A variety of defects were easily seen by the naked eye. The mirror looks a bit milky, one side more than the other. There are a few radial stripes as well as a few areas that look smudged. The milkier side of the mirror shows record groove like smudges. This is not a very quantitative analysis method but a good mirror will look uniform with no visible defects. This was clearly not uniform and had many and varied defects.
We looked at a variety of places on the mirror. We imaged average areas as well as the visible problem areas. We also took two images on the bare SiC . Each image below represents about a 1mm spot on the mirror.
These are three typical areas 9mm, 42mm and 52 mms from the outer edge.
Figure 1 - 9 mm from edge(file fig9.JPG)
The white spots are defects. Probably places where imperfections in the material were ripped out during machining or polishing. You can also see scratches. I have included the file name for each image. They are stored in the same directory as this HTML. They are 1280x1024 jpegs and show more detail when blown up.
Figure 2 - 42 mms from the edge (file fig6.JPG)
Figure 3 - 52 mms from the edge (file fiig3.JPG)
You can see the white spots which are surface defects and a variety of scratches. The surface defects are mostly elongated in the same direction and appear to be tears occurring during diamond turning. A good mirror would not show these bright spots at anywhere close to this density.
The next two images show an image from the low scatter side compared to one from the high scatter side. The defect density is about the same but the high scatter side shows far more polishing scratches.
Figure 4 high scatter side(file fig1.jpeg)
figure 5 low scatter side (file fig11.JPG)
The following four images show defects that were visible to the naked eye. Figure 6 shows a portion of the radial stripe which looks like a polishing problem. Figure 7 is probably a surface defect. Figures 8 and 9 are clearly surface defects. Figure 8 shows what looks like the SiC/Si matrix showing through and figure 9 shows some strange crystallization of the substrate.
Figure 6 portion of radial stripe (file fig7.JPG)
Figure 7 surface defect (file fig5.JPG)
Figure 8 SiC/Si matrix print through (file fig8.JPG)
Figure 9 crystallization defect (file fig10.JPG)
The last two images show the bare SiC near the central hole. Figure 10 is right at the edge of the hole and figure 11 is half SiC and half aluminum coated SiC. Jean says that the two tone pattern is caused by two different materials that have different phase effects on the two beams presumably silicon and silicon carbide. The pattern does not seem to print through the aluminum coating in most places as shown in figure 11
Figure 10 at edge of central hole (file fig2.JPG)
Figure 11 half SiC half Al coated SiC (file fig3.JPG)
The upper half is coated with Al.
The scatter measurements were performed on the 16th by Joni Pentoney. The testing setup is relatively straight forward. It used a HeNe laser that passed through a chopper with reflective blades. When the blade was in the beam the laser was directed to a reference detector. When the chopper blades were not in the way the beam traveled to the mirror under test. A large,~18" salad bowl like mirror is with a hole in the bottom is positioned so that the incoming laser beam goes through the hole and the concave surface faces the test mirror and is very close. This way nearly 4pi of the scattered light is collected by the mirror. A detector is positioned at the focus of the mirror to measure the scattered light. A lockin amplifier is used to measure the signal from each detector. The laser output power does vary so the reference signal is used to correct for changes in output power. A standard scatter material made by Labsphere was used that had a quoted scatter of 1.4%
The scatter was measured at 106 positions on the mirror in at 5 different radii. The measured scatter followed the apparent scatter of the mirror showing 2 to 3 times the scatter on the milky side. The average scatter for each radii is shown below along with the Russian scatter measurements.
Position measured from the edge of the central hole outward. The mirror is 96.5 mm from the edge of the central hole to the outer edge. The Russian data stopped at 75mm which is about .8 inches from the edge. Since the scatter seemed to be climbing radially we decided to do the 90mm point. This was so close to the edge that Joni was afraid that we may have been falling off of the aluminum and did the 84mm set as a check.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The graph below shows one complete data set for the 90 mm radii. Figure 4 above shows a microscope image taken at the 280 degree position near the peak of the scatter curve. Figure 5 above was taken at the 100 degree position. It seems pretty obvious that the increased scatter is caused by a lot of long scratches and an increased number of small point defects. The number of large defects seems about the same. This implies that the increased scatter on the 280 degree side is caused by polishing problems.
The average scatter using all of the measurements is 2.14% which is about 4 times the .5% specification.
The last test was to measure the surface profile directly with the Talystep profiler. The machine uses a .8 micron stylus that rides directly on the surface with a .5 Mg loading. The stylus is moved across the mirror over areas of 0.1 to 0.5 mms and the height of the stylus is measured as a function of position. The instrumental noise is about .5 angstroms. We were unable to measure profiles on the aluminized portion of the mirror. Some sort of contaminant on the aluminum surface would gunk up the stylus and we saw big jumps in the data as it would build up and then jump over the stuff. Instead we measured the profile at the edge of the mirror where there was no aluminum.
Figure 12 (tip your head a little to the right) 300 micron scan - The curvature is caused by the curvature of the mirror and the angle at which we mounted it. This position was chosen away from any of the large defects to see what the roughness was in between the defects.
Figure 13 100 micron profile of position 90-190 on figure 12 - This portion was chosen because it was relatively flat and the RMS roughness calculation would have some meaning. These graphs show that the surface is quite smooth in between the scratches as was the case with the old mirror. From the scatter measurements above we calculate a surface roughness of 73.1 angstroms. While it was already obvious from the microscope pictures the Tallystep measurements seem to confirm that the scatter is caused by the point defects and scratches as opposed the bulk material.
