Dear David, thanks a lot for your answer and excuse me please for long
request.
In principal I solve my problem by varying parameters in frcmod file and
providing constraint minimisation to compare results with MP2
calculations. It seems to be manual (and badly time consuming) analog of
what script on
http://www.ub.edu/cbdd/?q=content/small-molecule-dihedrals-parametrization
should be doing. I got hc-c3-Si-c3 parameter by working up TMS molecule,
and then went to more comlicated molecules to parametrise c3-c3-Si-c3
and so on. My check in with c3-Si-c3 angle force constant shows that
parmgen gives rather ok parameters. Nevetherless taking force constant
for hc-c3-Si-c3 equal to hc-c3-c3-c3 gives rotational barrier nearly
twice higher than it has to be (and so far hc-c3-Si-c3 force constant
had to be decreased). Same situation has place with c3-c3-Si-c3. Scan
profiles with new parameters looks quite ok and i will see when how the
system will behave in MD when will manage to start an MD.
If you find my procedure wrong - please let me know, and if you are
interested in more details - do not hesitate to ask.
Best regards,
Dmitriy
On 21.07.2015 14:58, David A Case wrote:
> On Tue, Jul 21, 2015, Sharapa wrote:
>> I want to make MD on a small organic molecule containing Si.
>> (trialkylsilyl group)
>> I thought on usage GAFF, but there is no parameters for Silicon.
>> I found http://archive.ambermd.org/200804/0353.html, but i am badly
>> confused that as far as i understand parameters simply copied from
>> carbon. That looks little bit weird for me.
>>
>> I tried to generate parameters by my own
>> 1) bond length and angles from DFT and MP2 calculations
>> 2) force constants by parmgen. Seems this way they depends only on a
>> bond length and not on a type of atom, that isn't true generally for
>> gaff.dat, but ok, they looks not too strange so i accept them.
>> 3) wdv-radii and 6/12 parameter that i found in papers fits to mentioned
>> on a link (Si 2.220 0.320) and seems i can ignore polarisation.
>> (I put my parameters at the end of the email)
> These look OK for a start. Of course, you should do some simple simulations,
> say of the Si-containing molecule by itself to see what problems exist.
>
>> Nevetherless i have problem with dihedrals.
> Most torsion terms should be small, and probably not that much different from
> the corresponding carbon terms.
>
>> I tried to follow
>> http://www.ub.edu/cbdd/?q=content/small-molecule-dihedrals-parametrization
>> and perform set of Gaussian MP4//MP2 calculations but didn't succed even
>> with tetramethylsilane molecule (multiple unclear problems with
>> run_all_sander_param.sh script).
> This is not a good way to get help: saying you have "multiple unclear
> problems" is not specific enough.
>
>> I tried to search answer on
>> http://q4md-forcefieldtools.org/REDServer-Development/ but the part i
>> liked the most FFParmDev seems does not exist.
> Again, we (or Francois) would need more specific information. It's easy for
> pages on a web page to become stale or unavailable. You can help by giving
> detail info on what you tried and what the error was.
>
> ....good luck....dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jul 30 2015 - 05:00:02 PDT
