Predicting the bond dissociation enthalpies in lignin-derived molecules using quantum machine learning models

Bond dissociation enthalpy (BDE) is a fundamental chemical property for predicting molecular stability and reactivity. BDEs are crucial for understanding antioxidant efficiency, enzyme catalysis, surface functionalization chemistry, and drug discovery. This project will focus on predicting BDEs for C-O and C-C bond types in lignin-derived molecules, essential for efficient lignin decomposition processes in biofuel production and renewable chemicals. Computational approaches such as density functional theory (DFT) are computationally expensive which limit the size and diversity of BDE dataset for lignin-relevant systems. To address this challenge, we propose an active learning framework combined with Bayesian optimization which iteratively identifies the most informative BDE data points. We will use classical machine learning models to generate high-quality BDE data and minimize the requirements of costly DFT calculations. In addition, we will focus on the potential of quantum models, leveraging properties such as entanglement and superposition. We will develop a “quantum” active learning framework which will enable a more efficient exploration of chemical spaces in lignin-related systems. The use of quantum models will further enhance predictive capabilities by efficiently handling data-scarce scenarios. This work will contribute towards the efficient utilization of biomass for renewable energy applications.

Faculty Supervisor:

Viki Kumar Prasad

Student:

Partner:

Indian Institute of Technology Madras

Discipline:

Physics

Sector:

Quantum Science; Green/Alternative Energy; Artificial Intelligence

University:

University of Calgary

Program: