Search

A new basis set of triple-ζ quality for carbon, 3z-S, is developed and tested at the DFT, MP2, and CCSD(T) levels, taking into account solvent and vibrational corrections for a number of molecules ranging from the smallest fluoromethane, CH 3 F, to the largest 5,10,15,20-tetraphenylporphyrin, C 44 H 30 N 4 . The proposed highly economical 3z-S basis set has been proven to provide very good accuracy in all examinations, comparable to that of the NMR-oriented Jensen's pcS-2 basis set, which is about 50% larger than 3z-S.

An accurate quantum chemical modeling of 125 Te NMR spectra is of great importance in the NMR structural assignment for real-life tellurium compounds, which represent a growing interest in organic and inorganic chemistry nowadays. This work reports a computationally modest combined approach based on the density functional theory only, which provides an excellent accuracy against the experiment and can be effectively applied for the routine large-scale calculations of tellurium chemical shifts. The role of solvent, vibrational, and relativistic corrections has been thoroughly investigated. Special attention was paid to the effect of taking into account the scalar relativistic effects during the geometry optimizations on the calculated tellurium chemical shifts.

New basis sets, acvXz-J (X = 2, 3, 4), were designed for tellurium and selenium atoms for the calculation of indirect nuclear spin-spin coupling constants involving selenium or tellurium nuclei. Saturation of angular spaces was performed in an even-tempered manner till achieving the convergence of spin-spin coupling constants under consideration. The high-accuracy correlated SOPPA(CCSD) method was employed in this procedure. Saturated basis sets were contracted by approximately 30% using the coefficients obtained from the atomic self-consistent field (SCF) closed-shell calculations of selenium and tellurium anions, Se 2- and Te 2- . Final basis sets of acvXz-J showed sufficiently high accuracy in comparison with previous special basis sets.

Phosphorus-carbon spin-spin coupling constants in a series of salient heterocyclic phosphines were calculated at the SOPPA(MP2) level including evaluation of relativistic and solvent effects. A number of the locally dense basis set schemes were thoroughly investigated in terms of their accuracy versus computational demands. The most effective computational scheme was tested in a benchmark series to provide a very good correlation between 2 J PC calculated at the SOPPA(MP2) level and experiment. A marked dihedral angle dependence of 2 J PC was demonstrated, and this could be used in stereochemical studies of a wide series of organophosphorus compounds based on the phosphorus-carbon coupling constants.

A number of computational schemes for the calculation of 15 N and 31 P NMR chemical shifts and shielding constants in a series of azoles, phospholes, and phosphazoles was examined. A very good correlation between calculated at the CCSD(T) level and experimental 15 N and 31 P NMR chemical shifts was observed. It was found that basically solvent, vibrational, and relativistic corrections are of the same order of magnitude and alternate in sign, being, on average, of about 2-3 ppm in absolute value but, being much larger (up to 14 ppm) in the case of solvent molecules explicitly introduced into computational space. At the DFT level, the performance of nine exchange-correlation functionals including six conventional gradient functionals and three hybrid functionals was studied. The most accurate results were reached with the OLYP and Keal-Tozer's family of functionals, KT1, KT2, and KT3, while the most popular B3LYP and PBE0 functionals showed the most unreliable results. On the basis of these data, we highly recommend OLYP and KT2 functionals for the computation of 15 N and 31 P NMR chemical shifts at the DFT level in the diverse series of nitrogen- and phosphorus-containing heterocycles. Benchmark calculations of 15 N and 31 P NMR chemical shifts in a series of larger nitrogen- and phosphorus-containing heterocycles were performed at the DFT level in comparison with experiment and revealed the OLYP functional in combination with the aug-pcS-3/aug-pcS-2 locally dense basis set scheme as the most effective computational scheme.

Four-component relativistic calculations of 125 Te NMR chemical shifts were performed in the series of 13 organotellurium compounds, potential precursors of the biologically active species, at the density functional theory level under the nonrelativistic and four-component fully relativistic conditions using locally dense basis set scheme derived from relativistic Dyall's basis sets. The relativistic effects in tellurium chemical shifts were found to be of as much as 20-25% of the total calculated values. The vibrational and solvent corrections to 125 Te NMR chemical shifts are about, accordingly, 6 and 8% of their total values. The PBE0 exchange-correlation functional turned out to give the best agreement of calculated tellurium shifts with their experimental values giving the mean absolute percentage error of 4% in the range of ∼1000 ppm, provided all corrections are taken into account.

Relativistic calculations of (29)Si NMR shielding constants (chemical shifts) in the series of halosilanes SiX(n)H(4-n) (X = F, Cl, Br and I) are performed within a full four-component relativistic Dirac's scheme using relativistic Dyall's basis sets. Three different theoretical levels are tested in the computation of (29)Si NMR chemical shifts in comparison with experiment: namely, four-component relativistic GIAO-DFT, four-component relativistic GIAO-RPA, and a hybrid scheme of a nonrelativistic GIAO-MP2 with taking into account relativistic corrections using the four-component relativistic GIAO-RPA. The DFT results give larger relativistic effects as compared to the RPA data which might be rationalized in terms of the manifestation of correlation effects taken into account at the DFT level and not accounted for at the uncorrelated RPA level. Taking into account solvent effects slightly improves agreement with experiment, however, being not a matter of principle. Generally, relativistic pure nonempirical wave function methods perform much better as compared to relativistic DFT methods when benchmarked to experiment.