This paper describes how a thin, suspended silicon‑nitride membrane can convert light intensity into a large change of optical phase. The effect comes from radiation pressure: light pushes on a mechanically soft reflector, the reflector moves, and the extra path length shifts the phase of the reflected beam. Using measured device parameters, the authors predict a phase responsivity of 263 radians per watt and a device‑equivalent nonlinear index n2,eff = 5.4×10−7 m2 W−1 for a 50‑nm‑thick membrane “trampoline” suspended on serpentine springs.

The authors introduce a simple framework to compare mechanical phase elements. They separate mechanical properties (how easily the device moves and how that response holds up over large displacements) from optical properties (how much of the illuminated area actually contributes a common phase shift). From those pieces they define strength‑range‑aperture metrics: small‑signal compliance, retained‑response range, and phase‑coupled area. They then apply a common direct‑reflection protocol to a set of reported resonators so the systems can be compared on the same basis.

At a high level the device works like this: radiation pressure produces an optical force proportional to incident power. The force displaces the membrane by an amount z and the reflected phase shift is approximately 4πz/λ, where λ is the light wavelength. Two optical overlaps matter. One tells how well the light drives the mechanical motion. The other tells how much of that motion appears as a common phase shift in the collected reflection. Squaring this overlap gives an effective phase‑coupled area, which scales how efficiently radiation pressure converts light power into phase.

Why this matters: many common optical nonlinearities are weak, narrowband, or change the light frequency. A moveable reflector in the quasi‑static limit preserves the light’s carrier frequency, can be broadband in wavelength, and — according to the measured device numbers used here — can give a very large phase shift per unit intensity. The trampoline reported here stands out because it combines very high mechanical compliance, a practicable optical aperture for free‑space beams, and a retained response that remains 99.85% of the small‑signal responsivity through a full 2π phase swing. Under the paper’s reconstruction protocol the trampoline’s small‑signal slope of phase versus intensity is roughly 2,000 times larger than the next best system that met the same criteria.