Try to optimize again using the final output structure.

SCF or geometry convergence errors?

There is a SCF convergence error. I used in input file the low spin state (multiplicity = 2).

%nproc=16

%mem=14GB

%chk=cobalt.chk

# B3LYP/GEN FOPT=(Maxcycles=1024) SCF=(conver=7,maxcycles=2048)

11-12-2012

+2 2

C 8.84300000 3.31400000 0.93700000

H 9.01500000 2.40600000 0.81800000

C 9.36500000 4.19600000 0.04300000

H 9.88600000 3.89200000 -0.66600000

C 9.11600000 5.52100000 0.19500000

H 9.45900000 6.13600000 -0.41300000

C 8.36100000 5.94400000 1.24900000

H 8.18900000 6.84900000 1.37300000

C 7.84800000 4.99500000 2.14000000

C 7.04500000 5.35800000 3.32700000

C 6.87500000 6.65600000 3.75300000

H 7.23200000 7.36200000 3.26500000

C 6.17500000 6.89500000 4.90500000

H 6.06200000 7.76800000 5.20800000

C 5.64800000 5.85700000 5.60100000

H 5.17200000 6.00100000 6.38700000

C 5.83500000 4.60000000 5.11700000

H 5.47100000 3.88900000 5.59200000

C 4.63100000 3.49200000 1.43400000

H 4.49800000 4.13500000 2.09300000

C 3.93200000 3.59400000 0.26900000

H 3.34400000 4.30000000 0.13400000

C 4.10500000 2.64400000 -0.69600000

H 3.62800000 2.69100000 -1.49300000

C 4.97600000 1.63400000 -0.48200000

H 5.10900000 0.98500000 -1.13400000

C 5.66800000 1.57100000 0.71000000

C 6.63300000 0.51200000 1.02300000

C 6.80600000 -0.63600000 0.25900000

H 6.29100000 -0.75600000 -0.50800000

C 7.70700000 -1.57500000 0.60700000

H 7.82400000 -2.34000000 0.09200000

C 8.43800000 -1.37800000 1.73200000

H 9.06700000 -2.00800000 1.99800000

C 8.24000000 -0.23700000 2.47700000

H 8.74700000 -0.11400000 3.24600000

C 9.61700000 2.50100000 4.70500000

H 9.98800000 2.81600000 3.91100000

C 10.41900000 2.43100000 5.81100000

H 11.31400000 2.68000000 5.77100000

C 8.54900000 1.60700000 7.00500000

H 8.16200000 1.30600000 7.79500000

C 7.80300000 1.67800000 5.84600000

C 6.38600000 1.28100000 5.76400000

C 5.67800000 0.76700000 6.85500000

H 6.08800000 0.64500000 7.68000000

C 4.34100000 0.44800000 6.66300000

H 3.84200000 0.10600000 7.36900000

C 3.76300000 0.62600000 5.46100000

H 2.87000000 0.40400000 5.32800000

C 4.50600000 1.13500000 4.45000000

H 4.10300000 1.26800000 3.62200000

C 9.86100000 1.98300000 6.98100000

H 10.37400000 1.93800000 7.75600000

Co 6.94700000 2.45000000 3.18700000

N 8.09700000 3.68600000 1.97300000

N 6.51600000 4.33100000 3.98900000

N 5.49800000 2.50900000 1.66700000

N 7.35100000 0.70000000 2.13500000

N 5.79300000 1.45600000 4.59500000

N 8.33700000 2.14300000 4.70000000

C N H 0

6-31G

****

Co 0

LANL2DZ

****

--Link1--

%chk=cobalt.chk

%mem=14GB

%nproc=16

# B3LYP/GEN FOPT scf=(restart) Guess=read Geom=Checkpoint

11-12-2012-scf-restart

2 2

If you start directly from a Xray structure, the CH bond lengths are typically to short (as is also the case in your structure, just checked). Adjust them by hand to something reasonable (1.1 A). That usually resolves the SCF problem at the start of the optimization.

Dear Zendehdel,

As mentioned, before any new calculation, confirm the spin multiplicity and the ground state of your system (Co(II) triplet or higher ), remember that DFT method do better in the ground state. After this, you can try to test other basis sets and ECP, we used this procedure to find adequate basis sets and ECP for organotellurium compounds.

VEGA-TEIJIDO MA; ZUKERMAN-SCHPECTOR, J.; VENTURA, O.N.; CAMILLO, R. L.; CARACELLI, I. ; GUADAGNIN, R.C.; BRAGA,

A.L.; SILVEIRA, C.C.

Dichloro(cyclohexilidene-1-methylene)(phenyl)Te(IV).Looking for the theoretical treatment.. Zeitschrift für Kristallographie, v.: 219, p.: 652- 658, 2004

ISSN: 00442968 ; DOI: 10.1524/zkri.219.10.652.50816

http://www.oldenbourg-wissenschaftsverlag.de/olb/de/1.c.335324.de

Best regards

As Mauricio Angel Vega Teijido said, you have mistake in input line. If you specify the gen keyword, you need to use pseudo=read or just specify the genecp keyword. I will not give further comment of this problem. I'll give information about how to improve the scf convergence.

First, i want to ask you why do you choice multiplicity of 2 for cobalt (+2 cation). It is generally reported that Co2+ has multiplicity of 4 (high spin state is generally stable) Also, in this case, there is no convergence failure during the calculation. (probably)

However, ff there is proper reason, you need to consider the (1) lower convergence value or (2) fermi option or (3) qc option.

If you use the lower convergence value, you need to reoptimize the structure with default value.

If you use the fermi option, you need to do single point calculation without fermi option . This is because fermi option lead to change the orbital energy level. (cannot change the energy)

Hope this help

Sorry doublet or higher

If you need scf=qc you either try to converge to an unlikely spin state, or you have serious problem with the initial geometry, or your method is inappropriate.

Fixing CH bonds should always be the first step (besides selecting the correct spin state).

Wavefunctions for molecules containing transition metals are notoriously hard to converge. The best advice is to check the Gaussian manual and use some of the built-in convergence options. Turn the DIIS option on or off (I forget which is default). Use a higher convergence threshold for a few cycles then lower it later. Try a single-point HF calculation and if that converges use GUESS=READ for the DFT geometry optimization. Use the clean option on your CIF geometry just to be sure there are no odd close contacts or something. And,as was already mentioned, take the unconverged wavefunction and run it again with that as the initial guess (GUESS=READ). This may take some trial and error and patience! Oh, and check the literature - this seems like something someone would have done before.

Which is the error in the output file?

Thanks all. I'm checking your comments in Gaussian.

Jesus Baldenebro-López,

Annihilation of the first spin contaminant:

S**2 before annihilation 0.8315, after 0.7533

Convergence failure -- run terminated.

Raghavender Medishetty

To specify the ECP in input file, you can just follow this content

This is example using genecp keyword.

%chk=title

%mem=XXXMW

%nproc=X

#p B3LYP/genECP opt scf gfinput gfprint

Title

charge multiplicity

coordinate

specify the atom, which is not treated by ECP

specify the basis set

****

specify the symbol of metal atom 0

ECP potential (for example LANL2DZ, SDD etc..)

****

blank line

specify the simbol of metal atom 0

ECP potential (for example LANL2DZ, SDD etc..)

blank line

If you want to insert full information of ECP, you need to used gen and pseudo=read

This is example

%chk=title

%mem=XXXMW

%nproc=X

#p B3LYP/gen pseudo=read opt scf gfinput gfprint

Title

charge multiplicity

coordinate

specify the atom, which is not treated by ECP

specify the basis set to use atom

****

full information of ECP (you can find this information on web site https://bse.pnl.gov/bse/portal)

Best regards

Dear Mahmoud Zendehdel:

If you are spin contamination; then, the calculation method is not adequate. Can you show the input file?

Dear Jesus,

I have shared the input file in this question page (you can find it in previous comments).

You should increase the number of cycles for convergence.

You should increase the no of cycles for convergence and also check the for spin contamination

Dear Mahmoud Zendehdel:

What is the content of the .gjf or .com file for Gaussian?. I can not find it in the previous comments.

I put it as a text type. You can copy and paste it in a text editor program and change the file type to gjf.

You don't need to use a different set of basis functions on Cobalt. 6-31G includes basis functions for Cobalt. If you are running the jobs on your own computer, though, I can see why you would want to use pseudopotentials...

Also, you should definitely increase the maximum number of convergence cycles. You can do this with SCF=(maxcycles=n). 512 is a good number to try, but you may need even more sometimes.

In some cases, you may even need to increase the convergence criteria. You can do this with SCF=(conver=N), where N is the inverse power of ten.

One more thing! If you have a powerful enough computer, you can also use SCF=QC when you are having convergence issues. It takes longer, but it almost always works.

Even me I am facing the same problem I cannot optimized huge structure like binuclear and polynuclear complexes.So can anyone tell me the exact method for running metal complexes in guassian especially with M06.??? I also want to propose my own structure through Z-matrix that is by writing all the bond length, bond angle and dihedral angle manually or by just manipulate the values in the notepad/wordpad?? So anyone can help me solving this issue. ??   

Can You See This Video

https://www.youtube.com/watch?v=mi2DERG_3Eo

If you start directly from a Xray structure, the CH bond lengths are typically to short (as is also the case in your structure, just checked). Adjust them by hand to something reasonable (1.1 A). That usually resolves the SCF problem at the start of the optimization.

For molecules or clusters containing cobalt element, it's really difficult to get convergence, no matter of the functionals or basis sets being used. I tried various options and parameters for cobalt clusters, It turns out that SCF=XQC/QC mostly works.

H,

Try with B3LYP with 6-31G(d) on C, N, O, and H, and TZV on Co (with GEN) or M06/def2-SVP or B3LYP/def2-TZVP on all atoms (pls. see Inorg. Chem., 2018, 57 (21), pp 13397–13408). Also check, spin-state for your Cobalt(II)tris(2,2'-bipyridine) and fix the multiplicity.

I have a Fe complex of 85 atoms. Its been two weeks the convergence is not done yet. Please suggest the options which i can use for my calculations in Gaussian. I am using this recently.

# uwb97xd/6-31g(d) geom=connectivity scf=maxcycle=1024

Thanks & Regards