Supplementary Information for

Gate Control of Electronic Phases in a Quarter-Filled Manganite

T. Hatano1*, Y. Ogimoto1,2, N. Ogawa1, M. Nakano1,3, S. Ono1,4, Y. Tomioka5, K. Miyano6,7, Y. Iwasa1,8†, and Y. Tokura1,8

1RIKEN Center for Emergent Matter Science (CEMS), Wako 351-0198, Japan.

2 Fuji Electric Co. Ltd., Tokyo 191-8502, Japan.

3Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan,

4 Central Research Institute of Electric Power Industry, Komae 201-8511, Japan.

5 Electronics and Photonics Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8562, Japan.

6 Research Center for Advanced Science and Technology (RCAST), University of Tokyo, Tokyo 153-8904, Japan.

7 National Institute for Material Science (NIMS), Tsukuba 305-0047, Japan.

8 Quantum-Phase Electronics Center and Department of Applied Physics, University of Tokyo, Tokyo 113-8656, Japan.

*Correspondence to:

†Correspondence to:

The gate voltage dependence of the gate current and the estimation of accumulated carriers by electrolyte-gating

We used an ionic liquid of N,N-diethyl-N-(2-methoxyethyl)-N-methylammonium bis-trifluoromethylsulfonyl)-imide (DEME-TFSI) as a gate dielectric material.Figure. S1 displays the gate current as a function of gate voltage (VG) in the device recorded at 220 K. The rectangular shape in Fig.S1 indicates that the electrostatic mechanism of charge accumulation is dominant at -3 V < VG < 3 V.Here, it should be emphasized that the clear shift of metal-insulator transition temperature (TMI) accompanied by the temperature hysteresis (Fig.1c) indicates that the electric-field induced changes of transition temperatures seem to occur in the whole 5 nm-thick film. This class of field-induced metal-insulator (MI) transition beyond the standard screening length was recently reported in other electric double-layer transistors (EDLTs) based on strongly correlated electron systems16,17,20, and it was ascribed to the delocalization of localized carriers accompanied by the structural transition of materials20.Additionally, the present results demonstrated in PSMO-EDLT was the gate-induced conversion from conducting to insulating states for the negative VG, which provides additional evidence for the field-induced MI transitions of the whole films (see Fig. 2a).Using a specific capacitance value of anEDL with DEME-TFSI reported in the literature, CEDL = 10 F/cm2 (ref. 11), the charge carrier density accumulated in the PSMO channel was estimated to be 6.3  1013 /cm2/VG. This corresponds to 0.72 %/VG carriers per unit cell for the 5 nm-thick PSMO film, assuming that the carriers are distributed uniformly over the whole thickness of the film. In this estimation, the lattice parameters of the PSMO thin film grown on a (LaAlO3)0.3– (SrAl0.5Ta0.5O3)0.7 (110) substrate reported in ref. S1 was used to calculate the unit cell size. Thus the VGmodulation of ±50 mV wascalculated to be the x modulation of±0.0004(±0.04 % of Sr doping) in 5 nm-thick PSMO film.Note that there was no appreciable magnetic field effect on theVGdependence of the gate current [compare the black (0 T) and red line (9 T) marks in Fig. S1], indicating that the capacitance of the EDL formed at the interface between PSMO and DEME-TFSI is not affected by the magnetic field. This is in agreement with the naive expectation from the constituent non-magnetic elements of DEME-TFSI.

Fig. S1. Gate voltage (VG) dependence of the gate current (IG) for B = 0 T and 9 T recorded at 220 K.

Reversibility and reproducibility of PSMO-EDLT experiments

The EDLT operation of PSMO was very stable, when VGislimited within ±3 V. Figure S2 shows Rs-T curves of a PSMO-EDLT device before and after the gating experiments. The two curves show reasonable agreement with a shift ofTMI of 2.3 K, even though the red curve was taken after 10 days of continuous EDLT experiments.

Figure S3 displays the comparison of theRs-T curvesfortwo different PSMO-EDLT devices(D1 is the device for Figs. 1 to 3 in the main text).The TMIof D2 is slightly lower than that of D1. The origin of this difference in TMIis the very slight deviation of x, which is inevitable in the film growth. From the bulk phase diagram shown in Fig. 1a, this shift of TMIcorresponds tothe deviation of doping level of -0.2 % from x = 0.5.Figure S4 shows the temperature dependence of Rsfor device D2 with various VGs forB = 3.5 T and its three dimensional plot on the VG-T plane. Fair agreement between Fig. S4 and Figs. 2a, 2b shows good reproducibility of the ambipolar behavior and the robust insulating phase at around x = 0.5.

Fig. S2. Temperature (T) dependence of the sheet resistance (Rs) for the PSMO-EDLT device before (black line) and after (red line) gating experiments over 10 days.

Fig. S3. Temperature (T) dependences of the sheet resistance (Rs) for two different PSMO-EDLT devices.

Fig. S4.VG variation of Rs-T curves for the device D2 with varying VG from -3 V to 3 V under the magnetic field of B = 3.5 T.

Thickness dependence of PSMO-EDLT experiments

Figure S5 shows the resistivity - temperature (-T) curves of PSMO-EDLTs with different thicknesses(t) for VG = 0 and ± 3 V. In the device D1 (t = 5 nm, the same device inthe main text), TMIwas shifted by 40 K with varyingVGfrom - 3 V to +3 V (Fig. S5a). On the other hand, almost no gating effect was observed in the thicker deviceD3 (t = 10 nm), as shown in Fig. S5b. However, we observed peculiar gate responses as well as in D3 under the magnetic field.Figure S6 displays the VG variation of -T curves from VG = -3 V to 0 V atB = 4 T. AtVG = 0 V, D3 was almost metallic because the ferromagnetic state was stabilized by the magnetic field.By applying VG = -2.5 V,a strongly insulating state emerged. This behavior agrees fairly well with those of D1 and D2. The gate-induced insulating state indicates that the entire film was converted from the conducting to insulating state.This class of MI transition in the thicker films was recently reported in other EDLTs based on strongly correlated electron systems (ref. 16, 17, 20). Among them, one of the most recent papers on VO2-EDLT is suggesting a collective transition by the electrostatic modulation (ref. 20). We suppose that the similar situation was realized in our system, irrespective of the thickness. However, the mechanism of EDLT in strongly correlated electron systems is still in debate. Indeed, an alternative mechanism for VO2-EDLT is also presented (ref. S2). Further discussions and experiments including different material systems are left to be made in future research.

Fig. S5.The thickness (t) variation of the –T curvesat VG = 0 and ±3 V for a: t = 5 nm (D1) and b: t = 10 nm (D3).

Fig. S6.VG variation of -T curves for D3 (t = 10 nm) with varying VG from -3.5 V to 0 V under the magnetic field of 4 T.

Reference in SupplementaryInformation

S1. Wakabayashi,Y. et al. Orbital ordering structures in (Nd,Pr)0.5Sr0.5MnO3 manganite thin films on perovskite (011) substrates. J. Phys. Soc. Jpn.77, 014712 (2008).

S2. Jeong, J., Aetukuri, N., Graf, T., Schladt, T. D., Samant, M. G. Parkin, S. S. P.Suppression of Metal-Insulator Transitionin VO2 by Electric Field–InducedOxygen Vacancy Formation.Science, 339, 1402 (2013).

1