Electrochemical aspects of stress-corrosion cracking of sensitised stainless steels

Electrochemical aspects of stress-corrosion cracking of sensitised stainless steels

Corrosion Science, Vol. 23, No. 4, pp. 363-378, 1983 Printed in Great Britain. 0010-938X/83/040363-15 $03.00/0 Pergamon Press Ltd. ELECTROCHEMICAL A...

842KB Sizes 17 Downloads 101 Views

Corrosion Science, Vol. 23, No. 4, pp. 363-378, 1983 Printed in Great Britain.

0010-938X/83/040363-15 $03.00/0 Pergamon Press Ltd.

ELECTROCHEMICAL ASPECTS OF STRESS-CORROSION CRACKING OF SENSITISED STAINLESS STEELS R. C. NEWMAN and K. SIERADZKI Corrosion Science Group, Brookhaven National Laboratory, Upton, NY 11973, U.S.A. Abstract--The intergranular stress--corrosion cracking (SCC) of sensitised Type 304 stainless steel has been studied using sodium thiosulphate solutions at room temperature. Electrochemical aspects of the cracking have been examined using a potentiostatic scratching electrode technique applied to simulated grain boundary alloys. These results are compared with the current transients observed after intergranular fracture of embrittled specimens under the electrolyte. Dissolution kinetics within actual propagating cracks have been examined using load modulation. Rapid intergranular failure in a thiosulphate solution can be induced by successive load pulses of extremely short duration. Difficulties in accounting for the SCC velocities by an electrochemical mechanism are discussed. INTRODUCTION SENSITISED stainless steels often corrode intergranularly in the absence of stress, and most theories of intergranular stress-corrosion cracking in these materials have consequently emphasised the role of anodic dissolution. Recently some quantitative agreement between crack propagation rates and an electrochemical mechanism has been obtained for Type 304 steel in oxygenated high temperature water, t although in this case the nature of the environment inside a crack is unknown. In this paper we describe an electrochemical investigation of sensitisation and SCC in Type 304 steel, using sodium thiosulphate solutions at room temperature. Details of various mechanical aspects of the cracking are given elsewhere. ~-4 This system offers several features which favour the testing of the hypothesis that SCC is the result of anodic dissolution: rapidity and potential dependence of cracking, large currents flowing out of the cracks, possibilities of estimating the crack tip solution composition and ease of simulating the crack path using model chromium-depleted alloys. Here several transient electrochemical techniques are used to study both the SCC phenomenon and the distribution of chromium near the grain boundaries. The SCC of sensitised Type 304 steel in thiosulphate solutions, reported in 1981, ~ has obvious analogies with polythionic acid cracking where the main aggressive species is thought to be $4062-. 6 Thiosulphate has also been implicated in some cases of low temperature SCC in nuclear systems: and is active at concentrations as low as l0 -6 M. 2 Dhawale et al. a first pointed out the possibility of correlating the potential range of thiosulphate cracking with the expected potential range of active iron dissolution and elemental sulphur formation in a composite potential-pH diagram. Subsequently we have shown that the choice of a suitable crack tip pH gives a more convincing agreement. 4.9 This thermodynamic argument is illustrated in Fig. 1. Manuscript received l0 February 1982. Presented at the Meeting on "Electrochemical Techniques in Corrosion Testing and Research" held at the Corrosion & Protection Centre in the University of Manchester Institute of Science & Technology, 4-6 January 1982. 363

364

R.C. NEWMANand K. SIERADZKI I

I

FE203+ HSO~ ~2o3 + $40~-

0'4

FE203

t .

FE+++

"-.

HIS~f

+

SO~-

+S FE203

-0.4 --

FES

FE + It2S

.

FE304 +

-0'8

I

0

I

2

I

4

I

6

pH

I ¸-

8

I

10

I

12

FIG. 1. Potential-pH diagram for Fe-S-H20 at 298°K, not considering SO4=-. All equilibria involving dissolved species are drawn for unit activity; equilibria involving H=S are for 1 atm. gaseous H2S. tt° (FeS) ---- - 97.7 kJ mol-1; other values from Pourbaix "Atlas". Severe cracking ranges are indicated by a, b and c.

The rate of thiosulphate SCC can be extremely high. Figure 2 shows crack velocities measured at two potentials in a dilute solution; the steel was sensitised at 600°C for 24 h and had 0.07C. In this case the restricted potential range of cracking found in more concentrated solutions was not found owing to solution resistance effects. However, the potential range of cracking in a 0.5 M solution correlated closely with that of retarded repassivation of a simulated grain boundary alloy (9Cr, 10Ni) in a similar solution acidified to pH 3.0 (Fig. 3). Although this is by no means an exact simulation of the crack environment, disproportionation of thiosulphate helps to maintain a solution of about this pH near the crack tip following acidification by metal ion hydrolysis. This is easily seen by adding a chromium salt to a thiosulphate solution. Figure 3 also emphasises the main theoretical difficulty in this work" that the maximum crack velocities are 1-2 orders of magnitude higher than those expected from the electrochemical measurements. Currently it appears that to explain these results one must propose intermittent brittle fracture of the grain boundaries, assisted by local martensite formation and by the sharpening action of anodic dissolution.3. 4 In this paper, three related aspects of the SCC mechanism are considered. First, the repassivation behaviour of a series o f scratched iron-chromium-nickel alloys is examined. Secondly, after using a high temperature hydrogen charging technique,

Electrochemical aspects of stress-corrosion cracking I

I

I

I

I

365

I

T E

-5 X

>k-

X

O0

I-'4

t.~ 0 J bJ

-B

U CE

-7

X

X

?

>

@ /~ 0

0...-IBP__I

m"

X...

eSl~B

m'/

I

I

rd (3 0 _J

t

I

[email protected]

60

[email protected]

I

[email protected]

I

[email protected]@ [email protected]

K (ION m - 3 / 2 )

FIG. 2. Variation of crack velocity with apparent stress intensity for precracked specimens at two potentials in 6 × l0 -4 M Na2SzOa. intergranular fracture surfaces are created under an electrolyte and the resulting potentiostatic current transients analysed. Thirdly, load pulsing during potentiostatic polarisation of SCC specimens is used as a means of exposing fresh chromium depleted material to the solution present near the crack tip; the consequent current transients are used to assess various models of the cracking process. ~E - 5

I

I

0

I

o

I

I

e(~;

@

Z -6

0

H

Z 0

8Z Od HbJ Z Ld [L t~ 0 _S

-7

X

X

X X

X

(b) x

-8

--S

I

-488-288

I

I

I

I

@

[email protected]@

[email protected]

POTENTIRL

(m'/

....

SCE)

FIG. 3. Comparisonofpenetrationrates:(a)CERtestsin0.5M Na2S2Oaatl0-4s-1; (b) s ~ a ~ h i ~ ofFe-9Cr-10Nialloyin0.5M NazS203acidifiedtopH3.0.

366

R . C . NEWMAt~and K. SIERADZKI EXPERIMENTAL

METHOD

Background

The alloy used for all the SCC experiments in this and previous work 2-~ contained Cr 18.7 wt %, Ni 8.55, M n 1.70, Si 0.70, C 0.07, P 0.026, and S 0.005. The sensitisation treatment was 600°C for 24 h following solution annealing at 1100°C. According to Tedmon et al. 11 an alloy of this carbon content and sensitisation temperature could have a grain boundary chromium concentration as low as 7.5%. An alloy containing 9.2Cr, 10.1Ni has been used in our previous repassivation measurements, a,4 recognizing that stress-corrosion cracks are probably by no means atomically sharp and that the chromium-depleted zone typically extends to ca. 100 nm either side of the grain boundary. One purpose of this work is to address the possibility that the crack could be proceeding down a narrow zone of very low chromium content. Measurements of grain boundary chromium concentrations in these materials using scanning transmission electron microscopy are by no means routine; thus an electrochemical method has been used in conjunction with intergranular hydrogen embrittlement. Procedure Preparation o f a l l o y s f o r electrochemical measurements. The followingcompositionswere exam-

ined: Cr 4.01 6.09 9.22 10.8 11.7

Ni 10.2 9.13 10.1 8.89 8.74

Cr 12.6 15.0 18.7

Ni 10.1 9.27 8.55 (annealed steel)

The laboratory melts were cold-rolled, homngenised in vacuum at 1200°C for 16 h, and air-cooled. In this condition all the alloys were highly martensitic: this was not expected to be significant as rapid martensite formation is expected to occur on scratching or tensile deformation. Specimens for electrochemical measurements were cut as plates approx. 20 x 10 × 4 mm and set in a row with individual electrical connections, face upwards in a single rectangular block of epoxy resin. In this fashion all the specimens could be maintained at the same potential, or a single specimen could be selected. When using rather unstable solutions, accurate comparisons could be quickly made between different alloys. Scratched electrode measurements. The specimen block was ground to a 600 grit finish and attached to the rectangular base of a plastic container. Four 5 m m wide strips of platinum foil and a platinum wire were spot-welded together to f o r m a rectangular counter electrode, approx. 60 × 120 m m outer dimensions, with a wire connection. This was placed on the base of the cell around the specimen block. A saturated calomel electrode (SCE) was placed in the solution near one edge of the cell, with a double capillary barrier to its contents. The solutions were at room temperature (22 -4- 2°C) throughout and were exposed to air. Scratching was performed manually with a diamond pencil. The potential was controlled by a potentiostat (Stonehart model BC 1200). Within the scope of the investigation the cathodic component of the cell current due to oxygen reduction did not affect the current transient measurements. Typically the projected scratch area was 2.0 -4- 0.4 m m ~ (,~ 10 x 0.2 mm), with a contact time of 50 -4- 20 ms. The current transients were recorded by a Nicolet digital storage oscilloscope and stored directly on magnetic discs or transferred via GP-IB interface to a Hewlett Packard 9845 computer, where they could be integrated or otherwise processed. Embrittlement ofsensitised steel. Current transients were obtained from oxide-free grain boundary fracture surfaces using hydrogen embrittlement. The method of Elkholy et al. 12 was used to charge large quantities of hydrogen through the thickness of notched specimens so that they could be fractured intergranularly at room temperature. A mixture of acidic and neutral sulphates (40% KHSO~, 40% NaHSO4, 7% Na2SO4, 13% K~SO4) was melted in a cylindrical glass beaker using heating tapes, and water dripped in continuously through a hole in a Teflon lid. With the specimen as cathode and a 6 cm ~ platinum anode, a current density of ca. 200 mA cm -2 was passed at various temperatures between 200 and 325°C. For most of the measurements the sensitised steel was used in strip form (80 × 8 x 0.5 mm) after 50% cold reduction by rolling. In this condition it was highly magnetic with > 10% martensite. Double edge and face notches were cut at intervals of 10 mm using a triangular file, giving seven fractures from each 80 m m strip; the remaining cross section after notching measured about 3 x 0.2 ram, similar in width to the scratches made in the scratching

Electrochemical aspects of stress-corrosion cracking

367

electrode tests and therefore giving about the same ohmic potential drop in the electrolyte under identical conditions [13]. After charging with hydrogen at 275°C for 4 h, these specimens were easily fractured by manual bending under the electrolyte, using rubber gloves and multiple electrical connections. Several specimens were also used in the unworked condition (80 x 12 x 1 mm); these were less easily fractured intergranularly by notching and bending, and gave better results when notched in the centre of both faces and pulled in a constant extension rate device with a displacement rate of about 3 ~tm s-a. Under these conditions crack growth was rather gradual but showed useful discontinuities.

Load pulsing ofsensitised steel. A servo-hydraulic testing machine was used to deliver a variety of load or displacement waveforms to smooth tensile specimens (3.2 × 1.6 mm rectangular cross section and 25.4 mm gauge length). The test solution was contained in a small plastic beaker, slit and cemented around the specimen with a silicone rubber compound, containing counter and reference electrodes. In one set of experiments a large number of cracks was initiated by straining several per cent at a strain rate of ca. 10-~ s-1, then the system was switched to load control and load pulses applied at a constant potential (usually - 200 mV SCE). Repetitive load pulses from near zero of very short duration (~ 50 ms) were also applied to uncracked specimens at - 200 mV. The resulting current transients were recorded, as described earlier, and compared with those obtained by scratching various alloys in neutral and acidified solutions. E X P E R I M E N T A L R E S U L T S AND D I S C U S S I O N

ScratchhTg electrode measurements in neutral thiosulphate solutions Results for the F e - 9 C r - 1 0 N i alloy and the annealed Type 304 steel in 0.5 M NazS2Oa have been presented elsewhere3; in Fig. 4 the repassivation behaviour of a series o f alloys is shown for two potentials where SCC initiation was rapid. Retarded repassivation is observed for Cr concentrations < 11.7%; the 12.6% alloy was indistinguishable from the Type 304 steel in this test. The 4.01 and 6.09 % alloys showed no passivation and were therefore not investigated in detail as possible grain b o u n d a r y analogues, as it was k n o w n that no intergranular corrosion occurred in this solution in the absence o f stress. However, it was notable that very little difference in peak current density was ever observed between the scratched 4, 6 and 9 ~/o Cr alloys. Figure 5 shows the dependence on potential o f a simple repassivation parametcr for two alloys in 0.5 M Na2SzO3 with no acidification (pH ~ 8), These data serve as the baseline for the fracture surface measurements, which are also included. The primary comparison is made at -- 200 to -- 250 mV (SCE) where the repassivation transients o f the chromium depleted alloys are most characteristic and SCC is rapid. Fracture o f embrittled steel Figure 6 shows the fracture surfaces produced by bending the cold rolled, sensitised steel samples with and without hydrogen charging. The hydrogen-assisted fracture is clearly at least 8 0 ~ intergranular; higher magnification examination disclosed the usual sub-micron roughness associated with the grain b o u n d a r y carbides. When fractured at -- 200 mV (SCE) in 0.5 M Na~S~O3, the embrittled specimens developed an even interference tinge whereas the ductile fracture surfaces remained bright except for a few intergranular areas. This was indicative o f a reasonably uniform exposure o f the c h r o m i u m depleted material, as a dark film was observed on the scratches made on the alloys containing less than 12Cr. Clearly the retarded repassivation o f the intergranular fracture surfaces (Figs 4-5) is associated with their chromium depletion. There must be some degree o f coverage o f the fracture faces by chromium-rich carbides, with a m a x i m u m coverage o f 50 % if the original grain boundaries were continuously decorated. Transmission micro-

368

R . C . NEWMANand K. SIERADZKI

scopy studies 4 suggest that 20 ~ is a more reasonable figure. Scratching electrode tests have shown that both chromium and its carbides repassivate extremely rapidly below the potential range of transpassive dissolution, with the passage of less than 5 mC cm-2 i4; thus no more than 1 mC cm-Z of fracture surface area can be due to carbide reactions and this is confined within ~ 100 ms after scratching. The annealed steel repassivates only a little less rapidly, as reported previouslya; thus 20 ~o of ductile

~~._ 1

--,3

i

i

B

!

s

11

c d

--°9

W

O

- ! ,5

W -1,8 z •-a - 2 , 1 v N

~- - 2 . 4 z w o~ - 2 . 7 n, -n u -3 --,|

i

Cr Cr 12 C r fracture surTace

a

b

rr E - -1.2

i

I

|

I

I

I

I

I

I

I

.3

.7

1.1

1.5

1.9

2.3

2.7

3.1

3.5

3.9

TIME,

FIG. 4a.

1.5 a

I tc E

w o n, w N z

Z ba

a b

9 i1

\

d

\ \

e

13, C r fracture

.5

Cr

B

;

-1

[email protected]

-1,5

0.5M N a 2 S 2 0 3

-2

\ - 2 0 0

mV SCE

-~. 5 ~

u -1

-.5

e

.5

1

l

1,5

TIME.

2

i

l



2.5

3

3.5

s

FIG. 4b. FIG. 4. Comparison of transient electrochemical behavior of intergranular fracture surfaces with that of scratched F e - C r - N i alloys in 0.5 M Na~S20~: (a) - 250 mV (SCE); (b) - 200 mV (SCE). Fracture surface results are normalised to a single peak current.

Electrochemical aspects of stress-corrosion cracking

369

Od

I E

u 58

----

,

,

U E

,

---,

-'n----

o o 4 ~

(4

0 .... +.., X...

0

]~ Cr ~Z Cr qracture

OC

w

FL

surface

30

X)~

~:

X

>- 2 e

o~x x

I--4 U]

X++

z

caw

18

o

+

LI {]~ I

U

--0

0 I

+

X ÷ o+n+.o+_ o o 8

0

I

-408-288 POTENT

0

X +

0

i~

I

I

l

8

20~

488

IFIL

(rnV. SCE)

FIG. 5. Potential dependence of a repassivation parameter for two scratched Fe-Cr-Ni alloys, compared with results obtained on intergranular fracture surfaces of sensitised steel. fracture interspersed with the intergranular facets would contribute about another 1 mC cm 2 of fracture surface and its effect would be negligible after ~ 150 ms. Impurity segregation is expected to have only a small effect on transients which involve dissolution of many atom layers of alloy, particularly as the strongest adsorber present is sulphur generated from the solution; similarly, any hydrogen present on the fracture surface can only account for a small charge density provided that no hydride phases are formed. Current transients were obtained from embrittled grain boundaries in both the cold rolled bend samples and in one non-rolled constant extension rate specimen where a burst of fast intergranular fracture occured prior to final separation. These transients showed a time from peak current to 90 % repassivation of c a . 1 s, consistent with a mean chromium concentration (within the depth attacked) of a little above 10.8 %. The scatter bands in Fig. 4 were arrived at by taking 8 transients and normalising them all to a single peak current. The depth of metal attacked was estimated using the charge passed, the projected fracture surface area and a grain size roughness factor of 2, assuming that the fine scale roughness is compensated by the reduction in effective surface area due to the carbides and ductile fracture. This gave 10 nm as a typical value of the depth dissolved in 1 s at -- 200 to -- 250 inV. The results suggested that, if material of 10.8 % Cr or less is present, it is confined to less than about 2 nm on either side of a typical grain boundary. Examination of the transients in Fig. 4 indicates that the subtraction of even a 50 % area contribution from carbides and imperfect grain boundary fracture (both assumed to behave like the scratched 18Cr steel) would still lead one to deduce a grain boundary chromium concentration between 10.8 and 11.7 % for the first 10 nm.

370

R.C. NEWMANand K. Sm~A,DZKI

Use of smaller increments of Cr concentration in the scratching tests would enable the grain boundary composition to be established more precisely. Probably both stress-corrosion and hydrogen cracks preferentially sample those boundaries with least chromium, owing to enhancement of both martensite formation and dissolution kinetics.

Load pulsing The first experiment performed under this heading involved the application of short, successively larger load pulses from near zero to an uncracked sensitised steel specimen held at -- 200 mV (SCE) in 0.5 M Na=S=Os. This series, progressing up to a little above yield, is shown in Fig. 7(b) with its corresponding anodic current transients. These result from film fracture occurring rather randomly over the gauge length, and their shape reflects the repassivation of the matrix alloy as measured in the scratching tests. The repeated application of a considerably larger load pulse showed a dramatic difference: now the repassivation began to take many seconds as early as the second pulse (Fig. 7c). After only five pulses at about 10 s intervals, a large number of fine intergranular cracks was visible. The application of 30 pulses followed by overloading the specimen showed that many cracks were present varying in length from about 300 to 600 ~tm. This is therefore equivalent to a crack velocity of 1-2 ~tm s -1 calculated from the total exposure time, or 100-200 ~m s -1 calculated from the time spent under load! Several conclusions can be drawn from this behaviour: 1. In this form of active path SCC, the crack tip strain and corrosion can be allowed to occur alternately rather than simultaneously without much alteration in the mean crack velocity; however, in the absence of a periodic loading the penetration rate (indicated by the current flowing at zero load) eventually decays to zero. 2. The aggressive environment inside a crack is established extremely rapidly and is clearly different from both the bulk solution and the low pH solutions prepared using chromium sulphate or sulphuric acid additions? The long term repassivation transients for the load pulsing and scratching are compared in Fig. 8. 3. The current transients are very large considering the probable degree of crack tip strain involved, and do not become smaller with successive identical load pulses. Indeed, the longer the waiting period between pulses, the larger the next transient. Thus the cracks are growing appreciably as a result of each pulse, and the same size of load pulse can repeatedly cause extensive crack tip deformation or fracture. The above considerations highlight a possible contribution of this technique to the estimation of two unknown quantities: (a) the nature of the crack tip deformation producing these large current transients, and (b) the nature of the environment near the crack tip. In one specimen a relatively small number of cracks was produced, with a total crack tip length of about 20 mm, as estimated from optical microscope examination. After allowing the cracks to passivate at zero load, successive load pulses were applied from near zero and the current transients (similar in shape to those in Fig 7c) recorded. After allowing for an initial transient from some matrix material (decaying in < 1 s) the current due to exposure of chromium depleted material was deduced to be slowly decaying from about 1 mA. If this were to arise wholly from stretching of the crack tips exposing a 200 nm wide chromium-depleted zone, the subsequent anodic current density at these sites must have been decaying slowly from about 25 A cm -~.

FIG. 6. Fracture surfaces of sensitised steel after rolling, notching and bending (a) charged with hydrogen in salt bath; (b) uncharged. ( × 50).

Electrochemical aspects of stress-corrosion cracking

373

This is about the same unlikely crack tip current density as must be proposed to account for the maximum crack velocities of ~ 10 ptm s-l.~, 4 A more reasonable interpretation is that the crack advances in a brittle fashion on the application of each load pulse, as a result of previous sharpening by dissolution. If the chromiumdepleted zone extends 100 nm on either side of a grain boundary, then the bare surface

3000 2?eB

LORD

240B

PULSE

21BB 1BBB

ISSe

G

12ee

¢

9~B

0

G00

>

36e I

I

I

I

I

I

I

I

I

• 65

.?

.75

.8

.85

.9

.85

I

1.05

8 ,G

TIHE,

F m . 7a.

• fl55 2.3

.B'1 .2

.~]25

.~1

~-. ~m~ Ld O

-.82

2.0

N Z

~-.035 - . 95

1.9

hi rv

~ - . 8G~ -i U

-.08

4

B

12

1G 28 TIME,

FIG. 7b.

2't ¢

2B

32

36

4El

1.1

374

R . C . NEWMAN and K. SIERADZKI

-.55

--,G -.05 ~-.7

W-.75 0 N Z

~-.s5

304 SENSITIZED 0.5M Na2S203 -280 mV SCE

W

REPETITIVE LORD

PULSES

~ -095 U -[

I

-18 -t

8

2G 35

I?

44

53

G2

71

00

TIME, s

Flo. 7c. Fro. 7. (a) Load pulse used in crack growth experiment. 1 V = 890 N. (b) Potentiostatic current transients produced by load pulsing of smooth sensitised steel specimen. Peak loads are given in V, 1 V ---- 890 N. 0.5 M Na=S~Oa, - 200 mV (SCE). (c) Current transients produced by repetitive application of the load pulse shown in Fig. 7(a) to a smooth sensitised steel specimen, 0.5 M Na=S20~, - 200 mV (SCE).

B

-20fl mV SCE

--.4 --,8 E

-!

II

\

I~ \

.2

-

~ ~

"x,

~, \

~. - 1 . s W O

-2

w

-2.4

N Z 3 -2. B

~

Ib ~

I

9 Cr, pH 2.4 (e.sM Na2S~03~ LORDPULSE

"

'--r

v- - 3 . 2 z

w u

-3.G

-4 1.3

.

!

1.8

2.3

i

2.8

!

3.3

. , i 3.8

I 4.3

4.8

.

s

5.3

5.9

6.3

TIME, =

Fzo. 8. Comparison of scratch current transients for Fe-9Cr-10Ni alloy in two acidified thiosulphate solutions with load pulsing current transient for a cracked specimen at the same potential. Load pulse as in Fig. 7(a). Load pulsing transient is normalised to the same peak current amplitude as the scratch transients; actual peak amplitude was 1.3 mA.

Electrochemical aspects of stress-corrosion cracking

375

dissolution in the acidified thiosulphate solution could produce a crack tip radius of as little as 5-10 nm for a typical chromium concentration profile. 15 If this sharp penetration is embedded in a zone heavily transformed to martensite as a result of the high local strain, very high elastic stresses can be obtained on application of the next load pulse, leading to localised grain boundary separation. Hydrogen cannot be ruled out as a contributing factor. This is the model proposed elsewhere to account for the high crack velocities. 4 When the crack advances in this fashion, the area of fresh metal surface exposed to the solution is relatively large: for a i0 ~m jump distance (typically required to explain the crack velocities) and the same cracks described above, the area exposed is 4 × 10 -3 cm 2 and the current density is decaying from 250 m A c m -~. This correlates reasonably well with the maximum current densities in scratching tests in the pH 3 solution. 4 An experiment was carried out to explore the idea that load pulsing caused the cracks to advance discontinuously. A load pulse from zero was applied at - - 2 0 0 mV (SCE) and the current allowed to decay for 30 s. At zero load, the potential was switched to ÷ 400 mV (SCE) where even the 9Cr alloy repassivates very rapidly, 4 and the same load pulse repeated several times at intervals of ~ 5 s. In this case the first current transient at ÷ 400 mV ( ~ 1 mA peak current) was very much larger than those following, which soon became negligible in size ( < 20 ~A peak current.). One interpretation of this observation is that the cracks are initially sharp owing to the previous pulse at -- 200 mV and, independent of the potential, are then able to jump several ~m and give a high current transient. However, at + 400 mV there is no subsequent dissolution to sharpen the cracks and the second and subsequent load pulses cause only minor additional crack tip deformation. Investigation of these phenomena is continuing.

Further hTvestigation of the nature of the crack tip envh'onment Consideration was given to possible modifications of the crack tip environment other than the reduction of pH to 3 by Cr ~ hydrolysis, which accounts satisfactorily for the potential range of cracking 4 but not for the load pulsing transients. Four feasible alterations were investigated in order to identify conditions which could account for the slow repassivation observed after load pulsing: (1) Increase of thiosulphate concentration with the pH remaining near 3, (ii) maintenance of a crack tip pH substantially below 3 by continuous rapid injection of Cr a+, (iii) a high degree of homogeneous decomposition of thiosulphate to (SO~2 ÷ SO2 + S) at pH ~ 3, or at lower pH down to 1, and (iv) a high degree of electrochemical decomposition of thiosulphate, aided by the high surface-to-volume ratio in a crack, e.g. oxidation to sulphate. Scratching tests were carried out on the 9.2Cr alloy, giving the following results: 1. Increasing the thiosulphate concentration above 0.5 M caused more rapid repassivation at -- 200 mV (SCE) when the concentration exceeded ~ 1 M for pH 8 or ~ 2 M f o r p H 3 . 2. For -- 200 mV (SCE), 0.5 M Na~S20~ and pH values down to 1 [obtained by rapid addition of large amounts of either H2SO4 or Cr~(SO4)a], the peak current density (ip) and charge density passed in 1 s (qi) were increased by less than 20 ~ from their pH 3 values. However, the sudden repassivation phenomenon was progressively

376

R . C . NEWMANand K. SIERADZKI

shifted to longer times (Fig. 9). Measurements could not be made prior to sulphur precipitation, as the whole electrode activated for several seconds after the acid addition with a peak current density of ~ 50 mA cm-L Application of resistance compensation available on the potentiostat showed that both this active dissolution and the temporary scratch activation were not solution resistance limited. In the long term ( > 10 s), the electrode was again more passive than would be consistent with the load pulsing transients. 3. After longer exposure to dilute acid, the thiosulphate solutions were extensively converted to (S q- SOs -+- SOs~-) and were not only efficiently buffered (by SO3~- q2H ÷ ~- "H2SO8"), but consistently less aggressive than after short mixing times. No slow current decays were observed. The sulphite/sulphurous acid buffer dominated the Cr 3+ hydrolysis equilibrium when Cr2(SO4)3 was used. 4. Sulphate additions caused a remarkable activation of the alloy even at neutral pH and are quite capable of accounting for the prolonged dissolution observed within the cracks. Figure 10 shows the current transients obtained by scratching several alloys in a neutral (0.05 M Na2S203 -k 0.5 M Na~SO4) solution at -- 200 mV (SCE). In this case, acidification to pH 3 had a rather small effect. Neither solution component on its own was effective. The 9.2Cr alloy showed extensive dissolution, again with similar values of ip and ql as in the plain thiosulphate acidified to pH 3; thus the crack velocity correlations presented elsewhere 4 still hold, as does the conclusion that the crack velocities cannot be accounted for by the dissolution rates. The potential dependence of the transient again predicted a cracking range of approx. -- 300 to q- 100 mV (SCE). The 12.6 and 15Cr alloys showed a localised type of corrosion which is still under investigation. Once again, although the 6Cr alloy was uniformly active, it could not be induced to react faster than ~ 300 mA cm-L As Fig. 11 shows,

--2

-3.4

-288 mV SCE

-4.1

Fe-9.2 C r - l e . t N! 8.5M Na2S203

~--4.8

~

-5.5 -6.2 z

-6.9 -?.6

-8.3 --9

12

a

b

pH 2.8 pH 2 . 5 2.2

I

I

I

I

I

I

I

I

I

13

14

15

16

I?

18

19

2B

21

22

TIME, s

FIG. 9.

Effect of extreme acidification on scratch current transients for Fe-9Cr-10Ni alloy in 0.5 M Na2S~Os at - 200 mV (SCE).

Electrochemical aspects of stress-corrosion cracking

-5

i"

'

z

I

....

I

z

i

d

377

i

,'

-5.3 d

-5. G {K

E

-5.9 -G. 2

w

-G.5

bJ N Z Z3

-G.B

I-Z hi n,'

-7.4

o

U

-7.

c 0.05M Na2S203 + O.SM Na2S04 -200 mV SCE IB Cr 15 Cr 13 Cr S Cr

I

c d

-7.7 -8 0

i

!

i

f

I

B

IG

24

32

40

- - . - I

I

48

,

58

L

|

84

72

80

TIME,

FIG. 10.

Scratch current transients for a series of Fe-xCr-10Ni alloys in (0.05 M Na~S~O3 ~ 0.5 M NazSOa), pH ~ 6, - 200 mV (SCE).

a much retarded repassivation of the fracture surface is observed in this solution, giving a reasonable approximation to the behaviour shown in Fig. 8(c).

Mechanism of dissolution enhancement by sulphate additions The shapes of the transients obtained without sulphate additions at potentials around -- 200 mV (SCE) (e.g. Fig. 4, 9.2Cr) are characteristic of passivation initiated by precipitation. 16 No noticeable effect of stirring was observed; thus the precipitation -5 -5.3 -5.G -5.9

FRRCTURE SURFRCE O.B5M Na2S203 + O.5M Na2SO4 oH 3 -200 mV SCE

E -6.2 w -E. 5 o hJ -B. B N Z -7. I iZ

-7.4 W [g n," -7.7 U

-B IG

I

I

I

I

I

I

I

I

I

16,5

17

17.5

18

18.5

19

19.5

20

20.5

21

TIME,

FIo. 11.

Current transient for intergranular fracture surface of sensitised steel in (0.05 M Na2S203 + 0.5 M Na~SO4), p H 3, - 200 mV (SCE).

378

R.C. NEWMANand K. SIERADZKI

must occur within the dark, porous surface deposit formed during the period of activity. The sulphate additions m a y provide an inert ion which can hold iron, nickel and c h r o m i u m in solution within the pores o f the deposit and thereby retard or prevent precipitation. The ubiquitous dark deposit accounts for the limitation o f the maximum sustained current density (100-300 m A cm -~) observed in all the solutions examined for all alloys with > 4Cr. CONCLUSIONS I. High temperature hydrogen charging can be used in conjunction with transient electrochemistry to estimate grain b o u n d a r y c h r o m i u m concentrations in a sensitised steel. For a 0.07C material sensitised at 600°C for 24 h, the mean concentration within l0 nm o f the grain b o u n d a r y is between 10.8 and I 1.7 ~ . 2. Measurement o f current transients following potentiostatic load pulsing o f SCC specimens can give information on the crack tip solution composition. In this system, rapid cracking can be induced by periodic load pulses o f very short total duration. 3. Thiosulphate SCC probably involves intermittent brittle fracture; the role o f dissolution is to sharpen the crack. The behaviour o f the alloy in the crack tip solution can be simulated by using an excess o f sulphate. Acknowledgements---This work was supported by the U.S. Department of Energy, Division of Basic

Energy Sciences, under Contract No. DE-AC02-76CH00016. The assistance of Kenneth Sutter and Ronald Graeser in the experimental work is gratefully acknowledged. Yako Sanborn and Robert Sabatini performed the computer programming and scanning electron microscopy, respectively. REFERENCES 1. F. P. FORDand M. SILVERMAN,Corrosion 36, 558 (1980). 2. H. S. ISAACS,B. VYASand M. W. KENDIG, Corrosion 38, 130 (1982). 3. R. C. NEWMAN,K. SIERADZKIand H. S. ISAACS,Proc. Conf. on Environmental Degradation of Engineering Materials, p. 163. Virginia Polytechnic Institute (1981). 4. R. C. NEWMAN,K. SIERADZKIand H. S. ISAACS,Metals Trans. 13A, 2015 (1982). 5. H. S. ISAACS,B. VYASand M. W. KENDm, Paper No. 26, Corrosion 81, Toronto (1981). 6. A. DRAVNIEKSand C. H. SAMANS,Proc. Am. Petr. Inst., 37 IH, 100 (1957). 7. U. S. NUCLEARREGULATORYCOMMISSION,Report NUREG-0691, Sept. 1980, Chapt. 2 (1980). 8. S. DHAWALE,G. CRAGNOLINOand D. D. MACDONALD,EPRI Project RP 1166-1, Progress Report July-December (1980). 9. K. SIERADZKI,H. S. ISAACSand R. C. NEWMAN,Paper No. 224, Corrosion 82, Houston (1982). 10. I. MATSUSHIMA,Boshoku Gijutsu 22, 141 (1973). I I. C. S. TEDMON,D. A. VERM1LYEAand J. H. ROSOLOWSKI,J. electrochem. Soc. 118, 192 (1971). 12. A. ELKHOLY,J. GALLAND,P. AZOU and P. BASTIEN,C. r. Acad. Sci., Paris, Ser. C 284, 363 (1977). 13. H. J. PEARSON,G. T. BURSTEINand R. C. NEWMAN,J. electrochem. Soc. 128, 2297 (1981). 14. J. WOODWARD,Ph.D. Thesis, University of Cambridge (1981). 15. M. G. LACKEYand F. J. HUMPHREYS,Proc. Conf. on Hydrogen Effects in Metals, Jackson Hole, Wyoming, 1980 (eds. I. M. BERNSTEINand A. W. THOMPSON),p. 666. AIME (1981). 16. W. J. MULLER,Trans. Faraday Soc. 27, 737 (1931).