Diffusion of SiO2 in Rhyolitic Melt

Abstract:

SiO₂ is the most major component in silicate melts, and the diffusion of SiO₂ plays a controlling role in growth or dissolution of quartz from or into silicate melts. The diffusivity of SiO₂ is small and highly dependent on melt compositions (Lesher and Walker, 1986; Koyaguchi, 1989), making it difficult to extract high-quality Si diffusivity data.

We conducted quartz dissolution experiments in rhyolitic melt (0.1wt% H₂O, 73 wt% SiO₂) at 1300-1600 °C. We also have preliminary data on quartz dissolution in basaltic melt at 1300 °C , one quartz dissolution experiment (Zhang et al., 1989) in andesitic melt at 1300 °C , and five cassiterite dissolution experiments (Yang et al, in review) in various hydrous rhyolitic melts (containing 0.1-5.9 wt% H₂O, and 74-77 wt% SiO₂) at 900-1100 °C. All experiments were conducted at 0.5 GPa using piston cylinder apparatus. All data were combined to examine the dependence of D_SiO2 on melt compositions. Though in individual experiments lnD_SiO2 is a linear function of SiO₂ concentration as shown in literature, the combined data show that lnD_SiO2 decreases linearly with X_Si+Al, where X_Si+Al is defined as cation mole fraction of Si+Al in melts. By fitting concentration profiles at different temperatures using D_SiO2 = D₀ e^a(1-X_Si+Al), the results show that the parameter a is roughly composition-independent across all experiments and is linear to 1/T:

a = 2.603(±0.451) + 35282(±627)/T, r²= 0.996.

D₀ is the extrapolated effective binary diffusivity of SiO₂ in pure silica melt (X_Si+Al = 1 roughly corresponds to X_Si = 1 in quartz dissolution experiments). For quartz dissolution experiments in rhyolitic melt, the dependence of D₀ on T is:

lnD₀ = -14.041(±1.915) - 34719(±3125)/T, r²= 0.875

For cassiterite dissolution, lnD₀ values do not follow the above trend because the dominant SnO concentration gradient can affect interdiffusion between SiO₂ and other components. That is, X_Si+Al alone is not enough to account for how D₀ depends on other components. The effect of H₂O on both a and D₀ is roughly accounted for by simply including H as a cation when calculating X_Si+Al. Namely, adding H₂O equals to lowering X_Si+Al, e.g. at 1300°C, every addition of 1 wt% H₂O (i.e. lowering X_Si+Al by ~0.05) would increase D_SiO2 by ~1.2 lnD units for a melt with initial X_Si+Al = 0.85 on anhydrous basis.