Perovskite grain wrapping by converting interfaces and grain boundaries into robust and water-insoluble low-dimensional perovskites

Abstract

Stabilizing perovskite solar cells requires consideration of all defective sites in the devices. Substantial efforts have been devoted to interfaces, while stabilization of grain boundaries received less attention. Here, we report on a molecule tributyl(methyl)phosphonium iodide (TPI), which can convert perovskite into a wide bandgap one-dimensional (1D) perovskite that is mechanically robust and water insoluble. Mixing TPI with perovskite precursor results in a wrapping of perovskite grains with both grain surfaces and grain boundaries converted into several nanometer-thick 1D perovskites during the grain formation process as observed by direct mapping. The grain wrapping passivates the grain boundaries, enhances their resistance to moisture, and reduces the iodine released during light soaking. The perovskite films with wrapped grains are more stable under heat and light. The best device with wrapped grains maintained 92.2% of its highest efficiency after light soaking under 1-sun illumination for 1900 hours at 55°C open-circuit condition.

Tributyl(methyl)phosphonium iodide is added to perovskite to form a robust grain-wrapping layer to increase solar cell stability.

INTRODUCTION

With the perovskite solar cells at the cusp of commercialization, the instability issue of perovskite solar cells is increasingly important and urgent to be solved. Many different strategies have been applied to improve the intrinsic stability of the perovskite layers, including compositional engineering, morphology improvement, defect passivation, charge transport layer modifications, strain engineering, etc. (1–6). Incorporating metal cations such as Cs⁺ was found to enhance the thermal stability and, more importantly, the light stability of methylammonium lead iodide perovskite (MAPbI₃) (7). It is also vital to stabilize the black phase of formamidinium lead iodide perovskite (FAPbI₃) by tuning the Goldschmidt tolerance factor toward a region in favor of a stable perovskite structure. Triple cations [methylammonium (MA⁺), formamidinium (FA⁺), and Cs⁺] or quadruple cations (MA⁺, FA⁺, Cs⁺, and Rb⁺) and double anions or triple anions (Br⁻, I⁻, and Cl⁻) were broadly reported to have better thermal stability than MAPbI₃ (8, 9). Recent studies showed that removing or markedly reducing methylammonium cations would noticeably improve perovskites thermal, chemical, and light stability (10). To improve the film morphology, many solid or liquid additives, such as dimethyl sulfoxide (DMSO) (11, 12), carbohydrazide (13), hydroxyl-substituted polyamide derivative (14), and n-heptylamine (15) in a combination of antisolvents and annealing recipes, were used to tune the film drying and crystallization process, which effectively increased the grain sizes and improved the grain crystallinity. Surface passivation, including adding organic molecules with unique functional groups, which have secondary chemical bonding with perovskites (16, 17), or converting perovskites into robust two-dimensional (2D) perovskites (18–21), oxysalts (22), or sulfide (23), is effective in not only passivating the surface defects but also enhancing the stability of perovskites.

Most of the strategies to improve the perovskite stability pointed out the importance of strengthening grain surfaces and grain boundaries (GBs) in polycrystalline perovskite films. The defective sites and GBs usually initialize degradation when the materials interact with moisture, oxygen, or other stressors such as high temperature, light [particularly ultraviolet (UV) light], and bias. Moisture tends to absorb protons from the ammonium cations at the GBs and then form an intermediate complex that could lastly decompose (24, 25). Oxygen can attack surface iodide vacancy to form superoxides species with the assistance of photoexcited carriers that can further damage perovskites (26, 27). Ion migration, accelerated under bias and high temperature, is generally much faster through defects such as GBs (28–30). At high operational temperatures, highly volatile species such as MA⁺ may depart the perovskite films from the material interfaces and GBs (31) or interact with the charge transport layers or etch electrodes. Therefore, it is essential to passivate both the interfaces and GBs and make them more robust toward all stressors, which affect not only the efficiency but also the stability of perovskite solar cells. However, most past studies only passivated the top surfaces, while the GB passivation has not been well studied. Our recent study shows that GBs have at least comparable or even more severe charge recombination rate than film surface from a direct measurement (32). Compared to the material surface, the GBs are more difficult to access by passivation molecules if the surface treatment is used for passivation. Although the passivation of GBs was claimed as a result of surface treatments, it is not clear whether these surface treatments really allow the passivation molecules to penetrate the already formed GBs and passivate them (33, 34). On the other hand, it is often speculated that some passivation molecules added in precursor solution would be expelled to GBs during film formation, and there was no rigorous study to determine whether these passivation molecules would really stay in GBs (35–38).

In this work, we report a molecule, tributyl(methyl)phosphonium iodide (TPI), which can convert the perovskite GBs to a thin and robust 1D perovskite when it was added to the perovskite precursor. This 1D perovskite is very stable against water/moisture, thermal, and light stressors. We directly observed that the 1D perovskite could wrap the perovskite grains, resulting in improved device efficiency and much better long-term light stability for perovskite solar cells.

RESULTS

Design of grain-wrapping material and its property

The protection of perovskite films by surface passivation, or even converting the perovskite surface into robust inorganic shells, might not be sufficient to protect perovskites, because degradation can precede along in-plane direction once degradation starts at some defective sites or imperfect surface protection. We propose a material structure with each perovskite grain wrapped by a thin, robust material that can be more tolerant to nonperfect passivation. It adds another barrier to lateral precession of degradation, as illustrated in Fig. 1A. When we searched for GB wrapping materials, we found that TPI behaved very differently from many other passivation molecules. In contrast to the weak coordination bonds for many passivation molecules, TPI reacted with perovskites to partially convert it into a wider bandgap material that shows excellent water resistance.

Fig. 1. 1D TPPbI₃ for grain wrapping.

(A) Schematic diagram of TPPbI₃-wrapped perovskite grains. ETL, electron transport layer; HTL, hole transport layer. (B) A photograph of TPPbI₃ single crystals. Scale bar, 0.2 mm. (C) Absorption spectrum of a TPPbI₃ film. OD, optical density. (D) The single-crystal structure of TPPbI₃. (E) XRD patterns of TPPbI₃ film before and after 2.5-sun illumination at105°C in N₂ degradation test for 170 hours. cps, counts per second. (F) Photographs of MAPbI₃ powder before and after soaking in water for 10 s. (G) Photographs of TPPbI₃-wrapped MAPbI₃ before and after soaking in water for 12 hours.

To understand what is the reaction product, we applied the Fourier transform infrared (FTIR) spectroscopy and x-ray diffraction (XRD) measurements to investigate the reaction product(s) between MAPbI₃ and TPI as well as PbI₂ and TPI. We first examined the FTIR spectra of the perovskite precursor solution with TPI additive to find out whether they form complex in solution. As shown in fig. S1, no peak shift or peak broadening was observed in the FTIR spectra, indicating that TPI does not strongly coordinate with MAPbI₃ or PbI₂ in N,N′-dimethylformamide (DMF) solution. This is actually important to apply TPI in device fabrication, because it would be difficult to disperse TPI to GBs if TPI reacted with perovskite precursors in solution to form insoluble substances. Then, we directly mixed MAPbI₃ powder with TPI powder and then heated the mixture to 150°C. TPI melted and thoroughly mixed/reacted with the help of stirring. Then, the reaction products were ground into powder for XRD measurements. We applied four different molar ratios of MAPbI₃:TPI from 3:1 to 1:2, considering that these ratio may be applied in perovskite solar cell fabrication. As shown in fig. S2, the diffraction peaks from the XRD patterns remained to be the same for the MAPbI₃:TPI mixtures with different ratios. We thus conclude that the reaction product should be the same with different perovskite:TPI ratios. When TPI was mixed with methylammonium iodide (MAI) or PbI₂ and heated up to 150°C, the same reaction product was found from the TPI:PbI₂ 1:1 mixture, while no reaction was found between TPI and MAI. To determine the product’s chemical structure, we grew single crystals of the reaction product, with a photo of the crystals shown in Fig. 1B, light yellow strip-like crystals. A bandgap of 2.88 eV is determined from the absorption spectrum shown in Fig. 1C. We conducted the single-crystal XRD measurement. The crystal data and structural refinement data are shown in table S1 (crystal data and structural refinement data for TPPbI₃ single crystal), which concludes that the crystal is TPPbI₃. As shown by the crystal structure in Fig. 1D, TPPbI₃ is a 1D perovskite, which has TP⁺ cations surrounding [PbI₆]⁴⁻ 1D octahedral cages connected by face sharing.

We characterized the light and thermal stability of spin-coated TPPbI₃ films. We put the TPPbI₃ film and FA_0.91Cs_0.09PbI₃(FACs perovskite) control film as comparison under 2.5-sun illumination in a N₂-filled box at 105°C for the degradation test. Then, we compared their XRD patterns before and after the heating and light soaking. As Fig. 1E shows, for the TPPbI₃ film, the intensity of the 7.49° and 10.01° peaks increased after a 170-hour test, which means that it had higher crystallinity and no formation of PbI₂ was detected. It indicates that the product is very stable under the light. In contrast, the whole FACs control film degrades within 6 hours as shown in fig. S3. Then, we annealed the bare TPPbI₃ films without any capping layer at 120°C for 12 hours in a N₂-filled glove box. As shown in fig. S4, the main peak around 7.49° intensity increased substantially, indicating that its crystallinity became better under 120°C and other peaks almost did not change, which means that the TPPbI₃ has good thermal stability.

To demonstrate the protective effective of TPPbI₃ against water, we mixed MAPbI₃ and TPI powders at a molar ratio of 3:1. At this ratio, the TPI can be fully reacted, and there are some MAPbI₃ left unreacted. The solid mixture did not show obvious change at room temperature, but the solid reaction was seen by the naked eyes when they were annealed at 150°C, which is above the melting point of TPI (123°C). The white TPI power was melted and turned into a transparent liquid. After mixing with MAPbI₃ powder with stirring for 5 min, the liquid turned light yellow. The shell of MAPbI₃ particles was quickly turned into TPPbI₃, while some black MAPbI₃ at the core of these particles remained unreacted, indicating that the MAPbI₃ is excessive. After cooling to room temperature, the products went back to a solid state. We then soaked the powder in water. As shown in Fig. 1F, the powder did not show noticeable color change even after water soaking for 12 hours. In notable contrast, MAPbI₃ powder turned yellow within 10 s when it was soaked in water. After we ground the core/shell-structured powder to expose the MAPbI₃ core, it also turned yellow within 10 s. It shows that the formed TPPbI₃ shell has a strong resistance to water or moisture.

We also used MAPbI₃ films treated by TPI for surface passivation to check whether the TPI could stabilize perovskites under illumination. We dissolved TPI into toluene (TL) and isopropyl alcohol (IPA) mixed solvent with a volume ratio of 1:4 at a concentration of 4 mM. A mixed solvent was used, because IPA dissolves TPI better, while the TL was added to reduce the damage of IPA to perovskites. After surface treatment by spin-coating the TPI solution on MAPbI₃, both the unencapsulated control and TPI-treated MAPbI₃ films were illuminated by simulated sun light with an intensity of 100 mW/cm² in ambient air. As we can see from the XRD patterns in fig. S5, the control MAPbI₃ film started to decompose into PbI₂ in a few minutes, and almost the whole film was converted to PbI₂ in 60 min. In contrast, PbI₂ did not show up in TPI-treated film until light soaking for 120 min. This showed that the conversation of MAPbI₃ surface into 1D perovskite could markedly enhance the stability of MAPbI₃ against moisture under light.

Evidence of grain wrapping

We added TPI into precursor solation to realize grain wrapping in perovskite films. Because TP cations are too big to incorporate into the perovskite lattice, they should be expelled to the GB during the perovskite crystallization process. We chose a double-cation perovskite absorber of (FA_0.91Cs_0.09)PbI₃ (donated as FACs perovskite), fabricated by blade coating for the device stability study, because it demonstrated much superior light and thermal stability (39). To find out whether TPI can really wrap the perovskite grains, we applied atomic force microscopy-based IR (AFM-IR) spectroscopy to investigate the distribution of TP cations in the perovskite films. By measuring the thermal expansion induced by the absorption of infrared light of a chemical bonding–specific wavelength using AFM, AFM-IR can map the distribution of a specific moiety with the spatial resolution of regular AFM (40). We first checked the FTIR signals of a control FACs perovskite film and a pure TPPbI₃ film. As shown in Fig. 2A, we found that the strongest absorption peak of the FACs perovskite is at 1713 cm⁻¹, which can be ascribed to C═N stretching, C─H bending, or N─H bending from FA based on previous studies (41). TPPbI₃ showed two distinguished absorption peaks at around 975 and 1450 cm⁻¹ (Fig. 2B). The first absorption peak should be caused by ─CH₃ rocking from TP⁺, and the absorption peak at 1450 cm⁻¹ should come from the asymmetric ─CH₃ stretching from TP⁺ (42). The FTIR spectrum of TPI has both peaks at 975 and 1450 cm⁻¹ (43), confirming that they are related to TP⁺ cations. To map the distribution of FACs perovskite and TP⁺ cations, we excited the samples at 1713 and 975 cm⁻¹, respectively. The perovskite film in this study contains the optimal TPI amount of 0.6% molar ratio to Pb²⁺ in FACs perovskite, which shows the best device performance. The topography image of the film is shown in Fig. 2C. As shown in Fig. 2D, under excitation with IR light at 975 cm⁻¹, a very strong and clear response showed up at the GBs, suggesting that the grains were well wrapped by TPPbI₃. We estimate the thickness of the TPPbI₃ wrapping layer to be 1.7 nm based on a cuboid-shaped grain with the size of 0.8 μm by 0.8 μm by 0.8 μm. In contrast, no clear TP⁺-related signal could be observed for the control film without TPI (Fig. 2F). Its topography image is shown in Fig. 2E. When IR light at 1713 cm⁻¹ was used to excite the same location of the samples, a complementary distribution of perovskites could be clearly seen in fig. S6. The results showed clear evidence that the TPPbI₃ can wrap the GBs using scalable blading fabrication.

Fig. 2. AFM-IR study of perovskite films surfaces with and without TPI additive.

FTIR spectroscopy in AFM-IR equipment of (A) FACs control perovskite film and (B) TPPbI₃ film. (C) Topography image of FACs perovskite film with 0.6% TPI additive. (D) AFM-IR absorption of FACs perovskite film with 0.6% TPI additive at 975 cm⁻¹, corresponding to TPPbI₃. (E) Topography image of FACs control film. (F) AFM-IR absorption of FACs control at 975 cm⁻¹, corresponding to TPPbI₃.

We further confirm that the TPI could wrap the grains and also stay at the top and bottom surface by conducting a cross-sectional AFM-IR measurement. To prepare the sample, we blade coated a layer of FACs perovskite with 3% TPI additive on a Si substrate for a smooth cross section, since the roughness tolerance of the AFM is less than 3 μm. The higher concentration of TPI can provide better contrast between TPPbI₃ and FACs perovskite. Afterward, we blade coated a layer of CYTOP on top of the perovskite as both perovskite encapsulant and the mechanical support for AFM tip, similar to the samples prepared for high-resolution tunneling electron microscopy (the illustration of the measurement is shown in Fig. 3A). The cross-sectional topography image of a sample is shown in Fig. 3B, which shows the clear interface of the substrate with perovskites. Figure 3C shows the phase image of the cross-sectional sample, where the phase contrast comes from the difference in adhesive, stiffness, and frictional properties of the different materials. We determined the red region to be TPPbI₃ and the blue region to be FACs perovskite. As shown in Fig. 3D, except for the lower phase signal area, which refers to the cleaved perovskite grain, the image at 975 cm⁻¹ showed that the TPPbI₃ stayed across the whole film while enriched at GBs and interfaces of perovskites with substrate and encapsulant. The appearance of more TPI in the upper region labeled as “A” indicates the cross section cleaved through GBs at those regions, rather than grain interior. The distribution of perovskites (Fig. 3E) shown by excitation at 1713 cm⁻¹ is almost complementary to that of TPPbI₃. The top surface and the bottom surface of the perovskite layer were fully covered by the TPPbI₃. The cross-sectional AFM-IR image showed complete wrapping of the grains by TPPbI₃. It is noted that some previous studies showed that a combination of 1D/3D heterostructure structure could improve the perovskite stability. However, they were mainly limited to surface passivation. For example, Yang et al. (33) used a cross-linking polymerizable propargylammounium (PA⁺) on top of the perovskites, while no obvious evidence was shown that 1D perovskites could get into the GBs. Ge et al. (44) applied a 1D trimethyl sulfonium lead triiodide into the 3D perovskite; nevertheless, still no obvious evidence showed that the 1D nanoroid could stay at the GBs. Kong et al. (35) used 2-diethylaminoethylchloride cation to form a 1D perovskitoid as the perovskite growth template, but still no distinct evidence showed that the perovskite would be at GBs. Here, we showed that incorporation of the 1D precursors into perovskite solution could result in the formation of a 1D wrapping layer at both GBs and interfaces.

Fig. 3. AFM-IR study about cross section of perovskite films with TPI additive.

(A) Illustration for the AFM-IR measurement on cross-sectional area of a sample. (B) Topography image of FACs perovskite film with 3% TPI additive. (C) Phase image of FACs perovskite film with 3% TPI additive; AFM-IR absorption of FACs perovskite film with 3% TPI additive (D) at 975 cm⁻¹, corresponding to TPPbI₃, and (E) at 1713 cm⁻¹, corresponding to FACs perovskite. Scale bars, 500 nm.

Characterization of perovskite films and devices

We characterized the optoelectronic property of the FACs perovskite films with and without TPI additives to evaluate the impact of the grain wrapping on FACs perovskite. The conversion of defective perovskite surface or GBs should also reduce the defects in the perovskite films, which was verified by photoluminescence (PL) mapping. PL mapping was also used to investigate the nonradiative carrier recombination of perovskite films before and after thermal degradation. Figure 4 (A and B) shows that the film with wrapped grains has higher and more uniform PL intensity than the FACs control film. Figure S7 shows the PL intensity distribution histogram image of the samples above. The mean value of the emission intensities of the perovskite films with wrapped grains was 2.3 times stronger than the control perovskite film. We estimate that nearly 80% of the perovskite grains have enhanced PL intensity, indicating a grain-wrapping efficiency, and the other 20% are already passivated without grain wrapping. Additional characterizations were performed using the thermal admittance spectroscopy and transient photovoltage (TPV) decay to quantify the effects of grain wrapping on the defect generation and charge recombination dynamics in the devices under operation conditions. Admittance spectroscopy is a well-established method to quantify the trap density of states (t-DOS) in thin-film photovoltaics. As shown in Fig. 4C, adding TPI to the FACs perovskites markedly reduced the trap state density, particularly in the trap depth region of 0.3 to 0.5 eV, which corresponded to the defects of positively charged iodide interstitials (32). The TPV decay measurement (Fig. 4D) showed that the carrier recombination lifetime increased from 0.75 μs in the control device to 1.16 μs in the targeted device. Then, we encapsulated both films and kept them at 70°C for 456 hours in the dark to evaluate the thermal stability. As shown in Fig. 4 (E and F), the emission intensity of the perovskite film with TPI is 1.5 times stronger than the control film after thermal stress testing, and its average carrier recombination lifetime was also 1.6 times longer than that of the control film. The stronger PL intensity and longer lifetime prove that the grain wrapping slowed down the defects generated during the thermal stress testing.

Fig. 4. Performance of perovskite devices with wrapped grains.

PL mapping of fresh (A) FACs control and (B) FACs film with 0.6% TPI additive. Cnts, counts. (C) The t-DOS obtained from thermal admittance spectroscopy measurement of the perovskite devices with and without TPI additive. (D) TPV measurement of devices with and without TPI additive. (F) Time-resolved photoluminescence (TRPL) mapping results of FACs film with the TPI additive under 70°C in the dark for 456 hours. (E) TRPL mapping results of FACs control film under 70°C in the dark for 456 hours. (G) UV-Vis absorption of I₂ of films with different concentrations of TPI additives under light soaking degradation. (H) J-V curves obtained under AM 1.5 G 1-sun illumination at room temperature from FACs control solar cell and FACs solar cell with 0.6% TPI additive. PVSK, perovskite. (I) Stability test results of encapsulated control and devices with 0.6% TPI additive under 1-sun illumination at 55°C open-circuit condition in ambient air.

The TPI in the perovskite films could further reduce the detrimental iodine formation in perovskite during the light soaking. We immersed FACs perovskite control films and films with different concentrations of TPI (0.6, 3, and 6%) into vials filled with 5 ml of TL. Then, we illuminated them for 7 days under 1-sun light-emitting diode (LED) with UV under 55°C in an N₂-filled glovebox. Then, we used UV–visible (Vis) absorption spectra to detect the I₂ extracted by the TL. As shown in Fig. 4G, the peak position around 500 nm is used to indicate the I₂ concentration. With the concentration of TPI increasing, the formation of I₂ decreased.

We also measured the whether TPI additives affect strain in the FACs perovskite films. We compared the XRD pattern of control FACs perovskite and those with 0.6 or 3% TPI additive. Figure S8 shows that the peak position changes are less than 0.01° after adding TPI, which indicates that no extra strain was introduced with the addition of the TPI additive.

The photovoltaic performances of devices with and without TPI were then investigated. We applied the grain-wrapping technique into p-i-n structure solar cells, using poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA) and C₆₀ as the hole and electron extraction layers, respectively. A layer of bathocuproine (BCP) was inserted between the C₆₀ and copper (Cu) as a buffer layer. The device structure is shown in the inset of Fig. 4H. The champion control solar cell without TPI showed an open-circuit voltage (V_OC) of 1.11 V, a short-circuit current density (J_SC) of 24.7 mA cm⁻², a fill factor (FF) of 80%, and thus a power conversion efficiency (PCE) of 22.0% under air mass (AM) 1.5 global (G) illumination. We optimized the percentage of TPI in perovskite precursors. As shown by the photocurrent curves in fig. S9, too much TPI (e.g., 1.8% mole ratio to lead) would make the device less efficient, mainly due to the insulating nature of TPPbI₃, while too less TPI does not really make a difference in device efficiency or stability. The optimal TPI concentration was 0.6 mole percent (mol %) (compared to Pb). After adding 0.6 mol % TPI in FACs perovskites, the resultant champion PCE was slightly improved to 22.9%, along with an enhanced V_OC of 1.15 V, a same J_SC of 24.70 mA cm⁻², and a slightly increased FF of 81%. Statistical photocurrent data with 20 devices of each condition are shown in fig. S10; the average PCE was increased from 21.15 to 21.96% by grain wrapping. We also compared the light stability of encapsulated FACs perovskite solar cells with and without TPI by light soaking the devices under 1-sun illumination for 1900 hours at 55°C under ambient conditions. As shown in Fig. 3I, the efficiency of the best control device decreased by 20.3% from its highest PCE after light soaking for 1877 hours. In contrast, the best device with grain wrapping showed only a 7.8% reduction of its highest PCE after light soaking for the same period. We extracted a duration of 3783 hours for the devices that reaches 80% of the highest PCE (T₈₀) for the device with grain wrapping, which is twice longer than the T₈₀ of the best control device. It is noted that this stability test was conducted at open-circuit condition, which is a more harsh condition than maximum power point tracking (45).

DISCUSSION

In summary, we have demonstrated that adding the TPI into FACs perovskite precursor solution to convert the GBs into 1D perovskites can effectively passivate the perovskite grains. This protective layer is robust toward stressors such as heat, light, and moisture, and it can substantially increase the film’s thermal and light stability. Moreover, the grain wrapping could markedly reduce iodide-related defect generation, evidenced by a slowed down formation of I₂ vapor of the perovskite films under light. After forming the protective layer at GBs, the device could reach 22.9% PCE and a T₈₀ lifetime of 3783 hours under light soaking conditions. Adding TPI into the perovskite precursor solution to wrap perovskite grains is a facile way to improve the intrinsic stability of perovskites.

MATERIALS AND METHODS

Materials

Unless stated otherwise, all the materials and solvents were purchased from Sigma-Aldrich. Formamidinium iodide (FAI) was purchased from GreatCell Solar. CYTOP was purchased from AGC Chemicals Americas Inc. Epoxy was purchased from Devcon Industries Inc.

Device fabrication

Indium tin oxide (ITO)/glass substrates were cleaned with acetone in an ultrasonic cleaner and dried in an oven at 60°C. Then, the ITO/glass substrates were treated with UV ozone for 15 min. PTAA (2.2 mg/ml) in TL solution was first blade-coated without posttreatment for the PTAA film. The perovskite layer was then blade-coated with an N₂ knife blowing at room temperature using a premixed A-B ink (39). The A ink was prepared by dissolving 2.2 M FAI and PbI₂ in 2-mercaptoethanol (2-ME) at room temperature and then further diluted by 2-ME to obtain a concentration of 1.1 M according to our previous solvent system recipe. The B ink was prepared by dissolving CsI and PbI₂ into DMSO at a ratio of 1:1 and stirring the solution at 60°C to fully dissolve the materials at 2.0 M. Formamidinium chloride, phenylethylammonium chloride, and formamidinium hypophosphite were added to the A ink as additives at molar percentages of (1.5, 0.15, and 1.0%) relative to Pb²⁺ ions, respectively. (Formamidinium chloride and formamidinium hypophosphite were used to passivate nonradiative recombination defects and stabilize the perovskite phase.) TPI, at a molar percentage of 0.6% relative to Pb²⁺, was then added to the mixed solution. The coating around a 800-nm-thick film was annealed at 150°C for 3 min and 100°C for 3 min in ambient air to get the perovskite phase. The perovskite film was then thermally evaporated with C₆₀ (30 nm), BCP (6 nm), and Cu (150 nm). Last, the devices were encapsulated by a cover glass with epoxy. The epoxy was then aged for 12 hours before characterization. A thin layer of CYTOP was blade-coated on top of the perovskite film for thermal stability tests.

Fabrication of FTIR sample

One and 0.5 mol of MAPbI₃ and TPI were dissolved in 1 ml of DMF separately. Pure DMF was used as background, and then we measured the FTIR signal of the 0.5 M MAPbI₃ and TPI solutions. Afterward, we mixed the 1 M MAPbI₃ and TPI solution at a volume ratio of 1:1 and measured its FTIR.

Single-crystal synthesis

We dissolved TPI and PbI₂ in 8 ml of 47% hydrogen iodide solution in water with a concentration of 0.5 M. Then, we heated it up to 110°C with stirring. Afterward, we cooled the solution down to room temperature and waited for precipitation.

Fabrication of TPPbI₃ film

A total of 0.5 mol of PbI₂ and TPI were dissolved in 1 ml of DMF. Then, we dropped 30 μl of solution on ITO/glass substrate and spin-coated it at the condition of 5000 rpm for 30 s. Then, we annealed the sample at 100°C for 5 min.

Characterizations

The J-V curves of perovskite devices were obtained with a Keithley 2400 SourceMeter under simulated AM 1.5 G irradiation produced by a xenon lamp–based solar simulator (Oriel Sol3A, Class AAA Solar Simulator). The light intensity was calibrated by a silicon reference cell (Newport, 91150 V-KG5). The scan rate was about 0.15 V s⁻¹, and only the reverse scan was performed. The device temperature was not controlled during measurement. To measure the long-term operational stability of perovskite solar cells, the devices were encapsulated, illuminated by a 1-sun–equivalent LED and under V_OC conditions. Single-crystal XRD measurement was performed on a Bruker SMART Apex II x-ray diffractometer with Cu Kα radiation (λ = 1.54178 Å). The diffraction data were handled by the Olex2 software (46), and the crystal structure was solved and refined by the ShelXT (47) and ShelXL (48) programs, respectively. The thermal ellipsoid model was rendered by VESTA (49). The surface AFM-IR was measured by Bruker nanoIR3 with contact mode. The IR laser range was from 900 to 1900 cm⁻¹. The IR input power is around 0.5 and 0.82%. The measurement was first calibrated by a polymethyl methacrylate (PMMA) sample for accurate topography, phase, and IR images. After calibration, the control blade-coated perovskite and spin-coated TPPbI₃ films FTIR in the range between 900 and 1900 cm⁻¹ were measured. After selecting distinguished IR peaks, 1713 cm⁻¹ for perovskite and 975 cm⁻¹ for TPPbI₃, topography, phase, and IR image of control perovskite film and perovskite film with 0.6% TPI additive were measured. The cross-sectional AFM-IR was measured by Bruker nanoIR3 with tapping mode. The IR laser range was from 900 to 1900 cm⁻¹. The IR input power range is around 2.1%. The measurement was first calibrated by a PMMA sample for accurate topography, phase, and IR images. After calibration, the perovskite film cross-sectional sample with a 3% TPI additive was measured. FTIR was measured by PerkinElmer FTIR spectrometer. The scanning range is from 400 to 4000 cm⁻¹ with a resolution of 4 cm⁻¹ and eight scans each time. Film and powder XRD patterns were recorded with a Rigaku MiniFlex x-ray diffractometer. Time-resolved PL imaging was measured by PicoQuant MicroTime 100 confocal microscopic system. Thermal admittance spectroscopy (t-DOS) measurements were performed by using an Agilent E4980A precision LCR meter, and the amplitude of AC and DC bias was fixed at 0 V, and the amplitude of the AC bias was 20 mV. The scanning range of the AC frequency was 0.02 to 2000 kHz. The t-DOS [N_T(Eω)] is calculated by using an equation N_T(Eω) = −$\frac{1}{\mathit{qkT}}\frac{\mathrm{\omega }\mathit{dC}}{d\mathrm{\omega }}\frac{V_{\text{bi}}}{W}$, where W and V_bi are the depletion width and build-in potential, respectively, which were derived from Mott-Schottky analysis of the C-V measurement; q, k, T, ω, and C are the elementary charge, Boltzmann’s constant, temperature, angular frequency, and specific capacitance, respectively. TPV decay was measured under 1-sun illumination and recorded by a 1-GHz Agilent digital oscilloscope.

Supplementary Materials

This PDF file includes:

Figs. S1 to S10

Table S1

Other Supplementary Material for this manuscript includes the following:

Data S1