Exploring Real Materials with DFT: Ground-State Theory and Beyond

[assed]

Hello. I started studying DFT and using it for a bit of work I needed to do in College. I found it a bit interesting, because of the possibility of getting in touch with more "applied" science. However, after a time reading papers related to it I got a little disappointed. It seems to me that all papers resulting from work of these softwares (WIEN2K, VASP, SIESTA, QUANTUM-ESPRESSO, etc) all mostly just report results of calculations for DOS, Band Structure, etc. The problem is these publications usually do not give a relevant result about the studied material, and for me it seems there's no way that kind of result is actually used in applied materials science. Also, they keep discussing about the problems with the method, and that I think is very redundant and not productive at all: like why do these people calculate bad gap with DFT if everyone knows results are usually terrible. But what atracts me to this area of research is again, the possibility of take real materials and study them with calculations.
After this long introduction (if anyone is still reading) I ask to those with good understanding about this area: what do you think about what I said? Am I wrong? Finally, I have great curiosity about something: is it possible, using this method of research, to relate to more basic solid state physics, like observing an interesting fenomena or property of matter( I mean related to basic theory; this question is from someone who likes theory a lot but want to deal with thing closer to "real world").
Thanks.

 

[DrDu]

Your complaint about people discussing mainly the problems of the methods is well justified, however, this is good scientific practice (both to complain and to discuss problems :-). You can't just go out and say you have done some work and report some results, but you have to be earnest about the possible shortcomings.

Of course it is well known that band gaps are wrong in most DFT calculations and you have to say so. However you have to ask yourself whether you are really interested in the correct prediction of band gaps or whether it is not more important for you to have, e.g. the correct k-dependence, effective masses, transition moments and the like.

To me it sounds well possible that you just haven't found the relevant articles. As you are not very specific about what exactly you have in mind it seems quite hard to help you. Are you planning to study some specific substance or some specific properties?

 

[assed]

Thanks for your answer.
Concerning the possible areas I wonder if some useful materials research can be done qith DFT(of course I intend to learn another methods used in quantum physics/chemistry, but my main question is regarding this specific type of DFT softwares), is about electronic transport or (which interests me the most) magnetism (and not only take some important material used for its magnetic properties and do the DFT calculations to obtains DOS, Bulk Modulus, lattice parameters, etc.)
Thanks again.

 

[bsmile]

magnetism is the poorest thing LDA addresses, I would say. LDA is still developing, targeted on most of the things you are complaining about, although personally I don't think LDA itself is able to address these issues.

 

[collpitt]

DFT by itself cannot be used for studying electronic transport. Using DFT one basically derives the Hamiltonian for a many body system. The derived Hamiltonian is then transformed to obtain various electronic properties of the system. For transport calculations, one has to further apply a transport formalism; popular ones being NEGF, S-Matrix etc. For a stronger conceptual understanding you can look at "Mesoscopic transport" by S. Datta.
Magnetism is included in the Hamiltonian in the form of a vector potential. ("Electron Theory of Magnetism" Gustav Bihlmayer).

 

[t!m]

It is important to remember that DFT is a ground-state theory. As such, it can be very accurate for ground-state properties such as lattice constants, phonon densities of states, or relative stability of competing structures. It is not a theory of excited states and so (strictly speaking) DFT should have no right predicting band gaps or other excited state properties. In principle, time-dependent DFT is a theory for neutral excited states. Alternatively, many practitioners employ the GW approximation to correct DFT band structures (including the "band gap problem"), which can then be used to calculate densities of states (or photoemission). The Bethe-Salpeter equation can then be used to calculate the neutral excited states and optical properties of the material (remember, the band gap and the optical gap are not the same thing!). This should give you enough information to continuing learning about modern 'best practices' for rigorously studying real materials.