Efficient Inverted Perovskite Solar Cells Using Dual Fluorinated Additive Modification

Materials engineering is key to improving the stability and photovoltaic parameters of inverted perovskite solar cells (PSCs). This work presents the effect of two different fluorinated additives on the performance of PSCs containing the archetypal three‐dimensional perovskite, methylammonium lead triiodide (MAPbI3). 3‐(2,3,4,5,6‐Pentafluorophenyl)propylammonium iodide (FPAI) is added to the anode modifying layer and (2,3,4,5,6‐pentafluorophenyl)methylammonium bromide (FMABr) is blended into the perovskite layer. The inverted devices containing FPAI in the anode modifying layer and 0.32 mol% of FMABr from the perovskite precursor solution had hysteresis‐free current density‐voltage characteristics and a maximum power conversion efficiency of 22.3%, which is an absolute increase of 1.7% compared to the MAPbI3 device (20.6%) of the same architecture but without the additives.


DOI: 10.1002/admi.202201939
interlayer between the p-type organic semiconductor and the perovskite layer. The amphiphilic interlayer improves the wettability and interfacial contact of the perovskite precursors on the anode modifier surface, reducing charge carrier recombination, improving the energy level alignment, and facilitating the formation of uniform perovskite films. [1,[5][6][7][8][9] It is believed that the improved wetting arises from hydrophobic-hydrophobic interactions between the anode modifier and the hydrophobic component of the amphiphilic interlayer with the hydrophilic functional groups of the latter coordinating with lead atoms in the perovskite layer. The multifunctional compound, 3-(1-pyridinio)-1-propane sulfonate, is an example of an interlayer material that can π-stack with a PTAA anode modifying film and coordinate with the lead in the perovskite layer. [8] The amphiphilic polymer poly[ (9,9-bis(3-(N,N-dimethyl-N-ethylammonium)propyl)-2,7-fluorene)-alt-2,7-(9,9-di-n-octylfluorene) dibromide] (PFN-P2) has also been demonstrated to be an effective interlayer material, containing a hydrophobic backbone and hydrophilic ionic functional groups that interact with PTAA and the perovskite film, respectively. The use of PFN-P2 as an interlayer has been shown to give high-quality perovskite films and efficient PSCs. [5] In addition to the layer onto which the perovskite is deposited, the interface between the perovskite film and the other charge extraction layer within the device (hole and electron for the standard and inverted architectures, respectively), and perovskite grain boundaries can be sources of defect-mediated nonradiative losses. To improve the optoelectronic properties of polycrystalline perovskite films, various defect passivation strategies have been employed, such as the use of additives in the perovskite precursor solution [10][11][12][13] and interfacial posttreatment modification. [6,[14][15][16][17] Using additives in the perovskite precursor solution have led to in situ passivation of the grain boundaries and surface defects during perovskite film formation. [17,18] For example, fluorinated additives have been shown to migrate to the perovskite/air interface forming a hydrophobic fluorine-based passivation layer that reduces the diffusion of moisture into the perovskite films, even when small amounts are used. [7,10,19] A key advantage of using small amounts of additives in the precursor solution is that the crystal structure and dimensionality of the bulk perovskite film are unchanged. Instead, the additive can modify the crystal orientation and Materials engineering is key to improving the stability and photovoltaic parameters of inverted perovskite solar cells (PSCs). This work presents the effect of two different fluorinated additives on the performance of PSCs containing the archetypal three-dimensional perovskite, methylammonium lead triiodide (MAPbI 3 ). 3-(2,3,4,5,6-Pentafluorophenyl)propylammonium iodide (FPAI) is added to the anode modifying layer and (2,3,4,5,6-pentafluorophenyl)methylammonium bromide (FMABr) is blended into the perovskite layer. The inverted devices containing FPAI in the anode modifying layer and 0.32 mol% of FMABr from the perovskite precursor solution had hysteresisfree current density-voltage characteristics and a maximum power conversion efficiency of 22.3%, which is an absolute increase of 1.7% compared to the MAPbI 3 device (20.6%) of the same architecture but without the additives.

Introduction
Soluble p-type organic semiconductors have been widely utilized as anode modifiers and hole extraction layers in inverted perovskite solar cells (PSCs) as their low-temperature processability is compatible with the promise of low embedded energy, large-scale PSC manufacture. [1] Poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA) is a commonly used p-type material for modifying the anode of inverted PSCs, but its intrinsic hydrophobicity makes it difficult to deposit high-quality perovskite films directly onto its surface. [2] A poor interface between the perovskite film and anode modifying layer can lead to less efficient charge extraction. [3] Indeed, interface quality is at least in part the cause of inverted PSCs having power conversion efficiencies that lag behind those of PSCs in a standard device architecture. [4] A method to overcome the hydrophobicity of the anode-modifying layer is to include a thin amphiphilic grain size, improve the interfaces, and reduce the number of pinholes. These modifications can result in improved charge extraction at both the hole and electron extraction interfaces and charge carrier density in the bulk of the perovskite film, leading to improvement in the photovoltaic parameters, especially, the open-circuit voltage (V oc ) and short-circuit current (J sc ).
In this study, two fluorinated materials with pentafluorophenyl and ammonium functional units, namely 3-(2,3,4,5,6-pentafluorophenyl)propylammonium iodide (FPAI) and (2,3,4,5,6-pentafluorophenyl)methylammonium bromide (FMABr) were used to modify the critical interfaces in inverted PSCs. The FPAI was used in conjunction with PFN-P2 to improve the properties of the anode modifying layer. We have previously reported the effect of alkyl chain length between the pentafluorophenyl ring and ammonium moiety and showed that while FPAI could passivate defects in the perovskite film it was detrimental to device performance. [20] In contrast, the methylene-linked equivalent, 2,3,4,5,6-pentafluorophenylmethylammonium iodide, gave the best performance and hence the use of FMABr in this work. Applying both modifications resulted in an increase in each of the photovoltaic parameters, with the best performing inverted devices achieving power conversion efficiencies (PCEs) of over 22%.

Results and Discussion
The structure of the inverted devices was glass/ITO/PTAA/ PFN-P2/Perovskite/PC 61 BM/BCP/Cu where BCP = bathocuproine and PC 61 BM = [6,6]-phenyl-C 61 -butyric acid methyl ester. To simplify the discussion, the device architecture containing 0.32 mol% of FMABr in the perovskite layer with or without the FPAI interlayer on the PTAA are henceforth referred to as FPAI/FMABr (glass/ITO/PTAA/FPAI/PFN-P2/ MAPbI 3 -FMABr/PC 61 BM/BCP/Cu) and FMABr (glass/ITO/ PTAA/PFN-P2/MAPbI 3 -FMABr/PC 61 BM/BCP/Cu), respectively. The FMABr concentration of 0.32 mol% was chosen for the optimized device based on previous reports that showed low additive concentration enhanced device performance. [10,20] The FMABr synthesis is described in the Supporting Information. All the devices were made under the same conditions, including solution concentrations and processing conditions. We had originally planned to modify the PTAA anode modifier layer through treatment with FPAI (concentration = 1 mg mL −1 in methanol). However, the perovskite films deposited onto the PTAA/FPAI layer suffered from poor coverage ( Figure S1, Supporting Information), which we ascribed to the low surface energy of the fluorinated FPAI. [21] We, therefore, spin-coated a thin interlayer of PFN-P2 (from a 0.5 mg mL −1 in methanol solution) onto the PTAA/FPAI bilayer to improve the wettability of the perovskite solution. Given that both FPAI and PFN-P2 are soluble in methanol, we used X-ray photoelectron spectroscopy (XPS) to determine whether any FPAI remained in the modified film (Figure 1a). Surface scans of the glass/PTAA/FPAI/PFN-P2 sample (with an average analysis depth of around 5 nm) displayed peaks associated with the C 1s, N 1s, F 1s, Br 3d, and I 3d at binding energies of 285 eV, 403 eV, 688 eV, 64 eV, and 613-634 eV, respectively. The presence of the Br 3d, F 1s, and I 3d indicates that both PFN-P2 and FPAI were present in the anode-modifying interlayer. PFN-P2 contains the bromine whilst FPAI is comprised of fluorine and iodine atoms.
Ultraviolet photoelectron spectroscopy (UPS) was used to evaluate the work function (W F ) of the PTAA/FPAI/PFN-P2 on glass film by taking the cut-off at high binding energy and the ionization potential (IP) corresponding to the difference in excitation energy and width of the spectrum. As observed in the data shown in Figure 1b, there was no difference in W F (4.45 eV) between the film with and without the FPAI being present in the interlayer and the ionization potentials were essentially the same at 5.20 eV and 5.15 eV, respectively. An energy level diagram of the two device architectures is shown in Figure S2, Supporting Information.
In the next phase of the study, the effect of FPAI and FMABr on the photophysical properties of the perovskite films was investigated. The UV-vis absorption and steady-state photoluminescence (PL) spectra and lifetimes of the perovskite films with and without FMABr deposited onto substrates coated with the

www.advmatinterfaces.de
anode-modifying materials with and without FPAI are shown in Figure 2. The introduction of a small amount of FMABr into the MAPbI 3 precursor and the modification of PTAA were found not to change the absorption edge or the absorption coefficient (note-the films were all ≈270 nm thick) of the perovskite films (Figure 2a). This was not surprising given the low mol% of additive used. In contrast, for the films deposited on glass, the steady-state PL of the films containing FMABr had lower peak intensity. Furthermore, there was a difference in the effect of the FMABr on the steady-state PL intensity when the films were deposited onto the different interlayers. A decrease in PL intensity has previously been ascribed to more efficient hole extraction and/or reduction in interfacial defects leading to an increase in V oc for devices that have passivated interfaces. [20][21][22][23] In the case of MAPbI 3 a measurable decrease in the steady-state PL intensity was observed when the film was deposited onto the interlayers, which would indicate that more efficient charge extraction was occurring. In contrast, the films containing FMABr all had a similar steady-state PL intensities regardless of whether the film was deposited on glass or an interlayer ( Figure 2b). Hence, any change in V oc would be more likely to arise from interfacial passivation. It is interesting to note that despite the differences in the steady-state PL spectra of the different perovskite films on the different surfaces, the longest lifetime was comparable for all the films within the experimental uncertainty ( Figure 2c and Table S1, Supporting Information). The PL results clearly show that the FMABr has subtle effects on the photophysical properties of the perovskite films when deposited onto different surfaces.
To determine whether addition of FPAI to the anode modifying layer and/or FMABr into MAPbI 3 induced a change in the crystal structure of the deposited perovskite layer, X-ray diffraction (XRD) and grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements were performed (Figures 3a and S3, and Tables S2 and S3, Supporting Information). Apart from small changes in the relative intensities of the major peaks (Table S2, Supporting Information), the diffraction patterns were very similar. The full-width-at-half-maximum (FWHM) values of the peaks in the XRD were also similar (Table S3, Supporting Information), which indicates that the FPAI and FMABr additives do not substantially change the crystal structure of the perovskite films. In the GIWAXS experiments, the dominant scattering component at Q = 1 Å −1 corresponds to the (110) plane of the perovskite. The ring-like intensity suggests that the crystallites are mostly randomly oriented, although there was a slight preference for stacking perpendicular to the substrate, as indicated by the bright spot at Q xy = 0 ( Figure S3, Supporting Information). Both sets of X-ray measurements support the conclusion that there was no explicit change in the perovskite film structure when either of the additives was used, that is, FMABr in the perovskite and FPAI in the anode modifying layer. Furthermore, scanning electron microscopy (SEM) images of the surface of the perovskite films showed that independent of whether one or more additives were used, the grain sizes on the surface of the film were all similar at around 160 nm (Figure 3b-d). Interestingly, the films containing FMABr deposited onto the glass/PTAA/FPAI/PFN-P2 substrate had fewer visible surface pinholes (Figure 3d) than the films deposited on glass/PTAA/PFN-P2. Such surface topology would suggest that a favorable film morphology exists, that would be expected to lead to an improvement in the device performance.
The effects of the FPAI modifier and FMABr additive on the performance of PSCs were compared using devices with the structure of glass/ITO/PTAA/PFN-P2/Perovskite/PC 61 BM/ BCP/Cu and glass/ITO/PTAA/FPAI/PFN-P2/Perovskite/ PC 61 BM/BCP/Cu. The current density-voltage (J-V) characteristics are shown in Figure 4a,b and the corresponding summary of the photovoltaic parameters is shown in Figures 4c,d, and Table 1. The external quantum efficiencies of representative devices are shown in Figure S4, Supporting Information, along with their integrated J sc . The best FPAI-free device fabricated containing the FMABr additive (0.32 mol%) was found to have a PCE of 22.1% with a V oc of 1.13 V and a J sc of 24.1 mA cm −2 , which were higher than the parameters of the reference device with the same architecture of glass/ITO/ PTAA/PFN-P2/MAPbI 3 /PC 61 BM/BCP/Cu but without the

www.advmatinterfaces.de
FMABr (PCE = 20.6%). When the anode modifying layer was treated with FPAI, the best devices were found to have a PCE of 22.3% with a V oc of 1.13 V and a J sc of around 24.8 mA cm −2 . The improved performance of the devices containing the FPAI interlayer and FMABr additive mainly arises from an enhancement in V oc (≈40 mV) and J sc (≈1.6 mA cm −2 ). These results are consistent with the additives reducing the interfacial trap states/defects between the different layers in the device and at the grain boundaries of the perovskite. In addition, the increase in the V oc and J sc was achieved without compromising the FF.
Furthermore, the glass/ITO/PTAA/FPAI/PFN-P2/Perovskite/ PC 61 BM/BCP/Cu devices were found to be hysteresis-freethe forward and reverse J-V curves are shown in Figure 4b. In contrast, the devices without FPAI in the anode modifying layer were found to show slight hysteresis in the J-V curves independent of the presence of FMABr in the perovskite layer ( Figures S5 and S6, Supporting Information). Thus, the use of FPAI in the PTAA/PFN-P2 modifying layer eliminated the hysteresis in the FPAI/FMABr devices, which is again indicative of passivation of interfacial traps and in accordance with the

www.advmatinterfaces.de
PL results and SEM images. To evaluate whether the FPAI and FMABr additives improved the stability, we measured devices with and without the additives under continuous AM1.5G 1 sun illumination without encapsulation in an inert atmosphere. The PCE was measured every minute over a 115-hour period ( Figure S7, Supporting Information). The device containing FMABr in the perovskite and FPAI in the anode modifying layer had improved stability under the measurement conditions, retaining 30% of its initial PCE at maximum power point compared to 20% of the control. The difference in the stability primarily arises from a decreased rapid burn-in for the additivecontaining devices, which could originate from the elimination of trap states and surface passivation. Interestingly, when the scanning software was stopped at the 46 th and 52 nd hour for ≈2 min to save data, we observed a small, sharp increase in the PCE when the scan was recommenced. The origin of the increase is not clear, although we note that the devices were continuously illuminated whilst the data was saved. The increase in the PCE could arise from redistribution of the ions whilst the device was in an open circuit state or interface and bulk trap neutralization by photo-generated carriers. [24]

Conclusion
In this work, FPAI and FMABr were employed to modify the properties of the anode and perovskite film, respectively, for inverted PSCs. While the addition of small amounts of the two fluorinated additives did not change the structural or optical properties of the films dramatically, their combination in the device led to notable improvements in the current density and open circuit voltage. Furthermore, the J-V characteristics of the materials were found to be hysteresis free. The use of both additives in the devices led to an increase of PCE from 20.6% for devices without the additives to over 22%. Thus, we believe that fluorinated additives are a fruitful area of investigation for high-performance and stable perovskite solar cells.
Perovskite Solution Preparation and Device Fabrication: For the MAPbI 3 precursor solution, lead(II) iodide (829.8 mg, 1.8 mmol) and MAI (286.1 mg, 1.8 mmol) were dissolved in a mixed solvent of γ-butyrolactone (GBL, ≥99%) and dimethyl sulfoxide (DMSO, ≥99.9%) (1.5 mL, 7:3 v/v) with stirring and heating at 65 °C for 2 h until the mixture dissolved completely, before being stirred overnight at room temperature before use. For preparing the films containing the additive, the corresponding amount of MAI was replaced with the desired fluorinated salt. For example, MAPbI 3 with 0.32 mol% of the additive was formed from a precursor solution containing 3.2 mg of FMABr, 285.2 mg of MAI, and 829.8 mg of lead(II) iodide dissolved in GBL:DMSO (7:3 v/v, 1.5 mL). The PSCs were fabricated on commercial ITO (20 Ω sq −1 : Xinyan) in a class 1000 clean room. The ITO substrates were sequentially cleaned in an ultrasonic bath with Alconox, Milli-Q water, acetone, and 2-propanol for 15 min each. After being blow-dried with nitrogen, the substrates were treated with UV-ozone for 30 min. All the prepared solutions were filtered before use. PTAA in toluene (2.5 mg mL −1 ) was then spin-coated at 6000 r.p.m. for 30 s onto the ITO. The substrates were put onto a hotplate for 10 min at 100 °C. After cooling for 5 min, FPAI (1 mg mL −1 in methanol) and PFN-P2 (0.5 mg mL −1 in methanol) were sequentially spin-coated at 5000 r.p.m. for 20 s onto the PTAA film. Note -PTAA/PFN-P2 films without the FPAI were prepared similarly. The perovskite precursors were spin-coated through a two-step process (1000 r.p.m. for 10 s and then 5000 r.p.m. for 79 s). During the second step, toluene (200 µL, anhydrous grade, 99.8%) was dropped rapidly after 15 s onto the middle of the rotating substrates. The film was then annealed at 100 °C for 30 min. After deposition of the perovskite layer, a thin PC 61 BM (23 mg mL −1 in chlorobenzene, prepared at least 2 days before and stored in the glovebox) layer was spin-coated at 1500 r.p.m. for 30 s onto the perovskite layer. Then a BCP (0.5 mg mL −1 in methanol) layer was deposited at 4000 r.p.m. for 20 s. Finally, 100 nm of Cu was deposited as the cathode under high vacuum (10 −7 -10 −6 Torr). The active area was defined by a shadow mask to be 0.08 cm 2 . All the films were fabricated under inert conditions (<1 ppm O 2 ; < 1 ppm H 2 O) in a nitrogen-filled glove box.
Film Characterization and Device Testing: Steady-state spectroscopy was carried out on the perovskite samples coated on soda-lime glass without encapsulation. The absorption spectra were measured with a Cary 5000 UV-visible spectrophotometer. The steady-state PL spectra were measured using an Edinburgh Instruments FS5 fluorometer with an excitation wavelength of 520 nm. The PL lifetimes were measured using a Fluorolog-3 spectrofluorometer with an integrated timecorrelated single photon counter with an excitation wavelength of 441 nm and a repetition rate of 1 MHz and pulse duration of 1.3 ns. X-ray photoelectron spectroscopy (XPS) and ultra-violet photoelectron spectroscopy (UPS) were performed using a Kratos Axis Supra+ with an Al Kα (λ = 1486.6 eV) and HeI (λ = 21.22 eV) source, respectively. Films were prepared on ITO on glass, such that the samples were grounded sufficiently so further charge neutralization was not necessary. XPS peaks were analyzed with CasaXPS. UPS data were treated to remove the satellite emission arising from He α, He β and He γ. The high binding energy cut-off was calculated using linear extrapolation while the valence band maximum was calculated by linear extrapolation on a log scale. [25] The surface work function was calculated by subtracting the excitation energy of the He(I) source from the high binding energy cut-off. The ionization potential was calculated by subtracting the excitation energy from the difference between the high binding energy cut-off and valence band maximum. The structures of the perovskite films were evaluated using X-ray diffraction (Bragg-Brentano geometry) on a Rigaku Smartlab diffractometer equipped with a 9 kW Cu rotating anode operating at 45 kV and 200 mA with a scanning range from 10° to 55° (2θ). Grazingincidence wide-angle X-ray scattering (GIWAXS) measurements were performed on a Xenocs Xeuss 2.0 using a Cu kα source (λ = 1.54 Å) with a Pilatus 1 M detector. A grazing-incidence angle of 0.3° was used and image corrections and analysis were performed using the GIXSGUI MATLAB package. [26] Scanning electron microscopy (SEM) images were obtained using a field emission scanning electron microscope (JOEL JSM-7001). The J-V performance characteristics and stability of the devices were tested inside a nitrogen-filled glovebox (O 2 < 1 ppm and H 2 O < 1 ppm) using a Keithley 2400 source-measurement unit under 1 sun illumination (AM 1.5 G, ≈100 Mw cm −2 ) with a scan rate of 200 mV s −1 . The illumination intensity was 100 mW cm −2 from the solar simulator (Sun 2000, class AAB Abet Technologies) which was calibrated against a National Renewable Energy Laboratory (NREL) certified standard 2 cm × 2 cm silicon photodiode with a KG3 filter. The cells were excited through a 0.3 cm diameter mask that matched the size of the cells. The mismatch factor between the certified KG3-filtered silicon photodiode standard and the perovskite cells was determined to be 1.09. The reported PCEs are corrected for the mismatch factor.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.