DFTB simulations of liquid water

[Guillaume Stirnemann]

Hi all, 

I am trying to simulate a box of liquid water using DFTB with periodic boundary conditions. My initial configuration is taken from a previous classical MD run performed in the NVT ensemble at the experimental density. 

I have tried different set-ups, including SCC-DFTB ("DFTB2") and DFTB3-diag, both with or without Grimme's D3 corrections, all ran in the NVT ensemble for 50 ps. 

In all cases, the simulations run well, but I end-up with a large hole in the simulation box. When comparing my radial distribution functions that of Cui who used DFBT+ (JPCB 115, 6790 (2011)), the first peak of the O-H rdf is usually a bit more pronounced but the contraction of the whole system is more visible on the second hydration shell. The differences are less visible for the O-O rdfs. The hole occupies about 1/2 of the simulation box, so the final density of my liquid phase is about 2! 

I attach the initial configuration, and the cp2k input file for DFTB2 with no D3. I tried on two different machines, one with 6.1 and the other one with 5.1. 

I am not an expert of CP2K so there might be some issues with my input, and your help would be very much appreciated. 

Best wishes

Guillaume 

CNRS Institut de Biologie Physico-Chimique, Paris, France

[Thomas Kühne]

Dear Guillaume, 

I have a couple of (hopefully helpful) comments using DFTB within CP2K: 

1. To the best of my knowledge only the mio and znorg parameter sets, 

i.e. the ob2 parameter set needs to be downloaded from dftb.org

2. IMHO, the usage of a damped gamma does not represent a general 

improvement, it marginally improves the 1st peak for water but appears 

to be worse for a lot of other systems

3. I rather suggest to use the improved parametrization of Dömer et al., 

see e.g. https://arxiv.org/pdf/1408.5161.pdf and references therein. 

Since I couldn’t find the files anywhere else, I have attached them to the 

present email. We will check them again and upload them to the repository 

next week. 

4. For comparison have a look at Fig. 1-3 in https://arxiv.org/pdf/1512.08206.pdf 

for a plain DFTB calculation of liquid water by CP2K with the provided original 

parametrization, without the modified gamma and no vdw correction, but including 

nuclear quantum effects. No inhomogeneity in the particle density have been observed. 

5. Please also have a look at the recent discussion about the SPME settings, 

which may play a role in your observed behavior.

Best, 

Thomas

[Guillaume Stirnemann]

Dear Thomas, 

Thanks a lot for your comments. I ran quick tests following some of your suggestions and I think I have figured out where the problem might come from. It seems to be related with the use of damped gamma. I tried a short simulation (20 ps) without the gamma scaling on and using the standard cp2k parameters for SCC-DFTB, and everything seems fine. Now, this is very strange because this did not seem to cause any problems when used in DFTB+ (JPCB 115, 6790 (2011)). Has anyone some experience with the use of damped gamma in cp2k?

Thanks for pointing out some recent optimizations of the DFTB2 parameters (by the way, in the files you sent the hh spl file seems to be missing, right?). However, I would eventually need to perform simulations on molecules containing atoms other than C, N, O or H, so I may have to stick with a less-accurate set of parameters. Some more related questions below: 

1) From what I understand, the spl files provided with cp2k corresponds to the mio 0-1 parameter set. Is this right? Some parameters (e.g. for phosphorus) seem to differ from the current mio 1-1 version. 

2) For the molecules I am interested in, DFTB3-diag would probably be more suitable. But the problem is that it was parametrized with the damped gamma for hydrogens, for which there is the issue discussed above. 

3) I would like to try to include the D3 corrections for dispersion interactions. The only thing which is not clear to me is the values I have to specify for D3_SCALING. Should I follow the values mentioned for the zero-damping case in the DFTB+ manual (see https://www.dftbplus.org/fileadmin/DFTBPLUS/public/dftbplus/latest/manual.pdf p45)? However, they seem to correspond to full DFTB3, whereas cp2k only has SCC-DFTB and DFTB3-diag. If anyone has experience regarding which D3_SCALING to use in that case, this would be great. 

Thanks again! 

Best 

Guillaume 

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.
<input-DFTB2-gamma-noD.in><h2o.xyz>

==============================

Prof. Dr. Thomas D. Kühne

Dynamics of Condensed Matter

Chair of Theoretical Chemistry

University of Paderborn

Warburger Str. 100

D-33098 Paderborn

Germany

tdku...@mail.upb.de

+49/(0)5251/60-5726

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.

[Thomas Kühne]

Dear Guillaume, 

as written in the paper only modified O-O & O-H parameter sets were 

provided, i.e. the H-H interactions should be taken from the original mio 

parameter set. To the best of my knowledge, the latter is indeed the 

0-1 version. 

Best, 

Thomas

[hut...@chem.uzh.ch]

Hi

this phenomenon has been seen by us and others before. See
Choi et al., JPC B 117 5165 (2013). We have reparametrized
the gamma_H correction to get better results for pure water
(only published in the PhD of Yannick Misteli, chapter 3.7)

best regards

Juerg Hutter
--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491
Institut für Chemie C FAX : ++41 44 635 6838
Universität Zürich E-mail: hut...@chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp...@googlegroups.com wrote: -----
To: "cp2k" <cp...@googlegroups.com>
From: "Guillaume Stirnemann"
Sent by: cp...@googlegroups.com
Date: 12/26/2018 04:33PM
Subject: [CP2K:11087] DFTB simulations of liquid water

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.

[attachment "input-DFTB2-gamma-noD.in" removed by Jürg Hutter/at/UZH]
[attachment "h2o.xyz" removed by Jürg Hutter/at/UZH]