Analytical studies in geology push constantly towards the ideals of being fast, in situ and micro. Sputtering using a continuous and focussed primary ion beam generates a continuous stable beam of secondary ions well-suited to precise ratio measurements but it forms many different molecular ions in addition to single nuclides. This results in isobaric interferences when analysing multi-element minerals which become crucial if trace-element abundances are to be determined accurately. A double focussing mass spectrometer of high mass-resolution must therefore be used to separate the ions of interest from their unwanted isobars by the small nuclidic mass-differences between them.

High mass-resolution limits the sensitivity of mass-analysers. Wide slits are needed for high sensitivity but narrow ones for high mass-resolution. If we choose an object slit as large as 100 microns (small compared with most sector instruments), then a symmetrical sector magnet needs to be 1000 mm in turning radius to achieve mass resolution of 10⁴. This means use of a very large magnet which flies against the philosophy of building smaller instruments. Nevertheless, it was the only solution in 1974 when we first faced the problem, and it remains the best established solution today .

The ion optical design of any high-acceptance mass-analyser must minimise the complex image aberrations of the resolved beam that are strongly dependent on the acceptance. A publication in 1974 by H. Matsuda of Osaka University addressed exactly this problem: what combinations of parameters for a cylindrical electrostatic analyser and sector magnetic analyser combined with other ion optical components will produce the smallest summed image-aberration? Matsuda found an optimum family of 'forward geometry' solutions, which were confirmed by my colleague S. Clement using an independent analysis based on beam transport theory. In response to this solution and to the need for micro isotopic analysis in geology, the Research School of Earth Sciences at ANU supported our building a large ion probe as a long-term project. The first so-called geological ion probe, termed SHRIMP for Sensitive High Resolution lon MicroProbe, was coaxed to its performance specifications in 1981 (aberrations summed to not more than 6 microns in a total image width of 45 microns). Despite the large sizes of the ESA and the magnet, fields of the required uniformity and shapes can be made sufficiently well to conform to the Matsuda geometry. Several more SHRIMPs have since been built, and the field of isotopic dating of U-bearing minerals has been transformed.

Evolution of these sector dinosaurs is continuing, based on new mass analyser ideas published by Matsuda in the early 1990s. Matsuda calculated that it is possible, using particular ion optical configurations with a reverse-geometry sector mass analyser, to demagnify the image-width at the collector without also demagnifying the mass-dispersion. Image aberrations must be still smaller to utilize this. We will describe a specific combination that will give still-higher mass resolution with a large magnet while maintaining the present SHRIMP-level of sensitivity. While the existing SHRIMPs operate routinely at 5500R with flat-topped peaks and an 80 micron object-slit, the new model should achieve 21000R with the same ratio of image to collector-slit width, the same object-slit width, and the same sensitivity. The higher resolution will open up a number of valuable geological applications that are denied at 5500R owing to particular isobaric interferences. An example is the interference of Ca⁺ dimers with the Sr isotopes in Ca-rich minerals: operation at 15000R will allow the precise and reliable measurement of ⁸⁷Sr/⁸⁶Sr ratios in limestones, which has geological applications such as tracing the secular evolution of seawater geochemistry.