Solution‐Processed AgSbS2 Thin Films for Photodetection

Ternary chalcogenide, silver antimony sulfide (AgSbS2), has emerged with great potential for optoelectronic applications, thanks to its excellent optical properties, facile processability and superior stability. However, high‐performance AgSbS2‐based photodiodes have not been realized mainly due to the large dark current caused by the poor morphology and device leakage. Here, compact AgSbS2 and porous AgSbS2 thin films are fabricated mainly via modulating the sol–gel processes of the precursors. After optimizing the Ag content and the thermal annealing temperature of the AgSbS2 films, AgSbS2 photodiodes with excellent performance metrics, including low dark current density, decent specific detectivity, and superior stability, are achieved. Furthermore, the optimized AgSbS2‐based photodetectors are also introduced for heart rate detection, which exhibit excellent sensitivity and great potential for real applications.


Introduction
Inorganic chalcogenide semiconductors have widespread applications in solar cells, photodetectors, and transistors, due to their nontoxic and earth-abundant elements, low-cost deposition techniques, and superior stability compared with other solution-processed semiconductors (e.g., organics and metal halide perovskites). [1][2][3] In recent years, antimony-based chalcogenides such as antimony sulfide (Sb 2 S 3 ), [4] antimony selenide (Sb 2 Se 3 ), [5,6] silver antimony sulfide (AgSbS 2 ), [7] copper antimony sulfide (CuSbS 2 ), and silver bismuth sulfide (AgBiS 2 ) have emerged as promising candidates for next-generation thin-film photovoltaics due to their superior stability, tunable bandgap, and DOI: 10.1002/apxr.202200111 decent charge transport properties. [8][9][10] In particular, AgSbS 2 possesses high absorption coefficient covering the whole visible and near-infrared regions, and have attracted great attention in the field of optoelectronics. [11,12] For instance, Ho and Lee fabricated AgSbS 2 nanoparticles via successive ionic layer absorption and reaction (SILAR) for sensitized solar cells, and the devices achieved a power conversion efficiency (PCE) of 0.34%. [7] Yang and Lee prepared the AgSbS 2 nanoparticles using SILAR process and investigated the effect of annealing process, resulting an improved efficiency of 0.79%. [13] Lv et al. reported the preparation of AgSbS 2 films by the pyrolysis of Ag-butyldithiocarbamate and Sbbutyldithiocarbamate complex solution. By adjusting the annealing temperature of the crystal phase and the cycles of spin-coating, the AgSbS 2 -based solar cells achieved a PCE of 2.09%. [11] Furthermore, Zhang et al. deposited AgSbS 2 films via spin-coating of a precursor, containing antimony acetate, silver nitrate, and thiourea and studied the effect of annealing temperature on the crystallinity and morphology carefully. Then, a champion PCE 2.25% was realized. [14] Despite the recent progress of AgSbS 2 solar cells, AgSbS 2based photodetectors have been barely reported. More recently, Yang et al. adjusted the composition of AgSbS 2 compounds and fabricated high-performance phototransistors with relatively high detectivity and outstanding stability. [15] However, phototransistors possess relatively slow response and poor linearity, compared with photodiodes. In addition, the AgSbS 2 devices were mainly fabricated by various methods, including SILAR, chemical solution method, pulsed laser deposition, etc., [16][17][18][19] and the morphology of AgSbS 2 includes nanoparticles, nanostructured flower-like clusters, and nanoarrays. Hence, it is very challenging to fabricate layer-by-layer thin film photodiodes, due to the large leakage current caused by the pin-holes.
Motivated by the above concerns, we first investigate the solgel processes of the AgSbS 2 precursors and try to optimize the film morphology and their optoelectronic properties. The morphology could be mainly tuned by varying the Ag concentration and the related porous structure, and the charge carrier transport properties could be enhanced by optimizing the thermal annealing conditions. Then, we compare and investigate the influence of AgSbS 2 film morphology on the device performance of AgSbS 2 photodiodes, and try to minimize the dark current and noise of the photodiodes. Furthermore, we also demonstrate the optimized devices for real applications, e.g., heart rate detection.

Results and Discussion
The AgSbS 2 precursor was mainly composed of silver nitrate, antimony acetate, and thiourea. Intriguingly, we obtained two different forms of AgSbS 2 precursors by adjusting the adding order of precursor materials. Recipe A: we mixed silver nitrate, antimony acetate, and thiourea, and dissolved in a mixed solvent of dimethyl sulfoxide and water. Recipe B: we first dissolved antimony acetate and thiourea in the mixed solvent to obtain the Sb 2 S 3 precursor, and then, we mixed the silver nitrate with the Sb 2 S 3 precursor. More details could be found in the Experimental Section of the Supporting Information. Interestingly, these two precursors could result distinct AgSbS 2 film morphology, that is, we can obtain dense and uniform AgSbS 2 films from Recipe A, and Recipe B results porous AgSbS 2 films. Figure S1a,b of the Supporting Information depicts the optical photos of the AgSbS 2 precursor and spin-coated compact films, respectively. The surface morphology of the compact AgSbS 2 films was recorded by scanning electron microscopy (SEM) as shown in Figure S1c of the Supporting Information, which exhibited small grain size of 20-100 nm and noticeable pin-holes. The corresponding XRD patterns were shown in Figure S1d of the Supporting Information which can be indexed to the AgSbS 2 phase. Subsequently, we have also characterized the porous AgSbS 2 films. Figure 1a schematically illustrates the preparation of the porous AgSbS 2 films. By adjusting the ratios of AgNO 3 and Sb 2 S 3 precursor, various precursor and AgSbS 2 films could be obtained. Figure  S2 of the Supporting Information shows the optical photos of the solutions and the resulted films. The color of the precursor gradually changed from transparent to brown with the increase of Ag content, and the morphology of the obtained AgSbS 2 films also changed dramatically. The precursors containing 10 Ag% and 20% Ag resulted very hazy films, which are full with large pin-holes and aggregation of grains. With the increase of Ag concentration, the spin-coated AgSbS 2 films exhibited reflective and uniform surface morphology, which is promising for device fabrication. Figure 1b compares the XRD patterns of the porous AgSbS 2 films with various concentration of Ag. The 30% and 40% films still contain characteristic peaks of Sb 2 S 3 , and with the increase of ratios of Ag up to 50% and 60%, the obtained films only present AgSbS 2 phase. Figure 1c displays the absorption coefficients ( ) of AgSbS 2 and Sb 2 S 3 films determined by the reflectance corrected transmittance spectra and film thickness. Even with the addition of 30% Ag, the AgSbS 2 films exhibited significantly enhanced and extended absorption. It is also worth noting that the value of varied slightly for the Ag x SbS 2 thin films, which is mainly influenced by the light scattering effect of the porous structure. We also characterized the chemical states of porous AgSbS 2 films via X-ray photoelectron spectroscopy (XPS) as shown in Figure S3 of the Supporting Information, which was corrected with the standard C 1s peak at 284.8 eV as a reference. The statistics of binding energy values (in eV) of the main peaks in the XPS spectra of AgSbS 2 and Sb 2 S 3 films were summarized in Table S1 of the Supporting Information. The Ag 3d 5/2 and Ag 3d 3/2 peaks with a separation energy of 6.06 eV could be referred to Ag + state. [14,20] The binding energy difference between Sb 3d 5/2 and Sb 3d 3/2 peaks was 9.35 eV that is comparable to the reported values, indicating the existence of Sb 3+ . [20][21][22][23] The Sb 3d peaks shifted with the increasing of Ag, indicating the change of the binding energy of Sb-S, and the peaks of S 2p 3/2 and S 2p 1/2 were indexed to the reported value of S 2− . [14,24] More interestingly, we also recorded the morphology of these silver antimony sulfide films with SEM images as shown in Figure 1d-g. With the adding of Ag into the Sb 2 S 3 precursor, porous structure formed within the AgSbS 2 films. Furthermore, the morphology could be slightly modulated with Ag content, and the sample with 50% Ag exhibited the smallest pores. We have also converted these SEM images to monocolored pictures with the same threshold, and statistically determined the pore sizes of the 30% Ag, 40% Ag, and 50% Ag samples to be 101, 75, and 67 nm, respectively, as shown in Figure S4 of the Supporting Information. The formation of the pores is mainly attributed to the crystallization processes of the Sb 2 S 3 and AgSbS 2 crystals. In this work, we introduced sol-gel method to prepare Sb 2 S 3 nanocrystal first. Then, we added AgNO 3 as the Ag source to react with the Sb 2 S 3 precursor, and the recrystallization and aggregation of the nanocrystals may result various pores.
Having established the processing techniques of compact and porous AgSbS 2 films, we now turn to the device considerations. Here we fabricated the devices according to the following structure: FTO/c-TiO 2 /compact AgSbS 2 /Spiro-OMeTAD/MoO 3 /Ag. We first optimized the AgSbS 2 photodiodes to improve the photocurrent and PCE, mainly including the annealing temperature of the compact electron transport layer, AgSbS 2 active layer thickness and preannealing temperature, as shown in Figures S5-S7 of the Supporting Information. We found the devices fabricated with 450°C annealed TiO 2 , active layer thickness of ≈200 nm, and preannealed at 120°C exhibited the best device performance. The annealing temperature of TiO 2 layer could affect the crystallinity of the TiO 2 layer, which mainly influence the ability of electron extraction. The preannealing temperature mainly is crucial for the film morphology, which has significant impact on the device leakage and charge carrier blocking. The optimal active thickness mainly reflects the charge transport capability and the diffusion length of the charge carriers. The histogram of the compact AgSbS 2 devices performance was shown in Figure S8 of the Supporting Information, indicating the reproducibility of the devices. Eventually, we achieved a champion PCE of 1.61% (Figure  2a) based on the spin-coated AgSbS 2 thin films, which is high compared with the literature reported values. [11,13,14] Despite the high photocurrent and efficiency, the devices based on the compact AgSbS 2 layers possess extremely high dark current and noise as shown in Figure 2b,c, which is detrimental for photodetection. The large dark and noise current could be mainly attributed to the ultrathin films and the pin-hole induced device leakage. On the contrary, the devices fabricated based on the porous AgSbS 2 films exhibited slightly reduced photocurrent, but suppressed dark current and noise as shown in Figure 2d-f. It is worth noting that the noise current has less dependence on the reverse bias, and the devices reached a low noise density of ≈0.2 pA Hz −1/2 at −1 V, which is very promising for photodetection. The reduced dark current could be accounted to the thick porous films. The increase thickness can effectively eliminate the device leakage. However, large film thickness can also limit the charge transport. However, the porous structure offers a great platform for the hole transport material (HTM), Sprio-OMeTAD, to be penetrated into the AgSbS 2 layer, forming bulk heterojunctions, which is beneficial for charge separation and extraction.
To further optimize the device performance of the AgSbS 2 photodetectors, we carefully evaluate the influence of Ag content, annealing temperature and film thickness as shown in Figure 3ac. The 30% Ag and 50% Ag devices showed similar light/dark ratios, which are higher than the devices based on 40% Ag and 60% Ag. More importantly, the 50% Ag devices exhibited lower dark current density than the 30% Ag-based devices. We have to note that the device performance do not follow a trend of Ag concentration, but more related to the AgSbS 2 morphology. The 50% Ag samples resulted the smallest pores of ≈60 nm, can offer better charge extraction and reduce the device leakage for the bulk-heterojunctions. Therefore, we further optimized the 50% Ag incorporated devices. By adjusting the preannealing and annealing temperature, we confirmed the optimized preannealing and annealing temperature at 200 and 280°C, respectively. In addition, the device performance also showed less dependence on the active layer thickness, and the 390 nm thick devices showed the lowest dark current and slightly lower photocurrent. Furthermore, Figure 3d-f presents responsivity (R) of the 30% Ag, 40% Ag, and 50% Ag based porous AgSbS 2 photodiodes. All of these devices exhibited wide spectral response in the range of 350-900 nm, which is consistent with their absorption spectra. The highest R of the 50% Ag based AgSbS 2 photodiodes reached ≈0.09 A W −1 at 620 nm. The external quantum efficiency (EQE) spectra of these devices with −1 V bias were as shown in Figure S9 of the Supporting Information. The value of EQE decreased slightly with the increase of Ag. Figure 4 further presents other key performance metrics of the optimized AgSbS 2 photodiodes. The photocurrent transient to a modulated 530 nm LED light illumination was shown in Figure 4a. The rise times (t r ) and fall time (t f ) of the devices were determined to be 6 and 34 s, respectively. Moreover, the temporal photoresponse for a number of cycles at reverse bias voltage were shown in Figure S10a of the Supporting Information, indicating a repeatable response. Figure S10b of the Supporting Information demonstrates the photoresponse of the devices under LEDs with various wavelengths, which shown wide response at visible range. Then, we also calculated the specific detectivity (D*) of the optimized 50% Ag based AgSbS 2 photodiodes based on the measured noise and responsivity, given by [25][26][27][28] where A is the device area (0.05 cm 2 ) and B is the testing bandwidth. The specific detectivity increased under reverse bias and the highest D* reached >1 × 10 11 Jones at −1 V (Figure 4b). In addition, we tested the devices stability of the photoresponse as shown in Figure 4c. We observed negligible degradation of photocurrent under modulated 530 nm LED light illumination after numerous cycles. We also compared the device performance of these Ag x SbS 2 based photodiodes with the literature reported AgSbS 2 solar cells and photodetectors as shown in Table S2 of the Supporting Information. Our devices showed state-of-theart power-conversion efficiency and detectivity. Moreover, these photodiodes showed superior response time, compared with Ag x SbS 2 based phototransistors.
Considering the decent performance metrics of AgSbS 2 photodiodes, we further introduced the fabricated devices for heat rate monitoring. The human tissues are mainly made up of hemoglobin and water, which primarily absorb the photons with wavelength < 650 and >900 nm. The spectral response of the optimized devices covering the whole range of 350-900 nm, which can effectively detection the transmitted light from the biological tissues. [29][30][31] The schematic illustration of heart rate detection is displayed in Figure 5a. We used 670 nm LED light as the illumination background, then the finger was placed between the devices and LED. The intensity of transmission light was dependent on the increasing and decreasing blood flow in vessels led by systole and diastole of heart. The photodetectors can transform the periodic light signal to electrical signal. [32,33] In Figure 5b, the photoplethysmography signals were observed as the temporal photocurrent response of the devices without bias. The blue line was the heart rate under state of rest and the orange line was after cardio exercises. The signals indicated that the photodiodes can detect the heart rate sensitively in different states. As shown in Figure 5c, we can further obtain the characteristic frequency via the Fourier transformation of the temporal photocurrent response, which corresponding to heart beat of 67 and 112 beats per minute, respectively, which is close to the value recorded by commercial electronic wristband as shown in the insets. We also explored the ability of the devices under various wavelength light illumination as shown in Figure S11 of the Supporting Information. They also exhibited decent signal resolution for accurate heart rate evaluation, indicating the great potential for real applications.

Conclusion
In summary, we systematically investigate the solution-processed AgSbS 2 thin films. By tuning the preparation recipe of the precursors, we successfully achieved both compact and porous AgSbS 2 thin films. Then, we fabricated photodiodes based on these two types of AgSbS 2 thin films, and investigated their optoelectronic performance. Interestingly, the compact AgSbS 2 films resulted relatively high photocurrent and efficiency. However, the device leakage induced ultrahigh dark current and noise limited the application for photodetection. On the contrary, thick porous AgSbS 2 films can form bulk heterojunctions with the HTM, resulting not only decent charge extraction but also extremely low dark current and noise. Then, we further optimized the porous AgSbS 2 photodiodes, and obtained the first set of viable AgSbS 2 Figure 5. a) Schematic diagram of heart rate detection, b) temporal photocurrent response under 670 nm LED illumination, before and after cardio exercises, and c) Fourier transformation of the photocurrent curves. The insets display the heart rates recorded by a commercial electronic wristband.
photodetectors, which also exhibited great performance for heart rate detection.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.