A universal ligand for lead coordination and tailored crystal growth in perovskite solar cells

Chemical environment and precursor-coordinating molecular interactions within a perovskite precursor solution can lead to important implications in structural defects and crystallization kinetics of a perovskite film. Thus, the opto-electronic quality of such films can be boosted by carefully fine-tuning the coordination chemistry of perovskite precursors via controllable introduction of additives, capable of forming intermediate complexes. In this work, we employed a new type of ligand, namely 1-phenylguanidine (PGua), which coordinates strongly with the PbI2 complexes in the perovskite precursor, forming new intermediate species. These strong interactions effectively retard the perovskite crystallization process and form homogeneous films with enlarged grain sizes and reduced density of defects. In combination with an interfacial treatment, the resulted champion devices exhibit a 24.6% efficiency with outstanding operational stability. Unprecedently, PGua can be applied in various PSCs with different perovskite compositions and even in both configurations: n–i–p and p–i–n, highlighting the universality of this ligand.


Synthesis of PGua
Aniline (1.86 g, 20.0 mmol, Sigma-Aldrich) was dissolved in hydrochloric acid solution (36.5% in water, 4 mL), cyanamide (2.1 g, 50 mmol, Sigma-Aldrich) was added into the solution, then the mixture was heated to 85 °C for 12h.Na2CO3 solution (10% in water, 20 mL) was dropped into the mixture, a white solid was precipitated.The solid was filtered to get phenylguanidine (2.2 g, 81 % yield) with further purification.

Substrate
Fluorine doped tin oxide (FTO) substrates (NSG-10) were chemically etched by zinc powder and 4 M HCl solution and sonicated in 2% Hellmanex water solution for 30 min, acetone for 15 min and ethanol for 15 min, respectively.Then, all substrates were further cleaned by UV-Ozone for 15 min.Then, a compact TiO2 layer was deposited on cleaned FTO substrates via spray pyrolysis deposition from a precursor solution of titanium diisopropoxide bis(acetylacetonate) (Sigma-Aldrich) in anhydrous ethanol (Acros), with oxygen as carrier gas.Substrates were heated at 450 o C and kept at this temperature for 15 min before and 30 min after the spray of the precursor solution, then left to cool down to room temperature.Mesoporous TiO2 layer was spin-coated at 4000 rpm for 20 s, with the acceleration rate of 2000 rpm/s, using a 30 nm TiO2 paste (Dyesol 30 NR-D) diluted in ethanol with 1:6 volume ratio.After the spin-coating, the substrates were dried at 80 ℃ for 10 min and then sintered at 450 o C for 30 min under dry air flow.

Perovskite layer
The perovskite precursor solution was prepared by dissolving a mixture of cesium iodide (0.075 mmol, TCI Co. Ltd.), methylammonium bromide (0.15 mmol, Dyenamo), formamidinium iodide (1.1275 mmol, Dyenamo), lead iodide (1.575 mmol, Alfa Co. Ltd.) in 1 mL mixture of DMF and DMSO (DMF:DMSO=4:1 v/v, Acros).For the MAPbI3•0.05PbI2perovskite precursor, methylammonium iodide (1.4 mmol, Dyenamo), lead iodide (1.47 mmol, Dyenamo) were dissolved in 1 mL mixture of DMF and DMSO (DMF:DMSO=4:1 v/v, Acros).For the MAPbBr3•0.05PbBr2perovskite precursor, methylammonium bromide (1.4 mmol, Dyenamo), lead bromide (1.47 mmol, Dyenamo) were dissolved in 1 mL DMSO (Acros).The PGua was dissolved in the precursor solution with different concentration.The perovskite solution was spin-coated through two-step program (1000 rpm for 10 s and 6000 rpm for 20 s) with pouring chlorobenzene as an anti-solvent 5s before the end of the second step.Then the substrates were annealed at 100 o C for 40 min in dry air.The MA vapor-treatment was employed to form high-quality MAPbBr3•0.05PbBr2perovskite films according to the literature. 1 The CEAI was dissolved in IPA (5 mg/mL) and the solution was spin-coated at 4000 rpm for 20s on the as-prepared perovskite films and dried on a hot plate at 100 o C for 10 min.The substrates were cooled down to room temperature after annealing the perovskite.

Preparation of triple cation perovskite precursor for p-i-n PSCs:
For 1.45M of perovskite precursor solution, Cesium Iodide (Sigma-Aldrich), Methylamonium Bromide (Dyenamo), Formamidinium Iodide (Dyenamo) and Lead Iodide (PbI2, TCI) are being weighed in one vial.The final stoichiometry corresponds to Cs0.05FA0.85MA0.10Pb(I2.9Br0.1)with 5% of PbI2 excess.DMF (Sigma-Aldrich) and DMSO (Sigma-Aldrich) used as solvents with ratio is 4:1.The precursor is stirred for 1 hour at 40°C.After the complete solution of the precursor 0.25 mg of PGua is added and left to stir until the precursor is deposited.

Preparation of double cation wide band gap perovskite precursor for p-i-n PSCs:
For 1.2M of perovskite precursor solution, Cesium Iodide (Sigma-Aldrich), Formamidinium Iodide (Dyenamo), Lead Iodide (TCI) and Lead Bromide (TCI) are being weighed in one vial.The final stoichiometry corresponds to Cs0.17FA0.83Pb(I1.8Br1.2).DMF (Sigma-Aldrich) and DMSO (Sigma-Aldrich) were used as solvents with ratio DMF:DMSO 4:1.The precursor is stirred for 2 hours at 40°C.Afterwards, the precursor is being filter with a 0.22µm filter pores and then 1-0.5 mg of PGua is added and left to stir until the precursor is deposited.

Stack preparation for p-i-n PSCs:
The substrates, sputtered with Indium Tin Oxide (ITO) were cleaned with Ethanol, Isopropanol and Water sequentially for 5 minutes each.After drying them, they are processed in an UV-O3 chamber for 20 minutes.[2-(3,6-Dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic Acid (TCI) is statically spin coated on the substrate with 3000rpm for 30 seconds and then the substrates are annealed for 10 minutes at 100°C.For the triple cation perovskite deposition, a two-step static spin coating program is used.The first step lasts 10 seconds and rotates with 2000 rpm, while the second step utilizes 6000 rpms for 20 seconds.5 seconds before the end of the second step 150µl of Chlorobenzene are used as antisolvent.Then the substrates are annealed at 100°C for 60 minutes.While for the double cation wide band gap perovskite deposition a two-step static spin coating program is used.The first step lasts 10 seconds and rotates with 2000 rpm, while the second step utilizes 4000 rpms for 40 seconds.10 seconds before the end of the second step 200µl of Chlorobenzene are used as antisolvent.Then the substrates are annealed at 100°C for 15 minutes.Phenyl-C61-butyric acid methyl ester (Ossila) in concentration of 10mg/ml in Chloroform is dynamically spin coated at 4000 rpm for 30 seconds.Bathocuproine (Sigma-Aldrich) in concentration of 0.5mg/ml is also dynamically spin coated at 4000 rpm for 60 seconds.Lastly, 100nm of aluminium are being thermally evaporated in a vacuum chamber.For the first 5nm of the material's deposition, a rate of 1nm per second is utilized.After 5nm of the electrode's thickness have been evaporated, the deposition rate is increased to 10nm per second and kept constant until the end of the process.

Devices Characterization
The solar cell devices were measured using a 300 W Xenon light source (Oriel).The spectral mismatch between AM 1.5 G and the solar simulator was adjusted by a Schott K113 Tempax filter (Prazosopms Gas & Optik GmbH).The light intensity was calibrated with a silicon photodiode with an IR-cutoff filter (KG2, Schott).Current-voltage characteristics were applied by an external voltage bias while measuring the corresponding current with Keithley 2400 source meter in ambient air.The voltage scan rate was 100 mV/s.The devices were covered with a black metal mask with an active area of 0.16 cm 2 . 1 H NMR measurements were performed on Bruker AvanceIII-400 MHz NMR spectrometer.Incident photon to current efficiency (IPCE) was carried by a commercial apparatus (Aekeo-Ariadne, Cicci Research s.r.l.).The top-view and cross-section morphologies of the samples was characterized using a high-resolution scanning electron microscope (Zeiss Merlin) with an in-lens secondary electron detector.The operational stability of the devices was measured under a white light-emitting diode lamp with biologic MPG2 potentiostat under N2 gas flow at maximum power point tracking (MPPT).The thermal test was carried out on a hotplate under N2 gas flow in the dark.

PL mapping and PLQY measurement
The PL images were optained by OLYMPUS BX50 stereomicroscope and sCMOS camera ("Zyla 5.5 sCMOS" by Andor) with a long-pass filter while the partial illumination of the sample was provided by a 623 nm red light-emitting diode (Thorlabs, SOLIS-623C).Photoluminescence quantum yield (PLQY) measurements were performed using an absolute photoluminescence characterization setup from Quantum Yield Berlin (QYB).The sample was radiated with 532 nm laser light at different photon fluxes (equivalent to 0.007-10 Suns) focused on a sample (placed in an integrating sphere) with an illumination spot size of 0.1 cm2.The measured counts were obtained with an integration time of 0.5 s and averaged over 10 spectra.

TrPL measurement
Transient photoluminescence measurements were performed with a UV-vis photomultiplier tube and a single-photon counting device (Timeharp).A 515 nm laser (Omicron) was used as a light source and it was modulated digitally with a trigger signal generated by an arbitrary wave-form generator.The laser power during on-time was adjusted to match the Jsc of a perovskite solar cell under 1-sun illumination (AM 1.5G) and lower intensities.The on-time was set to 240 µs and the off-time to 10 µs.The integration time was set to 300 s.The spot size was around 0.785 cm².

XRD measurement
The X-ray diffraction patterns were recorded with PANalytical Empyrean system with a PIXcel-1D detector, Bragg-Brentano beam optics and parallel beam optics.Light source is from copper Kα beam filtered with nickel β filter.Diffraction spectra were characterized between 2-theta of 5 o and 50 o at a scan rate of 1 o per minute with the step width of 0.02 o Other measurements 1 H NMR measurements were performed on Bruker AvanceIII-400 MHz NMR spectrometer.XPS measurements were carried out on an Axis Supra apparatus (Kratos Analytical) using the monochromated Kα X-ray line of an aluminum anode.The pass energy was set to 20 eV with a step size of 0.1 eV.Scanning electron microscopy was performed on a ZEISS Merlin high-resolution (HR)-SEM.
,-1 represent the concentrations of electrons and holes in the trap states, which are functions of the relative trap energy level  / and temperature T, with k being the Boltzmann constant.Furthermore, the so called carrier lifetimes correspond to the minority carrier lifetimes for electrons and holes, which depend on the product of trap density  ' , capture cross section σ ! of the respective charge carrier and the thermal velocity  '( . If the electron and hole concentrations are about equal ( = ) and significantly exceed the intrinsic carrier concentration ( $ ), the SRH recombination rate simplifies to: Where  SRH,E1 is the SRH recombination rate via a trap level at energy ET,1 and  SRH,E2 is the SRH recombination via a second trap level with energy ET,2.The other parameters are the generation rate  and   as the external radiative recombination coefficient, respectively.
The steady state PLQY was then calculated as Therein, G is the averaged generation rate calculated from the photon flux density divided by the layer thickness of 470 nm.An effective density of states of  ) =  .= 2.0 10 &7 cm -3 was assumed. [1]The external radiative recombination coefficient was set to  8,9:; = 10 +&< cm 3 s -1 .The energy level of the second trap level T2 is set to be approximative mid-bandgap  -,% = −4.75(eV) between the conduction band edge  ) =-4.0 (eV) and the valence band edge  = =-5.53(eV) The free parameters from the steady-state equation were chosen to minimize the squared difference as for the different generation rates G used for the intensity dependent PLQY measurements.

PGua modified sample
Reference sample ',%/(,% (s) 2.34^-5 1.75^-5 ET,1 (eV) -4.0269 -4.0643 The data from the transient photoluminescence (TRPL) was normalized and fitted with the biexponential decay formula With the normalized overall count rate , the two initial Intensities  DFT calculations have been carried out on the (001) MAPbI3 surface within supercell approach by using the Perdew-Burke-Ernzerhof (PBE) [2] functional.Slab models have been built starting from the tetragonal phase of MAPbI3, by fixing cell parameters to the experimental values. [3]his approach has been already applied previously to ensure a proper comparison with the MAPbI3 systems. [4]A 15 Å of vacuum were added along the non-periodic direction perpendicular to the slabs in all cases.
Perovskite models are simulated using the Quantum Espresso package. [5]PBE calculations have been performed by using ultrasoft pseudopotentials (shells explicitly included in calculations: F 2s, 2p; Br 4s, 4p; I 5s, 5p; N, C 2s, 2p; H 1s; Pb 6s, 6p, 5d) and a cutoff on the wavefunctions of 40 Ryd (320 Ryd on the charge density).DFT-D3 correction were also included. [6]Electronic structures of the pristine bulk and the PGua passivated bulk supercells were refined using the hybrid HSE06 exchange correlation functional with α=0.43 and inclusion of spin-orbit coupling corrections.The calculated DOS reported in the diagrams in Figure S9 have been aligned to the respective VB level in all cases.
Starting from the flat PbI2-terminated (001) surface, PGua molecules were added in the two defective slabs: VPbI2 and Iodine Frenkel (VI + /Ii -).The calculation of additives inside the MAPbI3 bulk was carried out on the 2x2x2 tetragonal supercell with a PbI2 vacancy (VPbI2), with optimized ionic and cell parameters to estimate volume changes.
The formation energy of the PbI2 vacancy at the surface and in the bulk is given by: where (prist.) is the pristine reference system (bulk or surface), (def.) is the respective system with the  KLM% , and µ(PbI % ) is the chemical potential of a PbI2 unit, obtained from the geometry optimized PbI2 phase, COD ID 9009114. [7]e passivation energy when adding PGua (or other additives in the bulk) at the respective surface/bulk defects is calculated as: where (PVK + additive) is the additive-passivated perovskite, and (additive) is the energy of a single additive molecule.
The formation energy of the iodine Frenkel (VI + /Ii -) defect at the perovskite surface is directly given by: Note that, as all considered defects are charge neutral, no electrostatic corrections are required.
Molecular complexes simulations have been carried out using Gaussian09 program package [8] with the B3LYP functional [9] along with the lanl2dz basis set for Pb and I and 6-31G* for the other species, including empirical DFT-D3 dispersion corrections. [10]    Since the Voc of the PSC with PGua and CEAI is even higher than the QFLS of the PGuapassivated perovskite film, we rely on the assumption that the QFLS must be at least 1.181 V. Hence, the remaining 79 mV (or less) difference between the QFLSmax and the QFLS of the PGua+CEAI sample is attributed to an additional non-radiative recombination in the bulk and at perovskite surface.Considering that CEAI is a surface passivation applied as a compact layer on top of the perovskite film, we assume that it cannot significantly alter the bulk non-radiative recombination and is primarily passivating the surface.Hence, for the arbitrary calculation of the PCE losses due to perovskite surface recombination, we consider PGua-and CEAIpassivated sample to be limited only by the non-radiative recombination in the perovskite bulk.
The maximum QFLS of a semiconductor ( >4I ) can be found from the material bandgap and absorptivity using the following equation: Using the common notion that the ( >  6 ) = 1 and ( <  6 ) = 0, the equation X can be solved to find the maximum QFLS for a specific  6 .Based on the emission peak from PLQY measurements at 1.53 eV, we calculated the  >4I of perovskite studied in this work to be 1.26 eV.

Figure S1. 1 H
Figure S1. 1 H NMR spectra of (a) FAPbI3 solution and (b) the mixed solution of PGua and FAPbI3 (the molar ratio of PGua and FAPbI3 is 1:2.5), which were dissolved in DMSO-d6.The integration of peaks shows the NH peak from guanidine is mixed with the NH peak from FA + .(c) The UV-Vis absorption of PbI2, PGua, the mixture of PbI2 with PGua, FAPbI3 and the mixture of FAPbI3 with PGua solution, which were dissolved in DMSO solution with the concentration of 0.3 mM.XPS core level signals of N 1s of PGua (d) and the mixture of PbI2 with PGua (e).(f) The UV-Vis absorption of PbI2, PGua, the mixture of PbI2 with PGua film.The molar ratio of PGua and PbI2 (FAPbI3) is 1:1 in the above experiments (c-f).

Figure S6 .
Figure S6.UV-vis absorption spectra of (a) reference perovskite film and (b) PGua-treated perovskite film at various annealing times.

Figure S7 .
Figure S7.Absolute PL spectra of perovskite films on glass under different incident illumination intensities.

Figure S8 .
Figure S8.Generation and Recombination pathways in a two-trap level SRH model.The rate equation to describe the intensity dependent change in charge carrier density was modeled with generation, radiative recombination and a two-trap level Shockley-Read-Hall mechanism as depicted in Figure S8.While radiative recombination is described as classical band to band recombination, two nonradiative recombination pathways are mediated by trap states.The recombination of charge carriers (electrons  and holes ) via trap states, known as the Shockley-Read-Hall (SRH) mechanism, is governed by the following equation: D and  & and the two lifetimes  D and  & .Using a linear loss function  = X >?4@A8?B,-EFG −  H$+?I' X & leads to an good approximation of the measured data except for the TRPL-Data of the PGua sample measured at 1 Sun equivalent fluence.Using a root loss function  = X >?4@A8?B,-EFG −  H$+?I' X &/% increases the fit quality for the that sample but decreases the perceived fit quality with most other measurements.The differential lifetime τ TPL τ TPL = ^− 1   ln( TPL )  b +& τ TPL / = − ^ ln( TPL )  b +& from the bi-exponential fit is depicted in Figure S9 b) and d).

Figure S9 .
Figure S9.Light intensity dependent, normalized Time-resolved PL (TRPL) decays with corresponding bi-exponential fits for using a linear (a) and square-root (c) loss function for reference perovskite (black) and PGua doped perovskite (red).(b) and (d) show the differential lifetimes extracted from the biexponential fit (white lines in (a) and (c)) with the linear and square-root loss function, respectively.

Figure S10 .
Figure S10.Top view of the PbI2-terminated surface layer with (a) PbI2 Schottky defect, VPbI2, and (b) after passivation with PGua.All remaining layers of the slab have been removed for an improved visibility of the Schottky defect.

Figure S11 .
Figure S11.Density of states of (a) pristine and (b) passivated bulk perovskite at the HSE06+SOC level of theory.In (b), PGua passivates a neutral PbI2 vacancy.

Figure S12 .
Figure S12.Surface and bulk passivation by additives: (a) Structural representation of an additive molecule bonded to a undercoordinated Pb ion at the PbI2-terminated surface.(b) MAPbI3 bulk with an additive molecule passivating a PbI2 vacancy.

Figure S14 .
Figure S14.Absorption coefficient measurement and obtained Tauc plot of reference perovskite (black) and PGua doped perovskite films (ref) on glass.

Figure S15 .
Figure S15.(a) Light intensity-dependent measurement of Voc of reference device (black), PGua treated device (red) and combined CEAI passivated device (blue).(b) cell JV-curves compared with the pseudo JV-curves, reconstructed from Fig. 4a of the reference and PGuapassivated perovskite solar cells, where the difference between them represents the charge transport losses.

Figure S16 .
Figure S16.Product of Voc and FF of the manufactured PSCs found in the perovskite database as a function of the absorber bandgap (EG) in comparison to the champion cell shown in this work.The data of over >28,000 PSCs shown in this plot was gathered from perovskite database.

Figure S17 .
Figure S17.The linear fitting of MPPT of combined passivated device, the slope of the fitting line is 7.17*10 -6 .

Figure S22 .
Figure S22.JV-curves of the champion p-i-n devices made with (left) triple-cation and (right) double-cation perovskites with and without addition of PGua.

Table S1 :
Additive binding energy with PbI2 precursor complex.
( For shallow traps it follows that  & >>  & or  & >>  & dependent on whether the trap energy is close to the conduction band or the valence band.For low generation rates it can also be assumed that  <<  & or  <<  & .Thus, as either  & or  & govern the denominator of the SRH recombination rate, we only must consider either  ' (if  & >>  & ) or  ( (if  & >>  & ).For simplicity we therefore set  ' =  ( .

Table S2 :
Additive binding energy with undercoordinated Pb ions at the perovskite grain surface shown in FigureS10a.

Table S3 :
Passivation of PbI2 vacancies in the MAPbI3 bulk shown in

Table S4 .
Champion Photovoltaic Parameters of PSCs with, without PGua dopant and with CEAI treatment.