Giant electrostatic modification of magnetism via electrolyte-gate-induced cluster percolation in $\mathrm{L}{\mathrm{a}}_{1-x}\mathrm{S}{\mathrm{r}}_x\mathrm{Co}{\mathrm{O}}_{3-\delta }$

Electrical control of magnetism is a long-standing goal in science and technology, with the potential to enable a next generation of low-power memory and logic devices. Recently developed electrolyte gating techniques provide a promising route to realization, although the ultimate limits on modulation of magnetic properties remain unknown. Here, guided by a recent theoretical prediction, we demonstrate large enhancement of electrostatic modulation of ferromagnetic order in ion-gel-gated ultrathin films of the perovskite $\mathrm{L}{\mathrm{a}}_{0.5}\mathrm{S}{\mathrm{r}}_{0.5}\mathrm{Co}{\mathrm{O}}_{3-\delta }$ by thickness tuning to the brink of percolation. Application of only 3–4 V is then shown capable of inducing a clear percolation transition from a short-range magnetically ordered insulator to a robust long-range ferromagnetic metal with perpendicular magnetic anisotropy. This realizes giant electrostatic Curie temperature modulation over a 150 K window, outstanding values for both complex oxides and electrolyte gating. In operando polarized neutron reflectometry confirms gate-controlled ferromagnetism, additionally demonstrating, surprisingly, that electrostatically induced magnetic order can penetrate substantially deeper than the Thomas-Fermi screening length.

Figure 1

(a) $\mathrm{L}{\mathrm{a}}_{1-x}\mathrm{S}{\mathrm{r}}_x\mathrm{Co}{\mathrm{O}}_{3-\delta }$ (LSCO) EDLT schematic. S/D represents source/drain, $V_{\mathrm{g}}/V_{\mathrm{SD}}$ the gate/source-drain voltages, red/blue charges the ion-gel cations/anions, and yellow/gray charges electrons/holes, respectively. The LSCO film has thickness $t$ and finite clusters are shown in green. (b) Solid curves (color coded for different theoretical thickness, $t$) show the theoretical surface charge density required to induce percolation, $\mathrm{\Delta }{s}_{\mathrm{c}}$ (left axis), vs bulk chemical doping, $x_{\mathrm{eff}}$. These are obtained from Ref. [41], by rescaling to the LSCO experimental percolation threshold, $x_{\mathrm{c},\mathrm{LSCO}}$ [43]. Data points (right axis, color coded to the experimental thickness, $t_{\mathrm{expt}}$) show the maximum experimental surface charge density achieved, $\mathrm{\Delta }{s}_{\mathrm{expt}}$. Determination of $\mathrm{\Delta }{s}_{\mathrm{expt}}$ and $x_{\mathrm{eff}}$ is discussed in the Supplemental Material, Sec. D [44]. The shaded region is discussed in the text. (c–g) Temperature $T$, dependence of resistivity ρ (log scales), for LSCO films with nominal $x=0.15$, 0.22, 0.5, 0.5, and 0.5 at $t_{\mathrm{expt}}=12,12,8,6$, and 5 unit cells, respectively, at $V_{\mathrm{g}}=0$ to $-4\mathrm{V}$.

Figure 2

(a–c) Temperature $T$, dependence of (a) zero magnetic field resistivity ρ, (b) 9 T out-of-plane magnetoresistance MR, and (c) low-field (out-of-plane field, $B_{\mathrm{OP}}=0.02\mathrm{T}$) transverse conductivity ${\sigma }_{xy}$, at gate bias, $V_{\mathrm{g}}=0$ to $-4\mathrm{V}$. (d–f) 5 K $B$ dependence of (d) out-of-plane MR, (e) in-plane MR (with current, $I\left|\right|B$), and (f) ${\sigma }_{xy}$, at $V_{\mathrm{g}}=0$ to $-4\mathrm{V}$. (g–i) $V_{\mathrm{g}}$ dependence of (g) ρ at 5 K, (h) out-of-plane MR at 5 K (left axis) and 170 K (right axis), and (i) ${\sigma }_{xy}$ at 5 K and $B_{\mathrm{OP}}=9\mathrm{T}$ (left axis), and the Curie temperature, $T_{\mathrm{C}}$ (right axis). All data are from the $x=0.5$, $t_{\mathrm{expt}}=6$ unit cell $\mathrm{L}{\mathrm{a}}_{1-x}\mathrm{S}{\mathrm{r}}_x\mathrm{Co}{\mathrm{O}}_{3-\delta }$ film in Fig. 1.

Figure 3

(a) Neutron reflectivity $R$, vs scattering wave vector magnitude $Q$, from an $x=0.5$, $t_{\mathrm{expt}}=6$ unit cell $\mathrm{L}{\mathrm{a}}_{1-x}\mathrm{S}{\mathrm{r}}_x\mathrm{Co}{\mathrm{O}}_{3-\delta }$ film at gate bias, $V_{\mathrm{g}}=-3\mathrm{V}$, 30 K, and 1 T (in plane). Black and red denote non-spin-flip “$R^{++}$” and “$R^{--}$” channels, for both data (points) and fits (solid lines). (b) Spin asymmetry vs $Q$ for $V_{\mathrm{g}}=0,-3\mathrm{V}$. Lines are fits with the depth profiles in (c). (c) Depth profiles of the nuclear scattering length density SLD (left axis) and magnetization $M$ (right axis) for $V_{\mathrm{g}}=0$, $-3\mathrm{V}$; the film/substrate interface is at $z=0$.