Swift UVOT Calibration Report

Zemax optical models for the UV-Grism study:

Second order wavelength scale for sources near the axis


Date: 2008/06/17
Version: 1.0
Prepared by: N.P.M. Kuin (MSSL/UCL)

Relevant Documents



  1. Zemax optical models for the UV-Grism Study: Zemax pixel scale factor, wavelength scale, and boresight, version 1.1, Swift UVOT Calibration Report.



Introduction


This report builds on, and extends, the previous report on the Zemax pixel scale, wavelengths scale and boresight for the nominal filter wheel observations and first order.  The same observations of WR52 and WR86 are used, with one change: the spectra of two observations were summed, after the removalof obvious artifacts.

Comparison of the observed spectra to the reference spectra.


A comparison of the observed and reference spectra was made in order to identify spectral features that were in the UVOT spectra, but not in the reference spectra, with the aim to identify candidate spectral lines in second order.  This is made more complicated, since the IUE spectra are not very reliable for wavelengths larger than 3200Å, and there is a gap from ~3300-3500 Å in the reference spectra.  Also, in the uvot spectra, the strong CIII line at 4649 Å is too bright, and coincidence loss and centroiding eat away in the spectrum around that line, so that the spectral region of ~4550-4770 Å is not useful for diagnosing second order lines, since they will be lost.  

In WR52 no good, clearly ourstanding, unidentified, spectral features are apparent, although there are possible excesses. In WR86, there are several which are apparent. (1) There seems to be excess emission around 3590, 3807, 3881, 4002, 4187, 4256 Å. In both spectra there is an unidentified feature at ~3316 Å, but there is no reference spectrum available. At ~4850 Å both spectra show a minor excess.

The zemax second order wavelength scale


The zemax model provides second order points, for the same model calculated for the first order. Therefore we can use the same anchor point as before (260nm in first order), and derive a polynomial through the calculated points. Similar to the first order, the polynomial needs to be at least of third order to remove systematic variations. The polynomial is:

 Table 1. The Zemax second order polynomial fit anchored at 260nm, first order.

a0

a1

a2

a3

λ(Å) = Σ an Dn

1613.8±5.95

1.096±0.039

(7.97±0.71).10-4

(-3.028±0.380).10-7



Using the wavelength scale, the features with excess emission found in WR86 in the  3590-4002 Å range should have corresponding features in the first order at 1987-2155 Å, which is not seen. Taking a possible uncertainty into account of several pixels will not change that conclusion. The features at 4187 Å and 4256 Åwere predicted by the second order wavelength scale to be from lines at 2234 Å and 2264 Å, but both features are about equally strong, and even if one of them is a misplaced 2297 Å feature, there is no corresponding second, equally strong line known to be some 30 Angstrom away. 

second order study

Using strong first order lines, to seach for a signature in second order.


Since the instrumental sensitivity cuts off below ~1700 Å and above ~7000 Å this is the only range to be considered. However, the second order does only extend to about 3200A on the detector for the observations under consideration, since the dispersion is about 1.5Å/pixel, about half that of the first order. The observed spectra were plotted with the zemax second order wavelength scale.  This can be useful to seach for the strongest lines expected to appear in the second order, which are the 1909, 2297, 2405, 2530, 2906, 3070 Å lines.  

Unfortunately, the 1909 Å line is predicted to form a blend with the strong 3409 Å O VI first order line, which shows asymmetry and is quite broad, but does not provided any conclusive evidence for the second order 1909 line. The 2297 second order line falls on top of the 4320 line in first order in both spectra, and cannot be identified. The 2405 Å line falls in the overexposed region arounf the first order 4649 Å CIII line, but the 2530 Å CIV line in second order is predicted to be present at ~4850 Å in first order. Indeed, in both spectra there is a small feature, which is not present in the reference spectra.  The second order 2906 Å  line is predicted to fall on top of the strong 5696 Å CIII line. The 3070 Å OIV second order line is predicted to fall in first order at ~6000 Å, but no credible spectral feature is found.    

The result of this exercise with these two spectra is thus that there is only one possible match for the 2530 Å  line, which would be consistent with the Zemax second order wavelength scale.  Unfortunately, we will have to use other objects, which have some strong emission lines in the 1900-3100 Å range, but have less spectral lines at longer wavelength causing confusion.

Assuming that the small feature is due to the 2530 Å line, and drawing a continuum, we can compare the relative strength of the second order feature to that in the first order. I find for WR86 that the peak value in second order is about 30% of the peak value in first order, and in WR52 ~ 25%. One has to remember, though, that the FWHM in second order is up to twice that in first order, so the flux does not scale with the peak value.  It is not clear what the FWHM in second order would be, as the width of the spectral lines is expected to be similar the the first order spectral resultion.

Conclusion



Other spectra

It is useful to be able to use the new dispersion relation in uvotimgrism. To that end, the distance to the anchor point from uvotgraspcorr was first determined.
The anchor point found by uvotgraspcorr (pre-release version), was determined for all four spectra, and used to determine the offset to the 260nm position in first order, which was also measured. The mean offset found is (543,-290) with a 1-sigma error in the mean of about one pixel in x and y. The anchor point, using  the on-axis zemax solution, thus fall on position (1520.5, 700.1), which is offset from the zemax zero order by about 8 pixels. Projecting the position found on the zemax zero order, the wavelength where the anchor falls is estimated to be ~575±15nm.

 
The pixel distance from the 260nm point in first order to the anchor point in the zero order, is estimated to be 615.6  pixels, with some uncertainty due to the 8 pixel offset of the derived point from the model zero order.

The dispersion relations can be derived as polynomials with respect to the anchor point, by calculating a fit to the zemax points starting at an interpolated position in zero order at 575nm.

The coefficients found have been put in table 2.

Table 2. the Zemax  polynomial fit starting at the zero-order anchor point from uvotgrascorr

order

a0

a1

a2

a3

a4

first

λ(Å) = Σ an Dn

1620.9±60

-0.855±0.39

(6.024±0.885).10-3

(-3.574±0.835).10-6

(8.137±2.80).10-10

second

λ(Å) = Σ an Dn

1307.7±42.3

0.169±0.128

(1.21±0.12).10-3

(-3.03±0.38).10-7

second

D(pix) = Σ an λn

-1497.8±72.1

1.779±0.088

(-4.236±0.353).10-4

(4.886±0.459).10-8

Using the dispersion relations, a spectrum extracted using uvotimgrism can now be displayed in the wavelength region 1800-3000 Å, by calculating the wavelength scales for first and second order from the Dispersion variable in the output file. The spectrum can then be plotted with both scales, and a strong first order line should also be visible in the second order at the same wavelengths. 

The spectrum of the variable HD109962 was examined this way, but showed no evidence of strong first order lines in second order.