Effects of prior cold work and sensitization heat treatment on chloride stress corrosion cracking in type 304 stainless steels

Effects of prior cold work and sensitization heat treatment on chloride stress corrosion cracking in type 304 stainless steels

Corrosion Science 43 (2001) 1519±1539 www.elsevier.com/locate/corsci E€ects of prior cold work and sensitization heat treatment on chloride stress c...

583KB Sizes 1 Downloads 40 Views

Corrosion Science 43 (2001) 1519±1539

www.elsevier.com/locate/corsci

E€ects of prior cold work and sensitization heat treatment on chloride stress corrosion cracking in type 304 stainless steels C. Garcõa *, F. Martõn, P. De Tiedra, J.A. Heredero, M.L. Aparicio  Area de Ciencia de los Materiales e Ingenierõa Metal urgica, Dpto: I.M.E.I.M. E.U.P. and E.T.S.I.I., Universidad de Valladolid, C/Paseo del Cauce s/n, Valladolid 47011, Spain Received 7 January 2000; accepted 9 October 2000

Abstract The e€ects of prior cold work (CW) and sensitization treatment on the stress corrosion cracking (SCC) behaviour of Type 304 stainless steels have been studied in chloride solutions using electrochemical tests and magnesium chloride tests with U-bend specimen. The results indicated that the SCC behaviour of the CW steel was essentially di€erent from that of the solution-annealed steel. Intergranular stress corrosion cracking (IGSCC) of solution-annealed material changed into a mixed mode or dominant transgranular stress corrosion cracking (TGSCC) when the degree of CW was increased. The sensitization treatment enhanced IGSCC susceptibility by shortening failure time and accelerating crack initiation and propagation rates, but this e€ect was di€erent for several other degrees of CW. These e€ects were revised taking into consideration the electrochemical and microstructural phenomena. The most dangerous degrees of deformation for di€erent sensitization conditions for the development of IGSCC and TGSCC processes have been determined. Ó 2001 Elsevier Science Ltd. All rights reserved. Keywords: 304 stainless steels; Cold work; Sensitization; Stress corrosion cracking

*

Corresponding author. Tel.: +34-983-423-389/+34-983-422-289; fax: +34-983-423-389. E-mail address: [email protected] (C. Garcõa).

0010-938X/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 0 - 9 3 8 X ( 0 0 ) 0 0 1 6 5 - 7

1520

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1. Introduction Stress corrosion cracking (SCC) and intergranular corrosion of austenitic stainless steels (SS) are the most important corrosion processes that a€ect the service behaviour of these materials. Multiple material and environmental factors a€ect SCC behaviour of austenitic SS. The environmental factors include the processes of forming by cold plastic deformation and seam welding. The latter justi®es that there is a growing interest in knowing the e€ect that prior cold work (CW) and sensitization treatment have on the SCC behaviour of these types of materials. The in¯uence of these factors on the susceptibility to SCC of Type 304 SS is beginning to be studied as a consequence of cracking in the boiling water reactor (BWR) piping systems produced with this type of steel. The BWR pipes are formed by either extrusion or by plate rolling and joined through seam welding and cracking has been precisely tracked down largely to the heat a€ected zone (HAZ). The intergranular cracking morphology observed, following the sensitized grain boundaries in HAZ, is due to a low temperature sensitization phenomenon [1±4]. A wide number of reports have shown that the sensitization treatment signi®cantly modi®es the SCC behaviour and the cause of this is the intergranular precipitation and the grain boundary chromium depletion [5±8]. However, the results on the e€ect of sensitization treatment on SCC susceptibility of austenitic SS have been contradictory. In general sensitization seems to be detrimental to IGCC under chloride [5,6,9±11], polythionic acid [12] and high temperature water [13±15] and also in some caustic conditions [7,16]; but at the same time it might be bene®cial under other caustic conditions [17,18]. Also, a clear interrelation between the susceptibility to IGC and to intergranular stress corrosion cracking (IGSCC) has been proven to exist [19,20]. Contradictory results have been reported on the e€ect of CW on SCC susceptibility. In this way the susceptibility to SCC of Type 304 SS has been proven to increase with an increase on the degree of cold working, especially above 40% [21±23]. Meanwhile other studies [24,25] have shown that low levels of CW seem to increase the resistance of chloride SCC up to a maximum value evaluated at 30%CW. Such a change in SCC susceptibility has been attributed to the formation of strain induced martensite. However in other studies [26±31], where a great variety of SS with di€erent tendencies to experience martensite transformation was used, all agreed in that low levels of CW increased the susceptibility to chloride SCC, while higher levels of CW had a bene®cial e€ect, regardless of the percentage of strain induced martensite in SS. With respect to the combined e€ect prior to CW and sensitization treatment, few investigations exist for high temperatures and short sensitization treatment times that simulate the conditions of CW and subsequent welding processes. The majority of these studies [4,32,33] are centred on low temperature sensitization. Nonetheless, it seems to be clear that a low degree of CW increases the IGSCC susceptibility in sensitized 304 SS the same as with the IGC susceptibility. As mentioned previously, it is necessary that these studies be completed in order to determine the susceptibility

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1521

to IGSCC and also to transgranular stress corrosion cracking (TGSCC) of Type 304 SS previously cold worked and subsequently sensitized. This will be one of the objectives of this paper. In a previous study [34] the authors examined the e€ect of prior CW on IGC and TGC using image analysis as a discrimination technique between both attacks. Deformations less than or equal to 10% have been found to cause higher levels of IGC which presupposes an IGSCC risk that should be tested in the study. On the other hand, deformations equal to or superior to 30% have been found to promote higher TGC, which in turn suggests that the TGSCC risk on these samples should be studied. Therefore, the objectives of this work are to identify, control and understand the e€ects of prior CW and to establish its connections with thermal treatments (in isolated and joint form) on SCC behaviour. The environment that has been selected is one of the most aggressive ones, chloride solutions; and the material that has been selected is a commercial grade Type 304 SS due to its large number of industrial applications. As previously suggested, our aim is to also quantify these e€ects and to establish relationships with microstructural factors, such as carbide precipitation, chromium depletion, grain size or secondary phases, which are modi®ed during steel processing. Lastly, the ®nal goal of this study is to ®nd practical solutions to prevent and reduce IGSCC and TGSCC failures. The temperature±time±strain e€ects on susceptibility to IGSCC and TGSCC are carried out by using well established standardized methods, such as the potentiodynamic anodic polarization test and standard practice for evaluating SCC resistance in a boiling magnesium chloride solution. This last test was completed with quantitative measurements of crack initiation and propagation rates using quantitative metallography by image analysis techniques. 2. Experimental 2.1. Material The material used in this work was a commercial grade Type 304 SS, 4.4 mm thick rectangular plate, selected due to the fact that it covers a great number of industrial applications and shows most of the drawbacks related to IGC and IGSCC, mainly in welded joints. The chemical composition obtained by emission spectroscopy was: 0.07% C, 17.5% Cr, 9.15% Ni, 0.51% Si, 1.75% Mn, 0.1% Cu, 0.025% P, 0.007% S and balance Fe. 2.2. Thermo-mechanical treatments The plates were received in mill-processed (MP) condition, equivalent to full annealing with some CW. In addition to this, some plates were cut into strips 300 mm long by 20 mm wide and were heated for 15 min at 1050°C under a stream of

1522

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

argon before being water quenched by a solution-annealing (SA) treatment. This treatment was applied to eliminate MP condition before deformation processing and will be the state used as a referential condition. Some solution-annealed strips were machined according to ASTM E-8=94 standard [35] and uniaxially deformed at 10%, 20%, 30%, 40% and 50% engineering strains (notation: 10%CW, 20%CW, 30%CW, 40%CW, 50%CW; or in generic terms: %CW). The set values for the tensile testing procedure were obtained by a simulation study using the A B A Q U S commercial programme, based on the ®nite elements method. The CW levels indicated above were obtained due to this. This simulation also ensures a uniform strain state through the whole probe calibration length. The MP, SA, %CW strips were heat treated at 675°C under a stream of argon for 15, 30 and 60 min. (notation: MPS15, MPS30, MPS60, SAS15, SAS39, SAS60, %CWS15, %CWS30, and %CWS60, where the percentage symbol indicates the ®ve possible levels of strain). The short sensitization times chosen were similar to the HAZ welded joints conditions. On the other hand, in all the published studies that we have knowledge of sensitization times superior to 1 h are used. These times could be much too long if the fact that CW accelerates the healing process is considered [36]. This is the phenomenon that we have tried to avoid in this study. 2.3. Experimental techniques 2.3.1. Potentiodynamic anodic polarization measurements Potentiodynamic anodic polarization curves, according to ASTM G-5 [37], have been used to screen the corrosion behaviour of the strips on the di€erent treatment conditions. These curves supply information on the susceptibility of the material to di€erent types of corrosion (pitting corrosion, IGC, and SCC). With respect to SCC susceptibility, the curves permit us to identify the range of critical potentials, where cracking usually develops ± active±passive transient ± which quantitatively indicate the SCC crack growing risk. The surface preparation of the sample was done with no. 800 emery paper. Nitrogen streaming and agitation were used throughout the whole test. The tests were carried out in acid solution with chlorides containing 1 M H2 SO4 ‡ 5 N, NaC1 and temperature at 30°C  1. The experimental parameters of the tests were the following: 30 s delay at open circuit (OC) potential, 2 min anodic attack at ±200 mVSCE (SCE: saturated calomel electrode), then a delay of 5 min at VOC , 1 min cathodic cleaning at ±600 mVSCE , and then anodic potentiodynamic scan that started at 50 mVSCE below VOC until 200 mVSCE: The potential scan rate used was 30 mV/h. The specimens used in the polarization scans were examined using optical microscopy upon completing the experiments without any further etching treatment. 2.3.2. Anodic polarization double scan measurements This is another potentiodynamic method of rehearsal used, as the previous one, to reasonably provide accurate predictions of the potential ranges and the kinetic factors controlling SCC. This test is not a standardized method but it was applied

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1523

successfully for C±Mn steels in CO3 ±HCO3 [38], low-alloy steels on ammonium acetate solutions [39], and on caustic solutions [40]. First of all, the test involves scanning a range of potentials in the anodic direction at a high scan rate (in this work the fast scan rate was 1000 mV/h) to identify regions where an intense anodic activity is produced; and secondly, at a relatively low scan rate (10 mV/h) to identify regions where relative inactivity is produced. The comparison of the two curves indicate the ranges of potential with which high anodic activity in a ®lm-free condition is reduced to insigni®cant activity when the time requirements for ®lm formation are met. Therefore, this will indicate the range of potential with which SCC will be more probable. The test was carried out in the same solution and with the same experimental conditions than the anodic polarization test, but the specimens used were polished to a 1 lm diamond grit ®nish. The experimental parameters of the test were the following: 10 min delay at VOC , polarization scan started at 50 mVSCE below VOC until 300 mVSCE at 1000 mV/h, then a delay of 10 min at VOC , and polarization scan at 10 mV/h in the same conditions as the previous. 2.3.3. Boiling magnesium chloride test This test was conducted according to ASTM G-36 [41] standard. It is one of the few standardized tests for chloride SCC that also provides high severe conditions. For the test designs, standard ASTM G-30 [42] recommendations are followed. The specimens used were U-bend samples, which were prepared according to ASTM G-30. Plate material on ``MP'', ``SA'' and ``%CW'' conditions were cut and machined to get the geometry and dimensions of Fig. 1. Two holes were drilled as shown in this ®gure. Some of the specimens were submitted to sensitization treatment and were grounded to a 600-grit ®nish. The ¯at specimens were bent on a drill press with a 10 mm diameter mandrel with an appropriate jig. The specimens were plastic deformed during bending and stressed in a complex manner. When the Vshape was produced with the bending jig it was then converted to a U-bend by tightening the bolt. To prevent galvanic corrosion, the bolt, nut and washer were insulated from the specimens with PTFE sleeves and washers. The total strain on the outside of the bend was 25%. The specimens were tested in boiling 42% MgC12 to 155  1°C, and a modi®ed Allihn condenser was used to prevent vapour loss.

Fig. 1. U-bend specimen dimensions.

1524

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

The following procedures were used to evaluate SCC resistance: (a) Failure time according to ASTM G-36: The standard did not indicate a duration time limit of testing. After various preliminary trials a time limit of 48 h was agreed upon which supposed a high resistance to SCC for the studied material. Prolonging this test above the maximum time stipulated could bring about irregularities in the test environment. (b) Crack initiation rate: An initial crack initiation rate was determined as the number of cracks per minute present in the stressed surface. This was done after the ®rst 30 min of starting the test. The mean crack initiation rate was also evaluated as well as the relation between the number of cracks and failure time of the sample. (c) Crack propagation rate: The crack propagation rate was seen as being inconsistent during the whole test. In the ®rst stages the cracks grew at greater speeds and moreover not all of them in the same way. This is why a series of initiated cracks in the ®rst exposure at 30 min were selected. From that the initial crack propagation rate was determined as the average length/time of exposure of 120 min and the mean crack propagation speed as the average length/time of failure. (d) A metallographic examination of exposed surfaces and of polished and etched cross-sections at the end of the test was used to establish the type of cracking: intergranular, transgranular or mixed. The frequency of visual inspection was at 30 min during the ®rst 180 min, after that periodical inspections were carried out every 60 min. To control the number and length of the cracks a quantitative metallography technique by image analysis was used. Being more speci®c, this procedure for every sample at di€erent times was the following: The more stressed area of the sample was selected. This area was observed with optical microscopy. The corresponding images were digitalized and saved. Every image was analysed by means of speci®c software. To determine the crack initiation rate magnitude the number of new cracks, in every inspection, were counted. Finally, in determining the crack propagation rate magnitude a set of 10 of the longest cracks for every sample were taken. Every one of these cracks was measured in length and then an averaged value was calculated. 3. Experimental results 3.1. Anodic polarization measurements The generic aspect of the potentiodynamic polarization curves coincides for the di€erent thermo-mechanical conditions with the typical anodic polarization Type 304 SS plot. Appropriate electrochemical conditions that promote SCC correspond to the transition regions, i.e. active±passive transient and passive±transpassive transient. This is where the thermo-dynamic requirements of simultaneous activity

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1525

and passivity essential for cracking are carried out. Equally the polarization curves inform us on the susceptibility of pitting corrosion. Considering this, the e€ect of sensitization treatment and/or CW in the polarization behaviour of Type 304 SS has been analysed. Table 1 and Fig. 2 show the e€ect of sensitization time on the MP and SA samples. With both conditions when sensitization time was increased the corrosion potential decreased, the passive potential was shifted to less noble values and the passive potential ranges decreased. The anodic current density was not a€ected by the sensitization time but the passive current density was higher as the sensitization time increased. The critical pitting potential experimented an abrupt decrease with sensitization and in turn suggests their negative e€ect on pitting corrosion. The effects mentioned are more evident in the MP condition, and moreover have less passive ranges than the SA condition. Fig. 3 shows the polarization curves of the SA and %CW samples. The most noticeable aspect was the change in form of the curves for CW equal to or superior to 20%, where a secondary active±passive transition at more noble potentials appeared. Fig. 4 shows the e€ect of sensitization treatment for 10%CW and 30%CW samples. The ®rst observation with respect to this was that CW had a much stronger e€ect on the anodic behaviour than the sensitization treatment did. In samples with low deformations, 10%CW, the e€ect of the treatment was similar to that mentioned for MP and SA samples, but somewhat more evident. Contrary to this, the sensitization treatment of higher CWs the sensitization treatment did not a€ect the passive region instead it was centred on the increase of the critical active±passive

Table 1 Electrochemical parameters obtained from anodic polarization curves for MP and SA conditions for di€erent sensitization times Sample

MP MP sensitization (15 min) MP sensitization (30 min) MP sensitization (60 min) SA SA (15 SA (30 SA (60

sensitization min) sensitization min) sensitization min)

Corrosion potential (mV)

Passivation Pitting potenpotential (mV) tial (mV)

Anodic current density (lA/cm2 )

Passive current density (lA/cm2 )

394 409

130 140

‡445 ‡189

5675.40 5597.60

272.50 647.14

413

211

‡187

4742.42

1164.10

414

213

‡146

5081.60

1520.05

374 387

200 211

‡475 ‡387

2280.34 1832.31

619.44 729.15

394

239

‡329

2055.89

1874.99

396

248

‡296

2070.14

1970.60

1526

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

Fig. 2. E€ect of sensitization treatment on potentiodynamic anodic polarization curves for MP and SA conditions. The sensitization temperature was 675°C and the sensitization time selected was 60 min.

Fig. 3. E€ect of prior CW on potentiodynamic anodic polarization curves for Type 304 SS. Polarization scans for SA and 10%, 20%, 40%, 50%CW samples.

potential range. There was also an increase in the anodic current density as the degree of sensitization increased at the primary and secondary active±passive transient. Despite the presence of a great number of pitting, the microstructural examination through testing has permitted us to detect di€erent attack mechanisms according to the degree of CW. For the non-deformed samples or samples with low levels of CW an IG attack was mainly observed, as in Fig. 5(a), which corresponds to a 10%CWS60 sample. Meanwhile, for higher levels of CW the attack was mainly transgranular. Fig. 5(b) corresponds to a 50%CWS60 sample.

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1527

Fig. 4. E€ect of sensitization treatment on anodic polarization curves for 10%CW and 30%CW conditions. The sensitization temperature was 675°C and the sensitization time selected was 60 min.

Fig. 5. Surface appearance of Type 304 SS after potentiodynamic anodic polarization test (a) 10%CW and sensitized 60 min at 675°C (10%CWS60) and (b) 50%CW and sensitized 60 min at 675°C (50%CWS60).

3.2. Anodic polarization double scan measurements Fig. 6 shows the potentiodynamic anodic polarization curves at two di€erent scan rates for MP and 20%CW conditions. Regardless of the condition we can observe

1528

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

Fig. 6. Potentiodynamic polarization scans for MP and 20%CW conditions showing the domains of behaviour predicted from the curves. The slow scan rate was 10 mV/h and the fast one was 1000 mV/h. The comparison between them indicate the range of critical potential for SCC.

that the slow scan passivity is reached at a much lower potential than in a rapid scan, at the same time the passive current density is inferior for the slow scan. This indicates that in the same potential range, the rapid scan has minimized the formation of passive ®lm while the slow scan rate allows ®lming to occur, therefore, the test allows us to know the potential range where SCC an occur. In Fig. 6 one can also observe that the form of the curves are modi®ed by the e€ect of CW. In samples deformed at 20% or more, a secondary active±passive transient appears the same that occurred in the previous test. For MP and SA samples no changes were detected on the critical potential range with sensitization time. Nevertheless, the range of passivity in the slow scan was reduced considerably with sensitization time. Microstructurally, after the test, a prior attack on the grain boundary and on the austenite±delta ferrite interface was observed. For %CW samples, Fig. 7 shows that once again the e€ect of sensitization depends on the degree of CW. For samples with 10%CW the critical potential ranges not only did not increase, but rather decreased with sensitization time (Fig. 7(a)). On the contrary, in these samples a clear loss in stability was observed on the passive layer during slow scan. For samples with greater deformation (Fig. 7(b)), the critical potential range increased as the sensitization time and prior CW increased. In this case, the microstructural examination showed that the attack was transgranular on speci®c sites generated by CW. Linear appearance showed that there was a prior attack along slip planes, but strain induced martensite could have also contributed to such an attack. The di€erence in the curve forms, depending on the thermo-mechanical sample conditions, prevented us from obtaining quantitative results of SCC susceptibility as those carried out on other steels [33±35], as well as kinetic considerations.

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1529

Fig. 7. In¯uence of sensitization treatment in potentiodynamic polarization scans (a) for 10%CW and (b) for 40%CW. The sensitization temperature was 675°C and the selected sensitization time was 60 min.

3.3. Boiling magnesium chloride test 3.3.1. Failure time Table 2 shows the failure time and fracture mode results for MP, SA and CW samples for di€erent sensitization times. The ®rst important fact is that the e€ect of sensitization is characteristic of the previous condition in the steel, this suggests that the most important metallurgical variable is the degree of CW. As expected, for MP and SA samples the sensitization signi®cantly reduced the failure time. Even after using lower sensitization times to those commonly referred to, very high e€ects were obtained. For MA sample a contradictory e€ect for short sensitization times is observed. This seems to bene®t susceptibility to cracking and will be discussed later on. Fig. 8 shows how the prior CW, on its own, promotes a strong decrease in times of failure with respect to the SA condition. The only exception was the 20%CW with a similar time failure to the latter mentioned. It is convenient to point out that the

1530

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

Table 2 Failure times (h) and cracking mode Sensitization time (min)

MP

SA

10%CW

20%CW

30%CW

40%CW

50%CW

0 15 30 60

18.51 (TG)a 19.53 (M)b 20.46 (IG)c 8.56 (IG)

>48 (IG) >48 (IG) 25.43 (IG) 17.3 (IG)

14.33 (IG) 13.50 (IG) 10.16 (IG) 7.26 (IG)

>48 (M) >48 (M) 30.08 (M) 29.41 (M)

18.60 19.00 19.95 16.08

18.66 20.50 12.25 12.25

21.72 (TG) 18.61 (TG) 9.76 (TG) 7.35 (TG)

(M) (M) (TG) (TG)

(TG) (TG) (TG) (TG)

a

TG ± transgranular cracking. M ± mixed cracking. c IG ± intergranular cracking. b

Fig. 8. E€ects of CW (10%, 20%, 30%, 40%, 50%) and sensitization treatment (temperature ˆ 675°C and time ˆ 15, 30 and 60 min) on the failure time in a boiling 42% MgCl2 solution of solution-annealed and CW 304 SS.

MP condition has an average behaviour between SA and 10%CW, which would indicate that steel has a certain degree of work hardness on as-received condition. With respect to the sensitization treatment in¯uence for %CW samples, its e€ect was once again di€erent according to the prior degree of CW. For low deformations, 10%CW, the sensitization dangerously decreased time of failure, being in whichever sensitization condition lower than those registered in MA and SA conditions. Contrary to this the 20%CW had times of failure similar to SA samples for whichever sensitization condition. For degrees of CW equal to or superior to 30%CW, very low times of failure were obtained and the e€ect of sensitization was less important than in the rest. In these cases the in¯uence of the degree of CW was more important than the level of sensitization. The microstructural examination carried out after the test to establish the type of cracking indicated that the mechanism was di€erent depending on the state of CW

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1531

Fig. 9. Surface appearance after the boiling 42% MgCl2 test: (a) solution-annealed and 60 min sensitized (SAS60), (b) MP, (c) MP and 60 min sensitized (MPS60), (d) 10%CW and 60 min sensitized (10%CWS60), (e) 40%CW and 60 min sensitized (40%CWS60).

and sensitization time. Therefore, for SA and sensitization SA samples the initiation and propagation of the cracks were mainly intergranular. Fig. 9(a) corresponds to the SAS60 sample. For MP samples the sensitization not only reduced the time of failure but also originated a change in the cracking mechanism, so Fig. 9(b) shows the transgranular cracking for MP sample while in Fig. 9(c) we can observe intergranular cracking for MPS60 sample. The CW also originated new changes in cracking mechanisms, therefore samples with low degrees of CW and subsequently sensitized su€ered intergranular cracking, Fig. 9(d) corresponding to 10%CWS60 sample, while those highly deformed essentially presented transgranular cracking, for whichever sensitization time, Fig. 9(e) corresponding to 40%S60 sample. 3.3.2. Crack initiation rate Fig. 10 shows the results obtained for initial crack initiation rate and mean crack nucleation rate in some samples depending on the sensitization time. It was possible

1532

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

Fig. 10. E€ect of degree of prior CW on crack initiation rate of sensitized (temperature ˆ 675°C and time ˆ 15, 30 and 60 min) and non-sensitized type 304 SS in a boiling 42% MgCl2 solution.

to observe that the evolution of the process of crack nucleation is strongly dependent on the degree of prior CW. Therefore, the e€ect of CW was not only decisive in the failure of the samples, but also in the initial cracking. This suggests that the crack initiation mechanism must have been related to the microstructural e€ects that CW had on the Type 304 SS that were studied in a previous paper [43]. The main facts considered were: aging and transformation of delta ferrite, in secondary austenite and chromium carbides, and the austenite transformation to strain induced martensite. These new precipitated phases can act as new crack nucleation sites. The e€ect of sensitization treatment was especially negative for MP, SA and 10%CW samples, if a softening tendency was observed for higher sensitization times. In a previous study this same tendency was observed in relation to the chromium depletion of the grain boundaries, which suggested that the process of initial cracking could be related to higher levels of IGC detected in the samples with lower degrees of CW. Important di€erences were detected between the initial and mean crack initiation rates that indicated that the crack nucleation was not a uniform process throughout the test. The majority of the cracks are initiated in the ®rst stages of the test, for which the initial nucleation rate could be a more suitable parameter to control initiation cracking.

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1533

Fig. 11. E€ect of degree of prior CW on crack propagation rate of sensitized (temperature ˆ 755°C and time ˆ 15, 30 and 60 min) and non-sensitized type 304 SS in a boiling 42% MgCl2 solution.

3.3.3. Crack propagation rate The process of crack growth was proven to be strongly dependent on the dimensions of the cracks studied, and the cracks of longer length were those that propagated at a higher rate. Therefore, the growth is a statistical process, where a ®nite growth probability exists; probability is much higher as the dimensions of the initial crack are higher. The growth of the shorter cracks seemed to be inhibited by the decrease of stress that the propagation of the larger cracks generated. Furthermore, as expected from the geometric con®guration of the specimen, the propagation rate was not uniform during the test, but rather decreased as it advanced. The latter obliges us to consider the e€ect of the parameters being studied on the initial and mean propagation rate and to select a ®nite population of cracks, the longer ones, from those formed in the ®rst 30 min exposure. In Fig. 11 one can observe in a joint way, the e€ect of CW and sensitization time on the initial and mean crack propagation rates. The same as thing that occurred in the crack initiation process, the propagation also seemed to be strongly dependent on the degree of prior CW. On the other hand, the results obtained were completely coherent with the failure times and initiation rates exposed. The higher crack initiation rates appeared in scarcely deformed samples or in samples with very high deformation levels, while the samples with deformation around 20% presented lower crack propagation rates. The e€ect of sensitization treatment was especially important in MP, SA and 10%CW samples. The di€erences between both propagation rates (initial and mean) were only signi®cant in highly deformed and subsequently sensitized samples, where the mean

1534

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

propagation rates were considerably higher than the initial crack propagation rates. This will be discussed later on.

4. Discussion All the tests carried out revealed that the degree of prior CW had a decisive in¯uence on the chloride SCC susceptibility of Type 304 SS, especially when it was submitted to a sensitization process like welded joints. The same conclusion was reached on previous studies where the in¯uence of these variable were being investigated against IGC and TGC behaviour of Type 304 SS using a quantitative discrimination by image analysis [34]. Results indicated a transition from IGC to TGC as a function of prior level of CW and thermal sensitization treatment. As a general rule, for lower CW levels susceptibility to IGC was at its highest, and for higher CW levels there was a greater susceptibility to TGC. The results obtained in this work seem to suggest the existence of a strong interrelation between the IGC±TGC processes with SCC behaviour of Type 304 SS. Taking this into consideration, these results are discussed hereafter. Potentiodynamic anodic scans showed that the prior CW is the most important metallurgical variable in the anodic behaviour of steel. It is capable of originating a secondary activity±passivity transient at potential intervals [( 200 mVSCE ) and (‡100 mVSCE )]. This new anodic peak can be associated with the anodic dissolution of a new phase i.e. strain induced martensite. This phase has been observed by microstructural examination and has been measured by a magnetic test [43]. Furthermore, the current density of this new transition is lower when sensitization time is increased. This coincides with the additional transformation of martensite into secondary austenite and chromium carbides by heat treatment detected by magnetic measurements. The correlation between the results of the two electrochemical tests carried out and accompanied by the microstructural examination after the tests, suggested that both are adequate for predicting, at least qualitatively, the SCC susceptibility. In our case it is especially interesting to distinguish between the TGSCC and the IGSCC risk. In this way, the susceptibility of TGSCC will be higher as the range of the active±passive transient is higher (primary and secondary) where the necessary conditions are given so that dissolution can exist on crack tip and passivation on the wall tips and bulk matrix. On the other hand, intergranular nucleation and propagation crack could occur over a wider range of potentials known between these two transitions, therefore, the IGSCC could be more probable if the stability of the passive layer were lower. Following this proposed criterion, the results indicated that the sensitization had a negative e€ect on the susceptibility to the IGSCC for MP, SA and 10%CW samples as the sensitization time increased the passive current density increased and the passive potential range decreased. Precisely in these samples had been detected high IGC susceptibility [34].

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1535

For the rest of the samples the e€ect of sensitization time on anodic behaviour was radically di€erent. In these cases an important increase of the critical potential range activity±passivity was observed where TG crack was the most likely to occur. This coincided with the high susceptibility to TGC registered in these samples. The microstructural analysis con®rmed the detected change on the SCC mechanisms originated by the CW. The anodic polarization double scan curves have a restricted use being excessively dependent on the analysed variables, but considering the correlation of the last test it could be successfully used to predict qualitatively the TGSCC and IGSCC susceptibilities. The susceptibility to TGSCC could be evaluated by the critical potential ranges as has been proposed by other authors, but the IGSCC susceptibility needs another parameter. This parameter could be the loss of stability in the passive layer. The use of this criterion as a measurement of IGSCC susceptibility would mean assuming that the slow scan is able to form a stable passive layer on the grain boundary when sensitization does not exist. While in the sensitized samples the existence of chromium depletion in the grain boundary will form a less stable layer (with less Cr content, and higher Ni and Fe content) which will break at more negative potentials. On the other hand in the fast scan, enough time does not exist to detect these di€erences with which the di€erences between the passive potential ranges of two scans can be used to predict the IGSCC risk. The hypothesis can be assumed taking into consideration that these scans have di€erent passive current densities that would indicate di€erent passive layer nature. Furthermore, these differences do not appear in samples with high deformations where the mechanisms of attack are transgranular. The MgC12 test permits us to quantify the e€ect of the variables being studied through the use of parameters such as ``failure time'', ``crack initiation rate'' and ``crack propagation rate''. The evaluation of all these parameters with the degree of CW and sensitization treatment has been proven to be analogous but the crack initiation rate and crack propagation rate allow us to discriminate with higher precision the e€ect of sensitization time. As expected the solution-annealed treatment carried out had a very bene®cial e€ect. An important increase in time of failure with respect to an as-received condition was obtained. In spite of using short times of sensitization, this treatment promoted an important reduction of failure time, and developed the intergranular cracking form. It seems obvious that these changes should somehow be related to the abundant carbide precipitation along the grain boundary generated by the sensitization treatment. Therefore chromium depletion mechanisms will explain the IGSCC of sensitized SS. With respect to the e€ect of CW, it was expected that TGSCC susceptibility would be higher the higher the degree of CW as a consequence of the increase of residual mechanical stress produced by strain. However, in this study it has been left clear that this is only accomplished for high CW, with which the e€ect of CW is much more complex. On the one hand, very low deformations, 10%CW, are able to originate exceptionally high reductions of time failure with respect to SA samples,

1536

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

while the slightly higher CW samples, 20%CW, present failure times similar to SA condition. On the other hand, a change in mechanisms has been detected from IGSCC to TGSCC as CW increased. Furthermore, this behaviour is not an isolated fact, but rather appeared once again in CW samples subsequently submitted to a sensitization treatment. The latter suggests that microstructural factors originated by CW that can explain this behaviour must exist, aside from the mechanical factor. As we have stated, low degrees of deformation have a clear undesirable e€ect on IGSCC susceptibility, that the higher the degree of sensitization. The most important microstructural factor that generated these low degrees of deformation was the increase on the dislocations of grain boundary. Firstly, this increased the number of crack nucleation sites and secondly decreased the stability of the passive layer provoking a fast crack propagation. This is in accordance with the intergranular crack path and with the high initiation and propagation rates detected. When the material is submitted to a sensitization treatment a massive chromium carbide precipitation is produced in these speci®c sites generated by CW, where the di€usion of chromium is higher; therefore, the depletion chromium mechanisms explain the high IGSCC susceptibility detected. The increase of crack nucleation sites is not proportional to the sensitization time, but rather a clear tendency of softening for higher sensitization times. This is because at such a prolonged time, the matrix is already being depleted of chromium and, therefore, there is a partial healing of grain boundaries. Passing a certain degree of critical CW a slight increase of resistance to SCC occurs, coinciding with the appearance of a transgranular mode of cracking. For even higher deformations, once again all parameters show a deterioration and the TGSCC is produced as a predominant failure factor in all of them. This occurs for all material with CW equal to or greater than 30%, for whichever degree of sensitization. This new change in the cracking mode is narrowly related, aside from mechanical factors, with the microstructural changes originated by CW. For suciently elevated deformations the fact that Type 304 SS are low stacking fault energy material, begins to be important. Because of this, the increase of CW drives to an increase in grain distortion and on the number of slip bands in the grain. At the same time, the existence of strain induced martensite has been veri®ed. The SCC results and the TG cracking mode suggests that the slip bands became electrochemically more active than the grain boundary to act as crack nucleation sites. Consequently, the anodic dissolution slip steps, directly ®xes the crack path. The sensitization e€ect is not as important for these samples with high degrees of CW. Only for CW higher than or equal to 40%, after 60 min of sensitization, are very low time failure and high crack velocities observed. In these cases an important transgranular cracking is observed (and even fractures) outside the maximum stress zone of the sample. This fact indicates that the mechanical stresses produced by cold working are sucient to induce, on its own, a SCC phenomenon. On the other hand, and as this appears after prolonged sensitization, we should think that other microstructural e€ects also have an in¯uence. Previous studies [34] showed that prior CW levels exceeding 30% caused higher TGC values for this degree of sensitization. An extensive chromium carbide pre-

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

1537

cipitation was detected in the matrix in the areas generated by deformation, also the existence of chromium depletion in those areas was veri®ed. Therefore, we can consider that the cracking of these samples is a simple phenomenon of transgranular corrosion assisted by stress. At the same time as IG assisted by stress is widely reported on, the transgranular corrosion assisted by corrosion is practically unknown. The condition of the actual steel supply, MP condition, means an important IGSCC risk. In this case can be observed intermediate results between SA and 10%CW conditions, this suggests that the steel presents a certain degree of prior CW. This explains the higher stress bends used to make U-bend specimens for MP with respect to SA. In this case the e€ect of sensitization is strongly dependent on the treatment time. The short sensitization times scarcely a€ect or even improve the resistance to IGSCC. The stress relief that the thermal treatment attributes could explain this improvement. On the contrary to this, treatment times of 60 min drive to a massive intergranular cracking, even outside the area of maximum stress. In this case the intense chromium depletion in grain boundary explains this fact. Cracking can occur outside the most stressed area by propagation inhibition that is made by the compressed stress existing in the area. This explains the di€erences found between the initial and mean propagation rates. The failure mode can be quali®ed as IGSCC assisted by stress. This can also be seen for 10%CWS60 samples. Considering the previous argument, we could conclude that a critical level of CW exists, evaluated at 20%. After this, a change in cracking mechanisms was detected. Initiation and propagation of transgranular crack was observed instead of intergranular mode, the latter being more closely linked to low degrees of strain. The competition between the active dissolution of grain boundaries and slip bands was the main cause of the e€ect of CW, aside from the strain induced martensite anodic dissolution, being virtually impossible to separate the contribution of these e€ects. Nevertheless, this is in agreement with some of the published results for other austenitic SS, where martensite transformation does not exist [28]. The competition of mechanisms equally appears after sensitization, the critical level of CW (20%) is the one that delimits the negative in¯uence of sensitization. Lower deformations originate IGSCC risk directly related to the IGC susceptibility and deformations superior to the limit originate a high TGSCC risk directly related to TGC susceptibility. As a result to the previously mentioned, an optimum level of CW exists ± estimated at about 20% for this steel ± with minimum sensitivity to IGSCC, which is even better than that for the solution condition, and where the risk of TGSCC is practically negligible. Therefore, this would be an optimum registered state for the Type 304 SS steel, if we want to obtain the best IGC, TGC, IGSCC, and TGSCC behaviour. 5. Conclusions · Research on SCC susceptibility of solution-annealed, CW and sensitized Type 304 SS has shown a transition from IGSCC to TGSCC as a function of prior degree of CW and thermal sensitization treatment. As a general rule, the lower levels of CW

1538

· ·

·

·

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539

have been found to increase the IGSCC susceptibility and for higher deformations there is a greater susceptibility to TGSCC. The potentiodynamic anodic polarization methods have been shown to be suf®cient for qualitative discrimination IGSCC±TGSCC. The criterion that should be used to discriminate between both forms of failure has been established. CW Type 304 SS was susceptible to SCC but the susceptibility depends to a great degree on the prior CW. The study shows that CW lower or equal to 10% caused higher levels of IGSCC. The susceptibility of IGSCC was signi®cantly better for prior CW levels exceeding 30%. The SCC mode transformed from IGSCC to TGSCC mode with increasing levels of CW. The sensitization treatment was very dangerous for the IGSCC susceptibility for solution-annealed, as-received and low degrees of CW materials. A direct correlation has been veri®ed to exist between IGSCC risk and IG risk. The sensitization treatment has also been veri®ed as being very dangerous for SCC in materials with deformations equal to or superior to 40%, where risk of failure by transgranular corrosion assisted by CW exists. CW at 20% has been shown to be optimal for the corrosion behaviour of Type AIAI 304 SS studied here. This CW state would be the best commercial condition for the purchase of Type 304 SS, especially for welding. However, the current normal condition is MP, which in fact contains some CW, and can be susceptible to extreme IGC and IGSCC following sensitization.

References [1] R.E. Smith, Progress in reducing stress corrosion cracking in BWR piping, American Power Conference Proceedings, vol. 39, 1977, pp. 232±240. [2] H.H. Klepfer, et al., Investigation of causes of the cracking in austenitic stainless steel piping, NEDO 21000-GE, vols. 1 and 2, General Electric Co., 1975. [3] R.E. Hanneman, Model for intergranular stress corrosion cracking of welded type 304 stainless steel piping, BWRs, EPRI WS-74-174, vol. 1, 1980. [4] J.C. Danko, M.E. Indig, In-reactor and ex-reactor electrochemical methods for corrosion and stress corrosion cracking testing, Predictive Methods for Assessing Corrosion Damage to BWR Piping and PWR Steam Generator Proceedings, NACE, 1982. [5] F. Zucchi, G. Trabanelli, G. Rocchini, G. Perboni, Corros. Sci. 29 (4) (1989) 417. [6] M.O. Speidel, Metall. Trans. A 12 (1981) 779. [7] K.L. Money, W.W. Kirk, Mater. Perform. 17 (7) (1978) 28. [8] J.H. Zheng, W.F. Bogaerts, Corrosion 49 (1) (1993) 42. [9] C.P. Dillon, Mater. Perform. 29 (12) (1990) 66. [10] S. Torchio, Corros. Sci. 20 (4) (1980) 555. [11] A.I. Maier, C. Manfredi, J.R. Galvele, Corros. Sci. 25 (1) (1985) 15. [12] S. Ahmad, M.L. Mehta, S.K. Saraf, I.P. Saraswat, Corrosion 38 (7) (1982) 347. [13] B.M. Gordon, Mater. Perform. 19 (9) (1980) 29. [14] P.M. Scott, Corros. Sci. 25 (8) (1985) 583. [15] Z. Shenghan, T. Shibata, T. Haruna, Corros. Sci. 41 (5) (1999) 853. [16] K. Yamanaka, M. Kowaka, Corros. Sci. 32 (1) (1983) 23. [17] K.H. Lee, G. Cragnolino, D.D. Macdonald, Corrosion 41 (9) (1985) 540. [18] J.R. Crum, Corrosion 38 (1) (1982) 40.

C. Garcõa et al. / Corrosion Science 43 (2001) 1519±1539 [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43]

1539

S. Pednekar, S. Smialowska, Corrosion 36 (10) (1980) 565. A. Bose, P.K. De, Corrosion 43 (10) (1987) 624. M.J. Povich, P. Rao, Corrosion 34 (86) (1978) 269. L. Lungberg, Low temperature sensitization studies in area atom of type 304 SS, Korrosions Problem Reactor Material, Symposium, vol. 21, 1983. J. Kuniya, Corrosion 44 (1) (1988) 21. K. Takizawa, Trans. ISIJ 20 (5) (1980) 454. Y.M. Yeon, S.I. Kwun, J. Korean, Inst. Met. 21 (2) (1983) 105. C.L. Briant, A.M. Ritter, Metal Trans. A 12 (5) (1981) 910. C.L. Briant, A.M. Ritter, Corrosion 38 (11) (1982) 596. J.H. Zheng, W.F. Bogaerts, Corrosion 49 (7) (1993) 585. M. Kowaka, H. Fujikawa, Sumitomo Search 7 (1972) 10. J.E. Truman, The e€ects of composition and structure on the resistance to stress corrosion cracking of stainless steels, British Nuclear Energy Society Symposium on E€ects of Environment on Material Properties in Nuclear Systems, paper no. 10, Institute of Civil Engineers, 1971. R.W. Cochran, R.W. Staehle, Corrosion 11 (3) (1968) 369. H.D. Solomon, M.J. Povich, T.M. Devine, Slow strain testing in high temperature water, ASTM STP 665, American Society for Testing and Materials, Philadelphia, 1979, pp. 132±148. W.L. Clarke, R.L. Cowan, J.C. Danko, Dynamic straining stress corrosion test for predicting BWR materials performance, ATMSTP 665, American Society for Testing and Materials, Philadelphia, 1979, pp. 149±169. C. Garcia, F. Martin, P. Tiedra, J.A. Heredero, M.L. Aparicio, Corrosion 56 (3) (2000) 243. ASTM Standard E-8-94. Standard test methods for testing of metallic materials, American Society for Testing and Materials, Philadelphia, 1994, p. 60. A.H. Advani, L.E. Murr, D.G. Atteridge, R. Chelakara, S.H. Bruemmer, Corrosion 47 (12) (1991) 939. ASTM Standard G-5-87, Standard reference test method for making potentiostatic and potentiodynamic anodic polarization measurements, American Society for Testing and Materials, Philadelphia, 1993, p. 71. R.N. Parkins, Corros. Sci. 20 (1980) 147. N.J. Holdroyd, R.N. Parkins, Corros. Sci. 20 (1980) 707. G.J. Theus, J.R. Cels, Slow strain-rate technique: application to caustic stress corrosion cracking studies, ASTM STP 665, American Society for Testing and Materials, Philadelphia, 1979, pp. 81±96. ASTM Standard G-36-87, Standard practice for evaluating stress corrosion cracking resistance of metals and alloys in a boiling magnesium chloride solution, American Society for Testing and Materials, Philadelphia, 1993, p. 141. ASTM Standard G-30-90, Standard practice for making and using U-bend stress corrosion cracking, American Society for Testing and Materials, Philadelphia, 1993, p. 101. C. Garcia, F. Martin, P. Tiedra, J.A. Heredero, M.L. Aparicio, Metal. Trans., submitted for publication.