Flattening Grain-Boundary Grooves for Perovskite Solar Cells with High Optomechanical Reliability

Opto-mechanical reliability has emerged as an important criterion for evaluating the performance and commercialization potential of perovskite solar cells (PSCs) due to the mechanical-property mismatch of metal halide perovskites with other device layers. In this work, grain-boundary groove, a rarely discussed film microstructural characteristics, is found to impart significant effects on the opto-mechanical reliability of perovskite-substrate heterointerfaces and thus PSC performance. By pre-burying iso-butylammonium chloride additive in the electron-transport layer, we flattened GB grooves and created an opto-mechanically reliable perovskite heterointerface that resists the photothermal fatigue. The improved mechanical integrity of ETL-perovskite heterointerfaces also benefits the charge transport and chemical stability by facilitating the carrier injection and reducing the moisture or solvent trapping, respectively. Accordingly, we achieved high-performance perovskite solar cells which exhibit efficiency retentions of 94.8% under 440 h damp heat test (85% RH and 85 °C), and 93.0% under 2000 h continuous light soaking. This article is protected by copyright. All rights reserved.


Introduction
Heterointerfaces of perovskite and carrierextracting substrate play a critical role in determining the performance of perovskite solar cells (PSCs), which have drawn significant attention in the field. [1,2] In the literature, perovskite heterointerfaces have been most extensively studied regarding their optoelectronic properties, whereas the heterointerface microstructures and associated optomechanical behavior have rarely been investigated. [3][4][5][6][7][8][9] In a typical device-fabrication practice, the perovskite layer is fabricated on a conducting substrate with a pre-deposited inorganic electron-transport layer (ETL), followed by solution deposition of an organic holetransport layer (HTL). [10] Therefore, the intrinsic surface morphological characteristics of perovskite films can strongly determine the resultant heterointerface microstructures. [1,2,4] Because state-of-the-art perovskite layers are invariably polycrystalline, grain boundaries (GBs) are the most prominent microstructural feature that may emerge onto the film surface and inherently create grooving morphologies due to the solid-sate ion migration driven by the difference between film surface/heterointerface and GB energy. [11,12] While there are numerous studies focusing on engineering the density and various other characteristics of GBs, [13][14][15][16] the effects of GB grooves (GBGs) have been rarely discussed.
In PSCs, buried GBGs can create nanoscale physical voids between perovskite layer and bottom substrate, which can impart detrimental effects for a range of reasons: First, these voids can hinder the carrier injection and transport at the heterointerface; Second, these voids can trap environmental gaseous species and processing solvent molecules, influencing the chemical stability; Third and most critically, these voids can trigger interfacial delamination and affect the optomechanical properties of perovskite-substrate heterointerface. For all these considerations, we need to tailor GBGs on the buried perovskite surface for achieving stable heterointerface with optimal optomechanical and opto-electronic properties, but this remains unexplored to the best of our knowledge.
In this work, we leveraged high-resolution atomic force microscopy (AFM) to determine the GBG angles at the buried perovskite-substrate heterointerface of delaminated and flipped perovskite films. We developed a facile method to substantially flatten GBGs by incorporating iso-butylammonium chloride (i-BACl) additive into the SnO 2 ETL that triggers the change in the bottom perovskite-ETL heterointerface energies. We revealed that morphologically flattened GBGs improve the mechanical adhesion strength of the perovskite layer onto the ETL-coated substrate, making this essential heterointerface resist long-term photothermal fatigue. The enhanced mechanical integrity owing to small GBG sideangles also imparts excellent optoelectronic properties and high chemical stability. Owing to such dominating mechanical effects induced by GBG engineering, the resultant PSCs are not only highly efficient, but are also robust under rigorous and standardized stability tests including maximum-power-point tracking, damp-heat test, and light-soaking test.

Results and Discussion
In Figure 1a, the geometrical morphology of a regular GBG cross-section is schematically illustrated. Herein the GBG geometry at a perovskite-ETL heterointerface is inherently determined by the ratio of GB energy to heterointerface energy, [17,18] γ γ ϕ θ ( ) where γ GB is the GB energy, γ HI is the heterointerface energy, ϕ and θ is the dihedral and side angles of a GBG as illustrated in the inset of Figure 1c, respectively. According to the equation, by maintaining the GB energy, an increase in heterointerface energy can lead to the flattening of GBG morphologies via a reduction in the GBG side-angle (θ). In this study, the GB energies for the films with similar perovskite compositions and preparation conditions are considered close. In practice, tailoring of heterointerface energies can be achieved using interfacial chemical modifications. Therefore, we deliberately pre-buried an additive phase of iso-butylammonium chloride (i-BACl) in the deposited SnO 2 ETL and this is found effectively to flatten all GBGs at the buried perovskite-ETL heterointerface.  We also confirm that the surface morphology of SnO 2 ETL with pre-buried i-BACl is as uniform as that of pristine SnO 2 ETL ( Figure S1, Supporting Information). There is no traceable segregation of i-BACl on the SnO 2 film surface. Therefore, upon the ETL deposition, i-BACl is expected to be uniformly coated on the surfaces of individual SnO 2 nanocrystals in an ultrathin manner. Once the perovskite layer is sequentially deposited, we hypothesize that, in addition to the i-BACl originally existing on the ETL film surface, the volatile i-BACl originally buried inside the ETL film bulk can also diffuse to the heterointerface during the high-temperature annealing. These interfacial i-BACl phase can modify the heterointerface energy and result in GBG differences at ETL-perovskite heterointerface. Detailed distribution and dynamics of i-BACl are interesting to investigate in the future. Figure 1b schematically illustrates that the reduction of the GBG side-angle immediately enhances the interfacial contact at the perovskite-ETL heterointerface in the film structure. In order to quantify the GBG angles, we peeled off the perovskite films from the ETL-coated substrate and performed high-resolution atomic force microscopy (AFM) scanning onto the flipped perovskite films. Figure 1d,e shows the 3D AFM views of the flipped perovskite bottom surfaces with and without the GBG engineering. The corresponding surface height profiles extracted from Figure 1d,e are compared in Figure 1f, revealing a dramatic decrease in the GBG side-angle from 18.88° to only 7.29°. In order to obtain statistical evidence for this effect, we have acquired surface height profiles based on at least 100 measurements onto different GBG locations within a relatively large scanned area (2 × 2 µm 2 ). The AFM topography images of perovskite films with and without the GBG engineering are presented in Figure 1g,h. According to the histograms of the measured GBG side-angles, the mean GBG side-angle is reduced from 11.01° to 7.56° by the GBG engineering. We further performed first-principles calculations based on density functional theory (DFT) to reveal the impact of i-BACl chemical modification on the heterointerface energies. The crystallographic models are illustrated in the insets of Figure S2 (Supporting Information). The calculation results reveal that when the heterointerface is modified with i-BACl, the heterointerface relative energy is increased to 1.387 eV nm −2 when setting the pristine heterointerface as 0 eV nm −2 , well coinciding with our expectation (Figure 1c). We also tried to employ FA 0.6 i-BA 0.4 Cl to tailor the film bottom surface, which leads to a mean GBG side-angle of 9.05° ( Figure S3, Supporting Information). The GBG side-angle variation trend for the three samples (pristine, FA 0.6 i-BA 0.4 Cl, i-BACl) is consistent with the heterointerface energy change when different organic cations are used ( Figure S2, Supporting Information), attesting our proposed mechanism for the tailoring of GBG side-angle. We then examined the effect of GBG side-angle reduction on the (opto-)mechanical behavior of the buried perovskite-ETL heterointerface. First, we performed the interlayer delamination tests to compare the intrinsic mechanical reliability of the perovskite-ETL heterointerfaces with and without the GBG engineering. Here the film sample specimens are made with a multilayer structure of glass/ITO/SnO 2 (ETL)/FA 0.9 Cs 0.1 PbI 3 perovskite/poly(methyl methacrylate) (PMMA)/epoxy/glass. As schematically illustrated in Figure 2a, by applying a constant force (F) from one edge of the film, the perovskite film is gradually delaminated from the ETL-coated ITO glass. Note that in general, the perovskite-ETL heterointerface is mechanically weakest as compared with other interfaces in this sample setting. [19] In this regard, the delaminated perovskite film area (normalized using the exact area being divided by the original area) qualitatively reflects the mechanical adhesion strength of the perovskite film onto the ETL-coated substrate. [20,21] Figure 2b,c show the photographs of typical delaminated perovskite films without and with the GBG engineering, respectively. As seen, the perovskite layer with regular GBGs can be peeled off near-completely from the substrate. But when GBGs are flattened, only a small fraction of the film area can be delaminated, indicative of significant improvement in the mechanical integrity of the perovskite-ETL heterointerface. We performed film delamination tests for 50 samples for each case for statistical evidence. As shown in Figure 2d, after the GBG engineering, the mean normalized delaminated area of the standard perovskite film sample (0.8 × 0.8 cm 2 ) is reduced to less than 0.1. In comparison, the perovskite film with regular GBGs exhibits a mean normalized area of 0.65. These results confirm the superior mechanical reliability of the perovskite-ETL heterointerface with the GBG engineering. In addition, there is microstructural evidence of the greatly enhanced heterointerface. We noticed that once GBGs are engineered, in some areas of the delaminated, then flipped film, traces of SnO 2 nanoparticles are evident as shown in Figure S4 (Supporting Information), reflecting the strong perovskite-ETL heterointerface that pulls out SnO 2 from the ETL during the interlayer delamination. We then performed quantitative measurements for the mechanical strength of the perovskite-ETL heterointerface by using a double-cantilever beam delamination technique (see experimental details in Experimental Section and Figure S5a, Supporting Information). Figure 2e shows the traction stress changes applied to the sample until the full detachment of the perovskite layer from the SnO 2 ETL. To compare the mechanical strength of regular GBGs and flattened GBGs samples, we calculated the maximum traction stress and specific fracture energy of the perovskite-ETL heterointerface (See calculation details in Experimental Section ). After calculation, the sample with flattened GBGs delivers a significantly increased maximum traction stress of 3.583 MPa as compared to that (1.079 MPa) for the regular GBG case, which means the heterogeneous interface with flattened GBGs is more resistant to tensile forces. Then we compared the specific fracture energy of the two samples. The sample with flattened GBGs has a significantly higher specific fracture energy of 12645 kJ m −3 than the regular GBG case (3372 kJ m −3 ), which means higher energy was required to create one unit area of a crack between the perovskite-ETL with flattened GBGs. The mechanical strengths of both film structures have been also measured statistically ( Figure S5b,c, Supporting Information), showing average traction stresses of 2.425 MPa and 1.016 MPa and average specific fracture energy of 8855 kJ m −3 and 5763 kJ m −3 for the cases of flattened and regular GBGs, respectively. All the above results support that the flattened GBGs lead to enhanced mechanical reliability of the perovskite-ETL heterointerface.
We then performed photothermal fatigue experiments to compare the optomechanical reliability of the two perovskite-ETL heterointerfaces by placing the films (in the PSC device setting) under continuous one-sun-intensity illumination, coupled with a photothermally induced sample temperature of 50 °C. As schematically illustrated in Figure 2f, the light induces thermal effects that can trigger the morphological evolutions of GBGs and the heterointerface. As recognized, perovskites exhibit much higher expansion coefficients (3.3 to 8.4 × 10 −5 K −1 ) than those of inorganic ETLs (0.37 to 1 × 10 −5 K −1 ). [22][23][24] Upon photothermal effects, the film tends  to expand more than the ETL-coated substrate, causing tensile strain and thus interfacial sliding. The enhanced mechanical strength, coupled with the flattened GBGs, is expected to prevent the generation of interfacial voids and delamination more effectively. We acquired the film cross-sections using scanning electron microscopy (SEM) to compare the photothermal fatigue degree of both perovskite-ETL heterointerface types in response to identical illumination stress. As seen in Figure 2g,h, after 100 h continuous photothermal stress, significant voids are formed at the perovskite-ETL heterointerface with regular GBGs, mostly in the near-GB regions compared with the compact pristine regular GBGs sample ( Figure S6a, Supporting Information). The higher-resolution SEM image in Figure 2h clearly shows the enlarged details of the voids with diameter of about several nanometers and the tendency of perovskite-SnO 2 ETL interlayer delamination. However, for the perovskite film with flattened GBGs, the perovskite-ETL heterointerface is still very compact even after 100 h illumination compared with the pristine flattened GBGs sample ( Figure S6b, Supporting Information), without noticeable voids generation or interface delamination (Figure 2i,j). The AFM images of the buried perovskite-ETL heterointerface can further confirm the different morphological changes of the films with regular GBGs and flattened GBGs ( Figure S7, Supporting Information). On further consideration, the high surface activities at GBGs may lead to the migration of loosely bonded ions, further contributing to the volume expansion (deepening and widening) of GBG structures. To further confirm this, we used AFM to study the GBGs of the bottom perovskite surface after 100 h photothermal stress ( Figure S8a,b, Supporting Information). The mean GBG side-angle increases by 50.1% for the regular GBGs case (from 11.01° to 16.53°) while only a slight increase (12%; from 7.56° to 8.49°) is observed for the flattened GBGs case, as shown in Figure 2k and (Figure S8c,d, Supporting  Information). The X-ray diffraction (XRD) pattern in Figure S9 (Supporting Information) reveals no occurrence of phase degradation at this stage, attesting that the enlargement of voids is a result of GBG and interfacial morphological evolutions and excluding the effect of phase degradation.
We further compared the chemical stability and optoelectronic properties of perovskite thin films with regular and flattened GBGs. Regarding the chemical stability, we used accelerated aging test conditions (three-sun-intensity light generated using the white LEDs, coupled with a sample temperature of 80 °C) to monitor the degradation kinetics of both film samples. Figure 3a,b respectively shows the XRD and UV-vis absorption spectra of perovskite films with regular GBGs after different times (0, 25, 50, 75, 100 h) of aging, revealing quick degradation of perovskite phases into PbI 2 . When GBGs are flattened, only trace amounts of PbI 2 are observed after 100 h of aging, and the optical absorption behavior is mostly retained (Figure 3c,d). Flattened GBGs also lead to obvious improvement in the ambient stability of the films under the testing conditions of 22 °C and 65% RH, as shown in Figure S10   which can be attributed to retarded ingression into the film of moisture from the air. The trap densities in both perovskite film samples were estimated using the space-charge-limited current (SCLC) measurements. An electron-only capacitor-like device structure of SnO 2 /perovskite/PCBM/Au was fabricated. As shown in Figure 3e, the trap-filled-voltage V TFL can be obtained through the kink point of the ohmic and trap-filled regimes. By using the equation of where e is the elementary charge, L is the thickness of perovskite film, ε is the relative dielectric constant of the perovskite and ε 0 is the vacuum permittivity. The trap density in the perovskite film after the GBG engineering is reduced to 1.54 × 10 14 cm −3 as compared with 3.22 × 10 14 cm −3 in the regular perovskite film, calculated based on the extracted parameters shown in Table S1 (Supporting Information). Time-resolved photoluminescence spectroscopy (TRPL) spectra of the delaminated, flipped perovskite film samples for both cases are further acquired. As shown in Figure 3f, the TRPL spectra are fitted well with double exponential decay function f(t) = A 1 exp(−t/τ 1 ) + A 2 exp(−t/τ 2 ), where A 1 and A 2 are the decay amplitudes, and τ 1 and τ 2 are the decay time constants. The fitted parameters are summarized in Table S2 (Supporting Information). The average PL lifetime (τ) for the film with flattened GBGs is 383.70 ns, significantly larger than that (86.21 ns) for the regular GBGs case, suggesting the suppression of non-radiative recombination. [25] We then compared carrier-extraction properties of the perovskite-ETL heterointerfaces with regular and flattened GBGs using TRPL (Figure 3g). Although the intrinsic PL lifetime of the film is increased with the flattened GBGs, the PL quenching dynamics is faster in this film, showing τ of 3.33 ns compared with that of 16.76 ns for the sample with regular GBGs (see the fitted parameters in Table S3, Supporting Information). Steady-state PL spectra and PL mapping are also acquired, showing more effective PL quenching and uniform emission distribution in the film with flattened GBGs (Figure S11, Supporting Information). All these confirm not only better optoelectronic quality of the perovskite film with flattened GBGs, but also more effective charge extraction across the resulting perovskite-ETL heterointerfaces. It is also worth noting that in our experimental conditions, the perovskite-ETL heterointerface engineering does not change the average grain size of the perovskite film from the top view ( Figure S12, Supporting Information), attesting the effects caused by the flattened GBGs.
We fabricated PSCs to evaluate the effects of GBG engineering on device PCEs and stability. Figure 4a shows the J-V curves of the PSC devices with regular and flattened GBGs at reverse scan under stimulated AM 1.5G one-sun illumination. The GBG engineering has clearly improved the device operation, demonstrating a PCE of 23.75%, with a short-circuit current density (J SC ) of 24.55 mA cm  Table S4 (Supporting Information). External quantum efficiency (EQE) spectrum of the device with flattened GBGs is shown in the inset of Figure 4a, showing consistent an integrated current density. We further compared the J-V parameters statistics of the 60 devices with and without the GBG engineering, as shown in Figure S13 (Supporting Information), demonstrating very good reproducibility of the PCE improvement. Note that while the GBG microstructure is considered a proven dominating factor that influences device PCEs, other factors such as interfacial energetics ( Figure S14, Supporting Information) and defect passivation as a result of the i-BACl incorporation can also be contributing factors.
The PSCs with GBG engineering exhibit improved stabilities under various test conditions. We first monitored the PCE evolution upon MPP tracking under one-sun-intensity illumination ( Figure S15, Supporting Information). The PSC with the GBG engineering demonstrates 94.9% PCE retention after 180 h as compared with 67.1% for the regular PSC, demonstrating excellent operational stability improvement. Note that more operational test time is not used, since the observed stability difference has already been striking at 180 h. We also performed damp heat tests based on ISOS-D-3 protocol (85% RH and 85 °C) for encapsulated PSCs. [26] With the GBG engineering, the PCE retention is as high as 94.8% after 440 h testing. For comparison, the control device shows a rapid degradation in the initial 20 h with only 59.0% retention of the original efficiency. Finally, we evaluate the light soaking stability of PSCs under one-sun-intensity illumination in the nitrogenfilled glovebox at a stable temperature of 50 °C based on the ISOS-L-2I protocol. [26] As shown in Figure 4d, a typical GBGengineered PSC delivers a high retention (93.0%) of initial PCE after 2000 h testing is demonstrated, which clearly outperforms a regular PSC. These evidently increased device stabilities confirm the beneficial role of flattened GBGs at the perovskite-ETL heterointerface in terms of enhancing chemical stability and optomechanical reliability.

Conclusion
We have demonstrated the significant role of GBG geometry in optoelectronic properties, chemical stability, and most critically photothermal fatigue resistance of buried perovskite-ETL heterointerface. We find that a chemical modification of this heterointerface can effectively increase the interfacial energy and result in flattening of GBGs. Morphological, optical, and electronic characterizations on the flattened GBGs show that they exhibit improved charge transport characteristics, lower trap density, and higher stability upon light soaking and environmental stressors. We translate this fundamental knowledge to the device practice, leading to highly efficient PSCs with superior operational, environmental, and optomechanical reliability.
Preparation of Solutions and Thin Films: For preparing the ETL solution, the acquired SnO 2 nanoparticles were diluted using distilled water 5 times. For pre-burying i-BACl into SnO 2 ETLs, 1 mL SnO 2 nanoparticles solution was modified by the addition of 5 mg i-BACl. For preparing the HTL solution, 72.3 mg Spiro-OMeTAD, 28.8 µL 4-tertbutylpyridine, 17.5 µL lithium bis (trifluoromethyl sulphonyl) imide acetonitrile solution (520 mg mL −1 ) were dissolved into 1 mL CB. FA 0.9 Cs 0.1 PbI 3 perovskite films are used for the fundamental investigation of GBGs. For preparing FA 0.9 Cs 0.1 PbI 3 perovskite thin films with regular GBGs, typically, a perovskite precursor solution with 1 m molar concentration was first prepared by co-dissolving 154.8 mg FAI, 26.0 mg CsI and 461.0 mg PbI 2 in1 mL mixed solvent of DMF/DMSO (v/v, 7:3). Then, 50 µL of this precursor was spread on the SnO 2 ETL coated ITO substrates, followed by a 3-stage spin-coating process (500 rpm for 5s, 1000 rpm for 10 s and 6000 rpm for 30 s). During the third stage, 400 µL of TB was dripped at the center of the spinning substrate. Then, the spun films were then annealed at 170 °C for 5 min to form the final perovskite films. For preparing FA 0.9 Cs 0.1 PbI 3 perovskite thin films with flattened GBGs, additional 23.0 mg PbI 2 was added into the above perovskite precursor. The films are fabricated on SnO 2 ETLcoated ITO substrates with i-BACl pre-buried according to a method described above. FA 0.9 Cs 0.07 MA 0.03 Pb(I 0.92 Br 0.08 ) 3 perovskite films are used for device demonstration. For preparing FA 0.9 Cs 0.07 MA 0.03 Pb(I 0.92 Br 0.08 ) 3 perovskite thin films with regular GBGs, typically, a perovskite precursor solution with 1.35 m molar concentration was first prepared by co-dissolving 6.4 mg MAI, 13.5 mg FABr, 24.6 mg CsI, 39.6 mg PbBr 2 , 190.4 mg FAI, and 574 mg PbI 2 in 1 mL mixed solvent of DMF/DMSO (v/v, 1:4). Then, 50 µL of this precursor was spread on the SnO 2 ETL coated ITO substrates, followed by a 2-stage spin-coating process (1000 rpm for 10 s and then 4000 rpm for 30 s). During the second stage, 300 µL of EA was dripped at the center of the spinning substrate. Then, the spun films were then annealed at 100 °C for 40 min to form the final perovskite films. For preparing FA 0.9 Cs 0.07 MA 0.03 Pb(I 0.92 Br 0.08 ) 3 perovskite thin films with flattened GBGs, additional 17.2 mg PbI 2 was added into the above perovskite precursor. The films are fabricated on SnO 2 ETL-coated ITO substrates with i-BACl pre-buried according to a method described above.
Adv. Mater. 2023, 35, 2211155   Figure 4. Influence of GBGs on photovoltaic performance and stability of PSCs. a) Current density-voltage (J-V) curves and EQE spectra of the champion PSC devices with regular and flattened GB grooves and b,c) corresponding current/PCE outputs at the maximum power points (MPPs). d) MPP tracking under one-sun-intensity illumination; e) damp heat test under 85% relative humidity (RH) and 85 °C heating; and f) light-soaking test (one-sun-intensity illumination, coupled with a sample temperature of 50 °C; in the nitrogen-filled glovebox) stabilities of PSCs with regular and flattened grooves.