order layout¶
The spectrum of any object in the field of view of the telescope will be dispersed. For each order of dispersion, a separate spectrum appears on the detector. Each order also has a unique sensitivity which depends on the wavelength. There is also some variation of all that over the face of the detector, meaning that a spectrum in the centre will be slightly differently dispersed and have a different response than a spectrum near the bottom or the top, etc., of the detector.
The response for the uv grism, which was not blazed, is different than that for the visual grism, which was. Read for example about blazed grisms in Appenzellers book.
The uv-grism¶
The design goal for the uv-grism was to optimize the response in the 2000-3400A region. The design was made for the centre of the detector, since that is where the main target would be located during operations.
The zeroth order in the final design is not a point, but dispersed. While most of the optical light falls on about three pixels, the zeroth order of the uv-grism has an extended tail with a dispersion propotional to the inverse of the wavelength. For most sources the uv tail is hidden by the background noise. However, uv-bright sources, like White Dwarfs, Gamma Ray Bursts, and such show an extended zeroth order tail, often to appear disconnected from the visual part due to the reduced flux around the 2200A region due to the interstellar absorption.
The zeroth order main peak is separated from both the first and minus-first orders by about 300 pixels in the uv-grism. The minus-first order has a terrible point spread function, and therefore is not useful for science.
The second and third order wholy or partially overlay the first order. The part of the first order completely free from second order contamination varies over the detector. In the bottom half of the detector it is fair to say that the overlay starts in the first order frame of reference around 2750A, but if the source is not very blue the limit is more like 2950A. For a red source, it the wavelength at which the second order contamination rises above the background noise can be a a larger wavelength, of course.
In the upper half of the detector the uv-part of the second order curves away from the first order, and also shows some lateral displacement. There, even for blue sources the second order will not contaminate the first order until higher wavelengths are reached. The best result is for an offset of about +3.5 arcmin in X, and Y on the detector, and the region up to about 4500A shows second and first order emission not overlapping.
The second order sensitivity peaks further in the uv than that of the first order which peaks around 2700A. With also the width of the spectral track of the first order being larger than the first, i.e., 4.5 versus 3.1 pixels, the signal per pixel for the second order is generally much lower, with the exception of bright spectral lines.
The third order overlap starts close to 4900A of the first order spectrum. If the source has a blue spectrum, there is sometimes excess emission causing a jump in the flux. However, for most sources the third order flux is small being spread over about 6 pixels wide, and in the top part of the detector the offset is usually large enough to make it irrelevant for the forst order spectrum.
The visual grism¶
In the visual grism, the second order is much fainter than the firs tone since the grism was blazed. The first second, third, etc. orders all fall on one line, thus partially overlap. Like the uv grism, the zeroth order lies apart. The dispersion is about half that of the uv-grism, 6A/pixel in the visible grism versus 3A/pixel in the uv-grism. The response below 3000A (3000-2700A) is not very good, so it covers a slighly bluer region than ground based spectroscopy achieves.