Some questions about the use of ABINIT for computing formation energies

Running Abinit Ground state
1

Hi everyone.

I am currently trying to make use of ABINIT in my research, but I am incurring in some difficulties on how to use it, especially given that I lack a strong chemical background.

I understand the time that it would require to help me clarify these concepts. I will ask them in increasing order of complexity. Even understanding the simplest one would be very helpful to me. Thank you all in advance for your help.

Computing Formation Energy
I would like to compute the formation energy of some structures but I have some doubts regarding the procedure. From my understanding, given a structure with composition S = AnBm, the formation energy should be Ef(S) = E(S) - nE(A) - mE(B) (and similarly for compositions with more species). My question is: how do I compute the energy of single, isolated elements?

I followed the Base1 tutorial until the end, when it is explained how to compute the formation energy for H2. The main difficulty that I am facing is that, in that tutorial, in order to compute the total energy of H, some parameters need to be manually tuned (nsppol, nsym, etc. etc.) How can I tune them for arbitrary elements? Is there a general procedure or a reference list that I can use?

Replicating results of OQMD
For the most part, I am trying to replicate the results obtained by the Open Quantum Materials Database by referring to their DFT settings on a different simulator, VASP.

The problem is that for many parameters I can’t seem to find the analogous for ABINIT. For example: number of k-points per reciprocal atom (KPPRA), initial magnetic moments, or different approaches to pseudopotentials.

Is there a way I can replicate their results in ABINIT? In the case not all parameters are transferrable, would the discrepancy between running ABINIT simulations and their results be too large?

In order to answer that second question I could simply compute the results and compare, however I first need to understand how to compute the formation energy reliably, hence the first question.

Thank you everyone for your help. I recognize that I am probably asking for a lot, but I would be really grateful even for some slight hints and suggestions, not necessarily the full answers.

2

First, if you want the energy of a neutral atom in abinit, you have to do the calculation in an empty box (as is done for H2 in the tutorials). This means, for example, acell 20 20 20, xred 0 0 0,
ntypat 1, typat 1, natom 1, a single k point at 0 0 0, and so forth. Some atoms have filled shells, these are easy–start here to make sure you know what you’re doing, say with Neon. For other atoms, you’ll have to access a first-year chemistry book to have at hand typical orbital filling patterns, that will help you set nsppol and nspden appropriately.

As for a translation table between VASP and abinit, I wouldn’t do it that way, I’d suggest instead carefully doing the first tutorials of abinit, what we call the “base” tutorials, and then keep rigorously checking yourself for convergence and so forth as you study various compounds. Regardless of whatever marketing says, no code, including VASP and abinit, have a set of parameters that “just works” for everything. It’s critical to test as you go and be skeptical all the time.