Magnetic semiconductors (MSs) are promising materials for spintronic applications, which would be useful for non-volatile and low-power-consumption electronic devices^[1−5]. The ideal MSs are expected to at least satisfy the following requirements: first, their Curie temperature should be higher than room temperature (300 K); and second, both p-type and n-type MSs could be realized. Mn-based Ⅲ−Ⅴ MSs have been intensively investigated, such as p-type (In,Mn)As^[6−10] and (Ga,Mn)As^[11−15], while the maximum T_C are still much lower than room temperature (90 K in (In,Mn)As^[10], 200 K in (Ga,Mn)As^[13]). However, the effort made by previous research could not realize a MS satisfying these two requirements, until new results were found in Fe-doped Ⅲ−Ⅴ MSs. Both n- and p-type MSs could be achieved by doping Fe into Ⅲ−Ⅴ host semiconductors, such as p-type (Ga,Fe)Sb (with maximum T_C = 400 K)^[16−20] and n-type (In,Fe)Sb (with maximum T_C = 385 K)^[21−23]. Moreover, plenty of recent research has focused on the microscopic origin of the ferromagnetism of (Ga,Fe)Sb, revealing the intrinsic ferromagnetism of (Ga,Fe)Sb films ^{[19, 20, 24−26]}.

The magnetic anisotropy of (Ga,Fe)Sb (7.6 × 10³ erg/cm³ at x_Fe = 29.8%) is much smaller than the typical value of (Ga,Mn)As (3.6 × 10⁴ erg/cm³)^[27], which makes it difficult to be utilized for practical applications. In this work, we co-doped Ni with Fe into GaSb host semiconductor and achieved enhanced magnetic anisotropy of Ga_1-x-yFe_xNi_ySb films than that of (Ga,Fe)Sb. As a result, the K_u of Ga_1-x-yFe_xNi_ySb is negligible at y = 1.7% but increases to 3.8 × 10⁵ erg/cm³ at y = 6.1%, while the Curie temperature of samples was maintained above 300 K. Additionally, the hole mobility of Ga_1-x-yFe_xNi_ySb films reaches at 31.3 cm²/(V∙s) at x = 23.7%, y = 1.7%, which is much higher than the mobility of Ga_1-xFe_xSb at x = 25.2% (µ = 6.2 cm²/(V∙s)).

The Ga_1-x-yFe_xNi_ySb films were grown on semi-insulating GaAs(001) substrates by low temperature molecular beam epitaxy (LT-MBE), a schematic sample structure is shown in Fig. 1(a). At first, we grew a 150-nm-thick GaAs buffer layer to obtain a smooth GaAs surface at 550 °C. Then, a 5-nm-thick AlSb was deposited at 470 °C as a seed layer, following with a 100-nm-thick Al_0.9Ga_0.1Sb buffer layer grown at 470 °C to relax the strain induced by the lattice mismatch between Ga_1-x-yFe_xNi_ySb and GaAs. Compared with AlSb, Al_0.9Ga_0.1Sb buffer achieves a smoother surface, while the buffer layer keeps insulating due to the low concentration of Ga. After that, a Ga_1-x-yFe_xNi_ySb layer of 100 nm with almost the same Fe concentration x and different Ni concentration y was grown at the growth rate of 0.4 μm/h at 250 °C. Finally, a 2-nm-thick GaSb cap layer was grown at 250 °C to prevent oxidation of underlying Ga_1-x-yFe_xNi_ySb layer. We grew two series of Ga_1-x-yFe_xNi_ySb samples, A1−A4 and B1−B4, and the sample information is listed in Table 1. We changed x of samples A1−A4 (x = 25.2%−29.8%) and kept y = 0, to investigate the magnetic anisotropy and hole mobility of (Ga,Fe)Sb. Samples B1−B4 with nearly constant x (x ≈ 24%) and changed y (y = 1.7%−6.1%) were used to study the y dependence of the magnetic and electronic properties of Ga_1-x-yFe_xNi_ySb samples. In order to manifest the evolution of K_u and µ of Ga_1-x-yFe_xNi_ySb, the x of samples A1−A4 was roughly equal to the sum of x (B1−B4) and y (B1−B4), respectively.

The reflection high-energy electron diffraction (RHEED) was used to observe the surface morphology and crystallinity of the samples. The lattice constant of Ga_1-x-yFe_xNi_ySb layers was investigated by X-ray diffraction (XRD). Scanning transmission electron microscopy (STEM) and selective area electron diffraction (SAED) were employed to characterize the crystal structure and impurity concentration of the Ga_1-x-yFe_xNi_ySb layers. The magnetic and electronic properties of Ga_1-x-yFe_xNi_ySb were separately characterized by superconducting quantum interference device (SQUID) magnetometer and physical property measurement system (PPMS).

The RHEED pattern of an undoped GaSb film grown with the same condition as the Ga_1-x-yFe_xNi_ySb samples is displayed in Fig. 1(b). Figs. 1(c)−1(f) present the RHEED patterns of samples B1−B4 taken along the [11¯0] axis after the growth of Ga_1-x-yFe_xNi_ySb layers. It is obvious that the RHEED patterns of Ga_1-x-yFe_xNi_ySb layers are streaky with surface reconstruction of 1 × 3, which is similar with that of GaSb. The results imply that Ga_1-x-yFe_xNi_ySb layers grown by LT-MBE keep the zinc-blende crystal structure.

XRD spectra of samples A1−A4 are shown in Fig. 2(a), while that of samples B1−B4 are shown in Fig. 2(b), which are both performed using the Cu-K_α radiation (wave length λ = 0.15406 nm). No phases other than Al_0.9Ga_0.1Sb and Ga_1-x-yFe_xNi_ySb can be observed in the XRD spectra, and in particular there are no Fe−Sb, Ni−Sb intermetallic compounds or metal clusters. The concentration of Fe and Ni is preliminary determined by a series of energy dispersive spectra, which are measured repeatedly at different positions in Ga_1-x-yFe_xNi_ySb layer. According to the dependence between the Fe (Ni) doping concentration and Fe/Ga (Ni/Fe) flux ratio, the x and y could be further confirmed, as listed in Table 1. The lattice constant of (Ga,Fe)Sb and Ga_1-x-yFe_xNi_ySb could be given by a₁ = (1−x)a_GaSb + xa_FeSb and a₂ = (0.76−y)a_GaSb + 0.24a_FeSb + ya_NiSb. Here, a_GaSb, a_FeSb and a_NiSb are the lattice constant of GaSb, hypothetical zinc-blende FeSb and NiSb, respectively, which are estimated to be 0.60648 nm (a_GaSb), 0.55027 nm (a_FeSb) and 0.58233 nm (a_NiSb) as plotted in Figs. 2(c) and 2(d). The results are consistent with previous research^[16].

The crystal structure and impurity concentration of samples A1−A4 and B1−B4 were characterized by STEM and SAED. Fig. 3(a) shows the STEM lattice image of sample B4 taken along the [110] axis, showing the clear interface between Ga_1-x-yFe_xNi_ySb layer and buffer layer. Through the enlarged picture area marked by red rectangle in Fig. 3(a), the crystal structure of Ga_1-x-yFe_xNi_ySb layer is displayed more clearly in Fig. 3(c). Fig. 3(b) shows the SAED pattern of Ga_1-x-yFe_xNi_ySb layer of representative sample B4 (x = 24.3%, y = 6.1%). Due to the nonuniform Fe distribution in sample B4 with high doping concentration^[17], the shaded areas in Fig. 3(c) probably are caused by the influence of the strong ferromagnetism. Generally, there are no visible defects in Ga_1-x-yFe_xNi_ySb layer and no phases other than Ga_1-x-yFe_xNi_ySb.

A SQUID magnetometer was used to investigate the magnetic properties of Ga_1-x-yFe_xNi_ySb samples. Fig. 4 shows the magnetization hysteresis curves (M−H) of samples B1−B4 (x ≈ 24%, y = 1.7%−6.1%) measured at 10 K, with the magnetic field H being applied along the [110] and [001] axes. Clear hysteresis curves are observed and shown in Fig. 4, which demonstrate the presence of ferromagnetic order of Ga_1-x-yFe_xNi_ySb layers. For the samples with y = 1.7%, the in-plane magnetization and perpendicular magnetization show similar characteristics, suggesting small magnetic anisotropy in this sample. For the samples with y = 2.8%−6.1%, the difference between in-plane and perpendicular magnetization is remarkable. The perpendicular magnetization saturates at a much smaller magnetic field than the in-plane magnetization, indicating that the easy axis of magnetization is perpendicular to the plane and there is a strong magnetic anisotropy in these samples. As shown in Fig. 4, the perpendicular saturation magnetic fields of samples B1−B4 are marked by the blue arrows. In addition, the saturation magnetization M_S values increase as y increases, which are 148, 152, 159 and 165 emu/cm³ at y = 1.7, 2.8, 4.5 and 6.1%. The magnetic anisotropy constant could be estimated by Ku=∫0HS(M⊥−M∥)dH+2πMS2, where M_⊥ and M_|| represent the perpendicular and in-plane magnetization, respectively. Here, the saturation field H_S means the intersection of the in-plane and perpendicular M−H loop, which is marked by the red arrows in Fig. 4. It is found that the K_u is negligible for y = 1.7% but increases to 2.3 × 10⁵ erg/cm³ for y = 4.5% and 3.8 × 10⁵ erg/cm³ for y = 6.1%, suggesting the improvement of Ni doping to the magnetic anisotropy of Ga_1-x-yFe_xNi_ySb. Additionally, the magnetization hysteresis curves of samples A1−A4 (x = 25.2%−29.8%, y = 0) show similar characteristics with that of sample B1, of which the K_u can also be neglected. The results imply that Ni doping can effectively improve the magnetic anisotropy of Ga_1-x-yFe_xNi_ySb.

Hall data of samples B1−B4 was collected to characterize the magnetic transport properties of the Ga_1-x-yFe_xNi_ySb films. Samples B1−B4 were fabricated into Hall bars with a size of 200 × 40 μm². The hole mobility of Ga_1-x-yFe_xNi_ySb is determined by μ=lΔRxywΔBRxx, where the R_xy, R_xx , l and w are the anomalous Hall resistance, the longitudinal Hall resistance measured at zero-field, and the length and width between the electrodes of Hall bar, respectively. In order to accurately determine the hole mobility of Ga_1-x-yFe_xNi_ySb, the influence of AHE should be excluded. Thus, we measured R_xy at low temperature (10 K) and calculated the ΔR_xy/ΔB at high magnetic field (up to 16 T). As shown in Fig. 5, we measured the magnetic field dependence of Hall resistance curves (R_Hall−H) of samples B1−B4 (x ≈ 24%, y = 1.7%−6.1%) at 10 K. As plotted in the insets of Fig. 5, the obvious hystereses at low temperature reveal that R_Hall is dominated by the AHE in the low magnetic field range. Moreover, the red dashed lines in Fig. 5 almost coincide with the R_Hall−H curves at high magnetic fields, implying that the magnetization is nearly saturated. The mobility of Ga_1-x-yFe_xNi_ySb samples B1−B4 (x ≈ 24%) is listed in Table 1, which decrease from 31.3 cm²/(V∙s) at y = 1.7% to 4.6 cm²/(V∙s) at y = 6.1%. As for samples A1−A4 (y = 0), µ values also decline from 6.2 cm²/(V∙s) at x = 25.2% to 1.1 cm²/(V∙s) at x = 29.8%. Both Ni doping and Fe doping would reduce the hole mobility of Ga_1-x-yFe_xNi_ySb, however, samples co-doped with Fe and Ni exhibit higher mobility than that doped with Fe solely.

The Curie temperatures of samples B1−B4 (x ≈ 24%) are estimated by the Arrott plots of the R_Hall−H curves, which rise from 319 K (y = 1.7%) up to 354 K (y = 6.1%). With regard to samples A1−A4, the T_C values of these samples also increase from 337 K at x = 25.2% to 375 K at x = 29.8%, demonstrating similar impurity concentration dependence characteristics. The effects of Ni doping on mobility and T_C behave similar but weaker than that of Fe doping.

In summary, we co-doped Ni with Fe into GaSb host semiconductor by using LT-MBE, and achieved enhanced magnetic anisotropy and hole mobility of Ga_1-x-yFe_xNi_ySb films compared with that of (Ga,Fe)Sb. XRD and STEM results reveal that there are no second phases in Ga_1-x-yFe_xNi_ySb films. Moreover, we obtained the anisotropy constant K_u (up to 3.8 × 10⁵ erg/cm³) and hole mobility µ (up to 31.3 cm²/(V∙s)) in Ga_1-x-yFe_xNi_ySb films, which are much higher than that of (Ga,Fe)Sb (with the maximum K_u = 7.6 × 10³ erg/cm³ and maximum µ = 6.2 cm²/(V∙s)). Our results indicate that Ga_1-x-yFe_xNi_ySb is a promising magnetic semiconductor for spintronic applications.