(558cc) Combining Experimental Kinetics and Thermodynamic Modeling with IR Spectroscopy and Machine Learning for Fundamental Studies and Fast Product Quantification

Rodriguez Quiroz, N., University of Delaware
Lansford, J., University of Delaware
Tsilomelekis, G., Rutgers, The State University of New Jersey
Vlachos, D. G., University of Delaware
The chemo-catalytic conversion of non-edible lignocellulosic biomass into furanic compounds such as 5-hydroxymethyl furfural (HMF) and furfural represent critical steps in biomass upgrading to fuels and chemicals. Metal salts enhance the yield and selectivity in the biphasic dehydration of fructose to HMF by improving the extraction of the product from the aqueous phase reducing the formation of byproducts. Furthermore, the dehydration reaction carried out in microreactors enhances the mass and heat transfer rates resulting in further increase in the HMF extraction at short reaction times. However, our understanding of the effect of the salts on the rate of dehydration and selectivity is limited, and the quantification of the products is slow compared to the fast throughput of microreactors.

In the present work, we couple thermodynamic modeling with experimental kinetics to understand the effect of different metal salts and Brønsted acid species on the dehydration of fructose to HMF. We focus on different alkali and alkaline earth metal salts on the acidity and speciation. Furthermore, we develop a method for fast online product quantification by utilizing machine learning for the systematic analysis of IR spectra.