Monitoring the purity of recovered solvents by NIRS

Significance of the Topic

Monitoring the purity of recovered solvents is critical in chemical and pharmaceutical manufacturing to reduce waste, control costs and ensure product quality. Near-Infrared (NIR) spectroscopy offers a rapid, reagent-free method to assess solvent purity and quantify impurities such as water and methanol, enabling timely decision making in recovery processes.

Objectives and Study Overview

The study aimed to develop NIR-based calibration models for determining water and methanol content and overall purity in distilled methylene chloride. Samples were collected from a solvent recovery distillation unit covering typical impurity ranges and reference values were obtained by Karl Fischer titration (water) and gas chromatography (methanol, other organics).

Methodology and Instrumentation

Samples were scanned in glass jars without preparation using a NIRS XDS SmartProbe Analyzer equipped with a fiber-optic immersion probe and a 4 mm path length (2 mm gap from probe to reflector). The spectrometer covered 400–2500 nm in under 30 s, though spectral regions above 2200 nm were excluded due to fiber absorption. Sixteen co-added scans of sample and reference were ratioed to obtain absorbance. Second-derivative pretreatment enhanced spectral features and minimized baseline effects.

Main Results and Discussion

• Water Calibration (1800–2000 nm)

• Partial Least Squares (PLS) with three factors.
• SEC = 0.0156 % (w/w), SEP = 0.0133 % (w/w), comparable to Karl Fischer accuracy.

• Methanol Calibration (2000–2100 nm)

• PLS model based on GC reference values.
• SEC = 0.0031 % (w/w), SEP = 0.0034 % (w/w), similar to GC precision.

• Overall Purity Calibration (1120–1680 nm and 1720–2200 nm)

• PLS prediction of total non-methylene chloride impurities.
• SEC = 0.043 % (% purity), SEP = 0.0455 %, matching GC determination.

The consistent SEC and SEP values across calibration and validation sets indicate robust models with low risk of overfitting.

Benefits and Practical Applications

• Analysis time below 30 s with no chemical reagents or consumables.
• Simple grab-sample operation or potential inline integration.
• Eliminates the need for trained analysts in routine monitoring.
• Can be implemented on lab, at-line or in-process configurations (SmartProbe, Rapid Liquid or Process Analyzer).

Future Trends and Potential Applications

• Inline process control of distillation or solvent purification streams.
• Extension to other solvents and impurity profiles through tailored calibrations.
• Integration with advanced chemometric algorithms and machine learning for real-time quality assurance.
• Portable and handheld NIR devices for field or small-scale production environments.
• Data connectivity for automated process analytical technology (PAT) frameworks.

Conclusion

NIR spectroscopy using a fiber-optic SmartProbe provides rapid, accurate quantification of water, methanol and overall purity in recovered methylene chloride. Calibration models demonstrate precision on par with reference methods, enabling cost-effective, non-destructive monitoring in both laboratory and process settings.