The heat capacity of multicomponent hydrous natural melts and glasses

The thermophysical properties of silicate melts and glasses are of fundamental importance for the characterization of the dynamics and energetics of silicate melts on Earth and terrestrial planets. The heat capacity of silicate melts is of particular importance because of its implications for the temperature dependencies of melt enthalpy and entropy and for the potential relationship to melt structure and transport properties. Currently, there are reliable models for predicting the heat capacity of simple and multicomponent silicate glasses (C-p(glass)) as a function of composition and temperature. Recent differential scanning calorimetry (DSC) measurements of heat capacity for multicomponent silicate liquid (C-p(liquid)), however, have shown that published models do not accurately reproduce heat capacity measurements on some silicate melts. Here, we have compiled a database of heat capacity values for hydrous and anhydrous multicomponent natural samples. The measurements are on pairs of glasses and melts over the compositional range (wt%) of: SiO2 (44-79), Al2O3 (5-35), TiO2 (0-3), FeOtot (0 - 11); Na2O + K2O (0-27); CaO + MgO (0-39), H2O (0-6.3) and minor oxides. The compiled data show strong correlations between silica content (X-SiO2) and the configurational heat capacity (C-p(config)) defined as C-p(liquid) - C-p(glass) measured across the glass transition temperature (T-g). This correlation is used to establish an empirical model for predicting C-p(liquid) as a function of melt composition (i.e. SiO2 content) and values of C-p(glass) measured at the onset of the glass transition: C-p(liquid) = 52.6-55.88X(SIO2) + C-p(liquid) The model reproduces values of C-p(liquid) to within an average relative error of similar to 2.4%. Published models for the heat capacities of silicate melts (e.g., Stebbins, 1984; Richet and Bottinga, 1985; Lange and Navrotsky, 1992) applied to the same dataset have average relative errors in excess of 5.5%.