---
title: X-ray to Optical Parametric Down-Conversion
url: https://www.emergentmind.com/topics/x-ray-to-optical-parametric-down-conversion
type: topic
---

# X-ray to Optical Parametric Down-Conversion

Searching arXiv for recent and relevant papers on x-ray to optical parametric down-conversion.
X-ray to optical parametric down-conversion denotes a second-order nonlinear process in which an incoming x-ray pump photon of frequency $\omega_p$ splits into a lower-energy x-ray signal photon at $\omega_s$ and a long-wavelength idler photon at $\omega_i$, subject to energy conservation $\omega_p=\omega_s+\omega_i$ and crystal-momentum conservation through a reciprocal-lattice vector $\mathbf G$ [2002.12822]. In the non-degenerate regime of principal interest, the idler lies in the optical, ultraviolet, or soft-x-ray range while the signal remains near the Bragg-diffracted x-ray wavevector. The subject sits at the intersection of nonlinear x-ray optics, crystallography, and condensed-matter spectroscopy: it has been proposed as a route to probe valence-electron density fluctuations and band-selective response, yet its experimental status is heterogeneous across spectral regimes and materials, with strong claims in some non-centrosymmetric systems, null results in bulk diamond for x-ray-to-visible conversion, and more recent resonant studies linking XPDC to polaritonic hybridization near absorption edges [1904.13146], [2002.12822], [2606.02619].

## 1. Fundamental process and kinematic constraints

XPDC is treated as a $\chi^{(2)}$ nonlinear interaction in which the pump photon decays into two lower-energy photons. In crystal form, the process is constrained by

$$
\omega_p=\omega_s+\omega_i
$$

and

$$
\mathbf k_p+\mathbf G=\mathbf k_s+\mathbf k_i
\quad\Longrightarrow\quad
\Delta\mathbf k=\mathbf k_p-\mathbf k_s-\mathbf k_i+\mathbf G=0,
$$

where $\mathbf G$ supplies lattice momentum [2002.12822]. In the highly non-degenerate x-ray-to-optical regime, $\omega_s\gg\omega_i$ and $\mathbf k_s$ lies only a few tens of millidegrees from the Bragg-diffracted wavevector, so the expected XPDC signature appears as a small ellipse in a rocking-curve map over sample angle and scattering angle [2002.12822].

A standard perturbative description expresses the small-signal conversion efficiency $\eta$ as proportional to the square of the second-order susceptibility and to phase-matching quality. One summary provided for the diamond null-result study writes

$$
\eta \propto |\chi^{(2)}|^2 L^2 \mathrm{sinc}^2\!\Bigl(\frac{\Delta k\,L}{2}\Bigr),
$$

with $L$ the effective interaction length [2002.12822]. A more explicit undepleted-pump expression reported for GaAs and LiNbO$_3$ is

$$
\eta(\omega_s,\omega_i)\simeq
\frac{\omega_s^2\omega_i}{\varepsilon_0^2 c^3 n_p n_s n_i}
\bigl|\chi^{(2)}_{\rm eff}\bigr|^2 I_p\,L^2\,
\mathrm{sinc}^2\!\Bigl(\tfrac{\Delta k\,L}{2}\Bigr),
$$

where $I_p$ is the pump intensity, $n_p,n_s,n_i$ are refractive indices, and $\chi^{(2)}_{\rm eff}$ is the polarization-projected susceptibility [1904.13146]. This suggests that practical observability is jointly limited by the intrinsic weakness of x-ray nonlinearities, the short effective length imposed by absorption, and the narrow angular acceptance of phase matching.

The same conservation laws also underpin related x-ray optical wavemixing channels such as sum-frequency and difference-frequency generation. A non-relativistic QED treatment places XPDC and x-ray optical sum-frequency generation on a common footing and identifies the observable scattering pattern with an underlying response function of the medium [2104.05838]. In that formalism, XPDC corresponds to replacing the sum-frequency condition with $\omega_X=\omega_f+\omega_O$ and $\mathbf k_X=\mathbf k_f+\mathbf k_O+\mathbf G$ [2104.05838].

## 2. Quantum and response-theoretic descriptions

A central theoretical development is the non-relativistic QED framework for parametric x-ray optical wavemixing, formulated from the Coulomb-gauge Hamiltonian

$$
\hat H=\hat H_{\rm mat}+\hat H_{\rm em}+\hat H_{\rm int},
$$

with minimal-coupling interaction

$$
\hat H_{\rm int}
=\int d^3x\;\hat\psi^\dagger(\mathbf x)\Bigl[\alpha\,\mathbf p\!\cdot\!\hat{\mathbf A}(\mathbf x)
+\tfrac{\alpha^2}{2}\,\hat{\mathbf A}^2(\mathbf x)\Bigr]\hat\psi(\mathbf x)
$$

[2104.05838]. The resulting S-matrix leads to a double-differential scattering probability that depends on a microscopic matter correlator, and the key material quantity is a time-ordered density-momentum correlator whose Fourier transform defines a $\chi^{(2)}$-like kernel $\mathcal K_I(\mathbf k_1,\mathbf k_2;\omega)$ [2104.05838]. In a mean-field KS-DFT approximation this kernel is expressed as a sum over occupied and unoccupied states, making the band- and orbital-resolved structure explicit [2104.05838].

Within that framework, the measured nonlinear scattering intensity for an ideal plane-wave probe factorizes as

$$
I(\mathbf G;\omega_O)\propto |\mathcal K(0,\mathbf G;\omega_O)|^2,
$$

which motivates a nonlinear analogue of crystallographic reconstruction: measuring many $\mathbf G$ and $\omega_O$ values provides access to the modulus of the response kernel, with phase retrieval then required for real-space inversion [2104.05838]. This establishes a formal basis for using x-ray/optical mixing to reconstruct microscopic response functions rather than only integrated conversion efficiencies.

A separate quantum-mechanical theory was invoked to interpret polarization-dependent x-ray-to-UV PDC in GaAs. There the nonlinear conductivity at the signal frequency is decomposed into two orthogonal polarization channels, one parallel and one perpendicular to the pump polarization, with coefficients $A(\omega_p,G)$ and $B(\omega_p,G)$ depending on Wannier functions, momentum matrix elements, and a Brillouin-zone integral encapsulating the joint density of states [2012.07299]. The observable analyzer-angle dependence takes the form

$$
I_s(\theta)=I_0\cos^2(\theta+\theta_0),
$$

so a nonzero polarization shift $\theta_0$ indicates a nonzero perpendicular channel and implies that the generated signal need not remain parallel to the pump [2012.07299]. The interpretation advanced there is that only the parallel channel carries direct information on the Fourier component of the induced valence charge density, while the perpendicular channel reflects interband coherence effects [2012.07299].

These formalisms collectively shift XPDC away from a purely phenomenological $\chi^{(2)}$ picture toward a microscopic description in terms of density response, band structure, Wannier functions, and polarization-resolved current operators. A plausible implication is that discrepancies between experiments may partly reflect differing sensitivity to specific microscopic channels rather than a single universal bulk $\chi^{(2)}$ parameter.

## 3. Experimental implementations and observables

The best-documented non-degenerate bulk-diamond search employed the ESRF ID20 high-resolution diffractometer with a Si(111) double-crystal monochromator, optional downstream Si(311) high-resolution monochromator, angular divergence of about $1.1$ mdeg, and a beam size of $0.2\times0.2$ mm$^2$ [2002.12822]. The sample was single-crystal diamond with a $<100>$ surface and $500\,\mu$m thickness, measured in Laue transmission geometry using the 220 orientation with $\pi$-polarization [2002.12822]. The scattered x-ray energy $\omega_s$ was selected by a Si(220) channel-cut analyzer with $\Delta E\approx0.3$ eV, and a $256\times256$ pixel photon-counting detector with $55\,\mu$m pixels provided $2.5$ mdeg/pixel angular resolution [2002.12822]. Systematic scans over sample angle $\Delta\Omega$ and analyzer detuning $\Delta E=\omega_p-\omega_s$ were designed so that an XPDC signal would trace the predicted phase-matching ellipse in the $(\Delta\Omega,2\theta_s)$ map [2002.12822].

The earlier high-efficiency observations in non-centrosymmetric media used different materials and a somewhat broader instrumentation profile. In GaAs at Diamond Light Source beamline I16, a monochromatic, collimated $10.3$ keV beam of about $10^{10}$ photons/s was focused to $20\,\mu$m $\times 180\,\mu$m, while LiNbO$_3$ measurements at ESRF ID-20 used $10$ keV and about $5\times10^{10}$ photons/s with beam size $0.4$ mm $\times0.4$ mm [1904.13146]. Multi-bounce Si analyzers selected the transmitted x-ray signal with combined energy resolution $\Delta E\simeq1$ eV in GaAs and $\simeq0.3$ eV in LiNbO$_3$, and a Medipix detector recorded two-dimensional rocking curves with horizontal axis corresponding to deviation from Bragg angle and vertical axis to deviation in energy from the pump [1904.13146].

The polarization-dependence study in GaAs used Diamond Light Source I16 with $E_p=8.388$ keV, $\Delta E\approx1$ eV, beam spot about $100\,\mu$m, and divergence about $11\,\mu$rad [2012.07299]. A multi-bounce Si(333) channel-cut crystal at Bragg angle about $45^\circ$ served as both polarization analyzer and energy filter, and a MerlinEM detector at $1.25$ m from the analyzer recorded the PDC feature [2012.07299]. The protocol fixed the idler energy near $10$ eV and scanned the analyzer rotation $\theta\in[0,180^\circ]$ to extract the polarization shift from $I_s(\theta)$ [2012.07299].

More recently, resonant XPDC near the diamond K-edge has been implemented at ESRF ID20 with a Si(111) pump monochromator tuned between $9.96$ keV and $9.70$ keV, a high-purity single-crystal diamond sample, and a spherically-bent Si(660) analyzer in near-backscattering geometry set to $\hbar\omega_s=9.69$ keV [2606.02619]. A 2D detector placed slightly upstream of the analyzer focus recorded the full XPDC cone angular distribution in both in-plane and out-of-plane directions, with combined energy resolution about $1.25$ eV [2606.02619]. The experiment built a two-dimensional polariton spectral map by stacking positive-$\chi$ half-cones as a function of effective idler momentum and detection energy [2606.02619].

## 4. Reported observations, null results, and comparative interpretation

The literature contains two sharply different classes of result. One class reports strong non-degenerate PDC into ultraviolet or visible wavelengths in non-centrosymmetric crystals. In GaAs and LiNbO$_3$, conversion efficiencies were reported in the range $\eta\simeq10^{-8}$ to $10^{-6}$, described as about four orders of magnitude larger than efficiencies measured before in diamond, silicon, or other centrosymmetric crystals [1904.13146]. The measurements displayed dependence on crystallographic plane and idler energy: in GaAs, $\eta$ ranged from about $10^{-8}$ near $\omega_i\approx10$ eV down to about $10^{-9}$-$10^{-10}$ near $\omega_i\approx30$ eV, while in LiNbO$_3$ values reached about $10^{-6}$ near $\omega_i\approx5$ eV on the polar (006) planes and fell to about $10^{-8}$ near $\omega_i\approx20$ eV [1904.13146]. Spectral peaks were reported near known band-gap transitions and deeper resonances, including the $1.43$ eV direct band-gap transition in GaAs and Li-2s and Nb-4s resonances in LiNbO$_3$ at $5$-$10$ eV [1904.13146].

A second class of result challenges the interpretation of x-ray-to-visible XPDC in bulk diamond. In the high-resolution ESRF study, scans targeting an optical idler around $2.2$ eV showed only a strong central spot attributed to residual Bragg scattering and diffuse wings shifting linearly with $\Delta\Omega$; no isolated ellipse or peak consistent with XPDC phase matching was found [2002.12822]. The measured features did not coincide with the calculated XPDC locus, and when analyzer detuning was increased to $10$ eV the angular patterns remained fixed in position while diminishing in overall intensity, behavior judged inconsistent with XPDC but consistent with elastic scattering from residual flux in monochromator spectral tails [2002.12822]. High-resolution runs at $10$ keV pump energy suppressed the wings by orders of magnitude yet still showed no XPDC signature [2002.12822]. The empirical upper bound extracted for the conversion efficiency was $\eta_{\rm max}\lesssim10^{-11}$ within the resolution of that setup [2002.12822].

The same study explicitly reassessed earlier positive claims by degrading its own high-resolution data to emulate the lower-resolution APD-based configuration used previously. Under flux- and resolution-normalized convolution, the resulting peak shapes and count rates reproduced those earlier observations, leading to the conclusion that the previously reported peaks can be explained entirely by elastic scattering backgrounds [2002.12822]. This is a direct controversy in the field: the disagreement is not only quantitative but interpretive, concerning whether observed off-Bragg features in the visible-idler regime represent true XPDC or unresolved elastic background.

A further experimentally grounded complexity is added by polarization measurements in GaAs. There, the fitted PDC polarization shifts were reported as $+7.4^\circ$ for (111), $+5.1^\circ$ for (200), and $-17.6^\circ$ for (333), whereas the elastic Bragg reflection remained centered at $\theta_0\simeq0^\circ$ [2012.07299]. The reported result was that classical $\chi^{(2)}$ theory predicts $\theta_0=0$ for all reflections, while a quantum model allowing distinct parallel and perpendicular nonlinear channels qualitatively agrees with the existence of nonzero shifts [2012.07299]. This does not resolve the broader efficiency controversy, but it indicates that at least some observed x-ray-to-UV PDC observables display structure not captured by the simplest classical treatment.

## 5. Resonant regime and polaritonic XPDC

A distinct regime emerges when the idler approaches a strong electronic resonance. Around the diamond carbon K-edge, XPDC has been used to access high-energy polaritons formed by hybridization of the down-converted idler photon with electronic excitations in the nonlinear medium [2606.02619]. The theoretical description uses a two-level Hopfield model in the basis of bare idler photon and core-excited electron, with polariton Hamiltonian

$$
H^{\rm pol}=\hbar
\begin{pmatrix}
\omega_\gamma & 0\\
0 & \omega_e
\end{pmatrix}
+
\begin{pmatrix}
0 & V\\
V^* & 0
\end{pmatrix},
$$

yielding polariton branches

$$
E_{\pm}(\mathbf k_i)=\frac{\hbar(\omega_\gamma+\omega_e)}{2}
\pm
\sqrt{\frac{\hbar^2}{4}(\omega_\gamma-\omega_e)^2+|V|^2}
$$

[2606.02619]. Strong coupling is reached when $2|V|\gtrsim\hbar\Gamma_{\rm pol}$, and at the diamond K-edge the reported ratio was $2V/\hbar\Gamma\approx2.7$ [2606.02619].

Experimentally, the polariton spectral map plots XPDC intensity against detection energy $\omega_d=\omega_p-\omega_s$ and effective idler momentum $c|\mathbf k_i|$, revealing an anti-crossing nodal line that directly visualizes the upper and lower polariton branches [2606.02619]. A horizontal suppression at $\omega_d\approx302.5$ eV was associated with the second band gap in the diamond p-projected density of states, and Gaussian-broadened simulations of the polariton branches reproduced the key features of the measured map [2606.02619]. The same work reported that the XPDC count rate follows the p-projected density of states around the $1s$ edge [2606.02619].

The resonant study also extracted the refractive index $n(\omega)$ of bulk diamond from phase-matched cone diameters, reporting typical modulation of about $\pm1\%$ with fitted error bars of at most $10^{-3}$ [2606.02619]. The measured refractive index systematically exceeded older Kramers-Kronig-inverted reflectivity data while reproducing the same spectral features, and ab initio DFT calculations using FHI-aims with the HSE06 hybrid functional were reported to confirm the higher absolute magnitude and fine spectral detail [2606.02619]. This suggests that resonant XPDC is not merely a weak-frequency-conversion channel but can function as a bulk-sensitive soft-x-ray spectroscopic probe of optical constants and polaritonic dispersion.

## 6. Materials dependence, selection rules, and nanostructure control

Material symmetry is central to the interpretation of XPDC efficiencies. The strong efficiencies reported in GaAs and LiNbO$_3$ were attributed to a strong $\chi^{(2)}$ electric-dipole channel in non-centrosymmetric media, in contrast to intrinsically weak quadrupolar nonlinearities available even in centrosymmetric systems [1904.13146]. The same study argued for orbital- and band-selective valence contributions: in GaAs, different reflections emphasized either atomic-resonance peaks or the direct band-gap transition, while in LiNbO$_3$ the polar (006) planes yielded efficiencies two orders of magnitude above (110), indicating enhancement tied to the ferroelectric dipole and nonzero $\chi^{(2)}$ tensor components along the c-axis when $E_i\parallel c$ [1904.13146]. This suggests that reflection choice and polarization geometry act as microscopic selectors for particular electronic transitions.

The polarization study sharpened that point by linking the nonlinear conductivity to Wannier-function matrix elements and arguing that spectral and polarization dependencies can probe the real-space symmetry and coupling of Wannier orbitals in solids [2012.07299]. In that view, XPDC is not only sensitive to whether a crystal lacks inversion symmetry, but to how specific Bloch and Wannier amplitudes project onto the chosen reciprocal-lattice vector and analyzer geometry.

A more recent theoretical extension shows that optical nanostructures can control x-ray/optical nonlinear processes even though the x-ray wavelength itself remains extremely short [2507.22302]. In that framework the three-wave-mixing Hamiltonian is written in terms of a modulation of valence-electron charge density by the optical idler mode and leads to a differential production rate

$$
\frac{d\Gamma_{i,s}}{d\omega_s\,d\Omega_s}
=
\frac{\omega_i\,\omega_s^3\,E_p^2}{16\pi^2 c^3}\,
\bigl|\hat\epsilon_p\!\cdot\!\hat\epsilon_s^*\bigr|^2
\sum_G|\chi_G^{(2)}|^2
\Bigl|\int_V d^3r\,[\hat G\!\cdot\!F_i^*(r)]\,e^{i\,(G+k_p-k_s)\cdot r}\Bigr|^2
\delta(\omega_p-\omega_s-\omega_i)
$$

[2507.22302]. The optical idler eigenmode $F_i(r)$ of the nanostructure appears explicitly, so photonic-band engineering reshapes both the spectral and spatial properties of the emitted x-rays [2507.22302].

For a GaAs woodpile photonic crystal with $\omega_p=6.6$ keV, the reported fill-factor-normalized rate enhancement was about $2.2$ at $\omega_i\approx1.6$ eV relative to bulk [2507.22302]. The analysis attributes enhancement to both modified optical density of states and improved overlap integrals under phase matching, while symmetry and reciprocal vectors govern the directionality of x-ray emission [2507.22302]. A plausible implication is that nanophotonic engineering may become especially relevant in regimes where bulk XPDC is real but too weak or too diffuse to exploit without spectral and angular mode shaping.

## 7. Applications, limitations, and open questions

Several application directions recur across the literature. The general QED theory emphasizes imaging capabilities analogous to x-ray diffraction but with additional spectroscopic selectivity tunable through the optical admixture, and it frames nonlinear “crystallography” as reconstruction of a microscopic response function from nonlinear scattering measurements [2104.05838]. The strong-efficiency reports in GaAs and LiNbO$_3$ propose orbital- and band-selective spectroscopy, access to valence electronic density of states and band gaps at atomic spatial resolution, and sensitivity to anisotropic charge distributions in ferroelectrics, multiferroics, and complex oxides [1904.13146]. The polarization study argues that full reconstruction of valence charge density requires polarization-resolved measurements over many reciprocal-lattice vectors because only one polarization channel is directly tied to induced charge density [2012.07299]. The resonant diamond work extends the scope further to bulk refractive-index metrology and EUV/soft-x-ray polariton spectroscopy [2606.02619].

At the same time, the field is constrained by major unresolved issues. The most explicit controversy concerns the interpretation of x-ray-to-visible XPDC in bulk diamond, where improved angular and energy resolution led to a null result and to the claim that prior evidence should be reexamined as elastic scattering background [2002.12822]. This places stringent constraints on any theory predicting observable non-degenerate XPDC rates in bulk diamond, and the same study stated that preliminary quantum-electrodynamical calculations by Krebs and Rohringer were already consistent with the upper bound $\eta\le 10^{-11}$ [2002.12822]. The contrast with the much larger efficiencies reported in GaAs and LiNbO$_3$ leaves open whether material symmetry and resonance structure alone account for the discrepancy, or whether some reported positive observations still await a fully predictive microscopic theory.

The literature also identifies concrete methodological priorities. The diamond null-result study recommends stronger suppression of elastic-scattering background via improved monochromators, tighter collimation, and UHV beam paths; coincidence detection of signal-idler pairs to unambiguously tag correlated photons; materials or nanostructures with enhanced $\chi^{(2)}$ at x-ray frequencies; and time-domain approaches using femtosecond x-ray/optical pump-probe schemes to exploit transient valence-electron coherences and dynamical phase matching [2002.12822]. The nanostructure proposal likewise points toward more monochromatic heralded x-ray sources, enhanced ghost imaging of lattice and electronic dynamics, and spectroscopy beyond the standard quantum limit [2507.22302].

Taken together, the current state of x-ray to optical parametric down-conversion is best understood as a technically mature but experimentally non-uniform research area. Its kinematics and microscopic response theory are well developed; its resonant and polarization-resolved manifestations reveal rich electronic-structure sensitivity; its bulk non-resonant observability remains strongly material- and background-dependent; and its most credible future advances appear likely to come from coincidence-based detection, resonant-edge operation, and optical-mode engineering rather than from straightforward extrapolation of conventional bulk nonlinear-optical intuition [2104.05838], [2002.12822], [2606.02619], [2507.22302].

Source: https://www.emergentmind.com/topics/x-ray-to-optical-parametric-down-conversion