How Is a Telescope Mirror Made? Grinding Glass to a Millionth of an Inch

A reflecting telescope works by bringing light from the whole mirror to a single focus, and light arriving from different parts of the surface must stay in step to within a small fraction of a wavelength or the image degrades. The usual working criterion allows a surface error of about one-eighth of a wavelength, which for visible light is roughly seventy nanometres, and better mirrors do considerably better than that. Scaled up, a four-metre mirror held to that tolerance is equivalent to a surface the size of Britain deviating nowhere by more than a couple of centimetres. The required shape is a paraboloid rather than a sphere, because a sphere focuses light from different radii at slightly different points, and the difference between the two surfaces at the edge of a large mirror is a substantial fraction of a millimetre, so the whole task is to remove exactly that much glass from exactly the right places. Infrared and radio instruments tolerate proportionally larger errors because their wavelengths are longer, which is why radio dishes can be made of mesh.

The sequence has been broadly the same since the nineteenth century, with the scale and the metrology transformed:

• •Casting the blank, historically in plate glass and now in low-expansion ceramics or borosilicate, with very slow cooling over months to avoid internal stress
• •Spin casting for some large blanks, rotating the furnace so the molten glass forms an approximate parabola under rotation and far less material needs removing
• •Rough grinding with coarse abrasive between the blank and a tool, generating an approximate spherical curve
• •Fine grinding through successively finer abrasives, each removing the pits left by the last
• •Polishing with a pitch lap and a fine compound, which is a chemical and mechanical process rather than simple abrasion and which produces the optical surface
• •Figuring, the final selective polishing guided by measurement, removing tiny amounts from specific zones to correct the surface towards a true paraboloid
• •Coating, usually by depositing a thin aluminium layer in a vacuum chamber, which is what actually reflects and which must be renewed every few years

Nothing can be corrected that cannot be measured, and the measurement techniques are as important as the grinding. The classical method places a point source at the mirror's centre of curvature and uses a knife edge to cut the returning beam, producing a shadow pattern across the surface that reveals which zones are high or low, a test devised by Foucault in the 1850s that remains in use by amateur makers because it is sensitive, cheap and entirely visual. Modern production uses interferometry, in which light reflected from the mirror is combined with a reference beam and the resulting interference fringes map the surface directly with sub-wavelength precision, and computer-generated holograms provide reference surfaces for aspherical shapes that no simple optic can supply. Measurement must be performed in a thermally stable environment, because air currents and temperature gradients distort the result, and large mirrors are tested in towers with careful control. Testing an off-axis segment is harder still, and an error in a null-testing arrangement is what produced the famous defect in the Hubble Space Telescope's mirror, which was polished with great precision to the wrong shape.

Monolithic mirrors reach a practical limit because a thick disc sags under its own weight and takes far too long to reach thermal equilibrium, so the field solved the problem twice. Honeycomb casting removes most of the interior mass, leaving a stiff ribbed structure that is light and thermally responsive. Thin meniscus mirrors go further, making the glass deliberately too flexible to hold its shape and supporting it on hundreds of computer-controlled actuators that push and pull continuously to maintain the figure, which is called active optics and is why an eight-metre mirror can be twenty centimetres thick. Segmentation divides the aperture into many smaller hexagons aligned to nanometre precision, which is how the Keck telescopes and the James Webb Space Telescope achieve their apertures and how the extremely large telescopes now under construction reach tens of metres. Adaptive optics is a separate and complementary technology, correcting atmospheric distortion hundreds of times a second using a small deformable mirror in the light path and a reference star, sometimes created artificially by a laser exciting sodium in the upper atmosphere.

The surface must hold its shape to around seventy nanometres, which scaled up is a country flat to a couple of centimetres. Grinding moves from coarse abrasive to pitch polishing to selective figuring guided by measurement, and the knife-edge test from the 1850s still works. Large mirrors abandoned rigidity, either by honeycombing the back or by making the glass flexible and holding its shape with hundreds of actuators.