Vapor transport deposition of antimony selenide thin film solar cells with 7.6% efficiency

Vapor transport deposition of antimony selenide thin film solar cells with 7.6% efficiency ARTICLE DOI: 10.1038/s41467-018-04634-6 OPEN Vapor transport deposition of antimony selenide thin film solar cells with 7.6% efficiency 1,2 1,2 1,2 1,2 1,2 1,2 Xixing Wen , Chao Chen , Shuaicheng Lu , Kanghua Li , Rokas Kondrotas , Yang Zhao , 1,2 1,2 1,2 1,2 1,2 1,2 Wenhao Chen , Liang Gao , Chong Wang , Jun Zhang , Guangda Niu & Jiang Tang Antimony selenide is an emerging promising thin film photovoltaic material thanks to its binary composition, suitable bandgap, high absorption coefficient, inert grain boundaries and earth-abundant constituents. However, current devices produced from rapid thermal eva- poration strategy suffer from low-quality film and unsatisfactory performance. Herein, we develop a vapor transport deposition technique to fabricate antimony selenide films, a technique that enables continuous and low-cost manufacturing of cadmium telluride solar cells. We improve the crystallinity of antimony selenide films and then successfully produce superstrate cadmium sulfide/antimony selenide solar cells with a certified power conversion efficiency of 7.6%, a net 2% improvement over previous 5.6% record of the same device configuration. We analyze the deep defects in antimony selenide solar cells, and find that the density of the dominant deep defects is reduced by one order of magnitude using vapor transport deposition process. Sargent Joint Research Center, Wuhan National Laboratory for Optoelectronics and School of Optical and Electronic Information, Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan 430074, Hubei, China. Shenzhen R&D Center of Huazhong University of Science and Technology, Shenzhen 518000, China. Correspondence and requests for materials should be addressed to J.T. (email: jtang@mail.hust.edu.cn) NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ntimony selenide (Sb Se ) has recently emerged as a fast-turnaround manufacturing method for commercial CdTe 2 3 17, 18 promising green alternative to CdTe solar cells because it solar cells , and adopting this technique for Sb Se technology 2 3 Apossesses very attractive optoelectronic properties such as could facilitate the commercial competitiveness of Sb Se solar 2 3 proper bandgap (about 1.03 eV indirect and 1.17 eV direct) for cells. After optimizing the Sb Se films via VTD process, we 2 3 the absorption of a significant portion of the solar spectrum, high obtain the champion indium tin oxide (ITO)/CdS/Sb Se /Au 2 3 5 −1 optical absorption coefficient (greater than 10 cm ) and decent solar cells with a certified efficiency of 7.6%, which is a record of 2 −1 −1 1–3 1, 2, 4–12, 19, 20 carrier mobility (about 10 cm V s ) . Besides, because it is all Sb Se thin film solar cells by far and represents a 2 3 a binary compound with high vapor pressure, a fast, low tem- net 2% efficiency gain over previous report with the same device perature vacuum-based deposition technique can be employed configuration . We comparatively analyze the physical properties like the ones established in CdTe photovoltaics. It also possesses a of VTD-fabricated and RTE-fabricated Sb Se devices by X-ray 2 3 one-dimensional crystal structure with loose van der Waals diffraction (XRD), transient absorption (TA) spectroscopy and interaction between ribbons , thereby enabling grain boundaries deep-level transient spectroscopy (DLTS). We find that the VTD- free of dangling bonds in c-axis-oriented films and minimizing fabricated devices possess much higher film crystallinity, longer recombination losses therein. Furthermore, the earth-abundant carrier lifetime and fewer bulk and interfacial defects than RTE- elemental compositions of Sb Se as well as its easy fabrication fabricated devices. Especially, the reduced density of defect 2 3 promise low-cost manufacturing. All these attributes suggest the complex (Sb +Se ) enhances the device performance by Se Sb great potential of Sb Se for high-efficiency thin film solar cell decreasing the photo-generated carrier recombination. 2 3 and commercial application. However, the highest power conversion efficiency (PCE) of Results Sb Se thin film solar cells with a CdS/Sb Se superstrate con- 2 3 2 3 Fabrication of Sb Se films via VTD process. A schematic 2 3 2, 4–11 figuration is so far 5.6% , and with a ZnO/Sb Se superstrate 2 3 representation of VTD system is shown in Fig. 1a. The Sb Se 2 3 configuration 5.93% . Employing PbS colloidal quantum dots powder and the substrate were placed in the center and at the (QDs) as the hole transport layer, as-fabricated device CdS/ right end of the heater, respectively. The full deposition details are Sb Se /PbS QDs demonstrated an efficiency of 6.5% . Rapid 2 3 provided in experimental section. Hereby, the heating tempera- thermal evaporation (RTE) was used to fabricate all above- ture of the evaporation source, the pressure in quartz tube and the mentioned Sb Se solar cells. RTE is a process similar to close- 2 3 substrate temperature are the key factors that determining the space sublimation , in which the vapor is limited to a confined quality of final Sb Se films. The distance between the substrate 2 3 space and rapidly deposits on the substrate. The distance from the and the heating center was varied to achieve adjustable substrate substrate to the evaporation source is only 0.8 cm, and the whole temperature. We deposited Sb Se films at different heating 2 3 deposition is performed within 35 s. The small confined space temperatures, pressures and substrate temperatures. XRDs of all and the rapid deposition increase the difficulty of mixing the Sb Se films were measured to characterize the crystallinity and 2 3 vapor particles (Se, Sb and Sb Se ) evenly, potentially promoting x y orientation of these films, which is shown in Supplementary defect formation. The formation of defects, such as interstitial and Fig. 1. All the Sb Se films had a preferred [221] orientation with 2 3 antisite defects, accelerates nonradiative recombination and the only difference being the intensity of the diffraction peaks. As 13–16 degrades device performance . Furthermore, limited by our Sb Se is composed of one-dimensional (Sb Se ) ribbons stack- 2 3 4 6 n RTE facility, a simple one-zone tube furnace, the source and ing together via weak van der Waals force, a proper film orien- 1, 2 substrate temperatures are strongly correlated, which seriously tation is crucial for facile carrier transport in the film . Our hinders the independent optimization of both substrate and previous reports have demonstrated that [221]-orientation 1, 2, 6, 8 source temperature. These problems in our RTE technique have enhances carrier transport across Sb Se film . Therefore, 2 3 seriously restricted the development of Sb Se solar cells. There- 2 3 with the film thickness and measurement details kept strictly fore, we urgently need to explore alternative strategies to further identical, we utilized the intensity of (221) XRD peaks in Sb Se 2 3 improve Sb Se film quality and device performance. 2 3 films to evaluate their crystallinity. The crystallinity evolution of Here, we develop a vapor transport deposition (VTD) process Sb Se films deposited on CdS layers with varied evaporation 2 3 to fabricate record efficiency Sb Se thin film solar cells. In the 2 3 temperature, pressure and distance was investigated, as shown in VTD process, both the substrate temperature and the distance Fig. 1b. By carefully optimizing the three key factors, Sb Se film 2 3 between source and substrate are adjustable, enabling not only with the highest crystallinity was obtained when the evaporating highly oriented Sb Se film, but also enormously improved film 2 3 temperature was set at 510 °C, the pressure was 3.2 Pa and the crystallinity and reduced bulk and interfacial defects in Sb Se 2 3 distance from the substrate to the heating center was 21 cm. Here, solar cells. In addition, the VTD process is a proven low-cost and the actual temperature curves of powder and substrate were ab c Heater Au Sb Se vapor 2 3 Sb Se 2 3 Pump CdS Au Sb Se powder Substrate 2 3 ITO Glass 480 500 520 34 5 23 21 19 Temperature (°C) Pressure (Pa) Distance (cm) Fig. 1 Fabrication of CdS/Sb Se solar cells. a Schematics of our VTD system. b Evolution of crystallinity with the optimization of deposition condition. c 2 3 The device structure of our Sb Se solar cells 2 3 2 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | Evolution of crystallinity XRD intensity of (221) peak (×10 counts) NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab 0.44 0.42 0.40 0.38 0.36 0.34 480 500 520 34 5 19 21 23 480 500 520 345 19 21 23 Temperature (°C) Pressure (Pa) Distance (cm) Temperature (°C) Pressure (Pa) Distance (cm) cd 28 60 18 45 480 500 520 345 19 21 23 480 500 520 345 19 21 23 Temperature (°C) Pressure (Pa) Distance (cm) Temperature (°C) Pressure (Pa) Distance (cm) Fig. 2 Deposition condition-dependent photovoltaic characteristics. a Power conversion efficiency (PCE), b open-circuit voltage (V ), c short-circuit OC current density (J ) and d fill factor (FF) of the CdS/Sb Se solar cells fabricated by VTD process. A total of 135 devices are included for the statistics SC 2 3 analysis. Solid sphere symbols and error bars indicate average values and standard deviations, respectively monitored during the heating and cooling steps. The input Device performance. The Sb Se absorber layers were fabricated 2 3 temperature program and the measured temperature curves of under the above-mentioned conditions, and the corresponding substrate and powder are shown in Supplementary Fig. 2. This device structure ITO/CdS/Sb Se /Au is shown in Fig. 1c. The 2 3 heating profile and substrate distance corresponded to the max- dependence of device performance on different evaporation imum actual source and substrate temperatures of 540 °C and temperatures, pressures and distances is summarized in Fig. 2. 390 °C, respectively, as monitored by the thermocouples. The variation trend of PCE is similar to the crystallinity evolution The crystallinity evolution of Sb Se films indicates the of Sb Se films (Fig. 1b), which indicates that device performance 2 3 2 3 importance of evaporation temperature, pressure and substrate strongly correlates with the crystallinity of Sb Se film. As can be 2 3 temperature in VTD process. High evaporation temperature seen from Fig. 2, the best PCE, open-circuit voltage (V ), short- OC increases the kinetic energy of vapor particles and the surface circuit current density (J ) and fill factor (FF) were obtained SC mobility of adatoms on the substrate . On the other hand, high under the deposition condition where the highest film crystal- pressure can increase in-flight collisions with background gas linity was achieved (Fig. 1b). After carefully optimizing Sb Se 2 3 21, 22 atoms, and then reduced the kinetic energy of vapor particles . films and devices, the champion device with a certified power In addition, increasing the substrate temperature will lead to the conversion efficiency of 7.6% was obtained (certificate included in increased surface mobility of adatoms . Hence, the kinetic Supplementary Fig. 3). This value represents the highest PCE of 1, 2, 4–12 energy of vapor particles and surface mobility of adatoms are all Sb Se thin film solar cells reported so far , which is 2% 2 3 ultimately defined by combination of evaporation temperature, higher than previous 5.6% certified efficiency with the same pressure and substrate temperature. During the Sb Se film device configuration . Figure 3a shows the light current density- 2 3 deposition, vapor particles with proper kinetic energy and voltage (J-V) curve of the champion VTD-fabricated CdS/Sb Se 2 3 adatoms with proper surface mobility are mandatory for high- solar cell with certified PCE of 7.6%, V of 0.42 V, J of 29.9 OC SC −2 quality Sb Se films. Energetic vapor particles and high mobility mA cm and FF of 60.4%. The J-V curve of a RTE-fabricated 2 3 −2 of adatoms may increase the instability of adatoms, causing solar cell with PCE of 5.6% (V = 0.39 V, J = 25.3 mA cm OC SC displacements of lattice and consequently creating lattice and FF = 56.4%) is also included in Fig. 3a for comparison. 21, 22 defects . Such defects may act as new nucleation sites and Obviously, every performance parameter of the VTD-fabricated increase nucleation density of adatoms . In the extreme, the solar cell, especially J , is larger than that of the RTE-fabricated SC increased lattice defects and nucleation density may result in device. decreased crystallinity and quality of the film. That is why the We further checked the external quantum efficiency (EQE) of crystallinity of the Sb Se film decreased when deposited at overly the two devices (Fig. 3b) to investigate their photo-response. The 2 3 high evaporation temperature, lower pressure and closer distance EQE spectrum of VTD-fabricated device demonstrated a higher to evaporation source. photo-response from 520 nm (the absorption onset of CdS layer) NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 3 | | | –2 J (mA cm ) SC PCE (%) V (V) OC FF (%) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 a b 100 35 VTD RTE 15 60 RTE VTD V (V): 0.39 0.42 OC –2 J (mAcm ): 25.3 29.9 SC FF (%): 56.4 60.4 PCE (%): 5.6 7.6 10 CdS/VTD-Sb Se 2 3 –5 CdS/RTE-Sb Se 2 3 0 0 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 300 400 500 600 700 800 900 1000 1100 Voltage (V) Wavelength (nm) c d 0.44 0.90 VTD slope = 1.23 k T/q VTD J ∝ I B SC 0.90 0.42 RTE slope = 1.51 k T/q RTE J ∝ I B SC 0.40 0.38 0.36 0.34 0.32 0.30 0.28 0.26 10 100 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 –2 –2 Intensity (mW cm ) ln[Intensity (mW cm )] Fig. 3 Device performance and light intensity-dependent J and V of devices. a The light J-V curves of VTD- and RTE-fabricated devices under AM1.5 G SC OC illumination. The J-V curve of the VTD-fabricated device with certified efficiency of 7.6% (area= 0.091 cm ) was measured by National Institute of Metrology on 1 September 2017. The calibration certificate number is GXtc2017-1987 (Supplementary Fig. 3). b EQE and integrated J of the VTD- and SC RTE-fabricated devices. Light intensity-dependent c J and d V . Neutral-density filters (THORLABS) were used to control the light intensity SC OC to 1100 nm than that of RTE-fabricated device. The maximum Figure 3c illustrates J as a function of light intensity (I). Herein, SC EQE of VTD-fabricated device was close to 90% between 540 and J was fitted according to the power law dependence (J ∝ I ) SC SC 720 nm. For comparison, the device fabricated by our RTE by the log–log scale plot. The power value α for both devices is technique only had the maximum EQE of about 85%. This 0.9, close to unity (first-order). This means that trap-assisted suggests the VTD-fabricated devices possess very low recombina- recombination is present in both solar cells and is the dominating 24, 25 tion losses of photo-generated carriers and long carrier lifetimes loss mechanism . The light-intensity-dependent V can OC at CdS/Sb Se interface and in the whole Sb Se absorber .In provide critical insights into the recombination mechanism in the 2 3 2 3 −2 26 Fig. 3b, we derived the current density of 29.8 and 25.1 mA cm solar cells . For V measurement, the device is open circuit, so OC by integrating the EQE spectra with standard AM1.5 spectrum to there is no current extraction from the devices, and all photo- further validate our J values, which agreed with the experi- generated carriers recombine in Sb Se film. Thus, the carrier SC 2 3 −2 mental values of 29.9 and 25.3 mA cm very well, respectively. recombination process can be reflected based on the relationship Furthermore, a solar cell with initial PCE of 7.25% was stored in of V ∝ n(k T/q)ln(I), where k is the Boltzmann constant, T is OC B B 26–29 ambient air for about 40 days without encapsulation. This the temperature and q is elementary charge . n(k T/q) is the representative device was measured every week to monitor the slope of V vs. the natural logarithm of light-intensity ln(I). For OC stability of Sb Se film solar cell, and the results are displayed in trap-free solar cells, the slope of V vs. ln(I) should be k T/q 2 3 OC B 26, 29 Supplementary Fig. 4a. The PCE remained unchanged during the (i.e., n = 1) . For our devices, as shown in Fig. 3d, the slopes whole process. For the stability of device under continuous obtained by linear fitting were 1.23(k T/q) for the VTD- illumination, Supplementary Fig. 4b shows slight decrease for fabricated device and 1.51(k T/q) for the RTE-fabricated device. CdS/Sb Se device but no decrease for ZnO/Sb Se device, which This, again, indicates the presence of trap-assisted 2 3 2 3 1 29, 30 is consistent with the previous report . These results indicate that Shockley–Read–Hall (SRH) recombination in both devices . VTD process is a simple and effective technique for producing The slope was decreased from 1.51(k T/q) to 1.23(k T/q) by using B B high-efficiency and stable Sb Se thin film solar cells. VTD process, suggesting reduced trap-assisted recombination in 2 3 28, 31 VTD-fabricated devices . We measured the TA decay of the two devices to investigate the carrier lifetime (Supplementary Carrier recombination in CdS/Sb Se solar cells. From the Fig. 6). For the steady-state absorption (Supplementary Fig. 7)of 2 3 VTD- and RTE-fabricated Sb Se films, the absorption rose to the improved EQE of VTD-fabricated devices, we inferred that the 2 3 carrier transport was enhanced, and recombination loss was maximum at around 940 nm. Thus, the transient kinetic decay was monitored at 940 nm. By fitting the kinetic decay data, the reduced in the active Sb Se layer by VTD process. Therefore, to 2 3 further clarify the carrier recombination processes in Sb Se solar longer carrier lifetime (1339 ps) was obtained in VTD-fabricated 2 3 cells, we performed light-intensity-dependent J and V mea- device than that (1149 ps) of RTE-derived device. Longer carrier SC OC surements on the VTD- and RTE-based photovoltaic devices. The lifetime not only permits more efficient carrier collection and −2 −2 complete sets of J-V curves are given in Supplementary Fig. 5. hence larger J (VTD 29.9 mA cm vs. RTE 25.3 mA cm ), SC 4 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | –2 Current density (mA cm ) –2 J (mA cm ) SC V (V) EQE (%) OC –2 J (mA cm ) SC NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab c VTD RTE 500 nm 500 nm RTE VTD 100 200 300 400 500 600 Grain size (nm) de f Sb Se 900 nm 2 3 Sb Se 900 nm 2 3 VTD CdS 70 nm CdS 70 nm ITO ITO RTE 500 nm 500 nm RTE VTD 10 20 30 40 50 60 2 (degrees) gh i Depth (nm) RTE VTD Depth (nm) 0 290 580 870 1160 1450 0 290 580 870 1160 1450 0 14 RTE DLCP Sb RTE CV profiling Se VTD DLCP ITO Cd VTD CV profiling Sb Se 10 Sb Se 2 3 Sn 2 3 –1 –1 10 10 CdS Sb Se ITO –2 –2 4 Cd 10 10 Sn CdS 0 200 400 600 800 1000 0 200 400 600 800 1000 150 200 250 300 350 Sputter time (s) Sputter time (s) Width (nm) Fig. 4 Characterization of Sb Se films and interface analysis of CdS/Sb Se devices. SEM top-view images of a VTD-fabricated and b RTE-fabricated 2 3 2 3 Sb Se films. c Histogram of grain size in VTD-fabricated and RTE-fabricated Sb Se films. Cross-sectional SEM images of d VTD-fabricated and e RTE- 2 3 2 3 fabricated CdS/Sb Se devices. f XRD of VTD-fabricated and RTE-fabricated Sb Se films. SIMS depth analysis of g VTD-fabricated and h RTE-fabricated 2 3 2 3 devices. i C-V profiling and DLCP for VTD- and RTE-fabricated devices but also enables higher carrier concentration within the absorber cases were the same: ITO/CdS/Sb Se /Au. The Sb Se layers were 2 3 2 3 layer and therefore wider quasi-Fermi level splitting and compact and well adherent to the CdS with the same thickness of improved V (VTD 0.42 V vs. RTE 0.39 V). about 900 nm. We compared XRD patterns of the two Sb Se OC 2 3 layers, measured under identical conditions, as shown in Fig. 4f. Obviously, both films were favorably orientated along [221] Morphology and crystallization of Sb Se films. We now study 2 3 direction. However, the XRD intensity of VTD-fabricated film the material origin of the improved device performance. The was much stronger than that of RTE-fabricated film, indicating scanning electron microscopy (SEM) top view images of Sb Se 2 3 much higher crystallinity was obtained by VTD process. The SEM films deposited on glass/ITO/CdS substrates by VTD and RTE and XRD results both demonstrate VTD-fabricated Sb Se films are shown in Fig. 4a, b, respectively. The grains of VTD- 2 3 have larger grains and higher crystallinity compared with RTE- fabricated film are obviously larger than that of Sb Se film from 2 3 fabricated films. It should be noted that, for RTE-fabricated RTE process. We statistically analyzed the distribution of grain Sb Se film, the distance between evaporation source and sub- size from the two SEM top view images. Figure 4c depicts the 2 3 strate was merely 0.8 cm, and the deposition was carried out on histograms of the grain size. The average grain size of VTD- 1, 2 350 °C substrate for 35 s , while the corresponding value in the fabricated film was 382 nm with a standard deviation of 79 nm, VTD process was 21 cm, 390 °C and 2 min, respectively. Longer whereas the RTE-derived Sb Se film had an average size of 297 2 3 traveling distance could increase the collision probability between nm and a standard deviation of 82 nm. The cross-sectional SEM Sb Se particles and gas molecules, reduce the momentum of images of Sb Se solar cells fabricated by VTD and RTE methods x y 2 3 these particles when impinge onto the substrate and hence are shown in Fig. 4d, e, respectively. Device structures in both NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 5 | | | Normalized SIMS Normalized SIMS 16 –3 N (×10 cm ) 3 Frequency (%) Intensity (×10 counts) (120) (130) (230) (211) (221) (301) (321) (041 ) (141) (431) (002) (061) (422) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 decrease the occurrence of nucleation. Less nucleus within the device are filled by carriers through applying a voltage pulse to film, accompanied with higher substrate temperature and longer the device, which changes the capacitance associated with p–n film growth duration, enabled the VTD-derived Sb Se film with junction of the device . 2 3 larger grain size and better grain crystallinity . We comparatively analyzed the deep-level defects in VTD- fabricated and RTE-fabricated Sb Se films by DLTS. Herein, 2 3 we adopted DLTS with minority carriers injection (inj-DLTS) Interfacial properties of the devices. For all thin film photo- 37, 38, 40, 43 to detect both electron and hole traps in Sb Se films . 2 3 voltaic devices with CdS as buffer layer, Cd diffusion at the As shown in Fig. 5a, during the measurement, a quiescent 32, 33 interface is always observed . To check the Cd diffusion in our reverse bias (V )was first applied to the junction for reverse devices, we measured their composition distribution by secondary forming a depletion width in Sb Se layer. Then, a forward 2 3 ion mass spectroscopy (SIMS) depth profiling. As shown in Fig. 4g, pulse voltage (V )was appliedto fill the traps in the depletion fill h, Cd diffusion was observed in both VTD-fabricated and RTE- region. If we set V < 0, the junction was under reverse bias, fill fabricated CdS/Sb Se devices. The deeper Cd diffusion (greater 2 3 and there were only holes injected into the depletion region to than 200 nm) than that (greater than 100 nm) in RTE-fabricated fill the traps, and only hole traps were detected. If we set V > fill device was found in VTD-fabricated device. As CdS buffer layer 0, under forward bias condition, both electrons and holes were was prepared following identical procedures, we deduced that the injected into the depletion region, and thus both electron and different Cd-diffused depth was certainly caused by the different 37, 38, 40 hole traps could be detected (inj-DLTS) . deposition techniques: for VTD process, substrate temperature was Here, we take the hole traps as an example to elaborate the about 390 °C and deposition lasted for 2 min, while in RTE process change of transient capacitance caused by holes capture and 1, 2 the corresponding values were 350 °C and 35 s, respectively . emission (Fig. 5a). Figure 5b shows the corresponding variation Because diffusion is driven thermally, higher substrate temperature of depletion width (W ) and the process of holes being trapped and longer deposition time resulted in more Cd diffusion in VTD- and emitted, before and after the pulse voltage applied. C 32 0 fabricated device. As we reported before , Cd diffusion converted represents the steady-state junction capacitance at V bias. reverse p-type Sb Se into n-type, and resulted in a buried homojunction 2 3 When V was applied and held for a while, W narrowed down fill d at CdS/Sb Se interface. This could reduce the interface defects and 2 3 to W and the hole trap defects were filled. Once V pulse 34 t0– fill recombination at the heterojunction interface , being beneficial relaxed, W broadened to W and capacitance decreased to C d t0+ t0 for device performance. instantaneously. W was even larger than W because some + t0+ 0 Therefore, we further measured interfacial defects using holes had been trapped in depletion region. Over the course of capacitance-voltage (C-V) profiling and deep-level capacitance time, the trapped holes were gradually and eventually completely 1, 35, 36 profiling (DLCP) techniques . Generally, the defects density emitted from the occupied deep level. Then, W shrank to W d 0 obtained from C-V profiling (N ) includes the response of free CV and the capacitance returned to steady state (C ). As the state of carriers, and bulk and interfacial defects, while the defect density activated defects was determined by the temperature, a sequence obtained from DLCP measurement (N ) represents the DLCP of transient capacitance was thus measured at different sample 1, 36 response only from the free carrier and the bulk defects . temperature, and capacitance changes within a time window vs. Thus, we can characterize the defect density at CdS/Sb Se 2 3 the different temperatures was sampled as DLTS signal. We took interface by the difference between N and N . As shown in CV DLCP a fixed time window between t and t (Fig. 5a), and then the 1 2 Fig. 4i, obviously, the difference between N and N of RTE- CV DLCP corresponding hole emission rate e can be expressed by Eq. (1): fabricated device is much higher than that of VTD-fabricated device, indicating lower defect density at VTD-fabricated CdS/ lnðÞ t =t 2 1 e ¼ : ð1Þ Sb Se interface. Because the doping concentration of CdS was 2 3 t  t 2 1 3, 6 much higher than in Sb Se film , almost all depletion width 2 3 (W ) extended in Sb Se layer. Here the volume to surface ratio is d 2 3 W , the interfacial defect density of RTE-fabricated and VTD- As shown in Fig. 5a, the capacitance change within the time 11 −2 fabricated devices was calculated to be about 2.1 × 10 cm and window is ΔC=C −C , which depends on the sample tempera- t2 t1 10 −2 2.8 × 10 cm , respectively. The interfacial defect density in ture. DLTS signal can be reflected from the variation of ΔC/C VTD-fabricated Sb Se solar cell is also much lower than that 2 3 with different sample temperatures. Besides, based on the 11 −2 (1.22 × 10 cm ) in RTE-fabricated Sb Se solar cell with ZnO 2 3 changing of capacitance during the discharging process of traps, as buffer layer . Consequently, we concluded that the promoted the hole traps and electron traps can be differentiated by positive interfacial diffusion had effectively reduced interface defects. This and negative ΔC, respectively. is reminiscent of our previous observation that the performance DLTS results of VTD-fabricated and RTE-fabricated CdS/ of RTE-fabricated and thermally evaporated CdS/Sb Se solar 2 3 Sb Se devices are shown in Fig. 5c. One negative and two 2 3 cells always improved when stored in ambient air for a few positive peaks were found in both devices, indicating one electron 2, 6 days . Similar effect has also been demonstrated in CdS/CdTe trap (E1) and two hole trap defects (H1 and H2) in Sb Se films. 2 3 solar cells that the interfacial diffusion reduced interfacial lattice The activation energy and capture cross-section of traps can be 15, 38 mismatch and defects, then reduced the current loss and obtained from Arrhenius plot based on Eqs. (2) and (3): improved device efficiency . e 16πk m B E  E p p T V ln ¼ ln σ  ; ð2Þ 2 3 T h k T Deep defects in VTD- and RTE-fabricated devices. We subse- B quently investigated the deep defects in CdS/Sb Se solar cells 2 3 fabricated via VTD and RTE processes to demonstrate how the e 16πk m E  E n B C T trap-assisted recombination was reduced in VTD-fabricated n ð3Þ ln ¼ ln σ  ; 2 3 T h k T device. DLTS is a well-accepted powerful tool to investigate defect energy level, type and concentration in thin film 14, 15, 37–40 photovoltaics . It uses the transient capacitance of p–n junction at different temperature as a probe to monitor the where e and e represents electron and hole emission rate, n p changes in charge state of a deep defect center . Traps in the respectively, which can be obtained by Eq. (1); σ and σ are p n 6 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab t t t 0 1 2 C=C C=C 0 t 0– Time fill fill W W 0 t 0– reverse t 0– C=C C=C t t 0+ ≈ ≈ t 2 t 0+ ΔC = C – C t 2 t 1 t 1 n-region Depletion region p-region t 0+ cd 0.25 –3 VTD-E1 0.20 RTE-H1 VTD-H2 VTD VTD-H2 –4 RTE VTD-H1 0.15 0.10 RTE-H2 –5 RTE-H1 VTD-H1 0.05 –6 0.00 –0.05 RTE-E1 RTE-E1 –7 –0.10 –0.15 –8 RTE-H2 VTD-E1 –0.20 160 200 240 280 320 360 32 36 40 44 48 Temperature (K) 1/k T Fig. 5 DLTS analysis of VTD- and RTE-fabricated CdS/Sb Se solar cells. a Schematic demonstration of the mechanism of DLTS measurement. b Variation 2 3 of depletion width and the process of holes being trapped and emitted during the measurement. c DLTS signals of VTD-fabricated and RTE-fabricated devices at t /t = 1 ms/10 ms. d Arrhenius plots obtained from DLTS signals. C and W are the junction capacitance and the depletion width at the 1 2 t0– t0– moment before pulse voltage ended, respectively. C and W are the junction capacitance and the depletion width at the moment after pulse voltage t0+ t0+ ended, respectively capture cross-section of hole and electron traps, respectively, T is the temperature, and k T is the thermal energy. In addition, m B p Table 1 Defect parameters of VTD- and RTE-fabricated and m respectively represent effective mass of hole and electron, Sb Se solar cells 2 3 and E , E and E are the energy level of defect, conduction and T C V valence bands, respectively. Figure 5d shows the Arrhenius plot 2 −3 Defects E (eV) σ (cm ) N (cm ) T T obtained from Fig. 5c, by the varied e (hole or electron emission −17 15 VTD-H1 E + 0.48 ± 0.07 1.5 × 10 1.2 × 10 rate) corresponding to the DLTS peak positions in temperature −13 14 VTD-H2 E + 0.71 ± 0.02 4.9 × 10 1.1 × 10 according to Eqs. (2) and (3). Activation energy (E − E or E −13 14 T V C VTD-E1 E − 0.61 ± 0.03 4.0 × 10 2.6 × 10 −16 14 − E ) of the trap can be calculated from the slope of ln(e /T )or T p RTE-H1 E + 0.49 ± 0.03 2.2 × 10 1.2 × 10 −13 15 ln(e /T ) vs. 1/k T plot, and the capture cross-section could be n B RTE-H2 E + 0.74 ± 0.04 7.7 × 10 2.3 × 10 −12 15 extracted from the y-intercept. The trap concentration (N ) can RTE-E1 E − 0.60 ± 0.02 1.6 × 10 1.7 × 10 T C 15, 39 be obtained from the Eq. (4) : 2ΔC max measured the variation of depletion width with the applied bias, N ¼ N ; ð4Þ T A C as shown in Supplementary Fig. 8. The depletion width of VTD- and RTE-fabricated devices decreased from 430 and 324 nm to here N is the net acceptor concentration in Sb Se film, which 240 and 255 nm with the bias pulse changing from −0.5 V to 0.4 A 2 3 can be obtained from C-V profiling (Fig. 4i); ΔC equals to the V, respectively. However, the Cd diffusion depths in VTD- and max difference between C and C (Fig. 5a). RTE-fabricated Sb Se layers are about 200 nm and 100 nm from t0+ 0 2 3 Based on the above fitting results and calculation, the defect SIMS results, respectively. Thus, the Cd ions located outside the parameters of VTD- and RTE-fabricated Sb Se solar cells are detected depletion region and had no effect on DLTS result. The 2 3 summarized in Table 1. The properties of deep defects in Sb Se defects only originate from the intrinsic Sb Se . 2 3 2 3 film are experimentally uncovered. By comparing the activation In our VTD and RTE facilities, Se vapor is always excess for its energies of these trap defects, we find that they are similar in both higher vapor pressure than Sb and Sb Se ,so the Sb Se films are 2 3 2 3 1, 2 devices, indicating the same origins of these defects. To actually slightly Se rich . Our previous ab initio calculation of the investigate whether Cd diffusion affects the DLTS results, we intrinsic defects in Sb Se has demonstrated that the dominant 2 3 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 7 | | | DLTS (ΔC/C ) 2 2 ln(e / T ) or ln(e / T ) n p ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ab VTD-Sb Se RTE-Sb Se 2 3 2 3 –4.04 eV –4.07 eV E1: ΔE = 0.61 eV E1: ΔE = 0.60 eV E E F F –4.69 eV –4.69 eV H2: ΔE = 0.71 eV H2: ΔE = 0.74 eV H1: ΔE = 0.48 eV H1: ΔE = 0.49 eV –5.23 eV –5.26 eV cd E Sb Se Sb Se 2 3 2 3 Fn H2 H2 CdS E1 E1 Fp H1 H1 CdS E E V V Fig. 6 Influence of defect levels on the CdS/Sb Se solar cells. Energy states and defect level of a VTD-fabricated and b RTE-fabricated Sb Se films. Energy 2 3 2 3 band diagrams at CdS/Sb Se interface c in the dark and d under illumination 2 3 acceptor defects are antimony vacancy (V ) and selenium antisite We now discuss influence of defects in working conditions. Sb (Se ) defects under Se-rich condition . Therefore, we tentatively The position of E is always dependent on illumination intensity, Sb Fp attributed H1 and H2 defects to V and Se defects, respectively. and Fig. 6d revealed the situation under AM1.5 irradiation. The Sb Sb As reported by Tumelero et al. , the antisite defects dominated the intersections of E and defect levels can be used as boundaries to distribution of defects in trichalcogenides due to the similar sizes of differentiate whether the defects are charged or not during the the constituent atoms. Our previous simulation also showed that shifting of quasi-Fermi level . Clearly, H1 state is under Fermi Se and antimony antisite (Sb ) are acceptor and donor defects, level and submersed in electrons, so H1 defects always stay inert. Sb Se respectively . Consequently, the E1 defect is most likely associated In contrast, H2 and E1 defect states are mostly above E and they with the formation of Sb antisite defects. Interestingly, E1 and H2 are active in trapping holes and electrons, respectively. These Se always have similar densities with each other in both VTD- and trapped photo-generated carriers would most probably contribute RTE-fabricated samples, which indicate that antisite defect pairs to recombination loss. Furthermore, the energy levels of H2 and formed in Sb Se films, presumably forming [Sb +Se ]complex. E1 are located near to the midgap, which significantly increase the 2 3 Se Sb Please also note that using VTD process reduced the density of recombination possibility of photo-generated carriers . There- these antisite defect pairs by more than an order of magnitude fore, the H2 and E1 are the dominant defects that influence the (Table 1). shift of quasi-Fermi levels and trap-assisted recombination, and The energy levels of defects in the two samples are depicted in then the V and J of the solar cells. Moreover, due to the OC SC Fig. 6a, b, respectively. Herein, the conduction band, valence band higher defect density of H2 and E1 than carrier concentration 13 −3 3 and Fermi levels were obtained from ultraviolet photoelectron (about 10 cm )inSb Se layer , the E would be more likely 2 3 Fp spectroscopy (UPS) and Tauc plots by transmission spectra of to be pinned near E1 and H2 levels. Obviously, both of H2 and E1 Sb Se films (Supplementary Fig. 9). Clearly, the energy levels of have lower defect density in VTD-fabricated sample, which could 2 3 H2 and E1 are above the Fermi level (E ) in both samples, and H1 lead to relatively larger E downshifting and suppressed trap- F Fp is under E . To analyze the influence of defects on the photo- assisted recombination, explaining the better V and J in F OC SC generated carrier recombination, energy band and defect energy VTD-fabricated Sb Se solar cells. 2 3 levels of CdS/Sb Se device in the dark and under illumination are 2 3 depicted in Fig. 6c, d, respectively. Due to thermal equilibrium, Discussion the energy band of Sb Se bended downward around the hetero- 2 3 We have demonstrated that VTD technique can greatly enhance the interface, which led to the same Fermi level in CdS and Sb Se 2 3 performance of Sb Se thin film solar cells by increasing the crys- 2 3 layers. Under illumination, as depicted in Fig. 6d, the photo- tallinity of Sb Se films, reducing the interface and bulk defects in the 2 3 generated electrons were driven into CdS layer, prompting devices, and prolonging the carrier lifetime. Specifically, we believe electron quasi-Fermi level (E ) to move upward in CdS. Fn the advantages of VTD over RTE are based on two features. First, Meanwhile the photo-generated holes resulted in the hole the substrate temperature in VTD process can be regulated inde- quasi-Fermi level (E ) shifting downward in Sb Se layer. The Fp 2 3 pendently by changing the distance between source and substrate, shift of quasi-Fermi levels is positively correlated with the thus permitting higher substrate temperature (VTD 390 °C vs. RTE nonequilibrium carrier concentration. The difference between the 350 °C). The higher substrate temperature resulted in improved two quasi-Fermi levels determines V of the solar cell . OC crystallinity (threefold enhancement in XRD peak intensity) and 8 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE average grain size (VTD 382 nm vs. RTE 297 nm) of Sb Se films. Material characterization. Material and device characterization are similar to 2 3 1, 2, 11 previous report . XRD of Sb Se films was performed using a Philips X’Pert 2 3 Second, the slower deposition of Sb Se films (VTD 2 min vs. RTE 2 3 Pro diffractometer with Cu Kα radiation (λ = 1.54 Å). SEM measurement was 35 s) enables less film imperfection (Se +Se antisite complex and Sb Sb carried out with FEI Nova Nano SEM450. SIMS (IMS-4f, CAMECA instruments) V vacancy), as reflected by the reduced bulk defect density (VTD Sb was used to investigate the element distribution along the depth in the devices. UPS 14 −3 15 −3 10 cm vs. RTE 10 cm ) and interfacial defect density (VTD was used to investigate the Fermi level and valence band of Sb Se films. Experi- 2 3 10 −2 11 −2 ments were performed using a He I (21.21 eV) gas discharge lamp in a Kratos 2.8 × 10 cm vs. RTE 2.1 × 10 cm ), and increased carrier AXIS-ULTRA DLD-600W x-ray photoelectron spectroscopy measurement system lifetime (VTD 1339 ps vs. 1149 ps). Overall, all these advantages and recorded at 0 V samples bias in an ultrahigh vacuum chamber. The surface of facilitated the fabrication of high-quality Sb Se films, leading to 2 3 Sb Se films was etched before UPS measurement. Ultraviolet–visible near-infrared 2 3 −2 increased V , J as well as FF (VTD: 0.42 V, 29.9 mA cm , 60.4% transmission spectra (Perkin Elmer Instrument, Lambda 950 using integrating OC SC −2 sphere) were measured to determine the bandgap of Sb Se films. vs. RTE: 0.39 V, 25.3 mA cm , 56.4%). The champion device 2 3 achieved the efficiency record 7.6%, much higher than SnS, Cu O 47–49 Device characterization. J-V curve and PCE of the champion CdS/Sb Se solar and FeS solar cells which have been studied for many years . 2 3 cell was independently measured by National Institute Metrology (NIM), using the The fabrication of record efficiency Sb Se solar cells employing 2 3 Class AAA Solar Simulator with double-light source (SAN-EI ELECTRIC, XHS- vapor transport deposition, a technique with proven high turn- −2 2350M1, 100 mW cm , AM1.5 illumination) in air ambient at room temperature. around and low cost for commercial CdTe solar cells, further A metal mask was used to define the area of incident light. The area (9.099 mm ) strengthens the great potential of our Sb Se thin film photovoltaics. was also measured by NIM. The light intensity was calibrated by a standard Si- 2 3 reference solar cell. A Keithley 2400 Source Meter was used to acquire J-V data. On the other hand, we find that the deep defect density in the The external quantum efficiency of solar cells was measured using the light source 14 15 −3 best-performing Sb Se solar cells is about 10 to 10 cm , 2 3 generated by a 300 W xenon lamp of Newport (Oriel, 69911) and then split into 11 13 −3 which is much higher than that (10 to 10 cm ) in CdTe solar specific wavelengths by a Newport Oriel Cornerstone 130 1/8 Monochromator cells . These abundant deep defects could pin the quasi-Fermi (Oriel, model 74004). A standard silicon detector (70356_70316NS_455) was used for the calibration. The capacitance-voltage (C-V) profiling and DLCP data were level near H2 and E1 defect states, which provides a plausible measured using Keithley 4200. C-V measurements were performed at room tem- explanation for the low V observed in all Sb Se solar cells OC 2 3 perature in an electromagnetic shielding box at a frequency of 100 kHz and a.c. reported so far, despite various device configuration and film amplitude of 30 mV. The d.c. bias voltage was scanned from −1.0 V to 0.3 V. 1–12, 19, 32, 51 preparation methods have been explored . We suggest DLCP measurements were performed with a.c. amplitude from 0.02 V to 0.14 V and d.c. bias voltage from −0.2 V to 0.2 V. DLTS measurement were performed by that future research on tightly controlling Se and Sb components Semetrol DLTS system (Semetrol, LLC, USA) on the two typical devices. The in the vapor, their strictly stoichiometric condensation into the −1 temperature was scanned between 100 and 380 K, at a heating rate of 2 K min . film and the growth of highly crystalline film should be carried out The reverse bias voltage was −0.5 V. The filling pulse voltage and width were 0.4 V to minimize these deep defects and maximize device performance. and 1 ms, respectively. At every scanning temperature point, the transient capa- citance was measured 20 times for obtaining the average value. Transient In summary, we have obtained the superstrate CdS/Sb Se thin 2 3 absorption spectroscopy was pumped by 500 nm laser pulses (Light Conversion, film solar cell with a certified efficiency of 7.6% (V of 0.42 V, OC Pharos, 350 fs duration pulses, 5 kHz repetition rate) and probed at 940 nm. The −2 J of 29.9 mA cm and FF of 60.4%), which was promoted by SC time delay was adjusted by changing the path length of the probe. the VTD-fabricated Sb Se absorber layer. Compared with the 2 3 1–6, 8–12, 19, 20 previous reports on Sb Se solar cells , VTD process 2 3 Data availability. The data that support the findings of this study are available reduced the density of deep defects and subsequently suppressed from the corresponding author on request. trap-assisted recombination in Sb Se films, resulting in longer 2 3 carrier lifetime and better device performance. These encouraging Received: 22 December 2017 Accepted: 19 April 2018 results highlight the potential of Sb Se solar cells for high- 2 3 efficiency photovoltaic devices. Methods References Sb Se solar cell fabrication. All solar cells were deposited on ITO (In O :Sn) 2 3 2 3 1. Wang, L. et al. Stable 6%-efficient Sb Se solar cells with a ZnO buffer layer. 2 3 transparent conductive glass supplied by Kaivo, with sheet resistance of 6.5 to 6.8 Nat. Energy 2, 17046 (2017). −1 ohm sq , transmittance of 78.8 to 79.6% and ITO thickness of about 200 nm. The 2. Zhou, Y. et al. Thin-film Sb Se photovoltaics with oriented one-dimensional 2 3 ITO substrates were cleaned using detergent, acetone, isopropanol, ethanol and ribbons and benign grain boundaries. Nat. Photon. 9, 409–415 (2015). deionized water in sequence. CdS buffer layer was deposited by chemical bath 3. Chen, C. et al. Characterization of basic physical properties of Sb Se and its 7 −1 2 3 deposition . CdS layers were treated with H O (30 wt%) and 20 mg ml CdCl 2 2 2 relevance for photovoltaics. Front. Optoelectron. 10,18–30 (2017). (Aladdin) absolute methanol solution by spin coating, respectively, baked on the 4. Liu, X. et al. Thermal evaporation and characterization of Sb Se thin film for 2 3 hotplate at 400 °C for 5 min in air ambient and then cooled down naturally. Fol- substrate Sb Se /CdS solar cells. ACS Appl. Mater. Interfaces 6, 10687–10695 2 3 lowing that, VTD process was used to deposit Sb Se films. As shown in Fig. 1a, 2 3 (2014). 0.25 g Sb Se powder (99.999% purity, Jiangxi Ketai) was put into a quartz crucible 2 3 5. Leng, M. et al. Selenization of Sb Se absorber layer: an efficient step to 2 3 and placed in the center of VTD system (a single temperature zone tube furnace, improve device performance of CdS/Sb Se solar cells. Appl. Phys. Lett. 105, 2 3 MTI, Hefei, China). The ITO/CdS substrate was immobilized on a graphite support 083905 (2014). and then placed at the right end of the quartz tube. Substrate temperature was 6. Luo, M. et al. Thermal evaporation and characterization of superstrate CdS/ regulated by changing the distance between substrate and the center of the heater. Sb Se solar cells. Appl. Phys. Lett. 104, 173904 (2014). Vacuum was pumped by a mechanical pump and the stabilized chamber pressure 2 3 7. Wang, L. et al. Ambient CdCl treatment on CdS buffer layer for improved was controlled by varying the ventilation power of the pump. The heating tem- 2 performance of Sb Se thin film photovoltaics. Appl. Phys. Lett. 107, 143902 perature of VTD system was raised to the targeted evaporation temperature with a 2 3 −1 ramp rate of 20 °C min and kept for 2 min to obtain a desired Sb Se film (2015). 2 3 thickness. Then, we turned off the power to stop the deposition and finally took the 8. Liu, X. et al. Improving the performance of Sb Se thin film solar cells over 4% 2 3 sample out when it was naturally cooled down to about 100 °C. After that, gold by controlled addition of oxygen during film deposition. Prog. Photo. Res. back-contact electrodes (0.091 cm area, 100 nm thick) were evaporated by the Appl. 23, 1828–1836 (2015). resistance evaporation thin film system (Beijing Technol Science) under a vacuum 9. Li, Z. et al. Sb Se thin film solar cells in substrate configuration and the back 2 3 −3 pressure of 5 × 10 Pa. To optimize the quality of Sb Se film and the performance contact selenization. Sol. Energy Mater. Sol. Cells 161, 190–196 (2017). 2 3 of device, we systematically investigated the evaporation temperature of VTD 10. Yuan, C., Jin, X., Jiang, G., Liu, W. & Zhu, C. Sb Se solar cells prepared with 2 3 system from 480 to 520 °C, the pressure from 2.8 to 5.3 Pa and the substrate selenized dc-sputtered metallic precursors. J. Mater. Sci. Mater. Electron 27, distance from 19 to 23 cm. When the evaporation temperature was investigated, 8906–8910 (2016). the pressure and the distance were set as 4 Pa and 22 cm, respectively. The opti- 11. Liu, X. et al. Enhanced Sb Se solar cell performance through theory-guided 2 3 mized evaporation temperature and the distance of 22 cm were adopted for pres- defect control. Prog. Photo. Res. Appl. 25, 861–870 (2017). sure investigation, and then the optimized temperature and pressures were used to 12. Chen, C. et al. 6.5% certified Sb Se solar cells using PbS colloidal quantum 2 3 investigate the substrate distance. dot film as hole transporting layer. ACS Energy Lett. 2, 2125–2132 (2017). NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 13. Leijtens, T. et al. Carrier trapping and recombination: the role of defect 43. Dharmarasu, N. et al. Majority- and minority-carrier deep level traps in physics in enhancing the open circuit voltage of metal halide perovskite solar proton-irradiated n+/p-InGaP space solar cells. Appl. Phys. Lett. 81,64–66 cells. Energy Environ. Sci. 9, 3472–3481 (2016). (2002). 14. Heo, S. et al. Deep level trapped defect analysis in CH NH PbI perovskite 44. Tumelero, M. A., Faccio, R. & Pasa, A. A. Unraveling the native conduction of 3 3 3 solar cells by deep level transient spectroscopy. Energy Environ. Sci. 10, trichalcogenides and its ideal band alignment for new photovoltaic interfaces. 1128–1133 (2017). J. Phys. Chem. C 120, 1390–1399 (2016). 15. Lang, D. V. Deep-level transient spectroscopy: a new method to characterize 45. Boix, P. P. et al. Determination of gap defect states in organic bulk traps in semiconductors. J. Appl. Phys. 45, 3023–3032 (1974). heterojunction solar cells from capacitance measurements. Appl. Phys. Lett. 16. Yang, W. S. et al. Iodide management in formamidinium-lead-halide-based 95, 233302 (2009). perovskite layers for efficient solar cells. Science 356, 1376–1379 (2017). 46. Sherkar, T. S. et al. Recombination in perovskite solar cells: significance of 17. Kestner, J. M. et al. An experimental and modeling analysis of vapor transport grain boundaries, interface traps, and defect ions. ACS Energy Lett. 2, deposition of cadmium telluride. Sol. Energy Mater. Sol. Cells 83,55–65 1214–1222 (2017). (2004). 47. Sinsermsuksakul, P. et al. Overcoming efficiency limitations of SnS-Based 18. Wangperawong, A. et al. Bifacial solar cell with SnS absorber by vapor solar cells. Adv. Energy Mater. 4, 1400496 (2014). transport deposition. Appl. Phys. Lett. 105, 173904 (2014). 48. Minami, T., Nishi, Y. & Miyata, T. Heterojunction solar cell with 6% efficiency 19. Ngo, T. T. et al. Electrodeposition of antimony selenide thin films and based on an n-type aluminum-gallium-oxide thin film and p-type sodium- application in semiconductor sensitized solar cells. ACS Appl. Mater. doped Cu O sheet. Appl. Phys. Express 8, 022301 (2015). Interfaces 6, 2836–2841 (2014). 49. Prabukanthan, P., Thamaraiselvi, S. & Harichandranb, G. Single step 2+ 20. Choi, Y. C. et al. Sb Se -Sensitized inorganic-organic heterojunction solar cells electrochemical deposition of p-type undoped and Co doped FeS thin films 2 3 2 fabricated using a single-source precursor. Angew. Chem. Int. Ed. 53, and performance in heterojunction solid solar cells. J. Electrochem. Soc. 164, 1329–1333 (2014). D581–D589 (2017). 21. Mattox, D. M. Particle bombardment effects on thin-film deposition: A 50. Versluys, J., Clauws, P., Nollet, P., Degrave, S. & Burgelman, M. review. J. Vac. Sci. Technol. A 7, 1105–1114 (1989). Characterization of deep defects in CdS/CdTe thin film solar cells using deep 22. Rossnagel, S. M. Thin film deposition with physical vapor deposition and level transient spectroscopy. Thin Solid Films 451-452, 434–438 (2004). related technologies. J. Vac. Sci. Technol. A 21, S74–S87 (2003). 51. Liang, G. X. et al. Thermally induced structural evolution and performance of 23. Lin, X. Z. et al. 11.3% efficiency Cu(In,Ga)(S,Se) thin film solar cells via drop- Sb Se films and nanorods prepared by an easy sputtering method. Sol. Energy 2 2 3 on-demand inkjet printing. Energy Environ. Sci. 9, 2037–2043 (2016). Mater. Sol. Cells 174, 263–270 (2018). 24. Quilettes, D. W. et al. Impact of microstructure on local carrier lifetime in perovskite solar cells. Science 348, 683–686 (2015). Acknowledgements 25. Sun, Z. H., Sitbon, G., Pons, T., Bakulin, A. A. & Chen, Z. Y. Reduced carrier This work was financially supported by the Major State Basic Research Development recombination in PbS-CuInS quantum dot solar cells. Sci. Rep. 5, 10626 Program of China (2016YFA0204000), the National Natural Science Foundation of (2015). China (61725401, 91433105 and 51602114), the Special Fund for Strategic New Devel- 26. Zhao, B. D. et al. High open-circuit voltages in Tin-rich low-bandgap opment of Shenzhen, China (JCYJ20160414102210144), and the Project funded by China perovskite-based planar heterojunction photovoltaics. Adv. Mater. 29, Postdoctoral Science Foundation (2017M622445). The authors thank the Analytical and 1604744 (2017). Testing Center of HUST and the facility support of the Center for Nanoscale Char- 27. Gao, F. et al. Trap-induced losses in hybrid photovoltaics. ACS Nano. 8, acterization and Devices, WNLO. We also thank J. Luo and J. Li for X-ray diffraction 3213–3221 (2014). measurements, Q. Hu for his help in SEM measurements, and Y. Zhong and S. Sun from 28. Abbaszadeh, D. et al. Elimination of charge carrier trapping in diluted Zhengzhou University for their help in DLTS measurements. semiconductors. Nat. Mater. 15, 628–633 (2016). 29. Szendrei, K., Gomulya, W., Yarema, M., Heiss, W. & Loi, M. A. PbS nanocrystal solar cells with high efficiency and fill factor. Appl. Phys. Lett. 97, Author contributions 203501 (2010). X.W. and J.T. conceived the idea, designed the experiments and analyzed the data. X.W., 30. Li, Y. W. et al. High-efficiency robust perovskite solar cells on ultrathin C.C., and S.L. carried out the device optimizations. Y.Z. prepared the gold electrodes. X. flexible substrates. Nat. Commun. 7, 10214 (2016). W. carried out most of material and device characterizations. L.G. carried out the TA 31. Jahandar, M. et al. Highly efficient metal halide substituted CH NH I(PbI ) 3 3 2 1- characterization. C.C., S.L., K.L., W.C., R.K., C.W., J.Z. and G.N. assisted in data analysis. (CuBr ) planar perovskite solar cells. Nano Energy 27, 330–339 (2016). x 2 x X.W. and J.T wrote the paper. All authors commented on the manuscript. 32. Zhou, Y. et al. Buried homojunction in CdS/Sb Se thin film photovoltaics 2 3 generated by interfacial diffusion. Appl. Phys. Lett. 111, 013901 (2017). 33. Kumar, S. G. & Rao, K. S. R. K. Physics and chemistry of CdTe/CdS thin film Additional information Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- heterojunction photovoltaic devices: fundamental and critical aspects. Energy Environ. Sci. 7,45–102 (2014). 018-04634-6. 34. Liu, F. Y. et al. Nanoscale microstructure and chemistry of Cu ZnSnS /CdS 2 4 interface in Kesterite Cu ZnSnS solar cells. Adv. Energy Mater. 6, 1600706 Competing interests: The authors declare no competing interests. 2 4 (2016). 35. Heath, J. T., Cohen, J. D. & Shafarman, W. N. Bulk and metastable defects in Reprints and permission information is available online at http://npg.nature.com/ CuIn Ga Se thin films using drive-level capacitance profiling. J. Appl. Phys. reprintsandpermissions/ 1−x x 2 95, 1000–1010 (2004). 36. Duan, H. S. et al. The role of sulfur in solution-processed Cu ZnSn(S,Se) and Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in 2 4 its effect on defect properties. Adv. Funct. Mater. 23, 1466–1471 (2013). published maps and institutional affiliations. 37. Auret, F. D. & Nel, M. Detection of minority-carrier defects by deep level transient spectroscopy using Schottky barrier diodes. J. Appl. Phys. 61, 2546–2549 (1987). 38. Fourches, N. A quantitative treatment for deep level transient spectroscopy Open Access This article is licensed under a Creative Commons under minority-carrier injection. J. Appl. Phys. 70, 209–214 (1991). Attribution 4.0 International License, which permits use, sharing, 39. Kerr, L. L. et al. Investigation of defect properties in Cu(In,Ga)Se solar cells adaptation, distribution and reproduction in any medium or format, as long as you give by deep-level transient spectroscopy. Solid-State Electron 48, 1579–1586 appropriate credit to the original author(s) and the source, provide a link to the Creative (2004). Commons license, and indicate if changes were made. The images or other third party 40. Fleming, R. M., Seager, C. H., Lang, D. V. & Campbell, J. M. Injection deep material in this article are included in the article’s Creative Commons license, unless level transient spectroscopy: An improved method for measuring capture rates indicated otherwise in a credit line to the material. If material is not included in the of hot carriers in semiconductors. J. Appl. Phys. 118, 015703 (2015). article’s Creative Commons license and your intended use is not permitted by statutory 41. Khan, A. et al. DLTS: A promising technique for the identification of the regulation or exceeds the permitted use, you will need to obtain permission directly from recombination and compensator centers in solar cell materials. 1763–1768 the copyright holder. To view a copy of this license, visit http://creativecommons.org/ (Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic licenses/by/4.0/. Energy Conversion, Waikoloa, 2006). 42. Nguyen, T. P. Defects in organic electronic devices. Phys. Stat. Sol. (a) 205, 162–166 (2008). © The Author(s) 2018 10 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

Vapor transport deposition of antimony selenide thin film solar cells with 7.6% efficiency

Free
10 pages

Loading next page...
 
/lp/springer_journal/vapor-transport-deposition-of-antimony-selenide-thin-film-solar-cells-f2HCdzqT4z
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Science, Humanities and Social Sciences, multidisciplinary; Science, Humanities and Social Sciences, multidisciplinary; Science, multidisciplinary
eISSN
2041-1723
D.O.I.
10.1038/s41467-018-04634-6
Publisher site
See Article on Publisher Site

Abstract

ARTICLE DOI: 10.1038/s41467-018-04634-6 OPEN Vapor transport deposition of antimony selenide thin film solar cells with 7.6% efficiency 1,2 1,2 1,2 1,2 1,2 1,2 Xixing Wen , Chao Chen , Shuaicheng Lu , Kanghua Li , Rokas Kondrotas , Yang Zhao , 1,2 1,2 1,2 1,2 1,2 1,2 Wenhao Chen , Liang Gao , Chong Wang , Jun Zhang , Guangda Niu & Jiang Tang Antimony selenide is an emerging promising thin film photovoltaic material thanks to its binary composition, suitable bandgap, high absorption coefficient, inert grain boundaries and earth-abundant constituents. However, current devices produced from rapid thermal eva- poration strategy suffer from low-quality film and unsatisfactory performance. Herein, we develop a vapor transport deposition technique to fabricate antimony selenide films, a technique that enables continuous and low-cost manufacturing of cadmium telluride solar cells. We improve the crystallinity of antimony selenide films and then successfully produce superstrate cadmium sulfide/antimony selenide solar cells with a certified power conversion efficiency of 7.6%, a net 2% improvement over previous 5.6% record of the same device configuration. We analyze the deep defects in antimony selenide solar cells, and find that the density of the dominant deep defects is reduced by one order of magnitude using vapor transport deposition process. Sargent Joint Research Center, Wuhan National Laboratory for Optoelectronics and School of Optical and Electronic Information, Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan 430074, Hubei, China. Shenzhen R&D Center of Huazhong University of Science and Technology, Shenzhen 518000, China. Correspondence and requests for materials should be addressed to J.T. (email: jtang@mail.hust.edu.cn) NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ntimony selenide (Sb Se ) has recently emerged as a fast-turnaround manufacturing method for commercial CdTe 2 3 17, 18 promising green alternative to CdTe solar cells because it solar cells , and adopting this technique for Sb Se technology 2 3 Apossesses very attractive optoelectronic properties such as could facilitate the commercial competitiveness of Sb Se solar 2 3 proper bandgap (about 1.03 eV indirect and 1.17 eV direct) for cells. After optimizing the Sb Se films via VTD process, we 2 3 the absorption of a significant portion of the solar spectrum, high obtain the champion indium tin oxide (ITO)/CdS/Sb Se /Au 2 3 5 −1 optical absorption coefficient (greater than 10 cm ) and decent solar cells with a certified efficiency of 7.6%, which is a record of 2 −1 −1 1–3 1, 2, 4–12, 19, 20 carrier mobility (about 10 cm V s ) . Besides, because it is all Sb Se thin film solar cells by far and represents a 2 3 a binary compound with high vapor pressure, a fast, low tem- net 2% efficiency gain over previous report with the same device perature vacuum-based deposition technique can be employed configuration . We comparatively analyze the physical properties like the ones established in CdTe photovoltaics. It also possesses a of VTD-fabricated and RTE-fabricated Sb Se devices by X-ray 2 3 one-dimensional crystal structure with loose van der Waals diffraction (XRD), transient absorption (TA) spectroscopy and interaction between ribbons , thereby enabling grain boundaries deep-level transient spectroscopy (DLTS). We find that the VTD- free of dangling bonds in c-axis-oriented films and minimizing fabricated devices possess much higher film crystallinity, longer recombination losses therein. Furthermore, the earth-abundant carrier lifetime and fewer bulk and interfacial defects than RTE- elemental compositions of Sb Se as well as its easy fabrication fabricated devices. Especially, the reduced density of defect 2 3 promise low-cost manufacturing. All these attributes suggest the complex (Sb +Se ) enhances the device performance by Se Sb great potential of Sb Se for high-efficiency thin film solar cell decreasing the photo-generated carrier recombination. 2 3 and commercial application. However, the highest power conversion efficiency (PCE) of Results Sb Se thin film solar cells with a CdS/Sb Se superstrate con- 2 3 2 3 Fabrication of Sb Se films via VTD process. A schematic 2 3 2, 4–11 figuration is so far 5.6% , and with a ZnO/Sb Se superstrate 2 3 representation of VTD system is shown in Fig. 1a. The Sb Se 2 3 configuration 5.93% . Employing PbS colloidal quantum dots powder and the substrate were placed in the center and at the (QDs) as the hole transport layer, as-fabricated device CdS/ right end of the heater, respectively. The full deposition details are Sb Se /PbS QDs demonstrated an efficiency of 6.5% . Rapid 2 3 provided in experimental section. Hereby, the heating tempera- thermal evaporation (RTE) was used to fabricate all above- ture of the evaporation source, the pressure in quartz tube and the mentioned Sb Se solar cells. RTE is a process similar to close- 2 3 substrate temperature are the key factors that determining the space sublimation , in which the vapor is limited to a confined quality of final Sb Se films. The distance between the substrate 2 3 space and rapidly deposits on the substrate. The distance from the and the heating center was varied to achieve adjustable substrate substrate to the evaporation source is only 0.8 cm, and the whole temperature. We deposited Sb Se films at different heating 2 3 deposition is performed within 35 s. The small confined space temperatures, pressures and substrate temperatures. XRDs of all and the rapid deposition increase the difficulty of mixing the Sb Se films were measured to characterize the crystallinity and 2 3 vapor particles (Se, Sb and Sb Se ) evenly, potentially promoting x y orientation of these films, which is shown in Supplementary defect formation. The formation of defects, such as interstitial and Fig. 1. All the Sb Se films had a preferred [221] orientation with 2 3 antisite defects, accelerates nonradiative recombination and the only difference being the intensity of the diffraction peaks. As 13–16 degrades device performance . Furthermore, limited by our Sb Se is composed of one-dimensional (Sb Se ) ribbons stack- 2 3 4 6 n RTE facility, a simple one-zone tube furnace, the source and ing together via weak van der Waals force, a proper film orien- 1, 2 substrate temperatures are strongly correlated, which seriously tation is crucial for facile carrier transport in the film . Our hinders the independent optimization of both substrate and previous reports have demonstrated that [221]-orientation 1, 2, 6, 8 source temperature. These problems in our RTE technique have enhances carrier transport across Sb Se film . Therefore, 2 3 seriously restricted the development of Sb Se solar cells. There- 2 3 with the film thickness and measurement details kept strictly fore, we urgently need to explore alternative strategies to further identical, we utilized the intensity of (221) XRD peaks in Sb Se 2 3 improve Sb Se film quality and device performance. 2 3 films to evaluate their crystallinity. The crystallinity evolution of Here, we develop a vapor transport deposition (VTD) process Sb Se films deposited on CdS layers with varied evaporation 2 3 to fabricate record efficiency Sb Se thin film solar cells. In the 2 3 temperature, pressure and distance was investigated, as shown in VTD process, both the substrate temperature and the distance Fig. 1b. By carefully optimizing the three key factors, Sb Se film 2 3 between source and substrate are adjustable, enabling not only with the highest crystallinity was obtained when the evaporating highly oriented Sb Se film, but also enormously improved film 2 3 temperature was set at 510 °C, the pressure was 3.2 Pa and the crystallinity and reduced bulk and interfacial defects in Sb Se 2 3 distance from the substrate to the heating center was 21 cm. Here, solar cells. In addition, the VTD process is a proven low-cost and the actual temperature curves of powder and substrate were ab c Heater Au Sb Se vapor 2 3 Sb Se 2 3 Pump CdS Au Sb Se powder Substrate 2 3 ITO Glass 480 500 520 34 5 23 21 19 Temperature (°C) Pressure (Pa) Distance (cm) Fig. 1 Fabrication of CdS/Sb Se solar cells. a Schematics of our VTD system. b Evolution of crystallinity with the optimization of deposition condition. c 2 3 The device structure of our Sb Se solar cells 2 3 2 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | Evolution of crystallinity XRD intensity of (221) peak (×10 counts) NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab 0.44 0.42 0.40 0.38 0.36 0.34 480 500 520 34 5 19 21 23 480 500 520 345 19 21 23 Temperature (°C) Pressure (Pa) Distance (cm) Temperature (°C) Pressure (Pa) Distance (cm) cd 28 60 18 45 480 500 520 345 19 21 23 480 500 520 345 19 21 23 Temperature (°C) Pressure (Pa) Distance (cm) Temperature (°C) Pressure (Pa) Distance (cm) Fig. 2 Deposition condition-dependent photovoltaic characteristics. a Power conversion efficiency (PCE), b open-circuit voltage (V ), c short-circuit OC current density (J ) and d fill factor (FF) of the CdS/Sb Se solar cells fabricated by VTD process. A total of 135 devices are included for the statistics SC 2 3 analysis. Solid sphere symbols and error bars indicate average values and standard deviations, respectively monitored during the heating and cooling steps. The input Device performance. The Sb Se absorber layers were fabricated 2 3 temperature program and the measured temperature curves of under the above-mentioned conditions, and the corresponding substrate and powder are shown in Supplementary Fig. 2. This device structure ITO/CdS/Sb Se /Au is shown in Fig. 1c. The 2 3 heating profile and substrate distance corresponded to the max- dependence of device performance on different evaporation imum actual source and substrate temperatures of 540 °C and temperatures, pressures and distances is summarized in Fig. 2. 390 °C, respectively, as monitored by the thermocouples. The variation trend of PCE is similar to the crystallinity evolution The crystallinity evolution of Sb Se films indicates the of Sb Se films (Fig. 1b), which indicates that device performance 2 3 2 3 importance of evaporation temperature, pressure and substrate strongly correlates with the crystallinity of Sb Se film. As can be 2 3 temperature in VTD process. High evaporation temperature seen from Fig. 2, the best PCE, open-circuit voltage (V ), short- OC increases the kinetic energy of vapor particles and the surface circuit current density (J ) and fill factor (FF) were obtained SC mobility of adatoms on the substrate . On the other hand, high under the deposition condition where the highest film crystal- pressure can increase in-flight collisions with background gas linity was achieved (Fig. 1b). After carefully optimizing Sb Se 2 3 21, 22 atoms, and then reduced the kinetic energy of vapor particles . films and devices, the champion device with a certified power In addition, increasing the substrate temperature will lead to the conversion efficiency of 7.6% was obtained (certificate included in increased surface mobility of adatoms . Hence, the kinetic Supplementary Fig. 3). This value represents the highest PCE of 1, 2, 4–12 energy of vapor particles and surface mobility of adatoms are all Sb Se thin film solar cells reported so far , which is 2% 2 3 ultimately defined by combination of evaporation temperature, higher than previous 5.6% certified efficiency with the same pressure and substrate temperature. During the Sb Se film device configuration . Figure 3a shows the light current density- 2 3 deposition, vapor particles with proper kinetic energy and voltage (J-V) curve of the champion VTD-fabricated CdS/Sb Se 2 3 adatoms with proper surface mobility are mandatory for high- solar cell with certified PCE of 7.6%, V of 0.42 V, J of 29.9 OC SC −2 quality Sb Se films. Energetic vapor particles and high mobility mA cm and FF of 60.4%. The J-V curve of a RTE-fabricated 2 3 −2 of adatoms may increase the instability of adatoms, causing solar cell with PCE of 5.6% (V = 0.39 V, J = 25.3 mA cm OC SC displacements of lattice and consequently creating lattice and FF = 56.4%) is also included in Fig. 3a for comparison. 21, 22 defects . Such defects may act as new nucleation sites and Obviously, every performance parameter of the VTD-fabricated increase nucleation density of adatoms . In the extreme, the solar cell, especially J , is larger than that of the RTE-fabricated SC increased lattice defects and nucleation density may result in device. decreased crystallinity and quality of the film. That is why the We further checked the external quantum efficiency (EQE) of crystallinity of the Sb Se film decreased when deposited at overly the two devices (Fig. 3b) to investigate their photo-response. The 2 3 high evaporation temperature, lower pressure and closer distance EQE spectrum of VTD-fabricated device demonstrated a higher to evaporation source. photo-response from 520 nm (the absorption onset of CdS layer) NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 3 | | | –2 J (mA cm ) SC PCE (%) V (V) OC FF (%) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 a b 100 35 VTD RTE 15 60 RTE VTD V (V): 0.39 0.42 OC –2 J (mAcm ): 25.3 29.9 SC FF (%): 56.4 60.4 PCE (%): 5.6 7.6 10 CdS/VTD-Sb Se 2 3 –5 CdS/RTE-Sb Se 2 3 0 0 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 300 400 500 600 700 800 900 1000 1100 Voltage (V) Wavelength (nm) c d 0.44 0.90 VTD slope = 1.23 k T/q VTD J ∝ I B SC 0.90 0.42 RTE slope = 1.51 k T/q RTE J ∝ I B SC 0.40 0.38 0.36 0.34 0.32 0.30 0.28 0.26 10 100 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 –2 –2 Intensity (mW cm ) ln[Intensity (mW cm )] Fig. 3 Device performance and light intensity-dependent J and V of devices. a The light J-V curves of VTD- and RTE-fabricated devices under AM1.5 G SC OC illumination. The J-V curve of the VTD-fabricated device with certified efficiency of 7.6% (area= 0.091 cm ) was measured by National Institute of Metrology on 1 September 2017. The calibration certificate number is GXtc2017-1987 (Supplementary Fig. 3). b EQE and integrated J of the VTD- and SC RTE-fabricated devices. Light intensity-dependent c J and d V . Neutral-density filters (THORLABS) were used to control the light intensity SC OC to 1100 nm than that of RTE-fabricated device. The maximum Figure 3c illustrates J as a function of light intensity (I). Herein, SC EQE of VTD-fabricated device was close to 90% between 540 and J was fitted according to the power law dependence (J ∝ I ) SC SC 720 nm. For comparison, the device fabricated by our RTE by the log–log scale plot. The power value α for both devices is technique only had the maximum EQE of about 85%. This 0.9, close to unity (first-order). This means that trap-assisted suggests the VTD-fabricated devices possess very low recombina- recombination is present in both solar cells and is the dominating 24, 25 tion losses of photo-generated carriers and long carrier lifetimes loss mechanism . The light-intensity-dependent V can OC at CdS/Sb Se interface and in the whole Sb Se absorber .In provide critical insights into the recombination mechanism in the 2 3 2 3 −2 26 Fig. 3b, we derived the current density of 29.8 and 25.1 mA cm solar cells . For V measurement, the device is open circuit, so OC by integrating the EQE spectra with standard AM1.5 spectrum to there is no current extraction from the devices, and all photo- further validate our J values, which agreed with the experi- generated carriers recombine in Sb Se film. Thus, the carrier SC 2 3 −2 mental values of 29.9 and 25.3 mA cm very well, respectively. recombination process can be reflected based on the relationship Furthermore, a solar cell with initial PCE of 7.25% was stored in of V ∝ n(k T/q)ln(I), where k is the Boltzmann constant, T is OC B B 26–29 ambient air for about 40 days without encapsulation. This the temperature and q is elementary charge . n(k T/q) is the representative device was measured every week to monitor the slope of V vs. the natural logarithm of light-intensity ln(I). For OC stability of Sb Se film solar cell, and the results are displayed in trap-free solar cells, the slope of V vs. ln(I) should be k T/q 2 3 OC B 26, 29 Supplementary Fig. 4a. The PCE remained unchanged during the (i.e., n = 1) . For our devices, as shown in Fig. 3d, the slopes whole process. For the stability of device under continuous obtained by linear fitting were 1.23(k T/q) for the VTD- illumination, Supplementary Fig. 4b shows slight decrease for fabricated device and 1.51(k T/q) for the RTE-fabricated device. CdS/Sb Se device but no decrease for ZnO/Sb Se device, which This, again, indicates the presence of trap-assisted 2 3 2 3 1 29, 30 is consistent with the previous report . These results indicate that Shockley–Read–Hall (SRH) recombination in both devices . VTD process is a simple and effective technique for producing The slope was decreased from 1.51(k T/q) to 1.23(k T/q) by using B B high-efficiency and stable Sb Se thin film solar cells. VTD process, suggesting reduced trap-assisted recombination in 2 3 28, 31 VTD-fabricated devices . We measured the TA decay of the two devices to investigate the carrier lifetime (Supplementary Carrier recombination in CdS/Sb Se solar cells. From the Fig. 6). For the steady-state absorption (Supplementary Fig. 7)of 2 3 VTD- and RTE-fabricated Sb Se films, the absorption rose to the improved EQE of VTD-fabricated devices, we inferred that the 2 3 carrier transport was enhanced, and recombination loss was maximum at around 940 nm. Thus, the transient kinetic decay was monitored at 940 nm. By fitting the kinetic decay data, the reduced in the active Sb Se layer by VTD process. Therefore, to 2 3 further clarify the carrier recombination processes in Sb Se solar longer carrier lifetime (1339 ps) was obtained in VTD-fabricated 2 3 cells, we performed light-intensity-dependent J and V mea- device than that (1149 ps) of RTE-derived device. Longer carrier SC OC surements on the VTD- and RTE-based photovoltaic devices. The lifetime not only permits more efficient carrier collection and −2 −2 complete sets of J-V curves are given in Supplementary Fig. 5. hence larger J (VTD 29.9 mA cm vs. RTE 25.3 mA cm ), SC 4 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | –2 Current density (mA cm ) –2 J (mA cm ) SC V (V) EQE (%) OC –2 J (mA cm ) SC NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab c VTD RTE 500 nm 500 nm RTE VTD 100 200 300 400 500 600 Grain size (nm) de f Sb Se 900 nm 2 3 Sb Se 900 nm 2 3 VTD CdS 70 nm CdS 70 nm ITO ITO RTE 500 nm 500 nm RTE VTD 10 20 30 40 50 60 2 (degrees) gh i Depth (nm) RTE VTD Depth (nm) 0 290 580 870 1160 1450 0 290 580 870 1160 1450 0 14 RTE DLCP Sb RTE CV profiling Se VTD DLCP ITO Cd VTD CV profiling Sb Se 10 Sb Se 2 3 Sn 2 3 –1 –1 10 10 CdS Sb Se ITO –2 –2 4 Cd 10 10 Sn CdS 0 200 400 600 800 1000 0 200 400 600 800 1000 150 200 250 300 350 Sputter time (s) Sputter time (s) Width (nm) Fig. 4 Characterization of Sb Se films and interface analysis of CdS/Sb Se devices. SEM top-view images of a VTD-fabricated and b RTE-fabricated 2 3 2 3 Sb Se films. c Histogram of grain size in VTD-fabricated and RTE-fabricated Sb Se films. Cross-sectional SEM images of d VTD-fabricated and e RTE- 2 3 2 3 fabricated CdS/Sb Se devices. f XRD of VTD-fabricated and RTE-fabricated Sb Se films. SIMS depth analysis of g VTD-fabricated and h RTE-fabricated 2 3 2 3 devices. i C-V profiling and DLCP for VTD- and RTE-fabricated devices but also enables higher carrier concentration within the absorber cases were the same: ITO/CdS/Sb Se /Au. The Sb Se layers were 2 3 2 3 layer and therefore wider quasi-Fermi level splitting and compact and well adherent to the CdS with the same thickness of improved V (VTD 0.42 V vs. RTE 0.39 V). about 900 nm. We compared XRD patterns of the two Sb Se OC 2 3 layers, measured under identical conditions, as shown in Fig. 4f. Obviously, both films were favorably orientated along [221] Morphology and crystallization of Sb Se films. We now study 2 3 direction. However, the XRD intensity of VTD-fabricated film the material origin of the improved device performance. The was much stronger than that of RTE-fabricated film, indicating scanning electron microscopy (SEM) top view images of Sb Se 2 3 much higher crystallinity was obtained by VTD process. The SEM films deposited on glass/ITO/CdS substrates by VTD and RTE and XRD results both demonstrate VTD-fabricated Sb Se films are shown in Fig. 4a, b, respectively. The grains of VTD- 2 3 have larger grains and higher crystallinity compared with RTE- fabricated film are obviously larger than that of Sb Se film from 2 3 fabricated films. It should be noted that, for RTE-fabricated RTE process. We statistically analyzed the distribution of grain Sb Se film, the distance between evaporation source and sub- size from the two SEM top view images. Figure 4c depicts the 2 3 strate was merely 0.8 cm, and the deposition was carried out on histograms of the grain size. The average grain size of VTD- 1, 2 350 °C substrate for 35 s , while the corresponding value in the fabricated film was 382 nm with a standard deviation of 79 nm, VTD process was 21 cm, 390 °C and 2 min, respectively. Longer whereas the RTE-derived Sb Se film had an average size of 297 2 3 traveling distance could increase the collision probability between nm and a standard deviation of 82 nm. The cross-sectional SEM Sb Se particles and gas molecules, reduce the momentum of images of Sb Se solar cells fabricated by VTD and RTE methods x y 2 3 these particles when impinge onto the substrate and hence are shown in Fig. 4d, e, respectively. Device structures in both NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 5 | | | Normalized SIMS Normalized SIMS 16 –3 N (×10 cm ) 3 Frequency (%) Intensity (×10 counts) (120) (130) (230) (211) (221) (301) (321) (041 ) (141) (431) (002) (061) (422) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 decrease the occurrence of nucleation. Less nucleus within the device are filled by carriers through applying a voltage pulse to film, accompanied with higher substrate temperature and longer the device, which changes the capacitance associated with p–n film growth duration, enabled the VTD-derived Sb Se film with junction of the device . 2 3 larger grain size and better grain crystallinity . We comparatively analyzed the deep-level defects in VTD- fabricated and RTE-fabricated Sb Se films by DLTS. Herein, 2 3 we adopted DLTS with minority carriers injection (inj-DLTS) Interfacial properties of the devices. For all thin film photo- 37, 38, 40, 43 to detect both electron and hole traps in Sb Se films . 2 3 voltaic devices with CdS as buffer layer, Cd diffusion at the As shown in Fig. 5a, during the measurement, a quiescent 32, 33 interface is always observed . To check the Cd diffusion in our reverse bias (V )was first applied to the junction for reverse devices, we measured their composition distribution by secondary forming a depletion width in Sb Se layer. Then, a forward 2 3 ion mass spectroscopy (SIMS) depth profiling. As shown in Fig. 4g, pulse voltage (V )was appliedto fill the traps in the depletion fill h, Cd diffusion was observed in both VTD-fabricated and RTE- region. If we set V < 0, the junction was under reverse bias, fill fabricated CdS/Sb Se devices. The deeper Cd diffusion (greater 2 3 and there were only holes injected into the depletion region to than 200 nm) than that (greater than 100 nm) in RTE-fabricated fill the traps, and only hole traps were detected. If we set V > fill device was found in VTD-fabricated device. As CdS buffer layer 0, under forward bias condition, both electrons and holes were was prepared following identical procedures, we deduced that the injected into the depletion region, and thus both electron and different Cd-diffused depth was certainly caused by the different 37, 38, 40 hole traps could be detected (inj-DLTS) . deposition techniques: for VTD process, substrate temperature was Here, we take the hole traps as an example to elaborate the about 390 °C and deposition lasted for 2 min, while in RTE process change of transient capacitance caused by holes capture and 1, 2 the corresponding values were 350 °C and 35 s, respectively . emission (Fig. 5a). Figure 5b shows the corresponding variation Because diffusion is driven thermally, higher substrate temperature of depletion width (W ) and the process of holes being trapped and longer deposition time resulted in more Cd diffusion in VTD- and emitted, before and after the pulse voltage applied. C 32 0 fabricated device. As we reported before , Cd diffusion converted represents the steady-state junction capacitance at V bias. reverse p-type Sb Se into n-type, and resulted in a buried homojunction 2 3 When V was applied and held for a while, W narrowed down fill d at CdS/Sb Se interface. This could reduce the interface defects and 2 3 to W and the hole trap defects were filled. Once V pulse 34 t0– fill recombination at the heterojunction interface , being beneficial relaxed, W broadened to W and capacitance decreased to C d t0+ t0 for device performance. instantaneously. W was even larger than W because some + t0+ 0 Therefore, we further measured interfacial defects using holes had been trapped in depletion region. Over the course of capacitance-voltage (C-V) profiling and deep-level capacitance time, the trapped holes were gradually and eventually completely 1, 35, 36 profiling (DLCP) techniques . Generally, the defects density emitted from the occupied deep level. Then, W shrank to W d 0 obtained from C-V profiling (N ) includes the response of free CV and the capacitance returned to steady state (C ). As the state of carriers, and bulk and interfacial defects, while the defect density activated defects was determined by the temperature, a sequence obtained from DLCP measurement (N ) represents the DLCP of transient capacitance was thus measured at different sample 1, 36 response only from the free carrier and the bulk defects . temperature, and capacitance changes within a time window vs. Thus, we can characterize the defect density at CdS/Sb Se 2 3 the different temperatures was sampled as DLTS signal. We took interface by the difference between N and N . As shown in CV DLCP a fixed time window between t and t (Fig. 5a), and then the 1 2 Fig. 4i, obviously, the difference between N and N of RTE- CV DLCP corresponding hole emission rate e can be expressed by Eq. (1): fabricated device is much higher than that of VTD-fabricated device, indicating lower defect density at VTD-fabricated CdS/ lnðÞ t =t 2 1 e ¼ : ð1Þ Sb Se interface. Because the doping concentration of CdS was 2 3 t  t 2 1 3, 6 much higher than in Sb Se film , almost all depletion width 2 3 (W ) extended in Sb Se layer. Here the volume to surface ratio is d 2 3 W , the interfacial defect density of RTE-fabricated and VTD- As shown in Fig. 5a, the capacitance change within the time 11 −2 fabricated devices was calculated to be about 2.1 × 10 cm and window is ΔC=C −C , which depends on the sample tempera- t2 t1 10 −2 2.8 × 10 cm , respectively. The interfacial defect density in ture. DLTS signal can be reflected from the variation of ΔC/C VTD-fabricated Sb Se solar cell is also much lower than that 2 3 with different sample temperatures. Besides, based on the 11 −2 (1.22 × 10 cm ) in RTE-fabricated Sb Se solar cell with ZnO 2 3 changing of capacitance during the discharging process of traps, as buffer layer . Consequently, we concluded that the promoted the hole traps and electron traps can be differentiated by positive interfacial diffusion had effectively reduced interface defects. This and negative ΔC, respectively. is reminiscent of our previous observation that the performance DLTS results of VTD-fabricated and RTE-fabricated CdS/ of RTE-fabricated and thermally evaporated CdS/Sb Se solar 2 3 Sb Se devices are shown in Fig. 5c. One negative and two 2 3 cells always improved when stored in ambient air for a few positive peaks were found in both devices, indicating one electron 2, 6 days . Similar effect has also been demonstrated in CdS/CdTe trap (E1) and two hole trap defects (H1 and H2) in Sb Se films. 2 3 solar cells that the interfacial diffusion reduced interfacial lattice The activation energy and capture cross-section of traps can be 15, 38 mismatch and defects, then reduced the current loss and obtained from Arrhenius plot based on Eqs. (2) and (3): improved device efficiency . e 16πk m B E  E p p T V ln ¼ ln σ  ; ð2Þ 2 3 T h k T Deep defects in VTD- and RTE-fabricated devices. We subse- B quently investigated the deep defects in CdS/Sb Se solar cells 2 3 fabricated via VTD and RTE processes to demonstrate how the e 16πk m E  E n B C T trap-assisted recombination was reduced in VTD-fabricated n ð3Þ ln ¼ ln σ  ; 2 3 T h k T device. DLTS is a well-accepted powerful tool to investigate defect energy level, type and concentration in thin film 14, 15, 37–40 photovoltaics . It uses the transient capacitance of p–n junction at different temperature as a probe to monitor the where e and e represents electron and hole emission rate, n p changes in charge state of a deep defect center . Traps in the respectively, which can be obtained by Eq. (1); σ and σ are p n 6 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE ab t t t 0 1 2 C=C C=C 0 t 0– Time fill fill W W 0 t 0– reverse t 0– C=C C=C t t 0+ ≈ ≈ t 2 t 0+ ΔC = C – C t 2 t 1 t 1 n-region Depletion region p-region t 0+ cd 0.25 –3 VTD-E1 0.20 RTE-H1 VTD-H2 VTD VTD-H2 –4 RTE VTD-H1 0.15 0.10 RTE-H2 –5 RTE-H1 VTD-H1 0.05 –6 0.00 –0.05 RTE-E1 RTE-E1 –7 –0.10 –0.15 –8 RTE-H2 VTD-E1 –0.20 160 200 240 280 320 360 32 36 40 44 48 Temperature (K) 1/k T Fig. 5 DLTS analysis of VTD- and RTE-fabricated CdS/Sb Se solar cells. a Schematic demonstration of the mechanism of DLTS measurement. b Variation 2 3 of depletion width and the process of holes being trapped and emitted during the measurement. c DLTS signals of VTD-fabricated and RTE-fabricated devices at t /t = 1 ms/10 ms. d Arrhenius plots obtained from DLTS signals. C and W are the junction capacitance and the depletion width at the 1 2 t0– t0– moment before pulse voltage ended, respectively. C and W are the junction capacitance and the depletion width at the moment after pulse voltage t0+ t0+ ended, respectively capture cross-section of hole and electron traps, respectively, T is the temperature, and k T is the thermal energy. In addition, m B p Table 1 Defect parameters of VTD- and RTE-fabricated and m respectively represent effective mass of hole and electron, Sb Se solar cells 2 3 and E , E and E are the energy level of defect, conduction and T C V valence bands, respectively. Figure 5d shows the Arrhenius plot 2 −3 Defects E (eV) σ (cm ) N (cm ) T T obtained from Fig. 5c, by the varied e (hole or electron emission −17 15 VTD-H1 E + 0.48 ± 0.07 1.5 × 10 1.2 × 10 rate) corresponding to the DLTS peak positions in temperature −13 14 VTD-H2 E + 0.71 ± 0.02 4.9 × 10 1.1 × 10 according to Eqs. (2) and (3). Activation energy (E − E or E −13 14 T V C VTD-E1 E − 0.61 ± 0.03 4.0 × 10 2.6 × 10 −16 14 − E ) of the trap can be calculated from the slope of ln(e /T )or T p RTE-H1 E + 0.49 ± 0.03 2.2 × 10 1.2 × 10 −13 15 ln(e /T ) vs. 1/k T plot, and the capture cross-section could be n B RTE-H2 E + 0.74 ± 0.04 7.7 × 10 2.3 × 10 −12 15 extracted from the y-intercept. The trap concentration (N ) can RTE-E1 E − 0.60 ± 0.02 1.6 × 10 1.7 × 10 T C 15, 39 be obtained from the Eq. (4) : 2ΔC max measured the variation of depletion width with the applied bias, N ¼ N ; ð4Þ T A C as shown in Supplementary Fig. 8. The depletion width of VTD- and RTE-fabricated devices decreased from 430 and 324 nm to here N is the net acceptor concentration in Sb Se film, which 240 and 255 nm with the bias pulse changing from −0.5 V to 0.4 A 2 3 can be obtained from C-V profiling (Fig. 4i); ΔC equals to the V, respectively. However, the Cd diffusion depths in VTD- and max difference between C and C (Fig. 5a). RTE-fabricated Sb Se layers are about 200 nm and 100 nm from t0+ 0 2 3 Based on the above fitting results and calculation, the defect SIMS results, respectively. Thus, the Cd ions located outside the parameters of VTD- and RTE-fabricated Sb Se solar cells are detected depletion region and had no effect on DLTS result. The 2 3 summarized in Table 1. The properties of deep defects in Sb Se defects only originate from the intrinsic Sb Se . 2 3 2 3 film are experimentally uncovered. By comparing the activation In our VTD and RTE facilities, Se vapor is always excess for its energies of these trap defects, we find that they are similar in both higher vapor pressure than Sb and Sb Se ,so the Sb Se films are 2 3 2 3 1, 2 devices, indicating the same origins of these defects. To actually slightly Se rich . Our previous ab initio calculation of the investigate whether Cd diffusion affects the DLTS results, we intrinsic defects in Sb Se has demonstrated that the dominant 2 3 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 7 | | | DLTS (ΔC/C ) 2 2 ln(e / T ) or ln(e / T ) n p ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ab VTD-Sb Se RTE-Sb Se 2 3 2 3 –4.04 eV –4.07 eV E1: ΔE = 0.61 eV E1: ΔE = 0.60 eV E E F F –4.69 eV –4.69 eV H2: ΔE = 0.71 eV H2: ΔE = 0.74 eV H1: ΔE = 0.48 eV H1: ΔE = 0.49 eV –5.23 eV –5.26 eV cd E Sb Se Sb Se 2 3 2 3 Fn H2 H2 CdS E1 E1 Fp H1 H1 CdS E E V V Fig. 6 Influence of defect levels on the CdS/Sb Se solar cells. Energy states and defect level of a VTD-fabricated and b RTE-fabricated Sb Se films. Energy 2 3 2 3 band diagrams at CdS/Sb Se interface c in the dark and d under illumination 2 3 acceptor defects are antimony vacancy (V ) and selenium antisite We now discuss influence of defects in working conditions. Sb (Se ) defects under Se-rich condition . Therefore, we tentatively The position of E is always dependent on illumination intensity, Sb Fp attributed H1 and H2 defects to V and Se defects, respectively. and Fig. 6d revealed the situation under AM1.5 irradiation. The Sb Sb As reported by Tumelero et al. , the antisite defects dominated the intersections of E and defect levels can be used as boundaries to distribution of defects in trichalcogenides due to the similar sizes of differentiate whether the defects are charged or not during the the constituent atoms. Our previous simulation also showed that shifting of quasi-Fermi level . Clearly, H1 state is under Fermi Se and antimony antisite (Sb ) are acceptor and donor defects, level and submersed in electrons, so H1 defects always stay inert. Sb Se respectively . Consequently, the E1 defect is most likely associated In contrast, H2 and E1 defect states are mostly above E and they with the formation of Sb antisite defects. Interestingly, E1 and H2 are active in trapping holes and electrons, respectively. These Se always have similar densities with each other in both VTD- and trapped photo-generated carriers would most probably contribute RTE-fabricated samples, which indicate that antisite defect pairs to recombination loss. Furthermore, the energy levels of H2 and formed in Sb Se films, presumably forming [Sb +Se ]complex. E1 are located near to the midgap, which significantly increase the 2 3 Se Sb Please also note that using VTD process reduced the density of recombination possibility of photo-generated carriers . There- these antisite defect pairs by more than an order of magnitude fore, the H2 and E1 are the dominant defects that influence the (Table 1). shift of quasi-Fermi levels and trap-assisted recombination, and The energy levels of defects in the two samples are depicted in then the V and J of the solar cells. Moreover, due to the OC SC Fig. 6a, b, respectively. Herein, the conduction band, valence band higher defect density of H2 and E1 than carrier concentration 13 −3 3 and Fermi levels were obtained from ultraviolet photoelectron (about 10 cm )inSb Se layer , the E would be more likely 2 3 Fp spectroscopy (UPS) and Tauc plots by transmission spectra of to be pinned near E1 and H2 levels. Obviously, both of H2 and E1 Sb Se films (Supplementary Fig. 9). Clearly, the energy levels of have lower defect density in VTD-fabricated sample, which could 2 3 H2 and E1 are above the Fermi level (E ) in both samples, and H1 lead to relatively larger E downshifting and suppressed trap- F Fp is under E . To analyze the influence of defects on the photo- assisted recombination, explaining the better V and J in F OC SC generated carrier recombination, energy band and defect energy VTD-fabricated Sb Se solar cells. 2 3 levels of CdS/Sb Se device in the dark and under illumination are 2 3 depicted in Fig. 6c, d, respectively. Due to thermal equilibrium, Discussion the energy band of Sb Se bended downward around the hetero- 2 3 We have demonstrated that VTD technique can greatly enhance the interface, which led to the same Fermi level in CdS and Sb Se 2 3 performance of Sb Se thin film solar cells by increasing the crys- 2 3 layers. Under illumination, as depicted in Fig. 6d, the photo- tallinity of Sb Se films, reducing the interface and bulk defects in the 2 3 generated electrons were driven into CdS layer, prompting devices, and prolonging the carrier lifetime. Specifically, we believe electron quasi-Fermi level (E ) to move upward in CdS. Fn the advantages of VTD over RTE are based on two features. First, Meanwhile the photo-generated holes resulted in the hole the substrate temperature in VTD process can be regulated inde- quasi-Fermi level (E ) shifting downward in Sb Se layer. The Fp 2 3 pendently by changing the distance between source and substrate, shift of quasi-Fermi levels is positively correlated with the thus permitting higher substrate temperature (VTD 390 °C vs. RTE nonequilibrium carrier concentration. The difference between the 350 °C). The higher substrate temperature resulted in improved two quasi-Fermi levels determines V of the solar cell . OC crystallinity (threefold enhancement in XRD peak intensity) and 8 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 ARTICLE average grain size (VTD 382 nm vs. RTE 297 nm) of Sb Se films. Material characterization. Material and device characterization are similar to 2 3 1, 2, 11 previous report . XRD of Sb Se films was performed using a Philips X’Pert 2 3 Second, the slower deposition of Sb Se films (VTD 2 min vs. RTE 2 3 Pro diffractometer with Cu Kα radiation (λ = 1.54 Å). SEM measurement was 35 s) enables less film imperfection (Se +Se antisite complex and Sb Sb carried out with FEI Nova Nano SEM450. SIMS (IMS-4f, CAMECA instruments) V vacancy), as reflected by the reduced bulk defect density (VTD Sb was used to investigate the element distribution along the depth in the devices. UPS 14 −3 15 −3 10 cm vs. RTE 10 cm ) and interfacial defect density (VTD was used to investigate the Fermi level and valence band of Sb Se films. Experi- 2 3 10 −2 11 −2 ments were performed using a He I (21.21 eV) gas discharge lamp in a Kratos 2.8 × 10 cm vs. RTE 2.1 × 10 cm ), and increased carrier AXIS-ULTRA DLD-600W x-ray photoelectron spectroscopy measurement system lifetime (VTD 1339 ps vs. 1149 ps). Overall, all these advantages and recorded at 0 V samples bias in an ultrahigh vacuum chamber. The surface of facilitated the fabrication of high-quality Sb Se films, leading to 2 3 Sb Se films was etched before UPS measurement. Ultraviolet–visible near-infrared 2 3 −2 increased V , J as well as FF (VTD: 0.42 V, 29.9 mA cm , 60.4% transmission spectra (Perkin Elmer Instrument, Lambda 950 using integrating OC SC −2 sphere) were measured to determine the bandgap of Sb Se films. vs. RTE: 0.39 V, 25.3 mA cm , 56.4%). The champion device 2 3 achieved the efficiency record 7.6%, much higher than SnS, Cu O 47–49 Device characterization. J-V curve and PCE of the champion CdS/Sb Se solar and FeS solar cells which have been studied for many years . 2 3 cell was independently measured by National Institute Metrology (NIM), using the The fabrication of record efficiency Sb Se solar cells employing 2 3 Class AAA Solar Simulator with double-light source (SAN-EI ELECTRIC, XHS- vapor transport deposition, a technique with proven high turn- −2 2350M1, 100 mW cm , AM1.5 illumination) in air ambient at room temperature. around and low cost for commercial CdTe solar cells, further A metal mask was used to define the area of incident light. The area (9.099 mm ) strengthens the great potential of our Sb Se thin film photovoltaics. was also measured by NIM. The light intensity was calibrated by a standard Si- 2 3 reference solar cell. A Keithley 2400 Source Meter was used to acquire J-V data. On the other hand, we find that the deep defect density in the The external quantum efficiency of solar cells was measured using the light source 14 15 −3 best-performing Sb Se solar cells is about 10 to 10 cm , 2 3 generated by a 300 W xenon lamp of Newport (Oriel, 69911) and then split into 11 13 −3 which is much higher than that (10 to 10 cm ) in CdTe solar specific wavelengths by a Newport Oriel Cornerstone 130 1/8 Monochromator cells . These abundant deep defects could pin the quasi-Fermi (Oriel, model 74004). A standard silicon detector (70356_70316NS_455) was used for the calibration. The capacitance-voltage (C-V) profiling and DLCP data were level near H2 and E1 defect states, which provides a plausible measured using Keithley 4200. C-V measurements were performed at room tem- explanation for the low V observed in all Sb Se solar cells OC 2 3 perature in an electromagnetic shielding box at a frequency of 100 kHz and a.c. reported so far, despite various device configuration and film amplitude of 30 mV. The d.c. bias voltage was scanned from −1.0 V to 0.3 V. 1–12, 19, 32, 51 preparation methods have been explored . We suggest DLCP measurements were performed with a.c. amplitude from 0.02 V to 0.14 V and d.c. bias voltage from −0.2 V to 0.2 V. DLTS measurement were performed by that future research on tightly controlling Se and Sb components Semetrol DLTS system (Semetrol, LLC, USA) on the two typical devices. The in the vapor, their strictly stoichiometric condensation into the −1 temperature was scanned between 100 and 380 K, at a heating rate of 2 K min . film and the growth of highly crystalline film should be carried out The reverse bias voltage was −0.5 V. The filling pulse voltage and width were 0.4 V to minimize these deep defects and maximize device performance. and 1 ms, respectively. At every scanning temperature point, the transient capa- citance was measured 20 times for obtaining the average value. Transient In summary, we have obtained the superstrate CdS/Sb Se thin 2 3 absorption spectroscopy was pumped by 500 nm laser pulses (Light Conversion, film solar cell with a certified efficiency of 7.6% (V of 0.42 V, OC Pharos, 350 fs duration pulses, 5 kHz repetition rate) and probed at 940 nm. The −2 J of 29.9 mA cm and FF of 60.4%), which was promoted by SC time delay was adjusted by changing the path length of the probe. the VTD-fabricated Sb Se absorber layer. Compared with the 2 3 1–6, 8–12, 19, 20 previous reports on Sb Se solar cells , VTD process 2 3 Data availability. The data that support the findings of this study are available reduced the density of deep defects and subsequently suppressed from the corresponding author on request. trap-assisted recombination in Sb Se films, resulting in longer 2 3 carrier lifetime and better device performance. These encouraging Received: 22 December 2017 Accepted: 19 April 2018 results highlight the potential of Sb Se solar cells for high- 2 3 efficiency photovoltaic devices. Methods References Sb Se solar cell fabrication. All solar cells were deposited on ITO (In O :Sn) 2 3 2 3 1. Wang, L. et al. Stable 6%-efficient Sb Se solar cells with a ZnO buffer layer. 2 3 transparent conductive glass supplied by Kaivo, with sheet resistance of 6.5 to 6.8 Nat. Energy 2, 17046 (2017). −1 ohm sq , transmittance of 78.8 to 79.6% and ITO thickness of about 200 nm. The 2. Zhou, Y. et al. Thin-film Sb Se photovoltaics with oriented one-dimensional 2 3 ITO substrates were cleaned using detergent, acetone, isopropanol, ethanol and ribbons and benign grain boundaries. Nat. Photon. 9, 409–415 (2015). deionized water in sequence. CdS buffer layer was deposited by chemical bath 3. Chen, C. et al. Characterization of basic physical properties of Sb Se and its 7 −1 2 3 deposition . CdS layers were treated with H O (30 wt%) and 20 mg ml CdCl 2 2 2 relevance for photovoltaics. Front. Optoelectron. 10,18–30 (2017). (Aladdin) absolute methanol solution by spin coating, respectively, baked on the 4. Liu, X. et al. Thermal evaporation and characterization of Sb Se thin film for 2 3 hotplate at 400 °C for 5 min in air ambient and then cooled down naturally. Fol- substrate Sb Se /CdS solar cells. ACS Appl. Mater. Interfaces 6, 10687–10695 2 3 lowing that, VTD process was used to deposit Sb Se films. As shown in Fig. 1a, 2 3 (2014). 0.25 g Sb Se powder (99.999% purity, Jiangxi Ketai) was put into a quartz crucible 2 3 5. Leng, M. et al. Selenization of Sb Se absorber layer: an efficient step to 2 3 and placed in the center of VTD system (a single temperature zone tube furnace, improve device performance of CdS/Sb Se solar cells. Appl. Phys. Lett. 105, 2 3 MTI, Hefei, China). The ITO/CdS substrate was immobilized on a graphite support 083905 (2014). and then placed at the right end of the quartz tube. Substrate temperature was 6. Luo, M. et al. Thermal evaporation and characterization of superstrate CdS/ regulated by changing the distance between substrate and the center of the heater. Sb Se solar cells. Appl. Phys. Lett. 104, 173904 (2014). Vacuum was pumped by a mechanical pump and the stabilized chamber pressure 2 3 7. Wang, L. et al. Ambient CdCl treatment on CdS buffer layer for improved was controlled by varying the ventilation power of the pump. The heating tem- 2 performance of Sb Se thin film photovoltaics. Appl. Phys. Lett. 107, 143902 perature of VTD system was raised to the targeted evaporation temperature with a 2 3 −1 ramp rate of 20 °C min and kept for 2 min to obtain a desired Sb Se film (2015). 2 3 thickness. Then, we turned off the power to stop the deposition and finally took the 8. Liu, X. et al. Improving the performance of Sb Se thin film solar cells over 4% 2 3 sample out when it was naturally cooled down to about 100 °C. After that, gold by controlled addition of oxygen during film deposition. Prog. Photo. Res. back-contact electrodes (0.091 cm area, 100 nm thick) were evaporated by the Appl. 23, 1828–1836 (2015). resistance evaporation thin film system (Beijing Technol Science) under a vacuum 9. Li, Z. et al. Sb Se thin film solar cells in substrate configuration and the back 2 3 −3 pressure of 5 × 10 Pa. To optimize the quality of Sb Se film and the performance contact selenization. Sol. Energy Mater. Sol. Cells 161, 190–196 (2017). 2 3 of device, we systematically investigated the evaporation temperature of VTD 10. Yuan, C., Jin, X., Jiang, G., Liu, W. & Zhu, C. Sb Se solar cells prepared with 2 3 system from 480 to 520 °C, the pressure from 2.8 to 5.3 Pa and the substrate selenized dc-sputtered metallic precursors. J. Mater. Sci. Mater. Electron 27, distance from 19 to 23 cm. When the evaporation temperature was investigated, 8906–8910 (2016). the pressure and the distance were set as 4 Pa and 22 cm, respectively. The opti- 11. Liu, X. et al. Enhanced Sb Se solar cell performance through theory-guided 2 3 mized evaporation temperature and the distance of 22 cm were adopted for pres- defect control. Prog. Photo. Res. Appl. 25, 861–870 (2017). sure investigation, and then the optimized temperature and pressures were used to 12. Chen, C. et al. 6.5% certified Sb Se solar cells using PbS colloidal quantum 2 3 investigate the substrate distance. dot film as hole transporting layer. ACS Energy Lett. 2, 2125–2132 (2017). NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications 9 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04634-6 13. Leijtens, T. et al. Carrier trapping and recombination: the role of defect 43. Dharmarasu, N. et al. Majority- and minority-carrier deep level traps in physics in enhancing the open circuit voltage of metal halide perovskite solar proton-irradiated n+/p-InGaP space solar cells. Appl. Phys. Lett. 81,64–66 cells. Energy Environ. Sci. 9, 3472–3481 (2016). (2002). 14. Heo, S. et al. Deep level trapped defect analysis in CH NH PbI perovskite 44. Tumelero, M. A., Faccio, R. & Pasa, A. A. Unraveling the native conduction of 3 3 3 solar cells by deep level transient spectroscopy. Energy Environ. Sci. 10, trichalcogenides and its ideal band alignment for new photovoltaic interfaces. 1128–1133 (2017). J. Phys. Chem. C 120, 1390–1399 (2016). 15. Lang, D. V. Deep-level transient spectroscopy: a new method to characterize 45. Boix, P. P. et al. Determination of gap defect states in organic bulk traps in semiconductors. J. Appl. Phys. 45, 3023–3032 (1974). heterojunction solar cells from capacitance measurements. Appl. Phys. Lett. 16. Yang, W. S. et al. Iodide management in formamidinium-lead-halide-based 95, 233302 (2009). perovskite layers for efficient solar cells. Science 356, 1376–1379 (2017). 46. Sherkar, T. S. et al. Recombination in perovskite solar cells: significance of 17. Kestner, J. M. et al. An experimental and modeling analysis of vapor transport grain boundaries, interface traps, and defect ions. ACS Energy Lett. 2, deposition of cadmium telluride. Sol. Energy Mater. Sol. Cells 83,55–65 1214–1222 (2017). (2004). 47. Sinsermsuksakul, P. et al. Overcoming efficiency limitations of SnS-Based 18. Wangperawong, A. et al. Bifacial solar cell with SnS absorber by vapor solar cells. Adv. Energy Mater. 4, 1400496 (2014). transport deposition. Appl. Phys. Lett. 105, 173904 (2014). 48. Minami, T., Nishi, Y. & Miyata, T. Heterojunction solar cell with 6% efficiency 19. Ngo, T. T. et al. Electrodeposition of antimony selenide thin films and based on an n-type aluminum-gallium-oxide thin film and p-type sodium- application in semiconductor sensitized solar cells. ACS Appl. Mater. doped Cu O sheet. Appl. Phys. Express 8, 022301 (2015). Interfaces 6, 2836–2841 (2014). 49. Prabukanthan, P., Thamaraiselvi, S. & Harichandranb, G. Single step 2+ 20. Choi, Y. C. et al. Sb Se -Sensitized inorganic-organic heterojunction solar cells electrochemical deposition of p-type undoped and Co doped FeS thin films 2 3 2 fabricated using a single-source precursor. Angew. Chem. Int. Ed. 53, and performance in heterojunction solid solar cells. J. Electrochem. Soc. 164, 1329–1333 (2014). D581–D589 (2017). 21. Mattox, D. M. Particle bombardment effects on thin-film deposition: A 50. Versluys, J., Clauws, P., Nollet, P., Degrave, S. & Burgelman, M. review. J. Vac. Sci. Technol. A 7, 1105–1114 (1989). Characterization of deep defects in CdS/CdTe thin film solar cells using deep 22. Rossnagel, S. M. Thin film deposition with physical vapor deposition and level transient spectroscopy. Thin Solid Films 451-452, 434–438 (2004). related technologies. J. Vac. Sci. Technol. A 21, S74–S87 (2003). 51. Liang, G. X. et al. Thermally induced structural evolution and performance of 23. Lin, X. Z. et al. 11.3% efficiency Cu(In,Ga)(S,Se) thin film solar cells via drop- Sb Se films and nanorods prepared by an easy sputtering method. Sol. Energy 2 2 3 on-demand inkjet printing. Energy Environ. Sci. 9, 2037–2043 (2016). Mater. Sol. Cells 174, 263–270 (2018). 24. Quilettes, D. W. et al. Impact of microstructure on local carrier lifetime in perovskite solar cells. Science 348, 683–686 (2015). Acknowledgements 25. Sun, Z. H., Sitbon, G., Pons, T., Bakulin, A. A. & Chen, Z. Y. Reduced carrier This work was financially supported by the Major State Basic Research Development recombination in PbS-CuInS quantum dot solar cells. Sci. Rep. 5, 10626 Program of China (2016YFA0204000), the National Natural Science Foundation of (2015). China (61725401, 91433105 and 51602114), the Special Fund for Strategic New Devel- 26. Zhao, B. D. et al. High open-circuit voltages in Tin-rich low-bandgap opment of Shenzhen, China (JCYJ20160414102210144), and the Project funded by China perovskite-based planar heterojunction photovoltaics. Adv. Mater. 29, Postdoctoral Science Foundation (2017M622445). The authors thank the Analytical and 1604744 (2017). Testing Center of HUST and the facility support of the Center for Nanoscale Char- 27. Gao, F. et al. Trap-induced losses in hybrid photovoltaics. ACS Nano. 8, acterization and Devices, WNLO. We also thank J. Luo and J. Li for X-ray diffraction 3213–3221 (2014). measurements, Q. Hu for his help in SEM measurements, and Y. Zhong and S. Sun from 28. Abbaszadeh, D. et al. Elimination of charge carrier trapping in diluted Zhengzhou University for their help in DLTS measurements. semiconductors. Nat. Mater. 15, 628–633 (2016). 29. Szendrei, K., Gomulya, W., Yarema, M., Heiss, W. & Loi, M. A. PbS nanocrystal solar cells with high efficiency and fill factor. Appl. Phys. Lett. 97, Author contributions 203501 (2010). X.W. and J.T. conceived the idea, designed the experiments and analyzed the data. X.W., 30. Li, Y. W. et al. High-efficiency robust perovskite solar cells on ultrathin C.C., and S.L. carried out the device optimizations. Y.Z. prepared the gold electrodes. X. flexible substrates. Nat. Commun. 7, 10214 (2016). W. carried out most of material and device characterizations. L.G. carried out the TA 31. Jahandar, M. et al. Highly efficient metal halide substituted CH NH I(PbI ) 3 3 2 1- characterization. C.C., S.L., K.L., W.C., R.K., C.W., J.Z. and G.N. assisted in data analysis. (CuBr ) planar perovskite solar cells. Nano Energy 27, 330–339 (2016). x 2 x X.W. and J.T wrote the paper. All authors commented on the manuscript. 32. Zhou, Y. et al. Buried homojunction in CdS/Sb Se thin film photovoltaics 2 3 generated by interfacial diffusion. Appl. Phys. Lett. 111, 013901 (2017). 33. Kumar, S. G. & Rao, K. S. R. K. Physics and chemistry of CdTe/CdS thin film Additional information Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- heterojunction photovoltaic devices: fundamental and critical aspects. Energy Environ. Sci. 7,45–102 (2014). 018-04634-6. 34. Liu, F. Y. et al. Nanoscale microstructure and chemistry of Cu ZnSnS /CdS 2 4 interface in Kesterite Cu ZnSnS solar cells. Adv. Energy Mater. 6, 1600706 Competing interests: The authors declare no competing interests. 2 4 (2016). 35. Heath, J. T., Cohen, J. D. & Shafarman, W. N. Bulk and metastable defects in Reprints and permission information is available online at http://npg.nature.com/ CuIn Ga Se thin films using drive-level capacitance profiling. J. Appl. Phys. reprintsandpermissions/ 1−x x 2 95, 1000–1010 (2004). 36. Duan, H. S. et al. The role of sulfur in solution-processed Cu ZnSn(S,Se) and Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in 2 4 its effect on defect properties. Adv. Funct. Mater. 23, 1466–1471 (2013). published maps and institutional affiliations. 37. Auret, F. D. & Nel, M. Detection of minority-carrier defects by deep level transient spectroscopy using Schottky barrier diodes. J. Appl. Phys. 61, 2546–2549 (1987). 38. Fourches, N. A quantitative treatment for deep level transient spectroscopy Open Access This article is licensed under a Creative Commons under minority-carrier injection. J. Appl. Phys. 70, 209–214 (1991). Attribution 4.0 International License, which permits use, sharing, 39. Kerr, L. L. et al. Investigation of defect properties in Cu(In,Ga)Se solar cells adaptation, distribution and reproduction in any medium or format, as long as you give by deep-level transient spectroscopy. Solid-State Electron 48, 1579–1586 appropriate credit to the original author(s) and the source, provide a link to the Creative (2004). Commons license, and indicate if changes were made. The images or other third party 40. Fleming, R. M., Seager, C. H., Lang, D. V. & Campbell, J. M. Injection deep material in this article are included in the article’s Creative Commons license, unless level transient spectroscopy: An improved method for measuring capture rates indicated otherwise in a credit line to the material. If material is not included in the of hot carriers in semiconductors. J. Appl. Phys. 118, 015703 (2015). article’s Creative Commons license and your intended use is not permitted by statutory 41. Khan, A. et al. DLTS: A promising technique for the identification of the regulation or exceeds the permitted use, you will need to obtain permission directly from recombination and compensator centers in solar cell materials. 1763–1768 the copyright holder. To view a copy of this license, visit http://creativecommons.org/ (Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic licenses/by/4.0/. Energy Conversion, Waikoloa, 2006). 42. Nguyen, T. P. Defects in organic electronic devices. Phys. Stat. Sol. (a) 205, 162–166 (2008). © The Author(s) 2018 10 NATURE COMMUNICATIONS (2018) 9:2179 DOI: 10.1038/s41467-018-04634-6 www.nature.com/naturecommunications | | |

Journal

Nature CommunicationsSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off