (e.g., from the dark current of the multiplier
photo tube) are eliminated. The high voltage supply (HV) must be checked
and, if necessary, readjusted from time to time.
The scaler records the individual random pulses. Decimal scalers with
five decade stages are most frequently used, since a large number of events
10. Radiometric Techniques 289

scaler

cc=f[JC- - -- -- -- -
emitter
follower
..
detector probe unit

rate meter

Fig. 13. Schematics of a scintillometer for integral gamma measurements.

must be recorded to attain adequate precision. The first stage records up

to 10 pulses. The tenth pulse cleans out the record in the first stage and
transmits one pulse to the second stage. When 10 pulses have been received
by the second stage, it in turn transmits a pulse to the third stage, and so on.
In general, with n decade stages IOn pulses can be stored. In a counter with
five decades the first stage counts the individual pulses, the second stage
counts the tens, the third stage the hundreds, the fourth stage the thousands,
the last stage the tens of thousands. If the numbers 6, 3, 0, 1, 0 show up,
respectively, the total number of pulses counted is

6 xl +3x 10 +0 x 100 +1x 1000 +0 x 10,000 = 1036 pulses (19)

The precision of this measurement (or of any measurement of N pulses) is

a= VN = .YI036 = 32.2 pulses (20)

The probable error p is

p(%) = ~ (21)
VN
which means a probability of 0.50 that the experimental value N differs
from the true value by p. The count rate is given by

A = (~ V;V) cpm or cps (22)

where t is the time (in minutes or seconds) during which the N pulses were
accumulated.
290 L. Rybach

Above a certain limit (> 10 cps) direct indication of the count rate can
be given by ratemeters. In an integrating RC circuit (Fig. 13) the current
through the resistor R is a measure of the time average of the count rate A
and is indicated on a meter or recorder. The time constant R C is variable
on most instruments (1-20 sec) and the precision of the reading is
100
a(%) = --- (23)
V 2ARC

This equation is valid if the reading has reached its "equilibrium" state (the
capacitor C must be charged) which is established after a time of

In = RC(O.S In 2ARC + 0.394) (24)

In is short for short time constants; on the other hand, the precision of the
reading will be relatively low. Usually a compromise must be made by se-
lecting a medium time constant.
The time constant can be determined, if unknown, by following the
decrease of the reading after removing a radioactive source: during the time
constant RC the reading decreases to 37% of its original value (1 Ie ~ 0.37).
More complicated electronics is necessary if pulse-height analysis (for
a and y spectrometry) should be performed. As described in Sections 3.1
and 3.2 NaI(TI) crystals and Si or Ge(Li) solid-state detectors supply
electric pulses with pulse heights proportional to the incident a or y energy.
The basic principle of pulse-height analysis is single-channel discrimination:
since only pulses above a selectable limit can pass a discriminator, two dis-
scriminators followed by an anticoincidence unit can select pulses with
amplitudes between the upper (Uu ) and lower (UI ) discriminator level
(Fig. 14). Pulses above Uu (b, e, and f in Fig. 14) pass both discriminators
and are eliminated by the anticoincidence circuit; pulses below UI (a and d)
are rejected by the discriminators; pulse c which lies between Uu and U

discriminators

PUheight
H l s e.u" }~U
UI
- - _ to scaler
detector ampli
time fier
abc def

MJI

Fig. 14. Principle of pulse-height analysis (see text).

10. Radiometric Techniques 291

(within the "window" or "channel") appears only after d2 , and thus the
anticoincidence unit will deliver a pulse to the counter. Uu - Ul = Ll U is
called the channel width.
Operating several "windows" side by side the whole pulse-height
spectrum can be analyzed at the same time. Multichannel analyzers with at
least 100 channels are now routinely used in radiometric practice. During
a preselectable time the pulses per channel are accumulated and the number
is stored in a ferrite-core memory. The information is usually stored in BCD
format and can be made available by various means: the decimal number
may be recorded by using a typewriter or printer; BCD information may
be recorded in computer format on punched cards, punched tape, or on
magnetic tape. Digital information also may be converted to analog voltages
which are used to display spectra on a cathode-ray tube and may also be
used to drive an X-Y curve plotter.

3.4. Autoradiographic Methods

Autoradiography is certainly the oldest method for radioactivity deter-
minations and is still the best way to study the distribution of radioactivity
on the hand-specimen scale. In uranium ore-grade materials the radioactive
constituents (e.g., pitchblende, brannerite grains) are often inhomogeneously
distributed (Fig. 15) and very small (50-300 p,). Autoradiographs taken in
direct contact with a polished section of a rock sample permit the exact
localization of its radioactive mineral constituents.
The experimental procedure is very simple: the autoradiographs are
taken in a light-proof box by placing the specimen under study on the photo-
graphic plate or film. In case of uncovered thin sections a good contact
must be maintained by superimposing a weight or by clamping. Movement
of the rock specimen relative to the emulsion during the exposure must be
avoided. Exposure times depend upon the emulsion used, as well as on the
specific activity of the radioactive minerals present, and range from a few
minutes to several hours.
Using a-sensitive emulsions (Kodak NTB, Ilford K.2) estimates of
the U and Th contents can be made by counting the number of the individual
a tracks (Na.) per cm 2 under a microscope: 16
(25)

where T is the exposure time. The ThjU ratio (roughly 4 for common rocks,
0.03 for many types of uranium mineralization) must be assumed. For
detailed discussion see Ref. 17.
292 L. Rybach

Fig. 15. (a) Contact autoradiograph of a quartz

conglomerate (permian Verrucano, Switzerland)
showing intergranular distribution of brannerite
(white). Kodak single-coated medical x-ray film,
exposure 30 h. Natural size. (b) Contact auto-
radiograph (positive) of rock specimen and stan-
dards (bottom) for quantitative determinations.
Standard pellets have uranium concentrations
(from left to right) of 0.25, 0.5, 1.0, 2.0, and
4.0% U.
b
10. Radiometric Techniques 293

Comparing the image density produced by chemically analyzed stan-

dards to that of an unknown sample, quantitative assessments can be made.
A photodensitometer is used for this purpose. IS

3.5. Background Reduction for Laboratory Measure-

ments

For all radiometric determinations the background count rate Ao (cpm)

must be subtracted from the measured, gross count rate (sample + back-
ground) A. The statistical standard deviation (J'A in the net-sample count
rate is given by

(26)

where t and to are the counting times for the sample and background
measurement, respectively. In normal laboratory practice Ao is determined
periodically by counting the background for much longer times than sam-
ples. To compare different counting setups the sensitivity A2jAo is used;
for assaying low-level samples improvements can be made either by increas-
ing the count rate for a given sample or by reducing the background.
The simplest method of background reduction is the installation of
a massive shield surrounding the detector. Different materials and shield
thicknesses are used. The most common shielding material is lead. Steel
and (triply distilled) mercury are also frequently utilized. Further back-
ground reduction can be obtained by installing the counting laboratory in
the basement of a building.I9 Careful selection of the construction materials
(concrete ingredients, cements) may also help.20 A shield of 5 in. oflead or
an equivalent thickness of other nonradioactive material provides about

h Is'
4" Pb"
.-f.

Nol

HV J j
119.; ;:j
L]4'
Fig. 16. Typical shielding construction for
laboratory scintillation detectors.
294 L. Rybach

the optimum shielding from the external natural radioelements. Figure 16

shows a shielding type which is used in many laboratories.
Cosmic radiation (see Section 2.2) can be drastically reduced by a
Geiger anticoincidence ring,21 but the intrinsic radioactivity of the detector
itself (see Section 2.2) will still be present. Special care must be taken to
select low-level detector components (Ref. 10, pp. 81-84).

4. RADIOMETRIC PROSPECTING

4.1. Airborne Radiometric Survey

The most typical geophysical field around a radioactive ore deposit

(like the magnetic field caused by a magnetite ore body) is the radioactive
anomaly. By definition the radiation intensity reaches at least 4 X the "nor-
mal" background at such locations. Various instruments, methods, and units
are used, depending mainly upon the scale of the investigation.
Wide-range reconnaissance surveys are carried out with small air-
planes or helicopters. Aeroradiometric prospecting provides a great amount
of information in a short time. Several conditions must be fulfilled, however,
to justify the high cost of such a survey: (a) moderate topographic relief to
enable a nearly constant surveying altitude of 500 ft above ground; (b) the
area to be surveyed should be free of glacial or alluvial deposits, * swamps,
or extensive vegetation; (c) the airborne detection instruments must be
sensitive enough to locate anomalies down to 0.1 f-lCijm 2. One or more
NaI(TI) crystals of considerable size (4 x 2 in., 9 X 3 in.) with several
photomultipliers connected in parallel are used.
Since the radiation intensity from the ground decreases rapidly with
increasing flight altitude (Fig. 17) readings must be reduced to the standard
500-ft elevation: the accumulation time for the pulses coming from the
detector (1 sec for 500 ft) is automatically changed according to the signals
from the radar altimeter. The cosmic-radiation effect is also automatically
subtracted. In aeroradiometric surveying the unit cps is used for integral
counting as well as for gamma spectrometry discrimination. 22
Various survey-line arrangements are employed depending on the
problem encountered: short loops normal to the strike direction of length-
extended rock units; grid flying over disjointed, patternlike outcrops; "con-

The terrestrial component of gamma radiation found at 500 ft above terrain comes,
due to self-absorption, from the radionuclides in the surficial 12 in. of earth materials
within a circular area about 2000 ft in diameter.
10. Radiometric Techniques 295

210' cps

1010' 1510'
feet

Fig. 17. Gamma-ray intensity vs flight altitude

over rocks with no radiometric anomalies
(after Hand").

tour-line flying" over flatlying sedimentary formations. In order to couple

radiometric data with ground locations a film is taken simultaneously with
a vertical camera. Flight speed is commonly around 150 mph.
Calibration includes periodic testing of the crystal-detector-assembly
response by a small 137Cs source aboard the aircraft. Care must be taken to
shield this source during actual measurements. Luminating dials on the
flight control panel should make a minimum contribution to the measured
radioactivity. The overall uncertainity associated with the measured radia-
tion value is about 10% in the aircraft position data of about 0.5% of
the distance flown. 23
Aeroradiometric survey data (radioactivity profiles of the flight lines)
are compiled either as a radioactivity unit map or as a contour map. The
flight line spacing is usually 1 mile. Due to different radionuclide content
of rock units their geologic mapping can be done in many cases by aerora-
diometric measurements. Pitkin et al. 24 quote impressive examples. Under
favorable conditions empiric calibration for concentration determinations
can be made: Moxham 25 gives the equation

CeU (ppm) = kIYa (27)

where ceU is the mean uranium equivalent content (see Section 5.2) of the
surface rocks in the 2000-ft circle below the aircraft, IYa is the measured net
aeroradioactivity (cps), and k a conversion factor.
More recently four-track registration has been employed :26 three
channels for spectral U, Th, and K discrimination and one track for the
integral count rate. Aeroradiometric surveys are often performed simul-
taneously with aeromagnetic measurements.
296 L. Rybach

4.2. Carborne Prospection

Interesting areas and anomalies found by aero radiometric prospecting

can be studied in more detail by car borne methods. Carborne radiometric
traversing is especially successful in areas having extensive road networks.
The equipment used is quite similar to the airborne devices. The detector
is mounted over the roof of the vehicle to screen maximum area and to
reduce the influence of the road pavement (a detector 20 in. above ground
"sees" mainly a circular area 20 ft in diameter; from 6 ft this diameter is
about 100 ft). Expandable masts can set the detector in a measuring position
up to 20 ft above ground. Continuous ratemeter recording enables detecting
of anomalies during and after survey. The vehicle speedometer is connected
to the recorder chart-drive mechanism. Every mile a mark is automatically
made on the record. The ratemeter time constant RC depends on the survey-
ing speed v and must be selected to fulfill

RC~~
- v (28)

where d is the minimum width of anomaly to be resolved. Surveying speeds

up to 40 mph can be used. An acoustic alarm signal indicates anomalies
above a certain level.
Interpretation of carborne measurements requires careful investigation
of different factors influencing the radiometric reading at a given locality,
of which the "geometry" (relative configuration of radioactive source-detec-
tor) is by far the most important one. The gamma-ray intensity Iy in any
given array is, according to Baranow,27

I = k Qe (29)
y a

where Q is the solid angle of the array, e is the content of gamma-emitting

radioisotopes* (e.g., in U equivalent, see Section 5.2), a = {lIe, the mass
absorption coefficient (see Section 3.1), and k a conversion factor relating Il'
(e.g., in cps) to e. For a small change in Il'

LlI = all' LI.Q + all' LIe + all' Lla = ~ LlQ + k ~ LIe _ kQe Aa.
y aQ ae aa a a a2
(30)

* The gamma-emitting radioisotopes are assumed to be homogeneously distributed.

10. Radiometric Techniques 297

Since a varies only moderately over different gamma energies,28 Lla = 0 and

Lily =
ly LlQ
Q + -c-
Iy LI
e = ly
( LlQ Lie )
---sF + -e- (31 )

Thus, variation in the solid angle Q can affect the reading seriously. If the
detector is wholly surrounded by radiating-rock material (e.g., in a tunnel),
Q = 4n. At a flat-rock surface Q = 2n. Under field conditions the geometry
may vary rapidly which is in many cases difficult to compensate for exactly.
Therefore field measurements have only a limited quantitative meaning.

4.3. Prospecting on Foot

Prospecting on foot is widely applied for small-scale surveys (to study
radioactive outcrops in detail, to verify the anomalies found by airborne or
car borne prospecting). Here also the influence of the geometry is important.
Measurements are made in profile points (e.g., for the investigation of
a length-extended ore body) or in grid points (in case of irregularly shaped
ore bodies). The detector is held at a distance of about 3 ft from the rock
surface. For a detailed "spot examination" a closer position and the use
of a lead collimator surrounding the detector head is convenient. The
measured values are, after subtraction of the background, plotted and an
isoradiation map is drawn on which the surface outlines of radioactive ore
bodies will show up.
The unit on such surface maps is usually ,uR/h. * Since most portable
instruments register in counts per second (cps) the converting calibration
factor must be determined by measuring the radiation intensity of a source
with known dose rate at a given distance from the detector. Commonly,
small Ra needles (encapsulated in 0.5 mm Pt) are used for this purpose.
The conversion factor k can be calculated according to

k = 8~'~2 q (,uR/h per cps) (32)

y

where q is the Ra source strength (,uCi) and ly is the net gamma-ray intensity
of the source (cps) measured at the distance r (cm).
Due to the influence of several factors which cannot be exactly enumer-
ated (geometry changes, background variation) it is usually difficult to
perform quantitative determinations in situ. Therefore, representative sam-
ples are taken in the field for laboratory investigations under standardized
conditions. One exception is the interpretation of borehole measurements:
here the constant geometry (4n) enables quantitative determinations. 29 ,3o

pR/h = microroentgens per hour.

298 L. Rybach

It must be emphasized at this point that because of the absorption of

gamma rays in rock (90% of the radiation observed above an outcrop comes
from the outermost 7 in. of the rock material) the methods mentioned
above are in general unable to detect deep-seated ore bodies. Special techni-
ques may help: radon sampling in soil and measuring its concentration;22
infrared photographs taken from an airplane, which may show surface
temperature anomalies caused by the heat produced in the radioactive body.

5. QUANTITATIVE DETERMINATION OF NATURAL

RADIOELEMENTS

5.1. Sample Preparation

For quantitative radiometric determinations the measurement of
gamma rays by scintillation detectors is most frequently used. Due to the
relatively high penetrating power of gamma rays self-absorption within a
relatively small sample can be kept fairly constant for materials of quite
different compositions. 28
Sample preparation commonly includes crushing, pulverizing, and
homogenizing the material to be studied; by filling a convenient sample
container with the rock or mineral powder the geometry can be standardized.
There are no limitations on the grain size (which is in most cases < 1 mm)
if it is roughly constant for the series of samples under study. Sometimes
solid specimens (rock cores with constant diameter, tektites with nearly
spherical shape) can be measured directly.31,32 Figure 18 shows different
sample-detector arrangements.

5.2. Integral Counting for U and K Determinations

Integral y counting has its limitations, but as a reconnaissance tool, its
speed and simplicity are particularly valuable. In many cases, for example,
if the main sample activity is due to its uranium content (ore-grade ma-
terials), a simple "total-y" measurement (see Section 3.3) gives excellent
results. With about 350 g of crushed and/or pulverized rock material filled
into an 8-oz, aluminum-sample-container canister (Fig. 18g) the measure-
ment can be performed under field conditions with a portable counter. The
specific activity a of the sample, which is proportional to the concentration
of the corresponding radioisotope in the sample, is given by
Iv- 10
a = G (cps/g or flR/h/g) (33)
10. Radiometric Techniques 299

H
~I--ti
",.

Q
/1 ,I,
solid sample
J

o 0 [] powder sample

Fig. 18. Various sample-detector arrangements for (a) rock core, (b) tektite,
(c) small sample (up to 10 g) in well-type detector, (d) small sample (few g),
(e, f) large sample (200-1000 g), (g) measurement with portable counter (see
Section 5.2). To reduce background shielding is used for the arrangements
(a)-(f).

where Iy is the measured "sample + background" activity, 10 is the back-

ground (both in cps or ,uR/h), and G the sample weight (g). For rapid
determinations the use of a calibration curve (a vs concentration) (Fig. 19a)
is convenient; this can be drawn by measuring and plotting the a values
for a series of samples (in the same type of container) with known concen-
trations* of the radioisotope in question.
For rapid determinations in U ores in radioactive equilibrium (see
Section 2.1) this method is best suited. U concentrations > 200 ppm can
be determined without shielding. The precision depends mainly on the
fluctuation of the ratemeter indicator (see Section 3.3) and is usually better
than 10%. A measurement takes about 10 sec; an average reading is
obtained by watching the fluctuations of the meter needle (the indication
should be in the upper one-third of the meter scale; field ratemeter instru-
ments have a range-selector switch for this purpose). Results are expressed
in ppm eU, i.e., in "uranium equivalent" which is equal to the true uranium
value only if there is no other significant radiation from Th and K present

* Powder standards containing known amounts of U and Th (in radioactive equilibrium)

can be purchased from the USAEC New Brunswick Laboratory. Dry, pure KCl or
KCO s is used as K standard.
300 L. Rybach

10.0 cPSfg

2. rcpm=,7""g--------------,

1.0

1.
0.1

200 10010 10 20 30 40

Fig. 19. Working curve for uranium determinations with a portable counter (a), po
tassium calibration curve for determinations with a laboratory scaler (b).

in the sample. This, of course, affects the accuracy. In Table IV a few

examples are given.
With a better geometry (Fig. 18f, sample weight ca. 350 g) and a labo-
ratory scaler (Fig. 13) instead of a portable ratemeter and a shielding of
about 2 in. of lead surrounding the sample and detector the sensitivity can
be improved and extended down to 0.1 ppm eU. In this low concentration

Table IV
Comparison of Equivalent Uranium (eU) with Effective Uranium Concentrations

Portable field Laboratory

Rock type ppmU Reference
ratemeter, ppm eU scaler, ppm e U

Sericite gneiss with 750 700 730 55

U mineralization 1500 1450 1600
(Verbier, Switzerland)
Triassic dolomite with 1100 1030 1010 56
U mineralization 500 460 450
(Ferrera, Switzerland)
Rotondo granite 27 10.2 57
(Switzerland)
Mont Blanc granite 28 12.5 54
(Switzerland)
10. Radiometric Techniques 301

region, however, the influence of K and Th in the sample is serious and

can be resolved only by gamma-ray spectrometry (see Section 5.3).
Materials which are mined for potassium (usually a mixture of KCI
and NaCl) contain only minute amounts of U and Th. K contents in such
materials can be determined by integral counting (counting time in the order
of 1~0 min) with a precision of 2-10%. The statistical error (la. of a deter-
mination with a laboratory scaler is

(la. ~ vi NT + tAo (cpmjg) (34)

Gt

where NT is the number of counts of the sample (plus background) registered

during the time t (min) and Ao is the background count rate (cpm). a is here

a = G1 (NT
-t- - Ao ) (cpmjg) (35)

5.3. Gamma-Ray Spectrometry for Simultaneous U,

Th, and K Determinations

If the sample contains more than one radioelement gamma-ray spectrom-

etry can resolve the individual radiations. For U and Th the most char-
acteristic y rays are emitted by the daughter isotopes 214Bi and 208TI, res-
pectively (see Table II). The K content can be determined by measuring the
y rays from 4K (for isotopic composition and y-ray abundance see Table I
and Fig. 1). The measurements are most frequently carried out with the
arrangement shown in Fig. 18f, using a 3 X 3-in. Nal(TI) crystal. The y-ray
spectrum obtained by these means (upper solid line in Fig. 20) displays
characteristic photopeaks at 1.46 MeV (4K), 1.76 MeV (214Bi of the 238U
series), and 2.62 MeV (208Tl of the 232Th series).
A "window" is opened at each of these peaks. The window is centered
at the middle of the peak, and has a width 2L1E, where L1E is about 0.05E:
e.g., for potassium L1E = 70 KeV, and the window extends from 1.39 to
1.53 MeV. In multichannel pulse-height analysis these energies correspond
to channel numbers; for a given gain setting the energy-calibration curve
can be obtained by identifying the photopeaks in the different channels.
For peak identification see the U, Th, and K spectra in Crouthamel10 and
Heath. 13
The unknown U, Th, and K concentrations in the sample can be calcu-
lated after evaluating the calibration constants. This involves the measure-
ment of standards with known U, Th, and K contents under the same
302 L. Rybach

2.0 Mev 3D

_ _ SOfr4)I"
______ . U
_ ._._
10 . +-~~-'-l-+t-t- _ __ Th
K

1.0 +-------~_T_;I__~Hr____l

0 .1 -t------+-1++--*~+_-_l

20 40

Fig. 20. Gamma-ray spectrum of a rock sample

(solid line), contributions from K ( ... ), U
(-), and Th (-.-). Windows 1, 2, and 3
(see text).

experimental conditions as for the unknown samples. If one assumes that

the effect of K is negligible (this is true in the case of accessory minerals like
zircon or sphene which contain almost no K). The following symbols may
be used
Cu = U content of the sample (ppm)
CTh = Th content of the sample (ppm)

cu' = U content of the U standard (ppm)

CTh = Th content of the U standard (= 0 in common U standards)
Cu = U content of the Th standard (ppm)
drh = Th content of the Th standard (ppm)
al = specific activity (cpm/g) of the sample in window 1 (at 2.62 MeV)
a2 = specific activity of the sample in window 2 (at 1.76 MeV)
aI' = specific activity of the U standard in window 1
a2' = specific activity of the U standard in window 2
al" = specific activity of the Th standard in window
a 2" = specific activity of the Th standard in window 2
10. Radiometric Techniques 303

In this list the first two items are unknown, and the rest are known or can
be measured. In the determination of the a's the background count rate in
the corresponding window must be taken in account (see Table V).
For the sample one can write two equations with two unknowns

a1 = !tcu + hCTh (36)

a2 = ,hCu + hCTh
which means that both the U and Th contents contribute to the activities
measured in windows I and 2 (the f's are the calibration factors to be
determined). Similarly for the U standard we have

(37)

Table V
Digital Gamma-Ray-Spectrum Read Out (Counts in Channels 65-128), Calculation
of al and a,a

1751 1728 1670 1610 1521 1462 1369 1392

1434 1683 2031 2128 2027 1790 i501 1232
1012 862 750 688 614 617 612 577
638 671 723 769 785 841 810 760
696 592 539 499 463 422 391 419
397 409 374 393 449 559 594 712
709 660 524 483 383 323 240 233
229 200 198 227 214 197 199 168

( sum of cou~ts in window 1

a1 = - background in window 1) =
sample weight time

= - 1 (5840.0
- - - 1.50 ) = 1.808 cpm/g
1.9 1183.0

a, = - 1 (15,218.0 - 3.10 ) = 5.138 cpm/g

1.9 1183.0

a Spectrum of a mineral concentrate (1.9 g sphene) recorded by a Nuclear Data multichannel

analyzer; accumulation time 1183 min. Window 1 for Th, channels 107-117; window 2 for U,
channels 72-80. Background in window I, 1.50cpm; in window 2, 3.IOcpm.
304 L. Rybach

from which the calibration factors!t and fa can be calculated:

a'
~=_1_
Jl cu'
(38)
fa = c;,
a'

in cpm/g/ppm. For the Th standard one can write

a{' = };cu + hCTh

(39)
a;' = facTJ + i4CTh
leading to
a{' - !tCl]
C~h
ar- faclJ
(40)

C"
Th

From Eq. (36) it follows that

(41)

and

from which

(43)

where

(44)

k2 and k4 are also called the "stripping constants." In practical work CTh
is determined first; the result is then used to evaluate Cu.
10. Radiometric Techniques 305

The application of this method is demonstrated by U and Th deter-

minations in accessory minerals separated from the Silvretta gneiss and
from the Giuv syenite (Switzerland). The mineral concentrates as well as
the standards were measured in small aluminum sample containers with a
thin Mylar-foil bottom. Sample and standard weights were about I g. The
sample containers (diameter, I in.) were centered on the top of a 3 X 3-in.
NaI crystal (Fig. 18d). Counting times ranged from 5 to 24 h. The follow-
ing calibration equations and factors have been found:

CTh = 517(at - O.0291a2 )

(45)
Cu = 115(a2 - O.00015cTh)
The results are given in Table VI, together with mass-spectrometric U deter-
minations. The precision of the results is in the order of 5% as evaluated
from repeated measurements. Taking the mass-spectrometric results as
correct the accuracy is better than 5%.

Table VI
Nondestructive V and Th Determination in Accessory Mineral Concentrates by
Gamma-Ray SpectrometryG

Mineral Rock U, ppm Th, ppm U, ppm b

Zircon, SiIyretta gneiss 550 160 525

53-75 f1, fraction (Switzerland)
Zircon, SiIYretta gneiss 590 175 571
42-53 f1, fraction (Switzerland)
Zircon, SiIvretta gneiss 600 190 615
30-42 f1, fraction (Switzerland)
Sphene Giuy syenite 575 850
(Switzerland)
Epidote-allanite Giuy syenite 315 1150
(Switzerland)
Apatite Giuy syenite 65 130
(Switzerland)
Zircon Giuy syenite 2800 2370
(Switzerland)

G Sample weight approximately 1 g.

b Mass-spectrometric determinations.5.
306 l. Rybach

For simultaneous potassium determinations a very similar procedure

(solving three equations) gives the constants in the potassium equation:

(46)

where CK is the K content of the sample (%) and '13 is the specific activity
in window 3 (centered at 1.46 MeV).
For powdered rock samples (ca. 350 g) measured in the arrangement e
of Fig. 18 the following factors have been found:

CTh = l120(a1 - 0.02la2)

Cu = 370(a2 - 0.00048cTh) (47)
CK = 31.8 [a3 - (2.2 X IO-3 CU + 5.6 X IO-4 CTh )]

It should be mentioned that the sample containers should always be filled

to the top; a 2-mm deviation (incomplete filling) causes a 4% error in
concentration. Density variations have a much slighter effect.33
Table VII indicates the general accuracy and precision of this method.
The precisions indicated were calculated from repeated measurements.
Counting times range from 40 min for an average granite sample to 24 h
for basic rocks.
The method described above does not require too much calculation
(a desk computer will do in most cases). Its major disadvantage is that it
utilizes only a fraction 25%) of the information held by the recorded

Table VII
Accuracy and Precision of Gamma-Spectrometric Results a

Rock K,% U, ppm Th, ppm Method

Standard diabase W-l 0.53 0.52 2.2-2.4 Recommended values'

0.52 0.53 2.16-2.24 y spectrometry
Granite G-2 3.710.05 2.10.3 25.70.5 y spectrometry
3.740.01 Flame photometry
2.160.07 24.10.1 Neutron activation
Basalt BCR-l l.390.03 1.60.1 6.10.4 y spectrometry
1.410.01 Flame photometry
1.81 0.14 6.000.19 Neutron activation

a From Refs. 33 and 60.

10. Radiometric Techniques 307

"multichannel" spectrum which therefore must be measured with a relatively

high statistical accuracy for long counting times.
In recent years the availability of computers facilitates the analysis of
gamma-ray spectra and this analysis can be extended to the whole spectrum.
The data obtained by multichannel analyzers offer an excellent opportunity
for the application of least-squares methods. Salmon34 used the unweighted
least-squares method for the unscrambling of relatively simple gamma-ray
spectra of monoenergetic, artificial radioisotopes. In our case, however,
there is in addition to the complex components another complication pre-
sent: the spectra extend over several orders of magnitude due to scattering
effects (higher count rates in the low-energy channels). This fact suggests
the introduction of weights.
Let us define the following symbols:
n = number of channels
ai = count rate of the U standard in channel i (cpm)
bi = count rate of the Th standard in channel i (cpm)
Ci = count rate of the K standard in channel i (cpm)
di = count rate of the sample in channel i (cpm)
Vi = :1 Tflndom error

Wi = weight of d i
mu = U in the sample (g)
mTh = Th in the sample (g)
mK = K in the sample (g)
Mu = U in the U standard (g)
MTh = Th in the Th standard (g)
MK = K in the K standard (g)
x = the ratio mu/Mu
Y = the ratio mTh/ M Th
Z = the ratio mK/MK
ax, ay, az = the corresponding standard deviations
Qx, Qy, Qz = "weight factors"
G = sample weight (g)

Due to the linear superposition of the U, Th, and K contributions one can
write for each channel (error equations)

di = aix + biy + CiZ + Vi' weight Wi (48)

Having n such equations for only three unknowns the application of the
weighted least-squares method means the determination of the "best values"
308 L. Rybach

for x, y, and z in the sense of

n n
~ WiV;2 = ~ wild; - (aix + biy + CiZ)]2 = minimum (49)
>=1 >=1

It is easy to show that for multichannel operation (constant counting time

for each channel)
1
Wi =7-, (50)

where with the normal equations (Ref. 35, p. 107)

x
~ ai 2
/.-J - -
+ Y ..,
~ aibi + f aici _
-- z .., - - - ~
.., aj
,=1 d; i=1 di i=1 di ;=1

(51)

~ aici
X .., - -
+ y ~.., -biCj- + z ".., -Ci-
2 _
-
~
.., C;
i=1 d i ;=1 d; d; i=1

The matrix solution which is especially suitable for computer data processing
gives the values x, y, and z with which the "best fit" can be computed:
d/ = aix + b;y + CiZ (52)
With
v= [_1_t
n- 3 >=1
(d/ -
di
di )2 ] (53)

the standard deviations of the unknowns are

Gx = vQ!/2
Gy = VQ~/2 (54)
Gz = vQ}/2

The Q's can be calculated (Ref. 35, p. 95) from the matrix elements (normal
equations ).
Finally, the U, Th, and K concentrations in the sample are given by
MU
ppm U = (x Gx ) G x 106

ppm Th = (y Gy )
At
Jh X10 6 (55)

MK
% K = (z G z) G X 102
10. Radiometric Techniques 309

Experimental precautions must be taken to fulfill the assumptions necessary

for this method: (a) the standard spectra should be measured to high
statistical accuracy; (b) the background is constant; (c) there is no drift.
The latter can be attained by utilizing a gain stabilizer. 36
The measured spectra can be obtained in digital form on punched-tape
readout which serves as the input for the computer. With a computer
program the best fit and the U, Th, and K concentrations are calculated
from the spectra using the channels above 0.35 MeV. Figure 21 gives an
example of the best fit calculated by the least-squares method. The major
advantage of this method is that counting times can be reduced by a factor
of five to seven.
For the analysis of solid-rock core samples an automatic sample changer
has been constructed. 19 Rapid analysis can be obtained by combining this
device with computer-data interpretation. 31
Several successful attempts have been made to develop portable y-ray
spectrometers for in situ U, Th, and K determinations. 37 - 39 The weight of

1000~ ____________ ~

1.0 2.0 Mev 3D

count%hannel

100+---------{>-+-,I~ _ _ _ _ _--I

.'

10+-_ _ _ _ _ _ _ _ _ ~~~--I

1.0-L-_-'--_-'--_-'--_-'--_-'--_--'-~
20 40 60 80 100 120 channel

Fig. 21. Measured gamma-ray spectrum and

computed best fit for a Mont Blanc granite sample
(3.5 ppm U, 13 ppm Th, 2.5% K). Dots: counts
in 1000 sec. Solid line: best fit. Sample weight:
595 g (from Rybach et al. 54 ).
310 l. Rybach

the equipment is in the order of 20 kg. The major benefit of determining U,

Th, and K in the field is the greater volume sampled (approximately 300 kg,
which is far more representative than a 300-g laboratory sample). Measure-
ments have to be performed in 2n geometry (flat-rock outcrop). Foran
average granite a 5-min counting operation yields a precision of about
10% for K and about 15% for U and Th.
Measuring the natural radiation from U, Th, and K, radiometric
methods give satisfactory results down to about 0.2 ppm U and Th and to
about 0.1 ppm K (for 0.09-ppm Th and 0.05-ppm U precision is in the order
of 50%.40 For lower concentrations, as are common only in ultrabasic
rocks, fission track counting (see Section 5.6) or neutron activation analysis
are better suited. 41 ,42

5.4. Beta-Gamma Counting for Equilibrium Studies

The foregoing methods assume radioactive equilibrium (see Section 2.1)

between the parent isotopes 238U and 232Th (which are to be determined)
and their y-emitting daughter isotopes 214Bi and 208Tl (which can be detec-
ted). In most cases this assumption has been found to be valid (for unal-
tered, fresh rocks). In some cases, however (in altered rocks, leached U ores,
miocene coals containing uranium, deep-sea sediments), serious errors can
arise. To avoid tedious chemical determinations a simple radiometric
method 43 can be used even in the case of completely disturbed radioactive
eq uili bri um.
The method is based on the fact that disequilibrium affects the {J and y
activity of a given sample differently. Thus, for example, the U content
found by {J and y measurements will be different, depending on the state of
equilibrium. For the y measurement the arrangement f in Fig. 18 (open
annular sample container) is used; the {J's are counted with an end-window
GM detector placed a few mm above the sample-powder surface. An extra
calibration curve has to be drawn for the beta measurements by plotting
the net {J cpm against U concentration of the standards. If cUll is the U
content as determined by {J measurements and CUy is the value found by y
counting, the true U content Cu can be calculated according to
Cu = 2.30 cUll - 1.30 CUy (56)

The ratio q = CUy/cu is a measure of the state of equilibrium: q = 1 for

equilibrium. Table VIII gives examples.
A special gamma-spectrometric method 44 enables determination of 235U
by measuring its 0.184-MeV photopeak. Since the ratio 235U /238U = 1/137
10. Radiometric Techniques 311

Table VIII
Radioactive Equilibrium and Disequilibrium

Sample CUy, ppm cup !f Cu cux a

Triassic dolomite, 445 435 0.95 460 440

with U mineralization 895 875 0.95 920 940
(Ferrera, Switzerland)
Miocene coal 310 76 0.12 615
(SellenbUren, Switzerland)

a X-ray-fluorescence determinations from Ref. 56.

is constant in nature 238U abundances can be determined even in case of

completely disturbed equilibrium. Absorption effects cause difficulties, how-
ever, in this low-energy region.

5.5. Alpha Counting, Alpha Spectrometry

Most naturally occurring radioisotopes are a emitters. Despite the short

range of the a particles (see Section 2.3) their detection possesses certain
advantages as compared to fJ and y measurements, the major advantage
being the greater sensitivity due to the low background in a counting.
In total-a counting the sample is measured as a "thick source" (the
thickness of the source is much greater than the maximum range of a
particles, which is about 25-50 fl in solid samples). The measured a activity
is expressed in units of a/cm 2 /h, taking the surface area into consideration.
The number of a particles per unit area and time (N) is proportional to
the uranium and/or thorium content. For a series of zircons Holland 45
found the relationship

Cu (ppm) = 0.463N (a/cm 2/h) (57)

In general two detector types have been applied for total a counting: ZnS
screens optically coupled to a photomultiplier tube and proportional gas-
flow counters with 2n geometry. Solid samples can be measured directly
by placing a polished section on a thin ZnS foil (e.g., ATP-3 alpha strip
with a 4.5-mg/cm 2 ZnS emulsion on Mylar) overlaying a photomultiplier.
For an irregular sample-surface shape a matching piece of foil is cut; its
weight is a measure of the sample surface area. For gas-flow GM counters,
312 L. Rybach

Table IX
Total Alpba Counts vs Radioelement Content in Various Sedimentary Rocks a

Contents, ppm (by y-ray spectrometry)

Net a counts
Rock type
per em", h
U Th

Phosphate 108.0 240.0 4.0

Shale 28.3 80.0 5.0
Limestone 1.3 1.7 1.6
Bauxite 36.7 6.4 96.0

a From Ref. 62.

powdered samples (grain size <100 mesh) are filled into sample-holder
planchets (typical dimensions, 2 in. in diameter, 1/8 in. deep, taking up
about 1.5 g of powder). Background is very low for both detector types
(about 0.15 a/cm 2/h); sensitivity is mainly a function of the ratio of the
sample-to-detector area.
The counting time per sample ranges from 8 to 48 h in order to attain
a reasonable statistical error. The major advantages of the method are the
simple sample preparation and the small sample size required. Though less
quantitative than y-ray methods it gives informative results by simple means
(see Table IX).
Since the introduction of multichannel analyzers, total-a counting has
been frequently replaced by spectrometric techniques. After chemically sep-
arating uranium from thorium, single isotopes in the 238U and 232Th decay
series can be detected (Figs. 22 and 12). Sample preparation is a rather

4 cpm

oc. particle energy

Fig. 22. Alpha spectrum of uranium isotopes.

Resolution is 1.0% FWHM for the "SOU peak.
Counting time 24 h (after Rosholt et al.").
10. Radiometric Techniques 313

Table X
Uranium and Thorium Abundances, Isotopic Ratios Determined by Thin Source
Alpha Spectrometrya

Sample U, ppm Th, ppm 234U/238U 23OTh/ 232Th

Roanoke River sediment, 4.120.15 13.370.74 0.980.03 1.010.05

2-20 '" fraction
Mississippi River sediment, 3.290.12 7.930.57 0.940.13 1.190.06
2-20 '" fraction
Red River sediment, 2.730.08 7.700.39 0.940.03 0.960.02
2-20", fraction

a From Ref. 48.

elaborate chemical procedure (U and Th extraction) followed by electro-

deposition to obtain a "thin source." Chemical extraction includes de-
composing of 5-20 g of sample material in concentrated acids, sodium
carbonate fusion of residues, anion exchange, extraction, purification, and
determination of the chemical yield (for details see Ref. 46). For electro-
plating l-in.-diameter discs (made of stainless steel, titanium, wolfram, or
platinum) are used. Electrodeposition is carried out at 6-8 V and 1-2 A for
20-40 min (for details see Ref. 47).
During measurement the sample-holder discs are placed in a vacuum
chamber (approximately 20 f1, pressure) at a distance of 1-2 mm from a
silicon-barrier semiconductor detector with a sensitive area of about 3 cm 2
(see Section 3.2). The spectra are recorded in digital form with a multi-
channel analyzer. For isotopic-abundance determination the numbers of
counts/channel within the corresponding peak are summed up. For further
details see Scott,48 from which the figures in Table X are taken.

5.6. Fission Track Counting

Besides its application to age dating this method is especially powerful
for the determination of the abundance and distribution of uranium in single
mineral phases 63 or in whole rocks. 64 Polished thin sections or pellets pressed
from sample powder and cellulose binder are covered with a piece of Lexan*
(polycarbonate plastic). Good mechanical contact is essential between

* Available in sheets (10", to 100 '" thick) from Chemical Dept., General Electric Co.,
Pittsfield, Massachusetts. USA. The uranium content is very low ( < 10-5 ppm).
314 L Rybach

sample and Lexan foil. Standards (glasses with known uranium content or
pellets prepared from powder standards), covered the same way, and sam-
ples are mounted together into sandwiches which are irradiated in the ther-
mal column of a nuclear reactor.
Thermal neutron doses of the order of 1015 n/cm 2 are applied for sam-
ples in the ppm range, lower uranium concentrations requiring correspond-
ingly higher doses. The homogeneity of the neutron flux must be monitored.
Irradiation induces fission of some 235U nuclei in both sample and standard,
and the massive, charged fission fragments strike the attached Lexan foils
leaving tracks approximately 10 fl long.
After etching of the foils, these tracks can be seen with an ordinary optical
microscope (Fig. 23). By means of location marks, the occurrence of fission
tracks in the Lexan can be correlated with the mineral phases in the underly-
ing thin section. The density (tracks per mm2) is a measure of the uranium
concentration. Track counting is performed visually, under the microscope,65
or with a discharge counter 66. The uranium content can be calculated from
(di
Cu = Cs (diS (58)

where Cs is the uranium content of the standard and (di and (diS are the in-
duced track densities in the Lexan foils covering sample and standard,
respectively. Equation (58) assumes the same fission fragment range in sam-
ple and standard.
The major advantage of this method is, besides its simplicity, its great
sensitivity; determinations in the ppb range can easily be performed. 67

Fig. 23. Fission tracks in a Lexan foil ( x 522).

Etching conditions: 8 min in 6 N NaOH at 70C.
10. Radiometric Techniques 315

6. RADIOMETRIC AGE DETERMINATIONS

In the early days of geochronology age dating was carried out on radio-
active accessory minerals by the lead-alpha method 49 according to

kCPb
t=-- (59)
a

where t is the age of the accessoric mineral in 106 yr, Cpb is the lead content
determined by optical spectroscopy, a is the alpha radioactivity, and k is
a constant depending on the Th-U ratio in the mineral to be dated. Thus, a
radiometric measurement was necessary to evaluate the a value of a given
mineral sample and, in turn, its age. Because of several serious limitations
(one of which is the influence of nonradiogenic lead) this simple and rapid
method was successively replaced by more sophisticated methods such as
238U /206Pb, 235U /207Pb, 206Pb /204Pb, 87Rb/ 87 Sr, and 40Kj40 A dating. 50 These
methods require, however, mass-spectrometric determinations. Though the
term "radiometric age" appears everywhere in the literature one should be
aware of the fact that most probably no radiometric technique was applied.
Radiometric methods are used only in very special cases. For instance
gamma-ray spectrometry makes the dating of very young volcanic rocks
possible (e.g., historic eruptions of Vesuvius;51 from 234U/238U ratios de-
termined by alpha spectrometry (see Section 5.5) ages in carbonates (e.g.,
corals) younger than 106 yr can be calculated. 52

REFERENCES

1. Miesch, A. T., Methods of computation for estimating geochemical abundance,

U.S. Geol. Survey Prof. Paper 574-B, 1967.
2. Adams, J. A. S., Osmond, J. K., and Rogers, J. J. W., The geochemistry of thorium
and uranium, in Physics and Chemistry of the Earth, Pergamon Press, New York,
1959, Vol. 3, pp. 298-348.
3. Yokoyama, Y., Tobailem, J., Grjebine, T., and Labeyrie, J., Determination de la
vitesse de sedimentation oceanique par une methode non destructive de spCctro-
metrie gamma, Geochim. Cosmochim. Acta 32, 347 (1968).
4. Birch, F., Heat from radioactivity, in Nuclear Geology, H. Faul, ed., John Wiley
& Sons, New York, 1954, pp. 148-174.
5. Wollenberg, H. A., and Smith, A. R., Radiogeologic studies in the central part of
the Sierra Nevada batholith, California, J. Geophys. Res. 73, 1481 (1968).
6. May, H., and Marinelli, L. D., Cosmic-ray contribution to the background of low-
level scintillation spectrometers, in The Natural Radiation Environment, Univ. of
Chicago Press, Chicago, 1964, pp. 463-480.
316 L. Rybach

7. Kraner, H. W., Schroeder, G. L., and Evans, R. D., Measurements of the effects of
atmospheric variables on Radon-222 flux and soil-gas concentrations, in The Natural
Radiation Environment, Univ. of Chicago Press, Chicago, 1964, pp. 191-215.
8. Pensko, E., Wardaszko, T., and Wochna, P., Natural atmospheric radioactivity and
its dependence on some geophysical factors, Atompraxis 14, 255 (1968).
9. Gustafson, P. F., and Brar, S. S., Measurement of y-emitting radionuclides in soil
and calculation of the dose arising therefrom, in The Natural Radiation Environment,
Univ. of Chicago Press, Chicago, 1964, pp. 499-512.
10. Crouthamel, C. E., Applied Gamma-Ray Spectrometry, Pergamon Press, Oxford,
1960, 443 pp.
11. Taylor, J. J., Application of gamma-ray buildup data to shield design, WAPD-
RM-217, 1954.
12. Taylor, D., The Measurement of Radio Isotopes, Methuen & Co., London, 1957,
132 pp.
13. Heath, R. L., Scintillation Spectrometry Gamma-Ray Spectrum Catalogue, Phillips
Petroleum Co., 1964.
14. Hill, C. R., Osborne, R. V., and Mayneord, W. V., Studies of a radioactivity in
relation to man, in The Natural Radiation Environment, Univ. of Chicago Press,
Chicago, 1964, pp. 395-405.
15. Bowie, S. H. U., Portable X-ray fluorescence analysers in the mining industry, Mining
Magazine 106, 1 (1968).
16. Yagoda, H., Radioactive Measurements with Nuclear Emulsions, John Wiley & Sons,
New York, 1949, 356 pp.
17. Ragland, P. C., Autoradiographic investigations of naturally occurring materials, in
The Natural Radiation Environment, Univ. of Chicago Press, Chicago, 1964, pp. 129-151.
18. Bowie, S. H. U., Autoradiography, in Physical Methods in Determinative Mineralogy,
J. Zussman, ed., Academic Press, London, 1967, pp. 467-473.
19. Adams, J. A. S., Laboratory y-ray spectrometer for geochemical studies, in The
Natural Radiation Environment, Univ. of Chicago Press, Chicago, 1964, pp. 485-497.
20. Wollehberg, H. A., and Smith, A. R., Radioactivity of cement raw materials, Sym-
posium on Geology of Cement Raw Materials, Indiana University, 1966, pp. 129-147.
21. Sonntag, C., Extremely low-level scintillation spectrometer. In: Radioactive Dating
and Methods of low-level Counting, IAEA Vienna, 1967, pp. 675-686.
22. Bowie, S. H. U., and Bisby, H., Methods of detecting and assessing low grade uranium
deposits, J. Brit. Nucl. Ener. Soc. 14, 169 (1967).
23. Hand, J. E., Instrumentation for aerial surveys of terrestrial y radiation, in The
Natural Radiation Environment, University of Chicago Press, Chicago, 1964, pp. 687-
704.
24. Pitkin, J. A., Neuschel, S. K., and Bates, R. G., Aeroradioactivity surveys and geo-
logic mapping, in The Natural Radiation Environment, Univ. of Chicago Press,
Chicago, 1964, pp. 723-736.
25. Moxham, R. M., Some aerial observations on the terrestrial component of environ-
mental y radiation, in The Natural Radiation Environment, Univ. of Chicago Press,
Chicago, 1964, pp. 737-746.
26. Morris, D. B., personal communication, 1969.
27. Baranow, V. I., Aeroradiometric prospecting for uranium and thorium deposits and
the interpretation of gamma anomalies, Proc. Int. Con! on Peaceful Uses of Atomic
Energy, Geneva, 1956, Vol. 6, pp. 740-743.
10. Radiometric Techniques 317

28. Davisson, C. M., and Evans, R. D., Gamma-ray absorption coefficients, Rev. Mod.
Phys. 24, 79 (1952).
29. Scott, J. H., Dodd, P. H., Droullard, R. F., and Mudra, P. J., Quantitative inter-
pretation of gamma-ray logs, Geophysics 26, 182 (1961).
30. Scott, J. H., Computer analysis of gamma-ray logs, Geophysics 28, 457 (1963).
31. Rybach, L., and Adams, J. A. S., Automatic analysis of the elements U, Th and K
in solid rock samples by nondestructive gamma spectrometry, Proc. Int. Anal. Chem.
Con/. Budapest, 1966, Vol. 2, pp. 323-330.
32. Rybach, L., and Adams, J. A. S., The radioactivity of the Ivory Coast tektites and
the formation of the Bosumtwi Crater (Ghana), Geochim. Cosmochim. Acta 33,
1101 (1969).
33. Heier, K. S., and Rogers, J. J. W., Radiometric determination of thorium, uranium
and potassium in basalts and in two magmatic differentiation series, Geochim. Cosmo-
chim. Acta 27, 137 (1963).
34. Salmon, L., Analysis of y-ray scintillation spectra by the method of least squares,
AERE-Rept. 3640, 1961.
35. Grossman, W., Grundztige der Ausgleichsrechnung, Springer Verlag, Gattingen,
1961, 406 p.
36. Comunetti, A. M., A new gain stabilizing system for scintillation spectrometry, Nucl.
Instr. Meth. 37, 125 (1965).
37. Adams, J. A. S., and Fryer, G. E., Portable y-ray spectrometer for field determina-
tion of thorium, uranium and potassium, in The Natural Radiation Environment,
Univ. of Chicago Press, Chicago, 1964, pp. 577-596.
38. Lovborg, L., A portable y-spectrometer for field use, Danish Atomic Energy Com-
mission, Risa Rept. No. 168, 1967.
39. Doig, R., The natural gamma-ray flux: in-situ analysis, Geophysics 33, 311 (1968).
40. Heier, K. S., Uranium, thorium and potassium in ec10gitic rocks, Geochim. Cosmo-
chim. Acta 27, 849 (1963).
41. Lovering, J. F., and Morgan, J. W., Uranium and thorium abundances in possible
upper mantle materials, Nature 197, 138 (1963).
42. Nagasawa, H., and Wakita, H., Neutron activation analysis of potassium in ultra-
basic rocks, Geochem. J. 1, 149 (1967).
43. Eichholz, G., Hilborn, I., and McMahon, c., The determination of uranium and
thorium in ores, Can. J. Phys. 31, 613 (1953).
44. Bunker, C. M., and Bush, C. A., Uranium, thorium and radium analyses by gamma-
ray spectrometry (0.184-0.352 million electron volts), U.S. Geol. Survey Prof. Paper
550-B, 1966, pp. 176-181.
45. Holland, H. D., Radiation damage and its use in age determination, in Nuclear
Geology, H. Faul, ed., John Wiley & Sons, New York, 1954, pp. 175-180.
46. Rosholt, J. N., Doe, B. R., and Tatsumoto, M., Evolution of the isotopic composi-
tion of uranium and thorium in soil profiles, Bull. Geol. Soc. Amer. 77, 987 (1966).
47. Goldberg, E. D., and Koide, M., Geochronological studies of deep sea sediments by
the ionium/thorium method, Geochim. Cosmochim. Acta 26, 417 (1962).
48. Scott, M. R., Thorium and uranium concentrations and isotope ratios in river sedi-
ments, Earth Plan. Sci. Letters 4, 245 (1968).
49. Larsen, E. S., Keevil, N. B., and Harrison, H. c., Method for determining the age of
igneous rocks, using the accessory minerals, Bull. Geol. Soc. Amer. 63, 1045 (1952).
50. Hamilton, E. I., Applied Geochronology Academic Press, London, 1965, 267 pp.
318 L. Rybach

51. Vitozzi, P., and Rapolla, A., Radioactivity of vesuvian effusive products former to
1631, Bull. Volc.32, 136 (1969).
52. Thurber, D. L., Broecker, W. S., Blanchard, R. L., and Potratz, H. A., Uranium
series ages of Pacific atoll coral, Science 149, 55 (1965).
53. Cherry, R. D., Richardson, K. A., and Adams, J. A. S., Unidentified excess alpha-
activity in the 4.4-MeV region in natural thorium samples, Nature 202, 639 (1964).
54. Rybach, L., von Raumer, J., and Adams, J. A. S., A gamma spectrometric study of
Mont Blanc granite samples, Pure and Applied Geophysics 63, 153 (1966).
55. Hiigi, Th., Koppel, V., De Quervain, F., and Rickenbach, E., Die Uranvererzungen
bei Iserables (Wallis). Beitr. Geol. Schweiz, Geotechn. Serie, 1967, Lfg. 42.
56. Dietrich, V., Huonder, N., and Rybach, L., Uranvererzungen im Druckstollen Fer-
rera-Val Niemet. Beitr. Geol. Schweiz, Geotechn. Serie, 1967, Lfg. 44.
57. Rybach, L., Hafner, S., and Weibel, M., Die Verteilung von U-Th, Na, K und Ca
im Rotondogranit, Schweiz. Min. Petro Mitt. 42, 307 (1962).
58. Adams, J. A. S., and Lowder, W. M., eds., The Natural Radiation Environment,
Univ. Chicago Press, Chicago, 1964, p. 1069.
59. Grauert, B., and Arnold, A., Deutung diskordanter Zirkonalter der Silvrettadecke
und des Gotthardmassivs (Schweizer Alpen), Contr. Mineral. and Petrol. 20, 34 (1968).
60. Morgan, J. W., and Heier, K. S., Uranium, thorium and potassium in six U.S.G.S.
standard rocks, Earth Plan. Sci. Letters 1, 158 (1966).
61. Stevens, R. E., and others, Second report on a co-operative investigation of two
silicate rocks, U.S. Geol. Survey Bull. 1113, 1 (1960).
62. Adams, J. A. S., Richardson, J. E., and Templeton, C. C., Determinations of thorium
and uranium in sedimentary rocks by two independent methods, Geochim. Cosmo-
chim. Acta 13, 270 (1958).
63. Price, P. B. and Walker, R. M., A simple method of measurering low uranium con-
centrations in natural crystals, App. Phys. Let. 2, 23 (1963).
64. Fisher, D. E., Homogenized fission track analysis of uranium: A modification for
whole rock geological samples, Anal. Chem. 42, 414 (1970).
65. Kleeman, J. D. and Lovering, J. F., Uranium distribution in rocks by fission-track
registration in Lexan plastic, Science 156, 512 (1967).
66. Bertine, K. K., Chan, L. H., and Turekian, K. K., Uranium determinations in deep-
sea sediments and natural waters using fission tracks, Geochim. Cosmochim. Acta 34,
641 (1970).
67. Fisher, D. E., Homogenized fission track analysis of uranium in some ultramafic
rocks of known potassium content, Geochim. Cosmochim. Acta 34, 630 (1970).
Chapter 11

NUCLEAR ACTIVATION ANALYSIS

L. E. Fite, E. A. Schweikert,
and R. E. Wainerdi
Texas A & M University
College Station, Texas

and
E. A. Uken
Scientific Advisory Council
Pretoria, South A/rica

1. INTRODUCTION
The importance of the minor and trace elements in mineral chemistry is
becoming more widely recognized with advances in technology and inter-
pretation. The characteristics of many materials can be influenced by the
trace-element impurities found in various systems. For example, the quality
of steel is altered by its trace oxygen content! ; the presence of trace vanadium
in crude-oil poisons the platinum catalysts used in the cracking process by
the oil industry 2; sodium and chloride concentrations in sweat are used for
the detection of cystic fibrosis in children today 3; and the concentration of
trace pesticides in foodstuffs have importance. 4 Numerous other practical
applications of trace-element determinations have been investigated and
reported on in the literatureS in virtually all of the sciences.
The determinations of ultratrace, trace, and minor constituents in
natural systems may also provide a useful framework for the classification
of geochemical samples and may provide the basis of a method for mineral

319
320 L. E. Fite, E. A. Schweikert, R. E. Wainerdi, and E. A. Uken

exploration. To obtain useful information pertaining to possible mineral

deposits, or to the geochemical status of a given region, would require the
analysis of many properly selected samples. One method by which this may
be accomplished is atomic absorption spectroscopy (see Chap. 8) as de-
scribed by West. 6 A generally more sensitive method is nuclear activation
analysis. It can be automated under certain conditions; and, with the de-
velopment of solid-state detectors and the use of computers, the technique
can be made applicable to large-scale geochemical analyses.
Nuclear activation analysis is a technique for determining qualitative
and quantitative elementary composition by means of nuclear transmutation
and the subsequent measurement of emitted radiation from an unknown
substance. Basically, when a material is irradiated by nuclear particles
produced in a nuclear reactor, particle accelerator, cyclotron, or other
suitable source, some of the atoms present in the material interact with the
bombarding particles and, in many cases, are converted into radioactive
isotopes with known nuclear characteristics.
The separation and identification of each source of induced radioactivity
and a cataloging of the intensity of each becomes the basis for the measure-
ment of the quantity of the original element present in the irradiated material.
The radiation may be emitted either during the bombardment or in the
course of radioactive decay, wherein elements (actually isotopes) present
in a sample are identified by the characteristic energy and half-life of the
emitted radiation. The interactions of major importance in nuclear activa-
tion analysis are obtained from nuclear-particle-induced decay gamma rays.
The three basic steps in nuclear decay activation analysis are:
(a) Irradiation of the sample to induce the desired radioactivity.
(b) Measurement of the emitted radiations from the sample.
(c) Interpretation of the data to obtain the qualitative and quantitative
analysis of the samples.

1.1. Irradiation
The activating irradiation can be neutrons, protons, deuterons, tritons,
helions, or even high-energy gamma-ray photons. 7 The most common
methods of irradiation are summarized in Table I. The technique of activa-
tion analysis involving neutrons can be further separated into three types of
neutron-interaction analysis that produce gamma rays: inelastic scattering,
capture, and decay studies. Each process emits a gamma ray that is analyti-
cally characteristic for a given interaction and radioisotope. 8 Analyses based
on decay gamma rays are, by far, the most commonly used.
11. Nuclear Activation Analysis 321

Table I
Methods of Irradiation

1. Nuclear reactor
Advantages
High thermal neutron flux density
Simultaneous irradiation of many samples
Can provide long irradiations
Disadvantages
High initial cost
Suitable reactors are not widely available for service irradiations
Flux variability from one sample location to another

2. 14-MeV neutron generators (Cockcroft-Walton type)

Advantages
Relatively low initial cost
Monoenergetic 14-MeV neutron output
Portability
Disadvantages
Low flux compared to reactors
Difficult to achieve uniform flux dosage in sample

3. Isotopic neutron sources

Advantages
Low cost
Extremely reliable, with no maintenance
Reasonably constant neutron output
Disadvantages
Neutron production cannot be turned off
Flux levels generally lower than either generators or reactors

4. Cyclotron (charged particles)

Advantages
Suitable for activating lighter elements
By proper selection of particle type and energy interferences can be avoided-
surface analyses possible
Penetration depths can be controlled
Disadvantages
Comparatively expensive
Not presently suitable for large numbers of samples
Bulk analyses very difficult
322 L. E. Fite. E. A. Schwei kert. R. E. Wainerdi. and E. A. U ken

1.2. Measurement of Induced Radioactivity

The two principal approaches to evaluating induced radioactivities are:

(a) radiochemical separations followed by counting of the separated activi-
ties and (b) instrumental gamma-ray spectrometry. The instrumental ap-
proach to activation analysis, although generally simpler and more rapid
than the radiochemical approach, usually requires that great care be given
to the selection of bombarding particles, to the duration of bombardment,
and to the counting procedures employed. It also tends to be more limited
in application, particularly when the sample contains a number of trace
elements in a matrix that activates very easily. The analyses of geochemical
samples are usually performed instrumentally.
The various radioisotopes formed in a sample can be identified by a
device called a "gamma-ray spectrometer," which usually consists of a
gamma-ray detector, a pulse-height analyzer, and some form of read-out
device, and which measures the energies and intensities of the various gamma
rays that emanate from an activated sample. The two gamma-ray detectors
most commonly used are thallium-activated sodium iodide crystals [NaI(Tl)]9
and high-resolution lithium-drifted germanium [Ge(Li)], a solid-state device.

1.3. Gamma-Ray Interactions

Gamma rays interact in matter by three principal processes: photo-

electric capture, Compton scattering, and pair production.
A photoelectric interaction occurs when the complete energy of an
incident gamma ray is deposited within the detector. The signal output of
the detector can thus be proportional to the total energy of the interacting
gamma ray.
The Compton-scattering process involves the transfer of some of the
incident gamma-ray energy to an atomic electron, leaving the remainder of
the scattered gamma ray with reduced energy. The scattered gamma ray
may then interact by one of the three processes, or may escape the detector.
The effect is, therefore, a count of less energy than results from a photo-
electric interaction.
Pair production results in the formation of a positron and electron
pair during the capture of a gamma ray. This requires that a minimum of
1.02 MeV of energy be possessed by the incident gamma ray, with any excess
energy of the gamma ray appearing as kinetic energy shared between elec-
tron and positron that are produced as a pair. The positron will eventually
11. Nuclear Activation Analysis 323

recombine with some free electron, annihilating itself and the electron,
and producing two gamma rays of 0.51-MeV energy, moving 1800
apart. The annihilation radiation may interact in, or may escape from,
the detector.
The Compton continuum, caused by Compton interactions, makes
identification and recognition of low-energy gamma-ray photopeaks ex-
tremely difficult. Anticoincidence detectors have been designed for use with
not only NaI(Tl) crystals10 but also with Ge(Li) solid-state detectors.H
These detectors usually consist of a second scintillation detector surround-
ing the primary detector and serve the purpose of reducing the number of
Compton interactions that are counted.
When gamma rays interact with the atoms of a NaI(Tl) crystal, the
crystal emits visible light, proportional in intensity to the gamma-ray energy
deposited in the crystal. A photomultiplier tube, optically coupled to the
detecting crystal, produces a voltage pulse proportional in magnitude to
the intensity of the light striking the photocathode, and thus proportional
to the energy of the interacting gamma ray.
The solid-state detector [Ge(Li)] is a reverse-biased diode, and when
gamma rays interact with the atoms of this type of detector, free electrons
are produced. The magnitude of the current flow from the solid-state detec-
tor is proportional to the energy deposited within the detector by the gamma
ray that interacts with it.
Although the gamma-ray resolution of the Ge(Li) solid-state detector
is superior by a factor of approximately 30-35 to that observed with a
NaI(Tl) crystal, the efficiency of the Ge(Li) detector is significantly less
than that of a NaI(Tl) crystal. The magnitude of this reduction in counting
efficiency is, of course, dependent upon the respective sizes of the two de-
tectors, but, typically, the efficiency of a 30-cc Ge(Li) detector is 30 x lower
than that of a 3 x 3 in. NaI(Tl) crystal.
The current pulse produced by the Ge(Li) solid-state detector or the
voltage produced from a NaI(Tl) crystal, is then amplified and stored ac-
cording to its height in the memory of a pulse-height analyzer. The channels
of the analyzer correspond to energy intervals of the gamma-ray spectrum.
The number of pulses counted and stored in the various channels are,
therefore, a function of the characteristic gamma-ray spectrum of the radio-
active material being examined. A plot of this information will show peaks
in the channels corresponding to the particular gamma-ray energies that
were detected and stored in the pulse-height analyzer.
The development of the high-resolution Ge(Li) solid-state detector has
324 L. E. Fite. E. A. Schweikert. R. E. Wainerdi. and E. A. Uken

been very valuable for geochemical applications of neutron activation anal-

ysis. A gamma-ray spectrum from a geochemical sample can contain 50-70
distinct gamma-ray energies from 15-20 different elements.
The gamma-ray spectrum shown in Fig. 1 was obtained from a sulfide
standard of the Canadian Association of Applied Spectroscopists (CAAS)
using a 30-cc Ge(Li) solid-state detector. The sample had previously been
bombarded in a thermal-neutron-reactor flux density of 2 X 1012 n cm- 2 sec-1
for 4 h. Following a waiting time of 6 days to permit the decay of the short
half-life interfering radioactivities, the sample was counted for I h. From
the 26 observed photopeaks in the spectrum, 14 elements have been identi-
fied.

/.~p'.' '~'::f?:/i{:;:;;::~''.J .
'i,: ,.
WIGHT OF IAW'l.IE "'27 IIfG
._ '~:< J.~:/ _. ~TtOH
M[VT1It(l
TW(
f'wa
.. /'fits.
t . tOII NI(M2 xc
--~":: DELft TWIE'
CCUfTtNG TIllIE
6."
I 1M.
00tiTI

ot:T(CTat SlU ]0 CC

Fig. 1. Gamma-ray spectrum of CAAS sulfide ore-! using germanium-lithium detector.

11. Nuclear Activation Analysis 325

1.4. Data Interpretation

Computer techniques are highly suitable for the interpretation of data

obtained through nuclear activation analysis and may even be considered
essential for large numbers of geochemical samples in view of the speed with
which data can be accumulated from automatic activation-analysis systems.
Mathematical techniques are used to resolve the complex spectrum
into components that correspond to individual radioactivities. This proce-
dure is based on the principle that the combined gamma-ray spectrum,
measured under carefully controlled conditions, can be considered as a
function of the sum of the individual constituent radioactivities.
The activation equation, which describes the absolute method of quan-
titative analysis, is
A = N(1'/p(1 -- e-0.693Tl/lt/s) (e-O.693Tsltl/2)

where A is the intensity of the activity or the disintegration rate of the

radioisotope (dps), N is the total number of atoms of the target nuclide
present in the sample, (J is the activation cross section (cm2), 1p is the
neutron flux density (n cm- 2 sec-I), t1l2 is the half-life of the radioisotope,
TI is the irradiation time (same units as t1l2 ), and T2 is the delay time
from the end of irradiation to the beginning of the measurement of the
induced activity (same units as 1112 ),
The activation cross section is the probability that a neutron will inter-
act with the nucleus of a target atom and produce an unstable nuclide.
The quantity 1 - e-O.693Tlltl/s in the activation equation is known as the
saturation factor and e-O.693T.1t1/2 is designated as the decay factor.
Usually, instead of applying the above basic equation, the quantitative
analysis of an unknown sample is accomplished by comparing the intensity
of the radioactivity Au induced in the unknown sample with As produced
in a known standard of the same element. Thus,
Nu Au
Ns = As
where Nu is the number of atoms in the unknown and Ns is the number of
atoms of the same element in the comparison standard.
Both the unknown and the standard must be subjected to identical
irradiation and counting conditions in order to make this simple relation
applicable.
Various computer programs have been developed to process analytical
information on unknown samples. The principal mathematical methods of
326 L. E. Fite, E. A. Schweikert, R. E. Wainerdi, and E. A. Uken

data reduction utilize linear superposition to resolve a complex spectrum

into its component parts, thereby providing a qualitative and quantitative
analysis. This procedure requires that the conditions under which the spectra
are accumulated be such that the component spectra add linearly. Therefore,
in any energy interval LI E the total number of counts in the complex spectrum
is due to the summation of the contributions of all the isotopes present.
The most commonly used methods for separating a complex spectrum into
its components are spectrum stripping,12 matrix methods,13 restricted and
unrestricted least-squares curve fitting,u.I5 and peak-area calculations.I6
Each of these methods of data reduction requires that computers be sup-
plied with "library" information concerning each radioisotope that may be
observed in the complex spectrum. The library information contains the
known characteristics of the radioisotope, such as its half-life and emitted
gamma-ray energies, and a spectrum of the pure radioisotope. This spec-
trum is obtained by irradiating and counting the pure radioisotope and the
unknown sample under identical conditions. A description of each of the
four methods mentioned above follows.

1.4.1. Spectrum-Stripping Method

The spectrum-stripping method of data reduction is accomplished by
first identifying the highest energy, statistically significant photopeak in a
spectrum. The photopeak can be identified by the gamma-ray energy or
energies and half-lives of the library radioisotopes. Once identified the
library spectrum of the pure radioisotope is adjusted to yield the same peak
area as observed in the complex spectrum; this component is stripped, or
subtracted from, the complex spectrum. Since the library spectrum was
obtained from a known amount of the pure element, the quantitative
analysis is determined by the amount of the library radioisotope removed
from the complex spectrum. This process is repeated until all the peaks have
been examined and eliminated. The principal limitation of spectrum strip-
ping is the inaccuracy resulting from the inherent compounding of statistical
errors by successive subtractions.

1.4.2. Matrix-Solution Method

The matrix solution for a complex spectrum is based on the fact that
the total activity in any given energy interval LIE is the sum of all the con-
stituent activities within that interval. Thus, for any energy interval
11. Nuclear Activation Analysis 327

where Uj is the activity of the complex unknown spectrum for the energy
interval j, Sji is the activity in the energy interval j for the standard library i
and ai is the ratio of the activity of the unknown in the energy interval j
to the standard library spectrum i.
The value of ai is determined by solving a set of simultaneous equations
that are constructed by measuring the height of the complex spectrum at
different energy intervals. The number of equations is determined by the
number of components in the spectrum.

1.4.3. Least-Squares Curve Fitting

The least-squares technique forms a synthetic spectrum from the library
spectra in such a manner as to minimize the square roots of the variations
between the observed spectrum and the synthetic spectrum. The factors by
which the library spectra are adjusted to yield the best fit of the synthetic
spectra to the observed spectra provide the quantitative analysis of the un-
known sample. Certain restrictions should be employed when using the
least-squares method to eliminate negative results. This can be accomplished
by omitting the isotope or isotopes that yield negative results from the solu-
tion of the least-squares equations.

1.4.4. Peak-Area Calculations

The quantitative analysis of an unknown sample may also be deter-
mined by peak-area calculations. This is accomplished by comparing the
peak areas observed in the complex spectrum to the photopeak areas obtained
in the library spectra under the same irradiation and counting conditions.
Various methods of determining the peak boundaries and the peak areas
have been investigated and reportedP The peak-area method generally
works very well unless there are interfering peaks, and, by using Covell's
method,18 the boundaries can be chosen to minimize the effect of the inter-
ference. The Covell and classical methods are widely used in activation-
analysis calculations and generally give accurate results.

1.5. Automated Nuclear Activation-Analysis Systems

The automation of both radiochemical and instrumental separation

methods has expanded the applicability of activation analysis to routine
geochemical problems involving large numbers of samples. Automation of
purely instrumental systems can be multipurpose19 or of special design for a
particular analysis, such as the determination of oxygen, aluminum, silicon
328 L. E. Fite. E. A. Schweikert. R. E. Wainerdi. and E. A. U ken

magnesium, and iron in rocks.20 Automated analysis systems capable of

handling large numbers of similar samples using both reactors and acceler-
ators as the activation source have been designed and are routinely used
for the trace-element analysis of various materials. 21 - 23
The sampling and preparation methods as well as the parameters for
the analysis are determined by the trace elements of interest in the unknown
sample, and by the interferences present in the sample matrix.

1.5.1. Sampling and Sample Preparation and Manipulation

Considerable attention must be given to the collection of the samples

to obtain representative geochemically valid data. This is of particular
importance when considering geochemical samples since the weathering
process may extend several hundred feet into the earth's crust.24 It may be
desirable to crush the specimens to improve the homogeneity of the sample,
but grinding the sample could also introduce contamination. As a general
rule, preparation of the samples should be kept at a minimum, and it re-
quires meticulous care to minimize contamination. Because the method of
neutron activation analysis is a bulk method, it is usually desirable and
sufficient to obtain samples in the mg range.
Since instrumental activation analysis is a nondestructive technique, it
may be desirable to reanalyze a given sample to verify the previous results
or to recount the sample at specified intervals of time to follow the decay
of the radioisotopes produced by the bombardment of the sample. The
ability to reactivate and/or recount samples as desired is one of the principal
advantages of activation analysis in comparison to destructive methods.
Samples for activation analysis are usually placed in small, clean poly-
ethylene vials for irradiation and counting. Very small and usually known
amounts of sodium, chlorine, aluminum manganese, and vanadium are
among the common contaminants observed in the polyethylene vials. Such
vials become brittle and crack owing to irradiation damage during high
reactor-power irradiations; therefore, quartz or aluminum containers are
sometimes used where very high neutron fluxes or high temperatures are
required. It is occasionally necessary to remove the sample from its irradia-
tion container and place it in a nonirradiated vial for counting because of
the impurities common to the different types of vials. The transfer step is
difficult to automate and is, for the most part, done manually. Great care
must be exercised to avoid partial or total loss of the irradiated sample;
but postirradiation contamination is impossible, which is another principal
advantage of activation analysis.
11. Nuclear Activation Analysis 329

1.5.2. Analysis Parameters

Prior to the routine analysis of large numbers of samples it is desirable
to optimize the parameters of analysis, which include irradiation, waiting
and counting times, the type of irradiation source, and counting-system
geometry. These parameters are dependent upon the gamma-ray energies
of the emitted radioactivities, the activation cross section of the elements
present in the sample matrix, and the half-life of the induced radioisotopes
from not only the trace or minor elements of interest, but also the other
elements within the sample that become radioactive. Large numbers of
similar samples can then be processed automatically and routinely at rapid
rates by unskilled personnel, provided no significant interferences are present.

2. REACTOR-THERMAL-NEUTRON ACTIVATION
ANALYSIS

The basic components required for the analysis of short-half-life iso-

topes, i.e., radioisotopes with a half-life in seconds or minutes, are a pneu-
matic system, an irradiation and delay timer, a pulse-height analyzer with
a suitable gamma-ray detector, and a computer-compatible read-out system.
The automatic activation-analysis system shown in Fig. 2 is capable of
irradiating, counting, and analyzing large numbers of similar samples. The
system is primarily used for the determination of short-half-life radioiso-
topes, which are induced in samples by activation with thermal neutrons.
The sample is pneumatically transferred to the reactor core for irradiation
and returned for immediate counting and analysis. After the previously
mentioned experimental parameters have been established, a technician
following simple instructions can irradiate and count the activated samples
on a routine basis. This system has been used to determine vanadium and
titanium in geochemical samples on a routine basis.25
Standards of the radioisotopes of interest should be analyzed periodical-
ly to serve as flux monitors and also as checks on the activation-analysis-
system performance. If the same samples are periodically reanalyzed, the
technician operating the system can recognize a system malfunction by the
altered location of the gamma-ray photopeaks and their relative peak heights
after irradiation and counting, since the same composition should result for
each analysis of the same sample.
The analysis of long-life radioisotopes can be performed on an auto-
matic activation-analysis system, such as shown in Fig. 3, or by using the
330 L. E. Fite, E. A. Schwei kert, R. E. Wainerdi, and E. A. U ken

TO ~[A(.TOR
COO.

"

Fig. 2. Mark I-Ia automatic activation-analysis system.

high-resolution spectrometer, shown in Fig. 4. The Texas A & M Mark II

Automatic Activation Analysis System shown in Fig. 3 is a very-high-
capacity automated spectrometer. The system consists of three gamma-ray
spectrometers, i.e., three 3 X 3 in. NaI(TI) crystals and three 400-channel
pulse-height analyzers with digital gain and baseline stabilization, two sample
storage libraries, a pneumatic transfer system, a real-time clock, a computer-
compatible magnetic tape read out, and a control system.
The samples located in the storage library are selected automatically by
the control system and transferred to one of the detectors via the pneumatic
tube. When the count starts, the time is recorded to the nearest 0.01 min
on magnetic tape along with the sample identification number and the
analyzer code number. The analyzer code number is used by the computer
to determine which samples were counted by each pulse-height analyzer.
At the end of the counting period the time and the data stored in the memory
of the pulse-height analyzer are recorded on the magnetic tape. This system
is capable of analyzing samples at the rate of 100jh and is routinely used in
biomedical applications of activation analysis. 26 The use of the system for
geochemical applications would be limited by the resolution of the NaI(TI)
crystals; alternatively, Ge(Li) detectors could be used.
11. Nuclear Activation Analysis 331

LAfJOli
'CiHf~~f~S
OE
Fig. 3. Mark II automatic activation-analysis system.

Another automated system is shown in Fig. 4 and is designated as the

high-resolution spectrometer. This device consists of a shielded 30-cc Ge(Li)
detector, a 3200-channel pulse-height analyzer, a computer-compatible mag-
netic tape read out, a storage "library" for the samples after activation, a
pneumatic system, a real-time clock, a sample identification, and a control
system. The sample identification number, the time of day at the beginning
and end of data accumulation, and the live time of counting are punched
onto paper tape, and the spectral data is read onto magnetic tape. After
converting the information on the paper tape to punch cards, the cards and
the magnetic tape are processed by the computer to determine the analysis
of the elements in the sample.
The routine analysis of 350 geochemical samples for gold content was
recently completed using this automatic activation-analysis system. In addi-
tion, zirconium, lanthanum, and samarium have been routinely determined
in geochemical samples with this device. The system was also used in a
research program to study the distribution of iron, scandium, cobalt, lan-
thanum, europium, hafnium, and tantalum in a specific geologic area. 27
Samples to be counted on the two automated systems described above
are usually irradiated in a reactor-mounted rotisserie that is capable of
holding 100 samples. Standards are also distributed throughout the rotisserie
positions to serve as comparison mass standards for the analysis and as
neutron flux monitors. It is assumed that all of the samples in the rotisserie
are exposed to known levels of neutron flux, and this assumption is verified
by analyzing known standards.
 CAl
CAl
N

30cc Ge (Li)
DECTECTOR KENNEDY MAGNETIC
. TAPE UNIT SAMPLE STORAGE
,
.
3200 CHANNEL AUTOMATED SYSTEM :
I
ANALYZER
, m
."
;:;:
!D
rn

en
o
::T
~
~.
~
III
;:.
:zI

rn
~
I
:::l
III
...Q.
I
:::l
Q.
m
Fig. 4. High-resolution spectrometer.
?>
C
~
CD
:::l
11. Nuclear Activation Analysi s 333

Table II
Summary of Experimental Detection Limits Determined with Reactor Thermal
Neutrons a

Atomic Minimum detection ranges, fig

No.
range 10-5 10-4 10-3 10-" 10-1 100 101 10' 103 10'

8-9 F 0
11-18 Ar Na,Al Cl Mg Si, S
19-35 Mn Sc, V, Co, Cu, K, Ti, Cr,Ni Ca Fe
Br Ga, As, Zn, Ge
Se
37-53 In Sr,Rh, Rb,Zr, Y,Nb
Ag, Cd, Mo, Ru,
Sn, Sb, I Pd, Te
55-56 CS,Hf Ba,W Ta, Pt, Pb
Os
57-71 EU,Dy Sm, Ho, La, Nd, Ce,Pr, Tm
Er Gd, Yb Tb
Lu
72-82 Au Re,Ir Hg

a The thermal neutron flux is 4 X 1012 n cm- 2 sec- 1

The detection limits for various elements that have been experimentally
determined are given in Table II. These detection limits were obtained using
a NaI(Tl) crystal as the gamma-ray detector and would vary somewhat for
Ge(Li) solid-state detectors because of their lower counting efficiency.
Since most of the elements within a sample become radioactive when
bombarded with thermal neutrons, often radioisotopes are produced that
emit radiation that will interfere with the measurement of the desired radio-
activities; therefore, it may be desirable to employ radiochemical separation
techniques to remove the induced activity of interest from the other induced
activities in a given sample. Generally, radiochemical techniques are based
on postactivation radiochemical separations, and the automation of these
techniques may expand the applicability of activation analysis. Automated
radiochemistry systems are still in the developmental stage but tend to be
specialized and designed to separate one or more elements from a given
matrix. Radiochemical separation techniques have been applied to geo-
334 L. E. Fite. E. A. Schwei kert. R. E. Wainerdi. and E. A. U ken

chemical analysis of rock to reduce the matrix interference from the radio-
activities of interest,28 but for large numbers of analyses at modest cost
instrumental methods are recommended at the present time.

3. 14-MeV NEUTRON ACTIVATION ANALYSIS

For the determination of certain minor elements in geochemical samples

it may be desirable to use a 14-MeV neutron generator as the irradiation
source. A typical automated activation-analysis system using a 14-MeV
neutron generator is shown in Fig. 5. The sample is transferred to the
14-MeV neutron generator by the pneumatic tube and irradiated for a
preset time as determined by the control system. Various neutron monitors
are utilized to monitor the output of the neutron generator. At the end of
the irradiation the sample is transferred to a position between two NaI(TI)
detectors, which are connected to a pulse-height analyzer, and, after a pre-
determined delay time, the data is accumulated. The spectral data is usually
placed on paper tape while the next sample is being irradiated. This system
is routinely used in the analysis of geochemical samples for their aluminum,
iron, magnesium, silicon, and oxygen content,29.30 including the deep-sea
core samples recovered by drilling operations on the ocean floor. 31 Table III
indicates the experimental detection limits for numerous elements 32 for a
14-MeV neutron flux of 2 X 108 n cm- 2 sec-I. This system, as well as the

400 CH.ANNEL. ANAt..YZER

PNEUMATtc CONTROL.
t7:, DETECTOR
I 2-3 X3- NoltTIl
N

CRYSTALS

NEuTRON GENERATOR
CONTROl. CONSOLE

Fig. 5. Mark III automatic activation-analysis system.

11. Nuclear Activation Analysis 335

Table III
Summary of Experimental Detection Limits Determined with 14-MeV Neutrons a

Atomic Minimum detection ranges, mg

No.
range 0.001 0.01 0.05 0.1 0.5 1.0 5.0 10.0

8-9 o F
11-18 Si P AI Na Mg, CI
19-36 Cr Mn Ni Fe K, Co
Cu Se Br As
37-54 Sr Rb Pd Mo,Ru Nb
Sb Cd Sn Ag
In
55-71 Pr Ba Dy Tb Sm Eu Cs Gd
Nd Ce Er Lu
72-86 Ta Hf Ir
Hg Au Pb

a The neutron flux is 2 X 10' n cm- 2 sec-I.

others previously described, uses polyethylene pneumatic tubes, and all

transfer boxes are built from plastic without lubricants to further reduce
the possibility of sample contamination.
A portable system similar to the one shown in Fig. 5 has been success-
fully used for the routine analysis of samples in the field. A sealed-tube
neutron generator was placed in a 6-ft-deep, 3-ft-sq. hole while the detector
and the counting equipment were located in a trailer adjacent to the hole.
By placing the neutron generator in the hole, adequate shielding was pro-
vided for the system operator and the counting equipment from the neutrons
produced by the generator.
An automatic activation-analysis system that is capable of performing
in situ compositional analysis of a geologic area of interest is shown in Fig. 6
and is called the "turn-around" system. This system consists essentially of
a small 14-MeV neutron generator and a NaI(TI) crystal. The system is
rotated about the center axis of the device to allow either the detector or
the neutron source to be placed directly above and close to the sample.
This configuration permits the maximum practical distance between the
neutron generator and the detector, to prevent activation of the detector
crystal, while providing the optimum geometry for irradiation and counting.
The gamma rays detected in the scintillation crystal are stored in a pulse-
336 L. E. Fite, E. A. Schweikert, R. E.Wainerdi, and E. A. Uken

NEUTRON
Q[NERATC)ft

. ,..... AXIS Of ROTATION

.,'""1'' '' '''''/

Fig. 6. "Turn-around" system for remote activation analysis.

height analyzer (not shown). This system has been used in the analysis of
numerous rock samples33 34 and was developed as part of a proposed moon
probe.
A similar system, which has no moving parts, is shown in Fig. 7. In
addition to decay activation analysis, this system can be used for inelastic

---IT3-'-E NEUTRON ATTENUATOR 3 ' ~RON SOURCEJ- ' -

.. ' .:: '.:' " -. . ~ .

Fig. 7. Physical arrangement of major components for a combined activation

analysis system.
11. Nuclear Activation Analysis 337

scattering, capture gamma ray, and cyclic activation analysis of large geo-
chemical samples. When considering decay activation analysis, this system
has the disadvantage of poor irradiation and counting geometry when com-
pared with the previously mentioned system. 35

3.1. Cyclic Activation Analysis

In this method of analysis the sample is bombarded with a short burst
of neutrons and the induced activity is measured between bursts. A suitable
delay between the neutron burst and the counting must be made to allow
for the depletion of the neutron-capture gamma rays. This technique has
been demonstrated in the analysis of rock samples for aluminum, silicon, and
oxygen content. 36 A serious limitation of the method results from the poor
irradiation and counting geometry arrangement; however, the method is well
suited for the determination of radioisotopes with a relatively short half-life.

4. CHARGED-PARTICLE AND PHOTON ACTIVATION

ANALYSES

4.1. Introduction
In addition to thermal and fast neutrons, photons and charged particles
can also be used for activation analysis. Almost all stable elements can be
determined with excellent sensitivities (10- 6_10-10 g) using photons, protons,
deuterons, 3He, or 4He ions of energies up to 40 MeV. More specifically
the interest for these activating particles lies in that they offer high sensi-
tivity for most of the elements which cannot be determined with neutron
activation analysis or can only be determined with limited sensitivity. This
is the case for the elements with atomic numbers 3 to 23 and a few
heavier elements such as iron and lead.
For a comprehensive survey on the field of photon and charged-particle
activation analysis the reader is referred to the detailed discussions given by
Albert,37.38 Engelmann,39.40, Baker,41-43 and Tilbury.45

4.2. Photon Activation Analysis

4.2.1. Nuclear Reactions and Sources of Photons

The type of nuclear reaction induced by photons depends upon
their energy. At low photon energies (3-6 MeV) generally only y, y' re-
actions can occur. At higher photon energies (Ey > 7 MeV), the most
338 L. E. Fite, E. A. Schweikert, R. E. Wainerdi, and E. A. Uken

prominent reaction is y, n, which for photon energies 7-10 MeV has cross
sections typically 103 higher than y, p and 105 higher than y, a reactions,
for example. However, even for y, n reactions the absolute values of the
cross sections are usually small (10-100 mb), thus for determinations featur-
ing high sensitivity, high-intensity sources of photons are needed. The
threshold energies for y, n reactions on light elements range from 10 to
15 MeV and from 7 to 10 MeV for the heavier elements, the resonance
energies giving maximum cross sections are usually 10 MeV above the
threshold values.
Photons of the required energy and high intensity can be obtained as
bremsstrahlung photons produced by bombarding a high-Z-number target
with electrons. The accelerators used for producing high-energy electrons
are Van de Graaff accelerators, cyclic machines, such as the betatron, or
linear accelerators, "linacs." Linacs are used mostly for photon activation
analysis since they feature two important advantages, production of high-
intensity (up to 250-,uA average electron currents) and high-energy (up to
45-MeV electron energies) electrons and thus bremsstrahlung fluxes.

4.2.2. Analytical Applications

This discussion is divided into two parts, arbitrarily termed low-energy
photon activation using photons up to 6 MeV, and high-energy photon
activation with photons of energies higher than 7 MeV, typically photons
of 15-40 MeV and as high as 70 MeV.
Low-Energy Photon Activation Analysis. This type of activa-
tion has been used for the determination of heavier elements (Se to Mg)
where y, y' reactions yield short-lived isomers as, for example 77Se(y, y')77mSe
or 179Hf(y, y')179mHf. The cross sections for y, y' reactions are low (a few
mb), thus limiting the sensitivity which is generally in the range of a few
milligrams. 45 ,46
Two y, n reactions can also occur in the low-energy range: 9Be(y, n)8Be
and 2H(y, n)lH. In both cases the reaction products are not radioactive,
and the analytical applications developed are based on the detection of the
neutrons emitted during irradiation. 47 The thresholds for these reactions
are sufficiently low, 1.67 MeV and 2.23 MeV, respectively, that photons
from radioactive sources can be used for activation. Beryllium can be de-
termined at concentration levels as low as I ppm48 ,49 using relatively low
intensity sources of up to I Ci 0[124Sb (Ey , 2.1 MeV) and 8Sy (Ey, 1.84 MeV)
Deuterium down to 0.02 vol. % of D 20 has been measured 50 ,51 using 24Na
(Ey , 2.75 MeV).
11. Nuclear Activation Analysis 339

High-Energy Photon Activation Analysis. A very large range

of elements can be determined with excellent sensitivities using y, n reactions
with bremsstrahlung photons of 20-35 MeV. To illustrate this, some detec-
tion limits reported 52 are given in Table IV. These detection limits are cal-
culated for an interference-free irradiation equal to one half-life or l-h
maximum using a 27-MeV electron beam of 50 pA with the following
counting conditions: minimum of 100 counts/min for radioisotopes with
half-lives of less than 6 days and 500 counts/600 min for radioisotopes of
half-lives longer than 6 days, counting on a 3 X 3 in. NaJ(Tl) detector.
Additional information on calculated detection limits for a large number of
elements at different irradiation energies can be found in the papers pub-
lished by McGregor, 54 Engelmann,39.4o Wilkniss and Linnenboom,55 Ba-
ker,42.43 Anderson et al.,56 and Owlya et al. 57
The analytical procedures for destructive or nondestructive determina-
tions are similar to those used in neutron activation analysis. Two particular
points must be mentioned, however, one concerning the correct and repro-
ducible positioning of the sample in the bremsstrahlung beam and the other
concerning the selection of the proper irradiation energy. These questions
have been discussed by Engelmann,39.4o Baker et al.,58 Wilkniss and Linnen-
boom,55 and Lutz.44
Most applications reported have dealt with light-element analyses such
as the determination of oxygen and especially carbon and nitrogen in high-
purity materials. 59 - 61 Rocco et al. 62 have used this technique for determining
nitrogen in diamonds. The sensitivity limit for nondestructive analyses was
estimated at 0.2 ppm of nitrogen for 40-mg samples.
Several investigators41-43.52.53.55.56 have studied interesting analytical
applications for determining heavier elements (Z > 10), where photon acti-
vation has definite advantages over neutron activation for the following
reasons:
(a) Photon activation gives higher sensitivity or the reaction product
has a more readily identifiable gamma spectrum or half-life (de-

Table IV
Examples of Detection Limits in Photon Activation Analysis

0.1-1 fAg Cr, Mn, Te, Co, Ni, Zn, Mo, Hf, Pb
0.01-0.1 fig C, N, 0, F, Ti, As, Sr, Zr, Nb, Ag, Sb, Ta, Pt
0.001-0.01 fAg Cu
340 L. E. Fite, E. A. Schweikert, R. E.Wainerdi. and E. A. Uken

terminations of titanium, iron, magnesium, and calcium, for

example).
(b) An interference is eliminated (analysis of nickel in copper).
(c) Self-shielding problems as occurring in thermal-neutron activation
with matrices composed of boron, tungsten, manganese, cadmium,
and hafnium are eliminated.
Specific applications in geochemical analysis are illustrated in the
references cited below. Berzin et al.63 have reported on the analysis of
rocks and ores with selective irradiations using successively increasing
photon energies. With this procedure interfering contributions could be
accounted for. Copper, zinc, barium, silicon, and iron were determined in
ores using successively a 10.8-MeV irradiation giving barium, an 11.6-MeV
irradiation giving barium and copper, a 12.8-MeV irradiation giving barium,
copper, and zinc, and, finally, a 15.6-MeV irradiation giving copper, zinc,
barium, silicon, and iron. Schmitt et al.64 have described elegant non-
destructive methods combining photon activation and high-resolution
Ge(Li) gamma-ray spectrometry. Several meteorites and USGS Standards
were analyzed for their major- and minor-element content. Magnesium,
calcium, titanium, and nickel were determined nondestructively. In addition,
this method offered similar sensitivities to those attained in instrumental
thermal-neutron activation for sodium, scandium, chromium, manganese,
iron, cobalt, rubidium, yttrium, and cerium.

4.3. Charged-Particle Activation Analysis

4.3.1. Nuclear Reactions and Sources of Charged Particles

As for photon activation, this technique provides a wide analytical po-
tential complementing the capabilities of neutron activation. A number of
analytically useful nuclear reactions can be induced by charged particles of
low (a few hundred keY to a few MeV) and medium energies (up to 50 MeV).
Each reaction has its characteristic threshold and resonance energies
with cross sections which can be quite large (up to the barn range). Examples
of the different types of nuclear reactions which can be used for the determi-
nation of oxygen are given in Table V.
The minimum energy required for a nuclear reaction to occur (except
for low-energy elastic scattering or Rutherford scattering) is set by the
Coulomb barrier of the target nucleus with respect to the incident and out-
going particles and, in addition, for endothermic reactions, by the Q value
(energy balance) of the reaction. 44 A variety of sources of charged particles
11. Nuclear Activation Analysis 341

Table V

Low-energy reactions Medium-energy reactions

160(p, P )160 180(p, n)18F

160(p, y)17F 160(p, a)13N
180(p, a)lN 180(d, n)18F
180(p, y)19F 160("He, p )18F
160(d, p) 170 180(3He, pn)19F
180(d, p)190 160(4He, pn)18F
160(d, a)l<N
180(d, a)16N

are used for charged-particle activation analysis depending upon the type
and energy of particle required: isotopic sources (a-emitting heavy radio-
isotopes), Cockcroft-Walton and Van de Graaff accelerators, heavy-ion
linear accelerators ("hilacs"), and cyclotrons.
Activation techniques based on the use of charged particles must take
into account the fact that their penetration is very limited and dependent
upon the kind and energy of the bombarding particle and the chemical
composition of the target. From the standpoint of analytical applications,
two important consequences result from the rapid loss of energy of charged
particles in matter:
(a) The limited penetration depth and the corresponding small sample
volumes analyzed are of disadvantage in analytical applications
directed toward obtaining meaningful average values of an impurity
content. On the other side the energy-dependent penetration opens
interesting possibilities for investigating impurity distributions in
successive sample thicknesses. The only applications in this direc-
tion have, however, been limited so far to utilizing low-energy
charged particles for surface-impurity determinations. It must also
be noted here that the energy lost through the stopping of the
bombarding particles is given up in the form of heat (for example,
a lO-MeV proton beam of 10 flA intensity dissipates 100 W). Thus,
the target must be heat resistant or cooled in an appropriate way.
(b) A nuclear-reaction cross section varies with the energy of the
bombarding particle; thus, the specific radioactivity induced will
vary with the thickness of the sample. This nonuniform activation
implies that particular attention be given to ensure correct quantiza-
342 L. E. Fite, E. A. Schweikert, R. E.Wainerdi, and E. A. Uken

tion. This question has received thorough attention and different

investigators have proposed various quantization methods. 65 - 67 A
discussion on quantization and flux monitoring in geochemical
analyses has been given by Sippel et al. 68

4.4. Analytical Applications

4.4.1. Prompt Detection Methods

These methods, developed for surface analysis, use low-energy charged
particles and are based on the measurement of the prompt y radiation
emitted or of the elastically or inelastically scattered particles.
Nuclear interactions emitting prompt y radiation such as p, y reactions
have strong resonances for protons up to 4 MeV. Excellent sensitivities for
p, y activation have been reported for a number of light elements by Pierce
et al. 69 A specific application studied by these authors dealt with the mag-
nesium, aluminum, silicon, and iron distribution in copper pyrites and the
aluminum distribution in quartz-tourmaline samples. 70
Turkevich 71 has demonstrated the usefulness of elastic and inelastic
scattering of charged particles in geochemical analysis. He has, in particular,
applied this technique for the first determination of the chemical composi-
tion of the lunar surface. A Cm-242 source emitting ex particles of approxi-
mately 6 MeV was used for this purpose. On the heavier elements these ex
particles undergo Rutherford scattering, whereas on the light elements
nuclear interactions take place; in particular, ex, p reactions were used for
their detection. The energy of the scattered particles is dependent upon the
mass number of the scattering nucleus; thus, by measuring the energies
and intensities of the particles scattered at a given optimum angle with
respect to the incident beam, qualitative and quantitative information on
the chemical composition is obtained. Three different sites were analyzed
on three different Surveyor missions 72 - 75 and eight elements or groups of
elements were determined: carbon, oxygen, sodium, magnesium, aluminum,
silicon, "calcium" or elements with mass numbers between 28 and ap-
proximately 45, and "iron" or elements with mass numbers ranging from
approximately 45 to 65.

4.4.2. Techniques Involving Delayed Counting

Most analytical work utilizing high-energy charged particles has dealt
so far with the determination of trace amounts of the light elements (boron,
11. Nuclear Activation Analysis 343

carbon, nitrogen, and oxygen, particularly) in high-purity metals and semi-

conductors. 76 - 81
The main roadblock preventing, up to now, the application of this
technique to geochemical problems has most probably been the lack of
access to sources of high-energy charged particles such as Van de Graaff
accelerators and cyclotrons. It can be expected that with the multiplication
of these machines and the advent of small cyclotrons,82 this technique will
certainly be used more often. Most likely future applications should be forth-
coming on such problems as the determination of the isotopic ratios of light
elements as, for example, 12C_13C in meteorites and the quantitative analysis
of trace elements, as, for example, lead, for which charged-particle activa-
tion analysis appears to offer outstanding sensitivity.83

5. ACTIVATION ANALYSIS IN THE GEOSCIENCES

Geoscientists are making ever-increasing use of the various methods of

activation analysis. A summary84 of results obtained for the complete anal-
ysis of standards G-I and W -1 indicates that over 50 elements were deter-
mined by neutron activation analysis. In most cases the reported results
compare favorably with those obtained by other analytical procedures,
especially for those elements that are present in trace amounts only. The
applications are, however, no longer confined to laboratory assaying, but
a number of field tests and in situ analyses have already been and are still
being developed. Some examples of these trends are briefly outlined in the
following sections.

5.1. Mineral Prospecting

A feasibility study85 has revealed that some 30 elements may be detected
by irradiating the ground surface for periods of a few minutes with fast neu-
trons. It has been found that 3-MeV neutrons in penetrating the ground are
thermalized to produce a relatively uniform flux of neutrons of energy below
the I-keV range, at an approximate depth of 20 in. There is a strong possibil-
ity that elements that are not amenable to neutron activation analysis may
still be detected on grounds of their association with certain other elements
in nature. The distribution of lead, for example, may be mapped by deter-
mining silver. Fluorine and chlorine may be indicative of mineralization,
thus leading to the detection of an ore deposit.
The formal study of the deposition and distribution of minerals in the
earth's crust, called metallogeny, may be carried out by investigating the
344 L. E. Fite. E. A. Schwei kert. R. E. Wainerdi. and E. A. U ken

areal distribution of the trace elements. According to proponents of metallo-

genesis,86-90 the concentrations of groups of elements in ores are basically
determined by the presence of superimposed structures in so-called "metallo-
genic provinces." A study of the distribution of some trace elements, as
determined by purely instrumental, reactor-thermal-neutron activation
analysis, has revealed the potential of this approach. 91 .92 The simultaneous
determination of the elements Fe, Sc, Co, La, Eu, Hf, and Ta in uncrushed,
whole-rock specimens, not only revealed the compositional trends of these
elements across the different zones of a batholith, but a later intrusive was
also identified.91 An extension of this mapping procedure has indicated the
existence of a particular gold deposit in the Philippines. 92 With some re-
finements at least 15 elements may be determined simultaneously in rock
samples, by instrumental activation analysis. 93 Large-scale mineral explora-
tion based on activation analysis therefore appears feasible. It has been
suggested that a fast-neutron source may be used for the remote analysis
of major constituents of relatively inaccessible areas, including lunar and
planetary surfaces. 94 The availability of the high-thermal-neutron output,
252Cf spontaneous-fission source, will favor the determination of minor rock
constituents in the field. The continental-drift theory and other related stud-
ies that are aimed at establishing the similarity of rock types may also be
studied. Activation-analysis techniques are particularly useful where only a
small amount of sample is available, as in meteorites. After the analysis the
sample material may still be studied mineralogically, since this instrumental
technique is nondestructive.

5.2. Logging Techniques

The major advantage of nuclear logging techniques over other methods

is that they may be applied to encased holes. Both neutrons and gamma rays
will penetrate steel or cement casings.
In the natural-gamma-ray-Iogging technique naturally radioactive ele-
ments such as uranium, thorium, and potassium may be detected by passing
a sodium iodide probe down the borehole. A full description of radiometric
techniques is given in Chap. 10.
In the gamma-gamma or bulk-density-logging technique a gamma-ray
source is used such as Co-60, for example. The gamma rays emitted are
reflected from the surrounding material to reach a sodium iodide detector,
which is shielded from the source. The number of reflected gamma rays
reaching the detector is statistically related to the density of the surrounding
material. For a fixed-source strength, distance between source and detector,
11. Nuclear Activation Analysis 345

and borehole geometry, the instrument may be calibrated in units such as

gJcm 3
In neutron-neutron or moisture logging a neutron source emitting fast
neutrons is used in conjunction with a thermal-neutron detector. The prin-
ciple underlying the slowing down of neutrons may be illustrated as follows:
if a tennis ball (i.e., a fast neutron) is thrown against a solid or heavy wall
it bounces back, and little energy is lost. If the ball is thrown against a cur-
tain which is relatively light, the ball does not recoil, but drops straight
down. The fast neutron loses its energy in a similar fashion if it collides
with a hydrogen nucleus. If a large number of hydrogen atoms are present,
a large number of incident neutrons will be slowed down. The number of
fast neutrons surviving may therefore be said to be inversely proportional
to the hydrogen content. On an atom-per-volume basis water contains much
more hydrogen than any other naturally occurring inorganic material.
Neutron-flux variations as determined by the neutron detector are there-
fore a measure of the variation in the water or moisture content of the ma-
terial surrounding the borehole. The instrument may be calibrated to deter-
mine the weight of water per unit volume. This technique has been success-
fully applied by the Kennecott Copper Corp., Utah, to study dump leach-
ing processes. 95 If a gamma-ray detector is used in conjunction with a
neutron source, the activation products may also be determined, of course.

5.3. Ore Sorting

The major advantage of sorting by nuclear techniques rather than by
optical methods based on some physical property such as color, for example,
is that activation analysis is not restricted to the shape, appearance, or sur-
face composition of a particular sample. Shielding requirements do create
a problem, but these are surmountable.
The principles involved in the continuous on-stream monitoring of a
particular isotope have already been layed down. 96,98 A good example of an
automated neutron-activation-analysis system in industry may be found in
the routine determination of oxygen in steel,97,l as is currently carried out
by the South African Iron and Steel Corp., Pretoria. The gamma-ray activity
resulting from the 160(n,p)16N reaction with l4-MeV neutrons is measured
by integrating the gamma-ray activity above 4.5 MeV.
Signals from such a monitoring device may be fed into a sorter, in order
to pick out or divert a particular fraction of the material passing along a belt.
A promising application is the separation of rock from coal, by monitoring
the aluminum content. 99 The determination of carbon, oxygen, aluminum,
346 L. E. Fite. E. A. Schweikert. R. E.Wainerdi. and E. A. Uken

and silicon in coal has also been investigated.1oo Taconite ore may be sorted
on the basis of the continuous analysis of iron and silicon by 14-MeV-
neutron activation analysis. Copper101 and fluorine 102 in their respective ores
may also be measured. A method for monitoring the addition of fluorine to
drinking water has been described.1 03 An isotopic neutron source such as
americium-beryllium may be used for the determination of fluorine, since
the isotopic neutron source is not only cheaper and more plant compatible
than a neutron generator, but the reaction cross section for fluorine with
3-5-MeV neutrons is also larger.
Other elements with short-lived radioactive products such as Au-97m104
are currently receiving attention. More sophisticated techniques such as
neutron time-of-flight methods may also be employed in order to extend
the application of nuclear techniques in industry.

REFERENCES
1. Van Wyk, J. M., et al., A study of the macroscopic distribution of oxygen in a steel
rod by neutron activation and vacuum fusion techniques, Analyst 91 (1082),316 (1966).
2. Lenihan, J. M. A., Activation analysis in the contemporary world: questions and
answers, Modern Trends in Activation Analysis, James R. De Voe, ed., National
Bureau of Standards, Washington, 1969 (special publication 312, Vol. 1, pp. 10-12).
3. Fite, L. E., et al., Sampling methodology development on a large scale cystic fibrosis
screening program based on automated neutron activation analysis, Modern Trends
in Activation Analysis, James R. De Voe, ed., National Bureau of Standards, Wash-
ington, 1969 (special publication 312, Vol. 1, pp. 147-155).
4. Wainerdi, R. E., and Menon, M. P., Comparison of the various instrumental methods
for the analysis of traces of bromine pesticides in plant materials, by activation, in
the limit of sensitivity range, Nuclear Activation in the Life Sciences, International
Atomic Anergy Agency, Vienna, 1967, pp. 33-50.
5. Guinn, V. P., The current status of neutron activation analysis applications, Modern
Trends in Activation Analysis, James R. De Voe, ed., National Bureau of Standards,
Washington, 1969 (special publication 312, Vol. 2, pp. 679-697).
6. West, T. S., Atomic flame spectroscopy in trace analysis, Minerals Sci. Engng. 1 (2),
3 (1969); 2 (1), 31 (1970).
7. Wainerdi, R. E., and Dubeau, N. P., Nuclear activation analysis, Science 139 (3559),
1027 (1963).
8. Taylor, D., Neutron Irradiation and Activation Analysis, Van Nostrand, Princeton,
N. J., 1964.
9. Siegbahn, K., Alpha-, Beta-, and Gamma-Ray Spectroscopy, North Holland, Am-
sterdam, 1965.
10. Lam, C. F., and Wainerdi, R. E., A CsI-NaI dual crystal Compton reduction de-
tector, 1965 International Conference on Modern Trends in Activation Analysis,
College Station, Texas, 1965, pp. 189-194.
11. Kantele, J., et al., Gamma spectrometer systems employing an anti-Compton an-
nulus, Nucl. Instrum. Meth. 39 (2), 194 (1966).
11. Nuclear Activation Analysis 347

12. Kuykendall, W. E., and Wainerdi, R. E., The analysis of a complex gamma spectrum
utilizing the IBM 704 computer, Trans. Am. Nucl. Soc., 95 (1960).
13. Fite, L. E., et al., Computer coupled activation analysis, Washington Office of
Technical Services, TEES 2671-1, May 1969.
14. Drew, D. D., et al., Computer coupled activation analysis, Application of Computers
to Nuclear and Radiochemistry, National Academy of Sciences - National Research
Council, Washington, 1962, pp. 237-242.
15. Salmon, L., Analysis of gamma rays scintillation spectra by the method of least
squares, Harwell, Atomic Energy Research Establishment, R-3640, April 1961.
16. Yule, H. P., HEVESY, a computer program for analysis of activation analysis gamma-
ray spectra, Modern Trends in Activation Analysis, James R. De Voe, ed., National
Bureau of Standards, Washington, 1969 (special publication 312, Vol. 2, pp. 1108-
1110).
17. Yule, H. P., Computation of experimental results in activation analysis, Modern
Trends in Activation Analysis, James R. De Voe, ed., National Bureau of Standards,
Washington, 1969 (special publication 312, Vol. 2, pp. 1155-1204).
18. Covell, D. F., Determination gamma-ray abundance directly from the total absorp-
tion peak, Analyt. Chem. 31 (4), 1785 (1959).
19. Wainerdi, R. E., and Fite, L. E., Automated systems for nuclear activation analysis.
Automation and Instrumentation. L. Dadda, and U. Pellegrin, eds., Pergamon Press,
New York, 1967, pp. 557-569.
20. Fite, L. E., et al., A study of the feasibility of using nuclear activation analysis for
lunar compositional analysis, Office of Technical Services, Washington, TID-199999,
May 1963 (NASA).
21. Menon, M. P., et al., Activation analysis for nondestructive localization of impurities
in foils, Nucl. Applic. 2, 335 (1966).
22. Wainerdi, R. E., et al., The use of high speed, computer coupled automatic systems
for non-destructive medical and biological nuclear activation analysis, Proceedings
of the Third International Conference on Biology, Saclay, France, University of France
Press, pp. 171-198.
23. Dawson, E. B., et al., Activation analysis of placental trace metals, J. Nucl. Med.
9 (4), 160 (1968).
24. Bowen, H. J. M., and Gibbons, D., Radioactivation Analysis, Claredon Press,
Oxford, 1963.
25. Yule, H., private communication.
26. Fite, L. E., et al., Special considerations in the large scale analysis of biomedical
samples by automated nuclear activation methods, Proceedings of the Ninth Japan
Radio-Isotopes Conference, Tokyo (in press).
27. Wainerdi, R. E., et al., Neutron activation analysis and high resolution gamma-ray
spectrometry applied to area-elemental distribution studies, Nuclear Techniques and
Mineral Resources, Vienna, International Atomic Energy Agency, 1969, pp. 507-532.
28. Higuchi, H., et al., Use of a Ge(Li) detector after simple chemical group separation
in the activation analysis of rocks samples, Modern Trends in Activation Analysis,
James R. De Voe, ed., National Bureau of Standards, Washington, 1969 (special
publication 312, Vol. 1, pp. 334-338).
29. Santos, G. G., et al., Preliminary study on the use of fast-neutron activation analysis
for sea-floor compositional mapping, Nuclear Techniques and Mineral Resources,
Vienna, International Atomic Energy Agency, 1969, pp. 463-487.
348 L. E. Fite. E. A. Schweikert. R. E. Wainerdi. and E. A. Uken

30. Santos, G. G., and Wainerdi, R. E., Determination of silicon in rocks by fast-neutron
activation analysis using internal standards, J. Radioanalyt. Chem. 1, 509 (1968).
31. Hoffman, B., et al., The 14 MeV neutron activation analysis of deep sea core samples,
Trans. Am. Geophys. Un. 20 (4), 201 (1969).
32. Cuypers, M. Y., and Cuypers, J., Gamma ray spectra and sensitivities for 14 MeV
neutron activation analysis, J. Radioanalyst. Chem. 1, 243 (1968).
33. Cuypers, M. Y., et al., The measurement of the lunar surface composition using
nuclear activation analysis, Proceedings of the XIlIth Congresso Scienti/ica Interna-
zionale per I'Elettronica, Rome, 1966, pp. 17-28.
34. Hislop, J. S., and Wainerdi, R. E., Extraterrestrial nuclear activation analysis,
Analyt. Chem. 39 (2), 29A (1967).
35. Cuypers, M. Y., et al., Lunar and planetary surface analysis using neutron activation,
Radioisotopes for Aerospace, J. C. Dempsey and P. Polish uk, eds., Part 2, Systems
and Applications, Plenum Press, New York, 1966, pp. 292-308.
36. Givens, W. W., et al., Cyclic activation analysis, Modern Trends in Activation Anal-
ysis, James R. De Voe, ed., National Bureau of Standards, Washington, 1969
(special Publication 312, Vol. 2, pp. 929-937).
37. Albert, P., Chimia (Aarau) 21, 32 (1967).
38. Albert, P., Proc. of 2nd Int. Conf. on Practical Aspects of Activation Analysis with
Charged Particles, Euratom Report, Eur 3896 d-f-e, 1, 1967.
39. Engelmann, c., Commissariat a I'Energie Atomique Report, CEA 2559, 1964.
40. Engelmann, c., Commissariat a I'Energie Atomique Report, CEA R-3307, 1967.
41. Baker, C. A., U.K. Atomic Energy Authority Report AERE R-5265, 1967.
42. Baker, C. A., Hunter, G. J., and Wood, D. A., U.K. Atomic Energy Authority
Report R-5547, 1967.
43. Baker, C. A., and Wood, D. A., U.K. Atomic Energy Authority Report AERE
R-5818, 1968.
44. Lutz, G. J., Anal. Chem. 41, 424 (1969).
45. Tilbury, R. S., Activation analysis with charged particles, NAS-NS 3110, USAEC.
46. Lukens, H. R., Jr., Otvos, J. W., and Wagner, C. D., Int. J. Appl. Rad. and Isotopes
11, 30 (1961).
47. Kaminishi, K., Japan J. Appl. Phys. 2, 399 (1963).
48. Goldstein, G., Anal. Chem. 35, 1620 (1963).
49. Rook, H. L., Anal. Chem. 36, 2211 (1964).
50. Guinn, V. P., and Lukens, H. R., Trans. Am. Nucl. Soc. 9, 106 (1966).
51. Faries, R. A., Johnston, J. E., and Miller, R. J., Nucleonics 12 (10), 48 (1954).
52. Khristianov, V. K., Isotopenpraxis 3, 235 (1967).
53. Schweikert, E. A., and Albert, Ph., Radiochemical Methods of Analysis, Int. Atomic
Energy Agency, 1964, Vol. 1, 405 pp.
54. McGregor, M. H., Proc. 2nd Int. Con! on Peaceful Uses of Atomic Energy, 1958,
Vol. 20, p. 1771.
55. Wilkuiss, P. E., and Linnenboom, V. I., Proc. of 2nd Int. Conf. on Practical Aspects
of Activation Analysis with Charged Particles, Euratom Report, Eur 3896 d-f-e,
1967, p. 119.
56. Andersen, G. H., Graber, F. M., Guinn, V. P., Lukens, H. R., and Settle, D. M.,
Nuclear Activation Techniques in the Life Sciences, Int. Atomic Energy Agency,
1967, p. 99.
11. Nuclear Activation Analysis 349

57. Owlya, A., Abdeyazdan, R., and Albert, P., Proc. of 2nd Int. Conf. on Practical
Aspects of Activation Analysis with Charged Particles, Euratom Report, Eur 3896
d-f-e, 1967, p. 161.
58. Baker, C. A., Pratchett, A. G., and Williams, D. R., U.K. Atomic Energy Authority
Report, AERE R-5363, 1967.
59. Engelmann, C., Fritz, B., Gosset, J., Graeff, P., and Loeuillet, M., Proc. of 2nd Int.
Conf. on Practical Aspects of Activation Analysis with Charged Particles, Euratom
Report, Eur 3896 d-f-e, 1967, p. 319.
60. Engelmann, Ch., Gosset, J., Loeuillet, M., Marschal, A., Ossart, P., and Boissier, M.,
Proc. of Int. Con/. on Modern Trends in Activation Analysis, NBS Spec. Publ. 312,
1968, Vol. II, p. 819.
61. Revel, G., Chaudron, T., Debrun, J. L., and Albert, Ph., Proc. of Int. Con/. on
Modern Trends in Activation Analysis, NBS Spec. Publ. 312, 1968, Vol. II, p. 838.
62. Rocco, G. G., Garzon, O. L., and Cali, J. P., Int. J. Appl. Rad. and Isotopes 17,
433 (1966).
63. Berzin, A. K., Kunetsov, K. F., Sulin, V. V., Belov, V. I., Vitozhents, G. Ch., Mar-
tynov, Yu. T., Suslov, V. G., and Shoruikov, S. I., USAEC Report ORNL-tr-l097.
64. Schmitt, R. A., and Loveland, W. D., Geochim. Cosmochim. Acta 33, 375 (1969).
65. Albert, P., Sue, P., and Chaudron, G., Bull. Soc. Chim. France 2016, 97 (1953).
66. Engelmann, c., Acad. Sci. Paris 258, C4279 (1964).
67. Ricci, E., and Hahn, R. L., Anal. Chem. 40, 54 (1968).
68. Sippel, R. F., and Glover, E. D., Nucl. Instr. and Method 9, 37 (1960).
69. Pierce, T. B., Peck, P. F., and Cuff, D. R. A., Analyst 92, 143 (1967).
70. Pierce, T. B., Peck, P. F., Cuff, D. R. A., Nucl. Instr. and Meth. 67, 1 (1969).
71. Turkevich, A. L., Science 134, 672 (1961).
72. Turkevich, A. L., Franzgrote, E. J., and Patterson, J. H., Science 158, 635 (1967).
73. Ibid., Science 160, 1108 (1968).
74. Franzgrote, E. J., Patterson, J. H., and Turkevich, A. L., Science 162, 117 (1968).
75. Turkevich, A. L., Anderson, W. A., Economou, T. E., Franzgrote, E. J., Griffin,
H. E., Grotch, S. L., Patterson, J. H., and Sowinsku, K. P., Jet Propulsion Labo-
ratory, Technical Report 32-1265, 1968.
76. Lamb, J. F., Lee, D. M., and Markowitz, S. S., Proc. of 2nd Int. Conf. on Practical
Aspects of Activation Analysis with Charged Particles, Euratom Report, Eur 3896
d-f-e, 1967, p. 225.
77. Revel, G., and Albert, P., Proc. of 2nd Int. Conf. on Practical Aspects of Activation
Analysis with Charged Particles, Euratom Report, Eur 3896 d-f-e, 1967, p. 261.
78. Debrun, J. L., Barrandon, J. N., and Albert, P., Proc. of Int. Con/. on Modern
Trends in Activation Analysis, NBS Spec. Publ. 312, 1968, Vol. 2, p. 774.
79. Engelmann, c., and Cabane, G., Proc. of Con/. on Modern Trends in Activation
Analysis, Texas A & M University, College Station, Texas, 1965, p. 331.
80. Ricci, E., and Hahn, R. L., Proc. of 2nd Int. Conf. on Practical Aspects of Activa-
tion Analysis with Charged Particles, Euratom Report, Eur 3896 d-f-e, 1967, p. 15.
81. Rook, H. L., and Schweikert, E. A., Anal. Chem. 41, 958 (1969).
82. Fliesher, A. A., Hendsy, C. 0., Smith, C. G., Tom, J. L., and Wells, D. K., Proc.
of 2nd Int. Conf. on Practical Aspects of Activation Analysis with Charged Particles,
Euratom Report, Eur 3896 d-f-e, 1967, p. 435.
83. Schweikert, E. A., Trans. Am. Nucl. Soc. 13, 58 (1970).
350 L. E. Fite. E. A. Schweikert. R. E. Wainerdi. and E. A. Uken

84. Fleisher, M., Summary of New Data on Rock Samples G-l and W-l, 1962-1965,
Geochim. Cosmochim. Acta 29, 1263 (1965).
85. Senftle, F. E., and Hoyte, A. F., Using in situ Neutron Activation, Nucl. Instrum.
Meth. I, 42, 1966, pp. 93-103.
86. Bilibin, Y. A., Metallogenic Provinces and Epochs, Gosgeolteknizdat, 1955, Dept.
of the Secretary of State of Canada, Foreign Languages Division, Bureau of Trans-
lations, No. 58972.
87. Bryner, L., Metallogeny in Russia's Drive for Ore Deposits, Mining Engineering,
59, June 1963.
88. Burnham, C. W., Metallogenic Provinces of the Southwestern United States and
Northern Mexico, N. Mex. Bur. Mines and Mineral Res. Bull., 65 (1959).
89. Rose, A. W., Trace Elements in Sulfide Minerals from the Central District, New
Mexico, and the Bingham District, Utah, Geochim. Cosmochim. Acta 31,547 (1967).
90. Schuiling, R. D., Tin Belts on the Continents Around the Atlantic Ocean, Econ.
Geol. 62, 540 (1967).
91. Uken, E. A., Santos, G. G., and Wainerdi, R. E., Neutron Activation Analysis for
the Study of Metallogenic Provinces, Talanta IS, 1097 (1968).
92. Wainerdi, R. E., Uken, E. A., Santos, G. G., and Yule, H. P., Neutron Activation
Analysis and High Resolution Gamma-Ray Spectrometry Applied to Areal Ele-
mental Distribution Studies, Presented at IAEA Symposium on the Use of Nuclear
Techniques in the Prospecting and Development of Minerals Resources, Buenos
Aires, Argentina, November 4-8, 1968.
93. Gordon, G. E., Randle, K., Goles, G. G., Corliss, J. B., Beeson, M., and Okley, S. S.,
Instrumental Activation Analysis of Standard Rocks with High Resolution Gamma-
Ray Detectors, Geochim. Cosmochim. Acta 32 (4), 369 (1968).
94. Caldwell, R. L., Combination Neutron Experiment for Remote Analysis, Science
152, 457 (1966).
95. Howard, E. V., China Uses Radiation Logging for Studying Dump Leaching Pro-
cesses, Min. Eng. 20 (4), 70 (1968).
96. Anders, O. U., Activation Analysis for Plant Stream Monitoring, Nucleonics 20 (2),
78 (1962).
97. Coleman, R. F., and Perkin, J. L., The Determination of the Oxygen Content of
Beryllium Metal by Activation, Analyst 84, 233 (1959).
98. Anonymous, Employing nuclear methods to automate the grading of ores from
gold deposits, Sov. J. Non-ferr. Metals 5 (8), 20 (1964).
99. Rudanovskii, A. A., The Use of the Method for Induced Radioactivity for Automa-
tion of Rock Separation, Nucl. Sc. Abstr. IS, 15775 (1961).
100. Martin, T. c., Mathur, S. c., and Morgan, 1. L., The Application of Nuclear Tech-
niques in Coal Analysis, USAEC, Report TED-20080, 1963, pp. 1-1-2.
101. Gaudin, A. M., and Ramdohr, H. F., Induced Radioactivity for Course Copper -
Ore Concentration, Can. Metall. Quarterly 1, 173 (1962).
102. Jeffery, P. G., and Bakes, J. M., Neutron Activation Analysis with Neutron from
Isotopic Sources: Analytical Control of Fluorite Production, Warren Spring Lab-
oratory Report, Stevenage, Herts, U.K.
103. Nargolwalla, S., and Jervis, R. E., Continuous analysis of trace fluorine by 14 MeV
neutron activation, Trans. Am. Nucl. Soc. 8, 86 (1965).
104. Uken, E. A., Watterson, J. 1. W., Knight, A., and Steele, T. W., The Application of
Neutron Activation Analysis to Sorting Witwatersrand Gold-Bearing Ores, J.S.A.
[nst. of Mining and Metallurgy 67 (3), 99 (1966).
Chapter 12

MASS SPECTROMETRY

J. N. Weber and P. Deines

Pennsylvania State University
University Park, Pennsylvania

1. INTRODUCTION

The field of mass spectrometry originated with the work of J. J. Thomson

in the early part of this century. He anticipated the use of this tool for
chemical analysis and discussed some of its possible applications in his book
Rays of Positive Electricity and their Application to Chemical Analyses. After
Thomson's discovery that the element neon consisted of isotopes, Demp-
ster and Aston initiated a thorough study of the phenomenon. Aston's
research later resulted in the first precise determination of atomic masses
by mass spectrometry. The field advanced rapidly when the potentialities
of mass spectrometry were recognized by the petroleum and nuclear indus-
tries, and presently the instrumentation is used in a wide spectrum of
industrial and scientific research. Geochemical applications range from the
study of variations in isotopic composition produced by radioactivity, spal-
lation, fission, or physicochemical processes to investigations of chemical
composition by means of thermal-source isotope dilution or rf spark-source
mass spectrometry. In organic geochemistry mass spectrometry is also of
value in structural studies of organic compounds. The instruments used in
such research have been described in detail by Inghram and Hayden,!
Duckworth,2 McDowell,3 Herzog et al.,' and Roboz. 5 In this chapter de-
termination of carbon and oxygen stable-isotope ratios and spark-source
analysis of trace-element concentrations are discussed.

351
352 J. N. Weber and P. Deines

2. STABLE CARBON- AND OXYGEN-ISOTOPE MASS

SPECTROMETRY

2.1. Principles Underlying the Construction of an Iso-

tope-Ratio-Recording Mass Spectrometer

Over 20 years elapsed after Aston and Dempster developed techniques

to measure the relative abundance of stable isotopes before mass spectrom-
eters were produced commercially. Modern instruments have achieved a
high degree of sophistication but the basic principles of design remain little
changed. The mass spectrometer contains three essential parts in addition
to a vacuum system: an ion source, an analyzer, and an ion detector. The
ion source converts neutral molecules into ions, which can then be affected
by electric and magnetic fields. If the sample is gaseous (C02 , for example)
ionization can be effected readily by electron impact, the electrons coming
from a heated filament. The purpose of the analyzer is to separate these
ions into a mass spectrum using a magnetic field or a combination of mag-
netic and electric fields. The ion beams comprising the mass spectrum im-
pinge upon plates or "collectors" in the ion-detection system and the cur-
rents produced are amplified and measured.
For carbon- and oxygen-isotope-ratio analysis the gas most preferred
is CO2 , although C06 and molecular oxygen 7 can also be fed directly into
the mass spectrometer. The use of oxygen has some advantages in special
applications, because to obtain 1SO/160 measurements in natural samples
using CO2 , an assumption is made about the concentration 0[170. In tracer
experiments the relative abundance of 17 0 can be quite large and the for-
mulae 7 derived for use with molecular oxygen are mathematically correct
for all 160_170_1S0 abundance ranges. Samples of geochemical interest
contain carbon and/or oxygen in many different forms of chemical combina-
tion, such as carbonates, water, silicates, oil, etc., but techniques have been
developed to prepare carbon dioxide from virtually any type of starting
material.
Absolute measurements of isotope ratios can be made only with great
difficulty,S whereas for certain elements, differences in isotopic composition
can be determined easily and with high precision. By introducing two gases
alternately into the mass spectrometer over a short period of time, it is
possible to distinguish, for example, between a sample containing 2040.1
ppm ISO and another containing 2040.0 ppm ISO, that is, a difference of
0.1 ppm in the natural range of ISO concentration. The pressure in the ion-
ization region of the ion source is low (,...." I 0-4 mm Hg) and only a small
12. Mass Spectrometry 353

quantity of CO2 is actually admitted to the source. The gas-inlet system

must therefore be designed so as to reduce the pressure between the source
and the sample by several orders of magnitude with a minimum of isotope
fractionation in the process. Furthermore, a device to permit the admission
of one or the other of two gas samples in alternate sequence must be incor-
porated into the inlet system.
Pressure reduction is accomplished by inserting a constriction or leak
between the sample reservoir and the source. The type most commonly
used is a long length of capillary tubing which forms a viscous leak. Isotope
fractionation is minimal if the gas moves by viscous flow, that is, where the
mean free path of the molecules is much smaller than the dimensions of the
leak. As the mean free path is inversely proportional to pressure, it is
desirable to design the leak so that the gas flows through the greater part
of it at a pressure high enough to maintain viscous flow. Kistemaker 9
emphasized the importance of careful design in the construction of gas-inlet
systems to avoid fractionation, and demonstrated how fractionation effects
can be calculated for various combinations of leak dimensions, pressure,
and size of the sample container. Similar calculations for viscous-flow gas-
inlet systems are given by Boerboom and Tasman. lO A period of time
(usually minutes) is required to reach steady-state conditions after the flow
of gas through the capillary leak has begun. In order to compare two gas
samples rapidly, each sample flows continually through its own leak and a
solenoid-operated gas valve is employed at the low-pressure end of the leak
to switch the inflowing gas either into the ion source or into a "waste"
pump. As soon as steady-state flow is attained at the beginning of an anal-
ysis, one sample or the other can be directed into the mass spectrometer
by the solenoid valve, and the change from one gas to the other is almost
instantaneous. In this way, the effects of small and uncontrollable secular
changes in instrumental conditions are essentially eliminated, as these
variations should alter the isotope ratio of both gas samples to the same
extent, leaving the difference unchanged.
Smaller gas samples can be accommodated by using a molecular leak
such as a pinhole in the sample reservoir. Unlike the viscous leak arrange-
ment, the pressure of the gas in the reservoir is low, but unfortunately,
isotope fractionation occurs in molecular flow. This is not a random error
but a regular change with time, and the fractionation effects can be largely
compensated for by automatic inlet valves which permit both gases to flow
through their respective leaks for equal periods of time. Shackletonl l reports
a reproducibility of 0.0l %0 for such a system, using the CO2from CaCOa
samples weighing as little as 0.4 mg.
354 J. N. Weber and P. Deines

The major disadvantage of the viscous leak is that samples larger than
a micromole are necessary, but this can be overcome by adding an inert
gas such as nitrogen or helium to very small CO2 samples. In this way, the
mean-free path of the gas molecules through the capillary leak is smaller
because the total gas pressure is high. Han and Fritz12 analyzed sub micro-
mole samples of CO2 by this technique and reported no detectable isotope
fractionation in the leaks with partial pressures of CO2 as low as 0.1 fl.
Most modern CO2 mass spectrometers are equipped with a two-channel
(dual) gas-inlet system and capillary viscous leaks. The sample reservoirs
are fitted with mercury pistons which permit accurate and rapid adjustment
of the gas pressure, and automatic recycling timers can be used to activate
the solenoid gas-switching valve at a preselected time interval. Different
gas-inlet systems have been described,13-15 including a three-channel de-
sign. 16
A heated filament in the ion source emits electrons which are acceler-
ated by an electric field (40-100 V) and then collimated by a weak mag-
netic field. These electrons, whose emission is carefully regulated, collide
with the CO2 molecules of the sample, forming positive ions. The latter are
drawn out of the ionizing region by an electric field and accelerated to form
a beam of ions traveling along the analyzer tube between the poles of a
large magnet. The charged particles are forced into circular paths whose
radius of curvature is a function of m/e (mass divided by charge) for a given
magnetic field strength. Ion collectors are appropriately located to intercept
the ion beams of interest. For a constant magnetic field strength the mass
of the ions collected is inversely proportional to the accelerating potential.
For natural samples, CO2 ions of mass 44 (12C160 16 0) to mass 49
(13C180 180) are formed in the mass spectrometer but it is not necessary
to collect and measure the intensity of each of the ion beams in this mass
range in order to obtain 13Cp2C and 180P60 ratios. The ion collector is
constructed with two Faraday cups called the "single collector" and the
"multiple collector." Their design is such that for an appropriate value of
the accelerating potential, the output of the single collector can be attri-
buted to the mass-45 ion beam while that of the multiple collector derives
from the mass-44 ion beam. The potential drop across a high resistance in
the single and multiple collector circuits is amplified by a vibrating-reed
electrometer and a dc amplifier, respectively. A decade-resistor potentiom-
eter is connected with the two collectors so that with proper selection of
the decade resistors, no current flows through the circuit (Fig. 1). By this
"null" method, the ratio of the number of mass-45 ions collected to the
number of mass-44 ions collected is obtained from the value of the resis-
12. Mass Spectrometry 355

maSI

mals
.s

Fig. I. Ion-collector circuitry for determining

the ratio of mass-45 to mass-44 ion beams.
DCA, dc amplifier; VRE, vibrating-reed elec-
trometer; REC, recorder; R, = 1.5 X 1011 Q;
R2 = 1.5 X 10'0 Q; Ra = decade resistance
potentiometer. (After McKinney et al.'")

Fig. 2. Strip-chart record obtained from the re-

corder shown in Fig. I. Traces for sample and
standard gases alternate at about 1.5-min inter-
vals. The displacement of the sample trace from
the standard trace corresponds to a difference
in isotopic composition of about 0.25 permil.
356 J. N. Weber and P. Deines

tance inserted in the circuit. At a different accelerating potential, the mass-

46 ion beam enters the single collector while ions of masses 44 and 45 strike
the multiple collector.
The small residual current which cannot be reduced to zero by the
decade resistors is displayed by a strip-chart recorder (Fig. 2). From this
information and the values of the decade resistors, the following ratios are
determined: (a) number of mass-45 ions to number of mass-44 ions, and
(b) number of mass-46 ions to number of mass-44 and mass-45 ions. A
number of instrumental correction factors17- 19 are determined and applied
to the analytical data. Isotope ratios in delta notation are then calculated
from the two ion ratios using equations derived by Craig,17 More detailed
descriptions of isotope-ratio-recording mass spectrometers have been pub-
lished.13 ,14 ,20-24

2.2. Reporting of Carbon- and Oxygen-Isotope Data

Determination of the absolute isotope ratio is an exceedingly difficult
task, but it is possible to measure differences in isotope ratio between two
samples with a high degree of accuracy and precision. For this reason, iso-
tope data are reported in delta notation as deviations from an arbitrarily
selected reference (standard):

!513C = ( 1aCf12Csample _
lac /12Cstandard
1)
Usually the difference is expressed in parts per thousand (Ofoo) and the
expression on the right-hand side of the equation is then multiplied by 1000.
The rapidly expanding application of stable isotopes to geochemical
problems has created some difficulties and confusion in the area of isotope
standards. Early investigators in different laboratories selected their own
standards or reference samples. The form of the standard chosen, such as
water, carbonate, silica, etc., depended upon the type of materials to be
analyzed. While this situation was adequate for a number of years, in-
creased precision and accuracy resulting from developments in analytical
techniques and instrumentation could not be fully realized in comparing
different types of samples or data obtained in different laboratories unless
certain criteria for adequate standards were fulfilled. A great many standards,
such as "Moscow tap water," "Stockholm BaCOa," or even tank CO2,
have been used to report isotope data, but most of the carbon- and oxygen-
isotope-ratio measurements published today are related to two important
"reporting standards."
12. Mass Spectrometry 357

Isotope standards should serve two major purposes and permit inves-
tigators to report their measurements with respect to the same reference
point so that isotope-ratio data obtained in different laboratories or for
different types of materials can be compared, and calibrate their mass
spectrometers and determine instrumental correction factors. Theoretically,
only one standard would be necessary to serve the first purpose, but in
practice it is desirable to have at least two so that a standard with an iso-
topic composition not too far removed from those of the samples can be
chosen. For the second purpose, i.e., calibration of a mass spectrometer,
it is necessary to have at least two standards.
The criteria for an adequate isotope standard are: (a) the isotopic
composition must not change with time or be subject to alteration during
any handling or sample-preparation procedures; (b) the isotopic composi-
tion should be within the range of compositions of the analyzed samples;
and (c) the standard must be readily available to all investigators and each
aliquot of standard must have exactly the same isotopic composition. Most
of the standards used by earlier workers have not met the third criterion and
these standards either no longer exist or are available in such limited quan-
tity that they cannot be distributed.
The u.s. National Bureau of Standards established a program 25 to
distribute a stock of isotope reference samples and to serve as a clearing-
house for data concerning these "standards." Included in the isotope ref-
erence samples are water, air, CaC03 , C, and Si02 26 The two widely used
"reporting standards" are PDB-I and SMOW, neither of which is actually
available. However, the differences in isotopic composition between these
reporting standards and reference samples that can be obtained by an in-
vestigator are known. Thus it is possible to use one or more of the reference
samples as a measuring or "working standard" (i.e., for determining the
difference in isotope ratio between a sample and the measuring or working
standard in a mass spectrometer) and then mathematically convert the
result to a delta value expressing the difference in isotope ratio between the
sample and the reporting standard. For example, NBS-20 can be used as a
"working standard" in the laboratory, but for the purpose of publication,
the measurements can be converted mathematically so that the isotopic
data are referred to the PDB-I or SMOW standards.
The PDB-I standard is CO 2 prepared from PDB CaC03 by reaction
with 100% H3P0 4 at 2S.2C. The standard is not the carbonate itself, which
is a particular sample of a belemnite rostrum (Belemnitella americana)
collected from the Cretaceous Peedee formation of South Carolina. A large
quantity of data, especially in the isotopic paleotemperature field, has been
358 J. N. Weber and P. Deines

published using PDB CO 2 as the standard. For this reason, many investi-
gators prefer to continue the use of PDB as a reporting standard even
though the standard itself can no longer be distributed.
SMOW, or "standard mean ocean water," was proposed by Craig 27 as
a standard for both 180/16 0 and hydrogen-deuterium measurements. Al-
though most of the isotope-ratio data for carbonates was referred to PDB
at this time, most of the 180/16 0 determinations for water were expressed
with respect to a sea-water standard. Use of the latter was advantageous for
two reasons: (a) it provided a reference standard whose isotopic composi-
tion was similar to most of the analyzed samples, and (b) the preferred
analytical technique for water then involved the equilibration of CO2 (the
gas to be fed into the mass spectrometer) with water at a known tempera-
ture. If a standard water sample were treated in exactly the same manner,
differences in the isotopic composition of the CO2 equilibrated with standard
and sample waters would be identical to differences in the isotopic composi-
tion of the sample and standard waters; it is therefore not necessary to know
the fractionation factor defining the distribution of oxygen isotopes be-
tween water and CO 2 in equilibrium at a given temperature.
For the isotopic analysis of water some investigators had been using
an "average-ocean-water" standard. 28 Unfortunately, no single sample of
"average ocean water" which could be circulated widely actually existed.
An NBS water sample (No.1) is, however, readily available for distribution,
but it differs appreciably from sea water in isotopic composition. The
problem was solved by creating a hypothetical standard suitable for report-
ing sea-water analyses but which is defined in terms of NBS-I. This hypo-
thetical reporting standard, SMOW, is related to NBS-I by the equation: 27
18 0/160 (SMOW) = 1.008 180 P6 0 (NBS-I), and is very close to the "average

ocean water" of Epstein and Mayeda. 28 Using delta notation, the 15180
value of NBS-I with respect to SMOW is -7.94%0' Attention should be
drawn to a minor oversight 27 where the statement is made that the 15180 of
PDB-I (i.e., CO2 from PDB carbonate obtained by reaction with anhy-
drous H 3P0 4 at 25.2C) is +0.22%0 on the SMOW scale. The difference
(+0.22%0) referred to by Craig is between PDB CO2 and CO2 in isotopic
equilibrium with SMOW at 25C, not between PDB CO2 and SMOW.
Many of the longer established laboratories use secondary standards
for the actual mass-spectrometric measurements and have calibrated their
working standards against PDB-I when PDB was still available for this
purpose. A number of these standards have been compared. 17 29 NBS Iso-
tope Reference Samples can be used as working standards by those who do
not have adequately calibrated working standards. In many cases the work-
12. Mass Spectrometry 359

ing standards are for the laboratory's internal use only and are never
referred to in publications. Establishing new working standards by calibra-
tion against some other working standard which in turn may have been
calibrated by still other standards is not a recommended practice, nor are
other complicated procedures 30 used by some to relate their measurements
to the POB standard.
Isotope geochemists are not completely satisfied with all of the isotope
reference samples presently available. The problems involving isotope
standards were discussed at both the Third International Conference on
Nuclear Geology (Spoleto, Italy, 1965) and the Fifth International Con-
ference on Stable Isotopes (Leipzig, DDR, 1967). There was unanimous
agreement on the need for additional standards. A new calcite standard
closer to POB in isotopic composition than is NBS-20 was proposed to
eventually supplant NBS-20, and a new ocean-water standard to comple-
ment NBS-l has also been suggested. In addition, there is a need for two
CO 2 samples about 40%0 different in isotopic composition to permit the
calibration of mass spectrometers. The International Atomic Energy Agency
(IAEA) has distributed three water samples to each of 12 laboratories to
determine the extent of systematic analytical errors and unreliable working
standards. 31 The IAEA is establishing two new water standards for the
isotopic analysis of both hydrogen and oxygen.
Isotope-ratio data expressed relative to one standard can be converted
so that the same sample data are expressed relative to another standard.
b values with respect to standard S, for example, can be recalculated to give
b values relative to POB by using the equation

where bU:- PDBi is the isotopic composition of the sample x relative to the
POB standard, b(x-S) is the isotopic composition of the sample relative to
a standard S, and b(S-PDB) is the isotopic composition of standard S relative
to the POB standard, all 15 values being expressed in %0' 15 values for various
standards relative to POB have been reported by CraigP

2.3. Sample-Preparation Procedures

2.3.1. Water
It is undesirable to introduce water directly into the mass spectrometer. 32
An oxygen-containing gas which is very convenient for mass-spectrometric
isotopic analysis of oxygen is CO2 The problem then is to relate the 180/ 160
360 J. N. Weber and P. Deines

ratio of a water sample to the 180/ 16 0 ratio of a COz sample so that the
analysis of the latter will provide a measurement of the 180 content of the
water. A variety of techniques to accomplish this have been developed and
it is possible to obtain high-precision isotopic analyses of water samples
ranging from many liters to a fraction of a ,ul in size. These techniques can
be divided into two categories: (a) techniques that yield Oz, CO, or COz by
chemical reaction including the subsequent conversion of Oz and CO to
COz, and (b) techniques that involve isotopic equilibration of water and COz.
Procedures of the first type should produce yields as close to 100% as
possible, as isotope fractionation generally occurs. Methods such as elec-
trolysis are thus excluded. COz is evolved directly by heating water with
guanidine-HCl and good results have been obtained with water samples
as small as 0.1 ,ul,33.34 Also suitable for very small samples is a procedure
by which HzO is reduced to yield CO and Hz on a graphite rod heated to
1100-1200C.35 Hydrogen is removed from the system by diffusion through
a palladium tube and the CO remaining is converted to COz on a nickel
catalyst. Water vapor can also be reduced by a mixture of Fe and C at
500C.36 Iron oxide and Hz are formed without producing a significant
amount of CO. The iron oxide is later reduced by carbon at 1000C and the
resultant CO is catalytically converted to COz over nickel. With this method,
samples of water as small as 10 mg may be analyzed.
Molecular oxygen may be liberated from water by a variety of chemical
reactions. 37 The use of bromine pentafluoride for this purpose provides many
advantages,38 among which are 100% yields, good reproducibility (O.l 0/00)'
lack of a measurable memory effect, and rapidity (15-20 min per sample).
The reaction BrF5 + HzO -+ BrF3 + 2HF + toz takes place explosively
in a nickel vessel, but yields are erratic at room temperature. Between 80
and 100C, however, the reproducibility is greatly improved. A cold trap
cooled by liquid nitrogen removes all of the reaction products except Oz,
which is then quantitatively converted to COz over a heated carbon rod.
Equilibration methods are based on the principle that the ratio,
a = (180/ 160)Aj(l 80p 60)B is a constant for two phases A and B in isotopic
equilibrium at a given temperature. If the quantities of A and B and the
fractionation factor a are known, the isotopic composition of one phase
can be calculated from mass-balance considerations if the isotopic composi-
tion of the other phase has been measured before and after equilibration.
COz is most widely used for this purpose although for solid-source mass
spectrometry, compounds such as KZC03and KMn0 4 have been utilized. 39 .40
A considerable period of time, usually days, is required for isotopic equi-
librium to be established between HzO and COz at room temperature. In the
12. Mass Spectrometry 361

procedure reported by Epstein and Mayeda,28 the water samples are frozen
and the air is quickly pumped away. Approximately 175 cc (STP) of CO2
are added to 25 g of water, the pH is adjusted to 6 or lower to facilitate
equilibration, and the containers are placed in a water bath maintained at
25.3C. Aliquots of gas are withdrawn at 2-day intervals and analyzed
until a constant value 0[180/ 16 0 indicates that isotopic equilibrium has been
obtained.
If a "standard" water is treated exactly as the water samples, the delta
value calculated from the 18 0/ 16 0 ratio of the CO 2 equilibrated with standard
water and the 180/ 16 0 ratio of the CO 2 equilibrated with a water sample
will be the same as the delta value expressing the difference in the 180/ 16 0
ratio of the two water samples themselves. It is thus unnecessary to deter-
mine the fractionation factor, a = (l80jl 60)C0 2 /(l80jl 60)H 2 0' In order to
compare the oxygen-isotope ratios of water with those of oxides, silicates,
carbonates, etc., the value of a must, however, be known. It is not possible
to determine the difference in the 18 0/ 16 0 ratio of a sample of water and the
180 jl60 ratio of a sample of quartz by a direct mass-spectrometric compari-
son of CO 2 containing all of the oxygen originally present in the sample of
Si02 with the CO 2 which has been equilibrated with the sample of water,
unless a is known. Experimentally determined values of a at 25C are
1.0407,36 1.0424,41 1.0417,35 and 1.04073. 38 The value of a also depends on
the 13C/12C ratio of the CO 2 and the H/D ratio of the water, but variations
in carbon- and hydrogen-isotope ratios in natural samples are too small to
affect the value of a used for practical purposes. The oxygen isotope frac-
tionation factor at 25C has been calculated by statistical mechanics for the
various isotopic end members in the CO 2-H 20 system,42 and for the system
CaC03-H 20-C02. 43
The major disadvantage of the CO 2-HP isotopic equilibration method
is the length of time required to achieve isotopic equilibrium at room
temperature. A number of attempts to increase the rate of equilibration by
using higher temperatures have been reported,44,45 but comparative studies
have shown that greater reproducibility is obtained at lower temperatures. 46

2.3.2. Carbonates
CO2 can be prepared from carbonates in three different ways: (a) ther-
mal decomposition, (b) acid decomposition, and (c) liberation of oxygen
which is subsequently converted to CO 2, In thermal decomposition two of
every three oxygen atoms form CO 2 while the third combines with the cation
or cations to form an oxide. If the reaction is carried to completion the
362 J. N. Weber and P. Deines

carbon originally present as carbonate is quantitatively converted to CO2 ,

hence any fractionation of carbon isotopes during decomposition is of no
consequence. In the case of oxygen, however, fractionation effects are im-
portant as the 180 Jl 6 0 ratio of the two products, CO2 and oxide, is appre-
ciably different. If the fractionation of oxygen isotopes between CO2 and
oxide were highly reproducible, thermal decomposition would -serve as a
rapid and simple method of obtaining CO2 for both carbon- and oxygen-
isotope analysis. McCrea47 demonstrated, at least for calcite, that the kinetics
of thermal decomposition are such that CO2 with desired reproducibility of
isotopic composition cannot be obtained. Although thermal decomposition
of carbonates is used,48,49 decomposition by phosphoric acid or BrFs yields
better results.
As in the case of thermal decomposition the reaction of acids with
carbonates liberates only two-thirds of the total oxygen of the carbonate,
but for certain acid reactions the fractionation of oxygen isotopes is highly
reproducible. McCrea47 investigated a number of different acids and ob-
tained the best results with anhydrous H3P04 In this method powdered
carbonate (generally between 80- and 200-mesh grain size) and H3P0 4 are
mixed under vacuum and permitted to react in a closed vessel until the
evolution of CO2 appears to cease. The configuration of the reaction vessel
is such that acid and sample remain separate until the vessel is evacuated
and closed. A convenient design is shown in Fig. 3. The acid is placed in the
side arm of the reaction vessel and can be admixed with the carbonate by
tipping the vessel on its side. After the reaction appears to be complete,
the CO2 and other volatile phases, which include water and acid vapor, are
frozen in a trap cooled by liquid nitrogen and any vapors that are non-
condensable at this temperature are rapidly pumped away. The trap is then
brought to the temperature of a mixture of dry ice--chloroform-carbon
tetrachloride in order to release the CO2 while retaining water.
The fractionation factor for oxygen in the phosphoric acid reaction is
a = (180j160)co./(180Jl60)carbonate, where (180j1 6 0)co. refers to the carbon
dioxide liberated from the carbonate by H3P04 The value of a is dependent
upon the temperature of the reaction and the type of carbonate, but the
time of reaction for a monomineralic carbonate sample and the ratio of
sample to excess acid have no measurable effect. so Reactions for Ca, Sr,
Ba, and Pb carbonates are essentially complete in 1 day, but 2-3 weeks are
required for dolomite and for the carbonates of Zn, Cd, and Mn. Sl Complete
reaction, however, is not necessary as no variations in 13C/12C or 180/160
with time have been observed. Sl The relationship between a and tempera-
ture for calcite over the range 15-40C is a = -4.1 X 1O-6 t + 1.0111,
12. Mass Spectrometry 363

Fig. 3. Vacuum system for the preparation of CO 2 gas samples from carbonates by
reaction with anhydrous phosphoric acid. Reaction vessels with side arms for H 3 PO. are
shown above a series of sample tubes used to transfer CO 2 from the reaction system to
the mass spectrometer.

where t is the temperature in 0c. 50 The fractionation factor is different for

different carbonates, and this fact must be taken into account when the
180jI60 of two different carbonates, each prepared for analysis by the
H 3P0 4 -reaction method, are compared. Values of a at 25C have been
determined for the carbonates of Zn, Pb, Sr, Ba, Cd, and Mn, in addition
to calcite, aragonite, and dolomite.51
The duration of the reaction time is important when the sample contains
two or more carbonate minerals of different isotopic composition. Some
carbonates react much more slowly than others, and both the carbon- and
oxygen-isotope ratios of the liberated CO 2 will change during the reaction.
In some cases, advantage may be taken of the differential reaction rates to
364 J. N. Weber and P. Deines

determine the isotopic compositions of the individual carbonates comprising

a mixture. Some carbonate rocks contain both dolomite and calcite so fine-
grained and intimately admixed that physical separation is precluded. As
dolomite reacts more slowly than calcite with HaP0 4 , the CO2 released at
the beginning of the reaction is almost entirely from the calcite, whereas
that evolved near completion of the reaction is almost entirely derived from
the dolomite. Good results have been obtained for calcite-dolomite mixtures,
about - 200 mesh in particle size, prepared from calcite and dolomite each
of known isotopic composition. 52 The CO2 extracted after a reaction time
of 1 h represents a 100% yield from calcite but only a 5% yield from the
dolomite.
The reproducibility of 15 180 measurements is greatly improved if bio-
genic carbonate samples are heated in flowing helium prior to reaction with
phosphoric acid. 53 Numerous contaminants will contribute to ion beams
between masses 43 and 47, and the heating process apparently removes such
organic impurities. The measured 180 content is usually lower in samples
that have been heated. Twenty to 40 min at 420-430 o C is generally suffi-
cient. 54 A vacuum may be substituted for the helium atmosphere. 55
Oxygen can be extracted quantitatively from carbonates by carbon
reduction or by reaction with fluorine or halogen fluorides such as BrFa or
BrF5' 51,56 At 700C the decomposition of calcite by BrF5 is represented by
3BrF5 + CaCOa ---+ CaF2 + 3BrFa + CF 4 + i02' This method is especially
useful for carbonates that react very slowly with HaP0 4 . Carbon reduction
and halogen fluoride extraction are described in more detail in the following
section.

2.3.3. Silicates, Oxides, Phosphates, Sulfates, etc.

Three types of methods can be used to extract oxygen from a va-
riety of inorganic compounds: (a) reaction with agents such as mercuric
cyanide or guanidine-HCI to yield CO2; (b) reduction by carbon at high
temperatures to yield CO; and (c) oxidation by fluorine or halogen
fluorides to yield O2, For most minerals the latter method offers more
advantages.
Anbar and Guttmann 57 described techniques by which CO2 is formed
when various inorganic compounds are heated at 400C with a mixture of
mercuric chloride and mercuric cyanide. They reported good results for
oxides of Fe, Co, Cu, Zn, Pb, Hg, B, Ba, U, Ag, and Mn, and for some
phosphates, sulfates, and arsenates, but the method is unsuitable for sili-
cates and for the oxides of Al and Cr. AgCN can also be used in this way.58
12. Mass Spectrometry 365

With guanidine-HCI, oxygen can be extracted from orthophosphate samples

as small as 1-2 mg. aa ,a4,4S
The carbon-reduction method s9- 61 has been used to convert the oxygen
of silicates, oxides, and some carbonates to CO, but recent studies have
demonstrated the superiority of halogen fluorides, especially BrFs, for this
purpose. To produce CO the oxygen-bearing sample is intimately mixed
with finely powdered graphite and reacted at a temperature in the range
l600-2000C. The major disadvantage of carbon reduction is that complete
reaction and 100% yields are sometimes difficult to obtain, resulting in con-
siderable isotopic fractionation for some minerals. A modified version of
carbon reduction has been successfully applied to the isotopic analysis of
oxygen in sulfates. 62 ,6a
The reaction of fluorine or chlorine trifluoride with silicates is illustrated
by the following examples :64

3Mg2Si04 + SClFa ~ 6MgF2 + 3SiF4 + 4Cl2 + 602

Mg2Si04 + 4F2 ~ 2MgF2 + SiF4 + 202

These oxidizing agents are hazardous, however, and the oxygen yields for
some minerals are below 100%. Better results can be obtained with bromine
pentafluoride, which at room temperature is a colorless liquid. As BrFs
reacts readily with glass, a metal vacuum system of monel or nickel is
required. In the procedure developed by Clayton and Mayeda,6S commer-
cially available BrFs is distilled from a steel cylinder and transferred to a
nickel reaction vessel containing between 5 and 30 mg of the mineral to be
analyzed. The quantity of BrFs used is approximately 5 X the stoichiometric
requirement.
The reaction between BrFs and orthoclase is represented by

KA1SiaOg + SBrFs ~ KF + AIFa + 3SiF + SBrFa + 402

4

For a l2-h period of reaction, Clayton and Mayeda6S recommend a temper-

ature of 450C for quartz, micas, and feldspars; 600C for magnetite,
hematite, and ilmenite; and 690C for garnet and olivine. A trap cooled by
liquid nitrogen retains all of the volatile reaction products except oxygen.
The latter is transferred by means of a Toepler pump into a calibrated
volume to determine the yield and is subsequently passed over a carbon
rod heated to 550-600C to obtain CO2, In a method specifically developed
for the oxygen-isotope analysis of orthophosphate the sample is first pre-
cipitated as BiP04 and then reacted with BrFa.66
366 J. N. Weber and P. Deines

2.3.4. Carbon, Organic and Biologic Compounds, Atmospheric

CO 2 , and Volcanic Gases

The preparation of CO2 from carbon, coal, petroleum, organic matter,

and hydrocarbon natural gases for carbon-isotope analysis is a simple
procedure in which all of the carbon of the sample is oxidized to CO2, 67-69
Samples are oxidized at 800-900C over CuO in the presence of oxygen,
and if the gases are recycled through a trap cooled by liquid nitrogen, 100%
yields in the form of CO2 can be attained. Volatile impurities are separated
from the CO2 by one or more distillations using a trap cooled by a mixture
of dry ice, carbon tetrachloride, and chloroform, and also by passing the
gas over manganese dioxide and copper at 500C. This method is useful
for a wide variety of carbon-containing compounds such as organic matter
in sediments,70 natural gases,71 hydrocarbons,72 and biogenic organic
compounds. 73
Carbon dioxide can be prepared from organic matter by heating the
sample at fairly low temperatures with a strong oxidizing agent. A mixture
of chromium trioxide, potassium iodate, phosphoric acid, and fuming sul-
furic acid is capable of oxidizing many carbon-containing substances in a
few minutes at 150C.74
Isotopic analysis of the oxygen of organic compounds is more difficult
than for carbon. The 180jI60 ratio of some organic compounds can be
measured by directly introducing the sample into the mass spectrometer
without any chemical conversion,75.76 but both sensitivity and precision are
insufficient for the determination of natural variations in 180 abundance.
The method can be used, however, in biogeochemical tracer experiments
provided tracer enrichment is 5% or more. Greater sensitivity is obtained
by chemically converting the oxygen of organic compounds into CO2,
but yields close to 100% are required because of isotope fractionation during
the conversion process. Rittenberg and Ponticorvo 77 oxidized oxygen-con-
taining organic compounds by heating the sample with HgCl2at 360-530C.
Isotopic analysis of the CO2 produced by this reaction yields good results
for amino acids and a variety of other organic compounds. Gas-liquid
chromatography can be used to purify the CO2, 78
Atmospheric carbon dioxide can be absorbed in basic solutions or
extracted from the air by freezing with liquid nitrogen. As large isotopic
fractionation occurs with the first method, condensation by cooling is more
desirable. Keeling 79 .8o used evacuated 5-liter flasks to collect air samples
in the field. In the laboratory CO2 was extracted from these samples by
passing the air through a trap cooled with liquid nitrogen. Water is easily
12. Mass Spectrometry 367

removed by fractional distillation at dry-ice temperature. This procedure,

however, does not separate atmospheric N 20 from the CO2 , These two
gases have almost identical physical properties, induding similar vapor
pressure; N 20 follows CO 2 through the purification stages and enters the
mass spectrometer as a contaminant having molecular masses in the same
range as CO2 , Although the abundance of N 20 in the atmosphere over the
continents is only about 0.5 ppm by volume, i.e., about 1.60 / 00 of atmospheric
CO2 , the effect of N 20 on the measured b13C value should be corrected for. 81
Evacuated gas-sample bulbs or flasks are also used to collect other natural
gases such as those emanating from volcanos and fumaroles. 82

2.4. Geochemical Applications

Isotopes differ in both chemical and physical properties. Although

these differences are small, for carbon and oxygen they are sufficiently
large to give rise to measurable "isotope effects" in many geologic systems.
The behavior of two isotopes in a chemical reaction, which may be part
of a geologic process, is not exactly the same, and by measuring natural
variations in isotope-abundance ratios, information may be obtained to
either elucidate the process or determine conditions under which the process
operated. Equilibrium and kinetic isotope effects depend on the relationship
between isotopic mass and chemical bonding. If, for example, a heavy
isotope, 180, is substituted into a molecule of oxygen, 160 2, at ordinary
temperatures, the vibrational frequency between the two atoms is decreased
and the bond is strengthened. More energy is therefore required to dissociate
160 180 than 16 0 160. In addition to chemical reactions isotopes can be frac-
tionated by physical processes such as diffusion, evaporation, etc. HrO
and Hl80, for example, have different physical properties, induding spe-
cific gravity and vapor pressure. In Thode's83 words, stable isotopes are a
"key to our understanding of natural processes." A summary of isotope
effects, induding their quantitative treatment in terms of equilibrium con-
stants and free-energy changes in isotope-exchange reactions, has been
published by Bigeleisen. 84
The types of problems to which carbon and oxygen stable isotopes have
been applied cover almost every aspect of geochemistry. The literature of
this field up to 1968 comprises at least 1800 papers and publications. 85 In
this summary only a few papers can be cited to exemplify the different
applications to geochemistry.
368 J. N. Weber and P. Deines

2.4.1. Ore Deposits and High- Temperature Isotope Thermo-

metry

By means of stable-isotope ratios the mechanisms by which native

sulfur deposits form in salt domes have been elucidated. 86 Isotopes can be
used as a guide to the location of mineral deposits as well as to reveal their
origin. 87 - 92 From the fractionation of oxygen isotopes between two or more
coexisting minerals such as quartz, calcite, magnetite, etc., it is often possible
to determine the temperature at which the mineral deposit was formed. 93- 98
In certain cases the isotopic composition of the hydrothermal fluid can be
estimated.

2.4.2. Petroleum, Coal, and Natural Gas

Carbon-isotope data have yielded much information on the fascinating
problems concerned with the origin of petroleum and the mechanisms by
which oil evolves,799-101 and the technique may be used in the search for
carbonate reservoir rockS. 102 .103 13C/12C ratios of natural gases, coal, other
fossil fuels, and organic matter in sediments have been extensively
studied.104-113

2.4.3. Biogeochemistry
Abelson and Hoering114 .115 showed how carbon-isotope-ratio measure-
ments are useful in various biogeochemical investigations. One of the more
interesting aspects involves the interpretation of the isotopic composition
of carbon found in highly metamorphosed Precambrian rocks, implying the
existence of very ancient life forms.u 6- 132 Assigning a biologic or nonbiologic
origin to ancient graphites solely on the basis of carbon-isotope data has,
however, been criticized.133.134

2.4.4. Oceanography, Limnology, Glaciology, Hydrology, and

Meteorology
The hydrologic cycle is analogous to a large and complex fractional
distillation column. Because of the effect of 180/16 0 on the vapor pressure
of water, considerable fractionation of oxygen isotopes occurs in meteoro-
logical and related processes. Isotope-ratio measurements can thus be used
to help elucidate processes involving the oceans28.135-141 and lakes.142-144
As the 180 content of meteoric precipitation depends on such variables as
temperature, latitude, altitude, and origin of the air mass and its previous
12. Mass Spectrometry 369

history, 180jI60 measurements on snow, tim, and glacial ice can often
provide information about past climatic conditions and the history of
glaciers.145-160 Other hydrological applications involve ground and forma-
tional waters,161-166 and hot springs and geysers.167-174

2.4.5. Paleoenvironmental Analysis, Paleotemperatures, Sedi-

mentation, and Diagenesis
Under some circumstances, 13CjI2C and/or 180/16 0 ratios of dissolved
bicarbonate are directly or indirectly related to the salinity of the water,
or to the nature of the environment in which carbonate sediments are depos-
ited. Isotope studies of modern carbonates, including the skeletons of sedi-
ment-contributing organisms, have made it possible to interpret the isotopil-
composition of ancient carbonate rocks in terms of the depositional envi-
ronment. 175 - 189 Paleotemperature analysis is based on the fact that the
fractionation of oxygen isotopes between two phases is temperature de-
pendent. In-the method established by Epstein et al.,53.54 the two phases are
biogenic calcite (fossil invertebrate skeletons) and sea water. Long-term
trends in climates of the past have been revealed by paleotemperature deter-
minations.190-195 The interpretation of some of this work, however, has been
questioned.196-199 Longinelli2oo.201 is investigating the possibility of using
biogenic phosphate in sedimentary rocks for isotopic paleotemperature
measurements. The abundance of dolomite in ancient sedimentary rocks
has never been adequately accounted for, and numerous isotope studies,
both of natural occurrences and by hydrothermal synthesis, have been direc-
ted toward tris problem.52.202-213 Evaporite deposition has also received
considerable attention.214-218 Carbon and oxygen isotopes can provide infor-
mation concerning the diagenesis of sediments.219-224

2.4.6. Magmatic Processes and Metamorphism

Isotope-fractionation effects are smaller at higher temperatures but
they can be measured and used to elucidate the nature of magmatic pro-
cesses.225-229 Of particular interest are the origin of carbonatites,230-233 the
sources of carbon in diamonds,234 and the nature of metamorphic
processes.235-242

2.4.7. Meteorites and Tektites

A number of studies are described in the literature.243-252
370 J. N. Weber and P. Deines

2.4.8. Miscellaneous
l3C/12C analysis can be used to "correct" radiocarbon age determina-
tions for possible fractionation effects.253.254

2.5. Future Developments of Stable-Isotope Mass

Spectrometry
Stable-isotope mass spectrometry has advanced to the stage where
further improvement may be limited by sample-preparation techniques
rather than instrument performance. Unless great care is taken it is possible
to alter the isotopic composition of the sample, either during the preparation
and purification procedures, or by isotope fractionation during gas handling
and transfer, to the extent that the differences to be measured are smaller
than the variations introduced. This is especially true of very small samples
where contamination and isotope fractionation by adsorption on surfaces of
the sample reservoir, etc., become more important. 2s5
In the future the laborious task of manually recording the many
measurements made during analysis will likely be taken over by an analog-
to-digital converter system which will prepare punched cards, punched tape,
or magnetic tape records for computer processing. A recorder chart-to-
punched card converter has already been developed for this purpose. 256
McCullough and Krouse 257 incorporated a five-figure integrating voltmeter-
ratiometer and a digital printer in the measuring circuit of a conventional,
simultaneous-ion-collection mass spectrometer. They reported increased
precision in addition to ease of measurement, as the device eliminates
human measurement errors involved in drawing lines through null balance
traces on a recorder chart, and averaging of the signals may be accomplished
electronically. Other data-processing techniques have also been pro-
posed.258-260
Improvements are expected in the use of isotope standards, especially
in the availability of adequate standards for mass-spectrometer calibration.

3. TRACE-ELEMENT-CONCENTRATION
DETERMINATIONS BY SPARK-SOURCE MASS
SPECTROGRAPHY

The determination of trace-element concentrations by conventional

thermal- and gas-source mass spectrometry, using stable-isotope dilution
techniques, has become a widely used analytical method in geochemistry.
12. Mass Spectrometry 371

In spite of its superior accuracy the method has two serious limitations:
(a) it may be applied only to gaseous elements and those which may be
efficiently ionized in a thermal-ion source and (b) only one or at the most
a few elements can be determined simultaneously (the rare earths are an
exception).
These two limitations have been overcome by the recently developed
technique of spark-source mass spectrometry.5.261.262 In this method a very
large number of elements, potentially all, may be determined simultaneously,
and a wide concentration range from major elements down to trace con-
stituents may be covered in a single analysis. The sensitivity is approximately,
within a factor of two to three, the same for all elements, and the detection
limits achieved today for most of the elements are in the ppb range. A
decided advantage of this technique is the limited amount of sample prep-
aration required; conductors and semiconductors may be analyzed direct-
ly, whereas nonconductors require mixing with a conductor before analysis.
The amount of sample consumed during a trace analysis is very small,
and I-ppm detection limits have been achieved on O.OI-mg samples. Whereas
this may be of considerable value in the case of small samples, it will pose
a problem if the bulk composition of a large sample has to be determined,
since it requires that the sample be homogeneous at the scale of volume
which is vaporized and ionized in the analysis.

3.1. The Radiofrequency Spark-Source Mass Spectro-

graph

The mass spectrographs used for trace-element determinations by spark-

source techniques (a schematic drawing of a typical instrument is shown in
Fig. 4) incorporate as essential components: (a) an ion source, which
produces an ion beam representative of the total sample; (b) a mass analyzer
(electrostatic analyzer and magnetic analyzer), that resolves the ion beam
into a spectrum of mass lines; (c) a detection system (photoplate), which
permanently records the lines of the resolved mass spectrum and their in-
tensities; and (d) a vacuum system, that encloses the whole apparatus.
A rf spark source is commonly used as the ion source for trace analysis
by mass spectrography because it ionizes all elements with approximately
the same efficiency and demonstrates, in addition, high sensitivity, both in
terms of the amount of sample consumed during the analysis and the con-
centration of the element detected. Two disadvantageous features of this
type of ion source largely determine the design of the rest of the instrumen-
tation. Firstly, the ions leaving the source show a wide range in energy.
372 J. N. Weber and P. Deines

Energ) Dltl ~ n~1

Silt ~

\
\
'"
\~ostotic
\ Anol)zer

\
"'\ Ion Beam
Chop per ~ap Moni tor Electrode

Grounded Collimating Sl its " ,~ '!:, Energ) Def ining

Acce lerat ing Potenti al Energy Def i ni ng ...., ,_. Slit
R-f Spark Source Slit - ,
Bottom Moni t or Electrode
Front View of Energy Def ining -
and Tola l Beam Monitor System
Fig. 4. Diagram of a Mattauch-Herzog double-focusing mass spectrograph.

A mass analyzing system is therefore required that separates and focuses

ions irrespective of their initial direction and energy. A combination of
cylindrical, or spherical, electric field and magnetic sector field has been
found to have this property. The mass-analyzing system most frequently
used is based on the design of a double-focusing mass spectrograph by J.
Mattauch and R. Herzog. Secondly, the ion beam leaving the sourCe also
shows erratic intensity variations in time; hence, the abundance of the mass-
resolved ion beams of all elements of interest should be recorded simul-
taneously and averaged over a certain length of time. This is done con-
veniently with a photographic plate, which is hence widely used as the ion
detector, although electric recording systems have also been employed.
A wide variety of sample types may be analyzed by spark-source mass
spectrography; materials studied include conductors, semiconductors, insu-
lators (ceramics and rocks), liquids, and organic materials.

3.1.1. The rf Spark Source

In the rf spark source ionization of the sample is accomplished by the
application of a rf potential (50-100 kV at about 106 cps) to two conducting
electrode rods (1 mm diameter, 10 mm long) of the sample material. A
discharge occurs resulting in localized heating of the solid, vaporization,
and ionization. In order to keep the temperature of the electrodes low, since
12. Mass Spectrometry 373

excessive heating would cause selective losses of the more volatile elements,
the rf field is pulsed. The discharge is sustained only about 3% of the time
at most, usually smaller duty ratios are used.
At the effective temperatures in the spark (50,000-60,0000K) differences
in the ionization efficiencies between elements are substantially reduced (to
within a factor of two to three) compared to other types of ion sources.
Thermal-source ionization efficiencies, for example, vary over many orders
of magnitude. The relative uniform ionization efficiency is in fact the most
attractive feature of the rf spark source for trace analysis. The overall
efficiency of the spark-source method is given by the ratio of atoms vapor-
ized to the number of ions detected, a typical atom-ion ratio may be of the
order of 5 X 107 /1.
In the rf discharge singly charged positive and negative ions as well as
multiply charged ions are formed. For a given element the singly charged
species always produces the most intense lines, and the abundance of the
multiply charged ions decreases by approximately a factor of 10 for each
added charge. The relative-intensity distribution of the multiply charged
ions differs from element to element and changes with the operating char-
acteristics of the spark.
Polyatomic ions of the electrode material have been observed; charged
carbon clusters as large as Ct4 and C33 are formed from carbon electrodes,
and different polyelemental combinations have also been reported. In ge-
neral one can count on the formation of polyatomic clusters containing the
major constituents of the sample electrodes, and the relative concentration
of different species can be related to bond characteristics of the sample
compounds. The occurrence of this type of ion will complicate the mass
spectra obtained with a rf spark source and has to be taken into account in
their interpretation.
The residual gas in the ion source, as well as the gas adsorbed on the
electrode surfaces, is ionized in the rf discharge. Whenever the concentra-
tion of elements that are present in the background gas have to be deter-
mined, thorough cleaning of the electrode surfaces under vacuum and low-
ion-source pressures (10- 9 torr) are required.
The ions formed in the spark acquire their energy by acceleration in
the rf field between the electrodes; accordingly, the ion energy will depend
on where and when in the rf discharge ionization took place. Ions of dif-
ferent elements, charge states, or polyatomic clusters are formed at different
locations between the electrodes and at different times in the rf cycle; hence,
they experience different accelerations and accordingly show different energy
distributions. Due to the high potential of the rf field ionization in the spark
374 J. N. Weber and P. Deines

leads to a considerable energy spread (several thousand V) of the ions. Not

only must a double-focusing mass spectrometer be therefore used in con-
junction with the rf source, but an additional problem is also introduced.
As, on the one hand, the ions of different elements formed in the rf spark
show varying energy distributions, on the other hand the energy range which
can be accepted by the mass-analyzing system is limited (300 V), the
composition of the total resolved ion beam is not necessarily identical with
the ion composition of the spark plasma, and hence corrections may have
to be applied.
The ion-acceleration system of the spark source is relatively simple.
Since ionization takes place rather erratically, and the ions show a con-
siderable energy spread, little would be gained by employing complex-ion
optical-focusing arrangements. Hence, the electrodes are usually placed in
front of a large plate with a small, round hole at the optical axis of the
instrument. The main accelerating potential (about 20 kV dc) is applied to
this plate, and one of the electrodes is electrically tied to it. At a suitable
distance two narrow, grounded, collimating slits are provided to limit the
width and angular spread of the ions drawn from the source. In addition
a wider split plate may be located between the two grounded slits to be
used, as an electronic shutter, by the application of an electric potential.
For maximum analytical reproducibility all controllable conditions
under which the samples are analyzed have to be kept constant. This in-
cludes spark voltage, pulse length and repetition frequency of the rf field,
and the width of the spark gap and its position with respect to the acceler-
ating system, as well as the accelerating potential, the electrostatic analyzer
potential, and their ratio.
Since the use of the rf spark source for the analysis of solids is asso-
ciated with difficulties, other types of ion sources have been considered for
this purpose; these include the vacuum-vibrator source, the triggered, dc arc
source, the laser-discharge source with or without auxiliary electron beams,
and ion-bombardment sources. However, presently only the rf spark source
is widely used for trace analysis by mass spectrography.

3.1.2. The Mass-Analyzing System

Since the ion beam that leaves the source possesses an angular divergence
and a certain spread in energy, the mass-analyzing system should be able to
focus ions of the same mass at one point irrespective of their initial direction
and energy. This can be accomplished with an electric cylindrical field and a
magnetic field in series. Since such an arrangement focuses an ion beam
12. Mass Spectrometry 375

regardless of variations in two of its properties, it is called a double-focusing

mass-analyzing system.

The Magnetic Field. In many mass spectrometers mass separation

is accomplished with a sector magnetic field. Ions entering such a field are
deflected on circular paths of radius am, which is given by

=~ 1/2mV
a
m H r ez

where c is the velocity of light, H is the magnetic field strength, m is the

mass of the ion, V is the electric potential through which the ion has been
accelerated, e is the elementary charge, and z is the number of unit charges
carried by the ion (units in cgs-esu system). For constant H, V, e, and z
the deflection of ions in the magnetic field will produce a mass separation.
Since ion beams that are drawn from a rf discharge show energy variations
(the potential V is not the same for all ions) a magnetic field is insufficient
to perform a mass separation on them.
In addition to the deflecting properties the sector magnetic field pos-
sesses focusing properties, i.e., ions emerging with a small angular divergence
from a point in front of the magnetic field will be recombined in a point
behind the magnetic field. If H, e, z, and V are constant for the ion beam
entering the magnetic field, ions of a given mass will be focused at this point.
If H, e, z, and m are constant, and the ion beam contains ions of two distinct
energies, e VI and e V2 , where e V2 - e VI = eLi V, two separate images of
ions of the same mass will result. The spacial separation of the two images
will be proportional to the relative energy difference LI VIV. This action of
the magnetic field is called energy dispersion.

The Electric Field. The electric cylindrical field is produced by a

capacitor consisting of two cylindrical plate sections with a gap of about
1 cm between the two plates. The dimension of the capacitor in the direction
of the axis of the cylinder is about 5-10 X larger than the gap width. A
potential of 1000-2000 V is applied between the two plates producing a
homogeneous, radially symmetrical field. Positive ions entering the field are
deflected towards the center if the outer plate is kept at a positive and the
inner plate at a negative potential. The amount of deflection will depend on
the kinetic energy of the charged particle. The actual paths taken by indivi-
dual ions are rather complex and cannot be determined as easily as in the
case of the magnetic field. The electric cylindrical field also possesses focus-
ing properties. A monoenergetic ion beam of small angular divergence
376 J. N. Weber and P. Deines

~~-~~,.~~
,

,
'5'

Fig. 5. Principle of double focusing with an electric and a magnetic field in series
(schematic).

emerging from a point in front of the electric field will be focused at a

certain distance behind the field, as shown in Fig. 5. For a given dE"fiecting
potential of the electric field ions of a certain kinetic energy will be focused
on the central ion path, indicated by point A in Fig. 5; if ions of a different
energy are present in the beam their image will still lie at the same distance
behind the electric field, however, it will be displaced from the central path
by a certain amount (point B in Fig. 5). The distance between the two points
of focus is directly related to the relative energy difference between the
ions that are recombined in them. This action of the electric field is called
energy dispersion.
The Principle of Double Focusing. Consider an ion beam con-
taining two masses Ml and M2 that shows a certain energy spread. If such
a beam enters the mass-analyzing system shown in Fig. 5, ions of the small-
est energy present in the beam may be focused at point A, and ions with
the largest energy at point B. If a slit S" is provided in the plane of focus of
the electric field and adjusted so that only ions focused at point A will pass
through it, these ions will be separated in the magnetic field according to
their mass and will come to a direction focus behind the magnetic field, for
12. Mass Spectrometry 377

example, at points AMI and AM., corresponding to the masses Ml and M 2 ,

respectively. Mass separation in the magnetic field is possible because ions
passing through A are monoenergetic. Likewise, if the slit is set such that
only ions focused at point B will pass through it, the ions will be recombined
at points BMI and BM corresponding to the same two masses Ml and M 2 ,
representing, however, ions of a different energy. The line along which ions
of a given energy and small initial angular divergence are focused according
to their mass, is called the direction focusing curve d. If the slit is opened
so that ions of an energy range between the limits given by A and B pass
through it the resulting mass line for mass Ml would be spread over the
distance AMI to BM1 . The paths of the central beams through A and B
intersect at points ABMI and AB.li .; likewise, all other central beams of ions
of masses Ml and M2 and of energies between A and B will pass through
these two points. At points ABJ!1 and ABM the central beams of ions of
the same mass but different initial energies are recombined. The line defined
by the focus of ions of the same mass but different initial energies and
vanishingly small angular divergence (remember, central beams were con-
sidered only) is referred to as the velocity focusing curve, v. In general the
velocity and direction focusing point of a mass do not coincide. Where the
velocity focusing curve and the direction focusing curve intersect one ano-
ther, ions of the same mass are, however, brought to focus, irrespective of
their initial direction or energy, i.e., double focusing occurs (point D). By
consideration of the paths of the ions which are recombined in point D it
is found that double focusing of an ion beam is achieved whenever the
energy dispersion introduced into the ion beam in the electric field equals
that of the magnetic field, traversed in the opposite direction.
By judicial choice of instrument parameters J. Mattauch and R. Herzog
were able to design a mass spectrograph in which the line of direction focus
and velocity focus coincide for all masses, the line of double focusing is
straight and located at the exit of the magnetic field. This basic design, with
only slight modifications, is still being used in most of the mass spectro-
graphs today.

3.1.3. Detection System

Although electrical detection of the resolved ion beams has been used
in spark-source mass spectrographs and will become more important in
the future, the mass spectra are presently still commonly recorded on photo-
graphic plates. Photoplates used in spark-source mass spectrographs consist
generally of 2-in.-wide and 10, 12, or 15-in.-Iong pieces of thin glass covered
with a thin layer of gelatin containing an emulsion of silver halide salts.
378 J. N. Weber and P. Deines

When ions strike the photographic emulsion they affect the silver halide
grains and a latent image is formed. Development of the latent image reduces
to metallic silver, the silver salts in those grains which have been affected by
ion impact. The resulting cluster of silver grains darkens the emulsion locally.
To make the darkening on the plate permanent and to clear the unexposed
plate portion, the unreduced silver halide grains are dissolved in a suitable
chemical solution; this process is known as fixing. The fixing solution con-
tains, also, a hardening agent which toughens the gelatin. After developing
and fixing the plate is washed and dried.
The intensity of darkening on the photo plate is determined with a
microdensitometer by measuring the attenuation of a light beam when
passing through the darkened area, relative to that when passing the same
light beam through an unexposed, clear portion of the plate.
The photographic response measured with the densitometer is in some
manner related to the number of ions which have struck the plate over the
line area. The exact form of this relationship has to be established by plate
calibration. Higher ion-beam intensities and longer exposures will cause
more intensely blackened lines than weak ion beams or short exposures.
If the blackening of the plate is a function of the product of signal intensity
and exposure time, i.e., if the reduction of the exposure length by a given
factor may be compensated for by increasing the intensity of the ion beam
by the same factor, the law of reciprocity is valid for the plate emulsion.
Special types of photographic plates are required for successful ion-
beam-intensity measurements. Desirable emulsion properties include fine,
uniform grain; low gelatin content; and uniform, near-surface distribution
of the silver halide grains. The Q2 plates (one of the Q emulsions developed
for ion detection by the Ilford Company in cooperation with F. W. Aston)
are commonly used today for analytical work with spark-source mass
spectrographs. These more conventional gelatin-containing plates show,
however, a number of undesirable characteristics: (a) emulsion deterioration
at room temperatures; (b) sensitivity dependence on mass and energy of the
ions; (c) unsharp images due to scattering of ions in the emulsion; (d) line
broadening due to poor electrical conductivity of the emulsion; (e) excessive
fogging around the lines of heavy exposure; (f) poor vacuum characteristics
due to the absorption of water by the gelatin; and (g) susceptibility to
abrasion. Since a number of these poor characteristics are related to the
presence of gelatin, gelatin-free photoplates have been developed. Their
photosensitive silver salts are deposited without a binder directly on the glass
backing. The use of these plates has not become widespread, however.
The validity of the law of reciprocity for the Q emulsions has been
12. Mass Spectrometry 379

investigated and verified a number of times. Studies of the ion specific

properties of the plate emulsions have shown that their sensitivity is reduced
with increasing ion mass and decreasing ion energy, i.e., for a given ion-
beam intensity the blackening of a line will be more intense for ions of
lower mass and higher energy. The saturation density decreases with in-
creasing mass and decreasing ion energy. The contrast of the Q2 emulsion
is also affected by the ion energy; with increasing ion energy the contrast of
the plates is reduced.
In order to use the photographic plate darkening as a measure of the
size of the ion current producing it, the relationship between ion exposure
and photographic response should be determined over the total range of
darkening of the emulsion from barely visible images to saturation black-
ening. A number of methods have been proposed to accomplish this calibra-
tion; although they differ considerably in the details of their execution, they
are all based on the same principle, namely, that on the plate a number of
lines of different densities are produced which possess known relative
abundances. The relative intensities of the signals producing a set of cali-
bration lines of varying density on a photoplate may be known from:
(a) the length of time an ion beam of constant intensity was striking the
plate; (b) the relative abundance of different molecular ion species in the
beam; (c) the isotopic composition of elements; and (d) by electrical measu-
rement of part of the total ion beam which is producing the calibration
spectra. This can, for example, be done with the total beam monitor, as
shown in Fig. 4.
Quite frequently the Churchill two-line method, utilizing the known
relative abundance of two isotopes of an element, has been employed for
emulsion calibration. The use of the Churchill two-line method has been
criticized, however, since only a fraction of the total information usually
available on an analytical plate is evaluated for plate calibration. The
method also is quite sensitive to small deviations of the assumed isotopic
composition from the isotopic abundance which is actually producing the
lines on the photoplate (instrumental bias). This latter point is of special
importance because of the additive manner in which the emulsion calibra-
tion is established by the Churchill method.
A calibration which avoids this problem and utilizes all the available
information on the plate may be carried out as follows. In a series of mass
spectra of increasing exposure length, recorded below one another on a
photoplate, the density of the lines of an element including all isotopes in
all exposures is measured. The relative-exposure values of a given mass line
in different exposures are given by the exposure ratios determined with the
380 J. N. Weber and P. Deines

total beam monitor. The relative-exposure values of different isotopes in one

exposure are given by their isotopic abundances. The optical density of the
lines plotted versus the logarithm of the known relative-exposure values
yields the plate calibration. Background, line-width variations, and ion
specific properties of the emulsion have to be considered.
The calibration methods can be made absolute by measuring simul-
taneously one of the isotopes of an element electrically while another is
recorded photographically. Typically approximately 104-105 ionsJmm 2 must
strike the photoplate to produce a barely detectable image. The actual
number of ions required will vary with the emulsion and the mass and energy
of the ions.

3.2. Analytical Procedure

In order to obtain a meaningful analysis of a material by rf spark-
source mass spectrography, one has to ensure that the resolved ion beam
is truly representative of the sample. Biases that may occur in the analytical
process have to be evaluated and corrected.
It is imperative that the electrodes be representative of the sample to
be characterized, and that they are compositionally homogeneous at the
scale of volume that is consumed during an analysis. In a trace-element
determination about 0.1 mg of the electrodes are removed and only a frac-
tion of this material is actually ionized. Once the ions are formed only part
of them are drawn from the ionization region, and only a fraction of these
ions is permitted to pass through the energy-defining slit of the spectro-
graph. Additional ions are lost when the beam passes through the rest of
the mass-analyzing system. In the final analysis the analyst relies on about
1()13 ions to represent the electrode material, which, in turn, one would like
to interpret as being representative of a rock sample of several lb, for ex-
ample. In order that such inferences can reliably be made, one has to ensure
that anyone of the sampling steps in the analytical procedure yields a
representative sample, and that wherever biases are introduced these are
known and can be taken into account. Attempts have been made to evaluate
separately some of the biases involved in mass-spectrographic analysis.
Since most of these biases may not be dealt with directly, however, overall
correction factors have been determined by the use of samples of known
composition. The correction factors depend on the operating characteristics
of the spectrograph and may be applied only to analyses that are carried
out under the same conditions under which these factors were determined.
The relative amounts of charge deposited on the photoplate in different
12. Mass Spectrometry 381

mass spectra may be determined by intercepting a fraction of the total ion

beam entering the magnetic field. This is accomplished with a beam monitor
located behind the energy stop of the instrument, as shown in Fig. 4. The
ion currents striking the monitor are measured and integrated with a dc
amplifier.
For the analysis of a sample a series of exposures is produced on the
photoplate covering, normally, an exposure range from 10-13 to 10-6 C.
After an exposure has been made the plate is moved vertically, with the aid
of the photoplate elevator, so that an unexposed portion of the plate is
located behind the magnet gap, and the next exposure is laid down. Normal-
ly, 15-20 exposures may be recorded on a plate. The lowest exposures are
chosen so that some of the isotopes of the major constituent(s) are recorded
in the linear portion of the response curve of the photographic emulsion;
with an increased exposure length mass lines of lower and lower concentra-
tions will appear. The exposure is normally increased by a factor of two to
three depending on the concentration range to be covered on one plate. In
most instruments practically the whole elemental mass range may be covered
in a single spectrum.
The first step in the evaluation of a recorded mass spectrum is the de-
termination of the masses of the different lines. The orientation in a spec-
trum is usually not very difficult once a certain amount of experience has
been gained, because in most cases the major elements in a given sample
are known and the operator soon learns to recognize the typical isotopic
composition pattern of different elements. A mass scale along the photo-
plate may be determined by using the fact that the displacement of mass
lines along the line of focus follows a defined relationship. In the case of the
Mattauch-Herzog geometry a quadratic relationship holds. If dx is the
displacement of a line of unknown mass Mx from an arbitrary reference
point in the spectrum, the mass of this line is given by

where do and h are constants which can be evaluated from the knowledge
of the masses of two lines and their respective displacement from the arbi-
trary reference point.
After the masses of all lines in the spectrum have been established the
ion species producing them may be identified. In view of the large variety
of ion types formed in a rf discharge the unambiguous identification of
lines, especially of weak ones, may be difficult. The spectrum may be further
complicated by the occurrence of charge-exchange lines, i.e., lines which are
382 J. N. Weber and P. Deines

due to ions, such as An+, which have changed their charge during passage
through the instrument by collision with rest-gas atoms or molecules, B, to
yield the following:

The effect on the mass spectrum depends on where in the spectrograph the
collision occurred. Charge exchange within the electrostatic analyzer or
magnetic analyzer will lead to easily recognizable continua. Charge exchange
between the electrostatic analyzer and the magnetic analyzer will lead to
well-defined mass lines at mass positions (n/m 2 ) X M, where n is the charge
of the ion while passing through the electric field, m is the charge while
passing through the magnetic field, and M is its mass.
The following possible contributions to a mass have to be considered:
(a) singly charged ions, all unresolved isobars; (b) multiply charged ions;
(c) singly charged polymer monoelemental ions; (d) singly charged polymer
polyelemental ions; (e) multiply charged polymer ions; and (f) charge-
exchange ions. In general the abundance of polymer ions, mono or poly-
elemental, and of charge-exchange ions is relatively low and restricted to
the major elements in the sample.
The presence of an element in the sample is recognized by the occur-
rence of its characteristic isotope-abundance pattern. In order that a mass
line may be definitely assigned to an element one should be able to detect
at least one other isotope of the element and the intensity ratio of the two
lines should give the correct isotope ratio within 5% or better.
In the quantitative evaluation of a set of mass spectra absolute con-
centrations are not derived directly, but different constituents are referred
to an isotope of the matrix or of a trace element, or, alternatively, to an
isotope of an added internal standard. In order to arrive at absolute con-
centrations the concentration of the reference element either has to be
known or has to be determined by an alternate method. The initial con-
centrations evaluated from the mass spectra are always in terms of atom
fractions (at. %, ppm atomic, or ppb atomic), since they are based on iso-
topic ratios.
A large number of methods are being used to convert the recorded mass
spectra into absolute concentrations, varying in the number of correction
factors involved, the manner in which they are established, and the types
of biases which are removed by their application.
In the simplest (semiquantitative) evaluation method the ratio of the
impurity to that of a reference element is visually estimated from the ratio
of the minimum exposures in which impurity element and reference element
12. Mass Spectrometry 383

are barely detectable, with appropriate corrections for the relative isotopic
abundances of the isotope lines used. If in a series of graded exposures an
isotope of the reference element with relative abundance Ir becomes just
detectable in an exposure of Er relative units, where E is obtained from the
integrated charge collected with the total beam monitor, and if an isotope
of the trace element with relative abundance It is first detected in an exposure
of Et relative units, then if Cr is the concentration of the reference element
in the sample, in at. %, the concentration of the trace element Ct is given by

Ct = Cr ~ i (at. %)

This method will naturally yield only very rough estimates, since it is based
on a visual estimate of the optical density of the lines, neglects background
corrections and line-width variations, and assumes that all elements are
treated and recorded identically in the analytical process. The results
obtained in this manner normally agree within a factor of two to three with
those obtained by other analytical techniques.
The method will become considerably more refined if the line densities
are measured with a densitometer and are adjusted for line-width variations
and background and if a correction factor for different elemental sensitivies
R t is incorporated in the computation. In this case the ratio ErlEt is obtained
in the following manner. For both lines the density is measured with a
microdensitometer in several exposures of increasing length and plotted
versus the logarithm of the exposure. The exposures required to achieve a
line of the same density for the reference element and the unknown are read
from the plot; the density for which the comparison is made is chosen so
that it will fall on the linear portion of the response curve of the photo-
graphic emulsion.
The relative-sensitivity coefficient R t is a measure of the sensitivity of
the total measuring procedure for the line of the trace-element isotope
compared with the sensitivity for the line of the reference-element isotope;
to the latter an arbitrary value of unity is assigned. R t is determined by the
use of a standard which contains the trace element and the reference element
in known proportions. The concentration of the trace element is then cal-
culated from

R t varies from element to element and the charge state of the ions, as well
as with the operating conditions of the mass spectrograph. A particular
384 J. N. Weber and P. Deines

value of R t may, therefore, only be applied to analyses carried out under

the same experimental conditions under which the correction factor was
determined. The relative sensitivity coefficients may be thought of as a
composite of: (a) an ionization correction; (b) a transmission correction;
and (c) a detection correction due to ion mass and energy dependence of
the emulsion sensitivity. In some cases line-width-variation corrections are
also incorporated in R t
The acquisition of densitometer data and their reduction to concen-
trations is time consuming and prone to the introduction of human error,
especially if a large number of elements have to be determined; conse-
quently, in many cases only a fraction of the available information is ex-
tracted. The plate evaluation has been made more efficient, however, by
automatization of the densitometer measurements and the inclusion of
computer programs for the determination of emulsion calibration and con-
centration computations from densitometer data.

3.3. Sample-Handling Techniques

Conducting and semiconducting materials may be used after appro-
priate shaping directly as electrodes. Care has to be taken, however, that
in the shaping process none of the impurities that are being studied are
introduced into the sample. The analysis of such electrodes characterizes,
naturally, only the electrode material consumed in the analytical process.
Whether this is a representative sample of the bulk of the material from
which the electrodes were formed is a different question. If the composition
of the material is not uniform, individual analyses by spark-source mass
spectrography may deviate considerably from those obtained by other
analytical methods which utilize comparatively larger amounts of sample.
Metal powders may be formed into electrode rods by compacting them
under pressure. In cases where the physical properties of the metal prohibit
this the powder may be mixed with graphite as a binder. Analyses of low-
boiling metals such as mercury or gallium can be accomplished by the use
of liquid-nitrogen-cooled electrodes.
Nonconducting materials cannot be ionized directly in a rf spark source,
since for the initiation and maintenance of the discharge at least one of the
electrodes must be conducting. A number of electrode types have been
described which will permit the maintenance of a spark and the ionization
of insulating material in it. Some investigators have packed hollow, high-
purity metal tubes with insulator powder. Others have backed electrodes
of the nonconductor with a high-purity metal strip, or used cup electrodes
12. Mass Spectrometry 385

packed with the sample and faced by a pointed, conducting-probe electrode.

All of these electrode types have the disadvantage that the relative contri-
bution of conductor and insulator to the total ion beam is not very repro-
ducible. This makes the quantitative evaluation of the mass spectra obtained
with these electrodes difficult. The contribution of sample to the ion beam
is more reproducible, if the powdered insulator is well mixed with a high-
purity conducting material and pressed into electrode rods. Elements that
have been used for this purpose include gold, silver, copper, titanium, tin,
nickel, aluminum, and carbon. The selection of a particular conductor will
vary with the analytical problem, since the mass spectrum of the added
conductor is superposed on that of the sample. Considerable effort may
have to be spent to obtain a truly homogeneous mixture of conductor and
insulator.
Impurities of liquids have been determined successfully by drying or
electroplating them on high-purity metal electrodes; water impurities in the
ppb range have been determined on 30-mg samples.
The analysis of traces of carbon, nitrogen, oxygen, and hydrogen is
difficult by rf spark-source mass spectrography. This is due to the fact that
these elements are present in the residual gas of the mass spectrograph and
are ionized in the rf discharge along with the sample. Reduction of the
analytical background has been achieved by: (a) flushing the instrument
parts which are opened to atmospheric pressures during the changing of
a sample with dry argon; (b) removing surface contaminations by cathode
etching through the initiation of a glow discharge in the ion source at
about 25 fl pressure; (c) the use of cryogenic pumping in the ion source;
and (d) the presparking of the electrodes. The ion-source pressure has to
be kept in the 10-9 torr range during the analysis for these elements. Al-
though concentrations of oxygen, nitrogen, carbon, and hydrogen of a few
ppm and somewhat lower may be detected, quantitative determinations are
further complicated by the unavailability of certified gas-free samples, lack
of standards containing gases in the low-ppm range, and the heterogeneity
of the gas distribution in most samples.
Trace-element determinations of organic materials have also been
carried out by mass spectrography. Here again the addition of a conductor
to the sample is generally required. In some cases the organic sample was
mixed directly with graphite and pressed into electrode rods. This results,
however, in excessive background spectra, and it is preferable to first carry
out a low-temperature ashing of the sample; the ash is then mixed with
graphite and formed into electrodes.
As the sample consumption in the spark-source method is rather small
386 J. N. Weber and P. Deines

the technique may be readily applied to the analysis of microsamples. Small,

rigid conducting particles have been handled by constructing special elec-
trode holders of high-purity materials. Small powdered samples have been
analyzed by pressing electrodes from metal powders in which the sample
proper is confined to the very tip of the electrode rod. With this "tipped-
electrode" type trace-element concentrations at the ppm level were deter-
mined in 0.01 mg of sample.
Other analytical problems involving small samples are the analysis of
small compositional irregularities, such as inclusions, segregations, or layer-
ing. In conducting samples such compositional inhomogeneities have been
studied by sparking a pointed-probe electrode against a flat sample electrode.
With careful probe control impurities of a few ppm may be detected in
zones as narrow as 10 mils. The shape of the probe electrode is critical for
obtaining sufficiently well-defined spark areas. The contribution of the
probe electrode to the total ion beam reduces the sensitivity of the method,
and appropriate corrections have to be applied for its presence if quantita-
tive determinations are to be carried out.
The isotope-dilution technique, generally used only for elements which
can be ionized in a gas source or thermal source, has been applied to a
limited extent to trace analysis by rf spark-source mass spectrography.
Concentration determinations of copper in nickel oxide, antimony in tin,
and lead in cadmium have been described. The samples are dissolved, the
spike solutions are added, and part of the solution is either dried or electro-
plated on high-purity electrodes.

3.4. Detection Limit, Precision, and Accuracy

The minimum number of atoms required to produce a measurable
signal at the detector is defined as the absolute detection limit. If no disturb-
ing influences are present, such as line overlap or excessive background, the
absolute detection limit for an element is determined by the sensitivity of
the detector for the element, its isotopic composition and ionization effi-
ciency, as well as the transmission efficiency of the instrument. Based on the
following observations the absolute detection limit may be estimated to be
of the order of 5 X 1011 to 5 X 1012 atoms: (a) one out of 5 X 10 7 atoms
removed from the electrodes reaches the detector; (b) 104-10 5 ions/mm 2
are required to produce a detectable mass line; and (c) an average line
covers about 0.1 mm 2 It was assumed in the calculation that all ions from
the element are combined in one line, i.e., a mono isotopic element, and the
formation of multiply charged and polyatomic ions was neglected. Measured
12. Mass Spectrometry 387

absolute detection limits have been quoted as 6 X 1011 and 1013 atoms. The
modifier "absolute" in conjunction with detection limit should, however,
not be construed to indicate that the given figures present the absolute limit
of detection of this method since the number of atoms required to produce a
detectable line may be lowered by: (a) improved ion-source design, per-
mitting higher ionization and extraction efficiencies; (b) improved trans-
mission of the spectrograph; (c) increased detector sensitivity; and (d) im-
proved signal-to-noise ratio of the analytical process.
The detection limit is defined as the minimum concentration of an
element that may be determined in an analysis. The detectable concentra-
tion is reduced by increasing the exposure length. However, limitations exist
as to the length of time over which an exposure can reasonably be extended
and the size of the sample that is available. A maximum exposure of 10-6 C
can be obtained within a reasonable length of time. Measurements of samples
containing elements with isotopic concentrations at the ppb level as well
as extrapolations of concentration determinations of elements in the ppm
range show that in a 1O-6-C exposure detection limits of a few ppb can be
obtained for almost all elements; in some cases detection limits even below
1 ppb have been reported. Exceptions to these low detection limits are the
gases hydrogen, nitrogen, and oxygen, and carbon where it lies in the low-
ppm range for reasons discussed above.
Apart from sample-size limitations and reasonable exposure length a
number of other factors may prohibit the detection of the lowest possible
concentrations: (a) spectral overlap due to background lines, charge-ex-
change lines, multiply charged ions, molecular ions, and isobars may occur;
(b) the continuous plate background may become excessive. The general
plate background increases with the exposure length, and continua and
halation around major lines occurs. These phenomena are related to ion
scatter in the instrument, charge-exchange processes, surface charging of
the photoplate, and secondary ion impact.
Little can usually be done if the detection of an element is limited by
line overlap. An increase in the resolution of the mass spectrograph would
help; limits are, however, set by the fact that with a given instrument only
a certain maximum resolution can be practically achieved, and that increased
resolution will result in a loss of sensitivity since narrower collimating slits
have to be used in order to achieve it.
The possibilities of improving detection limits which are imposed by
excessive plate background have been and are still being investigated very
actively. Obviously a decrease of the residual pressure throughout the mass
spectrograph would be beneficial. Background reductions have been achie-
388 J. N. Weber and P. Deines

ved by preventing the very intense ion beam of the major component to
strike the plate, by breaking it at the point where this ion species would
be recorded, and leaving a gap. The halation around intense lines has been
reduced also by placing grounded, conduction strips on the plate at the
point of impact of the strong ion beams. Special developers and developing
conditions can suppress the development of the background and produce
clearer photoplates; very considerable improvements in this respect have
been achieved by the use of gelatin-free photoplates.
The smallest concentration difference that may be detected is deter-
mined by the precision and sensitivity, i.e., the slope of the analytical curve,
of the analytical method. The precisions quoted for mass-spectrographic
analyses vary considerably. Visual concentration estimates can be carried
out within a factor of two to three. Generally, precisions of 20-30%, and in
favorable cases of 10%, are obtained, if densitometry is used for plate
evaluation and appropriate correction factors are applied. In view of the
fact that the early electrical concentration determinations with rf spark-
source mass spectrometry showed precisions of 2.5%, i.e., about a factor
of 10 better than the results obtained with photographic detection, it has
been widely believed that the photoplate recording technique is responsible
for the poor reproducibility. This was, however, at variance with the obser-
vation that isotopic composition measurements could be made on photo-
plates with a precision of 5% and better. It has been recently shown that
with proper experimental methods the photoplate is inherently capable of
yielding results with a precision of 1-2 standard deviations.
Major sources of poor reproducibility are: (a) variability of spark
parameters, (b) sample inhomogeneities, and (c) densitometer errors. The
recognition that fluctuations in the ionization process of the rf spark source
are a major cause for poor reproducibility has led some investigators to
abandon the rf spark source completely. Others have attempted to obtain
better control over the spark parameters, such as rf pulse length, repetition
frequency, spark voltage, spark-gap width, and spark location. In this
second approach the spark parameters are fixed as closely as possible
throughout the calibration and during the recording of the analytical mass
spectra. As the average ion current flowing through the instrument has to
be varied in order to be able to record exposures of different intensity, an
ion-beam chopper, shown in Fig. 4, is set into the beam path. This device
interrupts the constant ion beam with a variable frequency, thus providing
a control over the average ion current flowing through the instrument with-
out changes in the sparking conditions. The use of the ion-beam chopper
has led to analytical precisions of 4-5%. Similar reproducibilities have been
12. Mass Spectrometry 389

achieved by using a low-voltage discharge source in place of the rf spark

source. The use of this type of ion source has, however, so far been very
limited.
For accurate measurements a comparative standard of sufficient homo-
geneity is essential in order to correct for differences in elemental sensitivities.
The accuracy of the method will critically depend on the constancy of these
relative-sensitivity coefficients and on the availability of standards which
match the sample in chemical composition and physical form. In general
one can expect that the achievable accuracy is slightly less than the precision
obtained by the same analyst.

3.5. Geochemical Applications

Although rf spark-source mass spectrographs have been used extensively

in solving a wide variety of analytical problems in industrial research their
application in geochemical studies has been very limited.
Leipziger263.264 was the first to determine 207Pb_ 206 Pb ages of uraninite
and pitchblende samples by rf spark-source mass spectrography. After
suitable shaping the minerals may be used directly as electrodes. The ana-
lytical procedure, therefore, requires no chemical treatment, which presents
a certain advantage over the more conventional methods. At the time the
determinations were carried out the limited precision with which the isotope
ratios could be measured photographically (5%) presented a drawback.
With the present better understanding of correct plate evaluation a con-
siderable part of the uncertainty can, however, be removed.
The early applications of the technique to the determination of trace
elements in geologic materials were of a preliminary nature, aimed largely
at the demonstration of the usefulness of the method.265-268 The standard
rock samples G-I, W -1, and S-l, and a number of native metals and sulfides,
were used as samples. Natural graphite crystals were also analyzed. 269
The quantitative aspect of the method for geologic samples was deve-
loped through the work of Taylor270.271 and Nicholls and his associates.272
The technique is generally used to supplement concentration determinations
carried out by wet chemical, emission spectrographic, x-ray-fluorescence,
and other methods. Examples of its application include an investigation
concerning the composition of meteorite-impact glass 273 and trace-element
studies of andesites 274 and rhyolitic volcanic rocks. 275
The main advantageous features of the method, such as small sample
size, the low detection limits, the wide range of elements that can be de-
termined simultaneously in a single analysis, and the wide concentration
390 J. N. Weber and P. Deines

range that may be covered in a single analysis, will no doubt lead to further
applications in geochemical studies. Complete analysis of many natural
waters by conventional methods is difficult without preconcentration. With
rf spark-source mass spectrography impurities in the ppb range have been
detected in 30-mg-water samples and the whole element range was covered
in the analysis. The technique will be especially useful where it is difficult
to obtain large sample volumes as in the study of connate waters, fluid
inclusions, or liquids equilibrated with solid phases in hydrothermal experi-
ments. Further applications making use of the small sample techniques
will include the analysis of small single crystals and of solid phases
which have been precipitated in hydrothermal-equilibrium studies. If the
rf-probe technique can be extended to insulators it will become possible to
study major-, minor-, and trace-element-concentration variations simulta-
neously in zoned crystals. In view of the low detection limits and the pos-
sibility of carrying out simultaneously isotopic and elemental abundance
studies, the method may also be applicable to the study of spallation pro-
ducts in meteorites.

REFERENCES

1. Inghram, M. G., and Hayden, R. J., A Handbook on Mass Spectroscopy, Nuclear

Science Series, Report No. 14, National Academy of Sciences-National Research
Council, Washington, 1954.
2. Duckworth, H. E., Mass Spectroscopy, Cambridge University Press, Cambridge,
1958.
3. McDowell, C. A., ed., Mass Spectrometry, McGraw-Hill, New York, 1963.
4. Herzog, L. F., Marshall, D. J., Kendall, B. R., and Cambey, L. A., in Advances in
Analytical Chemistry and Instrumentation, C. N. Reilley, ed., Interscience, New York,
1964, Vo!. 3, pp. 143-183.
5. Roboz, J., in Trace Analysis, Physical Methods, G. H. Morrison, ed., Interscience,
New York, pp. 435-509.
6. Chupakhin, M. S., Zh. Anal. Khim. 14, 331 (1959).
7. Dattner, J., and Fischler, J., Brit. J. Appl. Phys. 14, 728 (1963).
8. Nier, A. 0., Phys. Rev. 77, 789 (1950).
9. Kistemaker, J., Nat!. Bur. Standards (US) Circ. 522, 1953, p. 243.
10. Boerboom, A. J. H., and Tasman, H. A., Physica 24, 683 (1958).
11. Shackleton, N. J., J. Sci. Instr. 42, 689 (1965).
12. Han, I., and Fritz, G. J., Anal. Chem. 37, 1442 (1965).
13. McKinney, C. R., McCrea, J. M., Epstein, S., Allen, H. A., and Urey, H. C., Rev.
Sci. Instr. 21, 724 (1950).
14. Epstein, S., Nat!. Bur. Standards (US) Circ. 522, 1953, p. 133.
15. Akers, R. J., J. Sci. Instr. 44, 871 (1967).
16. Chupakhin, M. S., Zh. Anal. Khim. 15, 155 (1960).
12. Mass Spectrometry 391

17. Craig, H., Geochim. Cosmochim. Acta 12, 133 (1957).

18. Galimov, E. M., Grinenko, V. A., and Ustinov, V.I., Zh. Anal. Khim. 20, 547 (1965).
19. Ustinov, V. I., Galimov, E. M., and Grinenko, V. A., ibid. 20, 1180 (1965).
20. Thode, H. G., Graham, R. L., and Ziegler, J. A., Can. J. Res. 23, 40 (1945).
21. Nier, A. 0., Eckelmann, W. R., and Lupton, R. A., Anal. Chem. 34, 1358 (1962).
22. Letolle, B., Marce, A., and Fontes, J. C., Bull. Soc. Franc. Miner. Crist. 88, 417
(1965).
23. Matus, L., Opauszky, I., and Kiss, I., Pribory i Tekhn. Eksperim. 11, 158 (1966).
24. Vetshtein, V. E., Metrod. Vopr. Izotopnoi Geol., Akad. Nauk SSSR 68 (1965).
25. Dibeler, V. H., Nat!. Acad. Sci.-Nat!. Res. Council Pub!. 400, 1956, p. 55.
26. Mohler, F. L., U.S. Natl. Bur. Standards Techn. Notes 51, 1 (1960).
27. Craig, H., Science 133, 1833 (1961).
28. Epstein, S., and Mayeda, T., Geochim. Cosmochim. Acta 4, 213 (1953).
29. Dansgaard, W., ibid. 3, 253 (1953).
30. Teis, R. V., Naidin, D. P., Zadorzhnyi, N. K., and Stolyarova, S. S., Geochem.
Intern. 1, 44 (1964).
31. Halevy, E., and Payne, B. R., Science 156, 669 (1967).
32. Washburn, H. W., Berry, C. E., and Hall, L. G., U.S. Nat!. Bur. Standards Circ.
522, 1953, p. 141.
33. Boyer, P. D., Graves, D. J., Suelter, C. H., and Dempsey, M. E., Anal. Chem. 33,
1906 (1961).
34. I1'in, L. A., Vestn. Leningr. Univ. 21, 85 (1966).
35. Majzoub, M., J. Chem. Phys. 63, 563 (1966).
36. Compston, W., and Epstein, S., Trans. Amer. Geophys. Union 39, 511 (1958).
37. Vetshtein, V. E., Khim. Prom. Ukraini 1, 38 (1966).
38. O'Neil, J. R., and Epstein, S., J. Geophys. Res. 71, 4955 (1966).
39. Trofimov, A. V., Zh. Anal. Khim. 8, 353 (1953).
40. Vol'nov, L. I., Chamova, V. N., and Salimova, K. M., ibid. 2, 1262 (1966).
41. Staschewski, D., Ber. Bunsenges. Physik. Chem. 68, 454 (1964).
42. Staschewski, D., ibid. 69, 426 (1965).
43. Bottinga, Y., J. Chem. Phys. 72, 800 (1968).
44. Dostrovsky, I., and Klein, F. S., Anal. Chem. 24, 414 (1952).
45. I1'in, L. A., Maksimov, N. A., and Pante1eeva, N. S., Vopr. Med. Khim. 12, 211
(1966).
46. Tamas, J., and Opauszky, I., Magy. Kem. Folyoirat 71, 352 (1965).
47. McCrea, J. M., J. Chem. Phys. 18, 849 (1950).
48. Bolotnikov, A. A., and Finkel'shtein, Y. B., Pribory i Tekhn. Eksperim. 9, 172 (1964).
49. Vedder, R., Kernenergie 3, 890 (1960).
50. Deines, P., Ph.D. thesis, Pennsylvania State University, 1967, 230 pp.
51. Sharma, T., and Clayton, R. N., Geochim. Cosmochim. Acta 29, 1347 (1965).
52. Epstein, S., Graf, D. L., and Degens, E. T., in Cosmic and Isotopic Chemistry,
H. Craig, S. L. Miller, and G. J. Wasserburg, eds., North Holland, Amsterdam,
1964, p. 169.
53. Epstein, S., Buchsbaum, R., Lowenstam, H., and Urey, H. C., Bull. Geol. Soc.
Amer. 62, 417 (1951).
54. Epstein, S., Buchsbaum, R., Lowenstam, H., and Urey, H. C., ibid. 64, 1315 (1953).
55. Galimov, E. M., Tr. Mosk. Inst. Ne/tekhim. i Gaz. Prom. 56, 92 (1966).
56. Clayton, R. N., J. Chem. Phys. 34, 724 (1961).
392 J. N. Weber and P. Deines

57. Anbar, M., and Guttman, S., Intern. J. Appl. Radiation Isotopes 5, 233 (1959).
58. Shakhashiri, B. Z., and Gordon, G., Talanta 13, 142 (1966).
59. Baertschi, P., and Schwander, H., He/v. Chim. Acta 35, 1748 (1952).
60. Schwander, H., Geochim. Cosmochim. Acta 4, 261 (1953).
61. Dontsova, E. I., Geokhimiya, 824 (1959), in translation.
62. Longinelli, A., and Craig, H., Science 156, 56 (1967).
63. Rafter, T. A., New Zealand J. Sci. 10, 493 (1967).
64. Baertschi, P., and Silverman, S. R., Geochim. Cosmochim. Acta 1, 317 (1951).
65. Clayton, R. N., and Mayeda, T. K., ibid. 27, 43 (1963).
66. Tudge, A. P., ibid. 18, 81 (1960).
67. Maass, I., Chern. Tech. 10, 17 (1958).
68. Maass, I., Kernenergie 3, 883 (1960).
69. Craig, H., Geochim. Cosmochim. Acta 3, 53 (1953).
70. Eckelmann, W. R., Broecker, W. S., Whitlock, D. W., and Allsup, J. R., Bull. Amer.
Assoc. Petrol. Geol. 46, 699 (1962).
71. Nesmelova, Z. N., and Dubrova, N. V., Tr. Vses. Neft. Nauchn.-Issled. Geologo-
razved. Inst. 227, 250 (1964).
72. Zhurov, Y. A., and Karpov, A. K., Nauchn.-Tekhn. Sb. Po Geol. Razrabotke i
Transp. Prirodn. Gaza 71 (1963).
73. Keith, M. L., Anderson, G. M., and Eichler, R., Geochim. Cosmochim. Acta 28,
1757 (1964).
74. Van Slyke, D. D., Folch, J., and Plazin, J., J. BioI. Chern. 136, 509 (1940).
75. Grosse, A. V., and Kirshenbaum, A. D., Anal. Chern. 24, 584 (1952).
76. Swain, C. G., Tsuchihashi, G., and Taylor, L. J., ibid. 35, 1415 (1963)
77. Rittenberg, D., and Ponticorvo, L., Intern. J. Appl. Radiation Isotopes 1, 208 (1956).
78. Jordan, R. B., and Odell, A. L., Anal. Chern. 39, 681 (1967).
79. Keeling, C. D., Geochim. Cosmochim. Acta 13, 322 (1958).
80. Keeling, C. D., ibid. 24, 277 (1961).
81. Craig, H., and Keeling, C. D., ibid. 27, 549 (1963).
82. Naughton, J. J., and Terada, K., Science 120, 580 (1954).
83. Thode, H. G., Can. Petrol. Geol. Bull. 12, 242 (1965).
84. Bigeleisen, J., Science 147, 463 (1965).
85. Weber, J. N., Bibliography: Geochemistry of the Stable Isotopes of Carbon and
Oxygen, Materials Research Lab. Monogr. 1, Pennsylvania State Univ., 125 pp.
86. Thode, H. G., Wanless, R. K., and Wallouch, R., Geochim. Cosmochim. Acta 5,
286 (1954).
87. Engel, A. E. J., Clayton, R. N., and Epstein, S., Intern. Geol. Congr. Sympos. Geo-
chern. Prosp., 1956, p. 3.
88. Cheney, E. S., and Jensen, M. L., Econ. Geol. 61, 44 (1966).
89. Perry, E. C., Jr., and Bonnichsen, B., Science 153, 528 (1966).
90. James, H. L., Bull. Geol. Soc. Amer. 70, 1623 (1959).
91. Petrovskaya, N. V., and Grinenko, L. N., Geol. Rudn. Mestorozhd. 2, 3 (1962).
92. Belevtsev, Y. N., Lugovaya, I. P., and Mel'nik, Y. P., Dokl. Akad. Nauk SSSR
173, 678 (1967).
93. Clayton, R. N., Nat!. Acad. Sci.-Nat!. Res. Council Pub!. 1075, 1963, p. 185.
94. Clayton, R. N., in Studies in Analytical Geochemistry, D. M. Shaw, ed., Roy. Soc.
Can. Spec. Pub!. 6, 1963, p. 42.
95. Clayton, R. N., and Epstein, S., J. Geol. 66, 352 (1958).
12. Mass Spectrometry 393

96. Clayton, R. N., and Epstein, S., ibid. 69, 447 (1961).
97. O'Neil, J. R., and Clayton, R. N., in Cosmic and Isotopic Chemistry, H. Craig,
S. L. Miller, and G. J. Wasserburg, eds., North Holland, Amsterdam, 1964, p. 157.
98. Sharma, T., Mueller, R. F., and Clayton, R. N., J. Geol. 73, 664 (1965).
99. Silverman, S. R., Amer. Assoc. Petrol. Geol. Bull. 44, 1256 (1960).
100. Silverman, S. R., in Cosmic and Isotopic Chemistry, H. Craig, S. L. Miller, and G. J.
Wasserburg, eds., North Holland, Amsterdam, 1964, p. 92.
101. Silverman, S. R., and Epstein, S., Amer. Assoc. Petrol. Geol. Bull. 42, 998 (1958).
102. Dakhnov, Y. N., and Galimov, E. M., Sov. Geol. 9, 117 (1966).
103. Dakhnov, Y. N., and Galimov, E. M., Geol. Nefti i Gaza 10, 59 (1966).
104. Colombo, U., Gazzarrini, F., Gonfiantini, R., Sironi, G., and Tongiorgi, E., in
Advances in Organic Geochemistry, Pergamon Press, London, 1966, p. 279.
105. Colombo, U., Gazzarrini, F., Sironi, G., Gonfiantini, R., and Tongiorgi, E., Nature
205, 1303 (1965).
106. Compston, W., Geochim. Cosmochim. Acta 18, 1 (1960).
107. Grinberg, I. Y., and Petrikovskaya, M. E., Naukova Dumka Kiev, 148 (1965).
108. Jeffery, P. M., Compston, W., Greenhalgh, D., and de Laeter, J., Geochim. Cosmo-
chim. Acta 7, 255 (1955).
109. Nakai, N., J. Earth Sci. Nagoya Univ. 8, 174 (1960).
110. Sackett, W. M., Marine Geol. 2, 173 (1964).
111. Sackett, W. M., Eckelmann, W. R., Bender, M. L., and Be, A. W. H., Science 148,
235 (1965).
112. Sackett, W. M., Eckelmann, W. R., Bender, M. L., and Be, A. W. H., Erdoel Kohle
19, 562 (1966).
113. Sackett, W. M., and Thompson, R. R., Amer. Assoc. Petrol. Geol. Bull. 47, 525
(1963).
114. Abelson, P. H., and Hoering, T. C., Geol. Soc. Amer. Bull. 71, 1811 (1960).
115. Abelson, P. H., and Hoering, T. C., Proc. Nat!. A cad. Sci. U.S. 47, 623 (1961).
116. Rankama, K., Geol. Soc. Amer. Bull. 59, 389 (1948).
117. Rankama, K., J. Geol. 56, 199 (1948).
118. Rankama, K., ibid. 58, 75 (1950).
119. Rankama, K., Science 119, 506 (1954).
120. Rankama, K., Geochim. Cosmochim. Acta 5, 142 (1954).
121. Rankama, K., Bull. Comm. Geol. Finlande 166, 5 (1954).
122. Rankama, K., Econ. Geol. 49, 541 (1954).
123. Hoefs, J., and Schidlowski, M., Science 155, 1096 (1967).
124. Hoering, T. C., Geo!. Soc. Amer. Spec. Paper 68, 1961, p. 199.
125. Hoering, T. c., Carnegie Inst. Washington Year Book 61, 1962, p. 190.
126. Hoering, T. C., Nat!. Acad. Sci.-Nat!. Res. Council Pub!. 1075, 1963, p. 196.
127. Bondesen, E., Pedersen, K. R., and Jorgensen, 0., Medd. Gronland 164, 1 (1967).
128. Prashnowsky, A. A., and Schidlowskii, M., Nature 216, 560 (1967).
129. Marmo, Y., Geol. Foren. i Stockholm Forh. 75, 89 (1953).
130. Mars, K. E., J. Geol. 59, 131 (1951).
131. Meinschein, W. G., Science 150, 601 (1965).
132. Welin, E., Geol. Foren. i Stockholm Forh. 87, 509 (1965).
133. Craig, H., Econ. Geol. 48, 600 (1953).
134. Craig, H., Geochim. Cosmochim. Acta 6, 186 (1955).
135. Schutz, D. F., Isotopics 2, 1 (1965).
394 J. N. Weber and P. Deines

136. Craig, H., Science 154, 1544 (1966).

137. Craig, H., and Gordon, L. I., Univ. Rhode Is. Occ. Pub!. 3, 1965, p. 277.
138. Dansgaard, W., Deep-Sea Res. 6, 346 (1960).
139. Eriksson, E., and Bolin, B., U.S. Atom. Energy Comm. Sympos. Ser. 5, 1965,
p. 675.
140. Lloyd, R. M., Geochim. Cosmochim. Acta 30, 801 (1966).
141. Weber, J. N., Geokhimiya p. 674 (1965).
142. Deevey, E. S., Jr., and Stuiver, M., Limnol. Oceanogr. 9, 1 (1964).
143. Deevey, E. S., Jr., Stuiver, M., and Nakai, N., Intern. Assoc. Theor. Appl. Limnol.
Congr. Proc. 15, 284 (1964).
144. Oana, S., and Deevey, E. S., Jr., Amer. J. Sci. 258, 253 (1960).
145. Aldaz, L., and Deutsch, S., Earth Planet. Sci. Letters 3, 267 (1967).
146. Dansgaard, W., Fysisk Tidsskrift, 49 (1958).
147. Deutsch, S., Ambach, W., and Eisner, H., Earth Planet. Sci. Letters 1, 197 (1966).
148. Epstein, S., Nat!. Acad. Sci.-Nat!. Res. Council Pub!. 845, 1960, p. 102.
149. Epstein, S., and Sharp, R. P., J. Geol. 67, 88 (1959).
150. Epstein, S., and Sharp, R. P., J. Geophys. Res. 72, 5595 (1967).
151. Epstein, S., Sharp, R. P., and Goddard, I., J. Geol. 71, 698 (1963).
152. Epstein, S., Sharp, R. P., and Gow, A. J., J. Geophys. Res. 70, 1809 (1965).
153. Epstein, S., Sharp, R.P., and Vidziunas, I., ibid. 65, 4043 (1960).
154. Gonfiantini, R., ibid. 70, 1815 (1965).
155. Gonfiantini, R., and Picciotto, E., Nature 184, 1557 (1959).
156. Gonfiantini, R., Togliatti, V., Tongiorgi, E., de Breuck, W., and Picciotto, E., ibid.
197, 1096 (1963).
157. Gonfiantini, R., Togliatti, V., Tongiorgi, E., de Breuck, W., and Picciotto, E., J.
Geophys. Res. 68, 3791 (1963).
158. Lorius, c., Polar Record 12, 211 (1964).
159. Macpherson, D. S., and Krouse, H. R., Nat!. Acad. Sci.-Nat!. Res. Council Pub!.
1488, 1967, p. 180.
160. Picciotto, E., Deutsch, S., and Aldaz, L., Earth Planet. Sci. Letters 1, 202 (1966).
161. Bodvarsson, G., Joku1l12, 49 (1962).
162. Clayton, R. N., Friedman, I., Graf, D. L., Mayeda, T. K., Meents, W. F., and
Shimp, N. F., J. Geophys. Res. 71, 3869 (1966).
163. Craig, H., Science 133, 1703 (1961).
164. Degens, E. T., Geol. Rundschau 52, 625 (1962).
165. Epstein, S., in Researches in Geochemistry, P. H. Abelson, ed., John Wiley & Sons,
New York, 1959, p. 217.
166. Graf, D. L., Friedman, I., and Meents, W. F., II!. State Geo!. Surv. Circ. 393, 1965.
167. Banwell, C. J., in Nuclear Geology on Geothermal Areas, E. Tongiorgi, ed., Lab.
Geo!. Nuc!. Pisa, Italy, 1963, p. 95.
168. Ferrara, G. c., Ferrara G., and Gonfiantini, R., ibid., p. 277.
169. Ferrara, G., and Gonfiantini, R., Trans. Amer. Geophys. Union 46, 170 (1965).
170. Hulston, J. R., and McCabe, W. J., Geochim. Cosmochim. Acta 26, 383 (1962).
171. Craig, H., in Nuclear Geology on Geothermal Areas, E. Tongiorgi, ed., Lab. Geo!.
Nuc!. Pisa, Italy, 1963, p. 17.
172. Craig, H., Boato, G., and White, D. E., Nat!. Acad. Sci.-Nat!. Res. Council Pub!.
400, 1956, p. 29.
12. Mass Spectrometry 395

173. Fontes, J. C., Glangeaud, L., Gonfiantini, R., and Tongiorgi, E., A cad. Sci. Paris
C.R. 256, 472 (1963).
174. White, D. E., Craig, H., and Begemann, F., in Nuclear Geology on Geothermal Areas,
E. Tongiorgi, ed., Lab. Geol. Nucl. Pisa, Italy, 1963, p. 9.
175. Allen, P., and Keith, M. L., Nature 208, 1278 (1965).
176. Baertschi, P., Schweiz. Mineral. Petrog. Mitt. 37, 73 (1957).
177. Clayton, R. N., and Degens, E. T., Amer. Assoc. Petrol. Geol. Bull. 43, 890 (1959).
178. Keith, M. L., and Parker, R. H., Marine Geol. 3, 115 (1965).
179. Keith, M. L., and Weber, J. N., Geochim. Cosmochim. Acta 28, 1787 (1964).
180. Keith, M. L., and Weber, J. N., Science 150, 498 (1965).
181. Letolle, R., and Tivollier, J., A cad. Sci. Paris C.R. 263, 1824 (1966).
182. Lloyd, R. M., J. Geol. 72, 84 (1964).
183. Taft, W. H., and Harbaugh, J. W., Stanford Univ. Publ. Geol. Sci. 8, 1 (1964).
184. Weber, J. N., Geochim. Cosmochim. Acta 31,2342 (1967).
185. Weber, J. N., Amer. J. Sci. 265, 586 (1967).
186. Weber, J. N., Geochim. Cosmochim. Acta 32, 33 (1968).
187. Weber, J. N., Bergenback, R. E., Williams, E. G., and Keith, M. L., J. Sed. Petrol.
35, 36 (1965).
188. Weber, J. N., and Raup, D. M., Geochim. Cosmochim. Acta 30, 681 (1966).
189. Weber, J. N., Williams, E. G., and Keith, M. L., J. Sed. Petrol. 34, 814 (1965).
190. Lowenstam, H. A., and Epstein, S., J. Geol. 62, 207 (1954).
191. Urey, H. C., Lowenstam, H. A., Epstein, S., and McKinney, C. R., Geol. Soc. Amer.
Bull. 62, 399 (1951).
192. Bowen, R., Paleotemperature Analysis, Elsevier, Amsterdam, 1966, 265 pp.
193. Bowen, R., Earth Sci. 2, 199 (1966).
194. Emiliani, C., Science 154, 851 (1966).
195. Emiliani, c., J. Geol. 74, 109 (1966).
196. Donn, W. L., and Shaw, D. M., ibid. 75, 497 (1967).
197. Donn, W. L., and Shaw, D. M., Science 157, 722 (1967).
198. Shackleton, N. J., Nature 215, 15 (1967).
199. Shackleton, N. J., ibid. 218, 79 (1968).
200. Longinelli, A., ibid. 207, 716 (1965).
201. Longinelli, A., ibid. 211, 923 (1966).
202. Clayton, R. N., Jones, B. F., and Berner, R. A., Geochim. Cosmochim. Acta 32,
415 (1968).
203. Degens, E. T., and Epstein, S., ibid. 28, 23 (1964).
204. Russell, K. L., Deffeyes, K. S., Fowler, G. A., and Lloyd, R. M., Science 155, 189
(1967).
205. Spotts, J. H., and Silverman, S. R., Amer. Min. 51, 1144 (1966).
206. Weber, J. N., Science 145, 1303 (1964).
207. Weber, J. N., Geochim. Cosmochim. Acta 28, 1257 (1964).
208. Bonatti, E., Science 153, 534 (1966).
209. Friedman, I., and Hall, W. E., J. Geol. 71, 238 (1963).
210. Northrop, D. A., and Clayton, R. N., ibid. 74, 174 (1966).
211. O'Neil, J. R., and Epstein, S., Science 152, 198 (1966).
212. Schwarcz, H. P., J. Geol. 74, 38 (1966).
213. O'Neil, J. R., and Epstein, S., Trans. Amer. Geophys. Union 45, 113 (1964).
214. Borshchevskiy, Y. A., and Khristianov, V. K., Geokhimiya, 844 (1965).
396 J. N. Weber and P. Deines

215. Fontes, J. C., Geol. Rundschau 55, 172 (1965).

216. Fontes, J. C., and Nielson, H., A cad. Sci. Paris C.R. 262, 2685 (1966).
217. Gonfiantini, R., and Fontes, J. C., Nature 200, 644 (1963).
218. Uklonskii, A. S., Tsapenko, M. N., and Gluschchenko, V. M., Dokl. Akad. Nauk
UzSSR 23, 44 (1966).
219. Berner, R. A., Science 159, 195 (1967).
220. Gross, M. G., J. Geol. 72, 170 (1964).
221. Gross, M. G., and Tracey, J. I., Jr., Science 151, 1082 (1966).
222. Hodgson, W. A., Geochim. Cosmochim. Acta 30, 1223 (1966).
223. Murata, K. J., Friedman, I. I., and Madsen, B. M., Science 156, 1484 (1967).
224. Degens, E. T., Neues Jahrb. Geol. Pal. Monatsh., 72 (1959).
225. Dontsova, E. I., Geokhimiya, 430 (1966).
226. Dontsova, E. I., and Milovskii, A. V., Geokhimiya, 655 (1967).
227. Garlick, G. D., Earth Planet. Sci. Letters 1, 361 (1966).
228. Taylor, H. P., Jr., and Epstein, S., Geol. Soc. A mer. Bull. 73, 461 (1962).
229. Taylor, H. P., Jr., and Epstein, S., J. Petrol. 4, 51 (1963).
230. Deines, P., Geochim. Cosmochim. Acta 32, 613 (1968).
231. Kukharenko, A. A., and Dontsova, E. I., Econ. Geol. USSR 1, 31 (1964).
232. Taylor, H. P., Jr., Frechen, J., and Degens, E. T., Geochim. Cosmochim. Acta 31,
407 (1967).
233. Vinogradov, A. P., Kropotova, O. I., Epstein, E. M., and Grinenko, V. A., Geokhi-
miya, 499 (1967).
234. Vinogradov, A. P., Kropotova, O. I., Orlov, Y. L., and Grinenko, V. A., Geokhi-
miya, 1395 (1966).
235. Anderson, A. T., J. Geol. 75, 323 (1967).
236. Hahn-Weinheimer, P., Geol. Rundschau 55, 197 (1965).
237. Paulitsch, P., and Hahn-Weinheimer, P., Naturwiss. 48, 597 (1961).
238. Garlick, G. D., and Epstein, S., Geochim. Cosmochim. Acta 31, 181 (1967).
239. Schwarcz, H. P., Geol. Soc. Amer. Bull. 77, 879 (1966).
240. Schwarcz, H. P., and Clayton, R. N., Can. J. Earth Sci. 2, 72 (1965).
241. Taylor, H. P., Jr., Geol. Soc. Amer. Spec. Paper 76, 1963, p. 163.
242. Taylor, H. P., Jr., Albee, A. L., and Epstein, S., J. Geol. 71, 513 (1963).
243. Boato, G., Geochim. Cosmochim. Acta 6, 209 (1954).
244. Clayton, R. N., Science 140, 192 (1963).
245. Trofimov, A. V., Dokl. Akad. Nauk SSSR 72, 663 (1950).
246. Reuter, J. H., Epstein, S., and Taylor, H. P., Jr., Geochim. Cosmochim. Acta 29,
481 (1965).
247. Taylor, H. P., Jr., Duke, M. B., Silver, L. T., and Epstein, S., ibid. 29, 489 (1965).
248. Taylor, H. P., Jr., and Epstein, S., in Cosmic and Isotopic Chemistry, H. Craig,
S. L. Miller, and G. J. Wasserburg, eds., North Holland, Amsterdam, 1964, p. 181.
249. Taylor, H. P., Jr., and Epstein, S., Science 153, 173 (1966).
250. Vinogradov, A. P., Dontsova, E. I., and Chupakhin, M. S., Geochim. Cosmochim.
Acta 18, 278 (1960).
251. Vinogradov, A. P., Kropotova, O. I., Vdovykin, G. P., and Grinenko, V. A., Geo-
khimiya, 267 (1967).
252. Walter, L. S., and Clayton, R. N., Science 156, 1357 (1967).
253. Keith, M. L., and Anderson, G. M., Science 141, 634 (1963).
254. Pearson, F. J., Jr., U.S. Atom. Energy Comm. Rept. CONF-650652, 1966, p. 357.
12. Mass Spectrometry 397

255. Hoering, T. C., and Bufalini, M., Carnegie Inst. Washington Papers Geophys. Lab.
7, 205 (1961).
256. Cassie, G. E., Cuthbert, J., and Silk, c., J. Sci. Instr. 43, 283 (1966).
257. McCullough, H., and Krouse, H. R., Rev. Sci. Instr. 36, 1132 (1965).
258. Hagan, P. J., and de Laeter, J. R., J. Sci. Instr. 43, 662 (1966).
259. Moreland, P. E., Jr., Stevens, C. M., and Waling, D. B., Rev. Sci. Instr. 38, 760
(1967).
260. Ridley, R. G., and Young, W. A. P., U.K. Atom. Energy Auth. Rept. AWRE 0-37,
1965.
261. Ahearn, A. J., ed., Mass Spectrometric Analysis of Solids, Elsevier, New York, 1966.
262. Deines, P., The Principles of Spark Source Mass Spectrography and its Application
to the Determination of Trace Element Concentrations, Experiment Station Cir-
cular 78, The College of Earth and Mineral Science, The Pennsylvania State Uni-
versity, 1970.
263. Leipziger, F. D., Appl. Spectry. 17, 158 (1963).
264. Leipziger, F. D., and Croft, W. J., Geochim. Cosmochim. Acta 28, 268 (1964).
265. Brown, R., and Wolstenholme, W. A., Paper 75, presented at the 11th Annual
Conference on Mass Spectrometry and Allied Topics, San-Francisco, 1963.
266. Brown, R., and Wolstenholme, W. A., Nature 201, 598 (1964).
267. Herzog, L. F., Eskew, T. J., and Deines, P., Trans. Am. Geophys. Union 45, 112
(1964).
268. Herzog, L. F., Eskew, T. J., and Deines, P., The Pennsylvania State University
Mineral Industries Experiment Station Special Report No. 1-65, 1965.
269. Cookson, G. G., Fletcher, W., and Tushingham, R., Conf. Ind. Carbon and Gra-
phite, 1965, London, 1966, p. 77.
270. Taylor, S. R., Geochim. Cosmochim. Acta 29, 1243 (1965).
271. Taylor, S. R., Nature 205, 34 (1965).
272. Nicholls, G. D., Graham, A. L., Williams, E., and Wood, M., Anal. Chem. 39, 584
(1967).
273. Taylor, S. R., Geochim. Cosmochim. Acta 30, 1121 (1966).
274. Taylor, S. R., and White, A. R. J., Bull. Volcanol. 29, 177 (1966).
275. Ewart, A., Taylor, S. R., and Capp, A. C., Contr. Mineral. and Petrol. 18, 76 (1968).