Interface roughness governed negative magnetoresistances in two-dimensional electron gases in AlGaN/GaN heterostructures

Negative magnetoresistances (NMRs) have been widely observed in two-dimensional electron gases (2DEGs). However, their origins are under debate. Here, we report on NMRs in the 2DEG in AlGaN/GaN heterostructures, aiming to uncover their origins by utilizing electric field gating. We systematically measured the magnetoresistances in magnetic fields up to 12 T and at temperatures between 3 and 260 K and observed NMRs over a wide range of temperatures from 3 to 170 K, which become more pronounced with decreasing temperature. We conducted electric field gating experiments to correlate the occurrence of NMRs with the relationship between electron mobility and density. The latter is governed by defect scattering sources in the sample and can be theoretically modeled. Comparison of the measured electron mobility and electron density relationship with theory reveals that interface roughness scattering plays a crucial role in obtaining large NMRs. Our work demonstrates that electric field gating provides a means not only to tune the values of NMRs but also to uncover their mechanisms.

Figure 1

(a) Schematic (not to scale) of the layer structuring of the AlGaN/GaN heterostructure. (b) Micrograph and enlarged view of the sample in Hall bar geometry with width of $L_y=200\mu \mathrm{m}$ and voltage lead distance of $L_x=250\mu \mathrm{m}$. CH1 And CH2 in the enlarged view denote the measured Hall bar sections. Each section has four voltage contacts, with one pair in the longitudinal direction for measuring $R_{xx}$ and one pair in the transverse direction for measuring $R_{xy}$. (c) Temperature dependence of the sample resistance in the absence of a magnetic field and the Hall resistance at $B=9$ T. Inset shows the linear field dependence of the Hall resistances at the lowest ($T=3$ K) and highest ($T=260$ K) experimental temperatures. (d) Temperature dependence of the electron density $n$ and mobility μ derived from data in (c).

Figure 2

(a) The magnetoresistance ${\gamma }_{\mathrm{MR}}$ at $T=3$ K. The inset is an expanded view to show the weak localization induced cusp at zero field. (b) Data in (a) plotted as ${\gamma }_{\mathrm{MR}}$ versus $B^2$ to show the parabolic field dependence. The main panel shows the linear section while the inset presents the complete set of data. (c) ${\gamma }_{\mathrm{MR}}$ obtained at various temperatures. The magnetic field is applied perpendicular to the 2DEG.

Figure 3

(a) Magnetic field dependence of the resistance components Δ$R$ induced by Shubnikov–de Haas (SdH) quantum oscillations at various temperatures. (b) Fast Fourier transform (FFT) spectra of the data in (a). (c) Temperature dependence of $\mathrm{\Delta }{R}_{\mathrm{M}}$ at $B=11.46$ T (positive maxima in the SdH oscillations). Circles are experimental data and the line is a fit to Eq. (4) to derive the effective mass $m^{*}$. (d) Magnetic field dependence of peak values $\mathrm{\Delta }{R}_{\mathrm{M}}$ in the SdH oscillations at $T=3$ K. Circles are experimental data and the line is a fit to derive the quantum time from Eq. (4).

Figure 4

(a) Gate voltage dependence of the sample resistance at $B=0$ and 12 T. The inset presents ${\gamma }_{\mathrm{MR}}$ versus $B$ curves obtained at various gate voltages. (b) Color map of the field and electron density dependences of ${\gamma }_{\mathrm{MR}}$. It is constructed by taking $R\sim V_g$ curves at magnetic fields from 0 to 4.5 T in intervals of 0.5 T and from 4.5 to 12 T in intervals of 0.25 T. The data were taken at $T=3$ K.

Figure 5

Gate voltage dependence of the electron mobility and density (a) and ${\gamma }_{\mathrm{MR}}$ (b). The electron density is calculated from the measured Hall resistance. The mobility is derived from the measured zero-field resistance and the calculated Hall electron density. The data were taken at $T=3$ K.

Figure 6

Comparison of the experimental and theoretical relationships between the electron mobility and density. Symbols are for experimental data and curves are calculated (equations and discussions are presented in the Supplemental Material [42]).