Origin of delay time in stress corrosion cracking of austenitic stainless steels

Origin of delay time in stress corrosion cracking of austenitic stainless steels

Corrosion Science, 1965, Vol. 5, pp. 291 to 299. Pergamon Press Ltd. Printed in Great Britain ORIGIN OF DELAY TIME IN STRESS CORROSION CRACKING OF AU...

1MB Sizes 0 Downloads 42 Views

Corrosion Science, 1965, Vol. 5, pp. 291 to 299. Pergamon Press Ltd. Printed in Great Britain

ORIGIN OF DELAY TIME IN STRESS CORROSION CRACKING OF AUSTENITIC STAINLESS STEELS* H . H . UHLIG a n d J. P. SAVA Corrosion Laboratory, Massachusotts Institute of Technology, Cambridge, Massachusotts Abstract--Stressed specimens of 25 ~o Cr, 20 Yo Ni stainless stools wore removed from boiling MgCI= solution, pickled, and re-immersed. Total time to failure was not altered. Also, spocimem, not stressed, when heat treated at 154°C in air or by immersion in boiling MgC12 solution for various times showed the same susceptibility to stress corrosion cracking. These experiments, combined with additional upporting evidence, prove that an oxide or passive film has little or no influence on time for crack snitiation and propagation. Instead, the important factor appears to he associated with the time for segregation at lattice imperfection sites, of nitrogen atoms and similar damaging impurities, aided by the elevated temperature of the test. This segregation of nitrogen causes an ageing effect and also produces crack-sensitive paths. A low-nitrogen 20 Yo Cr, 19 ~o Ni austenitic stainless stool, for example, neither quench-ages nor is it susceptible to stress corrosion cracking. R~sum6---Des &hantillons sous tension d'aciers inoxydables a 25 ~ Cr et 20% Ni ont ~t6 retires d'une solution bouillante de MgCI= d~cap6s, et r6immerg~s. Le temps total de rupture n'a pas 6t6 alt6r6. D'autre part, des &hantillons non sous tension, quand ils sent trait6s a 154° C dans l'air ou par immersion dans une solution de Mg CI= bouiUante pendant diverses duress, ont montr~ la mEme susceptibilit6 ~ la corrosion par fissuration sous tension. Ces experiences, c o m b i n ~ s a des preuves additionnelles, prouvent qu'un oxide ou un film passif n'a que peu ou pas d'influence sur le temps de d6but et de propagation de fissuration. Au contraire, le facteur important semble 6tre associ6 au temps de segration aux endroits o~ des inperfections apparaissent dans le r~seau, d'atomes d'azote et d'autres impuret6s dommageables, s6gr6gation favoris~e par la temp6rature 61ev~ de l'essai. Cette s6gr6gation de I'azote produit un effet de vieillissement et crY6 des chemins sensibles a la fissuration. Un acior inoxydable austEnitique ~t 20% en Cr, 19% Ni ~ basse teneur en azote, par exemple, n'est susceptible ni a la corrosion sous tension par fissuration ni au vieiUissement. Zusammenfassung---Gespannte Proben aus korrosionsbest~indigen Stfihlen mit 25~/o Cr und 20~o Ni worden aus kochender MgCl=-L0sung horausgenommen, gebeizt und wider eingesetzt. Die Gesamtzeit bis zum Bruch findorte sich nicht. Auch ungespannte Prohen, die verschieden lange Zeiten bei 154°C in Luft oder in kochender MgCl2-L6sung erhitzt wurden, zeigten die gleiche Anfiilligkeit gegen Spannungsriss-Korrosion. Diese und andere'Versuche weisen nach, dass eine Oxyd oder Passivschicht einen geringen oder gar keinen Einfluss auf die Zeit fiir den Rissbeginn und die Kissausbreitung austibt. Stattdessen scheint der massgehende Einfluss mit der Ausscheidungszeit yon Stickstoffatomen und iihnlichen scl'~dlichen Vorunreinigungen an Gitterfehlorn in Verbindung zu stehen, unterstiitzt dutch die erh6hte Versuchstemperatur. Diese Ausscheidung yon Stickstoff verursacht neben einem Alterungseffekt auch rissempfindliche Bereiche. Ein korrosionsbest~ndiger ausenitischer Stahl mit 20yo Cr, 19~o Ni und einem niedrigen Stickstoffgehalt zum Beispiel ist unempfindlich gegen Abschreckalterung und zugleich gegen Spannungsrisskorrosion. Pe~epax ~ H a n p a ~ e H H ~ e o S p a a t ~ Ha IIep~aBemmax cTaae~, co;~epraauax 2 5 % Cr, 2 0 % N i BBIHHManHcB ~3 RHiianlePo pacTnopa M g C h , ~e~an~poBaa~c~, ~ BHOBB IIorpymannc~. O6i~ee BpeMH KO p a a p ~ B a IIpn aTOM He HaMeH~IJ~0CB. TOqH0 Ta~ me, ~ e n a n p a m e n n ~ e o6paa~In, no~Beprm~eca TepMnqecI~ol~ o6paSoT~e n~i6o B~ep~(nol~ n p n 154°C B B0a~IymH0~ aTMOC~pepe, nil60 norpymeHneM B g~nanmi~ pac~B0p M g C h , oSHapymnBa0m o~nna~oBym CII0C06HOCTS I¢ I ~ o p p o a n O H H O ~ y p a c T p e c R H B a H H m . ~)TI~ 0maT~I B c 0 q e T a H H H C ~OIIOJIHHTeJIbHhIMH KaHHhIMH cJIyH~aT ~oHaaaTeJIhCTBOM TOP0~ qTO 0HHCHaH HJIH HaCCHBHaH IIJIeHHH OHa3hlBaIOT MaJI0e HJIH He 0Ha3IaIBaIOT H H g a R o P o BJIHHHHH Hg~ n e p n o K BpeMeHH, HeoSx0~HMhII~

~JIH BOaHHHHOBeHHH Tpen~nH n fix pacnpocTpaiieHHa.

Hanp0T~B, onpeRenamn~ll ~a~T0p

CBH3aH IIO-BHKHMoMy CO BpeMeHeM, Heo6xoKHMt,IM KJIH BI~I~eJIeHHH B MecTaX Re~eRTOB pemeTgn aTOMOBaaoTa H ~ p y r n x HO~OSHhIX spe~H~rX IIp~Mecel~; aTOMy IIp0~4eccy CHOCOSCTByeT

*Manuscript received 3~Soptembor 1964. 291

292

H . H . UHUG and J. P. SAVA

noB~memrarl TeMnepaTypa. Tar~oe cuon~erme auoTa Bhlah[BaeT a~eleT cTapeHHA I4 cos~aeT riFTS4, Har~5o~ee 6~IaronpnATa~e ~ pa8BHT~4A Tpe~lH. Tag, HanpHMep, aycTeHHTHaA nepmaBetom~aA CTa~b, cogepmalaaA 20~/o Cr, 19~o Ni H MaJIoe I¢O~Hqecwno a3oTa He cTapeew H ~e crc~om, a r¢ Koppoat~OmmMy pacwpecrcHBattt,m no~l HanpameHHe~. INTRODUCTION WHEN a stressed austenitic stainless steel such as t y p e 310 (25~o Cr, 20~o Ni) is exposed to boiling MgCl~, surface cracks are n o t observed immediately, b u t only after a certain p e r i o d o f exposure. T i m e to failure, therefore, consists o f an i n d u c t i o n time for cracks to initiate, plus the time for c r a c k p r o p a g a t i o n . F o r one c o m m e r c i a l type 310 alloy presently studied, the i n d u c t i o n time, as d e t e r m i n e d by m i c r o s c o p i c e x a m i n a t i o n o f plastically d e f o r m e d specimens exposed to MgC12 boiling at 154°C, was a b o u t l h (Pig. 1). T o t a l time to c o m p l e t e fracture o f a specimen 0.030-in thick was a b o u t 4½ h. H o a r and Hines ~ t h r o u g h potential and extension m e a s u r e m e n t s also observed a delay time in the case o f 18-8 wires exposed to h o t 42 ~o MgCl2. This i n d u c t i o n time decreased with applied stress a n d with a d d i t i o n o f h y d r o c h l o r i c acid to the MgCl~

FIG. 1. Stressed typo 310 stainless stool exposed to MgCI, boiling at 154°C showing an induction time f o r stress corrosion cracking. (a) exposed 45 rain showing no cracks. (b) exposed 60 min showing cracks on convex surface only.

Origin of delay time in stress corrosion cracking of austenitic stainless steels

293

solution. Uhlig and Lincoln, 2 on the other hand, using plastically deformed specimens of 18-8 found no evidence of an induction time. Instead, cracks appeared under the microscope within five minutes or less on exposure to the MgC12 solution, the total time to failure consisting, therefore, of crack propagation. It is supposed that cold work in this instance shortened the induction time to negligible values. In higher Ni stainless steels, on the other hand, the induction time is appreciable regardless of cold work or level of applied stress. It is of interest to the mechanism of stress corrosion cracking to know the reasons for this induction time or observed delay in appearance of cracks. Hoar and Hines ascribed the delay to the time necessary for damage of a surface oxide film by MgCl2 solution. They exposed 18-8 wires, unstressed, to MgCl~ for 75 rain, followed by the usual exposure with application of stress, reporting that total time to failure was the same as if stress had been applied throughout the test. They concluded that the effect of stress is important only after an initial period during which metal is acted upon by MgCl2. They also exposed wires to air at room temperature for 6 days, to air at 200°C for 10 h, and to air at 400°C for 2 h observing longer average cracking time for the 200°C exposure compared to air exposure at room temperature or at 400°C. This they ascribed to build-up of a more resistant surface film at 200°C compared to lower or higher temperatures. The effect of HCI additions to MgCI2 in eliminating the induction period was consistent with this explanation. However, in the latter instance, HC1 additions also shorten the total time to fracture which may, therefore, have obscured the induction time they described. Also, pre-exposure to MgClz at 154°C or to air at above-room temperatures constitutes a process of heat treatment. It is important, therefore, to evaluate metallurgical ageing effects as well as surface film effects in any proposed mechanism accounting for observed changes in cracking time. EXPERIMENTAL PROCEDURE AND RESULTS Two series of experiments were carried out, in both cases using a commercial type 310 alloy, the analysis of which is as follows: 24"8~o Cr, 19.5 ~o Ni, 0"03~o C, 1.7~o Mn, 0"05~o N, 0"03~o P, 0"015~o S, 0"53~o Si, 0 . 4 ~ Cu, and 0"14~o Me. The test procedure was the same as that described previously. ~,a Briefly, specimens measuring 13 × ~ x 0.030 in (4.45 x 0.48 x 0.08 cm) were sheared to size from a sheet cold-rolled to about 85 per cent reduction in thickness. They were then either tested as cold-rolled, or were annealed at 1050°C for 30 min and water quenched. In either case, they were pickled in 25 vol. ~o HC1, 25 vol. ~oH2 SO4 of the commercial concentrated acids at 90°C to remove visible oxide scale, followed by a pickle in 15 vol. nitric acid, 5 vol. ~o hydrofluoric acid, based on the commercial acids, at 90°C for 5 rain. Each specimen was then washed, it was bent in a vice to a span of l ~ i n (4.1 era), transferred to the test apparatus and the final span adjusted to 1-~ in (3-65 era) and maintained at this dimension by means of a compression spring. Cracking time was reported automatically using microswitches in series with electric docks. The first series consisted of pickling cold-rolled specimens as usual, then mounting them in the test apparatus and exposing them to concentrated MgCI~ solution boiling at 154°C for a definite period either short of or just beyond the induction time necessary

294

H.H. UHLIGand J. P. SAVA

for initiation of cracks. They were then re-pickled for 5 min in the nitric-hydrofluoric acid mixture at 90°C in order to remove any oxide or passive film on the surface, washed and re-exposed to the MgCI~ solution. The second pickling was carried out maintaining the normal span of 1 ~ in using a stainless steel holder, and the specimen was re-inserted into the test apparatus carefully avoiding spring-back. This procedure was successful judging from results on "pickled specimens which were mounted, then pickled, and again mounted in the test apparatus without intermediate exposure to MgCI2. Despite some small loss of metal by pickling, these specimens showed the same cracking times, within usual experimental deviations, as specimens pickled and mounted only once. Results are plotted in Fig. 2, each point being the average cracking time for three specimens with maximum deviation of 0.2-0.4 h. Total times to cracking are plotted, and also the times as measured after the second pickling procedure. Note that the latter times decrease with initial exposure time, but that the total time to cracking, which includes the initial period of exposure, is approximately the same for all specimens. Removing the surface film during the course of exposure, whether early or late in the test, has no influence on time to failure. The second series of experiments consisted of pre-exposing sheared, annealed specimens, pickled initially, and unstressed (not bent), for various times to MgCI 2 solution at 154°C, then bending each specimen to the normal span and testing as usual. For comparison, a second series was heat treated in air at the identical temperature by enclosing annealed pickled specimens in closed test tubes submerged in boiling MgClz at 154°C. Comparison of the two series should indicate, therefore, any effect of the MgCI~ solution on an oxide or passive film during the pre-exposure period. Results are plotted in Fig. 3, each point representing the average of three specimens. Maximum deviations from averages range from 0.1 to 1.1 h. It is observed that the cracking time decreases for short periods of heat treatment at 154°C, and then increases again for longer periods. The effect is identical whether the specimens are heat treated by pre-exposure to MgC12 or to air at the same temperature. 5-0

t

'

I

I

i

I

Time

D

4.5 (

,

r~

n

\Total

~ , Q

40

~3.5 o

-

T!~ after

to Fa i lure

to ~-oilu~e Seco~Pickle

-

~

~ 3-0 i= 2.52.0

i

0

I

i

I

i

I

• 0'5 I'0 1"5 Time of Immersion in MgCl2before

Second

Pickle,

h

FIO. 2. Effect of pickling off oxide film on type 310 stainless steel at various times during oxposurv to MgCln, 154°C, on total time to failure (cold-rolled specimens).

Origin of delay time in stress corrosion cracking or austenitic stainless steels

295

7[ 6 5

4 ta_ o oJ

E

3 -2 --

I --

0

Pre-exposed to MgCl 2 , 1 5 4 ° C

o

Heat Treated

! 0

I0 2~0 Heat Treotmenl

i

in Air, 154°C

I

I

.50 40 50 60 or Pre-Exposure Time, h

70

FIG. 3. Effectof previous heat treatment of type 310 stainless steel in air or pro-exposure to boiling MgCI2 at 154°C on time to failure in boiling MgCI~ (armoaled, quenched specimens). DISCUSSION Data of Fig. 3 show that time to failure depends only on total previous exposure time of unstressed specimens to a temperature of 154°C whether the environment is MgC12 solution or air. Data of Fig. 2, along the same lines, show that cracking time is not altered by interrupting the test early or late, then re-pickling the stressed specimens and re-exposing them to the boiling MgCI2 solution. The present results, therefore, do not support any supposed role of an oxide or passive film in determining time to failure by stress corrosion cracking in MgC12 solution. Changes in structure of the alloy through heat treatment at 154°C appear to be the important immediate cause of observed trends in time to failure as shown in Fig. 3. Such effects very likely also entered the experiments described by Hoar and Hines. In the present experiments, one reason for structure entering as a factor is that rate of crack propagation depends on the magnitude of tensional stress acting on the specimen. This was shown, for example, by data of Uhlig and Lincoln 2 who reported a rapid initial rate (0.5-1.0 cm/h) of crack propagation in 18-8 stainless steel until the crack reached the mid-section of the bent test specimen where stresses change from tension to compression. Because cracking does not occur in presence of compressive stresses, the rate of cracking at this point became very much less (0.0035 cm/h) than the initial rate. After a realignment of stresses, and consolidation of numerous cracks, the rate again became high until the crack penetrated the entire specimen thickness. Hence, any factor which increases flow stress of a specimen would, under the present test arrangement, increase tensional stress on the convex side and decrease time to failure. This effect occurs because the total load on the specimen necesssary to provide a standard span of I ~ in is greater the higher the flow stress. The effect of load on rate of cracking was also demonstrated by testing specimens of variable thickness. A thin specimen supporting a lighter load cracks in longer time compared to a thicker specimen supporting a heavier load, even though the reverse situation might be assumed intuitively. For example, in the present tests the average cracking time (3 specimens) of annealed, quenched type 310 specimens 0.02 in thick was 8.4 4- 0.5 h, for 0.03 in thick specimens it was 6"7 4- 1.i h, and for 0.04 in thick specimens it was 3.9 -t- 1.2 h.

296

H . H . UHLIC and J. P. SAVA

Hence, the decrease of cracking time on heat treatment, followed again by an increase of cracking time, suggests a metallurgical ageing process in which the alloy becomes initially stronger, followed by an over-ageing process leading again to decreased strength. This suspicion was confirmed by measuring the total load which is necessary to apply to specimens, aged or overaged, to maintain the usual span of 1~ in, identical to conditions used in the cracl
£~

I

Io

I

I

I

I

I

Type 310 (24.8 VoCr,19.5°YoNi,0.03%C, 005% N) ~ 15OOl~_ ~ O~03in. Thick I

~

LowNitrogen(18.9%Cr,200%NI,0001% C, o

1

I

I0

I

I

I-"-o4

.

_

I

I

20 50 40 50 60 70 Time of Heot Treolment in Air o) 154°C

1

80

Fie. 4. Relationof previousheat treatment of stainless steels in air at 154°Con load required to maintain standard span of 1~ in. in test for stress corrosion cracking.

Origin of delay time in stress corrosion cracking of austenitic stainless steels

297

A g e i n g o f t h e p r e s e n t t y p e 310 stainless steel is p r o b a b l y c a u s e d m o s t l y b y s e g r e g a t i o n o f N a t o m s at d i s l o c a t i o n s a n d b y p r e c i p i t a t i o n o f nitrides f r o m a u s t e n i t e s u p e r s a t u r a t e d w i t h n i t r o g e n at 154°C. O v e r - a g e i n g , r e s u l t i n g p e r h a p s f r o m g r o w t h o f n i t r i d e s t o l a r g e r p a r t i c l e size, a g a i n d e c r e a s e s t h e flow stress o f the. alloy, a n d also d e c r e a s e s susceptibility to stress c o r r o s i o n c r a c k i n g , b u t t h e c r a c k - s e n s i t i v e p a t h s r e m a i n . S u s c e p t i b i l i t y is a p p a r e n t l y e l i m i n a t e d o n l y i f t h e n i t r o g e n c o n t e n t a n d t h a t o f s i m i l a r d a m a g i n g i m p u r i t i e s are r e d u c e d t o l o w levels. Acknowledgement--This research was supported by the Office of Ordnance Research on Contract No. DA-19-020-ORD-5247 to whom the authors express their appreciation.

REFERENCES 1. T. HoAR and J. Hn,ms, J. Iron Steel Inst. 182, 124 (1956). 2. H. U m a o and J. L~coLtq, JR., J. Electrochem. Soc. 105, 325 (1958). 3. H. Urn.io and R. WmT~, Trans. Amer. Soc. Mineral. 52, 830 (1960). 4. e.g. As, P, Sb, Bi, Ru when present in 18 % Cr, 20% Ni stainless steels are also damaging according to F. S. LArqo, Corrosion 18, 378t (1962). DISCUSSION BY J. G. HINES Uhlig and Sava 1 mention a discrepancy between their results and those reported previously by Hoar and Hines, 2's but in my opinion dismiss this discrepancy rather too easily. The difference is too serious to be ignored, although it may well be that it is more apparent than real, in that the two sets of results do not refer to the same thing. It seems possible that comparison of the two sets of results may lead to a better understanding of both. Uhlig and Sava remark that when U-bend specimens of 18/8 steels are exposed to 4.2Yo Mg CI~ solution the induction or delay period is short, in contrast both with their results on 25]20 0-hend specimens and ours on axially loaded 18/8 wires. In fact we obtained a variety of results on 18/8 type steels. On heavily deformed specimens stressed at or near their proof stress (raised by cold working), the induction period was short 2,s and only slightly altered by changes in specimen preparation or test conditions; the behaviour was thus generally similar to that of U-bend specimens, 1'4 With softened specimens tested at stresses near the proof stress, the induction period was longer, usually around 1-2 h, and was sensitive to treatments that altered the "thickness" of the air-formed film and to the addition of acid or oxidizing agents to the test solution) On softened specimens tested at stresses well below the proof stress the induction period was longer (up to several days) and again insensitive to specimen preparation (except for accidental or intentional cold work) or changes in the environment.S Our original interpretation of our results was in terms of the classical breakdown and repairsequenee of oxide films developed to explain pitting corrosion, 5 and indeed at that time all the results appeared consistent with this explanation. It was assumed that cracks formed only after local film-breakdown (in the classical sense) occurred. Thus with heavily deformed specimens, the oxide film was taken to have been sufficiently disrupted mechanically to allow crack initiation almost at once, whereas with undeformed specimens tested near the proof stress film breakdown occurred by chemical means and the induction period was sensitive to the initial state of the film and the nature of the environment In both cases crack initiation occurred once local film breakdown was achieved, but with softened specimens at low stresses initiation was delayed after film breakdown because suitable sites for crack initiation were not present but were produced by a combination of corrosion and local yielding. At low stresses the second stage predominates and the induction period becomes insensitive to initial chemical factors. This simple picture is, of course, no longer acceptable, as it now seems certain that oxide films in the usual sense cannot exist for long in hot 42~o MgCI2 solution; thus Hoar and West 8 report that active dissolution occurs. Moreover, it does not directly explain the results now reported on 25/20. On the other hand it does not seem possible to explain all our results on the diffusion model proposed by Uhlig and Sara; in particular, the marked effect of addition of acid to the test solution does not involve any "heat treatment" but reduces the time to fracture to a value much less than the induction period under normal conditions, and thus, is completely inconsistent with their model. It seems, therefore, that there are two separate effects, a chemical surface conditioning and a mechanical conditioning. With 18/8 steels tested near or above the proof stress for softened material

298

H.H.

UmaG and J. P. SAVA

the mechanical features are present from the start, presumably as some form of dislocation arrays, but with softened material at low stresses or with more resistant alloys the mechanical stage involves the formation or conditioning (e.g. by diffusion) of appropriate dislocation arrays. Cracking can occur only in a relatively narrow potential range, 7,8 and it may well be that the essential feature of the chemical stage is the establishment of this potential by reactions involving the oxide film, preferential dissolution of iron and chromium to leave a nickel-rich surface, etc. This suggestion is supported by the fact that if excess HCI is added the pol~ential is in the cracking range from the start. REFERENCES 1. H. H. UI~LXGand J. P. SAVA,Corros. Sci. 5, 291 (1965). 2. T. P. HOAR and J. G. HiNES, J. Iron St. Inst. 182, 124 0956). 3. J. G. Hil,o~s, Corros. Sci. 1, 2 (1961). 4, J. G. HiNES, unpublished work (1956). 5. T. P. HOAR, Trans. Faraday Soc. 45, 683 (1949). 6. T. P. HOAR and J. M. WEST, Proc. Roy. Soc. (A) 268, 304 (1962). 7. J. G. HINES, Corros. Sci. 1, 21 (1961). 8. S. BRENNART, Jernkont. Annlr 144, 560 (1960).

REPLY BY H. H. UHLIG Dr. Hines' comments are appreciated regarding the effect of stress on induction (and presumably also crack propagation) times. Our results show similarly that crack initiation and growth are a function of applied stress. This, in fact, might also have been inferred from early observations of stress corrosion cracking by others, that without applied or residual tensile stress metals do not undergo failure by cracking whatever the environment. It is quite in order to examine the question of why stress is necessary for cracking to occur and to this end Dr. Hines has made several suggestions. We show in our present paper that metallurgical factors are a more important consideration for explaining the initiation and growth of cracks in stainless steels as compared with supposed breakdown of surface films. In our experiments the environment was kept constant, as indeed was required in order to reach suitable conclusions, but it was not implied thereby that varying the environment would not alter cracking time. In an earlier paper 1 we showed the appreciable part played by environment, including the effect of pH, on cracking of 18/8 stainless steel. We concur, therefore, with Dr. Hines that there are two separate effects. We would prefer to designate them as (1) environmental and (2) metallurgical. The importance of the metallurgical effect is shown by the presently reported results on effect of ageing, and the critical role played by nitrogen in the 25/20 stainless steels. The important part played by alloyed nickel, as shown by Copson, 2 might also be mentioned. For cracking to occur of any susceptible alloy, the environment, as mentioned above, is also critical. In general, austenitic stainless steels will crack only in presence of Cl- or OH-, and furthermore, the lower the concentration of either ion the longer is the time for cracking to appear. It is also known that the addition of depolarizers, e.g. Fe 3+ to MgCl~ solution, increases rate of cracking) ,s and the effect of pH is such as to increase cracking time in the alkaline region and to decrease cracking time in the acid region. ~ The latter effect includes the effect of HC1 additions mentioned by Dr. Hines. Since the mechanism accounting for effects of pre-exposure to MgCl~ solution cannot be ascribed to surface oxide films, there are two possibilities, either one or both of which may operate. The first possibility is that depolarizers or acid additions to magnesium chloride increase electrolyte conductivity or increase the potential difference between the apex and the walls of the growing crack in accord with the usually described elecrochemical mechanism of stress corrosion cracking. The second possibility is that oxidizing substances or increase of hydrogen ion activity (within a limited range) shift the surface potential of the metal in a direction which favours chemisorption of Cl- on imperfections of the plastically deformed metal either at incipient cracks or at the apex of a growing crack. A specific array of imperfections and their association with specific impurities of the metal are apparently essential for successful chemisorption of components of the environment. At the same time, the environment must contain those components that are chemically able to chemisorb on the imperfection arrays. The latter mechanism has been called "stress sorption cracking ''~ and relates to the tendency of chemisorbed films to reduce surface energy of the metal and thereby to allow metal atoms to separate under tensile stress, forming a crack. Applying a potential to the metal, as in cathodic protection, influences the extent to which ions can adsorb on the alloy surface, as well as influencing the corrosion process itself. It may well be true in some systems that both stress sorption cracking and the electrochemical mechanism operate jointly, the latter being necessary to account for

Origin of delay time in stress corrosion cracking of austenitic stainless steels

299

formation of a metal reaction product which is then the substance actually adsorbing on the metal surface leading to a crack, rather than adsorption of CI- itself. REFERENCES 1. H. H. UHLIG and J. LINCOLN, Jr., J. Electrochem. Soc. 105, 325 0958). 2. H. COPSON, in Physical Metallurgy of Stress Corrosion Fracture (ed. T. Rhodin), p. 247. Interscience, New York (1959). 3. C. EDELEANU, J. Iron St. Inst. 173, 140 (1953). 4. H. H. UHHG, in Physical Metallurgy of Stress Corrosion Fracture (ed. T. Rhodin), pp. 1-17. Interscience, New York (1959).