Semiconductor electronics has an energy problem: As the number of chips in information systems, computers, and

consumer electronics continues to rocket skywards, so does the collective energy demand, already a significant drain on the world’s electricity consumption. Dissipating the heat generated by the electric currents that drive the devices only compounds the problem. Magnetoelectric multiferroics may offer a way out by allowing voltages rather than currents to drive memory and logic devices, but so far researchers have been unable to find many materials capable of operation at room temperature or above, essential for practical devices.

A large collaboration led by researchers from Cornell University and the University of California, Berkeley, has now reported in

Nature that growing multilayers of ferrimagnetic LuFe₂O₄ and ferroelectric LuFeO₃ with atomic layer control can construct a new multiferroic superlattice in which the physical properties exceed those of the constituent materials. Using a suite of characterization tools, the researchers demonstrated strong magnetic ordering to 281 K and ferroelectricity up to about 900 K. “In my view, this is just the beginning of a means to extend multiferroic behavior to practical temperatures so that electrical control of magnetism can become a reality for real devices,” says Darrell Schlom, whose Cornell laboratory grew the new material.

A magnetoelectric multiferroic is simultaneously ferroelectric and magnetically ordered. Ideally, a voltage aligns the ferroelectric polarization, and the magnetoelectric coupling results in a similar magnetic alignment representing the data stored, a currentless process consuming relatively little energy. The same coupling can also be applied to logic and high-frequency devices, says Ramamoorthy Ramesh, lead member of the collaboration’s Berkeley contingent. The trick for commercial applications is to find a material that responds ferroelectrically to a low voltage but has a large enough coupling to generate a useful magnetization at room temperature or above. Structures based on multiferroic BiFeO₃ have been one promising avenue, but the intrinsic magnetism in BiFeO₃ is too small for applications.

Julia Mundy, the lead author of the Nature letter, who has since moved from Cornell to Berkeley, explained the choice of lutetium–iron–oxygen compounds. “Over the last decade, LuFe₂O₄ has received a lot of attention; it is strongly ferrimagnetic but not ferroelectric. We recognized that LuFeO₃, well known as a robust ferroelectric but not magnetic, has a similar structure and chemistry. So we decided to combine them at the atomic level.” A significant difference in the structures of the two materials is that the planes of lutetium atoms are smooth in the ferrimagnetic compound but strongly rumpled in the ferroelectric one. By growing single layers of LuFe₂O₄ on multiple layers of LuFeO₃, the researchers hoped that the rumpling would be imprinted on the former, stabilizing a ferroelectric state.

To pull this off, the collaboration used Cornell’s extensive molecular beam epitaxy (MBE) facility. Growing LuFeO₃ and adding extra layers of FeO at precise intervals, the researchers found that the optimum properties came from a repeating sequence comprising a single LuFe₂O₄ layer on nine LuFeO₃ layers, although property shifts became evident with as few as two LuFeO₃ layers. “What was surprising,” says Mundy, “was that the symmetry breaking due to the lattice distortion also increased the magnetic transition temperature to near room temperature, as well as enhancing the magnetic moment.”

The rumpled structure was confirmed by scanning transmission electron microscopy. The magnetic properties measurements included superconducting quantum-interference device (SQUID) magnetometry and neutron diffraction at the National Institute of Standards and Technology (NIST). Ferroelectric properties were measured with piezoresponse force microscopy (PFM) and with x-ray linear dichroism measurements at the Lawrence Berkeley National Laboratory’s Advanced Light Source (ALS). To directly verify the magnetoelectric coupling, the researchers combined electric poling to write a ferroelectric domain pattern and x-ray circular dichroism with a photoemission electron microscope at the ALS to image the matching magnetic pattern. Extensive density functional calculations verified the beneficial effects of the distortions.

As Schlom’s assessment indicates, the success is only the first step on the march toward commercially viable magnetoelectric multiferroic devices for microelectronics, including new variations under investigation, such as spintronics. In addition to making an actual device, Ramesh pointed to some specific requirements already evident: raising the magnetic transition temperature to above 400 K and lowering the voltage required to drive switching from today’s 15 V to more like 100 mV. Another question is how applicable is MBE to commercial-scale production? Can higher throughput techniques like chemical vapor or laser deposition achieve the atomic precision needed?

Meanwhile, Manfred Fiebig of the Swiss Federal Institute of Technology (ETH) in Zürich placed the creation of a new high-temperature multiferroic from atomic templates in a larger context, saying, “One of the hottest topics in creating oxide materials with new functionalities is to apply strain. By introducing the rumpled templates, Mundy et al. carry strain from the macroscopic level to the realm of local distortions. I expect that researchers will readily pick up this new concept of applying targeted distortions to specific atoms to generate new properties in other materials.”

Read the abstract in Nature.