Light-induced metal-insulator transition in ${\mathrm{SrTiO}}_3$ by photoresistance spectroscopy

Photoresistivity and its spectral response has been systematically studied in oxygen-deficient ${\mathrm{SrTiO}}_3$ single crystals for a wide range of resistivities $\rho$ and carrier densities $n$. At room temperature we found a persistent photoresistance that gradually decreases as $\rho$ is diminished or $n$ is increased, in addition to relaxation times of seconds to a few minutes, suggesting that trapping of carriers is playing a major role. An analysis of the photoresistance excitation spectra showed two distinctive features that are related to the indirect gap of ${\mathrm{SrTiO}}_3$ at $(3.25\pm 0.04)\mathrm{eV}$ and to a direct transition at $(3.40\pm 0.03)\mathrm{eV}$. The photoresistive crystals present a temperature-dependent resistivity under illumination that experiences a metal-insulator transition below $T\sim 85\mathrm{K}$. The low-temperature photoresistance spectrum reveals a suitable technique to understand the origin of this transition, pointing to an enhanced efficiency of the $\sim 3.25\mathrm{eV}$ gap to promote electrons to the bottom of the conduction band.

Figure 1

Photoresistivity [defined as $(\rho -{\rho }_{\mathrm{dark}})/{\rho }_{\mathrm{dark}}]$ as a function of the time for the sample with ${\rho }_{\mathrm{dark}}=1.7\times {10}^8\mathrm{\Omega }\mathrm{cm}$ at $T=298\mathrm{K}$. The applied incident wavelength was $\lambda =365\mathrm{nm}$ $\left(\sim 3.40\mathrm{eV}\right)$. The inset shows the corresponding photoresistance for the samples with ${\rho }_{\mathrm{dark}}=1.8\times {10}^6\mathrm{\Omega }\mathrm{cm}$ (black squares) and ${\rho }_{\mathrm{dark}}=8\times {10}^5\mathrm{\Omega }\mathrm{cm}$ (green triangles).

Figure 2

(a) Time evolution of electrical current $I$ (normalized at its initial value $I_0)$ during the application of an incident light of $\lambda =365\mathrm{nm}$ for the sample S1 at $T=298\mathrm{K}$. The data are well fitted by a two exponential decaying function (red line). The inset shows the corresponding data and fit for the sample S2. (b) Time dependence of electrical current $I$ (normalized at its initial value $I_0)$ after an applied incident light of $\lambda =365\mathrm{nm}$ was removed (sample S1 at $T=298\mathrm{K})$. Again, the data are well fitted by a two exponential decaying function (red line). The inset shows the corresponding data and fit for sample S2. Other trial functions have been tested to fit our data without success.

Figure 3

(a) Resistivity (normalized at its 3.0-eV value) as a function of the incident light energy for the sample S3 at $T=110\mathrm{K}$. The arrow marks the direction of the scan (denoted as L-H scan). The maximum slope is reached at $(3.25\pm 0.04)\mathrm{eV}$. (b) Resistivity (normalized at its 3.9-eV value) as a function of the incident light energy for the same sample at $T=110\mathrm{K}$. The arrow marks the direction of the scan (denoted in this case as H-L scan). A minimum is reached at $(3.40\pm 0.03)\mathrm{eV}$.

Figure 4

Sketch of the STO energy diagram. VB and CB denote the valence and conduction band, respectively. A direct transition of $3.40\mathrm{eV}$ (labeled as 1 and represented by a solid arrow) between the maximum of the VB and the minimum of the CB is attained at the X point. An indirect transition of $3.25\mathrm{eV}$ (labeled as 2 and also represented by a solid arrow) takes place between the maximum of the VB at point $\mathrm{\Gamma }$ and the minimum of the CB at point X.

Figure 5

(a) Resistivity (normalized at its room-temperature value) as a function of temperature under the application of an incident light of $\lambda =365\mathrm{nm}$. The curves correspond to sample S1 (blue circles), S2 (black squares), and S3 (green triangles). For sample S1, resistivity measurements below $\sim 200\mathrm{K}$ are above our instrument range. Inset: Resistivity (normalized at its room-temperature value) without illumination as a function of temperature for the STO crystal S1 (blue circles), S2 (black squares), and S3 (green triangles). (b) Resistivity (normalized at its 3.9-eV value) as a function of the incident light energy for sample S3 at $T=31\mathrm{K}$. The arrow marks the direction of the scan (denoted in this case as H-L scan). The scanning speed of the incident light wavelength was $0.1\mathrm{nm}{\mathrm{s}}^{-1}$. Instead of a minimum, a plateau is observed between $\sim 3.40$ and $\sim 3.25\mathrm{eV}$. Inset: Photocurrent, PC, as a function of the incident light energy for the L-H scan. The data follows a $P{C}^{1/2}$ dependence with the energy.