Incoming analysis of sugars and polysaccharides

Importance of the Topic

Rapid and reliable identification of incoming raw materials such as mono- and disaccharides and polysaccharides is critical in pharmaceutical, food, and specialty chemical supply chains. Traditional wet-chemistry and chromatographic methods provide high specificity but are time- and resource-intensive. Near-infrared (NIR) spectroscopy offers a fast, non-destructive alternative capable of analyzing powders directly in glass vials or bags, enabling at-line or near-line verification that shortens release cycles and reduces operational costs while improving supplier quality control.

Objectives and Study Overview

This application note demonstrates an experimental approach to identify and assess the similarity of lots of glucose, lactose, sucrose, and cornstarch using FT-NIR spectroscopy. Goals included developing a robust classification model applicable across multiple plant sites and spectrometer units, demonstrating spectrometer-to-spectrometer calibration transferability, and establishing practical sampling and preprocessing steps for routine quality-control classification.

Methodology

Powdered samples of each carbohydrate type were measured directly through the bottoms of small glass sample vials using an integrating-sphere accessory on a Thermo Scientific Antaris FT-NIR Analyzer. Spectra were collected across the NIR region (instrument covers approximately 12000 to 4000 cm-1) with a spectral resolution of 4 cm-1 and a 60-second acquisition time per sample. Multiple spectra per sample were acquired to capture real-world variation such as particle size and packing effects.

Two spectral subregions were selected for model development because they contained the most distinctive spectral features separating the compound classes. Linear baseline correction was applied across each region to compensate for baseline shifts introduced by differences in powder packing and sample presentation.

A classification model was built in Thermo Scientific TQ Analyst software using principal component analysis (PCA). Each sample was measured twice on three different Antaris analyzers (total of 24 spectra). The full spectral set was reduced to five principal-component spectra (PCS) which served as the basis for the classifier. The model computes distances from each spectrum to the center of each class; a distance threshold can be set to flag ambiguous or off-spec samples.

Used Instrumentation

• Thermo Scientific Antaris FT-NIR Analyzer (FT-NIR, 12000–4000 cm-1 operational range)
• Integration Sphere Module for diffuse reflectance measurements through glass vial bottoms
• TQ Analyst software for preprocessing (linear baseline correction) and PCA-based classification modeling

Main Results and Discussion

The combined calibration data from three separate Antaris instruments produced a classifier that cleanly separated the four carbohydrate classes. In the 2D PCA representation the class clusters were tight and well separated. All calibration and validation spectra (24 total) were correctly classified. For each spectrum, the distance to the expected class center was substantially smaller than the distance to the next nearest class, indicating reliable discrimination and good margin for classification.

Example spectra show clear, reproducible NIR features for sucrose and other carbohydrates in the selected spectral regions. The study also demonstrates practical detection of contamination: a sucrose sample spiked with a small amount of glucose produced an increased distance to the sucrose class center that exceeded a predefined threshold and was therefore flagged as a potential problem.

Benefits and Practical Applications

• Non-destructive, rapid verification of incoming powdered materials without reagent use or complex sample prep.
• Ability to analyze samples through common packaging (glass vials), facilitating easy integration into receiving and QC workflows.
• Scalable calibration strategy that combines data from multiple instruments to improve transferability across plant sites and instrument units.
• Configurable distance thresholds enable automated screening and flagging of off-spec or contaminated lots for follow-up testing.

Future Trends and Applications

Expected developments and opportunities include:

• Improved instrument sensitivity and stability in newer FT-NIR models to enhance detection limits for low-level contaminants.
• Broader adoption of standardized multi-instrument calibration libraries to support enterprise-wide transferability and centralized model maintenance.
• Integration with laboratory information management systems (LIMS) and process control layers for automated release decisions and trend monitoring.
• Application of advanced multivariate algorithms and machine-learning approaches beyond PCA to increase robustness against complex matrix effects and variable particle size distributions.

Conclusions

This application demonstrates that FT-NIR spectroscopy with PCA-based classification can reliably identify and discriminate among common carbohydrates (glucose, lactose, sucrose, cornstarch) measured non-destructively through glass vials. The approach is fast, practical for routine QC, and transferable across multiple instruments when calibration data are aggregated. Distance-to-class thresholds provide a practical mechanism to detect contamination or mislabeling and trigger further analysis.

References

• Thermo Fisher Scientific. Application Note AN508223_E, June 2022. Antaris FT-NIR Analyzer — Incoming analysis of sugars and polysaccharides. Thermo Fisher Scientific.