Direct Methods for Field Homogeneity Optimization in NMR Spectroscopy

NMR spectroscopy is one of the most useful analytical techniques available to the chemist. It is also extremely expensive, making efficient use of instrument time a priority. One of the most irritating tasks in maintaining an NMR instrument is the need continually to readjust the correction ('shim') coil currents used to ensure a uniform magnetic field. This process of 'shimming' has until recently proved surprisingly resistant to automation, largely because the traditional experimental indicator of field quality, the amplitude of the steady-state deuterium lock signal, offers no information on the nature of the field homogeneity. The time spent on shimming varies greatly according to the purposes for which a spectrometer is being used, but a figure of an hour a day spent either in manual shimming or on more or less unsatisfactory automatic methods is not unusual. Considerably longer times are needed for the initial or 'cold' shimming of a new magnet or probe, where no approximate shim settings are available; cold shimming adds significantly to the installation costs of new instruments.


We set out to develop 'gradient shimming' methods, based on magnetic resonance imaging techniques, which would allow a conventional high resolution NMR spectrometer to map its own magnetic field, allowing automatic shimming to be carried out rapidly and accurately without the need for expensive extra hardware. To be of general use in high resolution NMR, an automated shimming method should fulfil a number of criteria. It should be (i) robust; (ii) accurate; (iii) fast; (iv) applicable to all samples; (v) require little or no extra hardware; and (vi) be usable with all types of probe.


Our first development was a deuterium spin echo method for 1D (single axis) shimming. This method is rapid, accurate, highly convergent, and gives results for spinning samples which are better than those obtained manually by the most experienced and patient operators. The shim mapping process (which need only be carried out once) takes 5 - 10 minutes, and iterative correction rarely requires more than 2 -3 minutes even for the most severe field inhomogeneity. For the special case of 1D shimming, the method meets all the six criteria outlined above. This was the first gradient shimming method to be published that did not require a fast pulsed field gradient facility, and hence could be used with almost any spectrometer; it has very rapidly proved its worth in laboratories around the world, and is now a standard feature of most new commercial spectrometers.


The problem of full 3D shimming is considerably more difficult, because of the need to map the field using the shim coils, which take tens of milliseconds to settle. Our first approach was to use a profile edge frequency method for field mapping, because this makes only very modest demands on the spectrometer hardware. This worked surprisingly well given its limitations, and represented a considerable advance on manual shimming methods. It remains the first and only published 3D gradient shimming technique applicable to conventional instruments, but unlike the 1D method outlined above it fails to meet criterion (iv) above, being restricted to samples with a single strong, dominant proton resonance.
To circumvent the limitations of edge frequency mapping we turned to a 3D shimming method based on the same principles of phase mapping as 1D gradient shimming, using stimulated echoes to compensate for the very slow switching speeds of the shim coils. This method has been implemented on 400 and 500 MHz spectrometers, and satisfies all six of our desiderata. It is more than competitive in accuracy and speed with manual shimming, being capable of complete 3D mapping and shimming of a 400 MHz spectrometer in one hour, and requires only minor and inexpensive hardware changes for implementation on any modern high resolution NMR spectrometer.


The 1D and 3D phase mapping techniques essentially solve the problem of both routine and cold automated shimming on high resolution spectrometers. The 1D technique has already been widely accepted and implemented, and we anticipate equal enthusiasm for the 3D technique which should be published shortly.


This work was supported by EPSRC grant GR/L17443

 

GAM home page

Morris research group home page

 

Please send any comments or questions about these pages to:
g.a.morris@man.ac.uk

Most recent revision 4th October 2001