Thin Solid Films 355±356 (1999) 487±493 www.elsevier.com/locate/tsf

Fatigue behavior of a quenched and tempered AISI 4340 steel coated with an electroless Ni-P deposit Y. GarceÂs a, H. SaÂnchez a, J. BerrõÂos b, A. Pertuz a, J. Chitty c, H. Hintermann d, E.S. Puchi b,* a

School of Mechanical Engineering, Faculty of Engineering, Central University of Venezuela, Apartado Postal 47885, Los Chaguaramos, Caracas 1045, Venezuela b School of Metallurgical Engineering and Materials Science, Faculty of Engineering, Central University of Venezuela, Apartado Postal 47885, Los Chaguaramos, Caracas 1045, Venezuela c Department of Applied Mathematics, Faculty of Engineering, Central University of Venezuela, Apartado Postal 47885, Los Chaguaramos, Caracas 1045, Venezuela d Faculty of Sciences, University of NeuchaÃtel, NeuchaÃtel, Switzerland

Abstract The fatigue life of a quenched and tempered AISI 4340 steel has been evaluated in three different conditions: (a) uncoated, (b) coated with an electroless Ni-P (EN) deposit of a P content of approximately 12±14wt.%, as-deposited and (c) as-deposited, followed by a two-step postheat treatment (PHT): 473 K for 1 h plus 673 K for 1 h. The results indicate that plating the base steel with this kind of deposit leads to a signi®cant reduction of the fatigue life of the material, particularly if the deposit is subjected to a subsequent PHT. Such a reduction has been quanti®ed by determining the Basquin parameters from the fatigue life curves obtained for the uncoated, coated, coated and PHT substrates. It has been shown that the fatigue life of the base steel can be reduced by 78% in the as-deposited condition and a 92% after a subsequent PHT. The microscopic observation of the fracture surfaces of the samples indicate that the fatigue process is initiated at the surface of the deposit and, subsequently, transferred to the substrate, with the assistance of the metallic bonding established at the deposit-substrate interface. This belief is supported by the observation of some continuity of the fracture features between the coating and the substrate under low alternating stresses. In the present study, the bonding between the EN deposit and the base steel was observed to be rather poor. Extensive secondary cracking along the coating-substrate interface after fatigue testing as well as the complete separation of the deposit from the substrate during tensile testing support this view. Such a behavior is believed to be related to the signi®cant difference that exists between the elastic and plastic properties of the EN deposit and the base steel. Nonetheless, the slight degree of metallic bonding that remains after the ®rst stage of fatigue testing seems to be enough to allow the passage of the fatigue cracks, prior nucleated in the deposit, into the substrate. It is therefore concluded that, in the present case, the EN deposit acts as a surface crack source or surface notch which decreases the fatigue life of the coated material by reducing the crack nucleation stage. q 1999 Published by Elsevier Science Ltd. All rights reserved. Keywords: Fatigue behavior; Plain carbon steels; Electroless Ni-P deposits; Fatigue cracks

1. Introduction The fatigue and corrosion-fatigue behavior induced by EN deposits on plain carbon steels of different carbon content have been extensively investigated in previous studies [1±7]. For example, Riedel [1] has reported an increase in the fatigue life of two St 52 and AISI 1055 steels when these have been coated with an EN deposit of 12% P and subsequently post-heat treated (PHT). Izumi et al. [2] have also reported an increase of about 20% in the fatigue life of medium strength steels (UTS of approximately 440± 750 MPa) when the EN coating is in the as-deposited condition. However, after a PHT these authors have reported a * Corresponding author. Tel.: 158-2-662-8927; fax: 158-2-662-8927. E-mail address: [email protected] (E.S. Puchi)

decrease which has been attributed to the precipitation of Ni3P particles. Puchi et al. [3] have also reported an increase in fatigue life of both AISI 1010 and 1045 steels which is more marked as the mechanical strength of the substrate material decreases. Therefore, those steels with tensile strengths of the order of 250±440 MPa could experience an increase in fatigue life when coated with EN deposits, depending upon the coating thickness and the predominant residual stresses in the deposit itself. BerrõÂos et al. [5] have investigated the effect of the coating thickness of an EN deposit on the fatigue behavior of an annealed AISI 1045 steel. In this study, 7±37 mm-thick EN coatings were deposited and PHT at two different temperatures. The only coatings that behaved similarly to the uncoated

0040-6090/99/$ - see front matter q 1999 Published by Elsevier Science Ltd. All rights reserved. PII: S 0040-609 0(99)00673-2

488

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493

substrate was the 7 mm deposit PHT at 473 K for 1 h, whereas all the other samples displayed a reduction in fatigue life. Particularly, for thicknesses ranging between 17± 37 mm it has been suggested that tensile residual stresses in the coatings contribute to the observed fatigue behavior and that such residual stresses could be associated with a relatively low P content, crystallization of amorphous Ni and precipitation of Ni3P particles. According to Riedel [1], the P content of the deposit depends of the pH of the solution employed as source of Ni ions. More speci®cally, it depends on the molar ratio Ni 21:(H2PO2) 2, since as the concentration of hypophosphite increases, the pH decreases and the P increases. For a pH less than 5, the EN deposit could have a P content of 10% or higher. The studies of Parker and Shah [8] indicate that if the P content of the deposit is greater than approximately 11± 12%, the residual stresses would be of a compressive nature. In relation to high strength steels employed as substrate, Wu and coworkers [9] have conducted an investigation on the fatigue resistance of a 30CrMo steel (0.30 C, 1.09 Cr and 0.24 Mo) oil quenched from 1143 K and tempered at 893 K for 3 h. The source of Ni ions was NiSO4 with a pH of 4.5. The deposit was PHT a 473 K for 1.5 h. These authors found a reduction in the fatigue limit of approximately 39% for the plated substrate and a reduction of 20% when the substrate was previously shot peened before plating. It was also reported that the fatigue cracks initiated at the interface between the coating and the substrate, and that in the deposit some of the cracks were parallel to the stress axis. The low fatigue strength of the coating was found to be responsible for the decrease in the fatigue limit of the plated steel. Zhang et al. [10] have also carried out three-point bending fatigue tests on a 30CrMo steel coated with an EN deposit of 43 mm thickness and 9.5 wt.% P. In this investigation some of the samples were shot peened before plating and some of the deposited specimens were PHT at 200, 400 and 6008C. The residual stresses in the coatings were determined by means of the bent strip method and it was observed that for all the conditions investigated such stresses remained compressive after annealing, but decreased with increasing annealing temperature. Also, shot peening before plating was observed to increase the compressive residual stress within the coatings. Regarding the in¯uence of EN deposits on the fatigue limit of the material, it was determined that such coatings reduced this property in comparison with the unplated substrate. The decrease in fatigue strength was observed to be less marked for the shot peened specimens but became signi®cantly higher as the PHT temperature increased. In relation to the fractographic analysis of the plated samples, it was reported that without the application of shot peening, the fatigue cracks initiated at the surface of the specimens, leading to the fatigue failure of the coating. On the contrary, when the samples were shot peened previously to the coating deposition, the crack initiation sites were displaced to

the coating±substrate interface. The work conducted by Zhang et al. [10] allowed the conclusion that the fatigue properties of this material, when it is coated with EN deposits, depends primarily on the fatigue resistance of the coating itself. Thus, the present investigation has been conducted in order to study the fatigue behavior, above the fatigue limit, of an AISI 4340 steel which has been oil quenched and tempered prior to plating at industrial scale with an EN deposit of 24 mm in thickness and a P content ranging between 12 and 14%. 2. Experimental techniques The present investigation has been carried out with samples of an AISI 4340 steel with the following composition (%wt): 0.34 C, 0.50 Mn, 0.30 Si, 1.5 Cr, 0.20 Mo and 1.50 Ni. This alloy is widely employed in the manufacture of automotive crankshafts and rear axle shafts, aircraft crankshafts, connecting rods, propeller hubs, gears, drive shafts, landing gear parts and heavy duty parts of rock drills. The material was provided as bars of approximately 16 mm diameter and 6 m length. Such bars were cut to pieces of approximately 120 mm length for machining tensile specimens and of 90 mm length for machining the fatigue samples. Both type of specimens had a gage diameter of 6.35 mm, gage length of 12.7 mm, ®llet radius of 25.4 mm and shoulder diameter of 12.7 mm, according to the ASTM standard E 606. The alloy was already provided in the quenched and tempered condition. The specimens were subsequently ground with successive SiC papers grit 600±1200 and polished mechanically. Fifty-six of these samples were degreased in a 5% HCl solution at 348±353 K for 7 min, rinsed again in distilled water, rinsed in a sodium bicarbonate (100 g per liter) solution and rinsed in water. The deposition was conducted industrially employing a bath composed of 30 g/l nickel sulphate, 30 g/l sodium hypophosphite, 35 g/l malic acid, 1.5 ppm lead sulphate, 10 g/l succinic acid and a stabilizer. During deposition the pH was maintained at approximately 5, at a mean temperature of about 358 K. The deposition rate was of approximately 12 mm/h and the process was conducted for 2 h which allowed a thickness of about 24 mm to be achieved. Such thicknesses were corroborated by means of the ball cratering technique (Calotest, CSEM) and image analysis (LECO 500). Twenty-four of the deposited samples were PHT in an argon atmosphere following a two-step process that involved an initial treatment at 473 K for 1 h and a subsequent heating at 673 K also for 1 h. The chemical analysis of such deposits was determined by means of SEM techniques (Hitachi S-2400) with EDS facilities. The observations were conducted at a constant potential of 20 kV. Tensile tests were carried out on a computer-

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493

Fig. 1. SEM photomicrograph illustrating a typical microstructure of the substrate. A large number of relatively coarse martensite plates (M) and carbide particles, visible as small white particles, can be observed.

controlled servohydraulic machine (Instron 8502) at a cross head speed of 10 mm/min. At least three samples were employed for characterizing the monotonic mechanical properties of both the coated and uncoated substrate. Fatigue tests were carried out under rotating bending conditions (Fatigue Dynamics, RBF-200) at a frequency of 50 Hz and alternating stresses of 590, 611, 634 and 663 MPa, for the uncoated substrate and the specimens in the as-deposited condition, which corresponds to 80, 83, 86 and 90% of the yield stress of the unplated substrate. For the coated and PHT samples the tests were conducted at 442, 516, 590 and 663 MPa, corresponding to 60, 70, 80 and 90% of the base steel. A total of 40 samples were employed for evaluating the fatigue properties of the uncoated substrate, 28 for the coated material and 28 for the coated and PHT samples, which exceeds the minimum number of specimens required in S-N testing for reliability data according to the ASTM standard 739 (12±24 samples). Thus, the testing procedure followed in the present work allowed a replication greater than 80%. In order to make possible a meaningful comparison of the fatigue life of the coated and uncoated specimens, all the samples were mechanically prepared in order to have similar polished surfaces before testing. The fracture surfaces of the samples were examined by means of SEM techniques, particularly in relation to the initiation of fatigue cracks and the different stages of their propagation.

489

site plates together with carbides, visible as small particles which constitute a typical tempered martensite structure, can be observed. On the other hand, Fig. 2 shows a view of the interface between the EN coating and substrate prior to fatigue testing, illustrating the deposition of an apparently uniform coating with satisfactory adhesion characteristics. The evaluation of the coated material during tensile testing and the observation of the fracture surfaces of the specimens after fatigue testing however, would indicate otherwise. As it has already been mentioned, the coating thickness was corroborated by means of the ball cratering technique, optical microscopy and scanning electron microscopy, which allowed to determine a mean value of approximately 24 mm. As shown in Fig. 3, the EDS analyses conducted on the deposit allowed to determine that the P content ranged between 12 and 14wt.%. As has already been pointed out, Parker and Shah [8] conducted an investigation concerning the effect of the P content of an EN deposit on the residual stresses within the coating. Accordingly, for a P content of the order of that present in the samples under examination, the residual stresses in the coatings are expected to be of a compressive nature. This fact would be in agreement with the ®ndings of Wu et al. [9] and Zhang et al. [10], who also reported compressive residual stresses of the order of 80 MPa in the coatings deposited and subsequently PHT at 473 K for 1 h, even though the P content of such deposits was lower (9.5wt.%) than that contained in the coated specimens employed in the present study, which also were in the asdeposited condition. 3.2. Evaluation of mechanical properties In order to evaluate if this particular deposit had any in¯uence on the monotonic mechanical properties of the composite coating-substrate material, a number of tensile

3. Results and discussion 3.1. Characteristics of the deposit Fig. 1 illustrates the typical microstructure of the substrate evaluated on the scanning electron microscope. The presence of a large number of relatively coarse marten-

Fig. 2. SEM view of the interface between the EN coating (D) and substrate (S) previous to fatigue testing. The deposit seems to be uniform and has apparently satisfactory adhesion characteristics due to the absence of visible cracks along the interface.

490

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493

Fig. 3. Typical EDS spectrum for the EN deposits involved in the present work.

tests were conducted with samples in the uncoated and coated conditions. The deposits plated onto the substrate employed in the present study did not show any signi®cant change either in yield stress or in the UTS of both the coated and uncoated base steel. The mean yield stress was found to be approximately 737 MPa, whereas the UTS was found to be approximately 1287 MPa. During testing of the coated samples, the deposits were observed to tear off signi®cantly from the substrate, indicating a poor adhesion of such coatings to the steel, which could worsen as a result of the difference in mechanical properties (elastic and plastic) between coating and substrate. The fact that the monotonic mechanical properties are observed to remain virtually unchanged after the application of the EN coating is not surprising since even if the plated deposits contributed somewhat to the tensile mechanical properties of the composite material, in the present case the thickness of such deposits is so small that its effect would be negligible. In relation to the fatigue tests conducted in order to evaluate the fatigue life of both the coated and uncoated samples, the determination of the monotonic mechanical properties of the material allowed to establish a stress amplitude range of 590±663 MPa for the substrate and coated asdeposited samples, which corresponded to a fraction of the yield stress of approximately 0.80±0.90. The coated and PHT specimens were tested in the stress range of 442±663 MPa, that is to say, 0.60±0.90 of the yield stress. The data showing the mean number of cycles prior to fracture (Nf) in terms of the alternating stress applied to the material (S) for the uncoated, coated as-deposited and coated and PHT specimens, are presented in Table 1. The results obtained have been plotted in Fig. 4 in which it can be observed that at each alternating stress level for both the coated as-deposited and coated and PHT materials, at least ®ve tests were carried out, whereas the fatigue properties of the uncoated substrate were evaluated employing at least eight samples at each stress. As mentioned before,

Fig. 4. Mean number of cycles prior to fracture (Nf) as function of the alternating stress applied to the material (S) for the uncoated, coated asdeposited and coated and PHT specimens.

these conditions allowed the ful®llment of the reliability conditions prescribed in the ASTM standard E 739. The most important aspect highlighted in Fig. 4 is the fact that plating an EN deposit of these characteristics onto the substrate steel signi®cantly decreases the fatigue life of the material in relation to the uncoated substrate, even if the coating is in the as-deposited condition, a state in which the maximum compressive stresses would be expected. In the as-deposited condition, at elevated alternating stress levels (663 MPa) the curve obtained for the plated samples indicates a reduction in fatigue life, in comparison to the uncoated substrate, of approximately 49.4%, whereas at low stresses (590 MPa) the samples present a reduction of approximately 77.7%. However, for the coated and PHT specimens the situation is even worse since at 663 MPa the fatigue life is reduced by 74.8%, whereas at 590 MPa it is reduced by 91.7%. These results, in a sense, corroborate those obtained by Wu and co-workers [9] and also by Zhang et al. [10] regarding the decrease in the fatigue limit of the 30CrMo steel when plated with EN deposits and PHT at different temperatures for different periods. According to these authors, a PHT for 1 h at 673 K gives

Table 1 Mean number of cycles to failure (Nf) vs. stress amplitude (S) for the uncoated and coated specimens S (Mpa) 442 516 590 611 634 663

Substrate

278538 ^ 144864 169913 ^ 36699 95100 ^ 33011 65125 ^ 6752

As-deposited

62160 ^ 10999 49900 ^ 6321 40860 ^ 14830 32960 ^ 3128

Deposited and PHT 39520 ^ 8698 33780 ^ 8611 22980 ^ 5151 16400 ^ 3883

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493 Table 2 Parameters involved in the Basquin relationship for the conditions tested Condition

A (MPa)

m

Substrate As-deposited Deposited and PHT

1555.6 4479.7 44354.1

2 0.077 2 0.184 2 0.431

rise to a decrease of 52% in the fatigue limit of the material, which initially was reported to be of the order of 750 MPa. The linear relationship between the alternating stress and the number of cycles to failure in a double logarithmic scale indicates the validity of the simple parametric expression earlier proposed by Basquin [11] for the description of this type of data, of the form S ˆ ANf2m

…1†

where A and m represent constants that depend on both material properties and testing conditions. A represents the fatigue strength coef®cient of the material and m the fatigue exponent. Table 2 summarizes the values of the parameters A and m for the three set of data represented in Fig. 4. The appropriate determination of the Basquin parameters, particularly for the composite coating-substrate material, is important for the evaluation of the fatigue performance of any component made of this steel that could be EN-coated either for improving some of its properties, such as corrosion and wear resistance, or achieving the required dimensions in order for the part to ful®ll properly its role in service. 3.3. Evaluation of the fracture surfaces of the samples The specimens tested at 590 and 660 MPa were examined after failure by SEM in order to study more closely the sites of crack initiation and their microstructural features, as well

491

as the microstructural changes that take place in general during the subsequent propagation of such cracks, leading eventually to the ®nal fracture of the samples. As an example, Fig. 5a,b shows two photomicrographs of typical crack initiation sites for samples tested under these conditions, respectively. In both cases, localized areas where the cracks nucleated, can be clearly observed. These constitute the focal points of a number of radial lines that propagate along the fracture surfaces. In Fig. 5a (sample tested at 590 MPa), it would seem that the crack nucleation site was associated with an isolated surface defect of a blister type and that the deposit-substrate interface was not disturbed to a signi®cant extent. For the sample tested at 663 MPa (Fig. 5b), severe secondary cracking along the deposit±substrate interface can be observed, possibly as a result of both poor bonding between them and the application of higher tensile stresses during each loading cycle. The severity of such cracking leads to the complete separation of the coating from the substrate in some areas of the cross section of the sample. Fig. 6a,b corresponds to a magni®ed view of the coating± substrate interface of the specimen shown in Fig. 5a, particularly at the site where the main crack started to propagate towards the substrate. The deposit is observed to remain relatively sound, although some secondary cracking can be seen specially at some areas of the interface. Also, it is possible to observe clearly some continuity of the fracture marks between the deposit and substrate, despite the relatively poor adhesion between them. This last feature indicates that the fatigue failure probably started at the surface of the deposit and propagated towards the base steel. Fig. 7, on the other hand, illustrates a magni®ed picture of the crack nucleation site of Fig. 5b. It clearly shows the severe cracking of the deposit and its separation from the substrate on both sides of the point where cracking of the substrate is believed to have started. Here, the metallic

Fig. 5. Typical crack initiation sites for samples tested at (a) 590 and (b) 663 MPa, respectively. At high alternating stresses (b), the deposit (D) has been severely detached from the substrate (S). The radial markings on both pictures indicate the origin of the fatigue failure.

492

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493

Fig. 6. (a) Magni®ed view of the coating-substrate interface of the specimen shown in Fig. 5a. (b) Magni®ed view of the coating-substrate interface of the specimen shown in (a). The coating has been identi®ed as (D) and the substrate as (S). Continuity of the fracture features between the coating and substrate can be clearly observed, as well as some secondary cracking along the interface.

bonding between deposit and substrate is apparently maintained to some degree, which again indicates that the possible failure mechanism should be the passage of fatigue cracks from the coating to the substrate. Therefore, cracks would be formed at the surface of the deposit, propagate through it leading to its fracture in some localized areas and ®nally transferred to the substrate in a region in which the metallic bonding between the coating and the substrate is still preserved.

Thus, even though the deposit would be under compressive residual stresses due to its P content, its lower mechanical strength in comparison with the substrate material leads to the its prior failure with the consequent transfer of the propagating cracks to the substrate. Hence, the coating actually operates as a fatigue crack source or surface stress concentration which becomes very effective even if the adherence of the deposit to the substrate is rather poor, as in the present case. This mechanism would explain the

Fig. 7. Magni®ed picture of the crack nucleation site of Fig. 5b. Severe cracking (C) of the deposit (D) and its separation from the substrate (S) can be clearly observed. The main crack has propagated from the deposit into the substrate.

Y. GarceÂs et al. / Thin Solid Films 355±356 (1999) 487±493

reduction in fatigue life observed for the coated specimens in comparison with the uncoated samples tested at the same alternating stresses. These features are consistent with the previous ®ndings of Zhang et al. [10] and Pertuz and co-workers [7], who were able to observe fatigue marks in EN deposits plated onto different steels as substrates. Such observations support the view that if the mechanical strength of the substrate is greater than that of the EN deposit or similar to it, the coating is bound to undergo fatigue failure before the substrate and to transfer the fatigue cracks to it, giving rise to a reduction in its fatigue strength. If, however the fatigue strength of the deposit is higher than that of the substrate, it is possible to observe an improvement in its fatigue performance, as it has been reported by Puchi and co-workers [3] for an EN deposit on AISI 1010 steel.

493

Thus, it is concluded that due to the relatively lower fatigue properties of the EN coating in comparison with the substrate, the deposit actually operates as a source of surface fatigue cracks, i.e. as a surface notch capable of imparting a signi®cant reduction to the fatigue properties of the substrate by reducing the time required for the nucleation of cracks. Coating of strong and tough substrates with weaker deposits in order to impart wear and/or corrosion resistance can thus lead to the fatigue failure of the component prior to expected. Therefore, mechanical design under conditions of high cycle fatigue should be based on the Basquin parameters of the fatigue life curve of the coated material rather than that determined for the uncoated substrate.

Acknowledgements 4. Conclusions Plating of a quenched and tempered AISI 4340 with EN deposits leads to a signi®cant reduction in the fatigue life of the material. Such a reduction, at a stress amplitude of 590 MPa, can achieve up to approximately 78% if the coating is in the as-deposited condition and 92% after a PHT such as the one explored in the present work. Thus, according to the present results, the reduction in fatigue life for this material when coated with this kind of deposits is much more significant than previously reported. It has been shown that such a decrease in fatigue performance occurs as a result of the passage of fatigue cracks form the coating to the substrate, a process which is believed to be assisted by the metallic bonding established between them. The continuity of certain fracture features between coating and substrate observed from the analysis of some fracture surfaces supports this view. In the present case, due to the signi®cant difference that exists between the elastic and plastic properties of coating and substrate, the adherence of the deposit to the base steel is rather poor. This conclusion is supported by the observation of extensive secondary cracking along the deposit± substrate interface after fatigue testing and the actual separation of the coating from the base steel during tensile testing. Nevertheless, such degree of adherence is found to be enough to allow the transfer of cracks from the coating to the substrate.

This investigation has been conducted with the ®nancial support of the Venezuelan National Council for Scienti®c and Technological Research (CONICIT) through the project LAB-97000644. J.A. BerrõÂos is deeply grateful to the Organization of the American States for the ®nancial support received through the Multinational Material Project. He is also grateful to the School of Mechanical Engineering of the University of El Salvador.

References [1] W. Riedel, Electroless Nickel Plating, Vol 48, ASM International, Metals Park, OH, 1991, pp. 48±181. [2] H. Izumi, H. Sunada, Y. Kondo, J. Soc. Mater. Sci. Japan 24 (1975) 320. [3] E.S. Puchi, M.H. Staia, H. Hintermann, A. Pertuz, J.A. Chitty, Thin Solid Films 290±291 (1996) 370. [4] J.A. Chitty, M.H. Staia, A. Pertuz, H. Hintermann, E.S. Puchi, Thin Solid Films 308±309 (1997) 430. [5] J.A. BerrõÂos, M.H. Staia, E.C. HernaÂndez, H. Hintermann, E.S. Puchi, Surf. Coat. Technol. 108±109 (1998) 466. [6] J.A. Chitty, A. Pertuz, H. Hintermann, E.S. Puchi, J. Mater. Eng. Perform. (1999). [7] A. Pertuz, J.A. Chitty, H. Hintermann, E.S. Puchi, J. Mater. Eng. Perform. (1999). [8] K. Parker, H. Shah, Plating 58 (1971) 230. [9] Y. Wu, Y. Zhang, M. Yao, Plat. Surf. Finish. (1995) 83. [10] Y.Z. Zhang, Y.Y. Wu, M. Yao, J. Mater. Sci. Lett. 15 (1996) 1364. [11] O.H. Basquin, Proc. ASTM 10 (2) (1910) 625.