From jerry %-% at %-% dft.chem.cuhk.hk Wed May 1 21:50:15 1996 Received: from iris.chem.cuhk.hk for jerry {*at*} dft.chem.cuhk.hk by www.ccl.net (8.7.1/950822.1) id VAA04642; Wed, 1 May 1996 21:01:15 -0400 (EDT) Received: from dft.chem.cuhk.hk by iris.chem.cuhk.hk via SMTP (931110.SGI/940406.SGI.AUTO) for chemistry:~at~:www.ccl.net id AA18897; Thu, 2 May 96 09:00:33 +0800 Received: by dft (950511.SGI.8.6.12.PATCH526/930416.SGI.AUTO) id JAA09614; Thu, 2 May 1996 09:00:31 +0800 Date: Thu, 2 May 1996 09:00:30 +0800 (HKT) From: Jerry C C Chan To: chemistry-: at :-www.ccl.net Subject: Summary: NBO for transition metal Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear Netters, Some days ago I posted a message below: > I notice that NBO has already been applied to transition metal > containing systems. Could anybody give me some references which concern > the transition metal-ligand interaction using NBO analysis, in addition to > the work by Maseras and Morokuma CPL 1992. > I have tried applying NBO analysis to [Co(NH3)5Cl]2+ and I find > some difficulties in interpretating some of the results: > if LP ( 1) N 3 / BD*( 1)Co 1-Cl 2, "lone pair electrons of N (donor) / > antibonding orbital of Co-Cl (acceptor)", is interpreted as an > indication of the importance of the resonance N-Co <-> N(+)-Co Cl(-), how > should one interprete BD ( 1)Co 1-Cl 2 / RY*(16)Co 1 ? I would like to express my gratitude to all who response to my query. Special thanks are due to Profs. Frank Weinhold, Jay Badenhoop, Martin Kaupp and Eric Glendening. I wish my summary would contribute to the popularity of NBO analysis. ************* > 1. Is it justified if NBO scheme is used to analyze the SCF density > obtained by dft method? While not developed for DFT wavefunctions, NBO analyzes the 1st order density matrix, and therefore should be more compatible with density- based methods than methods which analyze the wavefunction to determine bond order. We have had much success (and yes, some surprising results which were found later to be consistent with experimental results) from NBO analysis of coordination complexes. > 2. Is there any physical meaning associated with the energy of the > natural bonding orbital? I don't understand the physical picture of > "Population inversion found on atom ..." although there is an > operational explanation in the NBO manual. The natural bond orbitals are calculated as the 1- and 2-center set of orthogonal orbitals which takes account of the greatest amount of the electron density in a maximum-occupancy sense. Each orbital has an energy or eigenvalue associated with it, just as the atomic orbitals and molecular orbitals do. However, sometimes one or more orbitals will be higher in occupancy, but also higher in energy than one or more other orbitals. Therefore the ordering from highest to lowest occupancy and the ordering from lowest to highest energy will not be the same. Often this occurs if you have a large number of orbitals which are fairly close in energy and population, which is common in these coordination complexes. This warning message about "population inversion" just alerts you to this different ordering of the orbitals, and does not indicate the orbitals are of inferior quality or are mislabeled in any way. > 3. How can the NBO result be considered as valid? NHO procedure > fail in CO2 (JACS, 1980) because of the miscount of the total number > of orbitals. However, if the total number of orbitals is counted > correctly but some of the BD and *BD occupancy are < 1.7 and > 0.4, > respectively, can we still accept the NBO result? First, the NBO procedure has been modified to handle a wider variety of cases than in 1980. I do not believe that the CO2 case fails with the latest version of NBO. ... However, you must not just consider the NBO's listed in the table. This is the "best" localized set of orbitals, but low-occupancy bonding orbitals (BD or LP) or high-occupancy anti-bonds (BD*) indicates a highly delocalized molecule. You must also consider the list of second order perturbative estimates of the donor-acceptor interactions. This list indicates the most important delocalizations from bonding to antibonding orbitals. You will find both Co --> ligand and backbonding ligand --> Co interactions. ... In addition, our group has developed a quantitative NBO-based resonance theory, which calculates a set of localized resonance structures and associated resonance weights which best describe the full density, then calculates not only a bond order but also partitions the bond into covalent and ionic contributions to this bond order based on the bond polarization coefficients. This bond order is more comparable to Bader's bond order than a measure of bond order only based on the "best" localized NBO structure (resonance structure of highest "weight") only. Unfortunately, the method has been developed for organic molecules, and cannot handle transition metal complexes without further development. Jay Badenhoop Department of Chemistry Southern Illinois University at Carbondale ******************* > In the Second Order Perturbative Analysis of Co(NH3)5H2O, I get > BD*( 1)Co 1- N 4 /299. BD*( 1)Co 1- N 5 2168.13 0.01 0.249 > I really do not understand the physical picture implied by these large > interactions. The BD*-BD* 2nd-order energies should be ignored, since the BD* is not even approximately "doubly occupied" and the PT picture is no longer valid. (NBO prints out these entries for convenience in analyzing excited states, which may have significant population in BD* orbitals.) Only the donor(LP,BD)-acceptor(BD*,RY*) entries have physical significance in this case. FW ******************* > I notice that NBO has already been applied to transition metal > containing systems. Could anybody give me some references which concern > the transition metal-ligand interaction using NBO analysis, in addition > to the work by Maseras and Morokuma CPL 1992. > We frequently use the NPA, and to a lesser extent the NBO analysis for TM compounds. Some references are: Inorg. Chem. 1994,33,2122; Chem.Eur.J. 1996,2,348; J.Am.Chem.Soc. 1996,118,3018. > I have tried applying NBO analysis to [Co(NH3)5Cl]2+ and I find > some difficulties in interpretating some of the results: > One principle problem with the NBO part is that for many TM compounds the program fails to find _one_ Lewis structure. Thus, you should be very careful in examining the Lewis structure you get. Does it make any chemical sense? How large are, e.g., the energy terms in the second-order perturbation theory analysis (values larger than, say, 30-50 kcal/Mol clearly tell you that your Lewis structure does not describe the density matrix well) ? Of course there are other criteria one may look at, such as the percentage of the density matrix described by the Lewis structure. To get around these problems, Weinhold and coworkers have meanwhile developed an extension they call 'natural resonance theory' which expands the density matrix in several NBO Lewis structures. While the theory is, as far as I know, at the moment just available as an internal report of the University of Wisconsin, some interesting applications (to main group species) may be found in: J. Chem. Educ. 1995, 72, 583. I expect this to give a much improved analysis for TM systems as well and hope it will be available soon (it's now part of the NBO 4.0 program). Of course the NPA is no problem, so charges and NAO populations may be used safely even with the present version. One point regards the use of NBO/NPA with DFT (Kohn-Sham) calculations: In principle the Kohn-Sham orbitals only describe a wavefunction for the so-called 'noninteracting reference system' and thus can not be interpreted as the wavefunction for the real system. In practice, however, this does not create any problems. Comparisons with HF, MP2, etc.... density matrices show that the KS orbitals give the expected characteristics (e.g. showing effects of electron correlation) and we have found them to provide a useful basis for qualitative interpretations, in particular for TM systems. Hope this helps a bit. Regards, Martin Kaupp ------------------------------------------------------------------ | Dr. Martin Kaupp | | Max-Planck-Institut fuer Festkoerperforschung, | | Heisenbergstrasse 1, D-70569 Stuttgart, Germany, | | Tel.: country-code+711/689-1532 | | Fax.: country-code+711/689-1562 | | email: kaupp ^%at%^ vsibm1.mpi-stuttgart.mpg.de | | | | and Institut fuer Theoretische Chemie, Universitaet Stuttgart, | | Pfaffenwaldring 55, D-70569 Stuttgart, Germany | | Tel.: country-code+711/685-4399 | | Fax.: country-code+711/685-4442 | | http://www.theochem.uni-stuttgart.de/~kaupp/ | ------------------------------------------------------------------ ************************ The Co-Cl bond that NBO calculates is probably highly polarized toward Cl and might alternatively be described as a Cl lone pair with fairly strong charge transfer interaction with Co. The interaction is apparently strong enough that NBO calculates a bond between these two atoms. Co-Cl --> RY*(16) is then a separate charge transfer interaction between the Cl lone pair and a different hybrid on Co (presumably one of the 3d orbitals?). I gather that you're using the NBO perturbative analysis to judge the degree of delocalization between the ligands and Co. If so, you should probably force NBO to treat the ligands consistently. That is, if NBO doesn't calculate any bonds between Co and NH3, then prevent the program for calculating a Co-Cl bond. This is accomplished with NBO $CHOOSE keylist, which is described in the NBO manual. $CHOOSE allows the user to stipulate the pattern of bonds (the Lewis structure) that NBO will calculate. For your complex, only ask for NH bonds; no bonds would then be calculated between Co and N or Cl. If you don't have an NBO manual and would like to try $CHOOSE, send me a copy of your input deck. I'll add the $CHOOSE keylist to it (it's fairly simple) and return the file to you. Eric Glendening Department of Chemistry Indiana State University Terre Haute, IN 47809 ericg-!at!-chem.indstate.edu ************************** The rule of thumb that I use is to neglect any interaction that is an order of magnitude weaker than the strongest appearing in NBO's perturbative analysis. This works well whenever one is looking for a qualitative description of the delocalization patterns in a molecule (or complex, as in your case). So I'd ignore the RY* interaction. ... It is certainly reasonable to judge the degree of covalent character from the presence or absence of a bonding NBO. If NBO calculates a bond, then the polarization coefficients give a quantitative measure of covalent character; if NBO calculates a lone pair, then covalency can be judged from the strength of the orbital interactions. You won't however be able to compare the covalency of two interactions when NBO assigns one a bond and the other a lone pair. While it is likely that the interaction with the bonding NBO should have higher covalent character, this may not always be the case. I really recommend $CHOOSE as it will allow a consistent comparison of all interactions based on lone pair delocalizations. Eric Glendening ************************ [Co(NH3)5Cl]2+ CO 0.00000 0.00000 0.00000 CL -1.72778 -0.00000 -1.47168 N 1.47849 0.00000 1.27277 N 1.25572 -0.00000 -1.50461 N -0.03050 -1.96150 -0.01613 N -0.03050 1.96150 -0.01613 N -1.27561 0.00000 1.51536 H 1.15559 0.00000 2.23183 H 2.07308 0.81190 1.16605 ... Here's the CHOOSE input for your Co complex: $nbo $end $choose lone 1 3 2 4 3 1 4 1 5 1 6 1 7 1 end bond s 3 8 s 3 9 s 3 10 s 4 11 s 4 12 s 4 13 s 5 14 s 5 15 s 5 16 s 6 17 s 6 18 s 6 19 s 7 20 s 7 21 s 7 22 end $end Note that the "resonance" keyword is not needed when the Lewis structure is specified in $choose. $choose is free-format and reads as follows: First the lone pairs (lone ... end); atom 1 (Co) has 3, atom 2 (Cl) has 4, etc. Then the bonds (bond ... end); a single bond between atoms 3 and 8, a single bond between atoms 3 and 9, etc. Attach the $choose...$end input to the end of your input deck and run; no need to modify your input deck in any other way. NBO always searches the input deck for $choose and uses it if available. Otherwise, NBO defaults to its standard search for a Lewis structure. Also, $choose input will in no way affect the population analysis. Just run NBO once; you'll get the atomic charges and $choose Lewis structure in one calculation. Eric ************************* From WEINHOLD {*at*} chem.wisc.eduThu May 2 08:35:05 1996 Date: Mon, 29 Apr 96 11:44 CST From: WEINHOLD:~at~:chem.wisc.edu To: jerry -A_T- dft Subject: Re: NBO LICENSE AGREEMENT: NATURAL BOND ORBITAL 4.0 PROGRAM Name ______________________________ Institution ______________________________ Address ______________________________ ______________________________ E-mail ______________________________ Telephone ______________________________ I agree that the NBO 4.0 program will remain under my direct control and that the program will not be redistributed to others in any form. ______________________________ ________ signature date Please return this form with $100 remittance (payable to Theoretical Chemistry Institute) to: Theoretical Chemistry Institute c/o Sherry Naffz Department of Chemistry University of Wisconsin Madison, WI 53705 _____ additional manuals : at : $30/each (includes shipping and handling)