Quantum Photonics
Quantum photonics exploits the quantum mechanical properties of light to enable capabilities in computing, communication, and sensing that classical optics cannot reach. Where classical optics treats light as a continuous electromagnetic wave, quantum photonics operates in the regime where the discrete nature of photons and their quantum correlations become essential. A single photon serves as an ideal carrier of quantum information: it travels at the speed of light, interacts weakly with its environment, and therefore preserves quantum coherence over long distances, including across deployed optical fiber.
The field encompasses three core capabilities. The first is the generation of non-classical light, including single photons, entangled photon pairs, and squeezed states whose noise falls below the vacuum level in one quadrature. The second is the manipulation of these states using linear optical elements such as beam splitters and phase shifters, together with nonlinear processes for state conversion. The third is the detection of light with single-quantum sensitivity, where individual photons are counted and, in the best detectors, their number is resolved. Integrated quantum photonic circuits bring these functions onto chip-scale waveguide platforms, providing the phase stability and reproducibility that bulk free-space optics cannot sustain at scale.
Quantum photonics is a convergence of quantum physics, optical engineering, and materials science. Progress in the field drives photonic quantum processors, quantum networks for provably secure communication, and quantum sensors that exceed classical measurement limits. This category provides comprehensive coverage of the sources, components, and systems that generate, manipulate, and detect light at the quantum level. Its emphasis is on the devices themselves and on the systems built directly from them; the companion article Quantum Photonic Devices and Systems, filed under photonic and optical computing, takes up the computing architectures—boson sampling machines, cluster-state processors, and photonic repeater networks—that those devices make possible.
Articles in This Category
Fundamental Concepts
Quantum States of Light
The quantum mechanical description of light differs fundamentally from the classical electromagnetic field. Photons are the elementary excitations of the quantized field, each carrying energy proportional to its frequency. Fock states, which contain a definite number of photons, form the basis for describing quantum light, although practical sources produce superpositions or statistical mixtures of these number states. Coherent states, the closest quantum analog of a classical laser field, carry Poissonian photon statistics, while squeezed states redistribute quantum noise so that one quadrature falls below the vacuum level at the expense of the other. Even the vacuum state, containing zero photons, exhibits quantum fluctuations that set fundamental limits on measurement precision.
Entanglement and Non-Classical Correlations
Quantum entanglement creates correlations between photons that exceed any possible classical explanation. Entangled photon pairs exhibit correlated outcomes across complementary measurement bases, the property that underlies entanglement-based quantum key distribution: any attempt to gain information about the shared state disturbs the correlations and reveals the intrusion. The violation of a Bell inequality, demonstrated in increasingly loophole-free experiments, confirms these correlations cannot arise from any local hidden-variable theory. Beyond two-photon entanglement, multiphoton states such as Greenberger-Horne-Zeilinger (GHZ) and cluster states provide the resources for measurement-based quantum computing and for sensing at the Heisenberg limit.
Photon Interference and Indistinguishability
Much of quantum photonics rests on interference between indistinguishable photons rather than on any direct interaction between them. Photons do not interact appreciably in ordinary linear media, so the effective nonlinearity used in photonic quantum gates arises from interference followed by measurement. The canonical demonstration is the Hong-Ou-Mandel effect: when two identical photons enter the two input ports of a balanced beam splitter, they always leave through the same output port, and the coincidence rate between the two outputs falls to zero. The depth of this dip measures how nearly identical the photons are in wavelength, bandwidth, polarization, arrival time, and spatial mode. Indistinguishability degrades quickly with spectral impurity or timing jitter, and multiphoton experiments scale poorly when visibility falls short, so Hong-Ou-Mandel visibility is one of the most widely reported figures of merit for single-photon sources.
Photon Detection and Measurement
Detecting individual photons requires devices with single-quantum sensitivity and low dark-count noise. Single-photon avalanche diodes operating in Geiger mode register a photon through self-sustaining carrier multiplication but cannot resolve photon number. Superconducting nanowire single-photon detectors (SNSPDs) offer the best overall combination of efficiency, timing resolution, and noise, particularly at telecommunications wavelengths near 1550 nanometers, where fiber-coupled devices have reached system detection efficiency of roughly 98 percent. Timing jitter of a few tens of picoseconds is routine, and careful optical filtering can hold dark counts below one count per second, although the best efficiency, the best jitter, and the lowest noise are seldom obtained in the same device. Transition-edge sensors add true photon-number resolution by measuring the heat deposited by absorbed photons, at the cost of microsecond-scale response and millikelvin operating temperatures. Because measurement projects a quantum state onto a definite outcome, detection is not a passive readout but an active step in many photonic protocols, supplying the effective nonlinearity in linear optical gates and heralding the presence of a photon in probabilistic sources.
Light-Matter Interfaces and Quantum Memories
Photons excel at carrying quantum information but not at storing it, because a photon in flight cannot simply be held in place. Quantum memories bridge this gap by mapping a photonic state onto a long-lived material excitation and releasing it later on demand. Established approaches use atomic ensembles in warm or cold vapor cells, rare-earth ions doped into crystals such as praseodymium or europium in yttrium orthosilicate, and single emitters such as trapped ions or diamond color centers coupled to optical cavities. The relevant figures of merit are storage efficiency, storage time, bandwidth, fidelity, and the number of modes a device can hold at once; no platform yet optimizes all of them together. Memories matter most for quantum repeaters, which must hold entanglement in one link while a neighboring link succeeds, and for synchronizing the probabilistic sources used in photonic computing. A related class of interface, the quantum transducer, converts a quantum state between the microwave domain in which superconducting processors operate and the optical domain in which it can travel through fiber. Cavity optomechanics supplies the leading route, using a mechanical resonator that couples to both an optical cavity and a microwave circuit as the intermediary; piezo-optomechanical and electro-optic devices pursue the same conversion by other means. Transduction efficiency, added noise, and bandwidth remain the limiting figures of merit, and improving them is a prerequisite for networking superconducting quantum computers over optical links.
Integrated Quantum Photonics
Integrating quantum photonic components onto a chip provides the stability, scale, and reproducibility that practical quantum systems require. A photonic integrated circuit builds interferometers from waveguides, directional couplers, and phase shifters whose path lengths are fixed by lithography rather than by mechanical alignment, so an assembly that would drift by a wavelength with a fraction of a degree of temperature change in free space instead holds phase over hours. Meshes of Mach-Zehnder interferometers, each with an electrically controlled phase shifter, implement arbitrary programmable linear-optical transformations across many modes, and single chips have carried hundreds of tunable elements. Co-integrating sources and detectors with these passive circuits, whether monolithically or by bonding, moves the field toward complete quantum systems on a single die. Just as important, fabricating in a commercial semiconductor foundry makes devices repeatable from wafer to wafer, which matters more than any single record device once systems must be built in quantity.
Applications
Quantum Communication
Quantum key distribution (QKD) uses the laws of quantum mechanics to establish shared cryptographic keys whose secrecy rests on physics rather than on computational difficulty. Any attempt to intercept quantum-encoded information disturbs the underlying states in a detectable way. Prepare-and-measure protocols such as BB84 encode bits on single photons, while entanglement-based protocols such as E91 distribute correlated pairs. Practical systems substitute attenuated laser pulses for true single photons and recover a rigorous security proof through the decoy-state method, which detects the photon-number-splitting attack that weak coherent pulses would otherwise permit. Measurement-device-independent variants remove the detector, historically the most attacked component, from the trust model entirely. Fiber QKD is a commercial technology over metropolitan distances; a laboratory twin-field experiment reported in 2023 carried keys over 1,002 kilometers of ultra-low-loss fiber without a repeater, and deployed backbones such as the Beijing-to-Shanghai link cover comparable distances using trusted relay nodes. Satellite links have distributed entanglement between ground stations more than 1,200 kilometers apart. Quantum repeaters, which combine quantum memories with entanglement swapping and purification, aim to extend secure links to continental and global scale without trusting the intermediate nodes.
Optical Quantum Computing
Photons serve as qubits, with information encoded in polarization, spatial path, time-bin, or frequency degrees of freedom. Linear optical quantum computing, following the scheme proposed by Knill, Laflamme, and Milburn in 2001, implements gates with beam splitters and phase shifters, using measurement and feed-forward to supply the effective nonlinearity needed for universal operation. The measurement-based, or one-way, model recasts the same idea as adaptive single-qubit measurements on a large entangled cluster state, which suits photonics because the cluster can be produced and consumed in a stream. Boson sampling provides a more specialized route to quantum computational advantage using only single photons or squeezed light, passive interferometry, and photon-number detection; the programmable Gaussian boson sampling machine Borealis, built by Xanadu and operating on 216 squeezed-light modes, produced samples in 2022 that were estimated to lie far beyond the reach of classical supercomputers. Continuous-variable approaches instead encode information in the quadratures of squeezed states and pursue fault tolerance through bosonic codes such as the Gottesman-Kitaev-Preskill encoding, offering trade-offs distinct from those of discrete-variable qubits.
Quantum Sensing and Metrology
Non-classical states of light enable measurements with precision beyond classical limits. Squeezed vacuum injected into the dark port of the Advanced LIGO and Advanced Virgo interferometers reduces quantum noise and improves sensitivity to gravitational waves. Frequency-independent squeezing already contributed a useful broadband gain during the third observing run; for the fourth run, LIGO added 300-meter filter cavities that rotate the squeezed quadrature as a function of frequency, so that photon shot noise is suppressed at high frequencies and radiation-pressure noise at low frequencies within the same measurement. Because detection volume scales with the cube of range, even a modest improvement in strain sensitivity multiplies the astrophysical event rate. Path-entangled states such as NOON states allow phase estimation whose uncertainty scales as the inverse of the photon number, the Heisenberg limit, rather than the inverse square root that bounds classical measurements, although loss erodes this advantage quickly. Quantum illumination uses entangled photons to improve target detection against a bright thermal background, a principle explored for low-power radar and lidar, and ghost imaging and sub-shot-noise microscopy apply related ideas where the probe power must stay low, as in the imaging of photosensitive biological samples.
Quantum Random Number Generation
Quantum random number generators exploit the intrinsic indeterminacy of quantum measurement to produce outcomes that are not merely difficult to predict but unpredictable in principle, unlike the deterministic algorithms that back conventional pseudorandom generators. Typical implementations send single photons onto a beam splitter and record which path each takes, measure the vacuum fluctuations of an optical field with a balanced homodyne detector, or digitize the phase noise of a semiconductor laser pulsed near threshold. The vacuum-fluctuation and phase-noise designs are attractive for deployment because they use ordinary photodiodes rather than single-photon detectors and reach gigabit-per-second output rates. Device-independent and semi-device-independent protocols certify the randomness of the output from measured data, reducing the trust that must be placed in the hardware itself. Quantum random number generation is among the most commercially mature quantum photonic technologies, available as rack appliances, expansion cards, and chip-scale modules that seed cryptographic key material.
Engineering Challenges
Optical Loss
Loss dominates nearly every quantum photonic system, because a lost photon cannot be recovered and the no-cloning theorem forbids the optical amplifiers that make classical fiber links practical. Every decibel of attenuation therefore converts directly into lost rate or lost fidelity. Standard single-mode fiber attenuates roughly 0.2 decibel per kilometer near 1550 nanometers, which limits point-to-point quantum key distribution to a few hundred kilometers before the secret-key rate falls below a useful level. On chip, silicon nitride waveguides routinely achieve propagation loss below 0.1 decibel per centimeter, and ultra-low-loss designs reach single-digit decibels per meter, but each fiber-to-chip facet, coupler, and switch adds a fixed penalty. In large interferometers these interface losses accumulate faster than propagation loss, so reducing insertion loss at the boundaries between components is often worth more than improving the components themselves.
Probabilistic Sources and Multiplexing
The most common photon-pair sources rely on spontaneous parametric down-conversion in nonlinear crystals or spontaneous four-wave mixing in waveguides and microresonators. Both are probabilistic: pairs appear at random times, and raising the pump power to increase the rate also raises the chance of emitting more than one pair in the same interval, which degrades heralded single-photon purity. The probability that many independent sources fire together falls exponentially with their number, so simply replicating sources does not scale. Two remedies dominate. Multiplexing operates many sources, time bins, or spectral modes in parallel and uses fast switches to route whichever one succeeded into a common output, trading switch loss for a higher heralding probability. Deterministic emitters such as semiconductor quantum dots in micropillar or photonic-crystal cavities instead emit a photon on demand when triggered, at the cost of cryogenic operation and, for most material systems, emission wavelengths outside the telecommunications bands.
Cryogenics, Packaging, and Wavelength Choice
The best single-photon detectors are superconducting and require closed-cycle cryocoolers operating near 1 to 4 kelvin, or lower still for transition-edge sensors. Cryogenics adds cost, volume, and power draw, and it constrains how detectors are packaged alongside room-temperature electronics and photonic chips. Fiber feedthroughs, thermal loading, and readout wiring limit how many channels one cryostat can support, which becomes the binding constraint as systems grow from a handful of detectors toward arrays of hundreds. Semiconductor avalanche diodes avoid cryogenics and integrate readily with CMOS electronics, but their InGaAs variants show higher dark counts, afterpulsing, and lower efficiency at telecommunications wavelengths. Wavelength choice compounds the problem, because many of the best solid-state emitters and memories operate in the visible or near infrared while fiber networks demand 1550 nanometers. Quantum frequency conversion bridges the gap by shifting a photon's wavelength in a nonlinear medium while preserving its quantum state, at the price of additional loss and added noise.
Material Platforms and Their Trade-offs
No single material platform supplies every function a quantum photonic system needs, so platform selection is a compromise. Silicon offers dense, low-cost circuits fabricated in existing foundries and a strong third-order nonlinearity for four-wave mixing, but it absorbs light below roughly 1.1 micrometers and, being centrosymmetric, provides no linear electro-optic effect. Silicon nitride gives very low loss and a transparency window reaching into the visible, yet it likewise lacks a second-order nonlinearity and depends on comparatively slow thermo-optic phase tuning. Thin-film lithium niobate supplies fast, low-voltage electro-optic modulation together with a strong second-order nonlinearity useful for both frequency conversion and photon-pair generation. Silica waveguides remain attractive where the lowest loss and best fiber matching matter more than density. Heterogeneous integration, which bonds or transfers one material onto another, aims to combine these strengths on a common substrate, and it is a principal reason that quantum photonics has converged on foundry-compatible processes.
Outlook
Quantum photonics sits at the frontier of quantum technology, translating fundamental physics into practical devices and systems. The intrinsic advantages of photons for quantum information, including room-temperature transmission, weak coupling to the environment, and compatibility with existing telecommunications infrastructure, make photonic approaches especially promising for near-term work in secure communication and sensing. The field is also unusually stratified in maturity: quantum random number generators and squeezed-light-enhanced interferometers are deployed and doing useful work today, metropolitan quantum key distribution is a commercial product whose role alongside post-quantum cryptography is still being settled, and fault-tolerant photonic computing remains a research program measured in orders of magnitude rather than percentages.
The central engineering challenges are consistent across those applications: deterministic single-photon generation, low-loss components and interfaces, efficient photon-number-resolving detection, practical quantum memories, and integration at a scale that foundry processes can sustain. None of these is a matter of discovering new physics; each is a question of engineering yield, loss budgets, and packaging. Progress on these fronts, surveyed across the topics in this category, will determine how far the quantum properties of light can be pushed in computing, communication, and metrology.