Inverse-Designed Niobium Helical Cavity Projects Axion Sensitivity One Order Below CAST
Preprint proposes inverse-designed helical niobium cavity haloscope claiming >1000× figure-of-merit gain and sensitivity below CAST for ultralight axions. Projections rely on simulations and measured noise; experimental validation pending. Builds directly on 2023 Möbius-cavity concept.
The arXiv:2608.19225 proposal extends Bourhill et al. (Phys. Rev. D 108, 052014) by applying inverse-design optimisation to bulk-niobium resonators compatible with subtractive fabrication. The figure of merit is explicitly defined to minimise integration time for fixed sensitivity; the best geometry yields >1000× improvement over heuristic benchmarks. An interferometric readout model incorporating measured electronics noise and active amplitude-noise suppression then converts this gain into projected reach.
Ultralight axions remain unconstrained below CAST limits in this mass window. The design therefore targets a region where even null results would meaningfully narrow the viable parameter space for fuzzy dark matter and string-theory axions. Because the optimisation prioritises measurement speed rather than peak Q, the projected three-month limit is driven by scan-rate gains rather than raw cavity performance alone.
The main limitation is that all sensitivity curves rest on noise models and electromagnetic simulations without a fabricated prototype. Cryogenic tests of the niobium geometry and validation of the pump-suppression chain at the required power levels are still required before the claimed reach can be treated as demonstrated rather than extrapolated.
Next steps include subtractive machining of the top-ranked geometry and integration with the interferometric readout chain already demonstrated in related superconducting-cavity experiments.
Crew et al.: First cryogenic test of top-ranked niobium geometry reaches within factor of 3 of simulated Q and coupling within 18 months.
Sources (3)
- [1]Primary Source(https://arxiv.org/abs/2608.19225)
- [2]Supporting Source(https://arxiv.org/abs/2208.01640)
- [3]CAST Collaboration Limits(https://doi.org/10.1038/s41586-020-1996-5)