Automated shimming

Despite 50 years of improvements in experimental techniques and spectrometer hardware, the vital step of obtaining a homogeneous static magnetic field in which to carry out experiments remained stubbornly resistant to automation until very recently. The skill of shimming, manually adjusting the currents through a set of 10 to 40 small correction coils positioned around the sample in order to flatten out errors in the magnetic field, is one which all high resolution NMR spectroscopists used to have to learn. Thousands of hours of instrument time are still wasted in shimming each year, despite the availability of effective automated techniques for the admittedly simpler problem of shimming magnetic resonance imaging systems. The difficulty lies in translating methods which use the strong, fast pulsed field gradient systems and strong proton signals found in MR imaging to the much more restricted hardware used in high resolution spectrometers, and in devising methods which can shim to much higher quality without the need for very strong signals.

We showed some time ago that the problem of shimming can be solved efficiently and generally for the restricted case of z shimming, on almost any modern high resolution spectrometer. The technique relies on measuring the phase change as a function of position in a sample when an extra delay is added to a deuterium spin echo experiment. The only field gradient required is that provided by the normal Z1 shim coil; 'homospoil' facilities for pulsing this coil are standard on most spectrometers. This method is faster, more accurate and more robust than manual shimming, typically taking less than a minute for routine shimming.

The problem of shimming in three dimensions is more difficult; the sample spinning used in many routine experiments sweeps most field errors under the carpet, but shimming transverse as well as z gradients is crucial both in the initial setting-up of a spectrometer and in experiments which require a non-spinning sample. The first method which has been shown to work for high resolution NMR without the need for fast 3D pulsed field gradients is edge frequency profile shimming. This is uncomfortably slow for routine use but is already cost-effective for initial shimming of new magnets and probes; our more recent techniques are both faster and more accurate, and have been adopted by a major manufacturer. We have also produced a simple solution to the problem of automated shimming in the presence of bulk sample motion such as that caused by convection.