Aristides P. Carneiro

Biorefining processes using ionic liquids (ILs) require proper solubility data of biomass-based compounds in ILs, as well as an appropriate thermodynamic approach for the modeling of such data. Carbohydrates and their derivatives such as sugar alcohols represent a class of compounds that could play an important role in biorefining. Thus, in this work, the pure IL density and solubility of xylitol and sorbitol in five different ILs were measured between 288 and 339 K. The ILs under consideration were 1-ethyl-3-methylimidazolium dicyanamide, 1-butyl-3-methylimidazolium dicyanamide ([bmim][DCA]), Aliquat dicyanamide, trihexyltetradecylphosphonium dicyanamide, and 1-ethyl-3-methylimidazolium trifluoroacetate. Comparison with the literature data was performed, showing good agreement. With the exception of [bmim][DCA], the solubility of these sugar alcohols in the other ILs is presented for the first time. The measured data as well as previously published solubility data of glucose and fructose in these ILs were modeled by means of PC-SAFT using a molecular-based associative approach for ILs. PC-SAFT was used in this work as it has shown to be applicable to model the solubility of xylitol and sorbitol in ILs ( Paduszyński ; et al. J. Phys. Chem. B 2013 , 117 , 7034−7046). For this purpose, three pure IL parameters were fitted to pure IL densities, activity coefficients of 1-propanol at infinite dilution in ILs, and/or xylitol solubility in ILs. This approach allows accurate modeling of the pure IL data and the mixture data with only one binary interaction parameter k_ij between sugar and the IL or sugar alcohol and the IL. In cases where only the pure IL density and activity coefficients of 1-propanol at infinite dilution in ILs were used for the IL parameter estimation, the solubility of the sugars and sugar alcohols in the ILs could be predicted (k_ij = 0 between sugar and the IL or sugar alcohol and the IL) with reasonable accuracy.

The Perturbed‐Chain Statistical Association Fluid Theory is applied to simultaneously describe various thermodynamic properties (solution density, osmotic coefficient, solubility) of aqueous solutions containing a monosaccharide or a disaccharide. The 13 sugars considered within this work are: glucose, fructose, fucose, xylose, maltose, mannitol, mannose, sorbitol, xylitol, galactose, lactose, trehalose, and sucrose. Four adjustable parameters (three pure‐sugar parameters and a k_ij between sugar and water that was allowed to depend linearly on temperature) were obtained from solution densities and osmotic coefficients of binary sugar/water solutions at 298.15 K available in literature. Using these parameters, the sugar solubility in water and in ethanol could be predicted satisfactorily. Further, osmotic coefficients and solubility in aqueous solutions containing two solutes (sugar/sugar, sugar/salt) were predicted (no additional k_ij parameters between the two solutes) reasonably. The model was also applied to predict the solubility of a sugar in a solvent mixture (e.g., water/ethanol) without additional fitting parameters. © 2013 American Institute of Chemical Engineers AIChE J, 59: 4794–4805, 2013

This study proposes three methodologies to define artificial neural network models through genetic algorithms (GAs) to predict the next-day hourly average surface ozone (O₃) concentrations. GAs were applied to define the activation function in hidden layer and the number of hidden neurons.