Canada-France-Hawaii Telescope

User's Manual for Scanning Fabry-Perot Spectrography

R. Arsenault

Version 3.01

July, 1997

1 Instrument Purpose & Performance

IMPORTANT NOTICE: You are consulting a CFHT manual presumably because you will have an observing run in the near future. Have you forwarded the Observation Preparation Form for your run to your support astronomer ? Please do if this applies to you. This form is very important to schedule the activities of the technical staff in preparation for your run. Include as much details as you think necessary. Unpleasant surprises at the telescope may be avoided, and your run may turn out more succesfull. In order to make this task easier, a number of HTML forms are available on the CFHT Web site. Look for the page " Tools and Information for Observers". It also contains information you might need to prepare your observations.

1.1 What is a Fabry-Perot & Why Use It ?

Fabry-Perot spectroscopy offers moderate resolution (~5000 to 10000) 2-D spectroscopy for the observation of various astronomical sources. The field of view varies between 1 and 10 arcmin depending on the intrumental configuration, and the spectral resolution depends essentially on the Fabry-Perot etalon inserted in the instrument. The spectral PSF is oversampled so that there is no lost of resolution resulting from a coarser sampling (except maybe at the edge of the field with a large gap etalon). The sampling is given in the table below and is adapted to the image quality expected from each instrumental configuration.

Fabry-Perot spectroscopy has been used particularly on extended objects like galaxies and nebulae. It is particularly efficient for emission lines, to obtain velocity or velocity dispersion fields. At the time of this writing (February 2005) the AOB/FP configuration has been fully tested on the telescope and has been used once the sky (but not with a CFHT etalon). Fabry-Perot spectroscopy is known to offer a very high product resolution-aperture (better than slit spectroscopy). That is why it is well adapted to faint surface brightness targets (HII regions, galaxies etc.). However, one might keep in mind that for high spatial resolution work with OSIS/FP or AOB/FP, each pixel sees a very small solid angle on the sky and observation will be flux limited. Success will be more likely on objects with relatively high surface brightness.

The data take the form of a data cube. One must scan a certain number of channels to obtain interferograms of the object. An interferogram is an image of the object modulated by the inteference pattern of the Fabry-Perot. The ring pattern shifts for each interferogram and covers the whole field if an appropriate number of channels has been selected. If the data cube is identified by 3 axis, alpha, delta and Z (the interferogram number), a cut along the Z axis, for a given pixel (or given position on the object) represents the line profile of the object convolved by the instrumental profile. The same is true of all position on the object, only the zero point of each line profile is shifted. This shift depends on the radius to the ring pattern center, and the correction for this shift is carried out with the help of the phase map, obtained from a calibration line. Once corrected one has a data cube composed of monochromatic images.

The previously described instrumental configurations are essentially used for Fabry-Perot interferometry. They are not well adapted to monochromatic imagery. For such program MOS or SIS in their original configuration are more appropriate. The FP etalon is inserted manually in the FP module or in the focal enlarger for the AOB and carrying monochromatic imagery would imply removing the etalon from the beam which is possible, but requires an operator to go to the instrument and performs the operation. This operation is not realistic for the AOB/FP since it requires removal of the detector.

Note also that MOS and OSIS, in their Fabry-Perot configuration or not, are considered as two separate instruments and therefore a switch between the 2 cannot be performed during the night. If one wishes to observed with both, it must be pointed out in the observing proposal, and this will be scheduled as 2 different runs, even if this is the same scientific project. Beware of CFHT's new policy (May 1995) that an instrument will not be mounted on the telescope for less than 8 consecutive nights, and it will not be set to a new configuration for less than 4 consecutive nights.

1.2 The Instruments: MOS, OSIS and AOB

A perspective drawing of MOS/SIS is shown here . One can clearly see the central octagon and the 2 serrurier struts for MOS and SIS. A simplified schematic of the instrument is shown . It attaches to the cassegrain acquisition/guiding/rotator unit ("bonnette") at the f/8 focus of the CFH 3.6m telescope. The image scale at the input focus of the instrument is 139 micron/arcsec. The wavelength and flat field calibration illumination system is located above the spectrograph in one of the bonnette ports and uses the back of the central viewing mirror to direct the light into the instrument when required. A large stiff octagonal structure with 4 openings or side ports attaches to the bonnette to support the two paths or trains. A central mirror slides in the octagon and allows the f/8 beam from the telescope to be directed into the MOS train, the OSIS train or to go directly through to an instrument mounted below. On the MOS side, the light passes through a collimator and f/2.8 camera with a 10x10 arcmin field when it is in imaging mode (note that the field size depend on the size of the filter used in FP mode). The optical train for OSIS is similar, except that the camera is f/8, offering a 3x3 arcmin field, and the mirror which directs the telescope beam into the spectrograph is mounted on piezo stacks which allow a fast tip-tilt correction using signals derived from a reference star in the field.

Both collimators (MOS and OSIS) produces a 46 mm pupil image approximately between the mirrors of the Fabry-Perot etalon (160 mm behind the last optical element of the collimator).

The image quality of MOS and OSIS is excellent. There is a low distortion of MOS field of view (0.95 % at the edge of the field) and a slight variation of image quality across the field of view. For this reason one should focus not at the center of MOS field of view but half way to the edge of the field. Image size is always smaller than 1 arcsec, and can be as good as 0.46 arcsec. No measurable image size variation can be seen across the OSIS field of view. The transmission of both optics is 72, 87, 96, and 97 % at wavelengths 3650, 4000, 5000, and 7000 A respectively (optics only without Fabry-Perot or filter).

The Fabry-Perot module (shown here as viewed from under the instrument) is a box that replaces the Grism and Filter box of MOS or OSIS. The installation of this module requires some time and therefore takes place during daytime . For this reason, one cannot change from MOS/FP to OSIS/FP during the night or even during a same run. If you plan to use both MOS/FP and OSIS/FP for the same program, you must specify this in the observing proposal. The Fabry-Perot etalon is attached to a mounting plate, that can be swung in the collimated beam . For this reason, a change during your run from scanning FP to imagery (without the etalon in the beam) is feasible but not advisable. This operation is not remotely controlled, and can change the instrumental configuration (bad matching of the phase map).

This setup also implies that the filters have to be installed in the mask slide, which then becomes the filter slide. The slide being located at the entrance focal plane, the filter are in focus (should be carefully cleaned before insertion of the slide) and limit the field of view if smaller than 80 mm square. This is usually the case for the 3 inch circular interference filters available for Fabry-Perot work. There are 4 positions of the slide, 3 filters plus an opened field. As a consequence, the telescope focus must be changed when filters are inserted or changed (if they are of different thicknesses).

The situation is similar for OSIS, although the field being much smaller (~40 mm), the field of view is not limited by the filter. Although OSIS mask slide has 6 positions, the SIS filter slide has only 4 (including the open position). The reason being that the normally used 2 inch filters are larger than the OSIS masks. At the beginning of 1996, CFHT has received a set of interference filters adequate for Halpha observation (rest and redshifted wavelength). If you plan using OSIS/FP at other wavelengths, make sure you have the required interference filters for your project ( 2 inch square, 7 mm maximum thickness; 2 inch circular can also be adapted).

For MOS and OSIS, there is the possibility to insert one filter (2 inch square) in front of the Fabry-Perot etalon in the collimated beam. This is however a manual operation. Three cells are available to mount filters in advance. Furthermore, the cell allows installation of filters up to 1 cm thick. If you plan to bring thick filters or need to use them at high inclinations, it would be appropriate to mount them in the collimated beam.

Note that when inserting a filter at the entrance focal plane, the telescope focus has to be changed to compensate for the focus change. This preserves the beam collimation onto the Fabry-Perot etalon. However, things are different in OSIS since the telescope focus must be kept constant to allow fast guiding (the guide star probe is located just before the mask slide or here the filter slide). Therefore, the procedure is to change the camera focus when using filters of different thicknesses. If the Fabry-Perot etalon used has a relatively low finesse, it is acceptable to have a small decollimation on the Fabry-Perot. Check with your support astronomer (tests have been carried out with an etalon of finesse 30, and the change in resolution is less than 4 %).

Consult the CFHT filter list here CFHT filters@ through this hyperlink. The WEB filter list (available through the previous link) is the official filter list of CFHT (as of Aug. '98). Please disregard any printed list, they are no more current. Remember that interference filters bandwidth are strongly affected by temperature changes. Here are a few rules:

Beam collimation: The central wavelength of a narrow band interference filter moves to shorter (bluer) wavelengths if the beam is faster. For beams slower than f/11 the effect is negligeable.

Incidence angle: The central wavelength of a narrow band interference filter is shifted to the blue as the filter is inclined. Inclinations of more than 25 degrees will widen the bandwidth and decrease the transmission; however, in practice inclinations are less than 5 degrees. Shift to the blue are 0.4A for 1 deg., 1.8A for 2 deg., 4A for 3 deg., 7A for 4 deg., and 11A for 5 deg.

Temperature: As a general rule, the bandwidth shift to the blue at cooler temperature. The rate of change is 0.18 A/deg. C (Angstrom per degree Celcius)

The following figure shows the scanning Fabry-Perot setup in the AOB focal enlarger . This image shows the enlarger and FOCAM. The top is bolted onto the AOB  (Adaptive Optics Bonnette) and the enlarger optics double the image scale from f/20 to f/40. All the optics is located in a tube screwed to the top of the enlarger structure. The Fabry-Perot etalon is mounted on the bottom surface of the enlarger (therefore in the f/40 beam), preceding FOCAM in the optical path. Focam contains the shutter and the interference filters. The CCD is mounted onto focam at the very bottom of the illustration (detector not shown on the image).

The Fabry-Perot etalon is located in a converging f/40 beam. This is not ideal, however in certain cases the finesse is not degraded by the decollimation. This implies that only small gap etalons can be used in this configuration, like the CFHT etalon #3. Moreover, the optics is not telecentric. This can introduce a severe radial degradation in the resolution (for example, with the Rutgers FP etalon which has a gap of 198 microns and a finesse of 25, only the central 10 arcsec of the field is useful to gather astronomical data). With this configuration, calibrations are very tricky and time consuming. For these reasons, AOB/FP cannot be offered to the general users unless they are very experimented in Fabry-Perot spectroscopy. Feasibility of the proposal must first be reviewed by the CFHT staff. Please, contact the FP instrument scientist if you are planning to apply for telescope time with AOB/FP!

1.3 The Fabry-Perot Etalons

Two etalons built by Queensgate Instruments of London are available at CFHT. Their working aperture is 50 mm and they are placed very close to the 46 mm exit pupil. All are piezo-scanned and servo-controlled; accurate capacitance measurements between the plates are used to detect deviations from paralelism or required spacings. For reliable operation they must be turned on during the entire run and continuously flushed with dry nitrogen. It is also not advisable to change the etalon during the night.

The etalons characteristics are listed in the following table. There are 2 finesses listed for each etalon. The highest is the reflective finesse. The Effective finesse that will be measured at the telescope is always lower due to a variety of factors like: non-perfect parallelism of the etalon, aging coatings, imperfect optics etc.

Etalon  Usable    Interference   Reflective  Effective  Scanning  Interfringe
       Wavelength    Order        Finesse     Finesse   Constant   A    Km/s

CFHT#1  4700-5400  1569 @ Hbeta      15         --        2.72    3.6    191
        6200-7500  1162 @ Halpha     20         13         "      5.6    258
CFHT#3  5600-7500    72 @ Halpha     --         62 @6400A 2.44    91     4170

These following plots show examples of the etalon CFHT#3 Finesse versus wavelength and of the transmission of the etalon versus wavelength (we have no data for wavelength shorter than 570 nm).

The parameters of a scan or setting of the etalon spacing are entered in the Pegasus session running on the HP acquisition computer. The values are passed to the Fabry-Perot controller CS100 (Queensgate Instruments) using an serial to parallel interface module in the MOS/OSIS controller. Our CS100 controller is relatively old but has regularly been returned to Queensgate for refurbishing or check out. It has in the past controlled a variety of Queensgate etalons from various institutes and observers are welcomed to bring their etalons if they wish. Connectors have changed somewhat over the years so it is better to contact your support astronomer if you think connectivity can be an issue. Controller and etalons at CFH were acquired in the early 80's and use the old Hugues type, high voltage rectangular connectors. We also have a jumper cable allowing to connect our older controller to recently built etalons. This picture shows a side view of one of our Fabry-Perot etalon and the connector configuration. This illustration shows a top view of one Queensgate Instrument etalon. The following picture shows the CS100, the etalon controller.

For some applications one might want to have the Fabry-Perot ring pattern off-centered. This is possible by mounting the etalons a bevelled ring. This also helps to prevent internal reflections, and is usually preferred for partial scanning. All together 3 mounting plates are available at CFHT: a 0 deg. (rings centered), 2.4 deg. (rings at edge of field for OSIS) and a 7 deg. (MOS).

1.4 The Detectors

This hyperlink@ lists the availlable detectors at CFHT. Observing with a Fabry-Perot is very demanding on the detector. The normal observing mode calls for numerous short exposures, which implies: short read-out time to maximise time on science target, and low read-out noise to maximise S/N ratio per exposure. Obviously, a high quantum efficiency also increases the S/N ratio per exposure. Large format is less critical since at these low flux the preference goes to large angular size pixel i.e. ~0.5", even to the detriment of PSF sampling. At the time of this writing, a 2K detector at CFH reads in approximately 150-170 sec. Read-out time decreases (but not linearly) if the detector is binned or a smaller raster is used.

The present set-up (CCD controller GENIII) allows a read-out of a 2Kx2K CCD in 160 seconds. This represents an upper-limit to the dead-time between channels because with a 3 inch circular filter a raster of 1500x1500 pixels is sufficient to cover the whole field.

There are a few high level features of the Detector-Controller that could be very useful for Scanning Fabry-Perot observations: skipper amplifier read-out and frame transfer area on the CCD. Both features go hand in hand because a transfer area would free the detector imaging area relatively fast so that the next exposure could take place. This is a necessary feature for skipper amplifier read-out because this process is relatively slow. In other words, the channel n+1 is being exposed while the channel n is being read-out. However, this feature has not yet been implemented and might not be in the near future first, because the detector group at CFHT has higher priorities to cope with, and second because such feature requires the CCD chip to be equipped with skipper amplifiers, which is not the case for any of our chips right now. However we keep this functionality in mind so that material and resources providing, it could be implemented quickly.

1.5 Limiting Magnitude and Performance

Fabry-Perot's offer a high throughput compared with slit spectrographs using grism or gratings. The price to pay for the high throughput and high spectral resolution is usually a very limited free spectral range. The total throughput of MOS/FP has been calculated using a calibrated planetary nebula (NGC 2022) to be close to 20 %. This hyperlink shows the details of the calculation. The reflectivity of the telescope mirrors have been monitored for about several years now and the results show an average 85% reflectivity. This is mainly due to our regular cleaning and aluminizing of the primary mirror. The coated MOS optics offers a very high transmission of 97% from 4500A to 1 micron. The 2 less efficient components are the interference filter and the CCD. It is extremely costly and difficult to fabricate narrow band interference filters (10 A bandwidth and less) with transmission higher than 70-75% and this is why we cannot expect too much improvement on this side. However the CCD quantum efficiency could greatly be improved by using a thinned device. The STIS2 device is of this type and when used with MOS/FP boosts the throughput by a factor of 2. The Fabry-Perot maximum transmission depends largely on the semi-reflecting mirrors absorption and reflectivity. The etalon CFHT#3 has a maximum transmission of 96% at Halpha for a reflectivity of the coatings of 93.3% and an absorption of 0.15% resulting in a reflecting finesse of 64. However, the previous etalon CFHT#2 had the corresponding figures: maximum transmission 82% at Hbeta, coatings reflectivity of 84%, absorption of 1.5% and reflective finesse of 25.4. These figures are not available for the etalon CFHT#1.

The previous figures imply that one can obtain a Signal to Noise ratio of 3.5 in 600 seconds on the boundary of a typical HII region (assumed to be 10E-15.6 erg/cm**2/sec/arcsec**2). This is for a single interferogram, a CCD QE of 0.45, and a pixel size of 0.3 arcsec. A complete scan in such conditions would take 6 hours for 30 channels, assuming 2 min. of read-out time per interferogram. This is assuming essentially no sky background. This may not be accurate if the pre-monochromator used accept a sky line. However, this line will also produce rings which will be easy to subtract from the data if not coincident with the observed line.

Pixel binning can be used to get a better sensitivity on truly extended sources.

1.6 Emission Lines versus Absorption Lines

In the past scanning Fabry-Perot observations have been carried out mainly on emission line objects. The reason being that it is much easier to isolate an emission line than it is to isolate a continuum. Therefore, absorption lines observation with Fabry-Perot have been far less popular. However, if the pre-monochromator (the filter) used is of high quality (4 cavities or more) and if it is well matched to the absorption line observed, this technique should work. A good match means that the interference filter must be centered on the absorption line and the FWHM of the filter bandpass must match very well or be smaller than the Fabry-Perot free spectral range. Pryor & collaborators have had good success using such a technique at CFHT (see CFHT Information Bulletin No. 32 p. 16).