On the calculation of 15N chemical shifts using linear regression formulae. A performance comparison of different methods

The calculation of NMR chemical shifts using linear regression formulae is revisited in the present work. 15N chemical shifts are considered as a test case. Mean unsigned errors of 4.8, 5.4 and 7.6 ppm are found for the B3LYP/6-31G(d)//B3LYP/6-31G(d), B3LYP/6-311+G(2d,p)//B3LYP/6-31G(d) and HF/6-31G(d)//B3LYP/6-31G(d) levels of theory, respectively, which are much more accurate than chemical shifts calculated by subtracting theoretical magnetic isotropic shieldings from the reference nitromethane. Moreover, the economical HF/6-31G(d)//B3LYP/6-31G(d) method is found to be about 15 times faster than the rather expensive B3LYP/6-311+G(2d,p)//B3LYP/6-31G(d) level of theory but only 2.5 times faster than B3LYP/6-31G(d)//B3LYP/6-31G(d). Our results indicate that the B3LYP/6-31G(d)//B3LYP/6-31G(d) level of theory constitutes an adequate compromise between accuracy and computational cost for 15N chemical shifts calculation using a linear regression formula.