BSSE problem with Iodine (problem with ECP/$freeze?)

[Tiuri]

Hi everyone,

I'm calculating RI-MP2 energies for a certain adduct system.

This system exists of two fragments and the second fragment can contain Cl, Br or I.
I am calculating MP2 energies using the

$ricc2
geoopt model=mp2

setup and using default values for $freeze, $cbas as chosen by default by define.

I did the normal MP2 singlepoint without problems with

dscf > outputfile1
ricc2 > outputfile2

and then the BSSE correction in a separate folder (but with the same input files) by

jobbsse -level cc2 -ri

Everything runs fine for Cl- and Br-containing systems, and I get a BSSE correction of about 2.5 kJ/mol (basis set is def2-TZVPP).

For iodine, I get strange results.


My RI-MP2 (no bsse correction) energy is

MP2 energy = -776.363....

but in bsse_out I get

CP-corrected total energy  -248.054...


Obviously this makes no sense. I have also noted that the "CP-Correction to total energy" is 528.30... hartree, so by far too large.


I'm using TM6.0 and standard settings for the basis sets (so my def2-TZVPP for Iodine has an ECP).

My $freeze reads "implicit core= 15  virt = 0".

I am guessing that the $freeze settings are wrong, so could you tell me what a recommendable setting would be, using ECP?
Or is there a general problem with jobbsse and ECPs?

Thanks a lot in advance!

[Arnim]

Hi,

in such a case I would use 'jobbsse -level cc2 -setup' first, and then go in the subdirectories and check/modify the $freeze and $atoms data groups.
If you use a def2 basis set, it would make sense to freeze 4 shells per iodine, because 14 shells are already in the ECP. Then you have a kr core frozen.
For the ghost calculations, only shells from atoms with a charge should be frozen.

Best,
Arnim