A spin metal–oxide–semiconductor field-effect transistor (spin-MOSFET) is highly required for innovative, ultra-fast, nonvolatile memory, and logic devices in the next generation of technology. A key focus in developing spin-MOSFETs is achieving high magnetoresistance (MR) ratios. In Si-based spin-MOSFETs, which have currently been the most actively studied, the MR ratios are limited to ∼1%. Recently, spin-MOSFETs based on perovskite oxide (La2/3, Sr1/3)MnO3 (LSMO) demonstrated large MR ratios of up to ∼140% at 3 K. The channel of the device was made using Ar irradiation to induce a metal–semiconductor transition in LSMO. In this method, however, the channel was formed in a dead layer, which was generated in LSMO near the interface with the SrTiO3 (STO) substrate and is magnetically inactive. This issue makes the full use of the potential of LSMO difficult. Meanwhile, the insertion of a LaMnO3 (LMO) buffer layer between LSMO and STO is known to suppress the formation of the dead layer in LSMO. In this study, we make an LSMO-based spin-MOSFET using an LMO buffer layer on STO. Although the channel length defined by lithography is approximately three times longer than that in the previous study, we observe a large MR ratio of up to ∼10%, which can be attributed to the suppression of the dead layer in LSMO. Furthermore, we demonstrate magnetic anisotropy modulation in LSMO driven by the shape magnetic anisotropy from the device structure. Our results suggest that LMO is a promising buffer-layer material capable of enriching the functionality of LSMO-based spin devices.
Nakamura et al. (Sun,) studied this question.