AUTHOR=Hada Subin , Solvason Charles C. , Eden Mario R. TITLE=Characterization-Based Molecular Design of Bio-Fuel Additives Using Chemometric and Property Clustering Techniques JOURNAL=Frontiers in Energy Research VOLUME=2 YEAR=2014 URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2014.00020 DOI=10.3389/fenrg.2014.00020 ISSN=2296-598X ABSTRACT=

In this work, multivariate characterization data such as infrared spectroscopy was used as a source of descriptor data involving information on molecular architecture for designing structured molecules with tailored properties. Application of multivariate statistical techniques such as principal component analysis allowed capturing important features of the molecular architecture from enormous amount of complex data to build appropriate latent variable models. Combining the property clustering techniques and group contribution methods based on characterization (cGCM) data in a reverse problem formulation enabled identifying candidate components by combining or mixing molecular fragments until the resulting properties match the targets. The developed methodology is demonstrated using molecular design of biodiesel additive, which when mixed with off-spec biodiesel produces biodiesel that meets the desired fuel specifications. The contribution of this work is that the complex structures and orientations of the molecule can be included in the design, thereby allowing enumeration of all feasible candidate molecules that matched the identified target but were not part of original training set of molecules.