EFFECT OF DISSOLVED WATER ON PHYSICAL PROPERTIES OF SODA-LIME-SILICATE GLASSES

Udaya K. Vempati

Research and Development, Owens-Illinois, Inc.
One Michael Owens Way, Perrysburg, OH 43551, USA

Terence J. Clark1

Bowling Green, OH 43402, USA

ABSTRACT

Dissolved gases in glass melts are known to influence properties of the melts as well as the resulting glass and dissolved water is thought to be one of the most influential of all the dissolved gases. In this work, the effect of vacuum processing and the ensuing changes in dissolved water concentration on various physical properties of soda-lime-silica glasses were studied. Glass melts with varying dissolved water concentration were prepared by melting frit at atmospheric and sub-atmospheric (~100 torr) pressures at 1450°C. The physical properties of these melts and the resulting glasses were determined by rotating spindle viscometry, beam bending viscometry, and spectroscopy. The densities of the glass samples were also determined. Results from these experiments are discussed in relation to prior work in the field and the implications of changes in properties on the glass making process are discussed.

INTRODUCTION

The presence of dissolved gases and their influence on properties of soda-lime-silica, as well as other silicate melts and glasses, is well known and characterized in detail [1-10]. A common route by which gases are introduced in glasses is through batch constituents, which may contain both physically adsorbed and chemically bonded gases. For example, both limestone (CaCO3) and soda ash (Na2CO3), commonly used in the synthesis of soda-lime-silica glasses, release CO2 during melting of the glass batch. Much of this gas escapes the melt in the form of bubbles during melting and fining, but some can remain dissolved in the melt. Depending on the raw materials used, many other gases can be present in a soda-lime-silica melt including H2O, SO2, O2, NO, and NO2. These gases may also remain in the resulting solidified glass. When the glass melt is subject to sub-atmospheric pressure or vacuum, the dissolved gas concentration is expected to change. Consequently, the physical and chemical properties of glasses obtained from vacuum-processed melts can be expected to be different from corresponding glasses obtained from atmospheric or ambient melts. Variations in atmosphere surrounding the melt, particularly in the oxygen content, could also influence the properties of glasses from vacuum-processed melts.

Several earlier reports have discussed the influence of vacuum processing on the properties of silicate glasses. For instance, Fenstermacher et al. [2] measured the dissolved water concentration and viscosity of soda-lime silicate glasses obtained from glass batches melted at both ambient pressure and under vacuum. They found that the softening point of glass samples resulting from vacuum melting can be as much as 5°C higher than the glasses obtained from ambient melts and attributed this difference to the lower dissolved water content in the vacuum melt samples. Similarly, Graff and Badger [9] measured the viscosity of glasses (presumably soda-lime silicates) obtained from melts subject to vacuum, as well as those saturated with carbon dioxide and water, and found that, at a given temperature, the samples from vacuum melts generally had higher viscosities than those from saturated melts. Sproull and Rindone [11] melted lithium-potassium silicates under vacuum and found that fibers drawn from these vacuum melts have lower strength, by as much as 25%, than those produced from ambient melts. They attributed the reduction in strength to a heterogeneous microstructure brought about by low levels of oxygen in the atmosphere surrounding the melt under vacuum.

Most of this early work focused primarily on a specific property of the glass, such as viscosity or mechanical strength. However a more thorough and comprehensive investigation of the effects of vacuum processing on various properties of soda-lime silica glasses and melts is lacking. To address this gap, viscosity, density, Fourier transform infrared (FTIR) and ultraviolet (UV)-Visible (Vis) transmission spectra of soda-lime silicate glasses obtained from vacuum melts were measured in the present study. These data were compared with the corresponding properties of glasses produced by standard practice, i.e. from melts prepared at atmospheric pressure. Any difference in properties between the two sets of glasses was explained in context of differences in dissolved water concentration and/or processing glass under an oxygen-deficient atmosphere. It should be pointed out that although many different gases can be dissolved in a silicate glass, the focus of this work was limited to water, because water is thought to be the dominant species among dissolved gases in soda-lime-silica glasses [12].

The central finding of this work was that vacuum processing has a measureable impact on the viscosity and UV-Vis transmission spectra of a soda-lime-silica glass. Specifically, at viscosities corresponding to melting and gob formation, the temperatures of vacuum processed melt can differ by about 10°C from the ambient melt. This result is consistent with those previously reported by others and can be attributed to the lower dissolved water concentration in the vacuum melt [2]. The dominant wavelength in the UV-Vis transmission spectrum of the vacuum processed flint glass is found to shift to lower values compared to a standard flint glass. Changes in redox brought about by the vacuum processing may have some bearing on this result. However, the density of the glass was not effected by vacuum processing. Mechanical properties, critical from an end use viewpoint, have not been characterized here and remain a topic for future work.

EXPERIMENTAL METHODS

Details of sample preparation and data collection methods are provided in this section. Note that x-ray fluorescence (XRF) measurements were carried out to establish composition of a vacuum processed glass sample. Within experimental error, the composition of this sample was found to be similar to a sample prepared by standard methods. Hence, except for their dissolved gas content, both kinds of glasses are treated to be of similar chemical composition in this work.

Batching and melting

Soda-lime-silica glass frit was prepared by melting batch materials in an electric furnace at ambient pressure. Raw materials from bulk suppliers were used in preparing glass batches of nominal composition, in mass fraction, 73.7% SiO2 - 13.6% Na2O - 11.3% CaO - 1.4% A12O3. The frit was then melted for 2.5 h in an evacuated furnace, where the pressure was maintained between 13.3 kPa and 17.3 kPa. This is referred to as "vacuum-processed glass". The frit was also melted in ambient, i.e. at 101.3 kPa, for 2.5 h to obtain glass samples that are representative of glasses produced by existing commercial practices. This is the "ambient-processed glass".

FTIR, UV-Vis transmission, and density measurements

Samples for FTIR experiments were prepared as described above by first melting ~300 g of frit at 1450°C for 2.5 h either in ambient or in vacuum. These melts were then poured onto a room temperature metal plate and pressed into discs. The discs were annealed at 550°C for 10 min and subsequently slow cooled to room temperature. They were then cut to size (3.81 cm diameter) using a core drill and polished on both sides. The polished discs were placed in an FTIR spectrometer sample chamber and purged with nitrogen gas for 15 min before collecting absorption spectra. A background scan was also collected with an empty sample chamber after a fifteen minute purge with nitrogen gas. Absorption spectra were used to calculate the dissolved water concentration in glass samples as will be described in the results section below.

Glass prisms measuring 35 mm × 35 mm × 13 mm for UV-Vis transmission were also prepared from the melts used for producing the FTIR samples. A small section, roughly 2 mm thick, was cut from the 35 mm x 13 mm face of the prism using a diamond saw. The cut faces were ground and polished to a mirror finish. A UV-Vis spectrophotometer was used for the transmission measurements.

Density measurements were carried out on crushed glass pieces obtained from the same sample set as used for FTIR and UV-Vis spectroscopy measurements by the Archimedes method. Deionized water with a few drops of surfactant was used as the weighing medium. A wire mesh, hung from a support frame connected to a balance, was used to suspend the samples in the water. A thermometer was clipped to the water beaker to continuously monitor the water bath temperature. Six glass samples were used from two different batches at each condition, vacuum and ambient processing, for the density measurement.

Viscosity

Viscosity measurements were carried out by two different methods. A rotating spindle viscometer was used to measure viscosities in the (101-104) Pa-s range, whereas a beam bending viscometer was used for measurements in the (108-1013) Pa-s range. Combining data from these two measurement series, viscosity of the glass in the (101-1013) Pa-s range was obtained by interpolation.

For the rotating spindle viscometer...