We need to be careful here. DFT is not just an approximation method. There are limitations of density functional theory calculations such as the number of atoms (<1,000) and the time frame (pico seconds) due to the extremely high computational demand. Scientists use DFT calculations in various ways. For example to understand experimental data. An example would be to understand and explain what causes a material to be brittle. Another example would be to predict the feasibility and properties of novel material that is yet to be made. For example, one could use it to explore what dopants, and in what concentrations, would optimize the ionic transport in solid oxide fuel cells. This would help the experimentalist to narrow down to a few select materials. The experimental results could then be compared with the DFT predictions. Other examples would be to explore properties when experiments are not feasible, such as at absolute zero, or at the center of the earth (high temperature and pressure), or for modeling planetary formations.
Usually, DFT is not used in alone. It may be used to calculate the activation energies, for example, and then use a combination of other techniques such as Monte Carlo (MC) and molecular dynamics (MD) to scale up to larger systems. Sometimes DFT is also combined with known experimental data to arrive at better predictions and understanding. Nowadays DFT is also being combined with machine learning (ML) and artificial intelligence (AI) to hone down to the results faster. We do live in a very exciting time!
Depending on your needs one could simply use MC or MD and get very good results with them. These are much less computationally intensive methods and they readily apply to much larger systems and larger time scales.
I would suggest two excellent books.
1) Density Functional Theory by David Sholl & Janice Steckel.
2) Introduction to Computational Materials Science by Richard Lesar.
One excellent review article that came out this week re DFT and AI is the following.
The central role of density functional theory in the AI age by Huang, et. al., Science 381, 170–175 (2023), 14 July 2023.
------------------------------
Shahriar Anwar
Sr. Research Specialist
Arizona State Univ
Chandler AZ
(480) 965-5696
------------------------------
Original Message:
Sent: 07-14-2023 14:57
From: Efrain De La Rosa Garcia
Subject: DFT in materials simulation
I have seen that some experts are not comfortable at all with the use of DFT in materials simulation because it is just an approximation method, so I want to know if there is a way to achieve a better accuracy at the results given by DFT without increase much more the computational cost.
------------------------------
Efrain De La Rosa Garcia
Instituto Tecnológico de Saltillo
Saltillo
528442908832
------------------------------