Enhanced perovskite electronic properties via a modi ﬁ ed lead( II ) chloride Lewis acid – base adduct and their e ﬀ ect in high-e ﬃ ciency perovskite solar cells †

Methylammonium lead triiodide (MAPbI 3 ) perovskite solar cells have gained signi ﬁ cant attention with an impressive certi ﬁ ed power conversion e ﬃ ciency of 22.1%. Suppression of recombination at the interface and grain boundaries is critical to achieve high performance perovskite solar cells (PSCs). Here, we report a simple method to improve the performance of PSCs by incorporating a lead chloride (PbCl 2 ) material into the MAPbI 3 perovskite precursor through a Lewis acid – base adduct. The optimal concentration of PbCl 2 that helps increase the grain size of MAPbI 3 with introduction of the ideal amount secondary phases (lead iodide and methylammonium lead tri-chloride) is 2.5% (molar ratio, relative to lead iodide). Examination by steady-state photoluminescence and time-resolved photoluminescence has shown that devices based on MAPbI 3 -2.5% of PbCl 2 facilitated longer charge carrier lifetime and electron – hole collection e ﬃ ciency which is ascribed to reduced defects and concurrent improved material crystallinity. Electrochemical impedance spectra (EIS) of the corresponding PSCs have revealed that, compared to the pristine MAPbI 3 perovskite ﬁ lm, the 2.5% PbCl 2 -additive increased the recombination resistance of the PSCs by 2.4-fold. Meanwhile, measurement of the surface potential of the perovskite ﬁ lms has indicated that the PbCl 2 - additive modi ﬁ es the electronic properties of the ﬁ lm, shifting the fermi-level of the MAPbI 3 ﬁ lm by 90 meV, leading to a more favourable energetic band matching for charge transfer. As a result, the performance of PSCs is enhanced from an average e ﬃ ciency of 16.5% to an average e ﬃ ciency of 18.1% with maximum e ﬃ ciency reaching 19% due to the signi ﬁ cantly improved ﬁ ll-factor (from 0.69 to 0.76), while the hysteresis e ﬀ ect is also suppressed with the PbCl 2 -additive.


Introduction
Perovskites based on organic-inorganic lead halides with a formula of APbX 3 (A ¼ methylammonium (MA), formamidinium (FA); X ¼ I, Br, and Cl) have attracted signicant research interest due to their excellent optoelectronic properties and low-cost fabrication process. 1 Perovskite solar cells (PSCs) which originally adopted the structure of a traditional dyesensitized solar cell using an iodide/triiodide based liquid electrolyte and MAPbI 3 nanoparticles as a light absorber were reported by Kojima et al. in 2009. 1 Following this seminal report, a signicant increase in device performance has been achieved through use of solid state hole transport materials for example, 2,2 0 ,7,7 0 -tetrakis-(N,N-di-4-methoxyphenylamino)-9,9 0 -spirobiuorene more commonly referred to in the literature as Spiro-OMeTAD.The power conversion efficiency (PCE) of solidstate PSCs has soared from 9.7% in 2012 to a validated PCE of 22.1% in early 2016. 3,43][4] Furthermore, these studies have shown that un-reacted PbI 2 is mainly located at the grain boundaries (GBs) of the MAPbI 3 lm potentially passivating GBs and reducing recombination in the perovskite lm, 2 while MAPbCl 3 acts as a template for the crystallization of MAPbI 3 perovskite. 5][8][9] Stranks et al. pointed out that the electron and hole diffusion length of perovskite lms made from the Cl-based precursor can be enhanced from about 100 nm to over 1 mm, 10 indicating a reduced recombination in the bulk perovskite layer or at interfaces between the perovskite layer and selective contacts.Lead chloride (PbCl 2 ) has been used as a Cl-source for preparing Cl-based perovskite lms. 11,12The use of lead chloride (PbCl 2 ) as an alternative to PbI 2 in the perovskite precursor was reported to improve the uniformity and surface coverage of MAPbI 3 , controlling its nucleation. 9Higher concentrations of PbCl 2 in the perovskite precursor induced increased numbers of perovskite crystallites, thus leading to improved lm morphology.Furthermore, Huang et al. added additional PbCl 2 (3 at% extra) to form a non-stoichiometric perovskite precursor with PbCl 2 -: MAI ¼ 1.03 : 3 (molar ratio) and found that the excess PbCl 2 facilitated the solubility of MAI in the perovskite precursor. 13A PCE up to 16.1% was achieved with an average PCE of 13.82% based on PbCl 2 -assisted heterogeneous nucleation crystallization.Nevertheless, the effect of excess PbCl 2 on the nucleation process of MAPbI 3 is still unclear.For instance, if PbCl 2 acts as a heterogeneous nucleation site, then the small increment in the content of PbCl 2 in the original precursor solution (from 1 to 1.03 molar ratio with respect to 3 MAI) actually did not signicantly increase the number of seeds (no more than 3% with respect to the original number of seeds) for crystal growth.
Although the benets of Cl on electronic and morphological properties of MAPbI 3 lms are excellent, the exact location and the amount of Cl as well as the formation of Cl-based products in the nal perovskite lms remain unclear.Colella et al. have suggested that Cl is preferentially located near the perovskite/ TiO 2 interface, which in turns induces band banding and improves charge collection efficiency of TiO 2 . 6Song et al. proposed that the perovskite lm prepared from a mixture of PbCl 2 and MAI consists of MAPbI 3 and secondary phases, e.g., PbI 2 and MAPbCl 3 . 4The amount of these secondary phases strongly inuences the PCE of PSCs.For instance, above the "optimal amount" of secondary phases, solar cells showed lower photovoltage while below that optimized value solar cells show lower photocurrent and thus lower PCE. 4 In mixed halide perovskite lms prepared from stoichiometric MAI : PbCl 2 (3 : 1, molar ratio), the amount of the secondary phases is oen controlled by tuning annealing temperature and time (100 C, 90 minutes) which is somehow unreliable due to the intrinsic thermal instability of MAPbI 3 perovskite. 14t has been established that due to the dewetting or agglomeration process of the as-deposited perovskite lms prepared from a simple one-step spin-coating method, upon annealing, pin-holes and voids are oen observed in the perovskite lm even with inclusion of PbCl 2 in the precursor and careful solution preparation. 126][17][18][19] Among them, the method based on antisolvent-dripping which takes place by sudden inducement of super-saturation in the lm is found to be very effective, 3,15,16 resulting in the formation of a smooth perovskite lm with full surface coverage.Recently, Ahn et al. reported a modied antisolvent-dripping method which can effectively retard the crystal growth process of the perovskite material by forming a Lewis acid-base (MAI-PbI 2 -dimethyl sulfoxide (DMSO)) adduct and make the MAPbI 3 lm with high quality and full surface coverage. 16To date, as reported, this is an effective approach to produce high performance MAPbI 3 solar cells.However, current-voltage hysteresis is still observed in the reported MAPbI 3 perovskite solar cell probably due to an inherent unbalanced electron-hole diffusion length ($130 nm for electrons, $90 nm for holes). 20Here we introduce a PbCl 2 additive in the Lewis acid-base adduct method to fabricate a hysteresis-less high PCE solar cell, which has not been previously reported.Since the impact of PbCl 2 on improving the morphology of the perovskite lm using the Lewis acid-base adduct method is marginal, this allows us to uncover the underlying reasons for the change of device performance by focusing on the modication of electronic properties of the perovskite material.We have found that this simple method successfully establishes effective secondary phases (PbI 2 and MAPbCl 3 ) in the perovskite material while minimizing the adverse effect of annealing.These secondary phases and the incorporation of chloride into the perovskite lattice enable high PCE solar cells with less hysteresis and an improved ll factor.

Device fabrication
Solar cells were fabricated using uorine-doped tin oxide (FTO) coated glass (Nippon Electric Glass, 15 U , À1 ) as the substrate which was rstly patterned through partial removal of FTO via etching using 35.5 wt% HCl and zinc powder.The substrates were then cleaned in sequence in 5% Decon-90 detergent, and a mixture of acetone, isopropanol and ethanol for 20 min each in an ultrasonic bath.Prior to use, the substrates were treated with ultraviolet ozone for 30 min to fully remove organic solvent residuals.An electron transporting layer based on the TiO 2 lm ($40 nm) was deposited in air via spin-coating a 0.15 M solution of titanium diisopropoxide bis(acetylacetonate) in 1-butanol at 2000 rpm for 20 s.The lm was then dried at 125 C for 5 min and annealed at 450 C for 30 min.A mesoporous TiO 2 (mp-TiO 2 ) layer ($200 nm) was spin-coated onto the compact TiO 2 lm using a diluted TiO 2 paste (0.12 g TiO 2 paste (Dyesol) in 1 mL of absolute ethanol) at 2000 rpm for 20 s, followed by sintering at 450 C for 30 min.Aer cooling to room temperature, the lm was treated with 20 mM TiCl 4 aqueous solution at 90 C for 10 min.The TiCl 4 -treated lm was cleaned with distilled water and annealed again at 450 C for 30 min.Aer this, the TiO 2 coated lm was treated in a UV-zone for 20 min before being transferred to an Ar-lled glovebox.Perovskite layers ($400 nm) with and without PbCl 2 in the perovskite precursor solution were deposited onto the prepared TiO 2 layer at 4000 rpm for 20 s.During spin-coating, 0.5 mL of diethyl ether was dropped on the center of the spinning substrate before it turned turbid.The perovskite layer was then dried at 65 C for 1 min, and annealed at 100 C for 2 min. 16The hole-transport layer ($200 nm) was deposited from the prepared Spiro-OMeTAD solution onto the as-prepared perovskite layer at 4000 rpm for 25 s.The device fabrication was nished by deposition of a 100 nm layer of gold lm for back contact on the prepared sample via an e-beam evaporation process under 10 À6 torr pressure.

Characterization
The top-view and cross-sectional scanning electron microscopy (SEM) images of the samples were taken using a eld emission scanning electron microscope (FSEM JOEL 7001F) at an acceleration voltage of 5 kV.The UV-vis absorbance spectrum was measured with a UV-visible spectrometer (Cary 50).The crystal structure of the perovskite lm as-deposited on FTO/compact TiO 2 /mp-TiO 2 was determined by X-ray diffraction (Rigaku SmartLab) with monochromatic CuKa (l ¼ 0.154 nm) as an excitation source.A scan rate of 1.5 per minute and a step size of 0.015 were used in the XRD measurement.The performance of perovskite devices was measured under irradiation of 100 mW cm À2 (AM1.5)provided by a solar simulator (Oriel Sol3A, Newport) equipped with a 450 W Xenon lamp.IPCE measurement was conducted by using a quantum efficiency system (IQE 200B, Newport) in AC mode.Electrochemical impedance spectroscopy (EIS) of the PSCs was performed in a frequency range from 1 MHz to 100 mHz using an electrochemical workstation (VSP BioLogic Science Instruments) at a forward bias of 0.5 V in darkness.An AC voltage with a perturbation amplitude of 10 mV was applied in the EIS measurement.X-ray photoelectron spectroscopy (XPS) (Kratos Axis Ultra) using mono Al Ka (1486.6 eV) X-rays was used to detect the elements in the perovskite lm.For XPS depth proling, a 4 keV Ar + ion was used for the charge-up effect.The photoluminescence (PL) spectrum was measured with a uorescence spectrometer (Edinburgh Instruments Ltd) at room temperature.The lm was photoexcited by using a laser (474 nm) with a pulse wavelength of 82.4 ps.Scanning Kelvin Probe Force Microscopy (KPFM) (Oxford instrument, Asylum Research) was performed on the prepared perovskite lm under ambient conditions using a NSG-03 Pt coated cantilever at room temperature.The work function of the cantilever was measured using a HOPG standard sample.The chloride analysis was performed on a Dionex RFIC ICS-2100 Ion Chromatography (IC) system (Thermo Scientic) with an ASDV auto sampler system.A Dionex-IonPac AS18 4 mm and a Dionex EGC-KOH II Cartridge were used as the column and eluent, respectively.The suppressor (ASRS 300, 4 mm) operated at 82 mA.The calibration standards were diluted from an IC standard (containing 1000 mg mL À1 of Cl) purchased from Choice Analytical.Perovskite/FTO samples (2 Â 1.5 cm) were immersed into 2 mL of Milli-Q water in a clean beaker until the perovskite layer was fully removed.The amount of dissolved perovskite material in water was used for further calculation.1.2 mL of the resulting solution was pipetted out and further diluted ve fold for IC measurements.

Results and discussion
Depicted in Fig. 1(a)-(e) are the top-view SEM images of perovskite layers with various concentrations of PbCl 2 in the precursor solution (from 0% to 10% with respect to the content of PbI 2 ).The MAPbI 3 lm consists of grains with the size of 100-300 nm which are connected closely, resulting in a pinhole-free and highly compact lm (Fig. 1a), which is desirable for PSCs.When only 1% PbCl 2 was added into the perovskite precursor, there is no observable change in the lm morphology in terms of grain size and lm compactness (Fig. 1b) compared to the pure MAPbI 3 lm.This indicates that a negligible amount of PbCl 2 in the perovskite precursor does not dramatically change the surface morphology of the perovskite lm in our experiments.This phenomenon is different from what has been reported previously where a small amount of PbCl 2 ($1%) signicantly improved surface coverage of the perovskite lm. 13t is ascribed to the effectiveness of the Lewis acid-base adduct method which results in high quality MAPbI 3 perovskite lms. 16,21However, when the PbCl 2 -additive increased to 2.5%, larger grains with the crystallite size in the range of 300-500 nm are observed in the perovskite lm (Fig. 1c).This effect could be attributed to a better crystallization of the perovskite layer, which leads to larger perovskite crystals in the presence of PbCl 2 .Probably MAPbCl 3 is formed rst and then acts as a template to aid grain growth. 5To gain more insight into the inuence of PbCl 2 on the morphological properties of the perovskite lm, the concentration of PbCl 2 in the MAPbI 3 precursor was further increased to 5%, 7.5%, and 10%, respectively.As shown in Fig. 1(d)-(f) under the same deposition and annealing conditions, there is no further increase in grain sizes of the perovskite material.Nevertheless, small voids/ pinholes can be seen clearly at the GBs on the surface of perovskite lms with precursor solutions in excess of 5% PbCl 2 .The number of voids/pinholes increases with the increase of PbCl 2 concentration.It is known that voids at the GBs can increase recombination in the device due to poor contact/ necking between the light absorbing particles, thus impacting device performance.Formation of voids in lms with over 5% PbCl 2 in the perovskite precursor might be a consequence of the different nucleation rates and competitive growth of nuclei MAPbI 3 , MAPbCl 3 and PbI 2 , which somehow affects the surface coverage of the lm. 9Furthermore, irregular grain sizes are also observed in the perovskite lm with 10% PbCl 2 in the precursor solution (Fig. 1f), which could be due to the coexistence of MAPbI 3 and methylammonium lead chloride (MAPbCl 3 ) as conrmed by XRD (see below).The above SEM results indicate that 2.5% PbCl 2 provides the optimal quality of surface morphology of the perovskite lm.
Since the optimal light absorption and energy bandgap are very critical for the performance of a solar cell, the light absorption properties of the MAPbI 3 lm made from different contents of PbCl 2 in the precursor were investigated by ultraviolet-visible (UV-vis) spectroscopy (Fig. S1 †).However, negligible change is found in the onset light absorbing wavelength and the intensity of light absorption of the perovskite layers with the content of PbCl 2 below 5%.A slight blue-shi in the UV-vis spectrum is observed with the lm made from a higher content of the PbCl 2 additive (above 5%), which could be related to incorporation of chloride into the MAPbI 3 crystal lattice, which reduces the lattice symmetry and thus increases the material bandgap 22 and the coexistence of a small amount of MAPbCl 3 .The reduction in intensity of light absorption of the lm with a higher content of PbCl 2 is associated with the existence of more voids as observed in Fig. 1(d)-(f) and possible secondary impurities such as PbI 2 and MAPbCl 3 in the lm as indicated in the XRD (below).
To gain insight into the effects of the PbCl 2 -additive on the crystal structure of the perovskite material, X-ray diffraction (XRD) data were collected for the diffraction patterns of the lms (Fig. 2).From the data, it is found that when the concentration of PbCl 2 is less than 5%, all XRD peaks can be well indexed to the reection of MAPbI 3 without any impurities detected within the instrumental limit of the XRD facility.However, a small peak at 12.8 which is ascribed to the (001) plane of PbI 2 is observed when the amount of PbCl 2 -additive exceeds 5%.For the lm with 10% PbCl 2 , a diffraction peak at 15.6 which belongs to the reection of the (110) plane of methylammonium lead tri-chloride (MAPbCl 3 ) is also observed.It is worth noting that no peak from PbCl 2 is observed in all the XRD spectra, indicating that the added PbCl 2 is completely consumed in the reaction with MAI to either form PbI 2 or MAPbCl 3 .This is in good agreement with a previous study. 16A moderate excess of PbI 2 in the MAPbI 3 perovskite lm which mainly exists at GBs is benecial because it forms an energy barrier that hinders leakage of both electrons and holes from MAPbI 3 for recombination. 23However, a large amount of PbI 2 in the perovskite lm is detrimental for light harvesting due to its much larger bandgap (2.41 eV for PbI 2 ) and low light absorption, which in turns reduces the current density of devices. 2 Furthermore, a small shi of the XRD patterns towards the smaller angle of MAPbI 3 was observed with the presence of PbCl 2 in the precursor (Fig. 2(b)-(d)).For example, by comparison of the XRD patterns, it is found that peaks at 14.14 ((110) plane, Fig. 2b), 28.47 ((220) plane, Fig. 2c) and 31.9 ((310) plane, Fig. 2d) for the pristine MAPbI 3 shi to 14.16 ((110) plane), 28.50 ((220) plane) and 31.93 ((310) plane) for the MAPbI 3 -2.5% PbCl 2 additive.The shis is slightly larger than the XRD measurement step (0.015 ), suggesting the effect of inclusion of Cl on the crystal lattice of MAPbI 3 .Furthermore, the trend illustrating that the more PbCl 2 in the MAPbI 3 precursor leads to the more shi of the XRD patterns towards the smaller angle of MAPbI 3 is observed, conrming that a certain amount of Cl might be incorporated into the crystal lattice of MAPbI 3 .
We have attempted to determine the chlorine content in the MAPbI 3 lms made from the PbCl 2 additive based precursor by a conventional method such as X-ray photoelectron spectroscopy (XPS) (Fig. S2 †).Nevertheless, no characteristic peak of chlorine was detected in the lm even in the XPS depth proling spectrum (Fig. S2 †).The failure of XPS for detection of Cl in perovskite lms could be due to the strong photon/ion energy, which destroys volatile Cl-based species.Unlike common techniques such as EDS and XPS, ion chromatography (IC) is a highly accurate non-destructive testing method, which guarantees that the measurement does not lose Cl-based species.The chloride content in the nal perovskite lms prepared from different concentrations of PbCl 2 in the perovskite precursor solutions was assessed by IC as demonstrated in Table 1.It reveals that Cl remains in perovskite lms aer a mild annealing (2 min, 100 C) which is consistent with previous reports. 5,24The preservation of Cl in our annealed perovskite lms could be due to the high compactness of the as-deposited perovskite lm which slows down the rates of diffusion of Cl to the surface and prevent its loss aerwards. 25The exact forms of Cl-based products are beyond the scope of this study; we speculate that the remaining chloride could incorporate into the MAPbI 3 lattice, at the GBs, interfaces, or in MAPbCl 3 form.The induced MAPbCl 3 is not detected in the XRD pattern in low concentration PbCl 2 -based perovskite lms (below 10%), which is presumably because of the poor crystallinity of the MAPbCl 3like form. 5,26Furthermore the unreacted PbI 2 passivates GBs which generally worked as recombination centers. 2 To explore the benets of the PbCl 2 -additive in the perovskite precursor on the performance of solar cells, solar cells based on different PbCl 2 contents were fabricated and the comparison of J sc , V oc , FF and PCE is summarized in Fig. 3(a)-(d).It is found that the PbCl 2 -additive does not signicantly change the V oc and J sc of the solar cells with the content of the PbCl 2 -additive up to 2.5%.Beyond this, J sc decreases marginally which is ascribed to reduced light absorption associated with formation of PbI 2 impurity and increased voids/pinholes.The most remarkable change is observed with respect to the FF of the solar cells.When the PbCl 2 content is increased from 0% to 2.5%, the FF is improved from 0.69 to 0.76 primarily as decreased series resistance (R s ) and increased shunt resistance (R sh ) (Table 2).The decrease in R s is presumably attributed to the reduction of GBs as observed in Fig. 1(a)-(c), while the increase in R sh corresponds to improved perovskite lm quality and more efficient interfacial charge transfer between MAPbI 3 and TiO 2 lm.The increase of PbCl 2 concentration to 5% led to an increased R s and reduced R sh , which could be due to the occurrence of more pinhole/voids and impurities as discussed above.With further increasing the PbCl 2 content to 7.5%, devices exhibited higher R s , yet higher average R sh .However, an extremely large variation was observed, which is possibly related to the non-uniformity of the lm due to the distribution of pinhole/voids and a large amount of impurities (PbI 2 and MAPbCl 3 ) in the perovskite lm.Thus the devices with the 2.5% PbCl 2 additive result in the maximum FF, which is in line with PCEs obtained.As a result, the best performance devices were obtained with the 2.5% PbCl 2 -additive in the precursor solution with an average J sc of $23.5 mA cm À2 , V oc of $1.04 V and FF of $0.75, leading to an average efficiency of 18.1% (based on a batch of 10 cells), which is much higher than that of devices without the PbCl 2 -additive which shows an average PCE of 16.5% (J sc of $23.5 mA cm À2 , V oc of $1.02 V, and FF of $0.69).The best performance of the 2.5% PbCl 2 -additive based cell produced an efficiency of 19% with details shown in the following section.The detailed cross-sectional SEM structure of a representative solar cell made from the precursor containing the 2.5% PbCl 2 -additive in MAPbI 3 solution is depicted in Fig. 4a.As depicted, the perovskite layer is homogeneous with large crystals which connect closely and seem to form a monolithic layer from the mesoporous bottom layer to top, suggesting a low density of GBs.
The J-V curves of the MAPbI 3 perovskite solar cells with and without 2.5% of PbCl 2 in the perovskite precursor measured under reverse (from V oc to J sc ) and forward (from J sc to V oc ) scanning are provided in Fig. 4b.The MAPbI 3 -based solar cell produces a PCE of 17.1 (15.1)%, with a V oc of 1.03 (0.98) V, a J sc of 23.7 (23.8) mA cm À2 , and a FF of 70 (65)% when measured under reverse (forward) scan, respectively, revealing a hysteresis (H) of 11.7% (H ¼ 100% Â (PCE rev À PCE fw )/PCE rev ).In contrast, the MAPbI 3 -2.5% PbCl 2 -based solar cell achieved a PCE of 18.9 (17.2)%, a V oc of 1.05 (1.03) V, a J sc of 23.7 (23.9) mA cm À2 , and a FF of 76 (70)% when measured under reverse (forward) scan, respectively, with only 8.9% hysteresis.
The external quantum efficiency (EQE) spectra of the MAPbI 3 -based and MAPbI 3 -2.5% PbCl 2 -based device exhibit a high EQE value between 80% and 86% across a broad wavelength range from 400 nm to 750 nm with the photon response spectral tail extending to around 800 nm as illustrated in Fig. 4c.The integrated current densities are 20.1 mA cm À2 for the device with MAPbI 3 and 20.8 mA cm À2 for the device with MAPbI 3 -2.5% PbCl 2 , which are slightly lower than those measured from J-V curves.This is ascribed to the presence of shallow trap states in the perovskite lms.Shallow trap states in perovskite and/or interfaces can be easily lled under illumination during the J-V testing through very short light soaking (10 seconds in our case), which in turn facilitates charge carrier transport.However, the IPCE measurement was performed under very low intensity of monochromatic light without bias illumination.The electron generated in the IPCE measurement rst needs to ll the trap states before being collected. 27This leads to a minor discrepancy in integrated J sc from the IPCE spectrum as compared to the J sc measured by constant steady-state illumination of the solar simulator.
The steady-state performance of the two representative cells (MAPbI 3 and MAPbI 3 -2.5% PbCl 2 ) at the maximum power point (MPP) was also measured (Fig. 4d).Both devices were preconditioned by illumination for 10 s prior to the measurement.The current densities at the MPP drop within a few seconds from 20.1 mA cm À2 to 19.1 mA cm À2 for the MAPbI 3 -based solar cell and from 20.4 mA cm À2 to 19.7 mA cm À2 for the MAPbI 3 -2.5% PbCl 2 -based solar cell under a continuous measurement duration of 100 s.The applied voltage bias near the MPP is 0.85 V for the MAPbI 3 -based device and 0.89 V for the MAPbI 3 -2.5% PbCl 2based device, which has a corresponding stabilized power output of 16.2% and 17.5%, respectively.This result suggests that addition of 2.5% PbCl 2 into the perovskite precursor also dramatically improves the steady-state efficiency of PSCs.
The steady-state photoluminescence (PL) emission and timeresolved photoluminescence (TRPL) were carried out to elucidate the charge carrier dynamics of MAPbI 3 and MAPbI 3 -2.5% PbCl 2 lms with and without the presence of charge transport layers (TiO 2 and Spiro-OMeTAD) (Fig. 5(a)-(d)).Compared to the pristine MAPbI 3 perovskite lm, the intensity of the PL peak of the MAPbI 3 -2.5% PbCl 2 lm increases while the full width half maximum (FWHM) remains constant (Fig. 5a), indicating that more electron-hole pairs in the lm lead to stronger radiative recombination.This indicates improved crystallinity of the asdeposited lm. 28Moreover, the steady-state PL spectrum of the MAPbI 3 -2.5% PbCl 2 lm in Fig. 5a depicts a blue-shi by 2 nm (from 769 nm to 767 nm) which is consistent with the blue-shi observed in UV-vis absorption spectra acquired for the lms, further suggesting the incorporation of chloride into the MAPbI 3 lattice.On contacting with charge transport layers,  a considerable decrease in PL intensity was observed.Compared to MAPbI 3 , the drop of PL peaks of MAPbI 3 -2.5% PbCl 2 is more signicant when in contact with the TiO 2 lm, suggesting a more efficient interfacial charge transfer.This is in good agreement with the enhanced performance of MAPbI 3 -2.5% PbCl 2 -based PSCs compared to pristine MAPbI 3 devices.
The TRPL spectra of samples are tted using a bi-exponential function of time (t) (F(t) ¼ P a i e Àt/s i , i ¼ 1, 2), which contains a fast decay time component and a slow decay time component. 29The fast decay component is associated with recombination behavior at the surface, while the slow decay component corresponds to recombination in the bulk of the perovskite lm. 30The result indicates that, compared to pristine MAPbI 3 , addition of PbCl 2 into the perovskite precursor results in lms with a marginal decrease in the fast decay component (from 2.1 ns to 1.44 ns) but longer in the slow component of lifetime (from 80 ns to 85 ns) (Fig. 5b).Therefore the overall average PL lifetime of the perovskite layer is increased from 75 ns to 80 ns.
Time-resolved photoluminescence measurements of the MAPbI 3 show a decrease in the PL lifetime of the perovskite layer from 75 ns to 56.7 ns in the presence of TiO 2 (40 nm compact TiO 2 /200 nm mesoporous TiO 2 ).When in contact with Spiro-OMeTAD, the lifetime ¼ 45 ns is obtained.Whereas the PL lifetime of the MAPbI 3 -2.5% PbCl 2 layer is reduced more signicant from 80 ns to only 49.5 ns and 26.3 ns, respectively, when in contact with the TiO 2 lm and with Spiro-OMeTAD layer, indicating a faster and more effective electron and hole transport.The more effective hole transport is crucial for MAPbI 3 perovskite due to the unbalanced electron-hole diffusion length ($130 nm for electrons, $90 nm for holes) according to a previous report. 20These results indicate improved charge collection efficiency of MAPbI 3 -2.5% PbCl 2based PSCs compared to that of pristine MAPbI 3 , which is consistent with the improved FF of devices fabricated in this study.
To further explore the possible origin of the superior optoelectronic properties of the MAPbI 3 -2.5% PbCl 2 based solar cell to the pristine MAPbI 3 based solar cell, electrochemical impedance spectroscopy (EIS) measurements were conducted and the corresponding Nyquist plots are depicted in Fig. 6.The intercept point at the real part of the Nyquist plots is associated with series resistance of the device (R s ), which is mainly contributed to the sheet resistance of FTO glass.The small semicircle at high frequency corresponds to the charge transfer process between perovskite and selective contacts while the large semicircle at low frequency demonstrates the recombination of charge carriers in the perovskite layer (according to previous reports). 31,32Fitting the EIS of the MAPbI 3 and MAPbI 3 -2.5% PbCl 2 solar cells using an equivalent circuit depicted in the inset in Fig. 6 indicates that both devices have comparable interfacial charge transport behavior.Nevertheless, the MAPbI 3 -2.5% PbCl 2 based solar cell has nearly 2.4-times higher resistance for recombination (1.68 Â 10 5 U) than the pristine MAPbI 3 solar cell (6.93 Â 10 4 U), whereas both devices exhibit similar capacitance (0.789 mF for MAPbI 3 versus 0.769 mF for MAPbI 3 -2.5% PbCl 2 ).As a result, the electron lifetime, a product of resistance and capacitance (s ¼ R Â C), of the solar cell with the 2.5% PbCl 2 -additive in the precursor (s r ¼ 0.12 s) is over 2-fold higher than the device made from pure MAPbI 3 (s r ¼ 0.055 s).The increased carrier lifetime is one of the factors that are responsible for the higher performance of the solar cell.
Kelvin probe force microscopy (KPFM) is a powerful technique not only to examine surface topography but also the work function of the surface.The KPFM topography of both pristine MAPbI 3 and MAPbI 3 -2.5% PbCl 2 is shown in Fig. 7a and c.It appears that the surface of both perovskite lms is uniform over a large area (5 Â 5 mm 2 ).However, the lm made from the 2.5% PbCl 2 based precursor consists of larger grains.The root mean square as an indicator of roughness of the lm is 15.75 nm for the PbCl 2 -additive lm and 14.27 nm for the pristine MAPbI 3  lm.The slightly rougher surface is attributed to its larger grain size.Measurement of the surface potential spectra of MAPbI 3 and MAPbI 3 -2.5% PbCl 2 lms (Fig. 7b and d) reveals an increase of mean values of contact potential difference (CPD) by $90 mV in the MAPbI 3 -2.5% PbCl 2 lm compared to the MAPbI 3 lm.It is known that the conduction band of MAPbI 3 perovskite is $80 meV more negative than that of n-type TiO 2 , which in principle forms an energy barrier for electron injection from MAPbI 3 to n-TiO 2 during the charge transport process. 33herefore, the increase of the electron quasi-Fermi level near the conduction band edge of MAPbI 3 -2.5% PbCl 2 favors improved energy level alignment with adjacent TiO 2 for electron transfer at the MAPbI 3 -2.5% PbCl 2 /n-TiO 2 interface.This work provides new insight into the mechanism that governs the impact of PbCl 2 on both morphology and electronic properties of the MAPbI 3 lm for high energy conversion efficiency.
The stability of un-encapsulated perovskite solar cells based on MAPbI 3 with and without 2.5% PbCl 2 was tested by aging the device at the maximum power point (MPP) under constant illumination (100 mW cm À2 ) at room temperature in ambient air (RH% ¼ 50-60%).It is found that both devices retained $65% power conversion efficiency aer 80 minutes stability testing (Fig. 8).The above results suggest that the impact of the PbCl 2 additive on the stability of the perovskite solar cells is negligible in this work.

Conclusions
We have demonstrated the effect of PbCl 2 on the morphology, crystallinity and recombination of the MAPbI 3 perovskite lm that was prepared via a one-step Lewis acid-base adduct method.Although addition of a low content of PbCl 2 (less than 2.5%) into the MAPbI 3 precursor did not pose a signicant change in the crystal structure and optical properties, the resulting lms show a larger grain size.More importantly, the PbCl 2 additive improved the electronic properties and dynamics of charge carriers in the MAPbI 3 lm.The electronic properties of the perovskite lm were modied by the PbCl 2 additive with a signicant downward shi of the surface work function by 90 meV, which facilitates electron injection from perovskite to TiO 2 .Analysis of TRPL indicated that the PbCl 2 -additive in the precursor led to perovskite lms with a more balanced charge (electron and hole) collection and longer carrier lifetime.The EIS results of the corresponding PSCs also conrm that devices with the 2.5% PbCl 2 additive have a 2.4-fold higher resistance for recombination compared to that in the pristine MAPbI 3based solar cell, indicating reduced trap density, resulting in a lower observed current-voltage hysteresis of solar cells.As a result, the performance of PSCs was increased from an average performance of 16.5% with pristine MAPbI 3 to 18.1% with MAPbI 3 -2.5% PbCl 2 .The nearly 9% enhancement in device performance is mainly due to the improvement of the FF from 0.69 to 0.76.

Fig. 3
Fig. 3 Statistical parameters of (a) V oc , (b) J sc , (c) FF, and (d) PCEs measured under reverse scanning (V oc , to J sc ) for 10 cells using MAPbI 3 with varied contents of PbCl 2 in the precursor.

Fig. 4
Fig. 4 (a) Cross-sectional SEM image of the completed MAPbI 3 + 2.5% PbCl 2 -additive solar cell; (b) J-V curves of perovskite solar cells using MAPbI 3 with and without 2.5% PbCl 2 in the perovskite precursor, under reverse and forward voltage scan; (c) corresponding IPCE spectra; and (d) steady-state photocurrent measured at a bias voltage (0.85 V for the MAPbI 3 device and 0.89 V for the MAPbI 3 + 2.5% PbCl 2 device) at maximum power point and stabilized power output 10 s pre-conditioned to 1 sun light illumination.

Fig. 6
Fig. 6 Nyquist plots of PSCs based on pristine MAPbI 3 perovskite (square black) and fitting (green), and MAPbI 3 -2.5% PbCl 2 perovskite (dot red) and fitting (blue) in the dark at 0.5 V forward bias.The inset depicts the corresponding equivalent circuit.

Fig. 8
Fig. 8 Relative performance of perovskite solar cells based on MAPbI 3 with and without the 2.5% PbCl 2 additive under continuous illumination for 80 minutes in ambient air with relative humidity: 50-60%.

Table 1
Concentration of chloride in the final perovskite films prepared by different contents of PbCl 2 in the perovskite precursor Content of PbCl 2 in the perovskite precursor (%, molar ratio relative to MAPbI 3 ) Content of Cl À ions in the perovskite precursor (%, molar ratio relative to MAPbI 3 ) Content of Cl À ions in the nal perovskite lms (%, molar ratio relative to MAPbI 3 )