Zero Point Energy in VASP

[askhetan]

I want to calculate change in Gibbs free energy during an electrochemical reaction using the formula d(G) = d(E) + d(ZPE) - Td(S), (where d - delta (change)). Some authors have previously used the E and ZPE(zero point energy) in this equation from DFT calculations.

While it is obvious that E is certainly the E0 value in the stdout file, where can I find ZPE ? I am confused because the definition of ZPE says it is the minimum energy at 0 K(wikipedia). But isnt that what DFT itself calculates by giving us E0???

Apparently, from what I see in others's posts in this forum, ZPE is obtained with tags such as VOSKOWN and POTIM. I am not doing MD or isolated molecules, but O2 molecule on an oxide slab, so I guess specifying these for periodic slab calculations isn't so crucial for such a system.... or is it??? Where can I get this ZPE value from and what does it mean?. The VASP manual unfortunately couldn't throw light.

[alex]

Hi askhetan,

for basic understanding I suggest

http://www.gaussian.com/g_whitepap/thermo.htm

for reading. It is a complete and general article about how to get thermochemical data from ab-initio/first-principles calculations. You might also try a physical chemistry textbook to answer your question in paragraph 2.

Cheers,

alex

[askhetan]

Cheers Alex! That document indeed has quite the kind of info that I like to have before black box calculations.
I am still wondering, though, how to get the ZPE values from VASP.

[alex]

Hi,

make a frequency calculation, add up all non-imaginary modes, devide by 2, done.

Cheers,

alex

[askhetan]

Thanks for the info, unfortunately I'm a beginner. I spent the day first trying to understand what kind of vibrations are being talked about and they are certainly not molecular vibrations.

vibrations contributing to ZPE are atomic vibrations of some residual kind which are not mechanical in nature but more electromagnetic.... please sorrect me if I am wrong.

Now what I need to know is:

1) What do you mean by a frequency calculation? which frequency is being implied here?
2) How do you do such a calculation?
3) In which output file do you find these frequencies? and are these really listed as frequencies? or rather energies corresponding to those frequencies?

ALSO, if possible can you direct me to a page with info about this (both theoretical info with equations and practical info for VASP). Searching the VASP manual takes up the day with no result at all

[alex]

Hi,

well, VASP is not just some kind of app. It's a complex program and unless you understand what you are doing you should read a lot firstly. E.g. textbooks, cited articles and maybe visiting a workshop is also a good idea.

This forum is not for education (as far as I understand), but for questions about 'how to do this or that'. From my experience it takes e.g. a PhD student about 3-6 months training to feel comfortable with the program.

My advice: search somebody or some group in your university who might be able to train you.

Cheers,

alex

[askhetan]

I know, but the problem here is that I am the sole member in my research group and have to learn it all by myself.

thanks anyways

[tlchan]

[quote author= d(E) + d(ZPE) - Td(S), (where d - delta (change)). Some authors have previously used the E and ZPE(zero point energy) in this equation from DFT calculations.

While it is obvious that E is certainly the E0 value in the stdout file, where can I find ZPE ? I am confused because the definition of ZPE says it is the minimum energy at 0 K(wikipedia). But isnt that what DFT itself calculates by giving us E0???
[/quote]
The Kohn-Sham equation in DFT considers only the quantum mechanics of the electrons. The zero-point energy of the electrons are indeed included in the total energy. Note that the dynamics of the ions is still classical, although the force is evaluated quantum mechanically. I believe the zero-point energy that you are seeking is related to the vibrational modes of the atoms. You can perform the calculations as suggested by Alex above.

[askhetan]

Thanks! tlchan - that stands pretty clear, that i will need to perform frequency calculations for vibration modes. what I am searching for is the file where there values are reported. from the hands on session III of vasp, i see that i need to just copy the contcar to poscar and do another calculation. Because I have been using IBRION =2, i don.t need to set POTIM... moreover, I checked for optimum POTIM and found trial step in the stdout quite close to 1. What extra parameters do I need to specify?

In the end of the OUTCAR file, I cannot find any frequency values being reported in any format. so could you please just describe to me in steps how to get to this value. the fundamental questions about its meaning and all I will try to get from some other sources.

Thanks

[alex]

Hi ashketan,

I see your point. To get you started, some INCARs:

for optimisation (very(!) tight criteria, you may experiment a little). This setting is for large slabs.:


SYSTEM = ABC

Electronic minimisation
GGA = PE
PREC = Accu
EDIFFG = -1.e-6
EDIFF = 1.e-8
ENCUT = 400.
NELMIN = 4

DOS related values
SIGMA = 0.1
ISMEAR = 0 ! Gaussian smearing for big cells

Ionic relaxation
NSW = 350 number of steps for IOM
IBRION = 2 ionic relax: 0-MD 1-quasi-New 2-CG
ISIF = 0 ionic relaxation + stress tensor calcn.

Do not write huge files
LCHARG = .FALSE.
LWAVE = .FALSE.

Performance options
NPAR = 2 ! adjust to number of processors
ALGO = V


For a frequency job you replace the section below "Ionic relaxation" with

IBRION = 7 # switches on the DFPT vibrational analysis (with no symmetry constraints, 8 does the job with symmetry)
ISIF = 0 ionic relaxation + stress tensor calcn.
LEPSILON = .TRUE. # enables to calculate and to print the BEC tensors
NSW = 1 # reduces nr. of ionic steps to 1; unlike for IBRION=5, here the setting is very important
NWRITE = 3 # affects OUTCAR verbosity: explicitly forces SQRT(mass)-divided eigenvectors to be printed
# well, and you have to remove NPAR like
# NPAR = 2


NWRITE = 3 creates huge(!) OUTCARs, be careful. In the online manual you may check what property is really needed by yourself.

To search for your frequencies, try
'grep THz OUTCAR' (no tics, capital H)


Cheers,

alex

[cappri]

Hello Alex,
May I have an answer from you about the same Question raised by Ashketan before in simple words like you run a calculation for an adsorbate on metal surface where I would go in outcome file to calculate the ZPE + TdS

[alex]

Hi cappri,

you'll have to do the calculations outlined above: Structure optimisations (tight criteria) and frequency calculation later on top of these optimised structures.
In the meantime you'll have to read and understand ;-) the white paper above. If there is any question left, please come back again.

Cheers,

alex

[askhetan]

Cheers Alex!

That was of great help. I come back to this thread because now I also have to use these same values from the vibrational analysis for the improved dimer method (IDM). I believe you could give me a tip or two. When I do the IDM, which vibrational modes do I need to set in? of the initial folder (00) or of the final folder (0X+1) ? The UT Austin page (from the Henkelmann group) also mentions about additional tags such as ICHAIN, etc. but I do not find any reference to those tags on the vasp website. WOuld be great if you could please help me on this! Danke Schoen!

[perez2013]

hi

have a look at this paper:
X Gonze, P Boulanger, M Cote, Ann Phys (Berlin) 523 No 1-2, 168-178 (2011)

Manuel