Raw materials qualification within a workflow: FT-NIR analysis using the Antaris II Analyzer

Raw materials qualification within a workflow: FT-NIR analysis using the Antaris II Analyzer — Application Note Summary

Significance of the topic:
Reliable and rapid raw material identification and physical characterization are essential front-end steps for any Process Analytical Technology (PAT) strategy in pharmaceutical and chemical manufacturing. Fast, non-destructive analytics at goods-in reduce the risk of incorrect materials entering production, speed release decisions, enable traceability and minimize waste compared with slower laboratory methods such as titration or HPLC.

Objectives and overview of the study:
This application note demonstrates integrating a Thermo Scientific Antaris II FT-NIR analyzer with RESULT software to create an automated, decision-driven workflow for raw material qualification. The workflow covers material identity confirmation, physical property discrimination (example: particle size), and automated reporting/archival with DCS/LIMS connectivity. The aim is to show a practical model for front-end PAT where spectroscopy and chemometrics enable rapid pass/fail decisions at the loading dock or warehouse without specialized spectroscopy expertise.

Methodology and experimental approach:

• Sampling approach: Spectra were acquired non-destructively through drum liners using a SabIR diffuse-reflectance fiber probe attached to the Antaris II, enabling analysis without opening packaging.
• Workflow design: RESULT software executes a sequence of events (Request → Measure → Perform If → archive/report). Inputs to the Request event can be manual IDs, barcode scans, text files, or process control system data (DCS/SCADA).
• Chemometric strategy: Unknown spectra are evaluated against multiple libraries (training sets) in parallel using qualitative algorithms available in TQ Analyst and RESULT: Similarity Match, QC Compare, Distance Match and Discriminant Analysis (including SIMCA and Mahalanobis distance outputs).
• Preprocessing: Data conditioning included Multiplicative Scatter Correction (MSC), second-derivative transformation, and Norris smoothing (segment length 9, gap 3). Known spectral contributions from polyethylene liners (approximately 4200–4350 cm−1 and 5650–5800 cm−1) were excluded from analysis to reduce container matrix interference.
• Decision logic: If a library match yields a PASS, the workflow proceeds to optional particle-size analysis (only run when identity matches lactose in this example). FAIL outcomes trigger archival, quarantine workflows, and notifications to inventory/LIMS.

Used instrumentation:

• Thermo Scientific Antaris II FT-NIR Analyzer
• SabIR diffuse-reflectance fiber probe for through-container sampling
• Thermo Scientific RESULT workflow software
• TQ Analyst chemometric software (SIMCA, Discriminant Analysis, Similarity, QC Compare)
• Cordless barcode reader for automated sample metadata capture
• Integration interfaces: OPC connectivity to DCS and links to LIMS/inventory systems

Main results and discussion:

• Material discrimination: The Antaris II plus chemometrics successfully distinguished multiple raw material classes using Discriminant Analysis and other library-based algorithms. Visualizations (scores plots on principal component axes and second-derivative spectra) illustrated class separation and within-class variability.
• Container interference management: Removing polyethylene liner bands and using preprocessing (MSC, derivatives, smoothing) effectively minimized container-related spectral features, allowing robust identification through packaging.
• Particle size discrimination: Two particle-size classification methods were developed: (a) Discriminant Analysis on raw spectra using SIMCA (typically requires ≥5 standards per class), and (b) second-derivative preprocessing with Norris smoothing. Both methods used Pharmatose lactose samples with mesh sizes of 50, 80, 90, 100, 110 and 125 µm and achieved discrimination of particle-size-related spectral differences once identity was confirmed.
• Automated decision-making and traceability: RESULT workflow issued PASS/FAIL outcomes and, on PASS, archived results and reported them to operators and DCS via OPC. On FAIL, the system archived failed spectra separately, flagged the loading location, and triggered quarantine/ supplier notification procedures.

Benefits and practical applications of the method:

• Speed and non-destructiveness: FT-NIR provides near-instant chemical identification without consuming the sample, enabling materials to proceed into process immediately when passed.
• Minimal operator training: The workflow and barcode-driven data capture allow warehouse staff to run qualified analyses without spectroscopy expertise.
• Flexible automation: RESULT workflows combine data acquisition, chemometric analysis, logical decision-making and integration with plant control and quality systems for streamlined materials management.
• Improved supply-chain control: Early detection of non-conforming materials reduces production risk, supports quarantine procedures, and enhances supplier feedback loops.

Limitations and practical considerations:

• Chemometric dependency: Robust classification requires representative training sets and appropriate preprocessing; some methods (e.g., SIMCA) demand multiple standards per class.
• Container and matrix variability: Although container bands can be removed, unexpected liner materials or strong matrix effects may complicate identification and require additional model development.
• Scope of quantitative claims: The note focuses on qualitative classification and particle-size discrimination; full quantitative assays (e.g., APIs or degradants at low levels) may require supplementary calibration and validation.

Future trends and possible applications:

• Tighter PAT integration: Expect broader implementation of NIR at goods-in combined with in-line and at-line probes for continuous monitoring and real-time release strategies.
• Advanced chemometrics and machine learning: More sophisticated algorithms, transfer learning and model update automation will reduce model maintenance and improve robustness across sites.
• Digital lab and factory connectivity: Deeper integration with LIMS, ERP and MES using standardized protocols (OPC UA, APIs) will enhance automated quarantining, traceability and regulatory record-keeping.
• Miniaturization and multiplexing: Smaller, ruggedized NIR probes and multi-point sampling will expand applicability in warehouses and confined loading areas.
• Regulatory acceptance and standardization: Increased use of validated spectroscopic workflows will support regulatory pathways for PAT-enabled release and reduced end-product testing.

Conclusion:
The Antaris II FT-NIR analyzer combined with RESULT workflow software and TQ Analyst chemometrics provides an effective, automated solution for front-end raw material qualification. The platform enables rapid, non-destructive identity checks, conditional physical-property analysis (e.g., particle size), and integrated reporting to plant control and quality systems, reducing risk and speeding material disposition at goods-in. Proper model development, representative training sets and attention to container effects remain important for reliable deployment.

Reference:

• Application note authored by Jeffrey Hirsch, Ph.D., Thermo Fisher Scientific — Antaris II FT-NIR Analyzer raw materials qualification workflow (Thermo Fisher Scientific, AN51088_E 03/22M).