CoChat

Multimodal electron microscopy of halide perovskite interfacial dynamics

Open access

Published version of a preprint

The DOI matches its Crossref record; 56 of the 56 references it cites were found in a registry or an index; 7 of the 30 sources its claims cite were checked for support.

Abstract

Abstract Halide perovskite light-emitting diodes promise high-efficiency 1–3 , low-cost optoelectronics, yet their operational instability remains a critical barrier to practical deployment. Here we develop a multimodal in situ electron microscopy approach that integrates four-dimensional scanning transmission electron microscopy, energy-dispersive X-ray spectroscopy and atomic-resolution imaging to directly visualize structural and chemical evolution in a working halide perovskite light-emitting diode with nanometre precision. Our in situ biasing measurements uncover nanoscale structural and chemical transformations initiated at transport layer interfaces, including the formation of metallic lead and lead-rich secondary phases, as well as strain-driven grain fragmentation. On biasing, we observe the partial transformation of the metallic Al contact to insulating AlCl 3 . Crucially, whereas the bulk of the perovskite emitter remains relatively intact, our experiment shows that degradation is localized at interfaces. By comparing in situ and ex situ measurements, these results establish a mechanistic link between interfacial strain, ionic transport and electrochemical reactions in working devices, and provide a broadly applicable framework for nanoscale degradation analysis in complex multilayered optoelectronic systems using multimodal in situ biasing microscopy.

Overview

CoChat Summary

## TL;DR

This paper uses in situ multimodal electron microscopy to follow a working halide perovskite LED under electrical bias and finds that failure is driven mainly by interfacial, not bulk, degradation. The authors link this degradation to strain, ion migration, formation of lead-rich phases, and corrosion of the Al contact into AlCl3.

## Key findings

  • A sky-blue PeLED showed an initial electroluminescence peak at 485 nm, a 15 nm full-width at half-maximum, and a peak luminance exceeding 9,500 cd m−2 at 10.8 V; its operational lifetime (T50) was ~3 min at 1 mA cm−2.
  • Under 0.2 nA constant current bias, low-dose 4D-STEM datasets were collected at 0, 3, 6, 15, 25, and 35 min, revealing a substantial nonlinear increase in driving voltage, consistent with rising device resistivity.
  • Biasing caused the emitter/transport-layer interfaces to lose orthorhombic CsPbX3 and form lead-rich phases, primarily CsPb2X5 and PbX2; by 35 min, the paper reports substantial degradation at both ETL/emitter and emitter/HTL interfaces.
  • The cathode also degraded: AlCl3 diffraction peaks appeared after 6 min of biasing, and the AlCl3 layer grew to around 10.8 nm after 35 min.
  • The authors report that the emitter bulk remained relatively intact, and conclude that interfacial reactions, rather than bulk perovskite decomposition, mainly account for electroluminescence loss.

## Methods

  • The study used a cross-sectional nanoLED PeLED mounted on a MEMS chip for in situ electrical biasing inside an aberration-corrected electron microscope.
  • The device architecture included ITO/PTAA/PVK/PVP/Cs1.2FA0.3Pb(Br0.65Cl0.35)3/TPBi/LiF/Al; the FIB-prepared lamella was about 100 nm thick.
  • Main analyses combined low-dose 4D-STEM, STEM-EDX, atomic-resolution HAADF-STEM, geometric phase analysis for strain, and NMF / template-matching diffraction indexing to track structural and compositional changes over time.

## Limitations

  • The paper states that Joule heating was not directly measured; it is discussed as a likely contributing factor based on correlations in the data.
  • The paper notes that the work was performed on a lamella/nanoLED geometry and then validated with ex situ bulk-device measurements; this makes some observations potentially sensitive to geometry, although the authors argue the mechanisms are intrinsic.
  • The text makes evident that the study is focused on a single device class and composition; broader generalization to other perovskite LED architectures is not stated in the paper.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41586-026-10238-8

Cites

56 of 56 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 In situ biasing multimodal electron microscopy of a perovskite blue light-emitting diode. a, Schematic diagram of the STEM detectors used for the galvanostatic study of the nanoLED device, with biasing configuration of the Al/LiF and indium tin oxide (ITO) contacts. The nanoLED cross-section is 40 μm wide and 100 nm thick. b, Current–voltage and luminance characteristics of the pristine device, showing an emission peak at 485 nm (inset). c, STEM-HAADF cross-section of the nanoLED device stack. d,e, Atomic-resolution STEM-HAADF image (d) of the perovskite emitter with the corresponding fast Fourier transform (e), showing an orthorhombic structure with a (020) interplanar spacing of 0.58 nm. f, Ideal orthorhombic lattice structure of the perovskite emissive layer seen along the [101] zone axis. Scale bars, 200 nm (c), 1 nm (d), 2 nm−1 (e). Source data
Fig. 1 In situ biasing multimodal electron microscopy of a perovskite blue light-emitting diode. a, Schematic diagram of the STEM detectors used for the galvanostatic study of the nanoLED device, with biasing configuration of the Al/LiF and indium tin oxide (ITO) contacts. The nanoLED cross-section is 40 μm wide and 100 nm thick. b, Current–voltage and luminance characteristics of the pristine device, showing an emission peak at 485 nm (inset). c, STEM-HAADF cross-section of the nanoLED device stack. d,e, Atomic-resolution STEM-HAADF image (d) of the perovskite emitter with the corresponding fast Fourier transform (e), showing an orthorhombic structure with a (020) interplanar spacing of 0.58 nm. f, Ideal orthorhombic lattice structure of the perovskite emissive layer seen along the [101] zone axis. Scale bars, 200 nm (c), 1 nm (d), 2 nm−1 (e). Source dataCC BY · PubMed Central
Fig. 2 Interfacial strain and structural distortion revealed by geometric phase analysis. a–c, Atomic-resolution HAADF-STEM images of the DCDH emitter at the ETL interface (a), emitter centre (b) and HTL interface (c). Boxes mark regions of elevated HAADF intensity that co-localize with Pb-rich products (for example, metallic Pb, PbX2, CsPb2X5), indicative of interfacial compositional heterogeneity. d–i, Corresponding geometric phase analysis strain maps showing the out-of-plane (εyy; y direction set by the FIB lamella geometry) (d–f) and in-plane (εxx; x direction) (g–i) components, corresponding to the ETL interface (d,g), emitter centre (e,h) and HTL interface (f,i), respectively. Red (positive) denotes tensile strain and blue (negative) denotes compressive strain; shear components are provided in Extended Data Fig. 1a–c. Localized tensile zones are observed adjacent to the Pb-rich regions (boxed), consistent with lattice expansion driven by compositional inhomogeneity, whereas compressive deformation near the ETL/emitter interface (d,g) suggests chemically induced lattice shrinkage. The emitter centre (b,e,h) remains largely undistorted under pristine conditions, whereas both interfaces exhibit misalignment, defects and inhomogeneity. An axis compass on panels indicates in-plane (x) and out-of-plane (y) directions used for strain analysis. Scale bars, 2 nm.
Fig. 2 Interfacial strain and structural distortion revealed by geometric phase analysis. a–c, Atomic-resolution HAADF-STEM images of the DCDH emitter at the ETL interface (a), emitter centre (b) and HTL interface (c). Boxes mark regions of elevated HAADF intensity that co-localize with Pb-rich products (for example, metallic Pb, PbX2, CsPb2X5), indicative of interfacial compositional heterogeneity. d–i, Corresponding geometric phase analysis strain maps showing the out-of-plane (εyy; y direction set by the FIB lamella geometry) (d–f) and in-plane (εxx; x direction) (g–i) components, corresponding to the ETL interface (d,g), emitter centre (e,h) and HTL interface (f,i), respectively. Red (positive) denotes tensile strain and blue (negative) denotes compressive strain; shear components are provided in Extended Data Fig. 1a–c. Localized tensile zones are observed adjacent to the Pb-rich regions (boxed), consistent with lattice expansion driven by compositional inhomogeneity, whereas compressive deformation near the ETL/emitter interface (d,g) suggests chemically induced lattice shrinkage. The emitter centre (b,e,h) remains largely undistorted under pristine conditions, whereas both interfaces exhibit misalignment, defects and inhomogeneity. An axis compass on panels indicates in-plane (x) and out-of-plane (y) directions used for strain analysis. Scale bars, 2 nm.CC BY · PubMed Central
Fig. 3 Multimodal analysis of structural and chemical degradation of a DCDH nanoLED under electrical biasing. a, Plot of voltage versus time required to drive the nanoLED at a constant current of 0.2 nA, for each biasing condition. b, Summed one-dimensional radial electron diffraction patterns for different biasing times, showing the main diffraction features under increasing biasing time at constant current. The red dots highlight diffraction peaks corresponding to Pb-rich phases. c, Calculated lattice parameters across all biasing conditions, revealing progressive distortion of the perovskite lattice. The reported uncertainties (shading) correspond to one standard deviation derived from the covariance matrix of the refinement. d, STEM-HAADF of the emitter cross-sections (pristine) from 0 to 35 min biasing, and corresponding vDF images taken after biasing from the same area. Panels labelled ‘pristine’ are low-magnification HAADF context crops taken from a single parent overview of the lamella at t = 0; post-bias images at each time point were acquired once in fresh, distinct regions of interest (ROIs) to avoid cumulative beam dose. e, The intensity–mass loss and cross-sectional area loss quantified by analysing changes in areas of emissive layer and local image intensity. Cross-sectional area loss refers to the projected emitter area decrease after rigid registration; intensity (‘mass’) loss is HAADF-integrated intensity within the emitter mask, normalized to ETL as an internal reference to mitigate thickness and brightness drift. f, Mean intragrain size and mean grain orientation spread (GOS), showing structural evolution with biasing. Error bars represent the standard deviation of the intragrain distribution within each ROI. g, Virtual bright-field (vBF) image of the nanoLED stack after 35 min biasing. h–j, NMF components (factors and loadings), to show the spatial distribution of the original DCDH stoichiometry (h), and of phases that evolve during biasing, such as Pb-rich degradation products (i) and AlCl3 (j). Scale bars, 100 nm (d,g). Source data
Fig. 3 Multimodal analysis of structural and chemical degradation of a DCDH nanoLED under electrical biasing. a, Plot of voltage versus time required to drive the nanoLED at a constant current of 0.2 nA, for each biasing condition. b, Summed one-dimensional radial electron diffraction patterns for different biasing times, showing the main diffraction features under increasing biasing time at constant current. The red dots highlight diffraction peaks corresponding to Pb-rich phases. c, Calculated lattice parameters across all biasing conditions, revealing progressive distortion of the perovskite lattice. The reported uncertainties (shading) correspond to one standard deviation derived from the covariance matrix of the refinement. d, STEM-HAADF of the emitter cross-sections (pristine) from 0 to 35 min biasing, and corresponding vDF images taken after biasing from the same area. Panels labelled ‘pristine’ are low-magnification HAADF context crops taken from a single parent overview of the lamella at t = 0; post-bias images at each time point were acquired once in fresh, distinct regions of interest (ROIs) to avoid cumulative beam dose. e, The intensity–mass loss and cross-sectional area loss quantified by analysing changes in areas of emissive layer and local image intensity. Cross-sectional area loss refers to the projected emitter area decrease after rigid registration; intensity (‘mass’) loss is HAADF-integrated intensity within the emitter mask, normalized to ETL as an internal reference to mitigate thickness and brightness drift. f, Mean intragrain size and mean grain orientation spread (GOS), showing structural evolution with biasing. Error bars represent the standard deviation of the intragrain distribution within each ROI. g, Virtual bright-field (vBF) image of the nanoLED stack after 35 min biasing. h–j, NMF components (factors and loadings), to show the spatial distribution of the original DCDH stoichiometry (h), and of phases that evolve during biasing, such as Pb-rich degradation products (i) and AlCl3 (j). Scale bars, 100 nm (d,g). Source dataCC BY · PubMed Central
Fig. 4 Interfacial degradation at transport layer–perovskite emitter interfaces under bias. a–d, Summed diffraction patterns (150 pixels) at the ETL/emitter and HTL/emitter interfaces under pristine conditions (tbias = 0 min) and after 35 min under bias. Red labels denote diffraction features assigned to PbX2; yellow labels indicate reflections matched to the pristine orthorhombic perovskite; green labels denote CsPb2X5. a,b, Show a good match to the orthorhombic perovskite emitter near the ETL (a) and evidence of PbX2 formation near the HTL (b). c,d, Pb-rich phases (CsPb2X5, PbX2) emerge at the ETL/emitter interface (c) and the emitter/HTL interface (d) after 35 min of biasing, which remains oriented along the [100] zone axis. e,f, Schematic illustrations show the pristine state of the interfaces with emitter at 0 min of biasing (e) and progressive phase transformations at the interfaces with the emitter after 35 min of biasing (f). The dashed lines represent the interfaces of ETL/emitter and emitter/HTL. At the pristine state, the interfaces show a small amount of the Pb-rich regions, which are negligible compared with the biased region. Scale bars, 2 nm−1.
Fig. 4 Interfacial degradation at transport layer–perovskite emitter interfaces under bias. a–d, Summed diffraction patterns (150 pixels) at the ETL/emitter and HTL/emitter interfaces under pristine conditions (tbias = 0 min) and after 35 min under bias. Red labels denote diffraction features assigned to PbX2; yellow labels indicate reflections matched to the pristine orthorhombic perovskite; green labels denote CsPb2X5. a,b, Show a good match to the orthorhombic perovskite emitter near the ETL (a) and evidence of PbX2 formation near the HTL (b). c,d, Pb-rich phases (CsPb2X5, PbX2) emerge at the ETL/emitter interface (c) and the emitter/HTL interface (d) after 35 min of biasing, which remains oriented along the [100] zone axis. e,f, Schematic illustrations show the pristine state of the interfaces with emitter at 0 min of biasing (e) and progressive phase transformations at the interfaces with the emitter after 35 min of biasing (f). The dashed lines represent the interfaces of ETL/emitter and emitter/HTL. At the pristine state, the interfaces show a small amount of the Pb-rich regions, which are negligible compared with the biased region. Scale bars, 2 nm−1.CC BY · PubMed Central
Extended Data Fig. 1 Complementary shear strain components and intrinsic interfacial degradation pathways. a–c, The Geometric Phase Analysis (GPA) in shear direction reveals pristine deformed strains at interfaces and main emitter, indicating mechanisms induced interfacial degradation. The region detected is consistent with areas present in Fig. 2. d, Schematic illustration of intrinsic degradation pathways in ETL/perovskite emitter/HTL systems, including formation of Pb-rich phases, point defects, interfacial chemical attacks, and lattice strain.
Extended Data Fig. 1 Complementary shear strain components and intrinsic interfacial degradation pathways. a–c, The Geometric Phase Analysis (GPA) in shear direction reveals pristine deformed strains at interfaces and main emitter, indicating mechanisms induced interfacial degradation. The region detected is consistent with areas present in Fig. 2. d, Schematic illustration of intrinsic degradation pathways in ETL/perovskite emitter/HTL systems, including formation of Pb-rich phases, point defects, interfacial chemical attacks, and lattice strain.CC BY · PubMed Central
Extended Data Fig. 2 Device operation induces phase changes in the perovskite emitter and Al cathode. a, Summed 1D radial diffraction profiles from the perovskite emitter before and after 35 min of operation, showing the emergence of metallic Pb peaks at (11¯1), (002), (311), (400), and (133). b, Atomic-resolution STEM-HAADF image of Pb0 nanoparticles located at grain boundaries after biasing, with corresponding FFT confirming the metallic phase. Scale bar, 1 nm. c, Summed radial diffraction profiles of pristine and biased Al cathodes, indicating the formation of AlCl3 via ion migration, with characteristic peaks at (2¯20), (003), and (33¯3).
Extended Data Fig. 2 Device operation induces phase changes in the perovskite emitter and Al cathode. a, Summed 1D radial diffraction profiles from the perovskite emitter before and after 35 min of operation, showing the emergence of metallic Pb peaks at (11¯1), (002), (311), (400), and (133). b, Atomic-resolution STEM-HAADF image of Pb0 nanoparticles located at grain boundaries after biasing, with corresponding FFT confirming the metallic phase. Scale bar, 1 nm. c, Summed radial diffraction profiles of pristine and biased Al cathodes, indicating the formation of AlCl3 via ion migration, with characteristic peaks at (2¯20), (003), and (33¯3).CC BY · PubMed Central
Extended Data Fig. 3 Interfacial structural and compositional evolution under galvanostatic bias. a–d, Summed diffraction patterns (integration over 150 pixels) from the ETL (a,c) and HTL (b,d) interfaces at successive biasing times (t = 3, 6, 15, 25 min), progressive structural changes with increasing bias duration. Colour legend, red labels denote features assigned to PbX2; yellow tick marks indicate matched planes of the pristine orthorhombic perovskite; green labels denote CsPb2X5. Patterns at each time point were acquired in fresh, non-overlapping ROIs to avoid cumulative dose history; intensities are normalised to the elastic background for comparison. e,g, STEM-EDS line profiles across the ETL (e) and HTL (g) interfaces at t = 0 min, establishing baseline elemental distributions in the pristine state. f,h, Corresponding profiles at t = 35 min, showing enhanced Pb-rich signal near both interfaces, consistent with the emergence/growth of Pb-rich secondary phases under extended operation.
Extended Data Fig. 3 Interfacial structural and compositional evolution under galvanostatic bias. a–d, Summed diffraction patterns (integration over 150 pixels) from the ETL (a,c) and HTL (b,d) interfaces at successive biasing times (t = 3, 6, 15, 25 min), progressive structural changes with increasing bias duration. Colour legend, red labels denote features assigned to PbX2; yellow tick marks indicate matched planes of the pristine orthorhombic perovskite; green labels denote CsPb2X5. Patterns at each time point were acquired in fresh, non-overlapping ROIs to avoid cumulative dose history; intensities are normalised to the elastic background for comparison. e,g, STEM-EDS line profiles across the ETL (e) and HTL (g) interfaces at t = 0 min, establishing baseline elemental distributions in the pristine state. f,h, Corresponding profiles at t = 35 min, showing enhanced Pb-rich signal near both interfaces, consistent with the emergence/growth of Pb-rich secondary phases under extended operation.CC BY · PubMed Central

Verification

Claims the paper makes

5 Valid · 2 Not valid · 8 Not checked · 70 Own results and interpretations

44 more claims

Create a free account to see all 85 claims and to verify the 3 references still open.

See all claims — free account
  • It may state two separate claims as one. 30
  • Probably not a substantial claim. 7

Create a free account to see these claims.

Citations

Cites

Show all 56 references

Cited by

See all 14

Metadata

Authors
Li, X., Gu, Q., Huang, W., Fairclough, S. M., Friend, R. H., Stranks, S. D., Liu, T., & Ducati, C.
Affiliations
University of Cambridge (GB)
Corresponding author
  • Liu, Tianjun
  • Ducati, Caterina
Venue
Nature
Published
2026-03-11
Volume
651
Issue
8106
Pages
614-620
ISSN
  • 0028-0836 (ISSN-L)
  • 1476-4687
OpenAlex
W7135016988
Cited by
14
References
54
Citations by year
2026: 14
FWCI
22.31
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • Apollo (University of Cambridge) (submittedVersion, cc-by)
Funders
  • Diamond Light Source
  • Henry Royce Institute
  • Engineering and Physical Sciences Research Council (R008779/1, EP/R008779/1)
Topics
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Perovskite Materials and Applications
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Structural Biology › Advanced Electron Microscopy Techniques and Applications
  • Physical Sciences › Materials Science › Materials Chemistry › 2D Materials and Applications
Keywords
halide perovskites, perovskite light-emitting diodes, device degradation, interfacial dynamics, in situ biasing, 4D-STEM, interfacial strain, grain fragmentation, ion transport, electrochemical reactions

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Multimodal electron microscopy of halide perovskite interfacial dynamics. https://app.cochat.ai/papers/10.1038/s41586-026-10238-8
The paper
Li, X., Gu, Q., Huang, W., Fairclough, S. M., Friend, R. H., Stranks, S. D., Liu, T., & Ducati, C. (2026). Multimodal electron microscopy of halide perovskite interfacial dynamics. Nature. https://doi.org/10.1038/s41586-026-10238-8

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.