Surface phase, morphology, and charge distribution transitions on vacuum and ambient annealed $\mathrm{SrTi}{\mathrm{O}}_3$(100)

The surface structures of $\mathrm{SrTi}{\mathrm{O}}_3$ (100) single crystals were examined as a function of annealing time and temperature in either oxygen atmosphere or ultrahigh vacuum (UHV) using noncontact atomic force microscopy (NC-AFM), Auger electron spectroscopy (AES), and low-energy electron diffraction (LEED). Samples were subsequently analyzed for the effect the modulation of their charge distribution had on their surface potential. It was found that the evolution of the $\mathrm{SrTi}{\mathrm{O}}_3$ surface roughness, stoichiometry, and reconstruction depends on the preparation scheme. LEED revealed phase transitions from a $\left(1\times 1\right)$ termination to an intermediate $c(4\times 2)$ reconstruction to ultimately a $(\surd 13\times \surd 13)-R33.7^{\circ }$ surface phase when the surface was annealed in an oxygen flux, while the reverse transition from $(\surd 13\times \surd 13)-R33.7^{\circ }$ to $c(4\times 2)$ was observed when samples were annealed in UHV. When the surface reverted to $c(4\times 2)$, AES data indicated decreases in both the surface Ti and O concentrations. These findings were corroborated by NC-AFM imaging, where initially $\mathrm{Ti}{\mathrm{O}}_2$-terminated crystals developed half-unit cell high steps following UHV annealing, which is typically attributed to a mix of SrO and $\mathrm{Ti}{\mathrm{O}}_2$ terminations. Surface roughness evolved nonmonotonically with UHV annealing temperature, which is explained by electrostatic modulations of the surface potential caused by increasing oxygen depletion. This was further corroborated by experiments in which the apparent roughness tracked in NC-AFM could be correlated with changes in the surface charge distribution that were controlled by applying a bias voltage to the sample. Based on these findings, it is suggested that careful selection of preparation procedures combined with application of an electric field may be used to tune the properties of thin films grown on $\mathrm{SrTi}{\mathrm{O}}_3$.

Figure 1

Selected LEED results representing the three main structures observed in tandem with their simulated LEED patterns: (1) $c(4\times 2)$ phase (a), obtained after a 30-min anneal in ${\mathrm{O}}_2$ flux at 1270 K, and simulation (b). (2) $(\surd 13\times \surd 13)-R33.7^{\circ }$ phase (c), prepared by extending the anneal time to 10 h, along with its simulated pattern (d). (3) $(4\times 4)$ phase (e), obtained by sequential heating in UHV to first 900 K and then 950 K for 30 min each following a preparation as in (2), and its corresponding theoretical LEED pattern (f). All LEED patterns have been recorded at 130 eV beam energy. The white dots in (b), (d), and (f) reflect the $\left(1\times 1\right)$ periodicity, the blue dots the periodicities of the reconstructions, and the red dots in (b) and (d) rotationally equivalent domains of the reconstruction.

Figure 2

Structural phase transitions on $\mathrm{SrTi}{\mathrm{O}}_3$ (100) as a function of temperature, time, and environment. (a) Under continuous ${\mathrm{O}}_2$ flux at $T=1270\mathrm{K}$, LEED patterns evolve from $1\times 1$ to $c(4\times 2)$ and finally $(\surd 13\times \surd 13)-R33.7^{\circ }$. (b) If, after 30 min in ${\mathrm{O}}_2$ flux, annealing is continued in UHV at a lower temperature $\left(T=900\mathrm{K}\right)$ and lower ${\mathrm{O}}_2$ pressure (${10}^{-6}$ mbar), the same trend is observed but with a longer time constant; the $c(4\times 2)$ is still visible even after 16 h of heating. (c) The transformation is reversed for annealing in UHV, where the $c(4\times 2)$ reconstruction can be recovered from a $(\surd 13\times \surd 13)-R33.7^{\circ }$-terminated surface (c-i–iii). Further annealing results in combinations of $c(4\times 2),(4\times 4),(2\times 2)$, and $(2\times 1)$ reconstructions (b-iv,v). (d) When starting from a $c(4\times 2)$-reconstructed surface, the results are qualitatively similar to those in (c), but with the transformations at slightly lower temperatures.

Figure 3

(a) AES derivative spectra obtained on the as-introduced sample (dark blue) as well as on the same sample after it had been heated to 600 K (light blue), 900 K (green), and 1000 K (red) for 35 min each; all spectra are normalized with respect to oxygen peak. Initial traces of carbon contamination are removed for annealing temperatures above 600 K. (b) Plot of the ratios of the Sr/O (red) and Ti/O (blue) peak heights with respect to their peak heights at $T=300\mathrm{K}$. The data suggest that while the surface overall loses oxygen, the surface concentration of strontium increases relative to titanium. The error of the normalized ratio was determined to be ±5% while the uncertainty in the temperature reading was estimated at ±25 K.

Figure 4

NC-AFM images illustrating the evolution of the surface morphology of a [100]-oriented $\mathrm{SrTi}{\mathrm{O}}_3$ single crystal sequentially annealed in UHV for 30 min to the temperatures indicated in the panels. Step heights evolve from unit cell step height (i.e., ≈4 Å; blue dotted lines) in (a) to a mixture between 0.5- (green dotted lines), 1-, and 1.5-unit cell step heights (red dotted lines) in (b) and (c) to almost exclusively 1.5-unit cell step heights in (d). In (e), 2-Å deep holes start to appear, while elsewhere in the image steps featuring 2.5-unit cell heights (cyan) have been formed. Image size is $500\times 500\mathrm{n}{\mathrm{m}}^2$ in all cases.

Figure 5

Root mean square surface roughness of $\mathrm{SrTi}{\mathrm{O}}_3$, as measured on otherwise “atomically flat” terraces, plotted for different UHV annealing temperatures. After an initial sharp decrease, it is found to rise again for values above ≈1100 K.

Figure 6

(a)–(e) NC-AFM images (180 nm × 180 nm) demonstrating substantial changes in the apparent surface roughness while ramping the bias voltage applied to the back of the crystal. All images cover the same $z$ range of 4.8 Å from darkest color to brightest color. (f) Plot of the root mean square (rms) surface roughness as a function of the applied bias voltage $V_{\mathrm{ts}}·$ (g) Cross sections along the lines highlighted in image (a) (red) and (e) (blue) visualizing the decrease in peak-to-peak roughness. The sample was a 0.7% Nb-doped $\mathrm{SrTi}{\mathrm{O}}_3$ crystal that was prepared by heating in ${\mathrm{O}}_2$ flow for 30 min to 1270 K followed by a 30-min anneal in UHV at 800 K. The minimum observed roughness [in (e)] is ≈0.4 Å, which is consistent with roughness data calculated from the terraces shown in Fig. 4.

Figure 7

Results of bias sweep experiments conducted on undoped $\mathrm{SrTi}{\mathrm{O}}_3$, where the vertical position of the tip relative to the surface (labeled as $z$ piezo) was recorded as a function of a bias voltage applied to the backside of the $\mathrm{SrTi}{\mathrm{O}}_3$ crystal. (a) $z$ piezo vs bias voltage plot. Three different regions (accumulation, transition, and inversion) were observed when ramping the bias. (b) Zoom into the transition region. (c) Plot of $V_{\min }$, representing the value of the applied bias voltage $V_{\mathrm{ts}}$ at the minimum of the parabola in the transition region, for bias sweeps carried out at 40 distinct lateral positions along a straight line with 5 nm spacing between individual points. (d) Normalized statistical distribution of the values plotted in (c); the blue line, which represents a Gaussian distribution, is added solely to guide the eye.