Legacy Extragalactic Survey (SMILES)

• Legacy Extragalactic Survey (SMILES) is a deep, multi-tiered JWST program mapping galaxy evolution, dusty star formation, and AGN using advanced mid-infrared imaging and spectroscopy.
• It employs continuous coverage with eight MIRI bands and supplemental NIRSpec MOS for detailed measurements of PAH features, redshifts, and ionizing photon efficiencies in galaxies at cosmic noon.
• Its rigorous data reduction pipeline and open catalog releases enable robust multiwavelength studies, advancing our understanding of environmental impacts on reionization and galaxy feedback.

The Legacy Extragalactic Survey (SMILES) is a comprehensive, multi-tiered observational program centering on the use of JWST’s Mid-Infrared Instrument (MIRI) and Near-Infrared Spectrograph (NIRSpec) to probe galaxy evolution, star formation, and active galactic nuclei (AGN) across cosmic history, with particular emphasis on the obscured Universe at “cosmic noon” (redshifts $z \sim 1$–3) and beyond. SMILES leverages continuous mid-IR imaging (5.6–25.5 $\mu$m), medium-resolution spectroscopy, and rich ancillary data in well-studied deep fields, notably GOODS-S/HUDF. The initiative is distinguished by its contiguous multi-band coverage, rigorous data reduction pipeline, legacy catalog releases, and capacity for new extragalactic diagnostics inaccessible to prior facilities.

1. Survey Design and Scientific Scope

SMILES is architected to produce a definitive census of galaxies and AGN in the mid-infrared, exploiting the sensitivity and resolution gains of JWST/MIRI—up to three orders of magnitude deeper than past missions. The program targets a $\sim$34 arcmin$^2$ region in GOODS-S/HUDF using fifteen MIRI pointings in a 5×3 arrangement, ensuring contiguous coverage across all eight MIRI bands. NIRSpec multi-object spectroscopy (MOS) supplements imaging by providing redshifts and rest-frame optical diagnostics.

Principle science drivers are:

• Quantifying dusty star formation and cosmic dust properties via multiple PAH features (e.g., 3.3–18 $\mu$m)
• Detecting and characterizing heavily obscured AGN, many previously unobserved at shorter wavelengths
• Mapping the efficiency and topology of ionizing photon production, especially in dense environments
• Investigating multi-phase outflows, feedback, and the impact of environment on galaxy evolution

The survey design draws on experience from previous legacy fields (e.g., CANDELS (Grogin et al., 2011), DESI Legacy Imaging Surveys (Dey et al., 2018), HELP (Shirley et al., 2021)) and incorporates multi-tier (“wedding cake”) sampling strategies for robust statistical coverage across luminosity, mass, and redshift.

2. Instrumentation and Observational Methodology

SMILES takes full advantage of JWST’s MIRI spectral capabilities: eight intermediate-width imaging bands from 5.6–25.5 $\mu$m ($\Delta W = \Delta\lambda/\lambda_0 \approx 0.2$), with additional spectroscopy from NIRSpec (G140M/F100LP and G235M/F170LP gratings covering 0.97–3.07 $\mu$m at $R\sim1000$). This configuration allows continuous, high-fidelity spectral sampling of PAH emission, stellar continuum, and AGN-heated dust, enabling SED models with minimal bandpass correction uncertainties.

The observational setup includes:

• Uniform coverage and sensitivity planning driven by representative mid-IR SED templates for AGN and star-forming galaxies, ensuring depth sufficient to detect SFRs $>10 M_\odot$ yr$^{-1}$ at $z\sim2$
• Three-shutter nod dithering pattern for background subtraction and artifact mitigation, adapted to source size and spatial extent
• Ancillary photometry from HST, JWST/NIRCam, radio/X-ray, and ALMA for comprehensive multiwavelength association

The technical design—exemplified in both initial (Alberts et al., 24 May 2024) and subsequent (Rieke et al., 5 Jun 2024) data releases—enables statistically complete detection of both rare, bright AGNs and the bulk of star-forming galaxies, even at moderate to high redshift.

3. Data Reduction, Calibration, and Catalog Creation

The SMILES pipeline consists of multi-stage data reduction and calibration:

• Detector-level corrections (non-linearity, cosmic-ray rejection, bad pixel masking, warm pixel maps via median stacking)
• Astrometric alignment using external catalogs and per-tile corrections, with final mosaics resampled at 0.06″ pixel scale
• "Super-background" subtraction, stacking source-masked, baseline-subtracted images to remove large-scale gradients, readout stripes, and detector artifacts
• Mitigation of $1/f$ noise in NIRSpec data via column-wise Chebyshev polynomial modeling and subtraction ($\sim256$ pixels/degree)
• Spectroscopic calibration (wavelength, flat-field, flux), robust nod-based sky subtraction for both compact and extended sources

Photometric catalogs are extracted through forced photometry across all bands, employing circular and Kron apertures, completeness simulations by artificial source injection, and cross-association with existing deep field catalogs. The released product comprises:

• Eight fully reduced MIRI imaging mosaics (5$\sigma$ sensitivities 0.2–18 $\mu$Jy)
• Forced photometry for $>$3,000 sources at $\geq 4\sigma$ significance in at least one band
• NIRSpec spectra, redshift catalogs, emission-line measurements (GELATO for emission-line galaxies, pPXF for quiescent), metadata, and value-added SED fits (using Prospector)

4. Key Scientific Results and Analytical Methods

Early analyses showcase MIRI’s unique ability to resolve PAH-emitting regions in galaxies at $z\sim2$ and beyond. Color composites using F1500W, F1000W, and F560W demonstrate spatially resolved dust emission structures, and SEDs assembled from all eight bands enable direct measurement of total PAH luminosity ($$L_\mathrm{PAH} = L_{6.2} + L_{7.7} + L_{11.3} + ...$$).

The NIRSpec component has allowed first-of-kind measurement of the efficiency of ionizing photon production ($\xi_\mathrm{ion}$), defined via: $$\xi_\mathrm{ion} = \frac{\dot{n}_\mathrm{ion}}{L_\mathrm{UV}}, \quad \dot{n}_\mathrm{ion} = 7.28\times10^{11} L(\mathrm{H}\alpha)$$ where $\dot{n}_\mathrm{ion}$ is the ionizing photon rate derived from extinction- and slit-loss–corrected H$\alpha$ luminosity, with UV continuum from SED fitting. (Zhu et al., 18 Oct 2024) reported a positive correlation between $\log\xi_\mathrm{ion}$ and local galaxy overdensity $\log(1+\delta)$, with a measured slope of $0.94^{+0.46}_{-0.46}$ and a mean ratio difference of $2.43$ in $\xi_\mathrm{ion}$ between overdense and less dense regions. This supports the notion that environmental factors enhance ionizing photon output—potentially impacting cosmic reionization topologies.

Additional applications include:

• BPT diagnostics ([O III]/H$\beta$ vs. [N II]/H$\alpha$) for identification of heavily obscured AGN
• Measurement of outflow velocities from blueshifted emission and absorption features ($v_\mathrm{out}=828^{+79}_{-49}$ km s$^{-1}$)
• Calibration of mid-IR star formation indicators using resolved PAH strengths to probe SFRs and metallicity dependence

5. Survey Legacy, Data Release Policy, and Community Impact

SMILES maintains a commitment to rapid, open data release for legacy value. Imaging and spectroscopic data, catalogs, and processing software (e.g., GELATO, pPXF) are made available to the research community, often within months of acquisition. This approach aligns with practices set by previous legacy surveys such as CANDELS (Grogin et al., 2011), DESI (Dey et al., 2018), and HELP (Shirley et al., 2021), enabling extended science exploitation, reproducible analyses, and cross-survey comparisons.

The catalog structure is designed to facilitate downstream analyses, with metadata, selection functions, completeness maps, and multiwavelength cross-identifications enabling robust statistical studies across cosmic epochs.

6. Role of Environment and Theoretical Implications

The combination of deep mid-IR imaging, NIRSpec medium-resolution spectroscopy, and rich ancillary data positions SMILES as a unique resource for studying how environment shapes galaxy evolution. Target selection includes galaxies in overdense regions and protoclusters, allowing direct measurement of environment-dependent physical parameters.

Analysis of $\xi_\mathrm{ion}$ in dense environments (Zhu et al., 18 Oct 2024), together with the spatial mapping afforded by SMILES imaging, supports a scenario in which galaxies in overdense regions produce ionizing photons more efficiently—potentially accelerating local reionization of the IGM. Though the escape fraction of ionizing photons ($f_\mathrm{esc}$) remains uncertain, SMILES data provide vital empirical constraints on models of reionization topology and the interplay between environment and galaxy-scale feedback.

7. Historical Context and Future Directions

SMILES builds upon a lineage of extragalactic surveys, from early mid-IR mapping missions (IRAS, WISE, Spitzer) to legacy optical/NIR projects (CANDELS, HELP). MIRI’s multifilter design, unprecedented sensitivity, and resolution represent a factor of four increase in band coverage and several orders of magnitude improvement in capability over prior space- and ground-based efforts (Rieke et al., 5 Jun 2024).

The survey’s initial and second data releases (Alberts et al., 24 May 2024, Zhu et al., 18 Aug 2025) set methodological templates for future extragalactic studies, while recommendations point to expansion into other rich fields (e.g., COSMOS) and deeper, ultra-sensitive mapping in smaller areas for discovery space. Given the absence of planned missions with similar mid-IR reach, SMILES is poised to anchor next-generation investigations into obscured star formation, AGN phenomena, and cosmic reionization for years to come.

---

In summary, the Legacy Extragalactic Survey (SMILES) leverages JWST’s MIRI and NIRSpec instruments to establish a new paradigm for deep, multi-band mid-infrared extragalactic surveys. Its strategy, data products, analytical frameworks, and findings—such as environment-sensitive ionizing photon production—extend the boundaries of galaxy evolution studies and provide an open, enduring foundation for extragalactic research.