Viscosity of crystal-bearing melts and its implication for magma ascent

Experiments were performed at high temperature and pressure to determine the effective viscosity of a crystal-bearing andesite using the falling sphere method. Because viscosity experiments with partly crystallized samples are difficult to realize (i.e., due to high sensitivity of phase equilibria to P, T and water content), we have added zircon crystals to adjust precisely the volume fraction and the size of crystals in a magma. Using this approach, the anhydrous melt composition does not vary significantly with temperature and water content of the melt and, hence, the effects of crystals on effective viscosity can be worked out accurately. The investigated systems (magma analogue) were composed of an andesitic melt containing 0.5 wt% to 4.1 wt% H2O and 15 vol% to 40 vol% of zircon crystals with grain size <100 μm. Most experiments were performed at 300 MPa and 1523 K. The ZrO2 content dissolved in the melt under these conditions is 1.61 wt% ± 0.16 wt% (0.5-4.0 wt% H2O in melt) with no significant dependence on water content, which is about twice the value predicted by the model of Watson and Harrison (1983). The falling velocity of large platinum spheres was measured in the experimental magmas. The radius of the spheres (between 130 μm and 510 μm) was always much larger than the crystal sizes and the inter-grain distances, implying that the falling velocity of the spheres can be used to calculate the effective viscosity of the magmas. For magmas containing 15 vol% zircon, the measured viscosity was 0.7 log to 1.6 log units higher than the melt viscosity. The effective magma viscosity is higher than expected from literature models. The spheres did not move in systems containing 30 vol% to 40 vol% of crystals, even after 16 h at 1523 K. We attribute this observation to the presence of yield strength of more than 100 Pa, i.e., a threshold of accelerating force needs to be passed before the sphere can move in the magma.