is often available from magnetization measurements. It is not possible to make a general statement about the applicability of (9) to real systems with their complex electronic structure covering the range from delocalized band behaviour to fully localized spin systems. Taking Ni as an example, it seems more a question of defining the fraction of atoms with a given state rather than of the validity of (9). Hofman et al. [56 Hofl investigated Fe, Co, Ni, Gd, and a large number of compound systems. They used the constraint (9) successfully in their deconvolution procedure in order to obtain the Debye temperatures from an entropy matching. Since they used (9) only for determining the physical parameters of nonmagnetic contributions the deconvoluted ca as such were not subjected to this constraint. Yet the resulting AS”= values came out very close to the values expected from (9). Consequently, Inden [76Ind, 81Ind] suggested to use (9) as a constraint for determining K* ,as did Hillert and Jar1 [78Hil]. That is the reason why the calculated cy values derived from Eq. (3) do not perfectly fit the experimental data of Fe, Co and Ni (Figs 4-6). Constraint (9) has been relaxed by Chuang et al. [85Chu] in their deconvolution of c, of Fe, Co and Ni. As a result they obtained a closer fit to their deconvoluted cy data which were based on assessed data in Hultgren’s tabulation rather than on the original c,‘s. However, in view of the considerable scatter of experimental c, data from different sources, (e.g., Fig. 6) there seems to be no point to give up constraint (9) for a more close fitting of one particular set of data. The major drawback would be the loss of predictability. In view of the scarcity of experimental c, data for elements, alloys and compounds, predictability cannot be sacritied. 5.2
Magnetic enthalny
Another constraint can be formulated on the basis of theoretical models. Starting from the work of Brown and Luttinger [55Bro], Paskin [57Pas] developed an expression for the fraction of magnetic enthalpy above T, (or T,): f=-
s+l s(z - 1)
( 10)
where s is the spin number and z the coordination number. Applying this relation to the fee structure and taking spin % (applicable to Co and Ni) one obtains f=O.27. For the bee structure the values f=O.43 (spin l/2) and 0.29 (spin 1) are obtained from (10). At present, the magnetic effects are treated with f=O.28 for fee and f=O.4 bee, irrespective of spin number, as suggested by Inden [76Ind, 81IndJ. This can be improved with constraint (10). 5.3
Conclusions
No argument has been found which could impair the validity of (9). Therefore, the description of c? should be made consistent with the two constraints (9) and (10). This makes sure that
511
k-TRANSITIONS
a high level of predictability is maintained to cover the majority of cases (particularly alloys and compounds) where cPdata are not existing.
6.1
Exnerimental evidence for two y-states
It has been assumed in all the previous sections that each element has only one magnetic state pertinent to a particular crystal structure. As the fee allotrope of pure iron, ‘y, is not stable at room temperature, various techniques have been used to extract the relevant magnetic properties. These include extrapolation from alloys [63Wei], the examination of a variety of phycical properties [78(a) and (b)Ben], and measurements on very small particles where the fee phase has been retained by coherency effects. When the results from these various techniques are compared, it is evident that, at low temperatures, fee y-iron behaves predominantly like an anti-ferromagnetic material with a low magnetic moment, state yz, but there is also evidence for a much higher moment at higher temperatures, state *(I. In Re-Ni and Fe-Co, there is compelling evidence for the co-existence of two magnetic states [78h4io], and this concept has also been extended to other 3d elements [79Wei]. 6.2
Theoretical treatment of multiple states
The simplest way to describe the equilibrium between various competing states is to use a Schottky model [63Wei, 63Kau], where two states are considered, the ground state yz and an excited state Tl, with an energy difference AE between them. If (g,, g,) are the degeneracies of the two states, respectively, the fraction v of excited states is given by
JLJL,, 1-v
g,
__AE
( 1 RT
(11)
Clearly if AE is large, there is effectively only one state (and one moment). However as AE becomes smaller there can be some quite marked effects, as evidenced by Invar alloys where AE is not far from zero. The Schottky model does not lead to a h-transition per se, but several variants of the Schottky model have been developed to take into account the situation where one of the two states undergoes magnetic ordering [77Mio, 8OMio], see Fig. 14. In such cases AH -will become temperature dependent, and there is a further change in the temperature dependence of the effective magnetic moment. 6,2
Thermodvnamic consequences of multiple states
Although there is extensive circumstantial evidence for the existence of multiple magnetic states, the necessary formalism has not, as yet, been incorporated into the current software for phase diagram calculations. This is despite the fact that its inclusion has been suggested as essential to the proper description of the phase transformations in iron base alloys [63Kau, 82Ben], and the explanation of a number of thermodynamic anomalies is facilitated by this concept [78Mio]. This can be attributed to the lack of theoretical backing for the concept, particularly in the early stages of its development.
D. DE FONTAINE
512
5.4
et al.
First urincinle calculations of multinle states
This situation has however changed dramatically in recent years. First principle calculations have not only confirmed the existence of multiple states, but have also confirmed the value of the moments, Table IV, and energy gaps (see Fig. 15) inferred previously by experimental techniques [88Mor, 90Mor (a,b,c), 93Asa(a,b)].
6.5 Potential advantac s of Muding amti-state model (a) A better descriptioi of the transition between ferromagnetic and anti-ferromagnetic behaviour in iron-base alloys and more appropriate values for the magnetic properties of metastable allotropes. (b) A better description of the low temperature behaviour of iron, which impinges on the calculation of other properties at low temperatures, such as stacking fault energies and martensite transformation temperatures. (c) A potential resolution of some of the discrepancies that seem to occur between magnetic parameters extracted from c, curves and determined by magnetic measurements.
6.6
Incornoration of multinle states into comnutinrr software
One can consider adding the necessary formalism either within the existing magnetic framework or as a separate module. A separate Schottky module would have the advantage that it could also be used in other areas such as a description of the glass transition. Such a module will not need to be invoked in the default situation where only one magnetic state is involved. While separation of multiple state effects and the h-transition would necessarily be an approximation, it is adequate as an interim solution, and much simpler to implement. The general availability of a Shottky function could also have further applications in cases where there is equilibrium between two states that differ with respect to some other electronic transitions and not just for the magnetic case.
7.
Reference states
7.1
Choices of reference states
The magnetic effect has been treated as a separate effect which has to be added nonmagnetic part of c,. Therefore, the integrated functions like enthalpy, entropy, energy also take up the magnetic effect as additive terms. In principle, there should difference whether the magnetic terms are added to the ground state at OK or paramagnetic high temperature state.
to the Gibbs be no to the
In practice, the thermodynamic functions are derived from high temperature data. ,,High temperatures” is synonymous with the range of temperatures where equilibria can be attained during the experiments. The assessment work that has been performed so far has not included,
A-TRANSITIONS
at high temperatures, a magnetic entropy contribution. If then the magnetic term is added there may appear, at low temperatures, a violation of the third law. So far, this has not been considered to be a problem for the calculation of phase equilibria since these. are usually not made at such low temperatures. However, if the low temperatures are being considered in the future, e.g., to treat the thermodynamic effects of martensitic transformations, then the relialoility of the thermodynamic functions needs to be extended to low temperatures, that means into the ranges where the mentioned conflicts may appear. Similar arguments can be put forward if the magnetic term is added, starting from the ground state. The various consequences have been discussed in [91Ind]. 7.2 Conclusions It is recommended to include, in the future a magnetic entropy contribution into the high temperature description of the thermodynamic functions during the assessment of high temperature thermodynamic data.
8.1
Procedure in the absence of cP&Q
If the necessary cP data are not experimentally available, the protocol suggested for stable phas#escannot be implemented and some alternative proposals are therefore necessary. In the absence of any other information, it has been common practice to set the magnetic properties of metastable phases as either identical to the values for the stable phases, or to zero. However this is clearly liable to significant error. As first principle calculations for stable phases now yield values for the Bohr magneton numlber p that are close to the values obtained by experiment, increasing confidence is given to the values calculated for metastable structures. As Debye temperatures can also be derived by the same route, this would mean that there can be significant input into leading terms in the generalexpression cP =cy +f(B,)+aT+bT’ from first princple calculations, even in the absence of experimental cP information. However, there is as yet no theory for calculating the critical temperature for magnetic disordering, so empirical methods have still to be used to obtain an estimation from the Bohr mag,neton number WTau]. It is therefore unavoidable that there will be more divergence between starting values entered from first principle calculations and the final values that may emerge from the optimization process. Nonetheless it is considered this is a better option than the :previously mentioned tendency to set values either equal to those of the stable allotrope or to zero.
Table IV combines data for both stable and metastable phases as well as data relating to altelmative magnetic states for a given phase. The occurrence of multiple states in practice will depend on the energy gap between the states, which can be estimated from [Fig. 151 for some
513
D. DE FONTAINE eta/.
of the cases (a more exhaustive treatment is considered to be beyond the remit of this report). However, several interesting features emerge from the data compiled in this table. 1. It is crystal crystal should
clear that the calculated J3values for metastable phases are generally different for each structure and vary between ferro- and anti-ferromagnetic configurations in a given structure. The assumption of identical values for stable and metastable allotropes therefore be considered the exception rather than the rule in most circumstances.
2. The values of l3 estimated indirectly by Weiss have been confirmed to a surprising degree by first principle calculations. The combined picture also confirms previously suggested systematic trends which should help reliable interpolation (Fig. 16). 3. There is a significant difference between the 0 values that have been extracted from the deconvolution of cP measurements and the values derived independently from magnetic measurements. There is also a significant spread between the 0 values obtained by using different methods to extract the magnetic specific heat capacity [64Lyt, SlMes, 85Chu] (as already mentioned in section 3). The earlier compilations of 13values to be used within the suggested formalism [87Chi] were in line with experimental I3 values and did not make the value of 13an adjustable parameter in the optimization process. 4. Some of these differences can be rationalised by the effect of temperature on mixed magnetic states, as most of the data in Table IV essentially refers to zero K. This is almost certainly the case for Cr and Mn [72Wei, 79Wei, 79Ben]. In the case of Co, the hexagonal phase is a close competitor near the Curie temperature and high stacking fault densities may therefore also lead the unusual effects [95Mio]. Other differences may arise because of the effects of magnetic forces on the Debye temperature [74Woh, 64Lyt]. Another possible source of error is the assumption that there are no orbital contributions, but the latter appear to be very small [9OEri]. 5. Although global optimization is recommended, the consistency of Table IV confirms that theoretical B values should be given a substantial weighting in any optimization procedure.
9. Overall conclusions The following procedure is suggested for the determination of ca
including the global
optimization of c,: l Equation (12) is used to describe the total cP in accordance with the most recent recommendation (see report of Group 1): cP =f(B,)+aT+bT’+cpmaS l
l
+c”
(12)
In order to comply with the high temperature c,-data of some particular cases Group 1 suggested to allow for a temperature dependence of f&,.The two Debye temperature approach discussed here (see Eq. (8)) should also be considered as an alternative. The first three terms in (12) are nonmagnetic contributions and require three parameters 8,, a, and b to be fitted in an optimization. The preferred equation for the magnetic term is the extended version of Eq. (3) as already suggested in section 4 (at least four terms, see appendix A). T, is fixed from
A-TRANSITIONS
l
experiment. According to the present recommendation not only the parameters K- and K+, but also m and n should be allowed to vary and thus be included in the optimization scheme. This optimization can be made with the contraints (A-9) and (A-10) (see appendix A) (P>+2 ~~ This ensures (P).(z-1) * that independent magnetic properties like (p) are included in the optimization and can imposed by the relations ASW(exp) = Rln((P)+ 1) and f =
l
act as a constraint against arbitrary changes. c” should be set to zero as the fist approximation.
The optimization may then have to be repeated with different weightings to obtain a meaningful overall picture, as is the standard practice in phase diagram optimization. The description of the magnetic effect has to be extended to include the effect of high spin numbers.
515
516
D. DE FONTAINE
et al.
10. References 42sto: SIT& 52Bus: 52Vle: 54Gri: 55Bro: 55Cat: 55Tau: 55Tom:
J.W. Stout and H.E. Adams, J. Amer. Chem. Sot. 64 (1942) 1535 S.S. Todd and K.R. Bonnickson, J. Amer. Chem. Sot. 73 (195 1) 3894 R.H. Busey and W.F. Giauque, J. Amer. Chem. Sot. 74 (1952) 4443 J.H. Van Vleck, Rev. Mod. Phys. 25 (1952) 221 M. Griffel, R.E. Skochdopole and F.H. Spedding, Phys. Rev. 93 (1954) 657 H.A. Brown and J.M. Luttinger, Phys. Rev. 100 (1955) 685 E. Catalano and J.W. Stout, J. Chem. Phys. 23 (1955) 1284 K.J. Tauer and R.J. Weiss, Phys. Rev. 100 (1955) 1223 J.R. Tomlinson, L. Domash, R.G. Hay and C.W. Montgomery, J. Amer. Chem. sot. 77 (1955) 909 55Zen: C. Zener, Trans. AIME 203 (1955) 619 56Boz: R.M. Bozorth in Ferromagnetism, van Nostrand (1956) J.A. Hofman, A. Paskin, K.J. Tauer and R.J. Weiss, J. Phys. Chem. Solids 1 56Hof: (1956) 45 R.J. Weiss and K.J. Tauer in: Theory of Alloys Phases, ASM Symposium, pub. 56Wei: ASM (1956) 290 57Pas: A. Paskin, J. Phys. Chem. Solids 2 (1957) 232 58Wei: R.J. Weiss and K.J. Tauer, J. Phys. Chem. Solids 4 (1958) 135-143 6OGoo: J.B. Goodenough, Phys. Rev. 120 (1960) 67 62Shi: G. Shirane and W.J. Takei, J. Phys. Sot. Japan 17 (Suppl. B-III) (1962) 35 63Kau: L. Kaufman, E. Clougherty and R.J. Weiss, Acta Met. 11 (1963) 323 63Wei: R.J. Weiss, Proc. Phys. Sot. 82 (1963) 281 64Bra: M. Braun, Dissertation, Universitlt Kiiln, 1964 64Lyt: J.L. Lytton, J. Appl. Phys. 35 (1964) 2397-2406 65Bra: M. Braun and R. Kohlhaas, phys. stat. sol. 12 (1965) 429 67Hil: M. Hillert, T. Wada and H. Wada, J. Iron Steel Inst. 205 (1967) 539 72Wei: R.J. Weiss, Phil. Mag. 26 (1972) 261 74Woh: E.P. Wohlfarth, Phys. Stat. Sol. (a) 25 (1974) 285-291 75Gri: G. Grimvall and I. Ebbsjo, Physica Scripta 12 (1975) 168 75Ind(a): G. Inden, Z. Metallk. 66 (1977) 725 75Ind(b): G. Inden and W.O. Mayer, Z. Metallk. 66 (1975) 725 76Gri: G. Grimvall, Physica Scripta 14 (1976) 63 761nd: G. Inden, Proc. Project Meeting CALPHAD V, Dusseldorf 1976, p. IIIA-1 77Ben: W. Bendick, H.H. Ettwig, F. Richter and W. Pepperhoff, Z. Metallk. 68 (1977) 103-107 77Mio: A.P. Miodownik, Calphad 1(1977) 133-158 77Roy: D.M. Roy and D.G. Pettifor, J. Phys. F 7 (1977) 183 78Ben(a): W. Bendick, H.H. Ettwig and W. Pepperhoff, J. Phys. F 8 (1978) 2525 78Ben@): W. Bendick and W. Pepperhoff, J. Phys. F 8 (1978) 2525 78Chi: S. Chikazumi and M. Matsui, J. Phys. Sot. Japan 45 (1978) 458 78Hil: M. Hillert and M. Jarl, Calphad 2 (1978) 227 78Mio: A.P. Miodownik, The concept of 2 gamma states in: Physics and Application of Znvar Alloys, Ed. Saito, Vol. 3 Maruzen (1978) pp. 288-309 79Ben: W. Bendick and W. Pepperhoff, J. Phys. F 9 (1979) 2 185 79Nis: T. Nishizawa, M. Hasebe and M. Ko, Acta Metall. 27 (1979) 817 79Wei: R.J. Weiss, Phil. Mag. B 40 (1979) 425-428
A-TRANSITIONS
79Wil: 80Leg: 80Mio: 811nd: 8 1Mes: 82Ben: 83Ros: 85Chu: 85Sun: 87Chi: 88Mor: 90Dav:
90Eri(a): 90Eri(b): 90Mor(a): 90Mor(b): 90Mor(c): 91Din: 91Ind: 93P,sa(a): 93&a(b): 94Fri: 95hIio: 951nd: 95Sch:
I.S. Williams, E.S.R. Gopal and R. Street, J. Phys. F 9 (1979) 431 S. Legvold in Ferromagnetic Materials, Vol. 1, E.P. Wohlfarth (Ed.), NorthHolland Publ., Amsterdam 1980, p. 184-295 A.P. Miodownik and M. Hillert, Calphad 4 (1980) 143 G. Inden, Physica B 103 (1981) 82-100 P.J. Meschter, J.W. Wright, C.R. Brooks and T.G. Kollie, J. Phys. Chem. Sol. 42 (1981) 861-871 W. Bendick and W. Pepperhoff, Acta Met. 30 (1982) 679 J. Rosen and G. Grimvall, Phys. Rev. B 27 (1983) 7199 Y.Y. Chuang, R. Schmid and Y.A. Chang, Met. Trans. 16A (1985) 153-165 B. Sundman, B. Jansson and J.-O. Andersson, The THERMO-CALC databank system, CALPHAD 9 (1985) 153-190 Chan-peng Chin, S. Hertzman and B. Sundman, TRITA-MAC 0203, revised Aug. 1987 V.L. Moruzzi et al., Phys.Rev. B 38 (1988) 1613-1520 R.H. Davies, A.T. Dinsdale, S.M. Hodson, J.A. Gisby, N.J. Pugh, T.I. Barry and T.G. Chart, MTDATA-The NPL Databank for Metallurgical Chemistry, in Proc. Conf. User Aspects ofPhase Diagrams, F.H. Hayes (Ed.), Institute of Materials, London 1990, p. 140-152 0. Eriksson et al., Phys. Rev. B 42 (1990) 2707-2709 G. Eriksson and K. Hack, Met. Trans. 21B (1990) 1013-1023 V.L. Moruzzi, Phys. Rev. B 41(1990) 6939-6946 V. L. Moruzzi et al., Phys. Rev. B 43 (1990) 8361-66 E.G. Moroni et al., Phys. Rev. B 41(1990) 9600-02 A.T. Dinsdale, SGTE Data for Elements, CALPHAD 15 (1991) 317 G. Inden, Stand. J. Metall. 20 (1991) 112 T.Asada, K. Terakura, in: Computer Aided Innovation of New Materials, M. Doyama et al. (Eds), Elsevier Publ., Vol. I (1993) 169-172 T. Asada, private communication to A.P. Miodownik K. Frisk and C. Qiu, Z. Metallk. 85 (1994) 60-69 A.P. Miodownik, unpublished work (in preparation) G. Inden, unpublished work (in preparation) C.G. Schon and G. Inden, submitted to Scripta Met.
517
D. DE FONTAINE
11.
et al.
Aonendices
The thermodynamic functions are given with reference to the complete paramagnetic state for which T== stands. Definitions: R=gas constant T=$
(or +) N
C
1 A=L+p+-+-+... l+m 3(1+3m) 1 B=L+-+-+-+... l-n 3(1-3n)
D=i
n(
1 5(1+5m) 1 5(1-5n)
1 7(1+7m) 1 7(1-7n)
l+$+&+-&+...
ai =2.i-1
11.1 Anpendix A Magnetic contributions to thermodynamic functions based on Eq. (3)
c, =2.K-.R.
=-2.K-
1 1 1 z~+--z~“‘+-~~‘“+-~~~+... 3 5 7
.R.‘&.
T1+Sll?
A-<-z_ l+m
+2.K+.R.T,
(A-l)
3(1+3m)
----...
5(1+5m)
T
*+7m
7(1+7m)
(A-2)
.B
Smag(T)-Srrug(=)= C_~_??-~-r?m-.~. m 9m 25m
(A-3) 49m
A-TRANSITIONS
519
Gw(T)-Gw(=)=Hw(T)-T.Sw(T)-(HMg(m)-T.S~(=+ =-2.R.Tc _2.K-
.(K- .A-K+ .R.Tc .i
.D)q(A-4)
a,*m(l+ aim)
1 ~-n+T~-3"+;+'
=:z.K+.R.
.(K- .C-K+
Z’+a’m
id
c;
.B)+2.R.T,
+1-T-7n+
(A-3
...
7
P
=z.K+.R.‘Q.
-+-+-+-+...
l-n
z I-3” 3(1-3n)
Smg(T)-S*(w)=-2.R.K+.
G’“‘~(T)-Gmg(m)=2.K+
z I-5n 5(1-h)
z l-70 7(1-7n)
n
.R.‘&
(A-7)
“-”
.i i=l
q*n(l - cqn)
K+.B
f=
(A-6)
K+.B-K-.A
(A-8)
(A-9)
(A-10)
J-LL&wndix
E
Magnetic contributions to thermodynamic functions based on Eq. (4)
211 -_ C, :=
R.k-.~.exp(-p(l-7))
(B-l)
H”@(T)-Hmg(=)=
S”“(T)-S’(~)=~exp(-p(l_~))-R. P
(B-3)
520
D. DE FONTAINE
eta/.
Gw (T) - Gmg (w) =
ci = R.k* sTmexp(q(l-T))
(B-5)
exp(q(1 - 2)) Gw(T)-Gm(-)=-k”q~‘Tc
SW(T) - SUE
= -y
exp(q(l-7))
e.xp(q(l - 2))
( I f=;(l-::.p*)t4&.l(l+;)
(B-6)
(B-7)
(B-8)
k’ I+’
ln((p)+l)=;+$-$e-.
(B-9)
(B-10)
521
A-TRANSITIONS
12.
Tables
Table I Integrated quantities below (-) and above (+) T, as obtained from treatments (2) and (3) m
Eq.#
AI-I-
AS-
(2) (3) (2) (3) (2) (3)
RTc 1.233 1.133 0.778 0.728 0.571 0.538
R 2.199 2.093 1.097 1.045 0.729 0.696
-
-
1 1 2 2 3 3
Parameters used for calculation of cy
Eq.#
AH’
AS+
(2) (3) (2) (3) (2) (3)
RTc 0.346 0.333 0.235 0.226 0.178 0.170
R 0.245 0.232 0.183 0.174 0.147 0.139
Table II with Eq. (3) in accordance with the SGTE database [9 lDin]
Table III Debye temperatures and spin values of the elements in various compounds [56HofJ %mpound - WL MnF, NiF, NiCl, NiO MnO Fe0 -
0R,/K
e;rK
P&n
T,
Structure
240 306 325 425 350 450
625 700 390 900 700 600
5 2 2 2 5 4
66.5 73 53 523 117.8 191
SnO, SnO, CdCl, NaCl NaCl NaCl
D. DE FONTAINE
522
et al.
Table IV Comparison of magnetic moments for selected elements Data from SGTE (Scientific Group Thermodata Europe), first principle calculations (FPC) anti-ferromagn. 8 in & Element Cr Cr Cr Cr Cr Cr Cr Mn Mn Mn Mn Mn Mn Fe Fe Fe Fe Fe Fe Fe Fe co
co co co co co co
co Ni Ni Ni Ni Ni Ni Ni Ni Ni
Source J87Chi 91Din (SGTE) 93Asa(b) (FPC) 88Mor, 90Mor(a)-(c) 79Wei (EXP magn.) (EXP cp ) 87Chi 91Din (SGTE) 93Asa(a)-(b) (FPC) 88Mor, 90Mor(a)-(c) 79Wei 1(EXP mag) /87&i 91Din (SGTE) 93Asa(a) (FPC) 88Mor, 90Mor(a)-(c) 79Wei (EXP mag) 85Chu I64Lyt I87Chi 91Din (SGTE) 93Asa(b) (FPC) 88Mor, 90Mor(a)-(c) 78Mio (estim.) (EXP mag) 85Chu 164Lyt I87Chi
88Mor, 90Mor(a)-(c)
BCC
FCC
0 0
0
1.0 0.9
Of4
BCC 0.4 0.008 010.5 010.6 010.7 0.4 O.O+
FCC 0.82 0.82 013.0
CPH 0 0 0
-
-
0
0.09
2.0
2.76
0.62 2.25
0 0.15
2.410 2.3 0.70 0.70 1.21 0.510 0.40 0.70 0.57
0
0.20 -
0
1.2
CPH 0 0 0
4.5 (1) 2.22 2.22 2.32 2.2 2.22 2.05 1.03 1.80 1.35 1.80 1.70 1.70 2.00
0.85 0.84 0.50 0.30 0.85
(1) 2.56 2.7 2.6 -
0 0 2.56
1.70 1.35 1.70 1.80 1.80 1.80
1.70 1.35 1.61
0.62 0.52 0.66
0.62 0.25 0.58 0.53
1.55 1.75
-
1.0
0 0 0
0
1.70 0.89 1.21 0.52 0.60
-
0
0.1
0.25
523
h-TRANSITIONS
Magnetic specific heat capacity versus reduced temperature according to Eq. (2) for different values of exponents m and n using K- = 0.89486, and K’ = 0.29830. These numerical values have been selected in order to provide a given step in cy at T, in the treatments (3) and (4) of Figs 2-3. Fig. 1a linear representation IFig. lb double-log representation
Comparison between Eq. (2) (solid lines) and (3) (broken lines) in linear representation for various values of m and n. The coefficients K* are given the same values as in Fig. 1 so that Eq. (3) yields the limiting values c;(Tc) = 3R and ci(Tc) =R. The series expansion (3) has been extended to include a fourth term. !Fig. 2a low temperature branch IFig. 2b high temperature branch
Comparison between Eq.(3) and Eq. (4) for the same situation as in Fig. 2, i.e., with limiting values c;(Tc) =3R and ci(Tc) =R, K* as in Fig. 1 and k-=3, k’=l. The exponents used are m=3, n=3 in Eq. (3) and p=4.25, q=3.8 in Eq. (4). IFig. 3a linear representation :Fig. 3b double-log representation
Magnetic specific heat capacity of bee-Fe. Deconvoluted
cy
values from [64Bra].
Solid line: calculated with Eq. (3) using the parameters in Table II. IFig. 4a. linear representation. IFig. 4b double-log representation.
Magnetic specific heat capacity of feeCo.
Deconvoluted
CT values from [64Bra].
Solid line: calculated with Eq. (3) using the paramaters in Table II. Fig. 5a linear representation. Fig. Sb double-log representation.
Magnetic specific heat capacity of fee-Ni. Deconvoluted cy values: l [64Bra], A [81Mes]. Solid line: calculated with Eq. (3) using the parameters in Table II. .Fig. 6a. linear representation IFig. 6b double-log representation
D. DE FONTAINE et al.
524
Magnetic specific heat capacity of Gd. Deconvoluted cy
values from [54Grifl.
Fig. 7a linear representation Fig. 7b double-log representation
Magnetic specific heat capacity of beeCr. Experimental total c,, -values from [79Wil].
Magnetic specific heat capacity per mole (formula unit) of MnO and NiO. Deconvoluted cy -values [95Ind] derived from [SlTod] (MnO) and [55Tom] (NiO) using the Debye temperatures in Table III. Fiy 1Q Magnetic specific heat capacity per mole (formula unit) of Fe,,O. Deconvoluted cy -values [95Ind] derived from [SlTod] using the Debye temperatures in Table III. Fig. 11 Magnetic specific heat capacity per mole (formula unit) of NiF, and MnF,. Deconvoluted c,” -values [95Ind] derived from [55Cat] (NiF,) and [42Sto] (MnF,) using the Debye temperatures in Table III. Fig. 12 Magnetic specific heat capacity per mole formular unit of NiCl, Deconvoluted c,“-values [95Ind] derived from [52Bus] using the Debye temperatures in Table III. Fig. 13 Magnetic specific heat capacity of MnO plotted in a linear representation. Data as in Fig. 9. Fig. 14 Relative position of energy levels according to the models of Weiss [63Wei], Hillert [8OMio], Chikazumi [78Chi] and Pepperhoff [78Ben(a)]. Fig. 15 Variation of the energy difference between competing states with electron concentration.
Variation of the magnetic moment 8 with crystal structure and position in the periodic table. (FM: ferromagn., AF: antiferromagn., MAG: from magnetic measurements, Cp: from specific heat capacity)
A-TRANSITIONS
Magnetic specific heat capacity versus reduced temperature according to Eq. (2) for different values of exponents m and n using K- = 0.89486, and K’ = 0.29830. These numerical values have been selected in order to provide a given step in c w p at T, in the treatments (3) and (4) o:I Figs 2-3. Fig. la linear representation Fig. lb double-log representation
525
D. DE FONTAINE
526
et al.
4-
a)
p! a VP
r ‘T/T=
T=T/T
C
Comparison between Eq. (2) (solid lines) and (3) (broken lines) in linear representation for various values of m and n. The coefficients K’ are given the same values as in Fig. 1 so that IQ. (3) yields the limiting values c;(Tc) = 3R and ci(T,-) =R. The series expansion (3) has been extended to include a fourth term. Fig. 2a low temperature branch Fig. 2b high temperature branch
527
A-TRANSITIONS
a.)
-~~ 2
0
r=T/Tc
Cotnparison between Eq.(3) and Eq. (4) for the same situation as in Fig. 2, i.e., with limiting values $(T,)=3R and ci(Tc) =R, K* as in Fig. 1 and k-=3, k+=l. The exponents used are m=3, n=3 in Eq. (3) and p=4.25, q=3.8 in Eq. (4). Fig. 3a linear representation Fig. 3b double-log representation
D. DE FONTAINE
528
et al.
b)
r=T/Tc
Fig. Magnetic specific heat capacity of bee-Fe. Deconvoluted c,“” values from [64Bra]. Solid line: calculated with Eq. (3) using the parameters in Table U. Fig. 4a. linear representation. Fig. 4b double-log representation.
A-TRANSITIONS
1
a)
0
b)
loo t
0.1
co
;
1
I
10
T’T/Tc
Magnetic specific heat capacity of fee-Co. Deconvoluted
cy
values from [64Bra].
Solid line: calculated with Eq: (3) using the paramaters in Table II. Fig. 5a linear representation. Fig. 5b double-log representation.
D. DE FONTAINE
530
et al.
0
7=T/T
C
b)
1 7=T/Tc
Magnetic specific heat capacity of fee-Ni. Deconvoluted cPW values: l [64Bra], A [SlMesJ. Solid line: calculated with Eq. (3) using the parameters in Table II. Fig. 6a. linear representation Fig. 6b double-log representation
531
A-TRANSITIONS
a)
8
Gd
i
Tc=BOK
i
r=T/T
C
10
Gd
P 3 E .=1.
i..
1
.
H o=
.
.
. ,
0.1
0.1
I
1
10
r =T/Tc
Magnetic specific heat capacity of Gd. Deconvoluted c,” values from [54Grifl. Fig. 7a linear representation Fig. 7b double-log representation
532
D. DE FONTAINE
et a/.
35
Cr TN=311K
. 30 P 5 E zd
0 0
O-25
20 0.6
1.4
1
r=T/T
N
Magnetic specific heat capacity of bee-Cr. Experimental total cP -values from [79Wil].
0.1
1
10
r=T/TN
Magnetic specific heat capacity per mole (formula unit) of MnO and NiO. Dcconvoluted c? -values [95Ind] derived from [51Tod] (MnO) and [STom] (NiO) using the Debye temperatures in Table III.
A-TRANSITIONS
533
Magnetic specific heat capacity per mole (formula unit) of Fe,,,,O. Deconvoluted c,“‘-values [95Ind] derived from [SlTod] using the Debye temperatures in TabIe III.
loo L
0.1
NiF2 1 : l ......._......_...... -I._._......_.__.._._,.__........,..,..................... TN=73.2K;
0.01 0.1
1 7 =
a
10
T/TN
Fig. 11 Magnetic specific heat capacity per mole (formula unit) of NiF, and MnF,. Deconvoluted c,” -values [951nd) derived from [55Cat] (NiF,) and [42Sto] (MnFJ using the Debye temperatures in Table III.
534
D. DE FONTAINE et al.
10 l-
NiClp a 0
rN=53K
l
B
8
?
0
‘ij
E
8
8
2 P
8
6a 0
l
8
0.1
10
1 r=T/TN.
EiLK Magnetic specific heat capacity per mole formular unit of NiCl, Deconvoluted cy -values [95Ind] derived from [52Bus] using the Debye temperatures in Table III.
0
1
r=T/T
2 N
Magnetic specific heat capacity of MnO plotted in a linear representation. Data as in Fig. 9.
A-TRANSITIONS
535
HILLERT
WEISS
-PEPPERHOFF
Relative position of energy levels according to the models of Weiss [63Wei], Hillert [80Mi0], Chikazumi [78Chi] and Pepperhoff [78Ben(a)].
Energy dlfterence between Two Gamma States 20
- FM Ground State 10 ‘; B E 3
..= 0.
i$ (10) -
lnvar Composition
C Q (20)
AF Ground State
-
I @Ok.5
&
7
,
I 7.5
8
,
,
8.5
9
electron-atom 63wEI
78MlO
WpI$~efl “*BI’”
77SEN
“p.“”
,
,
,
9.5
10
ratio
&EN(b)
7SSEN(J
QOMOR(b)
Elac~M
Sp.$sd
F,’
BOMOR(O)
%?A.%
1
‘.’
Variation of the energy difference between competing states with electron concentration.
536
D. DE FONTAINE
et al.
Variation of Magnetic Moment in the periodic table
0
V
CR
MN
FE
co
NI
CU
Element CPH
Fip. 16 variation of the magnetic moment I3with crystal structure and position in the periodic tat&,