Origin of Charge Density at ${\mathrm{LaAlO}}_3$ on ${\mathrm{SrTiO}}_3$ Heterointerfaces: Possibility of Intrinsic Doping

As discovered by Ohtomo and Hwang, a large sheet charge density with high mobility exists at the interface between ${\mathrm{SrTiO}}_3$ and ${\mathrm{LaAlO}}_3$. Based on transport, spectroscopic, and oxygen-annealing experiments, we conclude that extrinsic defects in the form of oxygen vacancies introduced by the pulsed laser deposition process used by all researchers to date to make these samples is the source of the large carrier densities. Annealing experiments show a limiting carrier density. We also present a model that explains the high mobility based on carrier redistribution due to an increased dielectric constant.

Figure 1

(a) NEXAS O $K_1$ of sample 0728 (deposited at ${10}^{-5}\mathrm{Torr}$) and sample 0802 (deposited at ${10}^{-6}\mathrm{Torr}$). (b) Vis-VUV-SE absorption constant $\alpha$ spectrum of the same samples as in (a).

Figure 2

Sheet carrier densities at 20 (blue symbols) and 300 K (red symbols) as a function of annealing temperature in 600 W atomic oxygen, for samples made at ${10}^{-5}\mathrm{Torr}$ of ${\mathrm{O}}_2$. 600 W of atomic oxygen corresponds to $\sim {10}^{17}$ oxygen atoms per ${\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$. The latter value was taken after Ingle et al. [15], who worked on the same system in our laboratory. The values at $25°\mathrm{C}$ indicate the as-deposited samples. The different symbol shapes indicate different samples made under similar conditions: two made at Stanford (circles and triangles), the other two made at the University of Twente (crosses and squares). Sample thickness ranges from 5 to 26 ML.

Figure 3

Calculations of the electron density as a function of distance from the interface for two different total numbers of electrons per ${\mathrm{cm}}^2$ ($3\times {10}^{14}$ and $1\times {10}^{16}$) at two different temperatures. The shaded blue area indicates the location of the oxygen vacancies for the high carrier density case ${10}^{16}$; for the low carrier density case, no oxygen vacancies were assumed, hence the different slopes at the interface.