What is the role of a chemist in computational chemistry?

Question

As someone that wants to learn about computational chemistry, mainly for its use in energy storage materials (and therefore DFT as the most practical method of it), I am wondering what is essential to learn about DFT. The theory behind it seems quite vague if someone does not have the practical mathematical formulations. When a chemist comes to use computational chemistry in software, it is something completely different compared to when they learn about DFT.

What I understand is that the most essential task is the improvement of exchange correlation functionals to give better and more accurate results, which must be then compared with experimental observations, but the course that I have followed never discussed any way to do it. The consideration of LDA and GGA are never clear for a computational chemist for what is better or worse to use and the development of them is a job for a quantum physicist for example.

So I am wondering, what is really the essential role of a chemist in computational chemistry and what does he/she have to learn to be effective?

Answer 1

In terms of the role of chemists for what you call the "most essential task":

"What I understand is that the most essential task is the improvement of exchange correlation functionals to give better and more accurate results"

"... the development of them is a job for a quantum physicist for example."

I am reminded of two questions we've had here so far:

• What are some recent developments in density functional theory?
• What is the closest thing we have to "the" universal density functional?

For the first question: several milestones have been explained in the answers so far:

100% of these paper's have authors in chemistry departments.

In all cases, the first/only author is a chemist, except for the last paper, where the senior-most author John Perdew has a dual affiliation with Physics and Chemistry.

For the second question: Three people have answered so far, regarding what the closest thing to a "perfect" exchange-correlation functional might be, and all three of them were chemists, as well as the authors in the papers cited in those answers.

Summary: Milestones in functional development over the years have come both from physicists and chemists, with many of the key papers since 1993 seemingly coming from people in chemistry departments (or in some cases with dual affiliation in both chemistry and physics).

I am also not trying to say that chemists are more important than physicists: Walter Kohn (inventor of DFT) got his undergrad degree in Mathematics and his PhD in Physics, before getting a Nobel Prize in Chemistry. My answers here and here also show that much of the early DFT development was done on the uniform electron gas, by particle physicists like Keith Brueckner and Jeffrey Goldstone (these two were also important players in the coupled cluster method, which is now developed and used much more by chemists).

Answer 2

As a computational chemist, I find my role is to act as a translator. In a molecular dynamics simulation, I need to know enough physics and maths to translate chemical structures into a physical model, and make sure that the physics at play produces a model that makes sense chemically. Then if I'm doing multiscale modelling, say at finite element level, I need to know enough engineering to translate my chemical model into a larger structure like a beam or panel.

Nike makes the good point that a lot of seemingly physics based papers are written by chemists. My view on this is that these people are best translators, those that understand enough physics that they can develop new tools to represent chemistry in new and innovative ways.

If you think of materials science as the crossroads of engineering, chemistry and physics, computational materials science pushes that even further by adding computer science into the mix.

Answer 3

The role of the chemist in computational chemistry is to ask the right questions. Then, the computational chemist can step in, and start computing. The computational chemist should know which models are good for the problem in question, and what kinds of pitfalls to watch out for.

Then, at the very deep end you have the method developers as well as the code developers, who know the intrinsics of the chemical models and how they are implemented. To develop code, you don't have to understand why a method works, but to develop better methods you do have to know this. What often happens is that people start by writing code, and then as their career advances the younger people step in...

Answer 4

The role of the chemist in computational chemistry is to provide the real physical evidence. Computational chemistry deals with models, and predictions. It can give you insight into the mystical quantum nature of molecules but it's just that, a prediction.

The experimental chemist can then use those models and predictions to guide their experiments, which can support or oppose the predictions and models. The new experimental data can then be used to improve the models, and the process repeats itself!

This diagram describes well the typical workflow of a computational chemist and experimental chemist.

Of course, any good computational chemist can do both. They can come up with a good chemical idea, find experimentally relevant details in the literature or get someone to collaborate with, make hypotheses, test those hypotheses, gain insight and repeat the process. This is why it is good to have a good all around training. Don't be a monkey who just pushes a button. Think for yourself!