Cavity Array Microscope: Parallel Quantum Imaging

• Cavity array microscopes are imaging platforms comprising 2D lattices of independent optical microcavities enabling diffraction‐limited, parallel quantum measurements.
• They exploit diverse architectures—including silicon, open-access, and MLA-based designs—to achieve high spatial resolution, strong emitter coupling, and uniform performance.
• The system offers rapid multiplexed readout with sub-nanometer sensitivity, making it essential for quantum optics, nanoscale biosensing, and parallel quantum simulation.

A cavity array microscope is a parallelized imaging and sensing platform constructed from a two-dimensional lattice of independently resonant optical microcavities, each acting as a localized high-cooperativity light–matter interface. By engineering such cavity arrays, the system enables simultaneous, diffraction-limited probing of multiple microscale sites—crucially allowing both strong coupling to individual quantum emitters and highly sensitive detection at each pixel. Cavity array microscopes have emerged as essential tools at the intersection of quantum optics, quantum information, and nanophotonics, where high spatial resolution, field enhancement, and parallel readout are demanded (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026, Derntl et al., 2013).

1. Cavity Array Architectures

Cavity array microscopes employ diverse optical architectures, typically distinguished as on-chip microfabricated arrays, open-access microcavity modules, and free-space lens-based multi-mode resonators.

• Microfabricated silicon platforms employ lithographically defined, concave silicon micro-mirrors coupled to fibre arrays, forming Fabry–Pérot resonators with radii of curvature $R_C \sim 50$–60 μm, cavity lengths $L_C \sim 43$ μm, and mode waists $w_0 \sim 2$ μm (at 780 nm) (Derntl et al., 2013).
• Open-access microcavity arrays use CO$_2$-laser-ablated, fused silica pyramidal substrates with dielectric coatings, achieving mode waists $w_0 \sim 10\,\mu$m, highly uniform per-site fineness ($\mathcal{F} \sim 10^5$), and robust mechanical isolation (Doherty et al., 2022).
• Free-space multi-site arrays integrate microlens arrays (MLAs), 4f telescope systems, and spatial light modulators (SLMs) to define hundreds of sub-wavelength-scale TEM$_{00}$ cavity modes within a single macroscopic resonator. For example, with $w_0 \sim 1\,\mu$m, pitch $d_0= 10\,\mu$m, and total numbers exceeding $600$ modes (Soper et al., 6 Feb 2026, Shaw et al., 12 Jun 2025).

These architectures are summarized below:

Architecture	Mode Waist ($L_C \sim 43$0)	Pitch	Max. Sites	Finesse ($L_C \sim 43$1)
Silicon microcavity	2.16 μm	250 μm	48	$L_C \sim 43$2 (proj.)
Open-access microcavity	10.6 μm	100 μm	4–16	$L_C \sim 43$3–$L_C \sim 43$4
Free-space MLA	1.15 μm	10 μm	600+	114–145

These geometries determine the fundamental optical parameters, spatial resolution, and compatibility with quantum emitter arrays (e.g., neutral atoms, quantum dots).

2. Optical and Cavity QED Parameters

Core performance metrics for each cavity pixel include mode waist $L_C \sim 43$5, mode volume $L_C \sim 43$6, free spectral range (FSR), cavity linewidth $L_C \sim 43$7, finesse $L_C \sim 43$8, and single-atom cooperativity $L_C \sim 43$9.

Cooperativity is defined as

$w_0 \sim 2$0

where $w_0 \sim 2$1 is the vacuum Rabi frequency, $w_0 \sim 2$2 the cavity energy decay rate, and $w_0 \sim 2$3 the atomic dipole decay rate. Alternatively, for an atom at the antinode on a cycling transition,

$w_0 \sim 2$4

Unity (or larger) cooperativity ensures that a majority of photons emitted by the atom are funneled into the cavity mode ($w_0 \sim 2$5) (Shaw et al., 12 Jun 2025).

Key measured values include:

• Silicon microcavity arrays: $w_0 \sim 2$6 GHz, $w_0 \sim 2$7 GHz, $w_0 \sim 2$8 (current), $w_0 \sim 2$9 (with high-$_2$0 mirrors) (Derntl et al., 2013).
• Open-access single mode: $_2$1 MHz, $_2$2 MHz, $_2$3 (Doherty et al., 2022).
• MLA-based free-space arrays: $_2$4m, $_2$5, $_2$6 (extendable to $_2$7 in next-gen) (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026).

Parallel operation imposes uniformity: cross-site $_2$8, mode waist variations $_2$9 %, frequency non-degeneracies $w_0 \sim 10\,\mu$0 linewidth for $w_0 \sim 10\,\mu$195 % of sites in latest MLA approaches (Soper et al., 6 Feb 2026).

3. Control, Tuning, and Stability

Independent tunability is achieved via:

• Electrostatic actuation in silicon arrays, enabling voltage-controlled length shifts with pm-range precision, far exceeding the free-spectral-range ($w_0 \sim 10\,\mu$2 nm max, $w_0 \sim 10\,\mu$3V/pm voltage-displacement gradient) (Derntl et al., 2013).
• Piezoelectric tip/tilt of planar mirrors in open-access modules, allowing $w_0 \sim 10\,\mu$4 MHz single-pixel frequency precision with crosstalk $w_0 \sim 10\,\mu$5 MHz per 100 MHz shift (Doherty et al., 2022).
• Lens translation or SLM-based beam steering in MLA-based platforms for mode degeneracy across hundreds/thousands of sites, with degeneracy sensitivity $w_0 \sim 10\,\mu$60.2 MHz/μm lens translation (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026).
• Feedback-stabilized PID locking yields cavity length stability $w_0 \sim 10\,\mu$7 pm over seconds for fully parallelized arrays (Derntl et al., 2013).

Mechanical and environmental factors (field curvature, astigmatism, birefringence, and loss) are systematically budgeted. For MLA arrays, large field-of-view (FOV) was demonstrated (140 μm for 10 μm pitch, i.e., 600+ cavities), with optical tolerances achieving sub-micron stability per site for day-scale operation (Soper et al., 6 Feb 2026).

4. Imaging Operation: Resolution, Speed, and Parallelism

Each cavity functions as a confocal, diffraction-limited pixel whose spatial resolution is determined by $w_0 \sim 10\,\mu$8, while signal sensitivity scales with the Purcell factor $w_0 \sim 10\,\mu$9 and cavity finesse.

• Transverse pixel size is set by the Gaussian mode waist: $\mathcal{F} \sim 10^5$0–10 μm for state-of-the-art platforms, corresponding to transverse spatial resolution $\mathcal{F} \sim 10^5$1–10 μm (Derntl et al., 2013, Soper et al., 6 Feb 2026, Doherty et al., 2022).
• Axial sensitivity derives from stabilization bandwidth: pm-scale length shifts allow sensing of sub-nm refractive index or film thickness changes (Derntl et al., 2013).
• Frame rates theoretically extend to the cavity linewidth $\mathcal{F} \sim 10^5$2 (e.g., $\mathcal{F} \sim 10^5$310 MHz), though practical rates are limited by electronics and detection hardware; demonstrated bandwidths are $\mathcal{F} \sim 10^5$4ms per site for single-atom readout (Shaw et al., 12 Jun 2025).
• Multiplexed readout is achieved by simultaneous photodiode or fibre-array detection of cavity transmission, with cross-talk $\mathcal{F} \sim 10^5$5 % and per-site discrimination fidelities $\mathcal{F} \sim 10^5$6 % (Shaw et al., 12 Jun 2025).
• Sample handling is flexible: planar substrates are inserted between fibre block and mirror chip (silicon approach), or within the science plane of free-space arrays, with translation stages for raster scanning (Derntl et al., 2013, Soper et al., 6 Feb 2026).

5. Applications: Quantum, Nanophotonics, and Sensing

Cavity array microscopes are uniquely suited to quantum technology and parallel biosensing due to single-emitter sensitivity, strong coupling, and addressability.

• Quantum error correction and mid-circuit readout: Fast (ms-to-μs scale), non-destructive measurement across $\mathcal{F} \sim 10^5$740 sites, compatible with quantum error correction in neutral-atom systems (Shaw et al., 12 Jun 2025).
• Quantum networking: Each resonator output is fiber-coupled, supporting high-rate, parallel entanglement distribution and quantum node multiplexing (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026).
• Many-body photonics: Engineered cavity–cavity coupling enables realization of Jaynes–Cummings–Hubbard or Bose–Hubbard-type Hamiltonians for quantum simulation (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026).
• Nanoscale biosensing: Sub-nanometre axial detection, single-molecule sensitivity, and detection of thin films via cavity-enhanced absorption or fluorescence (Derntl et al., 2013, Doherty et al., 2022).
• Continuous quantum-nondemolition (QND) scanning: QND measurement of atomic density with subwavelength (e.g., $\mathcal{F} \sim 10^5$837 nm) spatial resolution is possible in engineered dark-state architectures with strong cavity coupling and optimized measurement rate $\mathcal{F} \sim 10^5$9, $_{00}$0, and narrow bandwidths (Yang et al., 2018).

6. Scalability, Limitations, and Outlook

Scalability is governed by FOV, site pitch, array uniformity, and losses:

• MLA-based free-space cavities demonstrate 600-site operational arrays with prospects for $_{00}$1–$_{00}$2 sites by reducing pitch, increasing FOV (via wide-field microscope objectives), and improving AR coatings (Soper et al., 6 Feb 2026).
• Mode uniformity, degeneracy, and tuning tolerances become more stringent at higher site counts, demanding $_{00}$3300 nm transverse, $_{00}$4m longitudinal element stability (Soper et al., 6 Feb 2026).
• Losses are dominated by coating and aspheric lens contributions; switching to objectives/microscope-grade optics reduces field curvature and extends area.
• Intrinsic finesse increases (e.g., $_{00}$5) yield higher cooperativity and $_{00}$680% light collection efficiency (Soper et al., 6 Feb 2026).
• Next-generation systems will support glass-cell integration, higher $_{00}$7 ($_{00}$8), and robust operation with neutral atom, solid-state, or nanophotonic emitter arrays (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026).
• Engineering intracavity elements (acousto-optic/electro-optic) permits reconfigurable photon hopping or Hamiltonian engineering (Shaw et al., 12 Jun 2025).

The cavity array microscope thus constitutes a flexible, massively parallel platform for quantum-enhanced measurement, addressable quantum simulation, and nanoscale imaging, with clear pathways toward further increases in scale, sensitivity, and bandwidth (Shaw et al., 12 Jun 2025, Soper et al., 6 Feb 2026, Derntl et al., 2013, Yang et al., 2018, Doherty et al., 2022).