In part 1 of this series, Ming Zhao outlined the problem: how to get a nice, uniform spread of light across lots of pixels when trying to do sensitive infrared photometry? We discussed lots of options, but it was a conversation with another assistant professor here at Penn state, Suvrath Mahadevan, that showed us the way. The story starts way back in Suvrath’s grad student years…
[queue wavy flashback effects…]
Vintage photo of Suvrath, then a young graduate student in snowy Florida
Suvrath was at the University of Florida, working with his thesis adviser Jian Ge (both of them formerly of Penn State, actually) on the Kitt Peak Exoplanet Tracker. The Exoplanet Tracker was an externally dispersed interferometer (I worked on one, too — TEDI, at Palomar, with James Lloyd, Dave Erskine, and Jerry Edelstein, who was also my wife’s thesis adviser… wait, I’m getting off track).
Anyway, ET was a planet-finding instrument that measured precise radial velocities by passing starlight first through an interferometer then through a low- or medium-resolution spectrograph. The problem was that while the interferometer was great for precise velocities, the fringes it produced in the spectra were terrible for instrument calibration. Normally, one sends spectrally featureless light through a spectrograph to get a “flat field” and measure the instrumental response. They had a build a much more stable version of the interferometer than the one run by a piezo, but the new version couldn’t be jiggled around to wash out the fringes (sound familiar?). Without flat-fielding, the project wouldn’t work, and they couldn’t take flat fields.
What’s worse, they wanted to do do wavelength calibration to take out the effects of a tilted slit using an arc lamp, and the fringing of the interferometer made this very had, as well. The fringes were the whole secret sauce that made the ET and MARVELS projects work at all, and yet they were confounding all of the normal ways one calibrated instruments.
Suvrath had been working on the problem to no avail, and decided to put it down for a while and switch gears. As he turned and got up, a copy of the Edmund Optics catalog fell to the ground and opened up to a random page. Survath picked it up and…
light poured down from from the heavens, choirs sang, a gentle breeze tousled Suvrath’s flowing locks as his eyes widened and he saw the very answer he sought on the open page…
A holographic diffuser.
A holographic diffuser is an optic that scrambles the directions of light that passes through it. Sort of like frosted glass makes everything behind it (very very) fuzzy, these do the same except they don’t block or reflect any the way frosted glass does, so you don’t lose any light.
If you put a holographic diffuser in the light path of a camera or spectrograph, then light that normally would come to a nice sharp focus will get diffused out into a Gaussian, or tophat, or some other shape that the designers choose. Up close they look like clear wafers with little worm shapes etched into them — I think that the direction light gets redirected is basically a function of what part of the diffuser it hits.
So, Suvrath and Scott Fleming, another student of Ge’s, installed a holographic diffuser in ET (and, later, MARVELS) to make “milky flats” — these diffusers washed out the fringes caused by the interferometer and allowed them to calibrate their instrument. They also allowed for “non-fringing” arc exposures to be taken through (and despite of) the interferometer.
Problem solved. Project saved. Magic book of fate to the rescue.
Or, I should write, problemS solved, because years later in a conversation about Ming Zhao’s project, Suvrath would suggest these diffusers to solve the problem the telescope that wouldn’t defocus enough, and the seeing that was never quite bad enough. If we installed a holographic diffuser at Palomar, it might blur out the stars just the way we needed, and in a very stable, predictable way.
Maybe. These things have a way of being much more complicated than you think they should be. Stay tuned for the next installment…