Adhesion and Adsorption of Polymers Compress

POLYMER SCIENCE AND TECHNOLOGY
Volume 12A
ADHESION AND
ADSORPTION
OF POLYMERS
Editorial Board:
William J. Bailey C. A. J. Hoeve

University of Maryland Texas A&M University
College Park, Maryland College Station, Texas
Yoichi Ishida
J. P. Berry Osaka University
Rubber and Plastics Research Association Toyonaka, Osaka, Japan
of Great Britain
Shawbury Frank E. Karasz
University of Massachusetts
Shrewsbury, England
Amherst, Massachusetts
A. T. DiBenedetto Osias Solomon
The University of Connecticut Poly technical Institute of Bucharest
Storrs, Connecticut Bucharest, Romania
Volume 1 • STRUCTURE AND PROPERTI ES OF POLYMER FI LMS

Edited by Robert W. Lenz and Richard S. Stein
Volume 2 • WATER-SOLUBLE POLYMERS

Edited by N. M. Bikales
Volume 3 • POLYMERS AND ECOLOGICAL PROBLEMS

Edited by James Guillet
Volume 4 • RECENT ADVANCES IN POLYMER BLENDS, GRAFTS, AND BLOCKS

Edited by L. H. Sperling
Volume 5 • ADVANCES IN POLYMER FRICTION AND WEAR (Parts A and BI

Edited by Lieng-Huang Lee
Volume 6 • PERMEABILITY OF PLASTIC FILMS AND COATINGS

TO GASES, VAPORS, AND LIQUIDS
Edited by Harold B. Hopfenberg
Volume 7 • BIOMEDICAL APPLICATIONS OF POLYMERS

Edited by Harry P. Gregor
Volume 8 • POLYMERS IN MEDICINE AND SURGERY

Edited by Richard L. Kronenthal, Zale Oser, and E. Martin
Volume 9 • ADHESION SCIENCE AND TECHNOLOGY (Parts A and BI

Volume 10 • POLYMER ALLOYS: Blends, Blocks, Grafts, and Interpenetrating Networks

Edited by Daniel Klempner and Kurt C. Frisch
Volume 11 • POLYMER ALLOYS II: Blends, Blocks, Grafts, and Interpenetrating Networks
Edited by Daniel Klempner and Kurt C. Frisch
Volume 12. ADHESION AND ADSORPTION OF POLYMERS (Parts A and BI

A Continuation Order Plan is available for this series. A continuation order will bring
delivery of each new volume immediately upon publication. Volumes are billed only upon
actual shipment. For further information please contact the publisher.
Volume 12A
ADHESION AND
ADSORPTION
OF POLYMERS
Edited by
Lieng - Huang Lee
Xerox Corporation
Rochester, New York
PLENUM PRESS· NEW YORK AND LONDON

Library of Congress Cataloging in Publication Data
International Conference on Adhesion and Adsorption of Polymers, Honolulu, 1979.

Adhesion and adsorption of polymers.
(Polymer science and technology; v. 12A-B)

Includes indexes.
1. Polymers and polymerization-Congresses. 2. Adhesion-Congresses. 3. Adsorp-
tion-Congresses. I. Lee, Lieng-Huang, 1924- II. Title. III. Series.
00380.153 1979 547.8'4 80-262
ISBN-13:978-1-4613-3095-0 e-ISBN-13:978-1-4613-3093-6
001: 10.1007/978-1-4613-3093-6
First half of the proceedings of an International Conference on

Adhesion and Adsorption of Polymers, Honolulu, Hawaii, April 2-6, 1979
© 1980 Plenum Press, New York

Softcover reprint of the hardcover 1st edition 1980
A Division of Plenum Publishing Corporation
227 West 17th Street, New York, N.Y. 10011
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted,

in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, without written permission from the publisher
PREFACE
Honolulu is a most beautiful place, suitable for all occa-

sions. Its choice as the meeting site for the first Joint Chemical
Congress between the American Chemical Society and the Chemical
Society of Japan was praised by scientists from both sides. During
this Congress, the International Conference on Adhesion and
Adsorption of Polymers was held at the Hyatt Regency Hotel between
April 2 and 5, 1979. We had speakers from ten nations presenting
over forty papers related to the subject matter. It was a memorable
event.
Unlike our two previous adhesion symposia held in 1971 and
1975, this was the first time in the same conference that we
discussed both adhesion and adsorption of polymers simultaneously.
These two important phenomena are not only inter-related, but also
equally important in adhesive technology as well as biochemical
processes. The papers presented to this Conference deal with these
two phenomena from both fundamental and practical viewpoints.
Furthermore, with the advance of new surface analytical techniques,
the actual, microscopic happenings at the interfaces can be pin-
pointed. Thus, characterization of interface became one of the
major focuses of this Conference. As a result, a broad coverage of
the subject matter includes statistical thermodynamics, surface
physics, surface analysis, fracture mechanics, viscoelasticity,
failure analysis, surface modification, adsorption kinetics, bio-
polymer adsorption, etc. Thanks to the diligence of our contri-
butors, we are now able to publish the final papers in these two
volumes.
v
vi PREFACE
Fig . 1. From Left to Right: Dr. T. Hata,

Mrs. and Dr. L. H. Lee, Dr. K. Nakao .
Fig. 2. From Left to Right: Dr. and Mrs. A.

Silberberg, Dr. L. H. Lee.
PREFACE vii
In these Proceedings we present the revised papers and discus-

sions during the Conference in eight parts:
Part I: Polymer Surface Interactions

Part II: Characterization of Adhesive Interfaces
Part III: Polymeric Structural Adhesives
Part IV: Fracture Strengths of Polymeric Systems
Part V: Modification of Polymer Interfaces
Part VI: Kinetics of Polymer Adsorption
Part VII: Characterization of Adsorbed Interfaces
Part VIII: Adsorption of Biopolymers
Several papers were published elsewhere, so, instead, we have

included two "Quickies" and one contribution after the Conference.
I would like to take this opportunity to thank our co-

chairmen: Dr. K. Nakao (Fig. 1) and Dr. W.H. Grant. They helped
very much in soliciting papers for this Conference. We would like
to thank our plenary speakers: Professors I. Prigogine and A.
Bellemans, Professor T. Hata (Fig. 1), Dr. S. Tsai, and Professor A.
Silberberg (Fig. 2). We are grateful to Dr. Shiro Matsuoka for his
talk at our Symposium Banquet on "Energy Conservation Problems for
Future Plastics Industries." We also wish to thank our sponsoring
organizations, the Division of Organic Coatings and Plastics Chem-
istry of the American Chemical Society and the Chemical Society of
Japan.
Acknowledgement is made to the donors of the Petroleum Re-

search Fund, administered by the American Chemical Society, for
partial support of this Symposium.
Last but not least, I would like to thank Ms. Nancy Rickey for
her patience in typing the entire Proceedings in camera-ready
format.
Lieng-Huang Lee
November, 1979
Contents of Volume 12A
PART ONE: Polymer Surface Interactions
Introductory Remarks 3
LlENG-HU ANG LEE
Statistical Mechanics of Surface Tension and Adsorption 5

I. PRIGOGINE AND A. BELLEMANS
Surface and Interfacial Tensions of Polymer Melts

and Solutions 15
TOSH1O HATA AND TOMOYUKI KASEMURA
Donor-Acceptor Interactions at Interfaces 43
F.M. FOWKES
Surface Tension of Solids: Generalization and Reinter-
pretation of Critical Surface Tension 53
SOUHENG WU
A Role of Molecular Forces in Adhesive Interactions of

Polymers 67
V.A. BELYI, V.A. SMURUGOV AND A.T. SVIRIDYONOK
Discussion 77
PART TWO: Characterization of Adhesive

Interfaces
LlENG-HU ANG LEE
ix
x CONTENTS OF VOLUME 12A
Photoacoustic Spectroscopy for the Study of Adhesion

and Adsorption of Dyes and Polymers 87
LIENG-HU ANG LEE
Adhesive-Adherend Bond Joint Characterization by Auger

Electron Spectroscopy and Photoelectron Spectroscopy 103
J.S. SOLOMON, D. HANLIN AND N.T. McDEVITT
Techniques for Measuring Adhesive vs. Cohesive Failure 123

TENNYSON SMITH AND PIERRE SMITH
Failure Characterization of a Structural Adhesive 141

DAVID W. DWIGHT, EROL SANCAKTAR AND HALBERT F. BRINSON
Dielectric Relaxation Gradients in an Adhesive Bond 165

JOHN L. CROWLEY AND ARTHUR D. JONATH
New Approach to the Understanding of Adhesive Interface

Phenomena and Bond Strength 175
ARTHUR D. JONATH
Discussion 195
PART THREE: Polymeric Structural Adhesives

D.H. KAELBLE
Fracture Mechanics and Adherence of Viscoelastic Solids 203

D. MAUGIS AND M. BARQUINS
The Viscoelastic Shear Behavior of a Structural Adhesive 279

EROL SANCAKTAR AND H.F. BRINSON
Surface Energetics and Structural Reliability of Adhesive

Bonded Metal Structures 301
D.H. KAELBLE
Composition and Ageing of a Structural, Epoxy Based Film

Adhesive 313
C.E.M. MORRIS, A.G. MORITZ AND R.G. DAVIDSON
Viscoelastic Characterization of Structural Adhesive via

Force Oscillation Experiments 321
D.L. HUNSTON, W.D. BASCOM, E.E. WELLS, J.D. FAHEY
AND J.L. BITNER
CONTENTS OF VOLUME 12A xi
Crack Healing in Semicrystalline Polymers, Block Copolymers

and Filled Elastomers 341
RICHARD P. WOOL
Discussion 363
PART FOUR: Fracture Strengths in Polymeric

Systems

KAZU MUNE NAKAO
Criterion of Interfacial Fracture on Tensile Adhesive Joint 371

MINEO MASUOKA AND KAZUMUNE NAKAO
Deformation and Shear Strength of FRP Adhesive Joints 393

HIROO MIY AIRI, HIDEAKI FUKUDA AND
ATSUYOSHI MU RAMATSU
Stress Distribution and Strength in Shear on the Adhesive

Lap Joint Loading Bending M9ment 413
YUKISABURO YAMAGUCHI AND SUSUMU AMANO
Fracture Criteria on Peeling 421

T AKAH ASHI IGARASHI
Superposition of Peel Rate, Temperature and Molecular Weight

for T Peel Strength of Polyisobutylene 439
TOSHIYA TSUJI, MINEO MASUOKA AND KAZUMUNE NAKAO
Discussion 455
Author Index xvii
Subject Index xxxi

Contents of Volume 12B
PART FIVE: Modification of Polymer Interfaces

L.H. SHARPE
Role of Interface in the Strength of Composite Materials 463

S.W. TSAI AND H.T. HAHN
Surface Modification of Polyethylene by Radiation-Induced

Grafting for Adhesive Bonding 473
SHlNZO YAMAKAWA
Adhesive Tack 497

R.BATES
Adhesion Measurement of Polymeric Films and Coatings

with Special Reference to Photoresist Materials 503
K.L. MITTAL AND R.O. LUSSOW
Adhesion of Thin Plasma-Polymerized Films to Metals 521

PABITRA DATTA AND GRZEGORZ KAGANOWICZ
Adhesion of Filled Polyolefins to Metals and Glass 539

V.A. BELYI, N.I. EGORENKOV AND A.I. KUZAVKOV
Adhesive Characteristics of Ultraviolet Radiation Curable

Resins 551
K. NATE AND T. KOBAY ASHI
An Evaluation of the Phenomena and Their Final Effects

Resulting from a Corona Discharge on Low Density
Polyethylene 563
R. VAN DER LINDEN
Discussion 577
xiii
xiv CONTENTS OF VOLUME 128
PART SIX: Kinetics of Polymer Adsorption

WARREN H. GRANT
Adsorption of Polymers from the Melt and Solutions of

Finite Concentration 591
A. SILBERBERG
On the Structure of Boundary Layers of Polymers on

Solid Surfaces 601
YURY S. LIPATOV
Selective Adsorption of Polymers from Solution 629

RYONG-JOON ROE
Reaction in Models of Bio-Membranes in Connection with

Cell Motion 643
A. SANFELD, A. STEINCHEN, W. DALLE VEDOVE AND P.M. BISCH
Kinetics of Cell Adhesion to a Hydrophilic-Hydrophobic

Copolymer Model System 667
JONATHAN J. ROSEN AND MICHAEL B. SCHWAY
The Kinetics of Adsorption of Plasma Proteins to a Series

of Hydrophilic-Hydrophobic Copolymers 677
THOMAS A. HORBETT
Discussion 683
PART SEVEN: Characterization of Adsorbed

Interfaces

D.L. ALLARA
Surface Characterization of Hydrophilic-Hydrophobic Copolymer

Model Systems I. A Preliminary Study 691
BUDDY D. RATNER AND ALLAN S. HOFFMAN
Characterization of Adsorption Polymers at the Liquid-Solid

and Liquid-Liquid Interfaces by Interfacial Tension 707
TSUNETAKA MATSUMOTO AND KATSUHIKO NAKAMAE
CONTENTS OF VOLUME 128 xv
Adsorption of Polyelectrolyte Studied by Ellipsometry 729

AKIRA TAKAHASHI, MASAMI KAWAGUCHI AND TADAYA KATO
Adsorption Studies of Polymers on Metals by ~ourier

Transform Reflection Infrared Spectroscopy 751
D.L. ALLARA
Role of Coupling Agents in Metal-Polymer Adhesion: 1.

The Structure of the Silane Film at Metal-Polymer Interface 757
C.S. PAIK SUNG, S.H. LEE AND N.H. SUNG
Microstructure, Orientation, and Mechanical Properties

of Polytetrafluoroethylene 775
THEODORE DAVIDSON AND R.J. GOUNDER
Discussion 791
PART EIGHT: Adsorption of Bipolymers

WARREN H. GRANT
Adsorption of Proteins at Solid Surfaces 801

WILLEM NORDE
The Adsorption of Human Serum Albumin and y-Globulin

on Hydrophobic and Hydrophilic Surfaces 827
WARREN H. GRANT AND RONALD E. DEHL
Adsorption of Soluble Collagen and Adsorption to Adsorbed

Soluble Collagen 837
A. SILBERBERG
Conformational Changes of Structure of Immunoglobulin (IgG)

and its Fragments F band F at the Air-Water and Water-Solid
Interfaces a c 847
V.V. LAVRENTEV, L. CHASOVNIKOVA AND JU. SOROKIN
Discussion 855
About the Contributors 859
Author Index 867
Subject Index 881

PART ONE
POLYMER SURFACE
INTERACTIONS
Introductory Remarks
Lieng-Huang Lee
Wilson Center for Technology
Xerox Corporation
Webster, New York 14580
This is a wonderful opportunity for us to celebrate the first

Joint Chemical Congress between the American Chemical Society and
the Chemical Society of Japan. The Organizing Committee of this
International Conference on Adhesion and Adsorption of Polymers
would like to thank you and our speakers for making this meeting
possible.
There will be eight sessions of this Conference, dealing first

with adhesion and then with adsorption of polymers. The subject of
this session is "Polymer Surface Interactions." Some aspects of
this subject matter were discussed during the two previous confer-
ences (1,2). Today, as planned in the schedule, the first paper is
a plenary lecture on "Statistical Mechanics of Surface Tension and
Adsorption" by Professors 1. Prigogine and A. BeHemans. Dr.
Bellemans is scheduled to deliver this paper; however as a result of
the United Airlines strike, he is still staying at the Los Angeles
airport and waiting for an opportunity to fly out here. In the
meantime, I have brought some extra sets of my own slides and shall
briefly discuss the general concepts of surface tension, surface
energy and surface free energy.
The second paper is another plenary lecture on "Surface and

Interfacial Tension of Polymer Melts and Solutions," to be pre-
sented by Professor Toshio Hata, President of Gunma University,
Japan. Dr. Hata attended our 1971 Symposium on "Recent Advances on
Adhesion" and has been actively publishing papers on adhesion and
rheology of polymers. It is indeed fortunate for all of us to hear
Dr. Hata again during this Joint Chemical Congress.
3
LIENG·HUANG LEE
The third paper, "Acid-Base Interactions at Interfaces." will

be presented by Professor F.M. Fowkes of Lehigh University. Physi-
cists have frequently used a similar term, donor-acceptor inter-
action. Now, we chemists will be hearing a stronger argument about
a more familiar term, acid-base interactions. In this paper, Dr.
Fowkes points out that acid-base interactions are also occurring at
the polymer-polymer interfaces.
Dr. Souheng Wu will discuss the surface tension of solids in

the fourth paper of this morning. Dr. Wu has made experimental
measurements of surface and interfacial tensions of polymers. In
this paper, he is proposing a theoretical justification for criti-
cal surface tension introduced by Dr. W.A. Zisman. Though the
original title, "Surface Tension of Solids: An Equation of State
Approach," has raised strong comments by reviewers, the data as-
sembled could serve some useful purpose for practioners.
The last paper of this session is "A Role of Molecular Forces
in Adhesive Interaction of Polymers," by Professor Belyi et al. Dr.
Belyi could not attend this Conference. We understand their
situation in the U.S.S.R. This paper is by no means rigorous, but
it sums up their view on the effect of dispersion forces on
adhesion. Those who are interested in the subject matter should
read a new book, Dispersion Forces (3).
REFERENCES
1. L.H. Lee, Ed., Recent Advances in Adhesion, Gordon and Breach,
New York, London and Paris (1973).
2. L.H. Lee, Ed., Adhesion Science and Technology, Plenum Press,
New York (1975).
3. J. Mahanty and B.W. Ninham, Dispersion Forces, Academic Press,
London, New York (1976).
Statistical Mechanics of Surface Tension
and Adsorption
I. Prigogine and A. Bellemans
Faculty of Sciences
Universite Libre de Bruxelles, 1050 Bruxelles, Belgium
ABSTRACT
This introductory lecture is broadly divided into three parts.
In the first one we review some fundamentals of surface phenomena;
1. e., the conditions of mechanical equilibrium of an interface,
Gibbs thermodynamics of surfaces and the statistical formulation of
surface properties. The second part is devoted to a critical
discussion of current statistical models of the surface tension of
pure liquids and liquid mixtures (cell model, perfect and regular
solutions, corresponding state theories, r-mer mixtures, etc ... ).
The last part deals with the adsorption of a chain molecule at an
interface, in relation with different physical parameters: inter-
actions, length and concentration.
INTRODUCTION
This lecture will be divided into three parts. In the first

one we shall briefly survey some fundamentals of interface phenom-
ena; i.e., the conditions of mechanical equilibrium, Gibbs thermo-
dynamics of surfaces and the general statistical approach of inter-
faces. The second part will be devoted to a critical presentation
of some particular models developed for describing surface proper-
~ies of pure liquids and liquid mixtures. The third and last part
will concern one of the central topics of this Colloquium; i.e., the
adsorption of a chain molecule at an interface.
5
6 I. PRIGOGINE AND A. BELLEMANS
FUNDAMENTALS OF SURFACE PHENOMENA
From the macroscopic point of view the interface between two

adjacent fluid phases, a, b appears as a geometrical surface where
the local thermodynamic properties of the system change abruptly.
As originally stated by Young (1) in 1805, this surface is mechan-
ically equivalent to a non-rigid membrane with zero thickness,
isotropically stretched by the so-called interfacial tension y •
Insofar as this membrane may be regarded as massless, its equilibri-
um shape is determined by the Laplace equation (2)
(1)
where Pa' Pb are the hydrostatic pressures in phases a, b, and K1 ,

K are the principal curvatures at the considered locus. Moreover,
t~e equilibrium condition of an edge resulting from the encounter of
the interfaces separating three mutually coexisting phases a, b, c,
is expressed by the vectorial equation of Neumann
Ya~b
a Ybc
c
Yca
(assuming zero edge tension).
Obviously the assumption that two adjacent phases retain their

own intensive properties downright to a surface of discontinuity is
only a useful way of thinking which, when necessary, must be
completed by a detailed picture of what really happens at the
molecular level. The interface is actually formed by a thin layer of
matter operating the transition between the two bulk phases and both
the mechanical and physico-chemical properties of this layer usu-
ally differ markedly from those of these phases. In particular, as
originally pointed by Bakker (3), the interfacial tension is merely
a macroscopic manifestation of the anisotropy of the pressure
tensor within the interfacial layer.
The thermodynamics of surface phenomena was formulated by

Gibbs (4) in an extremely elegant way. Indeed classical thermo-
dynamics is devised for handling locally homogeneous systems; i.e.,
the change of any intensive variable is completely negligible over a
few molecular diameters. This is clearly not so within the transi-
tion layer, the density of which may e.g., drop by a factor of a
thousand over perhaps 10 or 20K when passing from a liquid to a gas
phase. Gibbs superseded the difficulty by introducing the concept
of dividing surface. This is a geometrical surface located in the
transition layer (or at least sensibly close to it) and passing
through "points which are similarly situated with respect to the
STATISTICAL MECHANICS 7
condition of the adjacent matter", as worded by Gibbs himself (4).

The actual system, consisting of two homogeneous bulk phases sep-
arated by a thin inhomogeneous layer, is then replaced by two
homogeneous bulk phases extending on each side of the chosen
dividing surface; this surface itself is treated as a homogeneous
two-dimensional phase and provided therefore with appropriate
mechanical and physico-chemical properties, in order to match the
actual system macroscopically. E.g., its free energy F is split
into three parts, respectively the two bulk phases and the surface
phase s
F = Fa + Fb + Fs
with the following differentials (5)

a (4a)
dF a = - Sa dT - Pa dV a + Ell l' dN 1·
b
dFb = - Sb dT - Pb dV b + Elli dN i (4b)
dF s = - Ss dT + Y dA + Ell 1. dN~1 (4c)
Note that we limit ourselves to systems in mechanical, thermal and

chemical equilibrium, and that we are essentially interested in
flat interfaces (for which p = P ). Al though eq. (4c) looks
strictly similar to eqs. (4a1- (4b9, one should realize that a
surface phase has no real autonomy in general; its very existence is
conditioned by the presence of the two "parent" bulk phases and its
composition follows from the values of the set of chemical poten-
tials {ll.}.
1
The variables {N~} specify the material content and the compo-
1
sition of the surface phase. In the same way as one defines bulk
concentrations (C~1 = N~1 / Va
, C?l=lN? / Vb)' it is convenient to
introduce the adsorption
as the mean number of molecules of species i adsorbed on a unit area

of the interface. Similar arguments which lead to the well known
Gibbs-Duhem equation for the bulk, may next be used to derive the
Gibbs adsorption formula
= - s S dT - E r. d II .
1 1
(6)
where Ss is the surface entropy per unit area. Actually the value
of r i depends on the precise location of the dividing surface
(which for a flat interface is somewhat arbitrary). The relative
adsorption of one component with respect to another is, however, a

perfectly well defined quantity, independent of the choice of this
surface.
Turning now to statistical mechanics, we see two main routes
for developing a molecular expression of surface tension and re-
lated properties. One starts from the statistical formulation of
the pressure tensor and was fully formulated by Kirkwood and
Buff (6). The other route proceeds from the partition (or grand
partition) function Z of the system, going to the free energy F and
finally to y by means of the two relationships
F =- kT In Z (7 )
y @
=aiF) T,V,{N.} (8)
l.
In particular, for a fluid of spherical molecules, the final
expression of y is
Y = (112) ff• du X2 - Z2
dR dR R
...
n2 ( z , R) dz (9)
where u{R) is the intermolecular pair potential and n2{z,R) is the

pair density within the transition layer, normally oriented to the z
axis; Ris the vector joining two molecules (with components Y,Y,Z).
It is only rather recently that this expression has proven really
useful, when reliable information became available on n2 {z,lt).
Generalizations of eq. {9} for polyatomic molecules have also
recently been established {7}.
STATISTICAL MODELS OF INTERFACES

Eq. {9} and its extensions to nonspherical molecules are of
limited use because of our rather poor information concerning the
pair density n2 • Furthermore one is often interested in complex
systems for whl.ch equations like (9) would not only be difficult to
establish but also completely useless. Hence the deep interest in
developing more or less sophisticated models specialized to dif-
ferent kinds of interfaces.
In this respect the Cell Model of the liquid state has been
applied rather successfully to describe the liquid-gas interface
far from the critical point. Let us recall here that each molecule
of the liquid is supposed to be locked in a cage; given w the
potential energy at the cage center and 'iJ the partition function of
a molecule in its cage, one gets
(10)
where indices L, s respectively refer to "liquid" cells and "sur-

face" cells, a is the area of a surface cell at the interface and
the gas phase is treated as a vacuum (8). The results obtained are
reasonably good for y itself but less satisfactory for the surface
energy and entropy. A considerable improvement is obtained by
allowing a lacunar structure of the surface layer; i.e., by intro-
ducing a certain number of empty cells.
Liquid mixtures of different molecular species obviously con-

stitute a more complicated problem as care must now be taken of the
preferential adsorption of some of the components at the interface.
The so-called model of refiul'!r solutions is already a reasonable
approach for binary mixtures of spherical molecules of similar
size. Given a lattice with each site occupied by either an A or a B
molecule and given three energies e: AA' e: AB e: BB for first neighbor-
ing pairs, one gets in the simplest verSion of the model, i. e. ,
monolayer approximation and random mixing in the bulk and at the
surface respectively (9):
s 2
kT xA w L2
y = YA +a- In - L +- [( 1-2m)x: - (l-m) xB ]
xA a
( 11)
S
kT xB w 2 L2
= YB +a- In - L +- [( 1-2m)x~ - (l-m) xA ]
xB a
Here xA' xB are the mole fractions (x A + xB = 1), w = z(e: AB - e: AA /2

- e: BB /2), z is the number of first neighbors of a molecule in the
liquId and z (l-m) is the number of first neighbors of a molecule at
the interface. In order to get Y, eqs. (11) need to be solved
numerically for x S (or x:). Only for w = 0 (i.e., for the so-called
perfect solutions~ does one obtain an explicit expression for y in
L L
terms of xA' xB
- y a/kT
e = xLAe - Y A a/kT +xBe
L - YBa/kT (12)
Note that the monolayer model has been shown to present a slight
thermodynamic inconsistency to which one can easily remedy by
considering a multilayer approach (10).
There are two natural extensions of the model of regular

solutions: the first one is to substitute to the lattice a more
realistic picture of the liquid, the second one is to generalize the

model to molecules occupying more than one lattice site (the so-
called r-mer solutions). The first problem has been worked out by
means of a combination of the cell model and the theorem of
corresponding states (the so-called average potential model or
corresponding states theory (11». The second problem has been
extensively studied and will be discussed at length in the third
part of this lecture.
The surface tension of electrolyte solutions was initially

tackled by Onsager and Samaras (12). In the dilute case image
charges arising at the junction of two media of different dielectric
constants play the fundamental role. As the concentration in-
creases many other factors come into play: the finite size of the
ions, their polarizability, the change of the dielectric constant
of the solvent itself at the interface and also saturation effects.
Another interesting type of liquids and liquid mixtures for
which no serious statistical theory of surface properties seems to
exist, are associated substances where hydrogen bonds may extend
from molecule to molecule.
ADSORPTION OF CHAIN MOLECULES
Even when presented in a very schematic way this problem is an

extremely difficult one: it involves both energetic and entropic
effects at the level of a single chain molecule and, if the
concentration is not small, which is frequently the case, the
situation becomes extremely complex.
The simplest model consists in confining the chain molecules
on a semi-infinite lattice, each site being occupied either by a
solvent molecule or by a chain segment. Even so there are already
three kinds of effects to be taken care of, at infinite dilution:
(a) An entropic effect related to the decrease of the avail-
able number of internal configurations of the chain near
the surface;
(b) A first energetic effect linked to the energy difference
resulting from the replacement of a solvent molecule by a
chain segment in the surface layer;
(c) A second energetic effect related to differences in the
pair interactions: solvent-solvent, chain segment-chain
segment and solvent-chain segment. (This effect vanishes
for athermal solutions.)
For finite concentrations, excluded volume effects and energetic
effects between different chain molecules come into play. The
difficulty of the problem is easy to figure out when realizing that
even in the bulk phase the problem of the average conformation of a
single chain can only be solved in an approximate manner, when

retaining the excluded volume effect between distant segments.
Broadly speaking, there seems to exist four kinds of approach

to the adsorption of chains at an interface:
(a) analytical approaches,
(b) mean field theories,
(c) computer simulations
(d) exact enumerations.
Analytical theories are very difficult to develop for long

chains. Even the very special case of dimers is not solved. Edges
effects for a two-dimensional assembly of pure dimers were exactly
treated (13,14) and accurate values were also obtained for pure
trimers (15). Graph theoretical approaches in two and three dimen-
sions have also been developed recently by one of us (A.B.). They
permit in principle to compute the surface tension of monomer-dimer
mixtures at all concentrations by means of an exact series expan-
sion. The number of terms required may, however, become prohibitive
in some cases (16).
Mean field theories of adsorbed chains have been developed by

very many people in different ways. One of us (I.P.) used this kind
of approach for studying short chains (17) (dimers and trimers). All
these works may in some sense be regarded as extensions of the
Flory-Huggins theory of bulk phases, to interfaces; the system is
usually sliced into monolayers and the Flory-Huggins combinatorial
formula is used in each of them. The calculations are very
complicated for long chains. The quantities of interest are usually
the mean number of adsorbed segments of the chain vs. the tempera-
ture, the mean length of "trains" of adsorbed segments and the mean
length of "bridges" between two such successive "trains". A recent
account of the situation has been given by Silberberg (18).
Computer simulations have been widely used in the past for
studying properties of chain molecules, either isolated or in the
bulk phase. During the last few years such simulations have been
extended to adsorbed chains. The method used is based on extensions
of the usual Monte Carlo method as it is applied in statistical
mechanics and chains counting 100 segments or more were con-
sidered (19). Both the influence of the temperature and of the
presence of other chains were studied in this manner (20).
Let us finally turn to exact enumeration techniques, which

were very successful to analyze the conformation and thermodynamic
properties of chains in homogeneous phases and are presently ap-
plied to adsorbed chains (21,22). Because of the direct experience
of one of us (A.B.) in this kind approach, we shall develop it here
in some detail, in the particular case of dilute athermal solutions.
Consider a regular lattice of V sites, A of which are (equiva-

lent) surface sites. Each site is occupied by a solvent molecule
(monomer) or by a segment of a chain molecule (n-mer ) . Let us
introduce the following quantities:
Vc : the total number of configurations allowed to a single

n
n-mer when boundary effec1~jlre neglected (i.e. assuming
periodic boundaries and V » n).
the number of configurations lost for a single n-mer on
account of the boundaries (these configurations can be
figured out as lost when a periodic system is "cut" by a
surface) .
Ag n (v): the number of configurations allowed to a single n-mer

when V of its segments occupy superficial sites.
The grand partition function of the system is then given by
=1 + Z {Vc
n
+ A [-1 n 12 + E
V
where z is the activity of the n-mers and
f = exp(c:/kT), (14 )
c: being the adsorption energy per n-mer segment. It then follows

straightforwardly that the modification of the surface tension with
respect to that of the pure solvent is
g (v) (fv -
n
1)] /c n (15)
where C is the n-mer concentration and a is the area of a super-

ficial site. The adsorption is similarly given by
(16)
and the average number of adsorbed segments of a n-mer touching the

surface is
<v> = (17)
Eg (v) fV
n
One may also define such quantities as the mean thickness of the
adsorbed n-mer, perpendicularly to the surface, and its mean exten-
sion, parallel to the surface.
Exact enumerations of c ,1 , g (V) and similar quantities have
n n n
been made or are presently studied (21,22). Obviously one cannot
hope to obtain them for very long chains, a reasonable limit being
n = 15 or 16 for a simple cubic lattice. This may appear far too
short to match the actual problem but one should realize that one
deals here with exact numbers and that powerful extrapolation
methods exist wherefrom reliable information can be drawn about the
asymptotic limit n + 00.
FINAL REMARKS
In spite of an enormous theoretical effort made along differ-
ent lines of approach and which has, without any doubt, led to a
general understanding of the problem, much remains to be done. The
lattice model is indeed very crude, but just dropping the lattice
and replacing it by a continuum would not do much good as this would
negate the solvent structure completely. Future efforts should
take care both of the specific nature of the chains and also of the
molecular structure of the interface in relation to solvent proper-
ties.
REFERENCES
1• T. Young, Phil. Trans. ROY; Soc. (London) 95, 65 (1805).
2. P.S. Laplace, "Mecanique celeste", suppl. 10th Vol. (1806).
3. G. Bakker, "Theorie de la couche capillaire plane des corps
purs", Scientia (Paris) 11, 14 (1911).
4. J.W. Gibbs, Collected works, 2 Vols., New York (1928).
5. R. De fay , 1. Prigogine, A. Bellemans and D.H. Everett, "Sur-
face Tension and Adsorption", Longmans Green and Co.
Ltd., London (1966).
6. J.G. Kirkwood and F.P. Buff, ~ Chem. Phys. 11, 338 (1949).
1. H.T. Davis, ~ Chem. Phys. 61, 3636 (1911).
8. I. Prigogine and L. Saraga, ~ Chim. Phys. 49, 399 (1952).
9. E.A. Guggenheim, Trans. Faraday ~ ii, 150 (1945).
10. R. Defay and I. Prigogine, Trans. Faraday Soc. 46, 199 (1950).
11 . A. Englert-Chowols and 1. Prigogine, ~ Chim. Phys. 55, 16
( 1958).
12. L. Onsager and N.N.T. Samaras, l..:. Chem. Phys. ,g, 528 (1934).
13. M.E. Fisher, Phys. Rev. 124, 1664 (1961).
14. P.W. Kasteleyn, Physica 21, 1209 (1961);.!!..: Math. Phys. !, 281
(1963). --
15. J. Van Craen, J. Chem. Phys. 63, 2591 (1915).
16. A. Bellemans and S. Fuk~sICa 50, 348 (1910); A. Bellemans,
Physica 65, 89 (1913); A. Bellemans, J. ColI. Interf.
Sci. 58, 521 (1911). - --
11. 1. Prigogine,- l..:. Chim. Phys. 41, 3 (1950); 1. Prigogine and L.
Saroltla, .!!..: Chim. Phys. 41, 801 (1950).
18. A. Silberberg, Faraday Discussion, Chem. Soc. 59, 203 (1915).
19. M. Lal and R.F.T. Steppo, ~ Pol. Sci.: Polymer Symposium £2,
401 (1971).
20. A. T. Clark and M. Lal,.1.:.. Chem. Soc. , Faraday Trans. II, 74,
1857 (1978).
21. A. Bellemans, ~ Pol. Soc.: Part C 39, 305 (1972).
22. M.N. Barber, A.J. Guttmann, K.M. Middlemiss, G.M. Torrie and
S.G. Whittington, ~ Phys. A 11, 1833 (1978).
Surface and Interfacial Tensions of
Polymer Melts and Solutions
Toshio Hata
Gunma University
3-39-22, Showa-cho, Maebashi-shi, Gunma Prefecture,
Japan, 371
and
Tomoyuki Kasemura
Department of Textile Industry, Faculty of Engineering
Gifu University, Kagamihari-shi, Gifu Prefecture,
Japan, 504
ABSTRACT
Work done in the last ten years at our laboratories is reviewed
on the following topics: (1) Methods (sessile bubble and drop
methods) for the determination of surface and interfacial tensions;
(2) Molecular dependence of surface tension (y); (3) Molecular
structure and y; (4) Composition dependence ofy in copolymers; (5)
Interfacial tension (Y12)' s between polymer melts; (6) y's of
polymer solutions and their Y12' s against water; and (7) y's and
Y12 's of polypeptide solutions. Molecular weight dependence of yis
explained on the basis of Sugden's equation and the effects of
molecular structure such as methylene sequence in the main chain and
the alkyl side chain are pointed out. The sequence effect plays an
important role especially in copolymers. Different behaviors of
copolymers with different combinations of monomeric units were
introduced. The importance of interfacial tension measurements is
emphasized for polymer melts and solutions, including polypeptide
15
16 TOSHIO HATA AND TOMOYUKI KASEMURA
solutions, to ch~racterize the surface properties of polymeric

materials.
INTRODUCTION
Surface and interfacial tensions arising from intermolecular

interaction of liquids are also directly related to adsorption,
adhesion, wetting, dyeing, printing and other surface processes,
which have become increasingly important for polymer systems.
Nevertheless, enough systematic studies have not been done to
enable one to generalize or characterize the polymer behaviors in
surface and interfacial tensions. In the present review, we want to
deduce possibly generalized laws or characteristics for polymeric
systems from our experimental work, which has been done in the last
ten years with our co-workers at the Tokyo Institute of Technology,
Gunma, and Gifu University. For other researchers' results, the
reviews by Wu (1,2) should be referred to.
1• METHOD (3)
Surface tension has been measured by the sessile bubble method

and interfacial tension by the sessile drop method. Both are
thought to be the most adequate for highly viscous substances such
as polymer melts and solutions. Figure 1 is the schematic represen-
tation of the sessile bubble method. This glass cell is placed in a
container filled with deoxidized and dehydrized nitrogen gas. The
supporting board can be raised by a lever as high as the horizontal
metal plate dips into the liquid sample. A bubble is formed by
passing nitrogen gas through a hole at the center of the plate. The
entire apparatus is placed in an air bath adjusted to a proper
temperature.
The bubble'S maximum radius (r) and the distance (h) from the
meniscus top to the r plane are measured by a cathetometer. Surface
tension (y) is calculated by Porter's equation (4)
(1 )
where ~p is the difference in density of the sample in gaseou8 and

liquid states, and g is the gravitational constant. Density is
measured by a proper method according to the sample's state at a
normal temperature and with dilatometry at higher temperatures.
In preliminary experiments, we examined the precision of the
approximate equations proposed by Porter, Wheeler and Staicopolas,
using some standard liquids. As a result we found that the Porter's
equation was most reliable in the range of O.6>h/r>O.4. The error
was within 2%.
SURFACE AND INTERFACIAL TENSIONS 17
Fig. 1. Scheme of sessile

bubble method.
Interfacial tension can be measured with the same method only

by replacing a bubble by a drop of a second liquid. The drop of
liquid with density higher than the matrix is placed on a teflon
plate at the bottom of the cell.
Here I wish to call attention to the fact that the measurement
of interfacial tension is extremely limited for liquid of low
molecular weight because of their compatibility or partial misci-
bility, while, in contrast, polymers are generally incompatible, so
we can measure it for almost all combinations of polymer homo logs
which are miscible in the low molecular wieght region. This fact is
of great importance for scientific research in surface chemistry.
2. MOLECULAR WEIGHT DEPENDENCE OF SURFACE TENSION
2.1 A Formulation and the Experimental Examination
Surface tension of polymer homologs varies with molecular

weight generally in the oligomer region. Whether surface tension
increases or decreases with molecular weight depends on the nature
of repeating unit and end groups, e.g. in n-alkanes it increases,
while in alkyl naphthalene and alkyl diphenyls decreases with alkyl
chain length as shown by Bascom et al. (5). Examples by us are
shown in Fig. 2 for poly (a-methyl styrene) (3) and in Fig. 3 for
ethylene and propylene glycol condensates with and without methyl
end group (6). (See also Fig. 7 for alkylene glycols.)
We have explained this dependence of surface tension on

molecular weight and composite units by using Sugden's equation
(8) •
E
~
c30
>.
~
Z
o
:nz 28
w Fig. 2. Surface tension vs.
t- degree of polymeriza-
tion (n) for poly
(a-methyl styrene).
10 20 30 40 144
n
45
PEG
-t/' Zi..... /C.>--- -
Fig. 3. Surface tension vs.

molecular weight at
25 0 C.
z
o EG, PEG: Ethylene glycol homologs
:nz 30 PG, PPG: Propylene glycol homo-
W
t- logs
MM: Monomethyl, OM: Dimethyl
(6: by A. K. Rastogi and L.E.
St. Pierre (7»
o 200 400 600 800 1000

MOLECULAR WEIGHT
_ j(PL"G)P l 13 _ (PLP)B _(f)B (2)

y- ~ M ~ - M - V
where PL' PG are density at liquid and gaseous states respectively;

M is molecular weight, and 13 is Macleod's exponent which was 4 in
Sugden's work. As Parachor P and molecular volume V can be assumed
to obey the additivity law, Eq . (2) is expressed as follows by
dividing P and V into two parts, repeating unit (suffix r) and end
groups (suffix e).
(3)
where m is the number of repeating unit and at present a homologous

series with two same end groups is considered. Equation (3) is
transformed as follows and expanded, then an approximate eq. (4) is
obtained
p )13 (1+-~
2 P -p )13 V -V )...£1
~
y= ( -1: + ..L ...L....!:. tJ
V m+2 P m+2 V
r r r
• Yr [1 ;!I(~:r- 11]
+
(4 )
where y =(P IV )
r rr
a,
y =(P IV )
e ee
a,
k= V IV , and m+2 can be pu t n
er
(degree of polymerization) in many cases. Equation (4) shows that,
if a would be invariable, y increases or decreases linearly with
1 I (m+2) or lin (11M) according to y > y or y <y respectively, and y
r ere
keeps constant when y =y. A similar treatment has been applied to
r e
glass transition temperature (T ) by Ueberreiter and Kanig (9),
obtaining the linear relationshi~of liT and 11M, or approximately
T and 11M. In the same way for speCific volume (v), we have the
r~lation,
2M
v = vr + ~ (v
Mer
- v ) (5)
where vr =Vr1M,
r ve =Ve1M
e and M, r Me are molecular weights of
repeating unit and end group respectively. Figure 4 shows that
these relations (r, T , v~l/n) hold for poly (a-methyl styrene), and
Fig. 5 does for r & polyethers in Fig. 3. Curves in Fig. 5
converge to definite values (y~ of respective polyethers at
infinite M, passing through different lines depending on the
presence or absence of methyl end group. Strictly speaking,
however, a
in Eq. (4) varies with M roughly from 5 to 3 in the range
of low molecular weight. a in the exponent of (relY r> acts as
compensating the change of a in the numerator of the coefficient.
This might be a reason why linearity is established in many cases.
The effect of a,
however, must be taken into consideration
especially in low molecular weight region, together with the effect
tha t the higher order terms are omit ted in Eq. (4) . An exact
relation of y to n (M) is given in the following discussion.
LeGrand and Gaines (10) have proposed the relation

A
Y = YCD - 2/3 (6)
M
where YCD is the surface tension at infinite molecular weight.
Though the authors' discussion on the coefficient A is question-
able, (e.g. according to their discussion, A must be a positive
constant, then Eq. (6) can't explain the decrease of Y with
molecular weight), Eq. (6) seems to represent very well the observed
values of each homologous series. It must be noticed, however, that
the plots of y vs. M-2/3 for poly(ethylene glycol)s and poly-
(propylene glycol)s with and without methyl end group do not
converge to the respective values ofyCD at infinite M, where the end
group effect should disappear.
Fig. 4. Specific volume (v),

160;::; glass transition tempera-
140...... ture (Tg) and surface
.....
IZO
tension (y) vs. lin.
100
I
o 10 15 20
lin l< 10'
E45
~
c
~ ____-.--.I--- PEG
>.
.B40
Z
o Fig. 5. Surface tension vs.
Vi35 11M at 25 0 C for polyethers
z in Fig. 3.
UJ
.....
30
UJ
u
<t
~25
:::>
Vl
•
o 5
11M
According to Wu the molecular weight dependence of surface

tension can be expressed well by the relation
-1/4 -1/4 B
Y = Yoo +-
M
n
or
114 1/4 (8) (2)
Y = Yoo
where M is the number-average molecular weight. If the co-

n
efficients Band C are positive, as pointed out by the author, Eqs.
(1) and (8) fail again to represent the decrease ofy with molecular
weight. Equations corresponding to Eqs. (1) and (8) can be easily
derived from Eq. (3) as
(9)
-1/8 ( 10)
y
where Yr ' Ye and k are the same as those in Eq. (4) and k'=Pe/P r •
(Hata (8) had expressed Eq. (9) as Y =y r {1+2/(m+2Ve/vJ.(P e /P r-
V IV )}4, putting 8=4.) As y is equal to Yoo, i f we can put 8=4,
e r r
k=1, k'=1 and m+2=n, Eqs. (9) and (10) reduce to Eqs. (8) and (1)
respectively. The derivation of Eqs. (9) and (10) doesn't involve
any operation such as expansion and omission of higher order termsj
therefore they are more accurate than Eq. (4), at least mathe-
matically. In fact too, the results shown in Figs. 2 and 3 give the
respective straight lines by the plots of y1/8 vs. 1/(m+2k) if we
take the experimental mean values for e.
The line holds also for
smaller m, which the approximate Eq. (4) failed to reproduce.
So far we have considered molecules of X-R n -x or X-R n -X' type.

For molecules of X-R -Y-R -x type, e.g., dialkyl phthalatesj how-
n n
ever, it was found that the additivity law above mentioned doesn't
hold. The results and discussions for them will be presented in the
next chapter. (3.4)
2.2 Estimation of Yr of Repeating Units and Ye of End Groups
On the basis of Eq. (4), we can estimate y and y from

experimental results on the molecular weight dependfnce of s~rface
tension. 8 is obtained by log y - log P plots, and k(=V IV ) by the
molecular weight dependence of specific volume using Eq~ (5). The
analysis of poly(ethylene glycol)s, poly(propylene glycol)s,
poly(isobutylene glycol)s, and their methyl ethers, for instance,

gives the values of y. and y. as listed in Table 1.
r e
The fact that y of poly(ethylene glycol) is almost independent
of the molecular weight arises from Yr(-CH 2-CH2-O-) = Ye(-OH).
Table 1. Surface l'enswn of' Repeahng Um.ts ( y. ) and
r
End Groups (y ) for Some Polyethers at 25 0 C
e
Repeating units Yr(dyn/cm) End groups Ye(dyn/cm)
-CH 2 -CH 2-O- 42.1 -OH 42- 44

-CH(CH )-CH -0- 31.0 - CH 3 11 - 13
3 2
-C(CH 3 )2- CH 2- 0- 21.0
3. MOLECULAR STRUCTURE AND SURFACE TENSION
3.1 Effect of Tacticity
Temperature dependence of surface tension of isotactic and

atactic polypropylene (11) is shown in Fig. 6. So far as this
result indicates, the effect of tacticity on the surface tension is
negligibly small. The flexibility of chain, however, is known to
influence various thermodynamic properties of polymer. Therefore,
it may be too early to conclude that the tacticity has no effect on
surface tension. The necessity to consider the chain flexibility
will be discussed again in the following section.
3.2 Methyl Group Effect

In general, surface tension decreases by the substitution of a
hydrogen for methyl radical, as shown in Figs. 3 and 5 for
poly(ethylene, propylene and isobutylene glycol)s. Further exam-
ples are polypropylene (y = 22.6 dynl cm) vs. polyethylene ( y = 21.3
dyn/cm for both at 150 oC). This effect of methyl group may arise
from the increase of free volume. For (1-hydrogen, however, the
methyl substitution doesn't always reduce y, e.g.
y(PE»y(PIB»y(PP) and y(PSt)~y (PXMSt). In the case of poly(iso-
butylene) it may be said that the flexibility of molecules (intra-
molecular motion), which is increased by the introduction of (1-
methyl, acts as hindering the surface orientation of low energy
atomic groups, thereby increasing y. In poly(isobutylene glycol),
this effect may be relatively small, because of the neighboring
E
~40
c l>. : al-PP
>-
-0 c : iso-PP
;30
o....
III
Z 20
w
I-
~1 0
~
a: OL-______-L______
~ ______~L-____ ~ ~
III 100 140 180 240

TEMERATURE (OC)
Fig. 6. Temperature dependence of surface tension

for isotactic and atactic polypropylenes.
oxygen bond which is able to rotate as well freely as before the

introduction of ex -methyl.
Methyl group on the chain end lowers more effectively than the
one on the side chain, if they are compared at the same methyl
content (see original reference) (6).
3.3 Effect of Methylene Sequence in Alkylene-glycols and Their

Condensates (12)
Here I discuss the effect of methylene sequence on y in

alkylene-glycols, HO- (CH 2 ) m-OH (AG) and their condensates, poly-
(alkylene glycol)s HO-[(CH 2 ) m-0]n -H (PAG). There has been given
Eq. (4) or Eq. (9) applicable for y vs. m of AG . For y vs. m (not n)
of PAG, Eq. (11) can be derived by the same way, putting 0+00 •
y = YCH r1 + ruL {(~) 1/8 - 1}] ( 11)

2 l m+1 YCH 2
where YCH and Y 0 are surface tension components of CH2 and 0

respecti ve~y, and k is ratio of their molecular volumes Vo/V CH .
The results are shown in Fig. 7, and plots of yvs. 1/(m+2), for ~G
and 1/(m+1) for PAG are shown in Figs. 8 and 9, respectively.
Equation (4) holds for AG, while for PAG, experimental curves go
through below the straight line expected from Eq. (11). (The exact
equation corresponding to Eq. (9) gives the same result.) The reason
~4Sr--------------------------~
E
u
C
>.
~
'0
;- 40
o
<f)
z:
W
I- 35
w PA G --------
u
<{
lI..
~ 30
<f)
o 2 4 6 8 m 10
Fig. 7. Surface tension vs. number of methylene

groups of alkylene glycols and poly
(alkylene glycol)s at 90 oC.
E '5
~
c
>.
z ~ '0
o
~ 40
z
o
IJJ
I- Vi 35
Z
W
w I-
u
~ 35
cr w JO
~ u
~
<f)
30 ~ 25
0.1 0.2 0.3 o 0.1 0.2 0.3 0.4 0 .5
Ym:2 I/(m.')
Fig. 8. Surface Tension vs. Fig. 9. Surface Tension vs.

1/(m+2) for alkylene 1/(m+1) for poly (alkylene
glycols. glycol)s.
may be attributed to the predominant adsorption of low energy

component, e.g. methylene sequence in this case. The difference of
the effect of methylene sequence between AG and PAG is probably due
to the cooperative adsorption of neighboring sequences in PAG, so
that the length of methylene sequence ficti tiously grows to the
longer one in the adsorbed states as seen in Fig. 9. The sequence
effect is observed more frequently in copolymers as shown in the
next section. (In Fig. 8 we can see that y of HO- (CH2 ) 5-0H (/l) lies
off the curves at 20 0 C and 50 oC. This is presumably cuased by the
ring &ormation of six members by hydrogen bonding, which disappears
at 90 C as shown in the graph.)
3.4 Side Chain Effect
It is well known that properties of vinyl polymers vary with

the length of alkyl side chain. For example, glass transition
temperature decreases rapidly at first and reaches the minimum, at
the chain length of Cl0 in the case of poly(methacrylate)s. As
surface tension is governed by cohesive energy density similarly to
T , it is expected the same behavior to be observed in surface
tlnsion. The dependence ofy on side chain length has been studied
for polyacrylates and poly(vinyl alkylate)s in molten state. In
polyacrylates y decreased monotonously from 31.0 dyn/cm for
poly(methyl acrJlate) to 21.2 dyn/cm for poly(2-ethyl-hexyl
acrylate) at 150 C (13). Samples of longer side chain have not been
examined. Here the results for poly (vinyl alkylate)s made up of n-
aliphatic acids (14) are shown in Fig. 10. Y reaches the minimum
value at poly (vinyl caprylate) of c8 side chain.
Similar results are obtained for dialkyl phthalates of various
alkyl chain lengths (15) as shown in Fig. 11. These materials are
or tho compounds, so that their branching structure rather likes
poly(vinyl alkylate)s.
Components of surface tension, y d and y p in terms of Fowkes
(16), are estimated by the measurements of interfacial tension
(y 12) against polyethylene at 150 0 C in the case of poly(vinyl
alkylate)s, and by the measurements of contact angle (6) on solid
paraffin in the case of dialkyl phthalat8s at 20 0 C, according to the
equation
Y 12 = Y1 + Y2 - 2~ ( 11 )
where y 1(= y~ + y~) is taken as the surface tension of the sample,

and Y2(= y~) is that of polyethylene or paraffin. As shown in Figs.
1S and 11 b t?e fi.rst decrease of y is caused by the de~rease of both
y and ~ 1n d1alkyl phthalates and by y Palone 1n poly(vinyl
30r--------------------,
40
"E 25
~
c r
>.
~
~20
z
o....
a
......
30
\
III
Z ,,
~10
w
W
~ ,,
~ UJ " n-Alkane
U
<
~ 5
::l
III
OL---~----~----~----~
o 5 10 15 20 5 9 13
NUMBER OF CARBON ATOMS NUMBER OF CARBON ATOM S
IN SIDE CHAIN 1 N ALKYl GROOP
Fig. 10. The effect of side Fig. 11. The effect of alkyl
chain length of poly(vinyl chain length of dialkyl
alkylate)s on surface tension phthalates on surfaced tension
and its components due to and its components, y and
dispersion force (yd) and f, at 20 0 C.
polar force (yp) at 90 0 C.
alkylate)s. Different behaviors of yd between the above two series

will be understood by comparing the phthalic and vinyl groups whose
dispersion interactions are far different.
After passing through the minimum, y increases fairly in

parallel with the increase of y d, which shows the predominant
interaction between alkyl side chains. In fact, the change in yd in
dialkyl phthalates is essentially the same as that in n-alkanes as
seen in the Figure. For the minimum y, it can be said that the so-
called HLB (hydrophile-lyophile balance, exactly in this case
"lyophile-lyophob balance") reaches an optimum condition at that
side chain length.
4. SURFACE TENSION OF COPOLYMERS
4.1. Sequence Effect in Random Copolymers
First we deal with random copolymers. Out of various combina-

tions of monomeric units, we have taken up copolymer samples
classified to each of the following three types of combination:

1) nonpolar/nonpolar: ethylene/propylene copolymers (E-P
series) ;
2) polar/polar: ethylene oxide/propylene oxide copolymers
(EO-PO series) and tetrahydrofuran/propylene oxide co-
polymers (THF-PO series); and
3) nonpolar/polar: ethylene/vinyl acetate copolymers (EVA
series) •
It is of interest to compare the probably different behaviors of y
for these three types of combination.
In Figs. 12 and 13, results are shown for E-P and THF-PO series
respectively (17). Here x2 is taken as the mole fraction of low
energy component, i.e. propylene in E-P series and propylene oxide
in THF-PO series. As seen in the figures, observed values deviate
from the straight line expected from the simple additivity proposed
by Rastogi and St. Pierre (18).
If the additivity of P and V in Sugden's equation can be

assumed again for monomeric units 1 and 2 in random copolymers, we
have
Y • Y, (:~ ) (1 + :::::)a (12)
where Y1=(P 1/v 1 )8, surface tension of homopolymer 1; k1 = P 1/(P 2 -

P1)' k2 = Vl(V 2 - V1) and x2 is the mole fraction of monomeric
unit 2. V1 , V2 in k2 are given by specific volume measurements and
P1 , P2 in k1 are calculated from the relation P = y1/8 v for each
polymer. Calculated curves are drawn with dotted broken lines (-
- -) in Figs. 12-14, using the mean values ofefor each copolymer
series. (8 = 3.0 for E-P series, and 3.5 for THF-PO and EO-PO
series.) As you can see, the observed values can't be represented
by Eq. (12), showing the limited applicability of Sugden's equation
or the simple additivity of Parachor to copolymers. Such a dis-
crepancy is probably due to the selective adsorption of low energy
component of large sequence length.
In random copolymers, there exists a distribution of sequence

length of each component, accordingly do the sequences of low energy
component, length of which is larger than the mean value corres-
ponding to x 2 . These sequences will preferentially be adsorbed to
surface and reduce y more than expected from the additivity law.
Mole fraction x2 larger than 0.4 will be necessary for the sequence
to be effective in its length as well as its quantity. The roles of
sequence length and its quantity are clearly demonstrated by block
copolymers described in the next section.
E J O r - - - - - - - -- ----. 40r-------------~
~ E
c
>. ~
;g c
§35
z
o.... z
VI
o
Vi
~ 25
I- ....~ 30
UJ w
U u
~ if.
a: ~ 25
::> VI
VI
20L--~-~ _ _ _ _L_-J
~
o 0.2 Q4 0.6 08 1.0 o 02 0.4 C6 10

Xl
Fig. 12. Composition dependence Fig. 13. Composition dependence
of surface tension of ethylen- of surface tension of tetra-
propylene copolymers at 150 o C. hydrofuran-propylene oxide
(- - - : calculated by copolymers.
Eq. (12».
4.2 Comparison Between Random and Block Copolymers
Both random and block copolymers of various compositions can

be synthesized from EO and PO monomers. We (19) compared random EO-
PO copolymers with block ones of the type (EO)~-(PO)m-(EO)~, where
the composition of PO was varied by ~, keeping m constant at 16, 30,
and 56. The studies for essentially the same materials had been
done by Rastogi and St. Pierre (18). Nevertheless we tried the
similar experiments by ourselves in order to examine the applica-
bility of the additivity law to random copolymers with additional
experimental values, and further to confirm the sequence effect by
the comparison with block copolymers. The results almost agree with
those of Rastogi et al. under the same conditions. An example in
our observations is shown in Fig. 14. The random copolymers (a in
the figure) behave just like E-P and THF -PO series, while in
contrast the block copolymers lower y drastically at a few mole
fractions of PO, and the more the sequence length m, the more the
decrease. Each plateau corresponds to each m. It's noteworthy that
the curve of random copolymers intersects and goes below the curve
of block copolymers of m= 16 at a mole fraction about 0.8. This
means that PO sequences larger than m=16 appears more frequently in
the random copolymers over this composition. This effect of block
EO-PO copolymers is quite similar to surface-active agents, there-
fore they are used, in practice, as surfactants. The introduction
of PO block (low energy component) of proper length is of greater
advantage for the reduction of y than their statistical occurrence
_40~--------------------~
E
u
...... \ ,
~ II\ '~
- la' '\.
" \
~351 \
..... I
I
VI ,\
,
Z
W
~
, \
o 0.2 0.4 0.6 0.8 1.0 100

MOLE FRACTION OF PO VAc CONTENT (wt %)
Fig. 14. Composition dependence Fig. 15. Composition depen-

of surface tension of ethylene dence of surface tension of
oxide-propglene oxide copoly- ethylene-vinyl acetate
mers at 60 C. copolymers .
a : random, b: block (m=16),
c : block (m=30),
d: block (m=56)
at the same composition .
4.3 Characteristics of Ethylene-Vinyl Acetate Copolymers

In preceding sections, it was pointed out that random copoly-
mers composed of nonpolar/nonpolar and polar/polar monomers have
rather common features in '( - x2 curves. Ethylene-vinyl acetate
copolymers as an example of nonpolar/polar pairs, however, behave
considerably differently as shown in Fig. 15, where,( goes through a
minimum at about 35 wt% of vinyl acetate. A minimum or a maximum
has also been observed in glass transition temperature for some
kinds of copolymers, and it has been attempted to modify the
additivity law to interpret the results. More theoretically,
however, behaviors of polymer surface tension must be founded with
liquid theories of polymers, and an explanation for the results of
EVA above described has been made by our co-worker, T. Nose (20), on
the basis of his hole theory of polymer liquids, applying it to
surface tension of mixtures. Qualitatively speaking, the minimum
appears as a result of the balance between energy and entropy
contribution with the introduction of vinyl acetate into poly-
ethylene.
5. INTERFACIAL TENSION BETWEEN POLYMER MELTS
Since Good and Girifalco, and Fowkes originally proposed the

formulae on interfacial tension (Y12)' several modified equations
have been reported in the last ten years. The validity of them,
however, has not been fully examined because of the lack of enough
experimental data especially on Y12 . Here I present some results of
Y12 for rather systematically selected combinations: the one is
polyacrylates (PMA, PEA, PnBA, and P2EHA) to each other and to
poly(ethylene) (PE) and poly(styrene) (PS) j the other is EVA of
various vinyl acetate contents to PE, poly(vinyl acetate) (PVA )
and one of EVA (VAc 25 wt%). c
In Figs. 16 and 17, Y12 of the former series are shown as a

function of alkyl chain length at 1500 C (21). For pairs of each
polyacrylate, the smaller the difference of alkyl chain length is,
the smaller Y12 is obtained as expected, and for pairs to PE, the
larger the chain length, the smaller Y12 . For pairs to PS, Y12 is
minimized about at poly(propyl acrylate), showing the best affinity
of the pair. Results for EVA series (22) are shown in Fig. 18
against vinyl acetate contents, from which a similar conclusion as
above mentioned can be drawn.
Using these data as well as y of each polymer, we examined the
proposed equations on Y12' Here the following three equations are
checked by the data of polyacrylates series.
-10~-----------------------'
E
~
c:
>-
:g
5.....
III
Z
UJ 5
.....
\E", PMA
---
.....0
.-0
Fig. 16. Interfacial ten-
sion of polyacrylates to
each other as a function
of Side chain length at
-'
oct
.....
u
{f
a:
\0 ,...L>
.....0
150 C.
UJ c
.....
....Z 0
7 9 PE
3 5
NUMBER OF CARBON ATOMS IN
ALKYL SIDE CHAIN
E1 O~-------------------------,
~
c
>-
~
z
:5. ~
8 0
Fig. 17 . Interfacial
tension of polyacrylates
against polyethylene and
~ \ PS o~~.--. polystyrene at 150 oC.
a::
UJ
~v~---.
I-
~ 0 '--.L...-....L..-~-'----'---I.--''--..L...--'-..... ,.,,'''''-.L.-'
J 5 7 9 PE
NUMBER OF CARBON ATOMS IN
ALK'Wl. SIDE CHAIN
E
~
c
>-
.:!;I
5 10
in
z
W
I-
..J
~
Fig. 18. Interfacial tension
..... of ethylene-vinyl acetate
u 5 5
copolymers against poly-
~ ethylene, poly(vinyl acetate)
a::
w and one of the copolymers
t-
Z (VAc 25 wt%) at 150oC.
50 100
VAc CONTENT (..... t%l
D.K. Owens and R.C. Wendt (23), D.H. Kaelble and K.C. Uy
(24) and R.K.S. Chan (25)
Y12 = Yl + Y2 - 2~ - 2~~ ( 13)

T. Hata and Y. Kitazaki (26)
S. Wu (27)
(15)
y ~ and y~ in Eqs. ( 13) and ( 15) were determined with the

observed value of Y1i against PE and y 1 of each polymer, putting
Y2::Y~ for PEe y~, Y1 and Y~ in Eq. (14) were determined with Y12
values against PE and PS, assuming ~=Y~+Y~ andy ~=o for the latter.
The results are summarized in Tables 2 and 3. As you can see in
Table 3, the calculated values are much smaller than the observed
ones, except for some values calculated with Wu's equation. The
terms of work of adhesion seem to be overestimated by these equa-
tions.
Table 2. Surface Tension Components of Polyacrylates

(at 150oC)
By Equations ( 13) and (14) By Equation (15) .

d
Polymers Y l yp
a b f/~
(x1 0)
Y Y
P PI
(i10b)
Y Y
PMA 31.0 24.5 6.5 3.4 3.1 21.0 24.5 6.5 21.0
PEA 26.4 21.0 5.4 4.4 1.0 20.0 i 21.0 5.4 20.5
PnBA 22.8 19.7 3.1 2.6 0.5 13.6 119.9 2.9 12.7
P2EHA 21.1 19.4 1.7 1.1 0.6 8.1 19.5 1.6 7.6
W-PE 23.2 23.2 0 0 0 0 23.2 0 0
PS 31.5 26.9 4.6 4.6 0 14.6 27 .1 4.4 14.0

Table 3. Comparison Between Calculated and Observed

Interfacial Tensions (at 150 0C)
Y12 cal. (dynl cm) y 12 0bs.

(dyn/cm)
Polymer Pairs
Eq. (13) Eq. (14 ) Eq. (15)
PS-PMA 0.2 0.5 3.2
PS-PEA 0.4 0.9 1.4
PS-PnBA 0.7 1.4 1.4
PS-P2EHA 1.3 2.5 2.5
P2EHA-PMA 1.8 1.9 3.5 5.7

P2EHA-PEA 1. 1 1.2 2.1 3.3
P2EHA-PnBA 0.2 0.3 0.4 1.2
PnBA-PMA 0.9 1.4 1.9 3.0

PnBA-PEA 0.3 0.3 0.8 3.0
PEA-PMA
I 0.2 0.8 0.4 1.3
6. SURFACE TENSION OF POLYMER SOLUTIONS AND THE INTERFACIAL

TENSION AGAINST WATER
Now I want to introduce some examples to show that y' s of

polymer solutions do not always realize the characteristics of
solute polymers, but Y12' s of the solutions against water can show
it clearly.
Figure 19 is surface tensions of toluene solutions of EVA

(Y sol) and their interfacial tensions to water (y sollH 0) plotted
against VAc contents in EVA (28). y l's do not vary ov~ the whole
so
range of EVA composition while Ysol/H 0 changes drastically with

the composition, passing through a min~um and a maximum. Toluene
is more surface-active than EVA of any composition in this solution,
accordingly it remains at surface keeping Yso 1 constant at the value
of toluene. On the other hand, at the interface. between the toluene
solution and water, EVA molecules are adsorbed more and more with
the increase of hydrophilic vinyl acetate component. The minimum
and the maximum are caused by the competition between lyophilicity
and the increasing compatibility of EVA with toluene in this region.
In fact, at the maximumYSOl/H20 both EVA(87.0 wt% VAc) and toluene
have an equal solubility parameter of 8.9.
Figure 20 is the case of toluene solutions of various poly-

acrylates (29). In this case, tOQ, Ysol is unchanged in spite of
the difference of alkyl side chain, while Y l/H 0 varies according
to the hydrophilic nature of side chain. so 2
y. l/H O's of toluene solutions of a styrene-maleic anhydride

so 2
(3/1) copolymer were measured against water of various pH's (30).
(Fig. 21) The rapid decrease of Ysol / H 0 near at pH 11 suggests the
dissociation of maleic acid at this pH?
E
2 35~l
.;; 30 .#0.
'? 25-
-<
(5 20
o ~
"20
~ (; 15
"tI
(5
~15 .,,: solvent 9 7 5 3 1
CARBON A TOM S
N lJ>1BER OF
IN SIDE CHAIN
o : )"sol at 25°C,.:(sOIl"30 at 25°C,
e : a t 45°C. 0: elt 65°C. e : tT/H.O
VAc CONTENT (wt%1 at 25°C. e:oI45·C. e : oI65°C.
Fig. 19. Surface tension of Fig. 20. Surface tension of

ethylene-vinylacetate copoly- polyacrylates/toluene solutions
(y 1) and their interfacial
mers/toluene solutions (Y sol ) so
and their interfacial tension tension against water (Y sol / H 0)
against water (Ysol/H 0) as a as a function of side chain 2
length of the solutes at
function of vinyl ace€ateocon- C=0.01g/100ml.
tent of the solutes at 20 C,
C=0.01g/100ml.
Fig. 21. Interfacial tension

between styrene-maleic
anhydride copolymer/toluene
solutions and water of dif-
ferent pH at 20 C,
C=0 . 05g/100ml .
O~~--~~--~~~~
2 4 6 8 10 12 14
PH
As demonstrated so far, interfacial tensions of polymer solu-

tions against water or other incompatible liquids provide important
information on the characteristics of solute polymers and their
adsorption behavior . It will be applicable as a tool to predict the
polymer adsorption on a hydrophilic solid surface.
7. SURFACE AND INTERFACIAL TENSIONS OF POLYPEPTIDE SOLUTIONS
Biological processes seem to take place very often at surface

or interface, where the selective adsorption may play an important
role in specifying the process . So it's useful and interesting for
studies on biological phenomena to make clear the surface-chemical
behavior of biomaterials . Now we want to introduce two topics in
our studies on polypeptide solutions: one is on the surface tension
in relation to the conformational change of polypeptide and another
is on different behaviors between surface and interfacial tensions
of polypeptide solutions .
7.1 A Change of Surface Tension Accompanied with the Conforma-

tional Transition of PolY-y -Benzyl-L-Glutamate in Ethylene
Dichloride Solutions (31)
We found, for solutions of poly-y-benzyl-L-glutamate (PBLG) in

ethylene dichloride (EDC), that the conformational transition
occurred with the change of concentration. The measurements of
optical rotatory disperions (ORO) and circular dichroism (CD) have
shown that PBLG takes the a-helical conformation predominantly in
concentrated solutions and the a-structure in diluted solutions.
In the concentration range of their transition, the drastic
decrease of surface tension (y 1) was observed as shown in Fig . 22,
so
where helical contents calculated by the Moffitt's equation, using
ORO data, are plotted together. y 1 of further diluted solutions
so
1//
-o~ 0 0- ~
0
0
30 100 Z~
E
u
IJJ
~
Fig. 22. Surface tension of
,. J
"- z poly-y-benzyl-D-glutamate in
c 0
>.
"0
U ethylene dichloride and the
25 50 :;i content of helical conforma-
0
....u tion at 25 0 C plotted against
....J
log C.
~ IJJ
:x:
/
,.
20 0
-1 0 1
Log C (c:g/dl)
(not included in the Figure) scarcely varies keeping the value of

solvent itself.
It is concluded from these facts that PBLG molecules in the S-

form and more or less in associated states may be more compatible
with EDC, accordingly tend to exist in the bulk phase, while in the
helical form they will become more lyophobic, because of the benzyl
side chains oriented towards outside the rod like molecules. At
20-30% of helical content corresponding to the minimum y ,the
lyophobic and lyophilic parts may reach the optimum balances~ike as
a surface active agent. The following increase of y so 1 with
increasing concentration may be attributed to the formation of
liquid crystal (32,33), where helical molecules attract each other
so as to stay in the bulk.
The changes of surface tension accompanied with the conforma-
tional transition have been reported by Neumann, Moscarello and
Epand (34) for aqueous solutions of poly-L-lysine, and by Morimoto
and Ueberreiter (35) for dimethylformide and dioxane solutions of
PBLG.
7.2 Surface Tension of Pol -DL-Alanine in A ueous Solutions of

Different pH and the Interfacial Tension Against Toluene 30
The change in Ysol of aqueous solutions of poly-DL-alanine, at

25 0 C
and C:0.02g/dl, are shown in Fig. 23 as a function of pH. y 01
becomes the maximum at pH 8.0 and the minimum at 4.5. Since PO~y
DL-alanine is a random copolymer of D- and L-alanine, its conforma-
tion may be thought to be almost random coil and not to vary with
pH. On the basis of the fact that intrinsic viscosities of its
55~------------------------------'
E SO
~
c
>-
"0
45
2 4 6 8 10 12 14
PH
Fig. 23. Surface tension of poly-DL-alanine in

aqueous solutions of different pH at
25 oC, C=O.01g/100ml.
aqueous solutions reach the minimum at pH 8.0 (Katagai et al. (36»,

this pH can be taken for the isoelectric point of the poly-DL-
alanine. However, poly-L-lysine (34) and egg albumin (37) are
quite different from poly-DL-alanine in that the formers show the
minimum Y 1 values near at their isoelectric points respectively.
so
In these cases, it must be taken into consideration that a change of
molecular structure occurs at the isoelectric point.
Our results can be explained as follows: Lower values of Y 1

at both sides of the isoelectric point are probably due to ~Re
increasing balance between lyophilic and hydrophilic (electrically
charged) parts in molecules, so that surface adsorption increases
and Ysol decreases. At the isoelectric point, molecules of poly-DL-
alanine decrease their dimension and more or less aggregate with
each other as known by the viscosity minimum. As a result, the
desorption from surface will be accelerated to increase \01. Below
pH 4.5, the point of the minimum Ysol' charged molecules tend to
exist in bulk phase and increase \01 again.
Interfacial tensions between toluene and aqueous solutions of

poly-DL-alanine (Ysol/T) behave differently from surface tensions
of the solutions (Y sol ). Figure 24 shows Ysol/T vs. pH, where we
can see a broad plateau of Ysol/T over the range of pH 4"'9. The
value of 16-17 dyn/cm at 25 0 C in this region is much smaller than
that between pure water and toluene (55 dyn/cm). These facts
indicate that the methyl side groups of poly-DL-alanine are more
20r-------------------------~
E
~
c::
~15
t-
.....
(510
~
2 4 6 8 10 12
PH
Fig. 24. Interfacial tension between toluene and
aqueous solutions of poly-DL-alanine as
a function of pH at 25 0 C, C=0.01g/100ml.
strongly adsorbed to the toluene/water interface than to the air/

water interface, and as a result the effect of pH changes on the
adsorbability is screened behind it.
On the behavior of ysolfT in the acidic and alkaline end
regions, discussions should be reserved at present, because YIlT
did not reach equilibrium with time even over 200 hrs. (Ir1'°t.he
Figure the values at 100 hrs. are plotted by simple open circles.)
This is probably due to the dissolution of poly-DL-alanine in
toluene in these regions.
Anyhow it can be concluded that comparative studies on surface

and interfacial tensions are quite useful to make clear the impor-
tant role of interface and contacting liquid and to characterize the
surface behavior of polypeptides in solution.
CONCLUSIONS
So far we have discussed several topics from our studies on

surface and interfacial tensions of polymer melts and solutions. As
you see, the measurements of these quantities provide much useful
information not only for technical purposes but also for basic
researches on polymer physics and surface chemistry. The reason is
that surface or interfacial tension is essentially a thermodynamic
quantity and, in addition, it sensitively reflects the change in
surface or interface. As above-mentioned, to make clear the
selective behavior of polymeric materials at interface becomes
increasingly important. The method is classical but essential and
simple. We hope that surface and interfacial tensions are recog-
nized more rightly as a tool for the studies on polymer science and
technology and experimental results can be further accumulated and

their theoretical foundations can be firmly established.
REFERENCES
1. S. Wu, "Interfacial and Surface Tension of Polymers", ~ Mac-

romol. Sci.-Revs., Macromol. Chem. Chem. Cl0, 1 (1974).
2. S. Wu, "Interfacial Energy, Structure, and Adhesion Between
Polymers", (Chap. 6) in Polymer Blends, Vol. 1, D. R. Paul
and S. Newman, eds., Academic Press, New York (1978).
3. T. Hata, "Surface Tension of Molten Polymers" (in Japanese),
Hyomen (Surface) 6, 281 (1968).
4. A.W. Porter, "The Calculation of Surface Tension from Experi-
ment. Part I, Sessile Drops", Phil. Mag . ..l2., 163 (1933).
5. W.D. Bascom, R.L. Cottington, C.R. Singleberry, "Dynami.c Sur-
face Phenomena in the Spontaneous Spreading of Oils on
Solids", in Contact Angle, Wettability and Adhesion
(Advances in Chemistry Series, No. 43), p. 355, Am. Chem.
Soc., Washington, D.C. (1964).
6. T. Kasemura and T. Hata, "Surface Tension of Poly(ethylene
glycol), Poly(propylene glycol) and Their Methyl Ether
Homologs--Molecular Weight Dependence and Methyl Group
Effect", (in Japanese), Kobunshi Ronbunshu 33, 192
(1976).
7. A.K. Rastogi and L.E. St. Pierre, "Interfacial Phenomena in
Macromolecular Systems V. The Surface Free Energies and
Surface Entropies of Polyethylene Glycols and Poly-
propylene Glycols", J. Colloid Interface Sci. 35, 16
(1971). - -- -
8. T. Hata, "Surface Chemistry of Polymers", (in Japanese),
Kobunshi (Polymers) 17, 594 (1968).
9. K. Ueberreiter and G. Kanig, "Die Ketteniangenabhangigkeit des
Volumens, des Ausdehnungs-Koeffizienten und der
Einfriertemperatur von fraktionierten Polystyrolen", Z.
Naturforsch. 6a, 551 (1951). -
10. D. G. LeGrand and G. L. Gaines, Jr., "The Molecular Weight
Dependence of Polymer Surface Tension", ~ Colloid Inter-
face Sci.TI, 162 (1969).
11. Y. Kitazaki and T. Hata, unpublished data.
12. T. Kasemura, N. Yamashita, K. Suzuki, T. Kondo and T. Hata,
"Surface Tension of Alkanediols and Their Condensates,
Poly(alkylene glycol)s", (in Japanese), Kobunshi Ronbun-
shu 35, 215 (1978).
13. H. Tozaki-and T. Hata, presented to the 26th Ann. Meeting of
Soc. Polym. Sic., Japan (1977).
14. T. Kasemura, F. Uzi, T. Kondo and T. Hata, "Surface Tension of
Poly(vinyl alkylate)s", (in Japanese), Kobunshi Ronbunshu
36 (1979), in press.
15. T. Kasemura, N. Yamashita, T. Kondo and T. Hata, "Surface

Tension of Dialkyl Phthalates", (in Japanese), Kobunshi
Ronbunshu 33, 703 (1976).
16. F.M. Fowkes, "Determination of Interfacial Tensions, Contact
Angle, and Dispersion Forces in Surfaces by Assuming
Additivity of Intermolecular Interactions in Surfaces",
~ Phys. Chem. 66, 382 (1962); "Additivity of Intermole-
cular Forces at Interface I", ibid. 67, 2538 (1963);
"Dispersion Force Contributions to Surface and Inter-
facial Tensions, Contact Angles, and Heats of Immersion"
in Contact Angle, Wettability and Adhesion (Advances in
Chemistry Series No. 43), p. 99, Am. Chem. Soc. (1964).
17. T. Kasemura, N. Yamashita, K. Suzuki, T. Kondo and T. Hata,
"Composition Dependence of Surface Tension of Copoly-
mers", (in Japanese), Kobunshi Ronbunshu 35, 263 (1978).
18. A.K. Rastogi and L.E. St. Pierre, "Interfacial Phenomena in
Macromolecular Systems III, The Surface Free-Energies of
Polymers", ~ Colloid Interface Sci. 1!., 168 (1969).
19. T. Kasemura, K. Suzuki, F. Uzi, T. Kondo and T. Hata, "Compos-
i tion Dependence of Surface Tension of Ethylene Oxide-
Proplyene Oxide Copolymers", (in Japanese), Kobunshi Ron-
bun shu 35, 779 (1978).
20. T. Nose, "A Hole Theory of Polymer Liquids and Glasses V.
Surface Tension of Polymer Liquids", Polymer ~ 1, 1
(1972), ibid VI, Mixtures, ibid 3, 196 (1972).
21. H. Tozaki and T. Hata, presented to the 26th Ann. Meeting of
Soc. Polym. Sci., Japan (1977).
22. H. Uraki and T. Hata, presented to the 13th Symp.on Adhesion
( 1975) .
23. D.K. Owens and R.C. Wendt, "Estimation of the Surface Energy of
Polymers", ~!E£h Polym. Sci. ..11, 1741 (1969).
24. D.H. Kaelble and K.C. Uy, "A Reinterpretation of Organic
Liquid-Polytetrafluoroethylene Surface Interactions", J.
Adhesion 2, 50 (1970). --
25. R. K. S. Chan, "Surface Tension of Fluoropolymers I. London
Dispersion Term", !!..:. Colloid Interface Sci. 32, 492
(1970), "II. The Polar Attraction Term", ibid. 32, 499
( 1970).
26. T. Hata, See Ref. (8); Y. Kitazaki and T. Hata, "Extension of
Fowkes' Equation and Estimation of Surface Tension of
Polymer Solids", (in Japanese), ~ Adhesion Soc., Japan
8, 131 (1972).
27. S. Wu-;-"Calculation of Interfacial Tension in Polymer Series",
~ Polym. Sci. C 34, 19 (1971).
28. M. Sudo, H. Uraki and T. Hata, presented to the 12th Symp. on
Adhesion (1974).
29. H. Arakawa, H. Tozaki and T. Hata, presented to the 24th Symp.
on Polym. Sci. (1975).
30. H. Tozaki, H. Uraki and T. Hata, presented to the 13th Symp.
on Adhesion (1975), see also Ref. 29.
31. H. Arakawa, M. Oya and T. Hata, presented to the 27th Ann.

Meeting of Soc. Polym. Sci., Japan (1978).
32. C. Robinson, "Liquid-Crystalline Structures in Solution of a
Polypeptide", Trans. Farad. Soc. 52, 571 (1956).
33. K. Kurotsu, K. Kikuchi, A. Tsutsumi and M. Kaneko, "Electric
Birefringence of Poly(y -benzyl-L-glutamate) in Ethylene
Dichloride", Polym. ~~, 571 (1974).
34. A.W. Neumann, M.A. Moscarello and R.M. Epand, "The Application
of Surface Tension Measurements to the Study of Conforma-
tional Transitions in Aqueous Solutions of Poly-L-
Lysine", Biopolym. 1£, 1945 (1973).
35. S. Morimoto and K. Ueberreiter, "Surface Tension of Polymer
Solutions. II. Poly- -Benzyl-L-Glutamate in N, N-
Dimethy1 formamide, Dioxane and Dichloracetic Acid
Solutions", Kolloid-Z. u. Z. fur Polymere 253, 1009
(1975). -
36. R. Katagai and J. Shinomiya, "The Viscosity Change of Poly-DL-
alanine in Aqueous Solution", Makromol. Chem. 127, 282
(1969). - - --
37. E.A. Hauser and L.E. Swearingen, "The Aging of Surfaces of
Aqueous Solutions of Egg Albumin", ~ Phys. Chem. 45,644
(1941) •
Donor -Acceptor Interactions
at Interfaces
F.M. Fowkes
Department of Chemistry
Lehigh University
Bethlehem, PA 18015
ABSTRACT
Nearly all intermolecular interactions in solution and at

interfaces can be reduced to two phenomena: London dispersion
forces, and electron donor-acceptor (acid-base) interactions.
Hydrogen bonds are included in acid-base interactions, and dipole
phenomena are usually negligibly small. Earlier popular notions
that all "polar" groups can interact with each other are shown
untenable; donor-donor and acceptor-acceptor interactions are
negligibly small compared to donor-acceptor interactions. Sup-
porting data include surface and interfacial tensions, contact
angles, and adsorption of polymers from organic sol vents onto
inorganic powders.
INTRODUCTION
Intermolecular forces between unlike molecules in solutions or
at interfaces determine solution and interfacial properties. There
are so many different kinds of intermolecular interactions that for
"practical" studies most investigators of interfacial phenomena
have tried to lump all interactions into two terms: dispersion
force interactions and "polar" interactions. The most widely used
equations for "polar" interactions are geometric mean equations
suitable for dipole-dipole interactions, but completely unsuitable
for acid-base interactions, as will be explained. This unfortunate
approach has also been used in solution studies.
43
44 F. M. FOWKES
R.S. Drago and co-workers have recently correlated enthalpies

of interaction of over thirty acids and bases in neutral solvents
such as carbon tetrachloride (1). The correlation gives predic-
tions of all hydrogen-bonds, Br,nsted acids and bases, and Lewis
acids and bases with good accuracy (except for some interactions
which are greater than 80 kJoules/mole), and does not need to take
into account any dipole-dipole interactions.
One might well distinguish between donor acceptor interactions

in which an electron is transferred to the acceptor and acid-ba~e
interactions in which protons are transferred to the base. However,
for the interactions discussed in this work the electrons are not
completely transferred from one molecule to the other, but shifted
in distribution; consequently these electron donor-acceptor inter-
actions and acid-base interactions are indistinguishable and both
are accurately predicted by the Drago relation. Drago's work
suggests that in solution studies one can assume that all intermole-
cular interactions can be closely approximated by just two pheno-
mena, dispersion force interactions and acid-base interactions, and
that dipole-dipole interactions are usually too small to need to
consider. This approach is very different from current practice in
both principle and in the kinds of equations used.
This paper used the above approach for evaluation of inter-

facial phenomena (work of adhesion, contact angles, and adsorption)
and it is found that many interfacial phenomena can be best pre-
dicted by treating interfacial forces as consisting of only disper-
sion forces and acid-base interactions, and neglecting dipole-
dipole interactions.
In 1961 the author proposed that since intermolecular attrac-
tions result from several fairly independent phenomena (such as
disperion forces (d), dipole interactions (p), hydrogen-bonds (h),
etc.), it is reasonable to separate out such terms in the work of
adhesion
(1)
and in the surface tension

d . .P h (2)
Y=Y+T+Y+
This principle leads to the widely used equation
with which interfacial tensions, contact angles, free energies of

adsorption, and Hamaker constants were successfully calculated
(2,3). In confining the use of this equation to dispersion force
DONOR-ACCEPTOR INTERACTIONS 45
interactions the geometric mean expression is correct. There is

also some reason to treat dipole-dipole interactions separately
with the geometric mean expression
Wp - - rp,-;y
P (4)
A - ¥lY2
The interaction energy between two dipoles is -~~~/3kTr~2' (where
~ is the dipole moment, k Boltzmann's constant, and T absolute
temperature) so if the distance between dipoles r 12 is the geometric
mean of r 11 and r 22 of the pure materials, then equation (4) is
correct.
The extension of equations (1), (2), and (4) to try to predict

hydrogen-bonding with a geometric mean expression is quite incor-
rect, for hydrogen-bond acceptors such as ethers, esters, or aro-
matic hydrocarbons cannot themselves form hydrogen-bonds, and
therefore yh is zero for such materials, even though these materials
have a large W~ with hydrogen-donors (4). Similarly some hydrogen-
donors such as chloroform have zero values of y h , but large values
h
of WA (23).
Zisman's extensive series of contact angle measurements of
organic liquids on polymer surfaces (5) have tempted several inves-
tigators (6-9) to try to solve equation (1) by assuming that all
polar and hydrogen-bonding interactions can be predicted by equa-
tion (4). This forced fit of data into incorrect equations makes
all conclusions at least a little wrong (e.g., finite yP values for
polyethylene and paraffin).
In solution studies the separation of heats of ml.Xl.ng (6HM)

into several terms soon followed. Blanks and Prausnitz (10) used
only two terms (polar and nonpolar) while Gardon (11) and Meyer and
Wagner (12) included terms for dipole-dipole and for dipole-induced
dipole interaction. Although hydrogen-bonding had been recognized
much earlier ( 13, 14), it was Hansen (15) who brought forth the
widely-used three-dimensional solubility parameter:~):
(5)
and
(6)
The dispersion force term (6Hd), or more properly the change in

internal energy on mixing (6U~~, can be correctly evaluated with
46 F. M. FOWKES
the geometric mean
6H~ = 6D~ = Vm~1~2(~~ _ ~~)2 (7)
and perhaps a dipole-dipole term (6U~) can be justified:
(8)
but for reasons given in the previo~ Pftragraph there is no way that
6ijh can be predicted with a term (~ -~2)2 and consequently any two
m 1
sets of correlations are in serious disagreement (16).
The extensive computer-forced correlations of heats of mixing

and of works of adhesion (7,15,17,18) are based on treating hydro-
gen-bonds by the geometric mean, as if the hydrogen-bond were a
dipole-dipole interaction, a notion abandoned about twenty years
ago (19). A more recent approach has been developed by R.S. Drago
(1) who has treated the hydrogen-bond as a Lewis acid-base inter-
action. Drago has correlated the enthalpy of acid-base interaction
~b (in a neutral solvent such as carbon tetrachloride) with two
constants (EB,CB) for each base and two constants for each acid
(EA,C A):
_AH~b CC EE
Ll_~ = A B + A B
Drago assumed that ~~ and ~HP were negligibly smal~ for the acids
and bases, and in his correla~ion all predicted 6H~ values (up to
80 kJoules/mole) checked measured values within about 5% or less.
Drago's correlation (treating interactions as due only to

dispersion forces and acid-base interactions) is much more success-
ful than Kaelble's and Hansen's correlations (treating interactions
as due only to dispersion forces and polar interactions predictable
from geometric mean equations). It is of special interest that in
Drago's correlation the neglect of dipole-dipole interactions gave
no prdblems, suggesting that dipole-dipole interactions are negli-
gibly small compared to acid-base and dispersion force interac-
tions.
We propose that the heat of mixing (6HM) be given by
6~ = P6VM+Vm~1~2(~~ - ~~)2 - Xp(CACB+EAEB) + 6U~2 (10)

where X is the mole fraction of acid-base pairs per mole of

componeJis present. Similarly, we propose for the work of adhesion:
_ ~ moles of acid-base pairs + wP

WA = ~AYB - f(CACB+EAE B) x unit area A ( 11)
in which the constant f (near unity) converts enthalpy per unit

area into surface free energy, and the last term is usually small.
The Acidity or Basicity of Polymer Surfaces
The literature abounds with contact angle measurements of

liquids on the surface of solid polymers (5,20) but few of these
measurements are suitable for defining acid-base interactions.
Consequently, we have recently concluded a study (21) of well-
characterized acidic and basic polymers, and measured the contact
angles e of certain acidic or basic liquids from Drago's tables.
In these studies the work of adhesion is given by:

ab p
WA = YL (1 + cos e) + ne = WAd + WA + WA + ... (12)
where n is any reduction of the surface energy of the polymer

resultin~ from adsorption of the vapor of the test liquid. In
general n is expected to be negligibly small for high energy
e
liquids on low energy solids, and we have measured n for several
e
such systems and find it to be indeed zero (22). The experimental
approach in this work has been to determine
(13)
for a series of Drago-characterized acidic or basic liquids on

polymers of controlled acidity or basicity.
The findings (Figs. 1 and 2) were that A w-w1-

ne is zero for
acidic liquids on acidic surfaces (phenolic liquids on acrylic acid
sites) or for basic liquids on basic surfaces (pyridine on vinyl
acetate sites), though quite appreciable with acid base-inter-
actions (phenolic liquids on vinyl acetate sites or pyridine on
acrylic acid sites).
In general, these two sets of contact angle measurements agree

with the general conclusions concerning the Drago correlation:
intermolecular forces are dominated by dispersion force interac-
tions and acid-base interactions, while dipole-dipole interactions
are negligibly small.
48 F. M. FOWKES
0.1 N NaOH 0
N
:;:
.....
'-'
C/O
3
""~
ex;
,,",'"
,
coc:::
~
<C
""
6
~
N
2
Q)
C/O
0
'-'
~,
;:---'
1
35% Phenol in TCP~

a
a 1% 27. 3% ACRYLIC ACID 5% 6% 7"L
Fig. 1. Effect of solvnnt-acrylic copolymer inter-

action on WA-WA-lf E.
7 ~ Phenol in
6 48
Pyridine
O~----------~------~--~'---------~-----
!l lOX VINYL ACETATE 20% 30~ 0
Fig. 2. Effect of solvent-v~nYlacetate copolymer
interaction on WA-WA-W E.
Acid-Base Interactions in Polymer Adsorption

A second route to explore the phenomenon of acid-base inter-
actions at interfaces has been to measure the amount of adsorption
of acidic or basic polymers onto acidic or basic inorganic powders
from solution in acidic or basic organic solvents (23). Polymethyl-
methacrylate (PMMA) was the basic polymer and post-chlorinated
polyvinylchloride (C1-PVC) was the acidic polymer. The methacryl-
ate ester groups are as basic as the acetate groups characterized by
Drago and C1-PVC is nearly as acidic as chloroform. The acidic
solid was silica; the SiOH silanol groups on the silica surface are
well-known to be acidic, and are known to form acid-base bonds with
methacrylate groups in the adsorption of methacrylate polymers
(24). The basic solid, calcium carbonate, is well-recognized for
its basicity.
The basic polymer (PMMA) adsorbed very slightly onto the basic
solid (CaC0 3); about 4% of a complete monolayer adsorbed from CCl 4
50 F. M. FOWKES
or benzene, and no adsorption was measurable from the other sol-

vents. In contrast, PMMA adsorbed strongly onto the acidic Si0 2 ,
especially from the more neutral solvents (Fig. 3).
It appears that the basicity of the methacrylate groups must be

less than the basicity of dioxane, but much greater than the
basicity of benzene. These considerations suggest that the heat of
interaction of PMMA with t-butanol should be less than the 12.3
kJ/mole of dioxane, but much greater than the 5.0 kJ/mole of
benzene. This is confirmed by Drago's 10.5 kJ/mole for the inter-
action of esters with t-butanol.
The adsorption of the acidic polymer (C1-PVC) onto acidic

sil~ga w~s so small that it was less than our limits of detection
(10 g/m, or 0.3% of a monolayer); this again illustrates the
unimportance of dipole-dipole interactions between two acids. On
the other hand, the acidic polymer was strongly adsorbed onto the
basic filler (CaC0 3 ) from the more neutral solvents (Fig. 4).
COMPETITION BETWEEN COMPETITION BETWEEN

SOLVENTS AND PMMA SOLVENTS AND Si02
FOR Si02 FOR PMMA
THF
~-----L----~~----7-----~----~~--~~
15 kJ/mole 5 0 5 10 15
I1H ab vs. BuOH aH ab vs. EtAc
Fig. 3. Competition between solvents and solid

surfaces during adsorption (PMMA vs. Si0 2 ).
COMPETITION BETWEEN COMPETITION BETWEEN

SOLVENTS AND CaC03 SOLVENTS AND CI-PVC
FOR CI-PVC FOR CaC0 3
o 5 10
6H ab VS. EtAc
Fig. 4. Competition between solvents and solid

surfaces during adsorption (CI-PVC vs. CaC0 3 ).
In adsorption of polymers from solution, the acidity or basi-

city of the solvent was found to play an important role; a solvent
more basic than the basic polymer is preferentially adsorbed, and a
solvent more acidic than the acidic solid is preferentially bound to
the basic polymer. Therefore, strong polymer adsorption is ob-
served only from solvents more neutral than the adsorbing polymer or
the adsorbent surface sites.
In retrospect we find these acid-base rules for adsorption to

govern most of the earlier work. The twenty-year old studies of
Koral, Ullman , and Eirich (25) for the adsorption of polyvinyl-
acetate (basic) onto iron powder (acidic) showed strong adsorption
from neutral (CCI~) or weakly acidic (C 2 H4 CI ) solvents but no ad-
sorption from acetonitrile (basic). Simllarty, the current studies
of G.J. Howard and co-workers (26) show strong adsorption of
polyacrylonitrile polymer (basic) on silica (acidic) from weakly
acidic solvents (trichloroethylene) but not when proprionitrile is
the solvent (equally basic, but more concentrated than the adsorb-
ing polymer).
ACKNOWLEDGEMENTS
The author is deeply indebted to several co-workers; without
their work this concept would be only a hazy notion. Special thanks
go to S. Maruchi (of Sanyo-Kokusaku Pulp Co., Tokyo), Prof. J.A.
Manson, M. Marmo (now at Eastman Kodak), M.A. Mostafa, and G. Sie.
Support has come from Ford Motor Company, Sanyo-Kokusaku Pulp Co.,
and Lehigh University.
52 F. M. FOWKES
REFERENCES
1. R.S. Drago, G.C. Vogel and T.E. Needham, J. Am. Chem. Soc. 93,
6014 (1971); R.S. Drago, L.B. Parr and~S. Chamberlain,
~ Am. Chem. Soc. 99, 3203 (1977).
2. F.M. Fowkes, ~ Phys. Chem. 66, 382 (1962).
3. F.M. Fowkes, Ind. ~ Chem. g, 40 (1964).
4. F.M. Fowkes, J. Adhesion 4, 155 (1972).
5. W.A. Zisman, Advances in- Chemistry "Contact Angles, Wetta-
bility, and Adhesion", ed. F.M. Fowkes (ACS Special Pub-
lication) 43, 1 (1964).
6. D.K. Owens and R.C. Wendt, ~ ~ Polymer Sci. .11, 1741
(1969) .
7. D.H. Kaelble, J. Adhesion 2, 66 (1970).
8. S. Wu, J. AdheSion 5, 39 (1973).
9. J. Kloubek, J. Adhesion 6, 293 (1974).
10. R.F. Blanks and J.M. Prausnitz, Ind. Eng. Chem. 1, 1 (1964).
11. J.L. Gardon, J. Paint Technol. 38, 43 (1966).
12. E.F. Meyer andR~agner, ~ Phys. Chem. 70, 3162; 75, 642
(1971) .
13. W. Goody, ~ Chem. Phys. 1, 93 (1939); ~, 170 (1949); ,2., 204
(1941).
14. P.A. Small, ~ ~ Chem. 1, 71 (1953).
15. C.M. Hansen, J. Paint Technol. 39, 104 and 505 (1967).
16. E.B. Bagley and-:r:M. Sciglian~ Chapter XVI "Solutions and
Solubilities", ed. M.R.J. Dack, Vol. VIII of "Techniques
of Chemistry", A. Weissberger, ed. Wiley, New York
( 1975) .
17. K.L. Hoy, ~ Paint Technol. 42, 118 (1970).
18. C.M. Hansen and A. Beerbower, "Solubility Parameters in "Ency-
clopedia of Chemical Technology", Supplement Volume, 2nd
Edition, Wiley, New York (1971).
19. G.C. Pimental and A.L. McClellan, "The Hydrogen Bond", Free-
man, San Francisco (1960).
20. J.R. Dann, J. Colloid Interface Sci. 32, 302 and 321 (1970).
21. F.M. Fowkes--and S. Maruchi, CoatingS-and Plastics Preprints
37, 605 (1977).
22. F.M. Fowkes and D.C. McCarthy, to be published.
23. F.M. Fowkes and M.A. Mostafa, Ind. Eng. Chem. Prod. R&D 11, 3
(1978).
24. B.J. Fontanta and J.R. Thomas, ~ Phys. Chem. 65, 480 (1961).
25. J. Koral, R. Ullman and F.R. Eirich, ~ Phys. Chem. 62, 541
( 1958).
26. G.J. Howard and M.J. McGrath, ~ Poly Sci. (Poly. Chem. Ed.)
.l2., 1721 (1977).
Surface Tension of Solids:
Generalization and Reinterpretation of
Critical Surface Tension
Souheng Wu
E.I. du Pont de Nemours & Company
Central Research and Development Department
Experimental Station, Wilmington, Delaware 19898
ABSTRACT
The concept of critical surface tension is generalized and
reinterpreted in terms of a proposed equation of state. The
equation defines a spectrum of critical surface tensions for a
given surface, and provides a method by which the surface tension
can be accurately determined from the contact angles of a series of
testing liquids. The surface tensions obtained for solid poly-
mers, organic solids and monolayers by this method agree remark-
ably well with those obtained from melt data (temperature depen-
dence), liquid homologs (molecular-weight dependence) and the
harmonic-mean equation. In contrast, those obtained by the geo-
metric-mean equation and Zisman' s critical surface tensions are
often too low. These results also support the validity of the
harmonic-mean equation.
CRITIQUE OF SOME EXISTING METHODS FOR DETERMINING THE SURFACE

TENSION OF SOLIDS
The surface tension of a solid is difficult, if not impos-

sible, to measure directly, because large reversible deformation
of a solid surface cannot be done. A number of indirect methods
based on empirical or semi-empirical relations between the wet-
tability and the surface tension have been proposed, including the
critical surface tension (1-4), the method based on the harmonic-
mean equation (5-9), and the method based on the geometric-mean
equation (10-12). Several theoretical treatments are also avail-
53
54 SOUHENG WU
able (13-20).
Zisman's Critical Surface Tension
This, first proposed by Fox and Zisman (1), has been widely
used. However, its value is quite variable, depending on the
nature of the testing liquids used. For instance, the critical
surface tension for poly(ethylene terephthalate) varies from 27 to
46 dyne/cm at 20°C (21) j the value for perfluorodecanoic acid
monolayer varies from 11 to 27.5 at 20°C (22). Moreover, the
critical surface tensions of liquids have been found to be signifi-
cantly lower than their surface tensions which can be measured
directly and accurately. For instance, the critical surface
tension of water is about 20 dyne/cm, compared with its surface
tension of 72.8 dyne/cm at 20°C (23,24). Zisman (22) has cau-
tioned against equating the critical surface tension to the sur-
face tension.
Geometric-Mean Method and Harmonic-Mean Method
Recently, semi-empirical equations relating the wettability

of the surface tension based on the theory of fractional polarity
(6-8) have been proposed, providing methods for determining the
surface tension and the polarity of solid surfaces from contact
angles of two testing liquids. These are, namely, the geometric-
mean method (10-12) and the harmonic-mean method (5-9). We will
show that the harmonic-mean method gives accurate results, whereas
the geometric-mean method does not.
The geometric-mean equation is an empirical extension of the
Fowkes I equation (25). The experimental proofs of the Fowkes'
equation consist mainly of the following. (a) The dispersion-
force component Yd of water as calculated from the interfacial
tensions between water and a series of n-alkanes is rather con-
stant (25). (b) The equation correctly predicts the interfacial
tension between water and mercury (25). However, when the equa-
tion is applied to water/organic liquids or DJe'i:'cupy!organic
liquids systems, large deviations are found. These deviations are
attributed to polar interactions which are, however, not accounted
for in the Fowkes' equation.
Moreover a the surface tension y and its dispersion and polar

components, y and yP, respectively, for a given solid surface as
calculated by the geometric-mean equation vary widely, depending
on the nature of the testing liquids used. For instance, for nylon
66 at 20° C, Y varies from 35.7 to 45.8 dyne/cm, y d from 20.7 to
40.2 dyne/cm, and yP from 3.0 to 14.9 dyne/cm (12). Such wide
SURFACE TENSION OF SOLIDS 55
variations show that the geometric-mean equation is only a rough

approximation.
On the other hand, Wu (5-8) has accurately measured both the

surface and interfacial tensions of many incompatible pairs of
polymer liquids and melts. In such systems, stable equilibrium
interfaces can be formed so that the surface and interfacial
tensions can be measured accurately and directly. Such data
provide an excellent opportunity for testing and verifying the
equations experimentally. By using the data for fifteen pairs of
polymer liquids and melts, Wu (5-8) showed the validity of the
harmonic-mean equation, whereas the geometric-mean equation was
found to be in serious error in many cases (5-8).
THE EQUATION OF STATE METHOD
Here, we would like to propose a novel equation of state,

generalize and reinterpret the critical surface tension, and pro-
vide a method by which the surface tension of solids can be
accurately determined from wettability data. We will also show
that the surface tensions of solids obtained by the present method
agree remarkably well with those obtained from the melt data, the
liquid homo logs and the harmonic-mean method. In contrast, the
values obtained by the geometric-mean method and the Zisman' s
critical surface tensions are often too low.
The Proposed Equation of State
Combining the Young's equation (26) with the equation of Good

and Girifalco (27) gives
cos e (1)
where e is the equilibrium contact angle, $ the interaction para-

meter, YS the surface tension of the solid in vacuum, Y LV the
surface tension of the liquid in equilibrium with its saturated
vapor, and ne the equilibrium spreading pressure. The $ has been
explicitly expressed in terms of molecular parameters (27), or in
terms of macroscopic polarities of the two phases. (6-8) The
critical surface tension may be defined as
Yc = Lim YLV (2 )
e-+-O
Thus, combining eqs. (1) and (2) gives
56 SOUHENG WU
2 2 2
~ = cp YS - 1Te + ( 1Te 14 cp YS ) + •••
which converges rapidly, and therefore, can be truncated to
(4)
where we have used Y '" to replace Y , i. e., Y = Y , to emphasize

c, '" c c, - c
the fact that Y is a function of <1>. Using eq. (14) in eq. (1) gives
c
1/2
cos e = 2(y c,/Y LV ) - 1 (5)
which is strikingly simple, and will be termed the equation of

state.
Application of the Equation of State
Eq. (5) can be rearranged to

2
Yc , = (114) (1 + cos e) YLV (6)
which can be used to calculate Y '" from the contact angle of any
c,'"
one testing liquid. When a series of testing liquids are used, a
spectrum of Y '" values will be obtained. The maximum Y. '" value is
equal to YS . cThis
'''' is because = 1 occurring when thec,polarities
'"
ma~
of the two phases are identical t6-8,21), and if 1T is small, then
from eq. ( 4 ) , e
Yc,,max = YS - lTe = YS (1 )
Thus, if the spectrum of Y is plotted vs. the polarity of the

testing liquids, the data pgt~ts will fall into a smooth curve with
the maximum Yc,,,,,max
'" equal to Ys .
However, since the polarity of many testing liquids are un-

known, the Yc , spectrum may instead be plotted vs. Y LV. In this
case, the data points will be somewhat scattered, since y. is a
function of not only YLV ' but also (or the polarity). Ho~~$er, a
smooth curve can be drawn just to envelop all the data points below,
as illustrated in Figs. 1-4 for some polymers, organic solid and
monolayers. Table 1 gives the Y '" spectra for polyethylene and
c,'"
polytetrafluoroethylene as calculated by eq. (6) as examples. All
contact angle data are from Refs. (1-4,11). It can be seen that
Y occurs at large e values, so that we may safely let 1T = o.
c,m,max e
ThuS, Yc,,,,,max
'" = Ys .
50
O~--~--~----~--~-- __ ~ __- L_ _ _ _L -_ _ ~
m ro ~ ~ ~ w ro 00 ~
SURFACE TENSION OF TESTING LIQUIDS, YLv, dyne/em ot 20·C
Fig. 1. Equation of State Plots for Polyethylene

and Polytetrafluoroethylene.
50 Ye,.", max
Poly (vinylidene chlor ide)
40
10
O~--~----~--~~--~----~--~----~--~
10 20 30 40 50 60 70 80 ~
SURFACE TENSION OF TESTING LIQUIDS, YLv, dyne/cm
Fig. 2. Equation of State Plots for Poly(vinylidene

chloride)and Poly(vinylidene fluoride).
58 SOUHENG WU
50
lYe .•.mal
Poly (methyl methacrylate)
40
I
g
..... 30
t
-0 o
o
o
10
O~--~-----L----~ ____L -_ _ ~ ____ -L____ ~ __ ~
10 20 30 40 50 60 70 80 90
SURFACE TENSION OF TESTING LIQUID, YLV' dyne /cm
Fig. 3. Equation of State Plots for Poly(methyl-

methacrylate) and Paraffin Wax.
50
40
.li
..... 30
~
-
-0
-e-
~ 20
10
OL---~----~----~--~----~----~----~--~
10 w ~ ~ ~ w ro 00 90
SURFACE TENSION OF TESTING LIQUID. YLV' dyne/em
Fig. 4. Equation of State Plots for =CCl 2 Surface

(i.e., Adsorbed Monolayer of Perchloro-2,4-
pentadienoic Acid) and -CH~ Surface (i.e.,
Adsorbed Monolayer of OctaOecylamine).
Zisman's critical surface tension, designated as y Z' can be

defined as yc, Z = Iyc ,~I' !e' -_ o· It can be seen in Figur~S 1-4 that
Ye, Z is determined only by the few low-contact-angle data. In fact,
Yc Z is little affected by all the higher contact-angle data in a
Zl.!man plot.
Comparison Among Various Methods

Table 2 lists the YS values obtained by the present equation
of state method for various solid polymers, organic solids and
monolayers. Also included for comparison are those obtained from
molecular-weight dependence of liquid homologs (7,8,28), tempera-
ture dependence of melts (5-10), the harmonic-mean method (5,6,9),
the geometric-mean method (5,11,12) and Zisman's critical surface
tension (4). The results obtained from the melt data and the liquid
homolog data are accurate within 0.1 to 0.5 dyne/em (8). It can be
seen that the results obtained by the equation of state method, the
melt data, the liquid homolog data are generally in excellent
agreement with one another, whereas those obtained by the geo-
metric-mean method and Zisman's critical surface tensions are often
too low.
Remarks on Some Materials

Polyethylene: The YS at 20°C is 35.9 dyne/cm by the equation
of state method, 35.7 dyne/cm from melt data (29), 34.7 dyne/cm from
liquid homo logs (8), 36.1 dyne/cm by the harmonic-mean method (5),
in contrast to 32.4 dyne/cm by the geometric-mean method (11,12)
and 31 dyne/cm for Zisman's critical surface tension (4).
Polytetrafluoroethylene: The '5 at 20°C is 22.6 dyne/cm by the

equation of state method, 23.9 dyne/cm from liquid homo logs (8),
22.5 dyne/cm by the harmonic-mean method (5), in contrast to 15.6-
19.1 dyne/cm by the geometric-mean method (11,12) and 18.5 dyne/cm
for Zisman's critical surface tension (4).
The surface tension of a fluorocarbon oil elF 44 has been
directly measured to be 21.5 dyn/cm at 20°C by Dettre and Johnson
(30). As the molecular weight increases, the surface tension should
increase. It is thus noteworthy that the YS values for solid
polytetrafluoroethylene at 20°C obtained by the equation of state
method, the liquid homo logs and the harmonic-mean method are all in
the range of 22 to 24 dyn/cm, consistent with expectation. In
contrast, the geometric-mean method and the Zisman's critical sur-
face tension are too low.
g;
Table 1. CALCULATION OF CRITICAL SURFACE TENSION SPECTRA FOR POLYETHYLENE
AND POLYTETRAFLUOROETHYLENE
YLV POl;y:eth;y:lene Pol;y:tetrafluoroeth;y:lene

dyne/cm e y c,<\> e Yc,<\>
Liquids at 20°C degree dyne/cm degree dyne/cm
n-Alkanes
Hexane 18.4 12 18.0
Heptane 20.3 21 19.0
Octane 21.8 26 19.7
Decane 23.9 35 19.8
Dodecane 25.4 42 19.3
Hexadecane 27.6 46 19.8
n-Alk;y:benzenes
Hexylbenzene 30.0 52 19.6
Benzene 28.9 46 20.7
Halocarbons
Methylene iodide 50.8 52 33.2 88 13.6
sym-Tetrabromomethane 49.7 79 17.6
Tetrachlorobiphenyl 45.3 81 15.1
a-bromonaphthalene 44.6 35 36.9 73 18.6
Hexachloropropylene 38.1 65 19.3 en
Perchlorocyclopentadiene 37.5 67 18.1 0
sym-Tetrachloroethane 36.3 10 35.7 56 22.1 c
J:
Hexachlorobutadiene 36.0 60 20.2 m
zG')
Tetrachloroethylene 31.7 49 21.7
Carbon tetrachloride 26.8 36 21.9 :E
c
Esters (I)
c
Tricresyl phosphate 40.9 34 34.2 75 16.2 :Il
'"T1
Benzylphenyl undecanoate 37.7 28 33.4 67 18.2 »
()
Bis(2-ethylhexyl)phthalate 31.2 5 31. 3 63 16.5 m
Bis(2-ethylhexyl)sebacate 31.3 62 16.8 -t
m
Z
(I)
Miscellaneous (5
z
t-Butyl naphthalene 33.7 7 33.4 65 17.0 0
'"T1
Formamide 58.2 77 21.8 92 13.5 (I)
Glycerol 63.4 79 22.5 100 10.8 0

r
Water 72.8 94 15.8 108 8.7 0
(I)
Mercury 485.0 150 2.2
~
e;
Table 2. COMPARISON AMONG SURFACE TENSION VALUES OBTAINED BY VARIOUS METHODS FOR
SOLID POLYMERS, ORGANIC SOLIDS AND MONOLAYERS
Zisman's
Critical
Surface Surface Tension, YS' dyne/cm at 20° C
Tension By Eq. of From From By Geometric
State Plot, Melt Liquid By Harmonic Mean Equa-
dyn~
Material Yc ~ cm Data Homo logs Mean Equation tion
YC z zmax
Polymers
Polyethylene 31 35.9 35.7 34.7 36.1 33.2
Polytetrafluoro- 18 22.6 23.9 22.5 19.1
ethylene
Polytrifluoro- 22 29.5 27.3 23.9
ethylene
Poly(vinylidene 25 36.5 33.2 30.3
fluoride)
Poly(vinyl 28 37.5 38.4 36.7
fluoride)
Poly(vinylidene 40 45.2 45.4 45.0
chloride)
Poly (vinyl 39 43.8 41.9 41.5
chloride) ~
Polystyrene 33 43.0 40.7 42.6 42.0 c
J:
Poly(methyl 39 42.5 41.4 41.2 40.2 m
zG)
methacrylate)
Polychlorotri- 31 32.1 30.9 30.1 27.5 =E
c
fluoroethylene
(I)
TFE/CTFE 60140 (a) 24 26.2 25.2 20.9 c

:II
TFE/CTFE 80120 (a) "T1
20 22.1 21.3 16.5 »
E/TFE 50/50 (b) 26.5 24.1 C')
27.1 27.6 m
Nylon 66 46 43.8 46.5 44.7 47.0 -I
m
Poly(ethylene 43 44.0 44.6 42.1 41.3 z
(I)
terephthalate) 0
z
Orsanic Solids 0
"T1
Paraffin Wax 23 32.0 35.0 31.0 25.4 ~

r
0
Hexatriacontane 21 23.0 31.4 23.6 19.1 en
MonolaIers (Adsorbed on Platinum)

Octadecylamine 22 30.5 28.1 25.8
Perfluorolauric acid 5.6 14.5
Perfluorodecanoic acid 11 26.2 19.4 14.1
-CF2H Surface (mono- 15 26.5
hydroperfluoro-
undecanoic acid)
=CC1 2 Surface 43 49.5 48.4 46.2
(Perchloro-2,4-
pentadienoic acid)
(a) TFE/CTFE = copolymer of tetrafluoroethylene and chlorotrifluorethylene.

(b) EITFE = copolymer of ethylene and tetrafluoroethylene.
f)
64 SOUHENGWU
Paraffin Wax: The YS at 20°C is 32.0 dyne/cm by the equation

of state method, 35.0 dyne/cm from melt data (31) and 31.0 dyne/cm
by the harmonic-mean method. In contrast, the geometric-mean
method (12) gives 23.7 dyne/cm; the Fowkes' equation gives 25.5
dyne/cm; Zisman's critical surface tension (4) is 15-22 dyne/cm.
The surface tension of molten paraffin wax at its melting point

of 65 0 C has been directly measured to be 32 dyn/cm ~ Padday (31).
As the molten paraffin wax solidifies and cools to 20 C, its surface
tension should increase. Thus, it is noteworthy that the YS values
for solid paraffin wax at 20 0 C obtained by the equation of state
method, the melt data and the harmonic-mean method are all in the
range of 31 to 35 dyn/cm, consistent with expectation. On the
contrary, the Fowkes' value of 25.5 dyn/cm and the Zisman's critical
surface tension of 15 to 23 dyn/cm seem much too low.
III. CONCLUSION
The present analysis clearly shows that the proposed equation
of state method gives accurate results. The surface tensions
obtained by the equation of state method, the harmonic-mean method,
the melt data and the liquid homo logs agree remarkably well with one
another. In contrast, those obtained by the geometric-mean method
and Zisman's critical surface tensions are often too low.
REFERENCES
1. H.W. Fox and W.A. Zisman, J. Colloid Sci. 5, 514 (1950).
4. W.A. Zisman, Advan. Chem. Ser. 43, 1 (1964).
5. S. Wu, !!..:. Polym. Sci. C34,--:r9(1971).
6. S. Wu, J. Adhesion 5, 39 (1973); also in "Recent Advances in
Adhesion", L. If. Lee, ed., pp. 45-61, Gordon and Breach,
New York, 1973.
7. S. Wu, J. Macromol. Sci. C10, 1 (1974).
8. S. Wu,in"Polymer Blends"~Vol. 1, D.R. Paul and S. Newman,
eds., pp. 243-293, Academic Press, New York, 1978.
9. S. Wu and K.J. Brzozowski, J. Colloid Interface Sci. 37, 686
(1971). - -- -
10. S. Wu, !!..:. Phys. Chem. 74, 632 (1970).
11. D.K. Owens and R.C. Wendt, !!..:. !EEl:. Polym. Sci. ..u, 1741
( 1969) •
12. D. H. Kaelble, "Physical Chemistry of Adhesion", Wiley-Inter-
science, New York, 1971.
13. D. Patterson and A.K. Rastogi, !!..:. Phys. Chem. 74, 1067 (1970).
14. K.S. Siow and D. Patterson, Macromol. ~, 26 (1971).
15. R.J. Roe, Proc. Nat. Acad. Sci. U.S. 56, 819 (1966).
16. E. Helfand and A~Sapse, J. Chem. Phys. 62, 1327 (1975).
17. H.W. Kammer, .!:. Phys. Chem. Leipzig 258, 1149 (1977).
18. C.W. Stewart and C.A. von Frankenburg, ~ Polym. Sci. A-2, ~,
1686 (1968).
19. H.W. Kammer, .!:. Phys. Chem. Leipzig 255, 607 (1974).
20. R.J. Good and E. Elbing, Ind. Eng. Chem. 62, 54 (1970).
21. J.R. Dann, J. Colloid Interface Sci. 32, 302 (1970).
22. F. Schulmanand W.A. Zisman, ~ Co"iIoId Sci. 1, 465 (1952).
23. R.E. Johnson, Jr. and R.H. Dettre, J. Colloid Interface Sci.
21,610 (1966). - --
24. E.G. Shafrin and W.A. Zisman, ~ Phys. Chem. 11, 1309 (1967).
25. F.M. Fowkes, in "Chemistry and Physics of Interfaces", pp. 1-
12, American Chemical Society, Washington, D.C., 1965.
26. P.C. Heimenz, "Principles of Colloid and Surface Chemistry",
Marcel Dekker, New York, 1977.
27. R.J. Good, in "Treatise on Adhesion and Adhesives", Vol. 1,
R.L. Patrick, ed., pp. 9-68, Marcel Dekker, New York,
1967.
28. D.G. LeGrand and G.L. Gaines, Jr., ~ Colloid Interface Sci.
31, 162 (1969).
29. S. Wu:-~ Colloid Interface Sci. 11, 153 (1969).
30. R.H. Dettre and R.E. Johnson, Jr., ~ Phys. Chem. 11, 1529
(1967); ~ Colloid Interface Sci. 11, 568 (1969).
31. J.F. Padday, Proc. 2nd International Congo Surface Activity 1,
136 (1957),
A Role of Molecular Forces in Adhesive
Interactions of Polymers
V.A. Belyi, V.A. Smurugov and A.I. Sviridyonok
Institute of Mechanics of Metal-Polymer Systems of
Byelorussian Academy of Sciences, Gomel, U.S.S.R.
ABSTRACT
The present paper discusses a possibility of developing new

ways of analytical and experimental study of molecular interactions
within polymer/polymer and metal/polymer contacts starting from the
general theory of molecular van der Waals forces. We consider that
polymers belong to a class of materials possessing a distinct
selective absorption of radiation. Formulas are suggested for
calculation of forces of interaction between solid surfaces having
discrete absorption spectra. It is shown that the intensity of
molecular interaction can be estimated by the number, intensity,
and mutual overlapping of absorption bands in the spectra of
materials in contact. A method of estimation of the contribution of
certain spectral ranges is suggested and an analysis is given of the
role of infrared region. Experimental and calculated data are
presented indicating the future of the research method chosen.
INTRODUCTION
Up till now there is no common opinion about the role of

molecular forces in adhesive interaction of polymeric bodies (1-
12). The need in a developed adhesion theory stimulates the study
of this problem.
It is of interest to analyze the electromagnetic theory of

molecular interactions betwen solid surfaces based on notions about
fluctuating electromagnetic fields always present in any body and
beyond it (13-16). From this theory it follows that the character
and intensity of molecular attraction are determined by the di-
67
68 V. A. BELYI ET AL.
electric properties of materials. Herein, depending on the corre-

lation between the clearance lI., between the surfaces, and major
wavelengths hO in absorption spectra of materials, the calculating
formulas include either dielectric permeabilities of imaginary
frequencies e:, (i 0, ~(H':) or their values at f,; = 0, Le., static
dielectric permeabilities of materials e:10' e:20' The dielectric
permeability e:(if,;) is related to the imaginary part of the complex
dielectric permeability of the real component of the frequency
through Kramers-Kronig relation (17):
e:(if,;) = 1 + .£ j [e:"(W)W ] dw (1)

1T 0 w2 + f,;2
where Wis frequency (Hz).
Difficulties met on application of the theory to solution of

practical problems are usually connected with the necessity of
accurate knowledge of the function of clearance between surfaces
and with uncertainty of e:" (w) for most materials, i.e., their major
macroscopic characteristic, that determines the intensity of the
interaction. It is important that the dielectric permeabilities
included into the expression of force are integral values of
function e:" Cw) related to the nature of materials, their chemical
composition, molecular structure, etc.
As applied to polymeric materials, one might say that their

absorption spectra look like bands of one or another width where
regions of transparency of the material exist along with absorption
regions of electromagnetic fields. Function e:"(w) for them can be
represented as in Fig. 1, if negligibly small absorption be neglect-
ed in the regions of transparency and if its value is taken as
constant in the absorption band. The fact that e:" (w) is zero on
certain areas allows expression (1) to pass from integration over an
infinite region to a sum of integrals with finite limits that bound
-·T ~(:n~~J
% %
~[+~(~~j dw :n:~2~)
the absorption regions:
.tit) = +
w1 w2 wn
The constancy of function e:" (w) within each interval of integration

gives an opportunity to put it beyond the integral sign; then
integration over the part (W 1TW 1+llw,) , for instance, gives
MOLECULAR FORCES IN ADHESIVE 69
which with relatively narrow absorption bands (6w is small)

approximates
W, 6w,
w,2 + ~
2
Herein, expression (1) has the form:
e:(i~) =1 + iT
2 ~
L.J e:
n
II (Wn) ~6wn2
wn
Wn 2
+ ~
J (2)
where wn ' frequency of the beginning of nth-absorption band (Hz);

6w n ' width of the absorption band; e:" (wn ), value of e:" (w) within
nth-band; n, the number of absorption bands in the spectrum of a
material.
Thus, there is a possibility to estimate the contribution of an

individual absorption band, or individual spectral region to di-
electric permeability of a material and, accordingly, to the inter-
action force. Herein, most vivid is the role of peculiar properties
,
£"(w~
€.(~)
t
Fig. 1. A schematic of function e:" (w) for polymeric

materials.
70 v. A. BELYI ET AL.
of materials associated with chemical composition and molecular
structure that determine the intensity and position of absorption
bands on the frequency axis. As applied to polymeric materials, of
greatest interest is the study of absorption spectra in the infrared
region (2 to 2~ m) where exist major absorption bands associated
with excitation of molecular vibrations allowing an estimation of
the contribution of this frequency range.
EXPERIMENTAL RESULTS
To check these suppositions, the following polymeric materials
were selected: polytetrafluoroethylene (PTFE), polyethylene (PE),
polypropylene (PP), polymethyl methacrylate (PMMA) , polycaproamide
(PCA) , polyethyleneterephthalate (PETP), for which molecular and
supermolecular structures, properties, and the shape of absorption
spectra, accordingly, differ considerably. Absorption spectra of
films of different thicknesses prepared from the above polymers
were obtained with a spectrophotometer UR-20. The values of
absorption indices X were calculated according to known techniques
for each absorption band, and widths 6w of those bands determined.
Values of e: " were calculated from expression e:" = 2n X (18), where n,
refractive index of a material in the optical region; its anomaly
within the absorption band has been neglected.
Based on the data obtained when processing absorption spectra

were estimated values of e:(if';) at f,; = O. These values are given in
Table 1. Table 1 shows that values of e:(if,;) obtained in such a way
for certain polymers slightly approach the values of e:, which
indicates a small contribution of IR-regions and a possib~e large
contribution of (6e:) of other spectral regions in this case. At the
same time, the variations in polymers properties (density, melting
temperature) that are determined by intermolecular interactions are
similar to those when going from one polymer to another in values
e:{~) estimated for IR-regions.
Recording of dielectric permeabili ties ff. i 0 for contacting

materials according to expression (2) where index n belongs to body
1, index K to body 2, and considering the statement from Levin and
Rytov (19) about primary contribution of spectral regions to adhe-
sive force when absorption bands overlap allows obtaining calcula-
ting formulas for various types of contacts.
When two polymeric bodies are in contact, the expression for

bonding force F will have the form:
where Wi' frequency of beginning of i-regions of overlapping of

absorption bands (Hz); /),w i' width of this region; e: (wi) and E2 1
Cw • ), are values of E" (w) for the first and second bodies, and R. is
1 '
the distance between the surfaces, respectively, within this fre-
quency interval. When materials are the same, then:
/),ww
CIO( n
E E"
1(W) n n
fi
F::! 8; R,3
J 1T +
n
E"
2
wn+.,t"2
~Oh
(4)
o n E1 ( w ) --'---"--
n
~+f
Table 1. Estimated Contribution of IR-Regions to Dielectric

Permeabilities of Materials
Materials E(i~) Eo /:).E Dens~ty, m.p.

g/cm °c
PE 1.015 2.27 1.255 0.92 115

PP 1.021 2.26 1.239 0.92 170
PMMA 1.048 2.84 1.792 1. 17
PCA 1.157 2.34 1.183 1. 14 220
PETP 1.354 3.70 2.346 1.40 260
PTFE 1.722 2.10 0.378 2.20 327
72 V. A. BELYI ET AL.
Formulas (3) and (4) allow an estimation of the effec t of

various factors on the intensity of molecular interactions and
underline its strong dependence on the number, intensity, and
mutual overlapping of absorption bands in the spectra of the
contacting materials.
For various polymeric materials, the values of sums were

estimated at ~ = 0:
included into Eq. (3). For polymeric materials in contact with

PMMA, these values were: for PP = 0.0245; PE = 0.0450; PCA = 0.138;
PETP = 1.06. These values and the sums ~, ~ calculated for each
material using its spectra were put intonEq~ (3) to obtain those
values of sub integral expression (A) which character of variations,
when going from one pair of materials to another, would determine
the variations in the force of molecular bonding, Table 2. Similar
calculations were carried out for combinations of various polymeric
materials.
Experimentally the force of molecular bonds of contacting
surfaces was estimated through the friction force of thin polymeric
films rubbed against PMMA and PTFE rollers having roughness
R = 0.25 to O. 2]Jm with the contact angle 1T rad. The tension of the
f!lm was 25N/m2 ; slidi~ speed 0.075 m/s; the temperature in the
friction zone was only 5 higher than that of the environment.
Table 2. Variations in the Force of Molecular Interactions

When Polymeric Materials Are in Contact
Materials A Friction Materials A Friction

of friction force, of fric- force,
pairs 10-2N tion on 10-2N
pairs
PMMA - PP 0.00245 4.31 PTFE - PE 0.0012 4.28

PMMA - PE 0.0045 4.45 PTFE - PP 0.0038 4.13
PMMA - PCA 0.0138 5.00 PTFE - PCA 0.21 4.62
PMMA - PETP 0.082 5.49 PTFE - PETP 0.39 6.98
Table 2 shows that the variations in molecular interactions

which were studied using absorption spectra according to the
technique described, coincide qualitatively (under the chosen
experimental conditions preventing strong chemical interactions
between surfaces, their melting, seizure, etc.) with the mode of
variations in the friction force. A lack of proportionality between
the calculated values of A and the friction measured, can probably
be explained by that during calculations not the whole spectral
region was considered, but only the infrared one. Table 1 gives a
more clear picture about the extent of its contribution.
The results of the calculation of bonding forces given in Table

3 for polymeric surfaces separated by a clearance 0.1~m are
comparable, in the order of value, with those available in the
literature and with experimental data (3,15).
From the view of the theory considered the molecular bonding

between contacting surfaces might be improved by enriching the
spectral composition of fluctuating fields by creating, for
instance, new types of bonds in materials, new functional groups,
etc. Thus, _'fen PE films were gamma-irradiated there was a band
1690-1750 cm in the absorption spectra which intensity increased
with the dose. The study of friction of irradiated films rubbing
against steel under conditions excluding large losses for mechani-
Table 3. Values of Forces of Molecular Interactions

Between Surfaces from Different Estimations
(Clearance R. = O. 1~m)
Materials F, N/m2 Remarks

PTFE - PETP -84 10-5 From formula (3)
PMMA - PE -13 10-5 From formula (3)
Quartz plates - 2 10- 5 From Lifshits' formula (15)
Quartz plates -(2-9)

. 10- 5 From Deryagin and Abrico-
sova (15)
Metallic plates _ 1.3 Values of absorption indi-
ces determining §" (w) are
approximately 10 times
higher for metals than
for polymers
74 v. A. BELYI ET AL.
3r-----------r---------~r_--------~120
z
'"o
-i
I
><
11/
2~--------~~~~~--_1----------~80 ~
~ .~. .. c::
o
.....
o
~----------r_--~~~~~--------~40 . Q.
o
1/1
..Q
«
~ ________ ~ __________ ~ ________ ~O
o 8 16 24
Irradiation dose, Mrad
Fig. 2. Variations in friction force and absorption

index within 1690 - 1750 cm- 1 after rubbing
of gamma-irradiated PE films against steel.
2
cal deformation and tribochemical reactions at P = 10 N/m,
V = 0.075 mis, showed (11) that the presence of an additional
component in the spectrum of fluctuating fields caused an increase
in the friction force; herein, it grows with increasing intensity of
the absorption bands (Figure 2). These observations are in good
agreement with those available in the literature when the normal
adhesion (9) and friction force (20) increased with the oxidation of
the polymer.
CONCLUSIONS
A satisfactory qualitative conformation between calculated and
experimental data obtained by the present authors provides future
applications of the trend chosen for the study of the role of
molecular forces in the adhesion interactions of polymers.
REFERENCES
1. B.V. Deryagin, What is Friction?, AN SSSR, Moscow (1963).

2. B.V. Deryagin, N.A. Krotova and V .~Smilga, Adhesion of
Solids, Nauka, Moscow (1973).
3. A.S. Akhmatov, Molecular Physics of Boundary Friction. Fiz-
matgiz, Moscow (1963).
4. V.A. Belyi, A.I. Sviridyonok, M.I. Petrokovets and V.G. Sav-
kin, Friction and Wear of Polymer-Based Materials, Nauka
i Tekhnika, Minsk (1976).
5. F.P. Bowden and D. Tabor, Friction and Lubrication of Solids

(Russian translation), Moscow (1968).
6. A.A. Berlin and V.E. Basin, Fundamentals of Polymer Adhesion,
Khimia, Moscow (1969).
7. S.B. Ainbinder and A. Ya. Loginova, "Adhesive Interaction Be-
tween Metal-Polymer and Polymer-Polymer Pairs", Mekh.
Polim. 5 (1978).
8. E.V. Teodorovich, "Contribution of Macroscopic van der Waals
Interactions to Friction Force", Izv. VUZov, Fizika, No.
11 (1976). ----
9. V.A. Belyi, N.l. Egorenkov and Yu. M. Pleskachevskii, Adhesion
of Metals to Polymers, Nauka i Tekhnika, Minsk (1971).
10. L.H. Lee, "Effect of Surface Energetics on Polymers Friction
and Wear", in Advances in Polymer Friction and Wear,
Plenum Press, New York 5A (1974).
11. V.A. Belyi, A. I . Sviridyonok and V. A. Smurugov, "Molecular
Changes in Surface Layers of Polymer Materials During
Rubbing", in Kurzfassungen der Berichte, Eurotrieb-77,
Dusseldorf 2/3 (1977).
12. V.A. Belyi, V.A~murugov and A. I. Sviridyonok, "Molecular
Interaction of Polymers Within Region of Friction Con-
tact", DAN SSSR 241, No.3 (1978).
13. E.M. LifshitS;-"Theory of Molecular Attraction Forces Between
Solids", ZhETF 29, No. 1 (1965).
14. I.E. Dzyaloshinskii:- E.M. Lifshits and L.P. Pitaevskii,
"General Theory of van der Waals Forces", UFN, No. 3
(1961) .
15. B.V. Deryagin, 1.1. Abricosova and E.M. Lifshits, "Molecular
Attraction of Condensed Bodies", UFN, No.3 (1958).
16. Yu. S.Barash and V.L. Ginsburg, "Electromagnetic Fluctuations
within a Substance and Molecular (van der Waals) Forces
Between Bodies", UFN, No. 1 (1975).
17. L.D. Landau and E.M. Lifshits, Electrodynamics of Continuums,
Nauka, Moscow (1959).
18. N.A. Borisevich, V.G. Vereshchagin and M.A. Validov, Infrared
Filters, Nauka i Tekhnika, Minsk (1971).
19. M.A. Levin and S.M. Rytov, Theory of Equilibrium Heat Fluctua-
tions in Electrodynamics, Nauka, Moscow (1967).
20. LA. Gribova, A.V. Vinogradov, S.A. Pavlov, A.N. Chumaevskii,
L.V. Dubrovina, A.P. Krasnov, V.A. Khomutov and V.V.
Korshak, "A Study of Physico-Chemical Processes During
Rubbing of Antifrictional Materials", Mekh. Polim., No.4
(1974). --
21. D. Tabor, "Surface Forces and Surface Interactions", ~ ColI.
Interface Sci. 58, No. 1 (1957).
Discussion
On the Paper by T. Hata and T. Kasemura
L.H. Lee (Xerox Corporation): I would like to take this

opportunity to thank Professor Hata for this excellent presenta-
tion. Most of the data presented here have not been published in
the western journals. This review serves to bring all of us up to
date about various aspects of surface and interfacial tensions of
polymer solids.
A. Silberberg (Weizmann Institute, Israel): Have you tried to
separate the surface tension into entropic and ethalpic parts? The
approach through the particular assigning to each group on equal
appearance opportunity at the surface implies a truly random occu-
pation of the surface. This is probably true for very short and
very long chains but might break down with very anisotropic mole-
cules and with copolymers of certain types. It would be good to
have some information in each case to what extent one might be
getting some ordering of the contact layer.
T. Hata (Gunma University, Japan): We have measured the
temperature dependence of surface tension for all samples, obtain-
ing surface entropy SS and surface energy (enthalpy) ES according to
the relation
s s
y =E - S T.
However, I did not present here these data and did not analyze the
change of y by these factors, because of the limited time for the
presentation. Generally speaking, when I said "selective adsorp-
tion" in the paper, it was always accompanied by the decrease of
surface entropy as well as surface energy. The change of them was
quite drastic especially for block copolymers similarly as shown in
y. It was noteworthy, however, that, for homologous series of the
similar type X-Rn-X' the molecular weight dependence of y is well
represented by the relation derived from the Sugden I s equation
n
78 DISCUSSION
based on the additivity of parachor over the whole range of molecu-

lar weight.
On the Paper by F.M. Fowkes
A. Jonath (Lockheed Palo Alto Research Laboratory): What is

temperature dependence of acid-base interaction as compared to
dipole-dipole interaction?
F.M. Fowkes (Lehigh University): The temperature-dependence

of any interaction is measured by its enthalpy (~H). Since acid-
base interactions appear to be far stronger than dipole interac-
tions, they should have a greater temperature dependence.
A. Jonath: What is electric field dependence, if measured, of

the acid-base interaction?
F .M. Fowkes: Acid-base interactions at interfaces always

involve a local electric field on the order of 10 5 volts/cm with
the base side positive and the acid side negative. An applied field
of opposite sign and greater field strength might weaken the bond,
but no such experiments have been tried.
F. Leo (De Soto Inc.): Will solvent-polymer combination with

the most interaction (acid-base) have the highest viscosity?
F .M. Fowkes: Two interactions control polymer swelling and

viscosity; dispersion free interactions and acid-base interactions.
Maximum viscosity is observed with good match of dispersion forces
(od) plus strong acid-base interaction.
F. Leo: Keeping in mind, "OSHA" and Air Pollution Regulations,

do you have any information on acid-base interactions of ketones
(MEK, MIBK, cyclohexanone), esters (butyl acetate, cellosolve ace-
tate, ethyl acetate), commonly used in preparing the polyester-
poly isocyanate coatings?
F.M. Fowkes: The ketones and esters are all bases; the esters
a bit stronger than the ketones. I suppose that these solvents
could bond to isocyanate or urethane groups or to any free acid
groups. These solvents should not be so basic as to interfere with
bonding of the basic polymer to acidic surfaces such as silica,
iron, etc.
F. Leo: Also what effect does structural changes in polyester
or poly isocyanate have on acid-base interaction with surfaces (i.e.
for use as adhesives)?
DISCUSSION 79
F.M. Fowkes: Structure can sometimes be important in shield-

ing acidic or basic groups from each other. A well-shield group
cannot interact with solvents or surfaces.
F. Leo: What is R-NCO before it becomes a urethane (while
still -NCO), acidic or basic?
F.M. Fowkes: I have not measured the acidity or basicity of

isocyanates, but would expect the oxygen to be an electron donor and
the adjacent carbon to be an electron-acceptor, thus providing
intermolecular association somewhat like a hydrogen-bond.
F. Leo: For a polyester with a finite acid number, at what
points does this basic polyester become acidic with increasing acid
number? Also if the acid is an aromatic acid (e.g. isophthalic aci~
instead of an aliphatic acid (e.g. adipic acids), will this change
over from base to acid occur sooner?
F.M. Fowkes: A polyester will be basic, and with increasing

acid number, the basicity will tend to remain and the polyester will
have both acidic and basic properties. The resulting internal acid-
base interaction could compete effectively against other acids or
bases if they are weaker. Thus, acid groups in a polyester could
preferentially bond to the ester groups and minimize interaction
with weakly acidic solvents (like CH 2CL2 ) or with weakly acidic
substrates such as silica. Stronger acids such as isophthalic
should be more effective than weaker acids such as adipic.
D. W. Dwight (Virginia Polytechnic Institute): What is the

basic significance of the Drago constants C and E?
F.M. Fowkes: The modern theory of Hard and Soft Lewis Acids
and Bases (HSAB) shows that two properties are important for bases
(electron-donor "strength" and polarizability of donor site) and
two for acids (electron acceptor strength and polarizability of
acceptor site). The "C" constants vary mostly with polarizability
and the "E" constants vary mostly with inner shell electron binding
energies.
P. Datta (RCA Laboratories): The copolymers of acrylic acid

and ethylene are known to form domain structure of acrylic acid and
ethylene. Were the surfaces of your copolymers analyzed with ESCA
or other spectroscopic technique?
F.M. Fowkes: No.
P. Datta: How surface energy and contact angle measurements

are affected by domain structures of the copolymer and acrylic acid?
80 DISCUSSION
F.M. Fowkes: Note in figures that WAab increases with %acid,

but less than proportionally, suggesting a smaller proportion is
available when acid content is greater.
On the Paper by Souheng Wu
F.M. Fowkes (Lehigh University): I believe Dr. Wu's wetta-

bility data should be considered just a small part of the field of
physical chemistry and therefore not be contradictory of well-
established general principles, one of which is the geometric mean
averaging of intermolecular energies, established by Berthelot, van
Lane, Scatchard, Hildebrand, Leonard-Jones, and London. Dr. Wu has
admitted in earlier papers that his averaging is not realistic in
the light of London equations; what purpose is served by continuing
this approach?
Secondly, Dr. Wu (and others) use the technique of defining Y

c
or Yc(max) to be the surface tension of the solid and then without
any proof assume they have established a truth. I agree that in
~ cases the value of YiB might be zero, but until it is proved,
such an assumption is very speculative. I suggest studying liquids
on mercury where Y12 can be measured and where at least water gives
a finite contact angle.
Thirdly, Dr. Wu suggests that the surface tension of poly-
ethylene can be estimated by extrapolation of values for the liquid
vs. temperature; this is unrealistic for the molar volumes are
different and intermolecular distances are shorter in the solid.
Fourthly, he and many others may not have heard my two papers
hydrocarbons which reduce yt

(not yet in print) on the specific interactions between long chain
to values of several dynes and below
Yt. It is therefore not surprising that YPE d is less than YpE .
I heartily agree that Y values are different with different

liquids on the same solid. Th~t is what I show in my Figs. 1 and 2,
but J.R. Dann made this quite clear in 1970 (see my reference No.
20) •
Souheng Wu (du Pont Company): Firstly, the harmonic-mean

relation for dispersion force interaction has a sound theoretical
basis and is in fact derived from the London theory, as I have
discussed in detail in J. of Adhesion, 5, 39 (1973) and in "Recent
Advances in Adhesion", L. H. Lee, ed., pp-: 45-61, Gordon and Breach,
1973. Both the harmonic-mean and the geometric-mean relations are
special cases of the London theory; each has its range of applica-
DISCUSSION 81
bility and limitation. I have shown previously that the harmonic-

mean relation applies more accurately for polymers than the geo-
metric-mean relation by testing them with the directly measured
surface and interfacial tensions of polymer liquids and melts, as I
have discussed in detail before and summarized in J. of Adhesion, 5,
39 (1973), in "Recent Advances in Adhesion", L.H. Lee, ed., pp. 45-
61, Gordon and Breach, 1973, in J. Macromol. Sci., ClQ, 1 (1974),
and in "Polymer Blends", Vol. 1, D.R. Paul and S. Newman, eds., pp.
243-293, Academic Press, 1978. Therefore, the harmcnic-mean rela-
tion is theoretically sound, and completely consistent with the
London theory and experimentally realistic. This is exactly the
reason why the harmonic-mean relation is preferred and used. If
these are not good reasons, what are good reasons? There has never
been an admission to the contrary. Please also note that earlier
this morning, Dr. Hata confirmed that the harmonic-mean relation is
better than the geometric-mean relation in his lecture.
Secondly, yc or y c,max has -----

never been defined as the surface
tension of a solid. On the contrary, I have shown that y ory
c c max
cannot equal the surface tension in general. I have shown that
y equals the surface tension both theoretically and experi-
c,cb ,max
mentally in the present paper. y LS cannot be zero in general. This
has been known for many years and by many people. The fact that YLS
cannot generally be zero is exactly the crux of the problem, and is
exactly what the present paper is about. I have measured the
surface and interfacial tensions of many pairs of incompatible
polymer liquids and melts. These data would provide a rigorous test
of my method.
Thirdly, the extrapolation of surface tension from the liquid
state to the amorphous solid state is based on the well-established
corresponding state concept of Guggenheim, Katayama, Prigogine and
others. The differences in molar volumes and intermolecular dis-
tances are accounted for in such an approach. Furthermore, remember
that only amorphous solid surfaces are dealt with here. Amorphous
solids may be regarded to have "liquid-like" structures. Extrapo-
lation from the liquid state is still one of the best methods, if not
the best method, for estimating the surface tension of a solid
having an amorphous surface.
Fourthly, the present approach is to address to the problem of

variable y values. I am pleased to see that Dr. Fowkes "heartily"
c
agrees that y values are variable. Dr. Dann made this quite clear
c
in his earlier paper as I quoted in the text. I believe that the
present equation of state approach successfully handles this
problem.
82 DISCUSSION
A. Silberberg (Weizmann Institute, Israel): I would be inter-

ested in knowing how the surface of your polymeric specimens were
prepared. One may expect differences when these surfaces are formed
in air, in contact with metal or with some other forming surface.
Were these surfaces formed by machining, i.e. by mechanical manipu-
lation. Their chemistry might be expected to be different in each
of these cases.
Souheng Wu: All the contact angle data used are taken from the
works by Dr. Zisman and his coworkers, as noted in the text. The
surfaces used were reported to be representative of the materials in
amorphous state free from contamination and scratches.
PART TWO
CHARACTERIZATION OF
ADHESIVE INTERFACES
Lieng-Huang Lee
Wilson Center for Technology
Xerox Corporation
The most important advance during the last several years is the
characterization of surface by various new techniques (1), e. g. ,
ESCA, AES, SIM, etc. With new techniques, we do not have to
speculate on what is at the interface; instead, we can actually see
what is there before and after adhesive bonding. In fact, because
of the new characterization techniques, we may have to modify some
adhesion theories formulated in the past.
In this session, we shall have five papers discussing charac-

terization of adhesive interfaces. My paper will deal with photo-
acoustic spectroscopy, a new technique developed by A. Rosencwaig
(2). This technique can be used to identify polymers, dyes and
other compounds containing chromophores at the interface.
The second paper, by J.S. Solomon and D. Hanlin, is about the

applications of AES (Auger electron spectroscopy) and ESCA (elec-
tron spectroscopy for chemical analysis) to the characterization of
an adhesive-adherend bond joint. Since Mr. Solomon. has not
arrived, I shall read one of my papers previously presented to the
Society of Photographic Scientists and Engineers. This paper is
about electroadhesion, and its title is "A Surface Interaction
Model for Triboelectrification of Toner-Carrier Pair" (3). I am
glad that I did bring extra sets of slides after I heard about the
UAL strike.
• Dr. Solomon finally arrived on Wednesday and his paper was

delivered on that morning to the Fifth Session.
85
86 LlENG-HUANG LEE
The third paper discusses techniques for adhesive vs. adhesive

failure. One of the techniques used was {3-ray back-scattering.
The count rate could be correlated with the fractional area that
failed adhesively and cohesively. This paper, by Dr. T. Smith and
P. Smith, was to be presented by Dr. Smith. Though Dr. Smith works
at Thousand Oaks, California, he, too, could not fly out from Los
Angeles. (When we realized that several scheduled speakers could
not be present, we started to draft participants for "Quickies".
One of our audience, Dr. R. Van der Linden from Belgium, presented a
very interesting "Quickie" on corona discharge on low density
polyethylene. As a result, his paper is now included in this
Proceedgings.)
The fourth paper is related to the locus-of-failure analysis
in adhesive bonding research and will be presented by Dr. David
Dwight. Dr. Dwight will discuss the use of SEM, EDAX (energy
dispersive analysis of X-rays) and ESCA for the failure analysis.
The last paper of this session is about the measurements of dielec-
tric relaxation gradients in an adhesive bond and will be presented
by Dr. John L. Crowley. Following Dr. Crowley's presentation, Dr.
Jonath will present an interesting "Quickie" on "New Approaches to
the Understanding of Adhesive Interface Phenomena and Bond
Strength."
At any rate, we shall make this session as full as if we did
not have the strike.
REFERENCES
1. L.H. Lee, Ed., Characterization of Metal and Polymer Surfaces,
Academic Press, New York (1977).
2. A. Rosencwaig, Phys. Today 28, No.1, 23 (1975).
3. L.H. Lee, Photo. Sci. and Eng. 22, 228, July/August, 1978.
Photoacoustic Spectroscopy for the
Study of Adhesion and Adsorption of
Dyes and Polymers
Lieng-Huang Lee
Webster Research Center, Xerox Corporation
ABSTRACT
Photoacoustic spectroscopy (PAS) has recently been applied for
the determination of the surface and bulk structure of solids. PAS
enables one to obtain spectra, similar to absorption spectra, on any
type of solid materials or composites. One of our studies involves
the examination of adhesion and adsorption of dyes and polymers.
This paper briefly describes basic principles, the instrument and
some of the new techniques developed in our laboratory for the study
of solid structure. The advantages and limitations of this analyti-
cal method will be discussed. Future applications of this technique
for coating, dyeing, printing, and related technologies will be
mentioned.
INTRODUCTION
During the last several years, photoacoustic spectroscopy (1-

5) (PAS) has been employed to study the composition and structure of
solids. Since many solids tend to scatter light and are either
insoluble in a solvent or different in structure in solution, this
technique offers advantages over many existing spectroscopic
methods in elucidating the structure of solid materials. By varying
the modulating frequency, one can study the bulk as well as the
surface. Thus, PAS may be applied to study adhesion and adsorption
of dyes and polymers on surfaces. Some of our preliminary studies
will be discussed in a later section of this paper.
87
88 LIENG-HUANG LEE
PAS as developed is not as versatile as originally claimed (1).

There are limitations such as the saturation problem (6,7). Several
methods have been reported in the literature (7) to alleviate this
problem; however, not all of them are successful. We shall describe
some of the improved methods used by us.
The applications of PAS can be greatly enhanced if the spectrum

can be extended to the infrared. Recently, Kirkbright (8) and Adams
(9) reported the extension of the spectrum to the near IR (2.5
microns and 3.4 microns, respectively). With these extensions, the
applications (8) have been broadened for the characterization of
pigments, blood, fats, thin-film coatings, polymers and lubricants.
New spectrometers with near IR capabilities are now offered by EG&G
Princeton Applied Research (10), Gilford Instruments (11) and EDT
Research (England) (12). It is likely that the use of the laser
(13,14) can bring forth a wider infrared range for future spectro-
meters.
Principles of Photoacoustic Spectroscopy

In the spectrometer, the sample is placed inside a specially
designed closed cell containing air (or other gases) and a sensitive
microphone. The solid is then illuminated with chopped or modulated
monochromatic light. Rosencwaig (15) and Parker (16) theorize that
the primary source of the acoustic signal arises from the periodic
heat flow from the solid to the surrounding gas. Nonradiative de-
excitation processes can convert a part or all of the light absorbed
by the solid into heat. The periodic flow of this heat into the
surrounding gas produces pressure fluctuations in the cell and thus
genera tes sound detected by the microphone. The signal from the
microphone is then recorded as a function of the wavelength of the
incident light.
Although the magnitude of the PAS signal is a function of the
heat evolved or the light absorbed by the samples, there are other
factors (15) involved, e.g., thermal properties of the sample and of
the cell and the chopping frequency. The interrelationship of these
parameters may appear to be somewhat complicated; however, in
general, the PAS spectrum corresponds closely to the optical
absorption spectrum of the sample, provided that the sample is
essentially in the same state in solid as in liquid or solution.
Our work further confirms this relationship.
Three important parameters are specified by the theory (15):

the sample thickness t, the optical absorption length ~b and the
thermal diffusion length ~ s; ~ b = 1.ta, where a
is the optical
2asi
absorption coefficient, and ~s = (w) , where asis the thermal
PHOTOACOUSTIC SPECTROSCOPY 89
diffusivity of the sample, and w is the modulating frequenoy in

radians per seoond. The thermal diffusion length determines how far
a heat wave of frequency w will travel before excessive damping
occurs. The thermal diffusion length increases with the decrease in
modulating frequency. The photoacoustic spectrometer responds to
the total amount of light absorbed in the sample within a depth
roughly corresponding to the thermal diffusion length. A sample may
be considered to be photoacoustically opaque if substantially all
of the light is absorbed within this depth.
The thermal diffusion length of air at 50 Hz is approximately

0.34 mm. The ranges of other photoacoustic parameters (16) are
listed in Table 1.
Table 1. Photoacoustic Parameters
Reflectivity R 4-90~
Absorption coefficient 8 0_10 6 cm- 1

Thermal diffusivity a 10- 3 _ 10 3 cm2/sec
s
Thermal diffusion length, ~s 10 -3 - 1. 0 cm
Thermal damping constant as 1_10 3 cm- 1
Chopping frequency f 10_10 4 Hz
EXPERIMENTAL
Instrument
The instrument is the Model 6001 Photoacoustic Spectrometer

manufactured by EG&G Princeton Applied Research. The optical
system (18) of this instrument is shown in Fig. 1. The light source
is a 1kW xenon arc lamp which is electronically modulated. The
monochromator incorporates a turret of three concave holographic
gratings covering the spectral range of 200 nm to 26 ~m. Relative
aperture is F/4.2. Grating selection is automatic and under
microprocessor control'. Spectral bandwidth of the monochromator is
controlled by the slit selection.
The sample cell is of fused silica. Samples (solid or liquid)

are easily loaded into a rectangular receptacle (4x8x2 mm) which is
inserted into a cuvette.
90 LlENG-HUANG LEF
I kW XENON ARC L AMP IN

F/4 2 ELLIPSOIDAL
REFLECTOR
,,- - ,
- - - - OPTIO NA l- - - - - - - - - -,
, JPLANE MIRROR I
I ,
, EXTERNAL '
, • - - lASER ,
L_ ~ ___ ________ __ ~~A~ ___ :
<:::-,,:~-- PYROEL EC TRIC
REFERENCE
DETECTOR
OFF - AX IS
Elli PSOI OA L
REFLEC TO R
LJ--
TURRET OF FUSED SI UCA
THREE BEAM SPLITTER
GRATINGS
SAMPLE CELL
r==J
Fig. 1. Optical system of the model 6001 photo-

acoustic spectrometer .by EG&G Princeton
Applied Research.
In the electronic system, there are two two-phase lock-in ampli-

fiers which separately process the microphone and pyroelectric
signals. The entire system is under microprocessor control, with
keyboard entry of experimental parameters. A 14-bit analog-digital
converter encodes the outputs of the two lock-in amplifiers and the
output of a follower potentiometer coupled to the sinebar mechanism
of the monochromator. Two 12-bit digital-analog converters serve
to output data to the X-Y recorder.
RESULTS AND DISCUSSION
PA Spectra of Polymers
Several polymers, e.g., polyester (Fig. 2), polycarbonate (Fig.

3), polystyrene (Fig. 4), polymethyl methacrylate (Fig. 5), have
absorption mainly in the UV region (Fig. 6). This absorption range
facilitates the study of adsorption of coloring dyes on the polymer
surfac~ In this paper, we shall illustrate the applications with
Mylar polyester, gelatin, and cellulose fibers (Fig. 7).
PA Spectra of Dyes
Since dyes have strong absorption in the visible region, the PA

spectra of solid dyes nearly all have saturation problems. Satura-
tion (7,15) is closely related to the optical absorption
coefficient, 8 . For a sample with a high optical absorption
coefficient, saturation of the PA signal can occur. For a sample
with a low optical coefficient, e.g., 8<10 3 cm- 1 , the PA spectrum in
the wavelength region will yield the true 0ftic~t absorption spec-
trum. In the wavelength region where 8>10 cm , saturation will
occur and the true optical absorption peak height will not be
observed in the PA spectrum.
SENSITIVITY 20
FREQUENCY 40 Hz
SCAN RtrrE loonm/mln
200 240 400

WAVELENGTH, nm
Fig. 2. Photoacoustic spectrum of mylar film.

92 LlENG-HUANG LEE
SENSITIVITY 100
FREQUENCY 40 Hz
SCAN RATE 100
nmlmin
...J
<l
Z
(!)
in
200 240 280 320

WAVELENGTH, nm
Fig. 3. Photoacoustic spectrum of polycarbonate .
SENSITIVITY 20
FREQUENCY 40 Hz
SCAN RATE 100 nm Imln
...J
<l
Z
~
VJ
200 240 320 360 400

WAVELENGTH, nm
Fig. 4. Photoacoustic spectrum of polystyrene .

SENSITIVITY 50
FREQUENCY 40 Hz
SCAN RATE IOOnm/mln
200 400
WAVELENGTH, nm
Fig. 5. Photoacoustic spectrum of polymethyl

methacrylate.
To alleviate the saturation problem, one can apply several

sample preparation techniques.
1. If the solid is soluble in a solvent, a liquid or solution
(6,13,19) can be used instead of the solid.
2. The solution, e.g. methyl violet 2B, can be applied as a
drop on a cellulosic paper (Fig. 8), or on a TLC plate.
3. The solution can be solidified into a gel, e.g . , potas-
sium dichromate, K2Cr201 , in gelatin (Fig. 9).
4. The solid can be diluted with a nonabsorbing powder,
e.g., MgO (8). The mixture, e.g., a-metal free phthalo-
cyanine and MgO, is co-ground (Fig. 10).
Since PA spectra may contain results due to both absorption and

reflection loss, one should be careful in interpreting the spectra.
94 LlENG-HUANG LEE
SENSITIVITY 50
FREQUENCY 40 Hz
SCAN RATE 100 nm/mln
200 240 280 320 360 400

WAVELENGTH, nm
Fig. 6. Photoacoustic spectrum of cellulose.
I. POLYCARBONATE
FINE POWDER
2. POLYESTER FILM
(MYLAR)
I ,
3. CELLULOSE PAPER
i ,
! (3)!
j!v-...../ 1
I "", ,
. ".
I! "'l. ,
I. \\. " ..... __
., - ...............
........................... ........ _._._. .........
200 250 300 350 400 450 500
WAVELENGTH, nm
Fig. 7. Photoacoustic spectra of polymers

I. ON FILTER PAPER
2. IN METHANOL SOLUTION
SEN.: 10
SCAN RATE: IOOrmlmln
FREQUENCY :40Hz
200 400 500 600 700

WAVELENGTH, nm
Fig. 8. Photoacoustic spectra of methyl violet 2B

on filter paper and in methanol solution.
Correlation Between UV-Visible and PAS Absorption Peaks of Dyes

The absorption spectra of dye solutions in the UV-visible
range and the PA absorption signals have been carefully compared.
Although there are some exceptions, the general agreement between
the two types of spectra is very good. The results are shown in
Table 2. There are some minor shifts of absorption peaks, but the
characterization with PA spectra appears to be well substantiated.
In Table 2, the PA absorption peaks of those dyes absorbed on paper
agree, in general, with those obtained directly from dye solutions.
This result suggests that PAS can be properly used for the study of
adsorption of dyes or polymers.
Adsorption of Dyes on Polyester
The adsorption of dye on a polyester surface was used to

illustrate the general application of PAS for the adsorption study.
96 LlENG·HUANGLEE
I. KZ CrZ C>-r Aq. SoI'n. (5%), SEN.:

200
Z. KZ CrZ C>-r IN CURED GEl..ATlN
(2%), SEN.: 500
3. GELATIN (BLANK), SEN.: 200
ALL AT FREQUENCY OF 40Hz,
AND A SCAN RATE OF 50 nmlmin
1~3)
\ ,
~--
200 300 400 500 600 700 800
WAVELENGTH, nm
Fig. 9. Photoacoustic spectra of potassium dichromate

in solution and in gelatin.
I. CO-GROUND WITH MOO

(I : 100), SEN. : 50
2. MIXED WITH MOO
(1 : 100), SEN.: 100
ALL AT FREQUENCY OF
40 Hz AND SCAN RATE
OF 100nm/min
...
200 400 600 800 1000 1200 1400
WAVELENGTH, nm
Fig. 10. Photoacoustic spectra of alpha metal-free

phthalocyanine in magnesium oxide.
(2)_/\
..
I)
! ,
.I . ,
I .
I. MYLAR POlYESTER
(0.5 mil), SEN.: 200
2.MYLAR IN RHODAMINE B
. .!~' SOLUTION (I x 10-3 MIL)
(3)
'H
.,' \.I,
,
...J FOR I HOUR. SEN.: 100
c{
z 3. MYLAR IN RHODAMINE B
£! I I.
SOLUTION (I It 10-4 MIL),
'1
C/) I I
I \
• ,I I SEN.: 100
I' I.
~H~T FREQUENCY OF
i ~\'
1 ~
'~ ...\
.--~,
200 300 400 500 600 700 800 900

WAVELENGTH. nm
Fig. 11. Photoacoustic spectra of adsorbed rhodamine

B on mylar.
A thin Mylar polyester film (12.5 x 1x5 cm) was_~laced in each

of the following rhodamine B solutions: (a) 1x10 mole/t, (b)
1x10 -4 mole/t, (c) 1x10 -5 mole/to The adsorption took place at room
temperature for one hour. The adsorbed films were dried and then
used for the PA analysis. The films from all three solutions showed
the absorption peak of the adsorbed dye, rhodamine B at 555 nm (Fig.
11). The advantage of PAS in this case is the examination of the
adsorbed species without any separation of the substrate.
PAS For Multilayer Films
Multilayer films can be formed in nature, e.g., the green

leaves, or by lamination or adhesion. The examination of the
composition of various layers without any destructive treatment is
one of the advantages of the technique. Examination of the protec-
tive layer of leaves with PAS reveals the waxy material synthesized
by the plant to protect the interior photosynthetic structure.
;g
Table 2. Comparison of UV-Visible Absorption Peaks with PAS Signals of Dyes
Dyes UV-Visible Absorption PAS Signal PAS Signal

Peaks, (in solution) (in solution) (on filter paper)
(nm) (nm) (nm)
Alkali blue, 4B 586 592 595

Aniline blue, SS 582 590 599
Anthraquinone violet R 548-550 536 554
Azure A 637 617 603
Biebrich scarlet (K20 sol.) 506 509 512
Crystal violet 593 589 555-595
Cyanine acid blue 572 580 596
4',5',-Dibromofluorescein 520 524 526
Eosin B 515 522 530
Indulin (SS) 595 598 600
Methyl green 632 630-645 645 cm
Methyl red 423-519 (dep. on pH) 515 483 z
C>
Rhodamine B (276,315)548,556 546 560 :I:
c
»z
C>
....
m
m
I. FREQUENCY - 10Hz. SEN.: 20

2. FREQUENCY - 40Hz, SEN. : 20
3. FREQUENCY - 500Hz, SEN.: 500
ALL AT SCAN RATE OF 5Onm/mln.
200 300 400 500 600

WAVELENGTH, nm
Fig. 12. Photoacoustic spectra of laminated paper

with Rhodamine B at various chopping
frequencies.
The examination of the solid surface is achieved by the

increase in the modulating frequency or the decrease in the thermal
diffusion length, as discussed in one of the earlier sections. The
same technique is used to reveal the surface layer of a laminated
film. A polyester film (12.511) was laminated on a filter paper
coated with a drop of rhodamine B dye. By increasing the modulating
frequency, the absorption peak of the underlayer dye decreased, as
shown in Fig. 12. In other words, the thermal diffusion length is
shortened and less dye in the bulk is detected by PAS. In practice,
this technique in changing the modulating frequency is not as
straightforward as it sounds. The results should be interpreted
with caution.
100 LlENG·HUANG LEE
Future Applications of PAS to Adhesion and Adsorption
In this paper, we briefly demonstrate the applications of PAS

to adsorption and adhesion studies, and the limitations of the
technique, e.g., the saturation problem. With a further broadening
of PAS into the IR range (13,14) the applications can be expanded
for the identification of organic functional groups. We can
visualize future applications of PAS in the following areas:
1. Adsorption and adhesion of biopolymers on various sub-

strates.
2. Adhesion of coloring materials on polymers or other sub-
strates.
3. Identification of inks, dyes and pigments in a polymer
matrix.
4. Photoadhesion phenomena.
5. Photoadsorption and photodesorption phenomena.
Technologically, PAS will find uses in the characterization of

coatings, inks, and dyes for textiles.
CONCLUSIONS
Recent developments in photoacoustic spectroscopy for solid

materials indicate new applications for the study of adhesion and
adsorption of polymers and biopolymers. Though there are still
problems associated with this technique, progress has been made,
especially in regard to the saturation problems. The comparison
between the optical absorption spectra and PA signals confirms the
general agreement. With the current state of the art, PAS can be
qualitatively used to characterize dyes or polymers as isolated or
as adsorbed species. However, because of the nature of the' PA
effect, the interpretation of spectra should be exercised with
caution.
ACKNOWLEDGEMENTS
The author would like to thank Mr. H. Reichard of EG&G Prince-

ton Applied Research for discussions about the techniques in using
the Instrument, Dr. A. Rosencwaig of the Lawrence Livermore Labora-
tory and Dr. R. Staley of the Massachusetts Institute of Technology
for the comments on the manuscript, and P. Alden for technical
assistance.
REFERENCES
1. A. Rosencwaig, "Photoacoustic Spectroscopy of Solids", Phys.

Today 28, No.1, 23 (1975).
2. A. Rosencwaig, "Photoacoustic Spectroscopy - A New Tool for
Investigation of Solids", Ana. Chem. 47, No.6, 592A, May
1976. -----
3. R. Harshbarger and M.B. Robin, "The Opto-acoustic Effect.
Revival of an Old Technique for Molecular Spectroscopy",
Accounts of Chem. Res. 6, 329 (1973).
4. A. Rosencwaig,-"Solid State- Photoacoustic Spectroscopy", p.
193 in Optoacoustic Spectroscopy and Detection, Ed. by
Y.H. Pao, Academic Press (1977).
5. F.W. Karasek, "Photoacoustic Spectroscopy", Res./Dev., p. 33,
September 1977.
6. J.F. McClelland and R.N. Kniseley, "Signal Saturation Effects
in Photoacoustic Spectroscopy with Applicability to Solid
and Liquid Samples", ~ Phys. Lett. 28, No.8, 467
( 1976) .
7. R.B. Somoano, "Photoacoustic Spectroscopy of Condensed Mat-
ter", Angew. Chem. Int. Ed. ~ lI, 238 (1978).
8. G.F. Kirkbright, "Optoacoustic Spectrometry and Its Applica-
tion to Examination of Solid and Liquid Samples", paper
presented to the Photoacoustic Spectroscopy Symposium,
Boston, October 1978.
9. M.J. Adams, "Optoacoustic Spectrometry of Solid Samples in the
Near Infrared", paper presented to the Photoacoustic
Spectroscopy Symposium, Boston, October 1978.
10. H.S. Reichard, "A Microprocessor-Based Photoacoustic Spectro-
meter", paper presented to the Photoacoustic Spectroscopy
Symposium, Boston, October 1978.
11 . R.E. Blank and T .D. Wakefield II, "A Newly Developed Photo-
acoustic Spectrometer with Selected Applications", paper
presented to the Photoacoustic Spectroscopy Symposium,
Boston, October 1978.
12. D.J. Johnson and G.F. Kirkbright, "A New UV-Visible and Near-
IR Optoacoustic Spectrometer for Solids and Liquids",
paper presented to the Photoacoustic Spectroscopy Sympo-
sium, October 1978.
13. P.E. Nordal and S.O. Kanstad, "Photoacoustic Spectroscopy on
Ammonium Sulfate and Glucose Powders and Their Aqueous
Solutions Using a CO 2 Laser", .Qpb. Commun. 22, 185
(1977 ).
14. M. J • D. Low and G. A. Parodi, "Infrared Photoacoustic Spectro-
scopy of Surface Species", paper ~esented to the Spring
Meeting of the American Chemical Society, April 1979.
15. A. Rosencwaig and A. Gersho, "Theory of the Photoacoustic
Effect with Solids", !!..:. ~ Phys. 47, 64 (1976).
16. J.G. Parker, "Optical Absorption in Glass: Investigation
Using an Acoustic Technique", ~.Qpb..1£, 2974 (1973).
102 LIENG·HUANG LEE
17. J • F. McClelland and R. N. Kniseley, "Photoacoustic Spectroscopy

with Condensed Samples", ~ Opt . .!2, 1658 (1976).
18. Technical Bulletin, "The Model 6001 Photoacoustic Spectro-
meter", EG&G Princeton Applied Research (1978).
19. J •W. P. Lin and L. P. Dudek, "Photoacoustic Spectroscopy. I.
Signal Saturation Effect and Techniques for Obtaining
Good Solid Spectra", paper submitted to !!.:. Anal. Chem.
Adhesive-Adherend Bond Joint
Characterization by Auger Electron
Spectroscopy and Photoelectron
Spectroscopy
J.S. Solomon and D. Hanlin
University of Dayton Research Institute
300 College Park Avenue, Dayton, Ohio 45469
and
N.T. McDevitt
Air Force Materials Laboratory (MBM)
Wright-Patterson Air Force Base, Ohio 45433
ABSTRACT
Chemical and physical information about intact adhesive bond

joints is usually inferred from data obtained from each isolated
component (adhesive and adherend) prior to bonding or after bond
failure. A technique is presented in which the chemical state as
well as the elemental distribution can be obtained from intact bond
joints using conventional surface characterization techniques such
as Auger electron and photoelectron spectroscopies. The technique
involves the use of thin film adheren~ adhesively bonded struc-
tures. The thin film adheren$1s (5x10- m) are prepared by vacuum
deposition, which, after various treatments and subsequent bonding,
can be ion beam etched away until the adherend-adhesive interface is
reached for characterization. A number of bonding parameters have
been investigated using this technique, including effects of adher-
end surface treatments (i.e. anodization, protection primers, cor-
rosion inhibitors) and cure conditions (i.e. time, temperature and
pressure).
103
104 J. S. SOLOMON ET AL.
INTRODUCTION
Adhesive bonding offers considerable weight and economic

advantages over conventional fabrication processes. Aircraft and
automotive applications of adhesive bonding have spawned consider-
able research into various related subjects involving a number of
scientific disciplines. Just a few years ago, adhesive bonding was
considered more of an art than a science and the majority of bonding
evaluations were based strictly on empirical results. With the
growing interest in adhesive bonding, the direction of related
research has been shifted somewhat away from the empirical to a more
systematic scientific approach.
Of empirical interest in any potential application of adhe-

sively bonded structures are strength and durability. Conse-
quently, those parameters which affect strength and durability are
at the heart of adhesive bonding research. Included in this
research are the material and physical properties of adhesives, the
bulk and surface characteristics of adherends, and the nature of the
forces involved in bonding. One of the recognized facts about an
adhesive bond is that these parameters are all interrelated and can
undergo changes once a bond joint is formed. Therefore, much
information can be gained if the same quantities (chemical and
physical) could be investigated within the bond joint as well as
from the separate components.
The chemical and physical characterization of the adherend

surface and adhesive can be performed using a number of surface and
energetic characterization techniques. However, these techniques
are generally applied to the characterization of the isolated bond
joint components before bonding or after failure (1-4). The reason
being, the size and location of the bond joint do not fulfill
conventional surface technique requirements.
Normally, when one thinks of an adhesive bond, the bond line is
thought of as a two dimensional "interface" between the adhesive and
adherend. In reality, the bond jOint is more than this. It is a
three-dimensional region which has been called an "interphase" (5).
This region incorporates all the material from some point in the
bulk adhesive toward and through the actual boundary between adhe-
sive and adherend to the point where the local properties approach
those of the bulk adherend. This generally amounts to a cross
section hundreds of nanometers in thickness, located perhaps,
several millimeters or more below the outer surface of one of the
adherends.
One method of assessing the chemical and physical properties

of an adhesive bond is to induce failure within the interphase and
characterize the failed surfaces (2,3). However, as reported by
Dwight (1), real fracture surfaces do not always exhibit a clear
ADHESIVE-ADHEREND BOND JOINT 105
failure mode. When failure is entirely cohesive within the adhe-

sive, surface analysis can be extremely difficult because of
(1) charging effects and (2) uncertainty as to the location of the
interphase with respect to the locus of failure. When mixed mode
failures are encountered, difficulties also arise since both sur-
faces may appear the same, when in fac~, some areas on the surfaces
may only be covered with a very thin carbon containing layer (2).
In the latter case, an ion beam etch would quickly remove the thin
carbon layer and reveal the local mode of failure. This approach
still leaves an uncertainty as to the size and location of the
interphase.
If the interphase volume was at or near the surface, composi-

tion profiling techniques could be employed to examine the bond
joint in an undisturbed state. This would remove some of the
difficulties and uncertainties described above. Auger in-depth
profile analysis of thin films is a well established technique with
sufficient depth resolution to provide composition and elemental
distribution information from thin films (6) and has been used
successfully in the past to provide chemical information from both
adherend and adhesive surfaces (1-9). The technique, however, is
not very practical for examining very thick specimens, such as
intact adhesively bonded structures, because of the relatively slow
etching rate of the ion beam and depth resolution dependence on
thickness (10,11). Consequently, unless one or both adherends of
an adhesively bonded structure are extremely thin (~111m) surface
probing techniques cannot be realistically applied. They can be
applied, however, if vacuum deposited films could be used for the
adherend. Since a relatively thick material is required for
bonding, the substrate on which the metal film is deposited would
serve as a "secondary" adherend. After bonding, the "primary"
adherend would be separated from its supporting substrate and
analyzed by Auger electron spectroscopy (AES) and x-ray photo-
electron spectroscopy (XPS) for a number of related parameters. For
example, Auger in-depth profiles through a bond interphase could
provide information concerning adhesive penetration as a function
of surface preparation, elemental distributions within an inter-
phase, and effects of corrosion inhibiting primers on adhesive
penetration. The state of chemical bonding or changes in chemical
bonding within an interphase could be characterized by XPS.
SPECIMEN PREPARATION
Since there is considerable interest in adhesively bonded

structures of aluminum and aluminum alloys in the aircraft indus-
try, aluminum was chosen as the "primary" adherend for the initial
study. The problem of separating the deposited aluminum from its
supporting substrate after bonding was easily solved.
The presence of fluorine on titanium surfaces has been sug-

gested as the primary cause of adhesive failure (failure at the
adhesive-adherend juncture as compared to cohesive failure which is
failure totally within the adhesive) (12,13). Our previous work
involving surface preparations for adhesive bonding of titanium
confirmed this finding. In addition to the presence of fluorine,
removal of surface oxides almost guaranteed an adhesive failure
mode. Therefore, to enable separation of the bonded thin film
aluminum adherend from its supporting substrate, titanium or 6A1-
4V-Ti were used as supporting substrates. These "secondary" adher-
ends were pretreated with an oxide removing etch containing fluor-
ine (HN0 3/HF) prior to vacuum deposition (14,15). Aluminum films
of approximately 0.5 to 1J.lm were then vacuum deposited onto the
etched substrates.
At present, the surface preparation of aluminum for adhesive

bonding receiving the highest fnterest in the aircraft industry is
anodization (16). Therefore, the deposited aluminum films, to-
gether with the supporting substrates, were anodized at constant
potentials, which could produce anodic oxide films less than 0.4J.lm
(4000l) in thickness. Figure 1 is a scanning electron microscope
(SEM) micrograph at 50,000x magnification of an anodic aluminum
oxide film grown on aluminum at 40 volts for 20 minutes at room
temperature in 1.0M H3P04 . The columnar structure (11-19) of the
porous anodic oxide ~s very evident. Figure 2 contains an SEM
micrograph of the cross section of the anodic oxide film grown on a
vacuum deposited aluminum film which was anodized for 5 minutes at
the conditions stated above. The pore diameters of the anodic oxide
film shown in Figs. 1 and 2 are approximately 400~. Comparing Figs.
1 and 2, no distinguishable differences are evident between the
anodic film grown on bulk aluminum and one grown on the thin film.
This is in agreement with Dell'Oca (20).
SEM micrographs of anodic oxide films grown on aluminum
alloys, such as 2024 aluminum, were also compared to micrographs of
anodic oxides grown on bulk and thin film aluminum. The conclusion
was that the anodic oxides grown on the alloys were structurally
indistinguishable from those grown on bulk or thin film aluminum.
Therefore, the anodized thin film aluminum model adhesively bonded
structures could be used to simulate the bonding of thick anodized
aluminum and aluminum alloy panels.
The anodized aluminum films attached to "secondary" substrates
were bonded to another anodized surface in an autoclave cham-
ber (21). The adhesives used were 3M's AF143 (22), and American
Cyanamid's FM13 and FM400 thermo-setting epoxy adhesive films (23).
A typical adhesively bonded structure is illustrated in Fig. 3.
Figure 4 illustrates the orientation of the bonded structure for ion
beam etching and analysis after the secondary adherend is de-
laminated from the aluminum film.
ADHESIVE-ADHEREND BOND JOINT
Fig. 1. Scanning electron micrographs at 50,000x

magnification of a cross section from an
anodic oxide film formed on aluminum with
1.0M H3P0 4 at 40 volts for 20 minutes.
EXPERIMENTAL
The Auger instrumentation consisted of a Physical Electronics

Industries (PHI) Inc. model 540A Thin Film Analyzer equipped with a
single pass cylindrical mirror analyzer (CMA) with a resolution
l:£/8"-{).6%. The coaxial electron gun was operated with a 4keV
potential at 0.1 to 0.5~ beam current. A peak-to-peak modulation
of 5eV was applied to the analyzer for phase sensitive detection and
the resultant N'(E) data was digitally recorded and computer pro-
cessed as previously reported (24). An electron flood gun biased
at -18v with respect to the specimen was used to minimize possible
charging effects. The ion beam was generated with a PHI model 94-
191 Sputter Ion Gun which was operated with a beam Fotential of 2keV
and an ion current density of approximately l~A/mm. When charging
effects were evident at the onset of analysis, the electron flood
gun and ion gun (at reduced current) were turned on prior to data
collection. Since the region of interest was well below the
aluminum metal surface, any ion etching during this time was not
detrimental to the analysis.
The x-ray photoelectron spectrometer was an AEI model ES-l00
with a magnesium x-ray source. The description of this instrument
Fig. 2. Scanning electron micrographs at 50,000x

magnification of a cross section from an
anodic oxide film formed on a 900 nanometer
thick vacuum deposited aluminum film anodized
in 1.0M H3P0 4 at 40 volts for 5 minutes.
BOND JOINT
- 5)(1 0 -7 m } INTERPHASE
i
"SECONDARY"ADHEREND
Fig. 3. Illustration of the bond joint cross section

of an adhesively bonded anodized thin film
aluminum adherend attached to a supporting
"secondary" adherend.
2 KeV
+
Ar
-7
~5x10 m
Fig. 4. Illustration of the bond joint cross sec-

tion shown in Fig. 3 with "secondary" ad her-
end removed for ion beam etching and subse-
quent AES and XPS analyses.
was previously reported (25).
Vacuum deposition of aluminum was performed in a Varian Asso-

ciates F-12E ultra high vacuum system from a hot tungsten conlcal
filament at a background pressure of approximately 1. 3 x 10- Pa
(1 x 10- 8torr). A Kronos Inc. 301 Film Thickness monitor was used
to measure the thickness of deposited aluminum.
RESULTS AND DISCUSSIONS
Elemental Distribution
Two thin film adhesively bonded structures were prepared using
aluminum films anodized in phosphoric acid and phthalic acid,
resulting in a porous anodic film in the case of phosphoric acid and
a dense nonporous oxide film with the phthalic acid. After bonding
with AF143, the supporting substrates were separated from the
aluminum thin films. AES spectra from the delaminated surfaces in
Fig. 5 show a clean and complete separation. Auger in-depth
elemental profiles through the respective interphases are shown in
Fig. 6. The method to construct Auger in-depth profiles was
previously reported (24).
"S[CONDAPN"AOHERENO
"PRIMARY· AOHERENO
260 .60 sbo 860 ,cioo '2'00 ,~oo GOO ,800

ENERGY (eV)
Fig. 5. Auger electron spectra of delaminated thin

film aluminum "primary" adherend and HNO -HF
etched 6A1-4V-Ti alloy "secondary" adherand
surfaces.
PHTHALIC ACIO AHOO

PHOSPHO"IC ACIO ANOO
UI n
0..
I: 0..
II:
eLl
......
0 (1<' )
I:
II: 0 " )
/:........
D..
fl
.... D..
.... . A '(":'), fl
.... *,.: ••••••
/"
0.. eL. ; " ' . " ' / ' ' " . , , '
0..
..
TIME [MIN) TIME [MIN)
Fig. 6. Auger elemental profiles of AI, 0, and C, as

a function of ion sputter time, through the
interphase regions of bonded phthalic and
phosphoric acid anodized thin film aluminum
adherend structures,
The abscissa in Fig. 6 is a measurement of sputtering time

which can be converted to a depth measurement if sputtering rates
are known. As reported by Kirschner and Etzkorn (26), converting
sputtering time to depth is not a straightforward task when multi-
layers of different elements and chemical states are profiled. The
problem is the sometimes large variance in sputter yields for pure
materials, mixtures, and oxides. Wehner has reported that there is
nearly a factor of two difference between the sputter yield of pure
aluminum and aluminum oxide (27). In the case of a porous anodic
oxide film, there is a thickness dependency on sputter rate because
the pore diameter decreases with thickness (28) and a more dense
barrier oxide layer is present between the pore base and the
metal (19). Sputter rate determination is further complicated
within an interphase because of the mixture of the adhesive with the
oxide and the oxide with the metal. Thus, the anodic oxide layer of
the phosphoric acid anodized aluminum film, as determined by the
oxygen profile in Fig. 6, appears to be 1.5 times thicker than the
phthalic acid anodic oxide, when in fact, the phthalic acid anodic
film is 0.1~m (10001) thick while the phosphoric acid anodic film is
0.45~ (45001) or 4.5 times thicker.
In addition to sputter rate considerations, attempts to relate

thickness to sputtering time can result in very large errors if
depth resolution is ignored. Depth resolution has been defined as
the quality of an in-depth profile compared to an ideal rectangular
signal as determined by the broadening of the trailing edges of the
profile (26,29). Figure 7 contains the smoothed and normalized in-
depth profiles of (A) aluminum, oxygen, and carbon from an adhe-
sively bonded phosphoric acid anodized aluminum film and (B) alum-
inum, oxygen, and titanium from an identical unbonded anodized film
on a 6A1-4V-Ti substrate. Based on predetermined sputtering rates
for anodically grown A1203,the oxide film thickness, as determined
from Fig. 7B is approximately 0.14~ (1400i).
The standard convention for determining depth resolution is to
divide the sputtering time, bt, between the points at which 16% and
84% of a particular dN(E)/dE signal are reached, by the total
sputtering time up to the point at which 50% of the signal is
recorded. The depth resolution, bt/t, based on the oxygen profiles
in Fig. 7, is 31% (7A) and 23% (7B). The latter is in good agreement
with Mathiew and Landolt's value of 22% for an A1/A1 203 interface in
anodic oxide films greater than 4 x 10-8m (4001) thick (11). The
difference in depth resolution between the oxygen profiles in Fig. 7
is most likely due to greater roughness effects produced by sputter-
ing (29) and, to a lesser degree, the change in sputtering rate with
the presence of carbon. The problem of identifying the exact point
of the interface between layers as represented by the profiles, when
profiles are subject in poor depth resolution (26), was easy to
solve in the case of aluminum-aluminum oxide because the AlKLL
bO 10
TIME [MIN] TIME [MIN]

8
A
Fig. 7. Normalized Auger elemental profiles as a
function of ion sputter time of (A) AI, 0,
and C through the interphase of a bonded
phosphoric acid anodized thin film aluminum
adherend and (B) AI, 0, and Ti through an
unbonded piece of the aluminum adherend
profiled in A to the supporting titanium
alloy substrate.
profile goes through a m1n1mum at the metal-oxide interface (30).

This artifact serves as a consistent marker for the purpose of
thickness calculations. This point also corresponds to the mid-
point of the positive going slope of the oxygen profile between the
16~ and 84~ levels. Therefore, when the AIKLL artifact is not
evident the oxygen profile can be used. The interface between the
metal oxide and the adhesive was determined empirically to be the
midpoint of the negative going slope of the oxygen profile, between
the 84~ and 16~ levels.
Returning to the profiles in Fig. 6, attention is directed to

the carbon profiles, which represent the adhesive in the inter-
phase. In the case of the dense, nonporous anodic film, there was
very little adhesive flow into the oxide. But it appears that
adhesi ve is present throughout the entire porous oxide film in
decreasing amounts towards the interface of the oxide and metal.
This, of course, should be expected if the anodic oxide density
increases towards the oxide-metal interface (19). One element
present in the anodic film formed with phosphoric acid, but not
profiled in Fig. 6, is phosphorous. Thompson et al. have reported
the presence of the phosphate in microcrystallites along the walls
of the columnar Al 20 3 with higher concentrations at the oxide
surface or pore openings (17). The AES spectrum from the surface of
a phosphoric acid anodized aluminum sheet in Fig. 8 does show the
presence of phosphorous, and the profile of phosphorous in Fig. 9
agrees with the model proposed by Thompson et ale Figure 10
contains the Auger in-depth profiles of AI, 0, C, and P from a 20
volt, 5 minute, 1.0m H3P0 4 anodized aluminum film bonded with FM400.
The phosphorous profile in Fig. 10 no longer shows a higher concen-
tration of that element at the oxide surface, but rather a Gaussian
like distribution throughout the oxide with a slightly higher
concentration towards the oxide-adhesive interface. Therefore, the
difference between the phosphorous profiles in Figs. 9 and 10 imply
that, during the cure cycle, the phosphate at the surface moved
ahead of the adhesive into the pore vacancies.
Comparison of the profiles in Fig. 10 with those from the
phosphoric acid anodized specimen in Fig. 6 shows less adhesive
penetration into the oxide in Fig. 10. Profiles from other speci-
mens with thinner oxide layers, bonded with FM400, indicate a limit
of penetration, probably related to the presence of inorganic
filler material in the adhesive.
In-depth profiles from an interphase can be used to assess cure

conditions such as temperature, time, and pressure. Too rapid
curing can result in inadequate adhesive flow and wetting. The
carbon profiles in Fig. 11 reflect subtle differences in adhesive
penetration with a 10 0 difference in cure temperature. In this case
the aluminum was anodized at 10 volts in phosphoric acid and bonded
with FM73.
rr
/ ____ _ ~_. _ . ___ I
-".J ~~ ,~
260 400 600 abo 1000 12'00 1';00 1600 1800

ENERGY (eV)
Fig. 8. Auger electron spectrum of the surface of
phosphoric acid anodized aluminum.
...J
<{
Z
C>
en
a::
w
C>
:::>
<{
o
W
N
...J
<{
:::!:
a::
o
z L-__~~~~~~~~~~~~~~~
o 5 10 15 20 25 30
ION SPUTTER TIME {min}
Fig. 9. Normalized elemental profiles of AI, 0,

and P, as a function of ion sputter time,
through the anodic oxide layer on phosphoric
acid anodized aluminum.
f------ 2800'& - - - -t--- 3100 ) \ - - - - 1
AI
P
...J
«
z(.!)
Vi
a::
w
C>
::>
«
0
w
N
...J
«
:!:
a::
0
z
0 10 20 30 40 50 60 70 80 90 100
ION SPUTTER TIME (min)

function of ion sputter time of AI, 0, P,
and C through the interphase of 20 volt,
5 minute, 1.0M H3P04 anodized thin film
aluminum bonded with FM400.
~1 750A ~
....J o
«
z
(!)
U'i
a::
w
(!)
=>
«
o
w
N
....J
«
~
a::
o
z
50 60 80
o 10 20 30 40
TIME (min)
ION SPUTTER

function of ion sputter time of AI, 0, and
C through the interphase of 10 volt, 1.0M
H3P04 anodized thin film aluminum bonded
with FM73 at cure temperatures of 245°F
versus 255°F.
A major problem associated with aluminum alloys, especially

when adhesively bonded, is corrosion. Consequently, corrosion
inhibiting primers are often used to protect the alloy surfaces.
This, of course, presents problems to adhesive bonding since the
presence of a primer may not be conducive to good bonding. Marceau
(31) reported adhesive failure modes in phosphoric acid anodized
2024-T3 clad aluminum bonded structures when BR227 (23) was used as
a corrosion inhibiting coating. He reported the presence of high
molecular weight molecules at the oxide-adhesive interface which
might have been too large to penetrate the pore voids in the anodic
oxide. When anodized aluminum is coated with a silane primer,
normally used to promote adhesive on glass or glass fibers, failure
modes of bonded structures of these primed anodized aluminum panels
are usually adhesive at the metal oxide-adhesive interface. The
elemental profiles in Fig. 12 from the interphase regions of a
silane primed and unprimed bonded aluminum thin film adherend
structures reveal the role of the primer, as represented by the
silicon profile, as preventing adequate penetration of the adhesive
into the oxide layer. In fact, adhesive penetration in the silane
primed specimen appears to be limited to the primer coating.
BONDED ANODIZED AI-NO PRIMER
---.I
<{
AI
r 3350
o
A----1
z
(!)
(J)
cr
w
(!)
::>
<{
Cl
W
N
...J
<{
::;!
cr
0
z
10 20 30 40 50 60 70 80 90 100
ION SPUTTER TIME (min)
BONDED ANODIZED AI-SILANE PRIMER
---.I 0
<{ Si
Z
(!)
(J)
c
cr
w
(!)
::>
<{
Cl
W
N
---.I
<{
~
cr
0
z
10 20 30 40 50 60 70 80 90 100
ION SPUTTE R TIME (min)
Fig. 12. Normalized Auger elemental profiles as

a function of ion sputter time of AI, 0,
and C through the interphase regions of
identical phosphoric acid anodized bonded
thin film aluminum structures in which
one was coated with a silane primer prior
to bonding.
A primer that has been used successfully for FM73 bonded

structures of anodized aluminum alloys is BR127 (23). This is a
modified epoxy phenolic primer containing strontium chromate which
quickly settles upon application (32). Figure 13 contains Auger
elemental profiles of the interphase of a thin film aluminum bonded
specimen in which the aluminum was phosphoric acid anodized and
coated with BR127 prior to bonding. Its profile shows carbon to be
present throughout the oxide layer. To the right of the profiles in
Fig. 13 are positive secondary ion mass spectra (SIMS) recorded
simultaneously at the specified time intervals during ion beam
etchi~. SIMS was used to monitor Sr because of its sensitivity for
the SR ions (33). At low Sr concentrations, such as that in BR127,
the Auger signal strength was insufficient for profile analysis.
The SIMS data shows Sr at the oxide-adhesive interface and not
throughout the anodic oxide.· This possibility infers that the
carbon detected within the interphase is associated with the cured
primer and not the adhesive. SEM micrographs confirm the presence
of large particles with mean separation distances of approximately
30~, which together with its epoxy solvent, could act as a barrier
for adhesive penetration into the anodic oxide.
AUGER PROFILE +SIMS
/\ Ie
r-720oA-i AJ+
..J
C(
Z I
0
,
~
~75m"
=c'6Om"
(l)
en
ct:
LIJ
(l)
:;)
C(
150min
0
LIJ
N
140min
::i
C(
2
ct: 125min
0
z
75min
, ,
40 160 80 90
ION SPUTTER TIME (min) A.M.U.
Fig. 13. Secondary mass spectra of Sr+ (right) and

normalized Auger elemental profiles, as a
function of ion sputter time, of AI, 0, and
C (left) through the interphase of a BR127
primed and bonded 10 volt, 1.0M H3P04 anodized
thin film aluminum structure.
Chemical Bonding
One of the major questions concerning the bond joint is what,

if any, chemical bonding occurs between adhesive and the oxide layer
of the adherend and to what extent this contributes to overall bond
strength and durability? In their work with the adhesive bonding of
aluminum with Epon 828, Lin and Bell suggested that chemical bonding
between the adhesive and aluminum during cure and the decrease in
adhesive bond strength may be related to a decrease in chemical
bonding (34).
One of the best means available to study chemical bonding is

photoelectron spectroscopy. When the exciting mechanism is an x-
ray beam, the technique is referred to as x-ray photoelectron
spectroscopy (XPS). AES can also provide some chemical bonding
information, but that capability is somewhat limited. For example,
Fig. 14 contains the low energy SiLMM AES spectra from (A) a surface
coated with the silane primer previously discussed, and from (B) a
region within the interphase region of the primed and bonded
specimen profiled in Fig. 12. The difference in shape between these
two spectra gives some indication of a change in the chemical
bonding of Si in the pure primer versus Si in the primer-adhesive
interphase. Likewise, the carbon AES spectra in Fig. 15 were
recorded at the oxide-adhesive interphase of the phosphoric acid
anodized specimen profiled in Fig. 6, and at a point within the bulk
adhesive. Of course, extreme caution must be exercised in applying
any interpretation to the differences between the spectra in Figs.
14 and 15 because of a number of unknowns, the foremost being
electron and ion beam effects (35). XPS, on the other hand, does
not introduce beam damage effects. Therefore, an identical bond
joint model was analyzed by (XPS) and the resultant carbon 1s
spectra are shown in Fig. 16. The solid line trace is the C1 from
the bulk adhesive and the dashed line trace is the C1 from aithin
the oxide layer. The main peaks with a binding energ~ of approxi-
mately 285 eV reflect carbon-carbon and carbon-hydrogen bonding
while unresolved features contributing to the asymmetry on the high
binding energy side of the C1 peaks reflect carbon-oxygen bond-
ing (36). The broadening of l-he C1 from the interphase region
where aluminum oxide is present indiScates an increase in carbon-
oxygen bonding. Thus, the differences in the CK~L spectra in Fig.
15 probably reflect real differences in the chem~cal bonding state
of carbon, although possible ion beam effects are unknown.
SUMMARY AND CONCLUSIONS
Surface characterization techniques, such as AES and XPS, have

been used to characterize the interphase region of an intact
adhesi vely bonded structure using thin film adherends which were
-Si LMM
ADHESIVE
A
INTERPHASE
1 1 1
"
,
40 80 120 250
eV eV
Fig. 14. SiLMM Auger electron Fig. 15. CKLL Auger electron
spectra from (A) a silane spectra from the interphase
primed surface and (B) inter- and bulk adhesive regions of
phase region of the silane the phosphoric acid anodized
primed specimen profiled in specimen profiled in Fig. 8.
Fig. 12.
thin enough to permit conventional depth profiling through the

interphase region. Attempts in the past to obtain chemical infor-
mation from the interphase region of a ruptured or failed bonded
structure with thick adherends using AES and XPS have been limited
because of severe charging at the onset of the analysis when the
surfaces are either nonconducting or there is uncertainty as to
failure location with respect to the interphase. It has been
demonstrated that elemental distribution and chemical' bonding
information obtained from an interphase can be used to evaluate the
effects of various bonding parameters and adherend surface condi-
tions on the performance of an adhesively bonded structure. When
combined with mechanical test data, chemical and physical informa-
tion from the interphase should lead to clearer understandings of
the nature of a particular failure mode in an adhesive bond.
c-c
or
C-H
~
ADHESIVE
INTERPHASE
o
c-o
" I
290 285 280 275

BINDING
IJ .... ...,II .. V
ENERGY
,-I"L-I'V'
eV)
VV I
Fig. 16. C1s photoelectron spectra from the inter-

phase and bulk adhesive regions of the
phosphoric acid anodized specimen profiled
in Fig. 8.
ACKNOWLEDGEMENTS
Research was sponsored in part by the Air Force Materials
Laboratory, Air Force Systems Command, United States Air Force,
Contract F33615-76-C-5185.
The authors wish to thank G. Hammer and J. Peters for the XPS
data and W. Moddeman and J.T. Grant for helpful discussions and data
interpretation.
REFERENCES
1. D.W. Dwight, ~ Colloid Interface Sci. 59, 441 (1911).
2. W.L. Baun, J. Adhesion 1, 261 (1916).
3. W.L. Baun, Experimental-Methods to Determine Locus of Failure
and Bond Failure Mechanism in Adhesive Joints and
Coating-Substrate Combination, in "Characterization of
Metal and Polymer Surfaces", Vol. 1, K.L. Mettal, ed.,
Academic Press, New York (1911). -
4. C.F. Garrett and E.F. Good, "Characterization of Bonding Sur-
faces Using Surface Analytical Equipment", 4th Inter-
national Symposium on Contamination Control, Washington,
D. C., September 1918; proceedings to be published by
Plenum Publishing Company, April 1919.
5. L.T. Drzal, "Summary of the Workshop Held on the Role of the
Polymer Substrate Interphase in Structural Adhesion", Air
Force Materials Laboratory Technical Report AFML-TR-11-
129, 1911.
6. P .W. Palmberg, Journal of Vacuum Sci. and Technol. .2" 160
(1912).
1. N.T. McDevitt, W.L. Baun, and J.S. Solomon, J. Electrochem.
Soc. 123, 1058 (1916). -
8. J.S. Solomon and W.L. Baun, "Surface Characterization of Con-
tamination of Adhesive Bonding Materials", 4th Inter-
national Symposium on Contamination Control, Washington,
D. C., Spetember 1918; proceedings to be published by
Plenum Publishing Company, April 1919.
9. W.L. Baun, N.T. McDevitt, and J.S. Solomon, Chemistry of Metal
and Alloy Adherends by Secondary Ion Mass Spectroscopy,
Ion Scattering Spectroscopy, and Auger Electron Spectro-
scopy, in "Surface Analysis Techniques for Metallurgical
Applications", ASTM STP 596, American Society for Testing
and Materials, 86 (1916).
10. J.W. Coburn and C. Kay, Critical Review of Solid State Science
4, 561 (1974).
11. H.J. Mathiew and D. Landolt, Depth Profile Analysis of Thin
Oxide Films by Auger Electron Sepectroscopy, in "Proceed-
ings of the 1th International Vacuum Congress and 3rd
International Conference on Solid Surfaces" (Vienna
1911), R. Debrozemsky, F. ~udenauer, F. Viehbock, and A.
Breth, ed., Dobrozemsky, Vienna (1911).
12. Y.H. Choo and O.F. Devereux, J. Electrochem. Soc. 123, 1868
(1916). - - - --
13. R.E. Pawel, J.P. Penler, and C.A. Evans, ~ Electrochem. Soc.
119, 24 (1912).
14. W.L. Baun, "Surface Analysis of 2024 and 1015 Aluminum Alloys
After Conditioning by Chemical Treatments", Air Force
Materials Laboratory Technical Report AFML-TR-15-122,
1915.
15. W.L. Baun and N.T. McDevitt, "Surface Characterization of

Titanium and Titanium Alloys, Part II: Effect on Ti-6A1-
4V Alloy of Laboratory Chemical Treatments", Air Force
Materials Laboratory Technical Report AFML-TR-76-29,
1976.
16. H.S. Schwartz, SAMPE Journal 13, 2 (1977).
17 . G. E. Thompson, R:"C:""" Furneau, G. C. Wood, J. A. Richardson, and
J.S. Goode, Nature 272, 433 (1978).
18. J.P. O'Sullivan and G.C. Wood, Proc. Roy. Soc. London, A317,
511 (1970).
19. J.U. Diggle, T.C. Downie, and C.W. Goulding, Chem. Rev. 69,
365 (1969). -- -- -
20. C.J. Dell'Oca, Thin Solid Films 26, 371 (1975).
21. W.H. Gutmann, "Concise Guide to Structural Adhesives", Rein-
hold Publishing Company, New York, 37 (1961).
22. Minnesota Mining and Manufacturing ACIS Division, 3M Center,
St. Paul, Minnesota 55101.
23. American Cyanamid Company, Bloomingdale Aerospace Products,
Havre de Grace, Md. 21078.
24. J. Solomon and W.L. Baun, I:. Vac. Sci. Techno!.~, 375 (1975).
25. T.N. Wittberg, J.R. Hoenigman, and W.E. Moddeman, J. Vac. Sci.
Techno!. 15, 348 (1978). --- --
26. J. Kerchner an~H.W. Etzkorn, Thin Film Analysis, from Sputter
Profiles to Depth Profiles by Combined Auger/X-Ray Analy-
sis, in "Proceedings of the 7th International Vacuum
Congress and 3rd International Conference on Solid Sur-
faces" (Vienna 1977), R. Debrozemsky, F. Rudenauer, F.
Viehbock, and A. Breth, ed., Dobrozemsky, Vienna (1977).
27. G.K. Wehner, The Aspects of Sputtering in Surface Analysis
Methods, in "Methods of Surface Analysis, A.U. Czanderna,
ed., Elsevier Scientific Publishing Company, Amsterdam
(1975).
28. G. Paolini, M. Masares, F. Sacchi, and M. Paganelli, I:.
Electrochem. Soc. 112, 32 (1965).
29. R. Honig and U. Harrington, Thin Solid Films .12, 43 (1973).
30. J. Solomon, Applied Spectroscopy 30, 46 (1976).
31. J.A. Marceau, SAMPE Quarterly~, 1 (1978).
32. K.K. Rice, S.L. Lehmann, D.K. Klappratt, and C.L. Mahoney,
"Exploratory Development of Corrosion Inhibiting Pri-
mers" , Air Force Materials Laboratory Technical Report
AFML-TR-77-121, July 1977.
33. A. Benninghoven, Surface Sci. 53, 593 (1975).
34. C.J. Lin and J.P. Bell, J. Appl. Polymer Sci. ~, 1721 (1972).
35. P. Braun, W. Faber, G. Betz, and F.P. Viehbock, Vacuum 27, 102
(1977) .
36. K. Siegbahn, C. Nordling, G. Johansson, J. Hedeman, P.F.
Heden, K. Hamrin, U. Gelius, T. Bergmark, L.O. Werme, R.
Manne, and Y. Baer, "ESCA Applied to Free Molecules",
North Holland, Amsterdam, 1969.
Techniques for Measuring Adhesive vs.
Cohesive Failure
Tennyson Smith and. Pierre Smith *

Rockwell International Science Center
Thousand Oaks, California 91360
ABSTRACT
Techniques have been developed for measuring the fractional
area of an adhesive joint that has failed adhesively (near or at the
substra te-adhesi ve interface) and cohesi vely (in the adhesi ve) •
a
These techniques involve ray backscattering, light reflection and
microscopy from an array of positions on the fracture surfaces of
the failed joint.
The a ray count rate depends upon the elemental composition and
is, therefore, different for the substrate and adhesive. The count
rate correlates with the fractional area that failed adhesively and
cohesively, for the localized spot that is monitored. A map of the
count rate values correlates with microscopic photographs which
also reveal adhesive vs. cohesive failure. The S ray and reflection
techniques are conducive to automatic computerized mapping and
averaging whereas the photographic technique is not.
INTRODUCTION
One of the common ways of characterizing an adhesive joint is
to measure the force required to fracture a lap shear bonded couple.
The lap shear bond strength is very useful for comparing adhesives
and surface preparations. Perhaps of equal or more importance is
the loci of fracture within the adhesive joint. Figure 1 outlines
the various loci that are possible. In the case of metal adherends
failure can occur in the adherend, in the metal oxide, in a weak
-.::--
Summer employee from Brigham Young University, Provo, Utah.
123
124 TENNYSON SMIT H AND PIERRE SMITH
contamination boundary layer, in the primer, in the adhesive and at

or near any of the boundaries between these phases. The fracture
surface is a topographical map of the weakest regions within the
joint. That is the fractograph is a map of those regions for which
the local stress first surpassed the local failure stress . Discov-
ery of the weakest regions can lead to corrective action in bond
design and thus lead to stronger more durable joints .
The usual procedure is to report the lap shear strength and a

qualitative statement that the failure was either cohesive or
interfacial. This judgement is based on a visual observation that
failure occurred at or near the adherend-adhesive interface (adhe-
si ve failure) i f the surface has the appearance of the metal, or
within the adhesive (cohesive failure) i f the surface has the
appearance of fractured adhesive . Although it is hoped that
eventually quick and inexpensive techniques will be developed for
characterizing the loci and mode of fracture in detail over the
fracture surface, a simple means for evaluating the fraction of the
surface that failed adhesively vs that which failed cohesively and
mapping these regions would be a step forward.
This paper reports new techniques for measuring the fraction

of interfacial failure, mapping these regions, and relates the
fraction to lap shear bond strength for brittle fracture.
Adhesive
Interface
Fig. 1. Schematic diagram of bond regions and

fracture topography.
ADHESIVE VS. COHESIVE FAILURE 125
EXPERIMENTAL
Three techniques have been investigated for measuring quanti-

tative values for the fraction or percent interfacial failure.
These three techniques are microscopic analysis of low magnifica-
tion micrographs, Bray backscattering and light reflection.
The Microscopic Technique (1)

The mating surfaces of a fracture joint are placed under a low
power microscope and photographed to provide an enlargement of
about 15X by piecing together an array of small areas. An 'example
of one piece of an array is given in Fig. 2. The areas that appear
to be the metal substrate (adhesive failure) are easily discern-
ible. The interfacial fractional area is determined by placing a
plastic grid over the photograph and counting small squares within
each interfacial region. To obtain the percent interfacial failure
the area of the squares counted is divided by the total area and
multiplied by 100. This technique is tedious and time consuming
(therefore expensive), but was done as a check on the other techni-
ques which are more easily automated.
Fig. 2. Photograph of microscopic view of lap shear

bond fracture region. Reveals cohesive ~
interfacial failure.
126 TENNYSON SMITH AND PIERRE SMITH
The ~ Backscattering Technique
A sample is irradiated with ~ rays (electrons) and the back-

scattered current is measured with a Geiger-Mueller counter. The
instrument used for this purpose has the trade name "Micro-Derm"
(2). The backscattered current depends on the elemental composi-
tion of the sample. Thus, if the sample is a metal with very thin
metal oxide and contamination layer, the backscattered current will
be characteristic of the particular metal. For oxides and con-
tamination of a few hundred Angstrom the effect of the oxygen and
carbon will be negligible. If the sample is adhesive the current
will be characteristic of the adhesive. For example, the metal
mostly used in this study was Al 1015-T6 alloy which gave a ~ count
of about 100 in 16 sec whether it was uncleaned, given the standard
FPL etgh (sulfuric acid-dichoromate) or the phosphoric acid anodize
(~4000A hydroxide) treatment.
The adhesive was a modified epoxy (FM13 from American Cyana-

mid) which gave a count of about 35 in 16 sec. The percent
interfacial failure was determined by measuring the ~ current at a
large number of equally spaced positions over the fracture surface,
summing the counts and dividing by the number of measurements (this
number is referred to as c). The percent interfacial failure is
determined by the following equation:
(1)
where cA is the ~ counts from the fracture surface adhesive and cM

is the ~ counts from the metal.
The Reflectance Technique

A sample is irradiated with light that has been focused to a
point. The light is scanned over the fracture surface and the
reflectance recorded for each position. The interfacial regions
are reflective whereas the regions of cohesive fracture scatter the
light strongly. We have used a HeNe laser beam for this purpose.
An equation similar to Eq. 1 is used to calculate the ~ interfacial
failure.
Computer Mapping (3)

The ~ backscattering and reflectance techniques are easily
automated to produce computerized scanning maps of the fracture
surface. This is done by a microcomputer operated X-Y positioning
table that scans the sample beneath the sensor head and stores the
count or reflectance value for each position. The computer is
programmed to either calculate the ~ interfacial failure area or to
ADHESIVE vs. COHESIVE FAILURE 127
produce a map of the fracture surface. The map is made by plotting

a dot pattern with the density of dots proportional to the e count
or reflectance signal value at a given position. The map is blank
in the cohesive regions and dotted in the interfacial failure
regions.
EXPERIMENTAL RESULTS
Calibration of e Backscattering with the Photographic Technique
Four samples of Al 7075-T6 alloy were phosphoric acid anodized

(PAA) in 10% H3P0 4 at 10V for 20 minutes and used as control samples
to mate with samples that deviated from the proper process to
deliberately produce weak bonds and increased interfacial failure.
Table 1 gives the sample treatment, the deviation and the resultant
% interface failure area as determined by the photographic and
back scattering techniques. The last column in Table 1 gives the lap
shear strength between the control sample and the damaged sample.
Samples lB and 2B were both given the proper PAA treatment,

sample 2A was exposed to the phosphoric acid for 20 minutes, but
without applied voltage; sample lF was PAA treated and then contami-
nated with stearic acid by placing a drop of pentane containing
stearic acid and spinning the sample to obtain a uniform deposit.
Samples lZ, 2Z, 10 and 20 were given the standard FPL (sulfuric
acid/dichromate) etch, then lZ was contaminated with silicone
vacuum grease and 10 was smudged with a clean cotton glove. Sample
lS and 2S were stainless steel samples that were anodized in
sulfuric acid/dichromate solution.
Figure 3 is a plot of % interfacial area as measured by the

Micro-Oerm (e backscattering) vs. that determined by the photo-
graphic picture technique. A good correlation exists between the
two techniques. The data would fall along the line in Fig. 3 i f
both techniques would have produced identical results. The fact
that all of the e backscattering data yield greater % I.F. than the
photographic technique is significant and expected. This is be-
cause it is difficult to estimate interfacial regions smaller than
the grid size on the clear plastic that are undoubtedly present in
the area that appears to be cohesive failure. The B back scattering
and reflection techniques do not suffer from this limitation and
thus pick up these areas as well as the larger ones that are
visible.
Locus of Failure
Having devised a means of identifying the locus of cohesive vs.

adhesive failure in a plane parallel to the adherend surface, by
....
~
CIO
Table 1. Correlation Between Interfacial Failure, as Measured by the Photographic
and B Backscattering Technique, and Lapshear Bond Strength
Sample ~ Interfacial Failure Area Lap shear

No. Material Treatment Deviation Photographic Backscattering Bond
Strength
(Ksi)
1B Al 7075-T6 PAA* None 30 32.7
2B " PAA None 10 13.4 4.3
46.1
1A " PAA None 15 17.4
2A " PAA O.V.+ 33 36.8 3.5
54.2
2F " PAA None 11 15.2
1F " PAA stearic acid 21 27.0 6.1
contamination 42.2
-!
2Z " FPL** etch None 2 5.3 m
1Z FPL etch silicone grease 80 90.4 1. 75 z
" z
contamination 95.7 -<
en
0
2D FPL etch None 6 14.0 Z
" en
10 " FPL etch cotton glove 49 54.0 4.7 3:
smudge 68.0 :::j
:J:
1S SS Anodize in H2S0 4/Cr 207 None 26 27.0 »
z
c
"ll
m
:II
* PAA = phosphoric acid anodize in 10% acid :II
m
** FPL etch = sulfuric acid/dichromate etch en
+ Zero applied voltage 3:
:::j
:J:
ADHESIVE VS. COHESIVE FAILURE 129
Fig. 3. Measurement of Interfacial Failure

(picture vs. micro-derm).
micrographs or computer maps, it is of interest to examine the loci

of failure perpendicular to the surface of any particular map
position. From the low magnification photograph in Fig. 2 and the
SEM micrographs in Fig. 4, three loci are identified. These loci
are fracture in the polymer matrix, fracture at the matrix-fiber
interface and fracture at the metal-matrix interface.
The ~ interfacial failure was based on fracture at the metal-
matrix interface and although these areas appear, from photographic
and SEM pictures, to be metallic with roll groove pattern and grain
boundaries, these pictures cannot reveal the presence or absence of
very thin oxide, contamination or adhesive. Thus Auger electron
spectroscopy (Fig. 5) was performed at the spot indicated in Fig. 4.
The AES spectrograms (Fig. 5) were taken prior to Ar+ sputter-
ing and after Ar+ sputtering for 2 minutes. Prior to Ar+ sputtering
only a large carbon, a small oxygen and perhaps sulfur peaks are
observed. This indicates that the fracture surface is not metal or
metal oxide, but either adhesive or stearic acid contamination. The
short sputter time required to reveal the aluminum oxide is in favor
of stearic acid contamination as the locus of failure.
~
w
o
AES
-I
m
Z
Z
-<
en
o
Z
en
:5:
=i
J:
»
z
o
"C
X40 m
X250 X1000 :0
:0
m
Fig. 4. SEM micrographs of interfacial failure area of 1F (Table 1). en
:5:
=i
J:
PR lOR TO Ar + SPUTIER
c AFTER 2 MIN Ar + SPUnER
Fig . 5. Auger electron spectroscopy of interfacial

failure regions indicated in Fig. 4.
Computer Mapping
The equipment has not as yet been modified to map with respect
to B backscattering, but Fig. 6 is a reflection map of a fracture
surface. The dotted regions faithfully correspond to the visual
observed interfacial failure regions.
Relationship Between Bond Strength and Interfacial Failure
One might expect that an increase in interfacial failure would

decrease the lap shear strength and this is observed in Table 1.
However, there is no obvious relationship between lap shear
strength and individual molecular bonds between adhesive and adher-
end because failure is initiated at a point (flaw) at which the
local stress first exceeds the local strength and then proceeds
sequentially from point to point as this criteria continues to be
met. Of course, once a crack is initiated this sequence of points
usually is at the crack tip, thus causing it to propagate along a
surface of least resistance.
1. 21
, Ii IJIlIl ;,),~ f.!~M:;:H :~:{J'li'&fi(I)P' ~ : :;?~?'i";'f'" " : ;, , l'"
llJ_,'"1\ ':1111 : '1111 ~\j ... /!\I
: : '"J II II II}' ' un
'till':1111",";:,,',d:~""
II , '11111111"',:
!'lIi ' N,l: : 1111 ""',, ,:c ,,~:<
1111 ','" ; 1," " ~'<" •
': ':'
1111""
~>"," ~ li.1I1:~',II:~;~ Y!~, , "c, • ':C
(,,','i lill , , , , • • •••
IIll
"','11'11 :,'1111 1111
"'I'
"'"
':"' ,
II!,IIIII, , ::
(j,,1 :, '
;; : 11111.'1'
,.',.
111.1 "',, ':';,
1111 "
II
e .48
Fig. 6. Computer map of reflectivity of a fracture
surface. The dotted regions failed inter-
facially.
The question we would like to answer is: what relationship is

there between the composite materials, the configuration of the
joint and the stress to initiate failure? However, we limit our
discussion to the question of why there is an approximately linear
relation between lap shear strength and the fraction of the bond
area that failed at the interface?
In the schematic Fig. 7a, the thickness dimension of the
adhesive layer has been exaggerated. To indicate a possible stress
distribution profile, curves have been drawn within the grid for
which the horizontal axis remains the position axis but for which
the vertical axis is used as a stress axis. A section of the grid
near a region of probable fracture initiation is expanded in Figs.
7b and c. In Fig. 7b a particular column has been selected to
identify possible materials (oxide, contamination, gas bubble,
primer and adhesive) that might occupy the net of elements between
adherends.
The average shear stress 0 (measured by the Instron tester)

will be the sum of the stresses 0. associated with each column j
modified by an interaction paramet~r a due to interaction between
columns,
O=Lao (1)
j"
j
For a given strain, o. will depend upon the compliance of each
column and 'lill therefore depend upon the materials within the
column. It should be noted, that although Eq. 1 relates the
measured stress and the stress distribution, it tells nothing about
STRESS DISTRIBUTION
(a)
,,- ,\ -
'
(l~ / v..... -"\\
1/ ,I /
/ - ......."
~ -
(l -
\
\ I~ ''t I
\ l
t2
\-:.J , ..... 1
--
'\"'- -;,,"7
"- '\..J
(l
(l - r-- __ _/
,/
......
..... _ / ' .....- .... " ..... _/"
tl
2 3 4 6
k
(e)
(bl
Fig. 7. Schematic representation of (a) lap shear

joint, (b) one element, and (c) stress profile
in the critical region.
the value of a at which failure will occur (OF).

To estimate OF' we invoke the model in Fig. 7c. Consider a
collection of connected columns (k in number), each column having
critical stress values 0., associated with the ith material within
the column. For simplicfty we consider only two regions, that near
the adherend interface and that in the adhesive. In Fig. 7c the
horizontal bars indicate the critical stress values near the inter-
face have a spacial distribution related to the surface contamina-
tion in column k = 1, 2, 4 and 6. If the amount of contamination in
each column is about the same, the values of o. will be close, with
average value designated OC for adhesive fa!lure. The bars at
a
higher levels are for failure in the adhesive with average cohesive
value O~.
Figure 7c is for the collection of columns in which the imposed
topographical stress distribution surface first contacts the topo-
graphical yield stress surface. Curves O't ' O't and 0'0. indicate the
hypothetical stress distribution long bef3re, ~ust prior to and at
failure. The compliance values of each column influence the
distribution curve O't' As 0' increases such that the distribution
curve O't reaches the 1critical stress values for interfacial fail-
ure, som€ adhesive failure occurs which changes the compliance and
thus modifies the stress distribution curve to show some relaxation
in the failed region. As the failed regions grow the stress
distribution curve is pinned in these regions, but continues to
increase in the other regions. When the stress distribution curve
contacts the upper cohesive levels, they fail and a flaw of critical
size is created. The crack propagates causing failure in the whole
joint.
It follows from this model that the average critical stress in

the set of k columns will depend on the fraction of the area, of the
k columns, that fail at the interface (8 ) and the fraction that
fails cohesively (8 ), i.e. a
c
(2)
Since the average stress, 0 k' in the critical flaw region,

should increase in proportion to the average stress for the total
joint, we make the approximation
o ~ l3 O'k
where l3 is a proportionality constant. It follows that
(4)
and
OF ~ 8a (O'a - 0c) + O'c (5)
where °a =ex l3kCJ ca (6 )
(7 )
The results in Table 2 can be used to test Eq. 5. A set of

PAA, Al7075-T6 samples were contaminated with silicone grease to
varying degrees. These samples were then bonded to uncontaminated
PAA samples, fractured and both fracture surfaces mapped with e
backscattering for %interfacial failure (% I.F.). The %I.F. for
»
Cl
::z:
m
S!2
<
m
<
!"
Table 2. Correlation Between Lapshear Strength and Percent Interfacial Failure 8

::z:
(PAA-Al 7075-T6-FM73) m
S!2
<
m
"T1
»
Lap Shear r
C
Strength :0
m
%Interfacial Failure B Backscattering (Ksi)
Contaminated
with Silicone
Sample Control grease Total
15 27 42 6.1
2 21 22 43 5.7
3 22 45 67 4.9
4 14 69 83 3.2
5 10 87 97 2.6
.-
Co)
~
the control samples (curve 1), the contaminated samples (curve 2)

and the sum of these (curve 3) are plotted vs. the bond strength in
Fig. 8.
The data from the contaminated surface and the data for the
total %I.F. fall near a visual straight fit in accordance with Eq.
5. As failure at the weak interface increases, failure at the
control interface naturally decreases (see lower curve, Fig. 8).
A plot of a F vs. % 1. F. for which ac is constant (common

adhesive) but a varies (different interfacial failure mechanisms),
should result i~ lines converging to a common value ofa f at zero %
I.F. but intercept at different values for 100% I.F. Th1s is shown
in Fig. 9 from the data in Table 1.
The intercepts at % I.F. = 100 and 0 give the extrapolated

values of the fracture strength of the interface and the cohesive
strength of the adhesive respectively. From Fig. 8 the cohesive
strength of the FM73 is about 8.6 Ksi for the 5 mil bond line
thickness. The fracture strength of the silicone grease contami-
nated surface is about 2.3 Ksi. Since the cohesive strength should
remain constant for a given adhesive and bond line thickness,
"'- .
,~ .
:::c 6 o.'o'Q.... ¢
o ) ..
a:i i ...... 0
"
0:
l-
V!
c 4 I " ..
~
cc
/
P d'"O
f "0
2 / 0 CONTROL ""-..
o CON - SILICONE GREASE
o TOTAL
00 100
Fig. 8. Plot of bond strength vs. % interfacial

failure.
ADHESIVE vs, COHESIVE FAILURE 137
....
'iii
0~~ __~__~__ ~~~~

4 -_ _ _ _- L_ _~~
o 20 40 60 80 100
% INTERFACIAL FAILURE
Fig. 9. Plot of bond strength vs. %interfacial
failure.
fracture strengths for various types of surface damage can be

obtained from the intercept at %I.F. = 100 as in Fig. 9. For the
lower curve Eq. 5 is not vali4, as %I.F. approaches 100 because OF
cannot be negative.
Another example of the utility of Eq. 5 is given in Fig. 10,
from a previous report (1). Aluminum samples were aged in humid
atmosphere with varying degrees of condensation. The thickness of
the oxide was measured by ellipsometry and ranged between about 200i
and 600i as the condensation increased. These samples were bonded
with a different adhesive from American Cyanamid (HT 424). This
adhesive is also a modified epoxy but contains a glass scrim rather
than a nylon scrim (in FM13). The larger diameter glass produces
bond line thickness (BLT) values of about 10 mil as compared to-5
mil for FM13. Except for two points the data fall along a straight
line as expressed by Eq. 5. For this adhesive and BLT, 0 '" 3.8
Ksi and 0 '" 2.5 Ksi. The value of 0 has been shown to cbe very
sensitive~o BLT and depends on the pr~sence of the scrim. It was
shown (4) that the shear strength of the matrix polymer in HT424 is
abou t 8.9 Ksi in good agreement with the value extrapolated from
Fig. 8 (i.e. 8.6 Ksi).
Curve 2 in Fig. 10 is a plot of oxide film thickness vs. %
LF., indicating a direct correlation between oxide thickness, %
4
~/
/
-......... ... /
/ 500
3 ~ ... /
0 /~ ... 400
<XC
VI / ~ V1
V1
,4"
~ w
Z
:I: :..:
~
<.!) / u
t5 2 / 300 :I:
0::
r- /A ~
~
VI
<I ......
0
z g..,/o w
0 0
w / 200
x
/
/0- 0
/ 100
0 0
0 100
Fig. 10. Bond strength and oxide thickness vs.

% I.F.
I.F. and thus bond strength. It was demonstrated in this case that
interfacial . failure was in the oxide film. The probability of
failure increases in direct proportion to the number of flaws in the
oxide and this increases with film thickness. Although the film
thickness at about 90% I.F. indicates that the bond strength should
fallon the theoretical curve 1 in Fig. 10, the index of refraction
of this film proved to be about 2.5 vs. about 2.0 for the other
films. The larger index corresponds to a more dense oxide (fewer
voids or flaws) and the observed larger value of a is therefore
predicted, i.e., curve 3 is appropriate for this oxi~e.
New techniques are described for measuring the percent inter-

facial (adhesive failure vs. cohesive) failure. The photographic
technique is tedious and time consuming, but is used to verify the
utility of the 13 backscattering and light reflection techniques
which are easily automated. These techniques can be used to produce
maps of interfacial failure regions, in addition to % interfacial
failure.
Plots of lap shear bond strength vs. percent interfacial

failure confirm a theoretical analysis relating these parameters.
An important consequence of this research is that one can extra-
polate the data to obtain the cohesive strength of the adhesive and
the fracture strength at the damaged surfaces. This procedure will
allow the separation of adhesive intrinsic properties from surface
preparation properties and thus lead to a better understanding of
the strength of adhesive joints.
REFERENCES
1. T. Smith, "Mechanisms of Adhesive Failures Between Polymers

and Metallic Substrates", 7th National SAMPE Technical
Conference, Vol. 7, pp. 349 (1975).
2. United Process Assemblies, Inc., 60 Oak Drive, Syosse~, New
York 11791.
3. T. Smith, "Nondestructive Inspection of Phosphoric Acid
Anodized Aluminum Panels for Contamination", AFML-TR-77-
42.
4. T. Smith, "Mechanisms of Adhesive Failure Between Polymers and
Metallic Substrates", AFML-TR-74-73, Part II.
Failure Characterization of a Structural
Adhesive
• +
David W. Dwight, Erol Sancaktar , and Halbert F.
Brinson
College of Engineering
Virginia Polytechnic Institute and State University
Blacksburg, Virginia 24061
ABSTRACT
Strength vs. rate-of-strain data show accurate, viscoelastic
time-dependence for a bulk epoxy adhesive (Metlbond 1113) in ten-
sion and shear, but not in the adhesively lap-bonded state. A
detailed analysis of the lap shear joints, combining results from
neutron radiography, scanning electron microscopy and roughness
measurements, indicated the dominance of internal voids and stress
concentrations not present in the bulk state. A qualitative
fracture mechanics approach was used to reconcile the results. A
general methodology to elucidate practical adhesive bonding prob-
lems is suggested that combines elastic, plastic and surface char-
acterization.
INTRODUCTION
Bulk Properties of Adhesives
Adhesive technology is growing rapidly; composite and light-

weight materials technology for aerospace and automotive applica-
•To whom correspondence should be addressed.

+Current address: Mechanical and Industrial Engineering Depart-
ment, Clarkson College, Potsdam, New York 13676.
141
142 DAVID W. DWIGHT ET AL.
tions are leading the development. Of critical importance to the

designer are strength and durability under anticipated service
conditions. Experience with organic, high polymer adhesives is
limited: for predicting the behavior in bulk or bonded state, static
tensile, dynamic fatigue, and fracture mechanics tests, among
others, are variously recommended.
We have been developing a unified approach that combines

microscopic information from modern surface analytical techniques
(1,2) with traditional viscoelastic studies of bulk tensile and
shear specimens. Using a structural epoxy adhesive (Metlbond 1113)
as a practical model, the fractographic results clarify scatter in
the lap shear strength data, identifying critical, uncontrolled
variables due to processing effects upon structure and properties.
Bulk stress-strain, strain rate, creep, relaxation, yield

and/or failure properties (including delayed failures) of a struc-
tural adhesive (Narmco Whittaker's Metlbond 1113 and 1113-2
adhesives) were measured first (3). These characterization studies
were performed on bulk materials using dog bone tensile coupons.
The data were correlated with a modified Bingham model and methods
described by Crochet (4), Naghdi and Murch (5), and Ludwik (6), to
produce linear viscoelastic models of the bulk adhesives.
It would be useful to determine a relationship between tensile
property data and the design data needed to predict the behavior of
lap-bonded joints. Thus, the present study was initiated to attempt
to find whether bulk tensile properties could be correlated in some
reasonable way to the shear behavior as obtained from bulk shear
coupons and lap shear adhesive jOints using Metlbond 1113. Of
further interest was the viscoelastic response of lap shear joints
of Metlbond 1113 tested with yield data plotted versus elastic
strain rate. Extensive fractographic analysis, combined with qual-
itative fracture mechanics arguments were employed to help explain
the results.
Adherend Surface Preparation
Since joint strength is a function of the adherend surface

preparation, chemical treatments of adherends were considered. Lin
and Bell (7) reported that the shear strength of thin-walled tubular
butt joints reached 9000 psi (62 mpa) after the joints were
machined, polished, vapor degreased, and treated with dilute sul-
furic acid-potassium dichromate solution. Without chemical treat-
ment the bond strengths decreased to 80% of that for chemically
treated jOints. For those joints with aluminum surface degreased
only, the strength was 65% of the maximum bond strength. Good (8)
reports that surface treatment of materials (e.g. etching and
particle bombardment) will help to obtain a system for which the
FAILURE CHARACTERIZATION 143
interfacial region is not the weakest. Dwight ( 1 ) reported

molecular-level details of adhesive bond strength to solid fluoro-
polymers and improvements when the surface was etched with sodium
complex solutions or exposed to an electrical discharge. Joints of
high strength are obtained when the controlling factors in chemical
bonding can be optimized.
In regard to surface roughness, Jennings (9) reports:

1) Roughening of the adherend surface can increase butt tensile or
shear strength of a joint with brittle adhesives. 2) Butt tensile
strength can be a function of the metallurgical state of the
adherend. 3) It is possible to have a joint stronger than the bulk
strength of the adhesive polymer. 4) A change in stress distribu-
tion at the adherend-adhesive interface will cause the joint
strength to vary with adherend treatment. The effect of surface
roughness depends on the type of adhesive mechanism. Mechanical
(interlocking) adhesion will be stronger when appropriate solidifi-
cation is achieved between the pores of the adherend.
A system for which the interfacial region is not the weakest

requires good wetting of surfaces to eliminate interfacial flaws,
and strong interfacial forces should be created by covalent bonding
or chemisorption. Also, the failure mode is strongly dependent on
the adherend surface treatment. In one study, all joints without
chemical treatment failed adhesively at the resin-metal interface.
For those specimens with surfaces treated, both cohesive and adhe-
sive types of failures were observed (7).
Locus of Failure and Mechanics Analysis
Scanning Electron Microscopy (SEM) is presently the best

method of qualitative analysis of micromechanisms in polymer frac-
ture. Recent studies illustrated the effects of weak links such as
voids, corrosion, and interfacial separation (1,2). Often fracture
patterns reflect stress distributions, as well as deformation and
fracture micromechanisms operating during crack propagation. For
example, a fracture mechanics model for "rubber toughening" effects
in epoxy polymers has been developed from SEM photomicrographs
(10) •
To explicitly and quantitatively account for the separate

contributions of interfacial properties and bulk rheology, Andrews
and Kinloch (11) proposed that the adhesive failure energy,
6 =60 • feR) (1)
where 6 0 depends only on the physical and chemical nature of the

fracture surface, and f is a function of the "reduced" rate of
failure propagation obtained from rate and temperature data using
the WLF equation. 60 is the work of bond fracture: the sum of the
interfacial, adhesive and adherend failure energies. Experimental
data required to test this theory are quantitative surface analysis
to determine the fractions of interfacial and bulk failure, the
intrinsic failure energies, and the overall joint failure energy
measured as a function of temperature and rate of crack propagation.
Andrews and Kinloch prepared joints with different surface-
energy substrates bonded to a single SBR rubber, cross-linked in
situ with one initial crack located at the interface at the edge of
the test specimen. Their theoretical development was corroborated
by the results in these idealized specimens. However, in recent
work they have employed qualitative surface analysis to elucidate
the case of mild steel bonded with an epoxy adhesive (12). The
nature and topography of fracture surfaces were examined visually
and by scanning electron microscopy as well as electron probe
microanalysis, using 5~ titanium dioxide tracer in the adhesive.
The initial locus-of-failure was cohesive through the adhesive, but
after water immersion, a complex locus-of-failure was found: the
path of fracture occurred between the oxide surface layer and the
epoxy adhesive, alternating into the adhesive layer.
In an approach to relate failure loads to shear strain energy,

Hart-Smith (3) developed an expression for adhesive joints with
idealized elastic-plastic adhesives and elastic adherends, showing
that bonded joint strengths could be characterized by the adhesive
strain energy (shear strength per unit area of bond). For double
lap joints with uniform shear distributions:
(2)
where
E = Adherend Young's Modulus
~(iy + y ) = Areaunder the stress-strain curve per unit
e p volume of adhesive
d~(iYe + Yp) = shear strain energy per unit area of bond

d = Bondline thickness
t = Adherend thickness
Y , Y = Elastic and plastic adhesive shear strain
e p
~ = Adhesive shear stress
P = Adherend load per unit width.

Generally, similar expressions arise from fracture mechanics

analysis, using Griffith-Irwin theory; for a solid of elastic
modulus E containing a crack of length L, the stress a f at which
fracture will occur is given by
a = K(EG/L)l (3)
f
where K = 1 is a constant and G(= 2y + W) is the total (irreversi-

s
ble) work dissipated per unit crack extension as new surface free
energy (y ) and/or deformation and other fracture process (W).
s
The following sections describe several experiments with a
practical epoxy/aluminum system and evaluate semi-quantitatively
the pertinent interfacial and bulk mechanisms.
EXPERIMENTAL
Materials and Preparations
Surface preparation of the aluminum adherend was done accord-
ing to the following specifications (14):
1) Solvent wipe with methyl ethyl ketone.
2) Vapor degrease in trichloroethylene by suspending in the
vapor until condensation on the surface no longer occurs.
3) Immerse for 20 minutes at 1500 F (66 0 c) in a solution of
68% distilled or de-ionized water, 23% sulfuric acid
(specific gravity 1.84), 9% sodium dichromate.
4) Rinse thoroughly.
5) Air dry so as to prevent retention of water. Thorough
draining is required.
6) Apply the primer at ambient temperature, agitating con-
tinuously during brushing, and air dry to a thickness of
0.00007 in. (1.78 x 10- 3mm) to 0.000025 in. (6.35 x 10-3
mm) •
Single lap shear specimens were prepared with an overlap
length of - 0.30 in. (7. 6mm) , adhesive thickness of -0.006 in.
(0.15mm). Aluminum adherends were 0.125 in. (3.17mm) thick for the
ASTM (D 1002-72) recommended single lap geometry. Symetrical
single lap specimens were also prepared, for they were recommended
by several authors (5,6) as a superior geometry which has lower
stress concentrations and more uniform shear stresses across the
glue line. These specimens differed from the single lap specimens
in that the adherends were machined to half thickness across the
overlap area to produce uniform specimen thickness and vertical

load application line. Aluminum adherends for all specimens were
1.00 in. (2.54cm) wide and were bonded at 260 0 F (121 0 C) for 30
minutes under 50 psi (345 kpa) pressure. Metlbond is a structural
epoxy adhesive commercially available from Narmco Materials Inc.
(600 Victoria Street, Costa Mesa, California 92621) as a 100% solid
film with (Metlbond 1113) or without (Metlbond 1113-2) a synthetic
carrier cloth. (Machining, preparation and bonding of the adhesive
lap joints were performed by the NASA-Langley Research Center.)
Neutron Radiography
Neutron radiographic examination of the symmetric lap speci-

mens was done at the VPI & SU Research Reactor. Using a direct
exposure method, x-ray film was placed in physical contact with a
converter screen in a light-tight container placed behind the
specimen. The converter screen was painted with a thin layer
containing gadolinium, which has a high mass absorption coefficient
for neutrons. (Other work (11) incorporated 5% Gd20~ directly into
Narmco 6800 adhesive, enabling detection of voids ~s fine as 0.02
in. (0.05cm).) Neutrons passing through the sample are absorbed by
the converter screen, which emits radiation and exposes the film.
Neutron flux is governed by thickness, material density, or absorp-
tion cross section, which can be determined from the image density
on the exposed film. Differences make it possible to detect voids,
flaws or other inhomogeneities in the glueline. The n~tro~ flux
emitted from the source was maintained between 10 n/cm and
106n/cm2.
Mechanical Behavior Testing
A symmetric rail shear test (Fig. 1) was used for bulk shear
studies. Specimens for this ~est were prepared with a ratio of 10:1
for the distance between tabs (d) to length (1) in accordance with
the analysis of Whitney, et al. (18). These specimens were sensitive
to tab application: Metlbond 1113 specimens had tabs attached with
Epoxy 901 adhesive and yielded at lower loads due to premature
tab/specimen failure, relative to the Metlbond 1113-2 specimens
that were glued with Eastman 910 adhesive. Loads were applied in
opposite directions on the center vs. the outside tabs and most of
the latter specimens produced 45 0 shear failures (Fig. 1).
i
Tabs are -0.15 in. x 5.2 in. (1.9 x 13.2cm) 0.014 in. (1.6mm)
thick aluminum pieces glued onto symmetric rail shear test speci-
mens to insure deformations in excess of elastic limits.
Fig. 1. Bulk shear test specimen. Load was applied

to the metal tabs as indicated by the arrows.
Typical 45° shear failure is shown at point
(A) and stress whitening at (B).
An Instron universal testing machine was us~~ with constant

strain rates ranging from 0.002 in./min. (5.08 x 10 mm/min.) to 2.0
in./min. (5.08cm/min.). Adjustable grips were used for single lap
specimens to avoid excessive bending of the specimen). Tests were
performed at 12°F (22°C) and 50% relative humidity.
Surface Analysis
Scanning electron photomicrographs were obtained with an

Advanced Metals Research Corporation Model 900 scanning electron
microscope. Operating at 20kV, high magnification views (500X)
gave the details of surface features, while survey scans at 50X
provided a check on the distribution of representative features
that characterize the surface. Specimens were cut to approximately
0.25 x 0.25 in. (6.4 x 6.4mm) with a band saw, and fastened to the
SEM stubs with conductive, adhesive-coated copper tape. To enhance
conductivity, a thin ~200R) film of AulPd alloy was vacuum evapor-
ated onto the samples. Photomicrographs were taken with the sample
inclined to accentuate the appearance of small asperities.
In an attempt to quantify the lap joint deformation and

fracture processes, the roughness of 0.24 in. (6.25mm) of represen-
tative fracture surfaces was measured with a Talysurf 4 with 2~m
sensitivity, scanning at 0.72 in./min. (18.3mm/min.) stylus travel
speed and vertical magnification of 243X. The analog output was
conditioned through a Tektronix data processor that scanned the
output at 20 msec. intervals, producing 1.024 discrete data points.
The best fit to these points produced curves for comparison with
calibrated standards, thus enabling a quantitative measure of sur-
face roughness.
Talysurf profiles on representative sections of 21 fractured

single lap specimens, combined with photographs of the fracture
surfaces, support the qualitative findings from the SEM photomicro-
graphs. Further interpretation of the Talysurf profiles was
attempted by measuring the actual length of the fracture surface.
To do this, horizontal sections that are either inherent voids or
interfacial failure sites were subtracted first from the total
profile length. Inclination angles of all remaining segments of the
profile were determined and components were adjusted to the same
scale. The adjusted profile lengths thus obtained are assumed to
represent the fracture surfaces, compara ti vely. The amount of
deformation and fracture surface created is assumed proportional to
the fracture energy. Hence, the relative profile gives an idea
about the contribution of deformation and fracture to the ultimate
load.
Strength in Bulk
Tensile and shear measurements on the bulk material (3) showed

ra te dependence, and revealed that elastic stresses and strains
increased with an increase in strain rate. An empirical equation
for shear was derived from bulk tensile results using '( = 0/2 as
maximum shear stress and Hooke's Law. Figure 2 shows an excellent

fit using tensile constants and experimental data for yield
stresses and strains of the bulk shear tests, in a Ludwik type
equation,
y
s
= y'
s
+ '[' log (3:..)
0,
(4)
y
where the yield stress, ~, is the elastic shear strain rate and yl,
y , and ~ are experimentally determined material constants. Many
s
more details of the mechanical property testing can be found
elsewhere (3,19).
Uniformity of the shear data depended upon the load transfer
mechanisms. Metlbond 1113 specimens without tabs had yield
stresses comparable to the bulk tensile predictions using Tresca
failure criteria. Metlbond 1113 ,and 1113-2 specimens with tabs,
however, had higher values than calculated from tensile yield
stresses. The yield strains were higher than values calculated from
tensile yield strains (using y = (1 + V)e:) • These results show
that it is possible to obtain shear stresses and strains higher than
4.5 31
r-------.--------r-------,--------r-------~------_,
Sulk Tensile Prediction
Y = 29.06 + 1.90 Log(';'I';'? Mpa
y' = l%lsec
() Experimental Points from the Bulk
4.2 Shear Data
25
VI
.>< 3.9 27 ~
'"
Cl.
V) V)
V)
W
a:: ....
Vl
a::
l- I-
V) V)
- -
Cl Cl
....
-'
3.6 25 ....
-'
>- >-
3.3
1~0--4.-----~,------L~------'-r------~------~------~
INITIAL ELASTIC STRAIN RATE (%/sec)
Fig. 2. Rate dependence of strength, showing corre-

lation between shear data and a Ludwik-type
equation with material constants derived
from tensile tests.
150 DAVID W. DWIGHT ET Al.
predicted by tensile data. Perhaps the same applies to the bonded

state.
Comparison of the yield stresses in the bulk tensile and shear

forms reveals an interesting phenomena. In the bulk tensile form,
Metlbond 1113 specimens had higher yield stresses compared to the
Metlbond 1113-2 specimens. This was expected, since Metlbond 1113
had a carrier cloth increasing its stiffness and strength. In the
bulk shear form, however, Metlbond 1113-2 produced higher yield
stresses than Metlbond 1113. This appeared to be due to the
differences in tab application and load transfer mentioned earlier.
Also, it is possible that shear forces accentuate fiber/matrix
interfacial failure and do not have the benefit of fibers held in
tension.
Hart-Smith suggested (13) that the failure load for a bonded

joint is a function of the area under the stress strain curve
(toughness) of the adhesive. Its magnitude will depend on whether
the failure is brittle or ductile, and the latter case is an
increasing function of the strain rate. Toughness will increase
with the strain rate only up to the ductile/brittle transition, then
toughness will decrease. With bulk materials we found that failure
strains were smaller at higher strain rates.
The effect of carrier cloth on toughness is illustrated in Fig.

3 where the area under the tensile stress-strain curve is plotted as
a function of the elastic strain rate. Maxima in toughness appear
for both adhesive forms, but shifted from _10- 3% sec -1 for epoxy
alone to -10 -1 % sec -1 for the epoxy on carrier cloth. The high
fiber stiffness apparently contributes resistance to rate effects.
On the other hand, interfacial sites for crack initiation lower
toughness by 20% on the average. The maxima indicate ductile-to-
brittle transition phenomena, confirmed by fractography. In fact,
the fracture profile-length results also show a maximum at approxi-
mately the same strain rates.
Joint Strength
In the adhesive bonded state there was scatter in the ultimate

shear strength vs. strain rate for both single lap (Fig. 4) and
symmetrical specimens. We feel the scatter results from different
mechanisms acting simultaneously during deformation and failure of
the joint. In other words, joint preparation (inherent voids and
variable stress states), viscoelasticity (rate dependence), and
fractUre modes (brittle and ductile) may effect failure of the joint
at random. The variation of stress in the lap shear geometry, as
well as voids and fiber/matrix failures are important, but uncon-
trolled variables. Therefore, a complete study of bonded joints
requires determination of rate effects in the bulk, combined with
800
METLBOND
1113-2
400
220
METLBOND 200
1113
10- 6 10- 5 10- 4 10- 3 10- 2 10-1 10°

ELASTIC STRAIN RATE (%/sec)
Fig. 3. Area under the stress-strain curve as a

function of test rate. The presence of
scrim cloth in Metlbond 1113 lowers the
toughness and shifts the ductile-to-brittle
transition to higher rates.
fracture behavior in the bonded state. All of these results are

necessary for a general model with parameters representing bulk
properties as well as the constraints and fracture mechanisms
unique to the bonded state of adhesives. The following section
describes our use of SEM to qualify these features. Preliminary
attempts were made to quantify the fractographic data using surface
profil~ measurements.
HEAD RATE (SHEAR) (m/min)

10-3 10- 2 10-1
1054
2200
954
o o
o o o
1800
8 o 854 ~
o
.....:I
~ o o o ~
i o 754 ffi
o g
~
~~ o § o o 654 H
S
S
1400
o
o o o 554
10001~0-~3~---------10~-~2~----------1~0--~----------lJOO
HEAD RATE (SHEAR) in/min

Fig. 4. Illustration of the unpredictable rate
effects upon strength of Metlbond 1113
in single lap shear adhesive bonds. (Cf.
bulk behavior in Fig. 2).
FractographY
Visual and photographic examination at low magnification of

the single lap joints revealed different failure mechanisms
including plastic deformation and matrix/primer interfacial fail-
ure. Figures 5 and 6 juxtapose representative photographs of
fracture surfaces and Talysurf profiles that provide information on
the new surface area generated during deformation and fracture.
Profile lengths (after subtracting horizontal sections, assumed to
be inherent voids or interfacial failure sites) provide semiquanti-
tative estimates of the strength-affecting mechanisms, at least
10
z
...... 8
M
<:)
6
>< 4
w
u
z 2
c::(
l-
V)
0
......
Cl -2
--l
c::(
u
-4
......
I-
0::
-6
w
> -8
-10
0 4 8 12 16 20 24
HORIZONTAL DISTANCE x 10 2 IN.
Fig. 5. Optical pnotomicrograph and corresponding
Talysurf profile of lap shear specimen 111.
differentiating bonding and fracture areas from voids.

Figures 7 and 8 juxtapose two fracture surfaces with their
corresponding Neutron Radiographs. In Fig. 7 it is clear that a
large, initial void on the Radiograph is detectable on the fracture
surface. This particular void has an area of approximately 6
z
10
.....
8
M
0
6
x
w
4
u
z 2
c:t:
r-
.....
(/) 0
0
.....J
-2
c:t:
u
..... -4
I-
0:: -6
W
:>
-8
-10
0 4 8 12 16 20 24
HORIZONTAL DISTANCE x 10 2IN.
Fig. 6. Optical photomicrograph and corresponding

Talysurf profile of lap shear specimen #23.
percent of the overlap area. Other, smaller voids can be detected

on the NR photograph in a pattern similar to that on the fracture
surface. In contrast, Fig. 8 provides an example of the variability
of size, shape and distribution of initial voids, and there is a
correspondence to the defect structure apparent on the fracture
surface. We believe that Neutron Radiography (or similar NDT
inspection) must be developed to the point that quantitative esti-
mates of the relative defect structure can be made. Clearly,
Fig. 7. Neutron radiograph and corresponding

optical photomicrograph showing the same
void (arrows) before and after fracture.
Fig. 8. Neutron radiograph and corresponding opti-

cal photomicrograph showing a different
distribution of void volume than Fig. 1;
The larger inherent voids lead to larger
cavities on the fracture surface.
equations 1-3 require values for this term in order to calculate

failure energies or loads for the overall joint.
Fractography of the tensile failure (Fig. 9) shows that the
bulk adhesive fails with uniform horseshoe pattern of dimensions
between 3.9 x 10- 3 in., 0.1 mm and 2 x 10-2 in., 0.5 mm, typical of
rubber-toughened epoxies (10). At high magnification, interfacial
failure between the fibers of the carrier cloth and the matrix is
clear, as is a ductile appearance due to spherical indentations
(- 1.0jJll diameter) thought to coincide with rubber inclusions.
Generally, our SEM study of nine fractured Metlbond 1113

single lap specimens showed plastic deformation in low strength
joints (Fig. 10). On the other hand, Fig. 11 illustrates that high
strength joints were characterized more by brittle cleavage cracks,
matrix/primer interfacial failure, and visible fiber beds in the
matrix with fibers extending out of the surface.
Examination of a high strength joint (2000 lb. - 907 kg) shows
large, smooth interfacial failure sites connected with brittle
cleavage crack belts. Figure 11 clearly shows that voids (A) and
fiber/matrix interfacial failure (B) are dominant. We conclude
that the relatively smooth surface shown on Fig. 11 did not absorb
much energy in creating new surfaces by deformation and fracture.
Therefore, in this case, the high strength of the joint probably was
due to high strain rate. However, examination of another surface
(Fig. 12) shows that the high strength obtained (1836 lb. - 933 kg)
was due to high surface free energy of the many void edges and
brittle cracks. This example shows the possibility of obtaining
high strengths at low strain rates through increased fracture
surface area. Talysurf profilometer scans of the three single lap
specimens discussed above provided further insight into the quali-
tative findings of the SEM studies. The profile length (assumed to
be proportional to the failure area) was relatively low for Fig. 11,
but relatively high for Figs. 10 and 12.
Although we are unable to predict adhesive bond strengths, at

least we can explain the scatter in Fig. 4 and the differences from
bulk fracture on the basis of variable micromechanisms revealed by
SEM.
It should be noted, however, that the rate effects and the

fracture energies are not the only factors to affect the joint
strength. Inherent voids and flaws and chemical impurities at the
adhesive/adherend interface also had an effect on the joint
strength. In fact EDAX (Energy Dispersive Analysis of X-ray
Fluorescence) examination of some low strength specimens revealed
the presence of silicone at the adhesive/primer interface acting as
a strength reducing factor.
158 DAVID W. DWIGHTET AL.
Fig. 9. Two magnification of scanning electron photo-

micrographs of bulk fracture in Metlbond 1113.
Tiny cavities (arrow) and larger "horseshoe"
pattern derive from rubbery inclusions in the
epoxy matrix.
Fig. 10. Two magnifications of scanning electron photo-

micrographs of lap shear specimen #17. Note
the plastic deformation at point A at the
lower left.
Fig. 11. Two magnifications of scanning electron

photomicrographs of lap shear specimen 123.
Note the voids (A), primer/matrix interfacial
failure (B) and brittle cleavage cracks (C).
Fig. 12 . Scanning electron photomicrograph of lap

shear specimen #9. Notice how different
is the void distribution from the previous
two specimens.
Above cases were chosen carefully for they represented the

whole fracture. Stress concentrations in crucial sections such as
overlap edges and in between voids act as strength reducing factors.
Much effort is needed to understand the process and materials
interactions that govern these inherent flaws .
CONCLUSIONS
Strength vs. rate experiments on an epoxy adhesive, in bulk

(tensile and shear) produce accurate and predictable time depen-
dence . However, the tensile data does not predict the highest
possible shear stresses and strains that can be obtained from the
bulk shear specimens. The bulk shear specimens constrained with
tabs produced about 60% higher yield stresses compared to the bulk
tensile predictions . The predic tions by the data of the bulk shear
specimens with tabs, however, is believed to be valid for joints
with ideal bond conditions (i . e . no inherent voids or flaws, strong
adhesive/adherend interfacial forces without chemical impurities,

uniform stress state, no residual stresses, and rate effects domi-
nant over fracture effects). In fact, fractographic study with
Neutron Radiography, optical and scanning electron microscopy
(SEM), and surface profiles showed the bonded state of the adhesive
was dominated by new constraints and random, inherent flaws and
stress states, leading to scatter in the lap shear data.
Comparison of the bulk tensile, bulk shear and bonded shear

data shows that only the elastic behavior is common to all three
modes. In the bulk and especially bonded shear states, however, new
stress states and fracture initiation processes appear to be the
dominating factors.
Therefore, characterization of bonded joint should include a

careful study of structure-process relationships to assess the
presence and effects of inherent flaws and voids, interfacial
chemical and physical state and possible residual curing stresses.
This information, combined with quantitative fractography and
viscoelastic modeling of the bulk adhesive, should make it possible
to characterize the bonded state accurately.
ACKNOWLEDGEMENTS
This work was supported by the NASA Langley Research Center

under NASA Grant NGR 47-004-090. A research associateship (E.S.) is
gratefully acknowledged.
REFERENCES
1. D.W. Dwight, J. Colloid and Interface Sci. 59 (3), 447 (1977).
2. D.W. Dwight, MoE. Counts and J.P. Wightman,in "Colloid Inter-
face Science, V. III. Adsorption, Catalysis, Solid Sur-
faces, Wetting, Surface Tension and Water", M. Kerker,
ed., Academic Press, New York, p. 143 (1976).
3. H.F. Brinson, M.P. Renieri and C.T. Herakovich, in "Fracture
Mechanics of Composites", ASTM, STP 593, p. 177 (1975).
4. M.J. Crochet, ~ ~ Mech. 33, 326 (1966).
5. P.M. Naghdi and S.A. Murch, ~ ~ Mech. 30, 321 (1963).
6. P.G. Ludwik, Elemente der Technologischen Mechanic, J.
Springer, Berlin, p. 9, 1909.
7. C.J. Lin and J.P. Bell, ~ ~ Polym. Sci. ~, 1721 (1972).
8. R.J. Good, ~ Adhesion~, 133 (1972).
9. L.W. Jennings, in "Recent Advances in Adhesion", L.H. Lee,
ed., Gordon and Breach, New York, p. 469 (1971).
10. W.D. Bascom and S. Mostovoy, in "Organic Coatings and Plastics
Chemistry", S.S. Labana, ed., Vol. 38 (1), American Chem-
ical Society, p. 152.
11. E. H. Andrews and A. J. Kinloch, .!h Polym. Sci.: ~ No. 46,

(1974) •
12. R.A. Gledhill and A.J. Kinloch, J. Adhesion 6, 315 (1974).
13. L. J. Hart-Smith, in nAir Force-Conference -on Fibrous Com-
posites in Flight Vehicle Design", Dayton, Ohio, Sep-
tember 1972.
14. D. Kutscha, Forest Products Laboratory Air Force Contract
Report No. ML-TDR-64-298, December 1964.
15. W.J. Renton and J.R. Vinson, ASTM Composites Reliability Con-
ference, STP-580, April 1974.
16. T.R. Guess, R.E. Alred and F.P. Gerstle, Jr., J. Test Eval. 5,
84 (1977). ----- -
17. W.E. Dance, D.H. Petersen, .!h ~ Polym. Sci.: ~ Polym.
~ 32, 399 (1977).
18. J.M. Whitney, D.L. Stansbarger and H.B. Howell, J. Composite
Mater. 2, 24 (1971).
19. E. Sancaktar and H.F. Brinson, This Volume.
Dielectric Relaxation Gradients in an
Adhesive Bond
John L. Crowley and Arthur D. Jonath

Lockheed Palo Alto Research Laboratory
3251 Hanover Street, Palo Alto, California 94304
ABSTRACT
The dielectric relaxation of a filled epoxy adhesive bond has
been measured as a function of the distance from one aluminum
adherend by means of dielectric depolarization spectroscopy. Four
thermally stimulated current peaks have been identified and charac-
terized using for each a single activation energy and relaxation
time approximation. A gradient in the magnitude of the peaks is
observed as one approaches the adhesive-adherend interface. The
difference in shape and magnitude of the curves can be explained in
two ways: (1) differences in the chemical composition of the
adhesive at the center of the bond and at the interface arise during
the curing process; or (2) dipoles used in the bonding process near
the interface are not free to respond to th~4electric field. The
gradients occur over distances up to 2.5 x 10 m. This is contrast-
ed with the typical range of tens of Angstroms for interfacial bond
forces involving dipole or van der Waals interactions.
INTRODUCTION
Aluminum to aluminum epoxy bonds are in widespread use in many
industries. The bond strength depends nqt only upon the mechanical
properties of the resin but also upon the surface preparation of the
adherend, the curing cycle of the resin, the ambient atmosphere and
several poorly understood aging characteristics. In addition, it
has been shown (1) that the locus of failure in an aluminum epoxy
bond depends upon the state of stress within the bond. For failure
in Mode I (opening) the fracture surface is most often observed to
occur at the center of the bondline, while the fracture surface in
165
166 JOHN L. CROWLEY AND ARTHUR D. JONATH
Mode II (shear) failure is characteristically found within the

adhesi ve ve.ry near the adhesi ve-adherend interface. The observa-
tion of such failure led to the investigation of possible structural
gradients in the cured resin within the interfacial region. Both
mechanical and dielectric relaxation methods were used to inter-
rogate the adhesive material properties within the adhesive-adher-
end interfacial region. The results of the thermally stimulated
dielectric relaxation (TSDR) measurements will be discussed in this
paper.
THERMALLY STIMULATED DIELECTRIC RELAXATION
The technique of TSDR, first used to study the impurity defects

in alkali-halide crystal (2), has been used to probe the micro-
structural relaxations in an aluminum-epoxy bond. A recent review
of its application to polymers has been given by Hedvig (3). The
TSDR experiment measures a thermally stimulated current in a di-
electric material which has been polarized by means of an externally
applied d.c. electric field. The measured current can be due to the
release of trapped mobile charge, relaxation of induced dipoles or
to the movement of polar side groups or chain ends in the polymer.
To describe what is happening in the experiment, it is assumed

that the thermally stimulated current is due to the reorientation of
dipoles. The sample is cooled to immobilize the dipoles in the
applied field orientation. The electric field is then removed and
the sample is connected to a sensitive current measuring instru-
ment. Next, the sample is heated and the dipoles begin to thermally
randomize. This decay of the internal field due to oriented dipoles
induces a current in the external circuit. The resultant cur-
rent/temperature thermogram can be analyzed for dipole density,
activation energy, and relaxation time. The current peaks recorded
in this way are found to correlate well with the transition tempera-
tures of the polymer as measured by mechanical relaxation or
conventional (a.c.) dielectric spectroscopy (3).
The motion of dipoles is assumed to be thermally activated.
The relaxation time(s) for dipole motion, '[(T), can therefore be
described by an Arrhenius relation, '[ (T) = '[ exp(E/kT), where '[
is the extrapolated relaxation time at infin~te temperature, E i~
the acti va tion energy for dipole motion and k is Boltzmann's
constant. The reorientation is carried out at a temperature at
which the relaxation time, '[(T), is much less than the time, t , for
which the electric field is applied. After reorientatio~ and
immobilization of the dipoles, the induced current caused by their
thermal randomization can be expressed by the following relation-
ship (4):
I
DIELECTRIC RELAXATION GRADIENTS 167
NdFp -1 -1
i(T) = ~2 [T 0 exp(E/kT)] exp
T
-[ [bToexp(E/kTI)]
where F is the applied electric field, p is the dipole moment, Nd is

the density of dipoles, b is the heating rate, and T is the
temperature at which the dipoles were polarized. These Rata were
evaluated with this equation using the solution provided by Jen-
kins (5).
EXPERIMENTAL
The epoxy used in this experiment was Epoxylite 810 (Epoxylite
Corporation, Anaheim, California), a silicate loaded, dianhydride-
cured glycidyl ether of a phenolformaldehyde Novolac resin. TSDR
and mixed-mode fracture samples were prepared with resin for bond-
ing to aluminum surfaces which were first polished with 111m diamond
polish and subjected to a common acid-dichromate etchant. After
bonding, the resin was cured at 120°C for two hours. Fracture
characteristically left a thin film of material or interfacial
accommodation zone (IAZ) on one aluminum adherend and a bond
thickness (bulk) of adhesive on the other. After fracture, addi-
tional TSDR samples were cut from the adherends. An aluminum
electrical contact, 0.635 cm in diameter, was evaporated onto the
adhesi ve surface of each specimen, while the aluminum substrate
served as the second electrical contact. All samples discussed here
were polarized at 25°C for 20 minutes.
Sample heating rates were varied from 0.12 deg K/s to 0.50 deg
K/s. These rates, although higher than usual for tests on polymers,
were still sufficiently slow to maintain thermal equilibrium be-
cause the thin film under measurement was bonded to a very good
thermal conductor.
The TSDR thermograms of the bulk and IAZ material appearing in

Fig. 1 indicate that there are certain differences in the dielectric
relaxation behavior of the adhesive depending upon location within
the bond. First, there appears to be a general difference in the
density of dipoles contained in the bulk and in the IAZj i.e., the
TSDR curve for the IAZ materials is lower-valued than that for the
bulk material. Secondly, the data for the bulk material show peaks
C and D appearing in the size ratio of 3:1, while in the IAZ data,
this ratio is at best 1:1. Two explanations for the difference in
shape and magnitude of the curves are offered: (1) differences in
the chemical composition of the adhesive at the center of the bond
BUl K MATERIAL
I .... - ,
,I
/ ' B
"
---"
\
\
\ "
\
\ '\
\
\ \
\ \
\ \
\ \
\ '\
\
\
R. T. IAZ MATERIAL
10- 15
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5
103/T (K- l )
Fig. 1. The Thermally Stimulated Dielectric Current of

Epoxylite 810 Bulk and IAZ Samples, Polarized
with 9.55 x 10 2 V/cm at 2r:J> C for 20 Minutes.
Four resolvable peaks are indicated as A, B,
C, and D.
and at the interface may arise during the curing process; or

(2) dipoles used in the bonding process near the interface are not
free to respond to the applied electric field.
Of the four discernible peaks in Fig. 1, the magnitude of three

peaks, B, C, and D, were found to vary linearly with applied
electric field. The current maximum for peak A increases at a

sublinear rate with increase in field. It is also observed that the
peak A maximum shifts in temperature as the polarization tempera-
ture is changed. Therefore, it does not appear that peak A is due
to dipole reorientation. Rheovibron measurements of the loss
tangent made on sheet cast samples of the epoxy were able to
identify only one peak in the temperature range of present interest.
This peak appears to correspond to peak C in Fig. 1.
Figures 2, 3, and 4 show the results of the TSDR measurements
on epoxy samples whose thicknesses were sequentially decreased by
metallographic sectioning techniques. The magnitude of the peak
o 0 o
______ 4-
1.0 ,..----
,,0 o
/
0.9 I II
I-
o
Z O.B
~
Of!
::::> I
u 0.7
~,ll
,,
~
~
~
0 0.6
w
!:::! I
-'
« 0.5 I
~ I
0 I
Z 004 I
0.3 II SAMPLE 1
o SAMPLE 2
0.2 • IAZ POST-FRACTURE SAMPLE
0.1
O~--~--~--~----~--~ __ ~L-__~__- L__~

o 0.01 0.02 0.03 0.04 0.05 0.06 O.W 0.08 0.09
ADHESIVE THICKNESS (em)
Fig. 2. Magnitude of Peak B as a Function of Thickness

of Material Remaining on Adherend (Normalized to
a Peak Current of 4.75 x 10- 13 A at a Field of
1.59 x 103 V/cm).
currents have been normalized to a constant field and average peak

magnitude. Because the polarization temperature and heating rate
were the same for all tests, the graphs essentially represent the
variation of the average dipole density in the epoxy as a function
of distance from the adherend. The density of dipoles responsible
for peaks Band C is seen to remain constant throughout th: 4bulk of
the material and to decrease at distances less than 2 x 10 m. The
data for peak D do not justify a straight line approximation in the
bulk of the material, but its density is also seen to decrease as
the adhesi ve-adherend inter face is approached. This resul t by
itself is not surprising; it has long been known that bulk proper-
ties differ from surface properties; i.e., any material must accom-
A
o
1.0
,/
,0-----------
0
-li-
0.9 ~ 0 0
I-
Z 0.8
w
'"
'"::> 0/
,
I
0
u
looi: 0.7 A
/
~
Q"
I
0
w 0.6 I
!::!
..... I
0.5
~0
Z 0.4
0.3 t::. SAMPLE 1

o SAMPLE 2
0.1
o
o 0.01 0.02 0.03 0.04 0.05 0.06 0.C17 0.00
ADHESIVE THICKNESS {em}
Fig. 3. Magnitude of Peak C as a Function of Thickness
of Material Remaining on Adherend (Normalized to
a Peak Current of 2.0 x 10- 13 A at a Field of
1.59 x 103 V/cm).
A
D D
1.0 /'
----- -
D
D/
0.9 D
/
l- I
Z 0.8 D
~ A
or:
:::>
/
u 0.7 /
~
DA
~
a.. /l:l.
0 0.6
w
!:::!
.....
0.5
~
0
Z 0.4
0.3 A SAMPLE 1
0 SAMPLE 2
0.1
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.(17 O.IJI 0.09
Fig. 4. Magnitude of Peak D as a Function of Thickness
or Material Remaining on Adherend (Normalized to
a Peak Current of 2.7 x 10-13 A at a Field of
1.59 x 103 V/cm).
modate itself to molecular (atomic) bond discontinuities caused by
surfaces. Rather, it is the measure of the spatial range of the
accommodation zone, approximately 20~m, that is of interest. This
is contrasted with the tens of Angstroms typical range for inter-
facial bond forces involving dipole or van der Waals interactions.
A long-range mechanism, perhaps in conjunction with the short-range
phenomena, is needed to explain the result. For example, if the
polymer contains interacting dipoles, ordering due to interfacial
electric fields might propagate a significant distance from the
interface. Also, electrokinetic phenomena associated with col-
loids (6), postulated to explain the ordering of floccules seen in
two-phase epoxy polymers (7), might describe long-range effects
taking place in the epoxy cure process.
In addition to the TSDR measurements, an independent study was

made of the shear wave speed, C , in the adhesive as a function of
the bond thickness (8). Sample\ for this study were identical to
those used in the TSDR measurements. The shear wave speed is given
by the expression
Cs =fGIP
where G is the shear modulus, and p is the adhesive density. By
differentiating the above expression, the change in shear speed can
be expressed as
l1Cs _ 1 (l1G _ ~)
C - 2 G P
s
The results of the shear wave speed measurements are shown in
Fig. 5. The data show a variation in shear speed over the same
magnitude of bond thickness as the variation in normalized peak
current amplitudes shown in Figs. 2, 3, and 4. However, the
variation in the shear speed is 10 percent of the original speed
measured in the bulk, while the variation in the magnitude of peak B
at comparable bond thickness is approximately 50 percent. If it is
assumed that the variation in the magnitude of peak B is due
entirely to a change in the density of the adhesive in the IAZ, then
the change in shear speed represents a 70-percent decrease in the
shear modulus. Although this reduction in shear modulus is consis-
tent with the observed locus of failure for Mode II fracture,
density gradients alone do not account for the change in ratio of
peak magnitudes (see Fig. 1).
The normalized peak currents from the bulk materials shown in
Fig. 1 are a factor of three greater than the asymptotic values
shown in l"igs. 2, 3, and 4. This anomaly remains unexplained.
However, it is noted that the bulk fracture material underwent
severe deformation and microcracking.
SUMMARY
The dielectric relaxation of a filled epoxy adhesive has been
measured as a function of the distance from an aluminum adherend by
means of thermally stimulated dielectric relaxation. Four ther-
mally stimulated current peaks have been identified and character-
ized. A gradient in the magnitude of the peaks is observed as the
adhesive-adherend interface is approached. The difference in shape
and magnitude of the curves can be explained in two ways: (1) dif-
ferences in chemical composition of the adhesive at the center of
the bond and at the interface may arise during the curing process;
or (2) dipoles used in the bonding process near the interface are
not free to respond to the electric field. The gradients in the
1.6
e_e 0
~
~
1.5
V>
v
Cl
w
w
CL
V>
§
0::
<t:
w
:I:
1.4
V>
o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

REMAINING ADHESIVE THICKNESS (em)
Fig . 5. Shear Wave Speed Variation Within an Adhesive

Bond Thickness Measured Using Rayleigh Wave
Method (20 MHz). (See Ref. 8.)
dipole density occur over distances up to 2 x 10 -4 m.
ACKNOWLEDGEMENTS
This work was supported by Lockheed Missiles & Space Company,
Inc . Independent Research funds. The authors wish to acknowledge
the support of R. Holsinger and T. J. Stultz in preparing the samples
for test.
REFERENCES
1. W.D. Bascom, C.O. Timmons, and R. L. Jones, Apparent Inter-
facial Failure in Mixed-Mode Adhesive Fracture, J. Mate-
rials Sci. jQ, 1037 (1975). ---
2. C. Bucci, R. Fieschi, and G. GUidi, Ionic Thermocurrents in
Dielectrics, Phys. Rev . 148, 816 (1966).
3. P. Hedvig, "Dielectric Spectroscopy of Polymers", John Wiley
and Sons, New York, pp . 148-155 (1977) .
4. C. Bucci, Ionic Thermocurrents in Alkali Halide Crystals Con-
taining Substitutional Beryllium Ions, Phys . Rev . 164,
1200 (1967).
5. T.R. Jenkins, On Computing the Integral of Glow Curve Theory,

~ Compo Phys. 28, No.2 (1918).
6. A.W. Adamson, "Physical Chemistry of Surfaces", Interscience,
New York, pp. 119-226 (1960).
1. R.E. Cuthrell, Epoxy Polymers. II. Microstructure, ~ ~
Polym. Sci. 1£, 1263 (1968).
8. G.C. Knollman and J. J. Hartog, Lockheed Palo Al to Research
Laboratory, Palo Alto, CA (private communication).
New Approach to the Understanding of
Adhesive Interface Phenomena and
Bond Strength
Arthur D. Jonath
Lockheed Palo Alto Research Laboratory

Palo Alto, California 94304
ABSTRACT
Microscopic and macroscopic aspects of adhesion science are
currently under investigations in order to better understand the
fundamentals of adhesive behavior from bond formation to bond
deterioration/failure. Results of three novel experimental methods
are described: a) Measurements using an improved shear specimen
design indicate the absence of Mode II failure in a brittle epoxy
system. b) Results of ultrasonic Rayleigh wave measurements
indicate a possible mechanical properties gradient in the cured
resin within the interfacial region of an epoxy-aluminum bond. c)
Thermally stimulated dipole relaxation measurements indicate struc-
tural gradients within this region.
INTRODUCTION
Adhesive bond failure has been associated with missile mal-
functions in various aerospace systems. In most instances, the
contributing factors are not quantitatively identified. The prob-
lem is to determine the extent of failures resulting from improper
·This is an abridged version of the paper "Some Novel Approaches

to the Study of Adhesive Interface Phenomena and Bond Strength"
presented at the Structural Adhesives and Bonding Conference,
Technology Conferences Associates, P.O. Box 842, El Segundo,
Ca., March 13-15, 1919.
175
176 ARTHUR D. JONATH
adhesive selection, interface preparation, adhesive application, or

changes in adhesive characteristics due to age and environment. The
empirically established procedures for surface preparation prior to
application of adhesive do not guarantee reproducible quality
bonds, and tests for the achieved bond quality are generally not
related to the state of stress existing in an actual component.
Understanding the aging processes in adhesives is especially criti-
cal to the aerospace industry because debonding of insulation and
structural materials limit the reliability and lifetime of adhe-
si ve bonded systems such as solid propellant motors. Thus the
overall objective of the research program discussed herein is to
provide guidelines for selecting adhesive systems and surface con-
ditioning requirements and to predict the age environment behavior
of adhesive systems. To accomplish this, several novel experimen-
tal methods are being investigated for studying bonding/interface
phenomena and bond-joint failure processes. They include
o A novel shear test technique investigating assumptions
used in adhesive fracture toughness measurements.
o An ultrasonic surface wave method for measuring mechani-
cal properties in thin adhesive layers.
o A dipole relaxation method for experimental determination
of adhesive structural changes at adhesive/adherend
interfaces.
The purpose of this paper is to introduce the adhesives
researcher and applications engineer to these experimental con-
cepts, some of which are adapted from other fields, and to indicate
where results of such measurements can be of practical use.
FRACTURE MECHANICS
Fracture toughness measurement is one of the materials con-
cepts presently used to determine design data for adhesive bonded
structures. This is considered to be a measure of the ultimate
strength of a material in which the failure mode is by crack
propagation, causing failure at engineering stresses less than the
tensile or shear strengths. While fracture toughness is the
appropriate failure criterion for linear elastic brittle failure,
recent work at the Naval Research Laboratory and elsewhere has
highlighted several pitfalls in applying these methods to the
testing of adhesive joints (1). For example, joint geometry and
bond thickness effects dominate the fracture toughness measurements
in structural adhesive joints. Thus the approach for this aspect of
the study was to explore, in selected adhesives, the limitations of
fracture mechanics and to attempt to relate fracture energy
measurements to the adhesive polymer structure.
UNDERSTANDING OF ADHESIVE INTERFACE 177
Before discussing the novel adhesive shear test method, let us

describe the fracture toughness measurement results which led to
our need for measuring "pure" shear in an adhesive bond. Both
silicate filled and unfilled versions of the Epoxylite 810, a
dianhydride cured glycidyl ether of a phenolformaldehyde Novolac
resin were tested in the tapered double cantilever beam (TDCB) and
45 0 scarf-joint configurations shown in Fig. 1 to measure Mode I
(opening) and Mode II (shear) fracture toughness. The aluminum
adherend surfaces were prepared with dichromate acid etch. Table 1
lists the results. The Mode I fracture energy values, GIc ' are in
the range expected for unmodified epoxies. GIc in the filled system
is greater than that in the unfilled system implying that the
silicate filler contributes somewhat to toughening this epoxy.
In either filled or unfilled versions of this adhesive, the

opening-mode adhesive fracture energy GIc is nearly an order of
magnitude greater than the combined-mode GI,II value. This is
contrary to the results (G I IrGIc) generally found for the
fracture of isotropic materials'(2) and more like the results being
found for elastomer-modified and commercial structural adhesives
(3),
The locus of failure in the TDCB specimens precracked at the

center-of-bond (COB) was observed consistently to follow the COB.
In the scarf-joint samples, the failure occurred very near the
adherend-adhesive interface whether the precrack was at either the
COB or the interface. Although there is typically less than 50~m of
adhesive remaining on the adherend nearest the failure locus, the
failure is clearly cohesive in nature. In fact, adhesive failure
(no detectable adhesive remaining on an adherend) was never
observed in specimens where proper adherend surface preparation and
handling procedures are followed.
The questions arising are thus: Why in the presence of shear

stress does the failure locus follow a path so close to the
interface? Is it simply the goemetry of the scarf-joint which
drives the propagating crack to the interface? Is there an
asymmetric stress concentration about the crack tip forcing this
path? Or perhaps there are material properties changes (such as in
shear modulus) occurring within the bondline thickness close and
normal to the adherend surface? To answer these questions we sought
to understand from both microscopic and molecular views what is
happening in the interfacial accommodation zone (IAZ), this region
very close to the adherend through which the crack propagates when
significant shear stresses are present in the failing joint. More
about this later.
178 ARTHUR D . JONATH
CONTOURED TO
p 30 2 ADHESIVE
• -;:-r
B
•p
o.
1
t---+-- I IN .
(2.54 em)
12 IN.
(30.5 em)
1.5 IN . + I
~2IN.--l
(5.08 em)
b.
Fig . 1. Adhesive Fracture Test Specimens:
(a) Tapered Double Cantilever Beam for Mode I
and (b) 45 0 Scarf-Joint for Combined Modes I
and II.
Table
RESULTS OF FRACTURE TOUGHNESS MEASUREMENTS
Relative G KI KII
Crack Total
Length Nm/m2 in-lb/in2 N/m312 N/m3/2
FILLED SYSTEM
I G II
--=.c::::.
SCARF •1 93 0.53 2.08 x 10 6 1.54 x 10 6
.2 100 0.57 2.32 x 10 6 1.35 x 10 6
.4 110 0.63 2.47 x 10 6 1.28 x 10 6
Glc
TDCB .022 700 4.0 7.1 x 10 6

.045 700 4.0 7.1 x 10 6
•1 700 4.0 7.1 x 10 6
UNFILLED SYSTEM
G II
I
--=.c::::.
SCARF .2 51 0.29 1.66 x 10 6 9.63 x 10 5
.3 49 0.28 1.66 x 10 6 8.96 x 10 5
.4 39 0.22 1.49 x 10 6 7.75 x 10 5
Grc
TDCB .05 400 2.3 5.4 x 10 6

•1 320 1.9 4.8 x 10 6
First let us return to the fracture toughness results in Table

for both the filled and unfilled versions of the Epoxylite 810.
The critical stress intensity factor KII , calculated from the
scarf-joint data using the finite element displacement method of
Trantina (4), decreased as crack length increased. Since KII is
expected to vary with crack tip radius and not crack length, the
existence of Mode II fracture therefore became suspect. The concern
is that failure is shear might be determined by the shear strength

in the remaining ligament material and not by propagation of a shear
mode crack.
To test this hypothesis a modification of the Iosipescu speci-

men (5) used to measure shear in metals was adapted (6) for
measuring "pure" shear in an adhesive bond. This specimen with
simplified shear and moment diagrams is shown in Fig. 2. The
computed results of a finite element analysis (7) of the shear and
peel stresses along the bond length for this shear specimen is shown
in Figs. 3A and 3B respectively. In this case, the applied load is
1.0 lb. and the total bond length is 1.0 in. Young's mogulus and
Po~sson's ratio for the aluminum adherends are 10.67 x 10 psi and
0.3368; 2.65 x 10 4 psi and 0.32 were used for the adhesive. Note
that for this specimen the shear stress exceeds its average value by
less than 3% in magnitude for less than 5% of the bond length. Peel
stresses vary more than this, but they are two orders of magnitude
less than the shear stress. A typical lap joint calculation is
included in Figs. 4A and 4B for comparison (8). The differences in
a) uniformity of shear stress along the bond and b) magnitude of
peel stresses between the modified Iosipescu sample and the lap
shear are striking.
In Fig. 5 the adhesive shear strength, as measured (6) in the

above described shear specimen, is plotted as a function of the
normalized crack length. The bondwidth is 1.0 in. and thickness is
0.020 in. The l~ne in the figure represents the predicted behavior
if failure is determined by shear strength. Values above the line
at large crack lengths are probably volume effects and are being
investigated. Systematic deviation of the data to values below the
line implies the presence of Mode II. The data represented by open
triangles (center of bond moment = 0) are a strong indication that
Mode II is absent and that the failure is determined by the shear
strength of the remaining material. The measured shear strength
using this specimen is 4235 psi. The adhesive manufacturer reports
less than half this value measured in lap shear.
The solid triangles in Fig. 5 represent failure stresses

resulting when the zero moment position was deliberately moved
0.025 in. away from the centerline of the bond. The effect is
drastic--introduction of Mode I failure into the specimen. It is
evident that adhesive joint strength depends critically on the type
of loading experienced by the jOint.
As 1n the scarf-joint speciments, the failure locus in this

shear specimen lies within the IAZ. The symmetry of the specimen
and its loading tends to rule out a geometrical argument for this
behavior. Crack tip asymmetries could not be detected micro-
scopically. Thus we turned our attention to search for variations
~
LOADTRANSFER~~~~~~~~~~~~~~~3
ROD (TYP. )
~----------b----------~
P (b - a )/( b + a)
Pa/( b + a} SHEAR DIAGRAM
~ Pa (b - a}/12 (b + all
MOMENT DIAGRAM
Fig. 2. Adhesives Shear Test Specimen with Simplified

Shear and Moment Diagrams.
in material properties and changes in material structure within the

IAZ.
MECHANICAL PROPERTIES OF THIN LAYERS BY ULTRASONIC RAYLEIGH WAVES
Elastic waves similar to surface gravitational waves can be

transmitted by an elastic medium, as was first shown by Lord Ray-
leigh. Their particle motions decrease rapidly with depth; their
speed, which depends on the value of Poisson's ratio, is always less
than that for a shear wave.
0.5 A
0.4
iii
.....,
a..
V!
0.3
V!
....~
V!
Ill::
< 0.2
w
:I:
V!
0.1
o~--~~ __~__~~__~____~
o 0.1 0.2 0.3 0.4 0.5
JOINT LENGTH (IN.)
*10- 3 6.0
B
4.0
iii
a..
2.0
V!
V!
w
....
Ill::
V! 0
...... 0.1 0.2 0.4 0.5
w
w
a.. JOINT LENGTH (IN.)
-2.0
-4.0
Fig. 3. Stress Distributions from Half-Length

Finite-Element Analysis for A.) Shear
and B.) Peel in Shear Specimen Shown
in Fig. 2.
6r-------------------------------------,
6 8 10
DISTANCE ALONG OVERLAP
mm
Q
g 6r----------------------------------------"
....
.
5
!:: B
Z
;:) 3
~
IU
2
... E 1
IIIIi: N
lit E
~?" 0 4 6 ~~8----~
~ 2-1 10
oZ DISTANCE ALONG OVERLAP
IU
mm
~
-3
IU
-4
~
Z -5
...~ -6~---------------------- ________________ ~
Fig. 4. Stress Distribution from Full Length

Double-Lap Finite-Element Analysis (8).
A. Shear Stress; B. Transverse Normal
Stress (Peel).
CENTER OF BOND
MOMENT:
4
6 0
• = 0
1-
O~ __ ~ __ ~ __- L__ ~ __~____L -_ _~_ _~_ _- L_ _~

o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
CRACK LENGTH, 2 (/l
Fig. 5. Shear Strength Versus Crack Length for

Epoxylite Using Adhesive Shear Test
Specimen Shown in Fig. 2 (Ref. 6).
When a solid is immersed in a liquid and an ultrasonic wave

impinges at some angle onto the solid from the liquid, several wave
modes may be generated in the solid by mode conversion. That is,
energy in the incident mode converts to another mode at the expense
of the reflected wave. The critical angle for a surface mode is the
angle of incidence which produces an angle of refraction of 90 for
tha t mode. According to Snell's law, the refraction angle is
related to the angle of incidence and the speeds of sound in the
liquid and in the solid.
CR Ci
= (1)
sin 90 0 sin aiR
where Ci is the incident sound speed in the liquid, CR is the
surface wave speed in the solid, and aiR is the angle of incidence
at which the surface mode conversion occurs. The shear wave speed,
CS ' is related to CR by
where k is a function of Poisson's ratio. The shear modulus, G, is

a function of Cs and density p
G = pCs2 (3)
Most of the energy in the Rayleigh wave is concentrated within a
distance less than a wavelength from the solid surface and thus the
shear wave speed in thin adhesive layers (such as that adjacent the
cohesive fracture surface in the IAZ) can be determined.
A simple laboratory arrangement for measuring the Rayleigh

angle appropriate for a thin layer of adhesive deposited on a
substrate is shown in Fig. 6. A cylindrical glass vessel is mounted
on a spectrometer table so as to be rotatable about its longitudinal
axis (normal to the plane of the figure). This vessel is filled
with a liquid, and the test specimen is positioned at the axial
center of the vessel as shown. Thereby, the specimen can be rotated
about its longitudinal axis by rotation of the spectrometer table.
A 20 MHz ultrasonic transducer is positioned as shown in the figure.
The stainless steel mirror allows the single transducer to be used as
both acoustic transmitter and receiver. By adjustment of the sample
and the transducer post the transmitted wave angle of incidence is
kept equal to the angle of reflection as the incidence angle is
varied. Then the received single displays an amplitude minimum and
a phase reversal at ~, the Rayleigh angle.
1.
The apparatus was evaluated by determining shear speed in

various materials. For metals immersed in water the reflection
minimum was used to determine a ~ and thus CS . The agreement with
published shear speeds was typically +3%, -0%. The relative error
for repeating a measurement (sensitivity) was less than 0.15%.
However, for typical polymer materials immersed in water. there is
no surface generation mode, Le. Ci > CR (see Eq. 1). Therefore,
Dow Corning Fluid DC-200A (of 1cs. viscosity) was selected as ftn
immersion medium and phase reversal was used to determine a i.
Agreement was +8%, -0% suggesting a systematic experimental error,
perhaps in the determination of the incident sound speed in the
PANAMETRIC
PULSER
MODEL
o
TECTRONIX
SCOPE
5052 PR MOD EL 545
SAMPLE
PANAMETRIC TRANSDUCER TEST VESSEL

TYPE V317. 20 MHz. D = 0.25 IN. MOUNTED ON
TABLE OF
ADJUSTABLE POST SPECTROMETER
TYPE 2300
Fig. 6. Arrangement for Measurement of the Rayleigh

Angle.
liquid. But the relative error of the measurement was found to be

approximately 1%.
Samples consisting of 0.03 in . thick Epoxylite 810 bonded to

single aluminum adherends were prepared. Figure 7 shows the results
of Rayleigh wave shear speed measurements (9) on these epoxy samples
whose thicknesses were decreased by metallographic sectioning tech-
niques. For adhesive thicknesses greater than 10011m, C is asympto-
tic (within 2 percent) to the bulk cast value. At ~m from the
aluminum adherend, C has decreased by about 11 percent, to 1.4
s
km/sec., the values measured in the 500lJm and thicker postfracture
specimens. The latter specimens were fashioned from scarf-joint
test specimens whose fracture surfaces were located within the IAZ
about 1311m from the adhesive/adherend interface. Since the surface
wave displacement is significantly reduced at depths greater than
20lJm it is not surprising that C values measured for the IAZ
surface of these post-fracture spescimens match those (points on
extreme left in figure) for samples whose thickness was reduced to
2511m by polishing. Computational models for stress-strain distri-
bution or crack propagation in adhesive bonds assume constant
mechanical property values throughout the bond thickness. Yet
density gradients are probable in polymers adjacent to interfaces.
1.6
e_e 0
"
~
E
...Yo
1.5
VI
u
0
w
w
Q..
VI
§
or::
4:
w
I
1.4 0 0
VI
o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

REMAINING ADHESIVE THICKNESS (em)
Fig. 7. Shear Wave Speed Variation Within an Adhe-

sive Bond Thickness for Samples 2 (0) and
5 (I) and the Post-Fracture Sample (CJ)
Measured Using Rayleigh Wave Method
(20MHz) (Ref. 9).
These are most likely caused by enhanced diffusion towards surfaces

of the low molecular weight portion of the polymer size distribu-
tion. The data for this adhesive/adherend system indicate exis-
tence of a shear speed gradient extending 100~ from the interface
into the adhesive bond. Relative contributions of shear modulus and
density to this property gradient (see Eq. (3» have not been sepa-
rated and must be determined before any conclusion can be drawn
regarding modulus variation effect on adhesive joint failure locus.
However, that this gradient region contains the failure surface
when shear is present indicates a necessity for reexamination of
concepts based upon homogeneity of adhesive properties throughout a
bond.
THERMALLY STIMULATED DIPOLE RELAXATION

In order to afford a molecular view of behavior in the inter-
facial region, the experimental technique of thermally stimulated
dipole relaxation (TSDR), originally used to study defects in
alkali halide crystals (10), was applied to probe the microstruc-
tural relaxation in the aluminum-epoxy system (11). The TSDR
experiment involves the reorientation of dipoles in a dielectric
material by means of an externally applied electric field. Details
of the experiment are given by Crowley and Jonath elsewhere in this
volume.
Typical TSDR thermograms for IAZ and bulk samples are shown in
Fig. 8. In obtaining these data, the samples were heated to 235 0 K
yielding peaks D and C and then quickly recooled to 125 0 K. Peaks B
and A were then obtained by reheating at the prior rate. In this
manner, individual dipole responses can be separately observed.
The TSDR response caused by any interaction between the di-
poles and the silicate particles can be seen by comparing the
control (unloaded) sample thermogram with that of the loaded sam-
ple. Peaks A and B are virtually unchanged in magnitude, while the
presence of the filler decreases the relaxation current due to
reordering of the dipoles related to peaks C and D. It is concluded
that these dipoles play a role in bonding the filler particles and
thus fewer are available to respond to the polarization field.
Bulk and IAZ thermograms for the loaded material in the figure
differ in two ways: (1) The magnitude of the current is less in the
IAZ material than in the bulk. (2) The ratios of peak magnitudes in
the IAZ material differs from those in the bulk material.
It was proposed (11) that dipoles taking part in the bonding

process near the interface, and thus not free to respond to the
polarizing electric field, would explain the first observation.
BULK MATERIAL (UNLOADED)
"- "
. . . "", I· BULK MATERIAL

/ ..... 8 ~ (SILICATE LOADED)
'''-, I \
\
\
\
\
\
\
\
\
,
\
'--
"-
IAZ MATERIAL
(SILICATE LOADED)
10-15~--~----~----~----~----~--~~--~
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5
103 IT (K- 1)
Fig. 8. The Thermally Stimulated Dipole Relaxation

Current of Epoxylite 810 Bulk and IAZ Samples,
Polarized with 9.55 x 10 2 V/om at 25°C for
20 Min. Four resolvable peaks are indioated
as A, B, C and D.
The same authors postulate ohemioal oomposition changes, perhaps

due to inhomogeneities in the cure prooess, to explain the seoond .
At the present time, the behavior of peak A presents a dilemma.
Contrary to expeoted dipole response, its magnitude does not vary
linearly with applied field (see Fig. 9) and its peak maximum shifts
with polarization temperature (temperature at whioh field is
applied). Yet its magnitude does ohange with looation in the
adhesive. The implioations of these results are unolear.
10-11r----------------------------------------,
,...
~
I-
Z
W
~
~
::>
u
~
10- 12
«
w
0..
~
::>
~
~
~
10-13~______~~~--~~--------L-------~--~
103
ELECTRIC FIELD {V/cm}
Fig. 9. The Dependence of the Peak Current on the

Strength of the Polarizing Electric Field
for Peaks A, a, C and D. Note sublinear
behavior of A.
From the telJlperature shift in the peak maximum with heating

rate and from the slope of the low temperature tail, an activation
energy of 0.11 eV and a relaxation time constant of 0.1 sec. at
room temperature was calculated for dipole B. Using these values
and assuming a value for the dipole moment, an estimate of the
dipole density can be computed (11). The identity of the molecular
species responsible for this dipole is now known at this time but
typical values of dipole strength for most organic molecules lie
between 1 and 2 debye. For a dipole moment of 1 debye, a dipole
de~~ity of 2 x 10 20 cm- 3 was calculated for the IAZ and 1.2 x 10 21
cm for the bulk material.
Experiments to determine spatial extent of dipole gradients in

the IAZ are reported elsewhere (11). Although there are differences
among the peaks, the results are similar to that of the Rayleigh
wave experiment reported in the previous section. An example of
these results is shown for peak B in Fig. 10. The scatter in the
data is considerably greater than in the acoustic experiment but it
is evident that the percent change in TSDR current (Le. dipole
density to first approximation) is greater than the percent change
in shear speed. (c f. Fig. 1).
We have observed interfacial cohesive failure using a symme-

tric shear specimen design newly adapted for testing shear strength
in adhesives. A finite element analysis of the Epoxylite
810/aluminum system shows shear stress nearly constant along the
bond length and peel stress two orders of magnitude less than the
shear stress. Using this specimen we have concluded that failure in
shear is determined by the shear strength of the epoxy and is not
caused by stress concentration at a crack tip (Mode II). A detailed
analysis of the limitations of this modified Iosipescu specimen and
comparison to the symmetric lap shear specimen analysis (12) has yet
to be made. Such an analysis might include effects due to load
rate, bond thickness, adherend surface preparation and less brittle
adhesives.
Work in progress to determine the extent and nature of the IAZ,

a region over which adhesive microstructure and properties adapt
from surface or interface to bulk conditions, was presented. Using
an ultrasonic Rayleigh wave generation method, we have observed a
shear speed gradient extending approximately 100~ from the adhe-
sive-adherend interface. Measurement of the density gradient in
this region is needed to complement the shear speed data for
determination of shear modulus gradient. Then we will understand
more about the nature of the interfacial cohesive failure which
occurs when significant shear stresses are present.
192
ARTHUR D. JONATH
c c c
1.0 "..----
------~
,,0
/
C
0.9 / A A
...Z A c
0.8
~ I
:::J
u 0.7
:w:
~o
9iA I
0.6 I
I
i
Yo!
t::!
0.5
I
I
oZ 0.4
I
0.3 A SAMPLE 1
o SAMPLE 2
0.1
o ~--~----~--~----~ __ ____
~ L __ _~_ _ _ _L __ __J
o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

Fig. 10. Magnitude of Peak B as a Function of
Thickness of Material Remaining on
Adherend (Normalized to a Peak Current
of 4.75 x 10- 13A at a Field of
1.59 x 103 V/cm) (Ref. 11).
With a sensitive thermally stimulated relaxation technique, we

have observed the response of several distinct dipoles. The
aotivation energy and room temperature relaxation time deduced for
one (0.71 eV and O. 1 sec.) are reasonable for dipoles associated
with molecular side groups in polymers. TSDR measured dipole
structure gradients are seen to extend up to 20~m from the inter-
face. Because these gradients and that of the shear speed differ
both in magnitude and spatial extent, it seems unlikely that a
single underlying cause explains the behavior. From the TSDR

results to date we infer a long-range dipole contribution to the
bonding process and chemical composition changes within the inter-
facial region. Fourier transform infrared (FTIR) and electron
spectroscopy for chemical analysis (ESCA) techniques are being used
to identify the dipoles.
Oxidation and moisture diffusion processes associated with
adhesives aging/environmental deterioration often are found to be
interfacial phenomena. Therefore details of the molecular struc-
ture in the IAZ will provide insight into these mechanisms.
ACKNOWLEDGEMENTS
Throughout the various phases of this work, the author has
received help and encouragement from discussions with John Crowley,
John Bjeletich and Gil Knollman. This work was supported by the
Lockheed Missiles & Space Co. Independent Research Fund.
REFERENCES
1. W. D. Bascom, private communication.
2. J. G. Williams and P. D. Ewing, Int. J. Fracture Mech., 8, 441
(1972). -- -- -
3. W. D. Bascom, C. O. Timmons, and R. L. Jones, J. Mat. Sci., 10,
1037 (1975). --- -
4. G. G. Trantina, ~ Composite Materials, ~, 371 (1972).
5. N. Iosipescu, J. of Materials, 2, 13, 537 (1967).
6. J. G. Bjeletic~and A. D. Jonath, to be published.
7. D. L. Flaggs, private communication.
8. R. D. Adams and N. A. Peppiatt, ~ of Strain Analysis, 2, 13,
185 (1974).
9. G. C. Knollman, J. J. Hartog and A. D. Jonath, to be published.
10. C. Bucci, R. Fieschi, and G. Guidi, Phys. Rev., 148, 816
(1966). - -
11. J. L. Crowley and A. D. Jonath, Adhesion and Absorption of
Polymers, ed. by L. H. Lee, Plenum Press, New York, New
York (1979).
12. W. J. Renton, Experimental Mechanics, Nov. 1976, p 409.
Discussion
On the Paper by L.H. Lee
A. Silberberg (Weizmann Institute, Israel): Have you any idea

about the thinnest layer that could be scanned by this technique,
particularly if there are no strongly absorbing chromophores
present. Could one detect the presence of solution adsorbed pOl~mer
layer which typically are present in the amounts of 2-10 mg/M of
surface?
L.H. Lee (Xerox Corporation): No, we are just beginn~ng to
work on adsorption and do not have any data on the thinnest layer
which could be scanned. The polymers without chromophores could
present problems.
A. Jonath (Lockheed Palo Alto Research Laboratory): Have you
measured PAS on indirect vs. direct semiconductors; i.e., how does
the phonon structure (for indirect semiconductors) appear in the
PAS spectra?
L.H. Lee: No. We have not measured the semiconductors you
mentioned.
A Jonath: What experimental "tricks" did you use to make
measurements in the U.V. region?
L.H. Lee: We used ozone-generating Xenon lamp for getting good
PAS spectra in the U.V. region. In this region, we also have to be
careful about false signals caused by the instability of the system.
It is important to repeat several measurements, particularly with
fresh samples, to confirm the spectra. The U.V. radiation tends to
degrade some polymers or dyes after repeating scans.
B.D. Ratner (University of Washington): In order to study the
nature of adhesive bonding on a molecular level, a higher level of
195
196 DISCUSSION
resolution than that seen in your spectra would probably be neces-

sary. Is there any prospect for increasing the resolution of this
photoacoustic tecfinique or for meaningfully deconvoluting the broad
peaks observed in the U.V. region?
L.H. Lee: So far, resolution appears to be a problem, es-

pecially in the U.V. region. We have tried to reduce the amount of
sample, or to use a diluent for the same purpose. I have not seen
any work on deconvolution. However, I have seen a paper on the
Fourier transformation.. It is likely that the latter approach can
improve the resolution in the future.
On the Paper by J.S. Solomon
D.W. Dwight (Virginia Polytechnic Institute): Why does alumi-

num drop at the metal-oxide interface?
J.S. Solomon (University of Dayton Research Institute): The

apparent decrease in the aluminum at the oxide-metal interface is an
artifact of the derivative recording mode which is related to the
peak shape change and shift of ALKRQ.
D.W. Dwight: Why is there no oxygen in the epoxy?
J.S. Solomon: Some of the sputter profiles shown were con-
structed using normalized signals which means that the minimum
signal levels were arbitrarily assigned a value of zero. Oxygen was
always detected in the bulk epoxy adhesive.
On the Paper by D.W. Dwight
B. Ratner (University of Washington): Infrared analysis of

the etched Teflon surface could not detect chemical changes. Could
you identify by ESCA what chemical structural changes occurred on
the surfaces to impart the darkened color which was observed?
D.W. Dwight (Virginia Polytechnic Institute): Yes, we dis-
covered evidence for unsaturation as well as carbonyl and carboxyl
groups. The details are in J. Colloid and Interface Sci. 47, 650
(1974).
--
• R.K. Burnham, M.M. Farrow and E.M. Eyring, paper presented to
the Symposium on Photoacoustic Spectroscopy, FACSS-V, Boston,
October 1978.
DISCUSSION 197
P. Datta (RCA Laboratories): Viscoelastic properties of the

surface of the polymer contribute to the strength of adhesive bond.
Do you have any method to measure modulus of the polymer surface?
D. W. Dwight: I believe this variable is the most important
parameter that still eludes even qualitative evaluation. Two
approaches have promise to help here: 1. tribological-friction and
wear and 2. surface wave method.
On the Paper by J.L. Crowley and A.D. Jonath
D. W. Dwight (Virginia Polytechnic Institute): Why do there

appear to be two spectra, each with two peaks?
J.L. Crowley (Lockheed Palo Alto Research Laboratory): There
appear to be two spectra in Fig. 1 because peak cleaning techniques
were used to better observe peak B. Specifically the measurement of
the thermally stimulated current was stopped after peaks C & D had
been discharged and the sample was recooled. When the measurement
was resumed there was no longer a contribution to the current from
peaks C & D and peak B could be observed more distinctly.
·D.W. Dwight: Can you interpret the peaks in terms of molecular
motions?
J .L. Crowley: We are not able to correlate specific peaks with'
known molecular motions in the adhesive at this time. We hope to
make some determinations based on a comparison with FTIR spectrum.
D.W. Dwight: What has been published on the surface wave
technique for these materials?
J.L. Crowley: See Sound Waves in Solids, H.F. Pollord, Pion
Ltd., 1977, pp. 65-77.
On the Paper by A.D. Jonath
L.H. Lee (Xerox Corporation): How do you foresee the shear

wave analysis as a complementary device in determining shear
strength?
A.D. Jonath (Lockheed Palo Alto Research Laboratory): In the

approximation that shear strength is proportional to shear modulus,
a shear wave analysis might give the relative strength of an
adjoint. This could be especially useful in determining adhesive
aging characteristics. However, the strength of an adjoint is often
flaw-determined and the shear wave analysis would only be useful in
198 DISCUSSION
predicting locus-of-failure.
D.W. Dwight (Virginia Polytechnic Institute): Has the G of

steel or other model materials been determined by surface waves?
A. D. Jona th : We have determined the shear wave speed of model

materials (as is mentioned in the written version of the paper)
using the surface wave method. Some of these measured values can be
compared with published data as follows:
Material Shear Wave SEeed (km/sec)
Measured Published
304 Stainless 3.14 3.13 (Ref. An
6061 Aluminum 3.18 3.10 (Ref. A2)
Glass Prism 2.60 2.56 " "
Microscope Glass Cover Slip 2.93 2.84 (Ref. A3)
Plexiglas 1.44 - 1.55 1.32 (Ref. A2)
Polystyrene 1.19 - 1.20 1.12 " "
Note that all surface wave (measured) values are greater than
corresponding published values. The latter are bulk values, usual-
ly determined by an ultrasonic transit time technique. The differ-
ence in the two might be explained by density variations between
bulk and surface materials. Thus to a first order approximation,
the shear modulus of model materials has been inferred from surface
wave measurements. But the technique is sensitive to mechanical
properties very near the surface (to a depth of one wavelength of
the incident acoustic wave) and so a density measurement in a very
thin surface layer is required to accurately measure the shear
modulus in the Rayleigh wave experiment.
REFERENCES
A1. R.S. Sharp, Research Techniques in Nondestructive Testing,
Academic Press, New York, 1970.
A2. D. Ensminger, Ultrasonics, The Low- and High-Intensity AEEli-
cations, Marcel Dekker, Inc., New York, 1973.
A3. W.P. Mason, Physical Acoustics and The ProEerties of Solids,
D. Van Nostrand Co., Inc., Princeton, 1958, p. 20.
D.W. Dwight: What is the effect of roughness on this techni-

que?
A. D. Jona th : Surface roughness, if small compared to the
incident acoustic wavelength (as in these experiments), has no
effect.
PART THREE
POLYMERIC STRUCTURAL
ADHESIVES
D.H. Kaelble
Rockwell International
Polymeric structural adhesives now find a wide range of appli-

cations in mobile structures where high structural reliability with
reduced weight and cost result by proper adhesive bonding. Compo-
nent mechanical design, materials selection and manufacturing
process control each play a critical role in achieving proper
adhesive bonding. In this session the papers are arranged so as to
discuss several aspects of fracture mechanics and rheology, fol-
lowed by discussions of surface chemistry and adhesive chemistry,
aspects of joint performance and reliability. Summary discussions
on crack healing and UV curing deal with important issues relating
to the microstructural aspects of adhesion intermediate between the
chemical and continuum mechanical levels of analysis. The evolu-
tion of the technical viewpoint which intimately dovetails chemi-
cal, physical and mechanical viewpoints is reflected in this
session which is a small but important part of this international
conference.
(Editor's Note: The paper on "Adhesive Characteristics of Ultra-

violet Radiation Curable Resins" originally presented to this
Session by T. Kobayashi and K. Nate is included in Part 5 of this
book. )
201
Fracture Mechanics and Adherence of
Viscoelastic Solids
D. Maugis and M. Barquins
Equipe de Recherche de Mecanique des Surfaces
CNRS 1 Place A. Briand 92190 Meudon, France
ABSTRACT
Contact of two elastic solids is treated as a thermodynamic

problem. It is shown that U = UE+US and UT = UE+Up+U S are thermo-
dynamic potentials respectively for transformations at fixed grips
and at fixed load conditions (UE, Up' Us are the elastic, potential,
interface energies). Equations giving the displacement 6 and the
strain energy release rate G as a function of the contact area A and
the load P appear to be the equations of state of the system. Two
bodies in contact on an area A are in equilibrium if G= w, where w is
the thermodynamic (or Dupre's) work of adhesion. This equilibrium
is stable if aG/aA is positive, unstable if negative. The quasi-
static force of adherence is the load corresponding to aG/aA = o.
But equilibrium may be stable at fixed grips and unstable at fixed
load, so that the quasistatic force of adherence may depend on the
stiffness of the measuring apparatus. When G>w, the separation of
the two bodies starts, and can be seen as the propagation of a crack
in mode I. G-w is the force applied to unit length of crack; under
this force, the crack takes a limiting speed v, which is a function
of the temperature, and one can write
The second term is the drag due to viscoelastic losses at the

crack tip and is proportional to w as proposed by Gent and Schultz,
and Andrews and Kinloch. The function~ is a characteristic of the
material (most probably linked to the frequency dependence of E' and
E", the real and imaginary part of the Young modulus) and is
independent of the geometry and loading system. In this proposed
203
204 D. MAUGIS AND M. BARQUINS
formula surface properties and viscoelastic losses are clearly

decoupled from elastic properties and loading conditions that
appeal" in G. If is known, this equation allows one to predict any
feature such as kinetics of detachment at fixed load, fixed grips or
fixed cross-head velocity 6. (This last point completely solves the
problem of tackiness). The only hypotheses are that failure is an
adhesi ve failure and that viscoelastic losses are limited to the
crack tip; this last condition means that gross displacements must
be elastic for G to be valid in kinetic phenomena.
Three geometries are investigated: adherence of punches,

adherence of spheres and peeling. The variation of energies with
area of contact is given, and the kinetics of crack propagation
under various conditions is studied. Experiments on the adherence
of polyurethane to glass confirm the theoretical predictions with a
high precision.
1. INTRODUCTION
Historically, studies of adherence have approached the subject

from two viewpoints: (i) the one of surface chemistry, trying to
correlate adhesive strengths to van del" Waals forces or to the
thermodynamic work of adhesion; ( ii) and the one of fracture
mechanics in an attempt to predict the debonding threshold from
analysis of stresses around preexisting flaws. Neither of these
approaches was successful in studying problems as simple as peel-
ing, or adherence of spheres or punches on elastic half spaces.
Our understanding of the adherence of elastic solids made a

marked advance with the introduction of the energy balance concept
by Kendall (1) and Johnson et al (2). This concept was successfully
applied to a number of cases by Kendall: adherence of punches
(Kendall, 1); peeling (Kendall, 2,3,4,5), lap shear joints (Ken-
dall, 5-8), composites (Kendall, 9-12), laminates (Kendall, 5,1),
and fibre pull-out (Kendall, 13).
This energy balance theory is based on optimizing the total

energy of the system at equilibrium, and the adherence force is
derived by equating the first derivative of the total energy to
zero. But this procedure cannot give information about the stabil-
ity of the system, that depends on the second derivative of the
energy. Thus, one is led to reintroduce the concepts of fracture
mechanics such as the strain energy release rate G, or the stress
intensity factor K, and to study the stability by sign of the
derivative of G or K as proposed in fracture mechanics by Rice (14)
and Huet (15). This approach enables one to study the kinetics of
crack propagation, and provides a synthesis that reconciliates
surface chemistry and fracture mechanics (Maugis and Barquins, 16).
FRACTURE MECHANICS AND ADHERENCE 205
2. GENERAL THEORY
2.1 The Thermodynamic Point of View
Let us consider the system of two elastic solids in contact

over an area A. This system can exchange work and heat with outer
systems, but no matter (closed systems). A force P (either compres-
sive or tensile) can be applied on the two elastic bodies, either by
a fixed load, as in Fig. 1a, or by a more complex system as in Fig.
1b.
At this stage two points must be emphasized:

a) The area of contact A is allowed to vary, so that the
geometry of the system can change, and linear elasticity
must be excluded.
b) This variation of A may be done independently of the load
P, or of the elastic displacement 0, so that the state of
the system depends in general on two independent vari-
ables P,A or o,A. As an example the area of qontact can
decrease at fixed P or fixed 0, until the separation of
the two bodies occurs, corresponding to the rupture of a
joint under fixed load or fixed grips conditions.
p
[aj
[bj
Fig. 1. Study of the state of equilibrium at fixed

load condition (a), and with a testing
machine with a finite stiffness k (b). Fixed
grips condition corresponds to k m= 00.
m
We will consider the reduction of the area of contact as the

propagation of a crack toward the center of the contact, the crack
extension being in mode I (opening mode, with rims moving perpendic-
ularly to the crack plane). Our purpose is to study the equilibrium
and stability of the system under various internal constraints. We
will follow the general method of thermodynamics (see in particular
the illuminating book by Callen (17». The general method is as
follows.
The state of a thermodynamic system is defined by a number of

independent extensive variables X.,1
as S entropy, V volume, N.1
number of moles of constituent i. The system is in equilibrium for
virtual variations of these extensive variables if the energy U is
minimum. The equation
u = U (S,V,X 1.... )
is called the fundamental equation and its first differential
au au au
dU = (as)v,x. dS + (av)s,x. dV + (ax.)s,v dX i + •••
1 1 1
can be written
with the following intensive parameters:

(au)
as v,x. -= T th e t empera t ure
1
au _
-(a-v)s,x. = p the pressure
1
au
( aX. ) s , V :: Pi
1
These intensive parameters are functions of S, V, Xi . .. and the

knowledge of all the equations of state:
T =T (S,V,x i ... )
p =p (S,V,X i ···)
Pi = Pi (S,V,X i ···)
is equivalent to the knowledge of the fundamental equation.
But, in general, the equilibrium must be studied with various

constraints, as no heat exchange, constant temperature, constant
pressure, constant volume, etc. The elegant solution used in
thermodynamics is to use Legendre transformations of the energy U

that exchange any variable Xj with the partial derivative Pj = (~~.).
The equilibrium of the system at constant Pj corresponds to th~
minimum of the Legendre transform
These Legendre transformed functions are called thermodynamic

potentials. So, the Helmholtz free energy F replaces the entropy S
by the temperature T
F = U - TS
dF = -SdT -pdV + P.dX
1
.•.•
1
the enthalpy H replaces the volume V by the pressure p
H = U + pV
dH = TdS + Vdp + P.dX.
1 1
and the Gibbs free energy ~ replaces the entropy by the temperature,
and the volume by the pressure
tt = U - TS + pV
d~ = -SdT + Vdp + PidX i
Thus, any partial derivative P. can be defined as
1
P.1 = (au )
aXi S'V'X j = (aF ) (aH ) (~)
aXi T,V,X j = aXi S'P'X j = aXi T,P,X j (1)
2.2 Equilibrium of the System. Griffith's Criterion
Let us now return to our particular system. Its energy

U = U(S,Q,A)
is a function, b~sides entropy S, of 15 and A, and can be divided

in elastic energy UE and interface energy US. This interface energy
is solely function of A and will be written
=- wA
where y 1 and Y2 are the surface energies of the two bodies, Y12
their interfacial energy, and w the Dupre's energy of adhesion, or
thermodynamic work of adhesion. The first differential of energy
au au au
dU = (as)6,AdS + (~)S,Ad6 + (aA)s,6dA
can be written
dU = TdS + Pd6 + (G-w)dA (2)
with the following parameters

au aUE
(a6)s,A = (~)S,A = P
au aUE aus
(aA)s,6 = (aA )S,6 + (aA )S,6 = G-w
G that describes the variation of elastic energy with A at constant

6 is called "strain energy release rate".
The three equations of state *
S = S(6,A)
P = P(6,A) (3)
G = G(6,A)
carryon the same information as the fundamental equation
U = U(S,6,A).
For equilibrium under various constraints, the following

thermodynamic potentials are to be used:
- at constant temperature, the Helmholtz free energy
F = U-TS
dF = -SciT + Pd6 + (G-w)dA (4)
• Note that the implicit equation (3) implies

aP aA a6
{aA)6 {a6)p {aP)A = -1 (3 bis)
- at fixed load, the enthalpy
H = U-P 0
dH = TdS - odP + (G-w)dA (5)
- at fixed load and temperature, the Gibbs free energy
<t = U - TS - Po
~ = - S:1T - odP + (G- w) dA (6)
Let us now simplify the problem by assuming that thermal

effects are negligible (T=o, S=o). One has
(8)
noting that Up = -Po in the potential energy of the load P. Rela-

tion similar to eq. (1) can be written for G-w, leading to
aU E aU E aup
G = (aA )0= (~)p + (aA )p (9)
Note also that the crossed derivatives (Maxwell relations) of eqs.

(7) and (8) give
(10)
( 11)
Equilibrium at fixed load conditions (dP=o) corresponds to an extre-

mum of F or U (eq. 8 », and equilibrium at fixed grips conditions
(d 0:0) to an extremum of q.. or H (eq. (7) ) • In either case,
equilibrium is given by
G = w (12)
This equilibrium relation may be called Griffith's criterion and

links two of the three variables 0, P, A of the equation of state
(3), so that the equilibrium curves o(A), A(P), P(o) are function of
w.
If G~~ the area of contact will spontaneously change so as to
decrease thermodynamic potentials. (In spontaneous evolution they
cannot increase). If G<w, eq. (7) and (8) show that A must
increase, and the crack recedes. Conversely, if G>w the area of

contact must decrease to have dF<o or di<o, and the crack extends.
GdA is the mechanical energy released when the crack extends by dA.
The breaking of interfacial bonds requires an amount of energy wdA,
and the excess (G-w)dA is changed in kinetic energy if there is no
dissipative factor. G is often called "crack extension force" but
properly speaking the crack extension force is G-w, which is zero
when G=w at equilibrium.
2.3 Stability of Equilibrium, and Stiffness of the Testing Machine
The equilibrium defined by eqs. (12) and (9) can be stable,

unstable or indifferent. A thermodynamic system under a given
constraint is stable if the corresponding thermodynamic potential
is minimum, i.e. if its second derivative is positive.
Stability at fixed grips conditions (d6=0) as deduced from eq.

(7) corresponds to
(13)
Stability at fixed load conditions (dP=o) corresponds to
(14)
But stability can be studied with a testing machine with a finite

stiffness k. Figure 1b is a schematic representation of such a
m
machine. The stiffness k is that of the spring. The crosshead
m
displacement 6 can be obtained by turning the screw (that is
thermodynamically equivalent to place a load P onto the spring, and
then to clamp it without external work), and is divided in elastic
displacement 6 of the spring, and elastic displacement 6 of the two
m
solids in contact. The spring exerts a force P = km6 m on the two
bodies and one has
6=6 +6 m =6 +L
km
(15)
Let us study the stability of the system involving the two elastic
bodies and the spring at fixed crosshead displacement 6. The energy
of the system includes now the elastic energy Um of the springs
U = UE(A,6) + Um(6m) + US(A)
and its first differential is

= GdA + Pdo + Pdo m - wdA
= Pd~ + (G-w) dA
Equilibrium at fixed ~ is still given by G=w, and the stability by
(16 )
but this stability depends on the stiffness k of the apparatus.

m
Intuitively, one can see the spring as a reservoir that provides
energy for crack propagation at constant ~. Let us compute
(~~)~ as a function of (~~)o by considering G[A,6(o,A)] as a
function G[A,o(~,A)]
( aG) aG) (a G) (M.) (17 )

aA ~ = (aA 0 + ao A aA ~
hence
(18 )
Taking into account eqs. (18) and (10), eq. (17) becomes
ap 2 1
(a A) 0 ---'--a-p- (19)
k m + (ao) A
ap
The quantity (ae)A = k is the stiffness of the two elastic solidS,
and is posi ti ve. aJ ~ can be zero, whereas (aaGA) 0 is still
So, (aA)
positive. Eq. (19) given by Courtel et al. (18) shows that the
stabili ty range monotonically increases with the stiffness, from
the fixed load case (k =0) to the fixed grips case (k ex». When
m m
km+o one approaches the fixed load condition and eq. (19) becomes
This last equation is more easily found by writing

aG aG aG aG
dG = (ai)pdA + (ap)A dP = (aA)odA + (ao)Ado = 0
and using eq. (3 bis) and Maxwell relations.
2.4 Quasistatic Adherence Force
The more general equation for stable equilibrium is thus
G =w
aG
(aA)~ > 0
If a fluctuation decreases A (dA<o), G incrementally decreases and

one has G<w: the crack recedes to its equilibrium position. It can
only advance if the load P or the displacement 0 is varied, bringing
back G to the value w: one is dealing with controlled rupture of an
adhesive joint. In this case one has
(20)
with the two particular cases

aG aG
dG = (aA)odA + (a8)A dO =0
aG aG
dG = (aA)p dA + (ap)A dP =0
Starting from a stable equilibrium with the two bodies com-
pressed (P>o, 0>0) let us decrease the cross-head displacement ~:
one generally encounters a progressive reduction of area of con-
tact, i.e. a controlled rupture with (~~)~>o up to a point where
(~~)~=Oj the equilibrium becomes unstable and the crack spontane-
ously extends toward the rupture under this given cross-head dis-
placement. It extends with acceleration, for the crack extension
force G-w increases as A decreases.
The load Pc corresponding to the limit of stability
G =w
=0
will be called the guasistatic adherence force of the two bodies.
As the limit of stability is a function of the stiffness km' the
measured adherence force can depend on the stiffness of the

apparatus.
The Fig. 2 outlines relations between the thermodynamic poten-
tial UT = UE+Up+US (for experiments at fixed load) and area of
contact A. (This figure displays the case of the contact of two
spheres) • The equilibrium curve is the locus of points with
aUT
(11\) p = 0 (1. e. G= w) . By decreasing the area of contact, one
encounters stable equilibrium (heavy line) and controlled rupture
up to the point C where
a 2uT aG
(-2-)P :: (aA)p = 0
aA
At this point spontaneous propagation occurs at fixed load, (dP=o)
and total energy decreases at constant load (quasistatic adherence
force) along the curve (UT)p'
From eq. (8), one has along equilibrium curve
dUT dP
dA = - <5 dA
and hence the point C corresponds to a horizontal taWbent on the

eguilibrium curve. It must be emphasized that neither (aAT) = 0 nor
dUT .
dA = 0 (see p01nt B) are suitable criteria for spontaneous crack
extension.
Similar curve can be drawn for experiments at fixed grips
conditions, with UE+US as the thermodynamic potential. The rupture
corresponds to an extremum of the equilibrium curve given by
2.5 Further Considerations About Forces and Displacements
Let us return to eq. (8); UT = UE+Up+US is the total energy of

a system like that of Fig. 1a and one can write:
aUT dUT
<5 = -Cap )A =- dP (21)
where total derivative holds for displacement along equilibrium

curve (G= w). One can also eliminate dUS = -wdA and write
....
......... [u,Jp slope. G-w
Fig. 2. Variation of the total energy UT with the

area of contact A in the case of the contact
of two spheres. By decreasing the area of
contact, equilibrium is stable (at fixed load
or fixed grips) up to the point C (heavy line);
between the points C and D, the equilibrium
is unstable at fixed load but stable at fixed
grips conditions; beyond the point D (dashed
line) the equilibrium is always unstable.
dUe = odP - GdA (22)
with Uc = -(UE+U p ) = -UT+Us ' a term we will call the complementary

energy. One has:
au c
G = -(aA )p
auC
15 = (aP)A
This last expression is known in classical elasticity as the
generalized Castigliano theorem, that gives the displacements in
au
nonlinear elasticity, when Castigliano' s theorem 15 = ~ PE) A is
irrelevant. The complementary energy is thus the thermodynamic
potential used in classical elasticity for transformation at fixed
load. In classical elasticity (Us=o)
Uc = PI5 - IPdl5 = Il5dP

but, in problems with surface energy, the elastic energy is no
longer given by IPdl5 as we will see below.
Eq. (1) can be rewritten as
dUE = Pdl5 + GdA (23)
that shows that evaluation of elastic energy in systems with surface

energy need\! special care. Besides external forces, there are
forces due to molecular attraction that cause elastic deformations
and lead to elastic energy storage. In classical mechanics the
derivative of the elastic energy gives the applied load. Here,
three derivatives can be defined:
(24 )
(25 )
dUS .
The force FS = - d 15 1S associated with displacements along the
equilibrium curve; the forcel1, also given from eq. (22) by
auc
11 = -(ar-)p
acts in addition to applied load P in a displacement 15 at constant

P.
2.6 The Stress Intensity Approach
It is shown in fracture mechanics that stresses vary as ~ with

the distance p near a crack tip, and the stress intensity fac€or K,
for crack propagation in mode I is defined by
KI
(J
z
= (26)
I2np
where (J is the normal stress in the crack plane. This stress

z
intensity factor is related to strain energy release rate by
K 2
I
G= E for plane stress (27)
2
G = K 2 1-v for plane strain (28)
I E
This intensity factor can also be seen as a discontinuity of

displacement factor (17), for discontinuity of displacement [u ]
can be written as Z
[u] = KI (k+1..,Jf; (29)

Z ~ 2n
with ~ the shear modulus and
k =b for plane stress (3Oa)
1+v
k = 3-4v for plane strain (30b)
2.7 Viscoelastic Losses
The force P corresponding to G=w and (oGloA)=O is the quasi-

o
static or thermodynamic force of adherence (not to be confused with
the thermodynamic work of adhesion w). When G>w, the force G-w is
applied to the unit length of crack that accelerates, until a limit
speed v imposed by viscoelastic losses is reached. In the case of
spontaneous crack extension, G-w increases with time under the
constant load P so that the crack velocity increases with time.
o
But there are cases (e.g. peeling) where G is independent of the
area of contact, so that the crack does not move under the force P .
A force P> P must be applied to move the crack at detectab£e
o
velocity, and the value Po can be obtained only by extrapolation
towards zero crack speed.
Viscoelastic losses vary with deformation rate and tempera-

ture, but results at various temperatures can be shifted to a
reference temperature Ts =Tg +50 (where Tg is the glassy transition
temperature) by using the WLF shift factor aT given (Ferry, 20) by
8.86 (T-TS )
(31)
log aT = - 101.6 + T-TS
So, for a given geometry, adherence forces may be studied as a

function of the reduced parameter a~v, where v is the crack speed: a
master curve is thus obtained (KaeIble, 21,22), (Gent and Petrich,
23) , (Gent, 24) , (Gent and Kinloch, 25). If one assumes that
viscoelastic losses are localized at the crack tip and proportional
to the thermodynamic work of adhesion- as suggested by peeling
experiments in various liquids (Gent and Schultz, 26) or on various
substrates (Andrews and Kinloch, 21) one can write
(32)
where ~ is a dimensionless function of crack speed and temperature,

independent of the geometry of the system. At high crack speed one
has G>'W, and the variation of crack speed with w, arises from the
multiplication term w in the right hand side of eq. (32).
Equation (32) can be rewritten as

(33)
where wf may be called apparent work of adhesion, or fracture work

of adhesion by analogy with fracture surface energy Yf. This
apparent work of adhesion is thus the product of two terms: the
thermodynamic work of adhesion wand a function of velocity and
tempera~ure, as proposed by Gent and Schultz (26) and Andrews and
Kinloch (21) • Equa tion (33) shows that influence of geometric
parameters entering in G can be tested by working at constant
temperature and crack velocity, i.e. working at constant wf •
Measuring adherence force with a non-zero crack speed, can lead to
very high values, since viscoelastic losses , and hence wf ' can be
enormous, even for 1 lJIIl S-1. Inspection of the WLF shift factor
shows that it is beneficial to work at high temperature to reduce
wf ·
- This means that viscoelastic losses can only arise if the inter-
face is itself capable of withstanding stress (Andrews and Kin-
loch, 21).
Application of eq. (32) implies that gross displacements are

purely elastic with G computed from the relaxed elastic modulus E ,
o
and that the frequency dependence of E only appears at the crack tip
where deformation velocities € are high. The loss factor :~ (where
E' and En are the real and imaginary parts of E) and its frequency
dependence are taken into account in the function ~(aTv).
Note that decreasing w at constant bulk properties and loading
system increases the crack velocity, so that the viscoelastic solid
seems to be less "viscous". At very low w values, the interface can
appear "brittle". Conversely, inserting a very thin film of glue
between the two solids considerably increases the work of adhesion
but does not change the variation of elastic potential energies, so
that the above treatment may be used for glued jOints.
3. EXPERIMENTAL METHOD
3.1 Apparatus
The apparatus used (Fig. 3) to study the adherence of glass
spheres and punches is a modified form of the apparatus described
elsewhere (Barquins and Courtel, 28) for the study of rubber
friction. The area of contact, illuminated by reflection of
monochromatic light, is observed through the punch or the spherical
cap with a microscope. Figure 4 is a diagrammatic view of the
mechanical system. The hemisphere or the punch is supported by a
balance. This balance has an axis of rotation given by two micro-
ball bearings; its sensitivity is about 10 dyn. The vertical
displacement of the arm can be recorded by a displacement transducer
with a precision of about 0.1 ~m. The whole mechanical system can
be laterally shifted to adjust the centre of the contact with the
microscope axis.
A 16 mm camera (25 or 50 frames per second) is located at the
top of the microscope, and the contact areas are recorded at a
magnification of about x10. Any sequence begins with the photograph
of a micrometer scale; the areas of contact are such that the view
through the glass ball does not introduce optical distortion. The
contact radii are thus obtained with a precision of about 1 ~m.
Experimental results of sections 4.4 and 5.5 correspond to about
160,000 frames, of which about 60,000 have been measured.
3.2 Material
The viscoelastic material chosen was an optically smooth poly-
urethane, recommended for dynamic studies in photoelasticity (PSM 4
Vishay) and delivered as plates of thickness h=3.175 mm (1/8 inch).
"T1
:0
»
~
c
:0
m
s::m
(")
:I:
»z
~
»z
o
»
o
:I:
m
:0
m
Z
(")
m
Fig. 3. General view of the apparatus used to study the adherence of glass
t.J
spheres and punches to viscoelastic materials. ~
-0
to..)
to..)
o
P+P'
,Ioac,D
"
"- elastomer
\
\
\ ~
COUnteMeight ~
x l>
C
G')
Fig. 4. Schematic arrangement of the apparatus shown in Fig. 3. en

l>
Z
o
~
OJ
l>
:0
o
C
Z
en
The surfaces were cleaned with alcohol, dried wi th warm air, and
left during 30 min for the equilibrium with room atmosphere to be
reached. Microscopic examination fails to reveal any surface
feature such as the bloom crystals observed by Roberts and Othman
(28 bis) on rubber surfaces. The Young's modulus E and the thermo-
dynamic work of adhesion w to glass were measured by the method
proposed elsewhere (Barquins and Courtel (28), i.e. the verifica-
tion of eq. (72) for the contact of a small glass ball. This method
was previously used to study the variation with temperature of E and
w for rubbers (Barquins et al., 29). The Young's modulus of this
polyurethane, so measured, is E=5x10 7 dyn cm- 2 , the Poisson's ratio
is taken as v=i, and w can vary from day to day, most proE~bly with
the relative humidity of the atmosphere from w=50 erg cm to w=85
erg cm-2 (high values corresponding to fine weather). So w was
always measured before any set of experiments, and the contact
during any set of experiment was always made at the same point. In
these conditions, the reproducibility is excellent.
4. ADHERENCE OF A RIGID PUNCH
4.1 Indentation By a Punch. The Boussinesq's Problem
Stresses and deformations provoked by a rigid circular flat

punch indenting an elastic half-space (Fig. 5a) were given by
Boussinesq (30). Under the punch (radius a) the elastic displace-
ment and the pressure are -
2
.bL
0= 2E aP (34)
P
(35)
2na2~_ 1':
a
The vertical displacement of the surface- for r>a is
1 2
u = ~ ~ Arc sin ~ (36)
z 1TE a I'
- From a theorem due to Way (31) the horizontal displacement in the

surface, outside the contact is independent of the load distribu-
tion inside the area of contact. It is thus the same as for a
concentrated force
(1-2y)(1+v) .E for I' > a
21TE 1"
The asymptotic value at large distances, £»1 is

a '
u (37)
Z
as for a concentrated force; the asymptotic value near the edge
u = <5 - b1 - ~ (38)
Z n ~~
du
thus, dr Z = 00 at the edge of the contact. The same equations hold
for an adhesive contact, when P is a tensile load (Fig. 5b).
4.2 Strain Energy Release Rate

The variation of energies in a fluctuation of area of contact
at fixed load or fixed displacement is difficult to "see" in Fig.
5b, and it is more convenient to consider Fig. 5c, a rig~ plate in
contact with a viscoelastic half space on an area A=na , under a
tensile load P.
In fact, it is not necessary to restrict ourselves to a rigid

plate, and the stresses developed at the interface will develop the
same kind of elastic deformation in the "indenter" (Fig. 5d). So,
as pointed out by Kendall (31 bis) adherence of a punch is represen-
tative of fracture experiment on a deeply notched cylindrical bar.
In Fig. 5d, the elastic displacement <5 is the sum of the displace-
ment of the two bodies and will be written
2P
= 3aK (39 )
with 2 2
1 1-\11
3 1-'-2
K = '4 (~+ E;- ) (40)
The elastic displacement varies with P at fixed a in Fig. 5a, but

can also vary with a at fixed P as in Figs. 5c and 5d. Equation (39)
is the equation of state (3).
The elastic energy is computed from the observed geometry of

the system, and from given values of <5 and a it is the same as in
Fig. 5c , i. e.
U =.! P <5 = 3aK
E 2 4
t -_p23aK
!
\U;!
'P i
i
t
\·,.....L..../
r
[a]
[c]
,
I
[d]
Fig. 5. Analogy between the adherence of a rigid
circular flat punch on elastic half-space
and a fracture experiment on a deeply
notched cylindrical bar.
With the potential energy, Up = -Po, one can compute G from eq.
(9):
p2
G 3Ko2
= 6na 3K = ana (41)
This is the second equation of state.
The equilibrium given by G=w is always unstable for

(42)
is negative. If, from the equilibrium point, A increases, it

will indefinitely increase; if it decreases, it will decrease until
the rupture is reached. If the punch is of finite size (radius a )
2 0
the area of contact, of course, cannot exceed A = na ; and the
only possibility when P reaches the critical valu~ 0
(43)
given by G=w, is a spontaneous reduction of area of contact down to

rupture. This value P , first given by Kendall (1) is the quasi-
-c
static adherence force of this punch. Similarly, if displacement 0
is imposed, spontaneous rupture occurs as soon as G=w, correspond-
ing to a critical displacement-
(44)
The Maxwell relations (10) and (11) becomes
cG cp P
(-a&)A = (cA)o = 2 (45)
2na
cG co P
(cp) A = -( cA)P = (46)
3na 3K
Note also the relation
(47)
4.3 Energies of the System; Berry's Representation
As said in sec. 2.2, the "enthalpy" UT = UE+Up+U S is the

thermodynamic potential for transformations at constant load (see
eq. (8». Figure 6 shows the variations of this "enthalpy" along
the equilibrium point, or during transformations at constant load
- As 0 may be seen as a crack opening displacement (COD), it appears

that a critical COD must be reached before the crack begins to
start.
P. At equilibrium points one has
UE = -2US = 2Aw (4S)
so that
(49)
aUT
and the curve UT(A) is a straight line with (aA )p = o everywhere
dUT
and dA = 0 nowhere.
For transformations at fixed displacement, the internal energy

U = UE+US is the thermodynamic potential and the Fig. 7 gives its
variations along equilibrium pOints with
For spontaneous transformation these potentials cannot in-

crease.
Berry (32) has shown that plotting the locus of the Griffith
criterion, and the loading curve on a stress-strain plot enables the
various energies involved to have a geometric interpretation.
Especially the surface energy of the system is the area between the
loading curve and the locus of the Griffith criterion. This locus
is the cubic
2
P = 9lC... 03 (50)
16nw
and is drawn on Fig. S. Points such as B or B1 on this locus have
the area A as parameter. For a punch of area A , the loading curve
o
is the straight line OB (constant A).o
At the equilibrium point B,
the elastic energy UE = iP 0 is given by the triangle OBI, and the
c c
surface energy I usl = iluEI by the double hatched zone (for a cubic
y=kx 3 , the area under the curve is ky x) • During crack propagation
at fixed load (Fig. Sb) or at fixed displacement (Fig. Sc) the area
of contact and the surface energy Us decrease, and the elastic
energy given by the triangle* OB 1 I mcreases toward infinity or
decreases to zero.
* Note that the elastic energy at the point B1 in Fig. Sb is not

given by
fPdo = iP c0 c + Pc (0-0 c )
(footnote continued on next page)
During crack propagation at fixed load, the variation of UE is given

by eq. (24) w1th
.
-r = - P2' aUE
Hence (a6'")p =2
P
and
<5
UE = iP c <5 c + J<5 (P+1r)d<5 = iP c <5.
c
A
0.5
r~~::~:::::G:~~O~~~~~~~~,~ P",OI
-1
-6
-7
Fig. 6. Total energy versus area of contact in reduced

coordinates. The dashed line is for equili-
brium points, and heavy curves are for varia-
tions at constant load. A schematic evolution
of the contact under the load P is also shown.
c
JL
Aow
/
/
/
/
/
/
/
- 0.5
-1
Fig. 7. Internal energy versus area of contact in

reduced coordinates. The dashed line is for
equilibrium points and heavy curves are for
variations at fixed grips condition. A schematic
evolution of the contact under the displace-
ment 0 is also shown.
c
In Fig. 8b the thermod~amic potential UT = UE+Up+Us at point

B, is the negative area HB,B 20 and its variation BB,B 2 (dotted zone)
during crack propagation is the work st(G-W)dA done by the system:
it corresponds to the kinetic energy (5f the crack and/or to heat
dissipated in viscous drag. The work done at rupture is infinite,
for rupture corresponds to 0 infinite.
Y.~
(pJ B
H ------ - ------
[aJ
o x.IDS.
w
[b]
[c] fZZJ u(
~UII
[:·. :.l~UT
Fig. 8. Berry's representation for the contact of

a rigid flat punch. Stored elastic and
surface energies are given for equilibrium
condition (a) and for crack propagation at
fixed load (b) and at fixed grip condition
(c). The variation of total energy during
propagation is also given.
f!
In Fig. 8c the thermodynamic potential U = UE+US is the posi-
tive area 0a2B1I and the work (G-w)dA done by the system is
still BB 1B2 • At rupture this wor~is ~UE = -US = Aow.
4.4 Crack Propagation at Fixed Load

Starting with a punch at P=o, 15 =0, let us suddenly apply a
tensile load P>P and study the kinetics of crack propagation by
measuring the va~iation of radius of contact with time.
Figure 9 is a photograph of the area of contact during this

crack propagation. Whatever the shape and size of the initial
contact, the final shape is that of a circle. The grey border
around the contact is due to light reflection along the polyurethane
profile shown in Fig. 5c. This profile varies with P and a.
Figure 10 shows the variation with time of the radius of
contact under loads varying from P = -3000 to -15000 dyne The
origin of times is that of rupture, so that all times are negative.
For a given tensile load, the crack speed increases as the radius a
decreases, and for a given radius a, the crack speed increases as
the tensile load P increases. These results can be analyzed by the
eq. (32) with G given by the eq. (41). On Fig. 11 the normalized
"moti ve" G-w computed from the value of P, wand the measured a is
w da
plotted as a function of the crack speed v = dt' The results
obtained at 21 0C were shifted to 23 0C by the WLF method (eq. (31)
with a glassy transition taken as T = -50 0C) to compare with the
peeling results (heavy curve, see gsec. 6). This curve is the
dissipative function $(aTv) that appears to vary as (aTv)0.6.
Once the dissipative function

$(aTv) = a(aTv)0.6 (51)
known, the differential equation (32), together with the eqs. (39)
and (41) allows for the prediction of any evolution of the system.
The kinetic is independent of a and only function of the instantan-
o
eous radius of contact a. It is the kinetic of the rupture of
plane/plane contact, under the action of a traction perpendicular to
the interface. In the particular case where the initial crack speed
is such that G>>W, it comes
1 6
t = -61 (61TKwa)0.6 ~
P10/3
t-)
t.>
o
o
3:
»C
C)
en
»
z
o
3:
!Xl
Fig. 9. View of the area of contact during the unloading of a glass »::c
flat punch from a polyurethane surface. The corresponding oC
profile of contact is shown in Fig. 5c. Z
en
"Tl
:0
»
~
300 c
:0
...... m
""'--'- ........... .....:-0" \. s:
.-.-... .'. E m
....."',~
......... ~.\,.'- ,\\ \ :::L.. (")
J:
(/)
.-.-~----.-.-.-.-.-.-.-. -~" ..... "~ »z
-._'-. ...... '\ :J
. . ""'a"a'D.o .... ,\,_. k - _ (")
'\. 2000 CJ)
----_.__ __ -.-.-.-.-
.-.-......--.
....'-.....
......
• \h
\ .I. ct
.... »
z
............... '"cr.,\., ~ o
..... \l< () »
........... DII '! ...., o
....
. ';'
• ir.
Z
"
............. \~ 0
~ J:
-~-'-'-'_0_._._._-.-.-.-._---
_0_0_ 0_ 0_ . _0_ . _ _0--'_0_ '........ \u• m
:0
._--._- .'--....
..............."\r;
' \. '"0 0 m
Z
(")
0 .... 0..0 ". m
P dynes '09'
. \~-
o -3000
• -4000
.~
+ -6000
• -8000
• -10000 8=21°C
• -12000 w= 75ergs.cm-2
6 -14000
• -15000 E=5.10 7 dynes.cm-2 , 10'=0.5
-20 -15 -10 -5 o

TIME ,s
Fig. 10. Adherence of a glass flat punch to polyurethane: radius of contact

versus time, for various tensile loads P. The origin of time is
that of rupture so that all times are negative. For a given tensile
load P, the crack speed increases as the radius a decreases, and
for a given radius ~ the crack speed increases as the tensile load
P increases.
I'.)
w
ro4'~ ______ ~ ________ ~ ________ ~ ________ ~ ______ ~ ________...
10
.."SA
.....'Y""
.~ . ~
:.•..J.
.,,,,+,#,,'
.~T
.,9+'1
...'
a
+ ...
P,dynes
• +
• -3000
10 .Jt'.~'
•• • -4000
,;r-: ~.
0.6
• -6000
• -8000
• ·\0000
.-12000
o 1 ··14000
.·15000
11~~~------~------~1~0--------1~0~2--------~~-------~~------~105
CRACK SPEED. pm.s·\
Fig. 11. "Normalized motive" versus crack speed for

adherence of a glass flat punch on polyure-
thane surface. The same master curve as for
peeling (heavy line) is obtained.
6 6
and the curves of Fig. 10 may be normalized by ~ and not by a 4 as
p1013 p
erroneously inferred by Maugis and Barquins (33).
The whole analysis is based on the hypothesis that eq. (41) is
still valid during crack propagation, i.e. that displacements 6 are
purely elastic, and related to P by eq. (39) with relaxed elastic
moduli. This was checked. Figure 12 shows the theoretical rela-
tions between 6 and a at equilibrium points, or during rupture at
constant load in the elastic hypothesis. The experimental dis-
placements are given on Fig. 13. As the origin of the displacement
6 was difficult to obtain, the points (6,a) corresponding to the
last frame (1/8 S before the rupture) of every unloading were put on
the theoretical curves. It can be seen that all the other pOints
fall on these theoretical curves. Thus 6 varies with a, at constant
P, in agreement with eq. (39), i.e. the displacements are purely
elastic, so that viscoelastic losses are limited to the crack tip.
This is why the expression for strain energy release rate is still
valid far from equilibrium.
10·~~
w
10
-0.5
-1
-1.5
Fig. 12. Relations, in reduced coordinates, between

elastic displacements 0 and radius a of contact
area of a flat punch with a viscoelastic
half-space. The dashed curve is for equilibrium
points and heavy lines are for variations
at constant load.
4.4 Crack Propagation at Fixed Grips
Starting with a punch at P=o, 0=0 one can suddenly apply a

displacement 0>0 c and record the variation of the force P during
crack propagation at fixed O. The variation ABO of the force with
a, is depicted on Fig. 14 (eq$. (39) and (43», but no experimental
verification was attempted.
4.5 Crack Propagation at Fixed Cross-head Velocit~ Tackiness
Starting with a punch (radius a o ) with an initial displacement

corresponding to unstable equilibrium (eq. (43» one can impose a
a,pm
O~ ________~I,OO~________~200~________~3~OO
- - ------------.£..-
6
6,pm
-20
-40
-60
P,dynes
0-3000
• -6000
• - 9000
, - 12000
• -15000
'·Ieooo
- 80
Fig. 13. Displacement versus radius of area of contact

for adherence of a glass flat punch on a
polyurethane surface. Experimental and theo-
retical curves exactly coincide. The dis-
placements are thus purely elastic, so that
viscoelastic losses are limited to the crack
tip .
fixed cross-head velocity 0. = (it

do and compute the variation P (t )
from eqs . (32), (39), (41) and (51).
The crack velocity is given by
dt
(3 i
da = K
8'ITawa.
_1\0\
/
and its variation by

Fig. '4. Relations, in reduced coordinates, between

the load P and the radius a of contact area
of a flat-punch with a viscoelastic half-space.
The dashed curve is for equilibrium points
and heavy lines are for variation at constant
displacement o.
d2a = _,_ (da)0.4(3K02 ) (_ da + 2a do)

dt 2 0.6 dt 81lWa2(l dt 0 dt
From equilibrium values a and 0 , one increments 0 to 0, by

92 0
0, = 00 + ~llt, and 1~~ I, and 1:t~ ~ are compute~ by above equations
with a o and 0, values. An approximate value a, is obtained by:
2
da, d a,
from which the correct values dt and dt 2 are calculated, hence
the true a" by above equations with a, • and~, instead of ao'~'.
By such incrementations one can compute a i and the force
Pi = ~ aiK~i·
Figure '5 displays such computed curves for various.~. The
force goes through a maximum Pmax that is often taken as a defini-
tion of the tackiness of the viscoelastic solid. It must be pointed
out that P has no physical signification: for a given punch of
max
radius a o it increases with the cross-head velocity ~ • From eq.
(4') Pmax corresponds to
(52)
The increase of G with time is due to simultaneous increasing ~ and
decreasing a:
dG aG • aG·
dt = (at)~ a + (at) a~
and eq. (52) leads to
2(~~)~ = (~~)a
for Pmax. The left hand side is a material (and geometric)
characteristic but the right hand side depends upon ~.
Experimental verifications are in progress.
5. ADHERENCE OF TWO SPHERES

5.' The Hertz Problem
Stresses and deformation produced by the contact of two elas-
tic spheres (radius R, and R2 , Young moduli E, and E2 , Poisson's
ra tio v, and V2) were given by Hertz ('880). Under a load P, the
area of contact has a radius a H given by
FRACTURE MECHANICS AND ADHESENCE 237
p \
\
TIME
Fig. 15. Computed curves, for various fixed cross-

head velocities, of the variation with time
of the adherence force of a flat punch in
contact with a viscoelastic surface.
(53)
with
1 1 +1-
- - -
R - R1 R2
and K given by eq. (40). The elastic displacement and the pressure
are
a H2
ISH = If" (54)
Oz
3 P
="2-2
1Ta
P;
H 1--;
.;;.
(55)
a
Note that the relation between P and IS is not linear (Fig.
16), so that the elastic energy is not iPISH but
UE = ~ PIS H (56)
and that elastic displacement is not given by the Castigliano' s
theorem, but by the generalized Castigliano's theorem (eq. (21»:
4
_~L __________________________________________________ _
-1
Fig. 16. Equilibrium curve for Hertzian (w=o) and

adhesive (w~o) contacts of two spheres.
Berry's representation with stored elastic
and surface energies at the equilibrium
point B.
using the thermodynamic potential UT given by
UT =- .2.U
2 E
(57)
(as for punches, eq. (49». Note also that the relation
dUE
do = P
is still valid.
The tangent to the Hertzian curve P( c) at a point where the

radius of contact is a, has a slope (the stiffness):
dP 8
k = dc = 3aK
2 (5 )
and intersect the 15 axis in ~ (the Legendre transform $(A) = c-AP
that exchange P and the compliance A = ~ in the (P,c) relation) and
the P axis in - ~ (the Legendre transform $(k) = P-kc that exchange 15
and k). Note that compliance and stiffness are that of a punch of
radius a.
5.2 The Strain Energy Release Rate
The Hertzian results do not take into account the molecular

attraction between the two bodies leading to a larger radius of
contact a>.a H and elastic displacement c>c H. The correct calcula-
tion has been performed by Johnson et al.(2).
The actual radius of contact under a load P can be expressed by

an equation similar to eq. (53):
3 P 1R
a =~ (59)
where P1 )P is an apparent Hertz load. The actual elastic displace-
ment 15 and elastic energy UE are found as following (Fig. 16).
Assuming no surface energy, let us apply the load P1 (point J): t~e
displacement is:
2
(eq. (54» and the elastic energy is ~ -i--a
P
(Eq. (56». Then, at constant radius of contact let us decrease the

load from P1 to P by increasing surface energy. The variation of
displacement (J to B) and elastic energy are that of a punch of
radius a:
2(P 1-P)
M = 3aK
2
P1 p2
= 3aK - 3aK
Thus, at point B one has
2 2P
15 = ~R + 3aK (60)
p2 a5K
U =-+-- (61a)
E 3aK 15R2
2 2 5
= 3aK (0 _ ~) + L!L (61b)
4 3R 15R2
From eq. (9) it comes, with Up = -Po
G _ (P 1-p)2
- 6 nRP 1 (62a)
3
= 3a K (1 _ R 0) 2
(62b)
8nR 2 a2
Equations (60) and (62) are the equations of state of the system.
Note that eq. (62) gives eq. (41) when R+c:q The derivatives of G
are
P 2_p2
00 1 (63)
( ClA)P =
4,fa 2RP 1
dG 3K 2 2 (64)
(ClA) 0 = 2 2 3 (a -QR)(3a +OR)
16n R a
Note that eq. (47) is still valid, and that UE can be written
UE = '52 (P 6+-GA) (65)
The Maxwell equations (10) and (11) becomes

dG dP P1-P
(~)A = (ClA) 0 = - --2 (66)
21Ta
P1-P
00
(aP) A = _(M)
ClAP = (67)
- 311Ci 3K
5.3 Equilibrium Relations and Quasistatic Forces of Adherence
The equilibrium relation G=w links now P1 to P and two of the

three variatbles of the equation of state (60).
Figure 17 shows the relation between P1 and P from eq. (62a);in

reduced coordinates it gives a parabola inclined by 45 0 to the axis,
the top of which corresponds to P = -5P 1 , the slope being
x .. _p__ ...!l.!L_ ~n2R

3llwR - 3I1wR2-lWR'"
3 4
Fig. '7. Elastic contact of two spheres: apparent

Hertzian load P, versus normal load P in
reduced coordinates. This is a parabola
inclinated at 45 0 to the axis. The stability
with fixed load conditions is shown by
heavy curve. Fixed grips conditions allow
to increase the range of stability (dotted
curve) •
(68)
3 P,H 3 PH
This figure thus shows the relation between a =~ and a H = i:.
Figure ,8 gives the same information for pea) together with
variations at constant o. These last curves have their slope given
by eq. (66) and have their maximum on the Hertz curve (P,=P).
Figure '9 shows the relation between 0 and a from eq. (62b) in
reduced coordinates together with variations at constant load. The
10
-5
-20 -15 -10 -5
-10
Fig. 18. Relations between normal load P and radius

of area of contact a of two spheres, in
reduced coordinates. The equilibriym curve
is pea); curves (P)o show the variation of
P with a at fixed grips condition. The curve
PH is given by Hertz' theory. Displacement
from equilibrium value 0 to 0' leads to an
instantaneous unloading at constant a (branch
LM) followed by an evolution along the curve
(P)o' towards a new equilibrium if 0' >
2 2
_ (31T W R) 1/3
ljK2
6 \ \
r~R]! \ \
\
\ \
15 \ "- "-
\ "- ........
\
\
\
\
\
""
10
"-
, ,,
-5
,,
,
,,,
,,
,,
I I~I
x= 3"wR
'\
I
I
I
I
-10 \
Fig. 19. Relations between elastic displacement 8 and

radius of area of contact a of two spheres, in re-
duced coordinates. The equilibrium curve is 8 (a) j
curves (8)p' show the variation of 8 with a at fixed
load. The curve 8H is given by the Hertz' theory.
Unloading from P to P' leads to an instantaneous
displacement at constant a (branch LM or LM'), fol-
lowed by an evolution along the curve (8)p, towards
a new equiljbrium if P' > - ~nwR (branch MN), or
towards rupture if P' < - ~WR (branch M'Q).
6(a) curve, first given by Johnson (34), has a slope

d6 sP 1+P
(69)
dA = 61Ta 3K
The (6)p curves have their slope given by eq. (67) and have their
minimum on the Hertz curve (P 1=P).
Note that in Figs. 18 and 19 the curves (P) 6 and (6)p are
independent of w.
The equilibrium curve P( 6), first given by Johnson (34) is
displayed on Fig. 16. Its slope is
dP 9Ka [ P1+P ] (70)

d6=T 5P 1+P
The curves (P)a giving the variations of P with 6 at constant a, are

straight lines such as BJ, independent of w. Their slope
(ap) = 3aK = k
a6 a 2
is the stiffness of the contact. They intersect the 6 axis at
6H 2
~ = ;R (the partial Legendre transform w(X,a) = 6-XP that exchanges
P and the compliance X = (~p6) in the P,6 relationship) and the P
P1 a
axis at - 2" (the partial Legendre transform W(k,a) = P-k6 that
exchanges 15 and k).
The stable equilibrium under the load P is defined by the
positive value of (~~)p, eq. (63). The branches corresponding to
stable equilibrium are the heavy curves in Figs. 17, 18, '9. They
correspond to
P, = P + 31rWR + [61rwRP + (31rWR) 2] i (71)
a3 = ~R l, + 31r;R + [2 (31r;R)+ (31r;R)2Ji ~ (72)
given by Johnson et al. (2). The term between brackets is the

correction to the Hertz's theory. The equilibrium is unstable and
the crack spontaneously extends at fixed load (adherence force) for
P = -P, = +WR (73)

3 2
a min = 31fWR 7 2K
15 = a 213R.
At this point (point C) one has dP/da=O, dc/dP=oo, and the curves
15 (a) and (c)p are tangents as expected above.
The equilibrium is stable at fixed grips conditions if
(~~) 15 >0. The two terms between brackets in eq. (64) vanish
respectively for P=P, and P+5P,=0. The hatched zone of Fig. '7 are
for instability. It can be observed that, at fixed 15, P, can de-
crease down to the top of the parabola. At this point, corres-
ponding to the spontaneous rupture, one has
5
P = -5P, = - ~wR
a 3 = 1TwR 2 /6K
15 = -3a2 IR.
At this point d c/dP=O, and dOldA=O as expected. The equilibrium
curves are thus more extended at fixed grips (up to the point D) as
expected in sec. 2.3. Beyond the point C eqs.(7') and (72) are not
valid, for the sign must be changed before the square root. How-
ever, the relation
P, - P = (6n wRP , ) i
remains valid along the whole branch of equilibrium at fixed grips.
Figure 20 shows the geometry of an adhesive equilibrium con-

tact under various loads, for a hard sphere on an elastic half-
space, and Fig. 20a compares Hertzian contact and adhesive contact
under the same load. In a Hertzian contact, the displacement 15 H is
twice the height of the spherical cap h=a 2 12R. In an adhesive
contact, one has 15< 2h. The contact angle 8 arises from the
existence of surface and interface energies. (for Hertzian contact
w=O and 8 =1T). Such an angle was effectively measured (Barquins et
aI, 35) in the case of glass-rubber contact (8<'60 0 ). Whereas in a
Hertzian contact, the elastic material leaves the contact facing
upward, in adhesive contact it leaves it facing downward, as shown
in Fig. 20, unless high loads are applied, corresponding to
a IR>Sin 8.
* By analogy with Wenzel's relation for the wetting of rough

surfaces, it may be anticipated that the apparent contact angle
can vary with the roughness r of the surface. Reasoning as in the
thermodynamic derivation of the Wenzel's relation (Good, 36), one
can see that a variation dA in the apparent area of contact leads
to a variation dUS = -wrdA=wadA in the energy of surface, while Up
and UE are at first approximation not modified for slight rough-
ness. Equation G=w becomes G=w =rw, indicating that slight
a
roughness can increase the adhesion as compared with an ideally
smooth surface, as experimentally shown by Briggs and Briscoe
(37) •
The load P1 is the apparent Hertz load (for w=o) that would
give the same area of contact, but the displacement would be
different. To have the same displacement 0 as the one calculated
for w~o, the apparent Hertz load P' given by
P ,2 = P 2 P1+2P 3 (74)
1 (3P"")
1
should be applied, but the area of contact would be different.
5.4 Energies of the System. Berry's Representation
Figure 21 shows the variations of the thermodynamic potential

Ur = UE+Up+Us during transformations at constant load or along the
equilibrium curve. The second branch of unstable equilibrium is
given by
P1-P = -(6TIWRP 1)t
The variation of the total energy in the Hertz theory (w=O) is also
given for comparison. The relation
2
UE = - '3UT
for variations along the equilibrium curve, is the same as for
Hertzian contacts (eq (56», or for punches (eq. (49».
When eqs. (67) and (69) are taken into account, the forces due
to molecular interactions, eqs. (24) and (25) become
61TRP 1
FS = w 5P,+P (75)
I~=tl
+---------~~~~------~
6=h
Fig. 20. Geometry of an adhesive contact in equilibrium

under various loads, for a hard hemisphere on
an elastic half-space. The view a compares
Hertzian contact and adhesive contact under
the same load. Whereas in Hertzian contact
the elastic material leaves the contact upwards,
in adhesive contact it leaves it downwards.
P1- P
II= -2- (76)
and eq. (21) may be verified. The force FS corresponds to a virtual

displacement along the equilibrium curve at an equilibrium point
such as N in Fig. 19.
UT
A
",(3<nfR2)~
... .... 0.5 2.5
.... ,,
\
\
\
,
\
\
-1.5
,
\
\ UTeA) "UT(A)
(P,< P)
,,
\(p,zP)
\ I
I
-2
Fig. 2'. Relations between total energy UT and area

of contact A as outlined in Fig. 2, with
five curves (UT)p for negative, positive
and zero load. The two branches of equilibrium
corresponding to the two determinations
P,-P = + 16nwRP, (heavy line) and P,-P =
- 16nwRP, (dashed line) are shown. Variation
of total energy in the Hertz' theory (w=o) is
also given for comparison (dotted line).
The Point B corresponds to 0=0.
The force IT corresponds to a virtual displacement at constant

load for any point such as N (equilibrium) or M (non-equilibrium).
At the rupture at fixed load IT=FS=~WR is the adherence force. For
rupture at fixed grips FS becomes infinite.
For transformations at fixed displacement, the internal energy
U = UE+US is the thermodynamic potential, and the Fig. 22 displays
its variations at fixed 0 or along equilibrium points.
-1 B
Fig. 22. Relations between internal energy U and area

of contact A with five curves (U~ for negative,
positive and zero displacement. As in Fig. 21,
the two branches of equilibrium (heavy and
dashed lines) and the Hertz' theory curve (dotted
line) are given. The point B corresponds to
P=o.
As proposed by Berry (32), the various energies can be repre-

sented on the (P,~) plot, Fig. 16. The elastic energy UE is the
area under the loading curve OJB, and the surface energy of the
system US' the area between the loading curve OJB and the locus of
Griffith's criterion ODeB (see Appendix 2). For a variation along
the equilibrium curve, eq. (23) leads to
-llJs = llJE - jPd ~

that can be checked on the Fig. 16 for a variation ~a displacing the
points J and B.
5.5 Crack Propagation at Fixed Load
Kinetics of Unloading: Let us consider the equilibrium (G=w)

under a load P, with a radius of contact a and a displacement ~. At
the time t=O, let us apply a load P'<P. Figure 19 shows that one
250 D. MAUGISAND M. BARQUINS
must observe an instantaneous variation of 6 at a constant area of

contact, followed by a simultaneous variation of 6 and a along the
curve ~)pI. The instantaneous displacement at constant a is that
of a punch under a load p_pl, and it is
M = 2{P-P I )/ 3aK.
Figure 19 shows that evolution along the curve (6)p' can lead to a
new equilibrium if pI >- ~1lWR or to rupture i f pI <- ~WR. In the two
cases the k1netics of the crack propagation can be studied by eq.
(32).
During crack propagation, following the curve (6)p' the strain
energy release rate is given by
(P _pI)2
1
G = 611RP 1
(77)
where pI is the new fixed load, and where P1 varies with the radius
of contact a according to the eq. (59) • The only underlying
assumption is that the displacement 6 is purely elastic, and that
viscoelastic losses are only at the crack tip.
Thus, the knowledge of the dissipative function ~(aTv) is the

only factor, with w, needed to predict all the features. When the
new load P I is applied, P 1 has the value corresponding to the
initial area of contact, so that G'W and the crack starts, with a
speed corresponding to G given by eq. (77 ) • As the crack pro-
gresses, a decreases; hence P 1 and G-w vary, so does the speed. The
evolution of the "normalized motive" (G-w)/w with the actqal radius
of contact is given in Fig. 23. One can see that if P'>- ~WR (i.e.
x>-O.5 on the figure), G gently decreases towards the value w3-that
is, a decreases towards a new equilibrium value. If pI<- 2nwR a
minimum in G may be observed at a critical radius of contact given
by
i.e.
= _pl. (78)
60 X:--
P'
3<nwR
G
--1
W
50
40 x=-20
/
/
30 x=-IS
",....10
20
/
/
10
Fig. 23. "Motive" of crack propagation (G-w)~ versus

instantaneous load P, or (radius a) for
different unloadings pI, in reduced coordinates.
If pI> - ~~R, (G-w) gently decreases towards
zero value and crack slows down; if
a 3K
--1- < pI < - ~1\WR
(a being the radius of the initial area of
coRtact), (G-w) first decreases, then increases
and crack 3slows down then speeds up; if
a K
pI< _~, (G-w) immediately increases and
crack thus continuously accelerates. The
straight line P,=_PI is the locus of minimum
value of (G-w) at fixed load conditions.
The value of this minimum. is

2P'
G = - 31fR.
Thus, for a given initial radius of contact a , this minimum of G
o
ma~ or may not be observed according to the value of pl. If P'<-
a o K/R, G continuously increases with time, and the crack acceler-
ates up to the rupture. If - ~ WR>p'>-a0 3K/R, G first decreases,
then increases, i.e. the crack slows down, then speeds up, so that
on a plot a(t) one has an inflexion point at the critical radius of
contact given by eq. (78).
Unloading Experiments: Experiments were made with spherical

caps of radius R=0.219 and 0.715 cm. The radii of contact were so
small «350 Pm) that the polyurethane plates (h=3.175 mm) were
considered as viscoelastic half-spaces.
The equilibrium eq. (72) and the adherence force P=- ~R have
been well verified these last years on smooth solids of low elastic
modulus •• (Johnson et al.(2), Barquins and Courtel (28), Fuller and
Tabor (38), Barquins et ale (29» or with small radius metallic
spheres (Easterling and Tholen (39), Kohno and Hyodo (40), Maugis et
aL (41». As stated above, the work of adhesion w was measured
before any set of experiments by measuring equilibrium radii of
contact under various loads. The only care to be taken is to be
sure that the equilibrium is reached. Figure 24 shows that kinetics
of loading are different from kinetics of unloading as first pOinted
out by Roberts and Thomas (42); by loading the equilibrium is
reached in 3 min, and further dwell time has no effect; by unloading
this equilibrium is not attained after 20 min. Equilibrium pOints
were obtained by loading with a dwell time of 10 min. The
reproducibility of radii of contact during a set of experiments was
better than 0.2% ••• •
• The existence of such a minimum was noticed by Roberts and Thomas

(42). They rearranged eq. (72) to have w versus a, and differen-
tiated the expression thus obtained. Their w so defined is the
apparent work of adhesion wf' and their minimum of wf is, in fact,
the minimum of G (see eq. (33» •
•• The observation of finite areas of contact under negative loads
seems to have been first published by Drutowski (44) •
••• So, as in Roberts and Thomas (42) experiments, no true hystere-
sis between unloading and overloading was found, contrary to the
results of Kendall (45).
I- 200
~
J-______ .'5000":"."_0_
l-
R=O.219cm
o
I 5 10
TIME,mn
15 20
Fig. 24. Radius of area of contact versus time of a

glass hemisphere in contact with a polyure-
thane surface, during an unloading from
p = 5,000 dynes to pI = 0 and during a
loading from P = 0 to pI = 5,000 dynes.
Kinetics of unloading and loading are quite
different. The equilibrium is more quickly
reached by loading.
Figure 25 shows the variation of the radius of contact. with

time for unloading, from P= 10000 dyn to various loads P 1< P. For
light unloading, a slow return towards new equilibrium is observed,
with continuously decreasing crackspeed (for PI=O the equilibrium
is not yet reached after 4 h). As expected, a slowing down of the
crack, followed by an acceleration until the rupture is reached, is
observed for strong unloading, 1. e. pI <- ~11WR=-61 dyn (similar
curves were given by Brunt (43) and DrutowsRi (44). The critical
radii of contact given by eq. (18) are reported on the figure; the
agreement with the observed inflexion points is quite good. The
dashed curve corresponds to the boundary between the two kinds of
~for punches, a grey annulus due to reflection of light along

the profile surrounds the area of contact as soon as one unloads.
Taking the outer radius as the radius of contact would lead to
large errors.
o
~ I!>o
Z
o
u
100
!>O
TIME . s
Fig. 25. Variation with time of the radius of the

contact area of a glass hemisphere in con-
tact with a polyurethane surface, for
unlo~dings from P = 10,000 dynes to various
loads pI < P. If pI> - ~1TwR = -67 dynes,

a slow return toward a new equilibrium state
is observed and crack speed continuously
decreases. If P'< - ~1TWR, the crack speed
decreases, then increases up to the rupture.
The value of radius a at the inflexion pOints
exactly corresponds to the computed minimum
value of the motive G-w.
unloading (P'=- ~1TWR); it was computed from data of Fig. 28. Note
that the time to rupture for an unloading from P to P' = -P is the
time of detachment of a ball of weight P under gravity.
Figure 26 shows the relation between the strain energy release

rate computed by eq. (77) and the crack speed v (deduced from the
slope of curve a(t) for unloading from P=10000 dyn to P'=-4000dyn.
The points corresponding to slowing down «a G/aA) ,>0) and these
corresponding to acceleration «a GIe) A)p'< 0) are indeed on a single
105 , - - - - - - - - - . - - - - - - - - - - ,
G
e rgs.cni 2
10'
P . IOOOO dynes
PO. -4 000 dynes
R. O.219 cm
9-23 ' C
103 1--_ _ _ _ _ _....1...::_ _ _ _ _ _--',
10 102 103
CRACK SPEED . pm.5-1
Fig, 26 . Relation between strain energy release rate

G and crack speed v = ~~, for an unloading
from P = 10,000 dynes to P' = -4,000 dynes
(see Fig. 25). All the points, corresponding
either to decreasing or increasing crack
speed, fallon a single curve.
curve .
The kinetics of unloading at 23 or 29 0 C are compared in Fig.

27. The crack speed increases with temperature, but the inflexion
points correspond to the same critical radius,
Results concerning various starting loads P, various final

loads P', with two radii R, and two temperatures are collected on
the Fig. 28 (Barquins and Maugis, 46). The results for 8=29 0 C were
shifted on those corresponding to 8=23 0 C by the WLF transform (with
T =_50 oC). This master curve is still the same as for peeling
(~eavy curve), and of course for adherence of punches. This figure
thus gives the dissipative function $(aTv) of the polyurethane,
eq . (51),
Elastic displacements have been recorded during crack propaga-

tion (Barquins and Maugis, 47). Figure 29 shows unloading from
P=5000 dyn to various loads P'=4000, 2000, 0, -2000 and -4000 dyn .
In two cases, the load was varied during unloading (from P'=O to -
2000 dyn, and from P' =-2000 to -4000 dyn). The agreemen t wi th
theoretical prediction (cf. Fig. 19) is excellent. Thus, as for
punches, the displacements are purely elastic, and viscoelastic
E
::1..
(J)
~200
o
<
a::
U
~
z
o
u
100
P = 5000dynes
P'=-3000dynH
R= 0.219cm
0 2 4 6 8 10
TIME,s
Fig. 21. Experimental curves a(t) for an unloading

from P = 5,000 dynes to pI = -3,200 dynes
at two different temperatures 23 C and 29 0 C.
The crack speed increases with temperature
but the inflexion points correspond to the
same critical radius.
losses are limited to crack tips.

Loading Experiments: As shown in Fig. 24 kinetics of loading
are quite different from kinetics of unloading. This point was
studied (Maugis and Barquins, 49).
On Fig. 30 is shown the path LMN for unloading from equilibrium

under a load P to equilibrium under a load PI<P. Starting now from
equilibrium under the load pI (point NI radius aN) let us steeply
apply the load P. A two steps path from N to L can be predicted.
The first step NT begins with an instantaneous displacement NS at
constant radius of contact aN corresponding to the displacement
2t£·
flO = 3a NK
of a punch under the load

"T1
1~~i----------~--------~--------~----------'---------~----------' :u
»
~
C
.9._1 :u
m
w 3:
m
(")
J:
ld »z
o
C/l
••\ "I:.it:'
~~ .. »z
~fi;;' • o
»o
10' .;'
.,. ,-r"
. J:
~',.. m
:u
- ~.,... p- 10000 5000 2000 m
~R < m 0.2'19 no o. ~ tt
Z
.. ~ (")
~I't 13 15 '3 '3 ~ ' 3 2J
'" 0.'",. m
~. ~'"
. 2000
0 ~
*' tI.;.~~·1 ••
- 2000 .
", ,. ~
.. - 3000
- 4000
·· -
~ ,:-.
·· .,
10 -5000
- 6000 p
.. . ...~
: ......
-,• . .
, ," - 8000
-9000 ·
. 0.6
- 10000
•
~
./,t" -15000
- 20000
-."t'
-· .
I
I
10-' 10 10' 103 10' 10'
CRACK SPEED ,,..m.s·'
Fig . 28. "Normalized motive" versus crack speed for adherence of glass hemisphere
on polyurethane, for various unloadings, and two radii and two temperatures.
The same master curve as for peeling (heavy line) and of course for unloading
of flat punch is observed.
!»
~
30
6,pm
20 \
\
'\
"'- '-
10
a,pm
P·.dynes
• 4000
" 2000
..• -20000
-10 II _ 4000
• 0.-2000
o -2000.-4000
P . SOOOdynK
R.O.219cm
K.8.9 107ctynn.coii
W.83 e'll"cnf'
-20
Fig. 29. Displacement versus radius of area of

contact for adherence of a glass hemisphere
on polyurethane, for various unloadings.
Agreement with theoretical curves is good,
even when the load is varied during unloading.
t.p* - P'
Hertzian conditions are realized at point S, and the path follows

the Hertz curve from the load P'+~* to the load P at point T. The
second step TL corresponds to crack healing at constant load along
the curve (o)p during which the crack slows down and stops. For
30
5,pm
25
...
"... ... "-
.....
20
15
,,
\
\
10 "-
"
5
R= 0.219 em
K = 9,5 .10' dyne<.cm-'
W= 10efos.cm·t
Fig. 30. Relations between elastic displacemento

and radius of area of contact a of two
spheres in an unloading-loading experiment.
After an unloading from P to pI, already seen
on Figs. 19 and 23 (path LMN), a loading
from P' to P leads to an instantaneous
displacement firstly at constant a (branch
NS) and then along the Hertz' curve (branch
ST) . This path (NST) is followed by an
evolution along the curve (o)p towards the
equilibrium point L. Experimental results
for loading step from P'=O and P'=4,OOO
dynes to P=5,OOO dynes are given.
light loading, i.e. ~p = P-P' ~ ~p', the Hertzian branch ST dis-

appears .
Thus the first step is the elastic response of a system with

zero surface energy. It is likely that during this fast motion, a
sheet of air is squeezed between the two solids, preventing molecu-
lar interactions (analogy with elastohydrodynamics) . During the
second step, the sheet of air is extruded and the radius of contact
progressively takes its equilibrium value.
Experiments were performed with a glass ball (R=0.219 cm) on
the polyurethane. Area of contact was recorded at 50 frames per
second, and the elastic displacement was recorded on a memory
oscilloscope. To avoid percussion, the loads were applied with a
damping system. The experimental points on the Fig. 30 are for
loading to P=5000 dyn from P'=O and p'=4000 dyn. (The numbers are
frame numbers.) Experimental results are in perfect agreement with
theoretical predictions.
The Hertzian evolution of the contact during loading (branch
as ST) is confirmed by the study of the surface profile by Newton
rings. For the same radius of contact, Fig. 31a,b,c show the Newton
rings and the profile for loading equilibrium at P=O, and unload-
ing. At equilibrium the contact angle is 91110 0 ; during unloading
the profile is as shown in Fig. 5c (compare Figs. 31b and 9); but
during loading the tangential joining is characteristic of a Hertz-
ian profile.
5.6 Crack Propagation at Fixed Grips

Starting from an equilibrium point one can apply a new dis-
placement 0 and record the variation of the force P during crack
propagation at fixed o. The predicted variation LMN tOW~dS a new
equilibrium is displayed in Fig. 18. For 0 < -
final state is the rupture of the joint.
i (~~ )2/3 the
The kinetics of crack propagation can be studie'3by eqs. (62b)

and (64). Figure 32 shows the variation of G with a at fixed 0 in
normalized coordinates. For any given couple (G,a) the slope at
constant 0 is larger than at constant P, for ~ ~)o > (~)p (see eq.
(41». As for crack propagation at fixed load, it exists a minimum
for G, so that the crack speed can first decrease then increase.
No experimental verification was attempted.
5.1 Crack Propagation at Fixed Cross-head Velocity

Starting from an equilibrium point one can impose a fixed
cross-head velocity ~ = ~ and compute the variation pet) from eqs.
(32), (60), (62) and (51 J •
The crack velocity is given by
[a]
[b] a I
~~...::- --;-- ---- - I6=3~ +1a'R:

e
[c]
Fig. 31. View of Newton's rings and corresponding

diagram profile of the surface for loading,
equilibrium at P=O and unloading, with the
same radius of contact. The view@conrtrns
the Hertzian evolution of the contact during
loading (branch S T on Fig. 30).
and its variation by
From the same calculation method as for punch (see sec. 4-5) one can
compute the radius ai' corresponding to 0i = 0i_1+6~t, by
z- 6
-[~'3ItP
Fig. 32. "Motive" of crack propagation (G-w):3 versus

instantaneous load P, or (radius a) for
different fixed grips conditions, in reduced
coordinates 2
If 15 > - 3: (Le. z> -1.310), G-w gently
decreases towards ~ero value and crack slows
down. If 15 < - 3: ' the "motive" G-w can de-
crease and then increase, or can immediately
increase.
The straight line 5P 1 + P = 0 is the locus
of minimum values of G-w at fixed grips
conditions, and the straight line 2P 1+P = 0
is the locus of inflexion points.
2
da. d a. 2
A£.
a i = a i _1 - I dt ~I&_I-~I
dt2 2
LJI,
hence the force
a. 3K 3Ro.
Pi = ~R (--~ - 1)
ai
2
Figure 33 displays such computed curves for various S. The

maximum P depends upon 8 and the starting equilibrium point.
max
Experimental verifications are in progress.
6. PEELING
6.1 Theory
Consider (Fig. 34) the peeling at an angle a of a thin elastic

strip (thickness h, width b) adherent to a rigid substrate, on to an
area A. The angle a can vary from a=O (lap shear joint) to a =1T (1T
o
peeling); the film may have been extended to a small strain € by a
tensile force F before application to the substrate so that an
elastic energy
F2 2
u---=hEh (79)
- 2b 2Eh
is stored by unit area of the strip. Neglecting the peel bend
energy and end effects which remain constants, the potential and
elastic energies are
Up = -Po =~-cos
~ b
a + P-F ) P(A-A )
b2Eh 0
p2_F2 F2
U - - - - (A-A) + - - A
E - 2b 2Eh 0 2b2Eh 0
Hence the strain energy release rate
(P_F)2 P
G= 2 + b (1-cos a) (80)
2b Eh
and the equilibrium relationship
~
\
\
\
\
\
\
\
p.-(~)
" '- .......

....... --..----------
-----\
\
!='
~
.~.=.~
>
cC)
o TIME
Cii
>
2
o
Fig. 33. Computed curves, for various fixed cross-head velocities, ~
III
of the variation with time of the adherence force of a >
rigid sphere in contact with a viscoelastic surface. ::t1
o
C
2
CJ)
, ("
I I
I
I I
~ 'p
Fig. 34. Thin elastic strip peeling at an angle
~ from a rigid substrate .
(P_F)2 P
-=-:~- + -b (1-cos ~) = w. (81)
2b 2Eh
As dG/dA = 0 (G independent of A), the corresponding equilibrium
load P is the quasistatic force of adherence. Under this load the
c
peeled film is in indifferent equilibrium. For P>P , the crack
speed is constant during peeling because G is indepen~ent of A.
Equation (81), with F=O, was given and experimentally verified
(at constant crack speed, 1. e. constant w~ by Kendall (4). The
approximation (Rivlin (50), Deryagin and Krotova (49), Kendall (1»
P = bw/( 1-cos ~)
is better than 6% when 1-cos ~>2(2w/Eh)1.

The total energy of the system is
UT = (G-w)A-(G-u)A o (82)
and under the load P , it is independent of A. Figure 35 shows the
c
variation of the total energy with the area of contact, in reduced
coordinates. It is interesting to compare this figure with Fig. 6:
the normalized curves (UT)p have now a constant slope G/w-1. As
crack propagation corresponds to a positive slope, it is easy to
study spontaneous propagation (i.e. under zero load); for P=O one
has G=u, and Fig. 35 shows that spontaneous propagation occurs when
u/w>1. Thus, an elastic strip cannot tolerate a stored elastic
A
0.5 Ao
....
\V .. o\ •..•• ······
-2 G,,~ •• ••·••••
........
-.!:! •••••••
w
-5
Fig. 35. Variation of the total energy of peeling

with the area of contact, in reduced coor-
dinates. The equilibrium curve (dashed line)
is a straight line corresponding to a constant
load. All these curves are thus curves (UT)p
with a constant slope Q - 1. For P = 0,
w
spontaneous crack propagation is observed
when the elastic energy u stored by unit area
of the strip and due to prestress, is higher
than w.
energy per unit area higher than w. But the influence of a

prestress is subtle: a strip has two ends, and if one end is clamped
to avoid spontaneous propagation, the force needed to propagate the
other crack may be increased by a pre-stress.
The influence of prestress was studied by Kendall for <X=O

(Kendall, 5,6) and for ~1T/2 (Kendall, 5,3). In fact for any value
<x, there is a critical tensile prestrain € above which the adherence
force is lowered (Maugis, 53). Equation (80) may be rewritten as
G(F) - G(O) - F(F-2P) (83)
- 2b 2Eh
which shows that adherence is increased i f F <2P . This critical

prestrain is twice the strain in the peeled film, due to the load
P. For a:90 0 its value is about 2w/Eh, i.e. very small.
c
Note that the displacement 0 is still given by the generalized
Castigliano's theorem 0 = -(oUT/oP)A, and that the compliance of
the system is
00
= (op) A
6.2 Experiments
Polyurethane strips (thickness h=3.175 mm, width b=1-3 cIY'

length 10-20 cm) were peeled from glass at various peel angles 10 ,
30 0 , 60 0 , 90 0 • Experimental precautions to have reproducible
results are the following: (i) cleanliness; (ii) constant tempera-
ture (23 0 C+0.5 0 ); (iii) absence of prestress. (the strips were
gently laid-on the glass, and a dwell time of about 15 min was used,
during which stress relaxations by microslips and squeezing of air
pockets probably occurs); (iv) constant bending (to have no varia-
tion in bending energy during peeling). The weight of the peeled
strip was, of course, taken into account in the applied load.
Figure 36 shows the relation between the crack speed v and the
"normalized motive" (G-w)/w with G computed by eq. (80) and w=65 erg
cm- 2 From eq. (32) this curve thus displays the dissipation
function ~(aTv) to which the punch and sphere results were compared.
There is a marked dispersion in the results that can be assigned to
temperature variations during experiments, local variation of the
interfacial energy due to dirty areas, local prestresses in the laid
strips, or variation in room humidity. The dissipative function
~(aTv) varies as (aTv)O.6, a result often found in peeling.
Birch et al.(54) have shown that humidity increases the peel
rate of rubber on glass, and this result together with peeling
experiments in various liquids by Gent and Schultz (26) or on
various substrate by Andrews and Kinloch (27) can be analyzed by a
reduction of w in eq. (32). (As said in sec. 3.2 the Dupre energy of
adhesion was observed to vary from day to day most probably with the
• It was observed that a prestrain e:=3% decreases the adherence

force for a:1T/2 but increases the adherence force for a.=10 0 • More
precisely, G is increased by this prestrain for a.=1T 12 and the
crack speed increases; G is decreased by this prestrain fora. =10 0
and the crack speed decreases. This is in agreement with eq.
(80) .
G
--1
w
o 90.
o 60.
o 30·
a 10'
lL----~----~------~-----_7.~----~._-----~~
10" 10 10 2 10 3 10" h,-
CRACK SPEED pm.s-'
Fig. 36. Peeling of polyurethane strips from glass

surface: I1 normalized motive l1 of crack
versus crack speed at various peel angles
o
(temperature a = 23 C + 0.5 Cj n~ prestressj
0
dwell time 15 mnj w = 65 ergs/cm ).
humidity content of the atmosphere). Roberts (55) has measured the

rolling resistance T of a glass cylinder (length t) rolling on our
polyurethane samples as a function of the humidity content of the
atmosphere (Fig. 31). His data points clearly to a slope 0.6 in
agreement with our results, and as rolling on adhesive substrates
corresponds to peeling at the rear of the contact (Kendall, 45,
t-
Roberts and Thomas, 42), Fig. 35 and Fig. 31 can be compared with
G= As G >::w in Fig. 31, one has G"'w(j> (aTv) and this figure
clearly supports the multiplication effect of w on viscoelastic
losses, as proposed by Andrews and Kinloch (21).
7. CONCLUSION
The introduction of the concepts of fracture mechanics such as

the strain energy release rate G and its derivative a Gia A has
allowed us to give a general picture of the adherence of visco-
elastic bodies. Two bodies in contact over an area A are in
equilibrium if G=w, where w is the thermodynamic (or Dupre's) work
of adhesion. This equilibrium is stable i f aG/aA is positive,
~/ Y ..V
/
A
d"'~- A/-~
~y·/~e. ·
10'~· /'> /e
,,0 . ~O'6
o 6 y' RH.
• J6
4 85
elOO
~ > ~ ________- L__________ ~ ________ ~ ___
1 ~ ~ ~
velocity . pm.S-'
Fig. 37. Effect of relative humidity on the rolling

resistance of a glass cylinder on a poly-
urethane surface. e = 2'oC. (Roberts, 56)
unstable i f negative. The quasistatic force of adherence is the

load corresponding to aa/ ClA=O. But equilibrium may be stable at
fixed grips conditions and unstable at fixed load, so that the
quasistatic force of adherence may depend on the stiffness of the
measuring apparatus. When G>w, the separation of the two bodies
starts, and can be seen as the propagation of a crack in mode 1. G-w
is the force applied to unit length of crack; under this force, the
crack acquires a limiting speed v which is a function of the
temperature, and one can write
G-w=~ (aTv).
The second member is the drag due to viscoelastic losses at the
crack tip, and is proportional to w as proposed by Gent and Schultz
(26) and Andrews and Kinloch (27). For a viscoelastic adhering to a
rigid body, the dimensionless function ~ is a characteristic of the
viscoelastic material, i.e. independent (i) of the nature of the
rigid body, (ii) of the geometry of the contact and (iii) of the
loading system. When ~ is known, this equation enables us to
predict the kinetics of detachment, provided that the failure is an
adhesive failure, and the viscoelastic losses are limited to the
crack tip. This last condition means that gross displacements must
be elastic for G to be valid in kinetic phenomena. The interest of
the proposed formula is that surface properties and viscoelastic
losses are clearly decoupled from elastic properties and loading
conditions that appear in G.
Three geometries have been investigated: adherence of spheres,

adherence of punches, and peeling. The quasistatic force of
adherence force, as given respectively by Johnson et al. (2) and
Kendall (14) are, of course, found; the variation of energies with
the area of contact is given; and the variation of the "motive" G-w
with the reduction of the area of contact is studied: G-w is a
constant for peeling, it continuously increases for punches, it may
decrease, or decrease then increase, or continuously increase for
spheres, according to the applied load and the area of contact.
Crack propagation at fixed load, fixed grips and at fixed cross-head
velocity are studied with implications on tackiness.
ACKNOWLEDGEMENTS
This research was supported by a grant from Directron des

Recherches et Moyens d' Essais DRME 1177-1037 and a grant from Delega-
tion Generale a'la Recherche Scientifique DGRST #78-7-2667.
APPENDIX
The stress distribution for the punch problem is given by eq.

(35) with the following asymptotic value near the crack tip:
P j~ K1
(84)
°z = 21Ta "2p = 121TP
2
where K1 is the stress intensity factor:

P
(85)
K1 = 21T 112 a 312
The discontinuity of displacement at the crack tip is given, from

eq. (38) , by
4p ~
[uJ = o-u z 1-u z2 = 201T ~
a
- -
- 31TaK
£Q.
a
(86)
where indices 1 and 2 are for solids 1 and 2, and P = r-a. Eq.
(86) can be written
16 K1 Ji>
[u z l = ~ "2; (87)
that reduces to eq. (29) for plane strain in the case of two
identical solids.
It can be seen that for two identical solids K1 and G, eq.

(41), are linked by the Irwin relation, eq. (28), for plane strain.
This means that GdA is the work done by punch stresses during crack
propagation. Let us verify this point in the general case of
contact of two different elastic solids. Let us compute the work
dUE needed to close, at fixed grips, a crack such as in Fig. 5d,
from radius a to radius a+da, by progressively applying stresses on
the two sides so that the final stresses are given by eq. (84) when
the crack is completely closed.
=1
a+da
dUE ~
a z [U z] 21Trdr
a 1
where the factor 2" arises because of the proportionality between
traction and displacements. The relevant stresses are those across
the annulus, prior to extension, i.e., those corresponding to
p = a+da-r, and the displacements are those across the annulus
prior to closure, i.e. those corresponding to p = r-a:
8K 1
dUE = ~
2
f a +da r-a ~
(a+da-r) rdr
a
that gives after a lengthy but simple calculation, and for da~
2K 2
1
dUE = 3K
21Tada
P2dA
= GdA
The equatio~
2K1
G=~ (88)
is the Irwin relation for plane strain, eq. (28), in the case of two
different solids.
The stress distribution for adhesive spheres is given (Johnson

58) by way of the superposition of a Hertzian distribution with a
load P1 and a punch distributiion with a tensile load p 1-p
3P1 2 ~ P -P 2 -~
a
z = - --2 (1 - -r 2) +- 1 - (1 - -r 2) (89)
21Ta a 21Ta 2 a
The asymptotic value near the crack tip is only due to the punch
stresses, and is given by
272 D. MAUGIS AND M. BAROUINS
P 1-P ~@'
Oz = -2
1Ta
2 12P (90)
as given by Savkoor and Briggs (56). Relation between G (eq. (62»

and K1 is still given by eq. (88), that means that only the punch
stresses significantly contribute to the work GdA during crack
extension.
Thus, the stress intensity approach only takes into account

the asymptotic value of the punch stresses near the crack tip. A
similar method was followed by Pollock et al.(57) for calculating P1
in any axisymmetric case by the energetic approach. Suppose that w
is at first set equal to zero, while a load P1 (not necessary given
by eq. (59» is applied to give a contact radius a equal to the
observed valu y• The, elastic and potential energies at this stage
are denoted UE and Up. Suppose the load to be reduced to the final
value P while a remaining constant and the surface energy Us = _1Ta 2w
added: the elastic and potential energies changes are those due to
a punch displacement under the tensile load P 1-P. The total energy
is now
As for a punch G-w can be defined as the variation of UT with ~

varia~ion da at constant punch tensile load P1-P (during which UE
and Up remain constant). Equilibrium is thus given by
2
[ aaa 2
(-1Ta w -
(P1- P ) ]
3aK ) (P -P) = 0
1
Le.
(P 1-P)2
G = =w
6JT a 3K
valid for any axisymmetric case.
APPENDIX 2
Calculation of the sector area S bounded by the loading

curve OJB and the equilibrium curve BCDO.
To simplify, let us use the reduced coordinates
y =(~)
1TwR '/3 and x = 3~R in which:
3K2
the equilibrium curve equation is

2/3 2x
YJ = x, + ~
x,
P, .
with x, = 3nwR l1nked to x by x=x, - ~
the Hertz curve is defined by

2/3
YH = 3x
the loading straight line BJ is given by
YB
2X
= --v3
x,
+ x, 2/3
So, the area can be written:
x,
=I
o
= 59 x,
5/3
-
3 «-
y~ x,
7/6 3 2/3
+ x,
and x, 2X 2/3
IBJ yBdX = I (~ + x, )dX = 3 h" x, 116 -2 x, 2/3
x x,
0
and I yHdx = I 3l/3dx = - -9 x 5/3
JO 5 1
x,
Then S = +x,2/3, which exactly corresponds to the absolute value

of the surface energy at the equilibrium point B.
REFERENCES
1. K. Kendall, !!.:.. Phys. 12.: !P.Eh Phys. !!" 1186-95 (1971).
2. K.L. Johnson, K. Kendall and A.D. Roberts, Proc. R. Soc. Lond.
! 324, 301-13 (1971). -----
3. K. Kendall, !!.:.. Phys. Q.:!P.Eh Phys. 2" 1782-7 (1973b).
4. K. Kendall, !!.:.. Phys. 12.: !P.Eh Phys. !!, 1449-52 (1975g).
5. K. Kendall, !!.:.. Phys. 12.: !P.Eh Phys. !!, 1722-32 (1975h).
6. K. Kendall, !!.:.. Phys. 12.: !P.Eh Phys. !!, 512-22 (1975c).
7. K. Kendall, Proc. R. Soc. Lond. A 344, 287-302 (1975e).
8. K. Kendall, J. Adhesi0ri"7, 137-40' (1975f).
9. K. Kendall, Proc. R. Soc: Lond. A 341, 409-28 (1975a).
10. K. Kendall, J. Mat:-Sc:-11, 638-44-c1976a).
11. K. Kendall, ~ Mat. Sc. 11, 1263-6 (1976b).
12. K. Kendall, ~ Mat. Sc. 11, 1267-9 (1976c).
13. K. Kendall, J. Mat. Sc. 10, 1011-4 (1975d).
14. J.R. Rice, "Fracture: An Advanced Treatise", ed. H. Liebowitz
(New York: Academic Press), Vol. 2, p. 191~311 (1968).
15. C. Huet, Ind. Minerale 1, 128-41 (1973).
16. D. Maugis and M. Barquins, !!.:.. Phys. 12.: !P.Eh Phys . .l!., 1989-
2023 (1978b).
17. H.B. Callen, Thermodynamics (New York: John Wiley & Sons)
(1960) •
18. R. Courtel, D. Maugis and M. Barquins, Industrie Minerale !!"
137-43 (1977).
19. H.D. Bui, Mecanigue de la rupture fragile (Paris: Masson)
(1978 ).
20. J.D. Ferry, "Viscoelastic Properties of Polymers", 2nd edition
(New York: John Wiley & Sons) (1970).
21. D.H. Kaelble, J. Colloid Sc. 19, 413-24 (1964).
22. D.H. Kaelble, ~Adhesion-'; 102-23 (1969).
23. A.N. Gent and Ft.P. Petrich~ Proc. R. Soc. Lond. A310, 433-48
(1969). -- - -- ----
24. A.N. Gent, !!.:.. Polym. Sc. A-2 ~, 283-94 (1971).
25. A.N. Gent and A.J. Kinloch, !!.:.. Polym. Sc. A-2~, 659-68 (1971).
26. A.N. Gent and J. Schultz, J. Adhesion 3, 281-94 (1972).
27. E. H. Andrews and A. J. Kinloch, Proc. R:- Soc. Lond. A 332, 385-
99 (1973). --------
28. M. Barquins and R. Courtel, Wear 32, 133-50 (1975).
28 bis. A.D. Roberts and A.B. Othman~Wear 42, 119-33 (1977).
29. M. Barquins, R. Courtel and D. Maugis, Eighth World Conference
on Non-Destructive Testing, Cannes (France), Paper 4 B 11
(1976).
30. J. Boussinesq, "Application des Potentiels" (nouveau tirage),
(Paris: Blanchard, 1885 (1969).
31. S. Way, !!.:.. !PPl:. Mech. ASME I, 147-57 (1940).
31 bis. K. Kendall, J. Adhesion 5, 77-9 (1973a).
32. J.P. Berry, !!.:.. Mech. Phys. Solids!!, 194-206 (1960).
33. D. Maugis and M. Barquins, C.R. Acad. Sc. Paris B, 286, 1-4
(1978a). -- -- - -- - -
34. K.L. Johnson, "The Mechanics of the Contact Between Deformable

Bodies", eds. A.D. de Pater and J.J. Kalker (Delft: Delft
UP), p. 26-40 (1975).
35. M. Barquins, D. Maugis and R. Courtel, C.R. Acad. Sc. Paris ~
279, 565-8 (1974).
36. R.J. Good, J. Am. Chem. Soc. 74, 5041-2 (1952).
37. G.A.D. Briggs and Briscoe, B.J:, ~ Phys. Q: !E.Eh Phys. jQ,
2453-66 (1977).
38. K.N.G. Fuller and D. Tabor, Proc. R. Soc. Lond. A 345, 327-42
(1975). -- - - -- --
39. K.E. Easterling and A.R. Tholen, Acta Met. 20, 1001-8 (1972).
40. A. Kohno and S. Hyodo, ~ Phys. Q: Appl:-Phys. I, 1243-6
(1974) .
41. D. Maugis, G. Desalos-Andarelli, A. Heurtel and R. Courtel,
ASLE Trans. 21,1-19 (1978).
42. A.D. Roberts and A-.G. Thomas, Wear 33, 45-64 (1975).
43. N. Brunt, Rheologica Acta 1, 242-7 (1958).
44. R.C. Drutowski, Trans. ASME: ~ Lub. Techn. 91F, 732-7 (1969).
45. K. Kendall, J. Adhesion 7, 55-72 (1975b).
46. M. Barquins and D. Maugis, C.R. Acad. Sc. Paris B 285, 125-8
(1977). -- -- - -- --
47. M. Barquins and D. Maugis, C.R. Acad. Sc. Paris ~ 286, 57-60
(1978).
48. K. Kendall, Wear 33, 351-8 (1975i).
49. D. Maugis and M. Barquins, C.R. Acad. Sc. Paris B 287, 49-52
(1978c). -- -- - -- --
50. R.S. Rivlin, Paint Technol. 9, 215-6 (1944).
51. B.V. Deryagin and N.A. Krotova, Dokl. Akad. Nank SSSR 61, 849-
52 (1948). -- -- -- ---
52. B.V. Deryagin and N.A. Krotova, Chem. Abstr. 43, 2842 (1949).
53. D. Maugis, Le Vide No. 186, 1-19(1'977). -
54. D.A. Birch, J.T. Evans and J.R. White, ~ Phys. Q:!E.Eh Phys.
jQ, 2003-10 (1977).
55. A.D. Roberts, Rub. Chem. Tech. 52, 23-42 (1979).
56. A.R. Savkoor and G.A.D. Briggs, Proc. R. Soc. Lond. A 356, 103-
116 (1977). ---------
57. H.M. Pollock, D. Maugis and M. Barquins, !E.Eh Phys. Lett. 33,
789-9 (1978).
58. K.L. Johnson, Brit. ~ !E.Eh Phys. 1, 199-200 (1958).
NOTES ADDED IN PROOF
1) As K1 is proportional to the discontinuity of displacement

[ uzl (eq. 29), Kl and G cannot be nega ti ve. Hence the minimum value
of K, and G during crack closure (G < w) are K, = 0 and G = 0 at
which the stress singularities disappear. During crack closure at
G = 0 the work done by the stresses at the crack tip and the
viscoelastic losses are negligible. Crack closure under llP is a

very fast phenomenon compared with the reverse crack propagation
under -&.
In loading experiments with spheres (Fig. 30), the branch NS

corresponds to the transition from G = w to G = 0, and the Hertz
branch ST to propagation at G = 0 (Le. P1 = P) with no stress
singulari ty • No air sheet (w;iO) is needed to explain this fast
propagation. The same explanation probably holds for rolling of
cylind~rs on adhesive substrates: the crack closure at the front is
z = 0, i.e., G = 0, and the rolling resistance corresponds

with (u]
to ~ peeling at the rear with G » w.
2) The vertical displacement u of the surface of an elastic half-
space indented by a rigid elliptical paraboloid (Hertzian contact)
was given by Boussinesq (1885). For a sphere it reduces to
u
z
= 3P
4a
1-1
nE
~r22 _ 1) 1/2 _ (r2 _ 2) Arc tg (i
2 2
_ 1)-1/2.]
a a a
that is equivalent to the eq. (15) given in the paper by Derjaguin
et ale (ref. a). For an adhesive contact (w ;i 0) punch displacement
(eq. 36) under tensile load P1 - P must be added to Hertzian
displacement under load P1 ; hence the vertical displacement
a2 [ 2
r
Uz = nR (~-
a
1)
1/2 2 2
- (r2 - 2) Arc tg (r2 -
a a
1)
-1/2
4
- 3" Arc sin ~
2
+ 1 - v -P Arc Sl.n
. -a
n Ear
that describes the equilibrium profiles drawn on Fig. 20, as well as

the non-equilibrium ones on Fig. 31. The quantity
would be added to have the profile of an adhesive sphere in contact

with a rigid plane as in ref. b.
REFERENCES
a) B.V. Derjaguin, V.M. Muller, Yu. P. Toporov, ~ Colloid Inter-
face Sci. 53, 314 (1975).

b) D. Tabor, J. Colloid Interface Sci 58, 2 (1977); J. Colloid
Interface Sci. 67, 380 (1978). -
The Viscoelastic Shear Behavior of a
Structural Adhesive
Erol Sancaktar*, Instructor

H. F. Brinson, Professor
Virginia Polytechnic Institute and State University
Blacksburg, Virginia 24061
ABSTRACT
The increased use of adhesives in the automobile, aerospace
and other industries is noted. The need to understand the visco-
elastic nature of polymeric adhesives is assessed. Previous bulk
tensile constant strain-rate, creep, relaxation, and delayed fail-
ures studies on an epoxy adhesive are summarized. New constant
strain-rate and creep data for symmetric rail shear and symmetric
lap shear geometries of the same epoxy adhesive are reported. It is
shown that significant viscoelastic shear behavior, including de-
layed failures, do occur. The linear viscoelastic methods and
results from previous tensile studies are extended to the condition
of shear. Predictions are shown to be at variance with experimental
results. The reasons for deviations are discussed; Le., the
possibility that linear viscoelastic analytical procedures are
inappropriate and the likelihood that viscoelastic phenomena are
different in bulk tensile, bulk shear and lap shear geometries.
INTRODUCTION
In recent years a great deal of attention has been given to the

characterization of light but strong polymeric and polymer based
composite materials. The application of these materials in the
* Presently, Assistant Professor, Clarkson College, Potsdam,

N.Y.
279
280 EROL SANCAKTAR AND H. F. BRINSON
automotive, aerospace and other industries have made the reliable

design of adhesively bonded joints very desirable. Often adhesive
bonding techniques either augment or eliminate the need of bolted or
riveted joints. A desirable feature of adhesive bonding is that it
is readily adapted to assembly-line manufacturing procedures and
stress concentrations arising from penetration techniques may be
avoided or minimized.
The purpose of the present investigation was to study the

viscoelastic shear properties of a particular epoxy adhesive, Metl-
bond 1113 and 1113-2**. This was to be done using a symmetric rail
shear test method on bulk panels and symmetric single lap shear test
method for bonded joints. Specifically, constant strain rate and
creep tests were to be performed on these specimens to ascertain
viscoelastic properties including delayed failures. While such
properties had been previously determined using bulk tensile test-
ing (1,2), the primary objective of the current program was to
determine i f bulk and lap shear viscoelastic properties could be
predicted from the earlier bulk tensile results. Prior to describ-
ing the current effort, it is therefore appropriate to describe the
earlier bulk tensile studies and our characterization procedures
used there (1,2).
Tensile Properties and Their Characterization

First, consider an idealized stress-strain response which we
will subsequently show is a reasonable approximation to the adhe-
sives studied herein as well as other polymeric materials (3,4).
This idealized stress-strain response shown in Fig. 1 contains a
region of linear elastic behavior followed by a region of visco-
elastic behavior terminating in a region of plastic flow. It is
i~portant to note that the linear elastic limit stress, a, is not
identified as a yield stress as is often done. Rather, the maximum
stress, Y, is identified as the yield stress. The values e and Y
are rate and time dependent. Thus, the idealized material of Fig. 1
exhibits time dependent incipient yielding or failure as opposed to
the time dependent subsequent yielding discussed elsewhere (5).
The idealized response described above is compatible with a

mathematical theory of viscoplasticity by Nagdi and Murch (6) and a
delayed yield response of Crochet (7). The latter is shown schema-
tically in Fig. 1 and is written analytically as,
** Metlbond 1113 is described by its manufacturer, Narmco Corp.,

as a 100% solid modified epoxy film with a synthetic carrier
cloth and is supplied in tape form. 1113-2 is the same as 1113
without the carrier cloth. Each is rubber toughened.
STRUCTURAL ADHESIVE 281
Y(e --)
STRAIN
Fig. 1. Idealized Stress-Strain/Strain-Rate

and Yield Behavior.
Y( t) = A + B exp ( -C x> (1)
where Y is the time depenaent yield stress, A, B, C are material

constants and where X is a time dependent parameter defined by
(2)
In (2) the terms e:. .'J and €i' E represent the viscoelastic and
elastic strains resp~~tively. J
In order to use equations (1) and (2) it is necessary to have a

mathematical representation of € .
1J
.v
and € . . E. A relatively simple
1J
procedure to accomplish this task is through the use of a mechanical
model which includes a sliding friction or a yield element. Several
models which have been proposed for this purpose are shown in Fig.
2. The modified Bingham model used by Brinson (3) demonstrates the
same phenomenological behavior as that shown in Fig. 1, Le., a
region of linear elasticity, followed by a region of linear visco-
elasticity terminating with a region of plastic flow.
The differential one-dimensional constitutive equation for the

modified Bingham model can be written as,
e = alE (J < e
e<a < Y
where E is Young's modulus, ~ is the viscosity coefficient and e and

Yare as previously defined. It is interesting to note that
equations (3) are similar to those used by others such as Malvern
(10) and Perzyna (11) which were based on an empirical non-lineari-
zation of a simple Maxwell model. Again, the interpretation of
yielding herein is quite different than that of the persons just
cited.
Solution of equations (3) for constant strain rate, e:, gives,
a( e:) = Ee: a < e

a(e:) = e + ~E: [1 - exp -(e: -.~)EJ
~e:
e< a <Y (4)
BINGHAM MODEL[8] a
SCHWEDOFF MODEL[8]
a r
Y
a a
MODEL USED BY MODEL USED BY
CHASE &GOLDSMITH[9] BRINSON[3]
Fig. 2. Various Viscoelastic-Plastic Models.

Solution of equations (3) for the case of creep and substitu-

tion into equation (2) gives the following specific form for the
time parameter, X, needed in equation (1),
X =°0 - e t [1 - 2 J] i (5 )
lJ
where °
is the creep stress and" is Poisson's ratio which has been
assumedOto be a constant. Crochet's delayed yield equation can now
be written as
I(t) = A + B exp[-K(Oo - e)t] (6)
where K = C(1 + 2\)2)I/lJ is a constant and a modified Bingham model

has been used to describe the stress-strain behavior. Equation (6)
represents a creep to yield phenomenon.
It might be noted that Crochet's model is not readily adapted
to quantifying the variation of elastic limit stress and yield
stress as a function of strain-rate. As a result, in our previous
studies we have used the following empirical equations due to Ludwik
(1) ,
e = e' + e" log(E:le:')

I = I' + I" log(E:le:') (7)
where e', e", I', I" and e:' are material constants.
As mentioned earlier, constant strain-rate, creep and relaxa-
tion stUdies were performed on bulk tensile coupons of Metlbond 1113
and 1113-2. Figure 3a shows the constant strain-rate response of
1113. It may be observed that the material displays the same
elastic, viscoelastic and plastic response regimes as the idealized
material shown in Fig. 1 and the modified Bingham model shown in
Fig. 2. Further, it is demonstrated that the modified Bingham model
can be used to closely approximate the measured behavior. It
should, of course, be mentioned that a different relaxation time
must be used for each strain-rate in order to achieve a close fit of
the data. The relaxation times needed are shown to vary nearly
linearly with strain-rate in Fig. 3b. While the need for a variable
relaxation time renders the modified Bingham model to be empirical,
it does not negate the utility of the constant strain-rate stress-
strain response curve fitting procedure. Obviously, the procedure
can be used quite accurately for interpolation and extrapolation
purposes (4). That is, gi ven the stress-strain response for a
single strain rate, the response for other strain rates can be
closely predicted.
Figure 3c shows the variation of yield (maximum) stress with
strain-rate compared to equation (7). Apparently, Ludwik's equa-
70.---,----.--~r---~--=la
60 • EO· (7) 4
8
50 3
...
'"
~40
<.!:
0 2
...J
V') V')
V') V')
~ 30 E e V w
ex 0 1113
t-
V') o.~_ (1 4t-
V') 0 1113-2
II
0
20
Q O.68m ~ /5EC -4 -3 -2 -1 0
06. 80m ~ /5EC 2 LOr, €
c 65.0m ~ /SEC
10 o 670m %/5EC Fig. 3b. Relaxation Time
• RUPTURE Versus Strain Rate for the
Modified Bingham Model.
4 5
Fig.3a. Stress-Strain/Strain-
Rate Behavior of Metlbond
1113 and Comparison with the
Modified Bingham Model.
MACHINE HEAD RATE (CM/MIN)

10- 3 10- 2 10- 1 100
70,.,-----,-----.-----,----....;..;;.-----.10
~ 60'"
V')
III
~
VI
~ 50 .. 7.5~
t- V=V' + o'LOG €/€' ~
V')
o V'= 58.12 MPa] ~

~ 40 [0'= 3.79 MPa ~
>-
e'= 1%/5EC
5
10- 1 10 0 10 1
STRAIN RATE (X/SEC)
Fig. 3c. Variation of Yield Stress and Comparison

with Ludwick's Equation for Metlbond 1113.
tion is an excellent representation of the variation in 'tE::nsile

yield stress. Similar results were found for elastic limit stresses
also (1).
Figure 3d shows the time for yielding to occur at various creep

stress levels. It is to be noted from Fig. 3a that the yield
(maximum) and failure (rupture) stresses are the same. As a result,
the data points shown in Fig. 3d actually represent the time for
creep to failure (rupture) to occur. The results are compared with
the predictions of equation (6) which indicates Crochet's model to
be a reasonable representation of a bulk tensile delayed failure.
phenomenon.
70 .--.------.----,,--,...----::l10
60
8
50
o
§:'" 40 ~------------------------
t F, TIME TO FAILURE ·
V')
V')
410 MIN
~ 30 o = 40.64 MPa
l-
V')
- EQ. (6)
--0-- } EXPERIMENTAL
20
2
10
o ~__~__~____~__~__~
o 10 20 30 40 50
TIME (tHN)
Fig. 3d. Comparison of Creep to Yield Data and

Theory for Metlbond 1113 •
• It should be recognized that Crochet's model was developed

only for a delayed yield process. We have elected to interpret
his model to represent a delayed rupture process as well.
Shear Properties and Their Characterization
As stated in the introduction, the objective of the current

effort was not only to measure shear viscoelastic-plastic-failure
properties but to attempt to predict shear results from the tensile
results described in the previous section. The symmetric rail shear
and symmetric lap shear specimen geometries shown in Fig. 4 were
used to obtain bulk and bonded shear properties respectively. These
specimen geometries have been shown to be reasonable approximations
of pure shear behavior (12-14).
It is easy to reformulate equations (1) - (7) for the case of

pure shear. All that is necessary is essentially to replace tensile
stress and strain variables everywhere by their shear counterpart.
However, the prediction of shear properties from tensile properties
using equations (1) - (7) requires a clearly defined relationship
SPEC mEN
o o o
o o o
o o o AOHES I VE -+-.....A
AOHERENO
o o o
o o o
P
Fig. 4a. Symmetric Rail Shear Fig. 4b. Symmetric Lap Shear
Test Specimen. Specimen.
between shear and tensile elastic limit and yield variables. Equa-
tions are needed at the termination of two response regimes--linear
elastic and linear viscoelastic.
First consider the linear elastic region in which the follow-
ing relations are valid for uniaxial tension:
(8)
where the subscript, t, refers to tensile variables. Hookes law for

shear and the relation between moduli are
G = E/2( 1 + v) (9)
where the subscript s refers to shear variables. Using the Tresca

or maximum shear stress failure theory, it readily follows from (8)
and (9) that,
(10)
The use of the Tresca failure criterion and equations (8) - (10)
allows the calculation of elastic limit stresses and strains from
their tensile counterpart.
By using Laplace transforms, it is easy to show for linear
viscoelasticity that equations (8) - (10) are valid where the
stresses and/or strains are a function of time providing Poisson's
ratio is a time independent constant. Conversion of equations (4) -
(6) from tension to shear requires additional information on relax-
ation times or viscosity coefficients. If Tobolsky's assumption of
equivalent relaxation times in shear and tension is invoked, it
readily follows that (15),
llt
( 11)
Using relations (8) - (11) shear predictive equations from

tensile counterparts can easily be obtained for constant strain-
rate, creep to delayed yield (failure or rupture) and elastic limit
or yield stress variation with strain-rate phenomena. These may be
written respectively as,
'(s = Gys
'(s(Ys) = as + llsYs (12)
(13)
aS ' + aS n log C:.~ (14 )
y
s
= y , + Y " log
s s
(ys )
• ,
Y·s
where shear variables Ls' Ys' and shear material properties G, as'
~s' etc., are related to tensile quantities as previously
desoribed.
EXPERIMENTAL PROCEDURES
Constant strain rate, creep and relaxation tests were per-

formed on the speoimen geometries shown in Fig. 4. The tests were
performed utilizing an Instron testing maohine in a controlled
environment of approximately 25 0 C (70 oF) and 50% R.H. Electrical
resistance strain gages were used to measure strains at appropriate
locations on the rail shear specimen. Elongations in the lap shear
tests were measured using a crack opening displacement (COD) gage.
Small holes were drilled into the aluminum adherends on either side
of the glue line, steel pins were inserted in the holes and the COD
gage was mounted between the steel pins. In this manner a very
precise measure of elongation was obtained over a small gage length.
Glued tabs were employed under the steel rails for the rail
shear tests. Average shear stresses for both rail shear and lap
geometries were calculated by dividing the applied load by the area
in shear. Additional test procedures and data reduotion details are
given in reference (14).
Constant strain-rate shear stress-strain results of the bulk

1113 and 1113-2 adhesives from symmetric rail shear tests are shown
in Figs. 5 and 6. As may be observed, the bulk adhesive shows
elastic and viscoelastic response regimes muoh as the bulk tensile
data of Fig. 3a. Properties generally increased with increasing
strain-rate. The yield (maximum) stresses are not especially
meaningful as failure always occurred under a rail at a stress
concentration site and not at the location of the strain gage.
Prediction of shear results from tensile data using the equa-
tions (12) we~ successful only in the elastic region. That is,
viscoelastic shear stress-strain behavior was not predictable from
6.-----------------------------------~
40
5
,
( 3)
30
4
ELASTIC STRAIN RATES:
III
(1) y = 3.71 %/SEC '"
"-
(2) Y= 4.3 x 10- 1
.;,£ ::E:
%/SEC
(3) Y= 1.43 x 10- 3
In In
In
I.LI 3 %/SEC In
I.LI
c:::
I- 20~
In GAV 121,000 psi In
c:::
c:(
(838 MPa) c:::
c:(
I.LI I.LI
:J: :J:
In In
,
2
SPECIMEN FAILURE (MAJOR

DROP IN STRESS) 10
OL-____ ~ ____ ~~ ____ ~ ____ ~______ ~
o 2 4 6 8 10
SHEAR STRAIN (%)
Fig. 5. Symmetric Rail Shear Stress-Strain/Strain-
Rate Response of Metlbond 1113.
viscoelastic tensile stress-strain behavior. It was possible to

fit the data of Figs. 5 and 6 using a modified Bingham model with
new constants and new relaxation times. Apparently Tobolsky' s
assumption of equivalent relaxation times in shear and tension is
not appropriate for the material tested herein. The linear elastic
limit stresses for different strain-rates from Fig. 6 are shown in
Fig. 7. Here good agreement was obtained between tensile predic-
tions and measured values.
The maximum or ultimate (yield as defined herein) stresses are
not shown compared with tensile predictions of equation (14)
50
6
40
~(3)
~ (4)
VI
.:.t.
'"
"'-
::IE:
V> ELASTIC STRAIN RATES: 30 V>

V>
~4 w
0::
I-
(1 ) Y = 11. 50 0::
l-
V> (2) Y = 2.25 2 V>
0:: (3) Y= 5.81 x 10=3 %/SEC 0::
(4) Y= 1.94 x 10
c( c(
w
:x: %/SEC w
:x:
V>
GAV = 136,000 psi 20 V>
(939 MPa)
2
~ SPECIMEN FAILURE
(MAJOR DROP IN STRESS) 10
8
SHEAR STRAIN
Fig. 6. Symmetric Rail Shear Stress-Strain/Strain-

Rate Response of Metlbond 1113-·2.
because, as will be shown subsequently, large differences were

found.
Symmetric rail shear creep response of 1113-2 is shown in Fig.

8. It is apparent that large creep strains were found in bulk shear
which were not predictable from tensile creep values. Maximum
rupture strains for tension were about 4-6%. Further, rupture did
not occur in most cases and the creep tests were discontinued after
the times indicated. Examination of the creep data clearly indi-
cates the bulk shear response to by non-linear. These observations
would tend to indicate that Metlbond 1113-2 exhibits different
o
-
EXPERIMENTAL POINTS 20 -;;;
Q.
:E
--- TENSILE PREDICTION
VI
VI
Lo.I
a::
In
.....
~
(t-)
as = as ' + as" log y'
lOu
.....
::E
-'
as' = 2160 psi (14.9 MPa .....
~
as" 319 psi (2.2 MPa) VI
c:c
-'
l' 1 %/S C Lo.I
10- 2 10- 1 100

ELASTIC STRAIN RATE (%/SEC)
Fig. 7. Variation of Elastic Limit Stress of
Metlbond 1113-2 and Comparison with Tensile
Predictions.
viscoelastic deformation mechanisms in tension and shear.
Creep to 10% strain levels from Fig. 8 are shown plotted in

Fig. 9. Also shown is the maximum stress reached for the highest
strain rate of Fig. 7. The use of an arbitrary failure strain is
justifiable and clearly Crochet's delayed yield model of equation
(13), if valid at all, should fit such a definition as well as the
actual yield or rupture process. Equation (13) does represent the
data reasonably well but the constants A , B , C are not predict-
able from tensile values. s s s
The stress-strain response of Metlbond 1113 for five different

symmetric single lap specimens tested at about the same strain-rate
is shown in Fig. 10. The predicted initial slope from tensile data
is shown and agrees well with the measured results. However, quite
obviously, considerable data scatter was obtained in the visco-
elastic region giving rise to substantial variation in failure
stresses and strains. Again, generally large strains were found in
the viscoelastic region not predicted by tensile results.
Constant elongation (relaxation) tests were conducted on two

lap shear specimens with the results being presented in Fig. 11.
That is, the grips of the Instron machine were held fixed after
reaching a predetermined load and strain level. As is evident,
substantial adhesive strain increase was observed while the machine
load was relaxing. Even though the aluminum adherends were SUbstan-
tially stiffer than the epoxy adhesive, considerable strain trans-
fer occurred between the two materials giving rise to creep of the
15
(l)
~ 10
c:(
c::
l-
V)
c::
c:(
LLI
:I:
V)
SPECH1EN REACHED 20% STRAIN (3)
AFTER 38 HRS.
5 (4)
SPECIMEN REACHED 17% STRAIN AFTER
48 HRS.
(1) TO = 5.1 ksi (35.0 MPa)
(2) TO = 4.7 ksi (32.2 MPa)
(3) '0 = 4.4 ksi (30.3 MPa)
(4) TO = 4.2 ksi (28.8 MPa)
o ____
~ ~~ ____ ~~ __ ~~ ______ __
~ ~
o
Fig. 8. Symmetric Rail Shear Creep Response of

Metlbond 1113-2.
adhesive under fixed grip conditions. Obviously a servo-controlled

test procedure not used herein would be the only proper procedure to
determine adhesive relaxation for the symmetric lap shear specimen.
Creep tests were performed on four 1113 symmetric lap speci-

mens with the results presented in Fig. 12. Again large shear
strains not predictable from tensile results were encountered. At
the highest stress level tested rupture occurred during loading.
For two other specimens, large strain discontinuities were observed
without specimen rupture and were likely the result of an adhesive
fracture and fracture arrest process. Again, the measured creep
strains tend to indicate a non-linear viscoelastic deformation
50
7
6
40
[J EXPERIMENTAL POINTS
-- EQUATION (13)
V)
V)
LLI
ex:
l-
V)
ex:
[J 30 ~
::t:
V)
3
20
o 200 400 600 800

TIME (min)
Fig. 9. Creep to 10% Strain Data for Metlbond 1113-2
Symmetrical Rail Shear Specimens.
process.
Creep to failure stresses are shown plotted in Fig. 13 where
failure was defined as either rupture, a sudden strain discontinu-
ity or reaching a 10% strain level for creep at the lowest stress
level. Obviously, the latter definition has no influence on the
character of the delayed failure curve. Again, equation (13) due to
Crochet is a reasonable representation of the delayed failure
process for the adhesive in the bonded state, but the constants A ,
Bs and Cs are not predictable from tensile results. s
Table 1 gives a comparison between the predicted and measured

average shear properties of Metlbond 1113. As noted earlier,
average elastic shear properties can be predicted reasonably well
from tensile properties. However, maximum stresses and strains are
not predictable from tensile values and in general larger values are
4
_ _-----;2
, 30
ELASTIC STRAIN RATES:

:;::-3 0";;;'
~
III (1) y = 7.9 x 10- 3 %/SEC 0...
~
V'>
(2) Y = 8.B x 10- 3 %/SEC V'>
.....
V'>
c:<:
(3) Y = 8.1 x 10- 3 %/SEC V'>
W
c:<:
f-
V'> (4) y = 8.8 x 10- 3 %/SEC f-
15 c:<:
Vl
(5) y = 9.6 x 10- 3 %/SEC
,
c:<:
ct:
~2
:z::
w
:z::
V'> EXPERIMENTAL RESULTS Vl
SPECIMEN FAILURE
PREDICTED INITIAL SLOPE FRON 10
BULK TENSILE DATA FOR
Y = 8.6 x 10- 3 %/SEC
o 61\
Fig. 10. Symmetric Single Lap Adhesive Stress-Strain

Response of Metlbond 1113. (Specimens
Tested at Nearly the Same Strain-Rate.)
found in shear than would be expected on the basis of tensile

measurements.
Previously reported measurements of viscoelastic adhesive bulk

tensile properties and the analytical procedures to characterize
such properties were presented. Relations were developed by which
adhesive viscoelastic shear properties could be predicted from the
earlier tensile measurements.
ADHESIVE
STRAIN 30
24
///
3.4 ,/
23
/'
/ 26
/
/
3.2
22 /
/_--
'" ~
/
0..
=
:£ 22~
V>
V>
UJ
21 V>
V>
/ --/ «
0::
~3.0 l-
0::
..... .....
//
V>
V> V> 0::
0:: 0::
«
« «UJ
//
UJ
w 20 18~
1/
::I: ::I:
V> V>
2.8
19
2
18 6
L-______- L______~~------~~-L~~ 10
o 300 600 900 1200
TIME (MIN)
Fig. 11 . Adhesive Relaxation Behavior of Two Metlbond

1113 Symmetric Single Lap Specimens (Constant
Deformation was Maintained at the End of
Adherends . )
Constant strain-rate and creep measurements were presented for

the same epoxy adhesive used in earlier tensile studies. Results
were presented for both bulk and bonded adhesive shear properties
using symmetric rail shear and symmetric lap shear test specimens
respectively.
It was demonstrated that both bulk and bonded shear stress-

strain response include a region of near linear elasticity followed
by a region of viscoelasticity terminating with a yielding region as
was the case with the tensile response. Generally, elastic shear
stresses and strains were shown to be predictable from tensile
results while viscoelastic stresses and strains were not. Large
viscoelastic strains were obtained in shear which were not antici-
pated on the basis of tensile predictions. Relaxation times in
t SUDDEN JUMP IN STRAIN LEVEL OF: t

(2) 18% (3) 44%
40
(1) RUPTURE
STRESS LEVELS:
(1) to = 4,000 psi
(2) to = 3,290 psi
(22. 7 t~Pa)
(3) to = 3,160 psi
30 (21.8 MPa)
(4) t = 3,000 psi
o (20.7
MPa)
-
z
~
t-
en
10
(4)
SPECIMEN REACHED ~ 10%

STRAIN AFTER 41 HRS.
o ~__________~____________~________~
o 10 20
TmE (MIN)
Fig. 12. Symmetric Single Lap Creep Response of

Metlbond 1113.
tension and shear were not the same in contradiction to Tobolsky's

assumption. For these reasons, it is felt that the materials
investigated tended to exhibit different viscoelastic deformation
mechanisms in shear from tension. Such different deformation
processes may also be evidence that use of a simple Tresca failure
theory for viscoelastic processes is inappropriate and that a more
correct procedure would be to use the general viscoelastic-plastic
theory of Nagdi and Murch (6). Quite obviously, the materials
studied displayed both geometric and stress-strain nonlinearities
in shear. Recently, we hav~ made detailed measurements of tensile
4.5r----------------------------------,
3'0
4.
o EXPERIMENTAL POINTS
25~
~ 3.5 • RUPTURE "'
"-
::E
EQUATION (13)
V') V')
V') V')
UJ .....
0:: 0::
l- l-
V') V')
0:: 0::
c:(
~ 3.0 UJ
~P
::I:
V') 20~
TO 10%
STRAIN
2.5
15
2.0 ~ _ _ _ _ _ _ _ L _ _ _ _ _ _ _ _ ~_ _ _ _ _ _ ~ _ _ _ _ _ _ ~
o 800 1600 2400 3200

TIME (MIN)
Fig. 13. Creep to Failure Data for Symmetric Single

Lap Specimens Bonded with Metlbond 1113.
Table 1. Comparison of Average Shear Properties

for Metlbond 1113.
Elastic Yield (Max) Failure

Modulus Stress Strains
Tensile 119 ksi 3780 psi 5.5%

Prediction (818 MPa) (26.0 MPa)
Symmetric 122 ksi 5000 psi 6.8%
Rail Shear (839 MPa) (34.5 MPa)
Symmetric 136 ksi 3840 psi 36.4%

Lap Shear (937 MPa)
stress-strain non-linearities of Metlbond 1113-2 and have success-

fully modeled the results using a non-linear integral equation
approach due to Schapery (16). These results will be reported
subsequently.
ACKNOWLEDGEMENTS
This effort was supported by NASA I S Langley Research Center

under NASA grant NGR 41-004-090. Dr. John G. Davis, Jr., was the
NASA technical monitor and his assistance is gratefully acknow-
ledged. Thanks are also due to Mr. Ted Massa and Mr. W. I. Griffith
for their assistance in preparing portions of the manuscript and to
Ms. Peggy Epperly for her expert typing.
REFERENCES
1. H.F. Brinson, M.P. Renieri, and C.T. Herakovich, "Rate and

Time Dependent Failure of Structural Adhesives", Fracture
Mechanics of Composites, ASTM STP 593, 1915, pp. 111-199.
2. M.P. Renieri, C.T. Herakovich, and H.F. Brinson, "Rate and
Time Dependent Behavior of Structural Adhesives", VPI
Report, VPI-E-16-1, April 1916.
3. H.F. Brinson, "The Viscoelastic-Plastic Characterization of a
Ductile Polymer", Deformation and Fracture of High Poly-
mers, H. Kausch, et al., eds., Plenum Press, NY, 1914.
4. H. F. Brinson and A. DasGupta, "The Strain-rate Behavior of
Ductile Polymers", Experimental Mechanics, .J2., No. 12,
pp. 458-463, December 1915.
5. W. Prager, Introduction to Mechanics of Continua, Ginn and
Co., NY, 136, 1961.
6. P.M. Nagdi and S.A. Murch, "On the Mechanical Behavior of
Viscoelastic-Plastic Solids", !L.. of !EE.h Mech., 30, p.
321, 1963.
1. M.J. Crochet, "Symmetric Deformations of Viscoelastic-Plastic
Cylinders", !L.. of !EE.h Mech., 33, p. 326, 1966.
8. M. Reiner, Advanced Rheology, H.K. Lewis and Co. Ltd., London,
208, 1911.
9. K.W. Chase and W. Goldsmith, "Mechanical and Optical Charac-
terization of an Anelastic Polymer at Large Strain Rates
and Large Strains", Experimental Mechanics, li( 1), pp.
10-18, January 1914.
10. L.E. Malvern, "The Propagation of Longitudinal Waves of Plas-
tic Deformation in a Bar of Material Exhibiting a Strain-
Rate Effect", Journal of Applied Mechanics, ~, pp. 203-
208, 1951.
11. P. Perzyna, "The Constitutive Equations for Rate Sensitive
Plastic Materials", Quarterly of Applied Mathematics, 20,
p. 321, 1963.
12. Y.T. Yeow and H.F. Brinson, "A Comparison of Simpl.e Shear
Characterization Methods for Composite Laminates", Com-
posites, Jan. 1978.
13. W.J. Renton and J.R. Vinson, "Shear Property Measurements of
Adhesives in Composite Material Bonded Joints", ASTM-STP
580, 1974.
14. E. Sancaktar, "The Viscoelastic Shear Behavior of Adhesives",
Ph.D. Thesis, VPI & SU, June 1979.
15. A.V. Tobolsky, Properties and Structure of Polymers, J. Wiley
& Sons, Inc., NY, 1960.
16. J. S. Cartner, Jr., "The Nonlinear Viscoelastic Behavior of
Adhesives and Chopped Fiber Composites", M.S. Thesis, VPI
& SU August 1978.
Surface Energetics and Structural
Reliability of Adhesive Bonded Metal
Structures
D.H. Kaelble
Science Center, Rockwell International
ABSTRACT
All metals ,except gold, are chemically reactive with oxygen
and moisture at ambient temperature and tend to form oxide or
hydrated oxide layers. In addition to oxidative chemical reaction,
these high energy surfaces are subject to physical adsorption by
water vapor and volatile organic contaminants. Earlier studies
have shown for structural aluminum alloy AL-2024-T3 that the kine-
tics of surface aging, subsequent to acid etching, can be followed
by wettability measurements and expressed on surface energy maps of
dispersion and polar components of solid-vapor surface tension.
Weibull statistics have recently been applied by Wegman and co-
workers to investigate the effects of both surface exposure time
(SET) and bond exposure time (BET) to high moisture upon the shear
bond strengths of metal joints. In this study the surface chemistry
and related surface energy of both metal adherends and adhesives are
analyzed prior to bonding. The theory of interfacial adhesion and
thermodynamics of fracture are combined to predict the effects of
both SET and BET on long term bond strength and reliability.
Structural adhesive joints of both aluminum and titanium structural
alloys bonded with epoxy-phenolic adhesive are shown to degrade
according to surface energetics predictions. Aging is shown to
lower average strengths without broadening the Weibull strength
distributions. Results are discussed in terms of structural reli-
ability of adhesive bonded metal structures.
301
302 D. H. KAELBLE
INTRODUCTION
All metals, except gold, are chemically reactive with oxygen

and moisture at room temperature and tend to form oxide or hydrated
oxide surface layers (1) • In addition to oxidative chemical
reaction, these high energy surface films are subject to further
physical adsorption by water vapor and volatile organic contam-
inants. In structural metal to metal joints, the reliability of the
bond may be directly related to the chemical stability of the
interfacial bond. Weibull statistical analysis has recently been
applied to investigate the effects of both surface exposure time
(SET) and bond exposure time (BET) on the distributions of single
layer shear bonds of aluminum alloys (2). Smith and Kaelble (3)
have recently conducted a detahled study of combined SET and BET
aging in normal (50% R.H., 23 C), and high moisture (95% R.H.,
54 0 C). The experimental methods of the study by Smith and Kaelble
are outlined in Table 1 with full details contained in the published
report.
Table 1. Metal Joint Reliability Studies

1. Metal Adherends: Unclad 2024-T3 aluminum alloy
surface treated by standard FPL sulfuric chromate
etch and T8-6A1-4V titanium alloy treated by stan-
dard phosphate fluoride cleaning process. Coupon
size 0.063" thick, 1" wide, and 4" long.
2. Adhesive: HT 424 epoxy-phenolic film adhesive (from
American Cyanamid) with glass fiber carrier and
standard weight 0.0135 + .005 lb/sq. ft. Unfilled
HT 424 primer with parts-A and B used with adhesive.
3. Bonding Process: Treated metal coupons spray primed
with 0.001" thickness HT 424 primer solution using
clean dry argon carrier gas. Primer layers dried 30
min ambient 23 0 C and 60 min at 66 0 c. An adhesive
film is placed in the 1.000" x 0.500" overlap
between two metal adherends. Six such joints are
aligned in a bonding jig with the glass carrier
acting to provide constant glue line thickness
0.008". Cure cycle with 60 min temperature rise to
171 0 C and 60 min cure cycle at 171 0 followed by
cooling to room temperature.
4. Tensile Lap Shear Testing: 1.5" x 1.0" x 0.063"
aluminum alignment shims bonded to eliminate offset.
Tests at 23 0 C using 0.01"/min Instron crosshead rate
and 4.5" jaw separation.
ADHESIVE BONDED METAL STRUCTURES 303
8~------TI------~I------~I--------'-I------'
Al 2024-T3 []
Aged Surface
~O
~, "- -
-
......
E
u
6-
""
""
SVr
'"
1"ac~
. -. a
"-
c: 9 (> 0. !lif}g
~
HT424-' "
c:
4- ....... -10~ -
o
A1 2024- T3
(Fresh Surface)
2- -
O~----~I~----~I------_~I------~I----~
o 2 4 6 8 10
Polar - ~ (dyn/cm)t
Fig . 1. Dispersion (a) and polar (8) components of 2
the solid-vapor surface tension YSV = a~ = 8S
for HT 424 primer (phase 1) and Al 2024-T3
adherend (phase 3).
DISCUSSION
In this study the surface chemistry and related wettability of
both adherends and adhesive were analyzed prior to bonding. The
results of this surface energy analysis can be plotted on a surface
energy map where the ordinant a and abscissa 8 as shown in Fig. 1
respectively, refer to dispersion (non-polar) and polar components
of surface energy and interfacial bonding mechanisms. The theory of
interfacial adhesion experimentally verified in this analysis de-
fines the thermodynamic work of adhesion W by the following
relation (4,5): a
(1)
where a 1, 8 1 define the dispersion and polar surface properties of

adhesive and a 3 , 8 3 , those of the metal adherend.
304 D.H.KAELBLE
As shown in Fig. 1 and further documented in detailed kinetic

studies by Kaelble and Dynes (4), the <13 and B3 properties of
aluminum alloy change dramatically with surface aging time after
FPL etch. The <1 1 , B1 surface properties of HT 424 adhesive are
shown to lie below the curve for Al 2024-T3 at all stages of surface
aging which predicts proper bonding between adhesive and adherend
in air.
The interfacial work of adhesion W as defined by Eq. (1) will
decrease with surface aging time (SET) ~s shown by the upper curve
of Fig. 2. As shown in the lower curves of Fig. 2, the lap shear
bond strength varies with SET in a fashion that correlates closely
with the predictions from work of adhesion calculations. A simple
but now widely demonstrated correlation between surface energetics
and fracture mechanics is available in the following relations for
critical stress °
for Griffith type crack initiation under normal
stress loadings (5,6):
0c = (;~y (R2 - R02y ~0 (2)
where E and C are a characteristic modulus and crack length which

are assumed constant and the surface energy parameters Rand RO are
defined by the following relations (5,6):
2
RO = o. 25 (<11 - 3)
(1 + ( B1 - B3 )2 (3)
2
R2
= (<12 - H) + (B2 _ K)2 (4 )
H = 0.5 (<11 + 3)
(1 (5)
K = 0.5 ( B1 + B3 )· (6)
In Eq. (4) two new surface energy parameters <12 and B2 define the
environment (phase 2) at the crack tip. The model for critical
stress defined by Eq. (2) can be presented on surface energy
coordinates as shown in Fig. 3.
As the adhesi ve joint changes from dry air immersion with

<12 = B2 = 0 to equilibrium response with water immersion with
<12 = 4.67(dyn/cm)1 and B 2 = 7.14(dyn/cm)1, the predicted decrease
in critical stress
c
°
of the HT 424 to Al 2024-T3 interface is:
0c (H 20)
a (air) = 0.644.
c
ADHESIVE BONDED METALSTRUCTURES 305
100~------~-------r-------'--------r-------'--------'
..,
W11 (HT424 Adhesive) a 76.2 erg/cm 2
70~------~------~------~--------~------~------~
o 120
Reduced SET (hr) at 54°C and 95% R.H.
3100
,'~D ________________ ~ ____ _

o
2700 0~------..I-------:'--------'--------"--------l...10-0------....J120
SET (hr) at 54°C and 95% R.H.

Fig. 2. Dependence of interfacial work of adhesion
W13 (upper curve) and lap shear bond strength
0b (lower curve) at varied SET.
Extensive joint strength testing of this system was completed to

determine the response surface of lap shear strength vs. both SET
and BET under high moisture (95% R.H., 54 0 C). These results are
summarized on the response surface of Fig. 4 where each point
represents the average of six tests. Comparing joint strengths for
fully aged (20 hr. SET, 1000 hr. BET) and unaged (0 hr. SET, 0 hr.
w
o
0-
8i~--------~--------r-------~------~ 8~i--~~--~---r----~--~--~--~
-
.....E
u
-
......E
u
"?- .......
>, c: 5
"C
>,
"C
...... <.!J
" 4 >- 4
c:
0 ......
'"L I=N
up- 3
QI ~
Q.
u
'" Q;- 0
i5 2
8 00 2 3 4 5 6 7
Polar - 1\ (dyn/cm)! R (dyn/cm)+
o
Fig. 3. Modified Griffith analysis of the effect of H20 immersion in reducing ::I:
critical failure stress 01 for interfacial fa1lure between HT 424 and A
l>
etched Al 2024-T3. m
r
OJ
r
m
»
o
:I:
",<S- m
."> (J)
...,# <
m
OJ
0
_..# Z
! 0
;= m
0
'"i3 s:
0-
"" m
'"co< -I
7 <
... »
r
:z:
'" '"... ~
",'" '" ",' '" / " :5 :0
... C
____ ____ _","
____ J _____ J_ '" ,," 0
......... '" '" '" '" '" '" " /",
-I
'" " " C
,,'" '" """ " :0
m
'" '" '"
" '" ' "'" " " "''' (J)
"'" '" '" ",'" '" '" "
.... / '" " " '"
_--r----~----~-----~-----L / / / ,,,
/' "/, /
",'" "," ",'" "," "

-_"- _ _ _ _ _ ~ ______ ~ _____ .1~- ~ ___L
......"" '" '" '" " ","
~<....
..,¢
o 200 400
" 600 800 1000
BET ( hr)
Fig . 4. SET and BET response surface for lap shear bond strength
for Al 2024-T3-HT 424 . w
o
......
308 D.H.KAELBLE
BET) provides (see Table 2) the following experimental ratio:
0b (aged, wet) 2275 psi 6 (8)

= 3300 psi = O. 9
0b (unaged, dry)
which is in close agreement with the prediction of Eq. (7).
An essentially parallel detailed study of surface aging showed
a shift in a 3 and B3 for phosphate-fluoride cleaned titanium alloy
similar to that detailed in Fig. 1. Application of the modified
Griffith analysis as shown in Fig. 5 provides the following pre-
dicted moisture degradation of bond strength at the HT 424 to Ti-
6Al-4V interface:
°c(H 20)
a (dry air) = 0.84. (9)
c
Table 2. Weibull Strength Distributions

Composite Pollmer Strength DistributtO~S
~
R = exp -(ob/oo)m °
Metal-Adhesive Joint Single
AL2024T3.HT424 Epoxy Lap
SET(hr) BET(hr) Shear a (Kg/cm2 ) m(o)
0
0 0 N=12 232 14.5

0 165.449 12 184 15.4
0 808,1023 12 165 10.0
21 0 12 208 15.0
20 669,983 12 160 18. 1
Ti-6Al-4V-HT424 Epoxy
SET(hr) BET(hr) ° (Kg/cm2 )
0
mea)
0 0 N=12 270 7.65

0 (670,1016) 12 182 6.22
21 0 12 272 7.65
21 (591,997 ) 12 202 6.35
SET = surface exposure time, BET = bond exposure time at 54 0 C

and 95% relative humidity.
»0
J:
m
CJ)
<:m
OJ
8. 8. 0
Z
0
m
0
s:
m
-I
E 6 »
u r
-- ........
c::
>..
CJ)
-I
~ :D
c
--0 N n
-I
~
~HT424 ........ <!)
~
c
4 :D
m
CJ)
"
...........
N
........
-!+ I
/'<:
1- u
1=
2~ ~ u 2
'0
D_ 0' •
O 2 4 6 8 o 2 4 6
II (dyn/cm)i R (dyn/cm)i
Fig. 5. Modified Griffith analysis of the effect of H 0 immersion in

reducing critical failure stress 01 for interfacial failure
between HT 424 and phosphate-fluor1de treated Ti-6Al-4V.
Co)
~
310 D.H.KAELBLE
i:)\)
.,,(S
~r-------------------------------~/
.. _
-D--_.9
// ,,1-'- ___;- ____ --,.-____
~
~
--7--/
/'
- - _ .... _ _ _ _ , /
/ 0,; ,,/ ,/"./
/,; ~~
~ ~
V)
'"
. . . . . --7:.-_ ,,'/
,; ,;
----/-;r;"
7~ - - - -
I'l'l' ;//
-L ,; / /
~~t:l ~
'"
- - - "
/
./
,; -----,.-----..J----- ,
---,/
/ ,,/
/
/
./
"
7 '- /
"
./
/
"\
r§l
~
.....
\0 ,; ,.. - - _"
--,;"-----j ;
r-----r------.4 ,/ / ,; " ,; / / ,'5
0 '- --- ,;,;
---0-- ' " , ;
",
"0
",;,; ,; / ,;
o ----_/_--- ~---- " / /
/.~-----/7-----~/f-----~/L------/~----~/~
o 200 400 600 800 1000
8ET (hr)
Fig. 6. SET vs. BET response surface for lap

shear bond strength for Ti-6Al-4v - HT424.
An equivalent calculation for moisture degradation of the HT 424

cohesive bond produces the following prediction (3):
°c(H 20)
o c (d ry . ) = 0.63. (10)
a~r
Joint strength testing of the HT .424 to titanium alloy was

carried out under conditions of separate and combined SET and BET in
high moisture (95% R.H., 54 C). The SET vs. BET response surface of
shear bond strength 0 is shown in Fig. 6 where each point is an
average of six strengt~ tests. As shown in Fig. 6, the shear bond
strength reaches an equilibrium value under extended moisture
aging. Comparing joint strengths for fully aged (22 hr. SET, 1000
hr. BET) and unaged (10 hr. SET, 0 hr. BET) for Ti-6Al-4V to HT 424
bonds (see Table 2) provides the following experimental ratio:
ADHESIVE BONDED METAL STRUCTURES 311
~ (aged, wet)
2873 psi = 0.75
a b ( unaged, dry) = 3840 psi
( 11)
which lies intermediate between the cohesive failure prediction of

Eq. (10) and the interface prediction of Eq . (9) • Microscopic
visual inspection of the fracture surfaces for the HT 424 to
titanium lap shear points shows predominant (above 50%) cohesive
failure for lap shear bonds described in Fig. 6.
Structural Reliability
The Weibull plots of Fig. 7 show shear bond strength distribu-

tions for unaged and fully aged aluminum (upper view) and titanium
(lower view) joints. The titanium bonds show lower Weibull m values
in both unaged and aged states as evidenced in Fig. 7 and the data
summary of Table 2. A design requirement of high reliability shear
llt! (psi)
1636 1998 2441 zqs1 3641 4447
2
.066
0= 0 .37
oS,
0:
oS
-2 .87
-4
i I
0.98
7.4 7.6 7.8 ao a2 a4
In%
llt! (psi)
1636 1998 2441 2981 3641 4447
2
Adherend • Ti-6A I-4V 0.066
0 .37
0:
oS,
0:
c
-2 <:> .87
~'I,\
~'
"!P> "
-4 "7.4 7.6 7.8 ao a2 a4
0.98
In llt!
Fig . 7. Comparison of Weibull Shear strength
distributions for aluminum (upper view)
and titanium (lower view) adherends.
312 D. H. KAELBLE
strength where R = 0.98 or In(-lnR) = -4 is shown by Fig. 7 to

predict higher performance for the aluminum alloy joints in both
unaged and aged-wet states. Conversely, if mean strength with R =
0.37 and In(-lnR) = 0 is applied as a design criteria, the curves
show titanium alloy joints to display higher unaged and aged
strengths.
REFERENCES
1. J .C. Bolger and A.S. Michaels, in Interface Conversion for
Polymer Coatings, ed. P. Weiss and G.D. Cheever, American
Elsevier Publishing Co., NY, Chapter 1 (1969).
2. D.W. Levi, W.C. Tanner, R.C. Ross, R.F. Wegman and M.J.
Bodnar, !!..:..!E£h Poly. Sci. 20, 1475 (1976).
3. T. Smith and D.H. Kaelble, "Mechanisms of Adhesion Failure
Between Polymers and Metal Substrates", Technical Report
AFML-TR-74-73, Air Force Materials Laboratory, WPAFB,
June 1974.
4. D.H. Kaelble and P.J. Dynes, J. ColI. and Interface Sci. 52,
562 (1975). - -- - --
5. D.H. Kaelble, !!..:.. !E£h Poly. Sci. ~, 1869 (1974).
6. D.H. Kaelble, Polymer Eng. and Sci. 11 (7), 474 (1977).
Composition and Ageing of a Structural,
Epoxy Based Film Adhesive
C.E.M. Morris, A.G. Moritz and R.G. Davidson

Materials Research Laboratories
Department of Defence
Ascot Vale, Victoria, 3032, Australia
ABSTRACT
A 180 0 c curing, structural film adhesive has been analyzed by

IR and NMR and found to consist of three different epoxy compounds
with dicyandiamide as the curing agent. The ageing of this material
at 20 0 C has been examined by monitoring changes in the molecular
weight distribution, epoxy equivalent value and solubility. It was
found that the changes which occur are ascribable to a combination
of cross-linking and hydrolysis.
INTRODUCTION
The use of structural, film-form adhesives in aircraft con-
struction is now widely adopted. While the one-part nature of these
materials has many advantages in terms of ease of use, the accom-
panying limited shelf life presents significant problems. This
paper examines the composition and ageing of one such adhesive.
The material, a 180 0 C curing, supported film, is extensively

used as a honeycomb-skin adhesive in a number of aircraft presently
in service. The recommended shelf life of the uncured material is 6
to 12 months at _18 0 C. As this material must be imported into
Australia and only small amounts are needed at anyone time for
aircraft repairs, the short shelf life has caused considerable
problems. A programme has therefore been undertaken to study the
composition, ageing and cure reactions of this material. The
initial results of this programme are presented here.
313
314 c. E. M. MORRIS ET AL.
EXPERIMENTAL
Samples from two batches of the adhesive were used in this

work.
The GPC studies were performed on a Clanor instrument fitted

with two detectors (differential refractive index and evaporative
analyser); the performance of this equipment, which is an Austral-
ian instrument, is described elsewhere (1). The column set con-
sisted of four four-foot Styragel columns of nominal porosity
3 x 10 3 , 1 x 10 3 , 500 and 200i, the solvent (chloroform) flow rate
was 1 ml/min and the oven temperature 30 oC.
NMR spectra were determined on a Varian HA60-IL spectrometer
using deutero chloroform solutions and tetramethylsilane as the
internal lock. Infrared spectra were measured on a Jasco IR-C
spectrometer as liquid films.
Epoxy equivalent values were determined by a titration method

based on those of Jay (2) and Dijkstra and Dahmen (3).
The flow test employed consisted of curing a 3.75 cm square

piece of adhesive at 180 0 C and 0.03 MPa (45 psi), cutting the
resultant specimen back to the original dimensions and expressing
the flow· as the weight of material cut off as a percentage of the
original sample weight (4).
RESULTS AND DISCUSSON

Composition
The overall composition of the adhesive was found to be
Epoxy resins 72-74%

Dicyandiamide 7.5%
Asbestos 10%
Glass support 8.5%
There is also a small amount of iron and traces of chromium and

nickel compounds. The overall composition varies somewhat from
batch to batch and also, as the adhesive is not uniformly spread
over the glass support, from place to place within a roll.
The glass support is a plain weave of glass rovings of about

200 ends per roving. The weave is such that the roving is more
flattened in one direction than in the other and that the inter-
stitial spaces are rectangular rather than square. (Figure 1)
STRUCTURAL EPOXY·BASED FILM ADHESIVE 315
Fig. 1. Woven glass support (X6 mag.)
GPC analysis of chloroform solutions was used to examine the

molecular weight distribution and also as a means of fractionating
the material. The resin dissolves almost instantly in chloroform
but these solutions do not conta i n the curing agent as dicyandiamide
is only slightly soluble in chloroform. The molecular weight
distribution of two batches of the adhesive is given in Figure 2:
the molecular weight covers the range approximately from 300 to
25,000.
The material was collected as four fractions comprlslng

(1) the lowest molecular weight peak, (2) the next peak, (3) mate-
rial eluted between about 108 and 134 mls and (4) the highest
molecular weight portion (81 to 108 mls approximately) . NMR and 1R
analysis of these fractions showed (5) them to be the following:
(1) triglycidyl p-amino phenol (I)
Triglycidyl p-Qmino phenol

316 C. E. M. MORRIS ET AL.
(2) mainly, an oligomer of I
(3) chiefly, a cresol-formaldehyde-novolac oligomer (II).

The NMR spectrum shows little fine structure suggesting a
mixture of isomers and/or moderately high molecular
weight material.
l\
CH-CH-CH
/\
CH-CH-CH
122 I 2 2
o, 9 (II)
r r
Cresol novolQc epoxy
(4) a high molecular weight oligomer of the diglycidyl ether

of bis-phenol A (III).
oG E BA
As might be expected, there were minor amounts of III (of

smaller values of n) in fractions (1), (2) and (3) and significant
amounts of II in fractions (2) and (4).
It is estimated that the triglycidyl p-amino phenol and its

oligomers constitute about 45% of the epoxy resins and the DGEBA
component 10-15%. The overall epoxy equivalent weight for the resin
fraction of the adhesive is 174.
Ageing
The room temperature ageing of 2 batches was examined so that

the effect of short times out of a freezer, or loss of dry ice
during shipping, could be assessed. The samples were kept, with
STRUCTURAL EPOXY-BASED FILM ADHESIVE 317
Batch 1 ------
Batch 2 -- •,
,I
--------- ----- ------ ---- ---'

----------
90 110 130 150

ELUTION VOLUME, ml
Fig. 2. Gel permeation chromatogram of two batches

of adhesive.
their backing material in place, inside plastic bags in a laboratory

at 20 oC. At the commencement of this experiment batch 1 was some 14
months old and batch 2 about 2 months old. At intervals of
approximately 14 days, a piece was cut off and a GPC trace run. The
changes which occurred over 4 months are illustrated in Figs. 3 and
4.
Only minor changes were observable in the chromatograms for

the first 30-40 days (less than the initial differences between the
batches). At longer times a tailing at the low molecular weight end
of the chromatogram occurred which eventually became an additional
peak. IR analysis indicated that this material was hydrolysed
triglycidyl p-amino phenol having a very minor epoxy component and a
greatly increased hydroxyl content. Essentially the same material
was obtained from the aqueous phase after refluxing a fresh sample
of adhesive in water for several hours . TLC showed it to be a
mixture of five or six similar compounds.
A very small peak, of negative refractive index relative to

chloroform, was discernible in batch 1 at 77 days and in batch 2 at
71 days at an elution volume of about 74 ml, that is, at signifi-
cantly higher molecular weight than the rest of the material. No
further changes occurred in this region with increasing time. This
compound has not been identified.
Initial
42 Days
84 Days
111 Days
~2' 0
z
o
~
w
cr:
0::
110 130 150

ELUTION VOLUME , ml
Fig. 3. Room temperature ageing of batch 1.
Init ial
49 Days
78 Days
~
105 Days
""
'\
,r,
----::::=:-:::.' '\
-,-,-,-,-.-.-,-,-..::~::,--:..:-:.:::..-:::-.:.:::~=.:=~.;;:::::~~/ \.
"-
' .....,,~
~~~~--------~--------~
,,~
O ------ -----~1~
30~-------------- 1~
ELUTION VOLUME , ml
Fig. 4. Room temperature ageing of batch 2.

STRUCTURAL EPOXY-BASED FILM ADHESIVE 319
Solubility
~
u 80
.!O
w
...I
~
...I
60
0
CJ)
~
40
w 100
::>
...I Value
~
> 80
~
0-
w
...I
z 60
CI:
~
0::
0
40
'0
~
20
80
Flow
•
60
~
0
....
...I
40
~
20
0
I
0 40 80 120 160 200
TIME . days
Fig. 5. Change in resin solubility, epoxy value and

flow during room temperature ageing .
• Batch 1. ~ Batch 2.
After about 40 day_~ chloroform solutions of batch 2 (concen-

tration about 4.5 x 10 glml) became more difficult to filter
through a 0.4511 filter and by about 55 days the solutions were
extremely difficult to filter. Batch 1 displayed this feature after
a somewhat longer interval. After some 60 days the samples were
significantly slower to dissolve. However, the solubility in
chloroform changed very little until, after about 120 days, the
material only swelled. (Figure 5).
The overall epoxy equivalent value (initially 173 for batch 1

and 174 for batch 2) took about 100 days to fall to half its
original value (Figure 5). Flow tests on the two batches of
adhesive showed essentially no change up to 100 days for batch 1 and
about 90 days for batch 2 (Figure 5).
Correlations must be made with a programme of mechanical

testing before final conclusions can be drawn on the time which this
adhesive can tolerate at ambient temperature and still be used
satisfactorily. However, it appears that periods up to several
weeks in a temperate climate are accompanied by changes in the
adhesive which are minor when considered in relation to batch to
batch variations. Batch 2, although actually younger than batch 1,
displayed some ageing signs rather sooner than batch 1, but its
chromatogram showed less obvious signs of change. It appears that
the reactions which occur are a combination of cross-linking and
hydrolysis.
ACKNOWLEDGEMENTS
The authors wish to thank P.J. Pearce for determination of the

epoxy equivalent values and I. Grabovac for operation of the GPC.
REFERENCES
1. I. Grabovac and C.E.M. Morris, "Performance of a Gel Permea-

tion Chromatograph and its Usefulness in Examining Mix-
tures of Organic Materials", Materials Research Labora-
tories, Melbourne, MRL-R-725 (1979).
2. R.R. Jay, Anal. ~ 36, 667 (1964).
3. R. Dijkstra and E.A.M.F. Dahmen, Anal. Chim. Acta. 11, 38
(1964).
4. C.V. Cagle, Handbook of Adhesive Bonding, McGraw-Hill, p. 32-
36 (1973).
5. C.F. Poranski, W.G. Moniz, D.L. Birkle, J.T. Kopfle and S.A.
Sojka, "Carbon-13 and Proton NMR Spectra for Character-
izing Thermosetting Polymer Systems. I. Epoxy Resins and
Curing Agents", Naval Research Laboratory, Washington,
NRL Report 8092 (1977).
Viscoelastic Characterization of
Structural Adhesive via Force
Oscillation Experiments
D.L. Hunston, W.D. Bascom, E.E. Wells (1), J.D. Fahey
and J.L. Bitner
Chemistry Division, Naval Research Laboratory
Washington, D.C. 20375
ABSTRACT
A test device based on the Weissenberg Rheogoniometer was

developed to measure the dynamic viscoelastic properties of solids.
A computerized data acquisition and analysis system with graphics
capability is utilized in this apparatus. Equations containing
terms for the moment of inertia and damping of the torsion head were
developed for forced oscillation. The apparatus was employed to
characterize an epoxy with known fracture and fatigue properties.
Despite the epoxies complex morphology, time-temperature superposi-
tion could be used to obtain master curves. The shift factors (and
glassy modulus) for thea transition depended on the thermal history
of the specimen while the shift factors for the B relaxation gave a
linear Arrhenius plot with an activation energy of 15 kcal/mole.
INTRODUCTION
Polymeric systems offer strong, light-weight materials that

are finding important commercial applications and as a result they
are being subjected to a tremendous range of loadin!b conditions.
For example, the use temperature can range from -50 C to +250 0 C
while the material t s response to stress is often called upon to
exhibit good behavior during_~he ve..'Cf short time periods corres-
ponding to impact loads (10 - 10 sec.), during the somewhat
longer time periods corresponding to fatigue or rising load situa-
tion (10- 3 - 100 sec.), and during the very long intervals charac-
321
322 D. L. HUNSTON ET AL.
teristic of static loading (100 - 10 6 sec.). In order to understand

such complex behavior as fracture, fatigue and wear, it is important
to characterize the basic mechanical properties of the materials
involved. Since polymer properties exhibit a strong dependence on
time and temperature, it is desirable to characterize samples over a
wide range of temperatures and times (or, in the case of cyclic
loading, the equivalent variable-frequency).
At present most characterization experiments involve measuring

the linear viscoelastic properties of the materials over a very wide
range of temperatures at a single frequency (1 Hz) or over a small
range of frequencies (3 to 110 Hz). The low and high temperature
data are then used to qualitatvely interpret the behavior at short
and long times, respectively. Such qualitative predictions, how-
ever, are not entirely satisfactory and thus it is desirable to
directly measure the properties over a much wider range of time
scales. Such experiments can identify rheologically simple mate-
rials where time-temperature superposition can be used (2). More-
over, commercial polymers are often blends which contain fillers,
additives, toughening agents, etc. As a result they can have time-
temperature regions in which their rheological behavior is complex
and thus measurements over a wide range of conditions is essential
(3) •
Although some authors (4,5) have conducted measurements of

this type--generally using creep or stress relaxation experiments--
a much broader application of such testing techniques is needed.
Creep (6) and stress relaxation (7) experiments are particularly
useful for investigating behavior at the longer time scales, while
forced oscillation measurements (8,9) provide a convenient techni-
que for shorter time scale experiments. It is desirable therefore
to combine both methods to obtain data over a wide range of times.
The purpose of this paper is to report the development of a forced
oscillation instrument, based on the Weissenberg Rheogoniometer,
and the use of this apparatus to investigate the viscoelastic
properties of a thermosetting epoxy.
THEORY
There are several potential difficulties that arise in the

analysis of forced oscillation tests. These complications arise
from two particular features of the experimental design. First, the
stress and strain are measured at different ends of the sample and
this complicates the analysis of the results. A recent paper by
Schrag (10) has discussed this topic in some detail. Second, some
motion at the stress measuring device is always required to obtain
data. This can make the sample deformations more complex and
introduce effects due to the inertia and natural damping of the
FORCE OSCILLATION EXPERIMENTS 323
the stress measurement device. To minimize this problem, efforts

are often made to stiffen the stress detection system (11). For
some experiments, such as transient measurements, such modifica-
tions are extremely important. Some comments on the value of a
stiff system for the forced oscillation testing of solids will be
given later in this paper. Although the remarks presented here are
directed primarily at the use of a Weissenberg Rheogoniometer for
testing of solids, the ideas are equally valid for other forced
oscillation experiments.
The mathematics of the forced torsional oscillation test were

analyzed many years ago by Markovitz et al (12). They consider a
cylindrical sample of radius, R, and length, t. The bottom of the
sample (axis of the cylinder mounted vertically) is driven in
torsional oscillation with an angular displacement, eo (t), and this
produces a simple wave motion in the sample. The top of the sample
is connected to a torsion head conSisting of a torsion bar, sample
grip, air bearing, etc. The torsional elastic constant of this
device is k, the moment of inertia is I, and the natural damping
force association with torsional motion is assumed to be propor-
tional to angular velocity with a proportionality constant, U.
Although Markovitz et al. do not consider the natural damping of the
torsion head, this term can easily be included simply by replacing
(Lw 2 _k) with (Lw 2-k-iwU) where w is the angular frequency and
i = /=T. The motion at the top of the sample, et{t), is then found
by balancing the stress produced by the wave motion in the sample
with the elastic and damping properties of the torsion head.
By modifying the results of Markovitz et al. the motion at the

ends of the sample can be written as an infinite series involving
powers of the reciprocal of the complex shear modulus, G*.
{~W)2j {_p)j
(2j + 1)!
where
(2)
and P is the sample density. The relative motion at the two ends of
the sample et/eo = mei~ where m is the ratio of the amplitude at the
top to that at the bottom and ~ is the phase angle by which the top
leads the bottom.
The solution of this problem gives rise to an infinite series
because the stress and strain are measured at different ends of the
sample. As a result, the only way to increase the rate of
convergence for the series is to decrease the sample length. There
is a practical limit to this approach, however, since grip effects

become more of a factor for small t. Thus, a series solution cannot
be avoided. This has consequences not only for measurements in the
linear viscoelastic range but also for experiments at larger ampli-
tudes where a non-linear response is achieved. When analyzing the
results of non-linear experiments, it is not sufficient to solve a
constitutive equation for oscillation at a point. Instead, the
boundary value problem for the actual geometry must be addressed.
Such an analysis is needed to clarify both the effect of sample
length and the nature of the deformation in the sample.
To solve eq. (1) for G, * i t is necessary to terminate the

series. For most samples the convergence is very rapid and only the
first term is needed; however, certain lower modulus materials may
require two terms. Although the equations become rather complicated
in this case, the solution is straightforward. Since several
quantities in eq. (1) are complex numbers, it is difficult to
specify the exact conditions where the second term becomes impor-
tant. As an example, however, consider (t 2w 2pI2G*) which is
approximately the ratio of the second to the first term. Whln t = g
cm, W = 120 rad./sec. (19 Hz), p = 1 g/cm3 , and I G I = 10
dynes/cm2 , the magnitude of this quantity is 0.18 and thus, the
second term makes a major contribution. Obviously, this term must
be examined when testing low modulus elastomers.
For many materials, however, only the first term is needed and
thus
G' =
m[{(~) +U' } (m-cos ~) + tB" sin ~] (3)
m2 - 2m cos 4> + 1
G" =
m[ -K~} + tB' sin l
m2 - 2m cos 4>
4> + tB·
+
(m-cos"J (4)
where
B = B' + i B" (5)

These equations provide a method to calculate viscoelastic proper-
ties from suitably treated experimental data.
When the stress detection system is very stiff (small m) and

the elastic constant dominates the other terms, the equations can be
greatly simplified to give the well known relationships
2 $I, k
G' = m - - 4 cos ~ (6)
'If R
2 $I, k
Gil = m - - 4 sin ~ (7)
'If R
Although this simplification is convenient, it is not required to

measure linear viscoelastic properties. In the linear range the
compliance of the stress detection system does not affect the type
of deformation present in the sample or the accuracy with which it
can be measured (unless the compliance is very high). Although the
amplitude of the oscillation will be less in a more compliant
system, this is taken into account by the equations. Consequently,
linear viscoelastic properties can easily be measured. For non-
linear studies, a stiff detection system may be an advantage;
however, this can only be determined if a solution with an appro-
priate non-linear equation is obtained.
Another interesting aspect of eqs. (3) and (4) is the inclusion
of terms involving U. Many previous papers have addressed the
importance of inertia terms but damping is usually neglected. In
most cases this is understandable; however, it may not be justified
for very low loss samples. An examination of eqs. (3) and (4 )
indicates that the second term in the numerator of eq. (3) is
unimportant since both U and sin ~ are small. In eq. (4), however,
both terms in the numerator involve small quantities. Conse-
quently, when the sample becomes more elastic (smatler sin ~) or the
inertia contribution becomes larger (smaller Iw -k), the damping
term increases in importance. In the present series of tests, when
the higher frequencies were used, the damping of the torsion head
was often found to be more important than the inertia althougH the
contributions of both quantities were usually small. By using the
appropriate equations, however, all effects can be taken into
account.
EXPERIMENTAL SECTION
Apparatus
The modified Rheogoniometer is illustrated in Fig. 1. The
sample (S) is held between two specially deSigned grips. The top
and bottom sections of each grip are made from aluminum while the
center portion is constructed from glass-reinforced nylon to mini-
mize heat flow along the shaft. The lower grip can be driven in
constant-speed rotation, in torsional OSCillation, or in any com-
bination of the two; separate synchronous motors and 60-speed gear
boxes (A) are used for the oscillatory and rotational drives. The
AB
I
ITC
I
I
I
I _ _ _ _ _1
B
C
Fig. 1. Diagram of Modified Rheogoniometer:

(A) motors and speed adjustment gear boxes,
(B) main gear box, (C) minicomputer,
(D) flexible diaphragm, (E) shaft angle
encoder, (M) micrometer, (R) transducers
for rotational motion, (S) sample, (T) dis-
placement transducers, (AB) air bearing,
(CS) cantilever spring, (TB) changeable
torsion bar, and (TC) temperature control
chamber.
rotational speed and the frequency of oscillation can therefore be

independently varied over a range of almost six orders of magnitude
while the amplitude of oscillation is also variable (B). The stress
levels associated with the motion are measured by monitoring the
twist of the top grip which is attached to a changeable torsion bar
(TB). An air bearing (AB) is used to maintain alignment without
introducing significant frictional forces. A compressive or ten-
sile stress can also be applied to the sample along the axis of
rotation by a cantilever spring (CS) attached to the bottom grip
through a rotating connection and adjusted with a micrometer (M) and
displacement transducers (T). Since the torsional drive mechanism
is connected to the bottom grip through a flexible diaphragm (D),
the rotational and axial forces can be applied independently in any
combination. The sample temperature is controlled by a specially
designed chamber (TC) that maintains a constant sample temperature
over the range from -180 0 c to +275 0 C.
The most crucial aspect of the instrument, however, is the data
acquisition system, Fig. 2. To fully utilize the versatility of
DATA
EOC
DATA
EOC C
RESET
Fig. 2. Computer interface: (E) is the shaft

angle encoder (Baldwin Model 3100AZS),
(I) is indicator lights, (FF) is the
flip-flop, (R) is the transducer,
(AD) is the sample-and-hold unit and
A-to-D converter, and (C) is the com-
puter (Texas Instrument 960A).
this apparatus, it is necessary to obtain very accurate data over

the entire range of time scales. Furthermore, real-time data
manipulation and evaluation are essential for determining what
experiments must be performed. This problem has been solved by
interfacing the instrument to a minicomputer (C) using an expanded
version of the design developed by Krieger et al (13). The data
acquisition rate and timing accuracy are assured by using high-
speed A-to-D converters equipped with sample and hold units attach-
ed to the rotational displacement transducers (R). A shaft-angle
encoder (E) is attached to the oscillatory drive shaft after the
gear box. Since all of the speed adjustment occurs before the
encoder, each revolution of the encoder corresponds to one cycle of
oscillation regardless of the frequency (13).
The encoder provides one index pulse and 360 sector pulses per
revolution. The value of 360 was selected because it is the maximum
number of points that can be conveniently handled at the highest
frequencies. The data acquisition cycle proceeds as follows, Fig.
2. Wheri the computer senses the index pulse it clears the interface
by resetting a flip-flop (FF). The first sector pulse after the
interface is cleared triggers the flip-flop which initiates the
acquisition of a data point from each transducer and notifies the
computer that acquisition is in progress. When the computer senses
the end-of-conversion signal from each A-to-D converter, the data is
transferred. Only then is the interface cleared by resetting the
flip-flop. This procedure makes it impossible for the computer to
acquire the same data twice. To be certain that no points are

missed, after 360 data points are collected from each transducer,
the computer checks to see that the index pulse is obtained before
the next sector pulse appears. This verifies that exactly one cycle
has been measured (any number of resolutions other than one can
readily be detected in the data analysis). This interface design
also makes it possible to use the computer in a time-sharing mode so
that other tasks can be performed simultaneously. A set of lights
(I) indicates when the interface is in operation.
In testing a sample, it is essential that the measurements

begin only after all transient effects have decayed away. This is a
major concern at the lower frequencies; while at higher frequen-
cies, it is important to minimize the number of cycles to minimize
heat build-up. In addition, it is desirable, although not neces-
sary, to obtain data for at least 2 cycles so that the base-line
drift can be corrected accurately. For each transducer the computer
stores the sum of data at each sector pulse and the sum of all data
pOints in a given cycle; thus, for m cycles, 720 + 2m data points
are collected. The total-per-cycle values are used to compensate
for any linear base-line drift and offset from zero. A Fourier
series analysis is then performed on the data points from each
transducer to obtain the amplitude and phase shift (relative to a
sine wave starting at the index pulse) for the fundamental mode and
any harmonics that are present (generally 4 overtones are checked).
The series analysis approach was selected after testing correlation
methods and more general Fourier analysis techniques for two rea-
sons: it can provide a filtering effect that increases the signal-
to-noise ratio and it yields absolute information for the motion at
both transducers and this 1s useful 1n the analysis of the results.
When the equipment is operating properly, a pure sinusoidal
wave (less than 1~ harmonic content) is obtained for the motion at
the bottom grip. When the deformation is in the range of linear
viscoelastic behavior, the motion at the top grip is also a pure
sinusoidal wave. For the present tests the minimum amplitude of
oscillation was used and a linear viscoelastic response was always
obtained. Under these circumstances the amplitude ratio and dif-
ference in phase shifts (Le. the phase angle, ell) can be used in
eqs. (3) and (4) to calculate the dynamic shear modulus. The
computer also determines the contributions made by the damping and
inertia terms in the equations because, although they are included,
they contain quantities that may be difficult to calibrate. The
validity of the series termination necessary to obtain eqs. (3) and
(4) is also checked. The computer system used in this work has
graphic display capability so that the actual data points can be
displayed and examined visually, if desired. The computer not only
plots the data points but also draws the best-fit sinusoidal curves
through the points so this can also be checked (Fig. 3).
Fig. 3. Data points and best-fit curves for motion

at top grip (upper curve) and bottom grip
(lower curve). Some scatter was introduced
in top grip data so that best-fit curve could
be seen.
Before testing a sample it is essential that the instrument be

calibrated. The torsional elasticity, moment of inertia, and
damping constant of the torsion head were measured as described in
the literature (14). The transducer displacements and relative
phase shifts must also be corrected for effects introduced by the
mechanics and electronics of the system (15). The transducers are
first calibrated at each conditioner setting with a micrometer. The
grips were then clamped together and the amplitudes and relative
phase shifts were determined at each combination of frequency and
setting on the filter unit contained in the electronics package. No
effort was made to independently measure the absolute amplitude of
oscillation since it is the relative motion that is most important
for the calculations. Consequently, the readings for the trans-
ducer attached to the bottom grip measured with the highest filter
setting were used as the standard. The corrections determined in
these tests can be quite significant. For example, Fig. 4 shows the
difference in relative phase shifts between the two transducer
signals as a function of frequency for the three possible filter
settings. To obtain phase angle measurements that are reliable to
better than 0.1 degree, it is essential that this correction not be
in the rapidly rising region shown in the figure since the scatter
becomes significant in this area. Thus the filter setting must be
increased at much lower frequencies than those suggested by the
manufacturer. In the apparatus described here, the computer recom-
mends the best filter setting for each set of conditions and
+
6 +
*+
+
+ Filter
- ++ + 20Hz
+ It 50Hz
~ + o 150Hz
r.n :5
..
+
en
o 2
~
Q..
- 10 0(1/60)
-I
Fig. 4. Phase-angle correction as a function of

frequency, f, in Hz for the 3 filter settings.
automatically includes all of the appropriate calibration factors

in the calculations.
The experiments were performed using samples having both rec-
tangular and circular cross-sections. To analyze the results for
samples with rectangulnr cross-sections, eqs. (3) and (4) were
altered by replacing nR 12 with
nR 4 [_1 64 00 tanh (k nb/.2)]

2 • ba 3 3 - 5" ~ (8)
n n=o (2n+1)5
where
kn = (2n+1)n/a (9)
b is the width of the sample, and a is its thickness. Unfortu-

nately, it is not possible to obtain a complete solution for this
geometry without making some assumptions about the stress or strain
field in the axial direction (16). The validity of these assump-
tions in an experimental system is difficult to judge. From a
practical point of view, however, the experiments performed here as
well as similar tests by other authors (8) show that, within
experimental error, both rectangular and circular cross-section
samples generally give equivalent results in the linear visco-
elastic range. A number of samples with different lengths were also
tested to detect effects resulting from gripping of the sample at
the ends. As with previous studies (8), it was found that these
effects could be ignored when the sample length exceeded 6 cm.
Materials
To illustrate the use of this apparatus, it was employed to
determine the dynamic mechanical properties of an epoxy that has
been extensively tested for fatigue (17) and fracture (18) behav-
ior. This epoxy is a diglycidyl ether of bisphenol A (Dow Chemical
Co. DER-332) cured for 16 hours at 120 0 C with 5 phr piperidine.
After the heat treatment the samples were allowed to cool in the
oven in an effort to keep the thermal histories relatively constant.
After each specimen was machined to the proper dimensions, it was
annealed at 75 0 C for 6 hours. It was felt that even a short
annealing period could affect the sample since significant"thermal
heating was induced during the machining of the specimens.
All of the temperature tests were performed starting at the

lowest temperature and moving up in steps whose size was determined
by the rate the measured properties were changing. The importance
of thermal history, which will be illustrated later, means that
increasing temperature experiments are the only meaningful way to
obtain data. At each step the frequency dependence was studied
after the sample was given time to reach constant temperature. The
exact number and range of frequencies tested was determined by the
temperature stability and the rate at which the sample properties
changed. In each experiment the final frequency tested was always
the same as the first and the repeatability of this measurement was
used to establish that constant temperature had been reached.
When all of these experiments were completed, an untested

sample was placed in an oven and annealed for an additional 18 hours
at 75 0 C. This sample was then characterized in an effort to
evaluate, at least in a qualitative way, effects related to thermal
history.

The shear mechanical properties of the epoxy sample annealed
for 24 hours were measured as a function of frequency in the
temperature range from -100 0 C to +110 0 C. Some typical results are
illustrated in Fig. 5 which shows the thermomechanical data
obtained at 6 Hz. There is an a or glass-to-rubber transition at
about +95 0 C and a B peak at about _60 0 C. The a transition was
examined in detail by analyzing data obtained over a 41 decade range
of frequencies at temperature between +40 0 c and +110 0 C. The results
were then considered for the applicability of time-temperature
superposition. The availability of data over a wide range of
frequencies makes it possible not only to accurately determine the
shift factors, aT' but also to critically examine whether the points
really fall along a single master curve or not. Unless the
frequency range is large enough, a true test of superposition cannot
be obtained. In fact, when the range is too small, the data will
always appear to superimpose.
G'
10
·to -50 0
Temperature
Fig. 5. Shear storage modulus (dynes/cm2 ) and

loss tangent at 6 Hz as a function of
temperature.
Recent papers by Koutsky (19) have indicated that the morphol-

ogy of epoxies is complex in that a nodular structure is involved.
Despite this complex morphology, the epoxy samples tested here were
found to exhibit rheologically simple behavior in that all of the
data could be reduced to single master curves. Figure 6 shows the
master curves for the sample (annealed for 24 hours) at a reduced
temperature, T , of +90 0 C. A single set of shift factors super-
impose all thre~ curves. (The modulus was normalized with the usual
temperature-density ratio.)
The shift factors corresponding to the epoxy samples after
annealing for 6 and 24 hours are shown in Fig. 7. Unfortunately, it
is not possible to obtain data at high enough temperatures to
compare the results with the predictions of equations such as the
Williams, Landel, and Ferry (WLF) relationships (T > Tg + 10 where
Tg via DSC is about 80 0 C) because extended;xposure to temperatures
near and above the cure tempera ture (120 C) can change the cure
state of the system. There is evidence that WLF behavior might be
expected. A series of tests are now being performed on the same
epoxy system but with a small amount of an elastomeric toughening
1010 10
G'
_ ~l t-
t2 Q,.
~\....
10 9
~~"
(!)
0 00
c:
~
_JIQ,.
t-
(!) ~~x~-*""X
108 10- 1
107~~ __~__~__L-~__- L_ _~~~~_ _~_ _~~_ _~_ _~10-2

-4 -2 o 2 4 6 8
log (OrW)
a:: 4
o
.....
U
4
IJ.. 2
.....
IJ..
I
(/) 0
(!)
o
....J
-2
-4
40 60 80 100 120
TEMPERATURE (0 C)
Fig. 1. Shift factor as a function of temperature

for the ~ relaxation of the epoxy after
annealing for 6 hours (X) and 24 hours (0).
agent added. For this system the cure temperature is still 1200 C
but the Tg is lower and so more data above Tg can be obtained. For
the toughened epoxy the preliminary results show that the shift
factors follow the WLF equation. Regardless of the exact nature of
the behavior above Tg , however, the shift factors might be expected
to increase dramatically as the temperature is lowered and approach
infinity at a temperature somewh~t below Tg • Experimentally,
however, a much different behavior is observed. The shift factors
were not found to increase indefinitely but rather to gradually
level-off. Moreover, the shape of the curve was found to be a
function of thermal history of the sample; for the specimen annealed
for 24 hours the shift factors were much closer to the expected
behavior.
The explanation for this result lies in the non-equilibrium

nature of the glassy state (20). As a sample is cooled near T , the
time required to reach an equilibrium state increases and th~s the
cooling rate must be decreased. Eventually, it becomes impossible
to cool the sample slowly enough to maintain equilibrium (21). With

systems that are closer to an equilibrium state, for example
annealed or aged samples, the temperature dependence of the shift
factor is closer to the expected behavior. The difference in shift
factors shown in Fig. 7 also has important consequences for the
dynamic moduli. For .both thermal histories tested here, the master
curves were surprisingly similar. Consequently, the moduli for
both histories were the same at temperature well above Tg and at
very low temperatures. In the intermediate range the sample
annealed for 24 hours had a higher storage modulus and lower loss
modulus than the sample annealed for 6 hours. This suggests that
thermal history (and aging) can be viewed in terms of changes in the
temperature dependence of the shift factors. This idea is quite
similar to the results of Struik (21). Obviously, a more complete
study of these effects is needed. It is clear, however, that no
simple theory can adequately predict the viscoelastic properties in
this temperature range and thus experimental characterization of
the sample is essential if its behavior is to be understood and used
to predict more complex properties such as fracture.
In the 6 Hz data (Fig. 5) a 8 relaxation was observed at about
-60 0 C. An effort was made to obtain a master curve in this region
by superimposing data obtained over a 3! decade range of frequencies
and the results are shown in Fig. 8. Although the scatter was
somewhat greater in this range because the temperature control was
not as good, no systematic variations were observed and thus
superposition seems to be a valid procedure. Since the data were
obtained over a wide frequency range, it is possible to examine the
shift factors for the 8 relaxation in an Arrhenius plot. Figure 9
gives such a graph and the data are found to fall along a straight
line whose slope gives an activation energy for the process of 15
kcal/mole. This is typical of the values generally obtained for 8
relaxations (8).
Figures 7 and 9 illustrate another important point. People
often look at isochronal data such as that in Fig. 5 and assert that
the motions associated with the a and 8 processes are "frozen out"
at given temperatures, in this case +80 0 C and -100 0 C. For the
glass-to-rubber transition the term "frozen out" has some basis in
fact since presumably the motion would stop if equilibrium could be
maintained as the temperature 1S lowered. Figure 7 suggests,
however, that it is unlikely that the sample can be cooled slowly
enough to totally stop the motion. The 8 peak, on the other hand,
follows Arrhenius behavior over the entire range of temperatures
studied here and thus there is no evidence that the motion will
cease until absolute zero. Consequently, the expression "frozen
out" is not really appropriate. The results shown in Fig. 5
indicate only that the motion associated with the 8 peak is slow
compared to 6 Hz. In reality, the 8 peak can be shifted to any
'.---_____________ __________--;-__,°.1°
2.'
..
G'
'0
0
.. I.' ..
......
"\
-t;""" ..
Q
0.05 !
.
0. 2.0
'"
'0
1.0
..
~:+..1. . . . .tL"~~.
_~2~----~1---~0---~----~2~--~3~--~4~--~50
Fig. 8. Curves for dynamic shear modulus (dynes/cm2 )

and loss tangent asoa function of reduced
frequency (T = -50 C).
o
1.8
1.2
0 .6
.....I-
l1:li 0
3. 8 4 .0 4 .2 4.8 &.0
CI' x 10'
0
-0.6
-1 .2
-1.8
-2 .4
Fig. 9. Shift factor as a function of inverse

temperature for the B relaxation of the
epoxy.
temperature that is desired simply by selecting the appropriate

measurement frequency.
Another important question is the mechanism responsible for
the 8 relaxation. Recent studies have combined the results of the
present work with those obtained from the Carbon-13 NMR experiments
of Dr. A.N. Garroway to address this problem (22,23). This compar-
ison suggests that a rotation in the polymer backbone may be related
to the 8 relaxation (23). Complete details of this study will be
published elsewhere.
CONCLUSIONS
A forced oscillation test device, based on a Weissenberg

Rheogoniometer, was developed for measuring the viscoelastic
properties of solid samples. A computerized data acquisition and
analysis system that uses graphic display capability to monitor the
acquisition and analysis process was constructed. Appropriate
equations containing terms for the moment of inertia and natural
damping of the torsion head were developed and computerized.
A bisphenol-A epoxy (cured with piperidine) whose fracture and

fatigue properties have been extensively studied was characterized
with this equipment. Despite the complex morphology of epoxies,
time-temperature superposition could be used to obtain master
curves for the dynamic shear modulus. In the range of the
transition the shift factors (and the glassy modulus) were
dependent on the thermal history of the sample. The shift factors
for the 8 relaxation gave a linear Arrhenius plot corresponding to
an activation energy of 15 kcal/mole.
ACKNOWLEDGEMENTS
The authors wish to thank Dr. I.M. Krieger, Dr. H. Markovitz,
and Dr. J. Hassell for their helpful comments during the develop-
ment of this equipment. The assistance of Mr. C.M. Henderson and
Mr. T .R. O'Neal in the construction of the apparatus is also
gratefully acknowledged.
REFERENCES
1. Present address: Logicon Inc., 8300 Merrifield Ave, Fairfax,

Virginia.
2. J.D. Ferry, Viscoelastic Properties of Polymers, Chapter 11,
Wiley, New York (1970).
3. D. Kaplan and N.W. Tschoegl, "Time-Temperature Superposition

in Two-Phase Polyblend", Polym. ~ Sci. .1!, 43 (1974)
and references therein.
4. R.E. Cohen and Y. Sawada, "Extension of the Contour Plotting
Method for Representing Linear Viscoelastic Loss Proper-
ties of SBS Triblock Copolymers", Trans. Soc. Rheol. 21,
157 (1976). - -
5. C. B. Bucknall, J. C. Reid, W. W. Stevens, "Fracture Mechanics
Studies of High-Impact Polystyrene", Coatings and Plas-
tics Prep. 37 (1), 477 (1977).
6. E. Riande, H. Markovitz, D.J. Plazek, and N. Paghupathi,
"Viscoelastic Behavior of Polystyrene-Tricresyl Phos-
phate Solutions", !!.:. Polym. Sci. ~ 50, 405 (1975).
7. S.S.Sternstein and T .C. Ho, "Biaxial Stress Relaxation in
Glassy Polymers; Polymethylmethacrylate", !!.:..!EEl:. Phys.
~, 4370 (1972).
8. W.M. Davis and C.W. Macosko, "A Forced Torsional Oscillator
for Dynamic Mechanical Measurements", Polym. Sci. ~
17, 32 (1977).
9. R.E. Cohen and N.W. Tschoegl, "Comparison of the Dynamic
Mechanical Properties of Two Styrene-Butadiene-Styrene
Triblock Copolymers with 1,2- and 1,4-Polybutadiene Cen-
ter Blocks", Trans. Soc. Rheol. 20, 153 (1976).
10. J .L. Schrag, "Deviation of Velocity Gradient Profiles from the
"Gap Loading" and "Surface Loading" Limits in Dynamic
Simple Shear Experiments", Trans. Soc. Rheol. 21, 399
(1977). - -
11. R. H. Lee, L. G. Jones, K. Pandalai and R. S. Brodkey, "Modifica-
tion of an R-16 Weissenberg Rheogoniometer", Trans. Soc.
Rheol. 14, 555 (1970). --
12. H. Markovitz-,-P.M. Yavorsky, R.C. Harper, Jr., L.J. Zapas and
T .W. DeWitt, "Instrument for Measuring Dynamic Viscosi-
ties and Rigidities", Rev. Sci. Instrum. 23, 430 (1952).
13. LM. Krieger and T.F. Niu, "A Rheometer for Oscillatory
Studies of Nonlinear Fluids", Rheol. Acta]£, 567 (1973).
14. Sangamo Controls Limited, "The Weissenberg Rheogoniometer:
Instruction Manual, Model R.18", (Sangamo Controls
Limited, England).
15. K. Bogie and J. Harris, "An Experimental Analysis of the
Weissenberg Rheogoniometer", Rheol. Acta 2, 213 (1966).
16. I.S. Sokolniloff, Mathematical Theory of Elasticity, 2nd ed.,
pp. 128-134, McGraw-Hill, New York (1956).
17. W.D. Bascom and S. Mostovoy, "Adhesive Fatigue Failure of an
Elastomer-Modified Epoxy", Coatings and Plastics Prep.
38,152 (1978).
18. W.D. Bascom, R.L. Cottington and C.O. Timmons, "Fracture
Reliability of Structural Adhesives", !!.:. .!EEl:. Polym.
Sci.: !2Ph Polym. Sym. 32, 165 (1977).
19. J. Koutsky, "Etching of Polymeric Surfaces: A Review", Polym.
Plast. Technol. ~ i, 139 (1977).
20. D.J. Plazek, personal communication, 1978.

21. L.C.E. Struik, "Physical Aging in Amorphous Polymers and Other
Materials", Elsevier, New York (1978).
22. D.L. Hunston, E.E. Wells and A.N. Garroway, "Linear Visco-
elastic Characterization of Thermosetting Epoxy Poly-
mers", presented at the Society of Rheology Meeting,
Houston, October 1978.
23. D.L. Hunston, E.E. Wells and A.N. Garroway, "Time-Dependent
Mechanical Behavior of Epoxy Polymers", presented at the
Adhesion Society Meeting, Savannah, GA, February 1979.
Crack Healing in Semicrystalline
Polymers, Block Copolymers
and Filled Elastomers
Richard P. Wool
University of Illinois at Urbana-Champaign
Urbana, Illinois 61801
ABSTRACT
Crack healing was found to be a generic property of polymers.

Its effect on the mechanical behavior of some semicrystalline
polymers, block copolymers and filled elastomers was investigated
as a function of time, temperature and strain history. It was found
that cracks or microvoids could heal instantaneously or slowly
depending on the microstructural damage and molecular rearrange-
ments incurred during the debonding process. Crack healing rates
increased with temperature and no healing was observed below the
effective T of the active molecular healing components in each
material. gA semiempirical kinetic theory was developed which
quantitatively described mechanical recovery due to crack healing
in terms of time, temperature and strain and provided a basis for
construction of master healing curves by time-temperature super-
position. It is apparent that crack formation and healing provides
another mechanistic interpretation of viscoelastic behavior in
addition to IIlOlecular relaxation processes and whose effect on
mechanical and fracture properties of polymers is largely under-
estimated.
INTRODUCTION
Microvoid formation in the interfaces of contiguous micro-

structural elements is an important aspect of mechanical deforma-
tion in multicomponent polymer systems (1). In semi crystalline
polymers, filled elastomers, block copolymers and general com-
341
342 RICHARD P. WOOL
posites, the interfaces are created by several processes which

include impingement growth during solidification, heterogeneous
formulation and domain segregation. Cracks or voids may form during
deformation, especially in the interfaces and can vary in size from
a few angstroms to macroscopic dimensions. The cracks can be
permanent, instantaneously reversible (instant healing) or time
dependent reversible (slow healing). Figure 1 shows the generic
stress-strain behavior resulting from crack healing in multicom-
ponent polymer systems. The microstructural damage is created by
the first cycle of the triangular strain input. The stress response
for the second cycle depends on the time, temperature and humidity
or other environmental factors of the sample in the rest state
between cycles. As the cracks heal in the rest state, the mechani-
cal properties of the sample are restored partially or fully to the
virgin state depending on the extent of permanent damage.
Crack healing can be studied by a variety of experimental
methods. Direct observation of crack closure modes and healing
processes can be done in a transmission electron microscope using
thin films (2) which have been subjected to mechanical damage. Our
recent experiments indicate that solvent cast block copolymers with
segregated domain structures, such as styrene-butadiene-styrene
(SBS), are useful for this study. Similar work has been done on
polyethylene by Miles et al (3). They observed closure of voids in
films with a stacked lamellar (hard elastic) morphology. Photo
optical methods can also be used to measure the relative amount of
crack development and healing (2). The effect of scattered trans-
mitted light due to cracks can be investigated using a photoelectric
cell in the ocular of an optical microscope. This method was used
to study the contribution of voids during stress relaxation at
constant strain in semicrystalline and block copolymers. When the
sample is unloaded, the "stress whitened" films heal in a tempera-
ture dependent manner as judged by the gradual restoration of the
initial clarity and mechanical strength. The degree of void damage
which develops at constant strain is found to depend on the magni-
tude and duration of strain. In addition to direct observation of
crack healing by optical and electron microscopy, a number of
scattering techniques can be used to monitor crack history. These
include small angle X-ray scattering (SAXS), light scattering and
neutron scattering. SAXS complements photo-optical studies by
monitoring growth or healing of voids before they reach the critical
size necessary to scatter visible light. Other spectroscopic
techniques can be used in conjunction with the above methods to
evaluate mechanisms of healing. Broadline proton NMR has been used
to determine the behavior of the amorphous or fluid like chains
during deformation and healing of hard elastic polypropylene (4).
Infrared spectroscopy can also be used to investigate conformation,
orienta tion and other molecular changes during healing (5,7 ).
Other useful techniques include mass spectroscopy, electron para-
magnetic resonance (EPR) and luminescence methods. The success of
CRACK HEALING 343
iii
-...
.~
c
en Healing
Time Time
Fig. l{a). Shows the 2-cycle strain and thermal
history used to evaluate mechanical
recovery and crack healing in polymers.
III
-
III
...
Q)
en
Strain
Fig. l(b). The generic stress-strain behavior
resulting from crack healing in multi-
component polymers is shown schematically.
the experimental method depends on the nature of the healing

process.
Healing rarely involves reformation of primary bonds asso-

ciated with chain scission processes, although radical recombina-
tion can occur in crystal "cages" (8,9) and to a certain extent in
disordered molecular regions. Facile healing occurs by restoration
of secondary bonding between chains or microstructural components
( 1 ) . Thus van der Waals or London dispersion forces play a very
344 RICHARD P. WOOL
important role in healing. Other kinds of bonding which may be

important in healing processes include hydrogen bonding and chemi-
sorption of chains on filler particles. The rate of healing depends
on the transport of the separated molecular components into each
others vicinity and the subsequent reformation of the original or
equivalent bonding states. Transport can be achieved by diffu-
sional processes through Brownian motion and bulk translation of
microstructural components by release of strain energy. Since
molecular mobility is necessary to achieve time dependent healing,
it is expected that healing rates will be appreciably affected at
temperatures near the glass transition temperature(s) of the
healing component(s) in the material.
In this paper the general phenomena of crack healing will be

investigated in terms of mechanical properties. Mechanical recov-
ery due to crack healing will be examined as a function of time,
temperature and strain for semicrystalline polypropylene, SBS block
copolymers and carbon black filled natural rubber. A kinetic theory
of healing will be developed which can be applied to all these
multicomponent polymer systems. It will become apparent from the
physical behavior of damage and healing processes that there is a
microstructural analogy to molecular behavior which explains visco-
elastic behavior of polymers. The microstructural contribution is
very important for understanding nonlinear viscoelastic behavior in
addition to bulk or "noncrack" constitutive behavior. While crack
healing is advanced as a point of study herein, it is recognized
that other "non crack" mechanisms also contribute to the mechanical
behavior described in Fig. 1. However, the importance of crack
healing will be clearly demonstrated.
MECHANICAL RECOVERY
We can define an adiabatic energy balance for polymer mechani-
cal deformation in terms of three mechanistic quantities as (1)
000
W= U + D + § (1)
0
where W = rate of working (mechanical)
0
U = rate of change of strain energy
0
D = rate of dissipation due to viscous and plastic processes
0
and S = rate of change of damage or surface energy due to frac-
ture mechanisms at the molecular or microstructural level.
For the strain history shown in Fig. 1, the mechanical recovery

R can be defined in several ways using parameters from Eq. (1). For
example, R can be defined in terms of the total work W1 and W2 done
CRACK HEALING 345
in cycle 1 and cycle 2, respectively, as

W2
RW = W1 (2)
The use of Eq. (2) requires few assumptions but it does not provide
information regarding specific contributions of the deformation
mechanisms during recovery. More detailed information can be
obtained by analyzing the stress-strain data in terms of the
energetic parameters, U, D and 3. In the elastic-fracture case, = B
o and we can define recovery as
32
R =S
1
where 3 1 and 32 are the total fracture work to create voids, break
bonds, etc., in the first and second deformation cycles, respec-
tively, and can be obtained from Eq. (1) as
3=W-U (4 )
The total stored elastic energy U is obtained by integrating below

the unload tensile curve. Thus, during healing, it is assumed that
all mechanical recovery is expressed in changes of surface energy as
the cracks heal and bonds reunite while no change occurs in the
strain energy function of the material. The latter requirement
means that the tensile unload curves of both cycles superimpose.
Viscous contributions can be partially evaluated by instantaneously
reloading (zero healing time) and subtracting the resulting hyster-
esis energy from the first cycle work. This procedure is reasonable
with materials that exhibit a predominant elastic-fracture mechani-
cal response but it may create problems for highly viscous systems.
In this study, we have used materials whose viscous contribution is
minimal in order to examine the crack healing phenomena. Thus Eq.
(3) will be used as a working definition of mechanical recovery, R.
If a large viscous contribution is suspected Eq. (2) may be used to
represent the data and in this case the kinetics of healing devel-
oped later will need to be modified accordingly.
MECHANICAL RECOVERY IN POLYPROPYLENE

Hard elastic polypropylene (HPP) fibers with stacked lamellar
morphologies were studied. These materials have been the subject of
many recent studies (10-14). Their tensile properties are unusual
and are shown in Fig. 2. HPP is highly crystalline (density = 0.92
g/cc) with the molecular chain axes preferentially oriented along
the fiber axis. Large deformations are accommodated by void
formation between lamellae and aggregates of the microstructure
(10-14). The internal crack surface area varies approximately
346 RICHARD P. WOOL
12.0
800
If)
I 9.0 t\I
Q 600 E
u
)(
......
.~ ~
01
-
1/1
-
1/1
1/1
1/1 6.0 Q)
Q)
~ 400 ~
( J) ( J)
3.0 200
°OLL-L~4--===~~---L----~------L--JO
ro ~ 00 ~ 100
Stra in, '¥o
Fig. 2. The effect of healing time on the stress-
strain behavior of hard elastic polypropylene
fibers is shown at room temperature.
linearly with strain and is about 100 m2 /g at 50% elongation. Thus,

this material is excellent for studying crack healing. As shown in
Fig. 2, the second cycle stress response depends on the healing time
of the fibers between deformation cycles. At short healing times
considerable stress softening is observed but at longer times the
original mechanical properties of the fibers are restored as the
cracks heal (15).
Recovery in HPP fibers was investigated quantitatively as a

function of time, temperature and strain, using the strain history
shown in Fig. 1. A strain rate of 10/min was achieved on an Instron
tensile tester using a gage length of 2.0 in and a crosshead speed
of 20 in/min. The fibers were tested at ambient room temperature
and humidity for both first and second cycles. The effect of
temperature on healing was studied by placing the fibers in a
constant temperature controlled box, -11 o C to 800 C, between defor-
mation cycles for the desired healing time, t. The healing time was
measured as the time between unloading at zero strain in the first
cycle to the beginning of the second cycle. The maximum strain, €
m
CRACK HEALING 347
was the same for both cycles.
The effect of healing time and strain maximum on recovery of

HPP fibers at room temperature is shown in Fig. 3. About 200
"virgin" samples were tested, each sample being subjected to only
the first and second deformation cycles. Using Eq. (3) the recovery
R is plotted as the ratio of the fracture energies. As shown in
Fig. 3, R is approximately linear with log t and faster recovery
occurs at lower strain maxima. This data is qualitatively similar
to recovery data obtained by Cannon et al (11) who used the fiber
length ratio as a definition of recovery, 1. e. the initial "set"
recovered with time.
The effect of temperature on recovery is shown in Fig. 4 for

strain maxima of 50, 15, 125 and 150%. In each case, the recovery
is seen to increase with temperature as might be expected from the
enhanced molecular mobility at higher temperatures. At _11 0 C (Tg =
_10 0 C) no recovery was observed for any of the strain series. In
each case the initial or instantaneous recovery occurred at room
temperature but further recovery did not occur below Tg. Recovery
could be "restarted" by heating the sample above Tg and stopped by
dropping below Tg. It was also observed that at 150% strain where
considerable damage had been created, almost complete mechanical
recovery was ob~ained at higher temperatures. At temperatures much
higher than 80 C, annealing rather than healing processes become
Log Time (min)
Fig. 3. The mechanical recovery is plotted as a

function of log healing time for HPP samples
subjected to several strain maxima at room
temperature.
348 RICHARD P. WOOL
100
75
50
25 tOm = 75~o
Em:O 50 ~o
t'!
>-
~
CD
::- 0
0
c.I
CD
II::
Em = 125~o tOm:o 150 ~o
~O~O~--~----~---l'---~--~IO~O----~--~~--~~--~'-~105
Log Time (min)
Fig. 4. The effect of healing temperature on mechan-

ical recovery is shown for HPP samples.
important and the constitutive microstructural character of the

material may change, thus affecting the interpretation of the data
in terms of crack healing arguments. It should also be noted that
despite the thermal treatments between deformation cycles, super-
position of the unload curves was obtained with reasonable accuracy
thus allowing the substraction of the strain energy component from
the total work input. At strains less than 15%, little hysteresis
was observed even though cracks formed. As discussed elsewhere (1),
cracks can contribute elastically to the mechanical response with-
out hysteresis effects, i.e. cracks can open and close in a reversi-
ble elastic manner especially at low strains.
In HPP fibers crack healing involves molecular motion in the
amorphous regions between the crystalline lamellar surfaces and
healing of internal crystal defects. Broadline proton NMR studies
(4) of HPP fibers indicate that a large increase in the central
component of the derivative absorption spectrum associated with the
more mobile or fluidlike chains occurs with increasing strain. In
CRACK HEALING 349
the rest state between deformation cycles the intensity of this

fluidlike central component decreases with healing time in a manner
closely paralleling the mechanical recovery. The increased mobil-
ity with strain can be attributed both to the increase in internal
surface area providing additional configurational freedom to the
amorphous chains on the lamellar surface and the generation of new
amorphous material. During healing, the debonded lamellar surfaces
come together and the intervening amorphous chains intermingle and
readsorb on the surfaces assisted by van der Waals interactions.
The increasing restriction of mobility of chain segments during
healing causes the observed decrease in the fluidlike central
component of the NMR broad line spectrum. Densi ty measurements
taken on HPP fibers during healing show an increase in density from
0.86 to 0.93 g/cc. If one assumes little density change of the
lamellar crystals then very large density changes must occur in the
amorphous component. The increased density must decrease the chain
mobility which is consistent with the observed NMR behavior. Stress
infrared studies (6,1) of HPP films indicate both a decrease in
molecular orientation and helical conformation of the PP molecules
with deformation. During healing in the rest state, the molecular
c-axis orientation and conformational regularity is restored. The
increase in orientation is attributed to the deflected lamellae on
the void surfaces returning to their initial stacked morphology
with a subsequent restoration of the chain stem alignment. This
interpretation is supported by X-ray studies of lamellar bending in
hard elastic polymers (16,11). The increase in helical conformers
is consistent with the NMR results in which the amorphous chain
mobility diminishes with void closure and may involve partial
ordering of chains in the amorphous interface and elimination of
surface defects.
Crack healing in semi crystalline polymers can also be inter-

preted in terms of fracture mechanics as a process which restores
the adhesive fracture energy, r, at each point on the interface to
its original value. Pursuing this analysis in a recent paper (12)
it was observed that in the case of HPP, r varied along the crack
surface during healing. Thus, healing was observed to occur
simultaneously at all points on the crack surface but at different
rates. As the crack propagates by lamellar debonding, the molecular
damage or defects on the crack surfaces varies with position in a
manner such that the damage is least in the vicinity of the crack
tip and greatest towards the center of the crack away from the crack
tip. This concept of surface damage is useful for interpreting
modes of healing in molecular and macroscopic models (18). In the
case of PP, the adhesive fracture ene~y for lamellar debonding was
calculated to be about 100 ergs/cm (12,19). As suggested by
Kaelble (20), r values for interfacial debonding in polymers can be
interpreted as the sum of the reversible thermodynamic contribution
plus a diffusional demixing component. In PP, the diffusional
demixing of amorphous chains represents about 10~ of the total
350 RICHARD P. WOOL
adhesive fracture work. From this viewpoint healing involves a

rapidly reversible or instantaneous component due to the thermo-
dynamic surface tension effect followed by a slower healing rate
involving diffusional mixing of chains in the amorphous interface.
The latter effect is consistent with the T dependence of healing
and is also useful for understanding def8rmation rate sensitive
adhesive fracture energies.
BLOCK COPOLYMERS
Similar recovery experiments were conducted on SBS block co-

polymers (Kraton 1101). Film samples 3 x 0.5 x 0.01 in were cast
from a solution of THF/MEK (90/10). Under controlled evaporation
conditions, the styrene component segregates into domains embedded
in the polybutadiene matrix. The resulting material is a thermo-
plastic elastomer in which the hard styrene domains act as cross-
links for the rubbery butadiene matrix (21,22). During mechanical
deformation, voids form in the interfaces between domains and
domain clusters. Healing of these voids can be observed by photo
optical techniques or TEM (2).
Our initial results on healing in SBS films are shown in Fig.

5. The mechanical recovery, R, is plotted as a function of log time
for healing temperatures ranging from -10 to 43 0 C. Again, R was
obtained by separating the fracture energy component from the total
work input using the strain history shown in Fig. 1 and a maximum
strain of 300%. At the higher temperature of 43 0 C, complete
recovery is obtained in about 12 hours. At T = _10 0 C, time
dependent recovery is not observed. The effect of T on healing is
interesting in this material since we have two disc~ete T values
for the domains and the matrix as well as a range of possi~le T 's
in the interfacial regions due to chain mixing. T = 100 0 C ~or
polystyrene and T = -85 0 C for polybutadiene. S~nce complete
recovery is obser!ed below 100 C, the polystyrene chains do not
0
appear to play a major role in the healing process. However,

healing ceases at temperatures above T of polybutadiene and it is
highly probable that the mobile chain~ which are responsible for
healing are those in the domain-matrix interfaces. These inter-
facial regions are mixtures of styrene and butadiene segments and
thus have a resultant effective Tg higher than that of polybuta-
diene.
Recovery in SBS is more rapid than HPP. This is due to the

higher mobility of the butadiene segments compared to polypropylene
and to the crosslinked morphology providing a strong elastic memory
(21). Also, while SBS was subjected to much higher strains than the
HPP fibers, the degree of microstructural damage was considerably
CRACK HEALING 351
1.0 6 __------------~_;
0.8
.
>-
6)
0.6
...... - - - ..
:>
0
u
-- - - -
CI>
a:
0.4 - --
-70°C
To = ISoC
0.2 f:mQa =300 "1o
-I
f: =5.0 min
00 4
Log Time (min)
Fig. 5. Recovery of SBS block copolymers is shown

as a function of healing time and temperature.
lower in the block copolymer. The mechanical hysteresis and healing

observed in these experiments are in general agreement with results
of other investigators (20-25). A series of papers will be pre-
sented elsewhere detailing kinetics and molecular mechanisms of
healing in block copolymers (2).
FILLED ELASTOMERS
The stress softening effect in filled natural rubber was

extensively investigated by Mullins (26,27), Bueche (28) and others
(29). Many processes contribute to this effect but the dominant
molecular process is considered to be debonding of matrix chains
from the filler particles. Mullins observed that this effect was
reversible, i.e. healing would occur in the rest state after
deformation (27). He determined R values as a function of time and
temperature from stress ratios at the same strain level in the first
and second deformation cycle and found the healing rate to increase
with temperature with an approximate linear logarithmic time depen-
dence.
We examined mechanical damage and healing in carbon black

filled vulcanized natural rubber using the strain history shown in
Fig. 1. The initial results (30) on the effect of strain level and
352 RICHARD P. WOOL
temperature on R are shown in Figs. 6 and 7, respectively. Again,

we observe an approximately linear recovery with log time. The
lower strain causes the least damage and thus higher recovery (in
some cases, the recovery can be independent of strain) and as
Mullins observed (27), faster recovery occurs at the higher temper-
atures. Healing in filled elastomers is considered to involve
read sorption of chains on the filler particle surfaces and closure
of voids in the polymer matrix. Carbon black used in materials such
as auto tires is considered to be a very active filler such that
debonded chains would have litt~e difficulty reattaching once they
diffused to the particle surface. Recovery in these materials is
rapid, similar to recovery rates in SBS and is presumably controlled
by the high mobility of the elastomeric chains.
70
~
~
u
>
o
u
u
a:
Log Time (min)
Fig. 6. Recovery of carbon black filled natural rubber

(74 phr) is shown as a function of healing
time at room temperature for strain maxima of
100 and 200%.
CRACK HEALING 353
...
:>.
Q)
o>
u
Q)
a::
Log Time (min)
Fig. 7. Recovery of carbon black filled natural

rubber is shown as a function of healing
time at healing temperatures of 40 and
1a oC. The maximum strain was 200%.
KINETICS OF HEALING
The generic character of crack healing and its influence on
mechanical recovery has been demonstrated in a number of multi-
component polymers in the past few sections. In this section, a
general kinetic theory of crack healing is presented to describe
mechanical recovery as a function of time, temperature and strain
level.
The mode of crack closure and healing is an important factor in

determining recovery kinetics. Cracks can heal by a "line" mode, a
"point" mode or a combination of such modes. The point mode
involves a "zip-up" mechanism in which healing only occurs at the
crack tip. The line mode involves simulhneous healing at all
points on the crack surfaces. These healing modes are mechanically
and kinetically distinguishable (12,15). Crack healing in brittle
354 RICHARD P. WOOL
inorganic materials such as NaCl (31) or LiF (32-34) occurs prefer-

entially by the point closure mode but the line mode also contri-
butes especially at higher temperatures. Our observations of crack
healing in polymers indicates a line mode of healing is favored.
The following analysis is based on an averaged line mode of healing
but is also applicable to point mode when consideration is made for
geometric factors at the crack tip.
It is proposed that healing depends on the rate of disappear-

ance of damage or defects in the debonded interface between the
crack surfaces. The defects were created during the debonding
process to make the voids and result from molecular rearrangement
and dislocations on or off the crack surfaces. Healing occurs as
the molecular configuration at the interface is restored to its
original or equivalent bonding state in the virgin material. This
theory assumes that the complex process of healing is controlled by
the slowest step, namely defect or damage disappearance via molecu-
lar rearrangement,and that sufficient driving forces exist to allow
the debonded surfaces to approach each other. The driving force for
crack closure prior to healing may arise from a combination of
stored strain energy in the bulk material surrounding the crack, non-
bonded potentials, connecting chains or fibrils between surfaces,
etc.
Let D be the concentration of damage or defects per unit volume

of crack interface. It is assumed that healing occurs at the same
rate at all points on the crack surfaces, Le. via line mode.
Therefore D is a parameter averaged over the nonuniform damage
distribution on the crack surface. The rate law for damage disap-
pearance is unknown and the following general kinetic expression is
proposed, namely
_ dD _ kDn (5)
dt -
where k is a temperature dependent rate constant and n is the order
or power of the healing process. Ideally for crack healing in-
volving multiple cooperative molecular motion, k and n are multi-
ples and sums, respectively, of contributions from individual
kinetic processes but should better be regarded as averaged quanti-
ties reflecting a convolution of non-simultaneous molecular mecha-
nisms with varying rate laws.
The mechanical recovery R is related to D by
D = S1 tH 1 - R) (6)
where S is a proportionality constant and S1 is given in Eq. (3).

Equation (6) reasonably assumes that the time dependent damage is
linearly related to the difference between the fracture energies of
CRACK HEALING 3SS
the first and second deformation cycles.
The initial damage existing at t =0 is
D
o = S1 8(1 - R )
0
(1)
where Ro is the instantaneous recovery component. Integrating Eq.

(5) and substituting for D and Dusing Eq. (6) and Eq. (1),
o
respectively, the mechanical recovery is obtained as a function of
time, temperature and Ro as
(1-R )
_1_
R= _ 0 (8)
n-1 -E/kT } n-1
{1 + Do (n-1)k oe t
where n ~ 1 and k has been defined as an Arrhenius rate constant in

which ko is the pre-exponential term, E is the activation energy, k
is Boltzmanns constant and T is temperature.
RO is a strain magnitude dependent parameter which can be

obtained experimentally and for HPP fibers has the form
R = Je:- m (9)
o
where J and m are constants and e: is the maximum strain shown in
Fig. 1. The value of e: for which Ro = 1 would indicate the strain
range for perfectly reversible cracks andlor viscoelastic behavior
in each system.
To obtain the kinetic constants from mechanical recovery
studies, Eq. (8) is simplified and rearranged as
R = (1
( 1- o) + Kt)a. (10)
1 - R
where K = D~-1 (n-1 )koexp -E/kT
1
and a. = n-1
For Kt »1, we obtain from Eq. (10),
t 1- R 1\ = a.
log \1_R Oj log K + a. log t ( 11)
Thus, a plot of the left hand side of Eq. (11) versus log t, if
linear, provides a. from the slope, and K from the intercept a. log K.
356 RICHARD P. WOOL
Knowing K at different temperatures, the other kinetic constants

can be calculated.
When K and a are known, R can be expressed simply as
(1-R )
o (12)
R = 1 - (1+Kt)a
which is the general expression proposed for healing in multi-
component systems.
For a first order healing process not considered above, n = 1,

we obtain from Eqs. (5), (6) and (7),
1n (~~:o) = kt (13)
for which the kinetic constants can be readily determined experi-

mentally.
The recovery half life, T, is measured at R = 0.5 and can be

determined from Eq. (12) as
T=-k ~[2(1-Ro)J~-1~ (14)

These equations predict that R increases with time and temper-
ature and faster recovery occurs at lower strains (large R).
o
If R0
= 0, the recovery rate is independent of strain. The permanent
damage contribution, D can be determined by either modifying the
definition of R in tgrms of energy contributions for permanent
damage or experimentally from the relation
D
P
= - Reo (15)
where Reo is the observed recovery at very long times.

The recovery behavior predicted by this kinetic approach is
shown in Fig. 8. Using Eq. (12), R is plotted as a function of log
t. The sigmodial recovery curve is in fact the shape of a master
recovery curve which describes healing of cracks and microstruc-
tural damage for most polymeric materials subjected to any particu-
lar strain history, not necessarily that outlined in Fig. 1. At
short times, the lower plateau extrapolates to the instantaneous
recovery contribution and at long times the upper plateau describes
ei ther full recovery or residual permanent damage. The master curve
is associated with a reference testing temperature T and strain,
o
and can be generated experimentally by time-temperature superposi-
tion of R data at different temperatures (2). Further work is being
conducted to investigate the effect of To and Em on the master
curve.
CRACK HEALING 357
100.-----r-----r-----r-----r-----r-~~==--~
80
~
60 a = 0.2
...
>-
II)
K = 1.0
Ro = 0.2
>
0
u
II)
a::: 40
20~----
-2 o 2 10
Log Time
Fig. 8. The theoretical master healing curve is

shown.
APPLICATIONS
The kinetic theory of crack healing was evaluated by applying

Eq. (11) to the experimental data. Figure 9 shows a plot of log(1-
R )/(1-R) versus log t for HPP samples subjected to a maximum strain
o
of 150% (as in Fig. 1) and several healing temperatures. At long
times, Kt» 1, the data is linear from which we obtain the slope a =
0.09 ± 0.01 and n ~ 12. The extrapolated intercept a log K is
temperature dependent and yields an apparent activation energy E ~
28 kcal/mole. Similar behavior is obtained for other strain levels
over the same temperature range. Ro was determined as a function of
strain at room temperature and can be described by Eq. (9) as
R = 0.19 E- 0 • 635
o
At R = 1, E = 7.31% which can be interpreted as the maximum strain
o
limit for rapid crack reversibility in these hard elastic poly-
propylene fibers.
358 RICHARD P. WOOL
-
r2IQ:
- -
-
I I
CI
o
...J
Log Time (min)
Fig. 9. Log (l-R )/(1-R) is plotted as a function

of log hgaling time for HPP fibers at
several healing temperatures. The strain
maximum was 150%.
Using the kinetic constants and the value for Ro at each strain
level, the recovery R was determined as a function of time and
temperature using Eq. (12). Figure 10 shows a comparison of theory
with experiment for the room temperature recovery at several strain
maxima. The theoretical plots are slightly curved but fit the data
better than the linear log time plots used in Fig. 3. Figure 11
shows the comparison of theory with the experimental data for the
125 and 150% strain series at different temperatures. A reasonably
CRACK HEALING 359
IOO~------~--.--------.--~--------r---r-------'-~
- O+6xDV Theory
- - - - - - Experiment
80
.....,>-
>
o
u
a:•
~0~O~------~--~--------~--~------~----~------~--~104
Time (min)
Fig. 10. A comparison of theory with experimental

recovery data for HPP fibers is shown.
The strain maxima ranged from 25-150%
at 28 o C.
good correlation is observed. The effect of T is not intrinsically

g
incorporated in the theory although little recovery is predicted
near T at -17 oC. The temperature dependence can also be introduced
using ga WLF approach (2) and provides a rational basis for the
construction of master recovery curves using time-temperature
superposition principles and another experimental method of deter-
mining activation energies.
Despite the simplicity of the kinetic healing model, it is very

useful for quantitatively describing mechanical recovery data and
is in excellent qualitative agreement with the known physical
processes of crack healing in all materials investigated. The
theory has been applied to block copolymers (2) and filled elasto-
mers (30) with equal success.
SUMMARY
The concept of crack healing was discussed and shown to

play an important role in determining mechanical properties of
semicrystalline polymers, block copolymers and filled elastomers.
Important aspects and consequences of craek healing are summarized
below:
Co)
g
100ir----r----r----r----r----r----~--~----~--~--~
75
25
~
>-
~
Q) 0
>
0
u
Q)
a:
75
-17°C -17°C
E'm = 150 ,,"0 E xpt. E'm =150 ,,"0 Theory
90 0 100 10 5 :0
("')
Log Time (min) J:
»
:0
Fig. 11. The temperature dependent experimental recovery data for c
HPP fibers is compared with the theoretical prediction ~
for strain maxima of 125 and 150%. :E
o
o
r
CRACK HEALING 361
1• Crack healing is a general phenomenon which occurs in

many polymer systems. It affects mechanical behavior and is an
important consideration for mechanistically evaluating constitutive
relations.
2. Healing rates are controlled by temperature, strain his-
tory, environment and the nature of the microstructural damage.
Nonpermanent cracks can heal either instantaneously or in a time
dependent manner.
3. The concept of cracks opening and closing in a time
dependent manner provides an alternative microstructural inter-
pretation of viscoelastic behavior in addition to pure molecular
relaxation processes. Rate dependent mechanical properties which
are controlled by crack healing can be subjected to time-temper-
ature superposition analyses similar to the WLF approach with
viscoelastic polymers.
4. A semiempirical kinetic theory of crack healing based on
the rate of disappearance of damage adequately describes mechanical
recovery due to crack healing and provides the basic rationale for
constructing recovery master curves.
This work is continuing in an effort to further elucidate the
mechanics and molecular mechanisms of crack healing and to evaluate
other mechanical recovery processes suggested by several investiga-
tors (35-39).
ACKNOWLEDGEMENT
The author is grateful to the National Science Foundation for
financial support of this work with Grants ENG-76-09627 and DMR-78-
13942 and expresses thanks to Dr. T.J. Rowland and Dr. R.J. Gaylord
for helpful suggestions and comments.
REFERENCES
1. R.P. Wool, Poly. Eng.! ScL,.1!!., 1057 (1978).
2. R.P. Wool and K.M. O'Connor, papers in press; K.M. O'Connor and
R.P. Wool, BUll. A.P.S., 24, 289 (1979).
3. M. Miles, J. Petermann and H. Gleiter, Progr. Colloid! Poly.
ScL, 62, 478 (1977).
4. R.P. Wool, M.I. Lohse and T.J. Rowland, ~ Poly. Sci., Poly.
Letters Ed., 11, 385 (1979); M.I. Lohse, T.J. Rowland and
R.P. Wool, BUll. A.P.S., 24, 348 (1979).
5. R.P. Wool and w.o:-Statton, Chapter 12 in Applications of
POlymer Spectroscopy, Ed. E.G. Brame, Jr., Academic
Press, New York, 1978.
362 RICHARD P. WOOL
6. R.P. Wool, !!..:..Poly. SCi.,.1i, 1921 (1916).

1. R.P. Wool, Poly Eng.! Sci., in press (1919); C.C. Hsu and R.P.
Wool, BUll. A.P.S., 24, 349 (1919).
8. K.L. DeVries and D.K. Roylance, Progress in Solid State Chem-
istry, ~, 283 (1913).
9. L.A. Davis, C.A. Pampillo and T.C. Chiang, !!..:.. Poly. Sci., 11,
841 (1913).
10. LK. Park and H.D. Noether, Colloid! Poly. SCi., 253, 824
(1915).
11. S.L. Cannon, G.B. McKenna and W.O. Statton, J. Macromol. Sci.,
Part D, 11, 209 (1916). -
12. R.P. Wool, !!..:.. Poly. Sci., Physics Ed., in press (1919).
13. B.S. Sprague, !!..:.. Macromol. SCi., Phys., ~, 151 (1913).
14. E.S. Clark in Structure and Properties of Polymer Films, R.W.
Lenz and R.S. Stein, Eds., Plenum Press, New York, 1914.
15. R.P. Wool, paper #21S, presented at the 69th Annual Meeting of
A.I.Ch.E., Chicago, 1916.
16. H.D. Noether and W. Whitney, Kolloid Z.Z. Polymere, 251, 991
( 1913).
11. H. Ishikawa, H. Numa and M. Nagura, Polymer, 20, 516 (1919).
18. R.P. Wool, Int. J. Fracture, 14, 591 (1918).
19. R.P. Wool, Bull.~.P.S., 23, 386 (1918).
20. D.H. Kaelble and E.H. Cirlin, !!..:.. Poly. Sci., C43, 131 (1913).
21. S.L. Aggarwal, Polymer, 11, 938 (1916).
22. M. Morton, !!..:.. Poly. Sci., C60, 1 (1910).
23. J.L. LeBlanc, !!..:.. ~ Polym. Sci., ~, 2419 (1919).
24. E. Pedemonte et al., Polymer, ~, 531 (1915).
25. Y.D.M. Chen and R.E. Cohen, !!..:.. ~ POlym. Sci., ~, 629
(1911 ).
26. L. Mullins, Rubber Chem. Technol., ~, 281 (1948).
27. L. Mullins and N. R. Tobin, Rubber Chem. Technol., 30, 555
(1951). --
28. A.M. Bueche, !!..:.. Poly. Sci., 25, 139 (1957).
29. A. F • Blanchard and D. Parkinson, Ind. Eng. Chem., 44, 199
(1952).
30. R.P. Wool and S.K. Upadyayula, to be published; S.K. Upadya-
yula and R.P. Wool, Bull. A.P.S., 24, 259 (1919).
31. S.M. Park and D.R. O'Boyle, !!..:.. Mater. Sci., ~, 840 (1911).
32. R. Raj, W. Pavinich and C. N. Ahlquist, Acta Met., 23, 399
(1915). -- -- -
33. D. Nenow and S. Gueleva, J. Mater. Sci., 13, 1195 (1918).
34. D. Nenow and S. Gueleva, Kristall and Techn1k, 13, 429 (1918).
35. H. Brody, !!..:.. ~ POlym. Sci., 22, 1631 (1918):'"
36. J.B. Park and D.R. Uhlmann, !!..:.. ~ Phys., ~, 2928 (1910).
31. D.R. Uhlmann and J.B. Park, !!..:.. ~ Phys., 42, 3800 (1911).
38. J.B. Park and D.R. Uhlmann, !!..:.. ~ Phys., 44, 201 (1913).
39. H.H. Kausch and K. Jud, Macro-19 Proceedings of I.U.P.A.C.,
Mainz (1979).
Discussion
On the Paper by D. Maugis and M. Barquins
L.H. Lee (Xerox Corporation): Mr. Barquins and Dr. Maugis

should be congratulated for this excellent paper presented to this
Conference. Both the breadth and the depth of this paper can serve
as a good model for a general survey for the subject matter. We are
grateful to both authors especially for the excellent, final manu-
script to be appeared in the Proceedings. I am glad that Mr.
Barquins can come to present this paper.
On the Paper by E. Sancaktar, J.S. Cartner, W.I. Grif-

fiths and H.F. Brinson
A. Silberberg (Weizmann Institute, Israel): It seems to me

that failure of the joint is a falure of the material in bulk under
the applied deformation field. Failure does not seem to involve the
actual seam of attachment to the surface support except that
imperfection in the interface zone may be the first points of crack
formation. Is this correct and is it that you are attempting to
model?
H.F. Brinson (Virginia Polytechnic Institute): I do not fully
understand the question. If the discusser is referring to the rail
shear tests, then clearly bulk properties are measured. Failure
properties may not be very valid, however, since failure only
occurred at a stress riser. If the discusser is referring to bonded
joints, then I believe our photomicrographic studies (though none
are presented in the paper) prove that failures likely initiate at
an interface or at least near an interface. They are probably
caused by defects in bonding.
R.A. Gregg (Uniroyal): In your three rail test specimens,is

the sheet of adhesive chemically bonded as well as mechanically
fastened to the metal rails?
363
364 DISCUSSION
H.F. Brinson: No. However, metallic tabs were bonded to the

specimen to minimize the effects of stress concentrations. The
rails were bolted to the specimen with tabs. Stress concentration
was still a problem as all failures first occurred at a bolt hole.
P. Datta (RCA Laboratories): Which shear rates were the shear

measurements made? What was the temperature of the samples in the
shear measurements? Could this difference be related to the
differences in the shear and tensile measurements?
H.F. Brinson: The shear strain rates are given in each figure.
Tests were performed under controlled environmental conditions of
about 25 0 C (10 0 F) and 50% R.H. as identified in the text. Thus
temperature and humidity were not believed to be an influence.
A.D. Jonath (Lockheed Palo Alto Research Laboratory): (1) In

the symmetric rail shear specimen shown for the bulk adhesive, the
failure locus appeared to al terna te from one adhesi ve-adherend
interface to the other. Can you please comment on this? (2) Are
there thus significant tensile stresses introduced during the fail-
ure process?
H.F. Brinson: We attach no particular significance to the

actual fracture planes shown in the presentation for our rail shear
specimen (these photographs are not contained in the final manu-
script). The fractures always occurred at a stress concentration
site or hole under a rail where a complex stress state different
than pure shear prevailed.
The above comment does not negate our conclusion of different

deformation process in bulk shear and bulk tension. The tensile
stresses and strain measured are lower than would occur if the
material at the measurement site were stressed to rupture under pure
shear conditions. The low measured values are still higher than
those predicted from tension. Obviously, actual rupture values
would be higher still.
On the Paper by D.H. Kaelble
D.W. Dwight (Virginia Polytechnic Institute): (1) Would sol-

vent rinse (or other chemical probes) distinguish the SET mechan-
isms? (2) Fowkes argues that polar forces do not contribute to
bonding, so there appears to be disagreement. (3) Are your results
independent of locus of failure?
D.H. Kaelble (Rockwell International): (1) Ion sputter pro-

filing, Auger electron spectroscopy (AES), and ellipsometric anal-
ysis show definite correlations of oxide film morphology and SET
mechanisms (see ref. 3 for details). (2) The special relations
DISCUSSION 365
utilized in this discussion show that polar forces as analyzed by

contact angle measurements of unjoined surfaces directly influence
the strength of the bonded joint. A large number of studies (see
Ref. 7) now confirm the general validity of these surface energy
relations, and their ability to encompass the surface properties of
water and many hydrated surfaces. It is important to point out,
however, that dispersion-polar bonding is a subset of a much larger
class of covalent-ionic mechanisms in which hydrogen bonding and
other interfacial mechanisms can be described (see D.H. Kaelble,
Physical Chemistry of Adhesion, Wiley-Interscience, New York,
1971). (3) As pointed out by eq. 10, the fracture energy relations
can also be applied to predict cohesive failure criteria so as to
provide relative assessments of interfacial and cohesive degrada-
tion.
J. Perkins (AMMRC): In CrO = etching of AI, have you investi-
gated the possible interaction o#adherend on the aging of the resin
and/or the adhesive bond? By plasma etching (air or other), there
might be a change in aging characteristics.
D.H. Kaelble: The standard FPL sulphuric chromate etch process

was not varied in this study. Auger electron spectroscopy of the
surface treated aluminum alloy showed no evidence of residual
chromium at the surface or in the metal oxide film (see Ref. 3 for
details) .
P. Dryfuss (University of Akron): You have given data on the

effect of aging of the substrate in various environments. Have you
studied the corresponding effect of aging of the adhesive?
D.H. Kaelble: The bulk properties of the structural adhesive
have been studied by calorimetric and thermomechanical analysis in
dry N2 with no notable aging effects shown (see D.H. Kaelble and
E.H. Cirlin, J. Poly. Sci., Part C, 35 (1971), pp. 79-100).
P. Dryfuss: My colleagues at the University of Akron have
studied effect of aging of polybutadiene layers. We agree there is
minimal effect of aging in nitrogen, but we do find a very large
enhancement in autoadhesion after aging in air.
On the Paper by C.E.M. Morris, A.G. Moritz and R.G. Davidson
J. Perkins (AMMRC): Of the two batches tested, one older, were

the initial values observed close or diverse? (The graphs were in
terms of %of initial values.)
C.E.M. Morris (Materials Research Laboratories, Australia):

The initial epoxy values were very similar, 183 for Batch 1 and 174
366 DISCUSSION
for Batch 2.
M.A. Grayson (McDonnell Douglas Research Labs): We have

observed epichlorohydrin precursors in cuts from LC analysis of the
tetraglycidyldiaminomethane epoxy system. Have you found any of
these compounds in your analyses?
C.E.M. Morris: No, but we have not specifically looked for

such compounds.
L. Masters (National Bureau of Standards): Will the mechani-

cal tests you mentioned be performed on bonded components or on non-
bonded specimens of the epoxide?
C.E.M. Morris: Tensile shear tests will be conducted on

aluminum single overlap joints. It is also planned to look at
samples of the cured adhesive on a Rheovibron.
PART FOUR
FRACTURE STRENGTHS OF
POLYMERIC SYSTEMS
K. Nakao
3-12-1, Makayama-Sakuradai
Takarazuka-shi
Hyogo-ken, Japan 665
The analyses of tensile bond strength, shear strength and peel

strength are important problems which should be clarified in the
field of the adhesion science and technology. In the case of
interfacial fracture of cross lap joint, Mr. Masuoka suggests that
the failure of the tensile adhesive joint is governed by an energy
criterion. The stress distribution and the maximum stress in shear
at the circular area of lap joint carrying the bending moment have
been shown by Dr. Yamaguchi. On the peeling of flexible materials,
Dr. Igarashi shows: (1) The steady peeling strength for the elastic
adhesives is determined by the surface energy of the adhesives. (2)
The energy dissipation of the adhesives causes the increase in the
steady peeling strength. (3) On the contrary, the initial peeling
force is derived from the maximum stress in the adhesive layer.
The molecular weight of adhesives is also one of the most
important factors governing bond strength. Dr. Tsuji shows the
superposition of peel strength with molecular weight of adhesive,
temperature and time.
The composition and reaction of adhesives can also govern bond

strengths. Recently for making denture, a new technique, that is,
the fluid technique using self-curing acrylic resin has been de-
veloped. By Mr. Harashima, the effect of the crosslinking of PMMA
has been shown. These two papers are published elsewhere.
To my great regret, Dr. Miyairi could not attend our Conference

for personal reasons. However, his paper on the deformation and
shear strength of FRP adhesive joint is included in this part.
369
370 INTRODUCTORY REMARKS
I was pleased to accept Dr. Lee's invitation to preside at this

session concerning the "Fracture Strength of Polymeric Systems."
We, Japanese scientists, are grateful to the Chairman, Dr.

L.H. Lee and Co-Chairman Dr. W.H. Grant for organizing and conduct-
ing this Conference.
Criterion of Interfacial Fracture
on Tensile Adhesive Joint
Mineo Masuoka and Kazumune Nakao
Osaka Prefectural Industrial Research Institute
2-1-53, Enokojima, Nishi-ku, Osaka, Japan 550
ABSTRACT
The present investigation was undertaken to verify the cri-

terion of the interfacial bond failure. The time-temperature
superposition, the failure envelope, the dependence of the tensile
bond strength on the degree of crystallinity and the fractography of
the interfacial region of the adhesive were observed by the use of
steel-nylon 12-steel cross lap jOint. The ultimate tensile proper-
ties of the joint were found to be associated with the viscoelastic
nature of the adhesive rather than with a thermodynamic criterion.
The experimental results suggest that the Mises hypothesis can be
applied to the tensile adhesive joint as the fracture criterion.
INTRODUCTION
The performance of an adhesive joint is usually evaluated in

terms of a breaking load of the joint under different test condi-
tions. Therefore, the fracture criterion of the adhesive joint
plays an important role in predicting when the bond failure will
occur. In order to predict the bond strength quantitatively, the
fracture criterion must be combined with the stress distribution in
the adhesive layer. It has been reported by Gent (1) that the
observed bond strength of the joint subjected to tensile, pure shear
and peeling separations is consistent with an energy criterion for
adhesive failure and that a fracture mechanism based on Griffith's
energy criterion is very significant. In order to explain the
mechanical behaviors of the adhesive joint, a criterion for the
interfacial failure has been introduced by Hata (2): namely, the
bond failure will occur at the interface when the elastic work of
371
372 MINEO MASUOKA AND KAZUMUNE NAKAO
deformation of the adhesive reaches a critical value.

In the case of the internal fracture, we assumed the triaxial
stress state in the adhesive layer and derived a theoretical formula
for the tensile bond strength of a butt joint (3). The formula was
given by combining the stress distribution in the adhesive layer
under large plastic deformation with the fracture criterion of
Mises or Tresca, which is the hypothesis of the maximum shearing
strain energy or the maximum shearing stress. This formula includes
only measurable parameters and is expressed by the following form,
(ab/a t ) = 1 + 1 (a/hb ) 2 e: zb (1)
hb = ho (1 + e: xb )
where a b and at are the tensile bond strength and the yield strength
of the adhesive material, which was obtained from uniaxial tension
of dumbbell, e: b and e:z~ show the longitudinal tensile and the
lateral contra~ive stral.ns at breaking the joint, a /h is the
aspect ratio of the adhesive layer, and h means tHe °adhesive
thickness. In the case of aluminum-ethyl~ne vinylacetate co-
polymers(EVAs)-aluminum butt joint, the following empirical rela-
tion was obtained (Fig. 1).
2
(ab/a t ) = 0.35 + 1.8 (a/h b ) e: zb (2)
It is found from the agreement between both Eqs.(1) and (2) that the
hypothesis of Mises or Tresca is applicable to the criterion of the
internal fracture of the butt joint. The Mises and Tresca condi-
tions describe quite well the mechanical behavior of a ductile
material and are adopted frequently as the criterion of a ductile
fracture.
The present investigation has been undertaken to verify the
criterion of the interfacial fracture and the relationship between
the tensile bond strength and the mechanical properties of the
adhesive polymer by the use of steel-nylon 12-steel cross lap Joint
of the interfacial bond failure. It is our opinion for metal -
polymer adhering system that the bond strength is determined by the
mechanical and viscoelastic properties of the adhesive polymer, by
the stress state developed in the adhesive layer and by the local
stress which depends on the geometry of the adhesive layer. In
connection with them, one of the essential features on the mechani-
cal behaviors of the adhesive joint is the location of the site
where the bond failure is most likely to be initiated. Since
Bikerman (4) has pointed out that a true adhesional (or interfacial)
failure is improbable, the problems on the locus of failure and on
the weak boundary layer of the adhesive joint have been argued from
various viewpoints by many investigators (2,5,6,7). In the case of
the tensile adhesive joint such as butt or cross lap joint, a stress
concentration will be developed in the adhesive because the defor-
TENSILE ADHESIVE JOINT 373
.r::.
~ 5.0r-- - -----;-,-------;-I- - - - - - ,
~ a i~
"iii a_a ............
~ 4.0 ~cfJ• .,. . . .Adhesives :
.8 ,ce-. o'~.~
VAc·l. MI
~ ~ 3.0+-----:.,,""""_ •• --=----.., 0 19.0 20
~..n~ 0-;' .'~ • 18.5
a 31.1
300
24
~ 2.0 () 29.0 300
-5
& 1.0 +-....L...-....L...........---L.--+---'----JL--.L.-..L...--t--'---L..---L---J.---l
1.0 1.5 2 2.0 2.5
( Qo/hb )'£zb
Adhesive: ethylene vinylacetate copolymers (EVAs)

VAc: vinylacetate content in wt %
MI melt index number
Fig. 1. The plot of (0 b/o t) against (a/h b )2.
€zb for the internal fracture of the
butt joint, according to Eq. (1).
mation of the thinner adhesive layer in lateral (shear) direction is

restrained by the high modulus or rigidity of metal adherend. That
is, this restraint exerted by the high modulus material on the lower
one is expected to be highly localized at the interface and the
failure always occurs within the weaker phase. Therefore, we
consider that the existence of the interfacial fracture of the
adhesive joint may be valid and even in such case, the principle of
analytical mechanics can be applied. It is more practical, in this
sense, to define the interfacial fracture in terms of the state in
which no significant amount of residual adhesive remains on an
adherend surface after breaking the jOint by the use of a Simple
method. We confirm such a state by the use of microscopic methods
such as microscopes and X-ray microanalyzer.
However, it is very difficult to set up and solve the equation

describing the cross lap joint by the use of analytical mechanics,
because of a bend of the adherend in response to the applied tensile
loading. In this study, therefore, the criterion of the interfacial
fracture will be evaluated from experimental observations of the
ultimate tensile properties of the cross lap joint.
A crystalline polymer is essentially a composite material made

of two phases, that is, one crystalline and the other amorphous, and
its mechanical properties depend on the degree of crystallinity,
which changes with its thermal history. When a crystalline polymer
is used as an adhesive, there is a large merit that the mechanical
properties of the adhesive can be varied by thermal treatments

without any change in wetting behavior at making up the adhesive
joint by hot-melt bonding.
EXPERIMENTAL
Nylon 12 powder of XF 5000 supplied from Toray Co. LTD. was

used as the adhesive. Before making up the adhesive bond, the
powder was dried well under high vacuum at room temperature. The
adherend of mild steel with specification JI~ SPCC was machined into
25.2mm x 50.4mm x 5.0mm (bonded area = 1 in ) or 10.0mm x 50.0mm x
5.0mm (bonded area = 1 cm2 ). The adherend plates were polished by
Emery paper, degreased in distilled toluene by the use of ultrasonic
cleaner, and finally dried under vacuum at 100 0 C for 4 hrs.
The cross lap jOint was formed at 200 0 C for 10 min. in hot
press and allowed to cool or quenched by immersing it into dry ice -
acetone mixture. The average cooling rates were about 20 0 C/min and
730 0 C/min, respectively. In order to change the degree of crystal-
linity of nylon 12 adhesive, the bonded specimen was annealed at
150 0 C in nitrogen atmosphere. The annealed specimen was aged at
room temperature for a week in vacuum desiccator to relax the
residual stress developed in the adhesive layer during cooling
process (8). The tensile bond strength decreased to about one-half
during 4 days and then held constant for over two weeks.
The tensile bond strength was measured by the use of Instron

tension tester in accordance with ASTM D-1344-54T. A universal
joint and some special attachments were uSed to make straight pull
of the adhesive joint. The observed tensile bond strength was
averaged over five specimens. The deviations from the average were
within ± 7%.
The dependence of the mechanical properties of bulk nylon 12,

such as the yield strength, the yield strain, the Young's modulus
and the tensile modulus (ASTM D-638-61T), on the degree of crystal-
linity was determined from uniaxial tension of nylon 12 dumbbells
annealed. These quantities were also averaged.
The density of the adhesive or the dumbbell was measured at

25 0 C by the use of a density gradient column. The density was also
averaged over five pieces for each specimen. The degree of crystal-
linity was calculated from d =0.99 g/cm 3 and d =1.10 g/cm3 by the
a c
use of the following relation (9).
1/d = x/d + (1-x)/d

c a
where d and dc are the density of amorphous and crystalline
a
regions, respectively, and x is the degree of crystallinity of the

crystalline region having the density d.
Reflecting microscope and X-ray microanalyzer (CKa) were used
to observe the locus of failure and the fracture surface of the
boundary region of the adhesive.
The surface states of the steel adherend were investigated by

the use of reflecting microscope and X-ray microanalyzer to confirm
the locus of failure. The results are shown in Fig. 2. We did not
observe any change in surface appearance under the scope of reflect-
ing microscope (top photograph). White spots, which were detected
by CK ~ray and shown in photographs (B), (C) and (D) of Fig. 2, mean
a region where carbon atoms are concentrated on the adherend
surface. Photographs (B) and (C) show the surfaces before bonding
and after breaking, respectively. The density of white spots is the
same in both photographs (B) and (C). On the other hand, photograph
(D) shows the feature of the adherend surface coated with very thin
layer of organic compound, which would be identical with the state
having a residual adhesive or a weak boundary layer on the surface.
This surface was prepared by immersing a purified adherend into the
1 % toluene solution of polyethylene oxide oligomer having the
molecular weight of 700 for 1 min. and then by drying as to keep it
standing vertically in a vacuum desiccator. Evidently, the surface
of (D) shows the higher density of CKa spots than others. These
results indicate that there is no significant amount of a residual
adhesive or a weak boundary layer on the adherend surface after
breaking the joint. These photographs in Fig. 2 were chosen at
random from a large number of similar montages on the scope of both
analyses and certain features shown in chosen photographs are
considered to represent whole features. In the case of steel-nylon
12-steel cross lap joint adopted in this study, we can conclude that
the interfacial fracture occurred under all test conditions.
The temperature-time superposition and the failure envelope

are important means for characterizing the ultimate properties of
viscoelastic materials and will provide the criterion for predict-
ing when fracture will occur in a specimen subjected to different
test conditions. From these viewpoints, the dependence of tensile
bond strength on both temperature and. strain rate is examined by
applying these methods to the interfacial fracture of the cross lap
joint. In this study, the apparent strain rate, which was defined
as a crosshead speed divided by the adhesive thickness, was employed
to represent an average deformation rate of the adhesive because the
true rate could not be measured due to the bend of the steel
adherend under applied loading.
(A) The adherend surface after breaking

observed by reflecting microscope.
(B) The adherend surface before bonding
detected by CKa.
(C) The adherend surface after breaking
detected by CKa.
(D) The adherend surface having thin layer
of oligomer detected by CKa.
Fig. 2. The features of adherend surfaces charac-
terized by the use of reflecting microscope
(top) and X-ray microanalyzer (bottom).
The tensile bond strength increases continuously with the

apparent strain rate up to a critical rate and then decreases
abruptly, as shown in Fig. 3. An abrupt decrease in the bond
strength with the peel rate was reported for the peel separations by
Gent (10), Kaelble (1) and Fukuzawa (11). In their results, the
change in the fracture mode from "cohesive" to "adhesive" at the
transition point was 40bserved. 5he critical strain rate is in the
range between 3 x 10 and 1 x 10 %Imin. The transition occurs at
higher rate at higher test temperature, as shown in Fig. 3. In
accordance with our calculation on the basis of the rate process,
when the time to break is equal to the relaxation time of the
adhesive polymer, such transition will occur (12). This critical
o 1 .2 -1 3 4
Log E (min)
Adhesives: Nylon 12, 0.2 mm in thickness

and high density polyethylene
(R.D.-PE), 0.15 mm in thickness.
Fig. 3. The change in the tensile bond strength with
the strain rate (€) and the test temperature
in the case of the interfacial f~acture of
the cross lap jOint. The bonded area is
of 1 cm2 .
rate is identical with the "critical deformation velocity" spe~i

fiSd in the region of impact test for viscoelastic materials (10 ,.
10 %/min).
The fracture surfaces of the boundary region of the adhesives
are given in Fig. 4. It shows the change in the fracture pattern
with the strain rate. The fractography obtained from the inter-
facial fracture indicates that the failure mode changes abruptly
from the ductile to the brittle one at the critical rate because of
the glassy manner of the adhesive deformation under higher rate than
the critical velocity. The interfacial bond failure will be
initiated by ductile or brittle rupture of the boundary region. We
higher strain rate

. . .
E< E<E
c
Adhesives: (A) and (B) are high density
polyethylene.
(C) and (D) are nylon 12.
Fig. 4. The change in the fracture surfaces with the
strain rate. € is the critical deformation
velocity. c
wish to emphasize on the basis of the fractographic observation for

the ductile range that the shear stress or strain acting at the
boundary region plays a very important role in the initiation of the
interfacial fracture and consequently, in the determination of the
tensile bond strength. Needless to say, such shear components at
the boundary region depend on the adhesive geometry and on the
mechanical properties of the adhesive polymer.
The reduced tensile bond strength, 0b x To/T, is plotted in

Fig. 5 against the apparent breaking strain, lb' which was estimated
by the conventional method reported previously (12), where 0b is the
tensile bond strength observed at each test temperature (T) and T
means the reference temperature (T = T + 50 oK). The failurg
envelopes obtained for the interfac~al f~acture indicate a common
pattern which would be expected for a viscoelastic material (13).
The tensile bond strength, reduced to 363 0 K, is plotted in Fig.

6 against the apparent strain rate (€). The results obtained at
different test conditions are superimposed satisfactorily to get a
single master curve. The master relation for the interfacial bond
failure exhibits the Arrhenius's type. The activation energy of 51
kcal/mol is calculated from the dependence of the shift factor, aT'
on 1/T, as shown in Fig. 6. This activation energy agrees well with
that obtained from the rate-temperature dependence of ductile frac-
ture for viscoelastic materials. However, the activation energy of
51 kcal/mol is much larger than 9 to 13 kcal/mol reported by Illers
for a rheological rate process of nylon 12 (14).
The Ferry's shift factor, aT' is given approximately by aT =

n TIfl To· The shift factor is equal to the ratio of the viscosity (n)
at temperatures of T and T. Even though the time-temperature
. 0
dependence of the tenslle bond strength would result from that of
the viscosity, the activation energy will be expected to be 10 to 20
kcal/mol. The rate process on the fracture of the adhesive joint
consists of many elemental steps and the rate-determining step is
still uncertain. Consequently, it is very difficult to explain the
activation energy of 51 kcal/mol in terms of the rheological models
proposed.
The master relation and the failure envelope obtained for the
interfacial fracture of the cross lap joint are quite similar to
those obtained for a viscoelastic material. This similarity in the
ultimate tensile properties indicates that these tensile properties
of the adhesive joint are associated with the dynamic effect of a
viscoelastic nature of the adhesive polymer rather than with a
thermodynamic surface criterion, such as the work of adhesion,
wettability and so on. A similar conclusion has been obtained for
the internal bond failure, that is, Gent (15) has applied the WLF -
superposition and a fracture mechanism in terms of Griffith's
energy criterion to the cohesive strength. In accordance with
Gent's findings, the observed bond strength reflects the visco-
elastic responses of an adhesive polymer and the bond failure will
occur when the energy stored elastically in the adhesive in the
vicinity of the initial flaw and released by growth of the initial
20 I
- Nylon 12 (To=363K)
ee~
I 0 +-
ee
12
110
8
Temp. °C
0
100 CD
00 e
.~\ -
'. 90.
80 •
e Ie. 70 e
8(}
•• •
i 60 ~
·0··;-
50 e
<D • • - 25 e
60
o I
'0
0 0 00
I
I
e: . . ;
0
40
···.:.r·l
/.' H.D.-PE I
..
~~ e (To=20dK)
~CJ/i
~ I . . ..'.
20
<D.
~CD -. ..........
....
~
0
o 20 40 60 80 100
Apparent breaking strain(E b 0/0)
Fig. 5. The failure envelopes of the interfacial

fracture of the cross lap joint in the
lower strain rate region. The adhesive
thickness is 0.2 mm for nylon 12 and 0.15
mm for high density polyethylene (H.D.-PE),
respectively. The bonded area is of 1 cm2 .
flaw is sufficient to meet the energy requirement for growth. It is

suggested from the fractographs shown in Fig. 4 that the fracture
mechanism proceeded in the boundary region will be analogous to that
pointed out by Gent (15). Phenomenologically, the problem of
characterizing the ultimate tensile properties for the "inter-
facial" fracture would be essentially similar to that for the
"internal" bond failure.
In the case of the interfacial fracture of steel-nylon 12-

steel cross lap joint, the dependence of the tensile bond strength
on the degree of crystallinity of the adhesive is examined in order
to predict the fracture criterion from the relationship between the
tensile bond strength and the mechanical properties of the adhe-
1 . 2 3-2 -1 o 2 . 3 5
Log C (minI ) Log CaT
Fig. 6. The time-temperature superposition and

the dependence of the shift factor on the
temperature for the interfacial fracture
of the steel-nylon 12-steel cross lap joint.
The adhesive thickness is about 0.20 mm.
The bonded area is of 1 cm2 .
sive. An annealing of the bonded specimen resulted in the increase

in the density or the degree of crystallinity of the adhesive. The
results obtained are shown in Fig. 7. The density on the abscissa
in Fig. 7 shows the average value of the bulk adhesive. The degree
of crystallinity at the interfacial region is higher about 5 %or
more than that of the bulk nylon 12 adhesive, regardless of the
crystal morphology (16). The density of the interface at which the
bond failure occurred is expected to be higher than that shown in
Fig. 7. The cross section of the adhesive is given in Fig. 8. The
crystallites in the adhesive allowed to cool are larger than those
in the quenched adhesive. The transcrystalline region is not grown
at the interface in this case.
The observed tensile bond strength increases markedly with the

density but is independent of the crystallite size. For exa~ple,
the increase in the tensile bond strength from 50 to 200 kg/cm can
be attributed to the increase in the degree of crystallinity from 25
Degree of crystallinity(ofo)
25 30 35 40
25 0
.-..
N bonded I
~ J area 0
.:/+ ~
~20 0 e-cm 2
-
'-"
Ol
o-i n 2
-
C
~ 150
1/1
/,e 0
~ 8J 0'9/0
-0
C
.8 10() ,e 8 ,/ 0
~ ,60
'ijl ,0
c 0,,801
~ 50
1.015 1.020 1.025 1.030
Density of nylon 12 adhesive
(g Icm 3 )
Fig. 1. The dependence of the tensile bond strength

on the density or the degree of crystallinity
of the adhesive for the interfacial fracture
of the cross lap joint. The adhesive thickness
is about 0.10 Mm.
(A) The quenched adhesive.

(B) The adhesive allowed to cool.
Fig. 8. The cross sectional microphotographs of the
nylon 12 adhesive molded against the steel
adherend. The adhesive thickness is about
0.2 Mm.
to 40 %. Fig. 1 shows also the well-known fact that the larger is

the bonded area, the lower is the bond strength per unit bonded
area. The dependence of the tensile bond strength (Ob) on the
density (d) is expressed by the following equations.
For the bonded area of 1 in 2 ;

2
Log
°b
= 45.82d - 44.86 (kg/cm ) (1)
2
and for the bonded area of 1 cm ;
2
Log 0b = 32.02d - 30.54 (kg/cm ) (4)
The higher degree of crystallinity gives the higher modulus of the
adhesive. From the analysis of Gent (15) and Hata (2), the tensile
bond strength will be expected to increase with the Young's modulus
of an adhesive. On the other hand, a thermodynamic or surface
chemical interaction force operative at the interface would be
expected to be almost constant since an effective contact area at
the interface will be altered even slightly by the increase in the
density. It is very difficult to consider that the bond failure
occurred at the weak boundary layer, which will grow by migrating
the low molecular weight compound such as antioxidant to the
interface during annealing. A strength of such weak boundary layer
is independent of the degree of crystallinity of the adhesive.
Therefore, the increase. in the tensile bond strength with the
density, as shown in Fig. 1, could not be observed, even if the bond
failure would occur at the weak boundary layer.
When a crystalline polymer is crystallized in contact with
metal surface, transcrystal sometimes grows at the interfacial
region. The preferable effect of the transcrystalline region on the
adhesive joint in relation to the weak boundary layer has been
reported by Sharpe (5) and Schonhorn (11). On the other hand, it
has been pointed out by the authors (16) that the crystal morphology
of transcrystal itself can not contribute to the increase in the
tensile bond strength. The transcrystalline region has a higher
degree of crystallinity and higher modulus (18). The fact that the
transcrystalline region gives the higher bond strength is due to its
high degree of crystallinity rather than its crystal morphology.
Mechanical properties of nylon 12, such as the yield strength
(Ot)' the yield strain (€t)' the Young's modulus (E) and the tensile
modulus (E t ), were measured. The results are given as a linear
function of the density or the degree of crystallinity, as shown in
Figs. 9 and 10. These quantities were obtained from uniaxial
tension of nylon 12 dumbbells annealed under nitrogen atmosphere.
An example of the stress-strain curve of the dumbbell and the
definition of these quantities adopted in this study are shown in
Fig. 11.
DE"gree of crystall ini ty (010)
L
20 25 30 35 40
18000-rf---1----;+------:-+--~
-
N
I
TensilE"
I
~.;:••
.!
(() .
E 16
~ - -modulus dl» ° .tOo~ to
~Et~r~'cPo
0'1
~
"3
"8
.. •
•
~
~ •• --~~----+
E -
•
/ ----t----+---+---- ./
~ ~. 0"
-
III
C
~./
"oo()T· •
Qj to
to
~o
cit
III
"0'1
C
-+/°1 Young's
:::::J ~ •• modulus (E)
§?
;.~--~----~-----+----~
l010 t015 1.020 1.025 1.030

DE"nsity of nylon 12 (g/cm 3 )
011 - crystallize-d at 150°C for 30min
o - allowe-d to cool from the- me-It .
• - Qut'nche-d from tht" melt .
Fig. 9. The dependence of the Young's and tensile

moduli on the density or the degree of
crystallinity. These quantities were
obtained from uniaxial tension of nylon
12 dumbbells annealed.
The Young's modulus was measured wi thin elastic limit (e: <
0.05%). The tensile modulus adopted is identical with that of ASTM
D-638-61T. The similar relations for nylons have been reported by
Starkweather (19).
The dependence of the tensile bond strength on the degree of

crystallinity of the adhesive, as shown in Fig. 1, can not be
explained by any single mechanical quantity of the bulk polymer.
Therefore, two tensile properties were calculated by the use of the
data shown in Figs. 9 and 10. One is the product of the tensile
modulus and the yield strain, Et.e: t , which means the tentative yield
strength of bulk nylon 12 when the linear stress-strain relation is
assumed. The other is the half of the product of the yield strength
and the yield strain, i.ot.e:t, which shows the work of deformation
during yielding. The dependence of the values, both Et • e: t and
Degree of crystallinity(OJo)
_ 20 25 30 35 40
~ 15~------~------~------'1~-----.i
~ 10 £t I _~O~!'1d6~1·'a#~-
~
;
~
; E;: ---I - -
I- ID
~______- L______~______~______~
-a 1: 800.....----------;------;--1
CiI_
~l700 ~
:g ~ 60
.~ CiI
> ~ 5uu+-------r-----~r_----_;------~~
(f) 1.010 1.015 l020 1.025 1.030
Density of nylon 12(g/cm3 )
ID -crystallized at 150°C for 30min.
0- allowed to cool from the mell
_ - quencht'd from tht' mt'll
Fig. 10. The dependence of the yield strength and

strain on the density or the degree of
crystallinity. These quantities were ob-
tained from uniaxial tension of nylon 12
dumbbells annealed.
80
-~
N
60
t
6t
-
yield
~ strength
\ I)
\I)
~ 40
.....
c./)
tensile modulus
5 10 15 20
Strai n (0/0)
Fig. 11. An example of the conventional stress-strain

curve of nylon 12 dumbbell and the definition
of the mechanical properties adopted in this
study.
~.at.€ , on the degree of crystallinity or the density is given in

Fig. 1~. The density dependence of the former is similar to that of
the tensile bond strength, as shown in Fig. 1. This similarity
between Figs. 1 and 12 suggests that the tensile bond strength for
the interfacial bond failure would be expected to reflect the
yielding behavior of bulk polymer used as the adhesive. However,
the tensile bond strength observed is very much lower than the
tentative yield strength defined as Et.€t.
The cross lap jOint has a large stress concentration, resulted
from a bend of the adherend under applied tensile loading. From the
other experiment in our laboratory, the stress concentration co-
efficient of the cross lap joint adopted in this study was evaluated
to be 6 to 10 times a butt joint. The coefficient of a butt joint is
in the range of 1.0 to 1.2 as reported by Kobatake on the basis of a
stress analysis (20). When the stress concentration is absent, the
tensile bond strength will be expected to approach to the value of
the product, Et.€t.
Degree of crystallinity("Io)
,..... 20 25 30 35 40
N
E 50
~
C'I
0-
4 (} I I
~%, o
0 8 ~6\l>=
w 3:0
0_ 0 - 0 - <0()
~ 2:0
_ IN
0
I I I
........
2 -----+-----;------~--aP
N
OJ'
E
Jr! _--+-_-+-_-1- 8~
g o
w - -----r---o ~--~----+
90 9063
w -
><
800
~oOo
0 ___ °
I
I
l010 1.015 1.020 1.025 1.030
Density of nylon 12 (gJcm 3 )
Fig. 12. The dependence of the product of the tensile

modulus and the yield strain (E t x €t) and the
half of the product of the yield strength and
the yield strain (! x a x €). These quan-
tities were calculated ~y th~ use of the data
shown in Figs. 9 and 10.
When the thin layer is bonded between rigid metal platens and
tested under a uniaxial tension. Such specimen can not contract
laterally and the triaxial stress state develops in the specimen.
In accordance with our analysis (3,21), under triaxial stress
state, the stress required to rupture the thin layer is always
higher than the yield stress of the bulk polymer obtained from a
uniaxial tension. When the interfacial bond failure occurs, the
interfacial region of the adhesive would be subjected to the stress
field equivalent to the yielding stress state under uniaxial ten-
sion of bulk polymer. However, the observed bond strength will be
lower than the yield strength resulting from a stress concentra-
tion.
In order to determine the fracture mechanism of the inter-

facial bond failure, the fracture surfaces were observed by the use
of the reflecting microscope. The fractographic patterns are given
in Fig. 13, which shows the change in the crack patterns with the
density of the adhesive. The higher is the density, consequently,
the higher is the tensile bond strength, the larger is the crack
depth and width. These crack patterns of the boundary region is
quite similar to the appearance observed on cup-bottom when the cup-
and-cone fracture occurs in a specimen under tension. The fracto-
graphy shown in Fig. 13 indicates that the interfacial fracture of
the joint is attended by the ductile rupture of the interfacial
region of the adhesive. In other words, the interfacial bond
failure can be considered as such processes that the interfacial
region of the adhesive is ruptured in a ductile manner and at the
same time, the bond separation occurs at the interface. This
feature on the interfacial bond failure is supported by the depen-
dence of the ultimate properties of the joint on the time-tempera-
ture and on the degree of crystallinity. That is, this dependence
results from the ductile deformation and fracture manner of the
interfacial region of the adhesive.
In general, the ductile fracture will be caused by the shear
component of a stress or strain energy induced in a specimen under
tension. It suggests that the interfacial bond failure will occur
when the shearing stress or the shearing strain energy operative at
the interfacial region increases up to a critical value equivalent
to the yielding point of the adhesi ve polymer under uniaxial
tension. The mechanical behaviors of ductile materials are
described quite well by the hypothesis of Tresca or Mises. They are
employed frequently as the fracture criterion. In accordance ~ith
the Tresca condition, when the shear stress in a specimen reaches
the maximum value equivalent to the yielding point of uniaxial
tension and with Mises condi_ion, when the shearing strain energy in
a specimen reaches the maximum value equivalent to the yielding, the
ductile fracture will occur, respectively. Both hyoptheses are
valid criteria for the internal fracture of the butt joint as
pointed out by the authors (3,21). We can also consider that the
Fig. 13. The change in the fractographic patterns

observed on an interfacial region of the
nylon 12 adhesive with the density (d).
Mises and Tresca hypotheses can be applied reasonably to the

interfacial fracture of the tensile adhesive joint as the fracture
criterion.
When a stress distribution in the adhesive could be obtained
for the cross lap joint under the condition of a bend of the
adherend, for example, by the use of the finite element method, an
analytical equation describing the interfacial region could be set
up and solved by combining the stress distribution with the hy-
potheses. It would be possible to establish the fracture criterion
for the interfacial bond failure. It has been pointed out by Gent
(1) that the bond failure, both for "cohesive" and "adhesive", is
governed by the energy criterion generalized from Griffith fracture
criterion. However, it is very difficult to determine which kind of
the hypotheses, either the shearing stress or the shearing strain
energy, will be predominant for the interfacial fracture of the
adhesive joint.
In the case of the internal fracture of aluminum-EVA-aluminum
butt joint having a rectangular bonded area (10mm x 32mm), an
interesting result was found as seen in Fig. 14, which shows the
3
H
-i 101
b
go
-
I
0'
00' - ,
g~"-
I
I
~ _o
0 0
I cross-sectional
I
-
shope of a:lhe-s ive .
I61......i.. B'" . . .. j "~ -;T

" 0
..0.•0 0 -0- square
g .. ' ·D·· · rectang(e
I I
o I I I
I !
I
123 5 10 15 20
Adhesive thic:kness(mm) Aspect ratio (a. fh.)
Adhesive: ethylene-vinylacetate copolymer (EVA)

having the vinylacetate content of
19 wt %and the melt index number of 20.
Fig . 14. The lowering in the tensile bond strength
with the anisotropic shape effect of a
bonded area in the case of the internal
fracture of butt joint.
dependence of the reduced tensile bond strength (a b/a t) on the

adhesive thickness or the aspect ratio of the adhesive, where a b and
a mean the tensile bond strength and the yield strength of the EVA
05tained from uniaxial tension of dumbbell, respectively, and h is
o
the adhesive thickness . The observed tensile bond strength for the
rectangular joint decreases with the anisotropic shape effect to
about one-half that for the square joint even though both joints
have same bonded area. The effect of an anisotropic shape on the
tensile bond strength will be explained by the Mises hypothesis in
the same manner derived from Eq. (1) but not by the Tresca one. a ,
a , and a are the principal stresses. The x axis shows the tensile
dlrection~ The y and z axis are the major and the minor one in
lateral direction, respectively. The Mises criterion is expressed
by the following equation.
2 2 2
(ax - a)
y + (a y - a)
z + (a z - a x ) (5)
Eq. (5) can be rewritten:
For the square cross section (3);
ax - a y = at (6)
and for the rectangular cross section;

ax - a z = (2//j)at
Eq. (6) shows the same form of the Tresca criterion. Eq. (7) is
given by neglecting the lateral contract strain on the major axis.
(Ob/Ot)REC and (Ob/Ot)SQ is the tensile bond strength of the
rectangular and the square joint, respectively. The following
relation is finally given by the use of the same manner derived from
Eq. (1).
('l/Ot)REC = (/3/2) (Ob/Ot)SQ (8)
That is, the tensile bond strength of the rectangular cross section
lowers to about /3/2 that of the square one when both joints have
the same bonded area. Eq. (8) agrees well with the result shown in
Fig. 14. This anisotropic shape effect indicates that the Mises
hypothesis is significant as the fracture criterion of the tensile
adhesive joint. The analytical formulation of the anisotropic
shape effect which has not been achieved in detail will be reported
in a forthcoming paper.
CONCLUSIONS
In the case of the interfacial fracture of steel-nylon 12-

steel cross lap jOint, the importance of the viscoelastic or
mechanical responses of the adhesive polymer to the tensile bond
strength is shown in this study. The temperature-time super-
position, the failure envelope, the dependence of the tensile bond
strength on the degree of crystallinity of the adhesive and the
fractography of the interfacial region of the adhesive were inves-
tigated to verify the fracture criterion of the tensile adhesive
joint.
It will be considered that the interfacial bond failure is

identical with the ductile or brittle rupture of the interfacial
region of the adhesive. Such rupture manner depends on the test
conditions. Consequently, it is found to be reasonable that the
observed ultimate tensile properties for the interfacial fracture
of the adhesive joint are quite similar to those obtained from the
tensile test of a ductile material. The activation energy of 51
kcal/mol was estimated by the Arrhenius plot of the shift factors.
This value is much larger than that obtained from a rheological rate
process of nylon 12, suggesting that a complicated fracture
mechanism acts in the interfacial region.
The dependence of the tensile bond strength on the degree of

crystallinity is quite similar to that of the tentative yield
strength of bulk nylon 12, defined by the product of the tensile
modulus and the yield strain.
These facts observed in this study suggest that the Mises

hypothesis is very significant for the interfacial and internal
bond failure as the fracture criterion and is powerful especially to
analyze the anisotropic shape effect on the tensile bond strength.
REFERENCES
1. A.N. Gent and A.J. Kinloch, "Adhesion of Viscoelastic Mate-

rials to Rigid Substrates. (III) Energy Criterion for
Failure", !!..:. Polymer Sci., A-2, .2" 659 (1971).
2. T. Hata, "Rheology of Adhesive Failure. (II) Rheological Con-
sideration of the Transition between Interfacial and
Cohesive Failure", J. Adhesion Soc., Japan, 8, 64 (1972)
and "Mechanism of Adhesive Failure", J. Adhesion 4, 161
(1972). - -
3. M. Masuoka and K. Nakao, "Effect of Aspect Ratio on Tensile
Bond Strength for Butt Joint of Internal Fracture" in
"Adhesion Measurement of Thin Films, Thick Films and Bulk
Coatings", ASTM STP-640, K.L. Mittal, ed., American
Society for Testing and Materials, Philadelphia, (1978).
4. J.J. Bikerman, Chapter VI. "Final Strength of Adhints" in "The
Science of Adhesive Joints", 2nd. Ed., Academic Press,
New York, (1968).
5. L.H. Sharpe, "The Interphase in Adhesion", J. Adhesion 4, 51
(1972). - -
6. R.J. Good, "Theory of "Cohesive" vs "Adhesive" Separation in
an Adhering System", !!..:. Adhesion~, 133 (1972) and "Locus
of Failure and Its Implications for Adhesion Measure-
ments" in "Adhesion Measurement of Thin Films, Thick
Films and Bulk Coatings", ASTM STP-640, K.L. Mittal, ed.,
American Society for Testing and Materials, Philadelphia,
( 1978).
7. D.H. Kaelble, "Peel Adhesion: Influence of Surface Energies
and Adhesive Rheology", J. Adhesion 1, 102 (1969).
8. M. Masuoka and K. Nakao, "Effect of Residual Stress on Tensile
Bond Strength for Steel-Nylon 12-Steel Cross Lap Joint",
J. Adhesion Soc., Japan, 12, 41 (1976).
9. A. Muller and R. Pluger, "Eigenschaften chemischer Aufbau und
kristallinitat von Polyamid-Kunststoffen", Kunststoffe
50, 203 (1960).
10. A.N. Gent and R.P. Petrich, "Adhesion of Viscoelastic Mate-
rials to Rigid Substrates", Proc. Roy. Soc., A, 310,
433( 1969).
11. K. Fukuzawa, "Studies on Rheology of Pressure Sensitive Adhe-
sive Tapes (V) Theoretical Equation for 180 0 Angle Peel
Force at Constant Peel Rate", !!..:. Adhesion Soc., Japan, .2"
301 (1969).
12. M. Masuoka and K. Nakao, "Temperature-Time Superposition and

Failure Envelope for Interfacial Fracture of Cross Lap
Joint", .!!..:.. Adhesion Soc., Japan, ~, 341 (1979).
13. T .L. Smith, "Ultimate Tensile Properties of Elastomers (I)
Characterization by a Time and Temperature Independent
Failure Envelope", .!!..:.. Polymer Sci., A, 1, 3597 (1963).
14. K.H. Illers, "Der Einfluss von Wasser auf die molekularen
Beweglichkeiten von Polyamiden", Makromol. Chem., 38, 168
(1960). - -
15. A.N. Gent, "Adhesion of Viscoelastic Materials to Rigid Sub-
strates, (II) Tensile Strength of Adhesive Joints", J.
Polymer Sci., A-2, 2, 238 (1971). -
16. K. Nakao, H. Endo, and M. Masuoka, "The Relation Between
Tensile Bond Strength and Crystalline Properties of the
Adhesive on the Steel-Nylon 12-Steel System", in "Mechan-
ical Behavior of Materials", Proceedings of the 1971
International Conference on Mechanical Behavior of Mate-
rials, Vol. III, The Society of Materials Science, Kyoto,
Japan, (1972).
17. H. Schonhorn and F. W. Ryan, "Effect of Morphology in the
Surface Region of Polymers on Adhesion and Adhesive Joint
Strength", .!!..:.. Polymer SCi., A-2, &, 231 (1968).
18. T.K. Kwei, H. Schonhorn, and H.L. Frisch, "Dynamic Mechanical
Properties of the Transcrystalline Regions in Two Poly-
mers", .!!..:..!E.Q.h Phys., 38, 2512 (1967).
19. H.W. Starkweather, J.R., G.E. Moore, J.E. Hansen, T.M. Roder,
and R.E. Brooks, "Effect of Crystallinity on the Proper-
ties of Nylons", .!!..:.. Polymer SCi.,D, 189 (1956).
20. Y. Kobatake and Y. Inoue, "Theory of Adhesive Joint",
Proceedings of the First Japan Congress on Testing Mate-
rials, 156 (1958).
21. M. Masuoka and K. Nakao, "Effect of Aspect Ratio on Tensile
Bond Strength for Butt Joint of Internal Fracture", J.
Adhesion Soc., Japan, ji, 125 (1978). -
Deformation and Shear Strength of FRP
Adhesive Joints
Hiroo Miyairi, Hideaki Fukuda and Atsuyoshi Muramatsu
Institute for Medical and Dental Engineering
Tokyo Medical and Dental Univ., 2-3-10, Surugadai,
Kanda, Chiyoda-Ku, Tokyo, Japan 101
ABSTRACT
The structural adhesives used in this study were a hard type

polyester, a soft type polyester and an epoxy polyamide. The FRP
adherend consisted of polyester matrix and reinforcement of glass
mat/glass roving cloth combinations. The adhesive strength was
obtained by using the tension shear test of single lap adhesi-..re
joints. The characteristics of adhesive strength were examined for
the change in lap length, and the results were used to explain the
deformation of adherend and the fracture pattern of layers. Fur-
thermore, the peeling behavior of adhesive layers, the bending
moment of FRP adherend and the influential factors on the adhesive
strength of FRP adhesive jOints were investigated.
INTRODUCTION
Recently, fiber reinforced plastics (FRP) are used for large-

sized structural construction. Therefore, the adhesive construc-
tion is applied for the many purposes of practical use (1). FRP
adhesive construction shows the complicated fracture pattern in
comparison to other materials. In concerning with these problems,
Hart-Smith (2) have reported the design matters of single, double
and scarf joint, etc., for practical use. But in his paper, the
adherend of lap joints was used for aircraft structural materials,
boron/ epoxy, glass/epoxy and aluminum, etc. And they were con-
cerned with the adhesive layer to be separated.
393
394 HIROO MIYAIRI ET AL
However, the strength design for FRP adhesive construction has

not been well established in engineering work. Under the present
conditions the problem of adhesion is attracted by the attention to
the development of new adhesives and their technology. For the
isotropic or anisotropic adherend of single-lap joints, Renton and
Vinson (3) studied the theoretical approaches to adhesive layers
which are influenced by properties of adherend materials. But in
their paper, the deformation of adherend parts and, for concrete
example, FRP constitution of glass/polyester was not examined and
discussed.
Therefore, aiming at the usefulness for the design of adhesive

joints, we have investigated the deformation of the glass/polyester
adherend used for single-lap joints and, in practice, used for long
lap length of adhesive jOints, and the displacement and stress
distribution of a near-by adhesive single lap joint.
And so, this paper treated the single-lap joint as a basic

adhesive joint. Consequently we are establishing the guides for the
improvement in adhesive construction, and its technology and the
adhesive strength for practical use, on the basis of the influences
of their deformation and strength (4-6). The adhesives used in this
paper are of three types: a hard type polyester, a soft type
polyester and an epoxy-polyamide. A hard type polyester is the
same as the matrix of FRP adherend. A soft type polyester is of the
new type developed for FRP construction in this study. An epoxy-
polyamide is widely used as the structural adhesive. FRP adherend
is consisted of the combination of the glass mat and roving cloth
reinforcements. The adhesive strength was obtained by using the
tension-shear test. The characteristics of adhesive strength were
examined for the change in lap length of joint, and the results were
used to explain the deformation of adherend, the dimensional influ-
ence of lap length and the fracture pattern (7). Furthermore, the
factors affecting the adhesive strength, such as the peeling from
the single-lap ends, the bending moment at adherend, etc., and the
stresses at the adhesive layer related to the deformation of
adherend were investigated (8-10).
SPECIMEN
Constitution of FRP Adherend and Adhesives (11)
FRP adherend was composed of three layer reinforcements, such

as glass-mat/glass-roving cloth/glass-mat, and polyester as matrix,
and was made by hand layout method. The material description is
shown in Table 1.
The adhesive, as shown in Table 2, is a hard and soft type
polyester and epoxy-polyamide.
FRP ADHESIVE JOINTS 395
Table 1. Constitution of Glass Fiber Reinforced Plastics
Glass fiber reinforced Matrix Glass content

materials (wt %)
Mat+Roving cloth+Mat Polyester resin 39.0

(MC600 +WR570 +MC600) (Ester R110A)
Table 2. Kinds of Adhesive and Treatment of these Specimens
Adhesives Treatment
Hard type polyester 11450 chop mat 1/2 mat in

(Ester R110A) Cured at room temperature
Soft type polyester Cured at room temperature

(Ester AD5000)
Epoxy-polyamide Cured at room temperature
(Cemedine 1500)
A hard type polyester is widely used as the so-called primary

adhesive to FRP constructions. A soft type polyester was developed
for the use of FRP adhesive structural construction as the secondary
adhesive. An epoxy-polyamide is generally used as the so-called
secondary adhesive.
Dimension
The lap length R, of the single-lap adhesive joint is 12.5,

25.0, 50.0, 100.0 and 200.0 mm, respectively. Here the lap length
of 12.5 mm is based upon JIS (12) and ASTM standards; the width of
adhesive joint b is all 15 mm.
The deformation of adhesive joint was measured with strain

gages (Shinko Co., B-FAE-B-12T23) placed on both surfaces of adher-
end. The strain gage location is shown in Fig. 1. The gage length
of all specimens is 2BOmm between both chucks, and both grip ends of
specimen were fitted by the supporting plate of 25 mm to longi-
tudinal direction because of the working of the pure shear stress at
joint.
396 HIROO MIYAIRI ET AL.
;;=::.0= 40 f=30~;~
~:L=lII
l = 12.5 25.0 50.0 200.0 mm
100.0
un it : mm
Strain gage position ( Gage length L=280 mm)
Fig. 1. Dimension and shape of adhesive joint.
Mean Shear Stress-Elongation Diagram
Figure 2 shows the mean shear stress-elongation diagrams (T-A

diagrams) for the change in adhesive and lap length, when the
crosshead speed of the tension-shear test was 1.5 mm/min.
The mean shear strength for each adhesive was different, and
also the fractural elongation AB for each adhesive was different.
Therefore, the value of ~ of A-adhesive was relatively small with
increasing lap length, while those of Band C-adhesives were
relatively large. The apparent rigidity of joint is not explained
from the value of l:J:r/ t:.A because T decreases as lap length increases
as shown in Fig. 2. But considering that the gage length L is a
constant value of 280 mm the large value of t:.T/t:.A, within the limits
of small lap length, shows the elongation of R. to occupy a large
percentage in comparison with the elongation of adherend, of total
elongation of adhesive joints. And without regard to R., the
elongation at fracture of adhesive joints, As, decreased from
epoxy-polyamide, to soft and hard polyester adhesives in good
order. Especially in the hard polyester the elongation at fracture
AB was small in spite of the increase of lap length, and its mode of
fracture was brittle differing from the modes of others.
Figure 3 shows the mean shear strength T or PB/b being the

fracture load per unit width versus lap length~.
Generally, TB decreases with increasing 1 and in contrast to

this, PB/b increases, but in the case of GRP adherend, the increase
of PBib is small in comparison with other materials. It is
considered that the fracture of jOint occurred at the matlroving
cloth interface of adherend because of the stress concentration at
joint ends. For example, especially, the specimen with C-adhesive
showed such a fracture. These results show that the fracture of
0.9
0.8
" Hard Iype poIyesler adhesive
o 5011 Iypo polyesl., adhesive
0.7 " Epoxy- polyomid adhesive
9890 I.ngth L . 280mm
0.6
-E O.S
E
......
at
0 ..4
...
......
:II:
0.3
t -1010MM
0.2
0.1
1.0 2.0 3.0 4.0 5.0

,\ (mm)
Fig. 2. Shear stress-deflection diagram of some

adhesive jOints.
1.0 Epoxy-polyamide adhesive 40

_-a
__ - -- _4.
.,.--- ,-"",--
--
,
0.8 ,../
-'
-'
-' .-r- 30
-'
-'
-' E
"" E
.......
an
20 -'"
Soft type
[0
0.2
Hard type polyester
() LO-'-......I....-......I.---- . . . l - . . - - -- -----'....J 0
25 50 100 200
l ( mm )
Fig . 3. 1B' PB/b - R. diagram of some adhesive joints.

(TB:O~O, PB/b: .,~
joint is not caused only by the characteristics of adhesive but also

by the characteristics of FRP adherend.
Classification of Frantured Bonding Surface and Fracture Pattern

of Adhesive Joint
Figure 4 shows the five fracture modes of adhesive joint for
change in adhesive and lap length. By using five symbols such as
(a), (b), (c), (d) and (e) showing in Fig. 4, the fracture mode of
each specimen was classified as shown in Table 3, where the numeri-
cal number before each symbol is the number of fractured specimen,
and symbols a, b, c, •.• are of fracture mdoe showing in Fig. 4.
As a result, in the cases of the joint with A-adhesive of low

strength or the joint of short lap length, these fracture modes were
displayed by the symbols (a), (b) and (c), in the consideration of
the fracture all over the adhesive layer. Such a result is
explained by the following matters. The shear stress of the joint
with short lap length tends to distribute uniformly and the fracture
load of the joint with A-adhesive was relatively small. Therefore,
it is considered that, as these joints failed slowly through peeling
from the ends of lap length, the fracture of FRP laminates was not
observed. In the case of fracture mode (d) which the joint with C-
adhesive showed, the fracture occurred at the glass-mat/roving
cloth interface, for which the serious problem is the interfacial
strength of adherend rather than the characteristics of adhesive.
Glass .at
Glass roving cloth
(a ) ( b ) (e) (d) (e)
Fig. 4. Fracture pattern of adhesive joints.

Table 3. Fracture Mode of Adhesive Joint

L
\ 12.5 25.0 50.0 100.0 200.0
adhesive'
A 5a 4a,c 5b 4b,c 5b
B 5a 5b 4b,c 4b,c b,4c
C b,4c 5b 4d,e 4d,e 4d,e
Therefore, to use the expensive, high strength adhesive such

as an epoxy-polyamide to GRP adherend is not as efficient as other
adhesives in improving lap lengths.
Adhesive Joint Rigidity and Apparent Elastic Modulus of Adherend
The adhesive joint part has the discontinuous stress distribu-

tion and the variation of rigidity in comparison with the flat
adherend part. Therefore, in practical adhesive construction,
there are many cases of the joint fracture resulting in the discon-
tinuity of the joint deformation.
This tendency is especially seen on the flexible structure
like GRP, and then the influence of lap length is an important
problem to be associated with the joint rigidity. Here, this
problem was investigated by analyzing the apparent elastic modulus
of adherend, E'. Firstly, we divided the elongation of specimen, A,
into two parts: one with lap length ~ and the other part L-~ (where
L: gage length of 280 mm) when ignoring the shear deformation of
adhesive layer. The equation is as follows:
'(~ '(~ '( ~
A = 2E't + E't (L-~) = 2E't (2L-~)
2E't (1)
'( = (2L-~)
Where '(: shear stress

t: thickness of adherend
Hence E' is obtained by
E' = (2L-~)~ ~'( (2)
2t ~A
In Eq. (2) the shear deformation of adhesive layer and the

flexural deformation of adherend are discounted; the influence of
adhesive and the flexural load are examined.
Figure 5 shows E' versus lap length for some adhesives. In the
case of A-adhesive, E' increases with increasing lap length, but in
the case of Band C-adhesives it decreases.
~o, comparing with the elastic modulus of GRP adperend (E=1120

Kg/mm ), the values of E' obtained were small, especially in the
case of A-adhesive, E' was influenced by lap length. Therefore, in
the case of A-adhesive, the influence of shear deformation and
flexural load gradually decreased as the lap length increased. And
the decrease in the apparent elasticity E', with an increasing lap
length, was accompanied by the decrease in rigidity of joints
because of the joint's deformation. Therefore the increase of E'
indicates that rigidity of jOints according to large lap length
starts to increase and it is expected that these influences give
rise to the increase in the flexural rigidity.
Similarly, the apparent tension rigidity n of the adhesive

jOint is given as follows:
&/b 6P
n = t;S: = MA
where !PI/).A is the initial slope in P/b - A diagram as shown in Fig.
6. Comparing the initial slope in PIb - A diagram with that in
T - A diagram, the value of /). PIb /).A is a small change with the
variation in lap length t. And the value of /).P/b/).A increases as the
value of t increases.
o -Had 1ype polyes1er adhesive

• · ···Sof1 1ype polyester adhe5lVe
8001-+--1--1-
---- .
o _·-Epoxy·polyamld adhesive
()
'---
700 -.~--::-----I---------~
--~-I
() () • ()0 ...:.-~
__...- ..............
. -........ .......
.~
50 100 200
l lmm !
Fig. 5. E' - t diagram of some adhesive joints.

35 0 - Hard type polyester edheaive

t;. --- Soft type polyester edhesive
0 - - Epoxy - polyamid adhesive
30
25
E
E
~20
~
.0
~ 15
_ 21S.7mm
10
1.0 o 5.0
A(mm)
Fig. 6. P/b - A diagram of some adhesive joints.
By using Eq. (3), the relation between nand t is shown in Fig.

1. It is proposed that n is generally expressed with the increasing
function for t, but those behaviors were changed by adhesive, as
shown in Fig. 1. The n of A-adhesive increased in comparison with
other adhesives which showed respectively a similar pattern_ In the
case of Band C-adhesives, the increase in apparent tension rigidity
was small. It is considered that the discontinuity of rigidity was
weakened by the shear deformation of lap length. Therefore, the
rigidity of adhesive joint depends on the characteristics of adhe-
sive, i t is the same effect influenced especially by FRP large
structural construction.
Bending Moment Distribution of FRP Adherend
The test of single lap jOint gives erroneous results to the

occurrence of flexural load at adhesive position and adherend. In
this paper, the flexural influences were investigated by measuring
the strain at two sides of the adhesive joint. The different strain
occurred at two sides of adherend results in the combination of
12
1
10 .~---__vI
~
.
;,- ......
,-
.,_..... ...---- .--.--'
11'- ~
.~--/-o
8
6
I~
~o I I· 0 Hard type pOlyester adhesive
S,h ' ' ' '
~ Epoxy polyamid
pol,."..adhesive
odh"iw ~
4 I
12.525 50
I
100 ~oo
(mm)
Fig . 7. n -R. diagram of some adhesive joints.
tensile and flexural loads. Therefore, the standing € and € for

t m
tensile and flexural strains, or the strains at two sides of
adherend, respectively €1 and ~, are as follows:
(4)
And dividing the axial load acting on the adherend into tensile
and flexural loads, F and M, these equations are given by using €t'
€ and the elastic modulus of adherend E .
m
F = E €t b t
1/2 2E € b 2 E € b t2
M=2 fo { t m } y dy ={ m
6
} (5)
Here y is the distance from the neutral axis of the adherend.

Finally, after eliminating €t and €m in Eq. (5) by using Eq. (4),
Eq. (5) becomes:
F _ E b t
- 2
(6)
E b t
M = -'-2- (€, - €2)
Fig. 8 shows the normal and flexural forces, F and M, which were
obtained from Eq. (6), when treating the case of a soft type
polyester adhesive.
The bending moment near the joint increased with increasing

load, but decreased as the debonding at adhesive joint progressed.
And the maximum value of the bending moment increased with increas-
ing lap length. The bending moment near the supporting point showed
the same sign as that near adhesive joint in case of early loading
and was relatively small, but its bending moment reversed the sign
regardless of the lap length as load increased. Bending moment near
the clumping parts tended to decrease along the adherend in compar-
ison with that near the adhesive jOint. Bending moment near the
center of adherend tended to be influenced by lap length, and at its
location the normal load increased. Thus, the bending moment near
the supporting point and the adhesive joint changed its sign;
therefore, this tendency was shown to be especially at the center of
adherend.
50
-0- -0- -0- -0- -0- --0- --0- _

40 -if -<:r -0- - <l..
., '-0 7
,0'
Soft type polyester adhesive
;f'"
1 = 12.5 mm 1 2 3
30
!
/
P ..0- -0- -0-- '0...

I I
: i I
l = 25.0m m 4
I 9
I 5 6
6,I Jf' : ! ; : 9
'0--.--0... \
!
20 --0...·-04
E 1 = 50.0mm
E
tID
..l<:
If 7
i,40 140 i 40 I
8 9
9
10 "
~
- 10
- 20
Fig. 8. M - F diagram of soft type polyester

adhesive jOint.
Figure 9 shows the bending moment and normal force along the
longitudinal direction of joint. The normal force of the adherend
increased with increasing lap length, but in the case of short lap
length the curve tends to differ from others. And the normal force
F occurred at adherend is smaller than the bending moment Pt (where
t is thickness of adherend) as happened from the gap of both
supports, and these values did not change with the location of
adherend except on 1=12.5 Mm.
Figure 10 shows the bending moment of adherend which changed
with increasing load. Horizontal axis X, which started from the
center of lap length, denotes the distance along the longitudinal
direction of specimen. In the case of 1=12.5 mm, the bending moment
reversed the sign in the region of X=40-80 mm and approached to zero
near the supports. In the case of 1=50 mm, the bending moment
reversed the sign in the region of X=80 -120 mm; therefore in the
case of single lap jOints, the bending moment at adherend near
adhesion with constant gage length increased as lap length in-
creased and the sigrt of bending moment was reversed near the
supports, and then the single lap joint was deformed to the shape of
the capital letters. But in the case of 1:12.5 mm according to JIS
and ASTM standards, the bending moment was reversed in the region of
X=4cr 80 mm and X=120 mm, and the deformation pattern became very
complicated in this case.
1.2 ~~-----.-----~--1
..
7()-·_·_ · _· ~· -· - · -·~69
11. 5 -----
~10 4~""'::;p-_""~~I'-- "'----0,
0.8
r
02
___
4 - ____
~. -. ~~--~
i :t. _
7\1 M/ Pt 69
- 0.2 40 80 120
x (mml
Fig. 9. M/Pt - X Diagram of soft type polyester

adhesive joint.
40
_L-__
---- 1 = 12.5mm
=
- - - I 25.0 mm
- 1= 50.0mm
30
20
..
E
E
~
~ 10
0
,
o P = 50 kg
• P =HlOkg
CIP=I50kg __
- \0
o P=200kg
6 P =250kg
X P = 300 k~
-20 -+'- -
o 40
x ( mm )
Fig. 10. M - X diagram of soft type polyester

adhesive joint.
Influence of the Scarf Angle of Adhesive Joint
In the single lap joint the stress concentration of adhesive

would be reduced to some degree by providing a scarf angle to the
joint. Therefore, in this study we examined the mechanical effects
of the epoxy-polyamide adhesive joint with a scarf angle e = 35°.
Figure 11 shows the apparent shear stress - elongation diagram

(, -X diagram) with the change in lap length.
The rigidity of the scarf joint is low in comparison with that
of the foregoing single lap joint, and particularly in the small
short lap length the fracture elongation AB became scarf joint. And
the scarf joint was known to be that the deformation of its adherend
was very much followed by the applied load to fracture.
Figure 12 shows the apparent shear strength 'B or the adhesive

strength by unit width, PB/b, in the change in lap length, and also
the results of a single lap joint. However, the difference of both
strength properties was not recognized.
0.9
Epoxy-polyamid adhesive
0.8 o
0.7
.----
... 1= ,5. 2mm
/'
0.6 //
f ::: 25.1mm
'E
E O.S
....
CI
.:tI.
...
~0..4
------
1= ' 009~1 ='O' 8mm
0.3
------ 1=20J2mm
0.2
I = ZQ l omm
0.1
0
0 10 20 30 40 50
.\ (mml
Fig. 11. L - A diagram of epoxy polyamide adhe-

sive joint.
E
....E." ~E
.JI.
....E." Epoxy-polyamid adhesive
.. .= Ta{:~: } PYb
.D .JI.
........
~
40 1.0 ~
0.8
a -====-~
0"-------e
30
0.6
20
0.4
10
0.2
o o~~~----~--------~
12.5 25 50 100 200
L (mm)
Fig. 12. lR,PB/b - t diagram of epoxy polyamide

aahesive joint.
Figure 13 shows the distribution of bending moment of the scarf

jOint adherend in much the same way as the case of single lap joint.
The distribution of the bending moment of both joints showed nearly a
similar tendency, but these values of scarf joints in the center of
lap length are larger than that of foregoing single lap joint.
These data show that the decrease of rigidity of joint and of the
flexural load apportionment at the center of adhesive layer were
resulted by providing a scarf angle of joint. And in the large lap
length (t = 50 mm), the decrease of maximum bending moment at about
X = 40 mm was not shown; therefore, the scarf angle provided to
joint ends is not effective for the decrease of bending moment. But
the slight increase of fracture strength is caused by the average of
the distribution of bending moment.
Figure 14 shows the relation between the bending moment M and

normal force F with an increasing load in the change of lap length.
In the case of the scarf Joint dissimilar to foregoing single
lap joint, its bending moment M does not become M<O except the
positions of #5 and 118. Therefore, it is characterized that the
50 Epo;ll;V - polyam id adhesive
---- ! = 125
- -- 1=25
40 1--1--- - - 1 =50
o 50 kg
. ,00kg
C 150 kg
30 0200 kg
l> 250 kg
~ 300 kg
20
:I 0 I-t---+-
-10 r---'-- -
- 20
o
x (mm)
Fig. 13. M - X diagram of epoxy polyamide adhe-
sive joint.
50
40
30
20
E
E
Cl
.:>!
~
0
- 10
- 20
Fig. 14. M - F diagram of epoxy polyamide adhe-

sive joint.
bending moment at the adherend of adhesive joints with each lap

length becomes M>O.
Fracture Pattern and Lap Length of Adhesive Joint

Fracture pattern of adhesive joint is shown by the group of
Fig. 4 and the results of Table 3. And the state of fracture is
expressed by the initiation of the end of lap length to be caused by
stress concentration in spite of the increase of lap length. The
fracture mode (d) which epoxy-polyamide adhesive joint showed is
found as the mat/roving interfacial debonding, then the roving
layer is exposed at the fracture surface. But in the case of the
soft type polyester simiiar to the properties of the former adhe-
si ve, its fracture occurred at the adhesi ve interface and its
debonding surface did not progress to roving layer.
Therefore, it is considered that the soft type polyester

exhibited maximum adhesive properties for the glass fiber rein-
forced plastics (GRP) adherend. In the scarf joint of epoxy-
polyamide adhesive, the apparent shear strength 1R showed slightly
larger value than that of usual adhesive, but both ~racture patterns
were not different. This result was caused by the large scarf angle
e = 35 0 • There fore, the fracture of adhesive· joint started from the
lap length ends as the debonding and then progressed to the center
of adhesive layer as load increased.
These states of fracture can be prevented with the decrease of
stress concentration and bending moment of the adherend, but in this
experiment these effectiveness were not revealed. So the scaM'
angle a takes to be small and then the mechanical effects of
adhesive joint will be improved. And if the good epoxy-polyamide
adhesive was used for GRP adherend, we could not expect more than
the adhesive properties because of the debonding fracture of adher-
end. And so it is important to examine the decrease of the stress
concentration of the adhesive layer and the prevention of interface
fracture of GRP adherend.
Therefore, it is considered that the possibility of the pre-

vention of the joint fracture is to examine these fracture patterns
at the lap length ends. And in the adhesive design of FRP i t is
pointed out that we must consider the adhesive work to be a part of
engineering, including the decrease of stress concentration of the
adhesive layer, and not only the improvement of the material for
structural adhesive.
CONCLUSIONS
On the adhesive strength of glass-fiber reinforced plastics

(GRP), the characteristics of tension-shear strength of single lap
joint widely used for the secondary adhesive structure were ex-
amined practically in the consideration of lap length of the region
in 12.5-200.0 mm. The results are summarized as follows:
( 1 ) In the single lap joint of GRP adherend, the adhesive

strength P /b (fracture load per unit width) did not increase in
proportionElo lap length, and even in the case of an epoxy-polyamide
adhesi ve its strength was about 35.5 Kg/mm. Such results were
caused by the fracture of the mat/roving cloth interface of GRP
adherend. Therefore, even if a soft type polyester adhesive is used
we will be able to show the same adhesive strength as in the case of
an epoxy polyamide adhesive.
(2) Furthermore, such fracture is discussed reasonably in

term of the stress concentration which increasingly occurs at the
end of adhesive layer as the length increases, and the stress
concentration is influenced by the bending moment occurred at the
adherend. Therefore, in the case of such an adhesive joint as
fractured at the mat/roving cloth interface of FRP adherend, the
examination according to the remaking of jOint in consideration of
the relaxation of stress concentration and the adhesive technology
is required for the adhesive structure.
(3) The apparent elastic modulus E' of GRP adherend decreased

because of the bending moment and shear strain of the adhesive
layer, when changed with lap length and adhesive. In the case of a
hard type polyester, the value of E' increased as lap length
increased, but in the cases of a soft type polyester and an epoxy-
polyamide adhesive, it decreased. Consequently, the value of E' was
in the region of about 1/2-213 of the actual elastic modulus for
adherend.
(4) The reduction of stress concentration of adhesive layer

ends with the scarf angleo shows a little increased properties in the
case of scarf angle e=35 •
These situations are found in the fracture pattern and the

deformation of adherend. Therefore making the scarf angle to
decrease, we would expect the improvement in the adhesive shear
strength.
(5) Fracture modes of GRP joints were classified into five

types depending on the adhesives and lap length. In the case of
adhesive joint of short lap length showing low strength, the
fracture occurred at adhesive layer, and in the case of adhesive
joint of long lap length showing high strength it occurred at the
interface of GRP adherend.
(6) In the case of adhesive joint according to tension-shear

test with constant gage length, the bending moment occurred at the
adherend depended on the lap length; therefore in the case of lap
length 12.5 mm, it reversed the sign near the adhesive and support-
ing positions, but in the cases of lap lengths 25.0 mm and 50.0 mm
it showed a maximum value near the adhesive position and decreased
until reaching a minimum value near the supporting position.
REFERENCES
1. T. Hayashi, Fukugo Zairyo Kogaku (in Japanese), 840 (1971).
2. L.J. Hart-Smith, NASA Contractor Report, NASA CR-2218 (1974).
3. W.J. Renton and J.R. Vinson, J. Adhesion 7, 175 (1975).
4. C. Raphael, ~ Polym. Sym. l, 99 (1966).
5. H. Miyairi, Nippon Setchaku Kyokai-Shi (in Japanese) I, 4, 217
( 1971).
6. Fukugo Zairyo Gijutsu Shusei Henshu Iinkai, Fukugo Zairyo
Gijutsu Shusei (in Japanese), 690 (1976).
7. FRP Adhesion Committee, Reinforced Plastics (in Japanese) 14,
6, 313 (1968). -
8. L.M. Keer, ~ ~ Mech. ,2, 697 (1974).
9. R.D. Adams and N.A. Peppiatt, J. Strain Anal. 9, 3, 185 (1974).
10. F. Erdogan and M. Ratwant, ~ Compo Mat~~ 378 (1971).
11. H. Miyairi, H. Fukuda and A. Muramatsu, Zairyo (in Japanese)

26, 282, 649 (1977).
12. Japanese Industrial Standards K 6850 (1972).
Stress Distribution and Strength in Shear
on the Adhesive Lap Joint Loading
Bending Moment
Yukisaburo Yamaguchi and Susumu Amano

Kogakuin University
Nishishinjuku 1-24-2, Tokyo, Japan 160
ABSTRACT
The stress distribution and the maximum stress in shear at the
circular area of adhesive lap joint were discussed theoretically
and experimentally in this article. The joint carries the bending
moment due to external force P which is parallel to lap surface and
at a distance R. from the center of adhesive area. The following
equation with respect to the maximum shear stress T' was derived
and verified experimentally. max
P 2R.
T'max = -2
1TR
(- a
R
+ En
where R is radius of adhesive circular area; a and S are stress

concentration factors and are 1.2 and 2, respectively.
INTRODUCTION
There are two cases in which the bending moment acts on the
adhesive lap joint; in one case, tensile and compressive stresses
arise mainly from carrying the bending force perpendicular to the
adhesi ve surface (1,2,3) , and in the other, shear stress arises
mainly from carrying the bending force parallel to the adhesive
surface (4). The stress distribution and the maximum shear stress
or the shear strength of adhesive lap joint in the latter case (Fig.
1) are treated theoretically and experimentally in this article.
413
414 YUKISABURO YAMAGUCHI AND SUSUMU AMANO
f----.l-----l
d Y p
x
ar-~-;._-----~
Fig. 1. Adhesive Lap Joint Loading Bending

Moment pg,
INDUCTION OF THEORETICAL FORMULA
Preliminary Experiment
In Fig. 1, the bending force P that is at a distance g, from the

center 0 of the adhesive surface area and parallel to adhesive
surface was applied to a lap joint with color paste between two
transparent plates, and then it was found that the locus of relative
movement of every adhesive point was circular around the center 0 of
the adhesive area as shown in Fig. 2. Accordingly, we may assume
that the relative movement or shear strain in the adhesive layer is
directly proportional to the distance from the center 0, provided
that the adherends are much higher in elastic modulus than that of
adhesive. According to the total mechanical equilibrium system in
Fig. 1, bending moment P g, should be balanced to both actions a
couple P-P' and a compressive force pIt to the center O. Therefore,
the balance actions to bending moment pg, in the adhesive lap jOint
"may be divided into two reactions : the total shearing couples
around the center 0 in the adhesive area and the shear stresses
opposite to P".
Theoretical Formula
We assume that the adhesive area of lap jOint is circular with
radius R and similar to that of quadrangular abcd to make analysis
simple as shown in Fig. 3, though the adhesive surface area is
generally quadrangular. We may obtain the following relation from
the above assumptions and the preliminary experiment:
r
T = R To (1)
where T is shear adhesive stress at any point being r distant from

OJ R is the outside radius; T is the shear adhesive stress at the
outside of the circular adhes~ve area.
ADHESIVE LAP JOINT 415
Fig. 2. Photograph Showing the Locus of Movement

of Adhesive J)ints .
y
1--- - l -------
Fig. 3. Stress Analyzing Diagram in Adhesive Lap

Joint Due to Bending Moment.
The elementary moment dM of the circular elementary area dA

being at a distance r from the center 0 and dr in width to be
balanced to the corresponding couple p1 is presented in the follow-
ing equation
dM = 'r-dA'r
= T·21fr·dr·r
2
= 21fr ·T·dr (2 )
From Eqs. (1) and (2)

2 r
dM = 21fr R TO ·dr
r=R
M = p1 = f dM
r=O
r=R T
=f 21fr 3 ~ dr
r=O
(4)
or
2 p1
'to = 1f R
3 (5)
The distribution of shear stress T and the maximum shear stress

To due to the coupling action are also shown by the diagram in Fig.
3.
The distribution of the shear adhesive stress t on the line gh
due to compressive force P" parallel to yy as an example is shown by
the triangle gg'oand ohi in Fig. 3. Its maximum value i, is as
follows
P"
T1 =A B (6)
where A is the adhesive area, 1fR2 , and B is the stress concentration

factor concerning i 1 • The mean adhesive stress f JII due _to pn is to
be P" lA, so that the value of B being the ratio of T 1 to Tm should be
2, if the adhesive shear stress due to pn i is directly proportional
to the distance from the edge of the circle to the loading line XX
as shown in Fig. 3.
Accordingly, the total shear stress '("" at any point on the

adhesi ve layer must be the vector sum H+-) of vector '( and :r,
so
that the maximum total shear stress '("" may be obtained from Eqs.
max
(5) and (6) as follows:
_~-1-LQ (1)
- 3 "'T"'" 2 j.J
'If R 'lfR
It is clear in Fig. 3 that '("" arises at the point on the

line XX as follows: max
2 P1 P
'[~x = 'If R3 + 'lfR2 a
21
[R a + a] (8)
where a is the stress concentration factor concerning '[ 0 and

theoretically equal to about 1.
EXPERIMENTAL DISCUSSIONS
On The Stress Distribution
The distribution of the total adhesive shear stress '[~ '[ + t,

arose by the bending moment on the adhesive area, may be discussed
experimentally by the test using five strain gauges, 1-5, on the
line XX of the steel adherend surface, as an example, under various
bending forces P as shown in Fig. 4. Compressive strain e' and
tensile strain e" at every strain gauge under various bending forces
P are shown in Fig. 5. It is clear in Fig. 5 that compressive or
tensile strain on the surface is directly proportional to the
bending force and to the distance from the center o. And it is also
evident that the strain e' at the right side from the center 0 is
larger than e" of the left side, because the direction for the right
side is similar to that of '[; however that for the left side is
opposite to it. As the shear strain of adhesive layer may be
directly proportional to the compressive or tensile strain of
adherend, it may be recognized that the distribution of shear stress
shown in Fig. 5 corresponds to that in Fig. 3 or Eq. (1).
Total Maximum Shear Stress and the Shear Strength of the Joint
The shear strength of the double lap joint bonded wi th an

epoxy-poly-amide adhesive (BOND E and 0.05 mm in thickness) between
Fig. 4. Positions of Strain Gauges in Stress

Distribution Measurement.
0.1
'1'0
0
x
x
~·o.l
-0.2
- 03
•,
D
l>.
-0.4 0
-0.5
Fig. 5. Compressive and Tensile Strain on Adher-

end Surface at the Position of Every
Strain Gauge.
steel plates is about equal to the shear strength of the adhesive

itself L. The morphology of fracture surface of the adhesive
bonding ~hows the cohesive break at the adhesive layer (5). And
then, i f the value of L'" calculated from Eq. (8) in which the
max
values of P, ~, R, ~ and 8 were applied for the condition by which
the cohesive break of the adhesive bond due to bending moment P~
happened is equal to the strength La of the adhesive itself,
Equation (8) may be said to be reasonable.
Figure 6 shows the testing apparatus by which the actual shear

strength of the adhesive lap joint was compared with the value of
'['" in Eq. (8).
max
Figure 1 shows the values of '['" calculated from Eq. (8)

max
applying the actual bending moment Pma)!; 11 for each five test pieces.
In this experiment, R was 20 mm; 11 325 or 225 mm; P the maximum
bending force to break the adhesive joint; (l 1 or 1."~nd S 2. In
Fig. 1, A and A' show the values in the cases of 325 mm of 11 to be 1
or 1.2 of (l, and Band B' show those 225 mm of 1, to be 1 or '.2 of (l,
respectively. Line C shows the shear strength '[a of the adhesive
itslef which was measured by using the method of ASTM D132.
We may recognize that the maximum total shear stress '[~ax at

the breaking condition or the shear strength of lap joint due to
bending moment is about equal to the shear strength '[ of the
adhesi ve itself, and becomes more closely to '[ in the cfase of (l
being 1.2 than that being 1.0, as shown in Fig. a.,.
CONCLUSIONS
The interrelations among the maximum shear stress '['" arising

max
in the adhesive lap joint, the bending moment p1 loading parallel to
the adhesive surface, the radius R of the circular adhesive area and
the stress concentration factors (l and S were derived as follows,
[21 Sl (8)
max =
'['" (l +
R
It is verified from two kinds of experiments, the stress

distribution measurement and the breaking test of the adhesive lap
1----- - - - /"
clhesive ClI'('d
Fig. 6. Breaking Test Apparatus of Adhesive Lap

Joint.
5~------------------~
® d.V<1f"d8e v41ue ric('l")

---- --§- - --- -f --t
i fA B' faC
SA
2 ,.., 1
1 183
~ 225
n~ 12 183
110
~
1
225 1. 2 110
4cne ve S
Fig. 7. Actual Shear Adhesive Strength Due to

Bending Moment Under Various Conditions
and Shear Strength of Adhesive Itself.
joint, that the distribution of shear stress at the adhesive layer

was similar to the assumption by which Eq. (8) was derived . The
actual shear strength of adhesive lap joint also corresponded
closely to the relation presented by Eq. (8) .
REFERENCES
1. MIL- A - 5090B, "Standard Temperature Bend".

2. ASTM D 1184-69, "Strength of Adhesive on Flexural Loading" .
3. Y. Yamaguchi and S. Amano, Preprint of 14th Symposium on Adh.
of Japan, "Testing Method of Tensile Adhesive Bonding
Strength", 61 (1976).
4. L. R. Lunsford, Applied Polymer Symposia No . 3, Structural
Adhesive Bonding, "Stress Analysis of BondE!d Joint",
Interscience (1966).
5. Y. Yamaguchi, S. Amano and K. Furukawa, Prept'int of 15th
Symposium on Adh. of Japan, "Testing method of the Adhe-
si ve Streng th in Shear", 69 (1977).
Fracture Criteria on Peeling
Takashi Igarashi
Department of Technology, Gunma University
Kiryu, Japan
ABSTRACT
The mechanical conditions for the initiation and the propaga-

tion of fracture are considered in the case of trousers type of a
specimen consisting of two flexible canvas bonded with rubbery
adhesive. To reduce the difficulties in mathematical analysis a
mechanical model for peeling is adopted. It was found that the
condition for the initiation of peeling obeys the maximum stress
criterion and that for the propagation of steady peeling obeys the
minimum free energy criterion in thermodynamics. Here discussed
are limits of validity of thermodynamics for the problem of frac-
ture.
INTRODUCTION
When two canvas bonded with a rubbery adhesive are peeled off,
trousers type peeling as shown in Fig. 1 occurs. This peeling has
such an interesting character that there exists steady fracture:
that is, the ultimate state of break is being realized continuously
in this peeling. An external force required for its continuation
remains constant in this steady peeling process. Since the fracture
takes place continuously, measurement of the peeling force implies
carrying out continuous and infinite measurement.
On the other hand, the initiation of peeling or tearing in a

flawless sheet differs from the above mentioned steadily prop-
agating fracture. Peeling or tearing starts immediately after the
external force reaches some definite value. This feature, there-
fore, is thought to belong to an "ordinary" fracture like a tensile
421
422 TAKAHASHI IGARASHI
f
f
Adherend (canvas)
Adhesive layer
~
f
Fig. 1. Trousers type peeling.
fracture of a wire. How does the "ordinary" fracture occur? When

tensile load is applied on a thin and long bar, tensile stress
occurs in it, and the breaking of the specimen takes place when the
stress reaches some definite value. It is well known that in some
materials yielding phenomenon is observed at a certain state of
stress before breaking of the material showing "necking" in its
appearance. The necking extends along its longitudinal direction
with the increase in elongation of the specimen, then the stress in
the necking part increases again to the definite level that leads to
breaking of the specimen. Here is a series of a process of
"ordinary" fracture without presenting steady feature from loading
to breaking. Then in the case of peeling or tearing, it may be
thought that the initiation and propagation of it belongs to
different kinds of fracture and that each criterion of the fracture
obeys different mechanical laws. It is a well known experimental
fact that the force required to start tearing or peeling is much
different from that for its propagation.
When we discuss the fracture phenomena theoretically with

regard to the above facts, it may be important to distinguish the
problem of the initiation of the fracture of the adhesive that
completes after a series of successive stages from that on the
steadily continued fracture. As papers that discussed tearing or
peeling made no distinctions between the above two types of fracture
FRACTURE CRITERIA ON PEELING 423
from this standpoint; some papers discussed either the problem of

the initiation of fracture or that of steady fracture and others
discussed both problems inconsistently (1).
A mechanical theory on the fracture of solids was proposed by
Griffith (2) in 1920, on which a number of studies have been
successfully developed. It is then interesting to consider what the
Griffith theory does or does not teach us about these problems of
two types of fracture. One of the purposes of this article is to
make the structure of the Griffith theory clear with applying it to
the fracture of rubbery materials, to show its limit of validity and
to find out the possibilities of its further development.
Griffith, who has studied effects of a surface finishing of a

metal on its mechanical strength, found that the decrease in
material strength caused from existence of cracks was much greater
than the value expected from the calculated stress value based on
the elastic theory. This meant that the generally accepted hypo-
thesis at that time, i.e., fracture takes place when the maximum
stress in materials reaches a definite value, was not true.
Then, Griffith proposed a new version that the condition for

the fracture might be derived from minimizing the thermodynamic
potential (free energy) of the system. Thermodynamic potential of a
system consists of the free energy of the elastic deformation of a
solid caused by the external force and the surface free energy of
cracks existing in a solid. When work is done by the external force
on the solid, there occurs isothermal, reversible, elastic deforma-
tion of the solid storing the potential energy as free energy. In
the case that the external force exceeds a definite level, it leads
to lowering the increase in the total free energy to transform the
work done into surface free energy with opening the cracks rather
than to transform it into free energy of further deformation. This
condition is analogical to the gas-liquid equilibrium where evapor-
ation leads to lowering the increase in the total free energy at the
boiling point.
Two facts can be pointed out concerning the Griffith theory.

First, the theory is for the case that there exist cracks in the
solid from the beginning; second, it is the principle of minimum
free energy, which is expressed
(1)
where of is the variation of the total free energy; OFe is that of

the free energy of the deformation and OF s is that of the surface
free energy.
Thus, Griffith discussed tensile strength of a plate which was

in a two-dimensional plane stress state (Fig. 2). In this case, the
variation of free energy of the elastic deformation for change in
the crack length had been obtained from the stress analysis by
Inglis (3) as
of
e
=-(21TcfLoL)/E (2)
where 0 is tensile stress, E is Young's modulus. On the other hand,

the corresponding variation of surface energy is given as
of =4y&.
s
where y is the surface free energy. Then, from Eqs. (1), (2) and
(3), one obtains a well known result for fracture stress 0b
°b = (2EY/1TL)i (4)
It is noted that the free energy F has to be given in analytic

form. Significance should be emph~sized on the fact that with
introduction of thermodynamics into Inglis' stress analysis the
condition of fracture was given not as a consequence of a hypothesis
but as that of laws of nature; here exists the merit of Griffith
theory. Thus, the ultimate property of fracture strength could be
expressed in terms of physical quantities of the substance, i.e., so-
called material constants, such as elastic moduli and surface
energy. However, from the facts that the Griffith theory is the
equilibrium theory of cracks and that Eq. (4) includes length of the
(f
i i t
er·
y
Fig. 2. Griffith crack.

crack L, the problem of the strength of materials had to deal with

the substantial problem of the crack. In addition, the problem of
estimation of the surface free energy of solids is of importance
both from the viewpoint of strength of materials and as a link
between molecular theory and the Griffith theory.
Griffith's theory, i.e., principle of minimum free energy, was

extended to the studies on materials with large deformation by
Rivlin et al. (4) who discussed tearing of rubbery materials. On
the other hand, Hata (5) in Japan made a study on peeling a few
years earlier than Rivlin et al. For the peeling of a flexible and
inextensible tape, he proposed a simple and elegant relation,
f = W/(1-cos e) (5 )
where f is the peeling force per unit width of the tape; W is the
work of adhesion, and e is the peeling angle (Fig. 3). Though he
did not refer to the Griffith theory or thermodynamics, his study
included an idea of energy balance thought to be equivalent to
Griffi th theory. Here, the work of adhesion W had the same
dimension (force/length = energy/area) as that of the surface
energy and evalulated to the order of 10 3 erg/cm2 for vinylite(PE-
PVAc copolymer)-glass interface (6).
Since the Griffith idea came to be applied to the problems of

tearing or peeling, problems on the influence of a cut made in a
rubbery sheet on the steady fracture strength have been mainly
treated. The Griffith theory, however, should be of no validity in
discussing the problem of initiation of a crack because it is
essentially the equilibrium theory of a crack.
In this paper, first, the condition of initiation and the
following propagation of peeling are considered with the use of a
Fig. 3. Peeling of flexible tape from a rigid

plane.
426 T AKAHASH I IGARASH I
mechanical model to reduce the difficulties in the stress analysis.

It is valuable not only for simplifying the analysis but for
bringing the structure of the theory into relief. Second, the
influence of relaxation on the steady peel strength is discussed and
the effect of energy dissipation in materials is made clear.
EQUILIBRIUM THEORY ON PEELING (7)
The specimen, for which the peel strength is to be calculated,

is shown in Fig. 1. The experimental peeling force obtained for
such specimens is in the order of 10 kg/cm; therefore i t can be
assumed that the adherends of canvas are flexible but inextensible.
It follows from this assumption that no work is done on the
adherends by external forces during bending and also that no work is
done when a tensile force is applied during peeling. The rubbery
adhesive is made from unfilled synthetic rubber and thus can be
regarded as sufficiently elastic. Therefore energy dissipation
during deformation can be neglected.
Because of the above assumption a simple Hooke model for

peeling, as shown in Fig. 4, can be adopted. In order to simplify
the calculations, it is assumed that symmetry exists with respect to
the x axis and only the upper half needs to be considered. In Fig.
4, the adhesive layer has unit width, and y is the half thickness
of the adhesive. 0
Thus, one can obtain the following basic equations of mechani-

cal equilibrium with respect to x and y directions respectively:
;--r--r---,r----y- ~
Yo
o
Fig. 4. Mechanical model of peeling.
£....
dS
(#.!)
ds
=0 (6)
~S (~) - £E = 0 (7 )
where £ is the strain of adhesive layer directed to the y axis and E

is its Young's modulus. As the specimen has unit width, S is the
tension in the adherend for unit width and s is the length along the
adherend. From Eq. (2),
C<~
~ - cons t .:: S 0
- (8)
It follows from Eq. (8) that the x component of the tensile force is
constant and is equal to S everywhere along the adherend. Eqs. (6)
and (7) can be rewritten is
2
S d x + dSdx = 0 (6')
ds2 dsds
2
S I I + dSdy - £ E = 0 (7 ' )
ds2 dsds
Using the relation (dx/ds) 2+(dy/ds) 2 = 1, these equations can be

combined to obtain
dS_£E.Q!=O (9)
ds ds
Since £ = (y-yo)/yo and S=So at y=Yo' integration of Eq. (9) leads

to
2
S = Ey o£ 12 + S0 (10)
From Eqs. (3) and (6),

Ey
(~ £2 + S ) dx = S ( 11)
2 0 ds 0
Using the relations
dx dy/dx
ds = ll+(dx/dy)2] 1/2
and dy=yod£, Eq. (11) leads to

dx 2S
- 0
d€ = + -
E
(12)
in which the minus sign is adopted so that for x .... 0, € goes toward
+co,and thus the origin of the x axis corresponds to infinite
values. Integration of Eq. (12) gives
So 1/2 { exp [-(E/yoSo) 112xl }

€ = 4(E) 112 ( 13)
Yo 1 - exp[-2(E/y S) xl
o 0
Since € = (y/y )-1, the shape of the adherend under the action of
peeling force ~s given by
112
YoSo 112 { exp [-(Ely oSo) xl }
y = 4 (-E- ) 112 + Yo (13')
1 - exp[-2(E/y S) xl
o 0
It should be noted that the relation can be derived without intro-
ducing any criterion for fracture.
The variation of the free energy of the deformation, of, with

respect to the difference of the crack length can be expressed:
(14)
where € b is the € at the breaking point. The variation of the

surface free energy, of , is given like as Eq. (3):
s
of s = ro x (15)
where r is twice the surface free energy.

The second law of thermodynamics requires the relation
(Griffith criterion):
of :: of e + of s =0 (16 )
and this leads to the results with Eqs. (13) and (15):
(17)
or
(18)
On the other hand, the value of So is obtained for a given

peeling angle e using the relation
-
(dE)
dx xb
= 1Y tan e (19)
in Eq. (12):
~ (20)
tan e
Then
= EE2 {1 + sec e } (21)
So Yo b 2tan e
From Eqs. (10) and (11), the steady peeling strength f , which
s
corresponds to the stress S at the point of peeling, is derived as
2f s =r + 2S 0 (22)
Combining Eqs. (11), (21) and (22), one obtains

f= U{1-cos e) (23)
s
For the 90 0 peeling, Eq. (23) becomes:
f
s = r (24)
Equation (23) shows that the steady peeling force is deter-

mined only from the surface energy and the peeling angle in the case
of this peeling. Stress distribution obtained with much hard
working and expressed analytically does not appear explicitly in
the result (though in the case of Griffith too I cf. Eq. (4}) •
Further, Eq. (23) contains no Young's modulus nor the thickness of
the specimen that determines the strain or the stress at the peeling
point. Eq. (24) merely tells us that the work done by the external
force fs is equal to the work to produce the surface, that is:
(25)
Let us consider this fact in other thermodynamic form. As the

work done by the external force is stored as the potential energy
Fe' the following relation is obtained:
-F e (26)
o
where the integration with ds is taken to the displacement at which

the elongation of the adhesive layer reaches the breaking point.
The increase in the surface free energy has been given by Eq. (15).
The condition of equilibrium is given:
6 Fe + r dX b = 0 (27)
From Eqs. (26) and (27),
f __
af
(~)
s ds = r (28)
aX b
From the geometrical relations, one obtains:
ds = (1-cos e)dx b (29)
Then, from Eqs. (28) and (29), integrating df to the peeling force
f s , one obtains the same result as Eq. (23),
(f )
J s df = f
s
= r/(1-cos e) (30)
The above treatment is the feature of the Griffith theory

applied to the simplest example of the steady peeling. As for the
equilibrium or the stability of the crack or the cut in this
experiment, the peeling point is almost completely elongated along
the tensile direction and no shape of crack tip can be found as
shown in Fig. 1.
To compare this behavior with that of the initiation of the
peeling may be valuable.
THE CONDITION FOR INITIATION OF PEELING (7)

As mentioned above, the condition for the formation of the
initial crack leading to the fracture may not be derived thermo-
dynamically. Then we assume that peeling starts when the extension
of the spring reaches the breaking elongation ib or the breaking
stress 0b (these two are different from the values given by Eqs.
(17) and (18» at the peeling point x=xb in Fig~ 4. This assumption
is a working hypothesis which may not be derived from thermodynamics
or other mechanical laws. However, adopting this hypothesis in-
stead of Griffith condition, one can obtain the initial peeling
force, fi' as follows.
In this case, Eq. (10) gives
(31)
From Eqs. (29) and (31) the force required to ini tia te peeling
according to the maximum stress criterion is derived as
(32)
Since €bE:Ob' Eq. (32) can be rewritten as
(33)
For the 90 0 peeling, Eq. (33) becomes
(34)
r in Eq. (23) led from Griffith condition is replaced by €bOyyo/2

in these results. As €bO Yd2 is the work required for elongation of
unit volume of Hookean mIterials to the breaking point, Eq. (33)
shows that the fracture takes place when the external force fi has
done the work required for breaking a unit length of the adhesive
layer. However, the value of "fracture energy" obtained here is
generally not equal to the value of free energy in elastic deforma-
tion by which the condition of the steady peeling is led; nor has it
any relation with the free energy within this treatment thus far.
While f. is proportional to the thickness of the adhesive layer
obViousfy from Eq. (33), f is independent of the thickness.
Therefore, in the case of a Speeling experiment using a specimen
(Fig. 5) with thickness more than 2r/Ebob' the value of load
continues to rise until the start of peeling, then falls down to the
definite value of r at the same time as an initial crack appears,
and remains constant during steady peeling. The peak value is
proportional to the thickness.
The results of the experiment are shown in Fig. 6, which

suggests the validity of the maximum stress criterion for the
initiation and of the thermodynamic theory for the propagation of
peeling. Though the effectiveness of Griffith theory for the
continuation of peeling is thought to be elucidated from the
consideration above, we have no theoretical foundation why the
start of peeling is subject to the conditions of the maximum stress.
This problem is out of the frame of thermodynamics.
rC"nvos ~Adhesivel oyer
j
'J
i-: ----
Fig. 5. Peeling specimen.
20
;:;; 10
° 2 4
time (min)
6 8
Fig. 6. Peeling force - time curves for constant

rate of extension. y (mm): A,0.3; B,0.8;
o
C,1.0; D,1.6; E,3.5.
PEELING AND ENERGY DISSIPATION (8)
One of the problems which are required to solve within the

frame of thermodynamics is the treatment of relaxation phenomena in
fracture. Different from the thermodynamic equilibrium state, the
free energy cannot provide a condition for steady state in irrever-
sible process and it indicates only the direction of change in the
system. In the case that the steady state of a thermodynamic system
is realized, the system, being worked by an external force, converts
a part of the work done into heat. In such a system the free energy
is maintained at a higher level than the equilibrium state. It is
known that the condition for the steady state is determined by the
principle of minimum entropy production (9) • Therefore, the
criterion for the steady peeling of dissipative system should be

derived from this principle. However, it may be allowed to assume
that the condition for the steady state becomes near the condition
of minimum free energy with a small correction in the case of the
system close to the equilibrium.
From the above consideration, let us adopt the following
expression from the condition for the steady peeling of dissipative
materials:
OF = OF e + r dx + oU' =0 (35)
where U' is the energy dissipation during the deformation of the
adhesive in peeling. Therefore, in this expression the dissipative
energy is also to be formalistically regarded as a sort of free
energy equivalent to surface free energy.
Provided that u' is the dissipation of energy per unit volume
of the adhesive during the deformation to fracture and subsequent
recovery (Fig . 7), oU' becomes as:
oU' = hu'ox (36)

where h is the thickness of the adhesive layer. For 90 0 peeling,
the following relation can be derived from Eq. (36) by the same
treatment as deriving Eq . (24) from Eq . (16) :
f
s =r + hu'
Fig. 7. Stress-Strain curve.

Comparing Eq. (37) with Eq. (24), one can find that the peeling
strength is to be increased by the second term. In the peeling,
this additive term due to the energy dissipation represents the
dependeooe of the peel strength on the thickness of the deformed
layer.
The results of peeling experiment for the specimens with the

various values of u' are shown in Fig. 8 and in Table 1.
Concerning evaluation of the energy dissipation, a question

arises. As the energy dissipation takes place in the adhesive
layer, the value of u' can be determined from the area enclosed by
the hysteresis (Stress-Strain) curve in a simple extension test of
the adhesive. However, the extent of the elongation (i.e., 90 or
0b) cannot be determined. This value is theoretically given by Eq.
(17) or (18) in the case of elastic materials. Irreversible
thermodynamics, i.e., the principle of minimum entropy production,
must tell us a way to solve this problem. It is noted that assuming
some fracture condition in such a steady fracture will be always
tentative and conventional. Agreement of the values of u' obtained
from S-S curve elongated to the breaking point of simple tensile
test with those obtained by Eq. (37) from the dependence of peeling
strength on the thickness is rather satisfactory as in the case of
this kind of experiment.
100
No.1
/
........
E
~
en /
- I/°
~50
N
/
___.-_===-=_1 =I_- No .2
L:_!'.I'_~~I~==~.=.~..:.N:.::o::.3
c 234
h(mm)
Fig. 8. Dependence of peeling force on thickness

of the adhesive layer. (e),e=50min- 1
(x), e:=5min- 1•
"T1
::tI
l>
~
C
::tI
m
(")
Table 1. Results of Tensile and Peeling Tests ~
-t
m
::tI
. . -1
€,m1n I0b,kg/cm2 I ~ I u',kg/cm2 r ,kg/cm I u',kg/cm2 »~
Specimen No. ~
m
Tensile Peel ;;
G)
5 150 11.0 1 320 9.0 I 220
50 155 1.0 413 9.0 220
2 5 21 . 2 6.2 29 1.5 31
50 25.6 1.3 29 8.5 21
3 5 15.5 0.15 0 3.5 0
50 20 1.0 0 4.0 0
....
W
01
DISCUSSION
As shown in Fig. 6, the initial peeling force increases with
the thickness of the adhesive layer. On the other hand, the force
during the subsequent steady peeling process is practically inde-
pendent of the adhesive thickness. These results correspond to
those given by Eqs. (32) and (24), respectively. This is the
experimental verification for the fracture criteria on peeling of
elastic materials as mentioned above.
It may be interesting to classify the various formulas which

have been proposed for the peel strength according to the results of
this investigation and to examine whether they relate to the
initiation or to the steady propagation of peeling. The studies
made by Rivlin et al. (4) may be considered to refer to the
propagation where the tearing energy governed the peeling or
tearing strength. On the other hand, the treatment of peeling for
flexible ribbons made by Bikerman (10) may concern the initiation of
fracture, where the tensile strength of the cement is the major
factor in the peel strength.
As for the effect of energy dissipation on the peeling, the
relation between peel strength and thickness of the adhesive is
found experimentally to be linear as expected from Eq. (37). For
the adhesive with no energy dissipation, the strength is indepen-
dent of the thickness of the adhesive as expected from Eq. (37) by
setting u'=O.
Values of r, however, are of order of 106erg/cm2, whi~ is muc~
greate 2 than the values for a usual solid surface (10 to 10
erg/cm). This may be explained by the energy of rupture and by the
"draw out" of the polymer chains (11). This implies that r treated
as free energy formalistically has the character of the dissipation
energy. This consideration, however, may not require to rewrite the
minimum relations represented by Eqs. (16) and (35).
Since the Griffith theory and its extension have the gen-
erality as the phenomenological theory from thermodynamical
essence, it may be true that the theory needs not be modified in the
future. In other words, the theory can be considered as only
presenting the framework in which various phenomena can take place
according to the laws of nature. Cracks in a solid, for example,
govern the strength of the materials whose behaviors can be consi-
dered within this framework.
REFERENCES
1. T. Igarashi, ~ Polym. Sci. Phys. Ed. ~, 407 (1978).
2. A.A. Griffith, Ph!. Trans . .!!2L:. Soc. (London) A 221, 163

(1920). --
3. C.E. Inglis, Proc. Inst. Naval Architects, 14, March (1913).
4. R.S. Rivlin and A.G. Thomas, !!..:. Polym. Sci. lQ., 291 (1952);
H.W. Greensmith, !!..:. ~ Polym. Sci. 1, 183 (1960); H.W.
Greensmi th, L. Mullins and A. G. Thomas, Trans. Soc.
Rheo!. 4, 179 (1960). --
5. T. Hata, KObunshi Kagaku (Polym. Chem.) 467, 72,77 (1947).
6. T. Hata, Zairyo (Materials) 11, 341 (1964).
7. T. Igarashi, op. cit.
8. T. Igarashi, !!..:. Polym. Sci. Phys. Ed. 11, 2!29 (1~75). ~
9. I. Prigogine, "etude Thermodynamique des Phenomeneslrreversi-
bles", Desoer, Liege (1947).
10. J.J. Bikerman, !!..:. ~ Phys. 28, 1484 (1950).
11. T. Igarashi, !!..:. of The Adhesion Soc. of Japan ~, 5 (1972).
Superposition of Peel Rate, Temperature
and Molecular Weight for T Peel Strength
of Polyisobutylene
Toshiya Tsuji, Mineo Masuoka and Kazumune Nakao
Osaka Prefectural Industrial Research Institute
2-1-53, Enokojima, Nishi-ku, Osaka, Japan 550
ABSTRACT
Effects of molecular weight of adhesive polymer on T peel

strength were investigated over a wide range of temperature and peel
rate. Polyisobutylene fractions having narrow molecular weight
distribution were used as the adhesive, and low-density poly-
ethylene films as the substrate. The peel strengths were found to
yield a single master curve with respect to the peel rate and the
temperature for both the cohesive and the interfacial failures. It
was found that the shift factor a r agrees with that of the Williams-
Landel-Ferry (WLF) equation and 1S almost independent of the mole-
cular weight of the adhesive. Furthermore, the master curves
related to the rate-temperature are superimposed by the use of the
shift factor aM to yield a single master curve relation with respect
to the molecular weight of the adhesive. In the case f.f4 the
cohesive bond failure, aM is related to the ratio of (M/MO) . .
INTRODUCTION
Many studies on peel strength have shown that the time-

temperature reduced variable treatment can be applied to super-
imposed peel data in the similar way as the application to the
viscoelastic response of adhesive polymers (1-5).
It is well known that adhesion properties are greatly influ-

enced by the molecular weight of the adhesive polymer (6-9).
439
440 TOSHIYA TSUJI ET AL.
However, little has been reported in detail on the relationship

between them by the use of the adhesi ve polymer having narrow
molecular weight distribution. This will be attributed primarily
to the fact that the preparation of a large quantity of polymer
fraction having narrow molecular weight distribution is very
laborious. We were able to obtain a large quantity of polyiso-
butylene (PIB) fractions by the use of a large column fractionation
method.
The purpose of this study is to investigate the relationship

between the molecular weight of the adhesive polymer and the peel
strength in terms of the superposition principle by the use of PIB
fractions as the adhesive. T peel strength was measured in a wide
range '4f molecular weight of the adhesive between 2.0x10 4 and
gO.2x10 over a wide range of temperature and peel rate.
The peel strength obtained for each molecular weight of the

adhesive was found to yield a single master curve with respect to
peel rate, temperature, and molecular weight of the adhesive.
EXPERIMENTAL PROCEDURES
Two kinds of PIB were used as the starting materials for

fractionation, VISTANEX MML-80(~ :60X10 4 ) and VISTANEX LMMH(A :
5X104) both supplied by ESSO CH~MICALS Co. The fractionatiJn of
VISTANEX MML-80 was accomplished by the use of the theta elution
column method using benzene as the solvent and changing elution
temperature. In the case of LMMH, benzene-acetone mixtures of
increasing solvent power were used as the eluting solvent at
constant temperature, 25 o C. The load of PIB was about 80g per batch
and the same fractionation procedure was repeated several times.
The results of fractionation are summarized in Table 1. The ratio
of the weight- to number-average molecular weight, (M 1M ), indi-
cates fairly a narrow molecular weight distribution ~f €he frac-
tions obtained. The viscosity-average molecular weights, M , of
v
eight fractions selected as the adhesive from these fractions for
the present study were 2.0x10, 4 4 6.1x10,
3.1x10, 4 4
10.6x10,
21.8x10 4 , 36.0X10 4 , 46.6x10 4 and 90.2x10 4 , respectively.
The low-density polyethylene film of 0.13 mm thick, supplied

by MITSUBISHI JUSHI Co., was used as a substrate and, prior to use,
wiped with distilled acetone and dried in atmosphere for two to
three hours.
Test specimens were prepared as follows: Each substrate was
coated with a n-hexane solution of the adhesive, and after dried
overnight in vacuo at about 40 0 C, one film was placed over another
T PEEL STRENGTH OF POLYISOBUTYLENE 441
Table 1. Fractions of Polyisobutylene
a) b) c)
LMMH
d)
M M /M Mw Mn Mw/Mn
[n] v w n
fra.No.
dl/g (x10- 4 ) (x10- 4 ) (x10- 4 )
2 0.091 0.8 1.28

3 0.121 1.3 1.19
4 0.158 2.0 1.11
5 0.181 2.1 1.11
6 0.220 3.1 1.18
1 0.236 4.3 1.11
8 0.265 5.2 1.11 5.6 5.4 1.0 4
9 0.300 6.1 1.16
10 0.362 9.6 1.11
11 0.382 10.6 1.18 11.1 11.5 1.02
MML-80
fra.No.
2 0.560 21.8 1.41
3 0.130 36.0 1.34
4 0.838 46.6 1. 33
5 0.950 59.0 1. 33
6 1. 111 19.4 1. 31
1 1.190 90.2 1.31
a) GPe
b) light scattering
c) osmotic pressure
d) benzene, 25 0 e
with a spacer of 100~ thickness between them and then they were
pressed at 60 0 e for 10 min. Prior to testing, the bonded ~ecimens
were cut by 25 mm in width and conditioned overnight at 23 e and at
RH 60%.
T peel st~ength was measured at five temperatures in the range

from -30 to 40 e and at seven or eight rates in the range from 0.5 to
100 cm/min at each temperature by the use of Shimadzu Autograph DSS-
2000 tensile tester and a temperature control chamber. T peel
strength was averaged over three specimens.
Relationship between Molecular Weight of Adhesive and T Peel

Strength
The dependence of the peel strength on the molecular weight of

the adhesive is given in gigs. 1 and 2, in the case of the test
temperatures of 40 and -20 C, respectively.
As the molecular weight increases, T peel strength for any

rate, except with 50, 20 and 10 cm/min at -20 0 C, rises steadily up
to a maximum strength and the mode of bond failure in this region
was cohesive. Since the cohesive force of adhesive polymer in-
creases generally with the molecular weight in the region of low
molecular weight, the peel strength can be expected to increase with
the molecular weight. Beyond a maximum value, the peel strength
decreases abruptly and the mode of bond failure changed from
cohesive to interfacial failure, which had no visible adhesive
remaining on one side of the substrate, because the cohesive force
of the adhesive is expected to exceed the interfacial adhesive force
between the adhesive and the substrate. Finally, the peel strength
reaches an approximately constant value with increasing molecular
weight. The peel strength is altered with the molecular weight even
slightly at lower peel rate and markedly at higher rate. With any
molecular weight of the adhesives, the peel strength increases with
the peel rate for both modes of bond failure.
symbol peel ratt>(cm/min)

(!) 50
f---t ~ 20
coht>sive intt>rtac.al 0 10
E failure (i) 6 5
E (e) x 2
~ 1.0- .,
o 1
0.5
0.
25
r.
.oJ
01
C
Q.o
....
Vl 0,5
OL-~~~~~--~~~~~
4 5 6
log Mv
Fig. 1. Effect of molecular weight of poly iso-

butylene on T peel strength (40 0 C).
symbol pe-el rate- (cmimin)

o 50
() 20
o 10
ll. 5
3 )( 2
o 1
C> 0.5
slip-stick failure-(s)
Fig. 2. Effect of molecular weight of polyiso-

butylene on T peel strength (-20 0 C).
At the peel rates of 50 and 20 cm/min at -20 0 C, the slip-stick

failure was observed over the range of the molecular weight studied,
where the peel strength fluctuated widely with time as shown in Fig.
3. Therefore, the mean value of each peak strength was taken as the
peel strength. The schematic representation of slip-stick failure
is shown in Fig. 4. The slip-stick phenomenon would be associated
with the onset of the transition from rubberlike to glass like
behavior of the adhesive. The peel strength was almost independent
of the molecular weight of the adhesive in this slip-stick region.
Results similar to those shown in Fig. 1 were obtained at the

test temperatures of 20 and OOC. In the case of the adhesive having
M of 10.6x10 4 , a mixed failure of both cohesive and interfacial was
o~served at 20 0 C for the peel rates of 5, 10, 20, 40 and 100 cm/min,
and at OOC, 0.5, 1 and 2 cm/min. The dependence of the peel
strength on the molecular weight obtained at -30 0 C was similar to
that obtained at -20 0 C. The maximum peel strength shifted to ~igher
molecular weight side with increasing temperature from 2.1x10 at
-30 o C to 10.6x10 4 at 40 0 C.
444 TOSH IY A TSUJI ET AL.
t ime(distance peeled)
Fig. 3. Example of pattern of slip-stick failure.
t
rsubstrate
Fig. 4. Schematic representation of slip-stick

failure.
Superposition of Peel Strength
The dependences of the peel strength on both the peel rate and
the temperature, and the peel rate-temperature superposition of the
peel strength ar~ given in Fi!s. 5 and 6 in the case of molecular
weight of 3.7x10 and 90.2x10 ,respectively. In these and later
fi§ures, the peel strength was reduced by the ratio of 293fT, where
T( K) is the test temperature.
The solid line and the dashed line in Fig. 5 indicate the
cohesive and the slip-stick failure, respectively. On the left
side, the peel strength is related to the peel rate, R, for five
test temperatures. In the region of the cohesive failure, the peel
strength increases with the peel rate, but in the range of the slip-
stick failure, the peel strength decreases with the peel rate. A
single master curve obtained is shown on the right side, where the
T PEEL STRENGTH OF POLYISOBUTYLENE
~
-30\,
,
I
s\,
I
Fig. 5. Master relation for T peel stren§th

against peel rate, reduced to 20 C.
~ 2~
--
N
E!~
..c. 1
g;
...CIJ
tii
a;
Q.o
a.

against peel rate, reduced to 20 C.
TOSHIYA TSUJI ET AL.
factor aT relating equivalent rates at a temperature T and a

reference temperature (293 0 K) was estimated by superposing the
strength versus peel rate relation at any temperature upon that at a
reference temperature. The peel strength drops abruptly at the
transition region from cohesive to slip-stick failure.
Though not shown here, the results similar to khat in Fig. 4 5

were obtained for the adhesives having M of 2.0x10 and 6.7x10 ,
v
respectively.
The solid line and the dashed line in Fig. 6 refer to the
interfacial and the slip-stick failure, respectively. As shown on
the left side, in the region of the interfacial failure the peel
strength increases with the peel rate, but in the region of the
slip-stick failure the peel strength decreases with the peel rate.
A single master curve obtained is shown on the right side, where the
peel strength falls abruptly at the transition region from inter-
facial to slip-stick failure. Though not shown here, similar master
curves l(ere obtained for the adhesives having Mv of 36.0x10 4 and
46.6x10 .
It is noted that the single master curve relation is obtained

through both the cohesive and the slip-stick failure (Fig. 5) or
both the interfacial and the slip-stick failure (Fig. 6), respec-
tively.
- of 10.6x10 4 , however, data points
For the adhesive having Mv
were divided into two regions, one consisting of the cohesive
failure and the other consisting of the interfacial one. The
results obtained are shown in Fig. 7. The solid line ref'ers to
interfacial failure, the dotted line cohesive one, and the dashed
line slip-stick one, respectively.
The good superposition obtained under different test condi-

tions indicates that the dependence of the peel strength on the
temperature and peel rate is due to a response of viscoelastic
properties of the adhesive polymer for any failure mode.
Figure 8 shows the shift factor aT versus temperature relation

reduced to 20 0 C. It can be seen that aT is almost independent of
the molecular weight of the adhesive polymers. This fact would be
explained from the rheological rate process, that is, aT is approxi-
mated as
(1)
aT = nr/nT
o
where nT and nT are the viscosities measured at the standard
o
My: IO.6xl()4
IIII
0.
2
c .1\· \
;.
I \.
\
~
-;
EI~ s \
.bot
.,
\
A \
\
i;
'"1':
\
f...
\
:" I I
. . 'I
.7
A I·
1(
vi
10 ..... 0
·.. ·.,.. ····· /a
-I 0 I 2 -I 0 I 2 3 4 5
10~cmJmin)\ log IR a,.<c ml mi nll

against peel rate, reduced to 20 c.
Mv
o 2.0xl04
4 x 3.7
<:>6.7
10.6
3 ~ C
4360
El·, .90.2
a ---, _,8
2 A_ ,
,
... 0
,,
'- *-,,
IV
CI
0
0 ·' e. o
. . ... -..... . . ~
-1 •
-30 -20 -10 0 10 20 30 40
temperature :C
Fig. 8. Dependence of aT on temperature, reduced
to 20 oC.
temperature, TO' and the test temperature, T, respectively. On the

other hand, the relationship between viscosity, n, and molecular
weight, H, is expressed by
(2)
for the temperature T and
for the temperature TO' where K.r and KT are constants for each
temperature and a is equal to 3.4 abov.9 the critical molecular
weight, M (M of PIB:1.7x10 4 (10».
c c
Substituting Eqs. (2) and (3) into Eq. (1), we obtain
(4)
The ratio KT/KT is independent of molecular weight and dependent

exclusively on eemperature. It is found, therefore, from Eq. (4)
that the shift factor aT is independent of the molecular weight as
shown in Fig. 8.
In order to compare aT value in this study with that of the

literature, aT was re-estimated with respect to the reference
temperature of 243 0 K(T ) (11). The results obtained are given in
Fig. 9. The solid lin~ indicates the Williams-Lande I-Ferry equa-
tion (12) defined by the following:
-8.86(T-T )
log aT 101.6 + T-T -
= -=-=-~_-=-=s (5)
s
Ts is the temperature at which aT is unity. It can be seen that the
W-L-F equation is superposed on our data.
We attempted to superpose all the above master curves with
respect to mol~cular weight of adhesive. As the reference molecular
weight, 2.0x10 was adopted, Fig. 10 shows the results obtained. It
is found that a high portion of the data are superposed well on a
master curve.
The symbol ~ means the shift factor for the molecular weight,
by which the curves are shifted along the log RaT axis to superpose
the reference curve.
It is quite interesting that the dependence of the peel

strength on the time, the temperat~re and the molecular weight can
be unified by the use of the shift factors aT and aM as shown in Fig.
10. That is, the principle of the superposition based on the W-L-F
theory is extended to the effect of the molecular weight on the peel
strength in this study. The dependenc;.e of aM on the molecular
weight is shown in Fig. 11. The symbol 0 indicates the datum which
f;lv
o 2.0x104
x 3.7
e 6.7
o 10.6
II 36.0
• 90.2
o _ , log aT: -&86(T - T~)
101.6+-T - T.
( Ts= 243·k)
,; -1
Ol
o
- -2
-3
-30 -20 -10 10 20 30 40

temperature, ·c
Fig. 9. Dependence of aT on temperature, reduced
to -30 oC.
Mv
o . 2.0X 10'
X. 3.7
3
0.10.6
11. 36.0 0
,,
• • 90.2 ,
E 0
,
E
'"
N ""
0. 0
,
x:: :0
lb' ,
EI~ 2 "
L:
.....
C
......
41
III
~
41
a.
-1 o 2 3 4 5 6
I oglRaT"M(cm/min)\
Fig. 10. Superposition of peel rate, temperature

and molecular weight for T peel strength
of polyisobutylene.
corresponds to the maste~ curve consiijting of the cohesive failure

for the adhesive having M of 1o.6x10 .
v
In accordance with our consideration, the shift factor aT
includes all of the rate-temperature effect on the peel strength and
on the other hand, the shift factor ~ has the effect of the
molecular weight on the peel rate. From the viewpoint of the
rheological rate process, ~ is expected to be equal to nMftlM '
where MO is the standard molecular weight, in the similar manner ~o
Eq. (1). The relationship between the viscosity and the molecular
weight is expressed by
(6)
and
Thus, the relationship between ~ and molecular weight is given by

the following:
(8)
or
log ~ =
(l 10g(M/MO) (9)
That is, the plot of log aM against 10g(M/MO) must show a straight
line having a slope of (l. In general, (l is equal to 3.4. The
...,.
~
t
1.5
1S',
I
•
I
I
I
I
I
~ 1.0 I
tU I
~
01 I
0 I
I
- 0.5 I
I
I
I.
0
4 5 6
log Mv
Fig. 11. Dependence of aM on molecular weight
(Reference molecular weight: 2.0x10 4 ).
relation of log aM = 3.4 10g(M/MO) is given by the dashed line in

Fig. 11. As shown in Fig. 11, in the region of the molecular weight
below 10 5 where the cohesive bond failure occurred mainly, the
dependence of the shift factor aM. on the molecular weight agrees
well with that predicted by Eq. (9), which was derived from the 3.4
power's law between the viscosity and the molecular weight. On the
other hand, in the range of the higher molecular weight where the
interfacial failure occurred, the shift factor aM decreases
abruptly to much smaller value than that predicted from Eq. (9). It
is concluded, therefore, that the shift factor aM can be related to
the ratio of (M/M O)3. 4 under the test conditions where the bond
specimen shows the cohesive bond failure. The rate process on the
bond fracture consists of many elemental steps and the rate-
determining step is still uncertain. Consequently, it is very
difficult to give a reasonable clarification of the shift factor ~
for the interfacial bond failure.
It will be noted from Fig. 10 that the time-temperature

dependence of the peel strength can be further superimposed on the
molecular weight over a wide range of molecular weight by the use of
the shift factor aM' in other words, the principle of superposition
is extended so as to include an effect of the molecular weight on
the peel strength in terms of the shift factor aM defined in this
study.
CONCLUSIONS
The dependence of the T peel strength on the peel rate, the

temperature and the molecular weight of the adhesive polymer was
investigated by the use of the principle of the superposition. A
new parameter of the superposition is introduced by the rheological
rate process based on the W-L-F theory. This parameter is the shift
factor, aM' for the molecular weight, by which the master curves
related to the rate-temperature can be shifted along the axis of log
RaT to superimpose the reference molecular weight. The fraction-
ated polyisobutylene was used as the adhesive.
As the molecular weight of the adhesive increases, the peel
strength shows a maximum where the mode of the bond failure changed
from cohesive to interfacial failure, and the maximum point is
shifted to higher molecular weight side with increasing tempera-
ture.
The peel strength was found to yield a single master curve with
respect to the peel rate and the temperature for both the cohesive
and the interfacial failures. The shift factor aT agrees entirely
with that of W-L-F equation and is almost independent of the
molecular weight of the adhesive. The master curves related to the

rate-temperature are further superposed by the use of the shift
factor ~ to get a single master curve relation with respect to the
molecular weight of the adhesive. In the case of the cohesive bond
failure, the shift factor ~ is related to the ratio of (M/MO)3.4.
On the other hand, in the case of the interfacial bond failure, aM
is much smaller than the theoretical value. The shift factor ~
defined in this study on the molecular weight was found to be
explained qualitatively by the W-L-F theory in the range of rela-
tively low molecular weight.
REFERENCES
1. T. Hata, M. Gamo, K. Kojima and T. Nakamura, "Mechanics of
Peeling. II. Time-Temperature Superposition Principle in
Adhesive Failure". Chemistry of High Polymers (Japan),
22, 160 (1965).
2. D.H. Kaelble, "Theory and Analysis of Peel Adhesion: Rate-
Temperature Dependence of Viscoelastic Interlayers". J.
Colloid Sci., 12, 413 (1964).
3. A. N. Gent and R. P. Petrich, "Adhesion of Viscoelastic Mate-
rials to Rigid Substrates", Proc. ~ Soc. (London),
A310, 433 (1969).
4. Y. Nonaka, "Mechanics of Peeling. 5. Temperature Dependence
of Peeling Strength of Polyisobutylene", J. Adhesion Soc.
Japan, !, 207 (1968). ---
5. D.H. Kaelble, "Peel Adhesion: Influence of Surface Energies
and Adhesive Rheology", J. Adhesion, 1, 102 (1969).
6. K. Nakao, M. Matsunaga and H~Shakutsui, "Studies on Acrylic
Adhesives. Part 1. Bond Strength and Molecular Weight of
Polyethylacrylate", !h Adhesion Soc. Japan, 1, 15 (1965).
7. S.W. Lasoski and Jr.G. Kraus, "Polymer to Metal. The System
Polyvinyl Acetate-Steel", !h Polymer SCi., .!!!, 359
(1955) •
8. S.S. Voyutskii, "Adhesion and Autohesion of Polymers", Adhe-
sives ~, ,2, (4), 30 (1962).
9. A. D. McLaren and C. J. Seiler, "Adhesion. (III) Adhesion of
Polymers to Cellulose and Alumina", !h Polymer Sci., !,
63 (1949).
10. T.G Fox and S. Loshaek, "Isothermal Viscosity-Molecular Weight
Dependence for Long Polymer Chains", !h!P2.!.:.. Phys., 26,
1080 (1955).
11. M.L. Williams, R.F. Landel and J.D. Ferry, "The Temperature
Dependence of Relaxation Mechanisms in Amorphous Polymers
and Other Glass-forming Liquids", J. Am. Chem. Soc., 77,
3701 (1955). - - -- -- -
12. M.L. Williams, "The Temperature Dependence of Mechanical and

Electrical Relaxations in Polymers", ~ Phys. Chern., 59,
95 (1955).
Discussion
On the Paper by M. Masuoka and K. Nakao
A. Silberberg (Weizmann Institute, Israel): I find it very

interesting that your failure mechanism seems to be viscosity
related. Have you checked this by going below Mc into the region
where your <l should be about 1. Can you suggest why interfacial
failure should depend upon the bulk disentanglement mechanism which
accounts for the viscosity molecular weight dependence?
T. Tsuji, M. Masuoka and K. Nakao (Osaka Prefectural Indus-

trial Research Institute, Japan): Your questions are of great
importance from the viewpoint of the fracture mechanism of the
adhesi ve joint. (1) The polyisobutylene fractions below
M (:1.7x10 4 ) have very low viscosity, therefore, the appropriate
c
specimen for peel test was not prepared. (2) The temperature
dependence of the shift factor, aT' for both cohesive and inter-
facial failures fits the WLF equation exactly as obtained in our
results. It was found from the fractography of the interfacial
failure that the adhesive was deformed and ruptured in ductile
manner at the interfacial region. These facts show that the
viscosity of PIB adhesive plays a very important role in the
interfacial failure. In order to verify the shift factor, aM' an
application of the viscosity-molecular weight relation was attempt-
ed according to the WLF theory. This attempt is successful for the
cohesive failure but not for the interfacial one. The rate-
temperature-molecular weight superposition gave a fairly good idea
about what the interfacial failure means.
On the Paper by T. Igarashi
R. Evans (Mameco): What were the criteria for an adhesive that

had (or did not have) thickness dependence for peeling forces?
455
456 DISCUSSION
T. Igarashi (Gunma University, Japan): In both cases that the

adhesive had and had no energy dissipation, the steady peeling force
could be derived formalistically from principle of minimum free
energy. Since the dissipation energy shall be the quantity differ-
ent from the free energy, the criterion in the case of dissipative
adhesi ve must be the others, say, may be the minimum entropy
production principle. Because of the difficulty in theoretical
treatments, the dissipation energy in this paper was regarded as a
kind of free energy and was "renormalized" in it as shown in Eq.
(35) •
R. Evans: What adhesive had no thickness dependence? What

adhesive did?
T. Igarashi: The adhesives that have no energy dissipation do
not show the dependence of peeling force on the thickness. Such
adhesives behave quite elastically and do not show hysteresis in the
stress-strain curve; they, however, do not always obey Hooke's law.
In our experiment, the adhesive layers consisted of rubber vulcani-
zates which contain BR, SBR and no reinforcing fillers. To promote
adhesion the surface of adherend canvas was treated with RFL ..
AUTHOR INDEX
(The numbers in parentheses are references for discussion parti-

cipants. Pp. 1-456 are found in Volume 12A, pp. 457-858 in Volume
12B. )
A Avotin'sh, Y.Y., 549

Abraham, M.H., 825
Abricosova, 1.1., 75 B
Adam, N.K., 854
Adams, A.L., 682, 854 Baakker, G., 13
Adams, D.F., 472 Barich, V., 627
Adams, M.J., 101 Bach, D., 823
Adams, R.D., 193, 410 Bafr, Y., 122
Adamson, A.W., 174 Bagley, E.B., 52
Aggarwal, S.L., 362 Baier, R.E., 681, 822
Ah-Sing, E., 825 Baker, D.W., 790
Ahlquist, C.N., 362 Bale, W.F., 682, 835
Ahn, J., 519, 538 Barash, Y.S., 75
Ainbinder, S.B., 75 Barber, M.N., 14
Akhmatov, A.S., 74 Barquins, M., 203, 274, 275
Aleksandrova, N.A., 627, 641 Barrick, N.J., 824
Alexandrowicz, Z., 748 Bascom, W.D., 39, 162, 173, 193,
Allara, D.L., 549, 689, 751, 321, 338, 773
756, (794) Basin, V.E., 75
Alred, R.E' 4 163 Bates, R., 497
Amano, S., 13, 420 Bates, R.J., 501
Ananthanarayanan, S., 845 Baun, W.L., 121, 122
Anderson, H.R., Jr., 519 Baxt, L., 575
Andrews, E.H., 163, 274, 501 Beaumont, P.W.R., 472
Arakawa, H., 40 Beerbower, A., 52
Arakawa, M., 41 Bell, A.T., 538
Aris, R., 664 Bell, J.P., 122, 162
Armstrong, D.E., 823 Bellamy, L.J., 756
Armstrong, R.C., 664 Bellemans, A., 5, 13, 14
Ashbaugh, N., 472 Belyi, V.A., 67, 74, 75, 539,
xvii
xviii AUTHOR INDEX
549, 550 Briscoe, B.J., 275

Benko, B., 824 Brodkey, R.S., 338
Benninghoven, A., 122 Brody, H., 362
Benoit, H., 600 Bronsveld, W., 822
Bergh, A.A., 520 Brooks, R.E., 392
Bergmark, T., 122 Brownell, R.E., 824
Berlin, A.A., 75 Brunt, N., 275
Berry, J.P., 274 Brzozowski, K.J., 64
Bettelheim, F.A., 844 Bucc i , C., 173 , 193
Betz, G., 122 Bucknall, C.B., 338
Bigelow, C.C., 825 Bueche, A.M., 362
Biggs, H.G., 682 Buff, F. P ., 13
Bijsterbosch, B.H., 825 Bui, H.D., 274
Bikerman, J.J., 391, 437, Bulas, B., 747
549, 627 Bull, H.B., 824, 854
Binford, J.S., Jr., 640 Bullett, T.R., 519
Birch, D.A., 275 Burhnam, R.K., (196)
Bird, R.B., 664 Burns, H., 640
Birdi, K.S., 824 Busse, W.F., 501
Birkle, D.L., 320 Butler, D.W., 519, 538
Bisch, P.M., 643, 664, 665
Bitner, J.L., 321
Bjeletich, J.G., 193 C
Blanchard, A.F., 362
Blank, R.E., 101 Cagle, C.V., 320
Blumenthal, R., 664 Calien, H.B., 274
Bockris, J.O.M., 825 Cannon, S.L., 362
Bodnar, M.J., 312, 495 Cantow, H., 626
Boerio, F.J., 773 Carlson, D.E., 537
Bogie, K., 338 Carlson, E., 575
Bolger, J.C., 312 Carpenter, D.K., 640
Bonekamp, B., 824 Carr, S., 575
Boone, C.W., 673 Cartner, J.S., Jr., 299
Borisevich, N.A., 75 Chan, D., 600
Botham, R.A., 641, 728 Chan, M.G., 549
Boussinesq, J., 274 Chan, R.K.S., 40
Bowden, F.P., 75 Chandrasekhar, S., 664
Bradley, A., 537 Chang, C.C., 538
Brakere, R.R., 549 Changeux, J.P., 664
Brandts, J.F., 824 Chapman, B.N., 538
Brash, J.L., 682, 823, 824, 835 Chase, K.W., 298
Braun, P., 122 Chasovnikova, L., 847, 854
Breece, J.M., 537 Chattoraj, D.K., 854
Brewis, D.M., 495 Chen, Y.D.M., 362
Briggs, D., 495 Chesters, G., 823
Briggs, G.A.D., 275 Chia, C. Y., 472
Brinson, H.F., 141, 162, 163, Chiang, T.C., 362
279, 298, 299, Chiu, T.H., 825
(363), (364) Choo , Y. H., 121
AUTHOR INDEX xix
Chowols-Englert, A., 13 Davy, T. J ., 706

Chu, B., 627 Debenedictis, T., 495
Chumaevskii, A.N., 75 Debye, P., 627
Cirlin, E.H., 362 Deckert, C.A., 519, 520
Clark, A.T., 14, 626, 749 Defay, R., 13
Clark, D.T., 575, 706 Deferran, E., 472
Clark, E.S., 362 Degennes, P.G., 600
Cleary, E.G., 845 Dehl, R.E., 827, 835
Coburn, J.W., 121, 538 Dejardin, P., 747
Cohen, J.B., 790 Dell'Oca, C.J., 122
Cohen, R.E., 338, 362 Delollis, N.J., 495
Colson, J.P., 748 Derjaguin, B.V., 276
Comper, W.D., 845 Deryagin, B.V., 74, 75, 275
Conley, R.T., 773 Desalos-Andarelli, G., 275
Cottington, R.L., 39, 338 Dettre, R.H., 65
Cotton, J.P., 600, 627 Devanathan, M.A.V., 825
Counsell, P.J.C., 501 Devereux, O.F., 121
Counts, M.E., 162 Devine, A.T., 495
Courtel, R., 274, 275 Devore, D.I., 749
Coutts, M.D., 537 Devries, K.L., 362
Crane, L.W., 537 Dewitt, T.W., 338
Crochet" M.J., 162, 298 Dezelic, G., 823, 824
Crooks, H.N., 538 Dezelic, N., 823, 824
Crowley, J.L., 165, 193, (197) Diggle, J.U., 122
Cuthrell, R.E., 174 Dijstra, R., 320
Cuypers, P.A., 824, 825, 854 Dillman, W.J., 823
Dimarzio, E.A., 599, 640
Dini, J.W., 519
D Divens, W.G., 575
Doi, T., 728
Dahlquist, C.A., 501 Doner, D.R., 472
Dahmen, E.A.M.F., 320 Doumas, B.T., 682
Dakin, T.W., 575 Downie, T.C., 122
Dance, W.E., 163 Drago, R.S., 52
Danielli, J.F., 822 Drutowski, R.C., 275
Danil De Namor, A.F., 825 Dryfuss, P., (365)
Dann, J.R., 52, 65 Drylov, O.V., 824
Daoud, M., 600 Drzal, L.T., 121
Dasgupta, A.A., 298 Dubrovina, L.V., 75
Datta, P., (78), (197), (364), Dudek, L.P., 101
(579), (581),521 Duplessix. R., 600
Dautzenberg, H., 627 Dwight, D. W., (78), 121, 141,
Davidov, R.M., 824 162, (195), (197),
Davidson, R.G., 313 ( 198), (364), (577),
Davidson, T., 775, 790 (580), (581), (794)
Davis, H.T., 13 Dyatlovitskaya, E.V., 673
Davis, L.A., 362 Dynes, P.J., 312
Davis, W.M., 338 Dzyaloshinskii, I.E., 75
Davson, H., 822
xx AUTHOR INDEX
E Fieschi, R., 173, 193

Fisher, M.E., 13
Easterling, K.E., 275 Fixman, M., 749
Edelman, G.M., 854 Flaggs, D.L., 193
Egorenkov, N.r., 75, 539, 549, Fleer, G.J., 823
550 Flory, P.J., 641, 748
Eguchi, M., 575 Folkman, J., 673
Eirich, F.R., 52, 639, 640, 747, Fontana, B.J., 52, 641, 747
748 Forest, T.W., 706
Eisner, Y., 627 Foster, F.J., 824
Elbing, E., 65 Fowkes, F.M., 40, 43, 52, 65,
Elhorst, J.K., 824 (78), (79), (80),
Ellerstein, S., 640 728, 756
Ellison, H.A., 538 Fox, H.W., 64
Eltekov, A., 641 Fox, K.K., 641
Eltekov, Y.A., 640 Fox, T.G., 452
Endo, H., 392 Francis, S.A., 538
Ensminger, D., (198) Frank, H.S., 825
Epand, R.M., 41 Fridkin, V., 575
Erdogan, F., 410 Frisch, H.L., 392, 639, 640
Erman, V.Y., 627, 641 Frojmovic, M.M., 844
Eskin, V., 627 Fromageot, H.P.M., 825
Etzkorn, H.W., 122 Frommer, M.A., 748
Evans, C.A., 575 Fuks, S., 13
Evans, J.T., 275 Fukuda, H., 393, 411
Evans, R., (455), (456 ) Fukuzawa, K., 391
Everett, D.H., 13 Fuller, K.N.G., 275
Ewing, P.D., 193 Furasawa, K., 825
Eyring, E.M., (196) Furneau, R.C., 122
Furukawa, K., 420
F
G
Faber, W., 122
Fabulyiak, F., 627 Gabel, D., 824
Fahey, J.D., 321 Gaever, I., 673
Fairen, V., 665 Gail, M.H., 673
Farnoux, B., 600, 627 Gaines, G.L., Jr., 39, 65
Farrow, M.M., (196) Gall, W.F., 854
Feichtmayr, F., 575 Gamo, M., 452
Feinerman, A., 627 Gardon, J.L., 52
Felbeck, D.K., 472 Garrett, C.F., 121
Feng, M. Y., 472 Garroway, A.N., 339
Fenstermaker, C.A., 748, 824, Gebhard, H., 747
835, 854 Gelius, U., 122
Ferry, J.D., 274, 337, 452, Gemant, A., 575
501 Genkin, M.V., 824
Feyen, J., 824 Gent, A.N., 274, 391, 392,
Fiande, E., 338 452, 773
AUTHOR INDEX xxi
Gerald, J., 627 Grossman, R.F., 574

Gergelson, L.D., 673 Gueleva, S., 362
Gershman, H., 825 Guess, T.R., 163
Gersho, A., 101 Guggenheim, E.A., 13
Gerstlf, F.P., Jr., 163 Guidi, G., 173, 193
Gessler, A.M., 640 Gusev, S., 627
Gettings, M., 773 Gutmacher, R.G., 640
Gibbs, J .W., 13 Gutmann, W.H., 122
Gilliland, E.R., 640 Gutoff, E.B., 640
Gilman, A.B., 537 Guttmann, A.J., 14
Gilula, N.B., 673
Ginsburg, V.L., 75
Glansdorff, P., 665 H
Gledhill, R.A., 163
Gleiter, H., 361 Haager, 0., 854
Goan, J.C., 472 Hafner, J., 472
Goddard, R.L., 682 Hahn, H.T., 463, 472
Godde, J.S., 122 Hall, J .R., 495
Goldsmith, W., 298 Hall, J. T., 756
Golovachev, V., 627 Hammermesh, C.L., 537
Good, E. F ., 121 Hammes, J.P., 537
Good, R.J., 65, 162, 275, Hamrin, K., 122
391', 501 Han, C.C., 823
Goodland, R.L., 835 Hanlin, D., 103
Goodman, A.M., 537 Hanna, G.L., 472
Goody, W., 52 Hansen, C.M., 52
Goring, D., 575 Hansen, F.Y., 664
Goulding, C.W., 122 Hansen, J.E., 392
Gounder, R.J., 775 Hansen, R.H., 495, 575
Grabovac, I., 320 Hansma, P.K., 756
Graham, D.E., 822, 823 Hara, K., 640, 641, 728
Grant, W.H., 587, 641, 799, Harey, E.N., 728
823, 824, 827, 835, Harper, R.C., Jr., 338
(855), (856), (857) Harrington, U., 122
Grayson, M.A., (366) Harris, B., 472
Greeland, J., 626 Harris, J., 338
Greenler, R.E., 773 Harris, L.A., 538
Greensmith, H.W., 437 Harshbarger, R., 100
Gregg, R.A., (363) Hart-Smith, L.J., 163, 410
Gregor, L.U., 537 Hartog, J.J., 174, 193
Gregory, H.J., 495 Harvey, E., 627
Gribova, I.A., 75 Harvey, E.N.J., 641
Grievenkemp, J.E., 773 Hashimoto, N., 520
Griffith, A.A., 437 Hassager, 0., 664
Griffiths, P.R., 538 Ha ta, T., 15, 39, 40, 41,
Grinnel, F., 673 (77), 391, 437, 452,
Gross, B., 575 496
Gross, J., 844 Hauschka, S., 681
Grossberg, A.L., 854 Hauser, E. A., 41
xxii AUTHOR INDEX
Hayama, H., 728 Igarashi, T., 421, 436, 437

Hayashi, T., 410 (456)
Heaviside, 0., 575 Ignatova, T., 627
Hedeman, J., 122 Ikeda, M., 561
Heden, P.F., 122 Illers, K.H., 392
Hedvig, P., 173 Imoto, T., 640, 641, 728
Heimbuch, R.A., 472 Inagaki, H., 641
Heimenz, P.C., 65 Inglis, C.E., 437
Helfand, E., 65, 600 Inoue, Y., 392
Hellman, M.Y., 640 Inubushi, H., 728
Helmkamp, R.W., 682, 835 Iosipescu, N., 193
Hemker, H.C., 824, 825, 854 Ishida, H., 773
Hennenberg, M., 664, 665 IShikawa, H., 362
Herakovich, C.T., 162, 298 Israelachili, J., (683), (792)
Herbst, M., 854 Ito, T., 501
Hermens, W.T., 824, 9825, 854
Herzlinger, G.A., 825
Hesselink, F.T., 599, 747, J
748, 823
Heurtel, A., 275 Jackson, D.S., 845
Hill, T., 825 Jannink, G., 600
Hino, K., 501 Jay, R.R., 320
Ho, T .C., 338 Jenkins, T.R., 174
Hochmuth, R.M., 682 Jennings, L.W., 162
Hock, C. W., 501 Johansson, G., 122
Hoeve, C.A.J., 599, 626, 640, Johnson,·D.J., 101
747, 748, 823 Johnson, H.R., 519
Hoffman, A.S., 673, 681, 682, Johnson, K.L., 274, 275
691, 706, 822 Johnson, R.E., Jr., 65
Hohnson, W., 575 Jonath, A.D., (78), 165, 175,
Holly, F.J., 706 193, (195), (197 ) ,
Honig, R., 122 (198), (364)
Horbett, T.A., 677, 681, 682, Jones, L.G., 338
(684), (685),706, Jones, R.L., 173, 193
(855), (857) Jones, R.N., 773
Horh, K., 575 Joppien, G.R., 641
Howard, G.J., 52, 640, 641 Josefowitz, D., 501
Howell, H.B., 163 Jud, K., 362
Hoy, K. L., 52
Hsu, E.C., 773
Huft, C., 274 K
Huneke, J.T., (581)
Hunston, D.L., 321, 339 Kaasteleyn, P.W., 13
Hyndman, D., 790 Kaelble, D.H., 40, 52, 64,
Hyodo, S., 275 201, 274, 301,
312, 363, (364 ) ,
(365), 391, 452,
I 537, 538
Kaganowicz, G., 521
Ibanez, J.L., 665 Kagawa, I., 748, 749
AUTHOR INDEX xxiii
Kahn, A., 640 Kniseley, R.N., 100, 101

Kalnin, M.M., 549 Knollman, G.G., 174, 193
Kamagata, K., 501 Kobatake, Y., 392
Kambe, H., 501 Kobayashi, H., 538
Kammer, H.W., 65 Kobayashi, T., 551
Kanazashi, M., 575 Kochwa, S., 824
Kaneko, M., 41 Koenig, J.L., 773
Kang, A.H., 844 Kohno, A., 275
Kanig, G., 39 Kojima, K., 452
Kanstad, S.O., 101 Kolotyrkin, V.M., 537
Kaplan, D., 338 Kolthoff, I.M., 640
Karasaki, H., (580) Kondo, T., 39, 40
Karasek, F.W., 100 Konieczo, M.B., 495
Karlivan, V.P., 549 Kopfle, J.T., 320
Kasemura, T., 15, 39, 40 Koral, J., 52, 640
Katagai, R., 41 Korshak, V.V., 75
Katchalski, E., 824 Kosaka, H., 501
Katchalsky, A., 748 Koutsky, J., 338
Kato, T., 729, 748, 749 Kovaleva, N.V., 640
Kato, Y., 495 Krasnov, A.P., 75
Kausch, H.H., 362 Kratky, 0., 854
Kawaguchi, M., 729, 748 Kraus, G., Jr., 452
Kay, G., 121 Kresheck, G.G., 825
Kayayama, H., 728 Krieger, I.M., 338
Keer, L. M., 410 Krimm, S., 790
Kelly, A., 472 Krotova, N.A., 74, 275
Kelly, M.A., 706 Krugers Dagneaux, P.G.L.G.,
Kendall, K., 274, 275 824
Kerchner, J., 122 Kubacki, R., 537
Khomutov, V.A., 75 Kuno, M., 748
Khopina, V.V., 640 Kurotsu, K., 41
Khramova, T., 626 Kutscha, D., 163
Kikuchi, K., 41 Kuzavkov, A.I., 539, 549, 550
Killmann, E., 640, 728, 747 Kuznezova, D., 626
Killon, M., 575 Kwei, T.K., 392
Kim, G.Y., 575
Kim, S.W., 823, 854
King, R.T., 538 L
Kinloch, A.J., 163, 274, 391,
Lal, M., 14, 626, 749
501, 773
Kirkbright, G.F., 101 Laland, M., 13
Lambert, J.M., 501
Kirkwood, J.G., 13 Lampert, M., 575
Kiselev, A.V., 640, 641 Landau, L.D., 75
Kitazaki, Y., 39, 40, 496 Landel, R.F., 452, 501
Klappratt, D.K., 122 Landolt, D., 121
Klein, F., 822 Langley, P.G., 501
Klotz, I.M., 825 Laplace, P.S., 13
Kloubek, J., 52 Lasoski, S.W., 452
Knibbs, R.W., 501 Lavrentev, V.V., 847
xxiv AUTHOR INDEX
Leblanc, J.L., 362 Mahanty, J., 4
Lecomber, P.G., 537 Mahoney, C.L., 122
Lee, E.S., 854 Malcolm, B.R., 854
Lee, L.H., 3, 4, 75, (77), MaIers, A.Y., 549
85, 86, 87, (195), Malvern, L.E., 298
(196), (197), (363), Manne, R., 122
549, (582), (684), Manning, G.S., 749
(858) Marceau, J.A., 122
Lee, R.G., 823 Margis, D., 275
Lee, R.H., 338 Margolis, L.B., 673
Lee, R.L., 854 Maricic, J., 824
Lee, S.H., 757, 773 Mark, H., 501
Lefever, R., 664 Markovitz, H., 338
Legrand, D.G., 39, 65 Markus, G., 854
Lehmann, S.L., 122 Maroudas, N.G., 673
Leo, F., (78), (79) Marsden, J.G., 773
Levi, D.W., 312, 495 Martin, R.J., 537
Levin, M.A., 75 Martin, T.W., 472
Lewis, G., 501 Maruchi, S., 52
Lifshits, E.M., 75 Masares, M., 122
Lin, C. J ., 122 , 162 Mason, J.H., 575
Lin, D.G., 549, 550 Mason, W.P., (198)
Lin, J.W., (795) Masters, L., (366)
Lin, J.W.P., 101 Masuoka, M., 371, 391, 392, 439,
Lipatov, Y.S., 549, 601, 626, (455)
627, 728, 756 Matduura, R., 749
Loeb, G.1., 824 Mathiew, H.J., 121
Logan, M.A., 845 Mathot, C., 854
Loginova, A.Y., 75 Matsumae, K., 495
Lohse, M.I., 361 Matsumoto, S., 748
Lorensen, L.G., (794) Matsumoto, T., 101, 128, (193)
Loshaek, S., 452 Matsunaga, M., 452
Lott, A.D., 561 Matsuzawa, T., 520
Low, M.J .D., 101 Matthies, D., 575
Ludwik, P.G., 162 Mattson, J.S., 854
Lundberg, J.L., 640 Matuura, R., 599
Lunsford, L.R., 420 Maugis, D., 203, 274
Lussow, R.O., 503, 520 McCafferty, E., 728
Lygin, V.I., 641 McClellan, A.L., 52
Lyklema, J., 823, 824, 825, 835 McClelland, J.F., 100, 101
Lyman, D.J., 823, 824, 835 McConnell, P., 640
McCrackin, F.L., 538, 640, 748
McDevitt, N.T., 103, 121, 122
M MoGinniss, V.D., 561
McGrath, M.J., 52
Mack or , E.L., 641 McKenna, G.B., 362
Macosko, C.W., 338 McLaren, A.D., 452, 823
MacQueen, R.J., 519 McMillin, C.R., 824
Mac Ritchie , F., 823, 835 Means, A.M., 537
AUTHOR INDEX xxv
Memgeres, G., 627 N

Meyer, E.F., 52
Meyer, F.A., 844 Nagao, M., 728
Michaels, A.S., 312 Nagasawa, M., 748, 749
Middlemiss, K.M., 14 Nagdi, P.M., 298
Mihairi, H., 393 Naghdi, P.M., 162
Milan, M., 673 Nagura, M., 362
Miles, M., 361 Nakamae, K., 707, 728, (793)
Miller, I.F., 823 Nakamura, T., 452
Miller, I.R., 748 Nakanishi, K., 773
Mitchell, D.J., 600 Nakao, K., 369, 371, 391, 392,
Mittal, K.L., 496, 503, 519, 439, 452, (455)
538, (580), (581), Naono, H., 728
Miyairi, H., 410, 411 Nasehzadeh, A., 825
Miyamoto, T., 626, 641 Natarajan, R.T., 790
Miyoshi, S., 728 Nate, K., 551, ( 580 ), ( 581 ) ,
Mizuhara, K., 640, 641, 728 (582)
Moddeman, W.E., 122 Needham, T.E., 52
Mofnigman, J.R., 122 Nemethy, G., 825
Mohandas, N., 682 Nenow, D., 362
Moniz, W.G., 320 Nesterov, A., 627
Mons, J., 627 Neuman, R.E., 845
Montoya, 0., 495 Neumann, A.W., 41
Moore, G.E., 392 Nicolis, G., 665
Moreno, R., 575 Nicolson, G.L., 664
Moriguchi, H., 501 Niinomi, M., 538
Morimoto, T., 728 Ninham, B.W., 4, 600
Moritz, A.G., 313 Nishikawa, A., 823
Morris, C.E.M., 313, 320, (365), Niu, T.F., 338
(366) Noda, 1., 749
Morrissey, B.W., 823, 824, 835 Noether, H.D., 362
Morton, M., 362 Nonaka, K., 728
Mosback, K., 823 Nonaka, Y., 452
Moscarello, M.A., 41 Nordal, P.E., 101
Moseona, A., 613 Norde, W., (684 ), (685), 801,
Mostafa, M.A., 52, 728, 756 822, 823, 824, 825,
Mostovoy, S., 162, 338 (855), (856 )
Motomura, K., 599, 149 Nordling, C., 122
Moynihan, R.E., 790 Nose, T., 40
Moysya, E., 626, 621 Numa, H., 362
Mukai, S., 128 Nyilas, E., 825
Muller, A., 391
Muller, K., 825
Muller, V.M., 276 o
Mullins, L., 362, 437
Muramatsu, A., 393, 411 O'Boyle, D.R., 362
Murch, S.A., 162, 298 O'Connor, D.M., 361
Murphy, M.C., 472 O'Sullivan, J.P., 122
Mutschler, L.E., 682, 835 Okamoto, H., 575
xxvi AUTHOR INDEX
Okawa, T., 728 Petersen, D.H., 163

Onsager, L., 13 Petrich, R.P., 274, 391, 452
Oreskes, J., 823 Petrokovets, M.I., 74
Origlio, G.F., 790 Peyser, P., 640, 728
Orofino, T.A., 748 Phillips, D.C., 472
Othman, A.B., 274 Phillips, M.C., 822, 823
Otsuki, K., 728 Philofshy, H.M., 575
Outwater, J.O., 472 Picot, C., 600, 627
Owen, M.J., 472 Piez, K.A., 844
Owens, D.K~, 40, 52, 64, 575 Pilz, 1., 854
Oya, M., 41 Pimental, G.C., 52
Plazek, D.J., 338,339
Pleskachevskii, Y.M., 75, 549
P Plueddemann, E.P., 773
Pluger, R., 391
Padday, J.F., 65 Pollock, H.M., 275
Paganelli, J., 122 Poranski, C.F., 320
Paghupathi, N., 338 Porter, A.W., 39
Palm, W., 854 Poste, B.E., 664
Palmberg, P.W., 121 Powers, W.J., 495
Pappillo, C.A., 362 Prager, W., 298
Pandalai, K., 338 Prescott, R., 472
Paolini, G., 122 Pressman, D., 854
Papahadjopoulos, G., 664 Priel, Z., 844
Park, I .K., 362 Prigogine, " I., 5, 13, 437, 665
Park, J.B., 362 Prosser, J.L., 519
Park, S.M., 362 Puchwein, G., 854
Parker, J.G., 101 Pye, E.K., 824
Parkinson, D., 362
Parodi, G.A., 101
Pascalf, J.V., 495 R
Pasechnik, N., 627
Passaglia, E., 538, 640, 747 Raj, R., 362
Patat, F., 640 Rao, C.H.R., 773
Patterson, D., 64 Raphael, C., 410
Pavinich, W., 362 Rappaport, C., 673
Pavlov, S.A., 75 Rastogi, A.K., 39, 40, 64
Pawel, R.E., 121 Ratner, B.D., (195), (196),
Pademonte, E., 362 673, 681, 691,
Pefferkorn, E., 747 706, (791), (792)
Peilman, M., 575 822
Penler, J.P., 121 Ratwant, M., 410
Peppiatt, N.A., 193, 410 Rauhut, H.W., 495
Peria, W.T., 538 Read, B.E., 790
Perkel, R., 640 Reeves, O.R., 673
Perkins, J., (365) Refojo, M.F., 706
Perleberg, C.R., 538 Regan, R.J., 706
Perzyna, P., 298 Reichard, H.S., 101
Petermann, J., 361 Reid, J.C., 338
Peters, D.A., 520 Reiner, M., 298
AUTHOR INDEX xxvii
Renieri, M.P., 162, 298 Sarma, G., 600

Renton, W.J., 163, 193, 299, Sarolea, L., 13
410 Sasaki, T., 673, 706
Rentzepis, P.M., 495 Sato, T., 640
Rice, J.R., 274 Savkin, v.G., 74
Rice, K.K., 122 Savkoor, A.R., 275
Richards, F.M., 824 Sawada, Y., 338
Richardson, J.A., 122 Schaeffer, B.E., 664
Rivlin, R.S., 275, 437 Scharpen, L.H., 706
Robb, LD., 641 Scheraga, H.A., 825
Roberts, A.D., 274, 275 Scheutjens, J.M.H.M., 823
Roberts, R.F., 519 Schich, M.J., 627, 641, 728
Robin, M.B., 100 Schildenecht, C.E., 501
Robinson, C., 41 Schlag, J., 575
Roder, T.M., 392 Schliebener, C., 640
Roe, R.J., 65, 599, 600, 629, Schonhorn, H., 392, 495, 519,
640, 749, 790, 823 575
Rohrer, J., (582) Schrader, W.H., 495
Rosen, J.J., 667 Schrag, J.L., 338
Rosencwaig, A., 86, 100, 101 Schulman, F., 65
Rosenfield, R.E., 824 Schultz, J., 274
Ross, KD.L., 537 Schulz, R.A., 825
Ross, R.C., 312 Schumaker, R.V., 854
Rossmann, K., 495, 575 Schwartz, H.S., 122
Rothen, A., 854 Schwarz, E., 854
Rothstein, E., 747 Schway, M.B., 667
Rowland, F., 747 Scigliano, J.M., 52
Rowland, T.J., 361 Scott, J.R., 501
Roylance, D.K., 362 Scriven, L.E., 664
Rubin, R.J., 599 Searingen, L.E., 41
Ryan, F.W., 392 Seiler, C.J., 452
Rytov, S.M., 75 Semenovich, G., 626, 627
Sergeeva, L.M., 626, 627, 728,
756
S Shafrin, E.G., 65
Shakutsui, H., 452
Sacchi, F., 122 Shapira, R., 823
Sadakne, G.S., 641 Shapiro, A.B., 824
Sakai, l., 728 Sharp, R. S ., (198 )
Samak, Q.M., 823 Sharpe, L.H., 391, 459
Samaras, N.N.T., 13 Shem, M., 538
Samueld, R.J., 790 Sherriff, M., 501
Sancaktar, E., 141, 163, 279, Shifrin, V., 627
299 Shinomiya, J., 41
Sandorfy, C., 773 Shunji, 0., 549
Sanfeld, A., 643, 664, 665, Siegbahn, K., 122
(683) Silberberg, A., 13, (77), (82),
Sapse, A.M., 65 (195), (363),
Saraga, L., 13 ( 455 ), 591, 599,
xxviii AUTHOR INDEX
600, 640, 641, Stefan, J., 501

(683), (684), Stein, R.S., 790
747, (791), Steinbach, A., 673
(792), (793), Steinberg, G., 728
(794), 823, Steinberg, I.Z., 824
(856), (858) Steinburg, H.L., 538
Simha, R., 627, 639, 640 Steinchen, A., 643, 664, 665
Singer, J.M., 822, 823 Steingiser, S., 472
Singer, S.J., 664 Steppo, R.F.T., 14, 626
Singleberry, C.R., 39 Sterman, S., 773
Siow, K.S., 64 Sternling, C.V., 664
Skolnick, J., 749 Sternstein, S.S., 338
Small, P. A., 52 Stevens, W.W., 338
Smilga, V.P., 74 Stewart, C.W., 65
Smith, C.A., 854 Stoddart, C.T.H., 519
Smith, L.E., 640, 728, 823, Stradal, M., 575
824, 835, 854 Strasser, H.J., 640
Smith, O.E., 641 Stromberg, R.R., 538, 640, 641,
Smith, P., 123 747, 823, 824,
Smith, R., 641 835, 854
Smith, T., 123, 139,312,537 Strong, J., 538
Smith, T.L., 392 Strongberg, R.R., 728
Smolders, C.A., 824 Struik, L.C.E., 339
Smolinsky, G., 537 Stuart, P.R., 519
Smurugov, V.A., 67, 75 Sudo, M., 40
Soderquist, M.E., 825 Sumiya, T., 728
Sojka, S.A., 320 Sung, C.S.P., 757, 773
Sokolniloff, I.S., 338 Sung, N.H., 757, (795)
Soloman, P.H., 773 Suzuki, K., 39, 40
Solomon, J.S., 103, 121, 122, Sviridyonok, A.I., 67, 74, 75
(196 ) Swalen, J.D., 519
Solomonoas, I.N., 641
Somoano, R.B., 100
Sonnonstine, T., 575 T
Sorensen, T.S., 664, 665
Sorokin, J., 847 Tabor, D., 75, 275, 277,
Sowell, R.R., 495 Tadros, T.F., 835
Spaeth, E.E., 682 Takahashi, A., 729, 748, 749,
Spar, I.L., 682, 835 (793), (794)
Spear, W.E., 537 Tanaka, T., 600, 640
Sprague, B.S., 362 Tanford, C., 824, 825
Springer, J., 162 Tanner, W.C., 312
Srere, P.A., 673 Telisman, Z., 823
St. Pierre, L.E., 39, 40 Teodorovich, E.V., 75
Standefer, J.C., 682 Ter Minassian-Sarada, L., 854
Stansbarger, D.L., 163 Tesapnk, A.K., 537
Starkweather, H.W., 392 Thies, C., 627, 641
Statton, W.O., 361, 362 728
Stedman, M., 748 Tholen, A.R., 275
AUTHOR INDEX xxix
Thomas, A.G., 275, 437 Valasek, J., 748

Thomas, C., 854 Validov, M.A., 75
Thomas, H.R., 706 Van Craen, J., 13
Thomas, J.R., 52 Van Der Linden, R., 563
Thomas, R.L., 472 Van Der Waals, J.H., 641
Thompson, G.E., 122 Van Lamsweerde-Gallez, D., 664
Tien, R.K., 537 Van Romunde, L.K.J., 822
Timmons, C.O., 173, 193, 338 Van Turnhout, J., 575
Tishkov, N.l., 550 Vanderscheer, A., 823, 824
Tobin, N.R., 362 Vanderjagt, D., 682
Tobolsky, A.V., 299 Varoqui, ., 747
Todosyichuk, T., 626, 627 Vedove, W.D., 643, 665
Tomoshige, S., 641 Veis, A., 845
Tompkins, H.G., 756 Velarde, M.G., 665
Toriyama, Y., 575 Vendery, R.B., 501
Torrie, G.M., 14 Vereshchagin, V.G., 75
Toyama, T., 501 Viehbock, F.P., 122
Tozaki, H., 39, 40 Vignes-Adler, M., 664
Trantina, G.G., 193 Viksne, V.A., 549
Trapesnikov, A., 854 Vincent, B., (793), 825
Tsai, S.W., 463, 472, (577), Vinogradov, A.V., 75
(579) Vinson, J.R., 163, 299, 410
Tschoegl, N.W., 338 Vogel, G.C., 52
Tsuge, T., 749 Von Frankenburg, C.A., 65
Tsuji, T., 439, (455) Voyutskii, S.S., 452
Tsutsumi, A., 41 Vroman, L., 682, 854
Tsyganok, V.N., 549 Vuk-Pavlovic, S., 824
Turpin, M., 626
Tutas, D.J., 640, 747
728 W
Wagner, R.E., 52
U Wakamatsu, T., 728
Wake, W.C., 501
Ueberreiter, K., 39, 41 Wakefield, T.D., 101
Uhlmann, D.R., 362 Walton, A.G., 824, 825, (856),
Ullman, R., 52, 640 (857)
Uniyal, S., 823 Ward, C.A., 706
Upadyayula, S.K., 362 Ward, E., 673
Uraki, H., 40 Warner, C., 854
Usmanova, D.O., 641 Wasserman, J., 824
Uvarov, A.V., 627, 641 Watatani, K., 728
Uy, K.C., 40 Watson, W.A., 682
Uzgiris, E.E., 825 Way, S., 274
Uzi, F., 39, 40 Weathersby, P.K., 681, 682, 706
Weber, R.E., 538
Weber, T.A., 600
v
Vail, W.J., 664
xxx AUTHOR INDEX
Wecker, S.M., 790 Yamamoto, F., 495
Wegman, R.F., 312 Yamashita, N., 39, 40
Wehner, G.K., 122 Yamashita, S., 728
Wehs, E.E., 321 Yamori, S., 749
Weinberg, Z., 575 Yanazawa, H., 520
Weisman, Z., 844 Yavorsky, P.M., 338
WeIgand, H., 728 Yeow, Y.T., 299
Wells, E.E., 339 Yoshida, T., 640
Wen, W. Y., 825 Young, T., 13
Wendt, R.C., 40, 52, 64
Werme, L.O., 122
Westerdahl, C.A.L., 495 Z
Wetzel, F.M., 501 Zapas, L.J., 338
White, J.L., 641 Zbinden, R., 790
White, J.R., 275 Zettlemoyer, A.C., 728
White, L.R., 600 Zheludev, I., 575
White, M., 537 Zisman, W.A., 52, 64, 65, 501
Whitehouse, R.S., 501
Whitney, J.M., 163
Whitney, W., 362
Whittington, S.G., 14
Wielogorski, J.W., 706
Wightman, J.P., 162
Wilcox, D.L., 538
Williams, J.G., 193
Williams, M.L., 452, 453, 501
Winn, M.L., 854
Winters, H.F., 538
Wisniewski, S., 854
Wittberg, T.N., 122
Woermann, D., 627
Wolman, M., 822
Wood, G.C., 122
Wood, S.J., 641
Wool, R.P., 341, 361, 362
Wronski, C.R., 537
Wu, S., 39, 40, 52, 53, 64,
65, (80), (81), (82)
Wurstein, F., 575
Wyckoff, H.W., 824
Wydeven, T., 537
Yamaguchi, Y., 413, 420

Yamakawa, H., 627
Yamakawa, S., 471, 495, (579),
(580), (791)
SUBJECT INDEX
(Pp 1-458 are found in Volume 12A, Pp 459-866 in Volume 12B)
bonding, 473
bulk properties, 141
Abrasion characteristics, 551
of specimen, 625 debonding, 500
Absorption spectra, 68 failure, 123, 124
Acid-base interaction, 43, 78 failure energy, 143
Acrylate interaction in polymer, 67
oligomer, 558 interface, 175
resins, 582 metal structure, 301
Activation energy, 321 penetration, 113
Adaptation structural reliability, 301
conformational-, 591 tack, 497
Additivity law, 21 Adhesive-adherend
Adherence by Auger spectroscopy, 103
of polyurethane, 204 by ESCA, 103
of punch, 204 interface, 103, 124, 165
of spheres, 204, 236 Adhesive bond
Adherend by dielectric relaxation
aluminum, 165 gradient, 165
bondline thickness, 144 Adhesive joint
shear stress, 144 FRP, 393
surface, 758 tensile, 371
surface preparation, 142 Adsorbance
thickness, 144 of NaBr, 738
Adhesion of train, 7.44
cell, 667, 681 ratio, 639
measurements, 503 Adsorbed
mechanism, 491 monolayer, 751
of filled PE, 539 trains, 592
of polymer, 97 Adsorbent
of thin film, 521 ammonium chloride, 618
photoacoustic spectro- solution ratio, 602
scopy, 87 Adsorption
Adhesive affinity, 639
xxxi
xxxii SUBJECT INDEX
degree of, 605 Angle-resolved

degree of selectivity, 629 photoemission, 797
enthalpy of, 816 Anisotropic shape
kinetics, 736 effect, 390
of albumin, 684 Anodic oxide, 106
of dyes, 87 Antioxidant, 540
of gamma-globulin, 816 Anti"reflection
of hemoglobin, 677 coating, 521
of plasma, 681 Arrhenius
of polymer mixture, 618 rate constant, 355
of polymer on metal relation, 166
oxide, 707 Aspect ratio, 372
on polystyrene, 684 Athermal solution, 11
preferential, 638 ATP, 644
selective, 629 Auger
Adsorption isotherm, 724, 736 electron spectroscopy, 103,
binary, 635 129, 521, 528
of protein, 804 thin film analyzer, 528
of water, 710 Average absorption, 780
plateau, 814 Azide
Aerosil, 603 as buffer, 685
Ageing
of adhesive, 316
Aggregate B
fractionation of, 610
Aggregative Barry's representation, 234
character of Bending moment, 393, 419
adsorption, 602, 603, 608 distribution, 401
Albumin, 807 Benzene, 714
crosslinked, 809 Benzil, 553
radiolabeled, 804 Benzoin, 553
Alkyl diphenyls, 17 isopropyl ether, 553
Alkyl naphthalenes, 17 Benzophenone, 553
Alpha-metal free Beta-counter, 829
phthalocyanine, 93 Beta-ray scattering, 123
Aluminum, 105, 301 Beta-relaxation, 321, 335
alloy, 105., 301 Bimolecular
-aluminum oxide, 111 inhibition, 659
-epoxy system, 188 Bingham model
-ethylene vinyl mOdified, 281
acetate, 372 Biocompatibility, 677, 801
film (anodized), 106 Biomaterial, 827
-PE joint, 544 Biomembrane, 643
substrate, 751 Biopolymeric layer, 646
Aluminum oxide, 757 Bisphenol
as filler, 539 diglycidyl ether, 331
plate, 553 Blister method, 505
Aminosilane, 757 Block copolymer, 28, 341, 350
Y-Aminopropyltriethoxysilane, Blood
759
SUBJECT INDEX xxxiii
platelet adhesion, 831 membrane, 663

Boltzmanns constant, 45, 166, mobility, 661
355, 144 model, 8
Bond Cellulose fiber, 91
durability, 491 Chain
exposure time, 301 conformation, 612
jOint failure, 116 length, 633
strength, 131 molecule, 10
Bondability, 413 segment, 591
Bonding force, 70 Chandrasekhar principle, 594
Boundary layer Chaotropic ion, 816
structure of, 601 Characterization
Bovine of adsorbed polymer, 701
albumin, 678 of hydrophilic-hydrophobic
gamma-globulin, 828 polymer, 691
pancreas Charge
ribonuclease density, 563, 513, 811
(RNase), 802 distribution, 810
Brittle repulsion, 834
failure, 466 Chemical bonding, 118
fracture, 124 Chemisorption
Broadline of acetic acid, 752
NMR, 775, 185 Chip-coat, 582
Bromocresol, 618 Chlorinated
Brjtl1sted acid, 44 polyvinyl chloride, 49
Brush-like failure, 466 Chondroitin sulfate, 838
Bulk shear, 150 Chopping frequency, 89
Butt joint, 312 Chromatogram, 311
coefficient, 386 Chromium
oxide, 809
plate, 193
C Chromophores, 195
Circular dichroism, 35
Cadmium telluride Circumferential stress, 464
detector, 752 Coating glass, 691
Calcium carbonate, 49 Cobalt-60
Calf serum, 661 radiation, 618
Carbon black, 352 Cohesive
Carbon tetrachloride, 114 energy density, 25, 602
Carboxylic acid failure, 123, 124
absorption of, 542 force, 442
CASING, 414, 565 Coil conformation, 596
Castigliano theorem, 215, 237, Collagen, 831
261 fiber, 838
Cathetometer, 16 multimer, 837
Cationic polymer, 118 soluble molecule, 858
Cavitation, 645 Collagenase, 839
Cell Complementary
adhesion, 661 energy, 214
deformation, 643 Complex
xxxiv SUBJECT INDEX
shear modulus, 323 opalescence phenomena, 609

Compliance, 239 strain rate, 316
Composite strength, 463 stress, 134, 304
Computer maping, 126 stress intensity factor, 119
Concentration surface tension, 53
gradient, 636 wavenumber, 662
of damage, 354 Crochet's
of polymer segment, 631 delayed yield equation, 283
Conformational Crosshead
change, 841 displacement, 212
constraint, 591 velocity, 204
entropy, 634 Cross lap joint, 313
transition, 35 Crosslink density, 151, 195
Conservation Crosslinking, 564, 168
equation, 648 reactoin, 551, 551
Constitive equation, 281 Cross-section
Contact angle, 25, 55 circular, 330, 331
Contact area, 212, 218 rectangular, 330, 331
Contrast microscopy, 693 Crystalline band
Cooperative of PTFE, 118
trans con formation , 641 Crystallinity
Cooperativity constant, 641 index of, 183
Copper Cyclohexane
as filler, 539 for polycarbonate, 603
Corona discharge, 563 solution, 110
Corresponding states, theory, 10 Cytoplasmic spreading, 611
Corrosion
inhibiting primer, 115
Coulomb interaction, 805 D
Counter ions , 805
Coupling Debye length, 144
agent, 194, 195 Deformation
relation, 650 field, 363
Covalenties, 841 of adherend, 393
Crack ratio, 118
extension, 210, 213, 468 velocity, 218
healing of polymer, 341 Degradation, 564
initiation, 304 Degree
length, 145, 428, 468 of crystallinity, 311, 314
tip, 131 of hydrolysis, 418
velocity, 216, 234, 260 Dehydration, 102
Crack propagation, 144, 211 Density
kinetics of, 204 of gas, 18
with fixed cross-head of liquid, 18
velocity, 260 of packing, 540
with fixed grip, 260 Deprotonation, 801
Creep stress, 283 Depth of penetration, 612
Critical Desorption, 122
deformation velocity, 316 from hydrophobic
fiber length, 410 surface, 803
SUBJECT INDEX xxxv
Dextran energy of adhesion, 208

crosslinked, 685
Dichroic orientation
function, 780 E
Dicyandiamide, 313
Dielectric Eastman 910
constant, 10, 745, 811 adhesive, 146
depolarization spectro- Egg albumin, 37
scopy, 165 Elastic
layer, 521 deformation, 222
permeability, 68, 70 displacement, 239
relaxation of energy, 215, 222l 239, 263
epoxy, 165 modulus, 145, 21~, 399
Differential shear strain rate, 149
IR spectroscopy, 809 wave, 181
scanning calorimeter, 552 Electret formation, 566
Diffusion Electric field, 811
convection, 645 d.c., 166
demixing, 349 overlap, 817, 818
migration, 647 polarizing, 189
Dimer, 11 Electrical
Dipole, 165 double layer, 811
-dipole interaction, 45 Electrokinetic
moment, 45, 167 phenomena, 171
relaxation, thermally potential, 805
stimulated, 175 Electrolyte technique, 580
Dispersion equation, 651 Electromagnetic field, 68
Dispersion force, 45 Electron
Displacement gun, 107
equation, 651 micrograph, 319, 521
factor, 216 microscopy, 498, 780
transducer, 327 probe microanalysis, 144
Dissipative Electroneutrality, 811
energy, 433 Electrophoresis, 807
function, 229 Electrophoretic analysis, 828
Dividing surface, 7 Electrostatic repulsion, 742
Donnan Ellipsoid
equilibrium, 733 end-on, 831
exclusion, 738, 745 side-on, 831
Donor-acceptor Ellipsometer, 829
complex, 536 Ellipsometry, 521, 524, 729,
interaction, 43 848, 856
Double lap joint, 144 of protein, 809
Drago relation, 44 End-to-end distance, 742
Draw direction, 776 Energy
Ductile balance theory, 204
brittle transition, 150 dispersive analysis, 157
fracture, 372, 387 of adsorption, 814
Du Nouy ring, 703, 793 of rupture, 436
Dupre's Enthalpy, 209, 224
xxxvi SUBJECT INDEX
Entropy, 206 Fiber

conformational, 630 debonding, 466
of mixing, 630 efficiency, 470
Epoxy -matrix bond, 466
adhesive, 106, 141 -matrix debond energy, 469
~luminum bond, 175 -matrix failure, 150
equivalent weight, 316, 320 reinforced
phenolic adhesive, 301 plastics (FRP), 393
polyamide adhesive, 417 stiffness, 150
resin, 313, 618, 621 volume, 464
rubber toughened, 157 Fibrillar texture
tetraglycidyl in polypropylene, 783
diaminomethane, 366 Fibrinogen, 685, 687, 807
thermosetting, 322 on polyethylene, 857
Epoxylite, 191 Fickian diffusion, 646
Equation Filled elastomer, 341
of state, 55 Film
Equilibrium theory deposition, 522
of crack, 425 plasma-polymerized, 524
of peeling, 426 thickness, 857
ESCA, 509, 693 thickness monitor, 109
Etchant, 511 Fixation, 647
Ethyl methacrylate, 191 Flory
Ethylene equation, 742
acrylic acid copolymer, 79 -Huggins term, 632, 744
glycol condensate, 17 -Huggins theory, 11
oxide-propylene oxide Fluorescence
copolymer, 27 spectroscopy, 809
propylene copolymer, 22 Forced oscillation, 323
vinyl acetate, 27, 29 Fourier transform
Evaporative analyzer, 314 IR spectrometer, 752, 759, 760
Excluded volume, 10, 598 IR spectroscopy, 752, 757
function, 745 Fowkes equation, 54
Expansion factor Fractograph, 124
of loop, 742 Fractography, 152, 371, 377
External Fracture
force, 429 criteria of peeling, 421
polarization, 566 energy, 148, 431
reflection IR mechancis, 203
spectroscopy, 751 mode, 150
Extinction coefficient, 524, pattern, 408
751, 754, 755, 761, 778 stress, 424
toughness measurement, 176
Free energy
F criterion, 421
of deformation, 428
Failure of interaction, 632
envelop, 375 Free volume, 22
mode, 165, 166, 463 Friction
SUBJECT INDEX xxxvii
of irradiated film, 73 H
stress, 469
Hamaker constant, 44
Hard and soft acids and
G bases, 79
Harmonic mean equation, 53
Gadolinium, 146 Healing
Gamma-counter, 829 line mode, 354
Gamma-globulin, 807, 857 Heat
Gauss theorem, 811 of mixing, 46
Geiger Mueller Helical content, 36
counter, 126 Helium
Gelatin, 91 neon laser, 829
Geometric plasma, 473
mean equation, 53 Helmholz
mean expression, 45 free energy, 207
parameter, 217 plane, 805, 818
Gibbs Hemostasis, 837
adsorption formula, 7 Hertz
-Duhem equation, 7 load, 246
energy, 814 problem, 236
free energy, 209 Hertzian contact, 245
surface, 5 Hooke's Law, 149, 287
Glass, 758 Huggin's
mat, 394 coefficient, 840
-polymer joint, 544 constant, 843
roving, 394 Human
-rubber contact, 245 plasma albumin, 802
support, 314 serum albumin, 827
Glass temperature, 19, 217, 335 Humidity
of adsorbed layer, 610 absorption, 579
of UV resins, 555 Hydrodynamic
Globular protein, 801 boundary condition, 650
Glueline inhomogeneities, 146 equation, 683
Gold-palladium alloy technique, 730
film, 148 velocity, 644
Grafted Hydrogen
polyethylene, 791 abstraction, 553
silicone, 791 bonding, 25, 45, 565, 710,794
Grafting ion transfer, 817
radiation induced, 474 Hydrolysis
Graphite epoxy of silane monomer, 716
composite, 464, 577, 579 polymerization, 767, 795
Grating spectrometer, 778 Hydrophile
Griffith -lyophile balance, 26
criterion, 207, 209, 247 Hydrophilic
-Irwin theory, 145 -hydrophobic interaction, 668
theory, 423 surface, 667, 851
Hydrophilicity, 857
xxxviii SUBJECT INDEX
Hydrophobic Interlaminar
bond, 804 shear strength, 470
dehydration, 815 Intermoleculal'
interaction, 817 interaction, 16, 70
surface, 827 Internal
Hydrophobicity energy, 225
of protein, 804 slip, 788
of sorbent surface, 803 stress, 561
Hydroxyproline, 838 Interphase, 104
2-Hydroxyethyl Intrinsic viscosity, 714, 840,
methacrylate, 668, 677, 691 843
Iodine, 677, 827
Ion
I beam etch, 105
co-adsorption, 856
Immobilization of migration, 509
protein, 801 pair, 817
Immunoglobulin, 681, 847 Ionic strength, 722, 729, 847,
Incident 851
sound speed, 185 Iosipescu specimen, 180
Inconel, 528 Iron oxide, 758
Inelastic electron tunneling, 752 localization of, 707
Infrared Irwin relation, 271
dichroism, 775 Isoelectric point, 37, 805, 853,
spectroscopy, 521, 525, 541, 856
567, 617 Isotropic material, 177
Inorganic filler, 539
Instron test machine, 147, 522
Intensity J
of molecular interaction,
ratio, 72, 766 Joint strength, 150
Interaction parameter, 55, 132
Intercellular
communication, 668 K
Interface
energy, 207 Kaolin
roll of, 467 as filler, 539
Interfacial Kelvin-Zisman probe, 574
accomodation zone, 167 Kinetics
bond failure, 371 of adsorption, 677
failure, 126, 446 of cell adhesion, 667
fracture of joint, 371 of detachment, 204
free energy, 704, 855 of healing of polymer, 353
strength, 466 Kramers-Kronig
Interfacial tension, 6, 16 relation, 68
of polymer melt, 15
of polymer solution, 15
time-dependence, 792
water-oil, 707
SUBJECT INDEX xxxix
L Loop, 593
partition function, 743
Labeled protein, 677 -train-conformation, 744
Lag-phase, 673 -train-structure, 742
Lagrangian -train-tail conformation,
multiplier, 595, 633 729
Lamellar crystal, 780 Lorenz-Lorentz
Laminate, 577 equa tion, 732
Langmuir Loss factor, 218
adsorption, 829 Ludwik's equation, 149, 283
equation, 804 Lyophilicity, 34
isotherm, 638
Lap
joint, 414 M
joint deformation, 148
length, 395 Macleod's exponent, 18
shear joint, 141, 302 Macromolecule
shear strength, 124 nonadsorbed, 593
·shear test, 475 Macrophage, 643
Laplace Magnesium
equation, 6 X-ray source, 107
operator, 649 Magnetic
transform, 287 recording tape, 724
Lateral stress, 645 Matrix
Lauryl methacrylate, 555 -primer interfacial
Legendre failure, 152
theorem, 244 Maximum
transformation, 207 strain, 346
Length stress, 421
of train, 593 Maxwell
Limit model, 282
of stability, 212 relations, 209
Linear Mechanical
elasticity, 205 adhesion, 580
electron accelerator, 475 loss, 624
viscoelasticity, 281 properties gradient, 621
Lipid recovery, 345, 354
of membrane, 644 Melt index, 540
Liquid Mercaptoethanol, 851
crystal, 36 Mercury lamp, 552
state, 8 Metal
Locus carboxylate, 754
of failure, 127, 143, oxide interface, 112, 196
165, 482 Methacrylate
London dispersion resins, 582
force, 43 Methyl acrylate, 473
Longitudinal graft, 474
shear, 464 2-Methylanthraquinone,
strength, 464 553, 561
xl SUBJECT INDEX
Methyl group N
effect, 22
Methyl violet 2B, 91 Navier-Stokes
Methylol acrylamide equation, 644
graft, 474 Necking, 422
Michaelis-Menten Neumann equation, 6
desorption, 652 Neutron
Microvoid flux, 146
formation, 341 radiography, 141, 146
Migration Newtonian
site of, 514 fluid, 644
Minimum entropy liquid, 498
principle of, 434, 456 Nitrogen
Mises hypothesis, 371 reaction on surface, 521
Mixing condensation Non-crystalline phase
critical-, 609 PTFE, 775
Model biomembranes, 643 Non-linear
Modified epoxy, 126 viscoelastic deformation,
Moffitt's equation, 35 292
Moisture degradation, 310 Normal stress, 216
Molar Notched
enthalpy, 816 strength, 468
refraction, 732 Novolac resin, 167, 177
Molecular Nuclear magnetic resonance,
forces in adhesive, 67 high resolution, 603
mobility of adsorbed Nylon-12, 374
layer, 607
rearrangement, 855
Molecular weight o
critical-, 448
distribution, 440, 630 Optical
number average, 21, 440 absorption, 88
of polymer, 683 coating, 521
viscosity average, 440, density, 541, 766, 848
455 rotary dispersion, 35
weight average, 440 Organic-metal
Monochromator, 778 interface, 751
Monolayer Orientation function, 778
adsorption isotherm, 852 Oscillation regime, 658
Monosulphide Oxidation, 565
as antioxidant, 541 induction period, 546
Monte Carlo method, 11 Oxide
Multicomponent system, 601 etch rate, 509
Multilayer film, 97 thickness, 137
Multiple internal reflection, Ozonisation, 565
759
Muscle-like
contraction, 644 P
Packing density, 602, 617

SUBJECT INDEX xli
Packing perfection Platelet

degree of, 617 adhesion, 827
Papain, 848 aggregation, 837
Parachor, 18, 78 baboon, 792
Partition function, 12, 594 human, 792
Peel -platelet contact, 838
rate, 439 Platinum surface, 729
test, 475, 505 Poisson's ratio, 181, 221,
Peeling, 263 283
angle, 429 Polarity
bend energy, 263 of polymer, 712
force, 455 Polarizability
initiation of, 421, 430 of molecule, 10, 621
Peel Strength (T), 439 Polarized radiation, 776
of PE-epoxy, 473 Polarizer, 778
superposition, 444 Polyacrylamide, 792
Wet, 482 gel, 685
Penetration depth, 769 Polyacrylates, 25
Perfect solution, 9 Polyacrylic acid, 751, 794
Perfluorodecanoic acid, 54 Poly-dl-alanine, 37
Permittivity of medium, 811 Poly-alpha-methylstyrene, 17
Phagocytosis, 643 Polybutadiene, 718
Phase ageing, 365
angle, 329 carboxylate rubber, 618
reversal, 185 1,2-Polybutadiene
Phosphate buffer 841 dimethacrylate, 551
Phosphoric acid, 109 Polycaproamide, 70
Photoacoustic Polycarbonate, 91, 603
spectrometer, 89 Polychloroprene, 499
spectroscopy, 87, 196 Polyelectrolyte, 729
Photodegradative adsorption of, 794
etching, 486 polyester, 91
Photoelectron spectroscopy, soft, 393
103 hard, 393
Photoelasticity, 218 Polyethylene, 22, 56, 59, 70
Photoinitiator, 552 adhesion of, 539
Photopolymerization, 553 graft on, 677, 692
Photoresist, 503 high density, 540
coater, 668, 693 linear, 828
KTFR, 510, 580 low density, 440, 475
negative, 518 medium density, 475
positive, 510 oxidation of, 541
process, 516 surface treatment of, 563
Photovoltaic cell, 522 Polyethylene glycol, 21
Phthalic acid, 109 Polyethylene oxide
Piperidine, 331 oligomer, 375
Plasma, 685 Polyethylene terphthalate,
protein, 827, 838 54, 70
reactor, 475 Poly-2-ethylhexyl acrylate,
Plateau adsorbance, 821 25
xlii SUBJECT INDEX
Polyethyl methacrylate, 667, interfacial tension, 716

677 Polyvinyl alkylate, 25
Poly-gamma-benzyl-l-glutamate, Polyvinyl caprylate, 25
35 Polyvinyl chloride
Polyhydroxyethyl methacrylate, 1nconel interface, 528
667 metal, 529
Polyhydroxyproline, 838 Porter's equation, 16
Polyion, 729 Post-treatment, 569
coil, 742 Potassium dichromate, 93
Polyisobutylene, 439 Practical adhesion, 475, 504
fractions, 441, 455 Preferential localization, 712
Polyisobutylene glycol, 22 Prestrain, 266
Polyisoprene, 499 Prestress, 266
Poly-L-lysine, 37 Pressure, 216
Polymer sensitive adhesive, 497
chain flexibility, 602 Profile analysis, 105
melt, 30 Profilometry, 693, 697
on solid surfaces, 601 Proline, 838
solution, 33 Propylene glycol
solvation, 714 condensate, 17
solvent interaction, 610 Protein
Polymethyl methacrylate, 49, adsorption, 677, 801, 855
70, 91 dissociation, 794
film, 612 globular, 856
isotactic, 632 1-125 labeled, 685
solution, 631 iodination, 678
syndiotactic, 632 sorbent contact, 817
Polypeptide Pull test, 505, 534
chain, 804
fragmen t , 829
solution, 35
Polyproline, 838
Polypropylene, 22, 539 Quasi-isotropic
isotactic, 540 elastic moduli, 577
Polypropylene glycol, 21 strength, 578
Polysiloxane network, 757 Quasi-lattice model, 592
Polystyrene, 30, 91 Quasistatic force
in DeE, 603 of adherence, 203, 212,
interfacial tension, 716 240, 265
latices, 802 of punch, 234
solution, 631
Polytetrafluoroethylene, 56,
59, 70 R
grafted surface, 476
morphology, 775 Radial stress, 464
orientation, 775 Radius
semicrystalline, 780 of contact (actural), 239
Polyurethane, 218 of gyration, 714, 742
Polyvinyl acetate Rail
SUBJECT INDEX xliii
shear test, 146 conditions, 656

test, 363 Rubber friction, 218
Random walk
two choice, 594
Randomization S
of fiber orientation, 577
Rate Sapphire, 757
of change of damage, 344 Saturation
of change of strain energy, magnetic flux, 725
344 point, 743
of dissipation, 344 problem of PAS, 88
of healing, 344 Scanning electron
of working, 344 microscope, 106
Rayleigh microscopy, 141, 143, 693
angle, 185 Scarf
stability of jet, 683 angle, 405
-Taylor problem, 651 joint, 177
Reactive monomer, 551 Scintillation
methacrylate, 553 counting, 667
Recovery half life, 356 Scratch test, 506, 533
Reduced parameter, 217 Screen printing, 582
Reflecting microscope, 375 Second viral
Reflectivity, 89 coefficient, 745
Refractive index, 524, 612, Secondary ion mass
732, 794 spectra, 117
complex, 732 spectrometry, 758
of saphire, 759 Secular equation, 660
Regular solution, 9 Segmental mobility, 623
Relative humidity, 221 Selective adsorption, 27, 77,
Relaxation time, 166, 283
Resin 619, 633
Semi crystalline
-metal interface, 143 polymer, 341
-rubber binary mixture, Serological test, 801
498 Serum albumin, 837, 839
Reverse osmosis Serum-free media, 669
membrane, 521 Sessile drop method, 15
Rheovibron, 169
Rhodamine B, 97 Shaft-angle
Rigid punch encoder, 327
adherence, 221 Shear
Rigidity distribution, 144
of joint, 396 modulus, 172, 185, 624
Rolling resistance speed, 172
of glass cylinder, 268 strain rate, 364
Root mean square distance, strength, 150, 396
742 stress, 372, 413, 464
Rotational stress (maximum), 467
wave speed, 185
freedom, 820
mobility, 817 (metal), 198
Routh-Hurwitz stability Shift factor, 335, 379
xliv SUBJECT INDEX
Short fiber of surface tension, 5

composite, 577 Stearic acid
strength, 578 contamination, 129
Silane, 758 Steel
coupling agent, 557, 757 Nylon-12-steel, 381
primer, 115 Steric stabilization, 793
Silica, 49, 758, 794 Stern
adsorption on, 630 layer, 805
gel as filler, 539 potential, 805
surface, 581 Stiffness, 150, 210, 497, 577
Silicon oxide, 510, 511, 831 specific, 578
hydrophilic, 833 Stored
pretreatment, 515 elastic energy, 345
photoresist interface, 514 Strain
plate, 848 energy release rate, 203,
Silicone 208, 239, 250
dimethacrylate, 558 rate, 150, 282
rubber (graft), 691 Stress
Single lap joint, 394 concentration factor, 413
Sliding distribution, 372, 413, 417
friction, 281 distribution of punch, 270
speed, 72 elongation diagram, 396
Slip-stick intensity approach, 216
failure, 443 intensity factor, 204,
Small angle 216, 270
X-ray scattering relaxation, 778
(SAXS), 342 softening effect, 351
Sodium bromide, 730 -strain response, 291
Sodium chloride Strontium Chromate, 117
solution, 847 Structure
Sodium polyacrylate, 129 of adsorbed protein, 808
Solubility parameter, 45, 709 Structural adhesive, 177
Solvent power, 714 failure characterization,
Sorbent surface 141
degradation, 819 viscoelastic characteriza-
Span-to-depth ratio, 467
tion, 321
Specific
refraction, 616 viscoelastic shear
volume, 19 behavior, 279
Spontaneous rupture, 245 Structural rearrangement
Spreading pressure, 55 in protein, 807
Sputter ion gun, 107 Structurization
Sputtering degree of, 608
rate, 111 styrene, 521
time, 111 -butadiene rubber, 499
Squareness ratio -butadiene-styrene block,
of magnetization, 724 copolymer, 342
Stability of equilibrium, 210 -dimethylaminoethyl
Statistical mechanics methacrylate copolymer,
of adsorption, 5 718
SUBJECT INDEX xlv
-maleic anhydride adhesive, 497

copolymer, 34 Tack force, 498
Sudgen's equation, 15 Tackification, 499
Surface Tackiness, 233
ageing, 301 Tacticity
analysis, 148 of polymer, 22
area, 852 Talc
charge density, 818 as filler, 539
contamination, 133 Talysurf profile, 148
elasticity, 684 Tape test, 505
energetics of metal Tapered
structures, 301 double cantilever beam,
energy, 77, 301, 567 177
entropy, 77 Tearing energy, 436
generation mode, 185 Teflon plate, 17
hydrophilicity, 692 Temperature, 206
instability, 643 post bake, 516
modification of PE, 473 Tensile
potential, 574, 807 delayed failure, 285
pressure, 847 force, 263
roughness, 143, 705, 732 modulus, 374
site fraction, 638 stress, 424
topography, 705 stress (maximum), 467
turbulence, 643 tester, 441
viscosity, 684 Tensile strength, 372
wettability, 563, 567 of UV resins, 552
Surface free energy, 423, 428, reduced, 379, 389
521 Tetrafluoroethylene
dispersive component, 531 tensile deformation, 788
measurement, 529 Tetrahydrofuran
polar component, 531 -ionen block copolymer, 730
Surface tension, 15, 16, 644, -propylene oxide copolymer,
709 27
dispersion component, 54 Tetramethylbenzidine, 678
M.W. dependence, 17 Texture transition, 775
of polymer melts, 15 Thermal
of polymer solutions, 15 damping constant, 89
of solid, 53 diffusion length, 88
polar component, 54 diffusivity, 89
statistical mechanics, 5 stimulated current, 197
Surfactant, 28 stimulated discharge, 574
Symmetric Thermodynamic potential, 203,
lap shear, 286 207, 423
rail shear, 286 Thermosetting resin, 551
Theta
point, 742
T solvent, 608
Thick film of A-1100, 765
Tack Thickness
xlvi SUBJECT INDEX
dimension of adhesive, 132 Tyrode, 839, 840

of adhesive layer, 431
of adsorbed layer, 738
of oxide film, 509 U
of polymeric layer, 540
of silane film, 761 Ubellhode viscometer, 839
sample, 88 Ultrasonic
Thin film analyzer, 107 Rayleigh wave measurements,
Thrombus, 827, 838 175, 181
formation, 792, 831 Ultraviolet
T~umb tack test, 498 radiation, 551
H -Thymidine, 667 -visible absorption, 97
Time-temperature Undercutting constant, 509
superposition, 321, 322, Uniaxial
359, 375, 371 deformation, 780
Titanium tension, 287
alloy, 308 Un-notched strength, 468
dioxide, 794
Toluene, 714
Torsion bar, 323 v
Torsional
elastic constant, 323 Van der Waals
elasticity, 329 forces, 67, 204, 343,
Total internal 535
reflection (ATIR), 612 interactions, 165, 802,
Toughness 817
of adhesive, 150 Van Laar
Transconformation, 646 mixing energy, 632
Transcrystalline Vertical displacement, 221
region, 383 Vinyl
Transferred ions, 817 acetate grafts, 474
Translational entropy, 803 triethoxysilane, 759
Transmission spectrum, 753 Viscoelastic
Transverse losses, 203
failure strain, 578 material, 375
tension, 464 plastic theory, 296
Tresca shear behavior, 279
failure criteria, 149 solid, 218
hypothesis, 372 surface, 237
Trialkoxysilane, 758 Viscoelasticity, 150
Trichloroethylene, 145, 714 Viscoplasticity, 280
Triglycidyl p-aminophenol, 315 Viscosity coefficient, 282
Tris Void, 154
as buffer, 839, 840 Volume
Tropocollagen, 838 contraction, 557
Trousers-type fraction, 577
specimen, 421
Trypsin
-ETDA, 669
SUBJECT INDEX xlvii
Water
absorption, 482, 557
polymer solution, 716
Weak boundary layer, 372, 565
Weissenberg
Rheogoniometer, 321, 499
Wenzel's relation, 246
Wettability, 474
Welhelmy method, 793, 848
WLF
equation, 144, 439, 448,
455, 499
relationship, 333, 334
shift factor, 217
Work
of adhesion, 32, 46, 203,
204, 217, 252, 303,
425, 500, 510
of adsorption, 500
of fracture, 144, 468
x
X-ray
beam (monochromatized),
693
diffraction (wide angle),
775
microanalyzer, 373
photoelectron spectro-
meter, 107
Yield
strength, 372
stress, 149, 280
Young's modulus, 203, 282,
374, 424, 429, 557, 577
Zisman's
critical surface tension, 53

Adhesion and Adsorption of Polymers Compress

Uploaded by

Copyright:

Available Formats

Adhesion and Adsorption of Polymers Compress

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Adhesion and Adsorption of Polymers Compress

Uploaded by

Copyright:

Available Formats

POLYMER SCIENCE AND TECHNOLOGY

William J. Bailey C. A. J. Hoeve

Volume 1 • STRUCTURE AND PROPERTI ES OF POLYMER FI LMS

Volume 2 • WATER-SOLUBLE POLYMERS

Volume 3 • POLYMERS AND ECOLOGICAL PROBLEMS

Volume 4 • RECENT ADVANCES IN POLYMER BLENDS, GRAFTS, AND BLOCKS

Volume 5 • ADVANCES IN POLYMER FRICTION AND WEAR (Parts A and BI

Volume 6 • PERMEABILITY OF PLASTIC FILMS AND COATINGS

Volume 7 • BIOMEDICAL APPLICATIONS OF POLYMERS

Volume 8 • POLYMERS IN MEDICINE AND SURGERY

Volume 9 • ADHESION SCIENCE AND TECHNOLOGY (Parts A and BI

Volume 10 • POLYMER ALLOYS: Blends, Blocks, Grafts, and Interpenetrating Networks

Volume 12. ADHESION AND ADSORPTION OF POLYMERS (Parts A and BI

PLENUM PRESS· NEW YORK AND LONDON

International Conference on Adhesion and Adsorption of Polymers, Honolulu, 1979.

(Polymer science and technology; v. 12A-B)

First half of the proceedings of an International Conference on

© 1980 Plenum Press, New York

All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted,

Honolulu is a most beautiful place, suitable for all occa-

Fig . 1. From Left to Right: Dr. T. Hata,

Fig. 2. From Left to Right: Dr. and Mrs. A.

In these Proceedings we present the revised papers and discus-

Part I: Polymer Surface Interactions

Several papers were published elsewhere, so, instead, we have

I would like to take this opportunity to thank our co-

Acknowledgement is made to the donors of the Petroleum Re-

PART ONE: Polymer Surface Interactions

Statistical Mechanics of Surface Tension and Adsorption 5

Surface and Interfacial Tensions of Polymer Melts

A Role of Molecular Forces in Adhesive Interactions of

PART TWO: Characterization of Adhesive

Photoacoustic Spectroscopy for the Study of Adhesion

Adhesive-Adherend Bond Joint Characterization by Auger

Techniques for Measuring Adhesive vs. Cohesive Failure 123

Failure Characterization of a Structural Adhesive 141

Dielectric Relaxation Gradients in an Adhesive Bond 165

New Approach to the Understanding of Adhesive Interface

PART THREE: Polymeric Structural Adhesives

Introductory Remarks 201

Fracture Mechanics and Adherence of Viscoelastic Solids 203

The Viscoelastic Shear Behavior of a Structural Adhesive 279

Surface Energetics and Structural Reliability of Adhesive

Composition and Ageing of a Structural, Epoxy Based Film

Viscoelastic Characterization of Structural Adhesive via

Crack Healing in Semicrystalline Polymers, Block Copolymers

PART FOUR: Fracture Strengths in Polymeric

Introductory Remarks 369

Criterion of Interfacial Fracture on Tensile Adhesive Joint 371

Deformation and Shear Strength of FRP Adhesive Joints 393

Stress Distribution and Strength in Shear on the Adhesive

Fracture Criteria on Peeling 421

Superposition of Peel Rate, Temperature and Molecular Weight

Author Index xvii

Subject Index xxxi

PART FIVE: Modification of Polymer Interfaces

Introductory Remarks 459

Role of Interface in the Strength of Composite Materials 463

Surface Modification of Polyethylene by Radiation-Induced