Optimizing a High-Entropy System: Software-Assisted Development of Highly Hydrophobic Surfaces using an Amphiphilic Polymer

Abstract

In materials science, the investigation of a large and complex experimental space is time-consuming and thus may induce bias to exclude potential solutions where little to no knowledge is available. This work presents the development of a highly hydrophobic material from an amphiphilic polymer through a novel, adaptive artificial intelligence approach. The hydrophobicity arises from the random packing of short polymer fibers into paper, a highly entropic, multistep process. Using Bayesian optimization, the algorithm is able to efficiently navigate the parameter space without bias, including areas which a human experimenter would not address. This resulted in additional knowledge gain, which can then be applied to the fabrication process, resulting in a highly hydrophobic material (static water contact angle 135°) from an amphiphilic polymer (contact angle of 90°) through a simple and scalable filtration-based method. This presents a potential pathway for surface modification using the short polymer fibers to create fluorine-free hydrophobic surfaces on a larger scale.

Materiomically Designed Polymeric Vehicles for Nucleic Acids: Quo Vadis?

“…242 Adaptive experimental designs that integrate machine learning and data-driven experimental exploration have managed to solve multiobjective material optimization properties in a variety of domains 243 including MOFs, 244 perovskites, 245−247 peptide design, 248 and superhydrophobic surfaces. 249 Gianneschi and collaborators developed "Peptide Optimization with Optimal Learning" or POOL to identify short peptide substrates for enzymes and observed several benefits over conventional screening techniques such as phage display and directed evolution. 250 In their approach, model predictions informed peptide synthesis, ensuring a feedback loop between the model and experiment; they also accounted for uncertainty in model predictions by prioritizing diversity in peptide selection.…”

“…Bayesian optimization considers both exploration and exploitation in recommending candidates for synthesis; it typically prioritizes candidates where predicted biological performance is high, as well as candidates where uncertainty is very high . Adaptive experimental designs that integrate machine learning and data-driven experimental exploration have managed to solve multiobjective material optimization properties in a variety of domains including MOFs, perovskites, − peptide design, and superhydrophobic surfaces . Gianneschi and collaborators developed “Peptide Optimization with Optimal Learning” or POOL to identify short peptide substrates for enzymes and observed several benefits over conventional screening techniques such as phage display and directed evolution .…”

BOXVIA: Bayesian optimization executable and visualizable application

“…The application of BO has been increasing across various research fields because it can optimize high-dimensional and non-differentiable objective functions efficiently [1]. For example, in the field of materials science, BO has been used to explore new materials [3,4] and to optimize experimental parameters for improving material properties [5,6].…”

Application of Bayesian optimization for improved passivation performance in TiO x /SiO y /c-Si heterostructure by hydrogen plasma treatment

“…Accordingly, machine learning has begun to be used to optimize material processing conditions in practical experiments, also known as process informatics. [24][25][26] Bayesian optimization (BO) is a powerful method to optimize multiple parameters based on stochastic prediction, and it can reduce the number of experiments to optimize in multidimensional parameter space. 27) For example, Osada et al recently reported that BO can be successfully applied to optimize the epitaxial growth process of Si thin films and the crystal growth rate was increased to be approximately twice as high as that under standard conditions.…”