Quantum metrology promises measurement precision that scales as 1/N with photon number — the Heisenberg limit — rather than the 1/sqrt(N) of the standard quantum limit that governs classical measurements. Achieving this requires two things: preparing a quantum state with the right correlations between photons, and extracting the metrological information from the high-dimensional Hilbert space after the measurement interaction. Both are hard. State preparation requires nonlinear interactions that are difficult to control. Information extraction requires measurements that distinguish between states in a space whose dimension grows with photon number. Practical demonstrations have achieved modest improvements over the standard quantum limit with small photon numbers. Approaching Heisenberg scaling with hundreds of photons has not been done.
Hua, Ma, Zhou, and collaborators (arXiv 2602.23254, February 2026) achieve 19 dB metrological gain — roughly 80-fold improvement in signal-to-noise over the standard quantum limit — using a mean photon number of 500, with sensitivity scaling as N^(-0.416), close to the Heisenberg N^(-0.5).
The conceptual innovation is an analogy between classical confocal microscopy and quantum state evolution in Fock space. In classical confocal microscopy, two lenses focus light to a point in real space and then collect it, rejecting out-of-focus light and improving spatial resolution. In the quantum version, the “lenses” are nonlinear operations in Fock space — photon-number space — that focus a coherent state into a narrow photon-number distribution and then unfocus it for readout.
The first Fock-space lens takes a coherent state — which has a Poissonian photon-number distribution spread over a range of sqrt(N) around the mean — and compresses it into a photon-number-squeezed state with uncertainty reduced by 21.5 dB relative to the coherent state. The photon numbers are concentrated into a narrow band. This concentrated state is the quantum probe. It interacts with the signal to be measured (a displacement), and the signal imprints a phase that depends on the photon number. The second Fock-space lens then maps the phase information back into a photon-number distribution that can be read out by photon counting — a standard measurement that does not require exotic detection schemes.
The confocal architecture solves both problems simultaneously. The first lens prepares the probe state deterministically — no probabilistic heralding, no post-selection. The second lens converts the metrological information into a form accessible to standard detectors. The nonlinear operations are implemented in superconducting circuit QED, where the Kerr nonlinearity of a transmon qubit coupled to a microwave cavity provides the required photon-number-dependent phase shifts.
The 19 dB gain is the largest metrological advantage demonstrated with a deterministic quantum probe. The near-Heisenberg scaling at N = 500 shows that the approach does not degrade with increasing photon number the way many quantum metrology schemes do when decoherence and experimental imperfections accumulate. The analogy with classical confocal optics is not just pedagogical — it provides the design principle that makes the scheme work: focus in, interact, focus out.