NETZSCH launches AI-powered polymer qualification tool for recyclate quality control

Proteus Now Quantify analysis of a DSC curve from a recyclate sample. Subtle peak shifts, overlaps and variations in peak areas are evaluated across the full thermal fingerprint. The machine learning model resolves the transitions and converts a single DSC measurement into a quantitative polymer composition (PP, HDPE, LDPE, LLDPE). Source (All Images) | NETZSCH

As composites manufacturers face mounting pressure to incorporate recycled content, they encounter a persistent challenge: Knowing what is in the recyclate and in what quantities. To address this issue, NETZSCH Analyzing & Testing (Selb, Germany) recently introduced Proteus Now Quantify, a software solution that combines differential scanning calorimetry (DSC) with machine learning to identify and quantify unknown compositions from a single, fully automated measurement.

“A central challenge in plastics recycling is the lack of transparency around material composition,” says Dr. Sara Simon, sector manager polymers at NETZSCH. “Quick checks like melt flow index or FTIR are useful, but limited— often qualitative and not universal. They struggle with dark materials, multilayer films and complex mixtures. The result is materials get downgraded, products get over-engineered or batches are rejected — costing materials, energy and credibility.”

Featured Content

Report: Composites and Carbon Fiber Use in Hydrogen Storage READ MORE

Troubleshooting thermoplastic composite stamp forming READ MORE

Episode 55: Dave Noblin, Greene Tweed READ MORE

The software works by organizing variability in DSC data and resolving overlapping peaks that occur when multiple polymers crystallize at similar temperatures. Trained on calibrated polymer blends, it enables detection of contamination levels down to 1%. Currently focused on polymers such as PP, HDPE, LDPE and LLDPE, the system delivers quantitative composition results with a root mean square error (RMSE) values typically between 3-6%.

For composite applications, NETZSCH has developed a practical workflow: Users first determine the filler content via TGA or burn-off methods, then adjust the sample weight in the software to reflect the effective polymer mass. The company is actively seeking partners to share calibrated blends with and without fillers to expand the training dataset and improve the composite workflow.

“The variation across virgin polyolefin grades is often larger than the changes introduced during reprocessing in recycling,” Simon notes. “By training the model on a broad range of well-characterized virgin materials from different suppliers, we can reliably capture typical recyclate behavior.”

Comparison between target and predicted polymer concentrations for a calibrated mixture using Proteus Now Quantify (based on DSC curve in opening image). The prediction quality is assessed via RMSE (%) and deviation (%). RMSE gives an indication of model accuracy and is affected by the amount and quality of training data for each polymer class.

Regulatory developments are further accelerating adoption. From 2030, recyclate suppliers entering the European market will be required to meet EU standards plus will need to provide documented origin and composition of their materials. These regulations are designed to influence global markets, effectively making reliable material characterization a prerequisite for market access.

The software integrates directly with NETZSCH’s DSC instruments and operates under standardized test conditions — 10-milligram sample mass and heating and cooling rates of 10 Kelvin/minute — ensuring consistent and reproducible results. Once the DSC measurement is complete, analysis is performed within seconds.

Proteus Now Quantify is currently available for polyolefin analysis, while the material database continues to expand to include additional polymer classes such as PA, PET, ABS, HIPS, PC, PUR and other engineered materials.