Tables of mean opacities are often used to compute the transfer of radiation in a variety of astrophysical simulations from stellar evolution models to proto-planetary disks. Often tables, such as Ferguson et al. (2005), are computed with a predetermined set of physical assumptions that may or may not be valid for a specific application. This paper explores the effects of several assumptions of grain physics on the Rosseland mean opacity in an oxygen rich environment. We find that changing the distribution of grain sizes, either the power-law exponent or the shape of the distribution, has a marginal effect on the total mean opacity. We also explore the difference in the mean opacity between solid homogenous grains and grains that are porous or conglomorations of several species. Changing the amount of grain opacity included in the mean by assuming a grain-to-gas ratio significantly affects the mean opacity, but in a predictable way.

Previous computations of low temperature Rosseland and Planck mean opacities from Alexander & Ferguson (1994) are updated and expanded. The new computations include a more complete equation of state with more grain species and updated optical constants. Grains are now explicitly included in thermal equilibrium in the equation of state calculation, which allows for a much wider range of grain compositions to be accurately included than was previously the case. The inclusion of high temperature condensates such as Al$_2$O$_3$ and CaTiO$_3$ significantly affects the total opacity over a narrow range of temperatures before the appearance of the first silicate grains. The new opacity tables are tabulated for temperatures ranging from 30000 K to 500 K with gas densities from 10$^{-4}$ g cm$^{-3}$ to 10$^{-19}$ g cm$^{-3}$. Comparisons with previous Rosseland mean opacity calculations are discussed. At high temperatures, the agreement with OPAL and Opacity Project is quite good. Comparisons at lower temperatures are more divergent as a result of differences in molecular and grain physics included in different calculations. The computation of Planck mean opacities performed with the opacity sampling method are shown to require a very large number of opacity sampling wavelength points; previously published results obtained with fewer wavelength points are shown to be significantly in error. Methods for requesting or obtaining the new tables are provided.