Molecular Tailoring of Pyridine Core-Based Hole Selective Layer for Lead Free Double Perovskite Solar Cells Fabrication

To solve the toxicity issues related to lead-based halide perovskite solar cells, the lead-free double halide perovskite Cs2AgBiBr6 is proposed. However, reduced rate of charge transfer in double perovskites affects optoelectronic performance. We designed a series of pyridine-based small molecules with four different arms attached to the pyridine core as hole-selective materials by using interface engineering. We quantified how arm modulation affects the structure–property–device performance relationship. Electrical, structural, and spectroscopic investigations show that the N3,N3,N6,N6-tetrakis(4-methoxyphenyl)-9H-carbazole-3,6-diamine arm’s robust association with the pyridine core results in an efficient hole extraction for PyDAnCBZ due to higher spin density close to the pyridine core. The solar cells fabricated using Cs2AgBiBr6 as a light harvester and PyDAnCBZ as the hole selective layer measured an unprecedented 2.9% power conversion efficiency. Our computed road map suggests achieving ∼5% efficiency through fine-tuning of Cs2AgBiBr6. Our findings reveal the principles for designing small molecules for electro-optical applications as well as a synergistic route to develop inorganic lead-free perovskite materials for solar applications.


INTRODUCTION
Lead halide perovskite has shown remarkable optoelectrical and photovoltaics properties, and solution-processibility offers cost-effective deposition routes, arguably increasing its acceptability as a potential candidate for next-generation photovoltaics (PV). 1−4 However, its poor stability and the toxicity of lead restrict its immediate commercial endeavor. 5−8 To close this gap, significant attempts are being made to exploit interface materials to improve stability, 9,10 and the development of an eco-friendly perovskite for PV. 11,12 Double metal perovskite (Cs 2 AgBiBr 6 ) is an attractive alternative for thin-film PV, attributed to its enhanced thermal stability, less toxicity, and favorable photoelectrical traits including long carrier lifespan and moderate charge carrier mobility. 13−15 Efforts are being made to enhance the performance of Cs 2 AgBiBr 6 -based solar cells. This includes innovative deposition processes, 14,16,17 dye sensitization, 13,18,19 additive engineering, 20−22 binary solvent, 14,23 and interface engineering 24−27 to assist high-quality film formation, which in turn facilitate carrier transport, and thus power conversion efficiency (PCE). Despite these advancements, the PCEs are far below compared to the Pb-based perovskites with comparable bandgaps. This entails new approaches in the investigation to boost PV performance.
Small molecules with unique features offer prospective for application in lead-free Cs 2 AgBiBr 6 : low batch-to-batch variance, diversity via molecular design methodologies, and cost-effectiveness. 29 The pyridine can be served as an electrondeficient and passivation function unit to create new HSLs. 30 Pyridine-based HSLs and associated synthesis methodologies have been reported. 31,32 We put forward the linking topology of the pyridine core and common arms, as well as the Nposition effects on pyridine-based HSLs. 33,34 The arm modulation based on pyridine-based HSLs has not been probed yet. Notably, the incorporation of innovative small organic molecule as HSLs in Cs 2 AgBiBr 6 perovskite based solar cells remains unexplored. Here, we developed four pyridine-based HSLs using certain arms in the 2,6-position linking of pyridine: N,N-di-4-anisylamino (DAn), (N,N-bis(4methoxyphenyl)aniline) (PDAn), (N 3 ,N 3 ,N 6 ,N 6 -tetrakis(4-methoxyphenyl)-9H-carbazole-3,6-diamine) (DAnCBZ), and 4,4′-(10H-phenothiazine-3,7-diyl)bis(N,N-bis(4-methoxyphenyl)-aniline) (PTPDAn), named as PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn, respectively. We employed an array of techniques to quantify the chemical and optoelectronic properties. Moreover, we fabricated the Cs 2 AgBiBr 6 -based solar cells employing pyridine-based HSLs, and the PyDAnCBZ-based devices showed superior performance than other HSLs. Furthermore, we computed the photovoltaic simulation to unravel the road map for the lead-free device performance enhancement method.

RESULTS AND DISCUSSION
2.1. Synthesis. Herein, we critically analyze arm modulation employing small molecules with a pyridine core and its 2,6-position linking topology. The DAn, PDAn, and DAnCBZ moieties have frequently been utilized as arms in the design of organic hole-selective materials. 35 Moreover, PTPDAn, which consists of phenothiazine interacting with two PDAn arms, is the largest arm molecule. The synthetic routes of these materials are illustrated in Scheme 1. Four precursors were made using prior procedures that are detailed in the Supporting Information. The PyDAn, PyDAnCBZ, and PyPTPDAn were synthesized from the palladium acetatecatalyzed Buchwald-Hartwig cross-coupling reaction with Scheme 1. Synthetic Routes for PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn yields of 70, 46, and 52%, respectively, while PyPDAn was synthesized by the Suzuki cross-coupling reaction using palladium-tetrakis(triphenylphosphine) as a catalyst and resulted in the final molecule with a yield of 60%. The molecular structures were established by 1 H NMR, 13 C NMR spectroscopy, and HRMS ( Figure S1 in the Supporting Information). All the materials are soluble in common organic solvents.

Photophysical and Electrochemical Properties.
The normalized optical absorption spectra of pyridine-based small molecules are presented (Figure 1a), and the data are compiled in Table 1. The spectra revealed two principal absorption bands: 250−310 and 340−420 nm. This is in line with previous results on other pyridine-based HSLs. 33,36 The electronic transition causes an absorbed peak in the lower wavelength. The action of intramolecular charge transfer (ICT) between the armed group and electron-deficient pyridine core produced a substantial peak in the 340−420 nm region. According to the formula E g opt = 1240/λ onset , the optical bandgap (E g opt ) values were calculated to be 3.04, 3.06, 2.85, and 3.11 eV, respectively.
It can be deduced from Figure 1b that the peak maxima in photoluminescence spectra for PyDAn, PyPDAn, Py-DAnCBZ, and PyPTPDAn appears at 423, 506, 480, and 480 nm, with Stokes shifts of 145, 143, 172, and 128 nm, respectively. For organic materials, larger Stokes shifts result from a shift in the material's structure between the ground and excited states. It was observed that PyDAnCBZ has a larger Stokes shift than the others, which occurs from the modification of the geometrical configuration generated by fluorescence illumination and can enhance the capacity to fill pores, leading to higher charge transporting ability. 37,38 The energy levels of HSLs were measured using cyclic voltammetry (Figure 1c and Figure S2). The PyDAn and PyPDAn showed only one reversible event, while the PyDAnCBZ and PyPTPDAn showed two reversible anodic events. Two DAn units connect to a carbazole in the PyDAnCBZ's molecular configuration, and two PDAn units connect to a phenothiazine in PyPTPDAn's, which may form stable dications owing to charge delocalization throughout the conjugation chain, whereas PyDAn and PyPDAn can only undergo single cation transformation. The PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn have starting oxidation potentials of 0.62, 0.54, 0.36, and 0.60 eV, respectively. The DAnCBZ arm's greater electron donation than other arms contributed to PyDAnCBZ's lower oxidation potential. E HOMO values of these compounds were −5.02, −4.94, −4.76, and −5.00 eV, calculated from E HOMO = −4.4 + E ox onset . The corresponding E LUMO was calculated by adding E g opt to the E HOMO and summarized in Table 1. To unravel the energy level and geometrical structure of pyridine-based HSLs, we performed density functional theory (DFT) studies at the B3LYP/6-31G(d,p) ( Figure S3). The E LUMO and E HOMO of PyDAn spread over the pyridine core and diphenylamine arms, and the E HOMO of PyPDAn exhibits similarities in the electron densities with PyDAn, while the E LUMO of PyPDAn is distributed on the central pyridine and benzene bridge. The E HOMO of PyDAnCBZ was mainly distributed on DAnCBZ arms, and E LUMO was mainly distributed on the pyridine core. The E LUMO and E HOMO were distributed on the pyridine core and PTPDAn arms in the PyPTPDAn.
To study the thermal properties of pyridine-based materials, differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) were utilized. TGA curves ( Figure S4), the decomposition temperatures (T d ) corresponding to 5% with loss temperature, were 307.5, 372.5, 242, and 196.6°C for PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn, respectively. It can be deduced that the T d of pyridine-core based materials decreases as the arm size increases. The full and second heating run of DSC are presented ( Figure S5 and Figure 1d). The PyDAn displays an evident glass-transition temperature (T g ) at 81.0°C and melting of the crystals (T m ) at 165.4°C, whereas PyPDAn displayed an apparent T g at 82.0°C. The PyDAnCBZ and PyPTPDAn did not have an obvious T g , showing their amorphous nature. The larger arms can enhance T g , and PyDAnCBZ (PyPTPDAn) has improved its thermal stability and is beneficial for success in the real operating conditions of solar cells.
The electrical measurements were carried out to deduce the impacts of the arm on the HSLs ability for charge transport. The conductivity of the HSLs were determined in a perpendicular configuration using the structure FTO/HSL/ Au, and the J−V curves were measured in the dark.  Table 1. The pristine HSLs exhibited hole mobilities of 0.35, 1.85, 3.50, and 1.14 (× 10 −4 cm 2 V −1 s −1 ). The highest charge transporting value of PyDAnCBZ as compared to others derives from the improved intramolecular charge transfer and uniform morphology, subsequently discussed.
Electron paramagnetic resonance (EPR) spectroscopy is an advantageous technique to identify molecules with unpaired

ACS Applied Energy Materials
www.acsaem.org Article electrons and to evaluate the relative number of spins and their mobility. For this purpose, X-band EPR experiments on pristine and doped 25 mM toluene solutions of the four pyridine-based HSLs were performed at room temperature ( Figure 2). The EPR signal of the undoped samples was difficult to detect due to the number of spins inside the resonant cavity being practically below the detection limit of the spectrometer (about 10 9 ). In contrast, a dramatic enhancement of the signal was observed after the addition of lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) and 4tert butylpyridine (t-BP), confirming a remarkable number of holes in these materials after LiTFSI-doping (≈10 17 spin/mol). The structure of the spectra, three equally spaced lines with intensity ratios close to 1:1:1, indicates that the unpaired electrons are preferentially located on 14 N nuclei (I = 1), but the generation of other radicals with very short lifetime (<10 −6 s) cannot be disregarded. The absence of proton hyperfine structure is attributed to the presence of many protons with small hyperfine couplings. The spectra could be fitted to obtain the g value, the coupling parameter a N , and the peak-to-peak linewidth (Table S1). The calculated g values (2.0044, 2.0040, 2.0037, and 2.0043 for PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn, respectively) are in good agreement with those expected for N-centered cation radicals with the low orbital contribution to the ground state. The hyperfine coupling constant (a N ) and the peak-to-peak linewidth are relatively larger for PyDAnCBZ (6.8 and 4.8 Gauss, respectively) than for the other compounds, suggesting a higher spin density near the pyridine core. Before the fabrication of the device, the surface properties and film-formation capabilities of HSLs were investigated. The Cs 2 AgBiBr 6 film was prepared using the hot-casting method, and as a result of the rapid crystallization, the as-prepared films ( Figure S6) showed micrometer-sized agglomerates on the surface measured by scanning electron microscopy (SEM), which is in accordance with the results reported by Shao et al. and Bein et al. 14,39 The images of perovskite with HSLs atop the perovskite were presented ( Figure S7). The perovskite/ PyDAn exhibited subpar surface morphologies, whereas others show uniform coverage. To evaluate the morphology properties of HSLs, the HSLs were deposited on flat substrates. It showed that the PyDAnCBZ film had a remarkably uniform, smooth, and nanoparticle-free surface. In contrast, PyDAn and PyPTPDAn films display nonuniform coverage, whereas PyPDAn displayed aggregation and undissolved dopants. Room-temperature EPR spectra registered on 25 mM toluene solutions at the X-band. The dotted lines represent the best fits obtained using the spectral parameters listed in Table S1.

Device Performance.
We fabricated PSCs in the n-i-p structure using FTO/bl&mp-TiO 2 /Cs 2 AgBiBr 6 /HSL/Au to assess how the arm modulation of pyridine based-HSLs affected the PV performance. The cross-sectional SEM image of the PSC ( Figure S8) reveals distinct layered topologies, with perovskite and PyDAnCBZ having thicknesses of 232 and 180 nm, respectively. The mesoporous TiO 2 ESL with 216 nm was chosen to improve the thickness of Cs 2 AgBiBr 6 for absorbing more light. The same precursor weight concentration was used to uniform the HSLs. The J−V curves of the optimized device measured under standard AM 1.5G irradiation are presented in Figure 3a and summarized in Table 2.
The PSCs based on PyDAnCBZ attained a record PCE of 2.92% measured under reverse scan (V oc = 1059.0 mV, J sc = 3.73 mA cm −2 , and FF = 74.0%), whereas lower PCEs were achieved by the devices with PyPDAn under similar conditions. The PyPDAn-based PSCs delivered a PCE of 1.90% (V oc = 1053.2 mV, J sc = 2.76 mA cm −2 , and FF = 65.4%). The PyDAn-and PyPTPDAn-based PSCs yielded ∼1% efficiency. The performance of the devices based on the new HSLs was in the following order: PyDAnCBZ > PyPDAn > PyDAn > PyPTPDAn. The PCE of PSCs employing PyDAnCBZ is higher than that of Spiro-OMeTAD and PTAA, which measured 1.6 and 1.31%, respectively ( Figure S9 and Table S2 in the Supporting Information), pointing its effectiveness and suitability as a HSLs.
The reproducibility of the PSCs with the PyDAnCBZ was analyzed, and the PV performance is presented (Table S3) with an average PCE of ∼2.7%. Our findings demonstrate that PyDAnCBZ and PyPDAn as HSL achieved a higher V oc and FF than those of PyDAn and PyPTPDAn. The PyDAnCBZbased device showed the highest performance owing to the higher FF and J sc , which stems from the improved electrical conductivity and hole mobility and inhibited surface charge carrier recombination. The J−V curves of PSCs measured under forward and reverse scans ( Figure S10 and Table S4), and the calculated HI values for PSCs with PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn is 0.23, −0.07, 0.09, and −0.18, respectively. The PyPDAn-and PyDAnCBZ-based PSCs showed a reduction in hysteresis in comparison with the others, which is consistent with the device's PV performance, and we attribute to the improved charge-transporting abilities.
The external quantum efficiency (EQE) curves displayed that the PSCs with PyDAnCBZ have a higher EQE value between 300−500 nm than that of the others, with higher values of 69.8 and 72.4% at 365 and 440 nm were observed respectively (Figure 3b). Additionally, the PyDAnCBZ-based device displayed a higher EQE value across a wider wavelength range (700−1100 nm) as shown in Figure S11. It produced an enhanced integrated current density (J int ) of 3.48 mA cm −2 (Figure 3c), which is higher than the previously reported Cs 2 AgBiBr 6 -based devices. 40,41 Furthermore, only Cs 2 AgBiBr 6 is used as an absorber layer, no dye or other coabsorber is used. The outcomes are in line with those of the PyDAnCBZ with the highest J sc (3.73 mA cm −2 ). The maximum power point tracking for champion PyDAnCBZ HSL under 1 Sun condition and electrical bias was conducted. The stabilized current density and PCE values (Figure 3d) of the device were 3.60 mA cm −2 and 2.79%, respectively, which is in agreement with the PV performance.
We conducted the device stability to access the reliability of lead-free solar cells. The PCEs of PyDAnCBZ-based devices still had around 90% of their initial value after aging at the dry box with 10−30% relative humidity and room temperature for a thousand hours (Figure 3e). The V oc and J sc maintained the initial value, while the FF decreased to 90% of the initial value and was the cause of the slight deterioration of device performance. Suggesting PyDAnCBZ and Cs 2 AgBiBr 6 improved stability under dry air, with very little deterioration coming from the interfacial layers within the device.
The intrinsic ionic nature of double metal perovskite is prone to ion migration and impairs PV performance. 42,43 Therefore, we assessed how the PV performed when subjected to multistress with thermal (85°C) and ambient moisture conditions. The PCE still had 92 percent of its original value after aging around 150 h, as shown in Figure 3f. We attribute the improved reliability owing to the interaction between the pyridine core unit and Ag defects, as well as the improvement of the diffusion barrier that prevents the Ag and Br ions from migration. Importantly, the PyDAnCBZ possessed amorphous forms and could maintain their transporting properties under thermal aging. The phenomena constrained by PyDAnCBZ can be further quantified.
To elucidate the charge transport kinetics properties of the lead-free PSCs, complementary electrical characterizations were performed. Figure S12 represents the J−V curve of the PSCs measured under dark conditions, and the diode parameters extracted are represented in Table 2. The series resistance (R s ) is high for all PSCs, since typical lead-based PSCs present values of R s an order of magnitude of 1 Ω. Nevertheless, this value may be due to the resistive TiO 2 , having the HSL a low impact. The shunt resistance (R sh ) value is good in all the PSCs, indicating that there are no shunt sources between the electrodes, such as pinholes or conductive phases. The dark saturation current (J 0 ) can be associated with the carrier recombination in the cells, which correlates with the open circuit voltage. 7 Compared to the other PSCs, the PyDAnCBZ-based cell shows a lower dark saturation current, correlating with the V oc values measured in Figure 3a. Suggesting PyDAnCBZ employment largely mitigates the recombination, most likely in the back surface, while the other pyridine HSLs present larger surface recombination. The diode ideality factor close to 2 is a typical value for thin-film PV, which indicates that the carrier recombination occurs in the depletion area. Larger values of the ideality factor may indicate that other sources of recombination can affect the PV parameters, degrading the FF. PyDAnCBZ-based devices show a lower ideality factor, indicating the HSL ability to control the recombination process.
To deduce a comprehensive understanding of the enhanced PV performance, the interaction between Cs 2 AgBiBr 6 and PyDAnCBZ was investigated through a series of experiments. First, we conducted X-ray diffraction (XRD) characterization to analyze the crystal structures of perovskite films with and without PyDAnCBZ ( Figure S13). Both the films exhibited high-purity Cs 2 AgBiBr 6 and identical crystallinity, indicating that the spin-coating of PyDAnCBZ on top of the perovskite films had no impact on the perovskite crystal structure.
Further analysis was conducted using core-level X-ray photoelectron spectroscopy (XPS) to investigate the chemical states of Ag, Bi, Br, and Cs in both the control perovskite film and PyDAnCBZ-treated film (Figure 4a and Figure S14). The PyDAnCBZ-treated film showed higher binding energies for Ag 3d, attributed to the strong binding between surface silver ions and PyDAnCBZ. However, the characteristic peaks of Bi 4f and Br 3d showed a negligible shift after the PyDAnCBZ deposition. The finding suggests that the strong binding between Ag + and PyDAnCBZ can effectively passivate surface Ag, Ag i , and Ag Bi -defects that may have arisen from ion migration during the formation of the perovskite grains. 42,44,45 To quantitatively verify the passivation effect of PyDAnCBZ on defects, we utilized the space-charge-limited-current (SCLC) method to measure the hole-only device for perovskite films with and without PyDAnCBZ. The corresponding current−voltage curves are depicted in Figure 4b. The trap density (N t ) was derived from the equation V TFL = eN t L 2 /2εε 0 , where V TFL is the trap-filling voltage, L represents the perovskite film thickness, and ε 0 and ε are the vacuum permittivity and relative dielectric constant of Cs 2 AgBiBr 6 , respectively. The trap densities are calculated and summarized in Figure 4c. After the introduction of PyDAnCBZ, the trap densities decreased from 3.56 × 10 14 to 1.51 × 10 14 cm −3 . This reduction in trap densities further suggests that the incorporation of PyDAnCBZ can efficiently passivate defects at the surface and grain boundaries, as well as inhibit nonradiative recombination.
Additionally, we performed capacitance−voltage measurements and derived Mott-Schottky plots (Figure 5a−d). In a solar cell with a diode behavior, the region of the negative slope of a Mott-Schottky plot indicates the reduction of the depletion region with the increase of the voltage, from where the built-in potential can be calculated as in the case of the PyDAnCBZ. However, in the other solar cells studied, a positive slope is also found, which suggests a barrier for the charge carrier has emerged, due to their poor charge carrier ability. This barrier may explain the increase of recombination found in the dark J−V curves for these cells. Since this barrier  does not allow the extraction of more parameters in these cells with reliability, the rest of the measurements were made only in the PyDAnCBZ based solar cell. From the capacitance− voltage measurements, the carrier density and depletion width were calculated (Figure 5e), revealing the doping density of the Cs 2 AgBiBr 6 is 2.46 × 10 17 cm −3 , while the depletion at 0 V is 107 nm. This is found to be the first bottleneck for the performance of Cs 2 AgBiBr 6 as a light-absorbing layer, since the charge collection is limited, reducing the photocurrent that can be extracted. The measurement of the impedance in the vicinity of the turn-on voltage ( Figure S15) and we calculated the transfer time from this (Figure 5f). This parameter represents the time for a charge to recombine once is in the depletion region. A value in the order of 1 μs in the region of the maximum power point voltage is high enough to consider that losses by recombination in the junction are negligible.
To identify the barrier pushing the Cs 2 AgBiBr 6 solar cell performance, we simulated it using PC1D software, combining the results obtained from the characterization with the properties of Cs 2 AgBiBr 6 . 14, 46 The energy levels of the solar cell, including the FTO, TiO 2 , Cs 2 AgBiBr 6 , and HSL, are shown in Figure 6a. It can be observed that the depletion region of the solar cell simulated is in the order of 100 nm, in agreement with the value calculated by the capacitance− voltage, due to the high doping density of Cs 2 AgBiBr 6 . When simulating the device performance, it can be noted that the baseline simulation of the J−V curve and EQE calculated is close to the spectra of the PyDAnCBZ solar cell, as well as the PV characteristics ( Table 3). Tuning of some of the key properties identified as bottlenecks was performed. First, a reduction of the doping density to a value of 10 16 cm −3 allows for a higher collection of charge, increasing the current from 3.55 to 4.92 mA cm −2 , and the PCE by 1%. A decrease in the doping density could be possible by controlling the defects and crystallinity of the Cs 2 AgBiBr 6 films. Then, the effect of the excessive series resistance was analyzed. A reduction of an order of magnitude has a large impact on the FF, increasing from 69 to 77%. Cs 2 AgBiBr 6 -based PSCs are reported to have very low electron mobility, which impacts the charge collection and avoids the possibility of increasing the thickness of the Cs 2 AgBiBr 6 layer. 47 Improving the electron mobility by reducing the electron traps in the perovskite layer has a small impact in thin cells, but may allow for an increase in the light-absorbing layer thickness without affecting other parameters. The effect of increasing the electron mobility and thickness was simulated, founding a significant increase in the photocurrent (Figure 6b,c). Including all the modifications mentioned, the device based on Cs 2 AgBiBr 6 with the configuration used in this work has the potential of reaching a PCE of a 4.9%.
3.1.3. Synthesis of PyPTPDAn. 2,6-Bromopyridine (100 mg, 0.42 mmol), PTPDAn (0.86 g, 1.05 mmol), sodium tert-butoxide (366.9 mg, 3.822 mmol), and dry toluene (40 mL) were mixed in a flask and purged with N 2 for 10 min, and P( t Bu) 3 HBF 4 (9.23 mg, 0.05 mmol) and Pd(OAc) 2 (5.72 mg, 0.04 mmol) were added. The reaction was refluxed under N 2 for 48 h. After completion of the reaction, monitored with TLC, the reaction was allowed to cool to room temperature, and the solution was filtered to remove insoluble solids.  The filtrate was concentrated in vacuo and purified by column chromatography (silica gel, hexane: ethyl acetate 8:4 (v/v) as eluent) to give PyPTPDAn as a greenish-yellow solid with a yield of 52%. All of the characterization for the novel HSL including 1 H/ 13 C NMR spectroscopy and mass spectra are shown in the Supporting Information. The synthesis route of PyDAnCBZ has been reported by the reported literature from our group. 34

Device Fabrication.
The configuration of the PSCs is FTO/ b&mp-TiO 2 /Cs 2 AgBiBr 6 /HSL/Au. The FTO glass substrates (NSG10) were first cleaned with 2% Hellmanex solution, acetone, and isopropanol. The precleaned substrates treated with UV-ozone were heated to 500°C for over 30 min. The blocking TiO 2 was sprayed with the solution [1 mL of titanium(IV) diisopropoxide bis(acetylacetonate) was dissolved in 19 mL of ethanol]. These substrates were kept heating for 30 min after spray deposition. The mesoporous TiO 2 film was spin-coated at 4000 rpm for 30 s using the 0.13 g/mL 30NRD ethanol solution. The as-prepared mp-TiO 2 was postheated at 125°C for 15 min and 500°C for 30 min. Then, the mp-TiO 2 layer was impregnated in an aqueous TiCl 4 solution (67.5 μL of TiCl 4 in 10 mL of deionized water) at 70°C for 1 h, followed by sintered at 500°C for 30 min again.
The Cs 2 AgBiBr 6 precursor solution was prepared by dissolving 2 mmol CsBr, 1 mmol AgBr, and 1 mmol BiBr 3 in dimethyl sulfoxide (DMSO), which was stirred at 100°C until these salts were fully dissolved in DMSO. Both the FTO/b&mp-TiO 2 substrate and precursor solution were placed on the hot plate at 80°C, and the solution was then spin-coated on the aforementioned substrate at 3000 rpm for 30 s, which was subsequently annealed at 280°C for 5 min in the glovebox with argon. The ∼40 mg/mL PyDAn, PyPDAn, PyDAnCBZ, and PyPTPDAn with 75, 60, 30, and 25 mM were dissolved in chlorobenzene, respectively, the dopants were then added, and their molar ratios of LiTFSI and t-BP were fixed to 0.5 and 3.3 to the materials (Li salt solution is 520 mg mL −1 in acetonitrile). The precursor HSL solutions were then spin-coated atop the perovskite film at 3000 rpm for 30 s. Finally, a 70 nm thick gold electrode was thermally deposited.
3.3. Characterization. 1 H NMR spectra were recorded using a 400 MHz spectrometer. Chemical shifts are reported in delta expressed in parts per million (ppm) downfield from tetramethylsilane (TMS) using residual protonated solvent as an internal standard {CDCl3, 7.26 ppm}. 13 C NMR spectra were recorded using a 100 MHz spectrometer. Chemical shifts are reported in delta (d) units, expressed in parts per million (ppm) downfield from tetramethylsilane (TMS) using the solvent as an internal standard {CDCl 3 , 77.16 ppm}. The 1 H NMR splitting patterns have been described as "s, singlet; d, doublet; t, triplet; and m, multiplet". HRMS were recorded on a Bruker-Daltonics micrOTOF-Q II mass spectrometer. UV− visible absorption spectra of all compounds were recorded on a PerkinElmer Lamba 35 UV−visible spectrophotometer in dichloromethane solution. Cyclic voltammograms were recorded on a CHI620D electrochemical analyzer (potentiostat) in dichloromethane solvent using glassy carbon as the working electrode, Pt wire as the counter electrode, and saturated calomel electrode (SCE) as the reference electrode. The scan rate was 100 mV s −1 for cyclic voltammetry. A solution of tetrabutylammonium hexafluorophosphate (TBAPF 6 ) in dichloromethane (0.1 M) was used as the supporting electrolyte.
Thermogravimetric analysis (TGA) was carried out using Mettler Toledo TGA/DSC 1 at a heating rate of 5°C min −1 under a nitrogen atmosphere. The glass transition temperature for the pyridine-based HSLs was recorded by different scanning calorimetry (Pekin Elmer Diamond DSC) with a scan rate of 10 min −1 under N 2 gas flow. The second curve was analyzed. The SEM images were recorded by Hitachi S-4800.
EPR spectra were performed at room temperature using a Bruker ELEXSYS E500 spectrometer operating at the X-band. The spectrometer was equipped with a superhigh-Q resonator ER-4123-SHQ. Toluene solutions of undoped and doped samples were placed in quartz tubes, and spectra were recorded using typical modulation amplitudes of 1.0 G at a frequency of 100 kHz. The magnetic field was calibrated by an NMR probe, and the frequency inside the cavity (∼9.4 GHz) was determined with an integrated MW-frequency counter. The radical concentration was evaluated by integrating the EPR spectrum twice and by comparing it with a standard CuSO 4 · 5H 2 O sample. Data were collected and processed using the Bruker Xepr suite.
The J−V curves of the solar cells with an active area of 0.09 cm 2 controlled by a metal mask were measured under AM 1.5G illumination that was provided by a 3A grade solar simulator (Newport). The forward scan range is 0−1.3 V, and the reverse scan range is from 1.3 to 0 V (scan rate: 10 mV s −1 ). The EQE was measured by a 150 W Xenon lamp (Newport) attached to IQE200B (Oriel) motorized 1/4m monochromator as the light source. The impedance of devices was tested by Bio-logic SP-300 under different voltages in different frequencies. The devices with structures FTO/ HSL/Au and FTO/PEDOT:PSS/HSL/Au were fabricated and measured under dark conditions for calculating conductivity and mobility.
Impedance and capacitance−voltage measurements were performed inside a Faraday box with a Biologic impedance analyzer, following the reported literature, 48,49 by applying a 20 mV perturbation in a frequency range from 1 GHz to 100 Hz, and applied constant voltage from −0.5 to 1.5 V. EC-lab software was used to fit the equivalent circuits.

CONCLUSIONS
We developed and assessed four pyridine-based small molecules as a p-type selective material, investigated how the arm size affected the electro-optical properties, and deduced that arm modulation had an impact. Lead-free solar cells made with PyDAnCBZ outperformed the others. In comparison to the other compounds, PyDAnCBZ has a hyperfine coupling constant, and peak-to-peak linewidth is substantially larger, which suggests a higher spin density close to the pyridine core. Electrical measurements suggest that PyDAnCBZ facilitates improved charge carrier transport. The solar cells with PyDAnCBZ as a hole selective layer also displayed excellent thermal and moisture stability. Further, we suggested the roadmap through the photovoltaic performance simulation that, with fine-tuning, Cs 2 AgBiBr 6 solar cells can increase their power conversion efficiency to 4.9%. ■ ASSOCIATED CONTENT
Materials synthesis its spectroscopic and structural characterization, and the device electrical properties (PDF)