Computer Simulation of NMR Spectra

This is principally a report of developments since the last newsletter, so the previous article should be consulted for background. There has been steady but unspectacular progress in calculating and analysing magnetic properties useful for computer simulation of NMR spectra. I have continued to search for agreement with experiment and descovered some areas where our model is insufficiently responsive to chemical changes but have made rationalisations of these shielding tensors with their directional components from both experiment and DALTON calculations. There are solid state NMR spectra available which gives components of the tensor to compare with the great detail available from calculations. This work has been presented internally in student reports and talks but is not yet published.

However work producing more A and B Buckingham parameters was done at the same time and is in press [1]. An attempt was made to rationalise signs and magnitude of A, an intrinsically  much more difficult task than for the dipole moment vector. Some continuum solvation calculations were done at the same time, on TDMS, one of the commonest reference molecules used in NMR. There is extensive experimental data on the solvation shifts of this molecule. The calculations so far only partially predict this data.

Spin-spin coupling constants have been calculated with the DALTON program using the coupled-cluster propagator and have verified, (not surprisingly),  both the Karplus equation and the methodology [2]. Spin-spin coupling calculations are an active area in other research groups so there is serious competition here, (see bibliography).
 
 

The meaning of these numbers, including the orientational definitions, is described in the full text. What can be seen here is that there is a chemical sense to the values of this exotic property and within the limited data available that the correlation contribution is reasonably small, (4%), especially when compared to the basis error, which is expected to be also at least 4%.
 

New articles

[1] M.  Grayson and P.  Chittenden, "The Magnetic Shielding Polarizabilities of Some Tetrahedral Molecules", in press,  International Journal of Molecular Sciences.

[2] M. Grayson, and  S. P. A. Sauer, "The computation of Karplus Equation coefficients and their components using self-consistent field and second order polarization propagator methods'', in press,  Molecular Physics.
 

The Serious Competition in Spin-spin Coupling

[4] H. Sekino  and  R. J. Bartlett, Chem. Phys. Lett., 225, 486 (1994).

[5]  H. Fukui, H. Inomata, T. Baba, K. Miura and H. Matsuda, J. Chem. Phys., 103, 6597 (1995).

[6] M. Pecul and  M. Jaszunski  and J. Sadlej, Chem. Phys. Lett., 305, 139 (1999).

[7] J. Guilleme,  J. San-Fabián,  J. Casanueva and E. Díez, Chem. Phys. Lett.,  314, 168 (1999).
 
 

Materials Computation

There has been some progress here using quantum mechanical calculations to produce numbers in support of the phenomenological models used in liquid crystal simulations [1]. The quantum mechanics provides electrical moments, van der Waals parameters and molecular geometries. Often the calculations have had to be semiempirical, using MOPAC, rather than ab initio because of the size of the molecules.

The work on the larger molecules has also generated interest in accurate values of the properties of smaller molecules which share functional groups with the larger ones [2]. Some of the small molecule work should be publishable because of its fundamental science interest, as it contains calculations with different levels of correlation and methodology and can act as a benchmark on the accuracy of semiempirical calculations.

No new calculations of nonlinear optical properties have been done but modelling of these properties will ultimately be required in their liquid crystal context. Calculations on the nonlinear chromophore are well within the capability of our ab initio programs and the resulting tensors can be transformed to the local axis system imposed by the liquid crystal simulation.

The sort of molecules which required MOPAC calculations on the full systems were the (S)-2-chloroalkyl esters investigated by Goodby et al., (Liquid Crystals 14, 37 (1993)). The following pictures show one optimised geometry in the inertial frame as the halogen is varied from F to I. There are so many degrees of rotational and librational freedom that the choice of geometry, Boltzmann weighted geometry set or dynamics snapshots is almost arbitary. For the fluorine containing chloroalkyl ester the MOPAC timings on a Silicon Graphics 4400 chip were 2 hours. I estimated a CADPAC ab initio run with a double zeta basis would be a week and a half, (460 basis functions), but there are probably much better implementations  of CADPAC than ours running! Double zeta polarised bases are only of the order of a thousand basis functions so calculations are possible but at what geometry and how many.

The molecules differ only in the different sizes of the halogen atom which has a small effect on the space descriptors which might be used. The heavier halogens move the centre of mass nearer the chiral centre. Experimentally as the mass and size of the halogen increases and the
electronegativity decreases the temperature range of the smectic A phase increases. Sorting out a model which explains this phenomenon and separates steric, electronic and librational effects is a challenge for the future.

What is New?

[2] M. Grayson, "Monopoles, Dipoles and Quadrupoles, A Chemist's Perspective'', in Recent Research Developments in Quantum Chemistry, Transworld Research Network, Trivandrum, India, 1, ed. S.G. Pandalai, pp1-18 (2000), following a talk given at the Computational Collaborative Project 5, (Annual Meeting), 6-8th September 1999, at the University of Birmingham.