Reports: ND656195-ND6: Gas-phase NMR and Computations on Hydrocarbons in Volatile Fractions of Crude Oil

Gerard S. Harbison, PhD, University of Nebraska, Lincoln

As we proposed to PRF, we have collected and assigned a large database of 1H and 13C gas phase hydrocarbon NMR spectra, including n-pentane, n-hexane, 2-methylbutane, 2,3 dimethylbutane, 2-methylpentane, 3-methyl pentane, cyclopentane, cyclohexane, propene, cis- and trans-2-butene, and 1,3 butadiene.

We have obtained and analyzed several representative raw crude-oil samples, including South-Central Texas Heavy , Light Pennsylvania, Colorado Sweet, Arabian Sea Heavy, and Basra Light. We have developed protocols for collecting volatile fractions for gas phase NMR; this included building a customized stainless-steel gas manifold to fill NMR tubes at defined temperature and pressures, with and without adjunct gases to improve spin-relaxation. We have obtained and fully assigned volatiles from the Texas heavy crude, and identified signals from methane, ethane, propane, butane, various isomeric pentanes and hexanes as well as traces of cyclopentane and cyclohexane. We were struck by the lack of any downfield olefinic or aromatic signals. Our two-dimensional methods have been invaluable in disentangling these complex spectra. We have also begun the process of computing these spectra, including the use of vibrational self-consistent field methods for obtaining better vibrational averages. Comparison of a database spanning the full range of published gas-phase 13C chemical shifts, with shieldings computed using complete basis set extrapolation of the Dunning aug-cc-pVnZ basis sets, electron correlation via coupled cluster methods, and vibrational averaging using the normal mode approximation, gave a root mean square deviation between computation and experiment of 1.6 ppm, with systematic deviations between computation and experiment of less than 0.2 ppm. While these accuracies would be considered excellent for most chemical shielding calculations, at approximately 1% of the full shielding range, they are far less accurate than other computed small-molecule properties (e.g. the electric dipole moment). Why this is, is at present a mystery. Moreover, in most practical NMR computations, for reasons of computational feasibility, density functional theory replaces coupled-cluster and complete basis set extrapolation is impossible. Errors are therefore substantially greater.

We have developed new two-dimensional double and zero-quantum experiments for use in gas-phase NMR. The new sequences use a combination of phase cycling and gradients to achieve multiple-quantum selectivity. In the course of the research, we found that conventional gradient strategies will not work because of the high diffusion constants of low-molecular weight gases. We therefore adopted a different approach, choosing the lengths of the gradients so that unwanted coherences fall at nodes in the integrated magnetization, which has a Bessel-function-like time dependence. Finally, we developed a related sequence that allows measurement of relaxation of transverse magnetization of either multiple quantum sequence. In our paper, and we showed T2 relaxation curves for the 1H and 13C single quantum relaxation, as well as the zero and double quantum relaxation, in 13C1-acetylene at natural abundance.