As one of the most advanced light source technologies, X-ray free-electron lasers (XFELs) generate radiation spanning from terahertz to X-ray wavelengths.1 These facilities deliver unprecedented spatiotemporal resolution at subnanometer spatial scales and femtosecond-to-attosecond temporal regimes, revolutionizing ultrafast studies in physics, chemistry, and biology. The unique capability of free-electron lasers (FELs) to produce intense, wavelength-tunable coherent pulses further enables investigations of matter under extreme conditions. Concurrently, advancements in FEL theory and accelerator physics drive interdisciplinary innovations in laser physics and particle acceleration technologies. The evolution of FELs signifies a quantum leap in humanity’s ability to manipulate light–matter interactions, expanding fundamental scientific frontiers while bridging microscopic research with macroscopic applications.

The conceptual framework of FELs, first proposed by Madey et al.2 in 1971, relies on relativistic electron beams interacting with periodic magnetic fields in undulators to achieve exponential amplification through coherent stimulated emission.3^,4 Modern FEL facilities have transformed research paradigms across multiple disciplines. Central to FEL performance is electron beam quality, quantified through parameters, including pulse duration, bunch compression factor, charge quantity, divergence angle, energy spread, and transverse emittance. Temporal characteristics—particularly bunch length, current profile, and slice emittance—directly determine XFEL output properties such as peak power and pulse duration. However, transient beam diagnostics face dual challenges: (i) inherent time–energy correlations stemming from collective beam dynamics (e.g., space-charge effects and coherent synchrotron radiation) during acceleration and compression phases and (ii) the stringent requirements for attosecond-level diagnostics—although direct deflection methods remain theoretically viable,5 their practical implementation faces fundamental limitations in synchronization precision and field intensity control. Conventional radiofrequency transverse deflection structures (TDSs) provide femtosecond-level resolution through time-to-momentum correlation mapping but suffer from fixed deflection axes and insufficient resolution for attosecond science.6^,7

A collaborative effort by European Organization for Nuclear Research (CERN), Deutsches Elektronen-Synchrotron (DESY), and Paul Scherrer Institute (PSI) has yielded the PolariX TDS—a modular X-band (11.4 GHz) deflector system enabling polarization-tunable attosecond metrology.8 Implemented at SwissFEL’s Athos soft X-ray beamline, this system introduces critical advancements: dynamic polarization control via phase shifters (PS1/PS2) allows arbitrary deflection plane selection (a range of 0 to 180 deg; 5 deg resolution), achieving <10% amplitude variation across orientations. Experimental validation demonstrated record-breaking performance: 0.6 fs root mean square (RMS) resolution at 3.2GeV/200pC operation, reaching 300 as resolution under 10 pC low-charge mode. The system successfully reconstructed FEL power distributions with 290 as single-shot resolution and 0.57 fs average RMS duration using energy-loss difference reconstruction.

This breakthrough demonstrates three transformative dimensions. The integration of polarization agility with X-band technology achieves unprecedented multidimensional phase-space characterization (3D charge distribution) while advancing temporal resolution to attosecond regimes. Complementing this technical innovation, the SwissFEL experimental framework establishes a gold-standard methodology for free-electron laser diagnostics through synergistic hardware optimization (X-band cavities) and algorithmic reconstruction techniques. Most fundamentally, the PolariX TDS system transcends conventional diagnostics by bridging attosecond electron dynamics with photon pulse characterization, thereby enabling pump–probe experiments at intrinsic atomic timescales—a capability poised to redefine ultrafast science paradigms.

Looking forward, this technology holds multifaceted potential: enabling subfemtosecond XFEL pulses through precision beam control, revealing hidden beam dynamics via polarization-resolved diagnostics, and facilitating global standardization through modular design. As artificial intelligence permeates accelerator physics, the integration of PolariX TDS with machine-learning-driven optimization could transform XFELs from observation tools to programmable matter controllers, ushering in new eras of attosecond chemistry and extreme-condition material science.