A similar study was conducted in: "Study of gas-phase O–H bond dissociation enthalpies and ionization potentials of substituted phenols – Applicability of ab initio and DFT/B3LYP methods" by Erik Klein and Vladimır Lukes. I hope it would be a useful guide to you.

Helloo Babatunde

Thank you

It depends on size of your molecules and chemical composition. If molecule contains hydrogen, nitrogen, carbon, oxygen only (without transition metals e.g.) the correlation-consistent basis sets will be optimal, I suppose (For instanse cc-pVTZ basis set).

Dear Fortunatus Jacob,

Usually, we use DFT/B3LYP 6-31G.

If you want some further information about basis set, there is a description for basis set on Gaussian official website: http://gaussian.com/basissets/

And you can download some slides from other universities, for example, University of Southampton: http://www.southampton.ac.uk/assets/centresresearch/documents/compchem/DFT_L8.pdf

Dear Lei

Thank you for your good inputs, I will follow up.

Dear Oleg

Thank you for your answer, I have actually used Def2-TZVP for barium and aug-cc-pvtz/SDB-aug-cc-pvtz for the halogens.

Dear Fortunatus Jacob,

First and foremost " Inorganic compounds" is a vague description. Second, if the similar systems have not been explored computationally then benchmarking of the basis set is the must. Moreover, one cannot suggest you a basis set/functional until he/she is aware of the system.

Dear Fortunatus,

The basis set should be selected depending on the information and resource you have. In this situation, you can find two different approximations:

-Optimization with CCSD(T): This procedures are not common due to the extremely expensive calculation. Usually, the optimization is done with other methods, such MP2 or DFT, and then, only the energy is calculated with CCSD(T) to correct it, but, vibrations will come from the opt method only.

-Energy calculation: You can use both, Pople or Dunning basis sets (including ECP). Here, you can choose three different situations. 

  1) Use different basis set for each atom in order to reduce the calculation time.

  2) Use the same basis set for all atom

  3) Use Complete basis set extrapolation (CBS). In my opinion this is one of the best option for high accuracy calculations. Even so, you need to perform different CCSD(T) calculation.

In my opinion, a good option is to use Dunning basis set, taking into account that the algorithm for this technique with Dunning's basis set are improved in comparison with Pople ones. Then, depending on the type of calculation, you should select the basis set. For example:

If you use opt:  CCSD(T)/cc-pvDZ or CCSD(T)/aug-cc-pvDZ

If it is only energy and/or CBS calculation: CCSD(T)/aug-cc-pVnZ    n=2, T and Q which could be improved by quintuple zeta. etc.

[Attached] I provide you two references in CBS if you are interested in it and, in addition, a link to a web with all basis set for the atoms (including ECP) to be implemented manually.

If you can provide some extra information, maybe I could suggest some specific answer, but for the moment, I can only give you a brief and unspecific answer.

I hope it helps you,

Joaquim

PD: If you use other methods for opt and CCSD(T) for energy, take into account that the method you select will give limitation to your results. For example, if you want to find a enthalpy results from CCSD(T)/cbs//B3LYP/aug-cc-pvDz calculation, you must take into account that the error of your DFT can be limitative of your accuracy. For example, if the functional does not give you good vibrations, their contribution will not be corrected. The same with geometries, etc. So, I think that with this way, the first step is to validate the method correctly.

https://bse.pnl.gov/bse/portal

Article On the effectiveness of CCSD(T) complete basis set extrapola...

Article An efficient extrapolation to the (T)/CBS limit

Dear Joaquim,

Thank you for the good inputs, I real appreciate. Example of the species I am computing is Ba2Br3+. I have been employing DFT, MP2 and MP4 methods whereby for Barium atom I use Def2-TZVP basis set and for Bromine I use SDB-aug-cc-pvtz or aug-cc-pvtz, and I got good results. I now want to switch to other molecules and ions but I am not good and well knowledgeable on the selection of appropriate basis set. Thanks for your input, you have opened up my eyes more on this concern.

Dear Mohsin Yousuf Lone ,

It is true I used a general term 'inorganic compounds'. To be specific, I am working on Alkaline earth halides (monohalides, dihalides, dimers and ionic species). For example for ionic species I have computed the properties of Ba5Br9+.

For Barium atom I have been using Def2-TZVP basis set while for Bromine I have been using SDB-aug-cc-pvtz or aug-cc-pvtz. I have obtained good results so far as you can see from my published paper on barium dibromide.

Now I want to switch to other molecules but I am not well knowledgeable on the selection of basis sets that is why I asked that question so as to get some inputs that will help me in my upcoming works.

I really thank you for your inputs.

Dear Fortunatus--

One thing to consider is that, for best comparison of properties and especially energies (thermochemistry), one needs to be consistent as to the theory level you use, including the basis set.  

In other words, if you have thoughts of comparing/contrasting the Ba(x)Br(y)+ systems to systems using other alkaline earth metals or halogens (Let's say Ca(x)Cl(y)+) then you should if possible use the same basis set for Ca as you did for Ba, and the same basis set for Cl as you did for Br.  If you can't use the same basis set, use a basis set that is as similar as possible.  If you have good results for Ba(x)Br(y)+ with aug-cc-pvtz for bromine, you wouldn't want to go to a completely different basis set for chlorine.

I will also echo those other respondents who mentioned that doing a CCSD frequency job will take a lot of computational resources and that you'll likely end up less-accurate but faster procedures.  (Also, check whether your computational chemistry software has *analytic* frequencies for the level you want to use.  Being forced to use numeric frequencies is another way to spend a lot more in compuational resources.)

Dear David Shobe ,

Thank you for your advice, I take it. But from the output file of the program let say Firefly file, how do I know whether the frequencies are analytic or not? @David.

That information (which methods have analytical frequencies) should be available in the manual.

Can you suggest me about a suitable basis set for excited state energy calculation for IODINE molecule? I want to calculate UV-VIS aspectra of IODINE and I3- molecule computationally. Fortunatus R. Jacob David Shobe Joaquim Mª Rius Bartra

I like the aug-cc-pV_Z basis sets. You will want the set with the aug- prefix, since diffuse functions are necessary for excited states and for anions.

However, there are some basis sets that are optimized for different methods. Which methods are you planning to use? Not that I'm aware of any basis sets optimized for excited state calculations...

BTW, the EMSL basis set exchange has a new website, https://www.basissetexchange.org/