The influence of the mechanical test conditions on the corrosion fatigue behaviour of austenitic stainless steel in chloride solutions

The influence of the mechanical test conditions on the corrosion fatigue behaviour of austenitic stainless steel in chloride solutions

Corrosion Science, Vol. 29, No, 5, pp. 567-576, 1989 Printed in Great Britain 0010--938X/89 $3.00 + 0.00 © 1989 Pergamon Press plc THE INFLUENCE OF ...

432KB Sizes 5 Downloads 35 Views

Corrosion Science, Vol. 29, No, 5, pp. 567-576, 1989 Printed in Great Britain

0010--938X/89 $3.00 + 0.00 © 1989 Pergamon Press plc

THE INFLUENCE OF THE M E C H A N I C A L TEST CONDITIONS ON THE CORROSION FATIGUE B E H A V I O U R OF AUSTENITIC STAINLESS STEEL IN CHLORIDE SOLUTIONS T.

MAGNIN*,

D. DESJARDINSt and M. PU1GGALI'i-

* D 6 p a r t e m e n t mat6riaux, Ecole des Mines, 158 cours Fauriel, 42023 Saint-Etienne, France; and -t Laboratoire de M6canique Physique, U . A . CNRS 867, Universit6 de Bordeaux I, 351 cours de la Lib6ration, 33405 Talence Cedex, France

A b s t r a c t - - T h e corrosion fatigue behaviour of the 316L alloy in chloride solutions is analysed and compared to the stress corrosion cracking behaviour of the same alloy in the same electrochemical conditions. T h e kind of corrosion fatigue test (cyclic tensile stress, tension-compression strain) is shown to have a very sensitive effect on the resulting damage. The tension-compression test is m u c h more severe than the cyclic tensile stress and induces damage which can occur even if stress corrosion cracking is not observed. INTRODUCTION

IT HAS BEENsuggested that the same embrittlement mechanisms may be operative in both stress corrosion cracking (S.C.C.) and corrosion fatigue (C.F.) in metals and alloys such as the austenitic stainless steels in chloride solutions. 1Little attention has been paid to the interracial region between S.C.C. and C.F., particularly for crack initiation mechanisms. Moreover, even if the electrochemical damage is qualitatively similar in S.C.C. and in C.F., the mechanical surface damage is quite different in monotonic deformation, in cyclic tension-tension deformation and in tension-compression tests, e'3 This mechanical influence must be taken into account for the study of crack initiation mechanisms which are related to the surface interactions between the electrochemical and mechanical damage. The aim of the present study is: (i) to analyse the corrosion fatigue behaviour of a 316L austenitic stainless steel in chloride solutions as a function of the kind of tests (cyclic tensile stress and symmetrical tension-compression strain at low frequency) and (ii) to compare the C.F. properties with the S.C.C. properties of the same alloy in the same electrochemical conditions. EXPERIMENTAL METHOD C.F. and S.C.C. tests are performed on a water quenched passive 316L alloy in three different corrosive solutions: (i) a 30 g 1-I NaCI solution at 20°C and at free corrosion potential E0, (ii) a 44% MgCI~ solution at 153°C with a slight cathodic polarization ( - 40 m V under the free corrosion potential E0), and (iii) a 310 g I-t MgCI 2 + 110 g I -I NaC1 solution at ll0°C and at the free corrosion potential. Two different C.F. tests have been conducted: (i) cyclic tests in the 44% MgClz solution 4 on wires of 1.5 m m diameter (with a surface of 360 m m 2 in contact with corrosive solution), as indicated on Fig. 1. The loading period is conducted at a strain rate of 2.6 × 10 -4 s -I (at this strain rate cracking cannot occur during loading) and then an alternative stress is applied (R = Omin/Umax>0) with a given strain rate k which imposes the frequency of the test, and (ii) tension-compression tests in the other solutions on smooth Manuscript received 19 January 1988; in a m e n d e d form 24 April 1988. 567

568

T. MAGNIN,D. DESJARDINSand M. PUIGGALI

R = -F~io -

Looding period

T=l/f --1=

Fmox "I

f

I Fmox

?

Fmin

m

Time

Fzo. I. The experimental procedure for cyclic tensile tests. The definition of mechanical parameters.

specimens (15 mm gauge length, 6 mm diameter) at imposed symmetrical plastic strain amplitude + Aep/2and at constant strain rate ~. During cycling at imposed electrochemical potential (E = E0 or E = E0 - 40 mV), the corrosion current transient characteristics are recorded, paying particular attention to the peak current densities (for more details, see4'5). To quantify the evolution of the C.F. damage, replicas of the specimen surface were taken for different fractions of the fatigue lifetimes and were then observed using optical and scanning electron microscopy as described elsewhere. 6 S.C.C. reference tests are performed; (i)at imposed constant load in the 44% MgCI2 solution, and (ii) at imposed strain rate using the slow strain rate technique in the other solutions on the same specimens as for C.F. E X P E R I M E N T A L RESULTS

C.F. behaviour at imposed cyclic tensile tests D u r i n g the tests the elongation of the specimen and the c o r r e s p o n d i n g evolution of the corrosion current are simultaneously r e c o r d e d as shown by Fig. 2. The evolution of the current transients (showing anodic peaks) emphasizes a large change in the average current with a superposition of cyclic small amplitude events. A c o m p a r a t i v e analysis of the evolution with time of the specimen elongation in an inert a t m o s p h e r e and in the 44% MgC1 e solution allows the separation of the crack initiation period (or incubation period ti) and the crack p r o p a g a t i o n period as it is usually d o n e for S.C.C. tests. 4 T h e n , the crack growth rate Vf can be calculated as described elsewhere: 7 cracking s u r f a c e Vf = p r o p a g a t i o n time

So (1 _ _~_~)

tp

where, S Ois the initial cross section of the sample, tp the p r o p a g a t i o n time, a the initial stress and err the ultimate tensile strength. This p a r a m e t e r which represents the increase of the c o r r o d e d area per unit of time, will be expressed in m m z h - 1. Figure 3

Austenitic stainless steel in chloride solutions

1.0

"2 E

569

EvoLution of t h e c u r r e r ~ t r a n s i e n t s

\

during t h e t e s t

05

!jjjJ Ijjj,(

O --O.12 IO.fX:)Ill

211

22

2;

510

5LI

71l

712

7~3

= t (mn)

xlO 3 26

5

E current

25

.J

4

24

3

23

2

22

I

2i

Average eLongation

~ _....__...__---~ - I I0

I ti=45

E = E o - 4 0 mV

0

-05

i 20

316L MgCL2, 4 4 % , 153°C

I I0 z

t(mn) Fro. 2. Cyclic tensile tests at om~X = 3 3 0 M P a , R = 0 . 4 5 a n d i = 1.5 x | 0 - 4 s L T h e evolution of the average current i and of the transient current during the test.

shows the influence of the imposed strain rate k on the cracking velocity Vf for different mechanical conditions. It must be emphasized that the selected Omax and ami, values always correspond to the plastic deformation domain of the 316L alloy. From Fig. 3, it can be seen (i) the lower the strain rate, the larger Vfis, as it has been shown for S.C.C. tests, 8 and (ii) a critical value of g is observed (k - 10 - 4 s - 1 ) under which Vf and the corresponding C.F. damaging effect are more sensitive. Comparing the observed values of Vf in C.F. (Fig. 3) and for S.C.C. at imposed stress O'max ( V f i s about 1.1 mm 2 h -1 in this case for the same applied potential) it can be noticed that, whatever the mechanical conditions, Vfis always lower for C.F. than for S.C.C. at imposed load. The C.F. test at imposed cyclic tensile stress is less damaging than the classical S.C.C. test at imposed constant load. The proposed analysis is based on the determination of the transition between the crack initiation and crack propagation periods. Looking at the evolution of the current transients and of the specimen elongation (Fig. 2), it can be observed that the

570

T. MAGNIN, D . DESJARDINS a n d M. PUIGGALI f G,no~" 250 MPo • ~.R " 0 . 2 9 [
Cracking velocity for / constant load test


• ~R =0.45

MPo

1.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

316L

IjC:;

MgClz, 4 4 % , 153"C

%

~.•

vE

E= Eo-4OmV

.=

05

.

iO-6

I

I

I

10-5

10-4

10-3

(s-~) Cyclic tensile test. The influence of strain rate on cracking velocity.

FIG. 3.

cyclic evolution of the current is more regular during the propagation period determined from the value of the specimen elongation. This observation suggests that a more precise estimation of the transition from crack initiation to crack propagation is possible using the quantity of electricity generated per cycle all along the test, Figure 4 shows the cumulative value of the charge Q as a function of the number of cycles N. The transition between the crack initiation and the crack propagation periods is observed at N = 23 cycles for the applied mechanical conditions, which is very close to the value N = 20 determined from the specimen elongation.

70-

60-

3161.. MgClz, 4 4 % , 153©C

E=Eo-4OmV 50--

40-t) E 0

30-

20

I0

0

FIG. 4.

/•J I0

20

I ""130

N

(cycles)

I 40

I 50

Cyclic tensile test at amax = 250 M P a , R = 0 . 6 a n d ~ = 1.3 x 10 -5 S-1. T h e cumulative curve of the charge flowing during each cycle.

Austenitic stainless steel in chloride solutions

J

Corr. Air Corr.~t

~

11.

571

200

~Air'~-" I0 Corr Air

~

- 5 . to -4

b I00

o

I

I

I0 i

102

I 103

I I04

I I05

F1G. 5. Tension-compression

tests. The influence of the corrosive 30 g 1-1 NaCI solution on the 316L fatigue behaviour at k = 10 -2 s ~.

C.F. behaviour at imposed symmetrical tension-compression strain This kind of C.F. test is much more damaging than the previous one. Thus C.F. can occur in corrosive solutions in which S.C.C. is not observed. This phenomenon is illustrated by the comparative fatigue behaviour of the 316L alloy at 20°C in the 3.5 % NaCI solution and in air (Fig. 5). The fatigue lives in the corrosive solution are lower than in air although S.C.C. is impossible in this case whatever the type of test. 3 The influence of the 3.5% NaC1 solution is closely related to the applied strain rate as it is indicated on Fig. 6. This strain rate effect can be directly related to the cyclic evolution of the current transients (Fig. 7). At a low strain rate (/~ < 1 0 - 4 S - 1 ) the amount of depassivation is small and the repassivation at each cycle is complete, which limits the anodic dissolution: corrosion fatigue cannot occur. At a higher strain rate (for example ~ = 10 -2 s-l), the amount of depassivation is more pronounced

7000 '~ 5000

316L

--

Ni

3.5% NoCL

3000

I000 i0-~,

I iO-4

I 10-3

I 10-2

(s-') Fx6.6. Tension-compression tests. The influence of the strain rate on the reduction of the corrosion fatigue life at Aep/2 = 4 × 10 -3.

572

T. MAGNIN, D. DESIARDINSand M. PUIGGALI T

T

T

,o~

z 4 ~ ~p " IO-2s -I

oE
T

T

FIG. 7. Tension--compressiontests. The influence of the strain rate on the cyclic evolutionofJ (J = i/1.5 cm2) during the saturation regime at Aep/2 = 4 × 10 -3.

and a complete repassivation is not possible: anodic dissolution takes place continuously and corrosion fatigue occurs. This strain rate effect has been modelled elsewhere. 5 The very damaging effect due to the tension-compression tests can be explained in terms of localization of the plastic deformation and of the anodic dissolution which occur in this case. During the first cycles, the multiplication of the slip lines induces an increase of the current density, particularly of the anodic peak in tension, Jr, as shown in Fig. 8. Then fatigue slip bands form, which promote a localization of the dissolution process and a decrease o f J r. The mechanical formation of transgranular superficial microcracks due to an irreversibility of slip between tension and compression in the slip bands 2 is then accelerated by a localized anodic dissolution at the slip band emergence. This is shown on Fig. 8 which describes the simultaneous evolution of the cyclic peak stress a, the value of J r and the maximum length l of the observed superficial cracks as a function of the cumulative plastic strain (/?pcum = 2 Aep N). The evolution of the microcracks is sensitive to the environment and the transition of micro- to macrocracks induces a change in the evolution of JT. When the 100 t~m long cracks are formed, the repassivation at the crack tips is more difficult and Jr tends to increase. This clearly emphasizes the role of the localization of the mechanical damage on the corrosion fatigue mechanism and explains the much more damaging effect in tension-compression than for cyclic tensile stresses. It is particularly interesting to compare the C.F. behaviour in tension-compression with the S.C.C. behaviour of the 316L alloy observed using the slow strain rate technique. This comparison has been made on specimens of the same geometry in a 310 g 1-1 MgCI2 + 110 g 1-1 NaCI solution in which both C.F. and S.C.C. can occur. The S.C.C. resistance is quantified by the ratio between the elongation to fracture in the corrosive solution (8f(corr)) a n d the elongation to fracture in air (~'f(air))" The C.F. resistance is quantified by the ratio Ni(..... )/NiIair) where N~ is the number of cycles

Austenitic stainless steel in chloride solutions

573

Air (L =r~m) 4 0

120

~oo

.

.

t:k~m) 4 0 120

.

!i .

300

.

300

I000

-'}",.,

~

lOCK)~ , ~ 3 5%NoCL

All"

200

3.5%

n

NoCI

mm I

316L 2

-6 -5

=410-3

b T

4

~ = IO-Zs -d

I00

E

-3

-I

I 50

I ~00 ~pcum

F16. 8. T e n s i o n - c o m p r e s s i o n tests. The influence of the 3.5% N a C l solution o n the evolution of the crack length l during cycling a t Aep/2 = 4 x 10 3 and ~ = 10 -2 s -1.

corresponding to a rapid 1% decrease of the saturation peak stress (for more details, see Ref. 6). Figure 9 shows the comparison between the C.F. and the S.C.C. properties of the 316L alloy as a function of the imposed strain rate ~. T w o different domains can be observed (i) for 10 -4 s -1 < ~ < 10 -2 s -1, C.F. is encountered but S.C.C. is impossible. Moreover, the higher the strain rate, the more pronounced the C.F. damage is; (ii) for k < 10 -4 s -1 both S.C.C. and C.F. are observed and the damaging effect due to the corrosive solution increases when g decreases, for S.C.C. and for C.F. These results will be discussed later but it can be already noticed that, contrary to the situation described previously for the 316L in cyclic tensile stress, the C.F. test at

el !! 'I

I o i0-'r

I

I

I

1

I

10-6

10-5

10-4

10-3

lO-Z

(s -~ )

FIcJ. 9. A c o m p a r i s o n b e t w e e n the S.C.C. and the C.F. (at Aep/2 = 4 x 10 ~) behaviour of the 316L alloy in the 310 g 1- ~ MgCI 2 + 110 g 1- ~ NaCI solution at 110°C.

574

T. MAGNIN, D. DESJARDINS and M. PUIGGALI

imposed symmetrical tension-compression stress is more damaging than the S.C.C. test using the slow strain rate technique. DISCUSSION AND CONCLUSIONS The results obtained in the present study emphasize a marked influence of the kind of test on the C.F. behaviour of the 316L alloy and some differences between the S.C.C. and the C.F. damage, even if anodic dissolution may be involved in both S.C.C. and C.F. for the 316L alloy in the three chloride solutions at the corrosion potential. The C.F. damage is much more sensitive at imposed symmetrical tensioncompression strain than at imposed cyclic stress. The influence of the kind of corrosion fatigue test can be summarized and analysed as follows. With imposed cyclic tensile stress, the crack propagation velocities in the 44% MgCI2 at 153°C are always lower than those measured during S. C. C. at constant load. This kind of C.F. test can be considered as a sequence of slow strain rate tensile tests as suggested by other authors. 9'1° Thus the critical strain rate ~ = 10 -4 s - I under which the C.F. damage becomes sensitive (Fig. 3) is quite similar to the critical strain rate reported by other authors 11'12for S.C.C. Looking at the current transients during cycling, it is possible to determine the respective contributions to the C.F. damage of the mechanical and electrochemical mechanisms. The evolution with time of the current i is reported in Fig. 10 and can be described by the relation i'ibase = exp (-fit). The coefficient fl is proportional to the frequency of the test (Fig. 11). Thus, the slopes of the curves log (i-/base) = f(t) are not characteristic of the

i(mA) Imox

I ~ ' S ~

r/2

'

I

.

v!

.

.

r/2

o .

.

~

.

.

tc,~

.

et =127mn • t = 120mn

516L

• t • IOOmn D t -30ran

NIgCL2 4 4 % , 153"C E" Eo-4OmV

i - i==H-e -St

~o

0

I

I

L

2

I

3

I

I

4

5

t(s) FIG. 10. Cyclic tensile test at Crmax = 250 MPa, R = 0.6 and k = 1.5 × 10 -4 s -1. The decay in the current transient during one cycle at various m o m e n t s of the test (rupture time tr = 140 min).

Austenitic stainless steel in chloride solutions

575

316L 03

MgCL2,44%, 1 5 3 ~ E = Eo-4OmV

02

./

v

/ [ o'mo~~ 250 MPa

• ~R=06 f crmax"3:50 MPa • 1R " 0 4 5

A/

pm

i OI

I 0 2

I

I

03

04

f(Hz)

FIG. 11.

The evolution of the coefficient 13 of Fig. 10 with frequency for two different loading programs.

electrochemical phenomenon but are related to the mechanical parameters of the tests. The cyclic evolution of the current during the propagation period corresponds to the mechanical crack opening. Thus, the C.F. crack propagation essentially results from a S.C.C. mechanism which is influenced by the cyclic evolution of the stress but without any synergetic effect between fatigue and classical S,C.C. processes. With imposed symmetrical tension-compression strain, the damaging effect is more pronounced than observed during the S.C.C. at the imposed strain rate, contrary to the previous case. The C.F. damage is then related to a synergetic effect due to a very pronounced localization of the cyclic plastic deformation and of the anodic dissolution. Because of the slip irreversibility in the fatigue slip bands, crack initiation can take place without any corrosive solution (which was not possible for the cyclic tensile test) and the anodic dissolution accelerates the cracking process. The effect of strain localization leads to corrosion fatigue cracking in conditions for which S.C.C. cannot occur, for instance in the 30 g 1-1 NaCI solution at pH 6. When both S.C.C. and C.F. are possible (for instance in the MgCI2 + NaC1 solution at 110°C), two different domains are observed (i) when g >/~ (S.C.C.) where ~ (S.C.C.) is the critical strain rate for S.C.C., transgranular C.F. microcracks are formed in the 316L alloy by a combined effect of the plastic strain localization due to fatigue and the localization of the breakdown of the passive film which induces localization of the anodic dissolution. In this case, true C.F. occurs without S.C.C., and (ii) when g < ~" (S.C.C.), the C.F. damage is more pronounced than at g >/~ (S.C.C.): it is sensitivity enhanced by S.C.C. The same kind of correlation between S. C.C. and C.F. has also been reported recently on aluminium alloys. ]3 This paper emphasizes the relation and the differences between S.C.C. and true C.F. mechanisms which can occur in a 316L austenitic stainless steel subjected to a cyclic load in chloride solutions. This original approach will be extended to other metals in different solutions. It is the subject of a further study.

576

T. MAGNIN, D. DESJARDINSand M. PUIGGALI REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

D. A. JONES, Metall. Trans. 16A, 1133 (1985). T. MAGNIN,J. DRIVER,J. LEPINOUXand L. P. KUBIN, Rev. Phys. Appl. 19,467 (1984). J. M. LARDON, Doctorate Thesis, Saint-Etienne, (1986). P. DELIGNE, Doctorate Thesis, Bordeaux, (1986). Z. MAGNIN, L. COUDREUSEand J. M. LARDON, Mat. Sci. Forum 8, 177 (1986). T. MAGNIN, L. COUDREUSEand J. M. LARDON, Scripta Met. 19, 1487 (1985). J. J. PAUTHE,D. DESJARDINSand M. C. PETIT, Mere. Sci. Rev. Metall. 5,649 (1980). M. HELLE, D. DESJARDINS,M. PUIGGALIand M. C. PETIT, Passivity of Metals and Semiconductors (ed. M. FROMENT),p. 667 (1983). T. NAKAYAMAand M. TAKANO, Corrosion Nace 41,592 (1985). G. HERBSLEBand W. SCHWENCK,Corrosion Nace 41,431, (1985). A. DESESTRET,F. GAUTnEY,J. L. RANVlERandJ. C. COLSON,Mere. Sci. Rev. Metall. 6,393 (1977). H. BUHL, The Slow Strain Rate Technique, p. 333. Ugiansky and Payer Edit. (1979). T. MAGNINand P. RIEUX, Scripta Met. 21,907 (1987).