Multivariate data analysis of carbon-13 nuclear magnetic resonance substituent chemical shifts of 2-substituted naphthalenes

Abstract

The ¹³C n.m.r. chemical shifts of several 2-substituted naphthalenes have been analysed by principal components data analysis. In addition, the shift values have been related to the ¹³C n.m.r. chemical shifts of the corresponding monosubstituted benzenes by means of partial least-squares data analysis. This latter analysis showed that nearly all of the systematic variation in the naphthalene shift data set could be predicted from the shift data of the monosubstituted benzenes. Moreover, it is shown that the grouping of substituent effects noticed for most benzene derivatives is also present in the naphthalene data. The advantage of using partial least-squares data analysis compared with the conventional dual substituent parameter analysis is also demonstrated. In fact, the overall description of the ¹³C shifts using the σ_I, σ_R dual-substituent parameter correlation is not improved compared with the use of the mean values of σ_I and σ_R for the donor, acceptor, alkyl, and halogen subclasses. Finally, two existing methods for the determination of the relevant number of substituent parameters is discussed. A slightly conservative cross-validation method is argued to be better than methods where the variance is explained as being due to the experimental error.