Bridging Effects of Sulfur Anions at Titanium Oxide and Perovskite Interfaces on Interfacial Defect Passivation and Performance Enhancement of Perovskite Solar Cells.

Interfacial defects at the electron transport layer (ETL) and perovskite (PVK) interface are critical to the power conversion efficiency (PCE) and stabilities of the perovskite solar cells (PSCs) via significantly affecting the quality of both interface contacts and PVK layers. Here, we demonstrate a simple ionic bond passivation method, employing Na2S solution treatment of the surface of titanium dioxide (TiO2) layers, to effectively passivate the traps at the TiO2/Cs0.05(MA0.15FA0.85)0.95Pb(Br0.15I0.85)3 PVK interface and enhance the performance of PSCs. X-ray photoelectron spectroscopy and other characterizations show that the Na2S treatment introduced S2- ions at the TiO2/PVK interface, where S2- ions effectively bridged the TiO2 ETL and the PVK layer via forming chemical bonds with Ti atoms and with uncoordinated Pb atoms and resulted in the reduced defect density and improved the crystallinity of PVK layers. In addition, the S2- ions can effectively enlarge the grain size of the PVK layers. The average PCE of solar cells is improved from 15.77 to 19.06% via employing the Na2S-treated TiO2 layers. This work demonstrates a simple and facile interface passivation method using ionic bond passivation to afford high-performance PSCs. The bridging effect of S2- ions may inspire the further exploration of the ionic bond passivation and sulfur-based passivation materials.

Introduction

Organic–inorganic hybrid halide perovskite (PVK) materials have attracted enormous interest in solar cell applications, and the power conversion efficiency (PCE) of single-junction perovskite solar cells (PSCs) has improved from 3.8 up to 25.5% in 2021[1,2] due to their extraordinary optoelectronic properties such as the suitable band gap, large absorption coefficient, and high carrier mobility.[3−5] However, further improvements of PSC efficiency and stability are limited by the defects in the PVK layers and at the PVK/electrode interfaces, which could cause serious charge recombination and thus lower the short-circuit current.[6] Compact titanium dioxide (TiO2) is the most widely used electron transport layer (ETL) in PSCs for its suitable energy level alignment, easy fabrication, and good interface quality. However, the severe hysteresis effect and the TiO2/PVK interface of TiO2-based PSCs limit their further applications and tremendous attention has been devoted to the surface passivation in the TiO2/PVK interface.

There are two general strategies to passivate the ETL/PVK interface including the covalent bond passivation and the ionic bond passivation. The most well-studied covalent bond passivation is the Lewis acid–base chemistry method, based on the formation of covalent bonds between Lewis adducts and defects, which leads to the effective passivation of the interface defects and the improvement of the performance and stability of PSCs. Peng et al. introduced a double-side polymer poly(methyl methacrylate) (PMMA) at TiO2/PVK and PVK/spiro-OMeTAD interfaces and finally improved the PCE of PSCs to 20.8% with an impressive open-circuit voltage (Voc) of 1.22 V.[7] Characterizations of the PVK–PMMA molecular interface show that carbonyl (C=O) groups in PMMA can aid in the reduction of Pb2+ defect density at the ETL/PVK and PVK/HTL interfaces. Hou et al. effectively passivated ETL/PVK interface traps via hydrogen-bonding interactions (N–H/I), where the dopamine self-assembled monolayer on the SnO2 ETLs improved the carrier transport and significantly improved the PCEs of the corresponding PSCs from 14.05 to 16.65%.[8] Wang et al. demonstrated PSCs with stabilized PCEs of 22.6% via introducing a dopamine self-assembled monolayer on the top of the SnO2 ETL, which effectively passivate the surface antisite Pb (lead) defect by means of the assisted primary C=O binding.[9] Most recently, Zhou demonstrated a decreased open-circuit voltage deficit from 0.47 to 0.39 V with a stabilized efficiency of 22.99% by permeating a fluorinated perylene-tetracarboxylic diimide derivative on the PVK layer to passivate under-coordinated Pb2+ defects.[10] Chen utilized poly(propylene glycol)bis(2-aminopropylether) (PEA) additive in PVK layers, which interacted with the lead ions and considerably passivated surface and bulk defects.[11] Cai used carbonyl groups in 2,2-difluoropropanediamide to form chemical bonds with Pb2+ and passivate under-coordinated Pb2+ defects in PVK layers.[12] Besides, reduced graphene oxide-based materials with oxygen-based hydroxyl (−OH) groups and other Lewis bases containing amines (such as −NH2) or nitrogen functionalities have also been proved of effective passivation.[13−16] Meanwhile, chemical bonds reduced the crystallization rate, producing high-quality PVK films with fewer defects.[12]

Ionic bond passivation includes cation passivation and anion passivation. Cationic substances including metal ions and organic molecules form ionic bonds and other electrostatic interactions with negatively charged traps (such as uncoordinated I–, PbI3–, and MA+ vacancies) in PVK materials, proving to be effective.[17] Metal ions were reported by Bi et al. for the first time that can effectively passivate defects at grain boundaries of PVK layers.[18] Na+ ions on the PTAA [poly(bis(4-phenyl)(2,4,6-trimethylphenyl)amine)] HTL were found to diffuse into the grain boundaries of PVK layers, resulting in a lower defect density and a longer photoluminescence (PL) lifetime. Since the size and equivalent charge of Na+ are similar to those of MA+, the Na+ ions fill MA vacancies at grain boundaries and thus reduce negatively charged trap densities. Abdi-Jalebi et al. reported an impressive high internal PL quantum yield (larger than 95%) by adding potassium iodide (KI) to the Cs0.06FA0.79MA0.15Pb(I0.85Br0.15)3 PVK precursor, resulting in the accumulation of the K+ ions at the surface and grain boundaries to passivate those extended defects.[19] Beyond alkali metal ions including Cs+ and Rb+, Cu2+ and Ag+ are also reported to be capable of effective passivation.[20−23] Small organic molecules owning ammonium functionalities such as butylammonium (BA+), octylammonium (OA+), phenylethylammonium (PEA+), and diammonium derivatives have also been explored for their ability to enhance device performance and suppress trap-induced recombination.[24−27] However, there are only a few works on the anion passivation of insufficiently coordinated Pb2+ ions and halide vacancies. Reported anions for the trap passivation of PSCs include chloride ions (Cl–) and sulfide ions (S2–). Pool et al. reported that the Cl– ions show effective passivation on the traps at interfaces and/or grain boundaries and also enlarge the grain size of PVK layers.[28] Wang et al. reported a solution-processing method to form strong Pb–Cl and Pb–O bonds between PVK films and chlorinated graphene oxide layers to stabilize the heterostructure.[29] Chen and Cao et al. also discussed the bonding between S atoms with Pb atoms and demonstrated that the bonding is significantly effective for high-performance PSCs.[30,31] Sun achieved a PCE as high as 21.25% with an impressive high Voc of 1.22 V by doping Na2S into TiO2 layers, demonstrating the significant effect of S2– and Na+ ions.[32] However, the doping method is limited by the solubility of Na2S in the TiO2 precursors and the high-temperature calcination process of the TiO2 layer is not compatible with the flexible PSC fabrication process. In addition, we were curious whether the Na+ and S2– ions on the surface of TiO2 layers have better effects than those in the TiO2 layers. In addition, it is still unclear whether the crystal grain enlargement of PVK grains is caused by the surface roughness or the S2– ions on the surface of TiO2 layers. Therefore, it is necessary to come up with other methods to effectively introduce Na+ and S2– ions on TiO2 surfaces and is also very important to understand the interface passivation mechanism and explore the new interface passivation material.

Herein, we demonstrate that a simple Na2S treatment of the TiO2 surfaces can effectively passivate the defects at the TiO2/PVK interface and improve the quality of PVK layers, leading to an enhancement of the solar cell performance. The TiO2 layers were spin-coated with Na2S solution before the PVK layer deposition, and PSCs with a typical structure of fluorine-doped tin oxide (FTO)/TiO2/Cs0.05(MA0.15FA0.85)0.95Pb(Br0.15I0.85)3/Spiro-OMeTAD/Au were fabricated to investigate the effect on the surface passivation and device performance. Experimental results show that this simple Na2S treatment can effectively introduce S2– ions on the surface of TiO2 layers and improve the average efficiency from 15.77 to 19.06%. S2– ions can improve the crystallinity of PVK layers grown on the Na2S-treated TiO2 substrate, evidenced by a higher PL intensity, a longer photogenerated carrier lifetime, and a larger grain size. Space-charge-limited current (SCLC) and J–V curves indicate that devices with the Na2S treatment show the reduced carrier recombination loss at the TiO2/PVK interface and the suppression of interface recombination. This simple method will inspire the exploration of sulfide-based interface passivation in the research community of PSCs.

Results and Discussion

PSCs with the typical structure of FTO/TiO2/(Na2S)/PVK/Spiro-OMeTAD/Au shown in Figure a were prepared to investigate the effects of Na2S treatment on device performances. The Na2S treatment includes the conventional spin-coating process and the postannealing process right after the annealing of TiO2 layers, as shown in Figure b. Ethanol solutions of sodium sulfide with concentrations from 0.2 to 0.8 mg/mL were prepared as the precursor solution. Ten solar cells of each condition were tested under the AM 1.5 solar spectrum, and the device performances are shown in Figure c–f and details are listed in Table S1. It can be seen that the PCE first increases and then decreases with the increase of the Na2S concentration. The similar trend is also found in the parameters of Jsc, Voc, and fill factor (FF). The devices prepared with a Na2S solution of 0.4 mg/mL have the best performance, with an average PCE improvement from 15.77 to 19.06%, the average short-circuit current improvement from 21.39 to 22.39 mA/cm2, an average opening voltage increase from 1.149 to 1.191 V, and the average FF increase from 62.80 to 71.14%. The device performance degradation at a higher Na2S concentration is probably caused by the rough surface and the Na2S residue on the surface of TiO2 layers, as shown in Figure S1, which severely deteriorate the quality of the following PVK layer. Therefore, the surface treatment of TiO2 layers with a proper Na2S concentration can effectively improve the PSC performance, and the optimal concentration is 0.4 mg/mL in this work.

Figure 1

PSCs with TiO2 layers treated with different Na2S solutions. (a) Diagram of the device structure and photograph of PSCs; (b) schematic diagram of the fabrication process of Na2S surface treatment of the compact TiO2 layer; (c) short-circuit current density Jsc, (d) open-circuit voltage Voc, (e) FF, and (f) PCE of PSCs with TiO2 layers treated with Na2S solutions of different concentrations. (g) J–V curves of the devices with and without Na2S treatment under both reverse-scan and forward-scan directions and (h) EQE spectra of PSCs with and without the Na2S treatment of 0.4 mg/mL.

Figure g shows the J–V curves of PSCs measured under both reverse-scan and forward-scan directions with or without Na2S treatment. Obviously, devices with the Na2S treatment have smaller hysteresis. According to Shi,[6] the hysteric index (HI) factor of PSCs with the Na2S-treated TiO2 layers dropped from 0.156 to 0.061. The external quantum efficiency (EQE) of PSCs with the Na2S-treated TiO2 layers was higher than that of PSCs without treatment in the wavelength range of 300–900 nm, as shown in Figure h, indicating that the device with Na2S treatment has an effective charge collection. The integrated current density from EQE is smaller than that from the J–V measurements, probably due to the differences in the light sources used.[33] The light storage stability of PSCs with and without the Na2S treatment in the ambient under continuous sunlight (1 sun) is shown in Figure S2. The ambient humidity during the test is 86%, and the temperature of the device surface irradiated by sunlight is 48 °C. After a continuous radiation of AM 1.5G for 5 h, the unpackaged PSCs with the Na2S-treated TiO2 layers demonstrate a higher light storage stability in the high humidity ambient, with a 24% drop of the initial efficiency, compared with the 36% efficiency drop of the untreated devices. In conclusion, the Na2S treatment of the TiO2 surface can effectively reduce the hysteresis behavior and improve the charge collection efficiency and the stability of PSCs.

X-ray photoelectron spectroscopy (XPS) characterizations were conducted to verify the existence and the chemical bond states of Ti, O, Na, and S ions on TiO2 layers with and without Na2S treatment. The peaks of C 1s at 284.6 eV were used to calibrate the spectra. The characteristic peaks of Na+ 1s at 1071.4 eV and S2– 2p at 162.8 eV can be seen in the XPS spectra of TiO2 layers with Na2S treatment in Figure a,b, respectively. The corresponding peaks cannot be seen in the XPS spectra of untreated TiO2 layers shown in Figure S3. Meanwhile, the characteristic peaks of S6+ 2p at 168 eV are also shown in the XPS spectra, which were probably formed by the oxidation of S2– during the postannealing process.[34] XPS characterizations were also performed on the Na2S-treated TiO2 layers after a 5 mins’ sputtering of Ar+ ions to further investigate the elemental distribution of the Na+ and S2– ions. The existence of Na+ 1s peaks and S2– 2p peaks in the sputtered samples shown in Figure a,b implies that both Na+ and S2– ions diffuse into the TiO2 films. Figure c,d shows the XPS spectra of Ti 2p and O 1s peaks of TiO2 films with and without Na2S treatment. It can be seen very clearly that the Ti 2p peaks at 458.2 and 464.0 eV of the typical Ti–O bonds in untreated TiO2 layers[35] shift slightly to the direction of low binding energy in the spectra of the TiO2 layers with Na2S treatment. This result indicates that S2– ions may form chemical bonds with Ti4+ via substituting oxygen atoms or filling the oxygen vacancies,[36] resulting from the lower electronegativity of S2– ions compared with that of O2– ions. From Figure d, it can be observed that the peak height of adsorbed oxygen at 531.8 eV is higher and that of lattice oxygen peaks at 530.2 eV is lower in the TiO2 layer with Na2S treatment, which implies the existence of S2– ions on the Na2S-treated TiO2 layers. All these results prove that Na+ and S2– ions are successfully introduced into the compact TiO2 layers, and the S2– ions probably form the chemical bonds with Ti4+ ions.

Figure 2

XPS spectra (a) of Na 1s peaks and (b) S 2p peaks of the Na2S-treated TiO2 layer and the same sample sputtered with argon ions for 5 min and XPS spectra of (c) Ti 2p peaks and (d) O 1s peaks of the compact TiO2 layer with and without Na2S treatment.

The elemental scanning in the linear mode through the cross-section of the FTO/TiO2/PVK structure under high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) was carried out to further investigate the spatial distribution of Na and S ions in the TiO2 layers and TiO2/PVK interfaces. As shown in Figure a,b, the region of TiO2 layers can be roughly determined by the signal of Ti elements. Due to the small atomic numbers of Na elements and S elements, obvious background noise may appear in the energy-dispersive X-ray spectroscopy (EDS) scanning. It is interesting to note that the EDS intensity of Na atoms is lower and the distribution profile is more uniform compared with that of the S atoms. The elemental intensity (counts of the fingerprint EDS signal) profile indicates that elements S and Na are present in both TiO2 layers and PVK layers. We speculate that S ions at the TiO2/PVK interface may diffuse into the PVK layers and form bonds with Pb, while Na could diffuse into both PVK and TiO2 layers. Correlating the XPS results in Figure c, it can be concluded that part of S2– ions also forms bonding with Ti4+, and the bilateral bonds of S ions enhance the chemical binding at the TiO2/PVK interface. The appearance of the Na signal in the PVK layer indicates that alkali-metal elements can diffuse into the PVK layer and passivate the grain boundaries.[18]

Figure 3

(a) HAADF-STEM image of the cross-section of PVK films deposited on theTiO2 layer with Na2S treatment and (b) EDS mapping in a linear scan mode through the cross-section of FTO/TiO2/PVK stack.

The electron-only devices with the structure of FTO/TiO2/(Na2S)/Ag shown as the inset of Figures a and S4 were fabricated to investigate the effect of Na2S treatment on the electrical properties of TiO2 layers. The J–V characteristic curves of both structures measured under the dark are shown in Figure a, and the slopes stand for the conductivity of the samples. The conductivity of TiO2 layers was calculated from the linear fitting of the curves in Figure a, and the box diagrams are shown in Figure b. The conductivity of TiO2 layers with Na2S treatment is higher than that of the untreated samples, indicating that the improved conductivity of TiO2 layers resulted from the introduction of Na+ ions. The fact that the conductivity improvement of TiO2 layers is not as significant as that in the previous work is probably caused by the smaller Na atoms introduced by this spin-coating method.[32] This result is consistent with the results in Figure and the previous reports that alkali-metal ions can improve the TiO2 film conductivity.[18,32]Figure c,d shows the atomic force microscopy (AFM) images of the surface of TiO2 layers with and without Na2S treatment, where the surfaces of both samples are very smooth. The TiO2 layer with Na2S treatment has an average roughness (Ra) of 6.72 nm, which is almost the same as that of the TiO2 layer, indicating almost no effect of Na2S treatment on the morphology of the TiO2 layer.

Figure 4

Characterization of TiO2 layers with and without Na2S treatment. (a) J–V characteristic curves; (b) box diagrams of conductivity measurements of two samples; and AFM images [three-dimensional (3D)] of TiO2 layers without (c) and with (d) Na2S treatment.

To fully understand the effects of Na2S treatment, systemic characterizations of the TiO2/PVK interfaces and the following PVK layers were conducted. From the SEM images shown in Figure a,b, we can see that the crystal grain size of the PVK layers on the TiO2 substrates with Na2S treatment is larger than that on the untreated TiO2 substrates. The larger grain size leads to fewer grain boundaries which are beneficial to carrier transport.[37,38] Meanwhile, the contact angle of water on the TiO2 substrates with Na2S treatment is similar to that of the untreated TiO2 substrates, indicating that the Na2S treatment will not affect the hydrophobicity of the TiO2 substrates. It can be concluded that the larger grain sizes of the PVK layer probably are not caused by the hydrophobicity of the substrate in this work[39] but the bonding between S atoms and Pb atoms.[32]

Figure 5

Characterizations of PVK layers and TiO2/PVK interfaces with and without Na2S treatment. Scanning electron microscopy (SEM) images of PVK layers on TiO2 substrates (a) without and (b) with Na2S treatment, with the surface contact angles of water as the insets; (c) X-ray diffraction (XRD) diagrams, (d) UV–visible absorption spectra, (e) PL spectra, and (f) time-resolved PL (TRPL) spectra of the PVK layers on TiO2 substrates with/without Na2S treatment; (g) J–V curves of devices on TiO2 substrates with/without Na2S treatment under the dark; (h) J–V curves of electron-only devices on TiO2 substrates with/without Na2S treatment; and (i) Nyquist plot of electrical impedance spectroscopy (EIS) of devices on TiO2 substrates with and without Na2S treatment. Inset: the device structure in the test.

XRD characterization of PVK layers on different TiO2 substrates was conducted to illustrate the influence of Na2S treatment on the crystallinity of the PVK layers. It can be seen from the XRD diagrams in Figure c that the positions of all the diffraction peaks in both samples are the same. Main diffraction peaks at 14.1, 31.8, and 40.6° can be identified as those of the (100) (210) (300) crystal plane of the cubic PVK, respectively. There are no new characteristic peaks or shift of the peak position, indicating that the Na2S treatment does not change the crystal structure or the phase of the PVK layer. The peak intensity of the (100) crystal plane is enhanced in the samples treated with the Na2S solution of 0.4 mg/mL, indicating that the crystal alignment along the (100) crystal plane is improved, and the Na2S treatment is beneficial to improve the crystallinity of the PVK.

Optical properties of PVK layers grown on both TiO2 substrates were characterized and are shown in Figure d–f. Figure d shows the UV–visible absorption spectra of the PVK films on both TiO2 substrates. There is no obvious difference except the slight shift of the absorption edge in the light absorption spectra of PVK films on both samples. The absorption edges of PVK grown on TiO2- and Na2S-treated TiO2 are 787.7 and 785.5 nm, respectively. The calculated band gaps of PVK grown on TiO2- and Na2S-treated TiO2 are 1.574 and 1.579 eV, respectively, indicating that Na2S treatment has little effect on the energy band structure of the PVK layer. From the PL spectra of both samples shown in Figure e, it can be seen that the PVK film grown on the Na2S=treated TiO2 substrate exhibits a stronger PL emission, indicating a better crystal quality. From the TRPL spectra of both samples shown in Figure f, it can be seen that the PVK film grown on the Na2S-treated TiO2 film exhibits a longer carrier lifetime (656 ns vs 464 ns), which may result from a reduced nonradiative recombination. All these results show that the Na2S treatment improves the crystal quality of the PVK layers on the TiO2 substrates. The dark current–voltage characteristic curves of PVK solar cells on different substrates were measured and are shown in Figure g to investigate the influence of the Na2S treatment on the internal loss of carriers. The dark current can be written as the following equation[40]where J0 is the reverse saturation current density, q is the amount of charge, V is the voltage, n is the ideality factor, k is the Boltzmann constant, and T is the temperature. The reverse saturation current can be obtained from the intersection point by fitting the exponential region of the I–V curves with a tangent. The J0 of the PSC devices is reduced from 1.498 × 10–16 to 6.882 × 10–18 mA/cm2, with the introduction of the Na2S treatment. The decrease of the J0 indicates the reduction of the carrier recombination loss at the TiO2/PVK interface, which is beneficial to the improvement of the open-circuit voltage and the short-circuit current of the device.[40]

SCLC measurements of electron-only devices with a structure of FTO/TiO2/(Na2S)/PVK/BCP/Ag were performed to investigate the trap density at the TiO2/PVK interface and PVK layers. The J–V curves are shown in Figure h, and the trap filling limit voltage (VTFL) was obtained from the intersection point.[41] The fittings of the curves show that the VTFL values of the TiO2 and Na2S-treated TiO2 devices are 1.40 and 1.03 V, respectively. The corresponding trap state densities of the device are 2.15 × 1018 and 1.58 × 1018 cm–3, indicating that the Na2S treatment effectively passivates the traps.

Besides, the EIS measurements were performed under the dark to investigate the effects of Na2S treatment on the charge transfer at the interfaces. The Nyquist plots of PSCs fabricated on TiO2 substrates with and without Na2S treatment are shown in Figure i. The equivalent circuit diagram of the devices used to fit the plot is illustrated in the inset of Figure i, where Rs is the sheet resistance of electrodes and Rct is the charge-transfer resistance of the TiO2/PVK interface. The fitted Rct value of the control device is 2955 Ω and that of the device with a Na2S interfacial layer is 886 Ω. A smaller semicircle suggests a reduced resistance Rct, which means a suppressed recombination at the interface.[42] It can be concluded that the Na2S-treated TiO2 layer has a better interface for charge transfer.

Correlating with the above experimental results and discussion, we can summarize the mechanism of Na2S treatment at the TiO2/PVK interface in Figure . Na2S treatment can successfully introduce Na+ and S2– at the TiO2/PVK interface, where S2–ions form bonds with Ti4+ via substituting the oxygen atoms or filling the oxygen vacancies in the TiO2 layer and anchor-uncoordinated Pb atoms in the PVK layer, leading to the effective ionic bond passivation of interfacial defects, improved crystallinity of PVK layers, and the enhanced PSC performance.[18,20,30,31] Meanwhile, Na+ ions diffuse into the PVK layer and the TiO2 layer, resulting in the defect passivation at the grain boundary of PVKs and the improved conductivity of the TiO2 layer.[18] Compared with the doping method reported in the previous work, the spin-coating method is less effective in introducing Na+ ions into the TiO2 layers, which can improve the conductivity of TiO2 layers and thus enhance the device performance.

Figure 6

Schematic diagram of the mechanism of Na2S at the ETL/PVK interface.

Conclusions

In summary, we demonstrated a simple and effective S2– ionic bond passivation method to reduce defect densities and the corresponding recombination losses at TiO2/PVK interfaces and in Cs0.05(MA0.15FA0.85)0.95Pb(Br0.15I0.85)3 solar cells. The passivation method is to treat the TiO2 surface via spin-coating the Na2S precursor solution, followed by a postannealing process. XPS and other characterizations indicate that S2– ions can be effectively introduced at the TiO2/PVK interface and may function as anchoring sites for Pb2+ in the PVK layer, resulting in reduced defect densities both at the TiO2/PVK interface and in the PVK layer, enlarged PVK grain sizes, and inhibited nonradiative recombination. At the optimal Na2S precursor solution of 0.4 mg/ml, the efficiency of the best device has been improved to 19.8%. Compared with the doping method reported in the previous work,[32] the spin-coating method is less effective in introducing Na+ ions into the TiO2 layers and thus the smaller improvement in device performance. Our work on the bridging effects of S2– ions may inspire the further exploration on the interface passivation of the ETL/PVK interface and performance enhancement of PSCs.

Experimental Section

Solution Preparation

The TiO2 precursors were prepared by dissolving 250 μL of tetrabutyl titanate (Sigma-Aldrich) and 25 μL of hydrochloric acid (Kelong) in 3 mL of absolute ethyl alcohol (Aladdin). The Na2S precursors were prepared by dissolving Na2S (Aladdin) in absolute ethyl alcohol (Aladdin) according to the concentration of Na2S. The Cs0.05(MA0.15FA0.85)0.95Pb(Br0.15I0.85)3 PVK precursor solution consisting of 31.6 mg of cesium iodide (CsI, Alfa Aesar), 19.6 mg of methylammonium bromide (MABr, Lumtec), 67.2 mg of lead bromide (PbBr2, Sigma-Aldrich), 166.6 mg of formamidinium iodide (FAI, luminescence), and 470.2 mg of lead iodide (PbI2, Sigma-Aldrich) was dissolved in a mixture of 600 μL of dimethylformamide (Sigma-Aldrich) and 400 μL of dimethyl sulfoxide (Sigma-Aldrich) and kept stirring overnight at room temperature in the glovebox. The spiro-OMeTAD solution precursors were prepared by mixing 28.8 μL of 4-tert-butylpyridine (Sigma-Aldrich), 17.5 μL of Li-TFSI solution [Li-TFSI (520 mg, Sigma-Aldrich) in acetonitrile (1 mL, Alfa Aesar)], 60 μL of Co (III) TFSI salt solution (FK209 Co (III) TFSI salt (100 mg, luminescence) in acetonitrile (1 mL, Alfa Aesar), and 72.3 mg of 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene(spiro-OMeTAD, luminescence) in chlorobenzene (1 mL, Alfa Aesar) and kept stirring overnight at room temperature in the glovebox. The BCP (0.5 mg, J&K) was dissolved in absolute ethyl alcohol (1 mL, Aladdin).

Device Fabrication

The FTO glass (sheet resistance = 7 Ω sq–1) after washing by detergent, deionized water, acetone, and isopropanol, respectively, in an ultrasonic cleaner for 20 min was dried with a nitrogen gun and treated with UV–ozone for 15 min. The TiO2 compact layers were prepared by spin-coating the TiO2 precursor solution onto the FTO substrates at 5000 rpm for 30 s in air. Then, the samples were preannealed at 120 °C for 20 min on the hot plate and then heated in the muffle oven at 450 °C for 1 h to form the c-TiO2 layers. For the Na2S-coated samples, Na2S precursor solutions were coated onto the compact-TiO2 layers at 5000 rpm for 30 s in air and then annealed at 120 °C for 20 min on the hot plate to form the Na2S interfacial layers.

All samples were treated with UV–ozone for 15 min before the PVK deposition process, and PVK layers were prepared by a one-step process in a glovebox. The FTO/c-TiO2/(Na2S) substrates were spin-coated at 500 rpm for 5s and 5000 rpm for 50 during the spin-coating process of PVK layers, and 50 μL of chlorobenzene (Alfa Aesar) was injected onto the spinning substrate constantly. Then, the samples were annealed at 150 °C for 10 min on a hot plate. The Spiro-OMeTAD precursor was then deposited on the PVK layers via spin-coating at 5000 rpm for 50 s. At last, 80 nm thick Au electrodes were deposited on the top Spiro-OMeTAD layers using vacuum thermal evaporation.

Characterization

The I–V measurements and stability test were performed using a Keithley 2400 digital source meter under simulated sunlight from a Newport 94123A solar simulator matching the AM 1.5G irradiation (100 mW cm–2). The devices were measured from 1.4 to −0.2 V at a scan rate of 10 mV/s. XPS was carried out on PHI 5000 Versa Probe III. EQE was measured by Enlitech QE-R3011. UV-absorption spectra were recorded using a Youke UV-1901 UV–vis spectrometer. AFM scans were obtained by Bruker Dimension Edge. STEM was performed by an FEI Titan Themis (300 kV). The contact angle measurement was carried out by a contact angle meter (OSA25, Kruss). SCLC and conductivity were tested using a Keithley 2400 digital source meter. Ultraviolet photoelectron spectroscopy (UPS) was performed using a UPS system (Kratos, Axis Ultra) using HeI radiation of 21.22 eV.

XRD patterns were obtained using an EMPYREAN four-circle diffractometer operated at 40 kV and 30 mA at a scan rate of 20° per minute. SEM images were obtained using a field-emission scanning electron microscope (FEI inspect) at an acceleration voltage of 8 kV. Steady-state PL and TRPL were measured using Edinburgh Instruments, FLS 920 equipped with a light source with an excitation wavelength of 375 nm. EIS plots were measured using a CorrTest Electrochemical Workstation under dark conditions.