Enhancing the solar energy conversion efficiency of solution-deposited Bi₂S₃ thin films by annealing in sulfur vapor at elevated temperature

Abstract

Bi₂S₃ is a non-toxic n-type semiconductor, which has been commonly synthesized in the form of quantum dots or nanocrystalline films by solution deposition methods. Despite a favorable optical band gap of ∼1.3 eV, such films have not achieved high solar energy conversion efficiencies to date. We hypothesize that this is in part due to the presence of sulfur vacancies that, according to our density functional theory calculations, form a deep trap state in the band gap of Bi₂S₃, which can act as a strong recombination channel for photoexcited charges. Here, we report a microcrystalline Bi₂S₃ thin-film synthesized by annealing solution-deposited nanocrystalline Bi₂S₃ in a sulfur vapor environment at 445 °C, which simultaneously increases the grain size and phase purity of Bi₂S₃, fills in sulfur vacancies, and improves optical absorption. Time-resolved terahertz spectroscopy (TRTS) reveals that sulfur annealing increases the photoexcited carrier lifetime from sub-picosecond to ∼30 picoseconds, while the internal quantum efficiency of a photoelectrochemical solar cell device is increased 4-fold from ∼10% to ∼40%. In addition, TRTS reveals that the intra-grain carrier mobility in the sulfur-annealed films is ∼165 cm² V⁻¹ s⁻¹ and the long-range mobility is ∼111 cm² V⁻¹ s⁻¹ at short times, indicating that carriers are able to hop across grain boundaries. These results indicate that annealing in sulfur vapor can produce simultaneously high light absorption and charge separation efficiencies by achieving carrier diffusion length that is comparable to the light absorption depth, leading to high solar energy conversion efficiencies in Bi₂S₃.

---

Introduction

Solar energy is a sustainable option to satisfy the increasing world energy demand and to reduce CO₂ emission caused by burning fossil fuels.¹ Bi₂S₃, with a band gap of ∼1.3 eV, is a promising material for efficient generation of electricity or chemical fuels from sunlight.^2–13 It has been synthesized as either powder or thin film using techniques such as hydrothermal synthesis,^3,8,12,14 electrochemical deposition,¹⁵ chemical bath deposition,^16,17 vapor deposition,¹⁸ and successive ionic layer adsorption and reaction (SILAR).^4–6,9,13 Among these methods, SILAR is one of the most common due to its facile processing, versatile applications to different surfaces and nanostructures, and tunable coating thickness and packing density. For a typical SILAR deposition of Bi₂S₃, the substrate is successively immersed into separate solutions of Bi(NO₃)₃ as the cation precursor and Na₂S as the anion source, with intermediate washing procedures. However, Bi₂S₃ synthesized by SILAR consists of nanocrystalline films or quantum dots. The presence of a large number of grain boundaries, which are known to introduce carrier trap sites, negatively impacts the optoelectronic performance of the material. Additionally, the solution-deposited film without annealing has poor electric contact at the semiconductor/substrate interface, which leads to high resistance at the heterojunction.¹² To overcome these issues, some studies reported annealing of the solution-deposited Bi₂S₃ in argon at relatively low temperatures (<300 °C), as a post-synthesis method to increase the grain size to ∼20 nm.^5,6 Another study reported sulfurizing of Bi(NO₃)₃ in elemental sulfur vapor at 180 °C instead of Na₂S at room temperature to prevent possible sodium contamination and produced Bi₂S₃ quantum dots with an average diameter of 6 nm.¹³ These low-temperature annealing processes did not lead to highly crystalline Bi₂S₃. Therefore, it is necessary to explore alternative annealing methods to improve both crystallinity and phase purity of the solution-deposited Bi₂S₃.

In this work, we synthesize microcrystalline Bi₂S₃ thin films by high-temperature annealing of solution-deposited nanocrystalline Bi₂S₃ in a sulfur vapor environment. The sulfur annealing increases grain size and phase purity, fills in sulfur vacancies, and improves optical absorption. Time-resolved terahertz spectroscopy (TRTS) reveals that sulfur annealing increases the photoexcited carrier lifetime from sub-picosecond to ∼30 picoseconds, while the internal quantum efficiency of a photoelectrochemical (PEC) solar cell device is increased from ∼10% to ∼40%. In addition, TRTS reveals that the intra-grain carrier mobility in the S-annealed films is ∼165 cm² V⁻¹ s⁻¹ and the long-range mobility is ∼111 cm² V⁻¹ s⁻¹ at short times, indicating that carriers are able to hop across grain boundaries. These results indicate that annealing in sulfur vapor can produce simultaneously high light absorption and charge separation efficiencies by achieving a carrier diffusion length that is comparable to the light absorption depth, leading to high solar energy conversion efficiencies in Bi₂S₃.

Experimental and theoretical methods

Bi₂S₃ thin films were synthesized by a combination of solution deposition and sulfur vapor annealing. First, Bi₂S₃ thin films were deposited onto fluorine-doped tin oxide (FTO) substrates (2.5 × 1.5 cm, 2.2 mm thick, TEC 7, Hartford Glass) by spin-coating a bismuth nitrate solution as the Bi³⁺ source and reacting with a sodium sulfide solution to form Bi₂S₃. The bismuth precursor was prepared by dissolving 0.485 g of Bi(NO₃)₃·5H₂O (98%, Sigma Aldrich) in 10 mL acetic acid (≥99.7%, Sigma Aldrich). The sodium sulfide solution was prepared by dissolving 0.033 g of Na₂S (anhydrous, Sigma Aldrich) in 35 mL methanol. Each layer of Bi₂S₃ was first spin-coated onto FTO using 150 μL of the Bi(NO₃)₃·5H₂O solution at a spin speed of 2000 rpm for 40 s, then immersed in the Na₂S solution for 2 min, thoroughly washed with methanol, and completely dried under compressed air.^{11,13–15,18} After 5, 10 or 15 layers of Bi₂S₃ were coated, the samples were annealed in sulfur vapor in a tube furnace (Lindberg/Blue M 1100 °C, Thermo Fisher Scientific) equipped with a 1 inch diameter quartz tube (Quartz Scientific). Sulfur vapor was generated by subliming sulfur powder (2.5 g, 99.5%, Sigma Aldrich) at ∼110 °C in the upstream section of the tube furnace. Argon (99.995% purity, Praxair, 80 sccm flow rate) was used to convey the sulfur vapor to the downstream substrate. The substrates coated with solution-deposited Bi₂S₃ films were placed inside the hot zone at 400 °C, 445 °C and 470 °C, respectively. The annealing was conducted at atmospheric pressure for ∼60 min.

The morphologies, crystal structures, and chemical compositions of the Bi₂S₃ thin films were characterized by scanning electron microscopy (SEM, JEOL 7000F, 10 kV), transmission electron microscopy (TEM, JEOL 2010F, 200 kV), parallel beam X-ray diffraction (XRD, PANalytical Empyrean, Cu-Kα, 45 kV, 40 mA), and X-ray photoelectron spectroscopy (XPS, PHI 5600, Al-Kα, 13.5 kV, 300 W). The crystallite size of the un-annealed Bi₂S₃ nanocrystals was determined by TEM imaging of a single layer of Bi₂S₃ deposited directly onto a TEM grid using the above-described solution deposition method. The TEM grid consisted of ultrathin carbon film supported by a lacey carbon film on a 400 mesh copper grid (Ted Pella). The average crystallite size of the sulfur-annealed films was calculated from the measured XRD pattern using the Scherrer equation:¹⁹

where λ is the X-ray wavelength, β is the measured width of the peak at half-maximum intensity in radians, and θ_B is the Bragg angle.

The wavelength-dependent optical absorption properties of the samples were obtained using illumination from a Xe lamp (Model 66902, Newport). Two spectrometers (USB 2000+ and Flame-NIR, Ocean Optics) were used to measure the incident, transmitted and reflected light at UV-visible and near-infrared regions, respectively. Bi₂S₃ thin films were prepared on quartz slides (1 mm thick, Ted Pella) for the optical measurements to minimize diffuse scattering by FTO substrates. For both the transmission and reflection measurements, light was incident at a 45° angle to the back-side (quartz) surface of the sample. For the transmission measurements, the spectrometers were aligned with the incident light to capture the transmitted light (T). For the reflection measurements, the spectrometers were placed at a 90° angle to the incident light to capture the reflected light (R). The absorption efficiency was calculated using

A(λ) = 100% − T(λ) − R(λ). (2)

The light absorption depth (δ) was then calculated as

where α is the absorption coefficient and z is the film thickness measured from cross-section SEM images.

A PHI5600 XPS system acquired all photoelectron spectra. The instrument utilized a monochromated K_α Al source and a third-party data acquisition system (RBD Instruments, Bend Oregon). Analysis chamber base pressures were <1 × 10⁻⁹ torr. A hemispherical energy analyzer that was positioned at 90° with respect to the incoming X-ray flux and 45° with respect to standard sample positioning collected the photoelectrons. High-resolution XP spectra employed a 23.5 eV pass energy, 25 meV step size, and a 50 ms dwell time per step. High-resolution spectra quantified Bi 4f, S 2s (due to the spectral overlap and difficulty in resolving S 2p from Bi 4f), O 1s, and C 1s regions of the photoelectron spectrum. The XPS spectra were calibrated based on a binding energy of 284.8 eV for adventitious carbon. Post-acquisition data fitting of the spectral features utilized a Shirley-shaped background and GL(70) functional peak shapes for Bi 4f spectra while fits to the S 2s region utilized a linear background and GL(30) functional peak shapes.^20,21 The Bi 4f doublets were assigned identical full width at half maximum (FWHM) peak widths with areas for 4f_5/2 peaks containing 75% of the area of 4f_7/2 peaks.²² To prevent mathematically optimized but physically unrealistic peak widths, all features in the S 2s spectral region were constrained to identical FWHM values.

The photoconductivity and photoexcited carrier lifetime were analyzed using the TRTS technique. The charge carriers were excited in the sample with an ultrafast optical pump pulse with photon energy above the band gap, and the transient photoinduced conductivity was monitored in transmission using a time-delayed THz pulse. Here we used 100 fs duration, 400 nm pulses derived from a 1 kHz amplified Ti:sapphire laser source for excitation. Absorption of both un-annealed and S-annealed Bi₂S₃ films was similarly high at 400 nm. Thus, comparing transient photoconductive response at this excitation wavelength allows us to focus exclusively on the effect of annealing on microscopic conductivity and carrier lifetime. THz probe pulses were generated by optical rectification of 800 nm, 100 fs pulses from the same laser source in a [110] ZnTe crystal, and coherently detected by free-space electro-optic sampling in a second [110] ZnTe crystal.

The PEC measurements were performed in a three-electrode configuration, using a potentiostat (Model SP-200, BioLogic) under back-side broadband illumination from a Xe lamp. The incident light intensity from the Xe lamp at each wavelength was measured by a spectrometer. The integrated power of the Xe lamp output at wavelengths shorter than 950 nm (1.3 eV) was 81.7 mW cm⁻², as compared to 71.0 mW cm⁻² for the standard AM 1.5G spectrum (Fig. S1a†). Current density–voltage curves (J–V curves) were measured at a scan rate of 10 mV s⁻¹. J–V curves in aqueous electrolytes were measured in a three-electrode configuration with the Bi₂S₃ photoanode as the working electrode, a Pt wire (0.5 mm diameter, 99.99%, Sigma Aldrich) as the counter electrode, and a saturated calomel (SCE) reference electrode (Gamry). The aqueous electrolyte used was 0.3 M Na₂S electrolyte (pH ≈ 13). Potentials (in volts) in aqueous electrolytes are reported versus the reversible hydrogen electrode (RHE) using

V_RHE = V_SCE + 0.244 + [0.059 × pH]. (4)

The light absorption efficiency (η_abs) is the fraction of incident photons that are absorbed. The charge separation efficiency (η_sep) is the fraction of absorbed photons that reach the semiconductor/electrolyte interface. The product of light absorption efficiency (η_abs) and charge separation efficiency (η_sep) was calculated at each potential V by

where J_max is the maximum photocurrent density of the photoelectrode under the Xe lamp illumination, J_ph is the measured photocurrent density, and η_trans is the charge transfer efficiency at the semiconductor/electrolyte interface. The maximum photocurrent (J_max) for the Bi₂S₃ photoanodes was 43.5 mA cm⁻², which was obtained by integrating the photon flux from the Xe lamp at wavelengths shorter than 950 nm (1.31 eV). The surface charge transfer efficiency, η_trans, is assumed to be ∼100% for PEC measurements in a Na₂S electrolyte due to the fast kinetics of sulfide oxidation.

The incident photon-to-current efficiency (IPCE), also known as external quantum efficiency (EQE), was measured at 0.6 V_RHE, which is the onset potential for sulfide oxidation in the dark. A Xe lamp equipped with a monochromator (Cornerstone 130 1/8 m, Newport) was used as the illumination source. The spectral irradiance of monochromatic light at each wavelength was measured by a spectrometer and reported in Fig. S1b.† The IPCE was calculated using

where J_ph is the measured photocurrent density in mA cm⁻², P_mono is the intensity of the incident monochromatic light in mW cm⁻², and λ is the wavelength of the monochromatic light in nm. The absorbed photon-to-current efficiency (APCE), also known as internal quantum efficiency (IQE), was then calculated by

Density functional theory (DFT) calculations were implemented by the Vienna ab initio simulation package (VASP) code.^23,24 We used the generalized gradient approximation (GGA) exchange and correlation functionals as parameterized by Perdew, Burke, and Ernzerhof (the PBE functional).^25,26 The electron–ion interactions were treated within the framework of the standard frozen-core projector augmented-wave (PAW) method with valence configurations of 6s²6p³5d¹⁰ for Bi and 3s²3p⁴ for S.^27,28 An energy cut-off of 400 eV was used in the plane-wave basis-set expansion. Gaussian smearing with a width of 0.2 eV was used for ionic relaxation and the tetrahedron method with Blöchl corrections was used for density of states (DOS) calculations. The Grimme D3 correction method was used to account for dispersion interactions between layers of Bi₂S₃.²⁹ For calculations of the pristine Bi₂S₃ bulk unit cell (1 × 1 × 1), a 6 × 2 × 2 Monkhorst–Pack³⁰k-point sampling was used for ionic relaxation. For defect calculations, a bulk supercell (3 × 1 × 1) containing 60 atoms was constructed and spin-polarized DFT calculations were performed (Fig. S2†). For calculations of the bulk supercell, a 2 × 2 × 2 k-point sampling was used for ionic relaxation and a higher 8 × 8 × 8 k-point sampling was used for density of states calculations. Electronic band structure calculations were performed with 50 k-points in each high symmetry direction within reciprocal space of the crystal. Defect formation energies were calculated by

ΔE_defective = E_defective − E_{stoichiometric} + ∑n_defectμ_defect, (8)

where E_defective and E_{stoichiometric} are the ground-state energies of Bi₂S₃ with and without defect, respectively, n_defect is the number of atoms removed (added) from (to) the system to form the defect, and μ_defect is the chemical potential of the atom that is removed (added). n_defect = 1 if an atom is removed from the system, whereas n_defect = −1 if an atom is added to the system. The chemical potentials of S, O, and H atoms were calculated from the ground-state energies of S₈, O₂, and H₂ molecules, respectively. For visualization of the crystal structures, VESTA (Visualization for Electronic and Structural Analysis)³¹ was used.

Results and discussion

Both the un-annealed and sulfur-annealed Bi₂S₃ films were ∼150 nm thick (Fig. 1a and b). The crystallite size of the un-annealed Bi₂S₃ nanocrystals was measured to be ∼10 nm by TEM (Fig. 1c). Because of the small size of the Bi₂S₃ nanocrystals, the XRD pattern of the un-annealed Bi₂S₃ film (Fig. 1d) shows peaks only from the underlying FTO substrate, and no peaks indexed to Bi₂S₃. This is in contrast to a previous report, in which XRD peaks from a NaBiS₂ impurity phase were observed in the Bi₂S₃ synthesized by SILAR from saturated precursor solutions at 80 °C.¹³ On the other hand, annealing the solution-deposited Bi₂S₃ film in sulfur vapor at 445 °C results in a XRD pattern with peaks that can be indexed to pure orthorhombic Bi₂S₃ (ICDD PDF 04-014-6675) (Fig. 1d). The average crystallite size of the S-annealed Bi₂S₃ film was calculated to be ∼45 nm from the (112) reflections of the XRD pattern, and SEM micrographs show that the S-annealed film contains Bi₂S₃ grains with an average diameter that is very similar to this average crystallite size (Fig. 1b). Therefore, sulfur vapor annealing increases the crystallite size of the solution-deposited Bi₂S₃ from ∼10 nm to ∼45 nm.

Fig. 1 Top-view and cross-section SEM images of 10 layers of (a) un-annealed and (b) S-annealed Bi₂S₃ thin films on FTO substrate. (c) TEM image of un-annealed Bi₂S₃ nanocrystals. The background is the amorphous carbon film of the TEM grid. (d) XRD patterns of un-annealed and S-annealed Bi₂S₃ thin films.

The light absorption efficiency of the Bi₂S₃ thin film significantly increases at longer wavelengths after sulfur vapor annealing (Fig. 2a). From the measured optical absorption spectra, the indirect band gaps can be determined from the (αhν)^1/2vs. hν Tauc plot (Fig. 2b). The S-annealed Bi₂S₃ was thus determined to possess an indirect band gap of ∼1.24 eV. The experimentally determined band gap corresponds well to the theoretically predicted fundamental band gap, which is found to be indirect and occurs in the ΓX region of the Brillouin zone,³² with energy of 1.25 eV (Fig. 2c). However, the un-annealed Bi₂S₃ appears to have a larger band gap of ∼1.37 eV, which may be due to quantum confinement caused by nanoscale grain size. The S-annealed Bi₂S₃ has higher absorption at longer wavelengths (red-shifted) due to the increased crystallite size and lack of quantum confinement. The sub-bandgap absorption of the un-annealed Bi₂S₃ reaches a minimum value of ∼2.6% at 950 nm (1.31 eV), which suggests that the diffuse scattering by quartz substrate is negligible. Moderate sub-bandgap absorption is observed for the S-annealed Bi₂S₃, which is attributed to electronic transitions from the defects states to the conduction band, as will be discussed later along with IPCE. Overall, the S-annealed films absorb 67.0% of above-gap photons, compared to only 50.1% for the un-annealed films, despite the films having nearly identical thickness.

Fig. 2 (a) Optical absorption efficiencies. (b) Tauc plots to determine the indirect band gaps of 10 layers of un-annealed and S-annealed Bi₂S₃ thin films. (c) DFT electronic band structure of pristine Bi₂S₃.

X-ray photoelectron spectroscopy was used to quantify the chemical compositions of both un-annealed and S-annealed Bi₂S₃ films. Fig. 3 reports the XP spectra for the (a) Bi 4f and S 2p regions and the (b) S 2s region of the photoelectron spectra both for (top) un-annealed and for (bottom) S-annealed Bi₂S₃ films. Prior studies ascribed peaks at 158.9 eV to the Bi 4f_7/2 feature from Bi₂S₃ and 159.3 eV to the corresponding feature from Bi₂O₃.³³ For the Bi 4f region of the un-annealed film (Fig. 3a, top), fits reveal a feature at 158.6 eV (red) and a feature at 159.4 eV (blue). In agreement with prior studies, we ascribe the fit of the red doublet to Bi₂S₃ and the fit of the blue doublet to Bi₂O₃ in both the un-annealed and sulfur-annealed Bi 4f spectra. Notably, both Bi₂S₃ and a large amount of Bi₂O₃ are present on the surface of the un-annealed Bi₂S₃, which we interpret to indicate film oxidation either during the solution deposition process or by exposure to the air ambient. For the S 2s region of the un-annealed film (Fig. 3b, top), fits are dominated by a feature at 225.8 eV (red) that we attribute to S²⁻ in the Bi₂S₃. Additionally, the S 2s spectrum for the un-annealed film contains a smaller feature at 228.1 eV (green) that we attribute to neutral sulfur and two features above 231 eV (blue) that we attribute to highly oxidized sulfur species, denoted SO_x. The spectra at the bottom of Fig. 3 show each respective region following the sulfur annealing. The Bi 4f region of the S-annealed film (Fig. 3a, bottom) demonstrates significantly smaller Bi 4f features due to bismuth oxide as compared to the oxide features in the un-annealed film. Concomitant with the sulfur treatment, the S 2s region in the bottom of Fig. 3b demonstrates higher concentrations of the neutral sulfur (S⁰) relative to the un-annealed film. We attribute the excess S⁰ feature to trace sulfur from the annealing process that may be incorporated as interstitial sulfur in the Bi₂S₃ crystal and/or residual surface sulfur. Interestingly, the sulfur annealing step did not attenuate the SO_x features (blue) relative to the un-annealed film.

Fig. 3 X-ray photoelectron spectra of Bi₂S₃ thin films without annealing (top) and with sulfur vapor annealing at 445 °C (bottom). (a) Bi 4f and S 2p XP spectra and (b) S 2s XP spectra. Red traces correspond to fitted peaks ascribed to Bi₂S₃ for both the Bi 4f and S 2s spectral regions. Blue traces represent fitted peaks ascribed to oxidized species Bi₂O₃ and highly oxidized sulfur, denoted SO_x. Comparing the spectra of the un-annealed sample to the sulfur-annealed sample, the sulfur annealing dramatically decreased the quantity of bismuth oxide at the surface. However, the sulfur annealing did not remove oxidized sulfur species and somewhat increased the concentration of neutral sulfur species (green). The scale bar represents 2000 counts per second for the Bi 4f and S 2p spectra and 450 counts per second for S 2s.

The influence of defects on the electronic structure was then analyzed by calculating the DOS of pristine and defect-containing Bi₂S₃, along with their formation energies (Fig. 4 and S3†). We hypothesize that the sulfur vapor annealing fills in sulfur vacancies of the un-annealed Bi₂S₃. The defect states related to a sulfur vacancy (S_v) are occupied electronic states found deep in the band gap, at 0.63 eV above the valence band maximum (VBM). This finding is consistent with previous reports that S_v creates deep hole trapping states that allow electron–hole recombination.^34–37 Additionally, Bi₂S₃ containing two S_v possesses a higher DOS population for the mid-gap charge trapping states, which further shows that S_v in Bi₂S₃ can act as recombination sites (Fig. 4). Therefore, the un-annealed Bi₂S₃ may contain a higher concentration of mid-gap S_v states, which increases recombination and decreases photoexcited carrier lifetime. Moreover, the S-annealed Bi₂S₃ likely contains more sulfur interstitials (S_i) due to the low formation energy of this defect in a S-rich environment,³⁶ which is consistent with the greater amount of elemental sulfur species observed on the surface of the film by XPS. However, S_i only creates shallow hole trapping states that are about 0.12 eV above the VBM. In contrast to the S_v defect states, the trapped holes at S_i are less likely to recombine with the photoexcited electrons due to the smaller energy required to remove the holes to the valence band and avoid recombination (Fig. 4). On the other hand, the un-annealed Bi₂S₃ likely contains more oxygen impurities in the forms of oxygen substituting for sulfur (O_S) and oxygen interstitial (O_i), the formation of which is thermodynamically favorable due to the negative formation energies, −0.93 and −0.57 eV, respectively (Fig. S3†). O_S–Bi₂S₃ exhibits similar electronic structure to pristine Bi₂S₃. This is expected, since oxygen has the same number of valence electrons as sulfur and is more electronegative. O_i–Bi₂S₃ has a similar DOS to S_i–Bi₂S₃, resulting in shallow electronic states 0.12 eV above the VBM. Therefore, the oxide impurities in the un-annealed Bi₂S₃ seem to only cause shallow defect states, which may not contribute to the poor performance of the film.

Fig. 4 DFT-calculated density of states (DOS) and formation energies of pristine Bi₂S₃ and Bi₂S₃ containing sulfur vacancy (S_v), two sulfur vacancies (2S_v), and sulfur interstitial (S_i). Dark grey regions show the occupied electronic states, while light grey regions show the unoccupied electronic states. The valence band maximum (VBM) energy is set to zero eV, and is defined as the highest occupied non-defect state.

We then analyzed the impact of sulfur annealing on photoconductivity and photoexcited carrier lifetime using TRTS. We interrogated the pump-induced changes in the sample conductivity with sub-picosecond time resolution by varying the delay between the pump pulse and THz probe pulse. The change in transmission of the main peak of the THz probe pulse, –ΔT/T, as a function of time after excitation with 378 μJ cm⁻² pulse, for both un-annealed and S-annealed films is shown in Fig. 5a. It is proportional to the transient photoconductivity.^38–40 At the same excitation conditions, peak photoinduced conductivity is more than twice higher in the S-annealed film as a result of a less prevalent fast trapping and recombination of photoinjected charge carriers on a timescale that is shorter than our instrumental response time of ∼300 fs. Moreover, photoconductivity of un-annealed film is very short-lived and decays on sub-picosecond time scale due to both rapid carrier recombination inside the small Bi₂S₃ crystallites and fast trapping of photoexcited carriers in trap states at crystallite boundaries. Larger grain size, lower concentration of sulfur vacancies, and reduced number of interface defects in S-annealed Bi₂S₃ film translates into a significantly longer lifetime of mobile carriers (Fig. 5a). Photoconductivity in the S-annealed sample follows a bi-exponential decay with a fast (∼3 ps) and a slower (30 ps) component. The fast decay time is dependent on excitation fluence, increasing from 2.6 ps to 4.4 ps as excitation is decreased 6-fold to 63 μJ cm⁻² (Fig. S4†). This suggests that the process responsible for this fast photoconductivity decay is recombination of mobile carriers. On the other hand, the slower, 30 ps component is fluence-independent and represents trapping of carriers at sulfur vacancies and interface states that do not become saturated in the studied fluence range.

Fig. 5 Time-resolved THz spectroscopy of un-annealed and S-annealed Bi₂S₃ thin films. (a) Transient photoconductivity (lines) and instantaneous photoexcited carrier density (symbols) for S-annealed film. Smooth lines represent fits of experimental transient photoconductivity to single- (un-annealed) and bi-exponential (S-annealed film) decays. (b) Real and imaginary components of transient photoconductivity at different times after optical excitation for S-annealed film. Symbols represent experimental data, and lines – fits of experimental data to the Drude–Smith conductivity. (c) Drude–Smith c parameter for S-annealed film.

Fixing the pump-probe delay, and detecting the pump-induced changes in the amplitude and phase of the THz pulse transmitted through the sample allows us to extract the complex-valued, frequency-resolved THz photoconductivity spectrum at a given time following the photoexcitation. Fig. 5b shows the real (σ₁) and imaginary (σ₂) conductivity of the S-annealed Bi₂S₃ film at 2, 3, 5 and 10 ps after excitation with a 378 μJ cm⁻² pulse. The real component of conductivity decreases with time while the imaginary one stays almost unchanged and close to zero. We analyze this progression by fitting both real and imaginary conductivity to the Drude–Smith model, a phenomenological model of microscopic THz conductivity in granular materials where the grain size is comparable to the carrier mean free path.^38,41–48 It allows the extraction of the instantaneous photoexcited carrier density N, the effective scattering time τ_DS, and a measure of carrier localization within individual grains at a given time after excitation:

where the carrier effective mass, m*, is 0.246m_e for Bi₂S₃.⁴⁹ Degree of carrier localization is quantified by the parameter c, with c = 0 corresponding to free carriers and c = −1 corresponding to full localization of the charge carriers by the grain boundaries. For fully percolated systems, the Drude–Smith c parameter can be interpreted as the probability of carrier reflection at a grain boundary.⁴³ The effective scattering time τ_DS is determined by both intrinsic, bulk scattering time and the average time between collisions with the grain boundaries τ_boundary as^43,50

1/τ_DS = 1/τ_bulk + 1/τ_boundary. (10)

Lines in Fig. 5b are Drude–Smith fits to the experimental complex conductivity. The resulting carrier density is plotted as circles in Fig. 5a, and it follows the same trend as the bi-exponential decay of photoconductivity, indicating that trapping and recombination of mobile carriers is responsible for the observed transient reduction in photoconductivity while the intrinsic carrier mobility stays unchanged. Effective carrier scattering time τ_DS is unchanged over the first 10 ps and equal to 23 ± 5 fs, corresponding to the carrier mean free path of ∼3 nm. As the mean free path is significantly smaller than the ∼45 nm average grain size, contribution of grain boundary scattering to the effective relaxation time is negligible, and intrinsic, intra-grain mobility (μ_int) can be estimated as

However, long-range, inter-grain mobility decreases as carriers, while free to move within the individual grains, becomes more localized inside those grains. This phenomenon is reflected in the time dependence of the c parameter, which changes from ∼−0.43 ± 0.04 at 2 ps to −0.63 ± 0.04 at 10 ps after excitation, as has been observed in other polycrystalline systems.^38,43 There are two possible explanations for this behavior, which we are not presently able to distinguish between. One possibility is that, as carriers get trapped at inter-grain boundaries, the electrostatic field associated with those carriers increases the height of the potential barrier experienced by the free carriers.⁵¹ Another possible explanation is that the semiconductor goes towards a flat-band condition under illumination due to the large concentration of photoexcited carriers, resulting in a decrease in an already-existing potential barrier at the grain boundaries, and that this potential barrier subsequently returns to its original height as the photoexcited carriers recombine.⁵² Regardless of the mechanism, the long-range, dc-mobility of the film, μ_dc = (1 + c)μ_int decreases as a function of time after excitation from 111 cm² V⁻¹ s⁻¹ at 2 ps to 61 cm² V⁻¹ s⁻¹ at 10 ps. Based on a carrier lifetime (t) of 30 ps and an inter-grain mobility (μ_dc) of 61 cm² V⁻¹ s⁻¹, the carrier diffusion length (L_D) can be calculated as

This carrier diffusion length is similar to the light absorption depth at ∼700 nm (Fig. S5†), which suggests that simultaneously high light absorption and charge separation efficiencies can be achieved in the S-annealed Bi₂S₃ thin films, leading to high overall solar energy conversion efficiencies.⁵³ On the other hand, for un-annealed Bi₂S₃ films, low signal/noise ratio in the complex conductivity spectra did not permit accurate measurement of mobility.

The solar energy conversion efficiency of the 10 layers of Bi₂S₃ thin films with and without sulfur vapor annealing at 445 °C was evaluated by photoelectrochemical (PEC) measurements of the wavelength-dependent photon-to-current efficiencies and the potential-dependent light absorption and charge separation efficiencies (Fig. 6). The film thickness and annealing temperature were optimized for maximum photocurrent under white light illumination (Fig. S6†). The IPCE of the S-annealed films reaches a plateau of ∼40% at short wavelengths, compared to a plateau of ∼5% for the un-annealed films, and is higher than that of the un-annealed films at all wavelengths (Fig. 6a). The IPCE of the S-annealed films shows a sharp rise at wavelengths shorter than 1000 nm, which is consistent with the measured band gap of 1.25 eV, while the IPCE of the un-annealed films show a rise at wavelength shorter than 900 nm, which is consistent with the measured band gap of 1.37 eV. As mentioned earlier, the smaller band gap of the S-annealed Bi₂S₃ is consistent with the larger grain size of the films and a resulting lack of quantum confinement. In addition, IPCE shows sub-bandgap photon-to-current conversion up to 1100 nm for the S-annealed Bi₂S₃ film and up to 1050 nm for the un-annealed Bi₂S₃. The sub-bandgap IPCE for the un-annealed Bi₂S₃ is likely attributed to the electronic transition from the O_i defect states to the CBM (1.10 eV; see Fig. S3†) due to the expected abundance of O_i in the un-annealed film. The sub-bandgap IPCE for the S-annealed Bi₂S₃ is likely due to transition from the S_i defect states to the CBM (0.93 eV; see Fig. 4) due to the large amount of S_i expected in the film. Moreover, the APCE of the S-annealed films reaches a value of ∼50% at short wavelengths, compared to a maximum value of ∼15% for the un-annealed films, and is higher than that of the un-annealed films at all wavelengths (Fig. 6b). It should be noted that the measured quantum efficiencies do not represent the optimum performance of the Bi₂S₃ thin films as the applied voltage is lower than the built-in voltage that could be achieved in a photovoltaic device.¹¹ Additionally, the product of η_abs and η_sep is calculated from the J–V curves under white-light illumination (Fig. 6c), and reaches a value of ∼35% for the S-annealed films but only ∼5% for un-annealed films at 1 V_RHE. At the same voltage, η_sep (Fig. 6d) reaches a value of ∼55% for the S-annealed films, but only ∼10% for un-annealed films. The enhanced quantum efficiency and charge separation efficiency after sulfur annealing can be explained by an increased photoexcited carrier lifetime, consistent with the results of the THz spectroscopy characterization. Although the charge-carrier mobility was measured by THz spectroscopy only for the S-annealed films, it is very likely that the mobility is also higher than that for un-annealed films, further explaining the increased performance.

Fig. 6 (a) IPCE and (b) APCE of 10 layers of un-annealed and S-annealed Bi₂S₃ thin films measured at 0.6 V_RHE in 0.3 M Na₂S aqueous electrolyte at pH 13. (c) Products of η_abs and η_sep, and (d) η_sep of 10 layers of un-annealed and S-annealed Bi₂S₃ thin films at different potentials measured in 0.3 M Na₂S aqueous electrolyte.

The Bi₂S₃ thin films exhibit n-type conductivity according to these PEC measurements, which is consistent with other reports.^6,9,12 The n-type conductivity of Bi₂S₃ has been attributed to the presence of S_v and S_i donor defects.³⁶ However, for defect states to behave as effective donor levels, the energy difference between the defect states and the conduction band minimum (CBM) should be comparable to the thermal energy at room temperature (0.026 eV). Therefore, S_v and S_i defect states, which have energies that are 0.52 and 0.93 eV below the CBM, respectively, cannot act as effective donor levels, and cannot be responsible for the n-type conductivity of Bi₂S₃. We additionally calculated the formation energies and density of states for Bi₂S₃ containing hydrogen substituted for sulfur (H_S), or hydrogen interstitials (H_i). The DOS of H_S and H_i– Bi₂S₃ (Fig. 7) show occupied states within the conduction band of Bi₂S₃, which suggests that these hydrogen impurities can create effective donor levels in Bi₂S₃ without introducing mid-gap charge trapping states. In fact, hydrogen impurities have previously been to found to cause the n-type conductivity of other semiconductors, such as BiVO₄.⁵⁴ Moreover, the formation energy of H_i is negative and the formation energy of H_S is smaller than those of S_v and S_i, which indicates that the incorporation of hydrogen is energetically favorable. Hydrogen may be incorporated into Bi₂S₃via the decomposition of precursors during the solution deposition process and/or from water vapor in the annealing environment. Although this requires further investigation, we propose that hydrogen impurities may play a role in causing the n-type conductivity of Bi₂S₃.

Fig. 7 DFT-calculated density of states (DOS) and formation energies of Bi₂S₃ containing hydrogen substitution of sulfur (H_S) and hydrogen interstitial (H_i).

Conclusions

We have synthesized Bi₂S₃ thin film photoelectrode via a combination of solution deposition and sulfur vapor annealing. We have identified three major advantages of the annealing process. First, the sulfur vapor annealing improves crystallinity of the solution-deposited Bi₂S₃ nanocrystals by increasing the crystallite size from ∼10 nm to ∼45 nm. Second, sulfur vapor annealing converts the surface oxides of the un-annealed film to sulfides and thus improves the purity of the film. Third, sulfur vapor annealing may fill in sulfur vacancies of the un-annealed Bi₂S₃ to significantly reduce the concentration of mid-gap charge recombination sites, resulting in longer photoexcited carrier lifetimes and higher internal quantum efficiencies. Overall, the S-annealed Bi₂S₃ achieves carrier diffusion lengths that are comparable to the light absorption depth, which makes it promising for photovoltaic and photoelectrochemical energy conversion applications. The sulfur vapor annealing method could be utilized as a general approach to enhance the optoelectronic performance of solution-deposited metal sulfide materials. Moreover, DFT calculations show that hydrogen doping creates effective donor levels at the conduction band edge of Bi₂S₃, which should increase electron conductivity without introducing recombination sites. This direction can be explored in future experimental work.

Author contributions

Z. Z. and P. M. R. conceived the study. Z. Z. carried out the thin film synthesis, optical and PEC measurements, DFT calculations, XRD and XPS characterizations. S. K. I. and N. A. D. advised on the DFT calculations. K. K. and L. V. T. performed the TRTS characterization. A. D. C. and R. L. G. assisted with the XPS characterization. L. Z. performed the SEM and TEM characterizations. D. R. B. contributed to the design and interpretation of the experiments and simulations. Z. Z., R. L. G., L. V. T., N. A. D. and P. M. R. wrote the manuscript. All authors discussed the results and revised the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Z. Z. acknowledges support from the WPI Summer Undergraduate Research Fellowship (SURF).

References

1. N. S. Lewis, Toward Cost-Effective Solar Energy Use, Science, 2007, 315(5813), 798–801.
2. A. A. Tahir, M. A. Ehsan, M. Mazhar, K. G. U. Wijayantha, M. Zeller and A. D. Hunter, Photoelectrochemical and Photoresponsive Properties of Bi2S3 Nanotube and Nanoparticle Thin Films, Chem. Mater., 2010, 22(17), 5084–5092.
3. H. He, S. P. Berglund, P. Xiao, W. D. Chemelewski, Y. Zhang and C. B. Mullins, Nanostructured Bi₂S₃/WO₃ heterojunction films exhibiting enhanced photoelectrochemical performance, J. Mater. Chem. A, 2013, 1(41), 12826–12834.
4. Y.-C. Lin and M.-W. Lee, Bi₂S₃ Liquid-Junction Semiconductor-Sensitized SnO₂ Solar Cells, J. Electrochem. Soc., 2014, 161(1), H1–H5.
5. Q. Zeng, J. Bai, J. Li, Y. Li, X. Li and B. Zhou, Combined nanostructured Bi₂S₃/TNA photoanode and Pt/SiPVC photocathode for efficient self-biasing photoelectrochemical hydrogen and electricity generation, Nano Energy, 2014, 9, 152–160.
6. G. Ai, R. Mo, Q. Chen, H. Xu, S. Yang, H. Li and J. Zhong, TiO₂/Bi₂S₃ core-shell nanowire arrays for photoelectrochemical hydrogen generation, RSC Adv., 2015, 5(18), 13544–13549.
7. Y. Cao, M. Bernechea, A. Maclachlan, V. Zardetto, M. Creatore, S. A. Haque and G. Konstantatos, Solution Processed Bismuth Sulfide Nanowire Array Core/Silver Sulfide Shell Solar Cells, Chem. Mater., 2015, 27(10), 3700–3706.
8. C. Liu, J. Li, Y. Li, W. Li, Y. Yang and Q. Chen, Epitaxial growth of Bi₂S₃ nanowires on BiVO₄ nanostructures for enhancing photoelectrochemical performance, RSC Adv., 2015, 5(88), 71692–71698.
9. C. Liu, Y. Yang, W. Li, J. Li, Y. Li, Q. Shi and Q. Chen, Highly Efficient Photoelectrochemical Hydrogen Generation Using Zn_xBi₂S_3+x Sensitized Platelike WO₃ Photoelectrodes, ACS Appl. Mater. Interfaces, 2015, 7(20), 10763–10770.
10. L. Whittaker-Brooks, J. Gao, A. K. Hailey, C. R. Thomas, N. Yao and Y.-L. Loo, Bi₂S₃ nanowire networks as electron acceptor layers in solution-processed hybrid solar cells, J. Mater. Chem. C, 2015, 3(11), 2686–2692.
11. M. Han and J. Jia, 3D Bi₂S₃/TiO₂ cross-linked heterostructure: An efficient strategy to improve charge transport and separation for high photoelectrochemical performance, J. Power Sources, 2016, 329, 23–30.
12. C. Liu, Y. Yang, W. Li, J. Li, Y. Li and Q. Chen, In situ synthesis of Bi₂S₃ sensitized WO₃ nanoplate arrays with less interfacial defects and enhanced photoelectrochemical performance, Sci. Rep., 2016, 6, 23451.
13. I. Zumeta-Dubé, V.-F. Ruiz-Ruiz, D. Díaz, S. Rodil-Posadas and A. Zeinert, TiO₂ Sensitization with Bi₂S₃ Quantum Dots: The Inconvenience of Sodium Ions in the Deposition Procedure, J. Phys. Chem. C, 2014, 118(22), 11495–11504.
14. H. Bao, C. M. Li, X. Cui, Y. Gan, Q. Song and J. Guo, Synthesis of a Highly Ordered Single-Crystalline Bi₂S₃ Nanowire Array and its Metal/Semiconductor/Metal Back-to-Back Schottky Diode, Small, 2008, 4(8), 1125–1129.
15. X. S. Peng, G. W. Meng, J. Zhang, L. X. Zhao, X. F. Wang, Y. W. Wang and L. D. Zhang, Electrochemical fabrication of ordered Bi₂S₃ nanowire arrays, J. Phys. D: Appl. Phys., 2001, 34(22), 3224.
16. C. Gao, H. Shen, L. Sun and Z. Shen, Chemical bath deposition of Bi₂S₃ films by a novel deposition system, Appl. Surf. Sci., 2011, 257(17), 7529–7533.
17. A. Hussain, A. Begum and A. Rahman, Effects of annealing on nanocrystalline Bi₂S₃ thin films prepared by chemical bath deposition, Mater. Sci. Semicond. Process., 2014, 21, 74–81.
18. Y. Zhao, K. T. E. Chua, C. K. Gan, J. Zhang, B. Peng, Z. Peng and Q. Xiong, Phonons in Bi₂S₃ nanostructures: Raman scattering and first-principles studies, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84(20), 205330.
19. B. D. Cullity and S. R. Stock, Elements of X-ray Diffraction, Prentice Hall, Upper Saddle River, New Jersey 07458, 3rd edn, 2001.
20. N. Fairley, Peak Fitting in XPS2006 05/18/2017 [cited 2017 07/10/2017], http://www.casaxps.com/help_manual/manual_updates/peak_fitting_in_xps.pdf.
21. D. A. Shirley, High-Resolution X-Ray Photoemission Spectrum of the Valence Bands of Gold, Phys. Rev. B: Solid State, 1972, 5(12), 4709–4714.
22. J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy: Physical Electronics Division, Perkin-Elmer Corporation, 1993.
23. G. Kresse and J. Furthmüller, Efficient Iterative Schemes for Ab Initio Total-Energy Calculations using a Plane-Wave Basis Set, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54(16), 11169–11186.
24. G. Kresse and J. Furthmüller, Efficiency of Ab Initio Total Energy Calculations for Metals and Semiconductors using a Plane-Wave Basis Set, Comput. Mater. Sci., 1996, 6(1), 15–50.
25. J. P. Perdew, K. Burke and M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. Lett., 1996, 77(18), 3865–3868.
26. J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh and F. C. Atoms, Molecules, Solids, and Surfaces: Applications of the Generalized Gradient Approximation for Exchange and Correlation, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46(11), 6671–6687.
27. P. E. Blöchl, Projector Augmented-Wave Method, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50(24), 17953–17979.
28. G. Kresse and D. Joubert, From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59(3), 1758–1775.
29. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu, J. Chem. Phys., 2010, 132(15), 154104.
30. H. J. Monkhorst and J. D. Pack, Special points for Brillouin-zone integrations, Phys. Rev. B: Solid State, 1976, 13(12), 5188–5192.
31. K. Momma and F. Izumi, VESTA: a three-dimensional visualization system for electronic and structural analysis, J. Appl. Crystallogr., 2008, 41(3), 653–658.
32. H. Koc, H. Ozisik, E. Deligöz, A. M. Mamedov and E. Ozbay, Mechanical, electronic, and optical properties of Bi₂S₃ and Bi₂Se₃ compounds: first principle investigations, J. Mol. Model., 2014, 20(4), 2180.
33. W. E. Morgan, W. J. Stec and J. R. Van Wazer, Inner-orbital binding-energy shifts of antimony and bismuth compounds, Inorg. Chem., 1973, 12(4), 953–955.
34. M. Aresti, M. Saba, R. Piras, D. Marongiu, G. Mula, F. Quochi, A. Mura, C. Cannas, M. Mureddu, A. Ardu, G. Ennas, V. Calzia, A. Mattoni, A. Musinu and G. Bongiovanni, Colloidal Bi₂S₃ Nanocrystals: Quantum Size Effects and Midgap States, Adv. Funct. Mater., 2014, 24(22), 3341–3350.
35. V. Calzia, R. Piras, A. Ardu, A. Musinu, M. Saba, G. Bongiovanni and A. Mattoni, Atomistic Modeling of Morphology and Electronic Properties of Colloidal Ultrathin Bi₂S₃ Nanowires, J. Phys. Chem. C, 2015, 119(29), 16913–16919.
36. D. Han, M.-H. Du, C.-M. Dai, D. Sun and S. Chen, Influence of defects and dopants on the photovoltaic performance of Bi₂S₃: first-principles insights, J. Mater. Chem. A, 2017, 5(13), 6200–6210.
37. S.-Q. Zhan, H. Wan, L. Xu, W.-Q. Huang, G.-F. Huang, J.-P. Long and P. Peng, Native vacancy defects in bismuth sulfide, Int. J. Mod. Phys. B, 2014, 28(23), 1450150.
38. L. V. Titova, T. L. Cocker, S. J. Xu, J. M. Baribeau, X. H. Wu, D. J. Lockwood and F. A. Hegmann, Ultrafast carrier dynamics and the role of grain boundaries in polycrystalline silicon thin films grown by molecular beam epitaxy, Semicond. Sci. Technol., 2016, 31(10), 105017.
39. L. V. Titova, T. L. Cocker, X. Wang, A. Meldrum and F. A. Hegmann, Ultrafast Carrier Dynamics in Silicon Nanocrystal Films, in Nanoscale Luminescent Materials 2, ECS Transactions, ed. D. J. Lockwood and P. Mascher, 2012, vol. 45, pp. 21–29.
40. J. B. Baxter, C. Richter and C. A. Schmuttenmaer, Ultrafast Carrier Dynamics in Nanostructures for Solar Fuels, Annu. Rev. Phys. Chem., 2014, 65, 423–447.
41. M. C. Beard, G. M. Turner, J. E. Murphy, O. I. Micic, M. C. Hanna, A. J. Nozik and C. A. Schmuttenmaer, Electronic Coupling in InP Nanoparticle Arrays, Nano Lett., 2003, 3(12), 1695–1699.
42. J. B. Baxter and C. A. Schmuttenmaer, Conductivity of ZnO Nanowires, Nanoparticles, and Thin Films Using Time-Resolved Terahertz Spectroscopy, J. Phys. Chem. B, 2006, 110(50), 25229–25239.
43. L. V. Titova, T. L. Cocker, D. G. Cooke, X. Wang, A. Meldrum and F. A. Hegmann, Ultrafast percolative transport dynamics in silicon nanocrystal films, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83(8), 085403.
44. T. L. Cocker, L. V. Titova, S. Fourmaux, G. Holloway, H. C. Bandulet, D. Brassard, J. C. Kieffer, M. A. El Khakani and F. A. Hegmann, Phase diagram of the ultrafast photoinduced insulator-–metal transition in vanadium dioxide, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85(15), 155120.
45. D. G. Cooke, A. Meldrum and P. Uhd Jepsen, Ultrabroadband terahertz conductivity of Si nanocrystal films, Appl. Phys. Lett., 2012, 101(21), 211107.
46. L. V. Titova, T. L. Cocker, X. Wang, A. Meldrum and F. A. Hegmann, (Invited) Ultrafast Carrier Dynamics in Silicon Nanocrystal Films, ECS Trans., 2012, 45(5), 21–29.
47. G. W. Guglietta, B. T. Diroll, E. A. Gaulding, J. L. Fordham, S. Li, C. B. Murray and J. B. Baxter, Lifetime, Mobility, and Diffusion of Photoexcited Carriers in Ligand-Exchanged Lead Selenide Nanocrystal Films Measured by Time-Resolved Terahertz Spectroscopy, ACS Nano, 2015, 9(2), 1820–1828.
48. T. L. Cocker, D. Baillie, M. Buruma, L. V. Titova, R. D. Sydora, F. Marsiglio and F. A. Hegmann. Microscopic origin of the Drude-Smith model, arXivorg. 2017.
49. D. Guo, C. Hu and C. Zhang, First-principles study on doping and temperature dependence of thermoelectric property of Bi₂S₃ thermoelectric material, Mater. Res. Bull., 2013, 48(5), 1984–1988.
50. G. M. Turner, M. C. Beard and C. A. Schmuttenmaer, Carrier localization and cooling in dye-sensitized nanocrystalline titanium dioxide, J. Phys. Chem. B, 2002, 106(45), 11716–11719.
51. L. V. Titova, T. L. Cocker, D. G. Cooke, X. Wang, A. Meldrum and F. A. Hegmann, Ultrafast percolative transport dynamics in silicon nanocrystal films, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83(8), 085403.
52. R. Beranek, (Photo)electrochemical Methods for the Determination of the Band Edge Positions of TiO₂-Based Nanomaterials, Adv. Phys. Chem., 2016, 2016, 2.
53. R. E. Brandt, V. Stevanović, D. S. Ginley and T. Buonassisi, Identifying defect-tolerant semiconductors with high minority-carrier lifetimes: beyond hybrid lead halide perovskites, MRS Commun., 2015, 5(2), 265–275.
54. J. K. Cooper, S. B. Scott, Y. Ling, J. Yang, S. Hao, Y. Li, F. M. Toma, M. Stutzmann, K. V. Lakshmi and I. D. Sharp, Role of Hydrogen in Defining the n-Type Character of BiVO₄ Photoanodes, Chem. Mater., 2016, 28(16), 5761–5771.

---

Footnotes

† Electronic supplementary information (ESI) available: Spectral output of illumination sources, optimized geometries of defect-containing Bi₂S₃ supercell, additional results on DOS, TRTS, light absorption, and PEC performance. See DOI: 10.1039/c7se00398f

‡ Present addresses: Department of Materials Science and Engineering, Northwestern University, Evanston, IL 60208, USA.

---