An Improved Method for Heteronuclear Chemical Shift Correlation by Two-Dimensional NMR

Since the use of two-dimensional NMR for correlating the spectra of coupled heteronuclei was first proposed in 1977 (I), a considerable number of papers have appeared describing different techniques for heteronuclear correlation (2-8). The most widely used method (5, 9-12) produces a two-dimensional spectrum in which one signal appears for each directly bonded carbon-hydrogen pair in a molecule. The fi domain of the spectrum is governed by the proton resonance frequencies and thef, by the carbon-13, so that the correlation signal for a given CH, group appears at frequency coordinates which are just the proton chemical shift in fi, and the carbon-13 chemical shift in fi. Heteronuclear couplings are absent from both frequency domains, although proton-proton scalar coupling structure remains in fi. Such experiments are relatively simple to perform, and show good sensitivity; data for a typical shift correlation two-dimensional spectrum may be obtained in about 10 times the time required to produce a good protondecoupled carbon13 spectrum. A number of chemical applications of such shift correlation experiments have been published (9-ZZ), all using the basic pulse sequence of Ref. (5). This sequence generates proton-decoupled carbon-13 free-induction decays which are modulated as a function of r1 by the frequency difference between the proton chemical shift and the proton transmitter frequency. Since the signals are amplitude modulated with respect to t,, it is not possible to distinguish between positive and negative fi frequencies (13), so that to avoid ambiguity it is necessary to position the proton transmitter either to low field or to high field of the proton spectrum. This is unfortunate in that much of the proton transmitter power is thus wasted, which can lead to severe heating problems with ionic samples if the same proton frequency is used for pulses and for decoupling, as is usually the case. A second problem is that twice as many t1 samples are needed to digitize thef, domain of the spectrum if the proton transmitter lies to one side of the proton spectrum. Both from the point of view of optimum decoupling and from that of minimizing data storage needs and complexity of data processing, it would be desirable to be able to place the proton transmitter in the middle of the proton spectrum. This communication suggests the use of phase cycling of the proton and carbon-13 transmitter pulses, similar to that recently introduced in homonuclear correlation experiments (14-16). This converts amplitude modulation as a function oft, into phase modulation, allowing positive and negativef, frequencies to be distinguished. Thus in the usual experiment (5), examination off2 spectra for successive r1 values would show the amplitudes of individual signals oscillating while their phases remained fixed. The proposed modification would lead tof, spectra with signals oscillating between absorption and dispersion mode while