Skip to content

Polariton spectroscopy at the diamond K-edge via X-ray parametric down-conversion

FieldValue
JournalScientific Reports
Published2026-09-11
AuthorsFridtjof Kerker, Dietrich Krebs, Xenia Brockmüller, Ankita Negi, Christoph J. Sahle
DOI10.1038/s41598-026-70799-6

X-ray parametric down-conversion produced soft X-ray polaritons at diamond’s carbon K-edge with a fitted coupling strength of 3.32 eV. At 291 eV energy transfer, the fitted bandwidth was 2.44 eV, giving 2V/ℏΓ = 2.72. The interaction therefore reached the strong-coupling regime and exceeded the coupling reported for the authors’ earlier EUV experiment.

Diamond was selected for its high crystalline quality, resistance to radiation, and structured electronic excitations near the carbon K-shell ionization edge. In X-ray parametric down-conversion, a hard X-ray pump photon splits into a correlated signal and idler photon. The idler enters the soft X-ray range, is absorbed and re-emitted, and hybridizes with an electronic excitation. The correlation leaves the hybridization encoded in the higher-energy signal.

Measured XPDC patterns changed markedly with idler energy. Their radial intensity profiles and cone sizes carried the signatures of polariton dispersion. This response was much stronger at the K-edge than in the non-resonant EUV case studied previously.

The authors also describe polaritonic XPDC as non-resonant X-ray Raman scattering from polariton excitation modes. Unlike ordinary Raman scattering from excitations such as plasmons, the XPDC phase-matching condition is sharply defined. The signal and idler magnitudes are fixed by energy conservation, while their common rotation produces emission into cones.

The experiment used the X-ray Raman spectrometer at ESRF beamline ID20. A Si(111) double-crystal monochromator conditioned the pump beam, which had a focal spot about 5 μm in diameter. The diamond was initially aligned to the (220) Bragg reflection. A Si(660) spherically bent crystal analyzer selected a 9.69 keV signal and dispersed it onto a two-dimensional detector with 55 × 55 μm² pixels.

The analyzer fixed the signal energy, while pump-energy changes set the idler energy and measured energy transfer. Rocking-angle offsets of 3.6°, 3.7°, and 3.87° placed the 291 eV, 306 eV, and 310 eV cases at phase matching. Off-phase acquisitions at Ω + 0.7° were subtracted to remove diffuse inelastic scattering and instrumental artefacts.

An initial inelastic X-ray scattering scan covered 260 to 360 eV with 1.25 eV FWHM energy resolution. It resolved diamond K-shell peaks at 294 eV, 299 eV, and 306 eV, plus post-edge structure.

The authors stacked central lineouts into a two-dimensional map with energy transfer on the vertical axis and effective idler momentum, scaled to energy, on the horizontal axis. The map combined scans from 278 to 314 eV and retained the scattering signal without background subtraction.

The map was nearly empty below the 289.5 eV K-edge, apart from a weak tail near 287 eV. Above the edge, a diagonal nodal line revealed the anticrossing of the polariton branches. A separate horizontal suppression at 302.5 eV matched the dip in the inelastic scattering spectrum and the second band gap of diamond.

Summing across the horizontal momentum axis reproduced the inelastic scattering spectrum. The map therefore retained the measured spectral information while resolving its momentum dependence.

The lineouts changed sign and shape across the edge. At 291 eV, the profile had two outer peaks, inner dips, and a lowered central plateau. At 306 eV, it had two sharp negative features. At 310 eV, outer dips flanked inner peaks, with a slightly raised central plateau.

Each resonance was represented by a two-level system containing a photonic idler state and an electronic K-shell excitation. The coupling V opened an avoided crossing and produced upper and lower polariton branches. The forward model included the pump and analyzer bandwidths, an intrinsic polariton linewidth, the angular dependence of coupling, and a background prefactor.

Only the electronic component contributes to the electron-density correlation in the model. The photonic component would require simultaneous photon creation. The coupling is angularly modulated through the transversality constraint on the idler photon.

Selected fits from the full parameter table show how the extracted parameters varied across the edge.

Energy transferExcitation energyCoupling VBandwidth ℏΓRefractive index n
278 eV284.24 eV0.10 eV1.49 eV1.0007
289.5 eV293.15 eV2.14 eV1.71 eV1.0076
291 eV293.80 eV3.32 eV2.44 eV1.0026
302.5 eV302.26 eV0.90 eV0.92 eV0.9945
306 eV305.84 eV2.27 eV1.92 eV0.9974
310 eV308.27 eV1.53 eV1.60 eV0.9953

Coupling followed the scaled inelastic scattering signal and the density of states for dipole-allowed transitions. It fell before the edge and behind the pronounced edge structure. The fitted response also captured the 302.5 eV dip. The ratio 2V/ℏΓ reached 2.72 at 291 eV, confirming strong coupling.

The fit converted the effective idler momentum into photon energy through the dispersion relation containing n. It therefore determined n together with the excitation energy, coupling, and bandwidth from each momentum-resolved pattern.

The authors report an overall refractive-index modulation of at most about 5% across the measured edge region. Despite this small modulation, the fitted spectra resolved its energy dependence with small error bars. Error estimates used background-counting fluctuations and were propagated through the model-fit covariance matrix.

The method provides a bulk optical-constant measurement in the EUV and soft X-ray range, where direct bulk data are difficult to obtain. It can therefore complement measurements that mainly probe a material near its surface.

Earlier Kramers-Kronig results for natural diamond gave systematically smaller refractive-index values than the XPDC fits. Their spectral features, however, agreed well with the new data.

FHI-aims calculations with the HSE06 hybrid potential reproduced the measured features and supported the higher overall refractive index. The XPDC result therefore agrees with theory while separating from the earlier reflectance inversion.

The simple two-level model does not reproduce the measured left-to-right intensity asymmetry near the K-edge. This deviation concerns the upper and lower branches associated with the 293.8 eV electronic excitation. The authors leave its description outside the scope of the initial model.

The experiment also has a 1.25 eV FWHM energy resolution. The authors identify improvements to the setup and measurement bandwidth as future work. They expect stronger hybridization at isolated pre-edge resonances, such as those in graphite or hexagonal boron nitride, although those materials were not measured here.

For diamond optics, the results separate nonlinear XPDC response from ordinary absorption. The coupling changes sharply with idler energy, while the selected (220) lattice planes define the phase-matching geometry. A diamond soft X-ray polariton source therefore requires both the correct crystal orientation and tuning across the K-edge.

The authors argue that polaritonic down-conversion can complement near-surface reflectometry when optical constants are needed for EUV lithography or fundamental material science.

The paper requires a crystalline diamond nonlinear sample in the (220) reflection geometry for soft X-ray polariton spectroscopy. 6CCVD’s diamond materials catalog lists MPCVD single-crystal and polycrystalline diamond plates and wafers up to 125 mm. For an XPDC replication, engineering support can be used to review sample geometry and crystal orientation.

View Original Abstract

It has recently been shown that X-ray parametric down-conversion (XPDC) provides access to high-energy polaritons, resulting from the hybridization of down-converted photons with electronic excitations in a nonlinear medium. Here, we present a spectrally resolved study of this effect around the K-shell absorption edge in diamond. Our results exhibit pronounced signatures of polaritonic hybridization, which we visualize by introducing a polariton spectral map and analyze by help of theoretical modelling. We find that the hybridization at this absorption edge results in substantially higher coupling strength than previously reported for a non-resonant case and reaches well into the strong-coupling regime. In addition, we demonstrate how our measurements of polaritonic XPDC allow us to extract the refractive index for bulk diamond at high spectral resolution around the carbon K-edge.