Thin-film $(\mathrm{Sb},\mathrm{Bi}{)}_2{\mathrm{Se}}_3$ Semiconducting Layers with Tunable Band Gaps Below 1 eV for Photovoltaic Applications

Polycrystalline $(\mathrm{Sb},\mathrm{Bi}{)}_2{\mathrm{Se}}_3$ thin-film semiconductors are grown by coevaporation with a subsequent annealing process. It is shown that $\mathrm{Bi}$ can be incorporated into the ${\mathrm{Sb}}_2{\mathrm{Se}}_3$ lattice, substituting up to approximately 60% of the $\mathrm{Sb}$ atoms, while maintaining the orthorhombic crystal structure. Upon $\mathrm{Bi}$ substitution, the lattice expands mainly along the one-dimensional $(\mathrm{Sb},\mathrm{Bi}{)}_4{\mathrm{Se}}_6$ ribbons. In addition, the band gap decreases with a direct (indirect) band gap of 0.891 eV (0.864 eV) for a $({\mathrm{Sb}}_{0.4},{\mathrm{Bi}}_{0.6}{)}_2{\mathrm{Se}}_3$ thin film. A photovoltaic device based on a $(\mathrm{Sb},\mathrm{Bi}{)}_2{\mathrm{Se}}_3$ absorber is fabricated that displays an open-circuit voltage of 133 mV and a short-circuit current density of $18.4\mathrm{mA}/{\mathrm{cm}}^2$, demonstrating the potential of this material for infrared detection or multijunction solar-cell applications.

Figure 1

SEM micrographs of annealed $({\mathrm{Sb}}_{1-x}{\mathrm{Bi}}_x{)}_2{\mathrm{Se}}_3$ thin films on a soda lime glass substrate for different compositions in top view (top) and in the cross-section view (bottom). Note the different scale for the sample without any $\mathrm{Bi}$ content (x = 0). All films crystallize in the orthorhombic structure. Grain size seems to decrease with increasing $\mathrm{Bi}$ content.

Figure 2

(a) Lattice parameters a, b, and c of the orthorhombic Pnma structure of ${\mathrm{Sb}}_2{\mathrm{Se}}_3$ with respect to the incorporation of $\mathrm{Bi}$. Data from the literature are added from Wang et al. [20]. Open circles with crosses indicate two-phase systems, where the orthorhombic and the rhombohedral phases are observed. Open star at x = 0.6 marks the maximum $\mathrm{Bi}$ concentration with a pure orthorhombic phase. Light-gray region indicates the confidence in the composition for that data point. (b) Coherence length determined from the XRD peak broadening. Solid gray line is a guide to the eye for the coherence length for layers grown on SLG, annealed std (open gray circles).

Figure 3

XRD diffractograms for a $({\mathrm{Sb}}_{1\text{−}x}{\mathrm{Bi}}_x{)}_2{\mathrm{Se}}_3$ layer grown on SLG as grown (a), after the standard annealing process (350 °C) (b), and after a different annealing process at a nominal temperature of 360 °C (c). XRD diffractograms of the films shown in (a) and (b) need to be fitted including the rhombohedral phase, due to the (0 0 6) reflection between 18.4° and 18.8° (d). Film annealed at a nominally higher temperature does not show the rhombohedral (0 0 6) reflection, and thus, indicates an orthorhombic single phase. Vertical dashed lines show the position of the (0 0 6) reflection, according to ICDD PDF Card No. 00-033-0214. Fig. S1 within the Supplemental Material [30] shows the graph over the full $2\theta$ range from 10° to 90°.

Figure 4

(a) Raman spectra measured at room temperature for samples annealed with the standard process on SLG substrates. Spectrum for the sample with $x=1.0$ belongs to a sample in the as-grown state (not annealed) and serves to indicate the peak positions for the rhombohedral ${\mathrm{Bi}}_2{\mathrm{Se}}_3$ phase. Shifts of the three peak positions are indicated as dashed lines in (a) and are separately plotted in (b). Solid line represents a linear fit to the peak positions of P3.

Figure 5

(a) Absorption coefficient calculated according to Eq. (1) for the samples on SLG with the annealed std process. For higher $\mathrm{Bi}$ contents, the absorption edge redshifts, indicating a smaller band gap. (b) Normalized photoluminescence spectra for the same samples presented in (a), showing the redshift of the band gap as well. Samples are identical to those used for Raman measurements (see Fig. 4 for details on compositional values).

Figure 6

Band-gap variation upon $\mathrm{Bi}$ incorporation. Direct and indirect band gaps are estimated from spectrophotometry. Solid lines represent an orthogonal distance fit. Peak positions from PL measurements are redshifted by 70–100 meV with respect to the indirect and direct band gaps due to heating of the laser. Open green square represents the position of the PL peak maximum measured at room temperature and a laser power of 10 mW. Band-gap data from the literature are taken from Ref. [19].

Figure 7

(a) Current density-voltage characteristics in the dark and under one-sun illumination for a photovoltaic device with a $({\mathrm{Sb}}_{1\text{−}x}{\mathrm{Bi}}_x{)}_2{\mathrm{Se}}_3$ absorber with x = 0.431. (b) External quantum efficiency measured for several bias voltages. Linear extrapolation of long-wavelength response yields a band gap of 0.933 eV, as expected for the composition x = 0.431.