Vision of HyperScaleEM
Electron microscopy is essential to understanding structure-property-function relationships in modern materials engineering, condensed matter physics, chemistry, and structural biology. Yet, due to complicated scattering physics, today’s electron microscopes can only image tiny volumes with 3D atomic resolution. Within this project, we will turn the tables by utilizing and inverting the scattering physics to image scale-bridging volumes with atomic detail and chemical superresolution. Combining compressive data-acquisition protocols, state-of-the-art electron optics and detectors, and co-designed computational imaging algorithms will make this possible.
What we’re building
HyperScaleEM is building the microscopy workflow that doesn’t exist yet: an automated pipeline that can image hundreds of nanometres across with atomic-level resolution and chemical composition — in 3D, at the speed of a modern electron microscope, without human intervention at every step.
The core challenge isn’t the hardware — it’s that the computational methods required to reconstruct this data are too slow, too brittle, and too manual. We solve this by designing physics-informed AI that learns the microscope’s forward model and inverts it at scale, combined with closed-loop acquisition that decides where to measure next based on what it’s already seen.
What this enables: A biologist studying a virus particle, a battery engineer characterising a dendrite, or a chemist imaging a catalyst — all getting the same quality of 3D atomic structure data that today only a handful of world-leading centres can produce, in hours not weeks.
Project roadmap
Achieved
3D atomic resolution in volumes >10 nm
2023-2024
Multi-slice electron ptychographic tomography demonstrated three-dimensional phase-contrast microscopy beyond conventional depth-of-focus limits. See also arXiv:2512.19460.
In progress
Scale-bridging imaging pipeline
2025-2027
Pipeline work is connecting acquisition, reconstruction, and analysis for larger 4D-STEM volumes. Progress update.
Ramping up
Automated 3D chemical mapping across volumes
2026-2028
Integrating hyperspectral signals with scalable 3D reconstruction so structure and chemistry can be recovered together.
Ramping up
In-situ dynamics capture with autonomous acquisition
2027-2029
Closed-loop acquisition will target dynamic processes and decide where to measure next based on the evolving reconstruction.
Ramping up
Cross-scale integration: atomic-to-nanometre bridging
2028-2030
The final integration step links atomic detail with nanometre-scale context across large reconstructed volumes.
Team & open positions
HyperScaleEM is led by Prof. Philipp Pelz (PI) and a multidisciplinary team of postdoctoral researchers, PhD students, and visiting collaborators. The project is integrated with the broader ECLIPSE Lab and the CENEM Centre for Nanoanalysis at FAU Erlangen-Nürnberg.
Open positions funded through HyperScaleEM are advertised on the ECLIPSE Lab Opportunities page. We actively encourage applications from candidates with backgrounds in physics, materials science, computer science, or applied mathematics.
Interested in collaboration? Contact the lab →