How Can Beginners Start Computational Modeling in Heterogeneous Catalysis?

[jacone]

I have been working on experimental heterogeneous catalysis for a couple of years, specifically with metal oxides. I would like to learn computational modeling on heterogeneous catalysis (e.g hydrogen abstraction on acetone). I actually have no background on computaional calculations. What good references do you think might be useful to start with this? What software do I need for this? Are any of this reliable and free?

 

[meopemuk]

I have been working on experimental heterogeneous catalysis for a couple of years, specifically with metal oxides. I would like to learn computational modeling on heterogeneous catalysis (e.g hydrogen abstraction on acetone). I actually have no background on computaional calculations. What good references do you think might be useful to start with this? What software do I need for this? Are any of this reliable and free?

Hi jacone,

Welcome to the Forum! I used to work in this field 10 years ago. Some examples of my works should be close to your interests

M.A. Johnson, E.V. Stefanovich, and T.N. Truong, An ab initio study on the oxidative coupling of methane over a lithium-doped MgO catalyst: Surface defects and mechanism. Journal of Physical Chemistry 101 (1997), 3196-3201.

E.V. Stefanovich and T.N. Truong, Ab initio study of water adsorption on TiO2(110): Molecular adsorption versus dissociative chemisorption, Chemical Physics Letters 299 (1998), 623-629.

At that time GAUSSIAN was the most powerful package (I believe it still is). Unfortunately, it is not free, but as I recall it was sold at a big discount to universities. The great advantage of GAUSSIAN was that it was distributed with the source code, so you could implement your own models and go beyond the capabilities of the standard package. I am afraid that standard features may not be sufficient for such complex problems as reactions on oxide surfaces.

Eugene.

 

[jacone]

Eugene,

Thanks a lot for your help. I will look at these publications. However, I am still in need for a reference for beginners. Explanation of theories, applicability and limitations on assumptions, etc. I have no formal bachground on compiutational calculations so I'll be doing this by myself. Any websites or books that are useful?

 

[meopemuk]

I'll be doing this by myself.

That's tough. I believe you already have a solid knowledge of quantum mechanics. If yes, then you can proceed to more specialized books in quantum chemistry and solid state physics. There should be plenty of these books in any university library. I don't think there exists a texbook on quantum chemical modeling of reactions on oxide surfaces. This field is too young, and you would need to work your way through few journal articles. There are only few groups in the world, which are doing these calculations at a serious level (i.e., that can be compared with experiment). There is a huge number of people who are doing quantum chemistry of isolated molecules, or perfect crystals, or their surfaces, or defects in the bulk. To solve your kinds of problems you should be proficient in all these fields. Good luck.

Eugene.

 

[deepak1032]

I am presently trying to understand about a molecular modelling software wherein ti would be possible to theoretically study and analyze adsorption of mono and/or diatomic molecules onto a crystal or substrate.
Though Gaussian is quite good software for molecular modelling, I am unable to understand how to use it for adsorption study or Can it be used for such studies?
An understanding or any reference showing the above possibility shall be of great help.

 

[meopemuk]

Welcome to PF!

Yes, you can use Gaussian for modeling adsorption. Although you would need to employ a few non-trivial tricks. Check the two references that I've mentioned earlier (and other references cited there). All calculations there have been done with Gaussian and cluster model. The utility of this model strongly depends on the chemical bonding nature of the substrate. In my experience, the accuracy decreases and the required effort increases in the following order: ionic -> ionic-covalent -> covalent -> metallic.

 

[deepak1032]

Thank you for the prompt reply, which will definitely help me greatly. I really appreciate the details provided. Is it possible for you to provide the URL wherein I will be able to find the reference cited in the reply?
Thanks once again for the help.

With great thanks and warm regards
Deepak Thakur

 

[alxm]

I am presently trying to understand about a molecular modelling software wherein ti would be possible to theoretically study and analyze adsorption of mono and/or diatomic molecules onto a crystal or substrate.
Though Gaussian is quite good software for molecular modelling, I am unable to understand how to use it for adsorption study or Can it be used for such studies?

It's done all the time. However, you need knowledge of the methods used before you can get to work. QC methods are still pretty far from a 'black box' where you can stick in your model and get a value.

For instance, when you're talking about adsorption, I'm assuming there's London/vdW forces involved? If that's the case, you should know that LSDA-based density functionals (in other words; most DFT methods in use) won't usually accurately reproduce that, because it involves long-range correlation effects. OTOH there are specific DFT methods developed for that stuff that do work, but they have other drawbacks that need to be known about.

It's not that hard to do, but you do need to know what you're doing.

 

[meopemuk]

Thank you for the prompt reply, which will definitely help me greatly. I really appreciate the details provided. Is it possible for you to provide the URL wherein I will be able to find the reference cited in the reply?
Thanks once again for the help.

With great thanks and warm regards
Deepak Thakur

You can find PDF files of these articles by going to http://truong.hec.utah.edu/ and clicking on the "Publications" tab.

 

[deepak1032]

Dear Alxm/Prof. Eugene, Thank you for the specific information provided therein. It will definitely give me a direction for screening much study material that I already have. However, is it possible for you to kindly provide me with some material (may be research papers, thesis, books etc) that may be of help for the cited purpose therein?

Further, as you correctly analyzed, I'm about to work with London/Vander Waal's forces and thereby hunting in dark to find a way out. The only way I found is to just position (say H2 molecule to be adsorbed) near the other molecule (or substrate) and find out the changes. This is what I understand (really poor!) at present. That's why I badly need help.
Thanks once again.

Rgds
deepak thakur

 

[deepak1032]

Dear Prof Eugene,

Thank you for the link for the articles. That's a great help for a beginner

 

[alxm]

Further, as you correctly analyzed, I'm about to work with London/Vander Waal's forces and thereby hunting in dark to find a way out. The only way I found is to just position (say H2 molecule to be adsorbed) near the other molecule (or substrate) and find out the changes.

Right, well that's pretty much the way you'd do it. Create your model structure, position the adsorbed molecule in some decent guess of its position, do a geometry optimization to find the lowest energy conformation. You'd also want to try out a number of different guesses corresponding to the different binding modes; (e.g. side-on, end-on, etc) even if they're higher in energy you should often be able to find the local energetic minima for the various binding modes.

And so you have your difference in energy between the bound and unbound structure, \Delta E. But note that what you're calculating is the difference in electronic energy. To get a value comparable to experiment, you need to take into account the difference in zero-point vibrational energy, which involves doing a frequency calculation (= second-order derivatives).

So the practice of doing the actual calculation isn't too difficult. But since you've got vdW forces involved, finding a good DFT method isn't that easy; developing functionals that better reproduce dispersion forces is a pretty active research field. At the moment http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JCPSA6000116000022009620000001&idtype=cvips&gifs=yes" which is quite good for reaction energies, is rather infamous for not reproducing vdW interactions at all. (the molecules just don't bind)

The http://prola.aps.org/abstract/PRB/v45/i23/p13244_1" for adsorption studies. I don't really know how well they stack up against each other though, there's probably a benchmark study or two out there.

Those are functionals you'll probably be able to find in major existing packages (certainly Gaussian03, Jaguar 7). Since it's an active area of research there's a lot of http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=PRLTAO000096000014146107000001&idtype=cvips&gifs=yes" out there, but often implemented in less mature programs. (Taking a second here to acknowledge the work of Bengt Lundqvist who's spent as much time as almost anyone developing functionals that get vdW surface adsorption right)

So my best suggestion is to look around using your favorite tool (ISI Knowledge, Google Scholar, etc) and see what benchmark studies you can find on situations/compounds that most resemble what you want to study, and pick your method.

An important point as far as scientific rigor goes, BTW, is that once you do pick a method - stick with it. Don't fall for the temptation to pick whatever method happens to give the best results for in each particular case, or your results will be useless.

 

[deepak1032]

Thanks a lot Dear Alxm.
That will help me a lot.

 

[deepak1032]

Thanks a lot Dear Alxm.
That will help me a lot.

 

[deepak1032]

After a break I am back again. Can anyone help me with Gaussian software? I am frequently facing problem, particularly the link died problem, with server error #2070, even if I take a small molecule like formaldehyde and H2 system. Why does this error occurs? Can anyone help me with these error and other errors or tell me where can I find some details for the same. I am using Gaussian 03W.

Any hint shall be of great help to me.

Rgds
Deepak

 

[GC-Chem]

I am using gaussian calculation (DFT) for IR and Raman spectra of some organic molecules. Is there any possiblity to use gaussian to creat aborbsion and flurosence spectra?
what should to calculate HOMO and LUMO using DFT?

 

[honey43]

Hi Pals
I have a question in gaussian calculation(DFT). I want to know if it is possible to calculate polarization or not?

 

[alxm]

Hi Pals
I have a question in gaussian calculation(DFT). I want to know if it is possible to calculate polarization or not?

Welcome to PF, honey43.

Assuming of course that we're talking electrical here, do you mean polarization or polarizability?

If you really mean the former, the multipole moments are right there in every output file, just after the Mulliken populations. (which is why I'm not sure why you'd need to ask)
If you mean the latter, then it's the 'polar' keyword, look it up in the manual for the details.

 

[Rajini]

After a break I am back again. Can anyone help me with Gaussian software? I am frequently facing problem, particularly the link died problem, with server error #2070, even if I take a small molecule like formaldehyde and H2 system. Why does this error occurs? Can anyone help me with these error and other errors or tell me where can I find some details for the same. I am using Gaussian 03W.

Any hint shall be of great help to me.

Rgds
Deepak

could you please send the input file..do you just optimize the structure or just finding frequencies? Complete details can be seen in gaussian.org.

 

[Kukhun]

After a break I am back again. Can anyone help me with Gaussian software? I am frequently facing problem, particularly the link died problem, with server error #2070, even if I take a small molecule like formaldehyde and H2 system. Why does this error occurs? Can anyone help me with these error and other errors or tell me where can I find some details for the same. I am using Gaussian 03W.

Any hint shall be of great help to me.

Rgds
Deepak

Hi Deepak,

I am facing the exact same problem error # 2070, did you get to know, why this error is occurring all the time?

Thanks,
Kukhn

 

[rajavel]

Is it possible to do calculations for crystal field (for Solid state) with Gaussian software ?