MIT/Lincoln Labs CCID20 W19C1 CCD Test Results

This is the first standard epi, thinned, backside MIT/Lincoln Labs 2Kx4K CCD to be characterized in the UCO/Lick detector lab. A report is available.

Highlights for this device:

• Epoxy void. There is an apparent epoxy void between the thin CCD membrane and the supporting silicon wafer. This produces a significant depression in the device. Take a look at the "brick wall" data below for more explanation.
• Read noise is around three electrons.
• CTE is very good.
• Flatness is superb!

QE curve

The QE measurement is similar to the high resistivity QE curve (see W66C2) but with a more standard red response.

Surface plot

This is the flattest CCID20 we've seen so far! This measurement was made at room temperature with the device in its chip carrier.

Brick wall

This is the pattern which results from the incomplete backside laser anneal. Lincoln Lab personnel are investigating the possible causes and cures.

This surface plot shows a 30 mm X 30 mm section of W19C1. This is about one half of the area of the CCD and is the end closest to the bond pads and connector. We couldn't scan the other half of the device because (1) the dimple at the location of the epoxy void throws the laser beam off so rapidly that it can't be tracked and (2) the reflected laser beam is too faint to track on the third of the CCD farthest from the bond pads. We don't understand why there is such a difference from one end of the device to the other, but we've also seen this effect on W66C2.

At any rate, the important thing to note is the VERY small amplitude. In fact, about a third of the apparent curvature can probably be attributed to systematic errors from the measurement system. A visual inspection of the CCD indicates that the rest of the device is also very flat (except for the dimple at the glue void).

The following series of images shows a small area of the 2Kx4K CCD at different wavelengths. For each image we list the wavelength and amplitude of the pattern. The amplitude is determined by taking a single slice through the image and noting the maximum variation relative to the mean level. For some reason the amplitude of the pattern remains high even at long wavelengths. This is a puzzling result because the pattern is supposed to arise from annealing non-uniformity which occurs within a fraction of a micron of the surface. Long wavelength photons penetrate well beyond this surface and so those images should show a very low amplitude brick wall pattern. Lincoln Labs is investigating this problem.

Note the roughly circular pattern in the last two images. We think this is caused by a small void in the epoxy layer between the CCD and the supporting silicon substrate. Only at the longest wavelengths does light penetrate all the way to the back, where the reflection characteristics are changed in this location due to the absence of the glue. (We've had this CCD in a vacuum and it didn't blow up, so it isn't clear if there is air behind this void.) The width of the void is about 6.4 mm (426 pixels). The silicon membrane of the CCD is depressed over the void, so this might be an optical problem in some applications.