Quantum-Optical Induced-Coherence Tomography

• Quantum-Optical Induced-Coherence Tomography is a quantum imaging modality that uses SPDC to create correlated photon pairs for depth-resolved sensing.
• The method employs a hybrid interferometer with time- and frequency-domain acquisition to infer sample properties from interference in the detected signal photon.
• Experimental benchmarks include heralding efficiencies up to 63% and an axial resolution of around 0.194 mm, enabling precise, non-invasive imaging of multilayered structures.

Quantum-Optical Induced-Coherence Tomography (QICT) is a quantum imaging modality exploiting induced coherence in spontaneous parametric down-conversion (SPDC) to perform depth-resolved sensing with undetected photons. This approach leverages quantum correlations between signal and idler photon pairs: one photon (signal) is detected after traversing a visible-wavelength path, while its twin (idler) traverses or reflects from the sample at a distinct, often infrared, wavelength but remains undetected. Changes in the optical properties encountered by the idler are inferred by monitoring the interference in the detected signal photons. QICT enables non-invasive probing, precise tomography of multilayered structures, and operates in regimes inaccessible or impractical for direct detection, such as mid-infrared imaging. Hybrid interferometer architectures now permit parallel optimization of both arms and integration of time- and frequency-domain acquisition.

1. Foundational Principle: Induced-Coherence Quantum Interferometry

In QICT, two indistinguishable SPDC pathways are coherently pumped so that each can generate signal ($s$) and idler ($i$) photons as quantum correlated pairs. The foundation is that only the signal photon is detected, typically at a wavelength (e.g., 810 nm) well suited to efficient detection, while the idler propagates through the sample under test—possibly at a wavelength (e.g., 1550 nm) that is challenging for detection but advantageous for the application. Interference is observed in the signal arm if the two possible idler (probe) paths are made indistinguishable. Perturbations (phase shifts, delays, or reflections) imposed on the idler arm by the sample are mapped onto the interference fringes seen in the signal detection rate, enabling tomographic profiling with the idler never being directly detected (Kim et al., 2023).

The two SPDC processes, one forward, one backward through the nonlinear medium, produce (to first order) the quantum state

$$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$

where $s_n^\dagger, i_n^\dagger$ are the creation operators for signal and idler photons in each direction, and $C_{1,2}, p_{1,2}$ parameterize amplitudes and losses. The interference term is maximal when the idler paths are indistinguishable and is modulated by any phase imprinted by the sample.

2. Hybrid Interferometer Architecture and Detection Strategy

QICT as realized in Kim et al. (Kim et al., 2023) features a hybrid interferometer comprised of:

• A Mach-Zehnder–type arrangement for signal photons (visible, ~810 nm).
• A Michelson-type path for idler photons (infrared, ~1550 nm), where the sample is inserted in one arm.
• Double-pass SPDC in a single 5 mm periodically poled lithium niobate (PPLN) crystal, with a retro-reflecting mirror returning both pump and idler fields for backward generation and probing.

Signal photons traverse a variable-delay line (Δτ_s) before recombination, while the idler, after reflection by the sample, is recombined by the same retro-mirror. Only the output port for the signal is detected—no detection occurs in the idler arm. Spatial and spectral mode overlap is engineered to achieve heralding efficiencies up to 63% (signal-to-idler) and 49% (idler-to-signal), with the system monitored for interference visibility approaching these heralding limits (Kim et al., 2023).

3. Theoretical Formalism: Interference Contrast, Depth Encoding, and Domain Transformations

The singles count rate in the detected signal expresses as:

$$R_s \propto |C_1|^2|p_1|^2 + |C_2|^2|p_2|^2 + 2|C_1 C_2 p_1 p_2| \sqrt{\eta_s \eta_i} \cos(\Delta\phi_s + \Delta\phi_0)$$

where

$$\eta_{s\to i}^{(n)} = \frac{|p_n|^2}{|p_n|^2 + |q_n|^2}, \ \eta_{i\to s}^{(n)} = \frac{|p_n|^2}{|p_n|^2 + |r_n|^2}$$

are heralding efficiencies per path, and $\eta_s, \eta_i$ aggregate all loss processes. The maximal interference visibility $V$ is given in the symmetric case by:

$$V = \sqrt{\eta_s \eta_i} \approx \eta_s \approx \eta_i$$

Phase shifts or delays induced by the sample in the idler arm (complex reflection $i$0) map directly onto a phase shift in the observed signal photon fringes.

Depth information is extracted in two principal modes:

• Time-domain QICT: Scanning the signal delay (Δz_s) yields bursts of interference when the path length matches the roundtrip to a specific reflecting interface in the sample.
• Frequency-domain QICT: Fixing the delay and spectrally resolving the signal yields interferograms whose Fourier transforms present peaks at optical path lengths corresponding to interfaces, with amplitudes set by reflectivity $i$1.

4. Experimental Implementation and Performance Benchmarks

Kim et al. (Kim et al., 2023) achieved hybrid QICT with the following experimental parameters and outcomes:

• Heralding Efficiencies: Measured $i$2–$i$3, $i$4–$i$5.
• Visibility: Interference visibility $i$6, in line with theoretical prediction, verified by scanning a piezo-stage.
• Axial (Depth) Resolution: Determined by the coherence length of the signal photons:

$i$7

for $i$8.

• Transverse Resolution: USAF target imaging behind a 440 μm Si wafer produced lateral features of $i$9 (x) and $$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$0 (y), limited by idler spot and fiber-mode overlap.
• Tomographic Accuracy: Demonstrated for three- and four-layer samples. Extracted thicknesses in TD and FD QICT agreed with ground truth within 1–8%.
• Sensitivity and SNR: Fringe amplitude decays with delay per the signal coherence envelope; SNR scales as $$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$1 up to 1 s integration (shot-noise-limited regime).

5. Time- and Frequency-Domain QICT Methodologies

Time-Domain QICT

A motorized linear stage scans the signal arm delay $$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$2, producing clear interference burst envelopes when path-matching occurs with reflectors in the sample. Each main burst maps unambiguously to a sample interface. The depth resolution is limited by the spectral bandwidth ($$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$3) of the SPDC process. Calibration of zero delay is cross-referenced to the frequency-domain envelope center.

Frequency-Domain QICT

With fixed delay, a tunable bandpass or a single-photon spectrometer resolves the signal over the SPDC bandwidth. The measured spectrum $$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$4 encodes all interferometric path delays; a numerical Fourier transform yields peaks located at the optical thickness of each interface. The magnitude of the Fourier coefficients corresponds to interface reflectivity. Frequency-domain acquisition enables parallel depth mapping, limited mainly by spectrometer resolution.

6. Artifacts, Model-Based Correction, and Comparative Approaches

QICT and related quantum optical coherence tomography (QOCT) techniques can exhibit artifacts and echo signals resulting from multiple internal reflections and cross-interference effects. In the HOM-based QOCT paradigm, as analyzed in (Li-Gomez et al., 2022), multilayer samples produce not only primary dips at true interfaces but also virtual dips and peaks (“echoes,” “artifacts”) due to cross terms in the interferogram. The phase and amplitude of these artifacts depend on both geometry and pump properties. Fast genetic-algorithm post-processing, encoding candidate morphologies and reflectivities, minimizes the mean absolute error between measured and modeled traces over multiple pump wavelengths, enabling discrimination of true interfaces from artifacts (Li-Gomez et al., 2022). This multi-wavelength interrogation further leverages phase inversion properties to identify real features, marking a major advance for routine multilayer metrology.

Even-order dispersion, which limits classical low-coherence OCT, is canceled in QOCT due to symmetric energy conservation in SPDC pairs. Axial resolution is thus doubled compared to classical OCT for equal bandwidth, and artifacts persist only from cross-interference and echoes (Li-Gomez et al., 2022).

7. Applications, Extensibility, and Limitations

QICT enables spectroscopic depth profiling and 3D imaging in scenarios otherwise impractical for conventional detection, including:

• Biomedical imaging at mid-infrared wavelengths (biochemical specificity, stealth imaging).
• Industrial non-destructive probing of multilayered, photo-degradable, or optically opaque samples.
• Remote sensing and gas detection by adjusting SPDC parameters for target absorption bands (Kim et al., 2023).

Potential improvements include broader SPDC bandwidths (shorter crystals, aperiodic poling) for enhanced axial resolution, integration of SPAD arrays for rapid full-band acquisition, and extension to new wavelengths through crystal engineering. However, practical performance is limited by heralding efficiencies below unity (limiting SNR and contrast), the absolute calibration of optical delays (dependent on refractive index precision), and transverse (lateral) resolution set by idler beam focusing and mode-matching optics. Acquisition speed remains constrained by photon flux and detection statistics.

Summary Table: Principal Performance Figures (Kim et al. (Kim et al., 2023))

Metric	Value/Range	Determining Factor
Heralding Efficiency	43–63%	Mode overlap, losses
Fringe Visibility	~64%	$$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$5
Depth Resolution (axial)	0.194 mm	Signal photon bandwidth
Transverse Resolution	11–13 $$|\Psi\rangle = C_1 (p_1 s_1^\dagger i_1^\dagger + \ldots) |0\rangle + C_2 e^{i\phi} (p_2 s_2^\dagger i_2^\dagger + \ldots) |0\rangle$$6m	Idler spot, coupling
SNR Limiting Factor	Shot noise	Integration time; losses

QICT provides a shot-noise limited, high-resolution, and versatile approach to depth-resolved sensing using the principle of induced coherence, extending quantum metrology to undetected-photon regimes and permitting unique applications in scientific and industrial domains (Kim et al., 2023, Li-Gomez et al., 2022).