Morphology Tuning and Its Role in Optimization of Perovskite Films Fabricated from A Novel Nonhalide Lead Source

Abstract Usage of nonhalide lead sources for fabricating perovskite solar cells (PSCs) has recently attracted increasing attention as a promising route toward realizing high quality PSC devices. However, the unique role of nonhalide lead sources in improving perovskite film morphology and PSC performance has largely remained unexplored, impeding broader application of these materials. Here, it is demonstrated that by using a new nonhalide lead source, lead formate (Pb(HCOO)2), good control of perovskite film morphology can be achieved. With the usage of lead formate, PbI2 can nicely border the perovskite grain boundaries (GBs) and form domain “walls” that segregate the individual perovskite crystal domains. The PbI2 at the GBs lead to significant improvement in film quality and device performance through passivating the defects at the perovskite GBs and suppressing lateral carrier diffusion. An impressive carrier lifetime at the microsecond scale (τ 2 = 1714 ns) is achieved, further with an optimal power conversion efficiency of 20.3% for the resulting devices. This work demonstrates a promising and effective method toward fabricating high‐quality perovskites and high‐efficiency PSCs.

During the spinning process, about 58psi continuous nitrogen gas (room temperature) stream was blown over the film till the spin-coating finishes. At last, the substrates were annealed at 100°C on a hot plate for 20min. All of the procedures were carried out in a glovebox under the atmosphere of nitrogen.
Further, in order to improve photoelectric characteristics of solar cells, MACl (0mg/mL, 7mg/mL, 14mg/mL, 21mg/mL) was added into perovskite precursor solution with a Pb(HCOO)2:MAI molar ratio of 1:3.15, the champion device was obtained at a concentration of 14mg/mL. The perovskite films based on with/without additive of MACl were fabricated through the similar procedure.
Preparation of MAPbI3 films via the Pb(CH3COO)2 route. The MAPbI3 perovskite films were prepared using Pb(CH3COO)2 source, by varying the ratios of Pb(CH3COO)2 and MAI in DMF with DMSO additives (See Supplemental Information for more information). Then the precursor solution was spincoated on the glass substrate at 2800rpm for 40s. During the spinning process, about 58 psi continuous nitrogen gas (room temperature) stream was blown over the film until the spin-coating finished. At last, the MAPbI3 perovskite films using Pb(CH3COO)2 source were annealed at 100C for 15min (optimized conditions). These MAPbI3 films were used for characterize their surface morphology and carrier lifetime.
Device fabrication. Glass substrates with fluorine-doped tin-oxide (FTO) coating were cleaned with three different solvents in a sequence of deionized water, acetone and ethanol, followed by being purged in a plasma chamber for 8min. The compact SnO2 layer was formed via spin-coating diluted SnO2 colloid precursor (diluted in water, v/v 1:3) onto the pre-cleaned FTO glass substrates, at 3000rpm for 30s and then annealed at 150C for 30min. The perovskite films were deposited on top of the substrate (glass/FTO/compact SnO2 or glass/FTO/compact TiO2/mesoporous TiO2). The HTM layer was prepared by dissolving 2,2′,7,7′-tetrakis(N,N-dip-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD), lithium Li-TFSI and 4-tert-butylpyridine (4-tBP) in chlorobenzene and spin-coating the solution onto the perovskite layer. At last, an Au layer with a thickness of 100nm was coated to form metal electrodes via thermal evaporation.
Characterization. X-ray diffraction spectra were obtained with X-ray diffraction (Bruker D8-Advance, Cu Kα radiation). The field-emission SEM (Hitachi S-4300) was employed to acquire SEM images, the electron beam was accelerated at 10kV. Thermogravimetry analysis was conducted by TG-DTG (NETZSCH, STA-449-C) in the nitrogen atmosphere. UV-vis absorption measurements were measured at room temperature through UV-vis spectrophotometer (PerkinElmer Lamuda365). Femtosecond transient absorption spectroscopy was conducted by using an amplified Ti:Sapphire laser source (Coherent Legend, 800nm, 100fs, 2mJ per pulse, and 1kHz repetition rate). The laser output was split into two parts; the stronger beam was frequency doubled to generate 400nm excitation light, and the other one was focused into a 1mm thick sapphire plate to generate broad-band white-light probe pulses. The TA data were detected by a spectrometer at various time delays between the pump and probe pulses. The cross-sectional TEM image was prepared in a focus ion beam system (FEI HELIOS NanoLab 600i FIB). The surface roughness and phase were characterized by an AFM (KEYSIGHT Technologies 7500) with a Pt-coated conductive cantilever probe (Bruker, Model: SCM-PIT-V2) for KPFM technique. Fourier transform infrared spectroscopy (FTIR) spectra were collected by a Perkin-Elmer Spectrum GX FTIR spectrometer.
Steady-state PL and time-resolved PL measurements were conducted on a time correlated single photon counting (TCSPC) system (FluoTime 300, PicoQuant). The TRPL decay profiles were recorded at 768nm for all of the samples with a probing width of 5nm upon excitation by a 510nm laser (LDH-P-C-510, PicoQuant GmbH). X-ray Photoelectron Spectroscopy (XPS) was used to examine the presence of oxygen inside films etched by Ar + plasma (ThermoFisher ESCALAB 250Xi). The vacuum degree of the analysis chamber is about 5×10 -9 mbar. The space-charge-limited current (SCLC) measurements were carried out by a Keithley 2400 digital source-meter under dark condition. The J-V curves of the PSCs devices were measured using the Keithley 2400 series digital source-meter unit under simulated AM 1.5G irradiation (100mW cm -2 , xenon-lamp, Newport). The light intensity was calibrated by the standard reference of a Newport Si solar cell before measurements. The mask with a square aperture (0.09 cm 2 ) close to the top of the PSCs device was used to define the effective area. The J-V curves were measured from forward bias to reverse bias; the scan rate was 100mV/s. External quantum efficiency (EQE) was measured using a measurement system (model QE-R, Enli Technology Co., Ltd.). The devices testing were carried out in ambient air.   Figure S2a'-d' and a"-d" show the scanning electron microscope (SEM) images of the resulting films from 8 different processing conditions, from which we make the following observations. First, introduction of nitrogen flow during spin-coating improves the uniformity and crystallization of the resulting film, due to the super-saturation of the perovskite components as a result of fast volatilization of solvent in the wet film. This allows one to obtain uniform perovskite films without using highly toxic anti-solvents such as chlorobenzene, toluene, diethyl ether, which were often involved in lead halide routes. [1,2] Second, adding DMSO offers greater control of the crystal growth rate through improved solubility of the lead source, [1] resulting in high-quality perovskite films free of pin-holes ( Figure S2a"-d"). Third, post-annealing improves the film coverage, as the elevated temperature provides additional thermal energy to assist the crystallization of the perovskite material, and further enhances the efficiency of solvent removal ( Figure  S2d"). We further characterized the composition of the resulting films using Fourier transform infrared spectroscopy (FTIR), which provides information about the vibrational states of the molecules. ( Figure  S3e) shows the FTIR data of samples in (Figure S3 a-d) along with corresponding optical images. Films with nitrogen or thermal treatments exhibit dark colour (Figure S3 b-d), consistent with the two IR peaks around 3200~3450 cm -1 corresponding to the N-H stretching vibration ( Figure S3e), suggesting formation of perovskite phase. [2] In contrast, samples without these treatments show yellowish colour ( Figure S3a) and no such IR peaks, indicating absence of perovskite phase. Furthermore, we found that the S=O vibration peak at 1020 cm -1 ( Figure S3f), indicative of PbI2DMSO adduct and thus incomplete reaction, 1 can be significantly reduced by flowing nitrogen, and completely eliminated after annealing. As control, we also show data for sample without DMSO additive (dashed lines in Figure S3        For better understanding the spontaneous formation of PbI2 at GBs for such non-halide source, we also fabricate the perovskite film from another similar acid-based lead source, lead acetate (Pb(CH3COO)2, Pb(Ac)2) via the same route for comparison. As expected, small crystals also show up at GBs of perovskite films ( Figure S9 a-b). However, the distribution of these small crystals at GBs shows great difference between non-halide lead sources Pb(HCOO)2 and Pb(Ac)2 (Figure 2 a-b and Figure S9 a-b). The small crystals in white colours tightly and orderly wrap around the GBs of the perovskite via Pb(HCOO)2 route while it intersperses loosely at GB of perovskite via Pb(Ac)2 source. We suspect that the difference in crystal distribution is the result of non-halide anions induction，previous studies also indicated that some lead anions such as CH3COO -, Cland SCNhave a special contribution to crystal growth and film morphology. [3,4,5] For Pb(HCOO)2 route, the intermediate of MACOOH as an ionic liquid has an extremely low vapor pressure, [6] it remains in the perovskite film and retards the rate of crystal growth after the solvent evaporated, but the similar ionic liquid cannot be formed spontaneously in Pb(Ac)2 route. Therefore, excess cations of lead salts are evenly squeezed to GBs rather than dispersedly agglomerated.    We applied the charge carrier recombination-diffusion model involving surface recombination velocities (SRVs) to fit the PL decay measurement results, based on our previous work [29] . Especially, the SRVs reflect the defect density at grain surfaces. With no consideration on auger recombination process, the recombination-diffusion model is applied as: where n is the carrier density, D the diffusion constant, k1 and k2 correspond to the widely known monomolecular and germinate recombination rates, respectively. Initial and boundary conditions are given as: In which S0 and SL refers to the SRV at both surfaces of the film, respectively. More information can be found in this article. [29] Based on the above model and the experimental PL decays, the simulation results are shown in Figure  A1. With the listed parameters in Table S3, the simulation results are in good agreement with the measurements.
We can clearly find that the 1:3.15 sample (2=1714ns) possessed significantly reduced defects (S0 and SL) compared to the 1:3.30 sample (2=264ns). The monomolecular recombination rate k1 also shows a reduction in the 1:3.15 sample. We suggest that k1 is associated with the trap-assisted recombination, thus the reduction in k1 also verified the reduced defect density in the 1:3.15 sample.   was conducted under operational conditions in a glove box with N2 atmosphere, (b) the humidity stability of these devices were tested by exposing them to an environmental atmosphere with 45±5% RH, and their J-V curves were collected constantly.