Spin splitting tunable optical band gap in polycrystalline GdN thin films for spin filtering

Rare-earth nitrides, such as gadolinium nitride (GdN), have great potential for spintronic devices due to their unique magnetic and electronic properties. GdN has a large magnetic moment, low coercitivity, and strong spin polarization suitable for spin transistors, magnetic memories, and spin-based quantum computing devices. Its large spin splitting of the optical band-gap functions as a spin filter that offers the means for spin-polarized current injection into metals, superconductors, topological insulators, two-dimensional layers, and other novel materials. As spintronics devices require thin films, a successful implementation of GdN demands a detailed investigation of the optical and magnetic properties in very thin films. With this objective, we investigate the dependence of the direct and indirect optical band gaps ($E_g$) of half-metallic GdN, using the trilayer structure AlN (10 nm)/GdN ($t$)/AlN (10 nm) for GdN film thickness $t$ ranging from 6 to 350 nm, in both paramagnetic (PM) and ferromagnetic (FM) phases. Our results show a band gap of 1.6 eV in the PM state, while in the FM state the band gap splits for the majority (0.8 eV) and minority (1.2 eV) spin states. As the GdN film becomes thinner, the spin-split magnitude increases by 60%, going from 0.290 to 0.460 eV. Our results point to methods for engineering GdN films for spintronic devices.

Figure 1

(a) X-ray reflectivity for the rf-sputtered AlN (10 nm)/GdN (22 nm)/AlN (10 nm) and theoretical fit used to obtain the GdN film thickness. (b) High-resolution x-ray diffraction pattern for AlN (10 nm)/GdN (350 nm)/AlN (10 nm) on a (0001)-oriented sapphire (${\mathrm{Al}}_2{\mathrm{O}}_3$) substrate. (c) In-plane field cooled magnetization at 50 Oe for GdN thicknesses of 13, 30, and $80\mathrm{nm}$. (d) Magnetization ($M$) vs in-plane applied magnetic field ($H$) at $T=10\mathrm{K}$ for AlN (10 nm)/GdN ($t$)/AlN (10 nm) structures, with $t=13$, 30, and $80\mathrm{nm}$.

Figure 2

(a) Illustration of the optical density acquisition apparatus composed of light source, sample, aperture, and light detector. (b) and (c) show the optical density spectra for the set of samples at 300 and 6 K. (d) shows how the direct optical band gap, at room temperature, for a $t=30$-nm-thick film was extracted from the squared-absorption coefficient ($\alpha$) vs photon energy. The linear red fitted curve ($y=Ax+B$) is used to determine the band gap. (e) and (f) show ${\alpha }^2$ and ${\alpha }^{1/2}$ vs the photon energy $\mathit{E}=h\nu$ for determining the direct and indirect optical band gaps of the GdN thin films, respectively. The dashed lines indicate the linear extrapolation to $y=0$ where the intersection corresponds to the direct/indirect band-gap energy.

Figure 3

(a) Illustration of the band structure of a magnetic semiconductor in the paramagnetic (left) and ferromagnetic (right) phases. In the PM phase, the band gap does not depend on the spin's orientation. (b) and (c) compare the ${\alpha }^2$ vs $E$ plots for obtaining the direct optical band gap of sapphire/AlN (10 nm)/GdN (30 nm)/AlN (10 nm) in the PM ($T=300\mathrm{K}$) and FM ($T=6\mathrm{K}$) phases. In the FM phase, shown in the inset of (c), we observe more than one linear portion that is related to the spin-split band gap. (d) and (g) show the direct and indirect optical band gaps dependence with the GdN film thickness at 300 and 6 K, where the black dotted lines correspond to their average energy values, and (e), (f), (h), and (i) are the zoomed direct and indirect band gaps for the minority and majority-spin states at 6 K. The spin splitting of the band gap is also present in the indirect transitions. (j) Dependence of the magnitude of the spin splitting as a function of the GdN thickness, showing that the splitting becomes larger as the film thickness decreases.