64 Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part B Volume 47 Number 1 February 2006 Introduction Below the liquidus temperature, a crystalline phase can exist at equilibrium with molten glass. Since preventing crystallisation is of utmost importance for both glass processing and glass forming, the liquidus temperature is one of the properties that are taken into account when an optimised glass composition is being formulated. However, optimising glass formulation is a formidable task, considering the mul- ticomponent nature of glass and the need to constrain several properties that are limited by the processing technology or required for the inal product. A trial and error approach to this task would be rather costly and lengthy, but fortunately mathematical modelling simpliies it considerably. Mathematical modelling for optimised glass formu- lation is based on developing functional relationships between glass properties and composition. Each prop- erty is represented by a suitable mathematical formula that contains a number of coeicients to be ited to data obtained experimentally for a limited number of com- positions, which are usually statistically designed. Once each key glass property is mathematically represented, an optimised composition that satisies a number of requirements can be computed. Because the models are subjected to uncertainties, experimental veri ication and further reinement are necessary before an optimised glass is produced on a large scale. In a restricted composition region, a single sim- ple mathematical formula approximates actual the behavior for most properties, such as viscosity or chemical durability, with suicient accuracy. This, however, is not the case for the liquidus temperature, which is a smooth function of composition only within a single primary phase-ield. This paper re- ports the applicability of several published models to liquidus temperature data for silica, wollastonite, and devitrite within a narrow composition region typical of loat glass. To cover both primary and secondary crystalline phases, let us consider the maximum temperature (T J ) at which J-th crystalline phase is in equilibrium Effect of float glass composition on liquidus temperature and devitrification behaviour Pavel Hrma, (a),1 D. E. Smith, (a) Josef Matyáš, (b) J. D. Yeager, (c) J. V. Jones(d) & Edward N. Boulos (d) (a) Paciic Northwest National Laboratory, Richland, Washington, USA (b) Laboratory of Inorganic Materials of the Institute of Inorganic Chemistry of Academy of Sciences of the Czech Republic and the Institute of Chemical Technology, Prague, Czech Republic (c) Washington State University, Pullman, Washington, USA (d) Visteon Corporation, Nashville, Tennessee, USA Manuscript received 4 July 2003 Revised version received 9 November 2005 Accepted 18 November 2005 Liquidus temperatures (T L ) were measured for the following loat glass-type composition region (in mass%): 72·7 to 74·0 SiO 2 , 13·1 to 14·2 Na 2 O, 7·95 to 8·95 CaO, 2·97 to 3·97 MgO, 0·10 to 0·45 Al 2 O 3 , and 0·03 to 0·10 K 2 O. Glasses also contained constant minor fractions of Fe 2 O 3 (0·71) and TiO 2 (0·01). Fractions of silica, wollastonite, and devitrite were determined in glasses quenched from 900°C. Partial speciic values for T L were evaluated for silica and wollastonite phase-ields. The measured T L values were compared with values estimated using various models available in the literature. The diferences between predicted and measured T L for the loat glass composition region can be atributed to several causes, the most prominent being the neglecting of diferences in the slopes of liquidus surfaces within diferent primary phase-ields. Inaccurate estimates can also be expected when a model is applied to a glass with a smaller or a larger number of key components. Finally, erroneous estimates occur when the model is extrapolated beyond the composition region covered by data to which model equations were ited, or when a model that covers a large composition region is applied to a smaller subregion where the T L –composition relationship has a signiicant lack of it. 1 Corresponding author. Email pavel.hrma@pnl.gov Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, June 2005, 47 (1), 64–76