Electronics Guide

Quantum States of Light

The electromagnetic field must be described quantum mechanically to understand the properties that distinguish quantum light sources from classical sources. The quantum harmonic oscillator model represents each mode of the field with creation and annihilation operators that add or remove individual photons. The resulting Fock states, or number states, contain a definite number of photons and form the basis for describing any quantum state of light.

Classical light sources produce coherent states, which are eigenstates of the annihilation operator with Poissonian photon number statistics. Thermal light exhibits super-Poissonian statistics with photon bunching. In contrast, quantum light sources can produce states with sub-Poissonian statistics, where the photon number variance falls below the mean, exhibiting photon antibunching. This antibunching behavior has no classical analog and provides a definitive signature of non-classical light.

The density matrix formalism describes mixed quantum states and enables calculation of observable quantities. For practical sources that produce imperfect quantum states, the density matrix captures the statistical mixture of pure states arising from technical imperfections. Quantum state tomography reconstructs the density matrix from measurements, providing complete characterization of the quantum state produced by a source.

Photon Statistics and Correlation Functions

The statistical properties of light are characterized by correlation functions that describe the probability of detecting photons at different times. The second-order correlation function g2(tau) measures the normalized probability of detecting two photons separated by time tau relative to the probability expected for uncorrelated detections. For classical light, g2(0) is always greater than or equal to one, but quantum light sources can achieve g2(0) less than one, demonstrating antibunching.

The Hanbury Brown and Twiss experiment measures g2(tau) by splitting the light beam and detecting coincidences between two single-photon detectors as a function of their relative delay. For an ideal single-photon source, g2(0) equals zero because a single photon cannot trigger simultaneous detections at both detectors. A measured g2(0) below 0.5 is the conventional threshold for attributing the emission to a single quantum emitter rather than to two or more, and the best sources reach 0.01 or lower.

Higher-order correlation functions provide additional characterization beyond the second-order statistics. The third-order correlation g3 resolves three-photon events whose weight the second-order function alone cannot separate from a mixture of a single-photon signal and an uncorrelated background. Complete characterization of the photon number distribution requires either photon-number-resolving detection, such as transition-edge sensors or multiplexed click detectors, or reconstruction from a series of correlation orders.

Photon Antibunching

Photon antibunching occurs when photon detection events exhibit a minimum probability at zero time delay, the opposite of the bunching observed in thermal light. This effect arises from single-quantum emitters that must be re-excited before emitting another photon, creating a dead time during which no second photon can be produced. The antibunching timescale reflects the excited state lifetime of the emitter.

Measurement of antibunching requires detectors with timing resolution better than the antibunching timescale, typically nanoseconds or faster for solid-state emitters. The detector dark count rate and afterpulsing probability set lower limits on measurable g2(0) values. Time-correlated single-photon counting electronics record the arrival time of each detection event, enabling construction of the coincidence histogram that yields g2(tau).

The depth of the antibunching dip depends on both the source properties and the measurement conditions. Background light, multiphoton emission, and imperfect collection efficiency all increase the measured g2(0). Proper background subtraction and careful optical filtering are essential for accurate characterization. Pulsed excitation with time-gated detection further suppresses background contributions by rejecting events outside the expected emission window.

Hong-Ou-Mandel Interference

The Hong-Ou-Mandel effect demonstrates quantum interference between two indistinguishable photons incident on a beam splitter. When identical photons arrive simultaneously at opposite input ports, they always exit together through the same output port, never through opposite ports. This coalescence arises from destructive interference of the probability amplitudes for the two photons to exit separately, producing the characteristic dip in coincidence counts.

The visibility of the Hong-Ou-Mandel dip measures the indistinguishability of the two photons, with perfect indistinguishability yielding complete suppression of coincidences. Any distinguishing information in polarization, frequency, timing, or spatial mode reduces the interference visibility. Two genuine single photons interfere with full visibility in principle, whether they come from the same emitter or from two independent ones, because the effect requires no phase relationship between the inputs. The 50 percent ceiling often quoted for two-source interference belongs to attenuated laser pulses, where the Poissonian statistics allow both detected photons to have entered through the same port; it does not apply to Fock-state sources, and independent heralded sources have exceeded 90 percent visibility.

Hong-Ou-Mandel interference serves as both a characterization tool and a functional element in quantum photonic systems. Measuring the interference visibility between photons from different sources verifies their suitability for multiphoton quantum operations. In quantum computing and networking, the Hong-Ou-Mandel effect enables entangling gates and Bell state measurements that form the basis of linear optical quantum information processing.

Requirements and Figures of Merit

An ideal single-photon source produces exactly one photon on demand in a well-defined optical mode with unity probability. Real sources are characterized by several figures of merit that quantify their departure from this ideal. The single-photon purity, measured by g2(0), indicates the probability of multiphoton emission events. The efficiency quantifies the probability that triggering the source produces a detected photon in the collection optics. The indistinguishability measures how well consecutive photons interfere, essential for quantum information applications.

Brightness characterizes the rate of single-photon emission, typically specified as detected photons per second or per excitation pulse. High brightness is essential for practical applications where measurement times must be reasonable. However, increasing brightness often compromises purity or indistinguishability, requiring careful optimization for specific applications. The brightness into a single spatial mode suitable for fiber coupling is the relevant metric for many practical systems.

Additional source characteristics include the emission wavelength, spectral linewidth, polarization purity, and temporal profile of the emitted photons. Telecommunication wavelengths around 1310 nm and 1550 nm are preferred for fiber-based quantum communication due to low transmission loss. Narrow spectral linewidth enables high-visibility interference and efficient filtering. Transform-limited temporal profiles with controlled shape optimize the wavepacket for interference and detection.

Triggered versus Heralded Sources

Single-photon sources operate in two fundamentally different modes: triggered emission where the photon is produced on demand, and heralded emission where detection of a correlated photon signals the presence of the single photon. Triggered sources based on single quantum emitters produce photons in response to an excitation pulse, with the emission time determined by the excitation. This deterministic timing is valuable for synchronous quantum information processing but requires emitters with high quantum efficiency.

Heralded sources use parametric processes that create photon pairs, with detection of one photon heralding the presence of its partner. The probabilistic nature of pair generation means the single photon is not produced on demand, but the heralding detection provides precise timing information. Heralded sources achieve high purity because the heralding process projects onto a single-photon state, though the heralding efficiency limits the overall success probability.

Multiplexing schemes combine multiple probabilistic sources to approach deterministic operation. Spatial multiplexing uses an array of heralded sources with active switching to route the successful channel to the output. Temporal multiplexing stores photons from successful heralding events and releases them in synchronized time bins. Both approaches increase the probability of producing a photon on demand at the cost of system complexity and added loss from switching elements.

Source Architectures

Cavity-enhanced sources place the quantum emitter inside an optical resonator that modifies the emission properties through the Purcell effect. The cavity increases the emission rate into the cavity mode while suppressing emission into other modes, improving both brightness and collection efficiency. Photonic crystal cavities achieve small mode volumes for strong Purcell enhancement in monolithic structures. Fabry-Perot microcavities provide tunable coupling with open access for various emitter types.

Waveguide-integrated sources couple emitters directly to optical waveguides that collect and route the emitted photons. The high refractive index contrast of semiconductor waveguides enables efficient coupling of emission into guided modes. Diamond waveguides with embedded color centers and gallium arsenide waveguides with quantum dots demonstrate this approach. Photonic circuits can then process and route the single photons on-chip.

Fiber-coupled sources deliver photons directly into optical fiber for transmission and integration with fiber-based quantum systems. Achieving efficient fiber coupling requires matching the emission mode to the fiber mode, often using lens systems or tapered fiber couplers. Sources based on atoms or ions in free space typically achieve lower fiber coupling efficiency than solid-state sources designed for integrated collection.

Self-Assembled Quantum Dots

Self-assembled quantum dots form during epitaxial growth when a thin layer of semiconductor with larger lattice constant deposits on a substrate, with strain driving the formation of nanoscale islands. Indium arsenide dots on gallium arsenide substrates are the most developed system, emitting in the 900-1000 nm wavelength range. Indium arsenide dots on indium phosphide substrates emit at telecommunication wavelengths around 1550 nm, enabling direct fiber transmission.

The three-dimensional quantum confinement in these dots creates discrete energy levels analogous to artificial atoms. Optical excitation creates electron-hole pairs that relax to the ground state and recombine to emit single photons. The discrete level structure ensures that a single excitation produces a single photon, with the emission wavelength determined by the dot size and composition. Size variation between dots causes spectral inhomogeneity, requiring selection of individual dots with suitable properties.

Radiative lifetimes in self-assembled quantum dots range from hundreds of picoseconds to a few nanoseconds, depending on the dot structure and environment. The relatively fast emission compared to atomic systems enables high repetition rate operation. However, interaction with phonons causes pure dephasing that broadens the emission linewidth beyond the transform limit, reducing photon indistinguishability at elevated temperatures.

Excitation Schemes

Above-band excitation creates carriers in the surrounding semiconductor matrix that subsequently capture into the quantum dot and relax to the ground state before emission. This non-resonant approach is experimentally simple but introduces timing jitter from the capture and relaxation processes, broadening the temporal profile and reducing indistinguishability. The broad excitation spectrum also allows simultaneous excitation of multiple dots, requiring spatial or spectral filtering.

Resonant excitation drives the optical transition directly, eliminating the timing jitter associated with carrier relaxation. Two-photon excitation of the biexciton state enables resonant pumping while spectrally separating the pump from the emission. Phonon-assisted excitation uses slightly detuned pulses that drive transitions mediated by phonon emission, achieving the benefits of resonant excitation with less stringent laser filtering requirements.

Electrical injection offers a path toward practical devices by eliminating the need for optical pumping lasers. PIN diode structures incorporating quantum dots in the intrinsic region enable current-driven single-photon emission. Challenges include ensuring that only one dot lies within the injection area, achieving high injection efficiency, and maintaining coherence in the presence of the charge noise that the current itself introduces. Electrically driven devices have produced antibunched emission with good purity, but they have not matched the indistinguishability of the best optically pumped sources.

Diode structures earn their place in optically pumped sources as well, for a reason unrelated to injection. Embedding the dot in the intrinsic region of a gated diode fixes its charge state and screens fluctuating charges in the surrounding material, which is the dominant cause of spectral wandering. The same gate provides Stark tuning of the emission wavelength, allowing separate dots, or a dot and a cavity resonance, to be brought into alignment after fabrication. Nearly transform-limited linewidths from quantum dots depend on this electrical stabilization more than on any improvement in the dots themselves.

Photonic Structures for Enhancement

Photonic crystal cavities provide strong light confinement in wavelength-scale volumes, enhancing spontaneous emission through the Purcell effect. Their combination of high quality factor and mode volumes near a cubic wavelength yields measured Purcell factors of several tens, shortening the radiative lifetime and steering most of the emission into the cavity mode. Realizing the theoretical enhancement demands that the dot sit at the field maximum and be spectrally matched to the resonance, so fabrication relies on either site-controlled dot growth or post-fabrication alignment in which the cavity is patterned around a dot located by imaging.

Micropillar cavities use distributed Bragg reflector mirrors above and below the quantum dot layer to form vertical Fabry-Perot resonators. The circular pillar geometry provides lateral mode confinement and enables efficient coupling to external optics. Micropillars achieve high collection efficiency with moderate Purcell enhancement, representing a well-developed technology for high-brightness sources.

Circular Bragg gratings, or bullseye cavities, combine vertical cavity enhancement with broadband lateral mode engineering. The radial grating structure provides strong vertical emission directionality without requiring precise wavelength matching between dot and cavity. This broadband enhancement relaxes fabrication tolerances and enables use with spectrally diverse quantum dots, simplifying source development.

Performance and Applications

State-of-the-art quantum dot single-photon sources achieve simultaneous high purity, efficiency, and indistinguishability that place them among the best single-photon sources of any type. Demonstrated values include g2(0) below 0.01, Hong-Ou-Mandel visibility above 95 percent, and end-to-end efficiency into a single-mode fiber above 50 percent in the best laboratory demonstrations. Packaged, permanently fiber-pigtailed devices deliver less, with reported output-fiber brightness in the range of 20 to 35 percent, but they hold alignment through cooldowns and run for hours without retuning, which matters more than peak efficiency for a deployable instrument.

Quantum key distribution using quantum dot sources has demonstrated secure key generation at practical rates over metropolitan-scale fiber links. The combination of deterministic triggering, high brightness, and excellent single-photon purity provides advantages over attenuated laser sources that suffer from multiphoton emission. Integration with on-chip photonic circuits promises compact, scalable quantum communication transmitters.

Linear optical quantum computing requires sources of indistinguishable photons that can interfere with high visibility. Quantum dot sources producing streams of identical photons enable multiphoton experiments including boson sampling and photonic graph state generation. Scaling to larger numbers of photons requires multiple synchronized sources with matched emission properties, a significant engineering challenge being addressed through integrated photonic approaches.

Nitrogen-Vacancy Centers

The nitrogen-vacancy (NV) center in diamond consists of a substitutional nitrogen atom adjacent to a vacant lattice site, forming a point defect with remarkable optical and spin properties. The negatively charged NV center has a ground state electron spin triplet that can be optically initialized, coherently manipulated, and read out at room temperature. Single NV centers emit single photons while also providing a long-lived quantum memory in the spin degree of freedom.

Optical emission from NV centers occurs through a zero-phonon line at 637 nm and a broad phonon sideband extending to longer wavelengths. Only about 3-4% of emission occurs in the zero-phonon line, with the remainder spread across the phonon sideband. This low Debye-Waller factor is the central weakness of the NV center for photonic networking, because only the zero-phonon photons are useful for interference and the rest must be discarded; cavity enhancement raises that fraction by shortening the zero-phonon line lifetime relative to the sideband. The combination of room-temperature operation with an optically addressable spin remains the NV center's distinguishing strength, and it is the reason the defect underpins commercial magnetometry as well as quantum networking research.

Creating NV centers involves introducing nitrogen during diamond growth or through ion implantation, followed by vacancy creation through electron irradiation and annealing to form the NV complex. Controlled positioning of NV centers through masked implantation or delta-doping during growth enables integration with photonic structures. The diamond host provides excellent thermal conductivity and mechanical stability for practical devices.

Silicon-Vacancy and Group-IV Centers

Silicon-vacancy (SiV) centers incorporate a silicon atom in a split-vacancy configuration between two vacant lattice sites. The high symmetry of this structure results in strong zero-phonon line emission with about 70% of light in the ZPL at low temperatures. The narrow linewidth and high Debye-Waller factor make SiV centers excellent candidates for quantum networking where photon indistinguishability is critical.

The SiV center has an orbital doublet ground state that provides both advantages and challenges. The orbital degree of freedom enables fast optical transitions with large oscillator strength. However, the ground state splitting is only about 50 GHz, so single-phonon transitions between the orbital branches dephase the spin unless the device is cooled well below one kelvin. Operating near 100 millikelvin suppresses this channel by orders of magnitude and yields electron spin coherence times of roughly ten milliseconds under dynamical decoupling, with spin relaxation times exceeding one second. Nearby nuclear spins, such as the silicon-29 isotope of the defect itself, serve as longer-lived memories built on that electron spin.

Other group-IV color centers including germanium-vacancy (GeV), tin-vacancy (SnV), and lead-vacancy (PbV) offer different trade-offs between optical properties and spin coherence. The heavier group-IV elements produce larger ground state splittings, potentially enabling longer coherence times at more accessible temperatures. The tin-vacancy center, with a ground state splitting near 850 GHz, suppresses phonon-induced dephasing at temperatures around one kelvin that are far easier to reach than the millikelvin regime required by the silicon-vacancy center. These emerging color center platforms are being optimized for quantum network applications requiring both optical interfacing and spin-based quantum memory.

Integration and Photonic Engineering

Diamond photonic structures enhance light collection and emission properties of color centers. Diamond waveguides fabricated through angled etching or thin-film bonding route single photons to collection optics with improved efficiency. Photonic crystal cavities in diamond achieve Purcell enhancement, increasing the zero-phonon line fraction and reducing lifetime for faster operation.

Heterogeneous integration approaches bond diamond membranes containing color centers to other photonic platforms including silicon, silicon nitride, and gallium phosphide. This hybrid approach combines the quantum properties of diamond color centers with the mature processing and design capabilities of established photonic platforms. Careful interface engineering maintains optical quality while enabling complex photonic circuit functionality.

Fiber-coupled diamond devices provide a practical interface for quantum networking applications. Tapered optical fibers brought into near-field coupling with diamond nanophotonic structures achieve efficient single-photon extraction. Open Fabry-Perot microcavities with fiber mirrors enable tunable enhancement of color center emission with direct fiber integration. These approaches support field-deployable quantum communication systems.

Quantum Network Applications

Color centers in diamond enable quantum repeaters that extend quantum communication beyond the direct transmission distance limit. The spin-photon interface allows entanglement between the stationary spin qubit and the flying photonic qubit, enabling heralded remote entanglement between distant nodes. The first such link ran between two NV centers roughly 1.3 kilometers apart on the Delft campus in 2015. A metropolitan-scale version reported in 2024 entangled NV-center nodes in Delft and The Hague, about 10 kilometers apart and joined through a midpoint station by roughly 25 kilometers of deployed telecommunication fiber, using quantum frequency conversion to move the emitted photons into the telecom band.

NV center electron spins retain coherence for hundreds of milliseconds under dynamical decoupling at room temperature, approaching one second at cryogenic temperatures, which is long enough to serve as the working memory of a repeater node. Carbon-13 nuclear spins in the surrounding lattice, controlled through their hyperfine coupling to the NV electron spin, extend storage far further and form small multiqubit registers suitable for entanglement distillation and error correction. Scaling to larger numbers of coupled spins, and to many nodes operating simultaneously, remains an active research challenge.

The first loophole-free Bell test used NV centers in separated laboratories, closing the detection and locality loopholes in a single experiment. High-fidelity state preparation, fast random basis selection at each station, and space-like separation of the detection events together satisfy the requirements. Photonic Bell tests reported soon afterward accumulated far higher statistical significance, because entangled photon sources produce data at rates that heralded spin-spin entanglement cannot match; the diamond experiment remains notable for the entanglement swapping and remote state readout it demonstrated rather than for its raw significance level.

Types of Photonic Entanglement

Photon pairs can be entangled in various degrees of freedom including polarization, time-energy, and orbital angular momentum. Polarization entanglement is most commonly used because polarization states are easily prepared, manipulated, and measured with standard optical components. Time-energy entanglement exploits the correlation in emission time and energy conservation in pair generation, useful for long-distance quantum communication where polarization may drift.

The four Bell states form a maximally entangled basis for two-qubit systems. The phi-plus and phi-minus states have correlated polarizations (both horizontal or both vertical), while psi-plus and psi-minus have anticorrelated polarizations. Different source configurations naturally produce different Bell states, with the choice depending on application requirements. Complete Bell state analysis requires distinguishing all four states, which is impossible with linear optics alone without ancillary photons.

Hyperentanglement refers to simultaneous entanglement in multiple degrees of freedom. A photon pair can be entangled in both polarization and time-bin, for example, providing a larger Hilbert space for information encoding. Hyperentangled states enable more efficient Bell state analysis and increase the information capacity of quantum communication channels. The additional degrees of freedom also provide redundancy for error correction.

Spontaneous Parametric Down-Conversion

Spontaneous parametric down-conversion (SPDC) is the most widely used process for generating entangled photon pairs. In this second-order nonlinear optical process, a pump photon interacting with a nonlinear crystal spontaneously converts into two lower-energy photons called signal and idler. Energy and momentum conservation constrain the frequencies and directions of the generated photons, creating correlations that can be configured for various entanglement types.

Entanglement arises when the collection geometry erases every trace of which pathway produced the pair. The crossed-cone geometry uses a type-II crystal, whose signal and idler emerge with orthogonal polarizations on two cones that intersect along two directions; a pair collected from the intersections carries no information about which cone supplied which photon, leaving it in a polarization-entangled state of the psi family. The two-crystal geometry stacks a pair of thin type-I crystals with orthogonal optic axes and pumps them with diagonally polarized light, so the pair may originate in either crystal and the two amplitudes add coherently to give a state of the phi family. Both arrangements require compensation of the spatial and temporal walk-off that would otherwise reveal the emission pathway, and both rest on the phase-matching engineering treated under Parametric Down-Conversion below.

Time-energy and time-bin entanglement follow from the same process without any polarization structure. A pump with a coherence time much longer than the pair correlation time leaves the emission moment undetermined, and a Franson arrangement, in which each photon passes through an unbalanced interferometer of matched delay, converts that indeterminacy into two-photon interference in the coincidence rate; visibility above 71 percent violates a Bell inequality. Pulsed sources make the same encoding discrete by splitting the pump into early and late components before the crystal, which is the standard entanglement format for fiber links because time bins survive the polarization drift of deployed fiber.

Every SPDC source trades rate against fidelity. Pair emission per mode follows thermal statistics, so the probability of two pairs in one detection window grows roughly as the square of the single-pair probability, and the extra photons appear as accidental coincidences that dilute the measured entanglement. Raising the pump power therefore lifts the coincidence rate while lowering the coincidence-to-accidental ratio and the Bell violation, and practical sources are operated at a pair probability per pulse of a few percent or less, where the multipair penalty stays tolerable.

Four-Wave Mixing Sources

Four-wave mixing (FWM) in third-order nonlinear media provides an alternative to SPDC for photon pair generation. Two pump photons convert to signal and idler photons through the chi-3 nonlinearity, with energy and momentum conservation again creating the correlations. The process can occur in optical fibers, silicon waveguides, and other centrosymmetric materials where second-order nonlinearity is absent.

Spontaneous four-wave mixing in optical fibers benefits from the long interaction lengths possible in fiber and the natural fiber coupling of generated photons. However, Raman scattering produces noise photons that limit the signal-to-noise ratio, requiring cooling or spectral filtering. Photonic crystal fibers with engineered dispersion enable phase matching at convenient wavelengths while suppressing Raman noise through the modified phonon spectrum.

Silicon waveguides provide compact, CMOS-compatible sources of correlated photon pairs through FWM. The high nonlinearity of silicon enables pair generation in sub-millimeter to sub-centimeter devices, and ring resonator enhancement raises the generation rate for a given pump power while defining the signal and idler channels. Silicon carries a specific penalty at telecom wavelengths: two-photon absorption of the pump generates free carriers that absorb the very photons the device is making and shift the resonance, so pair rates saturate and then decline as pump power rises. Silicon nitride avoids this because its wider bandgap suppresses two-photon absorption, at the cost of a smaller nonlinear coefficient that must be recovered through longer devices or higher-quality resonators. Either platform supports integration of sources with modulators, filters, and detectors for complete quantum photonic systems on chip.

Source Characterization

Entanglement verification requires measurements that demonstrate correlations exceeding classical bounds. Bell inequality measurements compare coincidence rates for different measurement basis choices, with violations proving entanglement in a device-independent manner. The CHSH inequality is most commonly used, with the S parameter exceeding 2 proving nonlocality and values approaching 2.83 indicating high-quality entanglement.

Quantum state tomography reconstructs the full two-photon density matrix from measurements in multiple bases. Sixteen coincidence measurements combining horizontal, vertical, diagonal, antidiagonal, right-circular, and left-circular polarization settings are exactly sufficient to fix the 15 independent parameters of the two-qubit density matrix plus its normalization; larger overcomplete sets add redundancy that averages down systematic error. Maximum likelihood estimation finds the physical density matrix most consistent with the measurement data, from which entanglement measures such as fidelity and concurrence are calculated.

Practical source characterization also includes coincidence-to-accidental ratio (CAR), spectral properties, and timing jitter. High CAR indicates low multipair emission and background noise. Narrow spectral bandwidth enables efficient filtering and high-visibility interference. Timing jitter affects the precision of time-correlated measurements and the coincidence window required for pair detection.

Principles of Squeezed Light

Squeezed states of light have reduced quantum fluctuations in one quadrature at the expense of increased fluctuations in the conjugate quadrature, maintaining the minimum uncertainty product required by the Heisenberg principle. The quadrature with reduced noise falls below the standard quantum limit (SQL) that constrains coherent states and vacuum fluctuations. This noise reduction enables measurements with precision beyond the SQL, finding application in gravitational wave detection and quantum information.

Amplitude squeezing reduces fluctuations in the photon number, producing sub-Poissonian statistics. Phase squeezing reduces fluctuations in the optical phase. The squeezing angle determines which quadrature combination has minimum noise. Homodyne detection with a local oscillator measures quadrature amplitudes, with the local oscillator phase selecting the measured quadrature. Rotation of the squeezing ellipse relative to the detection basis degrades the observed squeezing.

The degree of squeezing is measured in decibels relative to the SQL, with higher values indicating stronger quantum noise reduction. The best laboratory demonstration remains 15 dB of directly measured squeezed vacuum at 1064 nm, reported in 2016, corresponding to roughly a thirtyfold reduction in noise power. Deployed systems typically work with 6-10 dB because optical loss mixes squeezed vacuum with ordinary vacuum and erodes the quantum correlations.

Optical Parametric Oscillators

Optical parametric oscillators (OPOs) below threshold generate squeezed vacuum through parametric amplification of quantum fluctuations. A pump beam drives a nonlinear crystal inside an optical cavity resonant at the signal frequency. Below the oscillation threshold, the parametric gain amplifies one quadrature while deamplifying the orthogonal quadrature, producing squeezed output. The cavity enhancement increases the effective interaction length, enabling strong squeezing with moderate pump power.

Continuous-wave OPO sources based on periodically poled crystals achieve the highest demonstrated squeezing levels. PPKTP crystals in bow-tie or linear cavity configurations produce squeezing exceeding 10 dB at stable operating points. The cavity must be actively stabilized to maintain resonance and optimal squeezing angle. Crystal temperature, cavity length, and pump power all require careful control for reproducible operation.

Pulsed squeezed light from synchronously pumped OPOs produces temporally localized squeezed states matched to pulsed detection systems. The squeezing bandwidth extends across the pulse spectrum, enabling high time-bandwidth product squeezed states. Broadband squeezing is valuable for quantum information applications where timing precision is critical.

Waveguide and Integrated Sources

Waveguide squeezed light sources confine the pump and signal modes to achieve high nonlinear interaction efficiency in compact devices. Periodically poled lithium niobate (PPLN) waveguides generate squeezing through single-pass parametric amplification without a cavity. The waveguide geometry enables direct fiber coupling for integrated quantum systems, though single-pass squeezing is typically limited to a few decibels.

Silicon nitride ring resonators produce squeezing through cavity-enhanced four-wave mixing. The high optical quality factor and engineered dispersion enable parametric oscillation thresholds below milliwatt pump powers. The CMOS-compatible platform offers a path toward integrated squeezed light sources, though current demonstrations achieve modest squeezing levels limited by propagation loss.

Integrated squeezers face challenges from waveguide loss, which admits vacuum fluctuations that degrade squeezing. Loss is unforgiving at high squeezing levels: a path with transmission efficiency eta caps the observable squeezing at minus ten times the base-ten logarithm of one minus eta, so ten percent total loss limits the measurement to about 10 dB no matter how much squeezing the crystal generates, and one percent loss caps it near 20 dB. This ceiling, rather than the nonlinear gain, sets the practical target for ultra-low-loss waveguide fabrication, low-loss couplers, and high-quantum-efficiency photodiodes. The combination of source, homodyne detector, and signal processing on a single chip remains an active development goal.

Applications of Squeezed Light

Gravitational wave detectors including LIGO and Virgo inject squeezed vacuum to reduce quantum noise, a practice that has been routine since around 2010 and is the clearest example of squeezed light earning its place in an operating instrument. Quantum noise limits these kilometer-scale interferometers across most of their band, with shot noise dominating at high frequencies and quantum radiation pressure noise on the test masses dominating at low frequencies. Because a fixed squeezing angle improves one at the expense of the other, LIGO added frequency-dependent squeezing for the fourth observing run in 2023, reflecting the squeezed vacuum from a 300-meter detuned filter cavity that rotates the squeezing ellipse below the cavity pole. The result was broadband improvement from tens of hertz to several kilohertz and a detection-range increase of roughly 15 to 18 percent, which translates into a substantially larger observable volume.

Quantum-enhanced sensing uses squeezed light to measure small signals with precision beyond classical limits. Biological microscopy, atomic magnetometry, and precision spectroscopy benefit from reduced measurement noise. The practical advantage depends on the degree of squeezing achievable in the specific experimental geometry and wavelength range.

Continuous-variable quantum information processing encodes information in the quadrature amplitudes of squeezed states. Cluster state computation uses entangled squeezed modes generated by combining squeezers on beam splitter networks. The deterministic state preparation and homodyne measurement of continuous-variable systems offer advantages over single-photon approaches for certain applications, though error correction requirements differ.

Phase Matching Configurations

Phase matching ensures that the pump, signal, and idler photons maintain a fixed phase relationship over the interaction length, enabling coherent buildup of the generated field. Birefringent phase matching uses the polarization-dependent refractive index of uniaxial crystals to achieve momentum conservation. The crystal orientation determines the phase-matched wavelengths and emission angles. Type-I produces co-polarized signal and idler, while type-II produces orthogonally polarized pairs.

Quasi-phase-matching through periodic poling inverts the nonlinear coefficient at intervals matching the coherence length. This approach compensates the phase mismatch accumulated in each domain, enabling any wavelength combination within the crystal transparency range. PPLN, PPKTP, and other periodically poled crystals have become standard for SPDC sources because of their design flexibility and high effective nonlinearity.

Non-collinear phase matching produces signal and idler beams at different angles from the pump. The angle between the beams depends on wavelength, creating spatial correlations that can generate entanglement. Collinear configurations with signal and idler propagating along the pump direction simplify collection but require post-selection or specialized crystal configurations for polarization entanglement.

Bulk Crystal Sources

Bulk crystal SPDC sources use free-space beams propagating through millimeter to centimeter-scale crystals. BBO is commonly used for visible wavelength pair generation due to its high damage threshold and broad phase-matching bandwidth. The relatively short crystal length limits brightness but provides broad spectral bandwidth useful for ultrafast applications.

Sagnac loop configurations with a type-II crystal generate polarization-entangled photon pairs through bidirectional pumping. The pump beam enters the loop and generates pairs propagating in both directions around the loop, which combine at the output with indistinguishable which-path information. This self-stabilizing design achieves high-fidelity entanglement without alignment-sensitive interferometer arms.

Sandwich sources stack multiple thin crystals with alternating orientations to generate pairs in a superposition of emission times, creating time-bin entanglement. The temporal separation between crystals defines the time-bin spacing, with pump coherence length exceeding this separation ensuring indistinguishability. Combinations with polarization entanglement create hyperentangled states.

Waveguide Sources

Waveguide confinement increases the photon pair generation rate by maintaining high intensity over extended interaction lengths. PPLN and PPKTP waveguides achieve brightness orders of magnitude higher than bulk crystals of the same length. The guided mode geometry also provides mode-matching to collection fibers, improving overall system efficiency.

Reverse proton exchange and titanium indiffusion create waveguides in lithium niobate with different mode properties and damage thresholds. Thin-film lithium niobate on insulator (TFLN) provides higher index contrast and smaller mode areas for further brightness enhancement. The single-mode operation of properly designed waveguides eliminates spatial mode cleanup required for bulk sources.

Silicon nitride and silicon waveguides generate pairs through four-wave mixing rather than SPDC, as discussed above. The CMOS compatibility of these platforms enables integration with other photonic components for complete quantum circuits on chip. Ring resonator enhancement provides additional brightness improvement for both FWM and SPDC waveguide sources.

Spectral Engineering

The joint spectral amplitude (JSA) describes the probability amplitude for generating signal and idler photons at specific frequencies. Engineering the JSA controls the spectral correlations between photons, critical for achieving high indistinguishability in heralded sources. A separable JSA with no frequency correlations produces pure heralded photons, while correlated JSA results in mixed states after heralding.

Group velocity matching makes the pump pulse travel at the same speed as the signal and idler, eliminating the timing correlations that cause spectral entanglement. Certain crystal cuts and wavelength combinations naturally satisfy this condition. Alternatively, pump bandwidth can be chosen to match the phase-matching bandwidth for approximate separability.

Spectral filtering narrows the bandwidth of collected photons, improving indistinguishability at the cost of reducing brightness. The filter bandwidth must be matched to the target application: narrow filtering for high-visibility interference, broad filtering for maximum pair rate. Filtering also enables wavelength multiplexing where different spectral channels carry independent photon pairs.

Conversion Principles

Quantum frequency conversion changes the wavelength of single photons while preserving their quantum properties including superposition and entanglement. The process uses sum-frequency generation or difference-frequency generation in nonlinear crystals, with a strong classical pump providing the frequency shift. Conversion between visible wavelengths where quantum emitters operate and telecom wavelengths suitable for fiber transmission enables hybrid quantum networks.

The conversion efficiency depends on pump power, crystal nonlinearity, and phase-matching bandwidth. High efficiency requires long interaction lengths and careful optimization of pump and signal modes. Internal conversion efficiencies exceeding 90 percent have been demonstrated in periodically poled waveguide devices, approaching the ideal of unity conversion, though the device-level figure that reaches a downstream detector is lower once coupling, filtering, and propagation losses are included. The added noise from the conversion process must remain low to preserve single-photon purity.

Noise sources in frequency conversion include spontaneous Raman scattering, spontaneous parametric fluorescence from the pump, and residual pump leakage. Cooling the crystal reduces Raman noise, while spectral filtering rejects fluorescence and pump. The signal-to-noise ratio after conversion determines the fidelity of the quantum state transformation.

Upconversion Detection

Upconversion detection shifts infrared photons to visible wavelengths where efficient silicon avalanche photodiodes can detect them. This approach provides an alternative to less-efficient infrared detectors for telecom-wavelength quantum communication. The conversion happens in a nonlinear crystal pumped by a strong continuous-wave or pulsed laser, with the sum-frequency output detected by a silicon counter.

The effective detection efficiency of an upconversion detector is the product of the conversion efficiency and the efficiency of the visible detector behind it. Overall figures near 30 to 40 percent are typical, which no longer competes with superconducting nanowire detectors on efficiency alone now that those exceed 90 percent at telecom wavelengths. The case for upconversion detection rests instead on operation without a cryostat, on the intrinsic noise rejection that comes from converting only a narrow spectral and temporal window, and on the ability to reuse mature silicon counting electronics.

Time-resolved upconversion enables ultrafast single-photon detection by using a short pump pulse as a temporal gate. Only signal photons present during the pump pulse are converted and detected. This technique has achieved femtosecond timing resolution limited by the pump pulse duration, useful for quantum timing and imaging applications.

Interface Between Different Wavelengths

Quantum network nodes using different physical platforms often operate at incompatible wavelengths. Trapped ions emit at specific atomic transitions in the visible or near-infrared, while long-distance fiber transmission requires telecom wavelengths. Quantum frequency conversion bridges these wavelengths, enabling heterogeneous quantum networks combining the best features of different platforms.

Bidirectional conversion maintains entanglement when both photons of an entangled pair undergo frequency conversion. The conversion must preserve the phase relationship between entangled states, requiring matched conversion processes for both photons. Demonstrations have shown preservation of polarization and time-bin entanglement through frequency conversion.

Bandwidth matching between the quantum emitter and the frequency conversion acceptance bandwidth affects the conversion efficiency. Narrow-linewidth emitters like atoms and ions match well to narrowband conversion processes. Broader quantum dot emission may require filtering or engineered broadband conversion. Optimization of the source-converter interface is essential for high-efficiency quantum interfaces.

Requirements for Determinism

A truly deterministic single-photon source produces exactly one photon in a specified mode with unit probability upon each trigger event. No demonstrated source fully achieves this ideal, but several approaches come close enough for practical applications. The relevant metric is the probability of delivering a useful photon to the next stage of the quantum system, combining source efficiency with collection, transmission, and coupling losses.

Triggered sources based on single emitters are inherently deterministic in the sense that each excitation pulse produces at most one photon. The limitation is the probability of actually collecting that photon, which depends on the extraction efficiency from the emitter and subsequent optical path losses. Cavity enhancement increases extraction efficiency, while waveguide integration improves collection into useful modes.

Heralded sources become quasi-deterministic through multiplexing, where successful heralding events from multiple probabilistic sources are routed to a common output. The overall success probability increases with the number of sources, approaching unity for large arrays with fast switches. The practical limit comes from switch losses and control complexity, requiring optimization of the multiplexing architecture.

Multiplexed Heralded Sources

Spatial multiplexing combines N heralded sources with an N-to-1 switch network that routes successful events to the output. If each source succeeds independently with probability p, at least one succeeds with probability 1-(1-p)^N, which tends to unity as N grows even when p is small. The gain is real only if the switch network is nearly lossless, because every decibel of insertion loss subtracts directly from the delivered photon probability and eventually cancels the multiplexing advantage. Fast optical switches with low loss and high extinction are therefore the enabling component, with electro-optic devices operating at megahertz to gigahertz rates and sub-decibel insertion loss.

Temporal multiplexing stores photons from successful heralding events and releases them in synchronized time bins. Fiber delay loops of different lengths route photons to align their arrival times at the output. This approach uses fewer sources than spatial multiplexing but requires fast switches at both the input and output of the delay network. The storage time is limited by fiber loss and switch speed.

Integrated multiplexed sources on photonic chips offer compact implementations with reduced losses. Ring resonators serve as both pair sources and delay elements. On-chip switches using thermo-optic or electro-optic effects provide the routing function. Current demonstrations achieve modest multiplexing factors, with scaling limited by waveguide propagation loss and switch crosstalk.

Progress Toward Practical Deterministic Sources

Quantum dot sources in optimized photonic structures currently achieve the highest demonstrated efficiency for triggered single-photon emission. Collection efficiency into the first lens exceeding 70% and system efficiency including fiber coupling above 50% have been reported. Combined with near-unity radiative quantum efficiency and high repetition rate operation, these sources approach practical determinism for specific applications.

The main remaining challenges for deterministic sources include matching the emission wavelength to application requirements, achieving simultaneous high efficiency and indistinguishability, and packaging for practical deployment. Quantum dots emitting at telecom wavelengths lag behind shorter-wavelength devices in performance. Operation at elevated temperatures would simplify cooling requirements but typically degrades coherence properties.

Hybrid approaches combining deterministic triggering from quantum emitters with heralding for photon number verification may offer the best practical path. A high-efficiency triggered source provides most photons on demand, while heralding identifies and discards the rare failure events. This combination could achieve near-unity success probability with current technology.

Single-Photon State Tomography

Complete characterization of a single-photon state requires measuring the density matrix elements that describe the quantum state. For polarization qubits, the two-dimensional density matrix has four elements determined by three independent parameters plus normalization. Measurements in three complementary bases (horizontal/vertical, diagonal/antidiagonal, and right/left circular) provide the data needed for tomographic reconstruction.

Maximum likelihood estimation finds the physical density matrix most consistent with the measurement data, enforcing constraints that the matrix be Hermitian, positive semidefinite, and unit trace. This approach handles measurement errors and finite statistics more robustly than direct inversion methods. Confidence intervals from the likelihood function quantify the uncertainty in reconstructed parameters.

Wigner function tomography characterizes the continuous-variable aspects of single-photon states, including the temporal wavepacket shape and spectral distribution. Homodyne detection with variable local oscillator phase measures the quadrature distributions from which the Wigner function is reconstructed. The characteristic interference fringes and negative regions of single-photon Wigner functions confirm the quantum nature of the state.

Two-Photon Tomography

Two-photon state tomography reconstructs the four-by-four density matrix describing the joint state of a photon pair. The 16 matrix elements require 15 independent parameters plus normalization, demanding more extensive measurements than single-photon tomography. Standard protocols use 36 measurement settings combining six basis states for each photon, with overcomplete data improving accuracy.

Entanglement measures extracted from the reconstructed density matrix include fidelity with respect to target Bell states, concurrence, and entanglement of formation. Fidelity above 0.5 with any Bell state indicates entanglement, with higher values demonstrating stronger entanglement useful for quantum information applications. These measures provide benchmarks for comparing different entangled photon sources.

Systematic errors in tomography arise from imperfect measurement bases, unequal detection efficiencies, and timing misalignment. Careful calibration of waveplates and polarizers defines accurate measurement bases. Efficiency corrections remove bias from unequal detection rates. Coincidence window optimization balances between including all correlated pairs and excluding accidental coincidences.

Advanced Characterization Methods

Process tomography characterizes quantum operations such as frequency conversion or quantum gates by determining how the process transforms an informationally complete set of input states. The reconstructed process matrix describes the operation's effect on any input state, enabling prediction of performance for intended applications. For optical processes, this involves preparing various input polarization states and measuring the output distribution.

Self-referencing tomography uses the source itself to provide reference states, eliminating the need for separately characterized inputs. For entangled photon sources, measurements on one photon of a pair herald known states of the other photon. This bootstrapping approach reduces calibration requirements and systematic errors from reference state preparation.

Compressed sensing techniques enable tomography with fewer measurements by exploiting the low rank of nearly pure quantum states. Rather than measuring all basis combinations, strategically chosen measurements provide the information needed to reconstruct low-rank density matrices. This approach reduces measurement time for sources producing high-quality quantum states.

Photon Counting and Correlation Measurements

Single-photon counting modules based on silicon avalanche photodiodes provide efficient, low-noise detection for visible and near-infrared wavelengths, with peak detection efficiencies around 65 to 70 percent and dark counts of a few tens of events per second. The detector specifications of importance include quantum efficiency, dark count rate, timing jitter, and afterpulsing probability. Dead time after each detection limits the maximum count rate and distorts coincidence measurements if it is not accounted for. Superconducting nanowire single-photon detectors dominate at telecom wavelengths, reaching system detection efficiencies above 90 percent at 1550 nm with timing jitter of a few tens of picoseconds, and below five picoseconds in devices optimized for timing. Their cost is the cryostat: operation near one to three kelvin.

Time-correlated single-photon counting (TCSPC) electronics record the arrival time of each photon detection relative to a synchronization signal. The timing resolution, typically tens of picoseconds, must exceed the source correlation timescale for accurate g2 measurements. Histogram accumulation over many events builds up the coincidence distribution, with total measurement time determining statistical precision.

Background subtraction removes contributions from detector dark counts and uncorrelated light. Measuring coincidences with large time offsets provides the accidental rate that would exist without source correlations. Subtracting this background from the raw coincidence data yields the true source correlation function. Proper background treatment is essential for accurate determination of g2(0) and other correlation parameters.

Brightness and Efficiency Measurements

Source brightness is commonly specified as detected singles rate, coincidence rate, or pair generation rate per milliwatt of pump power. The relationship between these quantities depends on detection efficiency, collection efficiency, and losses in the optical path. Inferring the intrinsic source properties requires careful accounting of all system efficiencies.

Klyshko efficiency provides a self-referencing measure of collection and detection efficiency using correlated photon pairs. Dividing the coincidence rate by the singles rate of the heralding arm gives the total efficiency of the other arm, including its optics, filters, fiber coupling, and detector. The measurement needs no externally calibrated reference, because every heralded event is known to have launched a partner photon into the signal path; the source acts as its own photon standard. Klyshko efficiency is therefore the honest figure to quote when comparing heralded sources, since it cannot be inflated by hidden losses.

Spectral brightness normalizes the generation rate to emission bandwidth, enabling comparison between sources with different linewidths. This metric is particularly relevant for applications requiring narrow bandwidth, where broad sources must be filtered with corresponding brightness reduction. The appropriate spectral brightness specification depends on the intended application bandwidth.

Indistinguishability Measurements

Hong-Ou-Mandel measurements between consecutive photons from the same source characterize the temporal stability and coherence of the emission process. The photons are delayed by one repetition period and interfered on a beam splitter. High visibility indicates that photons emitted at different times are quantum mechanically identical, essential for applications requiring multiphoton interference.

Two-source Hong-Ou-Mandel interference tests the compatibility of independent sources for networked quantum information processing. Two separate single-photon emitters can in principle reach unit visibility, and independent heralded sources have exceeded 90 percent, but doing so requires tuning the emitters into spectral resonance, matching their temporal wavepackets, and stabilizing them against drift. The 50 percent ceiling applies when the interfering inputs are attenuated laser pulses rather than single-photon states, as in measurement-device-independent quantum key distribution, and should not be used as the benchmark for a true single-photon source.

Spectral and temporal measurements provide complementary information about factors limiting indistinguishability. Spectrometers measure the emission linewidth and any spectral diffusion or wandering. Streak cameras and time-correlated counting reveal the temporal emission profile and timing jitter. Correlating these measurements with interference visibility identifies the dominant sources of distinguishability for targeted improvement.