A Census of Na D-traced neutral ISM and outflows at 0.6<z<4

arXiv:2604.18522 · astro-ph.GA · Submitted 2026-04-20 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: Today's paper: "A Census of Na D-traced neutral ISM and outflows at 0.6<z<4".

Jocelyn: A statistical census of neutral interstellar medium (ISM) and outflows in 309 galaxies at redshifts between 0.6 and 4 using JWST/NIRSpec spectroscopy provides a crucial view of how feedback mechanisms…

Vera: First, who's behind it and why it matters.

Paper summary: Vera: So, this paper, "A Census of Na D-traced neutral ISM and outflows at zero point six<z<four" gives us a really important statistical snapshot of what's happening in galaxies during cosmic noon. It claims to provide a crucial view into how feedback mechanisms regulate galaxy evolution across that redshift range using JWST/NIRSpec spectroscopy, which is super exciting for observational astronomy.

Jocelyn: I agree, Vera; the scope of this study is huge because it tackles the neutral interstellar medium and outflows, which we often struggle to trace compared to the ionized phases. What they are claiming here is that this census helps us understand how feedback acts on galaxy evolution at that specific cosmic time.

Subrahmanyan: From a theoretical standpoint, this type of dataset is incredibly valuable because it addresses the gap in our understanding of multiphase gas, which spans a wide range of temperatures and densities, making feedback characterization observationally difficult.

Vera: Exactly; the paper is focused on using Na D absorption to detect this cool neutral phase across three hundred nine galaxies at redshifts between zero point six and four using data from four different surveys: SMILES, JADES, Blue Jay, and Aurora.

Jocelyn: And what’s compelling is their approach of subtracting the stellar continuum using penalized pixel fitting, pPXF to model the Na D doublet at rest-frame five thousand eight hundred ninety Å and five thousand eight hundred ninety-six Å to actually detect this neutral ISM absorption in seventy-six galaxies.

Subrahmanyan: That method moves us past just seeing bright emission lines and starts constraining the gas reservoir itself, which is a big step toward modeling how energy and momentum from feedback drive galaxy evolution.

Vera: The key claims they make are about how the presence of Na D absorption depends on host-galaxy properties; they found that the percentage of detections is higher in massive galaxies, specifically those with a log(M*/M⊙) greater than ten and only fifteen percent in lower-mass systems.

Jocelyn: That dependency on host mass is fascinating because it suggests that the conditions necessary to produce detectable neutral absorption are tied directly to the galaxy's gravitational potential or star formation rate.

Subrahmanyan: If the detection rate scales with stellar mass, it implies that feedback processes, whether from supernovae or black holes, are more effective at driving outflows in more massive systems where the gas reservoir is larger.

Paper summary: Vera: They also pointed out that for lower-mass galaxies on or above the star-forming main sequence, detections are only found in a specific subset, which narrows down where we expect to find this neutral gas.

Jocelyn: Then they move into how dust and star formation history affect these observations, showing a strong correlation between Na D absorption and dust attenuation—the percentage of Na D-detected galaxies have a median AV higher than those without detections.

Subrahmanyan: That correlation with dust is significant because it suggests that the neutral phase we are observing isn't just randomly distributed; it’s intimately linked to the processes that produce dust, which usually involves star formation or intense radiation fields.

Vera: Furthermore, they found a statistically significant correlation between dust attenuation and the equivalent width of the Na D absorption, EWNaD,ISM with p less than zero point zero five.

Jocelyn: That statistical significance is what makes their connection between dust and the neutral gas phase really robust; it’s not just a visual trend in the data.

Subrahmanyan: This correlation helps us refine models of multiphase gas because it links a measurable property of the neutral phase directly to the processes that create obscuration, which is essential for understanding how galaxies quench.

Vera: They also looked at massive quiescent galaxies and found that Na D absorption is preferentially seen in older systems with larger four thousand Å breaks, like Dn4000 greater than one point four.

Jocelyn: That means we might be seeing the remnants of gas that was present when the galaxy was actively forming stars, even if it’s now quiescent, which is a key piece of information for quenching pathways.

Subrahmanyan: It suggests that these older systems retain a signature of their past star formation history through this neutral phase, which ties directly into how we model the transition from star-forming to quiescent galaxies.

Vera: And they also found that in younger, rapidly quenching galaxies with strong Balmer absorption HδA, Na D is detected, pointing toward a link with recent quenching history.

Jocelyn: It seems like the paper is building a coherent picture connecting the neutral gas phase across different galaxy types and evolutionary stages.

Subrahmanyan: The classification of outflows in this census, where they identified twenty-six Na D outflows, shows a clear dichotomy between star-forming and quiescent systems.

Paper summary: Vera: In star-forming galaxies, the Na D outflows correlate with star-formation properties, which supports the idea of a "star-formation-driven origin" for those outflows.

Jocelyn: But in quiescent galaxies, the situation is different because those outflows are not associated with residual star formation and often require more energy than ongoing star formation can supply.

Subrahmanyan: This difference in required energy is what drives the next big idea: that AGN dominate Na D-traced neutral outflows in these quiescent systems, with a high fraction of forty-six percent among the outflow-detected quiescent galaxies.

Vera: They even found five quiescent galaxies with what they call "AGN fossil outflows," which suggests that AGN-driven outflows can persist even after the active accretion phase has ended.

Jocelyn: So, the physical properties of these twenty-six Na D outflows showed that those in star-forming galaxies have energetics consistent with ongoing star formation, but the ones in quiescent galaxies need more momentum than low levels of current star formation can provide.

Subrahmanyan: When comparing neutral and ionized phase outflows, the paper notes that neutral outflows are "more readily detected, and possibly more prevalent, than ionized outflows in massive quiescent galaxies at z ∼ two" which they attribute to tracer-dependent detectability.

Vera: That points toward a picture where Na D-traced neutral outflows in quiescent galaxies are "more often linked to AGN feedback, either synchronous in AGN hosts or fossil in systems lacking current AGN signatures".

Jocelyn: They also noted that the cosmic-noon Na D outflows appear offset by roughly zero point three–zero point five dex in M˙out and E˙out compared to local ULIRGs and post-starburst galaxies, although they weren't sure if this indicated redshift evolution due to profile modeling differences.

Subrahmanyan: This comparison highlights how important the cool neutral phase is for understanding galaxy quenching at cosmic noon, showing that Na D outflows are driven by different mechanisms depending on whether the galaxy is star-forming or quiescent.

Vera: So, to wrap up this census of Na D-traced neutral ISM and outflows at zero point six<z<four we see a state-dependent picture where kinematics and energetics are consistent with ongoing star formation in star-forming galaxies but point toward AGN feedback in quiescent ones.

Paper summary: Jocelyn: That means the driving mechanism isn't universal; it really depends on the galaxy's current evolutionary state, which is a complex insight for us studying galaxy assembly across cosmic time.

Subrahmanyan: The implication here is that we need to move beyond just studying the ionized gas to get a complete picture of how feedback shapes the gas within and around galaxies, which is observationally difficult because galaxies host multiphase gas spanning a wide range of temperatures and densities.

Vera: I think this paper really pushes us to incorporate these cool neutral phases into our models, because the results show that Na D outflows are driven by distinct mechanisms in star-forming and quiescent galaxies.

Jocelyn: It opens up new avenues for future work regarding how we can better constrain these energy injections in both active and passive systems across cosmic time, which is where the real sky surveys come in next.

Subrahmanyan: The study's limitation, as stated by the authors, is that they are working under a standard ΛCDM universe with specific cosmological parameters—H0 = seventy km s−one Mpc−one omegaΛ = zero point seven, and omegam = zero point three —which limits the direct comparison to other cosmological models without further adjustments.

Vera: That’s a fair limitation; it sets the stage for future work where we can test these findings against different cosmological frameworks and see how robust these conclusions remain.

Jocelyn: Overall, the census provides a crucial statistical view of the cool neutral phase across the quenching sequence at cosmic noon, linking it to specific physical properties like dust and stellar population ages.

Subrahmanyan: This work contributes to understanding how feedback shapes galaxy evolution by providing a census of outflows across gas phases and evolutionary stages, which is what the authors set out to do.

Vera: It really shows that Na D-traced neutral outflows at cosmic noon are state-dependent, driven by different mechanisms in star-forming versus quiescent galaxies.

Jocelyn: That distinction between star-formation and AGN feedback in the quiescent regime is a really important finding for mapping out galaxy quenching pathways.

Subrahmanyan: The conclusion that "AGN fossil outflows may be common among cosmic-noon quiescent galaxies rather than rare phenomena" is a significant point because it suggests that the impact of past AGN activity can persist long after the AGN itself has stopped actively feeding.

Conclusion: Vera: So we've been diving deep into the data from this new study, "A Census of Na D-traced neutral ISM and outflows at zero point six<z<four" and now it's time to look at what all that means for galaxy evolution.

Jocelyn: I agree; the title itself tells us exactly what they're tackling—a census of neutral gas and outflows across a huge stretch of cosmic time, from early times to just before today.

Subrahmanyan: From my perspective as a theorist, the authors are using this data to build a statistical picture of how feedback mechanisms, like those from supernovae or black holes, manage the gas reservoirs in galaxies during this crucial epoch at cosmic noon.

Vera: Exactly; they're not just looking at pretty pictures of bright emission lines; they're trying to map out the invisible stuff that’s actually controlling how galaxies grow and change.

Jocelyn: And when you consider the authors, SMILES, JADES, Blue Jay, and Aurora all contributing to this dataset—that tells us this isn't just one team working on a single telescope; it's a massive effort pulling together multi-wavelength data from different surveys.

Subrahmanyan: That breadth of data is what makes their statistical census so robust; they’re trying to capture the complexity of the ISM across various galaxy types and redshifts using this Na D tracer.

Vera: It really shows how important it is to look at these neutral phases because they are often where the action—the regulation of star formation—is happening, not just the bright, ionized gas we usually see.

Jocelyn: And their conclusion seems to be that the way feedback works isn't uniform across all galaxies; it really depends on whether a galaxy is actively forming stars or has already settled into a quiescent state.

Subrahmanyan: That distinction between star-forming and quiescent systems, especially concerning the role of AGN in those quieter galaxies, is where this research connects directly to our larger models of galaxy quenching.

Vera: It suggests that we need to look at these cool neutral outflows not just as a byproduct, but as a critical component in understanding how galaxies transition from being actively star-forming to becoming passive.

Jocelyn: And the fact that they found evidence for "AGN fossil outflows" in quiescent systems really adds a layer of nuance, showing that past activity can leave an imprint long after the main event has passed.

Subrahmanyan: That concept of fossil outflows persisting is significant because it challenges simpler models where feedback effects are assumed to be instantaneous or purely synchronous with current star formation.

Vera: So, what this study really gives us is a detailed map showing that the physics driving galactic evolution changes depending on the galaxy's current evolutionary status at cosmic noon.

Jocelyn: It sets a clear direction for future observational work, pointing toward how we can better detect and characterize these neutral phases in other deep surveys across different redshifts.

Steward Observatory, University of Arizona · Kavli Institute for Cosmology, University of Cambridge · Cavendish Laboratory, University of Cambridge · DARK, Niels Bohr Institute, University of Copenhagen · Department of Physics, University of Oxford · Scuola Normale Superiore Pisa · Department of Astronomy & Astrophysics The Pennsylvania State University DepartmentofPhysics

astro-ph.GA

Submitted: 2026-04-20

Updated: 2026-10-01

Comments: 35 pages, 20 figures; accepted for publication in The Astrophysical Journal

Project page: https://ga-nifs.github.io

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 79/100

The gist: A statistical census of neutral interstellar medium (ISM) and outflows in 309 galaxies at redshifts between 0.6 and 4 using JWST/NIRSpec spectroscopy provides a crucial view of how feedback

Key concepts

Na D doublet
This refers to a specific spectral signature from sodium atoms in the neutral interstellar medium. By observing absorption at 5890 Å and 5896 Å, researchers can detect the presence of cool, neutral gas surrounding galaxies.
Star-formation-driven origin
In star-forming galaxies, Na D outflows are consistent with energy being supplied by ongoing star formation. This suggests that when a galaxy is actively forming stars, the feedback mechanisms driving these outflows are primarily linked to that current star production.
AGN fossil outflows
These refer to outflows in quiescent (non-star-forming) galaxies that might have been driven by an Active Galactic Nucleus (AGN) in the past. These 'fossil' signatures persist even after the AGN has stopped actively feeding, indicating a historical link to AGN feedback.
Main sequence (sSFR/yr−1)
The main sequence represents a baseline relationship between a galaxy's mass and its star formation rate. Galaxies above this line are forming stars rapidly, while those below it are forming stars more slowly. The study found Na D detections were concentrated in galaxies near or above this baseline at lower masses.

Terminology

Summary

A statistical census of neutral interstellar medium (ISM) and outflows in 309 galaxies at redshifts between 0.6 and 4 using JWST/NIRSpec spectroscopy provides a crucial view of how feedback mechanisms regulate galaxy evolution at cosmic noon.

How it works

The study presents a statistical census of the Na D-traced neutral ISM and outflows in 309 galaxies spanning the redshift range of 0.6 < z < 4, utilizing JWST/NIRSpec medium-resolution grating spectroscopy from four surveys: SMILES, JADES, Blue Jay, and Aurora. After subtracting the stellar continuum using penalized pixel fitting (pPXF), the researchers model the Na D doublet at rest-frame 5890 Å and 5896 Å to detect neutral ISM absorption in 76 galaxies.

Key Findings on Na D ISM Incidence

The analysis reveals a dependence of Na D detection on host-galaxy properties:

% of detections are found in massive galaxies (log(M∗/M⊙) > 10), and only 15% in lower-mass systems.

% of detections are found in massive galaxies (log(M∗/M⊙) > 10), spanning the full range from above to below the star-forming main sequence.

% of detections are found only in galaxies on or above the main sequence at the lower-mass end (log(sSFR/yr−1) = -10).

Dependence on Dust Attenuation and Star Formation History

The presence of Na D absorption is strongly correlated with dust attenuation:

% of Na D ISM-detected galaxies have a median AV higher than those without detections.

% of the correlation between AV and EWNaD,ISM is statistically significant with p ≪ 0.05.

Furthermore, the detectability in massive quiescent galaxies depends on stellar population properties:

  1. Na D absorption is preferentially detected in older systems with larger 4000 Å breaks (Dn4000 > 1.4).

  2. It is also detected in younger, rapidly quenching galaxies with strong Balmer absorption HδA, suggesting a link to recent quenching history.

Classification of Outflows and Driving Mechanisms

The study identifies Na D outflows in 26 galaxies, revealing a dichotomy between star-forming and quiescent systems:

% of Na D outflows are found in star-forming galaxies, where properties correlate with star-formation properties, consistent with a star-formation-driven origin.

% of Na D outflows are found in quiescent galaxies, where they are not associated with residual star formation and often require more energy than such star formation can provide.

The AGN fraction among outflow-detected quiescent galaxies is high (46% in the outflow subsample), suggesting that AGN dominate Na D-traced neutral outflows in cosmic noon quiescent systems. The researchers identify five quiescent galaxies with possible AGN fossil outflows, indicating that AGN-driven outflows can persist beyond the active accretion phase.

Outflow Properties and Quenching Link

The physical properties of the 26 Na D outflows are analyzed:

% of Na D outflows in star-forming galaxies have outflow energetics consistent with ongoing star formation.

% of Na D outflows in quiescent galaxies require more energy and momentum than can be supplied by low levels of ongoing star formation.

The comparison between neutral and ionized phase outflows suggests that neutral outflows are more readily detected, and possibly more prevalent, than ionized outflows in massive quiescent galaxies at z ∼ 2, likely due to tracer-dependent detectability. The census supports a picture where Na D-traced neutral outflows in quiescent galaxies are more often linked to AGN feedback, either synchronous in AGN hosts or fossil in systems lacking current AGN signatures.

Comparison with Lower-Redshift Outflows

The cosmic-noon Na D outflows appear offset to higher M˙out and E˙out by roughly 0.3–0.5 dex relative to local ULIRGs and post-starburst galaxies, but this difference is not interpreted as strong evidence for redshift evolution due to potential differences in Na D profile modeling methods. The results highlight the importance of the cool neutral phase for understanding galaxy quenching at cosmic noon, suggesting that Na D outflows are driven by different mechanisms in star-forming and quiescent galaxies.

Conclusion

The census supports a picture where Na D-traced neutral outflows at cosmic noon are state-dependent: in star-forming galaxies, their kinematics and energetics are broadly consistent with being powered by ongoing star formation, whereas in quiescent galaxies they are more often linked to AGN feedback, either synchronous in AGN hosts or fossil in systems lacking current AGN signatures. The study concludes that "AGN fossil outflows may be common among cosmic-noon quiescent galaxies rather than rare phenomena.

Improvements for AI systems

As a fastidious researcher, I have analyzed this scientific paper to extract high-value information that can directly inform and enhance AI systems, particularly those focused on astrophysics, galaxy evolution modeling, and feedback mechanisms.

Here are the specific improvements for an AI system based on this paper:

) Improvement Focus: Galaxy Quenching & Feedback Modeling (Deep Learning/Physics-Informed Neural Networks - PINNs)

The current AI likely struggles with the complex, multi-parameter dependencies of galaxy quenching. The paper provides a detailed empirical framework linking observable features to evolutionary history.

  1. Improvements:

Adaptive training of PINNs using the derived spectral diagnostics as constraints. The AI should be trained not just on raw spectra or photometry, but on synthetic data generated using the Paper's specific modeling pipelines (pPXF for stellar continuum, Prospector for SED fitting). The loss function must be augmented with terms derived from Section 5.1 and 5.1.3:

  • Instead of just minimizing residual spectral error, penalize predictions that violate the observed correlations between indices like [Dn4000] and HδA (Section 5.1.3).

  • Integrate a loss term based on the derived relationship between dust attenuation (AV) and Na D equivalent width (EWNaD,ISM) as defined in Section 5.1.2, ensuring the AI learns that dust shielding is a primary driver for neutral gas observability in star-forming galaxies but not quiescent ones.

  1. Improved System Capability:

The improved AI can perform Quenching Pathway Classification. It could ingest JWST/NIRSpec spectra and immediately classify a galaxy's quenching history into one of the distinct regimes identified:

  • Old Quenched System (High Dn4000, low AV dependence).

  • Rapidly Quenched System (Low Dn4000, high HδA dependence).

This moves beyond simple star-forming/quiescent labels to a nuanced understanding of the quenching mechanism itself.

) Improvement Focus: Outflow Mechanism Discrimination (Classification & Prediction)

The paper establishes a clear dichotomy in outflow drivers based on host galaxy type (star-forming vs. quiescent).

  1. Improvements:

Implement a multi-stage classification pipeline for detected outflows using the derived parameters from Table A4 and Figure 10:

  • Stage 1: Determine Host Type (SF vs. QG) via SED/Morphology fitting.

  • Stage 2: Calculate Mass Loading Factor (η = M˙out/SFR).

  • Stage 3: Classify Driver based on the relationship between Vout, η, and host type. The AI should learn that high Vout coupled with low η in a quiescent host strongly suggests AGN dominance (as seen in Section 6.1), whereas high Vout coupled with moderate/high η in an SF host suggests star-formation driving.

  1. Improved System Capability:

The improved AI can function as an Outflow Driver Diagnostic Tool. Given a JWST observation, it can provide a probabilistic assessment of the outflow origin:

  • If the host is quiescent and AGN is present, the AI predicts an AGN-driven outflow (high energy/momentum consistency).

  • If the host is star-forming and Vout correlates strongly with SFR, it predicts a star-formation driven wind.

This capability directly addresses the ambiguity highlighted in Section 6.1 regarding whether outflows are powered by ongoing SF or fossil AGN activity.

) Improvement Focus: Fossil vs. Active Feedback Modeling (Time Evolution)

The paper introduces the AGN fossil outflow scenario as a crucial mechanism for maintaining quiescence post-quenching, which is difficult to model with standard instantaneous feedback models.

  1. Improvements:

Train a recurrent neural network (RNN) or transformer model on time-series data if available, or use sophisticated attention mechanisms on spectral features that track different stellar ages (Dn4000 vs. HδA). The AI should be specifically tasked with identifying fossil signatures—i.e., residual Na D absorption in quiescent galaxies where the AGN signature is absent, but the outflow energetics are too high for residual SF to explain (Section 6.1).

  1. Improved System Capability:

The improved AI can perform Quiescence Persistence Prediction. For a galaxy with current low SFR and no obvious AGN, it can predict the likelihood of observing a persistent Na D outflow based on its stellar population age distribution (Dn4000) and quenching timescale (HδA). This directly tests the hypothesis that AGN feedback maintains disturbed gas states long after accretion stops.

) Improvement Focus: Observational Bias Mitigation (Robustness Check)

The paper spends significant effort mitigating observational biases related to continuum SNR and slit-loss correction.

  1. Improvements:

Integrate a Bias Correction Module into the input processing layer of the AI pipeline. This module must learn to adjust derived physical parameters (like EWNaD,ISM or Vout) based on the measured local continuum SNR (SNRcont). The system should explicitly learn the relationship between high SNR and plateaued detection rates versus low SNR and noise-limited detections.

  1. Improved System Capability:

The improved AI can provide Bias-Aware Inference. When analyzing a new dataset, it will not just report a result but will also quantify its uncertainty relative to the observational conditions. For example, it can state: "This Na D detection is robust because SNRcont > 5, consistent with the high-SNR regime where detection rates plateau." This moves the AI from producing potentially misleading results to providing scientifically rigorous constraints on its own inferences.

Abstract

We present a statistical census of the Na D-traced neutral interstellar medium (ISM) and outflows in 309 galaxies at 0.6<z<4 using JWST/NIRSpec medium-resolution spectroscopy from the SMILES, JADES, Blue Jay, and Aurora surveys. After subtracting the stellar continuum, we model the Na D λλ5890, 5896 Åand identify 73 Na D ISM absorptions, including 63 robust (4 in broad-line AGNs) and 10 tentative detections. Of the robust detections, 88% are found in massive galaxies ((M*/M)>10), and 12% in lower-mass systems. At high mass, ISM absorption is seen in both star-forming and quiescent galaxies, whereas in lower-mass systems it is observed only in star-forming galaxies. In massive quiescent galaxies, Na D detectability appears linked to star formation history: it is preferentially detected in older systems with larger 4000 Åbreaks, and younger, rapidly quenching galaxies with strong Balmer absorption H δ A. We identify Na D outflows in 25 galaxies, revealing a possible dichotomy in driving mechanisms between star-forming and quiescent galaxies. In star-forming galaxies, outflow properties correlate with star-formation properties, consistent with a star-formation-driven origin. In quiescent galaxies, however, outflows are not associated with residual star formation and often require more energy than such star formation can provide. Together with the high AGN fraction among outflow-detected quiescent galaxies, this suggests that AGN dominate Na D-traced neutral outflows in cosmic noon quiescent systems. We further identify four quiescent galaxies with possible AGN fossil outflows, suggesting that AGN-driven outflows can persist beyond the active accretion phase and may help maintain quiescence.

Sources

Related papers