Synthesis, Characterization, and Energetics of Solid Solution Along the Dolomite-Ankerite Join, and Implications for the Stability of Ordered Cafe(C03)2

American Mineralogist, Volume 81, pages 1141-1147, 1996

Synthesis, characterization, and energetics of solid solution along the dolomite-ankerite join, and implications for the stability of ordered CaFe(C03)2

L. CHAI* AND A. NAVROTSKY

Department of Geological and Geophysical Sciences, and Princeton Materials Institute, Princeton University, Princeton, New Jersey 08544, U.S.A.

ABSTRACT Samples along the dolomite-ankerite join were synthesized using a piston-cylinder apparatus and the double-capsule method. Some of the ankerite samples may be disordered. Thermal analysis and X-ray diffraction showed that all samples can be completely decomposed to uniquely defined products under calorimetric conditions (770°C, O2), and a well- constrained thermodynamic cycle was developed to determine the enthalpy of formation. The energetics of ordered and disordered ankerite solid solutions were estimated using data from calorimetry, lattice-energy calculations, and phase equilibria. The enthalpies of formation of ordered dolomite and disordered end-member ankerite from binary carbonates, determined by calorimetry, are -9.29 :t 1.97 and 6.98 :t 2.08 kllmol, respectively. The enthalpy of formation of ordered ankerite appears to become more endothermic with increasing Fe content, whereas the enthalpy of formation of disordered ankerite becomes more exothermic with increasing Fe content. The enthalpy of disordering in dolomite (approximately 25 kllmol) is much larger than that in pure ankerite, CaFe(C03h (approximately 10 kllmol), which may explain the nonexistence of ordered CaFe(C03)2'

INTRODUCTION the cation octahedra may be responsible for the instabil- FeH and MgH substitution is common in mineral sys- ity of ankerite with high Fe content. Reeder and Dollase tems. FeH and MgH form a complete solid solution in (1989) studied the crystal structures of ankerite with the the FeC03-MgC03 system, whereas FeH only partially use of mineral single crystals and found no obvious in- substitutes for MgH in dolomite to form ankerite dication of significant distortion of the coordination poly- [Ca(FexMg1_x)(C03)2, 0 ~ x ~ 0.7] (Reeder 1983; An- hedra. ovitz and Essene 1987; Reeder and Dollase 1989; David- Study of the energetics of ankerite solid solutions may son et al. 1993, and references therein). As a solid solu- explain the apparent instability of CaFe(C03)2' Recently, tion, ankerite has the same crystal structure as dolomite, we developed calorimetric methods for FeH-bearing car- with CaH occupying one set of cation sites and FeH and bonates (Chai and Navrotsky 1994, 1996). We used this MgH the other (Reeder and Dollase 1989). At high tem- method to determine the enthalpy of formation of syn- perature, ankerite is expected to undergo a disordering thetic ankerite. We also derived the enthalpies of for- transition, as does dolomite, to the calcite structure. End- mation of CaMg(C03h and CaFe(C03)2 with the use of member ankerite, CaFe(C03)2, has not been found as a calculated lattice energies. These results help to define the mineral. Furthermore, phase-equilibrium studies (Gold- energetics of mixing of ordered and disordered Ca(Fe,Mg)- smith et al. 1962; Rosenberg 1963, 1967) show that the (C03)2 solid solutions. composition CaFe(C03)2 lies in a two-phase field at tem- EXPERIMENTS peratures between 350 and 800°C consisting of calcite and siderite. Recently, Davidson et al. (1993) successfully Synthesis synthesized a disordered CaFe(C03)2 (calcite-type) phase Ankerite samples were synthesized using a piston-cyl- at 845°C and 30 kbar and suggested that ordered inder high-pressure apparatus and the double-capsule CaFe(C03)2 could exist at low temperature. method (Powell et al. 1984; Anderson and Crerar 1993). Several efforts have been made to explain the nonex- Precipitated CaC03 and synthetic MgC03 and FeC03 were istence of ordered CaFe(C03)2 (Rosenberg and Foit 1979; used as starting materials. As suggested in previous work Reeder and Dollase 1989, 1991; Rosenberg 1991). Ro- (Rosenberg 1967, 1968), precipitated CaC03 was used to senberg and Foit (1979) proposed that the distortion of promote the synthesis reaction. MgC03 and FeC03 were prepared from basic magnesium carbonate and iron ox- Present address: Department of Materials Science and En- alate, respectively, in a C02 atmosphere using a cold-seal gineering,* University of Pennsylvania, 3231 Walnut Street, Phil- vessel apparatus (Chai and Navrotsky 1993, 1994). adelphia, Pennsylvania 19104, U.S.A. Approximately 300 mg of the starting materials with 0003-004X/96/091O-1141$05.00 1141 1142 CRAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE appropriate stoichiometry were ground together in an ag- step of 0.02° 28 and a counting time of 4 s (per step). The ate mortar with acetone for about 1 h. A portion of the 20 range was usual1y from 10 to 90°. Si was used as an mixture was then packed into a gold capsule (0.38 cm external standard. The unit-cell parameters were refined diameter and 0.8 cm long). The gold capsule was placed using the subroutine in the controlling software. within a platinum capsule (0.46 cm outside diameter and Simultaneous differential thermal analysis and ther- 1.6 cm long) along with approximately 100 mg of Ag2C204 mogravimetric analysis (DT A/TGA) were conducted in and a small amount of graphite (as the reducing buffer a dynamic O2 atmosphere with the use of a Netzsch STA materials). One end of the platinum capsule was previ- 409C apparatus. The flow rate of O2 was approximately ously sealed by arc welding. After loading the charge, the 0.5 cm3/s. The experimental temperature ranged from other end of the platinum capsule was welded with the room temperature to 1050 °C, with a 10 °C/min heating part of the platinum capsule containing the charge im- rate. Al203 crucibles were used as sample containers. The mersed in water to prevent the decomposition of Ag2C204. buoyancy effect was corrected using experiments with The welded platinum capsule was about one-half the empty crucibles. length of the graphite furnace used in the piston cylinder. A heat-treated pyrophyllite cup was used to separate the Calorimetry platinum capsule from the graphite furnace. A W3%Re/ All calorimetric measurements were made using a Cal- W25%Re thermocouple and an Eurotherm 818 controller vet-type calorimeter operating at 770 OC(Navrotsky 1977). were used to measure temperature. Three types of calorimetric experiments were used: drop- The double-capsule technique yielded about 50-80 mg solution calorimetry, solution calorimetry, and trans- of sample in each experiment. To obtain a sufficient posed-temperature-drop calorimetry. In solution calo- amount of sample for characterization and thermochem- rimetry, a sample is preequilibrated at the calorimetric ical measurements, 2-3 successful experiments under temperature before it is dissolved in the solvent (molten similar conditions were needed. We found it was difficult lead borate, 2PbO. B203). The measured heat effect is the to reproduce temperature among the experiments, pos- enthalpy of solution. In drop-solution calorimetry, small sibly because it was difficult to control the positions of pellets of about 20 mg of sample are dropped from room the platinum capsule and the thermocouple when the temperature into a platinum crucible containing solvent whole sample assembly was under pressure. All samples in the calorimeter. The measured heat effects consist of used for later experiments were produced using power heat content and enthalpy of solution. In transposed- tem- control. The temperature for all experiments was approx- perature-drop calorimetry, small pellets of sample are imately 850 :t 50°C. dropped from room temperature into an empty platinum A typical experiment was run at 18 kbar for 24 h. At crucible in the calorimeter. The heat effects in this case the start of each experiment, temperature and pressure include heat content of the sample and any heat effects were increased simultaneously to reduce the possibility resulting from reactions (decarbonation, oxidation) in the of abrupt bursting of the capsule as a result of the decom- calorimeter. position of Ag2C204 and the expansion of gas during heat- Calorimetric methods for carbonates in general and mg. Fe2+-containing carbonates in particular were described At the end of each experiment, the power was cut to by Chai and Navrotsky (1993, 1994, 1996). Drop-solu- zero. The temperature dropped to about 300°C within 5 tion calorimetry in a flowing gas has proved effective for s. Pressure was then released. When the platinum capsule calcite, magnesite, and dolomite. However, for the Fe2+- was opened, the release of gas indicated that the experi- bearing carbonates, direct drop-solution calorimetry could ment was successful. The white color of the sample also not be used because the calorimetric reaction reached an indicated that the temperature was well controlled and irreproducible final state, presumably because of the slow that Fe was divalent. oxidation of Fe2+ in the lead borate solvent (Chai and The dolomite sample was made from a mechanical Navrotsky 1994). A two-step calorimetric procedure was mixture of precipitated CaC03 and basic magnesium car- used for the Fe2+-bearing carbonates. The first step of bonate at 700°C in C02 atmosphere with the use of a oxidative decarbonation in O2 was followed by dissolu- cold-seal vessel apparatus (Chai and Navrotsky 1993). tion of the FeH-containing oxide products. Once the Fe2+ components in both reactants and products in the anker- Characterization ite-formation reaction were transformed into the same Samples were characterized by X-ray diffraction (XRD) oxidation state (+ 3) in the solvent, the enthalpy of for- with the use of a Scintag PAD-V diffractometer with a mation could be calculated. Detailed equations are given solid-state detector and CuKa radiation. Because of the below. limited amount of sample, a slurry of the sample powder RESULTS AND DISCUSSION with acetone was mounted on a zero-background quartz sample holder. The sample holder was spun during data Cation disordering and structural changes collection to minimize preferred orientation. For unit-cell The compositions of the ankerite samples were calcu- parameters and structure refinements, powder X-ray dif- lated from the starting stoichiometry and are reported as fraction data were collected in a step-scan mode with a the mole fraction of Fe2+ in the dolomite structure [i.e., CHAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE 1143

TABLE1. Composition and unit-cell parameters of the ankerite solid solutions

Sample x a (A) c(A) V (A3) Figure of merit

F(15) CaMg(C03)2. 0.00 4.8076(3) 16.007(2) 320.40 = 113 F(13) ank1 0.1583 4.8102(6) 16.0524(43) 321.66 = 38 F(13) ank2 0.2990 4.8108(2) 16.1095(17) 322.88 = 98 x =1.0 F(13) ank3 0.4996 4.8131(1) 16.1200(9) 323.41 = 169 F(13) ank4 0.6480 4.8182(8) 16.1543(5) 324.79 = 355 F(10) ank5 1.00 4.8300(9) 16.221(8) 327.72 = 21 x =0.6480 CaFe(C03)2 .. 1.00 4.8291 (3) 16.210(1) 327.38

Note: The numbers in parentheses in unit-cell parameters are the errors x =0.4996 in the last decimal place. .Eugui dolomite. .. From Davidson et al. 1993. x =0.2990

x = 0.1583 x in Ca(FexMg1_J(C03)2, Table 1]. X-ray diffraction patterns of the ankerite solid solutions show that the order- x=O.O ing reflections gradually disappear with increasing Fe content (Fig. 1). The ordering reflections (015) and (021) can still be seen in the sample ankl with x = 0.1583 but disappear in the samples with higher Fe contents. When o 10 20 30 40 50 60 70 CuKa radiation is used the backgrounds of the XRD pat- 26 (degrees) terns of the Fe-bearing phases are usually high because of the fluorescence of Fe. To reduce the signal/noise ratio, FIGURE 1. X-ray diffraction patterns of the ankerite solid solution. Three ordering reflections in dolomite are indexed. a scan with longer counting time (10 s) was performed for ank3 with composition x = 0.4996. No ordering reflections were found. paratus. This sample is generally considered to be par- Goldsmith et a1. (1962) observed a wider composition tially disordered (Navrotsky and Capobianco 1987) range in which ordering reflections were found. Also, they because of the likelihood of reordering during the quench found that ankerite deviated from the ideal stoichiometry (Reeder and Wenk 1983). (to more Ca-rich composition) with increasing Fe con- The unit cell of ankerite expands as the amount ofFe2+ tent. These different results may be due to the higher increases because of the replacement ofMg2+ by the larg- temperature used in this study. er Fe2+. For a given composition, x, the a parameters of The disordering temperature is expected to decrease all synthetic samples (disordered) are smaller than those with increasing Fe content. The disordering of of natural samples (ordered), whereas there is a turnover CaMg(C03)2 occurs at about 1150 °C (Reeder and Wenk in the c parameters between ordered and disordered sam- 1983), whereas disordered CaFe(C03)2 has been made at ples. This trend is seen clearly in the volume-composition about 845°C and 30 kbar (Davidson et a1. 1993). There- plot (Fig. 2c). Disordering in dolomite increases unit-cell fore, samples ank3 and ank4 (x = 0.6480) are probably volume; disordering in end-member ankerite decreases at least partially disordered. To determine the degree of unit-cell volume. The thermochemical results, discussed disordering, Rietveld refinements were attempted for in the following section, support the above interpretation sample ank3, with ordered, disordered, and partially dis- of the ordering state. ordered dolomite structures used as the starting structural models. Unfortunately, no structural model showed a clear Decomposition products advantage over the others. It is essential to understand the composition change In general, unit-cell parameters of synthetic and natural during the decomposition and the behavior of the decom- ankerite samples increase with Fe content (Table 1 and position products at the calorimetric temperature and at Fig. 2). The unit-cell parameters of ank5 (x = 1.0) are room temperature to interpret calorimetric data. The two- consistent with those of disordered CaFe(C03)2 from Da- step calorimetric approach is based on the assumption vidson et a1. (1993), strongly suggesting that ank5 is com- that the energy state of the decomposition products after pletely disordered. The unit-cell parameters for natural the first step does not change before they are dissolved in ankerite and a Eugui dolomite are also included in this the solvent in the second step. This may be a reasonable figure. The natural ankerite samples are ordered, with assumption because the decarbonation of ankerite and Fe2+ occupying the Mg2+ site in the dolomite structure the oxidation of Fe2+ component are usually rapid and (Reeder and Dollase 1989). Mn2+ contents are up to 4.7 complete in the first step. Several experiments were per- mol% (a significant impurity) in those samples. The dis- formed to understand the behavior of the intermediate ordered Eugui dolomite in the study of Navrotsky and products. Capobianco (1987) was obtained by quenching the an- DT AlTGA showed that all samples decomposed below nealed sample from 1250 °C using a piston-cylinder ap- 800 °C in O2, XRD of the products after DT A/TGA in- 1144 CHAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE

4.84 16.25 328 o o .0 (a) 16.20 (b) ..0 (e) ....~ ~ 4.83 B .' ~~326 ~.~ .,' 16.15 .';Cf ~" S -$ 4.82 ~~0 -$ ~324 u 16.10 "0 '" > ..0 ~~.0'

FIGURE 2. Unit-cell parameters of synthetic and natural ankerite samples. The circles are from this work, the triangles are from Reeder and Dollase (1989), the squares are from Davidson et al. (1993), and the diamonds are from Navrotsky and Capobianco (1987). The dashed lines are the proposed changes for ordered samples, and the dotted lines are for disordered samples. dicated that the decomposition products for all ankerite position products were not measured directly because of solid solutions are CaO, MgO, and Ca2Fe20S' The sample the reactivity of CaO and MgO in the decomposition weight change was monitored during calorimetry. XRD products (mentioned above). The heat contents were cal- and weight changes confirmed that all samples, prepared culated from the individual oxide values of Robie et al. as pellets, completely decomposed under the calorimetric (1978). The enthalpies of drop solution of the decom- conditions to form oxide mixtures (CaO, MgO, and position products were also calculated from individual Ca2Fe20S)' oxide values rather than from data for the decomposition Weight changes ofthe decomposition products at room samples themselves. The enthalpies of drop solution of temperature in air were also monitored. The reactions of CaO and MgO were derived from the measured enthal- the decomposition products in air (presumably of CaO pies of drop solution of the corresponding carbonates as and MgO with H20 and C02) were slow and continuous. follows: This made direct calorimetric measurements of the de- CaC03 (25°C, s) --+ CaO (770°C, soln) composition products difficult. + C02 (770°C, g) (1) Energetics of ankerite solid solutions CaO (25°C, s) + C02 (25°C, g) The enthalpies of formation of ankerite solid solution, --+ CaC03 (25°C, s) (2) Ca(FexMg'_x)(C03h, were calculated from enthalpies of drop solution (fli/!,) of ankerite, calcite, magnesite, and C02 (770°C, g) --+ C02 (25°C, g). (3) siderite (Table 2). The enthalpies of drop solution of cal- Adding Equations 1,2, and 3, we have cite, magnesite, and dolomite were measured directly, whereas the enthalpies of drop solution of ankerite and CaO (25°C, s) --+ CaO (770°C, soln). (4) siderite were calculated from enthalpies of decomposition The enthalpy of Reaction 1 is the measured enthalpy of (AHdec),heat contents (AHhc), and enthalpies of drop so- drop solution (fli/d,) of calcite (Table 2), the enthalpy of lution (fli/d,) of the decomposition products. Enthalpies Reaction 2 is the heat of formation (fli/r) of calcite, and of decomposition of ankerite solid solutions were deter- the enthalpy of Reaction 3 is heat content (fli/hc) of C02' mined by transposed-temperature-drop calorimetry. Heat The latter two values are from Robie et al. (1978). The contents and enthalpies of drop solution of the decom- calculated enthalpy of drop solution of CaO (Reaction 4)

TABLE2. Calorimetric data for the ankerite solid solutions

Sample x ilH... ilHhct ilH..* ilH~ ilH, Dolomite 0.00 399.21 :t 0.90(6) -9.29 :t 1.97 ank1 0.1583 404.19 :t 2.10(4) 75.024 33.477 362.64 :t 3.11 6.80 :t 3.45 ank2 0.2990 375.54 :t 1.24(4) 76.974 42.344 340.91 :t 2.30 10.36:t 2.64 ank3 0.4996 333.89 :t 4.44(5) 79.750 54.968 309.11 :t 4.68 16.21 :t 4.80 ank4 0.6482 306.99 :t 1.45(5) 81.811 64.340 289.52 :t 1.88 16.54:t 2.10 ank5 1.00 253.72 :t 1.45(6) 86.690 86.525 253.56 :t 1.84 6.98 :t 2.08 Calcite 196.85 :t 0.49(6) Magnesite 193.07 :t 1.68(5) Siderite§ 63.69 :t 0.83

Note: Unit is kilojoules per mole. Errors are two standard deviations of the mean. The number in parentheses is the number of experiments. t Heat contents of oxides are from Robie et al. (1978). Calculated from data of the individual oxides (see text). *§ From Chai and Navrotsky (1996). CHAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE 1145

30.00 TABLE 3. Lattice energy of carbonates in the calcite-siderite- magnesite system -0 Phase Formula Lattice energy (kJ/moIJ IS 20.00- ...... Calcite CaCO, -3163.5 g Siderite FeCO, -3412.1 § Magnesite MgCO, -3425.2 Dolomite CaMg(CO,J2 -6600.8 ~ 10.00- Pure ankerite CaFe(CO,). -6580.2 13 .2 .... o ~ 0.00- 1995). A three-term Born equation (Coulomb interac- Id 8 tion, repulsion, and van der Waals force) was used for ~ - - - - the ion-pair potentials, and the parameters in the poten- f_ -10.00 ------~;de~~~ tials were obtained by minimization of total lattice energy with the use of crystal-structure data of calcite, magnesite, I I I I I I dolomite, and siderite. The structural data for ordered o 0.2 0.4 0.6 0.8 1 CaFe(C03)2 were derived by extrapolation from natural x in Ca(FexMgI_YC03)2 ankerite (Reeder and Dollase 1989). The calculated lattice energies (I::..HL;Table 3) were used to obtain the en- FIGURE 3. Enthalpy of formation of the ankerite solid solu- thalpy offormation of dolomite and ankerite (Fig. 3): Ililr tion. The circles are from calorimetry in this work, the squares (dol) = IlilL (dol) - IlilL (cal) - IlilL (mag) = -12.1 kJI are from lattice-energy calculations, the triangle is from Navrot- mol, and Ililr (ank) sky and Capobianco (1987), the inverted triangles are from Hol- = IlilL (ank) - IlilL (cal) - IlilL (sid) land and Powell (1990), and the diamond is calculated from the = -4.6 kJ/mol. This estimate of the enthalpy of forma- solvus (see text). The Xs are for disordered phases calculated tion of dolomite is in very good agreement with values from solid-solution-mixing parameter obtained from Davies and determined from experiments (-11.5 :t 0.5 kJ/mol, Na- Navrotsky (1983). The darkness of the symbols indicates the vrotsky and Capobianco 1987; -11.1 :t 2.5 kJ/mol, Chai degree of ordering: Solid symbols indicate complete order; open and Navrotsky 1993). Therefore, the value derived from symbols, complete disorder; and shaded symbols, partial disor- lattice energies (-4.6 kJ/mol) may give a reliable esti- der. mate of the enthalpy off ormation of ordered CaFe(C03)2, which has not been determined experimentally. This re- from the above reactions is -16.28 :t 1.19 kJ/mol. The sult indicates that pure ankerite is energetically less stable enthalpy of drop solution ofMgO (39.78 :t 2.44 kJ/mo1) than dolomite relative to binary carbonates. can be derived from similar thermochemical analysis. The The measured enthalpy of formation of disordered do- enthalpy of drop solution of Ca2Fe20S (86.525 :t 1.33 kJI lomite (Navrotsky and Capobianco 1987) and the en- mol for 0.5Ca2Fe20S) was determined by drop solution thai pies of formation of CaMg(C03)2 and ordered of the decomposition product of ank5 with composition CaFe(C03)2 from the Holland and Powell (1990) data set CaFe(C03)2 (Table 2). The heat content of Ca2Fe20S was provide additional constraints on the energetics of mixing also measured using the decomposition product ofank5. and disordering in the ankerite solid solution (Fig. 3). The The resulting value (86.51 :t 0.68 kJ/mol for 0.5Ca2Fe20S) enthalpy of formation of ordered CaFe(C03)2 from Hol- from this work is in good agreement with that (86.69 kJI land and Powell (1990) was derived from the data of mol) from Robie et a1. (1978). FeH-MgH partitioning in carbonates and carbonate-sil- The total enthalpy of drop solution of ankerite [Ca(Fex. icate equilibria. Mg1-X)(C03)2] can then be calculated as Ilil:, = Ilildec + If the quenched sample from Navrotsky and Capo- Ilild, - Ililhc' bianco (1987) is completely disordered, then the calori- The enthalpy of formation of ankerite [Ca(FexMgi_J- metric data from this study may define the enthalpy of (C03)2] from calcite, magnesite, and siderite may be cal. mixing of disordered ankerite (Fig. 3). When a regular culated as Ililr (ank) = Ilil:, (cal) + (1 - x)Ilil:, (mag) solution model is used, a curve fit to these data results in + xllil:, (sid) - Ilil:, (ank), where Ilil:, (cal), Ilil:, (mag), a regular solution parameter of 34.52 :t 4.89 kJ/mol. and Ilil:, (sid) are the enthalpies of drop solution of cal- This implies a critical temperature of approximately 1800 cite, magnesite, and siderite, respectively (Table 2 and °C, which contradicts the experimental results (only a sin- Fig. 3). gle phase was found in our synthesis conditions at 850 :t The enthalpy of formation of disordered CaFe(C03)2 50°C). An alternative explanation is that the sample from can be estimated from the solvus of the CaC03-FeC03 Navrotsky and Capobianco (1987) is indeed partially dis- system (Davies and Navrotsky 1983; Davidson et a1. ordered, and their measured enthalpy of disordering pro- 1993). This value (8.35 kJ/mol) agrees with the measure- vides a lower limit to the enthalpy of disordered dolo- ment from this study (6.98 :t 2.08 kJ/mol). mite. An empirical approach was applied using the WMIN Davies and Navrotsky (1983) systematically studied the program (Busing 1981) to calculate the lattice energies of relationship of energetics and volume mismatch in some carbonates in the CaC03-MgC03-FeC03 system (Chai solid solutions. They found that regular solution param------

1146 CHAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE eters of the solid-solution models (W) have a linear re- thalpy of disordering is 20-30 kllmol for dolomite but lationship with the difference in volume of the end-mem- only about 10 kllmol for iron dolomite. It should be bers (.:1V), W = 100.8.:1V - 0.4 (kllmol). Recently, we emphasized that the uncertainty of the above estimation determined the energetics of solid solution in the system for the energetics of disordered dolomite is relatively large. FeC03-MgC03 (Chai and Navrotsky 1996) and found that the regular solution parameter from the calorimetric data, Implications for the stability of ordered CaFe(C03)2 using a regular solution model, was in good agreement The above estimations may explain the nonexistence with that calculated from the above relationship. The sol- of pure ankerite. The enthalpy difference of disordering id solution in the system FeC03-MgC03 crystallizes in in dolomite is much larger than that in CaFe(C03)2' In the calcite structure, which is the same as that of disor- other words, there is insufficient energy gain on ordering dered ankerite. Therefore, the energetics of mixing of an- of CaFe(C03)2 to stabilize it relative to a two-phase mix- kerite solid solutions may be estimated from the above ture of siderite and calcite. relationship. The entropy off ormation of dolomite from calcite and The volume differences between disordered and or- magnesite is -1.6 J/(mol. K) (Robie et al. 1978). If the dered dolomite samples and pure ankerite are small. Thus, entropy of formation of ordered CaFe(C03)2 from calcite the predicted enthalpies of mixing are nearly ideal [ap- and siderite is approximately the same, then the less neg- proximately < 1 kllmol for Ca(FeosM&Js)(C03)2]' There- ative enthalpy of formation of CaFe(C03)2 is the reason fore, it may be reasonable to treat the enthalpies of for- for the instability of this phase. The - T.:1Sterm is always mation as linear with Fe content. With this assumption, positive and increases at high temperature, which is not the calorimetric data ofank3, ank4, and ank5, combined favorable for the formation of ordered ankerite. The dis- with the solvus data for pure ankerite (Davies and Na- ordered phase, on the other hand, is increasingly stabi- vrotsky 1983; Davidson et al. 1993), define the energetics lized at high temperature by its large positive configura- of disordered ankerite (upper dashed line, Fig. 3). The tional entropy. Thus, ordered CaFe(C03)2 could be made calorimetric data, combined with lattice-energy calcula- only at low temperature, but kinetic barriers may prevent tions for dolomite and ankerite, give the best estimate of its formation. the energetics of ordered ankerite (lower dashed line, Fig. The enthalpy of disordering of CaFe(C03h can be es- 3). The calorimetric data, then, for x = 0.158 and 0.299 timated (about 10 kllmol) from the data shown in Figure represent partially ordered samples. The intercept at x = 3. The configurational entropy of complete disordering of o of the upper line represents the extrapolated enthalpy CaFe(C03)2 is .:1S = 2R In 2 = 11.53 J/(mol. K). There- of formation of a totally disordered dolomite, implying fore, the temperature of disordering of CaFe(C03)2, treat- an enthalpy of complete disordering that is at least twice ed as a hypothetical first-order transition, would be 594 that measured by Navrotsky and Capobianco (1987). 0c. This is below the solvus (approximately 829 0C) of From Figure 3 it can be seen that the enthalpy of for- the system CaC03 -FeC03 at the composition of mation of ordered ankerite becomes more endothermic CaFe(C03)2 (Davidson et al. 1993). This result, again, with increasing Fe content, whereas disordered ankerite shows that ordered CaFe(C03)2 could exist only at low becomes more exothermic with increasing Fe content. temperature, if at all, which is consistent with the predic- The correlation between volume mismatch and mixing tion by Davidson et al. (1993). energetics of Davies and Navrotsky (1983) may also be It is also interesting that our data suggest that the vol- used to estimate the enthalpies offormation of disordered umes of disordered synthetic samples are smaller than ankerite and disordered dolomite from the mixing of two those of more ordered natural samples in the range of binary carbonates. For the reaction CaC03 (s) + FeC03 high Fe content, although a complication is that the latter (s) -- CaFe(C03)2 (s) (disordered), the enthalpy is esti- contain a few mole percent of the larger cation, Mn2+. mated to be 11.3 kllmol (the volume-mismatch param- This may suggest a negative volume of disordering (as eter of calcite and siderite is 0.2257), in reasonable agree- suggested by Davidson et al. 1993) and reflects the ex- ment with the measured value of 7.0 :t 2.1 kllmol. For pansion of the c axis in the more disordered structure. the reaction CaC03 (s) + MgC03 (s) -+ CaMg(C03h (s) This behavior is opposite that of CaMg(C03)2 and again (disordered), the enthalpy is estimated to be 13.6 kllmol points to the conclusion that the alternate stacking of Fe (the volume-mismatch parameter of calcite and magne- and Ca layers does not appear to be favorable, either site is 0.2742). Because the enthalpy of formation of or- structurally or energetically. This negative volume change dered dolomite from calcite and magnesite is -11.3 kJI in ordering of pure ankerite indicates that higher pressure mol, this calculation yields an enthalpy of disordering of is not favorable for the formation of the ordered ankerite. 24.9 kllmol, in the general range found by extrapolation of the calorimetric studies and significantly higher than ACKNOWLEDGMENTS the value of about 13 kllmol estimated by Navrotsky and Capobianco (1987), again suggesting that their sample had We thank Richard J. Reeder and Peter Heaney for many helpful dis- cussions. We thank William R. Busing for providing the WMIN program partially ordered upon quench. and the instruction manual, and Jeff E. Post for help with lattice-energy The above arguments placed limits on the enthalpies calculations. We also thank Robert Rapp for his enormous help in piston- of disordering of dolomite and pure ankerite. The en- cylinder synthesis. The manuscript was improved by the comments of J. CRAI AND NA VROTSKY: ENERGETICS OF DOLOMITE-ANKERITE 1147

William Carey, Peter A. Rock, and an anonymous reviewer. This work lations: The system K,O-Na,O-CaO-MgO-MnO-FeO-Fe,O,-AI,O,- was supported by funding !Tom the Department of Energy (grant DEFG02- TiO,-SiO,-C-H,-O,. Journal of Metamorphic Geology, 8, 89-124. 85ERI3437). This work also benefited from support of our laboratories Navrotsky, A. (1977) Progress and new directions in high temperature by the Center for High Pressure Research, an NSF Science and Technol- calorimetry. Physics and Chemistry of MineraIs, 2, 89-104. ogy Center. Navrotsky, A., and Capobianco, C. (1987) Enthalpies of formation of dolomite and of magnesian calcites. American Mineralogist, 72, 782- 787. REFERENCES CITED Powell, R., Condlilfe, D.M., and Condlilfe, E. (1984) Calcite-dolomite Anderson, G.M., and Crerar, D.A. (1993) Thermodynamics in geochem- geothermometry in the system CaCO,-MgCO,-FeCO,: An experimen- istry, the equilibrium model, 342 p. Oxford University Press, New tal study. Journal of Metamorphic Geology, 2, 33-41. York. Reeder, RJ. (1983) Crystal chemistry of the rhombohedral carbonates. Anovitz, L.M., and Essene, E.J. (1987) Phase equilibria in the system In Mineralogical Society of America Reviews in Mineralogy, II, 394 p. CaCO,-MgCO,-FeCO,. Journal of Petrology, 23, 389-414. Reeder, R.J., and Wenk, H.-R. (1983) Structure refinements of some ther- Busing, W.R (1981) WMIN, a computer program to model molecules mally disordered dolomites. American Mineralogist, 68, 769-776. and crystals in terms of potential energy functions. Oak Ridge National Reeder, R.J., and Dollase, W.A. (1989) Structural variation in the dolo- Laboratory Report, ORNL-5747. mite-ankerite solid-solution series: An X-ray, Mossbauer and TEM Chai, L. (1995) Synthesis, characterization, and energetics of carbonates study. American Mineralogist, 74,1159-1167. in the CaCO,-MgCO,-FeCO, system. Ph.D. thesis, Princeton Univer- - (1991) Structural variation in the dolomite-ankerite solid-solution sity, Princeton, New Jersey. series: An X-ray, Mossbauer, and TEM study-Reply. American Min- Chai, L., and Navrotsky, A. (1993) Thermochemistry of carbonate-py- eralogist, 76, 661-662. roxene equilibria. Contributions to Mineralogy and Petrology, 114, 139- Robie, R., Hemingway, B.S., and Fisher, J.R. (1978) Thermodynamic 147. properties of minerals and related substances at 298.15 K and I bar -(1994) Enthalpy of formation of siderite and its application in (10' pascals) pressure and at higher temperatures. U.S. Geological Sur- phase equilibrium calculation. American Mineralogist, 79, 921-929. vey Bulletin 1452. - (1996) Synthesis, characterization, and enthalpy of mixing of the Rosenberg, P.E. (1963) Subsolidus relations in the system CaCO,-FeCO,. (Fe,Mg)CO, solid solution. Geochimica et Cosmochimica Acta, in press. American Journal of Science, 261, 683-690. Davidson, P.M., Symmes, G.H., Cohen, B.A., Reeder, R.J., and Lindsley, - (1967) Subsolidus relations in the system CaCO,-MgCO,-FeCO, D.H. (1993) Synthesis of the new compound CaFe(CO,), and experi- between 350 OCand 550°C. American Mineralogist, 52, 787-796. mental constraints on the (Ca,Fe)CO, join. Geochimica et Cosmochim- -(1968) Subsolidus relations on the dolomite join, CaMg(CO,),- ica Acta, 57, 5105-5109. CaFe(CO,),-CaMn(CO,),. American Mineralogist, 53, 880-889. Davies, P.K., and Navrotsky, A. (1983) Quantitative correlations of de- - (1991) Structural variation in the dolomite-ankerite solid-solution viations from ideality in binary and pseudo binary solid solutions. Jour- series: An X-ray, Mossbauer, and TEM study-Discussion. American nal of Solid State Chemistry, 46, 1-22. Mineralogist, 76, 659-660. Goldsmith, J.R., Graf, D.L., Witters, J., and Northrop, D.A. (1962) Stud- Rosenberg, P.E., and Foit, F.F., Jr. (1979) The stability of transition metal ies in the system CaCO,-MgCO,-FeCO,: I. Phase relations; 2. A meth- dolomites in carbonate system: A discussion. Geochimica et Cosmo- od from major-element spectrochemical analysis; 3. Compositions of chimica Acta, 43, 951-955. some ferroan dolomites. Journal of Geology, 70, 659-688. Holland, T.J.B., and Powell, R (1990) An enlarged and updated inter- MANUSCRIPT RECEIVED OcrOBER 2, 1995 nally consistent thermodynamic dataset with uncertainties and corre- MANUSCRIPT ACCEPTED MAY 28, 1996