In order to better constrain the extent to which common sulfide minerals will retain their osmium isotopic composition subsequent to crystallization, we have conducted experiments to quantify the diffusion behavior of osmium in pyrite and pyrrhotite. Experiments consisted of either (1) isothermal soaking of diffusion couples consisting of natural pyrite or pyrrhotite crystals packed against powdered Os-bearing Fe-sulfide or (2) 'relaxation' of initially high near-surface osmium concentrations produced in the latter experiments (pyrite only). Osmium penetration into samples was characterized by depth profiling using Rutherford backscattering spectroscopy (RBS) (pyrite) or electron microprobe analyses across sectioned run products (pyrrhotite), Results of the first type of diffusion experiment involving pyrite show only limited osmium penetration into sample surfaces, with the extent of penetration uncorrelated with run duration. Images of pyrite samples using atomic force microscopy show roughening of initially smooth surfaces as a consequence of step formation and suggest that osmium incorporation into the near-surface occurred by solute uptake during step growth and not by volume diffusion. Prolonged (1000(+) h) 'relaxation' experiments revealed no additional osmium penetration into pyrite surfaces and based on the depth resolution for RES, a maximum diffusion coefficient of 2.5 x 10(-23) m(2)/s at 500 degrees C was calculated. Experiments involving pyrrhotite over the temperature range of 950-1100 degrees C showed extensive osmium uptake and osmium concentration gradients that conform with Fickian diffusion behavior. We found that pyrrhotite Fe/S could be varied by changes in the composition of the starting material and osmium source and over the range of Fe/S produced in experiments (molar Fe/S = 0.83-0.90), we observed no systematic variation in the osmium diffusion coefficient. Diffusion coefficients measured parallel to the a crystallographic axis were on average similar to 1.4x higher than values measured parallel to c and regression of the c-axis data yielded the Arrhenius relation: D(T) = 1.31 +/- 0.16 x 10(-5) m(2)/s exp-211.5 +/- 14.7 kJ/mol/RT The application of these diffusion data to simple models of diffusive exchange during static or polythermal timetemperature histories is used to assess the conditions under which radiogenic osmium will be retained. During isothermal annealing, calculations indicate that the cores of millimeter-sized spherical pyrrhotite crystals undergoing diffusive exchange with an external osmium reservoir will have their initial compositions perturbed in less than or equal to 0.5 Ma at temperatures exceeding 400 degrees C, Pyrite undergoing the same process at 500 degrees C requires in excess of lo hla before crystal cores are affected. The relatively short 'core retention' time-scales for pyrrhotite indicates that this mineral may be prone to isotopic resetting following relatively brief crustal thermal events, thus possibly accounting for the scatter that commonly occurs in Re-Os isochrons generated from massive sulfide samples. Calculated closure temperatures (T-c) for osmium exchange in pyrrhotite yielded values of 300-400 degrees C for grain sizes ranging from 10 to 1000 mu m. These values of T-c are similar to those calculated for Ar retention in biotite, and considerably lower than for Sr in apatite and plagioclase, for example. Such low closure temperatures for pyrrhotite suggest this mineral will date the final stage in the cooling of a magmatic system and possibly be susceptible to open system osmium exchange in the presence of late-stage hydrothermal fluids. This latter result infers that caution be applied when interpreting elevated initial osmium isotopic ratios as a product of crustal assimilation at the magmatic stage. (C) 2000 Elsevier Science B.V. All rights reserved.
