FT-NIR for Online Analysis in Polyol Production

FT-NIR for Online Analysis in Polyol Production — Executive Summary

Importance of the topic

Hydroxyl value, acid number and residual monomer content are decisive quality attributes for polyols and polyester resins because they reflect polymer chain length, conversion and purity — parameters that directly affect downstream processing and final product performance. Rapid, reliable measurement of these attributes enables timely process control, reduces off-spec production, lowers chemical exposure and operating costs, and supports quality-management programs such as Six Sigma and TQM.

Objectives and overview of the study

This application note evaluates Fourier transform near-infrared (FT-NIR) spectroscopy (Thermo Scientific Antaris FT-NIR system) as an alternative to conventional titration and gas chromatography for routine and near-line quality control in polyol/polyester production. The study aims to demonstrate method precision and accuracy for hydroxyl value, acid number and residual ethylene oxide (EtO) across three distinct sample sets (surfactants, liquid polyols, and translucent polyester solids) and to illustrate practical sampling approaches for liquids and solids.

Methodology

Samples: Proprietary liquid and solid polyol/polyester samples collected from multiple batches and production stages.

Spectral acquisition:

• Spectral ranges: 4000–10,000 cm-1 and 4000–12,000 cm-1 (depending on set)
• Resolution: 8 cm-1
• Scans averaged: 50–100 co-added scans
• Pre-treatments: Norris/second derivative and Multiplicative Scatter Correction (MSC) for baseline and scatter correction
• Quantification algorithms: Partial Least Squares (PLS) and Stepwise Multiple Linear Regression (SMLR)

Calibration and validation: Reference values obtained by titration (hydroxyl value, acid number) and headspace GC (EtO). Model performance assessed by correlation coefficient (R), RMSEC (Root Mean Squared Error of Calibration), and RMSECV (Root Mean Squared Error of Cross-Validation) using leave-one-out protocols. Replicate precision quantified as %RSD.

Instrumentation used

An Antaris FT-NIR Method Development Sampling (MDS) System with the following sampling modules and accessories:

• Three-position heated transmission holder for 7 mm disposable vials (temperature control to ±0.1 °C; RESULT software integrates temperature control and equilibration delay)
• External heater for sample preheating to reduce equilibration times
• Tablet/Standard Tablet Transmission Module for translucent solid samples (simple placement of resin pieces on the platen and lowering of the detector)
• Thermo Scientific RESULT software for data collection and TQ Analyst for model development and prediction

Main results and discussion

Sample Set 1 (surfactants):

• Hydroxyl value model (SMLR using two wavelengths at 6854 and 8011 cm-1): R = 0.9999, RMSEC = 1.89 mg KOH/g, RMSECV = 2.59 mg KOH/g; reference range 70–350 mg KOH/g; replicate precision 0.55% RSD.
• Acid number model (3-factor PLS, 4500–8900 cm-1 after outlier removal): R = 0.9980, RMSEC = 0.129 mg KOH/g, RMSECV = 0.186 mg KOH/g; reference range 0.8–6.0 mg KOH/g; replicate precision 1.65% RSD.

Sample Set 2 (liquid polyols — hydroxyl and residual EtO):

• Hydroxyl value (SMLR, three points): R = 0.9994, RMSEC = 0.780 mg KOH/g, RMSECV = 0.951 mg KOH/g; reference range 23.4–116.0 mg KOH/g. Temperature during measurement controlled at 50 ± 0.1 °C to mitigate particulate effects.
• Residual ethylene oxide (SMLR, three points): R = 0.9999, RMSEC = 0.400 (units consistent with ppt scaling), RMSECV = 0.460; EtO range 1.78–100 ppt.
• Inter-parameter correlation between hydroxyl and EtO was low (R ≈ 0.624), indicating independent calibrations were feasible.

Sample Set 3 (translucent polyester solids):

• Hydroxyl value (single-term SMLR with denominator wavelength for pathlength compensation): R = 0.9995, RMSEC = 0.552 mg KOH/g, RMSECV = 1.99 mg KOH/g; range 7.1–62.6 mg KOH/g. Denominator wavelength compensated for variable sample thickness in solid tablets.
• Acid number (two-term SMLR with a denominator point): R = 0.9950, RMSEC = 1.31 mg KOH/g, RMSECV = 1.73 mg KOH/g; range 8.6–47.8 mg KOH/g. One primary wavelength corresponded to the weak second carboxyl overtone region; secondary wavelengths provided matrix compensation.
• Some datasets exhibited intercorrelation between hydroxyl and acid values; inclusion of critical early samples (e.g., prior to final acid charge) was essential to preserve calibration independence.

Benefits and practical applications of the FT-NIR method

• Speed: Rapid spectral acquisition (seconds to minutes) versus lengthy titration or GC procedures, enabling near-line or at-line decision-making.
• Multiparameter capability: Single spectral scans can support simultaneous quantification of hydroxyl value, acid number and residual monomer content.
• Reduction of hazardous chemistry: Eliminates routine use of titration reagents and associated disposal and exposure risks.
• Improved process control: Tight temperature control (±0.1 °C) and at-line sampling accelerate detection of reaction end-points and reduce off-spec production.
• Operational simplicity: Disposable 7 mm vials and straightforward tablet mounting support high throughput and reduced operator effort.
• Analytical performance: Correlation coefficients close to unity and RMSEC/RMSECV values at or below typical QC tolerances demonstrated comparable accuracy to titration and GC for the tested properties.

Future trends and opportunities

• Broader inline implementation: Integration of fiber-optic probes or flow-through cells to move from at-line to true online monitoring and closed-loop control.
• Model robustness: Expansion of calibration sets to include wider feedstock variability, seasonal batches and new chemistries to improve long-term robustness and reduce maintenance.
• Advanced chemometrics: Use of nonlinear algorithms, variable selection and transfer learning to improve sensitivity for minor components and facilitate model transfer between instruments/facilities.
• Digital integration: Connection of FT-NIR systems with plant DCS/PLC and LIMS for automated data capture, alarm generation and historical trending to support predictive maintenance and quality assurance.

Conclusion

The Antaris FT-NIR system provides a fast, precise and accurate alternative to titration and GC for routine QC of hydroxyl value, acid number and residual ethylene oxide in polyol and polyester production. Proper sampling, temperature control and chemometric model development (PLS, SMLR with derivative and scatter correction) yield high-performing calibrations suitable for near-line and at-line monitoring. Adoption of FT-NIR can reduce cycle times, operator exposure to hazardous reagents, and production waste while enabling more responsive process control.

References

• Thermo Fisher Scientific. Application Note 51594: FT-NIR for Online Analysis in Polyol Production. 2008.