The Journey of a photon through SNAP

Mar 25, 2006
April 6, 2006

Once upon a time, a star in a distant galaxy exploded. Light from the supernova travelled for millions of years through space. One very lucky photon happened to fly in exactly the right direction to enter the SNAP telescope. Let's follow this photon as it traverses many obstacles within the telescope and detectors; it will tell us much about the procedures we must use to calibrate the collected data.

After it enters the telescope aperture, the photon must bounce off four mirrors: the primary, secondary, folding flat, and tertiary. At each surface, there is a chance that the photon may not reflect properly. The overall reflectivity will vary slowly with wavelength. We need to correct for this wavelength-dependent reflectivity.

We can measure the overall reflectivity of the mirrors by comparing two sets of long-term observations. First, the repeated measurements of stellar brightness in the SNAP field(s) over the course of months; we can choose many thousands of stars with high signal and no sign of variability, spanning a large range of color. Second, measurements of light from a set of LEDs which illuminate both the detectors and a number of photodiodes on the focal plane. We use the photodiodes to determine any change in the LED output over time, and correct for it. Then, we compare the long-term trends of the stellar light (which has bounced off all 4 mirrors) with the LED light (which has not touched any mirror). Any excess decrease in stellar light must be due to changes in the reflectivity of the mirrors.

After bouncing off all the mirrors, the photon will head towards the focal plane, which covers a large area more than half a meter in diameter. As a result, photons from different parts of the field will approach the focal plane at slightly different angles. Each photon will then enter a filter. Most photons will be reflected or absorbed by this filter; only a select few will pass through it. We need to know as accurately as possible the transmission of each filter as a function of wavelength.

Before flight, we can test each filter thoroughly by comparing beams of monochromatic light which do and do not pass through the filter. During flight, however, the transmission of filters may decrease overall, and/or the passband may shift in shape or position. We can determine some of these changes by repeatedly shining light from LEDs -- which have known, relatively narrow spectra -- onto the focal plane. If we have enough LEDs, we can measure the transmission of each filter (plus the efficiency of the detector behind it) at several wavelengths across its nominal passband. If a filter's properties change, the relative intensities of measured light from these LEDs will change. In a similar way, we can use the light from stars -- either a small set of bright stars with very well known spectra, or a very large set of fainter stars in the SNAP field(s) which are measured very frequently -- to check for shifts in the filter transmission curve.

Variations in overall filter transmission from place to place across the glass (and detector efficiency across the chip) will show up as spatial variations in the diffuse light from the LEDs and from stars. We can use these same measurements to verify the small, systematic variation in transmission as a function of angle of incidence within each filter.

Note that we cannot separate fully some properties of the filters from similar properties of the detectors behind them.

The photons which pass through the filters will then strike the SNAP detectors. Most of these photons will be registered by the detectors, but some fraction will reflect from the surface or fail to be recorded. The efficiency of each detector will change as a function of wavelength -- slowly over most of its range, more quickly at the extreme edges. The detectors may suffer from other problems as well: some of the electrons liberated in one pixel may diffuse through the subtrate and be recorded at a neighboring pixel location. The very large pool of electrons at the location of a bright star may not be recorded properly (non-linearity) or overflow the gates and flow across the chip (bleeding).

Pre-flight testing will give us most of the characteristics of each detector we need. However, it is possible that some properties of the detector may change during the mission. We can monitor changes in the quantum efficiency of a detector (and the filter in front of it) by shining light from LEDs onto the focal plane periodically, and comparing the light recorded by the detector to the light recorded by a nearby photodiode. If we have several LEDs which produce light with wavelengths near the edge of a detector's range, we may be able to track precisely variations in its cutoff region.

The infrared detectors are likely to suffer from at least one additional complication: inter-pixel capacitive coupling. The charge collected in the readout electronics of one pixel can exert an influence on the measurements made by its closest neighbors. As long as we characterize this effect carefully before launch in each detector, we should be able to correct for it easily when measuring the isolated bright stars which serve as calibration sources.