Geochimica et Cosmochlmrca Ado Vol. 54, pp. 2947-295 I Copyright 0 1990 Pergamon Press plc. Printed in U.S.A. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 0016-7037/90/$3.00 + 00 Oxygen diffusion in an Fe-rich basalt melt zyxwvutsrqponmlkjihgfedcbaZYXW DANTE CANIL* and KARLIS MUEHLENBACHS Department of Geology, University of Alberta, Edmonton, Alberta, Canada, T6G 2E3 zyxwvutsrqponmlkjihgfedcb (Received March 29, 1990; accepted in revised form August 13, 1990) Abstract-The self-diffusion of oxygen has been measured in an FeTi basalt melt at one atmosphere from 1320- 1500°C under different oxygen fugacities. Diffusivities were determined by measuring isotopic exchange between spheres of melt and flowing oxygen in a vertical tube furnace. The oxidation state of the melt was varied by performing experiments in either 02 or CO* gas. The diffusion data for experiments in both gases can be fit to the equation D = 1.4 X 10 exp(-60(+7) X 103/RT) cm2/s. The observed self-diffusivities for oxygen in the basalt melt are one to two orders of magnitude lower than published chemical diffusivities at one atmosphere. Diffusivities in basalt melt at each temperature in O2 and CO* gas are similar within experimental error (~0.2 log D units at the 1 u level) and indicate that oxygen self-diffusion is insensitive to the oxidation state of Fe in natural basalt melts. The redox independence of oxygen self-diffusion observed in this study suggests that oxygen transport in natural magmas will not be significantly affected by changes in their redox state caused by degassing and/or crystallization. INTRODUCTION KNOWLEDGE OF THE RATE and mechanism of oxygen dif- fusion in silicate melts provides insight into their structural constitution. Towards this end, previous investigations of the self-diffusion of oxygen in silicate melts have attempted to identify anionic species oxygen may be associated with in silicate liquids, and relate these to their structure and physical properties (HOFFMANN, 1980; DUNN, 1986). No study has yet directly investigated the dependence of oxygen diffusion in Fe-bearing melts to changing oxidation state. It has been demonstrated that the oxidation state of Fe-bearing melts has a significant effect on their viscosity (MURASE and MCBIRNEY, 1970; CUKIERMAN and UHL- MANN, 1974; DINGWELL and VIRGO, 1987, 1988), density (MURASE and MCBIRNEY, 1973), anionic structure (VIRGO and MYSEN, 1985; MYSEN, 1986), and phase equilibria (Os- BORN, 1959; DUKE, 1974; DICKENSON and HESS, 1987). In- sight into the role of oxidation state on oxygen diffusion in Fe-bearing silicate melts would complement the information deduced from studies of the above related melt properties. This study examines the effect of varying oxidation state on oxygen self-diffusion in an Fe-rich basalt melt. This melt was chosen so that its high Fe content (15 wt% FeO) may show more clearly the effect of changing oxidation state on oxygen diffusion in natural magmas. The results of this study demonstrate that oxygen transport in most natural magmas will not be significantly affected by externally or internally imposed changes in their oxidation state. The validity of the Eyring equation for describing the relationship between vis- cosity and 0 diffusion in silicate liquids is also demonstrated and discussed. * Present address: BayerischesGeoinstitut, Universitit Bayreuth, Postfach 10 12 5 1, D-8580 Bayreuth, Germany. EXPERIMENTAL METHODS The method used to determine oxygen dithtsivities in this study is similar to that of MUEHLENBACHS and KUSHIRO (1974). Spheres of melt were suspended from Ft loops in a vertical tube firmace. Oxygen of known isotopic composition was exchanged continuously with the spheres of melt by maintaining a constant flow of either COZ or O2 gas. This experimental arrangement models exchange be- tween a sphere and “infinite” reservoir of oxygen, and can be described by the following solution to the diffusion equation (JOST, 1960): (d’*Or - 6’*0~)/(6’*Oi - 6’*0,) = (6/r2) C (l/d) exp(-n2r2Dt/r2) (1) where 6180r is the final isotopic composition of the sphere, b’*Oi is the initial isotopic composition, 6r80, is the equilibrium composition of the sphere after 100% exchange with the gas, t is the experimental duration (set), r is the radius of the sphere (cm), and D is the diffusion coefficient (cm’/s). The use of Eqn. (1) to calculate diffusivities requires that volume diffusion be the rate-limiting process and that convection not occur in the 3-5 mm diameter melt spheres during oxygen exchange. Ex- periments which showed a process other than volume diffusion to be rate limiting were discarded (see below). Rayleigh numbers cal- culated for the highest temperature experiments, using values for the thermal diffusivity and expansivity of basalt melt given in SPERA (1980), were orders of magnitude less than the value required for convection (- 1500, SPERA,1980). Therefore, convection in the melt spheres is thought not to have occurred during the experiments. The FeTi basalt starting composition is from a suite of samples from the Galapagos Spreading Center (MUEHLENBACHS and BYERLY, 1982). Slurries of rock powder (20-50 mg) and alcohol were fused onto weighted Ft loops using a gas torch. The fused glass beads were then equilibrated in air, 02, or CO2 at 1350°C for 8 h to remove bubbles and homogenize their 0 isotopic compositions. The oxidation state of the melts was varied by running one set of experiments in CO* and a second set in 02. The range info, covered by these pure gases is somewhat limited (i.e., only 3 logfOr units at the lowest temperature investigated; Table 1) and higher than that recorded by most magmas in nature (e.g., HAGGERTY, 1978). A wider and more realistic range off0, values would have been obtainable by mixing two oxygen-bearing gases, for example CO and COr. However, mixed gases were not employed as exchange media in the 2947