4D-STEM Electron Diffraction

electron_diffraction Electron Microscopy Coherent Diffraction Wave Optics
View Benchmarks (1)

4D-STEM acquires a full 2D convergent-beam electron diffraction (CBED) pattern at each probe position during a 2D STEM scan, yielding a 4D dataset (2 real-space + 2 reciprocal-space dimensions). This enables simultaneous mapping of strain, orientation, electric fields, and thickness with nanometer spatial resolution. Phase retrieval from the 4D dataset (electron ptychography) can achieve sub-angstrom resolution. High data rates (>1 GB/s) from fast pixelated detectors create computational challenges.

Forward Model

Cbed Forward

Noise Model

Poisson

Default Solver

ptychography epie

Sensor

PIXELATED_DETECTOR

Forward-Model Signal Chain

Each primitive represents a physical operation in the measurement process. Arrows show signal flow left to right.

P e⁻ Convergent Beam F diffraction Diffraction Pattern D g, η₁ Pixelated Detector
Spec Notation

P(e⁻) → F(diffraction) → D(g, η₁)

Benchmark Variants & Leaderboards

4D-STEM

4D-STEM Electron Diffraction

Full Benchmark Page →
Spec Notation

P(e⁻) → F(diffraction) → D(g, η₁)

Standard Leaderboard (Top 10)

# Method Score PSNR (dB) SSIM Trust Source
🥇 DiffED 0.878 39.1 0.953 ✓ Certified Gao et al. 2024
🥈 PhysED 0.849 37.7 0.941 ✓ Certified Chen et al. 2024
🥉 SwinED 0.822 36.4 0.930 ✓ Certified Wang et al. 2023
4 TransED 0.786 34.8 0.912 ✓ Certified Li et al. 2022
5 PhaseGAN-ED 0.725 32.3 0.873 ✓ Certified Zimmermann et al. 2021
6 DnCNN-ED 0.658 29.5 0.833 ✓ Certified Cherukara et al. 2018
7 MicroED 0.586 26.7 0.781 ✓ Certified Shi et al. 2013
8 PEDT 0.517 23.9 0.738 ✓ Certified Kolb et al. 2007
9 Direct-Methods 0.450 21.2 0.694 ✓ Certified Hauptman & Karle 1985
Mismatch Parameters (3) click to expand
Name Symbol Description Nominal Perturbed
camera_length ΔL Camera length error (%) 0 2.0
center_offset Δc Diffraction center offset (pixels) 0 1.0
elliptical_distortion ε Elliptical distortion 0 0.005

Reconstruction Triad Diagnostics

The three diagnostic gates (G1, G2, G3) characterize how reconstruction quality degrades under different error sources. Each bar shows the relative attribution.

G1 — Forward Model Accuracy How well does the mathematical model match reality?

Model: cbed forward — Mismatch modes: scan distortion, detector saturation, dynamical scattering, specimen tilt

G2 — Noise Characterization Is the noise model correctly specified?

Noise: poisson — Typical SNR: 5.0–25.0 dB

G3 — Calibration Quality Are instrument parameters accurately measured?

Requires: camera length, convergence angle, beam center, rotation angle, detector gain

Modality Deep Dive

Principle

4D-STEM electron diffraction scans a convergent electron beam across the specimen and records a full 2-D diffraction pattern (convergent beam electron diffraction, CBED) at each scan position. The resulting 4-D dataset (2-D scan × 2-D diffraction) enables mapping of crystal structure, orientation, strain, electric fields, and charge density with nanometer spatial resolution.

How to Build the System

Use a STEM equipped with a fast pixelated detector (Medipix3, EMPAD, or Dectris ARINA) capable of recording diffraction patterns at >1000 fps. Set a small convergence semi-angle (1-5 mrad) for nanobeam diffraction or large (20-30 mrad) for CBED. The scan step should be comparable to the probe size. Data volumes are large (tens of GB per scan), requiring efficient data pipeline and storage.

Common Reconstruction Algorithms

  • Virtual detector imaging (synthesized BF, DF, iDPC from 4D data)
  • Center-of-mass (COM) analysis for electric field mapping
  • Ptychographic reconstruction from 4D-STEM data
  • Orientation mapping (template matching against simulated patterns)
  • Strain mapping via disk position analysis

Common Mistakes

  • Detector dynamic range insufficient for simultaneous central beam and weak diffraction
  • Scan step too large relative to probe size, under-sampling the specimen
  • Not accounting for specimen thickness variation in diffraction pattern interpretation
  • Excessive electron dose for beam-sensitive materials (organics, 2D materials)
  • Misindexing diffraction patterns due to double diffraction or overlapping grains

How to Avoid Mistakes

  • Use counting-mode detectors (Medipix) with high dynamic range or electron counting
  • Match scan step to probe size for complete spatial sampling
  • Simulate diffraction patterns at the measured thickness for accurate interpretation
  • Use low-dose 4D-STEM protocols with fast detectors to minimize beam damage
  • Carefully index patterns considering multiple scattering; compare with simulations

Forward-Model Mismatch Cases

  • The widefield fallback produces a real-space blurred image, but electron diffraction records the far-field diffraction pattern (reciprocal space) — Bragg spots encode crystal structure, lattice spacings, and symmetry, which bear no resemblance to a blurred image
  • The diffraction pattern intensity I(k) = |F{V(r) * P(r)}|^2 encodes the Fourier transform of the projected crystal potential — the widefield real-space blur cannot access reciprocal-space crystallographic information

How to Correct the Mismatch

  • Use the electron diffraction operator that models kinematic or dynamical scattering from the crystal lattice, producing far-field diffraction patterns with Bragg peaks at reciprocal lattice positions
  • Index diffraction patterns to determine crystal structure and orientation; use dynamical simulation (Bloch wave or multislice) for accurate intensity matching and structure refinement

Experimental Setup

Instrument

Thermo Fisher Titan with Medipix3 / JEOL ARM with EMPAD

Accelerating Voltage Kv

200

Convergence Angle Mrad

1.5

Step Size Nm

1.0

Detector

Medipix3 / Merlin (256x256 px)

Exposure Ms

1

Camera Length Mm

580

Reconstruction

ptychographic phase retrieval / WDD

Signal Chain Diagram

Experimental setup diagram for 4D-STEM Electron Diffraction

Key References

  • Ophus, 'Four-dimensional scanning transmission electron microscopy', Microscopy & Microanalysis 25, 563 (2019)
  • Jiang et al., 'Electron ptychography of 2D materials to deep sub-angstrom resolution', Nature 559, 343 (2018)

Canonical Datasets

  • 4D-STEM benchmark datasets (Ophus group, NCEM)

Benchmark Pages