Particle accelerators push charged particles to high energies using electromagnetic fields. Conventional accelerators use metal cavities and sustain accelerating gradients of roughly 50 MV/m — limited by electrical breakdown of the cavity walls. Plasma wakefield accelerators replace the metal cavity with a plasma wave, achieving gradients thousands of times higher. But plasma accelerators have their own problem: creating and controlling the plasma channel over the meter-scale distances needed for useful energy gain.
Maier, Bohlen, and collaborators (arXiv 2602.22841, February 2026) demonstrate a different approach. Instead of pre-forming a plasma channel or using a discharge to ionize the gas, they send a femtosecond laser pulse through a low-pressure gas. The laser self-guides through nonlinear interaction with the gas, creating a plasma filament — a thin, elongated column of ionized gas shaped by the laser's own propagation dynamics. Then they inject an electron beam into the filament and measure acceleration.
The filament sustains an accelerating gradient exceeding 250 MV/m — five times higher than conventional cavities and generated without any external confinement structure. The plasma channel creates itself from the laser interaction. No discharge, no capillary, no mechanical waveguide. The laser provides both the plasma and the channel geometry in a single shot.
The key advantage is repetition rate. Discharge-plasma channels rely on stochastic breakdown processes and recover slowly. Laser-produced filaments can be generated at kilohertz rates — thousands of plasma channels per second, each one fresh and reproducible. The stochastic step is eliminated.
The accelerating gradient is modest by plasma wakefield standards — some experiments have achieved tens of GV/m. But the filament method trades peak gradient for controllability, scalability, and repetition rate. The fastest path to a practical plasma accelerator may not be the steepest gradient but the most reliable channel.