Need advice for setting CP2K input for all electron CAM-B3LYP calculations

[Kun-Han Lin]

Dear CP2K users,

Recently, I want to use CP2K to do umbrella sampling with the QM/MM method. The system is composed of  a Cl anion and neutral molecule (closed shell system). 

I first compared the static energy-distance curve calculated by CP2K and Gaussian under the same QM condition (CAM-B3LYP, 6-311G**, I also tried adding diffuse function in Gaussian, the results are similar to that without diffuse function). However, the results from CP2K give incorrect long-range behavior and the SCF also does not converge (especially in when the monomer are largely separated, I put in CP2K result.pdf file). Therefore, I'd like to know if there is some error in my input file. I attach it in this file as well (CAM-B3LYP.inp). Note that we do need to use a range-separated functional for describing this system properly.

Finally, I'd like to know if there is some setting to make this long-range exchange computations accurate and efficient. Since we want to do sampling, efficiency is a critical issue. If you have advice or any suggestion, please let me know. Many thanks in advance.

Best,
Kun-Han Lin

PhD student LCMD
Ecole Polytechnique Fédérale de Lausanne (EPFL)
EPFL SB ISIC LCMD
BCH 5312 (Bât. BCH)
CH-1015 Lausanne
Switzerland

E-mail: kun-h...@epfl.ch

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

Hi

your definition of the HF section is not correct:

&HF
&SCREENING
EPS_SCHWARZ 1.0E-12
&END
&MEMORY
MAX_MEMORY 100
&END
&INTERACTION_POTENTIAL
POTENTIAL_TYPE MIX_CL
OMEGA 0.33
SCALE_LONGRANGE 0.94979
SCALE_COULOMB 0.18352
&END
&END

If you want to reproduce the G09 results, I would also take the
WAVELET solver (not MT), remove the XC_GRID section (no smoothing)
and maybe increase the cutoff.

You can find some examples in tests/QS/regtest-hybrid-3 and 4.

If you want to optimize speed I would use GPW and pseudopotentials
together with the ADMM method.

regards

Juerg
--------------------------------------------------------------
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: Kun-Han Lin
Sent by: cp...@googlegroups.com
Date: 12/06/2016 01:10PM
Subject: [CP2K:8462] Need advice for setting CP2K input for all electron CAM-B3LYP calculations

--

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 "CP2K result.pdf" removed by Jürg Hutter/at/UZH]
[attachment "CAM-B3LYP.inp" removed by Jürg Hutter/at/UZH]

[Kun-Han Lin]

Dear Professor Hutter,

Thank you for your advice. I did some test for my system. Since I need a large cell for the umbrella sampling coordinate space, it took me 

lots of time to converge. Maybe I'll try something else or other softwares. Thanks for your help.

Best,

Kun-Han

[Samir ABDELOUAHED]

Dear Juerg,

I am facing a serious problem getting the right HOMO-LUMO gap for an isolated C60. I've a pending draft on C60 adsorption and I don't have any more the ADF license. So I thought cp2k might be the  best alternative. So I started calculating the C60 gap. The problem is that I am still far from the C60 experimental gap of around 5 eV.

the PBE, B3LYP , and PBE0 (using either GTH or MOLOPT DZVP basis set) calculations gave 1.62 eV, 2.64, and 2.90 eV, respectively.

As the ADF B3LYP All electron-STO basis set calculations gave around 4.7 eV I thought that a cp2k B3LYP all electron-GTO basis set would give a similar value...But apparently it is not the case. Indeed, I've got similar values to the pseudopotential calculations: 1.65, 2.69, and 2.95 eV.

So I am wondering whether this is due to the fact that the ADF calculations were done using STO, which are more efficient than GTO. If it is the case which other GTO basis should I take to get the right gap. Or is there some problems in my settings (inputs).

Please find attached the inputs for these calculations (6 inputs).

I would like to tell you that the calculations are converged without any problem even the hybrid ones as I started these calculations from a PBE pre-converged calculation. 

Many many thanks for your help,

Samir

[Matt W]

This paper cites a couple of sources saying the C60 HOMO-LUMO gap as 2.77 at B3LYP/6-31G**

https://journals.aps.org/prb/abstract/10.1103/PhysRevB.83.115103

[Samir ABDELOUAHED]

Dear Matt,

Thank you so much for the quick reply.

One of the powerful feature of cp2k (which is not available in most of the other codes as far as I know) is the ability/possibility to use different XC for different parts of the same material.

As I am going to study the adsorption of C60 on graphene/graphite I would like to use B3LYP/PBE0 for C60 and PBE for graphene. This is because PBE is suitable for graphene or metals while B3LYP does a great job for insulators and molecules with a big HOMO-LUMO gap.

A very similar work is already published using cp2k for MgO on Ag(001): https://pubs.acs.org/doi/abs/10.1021/jp311141k

As I have the same atom kind for C60 and graphene I don't really know how to set B3LYP for C60 and PBE for graphene or Carbon cluster.

Many many thanks again for your help,

Samir

[Matt W]

Hi,

On Monday, June 18, 2018 at 1:28:14 PM UTC+1, Samir ABDELOUAHED wrote:

Dear Matt,

Thank you so much for the quick reply.

One of the powerful feature of cp2k (which is not available in most of the other codes as far as I know) is the ability/possibility to use different XC for different parts of the same material.

As I am going to study the adsorption of C60 on graphene/graphite I would like to use B3LYP/PBE0 for C60 and PBE for graphene. This is because PBE is suitable for graphene or metals while B3LYP does a great job for insulators and molecules with a big HOMO-LUMO gap.

A very similar work is already published using cp2k for MgO on Ag(001): https://pubs.acs.org/doi/abs/10.1021/jp311141k

  

As I have the same atom kind for C60 and graphene I don't really know how to set B3LYP for C60 and PBE for graphene or Carbon cluster.

You just call them different names in the coordinate file, anything you like. Then have a KIND section for each atomtype, including ELEMENT to tell CP2K what they really are.

 

Matt