In situ IR Spectroscopy on Hydrous Albitic and Rhyolitic Glasses and its Implications for Water Speciation and Water Species Reactions in Silicate Glasses and Melts

Abstract

The interpretation of spectroscopic and experimental data on speciation of water in silicate melts and glasses has been controversial. There are two experimental approaches using IR absorption spectroscopy for studying the equilibrium speciation of H20 as a function of temperature (and pressure) and total water content. One method is by holding the silicate glass at high temperature to reach equilibrium, then rapidly quenching it to room temperature and measuring IR spectra at room temperature (e.g. Zhang et al., 1991). In the other method IR spectra are measured in situ as the sample is held at high temperature and pressure (Nowak and Behrens, 1995; Shen and Keppler, 1995). The in situ spectra show systematic changes in band intensities with temperature even in the glass state. Assuming that the molar extinction coefficients do not depend on temperature the authors of the in situ studies have inferred that (i) the speciation varied in the glass state (25-400~ and (ii) the equilibrium constant for the speciation reaction depends more strongly on temperature than inferred from data using the quench technique. It is important to resolve whether the difference of the results from the two techniques is owing to the inability to quench species concentrations even from 400~ for the quench technique, or to the temperature dependence of extinction coefficients or other physical properties of the glass (e.g. density) for the in situ technique. In order to resolve the apparent contradictions of the two experimental approaches we have measured near-infrared spectra of hydrous albitic glass (0.5-9.2 wt.% total water) and rhyolitic glasses (0.2-6.0) in situ in a wide temperature range ( -192 to +600~ depending on water content) as well as after quenching from high temperature. Because different baseline corrections have been applied to determine the concentrations of OH groups and molecular H20 from the combination bands at 4500 and 5220 cm -1, respectively, we have tested the equivalence of these methods for the albitic composition (9 samples). The baselines chosen are (i) a linear baseline for the band at 5220 cm -1, extension of this line to low frequency combined with a gaussian fitted to the band at 4000 cm -1 as the baseline for the band at 4500 cm -1 (Behrens et al., 1996), (ii) linear baselines for both bands (Shen and Keppler, 1995) and (iii) flexicurve baselines for both bands (Newman et al., 1986). If the molar extinction coefficients were independent on total water, OH and H20 concentrations differ by at most 3 % relative if absorbances (peak heights) are evaluated (for baselines i, ii and iii) or if integrated intensities (peak areas) are evaluated (for baselines i and ii). However, integrated intensities gives systematically lower OH and higher H20 (up to 10% relative) at low total water and higher OH and lower H20 (up to 7 % relative) at high total water. We suggest that the systematic differences may be due to the variation of the linear molar extinction coefficient with water content in accordance with results of Zhang et al. (1997). IR spectra were measured in situ using a heating/ cooling stage fitted to an IR microscope. Fig.1 display an example which is typical for the albitic and rhyolitic glasses. Systematic variation with temperature even close to liquid nitrogen temperature are observed for both absorbances and integrated intensities. For the H20 band the absorbance has a maximum close to room temperature whereas the integrated intensity is continously decreasing with temperature. For the OH band both the absorbances and the integrated intensities are continously increasing with temperature. For albitic glasses the T-dependence of the integrated intensity of the H20 band is almost independent on water content (about 3% per 100K) and only is weakly effected by the type of baseline correction. In contrast, the OH band is