Accurate satellite drag coefficients are important for reducing orbit prediction errors and for inferring unbiased atmospheric mass density from measurements of satellite decay. Past use of fixed satellite drag coefficients when inferring atmospheric mass density from orbital decay has resulted in large biases in empirical atmospheric models. These biases can be reduced by using physical drag coefficients which model the interaction between the atmospheric gas particles and the satellite surface; however, physical drag coefficients require detailed knowledge of the gas-surface interaction which is most sensitive to the energy and momentum accommodation coefficients. State-of-the-art models reveal that the effective energy accommodation coefficient for satellites in low Earth orbit is strongly correlated with the adsorption of atomic oxygen. Previous work has modeled this dependence using a Langmuir isotherm which works well at altitudes below ∼500 km but fails to match data at higher altitudes. Therefore, Freundlich and Temkin isotherms are used here to test whether data at higher altitudes might be better fit with a different adsorption model. Mathematically, both the Freundlich and Temkin isotherms should (and do) better fit the data because they have two free parameters compared to only one for the Langmuir isotherm. Physically, the Freundlich isotherm allows for multi-layer adsorption and an exponential range of adsorption energies corresponding to a non-uniform sur-face. The Temkin isotherm also allows for multi-layer adsorption and accounts for the interaction between the adsorbing gas and the adsorbate whereas the Langmuir isotherm is limited to monolayer adsorption and constant adsorption energy corresponding to a uniform surface.