NIR Model Transferability Using Binary Mixtures of Talc in Iron Sulfate and Water in Ethanol

Importance of the topic

The transferability of near-infrared (NIR) calibration models between instruments is a practical bottleneck for routine deployment of FT-NIR across pharmaceutical, polymer and chemical production environments. Direct transfer without empirical post hoc corrections reduces operational cost and complexity, speeds implementation, and simplifies regulatory justification. This study evaluates straightforward method transfer across multiple Thermo Scientific Antaris FT-NIR analyzers using two chemically and spectrally contrasting model systems: a low-absorbance solid system (talc in ammonium iron(III) sulfate) and a strongly absorbing, concentration-dependent liquid system (ethanol/water mixtures).

Objectives and study overview

This work aims to determine whether calibration models developed on a primary Antaris FT-NIR instrument can be applied directly to vector instruments without algorithmic corrections. Two representative model problems were selected to probe limits of transferability: (1) quantification of talc (sharp, non-shifting low-intensity bands) in an inert solid matrix, and (2) quantification across the full concentration range of water in ethanol (broad, strong bands that shift with changing hydrogen-bonding environments). Instrument matching diagnostics and chemometric models developed on primary instruments were used to predict identical samples measured on vector instruments, and transfer performance was assessed using prediction error, reproducibility metrics and spectral diagnostics.

Methodology

Sample preparation:

• Ethanol/water series prepared gravimetrically in independent samples spanning 0.1% to 100% water; no serial dilutions.
• Talc mixed with ammonium iron(III) sulfate prepared in dry state; independent samples spanning low to a few percent talc.

Spectral acquisition protocols:

• Solid (talc/iron sulfate): diffuse reflectance using integrating sphere; spectral range 3800–12,000 cm-1; resolution 4 cm-1; 90 co-added scans (~67 s); Norton-Beer Medium apodization; InGaAs detector; samples in 1/2 dram glass vials.
• Liquid (ethanol/water): transmission through 0.5 mm quartz cuvette; spectral range 3800–12,000 cm-1; resolution 8 cm-1; 64 co-added scans (~32 s); Norton-Beer Medium apodization; InGaAs detector; measurements at ~21 °C (same-day paired acquisitions).

Instrument matching and diagnostics:

• Toluene subtraction test (transmission) used to assess x-axis (wavelength) matching and to detect spectral artifacts between instruments; polystyrene mentioned as alternative standard.
• Second-derivative and Savitzky–Golay derivatization used to highlight band positions and small multiplicative offsets.

C hemometrics and preprocessing:

• Thermo Scientific TQ Analyst software used.
• Models developed using Partial Least Squares (PLS) and Stepwise Multiple Linear Regression (SMLR).
• Pre-treatments included Multiplicative Scatter Correction (MSC) and Norris derivative filters.
• Calibrations were developed on primary instruments and applied unchanged to vector instruments (no post-transfer slope/bias or spectral matching algorithms applied).

Used instrumentation

• Thermo Scientific Antaris FT-NIR analyzers (multiple units of the same model family).
• Integrating sphere module for diffuse reflectance (solid samples) and standard liquid transmission module for liquids.
• InGaAs detectors for all experiments.
• Standard 0.5 mm quartz cuvette for liquid transmission and 1/2 dram glass vials for diffuse reflectance.

Main results and discussion

Instrument matching:
The toluene subtraction test indicated close x-axis alignment and low instrumental artifacts for the instruments used in both model systems. This initial confirmation of instrument sameness was a prerequisite for direct transfer without additional mathematical correction.

Talc/iron sulfate (solid):

• Talc shows a narrow, sharp band at 7185 cm-1 (width at half-height ≈ 6.1 cm-1). Such narrow features are sensitive to x-axis and photometric accuracy.
• Spectra from six Antaris instruments overlaid closely after simple derivative preprocessing; main differences were small multiplicative offsets (y-axis) readily handled by chemometric modeling.
• Predictions from the primary calibration applied to vector instruments produced correlation coefficients (R) ≈ 0.995–0.996, slopes ≈ 1.00–1.03 and small intercepts, all within 95% confidence limits of the primary calibration parameters.
• Absolute prediction differences typically appeared in the second decimal place of percent talc; %RSD across vector instruments was generally low (often <4% and down to ~1% for many samples). Repeated measures on the primary instrument indicated method precision accounted for a non-negligible portion (roughly one-third to one-sixth) of the transfer variability.

Ethanol/water (liquid):

• Water and ethanol bands are broad, strong and concentration dependent, exhibiting frequency shifts and changes in band shape due to hydrogen bonding. This represents a tougher transfer challenge because both x- and y-axis match are critical.
• A PLS model (2 factors) built on the primary instrument using spectra from 10–90% ethanol predicted the same samples measured on the vector instrument with excellent agreement. Calibration and validation points for the two instruments overlaid closely despite the system’s inherent nonlinearity.
• Spectral diagnostics: overlaid second-derivative spectra for key concentrations (e.g., 2% water) showed negligible differences between instruments; subtraction plots revealed only minor residuals. Principal component score plots for second-derivative spectra from 0.1–100% water showed coincident scores for paired measurements, with the first two PCs explaining ~90% of variance.
• Quantitatively, relative differences in second-derivative intensities between instruments were typically below 1% at key frequencies; the largest relative difference observed was about ~4% for an extremely low-absorbance sample (0.1% water), which corresponds to a very small absolute signal.

Overall, both models transferred successfully without instrument-specific mathematical corrections. The critical enabling factor was good instrument matching (wavelength and photometric stability), verified by the toluene test and spectral overlays.

Benefits and practical applications

• Direct calibration transfer reduces time and resource demands by avoiding recalibration or complex transfer algorithms for each vector instrument.
• Supports broader implementation of FT-NIR across manufacturing sites and QC labs, enabling consistent measurement protocols and reduced method maintenance burden.
• Applicable to both solids (diffuse reflectance) and liquids (transmission) across diverse spectral behaviors — from sharp, low-intensity bands to broad, shifting hydrogen-bonded bands.
• The talc and ethanol/water models can serve as practical readiness tests for instrument networks prior to deploying critical production calibrations.

Future trends and potential uses

• Expanded use of standardized spectral diagnostics (toluene, polystyrene, or equivalent standards) and automated instrument qualification routines to ensure x- and y-axis matching across instrument fleets.
• Development of robust, universal calibrations trained on multi-instrument datasets (inoculation approaches) to reduce sensitivity to instrument-specific artifacts while maintaining acceptable performance for regulated environments.
• Integration of instrument matching metrics into instrument manufacturing and factory acceptance procedures to enable plug-and-play transferability.
• Hybrid strategies combining hardware standardization, periodic spectral standards checks, and selective algorithmic correction (when needed) to maintain regulatory traceability and auditability.
• Potential extension to on-line and at-line process analytics where rapid deployment across multiple analyzers is essential for process control.

Conclusion

This study demonstrates that, when instruments are well matched in wavelength and photometric response, FT-NIR calibration models can be transferred directly between Thermo Scientific Antaris analyzers for both a challenging solid-state system (talc in iron sulfate) and a dynamic liquid system (ethanol/water). The combination of rigorous instrument diagnostics (toluene subtraction), appropriate preprocessing, and robust chemometric modeling enabled low transfer error without slope/bias correction or other transfer algorithms. The talc and ethanol/water model systems also provide practical benchmarks to evaluate instrument readiness for routine calibration transfer in industrial settings.

Reference

1. J.K. Drennen, E.G. Kramer and R.A. Lodder, Critical Reviews in Analytical Chemistry, 22(6), 443 (1991).
2. J.D. Kirsch and J.K. Drennen, Applied Spectroscopy Reviews, 30(3), 139 (1995).
3. K.M. Morisseau and C.T. Rhodes, Drug Development and Industrial Pharmacy, 21, 1071 (1995).
4. K.H. Norris, Journal of Near Infrared Spectroscopy, 4, 69 (1996).
5. E.W. Ciurczak, Pharmaceutical Technology, 15, 140 (1991).
6. W.F. McClure, Analytical Chemistry, 66, 43A (1994).
7. J. Lin, Applied Spectroscopy, 52, 1591 (1998).
8. J.S. Shenk, M.O. Westerhaus and W.C. Templeton, Jr., Crop Science, 25, 159 (1985).
9. J.S. Shenk and M.O. Westerhaus, Crop Science, 31, 1694 (1991).
10. Y. Wang and B.R. Kowalski, Applied Spectroscopy, 46, 764 (1992).
11. Y. Wang and B.R. Kowalski, Analytical Chemistry, 65, 1301 (1993).
12. Y. Wang, M.J. Lysaght and B.R. Kowalski, Analytical Chemistry, 64, 562 (1992).
13. Z. Wang, T. Dean and B.R. Kowalski, Analytical Chemistry, 67, 2379 (1995).
14. E. Bouveresse, D.L. Massart and P. Dardenne, Analytica Chimica Acta, 297, 405 (1994).
15. E. Bouveresse, D.L. Massart and P. Dardenne, Analytical Chemistry, 67, 1381 (1995).
16. E. Bouveresse, D.L. Massart, I.R. Last and K.A. Prebble, Analytical Chemistry, 68, 982 (1996).
17. T.B. Blank, S.T. Sum, S.D. Brown and S.L. Monfre, Analytical Chemistry, 68, 2987 (1996).
18. J. Lin, S.-C. Lo and C.W. Brown, Analytica Chimica Acta, 349, 263 (1997).
19. F. Despagne, B. Walczak and D.L. Massart, Applied Spectroscopy, 52(5), 732 (1998).
20. J. Workman and J. Coates, Spectroscopy, 8(9), 36 (1993).
21. S.R. Lowry, J. Hyatt and W.J. McCarthy, Applied Spectroscopy, 54(3), 450 (2000).