Transverse-Momentum Resolved Two-Photon Interferometry

• Transverse-momentum-resolved two-photon interferometry is an advanced measurement technique that encodes spatial displacement as oscillatory phase information in the momentum domain.
• It employs beam-splitter geometries and far-field detection to convert transverse displacements into observable quantum beats, enhancing precision in quantum optics and heavy-ion collision studies.
• By resolving momentum correlations, the technique extends the effective range of traditional interferometry, maintaining sensitivity despite spatial mismatch and partial photon distinguishability.

Searching arXiv for recent and foundational papers on transverse-momentum-resolved two-photon interferometry and closely related heavy-ion and HOM/spatial-interference work. Searching arXiv for recent and foundational papers on transverse-momentum-resolved two-photon interferometry and closely related heavy-ion and HOM/spatial-interference work. Transverse-momentum-resolved two-photon interferometry denotes a family of measurements in which two-photon coherence, or closely related photon-induced alternative amplitudes, is read out through transverse momentum variables rather than through an unresolved coincidence dip or an angle-integrated yield. In quantum-optical implementations, the canonical setting is a generalized Hong–Ou–Mandel interferometer in which a balanced beam splitter is followed by far-field detection that resolves the transverse momentum difference of the two output photons, converting a transverse displacement into momentum-space quantum beats (Triggiani et al., 2023). In relativistic heavy-ion physics, the same phrase overlaps with two distinct but adjacent domains: genuine $\gamma\gamma$-initiated pair production, where the measured pair transverse momentum probes the incoming photon transverse-momentum distributions, and photon-induced two-source interference in coherent vector-meson photoproduction, where the observed pattern is first-order interference between alternative production amplitudes rather than two-photon fusion (Collaboration et al., 2019, 0812.1063). The literature therefore uses a common momentum-space language across experimentally and conceptually different interferometric regimes.

1. Conceptual scope and formal distinctions

The subject is not exhausted by a single observable or a single notion of coherence. In the spatial HOM formulation, the resolved two-photon outcome is written in terms of the transverse-momentum difference $\Delta k$ and the output class $X\in\{A,B\}$, with

$$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$

where $\alpha(A)=-1$, $\alpha(B)=+1$, $C(\Delta k)$ is the momentum-difference envelope, and $\nu$ encodes non-spatial indistinguishability (Triggiani et al., 2023). Here the interference is fourth-order: it appears in joint two-photon detection statistics after a balanced beam splitter.

That structure differs sharply from the heavy-ion two-source interferometer realized in coherent $\rho^0$ photoproduction. There the relevant object is a coherent sum of two production amplitudes associated with the two nuclei,

$$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$

which at midrapidity reduces to

$\Delta k$0

The measured suppression at low $\Delta k$1 is therefore first-order interference between two indistinguishable production paths, not HOM bunching or antibunching (0812.1063).

A third formalism is direct-photon Hanbury Brown–Twiss interferometry, where the observable is the normalized two-photon correlation function

$\Delta k$2

This is second-order intensity interferometry, expressed in relative momentum $\Delta k$3 and pair momentum $\Delta k$4, and used to infer the spacetime structure of the photon-emitting source (Garcia-Montero et al., 2019).

A strict encyclopedia treatment therefore requires a distinction among three categories. First, there is genuine two-photon interference in the quantum-optics sense, based on fourth-order correlations. Second, there is photon-induced two-source interference in heavy-ion collisions, which is not $\Delta k$5 and not HBT. Third, there are genuine $\Delta k$6 processes in heavy ions whose pair $\Delta k$7 and azimuthal structure are transverse-momentum resolved but do not by themselves constitute a full interferometric reconstruction. Much of the literature’s terminological ambiguity arises from the fact that all three are organized by Fourier-conjugate transverse variables and by coherence vs distinguishability.

2. Generalized HOM interferometers in the transverse-momentum basis

The most direct realization of transverse-momentum-resolved two-photon interferometry is the beam-splitter geometry analyzed for two displaced single-photon wavepackets. Two quasi-monochromatic photons enter a $\Delta k$8 beam splitter, one in each input port, with transverse wavefunctions $\Delta k$9 and displacement $X\in\{A,B\}$0. Far-field cameras measure transverse momenta $X\in\{A,B\}$1 and $X\in\{A,B\}$2, or equivalently their difference $X\in\{A,B\}$3. The resolved probabilities are

$X\in\{A,B\}$4

$X\in\{A,B\}$5

and, after integrating over mean momentum $X\in\{A,B\}$6,

$X\in\{A,B\}$7

The displacement is thus encoded as the oscillation period of a momentum-difference interferogram rather than as a direct image-plane shift (Triggiani et al., 2023).

The metrological significance of this transformation is that the resolved measurement is optimal for $X\in\{A,B\}$8. The generator of displacement is

$X\in\{A,B\}$9

and the quantum Fisher information is

$$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$0

For identical photons, the classical Fisher information of the resolved measurement satisfies

$$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$1

so the measurement saturates the quantum bound for all $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$2. The asymptotic variance is

$$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$3

The same framework shows why ordinary unresolved HOM is only locally adequate: for $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$4, bucket detection can remain optimal, but resolving $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$5 preserves optimality beyond the overlap regime (Triggiani et al., 2023).

Partial distinguishability does not destroy the resolved protocol in the same way it destroys standard unresolved HOM visibility. For $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$6, the Fisher information obeys the large-separation asymptote

$$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$7

so the sensitivity is reduced only by a constant factor rather than collapsing with increasing spatial separation. This is the core formal reason that transverse-momentum resolution extends the operative domain of HOM-style sensing.

A direct experimental realization of the spatial HOM version was reported for transverse-deflection estimation. In that experiment, type-II SPDC photons filtered by single-mode fibers interfered at a balanced beam splitter, a small relative deflection $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$8 was introduced by rotating one input collimator, and far-field momentum sampling was implemented by a movable slit of width $$P_\nu(\Delta k,X)=\frac{1}{2}C(\Delta k)\left(1+\alpha(X)\nu\cos(\Delta k\,\Delta x)\right),$$9 together with APD coincidence detection. The coincidence law was

$\alpha(A)=-1$0

with $\alpha(A)=-1$1. In the ideal case, the classical and quantum Fisher informations coincide,

$\alpha(A)=-1$2

Experimentally, the fitted momentum width was $\alpha(A)=-1$3, the visibility was $\alpha(A)=-1$4, and the discussion states that a deflection sensitivity of $\alpha(A)=-1$5 can be obtained for $\alpha(A)=-1$6 detection events (Wang et al., 6 Apr 2025).

3. Independent photons, momentum entanglement, and conjugate-basis recovery

The resolved-HOM paradigm was extended experimentally to independent weak coherent states and SPAD-array detection. In that setting, a pulsed laser at $\alpha(A)=-1$7 was attenuated into the weak-photon regime, split into two arms, phase-randomized to suppress $\alpha(A)=-1$8, and recombined on a balanced beam splitter. Far-field detection with a linear 8-pixel SPAD array gave a discrete version of the resolved two-photon law,

$\alpha(A)=-1$9

and, with finite pixel width $\alpha(B)=+1$0,

$\alpha(B)=+1$1

The reported fitted values were approximately $\alpha(B)=+1$2, $\alpha(B)=+1$3, and $\alpha(B)=+1$4. Crucially, oscillatory coincidence beats remained visible even when $\alpha(B)=+1$5, where conventional unresolved HOM would lose visibility (Sgobba et al., 27 Aug 2025).

A complementary route uses momentum-entangled photon pairs. For the state

$\alpha(B)=+1$6

an unknown relative displacement $\alpha(B)=+1$7 appears as a phase in the momentum basis, and the resolved probabilities after polarization mixing become

$\alpha(B)=+1$8

The quantum Fisher information is

$\alpha(B)=+1$9

and the classical Fisher information of the resolving measurement is

$C(\Delta k)$0

The precision therefore improves with the momentum separation $C(\Delta k)$1, and remains independent of the value of $C(\Delta k)$2 itself (Triggiani et al., 2024).

A further development showed the converse Fourier-dual principle: if the photons are deliberately made distinguishable in transverse momentum, quantum interference can be recovered by resolving the conjugate variable, namely transverse position in the near field. For Gaussian modes with momentum offset $C(\Delta k)$3 and displacement $C(\Delta k)$4, the resolved coincidence map at the beam-splitter outputs is

$C(\Delta k)$5

When $C(\Delta k)$6, the interference term is $C(\Delta k)$7, so the fringe period in the relative coordinate is $C(\Delta k)$8. Bucket detection washes this structure out, yielding

$C(\Delta k)$9

This is not a far-field momentum-resolved measurement, but it establishes that transverse-momentum distinguishability does not annihilate two-photon coherence; it relocates the observable interference into the conjugate basis (Gonzalez et al., 20 Aug 2025).

4. Photon-induced two-source interference in ultra-peripheral heavy-ion collisions

In heavy-ion collisions, the closest analogue to a momentum-resolved double-slit interferometer is coherent $\nu$0 photoproduction in ultra-peripheral $\nu$1 collisions. The process

$\nu$2

admits two indistinguishable production paths: either nucleus can emit the quasi-real photon while the other acts as the scattering target. The production sites are separated by the impact parameter vector $\nu$3, so the system forms a genuine two-source interferometer. Because the $\nu$4 has negative parity, the midrapidity cross section is

$\nu$5

which is destructively interferometric as $\nu$6. After averaging over the impact-parameter distribution, the suppression scale is

$\nu$7

The measured $\nu$8 spectrum indeed shows a downturn for $\nu$9, consistent with the predicted interference (0812.1063).

The STAR measurement used the Time Projection Chamber in a $\rho^0$0 solenoidal field, together with the Central Trigger Barrel for topology-triggered exclusive $\rho^0$1 candidates and the Zero Degree Calorimeters for the minimum-bias sample with mutual Coulomb excitation. The selected $\rho^0$2 pairs satisfied $\rho^0$3, and the residual background estimated from like-sign pairs was $\rho^0$4. Corrected spectra were fitted with

$\rho^0$5

where $\rho^0$6 denotes no interference and $\rho^0$7 the full predicted Klein–Nystrand interference. The weighted average of the four fitted $\rho^0$8 values was first quoted as $\rho^0$9; after rescaling two errors by $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$0, the result became $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$1. Including systematics, the final interference strength was

$$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$2

of the expected level, implying

$$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$3

at $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$4 confidence level. The abstract quotes median impact parameters of $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$5 and $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$6 fm for the two event classes, which explains the broader dip in the minimum-bias sample (0812.1063).

This experiment is not standard $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$7 fusion and not HBT. Its final state is produced by photonuclear vector-meson photoproduction, and the interference is between two production amplitudes of one coherent meson system. The paper further emphasizes a nonlocality aspect: because $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$8, the $$\sigma(p_T,b,y)=\left|A(p_T,b,y)-A(p_T,b,-y)e^{i\vec p_T\cdot \vec b}\right|^2,$$9 decays before the amplitudes from the two sources can spatially overlap. The interference must therefore survive in the $\Delta k$00 final state, which is described as an entangled nonlocal wave function and explicitly connected to the Einstein–Podolsky–Rosen paradox (0812.1063).

A polarization-sensitive extension of the same two-source geometry was later formulated for coherent photoproduction. In the helicity frame, the decay angular distribution was written as

$\Delta k$01

with

$\Delta k$02

Because the quasi-real Coulomb-field photons are taken to be fully linearly polarized, $\Delta k$03, and the polarization plane coincides with the reaction plane. In the point-source approximation, the two-source amplitude is

$\Delta k$04

so that at midrapidity $\Delta k$05,

$\Delta k$06

The work predicts a periodic oscillation of the second-harmonic decay anisotropy with $\Delta k$07, providing a polarization-dimension readout of the same momentum-space double-slit structure (Zha et al., 2020).

5. Genuine $\Delta k$08 processes and the transverse-momentum structure of heavy-ion photon fields

The most direct heavy-ion realization of genuine two-photon production is the Breit–Wheeler process

$\Delta k$09

measured in ultra-peripheral $\Delta k$10 collisions at $\Delta k$11 GeV. The STAR analysis observed $\Delta k$12 exclusive $\Delta k$13 pairs, reported a sharply peaked pair transverse-momentum spectrum with

$\Delta k$14

and measured a large fourth-order angular modulation

$\Delta k$15

where the fit form was

$\Delta k$16

The smooth invariant-mass distribution showed vector meson contamination absent at the level of $\Delta k$17 of the observed yields. The measured low-$\Delta k$18 structure and $\Delta k$19 harmonic were described as consistent with QED calculations for collisions of linearly polarized photons from the Lorentz-contracted Coulomb fields of the nuclei (Collaboration et al., 2019).

Although this measurement is not formulated as an interferometric reconstruction, its observables are inherently transverse-momentum resolved on the photon side. At leading order, the pair transverse momentum is the vector sum of the incoming photon transverse momenta, so the pair $\Delta k$20 distribution probes the convolution of the two photon transverse-momentum distributions. The paper stresses that the $\Delta k$21 spectrum and its centrality dependence show that the energy spectrum of the colliding photons depends on impact parameter and therefore on the spatial distribution of the electromagnetic fields. It further argues that the observables are sensitive to the field’s spatial and momentum distribution, described as the Wigner function (Collaboration et al., 2019).

The same literature also contains a semi-coherent $\Delta k$22 baseline for finite-$\Delta k$23 dilepton production. In that treatment,

$\Delta k$24

so the pair transverse momentum is inherited from the harder photon. The differential cross section is built in $\Delta k$25, invariant mass $\Delta k$26, and rapidity $\Delta k$27, with the soft region identified as $\Delta k$28 and the contribution relevant up to roughly $\Delta k$29. The calculation uses the equivalent photon approximation with

$\Delta k$30

taking $\Delta k$31 and $\Delta k$32, and imposes a lower cutoff $\Delta k$33 from a single-track acceptance condition. Coherence is represented schematically through

$\Delta k$34

This framework is explicitly described not as a finished interferometric theory, but as a $\Delta k$35-resolved $\Delta k$36 production model and a coherence-based scaffold (Fu et al., 2011).

Taken together, these works show why genuine heavy-ion $\Delta k$37 measurements belong in the same encyclopedia entry but not under the same formal heading as HOM or HBT. They resolve the transverse momentum of the two-photon initial state indirectly, through the produced pair, and they expose coherence scales, impact-parameter dependence, and polarization structure. What they generally do not provide is an explicit two-photon correlation function or a beam-splitter-type amplitude exchange observable.

6. HBT source imaging and self-referenced tomographic generalizations

Direct-photon HBT interferometry provides the explicit second-order source-imaging branch of the subject. In central $\Delta k$38 Pb–Pb at $\Delta k$39 TeV, the two-photon correlator

$\Delta k$40

was evaluated as a function of the relative momenta $\Delta k$41, $\Delta k$42, $\Delta k$43 at fixed pair transverse momentum $\Delta k$44. In the Gaussian approximation,

$\Delta k$45

and the diagonal radii were extracted more robustly through moments,

$\Delta k$46

The paper compared a baseline thermal hydrodynamic source with two enhanced-yield scenarios: early pre-equilibrium photons from a bottom-up thermalization picture and a late near-$\Delta k$47 enhancement of thermal rates. The decisive finding was that the longitudinal channel is the most sensitive discriminator. In the summary, $\Delta k$48 changes by about $\Delta k$49 for the early-time enhancement and by about $\Delta k$50 for the late-time enhancement, whereas $\Delta k$51 changes by roughly $\Delta k$52 and $\Delta k$53 by roughly $\Delta k$54. The correlators are also non-Gaussian, particularly longitudinally, with normalized excess kurtosis

$\Delta k$55

and the abstract states that with projected $\Delta k$56 heavy-ion events, a direct-photon HBT measurement for $\Delta k$57 is statistically significant if only statistical uncertainties are considered (Garcia-Montero et al., 2019).

A distinct but related generalization moves from sensing to state reconstruction. The FRINGE protocol uses two identical copies of an unknown quantum field incident on a laterally displaced Young double slit, followed by photon-number-resolved far-field detection. In the Fraunhofer regime,

$\Delta k$58

so transverse detector position $\Delta k$59 is simultaneously a label of transverse wave vector and of relative slit phase. An $\Delta k$60-photon event projects onto

$\Delta k$61

after the flat-spectrum simplification $\Delta k$62. For the two-photon sector,

$\Delta k$63

For a pure single-slit state $\Delta k$64,

$\Delta k$65

The paper presents this as a self-referenced, photon-number-resolved double-slit interferometric tomography compatible with commercially available photon-number-resolving cameras (Tzur, 9 Dec 2025).

This tomographic branch is not HOM and not HBT. Its importance lies elsewhere: it makes explicit that far-field transverse coordinate, transverse momentum, and interferometric phase are interchangeable labels under paraxial Fourier optics. A plausible implication is that sensing, source imaging, and state reconstruction are not separate topics but different inverse problems built on the same transverse-phase sampling architecture.

7. Experimental architectures, operating regimes, and persistent misconceptions

Across the literature, three experimental architectures recur. The first is the balanced beam splitter plus far-field momentum readout, implemented with cameras, SPAD arrays, or scanned slits. The second is the ultra-peripheral heavy-ion “double slit,” in which the two nuclei act as coherent sources separated by the impact parameter $\Delta k$66. The third is a Young-type self-interferometer with photon-number-resolving far-field detection. All three convert a transverse displacement or source separation into a measurable phase in a conjugate momentum-like variable.

The instrumental bottlenecks differ by platform. In spatial HOM experiments they include finite momentum resolution, expressed through $\Delta k$67, limited visibility $\Delta k$68 or $\Delta k$69, absence of photon-number resolution for same-pixel bunching, adjacent-pixel crosstalk, and the need to suppress first-order interference by phase randomization (Sgobba et al., 27 Aug 2025). In momentum-entangled proposals, the requirements are narrow momentum peaks separated by a large $\Delta k$70, detectors resolving both the peak width and the fringe scale, and a loss model in which the two-photon information is reduced only by $\Delta k$71 (Triggiani et al., 2024). In heavy-ion UPC measurements, the dominant issues are the impact-parameter distribution, detector smearing, trigger modeling, form-factor assumptions, and the distinction between coherent and incoherent production (0812.1063, Collaboration et al., 2019).

Several misconceptions recur. One is to identify all low-$\Delta k$72 suppression phenomena with HBT. The $\Delta k$73 UPC interferometer is explicitly not HBT; it is interference between two production amplitudes of one meson system (0812.1063). A second is to equate all heavy-ion photon-induced interference with genuine $\Delta k$74 fusion. Coherent $\Delta k$75 photoproduction is photonuclear, not $\Delta k$76 (Zha et al., 2020). A third is to treat unresolved HOM failure under spatial mismatch as complete loss of useful interference. The resolved protocols show the opposite: spatial mismatch becomes an interference phase in the momentum basis, so the operational range is extended precisely by resolving the conjugate variable (Triggiani et al., 2023, Sgobba et al., 27 Aug 2025).

The field’s unifying lesson is that transverse distinguishability is basis dependent. In one basis it appears as which-path information and suppresses an integrated signal; in the conjugate basis it reappears as oscillatory phase structure. Whether the observable is $\Delta k$77, $\Delta k$78, $\Delta k$79, or the Fourier transform of a source function $\Delta k$80, the governing principle is the same: transverse momentum is the natural readout variable for spatially encoded phase information.