PHYSICAL REVIEW B VOLUME $2, NUMBER 14 1 OCTOBER 1995-11 Lattice-parameter variation with carbon content of martensite. I, X-ray-diffraction experimental study iu Xiao, Zhong Fan, and Zhang Jinxiu Department of Physics, Zhongshan University, Guangzhou, 510275, People’s Republic of China Zhang Mingxing, Kang Mokuang, and Guo Zhengi Department of Material Science and Engincering, Northwestern Polytechnical University, Xian, 710072, People's Republic of China (Received 7 February 1995) ‘An experimental law of the x-ray diffraction of polycrystalline martensite is found, whichis satisfacto- rily explained from x-ray-difraction theory. The experimental phenomenon is well consistent with the hypothesis that at low carbon content martensite is of eubic structure, The cubic-tetragonal transition of ‘martensite concerned with the distribution of earbon atoms is observed, and an experimental critical point of the transition is also obtained. Dependences of the lattice parameter of low-carbon martensite ‘upon carbon content are established. With the dependences, it becomes possible to determine the carbon ‘content ofthe earbon-supersaturated a-phase of tee. 1, INTRODUCTION In the early 1920s, it was known that high-carbon fer- ous martensite is of tetragonal structure.'"* The depen- dence of the lattice parameters (a and c) of martensite upon carbon content was also established by Kurdjumov and Kaminsky” at that time, that is, c=aytap, a Bp, c/a=1+yp , w where p is the carbon content (wt %), ao the lattice con- stant of a-Fe, a=0.116+0.002, B=0.013:+0.002, and 7 =0.046:+0.001 This dependence was confirmed by many later investi- gations, and found to be approximately appropriate for polycrystalline and alloyed martensite.’~"? The results of various early investigations are shown in Fig. 1.° It should be noted that all these experimental data were obtained by the x-ray-diffraction method, and all are in the high-carbon range (20.56 wt %). In a quenched specimen with low carbon content, due to the low tetra gonality of martensite, the tetragonal splitting of the x- ray-diffraction peaks is hidden by the stress widening of the diffraction peaks. As a result, the values of a and ¢ cannot be determined. (Several isolated observations showed tetragonality of martensite at carbon content as Tow as 0.2 (Ref. 5) and 0.05 wt %, but the steels used were alloyed, and some other effect had apparently ar- igen.) Therefore, in Fig. 1 the low-carbon part (<0.56 ‘wt %) of the lines showing the dependence is only a sub- jective extrapolation. Zener'* suggested that the tetragonality of martensite results from the preferential distribution of carbon atoms to one sublattice of the octahedral interstitial sites in the bee lattice. Kurdjumov and Khachaturyan'®'© developed the initial suggestion with an order-disorder transition theory based on elastic interactions. Accord- ing to this theory, there exists a critical point of the order-disorder transition depending on temperature and (0163-1829/95/52(14)/9970(91/506.00 2 the carbon content of the carbon-supersaturated a phase Of steel. At room temperature (300 K) this point corre- sponds to 0.2 wt % carbon content, ie., when the carbon content decreases below 0.2 wt %, at room temperature the martensite should be of cubic, not tetragonal struc- ture, but up to now the eubic martensite has not yet been ‘observed. "* (The value of the critical point at room tem- perature has been given theoretically with fluctuating re- sults, first 0.2 wt %," then ~0.5 wt %,"” and lastly 0.2, wt," and the experimental results vary as well owing to the difficulty of detecting low tetragonality and the self-temperting effect of low-carbon martensite."*") Fur- thermore, the theory anticipates that on elevating the temperature tetragonal martensite may become cubic martensite, owing to the order—-disorder distribution of carbon atoms. However, this phenomenon has also not yet been confirmed,’ owing to the difficulty of ‘confirming whether the decrease (or disappearance) of tetragonality is indeed due to the order->disorder transi- tion or due to other effects occurring in martensite (in- 307 3.02 297| 5 oofursumet ang erst Sumontrleme a 292] “Tipson ane arta Lattice Parameter (A) 287| 2849 02 04 08 08 10 42 14 16 18 Carbon Content (wt %) FIG. 1. Early investigations of lattice-parameter variation of| ‘martensite with carbon content, summarized by Roberts (Ref. 6. 9970 (©1995 The American Physical Society2 LATTICE-PARAMETER VARIATION WITH... I... volving the segregation of carbon atoms to lattice defects, spinodal decomposition, or the precipitation of carbide). If this cubic-tetragonal transition concerned with the distribution of carbon atoms actually exists, the depen- dence of the lattice parameters of martensite upon carbon content will not be so simple as that shown in Fig. 1. Un- fortunately, up to now neither an experimental nor a theoretical result for the dependence has been obtained to our knowledge. ‘This paper reports an experimental phenomenon of the xray diffraction of polycrystalline martensite, which is satisfactorily explained by x-ray-diffraction theory. The phenomenon is well consistent with the hypothesis that at Jow carbon content martensite is of cubic structure. The cubie-tetragonal transition concerned with the distribu tion of carbon atoms in martensite is observed, and the experimental critical point of the transition is’ also ob- tained. The dependence of the lattice parameters of low- carbon martensite upon carbon content are established. With the dependences, it becomes possible to determine the carbon content of the carbon-supersaturated a phase of steel. This is very important for the study of the bai tic transformation and the tempering transformation of martensite.?"?? ‘On the basis of these experimental results, the previous theory of the cubic-tetragonal transition developed by Kurdjumov and Khachaturyan is reconsidered in the fol- lowing paper. Some valuable evolution and expansion is made, With the reconsidered theory, the experimental phenomenon is well explained. The dependence of the lattice parameters of martensite upon carbon content is computed theoretically. The result shows good con- sistency with the experimental values. 11, EXPERIMENT Material I is a group of plain steels varying in carbon ‘content, These steels were homogenized for 60h at 1200°C, and austenitized at 980°C, then quenched in a brine using specimens of 20X18X3 mm? (named bulk specimens). After mechanically polishing, the specimens were deeply etched with 15% HNO, +85% H,O, for the purpose of reducing the surface macro stress. Material II (iron: C 0.015, Si £0.10, Mn<0.10, P=0.002, $=0.021 wt %) and material III (steel alloy: C oon TABLE I. Carbon contents (wt %) ofthe specimens. Bulk specimens ‘Thin-plate specimens ‘Material T Material It Material It Specimen Carbon Specimen Carbon Specimen Carbon No. _content__No. No. content Hoo) S136, 12 03s) 2sog? 20s 13 048307637 T4038 0886 Ts 020 «S06 ESO. 16 oll = 6027s 17 0007028? m8 020 M8023 19 ls El 110 000 10s WL 000 0.15, Cr 1.36, Mn 1.35, Mo 0.45, Si 0.74 wt %) were machined to 20X 180.8 mm? plates after homogenizing for 48 h at 1200°C. After mechanical polishing to a final thickness of 0.6 mm, these thin-plate specimens were put into a computer-controlled carburizing furnace which ‘has been set up with a series of carbon potentials (the car- bonization materials were methanol and kerosene) for 3 h at 940°C, and then quenched into brine with temperature =S°C. ‘The carburized specimens were immediately preserved in liquid nitrogen until the x-ray-diffraction ex- periment. The contents of all experimental steels are list- ed in Table I The diffraction analysis is carried out with a D/max- 3C diffractometer using graphite monochromatic Cu Ka. radiation at 40 kV and 20 mA. The sampling space is 0,002", so the maximum angular resolution of the meter 0.002", The diffraction-strength-angle (i-20) profiles which show the tetragonal peak splitting are separated with a standard computer program, and every peak has undergone the Ka,, Ka; separation with the program. The Bragg angles of the peaks are determined with the parabola method. The angle-meter error is determined with a standard sample of silicon powder. (The half- ‘width of the peak of the powder near the {200} peak of iron is no more than 0.08".) The low-angle divergence is corrected with the Wilson-Pike formula. FIG. 2. Comparison of the positions and shape of the [110} and {200} diffraction peaks ‘of specimen I: (a) {110} peaks; (b) [200] peaks99r III, RESULTS AND DISCUSSION ‘The experimental law ofthe polycrystalline martensite series Figure 2 shows the [110} and {200} diffraction peaks of specimens T (bulk specimens). Although the carbon content is high, up to 0.79 wt % (specimen I-1), the peaks do not display evident tetragonal splitting, but the deflections of these peaks have some regular pattern: With increasing carbon content, the {110} peaks gradual- ly deflect to lower angles [Fig. 2)], the {200] peaks seem at first to deflect to lower angles and then to higher an- ales [Fig. 2(0)]. Specimens II and III were also used to verify this phenomenon. These specimens have no surface stress caused by mechanical polishing and are so thin (0.6 mm) that the quenching stress can be released to a great ex- tent; therefore their diffraction peaks should become sharp and more advantageous for observing the angle variation. Figure 3 shows the {110} and {200} peaks of specimens II. The situation is the same as that of speci- mens I, but the {200} peaks display tetragonal splitting beginning at the carbon content of 0.56 wt % (Fig. 4 shows individually the {200| peak of specimen II-5 which shows the tetragonal splitting). The tetragonal splitting of the {110} peaks begin at 0.76 wt % (specimen 113). Figure 4(b) clearly shows that with increasing car- bbon content the {200} peaks deflect at fist to low angles, LIU XIAO et al. 2 ® = a FIG. 4. The original {200} diffraction peak of specimen 15, in which the tetragonal spitting begins to emerge. then fo high angles, and the lowest angle is at 0.20 wt % C (specimen II-8). In alloy steel (specimens IMD, the same situation still remains (shown in Fig. 5). Here at 0.64 wt % C (speci ‘men III-4) the {110} peaks show the tetragonal splitting, and at 0.18 wt % C (specimen III-9) the diffraction angle of the {200} peak deflects to the lowest value. ‘The phenomenon shown in Figs. 2, 3, and 5 was always the same after repeated experiments, and the situation is still the same with Cr radiation. The maximum angular resolution of the meter at the present experimental condi- a 7% m1 a us m2 ML. FIG. 3. Comparison of the positions and shape of the {110} ‘and {200} diffraction peaks of specimen TI: (a) {110} peaks; (6) {200} peaks. ae eae 28 FIG. 5. Comparison ofthe potions and shape ofthe {110} an 200) ditacton peaks of pcimen II: (a [110] peak; {200} peas.2 LATTICE-PARAMETER VARIATION WITH... . I 9973 TABLE Tl. Mult the four plane-pais of tetragonal martens Diffraction [110} {200}, 2u plane (101) 20) 4 8 4 2 16 8 4 8 tions is 0.002", but the angle variations of the diffraction peaks are larger than 0.01° (see Tables IIT and IV below), 0 itis far from a random phenomenon. ‘Analysis ofthe experimental law based on x-ray difraction theory According to x-ray-diffraction theory, the overall diffraction intensity (1) of a reflection plane [character- ized by (HKL)] of a polyersytal is determined by” = AF ig sPen P(e RB) 5 2) where A is a constant (depending on the wavelength and intensity of the incident x ray and the radius of the difractometer), Funny the structure factor, Prox) the ‘multiplicity factor, g4@) the angle factor, e~*™ the tem- perature factor, and R (8) the absorption factor. For tetragonal martensite, there exist four pairs of diffraction peaks, (110) and (011), (200) and (002), (211) and (112), and (220) and (022). If the tetragonality is not large enough to split the pairs in the 1-20 profile, the structure factors, the angle factor (9), the temperature factors e~?™, and the absorption factors R (0) of the two peaks in a pair (such as Fo) and Foi) will be scarcely different, but the multiplicity factors will differ. Table II lists the values. For the (200) and (002) pair, because Piagy)=2 and 1, itcan be written that @ relatively sharp, the -2 profile can be approximated with a Gaussian function, Fie ® where ig is the maximum intensity in the profile, o a pa- 04, 03! = 02) whee ay eas aF aaa aes FIG. 6, Variation of m versus Ap/o. ~ ia is 47 where W, is the half-width of the peak), p=20, and 8g, Where 8 is the Bragg diffraction angle (i.e. , with d the spacing of the diffraction plane and 2 the wavelength of the x ray used). ‘Then the {200} composite profile can be approximated 4s follows [with the assumption that the (200) and (002) peaks have the same value of o Migagye PFa 5200) =foxao0)€ Flogoane (7 P08". o Denoting jaws) fo, Owing to (3), it ean be obtained that ‘qa Ziq) ~2ig- When dino) /4@™0, the angle & is , 20) obtained where the maximum intensity is reached in the composite {200} profile. That is, 28-ouow e001" HE—Paomde Poem" 9, (6) Obviously Gyan) >o, m=0 (Fig. 6 shows the variation of m versus ‘Apa; hence the top of the {200} peak is always close to that of the (200) peak. Equation (7) also gives that the value of & depends on the Values of Ap and g, ie. the 0.00 002 008 006 008 010 012 O14 016 owt %) FIG. 7. Variation of £ versus carbon content [with the as- sumption thatthe relation between the tetragonality and carbon ‘content of martensite corresponds with Eq, (1), with several different values of W,,9974 tetragonality of martensite and the half-width (W,) of the diffraction peaks. With the assumption that the rela tionship between the tetragonality and carbon conten consistent with Eq. (D, Fig. 7 is drawn (with several different W,.), showing the variation of £ versus carbon content of martensite. With increasing carbon content, the top of the composite {200} peak will deflect first 10 low angles, and then to high angles. When W,=0.2, 0.35%, and 0.5%, the carbon contents of the reflection points are about 0.035, 0.06, and 0.09 wt %, respectively. ‘The experimental values of 1, of the {200} peaks of specimens 11-10 (0.00 wt % C) and II-9 (0.14 wt % ©) are 0.2" and 0.35%, respectively. From Fig. 7 the value of £ of specimen II-8 should be larger than that of specimen II- 10, but this obviously disagrees with the experimental re- sults, and up to carbon content of 0.27 wt % (specimen TL) the experimental value of £ of plain steel martensite ddoes not yet exceed 65.017 (the experimental value of specimen 11-10). This situation can occur, in fact, only when the values of the lattice parameters @ and c simul- taneously increase, with increasing carbon content of martensite. As will be shown in the following paper, even if'in this case the martensite is tetragonal, the simultane- ‘ous increment of @ and ¢ will then imply a very small or- der parameter concerned with the distribution of carbon atoms. However, the order-lisorder transition concerned With the distribution of the carbon atoms in martensite of first-order character, i. the variation of the order pa rameter is noncontinuous atthe transition point; no soon- er does the ordering of the distribution of carbon atoms ‘occur than the order parameter jumps to a comparatively large value from zero. Therefore the simultaneous incre- ment of a and c implies in truth a disordered state of the distribution of carbon atoms, i.., the cubic structure of low-carbon martensite. Thus one can now think, with sufficient reason, that Eq. (1) does not correspond with the experimental phenomenon, and an arbitrarily small carbon content cannot cause the tetragonal structure of martensite. Furthermore, for tetragonal martensite, itis obtained from (6) that ag /_ ae BPo.200) 1 BPo002) (o/dg)*—2m? _1—m . (o/Ag? 21m? @ Figure 8 shows the variation of n versus Ag/o. It can be shown from Fig. 8 that at) ae Bram ”” Bea ° Relation (9) means that the top position of the compos- ite {200} peak is decided principally by go. [the Brags angle of the (200) peak]; in other words, the deflection of the unsplit {200} ‘peak principally represents the varia- tion of the spacing of (200) plane. With the same con- siderations, it can also be concluded that the deflection of the unsplit {110] peak represents the variation of the spacing of the (101) plane. In the present experiment, with increasing carbon content, the {110} peaks deflect LIU XIAO et al. 32 oy 0305 a7 os an 13 45 a9l0 FIG. 8 Variation of m [ie (8§/8gusm))/(86/8pc)] versus Ag/o. to low angles, meaning an, increase of the spacing of the (O1N plane (djpyrae/Va" rer) and the (200) peaks deflect first to Tow angle, then in the opposite direction, meaning that the spacing of the (200) plane increases at first and then decreases, ic, the lattice parameter @ in- ‘creases at first and then decreases. The multiplicity factor should have a similar effect on another group of peaks, ({220} and {211}), ic. the deflection of the {211} and [220} peaks should be caused principally by the variation of the spacing of the (211) and (022) planes. In the present experiment the effect is also observed, but with slighter evidence due to the small diffraction intensity ofthese peaks. It should be pointed out that when the 20 profile is not so sharp the approximation of it will be optimal with Cauchy function, = 1 IT Rig— oF ' (10) or a combination of the Gaussian and Cauchy functions, iene 1 i 2 . Re eo? ay ‘These different approximations will not affect the results mentioned above, but will give more tedious mathemati- cal forms. ‘The quantitative dependences of the lattice parameters ‘of martensite upon carbon content When the carbon content is very low, it can now be concluded that the martensite is cubic and only has one lattice parameter a. With increasing carbon content, the ‘martensite will become tetragonal, but when the tetra- gonality is not yet large enough to split the diffraction peaks, itis not possible to determine the real values of a and c. Therefore we have to denote by a’ and c’ the ap- arent lattice parameters. Here 2d 90)» €'=V2dr10) + ay where d)290) and ds10) are given by the Bragg differentformula 2d sind=Ay.,, where 0 is the Bragg angle of the {200} or {110} composite peaks and Zc, the wavelength of the Ka line of copper radiation (= 1.5405 A). Its easy to show that the value a’ is a good approxi- mation to the real value of a, but ¢” is far below c. Only ‘when the carbon content is high enough so that the diffraction peaks display the tetragonal splitting can we LATTICE-PARAMETER VARIATION WITH... . I 997 obtain the real values of @ and ¢. The diffraction values and the calculated lattice parameters (a of cubic marten- site, a’ and c’ of peak-unsplit tetragonal martensite, and @ and c of peak-split tetragonal martensite) are respec- tively listed in Tables III (plain steel) and IV (alloy steel). The linear regression equations between carbon content and lattice parameters are as follows. For the martensite of plain steel, TABLE IIL. X-ray-diffrction data and lattice parameters of specimens Il. For the cubic structure, the diffraction plane indices are written as (HK; the lt parameter is only a, with the average of (110) and (200) diffraction values, (a3) ap) /2, used as the experimental value of a. For the tetragonal structure, when the tetragonality is not large enough to split the diffraction peak, the difraction plane indices are written as {HK}, the lattice parameters are the apparent values a’ and c' nd in the table this part ighlighted by thick lines. When the tretragonality is large enough to split the difraction peaks, the diffraction plane indices are ‘written as (HAL), the experimental values of real lattice parameters @ and ccan be obtained, and if both the two peaks have split, the (io) a))/2 are used asthe experimental values of @ and c, respectively. may Ay] Vanesofai No. | gala 26 aA) gi | Am | Son | Sem | | * : A | & ‘ti am a9) amy | ance] ac zo | 21 i ‘20 0H ra 7]. ns | oar | aise ow aoe Lae anos | a7 awa | 2s na] os | aa son 2oms| ase | 20007 wo | aon 2osai care | arr ‘as | 30850 za 2osi7 sea | sr | arms | anose | osasr | casan| 20220 | 20sey | ass | ase asst ws | ust [ate | ares | oss | convo aon | 20 | 1220 | 120 ani | soe9976 LIU XIAO er al. 2 TABLE IV. X-ray-dilfraction data and lattice parameters of specimens III. The notations are the same as in Table 11 aA) fA) Ne 20 aA) Ana |Aom | ean | com | | a © Ay | AD aio) 20m) aio) @ — | 2807 |2.s080 0.00 2.866 | 2.804 ass 64976 2on7s 130 2.8676 2 xors 2.8008 mo} ors | seo 64930 2.0290 10 280s fa.seae 2.807 2.2009] 2.8700 ms |ow | ser 1927 zona Lasso 2.noa | 2.9873 2.870 {110} 1200} 1110) 1200} ¢ uns | o2s 2.8634 ]2.8931 a8 ‘A957 2.0097 Lan 2.8688 [2.8708 m7 | o2s | wer A097 2009 1.336 2.8672 | 2.8706] a.scta|2.8054 us | asi | 44459 65.065 20389 14323 2.8646 | 2.8792 2.8598] 2.9256] ms | oss | 4a osm 20ee 1a 2.8003 2.8805} 2.soa]2 9290] aio) | cory | 2oo) | oma) | cro) | cory | 00) | (2) | 2.8005] 2.8005] 2.9440 ]2 9448 m4 | a6 2.assi| 2.9406 «44.766 | 44.108 | 65.167 | cx.08s] 2.0227] 2.0516| 1.4303 | vars] 2.8006 294 2.8592 ]2.8590| 2.904 |2 9008 uns | 0.77 | 44792 |43.990] 68.207] 62.705 }2.0018 | 2.08665 | 14295 | 1.4808 2.9ses]2.9ss7 2.8591 2.9606 2.8575 2.8872 |2.9902 [2.0955 ma | tos | asis |as.738 |68.254] 61.896 | 2.0205] 2.0579 | 1.4286 | 1.4977 12 s28|2.0nn2 2857 2.9958 28567 ]2.8575| 3.0276] 3.0278 ta | 136 | 44830] 43.519 | 65.20 | 1.175 | 2.0200] 2.0778 | age | 1.5197 sae? | 3.202 2.8571 30752 LATTICE-PARAMETER VARIATION WITH... . I 8664+(0.020:0.002)p (wt %), p <0.6 (when p $0.18, c” becomes a), 2.8723 —(0.01340.002)p (wt %) , p >0.18 (when p>0.55, a’ becomes a) , c (A)=2.8664+ (0, 116+0.002)p (wt %), p>0.55 rr) For the martensite of 1.63Cr-1.35Mn-0.54Mo-0.54Si al- loy steel, e' (Ay 8674+(0.0200.002)p (wt %), p <0.6 (when p £0.18 e* becomes a), 8711 —~(0.014:40.002)p (wt 9), p >0.18 (when p 20.55 a’ becomes a), 2.8674-+ (0,1200.002)p (wt %), p>0.55 . a4) 2.8664 and 2.8674 A are the lattice constants of the equi librium a phases of iron and the alloy steel, respectively, which are precisely determined with the extrapolation method, Equations (13) and (14) may contain the following er- rors: the diffractometer system error (this has been corrected for, so can be neglected), the stochastic errors Of the determination of diffraction parameters and carbon content (0.02 wt %), and the regression error. It is usual- ly considered that the regression error is larger than the other errors. Here the error due to the self-tempering ‘effect of martensite will not be important, because the specimens are very thin (0.6 mm) and the quenching medium is low-temperature (—5°C) brine which has strong quenching capability; in addition, the specimens were kept in liquid nitrogen before the diffraction experi ‘ment. ‘The comparison between the experimental values, Eqs. (13) and (14), and Eq. (1) is shown in Fig. 8. It may be seen from Fig. 8 that the values of c' and a (of the cubic structure) have quite good linearity; therefore in Eqs. (13) and (14) they were regressed together, and so were a’ and 4a (of the tetragonal structure). The tetragonal splitting of diffraction peaks occurs when p >0.55 wt %, and the ex- perimental values of ¢ of plain steel correspond well with Eq. (1), but the a(a’)-p line is moved upward a little. The existence of cubic martensite is more clearly shown in Fig. 9, and the experiment gives 0.18 wt % as the critical carbon content of cubic and tetragonal martensite of plain steel. It is believed that the transition is really ob- served in the present experiment. The tetragonality of the alloy steel is slightly larger than that of plain steel The c’(avp lines of plain steel and the alloy steel are of 1W. L. Fink and E. D. Campbell, Trans. Am. Soe. Steel Treat. 9, 717 1926) 2N, Sljakov, G. Kurdujmoy, and N. Goufdob, Z. Phys. 48, 384 (1927), oor 3.06 304 302 Lattice Parameter (A) 00 02 04 06 08 10 12 14 16 Carbon Content piwt %) FIG. 9, Lattice-parameter variation with carbon content of ‘martensite; comparison between the experimental values, Eqs (13) and (14), and Eq, (D. the same slope, but have different constant terms. There- fore, for the purpose of determining the carbon content of the carbon-supersaturated a phase of different steel al- loys, it is suggested that a general dependence of c'(a)-p can be written as '(A)=ay +(0.020+0.002)p (wt %), p 0.6 (when p $0.18, e* becomes a), (15) where ag is the lattice constant of the different steels. IV, CONCLUSION (1) An experimental law of the x-ray diffraction of polycrystalline martensite is found. The law is satisfacto- rily explained with x-ray-diffraction theory. (2) The experimental phenomenon is well consistent with the hypothesis that low-carbon martensite is of cu- bic structure, and the cubic-tetragonal transition of mar- tensite concerned with the distribution of carbon atoms is observed. The experimental value of the critical point of the transition is obtained; at room temperature it is 0.18 ‘wt % C, corresponding with the theoretical estimation. (G3) The dependences of the lattice parameters of mar- tensite upon carbon content are obtained for plain steel (Eq. (13)}, for the 1.63 Cr-13.SMn-0.54Mo-0.54Si alloy steel (Eq. (14), and for a general stee! alloy (Eq. (15)) With these dependences, it becomes possible to determine the carbon content of the carbon-supersaturated a phase of steel. ACKNOWLEDGMENTS. This work was financially supported by the Advanced Research Foundation (Hong Kong) for Zhongshan Uni- versity, and the authors express their special appreciation to Xian Weijian 3G. Kurdjumoy and E. Kaminsky, Nature 122, 475 (1928). 4G. Kurdjumov and E. Kamisky, Z. Phys. 83, 696 (1929). 5G. Kurdjumov, J. Iron Steel Inst, London 198, 26 (1960). 6C.8, Roberts, Trans, Am, Inst. Metall, Eng. 197, 203 (1953)9978 7G. Hage, J. Iron Steel Inst, London 130, 439 (1934). "E, Ohman, J. Iron Steel Inst. London 123, 445 (1931). °K. Honda and Z. Nishiyama, Trans. Am. Soc. Stes! Treat. 20, 464 1932). YH, Lipson and A. M. B. Parker, J. Iron Steel Inst. London 149, 123 (1944), NJ, Mazur, Nature 166, 828 (1950). "2G, Kurdjumov, Metal. Trans. 7A, 99 (1976). "8p. G, Winchell and M. Cohen, Trans. Am. Soc. Met. 55, 347, 11962). MC. Zener, Trans. Am. Inst. Metall. Eng. 167, $50 (1946); Phys. Rev. 74, 639 (1948) 5G. Kurdjumov and A. G. Khachaturyan, Acta Metall. 23, 1077 (1975). "8A. G. Khachaturyan, Theory of Structural Transformations in Solids (Wiley, New York, 1983). LIU XIA0 et al 2 7A. G. Khachaturyan and G. A. Shatalov, Fiz. Met. Metal- loved. 32, (1971). *E. Z. Kaminsky and M. D. Perks, Probl. Metalloved Fiz. Met. 4,21 (1949). 19M, Cohen, Trans. Am. Inst. Metall. Eng. 224, 638 (1962) 28G_ V. Kurdjumov and A. G. Khachaturyan, Metal. Trans. 3, 1069 (1972). 21C. M. Wayman, Martenstic Transformations I, Proceedings of the 6th International Conference on Martensitic Transforma- tions, Sidney, Australia, 1989, edited by B. C. Muddle (Trans Tech, Switzerland, 1990), p. 1. 22H. K. D. H. Bhadeshia and J, Christian, Metal. Trans. 21, 761 (1990) HL, Klug and L. Alexander, X-Ray Difaction Action Pro- ‘cedures for Polyerystalline and Amorphous Materials (Wiley, ‘New York, 1974, p. 135