NIRSpec Instrument Overview

• NIRSpec is a near-infrared spectrograph on JWST featuring a configurable micro-shutter array with nearly 250,000 MEMS shutters for simultaneous spectra of multiple objects.
• It offers four observing modes—including multi-object spectroscopy, fixed slits, integral field, and bright object time series—to support diverse astronomical studies.
• High spectral resolution, calibrated sensitivity, and robust data reduction techniques enable deep surveys of the early Universe and precise measurements of faint cosmic sources.

The Near-Infrared Spectrograph (NIRSpec) is the primary cryogenic, near-infrared multi-object spectrograph onboard the James Webb Space Telescope (JWST). It is optimized to enable moderate- to high-signal-to-noise spectroscopy of faint astronomical sources across a wide wavelength range (0.6–5.3 μm), with a strong emphasis on multiplexed surveys in the early Universe. NIRSpec introduces the first ever configurable multi-object spectrograph in space, employing a micro-shutter array (MSA) containing nearly 250,000 individually addressable microelectromechanical (MEMS) shutters. This advances near-infrared astronomy through rapid, simultaneous observation of 50–200 targets per field—orders of magnitude more efficient than previous single-object or slitless systems—while also supporting fixed-slit and integral-field spectroscopy modes (Ferruit et al., 2022, Jakobsen et al., 2022).

1. NIRSpec Instrument Architecture and Modes

NIRSpec comprises a reflective optical chain with the following major elements: the JWST pick-off mirrors select a 3.6′×3.4′ region of the focal surface, which is then reimaged by the three-mirror fore-optics onto the slit plane. The slit plane hosts the MSA, five fixed slits (ranging from 0.2″ to 1.6″ in width), and a 3.1″×3.2″ integral field unit (IFU). Downstream optics collimate the beam, introduce dispersion via seven selectable elements (six ruled gratings and one double-pass CaF₂ prism), and focus the spectral orders onto two 2048×2048 H2RG HgCdTe arrays (18 μm pitch, mated to a silicon carbide bench for thermal stability). The MSA is the default for multi-object spectroscopy (MOS), providing programmable aperture configurations on the sky (Ferruit et al., 2022, Jakobsen et al., 2022, Böker et al., 2023).

Four distinct observing modes are supported:

• Multi-object spectroscopy (MOS): Up to 200 sources observed simultaneously across a 9.18 arcmin² field, each through a 0.20″×0.46″ shutter, typically configured as 1×3 “slitlets” for nodding capability (Ferruit et al., 2022).
• Fixed Slits (FS): Five static slits at the MSA center, supporting high-contrast work, bright targets, and time-series (BOTS) observations.
• Integral Field Spectroscopy (IFS): A reflective image slicer remaps a 3.1″×3.2″ patch into 30 slices (∼0.1″ wide), forming a virtual slit for contiguous spectroimaging (Böker et al., 2022).
• Bright Object Time Series (BOTS): Dedicated subarray reads (e.g., S1600A1 1.6″×1.6″ slit) for minimized slit-loss and extremely low systematics in exoplanet transit studies (Birkmann et al., 2022).

2. Micro-Shutter Array: Design, Operation, and Multiplexing

The MSA consists of four quadrants (98″×91″ each, total 9.18 arcmin²), each with a 365×171 shutter grid (0.20″×0.46″ open area per shutter, 0.27″×0.53″ pitch, 0.069″ inter-shutter bar). The hardware employs a combination of permanent magnet arm sweeps and electrostatic latching for rapid, robust addressability. Full array reconfiguration for a science pattern requires two magnet sweeps (∼25 seconds each), ensuring stable aperture definition for exposures. Approximately 82.5% of the unvignetted shutter population is deemed operable post-commissioning, with the main limitations being electrical short-masking (∼10.5%) and a population of failed-closed shutters (∼7.0%) (Rawle et al., 2022, Böker et al., 2023).

MOS mode typically deploys 1×3 open “slitlets” to allow background subtraction via target nodding. The system can routinely configure ∼200 open shutters per exposure, yielding up to ∼100 non-overlapping science spectra, with multiplexing efficiency limited by detector real-estate, the fixed 4-quadrant array geometry, and contamination/overlap risk. Multiplexing is formally modeled as follows (for catalogue target surface density $\Sigma_T$ and shutter solid angle $\Omega_\mathrm{SH}$): $\bar n_{\rm MSA} = \Omega_{\rm MSA}\,\Sigma_T$

$$p_{\rm VS} = \frac{N_S}{N_0}\frac{\theta_x}{0.5}\frac{\theta_y}{0.5}$$

$$p_{\rm NC} \approx \exp\left(-5\,\Sigma_C\,\Omega_{\rm SH}\right)$$

$$\bar n_{\rm VS}=p_{\rm VS}p_{\rm NC}\bar n_{\rm MSA}$$

where $\bar n_{\rm VS}$ is the expected uncontaminated count of viable slitlets and the “Arribas Algorithm” is used to maximize the non-overlapping set for spectral extraction (Ferruit et al., 2022).

3. Spectral Configurations, Throughput, and Sensitivity

NIRSpec supports three primary spectral configurations:

• Prism (CLEAR): 0.6–5.3 μm, resolving power $R=30$–330, single shot, optimized for continuum slopes and drop-out selection.
• Medium-resolution gratings: $R=700$–1340 (e.g., G140M, G235M, G395M), spanning 0.7–5.2 μm. Enables robust nebular emission-line diagnostics and stellar population studies.
• High-resolution gratings: $R=1850$–3700, covering same bands, for precise kinematics and velocity dispersion studies.

Quantum efficiency of the H2RG detectors peaks at 85%, with system photon conversion efficiency reaching 59% for the prism and ∼46% for gratings. Observed point-source sensitivity (S/N=10, 10,000 s) is $\sim100$ nJy ($\mathrm{AB}\sim26.4$) in prism, $\sim1\,\mu$Jy ($\mathrm{AB}\sim23.9$) in medium $R$, and $\sim3\,\mu$Jy ($\mathrm{AB}\sim22.7$) in high $R$ (Ferruit et al., 2022, Böker et al., 2023).

A key complexity in the MOS mode is the strongly wavelength- and position-dependent transmission function, requiring on-orbit calibration of point-spread path losses as a function of intra-shutter offset and wavelength (the transmission varies with JWST’s diffraction-limited PSF scaling as $\propto\lambda$). Calibration reference files are derived by stepping sources within slitlets (Ferruit et al., 2022).

4. Calibration, Data Reduction, and Performance

Spectral and spatial calibration is built on a high-fidelity, parametric instrument model—forward and backward geometric transforms linking sky to detector pixels via plane-to-plane optical paths, incorporating fifth-order polynomial distortion terms. In commissioning, in-orbit fitting of >30,000 spectral/astrometric reference points yielded residuals $\lesssim0.1$ pixel in both dispersion and cross-dispersion—better than 1/8 of a resolution element, surpassing mission allocations (Lützgendorf et al., 2022, Dorner et al., 2016).

The calibration pipeline executes:

• Stage 1: Saturation/jump detection, reference-pixel correction, non-linearity correction, dark subtraction, up-the-ramp ramp fitting.
• Stage 2: Background subtraction, spectrograph and detector flat corrections (D-, S-, F-flats), wavelength zero-point for MOS sources, absolute flux calibration, and path-loss correction based on observed/expected source centering (Ferruit et al., 2022, Oliveira et al., 2018, Lützgendorf et al., 2022). Handling of extended or non-uniform sources adopts a pixel-area correction and differential slit loss calculation; ongoing algorithmic developments aim to accommodate complex intrinsic morphologies.

Systematic effects such as electrical short-induced shutter masking, failed-closed/open shutters, and cosmic ray hits are accounted for by dynamic operability mapping, path-loss calibration, and robust cosmic-ray rejection algorithms. With a cosmic-ray rate near $5\,\mathrm{cm}^{-2}\,\mathrm{s}^{-1}$, detector noise sees a $\sim$7% increase for 1,000 s subexposures, yielding a $\le$8% impact on sensitivity for detector-limited modes—well within science requirements (Giardino et al., 2019).

5. Operational Reliability and On-Orbit Contingency Management

MSA reliability is actively managed using telemetry monitoring, automated electrical short-diagnosis routines, and dynamic row/column masking strategies. Short-induced IR glow is the dominant operational failure mode; affected lines are masked to suppress contamination, while minimizing loss of multiplexing power ($\sim$0.1% per line masked). Advances in onboard automation decreased masking latencies from weeks to days, and recent campaigns successfully unmasked self-healed lines, recovering MOS capacity (Bechtold et al., 18 Aug 2025).

Probabilistic shutter-quality flagging is now advocated to make optimal trade-offs between MOS multiplexing and reliability. Instead of hard binary unusability, shutters are classified by empirical open-command success probability, and MOS configuration tools can weight the use of shutters accordingly to maximize science yield (Rawle et al., 2022).

Current on-orbit performance exceeds 99% shutter operability (excluding vignetting and scheduled maintenance), with typical MOS programs accommodating $\sim$100 targets per exposure with negligible multiplexing loss due to masking.

6. Science Impact and Applications

NIRSpec’s MOS mode enables rapid, deep spectroscopy of extremely faint galaxies, emission-line studies in the epoch of reionization, and comprehensive environmental, chemical, and structural diagnostics across cosmic time. The combination of $R=100$–$3700$ resolving power and wide spectral coverage (0.6–5.3 μm) allows detection of crucial features from Lyman-$\alpha$ to PAHs, supporting science from exoplanet atmospheres to early Universe galaxy formation (Ferruit et al., 2022, Birkmann et al., 2022, Maseda et al., 2018).

Deep MOS exposures provide a serendipitous channel, where even non-targeted objects yield detectable [O III] or H$\alpha$ in nearly every open 1×3 slitlet after $\sim$20 hours of integration (yielding $\sim$1 new detection per slitlet at $t_\mathrm{int}=75,000$ s in R=100 mode), further enhancing spectroscopic yield for the highest-redshift and faintest populations (Maseda et al., 2018).

7. Future Prospects and Lessons for MEMS Spectrographs

Procedures developed for NIRSpec’s MSA—especially autonomous short detection, dynamic masking/unmasking, and operability mapping—establish a baseline for the deployment of future MEMS-based multi-object systems, including the Habitable Worlds Observatory. With the extended JWST mission and robust MSA hardware design, NIRSpec will continue to deliver transformative multi-object spectra into the 2030s and beyond, providing an essential resource for extragalactic, exoplanet, and time-domain studies (Bechtold et al., 18 Aug 2025, Ferruit et al., 2022).