1. Introduction

Organic-inorganic hybrid PSCs have recently drawn tremendous attention because of their simple solution processing, tunable optoelectronic properties, high carrier migration, low photo exciton binding energy, excellent bi-polar carrier transport capacity, efficient conduction of electrons and holes, rapid increase in power conversion efficiencies (PCE) [1] - [8] . In 2009, Miyasaka and coworkers introduced a seminal work involving organic perovskite solar cells demonstrated with a low PCE of 3.8% [9] . After a decade of development, the highest recorded and certified PCE has been reached 25.2%, meeting the needs of commercial solar cells [10] . However, photovoltaic cells are more expensive than conventional energy sources, such as fossil fuels, wind, hydro, and nuclear, due to the manufacturing materials involved. Reducing the amount of materials used or synthesizing cheaper alternatives with similar functions to facilitate the commercialization of photovoltaic cells was necessary [11] . The PSCs device structure can be divided into two categories, a normal n-i-p architecture and an inverted p-i-n architecture [12] [13] [14] [15] . The normal n-i-p structure often needs mesoporous TiO₂ as the ETL, and the production of this material requires high temperature environment (>400˚C) [16] [17] . For PSCs with an inverted p-i-n structure, the conventional organic p-type material is usually PEDOT:PSS [14] [18] , poly (triarylamine) (PTAA) [19] [20] . In contrast, inorganic p-type materials have nickel (II) oxide (NiO_x) [21] [22] , copper thiocyanate (CuSCN) [23] [24] , and copper iodide (CuI) [24] [25] , while the n-type material is mainly fullerenes and its derivatives. Therefore, researchers prefer inverted p-i-n structures involving a low temperature preparation process and low hysteresis effect [26] .

In the production of PSCs, the commonly used electronic transport materials and electronic trap states passivator of p-i-n structure include PCBM, C₆₀ and other fullerene derivatives [27] [28] [29] . Many studies have proved that in high-efficiency PSCs, C₆₀ and its derivative PCBM have excellent electron acceptability due to the spherical structural strain, hence making this kind of material the most effective passivation agent. However, since the PCBM has a higher electronic acceptance and excellent passivation effect on the active layer, it is widely used as a material of ETL. However, as an ETL passivation layer, C₆₀ is capable of passivation depth defects, but its passivation effect is worse than that of PCBM. Due to its much cheaper price, the use of C₆₀ as a passivation layer device promotes the commercialization process. In this work, considering the high price of PCBM, the method of micro-doping not only avoids the tedious process of spin coating multi-layer film, but also produces high-quality perovskite-PCBM heterojunction photoactive film [29] . In the whole device, the quality of active layer is the most critical factor affecting the PCE of photovoltaic cells. To obtain dense perovskite film, the purity of precursor solution should be high, and the solute should be completely dissolved. The solubility of lead iodide (PbI₂), Methyl ammonium iodide (MAI), and PCBM in organic solvents is limited, especially after the incorporation of PCBM. Therefore, in the preparation of doping solution, PCBM is stirred first and dissolved with N,N-Dimethyl formamide (DMF) and Dimethyl sulfoxide (DMSO) mixed solvent, and then perovskite precursors are added to form a slightly doped perovskite precursor solution. The addition of PCBM affected the morphology of perovskite, and an appropriate amount of PCBM can increase the grain size of perovskite film. A simple and convenient one-step method is used to spin coat the perovskite films. Figure 1(a) shows the structure of the devices in this work. Although the PSCs have higher carrier mobility and longer carrier life-times, the whole device will still have non-radiative recombination, which limits their V_oc value below the Shockley-Queisser theory [30] [31] . In this experiment, it was found that doped PCBM could not only passivate perovskite, increase grain size, but also improve the cells’ V_oc. When the PCBM doping concentration in MAPbI₃ precursor solution was from 0 - 1 wt%, the J_sc, FF and V_oc of the devices increased, the V_oc of the planar ITO/PEDOT:PSS/MAPbI₃ with PCBM/C₆₀/BCP/Ag (p-i-n) device structure reached more than 0.99 V, and the V_oc loss was less than 1% compared with the theoretical value.

2. Experimental Section

2.1. Materials

Lead iodide (PbI₂, 99.8%) and bathocuproine (BCP, 99%) were purchased from the Xi’an Polymer Light Technology Corp, CH₃NH₃I (MAI) was purchased from Greatcell solar Australia Pty Ltd, PCBM (99.5%) was purchased Nano-C, PEDOT:PSS conductive liquid was purchased from Heraeus, C₆₀ (99.9%) was purchased from Liaoning Youxuan New Energy Technology Co., Ltd. We obtained the Silver (Ag, ≥99.9%) and the solvents, which include DMF (≥99%) and DMSO (≥99.9%) from Sigma-Aldrich. All chemicals were used as received without further purification.

2.2. Device Fabrication

The patterned ITO glass substrates were subsequently cleaned in an ultrasonicator with ITO detergent, acetone, isopropyl alcohol and deionized (DI) water for 15 minutes each. After that, the ITO substrates were treated in a UV ozone oven for 15 min. First, PEDOT:PSS was spin-coated at 4000 rpm for 30 s onto the ITO glass substrate, followed by annealing at 150˚C for 10 min. Pristine perovskite precursor solution was prepared by dissolving PbI₂ and MAI (1:1 molar ratio) in the solvent mixture of DMF and DMSO (9:1 v/v) for a total concentration of 1.5 m in the glovebox. For doped perovskite precursor solution, perovskite concentration remained unchanged, and PCBM concentration was calculated according to the amount of PbI₂. The solution was stirred at room temperature overnight. Then, the perovskite layer was spun onto the hole transport layer (HTL) at 3000 rpm for 10 s and 5000 rpm for 50 s. Next the substrate was annealed on a hot plate at 100˚C for 20 min. Finally, the devices were finished by thermally evaporating C₆₀ (50 nm), BCP (8 nm), and Ag (80 nm) in the order.

2.3. Optoelectronic Characterization

The perovskite thin film surface morphology images were characterized by atomic force microscopy (Hitachi AFM 5100N). The top-view scanning electron microscopy (SEM, JEOL JEM5610) of the perovskite, which were spin coated on ITO/PEDOT:PSS substrates. Steady state photoluminescence (PL) spectra and PL mapping were acquired by Raman spectrometer (LabRAM HR Evolution) with 532 nm laser unit, which was manufactured by HORIBA FRANCE SAS. The time-resolved photoluminescence spectra (FLS1000, TRPL) were obtained with the range of wavelength from 600 nm to 800 nm by exciting at 532 nm laser. The absorption spectra of the perovskite films were determined by UV-visible spectrophotometer. The crystallinity of the perovskite thin film was monitored by X-ray diffraction (XRD, Rigaku, SmartLab3kW). The current density-voltage (J-V) curves were measured using a solar simulator with a source meter (Keithley 2400) at 100 mW/cm², AM 1.5 G illumination. The J-V curves were measured by scanning from negative bias (−0.2 V) to forward bias (+1.2 V). The External quantum efficiency (EQE) spectra were measured with an EQE measurement system, and the light intensity at each wavelength was calibrated with a standard single-crystal Si photovoltaic cell. All optical and electrical measurements were performed at room temperature.

3. Results and Discussion

PL spectra are an effective and non-destructive method to characterize the optical properties of semiconductor materials. By analyzing a PL spectrum, the bandgap, impurity type and activation energy of semiconductor material can be obtained. The TRPL indicates that electron relaxation and non-radiative recombination physical mechanisms can be obtained by detecting the change of photoluminescence intensity at a certain wavelength with time by pulsed excitation. PL and TRPL were used to investigate the effect of the PCBM additive on recombination and carrier lifetime of the perovskite film. It can be seen from the pristine and PCBM-perovskite thin films PL spectrum shown in Figure 1(b) that the PL emission peak centered at 766 nm did not change even when the dose of PCBM in perovskite films increased, and the obtained band gap of perovskite was about 1.61 eV. However, the PL intensity increased with different concentrations of PCBM. When the concentration of PCBM was 1.0 wt%, the PL intensity was the highest, but when the additive concentration was more than 1.0 wt%, the PL strength was weak, indicating that the crystallinity of the photoactive layer was the best when the additive concentration was 1.0 wt%. Meanwhile, in order to analyze the uniformity of perovskite films, Figures 1(c)-(f) showed the PL mapping of the micro-region (20 μm × 20 μm) centered on the emission peak of 766 nm, showing the composition uniformity of perovskite thin films doped with PCBM of different concentrations. Furthermore, the TRPL spectrum was used to evaluate the carrier extraction from the devices. The spectra were fitted by biexponential decay function of $y=y_0+A_1\ast {\text{e}}^{-x/{\tau }_1}+A_2\ast {\text{e}}^{-x/{\tau }_2}$ , and the fast decay constant ${\tau }_1$ was correlated to surface property, while the slow decay constant ${\tau }_2$ was associated with bulk property [32] . Herein, we calculated the average decay lifetime according to the equation of ${\tau }_{avg}={\displaystyle \sum A_i{\tau }_i^2}/{\displaystyle \sum A_i{\tau }_i}$ . The obtained results suggest the existence of two quenching processes including the carrier extraction occurred at the interfaces of PEDOT:PSS/perovskite with different PCBM as well as the carrier trap recombination. As shown in Figure 1(g), the average carrier lifetime of ITO/PEDOT:PSS/perovskite films with different PCBM concentrations (W/O, 0.5 wt%, 1.0 wt% and 1.5 wt%) were 982 ns, 1057 ns, 1056 ns, and 904 ns, respectively. These results indicated a faster photogenerated carrier extraction for ITO/PEDOT:PSS/perovskite films with slight PCBM, which was beneficial to the J_sc and PCE values of the devices. However, when the additive of PCBM concentration continues to increase, the photogenic carrier life would decrease sharply.

The UV-vis absorption spectra of the investigated perovskite films are shown in Figure 2(a). It can be seen from the UV-vis spectra that the MAPbI₃ perovskite doped with the PCBM exhibited a higher optical absorption than the pristine

photoactive layer across the entire visible spectrum. Generally, UV-vis absorption is mainly caused by the transition of electrons from the ground state to the excited state. Meanwhile, in semiconductor materials, the transition of electrons from the ground state to the excited state is related to their energy band structure. Therefore, the absorption law of light must be related to their energy band structure. The differential of the absorption spectrum curve and PL spectrum were shown in Figure 1(b), and it was found that the emission peak position of PL was consistent with the cutoff point of the absorption spectrum. This confirms the bandgap size of perovskite film.

XRD patterns (Figure 2(b)) show the crystallinity of the perovskite films grown on ITO/PEDOT:PSS. The pristine perovskite and perovskite with PCBM films had the same diffraction peak. The addition of PCBM did not change the phase composition of perovskite. Due to the structure and large molecular weight of the additive, it did not enter the crystal lattice of perovskite, but occupied the void position at the grain boundary, thus passivating perovskite and forming a denser PCBM-perovskite film. However, excessive additive of PCBM would make the film chapped, and the overall quality of the light absorption layer film would decline, thus decreasing the photovoltaic performance of the devices. To intuitively compare the crystallinity of perovskite films with different doping concentrations, the diffraction peak at 14.01˚ was amplified and smoothened. When the additive concentration of PCBM was 1 wt%, the diffraction peak was the strongest, indicating that the film at this time had the best crystallinity, which was self-consistent with the experimental results.

In order to visually describe the flatness of perovskite films, the AFM and SEM images characterize the flatness of perovskite films grown by spin coating on ITO/PEDOT:PSS substrates doped with PCBM of different concentrations. The measured root mean square (RMS) roughness was shown in Figures 3(a)-(d). The RMS roughness of the pristine perovskite film was determined to be 12.9 nm, larger than that of the perovskite with 1 wt% PCBM (8.7 nm). The results show that the surface flatness of perovskite films can be increased by micro-doping. However, increasing the concentration of PCBM in the perovskite precursor solution would restrain the smoothness of perovskite film surface. When the concentration of PCBM reached 1.5 wt%, the RMS roughness increased to

27.4 nm. This can be confirmed by the SEM morphology of the involved perovskite film. The amount of additive PCBM amount in perovskite films played an important role in changing the surface of perovskite films and devices. It can be seen from Figures 3(e)-(h) that the grain size would increase with the addition of PCBM. When the doping concentration of PCBM was 1 wt%, macromolecular agglomeration began to appear. As the additive doping concentration continues to increase, the agglomeration of PCBM became increasingly evident, and the surface morphology decreased.

It was found from a series of doping concentration J-V curves (Figure 4(a)) that the concentration of PCBM in the perovskite precursor would affect the photovoltaic parameters of the whole device. The corresponding photovoltaic parameters are listed in Table S1 (Support information). The J_sc of the device is related to the band gap, thickness, surface morphology, and carrier transport properties of the perovskite layer [33] . The addition of a certain concentration of PCBM can improve carrier mobility and thus obtain a higher J_sc, as illustrated in Figure 4(b). It was found that the FF of the device increased when the additive was very small (Figure 4(c)). Since the FF of the device is affected by series and parallel resistance, it reflects the ideal degree of the diode of the device. Especially for the V_oc of the device, When PCBM was added to the photoactive film as an additive, the V_oc of the device changed from 0.82 to 0.99 V (Figure 4(d)). Generally, the inverted p-i-n PSCs with PEDOT:PSS as HTL and C₆₀ as ETL exhibit a V_oc less than 1.0 V. Theoretically, the V_oc of the PSCs are determined by the highest occupied molecular orbital (HOMO, −5.5 eV) level of the donor and the lowest unoccupied molecular orbital (LUMO, −4.5 eV) level of the acceptor [34] [35] [36] [37] , so the V_oc $\left(V_{oc}\approx 1/e\left(\left|E_{HOMO}^{donor}\right|-\left|E_{LUMO}^{acceptor}\right|\right)\right)$ , where $E_{HOMO}^{donor}$ is the HOMO level of the donor and $E_{LUMO}^{acceptor}$ is the LUMO level of the acceptor [38] [39] . Combining with the above equation, it can be seen from the energy level diagram (Figure 4(e)) that when the concentration of photoactive layer doped PCBM was 1 wt%, the V_oc loss of the device was 1%. In theory, for PSCs, the V_oc loss is related to the perovskite bandgap (E_g) and the elementary charge (e), determined by the equation of ( $V_{loss}=E_g/e-V_{oc}$ ) [40] [41] [42] . In general, the high V_oc loss can be attributed to the high trap density of the perovskite surface causing severe interface defects non-radiative recombination, and an undesired energy-level mismatching between the n-type layer and p-type in the device [40] [43] . The loss mechanism of V_oc is an important and necessary research topic for ultra-efficient PSCs. According to recent literatures, the research on high-efficiency PSCs and low-loss V_oc and the development of high efficient crystalline silicon solar cells show that it is very necessary to improve the V_oc of solar cells. In recent years, PSCs have made continuous breakthroughs in efficiency. The most important means is to improve carrier management by using interface passivation technology to reduce interface non-radiation recombination and increase V_oc to improve efficiency. In the solar cell measuring process, PCE (the efficiency of converting solar light energy into electrical energy) is the most important. This directly reflects the performance of the devices. The PCE

value of the cells at different of PCBM ratio is shown in Figure S1 (Support information). It can be seen from Figure 4(f) that a noticeable EQE enhancement was characterized for PCBM incorporated PSCs. Furthermore, the EQE spectrum was also measured for devices constructed from different additive concentrations of PCBM in order to evaluate the efficient photocurrent at various wavelengths and to provide an additional measure of material band gap from the beginning to the end of photocurrent generation.

4. Conclusion

In summary, through a series of optical characterization such as PL, SEM and UV-vis, it was shown that the addition of small amount of PCBM to the precursor can form a self-passivating, high-quality perovskite thin film. As a result, the solar cells obtain a higher V_oc, J_sc, and FF. From the theoretical formula, the attained of 0.99 V is among the smallest V_oc loss for p-i-n type PSCs with PEDOT:PSS as HTL and C₆₀ as ETL in the obtained perovskite band gap of 1.61 eV. Systematic perovskite thin film optical characterization of the PSCs V_oc enhancement indicated that the improvement is mainly attributed to the enhanced crystallinity of photoactive layer and interface efficient charge carrier dissociation with decreased non-radiative recombination. Therefore, PCBM as micro-additive may be applicable to other organic-inorganic hybrid photoelectronic devices.

Acknowledgements

This work was supported financially by the fund for Bagui Talent of Guangxi Province (No. T3120097921), Talent Model Base (No. AE31200065) and Guangxi University Foundation (No. A3120051010), China.

Support Information