Controllable magnetization and resistivity jumps of manganite thin films on BaTiO₃ substrate

In complex systems such as perovskite manganites, the electronic and magnetic properties are strongly coupled to the lattice structure.^1–4 Using substrate-induced epitaxial strain in thin films thus becomes a useful way to control the physical properties of manganite.^5–9 In particular, ferroelectric substrate such as BaTiO₃ (BTO) offers additional tunability of strain since the lattice structure of BTO can be influenced by electric field.^10–13 In bulk, on cooling the structure of BTO changes from cubic to tetragonal at ferroelectric Tc of ∼400 K, and from tetragonal to monoclinic at 268 K, and from monoclinic (M) to rhombohedral (R) at 183 K.^14,15 Even without applying electric field, it has been shown that the structural transitions of the BTO induces distinct jumps of magnetization of 3d metals,^16–21 Sr₂CrReO₆,^12,22–24 magnetites^25–28 and manganites^29–33 films grown epitaxially on BTO substrates. The magnetization jump becomes especially interesting for the La_1−xSr_xMnO₃ (LSMO) and La_1−xCa_xMnO₃ (LCMO) thin films grown on BTO substrate since it gives rise to a giant and reversible magnetocaloric effect.³⁴ For LSMO/BTO²⁹ and LCMO/BTO^30–33 thin films, resistivity jumps have also been observed during the BTO structural transition. Surprisingly, both upward and downward jumps have been observed for the LCMO/BTO system, implying that the physical origin of the jumps still remains unclear.

In this work, we clarify the puzzle by demonstrating that electric field can be used to control the direction of the jumps. Specifically, the electric field was applied to the BTO substrate during post growth annealing process, which determines the orientation of the c-axis of the BTO substrate at room temperature. Our results clearly indicate two physical mechanisms that govern the direction of the jumps: 1) At low magnetic field, the direction of magnetization jump is determined by the magnetic anisotropy of the system, although the direction of resistivity jump is not affected; 2) At high magnetic field, when the magnetization is saturated along the field direction and magnetic anisotropy no longer plays a role, direction of both magnetization and resistivity jumps is governed by the relative orientation of c axis in the “M” phase of the BTO substrate with respect to the LCMO film plane. These clarifications allow us to use electric field to control the jumps of the manganite films on BTO substrates in the desired manner.

In this work, we grew 50 nm epitaxial thin films of La_5/8Ca_3/8MnO₃ (Tc ∼ 260 K) on the (001)-oriented BaTiO₃ substrates by Laser molecular beam epitaxy (MBE) (248 nm, 2 Hz, 2 J/cm² fluence) under oxygen atmosphere (8% Ozone) of about 3.0×10⁻³ mbar. The substrate temperature was kept at 820 °C to allow atomically flat growth front, which produces thin film with high quality in terms of minimum oxygen deficiency and other structural defects. Thin film quality was examined by in-situ reflection high energy electron diffraction (RHEED), ex-situ atomic force microscope (AFM) and grazing incident X-ray diffraction (GIXRD) (see supplementary: sample growth³⁷).

The BTO substrate is in cubic phase at high temperatures during thin film growth and becomes tetragonal when cooling back to room temperature after growth. The c-axis of the tetragonal phase can be controlled to either align along surface normal ((001)-oriented) or lie parallel to the surface ((100)-oriented) by applying an electric field (∼23 kV/cm) in the corresponding direction during post growth treatment (details see supplementary: Fig. S2³⁷). We have performed XRD measurements on LCMO thin films on these two types of substrates, as shown in Fig. 1. Although residual peak from the other orientation is present in both (100) and (001) samples, the main peak in the XRD always corresponds to the orientation of the electric field. This implies that the orientation of the majority of the samples do follow the electric field.

The actual orientation of the c-axis at room temperature governs the surface symmetry of the BTO substrates and thus the lattice symmetry of the LCMO thin films upon further cooling. The lattice constants of LCMO/BTO(001) and LCMO/BTO(100) thin films have been determined using XRD and GIXRD at room temperature and upon the M→R structural transition, as shown in Fig. 2 (see supplementary Fig. S3 - Fig. S5³⁷). Following the convention, for the BTO substrate we define [100] as a axis, [010] as b axis, and [001] as c axis. Each of these three lattice constants has its own temperature dependence. The temperature dependence of the three LCMO films’ lattice constants appears to one-to-one follow that of the BTO. We simply define the a, b, c of the LCMO films based on their temperature dependence with respect to that of the BTO substrate. Upon the M→R structural transition, the in-plane symmetry of the (001)-oriented BTO substrate surface changes from two-fold in M phase (in plane a < b, c is along surface normal) to nearly four-fold in R phase. For the (100)-oriented substrate surface, the surface plane has four-fold symmetry (in plane b = c, a is along surface normal) in both M and R phases. Fig. 2 shows the strain of LCMO film growth on the BTO substrate have little relax but still in a strained condition compare with the bulk (∼3.856 Å). Lattice constants of the LCMO film are strongly affected by the surface symmetry of the BTO substrates, and a clear in-plane compress is observed for the LCMO/BTO (100) film upon the M→R structural transition.

The transport and magnetic properties were conducted on a Quantum Design Physical Property Measurement System (PPMS) using 4-probe transport geometry (keithley 2400, the Measuring current is 100 nA, dc) and on a Quantum Design Superconducting Quantum Interference Device (SQUID), respectively. Upon cooling, at the temperature point where the M→R transition of the BTO substrate occurs, we observe distinct jumps for both magnetization and resistivity. For magnetization, the direction of the jumps depends strongly on the amplitude and the direction of the applied magnetic field. Fig. 3(a) shows filed cooling measurements of magnetization of the LCMO/BTO (001) film. Upon cooling, both low (∼100 Oe) and high (∼1 T) magnetic fields have been applied along both a (i.e. [100]) and b (i.e. [010]) directions in the plane of the LCMO film. At 100 Oe, the magnetization jumps down and up with the field applied along [100] and [010] directions, respectively. At 1 T, the magnetization jumps down no matter which direction the field is applied along.

The direction of the jumps in Fig. 3(a) can be easily understood from the temperature-dependent behavior of the magnetic anisotropy of the system. Fig. 3(b) shows M-H loops measured along both [100] and [010] directions before (190 K) and after (180 K) the M→R transition of the BTO (001) substrate. At 190 K, the remnant magnetization (Mr) is considerably larger when field is applied along [100] direction than [010] direction, indicating that [100] is the easy magnetization axis before the M→R transition. This in-plane anisotropy is caused by the in-plane two-fold symmetry of the monoclinic structure of the BTO (001) substrate^11,13 shown in Fig. 2(a). At 180 K, the Mr becomes nearly identical for both [100] and [010] directions, in agreement with the in-plane four-fold symmetry of the rhombohedral structure of BTO (001) substrate after the M→R transition. The 180 K Mr value appears to be in between the 190 K Mr values measured along [100] and [010] directions. This explains why at a low field of 100 Oe (far below the saturation field), the magnetization jumps down in [100] direction and jumps up in [010] direction when the temperature decreases from 190 K to 180 K. At a larger field of 1 T, the magnetization is fully saturated along both [100] and [010] directions before the M→R transition (190 K). After the M→R transition, the net magnetizations drops to a smaller value as clearly indicated in Fig. 3(a). This means that downward jump of the magnetization at 1 T reflects the drop of net magnetization induced by the M→R transition, which is independent of the field direction.

Sharp jumps of the resistivity have also been observed at the M→R transition point, as shown in Fig. 4(a). Unlike the magnetization, the resistivity jumps upward independent of the magnitude and direction of the applied magnetic field upon cooling. Comparing the R-T curves measured with 1 T field applied along [100] and [010] directions, one can clearly see that the anisotropic magnetoresistance is negligible compared to the M→R transition induced resistivity jump. This implies that the sign of the resistivity jump should correlate closely with that of the net magnetization jump of the system. Since the net magnetization drops, i.e. the ferromagnetic volume reduces after the M→R transition (Fig. 3(a), (b)), the system becomes more insulating and thus the resistivity jumps upward as shown in Fig. 4(a).

Remarkably, the direction of the jumps for both magnetization and resistivity is reversed for the LCMO/BTO (100) film. Fig. 5(a) shows temperature dependent magnetization measured with small (100 Oe) and large field (1 T) applied long both [001] and [010] directions. No anisotropic behavior can be seen and in all cases the magnetization jumps up at the M→R transition point. This is backed by the M-H curves (Fig. 5(b)) measured before (190 K) and after (180 K) the M→R transition. While the net magnetization increases by more than 40% after the M→R transition (Fig. 5(b)), no magnetic anisotropy is observed along [001] and [010] directions before and after the M→R transition because the four-fold symmetry is preserved in the M→R transition of the LCMO/BTO (100) film (Fig. 2(b)). This explains why the magnetization consistently jumps up with both low and high fields applied along both [001] and [010] directions.

The large increase of ferromagnetic volume in the M→R transition results in a dramatic drop of resistivity as shown by the R-T curves in Fig. 4(b). At the M→R transition point, the resistivity drops by nearly an order of magnitude with field applied in both [001] and [010] directions. The anisotropic magnetoresistance is again negligible compared to the size of the resistivity jump.

We now turn to discuss the correlation between the resistivity/magnetization jumps and the symmetry changes of BTO, as well as the relationship between the magnetic and resistive jumps, For the LCMO/BTO(100) samples, the BTO M→R transition leads to reduction of both in-plane lattice constants (b and c), which enhances charge career hopping between neighboring Mn ions. In turn, the resistivity decreases giving rise to the downward jump. For the LCMO/BTO(001) samples, the BTO M→R transition leads to a slight reduction of one in-plane lattice constant (b) but a clear increase of the other in-plane lattice constant (a). Although the quantitative understanding relies on theoretical calculation, it appears that the increased a-lattice constant dominates and gives rise to an upward jump of resistivity. As for the relationship between the jumps of resistivity and magnetization, they are consistently in the opposite direction, which is expected based on the double exchange model in manganites.³⁵ 

Finally, we emphasize that our conclusion is not affected by the fact that some residual domains of the other orientation exist in either (100) and (001) oriented LCMO/BTO systems. This is because in each case, the film is dominated by one particular crystallographic orientation as indicated by our XRD data. The global transport and magnetic behavior in particular the direction of the jumps should thus reflect the majority part of the films. Other structural factors such as grain boundary and domain wall may affect local properties³⁶ and play a role in the amplitude of the jumps¹⁶ (shown in the Fig. 3(a), Fig. 4 and Fig. 5(a)), but not the direction of the jumps. Although we could not completely rule out the influence of the reconstruction induced by the two BTO, there is neither experimental evidence nor physical arguments why they can lead to jumps, let alone dictating the direction of the jumps. As for the polarization of the BTO, it could play a role in ultrathin regime of the films, but is negligible for 50 nm thick films as confirmed by the fact that it makes not difference for the jumps no matter we polarize the BTO in up or down directions. Therefore, the opposite directions of the jumps are mostly likely caused by the different surface symmetry of the two types of BTO substrates, which can be controlled by electric field.

In summary, we demonstration that the direction of the BTO M→R transition induced sudden jumps of the magnetization and resistivity of the LCMO/BTO thin films can be controlled by electric field. This allows us to show that the post-growth substrate orientation controls the relative orientation of the monoclinic c-axis of the BTO substrate with respect to the film plane, which determines the direction of the magnetization and resistivity jumps manganites films grown on BTO substrates. This knowledge should also be highly valuable for future applications of giant and reversible magnetocaloric effect of manganites grown on BTO substrates.

We would like to thank Hui Wang and Ruqian Wu for valuable discussions. This work was supported by the National Basic Research Program of China (973 Program) under the grant No. 2011CB921800, 2013CB932901 and 2014CB921104, National Natural Science Foundation of China (91121002, 11274071 and 11205235), Shanghai Municipal Natural Science Foundation (11ZR1402600), the Wuhan National High Magnetic Field Center (WHMFCKF2011008). The authors thank beamline BL14B1 of SSRF (Shanghai Synchrotron Radiation Facility) for providing the beamtime.