Low-Temperature Processed TiOx Electron Transport Layer for Efficient Planar Perovskite Solar Cells

The most frequently used n-type electron transport layer (ETL) in high-efficiency perovskite solar cells (PSCs) is based on titanium oxide (TiO2) films, involving a high-temperature sintering (>450 °C) process. In this work, a dense, uniform, and pinhole-free compact titanium dioxide (TiOx) film was prepared via a facile chemical bath deposition process at a low temperature (80 °C), and was applied as a high-quality ETL for efficient planar PSCs. We tested and compared as-deposited substrates sintered at low temperatures (< 150 °C) and high temperatures (> 450 °C), as well as their corresponding photovoltaic properties. PSCs with a high-temperature treated TiO2 compact layer (CL) exhibited power conversion efficiencies (PCEs) as high as 15.50%, which was close to those of PSCs with low-temperature treated TiOx (14.51%). This indicates that low-temperature treated TiOx can be a potential ETL candidate for planar PSCs. In summary, this work reports on the fabrication of low-temperature processed PSCs, and can be of interest for the design and fabrication of future low-cost and flexible solar modules.


Introduction
Solid-state organic-inorganic hybrid perovskite solar cells (PSCs) have been one of the most significant discoveries in the field of photovoltaics because of their advantages, including a low-cost device fabrication process, desirable energy harvesting characteristics, light weight, and flexibility [1][2][3][4]. The first report of PSCs with a power conversation efficiency (PCE) of 3.8% was by Miyasaka et al. in 2009 [5]. Intensive research efforts have subsequently been undertaken on PSCs, with a reported record PCE of 25.2% under laboratory conditions [6]. In conventional PSC architecture, a perovskite film is sandwiched in between an electron transport layer (ETL) and a hole transport layer (HTL), and both layers are also sandwiched between a transparent electrode and a metal electrode in order to fabricate complete devices. In terms of improving PSC performance, it is significantly important to produce a uniform, compact, pinhole-free, and full surface coverage titanium oxide (TiO 2 ) compact layer (CL) as the ETL in order to achieve more efficient electron transport, charge extraction, and low interfacial recombination [7]. PSCs can be fabricated entirely using low-cost solution processing [8] (IV) oxysulfate (TiOSO 4 ; purity 99.99%) was purchased from Sigma Aldrich (St. Louis, MO, USA). Hydrogen peroxide (H 2 O 2 ; purity 35%) and N, N-dimethylformamide (DMF; purity 99.5%) were supplied by Kanto Chemical (Tokyo, Japan).

Device Fabrication
FTO-and ITO-patterned glass substrates were sequentially cleaned in a sonication bath with a KOH solution (1.4 g of KOH dissolved in 50 mL of ultrapure water) followed by ultrapure water for 10 min each for two cycles. The cleaned FTO and ITO substrates were then dried with nitrogen flow and pre-treated using oxygen plasma for 20 min prior to use. A complex precursor solution was prepared by adding water as a solvent, TiOSO 4 as a titanium source, and H 2 O 2 as a complexing agent, followed by heating, and was deposited on FTO or ITO substrates, according to the procedure described by Kuwabara et al. [30]. Then, the as-deposited substrates were baked at a low temperature (<150 • C) for 1 h and at a high temperature (> 450 • C) for 30 min. Then, 0.482 g of PbI 2 and 0.168 g of CH 3 NH 3 I were dissolved and mixed in anhydrous DMF (652 µL):DMSO (163 µL) at a ratio of 4:1. The perovskite precursor solution was stirred at 70 • C for 60 min prior to spin coating. A perovskite precursor solution was spin-coated in three steps, as follows: first step, 0 rpm for 10 s; second step, 1000 rpm for 10 s; and third step, 5000 rpm for 30 s. In the third step, 400 µL of chlorobenzene solvent was dripped on the substrate 5 s before the end of spin-coating, and was transferred to a hot plate at 100 • C for 60 min in a glove box under an inert environment and then cooled to room temperature. The Spiro-OMeTAD solution, used as the hole transport layer, was prepared according to the report by Wakamiya et al. [35]. Finally, a gold (Au) electrode with a thickness of 100 nm was deposited at a 1.0 Å/s growth rate on the Spiro-OMeTAD layer so as to complete the device.

Characterization
Scanning electron microscopy (SEMSU1510, Hitachi High-Tech, Tokyo, Japan) together with atomic force microscopy (AFM; SII SPI3800N, Seiko, Japan) were used to analyze the surface morphology. The X-ray diffraction (XRD) patterns of the prepared films were measured using an X-ray diffractometer (SmartLab, Rigaku, Japan) with an X-ray tube (Cu Kα radiation, λ = 1.5406 Å). Surfcorder (ET 200, Tokyo, Japan) was used to measure the thickness of the films. The current density versus voltage (J-V) characteristics of all of the fabricated PSCs were measured at a scan rate of 0.05 V/s in forward scan (FS; from −0.2 to 1.2 V) and reverse scan (RS; from 1.2 to −0.2 V) directions. Each device had a 0.09 cm 2 active area. Measurements were obtained under 100 mW/cm 2 AM 1.5G irradiation from a solar simulator, and were measured using a Keithley 2401 digital source meter. The incident photon-to-conversion efficiency (IPCE) spectrum of each device was measured using a monochromatic xenon arc light system (Bunkoukeiki, SMI-250JA, Tokyo, Japan). Figure 1a shows a schematic of the facile chemical bath deposition process at a low temperature (80 • C) for fabrication of the TiO x films, as well as their corresponding thermal treatments at 150 • C for 1 h and 450 • C for 30 min. Figure 1b,c shows the XPS spectra of the TiO x and TiO 2 films. The Ti2p peak positions were unchanged for both the low-temperature processed TiO x and high-temperature treated TiO 2 films, indicating that the TiO x showed the same chemical bonding state as TiO 2 (Figure 1c). In addition, from the XPS spectrum, the peak intensity of TiO 2 O1s O-Ti and Ti2p3/2 increased compared with TiO x , as shown in Figure 1b,c. This peak can be attributed to the high crystallinity of the TiO 2 film; the high-temperature treated TiO 2 retained its crystallinity, while the low-temperature TiO x had an amorphous structure ( Figure S1). The obtained Raman spectra further confirmed these results. A peak for the anatase crystal structure of the TiO 2 film was confirmed at~150 cm −1 , while no peak (black line) was found in the TiO x film (Figure 1d). This observation is similar to those previously reported by Sangaletti et al. [36]. Therefore, it can be inferred that the bond strength of TiO 2 changed because of the crystallinity. roughness was nearly equivalent in both cases. It can be concluded for both samples that the surface morphology and roughness showed similar trends, regardless of the crystallinity. Cross-sectional SEM images of the TiOx and TiO2 films are shown in Figure 2e,f. Both samples clearly show smooth films deposited (60 nm thickness) on the FTO substrates, which efficiently blocked direct contact between the FTO and perovskites. This implies that low-temperature treated TiOx solely serves as a potential ETL candidate for planar PSCs.  Water droplets on the TiO x and TiO 2 films are shown in Figure 1e,f. The TiO x film shows a higher water contact angle of 36.7 • , while the TiO 2 film exhibits a lower contact angle of 6.6 • , indicating that the TiO x film had a lower wettability compared with the TiO 2 film. The superior wettability of the TiO 2 film may be another avenue to improve the PCE of the resulting PSCs (as discussed later). In addition, we performed contact angle measurements to evaluate the surface-free energy of the TiO x and TiO 2 films. The contact angles at three different points on each film surface with water, formamide, and ethylene glycol were obtained and the average values were used to calculate the surface-free energy of each film [37]. The surface-free energy of the TiO x and TiO 2 films was 56.9 mJm −2 and 51.8 mJm −2 , respectively, suggesting that a lower surface-free energy facilitates the formation of a more stable film surface. Therefore, it can be concluded that the surface state of the high-temperature treated TiO 2 film affords a higher stability than the low-temperature treated TiO x film.

Results and Discussion
Top-view SEM images of the TiO x and TiO 2 films are shown in Figure 2a,b. It can be seen that almost similar surface morphologies that are uniform, dense, and pinhole free on an FTO substrate were obtained with both films. These smooth, pinhole-free, and dense scaffolds may offer efficient charge extraction and hole blocking in the resulting PSCs. This was further confirmed via atomic force microscopy (AFM) analysis, as shown in Figure 2c,d. The root mean square (RMS) roughness for the TiO x and TiO 2 films was 6.58 and 6.34 nm, respectively, indicating that the film surface roughness was nearly equivalent in both cases. It can be concluded for both samples that the surface morphology and roughness showed similar trends, regardless of the crystallinity. Cross-sectional SEM images of the TiO x and TiO 2 films are shown in Figure 2e,f. Both samples clearly show smooth films deposited (60 nm thickness) on the FTO substrates, which efficiently blocked direct contact between the FTO and Nanomaterials 2020, 10, 1676 5 of 12 perovskites. This implies that low-temperature treated TiO x solely serves as a potential ETL candidate for planar PSCs.  Figure 3a,b shows the top-view SEM images of the TiOx/MAPbI3 and TiO2/MAPbI3 films. It can be seen that both samples had large crystal grains with a uniform and flat surface morphology. The film with a smooth surface morphology and large perovskite grains has fewer grain boundaries and fewer traps, which aids in reducing the charge carrier losses at the trap states in the grain boundaries [38]. We compared the cross-sectional SEM images of the TiOx-and TiO2-based PSCs, as shown in Figure 3c,d, respectively. Both samples had perovskite films with similar thicknesses of 300 nm.
The XRD patterns of the TiOx/MAPbI3 and TiO2/MAPbI3 films are shown in Figure 4a,b. Diffraction peaks were detected at 2θ angles of 14.1°, 28.5°, and 31.8° in the TiOx/MAPbI3 and TiO2/MAPbI3 perovskite films, and the peaks were assigned to the (110), (220), and (310) crystal planes, respectively. There was no peak from PbI2 at 12.6° for both samples, indicating the complete transformation of PbI2. The full width at half maximum (FWHM) of the TiOx/MAPbI3 and TiO2/MAPbI3 films was 0.246 and 0.247, respectively, suggesting a similar crystallinity for the perovskite films, despite the difference in crystallinity for their respective ETLs [39]. The photoluminescence (PL) spectra of the FTO/MAPbI3, FTO/TiOx/MAPbI3, and FTO/TiO2/MAPbI3 films were measured and the results are shown in Figure 4b. The peak intensity was lower, in addition to an increased PL quenching compared with the perovskite formed on the FTO substrate, which confirms that efficient charge extraction occurred from the perovskite film. When the perovskite films were formed on TiOx and TiO2, a similar PL quenching was obtained for both samples, which implied efficient electron transfer from the perovskite film to the ETLs. This suggests that the TiOx and TiO2 films have similar charge extraction capabilities [40,41].  Figure 3a,b shows the top-view SEM images of the TiO x /MAPbI 3 and TiO 2 /MAPbI 3 films. It can be seen that both samples had large crystal grains with a uniform and flat surface morphology. The film with a smooth surface morphology and large perovskite grains has fewer grain boundaries and fewer traps, which aids in reducing the charge carrier losses at the trap states in the grain boundaries [38]. We compared the cross-sectional SEM images of the TiO x -and TiO 2 -based PSCs, as shown in Figure 3c,d, respectively. Both samples had perovskite films with similar thicknesses of 300 nm.   The XRD patterns of the TiO x /MAPbI 3 and TiO 2 /MAPbI 3 films are shown in Figure 4a,b. Diffraction peaks were detected at 2θ angles of 14.1 • , 28.5 • , and 31.8 • in the TiO x /MAPbI 3 and TiO 2 /MAPbI 3 perovskite films, and the peaks were assigned to the (110), (220), and (310) crystal planes, respectively. There was no peak from PbI 2 at 12.6 • for both samples, indicating the complete transformation of PbI 2 . The full width at half maximum (FWHM) of the TiO x /MAPbI 3 and TiO 2 /MAPbI 3 films was 0.246 and 0.247, respectively, suggesting a similar crystallinity for the perovskite films, despite the difference in crystallinity for their respective ETLs [39]. The photoluminescence (PL) spectra of the FTO/MAPbI 3 , FTO/TiO x /MAPbI 3 , and FTO/TiO 2 /MAPbI 3 films were measured and the results are shown in Figure 4b. The peak intensity was lower, in addition to an increased PL quenching compared with the perovskite formed on the FTO substrate, which confirms that efficient charge extraction occurred from the perovskite film. When the perovskite films were formed on TiO x and TiO 2 , a similar PL quenching was obtained for both samples, which implied efficient electron transfer from the perovskite film to the ETLs. This suggests that the TiO x and TiO 2 films have similar charge extraction capabilities [40,41].   We carried out electrochemical impedance spectroscopy (EIS) to further investigate the electrical properties of each interface, including the charge transfer, carrier recombination, and inner series resistance. Figure 5 shows Nyquist plots of the TiOx-and TiO2-based devices at zero bias in the dark and under AM 1.5G-100 mW/cm 2 simulated sunlight irradiation. According to the equivalent circuit model shown in the Figure 5a inset, detailed Nyquist plots fitting the analysis parameters of the corresponding devices are summarized in Table S1. R1 can be considered a resistance component derived from the ETL, while R2 is termed as a resistance component derived from the interface between the ETL and perovskite. R1 is a resistance component derived from the ETL and is labeled as charge transferred resistance (RCT). The reduction in RCT contributed to the superior charge transfer in the TiO2-based device compared with the TiOx-based device, implying that a difference in the crystallinity of TiO2 may contribute to charge transfer at the interface. In addition, R2 is a resistance component at the interface and is considered to be a charge recombination resistance. The larger value contributed to a lower recombination at the interfaces. Mostly equivalent values were obtained for both devices for R2, which is consistent with the results from the corresponding PL spectra (Figure 4b). We carried out electrochemical impedance spectroscopy (EIS) to further investigate the electrical properties of each interface, including the charge transfer, carrier recombination, and inner series resistance. Figure 5 shows Nyquist plots of the TiO x -and TiO 2 -based devices at zero bias in the dark and under AM 1.5G-100 mW/cm 2 simulated sunlight irradiation. According to the equivalent circuit model shown in the Figure 5a inset, detailed Nyquist plots fitting the analysis parameters of the corresponding devices are summarized in Table S1. R1 can be considered a resistance component derived from the ETL, while R2 is termed as a resistance component derived from the interface between the ETL and perovskite. R1 is a resistance component derived from the ETL and is labeled as charge transferred resistance (RCT). The reduction in RCT contributed to the superior charge transfer in the TiO 2 -based device compared with the TiO x -based device, implying that a difference in the crystallinity of TiO 2 may contribute to charge transfer at the interface. In addition, R2 is a resistance component at the interface and is considered to be a charge recombination resistance. The larger value contributed to a lower recombination at the interfaces. Mostly equivalent values were obtained for both devices for R2, which is consistent with the results from the corresponding PL spectra (Figure 4b). Nanomaterials 2020, 10, x FOR PEER REVIEW 7 of 12  Figure 6a shows the complete device structure. The current density versus voltage (J-V) characteristics under 1 sun AM 1.5G (100 mW/cm 2 ) for the TiOx-and TiO2-based PSCs are shown in Figure 6b. The corresponding device parameters are summarized in Table 1. A comparison of the scan direction for the FS and RS is provided in Table S2. The PSC with the TiOx film exhibited a short circuit current density (Jsc) of 20.64 mAcm −2 , open-circuit voltage (Voc) of 1.12 V, fill factor (FF) of 0.63, and PCE of 14.51% in the RS direction. The PSC with the TiO2 film had a Jsc of 21.06 mAcm −2 , Voc of 1.08, FF of 0.68, and PCE of 15.50% in the RS direction. The enhancement of Jsc and FF was attributed to a lowering of the injection barrier at the interface between the TiO2 CL and perovskite; this is because the high-temperature treated TiO2 CL formed a smooth interface (Figure 1e), which facilitated efficient electron flow. The enhancement of Voc was unclear for the TiOx-based PSCs compared with the TiO2-based PSCs. Additionally, large hysteresis was observed in the J-V curves for both devices ( Figure S2a). Previous reports have demonstrated a similar hysteresis behavior for TiO2 CL-based PSCs because of the rough interface between the TiO2 and perovskite [42]. The PSC with TiO2 CL as the ETL exhibited a PCE as high as 15.50%, which is close to that of the PSC with the TiOx ETL (14.51%). This implies that thermal annealing at a high temperature is not necessary in order to achieve high-performance PSCs. The incident photon-to-conversion efficiency (IPCE) was measured in order to verify the reproducibility of the PSCs based on the TiOx and TiO2 films, as shown in Figure 6c. The PSCs with TiOx and TiO2 films produced integrated photocurrents of 19.1 and 20.02 mAcm −2 , respectively. The IPCE value of the TiO2-based PSC was slightly higher than that of the TiOx-based PSC over the same wavelength range, indicating that the electrons were efficiently collected at the interface between the TiO2 CL and perovskite, along with a reduction in the interfacial energy barrier. A histogram of the device PCEs for both the TiOx-and TiO2-based devices is shown in Figure 6d. The PCE distributions for both devices were nearly similar, implying that a lowtemperature processed TiOx film can be used as an efficient ETL in future flexible solar modules.   Figure 6b. The corresponding device parameters are summarized in Table 1. A comparison of the scan direction for the FS and RS is provided in Table S2. The PSC with the TiO x film exhibited a short circuit current density (J sc ) of 20.64 mAcm −2 , open-circuit voltage (V oc ) of 1.12 V, fill factor (FF) of 0.63, and PCE of 14.51% in the RS direction. The PSC with the TiO 2 film had a J sc of 21.06 mAcm −2 , V oc of 1.08, FF of 0.68, and PCE of 15.50% in the RS direction. The enhancement of J sc and FF was attributed to a lowering of the injection barrier at the interface between the TiO 2 CL and perovskite; this is because the high-temperature treated TiO 2 CL formed a smooth interface (Figure 1e), which facilitated efficient electron flow. The enhancement of V oc was unclear for the TiO x -based PSCs compared with the TiO 2 -based PSCs. Additionally, large hysteresis was observed in the J-V curves for both devices ( Figure S2a). Previous reports have demonstrated a similar hysteresis behavior for TiO 2 CL-based PSCs because of the rough interface between the TiO 2 and perovskite [42]. The PSC with TiO 2 CL as the ETL exhibited a PCE as high as 15.50%, which is close to that of the PSC with the TiO x ETL (14.51%). This implies that thermal annealing at a high temperature is not necessary in order to achieve high-performance PSCs. The incident photon-to-conversion efficiency (IPCE) was measured in order to verify the reproducibility of the PSCs based on the TiO x and TiO 2 films, as shown in Figure 6c. The PSCs with TiO x and TiO 2 films produced integrated photocurrents of 19.1 and 20.02 mAcm −2 , respectively. The IPCE value of the TiO 2 -based PSC was slightly higher than that of the TiO x -based PSC over the same wavelength range, indicating that the electrons were efficiently collected at the interface between the TiO 2 CL and perovskite, along with a reduction in the interfacial energy barrier. A histogram of the device PCEs for both the TiO x -and TiO 2 -based devices is shown in Figure 6d. The PCE distributions for both devices were nearly similar, implying that a low-temperature processed TiO x film can be used as an efficient ETL in future flexible solar modules. Nanomaterials 2020, 10, x FOR PEER REVIEW 8 of 12 To verify the reproducibility of the devices based on the TiOx and TiO2 films, we compared the average Jsc, Voc, FF, and PCE values of 18 and 26 individual devices, respectively, as shown in Figure  7. Reproducibility is one of the most important parameters for modularization. The low-temperature treated TiOx and high-temperature treated TiO2 samples showed similar reproducibility results. This indicated that low-temperature treated TiOx performs as well as the high-temperature treated TiO2 in PSCs. Furthermore, we investigated the effectiveness of the low-temperature (< 150 °C) processed TiOx on an ITO substrate, which exhibited a satisfactory photovoltaic performance compared with the FTO substrate ( Figure S2b). The corresponding PSC parameters are summarized in Table S3. It can be concluded that a similar trend was observed for the low-temperature treated TiOx on an ITO substrate device. Therefore, the results show that the low-temperature treated TiOx film had a beneficial contribution to the PSC devices, which could help in further lowering the cost of the module manufacturing process in the future.  To verify the reproducibility of the devices based on the TiO x and TiO 2 films, we compared the average J sc , V oc , FF, and PCE values of 18 and 26 individual devices, respectively, as shown in Figure 7. Reproducibility is one of the most important parameters for modularization. The low-temperature treated TiO x and high-temperature treated TiO 2 samples showed similar reproducibility results. This indicated that low-temperature treated TiO x performs as well as the high-temperature treated TiO 2 in PSCs. Furthermore, we investigated the effectiveness of the low-temperature (< 150 • C) processed TiO x on an ITO substrate, which exhibited a satisfactory photovoltaic performance compared with the FTO substrate ( Figure S2b). The corresponding PSC parameters are summarized in Table S3. It can be concluded that a similar trend was observed for the low-temperature treated TiO x on an ITO substrate device. Therefore, the results show that the low-temperature treated TiO x film had a beneficial contribution to the PSC devices, which could help in further lowering the cost of the module manufacturing process in the future.

Conclusions
In this work, we fabricated and compared as-deposited substrates with TiOx and TiO2 films sintered at low temperatures (< 150 °C) and high temperatures (> 450 °C), and investigated their corresponding photovoltaic properties. The TiOx-based PSCs exhibited a satisfactory photovoltaic performance compared with the TiO2-based PSCs. The PSC with a TiOx CL showed a PCE of 14.51%, which was very close to that of the TiO2 CL-based PSCs (15.50%). In addition, a similar reproducibility was observed for devices fabricated using the TiOx and TiO2 films. This suggests that TiOx CL serves as a potential ETL in the PSC. There was a difference in crystallinity between the TiOx and TiO2 films, while the chemical bonding states and surface morphology were similar. Furthermore, there was no difference between the perovskite films grown on TiOx and TiO2 films, and an almost similar device performance was achieved. This work can enable the fabrication of entirely low-temperature processed PSCs, and could possibly contribute to the fabrication of flexible solar modules in the future.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1: Figure S1: XRD pattern of low-temperature treated TiOx film on an FTO substrate; Figure S2: Forward scan and reverse scan J-V curves of (a) TiOx and TiO2 films deposited on an FTO substrate; (b) TiOx film grown on an ITO substrate; Table S1: Nyquist plots fitting the analysis parameters; Table S2: Summary of device performance characteristics with TiOx-and TiO2-based PSCs; Table S3: Summary of device performance characteristics with a TiOx film deposited on an ITO substrate-based PSC.

Conclusions
In this work, we fabricated and compared as-deposited substrates with TiO x and TiO 2 films sintered at low temperatures (< 150 • C) and high temperatures (> 450 • C), and investigated their corresponding photovoltaic properties. The TiOx-based PSCs exhibited a satisfactory photovoltaic performance compared with the TiO 2 -based PSCs. The PSC with a TiO x CL showed a PCE of 14.51%, which was very close to that of the TiO 2 CL-based PSCs (15.50%). In addition, a similar reproducibility was observed for devices fabricated using the TiO x and TiO 2 films. This suggests that TiO x CL serves as a potential ETL in the PSC. There was a difference in crystallinity between the TiO x and TiO 2 films, while the chemical bonding states and surface morphology were similar. Furthermore, there was no difference between the perovskite films grown on TiO x and TiO 2 films, and an almost similar device performance was achieved. This work can enable the fabrication of entirely low-temperature processed PSCs, and could possibly contribute to the fabrication of flexible solar modules in the future.
Supplementary Materials: The following are available online at http://www.mdpi.com/2079-4991/10/9/1676/s1: Figure S1: XRD pattern of low-temperature treated TiO x film on an FTO substrate; Figure S2: Forward scan and reverse scan J-V curves of (a) TiO x and TiO 2 films deposited on an FTO substrate; (b) TiO x film grown on an ITO substrate; Table S1: Nyquist plots fitting the analysis parameters; Table S2: Summary of device performance characteristics with TiO x -and TiO 2 -based PSCs; Table S3: Summary of device performance characteristics with a TiO x film deposited on an ITO substrate-based PSC.