AI-Enhanced Kinetic Monte Carlo Modelling of CVD Diamond Growth with NV Centre Defects

• Grant number: 25/27755-9
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate (Direct)
• Start date: January 19, 2026
• End date: July 03, 2026
• Field of knowledge: Physical Sciences and Mathematics - Physics
• Principal Investigator: Vladimir Jesus Trava-Airoldi
• Grantee: Mirelly Ferreira da Silveira
• Supervisor: Paul May
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil
• Institution abroad: University of Bristol, England
• Associated to the scholarship: 23/14476-9 - Growth studies of monocrystalline CVD-diamond with minimization of structural defects, BP.DD

Abstract

The main objective of this project, based on initial growth data of single-crystal CVD diamond using the MWPECVD technique, is to better understand the influence of growth parameters on the density ratio of color centers (NV¿/NV°). Additional data will be sought to support simulation studies, taking into account the characteristics of the plasma discharge. The research will combine corrected KMC algorithms, ab initio modeling, and advanced deep learning methods such as Non-Negative Matrix Factorization (NMF) to investigate the growth and formation of the so-called color centers or defects resulting from vacancy formation-particularly nitrogen vacancies, NV° and NV¿-in single-crystal CVD diamond.The main focus will be on vibrational properties and spin-phonon coupling, which are crucial for the development of quantum technologies. The project includes several secondary goals, such as: developing an AI-assisted KMC model capable of accurately predicting CVD-diamond growth rates and NV formation as a function of input parameters and, especially, nitrogen concentration; diagnosing and correcting convergence errors in existing models; modeling NV centers using Density Functional Theory (DFT) to understand their structure and vibrational signatures; applying deep learning to identify patterns related to growth parameters; and using NMF to determine the NV¿/NV° concentration ratio with good accuracy.The final outcome will be a hybrid AI simulation framework capable of diagnosing and correcting convergence issues, predicting optimal growth conditions, and thus determining the NV¿/NV° concentration ratio through advanced spectral analysis.