PHB (poly(3-hydroxybutyrate)) and its copolymer PHBV (poly(3-hydroxybutyrate-co-3-hydroxyvalerate)) are among the most widely studied bio-based, biodegradable plastics, produced through bacterial fermentation rather than refined from oil.

But turning them into reliable products is harder than it sounds. PHB is naturally brittle, and it starts to break down chemically at a temperature only just above its melting point. That leaves manufacturers a very narrow window to safely melt and shape it during processes like extrusion or injection molding. Even small changes in formulation, adding a different comonomer, a plasticiser or a filler, can shift these thresholds noticeably, making trial-and-error design slow and costly.

ANIPH partner, the Agricultural University of Athens (AUA) addresses this in a recent publication titled “Machine Learning Prediction of Thermal Properties of PHB/PHBV-Based Materials: A Quantitative Structure–Property Relationship Approach Using an Integrated Polymer Database”. The study show the possibility to predict three properties that describe how a PHB/PHBV formulation behaves as it heats up or cools down:

• Glass transition temperature (Tg) – the point where the material turns from rigid to flexible
• Melting temperature (Tm) – the point where its internal structure melts
• Crystallization temperature (Tc) – the point where it re-solidifies into an ordered structure on cooling

To build the models, the team put together a carefully checked database of 572 data points, drawn from 109 scientific studies plus 14 in-house experiments carried out with CETEC. The dataset records each sample’s composition, molecular weight and any additives used, whether a plasticizer, a filler, or a stabilizer, so the models could learn how these formulation choices affect thermal behaviour.

Two machine learning methods were trained and compared, and after careful tuning and testing, the best models could explain a large share of the differences seen between formulations:

• Melting temperature: correctly explained about 82% of the variation
• Glass transition temperature: about 77%
• Crystallization temperature: about 76%

The results also made physical sense: formulations with a higher share of the hydroxyvalerate building block tended to melt and crystallize at lower temperatures, since it disrupts the neat, orderly packing of the polymer chains. Higher molecular weight, on the other hand, tended to push these temperatures up, as longer chains become more entangled and harder to melt. The type and amount of additive used also mattered, in line with its intended role, whether that’s adding flexibility, encouraging crystal formation, or improving heat stability.

Because thermal testing conditions aren’t always reported the same way across studies, the team had to work with a smaller, high-quality subset of the data for the final models, between roughly 120 and 200 samples per property, and was careful to define the formulation ranges within which the predictions can be trusted.

The curated dataset and modelling code have been made openly available, in line with the open-data principles that guide all of ANIPH’s research, so that other researchers and manufacturers can build on this work to design bioplastic formulations with the processing behaviour they need.

Read the full article here

Cover photo by charlesdeluvio on Unsplash