Published online 12 March 2008 | Nature 452, 142-145 (2008) | doi:10.1038/452142a

News Feature

Astronomy: Eyes as big as the sky

Three teams are racing each other to build the next generation of telescopes that would dramatically dwarf the largest on Earth today. Eric Hand checks out the competition.


Buddy Martin gets nervous before setting foot on the glass. He reaches into his pockets and sets aside a mobile phone, coins and fingernail clippers. He checks the soles of his shoes for grit. Then he gingerly steps out onto the 8.4-metre-wide disc, parked underneath the University of Arizona football stadium in the hangar-like cavern of the Steward Observatory mirror laboratory in Tucson.

Still unpolished and milky white, the disc sits like a record on a turntable — except that this record is borosilicate glass instead of vinyl, spins at 1 revolution per minute instead of 45, and weighs as much as 3 buses. Martin, a polishing scientist, is the disc jockey. He needs to shape the surface like a saddle and get it right to within 20 nanometres. Proportionately, it is like bulldozing all of Brazil to a smoothness of less than a centimetre. Thus Martin's anxiety over errant scrapes: "We have rules such as 'no tools above the glass'," he says. "A wrench falling would wreck it."

When Martin can confirm that the glass is polished correctly, it will be ready to become a mirror. Technicians will seal a giant bell jar over the glass and vaporize a soda can's worth of aluminium over the surface. The result will be as large or larger than all but two of the telescope mirrors currently in use around the world. But that is only the beginning. The Giant Magellan Telescope (GMT), pictured above, which is under development by a consortium led by the Carnegie Observatories in Pasadena, California, and the University of Arizona, will require six more mirrors just as big. If the first mirror is Brazil, there is still the rest of the Americas, all of Europe and a good chunk of Africa to go.

Click to read this graphic.GMT/TMT/ESO

With its seven mirrors the GMT will be 25 metres across, dwarfing the biggest telescope mirrors in the world today — those in the twin 10-metre Keck scopes in Mauna Kea, Hawaii. Yet even the GMT will be the smallest of the bigs in a game of one-upmanship that has emerged among astronomers. A second group, led by the University of California and the California Institute of Technology (Caltech) in Pasadena, is planning a giant named the Thirty-Meter Telescope (TMT). And the European Southern Observatory (ESO), headquartered in Garching, Germany, aims to build the biggest 'light bucket' of all: a 42-metre design called the European Extremely Large Telescope (E-ELT). All told, these three competing behemoths could cost more than US$3 billion (see Table).

The giant size of these scopes leads exponentially to big science. A telescope's ability to gather light increases with the area of the primary mirror, and its ability to resolve details increases with the diameter. All told, tripling in diameter from the Kecks, this new generation of telescopes will be roughly nine times better at grasping the light from distant stars, and nine times better at distinguishing them from darkness.

Telescopes this big could tackle fundamental and currently unanswered problems, such as the direct imaging of light from an extrasolar planet or measurement of the mysterious repulsive force that's accelerating the expansion of the Universe. "It's not just pushing for bigger toys," says Matt Mountain, director of the Space Telescope Science Institute in Baltimore, Maryland. "They open up a new parameter space that we don't have access to."

The giant telescopes would rival, even in some ways surpass, the James Webb Space Telescope (JWST), the 6.5-metre replacement to the Hubble Space Telescope that is supposed to launch into orbit in 2013. By getting above Earth's atmosphere, the space telescope will be able to see deeper into the infrared portion of the spectrum. But the ground-based telescopes, with their larger mirrors, would be able to spot smaller and fainter objects. Both the JWST and the ground-based telescopes are aiming to spot 'first light', the time 400 million years after the Big Bang when stars first began their fusion fire. "We want to observe photons from the first stars in the Universe," says TMT observatory scientist David Silva. "Are we crazy or what?"

“We want to observe photons from the first stars in the universe. Are we crazy or what?”

David Silva

The three groups are lining up partner nations and some are competing for private donors. The Canadians have committed to the TMT, and the Japanese are likely partners too. The Australian National University in Canberra, joined eight American institutions in the GMT project's growing list of partners, and now South Korea is working to find the money to become the tenth partner. Private funding is also playing a large role; in December, Intel founder Gordon Moore, a Caltech graduate, gave the TMT project $200 million, making it the team to beat. It's far from clear if all three scopes will be built, but each is pressing ahead, hoping to commence construction within the next two years.

Part of the competitiveness stems from the importance of finishing first. The Keck telescopes beat other telescopes in their class to operation by a few years and skimmed the creamiest science off the top. And so the TMT's competitors want to keep it from drinking everyone's milkshake before they have their own straws ready. But the tension also derives from a historical discord between the three rival groups, groups that have generally got what they wanted even as they pursued telescopes of different sizes and shapes. "The worst job in the world would be to get these three teams in a room and say, 'How do we collaborate?'" says Mountain.

The acrimonious divorce

To understand two-thirds of the competition, one needs to go to Pasadena, home to the small, privileged institutions of Carnegie and Caltech. For astronomers through much of the twentieth century, this Los Angeles suburb was the sociological centre of the Universe. Initially united, Carnegie and Caltech worked together to finish in 1948 the 200-inch (5-metre) Hale Telescope at Palomar Observatory in the southern California mountains. For decades it was the biggest eye on the Universe, one that discovered the incredible redshift of quasars in distant galaxies.

But in 1979 the two institutions divorced. The director at the time, Maarten Schmidt, orchestrated the dissolution of the telescope's joint operation, frustrated by managing a place that had two bodies to approve every decision. Carnegie, the scorned spouse, left to establish the two 6.5-metre Magellan telescopes in Chile, while Caltech pursued the twin Kecks with the University of California system (and, later on, NASA). The two institutions are separated by only a few miles, but today they are divided by more than just fault-lines and freeways. Bitterness lingers.

Click for a larger version of this graphic.R. BERTRAM/STEWARD OBSERVATORY MIRROR LABORATORY

On one side is the TMT headquarters, a renovated Catholic hospital on Caltech's satellite campus, where staff scientists work in the former dormitories of nuns. These blue-ribbon scientists work as if for a blue-chip business, jetting around the world to nail down partners, vendors and site licences. Recently they gathered in a glass-walled conference room to debate the finer points of five possible sites for the telescope — even though many acknowledge that a site on Hawaii's Mauna Kea would do the most to seduce the Japanese, who are also flirting with the ESO. While one astronomer lectures on the emerging problem of light pollution at a potential Baja California site, the remaining scientists peck away on laptops, multitasking. Silva, who has also worked for the ESO programme and the US National Optical Astronomy Observatory in Tucson, finds the TMT culture different from that of his former workplaces. "It's a confidence, it's an arrogance, it's a sense of not being afraid to be leaders," he says. "It's a sense of 'We are the best and we're not going to be humble about that'." It's Caltech to a T.

Caltech's rivals work just a few miles away, at the shady Santa Barbara Street headquarters for Carnegie. The GMT programme is the smaller operation, with 20 full-time employees to the TMT's 100. For the moment they also have much less money — although that could change at any moment with partners such as Harvard University, with its $34-billion endowment, and two Texas universities that have recently benefited from Houston billionaire George Mitchell and his new-found interest in cosmology. Carnegie director Wendy Freedman says she's expecting several major donations in the coming months that could collectively rival Moore's to the TMT.

Grand design

But the biggest difference between the two projects is in the technological ambition of the GMT design. The TMT mirror is designed on the same basis as the Kecks, just on a grander scale: 492 hexagonal mirrors 1.44 metres across instead of 36 mirrors 1.8 metres across. The seven mirrors of the GMT, however, represent a huge step up — some say a leap of faith — from the Carnegie's single-mirrored Magellan telescopes. The organization has never before dealt with the difficulties of focusing light from multiple primary mirrors. To remind his colleagues of the challenge, staff astronomer Steve Shectman got the building crew to paint the outline of the GMT's 8.4-metre mirrors in the car park at Carnegie. The circles dwarf the parked sport utility vehicles. The intent, he says, was to scare Freedman into a more cautious approach. "But it didn't scare her. It just got her all excited."

The two groups have been jockeying for years, ever since a Keck follow-on called the California Extremely Large Telescope was reborn as the TMT in 2002. Caltech's leadership had realized just how expensive the TMT would be, and that they would need as many partners as possible, not just Californians. But by that time Carnegie had gone off to start its own designs, says Richard Ellis, a lead TMT astronomer and former Palomar director. "Great opportunities were lost to get everyone on board."

"The E-ELT breaks every assumption we've made about the art of building telescopes so far." Roberto GilmozziESO

Led by Freedman, the GMT board worked quickly to level the playing field, which they felt favoured its competition, particularly given the involvement of the Association of Universities for Research in Astronomy (AURA). AURA operates US public telescopes with taxpayers' money, and was a partner on the TMT but not the GMT project. In late 2005, the GMT board sent a letter to AURA president William Smith, complaining that the association had already "picked the winner" and noting that no federal support had gone to early GMT work. The National Science Foundation (NSF) then told AURA to back out of the TMT project and shepherd both designs along. The GMT group has since lobbied the NSF to split whatever support it offers the two projects. Many scientists at the TMT, predictably, given its lead in money and design progress, would prefer a winner-take-all competition.

Some astronomers suggest that it would have been difficult for the groups to work together anyway. Too much time, money and prestige, they say, has been invested in two rival technologies: the segmented-mirror approach pioneered by Jerry Nelson at the University of California, Santa Cruz, now being applied by Caltech, and the honeycombed, spun-cast mirrors developed by Roger Angel at the University of Arizona, used by Carnegie. "There are almost religious issues tied up in these designs," says Mountain.

Vanity mirrors

Nelson has been pushing fly-eyed telescopes for nearly three decades. The inspiration for using segments as opposed to a complete mirror, he says, wasn't hard to find: the Romans used small tiles to create larger mosaic wholes. The real challenge was in finding a way to shape the segments. Most telescope mirrors are parabolic — a curved surface that reflects incoming parallel light waves towards a central focus. These mirrors are relatively easy to polish to high precision. But in a segmented telescope, each off-axis mirror occupies just a portion of the parabola and needs a custom shape — a much more difficult polishing task. In the late 1970s at Lawrence Berkeley National Laboratory in California, Nelson discovered that he could warp the mirror precisely by hanging lead weights in buckets from the edges of a mirror segment. He polished the mirror as if it were a symmetric parabola, then released the weights. The mirror popped back into the precise asymmetric shape he needed. "Working with the edges was all we needed to do," he says.

At one level, the GMT is also a segmented mirror, but one with only seven giant segments — each one of which relies on the method long advocated by Angel. Nearly 30 years ago, using a backyard kiln constructed from bricks and heating coils bought from Kmart, Angel and his former graduate student John Hill began experimenting with Pyrex glass, fusing together glass bowls from Angel's kitchen cupboard. Eventually, Angel worked towards two innovations: honeycomb moulds, to create a stiff, lightweight mirror that was mostly hollow, and spin-casting. Spin-casting involves spinning the glass as it cools so that centripetal forces guide the glass naturally into a parabolic shape, granting the grinders and polishers a head start. And the honeycomb technique gives the glass a thinness and lightness that keeps it from deforming because of its own weight and thermal expansion.

But like Nelson, Angel faces the problem of asymmetry. The earlier 8.4-metre mirrors made in Tucson were polished as parabolas; the six off-axis mirrors for the GMT have to be saddles. And that means Martin, the polishing scientist, thinks very hard about what they call "The Test". The test is a problem of metrology — how to measure mirror curvature to a strict standard. Measuring incorrectly means failure, such as when the main mirror for the Hubble Space Telescope was ground ever-so-slightly wrong, resulting in blurry pictures. "We all stay awake at night trying to figure out how not to make that mistake," says Martin.

So he and other opticians have constructed an elaborate Rube Goldberg-like system in a 28-metre tower above the GMT mirror, which will bounce light through two more mirrors and a hologram before an interferometer measures the patterns that signify perfection or not. One of the test mirrors alone is 3.75 metres across, nearly as big as the Hale Telescope atop Palomar. All sides acknowledge that the test of the first saddle-shaped mirror, scheduled for July, will be the GMT's crucial juncture. The GMT has received good publicity by being the first group to cast a mirror. But what the GMT team is really trying to do, in casting a mirror early and moving quickly to the test, says TMT scientist Chuck Steidel, is "retire risk".

The other side

Meanwhile, the Europeans have remained out of the fray, quietly corralling a consensus among the ESO's 13 member states and working towards a design that borrows from both the TMT and the GMT. The E-ELT will be segmented like the TMT, but bigger, with nearly 1,000 mirror segments. These are similar in size to the TMT segments, which will allow the two projects to pit vendors against each other. The European design borrows from the GMT in that the adaptive optics will be integrated into the telescope's mirrors (whereas the TMT corrects for atmospheric effects with extra mirrors farther down the optical path). Proponents of the former approach say that it yields a cleaner image and a wider corrected field of view.

The Europeans have had to set their sights a bit lower than originally planned, abandoning the preciously named OWL, or OverWhelmingly Large Telescope, after its 100-metre primary mirror proved preposterous. But by developing designs for the 42-metre E-ELT, a mirror that can be made now rather than in 20 years, the Europeans are playing to their strong suit: the dependable ESO budget. "On a year by year basis, I don't have to go and ask for money," says Jason Spyromilio, director of the E-ELT project office.

Dave Hilyard, Jerry Nelson and Eric Williams (left to right) work on the Thirty-Meter Telescope.Dave Hilyard, Jerry Nelson and Eric Williams (left to right) work on the Thirty-Meter Telescope.

With its €900 million (US$1.37 billion) price tag, the European project is by far the most expensive of the three, roughly twice as expensive as the Very Large Telescope — a four-telescope array in Chile that the ESO finished in 2000 after more than a decade. But the Europeans still think they can build the E-ELT without private money. "Yes, it's a big step, a larger project, a much more complex project," says E-ELT principal investigator Roberto Gilmozzi. "It breaks every assumption we've made about the art of building telescopes so far. But we're restricting our worries to the technical side. Finding the funds is something we can do."

Others say that Spyromilio and Gilmozzi are optimists — that they will either have to ask the ESO's member states for more money or find more partners (the two American projects are united in their fear that Japan or the NSF will partner with the ESO instead of with them). A third option, Spyromilio says, is that the ESO programme takes out a loan against its future funding streams — mortgaging the ESO's future to pay for the E-ELT now.

All eyes are now on the NSF to see what it will do next. All three projects need more money just to begin building their telescopes. Yet the NSF will have none to disburse until it finishes the Atacama Large Millimeter Array (ALMA) radio telescope in Chile, in the next five years. NSF astronomy director Wayne van Citters told astronomers in November that after ALMA's completion, the agency could provide a similar contribution — about $500 million — towards the construction of a giant telescope.

Also on the minds of the Americans is the question of who will pay for operational costs, which typically amount to 10% of construction costs annually. At that rate it would cost more than $100 million each just to operate the GMT and the TMT each year, and the NSF would be likely to have to abandon its long-term support of telescopes in the 2-,4- and 8-metre classes. The astronomy community seems unlikely to want to make that sacrifice; a 2006 'senior review' from the US astronomical community found that these smaller telescopes provided valuable training and employment for the majority of astronomers. Small telescopes, the report noted, also provide excellent science; the first extrasolar planet around a Sun-like star, for instance, was discovered by a modest 2-metre telescope in France. And the discovery of dark energy — a huge discovery that surely will be afforded a Nobel prize some day — was achieved with a hodgepodge of telescopes, most of them 4 metres or smaller.

For now, the TMT seems to be holding on to its lead, although members of all projects know how tenuous such an advantage can be. The TMT programme is managed by Gary Sanders, who has shepherded through other massive science projects such as the Laser Interferometer Gravitational-Wave Observatory sites in Livingston, Louisiana and Hanford, Washington state. But he also worked on the Superconducting Super Collider in Waxahachie, Texas, a project that Congress killed in 1993 after construction had already started. "If the astronomy community spoke with unanimity and rallied around one telescope," Sanders says with the voice of experience, "the case would be clearer."


But the deep historical competition between American astronomers may end up working to everyone's benefit. The number of Gordon Moores in the world — billionaires with an eye to the heavens — is clearly finite. But Sanders remembers a time in the 1980s when astronomers thought that there was only enough money for a single 8-metre-class telescope in the world. In the end, astronomers found enough money to build a dozen.

And so if history is any guide, the future of astronomy may end up looking like its past. In 1996, Schmidt, the former Palomar director, gave an interview to a historian, and marvelled at how the Carnegie–Caltech divorce seemed to lead directly to the Keck and the Magellan telescopes, record setters in their time. "And I cannot believe that if we'd stayed together we would have had four of the largest telescopes in the world together." 

Eric Hand covers physics and astronomy from Nature 's Washington DC office.

Commenting is now closed.