Determination of Plasticizer Content in PVC by FT-NIR Spectroscopy

Determination of Plasticizer Content in PVC by FT-NIR Spectroscopy — Summary

Significance of the topic

Accurate measurement of plasticizer content in polyvinyl chloride (PVC) is critical for product performance, regulatory compliance and quality control across industries (toys, cables, films). Traditional approaches such as Soxhlet extraction followed by gravimetry are slow, solvent-intensive and susceptible to interference. Near-infrared Fourier transform spectroscopy (FT-NIR) combined with multivariate calibration offers a rapid, nondestructive alternative capable of delivering near real-time at-line or near-line results with minimal sample preparation.

Objectives and overview of the study

The study aimed to develop and validate FT-NIR methods to quantify the dioctyl phthalate (DOP) plasticizer in three PVC sample types: translucent plates, transparent plates and thin films. Key goals were to demonstrate method feasibility, produce robust partial least squares (PLS) calibration models, and compare performance metrics showing suitability for routine quality control.

Methodology

Samples and sample handling:

• Two calibration sets were prepared: 22 PVC plates (translucent and transparent) with DOP content spanning about 5–50% by weight and thicknesses ~0.57–1.18 mm; and 11 transparent PVC films with DOP content ~9–40% and thicknesses ~0.234–0.369 mm.
• Plates were kept separate to avoid DOP migration between samples.

Spectral acquisition and instrumental settings:

• Spectra collected on a Thermo Scientific Antaris FT-NIR analyzer, CaF2 beam splitter, InGaAs detector.
• Spectral range: 12000–3800 cm-1; resolution: 4 cm-1; 90 scans per sample.
• Reflectance mode with an integrating sphere for plates (sample placed on sphere window; internal gold flag used as background reference).
• Transmission mode for films using a transmission sample holder and an air background.

Data preprocessing and calibration:

• Preprocessing: multiplicative signal correction (MSC) and Norris second derivative smoothing (segment/gap settings tailored per sample type) to reduce scattering effects and baseline variation while controlling derivative-induced noise amplification.
• PLS regression was used for quantitative calibration due to complex spectral overlap and co-varying sample properties. Spectral regions were selected with software-assisted region selection.
• Model complexity: number of latent factors used—translucent plates: 4; transparent plates: 4; films: 3.

Used instrumentation

• Thermo Scientific Antaris FT-NIR analyzer
• CaF2 beam splitter
• InGaAs detector
• Integrating sphere accessory (reflectance measurements)
• Transmission sample holder (for films)
• Thermo Scientific TQ Analyst software for preprocessing, region selection and PLS calibration

Main results and discussion

Model performance:

• Translucent plates: calibration correlation coefficient 0.99965, RMSEC 0.375 (% DOP); cross-validation correlation 0.995, RMSECV 1.490 (% DOP).
• Transparent plates: calibration correlation coefficient 0.99931, RMSEC 0.591; cross-validation correlation 0.996, RMSECV 1.450.
• Films (transmission): calibration correlation coefficient 0.99983, RMSEC 0.175; cross-validation correlation 0.999, RMSECV 0.407.

Interpretation:

• All three calibrations showed excellent linearity and low calibration errors, with films producing the best predictive performance (likely due to stronger, less-scattered transmission signals and thinner pathlengths).
• Cross-validation errors are larger than RMSEC as expected, reflecting model generalization; values (particularly for plates) indicate that prediction uncertainty is on the order of ~1–1.5% DOP, which is adequate for many QC applications.
• Preprocessing (MSC + derivative) effectively compensated for thickness- and scatter-related spectral variation inherent in plate measurements, enabling robust PLS models despite broad, overlapping NIR bands.

Benefits and practical applications

• Rapid: spectral acquisition and prediction typically take less than one minute per sample, enabling high sample throughput and fast feedback to production.
• Non-destructive and reagent-free: eliminates lengthy solvent extraction and gravimetric steps, lowering cost and environmental impact.
• Minimal sample preparation: plates and films analyzed as produced, supporting in situ, at-line or near-line process control.
• Multicomponent potential: the approach can be extended to other additives or compositional attributes using appropriate calibration sets.

Limitations and practical considerations

• Model validity depends on representative calibration samples spanning expected variability (DOP concentration, thickness, stabilizer type, pigmentation, manufacturing variability).
• DOP migration between samples can alter apparent composition; physical sample handling and storage must minimize cross-contamination.
• NIR spectral features are broad and overlapping; robust chemometrics and careful spectral region selection are essential.
• Instrument-to-instrument model transfer and temperature/pathlength variations require additional standardization or transfer strategies.

Future trends and potential applications

• Extension to other plasticizers and polymer systems to build comprehensive additive-monitoring toolkits for polymer manufacturers.
• Integration of FT-NIR probes and analyzers for true inline monitoring of extrusion and film production lines for real-time process control.
• Development of standardized transfer protocols, calibration maintenance strategies and cloud-based model updates to support multi-site deployment.
• Advances in portable FT-NIR hardware and improved detectors could broaden field and on-site compliance testing.
• Use of advanced chemometric methods (e.g., domain adaptation, calibration update algorithms, machine learning approaches) to improve robustness against formulation and process drift.

Conclusion

The study demonstrates that FT-NIR spectroscopy combined with appropriate preprocessing and PLS calibration provides an accurate, fast and nondestructive method for quantifying DOP plasticizer in PVC plates and films. With sub-minute analysis times and competitive predictive performance, FT-NIR is a practical alternative to Soxhlet extraction/gravimetry for routine QC and at-line process monitoring, provided that robust, representative calibration sets and appropriate handling practices are implemented.

References

1. Thermo Fisher Scientific, Application Note 51593: Determination of Plasticizer Content in PVC by FT-NIR Spectroscopy, 2008.