Itinerant Single Microwave Photons

• Itinerant single microwave photons are propagating quantum excitations in transmission lines, serving as flying qubits in superconducting circuits.
• The article reviews generation techniques like qubit–cavity swap protocols and engineered emission methods to achieve high fidelity and tailored photon shapes.
• Detection strategies span destructive methods to QND and number-resolving schemes, delivering efficiencies from 58% to over 96% for robust quantum network applications.

An itinerant single microwave photon is defined as a single-photon excitation propagating in a transmission line or waveguide mode, as opposed to being confined to a localized cavity. The ability to generate, manipulate, and detect such photons is fundamental for advances in quantum communication, microwave quantum optics, and distributed quantum information processing using superconducting circuit platforms. This article systematically reviews the physics, implementation strategies, detection methodologies, performance limits, and applications of itinerant single microwave photons, drawing on state-of-the-art theoretical and experimental results.

1. Fundamental Principles and Definitions

An itinerant microwave photon occupies a propagating spacetime mode—typically a quantized pulse with envelope $f(t)$—in a 1D waveguide or open transmission line. In the quantum input–output formalism, the annihilation operator for such a mode is

$A = \int dt\,f(t)\,a_{\rm out}(t),$

where $a_{\rm out}(t)$ is the Heisenberg output field operator. The mode can be engineered to have high temporal and spectral purity, and is readily interfaced with stationary superconducting qubits or parametric amplifiers (Eichler et al., 2010, Kindel et al., 2015).

Crucially, these photons can be manipulated on chip, routed between distant modules, used to probe quantum devices, or serve as “flying qubits” in a quantum network (Reuer et al., 2021). Their quantum state can be reconstructed through homodyne/heterodyne tomography (Eichler et al., 2010), or detected directly via number-resolving or quantum non-demolition (QND) schemes (Kono et al., 2017, Royer et al., 2017).

2. Generation of Itinerant Microwave Photons

High-fidelity single-photon sources are necessary for controlled experiments with itinerant microwave quanta. Predominant methods in circuit quantum electrodynamics (cQED) include:

• Qubit–Cavity Swap Protocols: A superconducting transmon qubit is coupled to a high-Q cavity. A sideband (e.g., blue-sideband) pulse induces a $\ket{g,0}\leftrightarrow\ket{e,1}$ transition, after which the cavity emits the photon into the waveguide with controlled temporal shape. Input–output relations ensure the external field inherits the internal state with efficiency set by the external coupling $\kappa$ (Kindel et al., 2015, Eichler et al., 2010, Werner et al., 31 Jan 2026).
• Shape and Frequency Control: By modulating the effective emission rate $\Gamma_f(t)$ or the external drive amplitude/frequency, the frequency and envelope $f(t)$ of the photon can be tuned, enabling time-symmetric (e.g., $\mathrm{sech}$), Gaussian, or other tailored shapes (Miyamura et al., 7 Mar 2025). Up to 40 MHz frequency tunability with $>95\%$ fidelity is demonstrated.
• Directional Emission and Absorption: Combining two or more qubits with quarter-wavelength separation and engineered couplings allows dynamical selection of emission or absorption direction via destructive interference, necessary for fully reconfigurable quantum networks (Gheeraert et al., 2020).
• Photon Blockade Sources: Strong nonlinearity in a cavity–qubit system (Jaynes–Cummings Hamiltonian, $g \gg \kappa$) facilitates photon blockade, producing trains of anti-bunched photons, as evidenced by $g^{(2)}(0)\ll1$ in correlation measurements (Lang et al., 2011).

3. Detection Methodologies

Detection of itinerant microwave photons can broadly be categorized into destructive, QND, and number-resolving approaches.

3.1. Destructive Detection

• Current-Biased Josephson Junctions (CBJJ): A phase qubit is excited by the photon from $|0\rangle$ to $|1\rangle$, which then tunnels into the continuum, producing a voltage pulse (“click”)—a fundamentally absorbing process (Chen et al., 2010).
• Photo-Assisted Quasiparticle Tunneling: Incoming photons convert into quasiparticles which poison a superconducting island, sensed in real time with single-Cooper-pair-transistor readout. This approach yields 10% efficiency, $<50$ ns timing resolution, and $\sim1\,\mu$s dead time (Basset et al., 21 Nov 2025).
• Engineered Nonlinear Dissipation: By coupling a two-level system (transmon) and a resonator through a three-wave-mixing process (parametric pump), a single photon in the buffer mode triggers an irreversible transition $|g\rangle\to|e\rangle$ in the qubit. The process is dissipatively induced, insensitive to photon shape, and achieves a measured efficiency of 58% and dark-count rate of 1.4 ms$^{-1}$ (Lescanne et al., 2019).

3.2. Quantum Non-Demolition (QND) and Non-Absorptive Techniques

• Dispersive QND Detection: An incoming photon reflects off a cavity containing a dispersively coupled qubit, imparting a conditional phase shift. Ramsey-style pulse sequences on the qubit allow QND detection of the photon without destruction, achieving $\eta\simeq0.84$ detection efficiency with photon survival probability $s\simeq0.87$ (Kono et al., 2017, Besse et al., 2017).
• Dispersive Detection Using Quantum Dots: A dispersively coupled double quantum dot detects photon number via a conditional shift in charge levels, converted to a measurable tunnel event, while only producing cavity dephasing as back-action. Efficiencies of 70–90% are achievable with appropriate tunneling rates and sub-100 mK temperatures (Matern et al., 24 Nov 2025).
• Dark-State Engineering: Using a small number ($N=3$–$5$) of inhomogeneously detuned artificial atoms coupled to a waveguide, a photon absorbed into the super-radiant “bright” mode is coherently transferred to long-lived “dark” states, passively increasing interaction time for high-fidelity, continuous, QND detection. Achievable detection fidelities $F\approx96\%$ are reported (Royer et al., 2017).
• Cross-Kerr/Dispersive Readout and Cascades: A transmon-QND detector realizes a conditional probe-cavity frequency shift, observed via homodyne detection; the signal-to-noise ratio scales as $\sqrt{N}$ in cascaded devices, with distinguishability up to 90% for two units and 95% for four units (Fan et al., 2014, Sathyamoorthy et al., 2015).

3.3. Photon Multiplication and Number-Resolving Detectors

• Josephson-Photon Multiplication (JPM): Inelastic Cooper-pair tunneling in a voltage-biased Josephson junction converts single input photons into $n$-photon bursts at a higher frequency (e.g., $|1\rangle_a\to|n\rangle_b$), which are then detected in the classical regime with linear amplifiers (Albert et al., 2023, Danner et al., 9 Oct 2025). Cascading two multiplication stages ($n=4$ per stage, $m=16$ output photons per input) supports 84.5% detection probability at a dark count rate of $10^{-3}/T$, with near-zero dead time.

Table: Comparison of Detection Strategies

Detection Method	Efficiency	Dark Count Rate	QND	Dead Time
CBJJ (current-biased Josephson)	$\sim$70%	$\sim 10^6$–$10^7$/s	No	5–10 ns
Dark-state trapping (Royer et al.)	92–96%	$4.2\times 10^{-3}$/μs	Yes	Negligible
Ramsey interferometer QND (Kono/Besse)	71–84%	0.015 per trial	Yes	100 ns
Engineered nonlinear dissipation	58%	1.4/ms	No	$\mu$s–high
Photon multiplication (JPM, $\times$16)	84.5%	$10^{-3}/T$	No	None (cont.)
Dispersive QD (theory)	70–90%	(thermally limited)	Yes	Continuous

All entries refer to empirical implementations where available; theoretical “QND” means non-absorptive and projective onto the photon number basis.

4. Tomographic State Reconstruction and Characterization

Full characterization of itinerant single-photon states is achieved via quadrature detection and statistical analysis:

• Heterodyne/Phase-sensitive Amplification: Output fields are amplified (HEMT or Josephson parametric amplifiers), mixed down to $X$ and $P$ quadratures, and digitized at nanosecond resolution (Eichler et al., 2010, Kindel et al., 2015).
• Moment Analysis: Extracting normally ordered moments up to 4th order separates the signal from added noise, yielding the density matrix and allowing Wigner function reconstruction. State fidelities in excess of 90% for single-photon Fock states and 99.7% for superposition states have been demonstrated (Werner et al., 31 Jan 2026).
• Backaction and Detector-limited Noise: Amplifier backaction can be modeled (e.g., as an effective cavity thermal occupancy $n_\mathrm{back}$) and minimized through device isolation and pump-scheme engineering (Kindel et al., 2015).

5. Performance Metrics and Limitations

Critical figures of merit for itinerant photon detection are:

• Detection Efficiency ($\eta$): Probability of registering a “click” given a single incident photon. For optimal QND detectors, values up to 96% are achieved (Royer et al., 2017); photon-multiplier based schemes reach 84.5% (Danner et al., 9 Oct 2025); earlier approaches report 70% (Chen et al., 2010, Kono et al., 2017, Besse et al., 2017).
• Dark Count Rate: Rate of spurious detector “clicks” in absence of signal. Dark-state modes and improvement in threshold discrimination tradeoff between $\eta$ and dark count rate (e.g., raising threshold reduces dark counts but can decrease efficiency).
• Bandwidth: Fast detection requires matching detector response to photon spectral width. Multipliers and JPM devices achieve $>100$ MHz, while QND and dispersive detectors are typically limited to 10–20 MHz by cavity and transmon coherence (Albert et al., 2023, Koshino et al., 2015).
• Dead Time: The interval between successive valid detection events. Multiplication-based schemes and true QND detectors can operate with no dead time, while avalanche or single-shot approaches are limited by reset cycles ($\mu$s scale) (Basset et al., 21 Nov 2025, Lescanne et al., 2019).
• Backaction and QND Character: For QND schemes, photon number is measured without permanent absorption, preserving coherence when initialized in a Fock state but projecting superposed states to the number basis (Kono et al., 2017, Sathyamoorthy et al., 2015, Fan et al., 2014).
• Scalability and Parallelization: Arrays of Josephson-junction detectors or cascaded QND modules allow number resolution and higher composite detection probability (Chen et al., 2010, Kindel et al., 2015, Albert et al., 2023).

6. Applications and Implementations

Itinerant microwave photons serve as fundamental carriers of information in quantum technologies:

• Quantum Networking and Communication: Flying qubits transmit quantum states between distant nodes, enabling modular superconducting architectures (Reuer et al., 2021). Capture, emission, and deterministic photon-photon gates have been realized with single-photon and two-photon fidelities of 75% and 57%, respectively.
• Heralded Entanglement and Quantum Repeaters: QND detection enables heralded remote entanglement and error signaling; post-conversion fidelities above classical thresholds ($F\simeq 87.6\%$ for optical upconversion) make multi-node quantum communication feasible (Werner et al., 31 Jan 2026).
• Quantum Optics and Sensing: State tomography, antibunching, and correlation functions measured with itinerant photons have provided microwave demonstration of quantum optical effects previously accessible only in the optical regime (Eichler et al., 2010, Lang et al., 2011).
• Microwave-to-Optical State Conversion: On-demand microwave photon generation and upconversion to the telecom band with $<0.012$ quanta added noise and SNR up to 5.1 have been demonstrated, bridging cryogenic superconducting systems to room-temperature photonic links (Werner et al., 31 Jan 2026).
• Continuous Quantum Measurement and Sensing: Continuous and rapid event detection (timing $<50$ ns, dead time $1\,\mu$s) now supports applications in time-correlated photon counting and dynamic quantum feedback (Basset et al., 21 Nov 2025).

7. Frontiers, Trade-offs, and Outlook

The central technical axes separating available detection strategies are:

• QND vs. Absorptive: True QND operation is essential for protocols relying on non-demolition measurements and complex multi-photon quantum logic (Royer et al., 2017, Kono et al., 2017, Matern et al., 24 Nov 2025, Koshino et al., 2015).
• Parallelizability and Number Resolution: Photon-multiplication and CBJJ arrays provide scalable number-resolving capabilities but tend to saturate, while cascaded QND units offer higher fidelity and projectivity (Albert et al., 2023, Danner et al., 9 Oct 2025).
• Bandwidth/Timeliness vs. Efficiency: Broader bandwidth in photon-multiplier or photodiode architectures offers faster response but sometimes sacrifices single-photon sensitivity or QND character (Basset et al., 21 Nov 2025, Albert et al., 2023, Danner et al., 9 Oct 2025).
• Dark State Engineering: Hybrid architectures exploiting long-lived dark states circumvent the backaction bandwidth trade-off, passively trapping excitations and supporting continuous high-fidelity QND monitoring at minimal cost in topology (Royer et al., 2017).
• Photon-Number Dephasing and Coherence: Fundamental trade-offs between measurement-induced dephasing and detection SNR dictate the ultimate performance envelope of non-absorbing detectors (Fan et al., 2014, Sathyamoorthy et al., 2015).

The increasing sophistication of device architectures, leveraging nonlinear quantum optics, tailored dissipation, quantum-limited amplification, and robust tomography, is rapidly closing the gap between microwave and optical single-photon science. The ongoing development of time-resolved, QND, number-resolving, and high-bandwidth microwave photon detectors will further empower circuit-QED platforms and heterogeneous quantum technologies with single-quantum precision.