CaCl2-rich corner of the CaCl2-CaF2-CaO phase diagram, up to 50 mol.% CaF2 and 20 mol.% CaO, was investigated by Wenz et al. . These authors showed the existence of:
· a ternary eutectic point at the composition CaCl2-CaF2-CaO (79-17-4 mol.%) melting at 625˚C,
· two peritectic points at the composition CaCl2-CaF2-CaO (67-23-10 mol.%) and (64-23-13 mol.%) melting respectively at 660˚C and 670˚C.
CaF2-rich part (>50 mol.% CaF2) of the phase diagram is still unknown and needs to be investigated. To build it, binary phase diagrams constituting the ternary system, CaCl2-CaF2, CaCl2-CaO and CaF2-CaO, must be previously considered.
The CaCl2-CaF2 system was investigated by several authors  - :
· in the CaCl2-rich region, they all suggested a single eutectics in the composition range 18.5 - 25 mol.% CaF2 at around 650˚C as well as a peritectic point at 735˚C around 41 mol.% CaF2,
· the system exhibits CaFCl compound at 50 mol.% CaF2 which is known to have an incongruent melting point . In the composition range 41-50 mol.% CaF2, this phase changes to α-CaF2 at around 735˚C,
· the liquidus data are more scattered in the CaF2-rich part (>50 mol.% CaF2).
Moreover, some of these results were obtained by visual-polythermal method which is not very accurate according to Chartrand et al. . These authors proposed a calculated phase diagram using a modified quasi-chemical model, and confirmed it with data from several studies.
The CaCl2-CaO system was only investigated in the CaCl2-rich part and data were compiled by Shaw et al.     :
· the system exhibits an eutectic point in the 5 - 6.5 mol.% CaO composition range at around 750˚C. However, a different eutectic composition was found by Threadgill  (28.7 mol.% CaO and 593˚C),
· a second eutectic was proposed by Neumann et al.  at 21 mol.% and 800˚C. These authors also reported the CaO(CaCl2)4 compound at 20 mol.% CaO, which melts congruently at 839˚C. Similar results were obtained by Perry et al.  specifying that the 20 - 22 mol.% CaO region may exhibit a peritectic point instead of eutectic point. However, Wenz et al.  did not report it: they proposed a peritectic point at 18.5 mol.% melting at 835˚C and suggested the CaO(CaCl2)2 compound at 33 mol.% CaO which was considered in Shaw et al. study ,
· a sharp increase of the liquidus temperature for higher CaO amount above 22 mol.% was reported by all the authors     ,
· Neumann et al.  noticed a thermal plateau at 703˚C on the cooling curve, attributed to the CaO(CaCl2)4 allotropic phase transformation.
The liquidus values are very dispersed since finely colloidal suspension of CaO can be obtained, resulting that it is not completely solubilized according to Freidina et al. .
The liquidus temperatures of the CaF2-CaO phase diagram are higher than 1300˚C as the melting points of both pure CaF2 and CaO compounds are 1418˚C and 2572˚C respectively :
· in the literature, authors  -  indicated a single eutectic shape diagram, determined by calculations and experimental measurements. There is no consensus about an accurate eutectic point composition which varies in the 15 - 28 mol.% CaO range and the solidus temperature is around 1360˚C according to Zaitsev et al.  and Kim et al. ,
· no specific compound was identified in the diagram, but partial CaO solubility in solid CaF2 phase was reported by Kim et al.  with a maximum around 5 mol.% CaO at the solidus temperature. This author also identified the CaF2 allotropic transition α → β at 1146˚C without any effect on the CaO solubility,
· Baak  proposed a two-liquid phase region in the 0.8 - 10 mol.% CaO composition range, up to 1485˚C.
According to the literature, discrepancies exist on the binary phase diagrams constituting the CaCl2-CaF2-CaO ternary system. In this study, both CaCl2-CaF2 and CaCl2-CaO systems were investigated by combining experimental data and modeling. From these results, the ternary phase diagram was calculated and vertical cross-sections measurements were then performed to verify the model.
Differential Scanning Calorimetry (DSC) technique was used to measure solidus, liquidus and phase transition temperatures of mixtures. Experiments were carried out in a heat-flux DSC MHTC 96 from SETARAM. The calorimeter is a 3D quasi-Calvet type sensor and can measures up to 1400˚C. The standard deviation was estimated from three runs with each calibration metals (Sn: = 231.9˚C, Pb: = 327.5˚C, Sb: = 630.6˚C, Ag: = 961.8˚C, Au: = 1064.2˚C, Cu: T˚fus = 1084.6˚C) and salts (CaCl2: = 772˚C, LiF: = 848˚C) and is lower than 1˚C. From measurements, a correlation among the measured heat flow and the corresponding molar enthalpy was found. Boron nitride crucibles were used for chloro-fluoride samples and graphite crucibles for CaO-base system.
Calcium chloride (Alfa Aesar 99.99%), calcium fluoride (Alfa Aesar 99.99 %) and calcium oxide (Alfa Aesar 99.95%) were used. The CaO was previously heated up to 900˚C under vacuum (≈10−1 mbar) for 12 hours to remove the residual water. Samples were prepared in a glove box under argon atmosphere by mixing the pure chemicals and around 150 mg were analysed. They were heat-treated into the apparatus under vacuum (400˚C, ≈10−2 mbar). Then, a pre-melting step was carried out under argon to ensure a good mixing of the chemicals. DSC analyses were performed under argon flow (30 mL∙min−1) with a heating rate of 2˚C min−1. No mass loss was observed during the experiment.
Crystallographic phases were identified by X-Ray Diffraction analysis (XRD) with a Bruker D4-ENDEAVOR diffractometer. Patterns were recorded from 20˚ to 70˚ by 0.016˚ step (2θ) in the Bragg-Brentano geometry using the Kα radiation of the copper anode (40 kV, 40 mA). Samples were prepared in a glove box under inert argon atmosphere: the salt was protected by a Kapton film to prevent water absorption.
Binary and ternary phase diagrams modeling was performed using Phase Diagram module from FactSage® 6.3 software  . In the Phase Diagram module, the Gibbs energy minimization is used for calculations where axis and constant parameters can be set, such as the volume, the temperature, the pressure… In this study, FTsalt and FToxid databases were used and the pressure was fixed to 1 atmosphere.
3. Results and Discussion
3.1. Binary Diagrams
3.1.1. CaCl2-CaF2 System
A classical DSC-signal from a binary sample of CaCl2-CaF2 (70-30 mol.%) is shown in Figure 1 and exhibits a sharp endothermic peak at around 660˚C, solidus—followed by a second broad endothermic peak at around 710˚C, liquidus—typical of a binary mixture. The solidus temperature was measured as onset point whereas liquidus temperature at the peak maximum, as recommended by Höhne et al. . The values were reported on the diagram in Figure 2 and are consistent with other authors  -  and FactSage® model (FTsalt database).
Eutectic composition could not be accurately determined on DSC curves since the peaks are not well defined in the 15 - 23 mol.% CaF2 composition range. The determination of the eutectic composition was done by using of a Tamman diagram   , representing the molar enthalpy associated to a first-order phase transition versus the molar fraction, based on the lever rule application. The resulting diagram has a triangle shape and the baseline endpoints indicate specific compositions (eutectics, peritectics, stoichiometric compound, solid solubility…).
In this case, the Tamman plot was established using enthalpy in the 5 - 40 mol.% CaF2 composition range. Both left and right endpoints from the baseline
Figure 1. Classical DSC curve of a CaCl2-CaF2 (70-30 mol.%) sample at 2˚C∙min−1.
Figure 2. Phase diagram (upper part) and Tamman diagram (lower part) of the CaCl2-CaF2 system.
are close to 0 and 50 mol.% CaF2 respectively, indicating a negligible solid solubility into CaCl2 and the presence of CaFCl stoichiometric compound. The highest point indicates the eutectic composition, 20.0 mol.% CaF2 at 647˚C, close to the calculated one (18.3 mol.% CaF2 and 652˚C) in agreement with the model.
To confirm the presence of the CaFCl stoichiometric compound, a melted mixture of CaCl2-CaF2 (40-60 mol.%) annealed for 72 hours at 650˚C was prepared for phases identification by XRD analysis. The resulting pattern is shown in Figure 3 where CaFCl was identified, which is consistent with the model and the literature data . As it was expected for this composition, α-CaF2 was also detected without any segregation.
Unexpected signal occurred in the CaF2-rich part at around 647˚C and was not observed in the data available in the literature: it could be attributed to a segregation effect. One way to avoid it is to introduce an annealing step after the salt pre-melting: this procedure was successful as no more unexpected signal was observed.
3.1.2. CaCl2-CaO System
The CaCl2-CaO diagram was investigated in the 0 - 27 mol.% CaO composition range. Measured data are reported in the diagram of Figure 4 and are coherent with the FactSage® model (FTsalt database).
The eutectic composition was determined by a Tamman diagram using solidus enthalpy in the 1 - 17 mol.% CaO composition range at 750˚C. Due to the
Figure 3. XRD pattern of CaCl2-CaF2 (40-60 mol.%) sample.
Figure 4. Phase diagram (upper part) and Tamman diagram (lower part) of the CaCl2-CaO system.
triangle sides trend, hypothesis was made that solid solubility is negligible in pure CaCl2, and that stoichiometric compound exists at 20 mol.% CaO. It permitted to set the Tamman baseline endpoints at 0 and 20 mol.% CaO; the determined eutectic composition is then 6.3 mol.% CaO and is in agreement with the calculated value within the model and consistent with the literature    .
The presence of a second eutectic point at around 21 mol.% CaO mentioned by Neumann et al.  could not be clearly identified since the interval composition is extremely narrow. However, the abrupt increase of the liquidus temperature is observed at higher CaO contents: the temperature rises from 900˚C at 22 mol.% CaO up to 1200˚C at 27 mol.% CaO, which is consistent with other studies   but a 2 mol.% difference can be noted with Wenz et al. work  and even more with Wang et al. data  (4 mol.%).
Furthermore, the extrapolated curve reaches around 32 mol.% of CaO at 1400˚C and is in agreement with the CaO solubility value determined by Sano et al. .
To prove the existence of a stoichiometric compound in the binary system, a mixture of CaCl2 containing 13 mol.% of CaO was melted and annealed for 48 hours at 650˚C. It was then characterized by XRD for phase identification. The pattern is shown in Figure 5 and reveals the presence of CaCl2 and CaO(CaCl2)4 compound, confirming the endpoint composition of the Tamman triangle basis (20 mol.% CaO), as well as the presence of three peaks that could not be attributed.
A slight endothermic peak is observed at around 715˚C for every samples similarly to the thermal plateau observed by Neumann et al.  at 703˚C on the cooling curves; this signal can be attributed to the allotropic phase transition of CaO(CaCl2)4 compound.
3.1.3. CaF2-CaO System
Due to the high liquidus temperature (>1300˚C), the CaF2-CaO system was not investigated by DSC. However, the phase diagram was calculated using FTsalt database which offers satisfying models for both CaCl2-CaF2 and CaCl2-CaO systems. The model was improved by including additional CaF2/CaO data from FToxid database and was plotted in Figure 6; it exhibits an eutectic point at 19.8 mol.% CaO melting at 1305˚C, and the CaO solubility domain into α- and β-CaF2 solids mentioned in Kim et al. study .
3.2. The Ternary Diagram
To build the CaCl2-CaF2-CaO diagram, the three binary diagrams constituting
Figure 5. XRD pattern of CaCl2-CaO (87-13 mol.%) sample.
Figure 6. Calculated phase diagram of the CaF2-CaO system.
Figure 7. CaCl2-CaF2-CaO phase diagram calculated with FactSage®.
the system were used. The calculated ternary phase diagram is plotted in Figure 7 and exhibits an eutectic point at the composition CaCl2-CaF2-CaO (78.2- 15.7-6.1 mol.%) melting at 637˚C, similar as in Wenz et al. diagram . However, four peritectic points are identified in the CaCl2-rich region and another one in the CaF2-rich region contrary to Wenz et al. work  who identified only two peritectic points in the CaCl2-rich region; characteristics of specific intersection points are summarized in Table 1.
Three vertical liquidus cross-sections were investigated by DSC; a CaCl2-CaF2 composition (50-50, 40-60 and 30-70 mol.%) was set and the CaO amount was increased up to 18 mol.%. Measurements for richer CaF2 and CaO concentrations were not possible due to the temperature limitation of the apparatus.
The results are shown in Figure 8 and the following observations can be noticed:
· the general shape of the liquidus temperature is in agreement with the model for each cross-section: a plateau from 0 to a threshold composition around 11 mol.% CaO and then a sharp increase,
Table 1. Liquidus temperature and composition of invariant points.
Figure 8. Liquidus cross-sections of the CaCl2-CaF2-CaO phase diagram (solid lines correspond to the model).
Figure 9. Evolution of the CaO solubility logarithm vs. the inverse of the absolute temperature.
· a slight difference of the threshold composition between measurements and the model, which could be attributed to the data selected to calculate the CaF2-CaO phase diagram and was not experimentally verified.
At constant temperature, the CaO solubility decreases when the CaF2 portion in the melt increases, as observed at 1400˚C by Sano et al.  in the investigated area. To compare his data with the present work, the logarithm of CaO solubility (xCaO) above 11 mol.% was plotted as a function of the inverse of the absolute temperature in Figure 9. A linear relationship was obtained for all compositions and equations were extrapolated toward 1400˚C to determine the CaO solubility; data are reported in Table 2.
Table 2. Linear regression equation of the CaO solubility logarithm vs. inverse of the absolute temperature and CaO solubility values at 1400˚C
The CaO solubility calculated at 1400˚C is slightly different than the ones determined by Sano et al.  with a maximum discrepancy reaching 4.2 mol.% in the equimolar CaCl2-CaF2 medium. However, the CaO solubility evolution is similar with a decrease of the solubilized CaO amount while the CaCl2 is replaced by CaF2 at constant temperature (1400˚C).
In this work, binaries CaF2-CaO, CaCl2-CaF2, CaCl2-CaO and ternary CaCl2-CaF2-CaO phase diagrams system were investigated in the CaO low region (<40 mol.% CaO) by thermal analysis (DSC) and thermodynamic calculations (FactSage®-FTsalt/FToxid databases).
The ternary diagram exhibits an eutectic point in the CaCl2-rich region at 637˚C, CaCl2-CaF2-CaO (78.2-15.7-6.1 mol.%), and five peritectic points. Vertical cross-sections analysis was performed in three CaCl2-CaF2 compositions (50-50, 40-60 and 30-70 mol.%) up to 18 mol.% CaO. Each section exhibits the same shape: a plateau up to around 11 mol.% CaO and then a rapid increase of the liquidus temperature. The measurements show that CaO solubility decreases while CaCl2 is replaced by CaF2 in this diagram region.
The authors wish to express sincere acknowledgements to Dr. C. Tenailleau from the Centre Inter-universitaire de Recherche et d’Ingénierie des Matériaux (CIRIMAT) in Toulouse who provided valuable XRD analysis.
 Krause, I.E. and Bergman, A.G. (1942) Singular Irreversible-Reciprocal Systems without Separation into Layers from Potassium and Calcium and Calcium Chlorides and Fluorides, and Sodium and Calcium Chlorides and Fluorides. Doklady Akademii Nauk SSSR, 35, 20-24.
 Desyatnik, V.N., Kurbatov, N.N., Strelov, V.A. and Shchavelevc, V.V. (1979) Melting Point Diagrams of Systems Containing Potassium and Calcium Fluoride. Izv Vysshikh Uchebnykh Zaved Tsvetnaya Metall, 4, 65-68.
 Chartrand, P. and Pelton, A.D. (2001) Thermodynamic Evaluation and Optimization of the Li, Na, K, Mg, Ca//F, Cl Reciprocal System Using the Modified Quasi-Chemical Model. Metallurgical and Materials Transactions A, 32, 1417-1430.
 Neumann, B., Kroger, C. and Jüttner, H. (1935) Die Systeme Erdalkalichlorid-Erdalkalioxyd und die Zersetzung der Erdalkalichloride durch Wasserdampf. Zeitschrift für Elektrochemie und angewandte physikalische Chemie, 41, 725-736.
 Threadgill, W.D. (1965) The Calcium Chloride-Calcium Oxide Fused Salt Electrolytic System: Solubilities, Metal Contents, and Freezing Points. Journal of The Electrochemical Society, 112, 632-633.
 Wang, S., Zhang, F., Liu, X. and Zhang, L. (2008) CaO Solubility and Activity Coefficient in Molten Salts CaCl2–x (x=0, NaCl, KCl, SrCl2, BaCl2 and LiCl). Thermochimica Acta, 470, 105-107.
 Seo, W.-G., Zhou, D. and Tsukihashi, F. (2005) Calculation of Thermodynamic Properties and Phase Diagrams for the CaO-CaF2, BaO-CaO and BaO-CaF2 Systems by Molecular Dynamics Simulation. Materials Transactions, 46, 643-650.
 Seo, W.-G., Zhou, D., Hamano, T. and Tsukihashi, F. (2004) Calulation of Phase Diagrams for the CaO-CaF2 and CaO-BaO Binary Systems by Using Molecular Dynamics. 7th International Conference on Molten Slags Fluxes and Salts, Cape Town, 25-28 January 2004, 835-838.
 Zaitsev, A.I., Zemchenko, M.A., Litvina, A.D. and Mogutnov, B.M. (1993) Thermodynamic Calculation of Phase Equilibria in the CaF2-SiO2-CaO System. Journal of Materials Chemistry, 3, 541-546.
 Kim, D.-G., Van Hoek, C., Liebske, C., Van Der Laan, S., Hudon, P. and Jung, I.-H. (2012) Phase Diagram Study of the CaO-CaF2 System. ISIJ International, 52, 1945-1950.
 Kim, D.-G., Van Ende, M.-A., Van Hoek, C., Liebske, C., Van Der Laan, S. and Jung, I.-H. (2012) A Critical Evaluation and Thermodynamic Optimization of the CaO-CaF2 System. Metallurgical and Materials Transactions B, 43, 1315-1325.
 Bale, C.W., Chartrand, P., Degterov, S.A., Eriksson, G., Hack, K., Ben Mahfoud, R., Melancon, J., Pelton, A.D. and Petersen, S. (2002) FactSage Thermochemical Software and Databases. Calphad, 26, 189-228.
 Bale, C.W., Bélisle, E., Chartrand, P., Decterov, S.A., Eriksson, G., Hack, K., Jung, I.-H., Kang, Y.-B., Melancon, J., Pelton, A.D., Robelin, C. and Petersen, S. (2009) FactSage Thermochemical Software and Databases—Recent Developments. Calphad, 33, 295-311.
 Hohne, G.W.H., Hemminger, W. and Flammersheim, H.-J. (1996) Differential Scanning Calorimetry: An Introduction for Practitioners. Springer, Berlin, Heidelberg.
 Tammann, G. (1903) über die Ermittelung der Zusammensetzung chemischer Verbindungen ohne Hilfe der Analyse. Zeitschrift für anorganische Chemie, 37, 303-313.
 Rycerz, L. (2013) Practical Remarks Concerning Phase Diagrams Determination on the Basis of Differential Scanning Calorimetry Measurements. Journal of Thermal Analysis and Calorimetry, 113, 231-238.
 Sano, N., Tsukihashi, F. and Tagaya, A. (1991) Thermodynamics of Phosphorus in CaO-CaF2-SiO2 and CaO-CaF2-CaCl2 Melts Saturated with CaO. ISIJ International, 31, 1345-1347.