Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract
The low-cycle fatigue (LCF) behaviour of SUS304-HP austenitic stainless steel was investigated systematically using tension-compression
cycling under fully reversed total strain amplitude control conditions at room temperature in laboratory air. In addition to tests at constant strain
amplitudes, incremental step tests (IST) were also carried out. Cyclic stress response, during companion specimen tests (CST), revealed combinations
of a variable cyclic hardening, stable behaviour and softening, depending on the applied cyclic strain amplitude, while during incremental step tests
it exhibited cyclic hardening character at all strain levels. Microstructure observations using optical and transmission electron microscopy (TEM)
revealed that with increasing total strain amplitudes the slip band density increased and the dislocation structure changed from a planar array to
a more cellular-like structure. Cyclic deformation-induced austenite/martensite transformation was observed at higher cyclic strain amplitudes.
The change in microstructures during cycling is responsible for the fatigue hardening/softening behaviour of the material. The SEM micrographs
revealed that at low-strain amplitudes the inclusion-type nucleation occurred near the surface, while at the higher strain amplitudes crack initiation
characterized by cleavage cracking occurred not only near the surface but also in the interior of the specimen. Linear or single-slope behaviour
was seen both in cyclic stressstrain and Coffin-Mason plots. Masing cyclic stressstrain behaviour was presented only in the IST method but not
in the CST method.
2005 Elsevier B.V. All rights reserved.
1. Introduction Despite the fact that low-cycle fatigue behaviour of type 304
stainless steel has been investigated for many decades and there
SUS304-HP is an improved version of type 304 austenitic exists a considerable literature on this topic at present [14],
stainless steel through the addition of nitrogen (N) element in relative limited amount of information on LCF behaviour of
its composition for purpose of enhancing the corrosion resis- SUS304-HP stainless steel is available. In earlier studies [5,6] on
tance and mechanical properties. This alloy is currently being the influence of nitrogen on fatigue properties of lower alloyed
used in industrial installations, such as petrochemical plants, austenitic stainless steels, it has been indicated that nitrogen as
electric-power generating stations and process plants as pip- an alloy element is capable of enhancing cyclic softening at
ing and structural material. In these applications, the compo- low-strain amplitude but causing hardening at the higher strain
nents of the structures are often subjected to repeated thermal amplitudes. The addition of nitrogen increased the low-cycle
stresses as a result of temperature gradients, which occur on fatigue lifetime of austenitic stainless steels [7,8]. The pro-
heating and cooling during startups and shutdowns or during cess behind improved fatigue properties is considered to be an
variations in operating conditions. Therefore, resistance to low- increased planarity in the slip mode, suppressing cross slip and
cycle fatigue (LCF) is an essential requirement in the design of promoting slip reversibility [7]. On the other hand, nitrogen was
these structures and components against failure under dynamic also found to play an essential role in stability of the austenite.
loading. Addition of nitrogen as an interstitial element favours the for-
mation of stacking faults and the formation of martensite during
straining, which reduces crack growth and leads to rapid hard-
Corresponding author. ening, i.e. it increases the resistance to plastic flow [911]. It
E-mail address: duyi ye@zju.edu.cn (D. Ye). was also indicated [12] that the effect of martensite formed on
0921-5093/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2005.09.081
D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117 105
LCF life depended on the amount of martensite present, strain to determine the cycle stressstrain curve (CSSC). In this case,
amplitude, grain size and whether crack initiation or propagation a specimen was subjected to repeated strain blocks in which the
controlled fatigue life at a given strain amplitude. In one study strain amplitude increased from 0.2% linearly up to 1.2%, and
[13], the LCF properties deteriorated on martensite formation, then decreased. The IST was also run until the specimen failed.
owing to more crack initiating sites becoming available. From During testing, the load was continually monitored and hys-
earlier studies, it can thus be expected that the low-cycle fatigue teresis loops were recorded at appropriate intervals by means
behaviour of SUS304-HP stainless steel with a certain content of a personal computer with a data-collecting and analyzing
of nitrogen will be different from that of conventional type 304 software.
austenitic stainless steel without containing nitrogen content. The microstructures of the alloy after fatigue tests were
In this paper, low-cycle fatigue tests of SUS304-HP stain- examined by both optical microscopy (OM) and transmission
less steel were carried out at room temperature under total strain electron microscopy (TEM). Samples for microstructural
control using companion specimen test (CST) and incremen- analysis were prepared below the fractured surface in fatigue
tal step test (IST) methods, respectively. A detailed examina- failed specimens by cutting perpendicular to the tensile axils
tion of deformation microstructures and fracture features of the with a wire saw. The samples for OM were ground using emery
fatigue specimens cycled to failure at various strain amplitudes papers of various grinds from 230 to 600 grit, and then etched
was also performed. The low-cycle fatigue characteristics were for 30 s in a solution of 16 vol% of HNO3 , 32 vol% of HCl and
then discussed by means of competing and mutually interactive 50 vol% of glycerol at 313 K. Samples for TEM obtained from
influences of cyclic strain amplitude, concomitant mechani- thin slice with 0.3 mm thickness were mechanically thinned
cal response, intrinsic microstructural effects and dislocation- to 30 m, and then electropolished in a solution containing
microstructure interactions during cyclic straining. The main 5 vol% of perchloric acid and 55 vol% acetic acid, using a
purpose of the present work is to gain a more complete under- twin jet apparatus at a potential of 20 V and a temperature
standing of the low-cycle fatigue behaviour of SUS304-HP of 798 K. TEM examinations were performed on a HITACH
austenitic stainless steel, especially concerning the analytical H9000NA transmission electron microscope, operating
relationship describing its behaviour, so as to use it more effec- at 300 kV.
tively in practical structural designing against fatigue. The fractured surfaces of the fatigue failed specimens were
ultrasonically cleaned in trichloroethylene, and then observed
2. Experimental details using a low-magnification binocular microscope to locate rel-
evant feature, which were subsequently examined at higher
The material used for present investigation was SUS304-HP magnification with a JEOL JSM-T20 scanning electron micro-
austenitic stainless steel supplied in the form of a plate, 22 mm scope (SEM) to determine the predominant fracture mode and
in thickness. The plates were hot rolled at 1040 C for 0.2 h, to characterize the fine-scale topography of the fatigue fracture
followed by quenching in water. The nominal chemical compo- surface.
sition of the material in percentage weight is listed in Table 1.
Specimens used in the low-cycle fatigue tests were cylindri- 3. Results
cal, 6.0 mm in diameter and 14 mm in the gauge-length section.
Fully reversed, pushpull and total strain amplitude controlled 3.1. Initial microstructure and tensile properties
fatigue tests were performed at room temperature in an ambient
air using a closed-loop servohydraulic testing machine (Shi- The microstructure of the SUS304-HP stainless steel in the
madzu). A triangular strain waveform with zero mean strain as-heat treated conditions, as shown in Fig. 1, consisted of ran-
(R = 1) at a constant total strain rate of 5 103 S1 was domly oriented grains with a few annealing twins. Both the
used. The tests were continued until fracture, with the strain grain and twin boundary were covered with discrete carbide
amplitude limits lying between 0.30 and 2.0%. The test fre- particles. The heat-treatment conditions mentioned previously
quency (f) at a certain total strain amplitude could be defined by result in an equiaxed grain size, measured by the conven-
the following formula, f = /(4a ), where a is the total strain tional linear intercept method, of approximate 88 m (twins
amplitude. The incremental step test method was also employed not taken into account). Fig. 2 shows a transmission electron
microscopy micrograph of the material in the same condi-
Table 1
tion, which reveals that the dislocation structure in the ini-
Chemical composition (wt%) of SUS304-HP austenitic stainless steel tial condition consists of pinned dislocation lines and small
loops with low-dislocation density. Both optical microscopy
C 0.06
Si 0.45
and transmission electron microscopy observations indicate
Mn 0.81 that, in the undeformed condition, there was no metallo-
S 0.006 graphic evidence of austenite/martesitic transformation in the
P 0.029 material.
Cr 18.19 The tensile properties of the present material are summa-
Ni 8.64
N 0.052
rized in Table 2. The results reported are the mean values based
Fe Bal. on multiple (three) tests. The yield strength ( ys ) defined as
the stress corresponding to a plastic strain of 0.2% is 275 MPa.
106 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
Fig. 1. Optical micrograph of SUS304-HP stainless in the as-heat treated con- 3.2. Low-cycle fatigue behaviour
ditions.
Table 3
Cyclic stress response and fatigue life data of the SUS304-HP stainless steel
t /2 Values at N = 0.5Nf Nf
e /2 p /2 /2
To assess the effect of the type of loading on cyclic harden- 3.2.2. Cyclic stressstrain curve and Manson-Cofn plot
ing/softening behaviour of SUS304-HP stainless steel, the cyclic The results of low-cycle fatigue tests for the present material
response for the incremental step test method was determined, are reported in Table 3. The elastic strain amplitude (ea ), plastic
in which the stress amplitude responses at strain amplitudes of strain amplitude (pa ), total strain amplitude (ta ) and saturation
0.4, 0.6, 0.9 and 1.2% in each strain block are plotted as a func- stress amplitude ( a ) in this table are deduced from the hysteresis
tion of fraction of block (bn /bf ), as shown in Fig. 5. In this loops corresponding to half of the total number of cycles to
figure, the cyclic stress response determined by the companion fracture.
specimens test method, shown in Fig. 3, is also presented for Fig. 6 shows the cyclic stressstrain curves of the material
comparison purpose, where the stress amplitude during strain determined by the CST and IST methods, respectively, where the
cycling is plotted as a function of fraction of life (N/Nf ). It is values of stress and strain amplitudes for the IST method were
found from this figure that the cyclic stress response determined taken at half of the total number of strain blocks to fracture. As
by the above two methods exhibit quite different characteristic a comparison, the monotonic stressstrain curve (MSSC) of the
material is also presented in this figure. It is seen that both the
Fig. 5. Comparison of cyclic stress response curves determined by the CST and Fig. 6. Cyclic stressstrain curves (CSSCs) determined by the CST and the IST
the IST methods. methods and monotonic stressstrain curve (MSSC).
108 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
Fig. 8. Stressstrain hysteresis loops plotted with matched lower tips: (a) the CST method; (b) the IST method.
strain-hardening exponent (n ), The expression for calculating the plastic stain energy for a
non-Masing-type material was developed by Jhansale and Top-
n 1 per [25] and can be written as,
b , c (4)
1 + 5n 1 + 5n
1n 2n
These calculated values are listed in Table 5 and exhibit Wp = (p ) + (0 p ) (6)
1+n 1+n
agreement with those found from the test. This indicates
that, as a first approximation, the empirical relationship (4) is where n is the exponent of the equation of the skeleton curve,
applicable for the case of the SUS304-HP stainless steel. p = (/2E0 )1/n , and 0 represents the increase in the
proportional stress range due to non-Masing behaviour of the
3.2.3. Masing-type behaviour and cyclic plastic strain material.
energy From the tests data of SU304-HP stainless steel, the coef-
Tests were also carried out to determine whether SUS304- ficient n of Eq. (6), evaluated by a least squares technique, is
HP stainless steel exhibits Masing cyclic stressstrain behaviour 0.21. Note that the value of n is quite different from the cyclic
(or Masing-type behaviour) at room temperature. This prop- strain-hardening exponent n (see Table 4). The increase in the
erty is important for description of the hysteresis loop during proportional stress range, 0 , can be obtained from the cyclic
cycling. Fig. 8a and b shows the superimposed loops plotted stressstrain curve and is given by 0 = (/2) 60. The plas-
with matched lower tips for both the CST and IST methods. It is tic strain energy per cycle at half-life, determined by measuring
seen that the material exhibits Masing-type behaviour in the IST the area of the hysteresis loop and calculated from Eqs. (5) and
but does not follow an ideal Masing-type behaviour in the CST. (6), are reported in Table 6. It easily follows from this table that
According to Abdel-Raouf et al. [23], Masing-type behaviour the prediction precision of the plastic strain energy based on Eq.
could be determined using the Bauschinger strain (), defined as (6) has been improved greatly due to reckoning into the effect
the plastic strain in the reverse direction at 75% of the prestress in of non-Masing-type behaviour.
the forward direction. The Bauschinger strain () plotted against
plastic strain range (p ) is shown in Fig. 9. It follows from this
figure that for the case of IST the material exhibits Masing-type
behaviour, since Bauschinger strain () increased linearly with
cyclic plastic strain range (p ), while in the case of the CST
non-Masing-type behaviour could be inferred owing to the fact
that Bauschinger strain () increased with plastic strain range
(p ) in a parabolic manner after an initial linear region at low-
strain levels. The above two methods made the same predictions
of Masing-type behaviour for the material investigated.
It is known that for the material exhibiting a Masing-type
behaviour, the cyclic hysteresis energy per cycle Wp can be
calculated directly in terms of the cyclic stressstrain curve Eq.
(2), and can thus be expressed as [24],
1 n
Wp = p (5)
1 + n
where (=2 a ) is the total stress range and p (=ap ) the
plastic strain range. Fig. 9. Bauschinger strain vs. plastic strain range.
110 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
Fig. 11. Optical micrographes in the fractured specimen tested in LCF at a = 0.9% showing: (a) slip band features; (b) morphologies of martensite.
observations of the crack nucleation or formation region early growth region (Fig. 16d) revealed a crystallographic (stage
revealed the crack initiation to occur at inclusions near the sur- I) morphology of crack growth characterized by arrangements of
face (Fig. 16c). The inclusion-type nucleation can be understood wavy striations that have steps intersection at right angles. The
as cyclic slip localization due to stress-concentration at the inclu- arrangements of steps are comparable to brittle striations. The
sion, leading to either decohesion of the inclusionmatrix inter- overload region (Fig. 16e) comprised of microscopic voids of a
face or cracking of the inclusion [17]. Observations of the crack variety of sizes and shallow dimples indicating the highly ductile
Fig. 12. Optical micrographes in the fractured specimen tested in LCF at a = 2.0% showing: (a and b) slip band features; (c) different morphologies of martensite.
112 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
Fig. 13. TEM micrographs in the fractured specimen tested in LCF at a = 0.6% showing: (a) a planar array of dislocation structures; (b) stacking faults.
Fig. 15. TEM micrographs in the fractured specimen tested in LCF at a = 2.0% showing: (a) well-developed dislocation cell structure; (b) cellular structure penetrated
by individual striations with parallel orientation.
D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117 113
Fig. 16. (ae) Macroscopic fractograph and SEM micrographs of the fracture surface in the specimen tested in LCF at a = 0.4%.
4. Analysis and discussion [27], an increase in slip band density is responsible for the hard-
ening in austenitic steels. To verify this suggestion, specimens
The micromechanisms responsible for the cyclic stress tested at different total strain amplitudes were also examined
response during low-cycle fatigue are dependent on the for the present material. The number of slip bands per grain
microstructure and strain amplitude [17]. In the as-heat treated was divided by the grain diameter in a direction perpendicu-
conditions of SUS304-HP stainless steel, the initial microstruc- lar to the slip bands to estimate the average slip band spacing.
ture comprises pinned dislocation lines and loops with low- About 5075 grains were counted for each specimen. Fig. 19
dislocation density (Fig. 2). During cyclic straining, the unpin- shows the dependence of slip band spacing on imposed strain
ning and multiplication of dislocations as well as the mutual amplitudes, in which a decrease in slip band spacing with an
interactions of dislocations and the interactions of dislocations increase in the applied strain amplitude was observed. Thus,
with grain boundaries increases the resistance to plastic defor- consistent with earlier studies [27,28], it can be inferred that
mation, which is responsible for the observed initial hardening in an increase in the degree of cyclic strain hardening with strain
the cyclically deformed material. According to Lerch and Gerod amplitude for the present material, as shown in Fig. 4, can be
114 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
Fig. 17. (ae) Macroscopic fractograph and SEM micrographs of the fracture surface in the specimen tested in LCF at a = 0.9%.
related to the number of slip bands and slip systems activated in transformation [16,21]. Therefore, after initial cyclic hardening,
the cyclically deformed specimens. Since the material studied the stress response of the present material is governed by two
here is quasi-stable at room temperature due to addition of nitro- competitive processes, namely dynamic recovery (i.e. formation
gen, it underwent a deformation-induced austenitic/martensitic of cells and subgrains) and defomation-induced martenstic trans-
transformation during cyclic straining. Numerous investigations formation. The formation of cells and subgrains due to recovery
[913] have shown that the formation of martensites during mechanisms leads to an increase in the mean path for disloca-
cyclic loading strongly affects the fatigue mechanical behaviour tions, which favours cyclic softening [1], whereas the formation
of various types of steels. Magnetic measurements on the frac- of martensite by strain-induced processes drastically increases
ture surface of fractured specimens after the LCF deforma- the resistance to plastic flow that promotes cyclic hardening of
tion showed that the martensite content had increased with the the material [11]. According to TEM observations at low-strain
applied total strain amplitude [10]. The measurement of the vol- amplitudes, the saturation dislocation structures in the deformed
ume fraction of martensite during fatigue tests also documented specimen displayed a planar-array dislocation configuration
that a direct relation exists between hardening and martensitic (Fig. 13a). Both OM and TEM examinations revealed no marten-
D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117 115
Fig. 18. (af) Macroscopic fractograph and SEM micrographs of the fracture surface in the specimen tested in LCF at a = 1.2%.
site formation in austenitic grains tested at this strain amplitude. compensate for the softening effect associated with dislocation
This agrees with the findings of Baudy and Pineau [10] con- cellular structure, the material exhibits a saturation character as
cerning the critical total or plastic strain amplitude, which must has been observed in the cyclic stress response curves (Fig. 3),
be achieved in order to induce martensite formation in cyclic which is followed by secondary hardening as a result of the fact
straining. Consequently, at low-strain amplitude, the progres- that the volume fraction of strain-induced martensite transfor-
sive softening process after initial cyclic hardening is attributed mation further increased with the increasing number of straining
to the formation of a dislocation configuration that bears the plas- cycles. At high-strain amplitudes, the cumulation of the cyclic
tic strain more easily [17]. At the intermediate strain amplitude plastic strain leads to a continuous increase in the density of shear
austenite/martensite transformation takes place and the amount bands, and thus in the density of sites for strain-induced nucle-
of this transformation increases with an increasing number of ation of martensite (see Figs. 12c and 15b). In other words, more
cycles, as shown in Fig. 11b. Meanwhile, the tendency to form a martensite is formed with a less number of cycles in this case. As
dislocation substructure characterized by a developing cell struc- a result, the material displayed continuous strain hardening up
ture containing low-dislocation density/ill-defined cell walls to failure without displaying an intermediate stage of stress sat-
(Fig. 14) suggests localization of deformation in the material, uration at the higher strain amplitudes (see Fig. 3). On the other
which results in less stress for sustaining a given strain ampli- hand, the formation of more martensite also leads to more crack
tude, namely softening. Thus, when an increase of the resistance initiating sites in the fatigued specimen at high-strain amplitude,
to plastic deformation resulting from formation of martensite can which has been exhibited in the fatigue fracture (see Fig. 18ad).
116 D. Ye et al. / Materials Science and Engineering A 415 (2006) 104117
5. Conclusions
[2] K. Bhanu Sankara Rao, M. Valsan, R. Sandhya, S.L. Mannan, P. [17] M. Klesnil, P. Lukas, Fatigue of Metallic Materials, Elsevier, Amster-
Rodrignez, Metall. Trans. A 24 (1993) 913. dam/Oxford/New York, 1980.
[3] M.R. Bayoumi, A.K. Abd Ellatif, J. Mater. Sci. 30 (1995) 3944. [18] R.W. Landgraf, Am. Soc. Test. Mater. Spec. Tech. Publ. 467 (1970) 3.
[4] G.C. Kaschner, J.C. Gibeling, Mater. Sci. Eng. A336 (2002) 170. [19] B. Tomkins, Philos. Magn. 18 (1968) 1041.
[5] M.L.G. Byrnes, M. Grujicic, W.S. Owen, Acta Metall. 35 (1987) 1853. [20] T.S. Srivatsan, R. Auradkar, Int. J. Fatigue 14 (1992) 355.
[6] Y. Li, C. Laid, Mater. Sci. Eng. A186 (1994) 87. [21] D. Hennessy, G. Steckel, C. Altstetter, Metall. Trans. 7A (1976) 415.
[7] J.-O. Nilsson, Scr. Metall. 17 (1983) 593. [22] J.D. Morrow, F.R. Tuler, J. Basic Eng. 87 (1965) 275.
[8] J.B. Vogt, S. Degallaix, J. Foct, Int. J. Fatigue 6 (1984) 211. [23] H. Abdel-Raouf, T.H. Topper, A. Plumtree, Proceeding of the Fourth
[9] L.D. Murr, K.P. Stand Hammer, S.S. Hecker, Metall. Trans. A13 (1982) International Conference on Fracture, Waterloo, 1977, p. 1107.
627. [24] D. Lefebvre, F. Ellyin, Int. J. Fatigue 6 (1984) 9.
[10] G. Baudry, A. Pineau, Mater. Sci. Eng. 28 (1977) 229. [25] H.R. Jhansale, T.H. Topper, ASTM STP 519, American Society of Test-
[11] M. Bayerlein, H.J. Christ, H. Mughrabi, Mater. Sci. Eng. A114 (1989) ing and Materials, 1973, p. 246.
L11. [26] N.Y. Jin, C. Zhong, X. Chen, Acta Metall. Mater. 38 (1990) 2141.
[12] S. Ganesh Sundara Raman, K.A. Padmanabhan, Int. J. Fatigue 17 (1995) [27] B.A. Lerch, A. Gerold, Acta Metall. 33 (1985) 1709.
271. [28] S. Ganesh Sundara Raman, K.A. Padmanabhan, Fatigue 16 (1994) 209.
[13] G.R. Chanani, S.D. Antolovich, Metall. Trans 5 (1974) 217. [29] E. Mousavi-Torshizi, S. Degallaix, Proceeding of the Fourth International
[14] T.S. Srivatsan, S. Anand, J.D. Troxell, Int. J. Fatigue 15 (1993) 355. Conference on Fracture, Waterloo, 1977, p. 75.
[15] R.P. Skelton, High Temperature Fatigue (Properties and Prediction), [30] M. Nystrom, U. Lindstedt, B. Karlsson, J.-O. Nilsson, Mater. Sci. Tech-
Elsevier, London/New York, 1987. nol. 13 (1997) 560.
[16] M. Botshekan, S. Degallaix, Y. Desplanques, J. Polak, Fatigue Fract. [31] C. Zhong, N.-Y. Jin, X. Zhou, E. Meng, X. Chen, Acta Metall. Mater.
Eng. Mater. Struct. 21 (1998) 651. 38 (1990) 2135.