GEMSTONE-NOESY

GEMSTONE is a new highly selective NMR technique. Combining it with traditional homonuclear 2D experiments creates exceptionally selective 1D versions of these experiments, which can be useful when trying to characterize highly overlapped spectra. In this post I compare the GEMSTONE-NOESY experiment with the standard 1D selective NOESY.

To compare GEMSTONE-NOESY with a standard 1D selective NOESY I used the halogenated steroid dissolved in chloroform-d that I have used in many recent posts. The structure of the compound and the numbering of its carbon atoms is shown below.

Spectra were collected with selective excitation at 1.803 and 1.841 ppm, which correspond to the H5α and H16α resonances, respectively. An Rsnob shaped pulse with a selective excitation bandwidth of 40 Hz was used. Despite partial overlap in the 1D ¹H spectrum, GEMSTONE can separate these resonances, as seen in the previous post. A NOESY mixing time of 500 ms was used. 

The spectra are shown in the stackplot below. The GEMSTONE-NOESY spectra are the top two traces shown in yellow and purple. For reference, a standard 1D ¹H spectrum is shown in green in the middle with several of the resonances labelled. The lower two traces in red and blue are the standard selective NOESY spectra. The large negative peaks in the selective NOESY spectra are the excitation sites.

One of the first observations is that the selective NOESY spectra both show NOEs to H2α and H3β, whereas only the H5α GEMSTONE-NOESY spectrum does. The H16α spectra should not show NOEs to H2α or H3β. Observation of these signals in the H16α selective NOESY experiment indicates that the selective excitation was not selective enough.

Signals from protons on the opposite face of the molecule (H1β, H12β and H7β) from the selected protons (H5α and H16α) are not observed in any of the selective NOESY spectra, as expected. A signal from the degenerate H4αβ signals is seen in both selective NOESY spectra, but only in the H5α GEMSTONE-NOESY. Again, this demonstrates that the selective NOESY was not selective enough.

The H1α and H15α resonances are separated in the GEMSTONE-NOESY spectra, but the H16α selective NOESY shows them both, when it shouldn't. Finally, the H7α and H9α signals appear in the H5α GEMSTONE-NOESY spectrum but not the H16α, as they should. These two peaks are present in both selective NOESY spectra.

The GEMSTONE-NOESY experiment clearly performs better than the standard selective NOESY experiment. There is no loss in sensitivity associated with the extra selectivity, and setup is no more difficult. There appears to be no reason not to use the GEMSTONE technique when recording 1D selective spectra.

The GEMSTONE selection can also be applied to ROESY, TOCSY and COSY experiments. The COSY variant may not seem particularly useful at first, but when trying to identify highly overlapped multiplets being able to identify the adjacent protons may prove very useful.

Acknowledgments
Thanks again to Prof. Ted Molinski for preparing and providing the sample.