The James Webb Space Telescope Mission: Optical Telescope Element Design, Development, and Performance

Highly complex integrated modeling was a key enabling capacity throughout all stages of development. The optical performance estimates made use of ground test data, integrated models, and simulations including the uncertainties in wave front sensing and control to verify the pre-launch requirements (Figure 5). Component design and ground test performance was used to provide predictions for optical alignments, component-level wave front error, and ground-to-flight effects. The dynamic components of the error budget used test data as inputs to an extensive structural-thermal-optical (STOP) integrated modeling process that predicted wave front stability (Knight et al. 2012) and line-of-sight image motion (Johnston et al. 2004). The integrated modeling for telescope performance made use of models of the structure, deployed thermal, thermal distortion, optical performance, dynamics and attitude control, and stray light. Each model was validated upon test data and conservative model uncertainty factors were applied to bound the worst case performance. Image motion predictions made use of exported vibrations and a model of the deployed dynamics of the observatory. The telescope thermal distortion and pointing stability following a worst-case slew used a thermal model that balanced the steady state at the hot attitude (pitched toward the Sun) and the cold attitude (pitched away from the Sun) with a worst case roll. The small temperature changes, less than 15 mK, from those thermal extremes was predicted for each of the thermal nodes on the telescope and used to determine the mechanical displacements on the structure. The repository of test data and integrated modeling results were used as inputs to the Integrated Telescope Model (ITM) simulator developed by Ball Aerospace in order to predict the optical performance, simulate data products for the development of analysis tools, and to rehearse the telescope alignment process (Knight et al. 2012).

Zoom In Zoom Out Reset image size

The same optical models were also used to inform pre-flight modeling of point spread functions (in particular using the software package WebbPSF, Perrin et al. 2014), which were used extensively in science planning, in development of proposal planning tools such as the exposure time calculator, and in development of data analyses pipelines. The core Fourier optical simulations of PSFs were augmented over time to become part of comprehensive high-fidelity data simulators, such as MIRAGE for NIRCam, NIRISS, and FGS data (Chambers et al. 2019; Hilbert et al. 2022) and MIRISim for MIRI (Klaassen et al. 2021). These, along with ITM, became critical enabling tools for the long campaign of preflight rehearsals of the WFSC alignment process (Section 4.2.3).

Low areal density mirrors were recognized as a key technology gap to enabling a ∼25 m² aperture space telescope. The areal density of the Hubble primary mirror is 240 kg m⁻², while JWST's objective was < 26.5 kg m⁻², which was achieved. A mirror technology development program was convened to evaluate and advance mirror technologies through multiple programs, including the subscale Beryllium mirror demonstrator (SBMD, Reed et al. 2001) and the advanced mirror system demonstrator (AMSD, Stahl et al. 2004). The AMSD program evaluated ULE and Beryllium with a wide variety of parameters such as the optical performances achieved, control authority, mounting, and fabrication schedule. While ULE was deemed to have programmatic advantages, it was found to have an astigmatism as it cooled that was non-deterministic and would have added uncertainty to the development. The decision to select O30 Beryllium, a more isotropic form of Beryllium not previously used in space telescopes, was made following the recommendation from the Mirror Recommendation Board. Beryllium was selected largely due to its small coefficient of thermal expansion (CTE) within the telescope's operational temperature range, making it particularly advantageous during the cryo-polishing fabrication process and achieving the telescope's passive stability objectives by not requiring active thermal control. Beryllium is also light weight, advantageous given the very tight mass constraints for the telescope (Feinberg et al. 2012). Beryllium mirrors have flight heritage from previous space missions, including the Spitzer, the Infrared Astronomical Satellite (IRAS), and the Cosmic Background Explorer (COBE).

Active control of primary mirror segment position was achieved using actuators mounted in a hexapod arrangement, plus a center actuator for active control of radius of curvature (see Figure 6). Specialized actuator mechanisms were developed specifically for JWST in order to enable the active positioning of the large segmented mirrors and to support the mirrors during ground test and the launch environment. Each mechanism makes use of a fine stage flexure and coarse drive coupling to control the linear displacement (Warden 2006). The actuators themselves have remarkable performance parameters including a fine step size of 7.7 nm resolution, with 2 nm of fine repeatability, over a 10 μm fine range. A coarse drive coupling in the same mechanism provides a 58 nm step size over a full 21 mm. Further, unlike ground telescope active and adaptive optics, which often use electrostatic or piezoelectric actuators, JWST's actuators operate mechanically via a gear train and flexures; the mechanical gear trains hold position stiffly and precisely even when the actuator is entirely unpowered, which is necessary to avoid undesired waste heat into the cryogenic telescope.

Zoom In Zoom Out Reset image size

The primary mirror segment development process required new facilitization and made use of economies of scale to fabricate the multiple segments in parallel. The mirror blanks were made from O30 Beryllium through a hot isostatic pressing process by Brush Wellman. The blanks were then light weighted by removing over 92% of the material in a honeycomb shape by Axsys Technologies. Next, the mirror substrates were polished by Tinsley Labs and each mounted to its flexure and radius of curvature system. The mirrors were then optically tested at ambient and cryogenic temperatures at the X-ray and Cryogenic Facility (XRCF) at NASA's Marshall Space Flight Center, followed by another round of cryo-polishing to ensure each mirror achieved the correct optical figure at the intended cryogenic operating temperature (Cole et al. 2006). The mirrors were gold coated using a vacuum vapor deposition process by QCI, Inc. The gold coating provides high reflectivity across the operational wavelength range of 0.6–28.1 μm (Keski-Kuha et al. 2012). A protective SiO_x overcoat was applied that improved the durability of the coating and enabled cleaning at stages throughout the I&T process (see Lobmeyer & Carey 2018). Finally, flight acceptance testing for each segment was carried out in the XRCF facility (see Figure 7).

Zoom In Zoom Out Reset image size

A large precision cryogenic structure was necessary to enable the passive optical stability without active control. The telescope and the science instruments are supported by a composite optomechanical structure that must withstand the launch environment loads, deploy within the capture range of the mirror actuators, survive the stresses induced from cooling down to cryogenic temperatures, and have minimal thermal distortion.

Early in the JWST development, it was recognized that the materials database for composite structures was inadequate over the operational temperature range, the ability to measure deformations was inadequate for the JWST verification testing, the engineering modeling tools needed development, and manufacturing process controls needed improvement. The coefficient of thermal expansion for the materials needed to be measured to less than 30 ppb K⁻¹ at temperatures of < 25 K, which was more than 100 times more precise than the state of the art at the time (Atkinson et al. 2007). ATK implemented a technique to measure the CTE for large structures with the precision needed for JWST. With the materials characterized, a prescription for the composite structure was defined that used unidirectional prepreg made from M55J carbon fibers and resins into laminant mixtures tuned for the appropriate strength and thermal performance. Manufacturing controls were established to precisely align the fibers during layup and closely manage the fiber to resin ratio necessary for precise material properties. Controls were also put in place to achieve the desired bonded joint adhesive thickness used to connect the individual tubes into a truss. A prototype of the composite structure, called the Backplane Stability Test Article (BSTA), was built by ATK and tested at MSFC's XRCF (Figure 8). Verification of the structure's stability made use of a new Electronic Speckle Pattern Interferometer (ESPI) metrology approach (Saif et al. 2008), a technology development in itself, to confirm the structure was TRL-6.

Zoom In Zoom Out Reset image size

The image-based phase retrieval methods used with JWST have a heritage stretching back to the diagnosis and correction of the infamous spherical aberration in the Hubble primary mirror (Fienup et al. 1993; Krist & Burrows 1995). Phase retrieval using the science instruments elegantly avoids the need for substantial dedicated wave front sensing hardware, and ensures sensing of wave fronts directly at the science focal planes. However, to accommodate the evolving alignment of the mirrors (from initial deployment errors measured in millimeters to final alignments measured in nanometers) these methods must operate over a tremendous dynamic range, and must also sense dissimilar and degenerate degrees of freedom. As a result several distinct wave front sensing methods must be used. The primary tool is focus-diverse phase retrieval, using a hybrid diversity algorithm developed specifically for JWST (Dean et al. 2006). Focus diversity is provided at different stages by defocusing the secondary mirror or by using weak lenses within NIRCam that can be inserted into the beam path. The focus diversity method is augmented with dispersed Hartmann sensing for the measurement of segment piston (Shi et al. 2004).

The step-by-step sequence of sensing and control activities, as well as the associated algorithms and software, were developed at Ball Aerospace. To test and prove the implementation, a 1:6 scale model and functionally accurate Test Bed Telescope (TBT) was built (Kingsbury & Atcheson 2004, see Figure 9). Using the TBT, the complete end-to-end telescope alignment process was successfully demonstrated, achieving TRL 6 in 2006 (Acton et al. 2018; Feinberg et al. 2007).

Zoom In Zoom Out Reset image size

Even with those fundamental tools proven, a decade of work remained to mature them from lab-scale demonstration to flight-ready processes. Operational implementation of the commissioning plan was complicated by the need to begin operation of fine guidance control while still adjusting mirrors (see Section 5.2), and to interweave telescope alignment with prerequisite steps of instrument commissioning such as focal plane calibrations (see Section 5.3). The methods were refined and operational plans were prepared leading up to launch (Perrin et al. 2016), culminating in detailed implementation plans, procedures, and observing programs. Contingency plans were prepared in the event that nominal plans could not be followed, for many distinct contingency scenarios. Phase retrieval analysis software was similarly iteratively refined prior to launch; the initial delivery of flight WSS software occurred in 2011, and regular improvements continued thereafter as part of the I&T of the Science & and Operations Center (SOC).

These processes for WFSC were repeatedly tested together with the flight hardware at various stages of observatory I&T. The initial test and operation of the integrated OTE electronics and mechanisms took place in 2016 (see Section 4.3.1). During the cryovacuum test of the telescope plus instrument suite (Section 4.3.4) while the majority of that test program used GSE for metrology of observatory alignments, specific activities were included to test the flight scripts for wave front sensing and control using flight hardware. This was the first and only time that NIRCam was used to sense OTE mirror alignments on the ground (Acton et al. 2018). In parallel, the data generated by that activity was flowed back to STScI and used for a demonstration of sensing and control software processes using the integrated SOC. This was the first major demonstration of processing of JWST data in a flight-like manner at the mission operations center (Lajoie et al. 2018). The commanding for mirror moves and deployments was repeatedly exercised as part of regular OTE functional checks, up to and including at the launch site.

A necessary input to the WFSC process was accurate knowledge of instrument-specific wave front errors, to allow subtracting the instrument contributions from the results of the image-based phase retrieval to perform OTE wave front control. This objective was met through precise measurements at dozens of field points within all instruments, as part of instrument cryo–vacuum testing completed by 2016.

The wave front sensing and control activities demanded human-in-the-loop controls and required training a large wave front team for round the clock operations during commissioning. The telescope was aligned start-to-finish over a hundred times in simulation, individually by many members of the team and in collaboration. As part of this training process, there were iterative refinements of the methods, procedures, and documentation. The individual simulations built toward larger team rehearsals, including 20 internal wave front team practices and 25 mission operations or science operations team wide rehearsals. Many of these rehearsals were carried out throughout the COVID-19 pandemic under work-from-home conditions, remotely. The extensive rehearsal program was a critical, invaluable activity in building a smooth-functioning cohesive wave front team combining staff from multiple organizations and skillsets.

The major structural elements of the OTE (see Figure 2), as well as the associated electrical harnesses, were integrated and tested at Northrop in 2014–2015. Major tests included load testing of the mirror backplane and testing of the Deployable Tower Assembly (DTA). After precision integration of the DTA, the two telescope wing structures, and the Secondary Mirror Support Structure (SMSS) to the backplane, these subsystems were exercised for functionality and repeatability with ambient deployments (Glassman et al. 2016). Modal surveys were also carried out in the stowed and deployed configurations, with appropriate mass simulators for hardware to be integrated later.

After shipping of this hardware to GSFC, the integration of additional harnesses and small electronics boxes (e.g., for the mirror actuators) and the OTE optics took place in a dedicated assembly and alignment facility in GSFC's largest cleanroom. The PMSA shim prescription was determined using metrology from a coordinate measuring machine brought from Tinsley Labs, now Coherent Inc., and laser tracker measurements of the composite backplane structure. The PM segments were mounted to the backplane with the assistance of a traveling robot arm, as shown in Figure 11. Laser trackers measured the alignment state to guide the installation, with laser radar independently measuring. Custom ground shims and adhesive-filled pin gaps secured the location to mechanical tolerances that were a small fraction of the range budget for the PM actuators (see Atkinson et al. 2016 for details).

Zoom In Zoom Out Reset image size

Integration of the Secondary Mirror Assembly and the Aft Optics Subsystem (AOS) completed the OTE. With the integration of the Fixed and Aft Deployable ISIM Radiators and the ISIM itself (the instrument module and the associated electronics compartment), and an array of thermal and stray light control blankets (>900!), the OTIS was complete. The ISIM of course had been through its own comprehensive I&T program, including three cryo–vacuum tests totaling nearly 300 days of round-the-clock operations (Kimble et al. 2016). This integration of the OTE + ISIM to form the OTIS took place in 2016 May.

Using the fully assembled OTIS, the integrated mirror control system hardware and software was exercised, first in small steps starting in 2016 October, and eventually in partial deployments of all 18 PMSAs and the SM in preparation for ambient optical testing. This included operation of PMSA actuators and sensors (resolvers and linear variable differential transducer [LVDT] length sensors), operated by the Actuator Drive Unit electronics and controlled by the Wave front Sensing System software. One minor anomaly discovered during this time is that a small number of LVDT sensors do not operate nominally. This was accepted to "use as is," given the availability of other telemetry to confirm mirror motions for those segments (e.g., using resolver telemetry). For some of the affected LVDT sensors, a modified operations concept was developed that used temperature-dependent calibrations to make the sensor information usable, later used successfully in flight. Mirror control mechanisms and processes continued to be exercised throughout the remainder of OTIS I&T, in particular during the OTIS cryovacuum test.

The OTIS underwent proto-flight level ambient environmental testing (vibration and acoustics) at GSFC in 2016–2017. Ambient optical measurements were carried out before and after those mechanical tests utilizing a Center of Curvature test setup, including a null lens and Computer Generated Hologram (CGH, for working with the aspheric PM surface), as illustrated in Figure 12; see Saif et al. (2017). The Center of Curvature setup incorporated a high-speed interferometer for making figure measurements at rates up to 5.9 kHz. With this equipment, the static wave front of the PM segments was measured before and after the mechanical environmental testing, along with the dynamic response of the backplane and mirror mounts. With a vibration stinger to excite the payload, mechanical transfer functions were measured to look for any signs of damage after the vibe and acoustics tests e.g., cracks in the structure; loosening of joints. No such damage was seen, with the figure and dynamic measurements repeating pre- and post-test within expected tolerances. Electrical functional checks were also carried out before and after the mechanical environmental testing, along with "first motion" (flinch) tests of deployment systems that could not be fully deployed at GSFC in the one-g environment.

Zoom In Zoom Out Reset image size

An anomaly did arise during the OTIS-level vibration test, when a loud "bang" was heard. This was determined to have been caused by gapping at one of the preloaded interfaces of a Launch Release Mechanism in a PM mirror wing. Modifications to the procedures to properly set and maintain the preload of these interfaces resolved this issue, and no damage had been done. In addition, excessive resonant response was seen for the SMSS and the AOS at some frequencies due to the low damping of the large composite structure. The test vibration spectrum was notched at these frequencies to protect the hardware during the OTIS-level test. Particle dampers were subsequently designed and installed onto the AOS and SMSS to reduce these responses; they operated successfully later at Observatory-level testing and ultimately through the actual launch.

The final phase of I&T for the OTIS was an extraordinarily challenging cryo–vacuum test, previously described in Feinberg et al. (2011) and Kimble et al. (2018). This took place at NASA's Johnson Space Center in historic Chamber A, which is a US national historic landmark from the Apollo program. After shipment to JSC, the OTIS underwent electrical functional testing, the SMSS was deployed (with assistance—the necessary GSE for a powered deployment in one-g was only at Northrop), and the DTA and PM wings were deployed.

After these activities, the payload was configured for the cryo–vacuum test in Chamber A, which had been extensively refurbished for the thermal and contamination requirements of JWST and outfitted with optical and thermal GSE comprising an elegant test architecture, illustrated in Figure 13. This architecture supported a rich array of operational, thermal, and optical test goals, with 40 separate tests. End-to-end optical tests were carried out using sub-apertures of the primary mirror. The test campaign applied lessons learned from the Hubble program.

Zoom In Zoom Out Reset image size

Key components of the optical test equipment included the Center of Curvature Optical Assembly (COCOA, Wells et al. 2010), whose interferometers had a view of the entire PM; photogrammetry (PG) cameras on four rotating booms (Lunt et al. 2020), which provided remarkably accurate relative positions (sub-100 μm) of optical targets over the many-meter distances involved, through image triangulation; the AOS Source Plate Assembly (ASPA), which mounted light sources (fiber-fed or local) at the intermediate Cassegrain focus of the OTE - these provided downward (half-pass) images through the TM, FSM, and SIs and upward (pass-and-a-half images) through the SM, PM and then, after reflection off sub-aperture Auto-Collimating Flats (ACFs), back through the entire OTIS optical train; and fiducial light strips straddling the edges of the PM. The position of the ASPA sources made their images highly aberrated, but in a precisely known way, so that alignments and OTIS wave front measurements could be extracted nonetheless. The downward sources were used to test the guiding control loop.

The optical test equipment (see Figure 14) worked together to satisfy the critical optical verification goals (such as the verification of the non-adjustable AOS to ISIM alignment, verification of the radius of curvature) as well as various cross-check goals. A succinct description of the process is as follows:

• 1.  
  PM segments were aligned and phased via photogrammetry and COCOA interferometry.
• 2.  
  The SM was aligned via photogrammetry and checked with Pass-and-a-Half imaging to the NIRCam instrument.
• 3.  
  AOS to ISIM alignment was verified via Half-Pass imaging using inward facing sources and all science instruments.
• 4.  
  Fiducial lights above the primary mirror were used for verifying pupil alignment, using NIRCam's pupil imaging capability.
• 5.  
  End-to-end imaging and field tilt was cross-checked using Pass-and-a-Half imaging using outward facing sources, the autocollimating flats, and all of the science instruments.
• 6.  
  Wave front Sensing & Control hardware checks and demonstrations were performed via Pass-and-a-Half testing and NIRCam.

Zoom In Zoom Out Reset image size

The Pathfinder Program used spare and test equipment as surrogates for the flight hardware in order to prove out many of the integration and testing activities (see Feinberg et al. 2014, Section 7.3). The test article, referred to as the "Pathfinder," was comprised of two spare primary mirror segments, a spare secondary mirror, and composite structure representative of the center section and the secondary mirror supports. The Pathfinder was transported and integrated using the protocols and procedures that would later be used on the flight hardware, in some cases demonstrating the actual capabilities of the test equipment.

Following integration, the Pathfinder was used to prepare the cryovacuum testing facility and equipment at JSC in 2015/2016, while the OTE and OTIS were being integrated and tested at Goddard. The Pathfinder cryotest program utilized the thermal and optical GSE developed for the OTIS test and the Pathfinder structure itself (Matthews et al. 2015a). The first Optical GSE test (OGSE1) checked out the COCOA and PG operations with those systems, while the second (OGSE2), incorporated the flight AOS (hence requiring careful coordination with the flight I&T flow) and the ASPA to dry-run the half-pass and pass-and-a-half tests as well. A third Pathfinder test included thermal mock-ups of the remaining center-section PM segments and validated the cooldown and warmup procedures that would be needed, including the requisite contamination control (Matthews et al. 2015b).

Cryo–Vacuum testing of the flight OTIS took place in 2017. The 100 days, round-the-clock test campaign was remarkably successful, despite the many challenges, including the passage of Hurricane Harvey through the JSC area in the middle of the test, which devastated the local community and shut down JSC for normal operations with 55'' of rainfall. After a 5 days period of special hurricane operations, and nearly running out of liquid nitrogen, the cryotest narrowly missed being abruptly ended. Fortunately the hurricane did not preclude safely continuing the test. All significant goals for the test were achieved, including confirming the health of the OTIS payload after its environmental test program (Wolf et al. 2018), accomplishing the planned optical verifications and cross-checks (e.g., Hadaway et al. 2018), validating the OTIS thermal model and the OTIS thermal distortion model, both required for the Integrated Modeling of the observatory as a whole, and various operational validations and demonstrations.

Performance of the OTIS was overall excellent, with predictions that satisfy the mission-level requirements (Lightsey et al. 2018). But, there were several types of optical instabilities identified. One, caused by over-tightness of the soft-structure frill stray light blocker and PMSA closeouts at the cryogenic operating temperatures (such that they exerted temperature-dependent forces on the mirror backplane), was mitigated with post-test modifications to restore the intended slack where feasible. A second, which coupled temperature variations in heater-controlled radiator panels on the instrument electronics compartment (IEC) to the mirror backplane, inducing oscillating structural distortions, was demonstrated with ambient measurements and analysis to be caused by a rigid non-flight mounting of the IEC for the OTIS CV test. Both of these have been shown to have minimal wave front impact in flight (see Section 5.4.4).

A final instability, called "tilt events" referred to sudden, stochastic changes in the piston/tip/tilt pose of individual PM segments. Several such events were seen throughout the OTIS test period. Though not fully explained, these were ascribed to stick/slip release of stresses from cooldown thermal deformation in the OTIS structure. These reduced during the end of the test, and it was expected that these would fade away with time after cooldown as the various stresses in the system were gradually relieved. This behavior during the OTIS test informed expectations that such events could be seen in flight. This supposition appears to be confirmed by the flight behavior (see Section 6.2.1), though with the excellent sensitivity of the flight wave front sensing, most of the observed tilt events in flight are actually below the detection threshold of the OTIS cryo-vac analyses.

After the deconfiguration of the OTIS from the test configuration, it was shipped to prime contractor Northrop Grumman's Space Park in Redondo Beach, California. While at Northrop, powered deployment of the SMSS and DTA took place with appropriate one-g off-loading hardware. Both deployments were made from the spacecraft electronics to demonstrate the connections and scripts were working properly. For the SMSS, this represented the only post-OTIS-vibration powered deployment, confirming the health of the deployment system after that proto-flight-level vibration exposure.

In the summer of 2019, the OTIS was integrated with the spacecraft and sunshield to form the full-up JWST observatory. Alignment metrology was performed in the integrated configuration to characterize the OTE to star tracker boresights.

In the full-up observatory configuration, the payload underwent various deployment tests (e.g., off-loaded deployment of the DTA and PM wings). The observatory as a whole was then put through acceptance-level vibration and acoustic tests, with subsequent deployment and electrical functional tests. Like the OTIS, the spacecraft and sunshield had previously successfully undergone mechanical environmental testing at proto-flight levels.

Both the PM and the SM were cleaned of particulates at appropriate times in the Northrop flow, with a gentle brush technique described by Lobmeyer & Carey (2018). For the SM, which had the most challenging particulate contamination budget (so cleaning was desired as late as possible), this cleaning took place after the final stowing of the observatory into its transport (and launch) configuration, just before encapsulation of the observatory into a clean, environmentally controlled shipping container for transport by sea to the Guiana Space Centre (GSC) in Kourou, French Guiana. There, the observatory executed final ground functional tests, was fueled, and was encapsulated in the Ariane 5 rocket fairing for launch.

The major structural deployments of the OTE (DTA, SMSS, and PMBSS wing deployments) all completed successfully and nominally, with no notable issues.

Subsequent results from telescope commissioning confirmed the precision of these deployments: for instance, the 8 m multi-hinged SMSS deployment placed the SM within 1.5 mm of its nominal position, well within the correction range of the SM actuators. Further, the telescope boresight offset relative to the spacecraft star trackers was found to be 34, nicely consistent with the 305 1σ preflight prediction. Similarly the corrections required to align the PMSAs were small (Table 4), with only one segment requiring a corrective move larger than 1 mm in position.

For launch, all mirror segments were stowed in launch restraints in order to limit lateral displacements during launch and ascent (see Figure 16). To begin to align the PMSAs and SMA, the segments therefore had to be released from their launch restraint, a pure piston move of ∼12.5 mm.

Zoom In Zoom Out Reset image size

The deployment sequence of the PMSA and SMA was carefully designed to verify the actuator stepper motor aliveness and responsiveness, as well as confirm proper management of the segment envelope and workspace boundaries. The sequence was tested on the ground on multiple occasions (e.g., OTIS testing at JSC). As such, stepper motors on segment A1 only were first incrementally commanded to move 1 step, 1 revolution, 10 μm, 150 μm, and 340 μm. The remaining segments (except A3 and A6, see below) were then commanded the same sequence of steps, followed by 1-mm increments until all the segments were fully deployed to 12.5 mm. Early in these deployments, some LVDT sensor readings did not show as smooth a progression as expected, which led to additional small flinch moves to verify all actuators were moving. Once past the first few millimeters and out of the launch restraints, the LVDTs showed the expected linear response. The initial non-linearity was interpreted as due to friction of surface contacts with the launch restraints, which, however, posed no problem to the deployments.

Segments A3 and A6 were deployed separately and last as a result of a faulty LVDT on one actuator each (as noted above in Section 4.3.1). Although the sequence of moves was identical to that of the other segments, the LVDT readings, which provide a coarse direct measurement of actuator length, had to be calibrated as the sensors cooled down. As a result, they were deployed separately, without incident.

Finally, once at their deployed positions, the segments were commanded to their intended nominal positions, based on ground alignment test results corrected for 0-g via modeling. Mirror deployments completed successfully and with no major issues.

The next series of activities aimed at finding, identifying, and re-arranging the images produced by each of the mirror segments. The deployed but not-yet-aligned segments were each acting as its own aberrated ∼1.3 m telescope. When first light on NIRCam was obtained on 2022 February 2nd, all celestial objects were indeed duplicated 18 times (see Figure 17).

Zoom In Zoom Out Reset image size

To identify an image of each segment, an isolated bright star was observed. Budgeted pre-flight uncertainties for mirror segment initial deployments, as well as in the initial telescope boresight offset (see Section 5.3.1), predicted that segments might be scattered by up to ∼15'. As a result, the target star was selected to have no similar-brightness neighbors within such a distance, and a large half-degree-diameter mosaic around the target star was generated, taking around 25 hr to complete. Figure 18 shows a cartoon of the planned mosaic, overlaid on top of a catalog sky image around the target. A subset of the flight data where the segment images were all found is also shown. The segment images were found within ∼34 of the nominal target location on average and the segment image scatter was of similar magnitude, both better than requirements and expectations. The simplicity of the plan for this initial mosaic step proved beneficial to accommodate larger-than-expected coarse pointing uncertainties at this time (prior to ACS tuning and optimization), and increased levels of detector persistence due to operating NIRCam well before it had fully cooled to its nominal temperature.

Zoom In Zoom Out Reset image size

Following the initial mosaic observation, a secondary mirror focus sweep was performed in order to measure and move the SM to a best focus position. This improved image quality and enabled guiding later on during OTE commissioning (see Section 5.2.1). The analysis led to an SM move of −427 μm (i.e., away from the primary mirror).

Next, each mirror segment's image in the initial mosaic was identified by sequentially tilting each mirror. Once identified, segments were commanded to form the pre-defined image array shown in Figure 15.

In the hexagonal array configuration, all segment images could be observed on one NIRCam detector at once and segment-level aberrations could be addressed as part of the segment global alignment activities. To do so, the SM was moved away from its nominal best focus by ± 400 μm to collect focus-diverse imagery for the purpose of phase retrieval analysis. As a result of this analysis, the SM was corrected in X- and Y-translation by 0.94 and 1.06 mm, respectively. Focus corrections were applied to each segment, along with corrections to the radius-of-curvature actuators of two PMSAs.

During a later stage of commissioning, a second iteration of Global Alignment was executed as part of the iterative alignment approach. At that time, additional small clocking, radial translations, and radius of curvature corrections were applied to the PMSAs to correct astigmatism and power. At this point in the OTE phasing process, the observed wave front errors achieved excellent agreement with ground measurements and preflight modeling of the higher spatial frequency mirror maps (Figure 19).

Zoom In Zoom Out Reset image size

Following each instance of Global Alignment, a sequence of image stacking was executed to position segment images on top of each other (see Figure 15). However, this stacking does not mean the light paths from each segment are in phase with one another, so Dispersed Hartmann Sensing measurements (Coarse Phasing) were executed to establish phase errors as a function of wavelength in order to measure pairwise segment edge heights and derive an overall piston correction to all of the PMSAs. Three iterations of this coarse phasing sufficed to bring the PMSA piston offsets to less than 1 μm, where fine phasing with NIRCam weak lenses could finalize the OTE alignment.

Following these PMSA piston corrections, a fine phasing activity took place where the NIRCam weak lenses were used to collect focus-diverse measurements for phase retrieval analysis (see Figure 15). The results of such measurements were then used to better (re-)stack the segment images as well as correct residual piston offsets between segments.

On 2022 March 11, the fine alignment process completed, yielding a telescope aligned to roughly 50 nm rms as seen at the fiducial field point on NIRCam A3. Following this, a mosaic observation was carried out around the alignment star. This multi-purpose observation tested science-like dithered and mosaiced observations for the first time, confirmed excellent PSF quality over all of NIRCam's field of view and across NIRCam's full wavelength range, provided early measurements of observatory backgrounds, and yielded an early glimpse of JWST's sensitivity to the high redshift universe.

Aligning the telescope to only one field can lead to degenerate solutions, where PMSA and SM misalignments balance each other out. Multi-field measurements are therefore required in order to achieve optimal optical performances across all the science instruments.

Two types of multi-field sensing were carried out: a NIRCam-only multi-field activity (using only Module A) and two instances of multi-field, multi-instrument activities. The NIRCam-only alignment activity was aimed at initially removing most of the PMSA-to-SM misalignment using the unstacked segments as a Hartmann sensor, analyzed with a centroid-based approach to measure the field dependence of aberrations. This "coarse" multi-field sensing proved to work exceptionally well. As a result of this analysis, the SM was moved in translation in X and Y (−210 μm, 420 μm) as well as tilt in X and Y (−550 μrad, 34 μrad). The SM correction was also accompanied by compensating PMSA moves in order to maintain the hexagonal array configuration.

Finally, multi-field, multi-instrument measurements were made in order to assess the field dependence over the whole field of view, this time using focus diversity provided by moving the SM in piston by ±100 μm. Two instances of this activity were executed since, as expected, MIRI had not yet reached its operational temperature at the time of the first measurements. In both instances, the results indicated no significant correctable field-dependent aberrations, in other words the initial NIRCam-only multi-field activity had corrected the telescopes field dependent aberrations. Only a small SM focus correction with minimal wave front error gain was applied, mostly to balance the relative focus terms of the science instruments (see Figure 20). The observatory was fully aligned on 2022 April 23, and the commissioning of the science instruments continued thereafter.

Zoom In Zoom Out Reset image size

From that point, the operations team entered a new stage of conducting routine OTE maintenance, which will be ongoing throughout the mission, discussed further in Section 6.

JWST uses star trackers, rate sensors, reaction wheels, and a fine steering mirror to achieve a coarse pointing. To transition into guiding, one of two FGSs will attempt to identify the intended guide star, whose position and fluxes are normally provided by the Guide Star Catalog. The coarse position error of the guide star as seen by FGS is fed back to ACS for correction. The guidestar will usually then be moved to a pre-computed "science" location in the FGS field of view, after which closed-loop guiding will be attempted and typically engaged. During closed-loop guiding, FGS measures centroids every 0.064 s and reports the position to ACS, which then commands the fine steering mirror to move in order to maintain the guide star at the appropriate position on the FGS detector.

Early in OTE commissioning, when the OTE was still providing 18 images for every star in the field of view (including on FGS), guiding operations had to be modified. Per our plan, FGS used one of the segment images to guide on, along with reference segment images. This was done by overriding the guide star selection system to account for the segment position offsets and flux differences compared to the guide stars catalog position and flux. Using this approach, closed-loop guiding was successfully demonstrated during the FGS LOS Initialization Activity (PID 1410).

Later in commissioning, guiding operations became increasingly routine; once the PMSAs were stacked into a single PSF, the guide and reference stars selection and locations could be automatically provided by the ground system using the operational catalog and only the associated stellar fluxes had to be overridden. Once the PSF was phased, the fluxes as well were being supplied by the system. By the end of OTE commissioning, guiding required no special intervention.

The majority of instances of closed-loop guiding during OTE commissioning were successful, although some failures do occur for a variety of reasons, including bad pixels or mis-cataloged guide stars. Nevertheless, success rates have risen continuously throughout commissioning and into science operations with now >95% of planned visits being successfully executed. When guiding success is achieved (i.e., the intended guidestar is identified, acquired, and tracked with a settled closed loop), then the image stability of that pointing becomes the performance metric of interest.

Special commissioning tests were included in the baseline plan to obtain the power spectrum of the line-of-sight (LOS) jitter and, in particular, to assess the contribution of vibrations caused by the MIRI cryocooler (CC).

The dedicated test to probe excitations from vibrations in the LOS data showed no evidence of significant contributors and has revealed excellent stability performance at ∼1 mas rms radial, very close to the measurement noise floor and significantly better than expectations and the 7 mas requirement. The typical results from the first commissioning LOS jitter analysis (PID 1163, observation 2, from Hartig 2022) are shown in Figure 21. In this case, jitter was measured at 1.04 mas rms (radial). The same tests also revealed no need to tune the MIRI cryocoolers pulse frequency which remains at its initial settings.

Zoom In Zoom Out Reset image size

The analyses, which continue to be regularly performed as part of the 2 days routine maintenance program (Section 6), have consistently shown jitter levels around 1 mas radial (moderately correlated with guidestar brightness). The analysis is quite sensitive and has revealed low frequency, low power oscillations at ∼0.3 and 0.04 Hz. The 0.3 Hz feature has been attributed to bending modes at the 1 Hz isolator at the SC-to-OTE interface, whereas the 0.04 Hz feature has been shown to vary over time and might be attributed to fuel slosh. During the commissioning period there were no clear indications of any LOS jitter response to the reaction wheel assemblies. However during the first months of science operations, a handful of measurements have shown minor LOS jitter contributions which appear correlated with certain speeds of reaction wheel assembly #6; a resonance appears to be excited in the vicinity of 16–17 Hz.

A key design feature of JWST is an OTE that is well separated and isolated, thermally and mechanically, from the spacecraft by the Deployable Tower Assembly (DTA). Uncertainties in the DTA deployment and other contributions were expected to produce initial errors in the nominal alignment of the OTE V-frame with respect to the spacecraft's fundamental coordinate system, called the J-frame, of ∼10–15'.

The slew of the observatory to its intended field is controlled by the Star Tracker Assemblies (STAs), which reside on the spacecraft side and are calibrated to the J-frame. However, the FGSs must then be able to acquire and identify the intended guide star, and they reside on the OTE side. Capturing the alignment between FGS and ACS requires the updating of the FGS-to-J direction cosine matrix (DCM). This matrix is updated to account for deployment uncertainties, as stated above, as well as changes in telescope boresight, which occur every time the SM is moved in translation or tilt (e.g., Global Alignment, MIMF).

While the initial error in the OTE-to-spacecraft alignment was expected to be close to 10', the misalignment was found to be only ∼34 from the nominal ground values (see Figure 18). The FGS-to-J DCM was subsequently manually updated during commissioning after taking observations of the sky with an SI or FGS and expressing that celestial pointing in terms of the FGS1 frame, while obtaining from the ACS/STAs the contemporaneous mapping of sky to the J-frame, thus providing the information to relate FGS1 to J-frame. This operation was also successfully performed during subsequent commissioning activities to maintain the alignment.

During the science mission, this relationship will continue to be dynamic at levels much lower than seen in commissioning (i.e., a few arcsec), and the ACS will autonomously update this calibration based on observed FGS guidestar location error.

Using FGS1 to define the relationship of the OTE to the spacecraft implies that it is also the reference for the OTE-based frame to which the SI fields of view are calibrated. This "V-frame" was defined nominally as a conventional 3-axis coordinate system aligned with the OTE principal mechanical axes, having V1 pointing out along the Cassegrain axis of symmetry (1). In flight, however, the V-frame is used to specify SI and FGS fields and various fiducial field points used for science targeting. So, in this application, it is treated spherically as angles, with the axes V2 and V3 corresponding to "field angles" within the OTE field.

In this scheme, the FGS1 field location and orientation with respect to V2,V3 is fixed, and on-sky astrometric calibrations that determine the SI's and FGS2's fields relative to FGS1 in essence establish their locations, orientations, and higher order distortions with respect to the V2,V3 field angles. This astrometric calibration scheme, its tools and products, are thoroughly treated by Sahlmann (2019a).

These calibrations are required for successful target placement and were performed during OTE commissioning to (1) determine the post-launch changes to the ground-determined relationships and (2) update this knowledge to ensure successful multi-SI wave front measurements. The commissioning team used a specially calibrated ∼15' astrometric region of the LMC for this purpose (Sahlmann 2019b).

The first measurements of SI relative locations showed the ISIM to be stable, with ground-to-flight evolution in the V2,V3 field angles to be at or below the ∼1'', and orientation changes of the SI fields <1'. Although precise scales and distortion calibrations fall into the SI activities and continue into the science cycles, basic instrument scales were tentatively measured during these OTE commissioning activities, and were found to be <0.15% different from ground measurements. For comparison, Hubble's SIs through the generations typically saw 1''–2'' of V2,V3 shifts and 0.2%–1.0% scale change from ground to flight.

Table 5 summarizes the end-to-end, referred to as "observatory," static wave front errors measured at the end of commissioning. The static wave front errors are well below their allocations in all channels, at all field points.

JWST's unobscured collecting area was measured using the NIRCam pupil imaging lens to be 25.44 m², exceeding its requirement of 25 m². The telescope's wavelength-dependent transmission ranges from 0.786 at 0.8 μm to 0.933 at 28 μm, again better than requirements at each wavelength. The transmission values were determined from final pre-flight measurements of mirror witness samples, combined with NIRCam grism measurements confirming the absence of detectable ice deposits.

The product of the above observed values for OTE area and transmission was also projected to end of life using modeled degradation of the optics. This places the effective area × transmission value of 19.58 m² at 0.8 μm and 23.18 m² at 20 μm, compared to the OTE requirements of 15.37 m² and 22.00 m², respectively.

Some of the OTE commissioning activities described earlier obtained data that also supported a secondary goal of probing various types of vignetting. Analyses of these data show no indication of any field-of-view cropping, unexpected OTE structure incursion, or pupil vignetting. Establishing that the telescope was unobstructed (except for secondary mirror support structures) fulfilled a mission-level requirement and was an exit criterion for OTE commissioning.

A dedicated thermal stability test was carried out following the telescope alignment in order to characterize the wave front stability and image motion on various timescales following a large, stressing thermal slew (i.e., early 2022 May; PID 1445 and 1446). This activity started by performing a 4 days thermal soak at the hot (Sun-normal) attitude and making baseline measurements. Then, the telescope was slewed to the cold attitude where continuous wave front measurements were made for the first 24 hr and then every ∼8 hr for the following 7 days. The thermal stability test confirmed three predicted wave front drifts that were bounded by the modeling predictions: short-timescale (2–4 minutes) oscillations from IEC panel heater cycling, medium-timescale (∼1 hr) drift from soft-structure induced thermal distortion, and long-timescale (∼1.5 days) drift from the composite backplane induced thermal distortion. These drifts are reported in Table 6. Temperature sensors on the telescope were also monitored during this test and confirmed that the temperature changes observed were within the noise of the temperature sensors (<40 mK).

Some science observation modes are, however, sensitive to these levels of WFE drift. It is also important to note that the worst-case delta-T induced as part of this thermal slew test would rarely be realized during normal science. In practice, science pointings across the sky are subject to much smoother and smaller temperature changes.

Pointing stability immediately following the thermal slew was measured when slewing from hot-to-cold and from cold back to hot. As discussed in Section 3.1, thermal distortion at the star tracker could result in uncorrected roll about the location of the fine guide star on the fine guidance sensor. The roll about the guide star was measured at the NIRCam field location to be 0.0265 mas hr⁻¹ in translation and comparable to a measured radial displacement from the star of −0.0230 mas hr⁻¹, well below the allocation of 6.3 mas.

During commissioning, there were many instances of sudden positional changes in one or more mirror segments, referred to as "tilt events" (see Section 6.2.1). The largest of these produce brief violations of the nominal stability values reported in Table 1. These positional changes are typically very small but detectable. These tilt events are generally ascribed to strain release within the OTE structures following cooldown to cryogenic temperatures, although the sources of the tilt events is not fully understood. The frequency and magnitude of these events have declined significantly in Cycle 1.

On many occasions, so-called tilt events, where sudden and uncommanded tilts of individual or groups of segments (e.g., wing segments), have been observed throughout commissioning and science operations. These tilt events were first observed during OTIS cryo–vacuum testing at Johnson Space Center in 2017 and they have been ascribed to the stick/slip strain release stored in the OTE hardware and/or structure during cooldown. They are expected to decrease in numbers over time as the OTE structure and hardware relax into their new environment. Tilt events continue to occur in the science mission (e.g., Schlawin et al. 2023), though less frequently and at a lesser magnitude than during early commissioning. As shown in Figure 23, tilt events episodically punctuate weeks-long periods of wave front stability. In practice, the infrequent occurrences of large tilt events have been the dominant source of WFE degradation requiring PMSA correction. Note however, that not all tilt events have led to a PMSA correction, and those who did were all corrected as part of our routine maintenance program. Ongoing trending will track the nature and frequency of these events. The last four months did not have any tilt events that drove excursions to 80 nm control threshold, which supports the hypothesis that the OTE structure is relaxing to a stable state. An example of a tilt event that occurred between wave front sensing visits is shown in Figure 24.

Zoom In Zoom Out Reset image size

Impacts from micrometeoroids on the PMSA have been observed since the middle of OTE commissioning. Pupil imaging first revealed localized surface changes to individual mirror segments, and phase retrieval analysis has revealed, in some cases, WFE changes on the impacted segments. However, not all micrometeoroid impacts have resulted in measurable changes in WFE since some show up only on pupil images or when averaging large numbers of optical path difference (OPD) maps (Figure 25). Moreover, the cumulative effect of these micrometeoroids impacts has so far minimally affected the overall telescope throughput.

Zoom In Zoom Out Reset image size

Notably, however, a large impact on segment C3 was observed in phase retrieval analysis from sensing visits covering the period May 22–24 UT. The impact was such that the global WFE worsened by 9 nm rms, after compensation by applying segment corrections in all degrees of freedom.

The telescope wave front error is still well below the nominal requirement following the single C3-event. However models of similar events indicate that with about ten similar events, we could be at our end-of-life wave front error requirement of 150 nm rms. Due to the precision launch of JWST, the observatory has sufficient fuel for 20+ yr of mission life, considerably longer than the mission design requirement minimum of 5 yr. With the possibility of an extended mission and the uncertain rate of C3-type events (from only a single occurrence) and the unexpected resulting WFE, the project has implemented a meteoroid avoidance zone (MAZ) for Cycle 2. Models produced by NASA's Meteoroid Environment Office show that the greatest impact rate for higher energy micrometeoroid strikes from sporadic sources, occurs in the so-called ram direction, the direction of flight as JWST moves with the Earth around the Sun. The Cycle 2 pointing restrictions favor observing in the wake (anti-ram) direction whenever possible. The proposed MAZ would reduce the instantaneous field of regard by about 40%. Models which remove all pointings from the MAZ and redistribute them over the allowed field of regard, can lower the impact rate on the primary mirror by 55%–65%. The reduction we expect to achieve in practice will likely be 30%–40%, because some high-priority and time-critical pointings in the MAZ will be allowed. As the Cycle 2 detailed observing plan is constructed, the expected impact reduction rate will be determined and monitored throughout the Cycle. Detailed damage models are also being constructed to provide a better understanding of the true risk of further C3-events; preliminary models suggest this was a higher than average energy impact on a sensitive area of C3. These models will be supplemented by data from a series of ballistic tests on relevant samples which should be completed over the next few months, building upon the experimental testing carried out early in the JWST development (e.g., Cohen et al. 2002).

The impact rate appears consistent with pre-flight expectations and mitigation strategies are being implemented. Meanwhile, micrometeoroids are being partially corrected as part of our routine maintenance corrections using, in particular, radius of curvature actuators. Since the end of commissioning, we have gained experience over a longer time baseline and with larger-number statistics.