Recently, Wang et. al. formulated a criterion for identifying the source of deep interface sates at the SiC/SiO2 interface, based on the known inefficiency of H2 passivation. We apply this criterion to a variety of excess carbon defects at the interface, which we have predicted to be energetically feasible. We find that, also with respect to this criterion, the simple C split interstitials are the most likely cause of the deep interface states among the examined defects.