Data-Driven Approach to Decipher the Role of Triglyceride Composition on the Thermomechanical Properties of Thermosetting Polymers Using Vegetable and Microbial Oils

Abstract

Sustainable and renewable polymeric materials are gaining traction, and vegetable oils have been used directly or in modified forms to meet this demand. At the same time, microbial hosts (such as the oleaginous yeast Yarrowia lipolytica) are being touted as sustainable alternatives for petroleum and vegetable oils. However, the exact role of fatty acid composition and speciation on polymer performance remains unclear. Here, we explore a datadriven approach to explicitly relate the underlying oil composition with the thermomechanical properties of the resulting polymeric material. In doing so, we identify the C16:0, C16:1, and C18:0 fatty acid contents of vegetable oils as critical parameters for predicting thermal stability at maximum heat loss (T max ). Machine learning-based approaches were applied to study the link between thermal properties and monomer composition. In the end, application of multiple linear regression modeling indicated strong dependence on the C16:1 content as evident by the parameter loading (loading of +428 for T max ). As a more sustainable source of oil, Y. lipolytica oil-based polymer properties were also dictated by the C16:0 and C18:0 fatty acid contents but with an opposite impact as compared with vegetable oils (T max loadings of −208 and +36 for Y. lipolytica oils, +19 and −72 for vegetable oils, C16:0 and C18:0, respectively). Despite these differences, Y. lipolytica oilbased polymers showed similar strength and cross-linking density to vegetable oil polymers. This work is the first evaluation of polymer properties from a library of vegetable-and yeast-sourced oils and highlights a mechanistic understanding of thermal stability from both oil source (vegetable or microbial) and oil composition that can be used for future design.

Altering the fatty acid profile of Yarrowia lipolytica to mimic cocoa butter by genetic engineering of desaturases

“…This study does yield promising results concerning CBE production and demonstrates the possibility of creating a pallet of strains expressing a wide range of fatty acid profiles while still maintaining a high lipid content that can be used for other applications apart from food/feed supplementation [ 34 ], e.g. biodiesel with different melting temperatures [ 35 ] or biopolymers [ 36 ].…”

Concentration Prediction of Polymer Insulation Aging Indicator-Alcohols in Oil Based on Genetic Algorithm-Optimized Support Vector Machines

“…Afterward, Shaban et al used k-nearest neighbors (k-NNs) as a classification model to classify the furfural data of 731 field transformers and utilized the packaging method as a feature selection method, and a recognition rate of 90% was achieved [ 20 ]. In addition, machine learning is also rapidly emerging in polymer science and technology, providing important support for the design [ 21 , 22 ], thermal stability [ 23 ], surface area, and crystallinity [ 24 ] of polymer materials.…”