Surface-interface coupling in an oxide heterostructure: Impact of adsorbates on ${\mathrm{LaAlO}}_3/{\mathrm{SrTiO}}_3$

The microscopic origin of the two-dimensional electron system (2DES) at the ${\mathrm{LaAlO}}_3/{\mathrm{SrTiO}}_3$ heterointerface remains a topic of debate. Among others, mechanisms involving surface defects acting as electron donors for the 2DES have been proposed recently. Since samples are usually investigated after exposure to air, adsorbates are a plausible candidate for such surface defects. We therefore used in situ photoelectron spectroscopy and in situ resistivity measurements to study the impact of adsorbates on the 2DES. The adsorbate coverage caused by storage in air is indeed found to induce additional charge carriers at the interface, compared to the adsorbate-free surface. In controlled adsorption experiments with water and hydrogen we identify adsorbed atomic hydrogen to induce additional charge carriers at the interface, differing from molecular water adsorption reported for samples exposed to macroscopically large volumes of water in air.

Figure 1

(a) Film growth as monitored by RHEED. The regular intensity oscillations of the specular reflex (indicated by red arrow in the insets) confirm layer-by-layer growth. (b) LEED pattern taken at an electron energy of 92 eV. The intensity profile along the dashed red line highlights additional spots corresponding to a $\sqrt{2}\times \sqrt{2}$ reconstruction rotated by ${45}^{\circ }$. (c) AFM image of the atomically flat sample surface. A height profile along the red line shows that steps are approximately one unit cell high. (d) Left: Reciprocal space map of a sample with a 20-uc-thick LAO film showing pseudomorphic growth on STO; right: intensity profile along $Q_z$ through the STO (103) Bragg spot. (e) Scanning transmission electron microscopy (STEM) image of the interface between LAO and STO. La atoms are marked in green and Sr atoms in blue. (f) Temperature-dependent conductivity measurements as a function of LAO thickness reveal a critical thickness for 2DES formation of 4 uc.

Figure 2

(a) $\mathrm{O}1s$ core level spectra of a postoxidized (PO) and a non-post-oxidized (noPO) sample. The shoulder on the high-binding-energy side indicates water contamination of the sample acquired during postoxidation. The angle dependence reflects the surface location of the contaminants. (b) Baking the gas pipe system between oxygen cylinder and gas inlet valve of the growth chamber reduces the contamination. Further purification can be achieved by a gas dryer resulting in the same spectral peak shape for PO and noPO samples. (c)–(e) LEED patterns reflect the surface cleanliness, same color code as in (b).

Figure 3

[(a) and (b)] Samples exposed to air (ex situ) exhibit strong water and carbon contaminations. (c) The relative ${\mathrm{Ti}}^{3+}$ vs ${\mathrm{Ti}}^{4+}$ intensity ratio (see inset for a close-up of the ${\mathrm{Ti}}^{3+}$ emission), which is a measure of the interfacial charge carrier density, is increased after storage in air. Annealing in dry oxygen recovers the in situ measured spectral line shape [red data points in (a)–(c)] and LEED pattern quality [inset in (b)]. In (a), the spectra are normalized to equal integrated intensity of the $\mathrm{Sr}3p$ and $\mathrm{La}4s$ core levels.

Figure 4

(a) $\mathrm{O}1s$ spectra, normalized to the $\mathrm{Al}2p$ core level, of the clean and hydrogen- (left panel) and water-vapor- (right panel) dosed surfaces. The total $\mathrm{O}1s$ spectral weight is not increased after dosing and the water- and hydrogen-dosed surfaces exhibit a similar $\mathrm{O}1s$ line shape. This suggests that only hydrogen binds to the surface and oxygen is absent in the adsorbate layer. (b) Schematic model of the surface adsorption; see text for details.

Figure 5

(a)–(c) Core-level spectra illustrating the changes of the electronic structure induced by surface protonation. The enhancement of the ${\mathrm{Ti}}^{3+}$ signal in the $\mathrm{Ti}2p$ core level points to a transfer of additional electrons to the interface (a), which is accompanied by a shift of the LAO core levels to higher binding energies (b). No changes in the spectral line shape of the film core levels is detected (c). (d) ${\mathrm{Ti}}^{3+}/{\mathrm{Ti}}^{4+}$ spectral weight ratio and relative $\mathrm{Al}2p$ core level shift for the complete set of experiments. The $x$ axis represents the chronological order of the measurements. The dashed line serves as a guide to the eye.

Figure 6

In situ resistivity measurements: (a) Sample exposed to water vapor, hydrogen, and oxygen atmosphere with cold cathode gauge initially switched off. The sheet resistivity decreases only when the pressure gauge is activated. This suggests that a cracking of the molecules is crucial for the adsorption process. A control experiment in oxygen results only in minor changes of the resistivity, probably due to residual water/hydrogen contaminations in the gas. (b) The interface of a sample with a 3-uc-thick LAO film can be switched from insulating to conducting by exposure to water vapor.

Figure 7

Schematic picture of the adsorption processes and their effect on the electronic structure of LAO/STO. Molecules are decomposed in the gas discharge of the cold cathode pressure gauge. Hydrogen atoms and ions protonate the sample surface, whereas oxygen or ${\mathrm{OH}}^{-}$ ions remain in the atmosphere. The binding of atomic hydrogen generates one electron free to move to the interface, whereas no electron is donated when ${\mathrm{H}}^{+}$ binds to the surface. The electron transfer is accompanied by a shift of the LAO bands to lower energy (marked by red arrows).