Stress-corrosion cracking of high strength steels

Stress-corrosion cracking of high strength steels

Corrosion Science, 1968, Vol. 8, pp. 359 to 375. Pergamon Press. Printed in Great Britain REVIEW ARTICLE STRESS-CORROSION CRACKING OF HIGH STRENGTH S...

922KB Sizes 13 Downloads 121 Views

Corrosion Science, 1968, Vol. 8, pp. 359 to 375. Pergamon Press. Printed in Great Britain

REVIEW ARTICLE STRESS-CORROSION CRACKING OF HIGH STRENGTH STEELS* J. W. KENNEDY and J. A. WHITTAKER t Fulmer Research Institute, Stoke Poges, Bucks, England INTRODUCTION HIGH strength steels are being used in increasing quantities for aircraft components and are being considered for other applications where their high strength-weight ratio is required. When selecting such a material the primary requirements are usually mechanical properties, dimensional stability, machinability and fabricability. In almost all applications, however, the components will, at some time or another, come into contact with corrosive conditions and it is therefore necessary to take into consideration a cracking phenomena, believed to be stress-corrosion cracking (scc), which this group of alloys can exhibit to varying degrees. In addition to scc, high strength steels are also susceptible to other types of brittle fracture, particularly H embrittlement cracking, and brittle fracture which arises because the materials are notch ~ensitive. The difficulty of distinguishing between scc and H embrittlement cracking under service conditions was illustrated by Shank 1 who reported on the failure of rocket chambers during hydrostatic pressure tests. Extensive investigations showed that in this particular case H embrittlement was responsible; it was assumed that H was generated by corrosion. It is extremely important to be able to predict the mechanism of possible failure, either scc or H embrittlement, because this will determine the protective treatment to be used. If the mechanism of failure can be shown to be scc, then an anodic metal coating will provide protection at defects in the coating. If, however, the mechanism of failure is H embrittlement, the use of an anodic metal coating may be deleterious as any defect in the coating will become a cathodic site, and H evolved at this site may enter the steel and lead to delayed failure. For steels susceptible to H embrittlement, a protective treatment such as phosphating and painting may be more suitable. The work described in the present paper is referred to throughout as stresscorrosion but the only justification for this term, in the majority of cases, is that the experimental techniques which have been adopted are similar to those used in studying other more generally accepted stress-corrosion systems. These techniques involve the specimens being simultaneously stressed and exposed to the test environment, usually aqueous solutions, and they are to be contrasted with conventional H embrittlement *Manuscript received 14 March 1967. Received in revised form 21 September 1967. tNow at British Aluminium Ltd. 359

360

J.W. KEr,rt,mDV and J. A. WHITTAKER

techniques in which specimens are stressed in uncontrolled laboratory atmospheres after prior charging with H (from the gas phase or by cathodic polarization in aqueous solutions). It m a y well be that H evolved in the corrosion reactions is responsible for the failure even in the former case. F o r the purpose of this report, the term high strength steel has been interpreted as referring to materials with yield strengths in excess of 100 ton/in 2, although this arbitrary definition has not been rigidly adhered to throughout. EXPERIMENTAL VARIABLES There are a number o f variables to be controlled or pre-determined in any investigation concerning scc, i.e. stress, environment, alloy composition. Other factors such as the configuration of the specimen, its size, and its orientation with respect to the principal working direction of the metal, grain size, its metallurgical history, and test temperature, change only the degree of scc.

1. Stress level The high strength steels have tended to be used for applications where the design stresses are high, sometimes approaching the yield stress. This has resulted in investigators using high elastic, or even plastic, stresses when comparing or evaluating steels. There seems to have been little or no effort to determine whether any of these steels have a threshold stress, i.e. a stress below which failure does not occur. This is probably a reflection of the original high design stress requirement,but if these steels are to become more widely used commercially, tests at the lower stress levels could be important. Phelps and Loginow 2 investigated the effect o f stress by exposing various steels at 50, 75 and 95% of yield stress at the Kure Beach Site (Table 1). As expected, the results show that, in general, as the stress level is increased the life decreases, but Airsteel X200 and the hot-worked die steel show approximately the same life regardless o f the stress level. It is also seen that a possible threshold stress m a y exist for semiaustenitic stainless steel Type A, hardened by refrigeration, and then aged at 950°F. TXBLE 1. EffECTOF STRESSLEVELON STRESS-CORROSION Steel

Y.S. ton/in 2

Time to Failure (d) 50 ~* 75 ~* 90 ~0"

USS. Airsteel × 200 AISI. 4340 MBMC. No. 1 (long) MBMC. No. 1 (trans.) AISI. 4335 q- V (long) AISL 4335 + V (trans.) Hot work die steel type A USS. 12 Mo-V stainless USS. Stainless W Precipitation-hardenable stainless type A (T950) Precipitation-hardenable stainless type A (R950) Precipitation-hardenable stainless type B (R950)

106 97.3 113"5 115 97.8 115 107 91-5 90"0 90.0 91"0 90"5

6 12 28 16 NF NF 6 1"4 65 5 16 NF

*Stress as percentage of yield stress. NF: Not Failed in 175 d.

6 10 5 8 NF 25 9 1 11 1 5 15

8 8 7 5 NF 15 8 1 22 1 2 12

Stress-corrosion cracking of high strength steels

361

Davies 8 also examined the effect of stress on time to failure of several steels and the same general trend of increased stress level leading to shorter life was observed. As, however, the test times were relatively short, I000 h maximum, it is difficult to say whether a threshold stress exists for any of the steels. Rubin 4 also examined the effect of stress level on scc for various maraging Steels and for 1 ~o Cr steel. All steels showed increased life with decreasing stress, and there is little evidence of a threshold stress.

2. Environment Phelps and Bhatt 5 investigated the effect of pH on the scc of a 12 C r - M o - V steel. Specimens were stressed to 75 ~o of the yield strength and immersed in 3 ~o NaCI saturated with O, the pH being adjusted with HCI or NaOH. It was found that at pH 1 scc occurred relatively rapidly and H was evolved. The time to failure gradually increased and the degree of corrosion decreased as the pH was increased from 3-11. At pH levels > 11 scc did not occur and corrosion rates were negligible. When specimens were anodically polarized, scc occurred at pH 1, 6.5 and 12.5. Rubin 4 also studied the effect of pH on scc. Centre notch specimens of 18 ~o Ni, maraging steel and a 5~o Cr-steel, were stressed to 75~o of the notch tensile 0.1 Yo proof stress and then exposed to 3 ~o NaC1 solution, pH 3-11. The general trend of results shows that below pH 5 the life decreased whereas above pH 9 there was some increase, but there was little effect on life between pH 5-9. It is also interesting to note that pH has less effect on the 5 ~o Cr-steel than on the 18 ~o Ni-maraging steels with the exception of RMS 200. The RMS 200 steel, which was superior to all other maraging steels and to the 5 ~o Cr-steel, appeared to differ only in respect of its Ti and B content. The Ti level was 0.23 ~ compared with 0.5 ~o average for the other steels, and B was not detected at all, whilst the level of B in the other maraging steels was 0.003 ~o. This aspect of the composition of maraging steels will be discussed subsequently. 3. Composition As has been stated previously, every alloy steel is susceptible in a given specific environment, and it is difficult to generalize on the effect of alloying elements. However, the effect of the minor alloying elements on the see susceptibility of the high strength steels has been the subject of many investigations. Perry e found that of the range of Cr-martensitic steels with maximum stresses in excess of 90 ton/in ~, only FV 566 and high-C FV 448 had a high resistance to cracking. The main alloying elements were Mo, V and Nb, and the specimens were tested in 3 ~o NaC1. The alloy which contained both Mo and V showed a great improvement compared to alloys containing only one of these elements. Rubin 4 examined the effects of composition variations on the scc susceptibility of 18 ~ Ni-maraging steels. The main constituent to be varied was Ti which ranged from 0.23 Yo to 1-0 ~o. The results of tests on bent beams indicate that susceptibility increases with increase in Ti content. Dean and Copson 7 found that susceptibility of a 25 % Ni-maraging alloy increased when the Ti content was high and when Zr and B were absent. It seemed that failure

362

J.W. KENNEDYand J. A. Wm'rrAr.r.R

was by intergranular cracking in the absence of Zr and B and that additions of these elements eliminated the formation of critical grain boundary precipitates. The results of Truman and Perry 8 for maraging steels Rex 539 and En 30D are of particular interest. Maraging steel "110" cast 06682, failed in an intergranular manner in an industrial atmosphere while the other maraging steels failed transgranularly. There were no noteworthy differences in heat treatment or composition which could account for this difference in the mode of cracking. Similar anomalies can also be seen in the results from tests in 3 ~ NaC1. Trace elements may play an important part in promoting cracking susceptibility by acting as cathode poisons and thereby accelerating H absorption. The presence and composition of inclusions may also be important in this respect. 9 Sulphide-bearing inclusions, in particular, may cause local poisoning of the solution, thus promoting the absorption of H.

4. Specimen configurations One of the major difficulties in attempting to correlate published data arises from the variety of specimen configurations which are used in stress-corrosion testing. The main types of specimens which have been used are: a. Tensile test pieces have the advantage of producing a uniform stress field, but have been little used because of the cost of manufacture and the cost and size of the apparatus needed to apply a constant load. b. Two-point loaded bent beam specimens have been widely used because of the simplicity of manufacture and ease of use. The major disadvantages are the lack of a uniformly stressed area and the difficulty of producing a desired stress level. c. Four-point bending specimens have the advantage of providing, nominally at least, a uniform stress over their centre section. d. Taper cantilever specimens are designed to produce a uniform stress along the specimen by tapering the width without varying the thickness. e. C-ring specimens have found favour with airframe manufacturers since they can be easily manufactured from structural tubing. f. U-bend specimens are produced by bending a fiat bar into a U-shape and using tension bolts at the open end of the U to maintain the applied stress. As with the two-point loaded bent beam specimen it has found wide application because of the simplicity of its manufacture and ease of use. However, its major disadvantage is in calculating the actual stress in the material due to the complex stress distribution. A number of investigators have used different tests for evaluating a single material, e.g. the results obtained by Setterlund x° on an 18~o Ni-maraging steel (Table 2). These results cover the use of two-point loaded bent beam and notched tensile specimens. When trying to correlate these stress-corrosion results a difficulty arises because the two-point loaded bent beam specimens are basically constant strain tests, and the stress is relieved as the cracking proceeds, whilst with the constant load type of test any effective reduction in area by cracking results in a higher stress and, therefore, an increasingly rapid failure. The limitations of the above tests methods have been well recognized and several investigators have looked at other specimens in an attempt to improve testing techniques. Davies 11 combining experimental techniques with theoretical stress analysis investigated the two-point loaded bent beam specimen, and a tapered cantilever specimen (see type d). Following these investigations, Davies designed a specimen

Stress-corrosion cracking of high strength steels

....

363

°. . . . . . . . . . .

0 U

z

[t~

. . . ~ . . ~

6 t~ ~o

0

tt~

.1 t,..,

z Z 0

;>

0

0 e~ 0

rj

.1

0

8~ .&

Z N

364

J.W. KENNEDYand J. A. WHITTAKER

having a uniform stress over a large area, the resulting specimen is referred to as a U-bend specimen, but there are no indications at present of its widespread acceptance. Another important investigation into specimen geometry was performed by Brown and Beecham. lz These investigators introduced a stress raiser of controlled geometry into the specimen, which they claim gives the following advantages: 1. The overall testing time can be shortened, as the initiation period is less. 2. Tests for scc in alloys which do not pit can be performed. 3. Having a controlled stress raiser, the state of stress at the initiation of scc can be calculated fairly accurately, which is not possible when the stress raiser is in the form of an irregular pit. The last claim appears to be of doubtful validity since pitting is not an essential pre-requisite to scc and in any case, preferential chemical attack may well take place at the base of the notch, altering its geometry. Experiments were performed using cantilever specimens of AIS1 4340 which had been heat treated to approximately 100 ton/in z before being pre-cracked in fatigue. The cracks were propagated to various depth so that varying stress intensities were achieved and the specimens were then loaded after immersion in 3.5 ~o NaC1. From the results it would seem that the threshold stress intensity is 53 kg/mm 2. The next step was to determine whether these findings could be applied to another entirely different crack geometry, and different stress conditions. Tension specimens with "thumb-nail" fatigue cracks were used and the results show that the data from the cantilever test are applicable to the "thumb-nail" flaw situation. They also show the importance of the stress factor in testing. Perry 6 also argues the need for an accelerated test in order that experimental I% HCI at room temperature (Notched specimens) x EN 56A1 o EN 56C ~-0,c. 1050"(; o EN 56D.J

I000

~

'°° ~ x

~ o

200

400

600 700

Tempering ternp, ~C

FIG I. Effect of tempering temperature on time to failure.

Stress-corrosion cracking of high strength steels 3% NaCIat

365

room temperotme

x EN 56A1 o EN 56Cpa.c. 1050°C [] EN 56DJ

x~X~

~x /

I0000 .,¢

x

I000 ~. o ~oo

i:

I0 I

200

400

Tempering temp,

FIG. 2.

600 700 eC

Effect of tempering temperature on time to failure.

scatter may be reduced and the rate of development work increased. He suggests an accelerated test for high-tensile stainless steels, the specimen form and environment being based on the mechanism of failure which has been proposed for these steels. The mechanism is dependent upon the absorption of H at the steel surface and the cracking resistance could then be affected by the following: a. Rate of H absorption in the zone of tri-axial stress. b. H level necessary for crack propagation. c. Corrosion resistance of alloy. The rate of cracking would be expected to be greater if a ready supply of H were available, as in an acid solution, and an area of tri-axial stress can be introduced by machining a notch in the specimen. To determine the effectiveness of this accelerated procedure, a series of tests were performed using En.56A, En.56C, and En.56D and the results (Fig. 1) may be compared with those of the same steel tested as plain tensile test pieces in 3 ~o NaCI (Fig. 2). The general shape of the curves is similar, the sensitivity to tempering at 450°C is clearly shown, and the materials are ranked in the same order. This accelerated test would therefore seem to be quite satisfactory for these materials under the given environmental conditions. The ultimate aim in stress-corrosion testing must be to reproduce in the laboratory the conditions to which materials will be exposed in practice. This is, perhaps, even more important with the high strength steels because, in many applications, the design stresses are very high, and the environments severe. In most applications of these high strength steels the load is constant, e.g. rocket motor casings, undercarriage components, etc.and this would seem to favour the use of the constant load type. From the

366

J.W. ~ v

and J. A. Wia-rrrArdut

Failure

Applied s t r e s s - 8 1 , O 0 0 Ib/in z

,~ 190min. /I

c~ I(;

Crock growth periodjT

/

iI

I/

° 8 6

s .

Incubation period

.

.

.

s

j_j

I{ ' - r - -

2

" t 0

ncreose due to Inltlol loodlng

I

20

I

I

40

60

Time,

FIG. 3.

I

80

I O0

min

Crack initiation and discontinuous cracking of 300 M steel, 220,000 lb/in 2, strength level, in distilled water at 68°F.

point of view of producing a uniform stress field, the tension test would seem to be preferable. In the evaluation of a range of materials it is desirable to have an accelerated test, but if this involves pre-cracking it would eliminate the all-important stage of crack initiation which may represent a considerable proportion of the total fife. The relationship between the initiation and propagation times will depend upon the stress level and environment and, as an example of this relationship in high strength steels, the results of Hanna e t aL as are reproduced in Figs 3 and 4. These show crack growthtime relationships (determined using the change in electrical resistance of the specimen) and also curves for the incubation time and time to failure vs. applied stress for a 300 M steel heat treated to 220,000 lb]in ~, in distilled water (68°F). The rate of crack propagation varies with the applied stress, and at 100,000 lb]in 2 the rate of cracking is approximately 8 ram/h, while at 40,000 lb/in ~ the rate is reduced to 1 mm/hr. Cracking was by discontinuous growth, and in most cases crack propagation was accompanied by "audible noises" which corresponded with discontinuities in the resistance curve. The authors concluded that the mechanism was one of mechanical failure alternating with slow stages of reinitiation. There would seem to be an advantage in having a specimen incorporating a stressconcentration factor which is representative of that found in practice. It is thought that if this is to be incorporated the value should be K r -- 3, being of the order of that commonly used for fatigue test work. The influence of notch acuity was also investigated by Hanna et al. is who found that the time to failure showed a dependence on notch sharpness similar to that shown by the notch tensile strength. This could also be important when the mechanism of failure can be attributed to H embrittleme.nt, as it has been indicated that the presence of a stress gradient can be an important factor. Thus it seems that to produce see test results which would be meaningful in

Stress-corrosioncracking of high strength steels

N= 1

.= 00 o =

6 0 ~ 1 2 0

367

time

°

o

4O I I

I0 I

I00 I

Time, min

I000 I

I00000

FIo. 4. Delayed failure characteristics of 330 M steel, 220,000lb/in =, strength value, showing incubation times in distilled water at 68°F.

practice, the most desirable rest would be a constant load tension test using a specimen which incorporates a stress-concentration factor of approximately 3.0. 5. H e a t t r e a t m e n t and microstructure

Tempering temperature affects the microstructure which, in turn, affects the scc susceptibility. The general trend is that the higher the tempering temperature, the higher the resistance to scc. An example of this is the results which Phelps and Loginow 2 obtained for 127o Cr stainless steel and Airsteel X200. Hanna et aL 13 tested A.I.S.I. 4340 steel in distilled water at three strength levels. The results show, apart from the short life failures, thatthe most susceptible condition of this steel is the 600°F temper condition. Perry 6 tested a number of high tensile stainless steels, among these was En.56D (0.1 70 proof stress 69 ton/in 2, U.T.S.109 ton f/in 2. This steel was tested in several conditions of heat treatment giving a range of tensile strengths and the results of testing in 1 70 HC1 at room temperature show (Fig. 1) that tempering at 450°C gives the steel its most sensitive condition. Hughes et al. 9 tested a Ni-Cr steel (En.26), in six different temper conditions and in two environments, 370 NaC1 and 0.1N HC1. The results show that a threshold stress appears to exist for each temper condition, particularly in HC1. It was clearly shown that the susceptibility if the steel is greater in the short transverse direction than in the longitudinal. When tempered at 330°C, the susceptibility of the steel in the long transverse direction lies somewhere between these two, and Hughes reports the same overall trend with specimens tempered at 160°C and 235°C. Davies et al. 3 investigated low alloy steels, by alternate immersion in 3.5 70 NaCI and all steels appear to show good resistance to scc at an U.T.S. below 200 kg/in 2. Above this, the steels were most resistant in rather different strength ranges, e.g.

368

J.W. Km,rrc~Y and J. A. WHITTAKER

4340 M in the range 270-300 kg/in ~, 4330 M in the range 240-260 kg/in ~ and H.11 in the range 220-240 kg/in ~. Setterlund 1° examined one heat of 2 0 ~ Ni-maraging steel, and five heats of 1 8 ~ Ni-maraging steel. These were tested in three metallurgical conditions: (a) annealed and maraged; (b) cold reduced and maraged; (c) welded and maraged. The results for bent-beam specimens and centre notch tensile test specimens tested in several different environments, show both steels to be susceptible in all conditions of heat treatment, but not in all environments, e.g. the 2 0 ~ Ni-maraging steel was not susceptible in laboratory air or in 4 ~o soluble oil. It was found by Dean and Copson 7 that the resistance of maraging steels to scc was influenced by the processing history. Specimens were given different heat treatments, some of which represented bad practice, and were then exposed to an aerated 3.5 NaC1 solution (pH 6). Specimens given the standard anneal and maraging treatment showed no sign of cracking after 30 d. Specimens given the 2200°F solution annealing showed marked grain growth and cracked very rapidly. Precipitation at grain boundaries was more marked after this treatment, which would be expected to increase the susceptibility to scc. These authors suggest that cold rolling before maraging breaks up the coarse grain structure and is, therefore, beneficial. Good scc resistance would seem to be favoured by small grain size, grain boundaries free from harmful precipitates, and the absence of composition gradients in the vicinity of prior austenite grain boundaries. Rubin 4 also noted grain size variations encountered with 18 ~ Ni-maraging steels, but the examples which he gave were not accompanied by details or descriptions of processing differences which would account for these variations. Briggs la tested welded panels of A.I.S.I. 4130 in which the corrosion susceptibility was different for welded zone and base metal. It was suggested that slow cooling from the austenite temperature allowed more Cr to diffuse to cementite leaving the ferrite more anodic. Increasing the cooling rate increased the resistance to corrosion. The relationship between mode of cracking and. microstructure is not a uniform picture for each class of steels, a large number of variables controlling the microstructure along with variables encountered in various tests creates a situation which does not permit a simple uniform explanation. In the case of low alloy steels, it has proved difficult to satisfactorily etch the microstructure to show the grain boundaries and this has handicapped researchers in observing microstructural effects. 6. Correlation o f sce susceptibility with mechanical properties Davies et al. a in their evaluation of high strength steels, also measured Izod impact

values. From these tests it could be seen immediately that increased scc susceptibility was related to the minimum value of the impact energy-tempering temperature curve. Electron microscope studies of the surface of specimens fractured by scc and in impact tests, in the most suceptible temper condition, showed similar intergranular failure. Brown 15 suggests, however, that this grouping of scc, Izod impact, and notch sensitivity can be misleading, as the modes of fracture may be affected differently by heat treatment and composition. Davies et al. a also examined correlation between scc susceptibility and the notched-unnotched tensile strength ratio. Specimens used had a stress-concentration

Stress-corrosion cracking of high strength steels

369

factor of 9.5. The time to failure of specimens tested at each stress level vs. the notchedunnotched ratio gave a linear curve, the slope of which depended upon stress level, being greatest for the highest stress. In general, the results showed the time to failure increased as the notched-unnotched tensile stress ratio increased. Over the test period of 1000 h, it was found no specimens failed where the ratio was in excess of 1 t 20. Thus, it would seem there are grounds for further investigations with regard to a possible correlation, particularly if the mechanical test can reproduce a change in the mode of cracking, i.e. from intergranular to transgranular as can sometimes occur after a particular heat treatment. It is possible that this change in mode of fracture may be related to grain size. PREVENTION OF STRESS-CORROSION C R A C K I N G Scc can be a major problem when using high strength steels, and it is, therefore, necessary to give a good deal of attention to the detailed design, such as the elimination of residual surface tensile stresses, and to the use of protective treatments particularly those involving cathodic processes.

1. Surface stresses It has been generally accepted that the best method of removing surface tensile stresses is to employ shot peening to produce compressive surface stresses. A typical advantage claimed for shot peening is reported by Suss 16 who observed a greater resistance to scc by A.I.S.I. 410, hardened stainless steel in high purity water at 300°F after this operation. Unpeened specimens failed within a week, but there were no failure of peened specimens in 8 weeks. He concluded that at 45,000 Ib/in 2 (.~ yield stress A.I.S.I. 410), shot peening would eliminate scc completely. Davies n also noted an improvement in resistance to scc on shot peening 4330 M and 4340 steel. The presence of compressive stresses only seemed a partial explanation because other finishing treatments which were evaluated also showed a residual compressive surface stress without showing the expected advantage in terms of specimen life. On examining the specimen surfaces, it was found that those which were most susceptible to scc had a similar surface morphology. Metallographic sections showed that the metal had apparently been " t o r n " from the surface leaving sharp "crevices", which could be seen clearly with the ground and chemically milled surfaces. The best resistance to scc was a surface structure where the " t o r n " metal surface had been "sealed-up", i.e. where there were 1no "crevices". It was also indicated from these results that H may have been a contributory factor in the failure of specimens finished by chemical milling and electropolishing.

2. Coatings Phelps and Loginow 2 tested 127o C r - M o - V steels and a 5 ~ Cr hot worked die steel, using a wide range of coatings, and subjected them to marine and semi-industrial environments. Failure time at both locations was increased when coatings anodic to the base metal were used. These include Zn di-butyl titanate primers, sprayed AI, and a N i - C d electroplate. The coatings which were cathodic to the underlying steel, i.e. electroplated Cr and electroless Ni on both steels and electroplated Ni on the hot worked die steel, did not prevent or delay cracking. One exception to this general

370

J.W. ~ Y

and J. A. WhqTrAKER

rule was electroless Ni on 12 ~ C t - M o - V steels which increased the time to failure. It is possible that this coating was sufficiently thick and free from defects that it provided an impermeable mechanical barrier. Protected specimens were also tested by Setterlund 1° and Rubin3 Setterlund evaluated a range of coatings on a 5~o Cr-steel (H.11), and the most successful of these were used by Rubin to protect 18 ~ Ni-maraging steels. The results for the most successful coatings are shown in Table 3. Each of these coatings protects by a different mechanism, the polyurethane simply forms a dense barrier between the steel and the environment, the inorganic zinc coating provides cathodic protection of the base material and the inhibited epoxy coating protects the metal by inhibitory action of the chromate compounds as well as by forming a barrier. TABLE 3.

BENT-BEAM STRESS=CORROSION TESTS FOR COATING EVALUATION

3 ~ NaCI Soln. Base material

Coating

Failure ratio

H.11 H.11 HAl H.11

None Polyurethane 500 Inorganic zinc-ll Inhibited epoxy 454-1-1

4 3 2 0

:4 :3 :2 :2

18~ Ni-Maraging 18~o Ni-Maraging 18~ Ni-Maraging 18~ Ni-Maraging

None Polyurethane X 500 Inorganic zinc-11 Inhibited epoxy 454-1-I

3 :3 1:3 3 :3 0 :3

Median failure time (h) 1.5 1380 687 NF 3100 119 4488 288 N= 4990

140°F water-sat, air Failure ratio 2 6 2 3

Medium failure time (h)

:2 :6 :2 :3

64 3500 821 2720

3:3 3:3 3:3 2 :3

535 1560 140 1870

For both steels, but particularly the H.11 steel, all these protective systems extended the life very appreciably. The inhibited epoxy and polyurethane were superior in both environments. Coatings which rely on providing cathodic protection for the underlying steel, can if damaged by scratching, or chipping, or if the coating is porous, give rise to failure by another mechanism. These defects could provide a cathodic site for the formation and absorption of H, generated by the corrosion reaction, and failure could be brought about by H embrittlement cracking. This possibility was recognized by Phelps and Loginow 2 and they tested specimens in the scratched and unscratched conditions, but reported the performance for both conditions to be approximately the same. MECHANISM OF F A I L U R E As was indicated earlier, it could be of considerable practical importance to determine whether the delayed fracture phenomenon of the various high strength steels is caused by see or by H embrittlement since this would indicate the most suitable protective treatment to be applied to the steel. The probability of failure by scc would

Stress-corrosion cracking of high strength steels

371

be reduced by the application of an anodic metal coating such as Zn, AI, or possibly Cd, but such a protective scheme might be quite unsuitable if H embrittlement were the cause of cracking. If embrittlement was caused by H produced by the cathodic corrosion reaction it might be preferable to use a phosphate coating followed by a paint scheme. The usual post phosphating baking treatments would, of course, be applied. For those ultra high strength materials which are still in the development stage, it could well be that a non-aqueous phosphating treatment, indeed complete nonaqueous finishing treatments, will be necessary. A number of workers have attempted to distinguish between these two causes of failure and two techniques have been used. The first involves the changes in endurance produced by electrochemical polarization, principally cathodic polarization, on stressed specimens totally immersed in an electrolyte, and the second involves fractographie studies. Up to the present it has not been possible using the fractographic techniques to differentiate between the two modes of failure, but the definite way in which this techniques can be used to detect other ceuses of failure, particularly fatigue, should encourage further work in this direction. The first of these techniques is based on the argument that scc, almost by definition, is dependent upon an anodic reaction and that cathodic polarization should, therefore, reduce the rate of this reaction and hence reduce susceptibility to failure. It has, in fact, been demonstrated many times for other systems, which are more generally accepted as involving scc, that cathodic polarization does delay cracking. Alternatively, it is argued that if H produced by the cathodic corrosion reaction is the principal agent causing failure then cathodic polarization by increasing the rate of the reaction should increase susceptibility to failure. Evans 17 has suggested that cathodic polarization need not necessarily reduce the susceptibility to failure in cases in which H is involved in the failure mechanism. He argues that if a single isolated cavity becomes filled with high-pressure H, the stress exerted outwards in all directions by the compressed gas, combined with a tensile stress applied from an external source, may cause a crack to start even where the applied stress, if it acted alone, would be insufficient to cause cracking. The crack will extend in a plane roughly normal to the applied stress, but the increasing volume of the crevice will cause the pressure to drop, and the rate of extension will therefore be controUed by the rate of diffusion of atomic H through the steel; it becomes high-pressure/-/2 only on arrival at the crack. If, however, the original cavity is not isolated, but is surrounded by many neighbouring cavities, each containing high-pressure H, the pressure in these cavities will tend to support the walls of the first cavity; the inception and extending of the crack from the first cavity then becomes less probable. Clearly a highly non-uniform dispersal of cavities fiUed with highpressure gas represents the most dangerous situation. This latter case may arise either because the cavities of a type likely to form at places where H will turn into/-/2 are sparsely distributed, or because parts of the surface where concentration gradients of atomic H will build up are sparsely distributed (these might be places where there is S, Se or As, capable of poisoning steps in the H evolution reaction). He further suggests that it is possible to envisage a case where the cathodic reaction of corrosion would charge only one type of sparsely distributed cavity with high-pressure gas, thus producing a dangerous situation, but that the application of cathodic polarization from an external source could charge also a different type of cavity (.perhaps smaller, but of

372

J.W. KEarny and J. A. Wm'rrAr..E~

greater frequency) and thus render the situation safer. Although he agrees that this may seem fanciful and perhaps would not often occur, it would represent a case where cathodic polarization might increase the life of a system even though H was essentially involved in the mode of failure. When the cathodic polarization technique is being used to differentiate between the two modes of failure then inferences are, in fact, being made about reactions occurring at the natural corrosion potential from observations made at more negative potentials, and because of this it is important to pay particular attention to the potential of the specimens. Figures 5 and 6 show the results obtained by Hughes e t al. 18 for potentiostatic studies of En.26. From these it was inferred that the failure at the normal potential in the chloride solution is due to scc and in the acid solution to hydrogen embrittlement. The authors note (Fig. 5) that the peak occurring at 600 mV (SCE) is at about that of the reversible H potential at pH 6 and PH~ ---- 1. Similarly the peak in Fig. 6 occurs at about the reversible H for pH 1. Thus at the natural corrosion potential in 3 ~ NaCI (pH 6) the H reaction is improbable which lends support to the evidence of the polarization experiments that scc is the cause of failure under these conditions. At pH 1, however, all the evidence points to H embrittlement. This paper is a particularly good example of the use of the technique and, although the authors do not indicate whether the points were replicated, the trends are quite clear. Brown a9 was one of the first to develop the technique, and showed that for AISI 410 stainless steel a small cathodic current greatly extended the time to failure, indicating that the mechanism was scc in the absence of impressed currents.

I0,000

.E E I000 o

1-IOO

[ potential

I -300

I

l -600 Potential

FIG. 5.

l -900 v S.CE.,

-IzO0 mV

Effect of impressed potential on time to failure under sustained load in

3 ~ NaCI, for En. 26.

Stress-corrosion cracking of high strength steels

373

I000

O

IOC E O

o o_

I--

j potential

i I I -250

I -450 Potential

FIG. 6.

I --650

v S.C.E.,

I -850

I -1050

rnV

Effect of impressed potential on time to failure under sustained load in 0"IN HCI, for En. 26.

Weibull, 2° using the technique, came to the conclusion that observed failure in non-plated bolts of high strength steel were due to H generated by corrosion in atmospheric moisture. Phelps and Loginow 2 also used the technique to investigate the failure mechanism of a 12~o C r - M o - V steel in aerated 3 ~ NaCI. A similar pattern of results to those obtained by Brown were noted, i.e. that the life of specimens was decreased under anodic polarization, but with slight cathodic polarization there was a one hundred fold increase in life indicating scc. With increasing levels of cathodic polarization the life again decreased and the probable cause of failure was then H embrittlement. Rubin 4 investigated the effect of applied potential upon failure time, his results indicating that 18 ~ Ni-maraging steel could be protected by cathodic polarization, but that the current to achieve this is quite specific. The results were taken to indicate the failure mechanism under normal conditions was by stress--corrosion. Hanna e t al. la examined the effect of cathodic polarization on specimens of 300 M steel to determine the mechanism of failure. It was found that increasing the cathodic potential decreased the time to failure, indicating H embrittlement. As further evidence, additions of As to distilled water were found to decrease the failure time by a factor of about 3. This indicated that H absorption was the rate controlling mechanism in the embrittlement process. This technique has not always produced positive results. Bhatt and Phelps 5 using a 12~o C r - M o - V steel in a 3 ~ NaC1 at pH 1 found the results inconclusive, and it required H diffusion studies to show that failures were most probable due to H embrittlement. The authors related the potentials of unpolarized and polarized Ii

374

J.W. KENNEDyand J. A. WHrrl'AKER

specimens to the equilibrium potential of the H reaction at the various pH levels which they studied. Although the electrochemical polarization technique has found wider support, Truman e t al. ~1 have questioned the validity of some of the conclusions from the tests. In 3 ~o NaCI specimens tempered a t 450°C showed an increase in life with constant current cathodic polarization, the life decreasing again for higher cathodic c.d.s, and a decrease in life on similar anodic polarization. This is exactly the behaviour pattern which all other workers have interpreted as indicating stress-corrosion as the mechanism of failure but Truman 2x appears to have rejected it. Material tempered at 250 ° showed a decrease in life on both cathodic and anodic polarization in the same neutral solution and Truman appears not to have differentiated the behaviour of these two tempers, rather considering that material tempered at 450°C and below had a common mode of failure. Cathodic polarization in 5 ~o H2SO4 reduced the life for both tempers. Unfortunately no potential measurements were made in this study. In a recent paper Yamaoka and Wranglen 22 investigated three Swedish steels, although the majority of the work reported was on a steel comparable with the silicon modified-4340. Testing of wires was carried out in deionized water at a stress equivalent to 93 ~ of the tensile strength. In air the endurance was 600 h but this dropped to 40 h in water. Various other tests included the addition of small quantities of sulphide ions to the water, cathodic polarization and the effect of decreased pH, and all indicated that failure was due to H embrittlement. Potential measurements were made in 0.1N NaSO4, which gave endurances comparable with the deionized in water, which indicated that the H evolution reaction was possible at the natural corrosion potential. Polarization studies appear to be suitable as a laboratory technique for distinguishing scc from H embrittlement. It would be most valuable, however, if some distinct metallographic differences were found to exist between the two types of failure, and attempts to determine fractographic differences have been made by various investigators. Hughes et al. Is found no real differences between the paths followed by scc and H embrittlement cracks. This was attributed to the fact that the active path followed by the stress-corrosion cracks was also the mechanically weakest path, and therefore the most susceptible to H embrittlement cracking. Davies n examined the fractured surfaces of 4330 M and 4340 steel specimens which had failed under anodic and cathodic polarization and those which had failed under H embrittlement conditions. The failures were found to be intergranular and Davies concluded that this indicated some fundamental similarity between scc and H embrittlement cracking. Davies e t a L s made a fairly detailed compilation of the modes of cracking of a range of steels, showing how the mode changed with tempering temperature, etc. Dean and Copson, 7 in a paper on maraging steels, concluded that failures in cathodically protected specimens in sea water were the most likely caused by H embritflement. When the specimens were sectioned and etched, differences could be observed. Those specimens in which the failure was attributed to stress-corrosion cracking showed fine hairline cracking, whereas there were rounded voids along cracks caused by H.

Stress-corrosion cracking of high strength steels

375

O t h e r i n v e s t i g a t o r s h a v e also u s e d e l e c t r o n m i c r o s c o p e t e c h n i q u e s to d e t e r m i n e t h e c h a r a c t e r i s t i c c r a c k i n g m o d e o f v a r i o u s a l l o y e n v i r o n m e n t systems, b u t it seems t h a t as f a r as d i f f e r e n t i a t i n g b e t w e e n scc a n d H e m b r i t t l e m e n t c r a c k i n g , while t h e v a r i o u s t e c h n i q u e s s h o w p r o m i s e , a b s o l u t e d i a g n o s i s is n o t yet possible. REFERENCES M. E. SHANK, C. E. SPAE'rH,V. W. COOKEand J. E. COYNr~,MetalProg. 76, 74, 84 (1959). E. FL PHELPS and A. W. LOGINOW, Corrosion 16, 325 (1960). R. A. DAVIES, G. A. DREYERand W. C. GALLAUGHER,Corrosion 20, 93t (1964). A. RUBtN, Report 2914 U.S.A.F. Research Contract DA-04-495-ORD-3069 (1964). H. J. BHAT'r and E. H. PHELPS, Corrosion 17, 430t (1961). R. PERRY, Joint B.I.S.R.A.-LS.I. Conf. Metallurgical Developments in High Alloy Steels. Special Report 86 (June, 1964). 7. S. W. DEAN and H. R. COPSON, Corrosion 21, 95 (1965). 8. J. E. TRtlMAN and R. PERRY,J. Iron Steel Inst. 202, 12 (1964). 9. P. C. HUGHES, L R. LAMBOURNand B. B. LXEBERT,J. Iron Steel Inst. 203, 154 (1965). 10. R. B. SE'rrERLtn,rD. Report 2684. U.S.A.F. Contract DA-04-495-ORD 3-69 (1963). 11. R. A. DAWES, Corrosion 19, 454 (1963). 12. B. F. BROWN and C. D. BEACHEM,Corros. Sci. 5, 745 (1965). 13. G. L. HANNA, A. R. TROIANAand E. A. STEROERWALD,Trans. Am. Soc. Metalls. 57, 658 (1964). 14. D. C. BRIGGS, J. U. MACEWAN and H. H. YA'r~, Corrosion 16, 141 (1960). 15. B. F. BROWN, Corrosion 15, 399t (1959). 16. FI. Suss, U.S.A.E.C. Report No. K A P L - M - H O S - 9 (1959). 17. U. R. EVANS, Private communication. 18. P. C. HUGH,, I. R. LAMBOtmNand B. B. LIEBERT,J. Iron Steel Inst. 203, 728 (1965). 19. B. F. BROWN, Gases in Steel. Report of N.R.L. Progress, p. 40 (1958). 20. I. WEmULL, Advances in Aeronautical Sciences p. 335. Pergamon Press, Oxford (1962). 21. J. E. TRUMAN, R. PERRY and G. N. CHAPMAN,J. Iron Steel Inst. 202, 745 (1964). 22. H. Y~a~AO~CAand G. WR~'qGLEN, Corros. Sci. 6, 113 (1966). 1. 2. 3. 4. 5. 6.