Thermodynamic Properties and Phase Diagrams of the Fe-Co and Ni-Pt Systems BY K. C. MILLS,"M. J. RICHARDSON AND P. J. SPENCER National Physical Laboratory Teddington Middlesex Received 22izd October 1973 Heat capacities for Fe-Co and Ni-Pt alloys of various compositions have been measured between 300 and 1130 K. The enthalpies of the orderdisorder transformation occurring in these alloys were determined from these measurements. These enthalpy of transformation values were combined with extant thermodynamic data for the disordered phase to calculate thermodynamic properties for the ordered phase which were then used to derive more information on the equilibrium diagrams for the Fe-Co and Ni-Pt systems. Iron-cobalt and nickel-platinum alloys undergo order-disorder transformations over certain composition ranges in the solid state.The boundaries of the ordered and disordered phases in these systems are not known with certainty despite the com- mercial significance of the alloys. The aim of the present study was to obtain thermodynamic information which could be combined with extant data to calculate reliable order-disorder phase boundaries for the Fe-Co and Ni-Pt systems. For lower temperatures calculated boundaries are often more reliable than those obtained from metallographic examination for the latter frequently correspond to non-equilibriumconditions arising fromdiffusional delays. In contrast the thermodynamic values necessary for calculation of the same boundaries can be measured at higher temperatures where diffusion is sufficiently rapid to achieve equilibrium conditions then extrapolated to lower temperatures.Elliott has reported a phase diagram for the Fe-Co system which contains the three solid phases ' FeCo ' ' Fe,Co ' and ' FeCo '; however Hultgren has proposed that only the 'FeCo ' phase exists as the evidence for the 'Fe,Co ' and ' FeCo,' phases is tenuous. Asano et aL3 employed a variety of techniques to investigate the structural changes in Fe-Co alloys but were unable to obtain any positive evidence for the existence of ' Fe,Co '. The ' FeCo ' phase undergoes an order-disorder transition around 1000 K when the ordered (a') b.c.c. structure is transformed into a disordered (a) b.c.c. structure. Heat capacity measurements have been reported 3-10 which show that the Cp(T) curve undergoes a rapid increase of Cp during the order-disorder transformation culminating in a peak in the Cp(T)curve at the critical temperature T,.However the Cpmeasurements also revealed the elcistence of a small peak in the Cp(T)curve around 820 K ; this is known as the '820 K anomaly '. Gormankov et all1 reported neutron diffraction studies which indicated that the degree of long-range order de- creases in the temperature range 770-820 K. Yokoyama et aL6 have reported that the temperature at which the anomaly occurs increases with increase in heating rate and have proposed that this effect is caused by the disordering process requiring adequate time for equilibrium to be attained. Thermodynamic data for the disordered (a) phase have been determined by Satow 40 K.C. MILLS M. J. RICHARDSON AND P. J. SPENCER et aZ.,12 who measured the Gibbs energy change AG for the reaction Fe(in alloy)+ H,O(g) = FeO(s)+H2(g). However these data were considered unreliable by Hultgren et aL2in their assessment of the thermodynamic properties of the Fe-Co system. Thus the thermodynamic data for the disordered (a) phase employed in this study are those obtained by Kaufman and Nesor l3 from combination of the calorimetric enthalpy of formation values due to Muller and Hayes l4 with known characteristics of the phase diagram. Thermodynamic data for disordered (a) phase alloys were combined with enthalpies and entropies for the order-disorder transformation (AHtrans and AS,,,,) determined in the present work to yield thermo- dynamic data for ordered (a’) phase alloys and by calculating AG(x) curves for both the ordered and disordered phase at a series of temperatures the equilibrium order- disorder phase boundaries were obtained by the common-tangent method.Hansen and Anderko l5 have reported that the Ni-Pt phase diagram contains two ordered phases the cubic ‘Ni,Pt ’phase and the tetragonal ‘NiPt ’phase both of which transform into a disordered f.c.c. structure at higher temperatures. However the phase boundaries of neither ordered phase are known with certainty. Gibbs energies for the disordered phase have been measured by galvanic cell studies l6 but the thermodynamic data listed by Hultgren et aL2have been employed in the present calculations; these data are based on AH; values and nickel vapour pressure measure- ments due to Walker and Darby.17 The data have been combined with the present measurements to yield more information on the equilibrium diagram of Ni-Pt.EXPERIMENTAL MATERIALS The alloys 1-9 listed in table 1 were the samples used in the calorimetric studies made by Miiller and Hayes l4 and the preparation of these alloys has been fully described.14 Alloys 10-12 were kindly donated by Dr. H. B. Bell of Strathclyde University and were prepared by high-frequency melting of weighed mixtures of iron and cobalt powders in an atmosphere of hydrogen. TABLETH THE COMPOSITION OF THE Fe-Co ALLOYS alloy 1 2 3 4 5 6 7 8 9 10 11 12 nominal XF~ 0.3 0.4 0.48 0.5 0.5 0.5 0.52 0.6 0.7 0.25 0.5 0.75 nominalxc, 0.7 0.6 0.52 0.5 0.5 0.5 0.48 0.4 0.3 0.75 0.5 0.25 wt.%C 0.006 0.005 0.012 0.004 0.005 0.006 0.006 0.006 0.005 0.006 0.007 0.008 TABLE2.-THE COMPOSITION OF THE Ni-Pt ALLOYS alloy 1 2 3 4 5 6 7 XPt 0.2234 0.2542 0.2848 0.3972 0.4430 0.4886 0.5192 XNi 0.7766 0.7458 0.7152 0.6028 0.5570 0.5114 0.4808 All alloys were annealed for 4 weeks at 730 K in evacuated silica capsules and slowly cooled to room temperature. The alloys were chemically analysed for carbon the results being given in table 1. The Ni-Pt alloys were provided by Johnson Matthey Co. and were prepared by melting a mixture of high-purity elements in an atmosphere of argon using an arc-furnace with a non-consumable electrode. The samples were turned over and remelted several times to ensure that homogeneity of the samples had been fully attained.The chemi- cal analyses of these alloys were carried out by Johnson Matthey Co. the results being given in table 2. The alloys were annealed for 6 weeks at 700K in evacuated silica capsules and slowly cooled to room temperature. HEAT CAPACITIES FOR Fe-Co AND Ni-Pt ALLOYS THERMAL MEASUREMENTS The measurements were made using a Perkin-Elmer differential scanning calorimeter (DSC) model 2. The controls of the DSC were adjusted to extend the upper limit of the temperature range to 1130 K. The output signal from the DSC was measured with a digital voltmeter and recorded on paper tape. The treatment of the data to yield Cpand enthalpy values was identical with that described previous1y.l8 To check the method the heat capacity of silver was measured between 300 and 800 K,and the experimental Cpvalues obtained were in excellent agreement with the assessed CJT)relationship reported by Hultgren et aLL9 The scatter of individual experimental points from the reported CJT)curve had a standard deviation of 0.5 %.Temperatures are based on the IPTS-68 scale. RESULTS In this study " one mole " refers to the species Fe,Co,- and Ni,Pt,-,. R was taken as 8.3143 J K-l mol-l and the molar masses were taken from the IUPAC Tables (1970). The heat capacities of the Fe-Co alloys are shown in fig. 1-4 and thermodynamic properties for these alloys are listed in table 3. In this investiga- tion with a heating rate of 0.3333 K s-l the ' 820 K anomaly ' was displaced to 869K for Feo,5Coo.5.The heat capacities of the Ni-Pt alloys are shown in fig. 5 and 6. D lOOr I 300 500 700 900 1100 TlK FIG.1.-The heat capacity of Feo.sCoo.s. Values of AHtrans and AS,,,, for the order-disorder transformation were calculated from the CJT)curves. However the value calculated for AHtrans is dependent upon the value used for the heat capacity of the ordered phase C,(ord) in the transformation region. The Cp(T)curve for Feo,sCoo.5 shown in fig. 1 exhibits the following charac- teristics (a) a smooth CJT) relationship for the ordered phase (region AB) (b) a sudden increase in C culminating in a peak (at C) (c) a further increase in Cp culminating in a peak (at D) (d)a smooth Cp(T)relationshipfor the disordered phase (EFG).In this study measurements were restricted to temperatures up to 1130 K 0 300 500 700 900 1100 TIK FIG.2.-The heat capacity of Fe0.3C00.7 Feo.4cOi).6 and Feo.4sCoo.s2 ; --Feo.sCoo.,; --- Feo.4Coo.6; -y Feo.48Coo.52. t 101 1 1 I I I I I I 300 500 700 900 1100 TIK FIG. 3.The heat capacity of Feo.szCoo.48 Feo.6Coo.4 and Feo.7Coo.3 ; - Feo.5zCoo.4a ; --- Feo.&o0.~; --,Feo.7Coo.3. TABLE 3.-THERMODYNAhfIC PROPERTIES OF FeCO ALLOYS T/K 400 2 624 7.56 2 546 7.34 2 541 7.33 2 556 7.37 2 568 7.40 500 5 341 13.62 5 188 13.23 5 130 13.10 5 214 13.29 5 243 13.36 600 8 197 18.82 7 991 18.33 7 858 18.07 8 038 18.44 8 101 18.57 700 11 222 23.48 10 963 22.91 10 746 22.52 11 037 23.06 11 157 23.28 800 14 851 28.32 14 286 27.34 13 900 26.72 14 216 27.30 14 416 27.62 900 18 659 32.81 18 392 32.16 18 100 31.63 18 424 32.23 18 561 32.49 loo0 22 413 36.76 23 429 37.48 24 332 38.13 24 730 38.75 22 758 36.92 1100 26 324 40.49 27 684 41.53 29 542 43.11 28 600 42.00 26 854 40.82 804k 5 9423- 5 10073-5 9763- 5 870&5 K.C. MILLS M. J. RICHARDSON AND P. J. SPENCER and it was not possible either to establish or predict with accuracy the C,(T)relation-ship for the disordered phase (curve FO in fig. 1) for the various alloys. Thus it was not possible to determine C,(ord) in the transformation range (curve BJF) with cer-tainty. For this reason the C,(ord) in the transformation interval is assumed to be c I 500 700 900 110C' T/K ; ; FIG.4.-The heat capacity of Feo.t5C00.75 and Feo.75Coo.25 -- Fe0.25C00.75 -- Fee.7 5C00. 5. given by the line BHE in fig 1. This assumption could readily be applied to all the alloys studied and is denoted as method 1. Values of AHtrans and AStranswere calcu- lated respectively by evaluation of Tat J (Cp-Cp(ord)) dT and Tat B Tat B thus AHtrans is obtained from the area enclosed by BCDKEHB in fig. 1.t Pepperhoff 20' 300 500 700 900 1100 TIK FIG.5.-The heat capacity Of Nio.78Pt0.22 and Ni0.7zPt0.28 ; --,Ni0.78Pf0.22; Ni0.72Pf0.28. -9 f It was not possible to assign a reliable value to AHtrans from the plot of (HT-H~~~) against temperature. HEAT CAPACITIES FOR Fe-Co AND Ni-Pt ALLOYS and Ettwig have reported heat capacities for the temperature range 600-1250 K which are in good agreement with the present measurements ; the CJT) relationship for the range (1 100-1200 K) reported by these workers can be combined with the CJT)relationship obtained in the present study to yield a smoooth relationship for C,(ord) against T in the transformation range.Pepperhoff and Ettwig obtained AHt,,, from the area enclosed by BCDKJB in fig. 1 ; this is referred to as method 11. Values of AHtranscan only be obtained by method I1 for alloys of composition Feo.5Coo.5 and these are 100 J mol-' greater than those obtained by method I. The plot of AH,,,, against composition for the Fe-Co system is given in fig. 7. The value of AHtransobtained from the present work for Feo.4C00.6 was lower than that predicted from the smoothed AH,,,,,(x) curve and the values of AHtransused in the calculation of the equilibrium phase boundaries were taken from the smoothed AHt,,,&) curve.The values of AH:,,, and AS,,,, for the order-disorder transforma- tions in Ni-Pt alloys were calculated in an identical manner to those for the Fe-Co system. To calculate the order-disorder phase boundaries in the Fe-Co and Ni-Pt systems the following procedure was adopted. The experimental values of AH,,,, and AS,,,, for Fe-Co and Ni-Pt alloys were combined with AH; and AS; values for the disordered phase 139 at the appropriate alloy compositions to yield AH; and AS; values for the ordered phase. AG; values were then calculated for the two phases at K.C. MILLS M. J. RICHARDSON AND P. J. SPENCER m 4-4 1 -8 +-.y -$ % a 2-d 0 . I I 8 selected temperatures using the equation AG; = AH; -TAS; and AG(x) curves were hence derived for each temperature. Equilibrium boundaries were obtained by draw- ing common tangents to the AG(x) curves.2o*21 The slightly larger AHtransand AS,,,, values obtained by method I1 would have little affect on the calculated phase I100 900 700 500 0 0.2 0.4 0.6 0.8 xco FIG.8.-Calculated order-disorder phase boundaries for the Fe-Co system. HEAT CAPACITIES FOR Fe-Co AND Ni-Pt ALLOYS boundaries as the increased AH,,,, and AS,,,, terms in the equation AG = AH-TAS tend to cancel each other.The relevant portions of the equilibrium diagrams for the Fe-Co and Ni-Pt systems are given in fig. 8 and 9 respectively. 9 oc 0 00 700 6 OC 5 00 400 0.4 0.6 0.8 xpt FIG. 9.-Calculated equilibrium diagram for the Ni-Pt system; - caIcuIated boundaries ; --,tentative boundaries for the aft' and a" phases. DISCUSSION Fe-Co SYSTEM The heat capacities of the alloys for the temperature range 300-400 K are within 43% of the values calculated by Kopp's rule.? The experimental values of Cp recorded in this study for Feo,5Coo.5 agree with the values obtained by Pepperhoff and Ettwig * to within about 1 % and are in good agreement with preliminary Cp measurements made by Normanton et al.l0 However the Cpvalues recorded here are about 7 % lower than values reported by Japanese worker^,^-^ 10 % lower than the values recorded by Castanet and Ferrier,' and 5 % higher than the preliminary Cpvalues determined by Sale.g The CJT)curve for Feo.2,Co,., (fig.4) shows two small peaks at 760 and 980 K corresponding to the a' -+ a transformation and the onset of the a 4y transformation (see phase diagram 2 respectively.The peak in the Cp(T)curve for Feo.,5Coo.25 at 760 K is presumably associated with the a' -+a transformation but no ready explanation can be given for the small Cppeak observed at 880 K. The scatter of Cp values obtained with these two alloys (+2 %) was greater than that usually observed. t At higher temperatures both iron and cobalt undergo transformations at 1033 and 740 K respectively; at temperatures where the C' values of the elements are influenced by these transfor- mations agreement between Kopp's rule and the experimental values of C of the alloy would not be expected.K. C. MILLS M. J. RICHARDSON AND P. J. SPENCER The values of AHtrans for the order-disorder transformation were calculated by assuming that the ' 820 K anomaly ' is an integral part of the total enthalpy required for the a' a transformation (cf. ref. (7)). The values of AHtrans obtained depend upon the base line used i.e. C,(ord) ; however values obtained for Feo.5Coo.5 by methods I and I1 of 4.1 1 k0.2 and 4.21 k0.2kJ mol-1 respectively are in good agree- ment with the value AH,,,, = 0.5RTc = 4.17 kJ mol-' calculated by the Bragg- Williams theory.22 Values of AHtrans reported by other investigations are shown in fig.7 and are in reasonable agreement with those obtained in this study. Most order-disorder transformations in metallic systems are considered to be "first-order transitions ",2 e.g. the Ni-Pt tran~formation.~~ Although the trans- formation in Fe-Co alloys has been reported as a " second-order transition " the " order " of the transition cannot be considered to be firmly e~tablished.~~ A diffi-culty encountered with the application of DSC to the study of order-disorder trans- formations is that the dynamic nature of the method inevitably produces a broadening of the Cppeaks. Consequently it is difficult to distinguish some "first-order transi- tions " from " second-order transitions ".The equilibrium boundaries between the ordered and disordered phases have been calculated on the assumption that the trans- formation in Fe-Co alloys is "first-order ". Thus the ordered and disordered phases are separated by a two-phase region.23 However the phase boundaries shown in fig. 8 could still be considered to define the transformation limits even if a " second-order transition "pertained. Fig. 8 indicates that the ' FeCo ' phase occurs between Fe,.,Co,. and Fe0.7C00.3. No evidence was found for the existence of ' Fe3Co ' and ' FeCo ' phases; in fact from fig. 8 the compositions Fe0.,SCo0.25 and Feo.25Coo.,5 lie within calculated two-phase regions. It is for this reason that AHtrans values for these compositions were not shown in fig.7 as the measured AH,,,, values probably do not correspond to a complete order-disorder transformation. At temperatures below 800 K the boundary of the orderedphasefor iron-rich compositions was calculated to turn inwards markedly towards the equimolar composition. This behaviour is improbable. The extant thermodynamic data for the disordered phase are not sufficiently accurate to merit drawing the boundary in this manner and hence the boundary has been constructed by a smooth extrapolation of the high-temperature boundary. It is illustrated by a dotted line in fig. 8 to indicate the uncertainty associated with its position. Yokoyama et aL6 noted that the position of the anomalous peak around 820 K observed in dynamic measurements of heat capacity was dependent upon the heating rate used.They attributed the ' 820 K anomaly ' to the slowness of the disordering reaction and applied the Bragg-Williams theory 22 to account for their observations. If the sluggishness of the disordering process is such that an alloy at temperature T results in a dynamic degree of order corresponding to an equilibrium degree of order for a lower temperature 8 then it can be shown from theory 22* that eqn (1) is applicable where E is the activation energy k the Boltzmann constant and A a con- stant with a computed value of In [-(8-T)]= In A +In (dO/dt) +E/RT. (1) In this study peak temperatures of 852f5 869f5 880+5 and 887f5 K were observed for Feo,5Coo,5 with heating rates of 0.167 0.333 0.667 and 1.333 K s-' respectively.The best fit of these data and data from the literature could be obtained with E = 199 kJ mol-l and A = 1.7 x 10-lo in reasonable agreement with the values obtained by Yokoyama et aL6 HEAT CAPACITIES FOR Fe-Co AND Ni-Pt ALLOYS Ni-Pt SYSTEM The C,(T) curves for Ni-Pt alloys in the ‘ Ni3Pt ’ phase field (fig. 5) show two Cp peaks the first corresponding to the Curie temperature and the second to the trans- formation of the ordered cubic (a’) phase into a disordered f.c.c. (a) structure. The Cpvalues calculated by Kopp’s rule are within 2 % of the experimental value for the ‘NiPt ’phase. The C,(T)curves (fig. 6) for alloys of composition Ni0.51Pt0.49 and Ni0.48Pt0.52 in the ‘NiPt ’ phase field show two C peaks at high temperatures.It is proposed here that the first peak is associated with the transformation of the ordered tetragonal (a”) phase into an (a’”) superstructure and the second peak with the transformation of this (a”’) phase into the disordered f.c.c. (a) structure. The CJr) curves for both the other two alloys in the ‘ NiPt ’ phase field show only one Cppeak. However at temperatures slightly above the peak temperature where Cp is decreasing rapidly with temperature there is a small arrest. This arrest could correspond to the formation of the a”’ phase but might also be the result of experi- mental uncertainties associated with individual C values. The evidence for the a’‘’ phase is discussed below. The equilibrium diagram for the Ni-Pt system in fig. 9 is not complete as no structural observations of the a”’ phase field have yet been made.The boundaries of the ‘ Ni3Pt ’ phase were calculated from the experimental AHtrans and AS,,,, values for the three appropriate alloy compositions in an analogous manner to that deccribed for Fe-Co alloys. Unfortunately the position of the maximum in the AH,,,,,(x) plot could not be accurately determined from these three measurements alone with the result that the AG(x) curve for the ordered phase cannot be defined with certainty. Thus the calculated phase boundaries of the ‘Ni,Pt’ or a’ phase are estimated to be accurate to only +3 mol %. The only direct evidence for the existence of the a”‘ phase is the observation of the two maxima in the CJT) curve. However the ordered (a”) phase for ‘NiPt ’ is iso- typic with the ordered f.c.tetragonal phase (a;) for ‘ AuCu ’,and the phase diagram proposed for the Au-Cu system 15*24 shows that with increasing temperature the f.c. tetragonal structure is transformed first into an f.c. orthorhombic superstructure and that at higher temperatures a; transforms into the disordered f.c.c. phase (a). Hirabayashi 25 reported a C,(T) curve for ‘AuCu’ which showed two C peaks corresponding to the temperature for the two transformations. The behaviour of the ‘ AuCu ’ phase alloys appears to be similar to that observed here for the ‘NiPt ’ phase alloys. It was not possible to resolve the twin-peak area of the C,(T)curve (fig. 6) with anv certainty into two AH,,,, values corresponding to the a”-+ a”‘ and the a’” -+a transformations.Thus it was not possible to calculate phase boundaries for a” and a”’ in the usual way. Therefore we have proposed tentative phase boundaries for these phases in fig. 9 based on the observed temperatures of the Cpmaxima and reference to the phase diagram 23 of the Au-Cu system. We thank Johnson Matthey and Co. Ltd for providing the Ni-Pt alloys and Dr. H. B. Bell (Strathclyde University) for providing some Fe-Co alloys. We are grateful to D. G. Nunn and E. B. Lees for rendering assistance and to Dr. E. J. McLaughlan for the chemical analyses of the Fe-Co alloys. We also thank Dr. R. Castanet (Marseille) Dr. F. Sale (Manchester University) Dr. A. S. Normanton Dr. R. Buckley and Prof. B. B. Argent (Sheffield University) and Dr.J. F. Counsel1(NPL) for helpful discussions. K. C. MILLS M. J. RICHARDSON AND P. J. SPENCER R. P. Elliott Constitutionof Binary Alloys 1st suppl. (McGraw-Hill 1965). R. Hultgren R. L. Orr P. D. Anderson and K. K. Kelley Selected Values of Thermodynamic Properties of Metals and Alloys (John Wiley New York 1963) suppl. sheet May 1971. H. Asano Y. Bando N. Nakanishi and S. Kachi Trans. Jap. Inst. Metal. 1967 8 180. S. Kaya and H. Sato Proc. Phys. Math. Soc. Japan 1943,25,261. H. Masumoto H. Saito and M. Shinozaki Sci. Rept. Res. Inst. Tohoku Univ. 1954 6 523. T. Yokoyama T. Takezawa and Y. Higashida Trans. Jap. Inst. Metal. 1971 12,30. R. Castanet and A. Ferrier Coll. int. C.N.R.S. no. 201 Thermochemie (Marseille 19721 collected papers p.345 ; see also Compt. Rend. Ser. C 1971 272 15. W. Pepperhoff and H. H. Ettwig reported by G. Inden and W. Pitsch Chemical Metallurgy of Iron and Steel Proc. Int. Symp. Metallurgical Chemistry-Applications in Ferrous Metallurgy Sheffield 1971 (Iron and Steel Inst. London 1973) p. 314-316. F. R. Sale Chemical Metallurgy of Iron and Steel Proc. of Int. Symp. on Metallurgical Chemistry -Applications in Ferrous Metallurgy Sheffield 1971 (Iron and Steel Inst. London 1973) p. 330. lo A. S. Normanton P. Bloomfield and B. B. Argent (University of Sheffield Sept. 1973) private communication. l1 V.I. Gormankov,D. F. Litvin A. A. Loshmanov G. G. Lyashenko and I. M. Puzei Sou.Phys. Cryst. 1962 7 637. T. Satow S. Kachi and K. Iwase Sci. Rept. Res. Inst. Tohoku Univ.1956,8,502. l3 L. Kaufman and H. Nesor 2. Metallkunde 1973 64,249. l4 F. Miiller and F. H. Hayes J. Chem. Thermodynamics 1971,3 599. M. Hansen and K. Anderko Constitutionof Binary Alloys (McGraw-Hill 1958). l6 K. Schwerdtfeger and A. Muan Acta. Met. 1965 13 509. l7 R. A. Walker and J. B. Darby Jr. Acta. Met. 1970 18 1261. l8 K. C. Mills and M. J. Richardson Thermochim. Acta 1973 6,427. l9 R. Hultgren R. L. Orr P. D. Anderson and K. K. Kelley Selected Values of Thermodynamic Properties of Metals and Alloys (John Wiley New York 1963). 'O F. Muller and 0. Kubaschewski High Temp.-High Pressure 1969 1 543. P. J. Spencer and F. H. Putland J. Iron Steel Inst. 1973 211 293. 22 W. L. Bragg and E. J. Williams Proc. Roy. SOC. A 1935 151 540. 23 R. S.Irani J. Contempory Phys. 1972 13 559. 24 0. Kubaschewski The Carter Memorial Lecture March 1970 ; Metal. J. Uniu. Strathclyde 1971. 2s M. Hirabayashi S. Nagasaki and H. Maniwa Nippon Kinzoku Gakkai-si 1950,14B 1