Phosphorus distribution between potassic alkali feldspar and metaluminous haplogranitic liquid at 200 MPa (H2O): the effect of undercooling on crystal–liquid systematics

Abstract

To evaluate the applicability of P2O5 concentration in potassic alkali feldspar as a monitor of P2O5 in melt for undercooled systems, crystal–melt partitioning for P was evaluated via feldspar growth experiments in P-bearing ((3 wt% P2O5), water-saturated haplogranitic liquids at 200 MPa, with liquidus undercoolings (ΔT) of 25, 50, 100, 200, and 300°C. Increasing undercooling in the range ΔT=25–200°C shows an evolution of crystal morphologies, from euhedral and well-filled individuals at ΔT=25–50°C to radial clusters with increasingly skeletal habit at greater undercooling. Experiments at ΔT=100–200°C also document the development of P- (up to (9 wt% P2O5) and Si-enriched, more alkaline boundary layers adjacent to crystals. Experiments at ΔT=300°C show an additional change in crystallization fabric in which spherulites of skeletal crystals form in open (vapor) space created by the dissolution of bulk silicate, and compositional boundary layers are not observed. We interpret the changes in reaction products at ΔT=300°C to indicate conditions below a glass transition; hence, partition coefficients were not determined for this undercooling. Values of K d(P)Kfs/melt from experiments at ΔT=25–200°C, calculated from pairs of crystal and immediately adjacent liquid compositions (including boundary layers at higher undercooling), are mostly in the range of 0.25–0.55 and show no effective change with increased undercooling. Essentially no change in K d(P)Kfs/melt with undercooling apparently stems from an interplay between boundary layer composition and a change in the substitution mechanism for P in feldspar from AlPSi−2, common in peraluminous to metaluminous liquids near equilibrium, to increasing proportions of ([ ],P)(M+,Si)−1 with increased undercooling. Bulk glass and liquid beyond boundary layers in experiments with significant percentages of crystallization are homogeneous, and show pronounced fractionation primarily due to the removal of an orthoclase component. Because crystallization was still in progress in experiments with ΔT≤200°C, compositional homogeneity in the bulk liquid requires extremely rapid diffusion of most haplogranite components (Na, K, and Al), apparently resulting from chemical potential gradients stemming from the removal of components from the liquid by crystal growth. Similar homogeneity and bulk fractionation in experiments with ΔT=300°C requires rapid diffusive equilibration for the alkalis even at temperatures below an apparent glass transition. Unlike the haplogranite components, P is only concentrated in liquid boundary layers (ΔT≤200°C) or low-density aqueous vapor (ΔT=300°C) adjacent to crystals. Hence, the P2O5 contents of melt inclusions likely are not representative of bulk melt concentrations in significantly undercooled systems (ΔT≤50–100°C).