SUPPLEMENTARY INFORMATION Reversible Electric Control of Exchange Bias in a Multiferroic Field Effect Device S. M. Wu 1, 2, Shane A. Cybart 1, 2, P. Yu 1, 2, M. D. Abrodos 1, J. Zhang 1, R. Ramesh 1, 2 & R.C. Dynes 1, 2, 3 1 Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA 2 Department of Physics, University of California, Berkeley, CA 94720 USA 3 Department of Physics, University of California, San Diego, La Jolla, CA 92093 USA Figure S1 Structural and chemical characterization of BFO/LSMO heterostructures a, Scanning transmission electron microscopy (STEM) measurements on the interface between LSMO and BFO showing that the interface is smooth at the atomic level. b, Electron energy loss spectroscopy (EELS) measurements of the BFO/LSMO interface, showing little interfacial interdiffusion. nature materials www.nature.com/naturematerials 1
supplementary information Figure S2 Ferroelectric domain characterization. a, Out-of-plane piezoresponse force microscopy (PFM) images showing a square central region that has been polarized down. b, Inplane PFM of the same region showing a striped domain structure in both the as-grown up polarization state, and the poled down polarization state. Detailed analyses of the in-plane and out-of-plane PFM images obtained from several locations of the sample, clearly show that the domain structure is comprised of almost exclusively of 71 domains, consistent with previous studies 1. Furthermore, conducting AFM measurements of these same regions show no indication of domain wall conduction, which is again consistent with the previously published reports of conduction in 109 and 180 domain walls and no measurable conduction in 71 walls 2. 2 nature MATERIALS www.nature.com/naturematerials
supplementary information Figure S3 Analysis of temperature dependent R S data. The data from Fig. 4a are reproduced here along with curves generated by adding only a vertical translation (light blue), only a multiplicative factor (pink), and both a vertical translation and a multiplicative factor (green). It is seen here that neither a vertical translation nor a multiplicative factor alone can completely explain this effect, and that both a small amount of doping and change in temperature independent scattering is likely occurring in this sample. The tail of R S (+45V) does not fit any of the curves, regardless, due to 2D electron localization effects at low temperature that cause the tail of the R S -T curve to diverge faster at low temperatures as the resistance of the sample is increased. nature materials www.nature.com/naturematerials 3
supplementary information Figure S4 Hysteresis measurements for resistance and electric polarization. An alternate method of obtaining sheet resistance vs. gate voltage is presented in the red curve. This curve differs from the curve presented in Fig. 3c because the gate voltage was left on the sample and the resistance was measured continuously over a 100 sec sweep. Heating effects through gate leakage are minimal as demonstrated by the fact that the curve closes on itself and does not display any symmetric increases in resistance at the higher gate voltages. The corresponding electric polarization hysteresis curve was obtained using the Sawyer-Tower method, which shows a full electric polarization of BFO. 4 nature MATERIALS www.nature.com/naturematerials
supplementary information Figure S5 Heterointerfacial ferromagnetic moment in BFO. In the study presented in the main article, the LSMO/BFO heterostructures are grown on STO (001) substrates. On the [001] surface, the G-type antiferromagnetic BFO is magnetically compensated. This means that we cannot directly explain the exchange bias effect (Fig. 2b), within the classical exchange bias model 3, and thus motivates the study of the magnetic spin structure of BFO at the heterointerface 4. One of the most effective techniques to accomplish this is through x-ray magnetic circular dichroism (XMCD) measurements in the total electron yield mode. In the XMCD measurement, the transition metals (Fe and Mn) L edges ((2p 3d dipole transitions) are probed with circular polarized x-rays. The difference between the two spectra (left/right circular polarized x-rays) provides direct information on both the magnetic spin orientation and the amplitude of the magnetization after applying a so called spin sum rule 5. a and b show the XMCD spectra together with the integrated spectra for both Mn and Fe L-edges at 10 K. Clear XMCD spectra are observed for both cases. The dichroism of Mn is consistent with the previous measured values, while an unexpectedly large XMCD is observed for the Fe L-edge. This data strongly suggests that in the first few nanometers of the BFO film, at the interface, a new magnetic spin structure is present that is markedly different from that in the remainder of the BFO film. Additionally, the opposite signs for the Mn and Fe L-edges, suggest the coupling between the bulk Mn and interfacial Fe spins is antiparallel in nature. In summary, the spin nature materials www.nature.com/naturematerials 5
supplementary information structure of the heterostructure is determined to be as shown in c. An enhanced magnetism is observed in the BFO close to the interface and the induced spins are coupled antiparallel with the LSMO spins. 6 nature MATERIALS www.nature.com/naturematerials
supplementary information References 1 Zavaliche, F. et al. Multiferroic BiFeO3 films: domain structure and polarization dynamics. Phase Transitions 79, 991-1017 (2006). 2 Seidel, J. et al. Conduction at domain walls in oxide multiferroics. Nat Mater 8 (2009). 3 Nogues, J. & Schuller, I. K. Exchange bias. J Magn Magn Mater 192, 203-232 (1999). 4 Yu, P. et al. Interface ferromagnetism and orbital reconstruction in BiFeO3- La0.7Sr0.3MnO3 heterostructures. In Press. Phys. Rev. Lett. arxiv:1006.1194v1 (2010). 5 Chen, C. T. et al. Experimental confirmation of the X-ray magnetic circular dichroism sum rules for iron and cobalt. Phys Rev Lett 75, 152-155 (1995). nature materials www.nature.com/naturematerials 7