Date of this Version
In a recent Letter Tusche et al.  showed that a complete and coherent FeO layer forms on both interfaces of a FejMgOjFe|001| magnetic tunnel junction (MTJ) when using oxygen-assisted growth. Their ab initio model for these MTJs predicted tunneling magnetoresistance (TMR) of several thousand percent due to the contribution from interface resonances (IRs) perfectly matched at the two interfaces. In this Comment, we show that in practice the predicted giant TMR is unfeasible because the IRs are mismatched by structural disorder and/or by applied bias voltage resulting in a moderate TMR.