2626 March 2014March 201 4 sky & telescopesky & telescopeThe furnace and mold keep spinning as the glass cools
rapidly to 650°°C. At that point, the glass becomes suffi -
ciently solid that further cooling must be done very care-
fully to avoid fracturing the glass; the furnace adds heat to
gradually cool the glass over the next three months.
Once the glass reaches room temperature, technicians
lift the furnace off , remove the top, sides, and fl oor of the
honeycomb mold, and apply a high-pressure water jet to
wash out the mold’s hexagonal columns. Finally, the mir-
ror blank is ready for polishing.Polishing the Mirror
Though GMT’s fi rst mirror was spun-cast in 2005, the
polishing that ground the mirror’s surface to its asym-
metric shape was not completed until seven years later.
That’s because even before polishing could begin, the
Mirror Lab had to design and build a whole range of new
equipment capable of measuring defects in the mirror’s
surface as small as 5 nanometers high.
The opticians at the Mirror Lab were keenly aware of
the spherical aberration that plagued the Hubble Space
Telescope, caused by a measuring error during polish-
ing. “Every concern that Hubble had, we have in spades,”
explains Martin. “We can’t rely on just one test.”
So, unlike Hubble, they typically have at least two
independent measuring devices check a mirror’s shape
during polishing. For the GMT’s unusually shaped mir-
rors, though, they decided to develop four tests. Each test
requires technology that is in some manner entirely new.
For example, all telescope mirrors are measured with a
null corrector, a test that aims a laser beam at the mirror’s
surface. Engineers compare the refl ection’s wavefront with
the wavefront that would be produced by the ideal mirror;
any disagreement between the two tells them where on
the mirror they need to polish. But the GMT null corrector
is 10 times larger than any built before, and the mirror it
must measure is severely asymmetric. To build the test, the
lab had to manufacture a 3.75-meter mirror, a major invest-
ment in itself, as well as a 28-meter-tall test tower to house
the mirror along with other testing equipment.
The team then designed a second scanning pentaprism
test, to guard against errors in the fi rst test. Also housed
high in the test tower, the device beams a laser through
a fi ve-sided prism to scan the mirror’s face. The laser
beam simulates starlight, with all rays parallel as they hit
the mirror. If the mirror’s curvature is correct, no matter
where the beam hits the mirror, its refl ection should focus
to the same point. A camera records the focused beam, and
any deviations it spots indicate polishing errors.
For the third test, the lab built another laser system
inside the test tower. This laser beam tracks the position of
a small retrorefl ector that sweeps across the mirror’s face.
The retrorefl ector’s three mirrors, arranged like the corner
of a cube, refl ect the laser beam back to its source, so the
laser tracker can measure the shape of the surface directly.SPIN CASTING 1. Two sample hexagonal columns are shown. The one
on the left has been cut open to show its interior and how it’s bolted into
place. The glass will melt and solidify around (not inside) these cores.- The crew places 19.2 tons of glass chunks on top of the ceramic mold.
- After a long day, they pose in front of their handiwork: a mirror ready
for casting. 4. The furnace lid lowers to enclose the mold. It will take
about a week to heat the glass to a maximum temperature of 1165ºC. - The furnace and mold together spin at a rate of 5 revolutions per min-
ute to give the mirror surface a preliminary parabolic shape.
ROBERT ZIMMERMANR. BERTRAM / STEWARD OBS. MIRROR LAB / UNIV. OF ARIZ. (2, 3, 4)➎
➊ ➋
➌
➍
PAT MCCARTHYMegatelescope MirrorsGMT Mirror.indd 26 12/24/13 11:45 AM