The effect of test temperature on SCC behavior of austenitic stainless steels in boiling saturated magnesium chloride solution

The effect of test temperature on SCC behavior of austenitic stainless steels in boiling saturated magnesium chloride solution

Corrosion Science 48 (2006) 4283–4293 www.elsevier.com/locate/corsci The effect of test temperature on SCC behavior of austenitic stainless steels in ...

766KB Sizes 0 Downloads 15 Views

Corrosion Science 48 (2006) 4283–4293 www.elsevier.com/locate/corsci

The effect of test temperature on SCC behavior of austenitic stainless steels in boiling saturated magnesium chloride solution Osama M. Alyousif a, Rokuro Nishimura a

b,*

Department of Mechanical Engineering, Kuwait University, P.O. Box 5959, Safat 13060, Kuwait b Department of Applied Materials Science, College of Engineering, 1-1, Gakuen-cho, Sakai, Osaka Prefecture University, Osaka 599-8531, Japan Received 10 January 2006; accepted 27 January 2006 Available online 24 May 2006

Abstract The stress corrosion cracking (SCC) of the austenitic stainless steels of types 304, 310 and 316 was investigated as a function of test temperature in boiling saturated magnesium chloride solution (MgCl2) using a constant load method. Both of types 304 and 316 exhibited similar corrosion elongation curves, while the corrosion elongation curve of type 310 was different from those of types 304 and 316, in terms of the three parameters such as time to failure (tf), steady-state elongation rate (lss) and transition time to time to failure ratio (tss/tf) obtained from the corrosion elongation curves for these stainless steels. The relationship between the time to failure and a reciprocal of test temperature fell in two straight lines on a semi-logarithmic scale as well as the relationship between the steadystate elongation rate and a reciprocal of test temperature. These regions were considered to correspond to a SCC-dominated region and a hydrogen embrittlement (HE)-dominated region from the value of (tss/tf) and the fracture appearance. The relationship between the steady state elongation rates versus time to failure on a logarithmic scale became a straight line, whereas the slopes of the line for the stainless steels were different with the different fracture mechanism such as SCC and HE. It was found that the linearity of the relationship can be used to predict the time to failure for the stainless steels in the corrosive environment. In addition, type 310 did not suffer from HE, which means that type 310 showed only SCC. This would be explained by whether or not a formation of a 0 -martensite takes place.  2006 Elsevier Ltd. All rights reserved.

*

Corresponding author. Tel./fax: +81 72 254 9323. E-mail address: [email protected] (R. Nishimura).

0010-938X/$ - see front matter  2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.corsci.2006.01.014

4284

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

Keywords: A. Austenitic stainless steels; A. Magnesium chloride solutions; B. Steady-state elongation rates; C. Stress corrosion cracking; C. Hydrogen embrittlement

1. Introduction Stress corrosion cracking (SCC) behavior in austenitic stainless steels in chloride and other corrosive solutions has been extensively investigated using various methods [1–7]. The cracking mechanisms for austenitic stainless steels can be either the active path dissolution mechanism [1,2], the film-rupture mechanism [3,4] or hydrogen embrittlement [5–7]. Nishimura qualitatively inferred a transgranular SCC mechanism for type 304 in sulphuric acid [3] and Nakayama observed two cracking modes (transgranular and intergranular) for type 304 in 42% boiling MgCl2 using a SSRT [4]. On the other side, Whiteman and Holzworth both demonstrated that 310 and 304 austenitic stainless steel is susceptible to embrittlement by hydrogen [5,6] and Rhodes proposed a model for crack initiation and propagation for type 304 based on a hydrogen embrittlement mechanism [7]. It was also reported that austenitic stainless steels may undergo phase transformation from c ! a 0 due to applied stress or hydrogen charging [8,9] and a 0 -martensite is directly related to brittle fracture [10]. Hydrogen transport into the bulk of the material or a 0 -martensite transformation may be a great factor in the cracking behavior of austenitic stainless steels. The characterization of the cracking behavior however is still unresolved and needs further studying. Material structure and corrosion reaction kinetics play an important role on the material’s behavior in corrosive environments. The constant load method can produce a corrosion elongation curve which can be used to obtain three useful parameters namely, time to failure (tf), steady-state elongation rate (lss) and transition time to time to failure ratio (tss/tf). These parameters can be used in analyzing the failure behavior and the steady state elongation rate (lss) can be used to predict time to failure (tf) [1–3,11]. Recent studies found that the characteristics of the steady state elongation rate can be applied to SCC of type 430 ferritic stainless steels in acidic chloride and sulfate solutions by using the constant load method [12]. On the ground of such characteristics, some threshold values of the critical temperature for SCC were estimated. The objectives of this research work are: (1) to investigate the effect of test temperature on the susceptibility of austenitic stainless steels to stress corrosion cracking, (2) to evaluate the three parameters described above for stainless steels in saturated boiling magnesium chloride solutions, and (3) to determine the cracking mechanism for austenitic stainless steels in saturated boiling magnesium chloride solutions. 2. Experimental The specimens used were the commercial types 304, 316 and 310 austenitic stainless steels. The geometry for SCC experiments was as follows: the gauge length 25.6 mm, its width 5 mm and the thickness 1 mm. The specimen geometry is shown in Fig. 1. The chemical compositions (wt.%) for the specimens used were shown in Table 1. The specimens were solution annealed at 1373 K for 3.6 ks under argon atmosphere and then water quenched. Prior to the experiments, the solution annealed specimens were polished to 1000 grit emery paper, degreased with acetone in an ultrasonic cleaner and washed with

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

4285

Fig. 1. Specimen geometry (dimensions in mm).

Table 1 Chemical composition (wt.%) and mechanical properties of the austenitic stainless steels specimens used

SS304 SS310 SS316

C

Si

Mn

P

S

Ni

Cr

Mo

rYield (MPa)

rTensile (MPa)

0.06 0.05 0.06

0.35 0.84 0.70

0.96 1.27 0.96

0.027 0.016 0.031

0.004 0.001 0.004

8.13 19.30 10.15

18.20 24.76 16.98

– – 2.22

276 275 323

691 575 636

distilled water. After the pretreatment, the specimens were immediately set into a SCC cell. The test solution was boiling magnesium chloride solution at different concentrations for boiling test temperature variation. The constant applied stress of 300 MPa was used, which was associated with SCC dominated failure in this type of environment for austenitic stainless steels [13]. All experiments were carried out under an open circuit condition. A liver-type constant load apparatus (lever ratio 1:10) to which three specimens can be separately and simultaneously attached was used with a cooling system on the top to avoid evaporation of the solution during the experiments. The specimens were insulated from rod and grip by surface oxidized zirconium tube. A change in elongation of he specimens under the constant load condition was measured by an inductive linear transducer with an accuracy of ±0.01 mm. 3. Results 3.1. Corrosion elongation curve The corrosion elongation curves for austenitic stainless steels, types 316 and 310, in boiling magnesium chloride solution are shown in Fig. 2, where the corrosion elongation curve of type 304 exhibited a similar behavior as that of type 316. From these curves the three parameters were obtained for each specimen at every test temperature: time to failure (tf), steady-state elongation rate (lss) for the straight part of the corrosion elongation curve and transition time to time to failure ratio (tss/tf), where transition time (tss) is the time when the straight part of the elongation curve deviate from linearity. On the basis of the corrosion elongation curve, the relationship between the parameters (tf), (lss), (tss/tf) and the reciprocal of the test temperature for all the tested specimens were shown in a logarithmic representation to establish the characteristics of SCC as a function of test temperature. For type 316 at T = 416 K, the time to failure tf was considerably shorter than type 310 and tss was close to tf which resulted in a tss/tf value close to unity. While for type 310,

4286

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293 10 time vs SS 316-416K time vs SS 310-416K

Displacement / mm

8

Boiling Saturated MgCl2 and σ= 300 MPa

6

lss 4

lss 2

t ss

0 0

2000

tf 4000

t ss 6000

tf 8000

10000

12000

14000

Time / s Fig. 2. Sample of the corrosion elongation curve for types 316 and 310 at T = 416 K and r = 300 MPa.

the tss value was approximately half the time to failure and this will have an indication as discussed later. 3.2. Test temperature dependence of three parameters Fig. 3 depicts the logarithm of the parameter tf versus the reciprocal of the test temperature for the three stainless steel types used in the experiments. For type 304, the relationship between tf and 1/T falls in a straight line, shown as region-I, until a test temperature of about 413 K (1/T = 2.42 · 103), below which the relationship deviates from the linearity. The region from 413 K (1/T = 2.42 · 103) to a threshold test temperature, approximately 403 K (1/T = 2.48 · 103), below which little fracture takes place within a laboratory time scale, was called region-II. For type 316, the parameter tf was also grouped into two regions; region-I at test temperatures above 424 K (1/T = 2.36 · 103) and region-II with test temperatures below 424 K (1/T = 2.36 · 103) up to a threshold test temperature of about 408 K (1/T = 2.45 · 103). The tf values of type 310 showed slight sensitivity to test temperature. The tf values for type 310 were grouped in a straight line which constitutes region-I until a threshold test temperature of about 408 K (1/T = 2.45 · 103) without region-II. Therefore, in Fig. 2, the elongation curve for type 310 at 416 K (1/T = 2.40 · 103) falls in region-I, while that for type 316 at 416 K (1/T = 2.40 · 103) falls in region-II. The relationships between the logarithm lss and 1/T for all stainless steels are shown in Fig. 4. For type 304 the values of lss can also become one straight line with a deviation at 413 K (1/T = 2.42 · 103) which corresponds to region-I identified in Fig. 3 and at test temperatures below 413 K (1/T = 2.42 · 103) the lss values for type 304 is grouped in region-II. For Type 316 lss values were grouped in two regions; region-I for test temperatures above 424 K (1/T = 2.36 · 103) and region-II for test temperatures between 424 K

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

4287

Fig. 3. The logarithms of (tf) versus the reciprocal of the test temperature (1/T) for types 304, 316 and 310 in saturated boiling MgCl2 at r = 300 MPa.

4288

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

Fig. 4. The logarithms of (lss) versus the reciprocal of the test temperature (1/T) for types 304, 316 and 310 in saturated boiling MgCl2 at r = 300 MPa.

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

4289

1.0

Region II 0.8

Region I

tss/tf

0.6

0.4

0.2

SS 316 SS 304 SS 310

Saturated Boiling MgCl2 and σ= 300 MPa 0.0 2.32

2.34

2.36

2.38

2.40

2.42

K/T X 10

2.44

2.46

2.48

2.50

3

Fig. 5. The logarithms of (tss/tf) versus the reciprocal of the test temperature (1/T) for types 304, 316 and 310 at r = 300 MPa.

(1/T = 2.36 · 103) and 408 K (1/T = 2.45 · 103). In the case of type 310, the values of lss can be grouped in a straight line which corresponds to region-I in Fig. 3. The third relations between tss/tf and 1/T for all stainless steels are shown in Fig. 5. For type 304 the tss/tf values were grouped into two regions as well as those (region-I and region-II) in Figs. 3 and 4. The region-I had approximately a constant value of tss/tf = 0.57 ± 0.02 for type 304 at test temperatures less than 413 K (1/T = 2.42 · 103), while region-II has tss/tf values close to 0.82 ± 0.02 for test temperatures higher than 413 K (1/T = 2.42 · 103). These constant values were found to apply to region-I and region-II for types 316 and 310. 3.3. Fracture surface appearance The scanning electron microscopy photographs for the fracture surface appearance for the austenitic stainless steels tested in boiling magnesium chloride solution were investigated. Fig. 6 shows the fracture surface appearances of types 304 and 310, which were grouped as region-I. They showed that the cracking mode for types 304 and 310 was predominantly transgranular over the test temperatures in region-I, which stands for the occurrence of stress corrosion cracking (SCC) [14]. On the other hand, for the fracture surface appearances in region-II tests, as shown in Fig. 7 for type 316, the fracture surface appearances changed to a mixture of intergranular and transgranular mode (in Fig. 7(a)) or predominantly intergranular mode (in Fig. 7(b)), which means that hydrogen embrittlement (HE) would take place partly or predominantly [14]. The amount of transgranular cracking in comparison to that of intergranular cracking decreased with decreasing temperatures until reaching the cracking mode of completely intergranular. In the case of type 310 the cracking mode was transgranular over the test temperatures, indicating that the dominating cracking mechanism was stress corrosion cracking. It was estimated that the differences in the fracture appearances for these stainless steels were attributed to a difference in corrosion behavior of these stainless steels to strain induced martensite formation at the grain boundaries as described later.

4290

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

Fig. 6. (a) Transgranular cracking for type 304 at T = 527 K, and (b) transgranular cracking for type 310 at T = 414 K (·550).

Fig. 7. (a) Mixed mode cracking behavior for type 316 at T = 416 K, and (b) intergranular cracking for type 316 at T = 414 K (·550).

4. Discussion 4.1. A parameter for predicting time to failure and estimation of the difference in fracture mechanism Fig. 8 shows the relationships between the log-tf and log-lss for all stainless steels. The log-tf versus log-lss curves for types 304 and 316 became two straight lines with slopes m = 1.95 at region-I and m = 1.30 at region-II, while type 310 had the same slope as m = 1.95 at region-I. These linear relations can be used as a predictive tool for the time to failure for austenitic stainless steels by means of the steady-state elongation rate lss even in saturated magnesium chloride solutions as well as hydrochloric acid and sulphuric acid solutions [15,16] as presented as follows: log lss ¼ m log tf þ C

ð1Þ

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

4291

1e-5

Elongation rate (lss) / m.s-1

Region-I tss/tf=0.57 Region-I tss/tf=0.57

1e-6

Saturated Boiling MgCl2 and σ= 300 MPa

1e-7

1e-8

SS316 SS304 SS310

Region-II tss/tf=0.82

1e-9

1e-10 1e+2

1e+3

1e+4

1e+5

1e+6

1e+7

Time to failure (tf) / s Fig. 8. The relation between time to failure (tf) and the logarithms of (lss) for types 304, 310 and 316 in boiling magnesium chloride solution at r = 300 MPa.

where m in Eq. (1) is the slope depending on the test temperature and C is the stress dependant constant. In other words, the linear equation implies that the lss value becomes a useful parameter for predicting tf independent of chloride concentration and test temperature because the lss can be obtained at a time within 10–20% of tf from the corrosion elongation curve. On the other hand, the specimens have no SCC susceptibility within the laboratory time scale, when the order of the magnitude of the steady state elongation rate becomes 1010 m/s or lower. This implies that the steady state elongation rate also becomes a parameter for the assessment of whether SCC takes place or not. In addition, in the case that the value of tss/tf becomes larger than the constant value (0.57 ± 0.02), the failure of the specimens does not take place by SCC, but by HE. Thus, by using both the order of lss and the value of tss/tf, the more accurate assessment of the SCC susceptibility can be conducted and as a result the critical values of the environmental factors can be determined. The two straight lines corresponding to region-I and region-II in Fig. 8 were an indication of the existence of two cracking mechanisms as reflected by the change in slope in the log-tf versus log-lss line and this notion was implicated by the post fracture fractography for types 304 and 316. In the case of type 310 austenitic stainless steel, the log-tf versus loglss relation became a single straight line with slope m = 1.95 which was similar to the slope of region-I for types 304 and 316. This may be interpreted as an identical cracking behavior for these types of stainless steel at those test conditions. 4.2. Relationship between formation of martensite and fracture (SCC or HE) Type 304 is a metastable austenite and it is very susceptible to martensite formation [17]. This type of austenitic stainless steels was used extensively in investigating hydrogen embrittlement behavior because of its material composition, structure ‘‘instability’’, intergranular cracking mode and ability to phase transform upon hydrogen charging or

4292

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

induced strain. The material displayed two different behaviors on the basis of the test temperature dependency. Hydrogen embrittlement behavior occurred at test temperatures below the critical temperature and SCC behavior occurred at test temperatures above the critical temperature, where the critical temperature is the temperature between region-I and region-II. It can be concluded based on the results of using the constant load method that type 304 is very susceptible to hydrogen embrittlement as well as stress corrosion cracking in saturated boiling magnesium chloride. This notion will be further discussed in the following section. Type 316 has also a metastable austenite and it is moderately susceptible to martensite formation. The material exhibited a dual cracking behavior similar to type 304 and this may be attributed to the closely identical chemical composition for both types. Type 316 had a higher critical temperature than type 304. This type of austenitic stainless steel can also be branded as susceptibility to both of hydrogen embrittlement and SCC failure depending on the test temperature in magnesium chloride solution as shown in the post fracture micrographs. But when it comes to type 310 which have a stable austenite structure, the material did not exhibit any hydrogen embrittlement cracking features based on the post fracture analysis of the broken samples. This behavior can be explained by the material’s high content of chromium which is a natural inhabitant of martensite transformation and this may be the reason that this type of stainless steels is susceptible to SCC only in magnesium chloride solutions. Hydrogen is a product of the electrochemical reactions at the metal surface. For metastable austenitic steels like types 304 and 316, the strain induced martensite along the grain boundaries will enhance the hydrogen permeation [18–20]. Martensite structure has a very high diffusivity coefficient and very small hydrogen content compared to those of the austenite. This may explain the intergranular cracking as the crack preferential path is through the martensite structure facilitated by the higher diffusion rates of hydrogen. At high temperatures, the corrosion rate will be higher than the corrosion rate at low temperature and this will result in a predominant metal dissolution at the crack tip faster than the hydrogen entry to the structure and this will enhance the predominance of SCC mechanism. The lower temperature means lower corrosion reaction rates and this will prohibit the film rupture mechanism hence letting the hydrogen embrittlement to take place and thus controlling the cracking mechanism. It is worth noting that increasing the amount of a 0 -martensite content in the structure produced by applied stress (or strain) will increase the susceptibility of the material to hydrogen embrittlement [20]. Type 310 is a stable austenitic stainless steel and no evidence of martensite formation was found so that the only possible cracking mechanism will be the film-rupture mechanism [21]. 5. Conclusions The following conclusions can be drawn from this work: • Using a constant load method, the effect of test temperature on stress corrosion cracking of austenitic stainless steels can be evaluated. There was two regions obtained, one for SCC dominated failure and the other for hydrogen embrittlement dominated failure.

O.M. Alyousif, R. Nishimura / Corrosion Science 48 (2006) 4283–4293

4293

• The cracking mechanism for type 304 was transgranular between test temperatures of 413 K and 428 K. The cracking mode for type 304 changed to intergranular at test temperatures between 412 K and 403 K. The cracking mechanism for type 316 was transgranular between test temperatures of 424 K and 428 K. The cracking mechanism for type 316 was a mixture of transgranular and intergranular cracking for test temperatures between 424 K and 408 K. The cracking mechanism for type 310 was transgranular at test temperatures higher than 408 K and no intergranular cracking behavior was observed. • The reason for the intergranular cracking mechanism for types 304 and 316 might the formation of martensite at the grain boundary where as for type 310 no martensite transformation existed which contributed to the cracking mechanism uniformity. This behavior for types 304 and 316 requires further investigation.

Acknowledgements The authors acknowledge the partial funding provided by Kuwait University under grant number EM05/02. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

[11]

[12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

T.P. Hoar, J.M. West, Proceedings of Royal Society A268 (1962) 304. T.P. Hoar, J.C. Scully, Journal of Electrochemistry 111 (1964) 348. R. Nishimura, Y. Maeda, Corrosion Science 46 (2004) 343. T. Nakayama, M. Takano, Corrosion 42 (1) (1986) 10. M.B. Whiteman, A.R. Troiano, Corrosion 21 (1965) 53. M.L. Holzworth, Corrosion 25 (3) (1969) 107. P.R. Rhodes, Corrosion 25 (11) (1969) 462. H. Hanninen, T. Hakkarainen, Metallurgical Transactions A 10A (1979) 1196. N. Narita, C.J. Altstetter, H.K. Birnbaum, Metallurgical Transactions A 13A (1982) 1355. J.C. Scully, The role of hydrogen in stress corrosion cracking, in: A.W. Thompson, I.M. Bernstein (Eds.), Effect of Hydrogen on Behavior of Materials, Proceedings of an International Conference, AIME, 1975, p. 129. R. Nishimura, H. Sulaiman, Stress corrosion cracking of sensitized type 316 austenitic stainless steel in pure sulfuric acid solution, 12th International Corrosion Congress: Corrosion control for low cost reliability (1993) 4325. R. Nishimura, Y. Maeda, Corrosion Science 46 (2004) 755. O. Alyousif, R. Nishimura, unpublished data. R. Nishimura, Corrosion Science 34 (11) (1993) 1463. R. Nishimura, K. Kudo, Corrosion Science 45 (4) (1989) 308. R. Nishimura, Corrosion Science 34 (11) (1993) 1859. C.L. Briant, Metallurgical Transactions A 10A (1979) 181. M.R. Louthan, R.G. Derrick, Corrosion Science 15 (1975) 565. D. Eliezer, D.G. Chakrapani, C.J. Altstetter, E.N. Pugh, Metallurgical Transaction A 10A (1979) 935. H. Hanninen, T. Hakarainen, Corrosion 36 (1) (1980) 47. S.S. Birley, D. Tromans, Corrosion 27 (2) (1971) 63.