A preprint reports a switch in the relative surface-enhanced Raman signals of three related molecules. At 1406 cm⁻¹, the measured order was CMBT, then MBT, then EMBT; at 1463 cm⁻¹, it was EMBT, then MBT, then CMBT.
The study proposes an atom-resolved framework that combines atomic Raman tensors with Raman Intensity Densities, or RIDs—real-space maps of the calculated response—and Charge Density Differences, or CDDs. It uses them to relate atomic contributions to electronic-structure changes.
The measurements were made on Ag-island films, while the computational work used molecule–silver-cluster models. The study is a methods demonstration involving three MBT-family molecules and selected silver-cluster models.
What was measured
The experimental dataset comprised 1,500 MBT spectra, 3,000 EMBT spectra and 4,500 CMBT spectra. Researchers averaged the spectra, corrected their backgrounds, fitted features with Gaussian profiles and normalized the results over 500–3150 cm⁻¹.
Unsubstituted MBT was used as the reference, while CMBT and EMBT were compared with MBT and with each other. Calculated spectra were also compared with the experimental patterns.
The SERS setup used 7-nm silver-island substrates, and samples were held for 30 minutes in 10⁻³ M solutions. Spectra used 532-nm continuous-wave excitation, laser powers from 0.2 to 2 mW and acquisition times from 1 to 60 seconds.
In surface-enhanced Raman spectroscopy, or SERS, the silver substrate is part of the measured system. The analysis therefore treats the molecular response and the modeled substrate response as connected parts of the calculation.
Reading a Raman signal
The framework does not judge atomic contributions by magnitude alone. Their signs represent phase relative to the overall signal, so a contribution can reinforce or oppose the total response.
RIDs show how atomic motion is associated with changes in polarizability—the way the system responds to light—while CDDs connect those patterns to electronic structure. The authors use the combined view to interpret intensity differences through charge redistribution and phase relationships.
The modeled atomic pattern
In MBT, C15 had the largest atomic Raman intensity, N14 the second-largest positive contribution, and C11 and C12 also showed positive contributions interpreted as constructive coupling to the overall signal.
For CMBT at 1463 cm⁻¹, C11 and C12 had reversed CDD and RID patterns and negative atomic Raman intensities, alongside lower global intensity. EMBT showed the opposite pattern at that mode: C11 and C12 had dominant constructive RID areas and were described as important contributors to the global Raman signal.
The authors interpret these differences through molecule- and mode-specific phase patterns in the calculations. It is not presented as a general rule that a substituent will strengthen or weaken Raman scattering.
A caveat at 1406 cm⁻¹
For the 1406 cm⁻¹ CMBT peak, the text reports lower global Raman intensity than MBT and says that decreases in some polarizability modulations outweighed increases.
For EMBT, the discussion associates a lower Raman response with reduced polarizability modulation and electronic changes. It describes the ethoxy substituent's strong electron-donating +M effect as counteracting additional charge flow into the benzene π-system.
CMBT's 1406 cm⁻¹ feature contains two nearby mixed normal modes, and the analysis primarily assigns the peak to the first, higher-intensity mode. On that selected-mode comparison, the reported order was MBT, then CMBT, then EMBT.
The measured peak ranking and the selected-mode ranking are different comparisons: one describes experimental features, while the other isolates a mode within a feature that contains overlapping contributions.
Where the model agrees—and does not
Calculated spectra reproduced the reported intensity ordering for both peaks, but the EMBT calculation reversed the direction of the main-feature intensities. The authors considered that result adequate for relative normal-mode comparisons.
The agreement is best read as support for the proposed comparison within these models, not as full validation of the method; no quantitative uncertainty was reported for the mismatch.
The calculations used an Ag7 cluster and selected adsorption geometries, so the atom-level assignments are tied to those model choices.
Non-local RIDs extended into the silver cluster, which the authors interpreted as evidence that polarizability modulation in the SERS substrate contributes to the molecule's Raman intensity. The contribution was inferred from the computational RID model and was not independently quantified experimentally.
What remains untested
The evidence is limited to averaged SERS spectra from MBT, CMBT and EMBT on Ag-island films and calculations of molecule–silver-cluster models, with detailed interpretation focused on two peaks and their associated modes.
It does not establish how the framework performs for chemically or structurally diverse molecules, different substrates or other Raman and TERS configurations. Nor does it directly measure atom-resolved electron dynamics experimentally; the atomic patterns come from calculations matched to selected spectra and models.
RID maps depend on grid and origin choices, and projecting them onto ground-state electron density reduces some information. The modeled substrate contribution was not independently quantified experimentally.
The study reports no inferential statistical tests or uncertainty intervals; the spectra were averaged, background-corrected and normalized to a common spectral integral.
Further testing would need to examine additional molecules and substrates and determine whether non-local substrate contributions can be measured independently. The study does not claim universal accuracy across Raman, SERS or TERS systems.
Transparency and support
Additional information and molecular structures are presented in the Supplementary Information. The paper says scripts used to compute atomic Raman intensities and RIDs will be published with the article, but it does not identify a current repository or release.
The authors declare no competing interests. Acknowledgements list financial support for M.B. from Evangelisches Studienwerk e.V. Villigst, computing support through Baden-Württemberg's bwHPC and DFG, and Madrid support from MICIU/AEI funding, ERDF/EU and the Severo Ochoa program for Centers of Excellence in R&D.
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