NIR and Raman: Complementary Techniques for Raw Material Identification

Importance of the Topic

Raw material identification (RMID) is a critical quality-control activity across pharmaceutical, food and chemical manufacturing. Rapid, reliable, and non-destructive identification at receiving and throughout production reduces the risk of out-of-specification product, prevents material waste, accelerates operations and supports Process Analytical Technology (PAT) objectives. Vibrational spectroscopy — specifically near-infrared (NIR) and Raman — meets key RMID needs by enabling fast, non-destructive testing that can be performed through common sample containers with minimal operator training.

Objectives and Study Overview

This technical note evaluated complementary use of Thermo Scientific Antaris II FT-NIR and DXR SmartRaman instruments for RMID. The goals were to: establish a rapid workflow for routine screening, determine limitations of each technique, and show how Raman can resolve NIR-challenging cases. A representative set of 27 common pharmaceutical raw materials (excipients, active ingredients, lubricants and salts) was analyzed to demonstrate practical identification performance in a process environment.

Methodology

• Sample set: 27 materials chosen to represent a broad cross-section of pharmaceutical raw materials (sugars, polysaccharides, hydrated and anhydrous salts, stearates, APIs, polyethylene, talc, etc.). Samples were measured through polyethylene bags to simulate non-invasive, at-dock testing.
• NIR acquisition: Thermo Scientific Antaris II FT-NIR with integrating sphere; spectral range 10,000–4,000 cm-1; 16 scans per sample; 4 cm-1 resolution. Spectra were preprocessed by multiplicative signal correction (MSC) for pathlength effects, first derivative to reduce baseline offsets, and Norris smoothing (segment 9, gap 7). A Discriminant Analysis chemometric model was built using Thermo Scientific TQ Analyst.
• Raman acquisition: Thermo Scientific DXR SmartRaman with 780 nm laser on the Universal Platform accessory. Automated features (autoexposure, Smart backgrounds, automated alignment) were used; autoexposure adjusted parameters to reach S/N ≥ 100 with acquisition times up to two minutes. Raman identifications relied on spectral library searching with match scores (0–100).
• Performance metrics: For the NIR chemometric method, Mahalanobis distance to the nearest incorrect class was used as a diagnostic of classification robustness. For Raman, library match scores quantified identification confidence.

Instrumentation Used

• Thermo Scientific Antaris II FT-NIR analyzer with integrating sphere (for sampling through bags/containers).
• Thermo Scientific DXR SmartRaman spectrometer with 780 nm excitation and Universal Platform sampling accessory.
• Software: Thermo Scientific RESULT (NIR operation) and TQ Analyst (chemometric model development and Discriminant Analysis).

Main Results and Discussion

• NIR screening: Most of the 27 materials were rapidly and correctly classified by the Antaris II/TQ Analyst workflow. Typical NIR acquisition required approximately 15 seconds per sample, and the chemometric model delivered clear separations for many classes as quantified by large Mahalanobis distances to the nearest incorrect class.
• Failure modes for NIR: Four materials showed elevated risk of misclassification (Mahalanobis distance to nearest incorrect class <3). Calcium carbonate was a clear failure case: its NIR spectrum had negligible material-specific features, and the recorded signal originated primarily from the polyethylene bag. Other challenging materials included anhydrous polyatomic ionic salts and hydrophobic materials whose spectral signatures overlap container materials (e.g., magnesium stearate vs. polyethylene).
• Raman follow-up: All four NIR-problematic samples were correctly identified by DXR SmartRaman via library searching. Reported spectral match scores were high: calcium carbonate 95.5, magnesium sulfate 96.3, manganese sulfate 99.4, magnesium stearate 88.2. Raman provided sharper, well-resolved peaks (sensitive to non-polar bonds and C–C backbone features) enabling robust library matches even when NIR information was poor.
• Complementarity: The methods target different vibrational modes (NIR tends to probe polar bond overtone/combination bands; Raman probes non-polar bond polarizability changes). NIR advantages include deeper penetration, larger sampling volume (reducing sampling error), and lower sensitivity to container interference in many cases. Raman offers better peak resolution and direct chemical specificity, and can penetrate optically clear containers (glass/plastic), making it preferable for water-containing samples or when sharper spectral features aid library matching.
• Representative diagnostic visuals summarized: instrument photos, example raw NIR spectra, Mahalanobis distance plots (showing clear identification vs. misclassification scenarios), overlap of CaCO3 and polyethylene bag spectra in NIR, Raman vs library spectral match for CaCO3, and a qualitative suitability chart mapping sample types to NIR vs Raman preference (e.g., grains/tablets favor NIR; pure solids, non-polar materials and aqueous/unknown samples favor Raman).

Benefits and Practical Applications

• Operational workflow: Use NIR as the primary, rapid screening tool for incoming materials (fast, easy, minimal training). Deploy Raman as a secondary, confirmatory technique for samples that are ambiguous or known to be problematic for NIR.
• Non-destructive sampling: Both methods can analyze through containers, supporting archiving and reducing hazardous exposure.
• Process integration: The combined approach supports PAT initiatives by enabling at-line and near-line checks (loading dock to production line), minimizing hold-ups and material misuse.
• Economic impact: Faster identification lowers labour costs, reduces material waste and prevents process downtime associated with incorrect raw materials.

Future Trends and Potential Uses

• Integrated workflows combining NIR chemometrics and Raman library search in a unified RMID decision engine, with automated triage (screen with NIR, route ambiguous cases to Raman) and operator guidance.
• Expansion of comprehensive, curated spectral libraries (cloud- or enterprise-hosted) and use of advanced machine learning models to improve discrimination between closely related materials and detect out-of-library items.
• Advances in portable/handheld NIR and Raman instruments with automated calibration and simplified user interfaces will broaden at-line deployment and enable routine checks by non-specialist staff.
• Higher fidelity preprocessing, robust container-interference compensation algorithms, and hybrid chemometric + spectral-matching approaches to reduce false positives/negatives for weakly absorbing materials (e.g., some inorganic salts).
• Integration with manufacturing execution systems (MES) and LIMS for automatic logging, traceability and real-time QC decision-making.

Conclusion

Combining FT-NIR screening with targeted Raman follow-up provides a practical, efficient RMID strategy. NIR delivers rapid, operator-friendly screening with broad applicability, while Raman resolves cases where NIR lacks discriminating power — particularly for non-polar materials, inorganic salts with weak NIR features, and aqueous or unknown samples. A workflow that employs NIR first and Raman for ambiguous or difficult samples yields robust identification, supports PAT goals, and reduces operational risk and cost.

References

• Strother T. NIR and Raman: Complementary Techniques for Raw Material Identification. Thermo Fisher Scientific Technical Note 51768, 2009.
• Instruments discussed: Thermo Scientific Antaris II FT-NIR and Thermo Scientific DXR SmartRaman.