Gordon Group News

While enjoying a leave of absence this year at Tokyo Metropolitan University in Professor Nagase's group, I have had the chance to spend a lot more time on scientific matters.  It's been fun, although I feel sympathy for Brett Bode who has been called on to take care of the day-to-day GAMESS business.

As part of a larger project, I have had the chance to revisit silatranes, which we[1] examined shortly after developing a parallel SCF code.  The silatrane structure positions a nitrogen atom in such a way that a 5th bond, across the tricyclic ring system may be formed to silicon.  The strength of this bond has been a matter of interest ever since the discovery of the first such compound in 1961[2], and it is now known to be rather weak.  First, in the solid state, the SiN distance is now known to vary between 2.007 Angstroms[3] for thiocynate and 3.000 Angstroms[4] osmium complex substituents.  Second, the geometry of methyl[5] and fluorosilatranes[6] in the gas phase is known to have a longer  SiN bond, about 0.27 Angstroms greater than the solid state, and NMR evidence suggests solutions of silatranes adopt intermediate bond distances[7].  The great variation in bond distance with substitution or media change can only be due to a very small SiN bond energy.

The point of our original work[1] was to identify trends in SiN bonding as a function of equitorial as well as axial substituents, R= H, CH₃, NH₂, OH, F, SiH₃, PH₂, SH, Cl, and in addition to the silatranes Y= O, we also examined Y= NH and CH₂.

The parallel SCF program permitted us to optimize all these combinations, but the computed structures had SiN bond distances that were 0.28 and 0.21 Angstroms longer than the methyl and fluorosilatrane gas phase structures, meaning the computed values were all around 0.5 Angstroms longer than the more numerous X-ray structures.  While the trends may well have been computed properly, the absolute bond distance errors are such that one could not be sure.  Therefore, we probed both the basis set at the SCF level, and MP2 single point energies, to suggest that resolving the difference between computation and electron diffraction experiments would require both improved basis sets and dynamic correlation.  The recent paper of Anglada, Bo, Bofill, Creheut, and Poblet[8] shows that MP2 computations do in fact give a good structure for fluorosilatrane.

Since we[9] have completed a scalable parallel MP2 gradient program recently, it is now feasible to examine the silatranes with the more accurate MP2 model.  Our own results for three silatranes are shown in the table:
 

so that rather close agreement with experiment is obtained when the Si atom is given two sets of d's and triple zeta basis sets are used on all atoms.  As Anglada et al. had found, even the MP2/6-31G(d) result is acceptable.

Accordingly we have computed the full harmonic force field of these three silatranes, for comparison to experimental spectra, and to obtain force constants for the transannular SiN stretch as a function of substituent.  Early experimental work typically used KBr pellets to contain the sample, and this material is opaque at low frequencies.  Thus the first paper[10] to correctly observe the SiN stretch in R=organic silatranes reported that this frequency is in the range 320-390 cm^-1.  The parent (R=H) silatrane was observed in Raman experiments in France[11], and in IR and Raman experiments in Russia[12].

For the C₃ symmetry of silatranes, both a and e vibrational modes are active in both IR and Raman, but of course the intensity patterns for each will be different.  Accordingly, we have written a program to carry out the numerical differentiation[13] with respect to electric fields needed to obtain the necessary polarizability derivative tensor.  Given the existence of parallel MP2 nuclear gradients, this methodology yields the MP2/6-31G(d) Raman intensities in about 1/3 of the computer time required to obtain the harmonic force field at the same level.  Given both IR and Raman intensities for the theoretical force field, we may compare to experiment.

The French group[11] performed only Raman experiments, and thus do not observe the SiN stretch, which has a very low Raman intensity.  The French spectrum is shown below:

with just a hint of noise (signal?) at the relevant frequency 313 cm^-1.  The Russian group[12] observed a strong IR and weak Raman band at this position, but failed to make the assignment of this band as the SiN stretch.  Our computations lead us to make the following assignments for the low frequency part of the silatrane spectrum,
 

                                                   --- Ref 12 ----                                                  ---- MP2/6-31G(d) ----

where vertical bars group absorptions we believe are overlapping in the experiment.  Note that the computed band at 267 cm^-1 has a very large IR intensity and smaller Raman signal, and its normal coordinate analysis shows that it is 62% SiN stretch.  An illustration of this mode shows convincingly that the observed 313 band must be assigned to the SiN stretch.  The computed SiN force constant in the parent silatrane is 0.225 millidyne/Angstroms.
 
 

In conclusion, it is satisfying that MP2 calculations are able to resolve not only the problem of obtaining satisfactory computed geometries for the silatranes, but they are also able to give sufficiently accurate vibrational frequencies and intensities that the silatrane transannular stretch can be assigned at 313 cm^-1.

I would like to thank Jamie Rintelman and Akihiko Yoshikawa for their very valuable assistance in locating the experimental vibrational papers, and Prof. Nagase for his very kind hospitality as well as the computer resources which these computations required.

  1. M. W. Schmidt, T. L. Windus, M. S. Gordon J. Am. Chem. Soc. 117, 7480-6(1995).
  2. C. L. Frye, G. E. Vogel, J. A. Hall  J. Am. Chem. Soc. 83, 966-7(1961).
  3. S. P. Narula, R. Shankar, M. Kumar, R. K. Chadha, C. Janiak Inorg. Chem. 36, 1268-73(1997).
  4. M. T. Attar-Bashi, C. E. F. Rickard, W. R. Roper, L. J. Wright, S. D. Woodgate  Organometallics  17, 504-6(1998).
  5. Q. Shen, R. L. Hilderbrandt  J. Mol. Struct.  64, 257-62(1980).
  6. G. Forgacs, M. Kolonits, I. Hargittai  Struct. Chem. 1, 245-50(1990).
  7. V. A. Pestunovich, B. Z. Shterenberg, E. T. Lippma, M. Ya. Myagi, M. A. Alla, S. N. Tandura, V. P. Baryshok,
         L. P. Petukhov, M. G. Voronkov  Doklady Phys. Chem. (English Translation) 258, 587-90(1981).
  8. J. M. Anglada, C. Bo, J. M. Bofill, R. Creheut, J. M. Poblet Organometallics  18, 5584-93(1999).
  9. a) G. D. Fletcher, M. W. Schmidt, M. S. Gordon Adv. Chem. Phys. 110, 267-94(1999).
      b) G. D. Fletcher, M. W. Schmidt, M. S. Gordon, B. M. Bode Comput. Phys. Commun. 128, 190-200 (2000).
10. I. Birgele, I. Mazeika, E. Liepins, E. Lukovics J. General Chem.  (English Translation) 50, 711-714(1980).
11. a) M. Imbenotte, G. Palavit, P. Legrand J. Raman. Spectrosc.  14, 135-7(1983).
      b) M. Imbenotte, G. Palavit, P. Legrand, J. P. Huvenne, G. Fleury J. Mol. Spectrosc. 102, 40-55(1983).
12. I. S.  Ignat'ev, A. N. Lazarev, S. G. Shevchenko, V. P. Baryshok Bull. Acad. Sci. USSR, Chem. Sci.
        (English Translation) 1375-1381(1987).
13. a) A. Komornicki, J. W. McIver  J. Chem. Phys. 70, 2014-6(1979).
      b) G. B. Bacskay, S. Saebo, P. R. Taylor  Chem. Phys. 90, 215-24(1984).