Michelle Polarimetry

Introduction

Imaging polarimetry for Michelle was commissioned in December 2005/January 2006. This observation mode has been offered for science usage starting in semester 2006B. This page gives some preliminary guidelines on carrying out Michelle polarimetry observations. These guidelines may change as we get more experience with such observations.

Spectral polarimetry is expected to be commissioned in 2007A.

The Observing Process

Imaging polarimetry is carried out similarly to regular imaging, in chop-nod mode. The difference for polarimetric observations is that a half-wave plate is introduced into the optical path just after the light enters the instrument. The wave-plate can be rotated to four positions (wave-plate angles of 0 degrees, 22.5 degrees, 45 degrees, and 67.5 degrees). The different wave-plate positions modulate any polarized radiation from the target object.

The observations are taken in such a way that two cycles of the wave-plate positions are observed for each nod position. The wave-plate is cycled through the positions in the following order: 0, 45, 45, 0, 22.5, 67.5, 67.5, 22.5 degrees. This order of observation has the same effect as ABBA nodding--it tends to cancel out linear drifts in the sky emission with time or position. The differences between the 0 degree and 45 degree positions and the 22.5 degree and 67.5 degree positions select two orthogonal polarization directions. Only linear polarization can be measured by Michelle.

To keep the total time per nod position short, only about 5 to 6 seconds is spent at each wave-plate position. The observations are taken with regular chopping, the amplitude and direction of which are chosen by the user in the OT as with regular imaging observations. The maximum chop amplitude is 15 arc-seconds so objects that are larger than this in both dimensions as seen on the sky cannot be properly imaged with Michelle.

Due to the overhead of moving the wave-plate the efficiency of the observations is lower than for regular imaging. In terms of the total time on-source the efficiency is about 0.075, compared to about 0.25 for regular imaging. However this time on source is divided between the four wave-plate positions. Thus to obtain a total on-source integration time of 30 seconds per wave-plate position the time needed is about 2 minutes divided by 0.075 or close to 27 minutes. We are investigating ways to increase the efficiency. Starting in 2007B it is the total on-source time that should be used when defining observations in the Observing Tool or when calculating expected S/N using the Michelle integration time calculator. (Previously the time per wave-plate position was used, which is a factor of 4 different than the on-source time.)

There is a separate wave-plate for Q-band polarimetry. At this time only the Qa 18.1um filter is available for imaging polarimetry in Q-band.

Required Weather Conditions

The commissioning observations have shown that stable conditions are needed to get accurate polarimetry observations. Stability is required in both the seeing and the atmospheric transparency in the filter that is being used. Since poor seeing in the mid-infrared is usually associated with variability of the seeing, it makes no sense to carry out polarimetry observations in seeing worse than IQ=70%. The atmospheric transparency in the mid-infrared is much more difficult to judge, or to relate to the cloud cover. However it seems likely that CC=50% (i.e. clear conditions) is going to be required to obtain good quality polarimetry data. Therefore PIs should request observations in conditions of CC=50% and IQ=70% or better, and are discouraged from requesting observations in conditions worse than this.

The water vapour conditions required are strongly dependent on the filter that is being used. For the N' and Si-5 11.6um filters, which cover the best part of the N-band window, variations in water vapour column should have very little effect. The Si-2 8.8um filter is a bit more sensitive to water vapour. The Si-1 7.9um filter and the Si-6 12.5um filter are quite sensitive to the water vapour column. The Si-3 9.7um and Si-4 10.3um filters are not as sensitive to this but are most sensitive to the ozone band, which varies in an irregular manner. Finally, the Qa 18.1um filter is more sensitive to water vapour column than most of the N-band filters.

From these considerations it is suggested that PIs request better water vapour conditions for the filters that are most sensitive to it -- Si-1, Qa, and Si-6. They need WV=50% conditions. The other filters can be used in poorer WV conditions, and the N' and Si-5 filters can be used under most WV conditions that are seen at Mauna Kea. When the WV is very high it is likely the CC conditions will be too poor for polarimetry anyway, so WV=Any can be requested for the N' and Si-5 filters. Once we have more experience with imaging polarimetry more specific WV ranges that can be used for observations in different filters may be given.

As for most observations in the mid-infrared, the sky brightness (which refers to the lunar phase) should be set to SB=Any.

Other Considerations

As well as needing stable conditions, polarimetry observations require good signal-to-noise ratio for accurate results to be obtained. The requirement is that one needs S/N of 70.5 in the reduced Stokes I image (i.e. in the total intensity image obtained by combining the images from the four wave-plate positions) to have an accuracy of +/- 1% in measured polarization. Obtaining this accuracy is made more difficult because the half-wave plates have a through-put of about 0.3, so for a given total time on-source a polarimetry observation will have a S/N value about 3 times lower than that which would be obtained in regular imaging mode with the same filter and the same total time on-source. Therefore to get the same S/N in polarimetry mode as in regular imaging mode takes of order ten times longer.

This consideration combined with the need for stable conditions means that only relatively bright mid-infrared targets can be observed in the polarimetry mode. Roughly speaking a target with a flux density of 2.5 Jy in the Si-5 11.6um filter if observed for 2 hours should yield an accuracy of 1 percent in the polarization (per pixel). With binning of the pixels the accuracy of the polarization measurement can be increased by a factor of a few, at the cost of a loss of spatial information. Sources that are significantly fainter than 1 Jy most likely cannot be observed in polarimetry mode, and indeed it would be better if any sources to be observed are somewhat brighter than this if a low level of polarization is expected. The ITC gives S/N values for an aperture, not per pixel; the ratio of the overall S/N to the per pixel S/N is about equal to the square-root of the number of pixels in the aperture.

The situation is somewhat better for the N' filter, and somewhat worse for other of the N-band filters. Note that the Qa filter sensitivity is rather worse than those of the N-band filters and the on-source efficiency is lower. As a result, targets for which polarimetry in the Qa filter are requested should be significantly brighter--at least 10 times brighter than the fluxes given in the preceding paragraph.

The above estimates are for targets which are either point-like or small as seen on the sky by Michelle. It is difficult to give sensitivity estimates for extended targets at this time.

The Gemini ITC can be used to estimate the time required to reach a good enough S/N to detect polarization. These estimates are based on a limited set of commissioning observations and should be viewed with some caution. It is recommended that PIs allow some extra leeway in their time estimates until the Michelle ITC has been tested against the results of a larger set of observations.

The commissioning observations were taken in local winter, and under average conditions. Conditions are likely to be better on the average in the summer, and whatever the season the best nights will produce better results than we had in the commissioning.

Properties of the Instrument

The Michelle polarimetry observations are made in single-beam polarimetry mode.

The following estimates have been derived from the commissioning data and from the expected properties of the wave-plate.

• the transmission of the N-band wave-plate is about 0.3; the transmission of the Q-band wave-plate is slightly lower, about 0.25
• the N-band wave-plate has maximum efficiency, assumed to be 1, at 10.5 microns; it decreases by about 10 percent at the ends of the N-band window
• the Q-band wave-plate efficiency is near 1 for the Qa filter
• the wire grid efficiency is high, close to 100 percent
• the instrumental polarization signature is at most 0.2 to 0.3 percent
• the instrument does not introduce any measurable rotation in the polarization vectors (i.e. ANGROT as defined in the POLPACK software is zero within the uncertainties of the measurements)

The transmission values given in the first item of the above list are used in the Michelle ITC. They are conservative estimates.

Baseline Calibrations

The only baseline calibration that will be taken for each observation is the observation of a flux standard, as in normal imaging observations, but through the polarization optics. The normal set of photometric standard stars, mostly taken from the series of papers by Martin Cohen and his co-workers, will be used for this purpose. As these are all normal stars, mostly K-type giant stars, without any detected circumstellar dust, they are expected to have no intrinsic polarization. Experience from the commissioning showed that if conditions are variable such stars appeared to have a polarization signal of a few percent because the different wave-plate images varied in signal and in the stellar PSF. Thus the observation of a standard star will also allow a direct assessment of whether conditions are stable enough for the polarimetry to be accurate.

Additional calibration observations (i.e., an astrometric standard, a PSF standard, or the observation of a previously known polarized object) can be requested but they will be charged to the program as additional calibrations.

We will take occasional observations of known polarized targets to keep track of the instrumental characteristics, but these observations will not be done very often. When done these observations will be part of the general calibrations and will not be associated with any individual program however will be avaliable in the archive.

When selecting Cohen standards (and very bright science targets) please be aware of the "hammer effect" which, although diminished by the decreased throughput of the waveplates, may be significant for sources brighter than ~100 Jy for a point source.

Spectral Polarimetry

Spectral polarimetry is not commissioned and so will not be offered at this time.

Last update 28 February 2007; Kevin Volk