High Purity OAM Entangled Photons from SPDC with Reduced Spatial Spectral Correlations

Abstract: Entanglement generated by Spontaneous Parametric Down Conversion (SPDC) involves multiple, often mutually correlated degrees of freedom. These degrees of freedom are often treated independently, overlooking the intrinsic correlation between them. We focus on the spatial spectral correlations that, if left uncontrolled, introduce distinguishability and reduce coherence, undermining applications such as high-dimensional OAM encoding. We analyze the spatio spectral structure of the biphoton and identify source configurations enabling a strong reduction of such correlations. We then quantify how spatial spectral coupling degrades OAM spatial purity, mapping high-purity regions as functions of OAM order, crystal length, and pump/collection waists. The resulting design parameters enable engineering bright, high purity OAM entangled sources, reducing the need for loss-introducing filtering and therefore supporting scalable high-dimensional photonic quantum technologies.

• The paper presents a method to generate high-purity, high-flux OAM-entangled photons from SPDC by decoupling the spatial and spectral degrees of freedom.
• The study identifies for the fist time experimental conditions allowing room-temperature generation of spatial and spectral independency in type I SPDC
• The researchers develop a double-sinc phase-matching function approximation that holds for a 1 nm spectral window, eliminates photon loss while reaching 1 unity purity, and opens a path to scaleable high-dimensional quantum information processing.
• title 2 to compare for metadata exercise

Motivation and scope

Spontaneous parametric down-conversion (SPDC) generates entanglement simultaneously in polarization, frequency, transverse momentum, and orbital angular momentum (OAM). The energy- and momentum-conservation constraints that make this possible also tie the frequencies of a photon pair to its emission directions, so spatial and spectral degrees of freedom are generically correlated. In hyperentanglement experiments these degrees of freedom are usually treated as independent resources, but uncontrolled spatial–spectral coupling introduces distinguishability, reduces coherence, and degrades the purity of OAM-encoded states. The standard remedy—spectral or spatial filtering—restores purity at the cost of photon flux, which limits scalability in high-dimensional quantum information processing.

This work by de Brito, Kolenderska, and Kolenderski addresses the problem at the source level. Its stated goal is to identify experimental conditions under which spatial and spectral degrees of freedom can be engineered independently in type-I SPDC without sacrificing brightness, and to map how residual coupling degrades OAM spatial purity as a function of OAM order, crystal length, pump waist, collection waist, and pump pulse duration.

Spatial–spectral decoupling of the phase-matching function

Working under the paraxial and small-angle approximations, the authors expand each longitudinal wavevector $k_{z,j}$ to second order in transverse momentum and frequency detuning about degeneracy, assuming zero transverse walk-off. For type-I SPDC, where signal and idler share polarization and hence identical group velocity and group velocity dispersion (GVD), the dimensionless phase mismatch separates into three terms: a quadratic transverse-momentum mismatch proportional to $|\vec{q}_s-\vec{q}_i|^2/2k_p$, a quadratic dispersion mismatch proportional to $\beta(\Omega_s-\Omega_i)^2$, and a linear group-velocity-mismatch term $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$. The authors emphasize that both spectral terms must be retained: the linear term dominates for short pulses, while the quadratic term is required to give the single-photon spectrum a finite width even in the continuous-wave limit.

The central analytical step is inspired by Schneeloch and Howell's treatment of transverse correlations. In the small-argument regime the sinc-shaped phase-matching function (PMF) is approximated as a product of two sinc functions, one depending only on the transverse momenta $\Theta(q_s,q_i)$ and one only on frequencies $\tilde{\Theta}(\Omega_s,\Omega_i)$. Numerical comparison for a 5 mm LiIO$_3$ crystal pumped at 400 nm ($w_p = 28\,\mu$m) shows the double-sinc approximation reproduces the general PMF well over a signal-wavelength window of roughly 1 nm around degeneracy (800 nm), provided the relative transverse momentum satisfies $|q_s-q_i|\lesssim 0.02\,\mu\mathrm{m}^{-1}$; conversely, when the photons are wavelength-degenerate the approximation holds across essentially the full range of allowed transverse momenta. This asymmetric behavior is physically meaningful: the PMF strongly constrains wavelength matching while being comparatively permissive in momentum space, so within this parameter regime the joint intensity distribution is naturally localized in frequency but weakly constrained in angle—the condition under which spatial and spectral degrees of freedom are effectively decoupled without external filtering.

The appendix extends the derivation to type-II SPDC, where orthogonal polarizations introduce distinct GVD coefficients for signal and idler. There the spectral mismatch takes the form $$(\sqrt{\mathrm{GVD}_s}\,\Omega_s-\sqrt{\mathrm{GVD}_i}\,\Omega_i)^2/2$$ plus linear group-velocity terms, which the four-Gaussian model captures explicitly. The resulting joint spectral amplitude (JSA) ellipticity and tilt are shown to be governed by the sign and magnitude of the GVD asymmetry between signal and idler—an effect relevant for tailoring heralded single-photon sources and quantum-interference schemes.

Four-Gaussian approximation

Including Gaussian pump envelopes in space and time and replacing each sinc factor with a second-moment-matched Gaussian (with prefactor $|\vec{q}_s-\vec{q}_i|^2/2k_p$0), the biphoton amplitude becomes a fully factorable product of four Gaussians: two governing position/momentum correlations and two governing frequency sum and difference. The spatial correlation width is fixed by $|\vec{q}_s-\vec{q}_i|^2/2k_p$1, inherited directly from the crystal length. The spectral part is more delicate: because the short-pulse PMF is an asymmetric function of two arguments, a direct Gaussian substitution fails. Instead the authors fit Gaussians to the reduced single-photon spectral density, following Fedorov's formalism, and interpolate between long- and short-pulse asymptotes via effective widths $|\vec{q}_s-\vec{q}_i|^2/2k_p$2 and $|\vec{q}_s-\vec{q}_i|^2/2k_p$3, with a phenomenological coefficient $|\vec{q}_s-\vec{q}_i|^2/2k_p$4 set to 0.4 for short pulses and 1 for long pulses (and analogously $|\vec{q}_s-\vec{q}_i|^2/2k_p$5 versus 1 for type-II). Comparisons against the general model show good agreement of spatial profiles for both $|\vec{q}_s-\vec{q}_i|^2/2k_p$6ps and $|\vec{q}_s-\vec{q}_i|^2/2k_p$7fs, and of spectra for both pulse regimes, including the slightly broader type-II bandwidth. Within the validated regime, the four-Gaussian model is exactly separable and therefore yields unit spatial purity—a strong claim, but one conditional on the approximation remaining faithful to the general wavefunction.

OAM-resolved purity analysis

For a Gaussian ($|\vec{q}_s-\vec{q}_i|^2/2k_p$8) pump, OAM conservation enforces $|\vec{q}_s-\vec{q}_i|^2/2k_p$9. The authors model collection into Laguerre–Gaussian modes of radial index $\beta(\Omega_s-\Omega_i)^2$0 with equal signal and idler waists $\beta(\Omega_s-\Omega_i)^2$1, trace out the spectral subsystem from the full density operator, and compute the spatial purity $\beta(\Omega_s-\Omega_i)^2$2. To keep integrals tractable they retain only the dominant quadratic terms of the phase mismatch—a simplification whose consequences are discussed below.

Three design rules emerge from the numerical study:

Parameter varied	Effect on spatial purity
Collection-to-pump waist ratio $\beta(\Omega_s-\Omega_i)^2$3	Purity rises toward unity as $\beta(\Omega_s-\Omega_i)^2$4 increases
OAM order $\beta(\Omega_s-\Omega_i)^2$5	Higher $\beta(\Omega_s-\Omega_i)^2$6 requires larger $\beta(\Omega_s-\Omega_i)^2$7 for the same purity
Crystal length $\beta(\Omega_s-\Omega_i)^2$8	Longer crystals raise purity faster with $\beta(\Omega_s-\Omega_i)^2$9
Pump duration $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$0	No effect for wavelength-degenerate photons

The physical interpretation is consistent throughout: a small collection waist implies broad angular acceptance, admitting photons whose transverse momenta are strongly coupled to their frequencies, whereas enlarging $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$1 narrows the momentum distribution and suppresses the correlation channel. Higher-$(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$2 LG modes extend further in momentum space and therefore sample regions of stronger coupling, imposing stricter collection requirements; longer crystals tighten the phase-matching cone and admit fewer angle-dependent frequency components. Notably, for degenerate photons the purity is insensitive to pump pulse duration, meaning the decoupling condition does not restrict the choice between pulsed and continuous-wave operation—an important degree of freedom for source engineering. The practical implication is that high-purity, high-flux OAM sources can be designed by choosing $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$3, $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$4, and $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$5 jointly rather than by filtering after generation.

Limitations and open questions

Several assumptions bound the validity of these results. First, the entire framework rests on the paraxial and small-angle expansions and on neglecting birefringence-related technical effects such as walk-off; beyond-paraxial corrections to OAM properties are known to be non-negligible in focused geometries. Second, the purity analysis retains only the dominant quadratic terms of the phase mismatch, so quantitative predictions far outside the small-argument regime should be re-examined. Third, the double-sinc decomposition is accurate only for relative transverse momenta below approximately $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$6 and within a roughly 1 nm spectral window for the reference geometry; sources targeting larger bandwidths or larger angular separation fall outside the demonstrated validity range. Fourth, the interpolation coefficients $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$7 and $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$8 are fitted parameters rather than derived quantities, and their robustness across material systems is not established here. Finally, the analysis is restricted to radial index $(1/v_{g,p}-1/v_{g,s})(\Omega_s+\Omega_i)$9 and equal signal/idler waists; extension to higher-order radial modes and asymmetric collection remains open, as does experimental verification of the predicted purity scaling.

Conclusion

The paper provides a source-level framework for suppressing spatial–spectral correlations in SPDC: a factorization condition on the phase-matching function, a four-Gaussian model valid for arbitrary pump durations via density-matrix-preserving spectral fitting, and a mapping of high-purity operating regions for OAM entanglement in terms of OAM order, crystal length, and collection/pump waists. The main result is a set of concrete design rules under which OAM entanglement can be obtained at unit spatial purity without lossy filtering, with direct relevance to high-dimensional QKD, mode-multiplexed entanglement swapping, heralded sources, and OAM-enhanced quantum optical coherence tomography—applications that the authors note require both high mode purity and stable mode matching. The framework's reliance on paraxial, small-argument, and quadratic-phase-mismatch approximations defines the boundary within which those rules apply.