H2O Diffusion in Silicate Glasses and Melts

Abstract

Diffusion of H20 in silicate melts and glasses plays a critical role in bubble growth that powers explosive volcanic eruptions and in stability of glass. Although extensive work has been carried out, there are still some major gaps in our understanding of the diffusion process. This work is aimed at understanding how the diffusivity of H20 depends on the anhydrous composition, the pressure, and the dissolved total H20 concentration (H2Ot). Samples used in this study include KS and Rhy (natural obsidian glass from Mono Craters, California, USA; initial H2Ot 0.8% by weight for KS and 1.9% for Rhy), NSL (natural peralkaline rhyolitic glass from New Zealand, with 2.0% HaOt), Ab (synthetic albite glass with 1.8% initial H2Ot), AOQ (synthetic haplogranitic glass with nominal composition of Qz28Ab38Or34 containing either 1.3% HzOt or 2.2%), T3 (synthetic glass with nominal composition of Qz50An50 containing 2.0% H2Ot). Table 1 lists the composition of the samples measured by electron microprobe. Dehydration experiments were carried out at 584-605~ and 0.5 to 5 kbar Ar pressure in a rapid-quench cold-seal furnace. After heating of 16 to 73 hours, each quenched experimental wafer is sectioned to obtain a slice from which the H20 diffusion profile is measured by infrared spectroscopy. The Brthker IFS88 IR spectrometer with microscope (A-590) setup at Hannover is used with a slit width of 5 to 25 ~tm. The H2Ot concentration is obtained by summing up the OH and molecular H20 (H2Om) species concentrations measured at 4520 cm -~ and 5230 cm -1 combination bands. All H2Ot concentration profiles are well fit by assuming that H2Ot diffusivity (DH2ot) is proportional to H2Ot (Fig. 1). They can also be fit well by assuming that molecular H20 is the diffusing species and DH2Om is independent of H20 t. These results are consistent with both Zhang et al. (1991) and Nowak and Behrens (1997) since H20 t is relatively low. The effect of different fitting procedures to the resulting DH2ot is examined and found to be within 20% relative. Experimental data show that DH2Ot decreases with increasing pressure. For AOQ composition at 600 ~ and 1% H2Ot, DH2ot decreases from 0.25 I.tm2/s at 0.5 kbar, to 0.133 I/m2/s at 2 kbar, and to 0.080 ktm2/s at 5 kbar. The activation volume for H20 diffusion inferred from the data is 17___8 (2ty error) cm3/mol, somewhat greater than but still in agreement within experimental error with that inferred from high temperature diffusivity data for AOQ (5 to 11 cm3/ tool, Nowak and Behrens, 1997). The dependence of D~2ot on anhydrous composition is examined at 5 kbar Ar pressure and 600~ DHzot at 1 wt.% H2Ot is 0.152, 0.080, 0.070, and 0.053 pm2/s for NSL, AOQ, Ab and KS/Rhy samples. For the T3 sample, the diffusion profile is too short to be measured, implying a maximum DH2ot at 1% H2Ot of ~0.002 ~tm2/s, much less than that in other glasses. For comparison, DiJ2ot in SiO2 glass (Moulson and Roberts, 1961) at 600~ and 1 bar is 0.044 ktm2/s by a