A new preprint presents a computational framework that makes hidden cancellations inside a Raman peak visible. It breaks the calculated Raman response into contributions from individual atoms and treats the sign of each contribution as phase information. Under that interpretation, atoms can produce strong local changes in polarizability, yet the overall signal can remain modest when those changes oppose one another.
The approach is meant to bridge a gap between a Raman intensity and the local electronic changes behind it. Its Raman tensor, a mathematical description of how the system's polarizability changes as a particular molecular vibration moves, is decomposed into atomic terms. The authors also use Raman Intensity Densities, or RIDs, to map that modulation through real space. Atomic RIDs are built from individual atom displacement vectors, sum to the global RID, and are paired with charge-density differences that show the calculated electron-density change for a normal mode.
The comparison behind the model
To examine the idea, the authors compared surface-enhanced Raman scattering, or SERS, spectra from three related molecules: MBT, 5-chloro-2-mercaptobenzothiazole, or CMBT, and 6-ethoxy-2-mercaptobenzothiazole, or EMBT. The experiments used 7 nm silver-island films. Each molecular sample was held in solution for 30 minutes. The reported collection included 1,500 MBT spectra, 3,000 EMBT spectra and 4,500 CMBT spectra. Spectra were averaged, background-corrected, fitted with Gaussian peaks and normalized to the spectral integral from 500 to 3150 cm-1.
The rankings of two peaks changed with the molecule. At 1406 cm-1, the experimental order was CMBT higher than MBT, which was higher than EMBT. At 1463 cm-1, EMBT was highest, followed by MBT and CMBT. The calculations reproduced those relative sequences, but reversed the main-feature intensities for EMBT. The 1406 cm-1 CMBT feature contains two nearby mixed normal modes; when the first mode is assigned for comparison, the calculated mode-intensity order becomes MBT, CMBT, then EMBT.
What the atom-level numbers add
The displayed atomic-intensity entries put the 1406 cm-1 mode at 1.24 for MBT, 1.16 for CMBT and 0.25 for EMBT. At 1463 cm-1, the corresponding entries were 0.18, 0.10 and 1.05. At 1406 cm-1, the C15 entries were 0.57, 0.62 and 0.16 across MBT, CMBT and EMBT, while the N14 entries were 0.40, 0.39 and 0.10. The numbers were scaled for presentation and were not accompanied by uncertainty estimates.
The authors use those signs to add a second layer of meaning. In their framework, an atomic-intensity magnitude represents the strength of a local polarizability modulation, while its sign represents phase. Opposite signs mark counteracting contributions, so strong local terms can cancel into a low net polarizability change even while the total global intensity remains positive. These atom-level phase assignments come from calculations, not from direct atom-specific measurements.
The substrate is part of the calculation
The model also places part of the response in the substrate. For MBT's 1406 cm-1 mode, the RID extended from the molecule into an Ag7 cluster, a modeled group of seven silver atoms. The paper reports this as a substrate response associated with the molecular vibration and contributing to the molecule's Raman intensity. That finding depends on the selected Ag7 adsorbate model, rather than a full extended experimental silver surface.
To generate the calculated spectra, the authors used Gaussian16 and Ag7 adsorbate models with CAM-B3LYP, D3 dispersion with Becke-Johnson damping, SDD for silver and cc-pVTZ for the other elements. Harmonic frequencies were scaled by 0.95463, and the calculated spectra were broadened with a 20 cm-1 Gaussian full width at half maximum.
A proposed explanation for the chlorine case
One proposed explanation appears in the CMBT result at 1406 cm-1. The authors identify enhanced C15 and N14 polarizability modulation alongside reduced C11 and C12 charge-density oscillations. They interpret that combination in terms of the chloro substituent's -I, or inductive, effect, reduced mesomeric donation and altered mode displacement. In their accounting, the reductions outweighed the enhancements elsewhere, leaving a lower net 1406 cm-1 response than in MBT. Because this CMBT feature contains mixed modes, the explanation remains tied to the stated mode assignment.
What remains untested
The study's evidence remains tightly bounded. It combines SERS measurements for three MBT-family derivatives with quantum-chemical calculations built around selected Ag7 adsorbate models. No independent validation across other molecules, substrates or Raman modalities was reported. The 5-CMBT and 6-EMBT comparison also mixes electronic and structural substituent effects, so the calculations do not isolate those influences.
Interpretation of the measurements has limits as well. Spectra were averaged over extended measurements affected by setup misalignment, and the reported work did not provide replicate variability or measurement uncertainty. Normalization to a 500 to 3150 cm-1 spectral integral limits absolute intensity comparisons between molecules. The RID and related maps are origin-dependent, and their projection onto ground-state electron densities reduces information.
Taken together, the preprint offers a way to connect changes in Raman intensity with local charge redistribution and phase-sensitive atomic interference. It does not establish that the calculated atomic contributions can be measured directly, that the selected Ag7 geometry is the unique real-surface arrangement, or that the framework generalizes across Raman and SERS systems. Further tests would need additional molecules, adsorption geometries and extended metal surfaces, along with experiments that can independently check the calculated atomic phase assignments.
Paper data and sources
Original title: Visualizing and Quantifying Atomic Contributions to Raman Intensities governed by Spatially-Resolved Atomic Interferences
Authors: Marc Broeckel, Johannes Gierschner, Alfred J Meixner, Kai Braun
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-25
DOI: Not available
Original paper · Full text