CCL:G: ccsd (t): Excessive mixing of frozen core and valence orbitals

[Mariusz Radon <mariusz.radon===gmail.com>]

 Sent to CCL by: Mariusz Radon [mariusz.radon]|[gmail.com]
 Dear Valentin:
 > From my experience, strong core-valence mixing may be, indeed, observed for
 second-row transition metals (here you have Zr). Recently, I have observed a
 strong mixing between 4d (valence) and 4p (outer-core) orbitals of Mo in CASSCF
 calculations (see: PCCP 2015, 17, 14890-14902; DOI: 10.1039/c4cp04863f). This
 may be related to your problem of excessive core-valence mixing, preventing
 Gaussian from starting CCSD.
 The occurrence of strong mixing between core and valence orbitals suggests that
 you have to correlate also the outer-core electrons (for Zr: 4p and 4s probably
 too) by modifying the default frozen core settings in CCSD(T); see
 “Frozen Core Options” in the manual. Including the problematic
 outer core orbitals in one group with the valence orbitals should mitigate your
 problem. Moreover, correlating the outer core orbitals of TMs is generally
 recommended for obtaining accurate results.
 If you decide to correlate also the 4p and 4s electrons of Zr, I would recommend
 to use a  basis set designed for correlating these outer-core electrons. Your
 current basis set (def2-TZVP) is not suitable for doing that; you should try
 something from the ANO or cc-pwCVnZ family. Remember that for second-row TMs
 relativistic effects become quite important. Hence, for Zr atom, I would
 recommend to use either the Douglas-Kroll-recontracted basis (cc-pwCVTZ-DK) or
 the one with relativistic pseudopotential (cc-pwCVTZ-PP).
 Best wishes,
 Mariusz
 --
 Mariusz Radon, Ph.D.
 Faculty of Chemistry
 Jagiellonian University, Krakow
 ul. Ingardena 3, 30-060 Krakow, Poland
 E-mail: mradon++chemia.uj.edu.pl
 or mariusz.radon++uj.edu.pl
 http://www.chemia.uj.edu.pl/~mradon
 > On 07 Nov 2016, at 21:36, Valentin Vassilev-Galindo
 vvassilevgalindo,,gmail.com <owner-chemistry++ccl.net> wrote:
 >
 >
 > Sent to CCL by: "Valentin  Vassilev-Galindo"
 [vvassilevgalindo(a)gmail.com]
 > I did a coupled cluster calculation and I found the error shown below.
 > Could you give me some suggestion? Thank you!
 >
 > Input file:
 >
 > % MEM=28GB
 > # ccsd(t) Def2TZVP SCF=(MAXCYCLES=3000)
 >
 > ### COMMENT ###
 >
 > 0           1
 > F     2.026723000     -1.795391000     -0.062792000
 > Zr    1.082243000     -0.123016000     -0.206225000
 > F     1.282418000      1.817002000      0.513824000
 > C    -0.701823000     -0.128018000      0.765581000
 > Al   -0.596452000      1.853945000      1.068611000
 > Al   -1.552563000      1.331878000     -1.320150000
 > Al   -1.383337000     -1.123291000     -1.197280000
 > Al   -1.764653000     -1.639899000      1.417760000
 >
 >
 > Last lines of output file:
 >
 > Defaulting to unpruned grid for atomic number  40.
 > Defaulting to unpruned grid for atomic number  40.
 > Defaulting to unpruned grid for atomic number  40.
 > Defaulting to unpruned grid for atomic number  40.
 > Largest valence mixing into a core orbital is  4.51D-01
 > Largest core mixing into a valence orbital is  4.84D-01
 > Excessive mixing of frozen core and valence orbitals.
 > Error termination via Lnk1e in /storage/software/intel/g09/l801.exe at Mon
 Nov  7 11:44:53 2016.
 > Job cpu time:       0 days  0 hours  2 minutes 27.9 seconds.
 > File lengths (MBytes):  RWF=     31 Int=      0 D2E=      0 Chk=      2
 Scr=      1
 >
 >
 > With best regards,
 >
 > Valentin Vassilev-Galindo>
 >