Hi, Marcos,
Many thanks.
Previsously, I treated periodic structures, the LDA method (with CAR
XC.authors) worked well, but this time, it didn't work, and even gave the
strange bond lengths,
such as c-c bond length 1.08 Angstrom, big difference of bond length in
different phenyls of the same molecule, for example, bond length in one
phenyl is 1.35 A, and the one in
another phenyl is 1.12 A.
Ok, I will try as your suggested.
The detail information of the molecules I am dealing with:
1) All the molecules contain several phenyl, carbon-carbon bonds(single-,
double- and triple-bond), and this molecules are sandwiched by Au electrodes.
these clusters consist of other elements too, like S, H, N, O, at least
now. These molecules have about 50 atoms. These molecules are all interested
in by
experimentalists, and I also want to insight the origin of their
experimental phenoma.
2) I couldnot use supercell. Even my computer has 2 G memory, it still could
not cope with the systems as big as the supercells of these molecules-contained
clusters.
As to the above clusters, molecules-sandwiched clusters, I guess it could
be better when I use supercell even with LDA method, i.e. gives a reasonable
geometry
(maybe I am wrong at this point)
I am looking forward to your additional advices on the above molecules, and
your help and advice are greatly appreciated.
Best regards,
David
-----Original Message-----
From: Siesta, Self-Consistent DFT LCAO program, http://www.uam.es/siesta
[mailto:[EMAIL PROTECTED] Behalf Of Marcos Verissimo Alves
Sent: 2005年2月16日 0:44
To: SIESTA-L@LISTSERV.UAM.ES
Subject: Re: [SIESTA-L] problem with geometry optimization
Hi David,
How shorter or how longer are the bonds? What kind of XC functional are
you using? In my calculations for PPV, I found that GGA performs
adequately for obtaining the geometry of these molecules. Are they
polymers, or small moelcules? Depending on several factors, your
calculations may yield good or bad results... I know it seems vague, but
that's it... It would be nice to know:
1) What elements you are working with (Carbon for sure, but anything
else?);
2) Details on the energy shift (0.2 eV is fine for C)
3) Basis size (200-250 Ry for DZ, 300 for DZP gave me very good results)
4) Supercell size
5) XC functional - the best bet for these molecules would be a GGA. PBE
works fine, but there is one (not sure which, I think Becke's with LYP,
but I may be grossly wrong) that is more suitable for molecules than
PBE.
Of course you will never get perfect agreement with experimental results
for a number of reasons, but if bond lengths and angles are within 1-3%
shorter or longer than experiment, then you're doing fine, I guess.
Regards,
Marcos
On Tue, 2005-02-15 at 13:44, Xiao changyong wrote:
> Dear all,
>
> In my calculation, I always meet two problems:
>
> 1)While optimizing two organic molecule geometries, the bond lengths always
> deviate from normal values, to be too short or too long.
>
> I have tried several methods, (this has annoyed me for a long time)
> including:
> a)abandonded polarized orbitals,
> b) defining cutoff radius in PAO.Basis, or using smaller or default
> PAO.EnergyShift to define the cutoff radius.
> c) first get optimized geometries from quantum chemistry, and then
> using small MD.MaxCGDispl in siesta.
>
> 2) From time to time, I got the prompted information: FERMID: ITERATION HAS
> NOT CONVERGED.
>
> FERMID: QTOT, SUMQ= 447.999999999999500 448.000000000496700.
> I used the method by increasing DM.Tolerance to 1, but it still ddidn't
> work.
> In fact, I also found another method by increasing meshcutoff which can
> work effectively, but at end the meshcutoff will be too larege that beyond
> my computer capability.
>
> How could I cope with these problems?
> Does anybody kindly share the experience?
>
> Bye the way, i am using siesta1.5 version.
>
> Regards,
> David
> ==========================================================
> David, Changyong XIAO, Research Fellow
> Department of Chemistry, National University of Singapore
> 3 Science Drive, Singapore 117542, Singapore
> Tel: (65)6874-7880; Fax: (65)-6779-1691
> E-mail: [EMAIL PROTECTED]
> ==========================================================
--
Dr. Marcos Veríssimo Alves
Departamento de Física
Universidade Federal de São Carlos
São Carlos, SP - Brasil
http://joselitosn.df.ufscar.br/~dm/equipe/verissim.htm
***********************************
Dr. Marcos Verissimo Alves
Physics Department
Universidade Federal de Sao Carlos
Sao Carlos, SP - Brazil
http://joselitosn.df.ufscar.br/~dm/equipe/verissim-en.htm
