The design considerations for the electrostatic deflection field are basically the same as those outlined for the magnetic field. In order to reduce dispersion and preserve the resolution it is important to have the electrostatic field as uniform as possible over as small a volume as possible. In an ideal Thomson parabola analyser this field should be placed far enough from the image plane such that only small deflections are necessary. Also, unlike the magnetic field, the direction that the particle is deflected is in the same direction as the field. This means that any non uniformity in field that exists in the direction normal to the plates will have an effect on the ideal operation of the analyser.
Traditionally Thomson parabola analysers used in the laboratory have used the same volume for the magnetic and electrostatic field {Weber et a l , 1986, Rhee,
1984, Lu et a l, 1997). The gap between the magnetic pole pieces then directly determines the lower limit of the measurable velocity for a given length of analyser. To achieve a large dynamic range it would necessary to increase the magnetic gap. Conversely, as previously discussed a smaller gap provides a much more uniform field geometry. Sakabe et a l (1980) developed a modified Thomson parabola spectrometer where the two field regions are separated. The magnetic deflection region can then use a narrow gap providing the uniform field required and the electrostatic section can have as large a gap as required to increase the dynamic range.
The requirements for FONEMA are for a dynamic range of 20 in energy which can be achieved using either method described above. However, the need for a magnetic field return path meant that it was impractical to use the magnets as the high voltage plates for the electrostatic field as well. This also allowed electrostatic deflection tests to be carried out independently of the magnetic deflection tests. For FONEMA then the electrostatic deflection takes place after the magnetic deflection as in the high dynamic range Thomson analyser of Sakabe et a l. The amount of space that was available made it impractical to use a pair of parallel plates to define
the electrostatic deflection region. Instead a single plate made of Beryllium Copper is used for the high voltage plate and the body of the analyser acts as the ground plate.
5.2 Computer modelling
The electrostatic deflection model was produced by Roger Woodliffe at MSSL and acted as the central core for the full instrument simulator. In order that the electrostatic deflection can be calculated at any point along an ion trajectory within the model it is important to be able to determine the electric field at any point. This process is done in two parts and is described in detail by Woodliffe (1991). Firstly the region of interest is divided into a mesh and the derivative at each node replaced by a finite difference approximation derived from a Taylor series expansion. These values are then interpolated in two-dimensional planes using a cubic spline routine. Interpolation between planes is then carried out using a second bicubic spline. The data shown in this section are just the field values at discrete mesh points interpolated by the contouring graphics routine.
Figure 5-1 shows a two dimensional equipotential field plot superimposed on a mechanical drawing of the analyser. The plane shown is that which includes the focus point of detector 2. In the figure it is clear that there is some electric field within the magnetic deflection region. The dashed contours show fields of 1% to 9%
of the applied deflection potential in each mode. In order to keep the field profile similar for all three analysers the electrostatic plate is extended so that the deflection regions are not effected by the end field. Figure 5-2 shows the field equipotentials in a plane parallel to the electrostatic deflection plate and including the central rays of each collimator.
F ie ld e q u i p o t e n t i a l s fo r z =
2 7 . 5 0 0 0 m m
20 e e - 2 0 - 3 0 20 40 0 - 2 0 X axis (m m )Figure 5-1. Plot of field values from the simulation with the final analyser mechanical design superimposed. Contours show the field potential as a percentage of the applied voltage to the deflection plate. The set closest to the plate vary from 10% to 90% in 10% steps (solid line), the next set (dashed line) vary from 1% to 9%, and the final set 0.1 % to 0.9%.
Field e q u i p o t e n t i a l s fo r y = —2.5 0 8 3 3 m m
25
Del. 1 Del. 2 Del. 3
30 30 20 — 50 g B M X (Ü X 20 40 60 0 Z axis (mm )
Figure 5-2. Plot of equipotentials in the x-z plane for y = -2.508mm. Contours as in Figure 5-1.
5.3 Initial results
Initial deflection results were carried out using a single hole collimator for each of the three detectors in an analyser unit. Because of uncertainty in the beam properties at different energies the energy was kept constant and the deflection voltage was changed. A typical set of results is shown in Figure 5-3. These results are for detector 2 in a left hand analyser operating in mode 2, i.e. -87.7Volts pre acceleration voltage and 137.7Volts deflection voltage. As can be seen the results compare favourably over much of the range although at the lower energies the simulator is predicting larger deflections than actually occur. These differences are probably due to differences in the electrostatic field profile. In the experiment the magnets were removed so that the electrostatic deflection could be studied independently. This means that the field profile will be slightly different than that shown in Figure 5-1. Although the magnetic field is set to zero in the simulator the
physical outline of the magnets still remains and influences the field structure within the model.
The error bar shown on the experimental points is the resolution of the imaging detector, approximately 140pm. Ideally it should be possible to determine the peak position to a higher accuracy than this but practical uncertainties about the local non-linearity of the MCP/anode system prevent this. These peak positions are measured from a theoretical straight through position for the collimator. Any number of effects could cause a shift in the relative peak position for these experimental results, e.g. non-normal pointing of the collimator channel, positional accuracy of the deflection plate within the analyser body. Essentially though the simulator is showing deflection results similar enough to the experimental results to have confidence in its use as a predictive tool.
D e t e c t o r 2, Mode 1 E / S d e f l e c t i o n
2 3
E n er gy/D efle ction Voltage
Figure 5-3. Plot of electrostatic deflection versus the ratio of ion energy to deflection voltage for experimental results (crosses) and output from the instrument simulator (dots). The simulator output is for masses 28 and 18 which generally make up over 90% of the ions exiting from the ion gun.