Engineering of Co/MgO interface with combination of ultrathin heavy metal insertion and post-oxidation for voltage-controlled magnetic anisotropy effect

Recently, spin manipulation making use of the interface with broken inversion symmetry is attracting much attention in the rapidly evolving field of spin-orbitronics.¹ Since it enables spin manipulation with low power consumption, the use of the voltage-controlled magnetic anisotropy (VCMA) effect at the interface of a ferromagnet/non-magnetic insulator is a promising technology toward future spintronic devices.^2–4 The VCMA effect is an interfacial phenomenon that drives the change of the magnetic anisotropy of the ferromagnet by applying an external voltage through the adjacent insulator (see Fig. 1). Moreover, the purely electronic VCMA effect, free from chemical reactions and the displacement of ions,⁵ enables spin manipulation with high-speed response and high writing endurance,^6,7 which can be of considerable benefit for practical spintronic devices. For the mechanism of the purely electronic VCMA effect at ferromagnet/insulator junctions, contributions such as the voltage-modulation of the band structure, the electron occupation state,^8–10 the Rashba spin–orbit anisotropy,¹¹ and the magnetic dipole moment¹² have been proposed.

To operate voltage-controlled spintronics devices based on the VCMA effect with lower power consumption, materials exhibiting a large VCMA coefficient are desirable. The VCMA coefficient is defined by ΔK_PMAt/E, where ΔK_PMA is the change of the effective perpendicular magnetic anisotropy (PMA) energy induced by the applied electric field, t is the thickness of the ferromagnetic layer, and E is the applied electric field given by V_bias/t_barrier. Here, V_bias and t_barrier are the applied bias voltage and the thickness of the tunneling barrier, respectively. A positive bias voltage induces electron accumulation at the ferromagnetic layer/insulator layer interfaces, as shown in Fig. 1. A negative (positive) VCMA coefficient describes a reduction (enhancement) of the PMA by applying the positive bias voltage. Therefore, at the ferromagnetic layer/insulator layer interfaces with negative (positive) VCMA effect, the PMA decreases (increases) due to electron accumulation.

Since the VCMA effect is an interfacial phenomenon, it can be strongly tuned by interfacial engineering. One method involves the insertion of a sub-atomic layer of heavy metals at the interface between the ferromagnet and an insulator. In a previous study, the effect of inserting a layer of 4d and 5d transition metals at Fe/MgO junction has been theoretically calculated via the first-principles calculations.¹³ The results suggest that the magnitude of the spin–orbit coupling of the inserted materials and the position of the Fermi level with respect to the d-band level can be important factors to tune the VCMA effect. Thus, 5d transition metals, such as Pt and Ir, are promising materials for the insertion layer because they exhibit large spin–orbit interactions and magnetic proximity effects. This concept has been demonstrated in epitaxially grown Fe/MgO based junctions with Ir doping.^14,15 In our recent experimental studies on interface engineered Co/MgO junctions, we have demonstrated a strong enhancement of the negative VCMA effect by inserting a sub-atomic layer of Pt.^16,17

Another method to tune the VCMA effect via interfacial engineering is to control the oxidation state at the interface between the ferromagnetic material and the insulator. Although it is known that the sign of the VCMA effect is negative in most materials, a positive VCMA effect can be observed by exploiting an interface between surface oxidized Co and high-k materials, such as HfO₂¹⁸ and TiO_x.^19,20 Interestingly, a large positive VCMA coefficient up to +230 fJ V⁻¹ m⁻¹ can be obtained for Co/CoO_x/HfO₂ junctions. A positive VCMA effect has also been detected for ultrathin-Co/MgO/HfO₂²¹ and Ru/CoFeB/MgO²² structures with MgO barriers deposited by sputtering. However, for the case of a sputter-deposited MgO barrier, it has been pointed out that the sign reversal of the VCMA effect can occur due to a slight change in the oxidation state at the junction interface.²³ By controlling the oxidation state at the interface by inserting the light metal MgAl, a large VCMA coefficient and a large PMA have been demonstrated in epitaxially grown single crystalline Fe/MgO based structures deposited by molecular beam epitaxy.²⁴ Although a lot of efforts are made to elucidate the physical mechanism of the VCMA effect so far, the effect of interfacial engineering, including both interfacial oxidation and the insertion of sub-atomic layer of heavy metals, remains to be thoroughly investigated.

In this study, we report on the impact of dual interfacial engineering utilizing heavy metal insertion and post-oxidation process on the PMA and the VCMA effect of Co/MgO-based junctions. We used Pt and Re for the materials of insertion layers since Pt (Re) shows electron doping (depletion) for the adjacent Co layer, and thus systematic changes of the VCMA effect due to the insertion of this layer may be expected. We found that the PMA of the Co/MgO junction is enhanced (reduced) by conducting an oxidation process for the films with a sub-atomic layer of Pt (Re) inserted, which is quite different from the case of simply inserting heavy metal elements.^16,17 Moreover, the signs of the additional VCMA coefficient due to the insertion of the sub-atomic layer of heavy metals are reversed by conducting an oxidation process after heavy metal insertion, interestingly. These results suggest the crucial role of heavy metals in the magnetic anisotropy for both unoxidized Co/MgO junctions and surface oxidized Co/MgO junctions. These findings will provide crucial information toward further development of the interfacial spin–orbit engineering for control of the VCMA effect.

The sample system used for the VCMA measurements consists of a Ta(5 nm)/Ru(10 nm)/Ta(5 nm)/Pt(2 nm)/Ru(4 nm)/Co(0.7–1.5 nm)/X (X = Pt or Re)(0–0.28 nm)/oxidation process/MgO(3 nm)/ITO(20 nm) polycrystalline film deposited on a thermally oxidized silicon substrate, where the order of the layers is described from bottom to top. For the control measurements, we also prepared multilayer films without X layers and/or an oxidation process. The deposition of the multilayer film was performed at room temperature by a combination of sputtering and molecular beam epitaxy apparatus without breaking the vacuum. Most layers were deposited by magnetron sputtering, and only the MgO layers were formed by electron-beam evaporation. To evaluate the PMA and the VCMA effect by using a polar-MOKE apparatus with a perpendicular magnetic field, we utilized Ru as underlayers. Although a large change in the VCMA coefficient by inserting an X layer has been reported for Co/X/MgO junctions with Os underlayer,¹⁷ Os is known to be toxic when oxidized. Therefore, in the present study, we used Ru as the alternative of Os because Ru is expected to show similar structural and electrical properties as Os. The layer of heavy metal X and the oxidation process were introduced to change the magnitude of the PMA and the VCMA effect. The thicknesses of Co, t_Co, and X, t_X, were varied by using a linear shutter. The oxidation process in an oxygen atmosphere was conducted using pure oxygen gas at a pressure of 0.5 Pa for 3 min in an oxidation chamber directly connected to the sputtering apparatus. For the top electrode, transparent ITO (indium tin oxide) was used for evaluating the magnetic properties of the ultrathin Co layer using polar-MOKE. To prevent heavy metals from diffusing into the Co layer, no annealing was done for any of the multilayer films. The VCMA devices were fabricated by conventional optical lithography, ion-milling, and lift-off processes. The cross-sectional area of the tunnel junction was 8 × 10 µm².

The magnetic properties of the multilayer films were evaluated by a combination of VSM and polar-MOKE. For the VSM measurements, multilayer films of Ta(5 nm)/Ru(10 nm)/Ta(5 nm)/Pt(2 nm)/Ru(4 nm)/Co(0.94–1.58 nm)/MgO(3 nm)/ITO(20), Ta(5 nm)/Ru(10 nm)/Ta(5 nm)/Pt(2 nm)/Ru(4 nm)/Co(1.1–1.74 nm)/oxidation process/MgO(3 nm)/ITO(20), and Ta(5 nm)/Ru(10 nm)/Ta(5 nm)/Pt(2 nm)/Ru(4 nm)/Co(1.26–1.58 nm)/X (X = Pt or Re)(0.03 nm)/oxidation process/MgO(3 nm)/ITO(20) were used. For the polar-MOKE measurements, a semiconductor laser with a spot size of 1.5 µm was utilized, which was much smaller than the cross-sectional area of the tunnel junctions. The change in the Kerr rotation angle was measured by applying an external magnetic field perpendicular to the film plane. For the VCMA measurements, polar-MOKE apparatus equipped with a micro-probe system was utilized. All the measurements were conducted at room temperature.

To investigate the effect of the oxidation process on the values of the saturation magnetization M_S and the thickness of the magnetic dead layer t_dead of the Co layer, we first carried out vibrating sample magnetometry (VSM) measurements. Here, the oxidation process was performed using pure oxygen gas at a pressure of 0.5 Pa for 3 min for Co surfaces [also see Fig. 2(a) and experimental details]. In Fig. 2(b), the saturation magnetic moment per unit area for the Co/MgO films with and without interfacial engineering due to the oxidation process is plotted as a function of the nominal Co layer thickness t_Co. The results show that the saturation magnetic moment per unit area increases linearly with increasing t_Co for both with and without the oxidation process. Here, the linear fit to the data shown by the dashed line gives the values of M_S and t_dead, i.e., the values obtained from the slope and the intercept with the horizontal axis are M_S = 1417 kAm⁻¹ and t_dead = 0.12 nm for the films without the oxidation process. The magnitude of M_S obtained for films without the oxidation process is close to the value of the similar Co/MgO films grown on Pt or Os underlayers reported in previous studies,^16,17 indicating a similar quality of the Co/MgO films on the Ru underlayer. The obtained t_dead is also almost consistent with the Co/MgO films with an Os underlayer. On the other hand, the magnitudes of M_S and t_dead obtained for the films with the oxidation process are M_S = 1344 kAm⁻¹ and t_dead = 0.66 nm, respectively. These values are closer to those reported for the films with an oxidation process using TiO_x layers.¹⁹ The increase in the dead layer due to the oxidation process, which corresponds to the thickness of 0.54 nm, can mainly be due to the oxidation of the Co layer. Under similar oxidation conditions, it was demonstrated that CoO_x is mainly dominated by CoO in previous studies.^25,26 Therefore, here, we assumed CoO_x as CoO. Considering the lattice parameters of Co and CoO as 2.51 and 4.23 Å, respectively,¹⁹ the increase in the insulator layer thickness due to the oxidation can be calculated as 0.91 nm. Thus, we evaluated the PMA and the VCMA effect by considering the increases in the dead layer and the insulator layer thicknesses due to the oxidation process. Compared to the unoxidized films, the slight decrease with M_S may be due to the non-uniformity of the oxidation level. It should be noted here that only a small difference was observed in the saturation magnetic moment per unit area in the multilayer films after inserting a sub-atomic layer of heavy metal X (=Pt or Re), followed by the oxidation process [see Fig. 2(b)], suggesting that the oxidation state of Co cannot be controlled by inserting a sub-atomic layer of X. Here, the orange and the green plots are the data with insertion of X = Pt and Re, respectively. The thickness of X, t_X, was fixed at 0.03 nm to evaluate the PMA and the VCMA effect using a polar magneto-optical Kerr effect (polar-MOKE) apparatus with a perpendicular magnetic field, which will be discussed later. In addition, for samples without the oxidation process, it is known that almost no difference was observed in the saturation magnetic moment per unit area in the multilayer films with inserting a sub-atomic layer of heavy metals.¹⁶ Thus, hereafter, the values of M_S and t_dead for Co/MgO given above for each condition are used to determine the values of the PMA and the VCMA coefficient in the multilayer films with the insertion of heavy metal layers.

Next, we conducted polar-MOKE measurements to quantitatively evaluate the PMA for both films with and without the oxidation process. To clarify the effect of insertion of a heavy metal X layer and the surface oxidation, we measured the X thickness t_X dependence of the polar-MOKE hysteresis curves. In Figs. 3(a)–3(d), the perpendicular magnetic field H_perp dependence of the Kerr rotation angle for different X and thicknesses t_X are shown, where the effective thickness of the Co layer t_Co* (=t_Co−t_dead) is fixed at 0.8 nm. Due to the multilayer films with an in-plane magnetic easy axis, a gradual switching is obtained for Co/MgO junctions, where X is not inserted. The magnitude of the saturation magnetic field decreases with increasing X thickness, reflecting an enhancement of the PMA at the Co/MgO interface. For the films with the insertion of Re without the oxidation process [see Fig. 3(b)], the saturation magnetic field slightly increases with increasing X thickness when t_X > 0.08 nm, which may be caused by the deterioration of film quality caused by the sputtering of heavy metal Re. Therefore, we only utilized the devices with t_X < 0.08 nm for the VCMA measurements. For the case with the insertion of the Pt layer with the oxidation process [see Fig. 3(c)], a sharp switching due to the out-of-plane magnetic easy axis is obtained when t_Pt is only 0.08 nm, indicating a strong enhancement of the PMA.

Here, we show the effect of inserting a heavy metal layer on the VCMA effect. Focusing on films without an oxidation process, the typical polar-MOKE hysteresis curves of Co/MgO junctions without heavy metal layers inserted are shown in Fig. 5(a), and the case with heavy metal layers inserted are shown in Figs. 5(b) and 5(c), where the effective thickness of the Co layer is fixed at 0.8 nm. To show the differences in the effect of the insertion layer, here the thickness of the inserted Pt and Re layers are set to 0.03 and 0.01 nm, respectively. To measure the VCMA effect, we applied a DC bias voltage through the MgO barriers for these multilayer devices. In Figs. 5(a)–5(c), the red and blue data were measured under the application of V_bias = +0.8 and −0.8 V, respectively. The inset shows a magnified view of the main curves and reveals that clear shifts in the saturation field due to the V_bias are obtained in all the cases, indicating the occurrence of a VCMA effect. Here, the K_PMA of the ferromagnetic layer can be determined in the same manner as in the case of Figs. 4(a)–4(d) by using Eq. (1). Since the change of the Kerr rotation angle due to the bias voltage was negligibly small when enough large H_perp was applied, we assumed that the M_S is not changed by applying bias voltage. The magnitude of the K_PMAt_Co* values obtained for each bias voltage condition are shown in Figs. 5(d)–5(f), respectively. Here, linear changes in K_PMAt_Co* are obtained for all the samples, and the magnitude of the slope corresponds to the VCMA coefficient. Considering the thickness of the insulator layer [MgO(3 nm)], the VCMA coefficient of Co/MgO junctions without the insertion layer and with Pt and Re insertion layers are determined as −37, −80, and −23 fJ V⁻¹ m⁻¹, respectively. These results indicate that the sign of the additional VCMA effect induced by inserting X is negative (positive) for Pt (Re), which is the same as the case of inserting Pt (Os) for the Co/MgO junctions deposited on Pt and Os underlayers.^16,17 Since Pt (Re and Os) exhibits electron doping (depletion) for the adjacent Co layer, charge transfer between X and Co layers may affect the magnitude of the VCMA effect. The values of the additional VCMA coefficient induced by X insertion Δξ are maintained over the range of the inserted layer used in this study [see Fig. 7(a)]. Since the sign of the VCMA effect in most materials is negative, here the negative values are placed at the top of the vertical axis shown in Fig. 7.

Now, let us show the impact of the oxidation process in addition to the insertion of a sub-atomic layer of heavy metals. In a similar process as the case of the films without the oxidation process, the typical polar-MOKE hysteresis curves of Co/MgO junctions without heavy metal layers inserted are shown in Fig. 6(a), and the case with heavy metal layers inserted are shown in Figs. 6(b) and 6(c), where the effective thickness of the Co layers is fixed at 0.8 nm. Here, the thickness of the inserted X layers is set to 0.02 nm to show the differences in the effect of the insertion layers. The magnitude of the K_PMAt_Co* obtained for each V_bias are shown in Figs. 6(d)–6(f), respectively. Here, linear changes in K_PMAt_Co* are again obtained for all the samples. However, a positive VCMA effect is obtained for the films without X insertion and with Pt insertion, as shown in Figs. 6(d) and 6(e). These results are quite different from the case of the films with unoxidized Co/MgO junctions, suggesting the impact of surface oxidation of Co on the VCMA effect. Since the effective thicknesses of the Co layers t_Co* in the films with and without oxidation shown in Figs. 5 and 6 are the same, we cannot explain the sign reversal simply by a decrease in the effective thickness due to the oxidation process. By considering the thickness of the insulator layers [CoO_x(0.91 nm)/MgO(3 nm)], the VCMA coefficient of Co/MgO junctions without the insertion layer and with Pt and Re insertion layers are determined as +3, +13, and −8 fJ V⁻¹ m⁻¹, respectively. Interestingly, the signs of the additional VCMA coefficient induced by X insertion are reversed by conducting the oxidation process (see Fig. 7). Although further research is necessary, these results may imply the existence of a mechanism such as the voltage-induced atomic structural change²⁸ for the additional VCMA effect induced by X insertion in Co/MgO-based junctions with surface oxidation.

Here, we conducted control experiments to further explore the origin of the modulation of the VCMA effect via dual interfacial engineering. As a control measurement, first, we changed the oxidation conditions. While the oxidation level was fixed to one condition in the above-mentioned experiments, we also investigated the VCMA effect using a similar sample structure but using a different oxidation condition [also see Fig. S3(c) in Sec. S1, supplementary material]. By utilizing films with a weaker oxidation level, a negative VCMA effect is observed without the Pt insertion layer. On the other hand, the sign of the additional VCMA coefficient Δξ due to the insertion of a sub-atomic layer of Pt is positive, which is similar to the results shown in Fig. 7(b). These results suggest that the oxidation level strongly influences the magnitude of the VCMA effect, while the magnitude of the additional VCMA coefficient due to the insertion of X, Δξ, is maintained relatively constant even when the oxidation level changes. In another control experiment, we changed the order of the heavy metal insertion and the interfacial oxidation. Although the oxidation process was performed after the insertion of heavy metal layers in the above-mentioned experiments, we also investigated the VCMA effect using a film in which the heavy metal insertion was conducted after interfacial oxidation of the Co layer [also see Fig. S4(d) in Sec. S2, supplementary material]. In this case, almost no change is observed in the additional VCMA coefficient due to the insertion of a sub-atomic layer of Pt. Here, a change of the PMA energy due to the insertion of the Pt layer is not observed as well [see Fig. S4(c) in Sec. S2, supplementary material]. Therefore, a direct contact or alloying between the ferromagnetic metal Co and the heavy metal may be necessary to achieve an enhancement of the PMA and a sign reversal for the additional VCMA coefficient by using the dual interfacial engineering.

Finally, we discuss the prospect of dual interfacial engineering using the insertion of a sub-atomic layer of heavy metals and an oxidation process for voltage-controlled phenomena in terms of the technical aspects. Although we have only utilized Ru as the underlayer and sub-atomic Pt and Re as the insertion layer in this study, the influence of underlayer materials on the VCMA effect at interface engineered Co/MgO junctions with heavy metals has been reported elsewhere.¹⁷ The magnitude of the modulated PMA and VCMA via the dual interfacial engineering method might be further enhanced by using other materials as the underlayer and/or the insertion layer and by using different thickness insertion layer to control the oxidation state of Co. Therefore, further research is still necessary to clarify the correlation between the PMA, VCMA effect, interfacial oxidation state, and underlayer materials. While we have focused on experiments at room temperature in this study, CoO_x layers formed by naturally oxidized Co are known to exhibit anti-ferromagnetism below 200 K, and thus, an exchange bias can be observed for Co/CoO_x junctions at lower temperature.²⁹ By exploiting the dual interfacial engineering reported in this study, a modulation of the Néel temperature, the magnitude of the exchange bias, and their voltage-controlled efficiencies may be expected. Recent studies revealed that the exchange bias can be modulated by the spin–orbit torque.^30–32 Therefore, more efficient spin manipulation may be expected by utilizing dual interfacial engineering, exchange bias, and the combination of spin–orbit torque and voltage-controlled magnetic properties.

In summary, we have experimentally demonstrated dual interfacial engineering exploiting the insertion of sub-atomic 5d transition metals (Pt or Re) and post-oxidation process on the PMA and the VCMA effect of Co/MgO-based junctions. We found that the PMA of the Co/MgO junction is strongly enhanced by introducing an ultrathin heavy metal Pt layer, followed by a post-oxidation process, which is quite different from the case of merely inserting heavy metal Pt without an oxidation process.^16,17 Furthermore, the additional VCMA coefficient due to the insertion of the sub-atomic layer of heavy metals intriguingly shows sign reversal when introducing an oxidation process after the heavy metal insertion. Thus, our findings will offer another pathway for designing voltage-controlled spin-orbitronic devices based on the VCMA effect.

See the supplementary material for the magnetic properties and VCMA effect of multilayer films with post-oxidation in air atmosphere, and magnetic properties and VCMA effect of multilayer films with X insertion after oxidation process.

The authors thank T. Yamamoto, T. Ichinose, H. Onoda, and Y. Hibino of the National Institute of Advanced Industrial Science and Technology; H. Sukegawa and S. Mitani of the National Institute for Materials Science; and Y. Kageyama, L. Sakai, K. Hiraga, K. Ohba, Y. Higo, and M. Hosomi of Sony Semiconductor Solutions Corporation for their fruitful discussions. This work was partially supported by the Japan Science and Technology Agency (JST), PRESTO Grant No. JPMJPR23H6, Japan and the New Energy and Industrial Technology Development Organization (NEDO) Grant No. JPNP16007, Japan.

The authors have no conflicts to disclose.

Hiroyasu Nakayama: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Funding acquisition (equal); Investigation (lead); Methodology (lead); Writing – original draft (lead); Writing – review & editing (lead). Tomohiro Nozaki: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Writing – review & editing (supporting). Takayuki Nozaki: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Writing – review & editing (supporting). Shinji Yuasa: Funding acquisition (lead); Investigation (supporting); Project administration (equal); Writing – review & editing (supporting).

The data that support the finding of this study are available from the corresponding author upon reasonable request.