The storage result
A laboratory test stored about 300 millijoules of laser energy inside a burst-mode optical cavity for 20 minutes. Stored-power stability was about 10 percent on the study's full-width-at-half-maximum measure, with 7.6 percent root-mean-square jitter. Both figures describe how much the stored power fluctuated during the run.
The work evaluates a burst-mode optical enhancement cavity for inverse Compton scattering applications and examines how cavity filling and energy storage vary with carrier-envelope offset, or CEO, parameters. The design targeted 1,000 electron bunches per burst at a repetition frequency around 1 gigahertz. The laboratory setup used 880 megahertz, commercial high-reflectivity mirrors, a symmetric bow-tie geometry and a beam about 300 micrometres wide on planar mirrors.
How the cavity was driven
Both laser streams traveled through a common injection path: a continuously injected feedback beam known as Lockline and a high-energy burst beam called TANGOR. The system used Pound-Drever-Hall locking, a feedback method for keeping the laser and cavity aligned. In the low-finesse test, each injection lasted 5 microseconds and contained 4,400 pulses at 880 megahertz, with bursts repeated at 1 kilohertz. The individual pulses were about 3 picoseconds long at FWHM.
The first test cavity had a finesse of about 2,000, with an uncertainty of 200. Finesse indicates how selectively a cavity builds up light at its resonant frequency. Measurements of the cavity mirrors corresponded to an estimated power enhancement of about 240, with an uncertainty of 50.
The injected energy was raised to 2 millijoules per burst, or about 450 nanjoules per pulse. At the 1-kilohertz burst rate, that was about 2 watts of average input power. The storage run accumulated about 300 millijoules, while the effective-gain estimate carried an uncertainty of roughly 15 percent.
The run also exposed limits in the control system. It ended when feedback became unstable, and noise from TANGOR to Lockline had not been fully addressed. The researchers reported no significant thermal effects apart from slight environmental drift.
Why higher finesse changes the problem
A coupling mirror was damaged when the input reached about 10 millijoules across a full burst, equivalent to roughly 2 microjoules per pulse. The replacement cavity measured a finesse of 7,300, with an uncertainty of 130. It was more sensitive to perturbations and needed tighter feedback.
The next tests focused on the carrier-envelope offset, or CEO, which describes how the optical phase shifts from one pulse to the next. For each CEO-frequency setting, the team swept the laser repetition period across resonance, measured reflected power at resonance and far from resonance, and fitted the normalized ratio. The model left only the spatial coupling coefficient and a global CEO-frequency offset as free parameters.
The fit reproduced nearly all of the measured points. It estimated spatial coupling at 0.51, with an uncertainty of 0.03, and an optimal CEO frequency shift of 130 megahertz, with an uncertainty of 5 megahertz.
Numerical simulations showed stronger CEO phase sensitivity in the high-finesse configuration. The simulated profile for the 7,300-finesse cavity qualitatively matched the measurement, while the low-finesse simulation did not reproduce the observed features.
The simulations pointed to CEO-frequency mismatch as the more consequential sensitivity. Compared with an equivalent CEO phase shift, a mismatch produced a much larger change in maximum stored energy, with differences spanning several orders of magnitude. The shape of the stored-energy profile was also highly sensitive to the mismatch.
The proposed operating window
For a cavity with finesse of several thousand, the authors specify keeping the CEO frequency within a few tens of megahertz of 880 megahertz and stabilizing CEO mismatch within a few tens of kilohertz to keep losses below a few percent. Those figures are proposed operating targets drawn from the study's analysis, not a report of optimized high-finesse storage.
Taken together, the measurements and simulations describe a system that can store burst-mode energy while becoming more sensitive to control conditions as finesse rises. The analysis therefore puts particular emphasis on CEO settings and on the mismatch between the locking and burst chains.
This document is an arXiv preprint, version 1, dated 29 September 2026. The work was partially financed by Amplitude Laser. The authors declare no conflicts of interest. Underlying data are not publicly available but may be obtained from the authors on reasonable request.
Paper data and sources
Original title: Optical enhancement cavities operated at GHz repetition rate in burst mode regime
Authors: Kevin Dupraz, Frédéric Blanc, Ronic Chiche et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-09-29
DOI: Not available
Original paper · Full text