Systematic Solvent Screening and Design for Biomass Fractionation Processes

Abstract

In view of depleting fossil-resources and climate change, bio-based feedstocks are envisioned to serve as a renewable carbon source in a circular economy. Biorefineries play a vital role in the circular economy, since they enable resource efficient utilisation of biomass, such as lignocellulose or microalgae. Biorefinery processes fractionate the feedstocks into molecular building blocks to produce pharmaceuticals, chemicals, food, feed, and fuels, thereby closing material loops. In a fractionation process, these building blocks are extracted from the biomass and fur- ther separated, frequently utilising organic solvents. The use of organic solvents significantly impacts the product yields, the energy demand, greenhouse gas emissions, operational safety, and economic viability of a biorefinery process. Therefore, solvent selection is a molecular- level decision with far-reaching consequences for the overall process. Computational methods can significantly accelerate solvent selection and guide experiments toward the most promis- ing candidates. However, there is a lack of efficient computational methods suitable for rational solvent selection and design for biomass fractionation, limiting the development of innovative biorefinery strategies. The present thesis addresses this gap by introducing computational methods for solvent screening and design, which were experimentally validated on lignocellulose and microalgae. A high-throughput screening method was developed to evaluate a database containing more than 8000 potential solvents. The method applies computational models to predict important solvent properties. Based on these predictions, solvent candidates with undesired structural features, thermophysical and thermodynamic properties, as well as environmental, health, and safety properties can be eliminated. Thus, the search is narrowed, enabling targeted experimental tests. However, for highly constrained solvent selection problems, identifying a solvent within the database that meets the many selection criteria may become difficult. To allow for rational solvent selection beyond a pre-defined database, a computational solvent design method was developed. This method can be used to tailor the solvents’ molecular structure toward the desired properties using a graph-based genetic algorithm. The close combination of computational methods and experiments enabled the identification effective solvents applicable for the fractionation of lignocellulosic and microalgal biomass. For the fractionation of lignocellulosic biomass and further valorisation of the extracted lignin, high lignin solubility is an important solvent selection criterion. Common fraction- ation approaches treat the biomass with acids, high temperatures and the selected sol- vent, resulting in undesired lignin condensation reactions which limit further lignin valori- sation. Aldehyde-assisted fractionation, an innovative biorefinery approach, inhibits such condensation reactions by lignin stabilisation with aldehydes. In this way, not only the cellulose-rich pulp and the hemicellulose sugars, but also the lignin fraction can be effec- tively valorised. However, this approach employs the carcinogenic solvent 1,4-dioxane which should be replaced by more benign alternatives. The developed computational methods identified solvents with high, experimentally confirmed lignin solubilities ranging from 20 to 60 wt.% (T = 85 ◦C). Fractionation experiments showed, that lignin was effectively stabilised by an aldehyde in many of the identified solvents. Some of the tested solvents outperformed 1,4-dioxane in terms of hemicellulose sugar yield, and/or toxicity, with a slightly lower lignin monomer yield as a trade-off. Microalgal biorefining faces several challenges, involving the use of toxic solvents and energy-intensive biomass drying. To circumvent the drying step, wet algal paste, still con- taining about 85 wt.% moisture, was investigated as a feedstock. The moisture is commonly considered as a barrier that further complicates biomass fractionation. The developed screen- ing approach identified solvents applicable to the fractionation of wet biomass of the model alga P. tricornutum. By combining the computational solvent screening approach with ex- perimental methods, a lab-scale biomass fractionation process for wet P. tricorntum biomass was developed. Breaking with the current view of treating water as a barrier, this approach exploits the presence of water to fractionate the biomass into lipids, carotenoids, carbohy- drates, and proteins. The developed biorefinery approach does not require biomass drying and employs only two benign solvents to fractionate the biomass at ambient conditions. Overall, the combination of computational solvent selection methods and experimental work allowed to replace harmful solvents from current fractionation approaches and paved the way for developing innovative biorefinery processes.