Determining the structure of functionalized graphene for tailored thermomechanical properties using ML techniques

Area of Science:

• Materials Science
• Computational Chemistry
• Nanotechnology

Background:

• Chemical functionalization of graphene offers vast design possibilities for nanosheets.
• Traditional methods for inverse design of functionalized graphene are computationally expensive and intractable due to numerous variables.

Purpose of the Study:

• To develop a fast and accurate data-driven framework powered by molecular dynamics (MD) for identifying graphene layouts with specific thermomechanical properties.
• To overcome the limitations of conventional approaches in the inverse design of functionalized graphene.

Main Methods:

• Generated a dataset of 1200 functionalized graphene sheets with varying layouts and thermomechanical properties (Young's modulus, thermal conductivity, max stress/strain).
• Trained regression models (SVR, Ridge, GPR) with Label and Bag-of-Words encoding, achieving high predictive performance (R² > 0.9).
• Employed an evolutionary optimization process coupled with trained ML models to find graphene layouts matching user-defined properties.

Main Results:

• Machine learning models demonstrated high accuracy (R² > 0.9) and low error (MAPE < 1%) in predicting thermomechanical properties.
• The framework achieved a 7-order magnitude speedup in property estimation and up to 6-order magnitude faster layout identification compared to pure MD simulations.
• MD-validations confirmed the framework's applicability, showing acceptable deviations for thermal conductivity and good alignment for mechanical properties.

Conclusions:

• The developed MD-powered, data-driven framework enables rapid and accurate inverse design of functionalized graphene sheets.
• This approach significantly reduces computational cost and time, making the design of tailored graphene materials more accessible.
• The study highlights the potential of machine learning in accelerating materials discovery and design for advanced applications.