Unravelling Defect Passivation Mechanisms in Sulfur-treated Sb2Se3

Sb 2 Se 3 has emerged as an important photoelectrochemical (PEC) and photovoltaic (PV) material due to its rapid rise in photoconversion efficiencies. However, despite its binary nature, Sb 2 Se 3 has a complex defect chemistry, which reduces the maximum photovoltage that can be obtained. Thus, it is important to understand these defects and to develop passivation strategies in order to further improve this material. In this work, a comprehensive investigation of the charge carrier dynamics of Sb 2 Se 3 and the influence of sulfur treatment on its optoelectronic properties was performed using time resolved microwave conductivity (TRMC), photoluminescence (PL) spectroscopy and low frequency Raman spectroscopy (LFRS). The key finding in this work is that upon sulfur treatment of Sb 2 Se 3 , the carrier lifetime is increased by the passivation of deep defects in Sb 2 Se 3 in both the surface region and the bulk, which is evidenced by increased charge carrier lifetime of TRMC decay dynamics, increased radiative recombination efficiency and decreased deep defect level emission (PL), and improved long-range order in the material (LFRS). ABSTRACT Sb 2 Se 3 has emerged as an important photoelectrochemical (PEC) and photovoltaic (PV) material due to its rapid rise in photoconversion efficiencies. However, despite its binary nature, Sb 2 Se 3 has a complex defect chemistry, which reduces the maximum photovoltage that can be obtained. Thus, it is important to understand these defects and to develop passivation strategies in order to further improve this material. In this work, a comprehensive investigation of the charge carrier dynamics of Sb 2 Se 3 and the influence of sulfur treatment on its optoelectronic properties was performed using time resolved microwave conductivity (TRMC), photoluminescence (PL) spectroscopy and low frequency Raman spectroscopy (LFRS). The key finding in this work is that upon sulfur treatment of Sb 2 Se 3 , the carrier lifetime is increased by the passivation of deep defects in Sb 2 Se 3 in both the surface region and the bulk, which is evidenced by increased charge carrier lifetime of TRMC decay dynamics, increased radiative recombination efficiency and decreased deep defect level emission (PL), and improved long-range order in the material (LFRS). These findings provide crucial insights into the defect passivation mechanisms in Sb 2 Se 3 paving the way for developing highly efficient PEC and PV devices. low frequency Raman spectroscopy (LFRS) showed increased long-range order in the Sb 2 Se 3 semiconductor thin films upon sulfur treatment. These results explain the improved photovoltage of Sb 2 Se 3 upon sulfur treatment, which presents a simple and efficient defect passivation strategy. Time-resolved microwave conductivity reveals that a low temperature sulfurization treatment reduces both surface and bulk recombination in thin film Sb 2 Se 3 -based photocathodes for photoelectrochemical water splitting. Photoluminescence shows an improved radiative recombination yield, and low frequency Raman spectroscopy indicates improved long range order in the films following sulfur treatment.


INTRODUCTION
Photovoltaic (PV) and photoelectrochemical (PEC) cells based on Sb2Se3 have seen a rapid rise in interest in recent years and significant progress has been made in their power conversion efficiencies, [1,2] which is primarily due to its high absorption coefficient and excellent optoelectronic properties. However, a major challenge for Sb2Se3 is the low photovoltage obtained from such devices. Originally it was thought that Sb2Se3 would exhibit low defect concentration as it is a binary compound (better stoichiometric control) with benign grain boundaries. [3] However, recent computational studies have shown the complex defect chemistry of Sb2Se3 and the formation of deep defect states that are detrimental to PV/PEC performance. [4] Cation-anion anti-site defects have been shown to have a low formation energy and are energetically positioned close to the mid gap, serving as deep recombination centers. Therefore, the understanding and the suppression/passivation of such deep defects is crucial for the further development of Sb2Se3 as a PV/PEC material. Several strategies have been presented in the literature such as Te incorporation for defect passivation, [5] sulfurization treatment for improved photovoltage [6] and gradient sulfur doping in Sb2Se3. [7] More fundamental insights into the basic optoelectronic properties for defect passivation/suppression for Sb2Se3, however, have not yet been reported.
In this work, we employed time resolved microwave conductivity (TRMC) and photoluminescence (PL) spectroscopy to gain fundamental understanding of the defect passivation mechanism in Sb2Se3 by sulfur treatment. We investigated the charge carrier dynamics of Sb2Se3 before and after sulfur treatment using TRMC and observed the passivation of deep defects in Sb2Se3 as evidenced by longer carrier lifetimes. We also show that the passivation of deep defects reduces not only surface recombination, but also bulk recombination using a wavelength dependent TRMC study. Furthermore, we demonstrate increased radiative recombination efficiency by suppression of nonradiative recombination pathways in sulfur-treated Sb2Se3 using PL spectroscopy. In addition to this, low frequency Raman spectroscopy (LFRS) showed increased long-range order in the Sb2Se3 semiconductor thin films upon sulfur treatment. These results explain the improved photovoltage of Sb2Se3 upon sulfur treatment, which presents a simple and efficient defect passivation strategy.

RESULTS AND DISCUSSION
Sb2Se3 thin films were synthesized on quartz substrates by selenization of Sb metal. The sulfurization of these films was carried out in a sulfur atmosphere (see experimental section for details). No significant surface morphological changes of Sb2Se3 thin films were observed upon sulfur treatment and incorporation of sulfur into the films was confirmed by EDX ( Figure S1).
The XRD pattern of Sb2Se3 was indexed to the orthorhombic phase of Sb2Se3 (JCPDS 01-089-0821), and that of the sulfur-treated film was identical to the Sb2Se3 (Figure S2), indicating that the bare Sb2Se3 crystal structure was not affected by the sulfurization treatment. It is also evident from the absorption spectrum, that the onset of absorption does not change significantly with sulfur treatment indicating that the incorporation of sulfur does not change the optical band gap ( Figure   S3).
We investigated the charge carrier dynamics in Sb2Se3 and how sulfur treatment of Sb2Se3 affects the underlying carrier recombination processes using TRMC. In TRMC, photoexcited charge carriers are generated in a semiconductor by a short laser pulse, giving rise to a light-induced change of the conductance in the sample (photoconductance (ΔG), Figure 1 a) as compared to the dark conductance. [8,9] This photoconductance changes over time due to both radiative and nonradiative losses of free mobile charge carriers in the semiconductor and hence valuable information with regards to the loss mechanism of photogenerated charge carriers in the semiconductor can be obtained. The bare Sb2Se3 and sulfur-treated films were excited with ~ 3 ns pulses (with wavelengths between 350-1100 nm) at different intensities (10 9 -10 13 photons/pulse/cm 2 ) and the change in photoconductance was monitored over time (ns to ~80 μs). As there are no charge selective contacts to separate the charge carriers, and assuming that the carrier mobility does not change during the time window of our measurement, the decrease in the photoconductance with time is purely due to the decrease in the concentration of mobile carriers in the bands by trapping or annihilation by recombination. Figure S4 shows the photoconductance normalized to the amount of photogenerated charge carriers (ΔG/βeI0FA) as a function of time under 500 nm (2.48 eV) pulsed illumination with a light intensity of 3.69 x 10 9 photons/cm 2 per pulse (light intensity corresponding to 1 sun illumination where the laser power was chosen such that the number of photons absorbed would equal to what it would be under 1 sun) for Sb2Se3 and sulfur-treated films.
From the peak of the TRMC signal (the maximum photoconductance of the sample), we can estimate the sum of the mobilities of the individual carriers using the equation given below [10] = !" !"# # $ $%& % where I0 is the incident intensity per pulse, e is the elementary charge, β is the ratio between the inner broad and narrow dimensions of the waveguide, FA is the fraction of incident photons absorbed within the sample, φ is the charge carrier generation yield, and Σμ is the sum of electron and hole mobilities (Figure 1 b). The effective mobility for both Sb2Se3 and sulfur-treated films was greater than 10 cm 2 V -1 s -1 , which is considerably higher than other water splitting materials such as Fe2O3 (0.005 cm 2 V -1 s -1 ), [11] CuFeO2 (0.2 cm 2 V -1 s -1 ), [12] and BiVO4 (0.08 cm 2 V -1 s -1 ) [13] .
Interesting to note is the fact that at a relatively low intensity of photons from the laser (~10 9 photons/cm 2 per pulse) clear TRMC signals could be detected in contrast to the light intensities of >10 13 photons/cm 2 /pulse that are needed to detect signals from many other absorber materials. This observation reveals the excellent optoelectronic properties and low trap density of Sb2Se3. [14] The maximum of the (non-normalized) peak height is quite similar for the two different sample types showing that the sulfur treatment did not substantially change the effective mobility values of the charge carriers. The charge carriers generated are proportional to the incident light intensity on the semiconductor, and the recombination of charge carriers can be understood using the general recombination rate equation given by [15] = − − ' − ( where n is the photocarrier density, and a, b, and c are recombination rate constants. If the change in photocarrier density dn/dt (controlled by a change in incident light intensity) shows a linear dependence, then first order recombination is the dominant recombination process (i.e., a » 1 and b,c << 1), while a sub-linear dependence would indicate higher order recombination processes.
Upon comparing the light intensity dependence of the maximum of the photoconductance signal, it is observed that the TRMC signal is relatively constant at lower light intensities and then decreases with increasing light intensity >10 11 photons/cm 2 /pulse (Figure 1 c). Based on the slope of the log plot ΦΣμ vs I0 α-1 , the order of the mobile charge carrier loss process could be deduced (where α is the reciprocal of the order of the recombination process). [13] At lower light intensities, the concentration of photogenerated electrons and holes is relatively low and trapping is the dominant loss process for mobile charge carriers. At light intensities above 10 11 photons/cm 2 /pulse α was ~ 0.63, showing that second order recombination dominates. At these increased light intensities, the bands are populated by a high density of photogenerated charge carriers leading to increased second order recombination-probably due to direct electron and hole recombination-and a reduction in the TRMC signal height (Figure 1 c) already during the laser pulse. Moreover, it is interesting to note that the onset for second order recombination occured at lower photon flux in sulfur-treated films (as indicated by extrapolation in Figure S5). The explanation for this phenomenon is that the sulfur treatment reduces the amount of defect/trap states, which enables the accumulation of a higher concentration of photogenerated charge carriers in the bands at lower laser intensities as compared to the untreated sample. Therefore, the second order recombination process sets in at a lower light intensity. Qualitative examination of the normalized photoconductance decay at a single wavelength (500 nm) for the Sb2Se3 and the sulfur-treated films (Figure 2 a) shows a slower initial decay for the sulfur-treated film. This observation indicates a decreased loss of mobile charge carriers, which points to a lower amount of defects/traps in the sulfur-treated sample.
Usually, the TRMC decays are either fitted with a single or a double exponential function.
However, as is seen from Figure S6, both single and double exponential functions do not provide a satisfactory fit with our TRMC decay. In order to understand this better, we made double log plots of TRMC decays (Figure 2 b). It is evident from the plots that there are different decay processes that begin at different times (a fast process occuring in less than 100 ns, and a slow process occuring beyond 15 μs). If the two processes began at the same time, then a double exponential function would fit well with our data. As this is clearly not the case, we treated these processes separately and fitted them each with a single exponential function, as shown in Figure 2 c. This method leads to better fits of our TRMC decay, and the time constants of these processes (tfast and tslow) could be determined (Figure 2 c). The initial fast decay is probably due to defectmediated recombination, which is also the decay process active under moderate stationary illumination, i.e. in PV and PEC devices under standard operational conditions. The difference in tfast between the Sb2Se3 and the sulfur-treated Sb2Se3 was found to decrease with increasing light intensity due to progressively dominant higher order recombination (Figure 2d). At the highest light intensities (>10 12 photons/cm 2 /pulse), the difference in lifetime between Sb2Se3 and sulfurtreated films was found to be minimal as a result of direct electron and hole recombination. The dependence of the decay time of this process on the excitation density (Figure 2d) is difficult to explain, although a defect-mediated recombination process can depend on the excitation density.
Another possibility is that direct recombination within the pulse is still active after the pulse, but our fitting with a time-independent residue does not support this interpretation. In any case, an intricate numerical model seems necessary to fully resolve this question. It is noteworthy that upon sulfur treatment, the decay of the fast process is decelerated, indicating a slower loss of mobile charge carriers. The reason can be that the sulfur treatment reduces trap states or recombination.
Due to a reduced number of defects following sulfur treatment, the probability of defect mediated mobile charge carrier loss is reduced, hence a longer lifetime after sulfur treatment is observed.
On the other hand, the slow process with its characteristic time constant tslow is shorter for the sulfurized film. This slow process (>15 μs) is likely due to de-trapping of carriers from shallow defect states back into the bands followed by band to band recombination. [16] With reduced and/or shallower defect/trap states, the trapping and de-trapping process is faster for the sulfur-treated samples, leading to a faster repopulation of the bands by thermalization from the traps and increasing the probability of band to band recombination.  (Figure 3 a). The photogenerated charge carriers from 350 nm illumination would be generated close to the surface due to the high absoprtion coefficient at this photon energy, and would yield a low ΦΣμ signal due to surface recombination. [17] The ΦΣμ was compared for the different wavelengths using front vs back illumination. The shorter wavelengths (350-800 nm) exhibited significantly lower peak ΦΣμ signal when illuminated from the back, which is probably due to a high recombination velocity at the back interface (quartz/Sb2Se3) because of a high defect density. With decreasing energy of the excitation wavelength (950 and 1100 nm), the penetration depth increases and the photogenerated charge carriers have a similar and even distribution for both front and back illumination, which is evident by the similar peak ΦΣμ signal (Figure 3 b). This observation implies that the 950 nm and Next we used photoluminescence (PL) spectroscopy to understand the radiative recombination and to identify the passivation of non-radiative recombinative pathways of Sb2Se3 and sulfur-treated films. As the Sb2Se3 and sulfur-treated films exhibited low PL yield at room temperature, the PL spectra were acquired at 10 K (Figure 4 a). The PL yield from Sb2Se3 was an order of magnitude lower than the sulfur-treated samples, indicating higher non-radiative losses in Sb2Se3 before sulfur treatment (inset Figure 4 a). After sulfur treatment, these non-radiative recombination pathways are reduced, leading to a higher PL intensity. The PL spectra show a bandedge emission and a broad defect emission at lower energy. Investigation on the band-edge PL peak at 1.25 eV indicates that the transition is likely a free to bound state emission at 1.25 eV (as evidenced by no peak energy shift and the < 1 power law exponent of the PL intensity vs incident power plot (Figure S9)). [18,19] The defect emission band extends over a broad energy range from 0.75 to 1.15 eV for Sb2Se3. The defect emission in the PL spectra at energies close to 0.8 eV were reduced due to sulfur treatment suggesting a removal of a deep defect/trap state ( Figure S10). The  normalized PL spectra of Sb2Se3 and sulfur-treated films (excitation wavelength 633 nm). Inset shows the non-normalized PL intensity, illustrating the increased radiative recombination due to sulfur treatment. b LFRS spectra of Sb2Se3 films before and after defect passivation using sulfur.
Low-frequency Raman spectroscopy (LFRS) can be used to probe the acoustic vibrational modes (AVM) of materials. Long-range order induces AVM enhancement thus revealing otherwise hidden LFRS modes. [21,22] AVMs of Sb2Se3 have been extensively studied using DFT-based theoretical models. Figure 4b represents the LFRS spectrum of the Sb2Se3 samples before and after sulfur treatment. Experimental LFRS studies have optically resolved 15 of these theoretically predicted modes. The theoretical and experimental data agree within 5% deviation from theory. [23][24][25] The AVM modes reported in the literature (theoretically as well as experimentally) are compared with the modes we observed, in Table S1, and a graphical summary is presented in Figure S11.
All of the LFRS peaks observed prior to sulfur treatment agree with the previously reported modes, the vibrational symmetries of which are well documented in the literature. [24,25] Upon sulfur treatment, two new AVM peaks emerge in the LFRS at 95.18 cm -1 and 141.41 cm -1 , corresponding to the simulated values of 94 cm -1 and 142 cm -1 , respectively. We believe that this is the first time these modes have been resolved optically. We also observe that sulfur treatment enhances the in the literature that all the allotropes of S8 have similar spectral feature at 153cm -1 . [26] The changes we observe in the LFRS of Sb2Se3 can be attributed to defect passivation upon sulfur treatment.
Based on these observations, we propose that defect passivation induces long range order in the system, which results in enhancement of hidden AVM modes of Sb2Se3. It can also be inferred that the presence of sulfur allotropes at defect states induces new local force-fields that drive the material to reorient. This reorientation results in long range order, enhancement of hidden AVMs and blue-shifts in the LFRS peaks.
The defect passivation in sulfur-treated Sb2Se3 was further confirmed by investigating the photovoltage generated from Sb2Se3 PEC devices. To confirm the improvement of photovoltage upon sulfurization treatment of Sb2Se3, dual working electrode (DWE) devices were fabricated to directly probe the photovoltage generated by a p-n junction in a water splitting photocathode (in our case Sb2Se3/TiO2). As shown in Figure S12, the sulfurized sample (sulfur-treated /TiO2/Pt) exhibits an improved onset potential of ~350 mV vs VRHE, compared to the non-treated Sb2Se3/TiO2/Pt whose onset potential is ~210 mV vs VRHE. The J-ΔV (ΔV is the difference between the back contact potential WE1 and surface potential WE2) curves indicate that the Voc generated by the Sb2Se3/TiO2 heterojunction is ~210 mV, which agrees well with the value of the onset potential for the standard J-V curve. Figure S12 clearly illustrates the improvement of photovoltage upon sulfurization treatment. Improvement in the photovoltage would indicate passivation of defects in the semiconductor and hence would be a key strategy for the improvement of the efficiency of Sb2Se3 for PEC and PV applications. [27] CONCLUSION A comprehensive investigation of the charge carrier dynamics of Sb2Se3 and the influence of sulfur treatment was performed using time resolved microwave conductivity (TRMC), photoluminescence (PL) spectroscopy and low frequency Raman spectroscopy (LFRS) to gain fundamental understanding of the recombination pathways in Sb2Se3. The key finding is that upon sulfur treatment the carrier lifetime is increased (i.e. reduction of electron-hole recombination) by the passivation of deep defects in Sb2Se3, which is evidenced by two key observations. First, the lifetime of the fast process (which corresponds to electron-hole recombination) is increased while the lifetime of the carrier trapping and de-trapping process (which corresponds to reduced defect states) is decreased for the sulfurized film. Moreover, it was shown that the sulfurization treatment increases the lifetime of electron-hole recombination across all wavelengths (even for long wavelengths close to the bandgap energy) indicating that the treatment improves carrier lifetime in both the surface and bulk regions. The passivation of deep defects in Sb2Se3 also increased radiative recombination-a key requisite for high efficiency PV and PEC devices. The improved charge carrier dynamics is also supported by LFRS which shows improved long-range order in the Sb2Se3 thin films upon sulfur treatment, as evidenced by the enhancement of hidden AVMs and blue-shifts in the LFRS peaks. Unlocking the defect passivation mechanisms in Sb2Se3 provides crucial insights on the electronic properties of this emerging PV/PEC material, paving the way for highly efficient photoconversion devices.

EXPERIMENTAL SECTION
Synthesis of Sb2Se3 thin films, sulfurization and material characterization: Sb metal was sputtered from a high purity Sb sputtering target to generate films of 300 nm thickness on quartz substrates for time resolved microwave conductivity (TRMC) measurements, and on Au-coated FTO substrates for dual working electrode measurements. Selenization and sulfurization procedures were the same as used in our previous work. [6] The crystal structure of Sb2Se3 and sulfur-treated Sb2Se3 was determined by X-ray diffraction using a Rigaku SmartLab instrument (Cu Kα radiation). UV-vis absorption spectra were measured using a PerkinElmer Lambda 950 spectrometer fitted with an integrating sphere. X-ray photoelectron spectroscopy (XPS) was conducted using a Physical Electronics (PHI) Quantum 2000 X-ray photoelectron spectrometer featuring monochromatic Al-Kα radiation, generated from an electron beam operated at 15 kV and 32.3 W. The energy scale of the instrument was calibrated using Au and Cu reference samples. The analysis was conducted at 1×10 −6 Pa, with an electron take off angle of 45° and a pass energy of 46.00 eV. Charge compensation during the measurement was achieved using a low energy electron source. The sputter depth profile was performed using Ar ions (2 kV potential) on an area of approximately 4 mm 2 . Surface elemental concentrations were determined using the instrument specific sensitivity factors for calculation. The core level emissions were fitted to deconvolute spectra with contributions from multiple elements using Voigt profiles (GL30) after Shirley background subtraction.
Time-resolved microwave conductivity measurements: TRMC measurements were performed by mounting the samples in a microwave cavity cell and placing within a setup similar to the one described. [16,28]  Dual-working electrode (DWE) fabrication: For DWE measurements, the Sb2Se3 was synthesized as described above. The main difference when compared to the conventional single working electrode setup is that a second working electrode (WE2, which measures the surface potential) is fabricated on the catalyst (Pt), which enables the determination of the photovoltage (the difference between the back contact potential (WE1) and the front contact potential (WE2)).
The exposed Au-coated FTO served as the WE1 or back contact. To make the WE2 (or front contact), a 20 nm-thick Au layer was sputtered onto a part of the exposed Pt layer, as shown in Figure S12. A copper wire was then connected to the Au via Ag paint as the WE2 and covered with another layer of epoxy for protection from the electrolyte.

Table of Contents
Time-resolved microwave conductivity reveals that a low temperature sulfurization treatment reduces both surface and bulk recombination in thin film Sb2Se3-based photocathodes for photoelectrochemical water splitting. Photoluminescence shows an improved radiative recombination yield, and low frequency Raman spectroscopy indicates improved long range order in the films following sulfur treatment.