Technology Nugget—Mirror Segment Manufacturing: Part 2
  Jerry Nelson
  University of California, Santa Cruz

In the last TMT Newscast, we described the basic challenging of polishing the telescope’s many mirror segments, and the application of Stressed Mirror Polishing (SMP) to polish the desired aspheric surfaces of the segments. For technical reasons, this polishing is done on circular mirror blanks. The warping of the segments is much more readily done if forces can be applied along the smooth circular perimeter. The basic idea is that after the polishing, one cuts away the outer parts of the mirror, leaving the desired hexagonal shape for the segment.

There is another important reason to polish circular mirrors, and then cut them. Since the primary mirror is composed of many hexagonal segments (738 of them), there are lots of segment edges that are in the interior of the primary mirror. Hence, it’s important that the segment edges are polished to the same high accuracy as the interior of a segment. This is important because the polishing process involves a polishing tool moving over the mirror blank, and as the tool moves over the edge of the mirror, the material removal rate tends to change (as it tends to be proportional to the tool pressure).

Thus, it is almost universal that the edges of polished mirrors are rounded, and thus optically imperfect. The polisher tries to avoid this, so one ends up with edges that are sometimes too high, sometimes too low, but always messed up to some extent. With a monolithic mirror, this occurs only at the outer edge, and it’s usually masked off so the astronomer doesn’t see any light from this imperfect part of the mirror. For TMT segments this would be unacceptable. So, by cutting the hexagon out of the circular mirror we are also cutting away the imperfectly polished edge, leaving hexagonal segments with optical surfaces perfect right up to the edge.

So this sounds pretty good. For SMP, we prefer polishing a circular mirror, then cutting away to make the hexagonal shape. Polishing processes tend to round the edge, so we have cut this imperfection away. However, there is yet another tricky issue.

Most solids have built-in stresses in the material. Glass is no exception. Most mirror blanks have compressive stresses on the outer parts of the blank, and tensile stresses in the interior. Combined, they are in equilibrium. This is usually good, since glass is a brittle material, and if the outer surface is in compression it makes it more tolerant of small impact loads that could locally chip or crack the surface.

If the surface was in tension, then such accidental damage could cause cracks to propagate and the mirror would shatter. These stresses arise as the glass cools, and the outer part cools and solidifies while the interior is still a viscous fluid. Once the outer part solidifies, the volume is defined, and as the warmer interior cools it shrinks, putting the skin in compression. One often anneals the glass (cooling it VERY slowly) to minimize these stresses.

Now we have a curved mirror, and we cut a hexagon out of it. Thus the outer edge that was in compression is removed. This changes the interior distribution of stresses in the mirror segment and often this will cause the mirror to warp. So, after all the work of polishing the mirror, cutting it into a hexagon degrades the optical surface shape!

There are two tricks we can use to recover. In practice, the warping from cutting is a small effect, though not negligible. We can measure a part of the interior stress, measuring what is called birefringence, an optical quantity. If we measure this before we cut it, we can predict with moderate accuracy how much the mirror will warp when we cut it. Hence, we can actually apply this computed correction in the Stressed Mirror Polishing calculations, and polish out the predicted warping.

The second trick, a much more powerful one, is to take advantage of another polishing technique called ion figuring. Some 30 years ago, people discovered that if a piece of glass is bombarded by high-energy ions, these energetic atoms will ablate the glass surface when they collide with it. This is analogous to sandblasting on an atomic scale.

NASA developed ion engines for space with high-enough intensity to make this technique interesting. It is a very slow process compared to ordinary polishing, but it’s very predictable and produces no edge effects. So, any number of small optical defects (such as warping from cutting) can be optically measured and then removed by ion figuring. If the amplitude is small enough, then this becomes a practical approach. And, if we are careful, we expect that the amount of ion figuring we will need to do will be small enough.

This technique was used to finish the polishing of the Keck Observatory mirror segments and it worked exactly as hoped. In practice, one can only remove errors smaller than about one micron, but that’s a huge help.

So, we plan to use stressed mirror polishing on circular mirrors. Then we will cut them into hexagons. They will warp a modest amount, and we will remove this error by ion figuring the mirrors, thus producing the desired optical surfaces on the mirror segments.

This may sound very complex, and perhaps it is. However, when you want to achieve an optical surface accurate to 10 nanometers, or ten-thousand times smaller than the thickness of a human hair, you need to use all the tricks you can get!