Pressure-induced line wavelength shifts in the 384cm-1 setting

Graham Harper pointed out that strong atmospheric lines may be expected to shift due to pressure broadening. When I set the wavelength zeropoint for reduced spectra I assume that the telluric lines are where HITRAN predicts and are at zero velocity.

If the pressure shifting effect is significant it would be particularly bad for the 384cm-1 setting used for Betelgeuse and Antares, since I used very strong lines to set the wn zeropoint. 

To investigate the magnitude of the effect,  I took spectra of water lines using the laboratory gas cell. 

I set up EXES for HIGH_MEDIUM, 384 cm-1, 2nd order, with the 1.9" slit. Then I took a number of spectra with different fill pressures in the gas cell. The water content was supplied by ambient vapor in the room that doesn't get swept out by the scroll pump. I added dry pressure with N2 from a spent cryogen tank. 

Here's what the undistorted sky spectrum from observations on Antares during F198 looked like.

And here are a couple of spectra of the gas cell at different pressures.

And then I reduced them to 1-d spectra

And measured the width and wavenumbers of the strong lines, using the same fitting tool as I use for wavelength calibrating all piped spectra.

There is a clear linear trend between FWHM and wavenumber -- [Sorry, the units and y-axis are mislabeled. Should be "Central wavenumber (cm-1").]

The width of the real sky line data in F198 is shown with the vertical magenta line. If the gas cell lines at their narrowest represent the true rest wavenumber from HITRAN, then using the positions of the sky telluric lines would induce a spurious blueshift of 1-2 km/s, by my estimation. 

I also measured the weak H2O line at 383.510 cm-1. It was present in the sky data and the lower pressure gas cell data. It doesn't appear to have the same linear trend. Does that mean I can trust its HITRAN value? The fits to the higher pressure gas cell spectra were not very robust due to the small line strength next to similar scale spectral fringing.

Indriolo 1485cm-1 setting

We weren't able to observe at 1485cm-1 during our first flight series in April 2014 because of filter availability. Our best filter choice would have been N07053 but tried to make due with BP6600 + MgF2 and waisted some time on it.

The setting is requested again for Cycle 4 by N.Indriolo, so I looked at 1485cm-1 at 6th and 8th order to check for any issues. We already used N07053 on sky to get ~1490cm-1 at order 3 for the IRC +10216 / VY CMa spectral maps, so that it works is no surprise.


Here's the side-by-side comparison of ~1485cm-1 in 6th order, Left: BP6600+MgF2. Result: 1500 counts in 136 sec. Right: N07053+CaF2. Result: 1500 in 30s.

Left: Apr 10 commissioning flight. 1500 counts in 136 seconds. Right: 1500 cts in 30 sec.

Using the grating at intermediate angles (36 deg for 1485cm-1 in 6th order) incurs a penalty in thoughput due to groove shadowing. If we take 1485cm-1 in 8th order, the highest available, we get the same flux in about 15 seconds. Pictured below. Unfortunately, the slit is about 10.5" and maybe not suited for on-source nodding. So we may opt to go for coverage and a lower order. We were able to get 2500+ counts from the blackbody in 14 sec at 3rd order. However, the decker was only 7-pix.

N07053, 1485cm-1 at 8th order. ~1500 counts in 15 seconds on blackbody