Progress in in situ TEM investigations of halide perovskites

Nucleation and growth

Nucleation is the first step in the crystallization process, and it largely determines the film quality. In 2016, Qin et al. first investigated the crystallization process of CH₃NH₃PbI₃ (MAPbI₃) using in situ liquid-cell TEM^[86] [Figure 1A]. Initially, they demonstrated the effectiveness of using electron beam-induced solvent evaporation to directly precipitate MAPbI₃ from precursor solutions. The compact structural configurations of the MAPbI₃ NPs enable them to have a high tolerance to electron irradiation. However, it remains necessary to maintain a dose rate of less than 90 e^-/(Ų s) to prevent or minimize beam damage. By quantitatively analyzing the evolution of NPs in the solution, they discovered that the average NP radius (R_a) and time (t) follow a relationship approximately characterized by Ra ~ t^1/5. This relationship does not completely align with either the diffusion or reaction-limited growth models predicted by the Lifshitz-Slyozov-Wagner (LSW) theory^[87-89]. They observed the frequent coalescence phenomenon in solution regions with a high density of precipitates, demonstrating complex dynamical behaviors. To better understand this intriguing occurrence, they proposed an integrated approach considering both crystal orientations and chemical characteristics. This study presents a technique using liquid-cell TEM methods to investigate the nucleation and growth characteristics of organic-inorganic hybrid materials. The approach developed in this research lays the foundation for future comparative studies on hybrid organic-inorganic perovskites composed of different anions (e.g., Cl and Br) and cations (e.g., FA and Cs).

Figure 1. (A) The schematic diagram on the left side of the image shows the formation of hybrid organic-inorganic perovskite precipitation through the electron beam-assisted solvent evaporation method. The right side of the image displays two consecutive frames captured during TEM imaging of the MAPI₃ samples at a dose rate of 86 e^-/(Ų s) per frame. Reproduced with the permission of Ref.^[86] Copyright © 2016 American Chemical Society; (B) In situ TEM characterization exhibiting the evolution of FAPbI₃ film structure over time and temperature. Reproduced with the permission of Ref.^[55] Copyright © 2016 Royal Society of Chemical; (C) The bright field (BF) images show the nucleation process at different temperatures with water partial pressure of about 1 mbar measured at room temperature (RT); (D) The dark field images show the effect of high humidity on the nucleation process at different temperatures. The water partial pressure is about 10 mbar measured at RT. Reproduced with the permission of Ref.^[90] Copyright © 2016 American Chemical Society; (E) Sequential in situ TEM images captured from various sections of a heated liquid cell at 90 °C depict the progression of perovskite nanocrystals transforming into consistent thin films under conditions of both high and low (F) nucleation densities, respectively. For these SAED images, dashed red, green, orange, and yellow quarter circles correspond to {110}/{220}, {213}, {202}, and {400} planes of the tetragonal perovskite crystal. Reproduced with the permission of Ref.^[91] Copyright © 2022 American Chemical Society. TEM: Transmission electron microscopy; SAED: Selected area electron diffraction.

Nearly simultaneously, Aguiar et al. performed a series of in situ TEM experiments to investigate the crystallization behavior of FAPbI₃ perovskite under elevated temperatures^[55]. Their research yielded significant insights into the mechanisms governing perovskite formation and degradation, particularly regarding grain morphology evolution and interfacial chemical reactions. Figure 1B illustrates that the optimal processing temperature for FAPbI₃ occurs at 175 °C. Notably, the grain boundaries exhibit a higher lead content compared to the grain interiors. At this temperature, a partially reversible process occurs, involving the lead exchange between the inner and surface boundaries. Beyond 175 °C, the formation and growth of lead-rich precipitates become evident. Nevertheless, these precipitates can only be partially reintegrated into the material after extended annealing at the same temperature. Additionally, through a thorough microstructural analysis, it was confirmed that the absorption of both hydrogen and hydroxide ions occurs readily at the grain interfaces. This absorption phenomenon can be ascribed to the presence of residual water vapor trapped within the non-crystalline substance at the grain boundaries during the formation process. This discovery provides insight into the intrinsic mechanisms underlying the formation and degradation of HPs under humid conditions.

In 2016, Aguiar et al. conducted a series of controlled in situ gas experiments using a differentially pumped ETEM equipped with a heating stage^[90]. This experimental setup allowed for direct observation of the crystallization process of FA-based perovskite in the presence of water vapor, nitrogen, and oxygen gases. Under these conditions, the combination of low temperature and water vapor pressure did not significantly alter the precursor materials [Figure 1C]. Building upon these observations, they proposed a mechanism for the crystallization and solidification phenomenon in organic-inorganic perovskites, which involves a thermal process within a nonreactive atmosphere. Their hypothesis suggests that the system minimizes its free energy by reducing the free surface area as crystals grow. Furthermore, they proposed that recrystallization occurs along grain boundaries, where any residual water inhibits crystallization and grain growth in the absence of sequential annealing. Additionally, to explore the impact of high humidity on the crystallization process of FAPbI₃ under ambient conditions, they conducted in situ ETEM experiments using a water pressure of 10 mbar at RT. Figure 1D displays the evolution process in a dark field image. They observed rapid material degradation and loss under 1 mbar of water pressure after annealing at 200 °C for ten minutes. However, under 10 mbar of water pressure, no significant degradation occurred at the same temperature. They also observed crystal coarsening under these two different water pressures. Furthermore, they discovered that the crystallization process is similar under both argon and air atmospheres. Consequently, they concluded that the formation of perovskite materials through crystallization remains unaffected by the surrounding gas environment. This study showcased the crystallization process under water vapor and environmental gases, providing a crucial foundation for future research.

In a later study, Wang et al. employed in situ liquid-phase TEM to trace the nucleation process of MAPbI₃ from a precursor solution^[91]. The perovskite precursor solution was spin-coated onto a silicon nitride (SiN_x) window within an in situ liquid chip and heated to 90 °C. As depicted in Figure 1E, perovskite films formed through the aggregation of nanocrystals originating from the substrate surface. The growth of perovskite nanocrystals continued until they contacted each other, covering the entire available substrate surface (t - t₀ = 1164 - 1196 s). Subsequently, grain growth decelerated and eventually ceased. In another region of the liquid cell [Figure 1F], smaller nanocrystals merged to create larger single-crystal grains. Notably, images captured during the later growth phase (t - t₀ = 1229 s) revealed transformations in grain boundaries, leading to adjustments in the angles between grains at triple junctions. This phenomenon was attributed to the principle of minimizing grain boundary energy, resulting in equalized dihedral angles and enhanced grain boundary configurations. Additionally, the researchers explored mechanisms by which additives such as urea influence perovskite formation, leading to increased grain size. They observed that perovskite nanocrystals exhibited higher solubility in a precursor solution containing urea. This solubility enhancement was ascribed to coordinate bonds forming between Pb²⁺ ions in the perovskite and oxygen (O) in urea, weakening the Pb-I bond and elevating perovskite solubility. Their method of investigating perovskite film development provides valuable insights for enhancing the production of perovskites and other technologically significant crystalline films.

These in situ TEM studies reveal the detailed crystallization process and its influencing factors in HPs. While these results provide a microscopic view of the entire process within a range from micrometers to several hundred nanometers, there is still a lack of in-depth nanoscale characterization to reveal more structural details. It is expected that exploring the crystallization behaviors at the nanoscale or even the atomic scale-considering factors such as domain structures, grain boundary arrangements, uniformity of the perovskite phase, and defect-induced strain-may bring insights into the underlying mechanisms.

Phase transition

As previously mentioned, perovskite crystals formed at RT typically undergo a phase transition process. This transition significantly influences the final crystal quality and photovoltaic property. Therefore, gaining a detailed understanding of this transition process can enhance our comprehension of the properties of HPs. In an inorganic HP system, Zhang et al. investigated the phase transition process under different temperatures in a CsPbBr₃ single crystal^[92]. Initially, they identified two types of growth twin domains within the CsPbBr₃ crystal. Subsequently, they explored the relationship between growth twin domains and transformation twins using in situ TEM. By analyzing SAED patterns [Figure 2A], they confirmed the phase transition temperature for a single CsPbBr₃ crystal. The CsPbBr₃ crystal exhibited an orthorhombic structure at RT. As the temperature increased to 90 °C, it transformed into a tetragonal structure and further transitioned to a cubic structure at 133 °C. Notably, an intriguing phenomenon emerged-the evolution of twin domain walls. Figure 2B illustrates that the CsPbBr₃ crystal displayed a distinct twin domain feature at RT. However, as the temperature rose from RT to 120 °C, the boundaries between crystal domains gradually blurred due to slight orientation deviations caused by thermal drift. Surpassing 133 °C, the domain boundaries abruptly vanished, precisely coinciding with the temperature of the first-order phase transition. Upon cooling, the lamellae reappeared but were now horizontally oriented. This peculiar memory effect observed in the crystal twins suggests that the domains formed during crystal growth and phase transition exhibit ferroelastic properties. These domains effectively divide the crystal into distinct orientation states, which can be either identical or enantiomorphous in structure, differing in their inherent spontaneous strain characteristics. The researchers proposed that the transition of the sample to a cubic structure signifies complete relaxation of the spontaneous strain induced by structural distortions, resulting in the simultaneous disappearance of domain boundaries. Conversely, during cooling, as strain intensifies with decreasing temperature, domain walls emerge to minimize the total system energy. The SAED pattern before and after heating reveals a transformation of the twin relation from reflection twins to 90° rotation twins along $$ {[10 \overline{1}]} $$. They interpreted this formation of domain walls due to energy minimization during the phase transition. Consequently, $$ {[10 \overline{1}]} $$ rotation twins, possessing lower domain wall energy, are favored. Their research not only deepens our understanding of the imperfections in CsPbBr₃ but also paves the way for more refined applications in photovoltaics, offering a pathway for customization.

One of the advantages of HPs in the field of photoelectron lies in their ease of fabrication using solution methods. DMF and DMSO are two widely used polar solvents for dissolving raw materials. However, research has revealed that employing these two coordinating solvents leads to the formation of solvated intermediates^[93-95]. Consequently, Sidhoum et al. aimed to elucidate the evolution process-from precursor solution to solvated intermediates, and ultimately to perovskite-using in situ liquid and gas TEM^[96]. Initially, they filled the in situ liquid cell with toluene and then injected the perovskite precursor, resulting in precipitation. As depicted in Figure 2C, nuclei gradually appeared and grew into rod-like structures. Additionally, they conducted Small- (SAXS) and Wide-Angle X-ray scattering (WAXS) experiments, confirming that the intermediate species is 2MAI-3PbI₂-2DMF. Furthermore, they leveraged the in situ TEM chip to directly observe the phase transition of the intermediate. To simulate real experimental conditions during the transition process, they injected ambient air from the room into the TEM sample holder, maintaining a temperature of 80 °C. The phase transition process is illustrated in Figure 2D. After annealing for 45 min at 80 °C, the intermediates transformed completely into tetragonal perovskite. This work provides conclusive and unique evidence of crystal fragmentation concurrent with the structural transition from the intermediate to the final phases.

Figure 2. (A) The SAED patterns of CsPbBr₃ were captured at three different temperatures: at RT, at 90 °C, and at 133 °C, respectively; (B) TEM images show the morphology of the domain varies in situ as the temperature rises from RT to 133 °C, followed by cooling, and the corresponding SAED patterns. Reproduced with the permission of Ref.^[92] Copyright © 2020 American Chemical Society; (C) TEM images show the nucleation and growth processes of the 2MAI-3PbI₂-2DMF intermediate; (D) TEM images and corresponding SAED patterns show the phase transition process. Reproduced with the permission of Ref.^[96] Copyright © 2023 American Chemical Society; (E) DPC-STEM image of a single CsPbI₃ nanocrystal heated at 200 °C viewed from the [001] direction and the magnified images from the colored box, showing the phase transition dynamics. Reproduced with the permission of Ref.^[97] Copyright © 2023 Springer Nature. SAED: Selected area electron diffraction; RT: Room temperature; TEM: Transmission electron microscopy; DPC-STEM: Differential phase contrast-scaning transmission electron microscopy.

Previous studies investigating phase transitions primarily relied on morphological changes and SAED patterns, falling short of achieving atomic resolution characterization. However, thanks to emerging low-dose technology, genuine atomic resolution imaging can now be accomplished through in situ TEM characterization, providing a deeper understanding of the relationship between structure and performance. Ma et al. applied iDPC-STEM to observe the phase transition of CsPbI₃ nanocrystals with atomic resolution in real-time for the first time^[97]. They tracked the phase transition of α, β, and γ phases of CsPbI₃ nanocrystals at atomic scale across temperatures ranging from RT to 250 °C. More significantly, they discerned the gradual phase transition process from the surface to the center within a single CsPbI₃ nanocrystal. As depicted in Figure 2E the rotation angles of PbI₆ octahedrons, indicative of the extent of phase transition, were larger at the center than near the surfaces. This continuous distribution of the perovskite phase likely spans from γ (center) to near α (surface). Notably, this outcome suggests that the heat impact infiltrated the nanocrystal progressively rather than occurring abruptly throughout its entirety. These findings provide detailed insights into the atomic-level changes during the transition dynamics of perovskite phases, revealing a gradual transition process over a significant period, characterized by distinct intermediate stages and spatial arrangements.

In in situ TEM observations of HP phase transition processes, technical advantages bring new opportunities for in-depth understanding. Initially, researchers relied on morphological changes and SAED patterns to study these transitions. However, as low-dose technology advances, researchers can now directly capture the gradual evolution process of phase transitions with high spatial resolution using iDPC-STEM. Despite these advancements, most studies still focus primarily on the phase structure isolatedly. It is well-known that domain structures tend to disappear when a perovskite phase with high symmetry transforms into a lower-symmetry phase. This transformation can lead to the generation and interaction of lattice strain, yet its impact on perovskite material performance remains unclear. Additionally, defects may arise during the phase transition process. Therefore, it is preferred to employ a combination of imaging and analytic techniques, such as cathodoluminescence spectroscopy (CL), DPC, 4D-STEM, and EELS, to obtain more physical information on domain boundaries and defects related to perovskite performance.

Electron beam

Due to the weak ionic properties of HPs, they are susceptible to unavoidable secondary and radiolysis damage, which accelerates their degradation in TEM, thereby complicating the analysis of observed phenomena. When exposed to electron beam irradiation, halogen ions rapidly detach from the structure, and organic components readily decompose due to radiolysis, leading to structural collapse. Consequently, the primary task before conducting TEM experiments is to determine a safe dose to avoid observing distorted images. Wu et al. have summarized safe and damaging doses for different perovskite materials under TEM observation^[98]. Extensive investigation has been conducted regarding the deteriorating effects of electron beam exposure on HPs, with the prevailing notion attributing degradation to phase segregation and transition phenomena. Dang et al. conducted a research investigation to observe the effects of electron beam irradiation on the radiolysis process of CsPbBr₃^[56]. As depicted in Figure 3A, they found that with increasing electron beam dose, significant changes in CsPbBr₃ nanocrystal morphology were evident from HAADF-STEM images. Furthermore, EDX measurements indicated that the intensities of Cs and Pb remained relatively unchanged, while a decrease in Br intensity was observed, particularly at higher electron doses. The reduction in the number of Br atoms can be attributed to the desorption of ionic compounds induced by electrons. Additionally, by analyzing HRTEM and HAADF-STEM images alongside fast Fourier transform (FFT) patterns, they observed the reduction of certain Pb²⁺ ions to Pb⁰, along with spontaneous migration and clustering of Pb atoms. This study offers a microscopic perspective on the degradation patterns of inorganic perovskites when subjected to extended exposure to electron beams. Moreover, it emphasizes the importance of monitoring dynamic changes in intrinsic components and local phase evolution to effectively characterize HPs in TEM, thereby providing valuable information for regulating electron beam dosage.

Figure 3. (A) The degradation process of CsPbBr₃ nanosheets when exposed to electron beam irradiation. The perspective shown is along the [001] orientation. Reproduced with the permission of Ref.^[56] Copyright © 2017 American Chemical Society; (B) Phase segregation behavior of CsPb(Br_0.8I_0.2)₃ perovskite under beam irradiation is demonstrated through the HRTEM image and the corresponding FFT pattern. After exposure to an excess electron beam dose of 1140 e Å^-2, the formation of Pb nanoparticles is observed. Reproduced with the permission of Ref.^[99] Copyright © 2020 American Chemical Society; (C) HRTEM images with increased doses during the degradation into PbI₂ show the atomic scale images of the decomposition pathway. Reproduced with the permission of Ref.^[39] Copyright © 2021 Springer Nature. FFT: Fast fourier transform; HRTEM :High-resolution TEM; TEM: Transmission electron microscopy.

Funk et al. utilized in situ TEM to observe the CsPb(Br_0.8I_0.2)₃ perovskite in real-time, verifying the presence of a diverse phase originating from electron beam irradiation^[99]. HRTEM characterization revealed the formation of a more iodine-rich phase [approximately CsPb(Br_0.6I_0.4)₃] at the periphery of the particle, whereas a central region maintained a pristine bromide phase (CsPbBr₃) [Figure 3B]. These findings offer valuable insights into the phase separation mechanism and have the potential to guide the development of phase-stable perovskite semiconductors.

Chen et al. observed a two-step decomposition pathway of MAPbI₃ under an ultra-low dose condition in real space at the atomic scale [Figure 3C]^[39]. The initial stage of decomposition involved the loss of MA⁺ ions, followed by the collapse of the perovskite structure into 6H-PbI₂. Notably, an intermediate phase (MA_0.5PbI₃) with locally ordered vacancies was observed, demonstrating robust stability prior to the complete collapse of the perovskite structure. Additionally, the authors highlighted that C-N bonds could be readily disrupted upon irradiation, resulting in the release of NH₃ and HI and the formation of hydrocarbons. These results enhance the comprehension of the degradation process, offering detailed insights at the atomic level that contribute to understanding its core characteristics.

Heat

In situ heating TEM is a powerful technology in materials science allowing samples to be thermally stimulated for the study of dynamic properties such as catalysis, phase transitions, sublimation, and materials growth. Typically, in situ heating in TEM involves using metal resistive wires or resistive membranes deposited on SiN_x windows, which heat the sample through joule heating^[100,101]. Recently, Zhao et al. replaced traditional metal resistive heaters with 2D graphene material, significantly enhancing the performance of the in situ heating chip^[102]. Heat is an inevitable by-product during the operation of HP devices, whether they are solar cells or LEDs. This heat can significantly affect the crystal structure and ion migration due to the fragile ion properties of HPs, thereby accelerating the degradation process. Therefore, understanding the mechanism of thermally induced degradation is crucial for future studies and can guide efforts to improve stability.

Divitini et al. conducted an in situ TEM heating experiment on a non-encapsulated PSC with a typical device architecture [indium tin oxide (ITO)/TiO₂/MAPbI₃/2, 2,’7, 7’-tetrakis-(N,N’,di-p-methoxyphenylamine)- 9,9’-spirobifluorene (spiro-OMeTAD)/Ag]^[103]. They employed HAADF-STEM imaging in combination with the associated EDX elemental map to observe real-time alterations in morphology and chemical compositions [Figure 4A]. Surprisingly, the HAADF-STEM images of the perovskite layer exhibited no changes when heated to 150 °C, even though the power conversion efficiency (PCE) of the devices had irreversibly degraded above 90 °C. The researchers attributed this degradation to the inactivity of the hole transport material (HTM). As the temperature increased, holes began to form in the region of the perovskite layer close to the interface with the electron transport layer (ETL). The primary reason for the perovskite degradation was identified as the migration of iodine and lead. Furthermore, they highlighted that the device fabricated in ambient air exhibited more severe iodine migration. This study represents one of the earliest works focused on heat-induced degradation of HP materials and provides a solid foundation for subsequent research.

Figure 4. (A) In situ imaging using HAADF-STEM and characterizing with EDX reveals the variations in morphology and elemental migration of MAPbI₃ when exposed to rising temperatures. Reproduced with the permission of Ref.^[103] Copyright © 2016 Nature Research; (B) HAADF-STEM images of a PSC undergoing thermal degradation after beam damage. Reproduced with the permission of Ref.^[104] Copyright © 2016 American Chemical Society; (C) HAADF images and SAED patterns MAPbI₃ before and after the heat. Reproduced with the permission of Ref.^[105] Copyright © 2018 WILEY‐VCH Verlag GmbH & Co; KGaA (D) STEM images of a PSC sample based on MA (top line) and Cs/FA/MA (bottom line), undergoing in situ heating from room temperature to (85, 130, 150, and 170) °C. Reproduced with the permission of Ref.^[110] Copyright © Elsevier; (E) The left row shows the iDPC-STEM images demonstrating the surface degradation of a CsPbI3 nanocrystal within 1min at 200 °C. The middle row shows diagrams depicting the transformation process of CsPbI₃ into Pb and Cs₄PbI₆ under high-temperature conditions. The right row shows the iDPC-STEM images showing the Cs₄PbI₆ particles observed from the [211] orientation. Reproduced with the permission of Ref.^[97] Copyright © 2023 Springer Nature. EDX: Energy dispersive X-ray; SAED: Selected area electron diffraction; TEM: Transmission electron microscopy; STEM: Scanning TEM; iDPC: Integrated differential phase contrast; HAADF-STEM: High angle angular dark field-scanning transmission electron microscopy.

Concurrently with the work by Divitini, Yang et al. conducted similar research^[104]. As depicted in Figure 4B, when the MAPbI₃-based PSC device was heated for 4 h at approximately 50-60 °C, defects began to appear. This phenomenon was attributed to the migration of elements induced by thermal effects, which was expedited by the presence of Schottky or Frenkel defects in the crystal structure. Additionally, they pointed out that the oxygen tended to attack the CH₃NH₃⁺ rather than the PbI₆ component of perovskites when exposing them to the ambient air at RT. Their work provided significant insights into preventing perovskite degradation by tailoring its molecular structure, particularly the organic moiety.

Furthermore, Kim et al. proposed a new degradation pathway for the MAPbI₃ layer. They suggested that the formation of PbI₂ grains did not result directly from the crystalline MAPbI₃ layer, but instead from the amorphous MAPbI₃ layer. This contrasts with conventional ideas that posit the formation of PbI₂ starting from the grain boundary or interface^[105]. In their observation, the precipitation of the trigonal PbI₂ grains occurred from the amorphized MAPbI₃ layer via an asymmetric intermediate state [Figure 4C]. This new insight contributes to a more comprehensive understanding of the degradation mechanism.

Later, Seo et al. noted that previous studies have primarily focused on the forward (n-i-p) device structure based on the single MA cation perovskite material. However, the MA-FA-Cs mixed cation perovskite material has shown more promise for commercialization than the single MA cation material. Additionally, the Hole transport layer (HTL) materials used in the forward structure devices, such as spiro-OMeTAD or poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA), commonly contain various dopants and additives. These dopants, which implicate ions that are hygroscopic and/or mobile, may negatively affect the stability of the perovskite absorber and the device, making it challenging to identify the exact causes of perovskite degradation^[106-110]. Therefore, they studied the thermally-induced effects on the photovoltaic performances of Cs_x(FA_yMA_1-y)_1-xPb(I_zBr_1-z)₃ PSCs in comparison with MAPbI₃ PSCs, and further investigated the degradation phenomenon directly utilizing real-time in situ TEM^[110]. They opted for the inverted (p-i-n) device structure, composed of ITO/NiO NPs/perovskite/PC61BM/ZnO NPs/Ag, to avoid the impact of the aforementioned organic HTL materials. As illustrated in Figure 4D, the perovskite device based on MA demonstrated a rapid and uninterrupted deterioration process as the temperature increased, characterized by the emergence of particles and dark voids. In contrast, the device incorporating Cs/FA/MA remained stable until it reached 130 °C, with particle formation commencing at 150 °C. However, the degradation of the Cs/FA/MA-based device intensified at 170 °C, resulting in the formation of sizable particles and voids. Common factors contributing to degradation, such as decomposed Pb and I migration, were observed in both types of PSCs. The MA-PSC displayed accelerated void formation and migration extending beyond the perovskite film as the temperature increased. Conversely, the Cs/FA/MA-PSC demonstrated negligible changes until it reached 150 °C but subsequently exhibited increased elemental migration and the emergence of large Pb-based particles and voids near the ETL side at 170 °C. The researchers attributed the deterioration to the creation of volatile cations, namely MA⁺ and FA⁺. They noted that the better thermal stability of FA could mitigate the deterioration of device performance. Additionally, the degradation mechanisms of these two types of devices exhibited slight variations. By analyzing EDX elemental mapping and lattice planes at the particle regions, the researchers observed the coexistence of two types of Pb and PbI₂ in the Cs/FA/MA device, while the MA-based device only exhibited PbI₂. They proposed that this phenomenon resulted from the formation of Br vacancies in the crystal structure, inducing the conversion of Pb²⁺ to Pb⁰ and the subsequent formation of Pb-Pb bonds, leading to the generation of Pb particles in the Cs/FA/MA-based device. Consequently, this study offers novel insights into comprehending the degradation of perovskite materials.

Previous studies on thermally induced degradation of HPs primarily focused on morphological changes and could not achieve the atomic resolution characterization. However, with emerging low-dose technology, genuine atomic resolution imaging can now be achieved through in situ TEM characterization. Ma et al. observed the degradation process of single CsPbI₃ nanocrystals at the atomic scale. As shown in Figure 4E, heating the CsPbI₃ nanocrystals at 200 °C for 1 min caused the flat (110) surfaces in the particles to gradually disappear, exposing more (010) surfaces. With continued heating, they noted that the cubic CsPbI₃ nanocrystals started to deviate from their uniform structure and slowly merged into one another at temperatures over 200 °C. Combining the EDX results, they deduced that the conversion products at 300 °C may be pure Pb and Cs₄PbI₆ particles, whereas at 400 °C, only Pb is observed. Figure 4E shows the diagram of decomposition products and confirms its Cs₄PbI₆ structure. In another work, they also observed the thermal-induced degradation of CsPbI₃ single crystal^[111]. They confirmed that thermal vibrations at high temperatures disrupted the asymmetric binding of surface ions, resulting in the extraction of these ions-either by breaking Cs-surfactant complexes or collectively pulling them out. This process initiated the degradation and fusion of CsPbI₃ QDs. These studies not only uncover the long-term, previously unclear dynamics of thermally induced phase transitions in HPs but also instill greater optimism and assurance for investigating further physical and chemical phenomena in HPs at the atomic scale.

Electrical bias

Electrical bias plays a crucial role in the operation of perovskite devices, where solar cells generate a built-in electric field, and LEDs are directly driven by electrical bias. Researchers apply electrical bias to samples using two primary methods. The first method involves a MEMS chip platform, where the specimen is fabricated using FIB technology and then attached to a chip installed on a specially designed holder. The source measurement unit (SMU) outside the TEM controls this setup^[112]. The second approach utilizes a micro-metal probe that makes direct contact with the sample, driven by a small piezo-driver^[113]. Various studies have explored the impact of bias on HPs and devices, focusing on changes in ion migration, phase structure, morphology, and other behaviors through in situ TEM observation. These investigations provide valuable insights into the behavior of perovskite materials under electrical bias.

Jeangros et al. utilized in situ TEM in combination with a microchip setup to deliver a constant electrical bias to a TEM lamella of a nano solar cell^[112]. The lamella consisted of active layers of MAPbI₃, charge carrier transport layers, and ITO and Au electrodes. Building on this foundation, they provided an in situ TEM observation that revealed the nanoscale degradation mechanisms of MAPbI₃ perovskite cells caused by bias. As shown in Figure 5A, TEM imaging and EDX analysis demonstrated that the degradation mechanisms of MAPbI₃ under bias were varied, primarily localized at the interface influenced by a positive bias voltage. Particularly in the context of forward conditions, iodide ions migrated toward the positively charged charge transport layer, while also evaporating alongside organic compounds within the hole transport layer. This process initiated the formation of PbI₂ NPs and empty spaces, ultimately resulting in the deterioration of the device’s operational efficiency. This study provides an initial microscopic understanding of the failure mechanisms of HP devices when subjected to bias, including the development of microscopic appearance and the dynamic mechanism of ion migration. However, studies investigating atomic scale behavior in this regard are still lacking and require further advancements.

Figure 5. (A) Scanning electron microscopy (SEM) image depiction of a biasing chip connected to a TEM lamella prepared using FIB for conducting in situ electrical measurements. The accompanying HAADF-STEM image along with an EDX mapping showcases the initial microstructure. Acquisition of variable J-V curves through TEM observation, coupled with HAADF images, demonstrates the morphological evolution at different electrical biases. Reproduced with the permission of Ref.^[112] Copyright © 2016 American Chemical Society; (B) HAADF-STEM, Bright Field TEM (BFTEM), and SAED were employed for the real-time monitoring of the morphology and structural variations of MAPbI₃ under in situ electrical bias. Reproduced with the permission of Ref.^[114] Copyright © 2018 Wiley-VCH; (C) A diagram illustrating the arrangement of the in situ TEM sample with an applied electrical bias, accompanied by an SEM image of the nano-solar cell lamella prepared using FIB. The scale bar represents 2 μm. Sequential HRTEM images of perovskite subjected to an electrical bias of 1 V. Reproduced with the permission of Ref.^[115] Copyright © 2021 American Chemical Society. EDX: Energy dispersive X-ray; TEM: Transmission electron microscopy; FIB: Focused ion beam; HRTEM :High-resolution TEM; HAADF-STEM: High angle angular dark field-scanning transmission electron microscopy.

Jung et al. utilized in situ TEM to observe the oxygen diffusion process under bias, aiming to understand the degradation mechanism of HP devices^[114]. Interestingly, they discovered that under forward bias, oxygen ions from the TiO₂ layers migrated into the MAPbI₃, causing a phase transformation from crystalline to amorphous [Figure 5B]. This severe structural change was directly attributed to the electric field-driven oxygen migration. To quantify the oxygen content, EELS measurement was conducted, revealing that 8 atomic percent (at%) of oxygen was incorporated into the MAPbI₃. This incorporation could potentially result in the degradation of HPs. Notably, these alterations were somewhat reversible in the absence of light and bias. However, the authors suggested that isolating the effects of moisture and oxygen alone could not fully mitigate this internally driven degradation. Overall, this study provides valuable insights into the degradation mechanism of HP devices, particularly regarding oxygen migration and TiO₂ degradation, observed through in situ TEM.

Kim et al. developed an in situ TEM configuration for observing the amorphization progression of HP layers when subjected to electrical bias^[115]. The experimental setup [Figure 5C] enabled the researchers to observe the alterations in the structure in a nanoscale solar cell lamella subjected to a steady positive bias of 1 V. Series of TEM images captured over time vividly demonstrate the amorphization process of the HPs layers when exposed to the bias. Additionally, the morphology and phase transformation can be understood by analyzing HRTEM images and SAED patterns, which reveal the emergence of a diffuse amorphous zone. The X-ray diffraction (XRD) patterns of the perovskite materials, before and after the implementation of forward bias, further support this observation. It has been proposed that the inherent structural changes in HPs could play a role in the deterioration of device performance. Intriguingly, subjecting the samples to a temperature of 50 °C to facilitate the recrystallization process has been shown to successfully restore the lost performance. This study delves into the underlying causes of the structural instability of PSCs, particularly focusing on the fundamental mechanism of defect formation and ion migration. The observation of structural transformation at the atomic scale in real-time provides valuable insights into these factors, underscoring the considerable promise of in situ TEM technology in high-pressure research.

Light

Light plays a crucial role in the operation of PSCs. Therefore, investigating changes in the morphology and structure of perovskite materials under light is a promising avenue. This approach provides a comprehensive understanding of the relationship between structure, properties, and performance, ultimately facilitating the development of light-responsive materials-based technologies. Various techniques have been proposed to incorporate light into TEM, including the use of a light path setup, a TEM-STEM probe holder, and MEMS chips^[116]. The light path setup, which consists of mirrors and an optical lens, directs light onto the sample inside the TEM. However, implementing such setups requires significant modifications to the TEM system. Fortunately, advancements in MEMS technology have allowed for the integration of compact MEMS chips into the in situ optical TEM holder. In this configuration, the light source is positioned in front of the specimen, while the circuitry is integrated within the chips^[117].

Our research group pioneered a novel on-chip technique known as light-incorporated in situ TEM (LI²ST) for studying HPs at the nanoscale^[118]. This method enables real-time tracking of structural transformations triggered by light at specific locations. The in situ platform is based on a MEMS chip, specifically designed for adjustable light illumination with customizable multi-wavelength capabilities [Figure 6A]. We fabricated perovskite samples using the thermal evaporation method. Initially, we captured SAED patterns from neighboring CsPbBr₃ grains, revealing the crystallographic structure in its orthorhombic form along the [112] zone axis (indicated by the white circle) without light stimulation. No significant changes were observed after 60 and 120 min. Subsequently, we repeated the experiment with the light turned on. The initial SAED pattern from another pair of neighboring CsPbBr₃ grains showed a consistent orthorhombic phase with a [100] zone axis. After 60 min of light exposure, a new diffraction spot emerged (indicated by the blue circle), and the intensity of these additional diffraction spots increased further after 120 min [Figure 6B], indicating the formation of new phases induced by the light stimulus. Our calculations suggest that the newly formed phase is PbBr₂. To obtain detailed atomic information and observe microstructural changes, we employed aberration-corrected STEM for ex situ high-resolution imaging [Figure 6C]. After two hours of light exposure, nanoclusters appeared in the HAADF-STEM image, which we attributed to PbBr₂ generation. Interestingly, these PbBr₂ nanoclusters were primarily located within the intragrain regions rather than at the grain boundaries. Extending the illumination period to 12 h resulted in significantly larger nanoclusters, suggesting that localized expansion of PbBr₂ nanoclusters induces perovskite degradation. Our work proposes a distinctive microstructural mechanism for HP degradation under light illumination, attributed to ion migration induced by the photo effect.

Figure 6. (A)The schematic diagram of LI²ST (B) The TEM images and corresponding SAED patterns of HPs film in the light-on state and light-off state for 0, 60, and 120 min, respectively (C) HAADF-STEM images of the HP sample under different illumination times (0, 2, and 12 h) reveal the occurrence of intragrain nucleation and growth of PbBr₂ impurity clusters. (marked using white arrows). Reproduced with the permission of Ref.^[118] Copyright © 2023 American Chemical Society. TEM: Transmission electron microscopy; SAED: Selected area electron diffraction; HPs: Halide perovskites; HAADF-STEM: High angle angular dark field-scanning transmission electron microscopy.

However, due to the limitations of the TEM available at that time, we could only observe changes in the material’s morphology and diffraction pattern during the in situ experiment. Capturing the evolution process of phase transition at the atomic scale can provide us with a deeper understanding of the decomposition mechanism. Looking ahead, the integration of 4D-STEM technology could offer additional opportunities and enhanced capabilities for observing microstructural changes at the atomic scale.

Ambient atmosphere

We have mentioned that materials exhibit distinct behaviors under various atmospheric conditions compared to a high vacuum. In today’s context, moisture-induced chemical degradation presents a significant challenge to the advancement of HP technology. This degradation can result in the rapid decomposition of the perovskite material, potentially leading to lead dissolution into the environment. Consequently, it is essential to conduct thorough investigations into the molecular-level stability and degradation of materials across different environmental conditions. This knowledge will play a pivotal role in developing more stable compositions and designing robust device architectures.

For the first time, the degradation mechanisms of MAPbI₃ under different atmospheres were investigated using in situ gas-cell TEM by Fan et al.^[119]. The experimental setup maintained a controlled atmosphere of either vacuum or dry air, with a chamber pressure of 700 Torr during the control experiment. Instead of focusing on the FIB-prepared full device cross-section specimen, their attention was directed towards the atomic structural changes of the HPs, which provide deeper insights into the degradation mechanisms of perovskite materials. As shown in Figure 7A and B, traces of PbI₂ were observed after subjecting the sample to a temperature of 85 °C for 400 seconds under vacuum conditions. Similarly, Figure 7C demonstrates the same evolution pathway for MAPbI₃ under dry air. This evolution involved the gradual transformation of the crystalline structure of MAPbI₃ perovskite from tetragonal MAPbI₃ to trigonal PbI₂ layered crystals along a specific crystallographic direction. Consequently, the researchers concluded that thermal effects alone are responsible for the degradation of the MAPbI₃ perovskite in both dry air and ambient vacuum conditions. They also emphasized that surface degradation initiates crystal structure transitions through a low energy barrier pathway, implying a degradation process that occurs layer by layer, starting from the crystal surface. Additionally, they fabricated an encapsulation layer on the MAPbI₃ surface to suppress the layer-by-layer thermal degradation and achieved promising results. In this work, they proposed a new thermal degradation model, suggesting novel strategies for enhancing the stability of perovskite materials.

Figure 7. (A) HRTEM image and SAED pattern of MAPbI₃ before heating under vacuum; (B) HRTEM image and SED pattern of MAPbI₃ after heating at 85 °C for 400 s under vacuum; (C) HRTEM image and SED pattern of an individual MAPbI₃ showing the heat-induced degradation process under a dry air atmosphere. Reproduced with the permission of Ref.^[119] Copyright © 2017 Elsevier; (D) In situ liquid-cell TEM monitoring the transformation of MAPbI₃ to MAPbI₃·H₂O and PbI₂ (top line), as well as the dissolution of MAPbI₃·H₂O and subsequent recrystallization into PbI₂ (bottom line). Reproduced with the permission of Ref.^[120] Copyright © 2021 John Wiley & Sons. HRTEM :High-resolution TEM; TEM: Transmission electron microscopy; SAED: Selected area electron diffraction; SED: Scanning ED; ED: Electron diffraction.

Akhavan Kazemi et al. investigated the degradation mechanism of MAPbI₃ caused by moisture using in situ liquid-cell TEM and in situ XRD techniques at 85% Relative Humidity (RH)^[120]. Figure 7D provides complementary insights into the chemical transformation from MAPbI₃ to MAPbI₃·H₂O, a finding corroborated by 2D solid-state nuclear magnetic resonance (ssNMR) measurements. The researchers highlighted that the chemical transformation from MAPbI₃ to MAPbI₃·H₂O occurs via a front-phase migration mechanism at the micrometer scale. The formation of PbI₂ originates from the nanoscale dissolution of MAPbI₃·H₂O and the recrystallization of larger PbI₂ particles. Initially, the PbI₂ particles nucleate and grow isotopically before forming bulgy particles. Subsequently, they grow preferentially in a specific direction, resulting in the formation of more elongated PbI₂ particles and eventually leading to dendritic growth. This study represents the first time that live observations of PSCs under humid conditions have been conducted through in situ liquid-cell TEM experiments, providing a real-time and clear evaluation of the reaction mechanisms at the nanoscale.

In this chapter, we summarize the applications of in situ TEM in studying the degradation mechanism of HPs. The first crucial consideration when investigating degradation mechanisms is to ensure that the material remains undamaged by the electron beam. Initially, researchers primarily relied on observing morphological changes, SAED patterns, and Energy dispersive X-ray spectroscopy (EDS) mapping. However, EDS mapping sometimes fails to provide useful information due to the low count rate of perovskite materials. Fortunately, advancements in low-dose technology now allow us to observe structural changes at the atomic scale in situ. Nevertheless, observing perovskite materials with high organic content in the A site remains challenging. High-efficiency PSCs typically contain more than 90% organic content. Additionally, the detailed degradation pathways differ significantly between inorganic and organic perovskites. While ion substitutions in the B and X sites primarily modulate the bandgap, mixed ions in the X site may lead to phase separation^[121]. However, the substitution of different organic ions in the A site significantly influences stability. Among all organic and inorganic monovalent cations, MA⁺ possesses the most appropriate effective radius for achieving proper phase stability in HPs. Nonetheless, it exhibits chemical instability. Some literature suggests that MA⁺ undergoes deprotonation in the presence of O₂/H₂O when leaving the crystal in the atmosphere^[122-124]. Regarding FA⁺, its reduced acidity, greater mass, and smaller dipole moment contribute to its superior chemical stability compared to MA⁺. Consequently, FAPbI₃ exhibits notably enhanced thermal stability in an inert environment when contrasted with MAPbI₃. Additionally, Cs⁺ cations exhibit even higher chemical stability than FA due to their non-volatility and resistance to deprotonation^[125]. However, at RT, the cubic α-phase of FAPbI₃ and the orthorhombic γ-phase of CsPbI₃ are not thermodynamically stable; their stable forms are represented by the non-perovskite, non-photoactive δ-phases^[85]. It is worth noting that to realize a better stabilization effect, some literature has reported that triple-cation recipes such as FA_0.79MA_0.16Cs_0.05Pb(I_0.83Br_0.17)₃ usually show higher stability, which has been applied in recent studies^[126,127]. In previous in situ TEM investigations of HPs, MAPbI₃ is the most studied material, and more research should focus on the mixed HPs system. To illustrate, when investigating perovskite degradation under various atmospheres, both articles focused on MAPbI₃ as the research subject. In addition, due to the low endurance of organic cations to electron beam-induced damage, it is also important to facilitate the advancements in low-dose technologies to realize the in situ observations of the organic perovskites’ structural evolutions without unwanted damage.