AN EXPERIMENTAL-STUDY OF PHASE-EQUILIBRIA AND FE OXY-COMPONENT IN KAERSUTITIC AMPHIBOLE - IMPLICATIONS FOR THE F(H2) AND A(H2O) IN THE UPPER-MANTLE

Experiments have been carried out from 500 to 1200 degrees C, 1 atm to 10 kbar, and f(H2) from that of the IQF buffer to air, to quantify th e variation of Fe oxy-component content in a titanian pargasite megacr yst amphibole from Vulcan's Throne, Arizona, The results document the operation of the following substitution mechanism in the amphibole cry stal structure: Fe2+ + OH- = Fe3+ + O2- + 1/2H(2) whereby the Fe3+/Fe- tot of the amphibole is controlled by T, P, and f(H2). For the amphibo le composition that was investigated, there is a linear variation of l og f(H2) as a function of log (Fe3+/Fe2+) at fixed T and P of the form log f(H2) = a + b log(Fe3+/Fe2+). Values of a and b are: [GRAPHICS] T wo different expressions were defined for the equilibrium constant for the amphibole Fe end-member reaction Ca2Fe52+Si8O22(OH)(2) = Ca2Fe32Fe23+Si8O24 + H-2. In the random mixing model, in which it is assumed that Fe3+ and Fe2+ mix randomly on the five M1, M2, and M3 crystallogr aphic sites, [GRAPHICS] for which log K = 4.25 - 4363/T (K) + 0.11(P - 1)(kbar). From the appropriate values of K, f(H2) of the experiments can be predicted to within similar to 0.5 log units from knowledge of the absolute amounts of Fe3+, Fe2+, and OH in the amphibole. For the n onrandom mixing model, in which observed Fe3+ and Fe2+ site population s are used to define the mole fraction terms, [GRAPHICS] for which log K = 5.29 - 5903/T (K) + 0.13(P - 1) (kbar). With this model, f(H2) of the experiments can be predicted to within similar to 0.3 log units f rom knowledge of the relevant ionic contents. The random mixing model yields slightly poorer estimates of f(H2) but can be used for literatu re data in applications because Fe3+ and Fe2+ site populations for nat ural kaersutite are seldom reported. In order to use the K expressions , the OH content of the amphibole must be known. In kaersutite for whi ch H content has not been measured, OH apfu can be estimated as (2.0 - Fe3+ - Ti). Because both the reactant and product amphiboles in the e nd-member reaction refer to components in a single homogenous amphibol e phase, the K expressions should apply to any calcic amphibole in whi ch Fe3+ and Fe2+ mix on the five M1, M2, and M3 crystallographic sites , regardless of the amphibole bulk composition. The study confirms tha t the relatively high Fe3+/Fe-tot of most natural kaersutitic amphibol es can result from P-T-f(H2) conditions characteristic of the upper ma ntle, rather than from oxidation during ascent or eruption. Closed-sys tem cooling favors the reduction, not oxidation, of amphibole. With K values from the equations above, it is possible to predict f(H2) of am phibole crystallization, presumably from a melt, if P and T, as well a s the relevant amphibole composition terms, are known. Calculated valu es of f(H2) for the majority of kaersutitic amphiboles reported in the literature range from approximately 0.01 to 100 bars. Such f(H2) valu es are generally consistent with estimated redox states and H2O activi ties of mantle processes. If f(H2) estimates are combined with f(O2) e stimates made on the same xenolith assemblages, H2O activity in the en vironment of formation can be predicted.