[DFTB-Plus-User] Am I fully optimizing the charges of my structure?

[Alathea Davies]

Hello!



I am working on optimizing the geometry/lattice of COFs that have been synthesized by an experimental group I am collaborating with, and to extract properties of those COFs, such as charge difference, band structure, density of states, etc. The version of DFTB+ that is currently compiled with our HPC cluster is version 20.1. I’ve been having some issues generating a band structure, and I’m wondering if there’s an issue further back in the geometry optimization. I have two sets of questions related to that in here.



Since these structures aren’t published in literature, I’m using Gaussian/Gauss View to draw the unit cells and export the coordinates to use with DFTB+. I begin my geometry/lattice optimizations with very low SCC and forces tolerances (1e-1) to slowly bring the lattice and atoms into the correct arrangement. Once the geometry and SCC have converged at low tolerances, I’ll bring the tolerances down in a stepwise manner (1e-1, 5e-2, 1e-2, etc.) to the DFTB+ defaults. Since I am trying to find the energetic minimum, I’ve been trying to converge these structures at a Fermi Temperature = 0 K. I’ve attached an example of this script with the input geometry and charges that I used (COF-geo-opt-dftb_in.hsd, COF-geo-input-POSCAR, COF-geo-input-charges.bin). I have been having an issue getting this geometry to converge at force tolerances lower than 2e-3 without increasing the Fermi Temperature.



First set of questions: If I converge this geometry at a Fermi Temperature of 100 K, is that at a true energetic minimum? Will giving a Fermi Temperature above 0 K generate an inaccurate band structure or density of states, if the charges aren’t fully optimized?



I took the geometry that converged at 2e-3 force tolerances, optimized the SCC to 1e-5 to generate “optimized” charges (COF-charge-relax-dftb_in.hsd), and then tried to get a band structure from that output (COF-band-struc-dftb_in.hsd, COF-band-input.gen, COF-band-charges-input.bin). Initially, I got the following error:



-> External file of charges has a total charge:        672.000000, instead of         600.000000



I then added a line under “Hamiltonian = DFTB { … Charge = -72 … }” to attempt to offset this charge discrepancy. I have a suspicion this didn’t truly solve my initial problem and just created more of them. For instance, the Fermi level output from this successful run was 32.8 eV, which doesn’t seem realistic.



Another issue that I believe I’m having is that, despite the SCC and geometry converging previously and reading those output files into the band structure script, in the new script, which is reading in charges from the converged calculation, the SCC doesn’t converge in one iteration for the band structure as I expected it to, according to the DFTB+ Band Structure<https://dftbplus-recipes.readthedocs.io/en/latest/basics/bandstruct.html> recipe. In the example file attached, when I added the “Charge” command, it took 35 iterations for the SCC to converge (Band1.log).



Second set of questions: I’m suspecting that the charges are not fully optimized for this structure, and that might explain the total charge discrepancy and needing multiple iterations to converge, despite previously converging. If this is the case, what could have gone wrong? How do I make sure the charges are fully optimized for a band structure or density of states calculation?



Thank you so much for your time and help!

~ Alathea Davies

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Band1.log
Type: application/octet-stream
Size: 114600 bytes
Desc: Band1.log
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment.obj>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: COF-band-input.gen.txt
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment.txt>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: COF-band-inputs-charges.bin
Type: application/octet-stream
Size: 5996 bytes
Desc: COF-band-inputs-charges.bin
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment.bin>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: COF-band-struc-dftb_in.hsd.txt
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment-0001.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: COF-charge-relax-dftb_in.hsd.txt
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment-0002.txt>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: COF-geo-input-charges.bin
Type: application/octet-stream
Size: 6428 bytes
Desc: COF-geo-input-charges.bin
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment-0001.bin>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: COF-geo-input-POSCAR.txt
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment-0003.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: COF-geo-opt-dftb_in.hsd.txt
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220218/8bd07af3/attachment-0004.txt>