From dsmith@CTCnet.Net Thu Oct 31 13:25:24 1996 Received: from home.CTCnet.Net for dsmith@CTCnet.Net by www.ccl.net (8.8.2/950822.1) id NAA16118; Thu, 31 Oct 1996 13:16:32 -0500 (EST) Received: from fock by home.CTCnet.Net (SMI-8.6/SMI-SVR4) id NAA14166; Thu, 31 Oct 1996 13:14:45 -0500 Date: Thu, 31 Oct 1996 13:14:45 -0500 Message-Id: <199610311814.NAA14166@home.CTCnet.Net> X-Sender: dsmith@pop.dasgroup.com X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: CHEMISTRY@www.ccl.net From: dsmith@CTCnet.Net (Douglas A. Smith Ph.D.) Subject: isocyanates: summary Quite a while back I asked a question of the list regarding the calculated non-linearity of isocyanates, i.e. R-N=C=O. I did get a reasonable response, and apologize for the summary not appearing earlier (we changed Internet access providers and were without mail access for over a month!). Anyway, here is my original posting and the edited responses. The answer, in short, is that the bond is NOT linear. I hope that those who answered that it should be linear, or thought that it should be linear, are not offended. Doug ============================================================================ =========== My original posting: However, the results are not intuitively obvious. For example, using H-N=C=O as a test case, the bond angles and distances are as given in the table. distance angle N-C C-O N-C-O ------- ------- --------- AM1 1.23 A 1.20 A 166.7 PM3 1.25 A 1.18 A 166.7 RHF/6-31G* 1.20 A 1.15 A 174.2 Similar deviations from linearity are seen in the angles of substituted isocyanates. What is the expected N-C-O angle based on experiment and/or higher levels of theory? ============================================================================ =========== Should be linear. I would suggest trying an H-N=C=O CASSCF calculation. A CCSD(T) calculation should also be feasible with a basis set of this size on an RS/6000 type of machine. This smells like artifactual symmetry breaking to me. I find it extremely interesting that RHF/6-31G* gave a non=linear N-C-O angle. Also the 6-31G* has been known to give result in bent geometries for linear carbon clusters,( such as C4-C7). I would recommend using the pVDZ basis set for exploring this aspect of the problem in a cost efficient fashion. If you do not have any access to Xray diffraction data of similar systems, and are concerned about site modeling or solid state effects, I would reccomend fixing the N-C-O angle to be linear. -Galen F. Gawboy Pinnacle Technology Resources, Inc. ============================================================================ =========== The only experimental evidence I know of for isocyanic acid, HNCO, is from spectroscopic data: Hocking, Gerry and Winnewisser, Can. J. Phys., 53, 1869 (1975) Hocking, Gerry and Winnewisser, Astrophys. J., 174, L93 (1972) as quoted by Harmony et al, J. Phys. Chem. Ref. Data, 8(3), 619 (1979). I don't have copies of the primary literature with me, but Harmony's comments is that "it was necessary to assume linearity of NCO group. The reported values of NH and HNC are most strongly affected by this assumption, but the heavy-atom distances are also affected to a smaller extent." Values quoted: NC 1.209A CO 1.166 NH 0.986 HNC 128.0 Hope this is of some help to you. Please summarise for CCL, as I'd be interested in seeing responses that you get. Prof. Helder M. Marques, Department of Chemistry University of the Witwatersrand ============================================================================ =========== I can't give you any experimental results on isocyanates, but I have optimized HNCO (Cs symmetry) at various levels (see below). The simple answers to your last question are 1) HNCO is not linear at any level of theory (including DFT and MP2), and 2) semiempirical methods do not "fail"; they actually reproduce other theoretical models reasonably well. Here are my results (they are close to, but not identical to what you reported). Theory Energy r(NC) r(CO) angle(NCO) AM1 -15.15 1.232 1.202 166.7 PM3 -15.29 1.251 1.181 168.6 HF/6-31G* -167.761384 1.200 1.145 174.2 MP2/6-31G* -168.232430 1.224 1.184 171.7 HF/6-31G** -167.766024 1.199 1.148 174.3 MP2/6-31G** -168.242279 1.225 1.183 171.7 B3LYP/6-31G** -168.681498 1.218 1.174 172.2 BP86/6-31G** -168.689231 1.230 1.186 171.3 SVWN/6-31G** -167.794926 1.216 1.177 172.3 Comments on structures: 1. Basis set has little effect on HF and MP2 structures 2. HF bond distances are much shorter than MP2 (especially CO distance) and bond angles are larger 3. Both PM3 and AM1 duplicate MP2 structure at least as well as HF 4. Agreement between MP2 and DFT is good for all types of DFT, but best agreement is between BP86 and MP2 5. Bond angle at MP2 and DFT is approx 172. The NCO unit is NOT linear. Comments on calculations: 1. All calculations used Cs symmetry 2. AM1, PM3, HF, MP2 calculations performed with Spartan 4.0.3 3. MP2 optimizations done with "unfrozen" core 4. DFT calculations performed with Gaussian 94 5. Also tried SVWN/6-31G** with PS-GVB (fine mesh) and obtained essentially identical result. If I recall correctly, sometime back in the '80s (or possibly very late '70s),HF Schaefer published an article in JACS on the structure of cumulenes like 1,2,3-butatriene (H2C=C=C=CH2) and found that these were not linear. I don't have the reference, but you should be able to find it without too much trouble. Good luck! Alan Shusterman Department of Chemistry Reed College ============================================================================ ============ We have been doing some computational research involving isocyanates, and we have found that, with the exception of silyl isocyanates, these cumulenes are not linear. This result is in agreement with experimental data. The sp-hybridized carbon atom does not yield a bond angle of 180 deg. because the lone pair of the nitrogen atom interacts with the pi*(C=O)orbital. This stabilizaing two-electron interaction is enhanced when the O-C-N atoms are not linear. Therefore, the HF results, particularly those obtained at HF/6-31G* level, reflect correctly this interaction. Fernando Cossio ============================================================================ =========== I've performed _ab initio_ calculations on .alpha.-methylazo alkyl isocyanates some years ago (RT Kroemer et al., J. Chem. Soc. Perkin-II (1994) pp2129). These calculations were HF/6-31G* and MP2/6-31G*. The most remarkable geometrical change at these levels of theory was a decrease in the N-C-O bond angle from 171.5 to 167 degrees. This small angle can be partially explained by a perturbation induced by the nearby diazene group. However, as I was initially surprised by the fact that the N-C-O group is not linear, I performed calcs on CH3-N=C=O and found a bond angle of 174.9 degrees at the HF/6-31G* level. The non-linearity and the decrease in the angle comparing HF/6-31G* and MP2/6-31G* was also observed by other people (S Ellis et al., Can. J. Chem. 69 (1991) pp1000) for H2B-N=C=O and F2B-N=C=O. If you perform a search for R-N=C=O in the Cambridge Crystallographic Database, you will find at least 50 structures (either -N=C=O as a ligand in metalloorganic compounds or as part of "real" organic compounds). In none of these structures you will find absolute linearity, the N-C-O angle varying from approx. 165 to 178 degrees. Of course one can argue that the levels of theory mentioned above are not enough (one needs larger basis sets and more sophisticated schemes for taking electron correlation into account than MP2), and in the crystal structures the N-C-O angle is non-linear because of neighbour group effects and crystal packing etc. Therefore, it would be interesting for me to know where you found the "observed linearity of the N-C-O" angle (maybe spectroscopy?). Maybe you could try a really sophisticated calculation on H-N=C=O (such as QCISD(TQ) or a coupled cluster calculation with a large basis set, the system itself is not very large). Dr. Romano T. Kroemer Phys. & Theoret. Chem. Lab. University of Oxford ============================================================================ =========== -- Dr. Douglas A. Smith, President and CEO | voice: (814) 262-9091 The DASGroup, Inc. | fax: (814) 262-9337 P.O. Box 5428 | email: dsmith@dasgroup.com Johnstown, PA 15904-5428 | Contract R&D specialists in computational chemistry, process modeling, synthesis and design of novel compounds for chemistry, materials science, and biotechnology.