Suppression of Superconductivity and Electrostatic Side Gate Tuning in High Mobility SrTiO$_3$ Surface Electron Gas

Abstract

We report on the fabrication and characterization of patterned high-mobility two-dimensional electron gases (2DEG) formed on SrTiO$_3$ (STO) substrate surfaces by hydrogen plasma exposure. The resulting devices consistently showed high electron mobilities up to 7400 cm$^2$/V$\cdot$s. A large range of electron density was systematically explored by controlled aging of the sample between cooldowns, including the expected range for the STO 2DEG superconducting dome. No superconducting transition was observed down to the base temperature of approximately 10 mK. This suggests suppression of superconductivity in high mobility quasi-two-dimensional SrTiO$_3$ electron gas, likely linked to vertical confinement and electronic orbital rearrangement. We systematically explored electrostatic gate modulation in this 2DEG system and its scaling with electron density and side gate geometry. In contrast with our initial expectation, we observed an improvement of achievable total modulation for larger side gate to channel separation. At low electron density, stochastic channel pinch-off events were observed, creating quasi-ballistic constrictions with irregular conductance quantization. This epitaxy-free and high mobility oxide material platform offers a promising new route towards patterning quantum devices.

Figures (11)

• Figure 1: (a) Optical image of the device E at 5X magnification. The dashed line indicates the location of the cross-sectional schematic shown on the right. Scale bar is 100 $\mathrm{\mu}$m (b) Sheet resistance as a function of temperature for two devices measured in a pulse tube cryostat. (c) Aging of the 2DEG in nitrogen desiccator storage. Gradual decrease in electron density with time after hydrogen plasma exposure follows a power law ($N_\mathrm{H}(t=0) - At^{\alpha}$). $N_\mathrm{H}$ represents the measured Hall electron density at $\approx$10 mK and $N_\mathrm{E}$ is the estimated room temperature carrier density using a mobility value of 5.5 cm$^2$/(V$\cdot$s) Mikheev15. (d) Accelerated 2DEG aging by mild heating of the sample. Shaded region transitions indicate hot plate temperature changes.
• Figure 2: a) Hall mobility at milliKelvin temperatures as a function of carrier density for side-gated Hall bar devices and Van Der Pauw samples. Average value across three Hall bar voltage pairs are shown for each cooldown. (b) Superconducting transition temperature versus electron density shown. The representative LaAlO$_{3}$/SrTiO$_{3}$ dome trace following Jouan22Mikheev23 is shown as a dashed line, along with data from Mikheev21BenShalom10. Absence of detectable superconductivity in HPE devices is marked as $T_\mathrm{c} =$ 0. (c) The measured flat temperature dependence of resistance in HPE devices A-F, with trace color mapped to Hall density. Superconducting transition traces measured in ionic liquid gated SrTiO$_3$ single crystal and thin film (from padhye26) in the same dilution refrigerator system are shown for comparison.
• Figure 3: (a) Shubnikov de Haas Oscillations resistance oscillations in out-of-plane magnetic field and temperature shown in the first and second field derivatives. Markers indicate the maxima and minima. (b) Peaks and dips in $d^2R_{\mathrm{S}}/dB^2$ were used to assign Landau level indices $n_{\mathrm{LL}}$. The dashed line shows the linear fit to electron density at the lowest temperature (60 mK). (c) Temperature dependence of the amplitude of the oscillations in the first derivative at $B=6.5$ T. The dashed line shows the fit to the Lifshitz-Kosevich model with $m^*=3.1$.
• Figure 4: (a) Symmetric modulation of channel conductance by two adjacent side gates in device A, $w=$ 5 $\mu$m, $g=$ 20 $\mu$m. (b) Side gate modulation efficiency in conductance ($\eta_G$) and Hall density ($\eta_N$) as a function of electron density for devices A-D. Dashed lines highlight $N_\mathrm{H}^{-1}$ scaling. (c) Equivalent effective dielectric constant with lumped geometric factor. For conductance modulation, mobility modulation by is also lumped into $f\varepsilon_r$. (d,e) Conductance modulation vs side gate voltage in two representative cooldowns of device A, highlighting a higher total modulation in remote gates due to lower leakage and despite lower gating efficiency.
• Figure 5: (a) Initial side gate voltage ramping on the $g=$ 20 $\mu$m section of device A at 5.5 $\times$ 10$^{13}$ cm$^{-2}$. $V_\mathrm{gL}$ is ramped to 10 V and back to zero first, followed by $V_\mathrm{gR}$. The black trace is a stable sweep with $V_\mathrm{gR}$. (b) Conductance was measured by ramping $V_{\mathrm{gL}}$ at various constant $V_{\mathrm{gR}}$ values, showing repeatable steps that slowly evolve with disorder potential tuning by $V_{\mathrm{gR}}$ . (c) DC bias spectroscopy showing quasi-ballistic subbands. $V_{\mathrm{DC}}$ is the measured DC voltage drop across the relevant channel section. Both gates were swept at equal gate voltage $V_\mathrm{gL}=V_\mathrm{gR}=V_\mathrm{gL,R}$. Series resistance of 4 k$\Omega$ was subtracted in (b) and (c).