Melt inclusion analysis of the Unzen 1991–1995 dacite: implications for crystallization processes of dacite magma

Abstract

Dacitic magma, a mixture of high-temperature (T) aphyric magma and low-T crystal-rich magma, was erupted during the 1991–1995 Mount Unzen eruptive cycle. Here, the crystallization processes of the low-T magma were examined on the basis of melt inclusion analysis and phase relationships. Variation in water content of the melt inclusions (5.1–7.2 wt% H₂O) reflected the degassing history of the low-T magma ascending from deeper levels (∼250 MPa) to a shallow magma chamber (∼140 MPa). The ascent rate of the low-T magma decreased markedly towards the emplacement level as crystal content increased. Cooling of magma as well as degassing-induced undercooling drove crystallization. With the decreasing ascent rate, degassing-induced undercooling decreased in importance, and cooling became more instrumental in crystallization, causing local and rapid crystallization along the margin of the magma body. Some crystals contain scores of melt inclusions, whereas there are some crystals without any inclusions. This heterogeneous distribution suggests the variation in the crystallization rate within the magma body; it also suggests that cooling was dominant cause for melt entrapment. Numerical calculations of the cooling magma body suggest that cooling caused rapid crystal growth and enhanced melt entrapment once the magma became a crystal-rich mush with evolved interstitial melt. The rhyolitic composition of melt inclusions is consistent with this model.

Appendices

Appendix A:

Thermal evolution model in the magma body

The calculation method used for the modeling of a cooling magma body is described here. Piwinskii and Wyllie (1968) heated a granodiorite (sample 766 with 69.4% SiO₂, similar to Unzen low-T magma) under water-saturated conditions at 200 MPa, and reported the relationship between melt fraction and temperature. About 50 vol% melting or crystallization occurs within a small temperature interval near to solidus (Fig. A1a). In order to approximate this behavior, we used linear relationships between the temperature T and crystal fraction ?as

Fig. A1

a Relationship between melt fraction and temperature (solid line) for water-saturated granodiorite composition at 200 MPa (modified after Piwinskii and Wyllie 1968). Starting material is Sample 766 with 69.4 wt% SiO₂ in Piwinskii and Wyllie (1968). The stable phase assemblage changes at the intersections of solid line and dotted lines. b Assumption of a linear relationship between temperature and crystal fraction used in the thermal evolution modeling. T _S and T _L denote the solidus and liquidus temperatures, respectively

$$\phi=0.5(T_\text{L} - T)/(T_\text{L} - 730)\quad (730 \leqslant T < 1000)$$

(A.1)

and

$$\phi=0.5 + 0.5(730 - T)/(730 - T_S)\quad (700 \leqslant T < 730),$$

(A.2)

where T _L and T _S are the liquidus and solidus temperatures, respectively. The values used for calculation are shown in Table 3. We assume a dacitic composition for both the magma and the country rock, because a petrological study has suggested that the upper crust has dacitic (granodioritic) compositions in this area (Yokose and Yamamoto 1996). Even though the country rock is not saturated with water, the melting occurs under water-saturated conditions as far as melt, magma-derived vapor and solid can coexist at the solid-magma interface. We, therefore, consider a single relationship between crystal fraction and temperature (Fig. A1b) for both melting and crystallization event (Huppert and Sparks 1988a). The crystals in the magma body are assumed to be small enough to remain suspended in the vigorously convecting magma (Sparks et al. 1984) for simplification.

We consider a layer of hot dacitic magma to have initial temperature T ₀ and vertical thickness of 500 m, and that the solid roof has initial temperature T _∞ and is of large thickness. The mode of heat and mass transfer changes greatly at a critical melt fraction bounding fluidal magma and immobile magma with a large crystal content (partially molten solid, hereafter). Above the critical melt fraction (C&M stage), heat and mass are efficiently transferred by vigorous convection; below it (CC stage), such transfer occurs only by conduction and diffusion (Huppert and Sparks 1988a). Experimental and theoretical studies indicate that a drastic transition in mechanical properties of a crystal-melt mixture occurs within a narrow range of the melt fraction, between 30 and 50% (e.g. Shaw 1980; Marsh 1981). The temperature at the critical melt fraction is called the ‘effective fusion temperature (EFT)’ (Huppert and Sparks 1988a, 1988b; Koyaguchi and Kaneko 1999, 2000). For our calculations, the critical melt fraction and the EFT were assumed to be 50% and 730°C, respectively.

Because the temperature at the boundary between the partially molten solid and the fluidal magma is fixed at the EFT and the magma temperature is higher than the EFT in the C&M stage, the magma can convect vigorously due to the temperature gradient in the magma (Koyaguchi and Kaneko 2000). The heat flux from the fluidal magma into the partially molten solid can be estimated from the Rayleigh-Nusselt relationship Nu ∝ Ra^1/3 (Turner 1973; Turcotte and Schubert 1982) as

$$F_\text{m}=bk\left({\frac{{\alpha g}}{{\kappa \nu }}} \right)^{1/3} (T_\text{m} - T_{\text{EFT}})^{4/3}$$

(A.3)

, where k, α, g, κ, ν, T _m and T _EFT are the thermal conductivity, thermal expansion, gravitational acceleration, thermal diffusivity, kinematic viscosity, magma temperature and the effective fusion temperature, respectively, and b is a constant. In the C&M stage, the migration rate of the solid-magma boundary, da/dt, is given by the conservation of heat at the boundary as

$$\frac{{da}}{{dt}}=\frac{{F_\text{m} - F_\text{s} }}{{\rho L\Delta \varphi }}$$

(A.4)

, where F _s is the heat flux from the interface to the solid, ρ is the density, L is the latent heat, and Δ is the difference in crystal fraction between the interface and the magma body. In a quasi-steady state, the rate of boundary migration can be obtained analytically (Huppert 1986; Koyaguchi and Kaneko 2000) as

$$\frac{{da}}{{dt}}=\frac{{F_\text{m} }}{{\rho [c(T_\text{m} - T_\infty) + L\Delta \varphi ]}}.$$

(A.5)

This study adopts Eq. (A.5) for the calculation of boundary migration. Conservation of heat in the magma body requires that

$$ - F_\text{m} + \rho L\frac{{d\phi }}{{dt}}H - \rho L\Delta \varphi \frac{{da}}{{dt}}=\rho c\frac{{dT_\text{m} }}{{dt}}H + \rho c(T_\text{m} - T_{\text{EFT}})\frac{{da}}{{dt}},$$

(A.6)

where H is the height of the magma body and c is the specific heat. After convection ceases, the magma cools by heat conduction. This process can be expressed as

$$\rho \left({c - L\frac{{d\phi }}{{dT}}} \right)\frac{{dT}}{{dt}}=k\frac{{\partial ^2 T}}{{\partial z^2 }},$$

(A.7)

where d/dT is obtained from Eqs. (A.1) and (A.2). The heat conduction in partially molten crust (T>T _s) is also calculated using Eq. (A.7). Following Koyaguchi and Kaneko (2000), we assumed that the C&M stage and CC stage were bounded by T _EFT + 10°C, evaluating the time scale of the C&M stage by the time taken for T _m to reach T _EFT + 10°C, as the timescale for T _m to reach T _EFT exactly is infinite.

Appendix B:

Possible temperature increase due to decompression crystallization

During the magma ascent from deeper level (250 MPa) to the emplacement level (140 MPa), if no heat was lost to the surroundings, the crystallization caused by decompression inevitably increases the temperature of magma. When the density difference between crystal and melt is sufficiently small, the energy balance requires that

$$c\Delta T=L\Delta \phi .$$

(B.1)

The relationship between crystal fraction and temperature at 140 MPa can be expressed in the same way as Eq. (A.2). In Fig. 4, the quartz stability curve is sub-parallel with the solidus curve and about 50°C higher than the solidus. The crystal fraction between the solidus and the quartz stability curve at 140 MPa can be written as

$$\phi=0.5\text{ } + \text{ }0.5\,(780 - T)/(780 - T_\text{S})\quad (725\text{ }(T_\text{S}) \leqslant T < 780),$$

(B.2)

provided that crystal fraction on the quartz stability limit (780°C) is 50%. If the magma with 730°C and 50% crystals at 250 MPa is decompressed to 140 MPa, the magma temperature and its crystal fraction becomes 764°C and 65%, respectively.