ADVANCED PARTICLE ANALYSIS
Identify an unknown pharmaceutical particle. Count, size and characterize complex populations. SizeID.bio combines advanced microscopy and Raman spectroscopy with DeepMorph™ image-based Deep Learning, automated RamanMetrix® spectral Machine Learning, pharmaceutical references and process expertise—all within controlled, independently reviewed investigations.
Have an unknown particle? Explore particle identification →
From sample to scientific evidence
Seven steps connect your analytical question to a reviewable result. The workflow is modular: preparation, controls and analysis are selected for the sample and the decision the results need to support.
Sample Preparation → Particle Analysis → Interpretation & Reporting

STEP 01
Understand the analytical challenge
Define the investigation, its intended use and the decision it must support. Consider the product, process, sample history and available reference materials.

STEP 02
Sampling & sample preparation
Assess the matrix, particle stability, background and potential losses before choosing a collection and handling strategy.

STEP 03
Blanks, references & system suitability
Agree blanks, references, system checks and study-specific acceptance criteria before acquisition. Controls are part of the analytical method from the start.

STEP 04
Microscopy & selected-particle Raman
Count and size particles by calibrated microscopy. Where chemical information is needed, acquire Raman spectra from selected targets and link the evidence at particle level.

STEP 05
Chemical populations & advanced interpretation
Compare defined chemical populations and reference materials. Apply Raman spectral ML where appropriate to differentiate materials and investigate probable sources.

STEP 06
Chain of evidence & reporting
Connect the sample, preparation, particle records and interpretation. A separate technical reviewer and independent QA function review the report before controlled release.

STEP 07
Discuss findings & next steps
Review what the evidence supports, what remains unresolved and which next step will help answer the analytical question.
Different samples demand different approaches
Preparation should serve the question without obscuring the particles of interest. The matrix, particle stability, background and possible losses guide the choice of handling strategy.

Filtration & direct particle analysis
For suitable liquids, filtration collects particles on an analytical substrate for direct imaging, counting and Raman identification on that same substrate. Membrane background, preparation effects and potential collection or handling losses are evaluated for the sample.

Component rinsing & extraction
Controlled rinsing or particle-extraction procedures collect material released from packaging and device components. The release conditions, blanks, references and collection strategy are defined for the investigation.

DPI / MDI sampling
DPI/MDI, collected-dose and impactor-related investigations need suitable collection substrates and transfer procedures. Deposition, collection efficiency and blanks are considered before results are interpreted; method suitability is established for the project.

Suspension analysis
Nasal suspension characterization provides an application context. Direct or in-situ analysis depends on matrix and cell suitability, background, particle movement, stability and concentration.
Visible rejects & individual particle investigations
A single unknown particle can be an appropriate starting point. We agree how to handle or isolate the particle, acquire chemical evidence and compare relevant references in the context of the original finding.
The sample matrix and intended use guide preparation, controls and the assessment of collection and handling effects.
Explore nasal and inhaled product applications →
Microscopy, Raman and two distinct AI capabilities
GramRay i MK3
GramRay i MK3 combines automated microscopy, calibrated particle morphology and selected-particle Raman analysis. Optical detection and characterization reach approximately 500 nm under suitable measurement conditions. That optical capability does not imply universal Raman identification at 500 nm.
Optical Imaging & Morphology → DeepMorph™ image-based targeting (where useful) → Raman Spectroscopy → RamanMetrix® spectral classification (where useful) → Scientific Interpretation

Microscopic Particle Counting & Sizing
Independent counting and sizing studies deliver particle counts, size distributions, calibrated 2D morphology and size-bin-specific statistics. Concentration is reported where supported by sampling and preparation. Raman adds chemical assignments when required; it is not a prerequisite for optical measurements.
DeepMorph™ — Image-based Deep Learning
Established applications include AECP recognition and targeting in nasal sprays and image-based API candidate recognition. DeepMorph™ analyzes optical particle images to select relevant morphologies; Raman provides separate chemical confirmation. Further models can be rapidly developed and adapted for suitable customer applications, with application-specific qualification.

Raman measurement & identification
Spectra from selected targets are evaluated against suitable libraries and reference materials. A chemical assignment depends on spectral evidence and agreed criteria; some measurements may remain unresolved.
RamanMetrix® — Automated Raman Spectral ML
RamanMetrix® is integrated into the instrument analysis workflow for automated Raman spectral classification. Multiple established models are available, with cellulose-related classification used routinely. Additional application-specific models can be developed and integrated for customer investigations.
Explore DeepMorph™ image-based Deep Learning →
Explore GramRay i MK3 instrument technology →
Established DeepMorph™ examples are distinct from models requiring further development: DPI agglomerates, agglomerates containing engineered API particles, spray-dried API particles and other complex pharmaceutical structures are potential application-specific targets. Model development and qualification are agreed for the customer application.
Measured evidence: API size and foreign-particle chemistry
API-specific particle-size distribution

Generic fluticasone nasal suspension, measured with GramRay i MK3 and a 50× objective. The distribution covers 2,671 Raman-confirmed, size-calibrated API particles. Median circular-equivalent diameter (D50): 2.35 µm.
Open the full-resolution API particle-size distribution →
Read the nasal-suspension application context →
Foreign-particle material identification
Supplier A · 156 Raman-analyzed particles. PET was the dominant material in this subset.
PET · 126 · 80.8%
Polystyrene · 17 · 10.9%
Epoxy · 7 · 4.5%
Cellulose · 4 · 2.6%
Polyethylene · 1 · 0.6%
Unknown ester · 1 · 0.6%
Shares describe the Raman-analyzed subset. They do not estimate the complete optical population.
Particle counts, size bins & chemical subsets
Quantitative outputs include size-bin statistics, distributions, sample and reference comparisons, material-specific populations and suitable multivariate analyses. Selected Raman targets come from the full optical population; accepted and inconclusive or unavailable assignments are outcomes within that subset. Chemical percentages use the relevant assigned or measured denominator, stated with the result.
Pharmaceutical references & process expertise
SizeID.bio maintains an extensive collection of pharmaceutical material, packaging and contaminant references: primary packaging, elastomeric stoppers and seals, blister layers, adhesives, capsules, inhaler and device components, polymers, excipients and process materials. Manufacturing and drug-delivery expertise helps select meaningful comparisons. A DPI investigation, for example, may compare capsule, blister, adhesive, housing or sealing materials with unknown particles. Chemical fingerprints, customer-supplied references and process information support probable-source investigation; representative sampling and discriminating references determine the strength of an attribution.
Foreign particle investigations →
Population and reference comparisons →
BEYOND IDENTIFICATION
Beyond Spectral Matching: RamanMetrix® Chemical Fingerprinting
Identifying a particle’s chemical material is often only the first step. Automated RamanMetrix® classification complements conventional library matching when spectral correlation alone cannot sufficiently distinguish related materials. Established models and relevant pharmaceutical references add depth to material differentiation and probable-source investigations.
1 · Spectral similarity
Conventional library matching, including Pearson correlation, remains a relevant way to assess spectral similarity. An assignment depends on spectral quality, suitable references and defined acceptance criteria.
2 · Automated RamanMetrix® classification
RamanMetrix® automates classification of Raman spectra within the instrument analysis workflow. Multiple established models support defined reference groups; cellulose-related classification is already used routinely. PCA and other suitable multivariate methods support the analytical interpretation.
3 · Hierarchical Raman classification
Established models can organize defined materials into families, subgroups and reference classes. New customer-specific Raman models can be developed and integrated, with model evaluation and qualification or validation matched to the intended use. Availability of established models does not imply coverage of every material class.
4 · References & probable-source investigation
Compare unknown Raman fingerprints with customer-supplied reference materials, then interpret the result with sampling and process evidence. Adequate references and evaluated models can support probable-source investigation; a class assignment alone does not establish an exact source.
DeepMorph™ uses optical images for morphology recognition and targeting. RamanMetrix® uses Raman spectra for automated chemical classification. They are distinct, complementary technologies—not one combined optical/Raman classifier. Source interpretation still depends on spectral quality, representative references, model evaluation and process information.
Explore reference-based foreign particle investigations →
RamanMetrix® software is provided by Biophotonics Diagnostics GmbH.
Method development and traceable results
The intended use guides sample preparation, acquisition and evaluation. Calibration and system suitability checks support the study. Blanks establish the background, and relevant product or packaging references help set identification criteria.
For recurring measurements, we agree acceptance criteria and assess qualification or product- and matrix-specific validation needs. Reproducibility, robustness and ongoing suitability are evaluated as required.
The evidence record links particle images, coordinates and morphology to Raman spectra, classifications, instrument settings, references and software or model versions. Chain of custody records physical sample handling and transfers separately.
Discuss method-development support →
Quality framework and documented practices →
Explore the scientific examples and their context on our application pages:
Particle images and associated Raman evidence →
Population and reference comparison examples →
Nasal suspension and API characterization →
Discuss your analytical challenge
Advanced Particle Analysis—From Microscopic Counting to Chemical Identification and Source Investigation. Tell us about your sample, suspected materials and the decision you need to make.
Explore our services →