Fabrication of UV-Stable Perovskite Solar Cells with Compact Fe2O3 Electron Transport Layer by FeCl3 Solution and Fe3O4 Nanoparticles

Even though Fe2O3 is reported as the electron-transporting layer (ETL) in perovskite solar cells (PSCs), its fabrication and defects limit its performance. Herein, we report a Fe2O3 ETL prepared from FeCl3 solution with a dopant Fe3O4 nanoparticle modification. It is found that the mixed solution can reduce the defects and enhance the performance of Fe2O3 ETL, contributing to improved electron transfer and suppressed charge recombination. Consequently, the best efficiency is improved by more than 118% for the optimized device. The stability efficiency of the Fe2O3-ETL-based device is nearly 200% higher than that of the TiO2-ETL-based device after 7 days measurement under a 300 W Xe lamp. This work provides a facile method to fabricate environmentally friendly, high-quality Fe2O3 ETL for perovskite photovoltaic devices and provides a guide for defect passivation research.


Introduction
Organic-inorganic hybrid lead halide perovskites have attracted extensive attention [1,2]. Since the first reports in 2009, the power conversion efficiency (PCE) of PSCs has been improved to 25.7% within about one decade [3][4][5]. A typical planar PSC is composed of the structure of a cathode layer [6]. The ETL plays a significant role in electron extraction and transport from the perovskite absorber to the FTO [7]. To obtain highly efficient perovskite solar cells, a thin, transparent, and electrically conductive ETL without pinholes is crucial.
Currently, the most commonly used ETL material in PSCs is TiO 2 , owing to its high chemical stability, innate transparency, inexpensiveness, and appropriate conduction band (CB) level aligning with the perovskite layer [8]. However, TiO 2 -based devices are reported to suffer from hysteresis and high charge recombination, which severely restricts the wide use of the TiO 2 ETL and hinders the development of PSCs [9]. Moreover, the photocatalytic properties of TiO 2 could reduce the illumination stability of PSCs, resulting in poor UV light stability in PSCs [10]. Thus, a great deal of effort has been made to alleviate this problem. Meanwhile, many endeavors have been directed at searching for alternative semiconductor materials for ETLs, such as SnO 2 , ZnO, and Nb 2 O 5 [11].
As an n-type semiconductor, iron oxide (Fe 2 O 3 ) has attracted increased attention in photovoltaic applications, due to its high chemical stability, low cost, and suitable energy band position [12]. Considering its ultraviolet stability and visible light absorption, Fe 2 O 3 is one of the most promising candidates for the ETL in PSCs. However, only several studies have been reported on the application of Fe 2 O 3 in PSCs [13][14][15][16]. Wang et al. applied spin-coated Fe 2 O 3 as the ETL in PSCs, attaining a PCE of 10.7%, with stability over 30 days upon exposure to ambient air, indicating high stability but a poor efficiency [14]. Guo et al. reported the application of Ni-doped Fe 2 O 3 ETL, achieving an efficiency of 14.2%. They also reported the application of γ-Fe 2 O 3 ETL fabricated at room temperature. However, it is difficult to fabricate Fe 2 O 3 films with good conductivity and crystallinity [15,16].
Herein, we report an Fe 2 O 3 ETL fabricated with the water-dispersed Fe 3 O 4 nanoparticles and FeCl 3 solution. It is found that the addition of FeCl 3 in Fe 3 O 4 nanoparticles precursor reduces the defects and enhances the passivation ability. As a result, the improved electron transfer and suppressed charge recombination contribute to an improvement in the short circuit current density (J sc ) and open-circuit voltages (V oc ), eventually yielding a champion PCE of 12.61%.

Preparation of Fe 2 O 3 ETLs
The ITO substrates were rinsed by ultrasonic vibration with acetone, ethanol, and deionized water for 30 min, and then treated with UV-ozone irradiation for 15 min.
A total of 600 mg of 2.2 mM FeCl 3 ·6H 2 O (Alfa Aesar, 97%) and 300 mg of 1.5 mM FeCl 2 ·4H 2 O (Alfa Aesar, 99%) was dissolved in 5 mL deionized water. Next, 800 mg of polyglucose sorbitol carboxymethylether was dissolved in 10 mL deionized water. Then, both of the solutions were mixed in a three-neck bottle, and stirred vigorously (300 rpm) with nitrogen gas bubbling. Then, the bottle was immediately transferred to a water bath at 60 • C, and 900 µL of 28% ammonium aqueous solution was added (stirring at 800 rpm). The bottle was transferred to a cryogenic bath (containing cold water, ice water, and ethanol). After cooling to −5 • C (decline rate 0.28 • C min −1 ), Fe 3 O 4 nanoparticles solution was eventually obtained after workup by dialysis and filtration. The

Fabrication of Perovskite Solar Cells
Perovskite solar cells were fabricated by a modified two-step method. Firstly, a PbI 2 solution with 600 mg mL −1 in DMF was dropped on the ETL substrate with 3000 rpm for 30 s. A total of 50 µL of mixed solution (60 mg mL −1 FAI, 6 mg mL −1 , MABr, and 6 mg mL −1 MACl in isopropanol) was then rapidly dripped on the rotating substrate 10 s after the spin procedure started. The as-prepared film was heated at 150 • C for 10 min in air in order to obtain a dense perovskite film. After cooling to room temperature, the HTL solution (spiro-OMeTAD, 25 µL) was deposited by spin-coating at 2000 rpm for 30 s. The HTL solution consisted of 72.3 mg spiro-OMeTAD, 28.8 µL 4-tert-butylpyridine (TBP), and 17.5 µL of 520 mg mL −1 lithium bis(trifluoromethylsulphonyl)imide (LiTFSI) in acetonitrile dissolved in 1 mL of chlorobenzene. Then, devices were oxidized in air for 36 h.

Characterization and Measurement
The surface morphology and cross-section of the samples were observed by a fieldemission scanning electron microscope (FE-SEM, Hitachi, SU8010, Japan). The XRD results were measured with an X-ray diffractometer (XRD, Bruker, D8 Advance, Germany). The samples were also investigated by X-ray photoelectron spectroscopy (Thermo, Escalab 250Xi, USA). The photoluminescence (PL) and time-resolved photoluminescence (TRPL) were detected with a 530 nm laser (Edinburgh Instruments, LP320, UK). The absorption spectra were recorded on a UV-vis spectrophotometer (Shimadzu, UV-2600, Japan). The contact angle measurement was measured by DSA25E (KRÜSS, Germany). The currentvoltage characteristics of the solar cells were tested with a Newport solar simulator and a Keithley 2400 Source Meter under AM 1.5G irradiation (100 mW cm −2 ). The electrochemical impedance spectroscopy (EIS) was measured with an electrochemical workstation (Autolab, PGSTAT 302 N, Switzerland) under AM 1.5G light condition with an alternative signal amplitude of 10 mV and in the frequency range of 0.1 Hz-40 kHz in glove box.  In order to further improve the planarity and compactness of Fe 2 O 3 films, FeCl 3 solution was incorporated into the Fe 3 O 4 nanoparticle precursor solution, which could simultaneously retain the advantages of the two methods and reduce the defects, thereby facilitating an efficient ETL. Figure S1 shows the top-view scanning electron microscopy (SEM) image of blank and clean ITO substrate, as previously reported. As shown in Figure 1b, the Fe 2 O 3 film prepared by 0.075 M FeCl 3 solution shows a morphology with cracks and pin-holes, which could lead to direct contact between the perovskite absorber and ITO, resulting in aggravated charge recombination. Figure 1c shows the morphology of the Fe 2 O 3 film fabricated by spin-coating water-dispersed ten-nm-sized Fe 3 O 4 nanoparticles with a concentration of 6 mg mL −1 (measured by Fe), which demonstrates a flat and compact surface except for a few gathered spots. Figure 1d depicts the morphology of the Fe 2 O 3 film prepared by FeCl 3 /Fe 3 O 4 mixed solution. It can be observed that the as-prepared Fe 2 O 3 film exhibits a pin-hole-free coverage, as a result of the cooperation between the nanoparticles and FeCl 3 solution in the annealing process. Figure 2a shows the X-ray diffraction (XRD) pattern of the Fe 2 O 3 films prepared by different methods. XRD analysis confirms that both the samples prepared by Fe 3 O 4 nanoparticles and FeCl 3 /Fe 3 O 4 mixed solution display the same diffraction peaks, which match the standard α-Fe 2 O 3 perfectly (JCPDS, No. 80-2377) [17]. XRD peaks at 22.5 and 24 degree may be the peaks of iron chlorate formed by the incompletely volatilized Cl in the crystallization process and the reduced iron. While the sample prepared by FeCl 3 solution displays an extra peak at low angle. The XRD results indicate that Fe 3 O 4 is converted into Fe 2 O 3 and that the FeCl 3 /Fe 3 O 4 mixed sample has better purity. X-ray photoelectron spectroscopy (XPS) measurements were carried out to elucidate the chemical composition of Fe 2 O 3 films prepared by different methods.  The reduced defects and passivated surface of the Fe 2 O 3 films make a great contribution to a smaller contact angle, which is conducive to the diffusion of perovskite precursor solution on the surface, thus, accelerating the nucleation process of perovskite films [19]. Figure S3 illustrates the UV-vis absorption spectra of Fe 2 O 3 films prepared by different methods. The Fe 2 O 3 film prepared by FeCl 3 /Fe 3 O 4 mixed solution shows a slightly higher absorption in almost the whole wavelength region, which could prevent the perovskite from degrading under UV irradiation and enhance the UV-stable ability. Figure 3a shows the top-view SEM image of the perovskite layer deposited on the FeCl 3 /Fe 3 O 4 mixed sample, which exhibits compact surface and large grain size. Figure 3b shows the cross-sectional SEM image of the entire structure, from which we can see the perovskite layer is also compact and the thickness is about 500 nm. Figure S4 shows the XRD patterns of perovskite coated on as-prepared substrates, and all the peaks of the perovskite are presented with an asterisk. All of them display the same characteristic peaks of perovskite materials, which indicates excellent perovskite crystallinity [20]. Figure 3c presents the best current density-voltage (J-V) curves of the devices based on Fe 2 O 3 films prepared by different methods. All samples were measured under AM 1.5G (from 1.2 V to 0 V, scan step of 0.04 V, and scan rate of 100 mV s −1 ). The devices based on Fe 2 O 3 films are also compared with the TiO 2 -based device, as shown in Figure S5. The detailed photovoltaic parameters of the PSCs with the best PCE values including open-circuit voltage (Voc), short-circuit current density (J SC ), filling factor (FF), and PCE are summarized in Table S1. The device prepared with FeCl 3 displays the lowest PCE of 7.72% and the device based on Fe 3 O 4 nanoparticles provides a PCE of 10.64%. Expectedly, the optimal device prepared by mixed solution exhibits overall superior performance, including a Voc of 0.98 V, Jsc of 23.45 mA cm −2 , and FF of 54.74%, resulting in a PCE of 12.61%. Compared with the device based on single Fe 3 O 4 nanoparticles, Voc and Jsc are improved, which may be due to the reduced defects and passivated recombination with the addition of FeCl 3 . The forward and reverse scanning tests were also carried out to investigate the hysteresis effect by (PCE reverse − PCE forward )/PCE reverse . As shown in Figure S6, the mixed sample shows a minimum hysteresis of 0.09. As a contrast, the FeCl 3 prepared sample shows a hysteresis index of 0.15, and that of the Fe 3 O 4 prepared sample is 0.10. It is indicated that PSC based on the mixed sample shows a better charge-transfer ability. Further characterizations were performed to evaluate the trap state density of the devices. We prepared electron-only devices with structures of ITO/ETL/perovskite/PCBM/Ag to quantitatively assess the trap state density in ETL, as shown in Figure S7. Compared with the Fe 3 O 4 -based device, the V TFL of the mixed sample is reduced to 0.12 V. It is indicated that that addition of FeCl 3 can obtain high-quality Fe 2 O 3 film with compact and flat coverage, contributing to passivating the surface defect and effectively filling the electron trap density, which can greatly improve the electrical properties and accelerate electron extraction and injection at the ETL/perovskite interface.

Results and Discussion
To investigate charge transport and recombination in perovskite solar cells, electrochemical impedance spectroscopy (EIS) was conducted. Figure 3d shows the Nyquist plots of the devices based on Fe 3 O 3 ETLs prepared by different methods under AM 1.5 G illumination, and the fitted parameters are summarized in Table S2. The semicircle at high frequency is related to the transfer resistance (R ct ) at the interface and the semicircle at low frequency corresponds to recombination impedance (R rec ) of the device [21]. The device based on FeCl 3 /Fe 3 O 4 film exhibits a R ct of 178 Ω and R rec of 1023 Ω. The reduced R ct is conducive to the enhanced the carriers transfer at the interface, and the increased R rec is beneficial to the suppressed charge recombination. To further investigate the leakage capacity of Fe 3 O 3 ETLs prepared by different methods, a leakage current test is carried out, as shown in Figure S8. The FeCl 3 /Fe 3 O 4 -based sample shows the lowest leakage value, indicating a better leakage performance. Photoluminescence (PL) was carried out to explore the carrier transport dynamics at the Fe 2 O 3 /perovskite interface, as shown in Figure 3e. All the samples display a typical emission peak at 788 nm, in agreement with the absorbance edge of the perovskite. The FeCl 3 prepared sample presents the lowest PL intensity. This could mainly be correlated to poor coverage of the prepared film, which could cause the direct contact between perovskite and ITO, resulting in an illusion of great electron transfer and extraction. A higher PL intensity is presented in the sample with Fe 3 O 4 -prepared films. It should be ascribed to the imperfect surface and interface. The FeCl 3 /Fe 3 O 4-prepared film demonstrates a PL quenching, indicating that the addition of FeCl 3 can passivate the surface defect and accelerate electron extraction and injection at the ETL/perovskite interface. To further demonstrate the charge transfer and extraction, the time-resolved photoluminescence (TRPL) was performed. Figure 3f shows the TRPL spectra and the fitting curves with a bi-exponential decay function [22]. It is clear that the average recombination lifetime (τ ave ) is prolonged from 7.76 ns to 20.28, and 12.29 ns for samples with FeCl 3 , Fe 3 O 4, and FeCl 3 /Fe 3 O 4-prepared films, respectively. Compared to the Fe 3 O 4-prepared sample, the decreased carrier lifetime of the FeCl 3 /Fe 3 O 4-prepared sample indicates that the addition of FeCl 3 passivates defects of the Fe 3 O 4 prepared films and greatly accelerates the charge separation and transport, leading to suppressed charge recombination.
The transmittance spectra of Fe 2 O 3 films prepared by different methods are shown in Figure S9. The FeCl 3 /Fe 3 O 4 -prepared film shows a high transmission, but it is still slightly lower than that of the TiO 2 film. Figure 4a shows the long-time stability test of controlled TiO 2 and mixed Fe 2 O 3-ETL-based perovskite solar cell, which were tested under a 300 W Xe lamp with the condition of humidity of less than 20% and temperature of 25°C. In order to obtain more accurate stability test results, we used Au as the top electrode instead of the original Ag. The efficiency of the device prepared by controlled TiO 2 ETL decreases more than 70% after 7 days of continuous irradiation. As the most commonly used ETL material in PSCs, TiO 2 is reported as a serious issue that affects the stability of the PSCs. As a product of the TiO 2 photocatalytic effect, UV illumination can excite TiO 2 to generate strong oxidizing holes, which could cause the decomposition of perovskite into CH 3 NH 2 , HI, and PbI 2 , and eventually result in the degradation of the stability [23][24][25]. The device prepared with mixed Fe 2 O 3 ETL still has 70% efficiency, indicating a better stability performance. We speculate that it is due to the UV stability and lesser photocatalytic ability of Fe 2 O 3 , which slows the perovskite from degradation and, thus, enhances the UV-stable ability of PSCs. We also tested the TiO 2 and mixed-Fe 2 O 3 -based devices at the maximum power point (MPP) to investigate the stability under UV illumination (composed of 313 nm, 340 nm, and 351 nm) without encapsulation, as shown in Figure S10. Under the same conditions for 300 min, the mixed-Fe 2 O 3 -based device retains 86% of its initial current density, while the current density of the TiO 2 -based device only retains 52%, indicating no UV reaction of Fe 2 O 3 and perovskite, which makes a great contribution to the UV-stable devices. To further confirm our point of view, the XPS measurements were carried out to elucidate the valance change of Pb in the perovskite of controlled TiO 2 and mixed-Fe 2 O 3-ETL-based perovskite solar cells, which were tested for long-time stability for 7 days. Figure 4b shows the peak of the XPS spectra centered at 141.4 eV, corresponding to Pb 0 4f 5/2 of controlled TiO 2-based sample, which is in good agreement with the literature values of 141.7 eV [26]. The peak is higher than that of the Fe 2 O 3 -based sample, confirming the presence of unsaturated Pb, which results from the degradation of perovskite and could be detrimental to the instability of the sample. For the Fe 2 O 3 -based sample, the peak of Pb 0 4f 5/2 is successfully suppressed, indicating improved stability of the perovskite. We think that the improvement of the stability should be ascribed to no UV reaction of Fe 2 O 3 , which protects perovskite from degradation under continuous irradiation.

Conclusions
In summary, we present a facile modification with FeCl 3 solution to optimize the Fe 2 O 3 ETL prepared by water-dispersed Fe 3 O 4 nanoparticles. The device efficiency is improved by more than 118% for the optimized device. The stability efficiency of the Fe 2 O 3 -ETL-based device is nearly 200% higher than that of the TiO 2 -ETL-based device after 7 days measurement. The improved performance of the as-prepared solar cells is attributed to the reduced defects at the interface, enhanced passivation ability, excellent perovskite crystallization originating from the addition of the FeCl 3, and the UVstable ability of the Fe 2 O 3 -based devices. This work is dedicated to broadening the scope of perovskite photovoltaic devices and provides a way for defect passivation in commercial applications.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nano12244415/s1, Figure S1. Top-view SEM images of pure ITO; Figure S2. Fitted curves of the Fe 2P 3/2 of Fe 2 O 3 films prepared by different methods; Figure S3. UV-vis absorption spectra of Fe 2 O 3 films prepared by different methods; Figure S4. XRD patterns of perovskite coated on different substrates; Figure S5. J-V curves of PSCs based on Fe 2 O 3 and TiO 2 ETLs; Figure S6. Hysteresis measurement of PSCs based on Fe 2 O 3 prepared by different methods; Figure S7. Current−voltage curves of the PSCs with a structure of ITO/HTMs/perovskite/spiro-OMeTAD/Ag; Figure S8. Leakage current measurement of PSCs based on the Fe 2 O 3 ETLs prepared by different methods; Figure S9. Transmittance spectra of TiO 2 and Fe 2 O 3 films prepared by different methods. Figure S10. The continuous illumination stability of the TiO 2 and mixed Fe 2 O 3 based devices under UV illumination without encapsulation; Table S1. Summary of photovoltaic parameters of the PSCs based on the control TiO 2 ETLs and Fe 2 O 3 ETLs prepared by different methods (30 devices tested in the reversed direction); Table S2. Summary of the fitted parameters of solar cells based on the Fe 2 O 3 ETLs prepared by different methods.