Attosecond XFEL and Beyond

Jiawei Yan

My current main focus is the generation and application of ultra-high-power hard X-ray attosecond pulses. More broadly, I work on advanced X-ray free-electron laser science, especially when source physics can be pushed toward stronger temporal compression, higher peak power, richer pulse structure, and more useful experimental capability.

My research spans attosecond XFELs, spatiotemporal shaping of X-ray pulses, fully coherent high-repetition-rate FELs, and AI for Science. At the center is a broader question: how can we move toward full control over X-ray pulses?

About

Advanced XFEL research centered on hard X-ray attosecond science

Current work in the MXL group at DESY centers on extending ultra-high-power hard X-ray attosecond pulse generation toward new source concepts, diagnostics, and applications.

I am based in Hamburg and work as a Scientist in the DESY MXL group on advanced XFEL science across theory, numerical design, and experimental implementation. My current research program focuses on advancing hard X-ray attosecond pulse generation toward higher peak power, stronger controllability, and broader experimental impact, especially where pulse properties can be deliberately matched to demanding ultrafast measurements.

Around this main line, I work on several connected problems: shaping X-ray pulses in both time and space, developing routes toward fully coherent high-repetition-rate FEL operation, and using AI and optimization methods to make accelerator and light-source physics more efficient and more programmable.

I do not see these as separate topics. They are parts of one broader objective: full control over X-ray pulses, from pulse duration and wavefront structure to coherence, stability, diagnostics, and machine intelligence.

Scientist, MXL Group at DESY

Current emphasis

Attosecond XFEL
X-ray pulse shaping
Fully coherent FEL
AI for Science

Editorial Appointments

Editorial Board

Scientific Reports

Early Career Editorial Board

Advanced Photonics

PhotoniX Life

Google Scholar ResearchGate LinkedIn jwyan.com

Research

Four research themes

Attosecond XFELs, X-ray pulse shaping, coherent high-repetition-rate FELs, and AI for Science.

Attosecond XFEL

Hard X-ray attosecond pulses are a uniquely powerful route to resolving electronic motion and nonequilibrium dynamics on their natural timescales. The challenge is not merely to reach the attosecond regime, but to do so with sufficiently high peak power, higher photon energy, and pulse properties that are actually useful for experiments.

This is a primary focus of the current research program. The work emphasizes beam shaping, electron-beam compression, automatic optimization, diagnostics, and the transition from proof-of-principle generation to robust scientific use.

Spatiotemporal shaping of X-ray pulses

Once pulse duration becomes controllable, the next question is whether one can also shape waveform, phase, and transverse mode content at the source. This matters because many future X-ray experiments will need not just shorter pulses, but pulses with designed spatiotemporal structure.

This direction studies methods for directly shaping X-ray pulses in time and space, including orbital-angular-momentum control and other routes to wavefront-engineered or waveform-engineered X-ray emission.

Fully coherent high-repetition-rate FEL

Fully coherent FEL operation at high repetition rate is a major goal for next-generation light sources, but standard seeded FEL schemes run into practical limits, especially in seed-laser power and scalability. More generally, the question is how to retain seeded-FEL coherence while removing those bottlenecks.

A representative line of work is the self-modulation seeded FEL mechanism, developed and experimentally validated in collaboration with colleagues. More broadly, this direction examines how self-modulation HGHG and related mechanisms can support compact, high-average-power EUV FELs as well as more general routes toward coherent high-repetition-rate FEL operation.

AI for Science

Accelerator and FEL systems are already too high-dimensional and too nonlinear to rely only on manual tuning or brute-force scans. If we want light sources to become more adaptive and more precisely controlled, intelligent optimization must become part of the physics workflow.

This work uses Bayesian optimization, machine learning, and physics-informed algorithms for machine studies, FEL design, and operational optimization, especially where these methods can improve both physical insight and practical controllability.

Recognition