CCL:G: Calculation of Redox Potentials

 Sent to CCL by: Vincent Xianlong Wang [xloongw---yahoo.com]
 Hi Tobias,
 Is the BSSE (basis set superposition error) relevant to the big deviation you
 observed?  Have you done the correction to check if the results change?
 Vincent
 ----- Original Message ----
 > From: Tobias Kraemer tobiask * chem.gla.ac.uk
 <owner-chemistry##ccl.net>
 To: "Wang, Xianlong " <xloongw##yahoo.com>
 Sent: Friday, April 24, 2009 11:08:22 PM
 Subject: CCL:G: Calculation of Redox Potentials
 Sent to CCL by: Tobias Kraemer [tobiask|chem.gla.ac.uk]
 Hi....
 I'm trying to get experience in calculating some redox potentials of copper
 complexes to refer these data to experimental values.
 I have done a number of initial calculations on a test set of organometallic
 molecules, to get used to the method, using B3LYP/6-31G*
 and the default PCM model as implemented in Gaussian with methanol as solvent.
 As expected, deviations from experiment are big
 (in order of a few hundreds of millivolts). I also applied this protocol to the
 copper complexes I'm interested in and in a next step increased the
 basis set to TZVP for copper and ligating atoms, and SVP for remaining ligand
 atoms. I observed a big difference in the computed value for the redox
 potential, going from
 -0.12 (B3LYP/6-31G*) to +0.35 (B3LYP/TZVP). This is surely not so unexpected.
 There are significant changes in the solvation energy
 for  the reduced form of the complex, Cu(I), whereas the solvation energy of the
 oxidised form, Cu(II), is uneffected. Also, significant changes in the relative
 energies of the
 structures in gas phase contribute to the change in the reduction potential.
 After reading some papers about the topic, I feel that the use of Jaguar for
 calculations of this type is preferred.  This causes some difficulty in
 reproducing results, as I'm using
 Gaussian. So, in order to design some further test calculations, I would like to
 ask for some advise from people that have experience with this type of
 calculations using Gaussian and the
 solvent models therein. Maybe there is some further literature, which would help
 me as well, that report benchmark calculations using Gaussian.
 I would especially like to know, if it is advisable to do full optimisations in
 solvent, to improve the results, and which solvent models yield good results,
 although
 I'm aware, that this strongly depends on the system that is being investigated.
 Your comments are much appreciated.
 Thanks
 Tobi
 -- ____________________________________
 dipl.-chem. tobias kraemer
 westCHEM, department of chemistry
 university of glasgow
 joseph black building
 university avenue
 glasgow G12 8QQ
 scotland, uk
 |phone| +44 (0)141 330 8121
 |email| tobiask###chem.gla.ac.uk
 ____________________________________http://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp-:-//www.ccl.net/chemistry/sub_unsub.shtmlhttp-:-//www.ccl.net/spammers.txt