X-ray photoemission studies of the metal-insulator transition in ${\text{LaAlO}}_3/{\text{SrTiO}}_3$ structures grown by molecular beam epitaxy

${\text{LaAlO}}_3/{\text{SrTiO}}_3$ interfaces grown by molecular beam epitaxy show a metal-insulator transition at a critical film thickness of four unit cells of ${\text{LaAlO}}_3$ on ${\text{TiO}}_2$-terminated ${\text{SrTiO}}_3$ substrates. This transition had previously been observed in ${\text{LaAlO}}_3$ films grown by pulsed laser deposition, where defects related to the growth process have been suggested as playing a role in the interface behavior. X-ray photoemission was used to examine the band offsets and look for an electric field across ${\text{LaAlO}}_3$ for a range of film thicknesses on both SrO- and ${\text{TiO}}_2$-terminated ${\text{SrTiO}}_3$ substrates. These results are compared to the predictions of the polar catastrophe model for the emergence of a metallic interface in this system. We do not find the predicted electric field in the ${\text{LaAlO}}_3$ or any dependence on substrate termination. While the observation of metal-insulator transitions in ${\text{LaAlO}}_3/{\text{SrTiO}}_3$ structures grown by different techniques points to an intrinsic effect, the absence of an electric field in the ${\text{LaAlO}}_3$ layer is not consistent with the polar catastrophe model of interface metallicity.

Figure 1

(Color online) (a) Room-temperature conductivity of LAO/STO interfaces grown on ${\text{TiO}}_2$-terminated STO. The dashed line signifies the sensitivity limit of measurement. (b) $4\times 3\mu {\text{m}}^2$ AFM image of 5 uc LAO grown on ${\text{TiO}}_2$-terminated STO. (c) RHEED oscillations observed during LAO growth. The growth rate is 1 uc/min.

Figure 2

(Color online) (a) Polar catastrophe model for the appearance of conductivity above 4 uc on ${\text{TiO}}_2$-terminated substrates. The LAO valence band bends upward until injection of electrons into the STO conduction band can occur. The subsequent downward bending of the STO bands is not shown. The experimentally determined valence-band offset is 0.35 eV. Band-gap values are from Refs. 21, 22. (b) A schematic of a composite photoemission peak from the LAO core levels showing shift and broadening.

Figure 3

(Color online) (a) $\text{La}4d$ and (b) $\text{Al}2p$ photoemissions in annealed samples. To cancel out charging and surface photovoltage effects, the $x$ axis is the binding energy relative to the $\text{Sr}3d_{5/2}$ peak in the same sample. The thin solid line tracks the observed peak center, while gray bands follow the range of predicted shifts of the peak, according to the polar catastrophe picture for the ${\text{TiO}}_2$-terminated case. The predicted range is set by built-in fields of 0.6 and 0.9 V/uc taken from Refs. 7, 8. The upward shift expected to lead to the metal-insulator transition is not observed. Curves are shifted vertically for clarity.

Figure 4

(Color online) FWHM of $\text{La}4d_{5/2}$ peak in annealed samples. Lines follow the expected broadening of the peak, according to the polar catastrophe picture for the ${\text{TiO}}_2$-terminated case. Predicted built-in field magnitudes are taken from Refs. 7, 8. Broadening in the emission peak due to the potential variation across the film is not observed. No dependence on thickness or termination is observed, and the width is similar to that observed in thick LAO films.

Figure 5

(Color online) Left: $\text{La}4d_{5/2}\text{-Sr}3d_{5/2}$ separation in as-grown samples. After 3 uc thickness the two terminations diverge. The solid lines track simulated peak positions when the band bending in the LAO is caused by constant fields of 0.06 and 0.14 V/uc for ${\text{TiO}}_2$- and SrO-terminated cases, respectively. Right: $\text{La}4d$ levels in Ti-terminated (black, bottom) samples, 4 and 9 uc samples. 4 and 7 uc samples, Sr terminated (red, top). The vertical lines cross the centers of the 9 and 7 uc $\text{La}4d_{5/2}$ peaks.