I am looking for methods to determine the dissociation free energy in a precise way, ideally from a simulations. My hope is to get the most precise values possible that are specific to the structure of the chemical element, rather than the more general energies I have found that completely ignore the context of the molecule (like in "Bond Dissociation Energies of Organic Molecules", ACCOUNTS OF CHEMICAL RESEARCH 2003). A machine learning program that can assess the bond energy of an input molecule using parameters learned from a database would be ideal.
Alternatively, I would like to know what are the most accurate and reliable source for such values, as I have been having trouble finding sourced values.
The goal is to use these values to calibrate other properties of a system based on thermodynamic cycles. Since this calibration is for molecular simulations, I need accuracy but most importantly, consistency between methods. I can live with not so accurate parameters, as long as the overall force field is balanced, meaning the differences are balanced.