Two-dimensional electron gas in a metal/amorphous oxide interface with spin-orbit interaction

Abstract

The formation of novel two-dimensional electron gas (2DEG) with high mobility in metal/amorphous interfaces has motivated an ongoing debate regarding the formation and novel characteristics of these 2DEGs. Here we report an optical study, based on infrared spectroscopic ellipsometry, of nonmagnetic metal and amorphous semiconducting oxide (Cu/Bi2O3) interfaces that confirms the formation of a 2DEG with spin orbit coupling (SOC). The 2DEG optical response was simulated with a uniaxial diagonal dielectric tensor within a sub-nanometer thin layer, where its x and z components lineshapes resolved in both free-electron and peak-like contributions, resulted very similar to theoretical predictions [M. Xie et al., Phys. Rev. B 89, 245417 (2014)] of a two dimensional electron gas confined in the normal direction of a perovskite interface. In particular, the small but finite conducting character of the z component provides a unambiguous signature of the presence of the 2DEG in the Cu/Bi2O3 system. Although the original constituent materials do not possess spin-orbit coupling (SOC), the resulting interfacial hybridization of such states induce electronic asymmetric wave functions. This work demonstrates the detection of 2DEG in amorphous crystals allowing to study its challenging interfacial phenomena such as SOC and interface-bulk coupling, overcoming an experimental impediment which has hold back for decades important advancements for the understanding of 2DEGs in amorphous materials.

Mid-infrared optical properties of non-magnetic-metal/CoFeB/MgO heterostructures

“…To model ψ spectra in figures 2(a)-(c), an additional interface layer had to be included between the CoFeB and the NM layers for simulating the presence of a 2DEG, as shown in figure 1. This is similar to the one reported in [41]. The DF of the 2DEG is anisotropic and depends on the underlying NM as shown in figure 6.…”

“…Its assignment was made similarly to the theoretical analysis by Xie et al [62]. As previously discussed in [41], despite the theory developed for field induced split of d-bands in perovskite oxides, the presence of the SOC character seems general for 2DEGs. We note that the Drude contribution to ε xx , see figure 6(a), strongly depends on the NM: the conductivities of the 2DEG, here more evinced below 200 meV, are much higher for the Pt-and W-seeded samples than for the Cu one.…”

“…This is offset by (ii) a real ε ∞ = 14.635, but this is estimated by considering only the mid-IR range. A first-approach of percolation theory for conduction electrons at optical frequencies [49], successfully used in our previous work [41], did not provide a correct reproduction of experimental curves for CoFeB samples. The mentioned approach can be algebraically decomposed into two Drude contributions and a Lorentzian.…”

“…See equation ( 1)) as seen when comparing the top panels of figures 2(a)-(c). We used a previously reported [41] percolation-modified DF for Cu. The optical properties of the NM layers are shown in figure S7 in the supplemental information.…”

“…For the optical response of the whole stack, after most of each ψ spectrum could be well described using a material-only model based on the stratified optical medium in figure 1 neglecting the 2DEG layer, i.e. the one made by including all the deposited constituents' DFs and film thicknesses reasonably close to nominal values; then we considered the formation of an interfacial 2DEG layer similarly as reported previously for the Cu/Bi 2 O 3 interface [41]. This additional layer is used to resolve some discrepancies between the experiment and the material-only model, which are, e.g.…”