Influence of ${\mathrm{SrTiO}}_3$ capping layer on the charge transport at the interfaces of ${\mathrm{SrTiO}}_3/{\mathrm{LaAlO}}_3/{\mathrm{SrTiO}}_3$ (100) heterostructure

The experimental realization of quasi-two-dimensional hole gas at the complex oxide interface has long been a grand challenge that may open up new possibilities for oxide-based electronics. ${\mathrm{SrTiO}}_3$ (STO)-capped ${\mathrm{LaAlO}}_3$ (LAO)/${\mathrm{SrTiO}}_3$ (100) is a potential candidate that has been recently demonstrated to show possible electron-hole bilayer conducting channels near the interfaces of the heterostructure. In this work we present a systematic investigation on the structural and magnetotransport properties of STO-capped and uncapped LAO/STO (100) samples that were grown by an oxide molecular beam epitaxy system. The STO/LAO/STO (100) samples exhibit metallic interfaces down to $T$ = 2 K even with a LAO thickness of merely 3 unit cells (3-uc). On the contrary for uncapped LAO(5-uc)/STO (100) samples the conducting interface becomes insulating at lower temperatures due to the surface adsorbate on top of the LAO as revealed from x-ray photoelectron spectroscopy measurements. The magnetotransport data at different back-gate voltages for STO-capped and uncapped LAO/STO (100) samples were analyzed and compared to each other using the two-band model. The presence of two types of charge carriers with opposite signs is uniquely found for STO/LAO/STO (100) samples which is in big contrast to the uncapped LAO/STO (100) samples with only electron-type carriers. Our results support the coexistence of electron-type and hole-type carriers in the STO/LAO/STO (100) heterostructure where their sheet densities and mobilities can be tuned significantly by electrical gating effect and the layer thickness of the heterostructure.

Figure 1

(a) The in situ RHEED oscillations during the growth of the STO/LAO/STO (100) sample along with the shutters control operation for each cell. A 30-s pause was given between the LAO and STO growth. The RHEED patterns on a STO (100) substrate before and after the growth of the STO/LAO/STO (100) sample are shown in (b) and (c), respectively. The RHEED oscillations and streak lines after the growth indicate a high quality ultra-thin and single crystalline films for the STO/LAO/STO (100) sample.

Figure 2

(a) The AFM image of a ${\mathrm{TiO}}_2$ terminated STO (100) substrate before the film-growth. (b) The AFM image after the growth of 5-uc LAO layer and 6-uc STO capping layer. The lower panels of (a) and (b) display the averaged terrace step-height of about 0.4 nm and the averaged surface roughness of about 80 pm. After the growth, no significant changes in the surface roughness was found, but few random small mesas with a height of about one LAO unit cell are observed. A STEM image across the interfaces of the STO(6-uc)/LAO(5-uc)/STO (100) sample is shown in (c), where sharp interfaces and epitaxial growths of both LAO layer and STO capping layers were clearly identified.

Figure 3

(a) The temperature-dependent sheet resistance $R_{\square }$ for LAO(5-uc)/STO (100), LAO(10-uc)/STO (100), and STO(6-uc)/LAO($t$-uc)/STO (100). The LAO(5-uc)/STO (100) sample turns into an insulating nature rapidly below around 260 K, but samples with the STO capping layer show metallic behavior down to 5 K even with a merely 3-uc thickness in the LAO layer ($t$ = 3). The huge differences in the $R_{\square }-T$ curves for uncapped (blue line) and STO-capped (black line) LAO(5-uc)/STO (100) samples reveal the critical role of STO-capping layer on the conducting channels at the interfaces. (b) The XPS spectra of the LAO/STO (100) sample on the surface of LAO and below the surface are shown as black and red lines, respectively. The XPS spectrum reveals the presences of OH-ligands that is likely associated with the insulating behavior found in the uncapped LAO(5-uc)/STO (100) sample.

Figure 4

The variation of the Hall coefficient with applied magnetic field at different $V_g$ for (a) STO(6-uc)/LAO(5-uc)/STO (100), (b) LAO(10-uc)/STO (100) with low $n_{\text{2D}}$, and (c) LAO(10-uc)/STO (100) with high $n_{\text{2D}}$. The Hall coefficient of LAO(10-uc)/STO (100) with low $n_{\text{2D}}$ remains constant with a rather weak dependence on the applied magnetic field for all the $V_g$ values we used, but for (a) and (c), the Hall coefficients show apparent variations with applied magnetic field, where such variations grow larger as the $V_g$ increases.

Figure 5

Two-band model fittings to magnetotransport data. (a, c) The Hall conductance variations with respect to applied magnetic field at different $V_g$ values for STO(6-uc)/LAO(5-uc)/STO (100) and LAO(10-uc)/STO (100) with high $n_{\text{2D}}$, respectively. The Hall conductance vs. magnetic field data were fitted using the two-band model for $V_g$ $\ge$ 0 V as shown with dashed lines. (b, d) The corresponding magnetoconductance variations for STO(6-uc)/LAO(5-uc)/STO (100) and LAO(10-uc)/STO (100) with high $n_{\text{2D}}$, respectively. The ${\sigma }_{xx}-{\mu }_{0}H$ curves were reproduced using the two-band fitting parameters as shown with the short-dashed lines, which show excellent agreement with the experimental data.

Figure 6

$V_g$ dependence on the extracted sheet densities and Hall mobilities. (a, b) The variations of electron-type (solid circle) and hole-type (open circle) Hall mobilities and sheet densities, respectively, with $V_g$ for the STO(6-uc)/LAO(5-uc)/STO (100) sample. For LAO(10-uc)/STO (100) with high $n_{\text{2D}}$, the variation in the Hall mobilities and sheet densities of the $n_1$ (solid triangle) and $n_2$ (solid circle) are shown in panels (c) and (d), respectively. The solid pentagons and solid squares are the Hall slope data extracted by linear fits to the Hall resistance vs. field curves at lower fields of (0–1 T) and higher fields of (9–10 T), respectively. The difference in the Hall slopes at lower and higher field regimes is a direct indication of the presence of different types of charge carriers. The inset of panel (c) shows a schematic of the Ti $d$ orbitals near the interface between LAO layer and STO (100) substrate, where the $d_{XY}$ band is lower in energy as compared to the $d_{XZ}$ and $d_{YZ}$ bands due to the charge confinement effect.

Figure 7

(a) An illustration of the proposed scenario for the STO/LAO/STO (100) shows the presence of q2DEG and q2DHG at the lower and upper interfaces, respectively. The dimensions in the cartoons are highly exaggerated and shall not be scaled as the real device geometry. The existence of spatially separated q2DEG and q2DHG creates a built-in electric field $E_c$ even at zero $V_g$, where the influence to the q2DEG and q2DHG for positive and negative $V_g$ cases are shown in (b) and (c), respectively. The application of a large positive $V_g$ value tends to broaden the q2DEG at lower interface while the q2DHG was driven more close to the upper interface as demonstrated in the density profile shown on the right panel of (b). Similar arguments can apply for the case of a large negative $V_g$ shown in (c).