Pyridine-functionalized fullerene derivative as an independent electron transport layer enabling efficient and hysteresis-free regular perovskite solar cells

Suitable electron transport materials bearing good interfacial contact, improved electron transport ability, and matched energy levels are indispensable for developing efficient perovskite solar cells (PSCs). Herein, regular (n-i-p) planar Cs 0.05 FA 0.83 MA 0.12 PbI 2.55 Br 0.45 (CsFAMA) PSC devices were fabricated using a pyridine-functionalized fullerene derivative (C 60 -3-BPy) as an independent electron transport layer (ETL), delivering a decent power conversion efficiency (PCE) of 18.22%, which is dramatically higher than that of the control device based on [6,6]-phenyl-C 61 -butyric acid methyl ester (PCBM) ETL (15.70%). The energy level offset between C 60 -3-BPy and the perovskite is smaller than that based on PCBM ETL, which is beneficial for efficient ohmic contact in ETL/perovskite interface and improved open-circuit voltage ( V oc ). Moreover, C 60 -3-BPy affords strong coordination interactions with perovskite, leading to an improved film quality of the perovskite layer with enlarged grain size and decreased trap state density, which contribute to facilitated electron extraction as reflected by the increases of both the fill factor (FF) and the short-circuit current ( J sc ). C 60 -3-BPy-facilitated electron extraction further results in hysteresis-free devices.

innovation of improving device structure and modifying the perovskite/interface layer interface, tuning the perovskite composition, crystalline phase and morphology of the perovskite film, the certified PCE of PSCs have exceeded 25.5%. [7] Although great progress has been made on device efficiency and stability, continuous efforts are still being devoted to improving the development of highperformance PSCs. Regular (n-i-p) structure PSCs have exhibited a faster evolution compared to that of inverted (p-i-n) structure PSCs, and the classic device structure of ni-p PSCs consists of a perovskite layer sandwiched between an underneath electron transport layer (ETL) and an upper hole transport layer (HTL). [8,9] Specifically, ETLs featuring an excellent electron-transporting ability, good optoelectronic properties and a compatible conduction band are beneficial for the charge transfer, morphology and crystallization regulation of perovskite films. [10,11] The most commonly used ETLs within high-efficiency n-i-p PSCs consist of titanium dioxide (TiO 2 ). [2,10,11] However, the compact TiO 2 ETL obtained via a high-temperature sintering treatment (>450 o C) significantly limits their application in flexible devices. [12,13] Moreover, the TiO 2 ETL obtained from a low-temperature procedure has the disadvantage of presenting low electrical conductivity and a high trap state density, severely hampering the PCE elevation as well as the long-term stability of the devices. [14,15] A series of strategies, including morphology optimization, doping, and interface modifications applied in the low-temperature solution-processed TiO 2 ETL have been proposed to address the above-mentioned issues. [11,16,17] Therefore, developing novel metal-oxide-free ETLs to avoid the drawback of TiO 2 ETLs for improved device performance has become a desirable alternative.
Fullerene derivatives have become efficient ETLs applied in n-i-p PSCs due to the advantages such as excellent electron mobility, low-temperature solution process, and tailored molecular structure and energy level. [18][19][20] Seok et al. deposited [6,6]-phenyl-C 61 -butyric acid methyl ester (PCBM) onto a flexible FTO/PEI substrate to construct n-i-p MAPbI 3 -based PSCs, achieving a PCE of 11.1%. [21] Snaith et al. employed cross-linked fullerene layers as the ETL to prepare n-i-p MAPbI 3-x Cl x PSCs, achieving a PCE of 16.6%, in which cross-linked fullerene layers greatly improve the electron collection capability, benefiting to enhanced device performance. [22] Thereafter, Snaith et al. prepared composite ETLs consisting of C 60 and 3-dimethyl-2-phenyl-2,3-dihydro1H-benzoimidazole (N-DMBI) to fabricate CH 3 NH 3 PbI x Cl 3-x -based n-i-p planar PSCs, delivering a remarkable PCE of 18.3%. [23] Fang et al. reported that C 60 pyrrolidine tris-acid (CPTA) was applied as an ETL to covalently anchor the underneath ITO substrate for preparing n-i-p MAPbI 3 -based PSCs, delivering a PCE of 18.29 % and significantly suppressing current-voltage hysteresis. [24] However, there have been only few examples of fullerene-derived ETLs applied in n-i-p PSCs which involve complex fabrication or synthesis procedures. Therefore, exploiting novel and efficient fullerene-derived ETLs with facile synthetic approaches is desirable to promote the development of n-i-p PSCs, especially for flexible substrates and low-cost devices. Recently, we reported a novel pyridine-functionalized fullerene derivative (C 60 -3-BPy) by a facile one-step method, which was applied as the ETL of inverted (p-i-n) MAPbI 3 -based PSCs, exhibiting a higher efficiency and stability than that of the control devices with conventional PCBM as ETL. [25][26][27][28] Whether such a pyridine-functionalized fullerene derivative can be applied as an independent ETL in n-i-p PSCs affording improved PCE remains an open question.
Herein, C 60 -3-BPy was applied for the first time as an independent ETL to construct planar n-i-p Cs 0.05 FA 0.83 MA 0.12 PbI 2.55 Br 0.45 (CsFAMA) PSCs, achieving a high PCE of 18.22%, which outperformed that of the control devices based on PCBM ETL (15.70%). The superior performance of PSCs with C 60 -3-BPy ETL was attributed to the fact that C 60 -3-BPy enables effective regulation of the work function of the ITO cathode as well as the crystalline quality of the perovskite film.

RESULTS AND DISCUSSION
The preparation procedure for the n-i-p planar CsFAMA PSCs and the device architecture is shown in Figure 1A. Briefly, the chlorobenzene solution containing PCBM or C 60 -3-BPy with a concentration of 5 mg ml −1 was spincoated onto the underneath ITO substrates. Thereafter, a CsFAMA perovskite precursor solution was spin-coated onto the fullerene derivative ETLs. After depositing the Spiro-OMeTAD HTL and Au electrode, the resultant PSCs were measured to evaluate the performance of the ETLs. The current-voltage curves of the PSCs with PCBM or C 60 -3-BPy as ETLs are shown in Figure 1B and the detailed photovoltaic parameters were recorded in To ensure accuracy in device measurements, more than 30 independent devices bearing PCBM ETL or C 60 -3-BPy ETL were prepared, and the statistical photovoltaic parameters are portrayed in Figure S1. The narrower PCE distribution for the device with C 60 -3-BPy as the ETL shown in the histograms ( Figure. S2) suggests the reliability of the device fabrication procedure. Furthermore, according  to the in-depth comparison of these photovoltaic parameters, it is clear that the PCE enhancement of the device with a C 60 -3-BPy ETL is mainly due to the significantly increased FF (from 68.18% to 74.86 %, ∼9.79% enhancement) and slightly increased J sc (from 21.27 mA cm −2 to 21.82 mA cm −2 , ∼2.58% enhancement) and V oc (from 1.03 V to 1.07 V, ∼3.88% enhancement), relative to those of the PSCs with PCBM ETL. The change in J sc was verified by the comparison of the external quantum efficiency (EQE) for the devices with PCBM or C 60 -3-BPy ETL ( Figure 1C). The device with the C 60 -3-BPy ETL shows a more predominant EQE response in the region of 580-700 nm and a more integrated J sc relative to that of the control device with PCBM ETL, indicating improved effective photon utilization after using the C 60 -3-BPy ETL. The current-voltage hysteresis of PSCs with PCBM ETL or C 60 -3-BPy ETL was thereafter studied by the J-V curve measurements in different scan directions ( Figure S3). The hysteresis index can be calculated from Equation 1 as: The PSCs with C 60 -3-BPy ETL presented a hysteresis index of 1.65%, which is lower than that (6.99%) of the control device with PCBM ETL (see Table S1), indicating negligible hysteresis for the PSCs with C 60 -3-BPy ETL as well as a reduced trap state and charge accumulation in the perovskite layer.
Ultraviolet photoelectron spectroscopy (UPS) measurements were performed on the C 60 -3-BPy/ITO and PCBM/ITO films to reveal the influence of these deposited fullerene derivatives ETLs on the performance of PSCs ( Figure 2). The obtained work function (W F ) from the UPS was -4.14 eV and -4.09 eV for ITO/PCBM and ITO/C 60 -3-BPy, respectively. The W F of the pristine ITO was -4.39 eV. Notably, the fullerene derivatives deposited on the ITO substrate significantly decreased the W F of pristine ITO. Moreover, C 60 -3-BPy renders a stronger interaction with the ITO substrate relative to that of the PCBM, leading to a decreased energy offset between the W F of the ITO and the conduction band energy level (CB) of perovskite, suggesting that a lower energy loss is favorable for electron transport. Furthermore, the smaller energy offset between the ITO/C 60 -3-Bpy and the perovskite film is beneficial for the formation of an ohmic contact, facilitating electron transport from the perovskite film to the ITO electrode and resulting in increased V oc .
The morphology of the ETLs has been unveiled to possess a considerable effect on the upper perovskite film quality. Therefore, the film morphology of ITO/PCBM and ITO/C 60 -3-BPy was investigated by atomic force microscopy (AFM) (Figure S4), showing a smoother surface for the ITO/C 60 -3-BPy with a root-mean-square of 0.92 nm while that of ITO/PCBM was 1.01 nm. The smoother surface of the ITO/C 60 -3-BPy is beneficial for reduced interface traps and the growth of large-size grains of perovskite. Thereafter, the influence of the ITO/PCBM and ITO/C 60 -3-BPy substrates on the morphology regulation of the CsFAMA perovskite layer was studied by scanning electron microscopy (SEM). Figure 3A,B show that the CsFAMA perovskite precursor deposited on the ITO/PCBM and ITO/C 60 -3-BPy substrates delivers a compact surface perovskite film without obvious pinholes. According to the statistic grain-size distribution shown in Figure S5, the ITO/C 60 -3-BPy/perovskite film possesses a grain size of 238 nm, which is larger than that of ITO/PCBM/perovskite with a value of 226 nm, suggesting better film quality of perovskite after the incorporation of the C 60 -3-BPy ETL.
This hypothesis was verified by powder X-ray diffraction (XRD) and synchrotron-based grazing incidence X-ray diffraction (GIXRD) measurements. The XRD patterns of ITO/PCBM/perovskite and ITO/C 60 -3-BPy/perovskite films show featured diffraction peaks at 14.2 • , 28.5 • , and 31.9 • assigned to the (001), (002), and (012) diffraction planes of perovskite, respectively ( Figure 3C,D). The perovskite film deposited on the ITO/C 60 -3-BPy substrate exhibits enhanced diffraction intensity along with a drastically decreased PbI 2 diffraction peak at 12.8 • , indicating the improved crystalline quality of the perovskite film. The 2D-GIXRD patterns of the ITO/PCBM/perovskite and ITO/C 60 -3-BPy/perovskite films at an X-ray incident angle of 1.0 • show scattered rings at q≈10, 20, and 22.2 nm −1 assigned to the (001), (002), and (012) diffraction planes of the perovskite crystals, respectively. The ITO/C 60 -3-BPy/perovskite film exhibits brighter and sharper scattered rings that the ITO/PCBM/perovskite film, suggesting improved film quality. Furthermore, the weaker scattered rings assigned to PbI 2 within the ITO/C 60 -3-BPy/perovskite film suggests the effective suppression of PbI 2 harmful to device performance. [29] Steady-state photoluminescence (PL) measurements and time-resolved photoluminescence (TRPL) spectroscopy measurements were performed to unveil the charge transport property of the PSCs with PCBM or C 60 -3-BPy ETLs, as shown in Figure 4A,B. The steady-state PL spectra of the ITO/PCBM/perovskite and ITO/C 60 -3-BPy/perovskite films show the featured PL peak at 766.4 nm of CsFAMA perovskite under an excitation wavelength of 532 nm ( Figure 4A). It is noteworthy that the C 60 -3-BPy ETL sandwiched between the perovskite layer and the underneath ITO electrode results in remarkable PL quenching relative to that of the counterpart device with PCBM ETL, indicating the more powerful electron extraction and transport capability of C 60 -3-BPy ETL due to the special coordination interaction and the suitable energy level between C 60 -3-BPy and the perovskite layer. [30][31][32] In addition, the PL peak of the ITO/C 60 -3-BPy/perovskite film shows a blue-shift from 766 nm to approximately 757 nm relative to that of the ITO/PCBM/perovskite film, suggesting decreased trap states within perovskite with C 60 -3-BPy ETL resulted from the improved perovskite crystallinity, the reduced grain boundaries, and Lewis acid-base reaction between the pyridine group and Pb 2+ of the perovskite. [20,27,28,[33][34][35] The TRPL measurements were further used to study the charge transfer kinetics from the perovskite to the fullerene derivatives ( Figure 4B and Table S2). The life of 21.12 ns for C 60 -3-BPy/perovskite film is shorter than that of the PCBM/perovskite film (23.59 ns), suggesting the effective charge transfer from perovskite to C 60 -3-BPy ETL. [36] Space charge limited current measurements were performed using electron-only devices to further quantify the trap state density of perovskite after the incorporation of PCBM or C 60 -3-BPy ETLs, and the structure of ITO/PCBM or C 60 -3-BPy ETL + perovskite/PCBM/Ag was analyzed. [37,38] The obtained J-V curves in the dark are depicted in Figure 4C. According to the bias voltage at the kink point, which is defined as the trap-filled limit voltage (V TFL ), the corresponding trap-state density (n t ) can be determined as 1.59 × 10 15 cm −3 and 1.24 × 10 15 cm −3 for the PSC device with PCBM and C 60 -3-BPy ETL, respectively (Supporting Information S7). The decreased trap-state density of the PSC device with C 60 -3-BPy ETL demonstrates that the C 60 -3-BPy ETL enables a high-quality perovskite film and constructs an optimal ETL/perovskite interface. Such an improvement on the quality of perovskite film is expected to benefit the improvement of the ambient stability of the PSC devices.
Electrochemical impedance spectroscopy (EIS) measurements were performed to investigate the charge transport dynamics in the dark and a reverse potential of 1.0 V. The obtained Nyquist plots of the devices and the corresponding fitted curves through an equivalent circuit are shown in Figure 4D. [39][40][41] The fitted parameters from the Nyquist plots including the series resistance (R s ) and charge transfer resistance (R ct ) are summarized in Table  S3. Moreover, it was determined that the constant phase element is related to the non-ideal chemical capacitance. The distinct discrepancy on R ct for the devices with PCBM ETL (1053.6 Ω⋅cm 2 ) and those with C 60 -3-BPy ETL (472.9 Ω⋅cm 2 ) suggest that C 60 -3-BPy ETL enables suppressed hole-electron recombination and facilitated charge transport.

CONCLUSIONS
Regular n-i-p CsFAMA PSCs using C 60 -3-BPy as an independent ETL afforded a decent PCE of 18.22%, which outperformed the control device with PCBM ETL (15.70%). C 60 -3-BPy plays an important role in modifying the work function of the ITO electrode and providing a smooth surface, which is beneficial for obtaining high-quality perovskite films and the matched energy alignment of the device. Furthermore, the high electron affinity of C 60 -3-BPy contributes to improved charge extraction. Moreover, the PSC with C 60 -3-BPy ETL exhibits negligible current-voltage hysteresis. This study demonstrates the potential of fullerene derivatives in realizing large-area, low-temperature, and flexible PSC devices.

Device fabrication
The patterned ITO-coated glass was cleaned by sequential ultrasonic treatment in detergent, deionized water, acetone, and isopropanol for 15 minutes, and then treated with ultraviolet-ozone for 20 minutes. C 60 -3-BPy was synthesized via a facile one-step Prato reaction at 80 o C as reported previously. [26,27] C 60 -3-BPy was dissolved in chlorobenzene with concentrations of 5 mg ml −1 and spin-coated onto the ITO substrates and annealed at 100 o C for 10 minutes. Next, CsFAMA perovskite precursor solu-

Measurements and characterization
Current density-voltage (J-V) characterizations were conducted using a Keithley 2400 source measurement unit under simulated AM 1.5 irradiation (100 mW cm −2 ) with a standard xenon lamp-based solar simulator (Oriel Sol 3A). The solar simulator illumination intensity was calibrated with a monocrystalline silicon reference cell (Oriel P/N 91150 V, with KG-5 visible color filter) calibrated by the National Renewable Energy Laboratory. EQE measurements were carried out on an Oriel Intelligent Quantum Efficiency 200TM measurement system which was equipped with a tunable light source. SEM images were obtained using a field emission scanning electron microscope (Zeiss Gemini SEM 500, Germany). The contact angles were measured using a CAM instrument (Data Physics, Germany). AFM measurements were performed on a XE-7 scanning probe microscope in noncontact mode (Park Systems, Korea). UPS measurements were carried out at the Catalysis and Surface Science station at the BL11U beamline in the National Synchrotron Radiation Laboratory. The GIXRD measurements were performed at the BL14B1 beamline of Shanghai Synchrotron Radiation Facility using X-ray with a wavelength of 1.38 Å. The 2D-GIXRD patterns were obtained by a MarCCD detector mounted vertically at a distance of around 274 mm from the sample with an exposure time of 50 seconds at a grazing incidence angle of 1 • . The XRD patterns were recorded on a Rigaku SmartLab X-ray diffractometer with Cu-Ka radiation (0.154 nm). The steady-state photoluminescence (PL) spectra were recorded using an Edinburgh Instruments FLS920 fluorescence spectrometer with an excitation wavelength of 460 nm. For TRPL spectra, the samples were excited by a 543 nm picosecond pulsed diode laser with a pulse width of 104 ps (Picoquant Gmbh Solea Supercontinuum laser using time-correlated singlephoton counting method, and TimeHarp 260 software was used to record the decays.) Impedance spectroscopy (EIS) measurements were performed using an electrochemical workstation (Autolab 320, Metrohm, Switzerland) in a frequency range of 1 Hz to 1 MHz under 1.0 V in the dark. Alternating current 20 mV perturbation was applied with a frequency from 1 MHz to 1 Hz. The obtained impedance spectra were fitted with Z-View software.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.