BoltzTraP2 Tutorial | Plotting Onsager Coefficients (conductivities, Seebeck coefficient)

This video shows you how to use BoltzTraP2 to compute and plot Onsager coefficients, e.g. Seebeck coefficient, Electrical conductivity (sigma), Electronic contribution to thermal conductivity (kappa), etc.
We both use the built-in BoltzTraP2 plotting utility, and I show you how these plots can be customized starting from a script shown in the video.
The script for plotting is available here, please consider starring the repo (: github.com/alg...

[@ComicBook-vf7wq]

Great tutorial I wounder how to get the -m convergence. I mean what to check and how to know that you have a converged value?

[@algebraiccontinuation9291]

Thanks for the comment, I think the best way to do that in my opinion is converging w.r.t. your DFT band structure. E.g. vary -m until your band structure from BoltzTraP2 looks as close as possible to your DFT band structure. The logic behind this is that most of the quantities computed from btp2 are some form of derivatives of the band energies (meaning those wiggles in the band structure are important to minimize!) Therefore the closer the band structure, the more 'converged' the results.

[@arjuns2112]

Thank you for the very informative video sir, I have a question. How do we plot the trace of the tensor quantity using scalar as mentioned in the github page. I tried to use it but is constantly getting error. Pls do support sir

[@algebraiccontinuation9291]

Thanks, and it's hard to say without knowing exactly what you're trying to do. You are trying to plot, e.g. Sxx+Syy+Szz? What's you're code to do this and what's the error you're getting?

[@Namsusi]

Thank you for the tutorial, I have a question. I heard that BoltzTraP2 could use approximation for relaxation time as 1/DOS, could you please let me know how to do it if possible. Thank you for your support

[@algebraiccontinuation9291]

See point 1 in the wiki gitlab.com/sousaw/BoltzTraP2/-/wikis/library You could easily implement it but you'd need to write the code to do so. You'd need to decide whether you want this to be E, or E, k dependent, and then just implement it in the code using your DOS data. As for how to do it, it's pretty much contained in your question. Presumably you have some methodological framework to get an approximation to tau as a function of the DOS. Then just find the relevant portion of the code (linked in their GitLab, you can read the wiki for the relevant info on non-uniform relaxation time) and implement.

[@abs5894]

Thank you for nice tutorial! The vasprun.xml one should select is of SCF step and not nSCF I assume. The btp2 bandstructure in my case looks similar but is not as smooth as DFT one. How much smooth must it be to proceed further? I increase -m factor to 50, yet the smoothness didn't improve noticeably.

[@algebraiccontinuation9291]

You should definitely use an nscf calculation on as tight of a k-mesh as you can afford.

[@abs5894]

@@algebraiccontinuation9291 Thanks for reply. When I used nSCF vasprun.xml, the integrate command kept on running without giving out any output files. When I used SCF one, it outputted. Can you please shed light on this? And on -m factor I mentioned earlier?

[@algebraiccontinuation9291]

You'd probably need to share more info as to what error messages you're getting. Check that all the band info is in the required files for your nscf calculation as well. You can increase -m as much as you want, but if there isn't enough data from the DFT data, the interpolation won't mean anything.

[@abs5894]

@@algebraiccontinuation9291 Is it possible for you to share your email ID where I can send attachments on the input and output I obtained? It would be a great help.

[@abs5894]

@@algebraiccontinuation9291 Is it possible for you to share your email ID where I can send attachments on input and output I obtained? It would be a great help.

[@NiloyMahanta-tz9pv]

Sir, thank you for your valuable guide. I am confused to change the parameters in the script to plot the power factor with carrier concentration. I tried it the way you taught. But it shows the same plot : seebeck vs carrier conc. Please help me

[@algebraiccontinuation9291]

You have all of the values in the condtens files, you just need to find what you're trying to compute, compute it within the script by selecting the relevant columns, and then plot accordingly. Probably you need to learn some python for this if you're stuck.

[@NiloyMahanta-tz9pv]

@@algebraiccontinuation9291 thanks sir how to get plot chemical potential in electron Volt

[@algebraiccontinuation9291]

Same as above, if you want more customized plotting you unfortunately need to know enough Python to modify the scripts. Alternatively you can take the raw data into some other plotting software of your choice and manually make the changes.

[@RajnarayanDe]

Sir I have another question... I have used slab model and added vacuum layer to build a monolayer. Does the thickness of this vacuum layer affects sigma (conductivity) values extracted from boltztrap2? Actually I am getting one order less value (with 20 angstrom vacuum) as compared to the literature (with 15 angstrom vacuum)

[@algebraiccontinuation9291]

It depends on what you're comparing specifically but yes it does matter. The entire cell is used for any volumetric quantities in btp2, so you need to adjust all of those manually. You also need to be very careful about your input k-point mesh for 2D systems.

[@RajnarayanDe]

@@algebraiccontinuation9291 Thank you so much for the reply... I did nscf calculation with 30×30×1 k mesh.

[@RajnarayanDe]

Thank you sir for the nice informative video.. I am trying to compute PF and ZT using your matplotlib code. So, for PF I am calculating like, PF= ((Sxx+Syy+Szz)**2 * (sxx+syy+szz))/3 Is it correct? Also, please guide on how to get the kappa/tao values from the condtens file so that I can calculate ZT

[@algebraiccontinuation9291]

Hi, you need to determine the correct way to combine the components for your particular situation, but the division by 3 wouldn't make sense for the above. Get your finalized quantities before combining into the PF. Also for ZT, do you have lattice thermal conductivity from another code? Without that you can't compute ZT, even with the electronic thermal conductivity from btp.

[@RajnarayanDe]

@@algebraiccontinuation9291 Thank you so much for the reply... Should I have to compute the trace of S and s seperately and then combine to get PF? Please guide me.. For ZT, I am taking a fixed value ok kl as also used by some researchers. I tried to extract from phonopy but it seems the structure should be Ibrav=0 and I am using R3m system

[@algebraiccontinuation9291]

However you combine them to get S, and sigma is something that you need to decide depending on your material (which in this case sounds like 2D), that's not something someone else can decide for you. For ZT, I'd be very skeptical of just taking a value from literature. If you are going to do it anyway, you can find the columns in the condtens files that contain the kappa rows and then extract them in the same way that the Seebeck coefficients and conductivities are in the scripts.

[@RajnarayanDe]

@@algebraiccontinuation9291 Thank you Sir...

[@Maria_Clarita]

Thank you for this video. Could you please explain to me how can I get hole concentration and electrons concentration from carriers concentration [e/uc]?

[@algebraiccontinuation9291]

Hi Maria, just determine by the volume of your unit cell what the conversion factor should be to go from e/uc * uc/volume = e/volume. Also see my video on Doping Semiconductors where another way to do this is explained.

[@markedwincleofe5589]

Very informative video! I just subscribed! Thank you. I'm new to this field, could you please teach me how to get the cell volume you mentioned at 5:58? Right now, I'm using QE instead of GPAW. Thanks in advance!

[@algebraiccontinuation9291]

Hey you can either simply compute it from your cell vectors manually, or you can read your structure into ASE in the python script. I believe ASE can read QE input files, so it's as simple as: import ase.io atoms = ase.io.read("your_file") volume = atoms.cell.volume I believe that this volume will be in A^3, so you can do the same conversion below as in the script in the video.

[@markedwincleofe5589]

@@algebraiccontinuation9291 Thank you for replying, Sir. Just one follow-up question, I noticed that when plotting vs. chemical potential and vs. carrier concentration, the plot appears to be flipped(?) Is there a reason to this? Thank you and more power!

[@algebraiccontinuation9291]

Yes, think of what type of charge carriers are induced when shifting the chemical potential towards one of the band edges.

[@user-bc9zz2hd5x]

Sir is it the volume of conventional unit cell, primitive unit cell or the cell I have used in DFT calculation?

[@user-bc9zz2hd5x]

6:00 Respected sir , kindly tell me one thing that you have mentioned here to use volume of unit cell. But my question , is it the volume of conventional cell, primitive unit cell or the unit cell which I have used in DFT calculation?

[@algebraiccontinuation9291]

It should be the one from the DFT calculation. But you should never use anything like a supercell for BoltzTraP2.

[@user-bc9zz2hd5x]

@@algebraiccontinuation9291 I have used a primitive unit cell in DFT calculation, so I have to put the volume of primitive unit cell in boltzTrap2 calculation. Is it correct? And when I will use conventional cell in DFT, then I have to use the volume of conventional unit cell in boltzTrap2 calculation ,,, is it?

[@algebraiccontinuation9291]

Yes correct.

[@user-bc9zz2hd5x]

@@algebraiccontinuation9291 thankyou sir ,,, but sir, suppose I want to study the effect of vacancy on transport and thermoelectric properties. And for introducing vacancy, I need a supercell. In that case which cell volume should I use in boltzTrap2 ?

[@algebraiccontinuation9291]

You'd need to implement something to unfold bands, it's not implemented in BoltzTraP2.

[@esmasemassel5662]

hello, sir, can you help me, I installed Boltztrap2 and we displayed it in the Wien2k interface, when we execute the calculation step, you did not display the result, the .trace file is empty .Thank you

[@algebraiccontinuation9291]

You might have some sort of issue with the Wein2k calculation, there's no way I could know without more information though.

[@Arshadkhan-zz3nf]

How to plot Seeback vs Temperature at constnt chemical potential

[@algebraiccontinuation9291]

Check out my most recent video on Doping Semiconductors

[@arjunrathod6498]

how to generate .gpw file?

[@algebraiccontinuation9291]

See this video kzbin.info/www/bejne/g3aQoJeHmJugmKs