The Use of Transmission FT-NIR Analysis for the simultaneous analysis of Carbamazepine and PEG 2000 in Extrudates

Importance of the topic

Real-time monitoring of drug content during hot-melt extrusion (HME) addresses a core challenge in pharmaceutical manufacturing: transforming poorly soluble active ingredients into processable, bioavailable solid dispersions. Inline analytical techniques that quantify API and excipient levels during continuous extrusion enable process understanding, control and rapid release decisions, supporting Quality by Design and Process Analytical Technology initiatives.

Study objectives and overview

This application study evaluated the feasibility of transmission Fourier-transform near-infrared (FT-NIR) spectroscopy to quantify both carbamazepine (CBZ, the model API) and polyethylene glycol 2000 (PEG 2000, a plasticizer) in clear extrudates produced by twin-screw extrusion with Kollidon VA64 as polymer matrix. The goals were to (i) demonstrate spectral measurement in a transparent melt using transmission probes in the die, (ii) develop simple calibration models for API and plasticizer, and (iii) assess real-time prediction of changing feed compositions during extrusion runs.

Methodology

• Extrusion: Thermo Scientific PRISM Pharmalab 16 HME twin-screw extruder operated at 120 °C, 300 g/hr throughput and 50 rpm screw speed.
• Formulations: Binary/ternary blends of carbamazepine (5–20% w/w), PEG 2000 (2.5–25% w/w) and Kollidon VA64; ten calibration mixtures and two validation blends were prepared.
• Spectral acquisition: Thermo Scientific Antaris II Method Development System (MDS). Powdered components were scanned in an integrating sphere (diffuse reflectance), extrudates were measured in transmission using custom die-mounted fiber-optic probes with 2 mm path length.
• Acquisition parameters: 4,000–10,000 cm−1 range, 8 cm−1 resolution, 32 co-added scans (~16 s per scan). During extrusion runs spectra were collected continuously (one spectrum every 30 s).
• Data processing and calibration: Spectra were converted to second derivative to sharpen bands and reduce baseline effects. Single-wavenumber multiple linear calibrations were developed (selected peaks at 5064 cm−1 for CBZ and 5912 cm−1 for PEG). The study notes that multivariate PLS would be preferable with a larger training set; Thermo Scientific RESULT and TQ Analyst software were used for acquisition and modeling.

Instrumentation

• Thermo Scientific PRISM Pharmalab 16 HME twin-screw extruder.
• Thermo Scientific Antaris II Method Development System (MDS) FT-NIR spectrometer with integrating sphere for powders and transmission die-mounted fiber probes for melts.
• Analysis and chemometrics: RESULT workflow for online prediction and TQ Analyst for model development.

Main results and discussion

• Spectral features: Distinct NIR absorptions in the second-derivative spectra allowed selection of analytical wavenumbers—5064 cm−1 for CBZ (strong, low interference) and 5912 cm−1 for PEG (weaker, partially overlapped by polymer/API bands).
• Calibration performance: CBZ calibration produced a correlation coefficient ~0.994 with RMSEC ≈ 0.48% (excellent linear fit). PEG calibration gave a correlation coefficient ~0.97 with RMSEC ≈ 0.82% (good but more variable), reflecting overlapping absorptions and fewer calibration points.
• Process monitoring: Inline predictions tracked programmed composition changes. Example transitions (e.g., 12.5%→7.5% CBZ runs) showed that the instrument detected trends reliably, with final predicted concentrations typically within the calibration error (≈0.5–0.8% absolute). However, the melt composition required long residence times (many minutes) to stabilize at the die; under the chosen conditions stabilization often exceeded 14–18 minutes.
• Limitations observed: When CBZ loading reached 20% the melt became cloudy (incomplete solubilization), increasing spectral noise and baseline shifts that degraded measurement precision. PEG predictions were more susceptible to variability due to spectral overlap and limited calibration sample density. The study therefore used single-wavenumber calibrations for feasibility, while recommending richer sample sets and PLS modeling for robust multivariate quantification.

Benefits and practical applications

• Real-time simultaneous monitoring of API and plasticizer supports process control, faster troubleshooting and reduced offline testing.
• Transmission FT-NIR with die-mounted probes enables measurement of transparent melts where reflectance probes fail, extending PAT capabilities for a broader range of HME formulations.
• Quantitative inline data allow assessment of residence-time effects and help optimize screw speed, throughput and formulation to achieve desired dispersion and dissolution states.
• Implementable within existing RESULT/TQ Analyst workflows, enabling near-term deployment in development laboratories and pilot manufacturing for formulation screening and scale-up studies.

Future trends and opportunities

• Develop full multivariate calibrations (PLS, OPLS) using expanded calibration sets to separate overlapping absorptions and reduce prediction uncertainty, particularly for excipients like PEG that show weak, convoluted bands.
• Incorporate temperature and scattering corrections or advanced preprocessing to mitigate baseline drift and noise from partial phase separation or cloudiness in melts.
• Integrate inline NIR predictions with closed-loop process control to adjust feed rates or screw profiles in real time and shorten stabilization times.
• Combine transmission FT-NIR with complementary sensors (Raman, imaging, torque/pressure) to monitor solid-state transformations (crystallinity vs amorphous) and physical consistency of extrudates.
• Refine probe and die designs to increase signal-to-noise, enable variable path lengths and simplify cleaning/validation for GMP environments.

Conclusions

Transmission FT-NIR spectroscopy with die-mounted probes is a feasible and effective PAT approach for simultaneous, real-time monitoring of carbamazepine and PEG 2000 in transparent HME extrudates. The method delivered strong quantitative performance for the API and acceptable results for the plasticizer using simple calibrations. Key practical considerations include managing melt clarity and expanding calibration sample diversity to enable robust multivariate models. With further model development and integration into control workflows, inline transmission NIR can materially improve continuous-extrusion process understanding and product quality assurance.

References

• Kelly A., Halsey S., Terrell M. The Use of Transmission FT-NIR Analysis for the simultaneous analysis of Carbamazepine and PEG 2000 in Extrudates. Thermo Fisher Scientific Application Note AN52565, 2014.