Figure 1

(a) Schematic model of an ideal abrupt LaCrO${}_3$/SrTiO${}_3$(001) interface, viewed along the [010] direction. The atom radii are scaled by $Z$ number, to help highlight comparison to the $Z$-based contrast of HAADF-STEM and RBS. (b) Random HRRBS spectrum for a 5-u.c.-thick LCO/STO(001) film (blue circles), along with SIMNRA simulations for the abrupt and optimized intermixed interfaces (red and orange lines, respectively). It is clear that an intermixed model fits more accurately, particularly for the Sr (∼350 keV) and La (∼370 keV) signals.Reuse & Permissions

Figure 2

(a) In a representative cross-sectional HAADF-STEM image of a typical 130-u.c.-thick LCO/STO sample, there is clearly a gradual change in column intensity across the interface, indicating that the LCO/STO interface is not abrupt (image is adjusted for systematic sample drift during data acquisition, but otherwise unprocessed). (b) To more quantitatively analyze the extent of interdiffusion at the interface, the HAADF intensity has been integrated (from the original, unprocessed data) and averaged over 30, 0.8-Å-wide lines situated separately along the $A$-site (dark blue) and $B$-site columns (light orange).Reuse & Permissions

Figure 3

Comparison of individual atomic concentration profiles for a 25-u.c.-thick LCO film. (a) EDS line profiles extracted from the raw, integrated peak areas, and compared with concurrently collected HAADF intensity (gray line) to account for sample drift. The combined Cr-$K$ and La-$L_{\beta }$ signal should be predominantly indicative of Cr (in agreement with the adjoining La and HAADF profiles). (b) Line profiles extracted from chromatic-aberration-corrected EFTEM elemental maps (c), (d), for the Ti $L$ and La $N$ edges. Profiles acquired with linewidths of roughly 1 and 10 u.c. are included for each. Interfacial inhomogeneities are arguably visible in both maps. For both EDS and EFTEM profiles, the intensities were independently normalized to facilitate comparison, and the vertical axes are in arbitrary units.Reuse & Permissions

Figure 4

(a) Calculation of the charge per unit cell across the interface assuming no change in cation valency (i.e., a net charge equal to 3[La]+2[Sr]+3[Cr]+4[Ti]–6) using atomic concentrations derived from EDS data. This demonstrates that in the absence of local electronic modifications, there would be a net excess of ∼0.2 $e^{-}$/u.c. on the LCO side of the interface (or 10${}^{14}$ $e$/cm${}^2$). (b) However, measurement of the Ti $L$ (blue squares) and Cr $L$ edges (red circles) with EELS indicates that there is some degree of valence change for the $B$-site cations. A partial shift in the Ti valence, from Ti${}^{4+}$ towards Ti${}^{3+}$ going into the LCO, is demonstrated by fitting $L$-edge $t_{2g}$:$e_g$ ratios (blue squares). The Ti $L_3$ ratios are shown, as the $L_3$ splitting should be more pronounced, but the $L_2$ edge trend is the same. There is also a partial change in the Cr oxidation state approaching the interface from Cr${}^{3+}$ to Cr${}^{2+}$, evidenced by increasing $L_3$:$L_2$ ratios. There is also a minor chemical shift (∼−0.2 eV) as expected from a partial shift towards Cr${}^{2+}$. Error bars are based upon the fitting error, which increases for decreasing concentrations of Ti or Cr. (c) Representative Ti $L$- and Cr $L$-edge spectra (1) away from the interface and (2) near the interface on the LCO side are included for the points indicated in (b), with the raw data indicated by hollow circles (blue or red) and the result of peak fitting traced in black.Reuse & Permissions

Figure 5

(a) Averaged profiles for 5-, 15-, 25-, and 130-u.c.-thick LCO films on STO, as measured by HAADF-STEM intensity (shifted vertically for comparison). Each region included in this examination was ∼40–50 nm thick in cross section. One standard deviation of the measurement at each position is included as a colored range around the measured profiles. The position is labeled in unit cells based upon the spacing of STO, and the zero position of the interface is not quantitative here. The widths at 10$\%$–90$\%$ intensity of the best-fit curves for several such measurements are summarized in (b), rounded to the nearest unit cell. The 2 u.c. error bars on the 5 u.c. data point correspond to the $B$-site measurement. The data suggest an increasing extent of interdiffusion with increasing LCO film thicknessReuse & Permissions

Figure 6

(a) Random HRRBS spectra for LCO/STO films of various thicknesses, and (b) the La atom profiles resulting from fitting these spectra with simnra calculations. The original interface position was estimated by multiplying the areal density from the simulations by the atom number density in coherently strained LCO and shifting this scale by the actual film thickness from RHEED intensity oscillations. The abrupt steps are an artifact of the models, and do not imply discrete concentration shifts.Reuse & Permissions