American Mineralogist, Volume 81, pages 973-981, 1996

Heat capacity and thermodynamic properties of nearly stoichiometric

wiistite from 13 to 450 K
SVEIN ST0LEN,1 RONNY GLOcKNER,l FREDRIK GR0NVOLD,1 TOORU ATAKE,2 AND
SA TORU IZUMISA WA 2
1
Department of Chemistry, University of Oslo, P.O. Box 1033, Blindem, N-0315 Oslo, Norway
2Research Laboratory of Engineering Materials, Tokyo Institute of Technology,
4259 Nagatsuta-cho, Midori-ku, Yokohama, 226 Japan

ABSTRACT
The heat capacity of a three-phase sample ofFeo.99o:to.oosO,Fe304, and Fe (mole fractions
0.915, 0.078, and 0.007, respectively) has been measured by adiabatic shield calorimetry
at temperatures from 13 to 450 K. FeO.990 and magnetite are formed as a metastable
intermediate on heating of a quenched nonstoichiometric wiistite with composition FeO.93740.
The small amount of Fe present stems from the second disproportionation reaction, in
which the stable two-phase mixture of Fe and magnetite is formed from FeO.990.The value
of ~Sm (FeO.990' 298.15 K), 60.45 J/(K. mol), is derived from the entropy of the three-
phase sample and recommended standard entropies of Fe and magnetite. The character
of the magnetic order-disorder transition changes with composition and is strongly co-
operative in FeO.990' with TN ~ 191 K. A minor, seemingly higher order transition is
observed at '"- 124 K. It is caused by the Verwey transition in the magnetite. This mag-
netite, formed in metastable equilibrium with FeO.990'is presumably more Fe-rich than
magnetite in stable equilibrium with Fe.

INTRODUCTION 10 GPa and ~600 K (Fei and Saxena 1986). For higher
pressures, our knowledge about the stability field of wiis-
The typically non stoichiometric iron monoxide, wiis- tite is uncertain. On the Fe-rich side increasing nonsto-
tite, probably participates in two reactions in the Earth's ichiometry ofwiistite is observed as pressure is increased
interior that are of geologic significance. In the boundary (cf. Simons and Seifert 1979; Shen et al. 1983; Mc-
between the upper and lower mantle (T ~ 2750 K, P ~ Cammon and Liu 1984; McCammon 1993). At atmo-
24 GPa) magnesiowiistite, (Mg,Fe)O, and a magnesium- spheric pressure the O-rich phase boundary of Fe1_xO
iron perovskite, (Mg,Fe)Si03, are formed by dispropor- moves toward higher 0 content as temperature is in-
tionation of the main component of the upper mantle, creased.The limiting value of 1 - x is '"-0.83 at '"-1700
(Mg,Fe)2Si04 (Liu 1975). Wiistite is presumably also K. According to Shen et al. (1983) the trend with pressure
formed in the so-called D" layer between the lower man- is probably the same as for the Fe-rich phase boundary,
tle and the inner liquid Fe core (T ::::: 3500 K, P :::::136 i.e., at a given intermediate temperature the O-rich phase
GPa). The presence of liquid iron diminishes the solu- limit first moves in the direction of Fe, and then back as
bility of Fe in the perovskite-type silicate, resulting in the pressure is increased further. The high-pressure phase
formation of wiistite (Knittle and Jeanloz 1989, 1991; boundaries ofwiistite, as derived from a thermodynamic
J eanloz 1990). description of the Fe-O system, are seen to depend mark-
Although the compositional stability field of the grossly edly on assumptions regarding the compositional behav-
non stoichiometric wiistite, Fe1-xO, is sensitive to' both ior of the bulk modulus ofwiistite (Stelen and Grenvold,
temperature and pressure, details of the pressure-induced unpublished manuscript).
changes to the stability field are much disputed. Wiistite To derive the high-pressure properties and to model
with composition Fe_O.940 forms eutectoidally from Fe multicomponent phase diagrams involving wiistite, a
and Fe304, magnetite, at --835 K at ambient pressure. thermodynamic description of the wiistite phase is need-
The main effects of increased pressure up to '"-10 GPa ed that is also correct beyond the phase boundaries at
are a decrease of the eutectoid temperature for wiistite ambient pressure. Two of the more recent evaluators of
formation and further extension of the wiistite phase field. the voluminous literature on the thermodynamics of the
Most reports conclude that the Fe-rich phase boundary Fe-O system differ with regard to the properties of the
of the Fe1-xO phase approaches stoichiometry when pres- non stoichiometric wiistite phase, especially for the meta-
sure is increased to about 10 GPa: 1 - x increases from stable region near 1 - x = 0.500 (Sundman 1991; Haas
0.954 at ambient pressure and 1185 K to --0.99 at and Hemingway 1992). The compositional variation of
0003-004 X/96/0708-097 3$05 .00 973
974 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE

the enthalpy of formation at 1075 K is shown together mate was of less value because the enthalpy of the tran-
with experimental and estimated values in Figure 9 of sition depends decisively on the stoichiometry of the
Gr0nvold et ai. (1993). In the same paper, molar heat phase.
capacities of stoichiometric FeO and of Feo.900 were es-
timated in the temperature range 300-1000 K from the EXPERIMENTAL METHODS
results for Feo.9470 (Coughlin et ai. 1951; Todd and Bon- Sample preparation and characterization
nickson 1951) and Feo.93790 and Feo.92540 (Gr0nvold et
The nonstoichiometric wiistite used in this investiga-
ai. 1993). For compositions approaching stoichiometry,
tion was prepared from ferric oxide and Fe. The Fe203
the molar heat capacities in the 200-500 K region appear
(E. Merck no. 3294) was heated in alumina boats in an
to be lower than for the more Fe-deficient samples,
electric furnace at 1070 K until constant mass was at-
whereas the opposite, normal trend prevails at higher
tained. This required about 40 h and resulted in a mass
temperatures, 900-1000 K (Gr0nvold et al. 1993). This
loss of 0.05%. According to the manufacturer's analysis
peculiarity was related to the more cooperative nature of
the impurity limits are (in percent by mass): CI- 0.01;
the magnetic order-disorder process when stoichiometry
SO~- 0.01; N, Pb, Cu, Mg, Mn, Ni, Zn 0.005; insoluble
was approached, an assumption that was qualitatively
in HCl 0.01. Spectrographic analysis also showed 50 ppm
confirmed in a recent study of the decomposition of
Si02. A small part of the sample was reduced to pure
quenched wiistite (St01en et ai. 1995). On the basis of the
metallic iron with the use of dry H2 gas at 1120 K until
chosen Gibbs energy of formation and the estimated heat
the Fe203/Fe mass ratio was 1.4295 :t 0.0002, compa-
capacity, the standard entropy of stoichiometric FeO at
rable to the theoretical value of 1.4297. Mixtures of this
298.15 K was estimated to be 61 :t 1 J/(K. mol) (Gr0n-
Fe and the Fe203 in the nominal Fe/O ratio 0.94 were
voId et ai. 1993). Haas and Hemingway (1992) derived a
4% larger value, 63.46 :t 0.71 J/(K. mol) in his extensive heated in evacuated and sealed vitreous silica tubes at
1270 K for 2 d and cooled in the furnace. Fifteen smaller
thermodynamic analysis of the Fe-Si-O system, whereas
samples with a total weight of 72 g were made. After
Fei and Saxena (1986) gave a much lower value, 57.59
crushing, the sam pIes were reheated in evacuated vi tre-
:t 0.20 J/(K. mol).
ous silica tubes to 950 K for 3 d and quenched by being
Nearly stoichiometric iron monoxide can be prepared
dropped into a solution of water, ice, and salt.
as a metastable intermediate in a two-stage dispropor-
The 0 content of the calorimetric sample was deter-
tionation of quenched metastable wiistite (Castelliz et ai.
mined by oxidation or reduction of 1.0-1.5 g samples at
1954; St01en et ai. 1995). In the first stage an Fe-rich
1070 K in air or H2 to constant mass in alumina boats.
wiistite forms together with magnetite at '" 450 K and
The resulting 0 content of the samples corresponds to
remains metastable to '"
550 K. At this temperature the
stable two-phase mixture of Fe and magnetite slowly Feo.93740.For further details see Gr0nvold et ai. (1993).
Room-temperature X-ray diffraction patterns were
forms.
taken with the Guinier-Hagg technique, CrKal radiation,
In the present study, the heat capacity of a three-phase
and Si as the internal calibration substance (Deslattes and
mixture of FeO.990:tO.0050 ("FeO.990"), Fe304, and a small
Henins 1973). Magnetic measurements were performed
amount of Fe was determined over the temperature re-
with a SQUID quantum-design magnetic measurement
gion 13-450 K. The sample was prepared through con-
system. The specific magnetic moments were calibrated
trolled disproportionation of a sample with composition
using a standard platinum reference sample from NIST
FeO.93740 in a step-wise-heated adiabatic calorimeter. (NBS). The accuracy of the deduced specific moment is
Careful monitoring of temperature-drift rates in the
within 0.1 %.
equilibration periods ensured optimum sample quality
(see below). The composition of the Fe-rich intermediate
and the ratio of the phases in the sample were derived Calorimetry
from (1) the known variation of the lattice constant of Details of the construction and measurement proce-
wUstite with composition, (2) determination of the amount dures for the laboratory-made, low-temperature calorim-
of magnetite in the sample through accurate determina- eter have been reported earlier (Atake et al. 1990). The
tions of the specific magnetic moment (the Fe content of mass of the sample loaded in the calorimeter was 22.8612
the sample was also obtained from the compositions of g or 0.33448 moles of FeO.93740. The contribution of the
the originally quenched and of the intermediately formed sample to the total heat capacity of the calorimeter was
wiistite), and (3) refinement of the mole fractions of the about 25, 50, and 60% at 20, 100, and 300 K, respec-
different phases by Rietveld-type analysis of neutron dif- tively. Temperatures were measured with a platinum re-
fraction results (Fjellv~g et aI., unpublished data). The sistance thermometer calibrated at the National Physical
thermodynamic properties of FeO.990 are evaluated and Laboratory, U.K., on the basis of ITS-90.
discussed. The high-temperature calorimetric apparatus and mea-
Magnetite is formed as a by-product in the first dispro- suring technique have been described earlier (Gr0nvold
portionation step. Initially the enthalpy of the Verwey 1967, 1993). The calorimeter was step-wise heated and
transition was believed to give a third estimate of the surrounded by electrically heated, electronically con-
amount of magnetite contained in the sample. This esti- trolled adiabatic shields. The sample was enclosed in an
 ST0LEN ET AL.: HEAT CAPACITY OF WOSTITE 975

TABLE 1. Molar heat capacity of a sample with overall composition FeO.93740

T Cp,m T Cp,m T Cp,m T Cp,m T CP,m T CP,m

(K) [J/(K'mol)] (K) [J/(K'mol)] (K) [J/(K. mol)] (K) [J/(K'mol)] (K) [J/(K' mol)] (K) [J/(K' mol)]

Series 1* 120.766 29.45 224.848 41.55 75.532 13.14 159.384 38.83 248.902 42.75
313.02 49.95 122.938 31.82 227.130 41.64 77.618 13.77 161.044 39.21 251.221 42.86
319.80 50.11 125.111 31.60 229.405 41.75 79.629 14.38 162.689 39.74 253.533 42.95
326.55 50.13 127.359 30.56 231.673 41.85 81.574 14.96 164.316 40.39 255.875 43.09
333.26 50.29 129.685 30.28 233.934 41.97 83.459 15.53 165.924 41.08 258.247 43.21
339.96 50.37 132.030 30.59 236.188 42.07 85.290 16.09 167.512 41.84 260.611 43.33
346.66 50.28 134.287 31.16 238.435 42.19 87.070 16.63 169.122 42.67 262.968 43.44
353.35 50.24 136.452 31.79 240.675 42.31 88.806 17.16 170.794 43.60 265.319 43.56
359.98 50.69 138.640 32.49 242.908 42.43 90.501 17.67 172.540 44.68 267.663 43.70
366.60 50.84 140.850 33.23 245.134 42.44 92.159 18.18 174.334 45.94 270.000 43.80
373.22 51.08 143.023 34.00 247.354 42.65 93.781 18.67 176.117 47.38 272.331 43.92
380.63 50.86 145.161 34.81 249.567 42.78 95.370 19.15 177.860 49.02 274.655 44.03
392.82 51.01 147.262 35.64 96.929 19.63 179.561 50.95 276.974 44.14
401 .23 51.14 149.328 36.49 Series IVt 98.458 20.10 181.214 53.23 279.285 44.24
409.69 51.23 151.359 37.33 11.441 0.093 99.960 20.57 182.836 55.98 281.591 44.35
418.19 51.32 153.359 38.07 12.445 0.085 101.437 21.02 184.423 59.01 283.891 44.46
426.73 51.41 155.336 38.49 13.715 0.111 103.018 21.52 185.957 61.62 286.191 44.55
435.32 51.63 157.300 38.64 15.198 0.154 103.350 21.63 187.433 65.62 288.475 44.66
443.95 51.70 159.254 38.82 16.748 0.208 105.268 22.23 188.804 76.03 290.752 44.77
453.19 51.88 161.195 39.26 18.215 0.269 107.145 22.84 189.968 102.42 293.023 44.85
461 .88 52.11 163.116 39.91 19.655 0.344 108.983 23.47 190.782 196.14 295.290 44.94
471.01 52.25 165.012 40.69 21.196 0.441 110.807 24.09 191.419 169.74 297.550 45.05
166.933 41.57 22.864 0.567 112.639 24.74 192.265 104.56 299.805 45.15
Series 11** 168.930 42.57 24.589 0.721 114.414 25.40 193.399 79.36
316.30 46.48 171.006 43.73 26.347 0.906 116.110 26.11 194.739 66.25 Series vt
325.01 46.77 173.100 45.07 28.090 1.117 117.772 26.95 196.234 58.04 185.146 59.77
333.65 47.07 175.209 46.63 29.846 1.357 119.396 28.03 197.850 52.55 186.162 61.55
342.23 47.38 177.327 48.50 31.640 1.631 120.970 29.66 199.557 48.83 187.240 64.68
350.76 47.64 179.415 50.78 33.408 1.928 122.487 31.52 201.332 46.31 188.265 70.83
359.25 47.96 181.493 53.69 35.183 2.253 123.965 32.10 203.268 44.52 189.203 82.58
367.71 48.18 183.547 57.34 36.994 2.608 125.440 31.41 205.354 43.26 190.007 107.14
376.15 48.55 185.544 60.93 38.846 2.996 126.931 30.67 207.467 42.45 190.503 151.88
384.62 48.78 187.501 66.19 40.721 3.410 128.450 30.30 209.596 41.94 190.746 197.26
393.11 48.97 189.323 85.11 42.587 3.844 129.990 30.27 211.735 41.63 190.955 210.50
401.61 49.19 190.648 174.11 44.432 '4.290 131.541 30.4 7 213.878 41 .44 191.172 183.04
410.16 49.40 191.669 143.11 46.274 4.752 133.099 30.82 216.021 41.38 191.422 152.52
418.72 49.57 193.011 86.06 48.116 5.225 134.701 31.26 218.163 41 .40 191.712 127.94
427.33 49.82 194.795 66.06 49.990 5.721 136.402 31.77 220.301 41.41 192.045 109.94
435.98 50.02 196.870 55.66 51.895 6.235 138.231 32.35 222.435 41.46 192.417 97.03
444.66 50.30 199.150 49.63 53.769 6.752 140.126 32.97 224.564 41.53 192.823 87.55
201.567 46.08 55.633 7.274 141.994 33.62 226.687 41.62 193.259 80.34
Series lilt 204.072 43.98 57.526 7.811 143.836 34.29 228.805 41.72 193.722 74.62
101.275 20.98 206.479 42.79 59.454 8.364 145.652 34.98 230.917 41.82 194.209 69.96
103.164 21.57 208.762 42.11 61.374 8.922 147.442 35.69 233.023 41.92 194.718 66.01
105.229 22.23 211.060 41.70 63.296 9.484 149.205 36.43 235.188 42.03 195.247 62.66
107.407 22.94 213.364 41.48 65.227 10.05 150.943 37.15 237.410 42.14 195.794 59.78
109.609 23.68 215.668 41.39 67.180 10.63 152.659 37.82 239.660 42.26
111 .812 24.45 217.970 41 .40 69.159 11.22 154.355 38.30 241.937 42.38
114.042 25.27 220.269 41.41 71.201 11.83 156.039 38.55 244.242 42.49
116.298 26.21 222.560 41.47 73.361 12.48 157.715 38.64 246.576 42.61
118.546 27.44
Note: M(Feo.93740) = 68.3504 g/mol.
As-quenched sample, University of Oslo.
*
** Partially disproportionated sample, University of Oslo.
t Partially disproportionated sample, Tokyo Institute of Technology.

evacuated and sealed vitreous silica tube of about 50 cm3 usual accuracy obtained with this apparatus (:1:0.25%) to
volume. A central well in the tube contained the heater :1:0.5%. The heat capacity of the empty calorimeter was
and platinum resistance thermometer. The thermometer determined in a separate series of experiments.
was calibrated locally at the ice, steam, tin, zinc, and
antimony points according to IPTS-68. The temperatures RESULTS AND DISCUSSION
were subsequently converted to ITS-90 and are judged to The heat-capacity data for the quenched wiistite,
correspond to ITS-90 within 0.05 K up to 900 K and FeO.93740, are given in Table 1 and plotted for one mole
within 0.10 K above 900 K. The resistance was measured of atoms as a function of temperature in Figure 1 (solid
with an ASL F-18 resistance bridge. The heat capacity of circles). The dotted and solid lines in the figure represent
the ~70g sample represents about 25% of the total heat the heat capacity of the two-phase mixtures of Fe + Fe304
capacity of the calorimeter. The relatively small amount and FeO + Fe304, respectively, calculated for the overall
of sample used in the present investigation degraded the composition FeO.93740. Heat-capacity values for FeO and
976 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE

34
o
200 g

32 o
o
o

150 o
30 r--""1 o
~~0

a o

~'-"
;::;
i--..I 100
S
u~

50

22 100 200 300 400

300 400 500 600 700
T(K) T(K)
FIGURE 2. Molar heat capacity of (I/1.9374)Feo.93740. Open
FIGURE 1. Molar heat capacity of (I/1.9374)Feo.93740. Solid
circles = Tokyo Institute of Technology results; open triangles
circles = University of Oslo (UO) results for quenched wUstite;
open circles = UO results for the partially disproportionated = University of Oslo results.
sample. The heat capacities of the two-phase mixture of Fe +
Fe304 and FeO + Fe304, calculated for the overall composition
FeO.93740,are represented by solid and dashed lines, respectively. effect connected with the magnetic order-disorder tran-
The dotted and dash-dot lines represent equations for (II
sition in the Fe-rich wiistite at 191 K is typical of a high-
I.9427)Feo.92470 and (I/2)FeO, respectively (Grenvold et al.
1993). er-order transition, whereas the effect at 124 K is caused
by the Verwey transition in the magnetite present. An
anomalous shoulder on the heat-capacity curve at '" 160
K is of less certain origin.
Fe304 are from Grenvold et al. (1993), whereas those for The lattice constants of the initially quenched wiistite,
Fe are from Haas and Hemingway (1992). Near 470 K of the products of the first disproportionation reaction
positive temperature-drift rates, which indicate the onset (the present sample), and of the stable two-phase mixture,
of the first disproportionation step in the two-stage dis- Fe + Fe304, are given in Table 2. In addition to the
proportionation reaction, were observed in the equilibra- reflections from Feo.990 and magnetite, one very weak
tion periods. The calorimetric experiments were contin- reflection corresponding to the strongest reflection from
ued until the instrumental temperature-drift rate was Fe was detected by X-ray diffraction. All lines from the
almost regained (at'" 535 K), after which the sample was initial phase with a = 430.4 :t 0.2 pm were absent. Thus,
cooled in the furnace to ambient temperature. A "two- the sample went through the first disproportionation stage
phase" specimen (see below) of Feo.990 and Fe304, mag- and the second reaction began before the sample was
netite, was thus prepared in the calorimeter by careful brought to room temperature. Still, this method is judged
disproportionation of the quenched metastable wiistite. to represent the best controllable method for preparation
In a second series of experiments, the heat capacity of ofa nearly pure two-phase sample in the presence of two
the intermediately formed two-phase specimen was de- disproportionation reactions that take place simulta-
termined in the temperature range 300-440 K, after which neously.
low-temperature experiments were performed. The heat It has been shown that the relationship between lattice
capacities are presented in Figures 1 (open circles) and 2, constant and composition of wustite, expressed as Xo, is
and the experimental results are also given in chronologic approximately linear (see, e.g., Grenvold et al. 1993). The
order in Table 1. Two distinct phase transitions are ob- lattice constant observed for Feo.990, a = 432.6 :t 0.2
served at 124 and 191 K, respectively. The heat-capacity pm, corresponds to an Fe/O ratio of 0.99 with the as-
 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE 977

TABLE2. Lattice constants of the different phases observed 25

D
after different thermal treatments
g
Phase Lattice constant (pm) 250
Prepared D 20

"FeO"
Fe304
430.4 :t 0.2
After first disproportionation step
432.6 :t 0.2
840.5 :t 0.2 200
D
D

d:I 15 tf
I
I
Fe one very weak line
r--'I
D
After second disproportionation step ..-...
~0
Fe present
Fe304 839.5 :t 0.2 e D 10
~150
'-'
""-
~1.."",.,1

sumption that the lattice constant of stoichiometric FeO

S ~D 5
U~
is 433.0 pm. The uncertainty in the Fe/O ratio is rela- 8Q
tively large, probably about :to.005. 100 g
Do
With knowledge of the composition of the initially DO
[J
0
o 10203040506070 8090100
quenched specimen, FeO.93740, and of the intermediately
formed wiistite, the amounts of Fe and magnetite in the
three-phase sample were determined from measurement 50
of the specific magnetic moment of the sample at 4 K
and magnetic field strength up to 4 MAIm. By extrapo-
lation of these results to a magnetic field of zero, the 150 200 250 300
specific magnetic moment was found to be 22.3 (A. m2)/
T(K)
g. The values of the specific magnetic moment of mag-
netite and Fe are from St01en et ai. (1995) and Danan et FIGURE 3. Molar heat capacities of quenched wiistite from 5

ai. (1968) and gave the following weight percentages for to 300 K. Open triangles = Feo.92540(Grenvold et al. 1993); open
inverted triangles = FeO.93790(Grenvold et al. 1993); solid squares
FeO.990, Fe304, and Fe; 78.43, 21.00, and 0.57, respec-
tively. A closely concordant phase distribution was ob- = Feo.9470 (Todd and Bonnickson 1951); open squares = present
results.
tained from neutron diffraction results: 77.72,21.90,0.38
(cf. Fjellvc\g et aI., unpublished manuscript). In the fol-
lowing discussion, the averages of the two sets of values
are used, i.e., 78.08, 21.45, and 0.48 wt%, which corre- ably associated with the changing concentration of atom-
spond to the mole fractions 0.915, 0.078, and 0.007 of ic defects and defect clusters and is strongly related to the
FeO.990, Fe304, and Fe, respectively. degree of nonstoichiometry (Schweika et ai. 1995). The
complex defect clusters in grossly nonstoichiometric wiis-
Heat capacity and thermodynamic properties of FeO.990 tite constitute magnetically coupled entities, which weak-
With the above distribution of phases in our calori- en the cooperative magnetic order-disorder behavior and
metric sample, heat-capacity values for FeO.990 can be cause the transition to be spread over a wide temperature
derived. In the evaluation, the low-temperature heat ca- range. This view is supported by the large magnetic mo-
pacities of Fe304 and Fe were taken from Takai et ai. ment observed for FeO.990 (Fjellvc\g et aI., unpublished
(1994b) and Desai (1986), whereas the values above am- manuscript), which indicates that the number of defect
bient temperature were taken from Gr0nvold et ai. (1993) clusters decreases as stoichiometry is approached.
and Haas and Hemingway (1992), respectively. The heat The previous heat-capacity estimates for FeO by Gr0n-
capacity of FeO.990 is presented in Figure 3 together with voId et ai. (1993) agree well with the present experimental
those for three non stoichiometric wiistite samples. The values for FeO.990, whereas estimates by Haas and Hem-
molar heat capacity ofwiistite, given as [1/(2 - x)]Fe1-xO, ingway (1992), and especially by Fei and Saxena (1986),
varies considerably with composition above the Neel are too large. Sundman's (1991) assessed heat-capacity
temperature. Earlier reported experimental and estimat- values for FeO.9470agree at 300 K but are about 4% low
ed heat capacities for wiistite in the temperature range at 500 K.
300-500 K are shown in Figure 4. The heat capacity of The heat capacity of the intermediately formed three-
(1/1.99)Feo.990 is approximately 7% lower than that of phase sample was fitted to polynomials in temperature
(1/1.9254)Feo.92540 at 300 K. and integrated to give values for the thermodynamic
The low heat-capacity values for FeO.990above the Neel functions at selected temperatures (see Table 3). Ther-
temperature reflect the more strongly cooperative nature modynamic values for FeO.990 were obtained after sub-
of the magnetic order-disorder transition in wiistite as traction of the contributions of Fe [obtained by integra-
stoichiometry is approached. This behavior is presum- tion of the heat capacities recommended by Desai (1986)]
978 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE

TABLE3. Molar thermodynamic properties of the sample with

overall composition FeO.93470
T Cp,m dlHm dlSm <t>~(T,O)
(K) [J/(K'mol)] (J/mol) [J/(K'mol)] [J/(K'mol)]
26.5
10 0.05 0.18 0.03 0.012
15 0.15 0.63 0.07 0.028
20 0.36 1.85 0.14 0.048
25 0.76 4.58 0.26 0.077
26.0 30 1.38 9.84 0.45 0.122
35 2.22 18.76 0.72 0.184
40 3.25 32.35 1.08 0.271
45 4.43 51.49 1.53 0.386
25.5 50 5.73 76.84 2.06 0.523
60 8.52 147.90 3.35 0.885
70 11.46 247.76 4.88 1.341
80 14.48 377.49 6.61 1.891
90 17.51 537.47 8.49 2.518
25.0 100 20.53 727.71 10.49 3.213
110 23.80 949.36 12.60 3.969
120 28.64 1207.4 14.85 4.788
- 130 30.28 1532.8 17.45 5.659
24.5 .
-
- 140 32.93 1846.9 19.78 6.588
.- 150 36.76 2194.7 22.18 7.549
160 38.90 2576.4 24.64 8.538
- 170 43.10 2984.2 27.11 9.556
- 180 51.39 3451 .8 29.78 10.60
24.0 . ~- 191 (230) (4239.2) (34.01 ) (11.82)
- 200 48.27 4906.2 37.44 12.91
210 41.85 5345.4 39.58 14.13
220 41 .42 5760.5 41.51 15.33
300 350 400 450 500 230 41.76 6176.0 43.36 16.51
T(K) 240 42.27 6600.0 45.16 17.66
250 42.77 7025.2 46.90 18.80
FIGURE 4. Molar heat capacities of quenched wiistite from 260 43.28 7455.5 48.58 19.91
300 to 500 K. Solid inverted triangles = present experimental 270 43.79 7890.8 50.23 21.01
280 44.30 8331.3 51.83 22.08
results for (1I 1.99)Feo.990; open circles = presently recommend- 290 44.81 8776.8 53.39 23.13
ed values for (1/1.99)Feo.990. Dash-dot line = equation for (II 298.15 45.18 9136.2 54.64 24.00
1.90)Feo.900; dashed line = equation for (1/1.9254)Feo.92540; sol- 300 45.31 9227.4 54.92 24.16
id line = equation for (1/1.9379)Feo.93790; dotted line = equation 350 47.58 11552 62.08 29.07
400 49.26 13976 68.55 33.61
for (1/2)FeO (Grenvold et al. 1993); dash-dot-dotted line = re- 450 50.35 16464 74.41 37.82
sults for (1I 1.94 7)Feo.9470 (Todd and Bonnickson 1951); aster-
isks = equation for (1/2)FeO (Fei and Saxena 1986); plus signs Note: M(Feo.93740) = 68.3504 g/mol.

= equation for (1/1.947)Feo.9470 (Sundman 1991); bars = equa-

tion for (1/2)FeO (Haas and Hemingway 1992).
is 5.5% lower, and the one by Sundman (1991) is 2.5%
lower. The previously estimated thermodynamic prop-
and magnetite (Takai et al. 1994b) and are given for se- erties of stoichiometric FeO (Gr0nvold et al. 1993) are
lected temperatures in Table 4. retained here.
Previously reported experimental and estimated values The present thermodynamic values for FeO differ con-
of the standard entropy at 298.15 K are shown in Figure siderably from existing estimates and may prove useful
5. The present value for Feo.990, 60.45 J/(K.mol), agrees in the important task of extending the thermodynamic
with that estimated for FeO by Gr0nvold et al. (1993), description of the Fe-O system to high pressure.
61 :t 1 J/(K. mol). The main uncertainty in the present
evaluation relates to the composition of the nearly stoi- TABLE4. Molar thermodynamic properties of FeO.990
chiometric wiistite. A relatively large uncertainty in the
composition, :to.005, is reasonable because the compo- T Cp,m dlHm dlSm <t>~(T,O)
(K) [J/(K'mol)] (J/mol) [J/(K'mol)] [J/(K'mol)]
sition is derived from the value of the lattice constant of
the wiistite phase. A measure of the error in entropy in- 50 6.40 85.45 2.29 0.581
100 22.81 811.2 11.70 3.588
duced by an incorrect composition of the nearly stoichi- 150 41.26 2385.5 24.24 8.337
ometric phase may be obtained by recalculating the stan- 200 54.48 5558.9 42.05 14.26
250 45.4 7 7846.2 52.28 20.90
dard molar entropy of wiistite, assuming it to be 298.15 47.43 10073 60.45 26.66
stoichiometric FeO. The resulting value is ~Sm(FeO, 300 47.56 10170 60.74 26.84
298.15 K) = 60.9 J/(K.mol). Thus, ~Sm(FeO, 298.15 K) 350 49.45 12574 68.21 32.28
400 50.75 15082 74.88 37.18
= 61 :t 1 J/(K. mol) remains a reasonable estimate. The 450 51.59 17636 80.89 41.70
standard molar entropy estimated by Haas and Heming-
Note: M(Feo.990) = 71.28793 g/mol.
way (1992) is 4% larger, the one by Fei and Saxena (1986)
 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE 979

32 70
0 0
0
0
8
0
31 60

Q.
"'~ g
..-.. ~30 :f-++~
~=+-+-f =r+++-+:
,......,
,.--. 50
0 :+T =r++-+:
~+T =r++-+.: 0
S =r++-+:
T++++ 8
~....
~"-' '-"
..........
"""'- ....
L-..."j I....-,j ~~40
V')
s 29 .....p s
00 ~u~
0'\
~<J
n
30 "

28

20
27 100 150 200 250 300
0.00 0.02 0.04 0.06 0.08 0.10 T(K)
x in Fe1_xO FIGURE 6. Molar heat capacities of quenched wlistite near

FIGURE 5. Standard entropy of wustite at 298.15 K. Solid TN' Open triangles = FeO.92540(Grenvold et al. 1993); open in-
verted triangles = FeO.93790(Grenvold et al. 1993); solid squares
triangles = experimental results for (1/1.9254)Feo.92540, (II
= FeO.9470(Todd and Bonnickson 1951); open squares = present
1.9379)Feo.93790 (Grenvold et al. 1993), and (1.199)Feo.990 results. The dotted and solid lines represent background esti-
(present study); open squares = estimated values for (1/2)FeO
mates using eD = 430 K and a variable eD' respectively (see
and (1/1.90)Feo.9oO (Grenvold et al. 1993); solid circle = exper- text).
imental value for (1/1.947)Feo.9470 (Todd and Bonnickson 1951);
open triangle = estimated value for (1/2)FeO (Fei and Saxena
1986); solid line = equation given by Haas and Hemingway
(1992); plus signs = equation given by Sundman (1991); dotted magnetic entropy is only 7.73 J /(K .mol) if the latter ref-
line = compositional variation recommended here. erence heat capacity is used. The assumption of eo = 430
K above 100 K leads to ~magSm= 9.55 J/(K'mol) at 300
K and implies the presence of additional magnetic heat-
The magnetic order-disorder transition capacity contributions at higher temperatures. Further-
The present study shows that the magnetic order-dis- more, the assumption of negligible magnetic contribu-
order transition in Feo.990 is much more cooperative than tions to the heat capacity at temperatures below 100 K
earlier observed for wiistite with composition further from may be an oversimplification. A final conclusion has to
stoichiometry (Todd and Bonnickson 1951; Grenvold et await additional experiments and calculations of the pho-
al. 1993). To derive an estimate of the transitional entro- non and magnon contributions to the heat capacity of
py, a nonmagnetic reference molar heat capacity for wiistite. The expected spin-only magnetic entropy incre-
FeO.990has to be subtracted. Two alternative Debye-type ment, ~magSm= 0.99R In 5 = 13.38 J/(K. mol), still has
background curves were derived after subtracting a dila- not been reached.
tational heat-capacity contribution from the observed Cp The Neel temperature, as observed by adiabatic calo-
results. In one approach, a constant Debye temperature rimetry, does not vary significantly with composition, even
was used with eo = 430 K, as obtained from the maxi- though the concentration of defect clusters, and thus the
mum in a plot of eo vs. temperature. Another reference cooperative nature of the transition, vary considerably.
heat-capacity curve was obtained with a linear decrease TN = 189 K was observed for Feo.9470 (Coughlin et al.
in Debye temperature from 430 Kat 100 K to 350 Kat 1951) and Feo.93790, and 193 K for FeO.92540(Grenvold
260 K, where the cooperative part of the transition seems et al. 1993), whereas the presently observed Neel tem-
negligible. The reference heat capacities are shown in Fig- perature is 191 K.
ure 6 together with the heat capacities of different wiistite Investigators have previously reported variation of the
samples in the temperature region 100-300 K. The molar Neel temperature with composition. According to Koch
-------
980 ST0LEN ET AL.: HEAT CAPACITY OF WUSTITE

50 Fe can be obtained by low-temperature H2 reduction of

60 Fe304 (Colombo et al. 1967). The formation of an Pe-
rich magnetite is in the present case indicated by the large
~50 lattice constant of magnetite obtained after the first dis-
6.
6. proportionation reaction [in agreement with Colombo et
6.
al. (1967)]. The lattice constant decreases from 840.5 :t
40 6.
6. 0.2 to 839.5 :t 0.2 pm, which is typical for stoichiometric
6. magnetite, after further annealing, wherein the stable two-
40 phase mixture of Fe and magnetite is obtained (Table 2).
The heat capacity of the (FeO.990 + Fe304 + Fe) sample
~..---.
--' near the Verwey transition is shown in Figure 7. The
0
8 broad heat-capacity effect does not accord with the dis-
10 continuity expected from earlier studies that indicate the
~'-"" 100 110 120 130 140 150
'-~I...-."j Verwey transition to be of first order for nearly stoichi-
ometric Fe3(l_o)04 (Shepherd et al. 1991; Takai et al.
s 1994a). The transition becomes second order when {) >
u~ 30 0.004 and then occurs at appreciably lower temperatures.
The magnetite presently obtained in the disproportion-
ation reaction is thus nearly stoichiometric and probably
slightly Fe-rich. The nature of the Verwey transition in
such Fe-rich magnetite is not known. It may be of higher
order and occur at a relatively high temperature. To study
this phenomenon further, the calorimetric sample was
20 annealed at 500 K for 3 weeks and thereby converted to
100 110 120 130 140 150 the stable two-phase mixture of Fe and magnetite. The
heat capacity of this product is shown in the inset of
T(K) Figure 7 (the absolute value of the heat capacity is un-
FIGURE 7. Molar heat capacity of the partially dispropor- certain because the sample contained a small amount of
tionated sample near Tverwey.Corresponding data for the fully glass after accidental breakage of the sample container
disproportionated sample are shown in the inset. Dashed line during transport). The transition is somewhat more co-
represents estimated nontransitional heat capacity. operative, but not of first order as expected for stoichio-
metric Fe304. The sharpening of the heat-capacity peak
and Fine (1967) and Seehra and Srinivasan (1984) the accompanying the Verwey transition cannot unequivo-
Neel temperature decreases with increasing Fe content. cally be ascribed to a change in the composition of the
The opposite variation was reported by Mainard et al. magnetite. It may equally well represent an annealing ef-
(1968), whereas McCammon (1992) reported that TNde- fect, with broadening of the peak owing to a distribution
creases from 210 K for FeO.900to 192 K for FeO.9SO, and of magnetite grains with slightly differing compositions.
then TN increases to 202 K for FeO.9SO.The discrepancy The entropy of the first-order Verwey transition for
between the different studies indicates that the magnetic nearly stoichiometric Fe3-o04 varies from 6.0 J/(K. mol)
properties of quenched wiistite may be affected by the for {) = -0.0002 to 4.0 J/(K.mol) for {) = 0.0035. The
thermal history, through differences in type and amount Verwey transition is of higher order for the more O-rich
of the frozen-in defects. Such differences are, however, magnetite, and the entropy of transition is much lower,
not expected to influence the thermodynamic properties e.g., 1.8 J/(K.mol) for {)= 0.004. Thus, the enthalpy value
of wiistite within the present accuracy of measurement. of the Verwey transition cannot be used for accurate de-
termination of the amount of magnetite present in the
Higher order Verwey transition in Fe-rich metastable (Feo.990 + Fe304 + Fe) sample. Integration of the excess
magnetite heat capacity, shown in Figure 7, leads to the entropy 4.0
According to Dieckmann (1982), the composition of J/(K.mol) for the Verwey transition in Fe304. The entro-
magnetite exceeds the stoichiometric 3:4 ratio in equilib- py equals that observed for the first-order transition for
rium with hematite, Fe203, at high temperatures. Below Fe3-o04, with {) = 0.0035, and hence indicates a nearly
the eutectoid decomposition temperature of wiistite, stoichiometric composition for the magnetite studied here.
magnetite appears to be stoichiometric in equilibrium with The origin of the shoulder on the heat-capacity curve
Fe. Because the magnetite in the present study was formed near 150 K is uncertain. It is probably not related to the
in metastable equilibrium with FeO.990' and not in stable Verwey transition in the magnetite. If so, the molar en-
equilibrium with Fe, we suggest that a more Fe-rich mag- tropy of transition becomes 10.2 J/(K.mol) of Fe304. This
netite is formed because a metastable phase generally in- value is improbably high in comparison with earlier re-
creases the stability fields of the neighboring phase or ported values. Thus, the shoulder is tentatively ascribed
phases. It has been reported that magnetite with excess to the wiistite phase.
 ST0LEN ET AL.: HEAT CAPACITY OF WDSTITE 981

In conclusion, the magnetite formed in metastable oxygen, and copper-oxygen. V.S. Geological Survey, Open File Report
equilibrium with Feo.990 is probably Fe rich. It is prob- 92-267.
Jeanloz, R. (1990) The nature of the Earth's core. Annual Reviews of
ably burdened by local fluctuations in composition. The Earth and Planetary Science, 18, 357-386.
preparation method may, however, provide new means Knittle, E., and Jeanloz, R. (1989) Simulating the core-mantle boundary:
for studying the nature of the Verwey transition. Such An experimental study of high-pressure reactions between silicates and
magnetite is presumably more Fe rich than Fe3:t604 in liquid iron. Geophysical Research Letters, 16, 609-612.
- (1991) Earth's core-mantle boundary: Results of experiments at
equilibrium with Fe. Further calorimetric results might high pressures and temperatures. Science, 251, 1438-1443.
thus shed light on the ongoing discussion about the Koch, F.B., and Fine, M.E. (1967) Magnetic properties of FexO as related
changes in the Verwey transition properties of magnetite to the defect structure. Journal of Applied Physics, 38, 1470-1471.
with deficiency and surplus (see, e.g., Shepherd et al. Liu, L. (1975) Post-oxide phases of forsterite and enstatite. Geophysical
1991; ° Takai et al. 1994a). Research Letters, 2, 41 7-419.
Mainard, R., Boubel, M., and Fousse, H. (1968) On the specific-heat
anomaly in FexO. Comptes Rendus de l'Academie des Sciences, series
B(12), 1299-1301 (in French).
REFERENCES CITED
McCammon, C.A. (1992) Magnetic properties of FexO (x > 0.95): Vari-
Atake, T., Kawaji, H., Hamano, A., and Saito, Y. (1990) Construction of ation of Neel temperature. Journal of Magnetism and Magnetic Ma-
an adiabatic calorimeter for heat capacity measurements from liquid terials, 104-107, 1937-1938.
helium temperatures to 330 K. Report of the Research Laboratory of - (1993) Effect of pressure on the composition of the lower mantle
Engineering Materials, Tokyo Institute of Technology, 15, 13-23. end member FexO. Science, 259, 66-68.
Castelliz, L., de Sutter, W., and Halla, F. (1954) On the decomposition McCammon, C.A., and Liu, L. (1984) The effect of pressure on nonsto-
of the wiistite phase. Monatshefte ftir Chemie, 85, 487-490 (in Ger- ichiometric wiistite, FexO: The iron-rich phase boundary. Physics and
man). Chemistry of Minerals, 10, 106-113.
Colombo, V., Gazzarrini, F., and Lanzavecchia, G. (1967) Mechanism of Schweika, W., Hoser, A., Martin, M., and Carlsson, A.E. (1995) Defect
iron oxides reduction at temperatures below 400 °C. Materials Science structure of ferrous oxide Fe1-XO. Physical Review B, 51, 15771-15788.
and Engineering, 2, 125-135. Seehra, M.S., and Srinivasan, G. (1984) Magnetic studies of non-stoichi-
Coughlin, J.P., King, E.G., and Bonnickson, K.R. (1951) High tempera- ometric FezO and evidence for magnetic clusters. Journal of Physics C:
ture heat content of ferrous oxide, magnetite and ferric oxide. Journal Solid State Physics, 17, 883-892.
of the American Chemical Society, 73, 3891-3893. Shen, P., Bassett, W.A., and Liu, L. (1983) Experimental determination
Danan, H., Herr, A., and Meyer, A.J.P. (1968) New determinations of of the effect of pressure and temperature on the stoichiometry and
the saturation magnetization of nickel and iron. Journal of Applied phase relations ofwiistite. Geochimica et Cosmochimia Acta, 47,773-
Physics, 39, 669-670. 778.
Desai, P.D. (1986) Thermodynamic properties of iron and silicon. Journal Shepherd, J.P., Koenitzer, J.W., Aragon, R., Spalek, J., and Honig, J.M.
of Physical and Chemical Reference Data, 15, 967-983. (1991) Heat capacity and entropy of nonstoichiometric magnetite
Deslattes, R.D., and Henins, A. (1973) X-ray to visible wavelength ratios. Fe3(l-6)04: Thermodynamic nature of the Verwey transition. Physical
Physical Review Letters, 31, 972-975. Review, 43,8461-8471.
Dieckmann, R. (1982) Defects and cation distribution in magnetite (IV): Simons, B., and Seifert, F. (1979) High-pressure wUstite: Cell parameters
Nonstoichiometry and point defect structure of magnetite (Fe3-604). and Mossbauer spectra. Annual Report Geophysical Laboratory
Berichte der Bunsengesellschaft ftir Physikalische Chemie, 86, 112- (Washington, DC), 78, 625-626.
118. St0len, S., Glockner, R., and Gr0nvold, F. (1995) Nearly stoichiometric
Fei, Y., and Saxena, S.K. (1986) A thermodynamic data base for phase iron monoxide formed as a metastable intermediate in a two-stage
equilibria in the system Fe-Mg-Si-O at high pressure and temperature. disproportionation of quenched wUstite-thermodynamic and kinetic
Physics and Chemistry of Minerals, 13, 311-324. aspects. Thermochimica Acta, 256, 91-106.
Gr0nvold, F. (1967) Adiabatic calorimetry for the investigation of reac- Sundman, B. (1991) An assessment of the Fe-O system. Journal of Phase
tive substances in the range from 25 to 775 OC:Heat capacity of a-al- Equilibria, 12, 127-140.
uminium oxide. Acta Chemica Scandinavia, 21, 1695-1713. Takai, S., Atake, T., and Koga, Y. (1994a) Heat capacity anomalies at
- (1993) Enthalpy of fusion and temperature of fusion of indium, the Verwey transition of Fe3(l-6)04. Thermochimica Acta, 246, 1-10.
and redetermination of the enthalpy of fusion of tin. Journal ofChem- Takai, S., Akishige, Y., Kawaji, H., Atake, T., and Sawaguchi, E. (1994b)
ical Thermodynamics, 25, 1133-1144. Low temperature heat capacities and Verwey transition of magnetite.
Gr0nvold, F., St0len, S., Tolmach, P., and Westrum, E.F., Jr. (1993) Heat Journal of Chemical Thermodynamics, 26, 1259-1266.
capacities of the wiistites FeO.93790and FeO.92S40at temperatures from Todd, S.S., and Bonnickson, K.R. (1951) Low temperature heat capacities
5 K to 350 K: Thermodynamics of the reaction: xFe(s) + (l/4)Fe304(s) and entropies at 298.15 K of ferrous oxide, manganous oxide and va-
= FeO.7S00+xO(s) = Fe1-XO(s) at T = 850 K, and properties of Fe1_XO(s) nadium monoxide. Journal of the American Chemical Society, 73, 3894-
to T= 1000 K. Journal of Chemical Thermodynamics, 25,1089-1117. 3895.
Haas, J.R., Jr., and Hemingway, B.S. (1992) Recommended standard
electrochemical potentials and fugacities of oxygen for the solid buffers MANUSCRIPT RECEIVED AUGUST 14, 1995
and thermodynamic data in the systems iron-silicon-oxygen, nickel- MANUSCRIPT ACCEPTED MARCH 20, 1996
~