cPCN-Regulated SnO2 Composites Enables Perovskite Solar Cell with Efficiency Beyond 23%

The (SnO2-cPCN) ETL shows superior electron mobility of 3.3 × 10−3 cm2 V−1 s−1, which is about three times higher than that of pristine SnO2. The less wettable SnO2-cPCN leads to perovskite layers with reduced grain boundaries and enhanced qualities due to suppressed heterogeneous nucleation of perovskite. The PSCs based on SnO2-cPCN showed negligible J–V hysteresis and two champion PCE of 23.17% and 20.3% on devices with 0.1 and 1 cm2 active area, respectively. The (SnO2-cPCN) ETL shows superior electron mobility of 3.3 × 10−3 cm2 V−1 s−1, which is about three times higher than that of pristine SnO2. The less wettable SnO2-cPCN leads to perovskite layers with reduced grain boundaries and enhanced qualities due to suppressed heterogeneous nucleation of perovskite. The PSCs based on SnO2-cPCN showed negligible J–V hysteresis and two champion PCE of 23.17% and 20.3% on devices with 0.1 and 1 cm2 active area, respectively. Efficient electron transport layers (ETLs) not only play a crucial role in promoting carrier separation and electron extraction in perovskite solar cells (PSCs) but also significantly affect the process of nucleation and growth of the perovskite layer. Herein, crystalline polymeric carbon nitrides (cPCN) are introduced to regulate the electronic properties of SnO2 nanocrystals, resulting in cPCN-composited SnO2 (SnO2-cPCN) ETLs with enhanced charge transport and perovskite layers with decreased grain boundaries. Firstly, SnO2-cPCN ETLs show three times higher electron mobility than pristine SnO2 while offering better energy level alignment with the perovskite layer. The SnO2-cPCN ETLs with decreased wettability endow the perovskite films with higher crystallinity by retarding the crystallization rate. In the end, the power conversion efficiency (PCE) of planar PSCs can be boosted to 23.17% with negligible hysteresis and a steady-state efficiency output of 21.98%, which is one of the highest PCEs for PSCs with modified SnO2 ETLs. SnO2-cPCN based devices also showed higher stability than pristine SnO2, maintaining 88% of the initial PCE after 2000 h of storage in the ambient environment (with controlled RH of 30% ± 5%) without encapsulation.


Introduction
Organic-inorganic hybrid perovskite-based solar cells (PSCs) are considered the paradigm shift of traditional crystalline silicon photovoltaics due to their fascinating optoelectronic properties and easy solution processability [1][2][3]. Since the first report in 2009 [4,5], the certified power conversion efficiency (PCE) of PSCs has exceeded 25%, encouraging the research community to explore further the general principles behind the material design, film formation, device structure, and operational mechanism [6][7][8][9][10]. In the early stage of PSC studies, mesoscopic titanium oxide (m-TiO 2 ) was frequently used as the electron transport layer (ETL), facilitating the certified world record PCEs [11]. However, the high temperatures (> 500 °C) sintering process damaged the solution processability and limited the application of PSCs in flexible and tandem devices [12]. Besides, the m-TiO 2 presents high photocatalytic activity under the illumination of ultraviolet light, undermining the long-term stability of PSCs [13]. In this regard, planar-type PSCs become potential alternatives to mesoporous types and growing high-quality ETL at low temperatures (≤ 150 °C) has been emphasized to obtain high-efficiency planar-type devices. Among all the candidate ETLs, low-temperature-processed SnO 2 has demonstrated advantages such as high optical transmittance, congruous energy levels, robust chemical and UV stability, and balanced electron mobility (10 −4 cm 2 V −1 s −1 ) [14][15][16], enabling several scintillating PSCs with recordbreaking performances simultaneously [17,18].
However, charge accumulation may happen at the SnO 2 / perovskite interface resulting in severe hysteresis due to much lower electron mobility of SnO 2 ETL than perovskite absorbers (0.5-30 cm 2 V −1 s −1 ) [19][20][21][22]. To alleviate the hysteresis effect, dopants or additives such as metal cations and polymers were added to enhance the conductivity of SnO 2 and hence the PCE of devices. Ren et al. investigated the effect of Nb-doping in SnO 2 and decreased hysteresis due to the enhancement of electron mobility from 1.02 × 10 −4 to 2.16 × 10 −4 cm 2 V −1 s −1 [23]. Wei et al. prepared a polymer-incorporated SnO 2 colloidal ink to ameliorate the compactness and wetting property of the SnO 2 layer, and suppressed hysteresis emerged due to better coverage of perovskite film on the SnO 2 -in-polymer matrix [24]. The results revealed that the underlayer's surface properties have a crucial influence on the morphology and quality of the perovskite films above.
Shreds of evidence have indicated that these grain boundaries might cause charge recombination ascribed to the presence of charge-trapping recombination centers, and tremendous efforts have been devoted to reducing the grain boundaries of perovskite films. Liu and Huang's groups reported, respectively, that the incorporation of ethylene-diamine-tetra-aceticacid (EDTA) or red-carbon quantum dots (RCQs) into SnO 2 can increase the wettability at ETL/perovskite interface and generate high-quality perovskite films with enlarged grain size and reduced grain boundaries [15,19]. In contrast, Huang et al. demonstrated that the non-wetting polymeric HTL surface could also lead to perovskite layers with large grains [25]. Given the discussions above, the surface conditions of the underlayer should be well controlled to generate high-quality and ideally defect-free (less) perovskite films. The influence of the wettability of the underlayer on the perovskite films needs to be further explored.
Targeting the conductivity of SnO 2 and its influence on the quality of the perovskite layer, the presented work realized crystalline polymeric carbon nitride (cPCN)-composited SnO 2 (SnO 2 -cPCN) ETL with superior electron mobility of 3.3 × 10 −3 cm 2 V −1 s −1 , which is more than three times higher than that of pristine SnO 2 . The PSCs based on SnO 2 -cPCN exhibited negligible current density-voltage (J-V) hysteresis due to the decreased charge accumulation at the perovskite/ ETL interface under the increased electron mobility. Besides, the SnO 2 -cPCN surface became smoother and less wettable. On this basis, the perovskite absorber layers with reduced grain boundaries and enhanced qualities were realized due to suppressed heterogeneous nucleation of perovskite. Incorporating cPCN into SnO 2 not only enhanced the electron mobility of ETL via effectively filling the electron trap states but also affected the growth of perovskite grains, reducing non-radiative recombination. Finally, planar PSCs based on SnO 2 -cPCN presented a champion PCE of 23.17% on devices with a small active area (0.1 cm 2 ) and a promising PCE of 20.3% on devices with a large active area (1 cm 2 ).

Preparations of the g-CN, cPCN
The g-CN powder was synthesized by thermal condensation of the urea. Typically, melamine (8 g) was heated to 500 ºC for 4 h at a rate of 12 °C min −1 in a muffle furnace in an air atmosphere. The crystalline PCN was synthesized by an ionothermal approach. In detail, 1.2 g of melamine was mixed with KCl (6.6 g) and LiCl (5.4 g) in a glovebox, and the mixture was then heated to 550 ºC under an N 2 atmosphere (2 L min −1 ) in a muffle furnace.

Preparations of the SnO 2 -cPCN Precursor
The synthesized cPCN NCs powder was first dispersed in deionized water, ultrasonicated for 10 h, and then filtered with a 0.45 mm filter to obtain cPCN NCs water solution of different concentrations. Then, the SnO 2 colloid precursor (15 wt.%) was diluted with deionized water to the concentration of 2.5 wt.% and was stirred at room temperature for 2 h. The cPCN NCs and SnO 2 solutions were mixed with a volume ratio of 1:1 ratio and then ultrasonicated for 1 h.

Device Fabrication
Chemically etched FTO glass substrates were cleaned with a detergent solution, deionized water, acetone, and anhydrous ethanol for 15 min, respectively. Next, the substrates were further cleaned with plasma treatment for 15 min. The SnO 2 solution (Alfa) was spin-coated on the FTO substrates at 3000 rpm for 30 s, followed by annealing at 150 °C for 30 min. The SnO 2 -cPCN solution also underwent the same procedures. The substrate was then cool down to room temperature on a spin coater. The (FAPbI 3 ) 0.9 (MAPbBr 3 ) 0.1 perovskite solution (PbI 2 , FAI, PbBr 2 , MACl, MABr, in DMF: DMSO = 9:1 volume ratio) was spin-coated at 1000 rpm for 10 s and 5000 rpm for 30 s onto the FTO/ SnO 2 substrate. 200 μL of chlorobenzene was dropped on the spinning substrate at 8 s before the program finish, and the FTO/SnO 2 /perovskite sample was heat-treated at 150 °C for 15 min. Then, the hole transporting layer was deposited on top of the perovskite layer at a spin rate of 4000 rpm for 20 s using a spiro-OMeTAD solution. For the spiro-OMeTAD solution, 72.3 mg of spiro-OMeTAD was dissolved in 1 mL of chlorobenzene with additives of 17.5 μL of bis(trifluoromethylsulfonyl)imide lithium salt (Li-TFSI, Sigma-Aldrich) solution (520 mg mL −1 in acetonitrile), 28.8 μL of 4-tert-butylpyridine (TBP, Sigma-Aldrich). Finally, 120 nm of the silver counter electrode was thermally evaporated under a high vacuum.

Characterization
SEM measurements were performed using a SUPRA 55, Zeiss, Germany, operated at an acceleration voltage of 5 kV. The XRD patterns were measured using a SmartLab X-ray powder diffractometer with an angle range of 2θ = 3° to 60°. UV-vis absorption spectra were recorded on a spectrophotometer (Agilent Cary 5000) in the 350-850 nm wavelength range at room temperature. The steady-state PL spectra were obtained using a fluorescence spectrophotometer (FLS980, Edinburgh Instruments). Current density-voltage (J-V) characteristics were measured using a source meter (Keithley 2400) under 100 mW cm −2 simulated AM 1.5 G irradiation with a solar simulator (Enli Tech, Taiwan) by reverse (1.2 to -0.1 V) scans or forward scan (from −0.1 to 1.2 V) modes at a scan speed of 200 mV s −1 . The hysteresis indices (H hysteresis ) of the devices are calculated based on Eq. 1.
The active area of devices was defined by a metal shadow mask of 0.1 or 1 cm 2 . The dark current-voltage curves were recorded with an electrochemical workstation (Zennium Zahner, Germany). In detail, the measurement works as follows: For the first edge potential of -1.0 V, the second edge potential of 1.5 V was applied. The EQE was characterized by the QE-R systems (Enli Tech.), and the measurement scope was 300-900 nm. EIS measurements were carried out in the dark at 0.5 V applied voltage using an electrochemical workstation (Zennium Zahner, Germany) with an AC perturbation of 10 mV ranging from 100 to 1 MHz at room temperature with 60% humidity (Fig. S6). FTIR spectra were recorded on a Nicolet iS 50 Spectrometer. The roughness of the films was recorded using atomic force microscopy (AFM, Multimode-8J, America). The water contact angle was measured at a Data physics OCA-20 drop shape analyzer.
To gain insights into the charge transport, we measured electron mobility using different ETLs in the same device structure. Specifically, the electron-only device was designed and fabricated using FTO/ETL/PCBM/Ag structure, as shown in the inset in Fig. 2b (later). In this analysis, we assumed that the current is only related to electrons. When the effects of diffusion and the electric field are neglected, the current density can be determined by the SCLC [26]. The different ETLs were spin-coated on FTO. Then, 120-nmthick Ag was deposited on FTO/ETL/PCBM samples. The dark J-V curves of the devices were performed on a Keithley 2400 source at ambient conditions. The electron mobility (μ e ) is extracted by fitting the J-V curves using the Mott-Gurney law (Eq. 2): where J is the current density, L the thickness of different ETLs, ε 0 the vacuum permittivity, ε r the dielectric permittivity of various ETLs, V app the applied voltage, V r the voltage loss due to radiative recombination, and V bi the built-in voltage owing to the different work function between the anode and cathode.

Crystalline Polymeric Carbon Nitride Characterization
Carbon derivatives, such as carbon nanotubes, fullerene, and graphene (oxide), have been widely used as additives or interlayers in PSCs [27,28]. Particularly, traditional graphitic carbon nitride (g-CN) was reported to facilitate highefficiency PSCs. For example, Jiang et al. added g-CN in precursor solutions to improve the quality and conductivity of perovskite films [29]. Similarly, Li et al. developed a series of functionalized g-CN (SO 3 -, OH-, NH 3 -, or NO 3 -C 3 N 4 ) to modify the perovskite precursor solution and achieved the best PCE of 20.08% with NO 3 -C 3 N 4 [30]. Recently, Chen et al. used hybrid quantum dots of SnO 2 /g-CN was used as the ETL in PSCs and demonstrated that the g-CN could reform the electronic density distribution around the neighboring SnO 2 crystal unit to effectively eliminate the oxygenvacancy type trap centers and promote the electron transport [31]. Compared with the widely used g-CN, the cPCN is the highly crystalline counterpart with fully conjugated π-electron systems and higher conductivity [32,33], which may further enhance the functionality of SnO 2 based ETLs. It is worth noting that traditional poly-heptazines are often (and misleadingly) classed as 'graphitic carbon nitride,' 'graphitic CN' or 'g-C 3 N 4 ′ in literature, including our own reports [33]. To that end, we have proposed to correct the term of 'gCN' to 'PCN' in one of our review papers [32].
The preparation protocol of the g-CN and cPCN semiconductor is described in the experimental section. In this study, preheated melamine was used as a precursor in the presence of KCl/LiCl to synthesize cPCN with tri-s-triazine subunits. The high crystallinity nature of cPCN was verified by X-ray diffraction (XRD), as illustrated in Fig. 1a. The cPCN exhibited a narrower and shifted diffraction peak at 28.3° with a full width at half-maximum (FWHM) compared to traditional g-CN (27.4°), indicating the well-developed and condensed crystal structure with enhanced interaction between layers [33]. The peak at 13.0° corresponding to the inter-plane distance of 0.618 nm for g-CN is shifted to 8.0° with a repeat unit of 1.099 nm due to an unfolded inplane network associated with sufficient condensation of the conjugated framework ( Fig. S1) [33]. The FTIR spectra in Fig. 1b further exhibit the structure information of cPCN and g-CN. The broad peaks between 3500 and 3000 cm −1 stem from the terminal amino groups, while the peak at 2150 cm −1 originates from terminal cyano groups (C≡N) owing to the loss of ammonia on the surface of traditional g-CN [34]. The set of peaks between 1700 and 900 cm −1 are characteristic signals from tri-s-triazine derivatives [35].
The as-prepared cPCN sample exhibits porous structures under scanning electron microscopy (SEM) (Fig. 1c). The transmission electron microscopy (TEM) image presents stacked layers of nanosheets (Fig. 1d). The high-resolution TEM images of cPCN are illustrated in Fig. 1e, f, revealing a clear hexagonal lattice structure with two lattice fringes. The lattice fringe of 0.33 nm may be assigned to the interlayer distance, while the 0.98 nm lattice fringe is likely originated from the in-plane periodicity. Furthermore, more information and discussions about the difference between PCN and cPCN, including morphology, structure, optical properties, electrical property, and stability, can be found in one of our works [33].

Fabrication and Characterization of SnO 2 -cPCN Film
We plotted XRD patterns of SnO 2 and SnO 2 -cPCN based films to investigate the structural properties and phase composition (Fig. S2). These peaks match well with characteristic diffractions of SnO 2 (JCPDS No. 01-077-0452). Except the signals from the underlying FTO, no obvious peak relating to cPCN, SnO 2 , or any other impurity is detected. It is important to note that due to the very thin nature of the deposited layers (ca. 30 nm), minority phases may be challenging to be identified. Based on the previous studies, the novel cPCN with higher crystallinity and conductivity may have the potential to improve the performance of SnO 2 ETL. Specifically, X-ray photoelectron spectra (XPS) of SnO 2 and SnO 2 -cPCN films deposited on quartz substrates are conducted (Figs. 2a and S3) to explore the interactions between cPCN and SnO 2 . We calibrated the binding energy scale for all XPS measurements to the carbon 1 s line at 284.8 eV. After the cPCN incorporation, the binding energies of Sn 3d 3/2 and Sn 3d 5/2 at 486.34 and 494.77 eV shift slightly to higher values at 486.52 and 494.87 eV, respectively. The blue shift (toward higher binding energy) of Sn 3d signals indicates the electron transfer and may promote electron mobility [36][37][38][39]. Besides, the SnO 2 -cPCN film presents two additional peaks at 400.08 and 404.16 eV originating from the N1s of cPCN (Fig. S3a, b). Consequently, the cPCN is successfully composited and chemically bonded with SnO 2 .
The speculated enhancement in electron mobility is further confirmed by measuring the J-V characteristics of electron-only devices with a structure of FTO/ETL/ PCBM/Ag based on SnO 2 -cPCN and pristine SnO 2 . Unless stated otherwise, all the characterizations are based on SnO 2 -cPCN films fabricated by adding 0.6 mg mL −1 cPCN solution to the colloidal solution of SnO 2 . The optimization process toward the highest PSC performance was recorded in the device fabrication section. As illustrated in Fig. 2b, the electron mobility of these two ETLs can be calculated using the space charge-limited current (SCLC)  [40,41]. The electron mobility of SnO 2 -cPCN was 3.3 × 10 −3 cm 2 V −1 s −1 , which was one magnitude larger than those of pristine SnO 2 (9.95 × 10 −4 cm 2 V −1 s −1 ). This is consistent with the conclusion of the XPS test. Considering that the carbon and nitrogen atoms of cPCN are sp 2 -hybridized to form fully π-conjugated electronic structures, we speculate that the excess electron traps in SnO 2 could be consumed under the electron-rich conditions [27], which consequently increased the mobility and decreased the work function [36]. This, in turn, led to different way of band-bending and resulted in a decrease of the energy barrier at the SnO 2 /perovskite interface [36,42], which can significantly affect charge extraction, collection, and recombination in PSC. The higher electron mobility can effectively facilitate electron transfer in the PSCs, reduce charge accumulation at the ETL/perovskite interface, improve efficiency, and suppress hysteresis in the PSCs [43]. Additionally, the optical transmission spectra of SnO 2 and SnO 2 -cPCN films coated on quartz substrates are presented in Fig. 2c. Both samples exhibit high average transmittance in the visible region, demonstrating excellent optical quality to pledge that most light can pass through and be absorbed by the perovskite layer. Ultraviolet photoelectron spectroscopy (UPS) measurements are performed to estimate band position shifts after hybridizing cPCN in SnO 2 (Fig. S4a). According to the formula of work function (WF) = 21.22 eV − E cutoff (cutoff binding energy), the WF of SnO 2 -cPCN is calculated to be − 4.32 eV. Then, the E VBM of SnO 2 -cPCN can be calculated to be − 8.08 eV by E VBM = WF-E F, edge (Fermi edge). It was reported that the WF, conduction band (E CBM ), and bandgap (E g ) of the SnO 2 nanoparticles were − 4.36, − 4.31, and 3.79 eV, respectively [44]. Therefore, the E CBM values of SnO 2 -cPCN can be calculated to be − 4.29 eV considering that the bandgap (~ 3.79 eV) of SnO 2 films is not changed with or without cPCN. According to the theory study in literature [31], we reasonably speculate that the fully conjugated structure of cPCN changes the surface electron density of SnO 2 , resulting in such shift in the energy band of SnO 2 . Subsequently, the E CBM values of perovskite are deduced using the same method to be − 4.27 eV with an E g of 1.58 eV (Fig. S4b) [6]. The energy band diagram for a typical n-i-p PSC device (Fig. 2d) indicates that the SnO 2 -cPCN ETL is more favorable as an effective electron extraction channel conducive to reducing V OC loss.
To ascertain the change in film morphology of ETL after hybridization, we show atomic force microscopy (AFM) and scanning electron microscope (SEM) images of the SnO 2 and SnO 2 -cPCN films deposited on the FTO substrates in Figs. 2e, f and S7, respectively. The SnO 2 -cPCN film turns out to have a similar root-mean-square (RMS) roughness (37.7 nm) as that of the control film (41 nm). In Fig. S5, the water contact angle/perovskite solution of the SnO 2 -cPCN film is measured as 41°/47°, which is much larger than that of the pristine SnO 2 (33.5°/38°). Furthermore, the contact angle of both films also is obtained after UV-ozone treatment and show the same trend (Fig. S6). According to the previous report [25], the non-wetting under-layer may lead to the formation of perovskite films with high-aspect-ratio crystalline grains since the lessened dragging force can result in high grain-boundary mobility [45]. Besides, Deng et al. further proposed that the attraction between the solute ions and solvent molecules on the hydrophilic surface was comparably stronger than the hydrophobic surface [46]. The strong attraction of the surface to the pre-existing clusters will fix more clusters on the substrate surface and block the re-dissolution of ions from pre-existing cluster surfaces. So, the existing cluster becomes more stable and easier to grow into crystal nuclei. In addition, the reaction heat released during cluster growth can be dissipated more rapidly through the surfaces due to a relatively higher interaction intensity, which is beneficial to the formation of the nucleus. Meanwhile, the precursor ions and solvent that are close to the surface will be attracted and then captured by the hydrophilic surface, which will slow down the diffusion of ions, leading to a slower crystal growth rate and smaller grain size [46][47][48]. Conversely, the non-wetting (hydrophobic) substrates can provide a higher free-energy barrier for nucleation, faster crystal growth rate, larger grain size, and less grain boundary [49,50].

Perovskite Film Characterization
Generally, the quality of perovskite films can be defined by some conspicuous features, such as grain size, crystallinity, and surface coverage, which collectively affect the performance of perovskite solar cells. Specifically, the topographic SEM images of perovskite films (prepared with the same composition and process) deposited on SnO 2 and SnO 2 -cPCN are exhibited in Fig. 3a, b to investigate the influence of SnO 2 -cPCN hybridization on the morphology and crystallization of the perovskite films atop. Both films present dense and uniform morphology, with randomly interconnected grains. The statistical size distributions of perovskite grain clusters on different ETLs are plotted in Fig.  S8. The perovskites deposited on SnO 2 possessed an average grain size of about 1.08 µm. As the cPCN is incorporated into the SnO 2 , the appearance of pinholes decreased, and the average grain size significantly increased to 1.66 µm. These phenomena validated that more non-wetting surface after introducing cPCN did enlarge the grain size and decrease the grain boundaries of perovskite, in accord with the contact angle test.
To further study the effect of the cPCN incorporation on the solid structure of perovskite films, we collected XRD patterns of the perovskite films on two kinds of ETLs. It can be observed in Fig. 3c that both samples present a cubic perovskite phase structure, in which the prominent peaks around 13.98° and 28.22° correspond to the (110) and (220) planes. The diffraction peak at 12.6° in all films belongs to the PbI 2 phase [45]. When cPCN was added, the intensity of (110) and (220) diffractions are enhanced, and the calculated full width at half maximum (FWHM) of (110) diffraction becomes smaller (0.112°) compared with the pristine sample (0.129°), suggesting a better-grown film with improved crystallinity. Subsequently, the UV-visible absorption spectra of the perovskite film grown on SnO 2 -cPCN further demonstrates slightly higher absorbance than that of the pristine film (Fig. 3d), contributing to enhancing photocurrents in the PSCs (vide infra).

Photovoltaic Performance
The photovoltaic performance of the novel SnO 2 -cPCN ETL based PSCs was examined by fabricating a series of (FAPbI 3 ) 0.9 (MAPbBr 3 ) 0.1 -based planar-type PSCs with a structure of FTO/ETL/perovskite/Spiro-OMeTAD/Ag, as exhibited in the inset of Fig. 4a. The optimization results of cPCN concentrations in SnO 2 colloidal are provided in Fig. S9 and Table S1. Besides, the current density versus voltage (J-V) characteristics for the champion PSCs based on the SnO 2 and SnO 2 -cPCN ETLs under AM 1.5G illumination (100 mW cm −2 ) are illustrated in Fig. 4a. The main cell parameters of V OC , J SC , FF, and PCE are summarized in Table 1. The device based on the pristine SnO 2 substrate presentes a PCE of 21.3% with J SC of 23.4 mA cm −2 , V OC of 1.11 V, and FF of 82%. After optimization, the best device (0.1cm 2 ) is obtained with a PCE of 23.17%, a V OC as high as 1.126 V, a J SC of 24.9 mA cm −2 , and FF of 82.5%. It is one of the best performances in PSCs with modified SnO 2 [15,51,52]. The superior performance of PSCs with cPCNtreated SnO 2 is in line with the improved film quality and higher absorptions. In Fig. 4b, the incident photon-to-current efficiencies (IPCE) of the optimal solar cells based on SnO 2 -cPCN show significant enhancement over 400-770 nm wavelength than that based on unmodified SnO 2 , which can be attributed to the enhanced UV-visible absorption spectra of the perovskite film grown on SnO 2 -cPCN with improved crystallinity (Fig. 3c) over the same period. Notably, the SnO 2 -cPCN based device has a much faster response of less than 8 s to reach the SPO point than the pristine device (80 s) ascribed to the reduced trap-assisted recombination and the enhancement of electron mobility caused by the incorporation of cPCN [53]. Ion migration, high trap density, and unbalanced charge transport are currently considered the main reasons for the hysteresis [54,55]. To investigate the hysteresis effect in our system, we measured J-V curves along with both bias scanning directions in Fig. 4d and Table 1. The hysteresis indices (H hysteresis ) of the devices were calculated based on Eq. 1 [56]. Besides, the device using SnO 2 -cPCN ETL shows negligible hysteresis, contrasting to the much larger effect with the SnO 2 ETL. The crucial role of SnO 2 -cPCN in resolving the hysteresis issue was also checked with transient photocurrent decay (TPC) and photovoltage decay (TPV) (Fig.  S10a, b). It is revealed that the time constant of the photocurrent decay significantly decreases while the decay time of the photovoltage increases when the device was fabricated based on SnO 2 -cPCN ETL. Due to the comparable electron mobility of SnO 2 -cPCN ETL (3.3 × 10 -3 cm 2 V −1 s −1 ) to the hole mobility of the doped spiro-OMeTAD (~ 10 -3 cm 2 V −1 s −1 ), the electron flux (Fe) is essentially equal to the hole flux (Fh) considering that the interface areas of both ETL/perovskite and perovskite/HTL are the same. Therefore, charge accumulation on either side of the devices based on the SnO 2 -cPCN can be alleviated, which, consequently, exhibit negligible hysteresis (Fig. S11) [19,57]. In addition, we speculate that the K + in carbon nitride may also contribute to the reduced hysteresis in the devices [58,59]. The statistical analysis of all parameters (V OC , J SC , FF, and PCE) of PSCs based on SnO 2 and SnO 2 -cPCN are provided in Fig. 4e (20 cells counted for each case). Fascinatingly, the devices based on SnO 2 -cPCN exhibite excellent repeatability with a minimal standard deviation and reliability in contrast to the devices based on unmodified SnO 2 .

Charge transfer dynamics
Further insights into the origin of the performance enhancement for the device using SnO 2 -cPCN ETL can be obtained from the thermal admittance spectroscopy (TAS) measurement by examining the trap density of states (tDOS) in the devices with and without cPCN (Fig. 5a). TAS is a well-established, effective technique for characterizing both shallow and deep defects and has been broadly adopted to understand defects in the thin film [60,61]. The defect density (N T ) can be estimated by Eq. 3: where C is the capacitance, ω is the angular frequency, q is the elementary charge, k B is the Boltzmann's constant, and T is the temperature. V bi and W represent the built-in potential and depletion width, respectively, and can be obtained from the previous report [62]. The applied angular frequency ω defines the energetic demarcation, where ω 0 denotes the attempt-to-escape frequency and can be output by frequency-dependent capacitance plot [63]. As illustrated in Fig. 5b, the devices with SnO 2 -cPCN ETL have dramatically lower trap density than the SnO 2 ETL (above 0.4 eV). This demonstrates that the perovskite grown on SnO 2 -cPCN ETL possesses reduced deep-level defects at the grain boundary due to the enhanced grain size, contributing to reducing the recombination of carriers and enhancing the performance of PSCs device dramatically [64,65]. Furthermore, the electron-only devices with the structure of FTO/ETL/Perovskite/PCBM/Ag were also manufactured for space charge limited current (SCLC) measurement as circumstantial evidence to the decreased trap density of perovskite deposited on SnO 2 -cPCN. The dark current-voltage (I-V) curves are illustrated in Fig. S12a, b. The trap densities The charge carrier recombination dynamics of the perovskites deposited on different ETLs were investigated by steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) decay measurements. The characteristic PL spectra of all the perovskite films are located at around 784 nm (Fig. 5c). Compared with the pristine sample, the PL intensity of SnO 2 -cPCN based perovskite films is much weaker, demonstrating enhanced charge extraction with reduced recombination, leading to increased photocurrents in the PSCs [53]. For the TRPL spectra in Fig. 5d, all TRPL curves are fitted by a biexponential function as follows: where A 1 and A 2 denote the relative amplitude fractions for τ 1 and τ 2 , respectively. Generally, the fast decay lifetime (τ 1 ) results from the trap-assisted recombination at the interface, and the slow decay lifetime (τ 2 ) is related to the biomolecular recombination of photogenerated free carriers due to traps in bulk [66,67]. The fast decay component (τ 1 ) for perovskite film deposited on the SnO 2 layer is 176 ns. After cPCN is incorporated into the SnO 2 layer, τ 1 is shortened to 30 ns, confirming the much faster charge extraction and transport through the SnO 2 -cPCN compared to the bare SnO 2 . Besides, the reduced interfacial recombination was correlated with the improved conductivity in SnO 2 -cPCN ETL, preventing charge accumulation at the perovskite/ETLs interface [68]. A decrease in recombination at the perovskite/ETLs interface is conducive to promoting the V OC and J SC , resulting from the decreased trap density in ETLs. The result is highly consistent with PL and SCLC measurements.
Electrochemical impedance spectroscopy (EIS) was conducted to investigate the charge transport and recombination behavior. The semicircles in the low and  high-frequency regions are attributed to the recombination resistance (R rec ) and the transfer resistance (R tr ), respectively [69]. Figure 6a exhibits the Nyquist plots of the control and optimal devices in the dark with a bias of 0.5 V and a frequency from 0.1-10 5 Hz. It is revealed that the devices with the SnO 2 -cPCN ETL have larger R rec than the control device owing to the reduced surface/interface trap states and other recombination centers [70]. This behavior agrees with the increased carrier lifetime in TRPL and enhanced V OC and J SC of PSCs based on SnO 2 -cPCN ETL.
Since the SnO 2 -cPCN-based devices presented outstanding performance with a small active area (0.1 cm 2 ), the largearea (active area, 1 cm 2 ) devices were further fabricated to assess the authenticity of cPCN in ETL of PSCs. The J-V curves measured from both directions are illustrated in Fig. 6b

Stability Test
Stability tests are a pivotal characterization of the PSCs. The long-term stability of unencapsulated PSCs fabricated with or without cPCN hybridization was investigated with the devices stored under dark in a glove box filled with nitrogen. As indicated in Fig. 6c, the device based on SnO 2 -cPCN ETL maintains 95% of the original efficiency  Fig. 6 a EIS of planar-type PSCs with SnO 2 and SnO 2 -cPCN ETL, the insert picture is the fitting model. b J-V characteristics of the SnO 2 -cPCN and SnO 2 based devices with a 1 cm 2 area (active area) in forward and reverse scan; the inset is a picture of real 1 cm 2 area devices. c Long-term stability measurements of devices without any encapsulation under N 2 glovebox, and d Humidity and oxygen stability under an ambient condition (relative humidity: 30% ± 5%) after 2880 h (~ 4 months). However, the device based on SnO 2 only keeps 85% of its initial efficiency under the same storage condition. Regarding the stability of PSCs under humidity and oxygen, the SnO 2 -cPCN-based device also presents higher stability than SnO 2 -based devices, maintaining 88% of the initial PCE after 2000 h of storage in the ambient environment (with controlled RH of 30% ± 5%) without any encapsulation (Figs. 6d and S13). For the same test duration, SnO 2 -based devices remains only 64% relative to their initial efficiency. Apparently, the stability against oxygen, humidity, and long-term stability has been significantly improved with the increased quality of the perovskite film deposited on SnO 2 -cPCN. The enhancement of stability can be attributed to the suppressed charge accumulation on the photoexcited perovskite and superoxide-mediated degradation pathway under the increased electron transfer with the SnO 2 -cPCN ETL [71].

Conclusion
To sum up, a novel and effective SnO 2 -cPCN composite yielding superior electron mobility of 3.3 × 10 −3 cm 2 V −1 s −1 was produced. It is more than 3 times higher than that of neat SnO 2 . Notably, the champion PCEs of the planar PSCs based on SnO 2 -cPCN reached 23.17% on a small area device (0.1 cm 2 ), and the PCE of 20.3% was obtained on a 1 cm 2 device. The improved performance of the PSCs based on SnO 2 -cPCN can be attributed to the following advantages. First, the surface wettability of SnO 2 -cPCN slightly decreased after the cPCN addition, suppressing heterogeneous nucleation and enlarging perovskite grain size. Consequently, high-quality perovskite films with reduced grain boundaries and mitigated non-radiative recombination were generated. Second, the high electron mobility and improved band alignment of SnO 2 -cPCN ETL reduced the charge accumulation at the perovskite/ETL interface, leading to negligible current density-voltage hysteresis. This work provides a promising direction for developing high-quality ETLs and verifies the enormous potential of large-scale deployment of perovskite photovoltaics.