CCL: COSMO vs COSMO-RS

 Sent to CCL by: Andreas Klamt [klamt=cosmologic.de]
 Dear Sina,
there are several potential reasons for the large difference. On the one hand you used qute different basis sets for the COSMO calculations at eps=78 and the eps=infinity calculations. Admittedly I do not expect that this changes the results too much.
I am not completely sure how you did the COSMO-RS reaction calculation. Did you use the thermodynamic cycle inserting the gasphase reaction energy from the b3-lyp and a def2-QZVP calculation? If so, then the latter result, i.e. 
 GReaction-COSMO-RS-Water -  GReaction-gas= -2.05 Kcal/mol
should be nothing else than the dG_solvation difference of educts and products on BP-TZVP-COSMO-RS level, since GReaction-gas cancels out. 
 Can you just check and confirm that?
Please note that bare COSMO (unlike other implicit solvation models) is not parameterized for giving optimal dG-solvation values. It just trets the electrostatic component, and not the dispersion, and other non-electrosttic contributions. Hence you cannot expect good quantitative results for dG_hydrtion from using bare COSMO. COSMO-RS is my recommended method to take all these things into account.
Finally it should be noted that the recommended procedure for dG_reaction calculation requires to include the vibr., rot. and transl. free energy contributions in the gasphase, which will be very important for a reaction in which the number of particles changes from 1 to 4. These can be calculated for exmple with the freeH tool in TURBOMOLE. But this contribution would appear in the gasphase as well as in the liquid phase and thus should not influence your differential results.
In summary, I do not find your result too surprising. The more relible answer should be the COSMO-RS answer, but please note that your calculations are still suboptimal. 
 Best regards
 Andreas
 Am 31.01.2013 00:13, schrieb Sina Behtash behtash!=!email.sc.edu:
 
 Sent to CCL by: "Sina  Behtash" [behtash ~ email.sc.edu]
 I am using both COSMO (implemented in Turbomole 6.0) and COSMOtherm/COSMO-RS to
 calculate the solvent effects on the free energy of reaction for the following
 reaction:
 Glucose  -->  HMF + 3 H2O
 First, I optimized all the reactants and products in the gas phase using b3-lyp
 and a def2-QZVP basis set. Next, assuming that the geometry does not change
 significantly in the presence of water, I used the optimized structure to run a
 COSMO single-point energy calculation with the same basis set and dielectric
 constant of water at 298 K which is 78.39. According to the COSMO calculation,
 GReaction-COSMO -  GReaction-gas= -15.2 Kcal/mol
 Next, in order to generate relevant input for COSMO-RS calculations, I ran
 another COSMO single-point energy calculation using the BP functional and TZVP
 basis set. Then I used COSMOtherm/Reactions and the BP-TZVP data base to run
 COSMO-RS calculations for my reaction at T=298K.
 The result shows that,
 GReaction-COSMO-RS-Water -  GReaction-gas= -2.05 Kcal/mol
 I was expecting that for water, the COSMO and COSMO-RS calculations should give
 almost the same result. However, in this case the difference in the effect of
 water on the free energy of reaction is around 13 Kcal/mol.  Can anyone explain
 to me why the difference is so large? Im wondering if the effects of hydrogen
 bonding and electrostatic interactions can be this significant.
 Sina Behtash
 Ph.D. Student
 Department of Chemical Engineering
 University of South Carolina
 Phone: (803)777-0611
 Email: behtash-$-email.sc.edu>
 
 --
 Prof. Dr. Andreas Klamt
 CEO / Geschäftsführer
 COSMOlogic GmbH & Co. KG
 Imbacher Weg 46
 D-51379 Leverkusen, Germany
 phone  	+49-2171-731681
 fax    	+49-2171-731689
 e-mail 	klamt^^^cosmologic.de
 web    	www.cosmologic.de
 [University address:      Inst. of Physical and
 Theoretical Chemistry, University of Regensburg]
 HRA 20653 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt
 Komplementaer: COSMOlogic Verwaltungs GmbH
 HRB 49501 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt