A GLIMPSE into the very faint-end of the H$β$+[OIII]$λλ$4960,5008 luminosity function at z=7-9 behind Abell S1063

Using the ultra-deep GLIMPSE JWST/NIRCam survey behind the Abell S1063 lens, this study constrains the faint-end of the H$\beta$+[OIII] luminosity function at $z=7-9$ to reveal that very faint galaxies contribute minimally to the ionising photon budget and cosmic star formation rate density, suggesting they play a limited role in cosmic reionisation.

The Gist

The Big Picture: Who Lit the Cosmic Switch?

Imagine the early Universe as a giant, dark room filled with thick fog (neutral hydrogen gas). For a long time, no one could see anything. Then, the first stars and galaxies turned on, blasting out powerful ultraviolet light. This light acted like a giant "switch," burning away the fog and making the Universe transparent so we can see it today. This event is called Re-ionization.

For decades, astronomers have been arguing a simple question: Who flipped the switch?

• Team Bright: Maybe a few massive, brilliant galaxies did all the work?
• Team Faint: Or maybe it was the sheer number of tiny, dim galaxies that, when added up, provided enough light to clear the fog?

This paper uses the James Webb Space Telescope (JWST) to settle the debate by looking at the "faint end" of the galaxy population behind a massive cosmic magnifying glass.

---

The Tool: A Cosmic Magnifying Glass

The researchers used a galaxy cluster called Abell S1063. Think of this cluster as a giant, natural lens. Just like a magnifying glass focuses sunlight to burn a leaf, this cluster's gravity bends and amplifies the light from galaxies behind it.

Because of this "magnifying glass," the JWST could see galaxies that are usually too dim to detect. It's like being able to see a firefly in a forest from a mile away because someone put a telescope right next to it.

The Discovery: The "Dim" Galaxies Are Not as Loud as We Thought

The team measured the light coming from these faint galaxies, specifically looking at a "glow" caused by hydrogen and oxygen gas (the [O III]+Hβ line). This glow is a direct indicator of how much new star formation is happening right now.

The Surprise:
They found that the number of these faint galaxies doesn't keep increasing as fast as we thought.

• The Old Expectation: If you look at the UV light (the "blue" light from young stars), the number of faint galaxies shoots up like a steep cliff. It seemed like there were billions of tiny, dim galaxies, and they should have been the main source of the light that cleared the cosmic fog.
• The New Reality: When looking at the actual gas glow (the nebular light), the cliff flattens out. It's more like a gentle hill.

The Analogy:
Imagine a stadium full of people.

• The UV View (Old): You see thousands of people holding tiny, flickering candles. You assume that because there are so many of them, their combined light is blinding.
• The Gas Glow View (New): When you look closer, you realize that while there are many people, most of them are holding dead candles or very weak ones. They are there, but they aren't actually lighting up the stadium.

Why the Difference? (The "Bursty" Stars and Metal Problem)

The paper offers two main reasons why these faint galaxies aren't as bright as they look in UV light:

1. The "Bursty" Lifestyle: Small galaxies don't form stars steadily. They have "bursts" of intense activity followed by long periods of sleeping (quiescence).
   • Analogy: Think of a firework. It flashes brightly for a second, then goes dark. If you take a photo of a firework at the exact moment it's dark, it looks like a dud, even though it's the same object that was bright a moment ago. The faint galaxies are often in their "dark" phase.
2. The "Metal" Deficit: In astronomy, "metals" are elements heavier than hydrogen and helium. Young, faint galaxies have very few metals.
   • Analogy: Think of a campfire. A fire with dry wood (metals) burns hot and bright with lots of sparks. A fire with wet, green wood (low metals) smolders and produces less light. The faint galaxies are like wet wood; they just don't produce the same amount of glowing gas as the bright ones.

The Verdict: Who Flipped the Switch?

The researchers calculated how much "ionizing light" (the light that clears the fog) these galaxies produce.

• The Result: Because the number of faint galaxies flattens out (there aren't as many super-dim ones as we thought), they contribute much less to the clearing of the cosmic fog than previously believed.
• The Conclusion: The "Team Faint" theory is likely wrong. The job of re-ionizing the Universe was probably done by the brighter, more massive galaxies that we can already see, rather than a massive army of invisible, tiny ones.

Summary in One Sentence

By using a cosmic magnifying glass to look at the faintest galaxies, this study found that they are too dim and too "sleepy" to have cleared the Universe's fog on their own; the heavy lifting was likely done by the brighter, more obvious galaxies.

Technical Summary

1. Problem and Motivation

The epoch of reionization (EoR), occurring roughly between redshifts $z \sim 6$ and $z \sim 15$, remains a critical area of study in cosmology. A central debate concerns the primary drivers of reionization: whether it was driven by a large population of faint, low-mass galaxies or a smaller population of bright galaxies.

• The Discrepancy: Previous studies of the Ultraviolet (UV) luminosity function (LF) at high redshifts ($z > 7$) typically found a steep faint-end slope ($\alpha \le -2$). If this steep slope holds, faint galaxies would dominate the ionizing photon budget, potentially leading to an "overshoot" of the reionization rate unless the escape fraction of ionizing photons ($f_{esc}$) is very low or evolves significantly.
• The Gap: Direct measurements of nebular emission lines (tracers of instantaneous star formation) at $z \sim 7-9$ have been limited by sensitivity. Previous constraints on the $[O \text{III}] + H\beta$ LF were restricted to brighter galaxies ($L \gtrsim 10^{41} \text{ erg s}^{-1}$), leaving the faint end ($L < 10^{41} \text{ erg s}^{-1}$) unconstrained or indirectly inferred.
• Objective: This study aims to constrain the faint end of the $[O \text{III}] + H\beta$ luminosity function down to unprecedented depths ($L \sim 10^{39} \text{ erg s}^{-1}$) to determine if the steep UV LF slope translates to nebular lines, and to assess the true contribution of faint galaxies to cosmic reionization.

2. Methodology

Data and Observations

• Survey: The authors utilized the GLIMPSE survey, a Cycle 2 JWST/NIRCam large program (GO-3293) targeting the strongly lensed Hubble Frontier Field cluster Abell S1063 ($z=0.348$).
• Depth: The survey achieved 5$\sigma$ depths of $\sim 30.9$ mag in broadband filters (F090W–F444W) and $\sim 30.1$ mag in medium bands.
• Sample Selection:
  • Lyman-Break Galaxy (LBG) Selection: A color-color selection was applied to identify galaxies with a flux dropout in the F090W filter (rest-frame Lyman-$\alpha$ break), targeting the redshift range $7 < z < 9$.
  • Photometric Redshifts: SED fitting using Eazy confirmed redshifts ($z > 6$).
  • Final Sample: 164 unique lensed galaxies (after removing multiply imaged systems) were selected.

Analysis Techniques

• Gravitational Lensing Modeling: A new parametric strong-lensing mass model (Zitrin et al. method) was used to calculate magnification factors ($\mu$) for each object. The model achieved an RMS of $0.54''$.
• Effective Volume: The effective survey volume ($V_{eff}$) was calculated by integrating the comoving volume weighted by magnification and detection completeness.
• SED Fitting & Flux Measurement:
  • Software: CIGALE was used to fit Spectral Energy Distributions (SEDs) with a delayed-$\tau$ star formation history, fixed metallicity ($Z=0.004$), and nebular emission templates.
  • Line Fluxes: The combined $H\beta + [O \text{III}] \lambda\lambda 4960, 5008$ flux was derived from the SED fits, as these lines fall within the NIRCam medium bands (F410M, F480M) and are blended in broadband data.
• Luminosity Function Construction:
  • The LF ($\Phi$) was calculated using the $1/V_{max}$ method, accounting for completeness ($C$) and effective volume ($V$).
  • Schechter Fit: The LF was parametrized using a Schechter function ($\phi^*, L^*, \alpha$).
  • Combination: GLIMPSE data (faint end) was combined with FRESCO survey data (Meyer et al. 2024) to constrain the bright end.

3. Key Contributions and Results

A. The Faint-End Slope of the $[O \text{III}] + H\beta$ LF

The study successfully measured the LF down to $L \sim 10^{39} \text{ erg s}^{-1}$ ($M_{UV} \sim -12$).

• Measured Slopes:
  • $z = 7-8$: $\alpha = -1.78^{+0.06}_{-0.06}$
  • $z = 8-9$: $\alpha = -1.55^{+0.11}_{-0.11}$
• Comparison: These slopes are significantly flatter than the UV continuum LFs at similar redshifts ($\alpha_{UV} \le -2$).
• Discrepancy Resolution: The authors ruled out that this difference is solely due to selection bias or depth. They compared their results to Wold et al. (2025), who found a steeper slope ($\alpha \approx -2.07$) but only probed down to $10^{41} \text{ erg s}^{-1}$ and applied a strict equivalent width cut. When GLIMPSE data was restricted to the Wold et al. depth, the slope steepened, confirming that the inclusion of very faint, low-equivalent-width galaxies drives the flattening.

B. Physical Interpretations

The authors propose three mechanisms to explain the flatter nebular LF compared to the UV LF:

1. Bursty Star Formation: Faint galaxies may have "down-turning" star formation histories (bursty SFHs), leading to a lower $[O \text{III}]+H\beta$ to UV ratio compared to UV-selected galaxies.
2. Metallicity Evolution: Lower-mass (fainter) galaxies have lower metallicities, which reduces the $[O \text{III}]/H\beta$ ratio ($R_3$). The authors modeled an evolving $R_3$ ratio as a function of $M_{UV}$, which successfully reproduces the observed flattening when separating the $H\beta$ and $[O \text{III}]$ components.
   • Resulting Separated LFs: The derived $H\beta$ LF is steeper ($\alpha \approx -1.95$) but still flatter than the UV LF, while the $[O \text{III}]$ LF is even flatter ($\alpha \approx -1.66$).
3. UV LF Turnover: A potential intrinsic turnover in the UV LF at the faint end was simulated. However, current UV data does not show a sharp turnover, making this less likely than the SFH/metallicity explanations.

C. Ionizing Photon Budget and Reionization

Using the derived LF and assuming a constant Lyman-continuum escape fraction ($f_{esc} = 0.14$):

• Convergence: Because the faint-end slope $\alpha > -2$, the cumulative ionizing photon emissivity ($\dot{N}_{ion}$) saturates quickly. Extending the integration to even fainter, undetected galaxies adds negligible contribution.
• Quantitative Contribution:
  • At $7 < z < 8$: Galaxies contribute 31% – 90% of the required ionizing budget.
  • At $8 < z < 9$: Galaxies contribute 46% – 156% of the required budget.
• Implication: Unlike scenarios assuming a steep UV slope (which require massive contributions from undetected faint galaxies), this study suggests that the detected population of star-forming galaxies is sufficient to drive reionization without relying on a "hidden" population of ultra-faint galaxies.

D. Cosmic Star Formation Rate Density (SFRD)

• The SFRD was calculated by integrating the dust-corrected $H\alpha$ LF.
• Results: The SFRD is consistent with previous nebular emission line measurements but slightly higher than UV-based estimates (which suffer from burstiness biases).
• Faint Galaxy Impact: Integrating down to the faintest limits ($SFR \sim 0.005 M_\odot \text{ yr}^{-1}$) only increases the total SFRD by $\sim 0.1-0.3$ dex. This confirms that faint galaxies contribute very little to the global star formation density at these redshifts.

4. Significance

• Resolving the "Ionizing Photon Crisis": The paper provides strong evidence that the steep UV LF slope does not translate to nebular lines. The flattening of the nebular LF resolves the tension where UV-based models predicted an over-production of ionizing photons (the "crisis").
• Role of Faint Galaxies: Contrary to the "faint galaxy dominated" reionization scenario, this study suggests that faint galaxies ($M_{UV} > -17$) are numerous but have low ionizing efficiencies due to bursty star formation and low metallicity. They are minor contributors to the total ionizing budget.
• Methodological Advance: The combination of the deepest JWST survey (GLIMPSE) with strong lensing modeling allows for the first robust statistical measurement of the nebular LF at $L < 10^{40} \text{ erg s}^{-1}$, moving beyond indirect UV-to-line conversions.

In conclusion, the GLIMPSE survey demonstrates that the bulk of the ionizing photon budget at $z \sim 7-9$ is produced by the galaxies currently detected, and the Universe's reionization can be explained by the observed population without invoking a vast, unseen population of ultra-faint galaxies.