A Flexible Biorefinery using Machine Learning
A multidisciplinary consortium of researchers at Aalto University and 脜bo Akademi in Finland, and at Technical University Munich in Germany, are developing a flexible biorefinery concept to exploit the unique synergism of lignin-carbohydrate complexes (LCCs), which are naturally found in wood, to produce fully bio-based antioxidants, effective surfactants and polymer additives. LCCs are indispensable components of the wood cell wall, providing wood its rigidity and structure. In classical pulping processes, LCCs are disintegrated to isolate cellulose, hemicellulose and lignin. Within the consortium project 鈥淎I-4-LCC: Exploiting Lignin-Carbohydrate Complex through Artificial Intelligence鈥 funded by the Research Council of Finland, the teams of late Professor of Practice Mikhail Balakshin (鈥2022, Aalto University), Prof. Patrick Rinke (Aalto University, Technical University of Munich) and Prof. Chunlin Xu (脜bo Akademi) have developed a process to produce LCCs with tailored properties and high yield.
The new process is based on a modified hydrothermal treatment of wood followed by solvent extraction of the resulting solids, named AquaSolv Omni (AqSO) by Balakshin. The properties of the LCCs can be adjusted precisely by tuning the process conditions. Researchers from Rinke鈥檚 group employed Bayesian Optimization to iteratively collect data points of interest and evaluate the impact of the process conditions (P-factor, temperature, and liquid-to-solid ratio) on yield and carbohydrate content. Incorporation of Pareto front analysis allowed the team to maximize both yield and carbohydrate content. To evaluate LCC鈥檚 potential for high-value applications, the antioxidant potential, surface tension and glass transition temperature were measured in close collaboration with experts from Xu鈥檚 group of biomass chemists. The joint effort was recently published in the prestigious journal ChemSusChem: .
Doctoral candidate Daryna Diment explains: 鈥淗igh carbohydrate content of LCCs was found beneficial for reducing the glass transition temperature and surface tension of aqueous solutions, implying its potential use in thermoplastic formulations and as bio-based surfactants.LCCs produced in severe processing conditions (high temperature and extensive time) demonstrated a remarkable antioxidant activity.鈥
Overall, the developed idea represents an unparalleled advantage over traditional biorefineries for LCC isolation because it involves only water, temperature, and acetone extraction. On top of that, Bayesian Optimization provided flexibility to produce LCCs with targeted properties in high yield. This work demonstrates the potential of LCCs as a novel biorefinery product and showcases how machine learning can be used to accelerate the development of new technologies to valorize local and natural resources.
More information:
Enhancing Lignin-Carbohydrate Complexes Production and Properties with Machine Learning in ChemSusChem by Diment et al.

Read more news

Herd immunity may not work how we think
A new study from researchers at Aalto University suggests that our picture of herd immunity may be incomplete 鈥 and that understanding how people are connected could be just as important as knowing how many are immune.
Aalto computer scientists in ICML 2025
Department of Computer Science papers accepted to International Conference on Machine Learning (ICML)
New quantum record: Transmon qubit coherence reaches millisecond threshold
The result foreshadows a leap in computational capabilities, with researchers now inviting experts around the globe to reproduce the groundbreaking measurement.