A Molecular Precursor‐Based Copper Antimony Sulfide Photodetector with Enhanced Performance by Silver Doping

Ternary copper chalcogenide semiconductors made of copper antimony sulfide (CuSbS2) are promising absorbers for high‐performance photoelectric devices due to their non‐toxic, abundant constituent elements and superior optoelectronic characteristics. However, the presence of a large number of CuSb antisite defects and atomic disorder suppress its performance in photodetection. Herein, a non‐vacuum, facile spin‐coating method based on an organic molecular precursor solution is employed to fabricate the preferable CuSbS2 thin films. With the aid of good adhesion between precursor solution and substrate, compact CuSbS2 thin films are in situ grown on the substrate. Meanwhile, the isoelectronic element of Ag‐doped CuSbS2 thin films can be realized by introducing Ag of varying amounts in the precursor. As a result, the concentration of CuSb defects decreases monotonically as the ratio of Ag/(Ag + Cu) increases from 0% to 5%. Compared with the undoped CuSbS2 device, the 5% Ag‐doped CuSbS2 photodetector achieves the optimum responsibility (R) of 244.48 A W−1, the external quantum efficiency (EQE) of 749.28% and the specific detectivity (D*) of 8.73 × 1012 Jones, which are increased by 76.4, 76.5, and 38.0 times, respectively. This study provides a novel strategy to achieve high‐performance CuSbS2 photodetectors by suppressing the CuSb inversion defects in the Ag‐doped thin film.


Introduction
Ternary copper chalcogenide semiconductor of CuSbS 2 has attracted extensive attention as one of the most promising urgent to develop a novel method with low cost and facile fabrication process to obtain high-quality CuSbS 2 thin films.
][26] Therefore, it is crucial to restrain the formation of Cu Sb antisite defects in CuSbS 2 thin film for enhancing device performance.Although isoelectronic and homogeneous element doping is considered to be an effective strategy to reduce the formation of antisite defects in photovoltaic devices, [15,27,28] it remains unclear whether a similar physical phenomenon occurs in the thin film photodetector because there has been no study on isoelectronic element doped CuSbS 2 photodetectors and their deep physical mechanisms.Consequently, the fabrication of high-performance CuSbS 2 photodetectors is still a challenge by a low cost and facile fabrication process.
Herein, a non-vacuum, facile spin-coating method based on organic molecular precursor solution is employed to fabricate preferable CuSbS 2 thin films.With good adhesion between precursor solution and substrate, compact CuSbS 2 thin films are in situ grown on fluorine-doped tin oxide (FTO) substrates during the annealing process.Meanwhile, the preparation CuSbS 2 thin films doped with isoelectronic of Ag element can be realized by introducing varying amounts of Ag in the precursor.The concentration of Cu Sb defects decreases monotonically with the ratio of Ag/(Ag + Cu) from 0 to 5%.The 5% Ag-doped CuSbS 2 photodetector achieves the optimum responsibility (R) of 244.48A W −1 , external quantum efficiency (EQE) of 749.28% and specific detectivity (D*) of 8.73 × 10 12 Jones, respectively.Compared with undoped CuSbS 2 photodetector, the R, EQE, and D* values increase by 76.4, 76.5, and 38.0 times, respectively.The results demonstrate that moderate Ag doping lead to the high-performance photodetector by suppressing Cu Sb inversion defects in the CuSbS 2 thin films.Our work presents a promising strategy to enhance the performance of CuSbS 2 photodetector by a low cost and facile process, which paves the way to achieve high-performance photodetectors based on other copper chalcogenide thin films.

Results and Discussion
CuSbS 2 semiconductor thin films are fabricated by spin-coating the molecular precursors on FTO substrates. The reaction process between metal oxides and BDCA to prepare the molecular precursors is depicted in Figure 1a.The BDCA molecule contains one thionothiolic acid (─CSSH) group, which ionize to H + and C 5 H 10 NS 2 − ions after adding a buffer solvent of ethanol and dimethylformamide (DMF).The ionized H + enables the dissolution of Cu 2 O and Sb 2 O 3 , to generate CuSb(C 5 H 10 NS 2 ) x organometallic complexbased precursors as illustrated in Figure 1b.In addition, the incorporation of Ag can be realized via adding Ag 2 O into the BDCA solution.In order to obtain high-quality thin films, the conditions of annealing temperature and atmosphere, precursor ratios, and type of substrate are studied systematically (Figures S1-S7, Supporting Information).
The crystal structure of CuSbS 2 is shown in Figure 1c, which features a layered configuration belonging to the space group Pnma (a = 6.018Å, b = 3.795 Å, and c = 14.495Å). [31] Furthermore, a part of Cu atoms are substituted by Ag atoms after Ag 2 O is added to the precursor solution for Ag-doped CuSbS 2 .
Crystalline structure and phase of as-prepared CuSbS 2 thin films are determined by X-ray diffraction (XRD) as shown in Figure 2a, in which the labeled diffraction peaks with the yellow inverted triangles belong to FTO polycrystalline thin film.The contents of Ag doping in CuSbS 2 are controlled by changing the ratios of Ag and Cu in the molecular precursors.The diffraction peaks of CuSbS 2 comprising various amounts of Ag dopants all correspond to orthorhombic chalcocite structure (PDF#88-0822).In addition, the (002) peak at 12.19°gradually shifts toward a lower diffraction angle with the increase of Ag contents.The results indicate that the position of Cu atoms is occupied by Ag atoms with larger radius, causing lattice expansion in the chalcocite structure.Notably, the thin films with the [Ag/(Ag + Cu)] ratios from 10% to 20% exhibit two mixed phases of CuSbS 2 and AgSbS 2 , implying that the single-phase region of CuSbS 2 is limited in a narrow impurity substitution range (<5% Ag doping).
To investigate the characteristics of the bandgap, UV-vis absorption spectra were measured for the different contents Agdoped CuSbS 2 thin films.All thin films exhibit strong absorption in the visible light range (Figure 2b).The absorption edge of CuSbS 2 thin film without Ag doping locates at 812 nm, which shifts to 839 nm for [Ag/(Ag + Cu)] of 0.2.The observation indicates that the light absorption range of CuSbS 2 can be expanded to the near-infrared region via varying the [Ag/(Ag + Cu)] ratios.However, there is an extra absorption peak at 740 nm for [Ag/(Ag + Cu)] ratios of 0.1, 0.15, and 0.2, corresponding to band edge absorption of AgSbS 2 , consistent with the results of XRD that detected the diffraction peaks of AgSbS 2 for the thin films.Based on the Tauc plot of (hv) 2 versus energy, the bandgap values are determined to be 1.58, 1.56, 1.54, 1.52, and 1.51 eV for the thin films with the [Ag/(Ag + Cu)] ratios of 0, 0.05, 0.10, 0.15, and 0.2, respectively (Figure 2c).The E g slightly decrease from 1.58 to1.51 eV with the increase of Ag + ratios from 0 to 20%.There are different crystallinity and average grain sizes decreasing from ≈1 to 0.5 μm for the CuSbS 2 and Ag-CuSbS 2 thin films, which result in the different Urbach-tails.The steep and gradual Urbach-tail absorptions will bring out variation of bandgap. [32]he relationship between the bandgap of thin films and their compositions is shown in Figure 2d.The change of the bandgap presents a nonlinear monotonic relationship with the compositions of CuSbS 2 , which can be fitted with the modified Equation (1) as follows: The nonlinear relationship between bandgap and the ratios of [Ag/(Ag + Cu)] originates from the deformation of electron distribution caused by the electronegativity difference of the Ag and Cu alloy atoms.The behavior indicates that our research can offer a reference for controlling the bandgap of CuSbS 2 thin films via impurity doping.
X-ray photoelectron spectroscopy (XPS) is applied to identify the chemical states of each elements in the thin films.are shown in Figure 3b,c,e.Each peak shows a highly symmetrical shape without any splitting or satellite peaks, indicating that Ag and Cu have a single valance state.In addition, there is not a distinct difference for spectral peak positions of Ag, Cu, Sb, and S by increasing the [Ag/(Ag + Cu)] ratios from 0.05 to 0.2.The peaks of Ag 3d 5/2 (373.7 eV), Ag 3d 3/2 (367.7 eV), Cu 2p 3/2 (931.4 eV), and Cu 2p 1/2 (951.3eV) in core-level spectra belong to Ag + and Cu + states [33,34] (Figure 3b,c).The core-level spectra peaks located at the binding energy of 529.4 and 538.7 eV are assigned to the 3d orbit of Sb 3+ coordinated with S 2− [35] (Figure 3d).The absence of peaks at 35.5 eV (Sb 4d 5/2 ) and 36.7 eV (Sb 4d 3/2 ) confirmed the absence of Sb 5+ .The core-level spectrum showed 2p 1/2 peak at 162.6 eV and 2p 3/2 peak at 161.4 eV belonging to S 2p. [36](Figure 3e).
The Raman spectra of CuSbS 2 and Ag x Cu 1-x SbS 2 are measured to further investigate the purity phase (Figure 3f).[39] However, for the [Ag/(Ag + Cu)] ratios of 10%, 15%, and 20%, an extra Raman peak at 300 cm −1 corresponding to the AgSbS 2 phase appears, which are consistent with the results of XRD.
Top-view and cross-sectional FESEM images of CuSbS 2 absorbers with different ratios of [Ag/(Ag + Cu)] are shown in Figure 4, indicating that the undoped and 5% Ag-doped thin films have compact surface morphology and uniform grain sizes.However, there are a large number pinholes and nonuniform grain sizes with the increase of Ag contents from 10% to 20%.The average grain sizes of the CuSbS 2 thin films decrease from ≈1 to 0.5 μm compared with 5% Ag-doped thin films.42] A cross-sectional FESEM image of the thin film with a thickness of 916 nm can be seen as shown in Figure 4f.The as-prepared thin film is densely stacked with no cracks throughout the crosssection view, which will effectively prevent the gold from infiltrating into absorber during the device preparation process.The EDS elemental mapping for 5% Ag-doped CuSbS 2 is shown in Figure 4g.The Ag-doped CuSbS 2 only contains four elements, namely Ag, Cu, Sb, and S, which are evenly distributed in the thin film.
To further evaluate the influence of Ag contents on CuSbS 2 device performance, the photoresponse behavior of the molecular precursor-based photodetector with a device structure shown in Figure 5a is investigated under room temperature and atmosphere.With the increase of [Ag/(Ag + Cu)] ratios from 0% to 20%, the responsibility (R) of CuSbS 2 photodetectors first increases from 3.2 A W −1 without Ag to 244.48A W −1 (5% Ag) and then drops to 1.16 A W −1 (20% Ag).The corresponding parameters of the photodetector as a function of Ag contents under the incident light of 405 nm are displayed in    The results demonstrate that moderate Ag doping resulting in the enhancement of photodetector deriving from the Cu Sb inversion defects being significantly suppressed in the CuSbS 2 thin films.However, with the increase of Ag from 10%, 15% to 20%, there are numerous pinholes and impurity of AgSbS 2 .With the increase of Ag + ratios to 10%, the photodetection performance and external quantum conversion efficiency of the device have declined to some extent.Combined with XRD analysis, AgSbS 2 increases the recombination probability of carriers and reduces the device performance due to the formation of the impurity phase.This also shows that Ag/(Ag+Cu) should be controlled in a certain range, and the optimum value is 0-5%.It is obvious that moderate Ag doping is a promising strategy to improve the performance of CuSbS 2 photodetector based on the molecular precursor thin film.
The operational stability and repeatability are two critical parameters for photodetectors.The real-time photocurrent response curves for CuSbS 2 photodetector with different Ag contents are measured by periodically turning on and off under 405 nm illumination at 3 V as shown in Figure 5b.The photogenerated current instantly increases to the stable value and decreases quickly as the light illumination is turned on and off, respectively.The 5% Ag-doped CuSbS 2 photodetectors possess better on/off switching properties.From these curves, one can see that the photocurrent rises with the increase of illumination power.Furthermore, the switching behavior has been maintained over multiple cycles without fluctuation, implying its good stability and repeatability originating from its high chemical stability of CuSbS It is worth noting that the photoresponsivity of the thin film photodetector decreases with the increase of the light intensity.The reasons are as follows.A large number of electron hole pairs in the device are generated under the excitation of the incident light.When the light intensity is lower than 10 μW cm −2 , the photocurrent increases nearly linearly.However, with the incident light continuously increasing, the photon-generated electron hole pairs are saturated, bringing about a slow increase in the photocurrent.Therefore, the photoresponsivity decreases with the increase of the light intensity when the light intensity exceeds 10 μW cm −2 .
Besides the photoresponsivity (R), external quantum efficiency (EQE) is a crucial parameter for judging the performance of a photodetector.EQE is related to the number of electron hole pairs   Normalized detectivity is one of the main performance indexes for photodetector, that is, the specific detection rate.This index describes the ability of detector materials to detect weak light.It can be expressed by the following Equation ( 4): Besides the visible incident light of 405 nm, the device performance of the Ag-doped CuSbS 2 photodetector under the incident light of 532 and 635 nm also are tested, as shown in Tables S1 and   S2 and Figures S8 and S9 (Supporting Information).As shown Figure 5e, the quantum efficiency of the photodetector decreases with the wavelength of the incident light, while the detectability increases first and then decreases.
To further understand the effect of Ag dopants on the charge transport characteristics, photoluminescence (PL) spectra are characterized and shown in Figure 6a.There is no obvious shift for room temperature PL peak with the increase of Ag ratios from 0 to 20%.However, the bandgap of the thin film varied from 1.58 to 1.52 eV for the different Ag contents.According to the optical characteristic of semiconductors, the differences (Δ(Eg-PL)) between the room temperature bandgap and the position of the PL peak can reflect the information on their defects.There are some related research results on Δ(Eg-PL) and the defects, such as S vacancies (V S ) or antisites. [43,44]With the increase of Ag from 0% to 20%, the Δ(Eg-PL) consistently reduces from 0.08 to 0.02 eV (Figure 6b), showing that Ag doping suppress the defects in the thin films and is therefore beneficial for charge transfer and performance improvement.It originates from partial suppression of Cu Sb antisite defects by Ag doping, causing a reduction of recombination centers. [25]elvin probe force microscopy (KPFM) is further used to identify the surface potential difference associated with the relative work function to characterize the surface defect. [45](Figure 6) Surface photovoltage (SPV), the difference between the surface potential acquired in the dark and under illumination, is adopted to indicate the density of surface states. [46]The photoinduced SPVs of undoped and Ag-doped CuSbS 2 thin films are 4 and 3 mV, respectively.The decreased SPV signal of Ag-doped CuSbS 2 thin films suggests the low density of trap states at the surface.[49][50] The average surface potential differences decreased from −403 to −427 mV, and the corresponding carrier concentration enhances.
The carrier concentration and mobility of CuSbS 2 thin films are measured and summarized in Table 2.The carrier concentration of Ag-doped CuSbS 2 absorber gradually increases with the increase of the [Ag/(Ag + Cu)] ratios, although it still exhibits ptype conductivity.On the contrary, carrier mobility is monotonically decreased.It indicates that ionized impurity scattering tends to take an important role in carrier transport.Therefore, higher carrier concentration and lower trap density are associated with Ag doping, resulting in fewer recombination centers.
Finally, the responsivity and detectivity are summarized for this work and other copper chalcogenides-based photodetectors in Table 3 and Table S3 (Supporting Information).It is clear that the responsivity of Ag-doped CuSbS 2 thin films photodetectors in this work is highest among all other copper chalcogenidesbased photodetectors.The detectivity of Ag-doped CuSbS 2 thin film device is at a high level.The work provides a novel strategy to achieve high-performance CuSbS 2 photodetectors by suppressing the Cu Sb inversion defects in the Ag-doped thin film.

Conclusion
In summary, we demonstrated that the moderate amount of Ag can be doped into the lattice of CuSbS 2 thin films by spin-coating a low-toxic BDCA-based precursor solution.Furthermore, it is observed that the quality of the CuSbS 2 thin film is improved by Ag doping.The bandgap and free carrier density of CuSbS 2 thin films are tuned by changing Ag/(Ag + Cu) ratios.As expected, the concentration of Cu Sb antisite defects decreases monotonically with the ratios of Ag/(Ag + Cu) from 0 to 5%.The 5% Ag-doped
Preparation of CuSbS 2 and Ag-Doped CuSbS 2 Molecular Precursor Solutions: For CuSbS 2 molecular precursor solution, Cu 2 O (0.9 mmol), Sb 2 O 3 (1 mmol), ethanol (1 mL), DMF (1 mL), and CS 2 (1.5 mL) as the sulfur source were mixed in a 20 mL vial under magnetic stirring at room temperature.n-Butylamine (2 mL) was added dropwise into the vial slowly, and then the solution was stirred at 60 °C until all the solid powders dissolved.Finally, a clear and brown-yellow solution was obtained.Subsequently, the precursor solution was filtered with a syringe filter (0.22 μm) to remove undissolved impurities.For Ag-doped CuSbS 2 molecular precursor solutions, all conditions were the same with the preceding processes except Cu 2 O (0.9 mmol) were substituted by Cu 2 O and Ag 2 O (0.9 mmol Cu 2 O + Ag 2 O) with Ag/(Cu + Ag) ratios of 0, 5%, 10%, 15%, and 20%, respectively (Figure S10, Supporting Information).
Deposition of Ag-Doped CuSbS 2 Thin Films: FTO-coated glasses were cleaned by sonication in sequence with acetone, deionized water, and alcohol for 30 min and then UV-treated for 15 min.Afterward, Ag-doped CuSbS 2 thin films were deposited by spin-coating the precursor solutions at 4000 rpm for 30 s on FTO substrates.The obtained thin films were annealed at 200 °C for 2 min.The spin-coating process was repeated three times.The CuSbS 2 thin films were then annealed at 350 °C for 10 min in a tubular annealing furnace.
Fabrication of Ag-Doped CuSbS 2 Photoconductive Devices: Ag-doped CuSbS 2 thin films with a thickness of 900 nm were deposited on cleaned FTO substrates by spin-coating.Then, the thin films were annealed at 350 °C for 10 min.Finally, patterned Au back-contact electrodes with a thickness of 100 nm were then successively sputtered onto the top of Agdoped CuSbS 2 using an ion sputtering coater.
Characterizations: Thermogravimetric analysis (TGA) was recorded on a TGA SDT650 instrument under a nitrogen atmosphere with a heating rate of 10 °C min −1 .The X-ray diffraction (XRD) patterns were recorded on a D/max 2200 PC X-ray diffractometer with Cu-K radiation (0.154 nm) at a scan rate of 2°min −1 .The surface and cross-section morphologies of Ag x Cu 1-x SbS 2 thin films were characterized by a field emission SEM (Gemini SEM 300).The optical absorption spectra in the range of 200-1000 nm were recorded on a UV-Vis spectrophotometer.The Raman spectra of thin films were performed at room temperature using a LabRam-HR evolution system in the back scatting configuration with 532 nm Ar laser sources of with adjustable powers from 0 to 4 mW.The photodetectors based on the thin films were excited by a He-Cd laser (405, 532, and 635 nm), whose light power density can be adjusted directly and measured by a thermopile power meter.Several neutral density filters placed between light sources and devices were used to realize various power densities under illumination.I-V and I-T characteristics of the devices were measured by Keithley 4200 A-SCS.The Kelvin probe force microscope (KPFM) was performed using a Nanoscope V Multimode 8 scanning probe microscope from Bruker Corporation.All experiments were carried out with the same AFM probe under ambient conditions (temperature of 25 °C, relative humidity of 25%).The carrier density and mobility were measured by Hall measurements (HMS-7000).

Figure 1 .
Figure 1.a) Dissolution mechanism of metal oxides in BDCA solution.b) Schematic illustration for preparing precursor solution and CuSbS 2 thin films.c) Schematic crystal structure of CuSbS 2 and Ag-doped CuSbS 2 .

Figure 2 .
Figure 2. a) XRD patterns, b) UV-vis spectra, and c) Tauc plot of Ag x Cu 1-x SbS 2 thin films with the different ratios of [Ag/(Ag + Cu)].d) The relationship between bandgap of Ag x Cu 1-x SbS 2 and the ratios of [Ag/(Ag + Cu)].The solid curve has a quadratic fit to the measured values of the bandgap.
The elemental survey XPS confirms the presence of Cu, Sb, and S in CuSbS 2 and extra Ag in the doped CuSbS 2 thin films as shown in Figure 3a.The fine scanning XPS of Ag 3d, Cu 2p, Sb 3d, and S 2p in the thin films with different [Ag/(Ag + Cu)] ratios

Figure 3 .
Figure 3. a) The XPS survey of all elements.b-e) XPS spectra of Ag 3d, Cu 2p, Sb 3d, and S 2p, respectively.f) Raman spectra of the thin films with different ratios of [Ag/(Ag + Cu)] from 0% to 20%.
2 .The current-voltage (I-V) curves of photodetectors based on 5% Ag-doped CuSbS 2 thin film under 405 nm illumination at a power ranging from 10 to 200 μW cm −2 are shown in Figure 5c.These curves are approximately straight and sym-metric for the different light power densities, implying there is good ohmic contact between CuSbS 2 thin film and Au electrodes.The detailed relation between the light intensity and photocurrent is shown in Figure 5d.It increases in the form of a power function fitted by y = ax b , where a = 0.73, b = 0.26.The photoresponsivity (R) of the photodetector is further calculated by Equation (2): R = I light − I dark P in × A (2) where I light and I dark are the photocurrent and dark current of the device, P in is incident optical power, A is the photosensitive area.The calculated R values are 244.48,86.64, 52.28, 37.45, and 28.27A W −1 for the different incident light densities of 10, 50, 100, 150, and 200 μW cm −2 , respectively.It can be fitted by a power function of y = ax b , where a = 1161.36,b = −0.67 as shown in Figure 5c.The maximum value of R reaches nearly 244.48A W −1 at an incident light density of 10 μW cm −2 , indicating excellent photosensitivity of photodetector based on 5% Ag-doped CuSbS 2 thin film for weak light signal.

Figure 5 .
Figure 5. a) Schematic of Ag-doped CuSbS 2 photodetector under illumination.b) Time response of 5% Ag-doped CuSbS 2 photodetector under different intensities of incident light (405 nm at 3 V) for five cycles.c) The photocurrent as a function of incident light intensity (at 405 nm) for 5% Ag-doped CuSbS 2 photodetector.(d) The photocurrent and photoresponsivity as a function of light intensity.e) The EQE and D* as a function of light intensity.f) The dependence of EQE and D* on the wavelength of incident light.
where h, c, R  , and e are represented by the Planck constant, speed of light, responsivity, and elementary charge, respectively.From Equation (3), the calculated value of EQE is up to 749.28% under the incident light of 405 nm at a light intensity of 10 μW cm −2 , validating excellent photosensitivity of based on 5% Ag-doped CuSbS 2 thin film for weak light signal.The detailed relation between the light intensity and EQE is shown in Figure5e.As the optical power density increased to 200 μW cm −2 , EQE decreased to 86.64%.

Figure 6 .
Figure 6.a) PL spectra and b) derived E g for the undoped CuSbS 2 and Ag-doped CuSbS 2 at ratios of 5%, 10%, 15%, and 20%.c) Surface potential differences of undoped CuSbS 2 film in the dark and e) under illumination.d) Surface potential differences of 5% Ag-doped CuSbS 2 film in the and f) under illumination.

Table 1 .
Detailed Device Parameters of Ag-doped CuSbS 2 Photodetector with Different Ratios of + Cu) under the Incident Light of 405 nm.

Table 2 .
Hole carrier density, resistivity, and hole mobility calculated from the Hall measurements, at different Ag/(Ag + Cu) ratios.

Table 3 .
The comparison of performance parameters between this work and other chalcogenide film-based photodetector in the previous works.