In the quiet corridors of ETH Zurich, a nanoparticle floats in a laser trap, its tiny motion monitored not by peering through a detector and reading out a signal, but by letting light itself carry the control signal back to the particle. This experiment is a striking demonstration of measurement-free coherent feedback, a concept at the edge of quantum control where information travels in a loop without ever being recorded.
The researchers from ETH Zurich’s Nanophotonic Systems Laboratory and Quantum Center—with Nadine Meyer and Romain Quidant among the senior authors—show that you can cool the center of mass motion of a levitated nanoparticle to a few hundred phonons by tuning the phase and delay of an all-optical feedback loop. That is, the motion influences the light, which travels around a fibre, re-impacts the particle, and damps the motion without a photodetector at any point in the loop.
Why does this matter? In quantum optomechanics, measurement-based feedback has been powerful but comes with backaction from the act of measurement and electronic noise that creeps back into the motion. Coherent feedback sidesteps those hurdles by preserving quantum correlations and keeping the control signal in the optical domain. It’s a different philosophy: let the light do the listening and the talking at once, and keep the conversation in the same quantum language.
What is coherent feedback and why isn’t it just fancy math
Coherent feedback is not about measuring and computing a response. It uses a light field that interacts with the particle, and then is redirected back to create a new trap landscape. The phase and delay of that loop matter because the returning light can either shift the equilibrium position or tug on the particle in a way that mimics a velocity-dependent damping. In the researchers’ setup, a carefully chosen delay aligns the loop so that the feedback effectively damps the motion, cooling the particle more than would be possible with static trapping alone.
The trick is that the feedback signal is not parsed by a detector and a processor; it travels as a coherent optical field. The relative phase between the trapping light and the feedback light, plus the loop delay, determines whether the feedback cools or heats the motion. In their design, a delay corresponding to about a quarter of the particle’s oscillation period yields the desired velocity-proportional damping. This is not magic; it’s a precise choreography of light and motion in the quantum regime.
Because everything stays in the optical domain, the scheme preserves quantum correlations between light and motion and avoids injecting electronic noise into the loop. That advantage matters when you’re trying to push systems toward the quantum ground state, where even tiny imperfections can shake the delicate balance needed for quiet motion.
The experiment a long optical leash that cools a nanoparticle
The experimental stage is a single-beam optical tweezer at 1064 nm, focused by a high-NA lens to trap a dielectric nanoparticle about 97 nanometers in radius. The particle’s motion along the optical axis z sits at a resonance in the tens to hundreds of kilohertz. A forward-scattered light signal detects motion in x, y, z via balanced detection, while the backward-scattered light is routed into a long optical feedback path. This is where the loop begins to feel like a relay race: the particle’s past motion imprints a phase onto the light, which then re-enters the trap to influence its present state.
The feedback beam travels through a single-mode fiber of about 1.3 kilometers, introducing a delay of roughly 6.34 microseconds. At the far end, a mirror and circulator steer the beam back toward the particle, where it re-interferes with the trap. The total loop efficiency η controls how strongly the past position z(t−τ) pulls the trap toward a shifted equilibrium zeq(t). A piezo-mounted mirror tunes the accumulated phase ϕ0, letting experimenters dial in the right interference pattern. The interplay of phase and delay is what makes the loop act like a gentle, velocity-sensitive brake or, if misaligned, like a kick in the wrong direction.
In their cooling experiments, the phase of the feedback light is adjusted so that the loop effectively measures velocity: when the delay corresponds to a quarter of the oscillation period and the phase aligns, the particle experiences a damping force that cools its motion. If the phase flips by half a cycle, the same loop sets up anti-damping, heating the motion instead. The researchers verify cooling and heating by comparing the displacement noise spectrum with and without coherent feedback. It is a striking demonstration that light can control motion without ever recording a measurement in the loop.
With careful optimization, the team reports a minimum effective temperature of about 705 microkelvin for the center-of-mass motion, corresponding to roughly 344 phonons for the given frequency. That is a meaningful step toward the quantum ground state, yet the result makes explicit the bottleneck: phase noise in the delay line becomes the dominant source of heating as cooling strengthens. The findings align with a simple, transparent theory that balances the random kicks from gas, photons, and phase fluctuations against the designed damping. In the language of the paper, the noise budget is as important as the cooling channel itself.
Notably, the experiments also reveal the subtle effect known as noise squashing: in-loop measurements can appear cooler than the out-of-loop reality because the phase noise in the loop injects extra fluctuations that propagate back into the particle’s motion. The out-of-loop readout, which is not used to close the loop, confirms a more modest reduction in motion yet remains a faithful witness to the cooling power of the coherent feedback. This dual perspective is a reminder that reality in quantum experiments often hides in the details of what you choose to measure and where you place your sensors.
A glimpse of a future shaped by optical control
The study opens a new chapter in quantum control of levitated objects. It shows that non-mechanical information loops—light that talks back to the particle without being read by a detector—can sculpt motion with remarkable precision. This is not just a curiosity; the same toolkit could be used to engineer nonreciprocal dynamics, dissipative interactions, or entanglement between remote mechanical systems, all mediated by light. The principle resonates with the broader dream of building quantum networks where information moves through photonic channels without collapsing into classical records.
Yet the path to ground-state cooling with coherent feedback is not straightforward. The delay line’s length is a double-edged sword: longer delays can enhance cooling in some regimes, but they also bring bigger phase noise. The authors sketch a practical roadmap: operate at higher frequencies, explore librational modes or standing-wave traps, and push the optical components toward lower phase noise and higher stability. On-chip or integrated photonics could compress the loop and reduce loss, bringing the dream closer to a scalable platform. The vision is not just for one particle but for what a family of levitated objects could become when coherently spoken to by light.
Beyond cooling, the coherent-feedback approach hints at a broader philosophy: in quantum technology, it may be advantageous to keep information flowing in a single, coherent language—light—rather than translating measurements into classical bits and back. If researchers can tame noise enough, this could enable portable quantum sensors that exploit the correlations between light and motion, or even networks of levitated particles that exchange quantum information through light. The implications reach from precision metrology to the architecture of future quantum devices, where control is achieved not by measuring and reacting, but by letting light and motion dance in step together.
As the ETH Zurich team puts it, levitated nanoparticles are a playground for both fundamental physics and practical sensing. They are exquisitely isolated from the environment, yet surprisingly tunable, with control knobs for the trap strength, the feedback phase, and the geometry of motion. Coherent feedback adds a new knob to that toolbox: a purely optical, measurement-free dial that can be adjusted to chase the quantum ground state while preserving coherence along the way. The authors also remind readers that phase noise is not just a nuisance but a fundamental constraint to be engineered around, and they outline paths that could eventually push Teff closer to the truly quantum limit.
In short, this work from ETH Zurich demonstrates a bold, human approach to quantum control: let light carry information and influence, not just report back what it has observed. If the phase noise hurdle can be overcome, and if the method scales to more complex motions and multiple modes, coherent feedback could become a mainstay in the quantum engineer’s toolkit, shaping sensors, interfaces, and perhaps even tiny quantum machines powered by light. The study is a reminder that progress in quantum science often looks like patient tuning of a musical instrument—only here the instrument is a nanoparticle and the bow is a fiber loop.
