Coexistence of Large Voltage Controlled Magnetic Anisotropy, Large Surface Anisotropy, and Large TMR by a new MTJ structure having MgO/CoFeB/Ir/CoFeB

Abstract

In recent years, writing data in magnetic random access memory (MRAM) utilizing voltage controlled magnetic anisotropy (VCMA) has attracted much attention for its potential low power consumption [1]. We proposed voltage-control spintronics memory (VoCSM) which had high-efficient and deterministic writing properties [2]. In order to realize those memories, three features of a large VCMA, a large surface anisotropy Ks, and a large tunneling magnetoresistance (TMR) should coexist. In addition, a large spin-Hall angle is a must for VoCSM. Many challenges based on MgO tunneling barrier/ferromagnetic layer (FL) such as CoFeB thin films combined with various materials as an insertion layer at the MgO/FL interface or as an underlayer of FL showed improved VCMA but were concerned to fail in the coexistence of the feature because of very thin storage-layer or degraded lattice growth between MgO and CoFeB [3]–[5]. As a result, none of them have had a practical meaning as a memory cell so far. In this study, the experiments were conducted in which the insertion position of Ir was changed in MgO/CoFeB/Ta thin films. Each of the interface layer, the interlayer and the underlayer of Ir showed an increase in VCMA, and the largest VCMA was obtained in the case of inserting the Ir interlayer into the CoFeB layer. In addition, both the resistance-area product (RA) and TMR ratio decreased greatly when using the Ir interface layer, but clearly improved by employing the Ir interlayer. The base multilayer structure for VCMA measurement was Ta (5 nm)/MgO $(\sim 3$ nm)/CoFeB (1–2 nm)/Ta (5–8 nm), which was deposited on a thermally oxidized Si substrate. The CoFeB layer was set to in-plane magnetization, and the base stack of IrMn/ CoFe/Ru/CoFeB /MgO/CoFeB/Ta with a reference layer was prepared for RA and TMR measurement by using current in-plane tunneling (CIPT). The multilayers for VCMA were patterned and etched into the device size with one side of 3 to $50 \mu \mathrm {m}$ and their hysteresis curves were measured using the magneto-optical polar Kerr effect. The effective perpendicular magnetic anisotropy field Hk $_{eff}$ of the CoFeB layer was measured while bias voltage was applied to the device, and the variation of Ks depending on the electric field E was evaluated as the VCMA coefficient. Figure 1 shows the VCMA coefficients (–dKs/dE) of the MgO/CoFeB/Ta thin films as the “Base” sample, “Interface” sample in which Ir (0.2 or 0.3 nm) is layered at the MgO/ CoFeB interface, “Interlayer” sample in which Ir (0.3 nm) is inserted in the middle of the CoFeB layer, and “Underlayer” sample in which Ir (0.5 nm) is formed between the CoFeB and the Ta layer. All coefficients of the “Interface”, “Interlayer”, and “Underlayer” samples increased more than that of the “Base” sample in terms of each average value, although each coefficient had a certain degree of dispersion. The Ks in the “Interface” sample also increased more than in the “Base” sample at each average value, however, the largest Ks (maximum of 2.2 erg/cm $^{2})$ and VCMA (maximum of 190 fJ/ Vm) were obtained in the “Interlayer” sample. The relationship between RA and TMR ratio in the MTJ samples similar to Fig. 1 with the reference layer is plotted in Fig. 2. Both RA and the TMR ratio in the “Underlayer” sample were almost the same as those in the “Base” sample, but both decreased in the “Interface” sample and further decreased by increasing the Ir layer thickness from 0.2 to 0.3 nm. In the “Interlayer” sample, the deterioration of RA was not observed, and although the TMR ratio decreased, it still showed a high value of more than 120%. By comparison at the Ir thickness of 0.3 nm, it can be seen that both RA and TMR are clearly improved by changing from the Ir interface layer to the Ir interlayer. In summary, we successfully found the practical MTJ structure as a memory cell which realized coexistence of a large VCMA, a large Ks, and a large TMR for the first time. The structure is expected to have a large spin-Hall effect as well. This work was partly supported by the ImPACT Program of the Council for Science, Technology and Innovation (Cabinet Office, Government of Japan).