research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

IUCrJ
Volume 6| Part 2| March 2019| Pages 189-196
ISSN: 2052-2525

Robust manipulation of magnetism in LaAO3/BaTiO3 (A = Fe, Mn and Cr) superstructures by ferroelectric polarization

CROSSMARK_Color_square_no_text.svg

aInstitute for Computational Materials Science, School of Physics and Electronics, Henan University, Kaifeng 475000, People's Republic of China, bInstitute for Superconducting and Electronic Materials, Australian Institute of Innovative Materials, University of Wollongong, Wollongong NSW 2500, Australia, and cSchool of Physics, Southeast University, Nanjing 211189, People's Republic of China
*Correspondence e-mail: cheng@uow.edu.au, wangyx@henu.edu.cn

Edited by P. Lightfoot, University of St Andrews, Scotland (Received 2 October 2018; accepted 15 November 2018; online 15 January 2019)

Robust control of magnetism is both fundamentally and practically meaningful and highly desirable, although it remains a big challenge. In this work, perovskite oxide superstructures LaFeO3/BaTiO3 (LFO/BTO), LaMnO3/BaTiO3 (LMO/BTO) and LaCrO3/BaTiO3 (LCO/BTO) (001) are designed to facilitate tuning of magnetism by the electric field from ferroelectric polarization, and are systemically investigated via first-principles calculations. The results show that the magnetic ordering, conductivity and exchange interactions can be controlled simultaneously or individually by the reorientation of the ferroelectric polarization of BTO in these designed superstructures. Self-consistent calculations within the generalized gradient approximation plus on-site Coulomb correction did not produce distinct rotations of oxygen octahedra, but there were obvious changes in bond length between oxygen and the cations. These changes cause tilting of the oxygen octahedra and lead to spin, orbital and bond reconstruction at the interface, which is the structural basis responsible for the manipulation. With the G-type antiferromagnetic (G-AFM) ordering unchanged for both ±P cases, a metal–insulator transition can be observed in the LFO/BTO superstructure, which is controlled by the LFO thin film. The LMO/BTO system has A-type antiferromagnetic (A-AFM) ordering with metallic behavior in the +P case, while it shifts to a half-metallic ferromagnetic ordering when the direction of the polarization is switched. LCO/BTO exhibits C-type antiferromagnetic (C-AFM) and G-AFM orders in the +P and −P cases, respectively. The three purpose-designed superstructures with robust intrinsic magnetoelectric coupling are a particularly interesting model system that can provide guidance for the development of this field for future applications.

1. Introduction

Traditional information storage media, e.g. hard disks, which consist of non-magnetic and magnetic layers formed on a support, feature slow reading/writing speeds, high energy consumption and weak thermal shock resistance. Utilization of the magnetoelectronic coupling in multiferroic materials might provide an alternative approach to solving these problems by electrically writing magnetic bits with extremely low energy consumption. Unfortunately, ferromagnetism and ferroelectricity are naturally contradictory in their requirements for 3d transition metals, and it is difficult for them to coexist in single-phase materials. So far it has not been possible to identify a stable, single-phase multiferroic material in which the magnetization can be totally switched. The artificial superstructure, with its combination of materials that are abruptly different in their properties, provides an ideal platform to directly couple the different physical properties between adjacent layers or create new physical properties, such as magnetoelectric coupling, superconductivity, multiferroicity, colossal magneto-resistance, etc. (Dong, Yu et al., 2009[Dong, S., Yu, R., Liu, J. M. & Dagotto, E. (2009). Phys. Rev. Lett. 103, 107204.]; Zhai et al., 2014[Zhai, X., Cheng, L., Liu, Y., Schlepütz, C. M., Dong, S., Li, H., Zhang, X., Chu, S., Zheng, L., Zhang, J., Zhao, A., Hong, H., Bhattacharya, A., Eckstein, J. N. & Zeng, C. (2014). Nat. Commun. 5, 4283.]; Dong, Yamauchi et al., 2009[Dong, S., Yamauchi, K., Yunoki, S., Yu, R., Liang, S., Moreo, A., Liu, J. M., Picozzi, S. & Dagotto, E. (2009). Phys. Rev. Lett. 103, 127201.]). In recent years, there has been extensive research focused on perovskite oxide superstructure materials in the hope that one part of the interface can provide magnetism and the other can provide ferroelectricity (Weng et al., 2016[Weng, Y. K., Lin, L. F., Dagotto, E. & Dong, S. (2016). Phys. Rev. Lett. 117, 037601.]) as a result of chemical compatibility as well as similar lattice constants of the constituent perovskite oxides (Weng et al., 2015[Weng, Y. K., Huang, X., Yao, Y. G. & Dong, S. (2015). Phys. Rev. B, 92, 195114.]). In particular, some perovskite superstructures are composed of two compounds with respective ferroelectric polarization and ferromagnetism, which provide an ideal scheme for the possible electric field control of magnetism with reduced energy consumption; they are also suitable systems for the study of interface effects (Bousquet et al., 2008[Bousquet, E., Dawber, M., Stucki, N., Lichtensteiger, C., Hermet, P., Gariglio, S., Triscone, J. M. & Ghosez, P. (2008). Nature, 452, 732-736.]).

Actually, the robust control of magnetism by an external electric field has already been realized in non-perovskite metals. Fechner et al. (2012[Fechner, M., Zahn, P., Ostanin, S., Bibes, M. & Mertig, I. (2012). Phys. Rev. Lett. 108, 197206.]) have demonstrated a 180° switching of the magnetization in a PbTiO3/Fe/Au/Fe heterostructure, which can be mainly attributed to magnetoelectric coupling at the PbTiO3/Fe interface. This coupling is amplified by interlayer-exchange coupling in the Fe/Au/Fe trilayer. There are also many ways to control the magnetism for perovskite materials such as LaMnO3 (LMO). The magnetization of an LMO thin film grown on SrTiO3 (STO) (Kim & Christen, 2010[Kim, H. S. & Christen, H. M. (2010). J. Phys. Condens. Matter, 22, 146007.]; Roqueta et al., 2015[Roqueta, J., Pomar, A., Balcells, L., Frontera, C., Valencia, S., Abrudan, R., Bozzo, B., Konstantinović, Z., Santiso, J. & Martínez, B. (2015). Cryst. Growth Des. 15, 5332-5337.]) or LaAlO3 (Zhang et al., 2017[Zhang, A. M., Zhang, W. C., Wu, X. S. & Lin, J. G. (2017). AIP Adv. 7, 055837.]) substrates can be directly controlled by changing the oxygen partial pressure. Two phase transitions occur in the LMO thin film (Hou et al., 2014[Hou, Y. S., Xiang, H. J. & Gong, X. G. (2014). Phys. Rev. B, 89, 064415.]), namely the transition from the A-type antiferromagnetic (A-AFM) phase to the insulating ferromagnetic phase and then to the metallic ferromagnetic phase. Gibert et al. (2015[Gibert, M., Viret, M., Torres-Pardo, A., Piamonteze, C., Zubko, P., Jaouen, N., Tonnerre, J. M., Mougin, A., Fowlie, J., Catalano, S., Gloter, A., Stéphan, O. & Triscone, J. M. (2015). Nano Lett. 15, 7355-7361.]) investigated the LMO/LaNiO3 heterostructures and found that the interface-driven magnetic moment variations have a strong dependence on interface reconstructions. A robust ferromagnetic moment and large room-temperature magnetoresistance are demonstrated by the LMO thin films (Vila-Fungueiriño et al., 2015[Vila-Fungueiriño, J. M., Rivas-Murias, B., Rodríguez-González, B., Txoperena, O., Ciudad, D., Hueso, L. E., Lazzari, M. & Rivadulla, F. (2015). Appl. Mater. Interfaces, 7, 5410-5414.]). To date, a variety of methods to control magnetism in layered perovskite heterostructures and superstructures have been considered. Wang et al. (2015[Wang, X. R., Li, C. J., Lü, W. M., Paudel, T. R., Leusink, D. P., Hoek, M., Poccia, N., Vailionis, A., Venkatesan, T., Coey, J. M. D., Tsymbal, E. Y., Ariando, & Hilgenkamp, H. (2015). Science, 349, 716-719.]) synthesized high-quality ultrathin LMO films on TiO2 terminated STO (001) substrates. An atomically sharp transition from the non-magnetic phase to the ferromagnetic phase can be observed when the thickness of LMO reaches five unit cells, which is argued to be the result of charge reconstruction induced by polar discontinuity. This ferromagnetic ordering is generated by the self-doping effect (i.e. electrons are transferred from the surface to the interface), which contradicts previous theoretical results (Hou et al., 2014[Hou, Y. S., Xiang, H. J. & Gong, X. G. (2014). Phys. Rev. B, 89, 064415.]; Dong et al., 2008[Dong, S., Yu, R., Yunoki, S., Alvarez, G., Liu, J. M. & Dagotto, E. (2008). Phys. Rev. B, 78, 201102R.]; Lee et al., 2013[Lee, J. H., Delaney, K. T., Bousquet, E., Spaldin, N. A. & Rabe, K. M. (2013). Phys. Rev. B, 88, 174426.]). Through an optical second-harmonic generation experiment, Mishina et al. (2016[Mishina, E. D., Buryakov, A. M., Sherstyuk, N. E., Sigov, A. S. & Rasing, T. (2016). Ferroelectrics, 500, 37-46.]) found that the magnetic configuration remained the same for the situation with or without an external electric field applied on a La0.7Ca0.25MnO3/BaTiO3 superstructure. Some theoretical and experimental results show that a change in the interface magnetism can be achieved by switching the ferroelectricity. For example, the control of magnetism and conductivity via an external electric field has been demonstrated theoretically in La1−xDxMnO3/BaTiO3 (001) (D = Ca, Sr and Ba) interfaces (Burton & Tsymbal, 2009[Burton, J. D. & Tsymbal, E. Y. (2009). Phys. Rev. B, 80, 174406.]). Unfortunately, the control of magnetism via an external electric field is rarely observed in metallic materials since the electric field cannot penetrate more than a few unit cells before it is completely screened by conductive layers. On the other hand, the technique of magnetism manipulation by ferroelectric polarization has developed quickly. Duan et al. (2006[Duan, C. G., Jaswal, S. S. & Tsymbal, E. Y. (2006). Phys. Rev. Lett. 97, 047201.]) investigated the control of magnetism of the Fe/BaTiO3 multilayers. Only the magnitude of the magnetic moment, and not its magnetic ordering, is changed when the direction of the ferroelectric polarization alters. Dong & Dagotto (2013[Dong, S. & Dagotto, E. (2013). Phys. Rev. B, 88, 140404R.]) have also investigated the control of magnetism through the ferroelectric polarization of BaTiO3 (BTO). The origin of the magnetization control is the modulation of charge density induced in the interfacial layers to screen the polarization charges of BTO. In recent work, the polarization control of magnetization has been experimentally demonstrated in the La2/3Sr1/3MnO3/BaTiO3 superstructure (Cui et al., 2015[Cui, B., Song, C., Mao, H., Wu, H., Li, F., Peng, J., Wang, G., Zeng, F. & Pan, F. (2015). Adv. Mater. 27, 6651-6656.]). This manipulation of magnetism is mainly due to the interfacial orbital reconstruction of the superstructure, driven by the shuttle displacement of Ti atoms under ferroelectric polarization. The conductance of different La1−xSrxMnO3 heterostructures can be dramatically switched by the switching of ferroelectric polarization (Yin et al., 2013[Yin, Y. W., Burton, J. D., Kim, Y. M., Borisevich, A. Y., Pennycook, S. J., Yang, S. M., Noh, T. W., Gruverman, A., Li, X. G., Tsymbal, E. Y. & Li, Q. (2013). Nat. Mater. 12, 397-402.]). The highly spin-polarized MnO2 layers near the interface act as an atomic scale spin valve in series with the ferroelectric tunnel barrier, which creates a switch for the conductance. To date, only metallic materials (not insulating ones) have been considered for the control of magnetism by BTO. The experimental demonstration of tuning effects in the above superstructure provides a model system showing the effectiveness of tunneling effects imposed by ferroelectric polarization.

Since the conclusions of the above-mentioned experimental and theoretical investigations are different and still the subject of debate, it is highly desirable to analyze the modulation of magnetism and its underlying mechanism in perovskite superstructures. The modulation of magnetism in artificially designed thin films and superstructures is crucial to their implementation in magnetoelectronic devices (Takamura et al., 2013[Takamura, Y., Folven, E., Shu, J. B. R., Lukes, K. R., Li, B., Scholl, A., Young, A. T., Retterer, S. T., Tybell, T. & Grepstad, J. K. (2013). Phys. Rev. Lett. 111, 107201.]). Therefore, novel physical phenomena can only be observed near the interface, and the electric field has only a limited tuning effect on the whole system. Designing a novel superstructure system and making a real sample for experimental examination is very costly and time consuming, and more importantly, it is not generally applicable in many cases. Fortunately, first-principles modeling and calculations allow us to precisely control the superstructure structure, polarization and magnetism on an atomic level, and simulate the tuning of magnetism by electric polarization before real samples are fabricated and examined (Huang & Dong, 2014[Huang, X. & Dong, S. (2014). Mod. Phys. Lett. B, 28, 1430010.]). Our scheme is to directly control the magnetic moments of the magnetic atoms that are tuned by ferroelectric polarization, thus realizing the robust manipulation of the magnetism.

BaTiO3 is an important material because of its ability to maintain strong electric polarization that can be reoriented easily by an electric field (Callori et al., 2012[Callori, S. J., Gabel, J., Su, D., Sinsheimer, J., Fernandez-Serra, M. V. & Dawber, M. (2012). Phys. Rev. Lett. 109, 067601.]). We chose tetragonal BTO (space group P4mm) for several reasons: (i) it is non-toxic compared with the popular ferroelectric material PbTiO3; (ii) perovskites grown on BTO have attracted considerable interest since the BTO crystal can accommodate a large amount of lattice strain during epitaxial growth; (iii) BTO is a typical ferroelectric system with a strong spontaneous polarization of 27 µC cm−2 (Wei et al., 2017[Wei, L. Y., Lian, C. & Meng, S. (2017). Phys. Rev. B, 95, 184102.]), and its ferroelectric behavior can be easily and significantly tuned by, for example, Sr substitution for Ba to form a solid solution (Tabata et al., 1994[Tabata, H., Tanaka, H. & Kawai, T. (1994). Appl. Phys. Lett. 65, 1970-1972.]); and (iv) the strong ferroelectric polarization of BTO can be easily switched by application of an electric field, thereby realizing the robust manipulation of the magnetism of ferromagnets. The LaFeO3 (LFO, Néel temperature TN = 740 K) (Acharya et al., 2010[Acharya, S., Mondal, J., Ghosh, S., Roy, S. K. & Chakrabarti, P. K. (2010). Mater. Lett. 64, 415-418.]), LaMnO3 (LMO, TN = 140 K) (Murakami et al., 1998[Murakami, Y., Hill, J. P., Gibbs, D., Blume, M., Koyama, I., Tanaka, M., Kawata, H., Arima, T., Tokura, Y., Hirota, K. & Endoh, Y. (1998). Phys. Rev. Lett. 81, 582-585.]) and LaCrO3 (LCO, TN = 253 K) (Wang et al., 2013[Wang, S., Huang, K., Zheng, B., Zhang, J. & Feng, S. (2013). Mater. Lett. 101, 86-89.]) compounds are all antiferromagnetic insulators with interlayer antiparallel spin alignments. They all have the same orthorhombic structure (space group Pbnm) with a continuous crystal framework, but possess different properties. In particular, the high TN indicates a strong superexchange coupling in the LFO bulk. The material compatibility of BTO and LaAO3 (LAO, A = Fe, Mn, and Cr) makes the formation of the superstructures experimentally achievable, therefore we have selected perovskite superstructures formed from LAO and BTO layers as the focus of this study, and report robust full control of magnetism by polarization, demonstrating how the displacement of oxygen and octahedral tilting can affect the ferroelectricity, ferromagnetism and magneto-electricity in a class of LAO/BTO superstructures.

2. Results and discussion

Our superstructures consist of single LAO (A = Fe, Mn and Cr) unit cells alternating with four BTO unit cells grown along the (001) direction (Dong & Dagotto, 2013[Dong, S. & Dagotto, E. (2013). Phys. Rev. B, 88, 140404R.]). The lattice parameters a, b and c of the superstructures are fully optimized to obtain the ground state, and the results are given in Tables S2–S4 of the supporting information. On one hand, the ferroelectric titanate needs to be relatively thick to maintain its polarization; however, the ultrathin LAO components only involve bilayers, which polarization can effectively penetrate. We hope that robust control of magnetism can be realized experimentally when the bilayers are coupled to ferroelectric polarizations. It is worth noting that through modern digital synthesis techniques, such as laser molecular beam epitaxy, superstructures can be fabricated with layer thickness on the unit-cell level and with near-perfect interfaces on an atomic scale (i.e. with minimal roughness, no misfit dislocations or other defects observed), which opens up exciting opportunities for the design of novel materials with richer physics (Shah et al., 2010[Shah, A. B., Ramasse, Q. M., Zhai, X., Wen, J. G., May, S. J., Petrov, I., Bhattacharya, A., Abbamonte, P., Eckstein, J. N. & Zuo, J. M. (2010). Adv. Mater. 22, 1156-1160.]). Thus, the design of our materials is experimentally practical.

In order to explore the robust control of magnetism when the LaAO3 (A = Mn, Fe and Cr) bilayers are coupled to ferroelectric polarizations, density functional theory calculations were carried out to determine the electronic and magnetic properties of the different superstructures. More details of our Vienna ab initio simulation package VASP5.3.5 (Kresse & Furthmüller, 1996[Kresse, G. & Furthmüller, J. (1996). Phys. Rev. B, 54, 11169-11186.]) calculations can be found in the supporting information. Before the simulation of superstructures, it is essential to check the physical properties of bulk LaFeO3, LaMnO3 and LaCrO3, which is not a trivial task. The agreement between the calculated results and other works confirms the reliability of our calculation set up (Table S1).

By introducing four monolayers of LAO supercells (including two LaO and two AO2 layers) and eight monolayers of BTO supercells (including four BaO and four TiO2 layers), we built the LAO/BTO (A = Fe, Mn and Cr) superstructures, obeying the typical perovskite sequence within the √2 ×√2 in-plane supercells along the (001) direction. The four monolayers of LAO and eight monolayers of BTO are shown in layers 1–8 and 9–12 of Fig. 1[link](a), respectively. Two asymmetric polar interfaces are hereby taken into account: the TiO2–LaO–AO2 and TiO2–BaO–AO2 layers are defined as the n- and p-type interfaces, respectively. When LAO is deposited on the BTO substrate, the ferroelectric polarization of BTO breaks the space-reversal symmetry, making the interfacial LAO layers partially polarized. In this work, two types of ferroelectric states with positive and negative polarizations have been adopted for full structural optimization and atomic relaxation. The polarization pointing from the n-type to the p-type interface is defined as the +P case, whereas −P corresponds to the case where the polarization points from LAO to BTO.

[Figure 1]
Figure 1
(a) Schematic structures of the LAO/BTO (A = Fe, Mn and Cr) superstructures. The arrows denote the directions of ferroelectric polarization. (b) The average local out-of-plane displacements between anions and cations for the −P case. (c) The average local out-of-plane displacements between anions and cations for the +P case.

The layer-resolved local dipole D, which is defined as the average value of an oxygen atom (anion) displacement, relative to the metal atom (cation) perpendicular to the interface, is exhibited in Figs. 1[link](b) and 1(c). Oxygen octahedra rotations and tilts in varying degrees, together with the Jahn–Teller distortions can be found in the three superstructures, but are not the main reasons for the magnetism variations of the superstructures (see Table S5 and the following discussions in the supporting information). The polarization has enormous effects on the displacement of oxygens, namely, D is negative and positive for the +P and −P cases, respectively. As shown in Fig. 1[link](b), the three −P curves show wave-like characteristics in the BTO layers, and then decrease dramatically across the interface. Finally, the local dipoles became negative in the LAO layers. As shown in Fig. 1[link](c), the oxygens in BTO are displaced away from the interface (layer 9) under the +P condition, indicating a net polarization (in BTO) pointing to the interface. Interestingly, the +P curves increase slowly at first, and then demonstrate very different behavior near the interface: the LFO curve changes very slowly, while the LMO curve increases moderately and remains negative. The dipole of LCO goes through the positive region, but then drops back to the negative region. The average bond lengths between O and Fe, Mn and Cr atoms near the interfaces are 1.86, 1.84 and 1.85 Å, respectively. These values are slightly shorter than for bulk LFO (1.94 Å), LMO (1.95 Å) and LCO (1.96 Å) compounds, indicating stronger interactions between adjacent Fe/Mn/Cr and O atoms. The large oxygen shifts can be expected to determine the interface properties, because these atoms mediate the interaction between LAO and BTO. It is worth noting that the +P curves are almost twice as low as those of −P case in the LAO region. This agrees with the fact that +P will split the AO2 layer, thus enhancing the charge disproportionation. The mechanism leading to such variations is related to the interfacial chemical-bonding effect, which will be clarified below.

As shown in Table 1[link], the net magnetizations of the G-type antiferromagnetic (G-AFM) LFO/BTO and C-type antiferromagnetic (C-AFM) LCO/BTO are 0 and 0.03 μB, respectively. Theoretically, these two net magnetizations should be zero. This difference is mainly due to the low symmetry that arises when we build the superstructures, which prevents the two Fe atoms in the same plane from being strictly symmetric. The difference in the Fe magnetic moment m2 for the opposing polarization directions reaches 0.07 μB, which leads to a minute net magnetization change of ΔM = 0.01 μB. The magnetic structure (G-AFM) of LFO/BTO is not affected by the ferroelectric polarization, which remains the same as that of bulk LFO. Correspondingly, LaFeO3 can be used as the substrate for magnetic storage devices. The net M of LaMnO­3 bilayers is switched from −0.01 to 3.82 μB, accompanying the +P to −P switching. The ferromagnetic configuration displays a large M, giving rise to 99.7% modulation by switching the polarization. This is higher than the result (93.9%) for the La0.75Sr0.25MnO3/BTO system obtained by Dong & Dagotto (2013[Dong, S. & Dagotto, E. (2013). Phys. Rev. B, 88, 140404R.]). Compared with Dong's results, in our work, not only the net magnetization M itself, but also the magnetic ordering can be modulated. As shown in Table 1[link], the total energies indicate that the ground state of the LCO/BTO superstructure tends to form the G-AFM order under the −P condition. The most striking result is that the C-AFM state has the minimum energy under the +P condition. It can be seen that the reason why the G-AFM→C-AFM phase transition occurs is the increasing magnetic moments (m2) of the Cr atoms as the ferroelectric polarization reverses.

Table 1
The calculated energy difference (per Fe/Mn/Cr atom) between the reference ferromagnetic (FM) and antiferromagnetic (AFM) states

m1 and m2 are the local magnetic moments for A cations using Wigner–Seitz spheres. M is the net magnetization. All moments are in units of μB. G-AFM = G-type antiferromagnetic, C-AFM = C-type antiferromagnetic and A-AFM = A-type antiferromagnetic.

Superstructure Ferroelectric Order ΔE (meV) m1 m2 M
LaFeO3/BaTiO3 +P FM 0 4.26 4.16 4.75
+P G-AFM −152.26 4.13 −4.07 −0.01
P FM 0 4.31 4.29 4.88
P G-AFM −247.87 4.16 −4.14 0
LaMnO3/BaTiO3 +P FM 0 3.59 3.27 3.43
+P A-AFM −30.48 3.24 −3.55 −0.01
P FM −42.45 3.58 3.77 3.82
P A-AFM 0 3.77 −3.37 0.20
LaCrO3/BaTiO3 +P FM 0 2.77 2.52 2.82
+P C-AFM −45.28 2.76 −2.47 0.03
P FM 0 2.80 2.78 2.86
P G-AFM −95.95 2.79 −2.78 0

By using perovskite oxides with the same structure, we have observed the FM→AFM phase transition, the C-AFM magnetic order and an unchanged magnetic structure when the polarization direction changes in the LAO/BTO superstructures. Entirely different magnetic orders are obtained in these superstructures, which might be directly associated with hole accumulation (antiferromagnetic state) and depletion (ferromagnetic state) around the interfaces, thus corresponding to the strongest magnetoelectric effect. Robust control of the magnetic moments may be realized by the altered polarization of BTO, which can be easily realized by experimentation.

The layer-resolved density of states of the LMO/BTO (001) superstructure is presented in Fig. 2[link](a). For the −P case, the top of the valence band in both the LMO and BTO regions shows a downward shift to the lower-energy region when approaching the interface, indicating that the internal electrical field of the whole superstructure is greatly affected by the ferroelectric polarization. The electrons do not appear in the BTO region and are only found in the two MnO­2 layers, i.e. the electron generation is not `at' but `beneath' the TiO­2/LaO interface. When the polarization reverses from −P to +P, both the majority and minority states are shifted toward the higher energy region so that the hybridization is enhanced, since the Mn-3d and O-2p bands are more delocalized. In Fig. 2[link](b), the superstructure exhibits metallic conductivity, which is mainly due to the small but identifiable density of states at EF. The Mn atoms in the tenth and twelfth layers exhibit perfect half-metallic properties. It is clear that the valence band is mainly composed of Mn-3d states hybridized with O-2p states, while the conduction band near the EF mainly consists of Mn-3d antibonding states with mixed O-2p antibonding states, as shown in the spin-down channels.

[Figure 2]
Figure 2
(a) The layer-resolved density of states of the LMO/BTO (001) superstructure (black line: +P, red line: −P). (b) The projected density of states of the LMO layers for the +P case. (c) The projected density of states of the LMO layers for the −P case. (The Fermi level EF is located at 0 eV and indicated by the vertical dashed line.)

As shown in Figs. 2[link](b) and 2(c), the projected density of states reflects the coexistence of spin-polarized eg and t2g states at Mn sites, which are delocalized and broadly distributed through the valence band. With the partially occupied eg and t2g states, the contribution to the energy comes from both the superexchange and double-exchange interactions. In Fig. 2[link](b), strong interactions among the Mn-eg, Mn-t2g and O-2p states are observed in the spin-up channel. The superexchange interaction contributed by the t2g states may stabilize the antiferromagnetic phase of the LMO/BTO superstructure. In Fig. 2[link](c), there is a considerable number of eg electrons that appear around −5 eV. These states mainly originate from the chemical bonding between the Mn-eg and O-2p states. The eg density of states shifts toward the lower-energy region compared with that in Fig. S1(b), which means that the occupancy of eg electrons increases. The partially occupied eg states can mediate the double-exchange between the Mn-t2g core spins, which overcomes the antiferromagnetic superexchange, so that a ferromagnetic configuration is stabilized. As we can see in Fig. 2[link](c), a notable feature is that the minority-spin density of states is completely unoccupied at the Fermi level, which causes the half-metallic behavior. Only a flat majority-spin eg band exists near the EF. The bandwidth of the Mn-t2g state is narrower than for the Mn-eg state, which reflects the easier localization of t2g orbitals than the eg orbitals. Generally speaking, the half-metallic character is mainly contributed by Mn, while the BTO film makes almost no contribution to the half-metallic states. The LMO/BTO system is also semi-metallic, namely, the density of states at EF tends to zero. This behavior is associated with a Dirac-cone-type band, which is demonstrated in our band structure analysis [Fig. S4(c)]. The layer-resolved and projected density of states are significantly changed from those of a non-spin polarized metal [Fig. 2[link](b)] to those of a half-metal [Fig. 2[link](c)]. In particular, the spin splitting density of states shown in the MnO2 layers indicates the ferromagnetic order of LMO/BTO in the −P case, which generates the magnetic moment.

As illustrated in Figs. 2[link](b) and 2(c), the destruction (+P case) of the half-metallic property is caused by the charge imbalance at the interface, in which the MnO2 layers lack electrons because the TiO2 layer does not donate electrons to the MnO2 layers. Therefore, the Fermi level EF shifts to a higher energy, thus destroying the half-metallicity. The downward shift of the projected density of states corresponds to hole depletion, which is mainly due to the fact that the density of states at EF decreases under the −P condition. The TiO2/LaO/MnO2 interface layers act as a magnetic switch to favor either the antiferromagnetic state (hole accumulation) or the ferromagnetic state (hole depletion) depending on the polarization orientation, which leads to a large variation in the magnetic moment and thus a large magnetoelectric effect. The whole LMO/BTO system is changed to a magnetic superstructure in the −P case because of the spin splitting of electrons. The LMO/BTO superstructure can be used as a sensing material for detecting harmful and toxic gases, since the O2 molecule is paramagnetic (Sobhan et al., 2015[Sobhan, M., Xu, Q., Katoch, A., Anariba, F., Kim, S. S. & Wu, P. (2015). Nanotechnology, 26, 175501.]). It can also be used as the photoanode for photocatalytic water splitting to recombine electrons and holes, as well as for separating photoelectrons from holes (Ji et al., 2013[Ji, W., Yao, K., Lim, Y. F., Liang, Y. C. & Suwardi, A. (2013). Appl. Phys. Lett. 103, 062901.]).

The LFO/BTO superstructure remains G-AFM in both the ±P cases. No apparent 2D electron gas can be seen near the Fermi level in the +P case. A metal–insulator transition can be observed for the LFO/BTO system when the polarization of BTO reverses (see the density of states results in the supporting information, Fig. S1). For the LCO/BTO system, the metallic 2D electron gas can be observed near the EF of the BaO/CrO2 interface in the +P case. For the −P case, the superstructure becomes a semiconductor. We found that the electronic and magnetic properties could be greatly affected by the different atomic environments and interface states. The whole LCO/BTO system is G-AFM (C-AFM) in the −P (+P) case, which agrees well with the results listed in Table 1[link] (see the density of states results in the supporting information, Fig. S2). In Figs. S3–S5, we show the band structures of different tetragonal superstructures with different performances. The LFO/BTO system is a metallic semiconductor with an indirect band gap of 1.12 eV in the +P and −P cases. The LMO/BTO system shows a topological feature.

Table 2[link] shows the number of electrons for the 3d orbitals (both eg and t2g) of the magnetic atoms A (A = Fe, Mn and Cr). It is clearly seen that the 3d electrons of Mn in the LMO/BTO superstructure increase and these electrons come from the BTO part. It is also found that the 3d electrons of Fe and Cr decrease and the lost electrons are transferred to the BTO region. The charge transfer between the BTO and LAO regions has little effect on the magnetic configurations of the superstructures. The reason for the change of magnetic orders could be due to electron transfer between the eg and t2g states. When the ferroelectric polarization changes from +P to −P, we found that the electrons transferred from the t2g orbital to the eg orbital (approximately 0.14) lead to magnetic transition from A-AFM to ferromagnetic in the LMO/BTO system. In the LCO/BTO system, in fact, the eg electron shows a large reduction while the t2g electron shows a major increase compared with the bulk LCO case. The electrons transferred from the eg orbital to the t2g orbital (approximately 0.15) may be the reason for the C-AFM→G-AFM phase transition. The electron transfer in the LFO/BTO system is clearly much smaller with respect to the other two cases, indicating the unchanged G-AFM state in both the ±P cases. Polarization-induced distortion variation of oxygen octahedra may lead to further degeneracy of 3d orbitals and electron transfer between eg and t2g. It is worth noting that Aruta et al. (2009[Aruta, C., Adamo, C., Galdi, A., Orgiani, P., Bisogni, V., Brookes, N. B., Cezar, J. C., Thakur, P., Perroni, C. A., De Filippis, G., Cataudella, V., Schlom, D. G., Maritato, L. & Ghiringhelli, G. (2009). Phys. Rev. B, 80, 140405R.]) have investigated the magnetic properties of the LaMnO3/SrMnO3 superstructure through an X-ray linear dichroism technique. They demonstrated that the AFM→FM transition could be attributed to the electron transfer of the partially occupied Mn-eg orbitals. Nevertheless, the reason for the control of magnetism in our superstructures still needs to be verified by future experiments.

Table 2
The number of electrons for the 3d orbitals (both eg and t2g) of the magnetic atoms in the LAO (A = Fe, Mn and Cr) bulks and the LAO/BTO superstructures

Materials Ferroelectric Order eg t2g 3d orbital
LaFeO3 0 G-AFM 2.17 3.82 5.99
LaFeO3/BaTiO3 +P G-AFM 2.52 3.20 5.72
P G-AFM 2.57 3.16 5.73
LaMnO3 0 A-AFM 1.87 2.89 4.76
LaMnO3/BaTiO3 +P A-AFM 1.75 3.16 4.91
P FM 1.91 3.03 4.94
LaCrO3 0 G-AFM 1.53 2.60 4.13
LaCrO3/BaTiO3 +P C-AFM 1.17 2.80 3.97
P G-AFM 1.03 3.04 4.07

It was found that robust manipulation of the magnetism, including the exchange interaction energy and magnetic ordering in LAO/BTO (A = Fe, Mn and Cr) superstructures, can be achieved by ferroelectric polarization. Along with the manipulation of the magnetism, the electronic structure was also significantly modified by polarization, and half metallicity was observed in LMO and LCO/BTO with the appearance of a 2D electron gas at the interface. The ferroelectric polarization of BTO changes the Fe—O, Mn—O and Cr—O bond lengths of layers at the interface. Since the original bond lengths and bond strengths are different among the LFO, LMO and LCO bulks, the bond length variations of the three superstructure systems are also different. These bond length changes will lead to the tilting of oxygen octahedra at the interface, causing structural and electronic reconstruction. Ferroelectric polarization can modulate the carrier concentration by introducing an accumulation of spin-polarized electrons and a depletion of holes near the interfaces, and thus can control the interface magnetic moments and net magnetization correspondingly. These superstructures are stable, controllable, easily grown and low-cost, promising future applications in spintronics, chemical gas sensing and information storage.

3. Conclusions

Although some multiferroic materials have been extensively investigated, finding strong magnetoelectric couplings for the full control of magnetization remains challenging. Here, we have built LaFeO3/BaTiO3, LaMnO3/BaTiO3 and LaCrO3/BaTiO3 (001) superstructures as proof of the potential for robust control of the magnetism when these magnetic layers are coupled to ferroelectric polarizations. Both superexchange and double-exchange interactions exist in these superstructures. The superexchange interaction can be found in the LMO/BTO superstructure in the +P case, while the double-exchange interaction plays a role in the remaining five cases. The LFO/BTO system shows a G-type antiferromagnetic order for both polarization directions. The BTO region is insulating, and the conductivity of this superstructure is entirely controlled by the LFO films. For the +P condition, the LMO/BTO system exhibits a metallic character and has A-AFM order. The stronger superexchange interaction contributed by the t2g states stabilizes the antiferromagnetic phase of the LMO/BTO superstructure. In contrast, under the −P condition, it changes to ferromagnetic due to the spin splitting of the mixed Mn-3d and O-2p states. The LMO/BTO superstructure finally acquires half-metallic and semi-metallic character, which may be a result of the strong spin polarization. A maximal change of 99.7% of the net magnetization can be achieved by switching the ferroelectric polarization. Upon polarization switching, the magnetic moment of Mn in particular shows significant modulations, as listed in Table 1[link]. The TiO2/LaO/MnO2 interface acts as a magnetic switch, which leads to a large variation in the magnetic moment and thus, the largest magnetoelectric effect among the superstructures. The LCO/BTO system can retain the G- and C-AFM configurations in the −P and +P cases, respectively. The magnetism variations of the three superstructures are mainly due to valence-state changes of the Fe/Mn/Cr ions and charge transfer among the Ti, O and Fe/Mn/Cr ions. The strong magnetoelectric coupling mediated by the interfacial effect enables full control of the magnetism. The LFO/BTO and LCO/BTO superstructures show a metal-insulator transition when the polarization of BTO reverses. The direction change of the ferroelectric polarization leads to electron transfer between the eg and t2g orbitals, which determines the variation of magnetic order of the three superstructures. The robust control of magnetism demonstrated in this article will provide a feasible scheme for experimental work.

4. Related literature

The following references are cited in the supporting information: An et al. (2017[An, M., Weng, Y. K., Zhang, H. M., Zhang, J. J., Zhang, Y. & Dong, S. (2017). Phys. Rev. B, 96, 235112.]); Betancourt et al. (2017[Betancourt, J., Paudel, T. R., Tsymbal, E. Y. & Velev, J. P. (2017). Phys. Rev. B, 96, 045113.]); Blöchl (1994[Blöchl, P. E. (1994). Phys. Rev. B, 50, 17953-17979.]); Dabaghmanesh et al. (2017[Dabaghmanesh, S., Sarmadian, N., Neyts, E. C. & Partoens, B. (2017). Phys. Chem. Chem. Phys. 19, 22870-22876.]); Ding et al. (2010[Ding, J. L., Lü, X. M., Shu, H. M., Xie, J. M. & Zhang, H. (2010). Mater. Sci. Eng. B, 171, 31-34.]); Dudarev et al. (1998[Dudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. & Sutton, A. P. (1998). Phys. Rev. B, 57, 1505-1509.]); Elemans et al. (1971[Elemans, J. B. A. A., Van Laar, B., Van Der Veen, K. R. & Loopstra, B. O. (1971). J. Solid State Chem. 3, 238-242.]); Fang & Nagaosa (2004[Fang, Z. & Nagaosa, N. (2004). Phys. Rev. Lett. 93, 176404.]); Hashimoto et al. (2010[Hashimoto, T., Ishibashi, S. & Terakura, K. (2010). Phys. Rev. B, 82, 045124.]); He et al. (2010[He, J., Borisevich, A., Kalinin, S. V., Pennycook, S. J. & Pantelides, S. T. (2010). Phys. Rev. Lett. 105, 227203.]); Koehler & Wollan (1957[Koehler, W. C. & Wollan, E. O. (1957). J. Phys. Chem. Solids, 2, 100-106.]); Kotomin et al. (2005[Kotomin, E. A., Evarestov, R. A., Mastrikov, Y. A. & Maier, J. (2005). Phys. Chem. Chem. Phys. 7, 2346-2350.]); Li et al. (2017[Li, Y. R., Hou, Z. T., Wang, T. X., Li, Y., Liu, H. Y., Dai, X. F. & Liu, G. D. (2017). J. Phys. Conf. Ser. 827, 012015.]); Monkhorst & Pack (1976[Monkhorst, H. J. & Pack, J. D. (1976). Phys. Rev. B, 13, 5188-5192.]); Muñoz et al. (2004[Muñoz, D., Harrison, N. M. & Illas, F. (2004). Phys. Rev. B, 69, 085115.]); Perdew et al. (1996[Perdew, J. P., Burke, K. & Ernzerhof, M. (1996). Phys. Rev. Lett. 77, 3865-3868.]); Pinsard-Gaudart et al. (2001[Pinsard-Gaudart, L., Rodríguez-Carvajal, J., Daoud-Aladine, A., Goncharenko, I., Medarde, M., Smith, R. I. & Revcolevschi, A. (2001). Phys. Rev. B, 64, 064426.]); Prado-Gonjal et al. (2011[Prado-Gonjal, J., Arévalo-López, Á. M. & Morán, E. (2011). Mater. Sci. Bull. 46, 222-230.]); Scafetta et al. (2014[Scafetta, M. D., Cordi, A. M., Rondinelli, J. M. & May, S. J. (2014). J. Phys. Condens. Matter, 26, 505502.]); Selbach et al. (2012[Selbach, S. M., Tolchard, J. R., Fossdal, A. & Grande, T. (2012). J. Solid State Chem. 196, 249-254.]); Shein (2005[Shein, I. R. (2005). Phys. Solid State, 47, 2082-2088.]); Siemons et al. (2007[Siemons, M., Leifert, A. & Simon, U. (2007). Adv. Funct. Mater. 17, 2189-2197.]); Sushko et al. (2013[Sushko, P. V., Qiao, L., Bowden, M., Varga, T., Exarhos, G. J., Urban, F. K. III, Barton, D. & Chambers, S. A. (2013). Phys. Rev. Lett. 110, 077401.]); Yang et al. (1999[Yang, Z. Q., Huang, Z., Ye, L. & Xie, X. D. (1999). Phys. Rev. B, 60, 15674-15682.]).

Supporting information


Acknowledgements

The authors would like to thank Dr Tania Silver for her critical reading of the manuscript and valuable remarks.

Funding information

Funding for this research was provied by the National Natural Science Foundation of China (award Nos. 51571083; 11674083), as well as the Key Scientific and Technological Projects in Henan Province (award No. 162102210169).

References

First citationAcharya, S., Mondal, J., Ghosh, S., Roy, S. K. & Chakrabarti, P. K. (2010). Mater. Lett. 64, 415–418.  CrossRef CAS Google Scholar
First citationAn, M., Weng, Y. K., Zhang, H. M., Zhang, J. J., Zhang, Y. & Dong, S. (2017). Phys. Rev. B, 96, 235112.  CrossRef Google Scholar
First citationAruta, C., Adamo, C., Galdi, A., Orgiani, P., Bisogni, V., Brookes, N. B., Cezar, J. C., Thakur, P., Perroni, C. A., De Filippis, G., Cataudella, V., Schlom, D. G., Maritato, L. & Ghiringhelli, G. (2009). Phys. Rev. B, 80, 140405R.  CrossRef Google Scholar
First citationBetancourt, J., Paudel, T. R., Tsymbal, E. Y. & Velev, J. P. (2017). Phys. Rev. B, 96, 045113.  CrossRef Google Scholar
First citationBlöchl, P. E. (1994). Phys. Rev. B, 50, 17953–17979.  CrossRef Web of Science Google Scholar
First citationBousquet, E., Dawber, M., Stucki, N., Lichtensteiger, C., Hermet, P., Gariglio, S., Triscone, J. M. & Ghosez, P. (2008). Nature, 452, 732–736.  CrossRef CAS Google Scholar
First citationBurton, J. D. & Tsymbal, E. Y. (2009). Phys. Rev. B, 80, 174406.  CrossRef Google Scholar
First citationCallori, S. J., Gabel, J., Su, D., Sinsheimer, J., Fernandez-Serra, M. V. & Dawber, M. (2012). Phys. Rev. Lett. 109, 067601.  CrossRef Google Scholar
First citationCui, B., Song, C., Mao, H., Wu, H., Li, F., Peng, J., Wang, G., Zeng, F. & Pan, F. (2015). Adv. Mater. 27, 6651–6656.  CrossRef CAS Google Scholar
First citationDabaghmanesh, S., Sarmadian, N., Neyts, E. C. & Partoens, B. (2017). Phys. Chem. Chem. Phys. 19, 22870–22876.  CrossRef CAS Google Scholar
First citationDing, J. L., Lü, X. M., Shu, H. M., Xie, J. M. & Zhang, H. (2010). Mater. Sci. Eng. B, 171, 31–34.  CrossRef CAS Google Scholar
First citationDong, S. & Dagotto, E. (2013). Phys. Rev. B, 88, 140404R.  CrossRef Google Scholar
First citationDong, S., Yamauchi, K., Yunoki, S., Yu, R., Liang, S., Moreo, A., Liu, J. M., Picozzi, S. & Dagotto, E. (2009). Phys. Rev. Lett. 103, 127201.  CrossRef Google Scholar
First citationDong, S., Yu, R., Liu, J. M. & Dagotto, E. (2009). Phys. Rev. Lett. 103, 107204.  CrossRef Google Scholar
First citationDong, S., Yu, R., Yunoki, S., Alvarez, G., Liu, J. M. & Dagotto, E. (2008). Phys. Rev. B, 78, 201102R.  CrossRef Google Scholar
First citationDuan, C. G., Jaswal, S. S. & Tsymbal, E. Y. (2006). Phys. Rev. Lett. 97, 047201.  CrossRef Google Scholar
First citationDudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. & Sutton, A. P. (1998). Phys. Rev. B, 57, 1505–1509.  CrossRef CAS Google Scholar
First citationElemans, J. B. A. A., Van Laar, B., Van Der Veen, K. R. & Loopstra, B. O. (1971). J. Solid State Chem. 3, 238–242.  CrossRef CAS Google Scholar
First citationFang, Z. & Nagaosa, N. (2004). Phys. Rev. Lett. 93, 176404.  CrossRef Google Scholar
First citationFechner, M., Zahn, P., Ostanin, S., Bibes, M. & Mertig, I. (2012). Phys. Rev. Lett. 108, 197206.  CrossRef Google Scholar
First citationGibert, M., Viret, M., Torres-Pardo, A., Piamonteze, C., Zubko, P., Jaouen, N., Tonnerre, J. M., Mougin, A., Fowlie, J., Catalano, S., Gloter, A., Stéphan, O. & Triscone, J. M. (2015). Nano Lett. 15, 7355–7361.  CrossRef CAS Google Scholar
First citationHashimoto, T., Ishibashi, S. & Terakura, K. (2010). Phys. Rev. B, 82, 045124.  CrossRef Google Scholar
First citationHe, J., Borisevich, A., Kalinin, S. V., Pennycook, S. J. & Pantelides, S. T. (2010). Phys. Rev. Lett. 105, 227203.  Web of Science CrossRef PubMed Google Scholar
First citationHou, Y. S., Xiang, H. J. & Gong, X. G. (2014). Phys. Rev. B, 89, 064415.  CrossRef Google Scholar
First citationHuang, X. & Dong, S. (2014). Mod. Phys. Lett. B, 28, 1430010.  CrossRef Google Scholar
First citationJi, W., Yao, K., Lim, Y. F., Liang, Y. C. & Suwardi, A. (2013). Appl. Phys. Lett. 103, 062901.  CrossRef Google Scholar
First citationKim, H. S. & Christen, H. M. (2010). J. Phys. Condens. Matter, 22, 146007.  CrossRef Google Scholar
First citationKoehler, W. C. & Wollan, E. O. (1957). J. Phys. Chem. Solids, 2, 100–106.  CrossRef CAS Web of Science Google Scholar
First citationKotomin, E. A., Evarestov, R. A., Mastrikov, Y. A. & Maier, J. (2005). Phys. Chem. Chem. Phys. 7, 2346–2350.  CrossRef CAS Google Scholar
First citationKresse, G. & Furthmüller, J. (1996). Phys. Rev. B, 54, 11169–11186.  CrossRef CAS Web of Science Google Scholar
First citationLee, J. H., Delaney, K. T., Bousquet, E., Spaldin, N. A. & Rabe, K. M. (2013). Phys. Rev. B, 88, 174426.  CrossRef Google Scholar
First citationLi, Y. R., Hou, Z. T., Wang, T. X., Li, Y., Liu, H. Y., Dai, X. F. & Liu, G. D. (2017). J. Phys. Conf. Ser. 827, 012015.  CrossRef Google Scholar
First citationMishina, E. D., Buryakov, A. M., Sherstyuk, N. E., Sigov, A. S. & Rasing, T. (2016). Ferroelectrics, 500, 37–46.  CrossRef CAS Google Scholar
First citationMonkhorst, H. J. & Pack, J. D. (1976). Phys. Rev. B, 13, 5188–5192.  CrossRef Web of Science Google Scholar
First citationMuñoz, D., Harrison, N. M. & Illas, F. (2004). Phys. Rev. B, 69, 085115.  Google Scholar
First citationMurakami, Y., Hill, J. P., Gibbs, D., Blume, M., Koyama, I., Tanaka, M., Kawata, H., Arima, T., Tokura, Y., Hirota, K. & Endoh, Y. (1998). Phys. Rev. Lett. 81, 582–585.  Web of Science CrossRef CAS Google Scholar
First citationPerdew, J. P., Burke, K. & Ernzerhof, M. (1996). Phys. Rev. Lett. 77, 3865–3868.  CrossRef PubMed CAS Web of Science Google Scholar
First citationPinsard-Gaudart, L., Rodríguez-Carvajal, J., Daoud-Aladine, A., Goncharenko, I., Medarde, M., Smith, R. I. & Revcolevschi, A. (2001). Phys. Rev. B, 64, 064426.  Google Scholar
First citationPrado-Gonjal, J., Arévalo-López, Á. M. & Morán, E. (2011). Mater. Sci. Bull. 46, 222–230.  CAS Google Scholar
First citationRoqueta, J., Pomar, A., Balcells, L., Frontera, C., Valencia, S., Abrudan, R., Bozzo, B., Konstantinović, Z., Santiso, J. & Martínez, B. (2015). Cryst. Growth Des. 15, 5332–5337.  CrossRef CAS Google Scholar
First citationScafetta, M. D., Cordi, A. M., Rondinelli, J. M. & May, S. J. (2014). J. Phys. Condens. Matter, 26, 505502.  CrossRef Google Scholar
First citationSelbach, S. M., Tolchard, J. R., Fossdal, A. & Grande, T. (2012). J. Solid State Chem. 196, 249–254.  CrossRef CAS Google Scholar
First citationShah, A. B., Ramasse, Q. M., Zhai, X., Wen, J. G., May, S. J., Petrov, I., Bhattacharya, A., Abbamonte, P., Eckstein, J. N. & Zuo, J. M. (2010). Adv. Mater. 22, 1156–1160.  CrossRef CAS Google Scholar
First citationShein, I. R. (2005). Phys. Solid State, 47, 2082–2088.  CrossRef CAS Google Scholar
First citationSiemons, M., Leifert, A. & Simon, U. (2007). Adv. Funct. Mater. 17, 2189–2197.  CrossRef CAS Google Scholar
First citationSobhan, M., Xu, Q., Katoch, A., Anariba, F., Kim, S. S. & Wu, P. (2015). Nanotechnology, 26, 175501.  CrossRef Google Scholar
First citationSushko, P. V., Qiao, L., Bowden, M., Varga, T., Exarhos, G. J., Urban, F. K. III, Barton, D. & Chambers, S. A. (2013). Phys. Rev. Lett. 110, 077401.  CrossRef Google Scholar
First citationTabata, H., Tanaka, H. & Kawai, T. (1994). Appl. Phys. Lett. 65, 1970–1972.  CrossRef CAS Google Scholar
First citationTakamura, Y., Folven, E., Shu, J. B. R., Lukes, K. R., Li, B., Scholl, A., Young, A. T., Retterer, S. T., Tybell, T. & Grepstad, J. K. (2013). Phys. Rev. Lett. 111, 107201.  CrossRef Google Scholar
First citationVila-Fungueiriño, J. M., Rivas-Murias, B., Rodríguez-González, B., Txoperena, O., Ciudad, D., Hueso, L. E., Lazzari, M. & Rivadulla, F. (2015). Appl. Mater. Interfaces, 7, 5410–5414.  Google Scholar
First citationWang, S., Huang, K., Zheng, B., Zhang, J. & Feng, S. (2013). Mater. Lett. 101, 86–89.  CrossRef CAS Google Scholar
First citationWang, X. R., Li, C. J., Lü, W. M., Paudel, T. R., Leusink, D. P., Hoek, M., Poccia, N., Vailionis, A., Venkatesan, T., Coey, J. M. D., Tsymbal, E. Y., Ariando, & Hilgenkamp, H. (2015). Science, 349, 716–719.  Google Scholar
First citationWei, L. Y., Lian, C. & Meng, S. (2017). Phys. Rev. B, 95, 184102.  CrossRef Google Scholar
First citationWeng, Y. K., Huang, X., Yao, Y. G. & Dong, S. (2015). Phys. Rev. B, 92, 195114.  CrossRef Google Scholar
First citationWeng, Y. K., Lin, L. F., Dagotto, E. & Dong, S. (2016). Phys. Rev. Lett. 117, 037601.  CrossRef Google Scholar
First citationYang, Z. Q., Huang, Z., Ye, L. & Xie, X. D. (1999). Phys. Rev. B, 60, 15674–15682.  CrossRef CAS Google Scholar
First citationYin, Y. W., Burton, J. D., Kim, Y. M., Borisevich, A. Y., Pennycook, S. J., Yang, S. M., Noh, T. W., Gruverman, A., Li, X. G., Tsymbal, E. Y. & Li, Q. (2013). Nat. Mater. 12, 397–402.  CrossRef CAS Google Scholar
First citationZhai, X., Cheng, L., Liu, Y., Schlepütz, C. M., Dong, S., Li, H., Zhang, X., Chu, S., Zheng, L., Zhang, J., Zhao, A., Hong, H., Bhattacharya, A., Eckstein, J. N. & Zeng, C. (2014). Nat. Commun. 5, 4283.  CrossRef Google Scholar
First citationZhang, A. M., Zhang, W. C., Wu, X. S. & Lin, J. G. (2017). AIP Adv. 7, 055837.  Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

IUCrJ
Volume 6| Part 2| March 2019| Pages 189-196
ISSN: 2052-2525