0% found this document useful (0 votes)

330 views

STP 516-1981

Uploaded by

ahmadmubarakalias

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

330 views

STP 516-1981

Uploaded by

ahmadmubarakalias

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 329

Localized Corrosion-

Cause of Metal Failure

A symposium
presented at the
Sevent/'fourth Annual Meeting
AMERICAN SOCIETY FOR
TESTING AND MATERIALS
Atlantic City, N. J., 27 June-2 July 1971

ASTM SPECIAL TECHNICAL PUBLICATION 516

Michael Henthome, symposium chairman

04-516000-27

AMERICAN SOCIETY FOR TESTING AND AAATERIALS

1916 Race Street, Philadelphia, Pa. 19103

NOTE
The Society is not responsible, as a body,
for tiie statements and opinions
advanced in this publication.

Printed in Tallahassee, Fla.

October 1972
Second Printing, July 1981
Baltimore, Md.

The Symposium on Localized Corrosion—Cause of Metal Fatigue was

presented at the Seventy-fourth Annual Meeting of the American Society
for Testing and Materials held in Atlantic City, N. J, 27 June-2 July 1971.
The sponsor of this symposium was Committee G-1 on Corrosion of
Metals. Michael Henthome, Carpenter Technology Corp., presided as
symposium chairman.

Metal Corrosion in the Atmosphere, STP 435

(1968), $27.00 (04-435000-27)

Introduction 1

Exfoliation Corrosion of Aliuninum Alloys—s. J. KETCHAM AND

I. S. SHAFFER 3

Nitric Acid Weight Loss Test for the H116 and H117 Tempers of 5086
and 5456 Aluminum Alloys—^H. L. CRAIG, JR. 17

Simplified Exfoliation Testing of Aluminum Alloys—^D. O. SPROWLS,

J. D. WALSH, AND M. K. SHUMAKER 38

Intergranular Corrosion in Iron and Nickel Base Alloys—

MICHAEL HENTHORNB 66

Significance of Intergranular Corrosion in High-Strength Aluminum

Alloy Products—B. W. LIFKA AND D. O. SPROWLS 120

Investigations of Galvanically Induced Localized Corrosion—

ROBERT BABOIAN 145

Crevice Corrosion of Metals—w. D. FRANCE, JR. 164

Crevice Corrosion of Some High-Purity Ferritic Stainless Steels—

E. A. LIZLOVS 201
Crevice Corrosion of Some Ni-Cr-Mo-Fe Alloys in Laboratory Tests—
R. p. JACKSON AND D . VAN ROOYEN 210

A Study of Factors Affecting the Hydrogen Uptake Efficiency of

Titanium in Sodium Hydroxide Solutions—L. C. COVINGTON AND
N. G. FEIGE 222

Prevention of Hydrogen Blistering and Corrosion by Organic Inhibitors in

Hydrocarbon Systems of Varying Compositions—c. c. NATHAN,
c. L. DULANEY, AND M. J . LEARY 236

Pitting Corrosion—A Review of Recent Advances in Testing Methods

and Interpretation—A. P. BOND 250

Relative Critical Potentials for Pitting Corrosion of Some Stainless

Steels—M. J. JOHNSON 262

Exfoliation Corrosion Testing of 7178 and 7075 Aluminum Alloys—

S. J. KETCHAM AND P. W. JEFFREY 273

Evaluation of the Tendency for Dealloying in Metal Systems—

E. D . VERINK, J R . , AND R. H. HEIDERSBAtH, JR. 303

Copyright by ASTM Int'l

Downloaded/printed by
University of Washington (University
STP516-EB/Oct. 1972

Introduction

Localized corrosion is a major cause of metal failure in a wide variety

of industries. More importantly it often results in unexpected failures be-
cause it is difficult to predict and is generally undetectable in its early
stages. The total amount of metal attacked may be very small, but by
being localized it can have very serious consequences. For example, one
relatively small pit in the bottom of a liquid storage vessel will make it
useless. Several of the various types of localized attack are not understood,
and there is a general lack of standard test methods to evaluate suscepti-
bility.
For these reasons Committee G-1 on Corrosion of Metals sponsored
a symposium to review the state of the art for the various types of local
attack including exfoliation, intergranular corrosion, crevice corrosion,
pitting, and dealloying. These reviews covered: (1) practical aspects—
what conditions cause attack, how can it be prevented, etc.; (2) mech-
anisms; and (3) testing. Papers providing new data on mechanisms and
test methods were also presented, and the related topics of galvanic cor-
rosion and hydrogen blistering were covered. Stress-corrosion cracking
was not dealt with specifically, but much of the symposium content was
relevant to situations which either lead to or exist during stress-corrosion
cracking.
The contents will be useful to those who need to prevent, understand,
or test for localized corrosion. In almost all cases both the metallurgical
and the electrochemical aspects have been dealt with.
Testing methods receive the most emphasis. In particular, the new data
and test methods for exfoliation will be of interest to users and producers
of aluminum alloys. The papers on pitting, crevice corrosion, intergranular
corrosion, dealloying, and galvanic corrosion highlight the mechanistic
aspects and the need for better test methods. They will be of special in-
terest to users of stainless steel, nickel base alloys, titanium alloys, and
copper alloys.
The practical problems associated with localized corrosion are far from
being solved, particularly in the case of pitting and crevice attack. The
papers in this volume define the problems and offer some solutions. They
will serve as a basis for more work aimed at increasing our understanding
of the mechanisms of attack and for the development of standard test
methods. Meanwhile they contain much useful information which will

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by FM International
by AS www.astm.org
University of Washington (University of Washington) pursuant to License
2 LOCALIZED CORROSION-CAUSE OF METAL FAILURE

enable us to minimize, if not yet completely control, this insidious form

of corrosion.
Michael Henthorne
Supervisor, Corrosion Research,
R & D Center, Carpenter
Technology Corp., Reading,
Pa. 19603; symposium chairman.

Exfoliation Corrosion of Aluminum Alloys^

REFERENCE: Ketcham, S. J. and Shaffer, I. S., "Exfoliation Corrosion

of Aluminum AUoys," Localized Corrosion—Cause of Metal Failure,
ASTM STP 516, American Society for Testing and Materials, 1972,
pp. 3-16.
ABSTRACT: This paper is a general review of the subject of exfoliation
corrosion of aluminum alloys summarizing both published and unpublished
work. Present thinking on mechanisms is presented. Susceptible alloys and
the corrosive environments in which exfoliation occurs are discussed, as
well as effect of varying degrees of exfoliation on static and dynamic
fatigue strength of 7075-T6 and on life of an actual aircraft structure.
Protective coatings and special heat treatments to minimize exfoliation
are discussed.
KEY WORDS: corrosion, exfoliation corrosion, aluminum alloys, grain
structures, plastic deformation, corrosion tests, salt spray tests, fatigue
(materials), accelerated tests
Exfoliation is a term used to describe a type of corrosion in which
delamination takes place parallel to the metal surface. When this occurs
on the surface, flakes of metal peel or are pushed from the surface due
to internal stresses caused by the building up of corrosion products. An
example is shown in Fig. 1. This type of attack can also take place below
the surface, a condition which is much more insidious, and difficult to
detect.
For a metal to exfoliate a highly directional structure of some sort is
required. It is most common in rolled or extruded aluminum alloy
products in which grains are elongated and flattened. It has been reported
also on other alloys such as primary magnesium ingots in chloride solu-
tions [7],2 a magnesium sheet alloy which contained 14 percent lithium
and 2 percent aluminum [2], and cupro-nickel heater tubes [5]. Wrought
iron, produced by piling several plates on top of one another, subjected
to heat, and passed through a rolling mill, develops zones which behave
^ Head, Chemical Metallurgy Branch, and metallurgist, respectively. Naval Air
Development Center, Warminster, Pa. 18974.
* The opinions expressed are the private ones of the authors and are not to be
construed as official or reflecting the views of the Department of the Navy or the
Navy Service at large.
' T h e italic numbers in brackets refer to the list of references appended to this
paper.

Copyright by ASTM Int'l (all rights reserved); Fri Jan

Downloaded/printed
Copyright 1972 byFM International
by AS www.astm.org
University of Washington (University of Washington) pursuant to Li
LOCALIZED CORROSION-CAUSE OF METAL FAILURE

FIG. 1—Example of severe exfoliation on 7178-T6 aluminum alloy iXl).

differently from one another when subjected to corrosive conditions.

Evans considers this zonal corrosion of wrought iron to have much in
common with the layer corrosion of light alloys [4].

Exfoliation in Aluminum AUoys

Aluminum alloys are most prone to this type of attack. It occurs on
2000 (Al-Cu-Mg), 5000 (Al-Mg) and 7000 (Al-Zn-Cu-Mg) series
alloys. It usually is related to susceptibility of these alloys to intergranular
attack as a result of heat treatment which has created selective anodic
paths at the grain boundaries for electrochemical attack. When inter-
granular attack develops in a product that has an elongated grain structure
parallel to the surface, exfoliation will occur. Such grain structures are
found in cold-rolled material and in the unrecrystallized portion of an

extrusion. The severity of the exfoliation will depend on the extent to

which the grains have been flattened or elongated, as well as to the con-
tinuity of the anodic path at the grain boundaries.
In exfoliation susceptible alloys, as corrosion proceeds along multiple
narrow paths parallel to the surface, the insoluble products which are
formed occupy a larger volume than the metal consumed in producing
them. These voluminous corrosion products exert a wedging action which
develops lateral tensile forces [5]. This results in the spUtting, flaking, or
delamination of uncorroded layers of metal, as shown in Fig. 2.
The time required for exfoliation corrosion to occur depends greatly
upon the thickness of the recrystallized layer normally present on the
outside surface of a rolled or extruded product. This outer layer is usually
susceptible to intergranular corrosion but, having an equiaxed grain struc-
ture, does not exfoliate. In a solid plate or extrusion no exfoliation takes
place until the corrosion environment has penetrated to the laminar grain
structure of the underlying material which is not recrystallized because
it did not receive an equivalent amount of cold work.
Thus, variations in exfoliation susceptibilities and rates of attack can
be often attributed to differences in thickness of the recrystallized skin.
In service where actual commercial structures or parts have machined
surfaces or rivet or fastener holes, the corrosive electrolyte can have easy

FIG. 2—Example of splitting of layers on 7178-T6 due to buildup of corrosion

products {XlOO unetched).

access to the interior directional grain structure, and exfoliation can occur
without a long incubation period.
Liddiard et al working with extrusions developed corrosion time curve
plots for an aluminum-copper alloy [6]. The effective loss in thickness was
used as the indication for the degree of exfoliation. By measuring the
deflection of specimens under four-point loading before and after ex-
posure in exfoliation tests, the changes in the effective thickness of the
extrusion was determined. In contrast to the normal exponential corrosion
rates for most aluminum alloys, they found that the corrosion-time curves
for specimens that undergo exfoliation are linear. The corrosion-time
curves that they developed, for specimens tested with the as-extruded
face exposed, showed three distinct portions: the first represents inter-
granular penetration of the recrystallized skin and is slow; the second
represents the rapid undermining of the skin as the attack reaches the
highly directional material, and the rate is fast; and the third represents
the exfoliation attack on the interior structure which proceeds at a rate
intermediate between the first and second. Exposing machined faces which
had the recrystallized surface removed produced a curve which after an
incubation period showed a single linear rate. Differences in exfoliation
resistance on the T/10 plane* and the T/2 midplane* in a 0.375-in.
plate are shown in Fig. 3.
Ufka et al point up another factor which influences susceptibility to
exfoliation of sheet and plate material [7]. Grain directionality increases
with increased reduction in area of the ingot. Plates over 1 in. in thickness
do not show the same degree of flattening at the midplane that thinner
plates do and are, therefore, more resistant to exfoliation. However, there
is a greater susceptibility to intergranular corrosion because of possible
slower cooling on quenching with thick plate. In the thickness range from
0.150 to 0.750 in., the rolled structure is the most prone to exfoliation.
With further reduction to thin gage sheet a high degree of recrystallization
takes place, and susceptibility to exfoliation decreases.
It has been well established m aluminum alloys that the degree of
exfoliation that starts at an edge of a specimen is affected markedly by
the method in which the edge was prepared. Where prior plastic deforma-
tion at an edge is excessive, exfoliation will be much more severe. The
best example of this point is the difference in the degree of exfoliation
between a sawed and sheared edge. Since the shearing operation causes
severe microstructural deformation it is not surprising that the electro-
chemical attack along preferred microstructural paths such as grain bound-
aries would be accelerated. Another prime example of this effect of plastic
deformation is in machined holes. When fastener holes are drilled with a

*T/10 plane refers to plane exposed when one tenth of the thickness is removed
from the plate; T/2 when one half of the thickness is removed.

^e lid
«> X

I.&

•a

I
d

dull drill bit or when rivets are drilled out of holes, the grain structure
distortion that develops can greatly change the exfoliation susceptibility
of the end grains in the holes.
Exfoliation has been found in aluminum material which was not sus-
ceptible to intergranular corrosion. Godard has referred to cases where
it was believed that preferential corrosion paths were provided by in-
soluble cathodic constituents which formed during solidification in the
original cast ingot [5], During subsequent working, these constituents be-
came aligned in stringers unrelated to the final recrystallized grain struc-
ture. Electron probe investigations have indicated that the presence of iron
and titanium tend to promote exfoUation.
Service Experience
Exfoliation of high-strength aluminum alloys is most prevalent in
marine and industrial environments. Pearlstein and Teitell reported severe
exfoliation on 2024-T3 exposed to tropical marine enviromnent in
Panama [9].
It has been stated that exfoliation of light alloys occurs generally in
atmospheric exposure, and is not normally encountered under immersion
conditions [6]. In deep sea exposure, however, at various depths in the
ocean, 7002, 7075, and 7178 sheet specimens in the T6 temper exfoliated.
Not long ago, severe exfoliation problems were encountered in the bilge
area of service boats constructed of aluminum alloy 5456-H321 [70].
Investigation showed that corrosion of the hull plates was due to a high
degree of continuity of grain boundary precipitate.
It is in aircraft applications where exfoliation creates the most concern.
Several descriptions of service problems have been published [11-13].
High-strength aluminum alloys developed during and after World War II,
with their significant strength to weight advantages, were at first used
conservatively, predominantly in thin gages. The next generation aircraft
used extrusions and sculptured plates to produce integrally stiffened parts.
The first problem involving a major structural component occurred
about 15 years ago on thick skin material of a seaplane in an area sub-
jected to impingement by the engine exhaust of piston engines. Reinhart
has discussed the corrosion of aluminum alloys by exhaust gases [14].
Soon other problems developed. One interesting one involved exfoliation
in wing fuel tanks of a turboprop plane due to the presence of micro-
organisms.
Subsequently, the Navy began flying planes that were too large to be
stored in the hangar deck of an aircraft carrier and had to be stored on
the flight deck exposed to a combination of sea spray and sulfur bearing
compounds. Incidence of exfoliation problems, particularly around fas-
tener holes, became much greater. Exfoliation, however, is not restricted
to the exterior. Problems have also developed in the interior of wing and

fuselage sections where fastener or rivet holes allowed entrance of

moisture.
An example of such a problem was in the spar caps of air sea rescue
planes. Internal defects were revealed by ultrasonic inspection of the
spar caps which were L-shaped 7075-T6 extrusions, bolted and riveted to
the wing center section of the aircraft. The defects which were restricted
almost exclusively to the skin attachment leg of the spar caps resulted
from pitting corrosion that initiated deep within the rivet holes and de-
veloped as exfoliation along the highly directional interior grain structure
of the extrusions, as shown in Fig. 4.
An investigation was made to determine to what extent this internal
exfoliation had compromised the structural integrity of the spar caps, and
the results have been published [75].
Specimens were taken from locations with varying degrees of surface
corrosion and of internal defects as indicated by ultrasonic inspection.
Fatigue tests were conducted under constant amplitude and under spec-
trum loading. Results are presented graphically in Fig. 5.
The results of the constant amplitude fatigue tests indicated that the
amount of surface corrosion present had no deleterious effect on fatigue
life. Likewise sUght laminar attack had only a negligible effect. Three
specimens with severe exfoliation in addition to some surface pitting
showed an average life reduction of more than 40 percent and a maximum
life reduction of almost 60 percent compared to specimens which had
suffered no corrosion.
The results of the variable amplitude or spectrum loaded fatigue tests
indicated that at the limit load of 30,000 psi the average life was un-
affected for the slightly exfoliated material but was reduced more than
50 percent for the severely exfoliated material.
Illustrations of the internal exfoliation which adversely affected the
fatigue strength of spar caps are found in Fig. 6. These photomicrographs
show that the corroded laminar paths are preferential sites for the initia-
tion of fatigue cracks.
Since the effect of exfoliation on the fatigue life of the spar caps was
considerable, it was decided that a full scale wing fatigue test of the air-
craft should be conducted, to provide the basis for establishing service
life criteria for spar cap replacement or modification.
Subsequently an aircraft of the same model with a total of 7216 flight
hours was withdrawn from service and used as the test vehicle. During
the fatigue test on the wing 8200 test hours were accumulated before
catastrophic failure occurred. Post-failure examination revealed the pres-
ence of exfoliation in the wing main beam lower spar cap. At this point
the left hand wing was reinforced with steel straps, and testing continued.
After another 4000 h, the left wing failed. Post-failure inspection in-
dicated that this left wing had more exfoliation than the right wing which

FIG. 4—Cross sections through 7075-T6 spar cap rivet holes showing pitting
corrosion, and subsequent development into exfoliation (top) X5 and (bottom)
X250 unetched.

S 40

z so

SUIFtCI SUCNT sinii

ATTACI ilfOUAIIM illOUATION
ONtr

] CONSTANT AHPUTUDI risp.

SPECnUa lOADINt
000 pil UNIT lOAD
L I 30,1
MIAN STIESS = 0
AtTIINATING STIISS JS.OOO pil

FIG. 5—Results of fatigue tests on 7075-T6 spar cap specimens.

failed first. It was concluded that the existence, but not necessarily the
amount of exfoliation, is a determining factor in fatigue life. That is, the
existence of even a small amount of exfoliation can initiate a fatigue
crack [16].
Protection Against Exfoliation
Metallic aluminum, zinc, and magnesium coatings (or combinations
thereof) will provide protection against exfoliation. Carter et al describe
the use of sprayed metal coatings to prevent exfoliation [17]. Paint coat-
ings are inadequate to prevent exfoliation around fastener and rivet holes
in highly stressed areas because the films are too brittle and crack. On
older model military aircraft exfoliation is being minimized by the use
of elastomeric polysulfide sealants, particularly over fastener patterns and
around faying surfaces [18].
The other approach is the use of extended aging to produce exfoliation
resistant tempers [19]. The effect of these extended or overaging treat-
ments is to lessen the tendency for intergranular corrosion to occur.
During extended elevated temperature aging, additional precipitation takes
place in the subgrain structures, providing a more equipotential situation
which lessens the rate of grain boundary penetration and exfoliation. The
effectiveness of these special aging treatments which are used commer-
cially has been demonstrated by laboratory and atmospheric exfoliation
tests. The T73 aging temper for 7000 series and the artificially aged
tempers of the 2000 series aluminum alloys used to minimize or eliminate
susceptibility to stress corrosion cracking also understandably have proven
to produce immunity to exfoliation. There is, however, a significant de-

L—<-—^

I
FIG. 6—Photomicrographs showing initiation of fatigue cracks at sites of in-
ternal exfoliation in fatigue specimens fabricated from spar cap extrusions (top)
X250 and (bottom) XlOO.

crease in strength (10 to 12 percent) with the T73 temper compared to

the older, more commonly used T6 temper. It was found that less over-
aging was required for exfoliation immunity than for stress corrosion im-
munity. For this reason a T76 temper having exfoliation immunity and
less loss in strength from the T6 temper was developed. Exfoliation re-
sistant tempers (HI 16 and HI 17) were developed recently for 5085
and 5456 alloy boat hull plate.

Test Methods
When exfoliation corrosion problems began to crop up in service, users
became critical of laboratory tests which had not detected this possibility.
The standard 5 percent salt spray test (ASTM B 117-64) produced
pitting attack but not exfoliation. A test was developed using 5 percent
salt solution acidified to pH3 with acetic acid [20]. The test was cyclic,
alternately spraying and allowing the specimens to become dry. As more
experience was gained with this test, its ability to detect borderline con-
ditions was questionable.
Several researchers experimented with electrochemical techniques. Budd
and Booth used a potentiostatic approach [21]. In the authors' laboratory
a galvanostatic method was tried. Although both appeared promising, the
electrochemical techniques did not lend themselves to the multiple testing
required for production control of heat treatment.
In the meantime, a cyclic acidified, synthetic sea salt spray test was
proposed [22] and also a 5 percent salt spray test involving introduction
of sulfur dioxide into the cabinet at specified intervals [23]. The latter
test was developed to simulate the environment of an aircraft carrier
which combines sea spray and sulfur from the carrier stack gases.
Results of a comparison of the three salt spray tests and a new immer-
sion test were published which indicated the acidified synthetic sea salt
spray was the most discriminating for detecting behavior of exfoliation
resistant tempers of 7178 [23]. Figure 7 shows a comparison of the ex-
foliation produced by the acidified sodium chloride and acidified sea salt
spray tests on identical specimens of 0.375 in. plate of 7178.
Because of the exfoliation problems on the military boats mentioned
earlier. Navy authorities decided to incorporate a test for determining
susceptibility to exfoliation into the specification for 5456 alloy. However,
there was lack of agreement among producers and users of the exfoliation
resistant tempers as to the best test method. This prompted the formation
of a task group among Aluminum Association members for 5000 series
testing and a joint task group for 7000 series alloys of ASTM GO 1/05
and the Aluminum Association. Both have been conducting interlabora-
tory testing programs of several test methods. The object of the ASTM
program was to prepare a recommended practice for the method which
in the opinion of the participants was the most satisfactory for the 7000

series. A description of the 7000 series program is published elsewhere

in this book.

Summary
With the increased use of exfoliation resistant tempers and the improve-
ment in test methods for detecting susceptibility, the incidence of severe
exfoliation problems on current and future production of military weapons
systems or other equipment fabricated from aluminum alloys should sub-
stantially decrease. The exfoliation resistant tempers in the 2000 and
7000 series alloys are, however, more prone to pitting corrosion as a
result of the equidistribution of precipitates throughout the matrix. Good
protective finishing systems for these alloys, therefore, will still be neces-
sary in marine and industrial environments.
In older equipment where exfoliation prone tempers are stiU in use,
constant vigilance is required so that the inception of exfoliation can be
detected and arrested. Ultrasonic inspection equipment of the type used in
the spar cap investigation is useful for this purpose. The importance of a
continuing program of replacement of structural components fabricated
from exfoliation resistant tempers cannot be overemphasized, particularly
in those cases where structural integrity under dynamic conditions is
required.

References
[/] Schiebold, E. and G. Siebel, Zeitschrift fur Physik, ZEPYA, Vol. 69, 1931,
p. 459.
[2] Godard, H. P., Jepson, H. P., Bothwell, M. R., and Kane, R. L. in The Cor-
rosion of Light Metals, Wiley, New York, 1967, p. 271.
[31 Castle, J. E., Harrison, J. T., and Masterson, H. G., British Corrosion Journal,
BCRJA, Vol. 1, No. 4, Jan. 1966, p. 143.
[41 Evans, U. R. in The Corrosion and Oxidation of Metals, Edward Arnold Ltd.
London, 1961, p. 511.
[5] Nielsen, N. A., Physical Metallurgy of Stress Corrosion Fracture, Metallurgi-
cal Society Conferences, Vol. 4, Interscience, N. Y., 1959, p. 121.
[6] Liddiard, E. A. G., Whitaker, J. A., and Farmery, H. K., Journal, Institute of
Metals, JIMEA, Vol. 89, 1960-1961, p. 377.
[7] Lifka, B. W., Sprowls, D. O., and Kaufman, J. G., Corrosion, SBIIA, Vol.
23, No. 11, Nov. 1967, p. 335.
[81 Godard, H. P., private communication.
[91 Pearlstein, F. and Teitell, L., Proceedings, 7th Annual National Conference on
Environmental Effects on Aircraft and Propulsion Systems, 1967.
[101 Sprowls, D. O., Summerson, T. J., Romans, H. B., and Sansonetti, S. J.,
Aluminum Association Report, issued 1969.
[77] Tate, R., Chemistry and Industry, CHINA, 27 April 1957, p. 506.
[72] Klang, D. R. and Peters, W. L., Materials Protection, MAPRA, April 1964,
p. 16.
[131 Ward, C. B. and Cohen, B., Light Metal Age, LMAGA, Vol. 24, Feb. 1966,
p. 16.
[141 Reinhart, F. M., Corrosion, SBIIA, Vol. 10, No. 12, Dec. 1954.
[75] Shaffer, I. S., Sebastian, J. C , Rosenfeld, M. S., and Ketcham, S. J., Journal
of Materials. JMLSA, Vol. 3, No. 2, 1968, p. 400.

[16] Berman, L., "Determination of Life Remaining in Model XX Airplane Wing,"

Naval Air Development Center Report, NADC-ST-7007, June 1970.
[17] Carter, V. E. and Campbell, H. S., Journal, Institute of Metals, JIM£A, Vol.
89, 1960-1961, p. 472.
[18] Miller, R. N., Material Protection, Vol. 6, No. 2, 1967.
[19] Bell, W. A. and Campbell, H. S., Journal, Institute of Metals, JIMEA, Vol. 89,
1960-1961, p. 464.
[20] Lifka, B. W. and Sprowls, D. O., Corrosion, SBIIA, Vol. 22, No. 1, Jan. 1966,
p. 7.
[21] Budd, M. K. and Booth, F. P., Corrosion. SBIIA, Vol. 18, No. 5, May 1962.
[22] Romans, H. B., Materials Research and Standards, MTRSA, Nov. 1969.
[23] Ketcham, S. J. and Shaffer, I. S., Proceedings, 4th International Congress on
Metallic Corrosion, 1969.

Nitric Acid Weight Loss Test for the H116 and

H117 Tempers of 5086 and
5456 Aluminum Alloys

REFERENCE: Craig, H. L., Jr., "Nitric Add Weight Loss Test for the
H116 and H117 Tempers of 5086 and 5456 Aluminum Alloys," Localized
Corrosion—Cause of Metal Failure, ASTM STP 516, American Society
for Testing and Materials, 1972, pp. 17-37.
ABSTRACT: This test was developed to indicate the resistance of
aluminum-magnesium alloys 5086 and 5456 in the HI 16 and HI 17 tempers
to exfoliation and intergranular corrosion. The test consists of measuring
the weight loss per unit area of a 2 by 0.25 in. by gage specimen im-
mersed in concentrated nitric acid at 30 C (86 F) for 24 h. The acid
dissolves a second phase, an aluminum-magnesium intermetallic compound,
in preference to solid solution of magnesium in aluminum. When this
compound is precipitated in a continuous network along grain boundaries,
the effect of the preferential attack is to eat around the grains, causing
them to fall away from the specimen. Weight losses for materials proven
susceptible to exfoliation or intergranular corrosion amount to the hundreds
of milligrams per square inch, while samples of 5086 and 5456 in the
H116 and H117 tempers lose only from 10 up to 100 mg/in.".
Results of tests performed on commercially produced materials are
presented, as well as several examples using other tempers and alloys
which have shown varied resistance to exfoliation or intergranular corro-
sion in service.
KEY WORDS: corrosion, aluminum magnesium alloys, exfoliation corro-
sion, intergranular corrosion, corrosion tests, corrosion resistance, nitric
acid, salt spray tests, immersion tests (corrosion), etching
In order to ensure that the Department of Defense has available to it
materials with superior resistance to corrosion, members of the Aluminum
Association cooperated in an examination of test methods for use in speci-
fying 5086 and 5456 alloys for boat and ship hull construction [/].^ The
chemical composition and mechanical properties of these and a closely
related alloy, 5083, are given in Table 1.
The present Federal Specification [2] calls for a metallographic exami-
' Research supervisor. Metallurgical Research Division, Reynolds Metals Co.,
Richmond, Va. 23218.
' The italic numbers in brackets refer to the list of references appended to this
paper.
17

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by FM International
by AS www.astm.org
University of Washington (University of Washington) pursuant to License
18 LOCALIZED CORROSION—CAUSE OF METAL FAILURE

TABLE 1—Nominal chemical compositions and typical mechanical properties

of alloys 5086, 5083, and 5^56."
Tension
Element' Strer igth"
Elongation,
Alloy-Temper Mg Mn Cr Ultimate Yield percent in 2 in.
5086-H32 4.0 0,45 0.15 42 30 12
(H116, HUT)
5083-H321 4.45 0.6 0.15 46 33 16
5456-HS21 5.1 0.8 0.12 51 37 16
(H116, H117)
° Reference: Aluminum Standards and Data, 1970-1971, The Aluminum Associa-
tion, 1, Dec. 1969, Table 1.1, p. 14, and Table 2, pp. 28-29, Federal
Specifications QQ-A-00260/19 and 20, have replaced the H32 temper
for 5086 and the H321 temper for 5456 with the designations, H H 6 and
H117.
' Values in percent—balance, Al plus normal impurities.
'Ksi.

tion of each production lot for susceptibility to exfoliation and intergranular

corrosion. The criterion is that "the microstructure is predominandy free
of a continuous grain boundary network of aluminum-magnesium precipi-
tate." This examination is carried out by etching a polished specimen with
40 percent phosphoric acid solution at 95 F (35 C) for 3 min, and
examining the resultant etched surface at 500 magnification.
A reference micrograph is established on a "first-article" basis for sheet
and plate which is representative of mill production. The corrosion re-
sistance of this "first-article" reference material is judged by exposing
samples for one week to the synthetic seawater acetic acid intermittent
salt spray test (SWAAT) [2,3].
In an Aluminum Association round robin, the SWAAT test produced
exfoliation of samples of 5456-H321 which had experienced exfoliation
corrosion in actual service. The same test conditions caused only negligible
corrosion of materials believed to have improved resistance to exfoliation,
such as 5086 and 5456 in the HI 16 and H117 tempers [4].
This test, it is recognized, does not always cause exfoliation of materials
believed to exhibit borderline susceptibility. Most workers ascribe this to
the difficulty in conducting a reproducible salt spray test. Because of this,
further work is being carried out on immersion tests. This paper presents
the results obtained with one of these tests, the nitric acid weight loss test.
It also describes the development and reasoning behind the procedure.
The second test, called ASSET, is the subject of another report,*
Premise of Test Procedure
The nitric acid immersion test is based on the same philosophy as the
metallographic examination described in the pertinent Federal Specifica-
tion [2]. In fact, it might be regarded as a quantitative metallographic
"See p. 38.

procedure, as it measures a characteristic of the microstructure of the

metal. This test is not intended to supplant the metallographic examination
of each production lot, but rather is intended to provide a means of quan-
titative measurement of precipitate continuity. It does not produce the
visual appearance of exfoliation in the test specimens. However, the pre-
dominant corrosion attack is clearly intergranular as indicated by metallo-
graphic examination of the exposed specimens. An example is shown in
Fig. 1.
The attack is intergranular because nitric acid dissolves the magnesium-
aluminum intermetallic compound while not appreciably attacking solid
solution of magnesium in the aluminum matrix. Due to the nature of the
precipitation process, it is possible to have a continuous network of the
intermetallic compound surrounding each grain, while the grain interior
remains relatively free of this precipitate. It is this condition that gave
rise to the exfoliation and intergranular corrosion experienced in service
by materials in ships hulls [5].
Using the SWAAT test described previously and other tests, the alu-
minum producers developed two tempers for 5086 and 5456 alloys, each
of which imparts resistance to exfoliation and intergranular corrosion.
In this temper, the alloys do not contain this continuous network of grain
boundary precipitate. Consequently, immersion in nitric acid causes little
weight loss (see Fig. 2).
On the other hand, in susceptible material, prolonged exposure to nitric
acid dissolves the envelope of intermetallic compound precipitated around
the grains. Entire grains become dislodged and fall away from the speci-
men, thereby causing large weight losses, as illustrated in Fig. 1.
This test defines, on a quantitative scale, whether or not precipitation
is of such an extent as to cause grain dropping in nitric acid. The problem
comes in selecting a pass-or-fail value in relation to the material perform-
ance in various environments other than nitric acid. Reliance must be
placed, then, on a sampling of materials which have been in service, and
which may or may not have suffered exfoliation or intergranular corrosion.
Such a selection is presented in this paper.
The decision on the magniture of the weight loss value a corrosion
resistant material should exhibit is pending further evaluation within the
Aluminum Association.
Test Procedure (summary)
The test is carried out in the following manner: a specimen is machined
to size. A convenient size is 2 by 0.25 in. by gage, with the 2 in. dimen-
sion in the direction of working. It is convenient to limit the specimen
gage to 1 in., and the specimen should be cut to preserve one worked
surface.

•••%^mm^tm

. »•

(top) Lamellar grain structure, 5456 alloy, nitric acid weight loss,
177 mg/in."
{bottom) Equiaxed grain structure, 5086 alloy, nitric acid weight loss,
129 mg/in.'
FIG. 1—Intergranutar corrosion of lest specimens after immersion in nitric acid
(y.250; etch, nitric acid).

vS-P/-^1'-i^'V-f''''- .•••••' '•*

•I .• A' , ""'iJ--" r i • '

(/op) 5086 alloy, 8 mg/in." weight loss in test.

(middle) 5086 alloy, 23 mg/in." weight loss in test, slight intergranular
corrosion attack.
(bottom) 5456 alloy, 60 mg/in." weight loss in test, slight intergranular
corrosion attack.
FIG. 2—Corrosion attack on test specimens after immersion in nitric acid
(X250; etch, nitric acid).

The specimen is cleaned by caustic etching prior to weighing, and then

the weight is determined to the nearest tenth of a milligram. It is then
immersed in concentrated nitric acid and held at 30 C (86 F) for 24 h.
Upon removing the specimen from the acid, it is rinsed and any loose
particles (grains) of metal are brushed off with a stiff bristle brush. The
specimen is dried, weighed, and the weight loss per unit area (WLV) is
calculated. A detailed procedure is given in the Appendix.

Notes on Test Procedure

It has been observed that the transverse surfaces of the specimen are
attacked more heavily than the worked surfaces. This results in greater
weight losses per unit area on those surfaces. Therefore, WLV from two
different gages are not strictly comparable. This is a second order effect,
however, and for purposes of this test may be neglected. It is illustrated
in Fig. 3.
The effects of alloy composition, temper, and grain morphology also
appear to be of second order in magnitude. Whether these effects are
significant should be determined by the person running the test. It should
be pointed up that, with a large WLV, the poor reproducibility due to the
variation inherent in removing loose grains from the specimen may obscure
these effects.
There are undoubtedly environments in which materials with large
WLVs can be used with little concern for corrosion. Thus, material ac-
ceptability is a joint responsibility of the consumer and the producer. In
specifying the H116 and HI 17 tempers for 5086 and 5456 alloys, the
consumer is assured that these alloys are in their most corrosion-resistant
tempers.

Test Development
As stated previously, this test is an outgrowth of a metallographic pro-
cedure. Two acids are commonly used to reveal the precipitation of the
aluminum-magnesium intermetallic: phosphoric acid and nitric acid.
Some data from a comparative study are presented in Table 2; examina-
tion of the data illustrates the reasoning for the choice made between
the two acids. Condition A represents 5456 alloy which has been heated
at a temperature and for a time period which produces an intergranular
precipitate around the grains. Condition B, on the other hand, has a
structure with all the magnesium in supersaturated solid solution. Con-
centrated phosphoric acid attacks B to a significant extent whenever A is
attacked. Thus, phosphoric acid is not very selective towards the alumi-
num-magnesium intermetallic compound. Nitric acid, on the other hand,
produces only minimal weight losses for B, while registering relatively
large weight losses for A.
Another variable was considered, that of specimen size. A comparison

(top) Section through surface of specimen lying in the rolling

plane.
(bottom) Section through surface of specimen lying in a plane trans-
verse to rolling direction.
FIG. 3—Corrosion attack on worked and transverse surfaces of nitric acid iveight
loss test specimen, 5456 alloy, 115 mg/in.^ weight loss in test (x250; etch, nitric
acid).

TABLE 2—Acid variables affecting the attack on intergratiular precipitate

in 5i56-H3Sl aUoy."
HjPO,, 85 weight percent HNOa, 70 weight percent
15 Volume Concentra- 15 Volume Concentra-
Percent ted Percent ted
Time, Temperature,
mm degC A B A B A B A B
15 30 8.0 1.4 1.0 1.0
45 30 16.0 9.1 6.0 0.7
16 60 36.0 36.0 16.8 1.2
45 60 94.0 23.0 40.0 2.8
" Values in the table represent weight loss per square inch for 5456-H321 alloy, 1.25 in.
gage, cut into specimens, 0.5 by 0.5 by 1.25 in.
Condition 'A'—5 specimens—heated one week at 212 F to develop intergranular
precipitate.
Condition 'B'—3 specimens—heated 3 h at 800 F, cold-water quenched to minimize
intergranular precipitate.

was made between a 2 by 0.25 in. by gage rectangular bar and a 2 by 2

in. by gage square section coupon. Again, two materials were selected
for study. One was 0.25-in. gage plate that represented the same condi-
tion as B, namely, a structure with most of the magnesium in supersatu-
rated solid solution. To demonstrate the effect on the nitric acid weight
loss value, a series of specimens was heated for various time intervals
at 100 C (212 F) before determining the weight loss value. This pro-
cedure, often called "sensitizing," produces a continuous precipitate of
aluminum-magnesium compound around the grains. The second material
was 0.188-in. gage sheet that initially represented a structure with both
intergranular and intragranular precipitate. Specimens were also subjected
to the same heating program as the 0.25-in. gage plate and then tested.
The results in Fig. 4 demonstrate three points:
1. The WLVs apparently are not affected greatly by the shape dif-
ference.
2. When specimens are heated for prolonged periods at 100 C, (212 F)
the WLVs become erratic. This is no doubt due to a redistribution of the
intergranular precipitate during this heating period.
3. With the 0.25-in. gage material, a sharp increase in WLV occurs
between 10 and 20 h, due to the onset of precipitation around the grains.
The weight losses greater than about 50 indicate significant amounts of
grain dropping. The 0.188-in. gage material seems to be "stabilized" with
respect to the intergranular precipitation, and heating does not increase
the weight loss value.
Table 3 shows the effects of time, temperature, and concentration of
nitric acid. Again, two materials were selected for the comparison. One,
Material C, represents a weight loss value near the 50 mg/in.^ value in
the normal test procedure, that is, 24 h immersion in concentrated nitric
acid at 30 C (86 F). Material D specimens were taken from a sample of
5456-H321 plate which exhibited exfoliation corrosion in service. It was

100
Houu »T 212 f (100 Q
FIG. 4—5083-H113 sheet and plate comparison of specimen size and its effect
on HNOi weight loss.

TABLE 3—Study of factors affecting the nitric acid weight loss test.

Factor Concentration, in volume percent

Dilute (25%), Ci Concentrated (100%), Ca
MATERIAL C — R E S I S T A N T TO EXFOLIATION

ti tj t2/t, ti ts ts/ti tl ts
Time, h 6 24 4.0 6 24 4 C1/C2 Ci/Cj
Temperature.
T,, 30 C 31 234 7.5 4 42 11.7 7.8 5.6
Tj, 40 C 103 510 4.9 16 146 9.1 6.4 3.5
T2/T, 3 2.2 4 3.5

MATERIAI. D — S U S C E P T I B L E TO EXFOLIATION

tl t2 t2/tl tl U tj/ti tl t2
Time, h 6 24 4 6 24 4 Cl/Cj C1/C2

Temperature:
Ti, 30 C 78 379 4.9 13 108 8.3 6.0 3.5
T2, 40 C 195 709 3.6 49 281 5.7 4.0 2.5
T2/T. 2.6 1.9 3.8 2.6
D/C
Time, h 6 24 6 24
Temperature:
T,, 30 C 2.5 1.6 3 3 2.6
Tj, 40 C 1.9 1.4 3. 1 1.9

chosen from the available samples left from the Aluminum Association
round robin test program. It had the lowest weight loss value of those
samples, 108 mg/in.^. All corroded areas were removed before specimens
were taken from the plate.
The ratios of weight loss values were used to assess the effect of each
variable. The grand averages are shown in Table 4.
It is expected that the discrimination between Materials C and D would
be relatively low, as the samples were selected to represent conditions
that were near the high range for resistant materials an'^' on the low side
for susceptible materials. Earlier work showed that the effect of immer-
sion time is linear m each of the two branches of the curve for the 0.25-
in. gage material shown in Fig. 4. Therefore, if the attack has proceeded
far enough to cause grain dropping, the ratio should be 4.0. This condi-
tion was met when the smaller weight loss value in a given ratio exceeded
100 mg/in.*. The grand average of 7.0 for this ratio demonstrates that
TABLE 4—Grand averages of ratios of weight loss values
for factors affecting the nitric acid weight loss test.
Material D/C 2.3
Time tj/tj 7.0
Concentration ci/cj 4.9
Temperature T,/Ti S.O

many weight loss values obtained in these tests did not reflect the grain
dropping condition, since the weight losses are relatively low until grain
dropping occurs, and the ratio is therefore greater than 4.0.
The effect of concentration is typical of the behavior of aluminum in
nitric acid. With pure aluminum, for equal times and temperatures, a
maximum in corrosion rate is observed at a concentration of about 20
volume percent. This rate falls off as the concentration increases to 80
volume percent up to the concentrated acid [6]. Under the conditions
studied here, the dilute acid was about five times more corrosive than
the concentrated acid. This and other work demonstrates that the con-
centrated acid is the better agent for producing intergranular attack in
this weight loss test, as the attack on the grain bodies is minimized.
The usual effect of a 10 C increase in temperature is to double the rate
of a chemical reaction. This ratio was approached as a limit in the present
test. The same considerations enumerated above for the time effect appear
to apply to this factor: once a minimum weight loss value is obtained,
the ratio approaches 2.0. This value appears to be between 100 and 200
mg/in.2.
Testing the H116 and H117 Tempers
Another series of tests was performed on Samples of 5086 and 5456
alloys in the HI 16 temper. All these materials passed the Federal Speci-

fication, "first-article" qualification test—the SWAAT test for exfoliation

and intergranular corrosion. As can be seen from Table 5, the weight loss
values ranged from 9 to 80, with most values falling between 10 and 32.
There were no apparent trends between alloys, or among gages or labo-
ratories, with the exception that the value for the 2.5-in. gage was high.
This specimen, if it had been cut with no more than 1 in. of the transverse
surfaces exposed (that it, reduce the gage from 2.5 to P4 in. as spelled
out in the proposed procedure), would have probably had a much lower
weight loss, in tlte 25 to 30 range.
TABLE 6—Typical weight loss values for several gages
of 5086 and 5i6S in the H116 temper.
Gage, Weight Loss
Alloy in. Sample mg/in.' Remarks"
5086 0.188 A 12 specimen size: 2 by 0.250 in. by
B 11 gage, tested by Laboratory B, un-
0.250 A 13 less noted below
B 13
C 10 Laboratory C
D 12 Laboratory C
0.875 A 16
B 16
1.00 A 16
B IS
1.5 A 12 specimen size: 2 by 0.25 by 0.75 in.
B 12
5456 0.250 A 9 Laboratory B-2'
B 26 Laboratory C
C 32 Laboratory C
0.500 A 20
B 19
0.750 A 16
B 18
D 11 Laboratory B-2'
E 14 Laboratory B-2'
F 16 Laboratory B-2''
1.00 A 24
B 24
1.50 A 18 2 by 0.25 by 0.75 in.
B 18 2 by 0.25 by 0.75 in.
2.6 A 80 Laboratory C
All materials passed the SWAAT laboratory test for exfoliation corrosion suscep-
tibility.
' Laboratory B-2 uses specimens 2 by 2 in. by gage.

The test procedure was given to the other members of the Aluminum
Association Committee to study reproducibility among laboratories. The
results are shown in Table 6. The samples tested were the same lots that
were used in the first Aluminum Association round robin exfoliation
testing program [/], including three samples which were given treatments
to produce conditions of low and medium resistance to exfoliation. An
analysis of variance was performed on these results by A. I. Kemppinen,
with the following results: there were no significant differences at the 99
percent confidence level due to variations among the laboratories or be-

TABLE 6—Comparison of nitric acid weight loss data for 5086 and 5i56 alloys in the
HUB and H117 tempers among laboratories (A-F) participating in the
Aluminum Association Round Robin Teat Program Two.
Identification Weight Loss, mg/in 1

Alloy Temper Gage A B C D E P Avg

5086 H116 0.250 52 59 59 101 93 61 26 20 59
1.00 12 14 14 17 16 25 24 17
H117 0.250 8 8 8 10 4 3 6 7
1.00 7 8 8 8 0 2 4 6 7
5456 H116 0.250 14 17 17 21 36 8 16 18
0.750 16 18 18 18 22 10 10 15 16
H117 0.250 13 16 16 22 11 7 17 15
1.00 9 10 10 10 7 6 9 9
5086 a 0.250 277 283 269 269 278 315 184 221 262
5456 h 0.250 213 219 210 204 242 224 136 155 200
a 0.250 274 310 289 312 339 203 283 287
...
• Control—low resistance.
* Control—medium resistance.

tween the tempers or the gages in 5086 alloy. Alloys were not compared,
nor was the gage tested in S456 due to the lack of strictly comparable
values (that is, the 0.75-in. gage for 5456-Hl 16 rather than a 1.00-in. gage
sample). No statistical evaluation of the control samples was made. The
grand mean for 5086 alloy was 22 mg/in.* and for 5456 was 16 mg/in.''.
The wide spread between these values and those for the low- and medium-
resistance samples demonstrates the ability of the test to discriminate
among good and poor materials.
Table 7 compares the average weight losses with the individual exfoli-
ation test results. Except for the Q ratings in the ASSET test, there were
no indications of susceptibility in any of the three tests for HI 16 or HI 17
temper materials. However, in the controls, some question was raised in
the ASSET test (one of duplicates in twelve sets) and more in the
SWAAT test (four in twelve sets). With the nitric acid WLV, there was
good separation of the medium resistance (200) from the two low-
resistance values—262 and 287.

Correiation with Service Experience

Since the SWAAT test was correlated with service experience, the
nitric acid weight loss test was also used to examine samples removed
from actual in-service failures of 5456-H321. The WLVs ranged from
114 to 492 in these specimens, as shown in Table 8. Since there was only
slight variation among these samples in their exfoliation behavior, it is
difficult to assess the significance of these WLVs in terms of degree of
resistance. However, some differences of degree were available in two
AA lots—^Lot 2, with a weight loss of 180 was marginal. One of three
laboratories found slight exfoliation in this Lot. Lot 4, with a weight loss

TABLE 7—Correlation of nitric acid weight loit viitk exfoliation and intergranular
corrosion behaior 0/ 5086 and 5^56 alloys in the H116 and H117 tempers in the
Aluminum Association Round Robin Test Program Two (Laboratories A-D).
Id(entification Weight Loss Exfoliation "tests—Round Two*
Weight Loss, ASSlET (immersion) SWAAT 1(salt spray)
Gage, mK/in.'
Alloy Tempei in. Avg of 6 Labs A B C D A B C D
5086 H116 0.250 59 Q' P* Q* P' P* N JM JM
1.00 17 Q* P* Q' N P* N N P^
H117 0.250 7 Q* P* N N P* N fd pd
1.00 7 P* N N N P* P« pi P*
S456 H116 0.250 18 Q* I« Q* P" N N I" N
0.750 16 Q* P* 0* P' P* N N N
H117 0.250 15 Q* P* (f P' P* N P* N
1.00 0 F* N N N P* V P* P*
5086 t 0.250 262 ES ES ES ES EM EM Q Q
5456 e 0.250 200 EI EM EI ELN' Q EM P EI
h 0.250 287 ES ES ES ES ES EI ES ES
* Visual code—N = no pitting, exfoliation or appreciable attack.
P = pitting.
Q = questionable exfoliation (pit blister metallographic examina-
tion recommended to determine type of attack).
EM = exfoliation, mild.
EI = exfoliation, intermediate.
ES = exfoliation, severe.
' Control—low resistance.
' Control—medium resistance.
' Examined metallographically and passed test on this basis.
* Two specimens gave disparate results.

of 215 showed poor behavior. This same material, annealed, gave good
results—WLV, 15, and only pitting in the salt spray test.
Table 9 shows data for 5086 alloy similar to that given in Table 8 for
5456 alloy. Since the service failure was attributed to stress-corrosion
cracking, a stress-corrosion test was used to evaluate these materials, as
well as the exfoliation and intergranular corrosion tests [7,8]. The stress
relief treatment apparently reduced the susceptibility to stress corrosion
cracking to the extent that the material, so treated, passed the test. How-
ever, the 425 F (218 C) treatment did not reduce greatly the continuity
of the grain boundary precipitate and thereby did not eliminate the con-
commitant susceptibility to intergranular corrosion. The gap between
highly susceptible material (WLV = 204) and that which passes the
stress corrosion test (WLV= 193) is narrow, but both materials show
moderate to severe intergranular corrosion in both ASSET and SWAAT.
A borderiine material that was not in service (Sample 2) showed slight
pitting with an intergranular component in one of two laboratory tests,
and had a WLV of 134. Sample 1 after heating at 450 F (232 C) showed
slight to moderate pitting with an intergranular attack on the microscopic
scale. Since the material for this test was taken from the original sample
which was heavily pitted, this attack may not be characteristic of the 23
WLV but rather represents vestiges of prior structure and its exposure.

TABLE 8—Correlation of nitric acid weight loss with exfoliation and intergranular
corrosion behavior of 5i56-H3ll alloy plate in service and laboratory tests.
Exfoliation or Intergranular Susceptibility
Identification SVVAAT"
Surface Midp]lane
Service Boat Weight Loss,'
No. Frame mg/in.* In Service Visual Micro Visual Micro
i 4P 201 pitting and B P C I
7S 139 intergranular B I C I
d 492 corrosion B I B I
3 129 severe C Exf B Exf
ibs
lOP 138 exfoliation C Exf B Exf
6P 200 corrosion C Exf C Exf
4 7S 204 D Ext C Exf
7P 210 D Exf C Exf
lOP 217 D Exf C Exf
5 5S 127 D Exf B Ext
SS 123 D Exf D Ext
9S 138 B Ext A Exf
lOP 114 C Exf A Exf
i 299 D Exf D Exf
AAlot 2»' 180 A P + SI A P
(Reynolds shows s litht Exf)
AAlot i 216 C Ext C Exf
4 A' 15 A P A P
" Results of AA round robin test performed by Alcoa, Kaiser, and Reynolds, take i
from Ret 1.
Code—Visual rating of degree: A = no blistering or delamination.
B = slight blistering or "lipping" of pits.
C = large blisters or localized delamination.
D = severe delamination.
Microscopic examination
of type of attack: P = pitting.
P + SI = pitting plus slight intergranular.
I = intergranular.
Exf = exfoliation.
» 0.188-in. gage.
' 5456-0 annealed plate.
'' Reported by T. J. Summerson, Kaiser Aluminum.
" Nitric acid weight losses—performed by Reynolds Metals Co.

The conclusion from this series is that the nitric acid weight loss test is
as effective in evaluating 5086 alloy service failures as it is with 5456
alloy samples.

Long-Term Behavior
To show the broader application of the nitric acid test in the examina-
tion of service behavior, a series of samples that have been collected over
the past fourteen years is presented in Table 10. The samples are dis-
cussed briefly next.

Peraluman 50 (al-5 percent magnesium, 0.3 percent manganese)

Three samples were removed from a ferryboat which sails the Caribbean.
One sample (No. 1) was badly corroded, showing heavy exfoliation

TABLE 9—Correlation of nitric acid weight loss with stress-corrosion cracking and
intergranular corrosion heharior of 50S6-HS2 alloy sheet in service and laboratory tests.
Stress-Corrosion Cracking [8-9] and Intergranular
Susceptibility
Service Laboratory Test"
Weight Loss,
Identification mg/in.' sec ASSET SWAAT
5086-H3S sheet. failed by inter- failed P/IG-MOD-SEV P/IG-SEV
as produced. 204 granular corrosion
0.125-in gage: and stress-
Sample 1° corrosion cracking
Sample 2 ' 134 not in service passed N P/IG-SL
Above material
stress relieved
8 h at:
425 F
Sample 1° 193 passed P/IG-MOD-SEV P/IG-SEV
Sample 2 ' 10 passed N N
450 F
Sample 1" 23 passed P/IG-MOD P/IG-SL
Sample 2 ' 7 passed N N
° Taken from amphibious vehicle—welded ramp; chemical analysis, 4.36Mg.
' Commercial source: chemical analysis, 3.72Mg.
' Rating system—Visual/Microscopic;
P = pitting
IG = intergranular
N = no intergranular corrosion
SEV = severe
MOD = moderate
SL = slight

and corrosion product buildup. Metallographic examination revealed

severe intergranular corrosion. Etching a polished surface developed the
expected continuous grain boundary network. The WLVs were very high,
especially in comparison with those of the two control samples, which
had seen ostensibly the same service but had not corroded. It does not
seem reasonable to ascribe the diiference in behavior to the magnesium
content alone. The hardness and gages were comparable for three mate-
rials, so it is assumed the temper and production practices were the same.
Thus, on this evidence, it was judged that Sample 1 had probably been
exposed to elevated temperature aging somewhere in its history, but that
other two samples had not. It was suggested that the materials should be
replaced with a more resistant alloy and temper.

5083-0
These samples were removed from a full sized experimental bridge
section, which had suffered corrosion attack in the industrial marine
atmosphere of a large East Coast city while undergoing fatigue testing.
The corrosion product formed on inner surfaces where condensation
occurred but where rain could not wash off contaminants. The WLVs
were higher for specimens with heat affected zones caused by welding,

TABLE 10—Use of nitric acid weight lots to investigate precipitation phenomena in

aluminum-magnesium alloys.
Weight Loss,
Identification mg/in.* Description
Peraluman 50AI (5Mg,
O.SMn) cms-gage sheet:
Sample 1—5.16Mg 275 exfoliation and severe intergranular
443 corrosion in service
Sample 2—4.7lMg 157 not corroded in service
144
Sample 3—4.83Mg 168 not corroded in service
155
S083-0, 0.1S5 gage sheet as
produced:
Sample 1 22 pitting plus slight intergranular corro-
Sample 2 20 sion in service
Sample 3 41 Samples 3 & 4 contained weld beads and
Sample 4 27 heat affected zones.
Above, annealed, cold-water
quenched:
Sample 1 8 passed stress-corrosion test [8]
Sample i 8
Above, heated 7 days:
218 F (100 C) 156 passed stress-corrosion test
221 days 212 F (100 C) 380 failed stres.s-corrosion test
B083-Hm extrusions:
Sample 1 9 as produced
Sample 2 8 annealed and cold-water quenched
Sample 3 22 production material exposed to sun
(Phoenix, Arizona) 21 April-Sept. 27
5lt56-HllS extrusions:
Sample 1 10 as produced
Sample 2 7 annealed and cold-water quenched.
Sample 3 61 exposed to sun (Phoenix, Arii.)
21 April-27 Sept.
5i56-HX extrusions: experimental temper:
Sample 1 90 as produced
' Sample 2 8 annealed and cold-water quenched
Sample 3 85 exposed to sun as above
Ne-HH:
A15, Mg O.SMn 184 deck plating in naval patrol boat failed
by cracking and intergranular corrosion
in service
S08S-H113 welded plate: evaluation of weight loss samples:
As welded
Sample 1—0.388 in. 17 no corrosion attack in weld, hent
affected zone, or parent metal
Sample 2—0.600 in. 13
Heated 1 week 212 F (100 C)
Sample 1 114 intergranular attack in heat affected
Sample 2 108 zone; slight undercutting of welds at
rolled surface-weld interface
6083-H113 plate 0.3125 in.
Car A
Sample 1 (side of car) 43 samples removed from aluminum hopper
Sample 2 (end of car) 34 car body (Alumina Service); no suscep-
tibility to intergranular or exfoliation
attack in service; samples also passed a
stress-corrosion cracking test
Continued

TABLE 10—{Concluded)
Weight Loss,
Identification mg/in.* Description
CarB
Sample 1 (side of car) 27 (Petroleum Coke Service); slight pitting;
Sample 2 3t> maximum depth 8 mils; no evidence of
intergranular or exfoliation attack in
service
5083-11 lis plate 0.S1S5 in.: plates, 4 by 4 by 0.3125 in. were removed
Car A (few months in B5 from the sides of coke cars; samples were
service) removed and tests performed: no failures
Car B (three years service) 58 in the stress corro.sion test; overall eval-
Car C (three years service) ()S nation of panels was: superficial dis-
coloration and slight attack in service
(Average of two samples
for each car)
50H:l-H}]i plate O.noO in. panels, 4 by 12 by 0.500 in. were in-
stalled in steel cars used to haul petro-
leum coke—90 to 450 F
As produced (control) 14 slight pitting and traces of intergranular
attack in the nitric acid weight loss test
Car A (55 days) 37 in the nitric acid weight loss test, .sample
Car B (45 days) 41 showed more intergranular attack than
control—more grain boundary precipi-
tate than control
Car C (57 days) 30 pas.sed the stress corrosion test
Cars A,B,C (185 days) failed the stress corrosion test
As produced, then heated 103 failed the stress corrosion test; one of
a t 212 F (100 C) for one three specimens failed at atmospheric
month exposure stress corrosion test at Kure
Beach, N. C. [8]\ samples of this material
showed severe exfoliation in the SWAAT
test while the as produced condition
showed no exfoliation

but the atmospheric corrosion attack was no heavier in these areas than
on the rest of the surface. After lO-years' exposure at an industrial
atmospheric test site, the corrosion did not progress beyond the initial
slight pitting and intergranular attack. The WLVs indicate this material
to be corrosion resistant, and the subsequent performance bears it out.
Note that after prolonged elevated temperature aging, the WLV increases,
but susceptibility to stress corrosion cracking does not occur until a value
of 380 is reached.

5083, 5456 Extrusions

These samples were tested by the plant laboratory, located at Phoenix.
As can be seen from a comparison of the data, the heating which occurs
during the summer caused increases in the WLV, from 9 to 22 for 5083-
H112, and from 10 to 61 for 5456-H112. When the latter alloy was
produced with a small amount of randomly distributed precipitate in an
experimental temper, the condition was considered stable towards the
heating, since a continuous network did not occur and the weight loss
value did not increase.

N6-1/4H (J percent magnesium)

A sample removed from a German-built patrol craft, operated in the
North Sea, showed a weight loss of 184, in a plate that had failed by
cracking and intergranular corrosion (test by a foreign affiliate laboratory).

5083-HI13 Weldments*
This test was conducted to see if weld zones might become sensitized
to corrosion if exposed to elevated temperatures, such as the defrosting
procedures used with railroad coal cars. Even prolonged heating (one
week or 168 h) at 100 C (212 F) did not produce a severe weight loss,
and the extent of intergranular attack by the nitric acid was judged
superficial.

5083-H113 Plate*~Railroad Car Bodies

Several sets of samples were either removed from aluminum car bodies,
or aluminum test plates were carried in steel car bodies. The purpose was
to determine if the hot cargo caused heating of the metal to the extent of
producing extensive precipitation within the alloy. Since 5083 has some
fraction of magnesium in a supersaturated solid solution, this precipitation
is always a possibility.
The first set was removed from a car used in alumina service. It showed
WLVs of 34 to 43, which were typical of that gage and temper as pro-
duced in the late fifties. The cars had been in service over a year at the
time of the sampling. Since then, they have been in service over ten years.
No complaints or problems have developed in the corrosion behavior of
this material in this application.
The second and third sets were removed from cars in petroleum coke
service. The cargo is intrinsically more corrosive than alumina. The WLVs
again were moderate, the corrosion attack due to the coke liquors was
slight, and there was no intergranular component to the pitting. Some of
these cars had been in service three years. Comparison of WLVs from
these cars with one from a car in service for only a few months indicated
that no precipitation was occurring.
The last set is a series of plates that initially exhibited a very low
WLV, 14. In the field test, after a relatively short period, the WLVs
increased by a factor of two or more. The samples passed the stress-
corrosion test.
However, samples from the same cars, exposed for 185 days to hot
cargo (the actual test period was longer, as deadhead time was not
counted) did fail the stress corrosion test. WLVs were not obtained on
these specific samples, but heating the as-produced plates for one month

'This temper has been superseded by the H321 temper.

at 100 C (212 F) produced a WLV of 103, and specimens failed the

stress corrosion test.
This sample is one of a series that is included in a long range study
of 5083 alloy. When precipitation is produced by heating at 100 C (212
F) for one month, the metal also shows susceptibility to stress corrosion
in atmospheric tests. It also shows severe exfoliation in the SWAAT test.
The as-produced material, to date (eleven years) is immune to all these
problems, in immersion and atmospheric testing in Reynolds Metals
Company's comprehensive, multienvironment program [8].

Discussion of Examples (long-term behavior)

The precipitation of the aluminum-magnesium intermetallic compound
is a complex metallurgical process. Susceptibility to exfoliation or inter-
granular corrosion depends not only on the site and nature of the pre-
cipitate but also on the grain structure. Exfoliation can occur only when
the grains are elongated. Equiaxed grains produce intergranular corrosion
and deep pitting. Whether or not a material does corrode also depends
on its environment. The results from the previous quoted examples attempt
to put some of this behavior in its proper perspective. The following
factors are judged to be important:
1. The magnesium content of the alloy.
2. The degree of cold work.
3. The fabrication history.
4. The environment, including exposure to elevated temperatures for
prolonged periods.
There is only a slight difference in the magnesium concentration among
5456, 5083, and 5086 alloys. Yet, again and again, that slight difference
seems to be critical, especially under otherwise equivalent conditions of
cold work, exposure to elevated temperatures, and environment.
Cold work is an accelerator of precipitation. This, then, makes it a
negative factor in producing a corrosion resistant material. It also leads
to precautionary measures when a structure is fabricated. Our best judg-
ment at this time is that a structure with controlled precipitation, such as
produced in the HI 16 temper, is superior in its behavior to a structure
with all or most of the magnesium in supersaturated solid solution. Again,
this difference can be measured with the nitric acid weight loss test.
In addition to cold work, structures may be "warm-formed" or other-
wise exposed to elevated temperatures, as in a welding process. The
nitric acid test is a convenient measure of the effect of these processes.
When it comes to judging the severity of environments, it can only be
suggested that, first, concentrated nitric acid is representative of a real
life environment for many aluminum structures, used in a chemical plant.
Secondly, nitric acid is a "most severe" environment compared to immer-
sion in seawater, exposure to marine or industrial atmospheres, or other

chemical solutions in which aluminum-magnesium alloys may be expected

to offer excellent service. The nitric acid weight loss test can be used
to detect subtle changes in the metal structure and does so prior to the
onset of susceptibility as measured by other, more conventional laboratory
tests.
Despite the complex behavior and multifactored situation described
above, the nitric acid weight loss test provides a quantitative scale covering
all degrees of behavior. This scale may be used for "pass-or-fail" judg-
ments based on the expectation of service performance, or it may be used
to subdivide materials into pass, fail, and "borderline" cases.
I conclude that 5086 and 5456 alloys in the HI 16 and H117 tempers
with weight loss values less than 50 are resistant to exfoliation and inter-
granular corrosion, while those materials with weight losses greater than
50 to 100 are suspect. The degree of susceptibility increases with the
amount of the weight loss, and materials with values of 200, 300, or 400
should be increasingly susceptible to exfoliation or intergranular corrosion.
Acknowledgments
The permission of Reynolds Metals Company to publish these proce-
dures and results is gratefully acknowledged.
The development of this procedure was, in a large part, due to the
careful work of the late H. B. Romans. The assistance of L. E. House-
holder, is acknowledged, in developing the procedure for plant use, and
encouraging its use in plant laboratories.
The contributions of J. S. Prestley and A. I. Kemppinen are also
acknowledged.
I appreciate the contribution of the various members of the AA Com-
mittee and especially acknowledge those who granted their permission
for me to include their test results. The Committee members are. S. G.
Epstein, The Aluminum Association, secretary; P. W. Jeffrey, Alcan Re-
search and Development, Ltd,; F. E. Loftin, Reynolds Metals Co.; J. F.
Pashak, Dow Chemical Co.; B. R. Pathak, Alcan Research and Develop-
ment, Ltd.; W. C. Rotsell, Harvey Aluminum, S. J. Sansonetti, Reynolds
Metals Co., chairman; M. B. Shumaker, Alcoa Research Laboratories;
D. O. Sprowls, Alcoa Research Laboratories; and T. J. Summerson,
Kaiser Aluminum & Chemical Corp.

APPENDIX
Nitric Acid Weight Loss Procedure
1. Prepare specimens measuring 2.000 by 0.250 in. by gage with the 2 in.
dimension parallel to the rolling direction. If the gage exceeds 1 in. reduce
to Vi thickness while retaining one original surface. Machine all sawn surfaces.
2. Smooth all edges with a fine file or fine emery cloth.

3. Measure all three dimensions to the nearest 0.001 in. with a micrometer.
Calculate the total surface area.
4. Immerse the specimen in 5 percent by weight sodium hydroxide (NaOH)
solution at 180 F (82 C) for 1 min followed by a distilled water rinse, a
30 s immersion in concentrated, reagent grade nitric acid, and a distilled
water rinse.
5. Allow the specimen to air dry completely. Do not wipe dry with a rag
or paper towel. From this point on in the procedure, handle the specimens
with tongs or tweezers.
6. Weigh the specimens to the nearest 0.1 mg.
7. Fully immerse each specimen in 80 ml of concentrated nitric acid,^
contained in a 100-ml beaker." It is suggested that beakers without spouts
be us2d with watchglasses as covers. Place the beaker in a water bath con-
trolled to 30 ± 0 . 1 C (86 F).
8. After 24 h, remove the specimens, rinse in distilled water and brush
with a stiff nylon bristle toothbrush to remove loosely adhering particles,
then allow to dry.
9. Weigh the specimens and determine the weight losses.
10. Determine the weight loss per unit area and express it in terms of
mg/in.-.

References
[/] Sprowls, D. O., Summerson, T. J., Romans, H. B., and Sansonetti, S. J., "The
Aluminum Association Tentative Exfoliation Test for Al-Mg Alloys for Boat
and Ship Hull Construction," TTie Aluminum Association, New York, 1969.
[2] Federal Specification QQ-A-00250/I9 and QQ-A-00250/20—5086 and 5456-
H116 and HI 17 Sheet and Plate.
[3] Romans, H. B., Maleriah Research and Standards, MTRSA, Vol. 9, No. 11,
1969, pp. 31-34.
[4] Brooks, C. L., Naval Engineers Journal, NVEJA, Aug. 1969, pp. 29-32.
[5] Craig, H. L., Jr., and Loftin, F. E., Proceedings, 26th Annual Conference, Na-
tional Association of Corrosion Engineers, Houston, Tex., 1970.
[6] "Aluminum with Food and Chemicals." 2nd ed.. The Aluminum Association,
Feb. 1969, p. 59.
[7] Craig, H. L., Jr., and Romans, H. B. in Stress Corrosion Testing, ASTM STP
425, American Society for Testing and Materials, 1969, pp. 51-65.
[8] Romans, H. B. and Craig, H. L., Jr., in Metal Corrosion in the Atmosphere,
ASTM STP 435, American Society for Tes;ing and Materials, 1968, pp. 61-82.
° Reagent Grade, 70 percent by weight. Fresh acid should be used for each test.
"Beaker, Berzelius (Corning 1040).

Simplified Exfoliation Testing

of Aluminum Alloys

REFERENCE: Sprowls, D. O., Walsh, J. D., and Shumaker, M. B.,

'Simplified Exfoliatioa Testing of Aluminum Alloys," Localized Corro-
sion—Cause of Metal Failure, ASTM STP 516, American Society for
Testing and Materials, 1972, pp. 38-65.
ABSTRACT: A program was undertaken to develop accelerated tests to
determine susceptibility to exfoliation in aluminum alloys. Experimentation
was concentrated on total immersion procedures to develop a simple,
easily controlled test. Chloride solutions did not cause exfoliation during
immersion; however, formulations of chloride-nitra'.e solutions were found
that produced severe exfoliation of highly susceptible Al-Mg-Mn alloy
specimens in only one or two days. Investigation of the most effective
chloride-nitrate ratios, concentrations, solution pH, and temperature led
to the development of a new, simple, and rapid exfoliation test for Al-
Mg-Mn alloys. Additional investigation resulted in the development of
similar test media for Al-Cu-Mg, Al-Zn-Mg, and Al-Zn-Mg-Cu alloys, and
showed that the optimum test conditions differ for the separate alloy
families.
Test data are presented for alloys 2024, 5086, 5456, 7005, 7075, and
7178 in metallurgical conditions that provide comparisons of high and low
resistance to exfohation. Comparisons with exposures to natural environ-
ments also are included.
KEY WORDS: corrosion, exfoliation corrosion, seawater corrosion, corro-
sion tests, salt spray tests, immersion tests (corrosion), aluminum alloys

Exfoliation corrosion, sometimes referred to as layer, stratified, or

lamellar corrosion, is a form of corrosion resulting from a relatively rapid
lateral attack along electrochemically anodic strata parallel to the metal
surface. In the more familiar occurrences of exfoliation corrosion, attack
progresses along grain boundaries. Because it occurs most readily in
alloys and tempers that have a relatively low resistance to stress-corrosion
cracking in the short-transverse direction, exfoliation sometimes is re-
garded as a special form of intergranular attack or of stress-corrosion
cracking. Such a limited perspective of exfoliation, however, is an over-
simplification. Experience with corrosion testing a wide variety of alloys

^ Section head, technician, and senior research engineer, respectively, Chemical

Metallurgy Division, Alcoa Research Laboratories, New Kensington, Pa. 15068.
38

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 12

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
SPROWLS ET AL ON SIMPLIFIED EXFOLIATION TESTING 39

and tempers at Alcoa Research Laboratories has shown that exfoliation

also can result from in/ragranular corrosion, and that not all products
that are susceptible to //i/ergranular corrosion or stress-corrosion cracking
will develop exfoliation. Thus, the requisite of a highly directional micro-
structure is more essential than a specific mode of corrosion. Another
requisite is a specific type of corrosive environment.
In recent years the development of exfoUation-resistant alloys and
tempers [1-5]" has brought into focus a need for improved accelerated
corrosion tests. There are increasing numbers of U. S. Government speci-
fications for aluminum alloy products that include exfoliation test require-
ments [6,7]. A suitable accelerated exfoliation test must not only be rapid,
but it must also enable the investigator to distinguish between alloys and
tempers of differing susceptibility to exfoliation and to rate them in the
same order as service environments. Liddiard et al [8] in 1960, reported
that "no satisfactory accelerated test for layer corrosion has been found."
Liddiard and other investigators [9] at that time used intermittent spray
exposures to neutral or acidified sodium chloride (NaCl) solution. The
main disadvantage of such tests is that they require considerable time,
usually several months. Although improved tests have been developed in
recent years by Budd and Booth [10], Lifka and Sprowls [11], Romans
[12], and Ketcham [13], these tests require sophisticated apparatus, and
the test results can be subject to considerable variability (Fig. 1).
Liddiard et al [8], after experimenting with aluminum-copper (Al-Cu)
alloys immersed in solutions of 3 percent NaCl, 6 percent NaCl -f- 0.11
percent hydrogen peroxide (H2O2), 0.5 A^ ammonium chloride (NH4CI)
and 3 percent NaCl buffered at pH 11, concluded that layer corrosion
does not occur under total immersion conditions because corrosion prod-
ucts are formed away from the site of attack, and there is always con-
siderable attack on the grains themselves. Indeed, this was the opinion at
Alcoa Research Laboratories at the time the improved cyclic acetic acid
acidified salt spray test was developed [11]. However, a renewed investiga-
tion of immersion tests prompted by U. S. Navy interest in an accelerated
exfoliation test for screening aluminum alloy boat hull plate led to the de-
velopment of some very effective and reproducible test procedures. The
purpose of this paper is to present results of this investigation and to com-
pare test results from immersion tests with tests in service environments.

Procedure and Results

General Test Conditions
Many years of testing experience and consultation on customer service
problems have shown that exfoliation corrosion is most likely to occur in
'The italic numbers in brackets refer to the list of references appended to this
paper.

certain types of environments. More aggressive environments involve

chloride ions from sources such as seawater combined with alternate
wetting and drying. Other accelerating factors are acidity, elevated tem-
peratures, and galvanic stimulation. Tests by spray techniques generally
incorporate some or all of the aforesaid conditions, and similar conditions
were investigated in the evaluation of corrodents for immersion testing.
Test specimens IVi by 3 in. were sawed from rolled plate with the
1V2 in. dimension parallel to the rolling direction. Edges of the specimens
were dressed by machining or sanding, after which the specimens were
etched 1 min in 5 percent by weight sodium hydroxide (NaOH) solution
at 82 C (180 F), rinsed in water, desmutted 30 s in concentrated nitric
acid (HNOs) at 27 C (80 F), rinsed in distilled water, and air dried.
The experimental solutions were made with reagent grade chemicals and
distilled water. Tests were performed in glass beakers with a volume of
solution sufficient to provide a minimum volume-to-metal surface ratio
of 40 ml/in.-. Solution temperatures were maintained at the indicated
temperature within ± 1 C.

Aluminum-Magnesium (Al-Mg) (SXXX) Alloys

Initial emphasis was on 5XXX alloys. Four items of plate were se-
lected to provide a variation in microstructure and susceptibility to exfoli-
ation based on prior performance in service and in spray tests. The
microstructures and a description of the spray test performances are
shown in Fig. 2.
Exploratory Tests of Various Corrodents—Acetic acid acidified chlo-
ride solutions of the following formulations have been shown to develop
exfoliation of susceptible materials in seven days of intermittent spray
exposure at 49 C (120 F).
1. Five percent NaCl solution buffered to pH 3 with acetic acid [11]
2. Synthetic seawater, ASTM D 1141-52 (without heavy metal addi-
tions) plus 10 ml of glacial acetic acid per liter of solution [12].
Immersion in these solutions at 49 C (120 F) for seven days did not
develop even a trace of exfoliation of low-resistance material. Exploratory
immersion tests, therefore, were extended to a variety of acidified chloride,
chloride-acetate, chloride-nitrate, and miscellaneous solutions. A number
of these corrodents and visual ratings of the exfoliation effects are listed
in Table 1. It should be noted that in these exploratory tests the objective
was to find a solution that would develop exfoliation in susceptible mate-
rial within a 5-day period at temperatures not exceeding 82 C (180 F).
The most promising corrodents found were chloride-nitrate solutions with
an appropriate combination of chloride and nitrate ions. It is known that
exfoliation will occur in certain of the other solutions with longer expo-
sure periods, and it is likely that under suitable conditions exfoliation
might also occur in other types.

Wit ' . . ; • "J '

SSfciS£^'*«&-;&."r i'.v---:fsc F « H i y a ° - - - ^ LJ^I - • •• ••**

(a) 5456-H117 P
(ft) 5086-H32 P
(c) 5456 control Q
id) 5456 control EM
FIG. 2—Microstructures {phosphoric acid etch) of jour items of 0.250 or
0.375-in.-thick hot-rolled plate of Al-Mg alloys with differing resistance to ex-
foliation rated as above by an acidified cyclic salt spray test fS.ll]. {For explana-
tion of exfoliation rating refer to footnote a of Table 1.

Preliminary tests of chloride-nitrate solutions were performed with a

solution containing 4 M NaQ + O.S M potassium nitrate (KNO3)
+ 0.1 M HNO3 (pH 0.4) developed by Zaretskii [75] for stress corrosion
testing Al-Cu alloys, and adapted for exfoliation testing by Alcan Re-
search and Development, Ltd.* This solution was ineffective at room
temperature, but at 66 C (150 F) produced mild exfoliation of the 5456
control specimen. Fig Id, and rapid general corrosion and thinning of
the other test materials identified in Fig. 2. Extensive testing of additional
chloride-nitrate solutions, including variations in solution pH and tem-
perature, led to a solution containing 1 M NH4CI + 0.25 M ammonium
nitrate (NH^NOg) -f HNO3 at pH 3 that at a temperature of 66 C (150 F)

'Jeffrey, P. W., personal communication. 22 Oct. 1968.

t o t<> ZCLP-BHCM^B-SH ?:Z5?:5?;S?;S5Z;5?;ZS5 5?:;Z?;5?; J?;PH^

S "«
.2 §»<
•g-S ti
S; PHft-BHft^04 PH PH s?;2;5'.Z!z;zZft-^;^ZZ;z;:z; Izift-ft-
U a.

5^ a, cu OH 0-PH p-SH 5^5',>^;;?;J^:^Za-Z5^/5:z2::2;

3 ..- -,
o S 2? l>t~l-l>t-t-t-t- I>>0"*l010>0'00t-«'OOI5«>0
o.- _

oooooooo oooooooooooooo ) O O

CO 00 9 t 04 O m 00 «5 «
X
Q. GO-HO(pH»flao^i-i 50 '\> '>» - t -f -^ -^ a* (NO*-*

5
I
+
O O "O "^

6 6=;- i I
^ + + + =•
H
o
^§.98
o o — ri
+
PS 4 J * J •*-) ^
O
•< ..Z
« en tn ^
rt rt c8 W5
s ** - 1 " . 1 9 9 - 4) o aj w
en M w Q
• ^ ' ^ '*j -f-
§6 d
« u
tcCsc ^
o; oj V
4^ •«-) W '
1+ +
Cv Rl !C A n •-SO C
3^ gU
s"« ?: z X z ?; ;2; te m to K^ s cs a

o »c »o ^ ^ « 5 « ) 0 ) 0 < 9 < " o* »* »»'^

-""do ^ • * ' • * ^

to to Z:<yZ^S'.5?;2;5?;Z;J?;a-Z;^ ;?;Z;ZZ^;;2;

6
3
-a
1
m isC
Zw?5?;5^5?;a''5:2;>5&.?'.Z J?;^;?^??;^?;
^ b<

.r
S S
* i

>-; 2 1^tic
Ifi

11^ «5«J'Ii^O^^^«fiO>0'0«S

Q..S «

o o o o o o
Qt tQ U> U> ^ 'O

>> Si M»

a .3 *MS
0
CO

>-• « « c> ^ t» C
0
a.
•M
at) s0
XI u
4 ^

B.Ji«i
its.
•3
£
-2 =
C.2., si
«•;= £ ^
2 • = 2. X :
II II II II
Zf^O'g 3S|
oo o 0-? e
o
o teg
za s'A '^ Z ze
Z >. ZZJJZ
^ «»
M5
»» •§§ = => 6<s
«
O
s
+ • 0 O + + , + 4-
+ O
^^ + ++U " 4 . - - d
01 S
;2: Z ^o^^ZZZZ
^ ^ COoK
' 5 lO O rt a« PI n
zzz;^
^ ^ ^ o*

caused very little general corrosion and produced severe localized ex-
foliation of the lower resistance 5456 control material. Although the solu-
tion did not actually produce exfoliation of the less susceptible 5456
control specimen. Fig. 2c, it appeared most worthy of further evaluation
as a corrodent for exfoliation testing of Al-Mg alloys by immersion.
Optimization of Chloride-Nitrate Solution Test—^Experiments with in-
termittent immersion methods to get oxygen to the surface of the test
specimen to simulate the intermittent spray effect were unsuccessful. On
the other hand, aeration of the solution with fine bubbles of compressed

(a) Intermittent immersion (10-min immersion; 50-min drying).

(b) Continuous immersion, quiescent.
(c) Continuous immersion, aerated.
FIG. 3—Photograph (approximately Xl) of specimens of 5456 control. Fig. 2d,
exposed five days by the methods noted above to a solution containing 1 M
NH.CI + 0.25 M NHiNO, + HNO, to pH 3 at 66 C (150 F) (pH adjusted daily).

air caused a striking increase in the exfoliation of the lower resistance

control specimen (Fig. 3). However, oxygen saturation of the chloride-
nitrate solution with H2O2 instead of bubbling with air was more effective
and decidedly easier to control.
The addition of magnesium chelating compounds, such as potassium or
ammonium tartrate or tartaric acid, further increased the effectiveness
of the test solution with less susceptible plate samples (Fig. 4).
The immersion test solution {e) described in Fig. 4 was compared with
the specification salt spray test [6] in a round robin program by an Alumi-
num Association Task Group to develop a standard exfoliation test for
aluminum-magnesium (Al-Mg) alloys. Tests performed in six different

(a) 1 M NH.C1 + 0.25 M NH.NOa + HNOa to pH 3.

(fc) 1 M NH.C1 + 0.2S M N H . N a + 1 0 ml/liter H.O2 + HNOa to
pH 3.
(c) 1 M NHiCl + 0.25 M NH.NO3 + 0.005 M K^CH.Os + 10 ml/
liter H2O2 + HNO3 to pH 4.5.
(d) 1 M NH.C1 + 0.25 M NH^NOj + 0.01 M KoC.H.Oe + 10 ml/liter
H2O2 + HNO3 to pH 4.5.
(e) 1 M NH,Cl + 0.25 M NH.NOa + O.Ol M K^C.H.Oe + 10 ml/liter
H2O2, pH 5.4.
(/) 1 M NH.Cl + 0.01 M K^C.H.O. + 10 ml/liter H.O. + HNO. to
pH 4.5.
Solution (d) included in Interim Federal Specifications QQ-A-
00250/19, 20, 31 Dec. 1968.
FIG. 4—Photograph (approximately XO.5) of specimens of 5456 control, Fig.
2c, exposed three days at 66 C (150 F) in solutions varied as noted above to
evaluate the relative effectiveness of specific components. (Solution changed daily).

laboratories demonstrated the superior reproducibility of the immersion

test procedure. The immersion test not only effectively produced exfoli-
ation of susceptible materials but also caused objectionable pitting of
some specimens not susceptible to exfoliation. Subsequent studies of the
test variants showed that the pitting attack could be minimized by modifi-
cations such as: reduction of chloride:nitrate ratio (Fig. 5), increasing
the temperature (Fig. 6), and shortening the exposure. Certain of these
modifications were incorporated into two similar test procedures (Fig. 7).
The procedure referred to subsequently in this article as the "ASSET"
test is listed next:

FIG. 5—Photograph (approximately xO.67) of Al-Mg alloy snecimens with dif-

fering resistance to exfoliation exposed two days at 66 C (150 F) to solutions of
varied chloride-nitrate ratio.

FIG. 6—Photograph {approximately XO.67) of Al-Mg alloy specimens with dif-

fering resistance to exfoliation exposed two days at temperatures of 49, 66, and
82 C (120, ISO. and 180 F).

FIG. 7—Photograph (approximately y.0.67) of Al-Mg alloy specimens with dif-

fering resistance to exfoliation exposed to two slightly different versions of the
ASSET immersion test.

Solution:0.75 M ammonium chloride.

0.25 M ammonium nitrate.
0.01 M ammonium tartrate.
10 ml/liter hydrogen peroxide (30 percent H2O2).
Temperature: 80 C (180 F).
Volume-to-metal surface ratio: 40 ml/in.^, min.
Duration: 24 h.
The other procedure (Appendix A) has been similarly designated the
"ASSET" test by the Aluminum Association Task Group which recently
recommended it as the preferred method for exfoliation corrosion testing
of 5086 and 5456 alloys.
Tests on Aluminum-Zinc-Magnesium-Copper (Al-Zn-Mg-Cu) {7XXX)
Alloys—Preliminary testing of 7178 alloy test specimens by the ASSET
procedure showed that it was too severe for these less corrosion resistant
alloys. Additional studies, therefore, were initiated to modify the test for
this alloy system. Specimens were obtained from a sample of %-in.-thick
7178 alloy plate that was solution heat treated, quenched in cold water,
and artificially aged to provide a low, medium, and high resistance to
exfoliation.
Lowering the temperature of the ASSET solution to 27 C (80 F)
materially reduced the severity of the corrosion. Other experiments showed
that the ammonium tartrate addition was not essential. With the reduced
test temperature and deletion of the ammonium tartrate, tests were per-
formed to verify the optimum ratio and concentration of the ammonium
salts.
Variations in chloride : nitrate molar ratios ranging from 10 : 1 to 2 : 1
(the nitrate concentration was fixed at 0.25 M) showed that the most
favorable results were obtained with ratios of 4 or 5 : 1. Higher ratios
reduced the exfoliation of the low-resistance material and caused some
blistering and slivering of the high-resistance specimen. Lower ratios re-
duced and even eliminated exfoliation of the low-resistance plate. Lower-
ing the concentration of the ammonium chloride and ammonium nitrate
salts, while retaining the 4 : 1 ratio, reduced the aggressiveness of the
solution but did not affect the exfoliation of the susceptible specimens
(Fig. 8).
A solution of this type (Test A) and the Zaretskii solution mentioned
above (Test B) were tested by a task group formed under ASTM Sub-
committee GOl.05.02 to evaluate improved exfoliation tests for Al-Zn-
Mg-Cu alloys. At the conclusion of the round robin test program per-
formed in eight different laboratories, it was the consensus of the task
group* that the results of the immersion tests agreed with the results of
' The results of this task group study are given in more detail by S. J. Ketcham
and P. W. Jeffrey in another paper in this symposium.

FIG. 8—Photograph (approximately xH of midplane (T/2) of 7178 alloy ex-

posed 16 h at 27 C {80 F) to varied concentrations (4:1 chloride : nitrate ratio)
of the ASSET solution modified for exfoliation of Al-Zn-Mg-Cu alloys.

the salt spray test (Fig. 9), and the Zaretskii solution (Test B) gave the
most reproducible test results. The name "EXCO" was adopted by the
task group for this test procedure (Appendix B).
Experiments with variations in the EXCO test yielded the following
additional observations:
1. Substitution of 0.5 M sodium nitrate (NaNO.,) for 0.5 Af KNO;,
did not affect the performance.
2. Reduction of the concentration of the salts by 50 percent also did
not affect the performance so long as the 8 : 1 ratio of NaCl to KNO.^
(or NaNO.3) was maintained.
Aluminum-Zinc-Magnesium (Al-Zn-Mg) (7XXX) Alloys
These alloys have a very good resistance to corrosion, and exfoliation
corrosion has not been reported to be a problem with the artificially aged
tempers. Exfoliation will occur in F and W temper material, and it has
been shown to occur in the heat affected zone of weldments exposed in
natural environments such as seawater. Weldments of X7005 alloy sheet
which had exhibited these exfoliation characteristics in tests conducted
by intermittent immersion in a 3.5 percent NaCl solution were selected
to determine whether such exfoliation susceptibility could be detected by
immersion test procedures.
Exposure for a period of 24 h by the EXCO procedure produced

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 12:32:22 EST 2016
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions au
Test A: 0.50 M ammonium chloride.
0.125 M ammonium nitrate.
10 ml/liter hydrogen peroxide (30% HaOa).
Test B: 4.0 Af sodium chloride.
(EXCO)
0.5 M potassium nitrate.
0.1 M nitric acid.
FIG. 9—Photograph {approximately XO.4) of midplane specimens (T/2) of
7178 alloy exposed to two different immersion tests and the acidified salt spray
exfoliation test {QQ-A-OO250/24, 25) [11].
52

exfoliation in the susceptible specimens and provided a clear delineation

between susceptible and resistant specimens. Figure 10 shows the appear-
ance of welded specimens after exposure to the EXCO test, the 3.5 per-
cent NaCl alternate immersion test, and the acidified salt spray exfoliation
test. It is evident that the immersion exfoliation test provides a better
correlation with the salt water exposure than the acidified spray test.
Aluminum-Copper (Al-Cu) (2XXX) Type Alloys
Material in the naturally aged tempers, T3 and T4, of these alloys
have shown considerable susceptibility to exfoliation in the atmosphere,
and this susceptibility is revealed readily by acidified salt spray tests. In
fact, slowly quenched 2024-T42 control panels are employed as controls
for the spray test cabinets. However, as illustrated in Fig. 1, appreciable
variability has been encountered in the spray tests, even with such highly
susceptible material. Therefore, tests were conducted with these materials
to determine whether exfoliation could be developed by immersion test
procedures.
Room-temperature exposure in both the ASSET solution for 5XXX
alloys, and the EXCO solution for 7XXX alloys effectively developed
exfoliation in the 2024-T42 control specimens. The ASSET solution
developed exfoliation in less time than the EXCO solution (one day
versus four days).
Comparison with Natural Environments
When using highly accelerated corrosion tests, one is naturally con-
cerned that the test conditions may be too drastic and produce unrealistic
corrosion effects. Thus, it is essential that performance of known materials
in the accelerated test be compared with that of the same material in
service or in a service type of environment. The following section describes
several examples illustrating a good relationship between results of immer-
sion type exfoHation tests and exposures to service environments. Exfoli-
ation was produced in the immersion tests duplicating that which occurred
in actual service or in service type environments. And of equal signifi-
cance, alloys and tempers that did not exfoliate in the service environments
likewise did not exfoliate in the immersion tests.
5086-H32 and 5456-H321 WeldmeMs in Seawater
A weldment of 5456-H321 that developed exfoliation of the parent
plate during a 2-year period of immersion in seawater at Miami, Fla.,
is shown in Fig II. The microstructure of the plate is also shown, and
the marked outlining of the grain boundaries with precipitated Al-Mg
constituent indicates that exfoliation could be expected in some environ-
ments (compare with Fig. 2c). The test coupons exposed to the immer-
sion tests indicated an intermediate susceptibility. A specimen of welded

•owe
~: ~ B .
"a »--fec
<^ ~ «i
o 5i S S
w s -5

"S § « a

^ ^ ^ >0 k.
U -t3 ^ «^ ^

•^
^ 00 >^ (O

S o-g 1 *
"^ o s 'H •«
o « o gii
« u =^ I- §
^1
C O
^^
i ill

J''W • ^JA' ' /;•" ' ^ „ - ' VS ••""'••'-or '^•-

':•* . 1 1 . . :

mjw^'' ••••ntsn. %>-

aft Jnr.ii-

• • % • ,

FIG. 11—Correlation of immersion test and seawater exposure on 5456-H321

welded specimen. Inset illustrates the microstructure of the plate and the exfoliation
attack which was similar for the accelerated tests and the service environment.

5086-H32 immersed for a 4-year period in seawater without exfoliating

in shown in Fig. 12. The test coupons exposed to the immersion test like-
wise did not exfoliate, although the coupon immersed in the specification
test solution developed marked pitting.
5456-H321 Alloy Boats in Lake Maracaibo
Severe corrosion of the exterior surfaces of 5456-H321 hull plates
of two crew boats occurred during 2-years' service in the brackish water
of Lake Maracaibo, Venezuela, in the absence of normal protective sys-
tems. Figure 13 shows the severe exfoliation observed on Boat A, and
Fig. 14 shows the deep localized pitting observed on Boat B. A com-
parison of microstructures readily indicates that the continuity of precipi-
tate in the plate from Boat A was much greater than in the plate from
Boat B and that some difference in the type of corrosion could be antici-
pated. The immersion test coupons showed a performance identical with
that observed in service, as illustrated in Figs. 13 and 14. Both forms
of corrosion could have been prevented on the crew boats by the proper
use of established protective systems.

Exfoliation Resistant Tempers of 5456 Alloy [6]

Rolled plate in exfoliation resistant strain hardened tempers of 5456
and similar alloys can be produced by proprietary techniques [76] that

tBKHntae tMt

MoStftea

tMl'Aur

FIG. 12—Correlation of immersion lest and seawater exposure on 5086-H32

welded specimen. Inset illustrates the microstructure of the plate and the pitting
attack which was similar for the accelerated tests and the service environment.

*.••"?-'

&;4?T^ "^mrmrn^rrn. "*#st.

FIG. 13—Correlation of immersion test and service on a 54S6-H321 welded

specimen of Crew Boat A in Lake Maracaibo. Photomicrograph illustrates the
microstructure of the plate and the exfoliation attack which was similar for both
the accelerated test and the service environment.

. 1 * * - - . •

FIG. 14—Correlation of immersion test and service on a 5456-H321 welded

specimen of Crew Boat B in Lake Maracaibo. Photomicrograph illustrates the
microstructure of the plate and the pitting attack which was similar for both the
accelerated test and the service environment.

achieve either of two metallurgical conditions: (a) homogeneous Al-Mg

solid solution or {b) homogeneous Al-Mg solid solution plus random
dispersion of MgzAls precipitate. Intermediate conditions that have
heterogenous solid solution or selective precipitation along deformation
bands and grain boundaries or both may have some degree of suscepti-
bility to intergranular corrosion or exfoliation. The greatest tendency for
exfoliation is associated with a high degree of continuity of MgzAlg pre-
cipitate along deformation bands and grain boundaries. Examples of these
metallurgical conditions, together with the results of accelerated exfoli-
ation tests, are shown in Fig. 15. Neither plates in the H117 nor in the
H343 temper showed any tendency for exfoliation in the ASSET immer-
sion test or in the salt spray test.

2024-T4 and T6 Extrusions in Seacoast Atmosphere

Because it is difficult to quench thick sections of 2024 alloy extrusions
rapidly enough to keep all of the copper in solid solution, it is not un-
common to observe exfoliation with the naturally aged tempers, T3 and
T4, after several years' exposure in the atmosphere. Figure 16 shows
the exfoliation that occurred with a 2024-T4 extrusion during a 4-year

4
i 2J
"-I
4
i

it t1

' ^ tO«4-T 4 <-*V:;J: ap^.^^'- 2024-T 6

FIG. 16—Correlation of immersion test and seacoast atmospheric exposure.

exposure to the seacoast atmosphere at Point Judith, R. I. This problem

can be corrected by employing the artificially aged T6 and T8 tempers.
The ASSET immersion test, requiring but a 24-h exposure at room tem-
perature, produced exfoliation in the T4 temper and pitting in the T6
temper specimens. Therefore, the immersion test could prove very useful
both to determine whether an Al-Cu alloy is susceptible to exfoliation,
and, if so, whether the corrective treatment to be employed is valid.

7075-T6 and T76 Extrusions in Seacoast Atmosphere

The resistance to exfoliation of high-strength heat-treatable aluminum
alloy products is principally a function of variations in grain shape, quench
rate, and the artificial aging treatment. A markedly directional grain
structure and a slow quench rate are generally adverse in the case of

material in the T6 temper, and exfoliation may occur in these products

within a year in a seacoast atmosphere. However, exfoliation attack can
be virtually eliminated by the use of the resistant T76 temper. Figure
17 illustrates the exfoliation that occurred in a 7075-T6510 extruded
section during a 1-year exposure. It also demonstrates the excellent re-
sistance afforded by a 7075-T76510 section during a 3-year exposure
to the seacoast atmosphere at Point Judith, R. I. The EXCO immersion
test effectively demonstrated similar corrosion characteristics in these
T6510 and T76510 temper sections and required an exposure of only
48 h duration.

Ittsers Ion

., } yeara
Atnospnere

FIG. 17—Correlation of immersion test and seacoast atmosphere. Midplane of

extruded wing planks (.0.7 in. thick).

Discussion
With the optimum immersion test conditions for an alloy type, speci-
mens of most products develop easily distinguishable forms of corrosion
such as exfoliation, pitting, or etching. Exfoliation may initiate as blisters
that progress into lamellar corrosion, and, some specimens that approach
borderline susceptibility to exfoliation may develop an "in-between" form
of corrosion. The term "pit-blisters" has been used to describe this form
of corrosion (Figs. 12 and 18), where the attack has the appearance of
small blisters that have opened into pits with residua] undercutting
around the edges. Although, generally speaking, pit-blisters can be due to

FIG. 18—Visual appearance of specimens of 5086-H32 alloy, Fig. 2b, after 1,

2, 3, and 7-days' exposure to specification immersion test.

more than one microscopic type of attack, their occurrence in this very
aggressive test denotes negligible susceptibility to exfoliation. One way
to gain assurance of this is to extend the exposure period and watch for
a change in the form of corrosion. This procedure is illustrated in the
following example for 5XXX alloys, and the same procedure is applicable
to other alloys. As indicated in Fig. 18 the pit blisters present after the
first day of exposure became larger with extended exposure but did not
progress into exfoliation. Compare this with the specimen in Fig. 19
where the pit blisters present after the first day grew with extended expo-
sure into larger blisters and distinct exfoliation. (The difference in per-
formance of the specimens illustrated in Figs. 18 and 19 is consistent

FIG. 19—Visual appearance of specimens of 5456 control, Fig. 2c, after 1, 2,

3, and 7-days' exposure to specification immersion test.

with the greater tendency for the precipitated Al-Mg constituent to out-
line the elongated grain structure in the 5456 alloy specimens shown in
Fig. 19). To determine the path of the corrosion and its mechanism,
metallographic examination of the corroded specimens is required. In
these examples, the 5086-H32 alloy specimens revealed pitting and
undermining pitting, whereas the 5456-H321 alloy specimens showed a
combination of undermining pitting and intergranular corrosion (Fig. 20).
It is noteworthy that the intermittent salt spray test (QQ-A-00250/20)
[7] of the latter specimens did not indicate exfoliation susceptibility.
It is evident that the test methods just described involve aggressive
corrodents, and it is believed that results of these tests can be used to
predict the serviceability of aluminum alloy products in very corrosive
marine environments. In relatively mild marine environments, and in many
other types of service, the corrosion effects shown in these tests may
never occur. There is always a considerable risk associated with the use
of highly accelerated corrosion tests such as these, and the results of
such tests must be interpreted with caution. While the comparisons with
service experience just reported are very encouraging, still more compari-
sons are desirable to relate the test results with specific environments.
It is important that the optimum formulation of test solution and choice
of test duration be used for a given type of alloy. Although further ex-
perimentation with these test procedures may lead to additional refine-

*-••• * « • • . -•'•^••'''•^j«»ii.»i«.«yi»jB>t«i.^'«i.yii.

u SWAAT IMMERSION

FIG. 20—Comparison of attack after exposure of specimens of 5456 control,

Fig. 2c, to specification acidified salt spray test and immersion test.

ments, they, too, should be compared with exposures in service environ-

ments.

Advantages of an Immersion Test Method

Continuous immersion tests offer several readily apparent advantages
over other acceptable spray or electrochemical type corrosion tests. Re-
sults may be determined rapidly (one to four days) and visually, without
the aid of a magnifying glass or microscope. Test apparatus is simple
and economical, and the test can be performed by a laboratory technician.
Because of the simple technique a continuous immersion test has a high
inherent reproducibility. This has been demonstrated by round robin test
programs with two of the methods just described. Tests can be performed
with equal ease on test coupons or upon small fabricated parts and
structures. Although no special preparation of the surface of the test
specimen is required, a light caustic etch is preferred. Finally, and most
important, the preceding advantages can be realized with continuous
immersion test methods because it has been possible to demonstrate very

encouraging comparisons with exfoliation experience in service or with

tests in service environments. Additional comparisons of these test methods
with service experience should be obtained.

APPENDIX
A. ASSET Continuous Immersion Exfoliation Test Procedure^
Application
1. 5086 and 5456 aluminum alloys.
Procedure
1. Specimen size: Optional, a nominal size of 1.5 by 4 in. is recommended.
2. Degrease specimen and etch 1 min in 5 percent by weight sodium hy-
droxide at 82 C (180 F), rinse in water, desmut 30 s in concentrated
nitric acid at 27 C (80 F), rinse with distilled water, and air dry.
3. Solution:
1.0 M ammonium chloride.
0.25 M ammonium nitrate.
0.01 M ammonium tartrate.
10 ml/1 hydrogen peroxide (30 percent H2O2).
Use reagent grade chemicals and distilled water.
pH as made is about 5.3.
Operate test at 66 ± 1 C (150 ± 2 F).
4. Recommended minimum solution volume to metal area is 40 ml/in.^.
5. Suspend specimens vertically and completely immerse in solution. Use
suitable inert container (that is, glass, etc.).
6. Normal test duration is 24 h.
7. Upon test completion, soak specimens in concentrated nitric acid until
clean and rinse in water.
B. EXCO Continuous Immersion Exfoliation Test Procedure^
Application
1. Al-Zn-Mg-Cu (1.2 to 2.8 percent copper).
Procedure
1. Specimen size: Optional, a nominal size of 2 X 4 in. is recommended.
2. Degrease specimen with suitable solvent; etching or other pretreatment
is not necessary but will not interfere.
3. Solution:
4.0 M sodium chloride.
0.5 M potassium nitrate.
0.1 Af nitric acid.
Use reagent grade chemicals and distilled water.
pH as made is 0.4.
Operate test at room temperature, that is, 25 ± 3 C (77 ± 5 F)
4. Recommended minimum solution volume to metal area is 50 ml/in.*.
'Procedure recommended by joint Aluminum Association-American Society for
Testing and Materials Task Group for Exfoliation Testing 7000 Series Copper
Containing Aluminum Alloys.
• Procedure recommended by Aluminum Association Task Group on Exfoliation
Corrosion and Stress Corrosion of 5000 Series Aluminimi Alloys.

5. Immerse specimens in the solution. The horizontal orientation of the test

surface is preferred to prevent loss of corrosion products.
6. Normal test duration is 48 h. Inspection of specimens (without clean-
ing) at intermediate periods should be performed to detect early de-
velopment of exfoliation that might be obscured with continued exposure.
7. Upon test completion, soak specimens in concentrated nitric acid until
clean and rinse in water.

References
[/] Lifka, B. W., Sprowls, D. O., and Kaufman, J. G., Corrosion, SBIIA, Vol.
23, No. 11, 1967, p. 335.
[2] Rotsell, W. C. and Long, J. R., Metal Progress, MEPOA, Vol. 91, No. 5,
1967, p. 106.
[3] Vaccari, J. A., assoc. ed.. Materials Engineering, MAENB, Vol. 71, No. 6,
June 1970, p. 22.
[4] Luhan, J. V. and Summerson, T. J., Metals Engineering Quarterly, MENQA,
Vol. 10, No. 4. 1970, p. 35.
[5] Brooks, C. L., Naval Engineers Journal, NVEJA, Vol. 82, No. 4, 1970, p. 29.
[6] Interim Federal Specification QQ-A-00250/19, 20 (NAVY-Ships), 31 Dec.
1968.
[7] Federal Specifications QQ-A-00200/14, 15 (ASG), 12 Aug. 1969 and QQ-
A-00250/24, 25 (ASG), 8 Oct. 1969.
[8] Liddiard, E. A. G., Whittaker, J. A., and Farmery, H. K., Journal, of the
Institute of Metals, JIMEA, Vol. 89, 1960-1961, p. 377.
[9] Booth, F. F. and Godard, H. P., Proceedings, First International Congress on
Metallic Corrosion, 1961, p. 703, Butterworth, London.
[10] Budd, M. K. and Booth, F. F., Corrosion, SBIIA. Vol. 18. No. 5. 1962, p.
197t.
[11] Lifka, B. W. and Sprowls, D. O., Corrosion, SBIIA, Vol. 22, No. 1, 1966, p. 7.
[12] Romans, H. B., Materials Research and Standards, MTRSA, Vol. 9, No. 11,
1969, p. 31.
[13] Ketcham, S. J. and Shaffer, I. S., Proceedings, Fourth International Corro-
sion Congress, 1969, to be published by National Association of Corrosion
Engineers.
[14] Sigalovskoya, T. M. and Zaretskii, E. M., Zashchita Metallov, ZAMEA, Vol.
3, No. 6, 1967, p. 730.
[15] Zaretskii, E. M. and Kireeva, A. F., Zavodskaya Laboratoriya, ZVDLA, Vol.
29, No. 9, 1963, p. 1098.
[16] Anderson, W. A., U. S. Patent 3,232,796, 1 Feb. 1966.

Intergranular Corrosion in Iron and

Nickel Base Alloys

REFERENCE: Henthorne, Michael, "Intergranular Corrosion in Iron and

Nidcel Base Alloys," Localized Corrosion—Cause of Metal Failure, ASTM
STP 516, American Society for Testing and Materials, 1972, pp. 66-119.
ABSTRACT: Intergranular corrosion in stainless steels (austenitic, ferritic,
and two phase) and nickel base corrosion resistant alloys (nickel-
chromium-iron, nickel-chromium-molybdenum, and nickel-molybdenum)
is reviewed. Practical aspects discussed include thermal treatments produc-
ing susceptibility, media that cause attack, influence of alloy composition,
and prevention methods. The currently most plausible mechanisms are also
outlined. Test methods, particularly the standard methods for quality con-
trol, are dealt with as well as techniques used to simulate the thermal
effects of welding. The relationships between intergranular attack and other
forms of localized corrosion are noted.
KEY WORDS; corrosion, intergranular corrosion, stainless steels, nickel
alloys, tests, evaluation, grain boundaries, precipitation (chemistry), stress-
corrosion cracking, pitting

Grain boundaries are physically distinct regions separating different

or differently oriented grains [1].' The boundaries are favored sites for
precipitation and segregation which makes them chemically as well as
physically different. These differences may result in intergranular attack
as shown in Fig. 1.
Most metal alloys are susceptible to intergranular corrosion when
exposed to specific corrodents. This paper is concerned only with
aqueous media and certain iron and nickel base alloys used primarily
for their corrosion resistance. In these alloys the attack is usually de-
pendent upon prior thermal history and is a result of chemical rather
than just physical differences at the boundaries. The aims are to review:
1. practical information—metallurgical treatments that produce sus-
ceptibiUty, the types of media that corrode susceptible material, and
how one can avoid the problem.

^ Supervisor, Corrosion Research, Carpenter Technology Corp., Reading, Pa.

19603.
* The italic numbers in brackets refer to the list of references appended to this
paper.
66

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
HENTHORNE O N INTERGRANUUR CORROSION tl

(top) Plan view of surface showing some grains dislodged (X50).

(bottom) Polished section perpendicular to surface (xlOO).
FIG. 1—Inlergranular corrosion.

2. mechanisms—the currently most plausible explanations for inter-

granular attack.
3. testing—available test procedures, their limitations, advantages, etc.
The alloys discussed are all used commercially and are those for

which intergranular corrosion susceptibility can be a limitation to their

optimum use. The subjects covered in turn are: (a) austenitic stainless
steels, (b) ferritic stainless steels, (c) austenitic-ferritic stainless steels,
(d) high-nickel alloys, (e) testing, and (f) relationship to other forms
of corrosion.

Austenitic Stainless Steels

This family of alloys has received more attention from researchers
of intergranular corrosion than any other material. Most of the work
has been with the basic 18Cr-9Ni analysis—AISI Type 304.

Conditions Leading to Attack

The great majority of intergranular corrosion problems in austenitic
stainless steels are related to metallurgical changes occurrmg in the
range of about 800 to 1500 F (427 to 816 C). The temperatures most
likely to cause problems are usually close to the middle of this range
but are also dependent upon the time of exposure. For example, short
times are more likely to produce susceptibility at temperatures in the
upper part of the range, whereas long exposures (many thousands of
hours) may be required at the lower end. This is shown in Fig. 2. For
a typical Type 304 analysis an exposure of several minutes at a temp-
erature such as 1250'F (677 C) is usually adequate to cause notice-
able attack in a severe intergranular corrosion test. We shall see later
that susceptibility is attributed primarily to the precipitation of chromium
rich carbides at the grain boundaries although other grain boundary
phases and segregates play a role in some cases.
Austenitic stainless steels can be attacked intergranularly in a few

BOO 900 1000 UOO IZOO 1300 1400 ISOO

TEMPERATURE "F.
FIG- 2—Effect of sensitizing time and temperature on corrosion of Type 347
stainless steel in boiling 65 percent HNO, [2].

00000

TIME, h
FIG. 3—Time-temperature-sensitization curves for Type 347 stainless steel in
boiling 65 percent HNOi \2\.

Specific media even when they are in the solution annealed condition.
This is a rarity and will be discussed later.
The term sensitization is commonly used to describe heat treatments
that may make the alloy susceptible to intergranular attack. The word
is used in this context in this paper and does not necessarily imply
susceptibility to attack. Sensitization data usually are plotted as a func-
tion of the sensitizing temperature and time as shown in Fig. 3. These
curves are called TTS, C, or Rollason curves.
It is customary to plot corrosion rate contours like those in Fig. 3,
but sometimes the onset of metallurgical changes known to relate to
the problem are determined (for example, by metallography, X-ray dif-
fraction) and plotted as shown in Fig. 4. There is a relationship be-
tween intergranular corrosion and M^^Q precipitation in the grain
boundaries, although of course the position of the intergranular attack
curve will depend upon the severity of the corrosion test used to evaluate
susceptibility. At the higher temperatures in Fig. 4 there is grain bound-
ary precipitation but no intergranular corrosion. The reason for this
will become apparent when the mechanisms of attack are discussed.
In practice, susceptibility to intergranular attack will result usually
from one of three situations:
1. Slow cooling from an annealing treatment (or from casting or
forging). This is obviously more of a factor in large product sizes
where the rate of cooling through the sensitization range is slower.
2. Stress relieving for several hours in the sensitization range, for
example, 1100 F (593 C).
3. During welding.
Welding is probably the most common cause of intergranular corro-

1100

PRECIPITATION
1000 -
AT BOUNDARIES

900

' 800 INTERGRANULAR

CORROSION

700
PRECIPITATION
< IN GRAINS
u soo
IL
2
bJ
I-
500

400
.0015 . 1500

TIME, h

FIG. 4—Influence of lime and temperature on MuCe precipitation and inter-

granular corrosion in Type 304 stainless steel [3].

sion. Parent metal on both sides of a weld is heated in the sensitization

range for several seconds or minutes during welding, and this can re-
sult in susceptibility as shown in Fig. 5. This type of attack is sometimes
called "weld decay" or "low-temperature heat affected zone corrosion."
It will be shown later that the high-temperature heat affected zones of
a weldment can be attacked in some of the modified austenitic stain-
less steels.
It is very important to realize that material heated in the sensitiza-

FIG. 5—Preferential attack in low-temperature heat affected zones (A) of Type

302 stainless steel welded strip ( x 2 ) .

tion range may not show intergranular corrosion in service. This is

true even if the alloy is attacked in a laboratory intergranular corrosion
test. Whether or not the problem will occur depends upon the specific
service environment. A lack of information as to which media cause
attack and which do not was highlighted by Warren [4] who compiled
a list of media reported to have caused problems. Since then the results
of comprehensive test programs to study this aspect have been reported
by Auld [J] and the High Alloys Committee of the Welding Research
Council [6]. These programs covered a variety of chemical process and
naturally occurring corrodents, and listings of those media which did
and did not cause attack in several different welded and sensitized
stainless steels were obtained.
It is difficult to summarize the findings of Warren's survey and the
field test data because the ability of a corrodent to cause problems is
dependent upon the specific alloy, its degree of sensitization, and corro-
dent variables—concentration, temperature, impurity contents, etc. Most
service experiences of intergranular corrosion have been in oxidizing, in-
organic or organic acids, but there are exceptions (for example, attack
in seawater) and by no means all acid solutions cause problems.
Mechanism of Attack in Sensitized Austenitic Stainless
Intergranular corrosion susceptibility in these steels increases as their
carbon content increases [7]. Whereas the carbon (typically 0.07 per-
cent by weight) is soluble during annealing (1950 F (1066 C)) it has
very restricted solubility at lower temperatures, for example, less than
0.01 percent by weight at ambient temperature. Rapid cooling from
the annealing temperature results in supersaturation of carbon in solu-
tion. Subsequent holding in the sensitization range enables this carbon
to precipitate out as chromium rich carbides. For example, (CT,Fe).,^Ca
type carbides have been detected [8,9] with chromium contents as high
as 70 percent by weight [10]. There is little doubt that intergranular
corrosion is related to these carbides.
Relatively early in the history of these alloys a plausible theory for
their intergranular corrosion was suggested by Strauss et al [11] and
Bain et al [7], They proposed that chromium depletion occurs adjacent
to the high-chromium grain boundary carbides. These low-chromium areas
have lower corrosion resistance in certain media and so are attacked
preferentially. A modification was proposed by Hatwell [12] who sug-
gested that chromium depletion in the boundary itself (between carbide
dendrites) was the cause of attack, at least in the early stages of sen-
sitization.
Other roles have been assigned to the grain boundary carbides. For
example there is the strain theory of Kinzel [9] which attributes attack
to corrosion of the distorted lattice adjacent to the precipitate. This

theory is difficult to prove or disprove, but the fact that corrosion rates
of the bulk material do not increase significantly with applied strain
is inconsistent with it unless other arguments (for example, electro-
chemical effects) are invoked.
Also there is the electrochemical theory promoted by Stickler and
Vinckier [13,14] and suggested by Kinzel [9] and Stefanides [15]. Inter-
granular attack is attributed to galvanic corrosion between the more
noble carbides and the adjacent matrix. There is some evidence for
the more noble character of chromium rich carbides [9,14], but this
is disputed by Osozawa et al [16]. However, if one combines the chro-
mium depletion and galvanic corrosion concepts there is feasibly quite
a large potential difference between the carbide and the "anodic" de-
pleted zone. In fact, the grain itself could serve as a cathode to a de-
pleted zone, independent of the potential of the carbide.
Baumel et al [17] argue that a galvanic mechanism would result in
much more widespread corrosion than just the area immediately adja-
cent to the carbide. Again this objection would be invalid if one con-
sidered a mechanism dependent upon depletion and galvanic corro-
sion with the depleted layer being the anode.
Stickler and Vinckier [14] calculate from their experimental data that
the width of metal attacked is much greater than could be attributed to
corrosion of chromium depleted material. They use this as evidence
against the chromium depletion theory but do not acknowledge the
possibility that once corrosion starts in depleted material it could spread
to "good" material, that is, the alloy content of the steel might be sufficient
to maintain the passivity of an air formed protective but not enough
to passivate a more actively corroding surface.
The three main theories are demonstrated in Fig. 6a,b,c, and the
the combined "depletion-electrochemical" concept is shown in Fig. 6d.
Baumel et al [17] and Wilson [18] have reviewed the depletion, strain,
and electrochemical intergranular corrosion models and concluded that
the chromium depletion theory is the most plausible. One obvious ad-
vantage of this concept is that it is simple and can be understood by
users of stainless steel who must learn how to avoid the problem. It
can be related readily to the remedial measures discussed later, and
providing a simple understanding of the problem probably helps in
their successful application. The main disadvantage of the chromium
depletion theory has been the lack of direct evidence for the depleted
zone. Some have detected a depleted layer [19,20], but the data are not
conclusive.
In addition to the discussed arguments in favor of chromium deple-
tion it is interesting to note the work of Ceilings and deJongh [27]. They
argue that if chromium depletion is a viable theory there should be
preferential elevated temperature oxidation in susceptible material. Their

GRAIN T~
CARBIDE CARBIDE
(d) INTERFACE INTERFACE

r • IF NOBLE
CARBIDE
I— IF CARBIDE

4
SIMILAR POTEN-
TIAL TO GRAIN

—T~ CARBIDE
CARBIDE
INTERFACE INTERFACE

(a) Chromium depletion.

(/>) Strain.
(c) Electrochemical—more noble carbide.
(d) Electrochemical—more active chromium depleted region.
FIG. 6—Schematic representation of intergranular corrosion mechanisms for
sensitized stainless steels.

oxidation tests showed this to be the case. It is also worth noting that
in recent years there have been attempts by Stawstrom and Hillert [22]
and Tedmon et al [23] to calculate the chromium content in and adja-
cent to precipitated grain boundaries, and their findings in general
agree with experimental observations and the chromium depletion model.
They also indicate a very narrow depleted zone which is consistent with
the practical problems experienced in detecting it.
While discussing the mechanism of attack, it is also important to
note the influence of: (1) precipitate morphology and (2) solution
chemistry. Lack of attention to these variables is probably the main
reason for apparently conflicting evidence and theories.
It is not always necessary to have a continuous grain boundary pre-
cipitate to get intergranular corrosion because chromium diffusion (and
depletion) can occur within the boundary itself. However, having a con-
tinuous grain boundary precipitate will generally increase the chances
of rapid attack. Precipitate morphology is very much a function of
prior thermal and mechanical history as well as sensitizing tempera-
ture. Its importance has been recognized by several workers [8,9,12,13,
24,25] and should be considered in any intergranular corrosion evalua-
tion. This subject will be dealt with again when nonsensitized stainless
steels are discussed.
The chemistry of the corrodent is important because the same mech-
anism of attack may not apply to all situations. For example, it will be
shown later that very strongly oxidizing solutions can attack aimealed
material which does not contain grain boundary carbides. The utiliza-
tion of data from one specific solution to refute or support a theory
therefore, can be misleading.

A consideration of the information available at this time strongly

suggests that in the regular 18Cr-9Ni austenitic stainless steels, chro-
mium depletion is a prerequisite for intergranular attack in most media.
The electrochemical theory may work together with chromium depletion
as noted previously. It will be shown later that for very strongly
oxidizing solutions, molybdenum containing alloys, and some of the
high-nickel alloys, chromium depletion adjacent to carbides is not the
only cause of intergranular attack.
Influence of Alloy Composition
There are many different types of austenitic stainless steel. Major
and minor alloying elements are added to or removed from the typical
18Cr-9Ni analysis to provide specific benefits in corrosion resistance,
mechanical properties, or ease of fabrication. These changes can signi-
ficantly influence intergranular corrosion susceptibility [26-57].
When considering composition effects it is important to separate
those due to the element itself from those occurring as a result of struc-
ture changes. The following discussion assumes a completely austen-
itic structure. It will be shown later that the presence of a second phase
can have a marked influence, and therefore some of the elements dealt
with could have additional effects by promoting or removing ferrite.
It should be noted also that the effect of composition can vary with
the environment and the mode of sensitization (for example, welding
compared to stress relieving).
Elements which can be detrimental to intergranular corrosion resis-
tance are carbon, molybdenum, silicon, and nickel. The role of carbon
is relatively straightforward, since one of the major causes (in some
alloys the only cause) of susceptibility is the precipitation of chromium
carbides [7,30,31]. Molybdenum, at the 2 to 3 percent level present in
AISI Type 316 and its modifications, slightly decreases carbide precipi-
tation but can increase susceptibility in certain media primarily because
it promotes the formation of deleterious intermetallic compounds such
as <T-phase [4,31-34]. These compounds can apparently cause sus-
ceptibility to attack even when present in amounts too small to detect
metallographically [25].
Silicon (1 to 2 percent) has been shown to be detrimental [26,35,36]
and particularly so in molybdenum containing steels because it ap-
pears to accelerate the precipitation of deleterious intermetallic com-
pounds [29]. The role of this element in intergranular corrosion will be
discussed again in connection with the attack of annealed material. In-
creases in nickel content can be deleterious to intergranular corrosion
because they reduce the solubility of carbon in austenite and so pro-
mote chromium carbide formation. Figure 7 demonstrates the deleteri-
ous effects of carbon and nickel (and the beneficial effect of chromium).

CARBON %

^0"^^^ ^^

14 16 18
NICKEL %
FIG. 7—Effect of stainless steel chromium and nickel content on the tolerable
carbon content to avoid intergranular attack in the CuSOrHtSOs test after 650 C,
1 h m.

It will be shown later that higher levels of nickel than shown in Fig. 7
can be increasingly deleterious.
Elements which form stable carbides (for example, columbium [30,37]
and titanium [37-40]) are generally beneficial because they tie up carbon
which could otherwise form chromium carbide. Note that titanium can be
deleterious in certain strongly oxidizing media such as nitric acid (HNOJ.
The reason for this is dissolution of titanium carbide (TiC) [41,42]. Chro-
mium can be beneficial as shown in Fig. 7, but in the amounts normally
present (up to about 25 percent) the effect is not generally sufficient to
prevent attack.
Boron is an element which in amounts from less than 0.001 up to
0.01 percent has been reported to be both deleterious and beneficial
depending upon the boron content within this range, the base composi-
tion, the heat treatment, and unfortunately the investigator [43-47]. The
boron atom is intermediate in size between the interstitial element carbon
and the substitutional elements iron, chromium, and nickel. Therefore,
it would be expected to play a role in grain boundary areas where there
is considerable misfit and has been reported to influence the rate of
formation of grain boundary carbides and intermetallics [43,48]. Al-
though the role of boron is believed mostly due to its effect on chro-
mium carbide, cr-phase, etc., it should be noted also that chromium
borides have been detected [48,49]. The effect of this element is in need
of further investigation with careful consideration of base alloy content,
boron level, thermal and mechanical processing, precipitate morphology,
etc.
Nitrogen additions (typically 0.15 percent or more) are made to stain-
less steels to increase their strength or permit savings in nickel (nitro-
gen is also an austenite former). Manganese is also added as a sub-

stitute for nickel and to increase the solid solution solubility of the
nitrogen. High-nitrogen Cr-Ni, Cr-Mn-Ni (AISI 200 Series), and Cr-
Mn stainless steels are commercially available so it is pertinent to con-
sider their intergranular corrosion resistance. They can be described best
as approaching or equalling the resistance of the regular nitrogen (0.03
percent) AISI 300 Series grades [30,50-54], although there have been
claims that they are Slightly inferior [55-57]. It should be noted that in
one of the latter cases [56] the conclusions are questionable because of
carbon differences.
Probably the most significant difference between low- and high-
nitrogen stainless steels is that the latter cannot be effectively stabilized
with elements such as titanium or columbium. This is because these
elements when combined with nitrogen reduce both its effectiveness in
solid solution and their ability to tie up carbon.

Avoiding Intergranular Corrosion in Austenitic Stainless Steels

Intergranular corrosion is avoided generally by using annealed mate-
rial which has been cooled rapidly through the sensitization range to
avoid a continuous grain boundary network of susceptible material.
Other commonly used methods of avoiding the problem are to use
low-carbon grades such as Type 304L (<0.03 C) or stabilized grades
such as Type 321 (titanium added) or Type 347 (columbium added).
Neither of these techniques make an alloy immune to intergranular cor-
rosion, but generally they do make it sufficiently resistant to be used
in the welded or stress relieved conditions. The theory behind these ap-
proaches is quite simple. The low-carbon grade removes the source
of the problem, that is, it limits the amount of chromium carbide that
can be formed. The stabilized grades take the different approach of
putting in a strong carbide former to tie up carbon before the chro-
mium gets a chance to do so.
In welding applications, the stabilized grade can be at a disadvantage
(compared to the low-carbon alloys), because the titanium or columbium
carbide goes into solid solution at the high temperatures associated
with welding. This problem is evident immediately adjacent to the ac-
tual weld metal and is sometimes described as knifeline attack. It
normally shows up in stabilized material which has been multipass
welded or welded and then stress relieved in the sensitization range.
Knifeline corrosion should not be confused with fissure attack which
can occur in titanium stabilized alloys (for example, Type 321) exposed
to strongly oxidizing media such as boiling HNOg [39]. This problem
is related very probably to the fact that near the fusion zone in a weld
a continuous grain boundary network of TiC forms, and it is soluble in
HNO3 [41,42].
The three techniques used to avoid intergranular corrosion in almost

all cases are either annealed material, low-carbon alloys, or stabilized

grades. There are, however, other factors which influence intergranular
corrosion, and these have been reported as prevention methods. They
include stabilize annealing, cold working, and grain size control.
Stabilize annealing (for example 1600 F (871 C)) has been sug-
gested as a means of improving the stabilized grades and those contain-
ing molybdenum [25]. In stabilized material the intent is to optimize the
formation of the stabilizing carbide. However, it is possible that the
1600 F (871 C) temperature is too low because there are reports of
resistance deteriorating when this anneal is used [5,6]. In nonstabilized
grades the function of the treatment is probably to form "harmless"
chromium carbides. At this temperature they are more likely to be dis-
crete, and chromium has had an opportunity to diffuse back into de-
pleted regions. Again, however, there is evidence that such treatments
can be harmful [6].
Stabilization treatments very probably can be beneficial by forming
random precipitates without chromium depletion. Unfortunately, how-
ever, the morphology of grain boundary precipitates is dependent upon
prior thermal and mechanical processing. This and the fact that deleteri-
ous intermetallic compounds can form in some of the alloys during the
anneal is probably why there is conflicting evidence regarding this "pre-
vention method." In any event, stabilize annealing can be hardly con-
sidered as a reliable technique for improving resistance unless prior
evaluation of similarly processed lots of the material involved has shown
it to be so.
The beneficial effects of cold work prior to sensitizing were noted
relatively early in the history of the stainless steels by Bain and co-
workers [7]. They attributed the benefits to the introduction of carbide
nucleation sites within the cold-worked matrix so minimizing continuous
grain boundary networks.
More recently Tedmon et al [58] confirmed this as shown in Fig. 8.
Their specimens were wires, and after the corrosion exposures they
used tension tests to evaluate the degree of intergranular attack. Using
thin-film electron microscopy, they observed increased matrix precipitate
in the cold-worked alloy and also saw some recrystallization in material
that was cold reduced 55 percent and held at 1112 F (600 C) for
24 h. They noted that this recrystallization would have the beneficial
effect of removing any continuous grain boundary network that did
form.
Although there is apparent agreement in the literature as to the
effect of cold work, it seems likely that in practice it is not quite so
straightforward. For example, cold work in 18Cr-8Ni stainless steels
can lead to transformation to martensite which then may be preferen-
tially attacked. Also it seems likely that small amounts of cold work (the

1 - - I • 1

120

(0
M
U
a 80
»-
(0
ASOLUTIONIZED ONLY
HI OSOLUTIONIZED, SENSITIZED
K
3 OSOLUTIONIZEO, COLD
I- 40 WORKED, SENSITIZED

i 1
0 1 1
0.01 100

TIME IN STRAUSS SOLUTION, h

FIG. 8—Influence of cold work on intergranular attack of 18Cr-8Ni-0.1C steel.
A reduction in the fracture stress signifies intergranular attack [58].

previous discussion has concerned 50 percent or more) could be del-

eterious in inducing grain boundary precipitation, especially in grades
susceptible to (r-phase formation. Yet another area in which cold work
is potentially deleterious is in increasing end-grain attack (particularly
in boiling HNO3).
The influence of grain size has been studied in some detail [25,59-62],
but there is disagreement as to its effect. It is difficult to study because
thermal and mechanical-thermal treatments used to vary grain size can
cause grain boundary chemical and structural changes which also in-
fluence corrosion. For example, high-temperature anneals used to pro-
duce large grain size will increase the susceptibility of stabilized grades
by virtue of dissolving the stabilizing carbide. On the other hand these
high-temperature treatments can be beneficial by completely removing
deleterious compounds such as o-phase and residual grain boundary cold-
work effects which might accelerate precipitation.
Following the early work of Newell [59], which showed increasing sus-
ceptibility with increasing grain size, the most active investigator has
been Cihal [60,61]. Using calculated and experimental values he gener-
ally agrees with Newell but does acknowledge more of the exceptions
discussed next. Cihal explains the grain size effect on the basis that in
small grain size material there is more grain surface area and therefore
less chance of a continuous network of precipitate. Whether this argu-
ment is significant for the range of grain size in commercial materials
is questionable.
In an effort to minimize thermal effects in producing varying grain
size, Levin and Kaloshina [62] gave their small and large grain material

the same final anneal. For both Types 304 and 321 they found no effect
of grain size in the copper-acidified copper sulfate test.
Likewise, Streicher [25] did not duplicate the work of Newell and Cihal.
Using both weight loss and electrical resistance techniques he concluded
that the effect of grain size depends upon the method of measurement
and the testing solution. He also notes that small grains are removed more
readily and that small grains are surrounded most frequently by ditching
in the oxalic acid etch test (see later for discussion of this test). It
should be noted also that once a small grain has been removed, attack
in HNO3 is accelerated due to the buildup of hexavalent chromium in
the crevice.
In conclusion there is conflicting evidence on the effect of grain size.
The weight of available experimental evidence indicates a slightly greater
susceptibility for large grain material, but to a large extent this is most
likely a result of factors leading to it rather than an inherent effect of
grain size itself.

Intergranular Corrosion in Nonsensitized Austenitic Stainless Steels

Annealed (that is, nonsensitized) austenitic stainless steels can be
susceptible to intergranular attack in strongly oxidizing media [25,63-80].
The solution most commonly used to study this phenomenon is hot nitric
acid plus sodium dichromate, for example, boiling 5 N HNO3 + 4 g/liter
Cr+''. The attack is attributed to the presence of hexavalent chromium
and increases with the amount of this ion as shown in Fig. 9. The data
shown are reported to represent intergranular attack. Other ions (Fe+^,
Ce+^,Mn+^) also have been reported to accelerate attack [25,64],
The problem received renewed interest in the 1960s as a result of
intergranular failures (probably stress assisted) in boiling water nu-
clear reactor fuel cladding. Investigations in France and the United
States dupHcated each other to some extent, and, in addition, research-
ers on both sides of the Atlantic have each published similar data in
a variety of different journals.
The accelerating role of oxidizing cations such as Cr+* is attributed
to their depolarizing the cathodic reaction which in turn raises the cor-
rosion potential and anodic reaction rate [25,64]. There are also indi-
cations that some anodic depolarization occurs [25]. Although a rise in
the anodic reaction rate is consistent with the observed higher corro-
sion rates, it does not explain why grain boundaries are attacked prefer-
entially. To explain the latter there are two main theories—(1) that
grain boundaries are attacked because they are high-energy regions [25,64,
70,71] or (2) that elements segregated at grain boundaries result in a
chemical difference between the boundary and the matrix, and this serves
as the driving force for corrosion [74-80]. There have been attempts to

FIG. 9—Increase in intergranular corrosion rate of annealed Type 304 stainless

steel in 5 N HNO, {80 h at 108 C) as a result of hexavalent chromium addi-
tions [68].

combine the two theories [72] on the basis that the high-energy regions
are those most likely to show segregation.
Chaudron [65] has shown that high-purity steels are not susceptible
to attack. This supports the segregation theories as opposed to those
dependent solely on grain boundary energy. Armijo [79] confirmed Chau-
dron's findings and studied the influence of specific elements in more
detail. Using high-purity metals, the effects of carbon, nitrogen, man-
ganese, phosphorus, and sulfur were evaluated. The only elements
found to have a significant effect on annealed intergranular corrosion
resistance were phosphorus and silicon. Intergranular corrosion rates
in a nitric acid-sodium dichromate mixture increased with increasing
phosphorus content (with significantly increased attack at the 0.09 per-
cent level), whereas the role of silicon is more complicated as shown
in Fig. 10. Similar effects for silicon in commercial purity alloys were
also reported by Coriou et al [66,67,69].
Efforts to explain the effect of silicon have been numerous, but there
is disagreement as to its precise role. There is considerable evidence
[80,82.83] showing silicon to be present in surface oxide films on the
higher silicon alloys. Quite logically this led Armijo and Wilde [80] to
propose that high-silicon alloys are resistant due to polarization of the
cathodic reaction by the silicon rich film. They support this idea with

-r-i-r-ri

0 19 Cr- 9NI
M A 16 C r - l 4 N i •

E • 14 C r - l 4 N i •
u a20Cr-20Ni •

E •
0 \ •

I • /
hi _ _
<
• ^
*~''"A o' W
m
O
tc • AVO-
IE
o . . A
0.1 — 1 . 1 . 1 111 a 1
0.01 0.1

SILICON percent by weight

FIG. 10—Variation of corrosion rate with alloy silicon content, {boiling nitric
acid-dichromate solution). Peak attributed to intergranular attack [80].

electrochemical measurements showing increased polarization as silicon

increased from 1 to 4 percent. Armijo and Wilde attribute the higher in-
tergranular corrosion rates at intermediate silicon levels to silicon segre-
gation at the grain boundaries and increased attack of these due to gal-
vanic corrosion.
Scharfstein [2S] invokes film effects to explain the role of silicon at
all levels. He proposes that intermediate levels of silicon (1 to 2 percent)
are sufficient to maintain passive films on the grains but not the grain
boundaries. This accelerates the attack of the latter. At higher levels
(4 percent) there is sufficient silicon to promote a protective film over
both, whereas at very low levels (<0.1 percent) there is no film effect
of silicon on either grain or grain boundary.
Desestret et al [72] relate the high intergranular corrosion rates of
1 percent silicon heats to segregation and grain boundary energy. These
amounts of silicon create grain boundaries with higher energy than
lower or higher levels as shown in Fig. 11. They claim that this results
in maximum segregation. The reduced attack in the higher silicon alloys
is attributed to a lessening of the chemical concentration gradient be-
tween boundary and matrix. Streicher [73] had shown previously that
high-silicon alloys were much more resistant to chemical etching and
thermal etching. Desestret et al [72] disagree with Armijo and Wilde
[80]. The former claim that silicon influences anodic but not cathodic
polarization, while the latter show the opposite. Until these differences
are resolved the exact role of silicon is difficult to assess.
Vermilyea and Tedmon [81] suggest that the role of grain boundary
segregation may be to increase the electronic conductivity of the oxide

>-
(9

I
3
s
SILICON %
FIG. 11—Influence of silicon content on grain boundary energy of 16Cr-14Ni
alloys [71].

at the boundary. This increases attack because it accelerates the oxida-

tion of Cr+' in the oxidefilmto Cr+'' which is soluble.
The proponents of the solute segregation theory for attack in an-
nealed material—Aust, Armijo, Westbrook, and co-workers [74-80]—
support their arguments with heat-treatment studies which are summar-
rized in Fig. 12. For Type 304, quenching from about 1050 C (1922 F)
produces no grain boundary carbides but solute segregation as de-
picted in Fig. 12a. If the steel is held in the sensitization range, say
600 C (1112 F) then continuous grain boundary precipitation can occur
as shown in Fig. 12b. Material treated at slighdy higher temperatures

(a) solution treated at 1050 and 1300 C, solute segregation.

(b) sensitized at 500 to 700 C, continuous carbides or second phase.
(c) Stabilized at 800 to 900 C, discontinuous carbides incorporating
segregant.
(d) High-purity alloy, "clean" grain boundaries.
FIG. 12—Schematic illustration of local changes in microsiructure and compo-
sition near grain boundaries [74],

such as 800 to 900 C (1472 to 1652 F) contains isolated grain bound-

ary carbides which may also incorporate solute impurities from adja-
cent grain boundary regions as in Fig. 12c. High-purity alloys contain
neither precipitate nor segregate (Fig. 12c/).
In summary, annealed austenitic steels can be susceptible to attack
in specific highly oxidizing media. Grain boundary segregation of ele-
ments such as silicon is very probably the cause, but the exact mechanism
by which the segregate increases corrosion is not yet clear. Figures 12
c and d suggest two methods of avoiding intergranular corrosion in an-
nealed material, that is, use high-purity alloys or stabilize annealing
treatments. It should be noted that the latter reduces rather than elim-
inates attack and will depend upon alloy composition and prior process-
ing history. High-silicon alloys are an alternative solution to this specific
problem although they may be prone to fabrication difficulties. Again,
it should be emphasized that annealed stainless steels are resistant to
intergranular attack in almost all situations, and the discussion pre-
viously relates to just a few specific, highly oxidizing media.

Ferritic Stainless Steels

Ferritic stainless steels, like the austenitic grades, can be susceptible
to intergranular corrosion which is largely dependent upon prior thermal
history.
The ferritic alloys (Fe-Cr and Fe-Cr-Mo) are commercially attrac-
tive because of their low raw materials cost and their resistance to
stress-corrosion cracking. On the negative side they have generally poorer
formability than austenitics, can be susceptible to "885 F embrittle-
ment," and are more prone to brittle failure and intergranular corro-
sion (particularly in the welded condition). There are several com-
monly used grades including types 405 (Fe-12Cr), 430 (Fe-17Cr), 434
(Fe-17Cr-lMo), 446 (Fe-26Cr), and 406 (Fe-13Cr-3Al). Stabilized
and free machining varieties exist in some cases, and it should be noted
that new Fe-Cr-Mo ferritic alloys are being currently developed and
introduced commercially.

Conditions Leading to Attack

From a practical point of view the intergranular corrosion behavior
of these alloys is vastly different from the austenitic grades. Thermal
treatments which produce susceptibility in the ferritic alloys are those
used to minimize it in the austenitics and vice versa as shown in Table 1.
To be more specific, a sensitization treatment for a commercial purity
ferritic stainless steel will involve heating above about 1700 F (927 C).
Material heated above this temperature can show susceptibility to inter-
granular attack independent of whether it is water quenched or air
cooled from that temperature. Slow cooling (for example, cooling in the

TABLE 1—Contrasting effect of heat treatment on intergranular corrosion of

ferritic {Type JfSO) and austenitic {Type SOi) stainless steel.
Average Corrosion Rate, mpy," in Boiling 65 percent HNOj
Heat Treatment Prior to Test
Alloy 1300 F, 1 h, W Q ' 2000 F, 1 h, WQ'
Type 430 40^ 400
Type 304 400 lO*
" mpy = mils per year.
' WQ = water quenched.
' No significant intergranular attack.

furnace) from the sensitization temperature or annealing sensitized mate-

rial at about 1450 F (788 C) for several minutes or hours will usually
eliminate the problem. Note that for very low-carbon, low-nitrogen ferritic
stainless steels water quenching can be beneficial as discussed later.
In the welded condition a ferritic stainless steel becomes sensitized
in the weld metal itself and in the high-temperature heat affected zone,
that is, greater than about 1700 F (927 C). Figure 13 shows welded
Type 430 strip after an accelerated intergranular corrosion test and
should be compared with the Type 302 in Fig. 5. Of course, some
austenitic alloys (that is, the stabilized grades) can also show suscep-
tibility to attack in high-temperature heat affected regions as discussed
previously.
Still considering the practical aspects it should be noted that a sensitized
ferritic stainless steel is susceptible to intergranular attack in a wider variety
of media than a sensitized austenitic alloy. Environments generally con-
sidered quite mild (for example water) can attack sensitized Type 430
stainless [84-86]. Also, remedial measures used for austenitic alloys.

^•w^r*

FIG. 13—Intergranular attack in weld and high-temperature heat affected zones

of ferritic stainless steel ( X 2 ) .

such as lowering the carbon content to 0.03 percent maximum, do not

eliminate the problem and neither does the addition of stabilizing elements
although they do minimize it. These factors increase the practical signifi-
cance of intergranular corrosion susceptibility in this class of materials.
Ferritic stainless steels have received much less study and attention
than the austenitics because of their smaller usage. There have been
several studies in the past 20 years, however, and a marked increase in
interest has developed in the past 3 years. This is a result of several
factors. Periodic shortages and price increases in nickel, together with the
fact that stress corrosion cracking is related closely to nickel content,
have led to increased interest in alloys low in nickel. Also, continued
advances in melting techniques have led to the possibility of economically
making ferritic alloys sufficiently low in carbon and nitrogen to offset
some of their poor as-welded mechanical and corrosion properties.
Mechanisms of Intergranular Attack in Ferritic Stainless Steels
Several mechanisms have been proposed. The apparent difference be-
tween the phenomena in ferritic compared to austenitic steels has led
some investigators to believe the mechanisms of attack in these systems
must be different, whereas others have been able to reconcile these
differences into one common theory for all stainless steels.
By far the most work has been done on the Type 430 (Fe-17Cr)
analysis. In this alloy, austenite can form at temperatures above 1700 F
(927 C) in amounts dependent upon the precise analysis within the Type
430 range. When ferrite and austenite are present, chemical segregation
of certain elements occurs due to their different solubilities in these two
phases. For example, there is transfer of some carbon from the ferrite to
the austenite, whereas the ferrite is enriched with chromium.
The correspondence in Type 430 between the temperature when austenite
forms and that which must be exceeded to induce intergranular corrosion
has naturally led to theories which relate intergranular corrosion suscepti-
bility to austenite or the subsequent martensite that forms during cooling.
Hochmann [87] proposed that the lower chromium content of the austenite
compared to the adjacent ferrite cause it to be preferentially attacked.
Kiefer [88] claimed that the second phase is attacked because it is
martensite. It does seem reasonable that martensite (lower in chromium
and higher in carbon) than the adjacent ferrite should be preferentially
attacked in a test medium such as boiling 65 percent HNO3, and this
is indeed found to be the case [89]. It is very unlikely, however, that
this is a necessary condition for intergranular attack because corrosion
occurs in high-chromium alloys in which neither austenite nor martensite
are detected. Also, even in alloys that contain martensite there may not
be a continuous grain boundary network that would be necessary for
both of the previously mentioned theories.

This lack of a continuous austenite network even when austenite is

present can be actually used to explain an observation of Baumel [89]
and Baerlecken et al [90] that ferritic stainless steels containing austenite
(martensite) can be more resistant to intergranular attack than completely
ferritic material. The findings of Baumel and Baerlecken et al are con-
firmed by the author's experience, an example of which is shown in Fig.
14. The composition of the two Type 430 alloys was varied to yield
austenite in one weld but not in the other.

'•-<-.. » «

(top) Weld metal contains ferrite plus martensite.

(bottom) Weld metal essentially ferrite.
FIG. 14—Polished and etched sections through welded Type 430 stainless steel
after boiling 65 percent HNOi test (,XlO). Note less severe attack in two phase
weld metal.

Baumel logically points up that the austenite formed at high tempera-

tures acts as a sink for carbon. This decreases the supersaturation of car-
bon in ferrite so it is less prone to precipitate out at ferrite grain bound-
aries. Of course, during cooling, some precipitation occurs in the austenite
(later martensite) and particularly at the ferrite-austenite interfaces. This
precipitation does not result in intergranular corrosion provided the
austenite does not form a continuous grain boundary network.
The formation of grain boundary carbides and nitrides has been used
also to explain intergranular attack in ferritic stainless steels. Several
roles have been ascribed to these precipitates. Houdremont and Tofaute
[91] proposed that iron rich grain boundary carbides formed at about
1200 F (649 C) during cooling from temperatures above 1700 F (927 C).
These iron rich carbides had low-corrosion resistance in many enviroments
so leading to intergranular attack. Holding the steel at about 1400 F
(760 C) caused chromium to diffuse into the carbides, increasing their

corrosion resistance and reducing susceptibility to intergranular attack.

The lack of evidence for these iron carbides, while chromium rich carbides
and nitrides have been detected in susceptible material, is a serious
drawback to this theory.
The role of grain boundary carbides and nitrides is explained differently
by Lula et al [92] who attribute intergranular corrosion susceptibility
to strains produced in the matrix adjacent to these precipitates. This
strained material is postulated to be more anodic than the remaining
base metal and grain boundary precipitates. The beneficial effects of
annealing at about 1400 F (760 C) are explained on the basis of stress
relief adjacent to the precipitates. The arguments against this theory are
the same as discussed previously for austenitic steels.

ui
»-
<
o
60 «
I-
<
a: 40 Z

io
O
c
X
o
o

TIME, h
FIG. 15—Variation of Type 430 stainless steel corrosion rate (in boiling 65
percent HNO,) and isolated precipitate with time of sensitization at 600 C [89].

Baumel [89] explains the intergranular corrosion susceptibility of the

ferritic stainless steels on the basis of chromium depletion adjacent to
chromium rich grain boundary precipitates (M23C0) as is commonly pro-
posed for the austenitic alloys. His main argument is based on the cor-
respondence between intergranular attack and the chromium content of
extracted precipitates as shown in Fig. 15. The maximum corrosion rate
coincides with the greatest rate of increase in the chromium content of
the isolate. This does support the depletion theory, although of course
all the extracted precipitates were not necessarily from grain boundaries.
Baerlecken et al [90] also attribute the problem to chromium deple-
tion adjacent to chromium carbides. More recently Demo [93,94], Bond
and Lizlovs [95,96], and Hodges [97] have supported the chromium
depletion theory and noted that nitrides (for example, CrzN) probably
play a role also. All of the supporters of the chromium depletion theory
concur that precipitation occurs in the temperature range of about 800

to 1700 F (427 to 927 C). Heating above this range dissolves carbides
and nitrides (in the matrix as well as the grain boundaries) and super-
saturates the ferrite. It seems likely also that some grain boundary segre-
gation of carbon and nitrogen will occur at these higher temperatures.
On cooling, the solubility of carbon and nitrogen is greatly lowered, and
carbides and nitrides precipitate very rapidly. At the lower end of the
precipitation range, say 800 to 1300 F (427 to 704 C), chromium
depletion occurs unless the material is held for extended periods of time
(hours) because chromium diffusion rates are low at these temperatures.
At the higher end of the range chromium diffuses more rapidly and
decreases the degree of chromium depletion and intergranular corrosion
susceptibility. Some precipitate agglomeration would be expected also at
these higher temperatures, and this in turn would decrease the possibility
of continuous grain boundary attack. This behavior is shown in Fig. 16.

1 1 1 < 1 I I I 1 1 - I -
tk SOLUTION _j
900
\
PRECIPITATION
800 I +Cr DIFFUSION
A66L0MERATK)N' ^-
1
u
X -
^ 700
at PRECIPITATION
^ 600 -^^—^^~>^!_ • - ^

500 - NO PRECIPITATION
^
400
10 100 400

HOLD TIME, MINUTES

FIG. 16—TT5 curves for Fe-26Cr alloy (C + N = 180 ppm) treated 1100 C,
30 min, water quenched, ferric sulfate-sulfuric acid tests [94].

By using the chromium depletion theory as the most feasible mechanism

it can be seen that the ferritic stainless steels are behaving similarly to
the austenitics. The main difference is that because of the very low solu-
bility of carbon and nitrogen in ferrite (compared to austenite) the pre-
cipitation reactions occur much faster. This is demonstrated in Fig. 17.
The presence of molybdenum in the iron-chromium ferritic stainless
steels appears to complicate the problem. Bond and Lizlovs [95] noted
that the presence of 1 or 2 percent molybdenum in 18 percent chromium
alloys resulted in intergranular attack in strongly oxidizing solutions in
material furnace cooled from 1500 F (816 C). This problem did not
show up in less oxidizing media (for example, copper-acidified copper
sulfate test) and was attributed to the same metallurgical changes that
cause embrittlement in these alloys after prolonged exposure at 700 to

1000 F ("885 F embrittlement"). Hodges [98], working with high-purity

alloys, attributes the role of molybdenum to that of slowing down the
diffusion of nitrogen so that nitride precipitation (and attendant chromium
depletion) occur at longer times.

Avoiding Intergranular Corrosion in Ferritic Stainless Steels

The most satisfactory way of avoiding intergranular attack is to use
these materials in the annealed condition, say 1450 F (788 C).
The addition of stabilizing elements (for example, titanium, columbium)
does minimize although not completely eliminate the problem [91,92,95,
105]. Stabilizing elements increase the sensitizing temperature and can

ISOO
(a) AUSTENITIC STAINLESS
•|400

5 1300
<
B:i20O
bJ
a.
311 too
u
(- I 10 10'
1000
TIME, MINUTES

I 10 10^ io« 10'

TIME, MINUTES
FIG. 17—Comparison of TTS curves for austenitic (Type 304) and ferritic
(Type 430) stainless steels treated 1950 F (schematic).

decrease susceptibility in the welded condition. Columbium is more effec-

tive in preventing intergranular attack in highly oxidizing media [95].
Lowering the carbon content to less than 0.03 percent does not prevent
attack as it does in the austenitics. Much higher purity (for example,
<0.005C, <0.015N) does improve toughness [87,99-102] and reduce
susceptibility to intergranular attack [96-98,103,104], but certain thermal
treatments can make even these alloys very susceptible to attack. For
example, Hodges [97] reports a corrosion rate of 439 /xm/year for a
17Cr-0.002C-0.006N alloy water quenched from 1850 F (1010 C) and
tested in boiling 65 percent HNO3. The same alloy showed a rate of
5010 jum/year when air cooled.

Duplex Stainless Steels

Austenitic-ferritic stainless steels containing about 5 percent or more
delta ferrite can have improved intergranular corrosion resistance com-
pared to austenitic material with the same carbon content [706-709].
Colombier and Hochmann [706] discuss some of the probable reasons
for the beneficial role of ferrite. They are as follows:
1. The total area of grain boundaries and phase interfaces is increased
in the presence of ferrite, thereby reducing the quantity of carbide pre-
cipitates per unit area.
2. The ferrite is significantly higher in chromium than is the austenite.
Therefore, it can lose chromium to a grain boundary precipitate and not
become sufficiently depleted to be preferentially attacked.
3. Chromium diffuses more readily in ferrite than austenite so there is
less chromium depletion.
4. The islands of ferrite present minimize the formation of continuous
grain boundary precipitates in the austenite grain boundaries.
Item 4 is probably the predominant factor.
On the other hand, delta ferrite can be detrimental to corrosion resis-
tance in certain media, particularly if a continuous network is present
[110-115], The problem is most prevalent in welds of the molybdenum
containing alloys. Types 316 and 316L. Preferential attack may occur
in the ferrite itself or at the ferrite-austenite interface and in the adjacent
austenite, depending upon the alloy and the corrodent. The media causing
attack are normally those with insufficient oxidizing power to induce
passivity. Problems have occurred in urea service, acetic acid mixtures, and
certain inorganic acid mixtures. Susceptibility to attack is readily revealed
in boiling 50 percent by volume hydrochloric acid (HCl) [775]. Remedial
measures include annealing after welding (some critical level of cold
work between welding and anneaUng may be required) or specifying low-
ferrite weld metal. TTie latter has the disadvantage that ferrite is benefi-
cial in minimizing weld cracking, and it should be noted also that it is
the distribution of the ferrite that is important from the corrosion stand-
point and not just the total amount present.
Ferrite containing weldments of Types 316 and 316L also may be
preferentially attacked by very strongly oxidizing media for example
boiling 65 percent HNO3. In this case the problem is probably related to
transformation products of the ferrite (for example, x' ^^^ ir-phase)
[110,114].
Just as the presence of ferrite can be sometimes beneficial to the in-
tergranular corrosion resistance of austenitic alloys the presence of austen-
ite improves the ferritic grades. This was discussed earlier for Type 430
which can form austenite at elevated temperatures, that is, greater than
about 1700 F (927 C). In that case the austenite transformed to martensite

on cooling, but it should be noted that austenite is also beneficial in the

higher chromium ferritic-austenitic alloys such as Type 329 (27Cr-5Ni-
1.5Mo) in which the austenite does not transform to martensite [106,
116-122]. Figure 18 compares the as-welded corrosion resistance of
low-and high-austenite material, and Fig. 19 shows the microstructures
in the high-temperature heat affected zones.
Considerable work has been done in Europe and the USSR oniFe-
2lCr-5Ni alloys, with or without molydenum and titanium [116-l\l9],
and 18Cr-8Mn-2Ni alloys [120-122]. It appears that these types of alloy
(and based on the author's experience, Type 329 also) behave more Uke
ferritic stainless steels than austenitics as far as intergranular corrosion

(a) Austenite content in base metal about 10 percent.

(b) Austenite content in base metal about 30 percent.
FIG. 18—As-welded Type 329 stainless steel after 28-h exposure to 10 percent
HNO, 4- 3 percent HF at 70 C ( x 2 ) .

'&m.

•A ^„ -sg '^ *^-

FIG. 19—Weld metal microstructure for specimens shown in Fig. 18 (xlOO).

Note larger amount of austeniie in weld of the more resistant composition.

is concerned. High-temperature treatments or welding can bring about

susceptibility. Heat treatments which provide resistance to attack involve
annealing at about 1750 F (954 C). Susceptibility can be induced by
holding annealed material in the range of about 800 to 1550 F (427 to
843 C). Levin and Kochergina [776] note that 2 percent molybdenum
suppresses the development of susceptibility when holding at 840 to 975 F
(449 to 524 C). This is in agreement with the findings of Bond and
Lizlovs [95] and Hodges [98] for purely ferritic alloys.
Stabilizing elements (especially titanium) are sometimes added to du-
plex stainless to further improve intergranular corrosion resistance. Some
[775,727] claim this to be beneficial and others [776,777] detrimental.
Smce most published work on duplex steels is lacking in measurements
of the amount and morphology (both of which are dependent upon prior

mechanical and thermal treatment) of the two phases present it is very

difficult to assess the role of elements such as carbon and titanium which,
in addition to influencing carbide precipitation, also control the ferrite-
austenite balance.
In summary, two-phase alloys have good intergranular corrosion resis-
tance provided the amount and distribution of the second phase are
such as to prevent continuous networks of grain boundary carbide.
High-Nickel Alloys
There are many commercially available, austenitic stainless alloys con-
taining more than 15 percent chromium with nickel ranging from 9 percent
(the common austenitic stainless steel level) to about 80 percent. Grades
with 15 percent or more molybdenum will be discussed separately. First,
Fe-Ni-Cr alloys (for example, Fe-20Cr-33Ni, Ni-15Cr-7Fe) will be
considered, although it should be noted that many of them also contain
a few percent of molybdenum and copper for added corrosion resistance.
Major reasons for the use of these high-nickel alloys in corrosion service
are their improved performance in reducing media and resistance to
transgranular stress corrosion cracking [123]. The intergranular corrosion
resistance of these high-nickel alloys has not been studied in such detail
as the austentitic stainless steels, but the data available show clear simi-
larities and differences in the two systems.
Both the stainless steels and nickel base alloys appear to be similar
in that sensitization treatments which cause attack usually involve grain
boundary carbide or intermetallic compound precipitation or both in the
range 800 to 1600 F (427 to 871 C) [124-134]. Welding, stress relieving,
and the slow cooling of thick sections are, therefore, potential sources of
trouble. There is very little published information as to what service media
will attack sensitized material (susceptible as determined by accelerated
laboratory tests). Indications are that the high-nickel alloys are similar
in this regard to the austenitic stainless steels. Their higher alloy content
provides improved resistance in many media, but on the other hand they
are often used in more severe corrodents in the first place.
Although the types of thermal treatment which cause attack are similar
for high- and low-nickel alloys, nickel does have a significant effect on
intergranular corrosion. Unfortunately, intergranular corrosion resistance
in the sensitized condition usually decreases as the nickel content is raised
[3,124,125,127]. One reason for this is that the solubility of carbon in
austenite decreases as the nickel content increases. For relatively low-
nickel levels Cihal [3] has determined the carbon content required to
make Fe-Cr-Ni alloys susceptible as a function of nickel content (See
Fig. 7). The influence of nickel in the range 13 to 46 percent is shown
in Fig. 20. It should be noted that the evaluation test used for the data
in Fig. 20 was not a severe one.

t- 600
<
K
Ul
2 500
111

TIME, h
FIG. 20—Influence of nickel content on TTS curves for Fe-20Cr alloys in CuSOf
HiSOi acid test [125].

Practical implications of this decreased carbon solubility as nickel

is increased are:
1. Maintaining carbon less than 0.03 percent does not minimize the
problem to the extent it does in the 18Cr-9Ni alloys [3,124,127].
2. Much larger amounts of stabilizing elements (columbium, titanium)
are needed to effectively minimize susceptibility [124,125,128] due to
chromium carbides, and at nickel levels of about 40 percent or more it
is reported that effective stabilization is not achievable with titanium
alone [129].
The high-nickel alloys appear to be more sensitive to processing vari-
ables (prior mechanical working, annealing, etc.) than the common
austenitic stainless steels [126,130]. Raymond [129] has emphasized the
importance of grain boundary precipitate morphology in these alloys.
Many of the high-nickel grades contain 2 to 3 percent or more molyb-
denum, and this further complicates matters (as in the stainless steels—
compare Types 316 and 304) by introducing more complex carbides and
intermetallic compounds such as <j-phase [126,130-132]. That molyb-
denum can be as important as carbon in these high alloys is demonstrated
in Fig. 21. In spite of this, some of the alloys with nickel content just
outside the stainless steel limits can be made to have resistance similar or
approaching that of the common austenitic stainless steels by control of
processing and analysis [130].
In the annealed and water-quenched condition high-nickel alloys are
generally resistant to intergranular attack as are austenitic stainless steels.

MOO 1200 ISOO 1400 1500 ISOO 'f

S93 649 704 WO ai$ 8TI "C
SENSITIZATION TEMPERATURE (1 HOUR.AC

Number % C % Mo
27 0.005 0.01
28 0.061 0.01
29 0.005 2.31
30 0.062 2.37
FIG. 21—Effect of carbon, molybdenum, and sensitizing temperature on inter-
granular corrosion of Fe-20Cr-33Ni-3Cu alloys. Third 48-h period rate in boiling
65 percent HNOi [130].

Like the latter they can be susceptible to attack in very strongly oxidizing
media (for example, nitric acid plus sodium dichromate). This suscepti-
bility is reported to increase with nickel content as shown in Fig. 22.
Intergranular corrosion susceptibility is probably related to the incidence
of intergranular stress-corrosion cracking (or perhaps more correctly—
stress assisted intergranular corrosion) of Ni-15Cr-7Fe alloys in high-
purity elevated temperature water [135-138].
In spite of their susceptibility to intergranular attack and reported cases
of intergranular stress-corrosion cracking, high-nickel alloys find ex-
tensive use in both oxidizing and reducing media and in situations (chlo-
rides, hydroxides) which cause transgranular cracking in the lower nickel
stainless steels.
Nickel-Chromium-Molybdenum A Hoys
Nickel-chromium-molybdenum (Ni-Cr-Mo) alloys are used because
of their resistance to chloride pitting, stress corrosion cracking, and a
wide variety of acid media which are corrosive to stainless steels. The
most commonly used alloy contains about 56 percent nickel, 16 percent
molybdenum, and 16 percent chromium, although higher chromium
versions find some use in Europe and the USSR.
One of the major drawbacks of these materials has been their sus-

200

FIG. 22—Influence of nickel content on intergranular corrosion of Ni-I7Cr-Fe

alloys. Tested 70 h in 5 N HNOs+ I g/liter Cr*" at 108 C [135].

ceptibility to intergranular corrosion if inadequately heat treated. The

temperatures producing susceptibility are in the range of about 1100
to 2100 F (593 to 1149 C). Grain boundary precipitation occurs in
short times, that is, in seconds or a few minutes at most. This means that
welds are susceptible, and there is a potential problem that annealed
material may be attacked if not cooled rapidly. Also, it is necessary
to anneal above a certain temperature, for example, 2225 F (1219 C)
while staying below temperatures of about 2300 F that cause incipient
fusion. Intergranular corrosion has been detected in a variety of acid
media, and the sensitizing temperature producing maximum susceptibility
depends upon the corrodent [139] as shown in Fig. 23.
The mechanism of attack is usually attributed to the precipitation
of grain boundary carbides and intermetallic compounds. These alloy
systems have been studied in some detail by Streicher [139], Protzl
[140], Class et al [141,142], Samans et al [143], Babakov et al [144],
and more recently by Leonard [145] and Hodge [146]. There is some
disagreement as to the identity of the precipitated phases responsible
for susceptibility.
It is generally agreed that annealed material resistant to attack con-
tains randomly distributed M„C carbides. Intermetallic compounds such as

Oxalic Add Etch Structure

High
Step Groove Ditch Step Temp.
-4*

1500 2000 2500

Temperature ol Heat Treatment I I Hr. I, °F. •

FIG. 2i—Schematic summary of relationship of heat treatment, etch structures

and corrosion of wrought Ni-16Cr-16Mo alloys [139].

P-phase [145,1471 "7-6" compound [143,145,148], ^^.-phase [146,148],

and (T-phase [140-142,144,147] have all been detected, and transforma-
tions from one form to another have been reported. These compounds
contain some or all of the following elements—nickel, iron, cobalt, tung-
sten, molybdemum, chromium. It seems reasonable to conclude that
they induce intergranular attack due to depletion of the adjacent matrix
in elements such as chromium and molybdenum, although Hodge [146]
has suggested that attack in oxidizing media is due to attack of the
precipitate itself because of its high-molybdenum content. Chromium
rich M.^q, carbides have been detected [140-142,144,147], and chro-
mium depletion adjacent to this or some other chromium rich carbide
may be a contributing factor to susceptibility. M02C has been found
[140,144], and molybdenum depletion adjacent to it is a possibility also.
Improved intergranular corrosion resistance can be obtained by limit-
ing the silicon content in the alloy [141,142,144]. Also a low-carbon,
low-silicon alloy is available, and although it does not eliminate sus-
ceptibility it does alleviate the problem considerably [145]. This reported
improvement derived from low silicon and carbon lends support to a
role for carbides as well as intermetallics. Streicher's observation (Fig.
23) that the sensitization temperature causing optimum attack in a
reducing acid (hydrochloric) diflfered significantly from that for an
oxidizing acid (chromic) is also consistent with there being more than
one precipitate involved. Grafen [142] saw somewhat similar behavior
but suggests that only one phase (sigma) is involved and that its com-
position varies with sensitization temperature. The author's experience
has been that carbon plays a significant role, and tr-phase is not the
only problem.

The prime methods for avoiding intergranular attack in these alloy

systems are to use material in the annealed plus water-quenched con-
dition or to use a high-purity alloy. Thermal treatments to stabilize the
deleterious intermetallic compounds have been proposed by Samans
et al [143], and Streicher [139] notes that multiple short anneals can
provide better resistance than a single longer time treatment.
Nickel-Molybdenum Alloys
These alloys do not contain chromium and thus differ from all the
other materials discussed in this review. They are included because
they are used in severe corrodents and do complement the previously
discussed high-nickel grades. The most well-known alloy in this cate-
gory contains about 65 percent nickel, 28 percent molybdenum, and
5 percent iron. It has excellent resistance to reducing media and in
particular HCl.
The Ni-28Mo alloy can be susceptible to intergranular corrosion.
For example, as-welded material is susceptible to intergranular attack
in boiling 20 percent HQ and in some sulfuric acid (HoSO^) (for
example, boiling 60 percent). The attack may occur in the weld metal
itself, in the base metal almost immediately adjacent to the weld, and
in a lower temperature zone which was heated in the region of 1450 F
(788 C).
The intergranular corrosion resistance of this alloy system has been
studied in some detail by Flint [149], Grafen [142], and Svistunova and
co-workers [150-154]. In the annealed condition, 2100 F (1149 C), the
alloy contains randomly distributed MGC type carbides of approximate
composition Ni^MogC. As the temperature is increased toward the
melting point there is a tendency for the molybdenum content of the
M„C to increase, for example, to Ni^Mo^C [142,149] and for Mo^C
to form [142,149,152,153]. Grafen [142] also noted the presence of an
iron-molybdenum type phase which contained more than 60 percent
molybdenum at 2300 F (1260 C).
Intergranular corrosion in and immediately adjacent to weld metal is
attributed generally to preferential attack of molybdenum depleted zones
adjacent to high molybdenum MBC and M02C with the latter probably
playing the major role.
Susceptibility in material exposed to temperatures of about 1450 F
(788 C) is again attributed to molybdenum depletion, in this case ad-
jacent to the intermetallic compound NiiMo [142,147,150,152,154] pre-
dicted from the nickel-molybdenum phase diagram.
Attempts to totally eliminate intergranular corrosion susceptibility
in weldments have been unsuccessful. Lowering carbon ( < 0.005 per-
cent) has been reported to remove the high-temperature problem [149,
153] which is in line with the above mechanism relating the attack to

carbides. Lowering the iron content (from 5 percent to less than 2

percent) [149,153] or adding a few percent tungsten [152,154] are also
reported to be beneficial. Vanadium additions (1 to 2 percent) have
been studied in some detail, with Flint [149] and Svistunova et al [150]
claiming the addition to be beneficial. Svistunova et al [755] later
show it can be detrimental which is consistent with the author's ex-
perience. Intergranular corrosion susceptibility produced after exposure
to lower temperatures such as 1450 F (788 C) is not avoided by de-
creasing the carbon content as expected from the above mechanism re-
lating susceptibility to Ni4Mo. Vanadium has been reported to both re-
tard [150] and accelerate [153] attack and tungsten to retard and narrow
the temperature range producing it [154].
Commercial alloys completely immune to attack are not available.
Some of the above alloys variations may minimize the problem, but
the only completely satisfactory remedy for service in media capable
of producing attack is to use annealed material.
Intergranular Corrosion Tests
Corrosion tests can be classified as: (1) service, (2) simulated service,
or (3) accelerated. Most intergranular corrosion tests are a particular
type of accelerated test, namely, quality control. In the latter, the test
media may bear little relationship to the intended service environment.
However, it is capable of detecting, in several hours or days, a metal-
lurgical condition suspected or known to result in intergranular attack
in service.
The development of intergranular corrosion tests has centered pri-
marily on the austenitic stainless steels, but many of the methods devel-
oped can be applied to other stainless and high-nickel alloys. There
are currently five established test methods for detecting susceptibility to
intergranular attack in stainless steels and one etching technique which
can be used as a screening test. All of these are ASTM standards and
are listed in Table 2.
Although it is convenient and complimentary to use the inventor's name
to describe a test, the practice is of dubious merit. For example, its pre-

TABLE 2—Standard intergranular corrosion tests for stainless steels.

ASTM Standard Test Media Common Test Names
A 393-63 boiling 5.7% CuSOi + 15.7% USO, Strauss, Hatfield,
Krupp
A 262-70, Practice A 10% H2Cs04, electrolytic at Streicher
ambient temperature
A 264-70, Practice B boiling 50%, H,SO, + Fej(S04)3 Streicher
A 262-70, Practice C boiling 65%, HNOj Huey
A 262-70, Practice D 10%, HNO, + 3 % HF at 70 C Warren
A 262-70, Practice E same as A 393 but with specimen in Accelerated Strauss
contact with copper metal

cise origin may be in doubt and the test may be modified as the years
go by, for example, the copper sulfate-sulfuric acid (CuS04-H:;S04)
solution. Also, an inventor might have more than one test, for example,
in addition to the ferric sulfate-sulfuric acid (Fe^,(SO^).j-H^,S04) test
there is a pitting test named after Streicher, and the oxalic acid (H:;C204)
etch test is also labelled by some as the Streicher test. For these reasons
it is generally preferable to identify by the test media.
Copper Sulfate-Sulfuric Acid
Warren [4] has reviewed the history of this solution in some detail.
The ability of CuSO^-H^SO^ mixtures to detect susceptibility to inter-
granular corrosion in sensitized stainless steels was noted by Hatfield in
1926. The solution was used by Hatfield [155] as a test and also modified
and promoted by Strauss [//] of the Krupp organization. Several further
modifications were made to yield the current ASTM Standard Recom-
mended Practice for Conducting Acidified Copper Sulfate Test for In-
tergranular Attack in Austenitic Stainless Steel (A 393-63).
The test specimen is immersed in the solution for 72 h. There is
very little grain dropping so a weight loss evaluation is not pertinent. In-
tergranular attack can be evaluated by the ring test (when a severely
attacked specimen is dropped on a hard surface it will not give a clear
ring), electrical resistance measurements, or more commonly by bend-
ing the specimen to open up any attacked grain boundaries and ob-
serving at low magnification (5 to 15 magnification). Metallographic
examination of sections from the tested specimen may be used to further
evaluate the degree of attack. It should be noted that in European
countries and on some material specifications in the United States the
solution composition and severity of the bend to be given after the test
often differ from ASTM A 393.
The fact that the method does not give a number (unless resistance
measurements used) is a disadvantage. Also, it is not very severe and
times longer than 72 h are often required to approach the degree of
attack obtained with some of the other evaluation solutions. Plans are
underway to drop this test from the ASTM standards because of its lack
of severity.
Recently the test has been improved by placing the specimen in con-
tact with copper in accordance with ASTM Standard Recommended
Practice for Detecting Susceptibility to Intergranular Attack in Stain-
less Steel (A 262-70, Practice E). The copper increases the severity
of the solution probably because it maintains the specimen potential
in a region more aggressive to chromium depleted grain boundaries. It is
only necessary to expose the specimen for 24 h, and in some European
countries a similar test is run for 15 or 16 h.
The addition of copper to the solution was originated by Rocha in

the midl950s [156], and the work leading to the ASTM standard is de-
scribed by Scharfstein and Eisenbrown [157]. There has been some dis-
agreement as to whether contact between the specimen and the copper
is necessary. Contact with copper (as compared to just the presence
of solid copper in the solution) moves the specimen potential very
slightly in the active direction and apparently increases the rate of at-
tack [25,157], This has been disputed by Schwenk [158-160] who claims
contact is not necessary and suggests that the reported differences [157]
are due to the use of cold finger condensers compared to his bulb con-
denser which in turn influences the oxygen content of the solution [160].
Since the ASTM standard specifies contact with copper and a bulb
condenser it would seem to minimize these possible variations and rep-
resent the most severe test. A CuS04 test in 50 percent H2S04 and
with metallic copper in the solution but not in contact with the speci-
men also has been reported [39,103] as a more severe test.
Whether copper is added to the solution or not the CUSO4-H2SO4
solution detects intergranular corrosion susceptibility that is primarily
due to chromium carbides rather than intermetallics such as (r-phase
[4,25,32,161-163].

Boiling 65 Percent Nitric Acid

In this test (ASTM Standard A 262, Practice C) the specimen is
exposed for 240 h (usually five 48-h periods using fresh solution for
each). Weight losses are used to calculate corrosion rates for each
period. In susceptible material the rates will usually increase with in-
creasing test periods because whole grains are lifted from the surface.
The solution and testing method were first reported by Huey [164],
although he was not evaluating intergranular corrosion. It was later real-
ized that the solution could be utilized in quahty control work [165].
The influence of test variables has been studied in detail by Truman [63]
and Streicher [25].
There are three potential disadvantages to the HNO, test:
1. It takes a long time and is expensive.
2. The degree of intergranular attack is accelerated by hexavalent
chromium which builds up during the 48-h test periods. This can result
in differing corrosion rates for different solution volume to specimen
surface area ratios.
3. The solution is very sensitive, to surface preparation [25], end grain
[25] and structural changes other than chromium carbide, for example,
it attacks ^-phase [4,25,32,39,161-163] and TiC [41,42].
Over the years these disadvantages have been countered with the
argument that because the method is quantitative and severe it is the
best one to use for quality control. For many years, it and the CuSO^-
H0SO4 test were the only ones available, and these arguments were

somewhat valid. Now however, there are quantitative alternatives (for

example, Fe2(S04).s-H2S04 test), and also the CUSO4-H2SO4 test has
been made more severe with the addition of copper.
The expense of the test and the effects of Cr+* can be minimized
by using a multi sample tester (MST). These have been discussed in
detail by DeLong [166], Truman [63], and Alger et al [167]. The acid
is distilled continuously to remove Cr+* as shown in Fig. 24, and as
many as 60 specimens may be tested simultaneously. Because Cr+* is
removed continually in an MST the corrosion rates are lower than those
obtained in a regular flask test. When the rates are low (that is, little
susceptibility to intergranular corrosion) the difference is not usually
significant [765]. However, when there is considerable susceptibility to
intergranular attack a flask rate may be as much as 50 percent higher'
than that in an MST. The problem is complicated further by the fact
that if one has a lot of highly susceptible material in an MST then the
effects of Cr+'' buildup may not be completely eliminated.

Wot«r lnl«t-
Waltr Outlet

Sp«cim«n
Hold«r

T«st
Chambtr
(Surrounded by
26-ga wire)

t; "• -v/y*—Droin

FIG. 24—Multi sample tester for boiling 65 percent HNO> test [167].

MST's are satisfactory for quality control work provided the mate-
rial is good quality and shows a rate significantly below the acceptance
limit (which is usually based on flask data). Material showing rates
approaching the acceptance limit may have to be retested in a flask.
MST's may be used also in research and are ideal for running a com-
parison series where all the specimens can be tested simultaneously.
However, if one wishes to precisely compare the rates with past or
future studies there can be complications for the reasons discussed above.
The 65 percent HNO.s test is still popular in the United States but
less so in Europe. With all its problems it would seem that except for
evaluation prior to HNO3 service its use will decrease in favor of the
Fe2(S04)3-H2S04 test and the CUSO4-H2SO4 test with copper contact.
Ferric Sulfate-Sulfuric Acid
Fe^CSO^)^ like CuSO^, inhibits the general corrosion of stainless steels
in H._'S0^ [169J70]. Streicher [25,170,171], found that by adjusting the
Fe2(S04)3-H^S04 composition it was possible to inhibit the acid suffi-
ciently to avoid a high rate of corrosion on the grain faces but attack
grain boundary regions whose corrosion resistance had been lowered
by chromium carbide precipitation. Because the solution produces grain
dropping in susceptible austenitic stainless steels in 120 h or less, it
is amenable to weight loss measurements and quantitative evaluation
(ASTM Standard A 262, Practice B).
There are several practical aspects of the test worthy of note. First
it is necessary to use a bulb condenser (not a finger condenser) and en-
sure a perfect flask/condenser seal. Otherwise, some evaporation may
occur which increases the rate of general corrosion of the specimen.
Second it is important to remove all traces of scale from the test speci-
men otherwise it may go active and yield a high rate of general corrosion
throughout the test. Along the same lines it is important not to put the
specimen into the acid before all the inhibiting ferric sulfate has dis-
solved. Another disadvantage is the necessity to boil HCl in the flask and
condenser after the test to remove an iron oxide residue. An advantage
over 65 percent HNO3 is that corrosion products produced during the
test do not influence the result.
The Feo(S04)3-HoSO^ test is sensitive to grain boundary chromium
depletion. It primarily detects problems associated with chromium car-
bides and, to a lesser extent, tr-phase in Type 321 (titanium stabilized),
but not in the molybdenum bearing Type 316 steels [39]. It does not
produce severe end grain attack of the type sometimes obtained in HNO3.
The method is particularly suitable for evaluating high-chromium
stainless steels (for example, 26 percent chromium ferritic [123]) and
appears to be more sensitive than the HNO3 test for detecting chromium
depletion in these alloys. The test was applied by Streicher [139] to

Ni-Cr-Mo alloys and more recently has been applied to other high-
nickel alloys [126,130,145]. Its application to all of these high-nickel
alloys is covered in a new standard, ASTM Test for Detecting Sus-
ceptibility to Intergranular Attack in Wrought Nickel-Rich. Chromium-
Bearing Alloys (G 28-72).

Nitric Acid-Hydrofluoric Acid

HNO3-HF solutions have been used for many years to evaluate inter-
granular corrosion resistance in austenitic stainless steels. Warren [172]
reviewed the history of this solution, and also performed an exten-
sive evaluation of it. He was interested primarily in developing a
quantitative test, for AISI Types 316 and 316L, which would detect
intergranular corrosion susceptibility due to chromium carbide but not
that due to <T-phase.
Warren's work, which included an evaluation of many production
heats, resulted in the ASTM Standard A 262, Practice D, which con-
sists of two 2-h exposure periods. Corrosion rates are calculated from
weight loss data, and it is customary to compare these rates to those
for material known to be resistant.
Disadvantages of the method are: (a) there is a high rate of general
attack which varies greatly with alloy content; (b) maintaining a tempera-
ture at 70 C can be less convenient than using a boiling solution; and
(c) hydrofluoric acid requires special handling, for example, no glass
equipment. Even though the method is rapid it is not used widely for
these reasons and because tests such as the Fe2(S04)?,-H2S04 and the
copper accelerated, C U S O J - H S S O J tests are available.

Oxalic Acid Etch

HoC^O^ is commonly used in metallography as an electrolytic etch
for stainless steels. Streicher developed this etching technique into a
procedure (ASTM Standard A 262, Practice A) for screening out
material resistant to intergranular corrosion due to chromium carbides
[173]. The test requires no costly specimen machining and can be run
in less than 30 min.
The specimen is polished to a 000 finish and then etched for IVz
min at 1 A/in.- in room-temperature 10 percent H2C0O4. Based on the
degree of attack at the grain boundaries the material is either deemed
completely resistant to attack or must be subsequently tested with one
of the previously described methods. Note that the etch test does not
reject material—it only passes good quality material. The test is appli-
cable to AISI Types 304, 304L, and probably other nonstabilized aus-
tenitics for all tests and for Types 316, 317, and their L grades for
all tests except 65 percent HNO3.
The oxalic acid test is very useful from both a time and economics

Standpoint. Unfortunately the percentage screened is quite low for one

of the most common grades for which it is used (Type 304L). Streicher
[173] reports it as 56.4 percent for the 65 percent HNO.t test, that is,
the oxalic acid etch test avoided the HNOs test only 56.4 percent of the
time on "good" material. He has proposed a method of increasing this
percentage, namely, to sensitize the H2C2O4 acid etch specimen 20 min
at 1250 F (677 C) rather than 60 min [171]. This obviously does in-
crease the number of specimens passing the H2C;>04 test, and the time
more closely simulates that in most welding applications.
A screening test of this general type is needed for quality control of
the increasingly used ferritic stainless steels. They can be sensitized dur-
ing production by annealing at too high a temperature, and more im-
portantly they are then susceptible to attack in relatively mild media.
Electrochemical Methods
Intergranular corrosion in stainless steels is predominantly under an-
odic control, and grain boundary attack is largely due to less anodic
polarization at the grain boundaries [25]. It is not surprising, therefore,
that materials very susceptible to intergranular attack can show anodic
polarization behavior different to that of resistant material. Several ex-
amples of this have been reported [16,95,174-180], and typical be-
havior for austenitic and ferritic stainless steels is shown in Figs. 25
and 26, respectively.
Because the boundary area causing the increased current density in

1 1 1 1 1
1300
bl ^-NO SENSITIZATION ^,.g
-J
<
U
m
z 900 / / / /^lOOO h -
iii
(9
0
/ / r "'/
K
0
>-
X
500
*^ Lo.iti -
>E
-I
<
t-
100 ^*s^^
z111
1-
0
-100

1 1 1 1 1

10 10 10 10 10 10 10'

CURRENT DENSITY, (lA/em^

FIG. 25—Anodic polarization of Type 304 stainless steel after sensitization from
0 to 1000 h at 650 C. 2 "H H>SOt at 90 C [16],

ANNEALED
ISOOF.W.q,

111
u
10
m
>
m
o
>

z
Ul

CURRENT DENSITY, rnA/cm*

FIG. 26—Anodic polarization of ferritic 18Cr-2Mo stainless steel in 5 percent
Hi^Ot at 24 C [95].

Figs. 25 and 26 is very small, it is predicted that the chromium de-

pleted areas adjacent to the grain boundaries actually have the anodic
polarization behavior shown in Fig. 27. This is consistent with the anodic
polarization curves in Fig. 28 which are typical for Fe-Cr alloys with
chromium varied from 0 to 18 percent [181,182].

Ill
I-
o
a.

LOG. CURRENT DENSITY

(a) Solution annealed.
(b) Sensitized.
(c) Chromium depleted zone.
FIG. 27—Anodic polarization of Type 304 stainless steel in deaerated HtSO^
(schematic) [182].

10" 10 10' 10" 10' 10° 10° 10'

CURRENT DENSITY, (lA/em^

FIG. 28—Effect of alloy chromium content on anodic polarization in 2 N HtSOt
at 90 C. Nickel content in range 8.32 to 9.85 percent [182].

Although anodic polarization curves, such as shown in Figs. 25 and

26, can distinguish between susceptible and resistant material they are
probably not a very reliable technique unless the material is in an ex-
tremely susceptible condition. One reason for this is that the grain
boundary areas constitute a very small part of the total surface whose
behavior is reflected in the polarization curve. Also they do not dis-
tinguish between grain and grain boundary precipitation and do not
take account of precipitate morphology. Another aspect is that they
do not allow for the time factor often necessary for significant inter-
granular attack in service.
Anodic polarization curves can be useful in indicating which potential
regions are most likely to cause attack. Then by performing long-term
potentiostatic tests in these regions one can obtain potential—sinter-
granular susceptibility relationships for the test media and time involved.
This technique has been used by several [16,175,176,177,179,183] with
detailed studies by Osozawa et al [76] and France and Greene [183].
Discussing the latter paper, Streicher [184,185] emphasizes that the po-
tentiostatic tests are only good for the time they were run and the speci-
fic media (and temperature) involved. To demonstrate his argument
he shows that intergranular attack can occur in extended periods of
time under conditions claimed not to produce attack by France and
Greene. Possible reasons for these differences have been suggested [183],
but, taking care to avoid errors in experimental technique that could
influence these results, the writer's experience confirms that of Streicher.
Another argument made by Streicher concerning the limitations of

potentiostatic intergranular corrosion tests is that they can be achieved

less expensively in other ways. For example, the standard corrosion
tests previously discussed cover a wide range of potential (see Fig. 29),
and each one is in effect a "chemical potentiostat." Therefore, one can
select a standard test to represent potential ranges of interest.
Potentiostatic etch tests (several minutes in duration) have been
claimed to be suitable for evaluating susceptibility to intergranular at-
tack. The application of this technique has been reviewed by Linder
[187]. A polished specimen is held (usually in H2SO4) at a potential just

LOG. CURRENT DENSITY —

FIG. 29—Potential regions for Type 304 stainless steel in standard test solutions
{Table 2) compared to anodic polarization curve in H^^SOi (schematic).

more noble than the critical passivation potential. It is recommended

that such methods be carefully related to other quality control tests and
service experience before putting too much reliance on them.
In short it would seem that electrochemical tests (other than simple
etch tests for screening purposes) have little to offer over available
standard accelerated test media particularly when one considers the
cost of the equipment (potentiostat, etc.) needed for the former.
Other Evaluation Techniques
In most of the previously discussed quality control tests the method
of evaluation was a weight loss which in turn was converted to an
average corrosion rate. Ideally this rate is then compared to that for
well aimealed material (immune to intergranular attack). This pro-
cedure is generally satisfactory provided: (1) there is significant attack,
(2) the corrodent is one which dislodges grains in susceptible material,
and (3) the grain size is not unusually large.

The Other most commonly used nondestructive evaluation technique

is measurement of electrical resistance [78,188]. The method has the
advantage of providing accurate "depth of attack"—"exposure time"
information but does require specially shaped specimens (for example,
wire or thin strip) and sensitive measuring apparatus. Another non-
destructive technique for detecting the existence of intergranular corro-
sion is the use of dye penetrants. This is obviously attractive for in-
service inspection" or routine inspection during the overhaul of com-
ponents prior to reassembly and return to service.
Commonly used destructive methods include metallography (usually
to measure the depth of attack) and the measurement of a loss in me-
chanical strength as determined by bend or tension tests.
Etch tests, internal friction [189], magnetic measurements [790], Moss-
bauer spectroscopy [797], and thermoelectric methods [792] have all been
used to investigate susceptibility to intergranular corrosion (that is, with-
out doing a corrosion test). The latter four have not been investigated
very thoroughly and at this time appear to have little to offer over the
simpler, commonly used etch techniques.
Sensitization to Simulate Welding
Intergranular corrosion susceptibility is very often a result of welding.
There are many problems involved in the corrosion evaluation of a
weldment, and the conventional weight loss techniques are of limited
value. A corroded weldment may show one or more of the forms of
attack shown in Fig. 30. Suggested techniques for evaluating these types
APPEARANCE CORROSION TYPE

WELD METAL

I TT
BASE METAL

BASE METAL
O:
WELD METAL
M:
BASE METAL, HIGH
I 1r TEMPERATURE HEAT
AFFECTED ZONE

BASE METAL, LOW

TEMPERATURE HEAT
AFFECTED ZONE

FIG. 30—Types of weldment corrosion [193].

of preferential attack are currently being studied by the Corrosion Sub-

committee of the High Alloys Committee in the Welding Research
Council [193]. It should be noted also that Pinnow and Moskowitz [194]
have reviewed the corrosion of stainless steel weldments and collected
a comprehensive bibliography on the subject.
Here we shall discuss the use of artificial heat treatments used to
simulate welding problems. The one most commonly used for austenitic
stainless is 1250 F (677 C), 1 h. For most welding applications this
represents a much longer time at temperature than occurs during weld-
ing and is generally used with the justification that it is more severe
than the latter. This is generally true as shown in Table 3 which sum-
marizes data from a Welding Research Council test program [6].
TABLE 3—Comparatire /teverity of welding and heat treatment [6].
Number of Racks Showing
lutergranular Attack"
AISI
Type Welded Heat Treated*
Slfif- 0 11
316 1 15
317 G 22
318 1 21
302 15 19
304 1 10
304L 0 5
321 6 12
347 1 13
° Racks (24 specimens) exposed to a number of actual service environments for times
ranging from 30 to 1600 days.
' Specimens heat treated for 1 to 4 h at 1100 to 1250 F.

It should be noted, however, that a 1250 F (677 C) treatment on

annealed material may bear littie relationship to welding if some metal-
lurgical changes occur in the temperature range between the annealing
temperature and the melting point. For example a stabilizing carbide
may be solutioned during welding thus making the weld and high-tempera-
ture heat affected zones much more susceptible to sensitization than an-
nealed material. This can be simulated by giving a double "sensitization,"
say 2300 F (1260 C) + 1250 F (677 C). One has to be careful not
to change the bulk analysis (for example, by loss of carbon, nitrogen,
or chromium) during the high-temperature treatment.
To simulate the weldment corrosion of ferritic stainless steels, sen-
sitization consists of heating to above about 1700 F (927 C), that is,
the sensitization is above the annealing temperature which is typically
1450 F (788 C). The rate of cooling from the sensitization tempera-
ture can be very important in these alloys because the actual damag-
ing precipitation occurs (very rapidly) at lower temperatures such as
HOOF (593 C).

The simulation of welding in some of the nickel base alloys is even

more diSicult than for the stabilized austenitic stainless grades. This is
because there may be a combination of deleterious precipitates—some
formed by exposure at intermediate temperatures such as 1200 to
1600 F (649 to 871 C) and others formed (or dependent upon ex-
posure at) temperatures approaching the solidus temperature.
Sensitization treatments are used obviously with credibility to study
stress relieving, but to simulate welding they leave much to be desired.
Unless a correlation is known between weld performance and a specific
sensitization treatment it is usually preferable to test weldments. Of
course in many cases this also has limitations because it may be difficult
to simulate bulk welding techniques (for example, multipass) on speci-
mens suitable for corrosion testing.

Relationship Between Intergranular Corrosion and

Otiier Forms of Localized Corrosion
Stress-corrosion cracking is sometimes intergranular. In such cases
it is possible that the only role of the stress is to open up the crack
allowing ready access of corrodent through corrosion product which
might otherwise stifle the corrosion reaction. In such a case the phe-
nomenon might more correctly be called "stress accelerated intergranu-
lar corrosion." It is probable that many, but not necessarily all, cases
of intergranular stress corrosion cracking fit into this category. Specific
examples for some of the alloy systems discussed in this review are:
1. Polythionic acid cracking in stainless steels [195].
2. Cracking of Ni-Cr alloys in high-temperature water [135-138].
3. Cracking of annealed stainless steels in high-temperature water
[796-/95].
4. Cracking of sensitized stainless steels in marine environments.
5. Cracking of sensitized stainless steels in caustic environments [199,
200].
In all of these examples, factors which minimize intergranular corrosion
susceptibility also improve stress corrosion cracking resistance but do
not necessarily eliminate cracking.
Pitting can be connected with intergranular corrosion in at least
two ways. First an intergranularly corroded surface can act as a crevice
to develop the environmental conditions for pitting as shown for an
austenitic stainless steel in Fig. 31. In this example some intergranular
corrosion occurred during the pickling (HNOs-HF) of a carburized
surface prior to seawater service. Conversely the environmental condi-
tions within a pit may cause intergranular attack which propagate the
pit even though the bulk solution does not attack the grain boundaries.
Also, Streicher [73] has noted that even in the absence of intergranular

FIG. 31—Corrosion pitting initiated hy intergranular corrosion (y^WO).

attack, pits in austenitic stainless steels tend to initiate at grain bounda-

ries.
References
[1] McLean, D., Grain Boundaries in Metals, Oxford University Press, London,
1957.
[2] Ebling, H. F. and Scheil, M. A. in Advances in the Technology of Stainless
Steels and Related Alloys, ASTM STP 369, American Society for Testing
and Materials, 1963, pp. 275-282.
[51 Cihal, v . . Protection of Metals. PTNMA. Vol. 4. No. 6, Nov./Dec. 1968,
pp. 563-577.
[4] Warren, Donald, Corrosion, SBllA, Vol. 15. April 1959, pp. 213t-220t.
[51 Auld, J. R. in Advances in the Technology of Stainless Steels and Related
Alloys, ASTM STP 369, American Society for Testing and Materials, 1963,
pp. 183-199.
[6] "Intergranular Corrosion of Chromium-Nickel Stainless Steels—Final Report,"'
Bulletin, Welding Research Council, WRCBB, No. 138, New York, 1969.
[7] Bain, E. C , Aborn, R. H., and Rutherford, J. B., Transactions, American
Society for Steel Treating, TASTA, Vol. 21, 1933, pp. 481-509.
[8] Mahla, E. M. and Nielsen, N. A., Transactions, American Society for Metals,
TASEA, Vol. 43, 1951, pp. 290-322.
\9] Kinzel, A. B., Transactions, Metallurgical Society of American Institute of
Mining, Metallurgical, and Petroleum Engineers, TMSAA, Vol. 194, 1952,
pp. 469-488.
[10] Philibert, J., Henry, G., Roberts, M., and Plateau, J., Revue de Metallurgie,
REMEA, Vol. 58, 1961, p. 557.
[//] Strauss, B., Schottky, H., and Hinnuber, J., Zeitschrift Anorganische Chemie,
ZACMA, Vol. 188. 1930. p. 309-324.
[12\ Hatwell, H., "Study of the Precipitation of Chromium Carbide Responsible
for Intercrystalline Corrosion in Austenitic Stainless Steels," 3rd Annual
Colloquium on Metallurgy, Corrosion of Steel, British Iron and Steel Industry
Translation No, 2771, J'lr.o iP*-:.

[13] Stickler, R. and Vinckier, A., Transactions, American Society for Metals,
TASEA, Vol. 54, 1961, pp. 362-380.
[14] Stickler, R. and Vinckier, A., Corrosion Science. CRRSA, Vol. 3. 1963,
pp. 1-8.
[15] Stefanides, N. V., Transactions, American Society for Steel Treating. TASTA,
Vol. 19, 1931, pp. 742-746.
[16] Osozawa, K., Bohnenkamp, K., and Engell, H. J., Corrosion Science, CRRSA,
Vol. 6, 1966, pp. 421-433.
[17] Baumel, A., Buhler, H. E., SchuUer, H. J., Schwaab, P., Schwenk, W.,
Ternes, H., and Zitter, H., Corrosion Science, CRRSA, Vol. 4, 1964, pp. 89-
103.
[18] Wilson, F. G., British Corrosion Journal, BCRJA, Vol. 6, May 1971, pp.
100-108.
[19] Philibert, J., Crussard, C , Waclie, X., and Gerber, M., Comptes Rendus,
Vol. 251:2, 1960, pp. 1289-1291.
[20] Aim, S. and Kiessling, R., Journal, Institute of Metals, JIMEA, Vol. 91,
1963, p. 190.
[21] Ceilings, P. J. and deJongh, M. A., Corrosion Science, CRRSA, Vol. 7, 1967,
pp. 413-421.
[22] Stawstrom, C. and Hillert, M., Iron arid Steel Institute, Jan. 1969, pp.
77-85.
[23] Tedmon, C. S., Jr., Vermilyea, D. A., and Rosolowski, J. H., Journal, Elec-
trochemical Society, JESOA, Vol. 118, No. 2, Feb. 1971, pp. 192-202.
[24] Cihal, V. and Kasova, Irena, Corrosion Science, CRRSA, Vol. 10, 1970, pp.
875-881.
[25] Streicher, M. A., Electrochemical Society, Vol. 106, 1959, pp. 161-180.
[26] Dillon, C. P., Corrosion. SBIIA, Vol. 16, Sept. 1960, pp. 433t-440t.
[27] Savkina, L. Ya. and Fel'dgandler, E. G., Metal Science and Heat Treatment,
MHTRA, Nos. 11-12, Nov./Dec. 1968, pp. 857-860.
[28] Scharfstein, L. R. in Effects of Residual Elements on Properties of Austenitic
Stainless Steels, ASTM STP 4IH, American Society for Testing and Ma-
terials, 1967, pp. 90-104.
[29] Lennartz, G., Mikrochimica Acta, MIACA, Vol. 3, 1965, pp. 405-428, British
Iron and Steel Industry Translation No. 4639, Iron and Steel Institute, Lon-
don, 1965.
[30] Binder, W. O., Brown. C. M., and Franks, F., Transactions, American Society
for Metals, TASEA, Vol. 41, 1949, pp. 1301-1346.
[31] Binder, W. O. and Brown, C. M. in Evaluation Tests for Stainless Steels.
ASTM STP 93. American Society for Testing and Materials, 1950, pp. 146-
171.
[32] Warren, D., Corrosion, SBIIA, Vol. 15, No. 5, May 1959, pp. 221t-232t.
[33] Castro, R., Hochmann, J., and Granjon, H., Revue de Metallurgie, REMEA,
Sept. 1965, pp. 847-850.
[34] Bourrat, J. and Hochmann, J., Bulletin du Cerde d'Etudes des Metaux, Vol.
9, Dec. 1964, pp. 285-303, British Iron and Steel Industry Translation 4129,
Iron and Steel Institute, London, 1965.
[35] Franks, R., Corrosion Handbook, Uhlig, ed., Wiley, New York, 1948, p. 160.
[36] Brown, M. H., DeLong, W. B., and Myers, W. R. in Evaluation Tests for
Stainless Steels, ASTM STP 93, American Society for Testing and Materials,
1950, pp. 103-119.
[37] Rosenberg, S. J. and Darr, J. H., Transactions, American Society for Metals,
TASEA, Vol. 41, 1949, pp. 1261-1288.
[38] Phillips, F. J., Transactions, American Society for Metals, TASEA, Vol. 39,
1947, pp. 891-906.
[39] Streicher, M. A., Corrosion, SBIIA, Vol. 20, No. 2, Feb. 1964, pp. 57t-72t.
[40] Levin, I. A. (ed.) Intercrystalline Corrosion and Corrosion of Metals Under
Stress, Consultants Bureau, New York, 1962.

[41] Cihal, v., KaSova, I., and Kubelka, J., Metaux Corrosion-Industries, MTUXA,
No. 529, Sept. 1969, p. 281.
[42] Schwaab, P., Schwenk, W., and Ternes, H., Werkstoffe und Korrosion,
WKKOA, Vol. 16, 1965, p. 844.
[43] Levitin, V. V. and Syreishchikova, The Physics of Metals and Metallography,
PHMMA, Vol. 7, No. 2, 1959. pp. 152-154.
[44] Farrell, J. W. and Rosenthal, P.. Metal Progress, MEPOA, Feb. 1960, pp.
101-103.
[45] Voeltzel, J. and Henry, G., Revue de Metallurgie, Part I Memoires, RVMMA,
Vol. 61, Feb. 1964, pp. 123-128, British Iron and Steel Industry English
Translation 4493, Iron and Steel Institute, 1965.
[46] Rundell, G. R.. "Austenitic Stainless Steel." U. S. Patent 3,306,736. Feb.
1967.
[471 Costello, T. M., Pinnow. K. E., and Moskowitz, A., Materials Protection.
MAPRA, Vol. 8, Nov. 1969, pp. 15-18.
[4R] Bungardt, K. and Lennartz, G., Archiv fiir das Eisenhiittenwesen, AREIA,
Vol. 34, July 1963, pp. 531-546, British Iron and Steel Industry English
Translation 3439, Iron and Steel Institute, London, 1963.
[49] Plumensi. J. P. and Kohn, A., "Distribution of Boron in 18-10 Stainless Steel
as Detected by Autoradiography." Comptes Rendus. Vol. 258, 1964. pp.
5651-5653.
[50] Franks, R., Binder, W. O., and Thompson, J., Transactions, American Society
for Metals, TASEA, Vol. 47, 1955. pp. 231-259.
[51] Renshaw, W. G., and Lula, R. A., Transactions, American Society for Testing
and Materials, Vol. 56, 1956, pp. 866-889.
[52] Warren, Donald, Corrosion, SBIIA, Vol. 16, March 1960, pp. 119t-130t.
[53] Gooch, T. G., Metal Construction and British Welding Journal, MCBWA,
Vol. 1, No. 12, Dec. 1969, pp. 20-25.
[54] Jesper, H. and Wessling. W.. Werkstoffe und Korrosion, WKKOA, 16 June
1965, pp. 453-465, British Iron and Steel Industry Translations 4433, Iron
and Steel Institute. 1965.
[55] Hamstead, A. C. and Van Delinder. L. S.. Corrosion, SBIIA, Vol. 15, March
1959, pp. 147t-157t.
[56] Medvedev, Yu S. and Tomilina, T. D.. Protection of Metals, PTNMA. Vol.
5, No. 4 July-Aug. 1969, pp. 348-352.
[57] Uhlig, H. H., Transactions. Electrochemical Society. TESOA. Vol. 87, 1945,
pp. 141-154.
[58] Tedmon, C. S., Jr., Vermilyea, D. A., and Broecker, D. E., Corrosion, SBIIA,
Vol. 27, No. 3, March 1971. pp. 104-106.
[59] Newell. H. D., Transactions. American Society for Steel Treating, Vol. 19.
1932, pp. 673-732.
[60] Cihal, V., Hutnicke Listy, HUTLA, Vol. 12, No. 1, 1966, pp. 73-82.
[61] Cihal, v., Protection of Materials. PTNMA. Vol. 2. No. 2. March-April 1966,
pp. 127-133.
[62] Levin, I. A. and Kaloshina, Z. M., Protection of Materials. PTNMA. Vol. 4.
No. 3, May-June 1968, pp. 235-237.
[63] Truman, J. E.. Journal of Applied Chemistry, JACHA, Vol. 4, May 1954,
pp. 273-283.
[64] Coriou, H., Hure, J., and Plante, G., Electrochemica Acta, Vol. 5, 1961. pp.
105-111.
[65] Chaudron, G., EURAEC-976 Quarteriy Report No. 6, Oct./Dec. 1963.
[66] Coriou, H., Desestret, A., Grail, L., and Hochmann, J., Comptes Rendus, Vol.
254, 1962, p. 4467.
[67] Coriou, H., Desestret, A., Grail, L., and Hochmann, J., Revue de Metallurgie,
REMEA. Vol. 61, 1964, p. 177.
[68] Coriou, H., Grail, L., Mahieu, C , and Gay, D., Comptes Rendus des Seances
de I'Academie des Sciences. CRSAA, Paris. Vol. 258. March 1964. pp 2566-
2569.

169] Coriou, H., Desestret, A., Grail, L., and Hochmann, J., Corrosion et Anti-
corrosion, CRACA, Vol. 14, No. 4, 1966, pp. 163-169.
[70] Camus, A., Desestret, A., Froment, M., and Guiraldenq, P., Comptes Rendus
des Seances de I'Academie des Sciences, CRSAA, Paris, Vol. 265, Jan. 1967,
pp. 425-428.
[71] Desestret, A., Epelboin, I., Froment, M., and Guiraldenq, P., Corrosion Sci-
ence, CRRSA, Vol. 8, 1968, pp. 225-234.
[72] Desestret, A., Froment, M., and Guiraldenq, P. in Extended Abstracts, 4th
International Congress on Metallic Corrosion, 7-14 Sept. 1969, Amsterdam,
pp. 122-124. Proceedings to be published by National Association of Corrosion
Engineers, Houston, Tex.
[73] Streicher, M. A., Journal, Electrochemical Society, JESOA. Vol. 103. July
1956, pp. 375-390.
[74] Aust, K. T., Armijo, J. S., and Westbrook, J. H., Transactions. American
Society for Metals, TASEA, Vol. 59, 1966, pp. 544-556.
[75] Aust, K. T., Armijo, J. S., Koch, E. F., and Westbrook, J. H., Transactions,
American Society for Metals, TASEA, Vol. 60, 1967, pp. 360-372.
[76] Aust, K. T., Armijo, J. S., Koch, E. F., and Westbrook. J. H., Transactions,
American Society for Metals, TASEA, Vol. 61, 1968, pp. 270-277.
[77] Aust, K. T., Transactions, Metallurgical Society of American Institute of
Mining, Metallurgical, and Petroleum Engineers, TMSAA, Vol. 245. 1969,
pp. 2117-2126.
[78] Armijo, J. S., Corrosion, SBIIA, Vol. 21, July 1965, pp. 235-244.
[79] Armijo, J. S., Corrosion, SBIIA, Vol. 24, Jan. 1968, pp. 24-30.
[80] Armijo, J. S. and Wilde, B. E., Corrosion Science, CRSSA. Vol. 8, 1968. pp.
649-664.
[5/1 Vermilyea. D. A. and Tedmon. C. S.. Transactions, MTGTBF. Vol. 1. 1970,
p. 1076.
[82] Rhodin, T. N., Corrosion, SBIIA, Vol. 12, 1956, p. 123t.
[83] Mahia, E. M. and Nielsen, N. A., Transactions, Electrochemical Society,
JESOA, Vol. 93, No. 1, 1948, pp. 1-17.
[84] Espy, R. H., "How Composition and Welding Conditions Affect Corrosion
Resistance of Weldments in Type 430 Stainless Steel," preprint of paper pre-
sented at NACE Meeting Cleveland. March 1968. National Association of
Corrosion Engineers. Houston. Tex.
[85] Lincoln. R. A. and Pruger. T. A.. Iron Age. IRAGA. Vol. 171, 26 Feb. 1953,
pp. 127-131.
[56] Herbsleb, G. and Schwenk. W., Werkstoffe iind Korrosion. WKKOA. Vol.
19, No. 2, 1968, pp. 103-113.
[57] Hochmann, J.. Revue de Metallurgie, RVMMA, Vol. 48, 1951, pp. 734-758.
[55] Kiefer, G. C , Engineering Station News, Vol. 22, June 1950, p. 21, Ohio
State University, Columbus, Ohio.
[59] Baumel, A., Archiv fUr das Eisenhiittenwesen, AREIA, Vol. 34, Feb. 1963,
pp. 135-146, British Iron and Steel Industry Translation 3287, Iron and
Steel Institute, London, 1963.
[90] Baerlecken, E., Fisher, W. A., and Lorenz, K., Stdhl und Eisen, STEIA,
Vol. 81, No. 12, 1961, pp. 76S-nS.
[91] Houdremont, E. and Tofaute, W., Stahl und Eisen, STEIA, Vol. 72, No. 10.
1952. pp. 539-545.
[92] Lula. R. A., Lena, A. J.. Kiefer. G. C , Transactions. American Society for
Metals, TASEA, Vol. 46, 1954, pp. 197-230.
[93] Demo, J. J., "Effect of High Temperature Exposure on the Corrosion Re-
sistance and Ductility of AISI 446 Stainless Steel," NACE Preprint No. 23,
1968 Conference, Cleveland, National Association of Corrosion Engineers,
Houston, Tex.
[94] Demo, J. J., Corrosion, SBIIA, Vol. 27. Dec. 1971, pp. 531-544.

[95] Bond, A. P. and Lizlovs, E. A., Journal, Electrochemical Society, JESOA,

Vol. 116, Sept. 1969, pp. 1305-1311.
[96] Bond, A. P., Transactions, Metallurgical Society of American Institute of
Mining, Metallurgical, and Petroleum Engineers, TMSAA, Vol. 245, Oct.
1969, pp. 2127-2134.
[97] Hodges, R. J., Corrosion, SBIIA, Vol. 27, March 1971, pp. 119-127.
[98] Hodges, R. J., Corrosion, SBIIA, Vol. 27, April 1971, pp. 164-167.
[99] Binder, W. O. and Spendelow, H. R., Jr., Transactions, American Society
for Metals, TASEA, Vol. 43, 1951, pp. 759-772.
[100] Lowe, J. H. C , d'A.Hunt, C , and Harker. H. H., Transactions. International
Vacuum Metallurgy Conference, 1969.
[101] Oppenheim, R. and Laddach, H.. DEW Technishe Berichte, Vol. 11. 1971,
pp. 71-77.
[102] Wright, R. N., Welding Journal. WEJUA, Vol. 50. 19 Oct. 1971, pp. 434s-
440s.
[103] Streicher, M. A., "The Effect of Heat Treatment on the Corrosion of 16%
& 25% Cr Stainless Steels," NACE Preprint No. 25, 1968 Conference, Cleve-
land, National Association of Corrosion Engineers, Houston, Tex.
[104] Parker, E. A., Advances in the Technology of Stainless Steels and Related
Alloys, ASTM STP 369. American Society for Testing and Materials. 1965,
pp. 348-355.
[105] Perkins, R. A., discussion to Ref 92, pp. 225-229.
[106] Colombier, L. and Hochmann. J.. Stainless and Heat Resisting Steels, St.
Martins Press, New York, 1968.
[107] Payson, P., Transactions, American Institute of Mining and Metallurgical
Engineers, TAMDA Vol. 100, 1932, pp. 306-328.
[108] Masi, A. and Mancini. F.. Steel International. STINB. March/April 1969, pp.
67-69.
[109] Kadlecek, P. E., Materials Protection and Performance. Vol. 10. June 1971.
pp. 25-30.
[110] Bloom, F. K. and Carruthers, M. E., Evaluation Tests for Stainless Steels,
ASTM STP 93. American Society for Testing and Materials, 1950. pp. 87-
100.
[Ill] Carruthers, M. E., Welding Journal, WEJUA, Vol. 38, 1959, pp. 259s-267s.
[112] Espy, R. H.. Metal Progress, MEPOA, Vol. 86, Sept. 1964, pp. 109-115.
[113] Stenas, G., Proceedings, 5th Scandinavian Corrosion Congress, Copenhagen,
1968, Vol. 1, pp. 13-2 to 13-22, Danish Corrosion Centre, Copenhagen,
Denmark.
[114] Daeman, R. A. and Dept. F., Welding Journal. WEJUA. Vol. 49, 1970. pp.
33s-40s.
[115] Gurry, R. W., Jacob, E. M., and Allen, S. H., Industrial and Engineering
Chemistry Product Research and Development, lEEDB, Vol. 10, No. 1,
1971, pp 112-119.
[116] Levin, I. A. and Kochergina, D. G., Protection of Metals, PTNMA, Vol. 1,
May/June, 1965, pp 225-231.
[117] Vyklicky, M. and Voboril, J., Kovove Materiah. KOMAA. Vol. 3, No. 3,
1965, pp. 232-245.
[118] Babakov, A. A., Fel'dgandler, E. G., Kareva, E. N., and Savkina, L. Y.,
Protection of Metals, PTNMA, Vol. 3, March-April 1967, pp. 117-122.
[119] Kakhovskii, N. 1., Langer, N. A., Yushchenko, K. A., and Chalyuk, G. I.,
Avtomaticheskaia Svavka, AVSVA, No. 12, 1964, pp. 30-37.
[120] Zhadan, T. A., Protection of Metals. PTNMA, Vol. 4, Sept./Oct. 1968, pp.
500-503.
[121] Zhadan, T. A. and Gulyaev, A. P., Sbornik Trudov. Tsentral'nyi Naucho-
issledovatel'skii. Institut Chernoi Metallurgii, STIIA, Moscow, 1968, No. 65,
pp. 34-37.

[122] Zhadan, T. A., Protection of Metals, PTNMA, Vol. 6, Nov.-Dec. 1970, pp.
631-632.
[123] Copson, H. R., Physical Metallurgy of Stress Corrosion Fracture, Rhodin,
T. N., ed., Metallurgical Society Conference, Vol. 4. Interscience, New York
1959, pp. 247-267.
[124] Posysaeva, I. I., Babakov, A. A., and Petrovskaya, V. A., Metal Science and
Heat Treatment, MHTRA, No. 7-8. July-Aug. 1969, pp. 556-559.
[125] Babakov, A. A., Zakharov, Y. V., Levin, F. L., and Golovin, A. I., Protec-
tion of Metals, PTNMA, Vol. 5, No. 3. May/June 1969, pp. 292-294.
[126] Brown, M. H., Corrosion, SBIIA, Vol. 25, Oct. 1969. pp. 438-443.
[127] Gulyaev, A. P. and Tokareva, T. B., Metal Science and Heat Treatment,
MHTRA, Nos. 1 and 2, Jan./Feb. 1971, pp. 117-123.
[128] Levin, F. L., Babakov, A. A., Abramov, A. A., and Zakharov, Y. V., Metal
Science and Heat Treatment, MHTRA, No. 5, May 1970, pp. 373-377.
[729] Raymond, E. L., Corrosion, SBITA, Vol. 24. June 1968, pp. 180-188.
[130] Henthorne, M. and DeBold. T. A.. Corrosion. SBIIA. Vol. 27. June 1971. pp.
255-262.
[131] Copson, H. R., Hopkinson, B. E., and Lang, F. S., Proceedings. American
Society for Testing and Materials, Vol. 61, 1961, pp. 879-889.
[132] Scharfstein, L. R. and Maniar, G. N.. British Corrosion Journal. BCRJA,
Vol. 1, July 1965, pp. 36-41.
[133] Tedmon, C. S., Jr.. and Vermilyea. D. A.. Corrosion. SBIIA. Vol. 27. Sept.
1971, pp. 376-381.
[134] Harris, J. A. and Scarberry. R. C , Journal of Metals. KNZKA, Sept. 1971,
pp. 45-49.
[135] Coriou, H., Grail, L., Mahieu, C . and Pelas. M., Corrosion, SBIIA, Vol.
22, Oct. 1966, pp. 280-287.
[136] Copson, H. R. and Economy, G., Corro.sion. SBIIA. Vol. 24. March 1968,
pp. 55-65.
[137] Economy, G., "Alloys Resistant to Stress-Corrosion Cracking in Leaded High
Purity Water," U.S. Patent 3,573,901, 6 April 1971.
[138] Copson, H. R. and van Rooyen, D., "Nickel-Chromium Alloys Resistant to
Stress-Corrosion Cracking," U.S. Patent 3,574,604, 13 April 1971.
[139] Streicher, M. A., Corrosion, SBIIA, Vol. 19, Aug. 1963, pp. 272t-284t.
[140] Protzl, M., Archiv fur das Eisenhiittenwesen, AREIA, Dec. 1958, pp. 745-756.
[141] Class, I., Grafen, H.. and Scheil. E.. Zeitschrift fiir Metallkunde, ZEMTA.
Vol. 53, May 1962, pp. 283-293.
[142] Grafen, H. in Advances in the Technology of Stainless Steels and Related
Alloys, ASTM STP 369, American Society for Testing and Materials, 1963.
pp. 223-234.
[143] Samans, C. H., Meyer, A. R.. and Tisinai, G. P., Corrosion, SBIIA. Vol. 22,
Dec. 1966, pp. 336-345.
[144] Babakov, A. A., Svistunova, T. V., and Chermenskaya, N. F., Metal Science
and Heat Treatment, MHTRA, Nos. 11-12, Nov./Dec. 1965, pp. 781-785.
[145] Leonard, R. B., Corrosion, SBIIA, Vol. 25, May 1969, pp. 222-228.
[146] Hodge, F. G., Preprint, Paper No. 82 Annual Meeting NACE, Chicago.
March 1971, National Association of Corrosion Engineers, Houston, Tex.
[147] Schiiller, H. J., Schwaab, P., and Albrecht, K., Zeitschrift fiir Metallkunde,
ZEMTA, Vol. 57, April 1966, pp. 312-317.
[148] Svistunova, T. V., Chermenskaya, N. F., Smirnova, A. V., Yakovleva, E. F..
Stegnukhina, L. V., and Zharkova, D. N., Metal Science and Heat Treatment,
MHTRA, Nos. 7-8, July/Aug. 1969, pp. 613-616.
[149] Flint, G. N., Journal, Institute of Metals, JIMEA, Vol. 87. 1958-59, pp. 303-
310.
[ISO] Svistunova, T. V., Doronin, V. M., Kruzhkov, V. I., Topilin, V. V., Dzugutov,
M. Ya., Vinogradov, Yu. V., Chermenskaya, N. F., and Kordonov, B. A.,
Stal, STALE, Aug. 1966, pp. 659-662.

[757] Babakov, A. A., Svistunova, T. V., and Chermenskaya, N. F., Proceedings,

3rd International Congress on Metallic Corrosion, Moscow 1966, available
from National Association of Corrosion Engineers, Houston, Tex., Vol. 2,
pp. 380-389.
[752] Pavlov, S. S. and Svistunova, T. V., Metal Science and Heat Treatment,
MHTRA, No. 11, Nov. 1968, pp. 861-865.
[75i] Svistunova, T. V., Runova, Z. K., and Sakharnov, A. A., Metal Science and
Heat Treatment, MHTRA, No. 5, May 1970, pp. 359-363.
[754] Pavlov, S. S. and Svistunova, T. V., Metal Science and Heat Treatment,
- MHTRA, No. 10, Oct. 1970, pp. 830-832.
[755] Hatfield, W. H., discussion of paper by Pfeil, L. B. and Jones, D. G., Journal,
Iron and Steel Institute, JISIA, Vol. 127, 1933, pp. 380-383.
[756] Rocha, H. J., discussion of paper by Brauns, E. and Pier, G., Stdhl und
Eisen, STEIA, Vol. 75, 1955, p. 586.
[757] Scharfstein, L. R. and Eisenbrown, C. M. in Advances in the Technology of
Stainless Steels and Related Alloys, ASTM STP 369, American Society for
Testing and Materials, 1965, pp. 235-239.
[755] Schwenk, W. in Advances in Technology of Stainless Steels and Related
Alloys, ASTM STP 369, American Society for Testing and Materials, 1965,
p. 239.
[159] Brauns, E. and Schwenk. W.. Stiihl und Eisen, STEIA, Vol. 84, No. 13, June
1964, pp. 787-789.
[160] Herbsleb, G. and Schwenk, W., Corrosion Science, SCRRA, Vol. 7, 1967.
pp. 501-511.
[767] Baumel, A., Sidhl und Eisen, STEIA, Vol. 84, June 1964, pp. 798-807.
[762] Ternes, H. and Schwenk, W., Archiv fiir das Eisenhiittenwesen, AREIA.
Vol. 36, Feb. 1965, pp. 99-108, British Iron and Steel Industry Translation
4456, Iron and Steel Institute, London.
[163] Schwaab, P., Schwenk, W., and Ternes, H., Werkstoffe und Korrosion,
WSKRA, Vol. 16, Oct. 1965, pp. 844-852.
[164] Huey, W. R., Transactions, American Society for Steel Treating, Vol. 18,
1930, pp. 1126-1143.
[765] Brown, M. H., Bulletin, American Society for Testing and Materials, ASTBA,
No. 169, Oct. 1950, pp. 200-209.
[766] DeLong, W. B. in Evaluation Tests for Stainless Steels, ASTM STP 93,
American Society for Testing and Materials, 1949, pp. 211-214.
[767] Alger, J. V., Roberts, E. C , Lent, R. P., Anderton, G. W., Bulletin, American
Society for Testing and Materials, ASTBA, No. 214, May 1956, pp. 57-60.
[76S] Tisinai, G. F. and Samans, C. H., Bulletin, American Society for Testing and
Materials, ASTBA, May 1959, pp. 64.
[769] Hatfield, W. H., Transactions, Electrochemical Society, TESOA, Vol. 46,
1924, pp. 297-312.
[770] Streicher, M. A., Corrosion, SBIIA, Vol. 14, Feb. 1958, pp. 59t-70t.
[777] Streicher, M. A., Bulletin, American Society for Testing and Materials.
ASTBA, No. 229, April 1958, pp. 189-198.
[772] Warren, Donald, Bulletin, American Society for Testing and Materials,
ASTBA, No. 230, May 1958, pp. 199-210.
[173] Streicher, M. A., Bulletin, Aiherican Society for Testing and Materials,
ASTBA, No. 188, Feb. 1953, pp. 35-38.
[174] Brauns, E. and Pier, G., Stdhl und Eisen, STEIA, Vol. 75, No. 9, 1955, pp.
579-586, Henry Brutcher English Translation 3662, Altadena, Calif.
[775] Schiiller, H. J., Schwaab, P., and Schwenk, W., Archiv fiir das Eisenhiitten-
wesen, AREIA, Vol. 33, Dec. 1962, pp. 853-862, British Iron and Steel In-
dustry Translation 3177, Iron and Steel Institute, London, 1963.
[776] Schwenk, W., Schiiller, H. J., and Schwaab, P., Werkstoffe und Korrosion,
WSKRA, Vol. 15, Aug. 1964, p. 621-631.
[777] Herbsleb, G., Werkstoffe und Korrosion, WSKRA, Vol. 19, March 1968. pp.

204-209, British Iron and Steel Industry Translation 6640, Iron and Steel
Institute, London, 1969.
[178] Herbsleb, G., Archiv fur das Eisenhuttenwesen, AREIA, Vol. 41, Sept. 1970,
pp. 875-881.
[179] Gooch, T. G., British Welding Journal, BRWJA, Nov. 1968, pp. 10-22.
[180] Gooch, T. G., Honeycombe, J., and Walker, P.. British Corrosion Journal,
BCRSA, Vol. 6, July 1971, pp. 148-154.
[181] Kossyi, G. G. and Kuzub, V. S., Zavodskaya Laboratoriya, ZVDLA, (English
Translation available. Industrial Laboratory), Vol. 31, May 1965, pp. 582-
584,
[182] Osozawa, K. .and Engell, H. J., Corrosion Science, CRRSA, Vol. 6. 1966,
pp. 389-393.
[183] France, W. D., Jr., and Greene, N. D., Corrosion Science, CRRSA, Vol. 8,
1968, pp. 9-18.
[184] Streicher, M. A., Corrosion Science, CRRSA, Vol. 9, 1969, pp. 53-56.
[185] Streicher, M. A., Corrosion Science, CRRSA, Vol. 11, 1971, pp. 275-276.
[186] France, W. D., Jr., and Greene, N. D., Corrosion Science, CRRSA, Vol. 10,
1970, pp. 379-382.
[187] Linder, B. in Proceedings, 5th Scandinavian Corrosion Congress, 1969, pp.
6.1-6.27.
[188] Rutherford, J. J. B. and Aborn, R. H., Transactions, Iron and Steel Division,
American Institute of Mining & Metallurgical Engineers, TAMDA, Vol. 100,
1932, pp. 293-301.
[189] Prikhod'ko, V. N. and Savitskaya, A. N., "Susceptibility of 18-8 Type Steel
to Intergranular Corrosion Determined by Method of Internal Friction,"
Izvestiia Vuz-Chem. Met., Nov. 1961, pp. 185-192, Henry Brutcher English
Translation 5473, Altadena, Calif., 1962.
[190] Belenkova, M. M. and Mikheev, M. N., Defektoskopiya, (in Russian)
DEFKA, No. 5, 1967, pp. 65-75.
[191] Albritton, O. W., Corrosion, SBIIA, Vol. 24, Dec. 1968, pp. 389-392.
[192] Shadrunova, A. P., "Use of Thermoelectric Method for Investigating the
Processes Taking Place in Austenitic Steels," Izvestiia Vuz-Chem. Met., Oct.
1962, pp. 107-110, Henry Brutcher English Translation 5779, Altadena, Calif.
[193] Henthorne, M., "Corrosion Testing of Weldments," to be published.
[194] Pinnow, K. E. and Moskowitz, A., Welding Journal, WEJUA, June 1970,
pp. 278s-284s.
[195] Samans, C. H., Corrosion, SBIIA, Vol. 20, 1964, pp. 256t-262t.
[196] Armijo, J. S., Low, J. R., Jr. and Wolff, U. E., Nuclear Applications,
NUAPA, Vol. 1, 1965, p. 462.
[197] Pashos, T. J., Duncan, R. N., Williamson, H. E., Arlt, W. H., Baroch, C. J.,
and Hoffman, J. P., Transactions, American Nuclear Society, TANSA, Vol.
7, 1964, p. 416.
[198] Pickett, A. E., "Investigation of Intergranular Cracking of Nonsensitized
Austenitic Stainless Steels in Circulating Autoclaves," EURAEC-GEAP-5495,
U. S. Atomic Energy Commission, Contract AT(04-3)-189, June 1967.
[199] Wanklyn, J. N. and Jones, D., Journal of Nuclear Materials, JNUMA, Vol.
1, 1959, p. 154.
[200] Snowden, P. P., Journal, Iron and Steel Institute, JISIA, Vol. 194, 1960,
pp. 181-189.

Significance of Intergranular Corrosion in High-

Strength Aluminum Alloy Products

REFERENCE: Lifka, B. W. and Sprowls, D. O., "Significance of Inter-

granular Corrosion in Higli-Strengtli Aluminum Alloy Products," Localized
Corrosion—Cause of Metal Failure, ASTM STP 516, American Society
for Testing and Materials, 1972, pp. 120-144.
ABSTRACT: Most heat treatable aluminum alloys are, to some degree,
susceptible to intergranular corrosion, depending upon the metallurgical
treatments that have been given to the metal. Susceptibility to intergranular
attack results from a localized decomposition of solid solution at the grain
boundaries and the formation of identifiable precipitates in the grain
boundaries. Hence, any alloying addition, thermal treatment or mechanical
treatment that influences precipitation of alloying constituents can in-
fluence the susceptibility to intergranular attack. The exact nature of the
localized corrosion path and the specific roles of grain boundary pre-
cipitates and the parent solid solutions in the grain margins depend upon
alloy composition. The actual significance of the intergranular corrosion
tendency with regard to the serviceability of an aluminum alloy product
varies with the alloy and the temper.
Products of alloys such as 2024-T3, 7075-T6, and 7178-T6 become sus-
ceptible to intergranular attack usually as a result of heat treating proce-
dures that are less than the optimum, and in such a condition they may be
susceptible to stress-corrosion cracking or exfoliation corrosion. Such
products can be artificially aged to produce special tempers, 2024-T8,
7075-T76, and 7178-T76 that provide a high resistance to stress-corrosion
cracking and to exfoliation. Although products in these tempers may show
some susceptibility to intergranular attack in an accelerated corrosion test,
such as specified in MIL-H-6088E, and in some natural environments, this
intergranular attack does not indicate a poor service performance.
Hence, it should be understood that predictions of the serviceability of
a high-strength aluminum alloy product can be made on the basis of an
intergranular corrosion test only in certain instances, chiefly for thin sec-
tions of aluminum-copper alloys in the naturally aged temper (T3, T4)
and aluminum-zinc-magnesium-copper alloys in the maximum strength
temper (T6). The most reliable predictions can be obtained from appro-
priate stress-corrosion cracking and exfoliation tests; for some alloys other
criteria, such as electrical conductivity and potential measurements, are
feasible.
KEY WORDS: corrosion, intergranular corrosion, grain boundaries, ex-
foliation corrosion, temper (metallurgy), grain structure, aluminum alloys,
electrochemistry, stress corrosion cracking, pitting
' Senior research engineer and section head, respectively, Chemical Metallurgy
Division, Aluminum Company of America, New Kensington, Pa. 15068.
V20
Copyright by ASTM Int'l (all rights reserved); Fri Jan 1
Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
LIFKA AND SPROWLS ON INTERGRANUIAR CORROSION 121

Intergranular corrosion is attack on a metal that proceeds along grain

boundaries at a considerably faster rate than through adjoining grain
bodies. Most of the high-strength, heat-treatable aluminum alloys are to
some extent, depending upon the metallurgical treatments employed,
susceptible to intergranular corrosion. The mechanism of the corrosion
is electrochemical and can be related to precipitation in the vicinity of
the grain boundaries. The specific electrochemical roles of the grain bound-
ary precipitates and the adjacent solid solutions vary from one alloy
system to another [1-3].^
This paper reviews the causes of susceptibility to intergranular corro-
sion in the 2XXX and the 7XXX aluminum alloys, and places in perspec-
tive the susceptibility of the various tempers with regard to product service-
ability.
Service experience has shown that the principal corrosion problems
with high-strength aluminum alloys in aircraft and areospace vehicles
have resulted from stress-corrosion cracking (SCC) and exfoliation corro-
sion [4-5]. Although these are generally intergranular phenomena, it has
been shown [6-8] that the serviceability of aluminum alloy products can-
not be related in a general way to the susceptibility to intergranular corro-
sion or to any other single characteristic such as the microstructure,
tensile strength, or toughness. Corrosion test data from laboratory and
outdoor exposure tests are used to illustrate the fallacies in trying to
judge corrosion performance of high-strength aluminum alloys by an
accelerated intergranular corrosion test.
Historical Background
The embrittling effect of extensive intergranular attack in thin (0.012-
in.) duralumin (3.5Cu-0.5Mg-0.3Mn) sheet used as covering ma-
terial in one of the early "all metal" airplanes was reported in 1922
by Rawdon et al [9]. Subsequent studies at the National Bureau of
Standards [10] indicated that "the practical means for preventing this
deterioration of duralumin consists, first, in increasing the resistance of
the material toward intercrystalline corrosion by the proper heat treat-
ment of the duralumin, and second, in protecting the surface against
corrosive attack by a suitable coating." Materials problems caused by
exfoliation corrosion, SCC, or corrosion fatigue of the early alloys and
tempers were identified with intergranular corrosion, and the blame
often laid to faulty heat treatment, particularly with the quality of quench
following solution heat treatment. Thus, problems resulting from inter-
granular corrosion of duralumin came to be associated with "improper
heat treatment," and in 1944 an accelerated corrosion test for suscepti-
bility to intergranular corrosion was incorporated into a U. S. Govern-
' The italic numbers in brackets refer to the list of references appended to this
paper.

ment specification [11] for the heat treatment of aluminum alloys. Hence,
it is not surprising that the occurrence of intergranular attack in an
aluminum alloy product today is likely to be a matter of concern among
aerospace engineers.
The Bureau of Standards, around 1925, demonstrated that sprayed
coatings of commercially pure aluminum provided excellent corrosion
protection to duralumin sheet. Such treatments, however, were not con-
sidered practical for large sheets. Subsequently E. H. Dix, Jr., at Alcoa
Research Laboratories established methods to metallurgically clad com-
mercial aluminum to both sides of a 2017-T4 (then known as 17S-T)
sheet to obtain outstanding corrosion protection [72]. In 1928 Dix pub-
Hshed an article [13] describing the improved resistance to corrosion of
Alclad 2017-T4. The great value and versatility of this concept has been
demonstrated by the widespread use of alclad sheet. In many of today's
aircraft designs, however, the advantages of alclad products cannot be
realized because parts are machined frequently from thick plate and
extruded sections.
While a properly controlled solution heat treatment in some instances
may have been an appropriate preventative for the ills of intergranular
corrosion of the thin sheet used in the early aircraft designs, there is no
way to avoid susceptibility to intergranular corrosion by regulating the
heat treatment and quenching of the relatively thick plate, extrusions,
and forgings used in modern aircraft. Therefore, beginning in the 1950s
and continuing to date, special tempers [14-16] were introduced to
assure greater resistance to exfoliation and SCC. With many of the new
exfoliation and stress-corrosion resistant alloys and tempers, the signifi-
cance of a degree of susceptibility to intergranular corrosion is consider-
ably different from that for the early duralumin and superduralumin
alloys.

Al-Cu (2XXX) Type AUoys

Electrochemical Relationships in the Microstructure

Copper is an important alloying addition to give strength to aluminum,
and because of its appreciable solubility in aluminum the aluminum-
copper (Al-Cu) alloys respond to solution heat treatment. Brown et al
[17] showed that the retention of 5 percent copper in solid solution
also changes the electrode potential of aluminum about 200 mV in the
cathodic direction. Thus, the subsequent precipitation of copper rich
constituents from the supersaturated solid solution alters the potential
in the anodic direction, and pronounced electrochemical relationships can
develop in the microstructure. Initial precipitation of discrete particles
occurs in grain boundaries and copper depleted regions develop adjacent

to the boundaries. The copper rich grain boundary precipitates are also
relatively cathodic, and in a corrosive electrolyte such as sodium chloride
solution electrochemical attack occurs in the copper depleted grain bound-
ary regions. Other alloying elements such as magnesium, silicon, manga-
nese, cadmium, zirconium, and vanadium are added to the 2XXX alloys
to provide various mechanical properties. The susceptibility to inter-
granular corrosion of all of the alloys, however, is influenced primarily
by copper concentration gradients in the Al-Cu solid solution in the
grain boundary regions.
Brown and Mears [18] devised a method to demonstrate the electro-
chemical mechanism of intergranular corrosion by measuring the electro-
chemical potentials and electrical current flow between grain boundary
zones and grain bodies of large grained materials. Using this procedure,
the change in potential at the grain boundaries and grain centers of a
high-purity Al-4Cu alloy with time of aging was measured [79]. The data
in Fig. 1 indicate that precipitation occurred more rapidly at the grain
boundaries than within the grains, with the boundaries becoming anodic
to the grain centers. The maximum difference in potential between the
grain boundaries and grain centers, as shown by the lower curves oc-
curred after aging for about 4 to 8 h at 191 C (375 F). With more
extended heating, the precipitation within the grain centers began to
approach that at the grain boundaries until virtually complete precipita-
tion occurred, both at the grain boundaries and within the grains. Thus,
the difference in potential between the two was reduced almost to zero.
It is shown also in Fig. 1 that the precipitation, both in the grains and

STRETCHED
5%

ELECTROLYTE
53g NoCI-fSg H j O ;
'GRAINS PER LITER AT25'C

'GRAIN BOUNDARIES
>|2 -0

AGING TIME-HOURS AT 379 F

FIG. 1—The potentials of the grains and grain boundaries of an aluminum alloy
containing 4 percent copper which was heat treated at 930 F quenched in cold
water and aged at 375 F [19].

the grain boundaries, was accelerated by plastic deformation (5 percent

stretch).
Hunter et al [20], usmg specimens thinned to 1000 A for transmission
electron microscopy, succeeded in relating the path of intergranular at-
tack to microstructural features of aluminum-copper-magnesium (AI-
Cu-Mg) alloy 2024. Figure 2 shows an electron transmission micrograph
illustrating intergranular corrosion of the copper depleted grain margins.
Susceptibility to pitting attack associated with a high degree of precipita-
tion is shown in Fig. 3. This alignment of pits along grain boundaries
would appear as a broad form of intergranular attack at lower (light
microscope) magnifications.

FIG. 2—Electron transmission micrograph of a thin metal specimen of an Al-

Cu-Mg alloy [2024) corroded in NaCl-HiO, solution. The metal had been solution
heat treated, quenched in cold water, and aged 4 h at 191 C (375 F). Inter-
granular attack (shown by the white path) between points B and C followed the
Cu-depleted grain margins visible between B and A. Particles of the relatively
catkodic grain boundary precipitate are visible extending from B (surrounded by
corrosion) to the top of the print. A Widmanstatten pattern of S' precipitate has
formed in the grains. The large dark particles scattered through the grains are
relatively insoluble constituents of Al-Mn and Al-Fe-Si compounds. In this condi-
tion 2024 alloy is highly susceptible to stress<orrosion cracking [20].

FIG. 3—Similar to Fig. 2 except that the metal was aged 16 h at 191 C (375 F).
Shows rounded pits in the grain body and along a grain boundary. Note also the
advanced degree of precipitation in the grain bodies. In this condition 2024 alloy is
highly resistant to stress-corrosion cracking [20].

Metallurgical Treatments Affecting Susceptibility to

Intergranular Corrosion
The principal factors affecting precipitation and susceptibility to inter-
granular corrosion of Al-Cu alloy products are total alloy content, rate
of quench after solution heat treatment, plastic deformation after quench-
ing, and exposure to elevated temperatures.
Quenching Rate—The effect of cooling rate during quenching upon the
resistance to intergranular corrosion and to stress-corrosion cracking has
been studied by various investigators [21-23]. With Al-Cu alloys m the
naturally aged tempers (T3, T4) maximum resistance to corrosion is
obtained by rapid quenching typified by^ prompt immersion in cold water.
However, this procedure can cause excessive distortion of formed parts.

and so at one time, air and fog quenches were employed to minimize
distortion. Such slow quenching produced undesirable susceptibility to
intergranular corrosion and was not permitted by Government Specifica-
tions [77]. However, in the case of artificially aged (T6, T8 tempers)
Al-Cu alloys, rapid quenching is not essential for maximum resistance
to corrosion.
The contrasting effects of cooling rate during quench on the suscepti-
bility to intergranular attack and on the exfoliation and SCC perfor-
mance of 2024 alloy sheet in the naturally and artificially aged tempers
are shown in Fig. 4. The relationship of SCC and exfoliation of the T42
temper sheet to susceptibility to intergranular attack in the accelerated
test in sodium chloride-hydrogen peroxide (NaCl-HaOa) solution is very
striking. Equally striking, however, is the absence of a similar relation-
ship for the T62 temper. Although the T62 temper sheet went through
r 1 11 1 in-\ 1—r-f \ i-rrq— • T — r ~ •r p - m
2 0 2 4 ALLOY SHEET
0T42
• T62 (AGED 5 H R . / 4 0 0 « F )

o \ 0

I 1

u T42 -YES- -NO-

T62 -NO-

»g
4-«- "8- *i B8 s 8

9i I'INTERGRANULAR P-PITTIN8 P+SI "P+SLIGHT 1

T42 +^«IXED
T62 -+ MIXED- -4i-P + S M
' I I I • • • • ! • " ' I • I I I '
10 100 1000 10000

AVERAGE QUENCHING RATE ( 7 5 0 - 5 5 0 * RANGE)«F PER SECOND

FIG. 4—Comparative effects on the corrosion and stress corrosion resistance of

2024-T42 and T62 tempers of sheet as influenced by quenching rate.
'The stress corrosion tests in this experiment were conducted on 0.064-in.-thick
preformed specimens [24] stressed in frames as described by Sager et al [25] and
exposed to 3.5 percent NaCl solution by alternate immersion (60-min cycle with
10-min immersion followed by 50-min drying in air). The stressed preform is a
very severe test in which a high elastic stress is superimposed upon residual stresses
introduced by forming, giving a total sustained stress that is greater than the pro-
portional limit of the material.

a variation of type of attack with cooling rate similar to that of the T42
temper sheet, there was no SCC or exfoliation of any of the items in
the T62 temper. Note in Fig. 5 the similarity in appearance of the inter-
granular attack developed by the NaCl-HjOa test (MIL-H-6088D) in
both the T42 and T62 temper sheet quenched with a cooling rate of 120
F/s. Compare this with the difference in type of attack of the two tempers
when the sheet was highly stressed and exposed to 3.5 percent sodium
chloride (NaCl) solution by alternate immersion. It is significant that
even with a very high stress, but in a less aggressive environment, the
rate of propagation of intergranular attack in the T62 temper sheet was
slow compared to that in the T42 temper sheet.
It is impossible commercially to quench any but relatively thin sections
fast enough to avoid susceptibility to intergranular corrosion. These data,
therefore, indicate that aging the metal to a T62 (or T8) temper im-
proves the resistance to exfoliation and SCC even though susceptibility
to intergranular corrosion is not eliminated.
Artificial Aging—Although solution heat treated and quenched Al-Cu
alloys are hardened at room temperature, this hardening phenomenon
involves the formation of Guinier-Preston (GP) zones in the grain bodies,
with no detectable precipitate formation in the grain boundaries. Artificial
aging by heating the metal at a suitable temperature (usually in the 149
to 204 C (300 to 400 F) range) initiates precipitation in the grain bound-
aries and stimulates extensive precipitation in the grain bodies. As was
discussed previously in connection with Fig. 1, if the metal is aged for
a prolonged time there would be no susceptibility to grain boundary cor-
rosion. However, it has been found unnecessary to extend the aging treat-
ment to that point to achieve practical immunity to SCC and to exfoli-
ation corrosion. In other words, artificially aged products of alloys such
as 2011, 2021, 2024, and 2219 when aged by the usually recommended
practices (Specification MIL-H-6088E, 5 Feb. 1971) will provide maxi-
mum resistance to corrosion and stress corrosion even though they may
be susceptible to intergranular attack. In these instances the differences
in potential between the grain boundary regions and the grain bodies are
relatively small, and thus there is very little driving force for intergranular
attack. Exceptions are 2014-T6 and 2024-T6; the aging practices in
common use for these alloys were selected to produce maximum strength,
and a considerable extension of the aging time and reduction in strength
is required to achieve maximum corrosion resistance.

Artificial Aging to Avoid Corrosion Problems in Natural Environments

Two examples of the use o^ artificially aged tempers of Al-Cu-Mg alloy
2024 to avoid corrosion problems will be discussed next. Rolled plate of
2024-T351 generally is susceptible to intergranular attack in the standard
type of attack test (MIL-H-6088E) and to exfoliation corrosion in out-

^
^

2 "^

0 *^

II
M to
if> < >^
^ -i- ^
g!
ly o •^
ty d to
^ 2 N-
Si! ^ «
m I-
fO CO

door atmospheres, particularly in marine environments. When this plate

is artificially aged to the T851 temper by the recommended treatment of
12 h at 191 C (375 F) it still is susceptible to intergranular attack (Fig.
6), but not to exfoliation corrosion. The photographs in Fig. 7 illustrate
the high resistance of 2024-T851 alloy plate to exfoliation in marine
environments after extended periods of exposure.
In the second example, 2024-T3 alloy drawn tube swaged for a short
distance on the ends stress corrosion cracked during 5-months' exposure
in a seacoast atmosphere, whereas another piece of the same swaged
tube that had been aged to the T8 temper (12 h at 191 C (375 F))
showed no sign of SCC after eight years in the seacoast atmosphere (Fig.
8). Again it was noted that the NaCl-H^O^ corrosion test revealed a
somewhat similar susceptibility to intergranular attack for both tempers,
but the type of attack and the stress corrosion performance in the natural
environment was very different (Fig. 9).
Significance of Susceptibility to Intergranular A ttack
Although only a slight susceptibility to intergranular attack can result
in marked susceptibility to exfoliation and to stress-corrosion cracking
of the as-quenched and naturally aged tempers (T3, T4) of Al-Cu alloys,
it was shown that the same is not true for the artificially aged tempers
(T6, T8). It may be seen from Fig. 1 that if the aging treatment is not
quite at the optimum period there will be a small difference in potential
between grains and grain boundaries. Metal in this condition may be
susceptible to a mixture of relatively wide intergranular and pitting
attack but not necessarily to exfoliation or to SCC. The NaCl-HoO^
accelerated corrosion test tends to exaggerate the intergranular corrosion
of artificially aged products, whereas in other environments the tendency
for pitting is more pronounced. Thus, it may be concluded that the use
of such accelerated intergranular corrosion tests has no significance as
a criterion of the corrosion performance to be expected of artificially
aged products of Al-Cu alloys.

AI-Zn-Mg-Cu (7XXX) Type Alloys

The second family of alloys comprising the high-strength, heat-treatable,
aluminum alloys are the aluminum-zinc-magnesium-copper (Al-Zn-Mg-
Cu) alloys (7XXX series) that provide the highest strength in aluminum.
Alloy 7075 is a typical example and much of the following discussion is
based on experience with this alloy.

Mechanism of Intergranular Corrosion

Some controversy exists over the actual mechanism by which inter-
granular corrosion occurs in the Al-Zn-Mg-Cu alloys [26]. Under some
heat treat conditions, precipitation occurs as closely spaced particles

«
;< J.-
f t.*;

f . • -^ » • 11
•II
•. * I. .
ii
. ••* . • 4-

-.4 • - . 1 • • • '

4^
^ •

' A • • • *-

* -*^«^l«^'-

•.M^-
T3t5»

••* • r 4 # « v , * ' * : ' ' f '

resi

TIOEVWTER ACID SALT SEACOAST

I YEAR SPRAY 2 WEEKS ATMOSPNEHE
2YRtTa«l)
4YR(TB5I)
inJi!.?.J~^''^'''''"t Z'V ^"'^""* °f l-in.-thick 2024 alloy plate exposed as
toZi'3n7eZe!'l^l) ' " " ' " " " '" "^^'"''"'•'"' "^ ""^ ^«^^ '^'"P^'- compared

¥IQ, 9=4wtige(l endi @f Nn.'mMe'dlemelef drmn tuit§ eitpoMi u tht MO-

mm tmm^m m Mnt MUk, M. I, The MUm tube was imt^lblt i§ Inur-
trma^ mmk m^ umu mrmiim mmM wiiMnfivemmtht btmum of the
femM Mnu develepei bf the imgHi, The iMJfi tube, elm iumpiibk to
inmwfmMf atmk, wlihgisei t^^earif estprnm with m fttibk m mkmeoplc
mmmmrim emktm, iGm^m pt^rnu fretn metlmrlHg nm rmeved),

•''!' ••" •:l-t"-^. * * * ' - « * » * * .!.• -i.*

? f • '« '•• ^ * - - • •*..- ;•. !>*. '•»•,', • ••"# • « • "

• * " • " ' . - • , - ' . • ' ' . ' , , • .'

T3-5»«) SCACOAST TS-BYR SEAa>AST

FIG. 9—Metallographic cross sections (Keller's etch) of corroded 2024 alloy
tubes shown in Fig. 8. Contrast the shallow mixture of intergranular and pitting
corrosion present in the T8 temper tube after eight years with the sharp penetrating
intergranular attack in the T3 temper tube after only five months (reduced 50
percent for reproduction).

along grain boundaries. Although the grain boundary precipitates have

been identified as M or M' phase (MgZng), they may contain substantial
percentages of copper. Because the particles are too small to be analyzed,
their electrochemical characteristics cannot be determined accurately. In
a corrosive environment, the most anodic constituent corrodes preferenti-
ally. It was theorized, therefore, that intergranular attack in this family
of alloys could result from selective attack of anodic constituents pre-
cipitated in the boundaries. This was the theory advanced by Golubev [27]
and by Dix [3] and the generally accepted one until about 1961 [28].
As early as 1947, however, Fink and Willey [29] had suggested that
the partial depletion of copper (along with zinc and magnesium) from
the solid solution at the grain boundary region causes the grain margins
to become anodic to both the grain bodies and precipitates in the bound-
aries, much the same as for the Al-Cu system. With the advent of trans-
mission electron microscopy, Hunter et al [30] studied the corrosion in
NaQ-HgOa solution of thin foils of slowly quenched 7075-W sheet, They
observed extremely selective corrosion along the grain margins, leaving
particles of grain boundary precipitate unattacked, as shown in Fig. 10.
This supports the mechanism proposed by Fink and Willey. It is possible,
though, that either mechanism may be operative, depending upon the
chemical composition and thermal history of the particular specimen
involved.
As-Quenched (.W) Temper
The Al-Zn-Mg-Cu alloys in the freshly heat treated (as-quenched)
condition are unstable and spontaneously age at room temperature. This
condition, designated the W temper, is rarely used commercially, except
to facilitate certain forming operations, because of the unstable mechani-
cal properties and because it is also highly susceptible to exfoliation and
s e c . The factors governing resistance to corrosion are much the same
as for the commercial T6 temper discussed next.

Artificially Aged Tempers

T6 Temper—The T6 temper designation is given to materials that have
been artificially aged to their maximum mechanical strength. This aging
is performed at relatively low temperatures of about 100 to 135 C (212
to 275 F). The principal hardening precipitation is a coherent clustering
of solute atoms to form GP zones that have little or no effect on the
resistance to intergranular corrosion in NaCl-HgOg solution (MIL-H-
6088E). With slow quenching, however, small amounts of a transition
precipitate M' form in the grain boundaries, and the metal in either W
or T6 temper becomes susceptible to intergranular corrosion. Conse-
quently, the cooling rate during quenching from the solution heat-treat
temperature is the prime factor as to whether such alloys corrode inter-

FIG. 10—Electron transmission micrograph of a thin metal specimen of Al-Zn-

Mg-Cu-Cr alloy (7075-W) corroded in NaCl-HiO, solution. The bulk metal was
solution heat treated and then quenched in boiling water to cause a susceptibility
to intergranular corrosion. This micrograph shows selective corrosion that has
proceeded along the edges of three adjacent grains. The upper left and the lower
boundaries lie at about 90 deg to the surface and the corrosion has followed a
comparatively narrow path. The upper right boundary, on the other hand, lies at a
relatively low angle to the specimen surface and the corrosion path appears to be
wide. Particles marked P are grain boundary precipitates left intact and uncorroded.
The small dark particles scattered through the grains are the relatively insoluble
chromium rich constituents. In this condition (7075-W) or when artificially aged
to the T6 temper, 7075 alloy sheet is susceptible to strexs-corrosion cracking [30|.

granularly. The general considerations for a fast cooling rate to prevent

intergranular susceptibility are much the same as previously discussed
for Al-Cu alloys, with three exceptions.
The first difference is that the minimum cooling rate to avoid suscepti-
bility to intergranular corrosion is appreciably lower for Al-Zn-Mg-Cu
alloys, of the order of 300 F/s as shown by Fig. 11 [31].
Figure 11 also illustrates the second difference, a reversion to pitting
corrosion for Al-Zn-Mg-Cu alloys at very slow cooling rates. This prob-
ably results from gross overall precipitation during quenching, thereby
minimizing regions of large differences in solution potential and achieving
an improved resistance to SCC and exfoliation. Because of this phenome-

non, plots of corrosion performance of Al-Zn-Mg-Cu alloys versus cooling

rates generally show a region of lowest performance in contrast with
Al-Cu alloys that tend to worsen continually with decreasing quench rates.
The third difference is that cold work introduced between quenching
and artificial aging has negligible effect on both the strength and corrosion
resistance of Al-Zn-Mg-Cu alloys in the T6 temper.
TJX Tempers.—^The T7 temper designation is given to alloys that are
aged beyond their maximum strength to provide a specified degree of
resistance to exfoUation or SCC, or both. For example, material in the
T76 temper is very resistant to exfoliation and provides improved stress
corrosion resistance in the short-transverse direction over that in the
T6 temper; while material in the T73 temper \32\ is essentially immune
to both exfoliation and SCC.

> I 11 I I •!

in in 7 0 7 5 - T 6 SHEET
z o P-PITTING
H a. I-INTERGRANULAR
z 8 30

10,000

AVERAGE QUENCHING RATE ( 7 5 0 - 5 5 0 ' R A N G E ) "F/SECOND

FIG. 11—EQeci of quenching rate on the inherent type of attack and the cor-
rosion resistance of 7075-T6. Unstressed, transverse specimens from 0.064-in. sheet
exposed 3 months to 3.5-percent NaCI alternate immersion.

The T7 aging is a two step process. The first step involves heating at
T6 aging temperatures, 100 to 135 C (212 to 275 F) and often the
actual T6 age is used. The second step is performed commercially at
temperatures between 160 and 185 C (320 and 365 F). A similar aging
effect can be accompUshed with a single high temperature furnace setting,
provided slow (less than 1 F/min) heating rates are used to permit
sufficient low-temperature aging during the heat-up cycle [33].
Aging at temperatures higher than those that produce maximum
strength causes a growth in the zone size and the formation of the metasta-
ble M'-phase, Mg (Al, Cu, Zn), (Figs. 12 and 13). The effect of this
aging process is reflected in the electrochemical potential of the material.
The potential of rapidly quenched 7075-W sheet aged at 121 C (250 F)
shifts about 75 mV in the cathodic direction after 24 to 36 h of aging

FIG. 12—Thin foil of 7075-T6 plate. Minute zones are visible in the grains
(black specks Z) and numerous precipitate particles (P) are present in the grain
boundaries. Dislocations are pinned to chromium rich constituent (C) and to
particles of grain boundary precipitate (P). Short-transverse specimens susceptible
to stress-corrosion cracking at 25 percent yield strength [30].

and remains unchanged thereafter (Fig. 14). Isothermal aging at tempera-

tures of 135 and 149 C (275 and 300 F) produces essentially the same
result but in a shorter time. However, at temperatures around 160 C
(320 F) a second precipitation process is indicated, the potential first
shifting in the cathodic direction and then in the anodic direction after
about 8 h of aging. This change in the anodic direction is more pronounced
on T73 material (Fig. 15) which is aged 24 h at 121 C (250 F) and
then aged for 8 h at 177 C (350 F). The first change of potential with
aging corresponds with precipitation of zinc from sohd solution at the
lower temperature. The second aging step reduces the copper as well as
zinc in solid solution in the grain bodies with the result that the grains

. - -*«"*\.'

i^s"!-:*'*' •^*

FIG. 13—Thin foil of 7075-T73 plate. Compared to the 7075-T6 {Fig. 12) the
zones are larger and the spacing between them greater. Tiny platelets of M'
precipitate (M') also are present. Residual quenching dislocations (D) are still
visible. The density of the grain boundary precipitate has increased, and there is
a very narrow region devoid of zones immediately adjacent to the boundary pre-
cipitate. Short transverse specimens resistant to stress-corrosion cracking at 75
percent yield strength [30].

and grain boundary regions attain similar potentials, thereby providing

improved resistance to intergranular attack.
Corrosion tests of 7075-T6 sheet reheated at temperatures of 160 to
216 C (320 to 420 F) have shown [34] that for each temperature there
is a period of heating that causes high susceptibility to intergranular
attack that is eliminated by continued heating. The length of heating
required to eliminate the intergranular susceptibility varied with the
quenching rate following heat treatment and, to some extent, with the
amount of cold work introduced before aging. As such, freedom from
intergranular susceptibility in the T7 tempers is a function of the amount

-870
n I I I i I r
REFERENCE ELECTRODE-.IN CALOMEL
53gNoCI + 3g HjOj PER LITER, 25»C

S 12 16 2 0 2 4 28 32 36 40 44 48
TOTAL AGING TIME . HOURS

FIG. 14—Extent of isothermal artificial aging as indicated by potential.

of second step aging and adverse effects of slow quench on the T6 per-
formance are overcome by this aging.
Significance of Susceptibility to Intergranular Attack
The mere presence of intergranular attack in 7XXX series alloys is
not a reliable criterion as to whether exfoliation or SCC will occur in
natural environments. Hence, determination of the inherent type of attack
in an accelerated test is of limited usefulness and must be interpreted in
the light of the metallurgical history of the metal.

T I I r T 1 1 1 1 r
REFERENCE ELECTRODE - .IN CALOMEL
53a NoCI+3g H j O j PER LITER,25»C

^^^ 0

24 28 32 36 42 44 48 HR./250'F
0 4 8 12 16 20 24 HR./350'F

TOTAL AGING TIME, HOURS

FIG. 15—Extent of two step artificial aging as indicated by potential.

T6 Temper—Generally, items that are susceptible to intergranular

attack are more prone to exfoliate than corresponding items with suscepti-
bility only to pitting corrosion. Such is not the case, though, for the re-
sistance to sec. For example, in tests similar to those illustrated in Fig.
11, but evaluating both exfoliation and stress corrosion resistance, suscepti-
bility to exfoliation was encountered over the entire range of cooling rates
resulting in intergranular susceptibility, whereas SCC was encountered
only at those quench rates (40^50 F/s) for which reduction in tensile
strength by corrosion was at a maximum. In another investigation, the
resistance to exfoliation and SSC in seacoast atmosphere was evaluated
for 13 production lots of 0.063- and 0.080-in. gage 7075-T6 sheet that
had been tested previously for suceptibility to intergranular attack (MIL-
H-6088E). The atmospheric specimens were located in close proximity
to the type of attack specimens. The SCC test specimen was a transverse,
plastically deformed sheet specimen of the type described previously in
connection with Fig. 4. Although all 13 lots had shown some suscepti-
bility to intergranular corrosion, only 8 lots developed exfoUation during
the 4-year seacoast exposure and only 4 lots incurred SCC.
A more serious drawback to the use of the type of attack method to
predict service performance is the fact that some materials, which are
susceptible only to pitting corrosion when exposed in the absence of
stress, can still be susceptible to SCC, particularly when stressed in the
short-transverse or transverse direction. Table 1, excerpted from an
investigation conducted under government sponsorship [35\, shows that
high resistance to SCC in the transverse direction of 2Vi-in.-diameter
rod was provided only by 7075-T7351 and 2219-T87 even though all
three 7XXX alloys were tested, and both of the 2XXX alloys had shown
pitting corrosion in the type of attack test.
T76 Temper—The interpretation for the T76 temper is somewhat
analogous to the artificially aged Al-Cu alloys. The initial aging at the
elevated temperature induces susceptibility to intergranular attack which
is eliminated by continued aging. However, commercial practices are
T A B L E 1—Laelc oj correlation between type of attaek and SCC performance
2 llS-in.-diameter coldfintshed rolled rod [35].
Type of Attack
(MIL-H-6088D)
SCC Performance
Alloy Surface Interior in Seacoast Atmosphere"
2014-T651 P P failed at 2 5 % YS» (15 ksi)
2218-T87 P + SI P ok a t 7 5 % YS (39 ksi)
70r9-T661 P P failed a t 2 5 % YS (17 ksi)
7178-T651 P P failed a t 2 5 % YS (18 ksi)
7075-T7351 P P ok at 7 5 % YS (44 ksi)
" l/8-in.-diameter transverse tension specimens exposed seven years at Point Comfort,
Tex.
<• YS = yield strength.

T765I

T65I
(a) T7651 specimen showing primarily pitting corrosion with a slight
intergranular tendency.
(b) T7651 specimen showing broad, blunt intergranular corrosion but
with general grain dissolution as well.
(c) T651 specimen showing sharp, penetrating intergranular attack
with no appreciable corrosion of the grains.
FIG. 16—(As Polished, X79). Metallographic cross sections from 0.375-in.-
thick 7178 plate after 1-week exposure to the accelerated exfoliation test, acidified
(pH 3) 5 percent NaCl spray at 49 C {120 F).

aimed at maintaining tlie maximum strength that is consistent with a high

resistance to exfoliation. The amount of aging necessary to produce the
required resistance to exfoliation normally is not enough to eliminate
completely intergranular susceptibility. Thus, both the T76 and T6 tem-
pers may show intergranular attack (Fig. 16), even though they differ
markedly in their resistance to exfoliation (Fig. 17). Intergranular attack
in the T76 products may be either very slight and associated primarily
with pitting corrosion (Fig. 16a), or else it will be relatively broad and
blunt, resulting in general solution of the grains as well as attack along
the boundaries (Fig. I6b). In contrast, intergranular attack in T6 mate-
rial tends to occur solely at the boundary region and is much finer and
penetrating (Fig. 16c). In some cases, however, there is little, if any,
metallographic difference in the intergranular attack of the two tempers.
It does not appear practical, therefore, to set a metallographic standard
limiting the amount of intergranular attack that can be tolerated in the
T76 temper. As such, determination of the inherent type of attack is of
no real significance in predicting serviceability of T76 temper products.
T73 Temper—In the case of the T73 temper, the amount of second
step aging should be sufficient to eliminate all vestiges of intergranular
susceptibility so that only pitting corrosion occurs. Consequently, deter-
mination of the inherent type of attack can be of some assistance in de-

T765I

FIG. 17—T/IO plane panels of 0.500-in.-thick 7075 plate after 1-week exposure
to the acidified 5-percent NaCl spray test required by T76 specifications. The
T7651 specimen typifies the high resistance to exfoliation afforded by this tem-
per (.Xl/2).

ducing whether or not a full T73 condition was achieved. If appreciable

intergranular attack is present, it can be concluded that the material is
not in a full T73 condition; but this does not necessarily imply that such
material is susceptible to exfoliation or SCC. Also the converse is not true,
in that the presence of only pitting corrosion does not guarantee a T73
condition.
Summary
The susceptibility of aerospace heat-treatable aluminum alloys to inter-
granular corrosion and how this relates to exfoliation and stress corrosion
performance in natural environments has been discussed. The intent has
been to show that susceptibility to intergranular attack in an accelerated
corrosion test does not preclude, per se, reliable serviceability. As final
support for this statement, Table 2 shows that the relative rating for
resistance to exfoliation and SCC of the principal aerospace alloys and
tempers does not follow a consistent trend with regard to whether or not
intergranular attack can occur.
Historically, conventional type of attack tests of unstressed specimens
have been used to judge the adequacy of heat treatment for thin sections
of Al-Cu alloys in the naturally aged temper (T3, T4) and of Al-Zn-Mg-
Cu alloys in the maximum strength temper (T6). For these situations
such tests have value. The extension of this criterion to other tempers,
however, is not justified. The limitation of this criterion was noted as
long ago as the year 1949 by Golubev [27]. In the case of thick sections,
maltreatment is not implied necessarily because even the best known heat
treat methods cannot produce rapid enough quench-cooling rates to pre-
vent intergranular susceptibility. With regard to the artificially aged Al-
Cu alloys (T6, T8) and Al-Zn-Mg-Cu alloys in the T76 temper, the aging
practice itself may result in some susceptibility to intergranular attack,
but there is no relationship between this attack and resistance to exfoli-
ation or stress corrosion cracking. The adequacy of such tempers has to
be judged by actual exfoliation or stress corrosion tests or other criteria
which can be calibrated against such tests [36-39].
References
[1] Dix, E. H., Jr., Transactions, American Institute of Mining and Metallurgical
Engineers, TAMMA, Vol. 137, 1940, p. 11.
[2] Dix, E. H., Jr., and Brown, R. H., Metals Handbook, 1948 edition, American
Society for Metals, Metals Park, Ohio, 1948, p. 228.
[3] Dix, E. H., Jr., Transactions, American Society for Metals, TASEA, Vol. 42,
1950, p. 1057.
[4] Malloy, A. M., Naval Aviation News, May 1965, p. 26.
[5] Leak, J. S., Materials Protection and Performance, Vol. 10, No. 1, 1971, p. 17.
[6] Sprowls, D. O. and Brown, R. H., Metal Progress, MEPOA, Vol. 81, No. 4,
1962, p. 79; Vol. 81, No. 5, 1962, p. 77.
[7] Lifka, B. W., Sprowls, D. O., and Kaufman, J. G., Corrosion, SBIIA, Vol. 23,
No. 11, 1967, p. 335.

TABLE 2—Comparisons illustrating poor relationship between corrosion performance

and susceptibility to iniergranular corrosion.
Probable
Susceptibility
sec- Susceptible to Intergranular Corrosion
Alloy Rating'' to Exfoliation (MIL-H-6088E)
6061-T6 A no yes
3094-T3.T4 B» no' no'
S419-T3 B"- no' no'
7075-TB B' no' no'
7075-T73 B no no
'J0H-T8 B no yes
•iO'Jl-TS B no yes
•^094-T7» B no yes
'i024-T8 B no yes
^•eifl-TS B no yes
7ft7S.T7(> C no yes
717.5-T736 C no yes
7178-T76 C no yes
•iOU-Td C yes yes
'J0n-T3 1) yes yes
'iO«4-T3,T4 1) yes yes
•i'ilQ-TS 1) yes yes
7075-T8 1) yes yes
7175-T60 D yes yes
7079-T6 U yi's yes
7178-T6 D yes yes
" Uatiiigs are based on service experience (excluding special chemicals or elevated
temperatures) and on 4-year exposures to seacoast or industrial atmospheres or 84 days
to 3.5 percent NaCl alternate immersion.
A—No known instance of SCC in service or in laboratory tests.
B—No known instance of SCC in service; SCC may occur in extreme laboratory tests
of short transverse specimens.
C—SCC not anticipated in service or in laboratory tests at sustained tension below
typical design stresses or residual stresses resulting from heat treatment, welding, or con-
trolled assembly stresses.
D—Instances of SCC in service unlikely with metal stressed parallel to direction of
grain flow, but occasionally experienced with sustained tension in the transverse or short-
tran.sverse directions.
' Rating only for thin sections which can be heat treated and quenched to achieve a
high cooling rate.

[5] DiRusso, E., Alluminio e Nuova Metallurgia, ANOMA, Vol. 39, No. 3, 1970,
p. 117.
[9] Rawdon, H. S., Krynitsky, A. I., and Berliner, J. F. T., Chemical and Metal-
lurgical Engineering, CMENA, Vol. 26, No. 4, 1922, p. 154.
ilO\ Rawdon, H. S., Industrial and Engineering Chemistry, lECHA, Vol. 19, No.
5, 1927, p. 613.
[11] Army-Navy Aeronautical Specification AN-QQ-H-186a, 10 Feb. 1944, "Heat
Treatment of Aluminum Alloys, Process For," (Superseded by M1L-H-6088E,
5 Feb. 1971.
[12] Dix, E. H., U. S. Patent 1,856,089, 28 June 1928.
[13] Dix, E. H., Aviation, AVTNA, Vol. 25, 1928, p. 26.
[14] Nock, J. A., Jr., Holt, M., and Sprowls, D. O., Metal Progress, MEPOA,
Vol. 80, No, 3, Sept. 1961, p. 87.
[15] Baker W. A., Bryant, A. J., Durham, R. J., and Elkington, R. W., Journal of
the Royal Aeronautical Society, JRAOA, Vol. 70, 1966, p. 757.
[16] Rotsell, W. C. and Long, J. R., Metal Progress, MEPOA, Vol. 91, No. 5, May
1967, p. 106.

117] Brown, R, H., Fink, W. L., and Hunter, M. S., Transactions, American In-
stitute of Mining and Metallurgical Engineers, TAMMA, Vol. 143, 1941 p,
115.
[18] Brown, R. H. and Mears, R. B., Transactions, Electrochemical Society, TESOA,
Vol. 74, 1938, p. 495.
[19] Mears, R. B., Brown, R. H., and Dix, E. H. in Stress-Corrosion Cracking of
Metals, published jointly by American Society for Testing and Materials and
American Institute of Mining and Metallurgical Engineers, 1944, p. 329.
[20] Hunter, M. S., Frank, G. R., and Robinson, D. L. in Proceedings, Second
International Congress on Metallic Corrosion, published by National Associa-
tion of Corrosion Engineers, 1963, p. 102.
[21] Dix, E. H., Jr., in Physical Metallurgy of Aluminum Alloys, American Society
for Metals, Cleveland, 1949, p. 214.
[22] Ketcham, S. J. and Haynie, F. H., Corrosion, SBIIA. Vol. 19, No. 7, 1963,
p. 242t.
[23] Ketcham, S. J., Corrosion Science, CRRSA, Vol. 7, No. 6, 1967, p. 305.
[24] Stress Corrosion Testing, ASTM STP 425, American Society for Testing and
Materials, 1967, p. 7.
[25] Sager, G. F., Brown, R. H., and Mears, R. B. in Stress-Corrosion Cracking of
Metals, published jointly by American Society for Testing and Materials and
American Institute of Mining and Metallurgical Engineers, 1944, p. 255.
[26] Sprowls, D. O. and Brown, R. H. in Proceedings, International Conference on
Fundamental Aspects of Stress-Corrosion Cracking, published by National
Association of Corrosion Engineers, 1969, p. 466.
[27] Golubev, A. I., Journal of Physical Chemistry (Russian), JPCUA, Vol. 23,
No. 9, 1949, p. 1116.
[28] Dix, E. H., Brown, R. H., and Binger, W. W., Metals Handbook, 8th edition.
Vol. I, American Society for Metals, Metals Park, Ohio, 1961, p. 918.
[29] Fink, W. L. and Willey, L. A., Metals Technology. METVA, Vol. 14. No. 8,
1947.
[30] Hunter, M. S. and Robinson, D. L. et al in Proceedings, International Con-
ference on Fundamental Aspects of Stress-Corrosion Cracking, Ohio State
University, published by National Association of Corrosion Engineers, 1968,
p. 495.
[31] Lifka, B. W. and Sprowls, D. O. in Aluminum, Vol. I., K. R. Van Horn, ed.,
American Society for Metals, Metals Park, Ohio, 1967, p. 140.
[32] Sprowls, D. O. and Nock, J. A., Jr., U. S. Patent 3.198.676, 3 Aug. 1965
(Alcoa).
[33] Thompson, D. S., Singleton, O. R., McGowan, R. D., and Spangler, G. E.,
Metal Progress, MEPOA, Vol. 98, No. 3, 1970, p. 82.
[34] Adenis, D., Develay, R., and Guilhaudis, A., Memoires Scienlifique de la Revue
de Metallurgie. MRMTA, Vol. 64, No. 12, 1967, p. 1095.
[35] Lifka, B. W. and Vandenburgh, D. G., Final Report-Government Contract
NAS 8-5340, Control No. 1-4-50-01167-01(If), CPB-02-1215-64, 1966,
Phase I, pp. 3-20.
[36] King, W., Lifka, B. W., and Willey. L. A., Materials Evaluation, MAEVA,
Vol. 23, No. 2, 1965, p. 89.
[37] Horst, R. L., Hollingsworth, E. H.. and King. W., Corrosion, SBIIA, Vol. 25,
No. 5. 1969, p. 199.
[38] Horst, R. L., Jr., and Lifka, B. W., Corrosion, SBIIA, Vol. 26, No. 3, 1970,
p. 111.
[39] Hagemaier, Donald, Nondestructive Testing, NODTA, Vol. 21, No. 5, 1963,
p. 317.

Investigations of Galvanically
Induced Localized Corrosion

REFERENCE: Baboian, Robert, "Investigations of Galvanically Induced

Localized Corrosion," Localized Corrosion—Cause of Metal Failure, ASTM
STP 516, American Society for Testing and Materials, 1972, pp. 145-163.
ABSTRACT: Modern electrochemical theories and techniques are used
to explain galvanically induced localized corrosion of metals. Examples
used are low-carbon steel, aluminum and aluminum alloys, and stainless
steels. The relevance of the galvanic series of metals to predict this be-
havior and the value of critical pitting potentials and critical protection
potentials are discussed.
KEY WORDS: Corrosion, galvanic corrosion, pitting, concentration cell
corrosion, corrosion prevention, polarization (charge separation), anodic
polarization, potential theory, corrosion environments, oxidation reduction
reactions, electrolytes
Galvanic effects occur when two or more metals are in electrical and
electrolyte contact. To predict their behavior in galvanic couples, a gal-
vanic series of metals can be constructed according to their potentials
measured in a specific environment. Usually the corrosion rate of the
more noble metal is decreased, and that of the more active metal is in-
creased in a couple.
In addition to potential differences, polarization behavior of metals is
also important in predicting galvanic effects. For example, metals which
corrode uniformly when electrically isolated may undergo localized corro-
sion when galvanically coupled. Aluminum alloys and stainless steels
which normally have a protective film over a wide range of potentials are
susceptible to pitting and crevice corrosion in sodium chloride (NaCl)
solutions. Galvanic couples of these alloys with other metals can either
inhibit or induce localized corrosion.
This paper describes methods of predicting the occurrence of galvan-
ically induced localized corrosion. Mixed potential theory and polariza-
tion techniques are used on several stainless steels and aluminum alloys.
Couples of these alloys with other metals are evaluated, and supporting
data obtained in long-term experiments are presented.
' Manager, Corrosion Laboratory, Metallurgical Materials Division, Texas In-
struments Inc., Attleboro, Mass. 02703.
145

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 1

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
146 LOCALIZED CORROSION-CAUSE OF METAL FAILURE

Tbeory
Electrode Kinetics
Materials which contain several different metals and alloys are sus-
ceptible to galvanic corrosion. The difference in electrochemical potential
between two or more dissimilar metals in electrical and electrolyte
contact causes electron flow between them. Attack of the more noble
metal or metals is usually decreased, and corrosion of the more active
metal is usually increased. A galvanic series of metals can be constructed
according to their potential measured in a particular environment [7].^
However, the information concerning a galvanic couple obtained from
such a series is of limited value because the series is affected strongly by
environmental factors such as type of electrolyte, temperature, agitation,
etc. In addition, under the influence of galvanic coupling, appreciable
polarization of the metals may occur, producing a protective film on the
metal surface or causing breakdown of an already existing protective
film. This effect is commonly observed with stainless steel. Thus, an over-
all characterization of each metal in the galvanic couple is necessary to
evaluate the behavior of the metals in a particular corrosive environment.
Galvanic corrosion of metals can be treated by application of the
mixed potential theory first described by Wagner and Traud [2]. The
theory is based on two simple hypotheses: (1) any electrochemical re-
action can be divided into two or more oxidation or reduction reactions,
and (2) there can be no net accumulation of electrical charge during an
electrochemical reaction.
Under the simplest circumstance, metallic corrosion would involve only
two reactions—oxidation and reduction. The corrosion of iron in sulfuric
acid (H2SO4) involves the anodic dissolution of iron and the evolution
of hydrogen. This is demonstrated by the polarization curves for iron
in 0.52 N H2SO4 in Fig. 1 [5]. The first hypothesis of the mixed poten-
tial theory is satisfied if one considers that each reaction has its own
reversible potential and polarization parameters. The second hypothesis,
that the total rate of oxidation equals the total rate of reduction, is only
satisfied at the intersection £corr, the corrosion or "mixed" potential. At
this point the rate of iron dissolution is equal to the rate of hydrogen
evolution. The potential is so displaced from the equilibrium potential
that the reverse reactions occur at a negligible rate and do not influence
the corrosion rate.
In Fig. 1, the data indicate that iron will corrode at a rate of about
0.5 mA/cm^ and will exhibit a potential of about —0.52 V versus the
saturated calomel electrode (SCE).

"The italic numbers in brackets refer to the list of references appended to this
paper.

-0.4

-0.5

o
>

2
-0.6

-0.7
10-1 10"

CURRENT DENSITY, ma/cm^

FIG. 1—Polarization behavior of iron in 0.52 N HiSOi.

Galvanic Couples
When two different corroding metals are coupled electrically in the
same electrolyte, both metals are polarized so that each corrodes at a
new rate. In Fig. 2 the corrosion potentials and polarization parameters
for uncoupled Metals A and B are shown. Metal A is more noble than
Metal B in that the equilibrium potential is less negative. When the mixed
potential theory is applied to the individual reactions (A/A+,H2/H+,
B/B+,H2/H+) the uncoupled corrosion rates are /corr.A for Metal A and
4orr,B for Metal B. When equal areas of Metals A and B are coupled,
the resultant mixed potential of the system £corr,AB is at the intersection
where the total oxidation rate equals the total reduction rate. The rate of
oxidation of the individual coupled metals is such that Metal A corrodes

'o,H2{B)
io.HjIA)

TOTAL REDUCTION RATE

O
z

^orr AB

TOTAL OXIDATION RATE

LOG CURRENT DENSITY

FIG. 2—Mixed potential behavior of galvanically coupled metals.

at a reduced rate /'OO„,A and Metal B corrodes at an increased rate

* corr,B'
The information required to predict the corrosion behavior of gal-
vanically coupled Metals A and B is shown in Fig. 2. In addition to the
anodic polarization curves for Metals A and B, it is necessary to measure
the mixed potential of the galvanic couple (Econ.Ao) under actual en-
vironmental conditions because the nature of the cathodic reactions can
have a marked influence on the mixed potential.
Experimental
Polarization Measurements
Polarization measurements were made using the cell similar to that
described previously [4]. Purified argon was passed through a fritted

glass bubbler to continuously deaerate the solution. The potentiodynamic

method (continuous potential scan and current recording) described pre-
viously was used in this work [4-6]. The current from the potentiostat
to the auxiliary electrode is measured as a potential across a precision
resistor which is selected to provide the required logarithmic converter
input voltage. The filtered output from the logarithmic converter is plotted
on the A'-axis, and the working electrode potential versus the saturated
calomel electrode is plotted simultaneously on the Y-axis. In this investi-
gation potentials were scanned at a rate of 0.6 V/h.
All measurements were made in 5 percent NaCl solution. Solutions
were prepared by dissolving reagent grade NaCl in distilled water and
neutralizing with hydrochloric acid or sodium hydroxide. Measurements
were made in deaerated solution at a pH of 7 and 30 C. Disk-shaped
specimens for polarization measurements were punched from commercial
grade sheet material in the mill annealed condition. The specimen surface
was polished with 600-grit silicon carbide paper, followed by ultrasonic
cleaning in detergent solution, and finally rinsed with distilled water. The
specimen was mounted in the type of holder described previously by
Myers et al [7]. In use, the specimen holder exposed only Teflon, and
1 cm^ of specimen to the electrolyte. This design avoids solution contami-
nation and crevice effects, while allowing the use of flat metal specimens.

Mixed Potential Measurements

Mixed potentials for electrically isolated metals and galvanically coupled
metals were measured using a six-channel Orion digital printer system.
Potentials measured versus the SCE were printed hourly over the period
of time required.
The 5 percent NaCl solution was prepared as described previously for
the polarization work. Measurements were made in air saturated solutions
using a constant temperature bath to maintain the solution temperature
at30C±0.1C.
Rectangular specimens of commercial grade metals in the mill-annealed
condition were ground flat and polished through 600-grit silicon carbide
paper. Specimens for galvanic coupling were:
1. Partially immersed in a parallel configuration with a 0.5-in. (1.27-
cm) spacer and connected external to the solution interface so that the
desired area ratio was obtained.
2. Pressure roU bonded to form bimetal or trimetal strip.
3. Drilled and connected with nylon nuts and bolts.
Methods 1 and 2 were used for mixed potential measurements only.
Specimens were masked peripherally at the air-solution interface. Method
3 was used to observe the type of corrosion and weight loss of specimens.
Measurements were made after ultrasonic cleaning in detergent and dis-
tilled water.

Results and Discussion

Frequently, metals which normally corrode uniformly will undergo
severe localized corrosion when galvanically coupled to another metal.
Examples include galvanic pitting or edge corrosion in metallic coatings.
Although this effect normally can be predicted by referring to the gal-
vanic series of metals, more specific information on this behavior can be
obtained by referring to the polarization behavior of the metals involved.
Anodic polarization curves for DHP copper and 1006 low-carbon steel
(LCS) in 5 percent NaCl solution are shown in Fig. 3. The corrosion
potentials of the isolated metals, obtained under actual corroding condi-
tions in aerated 5 percent NaCl solution, are used to determine the rates
of corrosion /corr.cu and /corr.re from the polarization curves in Fig. 3.

O "•" io« 10-2 10" 10 10* 10« 10*

CURRENT DENSITY ( Ua/cm2 )

FIG. 3—Potentiodynamic polarization curves for DHP copper and 1006 LCS in
argon saturated 5 percent NaCl solution at 30 C (0.6 V/h scan rate).

When copper and LCS are coupled galvanically under these conditions,
the mixed potential is E'enrr, and, from the polarization curves in Fig. 3,
the rate of corrosion of copper, i'e»rr,r,i, is decreased and the rate of cor-
rosion of iron, i'mrr.Fe, IS incrcascd.
Increasing the copper/LCS area ratio in the galvanic couple shifts the
mixed potential (£'corr) to less negative values so that galvanic corrosion
of LCS is accelerated.
The corrosion behavior of copper coated 1006 LCS can be used to
illustrate galvanically induced localized corrosion. Normally, LCS cor-
rodes uniformly. However, at pores or damage sites in copper coatings
on LCS and at coating edges, extreme localized corrosion of the LCS
occurs. This behavior is illustrated by the results of total immersion
experiments in 5 percent NaCl.
The photomicrograph of a cross section in Fig. 4a shows galvanic
pitting of LCS at a damage site in the copper coating. Galvanically in-
duced tunneling occurred at the coating edge, as shown in Fig. 4b.
Data obtained from copper coated 1006 LCS total immersion experi-
ments are listed in Table 1. Notice the shift in mixed potential for isolated
1006 LCS and copper-coated 1006 LCS (Cu/LCS area ratio, 100/1)
is 0.14 V. From the Tafel slope (0.06 V) for 1006 LCS in Fig. 3, the

(a) Galvanic pitting at damage site in copper coating,

(fc) Galvanic tunneling at exposed 1006 LCS edge.
FIG. 4—Photomicrographs of cross sections {X50) of Cu/1006 LCS/Cu strip
showing corrosion behavior after total immersion in aerated 5 percent NaCl solu-
tion at 30 C for 62 days.

predicted increase in corrosion rate of LCS is over two orders of magni-

tude. This is in agreement with the measured increase listed in Table 1.
In contrast to LCS, aluminum and aluminum alloys are susceptible to
localized corrosion in NaCl solutions [8]. The problem is most severe
under galvanic conditions due to breakdown of the protective film on the
aluminum surface. This effect is explained by referring to the polarization
behavior of aluminum.
TABLE 1—G<ilvanic corrosion (5 percent NaCl) of 1006 LCS.
Corrosion
Metal Area Mixed Potential, Rate,
System Ratio V versus SCE mpy"
1006 LCS -0.61 a
Cu/1006 LCS 100/1 -0.47 570
" mpy = mils per year
A typical polarization curve for aluminum and aluminum alloys is
shown in Fig. 5. The solid line represents anodic potentiodynamic polari-
zation from active to more noble potentials. The dashed line represents
reversal of polarization durection back to more active potentials.
Below Ec, the critical breakdown potential, the corrosion rate remains
low and unchanged over a wide range of potentials. Above E^ pitting
and crevice corrosion occurs due to the breakdown of the protective film
on the metal surface and is represented by the rapid increase in current

mi
<
s

.=^Ec

tO-> 100 io> 10' 10' 10«

CURRENT DENSITY ( f i a / e i n 2 )

FIG. 5—Schematic potentiodynamic polarization curve for aluminum and alu-

minum alloys in NaCl solution at room temperature.

OAT

o-t-

kl
« -oa
«
>
^-OA

I
^ -0.«

. ^ Ccorr^ Al

corr, Al

-IJ-

10'' 10* 10' 10' 10> 10*

CUIIRENT DENSITY (fla/cin> I
FIG. 6—Polentiodynamic polarization curves for DHP copper and 3003 alu-
minum alloy in argon saturated 5 percent NaCl solution at 30 C (0.6 V/h scan
rate).

density in Fig. 5. Upon reversing the polarization direction, the current

density decreases rapidly at the critical breakdown potential.
The polarization behavior of aluminum and aluminum alloy indicates
that in 5 percent NaQ solution, initiation and propagation of pits and
crevice corrosion occur at potentials above Er, and protection from pit
propagation and crevice corrosion occurs at potentials below Er. The
galvanic behavior of aluminum and aluminum alloys can be explained
best by referring to this polarization information.
In Fig. 6, polentiodynamic polarization curves obtained in 5 percent
NaCI solution for 3003 aluminum alloy and DHP copper are superim-
posed. Mixed potentials for galvanic couples of these metals in various
area ratios are listed in Table 2.
TABLK 2—Mixed potentials for copper/SOOS aluminum
galvanic couples in 5 percent NaCl.
Cu/Al Mixed Potential,
Area Ratio V versus SCE
1/1 -0.755
160/1 -0.736
300/1 -0.704
1000/1 -0.686
2000/1 -0.620
4000/1 -0.668
8000/1 -0.548
2000/1 -0.484

Notice that polarization of 3003 aluminum alloys above Ec (—0.76

V versus the SCE) occurs in the galvanic couples. Extensive localized
corrosion of the aluminum alloy was observed in the series of measure-
ments.
Total immersion experiments in 5 percent NaCl solution were conducted
with copper coated 3003 aluminum strip. The high copper/aluminum al-
loy area ratio at the edge produced galvanic tunneling as observed in the
photomicrograph of a cross section in Fig. 7.
When coupled to stainless steels, galvanically induced localized cor-
rosion of aluminum and aluminum alloys occurs. Specimens of Type 434
stainless steel and 3003 aluminum alloy (area ratio, 2/1) were coupled
and immersed in 5 percent NaCl solution. Photomicrographs of the alu-
minum surfaces are shown in Figs. 8a and b. Although pitting occurs on
the surface of electrically isolated 3003 aluminum, severe pitting is ob-
served in the metal which was coupled galvanically.
Similar experiments were conducted with 3003 aluminum alloy coupled

FIG. 7—Photomicrograph of a cross section (X50) of Cu/3003 Al/Cu strip

showing galvanic tunneling after total immersion in aerated 5 percent NaCl solu-
tion at 30 C for 60 days.

(a) Electrically isolated.

(b) Coupled to Type 434 stainless steel.
FIG. 8—Photomicrographs (XlO) of 3003 aluminum alloy surfaces showing
galvanically accelerated pitting after total immersion for 40 days in aerated 5
percent NaCl solution at 30 C.

to zinc. The mixed potential for this couple (—1.07 V versus the SCE)
is below Ec. Localized corrosion was not observed on the surface of the
aluminum alloy provided that alkaline buildup did not occur.
Galvanic behavior of stainless steels is difficult to predict because of
the passive behavior exhibited by these alloys over a wide range of po-
tentials. In addition, pitting and crevice corrosion due to the localized
breakdown of the passive film occurs in aqueous chloride solutions.
Polarization techniques and critical breakdown potentials, Ec, have
been widely used as a measure of the pitting susceptibility of alloys in
chloride solutions [9-12]. However, Ec values are not useful for predicting

the galvanic behavior of stainless steels. Measurement of critical protec-

tion potentials, Ep, is more valuable for this purpose [12].
Polarization behavior of stainless steel in 5 percent NaCl solution is
shown in Fig. 9. The solid line represents anodic potentiodynamic polariza-
tion from active to noble potentials. The dashed line represents reverse
polarization back to more active potentials.
Polarization above the critical breakdown potential, Er, results in a
marked increase in current density due to initiation of pitting. Reverse
polarization below £„ does not reduce the current density appreciably,

REeioN i n

%
\I
REGION n

RCeiON I

JL _!_ _l_ -L. -J

10-t 10<> 10> 10' 10' 10*
CURIIENT D C N d T Y Ulm/cmi)

FIG. 9—Schematic potentiodynamic polarization curve for stainless steel in

NaCl solution at room temperature.

indicating that propagation of existing pits and crevices corrosion occurs.

At the critical protection potential, Ep, the current density approaches
zero. Below Ep propagation of existing pits and crevice corrosion does
not occur.
The corrosion behavior of stainless steels can be thus separated into
three potential regions: Region I where protection from pit propagation
and crevice corrosion is observed, Region II where propagation of existing
pits and crevice corrosion occurs, and Region III where initiation of
pitting and crevice corrosion occurs. This behavior is observed in oxygen
containing and oxygen free electrolytes, as shown in the potentiodynamic
polarization curves for Type 434 stainless steel in air saturated and argon
saturated 5 percent NaCl solutions in Fig. 10. The curves are similar

except that in the absence of oxygen, the full anodic polarization curve
is observed. Controlled potential data, superimposed in Fig. 10, are in
good agreement with the potentiodynamic polarization curves. At po-
tentials above the critical protection potential, E^, extremely high weight
losses were observed due to crevice corrosion and pitting. Thus, by this
treatment, the galvanic behavior of stainless steel can be predicted in
oxygen free and oxygen containing electrolytes.
The relative resistance to pitting of stainless steels is measured fre-
quently by the magnitude of the critical breakdown potential (_E^). In

AIR

^ -0.4.-

— 0.« v/Hr SCAN

X CONTROLLED POTENTIAL

-t- -+- 4- 4- 4- -I
10' I0»
CURIIENT DENSITY (Aa/CM> I
FIG. 10—Polarization behavior of Type 434 stainless steel in 5 percent NaCl
solution at 30 C (potentiodynamic polarization at 0.6 V/h scan rate; reverse
polarization; X-controlled potential weight loss data).

Fig. 11, polarization curves for Types 409, 430, and 434 stainless steels
are shown. Both the critical breakdown potential (£c) and the critical
protection potential (Ep) become more noble in the order, Type 409,
Type 430, and Type 434. The ranking of the pitting resistance of these
metals is in the same order.
However, these metals behave similarly when galvanically coupled to
copper because their mixed potentials lie in Region II (Fig. 9) where pit
propagation and crevice corrosion occurs.
For example, galvanic pitting occurs at breaks in copper coatings on
Type 409 stainless steel (Fig. 12a), and galvanic tunneling occurs at the

CUmCNT DENSITY Uit/cm' I

FIG. 11—Potentiodynamic polarization curves for Types 409, 430, and 434
stainless steels in air saturated 5 percent NaCl solution at 30 C (0.6 V/h scan
rate).

exposed edges (Fig. I2b). Mixed potentials versus time are shown in
Fig. 13 for DHP copper, Type 409 stainless steel, and the DHP copper/
409 stainless steel couple. Notice that the potential of Type 409 stainless
steel fluctuates over a wide range of potentials. The mixed potential for
the DHP copper/409 stainless steel couple is more stable and after
about 40 h remains more noble than the potential of monolithic Type 409
stainless steel. This potential is within Region II, the pit propagation and
crevice corrosion region.
Galvanic couples of stainless steels with metals which exhibit mixed
potentials within Region I (below £,) yield protection from pit propaga-
tion and crevice corrosion. Mixed potentials for couples of Type 409
stainless steel with lead, 1006 low-carbon steel, and 3003 aluminum alloy
in 5 percent NaCl solution are listed in Table 3. The potentials are below
Ep and within Region I. Complete protection of Type 409 stainless steel
TABLE 3—Mixed potentials of galvanic couples in 5 percent NaCl.
Galvanic Area Mixed Potential,
Couple Ratio V versus SCE
Cu/409 SS 100/1 -0.3
Pb/409 SS 100/1 -0.54
1006 LCS/409 SS 5/1 -0.6
3003 Al/409 SS 1/1 -0.76

(a) Galvanic pitting at damage site in copper coating (x2S0).

(b) Galvanic tunneling at exposed 409 SS edge (X50).
FIG. 12—Photomicrographs of cross sections of Cu/409 SS/Cu strip showing
corrosion behavior after total immersion in aerated 5 percent NaCl solution at 30
C for 190 days.

from pitting and crevice corrosion was observed in these couples. Similar
results were obtained with Types 430 and 434 stainless steels.
The photomicrograph of a cross section in Fig. 14 of lead coated

DHPCapm/40iM
(IMl MK* MTM)

•OS
M 7S
(HOUMI

FIG. 13—Mixed potential behavior of DHP copper and Type 409 stainless steel
in aerated 5 percent NaCl solution at 30 C.

FIG. \A~Photomicrograph of a cross section (XSO) of Pb/Cu/409 SS/Cu/Pb

strip showing galvanic protection of Type 409 stainless steel at exposed edge after
total immersion in aerated S percent NaCl solution at 30 C for 190 days.

Type 409 stainless steel shows that protection from tunneling occurs.
This behavior should be compared to that of copper coated Type 409
stainless steel in Fig. 12b.
The photomicrographs of surfaces in Fig. 15a and b show pitting of
Type 409 stainless steel occurs in 5 percent NaCl solution, while galvanic
protection occurs in Type 409 stainless steel/1006 low-carbon steel
couples.

Condosions
Although the galvanic series of metals can be useful in predicting gal-
vanic behavior of metals in sodium chloride solutions, use of the mixed
potential theory and polarization of metals can lead to more precise in-
formation.
Galvanically induced localized corrosion occurs in chloride solutions
under extreme area ratio conditions or due to breakdown of protective
films.
Galvanic tunneling and pitting in metal coatings such as copper on
1006 LCS occur due to the high copper/1006 low-carbon steel area ratio.
When aluminum and aluminum alloys are coupled to metals such as
copper, stainless steel, and 1006 low-carbon steel, galvanic polarization
above Ec. the critical breakdown potential, occurs. Severe localized cor-
rosion such as pitting and crevice corrosion of aluminum and aluminum
alloys occurs under these conditions in NaCl solutions. Galvanic polari-
zation below Ec yields protection from localized corrosion as long as al-
kaline buildup does not occur.
Galvanic behavior of stainless steels is explained best by referring to
the polarization behavior of these metals. Above £<,, the critical break-
down potential, pit initiation, and crevice corrosion occur. This behavior
occurs when stainless steels are galvanically polarized above E^. Between
Ep, the critical protection potential, and Ee, propagation of existing pits
and crevices occurs. Galvanic couples of copper and stainless steel have
mixed potentials in this region. Below Ep, protection from pitting and
crevice corrosion is observed. Couples of stainless steel with lead, 1006
low-carbon steel, aluminum, and aluminum alloys have mixed potentials
below Ep with the result that the stainless steel is protected from pitting
and crevice corrosion.
Stainless steels are listed in the galvanic series of metals for a particular
environment according to the steady-state potentials. The series is valu-
able in predicting the effect of stainless steels on other metals in galvanic
couples; however, the opposite is not true. Stainless steels should be listed
according to Ep, the critical protection potential, in order to predict ac-
curately galvanic effects of other metals on them. This would also avoid
the confusion resulting from, the placement of stainless steels in two posi-
tions (active and passive) in the galvanic series.

(a) Electrically isolated.

(b) Coupled to 1006 LCS.
FIG. 15—Photomicrographs (x50) of Type 409 stainless steel surfaces showing
corrosion behavior after total immersion in aerated S percent NaCl solution at 30
C for 219 days.

A cknowledgments
The author wishes to acknowledge the assistance of Gardner Haynes,
John Spadaro, and Daniel Jean in this work.
References
[1] Fontana, M. and Greene, N. in Corrosion Engineering, McGraw-Hill, New
York.
12] Wagner, C. and Traud, W., Zeitschrift fUr Electrochemie, ZEELA, Vol. 44,
1938, p. 391.
[i] Makrides, A. C , Journal, Electrochemical Society, JESOA, Vol. 107, 1960,
p. 869.
[4] Greene, N. D., "Experimental Electrode Kinetics," Rensselaer Polytechnic In-
stitute, Troy, N. Y., 1965.
[5] Littlewood, R., Corrosion Science, CRRSA, Vol. 3, 1963, p. 99.
[6] France, W. D. and Lietz, R. W., Corrosion, SBIIA, Vol. 24, 1968, p. 298.
[7] Myers, J. R., Gruewler, F. G., and Smulczenski, L. A., Corrosion, SBIIA, Vol.
24, 1968, p. 352.
[8] Bohni, H. and Uhlig, H. H., Journal, Electrochemical Society, JESOA, Vol.
116, 1969, p. 906.
[9] Brennert, S., Journal, Iron and Steel Institute, JISIA, Vol. 135, 1937, p. 101.
[10] Hospadaruk, V. and Petrocelli, J. V. Journal, Electrochemical Society, JESOA,
Vol. 113, 1966, p. 878.
[;;] Horvath, J. and Uhlig, H. H., Journal, Electrochemical Society, JESOA, Vol.
115, 1968, p. 791.
[12] Pourbaix, M., Klimzack-Mathieiu, L., Mertens, C , Meunier, J., Vanleugen-
haghe, C , DeNuncky, L., Laureys, J., Neelemans, L., and Warzee, M., Cor-
rosion Science, CRRSA, Vol. 3, 1963, p. 239.

Crevice Corrosion of Metals

REFERENCE: France, W. D., Jr., "Crevice Corrosion of Metals," Loca-

lized Corrosion—Cause of Metal Failure, ASTM STP 516, American
Society for Testing and Materials, 1972, pp. 164-200.
ABSTRACT: Shielded metal surfaces are susceptible to crevice corrosion
when exposed to corrosive environments. Crevice corrosion conimoniy
occurs at lap joints, threaded connections, and gasket fittings, in porous
welds, and under debris deposits. This review considers the testing methods,
the mechanisms of attack, and the procedures for controlling crevice
corrosion of metals under a variety of corrosive conditions.
KEY WORDS: corrosion, crevice corrosion, poultice corrosion, concentra-
tion cell corrosion, corrosion tests, corrosion specimens, stress corrosion,
anodic polarization, cathodic polarization, pitting, electrochemistry, in-
hibitors, pH, cathodic protection, seawater corrosion, electrolytes, atmos-
pheric corrosion, polarization

Crevice corrosion^ is a form of localized attack that occurs at shielded

areas on metal surfaces exposed to certain environments. The sites for this
type of corrosion often are unavoidable because of the structural or func-
tional design and may even arise during the exposure period. Examples in-
clude spot-welded lap joints, threaded or riveted connections, gasket
fittings, porous welds, valve seats, coiled or stacked sheet metal, marine
or debris deposits, and the meniscus at a waterline.
The penetration of corrosive solutions into these relatively inaccessible
areas, with widths that are typically a few thousandths of an inch, can
result in various types of failures: the metal surface may become stained
or perforated by the corrosive; the mechanical strength may be reduced
below tolerance limits so that fracture occurs from the applied load or
from the wedging action of the corrosion products; operating components
may seize; or protective coatings may be disbonded from the metal sur-
face. Some examples of crevice corrosion [1-6Y are pictured in Fig. 1 to
show the appearance of several metal surfaces after exposure to various
' Senior research chemist, Research Laboratories, General Motors Corp., Warren,
Mich. 48090.
' Also known as concentration cell, contact, deposit, differential aeration, fissure,
gasket, interface, poultice, waterline, and wedge corrosion.
' The italic numbers in brackets refer to the list of references appended to this
paper.
164

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
FRANCE ON CREVICE CORROSION OF METALS 165

(a) Iron diesel cylinder liner under rubber sealing gasket in contact with
water [1].
(b) Breakdown of titanium above solution level (dashed) under wet salt crust
from 5.3 M NaCl at 80 C [2].
(c) Type 316 stainless steel flange (after removal of rubber asbestos gasket)
exposed to acetic and formic acid solutions at 120 F (49 C) [3].
(d) Type 304 stainless steel bolt and washers exposed for six mon:hs in
quiescent seawater \4].
{e) Stainless steel Thornton nail orthopedic implant after an exposure of four
years in a human body [5].
(/) Titanium pump shaft after service in chlorinated brine solution [6].
FIG. 1—Examples of crevice corrosion.

environments. These and other examples [7-75] indicate that crevice cor-
rosion may be observed in some rather diverse industries, products, and
applications. Undoubtedly, such case studies have led to the conclusions
that crevices are design problems which cause most equipment corrosion
failures [19] and that concentration cell (crevice) corrosion costs industry
more than other types of corrosion [20].
Moody [21] recognized crevice corrosion over 60 years ago on iron
specimens in contact with glass vessels, which he prevented by coating the

ends of the specimens with paraffin. During the ensuing years, this type of
corrosion has been the subject of a significant amount of work. Particular
emphasis has been placed on developing mechanisms of crevice corrosion,
determining factors affecting this localized attaclc, and establishing pro-
tection criteria.
This review considers the testing methods, the theoretical mechanisms
of attack, and the procedures that have been investigated for preventing
or controlling the crevice corrosion of several metals and alloys in a variety
of corrosive environments.
Experimental Procedures
Various crevice corrosion tests ranging from the simple to the complex
have been devised, but the procedures and the specimen designs generally
are unique to the originating laboratory, with only a few examples gaining
widespread application. Consequently, there are virtually no standard prac-
tices specifically for crevice corrosion, although the examination of all
corrosion test specimens for localized attack, especially at specimen holders
or spacers, is generally recommended in ASTM procedures for corrosion
testing. ASTM methods [22,23] for establishing the effects of gasket mate-
rials on metals have been published, and these provide some guidelines
for the development of standard tests for crevice corrosion.

Test Specimens and Assemblies

The development of specimens for the evaluation of crevice corrosion
susceptibility has presented a challenge to the ingenuity of researchers.
Among the necessary considerations are crevice dimensions, shapes, and
relative areas. For example, Ellis and LaQue [24] demonstrated that the
extent of crevice corrosion between overlapping stainless steel panels ex-
posed to seawater for 87 days was proportional to the freely exposed area
outside the crevice. Another factor is the mating surface, whether metal
or nonmetal. Crevices formed with dissimilar metals will involve the added
factor of galvanic corrosion [25]; those formed with nonmetal mating
surfaces, such as plastics, may exhibit variable corrosion behavior that is
related to ions leached from the material [26] or to the degree of material
stiffness [27].
Various procedures have been used to obtain crevice specimens for ex-
posure to corrosive environments with the objective of simulating actual
components in service. Some typical sample configurations with metal/
metal or metal/nonmetal combinations include sand or debris piled on
sheet metal [28,29], rubber bands wrapped around pieces of metal [291,
spot-welded lap joints [28], specimens partially immersed in solution
[2,30], rubber grommets in sheet metal [31], stacked coupons with and
without spacers [32,33], bolted or riveted joints [3,8,34,35], wire wrapped
on a threaded bolt [25,36], and a hairpin-shaped tapered sheet [37].

i^ C. d
(a) Stacked metal and nonmetal coupons with rubber band straps.
(b) Sheet metal with gasketed bolt and rivet.
(c) Metal wire wrapped on nonmetal bolt.
(d) Hairpin tapered crevice in sheet metal.
FIG. 2—Specimens for crevice corrosion tests.

These last four examples are illustrated in Fig. 2. In each case the extent
of corrosion is determined by interrupting the test for visual examination
and the measurement of specimen weight loss.
The development of artificial crevice assemblies has been of particular
importance for the accurate determination of crevice corrosion potentials
and currents. Three recent designs are shown in Fig. 3 to illustrate dif-
ferent procedures for accurate control of crevice geometry and dimensions
during electrochemical corrosion tests [38-40]. With the Rosenfeld-
Marshakov device (a) the specimen is clamped in the plastic frame after
a fixed crevice width is regulated by calibrated spacers;* the Lizlovs cell
(b) allows reproducible changes in the crevice geometry and dimensions
during the course of an experiment; and the France-Greene assembly
(c) permits electrode potential measurements along the length of a crev-
ice. Some applications of these assemblies for the interpretation and evalu-
ation of crevice corrosion are discussed subsequently.
Evaluation of Crevice Corrosion Susceptibility
The evaluation of crevice corrosion susceptibility is often accomplished
with weight loss measurements and relative rankings based on visual
examination. Such procedures can require lengthy exposure times for the
localized attack to develop unless the initiation and propagation processes
are accelerated or evaluated in another way, as with electrochemical mea-
surements. For example, Pourbaix [41] has characterized seven types of
cyclic anodic polarization curves from which predictions can be made
* Rosenfeld and Marshakov also have described a unique sectional electrode for
the determination of current distribution within a crevice [38].

o
>.

o
•o
c
CO
u E
x> _o
t-i ^ •5 0
Ui
0. "a
0

S * ^^•a
V
.2
c 13
O o u 3
c u
0 13
0. CO
H « <s t<i •* >/^

60 O
c " I
2.
CO
-a
_ f It •S «

^-Lt v>
U S
•= 2

i
en
•?

^ a Js
e\j
I
\
^^«#i^\-^>^ <a
1^ o> « 73

regarding the possibility of general, pitting, and crevice corrosion. Dif-

ferentiation is made between curves that indicate no occurrence of crevice
corrosion and those that indicate crevice corrosion susceptibility under
conditions of differential aeration.
One of the more significant advances in the development of evaluation
procedures has been made by Wilde and Williams [31]. These authors
conducted anodic polarization measurements with several alloys that had
been previously evaluated in seawater for more than four years. Cyclic
potentiodynamic anodic polarization curves for metals containing syn-
thetic crevices exhibited two critical potentials—the pitting potential (Ec)
and the protection or repassivation potential (E^). The experimental pro-
cedure involved a potential scan in the noble direction from the corrosion
potential to a point beyond the pitting potential. When the anodic current
density reached 2000 fiA/cm', the scanning direction was reversed, and
the potential at which the downward scan intercepted the upward scan was
designated the protection potential.""' Three of their curves are reproduced
in Fig. 4 to illustrate some possible variations. The curve for Hastelloy C,
which is not susceptible to crevice corrosion, exhibits reversibility,
whereas the curves for Incoloy 825 and Carpenter 20 Cb-3 exhibit slight
and extensive hysteresis loops, respectively. Comparison of these data with
seawater exposures indicated that the susceptibility to crevice corrosion
is related to the presence of hysteresis during cyclic polarization. They
also found that the difference between the pitting and protection potentials
(difference potential) could be correlated directly with the extent of crev-
ice corrosion after more than four years in seawater. This is shown by the
relationship in Fig. 5 for several stainless steel alloys where pitting (rather
than a synthetic crevice) was the precursor to crevice corrosion. Such a
test that requires hours rather than years of service to provide informa-
tion on the susceptibility and extent of crevice corrosion certainly deserves
further consideration and interlaboratory testing in other environments.

In Situ Detection of Crevice Corrosion

The ability to detect and monitor the progress of crevice corrosion
without disturbing the system is hindered by the fact that the evidence of
attack usually is restricted to relatively inaccessible areas. Various ap-
proaches for circumventing this problem have been demonstrated, such as
the visual examination of crevice corrosion in the progress through a trans-
parent glass or plastic plate [43,44] that is held against a metal surface to

° This definition of the protection potential is different than that proposed by

Pourbaix, who defines the protection potential, below which corrosion ceases, as
the potential where the current density meets the ordinate axis (that is, a practically
zero current density value on a linear scale) during the downward scan of a cyclic
polarization experiment [42],

I ••- • 1 1

HASTELLOY C
1.4 3.S r>/0 NoCI, Z S C
AERATED

1.2 -

1.0 -

0.8 - -

0.6 -

0,4 - / -

0? ' 1 a
K) 100 1000

APPLIED CURRENT DENSITY, /tA/cm«

1 1 1

1.2 -

1.0 - -

08 -
''~'i
06 -
INCOLOY 825

0.4 i.i «/D NoCI, 2»C -

AERATED

1 1 1 t)
1000

APPLIED CURRENT DENSITY ,/lA/cm<

form the crevice.' A similar visual procedure in conjunction with electro-

chemical measurements has been used by Mueller [45] whose hemispherical
test electrodes are placed in contact with hemispherical grooves in a plexi-
glass plate.
For quantitative measurements of localized corrosion, Jones and Greene
• An interesting technique for demonstrating the detection of differential aeration
at a corrosive drop on a metal surface involves the incorporation of indicators into
the drop to differentiate between anodes and cathodes within this unique crevice [291.

I 1 1
CARPENTER 2 0 C b 3 ^ •-^
3.5 w/o NoCI , 25C »— J
1.0 AERATEn
—

-
_J

<
5 0,6
•
- —
o
Q.
uj 04
O
o
a:
(J
^ 0.2 - -
UJ

0 —

-0.2 1 1 ^
o
10 100 1000 10,000
APPLIED CURRENT DENSITY.^A/cm^

(a) Hastelloy C.
(b) Incoloy 825.
(c) Carpenter 20Cb-3.
FIG. 4—Cyclic potentiodynamic anodic polarization curves for metals contain-
ing synthetic crevices [31].

SCO

Ej DETERMINED AT 0 600 V/hour

E DETERMINED AFTER REVERSINI
3.5 w/o SODIUM CHLORIDE, AERATED, 25C
SAMPLE AREA,-I220cm'
CREVICE AREA,«APPR0X 2 0 c m '
(410)

lUSS 100)

HASTELLOY C

0 10 10
WEIGHT LOSS, mg / c m '

FIG. 5—Correlation between difference potential and corrosion weight loss of

stainless steel alloys exposed in seawater for 4.25 years [31].

[46] have demonstrated an electrochemical technique suitable for detecting

crevice attack before any visible evidence could be observed. Corrosion
potentials and corrosion rates determined by linear polarization were found
to provide the necessary data. The combination of a marked increase in
corrosion current, a significant (30 to 50 mV) shift in the corrosion poten-
tial to active values, and the initial occurrence of erratic corrosion poten-
tial fluctuations revealed the existence of localized corrosion. The spon-
taneous initiation of crevice corrosion resulted in as much as a hundredfold
increase in the measured overall corrosion currents.
For both laboratory and process control applications, the development
of experimental resistance probes for continuous detection and measure-
ment of crevice corrosion is particularly interesting [47]. These probes
were made from 6061 aluminum alloy wire fitted with Teflon disks to
form a crevice region. Experience had shown that this combination of
materials was susceptible to severe crevice corrosion. With this system the
resistance of the aluminum electrical conductor increases as the cross
sectional area is decreased by corrosion. Thus, the resistance is propor-
tional to the corrosion rate, and this special probe exhibited a dramatic
increase in the resistance reading when crevice corrosion was initiated.

Crevice Corrosion Mechanisms

Significant contributions to the understanding of crevice corrosion began
some 50 years ago with research by McKay [48] on metal-ion concentra-
tion cells and by Evans [44] on differential aeration. Through the years,
a number of mechanisms have been proposed, modified, invalidated, re-
jected, or accepted, often with exceptions." There is little doubt that corro-
sion in crevices is a complex process that involves several competitive
and synergistic factors, which have been investigated and discussed in
some detail by several researchers, particularly Evans [49], Rosenfeld
and Marshakov [38], and their colleagues.* Both the classical and modern
experimental work are contributing to the development of some gener-
alized, and possible unifying, concepts of mechanisms that deserve con-
tinued consideration and refinement.

Concentration Cells
Much of the theoretical interpretation of crevice corrosion has been
associated with some sort of concentration cell. This concept is based
on the fact that the initiation and propagation of localized attack on metal
surfaces is influenced by nonuniformity of the corrosive solution. For
example, variations in the concentration of ions or dissolved gases may
' There are some other mechanisms that are primarily of historical interest
(that is, the hydrogen bubble egress and rival interface theories) [43,50,51].
'Additional publications are cited in these two references.

arise along the same metal surface because of the existence of crevices
that are relatively inaccessible to fresh electrolyte compared with freely
exposed surfaces. Such evironmental conditions cause potential differences
and influence electrode kinetics to the extent that an electrochemical cell
is established, wherein an oxidation process (that is, corrosion) occurs
at the anodic sites, and some reduction process (for example, oxygen re-
duction) occurs at the cathodic sites.
A description and critique of several concentration cell mechanisms
that have contributed to the understanding of crevice corrosion are dis-
cussed in the following paragraphs.
Metal-Ion Cell—This mechanism [48] is based on the development of
differences in metal-ion concentration between the inside and outside of
a crevice. Laboratory experiments with an electrochemical cell consisting
of two short-circuited copper electrodes, each immersed in a different con-
centration of copper sulfate solution in electrical contact through a porous
barrier, have shown that corrosion occurs predominantly in the solution
with the low metal-ion concentration. From the Nernst equation, the electro-
motive force (emf) of such a divided cell with a 0.0001 normal solution in
the anode chamber and a 1.0 normal solution in the cathode chamber is
0.13 V. By anology, such a cell can be formed in practice after the metal
ions formed by the initial corrosion reactions accumulate within the
crevice where solution flow or diffusion is restricted. Once a difference in
metal-ion concentration is established, accelerated dissolution (that is,
M —> M+ + e ' ) is anticipated at anodes just outside the crevice, whereas
the reduction of metal ions (that is, M+ + e~ -^ M) is anticipated at
the cathodes within the crevice. A schematic representation of the operation
of this crevice cell is illustrated in Fig. 6a.
Differential Aeration Cell—This mechanism [44,49] depends on the
occurrence of different dissolved oxygen concentrations in the solution
adjacent to a metal surface." Electrochemical cell experiments, where one
half-cell consists of a metal in a deaerated solution and the other half-cell
consists of a metal in an aerated solution, indicate that accelerated dis-
solution occurs at low oxygen concentrations while the reduction of oxygen
occurs where the O2 levels are high. This type of cell is illustrated in
Fig. 6b. In a crevice situation the oxygen is replenished more easily on the
exterior surfaces than within the shielded crevice, where corrosion is ac-
celerated as the result of oxygen depletion.
Critique—Schafer and Foster [32] have considered the metal-ion con-
cept invalid because it depends on the M+ concentration at the anode
source being lower than elsewhere on the metal surface and because the
reaction is not self sustaining (that is, continued operation of the cell even-

" This discussion of differential aeration is limited to crevice corrosion, but there
are more extensive applications of this general mechanism [49].

Low M Concenh-Qfion

High O. Concentration Low O. Concentration

-• 4 0H

[e e e e (i5t^L.U^ 1

(a) Metal-ion cell.

(6) Differential aeration.
FIG. 6—Schematic representation of crevice corrosion mechanisms.

tually should eliminate the metal-ion concentration differences, and the

reaction would cease). These authors suggest that the actual cathodic re-
action is oxygen reduction rather than metal-ion reduction and that the
corrosion observed at the mouth of a crevice is simply a special case of
the differential aeration cell. They support this viewpoint with the fact that
corrosion at the crevice opening is stopped by removal of dissolved oxygen.
In a related example Evans [49] has shown that although iron and
zinc demonstrate the differential aeration mechanism, copper does not.
When a copper electrode in an aerated solution is coupled with one in a
deaerated solution, corrosion occurs in the aerated environment. Evans has
attributed this reverse effect to the operation of a metal-ion cell. The
bubbling action during aeration stirs the solution enough to remove the
cuprous corrosion products from the freely exposed surface where the
attack is then accelerated.
The reconciliation of these observations depends on some flexibility
in the interpretation of the metal-ion mechanism. Although the metal-ion

cell cannot be self sustaining without some additional cathodic process,

the data on copper are difficult to explain without this concept. It appears
probable that initially metal ions concentrate within the crevice during
corrosion which is supported by oxygen reduction. Subsequently, the con-
centration of M+ and the combination of M+ and O2 reduction contrib-
ute to the location of corrosion just outside the crevice.
Tomashov [53] has questioned the validity of the differential aeration
mechanism under conditions such as metal in chloride environments and
suggested that conclusions based on the external current of a laboratory
macrocell can be misleading. This is because the aerated specimen (so-
called cathode) of the electrochemical cell sometimes exhibits a greater
weight loss than the deaerated specimen to which it is coupled. He ex-
plains that the activity of microcells on the cathode is enhanced by aera-
tion. Although Evans [49] admitted that several investigators have en-
countered this apparently contradictory situation, he notes that this is an
experimental artifact. It is traceable to the high resistance of the electro-
chemical cell or circuitry, which can accentuate the action of local anodes
on the macrocathode, thereby causing a greater weight loss at the aerated
specimen.
Another observation has been that the metal-ion and the differential
aeration concentration cells are competitive [54]. A low oxygen concentra-
tion favors accelerated attack which generates a high metal-ion concentra-
tion in stagnant areas, thereby stifling further anodic reaction. If these two
mechanisms were the only ones operating, the final location of attack would
depend upon which one became dominant. However, some of the other
factors that must be considered are discussed in the following paragraphs.
Active-Passive Cell—This cell is sometimes considered to be another
example of differential aeration, but it deserves special attention; the de-
pletion of any oxidizer in a crevice can initiate localized attack on active-
passive metals because the establishment and maintenance of passivity
depends upon the availability of sufficient oxidizer concentrations. There-
fore, a deficiency of oxygen can prevent the reformation of protective
oxide films at damaged areas on a metal surface, which become active
and corrode at an accelerated rate. Other oxidizers (for example, Cu+- and
Fe+"^) can have similar effects, as indicated by the depletion of ferric
ions in a stainless steel crevice. After 9 h in a hot sulfuric acid solution
which contained 10 g/liter of ferric sulfate, the stainless steel crevice poten-
tial changed from a passive value of 0.53 V versus saturated calomel
electrode (SCE), indicative of a 0.3 mils per year (mpy) corrosion
rate, to an active value of —0.36 V versus SCE, indicative of 6000-mpy
corrosion rate [55].
Rosenfeld and Marshakov [38] have measured similar potential changes
with time for several metals in a neutral electrolyte and have studied the
changes in electrolyte composition and characteristics during crevice corro-

sion. With adjustable crevice assemblies (Fig. 3a) these investigators

determined anodic and cathodic polarization curves which demonstrated
the acceleration of anodic corrosion reactions and the retardation of
cathodic reactions within crevices. These data indicated that the extent of
anodic and cathodic control determines whether corrosion is accelerated
inside or outside the crevice. As shown schematically in Fig. la, with the
reduced effectiveness of the cathodic process, the negative shift in
electrode potentials should tend to reduce corrosion at active potentials
{A) and either increase (fi) or unaffect (C) corrosion at passive potentials,
depending on the proximity of the reduction curves to the anodic current
density maximum.
Critique—Such a graphical presentation facilitates the understanding
of crevice corrosion, especially for active-passive cells, but the conclusion
related to corrosion at active potentials may be misleading if the anodic
interior and cathodic exterior areas of the crevice are considered a corro-
sion couple. For example, with the oxidation-reduction reactions shown
in Fig. 7ft,'" the corrosion current inside the crevice (/'i„) is greater than
that outside the crevice (/'„ut) at the corrosion potential of the couple
(E' couple (in/out)), and crevice corrosion would be predicted. However, it
is evident that in the uncoupled state, the exterior is corroding at a faster
rate (/out) than the interior (An) of the crevice. Thus, some flexibility is
important in the theoretical interpretation of polarization curves for crevice
corrosion.
Other Concentration Cells—In addition to the concentration cells dis-
cussed above, Myers and Obrecht [72] have suggested that there are at
least three additional types, namely, hydrogen-ion, neutral salt, and in-
hibitor cells. For example, high chloride ion concentrations or low in-
hibitor concentrations generally are associated with anodic rather than
cathodic regions on a metal surface. Furthermore, Rosenfeld and Mar-
shakov [38] showed that acidification (low pH) of the solution within the
crevice was a major contribution to accelerated and autocatalytic crevice
corrosion, especially in near-neutral solutions.*' Subsequently, Korovin
and Ulanovskii [58] found that pH changes had a far greater influence on
corrosion rates and potentials than did changes in oxygen concentration.

^''Other publications [56,57] provide details on the manipulation and theoretical

analysis of such polarization curves for various combinations of exchange current
densities, reversible potentials, corrosion potentials, Tafel slopes, and limiting diffu-
sion currents.
" Acidification is the usual, but apparently not the only pH change that can
occur within crevices. Rosenfeld and Marshakov also discuss "linear-selective dis-
solution" and "differential concentration cells" in which alkalization is responsible
for accelerated corrosion. Their results with a divided cell and three phase bound-
aries (for example, iron/dielectric/acid) indicated that the corrosion at the anode
along the dielectric perimeter was associated with a pH increase due to acid deple-
tion. Thus, depending on the bulk electrolyte pH, the anode electrolyte may be
of higher or lower pH than the bulk solution.

LOG CURRENT DENSITY

N^'o. \j>

'1 v /' '^ / Z'^''

O
1 / / / •
1 y'^ / ^
1 /^^ / ^
-E Y /^
coir,' OufI 1 /!!'
out _/'^ ir' 1 / > • - T O T A L OXIDATION
J//'' CURRENT
-E' Z*^ v ^r
coup e lin. outi 1

. # ^ y^' I

—E
S^^.^ 1

1
I
1 *"" TOTAL REDUCTION
1 CURRENT

y^^ .»
b
LOG CURRErjT

(a) Active-passive metal dissolution \.3S\.

(h) Crevice corrosion couples at active potentials.
FIG. 7—Schematic anodic and cathodic polarization curves for crevice corro-
sion analysis.

For example, with a combined reduction in oxygen concentration (9 to

1 mg/liter) and change in pH (8.3 to 2.7), the potential shift in the
active direction was 372 mV, of which 355 mV was associated with the
pH decrease and 17 mV with the reduction in oxygen concentration. For a
bulk pH of 6 to 7 in halide solutions at 150 C, Griess [34] found that the
pH within a titanium crevice where corrosion occurred was about 1,
whereas the pH was 4 to 5 when significant corrosion was absent.
Besides acidification effects, Griess showed that the severity of titanium
crevice corrosion increased with the solution salt concentration. Although
his experiments indicated that the anion was relatively unimportant, other
investigators have demonstrated the significance of anions. For example,
Strocchi et al [59] examined the crevice corrosion behavior of several
activated and passivated austenitic stainless steels (Types 302, 305, 316,
and 347) in contact with nonconducting materials (for example nitrile
rubber O-rings) during immersion in acid-chloride solutions. These steels
exhibited different susceptibiUty to crevice corrosion, but a common feature
was that oxygen was necessary for the initiation of crevice corrosion at
three phase contacts (metal/nonmetal/solution) and that aggressive anions
(for example, chloride) contributed significantly to the breakdown of
passivity and corrosion.

Unified Crevice Corrosion Mechanism

Based on the type of information developed in the preceding sections,
Fontana and Greene [57] have proposed a basic mechanism of crevice
corrosion. Initially the anodic dissolution (that is, M -» M+ + e~) and
cathodic reduction (for example, O2 + 2H2O + 4e- -> 4 0 H - ) processes
occur uniformly over the entire metal surface, including the crevice in-
terior. The oxygen in the shielded crevice area is consumed after some in-
cubation period, but the decrease in the cathodic reaction rate is negligible
because of the small area involved. Consequentiy, the corrosion of the
metal inside and outside the crevice continues at the same rate.*- With
the cessation of the cathodic hydroxide-producing reaction, however, the
migration of mobile negative ions (for example, chlorides) into the crevice
area is required to maintain charge balance. These processes are shown
schematically in Fig. 8. The resulting metal chlorides hydrolyze in water
to insoluble metal hydroxides and free acid. Both the chloride anions and
low pH accelerate crevice corrosion in a manner similar to autocatalytic
pitting, while the reduction reaction cathodically protects the exterior sur-
faces. Also, for active-passive metals that depend on protective oxide
films for their corrosion resistance, the breakdown of passivity and active

" Equal corrosion rates inside and outside the crevice would be anticipated from
Fig. 76 if the metal oxidation reactions were considered to be represented by a
single line.

'ci;

High O j Concentration
enfratlon " ^ ' ^ \ ^ L o w O^ ond High M " , C l " , H " Concentrations
»C

°2 • 4 0H Cl" H' CI" M* M* M' M' CI" H" CI"

FIG. 8—Fontana-Greene mechanism of crevice corrosion [57].

corrosion in crevices are favored by the increased concentration of chlo-

rides and hydrogen ions.
This mechanism represents a rather inclusive and unifying description
of crevice corrosion. It takes into account metal-ion concentration, oxygen
depletion, hydrolysis and acidification, aggressive anion migration, and
both active and active-passive dissolution behavior. Undoubtedly, there
are exceptions to the strict interpretation of this mechanism, for example
the observed occurrence of crevice corrosion when aggressive anions are
absent from the solution. However, this mechanism, which might be ap-
propriately named "differential concentration cell",^^ represents a signifi-
cant contribution and starting point for specific refinements.
Crevice-Related Localized Corrosion Plienomena
The phenomenological and mechanistic relationship between various
forms of localized attack have been under increased scrutiny lately. With
an emphasis on similarities, Brown [60] has proposed an occluded corro-
sion cell concept in which local acidification by hydrolysis is a common
feature of localized attack, such as crevice corrosion, filiform corrosion,
pitting, stress-corrosion cracking, intergranular attack, exfoliation, and
tuberculation. Such an approach facilitates the classification of corrosion
phenomena. In the following paragraphs the specific role of crevice corro-
sion in the initiation or propagation of several types of localized corro-
sion is considered.

Filiform (Underfilm) Corrosion

Filiform corrosion has been described by Uhlig [61] as an example of
a differential aeration cell and by Fontana and Greene [57] as a special
type of crevice attack in which the growing filament becomes a "selfprop-
agating crevice." The characteristic long and narrow filaments with active
heads and colored corrosion product tails have been observed on steel,
aluminum, and magnesium surfaces under coatings that are permeable to
" This term has been used before, although not as a description of this particular
mechanism [38],

moisture. At 65 to 95 percent relative humidities, the attack is partic-

ularly severe, and the result is unacceptable appearance, especially on the
exteriors of canned products. In a recent study with aluminum, Hoch and
Tobias [62] found that the heads had a pH of less than 1 compared with a
range of 7 to 8.5 for the cathodic areas, which were approximately 100
times greater than the anodic areas. When these conditions were simulated
electrochemically with differential aeration cells, 60-/iA/cm2 corrosion
currents and 20-mV potential differences between anodic and cathodic
areas were measured. This corrosion rate corresponds to a growth rate
of 2 mm/week for a filament with a 0.5 mm width and a 0.1 mm depth,
which approximates observed rates. Kane [63] has reported the concen-
tration of chloride by electrochemical transport at the filament head on
magnesium. Thus, these data on differential aeration, acidic anodes, chlo-
ride concentration, and large cathode-to-anode ratios support the similari-
ties between filiform and crevice corrosion.

Pitting Corrosion
Pitting has been recognized often as a result of corrosive attack initiated
at crevices [20,31,61], and the extent and depth of pitting has been re-
corded as a measure of crevice corrosion [24]. A consideration of the
similarities between pitting and crevice corrosion led Schaffer et al [64]
to suggest that they are identical processes. In both cases the locations
for corrosion initiation were considered dependent on differential aeration
or inhibitor depletion, and the autocatalytic propagation was associated
with an acid mechanism. Their conclusion was that pitting at microscopic
sites is a limiting case of crevice corrosion at macrosites.
Actual similarities between the propagation of pits and crevices are
evident from the microscopic and electrochemical investigations of Rosen-
feld and Danilov [65]. They considered that nearly closed pits on stainless
steels severely retarded the diffusion of passivators and demonstrated the
inhibition of pit growth by destroying the shield layer to permit fresh
solution access to the pit interior.
Smialowska et al [66] recently examined the pitting of stainless steels
and found that predominant initiation sites were complex sulfide-oxide
inclusions in the metal. Dissolution of the sulfide shell around an oxide
particle resulted in a narrow crevice between the oxide and the metal.
From this active site, the propagation of pits can be considered one kind
of crevice corrosion.

Stress-Corrosion Cracking
For metals under tensile stress, solution stagnation in crevices can
create an environment in which stress-corrosion cracking occurs; tests de-
signed to evaluate both effects simultaneously with specimens like a

Stressed sheet containing a tapered crevice [67] can provide valuable in-
formation on synergistic interactions. Examples have been reported where
stainless and carbon steel specimens failed during operation of pilot plant
loops at elevated temperature because of previously entrapped caustic
solutions [68] in crevice areas. Such failures did not occur with simple
crevice geometries that could be rinsed effectively. When annealed and
sensitized Inconel 600, Inconel 625, Incoloy 800, and stainless steels
(Types 304, 304L, and 347) were stress corrosion tested in autoclaves
with high pressure water at 316 C, Copson and Economy [69] found that
a combination of high stress, high oxygen concentration, and crevices
where oxidizing cations (for example Cr+", Fe+'\ and Cu+-) could ac-
cumulate, promoted intergranular cracking. No attack was observed with-
out crevices, but Armijo [70] has reported subsequently that crevices were
not necessary for the initiation of cracking in sensitized Incoloy 800 and
Type 304 stainless steel. Besides the initiation of stress-corrosion cracking
from existing crevices, Sylwestrowicz [71] has proposed a mechanism for
stress-corrosion cracking of a copper-beryllium alloy where preferential
attack at grain boundaries of slip planes forms crevices with sharp apexes
from which cracking develops. Stress concentrations at a crevice tip
should be enhanced by the wedging action of solid corrosion products:
Pickering et al [72] have measured wedging pressure (7 ksi) which
resulted in crack tip stresses on the order of the metal yield strength.
Bombara [73] has suggested an electrochemical and geometric analogy
between stress-corrosion cracking and crevice corrosion. His model is
based on the initiation of stress-corrosion cracks from deep, narrow pits
that appear to be similar to crevices. A further generalization is that once
a crack is initiated from this site, the geometry of the advancing crack
resembles a tapered crevice; crack propagation then depends on factors
such as those outlined in Staehle's mechanistic interpretation of proc-
esses operating during stress-corrosion cracking [74].

Other Localized Corrosion Phenomena

Localized corrosion, such as exfoliation, fatigue, and intergranular
corrosion, also has been related to crevice effects. For example, exfoliation
corrosion of aluminum aircraft skin at countersunk fastener holes resulted
from corrosion when solution penetrated the crevice area between the
fastener and the unprotected periphery of the hole [75,76]. Also, the
occurrence of crevice corrosion in fatigue cracks has resulted in the in-
teresting development of an electrochemical crack detection system [77].
When cracked specimens are exposed to a chloride-ferrocyanide solution,
a visible precipitate is formed at the crack because of the initiation of
crevice corrosion. The technique has been applied to high-temperature
and high-strength alloys to detect cracks smaller than those detectable
with conventional liquid penetrants. Also, there is an inhibitive, non-

destructive aspect to this test because the corrosion action evidently

blunts the crack tip.

Prevention and Control of Crevice Corrosion

The primary method for reducing crevice corrosion susceptibility is
effective design of components and structures. However, when crevices
are unavoidable, cathodic protection and inhibitors may provide ade-
quate corrosion control. When these techniques are unacceptable or un-
suitable, a material change would be another possible solution. Each of
these prevention and control methods are discussed next along with
some examples to illustrate suitable applications and limitations.

Design Factors
The recognition and, of course, the elimination of crevices are the
most direct approaches for solving crevice corrosion problems. Numerous
specific recommendations for minimizing crevice corrosion through
proper design have been developed, and experience has shown that the
following procedures should be advantageous [19,20,57,78,79]: seal ex-
isting crevices with noncorrosive caulking compounds or by welding; weld
joints rather than bolt or rivet them; butt-weld joints instead of spot-
welding lap joints; use nonporous materials for gaskets or insulators
to avoid moisture absorption and retention; eliminate marine or debris
deposits by periodic cleaning; avoid designs that have sharp comers
or areas where moisture and debris can collect; paint or coat faying
surfaces before assembly; and inspect welds for porosity and material
components for fabrication or workmanship defects.
Although the intensity of crevice corrosion is considered greater with
metal/metal mating surfaces than with metal/nonmetal combinations
[43], the use of noncorrosive and nonporous materials for sealants and
gaskets is also important. As demonstrated by Wyche et al [80] the
amount of corrosion protection depends on the characteristics of the mate-
rial, and attack can be actually stimulated by substances within organic
compounds, such as sulfur-bearing rubbers that were more harmful than
rubbers without sulfur. Low moisture permeability is an important mate-
rials property during immersion conditions and especially during wet and
dry atmospheric exposures where corrosive moisture from condensa-
tion, leakage, or precipitation can be absorbed, retained, and concentrated
in contact with metal surfaces.
Evans [49] has suggested zinc coatings on steel windows to prevent
the seizing caused by voluminous ferrous corrosion products, and he
also has referenced a number of papers related to crevice corrosion in
buildings.
Among the more recent innovations are clad metals that can serve

as effective substitutes for mechanical connections [81]. Procedures like

explosion cladding offer solid metallurgical bonds that can provide gal-
vanic cathodic protection. One example is an explosion-formed transfer
joint between the aluminum superstructure and steel hull of a marine
vessel, which apparently provides resistance to crevice and galvanic
corrosion [82]. Another interesting case study has been reported by Bur-
ton [79] who observed that the penetration of chloride-containing water
into crevices between stainless steel heat exchanger tubes and tube sheets
can result in corrosion and stress cracking, as depicted in Fig. 9a. One
clever remedy for this problem involves material and design changes
(Fig. 9b) that result in cathodic protection of the stainless steel when
the steel corrodes in the crevice area. Other cathodic protection applica-
tions are considered in the next section.

S S STEEL STEEL

S S STEEL
CLAD

WATER
SIDE

a ~ b
FIG 9—Stainless steel tube cracked in a stainless steel tube sheet from chlorides
in the water around the tube (a) and cathodic protection by design with a stainless
clad steel tube for preventing cracking of the lube (b) [79].

Cathodic Protection
The application of cathodic currents has been investigated as a means
of controlling crevice corrosion of metals, especially in seawater. May
and Humble [83] exposed nickel, Monel, and stainless steels (Types
302, 316, 410, and 430) to quiet seawater at Kure Beach, and their
results showed that crevice corrosion and pitting were reduced effec-
tively when the specimens were coupled galvanically to magnesium sac-
rifical anodes. Later work by Vreeland and Bedford [84] with some
of these alloys (Types 304, 316, and Monel) under similar seawater
conditions confirmed these findings and provided statistical information

on the erratic occurrence of crevice corrosion. Various steel anode com-

binations were tested to determine their performance in reducing or
eliminating the extent and depth of attack.
Lennox et al [83] conducted experiments in quiescent seawater at Key
West with stainless steels (Types 304, 316, 17-4 PH, and 20Cb-3)
having crevices at insulators or rubber grommets. The results of exposures
for several hundred days demonstrated that crevice corrosion was the
predominant type of attack when compared with general or pitting corro-
sion. None of the alloys tested were immune to the attack in the absence
of cathodic protection, but 20Cb-3 was the most resistant. Additional
data for stainless steels Types 205, 304, 410, and 430 at 5600 ft depths
exhibited similar crevice corrosion characteristics. The usefulness of these
alloys for seawater service depended upon cathodic protection which
was applied with either stainless steel or aluminum anodes. Subsequent
work by these authors [86] demonstrated the application of cathodic
protection to Type 304 stainless steel O-ring seals in quiescent and flowing
seawater. Although flow rates of 0.5 ft/s (0.15m/s) were found to in-
crease the crevice corrosion damage, both steel and zinc anodes still
provided effective cathodic protection. For the seawater with pH 7.8
to 8.2, the pH in the active crevice was 1.2 to 2.0. However, with
steel and zinc sacrificial anodes, the pH in the protected crevice was
5.5 to 6.5 and above 10, respectively.
For some alloys, the elimination of crevice corrosion by cathodic cur-
rents can cause other problems. When aluminum anodes were used
to protect the 17-4 PH alloy, hydrogen embrittlement and cracking re-
sulted [85], and severe hydrogen blistering was observed on Types 410
and 430 stainless steels at the required current levels [83]. In this regard,
Vreeland's experiments [87] have indicated the careful potential con-
trol that is necessary for successful cathodic protection of precipitation
hardened stainless steel in seawater. For a steel with a 175 ksi yield
strength, the required cathodic potential was between —0.6 and —0.75 V
versus SCE. Potentials more noble than 0.6 V versus SCE could re-
sult in crevice corrosion and pitting, whereas potentials more active
than —0.75 V versus SCE could result in hydrogen embrittlement. The
effective range was extended to —1.3 V versus SCE for a 154-ksi pre-
cipitation hardened steel.
Cathodic protection for controlling crevice corrosion in nonmarine
environments has not been applied extensively. However, Mueller's lab-
oratory experiments [45] with several steels in solutions of sodium chlo-
ride with chlorine dioxide have demonstrated a decrease in general and
crevice corrosion with cathodic protection. For example, the crevice cor-
rosion rate of Type 316 stainless steel was reduced from an average of
3.9 mpy to less than 1 mpy with a 0.05-mA/cm2 cathodic current
at —0.06 V versus SCE. It was concluded that cathodic protection ap-

pears promising for bleached pulp washers, assuming that the neces-
sary cathodic current can be supplied to the very inaccessible crevices.

Inhibitors
The use of inhibitors for controlling crevice corrosion presents definite
problems because the relative inaccessibility of the crevice site makes
replenishment of depleted inhibitors difficult or impossible. For ex-
ample, Peers and Evans [88] found intense waterline attack of iron that
was partially immersed in chloride solution containing sufficient anodic
inhibitors (for example, phosphates and carbonates) to passivate the
immersed area. Their data supported the concept that this type of cre-
vice suffered attack because of the exhaustion and slow replenishment
of inhibitors to the meniscus. Rosenfeld and Marshakov [38] also found
that increased inhibitor concentrations were necessary to provide pro-
tection for metals containing crevices. For iron corroding in a near-
neutral solution, it was necessary to increase the inhibitor concentration
tenfold over that required for a freely corroding surface to protect a
0.05-mm crevice.
Accessibility to mating surfaces prior to assembly facilitates the applica-
tion of an inhibitor system. One approach is to cover these surfaces with
inhibited paints, such as red lead for steel and zinc chromate for aluminum
[89]. Other investigators [90] have wrapped metal foils in a paper that
was saturated with inhibitor solutions before exposing a stack of these
specimens to a humid sulfur dioxide atmosphere. They found that some
of the solutions under test actually enhanced corrosion and that the
addition of surface active agents improved the performance.
The chances for complete inhibition of crevice corrosion are im-
proved when only short-term service is required. In their laboratory
evaluation of Type 410 stainless steel in inhibited refinery cleaning solution
at 140 F (60 C), Freedman and Dravnieks [91] observed that the ex-
posed surfaces were protected adequately, but that the crevice areas
were attacked. The addition of 0.3 percent lead acetate or stannous
chloride eliminated crevice corrosion, but tests performed in the pres-
ence of hydrogen sulfide showed no beneficial effects from these salts.
Hatch [92] has investigated the inhibition of a differential aeration
cell composed of a bare steel panel in one chamber and a steel panel
wrapped with filter paper to restrict oxygen availability in the other
chamber. He conducted electrochemical measurements of potential and
the short-circuit current, which is a measure of attack on the wrapped
panel. A variety of inhibitors (for example, sodium nitrite, sodium
benzoate, and zinc sulfate) were evaluated, and the results indicated the
concentration ranges where the treated solution provided either inhibition
or stimulation of attack.

Materials Selection
The selection of materials for service where crevice corrosion is a
potential problem involves a comparison of their performance under
crevice conditions. However, with the possible exception of data from
seawater tests, the lack of standard tests has contributed to some erra-
tic and nonreproducible results that thwart attempts to establish meaning-
ful ratings of materials by a comparison of data from several sources. A
ranking by Tuthill and SchillmoUer [57,93] of metals and alloys in terms
of crevice corrosion resistance in quiet seawater is given in Table 1. An
TABLE 1—Relative crevice corrosion resistance of
metals and alloys in quiet seawater [93,57].
Metal or Alloy Resistance
Hastelloy C "1
Titanium j inert
90Cu-10Ni-1.6Fe
70Cu-30Ni-0.«Fe best
Bronze
Brass
Austenitic nickel cast iron
Cast iron [ neutral
Carbon steel
Incoloy 825 ]
Carpenter 20 j^^^
Ni-Cu alloy
Copper I
316 stainless steel ]
Ni-Cr alloys pit initiation
304 stainless steel at crevices
Series 400 stainless steels

additional factor that must be considered is the effect of seawater flow

rate, and these same authors [4,93] have provided guidelines for materials
subjected to seawater which is stagnant, quiescent (quiet), flowing mod-
erately, or flowing at high speed. For example, HasteUoy C and titanium
exhibit no attack over this flow rate range; aluminum alloys under stag-
nant or quiescent conditions exhibit little, moderate, or considerable
attack depending on the alloy; 70Cu-30Ni alloy exhibits little to no
attack over the entire flow rate range; and stainless steels exhibit little
to considerable attack over theflowrate range.
Although such lists provide useful guideUnes, the order of merit is
not necessarily applicable to other media or even to the specified media
imder varied test conditions. Consequently, actual materials evaluation
tests under the actual conditions of exposure are essential for reasonable
success in the prediction of crevice corrosion behavior. Some of the
more significant investigations of several materials in a variety of en-
vironments are discussed in the following paragraphs.
Titanium—A rather extensive amount of information is available on

the crevice corrosion behavior of titanium, particularly in halide en-

vironments [28^2,94-102]. Generally, this metal is considered quite re-
sistant to halide environments [28], especially seawater. However, Kane
[63] has indicated that the severity of attack increases with temperature,
especially above 250 F (121 C), halide concentration, and hydrogen
ion concentration. Some of these data have been represented in the
form of a corrosion diagram that Feige and Kane [96] prepared to show
the pitting and crevice corrosion of Ti-50A in brine solutions.
Ruskol and KUnov [97] found that cracks and fissures in titanium
corrode hundreds of times faster than the exposed surface in 0.75 to
2.0 N H2SO4. Schlain and Kenahan [98] reported that for other acid
solutions, including hydrochloric, oxalic, and formic, corrosion initiated
at crevice sites rapidly becomes general corrosion over the external sur-
face that was previously passive.
Bomberger [32] has demonstrated that wet chlorine attacks titanium
in deep crevices. He attributes this to the dehydration of the gas under
the stagnant crevice conditions. Once the moisture content is below
the amount required for inhibition, acidic and hygroscopic corrosion
products are produced and accelerate the attack. Reaction with available
water yields hydrochloric acid and titanium oxides.
Griess [34] has conducted anodic and cathodic polarization measure-
ments in addition to conventional corrosion tests with tapered crevices,
some of which were exposed in autoclaves. His results in high-tempera-
ture halide and sulfate solutions indicated that alloying with sufficient
concentrations of molybdenum, nickel, or palladium should improve the
crevice corrosion resistance of titanium. Several investigators [99,700] have
noted the significandy improved crevice corrosion resistance of a Ti-
0.2Pd alloy compared with titanium. Its performance is associated with
the surface enrichment of palladium and the resulting improvement in
the ease of passivation [707]. However, tests with this alloy in hot sea-
water loops have shown that it too is attacked after prolonged exposure
to chlorides at elevated temperatures [702].
Aluminum—In a recent review Godard [63] concluded that crevice
attack is not generally responsible for significant deterioration of alumi-
num. He associated practical problems with thin sections (for example,
less than 0.040 in. (0.1 cm)), metal appearance, and effects of bulky
corrosion products. Examples included the staining of stacked or coiled
aluminum sheet by moisture. Crevice corrosion in the atmosphere and
fresh water was considered negligible. However, such attack in the form
of pitting has been noted in seawater, where the extent is related to
the general corrosion resistance of the alloy, in soil, and in marine at-
mospheres where appreciable corrosion rates are probably proportional
to the salt concentration in the atmosphere. It has been shown that
when aluminum was exposed to sodium chloride solutions the corrosion

rate in crevices was ten times faster than that on the exterior surface
[103].
Other investigators [104] have conducted experiments in natural and
simulated seawater with sintered aluminum powder (SAP) sheets clad
with aluminum and aluminum alloys. They found that certain clad alloys
were more resistant to crevice attack than others.
Stainless Steels—Although this series of alloys has been discussed pre-
viously in this paper, some additional data are worth mentioning. With
regard to relative rankings, the Wilde and Williams [31] data for stainless
steels exposed for more than four years in seawater showed that the crevice
corrosion resistance decreased in the following order. Types 316, 304, 446,
430, and 410. Although improvement in crevice corrosion resistance of
stainless steels often has been associated with the addition of molybdenum
as an alloying element, other investigators [85] concluded that Type 316 is
not necessarily superior to Type 304 stainless steel for applications in quies-
cent seawater. Such discrepancies undoubtedly arise from the fact that the
same variables are not always controlled, or controllable, to the same ex-
tent in diflEerent experiments.
Data also have been reported for the crevice corrosion resistance
of stainless steel in contact with various sealing compounds [80] that
provide a considerable range of protection, with nitrile rubber O-rings
[59], and with powdered metal and oxide gaskets [705]. Pray and Berry
[106] found that seizing, as a result of ferric oxide formation at the pe-
riphery of a journal-sleeve arrangement, was more prevalent with Types
410 and 430 stainless steels than with Types 347 and 17-4.
Wyche et al [80] did not observe crevice corrosion on specimens of
stainless steels (Types 301, 302, and 316) with spot-welded lap joints
that were exposed for several years in a marine atmosphere. Evidently,
the conditions necessary for atmospheric crevice corrosion of these alloys
are more severe than those for crevice corrosion during immersion in
seawater.
Miscellaneous Materials—The crevice corrosion performance of vari-
ous other materials has been studied, although the information in the
literature is somewhat sparse. However, in addition to some of the practical
examples noted in the introduction, the following references should be
mentioned: crevice corrosion of zinc, cadmium, and lead in seawater
[107]; the dealuminization of aluminum bronze by chloride crevice at-
tack [108]; the dezincification of brass under deposits [105]; the filiform
and crevice corrosion of magnesium [63]; the waterline attack of zinc in
sodium chloride solutions [109]; the crevice corrosion and pitting of
nickel alloys [35] and a nickel-copper alloy [84] in seawater.
Electrolytic Crevice Corrosion
The phenomenon of electrolytic crevice corrosion was recognized in
1965 during a study of electrode mounting procedures [110]. In contrast

to electrochemical crevice corrosion, which generally becomes initiated

after some incubation period because of changes in the stagnant crevice
solution, electrolytically induced crevice corrosion can be initiated along
a narrow electrolyte path immediately when current is applied to a
metal-electrolyte system. Besides their contribution to the development
of this mechanism, France and Greene [40] have demonstrated the passi-
vation of crevices during anodic protection. A significant amount of
data on electrolytic crevice corrosion has been published recently by a
group of researchers [27,59,73,111-114], including Bombara, Sinigaglia,
Strocchi, Taccani, and Vicentini.

Phenomenological and Mechanistic Aspects

The early work [110] on materials for mounting electrodes revealed
that that presence of microscopic crevices (for example, 0.001 in. (0.0025
cm)) between the metal electrode and some encapsulating resins caused
a pronounced effect on the active-passive anodic polarization curves. At
passive potentials the measured current densities were as much as 100
times greater than the expected values, as shown in Fig. 10 [115]. Since
these observations were disclosed, several authors have reported similar
behavior for various metal-electrolyte systems [116-118]. These examples
demonstrate clearly that the avoidance of crevices in and adjacent to

-o.ao"-
1.0 10 lo' 10' 10

CURRENT DENSITY Ifulc/l

FIG. 10—Comparison of potentiostatic anodic polarization curves for systems

with and without crevices [115].

electrodes is necessary for the determination of accurate active-passive

polarization curves.
The electrochemical similarity between porous powder metal com-
pacts and crevices is illustrated in Fig. 11 with Tikkanen's data [119]
for sintered stainless steel. In this case the voids associated with in-
complete sintering apparently act as microcrevices giving a characteristic
increase in passive current density with increased compact porosity. The
near coincidence of the critical maximum current densities indicates
that changes in real surface area are insignificant compared with active
crevice effects.
1.10
SINTERED STAINLESS STEEL

CURRENT DENSITY ( / / V c m )

FIG. 11—Influence of powder compact density on stainless steel anodic polariza-

tion characteristics [119].

Besides recommendations for suitable electrode mounting procedures,

the initial study [110] established the basis for an electrolytic crevice
corrosion mechanism. Measurements showed that the crevice interior
can remain active and corrode rapidly (for example, 1200 /uA/cm^ at
—0.42 V versus SCE) even when the exterior metal surface is maintained
at passive potentials characteristic of low-corrosion rates (for example,
2 /*A/cm2 at 0.25 V versus SCE). Thus, the high passive current
densities represent the large current density contribution from active
crevice sites where the large potential (IR) drop between the passive
and active sites is a result of the high resistance along the narrow
electrolyte path within the crevice.

^^?- 12-;Crevice potential variation of CF^ stainless steel together with its
anodic polarization curve [40].

Other studies [40,55] with a special crevice assembly (Fig. 3c) dem-
onstrated that the crevice potential is not an absolute value, but varies
with the crevice length. Such an electrode potential distribution along the
length of a 0.009-in. (0.023-cm) wide crevice is shown in Fig. 12 along
with an anodic polarization curve for this CF-8 stainless steel in N H..SO4.
These data can be compared to data for Type 304 stainless steel, as
shown in Fig. 13. Although the exterior metal surface was controlled at
similar potentials in both experiments, the potential gradients are sig-
mficandy different. For example, at a control potential of 0.6 V versus
SCE, the crevice potential for the CF-8 alloy decreased to active values
within the first 0.5 in. (1.3 cm), whereas the crevice potential of the

FIG. n—Crevice potential variation of Type 304 stainless steel together with its
anodic polarization curve [40].

Type 304 alloy remained at passive potentials along the entire 2-in.
(5.1-cm) crevice length.
The appearance of specimens after extended testing was consistent
with the data in Figs. 12 and 13. For example, a CF-8 specimen that
was exposed in the crevice assembly at a control potential of 0.6 V
versus SCE for 75 h is shown in Fig. 14. The unattacked or passive
region is evident on the left, while the position of the current density
maximum is indicated by the area of severe attack. Corrosion actually
diminished to the right of this point (slight grain faceting) as expected
from the data in Fig. 12.

i - * :*;.**•?'': ^ f ^ ^ ^ ^ ^ H : ^ * * * : .vf^'v''"- '

FIG. 14—Crevice specimen of CF-8 maintained at a control potential of 0.6 V

versus SCE for 75 h in N H,50, at 25 C. Crevice face (bottom) on left [40].

Based on Ohm's Law, the resistivity equation, and the Fokin-Timonin

analysis of potential distributions along thin wires polarized in relatively
large diameter tubes [720], two equations have been derived for predict-
ing the length of crevice that could be passivated with an applied anodic
potential. The first equation defines the length of an active-passive crev-
ice that can be passivated, and the second equation defines the length
of a passivated crevice that can be maintained passive [40].

Jjp = AEp I (active-passive crevices) (1)

Lp = /^:ZL?. (passive crevices) (2)

P'»

where:
Lp = passive length,
AEp = passive potential range,
W = crevice width,
p = solution resistivity,
ie = critical anodic current density,
if = passive current density, and
AEa = active potential range.
The electrochemical terms are designated in Figs. 12 and 13, and for
the data shown, AEp = control potential —0.02 V.

A comparison of experimental and calculated passive crevice lengths

showed that there was good correlation, as indicated in Table 2.
Solovena et al [101] have made this same analysis for the electro-
lytic crevice corrosion of titanium in an acid-chloride solution at 98 C.
Their calculated and experimental values for the passive crevice lengths
showed the same good correspondence as the data in Table 2.
In another investigation of geometric factors associated with electro-
lytic crevice corrosion, Bombara et al [27] derived equations for the
location of the maximum crevice attack within a crevice for any value of
the anodic control potential. The solution to their final equation gave
two possible values, one related to a previously activated surface and
the other related to a previously passivated surface. Thus these two
calculations can provide supplemental information to Eqs 1 and 2.

TABLE 2—Comparison of calculated and experimental passive crevice lengths [40].

Stainless Control Lp (in.) Lp (in.)
Steel Poten tial (V versus SCE) E:xperimental Calculated"
CF-8 0.00 0 0
0.20 0.13 0.17
0.60 0.50 0.56
304 0.02 0 0
0.10 0.37 0.42
0.20 8.0 8.3'
" Calculated from Kq 1.
' Calculated from Eq 2.

Prevention and Evaluation oj Electrolytic Crevice Corrosion

Although anodic protection has been applied successfully as an in-
dustrial corrosion control method, the preceding data indicate that the
protection of crevices in complex plant structures can pose a problem.
However, the methods [40] for improving the feasibility of anodically
protecting structures containing crevices follow directly from Eq 1.
Because localized attack is avoidable only when the passive length equals
or exceeds the crevice length, variables that increase the numerator
and decrease the denominator of this equation are advantageous. How-
ever, there are some practical limitations, such as the fact that en-
vironments in industrial applications are usually fixed (that is, p is
constant) and engineering designs seldom permit the widening of the
crevice to increase the (W) term. Obviously, any attempt to eliminate
crevices with sealants must be complete because very narrow crevices are
the worst cases.
The most significant factor in Eq 1 was found to be the critical anodic
current density which can vary by orders of magnitude for different
metals and alloys. The importance of this finding is that a relative rank-
ing of electrolytic crevice corrosion resistance for metals and alloys in

a given environment can be established on the basis of the critical

current density (/,.). Such a ranking is presented in Table 3. Because
standard procedures were not used to determine each of the polarization
curves referenced and because there even can be variations in the critical
current density for specimens within a given alloy classification [729],
this list should be considered a guideline based on the referenced data
rather than an inflexible ranking.

TABLE 3—Predicted electrolytic crevice corrosion resistance^

{Hi saturated, N Ht^Oi, 25 C).
Representative ic Reference
Metal (iuA/cm') (for i„ value)
Cr-lONi 1 121
Ti-6A1-*V » 122
Hastelloy C 9 123
Titanium 15 123
304-L stainless steel 35 123
347 stainless steel 40 124
316 stainless steel 45 123
304 stainless steel 75 40
201 stainless steel 125 125
20Cr-80Ni 300 121
Cu-45.4Ni-2.35Al 350 126
CF-8 stainless steel 2 000 40
Chromium 4 000 121
430 stainless .iteel » 000 118
Nickel 12 000 121
Fe-16Cr 25 000 124
Iron 120 000 127
Fe-15Co 200 000 128
° Ranked in the order of decreasing resistance.

Sinigaglia et al [112] also have investigated the electrolytic crevice

corrosion susceptibility of several austenitic stainless steels in acid and
acid-chloride solutions. In a series of constant potential polarization
experiments with specimens having a neoprene ring to form a crevice
site, they determined a critical potential below which crevice corrosion
occurred and above which the specimen completely passivated. On the
basis of the critical potential, the crevice corrosion resistance of these
steels in sulfuric acid solutions decreased in the following order:
321>316^304>302, and in the presence of chloride ions the ranking
was: 316>321^302>304. Also, it was found that the intensity of at-
tack decreased as the crevice width increased, a confirmation of Eqs
1 and 2.
In other constant potential experiments Lizlovs [39] has evaluated
the electrolytic crevice corrosion resistance of several ferrous alloys in
sodium chloride solutions with a unique cell (Fig. 3fc). The initiation
of crevice corrosion was marked by an increase in polarization current
when the glass bead was moved against the specimen electrode to
create a crevice at the interface. Comparison of the current-time curves

at constant potential provided the following ranking in order of de-

creasing resistance to crevice corrosion: 25Cr-3.5Mo-Fe, 17Cr-3Mo, 316,
304 =: 25Cr-Fe, 17Cr-Fe. Besides the beneficial effects of molybdenum
as an alloying element, sulfate ions in acid-chloride solutions have been
shown to improve the electrolytic crevice corrosion resistance of ferritic
stainless steels [777].
Critique
The theoretical analysis of electrolytic crevice corrosion demonstrated
that the ability to passivate crevices is controlled by electrolyte resistivity,
crevice dimensions, and the electrochemical behavior of the metal-elec-
trolyte system. However, recent work has indicated that other factors,
such as solution composition and concentration can contribute to this
phenomenon. Such effects have been noted by Mazza [2] in his study of
the preferential corrosion of titanium in crevices and under wet salt
deposits on partially immersed specimens (Fig. ]h). Although short-
term electrolytic tests showed breakdown voltages of about 9 V for
specimens in hot, concentrated chloride solutions at 176 F (80 C), the
presence of crevices or salt crusts reduced this value to about 6 V
during 24-h exposures. His polarization curves showed passive rather
than active-passive behavior which was characteristic of the metals de-
scribed in the previous section, and probably as a result of this fact the
measured potential difference between the inside and outside of the
crevice was only about 250 mV after 15 min and then became nearly
zero after a short time. Because of this small potential difference,
Mazza concluded that crevice corrosion of titanium in this environment
occurred because of changes in the crevice solution rather than poten-
tial (IR) drops. The polarization current was assumed to cause polari-
zation concentration and a reduction in the stability of passivity due
to accumulation of corrosion products and chloride ions, hydrolysis, in-
teractions of complex ions, reactions of chlorine with water, and oxygen
and water depletion. Thus, there are other factors besides potential
differences that can contribute to electrolytic crevice corrosion, and
these must be considered before accurate predictions of passive crevice
lengths can be made.
Vermilyea and Tedmon [130] have derived mathematical equations for
predicting the variations in concentration and potential in a crevice with
inert sides. They found that small potential differences caused relatively
large concentration changes and correspondingly large corrosion cur-
rents at constant applied potential. During constant potential polari-
zation at 0.00 V versus SCE the current increased to a maximum value
of 25 fiA. Calculations indicated that the solution concentration changed
from 10-» M NajSOi in the bulk solution to 7.8 X 10-'' M at the
mouth of the crevice to 1.0 Af at the bottom of the crevice. The potentials

associated with these three positions were 0.00, —0.028, and —0.105
V versus SCE, respectively.
In summary, there are some unique aspects to electrolytic crevice
corrosion, such as calculable potential drops within a crevice and the
immediate initiation of attack when current is applied. Except in the
original research [40,55], investigators have not tended to differentiate
between electrolytic and natural corrosion, and its appears that further
research that would establish the extent of correlation between the two
phenomena would be worthwhile. Based on the success of electrochemical
techniques for the detection and quantitative evaluation of crevice corro-
sion, continued applications and further advances in this field can be an-
ticipated.

Conclusions
Based on the literature reviewed for this paper on crevice corrosion,
the following conclusions are pertinent: (1) numerous test assemblies
and procedures are available, but standard methods for crevice corro-
sion testing are definitely required; (2) significant progress is being
made in the development of unifying concepts of mechanisms for elec-
trochemical and electrolytically induced crevice corrosion; and (3)
appropriate designs are among the more effective ways of avoiding
crevice corrosion, but consideration of materials selection, anodic and
cathodic protection, and inhibitors is worthwhile when crevices are un-
avoidable.

Acknowledgment
The author wishes to thank the publishers and the following individ-
uals and their colleagues for permission to use material from their
publications: B. F. Brown and the Marine Corrosion Section, Naval
Research Laboratory; W. H. Burton, Allied Chemical Corp.; L. W. Gleek-
man. Materials and Corrosion Engineering Services; N. D. Greene, Uni-
versity of Connecticut; E. V. Kunkel, Celanese Chemical Co.; F. L. La-
Que, International Nickel Co.; E. A. Lizlovs, Climax Molybdenum Co.;
F. Mazza, University of Milano (Italy); I. L. Rosenfeld, Institute of
Physical Chemistry of the Academy of Sciences (USSR); M. H. Tik-
kanen, Institute of Technology (Finland); and particularly B. E. Wilde,
U. S. Steel Corp. for prepublished data. Thanks are also given to M. D.
Cooper, L. L. Lewis, L. C. Rowe, and R. L. Chance, General Motors
Research Laboratories, for their helpful editorial comments.
References
[1] Speller, F. N. and LaQue, F. L., Corrosion, SBIIA, Vol. 6, 1950, p. 209.
[2] Mazza, P., Corrosion. SBIIA, Vol. 23, 1967, p. 223.
[3] Kunkel, E. V., Corrosion, SBIIA, Vol. 10, 1954, p. 260.
[4] Brown, B. F., Machine Design, MBUIA, Vol. 40, 18 Jan. 1968, p. 165.

[5] Greene, N. D. and Jones, D. A., unpublished data (see Fontana, M. G. and
Greene, N. D., Corrosion Engineering, McGraw-Hill, New York. 1967, p.
288).
[6] Gleekman, L. W., panel discussion moderator, Materials Protection, MAPRA,
Vol. 6, No. 10, 1967, p. 22.
[7] Muroaka, J. S., Materials Protection, MAPRA, Vol. 9, No. 3, 1970, p. 23;
Rigo, J. H., Materials Protection, MAPRA, Vol. 5, No. 4, 1966, p. 54.
[8] Gray, K. O., Materials Protection, MAPRA, Vol. 3, No. 7, 1964, p. 46.
[9] Kirchner, O. E. and Morris, F. M., Corrosion, SBIIA, Vol. 7, 1951, p. 161;
staff feature. Corrosion, SBIIA, Vol. 17, No. 2, 1961, p. 14.
[10] Lechnick, W., Brown, P. E., and Schultz, B. G., "Induction of Crevice Cor-
rosion in Reactor Materials by Concentration of Coolant Additives," High
Purity Water Corrosion of Metals, National Association of Corrosion En-
gineers, Houston, Tex., 1968.
[11] Moehrl, K. E., Corrosion, SBIIA, Vol. 17, No. 2, 1961, p. 26; DePaul, D. J.,
Corrosion, SBIIA, Vol. 13, 1957, p. 75t.
[12] Myers, J. R. and Obrecht, M. F., "Deposit and Corrosion Problems in
Potable Water Systems of Buildings," Paper No. 90, 27th Annual Conference,
National Association of Corrosion Engineers, Chicago, 1971.
[13] Turnbull, J. M., Corrosion. SBIIA, Vol. 16, 1960, p. 11.
[14] Moskvin, V. M., Alekseev, S. N., and Novgorodskii, V. I., Protection of
Metals, PTMNA, Vol. 1, 1965, p. 499; Boyd, W. K. and Tripler, A. B., Jr.,
Materials Protection, MAPRA, Vol. 7, No. 10, 1968, p. 40.
[15] Suzuki, T., Saito, H., Yoshioka, K., and Kitamura, Y., Corrosion Engineer-
ing, BOGJA, Vol. 19, No. 3, 1970.
[16] Smith, L. L., Tappi, TAPPA, Vol. 47, 1964, p. 118A.
[17] Hutcherson, C. H., Materials Protection, MAPRA, Vol, 7, No. 6, 1968, p. 27.
[18] Hudson, J. C. and Wormwell, F., Chemistry and Industry, CHINA, Aug. 1957,
p. 1078.
[19] Landrum, R. J., Chemical Engineering, CHEEA, Vol. 118, 24 Feb. 1969.
[20] Coburn, S. K., Materials Protection, MAPRA, Vol. 6, No. 2, 1967, p. 33.
[21] Moody, G. T., Journal, Chemical Society, JCSOA, Vol. 89, 1906, p. 720.
[22] "Standard Method of Test for Corrosion Produced by Leather in Contact
with Metal," Designation: D 1611-60, Book of Standards, American Society
for Testing and Materials, 1969.
[23] "Standard Method of Test for Corrosive and Adhesive Effects of Gasket
Materials on Metal Surfaces," Designation: F 64—69, Book of Standards,
American Society for Testing and Materials, 1969.
[24] Ellis, O. B. and LaQue, F. L., Corrosion, SBIIA, Vol. 7, No. 11, 1951,
p. 362.
[25] Godard, H. P., Materials Protection, MAPRA, Vol. 2, No. 6, 1963, p. 38.
[26] Rasmussen, L. M., Corrosion, SBIIA, Vol. 11, 1955, p. 155t.
[27] Bombara, G., Sinigaglia, D., and Taccani, G., Electrochimica Metallorum,
ELMTB, Vol. 3, 1968, p. 81.
[28] Renshaw, W. G. and Bish, P. R., Corrosion, SBIIA, Vol. 11, 1955, p. 41t.
[29] Corrosion in Action, The International Nickel Company, Inc., New York,
1955.
[30] Peers, A. M., Transactions, Faraday Society, TFSOA, Vol. 51, 1955, p. 1748.
[31] Wilde, B. E. and Williams, E., "The Use of Current/Voltage Curves for the
Study of Localized Corrosion and Passivity Breakdown on Stainless Steels in
Chloride Media," submitted for publication in Electrochemica Acta, as part
of the Proceedings of the 3rd International Conference on Passivity, Cam-
bridge, England, July 1970.
[32] Bomberger, H. B., Titanium Review, Vol. 10, No. 1, Aug. 1962.
[33] Rice, J. K. and Loucks, C. M., Materials Protection, MAPRA, Vol. 4, No. 2,
1965, p. 14.
[34] Griess, J. C , Jr., Corrosion, SBIIA, Vol. 24, 1968, p. 96.

[35] Niederberger, R. B., Ferrara, R. J., and Plummer, F. A., Materials Protec-
tion and Performance, Vol. 9, No. 8, 1970, p. 18.
[36] Compton, K. G., Mendizza, A., and Bradley, W. W., Corrosion, SBIIA, Vol.
11, 1955, p. 383t.
\37] Fontana, M. G., Industrial and Engineering Chemistry, lECHA, Vol. 47,
Nov. 1955, p. 81 A.
[38] Rosenfeld, 1. L. and Marshakov, I. K., Corrosion, SBIIA, Vol. 20, 1964,
p. I15t.
[39] Lizlovs, E. A., Journal, Electrochemical Society, JESOA, Vol. 117, 1970,
p. 1335.
[40] France, W. D., Jr., and Greene, N. D., Jr., Corrosion, SBIIA, Vol. 24, 1968,
p. 247.
[471 Pourbaix, M., "Theoretical and Experimental Considerations in Corrosion
Testing." presented at the Symposium on State of the Art in Corrosion
Testing, Annual Meeting, American Society for Testing and Materials, To-
ronto, Canada, 22 June 1970.
[42] Pourbaix, M., Klimzack-Mathieiu, L., Martens, Ch., Meunier, J., Vanleugen-
haghe, CI., deMunck, L., Laureys, J., Neelemans, L., and Warzee, M., Cor-
rosion Science, CRRSA, Vol. 3, 1963, p. 239.
[43] Mears, R. B. and Evans, U. R., Transactions, Faraday Society, TESOA, Vol.
30, 1934, p. 417.
[44] Evans, U. R., Journal, Institute of Metals, JIMEA, Vol. 30, 1923, p. 239.
[45] Mueller, W. A., "Electrochemical Study of Crevice Corrosion of Steels in
Solutions of Sodium Chloride and Chlorine Dioxide," Paper No. 78, 26th
Annual Conference, National Association of Corrosion Engineers, Philadel-
phia, 1970; Mueller, W. A., Corrosion, SBIIA, Vol. 25, 1969, p. 473.
[46] Jones, D. A. and Greene, N. D., Corrosion, SBIIA, Vol. 25, 1969, p. 367.
[47] FitzPatrick, V. F. and Ayres, J. A., Materials Protection. MAPRA, Vol. 9.
No. 2, 1970, p. 27.
[48] McKay, R. J., Transactions, Electrochemical Society, TESOA, Vol. 41, 1922,
p. 201.
[49] Evans, U. R., The Corrosion and Oxidation of Metals, Edward Arnold. Ltd.,
London, 1960.
[50] McCulloch, L., Journal, American Chemical Society, JACSA, Vol. 47, 1925,
p. 1940.
[51] Evans, U. R., Journal, American Chemical Society, JACSA. Vol. 48, 1926,
p. 1601.
[52] Schafer, G. J. and Foster, P. K., Journal, Electrochemical Society, JESOA,
Vol. 106, 1959, p. 468.
[53] Tomashov, N. D., Theory of Corrosion and Protection of Metals, Macmillan.
New York, 1966, p. 470.
[54] Hibert, C. L., Product Engineering, PRENA, Vol. 27, No. 10, 1956, p. c6.
[55] France, W. D., Jr., "Controlled Potential Investigations of Localized Corro-
sion Phenomena." Ph.D. thesis. Rensselaer Polytechnic Institute, Troy, N. Y..
1966.
[56] Stern, M., Journal, Electrochemical Society, JESOA, Vol. 105, 1958, p. 638.
[57] Fontana, M. G. and Greene, N. D., Corrosion Engineering, McGraw-Hill,
New York, 1967.
[58] Korovin, Y. M. and Ulanovskii, I. B., Corrosion, SBIIA, Vol. 22, 1966, p. 16.
[59] Slrocchi, P. M., Vicentini, B., and Sinigaglia, D., Eleclrochimica Metallorum,
ELMTB, Vol. 1, 1966, p. 239.
[60] Brown, B. F.. Corrosion. SBIIA, Vol. 26, 1970, p. 249.
[61] Uhlig, H. H., Corrosion and Corrosion Control, Wiley, New York, 1963.
[62] Hoch, G. M. and Tobias, R. F., "Mechanisms of Filiform Corrosion," Paper
No. 19, 27th Annual Conference, National Association of Corrosion En-
gineers, Chicago, 1971.
[63] Godard, H. P., Jepson, W. B., Bothwell, M. R., and Kane, R. L., The Cor-
rosion of Light Metals, Wiley, New York, 1967.

[64] Schafer, G. J., Gabriel, J. R., and Foster, P. K., Journal, Electrochemical
Society, JESOA, Vol. 107, 1960, p. 1002.
[65] Rosenfeld, I. L. and Danilov, I. S., Corrosion Science, CRRSA, Vol. 7, 1967,
p. 129.
[66] Szklarska-Smialowska, Z., Szummer, A., and Janik-Czachor, M., British Cor-
rosion Journal, BCRJA, Vol. 5, 1970, p. 159.
[67] Copson, H. R. and Berry, W. E., Corrosion, SBIIA, Vol. 16, 1960, p. 79t.
[68] Larrick, A. P. and Weed, R. D., Materials Protection, MAPRA, Vol. 7, No.
4, 1968, p. 34.
[69] Copson, H. R. and Economy, G., Corrosion, SBIIA, Vol. 24, 1968, p. 55.
[70] Armijo, J. S., Corrosion, SBIIA, Vol. 24, 1968, p. 319.
[71] Sylwestrowicz, W. D., Corrosion, SBIIA, Vol. 25, 1969, p. 405.
[72] Pickering, H. W., Beck, F. H., and Fontana, M. G., Corrosion, SBIIA, Vol.
18, 1962, p. 230t.
[73] Bombara, G., Corrosion Science, CRRSA, Vol. 9, 1969, p. 519.
[74] Staehle, R. W., Corrosion, SBIIA, Vol. 23, 1967, p. 202.
[75] Erben, A. R., Materials Protection. MAPRA, Vol. 4, No. 8, 1965, p. 18.
[76] Miller, R. N., Materials Protection, MAPRA, Vol. 6, No. 2, 1967, p. 55.
[77] Industrial Research, Dec. 1970, p. 58 and letter from Griessel, R. W., Re-
search and Development Center, General Electric Co., Schenectady, N. Y.,
1971.
[78] Mears, R. B., Australasian Corrosion Engineering, ACSEA, Vol. 6, No. 12,
1962.
[79] Burton, W. H., Materials Protection, MAPRA, Vol. 6, No. 2, 1967, p. 22.
[80] Wyche, E. H., Voigt, L. R., and LaQue, F. L., Transactions, Electrochemical
Society, TESOA, Vol. 89, 1946, p. 149.
[81] Pocalyko, A., Materials Protection, MAPRA, Vol. 4, No. 6, 1965, p. 10.
[82] Iron Age, IRAGA, 4 Feb. 1971, p. 56.
[83] May, T. P. and Humble, H. A., Corrosion, SBIIA, Vol. 8, No. 2, 1952, p. 50.
[84] Vreeland, D. C. and Bedford, G. T., Materials Protection and Performance,
Vol. 9, No. 8, 1970, p. 31.
[85] Lennox, T. J., Ir., Groover, R. E., and Peterson, M. H., Materials Protection,
MAPRA, Vol. 8, No. 5, 1969, p. 41.
[86] Peterson, M. H., Lennox, T. J., Jr., and Groover, R. E., Materials Protection,
MAPRA, Vol. 9, No. 1, 1970, p. 23.
[87] Vreeland, D. C , Materials Protection, MAPRA, Vol. 9, No. 1, 1970, p. 32.
[88] Peers, A. M. and Evans, U. R., Journal, Chemical Societv, JCSOA, 1953,
p. 1093.
[89] Shreir, L. L., Corrosion, Wiley, New York, 1963.
[90] Shreider, A. V. and Gintsberg, S. A., Zhurnal Prikladnoi Khimii, ZORKA,
Vol. 33, 1960, p. 2541.
[91] Freedman, A. J. and Dravnieks, A., Corrosion, SBIIA, Vol. 14, 1958,
p. 567t.
[92] Hatch, G. B., Corrosion, SBIIA, Vol. 21, 1965, p. 179.
[93] Tuthill, A. H. and SchillmoUer, C. M., Transactions, Marine Technology
Society, MTSTA, Vol. 3, 1965.
[94] Awaya, H., Yoshimatsu, H., Kajiyama, Y., and Tomoguchi, S., Corrosion
Engineering, BOGJA, Vol. 16, No. 3, 1967.
[95] Jackson, J. D. and Boyd, W. K., Applications Related Phenomena in Titanium
Alloys, ASTM STP 432, American Society for Testing and Materials, 1968,
p. 218.
[96] Feige, N. G. and Kane, R. L., Materials Protection and Performance, Vol. 9,
No. 8, 1970, p. 13.
[97] Ruskol, Y. S. and Klinov, I. Y., Protection of Metals, PTNMA, Vol. 2, 1966,
p. 544.
[98] Schlain, D. and Kenahan, C. B., Corrosion, SBIIA, Vol. 12, 1956, p. 422t.
[99] Shimose, T. and Takamura, A., Journal, Japan Institute of Metals, NIKGA,
Vol. 29, 1965, p. 416.

[100] Takamura, Akira, Corrosion, SBIIA, Vol. 23, 1967, p. 306.

[101] Solovena, Z. F., Fokin, M. N., and Timonin, V. A., Protection of Metals,
PTNMA, Vol. 6, 1970, p. 187.
[102] Battelle Technical Review, BATRA, Vol. 16, No. 5, 1967.
[103] Rosenfeld, I. L. and Marehakov, I. K., Zhurnal Fizicheskoi Khimii, ZORKA,
Vol. 31, 1957, p. 2328.
[104] Tolstaya, M. A., Flegontova, L. N., and Dmitriev, Y. V., Protection of Metals,
PTNMA, Vol. 6, 1970, p. 514.
[105] Uhlig, H. H., The Corrosion Handbook, Wiley, New York, 1955.
[106] Pray, H. A. and Berry, W. E., Atomic Energy Commission Publication,
BMI-886, 24 Nov. 1953.
[107] Ulanovskii, I. B., Korroziya Metallov i Splavov, KMSPA, No. 2, 1965, p. 359.
[108] Gleekman, L. W. and Swandby, R. K., Corrosion, SBIIA, Vol. 17, 1961, p.
144t.
[109] Bianchi, G., Corrosion, SBIIA, Vol. 14, 1958, p. 245t.
[110] Greene, N. D., France, W. D., Jr., and Wilde, B. E., Corrosion, SBIIA, Vol.
21, 1965, p. 275.
[Ill] Strocchi, P. M., Sinigaglia, D., Vicentini, B., Electrochimica Metallorum,
ELMTB, Vol. 2, 1967, p. 38.
[112] Sinigaglia, D., Taccani, G., and Bombara, G., Electrochimica Metallorum,
ELMTB, Vol. 3, 1968, p. 297.
[113] Strocchi, P. M., Sinigaglia, D., and Vicentini, B. in Proceedings, Third In-
ternational Congress on Metallic Corrosion, Vol. 1, Moscow, 1969, p. 281.
[114] Strocchi, P. M., Taccani, G., and Sinigaglia, D., Electrochimica Metallorum,
ELMTB, Vol. 4, 1969, p. 49.
[115] France, W. D., Jr., Materials Research and Standards, MTRSA, Vol. 9, No.
8, 1969, p. 21.
[116] Bond, A. P. and Lizlovs, E. A., Journal, Electrochemical Society, lESOA,
Vol. 115, 1968, p. 1130.
[117] Lizlovs, E. A. and Bond, A. P., Journal, Electrochemical Society, JESOA,
Vol. 116, 1969, p. 574.
[118] Recommended Practice for Standard Reference Method for Making Potentio-
static and Potentiodynamic Anodic Polarization Measurements," Designation:
G 5-71, Book of Standards, American Society for Testing and Materials,
1971.
[119] Tikkanen, M. H. A., "Modern Methods on Studying Corrosion Resistance of
Sintered Stainless Steels," Hoeganaes Sponge Iron Corp., Stainless Steel
Powders Seminar, Detroit, 25 Feb. 1965, p. 41.
[120] Fokin, M. N. and Timonin, V. A., Doklady Akademii Nauk SSSR, DANKA,
Vol. 164, No. 1, 1965, p. 150.
[121] Hodge, F. G. and Wilde, B. E., Corrosion, SBIIA, Vol. 26, 1970, p. 146.
[122] Peters, J. M. and Myers, J. R., Corrosion, SBIIA, Vol. 23, 1967, p. 326.
[123] Greene, N. D. and Judd, G., Corrosion, SBIIA, Vol. 21, 1965, p. 15.
[124] France, W. D., Jr., and Greene, N. D., Corrosion, SBIIA. Vol. 24, 1968,
p. 403.
[125] Walker, M. S. and Rowe, L. C , Corrosion, SBIIA, Vol. 25, 1969, p. 47.
[126] Mansfeld, F. and Uhlig, H. H., Journal, Electrochemical Society, JESOA,
Vol. 115, 1968, p. 900.
[127] Steigerwald, R. F. and Greene, N. D., Journal, Electrochemical Society,
JESOA, Vol. 109, 1962, p. 1026.
[128] Mottern, M. M. and Myers, J. R., Corrosion, SBIIA, Vol. 24, 1968, p. 197.
[129] Wilde, B. E. and Greene, N. D., Jr., Corrosion, SBIIA, Vol. 25, 1969, p. 300.
[130] Vermilyea, D. A. and Tedmon, C. S., Jr., Journal, Electrochemical Society,
JESOA, Vol. 117, 1970, p. 437.

Crevice Corrosion of Some High-Purity

Ferritic Stainless Steels

REFERENCE: Lizlovs, E. A., "Crevice Corrosion of Some Higli-Purity

Ferritic Stainless Steels," Localized Corrosion—Cause of Metal Failure,
ASTM STP 516, American Society for Testing and Materials, 1972, pp.
201-209.
ABSTRACT: Crevice corrosion of some ferritic stainless steels of low-
interstitial content containing 2 percent molybdenum and chromium from 20
to 28 percent, and of 25Cr steels containing up to 5 percent molybdenum
was studied using the potentiostatic polarization technique and a special cell
developed at the laboratory. The crevice corrosion resistance in chloride
containing media increased with increased chromium content at the same
molybdenum level and with increased molybdenum content at the same
chromium level. The 26Cr-lMo steel was immune to crevice corrosion
at room temperature. As the molybdenum content was increased, the
immunity to crevice corrosion was extended to higher temperatures (up
to 60 C), and also, as the chromium content was increased from 25 to
28 percent for the 2 percent molybdenum level, the immunity to crevice
corrosion extended from room temperature to higher temperatures (up
to50C).

KEY WORDS: corrosion, crevice corrosion, polarization, ferritic stainless

steel, molybdenum, chromium, corrosion tests
High-purity ferritic stainless steels are involved in growing commercial
interest. The excellent corrosion resistance of this class of alloys has been
demonstrated in work at the Climax Research Laboratory and by other
investigators [1-5].- In particular, the earlier work in this laboratory indi-
cated that molybdenum-containing ferritic stainless steels of low-inter-
stitial content possess a remarkable resistance to crevice corrosion in
chloride media. In an extension of the previous work, the crevice corro-
sion resistance of several high-chromium ferritic alloys was investigated.
Materials
The experimental alloys were produced by induction mehing and casting
in vacuum. The chemical compositions are given in Table 1. The ingots
^Senior research associate. Climax Molybdenum Company of Michigan (a sub-
sidiary of American Metal Climax, Inc.) Ann Arbor, Mich. 48106.
' T h e italic numbers in brackets refer to the list of references appended to this
paper.
201

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
202 lOCAUZED CORROSION-CAUSE OF METAL FAILURE

TABLE 1—Chemical compositions of the oHojIS studied.

Composition, %
Steel
Type Cr Mo Ni Si P S N C Al
26Cr-lMo 26.15 0.99 0.016 0.008
20Cr-2Mo 19.70 2.01 0.010 0.003
22.5Cr-2Mo 22.52 2.00 0.011 0.003
25Cr 26.66 0.002 0.13 0.002
26Cr-2Mo 25.02 1.99 0.013 O.OOS
28Cr-2Mo 27.86 2.00 0.021 0.007
25Cr-S.SMo 25.58 3.49 0.13 0.002
25Cr-SMo 26.25 6.24 0.13 0. 019 0.012 0 002 0.013 0.002 0 006

were hot forged at 2000 F (1090 C) to %-in. (20-min)-thick plate,

then hot rolled to 0.30 in. (7.6 mm) at 1900 F (1040 C) and finally
cold rolled to 0.150-in. (3.8-mm)-thick strip. All materials were annealed
at 1800 F (980 C) for 1 h and water quenched.

Experimental Procedures
PotentiostcUic Crevice Corrosion Tests
Potentiostatic crevice corrosions tests were conducted in a cell developed
in the Climax Laboratory [6]. A crevice in this cell is generated by press-
ing a glass bead against the disk-shaped electrode. The unwanted crevices
are eliminated by pressing the electrode against a Teflon gasket by means
of a polyvinyl chloride bolt. The crevice corrosion cell was adopted for
use above room temperature by wrapping heating tape around the cell.
The applied heat was controlled by a thermistor immersed in the electro-
lyte. Temperature control was db 0.2 C.
All potentiostatic crevice corrosion experiments were performed in 1 N
sodium chloride (NaQ) solution at room temperature (21 to 23 C), 30,
40, 50, and 60 C. The electrolyte was purged continuously with prepuri-
fied nitrogen. Experiments usually were conducted at 0.600 and 0.400 V
and occasionally at 0.000 V with reference to saturated calomel electrode
(SCE). Polarization current and potential were recorded as functions
of time.
The electrolyte was allowed first to reach the desired temperature.
Because of the cell arrangement, the electrode was in contact with the
solution during this period. Then the potentiostat with a voltage preset
at 0.0 V or less was switched on. After the initial current surge subsided,
the electrode potential was adjusted manually to the desired value; and,
after the polarization cunent stabilized, the crevice was applied to the
electrode surface, that is, the glass bead was advanced until it pressed
against the electrode surface. The electrode with the crevice applied was
usually kept at a given potential for 2 to 4 h, and then the crevice was
removed from the electrode surface. Finally, the electrode was removed
from the assembly and inspected for corrosion damage. The results were

considered acceptable only when the corrosion damage occurred under

the bead and in no other place.
Crevice Corrosion Tests in Ferric Chloride Solutions
Crevice corrosion tests with most of the materials were also carried
out in 10 percent ferric chloride (FeCla) solutions (pHl). Specimens
for these tests consisted of 1 by % by 0.15 in. thick (25 by 20 by 3.8 mm)
surface ground and polished through No. 600 wet silicon carbide paper,
and finally through 0.3/xm alumina-distilled water slurry (same polishing
procedure as for the disk electrodes for potentiostatic tests). Crevices
were formed by stretching a rubber band around the specimeq with two
Teflon cylinders placed on the major faces of the specimen. Thus, four
crevices were generated, two where the rubber band contacted the edges
and two where the Teflon cylinders contacted the faces of the specimen.
Crevice corrosion damage was estimated by visual inspection and weight
loss determination.
Resulfs
Potentiostatic Crevice Corrosion Tests
Polarization currents as a function of time for various stainless steel
electrodes at different temperatures in 1 N NaQ are shown in Figs. 1
through 5. Since the electrode area for all specimens was the same (1.26

20Cr-2Mo

40 80 100 120 140 160 ISO 200

TME, MINUTES

^-Crevice applied to electrode surface.

•-Crevice removed from electrode surface.
FIG. 1—Variations in polarization current with time at 0.00 V (SCE) in 1 N
NaCl at 22±1 C.

— 1 r •• 1 1 1 1 1 1 I 1 1 1 1 1
10-3

28Cr-2Mo ^-^"^ V r
10-^
kMP^ 1

10-5
L ^
n\ 1:
10-* • -

10-^ 1 1 1 1 1 1 1 > 1 1 1 1 1 1

0 20 40 60 80 IOC 120 140 160 180 200 220 240 260 280
TIME, MINUTES

A-Crevice applied to electrode surface.

•-Crevice removed from electrode surface.
FIG. 2—Variations in polarization current with time at +0.600 V (SCE) in 1 N
NaCl at 50 ± 0.2 C.

1 1 I 1 I 1 1 1 1 1 1 1

t ;

10-3 26Cr-lMo .- -

10-^
!]

10-*
-
/ 25Cr-2Mo y

10-*
]
:
10-i 1 1 1 1 i 1 1 1 1 i 1 1

0 20 40 60 80 100 120 140 160 ISO 200 220 240 260

TIME, MINUTES

0-Crevice applied to electrode surface.

•-Crevice removed from electrode surface.
FIG. 3—Variations in polarization currents with time at +0.400 V (SCE) in 1
N NaCl at 30 ± 0.2 C.

25Cr-2Mo
r

t/i
Ui
tfi.

1 10 —
5 —
K
UJ
o:
-
tf
^
O
§
Sio-5
N
^
=
~

10-* —

10-' 1 1 1
1 1 1 1 1
60 80 100
TIME, MINUTES

-Crevice applied to electrode surface.

%'-Crevice removed from electrode surface.
FIG. A—Variations in polarization currents with time at +0.400 V (SCE) in 1
N NaCl at 40 ± 0.2 C.

20 40 60 80 100 120 140 160 180 200 220

TIME, MINUTES

(^-Crevice applied to electrode surface.

•-Crevice removed from electrode surface.
FIG, 5—^^riation in polarization currents with time at +0.600 V (SCE) in 1 N
NaCl.
205
Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 12:32:22 EST 2016
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
206 LOCALIZED CORROSION-CAUSE OF METAL FAILURE

cm^) and the same glass bead was applied to the surface, the polarization
current itself rather than current density may be used to compare the
relative rates of crevice corrosion. The results from the potentiostatic
crevice corrosion testing are summarized in Table 2.
Polarization currents as a function of time for various stainless steel
electrodes at different temperatures in 1 N NaCl are shown in Figs. 1
through 5. Only a small polarization current, 10~^ to 10-« A, is recorded
prior to the application of the glass bead to electrode surface, characteris-
tic of stainless steel in a passive state with virtually no corrosion. After
the glass bead is pressed against the electrode surface, the passive polari-
zation current either remains unchanged, if material is resistant to crevice

TABLE 2—Results of controlled potential crevice corrosion test in 1 N NaCl cell on

ferritic stainless steels.
Presence of Crevice Corrosion at
Indicated Test Temperature
Steel Type 21 to 22 C 30 C 40 C 50 C 60 C
20Cr-2Mo yes
22.6Cr-2Mo yes
25Cr yes
26Cr-lMo no yes yes
26Cr-2Mo no yes yes
28Cr-2Mo no no no yes
25Cr-3.6Mo no no no yes
2SCr-5Mo no no no no yes

corrosion, or rises sharply if crevice corrosion is initiated under given

experimental conditions. Since the only anodic reaction is the dissolution
of metal under the glass bead, the anodic current is directly proportional
to the amount of metal dissolved under the glass bead, and the time-
polarization current diagram shows the initiation and progress of crevice
corrosion. However, since corrosion rates are potential dependent, only
corrosion rates obtained at the same potential can be compared with each
other. The results from potentiostatic crevice corrosion testing are sum-
marized in Table 2.
In general, as the chromium content was increased, the resistance to
crevice corrosion was also increased. The lowest resistance to crevice
corrosion was shown by the 20Cr-2Mo alloy. The crevice corrosion for
this material was initiated readily at a rather low temperature (22 ± 1 C)
at 0.00 V. Upon removal of the crevice from the electrode surface, a
stable pit developed at the site of the crevice damage, as evidenced by
continuing high polarization current (Fig. 1). Under the same experi-
mental conditions, crevice corrosion was also initiated on the 22.5Cr-
2Mo alloy, but the crevice corrosion current in this case was two orders
of magnitude less than for the 20Cr-2Mo alloy, and the electrode re-
passivated upon removal of the crevice (Fig. 1). No crevice corrosion

at room temperature could be initiated on the 25Cr-2Mo alloy. For the

28Cr-2Mo alloy some crevice corrosion could be initiated at 50 C; how-
ever, this material repassivated after some initial crevice corrosion before
the bead was removed from the surface even at +0.60 V (Fig. 2).
Molybdenum content plays an even more striking role in increasing
resistance to crevice corrosion. High-purity 25Cr steel suffered crevice
corrosion at 0.00 V and room temperature [6]. Crevice corrosion current
for 25Cr alloy was approximately the same as for 20Cr-2Mo alloy except
the 25Cr steel repassivated after the bead was removed from the surface,
while 20Cr-2Mo continued to corrode. Both the 26Cr-lMo and 25Cr-
2Mo alloys were resistant to crevice corrosion at room temperature (up
to approximately 24 C). Crevice corrosion could be initiated on both
materials at 30 C, although the crevice corrosion current for the 25Cr-
2Mo alloy was about two orders of magnitude lower than for the 26Cr-
IMo steel (Fig. 3). On removal of the crevice, the 25Cr-2Mo electrode
repassivated while the 26Cr-lMo alloy continued to corrode at +0.40 V,
that is, a stable pit developed at the site of the crevice corrosion damage.
At 40 C both materials suffered severe crevice corrosion and stable pits
developed at +0.40 V. The crevice corrosion current for the 26Cr-lMo
alloy was about twice that for the 25Cr-2Mo alloy (Fig. 4); thus, the
25Cr-2Mo steel is more resistant to crevice corrosion than the 26Cr-lMo
steel. The 25Cr-3.5Mo and 25Cr-5Mo steels were completely resistant
to crevice attack at temperatures of 40 C or lower.
Crevice corrosion on 25Cr-3.5Mo steel was initiated at 50 C and on
25Cr-5Mo at 60 C (Fig. 5). The rate of crevice corrosion for 25Cr-
3.5Mo steel at +0.60 V was approximately the same as for 28Cr-2Mo,
except that 25Cr-3.5Mo steel did not repassivate but continued to cor-
rode, indicating that a stable pit was formed at the site of crevice damage.
Crevice corrosion on the 25Cr-5Mo at +0.60 V and 60 C could be
initiated only after repeated application of the bead to the surface. The
corrosion rate was quite low (not exceeding 30 /iA), and repassivation
occurred while the bead was still applied to the electrode surface (Fig. 5).
Crevice Corrosion Tests in Ferric Chloride
The results from crevice corrosion tests in acidified FeCls solution
are presented in Table 3. Comparison of Table 2 with Table 3 shows
that the FeCLs test was slightly less severe than the polarization test.
Nevertheless, both tests ranked the alloys in the same order of resistance.
It should be noted that the FeCls test, even at room temperature, is a
very severe test, and its severity increases greatly with the temperature.

Discussion
Comparison of the results from potentiostatic crevice corrosion tests
gave the following order of merit for resistance to crevice corrosion:

TABLE 3—Crevice corrosion tests in lOFeCU-GHiO + 0.1 N HCl.

Weight Loss in mg/dm' per Day Computed
over Entire Coupon Area (test duration
in days given in parentheses)
Test Tenjperature
Steel Type 25 C SOC 40 C 50 C
20Cr-2Mo 64 197
(1) (1)
22.5Cr-2Mo 0 3
(7) (1)
25 Cr 670
(2.5)
26Cr-lMo 0 6^2 229 764
(%3) (4) (1) (1)
25Cr-2Mo 0 0 0 210
(7) (1) (1) (1)
28Cr-2Mo 0 0 0 13
(7) (4) (1) (1)
26Cr-3.6Mo 0 7
(24) (5)
25Cr-5Mo 0 0
(63) (11)

20Cr-2Mo (least resistant) <25Cr <22.5Cr-2Mo <26Cr-lMo <25Cr-

2Mo <28Cr-2Mo <25Cr-3.5Mo <25Cr-5Mo. This order of merit is
also supported by the crevice corrosion tests in ferric chloride. Thus, an
increase in chromium content at the same molybdenum level, or an in-
crease of molybdenum content at the same chromium level, increases the
resistance to crevice corrosion and the temperature region over which
materials are immune to crevice corrosion. In other words, when the
ferritic stainless steels have a single phase microstructure, higher chromium
or molybdenum contents stabilize the passive film against localized break-
down that causes crevice and pitting corrosion. While pitting corrosion
was not studied in detail, it can be generalized that stainless steels which
are immune to crevice corrosion are also immune to pitting corrosion at
the same chloride concentration and temperature. No contradiction has
been found for this statement either in the work of this laboratory or in
the literature.
The materials in potentiostatic crevice corrosion tests showed three
types of behavior (excluding the case of immunity to crevice corrosion):
repassivation with the bead still applied to the electrode surface, repassi-
vation after the bead is removed from the surface, and continuation of
corrosion after the bead is removed from the surface.
Materials having the ability to repassivate with the bead still on the
surface could be highly resistant to crevice corrosion, provided the re-
passivation under the bead occurs at sufSciently high potentials (+0.60
V (SCE) or higher). It is reasonable to assume that materials that coil-
tinue to corrode after the removal of the crevice may pit at a given
chloride ion concentration and temperature; it is possible that these mate-

rials may not pit under some conditions in ferric chloride or in potentio-
static tests. If their passive films were damaged mechanically or a tempo-
rary crevice generated on a surface of the material, a growing pit could
be expected at the site of the damage.
The good agreement between the electrochemical crevice corrosion tests
and the ferric chloride exposure tests is considered especially important
because of the difficulty that has been experienced in evaluating crevice
corrosion resistance by laboratory procedures. Although further studies,
especially long-time exposure tests, will be still required to confirm the
results, it now appears that by electrochemical testing, conditions leading
to immunity to crevice corrosion can be defined, that is, a temperature
and chloride concentration can be established below which crevice corro-
sion would not occur under any circumstances for a given condition
within a reasonable time period.
Conclusions
1. Increase in chromium or molybdenum content or both for high-
purity single-phase ferritic stainless steels increased their resistance to
crevice corrosion in chloride environment.
2. Ferritic stainless steels containing 25 to 26 percent chromium were
immune to crevice corrosion at room temperature and 1 N NaCl concen-
tration provided they contained at least 1 percent molybdenum. Increase
in molybdenum content resulted in increase in temperature at which 25Cr
steels were immune to crevice corrosion.
3. The relative order of merit of resistance to crevice corrosion was:
25Cr-2Mo (least resistant <25Cr <22.5Cr-2Mo <25Cr-lMo <25Cr-
2Mo <25Cr-3.5Mo <28Cr-2Mo <25Cr-5Mo.
References
[/] Chernova, G. P. and Tomashov, N. D., Korroziya Metallov i Splavov, KMSPA,
Sb. 2, Moscow, 1965, pp. 7-13.
[2] Schwartz, C. D., Franson, I. A., and Hodges, R. J., Chemical Engineering,
CHEEA, 20 April 1970, pp. 164-169.
[3] Lizlovs, E. A. and Bond, A. P., Journal, Electrochemical Society, JESOA, Vol.
116, No. 5, May 1969, pp. 574-579.
[4] Bond, A. P. and Lizlovs, E. A., Werkstoffe und Korrosion, WSKRA, Vol. 21,
No. 5, May 1970, pp. 336-342.
[J] Lizlovs, E. A. and Bond, A. P., Journal, Electrochemical Society, JESOA, Vol.
118, No. 1, Jan. 1971, pp. 22-28.
[6] Lizlovs, E. A., Journal, Electrochemical Society, JESOA, Vol. 117, No. 10,
Oct. 1970, pp. 1335-1337.

Crevice Corrosion of Some Ni-Cr-Mo-Fe

Alloys in Laboratory Tests

REFERENCE: Jackson, R. P. and van Rooyen, D., "Crevice Corrosion

of Some Ni-Cr-Mo-Fe Alloys in Laboratory Tests," Localized Corrosion—
Cause of Metal Failure, ASTM STP 516, American Society for Testing
and Materials, 1972, pp. 210-221.
ABSTRACT: Results were obtained for six alloys, evaluated in the labo-
ratory for crevice corrosion resistance. Two methods were used: (1) direct
observation of corrosion damage in several electrolytes and (2) electro-
chemical determination of polarization curves. Results showed the ratings
of the alloys in the order of decreasing resistance using both methods. In
the direct crevice corrosion tests, it was found that the crevice size was
important, especially for the more resistant alloys. Some scatter was
found in those results. The polarization data provided more detail of cor-
rosion behavior over a wide range of potential; therefore, the electro-
chemical test was considered a superior tool in alloy screening and de-
velopment. Neither method, at present, can predict exact field performance.
KEY WORDS: corrosion, alloys, concentration cell corrosion, electro-
chemistry, electrolytes, polarization (charge separation), anodic polariza-
tion, pitting

Electrochemical methods have been used in corrosion studies for many

years. Potentiostatic steps or sweeps are now used frequently in the course
of obtaining polarization data, in addition to the older procedure based on
controlled impressed current. Both techniques are generally accepted, al-
though finer procedural details have varied quite widely. However, a cor-
relation of the electrochemical data with direct immersion corrosion results
has not been usually made in the literature.
The work in this paper was aimed at adding to the understanding of
some of the test variables in potentiostatic polarization and in the direct im-
mersion crevice corrosion experiments. The results obtained by these two
laboratory methods were found useful for alloy screening and alloy develop-
ment, and they provided some information about the mechanism of crevice
corrosion.

' Research investigator and section supervisor, respectively. The International

Nickel Co., Inc., Paul D. Merica Research Laboratory, Sterling Forest, Suffem,
N. Y. 1090L
210

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1

Downloaded/printed
Copyright 1972 by FM International
by AS www.astm.org
University of Washington (University of Washington) pursuant to License
JACKSON AND VAN ROOYEN ON CREVICE CORROSION 211

Experimental
Materials
Table 1 lists the compositions of the alloys used; heat treatments of the
alloys are listed in Table 2.
Procedure for Direct Immersion
Figure 1 shows a picture of a crevice corrosion coupon including nuts
and bolts made of Teflon^; these were screwed onto the specimen with a
0.25-mm Teflon washer in between, providing a "wide" crevice. In some
alloys a tighter crevice was examined, formed by placing six very small,
square pieces of thin foil of Teflon (0.0125 mm thick) under each of the
outside corners of the bolt, which was then hand tightened as much as
possible; the thicker washer was ehminated. In all tests the specimen size
was 5 by 5 cm, and the volume of test solution was about one liter.
Immersion tests were made for 72 h in 5 test solutions using wide crev-

FIG. 1—Makeup of the crevice corrosion specimens used in the experiments:

(a) as crevice corroded, (b) front view, (c) bolt, nut, and washers used, and
(d) side view.

' Registered trademark of E. I. duPont de Nemours and Co., Inc.

OS
m

O O 'O O

9) n 00
I-* ^ O ^ »i
o ooo o

00 TH
« o» • * » ) "O w>
o oooo o
e> o o e e> d
I
•S "o"
•T3

i.
e
o
O
I

pa
•<

0 . . . .
a « T3 • * «p 11

M « O O O ""
o
s) 00 to m a >»

OS t ^ O >0 U3 O

urn I
O.P.2 cn
iB-aill^a

TABLE 2—Heat treatment of the alloys used.

Hastelloy alloy CI commercial-
Incoloy alloy 826 [ mill annealed
Inconel alloy 625J as received
AISI 316
Exp Fe-base—1 h/2200 F/OQ»
Exp Ni-base—1 h/2000 F/WQ'
° OQ = oil quenched.
' WQ = water quenched.

ices. These test solutions were 5 percent FeQs at room temperature,

5 percent FeQa + 5 percent H Q at room temperature, 5 percent FeQs
at 50 C, 5 percent CuCU at 50 C, and 5 percent NaOCl at 50 C. The
tighter crevice specimens were exposed only to 5 percent ferric chloride
(FeCls) at 50 C, and Type 326 stainless and alloys 825 were omitted from
this test. Chemically pure materials and deionized water were used for
making the solutions.
The surfaces of the specimens were machine ground to about a 30-/*in.
finish. Before testing, they were degreased in acetone, followed by absolute
alcohol, and rinsed in distilled water.

Electrochemical Procedure
Specimens with an exposed surface of 10 cm^ for the stepwise potentio-
static polarization tests were prepared from sheet. All nonexposed surfaces
were masked with two coats of a single component epoxy that cured at
around 160 C. Just prior to testing, the specimens were abraded with No.
240 siUcon carbide paper, degreased in acetone, and washed in distilled
water.
The (anodic) polarization measurements were conducted with a Wen-
king potentiostat, and voltage increases were programmed on a Wenking
stepping potentiometer. Potentials, relative to a saturated calomel elec-
trode, and the logarithms of the current densities were recorded continu-
ously on an X-Y recorder. Tests were done in acids under several condi-
tions of temperature, pH, and chloride concentration, and hydrogen was
bubbled through the solutions during the total time of the exposure. No
potential corrections were made for IR drop in the solution. A fine probe
was used for the potential measurements, with its tip at a distance of be-
tween two and three times its outside diameter away from the specimen
surface. Under these conditions, IR drop corrections would be less than
2 mV at 103 ^ / c m ^ .
Since pH values approaching 1 have been reported by others for the
liquid inside crevices in stainless steel,* it was felt desirable to do the initial
electrochemical tests at this pH level; the four most resistant alloys, how-
ever, were found to show no active peaks, and the breakout potentials for
'Greene, N. D. et al, Corrosion, SBIIA. Vol. 21, 1965, p. 275.

these test materials were high and not well separated. Raising the test
temperature from 50 to 65 C did not affect these results significantly. Con-
sequendy, more concentrated electrolytes were chosen.
Results
Penetration Measurements
Using the wider of the two crevices (0.25-mm spacer), the tests were
severe enough to cause attack of Type 316 stainless steel and alloy 825
in all five test media, as shown in Table 3. In the same tests, crevice corro-

XABLE 3—Results in several solutions with a "wide" crevice {jfS-h exposures).

Average Penetration in Attacked Area, mmpy*
Type Alloy Exp Exp Alloy Hastelloy
Test 316 825 Fe-base Ni-base 625 Alloy C
6% FeCls, 50 C 410 200 35 0.01», 83 0.04' 0.02'
6% CuClj, 50 C 23 950 200 5 0.003'. 15 0.003' O.Ol'
5% NaOCl, 60 C 73 46 0.02"' 0.03', 10 0.003' 0.006'
6% FeCh + 5% HCl, RT» 250 360 21 0.006' 0.004' 0.02'
6% FeCl,, RT= 320 32 0.14' 0.004' 0.002' 0.02'
" Millimeters per year. Where large disagreement was evident between replicates of the
experimental Ni-base alloy, individual penetrations are listed for each specimen.
' General corrosion only; no visible local attack.
' RT = room temperature.

sion was not observed on Inconel'' alloy 625 or Hastelloy' alloy C,

while the two experimental alloys showed attack only in some instances.
Crevice corrosion was considerably more severe in very confined areas,
and undercutting of spacers by corrosion was prevalent.
Table 4 lists the results obtained in hot FeCls with a more restricted
crevice (0.0125-mm spacers). Very severe attack was observed on most
specimens of all four alloys tested. Hastelloy alloy C, with the lowest rate,
showed individual penetration rates of up to 10 mm per year, and an over-
all average of about 5.2 mm per year. Other overall average rates in milli-
meters per year increased as follows: Inconel alloy 625, 9.5; experimental

TABLE 4—Results in 5 percent FeCk at 50 C with "tight" crevice-

{72-A exposures).
Average Penetration in
Attacked Area,
Alloy mmpy° Avg
Hastelloy alloy C 0.02', 0.04', 4, 7, 10, 10 5.2
Inconel alloy 625 5, 7, 8, 12, 12, 18 9.5
Exp Ni-base 0.006', 9, 13, 17, 18, 21 13.0
Exp Fe-base 8,10,12,20,25,25 16.7
" Millimeters per year.
' General corrosion only; no visible local attack.

'Registered trademark of The International Nickel Co., Inc.

' Registered trademark of Cabot Corp.

Ni-base alloy, 13; and experimental Fe-base alloy, 16.7. Once again, it
was immediately noticed that much deeper penetrations had occurred un-
derneath the Teflon spacers that were used, although the crevice corosion
was not confined to the undercut regions.
Polarization Behavior
A solution of 0.5 hydrogen chloride (HCl) + 1 N sodium chloride
(NaCl) was found to provide a reasonable separation of the polarization
curves of the four most resistant alloys, as shown in Fig. 2. ITie order of
increasing resistance to pitting, judged by the potentials at which passivity

• • " " — "~ * HASTELLOY ALLOY C

INCONEL ALLC
ALLOY 625 „ —

L""'' EXP. Ni-BASE

^EXP. Fe-BASE

10'
fii/m^

FIG. 2—Polarization curves in 0.5 N HCl + i N NaCl at 50 C, and potential

increase of 3 steps of 2 mV every 10 min.

was lost (£p) was the same in the hot FeCU test, that is, exp Fe-base
< exp Ni-base < Inconel alloy 625 < Hastelloy alloy C. For the other
two alloys. Type 316 stainless steel and Incoloy* alloy 825, no passive
region was found, indicating that the severe test conditions favored high-
corrosion rates throughout for the less-resistant alloys. The two curves are
plotted in Fig. 3. The critical current density, h (Fig. 2), of alloy C was
the lowest in this series, but the other three were not much difiFerent from
one another. In this instance, therefore, the £i.-values were more impor-
tant in rating the alloys for crevice corrosion resistance.
After the tests were completed, the various specimens of the four resis-

INCOLOY ALLOY 825

" - 2

_1_
10,000
MA/CM'

FIG. 3—Polarization curves in 0.5 N HCl +1 N NaCl, and potential increase of

3 steps of 2 mV every 10 min.

tant alloys were examined for pitting. The only one without localized at-
tack was Hastelloy alloy C.
The effect of the rate of potential increase can be seen by comparing
Figs. 2 and 4. Although the relative ratings of the alloys remained un-
changed, the /c-values tended to increase in the slower tests (done at
2-mV step every 10 min, compared to 2-mV step every 3.3 min) while the
£p-values decreased. Apparendy equilibrium is not achieved in a few
hours and may require several days. It is interesting to note that Hastelloy
alloy C was not affected significantly by the change in stepping rate.
A similar effect of rate of potential change is shown for the experimen-
tal Fe-base alloy in Fig. 5, which contains data for a lower pH and higher
Cl~ concentration (total Q - = 3.5 N). In addition, the change in elec-
trolyte caused an increase in anodic current densities, and passivity was
obviously difficult to achieve at the test temperature, which was 50 C. The
experimental Ni-base alloy also showed the effects of Cl~ concentration
at two different CI" levels, as plotted in Fig. 6 for tests at 40 C.
Increasing test temperatures caused higher anodic current densities, as
shown in Fig. 7 for the Fe-base alloy in 1 N HQ -I- 2.5 N NaCl. At 40 C
there was reasonably good passivity, none at 60 C, and 50 C was interme-
diate between these two.
The usefulness of the polarization technique for alloy development and
processing optimization is demonstrated in Fig. 8. The experimental Fe-
base alloy was examined after various heat treatments, indicating an opti-
mum solution annealing temperature in the vicinity of 2150 F (1177 C).
Below this temperature, pitting resistance was indicated to be impaired.
Results of exposures to FeCla solutions agreed with the curves in Fig. 8.

HASTELLOV ALLOY C

INCONEL ALLOY (25

EXP. Fe-BASE

10 « 10 1
/la/cm^

FIG. 4—Polarization curves in 0.5 N HCl + i N NaCl, and potential increase of

one 2-mV step every 10 min.

^ fMVIS 10 MIN.

/ia/cm'

FIG. 5—Effect of potential increase rate on polarization curves for exp Fe-base
alloy in 1 N HCl + 2.5 N NaCl at 50 C.

1.0

/la/cn^

FIG. 6—Effect of chloride ion concentration in exp Ni-base alloy polarization

curves in 1 N HCl at 40 C, with potential increased at 2-mV step every min.

10' 10 J 10 •

FIG. 7—Effect of temperature on polarization curves for exp Fe-base alloy in 1

N HCl + 2.5 N NaCl, with potential increased at a 2-mV step every min.

21saan(l22lxrF
ZIOOT

AS HOT ROLLED

10" 10'
lil/cm'

FIG. 8—Effect of annealing temperature of exp Fe-base alloy on polarization

curves in 1 N HtSOt + / N NaCl, with potential increased at a 2-mV step every

IMscBssion
There was a dependence of crevice corrosion on crevice geometry, and
attack increased in severity as the crevice became tighter. This observation
has practical implications and warns that widely scattered results can be
expected if strict experimental control is not exercised. Even when it is,
and assuming that laboratory results are reproducible, it would remain ex-
ceedingly difficult to translate them into practical terms (field perfor-
mance).
In terms of mechanism, the direct immersion tests agree with sugges-
tions* that corrosion is accelerated by the accumulation of Cl~ in the an-
odic area. The greatest and earliest accumulation of corrosion products
(including H+ and Cl~) would be expected in the more confined areas,
where the greatest penetration would then result.
Anodic polarization data were found more reproducible than the pene-
tration determinations, and they provided some understanding of corro-
sion rates over a range of potential. The potential of the Fe-base experi-
mental alloy was measured in hot 5 percent FeCls, by means of a probe
through the Teflon used to make a tight crevice, while active corrosion
was occurring. It indicated that the reaction was taking place in the region

of the breakout (Ep) point of the anodic polarization curve, and not in the
active region, as found for a different system by Greene and Judd and
Jones and Greene.*
It seems that the environments in the polarization work were less severe
than the one inside the "tight" crevice (FeCU at 50 C), where alloy C
corroded rapidly, but more severe than those inside wide crevices, where
the experimental alloys and alloy 625 escaped crevice attack. The high-
corrosion rate of alloy C in the tight crevice, for instance, corresponds to a
calculated current density roughly of the order of 500 to 1000 /iA/cm*.
and it occurred at a potential in the neighborhood of 0.35 V versus satu-
rated calomel electrode (SCE). None of the curves showed such a high-
current density for this alloy. Instances where polarization data would
have predicted high rates of crevice attack for the experimental alloys and
alloy 625, but where very low weight losses were found, suggest that the
crevices were too wide to allow the buildup of corrosion products to a
dangerous level.
In pursuing this point further, an experiment was done to find out if
time was involved. An electrochemical test was carried out with alloy C
at 50C, with the potential held at +0.4 V (SCE) in 0.5 N HQ + 1.5
NaCl. After several days, a breakout was evident, but the attack was
located under the masking material and about an inch away from its
edge; slight seepage had occurred. This result tends to show that the
electrolyte conditions required for attacking Hastelloy alloy C at 50 C and
0.4 V (SCE) are more severe than the starting solution in the cell.
The electrochemical method ranked the alloys in the same order as the
exposures in chemical environments, but provided considerably more
detail, since a wide range of electrochemical potentials was covered.
Useful anodic polarization data were obtained in NaQ solutions acidified
by means of sulfuric acid (H2SO4) or HCl, at several concentrations
and temperatures. Furthermore, this method rated the alloys reproducibly
and appeared more reliable for screening purposes than weight loss ex-
periments.
Although they appeared superior, the electrochemical results still could
not provide exact details of how alloys would perform in the field.

Condusions
1. In direct immersion tests, penetration measurements were influenced
by crevice geometry. Relatively less resistant alloys (for example. Type
316 stainless steel) were readily shown to be susceptible. More highly
resistant alloys (for example, Inconel alloy 625, Hastelloy alloy C)

•Greene, N. D. and Judd, G., Corrosion. SBIIA, Vol. 21, 1965, p. 15; also
Jones, D. A. and Greene, N. D., Paper No. 67, Annual National Association of
Corrosion Engineers Conference, Houston, Tex., March 1969.

appear to require large changes in crevice electrolyte for attack, asso-

ciated with very tight crevices.
2. Polarization data, obtained potentiostatically, are very useful in the
screening of alloys for crevice corrosion resistance. There is a need to
exercise care in the choice of electrolyte and testing conditions in order
to obtain a spread in results for a given series of materials.
3. The electrochemical tests rated the alloys in the same order as the
immersion tests in various chemicals. However, the polarization data
were more reproducible and provided more detail.
4. The work indicates that a reproducible laboratory crevice test
should be developed for the more highly resistant alloys because of their
sensitivity to variations in crevice geometry.

A Study of Factors Affecting the Hydrogen

Uptake Efficiency of Titanium in
Sodium Hydroxide Solutions

REFERENCE: Covington, L. C. and Feige, N. G., "A Study of Factors

Affecting the Hydrogen Uptake Efficiency of Titanium in Sodium Hy-
droxide Solutions," Localized Corrosion—Cause of Metal Failure, ASTM
STP 516, American Society for Testing and Materials, 1972, pp. 222-235.
ABSTRACT: An electrochemical technique is described for estimating
the tendency of titanium to absorb hydrogen on prolonged exposure to
caustic solutions. Unalloyed titanium appears to be quite resistant. A
possibility of localized hydrogen absorption and corrosion due to iron
contamination is described. Ways of avoiding this problem are suggested.
KEY WORDS: corrosion, corrosion tests, titanium, hydrogen, sodium
hydroxide, electrochemistry, polarization (charge separation)

Nickel has long been the standard material of construction for equip-
ment handling hot caustic soda in the Chlor-Alkali industry. Due to the
increasing price of nickel an interest is developing in alternate materials
of construction. Titanium is a leading candidate.
Although no large scale service experience is presently available for
titanium in caustic solutions, corrosion tests [1,2]'' of titanium coupons
in chemical plant exposures have shown it to be resistant up to 37 percent
sodium hydroxide (NaOH) at 230 F.
Some metals with good general corrosion resistance, however, are
found to be embrittled severely by hydrogen on prolonged exposure to
caustic solutions. Conventional corrosion testing techniques require an
impractical amount of time to acquire the data needed to evaluate a
metal's susceptibility to hydrogen embrittlement under such conditions.
This paper describes an electrochemical technique which has been devel-
oped to predict the rate of hydrogen pickup with variations in time and
temperature.

' Senior research engineer and manager. Market Development, respectively.

Titanium Metals Corporation of America, West Caldwell, N. J. 07006.
"The italic numbers in brackets refer to the list of references appended to this
papa.
222

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 1

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
COVINGTON AND FEIGE ON HYDROGEN UPTAKE EFFICIENCY 223

Knowing the corrosion current and the hydrogen uptake efficiency of

a metal in a particular environment, it is possible to calculate the rate
at which hydrogen will be absorbed by the metal. The theoretical back-
ground for determining corrosion currents from electrochemical polariza-
tion curves has been well established by other investigators [3,4], so that
will not be discussed in this paper. This will simply be a description of
the techniques employed and the results obtained.
The relationship between hydrogen content and embrittlement of tita-
nium has been determined by previous investigators [5,6] so no attempt
is made here to relate hydrogen content to physical properties. This
study was aimed at developing a method of estimating the rate at which
hydrogen would be absorbed in specific enviroments.

Experimental Techniques
Corrosion currents (7^) and hydrogen uptake efficiency (HUE) were
measured for titanium in NaOH solutions covering the concentration
range from 10 to 60 percent and temperatures from 70 to 220 F.
Corrosion currents were determined from polarization curves. The
specimens for polarization measurements were Vi-in. disks punched from
sheet material. They were electropolished for 1 min at 20 V in a solution
consisting of 60 ml of 70 percent perchloric acid, 590 ml of menthanol,
and 350 ml of butyl cellusolve. Experience has shown that hydrogen
pickup using this procedure on bulk samples at room temperature is
virtually nil. The surface area exposed for polarization was 1 cm^.
The NaOH solutions used were prepared from reagent grade NaOH
and distilled water. A portion of each solution was titrated with standard
hydrogen chloride (HQ) solution to determine the exact concentration.
Electrochemical polarization data were obtained using a Wenking 6IRS
potentiostat. A polarization cell was constructed using a cylindrical Vycor
glass body and employing two electrode holders of the design of France,
Jr. [7]. A Luggin probe connected to a saturated Calomel reference
electrode was fused into the glass body. This cell is shown in Fig. 1.
Platinum-clad titanium was used as the auxiliary electrode. The solution
was circulated through the cell at the rate of 1700 ml/min from a
4000-ml reservoir. This resulted in a complete change of solution in the
cell about every 20 s. The solution was purged with argon to remove
all the air by bubbling the gas through the solution in the reservoir and
venting to the atmosphere.
Electrochemical polarization experiments were carried out potentiostati-
cally. The procedure followed was to start at the open circuit potential
and proceed in either the anodic or cathodic direction. The potential was
changed in increments of 20 mV for the first 100 mV on each side of
the rest or open circuit potential and then increased to 100 mV and

n o . I—Polarization cell.

finally 250 mV as the distance from the rest potential increased. Three
minutes were allowed at each potential setting before the current was
read. Different specimens were used for the anodic and cathodic portions
of the curve. This was done to avoid any effect from hydride or oxide
films formed in the preceding portion of the curve.
HUE data were obtained by cathodically charging specimens of tita-
nium which had been freshly pickled in nitric acid-hydrofluoric acid
(HNO3-HF) solution at a current density of 9 mA/cva^ for 1 h in each
of the caustic solutions investigated. These specimens were then analyzed
for hydrogen content by vacuum extraction at 1400 C.
The alloys used in this investigation were as follows:
AUoy Composition Heat Treatment
Ti-50A 0.08% max C; 0.05% max N; mill annealed
0.0019% Hg-, 0.20% O2; 0.09% Fe
Ti-0.2Pd 0.08% max C; 0.05% max N; mill annealed
0.0019% H/, 0.20% max Fe
Ti-2Ni 0.08% max C; 0.05% max N; mill annealed
0.0036% H2

Results and Discussion

A typical polarization diagram is shown in Fig. 2. Because of passiva-
tion effects the anodic portion of the curve was not used for determining
/„ the corrosion current. However, the cathodic portion of the curve
has a clearly discernible linear region which can be extrapolated to the
open circuit or corrosion potential (Ec). The intercept then gives the
corrosion current (/,,) [8].
The HUE of an alloy is simply the percentage of the total hydrogen
avaDable which was absorbed by the specimen. From our cathodic charg-
ing experiments we know the total amount of hydrogen that was generated

+ 1.50 -I—I I I 11II T 1—I I I I I 1—I 11JII

1.25 -

1.00
7\
.50
-®-
.25

S .25
u
>.50
<
^ .75
UJ
2 1.00

1.25
I.SO

1.75

-ZJOO
fiQ/cm^

FIG. 2—Polarization curve for Ti-SOA in 59 percent NaOH at 90 F. h = 2

inA/cm'. Ec = -1.556 V.

on the surface of the specimen from the ampere minutes of current that
was passed. If we assume that the cathodic reaction is entirely the dis-
charge of hydrogen ions, which\ should be true in a deaerated pure solu-
tion of NaOH, then we know f^om Faraday's law that 1 g of hydrogen
will be generated for each 96,500 c or 1608.3 A min of current passed.
A current density of 9 mA/cm^ maintained for 1 h, therefore, would
generate 0.000355 g of hydrogen per square centimeter of specimen
surface. A 1 cm* specimen totally immersed however, would, have 2 cm*
of surface area (two sides) and, therefore, would be exposed to 0.00071 g
of hydrogen. Analysis of the cathodically charged specimens will give the
amount of hydrogen (AW) expressed as weight percent which was picked

up. Kaowing the weight (W) of 1 cm^ of the specimen, the HUE can
be calculated from the following relationship:
(Ag) (W)
% HUE =
0.00071
Table 1 lists the HUE, h and corrosion rate values obtained for each
condition of NaOH concentration and temperature used in our investiga-
tion. From Faraday's law we have calculated that a corrosion current
of 1 fiA/cni' will generate 0.000327 g of Hg/year/cm^ of specimen

TABLE 1—HUE, h and corrosion rate values for Ti-SOA in NaOH solutions.
ppm Hs/Year ppm Hj/Year
Corrosion 1 X HUE 1 X HUE
NaOH, Temperature, 7c HUE, Rate," for 3/16 in. for 1/8 in.
% deg F MA/cm' % mpy Sheet Sheet
10.9 70 0.8 0.5 0.28 1 1
10.9 90 1.0 0.6 0.34 1 1.4
10.9 130 1.7 1.3 0.59 3 6
10.9 180 4.0 3.7 1.39 23 34
10.9 220 6.0 6.1 2.08 68 84
ie.8 70 6.2 0.7 1.89 6 8
16.8 90 7.6 1.0 2.60 12 17
16.8 130 13.0 2.7 4.61 56 80
16.8 180 80.0 6.8 10.4 276 398
16.8 220 60.0 8.4 20.8 801 1164
24.0 70 2.2 0.3 0.76 1 1.6
M.O 90 20.0 0.6 6.93 16 23
24.0 130 60.0 1.3 17.3 103 140
24.0 180 90.0 13.1 31.2 1874 2700
24.0 220 220.0 16.1 76.3 5629 8100
36.8 70 6.0 0.3 1.73 2 3.4
86.8 90 10.0 0.4 3.46 6 9
86.8 130 28.0 2.0 7.97 73 105
86.8 180 86.0 9.0 29.6 1216 1750
86.8 220 260.0 16.3 86.6 6470 9827
59.0 70 0.16 7.2 0.06 2 2.5
69.0 90 0.4 6.8 0.14 4 6
S9.0 130 0.9 6.1 0.31 9 12
69.0 180 9.0 6.0 3.13 86 124
69.0 220 40.0 7.0 13.9 602 723
NOTK
0.000827 X H U E X h X 10,000
ppm Hi/year for 8/16-in. sheet Ti-50A.
2.068
0.000827 X HUE X / . X 10,000
= ppm Hi/year for 1/8-in. sheet Ti-50A.
1.429
* Calculated from /« values using the relationship R(Mvy) = 0-13 I, e/p.
where
A(mpy) •^ corrosion rate in mils per year,
U = corrosion current density in MA/cm*,
« = equivalent weight of titanum in grams <= 12, and
p SB density of titanium in g/cm* = 4.6.

surface. The amount of hydrogen pickup that could be expected from a

year's exposure in each environment was calculated as follows:
ppm Hz/year = (0.000327) (h) (HUE) (10,000)
_

These values in Table 1 were calculated on the basis of 3/16-in. Ti-50A

sheet weighing 2.058 g/cm^ and 1/8-in. Ti-50A sheet weighing 1.429
g/cm^ since most titanium process equipment uses material in this thick-
ness range.
A plot of log ppm Hz/year versus the temperature at which the value
was determined gives an approximate linear relationship for each solution,
as shown in Fig. 3. By plotting the data for each solution it is possible to
construct a family of curves that permit one to estimate the hydrogen
pickup per year at any temperature and NaOH concentration within the
ranges covered by our data.
Figure 4 displays these data in the form of hydrogen isobars with

— " T — :

lj-

10*

~
A -
10'

1 ^

- -

y/^

:, 1 1^—
130
__J
150
TEMPERATURE 'r
170
1

FIG. 3—Predicted hydrogen pickup for Ti-SOA 3/16-in. sheet in 24 percent

NaOH solution.

1 \ r- n 1 1 "' •• 1 1

\ 1 \ \ -
\
1 1 \
1 1 ^
I -
\
\
-
\
\
I ^ -
J
\
-

- \ -

- \ ^ i 5 0 0 0 ppm HYDROGEN PER YR.

-
ppm HYDROGEN
• 50 ppm HYDROGEN P ER YR.
• 5 p pm HYDROGEN PE R YR. -
1 1 1 1 1 1
10 20 30 40 50
% NoOH
FIG. 4—Hydrogen isobars for 3/16-in. Ti-SOA sheet.

temperature as the ordinate and percent NaOH as the abscissa. It is in-

teresting to note that the point of maximum hydrogen pickup occurs in
solutions in the 20 to 30 percent NaOH range. At higher concentrations
there seems to be a passivation effect that reduces hydrogen absorption.
In this connection, we have visually observed that a film builds up on the
titanium in these solutions, and it appears to grow thicker with increasing
temperature and solution concentration. This probably accounts for the
passivating effect indicated by the data.
The question arises as to the relationship between the HUE measured
at a current density of 9 mA/cm^ and that existing in a corrosion situation
where the current density is two to three orders of magnitude smaller. It
would be most accurate if the HUE in each case could be measured at
current densities corresponding to the measured corrosion current den-
sities. However, this is impractical due to the long elapsed time that
would be required to accumulate sufficient hydrogen in the specimens
to be measured accurately.
An experiment was performed in which the percent HUE of Ti-50A
was determined at current densities of 9, 6, 3, and 1.5 mA/cm* in 59
percent NaOH at 180 F, The results are plotted in Fig. 5. This gives an
approximately linear relationship over the range covered. If this straight
line is extrapolated to zero current density, it indicates that the value for
HUE in the /<, current density region would be increased by a factor of
about three over that measured at 9 mA/cm^. However, it seems unlikely

za
1

j
26 -'
24
1
22
1 -
T
20 \

18
\
Ui \
• \
= 16
J*
_ \
14 Ic V
12 / ^Jv^
10 — ^ —
Y-
)'
8
9 ma/cm*^
/
6 -/ •
/
2
/

eOO 1600 2400 3200 4000 4S00 S600 6400 7200 8000 8800 9600
CURRENT DENSITY ^a/cm'

FIG. 5—Percent HUE versus current density in 59 percent NaOH at 180 F.

that the relation would continue to be a straight line function as the

current density approaches zero. The line may approach zero current
density asymptotically so that HUE values approach 100 percent or zero,
depending on the mechanism involved as indicated by the dotted lines in
Fig. 5.
Several investigators [9,10] agree that hydrogen is absorbed in metals
in the atomic form. A hydrogen ion (H+) is nothing more than a hy-
drogen atom stripped of its electron. The hydrogen molecule has a
diameter of 2.12 A. The diameter of the hydrogen atom is 1.058 A [9].
The diameter of the hydrogen ion or proton is about 10-^ A or one
hundred thousand times smaller than the size of the atom.
The titanium surface is covered with an oxide film which is an insulator
and will not readily conduct electrons. Therefore, the hydrogen ions (H+)
must diffuse through the film and contact the metal surface in order to
pick up an electron and be converted into a hydrogen atom, in which
form it diffuses into the metal lattice. Because of its extremely small size
the hydrogen ion should pass through the oxide film with ease. However,
when it picks up an electron its size is suddenly increased one hundred
thousand times, which may make it much more difficult for the hydrogen
atom to pass back through the oxide film again. This would suggest that
at higher current densities the hydrogen may build up considerable pres-
sures under the film. This is supported by the observation that at high-
current densities, scales on the surface of a titanium cathode will fre-
quently spall off due to the action of the evolving hydrogen.

This concept may also explain the rectifying behavior of titanium oxide
(Ti02) films. As an anode, titanium offers considerable resistance to the
flow of current in an electrolyte because an oxide film forms on the
surface which acts as an insulator. It will not readily conduct electrons,
and anions diffuse only slowly through the film to the metal surface. As
a cathode, however, Ti02 films do not block current flow because the
hydrogen ions (H+) diffuse through the film with ease.
The diffusion rate of hydrogen into titanium is proportional to the
square root of the hydrogen pressure [9]. At low-current densities, diffu-
sion rates through the oxide film might nearly equal the hydrogen genera-
tion rate so the hydrogen pressure would be very low and little or no
absorption would take place. However, at higher current densities the
hydrogen generation rate would exceed the diffusion rate through the
film with the result that the excess hydrogen would build up a pressure
that would drive it into the metal in accordance with the square root
rule S = K \/P, where: S = solubility, K = constant, and P = pressure.
Hydrogen analyses performed on specimens of Ti-50A immersed in
50 percent NaOH at 200 F for three months indicated a HUE of about
1 percent based on corrosion currents calculated from weight loss
measurements. The HUE measured from cathodic charging experiments
at 9 mA/cm^ in similar environments gave values near 13 percent. This
would indicate that for Ti-50A, the HUE tends to drop off towards
zero at low-current densities.
Since an accurate measurement of HUE seemed to be impractical for
a corrosion situation and since it appeared that HUE values measured at
9 mA/cm^ would tend to be higher than those encountered under corrosion
conditions, it was decided to use these values in estimating hydrogen
pickup in a corrosion environment. Any error involved would tend to be
on the high side, thus providing a safety factor, and it is believed these
values would tend to vary in the same manner from the effects of tempera-
ture and solution concentration as would HUE values obtained at corro-
sion current densities.
It is generally agreed [5,6,9] that for most applications hydrogen con-
tents of less than 300 ppm do not degrade the physical properties of
titanium sufficiently to cause any problems. Figure 4 indicates that Ti-50A
will endure several years exposure in solutions containing up to 59 per-
cent NaOH and at temperatures up to 130 F before hydrogen contents
begin to approach this level. Therefore, it could be expected to have
a useful life in NaOH solutions. Our experiment did not determine the
effect of time on the passivating film buildup. It seems reasonable to
assume that the film thickness would continue to increase with time to
some equilibrium value and that the corrosion current and hence the
hydrogen pickup would decrease in a corresponding manner. It would
seem logical, therefore, that the hydrogen pickup estimated from Fig.

4 would be a maximum value and in actual practice might be found to be

considerably lower. The low-corrosion rates reported by Gegner and
Wilson [2] in industrial exposures would tend to support this idea.
In order to establish some correlation between industrial exposures
and the predictions based on these electrochemical measurements, arrange-
ments were made with a major chemical company to expose specimens
of Ti-50A and Ti-2Ni in the caustic effluent from their chlorine cells. This
solution was 50 percent NaOH at a temperature of 200 F. Specimens
were weighed at one month intervals, and two specimens were withdrawn
and analyzed for hydrogen at the end of the second month.
Weight loss measurements indicated a corrosion rate of 10 mils per
year (mpy) for Ti-50A and 9.8 mpy for Ti-2Ni for the first month which
decreased to 5.5 mpy for Ti-50A and 4.7 mpy for Ti-2Ni the second
month. The hydrogen analyses at the end of the second month indicated
the Ti-50A had picked up 60-ppm hydrogen in two months, while the
Ti-2Ni had increased by about 5500 ppm and was severely embrittled. The
Ti-50A showed no localized corrosion and was still fully ductile. The hy-
drogen pickup of 60 ppm on the Ti-50A, 0.040-in.-thick specimen would
have amounted to only 20 ppm on ^/^-in.-thick sheet or 14 on YiQ-in.
sheet. Also, these specimens were totally immersed so hydrogen pene-
trated from both sides. In service, probably only one side would be exposed
so the hydrogen pickup could be expected to be less. The exposures are
continuing, and, if the corrosion rate continues to drop, the Ti-50A
should show sufficient passivation by the end of the third or fourth month
to have a very useful life in this environment.
The embrittlement of the Ti-2Ni alloy in the plant exposure could
have been predicted from HUE measurements made in this environment.
Table 2 compares values of HUE for Ti-50A and Ti-2Ni in 10.9 percent
NaOH at various temperatures. The Ti-2Ni alloy was found to have very
high HUE values.
TABLE 2—Comparison of HUE for various titanium alloys.
HUE
Temperature, Ti-50A Smeared Anodized
NaOH, % deg F Ti-50A Ti-2Ni Ti-0.2Pd with Fe Ti-50A
10.9 70 0.5 12. 1
10.9 130 0.9 26. 7
10.9 180 3.7 57. 8
10.9 220 6.1 70. 1
16.8 70 0.0 2.5
16.8 130 4.8 8.3
16.8 180 6.3 5.5
16.8 220 11.9 3.1
24.0 70 0.3 4.2 9.5 0.5
24.0 130 1.3 20.5 11.5 1.5
24.0 180 13.1 2.8 19.3 3.6
24.0 220 16.1 1.6 34.6 4.5

Nickel alloy additions to titanium form the Ti2Ni compound which is

beneficial in resisting anodic pitting attack in the neutral pH brine en-
vironment. Under cathodic control or general corrosion of NaOH, the
Ti2Ni compound provides a local site for hydrogen absorption. The signifi-
cant increase in HUE of the TuKi compound was demonstrated in our
experiments. Field experiments confirm its unsuitability for service in
the caustic environment. However, in neutral brine environments where
hydrogen pickup is not a problem, the ability of Ti-2Ni to resist pitting
and crevice corrosion can be used to advantage.
Localized Corrosion Due to Iron Contamination
Previous investigators [11] have demonstrated that the presence of
metallic iron at the titanium surface facilitates hydrogen ingress into the
titanium. The mechanism of this attack is based upon the premise that
the surface contaminant is initially anodically attacked, and, either at
an adjacent site or within the same site, hydrogen is charged into the
metallic lattice. If sufficient hydrogen is absorbed to exceed the solubility
limit for the system, titanium hydride (TiH^,) is precipitated, providing
a phase with very low anodic pitting resistance. At this point the process
becomes autocatalytic and can continue to proceed after the initial local
area has been consumed.
Iron additions as either smeared iron or alloy addition react the same
as nickel under cathodic control. The HUE values in Table 2 demonstrate
this point. These values were obtained on specimens of Ti-50A which had
been smeared with iron by abrading the surface with a blunt mild steel
rod mounted in a drill press. In every case, iron contamination greatly
increased the HUE.
The upper solubility limit of iron, in practice, to prevent the Ti2Fe
compound formation and the possibility of localized attack is 0.12 per-
cent maximum. Present U.S. titanium industry practice is to hold iron
levels below 0.12 percent in both ASTM Grades 1 and 2 formulations.
Of more serious concern in commercial practice is smeared surface iron
resulting from material handling and fabrication. Carbon steel wire
brushing of welds is a classic mode of failure in these marginal environ-
ments where general attack is occurring to the titanium. Steel wire brush-
ing should be avoided. Acid pickling or stone grinding has been found to
be effective ways of removing surface iron contamination. Also, anodizing
can remove surface iron contamination if it is carried out for a sufficient
length of time to permit complete dissolution of the iron.
In addition to removing surface iron, the anodizing process builds up
a passive oxide film which is quite effective in suppressing hydrogen ab-
sorption. This is also evident from the data in Table 2 which lists HUE
values obtained on specimens of Ti-50A that were anodized for 20 min
at 90 V in a saturated solution of NaNH4HP04.

The demonstrated ability of anodized films to suppress hydrogen pickup

of titanium in solution, on first thought, is not consistent with the concept
that hydrogen ions because of their small size diff'use through surface
films with ease. However, if we consider the subject more carefully we
can see why the film retards hydrogen pickup. The cathodic reaction
H+ -)- e~ -» H is balanced by an equal and opposite anodic reaction in-
volving anions which must diffuse through the anodized film to the metal
surface to give up an electron which then travels through the metal to
the cathodic site where it is picked up by a H+. Thus, in the absence
of an external electromotive force (emf), the cathodic reaction can pro-
ceed no faster than the anodic reaction supplies it with electrons. Thus,
by blocking the access of the relatively large anions to the metal surface,
the anodized film effectively suppresses the cathodic reaction also and
thus the absorption of hydrogen.
The data in Table 2 are graphically shown in Fig. 6 where we have
HUE values plotted versus temperature for the various titanium alloys.
From this it is apparent the titanium-palladium alloy has a reduced HUE
under general corrosion attack. The alloy's significant commercial appli-
cation over the years has demonstrated its usefulness when the system
has an occasional excursion into the highly acidic or basic range. It also
appears to be promising for applications where hydrogen embrittlement

130 150
TEMPERATURE "F

FIG. 6—Comparison of HUE values for various titanium specimens in 24 per-

cent NaOH.

may present a problem. The variability in our testing program is most

probably related to the high-current density imposed and the surface being
unable to cope with the solid hydride layer. At lower current densities, the
majority of the surface film is probably a passive oxide with the palladium-
rich areas being the site of hydrogen ion discharge.

Summary
The mechanism of titanium's corrosion in caustic solutions is slow,
general anodic dissolution of the surface film with the rate of attack ac-
celerated by an increase in temperature. This reaction, although not
detrimental in metal wastage from the corroded surface, is the cause of
hydrogen ion discharge to atomic hydrogen at the oxide-metal interface.
The basis of this experimental work was to evaluate the anodic dissolution
rate with temperature and caustic concentration and predict the hydrogen
discharge to atomic hydrogen at the metal surface.
The test program excludes the passivating effect of oxide film which
will significantly reduce the rate of attack and operates at current densi-
ties considerably higher than the predicted corrosion current /,.. The
test procedure allows the investigator the opportunity to segregate one
step; the hydrogen uptake at the metal-oxide interface, and compare the
differences between alloys and surface finish preparations under the worst
possible test conditions.
The experimental technique described measures over a 1-h period the
formation of a hydride film and the bulk diffusion of hydrogen with
time at temperature through the interface. HUE in the simplistic terms
of the test program, is the amount of hydrogen absorbed on the surface
versus the amount evolved on the surface.
This work has demonstrated that electrochemical methods can be
used to rapidly estimate the suitability of a particular titanium alloy and
surface film condition for service in caustic environments. The data
point up the wide variability of HUE dependent upon surface film con-
ditions and indicate what direction may be taken to mitigate the rate
of attack. The short-term laboratory data reflect the worst conditions
possible in the NaOH environment.

Acknowledgment
The experimental work reported in this paper was carried out by
J. J. Henderson of TIMET's Henderson Technical Laboratories.

References
[/] Sheppard, R. S., Hise, D. R., Gegner, P. J., and Wilson, W. L., "Performance
of Titanium vs. Other Materials in Chemical Plant Exposures," Corrosion,
SBIIA, Vol. 18, June 1962, pp. 211t-217t.
[2] Gegner, P. J, and Wilson, W. L., "Corrosion Resistance of Titanium and

Zirconium in Chemical Plant Exposures," Corrosion, SBIIA, Vol, 15, July

1959, pp. 341t-350t.
[3\ Steam, M. and Geary, A. L., 'Electrochemical Polarization," Journal, Electro-
chemical Society, JESOA, Vol. 104, No. 1, Jan. 1957, pp. 56-63.
[4] Greene, N. D., "Predicting Behavior of Corrosion Resistant Alloys by Po-
tentiostatic Polarization Methods," Corrosion, SBIIA, Vol. 18, April 1962,
pp. 136t-142t.
[5] Williams, D. N., Schwartzberg, F. R., and Jaffee, R. I., "Hydrogen Contamina-
tion in Titanium and Titanium Alloys, Part V: Hydrogen Embrittlemcnt,"
WADC Technical Report 54-616, Wright Air Development Center, Feb. 1959.
[6] Lenning, G. A., Berger, L. W., and Jaffee, R. I., 'The Effect of Hydrogen on
the Mechanical Properties of Titanium and Titanium Alloys," Battelle Memo-
rial Institute, Fourth Summary Report to Watertown Arsenal, Ordnance Depart-
ment, Department of the Army, Contract No. DAI-33-OI9-505-ORD-(P)-l.
[7] France, W. D., Jr., Journal, Electrochemical Society, Vol. 114, No. 8, Aug.
1967, pp. 818-819.
\_S\ Henthorne, M., "Polarization Data Yield Corrosion Rates," Chemical En-
gineering, CHEEA, 26 July 1971.
[9] Livanov, V. A. et al, Hydrogen in Titanium, published by the Israel Program
for Scientific Translations, 1965, Daniel Davey & Co., New York.
[10] Evans, U. R., "The Corrosion and Oxidation of Metals: Scientific Principles
and Practical Applications," Edward Arnold Publishers, Ltd., 1960, p. 397.
[//] Cotton, J. B., "Using Titanium in the Chemical Plant," Chemical Engineciinf;
Progress, CEPRA, Vol. 66, No. 10, Oct. 1970, pp. 57-62.

Prevention of Hydrogen Blistering and

Corrosion by Organic Inhibitors in
Hydrocarbon Systems of
Varying Composition

REFERENCE: Nathan, C. C , Dulaney, C. L., and Leary, M. J., "Pre-

vention of Hydrogen Blistering and Corrosion by Organic Inliibitors in
Hydrocarbon Systems of Varying Composition," Localized Corrosion—
Cause of Metal Failure, ASTM STP 516, American Society for Testing
and Materials, 1972, pp. 236-249.
ABSTRACT: The inhibition of the corrosion of steel, caused by hydrogen
sulfide, was studied at various pH ranges in the presence of an oil phase.
Oil-soluble filming amine-type inhibitors are effective in stopping acid cor-
rosion. At 7.7 pH, only two of the oil-soluble inhibitors tested had any
effect. At 10 pH, in the presence of cyanide ion, only the water-soluble
inhibitors are effective.
KEY WORDS: corrosion, corrosion tests, corrosion prevention, cyanides,
inhibitors, pH, evaluation, hydrogen sulfide, tests

Numerous papers have been written on the effects of hydrogen sulfide

on steel [1-23, 25-33].^ Effects are almost uniformly bad, and generally
due to the handUng or processing of sour (sulfur containing) petroleum
or natural gas.
Approximately 40 percent of the oil and gas wells in the United States
produce a sour product [72]. Sulfide caused corrosion occurs in well equip-
ment, crude storage tanks, and pipelines [1, 12], and also in crude dis-
tillation units, catalytic cracking units, and other units of refineries and
chemical plants [2, 3, 15, 22, 23].
Previous laboratory studies have been made on acidic systems {5-7, 9,
10, 14, 16, 19-22, 25, 26, 29, 30], or on basic systems in the absence
otoW.\,3,4,ll,17,27,31].
This paper presents a study of the effects of corrosion inhibitors on

' Senior research associate, senior research chemist, and research chemist, re-
spectively, Betz Laboratories, Inc., Trevose, Pa. 19047.
' The italic numbers in brackets refer to the list of references appended to this
paper.
236

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
NATHAN ET Al ON HYDROGEN BLISTERING AND CORROSION 237

the hydrogen sulfide induced corrosion of steel (Fig. 1). Tests were made
in the presence of oil at various pH ranges up to 10.

Procedures
Corrosion Test
The apparatus used, and the procedures followed, for the corrosion
tests were the same as those of Gardner [16] and Gardner et al [5].
Briefly, a glass reaction bottle was charged with test coupon, aqueous
solution, oil, and hydrogen sulfide. The reaction vessel was then rotated
at 15 rpm in a cabinet heated to 80 C for a 20-h period.
Distribution Coefficients
No coupons were used. The oil/water distribution for cyanide and for
hydrogen sulfide was determined.
Analytical Procedures
Cyanide was analyzed by ASTM Tests for Cyanides in Water (D 2036-
68), employing the modified Liebig titration procedure. Hydrogen sulfide
was analyzed by the method of Shannon and Boggs [14].

Experimental Work: Theoretical Basis

Hydrogen Sulfide Corrosion: Redox Potential
In addition to pH, another method of arriving at a corrosive tendency
is to measure the oxidation-reduction, or redox, potential of an iron elec-
trode immersed in the corrodent [17]. In the absence of oxygen, as is the
usual case in a refinery, the redox potential is essentially a linear function
of the pH. Therefore, throughout this paper, the pH is used as a measure
of the corrosive tendency.
Corrosion caused by hydrogen sulfide is pH dependent. In general,
four pH ranges may be distinguished [3, 27, 31].
1. At a pH level below 4.5, acid corrosion and hydrogen blistering
occur. Both increase in severity with decreasing pH. Hydrogen penetra-
tion occurs because of poisoning of the surface by sulfide ions (see next
section for discussion of hydrogen blistering).
2. In the pH range of 4.5 to 6.0, because of low concentration of both
hydrogen and bisulfide ions, low rates of corrosion and hydrogen pene-
tration are observed.
3. The least protective sulfide films are formed [27] at pH 6.0 to 8.0.
These films consist of Mackinawite (ferrous sulfide, previously called Kan-
site) [28]. If the film is mechanically broken, corrosion may occur.
4. At a pH level above 8.0, in the absence of cyanide ion, very low
corrosion rates occur because of the protective sulfide film. In the pres-
ence of cyanide ion, hydrogen blistering and corrosion can occur. Cyanide

ion is very strongly absorbed on steel; the heat of absorption is approxi-

mately 20 kgcal/mole [24]. Cyanide ion also dissolves sulfide films to
give iron-cyanide complex ions.
Hydrogen Blistering
Hydrogen atoms are formed on a steel surface by corrosion processes,
even in basic systems. If the surface is poisoned by strong adsorption of
an ion, such as sulfide, cyanide, or arsenic, the hydrogen atoms cannot
combine to form molecules. The atoms penetrate the surface and may
diffuse completely through fairly thick metal walls.
If imperfections are present, hydrogen atoms may combine to form
molecules which cannot diffuse farther. Concentrations of molecular hy-
drogen can occur which are sufficient to exceed the yield strength of the
metal.
Corrosion Inhibitors: Effect of Oil
Acid Corrosion—The majority of the organic corrosion inhibitors are
of the filming amine type. These are molecules that contain amine groups
or nitrogen ring compounds, with a long chain hydrocarbon. The nitrogen
is adsorbed on a metal surface, and the hydrocarbon tail helps make the
surface hydrophobic [33].
The presence of oil has a marked effect on inhibitor performance. In
the complete absence of oil, filming amines have been found to be in-
effective [9]. Nathan [18] found the inhibition by filming amines to be
proportional to the concentration of inhibitor in the oil phase. Boies [6]
found that inhibition increased as the aromaticity of the oil phase in-
creased.
An oil phase is always present in the case of refinery or chemical plant
corrosion. It thus appears that studies of filming amine inhibitors should
be made in the presence of an oil phase.
Almost all filming amines are formulated in oil. In laboratory tests in
aqueous systems, the amount of oil present in the inhibitor may be suffi-
cient to provide the necessary oil phase [10].
Basic Corrosion Inhibition
Most filming amines work best at a pH of 4.5 to 6.5, and little or no
work has been reported on the effect of inhibitors with an oil phase in
basic systems.

Experimental Results
Chemicals Used
All inorganic chemicals used were chemically pure and were used as
received. Water was demineralized. The oil used was "Odorless Insecticide

Base" from American Mineral Spirits Corp. This oil is a highly purified,
dearomatized kerosine.

Aqueous Solution for Corrosion Tests

The aqueous solution for the pH 7.7 test contained 0.77 percent ammo-
nium sulfide ((NH4)2S); 5 percent ammonium chloride (NH4CI). Any
pH adjusted to 7.7 by addition oi 6 N hydrochloric acid (HCl). Any
sulfur that precipitated was filtered out before use of the solution. The
(NH4)2S was Fisher light ammonium sulfide, 20 to 24 percent by weight.
The pH 10 aqueous solution consisted of 4 percent ammonia (NH3)
(from ammonium hydroxide (NH4OH)) and 1000 parts CN- (from
sodium cyanide). All tests were run with 60 ml of aqueous phase and
30 ml of oil, unless stated otherwise.

Inhibitors Used
With the exception of two, the corrosion inhibitors were oil soluble.
All were of the filming amine type. Both of the water soluble amines
have some oil solubility. Table 1 shows the inhibitors used and their
composition as determined by infrared analysis.

TABLE 1—Inhibitors used and their composition as determined

by infrared analysis.
Inhibitor Solvent Composition
A oil cyclic amine + linear amine
B oil cyclic amine + linear amine
C oil cyclic amine + linear amine
D oil linear amide amine
E oil linear amide amine
F oil cyclic amine + linear amide
G water linear amine
H water linear amine

Test Program
Distribution Experiment
The water-oil distribution coefficients for cyanide and hydrogen sulfide
were determined, as shown in Table 2.

TABLE 2—Water-oil distribution coefficients for cyaniie and hydrogen sulfide.

Conditions: 30-ml AMSCO odorless insecticide base and
60-ml water rotated for 20 h at 15 rpm at a
temperature of 80 C.
CYANIDE"

CN in Oil
Added Found in Water Oil CN in Water
526 478
1072 930
HYDROGEN SULFIDE

HsS in Oil
Added Found in Water Oil HzS1 in Water
352 91 261 2.87
680 158 522 3.3
760 170 590 3.46
1112 249 863 3.46
1787 407 1380 3.4
2508 588 1920 3.3
4070 1040 3030 2.91
"NOTE:—No cyanide was found in any oil layer.

Effect of pH on Water-Oil Hydrogen Sulfide System

The pH of the system was varied by adding on HCl or NH4OH to
adjust the pH. Data are shown in Table 3 and Fig. 2.

TABLE 3—Ejfect of pH on wat-er-oil hydrogen sulfide system.

Conditions: 30-ml oil and 60-ml aqueous phase rotated 20 h at 80 C. Hydrogen sulfide
maintained constant at 1100 ppm.
pH as pH After Average Weight Appearance
Aqueous Phase Prepared Test Loss, mg of Specimen
38 000-ppm HCl 1.2 1.0 690 a
19 000-ppm HCl 1.6 1.2 367 a
10 000-ppm HCl 1.5 1.4 242 a
5 000-ppm HCl 2.1 1.6 149
1 000-ppm HCl 2.5 2.4 54 3k"
Demineralized water 6.6 4.4 26 3
500-ppm NHiOH 9.6 5.1 10 2
1 000-ppm NH4OH 9.4 5.9 8 2
5 000-ppm NH40H 9.9 6.6 3.5 1
10 000-ppm NH4OH 10.2 7.7 3 1
SO 000-ppm NHiOH 10.7 8.3 2.5 1
100 000-ppm NH4OH 10.9 8.7 4 1
<* General acid attack.
' See Fig. 1 for comparison of degree of hydrogen blistering of coupons.

60
\

50 NUMEROUS - PINPOINT - NO BLISTE

\ f^YDROSE^t ISTERS
BLISTER s
6ENER ACID
40 — ACID ATTACK
ATTAC K PLUS
SMALL
e LISTER 3
30
L •

9 " ^
•
4 5 6
pH of AQUEOUS SOLUTION

FIG. 2—Effect of pH on corrosion and hydrogen blistering in presence of H,S.

BLISTERING INCR EASING HEAVILY BLISTER ED NO VISIBLE OIF FERENCE

WITH CONCENTF ATION WITH CONCENTRATION

•
j ^ • •
•
2S
'I
m/
20

/•

0 1,000 ZPOO ZSXlO 4J300 SflOO

TOTAL HjS CONCENTRATION, ppm

FIG. 3—Effect of HiS concentration on corrosion in oil-water system.

EQect of Hydrogen Sulfide Concentration on the Water-Oil System

Only hydrogen sulfide was added to the water and oil. Effects of the
inhibitors were studied, as shown in Table 4 and Fig. 3.

TABLE 4—Effect of hydrogen sulfide concentration on water-oil system.

Conditions: 30-ml oil and 60-ml water, rotated 20 h at 80 C. Inhibitors added at 100
ppm, based on total fluid. Inhibitors 25 percent active (weight basis).
Hydrogen Sulfide, ppm
Inhibitor 0 160 338 508 723 1284 1525 2271 2949 3705 4102
None:
Weight loss, mg 4 19 17 29 29 32 40 29 30 26 30
Appearance of
specimen 1" 2 2 4 4 5 5 6 6 6 6
Inhibitor A:
Weight loss, mg 4 7 7 11 7 7 7 5 4
Appearance of
specimen 1 1 1 1 1 1 1 1 1
Inhibitor B:
Weight loss, mg 5 6 9 3 21 6 8 10 8 11 4
Appearance of
specimen 1 1 1 1 E» 1 1 1 1 1 1
Inhibitor C:
Weight loss, mg 6 8 7 7 7 7 7 8 10 7 9
Appearance of
specimen E' 1 1 1 1 1 1 1 1 1 1
Inhibitor D:
Weight loss, mg 12 6 8 9 5 9 10 7 4 7 6
Appearance of
specimen 1 1 1 1 1 1 1 1 1 1 1
Inhibitor E:
Weight loss, mg 7 9 11 6 27 17 11 11 12 9 9
Appearance of
specimen 1 1 1 1 E 1 1 1 1 1 1
Inhibitor F:
Weight loss, mg 6 14 11 6 7 10 9 9 7 8 6
Appearance of
specimen 1 1 1 1 1 1 1 1 1 1 1
Inhibitor G:
Weight loss, mg 14 14 11 16 13 33 25 16 11 13 9
Appearance of
specimen E E E E E E E E E E E
Inhibitor H:
Weight loss, mg 7 6 7 6 9 6 17 8 17 6 6
Appearance of
specimen E E £ E E E E E E E E
" Compare with specimens pictured in Fig. 1.
' Indicates specimen etched.

Intermediate 7.7 pH Test

The effect of inhibitors on 7.7 pH is shown in Table 5.
TABLE 5—Effect of cyanide concentration in intermediate pH 7.7 teat.
Conditions: 30-ml oil and 60-ml aqueous phase, rotated 20 h at
80 C. Inhibitors added at 100 ppm, based on total fluid.
Inhibitors Weight Loss, mg Appearance of Specimen
None 43 etched
A 1 smooth
B 39 etched
C random (13,21,49,61) etched
D 60 etched
E 42 etched
F 51 etched
G 39 etched
H 47 etched

Effect of Hydrogen Sulfide Concentration in pH 10.0

System with Cyanide
Results with the various inhibitors are shown in Table 6 and Fig. 4.

TABLE 6—Effect of hydrogen sulfide concentration in pH 10.0 system with cyanide.

Conditions: 30-ml oil and 60-ml aqueous phase. Cyanide constant at 1000 ppm. In-
hibitors added at 100 ppm, based on total fluid. Bottles rotated 20 h at 80 C.
Hydrogen Sulfide Added, ppm
Inhibitors 160 338 508 723 1284 1525 2270 2950 3705 4100
None:
Weight loss, mg 8 10.5 9.5 9 9.5 15 14 17.5 17 16.5
Specimen' appearance 1 2 3 3 3 4 5 5 6
Inhibitor A:
Weight loss, mg 15 10 12 15 14 19 18 22 21 21
Specimen appearance 1 2 2 3 3 4 5 5 6 6
Inhibitor B:
Weight loss, mg 11 13 16 13 15 11 12 22 24 21
Specimen appearance 1 2 2 3 3 3 5 6 6 6
Inhibitor C:
Weight loss, mg 17 15 15 16 16 12 17 24 23 22
Specimen appearance 1 2 2 3 3 3 4 5 6 6
Inhibitor D:
Weight loss, mg 14 17 15 14 17 15 14 22 23 22
Specimen appearance 1 3 2 2 3 4 4 6 6 6
Inhibitor E:
Weight loss, mg 7 12 11 13 13.6 19 18 21.6 21 20
Specimen appearance 1 2 2 3 3 3 4 5 5 6
Inhibitor F:
Weight loss, mg 9 14 13 13 15 16 15 20 22 21
Specimen appearance 1 2 2 3 3 4 4 6 6 6
Inhibitor G:
Weight loss, mg 5 7 8 10 9 9 12 13 16 15
Specimen appearance 1 1 2 2 3 3 3 3 3 3
Inhibitor H:
Weight loss, mg 4 4 4 4 5 3 4 5 5 3
Specimen appearance 1 1 1 1 1 1 1 1 1 1
• Figure 1 shows typical specimen appearance.

LARGE B LISTERS ALL LARGE BLISTERS

TIN Y BLISTE RS
APP •AR NO VISIBLE DIFFERENCES
' "

•
«
^^^."-"^

O 10
•^
•

1,000 2,000 3,000 4,000 5,000 ejOOO 7,000 8,000 9,000 10,000
HjS CONCENTRATION, ppm

FIG. 4—E^ect of H,S concentration on corrosion in pH 10 aqueous-oil system.

Effect of Cyanide Concentration in pH 10.0 Test

A constant hydrogen sulfide content of 4100 ppm was used. Results
are shown in Table 7 and Fig. 5.

NO
HEAVY BLISTE RS y^
BLISTER-
_ NO VISIBLE OIFFEF
so " IN6 "

' '
40

500 1,000 2,000 3,000

CYANIDE CONCENTRATION, ppm

FIG. 5—Effect of cyanide concentration in pH 10 system, 4100-ppm HtS.

TABLE 7—Effect of cyanide concentration in pH 10.0 test:

Conditions: 30-ml oil and 60-nil aqueous phase, with hydrogen sulfide held constant
at 4100 ppm. Inhibitors added at 100 ppm, based on total fluid. Test bottles
rotated 20 h at SO C.
Cyanide Added, ppm
Inhibitors 0 50 100 250 500 1000 2000 3000
None:
Weight loss, mg 4 5 5 6 10 22 40 56
Specimen appearance 1° 1 1 1 2 5 6 6
Inhibitor A:
Weight loss, mg 15 22 37 48
Specimen appearance 0 5 6 6
Inhibitor B:
Weight loss, rag 16 22 32 49
Specimen appearance 2 5 6 6
Inhibitor C:
Weight loss, mg 16 23 30 46
Specimen appearance 3 6 6 6
Inhibitor D:
Weight loss, mg 18 26 39 47
Specimen appearance 3 6 6 6
Inhibitor E:
Weight loss, mg 13 19 39 55
Specimen appearance 2 5 6 6
Inhibitor F:
Weight loss, mg 15 26 41 41
Specimen appearance 3 6 6 6
Inhibitor G:
Weight loss, mg 10 17 25 11
Specimen appearance 2 2 3 2
Inhibitor H:
Weight loss, mg 3 5 5 8
Specimen appearance 1 1 1 1
" Compare with Fig. 1 for degree of blistering.

Discussion
Distribution Experiments
Data in Table 2 show that cyanide is not extracted into the oil phase,
although a slight amount may be adsorbed on the glass vessel. The con-
stancy of the coefficient for hydrogen sulfide indicates that only the total
concentration need be considered.
Because of sulfide interferences with the cyanide titration, it was im-
possible to determine joint coefiicients.

Effect of pH on Hydrogen Sulfide Corrosion

These experiments show that the same sort of results are obtained in
the presence of oil as in water alone [31]. Some hydrogen blistering is
shown down to a pH of 6. There is still some corrosion at a pH of 8.

Effect of Hydrogen Sulfide Concentration

Here, only hydrogen sulfide was added to an oil water system. Hydro-
gen blistering increased with the hydrogen sulfide concentration. How-
ever, corrosion rates levelled oflf at approximately 1500-ppm hydrogen
sulfide and seemed to decrease slightly at still higher concentrations.
The water-soluble inhibitors were ineffective. Oil-soluble inhibitors
were, on the whole, very satisfactory in stopping this acid sulfide attack.

Intermediate pH {.7.7) Experiments

These solutions were buffered heavily to maintain a pH of 7.7. The
corrosion rates are not exceptionally high, and no hydrogen blistering
occurred. Only two of the oil-soluble inhibitors. Inhibitors A and C,
seemed to have any appreciable effect. At the present, no firm explanation
can be given for the better performance of Inhibitors A and C.
Filming amine type inhibitors are generally reaction products of mix-
tures of long-chain amines and long-chain acids. The resulting products
are complex mixtures which are very difficult to analyze.

EQect of Hydrogen Sulfide in Presence of Cyanide at pH 10.

In this high pH range, corrosion rates are not high. Hydrogen blistering
increases with the increase in hydrogen sulfide concentration.
The oil-soluble inhibitors have essentially no effect at this pH range.
The water-soluble, linear amine, type inhibitors seem to perform best.
It was noted that noninhibited coupons were water wet, while inhibited
coupons had hydrophobic surfaces.

Variation of Cyanide in the pH 10 Test

Hydrogen sulfide was constant at 4100 ppm. Corrosion and blistering
increased linearly with the cyanide concentration. This indicates that,
even in the presence of oil, cyanide is able to dissolve the sulfide from
the steel.
As in the case of variation of hydrogen sulfide, water-soluble Inhibitor
H was by far the most effective.

Condusions
The general conclusions are:
1. There does not seem to be any major differences in uninhibited
hydrogen sulfide corrosion in the presence or absence of oil, at least for
the nonaromatic oil used.
2. Oil-soluble inhibitors in acid systems generally inhibit both hydrogen
blistering and corrosion.
3. At the intermediate pH level, some inhibitors appear to work, al-
though the corrosion load is not high.

4. At pH above 8 in the presence of cyanide, corrosion rates can be

fairly high. Oil-soluble corrosion inhibitors seem to be completely in-
effective under these conditions, but the water-soluble materials tested
gave good results in decreasing both overall weight loss and hydrogen
blistering. The difference in effectiveness between oil-soluble and water-
soluble inhibitors may be a function of solubility, or of chemical properties
of the inhibitor molecules.

References
[7] Bartz, M. H. and Rawlins, C. E., Corrosion, SBIIA, Vol. 4, 1948, pp. 187-206.
[2] Schofer, N., Corrosion, SBIIA, Vol. 5, June 1949, pp. 182-188.
[3] Skei, T., Wachter, A., Bonner, W. A., and Bumham, H. D., Corrosion, SBIIA,
Vol. 9, 1953, pp. 163-172.
[4] Ewing, S. P., Corrosion. SBIIA, Vol. 11, 1955, pp. 497t-501t.
[5] Gardner, G. S., Faigen, H. L., Gibson, G. L., and Hall, W. S., Journal of the
Franklin Institute, JFINA, Vol. 262, 1956, pp. 369-384.
[6] Boies, D. B., Corrosion, SBIIA, Vol. 12, 1956, pp. 371t-375t.
[7] Biehl, I. A. and Schnake, E. A., American Petroleum Institute, Division of
Refining, PDRAA, Vol. 37 (III), 1957, pp. 129-134.
[8] Neumaier, B. W. and Schillmoller, C. M., Petroleum Refiner, PEREA, Sept.
1957, pp. 319-326.
[9] Roebuck, A. H., Gant, P. L., Riggs, O. L., and Sudbury, J. D., Corrosion,
SBIIA, Vol. 13, 1957, pp. 733t-738t.
[10] Bradley, B. W. and Dunne, N. R., Corrosion, SBILA, Vol. 13, 1957, pp. 238t-
2421.
[II] Meyer, F. H„ Riggs, O. L., McGlasson, R. L., and Sudbury, J. D., Corrosion,
SBIIA, Vol. 14, 1958, pp. 109t-115t.
[12] "Corrosion of Oil and Gas-Well Equipment," American Petroleum Institute,
Division of Production, APIPB, Dallas, 1958, pp. 26-39.
[13] Comeaux, R. V., Corrosion, SBIIA, Vol. 15, 1959, pp. 189t-193t.
[14] Shannon, D. W. and Boggs, J. E., Corrosion, SBIIA, Vol. 15, 1959, pp. 299t-
302t.
[15] Ehmke, E. F., Corrosion, SBIIA, Vol. 16, 1960, pp. 246t-252t.
[16] Gardner, G. S., Corrosion, SBIIA, Vol. 16, 1960, pp. 312t-318t.
[17] Hutchison, C. B. and Hughes, W. B., Corrosion, SBIIA, Vol. 17, 1961,
514t-518t.
[18] Nathan, C. C , Corrosion, SBIIA, Vol. 18, 1962, pp. 282t-286t.
[19] Hughes, W. B. and Stromberg, V. L., Corrosion, SBIIA, Vol. 19, 1963, pp.
9t-llt.
[20] Riggs, O. L. and Radd, F. J., Corrosion, SBIIA, Vol. 19, 1963, pp. lt-8t.
[21] Sardisco, J. B., Wright, W. B., and Greco, E. C , Corrosion, SBIIA, Vol. 19,
1963, pp. 354t-359t.
[22] Storey, W. D., Oilweek, OLWKA, 20 May 1963, pp. 48-53.
[23] Auer, L. and Hewes, F. W., Materials Protection, MAPRA, Aug. 1964, pp.
10-14.
[24] Beck, W., Glass, A. L., and Taylor, E., Journal, Electrochemical Society,
JESOA, Vol. 112, 1965, pp. 53-59.
[25] Brickell, W. F., Greco, E. C , and Sardisco, J. B., Corrosion, SBIIA, Vol. 21,
1965, pp. 48-52.
[26] Sardisco, J. B. and Pitts, R. E., Corrosion, SBIIA. Vol. 21, 1965, pp. 245-253.
[27] Sardisco, J. B. and Pitts, R. E., Corrosion, SBIIA, Vol. 21, 1965, pp. 350-354.
[28] Milton, C , Corrosion, SBIIA, Vol. 22, 1966, pp. 191-194.
[29] Greco, E. C. and Brickell, W. F., Materials Protection, MAPRA, Oct. 1966,
pp. 29-33.
[30] Chance, M. R., Materials Protection, MAPRA, April 1967, pp. 33-37.

[31] Gutzeit, J., Materials Protection. Dec. 1968, pp. 17-23.

[32] Hudgins, C. M., Materials Protection, Jan. 1969, pp. 41-47.
[33] Hausler, R. H., Goeller, L. A., Zimmerman, R. P., and Rosenwald, R. H.,
"Contribution to the Filming Amine Theory: Interpretation of Experimental
Results," paper presented at the National Meeting, National Association of
Corrosion Engineers Philadelphia, 2-6 March 1970.
[34] Bamett, L. M., Betz, Trevose. Pa., 24 March 1971, personal communication.

Pitting Corrosion-A Review of Recent

Advances in Testing Methods and Interpretation

REFERENCE: Bond, A. P., Titting Corrosion—A Review of Recent

Advances In Testing Metliods and Interpretation," Localized Corrosion—
Cause of Metal Failure, ASTM STP 516, American Society for Testing
and Materials, 1972, pp. 250-261.
ABSTRACT: There is widespread disagreement as to the applicability of
various test procedures in the study of pitting corrosion in alloy systems.
A literature survey has been made to identify areas of agreement and to
rationalize areas of disagreement, especially in the area of test methods,
such as potentiodynamic scanning to determine the breakthrough pitting
potential, potentiostatic tests, and other electrochemical tests. The variables
causing disagreement and the lack of reproducibility of test results are
discussed. The current status of attempts to formulate a detailed mecha-
nism for pitting corrosion in a given system is reviewed.
KEY WORDS: corrosion, stainless steel, tests, evaluation, pitting, aluminum,
aluminum alloys, polarization, electrochemistry

Pitting corrosion has long been a serious problem for metals that owe
their corrosion resistance to passivity. Localized breakdown in passivity
can lead to corrosion of very small anodes coupled to relatively large
cathodes, a situation that creates very high rates of penetration even when
the overall corrosion rate is not high. Since the phenomenon of passivity
is still not fully understood, it is not surprising that the localized break-
down of passivity resulting in pitting is even less well understood.
A great deal of experimental work has been devoted to the study of
pitting corrosion. Much data have been accumulated using immersion
tests, but a completely satisfactory test never has been developed. A given
immersion test cannot differentiate between the pitting resistance of metals
that do not pit in it and usually can give only a crude ranking among
those that do pit. In order to rank accurately the pitting resistance of a
group of alloys, a series of immersion tests of varying severity is required
in order that a pitting threshold can be established for each alloy. In order
to change the severity of the environment, the concentration of the ag-
gressive anion, the type or concentration of the oxidizing agent, or the
' Research supervisor, Research Laboratory, Climax Molybdenum Company of
Michigan (a subsidiary of American Metal Climax, Inc.), Ann Arbor, Mich. 48106.
250

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
BOND ON PiniNG CORROSION 251

temperature can be changed. An example of this approach is shown in

Table 1, in which the results give a relative ranking of the pitting re-
sistance of some rather resistant alloys [I].^ Tests of this type could be
rather time consuming and open to the objection that numerous changes
in the test environment are made. Immersion tests are also greatly in-
fluenced by edge effects. Pitting is much more severe at the edges for
most alloys, and, in many cases when edge pitting occurs and no pits are
formed away from the edges, it is impossible to know whether the edge
pitting protected the rest of the specimen or whether the major surface
really was resistant to pitting in the given environment.

TABLE 1—Results of exposure of S5Cr ferritic stainless steels

tolOFeCh-6H,0(Refl).
Test Corrosion Rate (mdd)°' of Indicated Alloys
Temperature,
deg C OMo 2Mo 3,5Mo 5Mo
25 40 nil nil nil
50 300 nil nil
70 400 nil
" mdd = milligrams per square decimeter per day.

In attempts to avoid these problems, several investigators [2-4] intro-

duced electrolytic methods. In this way, the potential difference between
the metal and the electrolyte can be varied at will without changing the
electrolyte or the temperature. Breakthrough potentials were determined
for various stainless steels by applying an external electromotive force
to a cell made up of the test alloy as anode and a nonpolarizable cathode
[2] or use of an arrangement such that the polarization that did occur at
the auxiliary electrode did not interfere with the measurements [3,4].
These measurements indicated that increased chromium or molybdenum
content improved resistance of stainless steels to pitting in agreement with
practical experience. However, electrolytic determination of breakthrough
potentials was not generally used by other workers. Streicher [5] used a
somewhat different approach in that he applied a potential sufficient to
cause pitting to initiate on all the alloys tested and counted the number
of pits formed under standard conditions. In the general case, interpre-
tation of the results obtained by his technique requires knowing how
overall pitting resistance would be expected to correlate with the number
of pits found in this type of test.
None of the electrolytic tests for pitting corrosion received wide ac-
ceptance until the electronic potentiostat came into wide use. Since that
time, measurements of pitting or breakthrough potentials have been made
' T h e italic numbers in brackets refer to the list of references appended to this
paper.

on a wide variety of metals and under a wide variety of conditions. A

typical determination of pitting potential is shown in Fig. 1. Unfortunately,
the great variety of materials and conditions has led to many apparent con-
tradictions. As a result, pitting corrosion has become one of the very con-
troversial areas of corrosion research. The purpose of the present paper
is to review recent work on some representative alloy systems in an effort
to identify areas of agreement and to rationalize some of the areas of
disagreement, especially in test methods.

600 MV/HR 60 MV/HR

-0.2 -0.1 0 0.1 0.2

POTENTIAL (V) VERSUS SCE

FIG. I—Polarization curve obtained for 17.6Cr-13.6Ni-1.99Mo austenitic stain-

less steel in argon-saturated 0.1 N HCl. The scanning rate was 600 mV/h from
—0.3 to 0.1 V; it was decreased to 60 mV/h from 0.1 V to the pitting potential
{Ref 31).

Aluminum and Aluminum Alloys

Several investigators have used controlled potential techniques to de-
termine pitting potentials of high-purity aluminum in various halide con-
taining environments [6-9]. The results obtained are shown in Table 2.
The various determinations agree remarkably well in spite of many differ-
ences in experimental technique. In some cases, the test specimens were
electropolished; in others, abraded and then etched in sodium hydroxide.
Various electrode mounting techniques were used, and no investigator
reported any necessity to take precautions against crevice corrosion. The
details of the polarization technique varied considerably, and there was
no evidence that variations in the rate of change of potential had any

TABLE 2—Pitting potentials determined on aluminum at 25 C.

Pitting Potential,
Electrolyte Reference V versus NHE»
HlOH-PlTRlTY A L
0.1 Af NaCl 8 -0.40
0.1 ilf NaCl 9 -0.46
0.5 M NaCl 7 -0.50
1 M NaCl 6 -0.48
1 M NaCl 9 -0.52
0.1 M NaBr 8 -0.29
l.OMKBr 9 -0.42
I.O Jf KBr 6 -0.85
1.0 M K I 0 -0.26
l.OMJLl 6 -0.20
99.4AL
0.1 M NaCl 8 -0.41
AL-2.4MG
0.1 Jf NaCl 8 -0.44
AL-0.2CU
0.1 M NaCl 9 -0.46
A L - 4 C U (aonJTioN TREATED)
0.1 M NaCl 9 -0.36
" NHE = normal hydrogen electrode.

large effect on the results obtained. It was, in fact, found that pit growth
could be stopped and started at will by changing the potential from 20
mV active to the pitting potential to 20 mV noble to the pitting potential
[9]. This process could be repeated many times on the same specimen.
Oxygen content of the test solutions was not reported to be an important
factor.
From these results, it is clear that there is a definite pitting potential
for aluminum which is readily measured reproducibly by diflferent investi-
gators (Table 2). The results obtained from immersion tests are in excel-
lent agreement with predictions made on the basis of pitting potentials.
When aluminum is undergoing pitting corrosion, it exhibits a corrosion
potential equal to the pitting potential in that environment, and when no
pitting corrosion is occurring, the corrosion potential is more active than
the pitting potential [7,8].
Apparently the pitting potential observed for aluminum is relatively
insensitive to purity or deliberate alloy additions (Table 2.) An ex-
ception is copper in solid solution, since it shifts the pitting potential in
the noble direction [9]. This is not necessarily beneficial to the corrosion
resistance of the alloy; in fact, according to Gavele and DeMicheli [9],
it leads to intergranular corrosion in aged aluminum-copper alloys. Inter-
granular corrosion results because the alloy becomes depleted in copper
as precipitation occurs at the grain boundaries and thus has a more active
pitting potential in the areas immediately adjacent to grain boundaries
than is the case for the areas of the grain faces.
The results on aluminum and its alloys indicate that other factors
besides pitting potential are extremely important in determining whether

or not pitting will occur. In many cases it is the cathodic polarization

behavior of aluminum that determines whether or not the pitting poten-
tial will be reached in an exposure test. For example, it has been found
that in relatively pure aluminum (99.99 and 99.999 percent) it is the
distribution of impurities, mainly iron and copper, that determines whether
or not pitting will occur in 0.5 M sodium chloride (NaCl) containing
0.3 percent hydrogen peroxide (H2O2) even though distribution of im-
purities has no effect on the pitting potential [7]. It was found that micro-
segregation of iron and copper reduced the cathodic overvoltage greatly,
allowing the pitting potential to be reached and pitting to initiate. That
copper and iron in aluminum reduce cathodic polarization has been well
established [9-77].
In summary, several investigators have demonstrated that a well-defined
pitting or breakdown potential can be determined experimentally for
aluminum and some of its alloys. This information, coupled with knowl-
edge of cathodic polarization behavior, has proved useful in interpreting
the corrosion behavior of aluminum.

Stainless Steels
In contrast to the rather well-defined situation for aluminum, at least
from an empirical standpoint, electrochemical studies of pitting of stainless
steels are fraught with many contradictions because of the pronounced
effect of many experimental factors on the results obtained. Unfortunately,
the importance of some experimental variables is only now being appre-
ciated.
As might be expected, surface preparation exerts a decided effect on
the resuhs obtained on stainless steel. It has been reported often that
resistance to pitting increases with the degree of mechanical polishing
[5,72,75], while electropolishing or etching leads to less resistance to
pitting [5] (Fig. 2). In a study of the effects of surface preparation on
pitting potentials of a series of iron-chromium alloys, Pessall et al [14]
showed that differences in surface preparation explain an apparent contra-
diction in the results of several workers [75-77] on this system. The
reasons for the pronounced effect of surface preparation are rather ob-
scure. One suggestion is that mechanical polishing covers grain boundaries
and inclusions with flowed metal. Pitting which was observed to initiate
preferentially at grain boundaries and inclusions may explain the effect
of mechanical polishing [5]. On the other hand, Standifer [18], as quoted
by Greene and Fontana [19], reported that pit sites were distributed
randomly on stainless steel except on application of "higher" anodic
current densities. In a study of surface films on stainless steel, Vernon et al
[20] found that the chromium content of the oxide film increased with
the degree of mechanical polishing. This effect could be important in

V Diamond
Surface Finish
(continuous scan)

ii- 1
100

E 300
Etched or 3/0
Surface Finish
(continuous scan)
200

• III Diamond Surface Finish

(scratch test)

V Etched Surface

-200
1 2
iWo Addition to Fe - 25 Cr at 7o

FIG. 2—Critical pitting potentials of Fe-Cr-Mo alloys in deaerated synthetic sea-

water at 90 C, comparing the effects of differing surface preparation and also com-
paring results from a continuous scan and from the scratch test (Ref 'il).

establishing the initial conditions in a pitting test. It has been noted also
that the influence of surface treatment is much less pronounced in tests
carried out at 80 C than it is at 20 C [75].
A rather surprising result is the rather large effect of gases dissolved in
the test solution on pitting potentials as shown in Table 3 [21]. These
results have been used [22] to explain the lack of agreement between
pitting potential measurements and exposure tests for pitting of zirconium
and Type 430 stainless steel reported by France and Greene [23]. Thus,

TABLE 3—Effects of dissolved gases on the pitting potentials of

some stainless steels in 1 M NaCl at S5 C {Ref 18).
Pitting Potential, V versus NHE,°
of the Indicated Steels
Gas Type 430 Type 304
Hs 0.06 0.19
Nj 0.11 0.22
A 0.14 0.30
Oj 0.21 0.31
' NHE = normal hydrogen electrode.

in carrying out exposure tests, it must be kept in mind that oxidizing

agents added to change the redox potential of the solution may themselves
have an effect on the pitting potential.
The rate of change of potential and the initial potential are of con-
siderable importance in the determination of pitting potentials of stainless
steels. Many investigators have reported that there is an induction time
that varies with chloride concentration and potential [24] for pit initiation
and that for this reason a continuous potentiodynamic scan overshoots
the pitting potential for 18 Cr-8 to lONi austenitic stainless steels by an
amount that increases with increasing scanning rate [25-29]. On the other
hand, other investigators working with ferritic stainless steels [30] or
molybdenum containing austenitic stainless steels [31] have reported that
the pitting potential is shifted in the active direction as the scanning rate
is increased. The latter effect could be explained by considering that, as
the potential difference between the bulk metal and the bulk solution is
increased, the passive film grows until the field across it has the same
gradient as it did before the change in potential difference. Thus, the
potential difference across the film/solution interface, which controls the
reaction between the solution and film, remains relatively constant. On
the other hand, if the potential difference is increased more rapidly than
the film can thicken, then the film/solution potential difference is in-
creased. Therefore, the faster the scan, the greater is the potential differ-
ence favoring film breakdown [30].
Depending on the relative importance of effects of induction time for
pitting breakdown and of film thickening, it appears that pitting poten-
tials determined by potentiodynamic scanning may either increase or de-
crease as scanning rate is increased. The observed effect probably depends
on surface preparation, starting potential, and the nature of the test alloy.
For these reasons, it has been suggested that potentiostatic techniques
should be used for measuring pitting potentials [27,28]. However, this
technique is very time consuming, since a separately prepared and acti-
vated electrode must be used to determine each point on the polarization
curve. Apparently, for this reason, the technique has been little used. A
novel experimental approach to this problem has been suggested by
deWaard et al [32] who used a wire specimen as the test electrode and
passed a constant current the length of the wire by means of a separate
auxiliary circuit so that IR drop creates a potential gradient within the
wire from one end to the other. In most cases 1000-mV gradients were
used. The potential at one point on the wire was controlled, with respect
to a reference electrode, with the aid of a potentiostat and counter elec-
trode so that the metal-solution potential difference varied over the length
of the test electrode in a known way, as long as the current flowing in
the electrolytic circuit controlled by the potentiostat was much smaller
than the current in the. auxiliary circuit involving only the wire. By noting

the position of any pits that form, the pitting potential under potentiostatic
conditions can be determined in a single experiment.
Another approach has been recommended by Pourbaix and co-workers
[33-35] who introduce the concept of the protection potential, that is,
the potential at which actively growing pits are stopped completely from
further growth. Thus, what actually is determined is a sort of hysteresis
loop for pitting with the most important value being the potential at
which the pits that were initiated at the breakdown potential repassivate
as the potential is scanned back in the active direction.
This hysteresis has been used by Wilde and Williams [22] to explain
the observation by Steigerwald [76] that the corrosion potential exhibited
by iron-chromium alloys was more active than the pitting breakdown po-
tential in an environment of equivalent chloride concentration. Wilde and
Williams [22] monitored the corrosion potentials of several alloys in
ferric chloride solution and found that the corrosion potential was more
noble than the pitting breakdown potential for a fairly short time, then
drifted to potentials more active than the breakdown potential, but re-
mained more noble than the repassivation potential as long as pitting
continued.
Determination of the protection potential appears to present some ex-
perimental difficulties. Once pits are initiated and begin to grow, the
solution within the pits becomes more concentrated in the aggressive ion,
and the pH shifts in the acid direction. In fact, it has been shown that
opening actively growing pits and allowing free circulation of the bulk
electrolyte into the pits causes these pits to stop growing in many cases
[36]. Therefore, if pitting is allowed to develop, the protection potential
that is determined must be characteristic of the solution within the pits,
not of the bulk electrolyte. Thus, the protection potential would be placed
at a more active potential than would be expected from the bulk solution
composition.
A technique recently described by Pessall and Liu [37] may make it
possible to determine what amounts to the protection potential without
first initiating actual pits. This technique consists of polarizing the test
electrode to the desired potential and then lightly scratching the electrode
surface. While the potential is held constant, the current is recorded as a
function of time after scratching until the electrode either repassivates or
stable pit growth is taking place. This process is repeated until the mini-
mum potential is found at which the electrode does not repassivate after
scratching. This value was reported to be reproducible and independent
of the surface preparation and prior history of the electrode. By elimi-
nating the initiation step in pitting, most of the factors that have caused
the lack of agreement among pitting potential measurements of several
investigators are also eliminated.
The results obtained by this method are, in general, qualitatively com-

parable to those obtained by the scanning technique as far as the effects

of alloy composition is concerned [33]. In fact, for some alloys, especially
those containing molybdenum, only small differences were observed in
the pitting potentials determined by the scratch test and by scanning tests
(Fig. 2). From a practical standpoint, the pitting potential measured in
this way should be a very useful guide to pitting corrosion resistance
since it takes into account the possibility of mechanical damage that would
be likely to occur to the passive layer under service conditions. The
scratch technique for determining pitting potential is very promising, but
more experience with results from different investigators is required be-
fore the method can be assessed completely.
Several investigators [17,24,32,34,35] have reported that anions such
as sulfate, nitrate, chromate, and chlorate inhibit pitting in chloride con-
taining solutions. Inhibition takes different forms with the different anions.
Sulfate moves the pitting potential to more noble values [24,34], nitrate
is only effective at more noble potentials [28,38], and chromate acts
mainly as a cathodic inhibitor [28], preventing attainment of the pitting
potential in various environments.
In general, pitting potential measurements obtained by all techniques
are in qualitative agreement with exposure tests and practical experience
as far as the effects of major factors on the pitting resistance of stainless
steels are concerned. Thus, increasing chromium and molybdenum content
and decreasing exposure temperature and concentration of the activating
ion, such as chloride or bromide, all contribute to greater resistance to
pitting as has been indicated by pitting potential measurements.
Other Metals
Electrochemical determinations of pitting potential have been made on
several other metals as summarized in Table 4. Several investigators re-
port reproducible results for zirconium [17,23,40,41], but it has been
reported also that the electrochemical tests do not correlate well with
exposure tests [23]. Too little information is available to indicate how
the results on the other metals in Table 4 compare with exposure tests.
Mechanism of Pit Initiation
No mechanism of the pitting of passive metals has been developed to
the point that it is capable of making quantitative predictions as to the
pitting resistance of any metal. In rough terms, most of the proposed
mechanisms involve either an activating ion, such as chloride, penetrating
the passive film and destroying it locally or the activating ion being prefer-
entially adsorbed at the metal solution interface displacing the adsorbed
passive layer and thus locally destroying passivity.
The mechanism involving breakdown of an oxide film has been further
subdivided by Hoar [501 into the "ion-migration" and the "mechanical"

TABLE 4—Pitting potentials of Zr, Mg, and Ni.

Pitting Potential,
Metal Electrolyte V versus NHE« Ref.
Zr 1 M HjSO^ + 1 M NflCl 0.4 SO
Zr 1 M NaCl 0.38 17
Zr 1 M KCl 0.4 36
Zr 0.« M NaCl 0.42 37
Mg 1 M NaCl -1.4 17
Ni 0.1 M NaCl 0.28 15
" NHE = normal hydrofjen electrode.

models. In the ion-migration model, the activating anion enters the oxide
film lattice without exchange [5,42]. This greatly increases the ionic con-
ductivity of the film, leading to locally high anodic dissolution rates and
pitting. In the mechanical model, anions adsorb at the oxide/solution
interface, lowering the interfacial energy to the point that cracks or splits
develop in the protective film under the influence of the electrostatic
repulsion of the adsorbed anions. In either model, it is assumed that the
initial step involves adsorption of the activating ion on the oxide surface
and that it is this step that is affected by the applied potential and the
presence of other anions that may inhibit pitting.
The direct adsorption model has been discussed by many investigators
[8,15,17,25,26,38]. This mechanism is able to explain qualitatively the
effects of potential and especially of inhibiting anions.
In both mechanisms, adsorption of anions plays a key role, but not
enough information on the energies involved is available for any quanti-
tative treatment to be made. However, based on another approach, Ver-
milyea [43] has developed a theory that allows a semiquantitative pre-
diction of pitting potentials. He assumes that the pitting potential is the
potential at which the protective metal oxide and the salt of the metal and
the aggressive anion are in equilibrium. To achieve this equilibrium, the
activity of the aggressive anion in the pit is increased by the potential
difference between the inside of the pit and the bulk solutions. By using
these assumptions and some approximations concerning the behavior of
concentrated solutions, the pitting potential can be estimated. Reasonable
agreement with experimental data was achieved for aluminum, magnesium,
iron, and nickel, but not for zirconium, titanium, or tantalum. This theory
does not predict which ions should be aggressive. It also appears difficult
to apply this treatment to alloys. However, it represents a promising ap-
proach that should receive a thorough test.
Conclusion
Great care must be taken in applying electrochemical techniques to a
study of pitting corrosion to ensure that the technique selected is appro-
priate to the metal under study and the results required. If the proper
technique is selected, electrochemical results are just as valid as carefully

accelerated exposure tests and, in most cases, much more convenient.

With the present state of knowledge, no laboratory technique can be used
to predict the exact corrosion behavior that will be encountered in a
complex practical installation. Although some progress is being made in
the study of the theory of pitting corrosion, empirical investigation will
remain of prime importance in this area for some time to come.
References
[7] Lizlovs, E. A. and Bond, A. P., Journal, Electrochemical Society, JESOA,
Vol. 118, 1971, p. 22.
[2] Uhlig, H. H. and Wulff, J., Transactions, American Institute of Mining and
Metallurgical Engineers, TAMMA, Vol. 135, 1939, p. 494.
[3] Brennert, S. J., Journal, Iron and Steel Institute, JISIA, Vol. 135, 1937, p. 379.
[4] Mahla, E. M. and Nielsen, N. A., Transactions, Electrochemical Society,
JESOA, Vol. 89, 1946, p. 167.
[5] Streicher, M. A., Journal, Electrochemical Society, JESOA, Vol. 103, 1956,
p. 375.
\6] Kaesche, H., Zeitschrift fiir Physikalische Chemie Neve Folge, ZPYFA, Vol.
26, 1960, p. 138 and Vol. 34, 1962, p. 87.
[7] Bond, A. P., Boiling, G. F., Domian, H. A., and Biloni, H., Journal, Electro-
chemical Society, JESOA, Vol. 113, 1966, p. 773.
[8] Bohni, H. and Uhlig, H. H., Journal, Electrochemical Society, JESOA, Vol.
116, 1970, p. 906.
[9] Galvele, J. R. and DeMicheli, S. M., Corrosion Science, CRRSA, Vol. 10,
1970, p. 795.
[10] Hansen, D. J. and Wetmore, F. E. W., Canadian Journal of Chemistry,
CJCHA, Vol. 34, 1956, p. 659.
lU] Ketcham, S. J. and Haynie, F. H., Corrosion, SBIIA, Vol. 19, 1963, p. 242.
[12] Smith, H. A., Metal Progress, MEPOA, Vol. 33, 1938, p. 596.
[13] Steensland, O., Corrosion Prevention and Control, CRPCA, Vol. 15, 1968,
p. 25.
[14] Pessall, N., Hull, F. C , and Liu, C , "Development of a Low-Cost Iron-Base
Alloy to Resist Corrosion in Hot Sea Water," Research and Development Re-
port No. 478, Office of Saline Water, Westinghouse Electric Corp., Pittsburgh,
1969.
[15] Horvath, J. and Uhlig, H. H., Journal, Electrochemical Society, JESOA, Vol.
115, 1968, p. 791.
[16] Steigerwald, R. F., Corrosion. SBIIA, Vol. 22, 1966, p. 107.
[17] Kolotyrkin, Ja. M., Corrosion. SBIIA. Vol. 19. 1963, p. 261t.
[18] Standifer, L. R., Jr., "A Study of the Relations between Micro and Gross
Solution Potentials and Susceptibility to Intergranular Corrosion of a 24S
Aluminum Alloy and Some Austenitic Stainless Steels," Ph.D. dissertation.
The Ohio State University, Columbus, 1952.
[19] Greene, N. D. and Fontana, M. G., Corrosion, SBIIA, Vol. 15, 1959, p. 25t.
[20] Vernon, W. H. J., Worrawell, F., and Nurse, T. J., Journal, Iron and Steel
Institute, JISIA, Vol. 150, 1944, p. 81 p.
[21] Wilde, B. E. and Williams, E., Journal, Electrochemical Society, JESOA, Vol.
116, 1969, p. 1539.
[22] Wilde, B. E. and Williams, E., Journal, Electrochemical Society, JESOA, Vol,
117, 1970, p. 775.
[23] France, W. D., Jr., and Green, N. D., Corrosion, SBIIA, Vol. 26, 1970, p. 1.
[24] Hoar, T. P. and Jacob, W. R., Nature. NATUA, Vol. 216, 1967, p. 1299.
[25] Leckie, H. P. and Uhlig, H. H., Journal, Electrochemical Society, JESOA,
Vol. 113, 1966, p. 1262.
[26] Leckie, H. P., Journal, Electrochemical Society, JESOA, Vol. 117, 1970, p.
1152.

Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 12:32:22 EST 2016
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
BOND ON PiniNG CORROSrON 261
[27] Schwenk, W., Corrosion Science, CRRSA, Vol. 5, 1965, p. 245.
[28] Herbsleb, G., Werkstoffe and Korrosion, WSKRA, Vol. 16, 1965, p. 929.
[29] Herbsleb, G., Werkstoffe und Korrosion, WSKRA, Vol. 17, 1966, p. 649.
[30] Hoar, T. P., Corrosion Science, CRRSA, Vol. 7, 1967, p. 341.
[31] Bond, A. P. and Lizlovs, E. A., Journal, Electrochemical Society, JESOA, Vol.
115, 1968, p. 1131.
[32] deWaard, C., Nicholson, J. W., and Posch, W., Werkstoffe und Korrosion,
WSKRA, Vol. 19, 1968, p. 782.
[33] Pourbaix, M., Klimzack-Mathieiu, L., Mertens, Ch., Meunier, J., Vanleugen-
Haghe, CI., DeMunck, L., Laureys, J., Neelemans, L., and Warzee, M., Cor-
rosion Science, CRRSA, Vol. 3, 1963, p. 239.
[34] Pourbaix, M., Corrosion Science, CRRSA, Vol. 5, 1965, p. 677.
[35] Pourbaix, M., Corrosion Science, CRRSA, Vol. 26, 1970, p. 431.
[36] Rosenfeld, I. L. and Danilov, I. S., Corrosion Science, CRRSA, Vol. 7, 1967,
p. 129.
[37] Pessall, N. and Liu, C , Electrochimica Acta, ELCAA, Vol. 16, 1971, p. 1987.
[38] Schwenk, W., Corrosion, SBIIA, Vol. 20, 1964, p. 192t.
[39] Uhlig, H. H. and Oilman, J. R., Zeitschrift fur Physikalische Chemie, ZPCFA,
Vol. 226, 1964, p. 127.
[40] Maraghini, M., Adams, G. B., Jr., and VanRysselberghe, P., Journal, Electro-
chemical Society, JESOA, Vol. 101, 1954, p. 400.
[41] Hackerman, N. and Cecil, O. B., Journal, Electrochemical Society, JESOA,
Vol. 101, 1954, p. 419.
[42] Hoar, T. P., Mears, D. C , and Rothwell, G. P., Corrosion Science, CRRSA,
Vol. 5, 1965, p. 279.
[43] Vermilyea, D., Journal, Electrochemical Society, JESOA, Vol. 118, 1971,
p. 529.

Relative Critical Potentials for Pitting

Corrosion of Some Stainless Steels

REFERENCE: Johnson, M. J., "Relative Critical Potentials for Pitting

Corrosion of Some Stainless Steels," Localized Corrosion—Cause of Metal
Failure, ASTM STP 516, American Society for Testing and Materials,
1972, pp. 262-272.

ABSTRACT: The breakthrough potentials of stainless steel alloy Types 409,

304, and HWT were determined in solutions containing various concentra-
tions of chloride ion, ranging from 10 to 40,000 ppm. Four different salts
were used to introduce the chloride ion: sodium chloride, calcium chloride,
cupric chloride, and ferric chloride. The breakthrough potential values
obtained decreased with increasing chloride ion concentration. These break-
through potential values were plotted against the logarithm of the chloride
ion concentration and a straight line drawn through the lowest potential
points for each alloy. This line described the minimum pitting potentials
for each alloy and was used to compare the three alloys studied. Using
this method, HWT proved to be a more pit resistant alloy than Type 304
which in turn was more pit resistant than Type 409.

KEY WORDS: corrosion, pitting, stainless steels, austenitic stainless steels,

polarization, anodic polarization

Since the design of the potentiostat and its initial use for corrosion
studies, technicians and particularly researchers have worked toward
establishing the potentiostat as a tool for rapid, reproducible testing of
the corrosion resistance of metals. In 1968, after considerable work by all
involved, ASTM Committee G-l/VI, Section I, issued a "Recommended
Practice for a Standard Reference Method for Making Potentiostatic
and Potentiodynamic Anodic Polarization Measurements."^ This method
is well accepted and widely used for general corrosion resistance deter-
minations. However, investigating and establishing the pitting susceptibility
of a metal by potentiostatic or potentiodynamic methods remain un-
defined.
Many investigators have used the breakthrough potential as a measure

' Senior research chemical engineer, Allegheny Ludlum Industries, Inc., Research
Center, Brackenridge, Pa. 15014.
' Designation: G 5-71, 1971 Annual Book of ASTM Standards, Part 31, American
Society for Testing and Materials, pp. 1047-1056,
262

Downloaded/printed
Copyright 1972 by by
AS FM International www.astm.org
University of Washington (University of Washington) pursuant to License
JOHNSON ON PITTING CORROSION OF SOME STAINLESS STEELS 263

of susceptibility to pitting [l-6f. Others have used current density differ-

ences taken at the same potential from anodic polarization curves, obtained
in acid solutions with and without chlorides [7]. Still others have investi-
gated and proposed the use of the ratios of current density values as a
method of evaluating the pitting susceptibility of metals and alloys [8].
For this investigation the breakthrough potential was used as a measure
of pitting susceptibility. Although no attempt is made to establish or
refute any of the methods mentioned above as valid or invalid, it is felt
that this investigation, by use of the breakthrough potential, establishes
the relative pitting susceptibility of three stainless steel alloys.

Procedure
Materials
The alloys investigated were Types 409, 304, and HWT. All specimens
were taken from commercial heats. The chemical compositions of the
heats used in this investigation appear in Table 1.

TABLE 1--Chemical compositions of alloys studied.

Weight Percent
Alloy C Cr Ni Ti
Type 409 0.049 11.50 0.20 0.53
HWT 0.041 18.55 0.30 0.70
Type 304 0.062 18.60 9.25 0.012

The specimens were sheared from 0.060-in.-thick strip which was in

the asannealed and pickled condition produced by the mill. No additional
preparation was used on the specimens except to degrease them in a hot
alkaline solution, rinse them thoroughly in water, and dry them.
Polarization Tests
The potentiodynamic polarization tests were performed using standard
equipment, except for the electrochemical cell, which was designed to
accept and expose 1 cm^ of a flat specimen.
The cell consists of a 2.5 in. diameter by 5-in.-high glass cylinder
which has been ground flat along its length about 1 in. wide. (Fig. 1). A
hole was blown in the center of the ground surface and the insides tapered
so that gases or heavy corrosion products would not be trapped. A Teflon
gasket with a 1-cm^ hole fit flat on the ground surface of the cell. The
Teflon gasket was machined flat except for a ring around the 1-cm^ hole,
which was left raised Yic in. The specimen was exposed by placing it flat
on the Teflon gasket and pressing it with a clamp. This arrangement is
^ The italic numbers in brackets refer to the list of references appended to this
paper.

Samfil*

Holder €•« T«flon G«»lt«t

FIG. 1—Cell assembly and clamping system used for potentiodynnmir polariza-
tion tests.

very similar to the Stem-Makrides specimen holder specified in the ASTM

Recommended Practice, the only difference being the much larger Teflon-
to-specimen contact area [9]. Electrical contact to the specimen was made
externally.
The platinum counter electrode, salt bridge, and saturated calomel
electrode were of the usual design. A Wenking potentiostat was used in
conjunction with a two-way recorder.
All anodic polarization curves were conducted automatically at a
scanning speed of about 0.7 V/h. The solutions were used "as mixed,""
without any aeration or deaeration, and at a temperature of 72 F ±: 1.
The pH of the solutions was not adjusted, except where the effect of pH
was being investigated. The specimens were allowed to remain in contact
with the solution for Vi h before the curve was started from the corrosion
potential.
For the potentiostatic holding experiments, the desired potential was
set before the specimen was mounted and the electrolyte introduced. After
the experiment was prepared, the potential was introduced and held con-
stant for 72 h. Subsequently, the specimens were examined for evidence of
pitting by a low power microscope.

Results and Discussion

Typical potentiodynamic polarization curves obtained in this investiga-
tion appear in Fig. 2. The lines and arrows point at the potentials where
a rapid increase of current occurred with very litde change in voltage. This
is considered an indication of a breakdown of the passive, corrosion re-
sistant, stainless steel film, and initiation of corrosion, which in this case
is in the form of pitting.

1.0

g •"

-.2 I r I mini I I I mill i . • •mil I 1 llliul L. uL ' ' """I I I I mill

0.001 0.01 0.1 1.0 , 10 100
CURRENT DENSITY (mA/cm'^l
FIG. 2—Potentiodynamic polarization curves of Type 304 in various chloride ion
concentrations using NaCl as the salt.

Occasionally crevice as well as pitting corrosion occurred on the speci-

men. The breakthrough potential in this case occurs at a lower potential,
and the current increase is slower, until the pitting breakthrough potential
is reached (Fig. 3). Then both curves essentially merge and form similar

1.0

I I I lllirl I I I mill I I iiim I I I mill ' • • ' • "

0.001 0.01 0.1 1.0 10.0 101
CURRENT DENSITY ImA/cm^l
FIG. 3—Potentiodynamic polarization curves of Type 304 in chloride solution
showing a normal breakthrough (A) and a crevice initiated breakthrough (B).

slopes. It has been shown that crevice corrosion does indeed occur at
much lower potentials than pitting. Lizlovs indicated that Type 304 would
crevice corrode at 0.00 V saturated calomel electrode (SCE) in 1 JV so-
dium chloride (NaCl) [10]. When curves indicating crevice corrosion were
obtained, the test was repeated until duplicate pitting breakthrough po-
tentials were obtained on each alloy in each solution.
The results obtained are plotted against the chloride ion concentration
in the graphs shown in Figs. 4-6. As would be expected, the breakthrough
potential values are decreased as the chloride ion is increased. This was
true regardless of the type of salt used to introduce the chloride ion.
There was no apparent effect of the cation on the breakthrough potential,
except possibly on the Type 409. where the breakthrough potentials in
the NaCl are generally about 0.1 V (SCE) above the lowest breakthrough
potential obtained in any other salt solution. Breakthrough potentials in
NaCl on Type 304 resulted in a similar effect, but only in the hieh-
chloride ion concentration.
There is considerable scatter in each concentration. This is attributed
somewhat to the nature of the test and mostly to the nature of pitting
corrosion. The tests were conducted potentiodynamically, and, although
many potentiostatic tests were conducted before the potential change
rate of 0.7 V/h was considered adequate, the test method remains dy-
namic, and the breakthrough potential could be "overshot." However,
such overshooting and scatter can occur even in potentiostatic tests, where
5, 10, and even 60-min holds are used. The nature of the initiation of

ppm C.
FIG. 4—Effect of chloride concentration on the breakthrough potential of Type
409 in various chloride solutions: Q NaCl, A CaCU, D CuCU, O FeCL (*72-h
hold test—no pitting.)

1.0

_i • ' '
ppm •
''
i^m^
FIG. 5—Effect of chloride concentration on the breakthrough potential of Type
304 in various chloride solutions: Q NaCl, A CaCl,. D CuCh. O Fed.,. (*72-h
hold test—no pitting.)

pitting on stainless steels is even more of a problem, particularly at the

low-chloride ion concentrations studied. It is not unusual to observe only
one pit on a 4 by 6-in. stainless steel panel exposed to a mild chloride
containing environment. This one spot would then be more susceptible to

i.ocr

-I I ] I I 11 L -
100 1000 10,000
ppm c r
FIG. 6—Effect of chloride concentration on the breakthrough potential of HWT
in various chloride solutions: O NaCl, A CaCl,, D CuCU, <> FeCh. (*72-h hold
test—no pitting.)

chloride attack, and, if tested as in this investigation, it would result in

a lower breakthrough potential than the l-cm^ area adjacent to it.
Thus, by testing over 40 cm- in various chloride ion concentrations and
salts, a trend which is not affected by "scatter" is established. The trend
is shown in Figs. 4-6 where a line is drawn through the minimum break-
through potentials.
The "minimum pitting potential line" describes the potentials below
which pitting should not occur in the particular alloy. This was tested
by holding at the potentials shown on the graphs, at the chloride ion con-
centrations indicated, for 72 h without observing any pitting. The potentials
and results obtained are listed in Table 2.
The slopes of the minimum pitting potential lines were different for
each alloy. There is a shift in the active direction by about 0.17 V (SCE)
for a tenfold change in concentration for Type 409, 0.16 V for Type 304,
and 0.23 V for HWT. The 0.16 V for Type 304 is considerably higher
than the 0.09 V reported by Leckie and Uhlig [5] for Type 304 in NaCl
solutions. However, if only the NaCl breakthrough potentials are con-
sidered (Fig. 5), and a minimum pitting potential line is drawn through
the higher chloride ion concentrations (the concentrations investigated by
Leckie and Uhlig), the slope obtained, 0.08 V, is in excellent agreement
with that observed by these investigators.
The different slopes of the minimum pitting potential lines for the dif-
ferent stainless steel alloys is not unusual. Hospadaruk and Petrocelli [4]
reported the pitting potential of the austenitic steels Types (201 and 301)
to be practically independent of chloride concentration in the range of 0.1
to 5.0 N NaCl, while that of Type 434 decreased linearly with the in-
crease in log [Cl~].
It is evident by comparing Figs. 4-6, that the minimum pitting potential
line is characteristic of the alloy and its resistance to pitting in chloride
environments. Using this criterion for evaluating alloys, this investigation
indicates that Type 304 is more pit resistant than Type 409, and alloy
HWT is more pit resistant than both Types 409 and 304 (Fig. 7).
The effect of pH on the breakthrough potentials of these alloys was
also investigated. Sodium chloride was used to mix solutions containing
10, 100, 1000, and 10,000-ppm C1-. The results obtained are shown in
Figs. 8-10.
As reported by other investigators [4,5], the breakthrough potential was
not greatly affected by pH in the range of 3 to 8 for alloys Types 409
and 304. At pH values below three, the effect was different for the low-
chloride solutions than that with high-chloride concentrations. In the 10
and 100-ppm Cl~ the breakthrough potentials increased, while in the
high concentrations the breakthrough potentials decreased. In general,
increasing the pH above eight tended to increase the breakthrough poten-
tials. No breakthroughs and no pitting was observed on any of the alloys

ja ja
en
s
o
9
O
-a
O
4^ J3 pa
•3 „ •3 ^
"3 M eo *J 60 ^J tC ^ ee
m M C ji a M B .M S

ti ti ti
xi'a, ^ a. ^'a. ^ a>
§1
o o o o
a a

II 00 i-H

I I I I
Si-
o o
a a
ll II if
o o
a a
o o

o eo
o £
o d

liliiill
^ a l-U i-a
20 os o o o o
B B

" O o
I

B
O
I
13 08,.
o

n 0 S
0< a
s

"10 100 1000 10,000

ppmcr
FIG. 7—Minimum pitting potential lines oj Type 409, Type 304, and HWT.

in solutions adjusted to pH 12. This is in agreement with known informa-

tion that stainless steels do not pit in strong alkaline solutions.
The breakthrough potential values obtained on each alloy in calcium,
cupric, and ferric chloride solutions for estabUshing the minimum pitting
potential lines were also plotted in Figs. 8-10 (filled symbols). The pH's
of these solutions were not adjusted. Note, however, how well these

FIG. 8—Effect of pH on the breakthrough potential of Type 409: open symbols,

NaCl; filled symbols, ( i ) CaCl,, (2) CuCl,. and (3) FcCl,.

FIG. 9—Effect of pH on the breakthrough potential of Type 304: open symbols,

NaCl; filled symbols, (1) CaCl,, (2) CuCl., and (J) FeCl>.

breakthrough potential values fit the curves obtained with the adjusted pH,
sodium chloride solutions.
The effect of pH on the breakthrough potentials of HWT is somewhat
anomalous (Fig. 10). Although the curves for 10, 100, and 10,000-ppm
CI- solutions follow the general trend described for Types 409 and 304,

FIG. 10—Effect of pH on the breakthrough potential of HWT: open symbols,

NaCl; filled symbols, ( i ) CaCl,, (2) CuCU, and ( i ) FeCU.

the values obtained in 1000-ppm Cl~ solutions are "erratic." No attempt

will be made to explain this behavior at tins time.
Conclusions
Anodic polarization tests in chloride containing solutions have been
described in the literature as methods for measuring pitting susceptibility.
Such tests, however, result in substantial scatter due to the nature of the
test and the nature of pitting corrosion. Thus, in comparing the pitting
resistance of two alloys, it is conceivable that an erroneous choice could
be made.
Conducting anodic polarization tests in solutions containing various
chloride ion concentrations, a minimum pitting potential line may be
drawn for each alloy. The minimum pitting potential line is characteristic
of the pitting resistance of the alloy and may be used to compare alloys
in general or in some particular chloride concentration. Using this method
of evaluation, HWT proved to be a more pit resistant alloy than Type
304 which, in turn, was more pit resistant than Type 409.
There is very litde effect on the breakthrough potentials of Types 409
and 304 in the 3 to 8 pH range. The effect of lower pH solutions depends
on chloride ion concentration and at higher pH solutions the breakthroughs
increase. More data are needed to fully establish the effect of pH on HWT.

Acknowledgment
The author wishes to express his appreciation to James Fenoglietto for
his assistance in conducting all experimental laboratory work.
Refnences
[1] Nielsen, N. A. and Rhodin, T. N., Jr., Zeiischrift Electrochemie, ZEELA,
Vol. 6 No. 2, 1958.
[2] Kolotyrkin, J. M., Corrosion, SBIIA, Vol. 19, Aug. 1963, p. 261.
[i] Schwenk, W., Corrosion Science, CRRSA, Vol. 5, 1965, p. 245.
[4] Hospadaruk, V. and Petrocelli, J. V., Journal, Electrochemical Society, JESOA,
Vol. 113, 1966, p. 878.
[5] Leckie, H. P. and Uhlig. H. H., Journal, Electrochemical Society, JESOA,
Vol. 113, 1966, p. 1262.
[61 Horvath. J. and Uhlig, H. H., Journal, Electrochemical Society, JESOA, Vol.
115, Aug. 1968, p. 791.
[7] Walker, M. S. and Rowe, L. C , Corrosion, SBIIA, Vol. 25, Feb. 1969, p. 47.
IS] Greene, N. D. and Judd. G., Corrosion, SBIIA, Vol. 21, Jan. 1965, p. 15.
\9\ Stem, M. and Makrides, A. C , Journal, Electrochemical Society, JESOA, Vol.
107. 1960. p. 782.
[10} Lizlovs. E. A.. Journal, Electrochemical Society, JESOA, Vol. 117, Oct. 1970,
p. 1335.

Exfoliation Corrosion Testing of 7178

and 7075 Aluminum Alloys*

REFERENCE: Ketcham, S. J. and Jeffrey, P. W., "Exfoliation Corrosion

Testing of 7178 and 7075 Aluminum Alloys," Localized Corrosion—Cause
of Metal Failure, ASTM STP 516, American Society for Testing and Ma-
terials, 1972, pp. 273-302.
ABSTRACT: As special heat treatments to improve resistance of aluminum
alloys to exfoliation were developed, the need for accelerated testing
methods to control production heat treatment arose. A joint task group
formed by ASTM Committee GO 1.05 and the Aluminum Association has
conducted an interlaboratory testing program of various salt spray and
immersion tests for 7000 series alloys. A constant immersion test produced
the best results.
KEY WORDS: corrosion, exfoliation corrosion, corrosion tests, salt spray
tests, pitting, immersion tests (corrosion), aluminum alloys, etching
The development of heat treatments for 7075 and 7178 aluminum alloys
to produce resistance to exfoliation has created the need for a test method
sensitive enough to differentiate between susceptible and resistant mate-
rials and also capable of detecting borderline cases.
Descriptions of three types of salt spray tests and one total immersion
test developed for this purpose have been published [i-i].* Lack of agree-
ment among producers and users of the exfoliation resistant tempers as
to the best test method prompted the formation of a task group in ASTM
GO 1/05 which in cooperation with the Aluminum Association conducted
an interlaboratory testing program of several test methods. The object of
the program was to prepare an ASTM recommended practice for the
method which in the opinion of the participants was the most satisfactory
for the purpose.
Initial effort was concentrated on two salt spray tests. It was later
expanded to include three total immersion tests and corrosion potential
measurements in an organic electrolyte [4].
* Report of an ASTM GO 1/05 Interlaboratory Testing Program in cooperation
with the Aluminum Association.
^ Naval Air Development Center, Warminster, Pa. 18974.
'Alcan Research and Development Ltd, Kingston, Ontario, Canada.
' T h e italic numbers in brackets refer to the list of references appended to this
paper.
273

Downloaded/printed
Copyright 1972 by by
A S T M International www.astm.org
University of Washington (University of Washington) pursuant to License
274 LOCALIZED CORROSION-CAUSE OF METAL FAILURE

Description of Matoials
7275 Alloy Plate
Specimens 3 by 6 by 0.375 in. were supplied by the Alcoa Research
Laboratories for the salt spray tests in the as-machined condition. The
specimens had been step-machined along the rolling direction to expose
both the T/10 and T/2 surfaces on a single specimen. The T/10 suriface
refers to the plane exposed when one tenth of the thickness is removed
from the plate; T/2 when one half of the thickness is removed.
Triplicate V4 by V4 by 2-in. long transverse specimens were supplied
from each of the above lots for corrosion potential measurements in an
organic electrolyte.
For the immersion tests, 3 by 3-in. stepped panels were supplied, but,
due to the possibility that in an immersion test one step might cathodically
protect another, separate panels IVi by 3 in. were also supplied. These
were taken from the T/10 and the T/2 planes. All were in the as-
machined condition.

7075 Alloy Extrusion

Three by three-in. stepped panels were machined from three Vi by
6-in. extrusions supplied by Harvey Aluminum Co.; also VA by 3-in.
panels with the T/10 and T/2 planes exposed; and VA by ?4 by 2-in.
specimens for potential measurements.

2024 Alloy Plate

Specimens of 2024-T42 alloy 2 by 5 by 0.500 in. were also supplied
to be used as controls for the corrosion tests. These were machined to
expose the T/2 plane and had been heat treated to have a low resistance
to exfoliation corrosion. Heat treatment conditions for these materials
are given in the Appendix.

Experimental Procedures
Cyclic Salt Spray Tests (first phase)
An acidified 5 percent sodium chloride (NaQ) solution (MASTMAA
SIS*) [1] and an acidified (approximately) 5 percent synthetic sea salt
solution (SWAAT*) [2] were evaluated for their efficacy in assessing
susceptibility to exfoliation. The operating conditions prescribed for the
two salt spray tests are given in the Appendix. For the SWAAT test all
participants maintained wet bottom conditions in the cabinet, for the
MASTMAASIS, three operated with dry bottom conditions.

* MASTMAASIS is an acronym for Modified ASTM Acetic Acid Salt Intermittent

Spray; SWAAT for Seawater Acetic Acid Test.

Immersion Tests (second phase)

Solutions used for the immersion tests were:
1. NaQ-KNOs-HNOs (EXCO).«
2. NHiCl-NHiNOa-HaOo."
3. NaCl-S02[i].
Operating conditions for these tests are given in the Appendix.
Specimen Surface Preparation
Both degreased and etched specimens were used in the 7178 program.
In an earlier preliminary testing program there had been considerable
variation in results which was attributed to surface effects. Etching the
surface was proposed as a way to provide a more uniform surface. It
seemed worthwhile to the task group members to include the two surface
preparations at this point in the interlaboratory program to determine
whether etching was necessary to obtain better reproducibility of results.
Acetone or benzene was used for degreasing. Etching was accomplished
by a 1-min immersion in 5 percent sodium hydroxide (NaOH) at 82 C,
a dip in concentrated nitric acid (HNOs), and rinsing.
For the 7075 program etching was used throughout. Stepped panels
were not used except in two cases. At that stage, however, it was decided
to determine whether there were any galvanic effects between the back
of the specimens which had the as-extruded surface and the machined
surface. Therefore, specimens with the sides and backs masked with elec-
troplaters tape were included as well as unmasked.
System for Rating Degree of Exfoliation Attack
In order to compare results of the exfoliation tests among laboratories
it was necessary to have a precise rating system. The system finally adopted
is given in Table 1 and illustrated in Fig. 1. The questionable category
refers to a condition where raised blisters or pit-blisters are visible which
could be due to either mild exfoliation attack or severe pitting. Metallo-
graphic examination would be required to determine which type of attack
had occurred. In evaluation of heat treatments or materials metallographic

TABLE 1—Vuual examination.

Code Explanation
N no appreciable attack; surface may be etched or discolored
F pitting
Q questionable exfoliation (pit blisters)
EM exfoliation, mild
EI exfoliation, intermediate
ES exfoliation, severe

"EXCO is an acronym for Exfoliation Corrosion Test. The test was developed
by Alcan. The solution was used originally for stress corrosion testing [5].
•See p. 38.

examination would be desirable in questionable cases. The purpose of

this task group however was to compare test methods, and an important
criterion for final selection of a test method was that the test produce the
fewest visually questionable ratings. Therefore, metallographic examina-
tion was not conducted on specimens with questionable ratings.
Corrosion Potential Measurements
A test method involving measurement of corrosion potentials in
methanol-carbon tetrachloride (CH3OH-CCU) electrolyte to predict
exfoliation resistance of 7000 series alloys has been described [4]. The
method is very quick and simple and would make an ideal production
control test. This method was included in the testing program in order
to compare results with those from the exfoliation tests.
The method for conducting the test is given in the Appendix.

Results
5a// Spray Tests
Results of the two salt spray tests are given for 7178 plate in Table
2. On the first lot of low-resistance plate material, both tests gave results
ranging from pitting only through all degrees of exfoliation. However, the
SWAAT gave more reproducible results from laboratory to laboratory.
In the majority of cases, the degree of exfoliation produced by the SWAAT
test was more severe than that of the MASTMAASIS, particularly on the
T/2 plane.
On the low resistance lot 1 plate specimen, six out of eight laboratories
reported exfoliation on the T/10 plane with the SWAAT test; only two
with the MASTMAASIS. On the T/2 plane both tests produced exfoli-
ation at all eight laboratories, but ail eight specimens showed severe
exfoliation in the SWAAT test, whereas all degrees of exfoliation occurred
in the MASTMAASIS.
On the low-resistance lot 2 plate material seven out of eight laboratories
reported exfoliation on the T/10 plane with the SWAAT test, only three
with the MASTMAASIS; on the T/2 plane all eight obtained exfoliation
with the SWAAT test, five out of eight with the other test. Again the
degree of exfoliation was greater in the SWAAT test.
On the high-resistance material, both tests produced either pitting or
a questionable rating with the exception of one case of exfoliation reported
on the T/2 plane.
Figures 2, 3, and 4 show representative examples of these results.
Figures 2 and 3 illustrate the variability obtained with the MASTMAASIS
test by presenting results from two different laboratories. Litde or no
exfoUation is evident in Fig. 2 (Laboratory A) while a greater degree
of exfoliation is shown in Fig. 3 (Laboratory B).

•0
IV
JS ' P H C ^ -(IH <yft,cy :<y : : :
u
•^
EH • ' • ta
•<

<
^
cc
H
-a
(0
a
at
•Of -PHPHPU
ll : : :S :
s h

O. o
wo
• • • « .fLiaPM

no • ^

tui
H S b a h

o. o
CMb • PHQiri DH OriPH

!/l«t» .CC . . . cs
aa H

:a3 Itctfti-i "

H
I
J,
" a w • W OH PH » W « •« " aw iwawww-^w

:S„ : 0 WWW .CO . . . a

•WW •O'
0 WWW -W • ' ' J
'.5
d '.^
i^ ^ .2
0 JJ "o «M
•M Ji X
X -w
M . .CC .1/3 l-l V O : : '.w Imcoco "
W • -W ' W P M W <0 <K • • .W -WWWoi

• 9
H
o S
5^
« o l-HtC :i-lScCl-lM " °
aw • asHfew<y<»«» WW - W W W W W o ^

•osnufiwwon a <:nufiwwos a
e a9
9 W

fr-

1
I
I
1
.?

I
i

A011V 000/

i
X
O

I
z
u
z
Z « ?
< d
M •

i
«» iM
at <

O JO •i
R
u
z< 1
I

o i
V •! ••
Z S! •»
< 2 X
•- 2 'J
S Ik Z
•• ^ ~
Ml ^ e z
"o g
<
S"5

The salt spray results for 7075 are given in Table 3. The data for
7075 are less complete than that for 7178 because by the time the 7075
extrusion was tested primary interest of the task group had shifted from
salt spray tests to immersion tests.
The two surface preparations, degreasing or etching, did not appear
to have any significant effect on the results.
Results of the salt spray tests on the 2024 specimens are not presented
because this alloy was not of prime interest in this program. It is worth
mentioning however that the MASTMAASIS test produced better repro-
ducibility and more severe exfoliation on 2024 than did the SWAAT
test, the reverse of the results on the 7000 alloys. The appearance of
some 2024 specimens is shown in Figs. 2, 3, and 4,

Immersion Tests
7178 Plate Material—Results obtained by the various laboratories in
the three immersion tests are presented in Tables 4, 5, and 6 and Figs.
5, 6, and 7.
Low-Resistance Materials (Lot 1) {Table 4, Fig. 5)—On this lot of
material the EXCO test produced some degree of exfoliation on the T/10
plane in six out of seven laboratories; on the T/2 plane all laboratories
reported severe exfoliation.
With the ammonium chloride-ammonium nitrate-hydrogen peroxide
(NH4Q-NH4NO8-H2O2) test seven out of eight laboratories obtained
some degree of exfoliation on the T/10 plane; on the T/2 plane all labora-
tories reported exfoliation, but the degree varied from mild to severe.
With the sodium chloride-sulfur dioxide (Naa-S02) test five out of
six laboratories reported exfoliation on the T/10 plane; all six reported
either intermediate or severe on the T/2 plane.
On this low-resistance material the effects of pretreatment and of
whether the planes were tested separately or as a stepped panel seemed
negligible.
Low-Resistance Material {Lot 2) {Table 5, Fig. 6)—On the T/10
plane of this lot four out of seven laboratories which conducted the EXCO
test reported some degree of exfoliation. On the T/2 plane all seven
laboratories reported some degree of exfoliation.
The NH4CI-NH4NO8-H2O2 test on the T/10 produced a variety of
results ranging from no appreciable attack to intermediate exfoliation. On
the T/2 plane all six reported exfoliation, either intermediate or severe.
With this lot of material it was again difficult to ascribe any significant
effect to pretreatment or to testing the two planes separately or as a
stepped specimen.
High-Resistance Material {Table 6, Fig. 7)—On the high-resistance
7178 all laboratories reported either pitting or no attack in all three solu-
tions on the T/10 plane (with one exception, from the NaCl-S02 test).

. J3 ;6^ ..a
H a ee
o a
o
s a
a-
I
. -Of . ' P H ' ..o
be <9
s a
•s
•s
.2 w
•a S£•5e
. s.ti
9<
lofc :
a

I. :& :& : : :
'» •w • • •

•S

o
a
5
:w :
I :w : :PM : •c
• 0
H
1•5
"o
• M
M

^ ?< *>
S 2^ •: ' V •ft

I, • » »fA i« F^

:& : .t/3

1
PQ
<
:H :» :H
H
o

•<! CQ U c m b, O US S •<pQuaHi^os a C/3

s a
CO
9

:S : "S .1-1
a •« c •«
» eS
OS

V
a. ft 0.
a v
•H

:S :
•H • •H

»
'•Z» • HHHE4 . en

w e wccS
H •PH.2 HHH
eS

0 S
V O
1 HHCCS
1 WHH
•w^ pa
Hi n
P2 <

.U3
•w

fH
.CfitBCCM .03

• ".sg

•WS

•vieauaH i<(i3n g •oiPQUQH p^on a

s 3

cS
»»
t-
V
Ji h ; :S •s CO I IM !

I1
cd
*«— H•«-u> D.
01
-H
•« H
• '^ •§
!/2 1 3 ^
4; tn
H ft"<5 •S.2
It
T3
OJ sf a o
d
!/J ^a
C3
CO : ;? h-( to :
sCS
bo
a> o
CC
w • :'^<3 : C4 * »» » w • • ia&dP4t.
Y-, Q

! I—I S i-H M !t-H

•W!Z;HW •H • WWWW -H .

Pd
* HH S ^ >H .' S t-H
• cdS?;c!'PH ^I^IOf

fclzS!; H
•^ c g^2 : : > - < : «
•;2;.2 www • • W -(li.S
-»-'
OS 6 ^-.
• » ^

d 0

OS
X •4-' IN •S.ti
v o u o<
00 0« 1 SwS ! : :Si-ig'-
'•o ^ 1 WWW • • ww"^
w
H

.'ScocO .MCCCOl—I .to

• HHHh •w .WWWW •W

H
IcototcS .to
•wc?wcy • WWWW •W

SS« : : : : :e 1—1 w t o ; .' ; : :a

www • • • • -.2 WWW • • • • -.2
'.*3
l"-^

M
's03 .2
•4M rJ3
S" "o
«M
^(V •< * 3
« O •*
<x
x
a>
SS.H : : • • o» « i cci-ico : : '• '•-'i^
www • • : r^'-' www • • • -w"

<JP50QW won a «(<moQw w o M a

a
s
a
en to

c* ci
•%
4-)
0.
• 1 .
' ;; •H-t
en M "oo S '^'S' c
4J^ 6J: C
H
13 o. c •- «
d en
V
e6
a:1 «S
a d, .Qf C '••s •
a
ca V V
w
'A 0

:ii,;z;pHfc

• IS
'biZPHPu -Pia. •ft.O'

fcZ^;

II
•£§
•+ I -
5=
PH^IZ

•<

:ii.;z;pL,h iftnd.O'O' -a.

:a.;zpHBH .(i.fciO'e, -ft.

<«
CSHBH
J<!
o
M 5
cS ^
j ;•^
•5 « •*
.•S o o<
ac ^^ C. D-
ft,p-z OHM-V P^PMOH

<PCUAH
S

*'
.4.

^ ^ • • . • ' • ! * ^ | , . , i . ' ' .'. ''^^^

* •

1
I
J • • < * -
o
si
I
i

t
'-.V.
?s

£ X-
% • • '

•m^

i
a.

R
* i
^1

!
I
.1

-4
"":;¥'

J}
E

: 1 I* *^at>)"S^*--?^s

- ft--.

On the T/2 plane either pitting or questionable ratings were reported

from all three tests with the exception of mild exfoUation from the NH4CI-
NH4NO3-H2O2 test reported by one laboratory.
7075 Extrusion—Results of the three immersion tests on 7075 are
presented in Tables 7, 8, and 9. Since the results on masked and unmasked
specimens were similar, they will be treated together in the following
discussion.
Low-Resistance Material (Lot 1) {Table 7, Fig. 8)—The EXCO test
provided the most severe attack and the best reproducibility in results of
the three solutions. On both the T/10 and T/2 planes all laboratories
reported severe exfoliation with one exception and that was intermediate.
Results with the NH4Cl-NH4NOa-H202 test ranged from questionable
to severe. With the NaCl-S02 test mild and intermediate exfoliation were
reported.
It is interesting to note that very little difference in exfoliation suscepti-
bility was evident between the T/10 and T/2 plane of the low-resistance
7075 extrusion as opposed to the low-resistance 7178 plate (Table 4).
Low-Resistance (Lot 2) (Table 8, Fig. 9)—Again, most severe ex-
foliation was evidenced with the EXCO test, and reproducibility in re-
sults from laboratory to laboratory was very good; all reported either
severe or intermediate exfoliation. The NH4CI-NH4NO3-H2O2 solution
again gave results ranging from questionable to severe exfoliation; the
NaCl-S02 solution either mild or intermediate.
High-Resistance Material (Table 9, Fig. 10)—^With the exception of
two laboratories, all participants reported either pitting or a questionable
rating on the high-resistance material, pitting being predominant.
Corrosion Potential Measurements—Potentials measured in the CHs
OH-CCI4 electrolyte are presented in Table 10. Reproducibility of values
among the laboratories was excellent. The method clearly distinguishes
between material of low and high susceptibility to exfoliation.

Discussion
Salt Spray Tests
The SWAAT test was more effective than the MASTMAASIS test
in detecting exfoliation susceptibility of 7178 alloy with different levels
of resistance. Although both tests showed variability among the eight
laboratories testing common stocks of material, the variability was less
with the SWAAT test.
The greater variability of the MASTMAASIS results could be due to
any number of reasons. Since four laboratories operated the cabinet under
wet bottom conditions and four under dry bottom conditions, variations
in humidity during the purge and soak cycles could result. The concen-
tration of acetic acid might have an effect. Increasing the concentration of

S«
•a
v
> wu ' • » » ^

j<
Vi
d
s o § «
=i HH 'UH

WW W

WW W ^

3wcywwwww

N-tt—( 2 hH W2 S 2
WW(7WWWWW

WW a w •w

wwo^w 'W

a j CC CC U2 1/3!/! C« CC CC g
WWWWWWWWW^

cn i» 03 cn w cc 1/3 CD 1/1 g
WWWWWWWWW^

03 031/3 03 :03 03 £
WWWW^ •WW a>

03 031/3 03
WWWW •WW,

•4:muowwoni-i s
a
s
03

<« s„
•o «"|WW 'WW •«>
o
J*
^-*\
Ji
«
s o
r-4 sa
[A ^ WW •WW
H«
6
1 51 SS
O •" WW •W m
C8 ^

1
s
e I
I
o
P
j "
^
s„
WW •W m

51|MP-( SM!»
•^ wwawwwi:?^

^'WWOfWHWWW

•^ WWO'B :<y

WW9H :a
•w

^ Swg^SSSS^I

WWWWWWWWWC6

•S > 3SSS 'WWo

I •« WWWW 'WW«3
11
if
<moowwoa3„ a
a
03
s ill
Copyright by ASTM Int'l (all rights reserved); Fri Jan 1 12:32:22 EST 2016
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions
KETCHAM AND JEFFREY ON EXFOIIATION CORROSION TESTING 293

o- 0.
l^& m r-l
aa/
'. to
. a
V
AT
PLIPH PHC « i-<

a/Of : : : : : » ^
'. tic
. a
•J3
: : : : : 'acr

S 'S. o
PHfcpd ( S P H PL( PH P< <o cw

PH PL, p;] PH pH PH P< PH l>r-

Can

g ••• p
<( . . . ft O

•a1

ts 1^ ft O
PH PH O ' C P J PH PH PJ W <O 05

t»
«
<I <i ft O
PL, p^ ata^ PH PH PM pi; a; 03

ec
. . . ig.^^S
o . . . < ; ft.o
PHPnaC • • -PHH
BO
13 h
•rj <u
. . . is!-^*^
o ^ . . . p* ft o

HflpaoOwfiHOffli-H a
a

Z >
o
z

-s
Q
yi
X
< "*
<
gj
Z >
3

I O
,z

TABLE 10—Corrosion potential measurements in CHtOH/CCl,

versus saturated calomel electrode.
7178 (T/2) 7075 (T/2)
Potentials (mV)
Exfoliation Resistainee Exfoliation Resistance
Low 1 Low 2 High Low 1 Low 2 High
A 280 280 1030 300 280 1024
B 290 270 1033 293 380 1030
C 258' 268° 1057' 242 270 1047
D 280 277 1050 287 250 1023
265'' 275' 1120*
E 289 275 1045 32S« sgo" 1130'
V 102 112 1015
G 245 250 1050 270 272 1043
H 282 298 1029 265 270 1011
'• Values from corrosion test specimens; values from specimens received later were 267,
258, and 890 mV for low- and high-resistance material, respectively.
*• After 10 min.
" After 35 min.

acetic acid to that used for the SWAAT test might improve reproduci-
bility of the MASTMAASIS test.
The number of tests conducted on the 2024 material were insufficient
to provide for definite conclusions as to the best salt spray test, but the
available evidence suggests the MASTMAASIS test is more effective than
the SWAAT for this alloy.
Immersion Tests
The EXCO test proved to be most effective of the immersion tests in
distinguishing high and low susceptibility to exfoliation, and good repro-
ducibility was obtained in the various laboratories. The NH4Cl-NH4NO,s-
H^Oa test results had greater variability. The NaCl-SOg test showed good
reproducibility but is the most complicated to perform.
As with the salt spray tests, surface preparation is not critical when
conducting immersion tests, as both degreased and etched specimens gave
similar results. Masking back and sides of specimen in total immersion
tests proved not to be necessary since masked and unmasked specimens
gave similar results.
Similar results were obtained whether T/10 and T/2 planes were ex-
posed as a stepped specimen or separately.
Salt Spray Tests Versus Immersion Tests
With regard to reproducibility, the test results show the three immersion
tests to be better than the MASTMAASIS but not better than the SWAAT.
The switch in emphasis, therefore, toward immersion tests was based on
considerations other than reproducibility, such as equipment cost, speed,
simplicity, and better discrimination of borderline cases by reducing the
number of visual questionable ratings.

Corrosion Potential Measurements

The corrosion potential measurements in CH3OH-CCI4 electrolyte can
clearly identify material of a high and low resistance to exfoliation. As a
quality control test for heat treatment it has the advantages of being a
rapid and easily performed test. The chief disadvantage is that it produces
no visual evidence of exfoliation.
Conclusions
Based on the results of this round robin testing program it has been
concluded by the task group members that the EXCO immersion test is
the best laboratory exfoliation test method for the copper bearing 7000
series alloys.
Status
A draft of a recommended practice for the EXCO test has been pre-
pared and is being reviewed by the task group.
Natural environmental tests of the material used in this round robin
in marine and industrial atmospheres have been started by three of the
laboratories represented in the task group. The Alcoa test site is at Point
Judith, R. I., a marine atmosphere. Exposure dates were July 1970 for
the 7178 plate and June 1971 for the 7075 extrusion. The Kaiser expo-
sure site is at Daytona Beach, Fla., with the 7178 plate specimens exposed
in Aug. 1970, and the 7075 extrusions in Oct. 1970. The Reynolds site
is at McCook, 111., a severe industrial environment, and exposure began
there in Oct. 1970.
A cknowledgments
The authors gratefully acknowledge the participation and contributions
of the following collaborators in the interlaboratory testing program and
their companies: Chairman: D. O. Sprowls, Alcoa Research Laboratory;
S. G. Epstein, The Aluminum Association; B. R. Pathak, Alcan Research
and Development Ltd.; H. L. Craig, Jr., Reynolds Metals Co.; A. W.
Morris, McDonnell-Douglas Corp.; G. W. Kelly, LTV Aero Corp.; J. F.
Pashak, Dow Chemical Co.; T. J. Summerson, Kaiser Aluminum and
Chemical Corp.; W. C. Rotsell, Harvey Aluminum Inc.; and I. S. Shaffer,
Naval Air Development Center.

APPENDIX
MASTMAASIS and SWAAT Salt Spray Tests
Apparatus
1. A commercial fog cabinet designed to comply with ASTM B 117-64
specification for salt spray testing and modified to permit automatic cycling.
Both tests operated at 49 C.
2. 6-h cycle for MASTMAASIS test: 4 per day as follows:
(a) 45-min spray.
(A) 2-h purge with air.

(c) 3-h, 15-min soak period at 40 to 95 percent relative humidity.

3. 2-h cycle for SWA AT test: 12 per day as follows:
Alternately spray 30 min and soak 1 h and 30 min at 100 percent
humidity.
Solution
1. MASTMAASIS—5 percent NaCl (distilled water) buffered to pH 3 with
acetic acid.
2. SWAAT—Salt solution prepared by dissolving 42 g of synthetic sea
salt which meets ASTM D 1141-52 specification without heavy metal addi-
tions and 10 ml of glacial acetic acid in 1 liter of distilled water.
Test Procedure
1. Degrease specimens.'^
2. Support or suspend the specimens with the 6 in. dimension 45 deg from
the vertical, with the machined surfaces upward.
3. Test duration: 168 h.
4. Clean the corroded specimens gingerly by immersing in concentrated
nitric acid only for as long as necessary to remove all corrosion products,
taking care not to dislodge anymore flakes of exfoliated metal than can be
helped.
Immersion Tests
All the immersion tests had the following procedures in common:
1. Degrease specimens.
2. Use reagent grade chemicals and distilled or deionized water.
3. After completion of tests, clean corroded specimens the same as after
salt spray tests.
EXCO Immersion Test^
Solution: 4.0 M NaCl.
0.5 M KNO».
0.1 M HNO3.
pH: 0.4.
Test temperature: 25 C ± 2.
Ratio of solution volume to specimen surface area: 50 ml/in.^
Test duration: normally 48 h although 24 h is adequate to demonstrate
mode of attack.
Note—In the interlaboratory testing program the specimens were suspended
vertically. The ASTM recommended practice will specify the hori-
zontal position.
NHtCl-NHtNO,-HiO, Immersion Test
Solution: 0.5 M NH.Cl.
0.125 M NH4NO,.
10 g/liter of a 30 percent stock solution of H3O2.
pH: 5.2
' In the interlaboratory testing program both etched and unetched specimens
were tested.
' Different solution concentrations, various chloride/nitrate ratios, time periods
and temperatures (20 to 35 C) were explored by Alcan. The conditions recom-
mended were found to be the optimum.

Ratio of solution volume to specimen surface area: 40 to 100 ml/in.^

(including edge).
Test duration: 16 h.
Specimens suspended vertically with nonmetallic hangers. Top end of speci-
men at least 1 in. below surface of solution.
NaCl-SOi Immersion Test
In this method SO2 gas is produced, via chemical reaction of NazSO, +
H2SO4, above specimens immersed in a 5 percent NaCl solution. The chemical
reaction occurring is:
Na^SOs + H 2 S 0 . - > Na^SO. + H^SOs
- > H2O + SO2
This test was developed to simulate the environment of an aircraft carrier
with sea spray and stack gases containing sulfur.
Apparatus: See Fig. 11.

CONDENSER

NYLON CORD
(ATTACHED TO REACTION
PLASTIC LID TUKI

•PERFORATED
PYREXTUBE{8')

REACTION TUBE
(Z.5')
'Na,SQi CAPSULES

BATTERY JAR THERMOMETER

-A) SPECIMENS

FIG. U—NaCl-SO, immersion test.

Solution: 5 percent NaCl—in each reaction tube 2 capsules of Na-SOa

(1 capsule contains 1.4 ± 0.2 g) are mixed with not less than
8 ml of concentrated HjSOj.
Test temperature: 42 C ± 2.
Ratio of solution volume to specimen surface area: 50 to 100 ml/in.^
Test duration: 72 h.
Specimens tested in horizontal position.
Corrosion Potential Measurements
Apparatus: See Fig. 12. The voltmeter can be a pH meter generating on the
0 to 1400 mV scale.
Solution: CHaOH, absolute—350 ml; CCl.—150 ml; CuCU, anhydrous—
2.75 g.
Specimen selection: Triplicate specimens from each lot shall be tested. Speci-

SATURATED CALOMEL
REFERENCE ELECTRODE

GLASS CONTAINER
6 IN. DIA. X
2 '/, IN. HIGH

' 5 0 0 ML. C H i O H / C C I 4
FIG. 12—Assembled apparatus used in corrosion potential test.

men size shall be M by M by 2 in., wherever possible, with the 2 in.

dimension in either the long transverse or longitudinal direction. With
material thicker than % in., the specimen width shall be such that an
area close to Vi in.- per specimen is exposed to the solution when the
specimen is immersed to the prescribed depth of V> in. (For example, a
i/ie-in.-thick specimen can have a width of % in.).
Specimens shall not be taken any closer to an exposed edge (after aging)
than the thickness of the material.
Test duration: 35 min.
Suspend specimens so that the bottom edges are in the same horizontal
plane.

Heat Treatment Details

NOTE—^These treatments were not intended to produce commercial tempers
but rather were customized to result in different degrees of resistance to
exfoliation.
7178 Plate
Plant production solution heat treat at 875 F, cold-water quenched and
stretcher straightened.
Artificially aged in laboratory furnace as follows:
Low resistance—Lot 1—3 h at 250 F plus 4 h at 325 F.
Low resistance—Lot 2—3 h at 250 F plus 8 h at 325 F.
High resistance—3 h at 250 F plus 18 h at 325 F .
(NOTE—All plates from 1 ingot source, all laboratory aging done in a
single furnace load with each item removed after the above indicated
second step soak).
2024 Plate
Solution heat treated for 30 min at 915 to 920 F metal temperature followed
by a boiling water quench (nonstandard quench media used to induce
high susceptibihty to exfoliation).
7075 Extrusion
Plant production solution heat treat at 870 F, cold-water quenched and
aged 24 h at 250 F .

Subsequently artificially aged in laboratory furnace as follows:

Low resistance—Lot 1—8 h at 310 F.
Low resistance—Lot 2—2 h at 310 F.
High resistance—20 h at 325 F.

References
[1] Lifka, B. W. and Sprowls, D. O., Corrosion, SBIIA, Vol. 22, No. 1, Ian. 1966,
p. 7.
[2] Romans, H. B., Materials Research and Standards, MTRSA, Nov. 1969.
[3] Ketcham, S. J. and Shaffer, I. S. in Proceedings, 4th International Congress on
Metallic Corrosion, 1969 (to be published).
[4] Horst, R. L., Jr., and Lifka, B. W., Corrosion, SBIIA, Vol. 26, No. 3, March
1970.
[5] Zaretski, E. M. and Kireeva, A. F., Zavodskaya Laboritoriya, ZVDLA, Vol. 29,
1963, pp. 1098-1101.

Evaluation of the Tendency for

Dealloying in Metal Systems

REFERENCE: Verink, E. D., Jr., and Heidersbach, R. H., Jr., "Evaluation

of the Tendency for Dealloying in Metal Systems," Localized Corrosion—
Cause of Metal Failure, ASTM STP 516, American Society for Testing and
Materials, 1972, pp. 303-322.
ABSTRACT: Literature in the field of dealloying is reviewed. Special
emphasis is placed on promising new techniques which appear to have
predictive capacity in assessing the likelihood of dealloying. These involve
use of electrochemical hysteresis methods to generate experimental poten-
tial versus pH diagrams for alloys. Superposition of these experimental
diagrams over the theoretical potential versus pH diagrams (Pourbaix
diagrams) for the constituent metals of the alloy provides a basis for
prediction of the tendency for dealloying as a function of potential and
pH. Alpha brass (70Cu-30Zn) in 0.1 M sodium chloride is used as an
example. The method provides new insights into the explanation of the
mechanisms involved in dezincification. For initially copper-free chloride
solutions, ranges of potential are indicated in which selective leaching of
zinc predominates, where alloy dissolution with replating is expected,
and where alloy dissolution without replating occurs.
KEY WORDS: dezincification, electrochemistry, corrosion, tests, evaluation,
copper-containing alloys, brasses, potential theory, pH

This paper reviews the current status of dealloying corrosion in metal

systems giving particular emphasis to newer methods and techniques of
investigation and evaluation which offer special promise.
Dealloying is a corrosion process whereby one constituent of an alloy
is removed preferentially from the alloy leaving an altered residual struc-
ture [1,2].^
While dezincification is the most commonly experienced form of de-
alloying other examples have been reported in practice [3-13]. These
include loss of nickel [3,14-16], aluminum [10,17-23] and tin [24,25]

' Professor and assistant chairman, Department of Metallurgical and Materials

Engineering, University of Florida, Gainesville, Fla. 32601.
^ Department of the Army, Construction Engineering Research Laboratory,
Champaign, 111.
' The italic numbers in brackets refer to the list of references appended to this
paper.
303

Downloaded/printed
Copyright 1972 by FM International
by AS www.astm.org
University of Washington (University of Washington) pursuant to License
304 LOCALIZED CORROSION-CAUSE OF METAL FAILURE

from copper alloys; iron from cast iron [26]; nickel from alloy steels [27];
and cobalt from Stellite [28].
Since the phenomenon was first reported by Calvert and Johnson [29]
in 1866, the literature has been filled with reports of research efforts aimed
at clarifying the mechanisms of dealloying. Nonetheless, there still is no
general agreement as to the detailed mechanism involved. One group
contends that the entire alloy is dissolved and that one of its constituents
then is replated from solution [24,28,30-41]. Another contends that one
species is dissolved selectively from the alloy leaving a porous residue
of the more noble species [16,42-48]. Still others believe that both
mechanisms take place [17,49-55]. The authors agree with this third
group for reasons which will become evident.
The study of dealloying phenomena is fraught with a number of com-
plications. Generally, such reactions are relatively slow, and a lengthy
exposure period is required to cause dealloying of sufficient amount to
facilitate evaluation. Consequendy, there is considerable interest in ac-
celerated tests for evaluation of tendencies of alloys to dealloy. Many
techniques have been employed. For example, electrolyte compositions
have been adjusted by using more concentrated solutions or solutions
having variations in oxidizing power [53]. Specific ions have been added
to stimulate dealloying, for example, saturated cuprous chloride solutions
have been used to accelerate the dezincification of copper-base alloys [56].
Electrochemical stimulation also has been used. Unfortunately, all too
often the test methods employed can be criticized as having biased the
experimental result, and, although specific techniques are now available
which can cause dealloying to occur in the laboratory, nevertheless, there
still is no firm basis for predicting the likelihood of dealloying in service
based on these tests.

Recent Work on Mechanism

Much of the present dilemma regarding the mechanism of dezincification
has resulted from physical observations which seemed to support one or
the other of the two most accepted theories regarding mechanism.
X-ray diffractometer data from a-brass [7] exposed 20 to 30 days in
5 N hydrochloric acid (HCl) at 50 C show interpeak scattering between
the position of the (111) peaks for a-brass and pure copper. This indicates
a face-centered-cubic structure of lattice parameter intermediate between
the starting material and that of pure copper and is evidence of a selective
leaching mechanism. The intensity between the two peaks on the corroded
specimens is higher than would be obtained due to overlap between the
"tails" of the copper and the brass peaks. However, this intensity prob-
ably would go undetected on a standard powder pattern obtained by film
methods. Recent X-ray diffraction and electron microprobe studies have
produced data which indicate that a selective removal process of dealloy-

FIG. 1—Dezincificalion plug in 70-30 brass exposed for 79 days in 1 N NaCl at

room temperature (x200).

ing can occur in copper-gold and copper-zinc alloys under certain circum-
stances [1,57-59].
Figure 1 shows a "classical" example of plug-type dezincification in
70Cu-30Zn, a-brass. At the interface between the unattacked alloy and
the copper "sponge" there is a dense copper-rich metallic layer; the porous
nature of the sponge offers little resistance to the circulation of solution
to the reaction interfaces.
Figure 2 shows a dezincified region below the original metal surface.
This is one of a group of specimens which showed X-ray evidence of
selective leaching of zinc. The perturbations above the original surface are
large, dense deposits of copper. Thus, this specimen illustrates that both
mechanisms can occur on one specimen.
Abrams [31] first showed that copper deposition could occur where
flow was restricted. Pits and crevices are practical examples of flow-
restricted configurations. Figure 3 is a metallographic cross section of a
specimen of Monel 400 showing a deposit of copper which had occurred
in a crevice after exposure for 12 months in seawater.
Figure 4a is a scanning electron micrograph (SEM) of a copper de-
posit on an «-brass specimen which had been exposed for 10 days in 5 A'
HCl at 50 C. Figure 4b is a nondispersive X-ray pattern obtained using
the SEM and shows the formation to be a copper deposit.

FIG. 2—Copper deposit on surface and selective leaching of zinc {spongy area)
occurred on the same 70-30 brass specimen dezincified for 10 days in 5 N HCl at
100C{X121).

FIG. 3—Copper deposit on Monel 400 in crevice-corrosion specimen exposed to

seawater for 12 months (X268). Specimen, courtesy U. S. Naval Ship Research and
Development Center.

'"1i^|t^''^^lW^

I6r

8 12
X

^ 10

I
m
S

2 4 S 8
EMISSION ENERCY- KlV
FIG. 4—(a) Scanning electron micrograph of copper deposit on the surface of
a 70-30 brass specimen exposed for 10 days in 5 N HCl at 50 C (xlOOO) and (b)
nondispersive X-ray pattern of mineral shown in (a).

New Techniqaes
Work now underway at the University of Florida provides encourage-
ment that the tendency for dealloying may be predictable on the basis
of experimentally determined potential/pH diagrams constructed from

«5

.*» 1 •

S t rt
fc "=>•
5^ ? R
bq 4) w

S T3 «
"s C -3
S a 3
«>>»>!•
•c ;Z «>

ill
?si
C *2

3^1
•^ '^
St Cf
s;-s^
2
z
a
1
II
jji
M C "^

_ Q.S g
II .'''18
X

0)
11
"ll'^
z
a
^
f x i"i
s
!? *
1
•^•2.1.
d'?"3
"-I •« a
fcl S
ES

electrochemical information generated by the electrochemical hysteresis

method [60,61]. In this method, potentiokinetic polarization curves (po-
tential/current density) are run from minus potentials to plus and then
returned over a range of interest in solutions of various pH's. Electro-
chemical features such as "zero current potentials," "passivation poten-
tials," "rupture potentials," and "protection potentials" are noted and
plotted on potential/pH coordinates as indicated in Figs. 5a and b. Con-
necting the data points delineates regions of "immunity," "general corro-
sion," and "passive" behavior. Details of the experimental procedure
appear elsewhere [60]. Figure 6 is the equihbrium potential versus pH
diagram drawn by Van Muylder et al [62] for the CU-CI-H2O system
assuming a chloride ion concentration of 0.1 M. Figure 7 is a simplified
version of this diagram assuming ionic species to be present in concen-
trations of 10"® M. Figure 8 is a similar simplified potential versus pH
equilibrium diagram for the Zn-H20 systems [64]. The equations used in
constructing these diagrams are given in Table 1. Figure 9 is the experi-
mental potential versus pH diagram for 70-30 Cu-Zn alloy prepared by
Fort.* This diagram was constructed using techniques just described.
Superposition of the diagrams, Figs. 7-9 gives Fig. 10. In making such
a superposition the assumption is made that from an energy standpoint
behavior of zinc atoms in the 70-30 brass matrix will be similar to zinc
atoms in a zinc matrix to a first approximation. It is evident that the
experimentally constructed diagram shares a number of features with the
equilibrium (potential versus pH) diagram for copper. For example, in
acid solutions the line, #76, separating copper from CuCl~, (molarity
= 10-") virtually coincides with the zero current potential on the up-
ward scan of the electrochemical hysteresis circuit for a-brass (70Cu-
30Zn) in 0.1 M chloride.
The boundary between Cu^O and copper, line #12, corresponds to the
sloping line segment of the experimental diagram parallel to it extending
from about pH = 7.3 to 8.3. The displacement downward is believed to
be related to the zinc content of the alloy. The "jog" in this line at pH
s 8.3 may be caused by the presence of ZnO ott the surface of the
alloy above pH = 8.3.
Lines 12 and \2b represent the Cu/CujO equilibria for the non-
hydrated and hydrated species of CU2O, respectively. The corresponding
line on the experimental diagram coincides with the position of the calcu-
lated line for hydrated CuaO (see Fig. 7). The films formed on the
specimens were too thin for positive identification by X-ray methods. It
is hoped that optical methods will be successful in identifying them.
The line on the experimental diagram above and parallel to the Cu/
CugO equilibrium at pH's above 8.3 coincides with the calculated position
of the metastable equilibrium line for the coexistence of copper with
' Fort, W. C, University of Florida, private communication.

-1\ • 0

3 4 S 6 7 > 9 10 11 12 13 14 tS 16
pH
FIG. 6—Equilibrium potential versus pH diagram for the Cu-Cl-HtO system at
25 C for solutions containing 0.1 M chloride ion. Solid species were assumed to be
Cu. CuX>, CuCl, 3Cu{OH),CuCl,y. CuO. and Cu,0, hydrated [62].

3Cu(OH)2CuCl2T in 0.1 M chloride (shown as Eq 57 in Table 1 and in

Fig. 7). It also is fairly close to the calculated position of line 14fc (see Table
1) which involves hydrated CU2O in equilibrium with CuO. However,
the greenish color and the presence of chlorides in the reaction product
film strongly suggest that at least some metastable tri-hydroxy-chloride is
present. The difference in slope between —0.0443 required for Eq 57

CuPj hycr

t 9 10 11 '/2 n
pN

FIG. 7—Simplified Cu-Cl-HtO diagram at 25 C for solutions containing 0.1 M

chloride ions and concentrations of ionic species equal 10—" M. Redrawn from
Ref 62.

and -0.0591 required by Eq 14 is too subtle to discern as yet consider-

ing the scatter in the present data for a-brass (70Cu-30Zn).
On the return scan of the electrochemical hysteresis circuit a zero cur-
rent potential at 0.200 VSHE (SHE refers to the standard hydrogen elec-
trode) is observed which is independent of pH, at least to pH = 11.
This feature is observed with pure copper and with copper-rich alloys of
the copper-nickel and of the copper-zinc systems exposed in chloride solu-
tions. These loci coincide with the position of the Cu/CuQ equilibrium
potential in 0.1 M chloride.
Relevance to Dealloying
Although a vast amount of research remains to be done to verify the
suggestions which are presented herein, there is considerable encourage-

IJO T 1 r:;—I 1 r 1 r 1 r

(b •^(IK
6

Zn(OH),
U4-.2

s
HZio;
-e
• - ^

-.81-

-w
—®-
-17
Zn
-1.4

1 12 13
pH

FIG. 8—Simplified Zn-HiO diagram for concentration of ionic species equal

10-0 M_ Redrawn from Ref 63.

TABLE 1—Equation! used in construction of potential versus pH diagrams for the

Cu-Cl-HsO system and the Zn-BiO system.
C u - a - H , 0 System [6«]
Eq No."
12 2Cu + H , 0 = Cu,0 -f- 2H+ + 2e
(a) E = 0.471 - 0.0591 pH (no hydrated oxides)
(6) E = 0.572 - 0.0591 pH (assumes hydrated CusO)

14 CujO + H , 0 = 2CuO + 2H+ + 2e

(a) JB = 0.669 - 0.0591 pH
(6) iS = 0.568 - 0.0591 pH (assumes hydrated CujO, but nonhydrated CuO)

1« 2CuO + HjO = CujO, -f- 8H+ + 2e

(o) E = 1.648 - 0.0691 pH

TABLE 1 (continued)
17 Cu++ + H , 0 = CuO + 2H+
(a) log(Cu++) = 7.89 - 2 p H

43 CuCfc- = Cu++ + 2 C ^ + e
(Cu++)
(o) E = 0.46S = 0.0591 log + 0.1183 log ( C ^ )
(CuC/r)
51 aCuC/ 4- H , 0 = C u , 0 + 2 a - + 2H+
(a) loK(Cn = -5.66 + pH

53 3 C u ( 0 H ) r CuC/jT = 4CuCO + 2 0 " + 2H+ + 2HzO

(a) lo«{a-) = -7.40 + pH

66 Cu + a - = CnCl + e
(a) E = 0.137 - 0.0591 log{CI-)

57 4Cu + 6H2O + 2 0 - = 3Cu(OH)2 C u a , ^ + 6H+ + 8e

(a) E = 0.461 - 0.0443 p H - 0.0148 log(C^)

59 4 C u a + 6HjO = 3Cu(OH)2CuCfc^ + iCI' + 8H+ + 4e

(a) E = 0.785 - 0.0886 p H + 0.0295 log(C/-)

62 2CU2O + 4H2O + 2 a - = 3 C u ( O H ) 2 C u a , T + 2H+ + 4e

(a) E = 0.451 - 0.0295 log(,Cl~) - 0.0295 p H

67 2CuC/,- + H2O = CU2O + 4 C ^ + 2H+

(a) WiCuClc) = 4.45 - p H + 2 l o g ( a - )

76 Cu + iCl- = CaClc + e
(a) E = 0.208 + 0.0591 \og(CuCl-r) - 0.1182 log (Ci")

80 CiiCi.- + H2O = CuO + 2C?- + 2H+ + e

(o) ;? = 0.932 - 0.1182 p H - 0.059 log(CiiC/r) + 0.1182 log (C/")

82 CuCl = Cu++ + C ^ + e
(a) ^ = 0.537 + 0.0591 log(Cu++) + 0.0591 log(C^)

Zn-HjO System [6.?)

Eq No.
5 Zn + H2O = ZnO + 2H+ + 2e
(a) E = - 0 . 4 S 9 - 0.0591 p H

6 Zn++ + HcO = ZnO + 2H+

(a) Iog(Zn++) = 10.96 - 2 p H

7 ZnO + HoO = HZnOj- + H+

(a) log(HZnOr) = - 1 6 . 6 8 + p H
9 Zn = Zn++ + 2e
(o) E = - 0 . 7 6 8 + 0.0295 log(Zn++)
" Equation number.'! refer to circled numbers on diagram.<i, Fig.s. 9-13.

-1- -r -r

1i X 0 ZEM CUMEHT NTENTIU

• MItnnMM NTtNTIH
*• nMTccniw POTEimu
1.4 A FIMT MMIMIIN
A tECWMn MWMA

.8 ^-4

CORROSION

.%:
T.
,t PRSSIVKTION \

-^^-^^ # — * • + + W-

J L. •r®>
t 9 10 11 12 13
PH

FIG. 9—Experimental potential versus pH diagram for 70Cu-30Zn in 0.1 M

chloride at 25 C*

ment to believe that experimental potential versus pH diagrams, interpreted

in conjunction with the equilibrium Pourbaix diagrams of the constituent
metals, can provide useful information regarding the theoretical tendency
for dealloying. Such predictions must be verified by long-term tests to
determine reaction kinetics and the compositions and morphologies of
reaction products. Such collateral investigations have been facilitated
greatly by the development of modern research tools including new types
of X-ray analysis, scanning electron microscopy, electron microprobe analy-
sis, ellipsometry, atomic absorption spectroscopy, etc.
For pure copper in deaerated solutions, the zero current potential on
the upward potential sweep of the electrochemical hysteresis cir-
cuit indicates the position of the so-called "immunity" line at a given pH.
As in the case of theoretical Pourbaix diagrams, "immunity" does not
imply absolute freedom from corrosion but rather a negligible amount

T 1 1 1

FIG. 10—70Cu-30Zn alloy in 0.1 M chloride solutions. Superposition of Figs.

7-9. Small dots indicate the domain in which selective removal of zinc is expected
in solutions free of copper ions. Larger dots indicate the domain in which both
copper and zinc dissolve. Crosshatching indicates the region in which copper is
expected to deposit.

of corrosion. The potential at which this zero current is observed often

is fairly close to the calculated position of the metal/metal ion coexistence
potential for a metal ion concentration of 10-» M. Thus, the arbitrary
choice of 10"« M, often chosen as a definition of "noncorrosion," seems
appropriate. Alloys which are rich in one component (for example, 70-
30 Cu-Zn or 90-10 Cu-Ni) tend to have many features in common with
the diagram for the major component (in this case, copper). The "im-
munity" line, however, appears to have a somewhat different significance

for alloys than for pure metals. While the kinetics of alloy dissolution for
a-brass are slow below the alloy "immunity" line this does not rule out
the possibility of dealloying. Referring to Fig. 10, the immunity line for
70-30 Cu-Zn in 0.1 M chloride was about 0.000 VSHE- Between this
potential and approximately —0.940 VSHE (line #9) there is a theoretical
tendency for the selective removal of zinc from the alloy. Poly crystalline
specimens of 70-30 Cu-Zn potentiostated at —0.150 VSHE in 0.1 M chlor-
ide of pH = 4 (noted by "X" on Fig. 10) showed zinc ion pickup in
the solution without the presence of detectable amounts of copper after
107 h at room temperature, based on data obtained by atomic absorption
spectroscopy. The limit of detection by this method is less than 0.25 ppm.
Pickering and Byrne [57] showed zinc ions but no detectable copper ions
from a-brass exposed in sodium sulphate (Na2S04) solutions of pH = 5
after 20 h potentiostated at various potentials below about + 0.050 VSHE-
Some researchers add copper ions to accelerate dezincification. Ac-
celerated test methods in which copper ions are added to the solution
will result in deposition of copper on the surface of the alloy at any po-
tential more active than about +0.200 VSHE, as shown in Fig. 10. There-
fore, one might be led to the erroneous conclusion that only one mech-
anism exists (dissolution of both components of the alloy followed by
deposition of the more noble species). Below about 0.000 VSHE, as shown
in Fig. 10, selective leaching can occur in the absence of copper ions.
At potentials more positive than 0.000 VSHE, both constituents of the
70-30 Cu-Zn alloy go into solution. Atomic absorption studies indicate
that the ratio of copper/zinc for a-brass in the bulk 0.1 Af chloride ap-
proaches that of the alloy (above 1.5) very quickly above 0.000 VSHE,
perhaps within less than 100 mV. Pickering and Byrne [57] report that
"two modes of dissolution may occur depending on the potential: pref-
erential dissolution of the more noble metal to simultaneous dissolution
of both components." They report preferential dissolution below about
+ 0.200 VSHE and simultaneous dissolution above +0.200 VSHE-
Pickering and Byrne's studies were carried out in Na2S04 solutions at
pH = 5. According to those authors, Cu++ was believed to predominate.
Studies at University of Florida were carried out in chloride solutions.
Under these circumstances cuprous ion is favored as the result of the
stabilizing effect of the formation of CuCl~2 ion [65]. Consequently, for a
given amount of current passed, the equivalent weight of copper dis-
solved would be greater in chloride solutions than in solutions free of
chloride ion. This probably accounts for our observation that copper/
zinc ratios approaching that of the alloy were achieved very soon after
raising the potential above 0.000 VSHE- Observation of the specimens
under test in the potential range between 0.000 VSHE and + 0.200 VSHE
showed no visual evidence of copper enrichment on the specimen surface
when solutions were stirred. Copious quantities of copper were deposited

on the auxiliary electrode, however. Tests in which no stirring was done

also resulted in some deposition on the platinum auxiliary electrode; how-
ever, noticeable amounts of copper also were deposited loosely on the
surface of the specimens. These observations confirm the conclusions of
Abrams [31] on the effect of stagnation. Crevices and other occluded
cells may be considered extreme cases of stagnation in which concentra-
tions of various species within the occluded cell may vary drastically from
those in the bulk solution [66].
In summary, at potentials between 0.000 VSHE and 4-0.200 VSHE both
constituents dissolve, but, in the presence of occluded cells or stagnant
conditions, copper may deposit on the specimen without evidence of
selective leaching. Naturally in this potential range (as in the previously
mentioned range below 0.000 VSHE) accelerated tests involving copper ion
additions will result in deposition of copper.
Above -(-0.200 VSHE both constituents of the alloy dissolve in the
proportions found in the alloy without evidence of deposition of copper.
If the potential has been held above -f 0.200 VSHE for an appreciable
period of time, reducing the potential to + 0.200 VSHE or below will result
in deposition of copper presumably by the reaction CuCl -I- e -> Cu +
CI-.

Summary of Observations (Refer to Fig. 10)

Potential, VSHE Observation
- 0 . 9 4 0 to 0.000 1. Preferential dissolution of zinc from 70-30
Cu-Zn alloy essentially without dissolution
of copper. Kinetics controlled by "blockage"
of the surface by residual copper, and ability
of zinc to diffuse to the reaction front.
2. If copper ions (or other reducible species)
are added to the solution, deposition of the
added species will result.
0.000 to +0.200 1. Both constituents of the alloy dissolve. The
ratio of copper/zinc in solution will be in-
fluenced by the valence of the predominant
copper ion. In chloride solutions formation
of CuCl-2 stabilizes the monovalent copper
species and little or no selective leaching of
zinc is observed. Some selective leaching of
zinc has been reported in nonchloride solu-
tions [47].
2. Copper will deposit from unstirred solutions
or in occluded cells.
3.^ Copper ions added to the solution will

deposit, particularly in stagnant solutions

or in occluded cells.
Above +.200 VSHE 1- Both constituents of the alloy dissolve in
the ratio in which they appear in the alloy.
2. No deposition of copper occurs from 0.1 M
chloride solutions.
3. If potential is dropped to +0.200 or be-
low, deposition of copper will result.
90-10 copper-nickel alloys also undergo dealloying with loss of nickel.
The conditions leading to a tendency for dealloying of 90-10 Cu-Ni may
be predicted as in the case of a-brass. Figure 11 is a composite of the
CU-Q-H2O diagram [62] for 0.1 M chloride (Fig. 6) plus the equilibrium
diagram for nickel [64] (with Ni++ assumed to be 10-® M), plus the
experimental diagram developed by Parrish for 90-10 Cu-Ni in 0.1 M
chloride solutions [67].
In the potential range between about 0.000 VSHE and +0.200 VSHE,
both constituents of the alloy dissolve. Deposition of copper occurs in
crevices [61] and pits [68] and presumably in unstirred solutions.
Specimens which have been exposed at potentials above 0.200 VSHE
will experience deposition of copper upon reducing the potential below
0.200 VSHE in 0.1 M chlorides.
Verink and Parrish [63] have reported the theoretical possibility of
dealloying of 90-10 Cu-Ni in domains where a passive region in the
equilibrium diagram for one alloy constituent coincides with a general
corrosion region for the other constituents.
Summary
Dealloying phenomena are encountered in a number of metal alloy
systems. It is a problem of considerable practical importance in the use
of copper-base alloys, particularly those of the copper-zinc system.
For over 50 years there has been disagreement regarding the mech-
anism of dealloying. It now appears that dealloying may proceed by
either a selective leaching process or a dissolution and replating mecha-
nism or both.
The development of new, highly sophisticated investigative tools such
as the SEM, microprobe, special X-ray techniques, spectrophotometric
analysis, electrochemical techniques, etc., provides great opportunities for
characterizing and evaluating dealloying phenomena.
The organization of these observations on an orderly basis such as
appears possible by use of potential versus pH diagrams seems to hold
great promise in the quest for better understanding. In turn, a clearer
understanding of the mechanisms involved will permit development of
alloys and procedures with greater engineering reliability.

T 1 1 1 n—I ^ r

CU2O3 hydr 1

-1.0 Cu Nj

-1.2

-U j _ _L -L. _1_ _L JL J. J_
3 4 5 6 7 « 9 10 II 12 13
pN
FIG. 11—90Cu-10Ni alloy in 0.1 M chloride solution. Superposition of the
experimental potential versus pH diagram of Parrish on the Cu-Cl-HtO diagram
(Fig. 7) and the Ni-H,0 diagram for Ni** = JO-" M. Small dots indicate domain in
which selective dissolution of nickel is expected in a copper free solution. Larger
dots indicate domain in which both nickel and copper dissolve. Crosshatched area is
domain in which copper is expected to deposit.

A cJcnowledgments
The electrochemical hysteresis method used to construct experimental
potential versus pH diagrams was developed and a portion of the work
on 90-10 Cu-Ni alloy was performed as part of the research supported
by the Advanced Research Project Agency under Contract No. N-00014-
68-A-0I73 administered by the Office of Naval Research, Washington,
D. C. The balance of the copper-nickel alloy work and part of the copper-

zinc alloy work were supported by the Department of Interior, Office of

Saline Water, Washington, D. C, under Contract No. 14-30-2600. The
National Association of Corrosion Engineers, Houston, Texas, through
their Committee on Research, supported the studies of dealloying. To
each of these sponsors, the authors express deep appreciation.
Special thanks are due Marcel Pourbaix who served as consultant on
the A R P A contract and whose continued association with the University
of Florida as co-director of the Center of Applied Thermodynamics and
Corrolsion serves as an on-going source of stimulation.
Appreciation also is expressed to several graduate research assistants
in the Department of Metallurgical and Materials Engineering at the Uni-
versity of Florida; to W. C. Fort, III, for the use of some of his data,
to K. K. Starr and T. S, Lee, III, for their helpful review and comments
during the preparation of the manuscript.

References
[/] Heidersbach, R. H., Jr., Corrosion, SBIIA, Vol. 24, No. 2, 1968, pp. 38-
44; Vol. 25, No. 7, 1969, pp. 285-286; Vol. 26, No. 5, 1970, p. 217: Vol. 26,
No. 8, 1970, p. 256; Vol. 26, No. 10, 1970, pp. 445-447.
[2] Tammann, G., Zeitschrift fur Anorganisch Atlegemeine Chemie, ZAACA, Vol.
107, 1919.
[i] Bird, D. B. and Moore, K. L.. Materials Protection. MAPRA. Vol. 1. No. 10.
1962, p. 70.
[4] Blumer, A. F., Corrosion, SBIIA, Vol. 5. No. 144, 1949.
[5] Crenell, J. T. and Sawyer, L. J. E.. Journal of Applied Chemistry, JACHA,
Vol. 12, No. 4, 1962, p. 170.
[6] Matsuda. M.. Journal, Japan Institute of Metals, NIKGA. Vol. 26, No. 1, 1962.
p. 124.
[7] Niederberger, R. B., Modern Castings, MODCA, Vol. 45, No. 3, 1964, p. 115.
[8] Polushkin, E. P. and Shuldener, M., Transactions, American Institute of Min-
ing, Metallurgical, and Petroleum Engineers, TAIME, Vol. 161, 1945, p. 214.
19] Simmons, A. L., Metal Progress, MEPOA. Vol. 57. 1950, p. 496.
[10] Tracy, A. W. in Atmospheric Corrosion of Non-Ferrous Metah. ASTM STP
175, American Society for Testing and Materials, 1956, p. 67.
IN] Van der Baan. Sj., Corrosion. SBIIA, Vol. 6, 1950, p. 14.
[12] Vernon, W. H. H.. Transactions, Faraday Society, TFSOA. Vol. 23. 1927,
p. 170.
[13] Wilkins, R. A., Mechanical Engineering. MEENA, Vol. 58, No. 12, 1936, p. 57.
[14] Heriderson, J. L. and Roadhouse, C. L.. Journal. Dairy Science. JDSCA. Vol.
23, 1940, p. 215.
[15] Stewart, W. C. and LaQue, F. L., Corrosion, SBIIA, Vol. 8, 1952, p. 259.
[J6] Taylor, F. and Wood, J. W., Engineering, ENGNA, Vol. 149, 1940, p. 58.
[17] Britton, S. C , Journal, Institute of Metals, JIMEA, Vol. 67, 1941, p. 119.
[18] Gleekman, L. W. and Swandby, R. K., Corrosion. SBIIA, Vol. 17, 1961, p. 144t.
[19] Hamstead, A. C , Industrial and Engineering Chemistry, lECHA, Vol. 50, No.
10, 1958, p. 87A.
[20] Hesse, A. H., Myskocoski, E. T., and Loring, B. M., Transactions, American
Foundrymen's Association, TAFOA, Vol. 51, 1944, p. 821.
[21] Peters, B. F., Carson, J. A. H., and Barer. R. D., Materials Protection,
MAPRA, Vol. 4, No. 5, 1965, p. 24.
[22] Schussler, M. and Napolitan, D. S., Corrosion, SBIIA, Vol. 12, 1956, p. 107t.
[23] Serre, J. and Lawreys, J., Corrosion Science, CRRSA, Vol. 2, 1965, p. 135.

[24] Clark, W. D., Journal. Institute of Metals, JIMEA, Vol. 73, 1947, p. 263.
[25] Trautzel, P. and Treadwell. W. D., Helvetica Chimica Acta, HCACA, Vol. 34,
1951, p. 1723.
[26] Leedon, L. M., Journal, American Water Works Association, JAWWA, Vol.
38, 1946, p. 1392.
[27] Dvorak, A., Sirojirentsvi, STRJA, Vol. 12, No. 1, 1962, p. 39, referenced by
Corrosion Abstracts, CRNAA, Vol. 2, 1963, p. 11.
[28] Fontana, M. G., Industrial and Engineering Chemistry, lECHA, Vol. 39. No.
5, 1947, p. 87A.
[29] Calvert and Johnson, J., Journal. Chemical Society, JCSOA, Vol. 19, 1866, p.
436.
[30] Lynes, W., Proceedings, American Society for Testing and Materials, ASTEA,
Vol. 41, 1941, p. 859.
[31] Abrams, R. B., Transactions, American Electrochemical Society, TAESA, Vol.
42, 1922, p. 39.
[32] Nixon, C. F., Transactions. American Electrochemical Society. TAESA, Vol.
45, 1924, p. 297.
[33] Hollomon, J. H. and Wulff, J., Transactions, American Institute of Mining,
Metallurgical, and Petroleum Engineers. TAIME, Vol. 147, 1942, p. 297.
[34] Thompson, D. B., Australasian Engineer, AUEGA, Vol. 48, Oct. 1954.
[35] Zender, A. R. and Bulow, C. L., Heating, Piping, and Air Conditioning,
HPAOA, Vol. 16, 1944, p. 273.
[36] Dixon, E. S., Bulletin, American Society for Testing and Materials, ASTBA,
No. 102, 1940, p. 21.
[37] Rhodes, F. H. and Carty, J. T., Industrial and Engineering Chemistry, lECHA,
Vol. 17, 1925, p. 909.
[38] Scully, J. C , The Fundamentals of Corrosion, Pergamon Press, London, 1966.
[39] Uhlig, H. H., The Corrosion Handbook, Wiley, New York, 1948.
[40] Hashimoto, K., Ogawa, W., and Shimodaira, S., Journal, Japan Institute of
Metals, NIKGA, Vol. 4, No. 1, 1963, p. 42.
[41] Horton, R. M., Corrosion, SBIIA. Vol. 26, No. 7, 1971, p. 160.
[42] Tabor, L. E., Journal. American Water Works Association, JAWWA, Vol. 48,
No. 3, 1963, p. 239.
[43] Gastev, S. S., Izvestiia, Akademii Nauk SSR Mettally, IZNMA, Vol. 3, 1965,
referenced in Corro.sion Abstracts, CRNAA. Vol. 6, 1966, p. 446.
[44] Bialosky, J. M., Corrosion and Metal Protection, Vol. 4, 1947, p. 15.
[45] Marshakov, I. K., Bogdanov, V. P., and Aleikina, S. M., Russian Journal of
Physical Chemistry, RJPCA, Vol. 38, No. 7. 1964. p. 960; Vol. 38, No. 8,
1964, p. 104; Vol. 39, No. 6, 1965, p. 804.
[46] Bassett, W. H., Chemical and Metallurgical Engineering, Vol. 27, 1922, p. 340.
[47] Efrid, K. D., M.S. thesis. University of Florida, Gainesville, Fla., 1970.
[48] Rubin, B. T., Ph.D. thesis. University of Pennsylvania, Philadelphia, Pa., 1969.
[49] Piatti, L. and Grauer. R.. Werkstoffe und Korrosion. WSKRA, Vol. 14, No. 7,
1963, p. 551.
[50] Colegate, G. T., Metal Industry, MINYA, Vol. 73, No. 507, 1948, p. 531.
[51] Kenworthy, L. and O'Driscoil. W. G.. Corrosion Technology, CRRTA, Vol.
2, 1955, p. 247.
[52] Evans, V. R., The Corrosion and Oxidation of Metals, E. Arnold, 1960.
[53] Stillwell. C. W. and Turnipseed, E. S., Industrial and Engineering Chemistry,
lECHA, Vol. 26, 1934, p. 740.
[54] Verink, E. D., Jr., and Parrish, P. A., Corrosion, SBIIA, Vol. 26, 1970, p. 5.
[55] Fink, F. W., Transactions, Electrochemical Society, TESOA, Vol. 75, 1939,
p. 441.
[56] Lucey, V. C, British Corrosion Journal, BCRJA, Vol. 1, 1965, p. 7; Vol. 2.
1965, p. 53.
[57] Pickering, H. W. and Byrne, P. J., Journal. Electrochemical Society, JESOA,
Vol. 116, No. II, 1969, p. 1492.

[58] Pickering, H. W., Proceedings, Conference on Fundamentals of Stress Corrosion

Cracking, Columbus, Ohio, 1967, Staehle, R. W., cd., pp. 159-177, published
1969, National Association of Corrosion Engineers.
[59] Pickering, H. W. and Byrne, P. J., Journal, Electrochemical Society, JESOA,
Vol. 118, No. 2, 1970, p. 209.
[60] Verink, E. D., Jr., and Pourbaix, M., "Use of Electrochemical Hysteresis
Techniques in Developing Alloys for Saline Exposures," Corrosion, SBIIA,
Vol. 27, No. 12, Dec. 1971.
[61] Pourbaix, M., Corrosion, SBIIA, Vol. 26, 1970, pp. 431-438.
[62] Van Muylder, J., deZoubpv, N., and Pourbaix, M., Report 101, CEBELCOR,
Brussels, Belgium, July 1962.
[63] deZoubov, N. and Pourbaix, M., Atlas of Electrochemical Equilibria in
Aqueous Solutions, Pergamon Press, London, CEBELCOR, Brussels, Belgium,
1966, pp. 406-413.
[64] Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions,
Pergamon Press, London, CEBELCOR, Brussels, Belgium, 1966.
[65] Latimer, W. M., Oxidation Potentials, 2nd ed., Prentice-Hall, Englewood Cliffs,
N. I., 1961.
[66] Pourbaix, M., Report 127, CEBELCOR, May 1965.
[67] Parrish, P. A., M.S. thesis. University of Florida, Gainesville, Fla., June 1970.
[68] Van Muylder, J., Pourbaix, M., and Van Laer, P., CEBELCOR Report 127,
Brussels, Belgium, May 1965.

ASM Handbook of Case Histories in Failure Analysis, Vol 3, 2019
100% (13)
ASM Handbook of Case Histories in Failure Analysis, Vol 3, 2019
709 pages
Atlas of Fatigue Curves
92% (26)
Atlas of Fatigue Curves
534 pages
ASTM - STP 425 - Stress Corrosion Testing 1967 PDF
No ratings yet
ASTM - STP 425 - Stress Corrosion Testing 1967 PDF
390 pages
Environmentally Assisted Cracking - Predictive Methods For Risk Assessment and Evaluation of Materials, Equipment, and Structures (ASTM Special Technical Publication, 1401) PDF
100% (2)
Environmentally Assisted Cracking - Predictive Methods For Risk Assessment and Evaluation of Materials, Equipment, and Structures (ASTM Special Technical Publication, 1401) PDF
491 pages
STP700 EB.14153 1 Fracture Mechanics... 1980
100% (1)
STP700 EB.14153 1 Fracture Mechanics... 1980
586 pages
Failure Analysis of Heat Treated Steel Components (2008) PDF
100% (4)
Failure Analysis of Heat Treated Steel Components (2008) PDF
636 pages
Handbook of Residual Stress and Deformation of Steel PDF
100% (7)
Handbook of Residual Stress and Deformation of Steel PDF
478 pages
Forms of Corrosion Recognition and Prevention NACE
100% (2)
Forms of Corrosion Recognition and Prevention NACE
388 pages
STP 656 - Intergranual Corrosion of Stainless Alloys
100% (2)
STP 656 - Intergranual Corrosion of Stainless Alloys
266 pages
ASTM - Thermal Fatigue of Materials and Components
100% (3)
ASTM - Thermal Fatigue of Materials and Components
270 pages
ASM Practical Fractography PDF
100% (10)
ASM Practical Fractography PDF
159 pages
Scientific Study of Mechanical Relaxation of Residual Stress PDF
100% (1)
Scientific Study of Mechanical Relaxation of Residual Stress PDF
120 pages
Mechanical Behavior and Fracture of Engineering Materials PDF
100% (3)
Mechanical Behavior and Fracture of Engineering Materials PDF
253 pages
ASTM - STP436 Electron Fractography
No ratings yet
ASTM - STP436 Electron Fractography
236 pages
ASM Failure Analysis of Engineering Structures Methodology and Case Histories PDF
100% (1)
ASM Failure Analysis of Engineering Structures Methodology and Case Histories PDF
204 pages
Metallography
100% (3)
Metallography
480 pages
STP 576-1976
100% (1)
STP 576-1976
316 pages
STP 518-1983
No ratings yet
STP 518-1983
173 pages
STP 1546-2012
No ratings yet
STP 1546-2012
390 pages
STP 600-1976
No ratings yet
STP 600-1976
263 pages
ASTM - STP 407 - Temper Embrittlement in Steel 1968 PDF
100% (1)
ASTM - STP 407 - Temper Embrittlement in Steel 1968 PDF
260 pages
Handbook of Residual Stress and Deformation of Steel
100% (1)
Handbook of Residual Stress and Deformation of Steel
477 pages
John J. Friel - Practical Guide To Image Analysis (2000, ASM International)
100% (1)
John J. Friel - Practical Guide To Image Analysis (2000, ASM International)
299 pages
Fatigue of Welding
100% (4)
Fatigue of Welding
310 pages
STP965 Eb.1415051 1 PDF
No ratings yet
STP965 Eb.1415051 1 PDF
447 pages
STP1256 Eb.1415051 1 PDF
100% (1)
STP1256 Eb.1415051 1 PDF
842 pages
Standardization of Fretting Fatigue
100% (1)
Standardization of Fretting Fatigue
275 pages
Atlas-Of-Stress-Strain-Curves Ocr PDF
100% (1)
Atlas-Of-Stress-Strain-Curves Ocr PDF
822 pages
Atlas of Fatigue Curves
100% (2)
Atlas of Fatigue Curves
534 pages
Stainless Steels For Design Engineers
67% (6)
Stainless Steels For Design Engineers
301 pages
2018 Book FractographyAndFailureAnalysis PDF
100% (8)
2018 Book FractographyAndFailureAnalysis PDF
172 pages
STP 619-1977
100% (1)
STP 619-1977
74 pages
Fatigue Thresholds
100% (3)
Fatigue Thresholds
436 pages
ASTM - Residual Stress Effects On Fatigue and Fracture Testing and Incorporation of Results Into Design 2007 - ASTM STP 1497
100% (11)
ASTM - Residual Stress Effects On Fatigue and Fracture Testing and Incorporation of Results Into Design 2007 - ASTM STP 1497
164 pages
Failure Atlas
100% (11)
Failure Atlas
478 pages
Fatigue and Crack Initiation - George C. Sih
100% (1)
Fatigue and Crack Initiation - George C. Sih
386 pages
ASM - Failure Analysis of Engineering Structures Methodology and Case Histories
100% (17)
ASM - Failure Analysis of Engineering Structures Methodology and Case Histories
204 pages
STP1189 Foreword
100% (1)
STP1189 Foreword
9 pages
Michael P. Monahan (Editor) Thomas A. Siewert (Editor) - Pendulum Impact Testing - A Century of Progress (ASTM Special Technical Publication, 1380) (2000)
No ratings yet
Michael P. Monahan (Editor) Thomas A. Siewert (Editor) - Pendulum Impact Testing - A Century of Progress (ASTM Special Technical Publication, 1380) (2000)
402 pages
STP780 Eb.1415051 1 PDF
No ratings yet
STP780 Eb.1415051 1 PDF
191 pages
STP 415-1967
50% (2)
STP 415-1967
551 pages
Stainless Steel Microstructure, Mechanical Properties and Methods of Application - 2015
100% (3)
Stainless Steel Microstructure, Mechanical Properties and Methods of Application - 2015
291 pages
ART Threaded Fastener Design and Analysis
No ratings yet
ART Threaded Fastener Design and Analysis
40 pages
Practical Engineering Failure Analysis Dekker Mechanical Engineering
100% (3)
Practical Engineering Failure Analysis Dekker Mechanical Engineering
607 pages
Metal Failures: Mechanisms, Analysis, Prevention
From Everand
Metal Failures: Mechanisms, Analysis, Prevention
Arthur J. McEvily
No ratings yet
MCE 518 Scan 3
No ratings yet
MCE 518 Scan 3
18 pages
Corrosion and Degradation of Materials
No ratings yet
Corrosion and Degradation of Materials
18 pages
Corrosive Damage in Metals and Its Prevention
83% (6)
Corrosive Damage in Metals and Its Prevention
36 pages
Corrosion: Meta Fitri Rizkiana, S.T., M.SC
No ratings yet
Corrosion: Meta Fitri Rizkiana, S.T., M.SC
41 pages
Chapter 4
No ratings yet
Chapter 4
57 pages
Evironmental Degradation of Materials: Rushikesh Kulkarni
100% (1)
Evironmental Degradation of Materials: Rushikesh Kulkarni
30 pages
Accelerated Corrosion Test
No ratings yet
Accelerated Corrosion Test
5 pages
Amte 133 Prelim Lesson
No ratings yet
Amte 133 Prelim Lesson
35 pages
6.4PPT self
No ratings yet
6.4PPT self
87 pages
UNIT 1 Basics of Surface Properties
No ratings yet
UNIT 1 Basics of Surface Properties
54 pages
Corrosion
No ratings yet
Corrosion
8 pages
Types of Corrosion: Dr. Hamdy A. Kandil
No ratings yet
Types of Corrosion: Dr. Hamdy A. Kandil
53 pages
Lecture 08 - Corrosion and Degradation of Materials - 2020
No ratings yet
Lecture 08 - Corrosion and Degradation of Materials - 2020
61 pages
Cleaning and Corrosion Control: Prelim
No ratings yet
Cleaning and Corrosion Control: Prelim
54 pages
Corrosion
No ratings yet
Corrosion
3 pages
SU
No ratings yet
SU
1 page
Nickel Welding
0% (1)
Nickel Welding
87 pages
Ferrous Metals Classification
No ratings yet
Ferrous Metals Classification
5 pages
Electronic Materials and Components-Die Bonding Materials
No ratings yet
Electronic Materials and Components-Die Bonding Materials
23 pages
Design of Die Casting Dies (TH) - PGDTD&CC
No ratings yet
Design of Die Casting Dies (TH) - PGDTD&CC
27 pages
Recent Progress in The Development of Gamma Titanium Aluminide Alloys
No ratings yet
Recent Progress in The Development of Gamma Titanium Aluminide Alloys
22 pages
Extrusion Second Edition PDF
100% (1)
Extrusion Second Edition PDF
604 pages
Conductivity Nickel
No ratings yet
Conductivity Nickel
2 pages
Aerospace Material Specification: Nickel Plating (Electrodeposited)
100% (2)
Aerospace Material Specification: Nickel Plating (Electrodeposited)
12 pages
4-MET416-CM-Module-4..pptx
100% (1)
4-MET416-CM-Module-4..pptx
64 pages
Unclassified Ad Number Limitation Changes TO
No ratings yet
Unclassified Ad Number Limitation Changes TO
33 pages
Weld Repair of Steam Turbine Rotor With 12 CR Weld
No ratings yet
Weld Repair of Steam Turbine Rotor With 12 CR Weld
5 pages
Metallic Coatings For Corrosion Prevention
No ratings yet
Metallic Coatings For Corrosion Prevention
9 pages
Sapa Design Collection en Brochure 110508 2
No ratings yet
Sapa Design Collection en Brochure 110508 2
44 pages
Bearing Material
No ratings yet
Bearing Material
2 pages
S20 Pompa Pumps Metallic Design Level 1 s20mdl1sm-REV0708 AP000009 en
No ratings yet
S20 Pompa Pumps Metallic Design Level 1 s20mdl1sm-REV0708 AP000009 en
31 pages
Steel Castings, Chromium-Nickel-Iron Alloy (25-12 Class), For High-Temperature Service
No ratings yet
Steel Castings, Chromium-Nickel-Iron Alloy (25-12 Class), For High-Temperature Service
4 pages
Industrial Material: Subject Name
No ratings yet
Industrial Material: Subject Name
17 pages
Chapter 2: Aerospace Materials Characteristics: Biliyar N. Bhat, NASA Marshall Space Flight Center
No ratings yet
Chapter 2: Aerospace Materials Characteristics: Biliyar N. Bhat, NASA Marshall Space Flight Center
145 pages
Material Science Long Quiz
No ratings yet
Material Science Long Quiz
3 pages
CH Emetics
No ratings yet
CH Emetics
1 page
DPMT-II Engineering Materials and Metallurgy
No ratings yet
DPMT-II Engineering Materials and Metallurgy
10 pages
Reaffirmed 2017
No ratings yet
Reaffirmed 2017
17 pages
Access Solution Manual for Absolute C++, 5/E 5th Edition Walter Savitch All Chapters Immediate PDF Download
100% (7)
Access Solution Manual for Absolute C++, 5/E 5th Edition Walter Savitch All Chapters Immediate PDF Download
39 pages
Specification For General Requirements For Wrought Copper and Copper-Alloy Plate, Sheet, Strip, and Rolled Bar
No ratings yet
Specification For General Requirements For Wrought Copper and Copper-Alloy Plate, Sheet, Strip, and Rolled Bar
16 pages
Metals Metals: Chapter Contents
No ratings yet
Metals Metals: Chapter Contents
11 pages
Weldability Al Alloys
No ratings yet
Weldability Al Alloys
7 pages
Get Social Intelligence Daniel Goleman PDF ebook with Full Chapters Now
100% (7)
Get Social Intelligence Daniel Goleman PDF ebook with Full Chapters Now
50 pages
NDT Basic Presentation
No ratings yet
NDT Basic Presentation
207 pages
Chapter1 Professor Oystein Grong
No ratings yet
Chapter1 Professor Oystein Grong
41 pages