What could be the possible basis set that can be used for a metal base macrocyclic complex in gaussian09?

I am using transition metals such as $\ce{Ni}$, $\ce{Co}$, etc. Also, I have tried using DFT and HF theory for optimization but I am not getting appropriate results.

I am basically looking for binding of metal with the ligand in these coordinate complexes on basis of which i can use them for a particular application. The basis set that I am using right now for DFT is Opt=(Cartesian,maxcycles=500) b3lyp/gen nosymm pseudo=read SCF=(conver=5,vshift=400,maxcycles=1000)

Edit: Earlier I was using a specific basis set for transition metals and now I am using lanl2 as a single basis set for the DFT calculation. I am also using hf cal in which the sto3g basis set is used but iam getting two consistent errors in optimization which are convergence failure and number of step exceed error processed link 9999.

  • $\begingroup$ Could you also indicate what you want out of the calculations? For example, if you care about excited states, you may choose a different basis set than if you don't. $\endgroup$
    – Ben Norris
    Dec 25 '12 at 2:02
  • 1
    $\begingroup$ I approved an anonymous edit which I assume is you from another computer, however the line at the beginning of your file is not actually the basis set but rather a collection of directives that tells the calculation how to run. The basis set is a set of mathematical functions that are used to describe the AOs of each atom. In your case, the gen keyword tells G09 to read a basis definition included at the end of the file, as Deathbreath has mentioned. If the definition is missing your calculation will fail. $\endgroup$ Dec 25 '12 at 13:07
  • $\begingroup$ Additionally, Pseudo=read also means that the calculation will expect a pseudopotential definition. The other stuff in the directive line consists of geometry optimisation instructions (Opt=), a functional (B3LYP), manually disabled symmetry (nosymm) and SCF convergence information (SCF=). $\endgroup$ Dec 25 '12 at 13:09
  • $\begingroup$ replace scf=... with scf=xqc, if that fails try scf=qc and report back $\endgroup$ Dec 31 '12 at 23:52
  • $\begingroup$ It might be helpful if you would post your current input file. There are many places where things could go wrong, and all we can do is conjecture what the solution should be. $\endgroup$
    – Eric Brown
    Jun 26 '13 at 18:12

In order to check on the availability of basis sets you may want to look at the Basis Set Exchange. After determining which basis set suits your problem (this will likely involve some literature search, Nickel, DFT, basis set, in Web of Science should do the trick), you can enter basis sets into Gaussian using the Gen keyword. BSE will export any selected basis sets in Gaussian format, so that shouldn't be a problem.


You didn't say anything about size of the system. Also what computer hardware is available for your computations. For TM-complexes you need to get better basis than lanl2 and definitely you need much better basis for your ligands. sto3g is a joke. Start with 6-31G(D,P) and LANL2DZ. For good data go for 6-311+G(d) for non-metal and LANL2TZ+f for Ni and CO. Another possibility is def2-svp (it has all 4th raw metals in as well as non-metals).

General remark. I hope that you have pre-optimized your ligand before you started with the TM-complex.

  • 1
    $\begingroup$ Small remark: you would better start from 6-31+G(d,p) rather then 6-31G(d,p). See, for instance, dx.doi.org/10.1021/jp021590l. I quote from there: "Considering the performance vs cost tradeoff, we recommend that diffuse functions always be included on nonhydrogenic atoms in DFT and hybrid DFT calculations unless the sole goal of the calculation is absolute bond energies or absolute ionization potentials." $\endgroup$
    – Wildcat
    Sep 20 '14 at 11:52

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.