Extreme Photonics: Mega Mirrors

To see the big bang, you need big optics.

by Dan Drollette, Senior Editor

When Mary Edwards wanted to look for interior flaws in the glasswork her colleagues were making, she took off her jewelry, removed her shoes and belt buckle, and emptied her pockets of sharp objects. She then stepped very carefully onto the surface of a mirror 8.3 m in diameter. “The only way to inspect is to get on your hands and knees on the glass and examine it with a microscope,” said the senior product engineer at Corning Inc. in Canton, N.Y. “After you’ve worked with glass for a long time, you know it will be OK as long as you respect it and take precautions so that nothing will fall off and chip or scratch it. Still, there’s elements of black magic. ... Glass making is equal parts science and art at this level.”

The glasswork was destined to become the largest single, or monolithic, mirror ever made and the heart of Japan’s new Subaru telescope. (“Subaru” is Japanese for the Seven Sisters, a cluster of stars that Westerners call the Pleiades.) Like all telescope mirrors of extreme size, the Subaru mirror is a one-of-a-kind item because it was too expensive for the manufacturer or the customer to cast a backup. Consequently, there was little margin for error.

And the potential for disaster lurked everywhere. Cool the 45-ton mirror too fast after coming out of the 1700 °C furnace and cracks could form. The massive mirror also had to withstand the stresses that go along with being sagged into its rough shape and reheated to its final curve.

Just grinding a mirror is abrasive and abusive; computer-guided, diamond-impregnated grinding wheels actually tear particles off the glass surface. Probably the most nerve-racking moment came when it was time to carve excess weight from the mirror’s underside: The mirror-makers had to flip the monolith over like a giant pancake without marring it.

Handling an 8-m-class primary mirror is difficult enough, but astronomers already are sketching plans for telescopes 12 times as large as the Subaru. There are now plans for telescopes with mirrors that are 30, 50, even 100 m across. A mirror the size of a football field may make some astronomers scoff, but Roberto Gilmozzi, director of the European Southern Observatory’s Paranal Observatory telescope in Chile and promoter of the proposed 100-m telescope, is undeterred: “If we build this, it will have 10 times the collecting area of every telescope ever built put together.”

If astronomers succeed in making such quantum jumps in mirror size, they will overcome the forces of history. Every time the size of a conventional mirror is doubled, the complexity and the cost increase eight times; consequently, it used to take 20 to 30 years between each jump. Now engineers hope to shorten the cycle.

They expect to do so by using optical interferometry, in which the light beams from several mirrors, tens or hundreds of meters apart, combine at a central detector; and by using adaptive optics, in which thin, flexible mirrors move hundreds of times per second to compensate for the distorting effects of turbulence in the Earth’s atmosphere.

Earth vs Space

To someone outside the astronomical community, it is hard to understand why astronomers need such large “light buckets,” as telescopes are irreverently called. The Hubble Space Telescope neatly avoids the distortions of the Earth’s atmosphere by hovering above it to bring back clear images of the universe. And a newer, bigger space-based telescope is in the offing.

But while the Hubble has taken astonishing photographs and worked well in areas such as ultraviolet astronomy, it is only 2.4 m in diameter and is limited for fields such as spectroscopy. Choreographing the maintenance and operation of a telescope in space presents logistical problems, as evidenced by the recent space shuttle missions needed to replace the Hubble’s gyroscopes.

Researchers say Earth-bound telescopes complement the work of space-based telescopes. Sir Martin Rees, Astronomer Royal at the University of Cambridge in England, explained, “The new (Earth-bound) telescopes will enable us to see our entire Earth in a grand, cosmic perspective. We can presently see back 90 percent of the way to the big bang. If the new telescopes are indeed built, we could go back still further, to the most distant and first galaxies. We could see when the very first stars formed and even peer into the cosmic dark age, when the universe began cooling.”

He added that astronomers could even image planets around other stars. In the last few years, scientists have found tantalizing – if indirect -- indications of large, Jupiter-like planets outside our solar system. “But with very large telescopes, humans could actually see these planets — even small, Earth-sized ones,” Rees said.

To do so, however, astronomers need to overcome the vastness of space. If the Earth were the size of a grain of sand and our sun were scaled proportionately, it would be the size of a lightbulb and 40 ft away. If this lightbulb were at, say, the doorway to the University of California’s Lick Observatory, the next-nearest lightbulb would be in Chicago, 2000 miles away. In essence, an astronomer looking from Earth for a planet orbiting another star is like an observer looking from Southern California for a grain of sand circling a light bulb in Chicago. The next-nearest star would be in Europe.

Optical tradeoffs

To spot these distant specks, astronomers need telescopes with both high light-gathering ability, or sensitivity, and the capacity to distinguish two distant objects that are close together, or resolution. Unfortunately, it is hard to get both characteristics in one telescope, said Jerry E. Nelson of the Lick Observatory. He said interferometry, for example, offers tremendous angular resolution because of its longer baseline in comparison with conventional mirrors.

However, interferometry is less sensitive than conventional mirrors and has a very small field of view: “It’s like looking at the stars through a soda straw,” Nelson said.

In contrast, conventional monolithic mirrors offer lower resolution but greater sensitivity and a wider field of view. Thomas A. Sebring of the National Optical Astronomy Observatories in Tucson, Ariz., said, “For an astronomer, the equivalent of the old hot-rodder’s adage would be, ‘There’s no substitute for square meters.”

Monoliths are also easier to operate and control. Their main problem, besides low resolution, is that the larger they are, the more difficult they are to fabricate and ship. With current technology, 8 m seems to be a practical maximum size for monolithic mirrors, said Corning’s Edwards.

To get the best of both worlds, Nelson, Sebring and others interested in large telescopes are looking at filled aperture mirrors. Instead of a single monolith of glass, these telescopes contain several smaller mirrors in a modified form of interferometry. Unlike typical interferometry, however, these mirrors are not meters apart but are immediately adjacent to one another, sharing the same telescope support structure. Each mirror is just a segment of the whole and they behave as one. For example, each of the two Keck telescopes has a primary mirror composed of 36 smaller hexagonal mirrors. These give each telescope a combined viewing capability of a 10-m mirror.

The new telescope designs combine parts of interferometry and adaptive optics to varying degrees. To achieve large telescopes, astronomers seem to have settled into a few general approaches: ambitious projects, such as the European Southern Observatory’s appropriately named OverWhelmingly Large 100-m telescope; and 30-m telescopes that seem moderate by comparison.

The OverWhelmingly Large telescope proposal has drawn the most attention, mostly because of its vast size, which is about the height and width of the Great Pyramid of Egypt. Its most ardent promoter, Gilmozzi, admits that it is the most speculative of all the approaches. It strains the limits of adaptive optics and requires superfast computers that have not been built. However, he said, “Every other aspect of this design uses only proven technology. Both the mechanical and optical design of the 100-m is based on proven approaches. There is nothing there that cannot be done today.”

To illustrate the OverWhelmingly Large telescope’s practicality, he cited the design of its primary mirror. Instead of requiring many custom mirror segments of different curvatures, his approach calls for 2000 identical hexagonal mirrors. Instead of the usual parabolic shape -- which reflects light to a single precise focus point -- each hexagon would be a spherical mirror 2.3 m in diameter. Other, corrective mirrors down the optical path would correct for the inevitable spherical aberrations.

Gilmozzi said this "Lego approach" would enable mass production of the pieces. Because all the segments would be exactly the same, they would be easier to grind and polish. Fifteen would easily fit inside a standard shipping container.

Based on industry estimates, Gilmozzi calculated that if the hexagons were manufactured at the rate of about five per week, the telescope mirror could be completed in six or seven years. Astronomers could also use the telescope during its construction: It would start as a 2.3-m version and grow into a 100-m behemoth, with astronomers continuously using it to make observations as it expanded.

Gilmozzi predicted that the OverWhelmingly Large telescope, if started immediately, could be completed by 2015 at less than one-thirtieth the cost of a monolithic version. At a projected $1 billion, this telescope would still be expensive, but not overwhelmingly so.

Too big?

Nelson, however, is not convinced. “The OverWhelminglyLarge telescope is an incredibly ambitious project. It pushes the envelope and pushes the state of the art and gets the creative juices flowing. That’s great, I’m all for them. ... But the number and magnitude of technical challenges is such that it will not be completed in the near future.” Instead, he favors smaller telescopes, such as his proposed 30-m California Extremely Large Telescope.

Nelson said scaling up the dimensions of any structure has limitations. “You don’t just go from an existing design and make it ten times bigger,” he said. At 20,000 tons, the weight of the OverWhelmingly Large telescope alone would be far beyond that of any similar structure. Simple practicalities become more difficult, such as how to move the monstrous object into position for a night’s viewing. “Just putting a dome around a 100-m scope is a huge project,” he said.

Even Keck’s adaptive optics remains “very challenging,” Nelson said, because it is an area that is still in its infancy, with many kinks still to be ironed out:

- Nelson’s major concern is the adaptive optics system required to operate on such a scale. The largest telescope using this technology today — the Keck II telescope at Mauna Kea, Hawaii — uses only a few hundred actuators to move the mirrors. A ten-fold increase in the size of an adaptive mirror requires 100 times as many actuators and 10,000 times more computational power to control them. Gilmozzi estimates that his proposed 100-m telescope will need 100,000 to 500,000 actuators.

- Adaptive optics systems need a bright light source near the target object to calibrate for atmospheric distortion. Not all astronomical targets have the necessary bright reference point nearby. One solution is to bounce a laser off a known layer of the atmosphere and to use it as the “guide star.” The idea has been demonstrated at a few locations, but it remains far from an everyday operational tool.

- Adaptive optics can correct for only one tiny portion of the sky at a given moment. One possible cure is atmospheric tomography, in which the turbulence of several layers is calculated all at once. The combined results could then be used to determine the amount of warping needed on the mirror. However, this approach is even further in the future than the laser guide star.

And all of these approaches require massive amounts of computing power. “To say that we’ll have the necessary computational ability in ten years places too much faith in Moore’s Law,” said Nelson. “Such a large extrapolation for computing makes me nervous.”

Instead, he prefers smaller increments on the way to large telescopes. “Maybe, if we are successful, that will pave the way for something bigger, such as a 100 m. A sensible way to get there is in steps,” he said.

Whichever proposal — if any — is finally chosen, most astronomers said that they are thrilled that such large optical telescopes are being seriously contemplated. “A century ago, the public would vicariously share in the adventure of the exploration of the planet, following scientists as they traveled to Antarctica and the jungle,” Rees said. “Now the public can participate and enjoy in a similar manner the exploration of the universe.”

Photonics Spectra

Λ back to top