Both sites, or it didn’t happen

Folded to fit the fairing

Folded to fit the fairing

A mirror about half again as wide as the rocket's fairing could hold had to be creased, packed, and then opened by a sequence of deployments that left no room for a second attempt.

Technicians in clean-room suits inspect the hexagonal gold-coated mirror segments of a space telescope
Primary mirrorEach segment carries its own actuators, because the primary is only brought into figure after launch.
Photo: James Webb Space Telescope Mirror37 · Wikimedia Commons

The constraint that drove everything

The James Webb Space Telescope's primary mirror is 6.5 metres across. The Ariane 5 rocket that launched it on 25 December 2021 had a payload fairing — the nose-cone shroud — roughly 4.57 metres in internal diameter. The arithmetic was irreconcilable from the start, and it shaped every structural decision made over the following two decades of development.

NASA, ESA, and the Canadian Space Agency had committed to a mirror large enough to detect infrared light from the earliest galaxies. That aperture was not negotiable; it was the mission's scientific purpose. What was negotiable — what had to be engineered — was how to get something that size into a cylinder that couldn't contain it. The answer was origami at aerospace tolerances: fold it, launch it folded, and rely on a deployment sequence that engineers at Northrop Grumman eventually counted at 344 single-point failure steps.

The telescope had to be packed into a launch shroud and then unfold through several hundred steps, each of which had to work once.

A single-point failure is a step with no redundancy. If it does not complete, the mission fails. By comparison, most spacecraft are designed to survive the failure of any single component. Webb was designed knowing that several hundred components each had to work the first time, in sequence, in the thermal environment of space, after surviving a rocket launch, and could never be touched again. No crewed servicing mission was planned — Webb orbits at L2, the second Lagrange point 1.5 million kilometres from Earth, four times beyond the reach of any crewed vehicle ever built.

A clean-room team in coveralls around a large segmented mirror
Clean roomClean-room work is done in teams because a mirror segment is never handled by one pair of hands.Photo: Euclid Spacecraft in Cleanroom (PIA25783) · Wikimedia Commons

What unfolding actually meant

The deployment sequence broke into two major phases. First came the sunshield; then the optics.

The sunshield is a five-layer structure of Kapton film, each sheet coated with aluminium and, on the two sun-facing layers, silicon. Fully deployed it is roughly 21 by 14 metres — about the footprint of a tennis court. In the fairing it was folded and gathered around the telescope like a crumpled wrapper, held in place by dozens of tension release mechanisms, membrane release devices, and motor-driven spreader bars. After launch it unfolded across twelve and a half days, beginning with the deployment of the solar array and antenna tower, and then running through pallet deployments, mid-boom extensions, and a painstaking tensioning sequence across all five layers. The geometry had to be correct because the shield's function is thermal: it holds the telescope side at around 40 kelvin while the sun-facing side reaches roughly 360 kelvin. The sensitivity of the infrared detectors demands that cold environment absolutely. A misaligned layer, or one that failed to tension properly, would have compromised the cold side and degraded science performance permanently.

The deployment in numbers

6.5 mprimary mirror diameter, fully deployed
4.57 mAriane 5 fairing internal diameter (the constraint that forced folding)
344number of single-point failure deployment steps
18hexagonal beryllium mirror segments
~21 × 14 msunshield footprint when deployed
~40 Koperating temperature of the telescope side of the sunshield
~360 Ktemperature of the sun-facing side
150 nm RMSwavefront error requirement
~60 nm RMSwavefront error achieved; approximately 2.5× better than requirement
14 daystotal time from launch to completed deployment

Then came the mirror. The 6.5-metre primary is made of 18 hexagonal segments, each about 1.32 metres across, machined from beryllium and coated with gold. Beryllium was chosen for its stiffness-to-weight ratio and its behaviour at cryogenic temperatures — it contracts predictably and very little. The 18 segments were arranged in two wing sections of three segments each, folded back against the central twelve during launch. After the sunshield was deployed, the secondary mirror support structure extended forward on its three booms. Then the primary mirror wings rotated outward and latched: left wing first, then right, each driven by motors and checked by position sensors. The physical latch — a mechanism that had to click into alignment at the end of each wing's rotation — was the step engineers had spent years qualifying, because it represented the last large irreversible commitment of the deployment.

A multi-layer thermal blanket under inspection lighting
SunshieldThe layers only work while the Sun, Earth and Moon stay on one side of them.

What followed was slower and finer. Each of the 18 mirror segments sits on a hexapod of six actuators that can move the segment in tip, tilt, and piston with nanometre precision. Phasing the 18 segments into a single coherent optical surface — making them act as one mirror rather than 18 small ones — took the commissioning team several weeks after deployment, using images of a single bright star to detect and correct the wavefront errors at each segment. The final aligned mirror, measured by the NIRCam instrument's wavefront sensing mode, achieved a wavefront error well below the mission requirement of 150 nanometres RMS. The published wavefront performance was closer to 60 nanometres RMS — better than the requirement by a factor of roughly two and a half.

What the number 344 actually implies

The 344-step count became the figure journalists reached for, and it understates the engineering rather than overstating it. Each of those steps involved not one action but a suite of monitoring conditions: motor current readings, encoder positions, limit switches, temperature sensors confirming the thermal environment was within bounds before the next step was authorised. Ground controllers at the Space Telescope Science Institute in Baltimore and at mission operations centres watched telemetry in real time and made go/no-go calls between steps. The sequence was not purely autonomous; it was human-supervised, though the commands themselves were uplinked and executed by the spacecraft.

Chronology

  1. 25 December 2021launch on Ariane 5 from Kourou, French Guiana
  2. 8 January 2022final deployment step completed
  3. Spring 2022mirror phasing and instrument commissioning
  4. July 2022first public science images released

The reason no abort or reversal was possible on most steps is structural: deployment mechanisms are irreversible by design. A hinge that releases cannot re-latch from outside; a membrane that unrolls cannot be re-stowed. The engineering philosophy was therefore not to build in reversibility — which would have added mass and complexity — but to qualify each mechanism so thoroughly on the ground that reversal would never be needed. Testing included thermal vacuum runs simulating the space environment, acoustic testing to replicate launch vibration loads, and repeated actuation of every mechanism well beyond the single cycle it would perform in flight.

Cryogenic transfer lines wrapped in insulation running along a tunnel wall
CryogenicsInsulated lines carry helium to the magnets. Frost at a joint is a maintenance signal, not a fault.

That philosophy was tested most visibly with the sunshield membranes. The Kapton layers are thin enough to tear, and ground testing of the full-scale assembly required a clean room the size of an aircraft hangar and an elaborate system of offload cables to simulate zero gravity without introducing false loads. Even so, engineers acknowledged before launch that the ground test environment was an approximation, and that some behaviours of the membrane in true zero gravity could not be perfectly predicted. The deployment worked. But the margin between success and mission loss during those twelve and a half days was never comfortable, and the team's public communications during the period reflected that accurately.

Webb completed its full deployment on 8 January 2022, fourteen days after launch. Science operations began in earnest after several more months of commissioning and instrument calibration. The first science images were released in July 2022. The telescope was operating at its design wavelength range, at its design temperature, with its mirror performing above specification — and none of the 344 steps had failed.

Read next