[CPMD-list] VDW

[Robert Williams]

Dear List,

I have a few answers for the questions
I raised regarding VDW corrections a
few months ago, and I've submitted
a paper describing what I've found,
listing a set of parameters that
can be used for the implementation in CPMD.
Preprints are available on request.

I have answered some of my questions below,
and I have attached an updated CPMD manual
entry for VDW corrections that includes LaTeX
source. (Axel, can this be added to the manual?)

On 12/10/2004 Robert Williams wrote:
> Dear List,
> 
> With respect to VDW CORRECTION,
> it is not clear from the manual or
> from the examples how
> the parameters should be chosen.
> 
> Of the parameters:
> type, i, j, \alpha_{ij}, R^{ij}_m, \beta_{ij}
> only the "type" of C6 is explained.
> (My notation is in TeX.)
> 
> i,j: I gather from the source that i and j are
> the indexes of the atom types in the order that
> they are listed in the &ATOMS section.
This, above, is correct.  It is also possible to
declair multiple atom types for a single element.
So, for carbon there would be two separate
pseudopotential entries.
> 
> \alpha_{ij}: has to be
> calculated by hand using tables I and II and equation 7
> from Elstner et al., I gather.
> However, when I do this I do not get values
> that are close to the values used in the
> cpmd/contrib/CPMD-test/vdw examples.  Please someone
> correct me if I've made a dumb mistake.
I have compiled pre-calculated parameters using
AU in the available preprint.  These values for
the vdW strength, C$^{\alpha\beta}_6$,
are indeed calculated from Elstner's values, but I
have added a scale factor that is required to fit
the DFT results to high level results (near the CBS limit).
The examples in ~contrib/CPMD-test/vdw are misleading
because they exclude HH and CH interactions,
which actually dominate the dispersion interaction
for the methane dimer, and to compensate - they
multiply the vdW strength by a factor of about 10,
without stating what has been done.

> It would be very
> useful to know what units are expected, since
> three of the papers I've read all use different
> units for the C_6 coefficients, J nm^6 mol^{-1} (S. Grimme),
> \AA^6 kcal/mol (S. Serra et al. Chem. Phys. Let. 331 (2000) 339-345,
> one of the references Axel refers to below),
> and in the Elstner paper, eV/\AA^6 (!?
> This must surely be a typographical
> error in this paper.  It should be eV \AA^6/mol.)
> The source is not informative here.
Parameters listed in my preprint are in AU,
Hartrees and bohr, appropriate for direct entry into
the VDW section of CPMD.

> The reference, M. Elstner et al.,
> gives a damping term in equation 8
> that appears to be different from that given
> in the cpmd manual under "VDW CORRECTION",
> so it is not clear how R^{ij}_m is determined.
> Likewise, \beta_{ij} (Elstner's "d*" in his eqn. 8),
> is also in question because the form of the damping
> terms is quite different in the manual compared to
> Elstner et al., a source of concern.
The damping function given in the present version of the
manual is incorrect.  It references a different damping
function that has been commented out of the source code.
The attached manual entry includes the form implemented
in the CPMD source: vdw.F.
> 
> After a cursory reading
> of the cpmd source, and running the examples,
> I am a little confused
> about how to choose atom types (i,j), since the types
> chosen for the ch4-ch4 calculation in the
> example both appear to be for carbon,
> where i and j are 1.  Should not the
> dispersion terms here be for H?
The present example is misleading.  Although
the vdW strength for the C-C interaction is
about ten times the interactions for C-H and H-H,
there is only one C-C interaction in the methane dimer.
There are six C-H and six H-H interactions, and
the distances for these interactions are smaller,
so they dominate.
> 
> 
> I'm waiting for an interlibrary loan to read
> the references by Le Sar (1984, 1987),
> and Elstner does refer to several other
> earlier papers that I've not seen yet, so
> perhaps I'm a little premature to submit
The references to Le Sar are for the commented code.

> this question to the list.   Still,  the
> manual should probably contain a little
> more guidance about the parameters,
> and about the difference between the implemented
> damping term and the one given in the reference.
Again, see the attached manual page, and request
the preprint.

A word of caution:  I've optimized the parameters
for the methane dimer interaction.  Although
the other parameters listed in this preprint may
be good starting values, there is no evidence yet
that they will produce good results.

Bob Williams

> Axel Kohlmeyer wrote:
> 
>> On Tue, 7 Dec 2004, Song Huajie wrote:
>>
>> HS> Dear CPMD experters,
>>
>> dear hj song,
>>
>> please have a look at the 'vdw' subdirectory of the
>> CPMD-test archive. it contains two examples for the use of this feature.
>>
>> you may also want to have a look at
>> chem. phys. lett. 331(2000), 339-345
>> and
>> chem. phys. lett. 360(2002), 487-493
>>
>> for two successful applications of the VDW correction.
>>
>> HS> To study the dispersion bound systems (organic clusters and 
>> solids), I
>> HS> intend to use the functionality (&VDW) of CPMD that adds van der 
>> Waals
>> HS> (dispersion) term to DFT. Unfortunately, The Manual of CPMD does not
>> HS> offer more information about it. I would like to know whether or not
>> HS> to activate the &VDW when performing energy minimization and (or)
>> HS> molecular dynamics. I am very hopeful that he who has been used this
>> HS> &VDW can response or offer some input files for carring out this
>>
>> i have not used the feature myself, but i only makes sense to me, to 
>> use it consistently for all calculations.
>>
>> best regards,
>>     axel kohlmeyer.
>>
>> HS> functionality. Your help will be appreciated.
>>
>> HS> With best regards,
>> HS>      H.-J. Song HS> HS> HS> HS> HS> 
>> _______________________________________________
>> HS> CPMD-list mailing list
>> HS> CPMD-list at cpmd.org
>> HS> http://cpmd.org/mailman/listinfo/cpmd-list
>> HS> HS>
> 
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: manual-vdw.pdf
Type: application/pdf
Size: 38519 bytes
Desc: not available
Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050315/05f1ee70/attachment.pdf 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: manual-vdw.tgz
Type: application/x-gzip
Size: 1861 bytes
Desc: not available
Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050315/05f1ee70/attachment.gz