Computational materials science: out of the scalar sand box.

426 nature materials | VOL 7 | JUNE 2008 | www.nature.com/naturematerials the observation that most multiferroics are ferroelectric antiferromagnets, Binek and Doudin9 suggested focusing on exchange bias, the directional coupling occurring at the interface between a ferromagnet and an antiferromagnet. This proposal triggered much activity in the search for exchange bias using multiferroics, its connection to ferroelectric properties10 and ultimately its electrical control. Interesting results on the manipulation of exchange bias have been reported by Laukhin et al., but with the lowtemperature multiferroic YMnO3 (ref. 11). However, the number of multiferroics that could be used for MERAMS is quite limited, and so far only BiFeO3 (BFO) has emerged as a potential candidate. In bulk, BFO is a rhombohedral perovskite, exhibiting antiferromagnetism with a Néel temperature of 640 K and ferroelectricity up to 1,100 K (ref. 12). A magnetoelectric coupling between the two order parameters has been demonstrated recently13. Its large electrical polarization12, the highest among all known ferroelectrics, already makes BFO a strong candidate for next-generation FeRAMs14. As now shown by Chu et al., BFO also seems very promising for the design of future MERAM elements. Indeed, they report for the first time a magnetoelectric manipulation of magnetization at room temperature. In BFO, the ferroelectric polarization and the antiferromagnetic vector are coupled13, so that reversing the polarization by an electric field also rotates the antiferromagnetic spins. In line with the above MERAM scheme, Chu et al. show that the ferromagnetic domain structure of CoFe micrometre-size elements deposited on top of a BFO film also exhibits a systematic coupling with the antiferromagnetic spins in BFO. Therefore, when an in-plane electric field is applied, the magnetic domain structure of the CoFe dots is modified as the magnetization rotates by 90 degrees. The original magnetic state is recovered when a voltage with opposite polarity is applied. This result is an essential step towards the fabrication of prototype MERAM elements that are reproducibly switched by an electric field. However, many issues remain, as for future devices a perpendicular geometry, low write voltages, high-frequency operation (GHz) and nanoscale devices will eventually be needed. Now, the next step is to realize a full MERAM element where the electrical resistance of a spintronics device is controlled with a multiferroic. Nevertheless, the control of magnetism by a multiferroic achieved by Chu and colleagues is certainly a testament to the rude health of this field, and extends the potential of multifunctional oxides for novel devices.