top of page
Search

New Publication | Understanding SERS Spectral Shape Variability through Substrate Optics, Molecular Orientation, and Unsupervised Clustering

We are excited to share our latest publication (open-access) in The Journal of Physical Chemistry C, titled “Understanding SERS Spectral Shape Variability through Substrate Optics, Molecular Orientation, and Unsupervised Clustering.” The paper presents a combined experimental and chemometric study that explains why SERS spectra often change in shape, not just in intensity, even when the same molecule is measured on the same general type of substrate.


What We Did

Using 1,2-bis(4-pyridyl) ethylene (BPE) on oblique-angle-deposited silver nanorod (AgNR) substrates as a model system, we built a comprehensive SERS data set across six controlled experimental conditions: defect mapping, batch-to-batch variation, nanorod length tuning, concentration-dependent drop-casting, static immersion, and real-time immersion measurements. To analyze these spectra objectively, we applied hierarchical cluster analysis (HCA) and principal component analysis (PCA). HCA separated the spectra into seven reproducible clusters, including low- and high-signal-to-noise regimes, while PCA confirmed that the separation was driven by systematic redistribution of the characteristic BPE vibrational modes rather than random noise or baseline artifacts.  We then introduced intensity web plots using three different normalization strategies to isolate the physical origins of spectral variation: overall enhancement and analyte coverage, substrate-dependent optical reweighting, and orientation-selective enhancement referenced to the 1015 cm⁻¹ mode.  


⚡ Key Results

This study shows that SERS spectral shape variability is not random. Instead, it arises from physically meaningful and experimentally traceable factors, especially substrate optical response, electromagnetic enhancement strength, molecular adsorption orientation, and adsorption dynamics. A major outcome is that unsupervised clustering can successfully organize complex SERS data into physically meaningful groups. The clusters were strongly associated with specific experimental conditions such as substrate defects, nanorod geometry, analyte concentration, and immersion time. The work also shows that relying on a single peak ratio is often insufficient for interpreting SERS variability. Instead, spectral classification is governed by the collective redistribution of intensities across all five characteristic BPE modes, making spectral shape a genuinely multivariate feature rather than a simple two-peak comparison.


Why This Matters

SERS is widely used for ultrasensitive chemical and biological detection, but spectral variability often limits reproducibility, interpretation, and confidence in practical sensing. This work provides a physics-informed, data-driven framework for understanding where such variability comes from and how it should be interpreted.  More broadly, the study offers a transferable strategy for analyzing SERS data from heterogeneous substrates, low-concentration measurements, and dynamic adsorption environments. By linking chemometric classification with real physical mechanisms, this work helps lay the foundation for more robust, interpretable, and reproducible SERS measurements in chemical, biological, and environmental sensing.




 
 
 
bottom of page