Polarimetry with UFTI
This document describes the use of IRPOL2 with the facility 1-2.5 micron imager, UFTI. IRPOL2 was designed and built by the University of Hertfordshire for the UKIRT: more details of the design characteristics of the polarimetry module and its corresponding optics can be accessed from the IRPOL2 Homepage.
NOTE: Only brief notes are given below; for a more complete discussion of imaging polarimetry see the UIST Imaging Polarimetry web page.
The UKIRT and University of Hertfordshire would appreciate an acknowledgement in any publication which contains data obtained using IRPOL2.
UFTI Polarimetry: Introduction
IRPOL comprises a half-wave retarder (the waveplate), a focal-plane mask and (internal to UFTI) a Wollaston prism. The focal plane mask and half-wave retarder will be installed by the Telescope System Specialist, or T.S.S., at the beginning of the night. The prism splits the incoming radiation into orthogonally-polarised e- and o-beams. With extended sources (greater than about 20 arcsec in declination), these beams would overlap on the array. Hence the need for the mask. IRPOL2 projects the e- and o-beams onto parallel strips on the array (in the top half of the array). The two strips in the bottom half of the array are used by the polarimetry reduction recipes (written by Malcolm Currie) as blank sky, for subsequent sky subtraction. An example exposure of a standard star, showing the e and o “images”, is shown below. The usable area of each beam on the array (i.e. the size of each horizontal strip) measures approximately 90 x 18 arcsec
Note also that for the Data Reduction Recipes to work your source MUST be located in the northern/top half of the array (rather than the southern half).
UFTI and UIST imaging polarimetry observations are set-up and acquired in essentially the same way. The main differences between instruments are: the orientation of the long axis of the mask aperture (UIST’s imaging polarimetry aperture is elongated N-S), the pixel size (0.09 arcsec for UFTI; 0.12 arcsec for UIST) and, as a consequence of both of these, the field-of-view on sky (90 x 18 arcsec for UFTI; 18 x 120 arcsec for UIST). L-band imaging and spectro-polarimetry are also available with UIST.
For detailed notes on how to set-up an imaging polarimetry observation with either UFTI or UIST using the ukirt-ot, please see the UIST Imaging Polarimetry web page. An example UFTI imaging polarimetry MSB is shown below.
E- and O- beam separations
The alpha-BBO prism in UFTI acts as the polarimetric analyser, splitting the incoming radiation into orthogonally polarised e- and o-beams. The divergence of the beams is dependent on the refractive index of the prism material and is therefore wavelength dependent.
The beam separations given in the table below are measured for bright point sources and are given in arcseconds; the UFTI pixel scale is 0.091″. The offsets are of the top (e-beam) with respect to the lower (o-beam) – note also that in RA the offset increases to the left (East).
|BAND||E-W offset (arcsec)||N-S offset (arcsec)|
Note also that the dispersive properties of the prism will result in slight elongation of the images in the dispersion direction, i.e. in a North-South direction. This is only noticeable under conditions of half-arcsecond seeing or better.
Sky or “Dome” flats
A full discussion on flat-fielding is also given in the UIST Imaging Polarimetry guide. Below we list typical sky counts obtained with UFTI.
Background Sky Counts
(2am – Grey time)
|Filter||Integ. Time||Mean counts|
Ideally, a few thousand counts should be sought in flat-field exposures, since noisey flats from only a few hundred counts may limit your ability to get repeatable numbers on un-polarised or low-polarisation (few tenths of a percent) sources.
Recent test exposures of the back of the primary mirror covers (i.e. with the covers closed) illuminated with two halogen lamps (2x 150 Watt bulbs) positioned on the platform above the south column, with the telescope in its “park” position (elevation -20 degrees) and the dome closed and in darkness, yielded very promising results. 3000-4000 counts were measured in 4-second Z, J and K-band images, while 5000-6000 counts were measured in 4-second H-band images (note that 4 seconds is the minimum full-array exposure time with UFTI). Exposures through the mask and waveplate, at each of the WP angles, with the halogen lamps on then off, should be acquired (to do this, run the “IRPOL Imaging dome flats – UFTI” observation in the calibrations menu TWICE). It is recommended that these “dome flats” should be secured at the start of a night when imaging polarimetry is planned.
Which filter should I use, K98 or K’
In the K-band, the background with the warm waveplate in the beam is considerably higher with the K98 filter than it is with the K-prime filter. Although both filters have the same bandpass (0.35 microns), the transmission peak/centre of the latter is blue-shifted by almost a tenth of a micron, resulting in a reduction in the background by a factor of ~2. Consequently, if one ignores source colour, then one would expect an increase in S/N of the order of sqrt(2) if using the shorter-wavelength filter. The K-prime filter in UFTI might therefore be worth considering, particularly for imaging-polarimetry of fainter sources, although note that this filter is less-well characterised than the MKO standard K98 filter. Those requiring accurate (absolute) photometry should use the K98 filter.
Polarimetry with UFTI:
Instrumental Polarisation and Efficiency
Instrumental polarisation and efficiency
Instrumental polarisation has been measured at I,J,H and K with UFTI. Measurements are tabulated here. In all cases values are low, although some variation (on the 0.2-0.3% level) is often noted. This is thought to be due to non-linearity with these bright targets, as well as inaccuracies in flat-fielding. Note also that, because the present list of unpolarised standards are too bright for UFTI, high-proper motion CMC stars, which are assumed to be unpolarised, are usually observed.
The efficiency of IRPOL2 was also measured by observing bright stars through a Glan prism (which induces 100% polarisation). A much lower signal-to-noise ratio is needed for these measurements and consequently the position angles measured (and polarisation values) are relatively stable. The system is, in any case, almost 100% efficient at all wavebands.
|Source||Band||Polarisation (%)||Pos. Angle (degrees)|
Position Angle Calibration
To calibrate measurements of polarisation position angle with UFTI + IRPOL, deep observations should also be obtained of a polarised standard. These may then be compared with the values obtained periodically by UKIRT staff and visiting observers, which we list here. These data suggest that a correction of approximately -9 degrees should be applied to p.a. measurements.
Earlier measurements with IRCAM3 indicate the need for a correction of -6.7(+/- 0.1) degrees. Because the IRPOL waveplate is above the camera, the orientation of the polarisation angle should be the same for all instruments. Any differences may be due to the fact that the e-/o- beam dispersion axis for the prism in each instrument is not perfectly aligned with the columns on the array.
Note also that at JHK one would not expect a large angle change with wavelength because the JHK waveplate is achromatic.
IMPORTANT: The DR will apply a corrections of -9 degrees to observed angles automatically. See the output from the DR for full details. Additional corrections to the angle can be made in the Polarimetry toolbox in Gaia.
Latency and its effects on Polarimetry
Q. Does latency/persistence on the array affect polarisation measurements? During commissioning, two polarised standards were observed, each with a 3-point jitter pattern and then a 5-point jitter pattern. All observations were J-band. The aim was to check whether latency or “persistence” (i.e. residual signal left on the array after observing a bright source; usually at the 0.3-0.7% level) could affect the apparent polarisation of a source if that source revisits the same location on the array too frequently (e.g. as might be the case if one continuously repeated a 3-point east-west jitter pattern). The results given in the table below suggest that this may be the case (though note that the observations were obtained in less than perfect [variable cirrus] conditions). It is therefore recommended that when observing bright sources, a 5-point jitter pattern and/or small offsets of the telescope (0,0) position be applied between consecutive jitter patterns.
3-point versus 5-point jitters
|RCrA 10 (3pt mosaic)||0.370.10^7||-22272||34049||1.10%||75 degrees|
|RCrA 10 (5pt mosaic)||0.395.10^7||1115||57509||1.46%||44 degrees|
|BD+57 2615 (3pt mosaic)||0.130.10^7||1017||-12873||0.99%||137 degrees|
|BD+57 2515 (3pt mosaic)||0.136.10^7||4875||10613||0.86%||147 degrees|
errors are typically 10%.
Array persistence is likely to be less of a problem for diffuse, more extended sources. Indeed, the well-known young stellar object GSS 30 was observed with a 3-point jitter pattern. As is shown below, the polarisation pattern associated with this bipolar nebula compares very well with published results (Chrysostomou et al. 1996, MNRAS, 278, 449).