Large and persistent photoconductivity due to hole-hole correlation in CdS

Large and persistent photoconductivity (LPPC) in semiconductors is due to the trapping of photogenerated minority carriers at crystal defects. Theory has suggested that anion vacancies in II-VI semiconductors are responsible for LPPC due to negative-$U$ behavior, whereby two minority carriers become kinetically trapped by lattice relaxation following photoexcitation. By performing a detailed analysis of photoconductivity in CdS, we provide experimental support for this negative-$U$ model of LPPC. We also show that LPPC is correlated with sulfur deficiency. We use this understanding to vary the photoconductivity of CdS films over nine orders of magnitude, and vary the LPPC characteristic decay time from seconds to ${10}^4\mathrm{s}$ by controlling the activities of $\mathrm{C}{\mathrm{d}}^{2+}$ and ${\mathrm{S}}^{2-}$ ions during chemical bath deposition. We suggest a screening method to identify other materials with long-lived, nonequilibrium, photoexcited states based on the results of ground-state calculations of atomic rearrangements following defect redox reactions, with a conceptual connection to polaron formation.

Figure 1

Illustration of lattice distortions around ${\mathrm{V}}_{\mathrm{S}}$ in the ${\mathrm{V}}_{\mathrm{S}}^{\times }$ ground state (a) and ${\left({\mathrm{V}}_{\mathrm{S}}^{••}\right)}^{*}$ nonequilibrium state (b). The bond lengths are as calculated by Nishidate et al. [24]. The ${\mathrm{V}}_{\mathrm{S}}$ site is indicated by a gray sphere. The orthographic projection along a Cd-S bond allows easy comparison of the distortion of the $\mathrm{C}{\mathrm{d}}_4$ tetrahedron in the undistorted lattice (bottom right of each panel) and around the vacancy.

Figure 2

Structural characterization of CdS thin films. (a) XRD spectra measured in grazing incidence geometry for CdS films (CdS_160709_5-8) grown at 70 °C with different thickness for a fixed bath chemistry (0.0015 M cadmium nitrate, 0.075 M thiourea, and 1.87 M ammonia). Purple lines and labeled peaks correspond to the cubic sphalerite phase (ICDD #96-900-0109), and green lines to the hexagonal wurtzite phase (ICDD #04-004-8895). (b) TEM micrograph of a representative sample (CdS_160704_2) shows that it is well crystallized.

Figure 3

Variation of photoconductivity sensitivity and decay rate with CBD bath chemistry using AM1.5 illumination. (a) Varying cadmium source concentration. (b) Varying thiourea (TU) concentration. (c) Varying ammonia concentration. These and other studies point to the ratio of cadmium-to-sulfur activity in the bath as the determining factor for photoconductivity. Note that a decay rate approaching zero can result when the film is not photosensitive, e.g., for high [TU]. (d) Decay rate and sensitivity are negatively correlated for photosensitive films. (e) Relationship between photosensitivity and Cd/S ratio in CdS films. Error bars represent 68% confidence intervals from the counting statistics of the XRF measurements.

Figure 4

Photoconductivity excitation spectra for a representative sample with large photoresponse (CdS_160613_4). Measured current is normalized to the incident radiant power density.

Figure 5

Analysis of photoconductivity decay for a representative sample with large photoresponse (CdS_160613_4). (a) Bottom axis, blue curve: photoconductivity time-series data measured using a white LED, which was turned on at time $t=1\mathrm{h}$ and off at time $t=4\mathrm{h}$. Top axis, black curve: power-law dependence of steady-state photoconductivity on light intensity, measured using a 455 nm LED. (b) Stretched exponential fits (blue lines) to short-time decay data measured at different temperatures. (c) Arrhenius plot; red points are data, blue line is Arrhenius fit. (d) Probability distribution $\left[p\left(E_a\right)\right]$ of activation energy for short- and intermediate-time decay processes.

Figure 6

The power-law relationship ${\sigma }_{\mathrm{PC}}\propto I^b$ between the steady-state photoconductivity (${\sigma }_{\mathrm{PC}}$) and light intensity ($I$) measured at different temperatures for a representative sample with large photoresponse (CdS_160613_4). The data at 25 °C (black curve) is the same as plotted in Fig. 5. At low intensity $b$ is between 0.3 and 0.5. At higher illumination $b$ is between 0.6 and 0.7. Data were measured using a 455 nm LED.

Figure 7

Evidence for thermally activated recombination in the temperature dependence of photoconductivity for a representative sample with large photoresponse (CdS_160613_4). (a) Temperature dependence of steady-state conductivity in the dark and under illumination using the white LED. (b) Transient photoconductivity measured at different temperatures. The light is switched on at 1 h and off at 4 h. The transient under illumination shows a fast rise with time constant ${\tau }_1$ and a slow decay with time constant ${\tau }_2$. The decay with the illumination off has characteristic time constant ${\tau }_3$. (c) Positive correlation between time constants ${\tau }_1$ and ${\tau }_3$. (d) Numerical simulations of the standard model, showing how transient behavior changes with temperature.

Figure 8

Energy transition level diagrams. (a) Standard model of photoconductivity including trap and recombination levels. (b) Energy level diagram for photoconductive CdS at equilibrium in the dark. (c) One possibility for energy level diagram for CdS under illumination.