The optical design of the ESPaDOnS polarimeter follows
closely that of Semel's visitor stellar polarimeter,
developed at Observatoire de Paris-Meudon and
used with great success at the Anglo-Australian
Telescope. In particular, it involves Fresnel rhomb
retarders to perform a very efficient polarimetric analysis
over the whole spectral domain. In addition to being the most
achromatic retarders available, such devices do not produce
spectral ripples in high-resolution polarised spectra,
as opposed to more conventional superachromatic waveplates (like the
Halle
superachromatic
retarders for instance).
As shown on the optical layout, stellar/calibration light is collected at Cassegrain focus through a 220 micron = 1.58" circular pinhole at the centre of the mirror atop the polarimeter. Two triplets (with respective focal lengths 55 and 25 mm, working at infinite conjugate ratio and separated by an air-equivalent optical path of 80 mm) ensure that fibres are fed at a beam aperture of f/3.6, that the output beams at fibre level are telecentric (pupil at infinity) and that the Wollaston prism work in a parallel beam.
The two half-wave Fresnel rhombs can rotate about the optical axis and be oriented to predefined azimuths (regularly spaced at 22.5 deg intervals) with respect to the optic axis of the Wollaston prism. The upper half-wave Fresnel rhomb can also rotate at a constant rate (at a frequency of about 1 Hz) in order to average out linear polarisation when performing circular polarisation analysis of highly linearly polarised sources. The orientation/rotation of half-wave Fresnel rhombs is remote controlled. The quarter-wave rhomb is fixed.
The two beams produced by the Wollaston prism are imaged onto the two
fibres of the fibre link that feature a separation
of 110 microns (referred to as object fibres). The corresponding
deviation that the Wollaston produces for each beam (with
respect to the system optic axis) is equal to 0.126 deg
(at 500 nm).
In non polarimetric mode, the Wollaston prism is replaced by a wedged plate with an output surface tilted to produce a beam deviation equal (both in amount and direction) to that of the Wollaston prism. Note that only one of the two object fibres is used (light entering the second object fibre can be blocked further on the optical path, e.g. by the dekker at the output of the image slicer). The Wollaston prism and wedged plate are both mounted in a remote controlled slide.
In this mode, the second circular aperture in the mirror atop the polarimeter (located 1.1 mm = 7.9" south of the central one, with a diameter of about 300 microns) collects photons from the sky background and redirects them to the third fibre of the fibre link. Note that sky light is also collected through this fibre in polarimetric mode, and can be blocked on the dekker at the output of the image slicer.
The main technical specifications associated to the polarimeter are:
the mechanical structure of the polarimeter is rigid
enough to ensure that its lower and upper surfaces do not move laterally relative to each
other by more than 2 microns for all telescope positions (and thus
that the mirror pinhole and optical fibres always remain in perfect optical conjugation);
the azimuth of rotating half-wave Fresnel rhombs is accurate
to within 3';
the retardation of half-wave and quarter-wave Fresnel rhombs is
nominal to within 30' throughout the whole spectral domain;
the input and output surfaces of the Fresnel rhombs and Wollaston prism are
parallel to within 5";
the beam deviations of the Wollaston prism / wedged plate
are nominal to within 1%;
the birefringence induced by the first triplet is smaller
than 0.5 nm;
the spot diagrams at fibre level feature a rms diameter smaller
than 5 microns over the whole spectral range;
the total throughput of the whole polarimeter is higher than
89% (i.e. higher than 99% per air/glass interface).