The group of Professor Lee Cronin at the University of Glasgow has combined machine learning with a chemical reaction system to speed up the discovery of new chemical reactions, which is an inherently unpredictable and time consuming process. This new approach of an Organic Synthesis Robot uses a Spinsolve Benchtop NMR spectrometer as an integral component. Their work has just been published in the prestigious journal Nature: J. M. Granda, L. Donina, V. Dragone, D.-L. Long and L. Cronin, Nature559, 377–381 (2018), DOI: 10.1038/s41586-018-0307-8
Photograph of the chemical robot
The photo shows the impressive setup of the chemical robot with 27 pumps, valves and six reactors, as well as NMR, IR and MS spectrometers for real-time analytics.
In my first two posts on using 1D and 2D NMR methods to assign the peaks of quinine (Figure 1), I looked at the 1H and 13C spectra.
Figure 1. Structure of quinine
In this post, I’m moving on to look at the 1H-13C HSQC spectrum. It’s worth spending a brief moment recapping what HSQC is all about and what info it gives you. In a nutshell, the HSQC experiment correlates proton and carbon chemical shifts over one chemical bond. Another way to put this is that a cross-peak in an HSQC spectrum says, “The proton with this chemical shift is directly attached to the carbon with that chemical shift”. By convention, HSQC spectra are presented with 1H shifts along the horizontal axis and 13C shifts along the vertical axis.
Some variants of HSQC also encode into the phases of the cross-peaks additional information about how many hydrogen atoms are attached to each carbon atom. This is sometimes referred to as multiplicity or DEPT editing. In the multiplicity-edited HSQC spectrum, it is conventional for the CH and CH3 groups to have positive phase, and the CH2 groups to have negative phase, just as in a DEPT-135 spectrum. Figure 2 shows the multiplicity-edited HSQC (“HSQC-ME”) spectrum of our 400 mM quinine sample. The CH2 signals are shown in blue and the CH and CH3 signals in red.