Abstract

To expedite the valorisation of lignin as a sustainable component in materials applications, rapid and generally available analytical methods are essential to overcome the bottleneck of lignin characterisation. Where features of a lignin's chemical structure have previously been found to be predicted by Partial Least Squares (PLS) regression models built on Infrared (IR) data, we now show for the first time that this approach can be extended to prediction of the glass transition temperature (Tg), a key physicochemical property. This methodology is shown to be convenient and more robust for prediction of Tg than prediction through empirically derived relationships (e. g., Flory-Fox). The chemometric analysis provided root mean squared errors of prediction (RMSEP) as low as 10.0 °C for a botanically, and a process-diverse set of lignins, and 6.2 °C for kraft-only samples. The PLS models could separately predict both the Tg as well as the degree of allylation (%allyl) for allylated lignin fractions, which were all derived from a single lignin source. The models performed exceptionally well, delivering RMSEP of 6.1 °C, and 5.4 %, respectively, despite the conflicting influences of increasing molecular weight and %allyl on Tg. Finally, the method provided accurate determinations of %allyl with RMSEP of 5.2 %.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call