Monitoring of cellulose oxidation level by electrokinetic phenomena and numeric prediction model

• Published in: Cellulose (London) Vol. 27; no. 6; pp. 3107 - 3119
• Main Authors: Tarbuk, Anita, Grgić, Katia, Toshikj, Emilija, Domović, Daniel, Dimitrovski, Dejan, Dimova, Vesna, Jordanov, Igor
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.04.2020; Springer Nature B.V
• Subjects: acetates; Algorithms; artificial intelligence; Bioorganic Chemistry; Calcium acetate; cationic surfactants; Cellulose; Ceramics; Chemistry; Chemistry and Materials Science; Composites; Correlation coefficients; Cotton; disposable products; Electrokinetics; Functional groups; Glass; isoelectric point; Machine learning; Mathematical models; moieties; monitoring; Natural Materials; Organic Chemistry; Original Research; oxidants; Oxidation; Oxidizing agents; Physical Chemistry; Polymer Sciences; prediction; Prediction models; Regression analysis; Streaming potential; Surface charge; Sustainable Development; Titration; zeta potential; Oxidation system; Cotton; Electrokinetic phenomena; Machine learning
• ISSN: 0969-0239; 1572-882X
• DOI: 10.1007/s10570-020-03028-6

Cellulose with a low level of oxidation is suitable for producing stable long-lasting materials with high added value, while extensively oxidized once is applicable for disposable products. In our previous comprehensive research, the fundamental behavior of the cotton under the action of different oxidants has been explored. Different levels of oxidation, as well as the type of functional groups, have been achieved by properly selected oxidants while controlling their concentration and treatment time. In this research, the electrokinetic ζ-potential of KIO 4 and TEMPO-oxidized cotton and the isoelectric point are measured by the streaming potential method, while the surface charge is calculated from the adsorbed cationic surfactant by the back-titration method. The results of electrokinetic phenomena are compared with the amount of created carboxyl groups determined by the calcium acetate method. The machine learning algorithms Waikato Environment for Knowledge Analysis for regression analysis is employed to develop models that make numeric predictions of the ζ-potential values based on the known number of carboxyl groups. The model with the correlation coefficient between the actual and the predicted value of ζ-potential is given for the first time. Graphic abstract