The following components comprise the ISIS/FOS polarization system, in the order the light traverses them (Fig. 3):
NB The quarterwave and halfwave plates can be interchanged, to optimise a particular application; ISIS needs to come off the telescope for the interchange operation. With the present plates, halfwave last gives the best slit view and largest field for linear polarimetry, hence is recommended for general use.
(i) calcite slab, 2 beams, 330-1100nm.
(actually, a Savart plate, which equalizes focus for both
polarizations and reduces polarization anomalies within ISIS).
The calcite is located immediately below the slit and gives
two beams which are both 100 % polarized, but orthogonally;
the o- and e-beam.
The relative intensity of these beams depends on the polarization vector
(size and orientation) of the incoming beam.
(ii) Polaroid (HNP'B; 300-800 nm approx.), for occasions when full spatial detail is mandatory, without interruptions by the Dekker structure.
[ TIFF ]
Figure 3: ISIS polarization modulator and slit area.
[ TIFF ]
Figure 4: Theoretical performance of superachromat halfwave. Actual as-made properties of ISIS superachromat can be measured in situ.
It is instructive to regard the ISIS system as a modulation polarimeter with a double-beam analyser (the calcite plate) and a rotating halfwave plate modulator. Since it takes many seconds to read out a CCD frame, the modulation frequency in a CCD polarimeter is necessarily a small fraction of a Hz. With such slow modulation, the polarization-derived sine modulation in a single spectrum is contaminated by extinction variations, scintillation, image motion and other seeing variations. Under such circumstances one needs the second spectrum (orthogonal polarization), in which all such extraneous noise is in-phase with the first, while the effects due to the polarization of the light source are inverted. Dividing one spectrum by the other removes the extraneous noise, but introduces pixel sensitivity noise, which must be removed by special flat-fielding, or by relating 2 exposures for which the polarization is equal and opposite, while the pixel sensitivities involved are the same.
A set of 4 exposures through a beamsplitting analyser, with 4 different position angles of the wave plate (for details, see Section 3), in fact contain enough information to determine the fractional amplitude of the sine modulation (degree of polarization) and its phase (polarization angle). These 4 exposures allow calibration of the instrumental gain, and render the polarization measurement independent of sky transparency and scintillation. This is the basis of the standard method recommended in the next section. Measurements carried out with ISIS show that a precision of 0.1 % is feasible. This implies that relative pixel sensitivities remain constant to that extent over at least the time taken to complete a full observation. This unique combination of stability, simultaneous recording of many image points and re-usability of the same detector makes CCDs ideal for a great variety of astronomical polarimetry.