Particle Correlated Raman Spectroscopy (PCRS): A Workflow for Correlating Particle Morphology with Chemical Identification

Particle-correlated Raman spectroscopy (PCRS) provides a practical way to link the physical characteristics of individual particles to their chemical signatures. The PCRS workflow pairs high‑resolution optical imaging with Raman spectroscopy, and it uses automated routines to locate particles, measure their size and shape, and collect Raman spectra. With specialized software, each particle’s image, coordinates, and Raman spectrum are connected and compared against reference libraries, creating a traceable dataset that ties morphology directly to chemical identity. Because PCRS accommodates wet, dry, and semi‑solid samples, it supports consistent, high‑throughput analysis across a wide range of particulate materials. The result is a reproducible, data‑rich analytical approach for laboratories requiring particle‑level characterization with both visual context and confident chemical identification. The analysis of particulate matter in pharmaceutical nasal spray formulations is presented as an illustrative application demonstrating the technique’s performance in a representative real‑world analytical context.

Introduction

In particulate analysis workflows, optical imaging and Raman microscopy are frequently applied as disconnected techniques.^1,2 Optical microscopy is commonly used to locate and size particles, while Raman spectroscopy is subsequently applied to a limited subset of manually selected targets, or to bulk particle populations.^3,4 This separation restricts analytical scalability and complicates interpretation of particle‑level chemical variability, as morphological, spatial, and chemical information are not systematically linked within a single analytical framework.⁵

Particle-correlated Raman spectroscopy (PCRS) addresses these limitations by integrating automated optical particle detection with spatially correlated Raman microspectroscopy. In the PCRS workflow, particles are first identified and localized by optical imaging, enabling the extraction of morphological descriptors such as size, shape, and spatial distribution. Raman microspectroscopy is then performed at the corresponding particle coordinates to obtain chemically specific spectral information.^1,2 The defining feature of PCRS is its particle‑centric data structure, in which optical morphology, spatial location, and Raman spectral data are permanently linked for each detected particle. This structured correlation enables chemically specific interpretation of particulate systems while preserving particle‑to‑particle variability and spatial context, supporting both targeted interrogation of individual particles and statistically meaningful population‑level analysis without reliance on assumptions of sample homogeneity.⁵

In this work, PCRS is presented as a general Raman microscopy framework for particle‑resolved chemical analysis. The underlying methodology, data structure, and analytical workflow are described in detail in the “Materials and Methods” section. Particulate matter generated from a pharmaceutical nasal spray formulation is included as an illustrative example of PCRS implementation, demonstrating the practical application of the framework rather than to emphasize a specific formulation or use case.

Materials and Methods

Sample Preparation for PCRS Analysis

PCRS is compatible with a wide range of particulate sample types, provided that particles can be presented on an optically accessible, Raman‑compatible substrate while preserving spatial separation and particle integrity. Sample preparation strategies are therefore selected based on the physical state of the material being analyzed, with the objective of minimizing sample manipulation while maintaining representative particle distributions.

Dry particulate samples, including powders and loose solids, are typically prepared by gentle dispersion onto optically flat substrates such as glass, quartz, or polished metal surfaces. Dispersion may be achieved by light tapping, air‑assisted deposition, or electrostatic methods to reduce particle agglomeration. Where necessary, low‑density spreading aids or antistatic measures may be employed to promote uniform particle distribution without altering particle chemistry.

Wet or suspension‑based samples may be prepared either by direct deposition or by membrane filtration prior to analysis, depending on particle concentration, matrix complexity, and the need to isolate particulates from dissolved components. For direct deposition, a controlled aliquot of the liquid sample is deposited onto the analysis substrate and allowed to evaporate passively under ambient conditions, enabling suspended particulates to settle naturally while preserving relative spatial relationships.

For dilute suspensions or environmentally relevant samples, particulate matter is commonly collected by filtration through optically and spectroscopically compatible membrane filters. Filtration concentrates particles from larger volumes and facilitates the removal of the bulk liquid matrix. Following filtration, filters are dried and analyzed directly using the PCRS workflow, or particles may be transferred from the filter to a Raman‑compatible substrate when required. Filter materials and pore sizes are selected to minimize Raman background contributions while ensuring efficient particle capture and retention.

Both preparation approaches preserve particle morphology and spatial integrity while enabling compatibility with automated optical detection and Raman microspectroscopic analysis. Volatile carrier solvents are preferred to reduce residual background; however, nonvolatile aqueous or organic matrices can be accommodated, provided that resulting spectral contributions do not interfere with particle identification.

Semisolid materials such as gels, creams, or viscous formulations are prepared either by thin‑film spreading or by dilution with a Raman‑transparent solvent prior to deposition. In cases where dilution is employed, care is taken to avoid chemical modification or selective loss of particulate components. Thin‑film preparation enables optical access to embedded particles while maintaining sufficient confinement to prevent particle migration during imaging.

Spray‑based samples, including aerosolized pharmaceutical and industrial formulations, are prepared by direct actuation or spray deposition onto the analysis substrate, consistent with established Raman approaches for pharmaceutical analysis.⁶ This deposition approach closely replicates real‑world use conditions and avoids filtration or mechanical capture steps that could bias particle populations. Following evaporation of the volatile carrier phase, residual particulate matter remains immobilized on the substrate and is analyzed directly.

Across all sample types, substrates are selected to provide optical flatness, low Raman background, and compatibility with automated microscopy. No chemical labeling, staining, or destructive pretreatment is required, preserving both particle chemistry and morphology for subsequent PCRS analysis.

PCRS Workflow Overview

The PCRS methodology follows a structured, image‑guided workflow designed to preserve particle‑level correlation between optical, spatial, and spectroscopic data throughout the analysis. A schematic representation of this workflow is shown in Figure 1.

The PCRS workflow comprises the following core operations:

1. Optical Image Acquisition – High‑resolution reflected‑light or transmitted-light images are acquired using brightfield, darkfield, or mixed‑contrast illumination modes selected, based on particle transparency, surface roughness, and refractive index contrast (Figure 2). Illumination geometry, numerical aperture, and exposure settings are optimized to maximize particle‑to‑background contrast while minimizing saturation and shadowing effects.

For samples extending beyond a single field of view, automated tile acquisition is performed using a calibrated motorized XY stage. Individual image tiles are stitched using feature‑based or coordinate‑based mosaic algorithms to establish a continuous, distortion‑corrected coordinate space (Figure 3). All image coordinates are retained in instrument‑native reference frames to ensure accurate downstream targeting.

2. Automated Particle Segmentation and Localization – Digital image processing routines are applied to the stitched optical images to identify discrete particulate features (Figure 4).⁷ Preprocessing steps may include background normalization, flat‑field correction, and contrast enhancement to compensate for illumination nonuniformity.

Particle segmentation is performed using a combination of intensity‑based thresholding, edge detection, morphological filtering, and connected‑component analysis. Over‑segmentation and particle merging artifacts are minimized through size, shape, and proximity constraints. Each segmented particle mask is assigned a unique identifier, centroid coordinates, and boundary definition, which are retained for subsequent morphology calculation and Raman targeting.

3. Quantitative Particle Morphology Extraction – From each particle mask, quantitative morphology descriptors are automatically calculated (Figure 5).⁵ These include projected area, equivalent circular diameter, Feret minimum and maximum dimensions, aspect ratio, perimeter, circularity, and ellipse fit parameters, such as major/minor axis lengths and ellipse ratio.

Morphological parameters are calculated independently of spectroscopic data and stored as structured metadata linked to the particle identifier. This separation enables unbiased statistical evaluation of particle populations and facilitates chemically resolved morphology comparisons during downstream analysis.

4. Raman Spectral Acquisition – Using the preserved optical coordinates, the microscope stage is repositioned with submicron precision to target individual particles for Raman analysis. Coordinate transformations account for differences between optical imaging and spectroscopic configurations, including objective offsets and stage calibration.

Raman acquisition strategies are selected based on particle size, morphology, and optical heterogeneity (Figure 6). For small or optically homogeneous particles, single‑point spectra are acquired at the particle centroid. Larger or heterogeneous particles may be interrogated using multipoint averaging or full-particle mapping to obtain chemically representative spectra while minimizing laser‑induced damage.^4,8

5. Spectral Preprocessing and Chemical Assignment – Raw Raman spectra are subjected to automated preprocessing routines to improve spectral quality and consistency (Figure 7). These routines include cosmic-ray detection and removal, baseline correction to address fluorescence contributions, and spectral normalization, where appropriate.

Processed spectra are compared against reference Raman libraries using similarity‑based matching algorithms, such as correlation or distance metrics, to assign chemical identities.⁹ Confidence scores or match quality metrics are retained alongside each chemical assignment, enabling transparent assessment of identification reliability. Chemically classified particles can be visually represented by recoloring particle maps based on assigned identity (Figure 8).

6. Particle‑Level Data Integration and Population Analysis – All particle‑resolved data—including optical images, segmented masks, morphology descriptors, spatial coordinates, Raman spectra, preprocessing history, and chemical assignments—are retained within a unified, particle‑centric dataset. This data structure preserves one‑to‑one correspondence between physical and chemical attributes for every detected particle.

The integrated data set supports interrogation at both the individual-particle and population levels. Statistical analyses can be performed on chemically grouped particles to evaluate size distributions, shape metrics, number fractions, and spatial distributions, as illustrated in Figure 9. Importantly, all population‑level results remain traceable to the underlying particle‑resolved data.

This workflow defines PCRS as a particle‑centric Raman microscopy framework, in which chemical identification is intrinsically linked to optically derived morphology and spatial context. Unlike bulk or point‑based Raman measurements, PCRS preserves particle‑to‑particle variability and enables statistically meaningful, chemically resolved interpretation of heterogeneous particulate systems.

Nasal Spray Sample Preparation and Analysis

For the illustrative application, two pharmaceutical nasal spray brands containing identical formulation components—fluticasone propionate and microcrystalline cellulose—were analyzed. Each formulation was actuated directly onto Raman‑compatible mirrored stainless‑steel substrates using the commercial spray device, a deposition approach that replicated normal use conditions while avoiding filtration or chemical pretreatment and minimizing spectral interference from conventional glass slides. Following evaporation of the aqueous carrier, residual particulate matter remained immobilized on the substrate and was analyzed directly using the full PCRS workflow.

PCRS Implementation

The PCRS workflow for the nasal spray sample described above was implemented on a Horiba XploRA™ PLUS Raman microscope equipped with high‑resolution optical imaging, an automated XYZ stage control, and particle‑targeted spectral acquisition capabilities, including ParticleFinder™ software and a corresponding reference database. Following wide‑area optical imaging and automated particle detection of n ≈ 10,000 particles over a 2 mm x 2 mm wide area, the retained particle coordinates were used to direct Raman spectral acquisition precisely to individual particle locations.

Instrument parameters, including excitation wavelength (532 nm), laser power (~0.7 mW), grating (1200 gr/mm), acquisition time (2 s), and number of accumulations², were selected to maximize spectral quality while minimizing thermal or photochemical effects. Spectra acquired from individual particles were subjected to standard preprocessing, including cosmic‑ray removal and baseline correction, followed by reference library‑based chemical identification.

Importantly, this implementation preserves the particle‑centric data structure, ensuring that optical images, morphology descriptors, spatial coordinates, Raman spectra, and chemical assignments remain fully correlated at the individual‑particle level throughout the analysis.

Results

Optical Particle Population

Optical imaging of the nasal spray deposits revealed a heterogeneous population of discrete particles distributed across the analyzed substrate areas, as shown in Figure 10 below. Particles spanned a broad size range and exhibited varied morphologies, including irregular aggregates and more compact features. While some particles appeared visually similar under optical microscopy, morphology alone was insufficient to infer chemical identity.

Raman‑Based Chemical Identification of Individual Particles

Raman microspectroscopy enabled chemically specific identification of individual particles detected within the nasal spray deposits. Across all analyzed samples, the particle population comprised multiple chemically distinct classes, including particles associated with active pharmaceutical ingredient (API) material, excipient‑derived particulates, and additional non‑API components. Chemically distinct particles frequently exhibited overlapping size ranges and similar optical appearance, emphasizing the limitations of morphology‑only particle analysis.

Chemically resolved particle size distributions for both brands are shown in Figure 11, where fluticasone propionate particles are highlighted in blue and microcrystalline cellulose particles in red. The average particle diameter of fluticasone propionate in Brand A was larger than in Brand B, with mean diameters of 4.6 μm and 3.6 μm, respectively. In addition, Brand A exhibited a broader size distribution (D10 = 2.6 μm, D90 = 7.0 μm) compared to Brand B (D10 = 2.3 μm, D90 = 5.2 μm), as summarized in Table 1. Similar trends were observed for microcrystalline cellulose particles, indicating formulation‑independent but brand‑specific differences in particle generation.

Beyond particle size, differences in particle shape were also observed. Chemically resolved ellipse ratio distributions for both brands are shown in Figure 12. Brand B displayed clearer separation between the ellipse ratio distributions of the API and excipient particles, while Brand A exhibited greater overlap between chemically distinct populations. These differences demonstrate that PCRS enables simultaneous evaluation of both chemical identity and morphology, revealing subtle inter‑brand differences that are not apparent through optical inspection or bulk‑averaged spectroscopic measurements alone.

Discussion

The results of the nasal spray study demonstrate the analytical value of PCRS as a general Raman microscopy methodology. The ability to chemically identify every detected particle without presupposing population uniformity enables both confirmation of expected formulation components and detection of chemically distinct materials within the same analysis.

By integrating automated particle detection with spatially registered Raman microspectroscopy, PCRS extends conventional point‑based Raman analysis into a statistically meaningful, image‑guided framework. This approach preserves morphological, spatial, and chemical context, which is critical for the interpretation of complex particulate systems.

General Applications of PCRS

Beyond the illustrative nasal spray example presented here, PCRS is broadly applicable to a wide range of particle‑analysis challenges where both chemical identity and particle‑level context are required. The particle‑centric nature of the PCRS data model makes it particularly well-suited to complex or heterogeneous particulate populations.

In pharmaceutical and biopharmaceutical development, PCRS can be applied to the analysis of extrinsic and intrinsic particulates in injectables, inhalation and nasal products, ophthalmic formulations, and lyophilized drug products. In these contexts, PCRS enables correlation of particle chemistry with size, morphology, and spatial distribution, supporting root‑cause investigations and formulation development.

In energy storage and materials science, PCRS is well-suited for the characterization of particulate carbons, conductive additives, and electrode materials in battery and supercapacitor systems. Individual particle analysis allows differentiation of chemically distinct carbon types, additives, or degradation products that may be morphologically similar under optical inspection.

For environmental and microplastic analysis, PCRS enables chemically specific identification of micro‑ and nano‑scale polymer particles while preserving particle size, shape, and distribution information. This capability supports studies of environmental contamination, particle aging, and source attribution without reliance on bulk averaging.

In forensic and trace analysis, PCRS provides chemically definitive identification of low‑abundance particles such as fibers, residues, and contamination particles while maintaining spatial traceability to the original sample context. The ability to interrogate individual particles without destructive preparation is particularly advantageous in evidentiary workflows.

Together, these examples illustrate that PCRS is not limited to a specific sample type or industry, but rather constitutes a general Raman microscopy framework for chemically resolved, single‑particle analysis across diverse scientific and industrial domains.

Conclusion

PCRS provides a robust and general framework for particle‑resolved analysis using Raman microscopy. By preserving one‑to‑one correlation between optical morphology, spatial location, and chemical identity, PCRS enables chemically resolved interpretation of particulate populations. The pharmaceutical nasal spray study presented here illustrates the practical implementation of PCRS and demonstrates how the framework can be applied to real particulate systems while remaining broadly applicable across scientific and industrial domains.

References

1. Ferraro, J. R.; Nakamoto, K.; Brown, C. W. Introductory Raman Spectroscopy, 2nd ed.; Academic Press: San Diego, CA, 2003.
2. Smith, E.; Dent, G. Modern Raman Spectroscopy: A Practical Approach; John Wiley & Sons: Chichester, UK, 2005.
3. Long, D. A. The Raman Effect: A Unified Treatment of the Theory of Raman Scattering by Molecules; Wiley: Chichester, UK, 2002.
4. Everall, N. J. Confocal Raman Microscopy: Common Artifacts and Practical Limitations. Appl. Spectrosc. 2009, 63, 245A–262A.
5. Harris, R.; Pal, R. Particle Size Analysis in Pharmaceutical Formulations: Methods and Limitations. Adv. Drug Delivery Rev. 2010, 62, 203–218.
6. Lau, A. K. S.; Shurvell, H. F. Raman Spectroscopy in Pharmaceutical Analysis. Appl. Spectrosc. Rev.1997, 32, 409–454.
7. Liang, C.; Nguyen, T. T.; McCreery, R. L. Automated Raman Spectral Analysis Using Library Matching and Multivariate Methods. Appl. Spectrosc. 2016, 70, 1943–1956.
8. Schmitt, M.; Popp, J. Raman Imaging and Mapping of Biological and Pharmaceutical Samples. J. Raman Spectrosc. 2006, 37, 20–28.
9. S.T. Japan Europe GmbH, Raman Spectra Databases—Comprehensive Raman Reference Libraries for Material Identification. S.T. Japan Europe GmbH. https://www.stjapan.de/products-1/spectra-databases/raman-spectra-databases/ (accessed Apr 28, 2026)