Prediction of coating thickness for polyelectrolyte multilayers via machine learning.

Area of Science:

• Materials Science
• Nanotechnology
• Computational Chemistry

Background:

• Layer-by-layer (LbL) deposition is a versatile method for creating functional nanoscale coatings.
• Existing LbL development lacks predictive models linking components to final properties.
• The need for rapid development of biomedical solutions, like antiviral coatings, is critical.

Purpose of the Study:

• To analyze the impact of 23 parameters on LbL coating thickness.
• To develop a machine learning model for predicting LbL coating properties.
• To demonstrate the feasibility of using machine learning for LbL coating development.

Main Methods:

• Literature data and experimental results were used to analyze coating parameters.
• A predictive model was developed using coating parameters and polymer molecular descriptors from the DeepChem library.
• Machine learning algorithms were applied to predict coating thickness.

Main Results:

• The relative impact of 23 parameters on coating thickness was analyzed.
• A machine learning model was developed, though performance was limited by data scarcity.
• The study demonstrates the first use of machine learning for predicting LbL coating properties.

Conclusions:

• Machine learning can accelerate the development of functional nanoscale coatings.
• This approach has the potential to reduce experimental costs and enable rapid responses to health crises.
• Future applications could include predicting biocompatibility, cell adhesion, and antimicrobial/antiviral properties.