[gmx-users] Preparing a silicon bulk

[Kevin C Chan]

I apologize for the previous reply to the list with an ugly subject. I realise it only after I receive a digest. I hope it could be deleted if possible so that it would not be included in the archive as a new thread. Sorry, Users. 

Thanks for the reply.

In principle that is what I plan to do - generate a basic topology (with homogeneous charge information) and then manually (or a script) change the charge of each SI atom. Honestly I only have experience of writing a simple topology for a MARTINI system. So what I expect was first to include the gromos forcefield so I have the [defaults] [atomtypes] [bondtypes] defined and then manually write the [moleculetypes] and [atoms] in which I can define the charges (I assume it will overwrite the charge value defined in [atomtypes], right?) then the tens or hundreds of rows of entries were treated as one single molecule and finally the [system] and [molecules]. 

However I am now stuck at the step that it seems no forcefield provided by Gromacs gives bonded parameters for Si-Si so g_x2top could not work out. Or maybe I should give up g_x2top and just manually write the [bondtypes] according some parameters provided by the literature and combined with nonbonded parameters provided by gromos forcefield, for instance, but this then arises the consistency problem Oh god :(

Thanks in advance and looking forward to any comment or experience sharing.

Regards,

Kevin
ukevi at gmx.hk <mailto:ukevi at gmx.hk>


> 
> From: Justin Lemkul <jalemkul at vt.edu <mailto:jalemkul at vt.edu>>
> To: <gmx-users at gromacs.org <mailto:gmx-users at gromacs.org>>
> Date: 20 December, 2014 06:11:14 HKT
> Reply-To: <gmx-users at gromacs.org <mailto:gmx-users at gromacs.org>>
> Subject: Re: [gmx-users] Preparing a silicon bulk
> 
> 
> 
> 
> On 12/19/14 12:43 PM, Kevin C Chan wrote:
>> I am kind of confused, how could I actually reply and include my reply into a thread in the mailing list? As I am receiving digest email, I should not just press the reply bottom, or should I?
>> 
> 
> Replies work like any other email.  Click reply, change the subject, chop out the text that doesn't pertain to your issue, and send.  If digests are acting weird, switch to individual messages.  The list can be somewhat high-traffic, but certainly the last few weeks haven't been all that busy.
> 
>> Back to the question, thank you for providing the hint on unit of Gromacs, g_x2top could identify the bonds now. However g_x2top is still not working as:
>> 
>> Opening force field file /home/kevin/opt/gmx_fftw3_double/share/gromacs/top/gromos54a7.ff/atomname2type.n2t
>> There are 1 name to type translations in file gromos54a7.ff
>> Generating bonds from distances...
>> atom 8
>> Can not find forcefield for atom SI1-1 with 4 bonds
>> Can not find forcefield for atom SI2-2 with 4 bonds
>> Can not find forcefield for atom SI3-3 with 4 bonds
>> Can not find forcefield for atom SI4-4 with 4 bonds
>> Can not find forcefield for atom SI5-5 with 4 bonds
>> Can not find forcefield for atom SI6-6 with 4 bonds
>> Can not find forcefield for atom SI7-7 with 4 bonds
>> Can not find forcefield for atom SI8-8 with 4 bonds
>> 
>> And my atonname2type.n2t contains:
>> SI   SI         1     28.08   3    SI 0.2352     SI 0.2352      SI 0.2352
>> 
>> We could see that it identifies 4 bonds for each atom (but why not 3?), however no ff could be found. I look into gromos54a7 (I thought it contains Si forcefields) and I found no Si-Si bonded parameters in ffbonded.itp but a comprehensive set of non-bonded parameters of Si in ffnonbonded.itp. I am curious why it is like this. Did people only do single Si simulations?
>> 
> 
> Si should form 4 bonds; that's what satisfies its valence.  What g_x2top is telling you is: "I found 4 atoms within the bond distance, but you tell me in the .n2t file that there should only be 3."
> 
> The Si parameters in the force field were probably for silica, so no Si-Si bonds.
> 
>> Ultimately is it better (or easier) to manually construct the topology file?
>> 
> 
> In theory, you have a regular ordering of atoms on a lattice, so g_x2top is the perfect tool, but the desire to assign different charges makes that difficult. I would suggest building the topology with g_x2top and modifying as needed; that will require some scripting (or could be done by hand for a reasonably small system), but realize that changes in charges can affect the validity of bonded parameters, since everything in a force field is interconnected.  So even if you find suitable bonded parameters for the interactions you need, if you change aspects of the nonbonded parameters, the bondeds can be compromised as a result...
> 
> -Justin
> 
> -- 
> ==================================================
> 
> Justin A. Lemkul, Ph.D.
> Ruth L. Kirschstein NRSA Postdoctoral Fellow
> 
> Department of Pharmaceutical Sciences
> School of Pharmacy
> Health Sciences Facility II, Room 629
> University of Maryland, Baltimore
> 20 Penn St.
> Baltimore, MD 21201
> 
> jalemkul at outerbanks.umaryland.edu <mailto:jalemkul at outerbanks.umaryland.edu> | (410) 706-7441
> http://mackerell.umaryland.edu/~jalemkul <http://mackerell.umaryland.edu/~jalemkul>
> 
> ==================================================
>