Available online at www.sciencedirect.com
Procedia Engineering 10 (2011) 1176–1183
ICM11
Cyclic deformation behavior and low cycle fatigue life of normalized medium carbon steel with hydrogen charging Y. Tsuchidaa*, T. Watanabeb, G. Suzukib, and H. Yanoa a1 Department of mechanical engineering, Daido university, Takiharu-cho 10-3, Nagoya 457-8530, Japan bGraduate school of Daido University, Takiharu-cho 10-3, Nagoya 457-8530, Japan
Abstract The effects of hydrogen on low cycle fatigue properties have been fundamentally examined using normalized JIS S45C steel. With the aid of hydrogen thermal desorption analysis, the influence of the increase in carbon content has been discussed. The S45C steel easily pick-up hydrogen. But it is relatively insensitive to hydrogen absorption. This increased amount of hydrogen pick-up is firstly caused by large amount of pearlite structure. Further sulfur takes an important role for hydrogen pick up during cathodic charging. Hydrogen plays two kinds of roles during fatigue tests. One is hydrogen in dislocation core and makes plastic deformation easier in the first monotonic tension stage. The other is hydrogen/vacancy pair, and suppresses dislocation motion and increases stress amplitudes. The latter should be the fundamental issue for fatigue life. The amount of absorbed hydrogen itself is not only important, but the difference of hydrogen concentration between surface and inside is more important. When the fatigue crack initiates inside, the fatigue life is severely damaged to less than 10% of non-charged one. Taking the lower limit value of hydrogen to form internal fish-eye type cracking, S45C is less sensitive to hydrogen in cyclic straining. This comes from the pearlite structure that resists crack propagation.
© 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license. Selection and peer-review under responsibility of ICM11 Strain controlled low cycle fatigue; Hydrogen; Fatigue life; Pearlite; Ferrite; vacancy; dislocation;
1. Introduction Steels suffer from degradations by hydrogen in mechanical properties. This is known as hydrogen embrittlement. It is evaluated under static loading or slow strain rate tension testing[1]. The higher the
* Corresponding author. Tel.: +81-52-612-6111; fax: +81-52-612-5623. E-mail address:
[email protected].
1877-7058 © 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license. Selection and peer-review under responsibility of ICM11 doi:10.1016/j.proeng.2011.04.196
1177
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
Table 1 S45C S10C
Chemical composition of the S45C used in the present research C Si Mn P S Cu Ni 0.46 0.21 0.74 0.024 0.023 0.10 0.04 0.09 0.18 0.31 0.010 0.010 0.10 0.05
Cr 0.15 0.20
Mo 0.01 0.03
䟺mass%䟻 Al N 0.004 0.0127 0.015 0.0128
strength is, the higher is the susceptibility to hydrogen embrittlement. Unfortunately, time dependent straining conditions have not attracted attention so much. We have explored under strain-controlled low cycle fatigue conditions on normalized low carbon steel [2-3], which is the most basic type of steel products. This steel is unexpectedly strongly susceptible to hydrogen and its fatigue lives become less than one tenth of non hydrogen charged condition. Efforts have been continued to clarify the mechanism underlying the strong susceptibility to hydrogen. This report is concerning medium carbon steel, and describes the influence of hydrogen charging on cyclic stress-strain behavior and fatigue life. 2. Experimental procedure Commercial JIS S45C hot rolled bars are subjected to normalizing treatment at 1123K. Its chemical composition and microstructure are shown in Table 1, Fig. 1 and Fig.2. The microstructure is composed of almost equal fraction of ferrite phase and pearlite structure. Round bar specimens measuring 5mm in diameter and 50mm in length were machined from the center of the normalized bars. They were hydrogen charged in the 0.1N sulfuric acid aqueous solution containing 0.06mass% ammonium thiocyanate. The hydrogen pick-up is determined by Thermal Desorption Analysis (TDA) after reserving in liquid nitrogen within 1 day. Round bar fatigue test specimens with parallel portion of 13mm in diameter and 20mm in length were machined from the normalized S45C bars. They were hydrogen charged in the above mentioned solution for 2 weeks, and reserved within 1 day in liquid nitrogen before fatigue tests. These specimens were attached to the servo-hydraulic fatigue testing machine as quick as possible. Strain controlled low cycle fatigue tests were conducted. The leading time before tests were within 7 minutes including attaching the strain gauge. The details of fatigue test conditions are gauge length of 12.5mm, strain amplitude of 0.0045, strain ratio of -1, and strain rate of 0.001/s. The fracture surfaces were examined using scanning electron microscope (SEM).
40Pm
Fig.1 Microphotograph of S45C normalized from 1123K
Pm
Fig.2 SEM micrograph of lamellar pearlite
1178
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
㻙 㻖㻖㻖㻮
㻖㻙㻖㻮
㻔㻑㻚㻤㻒㼐㻕 㻖㻑㻖㻤㻒㼐㻕 㻕㻓㻤㻒㼐㻕
㻓㻑㻓㻛
㻨㼙㼒㼏㼙㼈㼇㻃㼋㼜㼇㼕㼒㼊㼈㼑㻃㻋㼐㼄㼖㼖㻐㼓㼓㼐㻌
㻨㼙㼒㼏㼘㼗㼌㼒㼑㻃㼕㼄㼗㼈㻃㻋㼐㼄㼖㼖㻐㼓㼓㼐㻒㼐㼌㼑㻌
㻓㻑㻔㻓
㻶㻗㻘㻦 㻓㻑㻓㻙
㻓㻑㻓㻗
㻓㻑㻓㻕 㻶㻔㻓㻦 㻕㻓㻤㻒㼐㻕
㻓㻑㻓㻓 㻖㻓㻓
㻘 㻶㻗㻘㻦 㻗
㻖
㻕 㻶㻔㻓㻦 㻔 㻔
㻖㻘㻓
㻗㻓㻓
㻗㻘㻓
㻔㻓
㻘㻓㻓
㻷㼈㼐㼓㼈㼕㼄㼗㼘㼕㼈㻃㻋㻮㻌
Fig.3 Hydrogen desorption spectra of specimens hydrogen-charged at various current densities
㻔㻓㻓
㻦㼘㼕㼕㼈㼑㼗㻃㼇㼈㼖㼌㼗㼜㻃㻋㻤㻒㼐 㻕 㻌
Fig.4 Variation of hydrogen content by current density measured by desorption spectrum analysis
3. Experimental results 3.1. Thermal desorption measurement The S45C specimens are hydrogen charged with various current densities. Then state and amount of hydrogen are measured by TDA. The results are shown in Fig.3 with the results of JIS S10C steel. The desorption peak is found around 330K, and the spectra blow out around 373K. These are same with those of S10C. The total desorption up to 450K are plotted against current densities in Fig.4. It seems to increase with the logarithm of current density. The total desorption is about three times larger in S45C than S10C at a given current density. Hydrogen is reserved much more in the present S45C than S10C in the normalized conditions. 3.2. Deformation behavior
㻨 㻠 㻕㻓㻛㻪㻳㼄
㻗㻓㻓 㻶㼗㼕㼈㼖㼖㻃㻋㻰㻳㼄㻌
Figure 5 compares the stress strain response in the monotonic tension stage. For non charged specimen, the relation is straight until so-called upper yield stress (UYS) of tensile test; the elastic modulus is close to 208GPa that is reasonable for ferritic steels. Hydrogen charging makes this straight relationship to curved below the UYS. The stress, where plastic deformation begins, becomes around 300MPa. The upper yield stress seems hardly affected, but lower yield stress (LYS) tends to decrease by increased cathodic current density. Maximum stresses of the succeeding tension/ compression cycles are plotted against number of cycles in Fig.6. For non-charged S45C, the maximum stress decreases continuously from the 1st cycle and stays constant from 10th-20th cycles to 500th cycle; then increases again to fracture. On the other hand, the stress amplitude of
㻙㻓㻓
㻕㻓㻓 㻱㼒㼑㻐㼆㼋㼄㼕㼊㼈㼇 㻔㻑㻚㻤㻒㼐㻕 㻖㻑㻖㻤㻒㼐㻕 㻓 㻓
㻓㻑㻓㻓㻕
㻓㻑㻓㻓㻗
㻶㼗㼕㼄㼌㼑
Fig.5 Comparison of first tensile cycles of noncharged and hydrogen charged specimens
1179
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
㻓㻑㻔
㻗㻘㻓 㻲㼓㼈㼑㻃㻝㻃㼑㼒㼑㻐㼆㼋㼄㼕㼊㼈㼇㻏 㻦㼏㼒㼖㼈㼇㻃㻝㻃㻫㼜㼇㼕㼒㼊㼈㼑㻃㼆㼋㼄㼕㼊㼈㼇
㻶㻗㻘㻦
㻱㻵㻬㻰㻃㻶㻗㻘㻦 㻶㻗㻘㻦 㻶㻔㻓㻦
㻶㼗㼕㼄㼌㼑㻃㼄㼐㼓㼏㼌㼗㼘㼇㼈
㻶㼗㼕㼈㼖㼖㻃㼄㼐㼓㼏㼌㼗㼘㼇㼈㻃㻋㻰㻳㼄㻌
㻗㻓㻓
㻖㻘㻓
㻖㻓㻓
㻓㻑㻓㻔
㻩㼌㼖㼋㻐㼈㼜㼈
㻓㻑㻓㻓㻔 㻕㻘㻓
㻫㼜㼇㼕㼒㼊㼈㼑 㼆㼋㼄㼕㼊㼈㼇
㻶㻔㻓㻦
㻕㻓㻓
㻓㻑㻓㻓㻓㻔 㻔
㻔㻓
㻔㻓㻓
㻔㻓㻓㻓
㻔㻓㻓㻓㻓
㻔㻓㻓㻓㻓㻓
㻦㼜㼆㼏㼈㼖
Fig.6 Comparison of stress amplitude between S45C and S10C with and without hydrogen charging; cathodic charging current is 1.7 and 3.3 A/m2 for S45C, and 20 and 50A/m2 for S10C
㻔㻓
㻔㻓㻓
㻔㻓㻓㻓
㻔㻓㻓㻓㻓
㻔㻓㻓㻓㻓㻓
㻔㻓㻓㻓㻓㻓㻓
㻱㼘㼐㼅㼈㼕㻃㼒㼉㻃㼆㼜㼆㼏㼈㼖㻃㼗㼒㻃㼉㼄㼏㼘㼕㼈
Fig.7 Coffin-Manson plot for S45C and S10C with and without hydrogen charge
hydrogen charged specimens is lower than the non-charged specimens up to 10 cycles; this corresponds to the decreased LYS in Fig.5. The maximum stress decreases quite slowly with number of cycles. It is 1520MPa higher than the non-charged specimen after 10 cycles. It is same with the case of S10C, which is shown in the lower part of Fig.6. 3.3. Fatigue life Figure 7 illustrates the relation between plastic strain amplitude and number of reversals, so called Coffin-Manson relationship. The results of S45C are plotted together with those of S10C. The fatigue lives of hydrogen charged specimens are about 50% of non-charged one. When the cathodic current density is increased to 20A/m2, the fatigue life reduces to less than 10%, revealing fish-eye type fracture surface that is common with the cases of S10C.. 3.4. Fracture surface The fracture surfaces are examined by SEM and are shown in Fig.8. In the case of non-charged specimen in Fig.8(a) and (d), the fatigue crack initiates at specimen surface and propagates accompanying striation. When hydrogen is charged by current density of 1.7A/m2, fatigue crack also initiates at specimen surface, but it propagates inside with sharp quasi-cleavage facet, as shown in Fig.8(b) and 8(e). With further high current density of 20A/m2, the crack initiation site moves inside and propagates radically, as shown in Fig.8(c). The facet inside of this fish-eye fracture surface is also sharp and quasicleavage one. 4. Discussion 4.1. Amount of hydrogen desorption
1180
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
(a)
(b)
(c)
(d)
(e)
(f)
1mm
10Pm
Fig.8
2
SEM images of fracture surface. (a) & (d): without hydrogen charging, (b) & (e): 1.7A/m , and (c) & 20A/m2
The S45C steel bars are also given spheroidizing heat treatment, which is featured by heating at 1033K for 4 hours followed by slow cooling at 10K/h until 873K. The carbide morphologies are shown in Fig.9. The lamellar cementite before spheroidizing heat treatment changes to globular ones measuring 1-2 m. This specimen is hydrogen charged with cathodic current density of 3.3A/m2 and TDA measurement is performed. The results are compared in Fig.10 with normalized S45C and S10C. Here the current density for S10C is as high as 20A/m2. The sub-peak or blow-out at 360K is correlated in the previous paper[4] with hydrogen trap at cementite/ferrite interface. The evolution rate at this temperature drops to about half by applying spheroidizing of cementite; fairly high fraction of pearlite structure should be the reason for the extraordinary amount of hydrogen pick-up. The evolution peaks exist at 330K in Fig.8. This peak is attributed to the hydrogen in lattice or temporarily trapped by dislocation. This peak height is still higher in spheroidized S45C than normalized S10C, not withstanding the small ferrite/pearlite interface areas in both steels. Another factor may be affecting the hydrogen pick-up. Looking at the chemical composition in Table 1, relatively high sulfur content of S45C takes our notice. Sulfur, which is in solid-solution or forming MnS nonmetallic inclusion, can produce H2S in the electrolytic solution. This chemical agent is known as catalytic poison that enhances hydrogen pick-up. 4.2. Effect of hydrogen on deformation Hydrogen plays two kinds of roles in the present strain controlled low cycle fatigue studies. One is softening in the early stage of monotonic tension straining in Fig.5. The other is hardening of stress amplitude after 10th cycle in Fig.6. There exist dislocations in the ferrite phase in normalized state of S45C. They are immobilized by the Cottrell atmosphere of carbon and nitrogen atoms, up to upper yield stress (UYS). When hydrogen enters, it is known to react with dislocation core. This makes the Peierls stress lowered, and dislocation can move
1181
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
at smaller stress. This might be the reason why the plastic deformation start stress becomes lowered. The Cottrell atmosphere being still working after small plastic deformation, the UYS is recognized and followed by stress drop to the lower yield stress (LYS). This LYS is much lower in specimens that are hydrogen charged by higher current density. The result in S10C is shown in the lower part of Fig.6. The stress amplitude of hydrogen charged specimens are higher than that of non-charged one. It has been reported in the previous paper[4]. The cyclic tension/compression straining enhances annihilation of dislocation dipoles of opposite sign. This dislocation reaction results in the formation of vacancies. Then the vacancies are coupled with hydrogen atoms. This hydrogen/vacancy pair is ascribed to the strengthening of stress amplitude in S10C. This mechanism should be applied to the strengthened stress amplitude of S45C that is shown in the upper part of Fig.6. Before that cycle, the softening by hydrogen in dislocation core is overcoming the hardening by hydrogen/vacancy pair. The formation of vacancy/hydrogen pair is effective in stabilizing vacancy and forming more complex vacancy clusters. This should be the essential issues for crack initiation and propagation. 4.3. Reasons for reduction of fatigue life Figure 7 tells that crack initiation site has a critical role in fatigue life reduction by hydrogen charging. When the fatigue crack initiates at specimen surface, the reduction may be smaller even if the absorbed hydrogen is much more. The tendencies of initiation at surface and inside are compared in Fig.11. It is inevitable that there is surface asperity brought about by machining, polishing, or corrosion pit in extreme case. So tendency for cracking at specimen surface is higher (b) than inside, in non-charged base specimen. As the hydrogen (a) diffuses out from surface, hydrogen content at surface is always smaller than inside. This causes cracking tendency inside is increased more. The overall tendency inside becomes higher, as shown in shaded arrow in Fig.11. Crack initiation is not decided by the total amount of hydrogen but cross-sectional distribution. Normalized S45C contains large fraction of pearlite structure, which acts as temporally hydrogen trap. This leads to the suppressed diffusion rate of hydrogen, and surface hydrogen concentration decreases slowly. As a result, surface crack initiation tendency is maintained.
㻓㻑㻔㻓
㻨㼙㼒㼏㼘㼗㼌㼒㼑㻃㼕㼄㼗㼈㻃㻋㼐㼄㼖㼖㻐㼓㼓㼐㻒㼐㼌㼑㻌
㻖㻖㻖㻮
㻖㻙㻖㻮
㻱㼒㼕㼐㼄㻃㻖㻑㻖㻤㻒㼐㻕 㻶㻤㻃㻖㻑㻖㻤㻒㼐㻕
㻓㻑㻓㻛
㻓㻑㻓㻙
㻓㻑㻓㻗
㻶㻗㻘㻦
㻓㻑㻓㻕 㻶㻔㻓㻦 㻕㻓㻤㻒㼐㻕 㻓㻑㻓㻓 㻖㻓㻓
2Pm
Fig.9 SEM microphotograph of spheroidized S45C steel
㻖㻘㻓
㻗㻓㻓
㻗㻘㻓
㻘㻓㻓
㻷㼈㼐㼓㼈㼕㼄㼗㼘㼕㼈㻃㻋㻮㻌
Fig.10 Hydrogen desorption spectra of normalized S45C and S10C in comparison with spheroidized S45C specimens
1182
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
㻶㼘㼕㼉㼄㼆㼈
㻔㻓㻓
㻶㼘㼕㼉㼄㼆㼈㻃
㻫㼜㼇㼕㼒㼊㼈㼑 㼅㼄㼖㼈 㻷㼈㼑㼇㼈㼑㼆㼜㻃㼒㼉㻃㼆㼕㼄㼆㼎㻃㼌㼑㼌㼗㼌㼄㼗㼌㼒㼑
㻤㼌㼕
㻫㼜㼇㼕㼒㼊㼈㼑㻃 㼊㼄㼖㻃
㻤㼌㼕
㻋㼄㻌㻃㻶㼘㼕㼉㼄㼆㼈㻃㼆㼕㼄㼆㼎
㻋㼅㻌㻃㻬㼑㼗㼈㼕㼑㼄㼏㻃㼆㼕㼄㼆㼎㻃
㻓 㻶㼘㼕㼉㼄㼆㼈
㻬㼑㼖㼌㼇㼈
Fig.11 Effect of hydrogen on crack initiation tendency for initiation at surface and inside
Fig.12 Influence of crack initiation site on crack propagation
Figure 12 illustrates the different situation of crack propagation for cracks that are open to air or closed. In the case of surface crack initiation, the crack tip is exposed to air where the hydrogen partial pressure is quite low. In the case of internal cracking, the situation is quite different. The hydrogen is confined in the small area as hydrogen gas and the crack tip is exposed this high pressure hydrogen gas atmosphere. The hydrogen concentration at crack tip is kept high, and the crack propagation happen easily accompanying quasi-cleavage facet. Effect of hydrogen charging is straightly evaluated by the lower limit value of hydrogen concentration, where internal cracking is formed. They are 3.5mass-ppm for S45C, and 1.5mass-ppm for S10C[5]. S45C is more resistant to hydrogen than S10C. The fracture surface is chemically etched by nitral solution. Figure 13 shows the fracture surface near the origin inside of the fish-eye in high magnification. Crack is propagation penetrating lamellar cementite. As cementite is not affected by hydrogen, lamellar pearlite is useful for retarding crack propagation. 5. Summary The effects of hydrogen on low cycle fatigue properties have been fundamentally examined using normalized JIS S45C steel. With the aid of hydrogen thermal desorption analysis, the influence of the increase in carbon content has been discussed. Major results obtained are as follows. 1) The S45C steel easily pick-up hydrogen. But it is relatively insensitive to hydrogen absorption. 2) This increased amount of hydrogen pick-up is firstly caused by large amount of pearlite structure. Further sulfur takes an important role for hydrogen pick up during cathodic charging. 3) Hydrogen in normalized S45C plays different roles during fatigue tests. At the first monotonic tension stage, hydrogen in dislocation core makes plastic deformation easier. After 10th cycle, formation of
5Pm
Fig.13 SEM images of fracture surface of 20A/m2 specimen after nital etching
Y. Tsuchida et al. / Procedia Engineering 10 (2011) 1176–1183
hydrogen/vacancy pair suppresses dislocation motion and increases stress amplitudes. 4) The amount of absorbed hydrogen is not only important, but the difference of hydrogen concentration between surface and inside is more important. 5) When the fatigue crack initiates inside, the fatigue life is severely damaged to less than 10% of noncharged one. Suppressing internal cracking is important for enhancing resistance to hydrogen. 6) Taking the lower limit value of hydrogen to form internal fish-eye type cracking, S45C is less sensitive to hydrogen in cyclic straining. This comes from the pearlite structure that resists crack propagation. References [1] [2] [3] [4] [5]
Y. Hagiwara, C, Ito, N. Hisamori, H. Suzuki. K. Takai, and E. Akiyama, Tetsu-to-Hagane, 94(2008), pp215-221 Murakami Y, Intern. J. of Fatigue, 28(2006), p1509-1520 M. Nagumo, H. Shimura, T. Chaya, H. Hayashi, and I. Ochiai, Materials Science and Eng. A348(2003), pp192-200 T. Yamashita, K. Honda, F. Fujisawa and Y. Tsuchida, J. High Preessure Institute Japan, , Vol. 48(2010), pp11-18. T. Yamashita, T. Ochiai, M. Iimi, and Y. Tsuchida, J. High Preessure Institute Japan, , Vol. 49(2011), in printing.
1183
