Enhanced Photovoltaic Performance of Inverted Perovskite Solar Cells through Surface Modification of a NiOx-Based Hole-Transporting Layer with Quaternary Ammonium Halide–Containing Cellulose Derivatives

In this study, we positioned three quaternary ammonium halide-containing cellulose derivatives (PQF, PQCl, PQBr) as interfacial modification layers between the nickel oxide (NiOx) and methylammonium lead iodide (MAPbI3) layers of inverted perovskite solar cells (PVSCs). Inserting PQCl between the NiOx and MAPbI3 layers improved the interfacial contact, promoted the crystal growth, and passivated the interface and crystal defects, thereby resulting in MAPbI3 layers having larger crystal grains, better crystal quality, and lower surface roughness. Accordingly, the photovoltaic (PV) properties of PVSCs fabricated with PQCl-modified NiOx layers were improved when compared with those of the pristine sample. Furthermore, the PV properties of the PQCl-based PVSCs were much better than those of their PQF- and PQBr-based counterparts. A PVSC fabricated with PQCl-modified NiOx (fluorine-doped tin oxide/NiOx/PQCl-0.05/MAPbI3/PC61BM/bathocuproine/Ag) exhibited the best PV performance, with a photoconversion efficiency (PCE) of 14.40%, an open-circuit voltage of 1.06 V, a short-circuit current density of 18.35 mA/cm3, and a fill factor of 74.0%. Moreover, the PV parameters of the PVSC incorporating the PQCl-modified NiOx were further enhanced when blending MAPbI3 with PQCl. We obtained a PCE of 16.53% for this MAPbI3:PQCl-based PVSC. This PQCl-based PVSC retained 80% of its initial PCE after 900 h of storage under ambient conditions (30 °C; 60% relative humidity).


Introduction
Organic-inorganic hybrid perovskite solar cells (PVSCs) have been attracting a tremendous amount of attention because of their outstanding photoconversion efficiencies (PCEs) and low production costs [1,2]. The PCEs and operational stabilities of PVSCs have improved dramatically within a very short period [3,4]. PVSCs fabricated from methylammonium lead halide perovskites (MAPbX 3 , where MA is a methylammonium (CH 3 NH 3 + ) cation and X is a halide anion) are particularly interesting owing to their good optical absorption properties, high ambipolar charge transporting abilities, weakly bonded excitons that readily dissociate into free charges, and long electron-hole diffusion lengths [5][6][7][8].
In general, PVSCs can be divided into two classes depending on whether they have mesoporous or planar structures. Regular PVSCs with mesoporous structures possess a mesoporous metal oxide (TiO 2 ) layer as the electron transporting layer (ETL), with the perovskite layer coated on top of a TiO 2 layer presenting one of many tested hole transporting layers (HTLs) [9][10][11][12][13]. Although the highest PCEs have typically been produced from PVSCs having mesoscopic structures, the use of TiO 2 is considered to be a disadvantage

Characterization of Cellulose Derivatives and Perovskite Layers
Absorption spectra of MAPbI3 films coated on cellulose derivative-modified NiOxdeposited fluorine-doped tin oxide (FTO) glass were recorded using a Hitachi U3010 UV-Vis spectrometer (Hitachi High-Tech Co., Tokyo, Japan). Photoluminescence (PL) spectra of the cellulose derivative-modified MAPbI3 films were measured using a Hitachi F-4500 fluorescence spectrophotometer (Hitachi High-Tech Co., Tokyo, Japan). Time-resolved PL (TRPL) spectra of MAPbI3 films coated on the PQF-, PQCl-, and PQBr-modified NiOxdeposited FTO glass were recorded using a Horiba Fluoromax-4 spectrometer and Delta Time TCSPC-MCS kit with 405-nm pulsed light emitting diode (LED). The pristine and cellulose derivative-modified MAPbI3 films were encapsulated for measurement of their UV-Vis, PL, and TRPL spectra. The morphologies of the cellulose derivative-modified NiOx and MAPbI3 layers were imaged using atomic force microscopy (AFM, Seiko SII SPA400, Chiba, Japan), performed in the tapping mode. Three runs of surface roughness measurements were performed for each MAPbI3 layer. The surface and cross-sectional morphologies of the MAPbI3 layers deposited on the cellulose derivative-modified NiOx layers were analyzed using cold field emission scanning electron microscopy (FESEM; Hitachi-4800; Integrated Service Tech. Inc., Hinchu, Taiwan; operating voltage: 1.5-2.0 kV). The crystalline structures of the MAPbI3 layers were determined using X-ray powder diffractometry (XRD, Shimadzu SD-D1, Shimadzu Scientific Instrument Co., Taipei, Taiwan), operated with a Cu target at 35 kV and 30 mA. The contact angles (CAs) of water droplets on the cellulose derivative-modified NiOx films were measured using a Kyowa DropMaster optical CA meter (Applied Trentech Inc., Taipei, Taiwan).

Fabrication and Characterization of PVSCs
The PVSCs in this study had the structure FTO-deposited glass/NiOx/cellulose derivative/MAPbI3/PC61BM/BCP/Ag (100 nm), where the NiOx layer was modified with a

Characterization of Cellulose Derivatives and Perovskite Layers
Absorption spectra of MAPbI 3 films coated on cellulose derivative-modified NiO xdeposited fluorine-doped tin oxide (FTO) glass were recorded using a Hitachi U3010 UV-Vis spectrometer (Hitachi High-Tech Co., Tokyo, Japan). Photoluminescence (PL) spectra of the cellulose derivative-modified MAPbI 3 films were measured using a Hitachi F-4500 fluorescence spectrophotometer (Hitachi High-Tech Co., Tokyo, Japan). Time-resolved PL (TRPL) spectra of MAPbI 3 films coated on the PQF-, PQCl-, and PQBr-modified NiO xdeposited FTO glass were recorded using a Horiba Fluoromax-4 spectrometer and Delta Time TCSPC-MCS kit with 405-nm pulsed light emitting diode (LED). The pristine and cellulose derivative-modified MAPbI 3 films were encapsulated for measurement of their UV-Vis, PL, and TRPL spectra. The morphologies of the cellulose derivative-modified NiO x and MAPbI 3 layers were imaged using atomic force microscopy (AFM, Seiko SII SPA400, Chiba, Japan), performed in the tapping mode. Three runs of surface roughness measurements were performed for each MAPbI 3 layer. The surface and cross-sectional morphologies of the MAPbI 3 layers deposited on the cellulose derivative-modified NiO x layers were analyzed using cold field emission scanning electron microscopy (FESEM; Hitachi-4800; Integrated Service Tech. Inc., Hinchu, Taiwan; operating voltage: 1.5-2.0 kV). The crystalline structures of the MAPbI 3 layers were determined using X-ray powder diffractometry (XRD, Shimadzu SD-D1, Shimadzu Scientific Instrument Co., Taipei, Taiwan), operated with a Cu target at 35 kV and 30 mA. The contact angles (CAs) of water droplets on the cellulose derivative-modified NiO x films were measured using a Kyowa DropMaster optical CA meter (Applied Trentech Inc., Taipei, Taiwan).

Fabrication and Characterization of PVSCs
The PVSCs in this study had the structure FTO-deposited glass/NiO x /cellulose derivative/MAPbI 3 /PC 61 BM/BCP/Ag (100 nm), where the NiO x layer was modified with a quaternary ammonium halide-containing cellulose derivative (PQF, PQCl, or PQBr). FTO-deposited glass (sheet resistance: 7 Ω square −1 ) was purchased from Solaronix. The FTO substrates of PVSCs with patterned electrodes were washed well and then cleaned through O 2 plasma treatment. The NiO x precursor solution was prepared by dissolving nickel(II) acetate tetrahydrate (100 mg) in isopropanol and ethanolamine, stirring at 70 • C for several hours, and then filtering through a 0.45-µm polytetrafluoroethylene (PTFE) based filter. The NiO x -based HTL was deposited on the FTO layer through spin-coating of the NiO x precursor solution [50]. The sample was dried at 80 • C for 10 min and then thermally treated at 450 • C for 60 min. Various amounts of PQX cellulose derivative (X = F, Cl, Br; 0.03, 0.05, or 0.1 wt.%) were dissolved in DI water. The resulting solution was deposited on the surface of the NiO x -based HTL. The sample was dried at 100 • C for 30 min. The NiO x layers modified with 0.03, 0.05, and 0.10 wt.% of PQF are named herein as PQF-0.03, PQF-0.05, and PQF-0.10, respectively; the films modified with 0.03, 0.05, and 0.10 wt.% of PQCl are named PQCl-0.03, PQCl-0.05, and PQCl-0.10, respectively; and the films modified with 0.03, 0.05, and 0.10 wt.% of PQBr are named PQBr-0.03, PQBr-0.05, and PQBr-0.10, respectively. The MAI and PbI 2 were stirred in a mixture of DMF and DMSO (4:1, v/v). The MAI and PbI 2 containing solution was deposited on top of the cellulose derivative-modified NiO x -based HTL. The MAPbI 3 deposited substrate was dried at 100 • C for 10 min. Next, a solution of PC 61 BM in CB (20 mg mL −1 ) was deposited on top of the MAPbI 3 layer. A solution (0.3 mL) of BCP in isopropanol (0.5 mg mL −1 ) was then deposited on the PC 61 BM layer. The Ag-based cathode was thermally deposited onto the PC 61 BM layer in a high-vacuum chamber. The photo-active area of the cell was 0.20 cm 2 . The PV properties of the PVSCs were measured using a programmable electrometer equipped with current and voltage sources (Keithley 2400) under illumination with solar-simulating light (100 mW cm −2 ) from an AM1.5 solar simulator (NewPort Oriel 96000).

Characterization of Cellulose Derivative-Modified NiO x Layers
Because the PV parameters of PVSCs is closely related to the morphology of their NiO x -based HTLs, we used SEM and AFM to investigate the morphologies of the cellulose derivative (PQF, PQCl, PQBr)-deposited NiO x layers. SEM images of the PQCl-coated NiO x layer on the FTO substrate are shown in Figure 2. In Figure 2a, we observe the nanosheet structure of the pristine NiO x . The shape and size of the NiO x nanosheets did not change significantly after coating with different concentrations of PQCl (Figure 2b-d), suggesting the presence of thin films of this cellulose derivative. The surface morphologies of the PQF-and PQBr-deposited NiO x layers were similar to those of the PQCl-deposited NiO x layers (Figures S1 and S2). Figure 3 presents topographic images of the PQClcoated NiO x layers on the FTO substrates. These AFM images indicate that the surface morphology of the NiO x layer changed after coating with PQCl, with a nanoparticle structure appearing. Nevertheless, the surface morphology was not changed significantly when coating with different concentrations of PQCl. The surface roughness of the PQClmodified NiO x layers was slightly enhanced when compared with that of the pristine NiO x layer (Table 1). We observed similar features in the AFM images of the PQF-and PQBr-deposited NiO x layers ( Figures S3 and S4). Furthermore, we used a CA meter to examine the hydrophobicity/hydrophilicity of the surface-modified NiO x layers. Figure S5 displays photographs of water droplets on the pristine and cellulose derivative-deposited NiO x layers. Table 1 reveals that the CAs of the cellulose derivative-modified NiO x layers were lower than that of the pristine NiO x layer. Shen et al. reported that enhancing the hydrophilicity of NiO x -based HTLs encourages the formation of more uniform and larger crystal grains in MAPbI 3 layers [45]. Nevertheless, the uniformity and crystal size of MAPbI 3 were not only affected by the wettability of the HTL.

Morphologies of Perovskite Films Deposited on Cellulose Derivative-Modified NiO x Layers
To investigate the effects of the quaternary ammonium halide-functionalized cellulose derivatives (PQF, PQCl, PQBr) as interfacial layers on the crystallization of the perovskite films, we used SEM to examine the morphologies and film qualities of MAPbI 3 deposited on the cellulose derivative-modified NiO x layers, thereby allowing us to determine the optimal processing conditions for the preparation of the PVSCs. Figure 4 and Figure S6 display the top-view and cross-sectional SEM images, respectively, of MAPbI 3 films that had been deposited on the interfacial modifiers PQCl, PQF, and PQBr that had been subjected to annealing at 100 • C for 10 min. Moreover, Figure 5, Figure S7 and Figure S8 present the crystal grain size distributions of the MAPbI 3 films deposited on the PQCl, PQF, and PQBr interfacial layers, respectively. Table 1 summarizes the average crystal sizes of the PQCl-, PQF-, and PQBr-based MAPbI 3 films, calculated using Image J1 software. The crystal grains that appeared after growing the MAPbI 3 layer on the PQCl-modified NiO x HTL were larger than those of the pristine NiO x HTL. The largest crystal grains of MAPbI 3 were those for the sample prepared using the PQCl-0.05-modified NiO x . Nevertheless, the standard derivation (SD) of the crystal grain size distribution of the PQCl-0.05-modified MAPbI 3 layer was slightly larger than that of the pristine sample. Figure 6 provides a schematic representation of the crystal growth of an MAPbI 3 film on the PQCl-modified NiO x -based HTL. The quaternary ammonium halide units of PQCl have a chemical structure similar to that of MAI, suggesting that they might participate in the perovskite crystallization process through partial substitution of the MA cations with the quaternary ammonium cations as well as of the I − anions with Cl − anions [76,77]. The quaternary ammonium halide-containing side chains of the cellulose derivative PQCl-0.05 on the surface of the NiO x layer appeared to help with the repair of the crystal defects and promoted the crystal growth of MAPbI 3 , encouraging the formation of more uniform and larger crystal grains in the perovskite film [76]. When we coated a higher content of the cellulose derivative (PQCl-0.10) on the surface of the NiO x layer, interfusion of the large polymer backbone into the perovskite layer occurred during crystal formation in the MAPbI 3 layer, thereby decreasing the average size of the perovskite crystals ( Figure 6) [67,76]. The corresponding effects of PQF and PQBr at repairing the crystal defects were much poorer than that of PQCl [67,77]. Moreover, the average sizes of the MAPbI 3 crystal grains coated on the PQF-and PQBr-deposited NiO x were smaller than that of the pristine NiO x . The average sizes of the MAPbI 3 crystal grains decreased upon coating the cellulose derivatives PQF and PQBr at higher concentrations onto the surface of NiO x layer. Cross-sectional SEM images indicated that the crystal grains of MAPbI 3 became more densely packed after inserting an interfacial layer of PQCl between the NiO x and MAPbI 3 layers. Relative to the pristine MAPbI 3 film, the grain boundaries between the various crystal grains became vaguer for the cellulose derivative-incorporated MAPbI 3 films, resulting in higher coverage of the perovskite films [67]. The repairing of crystal defects mediated by the quaternary ammonium halides presumably helped to modify the grain boundaries [77]. The minimization of grain boundaries and the enhanced packing density of crystal grains would presumably be favorable for charge transfer in the perovskite films. The crosssectional SEM images indicated that the thickness of the perovskite layer did not change significantly after increasing the PQCl or PQF content ( Figure 4 and Figure S6), but it did decrease for the PQBr-modified perovskite layer. A thinner MAPbI 3 layer would presumably result in a lower capacity to absorb solar light and poorer PV performance from the corresponding PVSC.
AFM microscopy confirmed the interfacial effects of the PQF, PQCl, and PQBr on the morphologies of the MAPbI 3 films. Figure 7, Figure S9 and Figure S10 present AFM images of the MAPbI 3 films deposited on the interfacial layers of PQCl, PQF, and PQBr, respectively. Table 1 summarizes the statistical surface roughness of the MAPbI 3 films deposited on the PQCl, PQF, and PQBr interfacial layers. The AFM images indicate that largest crystal grains appeared after growing the MAPbI 3 layer on the PQCl-modified NiO x HTL. Moreover, the average size of the MAPbI 3 crystal grains decreased when coating the NiO x layer with a higher concentration of PQCl. The surface roughness of the MAPbI 3 films coated on the cellulose-modified NiO x layers was slightly lower than that on the pristine NiO x layer. The surface roughness of the MAPbI 3 films was slightly higher when the NiO x film had been coated using a solution of 0.10 wt.% of the cellulose derivative, relative to those obtained using the 0.03 and 0.05 wt.% solutions. Inserting the PQCl at the NiO x -MAPbI 3 interface promoted the formation of more uniform and larger crystal grains, and decreased the surface roughness of the MAPbI 3 film, presumably through the defect passivation effect of PQCl [78]. We suspected that a lower degree of light scattering and a higher absorption capacity, both favorable for enhancing PV properties, would be obtained for MAPbI 3 films having smoother surfaces and better film quality [79,80]. Nevertheless, the modification effects of PQF and PQBr at the MAPbI 3 -NiO x interfaces were much poorer than that of PQCl. The average sizes of the MAPbI 3 crystal grains coated on the PQF-and PQBr-deposited NiO x were smaller than that on the pristine NiO x .

XRD Images of Perovskite Films Deposited on Cellulose Derivative-Modified NiOx Layers
XRD was used to examine the crystal structures of the MAPbI3 films deposited on the cellulose derivative-modified NiOx layers. Figures 8, S11, and S12 reveal that the patterns of the MAPbI3 films formed on the PQF-, PQCl-, and PQBr-modified NiOx layers featured the typical diffraction peaks of MAPbI3 based perovskites, including characteristic peaks at 14.2, 28.4, and 43.08° corresponding to the (110), (220), and (330) phases, respectively [81][82][83]. These diffraction peaks indicated the formation of tetragonal crystal structures having lattice constants a and b each equal to 8.883 Å and c equal to 12.677 Å [82]. Moreover, the intensities of the (110) peaks for the MAPbI3 films coated on the PQCl-0.03-and PQCl-0.05-modified NiOx layers were higher than that for the MAPbI3 coated on the pristine NiOx layer ( Figure 8). The highest intensity of the (110) peak was that for the MAPbI3 film deposited on the PQCl-0.05-modified NiOx layer. A higher (110) diffraction peak intensity correlates with a better crystal quality for MAPbI3 films [45,51,65,70]. An MAPbI3 film of better crystal quality tends to display improved electronic properties, including

XRD Images of Perovskite Films Deposited on Cellulose Derivative-Modified NiO x Layers
XRD was used to examine the crystal structures of the MAPbI 3 films deposited on the cellulose derivative-modified NiO x layers. Figure 8, Figure S11 and Figure S12 reveal that the patterns of the MAPbI 3 films formed on the PQF-, PQCl-, and PQBr-modified NiO x layers featured the typical diffraction peaks of MAPbI 3 based perovskites, including characteristic peaks at 14.2, 28.4, and 43.08 • corresponding to the (110), (220), and (330) phases, respectively [81][82][83]. These diffraction peaks indicated the formation of tetragonal crystal structures having lattice constants a and b each equal to 8.883 Å and c equal to 12.677 Å [82]. Moreover, the intensities of the (110) peaks for the MAPbI 3 films coated on the PQCl-0.03-and PQCl-0.05-modified NiO x layers were higher than that for the MAPbI 3 coated on the pristine NiO x layer ( Figure 8). The highest intensity of the (110) peak was that for the MAPbI 3 film deposited on the PQCl-0.05-modified NiO x layer. A higher (110) diffraction peak intensity correlates with a better crystal quality for MAPbI 3 films [45,51,65,70]. An MAPbI 3 film of better crystal quality tends to display improved electronic properties, including greater charge carrier transport [45,51,65,70]. The presence of quaternary ammonium cations and Cl − anions at the NiO x -MAPbI 3 interface can passivate the positively charged defects in the perovskite layer induced by the loss of I − anions. Furthermore, the ammonium unit can passivate Pb-I antisite defects through electrostatic interactions [76]. Therefore, we found that the crystal growth of MAPbI 3 was promoted through the crystal defect repairing effect of PQCl. In contrast, the diffraction intensities of the (110) peaks for the MAPbI 3 layers deposited on the PQF-and PQBrmodified NiO x layers were lower when compared with that of the MAPbI 3 deposited on the pristine NiO x layer (Figures S11 and S12). Relative to the effect of PQCl, the interfacial layers of PQF and PQBr led to poorer crystal growth of the perovskite. We suspect that greater electronegativity limited the dissociation of F − anions from the quaternary ammonium fluoride, such that fewer F − anions could compensate for the I − vacancies of the perovskite [79,84]. Furthermore, the relatively large ionic radius of the Br − anion would affect its ability to compensate for ion of I − vacancies. As a result, the crystal defect repairing effects of PQF and PQBr were both poorer than that of PQCl [70]. Table 1 summarizes the crystal sizes in the MAPbI 3 films coated on the PQF-, PQCl-, and PQBr-modified NiO x layers. According to the Scherrer equation, these crystal sizes were calculated from the full width at half maximum (FWHM) of the (110) diffraction peak [85]. The average crystal sizes were greatest for the MAPbI 3 layers that had been deposited on the PQClmodified NiO x layers. The largest crystals were those in the PQCl-0.05-based MAPbI 3 film. The crystal sizes were lower for the MAPbI 3 layers deposited on the PQF-and PQBrmodified NiO x layers, and they decreased for the NiO x layers that had been treated with higher concentrations of the PQF and PQBr solutions. We attribute the smaller crystals to the presence of a higher content of polymer chains at the MAPbI 3 -NiO x interface. The steric bulk of the cellulose derivative-based polymer backbone presumably inhibited the formation of crystals of MAPbI 3 , leading to smaller crystals of the MAPbI 3 [70]. greater charge carrier transport [45,51,65,70]. The presence of quaternary ammonium cations and Cl − anions at the NiOx-MAPbI3 interface can passivate the positively charged defects in the perovskite layer induced by the loss of I − anions. Furthermore, the ammonium unit can passivate Pb-I antisite defects through electrostatic interactions [76]. Therefore, we found that the crystal growth of MAPbI3 was promoted through the crystal defect repairing effect of PQCl. In contrast, the diffraction intensities of the (110) peaks for the MAPbI3 layers deposited on the PQF-and PQBr-modified NiOx layers were lower when compared with that of the MAPbI3 deposited on the pristine NiOx layer (Figures S11 and  S12). Relative to the effect of PQCl, the interfacial layers of PQF and PQBr led to poorer crystal growth of the perovskite. We suspect that greater electronegativity limited the dissociation of F − anions from the quaternary ammonium fluoride, such that fewer F − anions could compensate for the I − vacancies of the perovskite [79,84]. Furthermore, the relatively large ionic radius of the Br − anion would affect its ability to compensate for ion of I − vacancies. As a result, the crystal defect repairing effects of PQF and PQBr were both poorer than that of PQCl [70]. Table 1 summarizes the crystal sizes in the MAPbI3 films coated on the PQF-, PQCl-, and PQBr-modified NiOx layers. According to the Scherrer equation, these crystal sizes were calculated from the full width at half maximum (FWHM) of the (110) diffraction peak [85]. The average crystal sizes were greatest for the MAPbI3 layers that had been deposited on the PQCl-modified NiOx layers. The largest crystals were those in the PQCl-0.05-based MAPbI3 film. The crystal sizes were lower for the MAPbI3 layers deposited on the PQF-and PQBr-modified NiOx layers, and they decreased for the NiOx layers that had been treated with higher concentrations of the PQF and PQBr solutions. We attribute the smaller crystals to the presence of a higher content of polymer chains at the MAPbI3-NiOx interface. The steric bulk of the cellulose derivative-based polymer backbone presumably inhibited the formation of crystals of MAPbI3, leading to smaller crystals of the MAPbI3 [70].

UV-Vis Absorption Spectra of MAPbI3 Films Deposited on Cellulose Derivative-Modified NiOx Layers
We recorded UV-Vis spectra of the MAPbI3 films that had been deposited on the cellulose derivative-modified NiOx layers to examine the effects of the interfacial layers

UV-Vis Absorption Spectra of MAPbI 3 Films Deposited on Cellulose Derivative-Modified NiO x Layers
We recorded UV-Vis spectra of the MAPbI 3 films that had been deposited on the cellulose derivative-modified NiO x layers to examine the effects of the interfacial layers on the optical absorption properties of the perovskite films ( Figure 9). Compared with the pristine MAPbI 3 film, the MAPbI 3 films deposited on the PQCl-0.03-and PQCl-0.05-modified NiO x layers absorbed more strongly over almost the entire spectral range. The highest absorption intensity was that for the PQCl-0.05-based MAPbI 3 film, consistent with its greater crystallinity, minimized grain boundaries, increased coverage, and lower reflection. Notably, however, the absorption intensity of the PQCl-0.10-based MAPbI 3 film was lower than that of the pristine MAPbI 3 film. Furthermore, as compared with the pristine MAPbI 3 film, the absorption intensities were lower for the MAPbI 3 films deposited on the PQF-and PQBr-modified NiO x layers.

PV Properties of PVSCs Incorporating Cellulose Derivative-Modified NiOx Layers
The optimized spin-coating procedure was used to prepare PVSCs incorporating the PL spectroscopy was used to study the interfacial effects of the cellulose derivatives PQF, PQCl, and PQBr on the PL properties of the MAPbI 3 films. Figure 10a displays the PL spectra of the MAPbI 3 films deposited on the cellulose derivative-modified NiO x layers. The wavelength of maximal PL of the MAPbI 3 films appeared near 768 nm. Relative to the signal for the MAPbI 3 film coated on the pristine NiO x layer, the PL intensities were lower for the MAPbI 3 films deposited on the PQCl-0.03-and PQCl-0.05-modified NiO x layers, implying that the charge separation capacity was enhanced for the MAPbI 3 perovskite films deposited on the PQCl interfacial layers [24]. Moreover, the PL intensity of the MAPbI 3 film deposited on PQCl-0.05 was lower than those of the PQCl-0.03-and PQCl-0.10-based MAPbI 3 films. We attribute the low PL intensity of the PQCl-0.05-modified MAPbI 3 film to the decrease in the number of crystal defects and the excellent defect passivation occurring at the NiO x -MAPbI 3 interface. Relative to the signal for the pristine MAPbI 3 film, the PL intensities were enhanced for the MAPbI 3 layers deposited on the PQF-and PQBr-modified NiO x layers, implying that PQF and PQBr could not passivate the interfacial defects and, thereby, inferior interfacial contact and poorer charge-separation capacity occurred at their NiO x -MAPbI 3 interfaces.
TRPL spectra were used to study the influence of the interfacial modifiers PQF, PQCl, and PQBr on the charge recombination processes of the perovskite films (Figure 10b). The carrier lifetime was obtained by fitting the PL data to a double-exponential decay model [86,87]: where A and B are constants and τ 1 and τ 2 are the fast and slow decay constants, respectively. The fitting results for the TRPL spectra are summarized in Table 2. Here, the average lifetime of the cell was calculated from the average of the fast and slow decay constants, obtained using the equation τ avg = (Aτ 1 2 +Bτ 2 2 )/(Aτ 1 + Bτ 2 ) The constant τ 1 is related to defect recombination or interfacial charge transport from MAPbI 3 to the HTLs; τ 2 is related to radiative recombination [86]. The lifetimes for the PQCl-0.03-and PQCl-0.05-modified MAPbI 3 films samples were shorter than that for the pristine MAPbI 3 film, indicating that the addition of PQCl as an interfacial modifier could minimize the number of defects in the MAPbI 3 film, enhance the degree of charge extraction, and decrease the non-radiative combination loss. Nevertheless, inserting an excess of PQCl at the NiO x -MAPbI 3 interface did not lower the number of defects of the MAPbI 3 film, with the PQCl-0.10 sample exhibiting a longer lifetime than that of the pristine MAPbI 3 film. Apart from that, the lifetimes of the PQF-and PQBr-modified MAPbI 3 films were longer than that of the pristine MAPbI 3 layer, and they increased significantly when higher contents of PQF and PQBr were present at the NiO x -MAPbI 3 interfaces. The TRPL spectra indicated that the carrier lifetimes of the PQCl-modified MAPbI 3 films were much shorter than those of the MAPbI 3 films coated on the PQF and PQBr layers.

PV Properties of PVSCs Incorporating Cellulose Derivative-Modified NiO x Layers
The optimized spin-coating procedure was used to prepare PVSCs incorporating the cellulose derivatives at the NiOx-MAPbI 3 interfaces. Figure 11 presents the photocurrent density-voltage plots of the PVSCs fabricated using the NiO x layers modified with various contents of the cellulose derivatives PQF, PQCl, and PQBr. Figure 12 displays statistical box plots for the PV parameters of 20 un-encapsulated pristine and cellulose derivative-modified MAPbI 3 -based PVSCs. The statistical values of the PV properties of these PVSCs are summarized in Table 3, including their open-circuit voltages (V OC ), short-circuit current densities (J SC ), fill factors (FFs), and PCEs. We performed 20 runs of PV evaluation measurements for each cell. A value of V OC of 1.07 V, a value of J SC of 17.98 mA cm −2 , an FF of 69.3%, and a PCE of 13.33% were obtained for PVSC I, fabricated from the NiO x HTL prepared without modification with a cellulose derivative as interfacial layer. These PV parameters are comparable with those of other published PVSC having similar architectures [87]. Relative to the pristine PVSC I, we obtained superior PV properties for PVSCs V and VI, based on the PQCl-0.03-and PQCl-0.05-modified NiO x HTLs, respectively, but not for PVSC VII. The PV properties enhanced after increasing the content of PQCl at the MAPbI 3 -NiO x interfaces for PVSCs V and VI. The highest performance was that for PVSC VI, incorporating the PQCl-0.05 film: a value of V OC of 1.06 V, a value of J SC of 18.35 mA cm −2 , an FF of 74.0%, and a PCE of 14.40%. These high values of V OC , J SC , and FF are consistent with the higher crystallinity and stronger UV-Vis absorptions of the MAPbI 3 films deposited on the PQCl-modified NiO x layers. The presence of PQCl at the MAPbI 3 -NiO x interface, with Cl − anions in the quaternary ammonium units, promoted the crystal growth and enhanced the MAPbI 3 film quality, leading to efficient charge separation and extraction and a low degree of charge recombination [88]. Incident photon-to-current efficiency (IPCE) spectroscopy confirmed the improvements in the values of J SC of the PVSCs fabricated from the cellulose derivative-modified NiO x HTLs ( Figure 13). The IPCEs of the PQCl-0.03-and PQCl-0.05-based PVSCs V and VI, respectively, were higher than that of PVSC I fabricating from the pristine NiOx film. Thus, we conclude that the incorporation of PQCl at the MAPbI 3 -NiO x interface had significant effects on repairing the crystal defects and enhancing the crystallinity of the MAPbI 3 film. As a result, the PV properties were improved for the PVSCs fabricated from the PQCl-0.03 and PQCl-0.05 samples. In addition, the PV properties of the PQCl-0.10-based PVSC IV were poorer than those of PVSC I (based on the pristine NiO x HTL), consistent with the lower crystallinity (as evidenced from SEM and AFM images and XRD patterns) of the PQCl-0.10-modified MAPbI 3 film. Nevertheless, the presence of an excessive amount of PQCl at the MAPbI 3 -NiO x interface did not improve the PV performance. When a higher content of PQCl was coated on the surface of the NiO x layer, the sterically bulky polymer interfused among the MAPbI 3 crystals and limited their growth, thereby decreasing the crystallinity, the absorption intensity, and the PV performance of the PQCl-0.10-based PVSC IV. In addition, the PV properties of the PVSCs were poorer when the MAPbI 3 films were deposited on the PQF-and PQBr-modified NiO x HTLs, implying that the effects of PQF and PQBr on crystal defect repair were much poorer than that of PQCl. Moreover, the PV performance of the PVSCs decreased when higher amounts of PQF (PVSCs II, III, and IV) and PQBr (PVSCs VIII, IX, and X) were present at the MAPbI 3 -NiO x interfaces. Nevertheless, the PCEs of the PQF-modified PVSCs (PVSCs II, III, and IV) were slightly higher than those of the PQBr-modified PVSCs (PVSCs VIII, IX, and X). Compared with the PQBr-modified MAPbI 3 layers, the higher UV-Vis spectral absorption intensities and larger average crystal grain sizes of the PQF-modified MAPbI 3 films resulted in the higher IPCEs and PV performance parameters of PVSCs II-IV. Based on these findings, we conclude that the PV performance of MAPbI 3 -based PVSCs can be improved through modification of the MAPbI 3 -NiO x interface with an optimized amount of PQCl, which has a positive effect on crystal growth and crystal defect repair in the MAPbI 3 layer. Polymers 2023, 15, x FOR PEER REVIEW 16 of 24 Figure 11. Current density-voltage characteristics of illuminated (AM 1.5G, 100 mW cm −2 ) PVSCs.          We measured the hole mobility in the MAPbI 3 layers to further examine the passivation effects of the cellulose derivative-based interfacial modifiers on the perovskite layers ( Figure 14). We calculated the mobility (µ) of the perovskite in the space-charge limited current regime using the equation where J is the current density, ε o is the vacuum permittivity (8.854 × 10 −12 F m −1 ), ε r is the relative permittivity of MAPbI 3 (32), V is the base voltage, and L is the thickness of the MAPbI 3 layer (410 nm) [66,89,90]. The estimated hole mobilities of the pristine and PQF-0.05-, PQCl-0.05-, and PQBr-0.05-based hole-only devices were 3.92 × 10 −3 , 2.68 × 10 −3 , 4.20 × 10 −3 , and 2.14 × 10 −3 cm 2 V −1 s −1 , respectively. Thus, the hole mobility of the PQCl-0.05-modified MAPbI 3 layer was greater than that of the pristine sample, while those of the PQF-0.05-and PQBr-0.05-modified samples were lower. We infer that the passivation effect of PQCl on the perovskite layer was much better than those of PQF and PQBr.

PVSC
Interfacial Layer We measured the hole mobility in the MAPbI3 layers to further exa sivation effects of the cellulose derivative-based interfacial modifiers on t layers ( Figure 14). We calculated the mobility (μ) of the perovskite in the limited current regime using the equation where J is the current density, εo is the vacuum permittivity (8.854 × 10 −12 F relative permittivity of MAPbI3 (32), V is the base voltage, and L is the th MAPbI3 layer (410 nm) [66,89,90]. The estimated hole mobilities of the pris 0.05-, PQCl-0.05-, and PQBr-0.05-based hole-only devices were 3.92 × 10 −3 , 2 × 10 −3 , and 2.14 × 10 −3 cm 2 V −1 s −1 , respectively. Thus, the hole mobility of th modified MAPbI3 layer was greater than that of the pristine sample, whil PQF-0.05-and PQBr-0.05-modified samples were lower. We infer that the p fect of PQCl on the perovskite layer was much better than those of PQF and  To further examine the effects of the addition of PQCl on the morphologies and optical properties of the MAPbI 3 -based perovskite films, we prepared a perovskite film (MAPbI 3 :PQCl-0.06) from a blend of PQCl (0.06 wt.%) and MAPbI 3 deposited on the surface of the PQCl-0.05-modified NiO x layer. The SEM images in Figure 15a,b reveal that the average crystal grain size (111 nm) of the MAPbI 3 :PQCl-0.06 film was much lower than that (272 nm) of the pristine MAPbI 3 film (Figure 4c). We suspect that the steric bulk and low thermal mobility of the large cellulose derivative backbones suppressed the formation of crystals of MAPbI 3 , thereby decreasing their average size [70]. Nevertheless, these crystal grains of smaller size underwent denser packing. As compared with the pristine MAPbI 3 film, the grain boundaries among the various crystal grains were more vague for the MAPbI 3 :PQCl-0.06 film, resulting in a higher coverage of the perovskite film. The repairing of crystal defects induced by the quaternary ammonium halide units presumably helped to connect the crystal grains [76]. The decrease in the number of grain boundaries and the greater packing density of the crystal grains would both favor charge transfer in the perovskite film [78]. The XRD patterns for the perovskite films coated on the PQCl-0.05-modified NiO x layer indicated (Figure 15c) that the crystal diffraction intensity of the MAPbI 3 :PQCl-0.06 blend film was slightly lower than that of MAPbI 3 film. Moreover, the diffraction intensity of the MAPbI 3 :PQCl-0.06 blend film was greater than that of the MAPbI 3 film coated on the pristine NiO x layer. In addition, the PL intensity of the MAPbI 3 :PQCl-0.06 blend film was lower than that of the MAPbI 3 film, implying that the charge separation capacity was enhanced after the addition of PQCl in the MAPbI 3 perovskite layer [24]. Consequently, the PVSC XI device (FTO/NiO x /PQCl-0.05/MAPbI 3 :PQCl-0.06/PC 61 BM/BCP/Ag) fabricated from the MAPbI 3 :PQCl-0.06 blend film exhibited values of J SC and PCE higher than those of the MAPbI 3 -based PVSCs I and VI ( Figure 11). Indeed, the PVSC XI exhibited a PCE of 16.53%, a value of V OC of 1.06 V, a value of J SC of 21.93 mA cm −2 , and an FF of 71.0% (Table 3). A high efficiency of 16.53% from forward scanning and a comparable efficiency of 16.48% from reverse scanning were obtained for the PVSC XI ( Figure S13). A negligible hysteresis of the current density-voltage curve implies the balanced charge transport at the NiO X /MAPbI 3 interface and good charge transport inside the MAPbI 3 layer for the PVSC XI [19]. Furthermore, the IPCE of PVSC XI incorporating the MAPbI 3 :PQCl-0.06 blend film was higher than those of the MAPbI 3 -based PVSCs I and VI ( Figure 13).
The storage stability of the cellulose derivative-based PVSCs (PVSC I, PVSC II, PVSC VI, and PVSC-XI) measured at 30 • C and 60% relative humidity is displayed in Figure 16. The PCE-stability of the PQCl incorporated PVSC VI and PVSC XI was superior to those of the Pristine, PQF-0.03-, and PQBr-0.03-based PVSCs (PVSC-I, PVSC-II, and PVSC-VIII). The lifetime of PQCl-based PVSC VI and PVSC XI without encapsulation was more than 900 h. The incorporation of PQCl at MAPbI 3 /NiO X interface promotes the crystal growth and effective crystal defect passivation for stabilizing perovskite crystal structures. Moreover, the stability of the PVSC was further enhanced by the addition of PQCl-0.06 in the MAPbI 3 layer for the PVSC XI. The interfacial layer effect of PQCl on the PV stability was much better than those of PQF and PQBr.

Conclusions
We used a series of cellulose derivatives (PQF, PQCl, PQBr) individually as interfacial modifiers of MAPbI3-NiOx interfaces and prepared corresponding PVSCs. The presence of quaternary ammonium cations and Cl − anions at the NiOx-MAPbI3 interface can passivate the positively charged defects in the perovskite layer induced by the loss of I − anions. Moreover, the ammonium unit can passivate Pb-I antisite defects through electrostatic interactions. The deposition of an appropriate amount of PQCl on the NiOx layer led to repair of the grain boundary defects, promoted crystal growth, and increased the

Conclusions
We used a series of cellulose derivatives (PQF, PQCl, PQBr) individually as interfacial modifiers of MAPbI3-NiOx interfaces and prepared corresponding PVSCs. The presence of quaternary ammonium cations and Cl − anions at the NiOx-MAPbI3 interface can passivate the positively charged defects in the perovskite layer induced by the loss of I − anions. Moreover, the ammonium unit can passivate Pb-I antisite defects through electrostatic interactions. The deposition of an appropriate amount of PQCl on the NiOx layer led to repair of the grain boundary defects, promoted crystal growth, and increased the

Conclusions
We used a series of cellulose derivatives (PQF, PQCl, PQBr) individually as interfacial modifiers of MAPbI 3 -NiO x interfaces and prepared corresponding PVSCs. The presence of quaternary ammonium cations and Cl − anions at the NiO x -MAPbI 3 interface can passivate the positively charged defects in the perovskite layer induced by the loss of I − anions. Moreover, the ammonium unit can passivate Pb-I antisite defects through electrostatic interactions. The deposition of an appropriate amount of PQCl on the NiO x layer led to repair of the grain boundary defects, promoted crystal growth, and increased the light absorption and hole mobility of the MAPbI 3 film. Nevertheless, the deposition of an excess of POCl on the NiO x suppressed crystal growth of the perovskite through the effect of the steric bulk of the polymer backbone of PQCl. Relative to the effect of PQCl, the interfacial layers of PQF and PQBr led to poorer crystal growth of the perovskite. The PV properties of PVSCs fabricated with PQCl-modified NiO x layers were improved when compared with those of the pristine sample. Furthermore, the PV parameters of a PQCl-modified, NiO xbased PVSC were further enhanced after blending the MAPbI 3 with PQCl. As compared with the pristine MAPbI 3 film, the grain boundaries among the various crystal grains became more vague in the MAPbI 3 :PQCl-0.06 film, resulting in a higher coverage of the perovskite film. The decrease in the number of grain boundaries and the greater packing density of the crystal grains both promoted charge transfer in the MAPbI 3 film.