ALBERT

Notes: The present study deals with the experimental modelling of two different mechanisms of crystal-melt segregation in crustal rocks: (1) the buoyancy-driven compaction of the crystal + melt matrix and (2) melt filtering in a partially crystalline matrix due to differential stresses. These two segregation mechanisms have differing relative efficiencies in the deformation of crustal rocks and result in different texture scales depending on melt fraction, melt viscosity and tectonic stresses. A centrifuge furnace has been used in the present study for the modelling of melt migration in partially molten granitic rocks. Samples of Beauvoir granite (Massif Central, France) with a grain size of 0.16–0.5 mm and dimensions of diameter ∼5 mm, length ∼16 mm were used. These samples had been pre-fused at temperatures of 1000–1075 °C, yielding an initial average melt fraction of ∼45–50 volume per cent. The centrifuging of partially melted samples during ∼6 hr at an acceleration of 1000g (g is gravity) results in a linear vertical distribution of melt over the length of the sample without the development of a compaction layer. The gradient of the melt fraction (melt migrates to the top of samples) correlates with temperature: 1075°C ∼7 volume per cent mm-1; 1050°C ∼4 volume per cent mm-1; 1000°C ∼1.5 volume per cent mm-1. The calculated rate of melt migration varies from 3x10-5 cm s-1 (1075°C) to 2x10-6 cm s-1 (1000°C).Differential stresses of ∼0.7–1.4 MPa have been generated in the centrifuge by putting a piston (weight ∼1.02–2.05 g, diameter ∼4.5 mm) on the top of the partially melted sample, which is then centrifuged at ∼1000g. The rate of melt squeezing from the sample in this case is about two orders of magnitude higher than that observed without the piston. After centrifuging for 6 hr, a compaction layer below the piston is formed with a thickness of ∼2.5 mm and a crystal fraction of ∼70–65 volume per cent. Further centrifuging (∼15 hr) does not result in any increase of the compaction-layer thickness or volume percentage of crystals in it. The comparison of the two segregation mechanisms confirms the much greater efficiency of differential-stress-induced melt segregation and accumulation in veins and pockets than the compaction mechanism.

Notes: The present experimental study deals with the laboratory modelling of two different mechanisms of gravitational percolation in partially melted rocks: (1) diapiric percolation of heavy material and (2) the sedimentation of heavy particles. These two mechanisms of mass transport in partially melted rocks result in different, scales of the segregation process in the melt-crystal matrix. A centrifuge furnace was used to simulate the percolation of the heavy particle layer through the partially molten granite at temperatures of up to 1000 °C. Samples of Beauvoir granite (Massif Central, France, grain size 0.16–0.5 mm with an initial degree of partial melting ∼45 per cent) were used as a matrix. A layer of Pt powder suspended in a melt of the same composition as the partially melted matrix was placed on the top of the granite sample. After centrifuging for various times (up to 2 × 104 s), X-ray images of samples were obtained and the evolution of the percolation process of heavy suspension in the partially molten granite was monitored from the Pt particle distribution. The diapiric or finger regime of percolation starts when the growth rate of a Raleigh-Taylor instability of the heavy layer is faster than the Stokes sedimentation velocity of individual particles in the upper layer. This relationship is a complex function of the size and initial concentration of heavy particles, as well as the ratio of particle to crystal size, the permeability of the matrix, and the heterogeneity scale in the partially melted matrix. At small concentrations (several per cent) and at large concentrations (where close packing of heavy particles results in an anomalous viscosity increase in the upper heavy layer) Stokes sedimentation is dominant in the vertical percolation of the heavy material. The sinking velocity of the diapir decreases when the size of heavy particles in it becomes comparable with the size of crystals in the partially melted granite. In this situation the vertical sinking of the diapir is not stable and the horizontal instability of the vertical mass transport starts to become important. Mass transport via diapiric percolation results in more efficient crystal-melt segregation of partially melted rocks. The percolation of individual particles provides only local melt-crystal flow on a scale comparable with the heavy particle size. The diapiric percolation provides a much larger scale of partial melt segregation with a length-scale comparable with the diapir size.