Most soil carbon models use a static Q10 or the Arrhenius equation to parameterize the impact of temperature on decomposition rates, but field and laboratory experiments have found that Q10 and the Arrhenius activation energy E vary with time and place, making predicted carbon-climate feedbacks very uncertain. Microbial biogeochemistry, microbial community dynamics, substrate diversity, soil aggregates, and mineral surface sorptive reactions have been proposed to explain the variability of Q10 and E. However, only relatively complicated models have considered these mechanisms and to disentangle their roles in modulating the temperature sensitivity of decomposition has proven difficult. Here we explored interactions between three of these mechanisms in response to temperature fluctuations using a numerically robust, relatively simple numerical model. We show, because of adsorption surface competition between mineral surfaces, enzymes, and microbes, that polymeric organic carbon can be protected from degradation and dissolved organic carbon can be preserved for long periods, consistent with observations. We also demonstrate that the conventional Q10 and characterizations of labile and recalcitrant substrates are not system parameters but transient emergent responses of the tightly coupled nonlinear system. We conclude that carbon decomposition models should include the interactions between sorptive mineral surfaces and microbial processes to correctly represent the temperature sensitivity of decomposition.