5
$\begingroup$

Independent from my general kinetics question here - Estimating Surface Reactions - Liquid, Turbulent Flow - I am only interested in responses regarding Gaussian in this question.

I would like to know if it is feasible to predict or model the influence of a surface (metal) on a reaction between two hydrocarbons (aromatic and aldehyde for example) in Gaussian. If yes, how would I go about it? (Building a small metal grid and working with it is very computationally intensive - it is the only approach I can think off...)

(I have access to Gaussian at my university - however I do not have access to Linda, so no shared nodes.)

$\endgroup$
  • $\begingroup$ I'm no expert with surface chemistry and I avoid metals like the plague. I do use G09 on a daily basis so I'm only speaking from the other side of the fence here but... you may be interested in something like ADF (Amsterdam Density Functional). Also, I think NWChem may be something to look into as well. Its freely available unlike ADF. $\endgroup$ – LordStryker May 20 '13 at 15:03
  • $\begingroup$ I doubt my university wants to buy anything - but I came across NWChem once before... but never looked at it in greater detail. Maybe I should - also metal surface effects have only become a focus because they are the only explanation for the reactions I am looking at (liquid phase is what we initially though but Gaussian suggests no...). Thanks. $\endgroup$ – DetlevCM May 20 '13 at 17:27
  • 2
    $\begingroup$ Well, do yourself a favor and find an alternative to Gaussian. There are options out there that will be much more efficient and better suited to modeling metals and surfaces. $\endgroup$ – LordStryker May 20 '13 at 19:29
  • $\begingroup$ @LordStryker That is one option - though there is always the issue of support. Heck, I even ended up with chemistry coming from mathematics and originally planning to do CFD - my topic has lead me into chemistry... $\endgroup$ – DetlevCM May 21 '13 at 8:32
7
$\begingroup$

You did go to Gaussian's Documentation on the subject, right?

In principle, it should work. In practice, you should consider using massive firepower, and the plane-wave codes (NWChem, abinit, VASP, etc. etc.)

Also, I question whether any of these codes properly model bond breaking on a surface, i.e. metal-catalyzed reactions. Big simulations don't guarantee correct wave functions.

I should point out that surface reactions with plane waves are also difficult. At least they used to be. There are/were issues with Ewald summation having to be done in 3D, and so the surface normal dimension used to have to be HUGE to avoid image effects.

This is a tough business.

$\endgroup$
  • $\begingroup$ Thank you for pointing me at this, however I cannot quite see how I can use that. All this tells me is "an option exists"... - Maybe I need to dig around this... (I will also be attending the Gaussian course in Wroclaw in June) Having said that - when you mention "massive firepower" I suspect I might run into trouble with my university's cluster - I do not have Linda available and the biggest node I can use is 12 cores and 24GB of RAM... (More cores = more nodes, therefore no Gaussian for me) $\endgroup$ – DetlevCM May 21 '13 at 8:35
  • 1
    $\begingroup$ If you google: Gaussian PBC examples, you get this page: scuseria.rice.edu/gau/gdv_pbc/pbc_guide.html $\endgroup$ – Eric Brown May 21 '13 at 12:40
  • $\begingroup$ That is a start - lots of fun - I will have a read of it and try to understand as much as possible (Maths background). I was also thinking of trying to get NWChem to run on our computing cluster - being free that should be doable. For now I gave you a +1 on your answer. $\endgroup$ – DetlevCM May 21 '13 at 12:46
1
$\begingroup$

Well, I am sure you found the answer (it's been 4 years since you asked the question), but you can use VASP (the Vienna Ab-Initio Simulation Package) or if you don't have or can't afford a licence for VASP, then Quantum Espresso is great too (though it's a pain to get a hang of if you are teaching it to yourself). Also, Schrodinger now comes enabled with Quantum Espresso plane wave code for solid state chemistry calculations.

$\endgroup$
  • $\begingroup$ NWChem is another option for plane waves that I did briefly use - but never went beyond testing (with the resources at the time). But even with a plane wave code, the problem is non-trivial due to the cell size. (Should be a lot better with the computers of today, but I moved on to other details of the same science area.) $\endgroup$ – DetlevCM Jun 2 '17 at 17:20
0
$\begingroup$

I have done some work with reactions on organic surfaces, and hybrid QM/MM calculations helped me solve my problem, particularly using the ONIOM forcefield. I understand that a classical forcefield may not be the best way to treat a metal surface, however.

$\endgroup$
  • $\begingroup$ I actually gave up on that avenue eventually - due to a lack of computational resources. In principle, ONIOM is an option. A colleague suggested freezing bonds which is another option. (But introduces its own issues.) $\endgroup$ – DetlevCM Jan 20 '15 at 10:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.