Fig. 1. (Color online) Schematic illustration of advanced TEM characterization for halide perovskites.

Fig. 2. (Color online) (a) Electron beam irradiation damage observed in free-standing MAPbI₃ films. (i, ii). TEM images recorded initially and after the irradiation (9870 e/(Ų·s) for ~2 min), respectively^[47]. (b) Time-series of TEM images on MAPbI₃ single crystal showing the electron beam damage from 0 to 50 s, where bubble-like morphology (colored arrows) emerged and grew^[49]. (c) Time-series of TEM images obtained on BA₂PbBr₄ nanosheets^[53]. (d) TEM images of the (i) CsPbCl_3, (ii) CsPbBr₃ and (iii) CsPbI₃QDs where "dark spots" present at the QD corners indicated irradiation damage^[54]. (e) Schematic illustration of CsPbBr₃ degradation pathway^[56].

Fig. 3. (Color online) (a) Degradation of MAPbI₃ studied using SAED taken from a near-<1 0>_t-oriented grain: i) the initial, pristine phase and ii) after 1 min (total dose per area of ≈1 × 10² e/Ų), iii) 2 min (total dose per area of ≈2 × 10² e/Ų), iv) 18 min (total dose per area of ≈2 × 10³ e/Ų) of weak electron beam exposure (≈2 e/Ų)^[46]. (b) Degradation in MAPbX₃ by forming superstructured intermediate phase: i) atomistic structure of tetragonal MAPbI₃; ii) electron diffraction (ED) pattern along the [001]_c direction; iii) the observed ED of superstructure phase; iv) the simulated ED of superstructure phase MAPbI_2.5; v) the corresponding atomistic structure; vi) atomistic structure of MAPbBr₃; vii) ED pattern along the [001] direction; viii) the observed ED pattern with additional reflections; ix) the simulated ED of superstructure phase MAPbBr_2.5; x) the corresponding atomistic structure with ordered bromine vacancies^[69]. (c) TEM images and [110] oriented-SAED patterns taken from grain highlighted in yellow circles from FAPbI₃ films with (i, ii) 10% MA, (iii, iv) 20% MA, (v, vi) 30% MA, (vii, viii) 40% MA^[78]. (d) Stabilization of photoactive perovskites against degradation by tilted octahedral, as illustrated by structural model (i–vi), calculated energy difference (vii), AFM-IR characterization (viii–x), and TEM imaging (xi) with corresponding SAED (xii–l)^[79].

Fig. 4. (Color online) (a) HRTEM of CsPbBr₃ nanocrystals (i) where the coexistence of cubic and orthorhombic phases were demonstrated by FFT patterns (ii, iii), simulated diffraction patterns (iv, v), and illustrated structure (vi, vii)^[41]. (b) CTF-corrected denoised HRTEM image (i) of CH₃NH₃PbBr₃ with different CH₃NH₃ orientations, where (ii, iii) the structural model (left) and the simulated projected potential map (right) corresponding to region 1 and 2 in (i), respectively^[39]. (c) Ptychography reconstructed image of CsPbBr₃, with the scale bar of 5Å^[84]. (d) Atomic-scale structures of intragrain stacking-fault (i) and twinning interfaces (ii) obtained on orthorhombic FA_0.5Cs_0.5PbI₃ grains along the [100] projection direction^[86]. (e) Atomically resolved interface at the (2T)₂ PbI₄–(2T)₂ PbI₄–(2T)₂ PbBr₄ heterostructure^[87]. (f) Butterworth-filtered LAADF-STEM images of grain boundaries (i), triple junctions (ii), grain boundary (iii) and aligned vacancy defects indicated by red circle (iv), obtained from a 30-nm-thick film of FAPbI₃^[67].

Fig. 5. (Color online) (a, b) Atomically resolved HAADF-STEM images and corresponding EDX-mappings of CsPbBr₃ nanoplates^[92]. (c) STEM-EELS from a CsPbBr₃ nanosheet to determine bandgap, where (i) demonstrates measured data and (ii) shows as-calculated bandgap value^[93].

Fig. 6. (Color online) (a) Stacking faults observed in a MAPbI₃ with corresponding FFT patterns as inset (i), and corresponding magnified HRTEM (ii, v, vi) with structural model (iii) and (iv) the simulated HRTEM image^[96]. (b) Atomically resolved-cryo-TEM image of aged MAPbI₃ collected at a low dose condition (electron dose, ~5.96 e/Ų), with corresponding enlargement (ii, iv), structural model (iii) and polarization map (v)^[40].

Fig. 7. (Color online) (a) In-situ heating of MAPbI₃ based PSCs up to 250 °C, where the temperature evolution of morphology change and elemental migration was monitored by HAADF-STEM images and EDX mappings. The same scale bar applies to all panels^[42]. (b) In-situ electrical biasing on MAPbI₃, where morphology and structure change was monitored by HAADF-STEM, TEM and SAED, respectively^[109]. (c) In-situ TEM showing the impact of controlled humidity on the conversion of MAPbI₃ into MAPbI₃·H₂O and finally PbI₂, using liquid cell^[112].

Fig. 8. (Color online) Illustrated summary of safe dose and damage dose for different perovskite materials, plotted in coloured columns. Numbers in the figure correspond to the reference numbers as listed in the tables and references. Shade in each column indicates relatively-safe dose range versus damage-prone dose range. Generally speaking, 2D pvsk is suggested to be imaged below the dose of 50 e/(Ų·s), MAPbI₃/ MAPbBr₃ below ~100 e/(Ų·s), whereas CsPbBr₃ can tolerate dose up to more than 1000 e/(Ų·s).

Table 1. Summary of TEM characterization details for MAPbI₃ and related structures. No damage or little damage was reported unless specified in the table.

Table 2. Summary of TEM characterization details for MAPbBr₃. No damage or little damage was reported unless specified in the table.

Table 3. Summary of TEM characterization details for all-inorganic CsPbBr₃ and related structures. No damage or little damage was reported unless specified in the table.

Table 4. Summary of TEM characterization details for 2D halide perovskites. No damage or little damage was reported unless specified in the table.