A T 1 o p e © K 1 s o g s p c s i a a m f T i T T h 0 Available online at www.sciencedirect.com ScienceDirect Journal of the European Ceramic Society 34 (2014) 4053–4058 Phase equilibria study of K–O–Si system in equilibrium with air G. Akdogan a,b, H. Johto b,1, P. Taskinen b,∗ a University of Stellenbosch, Department of Process Engineering, Private Bag X1, Matieland 7602, South Africa b Aalto University, School of Chemical Technology, Metallurgical Thermodynamics and Modelling Research Group, Espoo, Finland Received 21 February 2014; received in revised form 30 April 2014; accepted 5 May 2014 Available online 13 June 2014 bstract he binary phase diagram in the silica rich corner of K–O–Si system in equilibrium with air has been studied at temperatures between 770 ◦C and 500 ◦C. Equilibration at high temperature in an appropriate containment material of pure silica, followed by rapid quenching and measurement f phase structures and assemblages using SEM-EDS confirmed by an electron probe X-ray microanalysis technique, was carried out to obtain the hase composition data at equilibrium. The results are in good agreement with previous experimental data and close to the earlier assessed phase quilibria at silica saturation. 2014 Elsevier Ltd. All rights reserved. eywords: Thermodynamics; Slag; Melting point; Solubility i a s b s o a i n c t m a r . Introduction Reliable thermodynamic descriptions for important systems, uch as K2O–SiO2, can contribute to a better understanding f the chemical reactions taking place during the processes in lass, ceramic technology and extractive metallurgy. The binary ub-systems are also the key to computational multi-component hase diagrams. In copper smelting silica sand is added as flux and this ontains sodium and potassium oxides as impurities. During melting these impurities are believed to play important role n viscosity and liquidus temperature variations. In property ssessments the low-order system properties, typically binaries, re derived by extrapolation from high-order systems and ay be very far from their true nature, due to lack of data rom the binary and ternary systems of industrial importance. his highlights the need for modelling of slag equilibria and ts properties in the copper smelting and refining conditions. he thermodynamics of selected copper smelting slags with ∗ Corresponding author. P.O. Box 16200, FI-00076 Aalto, Finland. el.: +358 40 501 7411. E-mail address:
[email protected] (P. Taskinen). 1 Present address: Boliden Harjavalta Oy, FI-29200 Harjavalta, Finland. p t f K q K F a ttp://dx.doi.org/10.1016/j.jeurceramsoc.2014.05.007 955-2219/© 2014 Elsevier Ltd. All rights reserved. mpurity components, i.e. K, and Na, becomes vital. Therefore, s an important part of the copper smelting slags, silica rich ide of the K2O–SiO2 binary is of particular interest. This binary system also has applications in combustion of iomass fuels in particular; the concentrations of potassium in ome agricultural fuels considerably exceed those in coal and ther more traditional fuels. Potassium catalysed pyrolysis has profound influence on the char formation stage; generally ncreasing the char yields. The presence of potassium has a sig- ificant influence on the rates and profiles of evolution of gas omponents generated during the pyrolysis processes.1 According to Nordin,2 high potassium contents can have he effect of lowering the melting point of the ash or bed aterial, thereby enhancing the ability of deposit formation nd bed sintering. In view of the fact that the bed mate- ial is composed mainly of quartz particles, the formation of hases rich in potassium silicates is likely to play an impor- ant role in bed sintering when combusting agricultural biomass uels. The equilibrium diagram in the composition range from 2SiO3 to SiO2 was first measured by Kracek et al.3,4 by uenching techniques. Melting points of K SiO , K Si O and 2 3 2 2 5 2Si4O9 were reported as 976, 1045, and 770 ◦C, respectively. our eutectics were reported at 780, 742, 769 and 767 ◦C t 45.5, 67.6, 72.5 and 73.6 wt% SiO2 for K2SiO3–K2Si2O5, http://crossmark.crossref.org/dialog/?doi=10.1016/j.jeurceramsoc.2014.05.007&domain=pdf http://www.sciencedirect.com/science/journal/09552219 dx.doi.org/10.1016/j.jeurceramsoc.2014.05.007 mailto:
[email protected] dx.doi.org/10.1016/j.jeurceramsoc.2014.05.007 4 pean C K r a s t d o s m G t o t d s m i C K p u a t t r e d b t K m t o r o 1 t o t c p a K ( n ( r o t t a t e a a d c w m b m u c c o a b w L r s o u s t o a u m a c u e o T o f s L t t r a r t r h w s r t 054 G. Akdogan et al. / Journal of the Euro 2Si4O9–K2Si2O5, Quartz-K2Si4O9, and Tridymite–K2Si4O9, espectively. The K2O–SiO2 system has also been thermodynamically nalysed by Eliezer et al.5 who used KO0.5 and SiO2 as the pecies and a Redlich–Kister polynomial for representing the hermodynamic excess properties of the liquid phase. Their escription does not extend to solutions richer in potassium xide than 55 mol% K2O, and a phase diagram is given for the ection SiO2–K2Si2O5. Kim and Sanders6 used a sub-regular odel with K2O and SiO2 as components to describe the excess ibbs energy of the liquid phase. The interaction parameters of he liquid phase were obtained solely from the liquidus curve f silica and estimates of the critical temperature and composi- ion of the metastable liquid–liquid miscibility gap. The phase iagram was roughly outlined and thermodynamic properties of olid compounds used in the calculations were not given. This akes comparisons with experimental data difficult. The thermodynamic data for the binary system, including ts liquid and solid phases, were assessed and optimised by a ALPHAD-type thermodynamic fitting of the binary data in the 2O–SiO2 system by Allendorf and Spear7. The basic data on hase formation, melting point, eutectic compositions, and liq- idus curves were obtained primarily from earlier work,8,9 and n associate model with four associates as species was used for he liquid silicate phase. The predicted melting points and eutec- ic compositions are given in the paper along with the values eported by Wu et al.9 While the liquidus temperatures were in xact agreement, the eutectic temperatures showed appreciable ifferences when compared with the assessed values obtained y Wu et al.9 Zaitsev et al.10,11 constructed the equilibrium diagram for he K2O–SiO2 system based on the activities of SiO2 and 2O in K2O–SiO2 melts as calculated from a thermodynamic odel of silicate melts. The model was developed from the heory of associated solutions12. An assumption of formation f Si–O cluster groups of arbitrary sizes and spatial configu- ations was presumed. Thermodynamic functions of formation f solid compounds and phase equilibria data were taken from 1 and 3, respectively. The model approximates the concentra- ion and temperature dependencies of thermodynamic properties f the K2O–SiO2 melts as well as phase equilibria. In addi- ion, this model takes into account the structure of melts and hanges caused by variation of the concentrations of the com- onents. Melting of the intermediate compounds K2O·SiO2 nd K2O·2SiO2 is congruent, but incongruent in the case of 2O·4SiO2. Correspondingly, two eutectic and the peritectic K2O·4SiO2, SiO2, and liquid) points were presented in the join. Forsberg13 on the other hand developed a new thermody- amic description of the K2O–SiO2 system using KSi0.25O 1/4K4SiO4) and SiO2 as species for the liquid phase. With espect to these components, the highly non-ideal interactions f potassium oxide and silica could be described by using only hree temperature-independent Redlich–Kister coefficients for he excess Gibbs energy of the liquid phase. This treatment llowed appropriate choice of species with respect to the varia- ion of thermodynamic properties with composition, thus the xcess Gibbs energy of the liquid phase was described with r t T p eramic Society 34 (2014) 4053–4058 small number of parameters. In alkali silicate systems, this pproach was further justified by the lack of experimental phase iagram and thermochemical information at high alkali oxide ontents. The calculated phase diagram was in good agreement ith the experimental liquidus data points from Kracek et al.3 Wu et al.9 reported an optimisation of the system using the odified quasichemical model (MQSM)14 for the liquid phase ased upon the phase diagram studies of Kracek et al.3,4 Their elting points of the compounds are close to the measured val- es but the eutectic temperatures as well as enthalpies of the ompounds are different. According to Saulov15, the most diffi- ult liquid binary oxide solutions to optimise are those, in which ne mixed component is a basic oxide, while the other is an cidic oxide. Liquids in such systems are usually characterised y very strong short-range ordering. One of the binary system ith very strong short-range ordering is the system K2O–SiO21. arge negative values of the Gibbs energy of the quasichemical eaction are usually required to fit available experimental data in uch systems. Furthermore, strong compositional dependencies f the Gibbs energy of the quasichemical reactions are usually sed. The feature was originated from the described non- moothness of the Gibbs energy of mixing expressed by he MQSM at the composition of maximum ordering. Previ- usly suggested values9 of the parameters of the MQSM were djusted15 to maintain acceptable fit of the experimental liq- idus data reported by Kracek et al.3 and the activities of K2O easured by Zaitsev et al.11, Frohberg et al.16, Ravaine et al.17, nd Steiler18. Saulov demonstrated that the compound whose omposition coincides with that of maximum ordering in the liq- id phase was required for reasonable representation of phase quilibria in the vicinity of this composition. Limiting slopes f the liquidus curve of the compound were also calculated. he thermodynamic model for the system K2O–SiO2 was re- ptimised. The new parameters of the model by Saulov15 were ound to fit the available experimental data and represent rea- onably the phase relations for the entire compositional range. Romero-Serrano et al.19 employed the structural model by in and Pelton20,21 for the liquid. The thermodynamic descrip- ion of the system was optimised by Yazhenskikh et al.22 with he modified associate species model23. Yazhenskikh et al.22 eported the set of model parameters that reproduce the avail- ble experimental data and enable representing reasonable phase elations for the entire phase diagram. In the model they used hree associates for describing the molten oxide phase. The most ecent assessment extending over the whole compositional range as been made by Zhang et al.24, using a two-sublattice model25 ith SiO44− and SiO2 associates in the anion sublattice. Wu et al.9, Zaitsev et al.12 as well as Romero-Serrano et al.19 upplied the calculated phase diagram only for the compositional ange with high silica content. In the present study, the phase equilibria of the K–O–Si sys- em in equilibrium with air have been investigated to revisit the esults reported by previous researchers as well as to confirm he positions of Quartz, Tridymite and Cristobalite liquidus. his work is part of the efforts to consolidate experimental hase equilibria data from K2O–SiO2 towards characterisation pean Ceramic Society 34 (2014) 4053–4058 4055 o s t w 2 d g ( c m t c w u d f r a s c f z e F f a a w G G. Akdogan et al. / Journal of the Euro f Cu–O–K2O–Na2O–SiO2 slags. It ultimately opens up pos- ibilities to extract property data from molten silica rich melts o develop different silicate fluxes in metallurgical processes as ell as to better control industrial furnace operations. . Experimental procedure The experimental technique adopted in this work has been eveloped by the University of Queensland and its Pyrosearch roup26. The initial mixtures were prepared from high-purity 99.99 wt pct purity) SiO2 powder from Umicore (Belgium) and arbonate powder of K2CO3 with 99.5+ wt pct purity (heavy etals ≤0.0005 wt pct) from Sigma–Aldrich (USA), by mixing horoughly in an agate mortar and pestle. The initial mixture bulk ompositions were selected to obtain liquid slag in equilibrium ith solid phases. The powder mixtures of less than 0.2 g were sed for each equilibration experiment. A quartz (SiO2) crucible with an inner diameter of 8 mm and epth of 5 mm, made by fusing high-purity silica rod, was used or experiments (Fig. 1). Equilibration experiments were carried out in a vertical eaction tube (Friatec AG, Germany; impervious recrystallised lumina; 45 mm OD and 38 mm ID) within electrical resistance ilicon carbide heated furnaces (Fig. 2). The experiments were arried out in a Lenton (Hope Valley, UK) PTF 15/-/450 tube urnace. During the experiments, the temperature of the hot one of the furnace was determined with a S-type thermo lement (Johnson-Matthey, UK) connected to Keithley 2000 ig. 1. Quartz crucible and sample suspended in the hot zone of the Al2O3 urnace work tube. c r e l m w s z i fi t t s e t w t t t w 1 l a b f t b w t i Fig. 2. Furnace arrangement for equilibration experiments. nd 2010 multimeter – data loggers (Cleveland, OH, USA) nd the ambient room temperature measurement was done ith a Pt100 sensor (Platinum Resistance Thermometer; SKS roup, Finland). The used thermocouple and Pt100 sensor were alibrated against the melting point of copper and ice water, espectively. The temperature measurement data during the xperiments were collected with a NI LabVIEW temperature ogging program. The overall accuracy of the temperature easurements was estimated as ±2 K. The sample hooked to the platinum wire (0.5-mm diameter) as introduced from the bottom of the reaction tube and was uspended in the hot zone of the furnace. After raised to the hot one, the sample crucible was kept at 50 ◦C above the equil- bration temperature for 30 min to homogenise the melt. Then nally it was held for equilibration at the final target temperature o complete the required equilibration period. The equilibration ime ranged from 4 h to 72 h depending on the sample compo- ition and temperature. Careful consideration was given to the quilibration time. Repeated experiments with longer equilibra- ion time were performed for a number of samples to check hether the equilibrium was achieved. Several dedicated sets of ime series experiments were carried out for obtaining informa- ion on equilibration time. It was found that the equilibrium in he K–O–Si systems in air at above 1300 ◦C was readily achieved ithin 4 h. The attainment of equilibrium was achieved within 6 h at temperatures between 1200 ◦C and 1300 ◦C. At relatively ow temperatures between 1000 ◦C and 1200 ◦C equilibrium was chieved within 24 h. In some cases such as at temperatures etween 770 ◦C and 1000 ◦C the equilibration period varied rom 24 h to 72 h. Experiments were completed with the bottom and the top of he reaction tube left open to the atmosphere. After the equili- ration time was reached, the bottom end of the reaction tube as attached with a glass bottle containing water and ice mix- ure and the specimen was then rapidly quenched by dropping nto the ice-water bath. Finally it was dried and mounted in 4 pean Ceramic Society 34 (2014) 4053–4058 e w t m d ( G I p e S e e i y F e a a e d e p d T c l d S f f q e w a o o l s w a r 3 i Q p I a w t Table 1 The measured compositions (EDS analysis) of liquid oxide and temperatures of primary crystallisation in K2O–SiO2 system along with the only previous experimental work in the literature, Kracek et al.3,30 Temperature Wt% SiO2 Solid phase ◦C Kracek et al.3,30 Present study 1713 100 Cristobalite 1635 95.5 1579 95 1541 93 1518 92 1505 91.5 1500 88.7, 87.7 147030 Cristobalite = Tridymite Transition 1460 89 Tridymite 1400 86.2, 85.0 1385 86 1300 84.2, 84.8 1250 82 1200 80.5 1149 79.8 1135 79.3 1129 78.9 1100 80.1, 80.5 1036 77.5 1015 77 1000 72.9, 72.8 935 75.8 905 74.8 900 71.8, 72.8 87030 Tridymite = Quartz Transition 840 73.7 Quartz 834 73.7 800 70.8, 68.8 770 71.23 71.4 7 c l T a i B t l the Tridymite liquidus obtained by the present study is in line with the findings of Kracek et al.3,30 Above 1400 ◦C the trend shows that the present results are slightly lower in SO2 than the Table 2 Comparison between phase compositions analysed with different methods, EDS and EPMA analysis, together with corresponding equilibration temperatures. Temperature Wt% SiO2 ◦C Present study EPMA Present study EDxS 1500 88.6(5), 88.0(8) 88.7, 87.7 1400 86.1(7), 84.7(8) 86.2, 85.0 056 G. Akdogan et al. / Journal of the Euro poxy resin. A polished cross section of the mounted specimen as prepared using dry metallographic grinding and polishing echniques. The polished specimens were carbon-coated and their icrostructures and phase compositions were examined imme- iately after experiment by Scanning Electron Microscopy SEM). A LEO 1450 (Carl Zeiss Microscopy GmbH, Jena, ermany) scanning electron microscope was used with a Link nca X-Sight 7366 Energy EDS analyzer (Oxford Instruments lc, Abingdon, Oxfordshire, UK). The accelerating voltage mployed was 15 kV and employed standards were Quartz (SPI upplies Ltd., USA) for Si and O and Sanidine (Astimex Sci- ntific Ltd., Toronto, Canada) for K. The spectral lines used for ach element were K�. The compositions of the phases in part of the samples were n addition measured by electron probe X-ray micro anal- ser (EPMA). The EPMA used at the Geological Survey of inland (GTK) was a CAMECA SX100 (Cameca SAS, France) quipped with five wavelength dispersive spectrometers. The nalyses were performed using 15 kV accelerating voltage and probe current of 10 nA. Particular attention was paid to the xposure time to the electron beam in EPMA in order to avoid epletion of potassium under the beam, as reported by Sawyer t al.27. The electron beam was defocused to 10 �m to avoid sam- le damage and intensity loss of alkalis. The raw measurement ata was corrected using PAP matrix correction procedure28. he following standards and X-ray lines were used for Cu pure opper metal (Cu K�), Al2O3 as Al and oxygen standard (K� ines for both elements), albite was used for sodium (K�) and iopside as Ca standard (K�). All standards were from Astimex cientific ltd. There was no difficulty in the quenching of the liquid rom which a glassy, fully non-crystalline phase was readily ormed on quenching. Examples of typical microstructures of the uenched samples in K–O–Si system (backscattered scanning lectron micrographs) are given in Fig. 3. The compositions of the liquid and the crystal phases there ere measured by the EDS-point analysis methods immediately fter the experiments. The point analysis mode was selected in rder to shorten the counting time, due to the high mobility f potassium in silicates under electron beam29. Three different ocations with at least six points in each were measured for every ample. Part of the samples were corrupted due to moisture while aiting to be analysed with EPMA, in spite of being stored in desiccator, and thus only for a part of the experiments EPMA esults can be provided. . Results and discussion The EDS measurement results of the phase equilibria exper- ments of the K–O–Si system in equilibrium with air in the uartz, Tridymite and Cristobalite primary phase fields at tem- eratures between 770 ◦C and 1500 ◦C are reported in Table 1. n Table 2, the EPMA results of those samples not corrupted re presented together with EDS results. No inhomogeneities ere found in the glassy phase. The agreement between these wo measurement methods is good. The reason for sample 1 1 6430 K2O*4SiO2(KS4) + SiO2 orruption under time is likely to be moisture. The uncertain ast digit of the EPMA results has been given in parenthesis in able 2. The trend in SiO2 concentrations in the liquid at the quartz nd Tridymite liquidus between 770 ◦C and 900 ◦C obtained n the present study is lower that reported by Kracek et al.30 etween 900 ◦C and 1100 ◦C the Tridymite liquidus obtained by he present study is found to be lower in SiO2 than the Tridymite iquidus reported by Kracek et al.30 From 1100 ◦C to 1400 ◦C 300 82.8(9), 83.6(5) 84.2, 84.8 000 72.9(5) 72.9, 72.8 800 70.6(9), 69.7(2) 70.8, 68.8 G. Akdogan et al. / Journal of the European Ceramic Society 34 (2014) 4053–4058 4057 F ples s 1 T c o t e a o n e r p s u F r M K t t m c 4 s l a ig. 3. Backscattered scanning electron micrographs of quenched K–O–Si sam 000 ◦C (b). ridymite-Cristobalite liquidus reported by Kracek et al.3,30 A omparison is presented in Fig. 4. The solid lines in Fig. 4 are calculated phase boundaries based n the Mtox database by NPL32. In the database, the experimen- al phase diagram observations by Kracek et al.3,30 and Eliezer t al.5 and activity data by Frohberg et al.16, Ravaine et al.17 nd Steiler18 have been assessed using an associate model with nly two associates in the liquid phase, in order to minimise the umber of parameters to be adjusted in the assessment.33 The calculated silica liquidus is in good agreement with the xperimental points of Kracek et al.3,30 but the current equilib- ium data suggest smaller silica solubility in the liquid phase, in articular at low temperatures, below 1200 ◦C. The current mea- urements below 1000 ◦C, where very long equilibration times p to 72 h, varying initial compositions, different preheating ig. 4. Pseudo-binary phase diagram of the “K2O”–SiO2 system in equilib- ium with air; the solid lines have been calculated using the Mtox database and TDATA software [31] (w denotes to weight fraction, KS2 to K2Si2O5, and S4 to K2Si4O9). o T e 1 t i o o p A E U I a R aturated with solid SiO2 (Q crucible) in equilibrium with air at 1500 ◦C (a) and emperatures and small samples of less than 0.2 g for ensuring he equilibrium were used, also suggest that KS4 (or K2O·4SiO2) elts incongruently, as proposed by Zaitsev et al.10 and not ongruently as assumed by the assessments.,13,19,23,24 . Summary and conclusions The phase equilibria and liquidus temperatures in the K–O–Si ystems have been determined in air. The compositions of the iquid and solid phases have been measured using SEM-EDS nd electron probe X-ray microanalysis techniques. The effect f K2O on the compositions of liquidus curves of Quartz, ridymite, and Cristobalite primary phases of the system in quilibrium with air have been investigated between 770 ◦C and 500 ◦C and the results are in relatively good agreement with he earlier work of Kracek et al.3,4,30 In the future thermodynamic assessments it must be taken nto account that the most silica-rich potassium silicate K4S bviously melts incongruently, forming solid silica and a molten xide phase with a slightly lower silica concentration than it was reviously suggested. cknowledgements The authors are indebted to Outotec for the provision of the PMA analyses and funding the experimental work at Aalto niversity. Financial support by Tekes and its ChemEnergy and SS (ELEMET programme of Fimecc Oy) projects is also kindly cknowledged. eferences 1. Nowakowski DJ, Jones JM. Catalysis by potassium in the pyrolysis pro- cesses of biomass and basic biomass components. In: Imbabi MS, Mitchell CP, editors. World Renewable Energy Congress (WREC 2005). Amsterdam: Elsevier Ltd.; 2005. p. 590–5. 2. Nordin A. Chemical elemental characteristics of biomass fuels. Biomass Bioenergy 1994;6(5):339–47. 4 pean C 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 058 G. Akdogan et al. / Journal of the Euro 3. Kracek FC, Bowen NL, Morey GW. Equilibrium relations and factors influencing their determination in the system K2SiO3–SiO2. J Phys Chem 1937;41(9):1183–93. 4. Kracek FC. The cristobalite liquidus in the alkali oxide-silica systems and the heat of fusion of cristobalite. J Am Chem Soc 1930;52(4):1436–42. 5. Eliezer N, Howald RA, Marinkovic M, Eliezer I. Vapor pressure measurements, thermodynamic parameters, and phase diagram for the sys- tem potassium oxide-silicon oxide at high temperatures. J Phys Chem 1978;82(9):1021–6. 6. Kim SS, Sanders TH. Thermodynamic modeling of phase diagrams in binary alkali silicate systems. J Am Ceram Soc 1991;74(8):1833–40. 7. Allendorf MD, Spear KE. Thermodynamic analysis of silica refractory cor- rosion in glass-melting furnaces. J Electrochem Soc 2001;148(2):B59–67. 8. Phase diagrams for ceramists, Vols. I–XII. Westerville, OH: The American Ceramic Society; 1964–1996. 9. Wu P, Eriksson G, Pelton AD. Optimization of the Thermodynamic Proper- ties and Phase Diagrams of the Na2O–SiO2 and K2O–SiO2 Systems. J Am Ceram Soc 1993;76(8):2059–64. 0. Zaitsev AI, Shelkova NN, Lyakishev NP, Mogutnov BM, Khim ZF. Mod- eling of the thermodynamic properties of silicate melts and calculation of phase equilibria in the K2O–SiO2 system. Russ J Phys Chem (Engl Transl) 2000;74(7):1033–8. 1. Zaitsev AI, Shelkova NN, Lyakishev NP, Mogutnov BM, Khim ZF. The thermodynamic properties of K2O–SiO2 melts. Russ J Phys Chem (Engl Transl) 2000;74(6):907–13. 2. Zaitsev AI, Mogutnov BM. Thermodynamic properties and phase equilibria in the MnO–SiO2 System. J Mater Chem 1995;5(7):1063–73. 3. Forsberg S. Optimization of thermodynamic properties of the K2O–SiO2 system at high temperatures. J Phase Equil 2002;23(3):211–7. 4. Pelton AD, Blander M. Thermodynamic analysis of ordered liquid solutions by a modified quasichemical approach – application to silicate slags. Metall Trans B 1986;17B(6):805–15. 5. Saulov D. Application of the modified quasichemical model to solutions with strong short-range order. J Non-Cryst Solids 2007;353(30–31):2845– 51. 6. Frohberg MG, Caune E, Kapoor ML. Measurement of the activity of oxygen ions in the liquid systems Na2O–SiO2 and K2O–SiO2. Arch Eisenhüttenwes 1973;44(8):585–8. 7. Ravaine D, Azandegbe E, Souquet JL. Mesures potentiometriques de chaines electrochimiques comprenent des silicates fondus; interpretation des resultants par un modele statistique. Silic Ind 1975;40:333–40. 3 3 eramic Society 34 (2014) 4053–4058 8. Steiler JM. Données thermochimiques pour la sidérurgie, vol. 2. Comm. Eur. Communities, Techn. Rept EUR 7820; 1982. 9. Romero-Serrano A, Gomez-Yanez C, Hallen-Lopez M, Araujo-Osorio J. Thermodynamic modeling of alkali metal oxide-silica binary melts. J Am Ceram Soc 2005;88(1):141–5. 0. Lin PL, Pelton AD. A structural model for binary silicate systems. Metall Trans B 1979;10B(6):667–76. 1. Romero-Serrano A, Pelton AD. Extensions of a structural model for binary silicate systems. Metall Mater Trans B 1995;26B(2):305–15. 2. Yazhenskikh E, Hack K, Muller M. Critical thermodynamic evaluation of oxide systems relevant to fuel ashes and slags. Part I: Alkali oxide-silica systems. Calphad 2006;30(3):270–6. 3. Besmann TM, Spear KE. Thermodynamic modeling of oxide glasses. J Am Ceram Soc 2002;85(12):2887–94. 4. Zhang L, Schmetterer C, Masset P. Thermodynamic description of the M2O–SiO2 (M = K, Na) systems. Comput Mater Sci 2013;66(1):20–7. 5. Hillert M, Jansson B, Sundman B, Ågren J. A two-sublattice model for molten solutions with different tendency for ionization. Metall Trans A 1985;16A(2):261–6. 6. see e.g. Hidayat T, Henao MH, Hayes PC, Jak E. Phase equilibria stud- ies of Cu–O–Si systems in equilibrium with air and metallic copper and Cu–Me–O–Si systems (Me = Ca, Mg, Al, and Fe) in equilibrium with metal- lic copper. Metall Mater Trans B 2012;43B(6):1290–9. 7. Sawyer R, Nesbitt HW, Secco RA. High resolution X-ray Photoelec- tron Spectroscopy (XPS) study of K2O–SiO2 glasses: evidence for three types of O and at least two types of Si. J Non-Cryst Solids 2012;358: 290–302. 8. Pouchou JL, Pichoir F. Un nouveau modèle de calcul pour la microanalyse quantitative par spectrométrie de rayons X. Partie I: Application a l’analyse d’énchantillons homogènes. La Recherché Aérospatiale 1984;3:167–92. 9. Gedeon O, Zemek J, Jurek K. Changes in alkali-silicate glasses induced with electron irradiation. J Non-Cryst Solids 2008;354:1169–71. 0. Kracek FC, Bowen NL, Morey GW. The system potassium metasilicate- silica. J Phys Chem 1928;33(12):1857–79. 1. Davies RH, Dinsdale AT, Gisby JA, Robinson JAJ, Martin SM. MTDATA – thermodynamic and phase equilibrium software from the National Physical Laboratory. Calphad 2002;26(2):229–71. 2. Vaajamo I, Gisby J, Taskinen P. An extensive slag database: Lead-ing the way. In: Proc. Copper-Cobre 2013 Conference. 2013. p. 621–34. 3. Dinsdale AT, Gisby JA. Notes on Progress 3, MIRO RC81 Phase II. Ted- dington, UK: National Physical Laboratory; 1995 (private communication). Phase equilibria study of K-O-Si system in equilibrium with air 1 Introduction 2 Experimental procedure 3 Results and discussion 4 Summary and conclusions Acknowledgements References