MIRI SMILES Program Overview

• MIRI SMILES Program is a comprehensive mid-IR survey using JWST/MIRI to probe dusty star-forming galaxies, AGN, and compact starbursts across a broad redshift range.
• The survey employs deep eight-band MIRI imaging in GOODS-S/HUDF, achieving sub-microJy sensitivity and an order-of-magnitude improvement in resolution over previous mid-IR studies.
• Its innovative approach leverages continuous spectral coverage to map PAH features, construct luminosity functions, and measure ionizing photon efficiency, unveiling obscured AGN and nascent starbursts.

The Systematic Mid-Infrared Instrument Legacy Extragalactic Survey (SMILES) is a flagship JWST mid-infrared (mid-IR) extragalactic program utilizing the unique continuous spectral coverage and sensitivity of MIRI, in conjunction with NIRCam and NIRSpec follow-up, to probe the physical conditions, demography, and evolution of dusty star-forming galaxies (SFGs), active galactic nuclei (AGN), and compact starbursts from $z\approx0.3$ to $z>7$. SMILES is the first deep, medium-area, contiguous 5–26 μm survey enabling robust separation of obscured AGN from compact starburst activity, detailed mapping of polycyclic aromatic hydrocarbon (PAH) features, and precise measurement of the luminous infrared (IR) population and its connection to environment and cosmic reionization.

1. Survey Design, Instrumentation, and Depth

SMILES covers a $34.5$ arcmin$^2$ field in GOODS-S/HUDF with an eight-band MIRI imaging mosaic (F560W, F770W, F1000W, F1280W, F1500W, F1800W, F2100W, F2550W; 5.6–25.5 μm), reaching 5σ point-source depths of 0.2–17 μJy, with effective exposure times from 655 to 2187 s per filter. The angular resolution ranges from 0.19″ (F560W) to 0.86″ (F2550W), an order-of-magnitude improvement over previous mid-IR surveys. Faint-source completeness is $\sim$80% at $\leq1$ μJy for F560W/F770W channels. MIRI imaging is complemented by corresponding NIRCam ($0.9$–$5\,\mu$m) and extensive HST multiwavelength coverage. A NIRSpec follow-up component (MSA, G140M/F100LP and G235M/F170LP, $R\sim1000$, 0.97–3.07 μm) provides medium-resolution rest-optical spectroscopy for emission-line diagnostics and redshift validation (Alberts et al., 2024, Rieke et al., 2024).

The survey strategy leverages the continuous MIRI spectral sampling to overcome earlier limitations (e.g., Spitzer/WISE) due to sparse filter coverage and confusion noise, which precluded reliable diagnostics of both obscured AGN and PAH features in distant galaxies (Rieke et al., 2024).

2. Scientific Goals and Target Classes

SMILES is designed for four principal science drivers:

1. Census of Dust-Obscured AGN and SFGs: Systematic discovery and classification of heavily obscured ($A_\mathrm{V}\gtrsim10$ mag) AGN and dust-buried compact starbursts, extending the reach for AGN/SFG identification to $z>7$0, and for starbursts and “Little Red Dots” (LRDs) beyond $z>7$1 (Lyu et al., 2023, Pérez-González et al., 2024).
2. Measurement of PAH Emission: Mapping the evolution and spatial variations of major PAH bands (3.3, 6.2, 7.7, 11.3, 12.8, 17 μm) to constrain interstellar dust and SFRs at “cosmic noon” ($z>7$2–3) (Alberts et al., 2024, Rieke et al., 2024).
3. Luminosity Functions and Cosmic Evolution: Construction of mid-IR/total-IR and AGN luminosity functions (LF), enabling derivation of cosmic star-formation rate density (CSFRD) and black hole accretion rate density (BHAD) from $z>7$3 to $z>7$4 (Ling et al., 24 Dec 2025).
4. Link to Environment and Reionization: Direct measurement of the ionizing photon production efficiency ($z>7$5) in relation to local overdensity for the first time, constraining the sources and topology of reionization (Zhu et al., 2024).

3. Data Acquisition, Reduction, and Catalogs

All data are processed through the JWST Calibration Pipeline with custom enhancements: detector-level corrections, median-stacked “super-background” subtraction, warm/hot pixel and stripe mitigation, astrometric alignment to $z>7$6 versus NIRCam, and creation of science and error mosaics at 0.06″ pixel⁻¹. Source detection is based on an inverse-variance-weighted F560W+F770W image; forced photometry is provided in all MIRI bands using empirical and WebbPSF aperture corrections. The initial catalog contains $z>7$7 sources with $z>7$8 detections, matched to deep JADES NIRCam coverage for ancillary photometry ($z>7$915 filters). Random-aperture noise estimation and completeness corrections are performed per band using injected simulated sources (Alberts et al., 2024, Lyu et al., 2023).

Spectroscopic redshifts are available for $34.5$0 of the sample, and the remainder achieve $34.5$1 in photometric redshift precision (Ling et al., 24 Dec 2025).

4. Classification Methodologies and Diagnostics

Galaxy classification utilizes both SED decomposition and color–color diagnostics:

• SED Fitting: SMILES employs the Prospector framework, extending with additive components for stellar population synthesis (FSPS), star-forming galaxy dust templates, and semi-empirical AGN dust torus models. Bayesian fitting (via dynesty) marginalizes over up to 12 parameters, including AGN obscuration ($34.5$2) and IR fraction (Lyu et al., 2023).
• Mid-IR Color–Color Cuts: Using synthetic photometry on $34.5$33,300 Spitzer/IRS spectra, color grids are established to isolate AGN, SFGs, and silicate-absorption-dominated nuclei across seven redshift intervals. Colors are defined as $34.5$4 with combinations selected to straddle PAH or silicate features in the appropriate bands depending on redshift. Gaussian mixture modeling (GMM) defines boundaries yielding $34.5$5 purity per class, e.g., for $34.5$6–0.30, $34.5$7 and $34.5$8 select AGN (Kilerci et al., 18 Dec 2025).

SED-based classification validates MIR-color methodology and vice versa, with $34.5$9 agreement between methods in overlapping samples. The combined approach enables robust identification of SFGs down to $^2$0 at $^2$1, AGN with obscured and non-obscured MIR SEDs, and the discovery of high-$^2$2 silicate-absorption objects.

5. Key Scientific Results and Astrophysical Implications

5.1 Obscured AGN and the Evolving AGN Population

SMILES identifies 217 SED-based AGN candidates, including 111 “primary” AGNs ($^2$3, $^2$4), with 34% of these undetected in prior Spitzer or Chandra surveys. The survey uncovers a substantial population of heavily IR-obscured AGN (70% in the primary sample), with the obscured fraction peaking at $^2$5 and gradually increasing with redshift to $^2$6 (Lyu et al., 2023, Rieke et al., 2024). The faint-end slope of the AGN LF ($^2$7–$^2$8) is flatter than previous estimates and persists out to $^2$9 (Ling et al., 24 Dec 2025).

5.2 Luminous IR and AGN Luminosity Functions

Monochromatic, $\sim$0, and AGN-specific LFs are constructed in seven redshift bins from $\sim$1 to $\sim$2. The total IR LF is fit with a modified Schechter function, and the AGN LF with a double power law. The evolution of the faint-end slope ($\sim$3 at $\sim$4 to $\sim$5 at $\sim$6) indicates a decline in low-luminosity dusty galaxies with cosmic time (Ling et al., 24 Dec 2025).

5.3 Dust-Obscured Compact Starbursts (“Little Red Dots”)

The SMILES LRD sample includes 31 $\sim$7 compact, red sources, representing $\sim$8 of massive galaxies ($\sim$9) at $\leq1$0. SED modeling confirms these are sub-kpc, extremely dust-enshrouded ($\leq1$1 mag), young ($\leq1$2 Myr), high-SFR ($\leq1$3) systems; stellar photospheres dominate the rest NIR/MIR SED, with a minority showing AGN torus signatures (Pérez-González et al., 2024).

5.4 Environment Dependence of Ionizing Photon Efficiency

For the first time, SMILES quantifies the correlation between ionizing photon production efficiency ($\leq1$4) and overdensity ($\leq1$5) using NIRSpec spectra. A statistically significant positive slope ($\leq1$6) links $\leq1$7 to $\leq1$8 for $\leq1$9. In overdense ($0.9$0) regions, galaxies show on average a factor $0.9$1 higher $0.9$2 than in less dense environments, independent of redshift. This suggests an environmental bias for ionizing photon production, likely due to burstier star formation in dense cosmic structures (Zhu et al., 2024).

5.5 Star Formation and Black Hole Accretion Histories

SMILES-derived $0.9$3 and AGN LFs allow direct measurement of CSFRD and BHAD evolution. The IR-luminous SFRD confirms a broad dusty peak at $0.9$4–3.0. The BHAD shows a flattened plateau at $0.9$5–4, with higher values at $0.9$6 relative to X-ray-based studies, reflecting a previously hidden population of high-$0.9$7 obscured AGN (Ling et al., 24 Dec 2025).

6. Technical Innovations and Data Products

Custom background subtraction, warm pixel masking, and astrometric refinement are critical for achieving the deepest possible point-source limits, especially at shorter MIR wavelengths (Alberts et al., 2024). The initial public data release comprises eight MIRI imaging mosaics, a multiaperture photometric catalog for $0.9$8 sources, and forced photometry in all bands, with cross-match to NIRCam and ancillary catalogs (Alberts et al., 2024, Rieke et al., 2024).

A classification tool using z-dependent MIR color cuts provides a fast, physically motivated means to isolate SFGs, AGN, and silicate-absorption systems by leveraging strong PAH and silicate features in seven redshift windows out to $0.9$9 (Kilerci et al., 18 Dec 2025).

7. Legacy and Future Prospects

SMILES demonstrates that wide, contiguous, multi-band MIRI imaging is indispensable for a comprehensive census of dust-obscured galaxies and AGN, revealing populations—particularly heavily obscured AGN and nascent compact starbursts—missed by optical/UV/X-ray techniques. Its continuous spectral coverage enables previously unattainable diagnostics of PAH features, SFRs, and buried AGN fractions.

Persistent themes include the necessity of combining deep MIR, NIR, and spectroscopic data for robust galaxy/AGN decomposition, the environmental dependence of galaxy physical properties, and the critical need for further expanded surveys (wide-shallow and ultradeep) with JWST/MIRI and future cooled IR missions to mitigate cosmic variance and achieve a definitive mid-IR legacy for extragalactic science (Rieke et al., 2024, Ling et al., 24 Dec 2025).