A statistical study of ionized outflows in low-mass galaxies from DESI DR1

arXiv:2611.15884 · astro-ph.GA · Submitted 2026-11-11 · 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: We're looking at the paper: "A statistical study of ionized outflows in low-mass galaxies from DESI DR1".

Jocelyn: Here's the gist of the paper. A statistical study of ionized outflows in low-mass galaxies from DESI DR1 presents a statistical analysis of ionized gas outflows in a large sample of low-mass galaxies at z < 0.45 from the first data release of the Dark Energy Spectroscopic Instrument survey (DESI DR1) Using

O III: and Hα emission lines, we identify ∼ 5800 dwarf galaxies exhibiting robust signatures of ionized outflows Most are star-forming systems located above the starforming main sequence, with a smaller fraction of composites and AGN hosts The derived mass-loading factors span η = 0.001 − 1, with a median η ∼ 0.01, confirming previous findings from smaller datasets at similar stellar masses These values are 1–2 dex lower than predicted by cosmological simulations, suggesting that the ionized phase traces only a minor fraction of the total outflow mass budget Identification of Outflow Candidates The identification of candidate ionized outflows involved cross-matching the dwarf parent sample with the EmFit catalog to obtain 636,086 galaxies To identify galaxies hosting candidate ionized outflows, we cross-matched the dwarf parent sample with the EmFit catalog to obtain 636,086 galaxies We focused on the

O III: λ5007 emission line as it is a good tracer of ionized outflowing gas in the optical and is less affected by contamination from nearby emission lines or AGN broad-line region emission We selected the sources for which this secondary component was classified as an outflow according to the criteria described in Pucha et al.

Vera: (2025) This shrinks the sample to 7621 candidate outflow-host sources We subsequently cross-matched these two candidate sets to ensure a more robust outflow classification, resulting in a final sample of 5781 dwarf galaxies exhibiting clear evidence of ionized outflowing gas in both emission lines Quantification of Outflow Rates and Efficiency To investigate the impact of ionized outflows in dwarf galaxies, we estimate here the outflow efficiency, or mass-loading factor (η), defined as the ratio between the gas mass-loss due to outflow (i.e., the outflow rate M˙out) and star formation (i.e., η ≡ M˙out/SFR) For consistency with other works from the literature, we assume here C = 1, implying a constant outflow rate over time Since we do not have direct spatial information on the outflow extent, we assume rout = 1 kpc, motivated by spatially-resolved observations of ionized outflows in nearby star-forming and dwarf galaxies Following Rupke et al.

Jocelyn: (2005), we compute the outflow velocities directly from the spectra as vout = vbroad − vnarrow + 2σbroad,corr The estimated outflow velocities are in the range log(vout/km s−1) ∼ 1.8−3.2, similar to what found in local dwarf sources (e.g., Marasco et al.

Vera: 2023) Following Carniani et al.

Paper summary: Vera: The paper starts by explaining their identification process, which involved cross-matching the dwarf parent sample with the EmFit catalog to get over six hundred thirty-six thousand galaxies. They focused on using O III lambda five hundred seven as a tracer because it’s good for ionized outflowing gas and less affected by other nearby emission lines or AGN broad-line regions.

Jocelyn: They started by selecting sources based on criteria from Pucha et al two thousand twenty-five which narrowed the sample down to seven thousand candidate outflow-host sources. Then they cross-matched those two sets to make it a final sample of five thousand seven hundred eighty one dwarf galaxies exhibiting clear evidence of ionized outflowing gas in both emission lines.

Subrahmanyan: So, the methodology is pretty rigorous here, moving from a broad catalog to a specific tracer and then cross-validating the results. That systematic approach is crucial when you're trying to pin down something like an outflow signature in noisy data.

Vera: Moving on to quantification, they estimate the outflow efficiency, or mass-loading factor eta. This is defined as the ratio between the gas mass loss due to outflow, which they call M dot out, and star formation rate SFR. They assume a constant outflow rate over time and set that characteristic scale for rout to one kiloparsec based on observations of nearby galaxies.

Jocelyn: They then calculate the mass of the outflowing gas using both O III and H alpha emission lines, giving them Mout for each tracer, using equations nine and ten from their paper. The resulting outflow efficiencies they find are in the range of zero point zero zero one to one, with a median value around zero point one <ref:2611.15884#pg1>.

Subrahmanyan: That median value of zero point one confirms what we see in other works from smaller samples, like those by Concas et al two thousand twenty-two and Marasco et al two thousand twenty-three <ref:2611.15884#pg3>. It’s consistent across different distances and galaxy types, which is a good sign for the general physics.

Vera: The paper also addresses whether these outflows escape their gravitational potential to enrich the circumgalactic medium or even the intergalactic medium. They compare their outflow velocities with the escape velocities associated with galactic disks and dark matter halos.

Jocelyn: They found that a large fraction of these galaxies have outflow velocities comparable to or larger than the disk escape velocity, vesc disk, which jumps from about seventy eight percent for star-forming galaxies up to eighty eight percent for both composite systems and AGN hosts.

Subrahmanyan: That high velocity component suggesting escape into the circumgalactic medium is strong evidence that the ionized gas isn't just staying trapped locally in these small halos. But they note something else, which is important for the next part of their argument.

Vera: They do find that only about eleven percent of star-forming galaxies have a velocity greater than the halo escape velocity, vesc halo, compared to fifty six percent for composite systems and sixty eight percent for AGN hosts.

Jocelyn: So it looks like star formation drives outflows that mostly redistribute gas into the circumgalactic medium, but AGN activity can more efficiently accelerate gas to velocities capable of escaping the entire dark matter halo and potentially enriching the intergalactic medium.

Conclusion: Vera: To wrap up, this statistical study of ionized outflows in low-mass galaxies from DESI DR1 really shows us a few things. The paper confirms that the mass-loading factor eta decreases as stellar mass increases, which qualitatively matches the behavior predicted by cosmological simulations.

Jocelyn: They also note that there's a discrepancy between their results and those simulations. They suggest this difference could be due to systematic uncertainties in how they estimate eta, which is something we always have to keep in mind when comparing observations to theory.

Subrahmanyan: From a theoretical standpoint, the finding that the ionized phase traces only a minor fraction of the total outflow mass budget is significant for our models of galaxy evolution. It means the primary feedback mechanism might not be as energetic in these low-mass galaxies as some earlier models proposed.

Vera: And this paper gives us a clearer picture about where that ejected gas goes, too. We see that star formation drives the gas into the circumgalactic medium, while AGN activity seems more effective at launching it out into the intergalactic medium.

Jocelyn: Overall, "A statistical study of ionized outflows in low-mass galaxies from DESI DR1" provides a solid observational handle on how these small galaxies regulate their gas through outflows and where that material ends up in the cosmic web.

Subrahmanyan: It’s an important piece of the puzzle for understanding structure formation because it constrains the efficiency of feedback mechanisms across different galaxy scales.

M. Romano, A. Saintonge, C. Circosta, C. Cicone, D. Donevski, D. Scholte, B. Hagedorn

Max-Planck-Institut für Radioastronomie · INAF - Osservatorio Astronomico di Padova, University College London Department of Physics and Astronomy, Institut de Radioastronomie Millimétrique (IRAM), Institute of Theoretical Astrophysics, National Center for Nuclear Research · SISSA · INAF - Osservatorio Astronomico di Trieste, Institute for Astronomy, University of Edinburgh

astro-ph.GA

Submitted: 2026-11-11

Updated: 2026-11-11

Comments: Accepted for publication in A&A

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

Importance score: 91/100

The gist: A statistical study of ionized outflows in low-mass galaxies from DESI DR1 presents a statistical analysis of ionized gas outflows in a large sample of low-mass galaxies at z < 0.45 from the first…

Key concepts

Mass-loading factor ($\eta$)
This measures how much gas is being ejected by an outflow relative to the rate at which stars are forming in a galaxy. A low value, like the median 0.01 found here, means that for every unit of star formation, only a tiny fraction of the total mass is being blown out as an ionized outflow.
Outflow Velocity ($v_{out}$)
This is calculated from spectral line broadening to determine how fast the gas is moving away from the galaxy. The study found these velocities are generally in the range of $\log(v_{out}/km s^{-1}) \sim 1.8-3.2$, similar to what is observed in local dwarf galaxies.
Escape Velocity ($v_{esc}$)
This represents the speed required for gas to overcome the gravitational pull of a galaxy's disk or dark matter halo and escape into space. The study found that high-velocity outflows can exceed this velocity, suggesting they can escape the central galaxy's potential.

Terminology

Summary

A statistical study of ionized outflows in low-mass galaxies from DESI DR1 presents a statistical analysis of ionized gas outflows in a large sample of low-mass galaxies at z < 0. p. 0

Where the paper says it

O Box 1029, Blindern, 0315 Oslo, Norway 6 National Center for Nuclear Research, Pasteura 7, 02-093 Warsaw, Poland 7 SISSA, Via Bonomea 265, 34136 Trieste, Italy 8 INAF - Osservatorio Astronomico di Trieste, via Tiepolo 11, I-34131, Trieste, Italy 9 Institute for Astronomy, University of Edinburgh, Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ, UK November 13, 2026 ABSTRACT We present a statistical analysis of ionized gas outflows in a large sample of low-mass (log(M∗/M⊙) ≲ 10) galaxies at z < 0. 45 from the first data release of the Dark Energy Spectroscopic Instrument survey (DESI DR1).

45 from the first data release of the Dark Energy Spectroscopic Instrument survey (DESI DR1) Using [O III] and Hα emission lines, we identify ∼ 5800 dwarf galaxies exhibiting robust signatures of ionized outflows Most are star-forming systems located above the starforming main sequence, with a smaller fraction of composites and AGN hosts The derived mass-loading factors span η = 0. p. 0
Where the paper says it

Using [O III] and Hα emission lines, we identify ∼ 5800 dwarf galaxies exhibiting robust signatures of ionized outflows. Most are star-forming systems located above the starforming main sequence, with a smaller fraction of composites and AGN hosts.

001 − 1, with a median η ∼ 0. p. 0
Where the paper says it

The derived mass-loading factors span η = 0. 001 − 1, with a median η ∼ 0.

01, confirming previous findings from smaller datasets at similar stellar masses These values are 1–2 dex lower than predicted by cosmological simulations, suggesting that the ionized phase traces only a minor fraction of the total outflow mass budget

Identification of Outflow Candidates

The identification of candidate ionized outflows involved cross-matching the dwarf parent sample with the EmFit catalog to obtain 636,086 galaxies To identify galaxies hosting candidate ionized outflows, we cross-matched the dwarf parent sample with the EmFit catalog to obtain 636,086 galaxies We focused on the [O III]λ5007 emission line as it is a good tracer of ionized outflowing gas in the optical and is less affected by contamination from nearby emission lines or AGN broad-line region emission We selected the sources for which this secondary component was classified as an outflow according to the criteria described in Pucha et al. (2025) This shrinks the sample to 7621 candidate outflow-host sources We subsequently cross-matched these two candidate sets to ensure a more robust outflow classification, resulting in a final sample of 5781 dwarf galaxies exhibiting clear evidence of ionized outflowing gas in both emission lines

Quantification of Outflow Rates and Efficiency

To investigate the impact of ionized outflows in dwarf galaxies, we estimate here the outflow efficiency, or mass-loading factor (η), defined as the ratio between the gas mass-loss due to outflow (i.e., the outflow rate M˙out) and star formation (i. p. 0

Where the paper says it

By comparing individual outflow detections with stacked spectra of the general low-mass star-forming galaxy population from DESI, we show that mass-loading factors inferred from the stacks are systematically lower than those of individual detections, but the difference decreases toward larger offsets above the star-forming main sequence. Together with a positive correlation between mass-loading factor and specific star-formation rate, this suggests that enhanced star formation and burstiness play an important role in regulating feedback efficiency and shaping galactic winds.

e., η ≡ M˙out/SFR) For consistency with other works from the literature, we assume here C = 1, implying a constant outflow rate over time Since we do not have direct spatial information on the outflow extent, we assume rout = 1 kpc, motivated by spatially-resolved observations of ionized outflows in nearby star-forming and dwarf galaxies Following Rupke et al. p. 5
Where the paper says it

2012; Cicone et al. Since we do not have direct spatial information on the outflow extent, we assume rout = 1 kpc, motivated by spatiallyresolved observations of ionized outflows in nearby star-forming and dwarf galaxies.

(2005), we compute the outflow velocities directly from the spectra as vout = vbroad − vnarrow + 2σbroad,corr The estimated outflow velocities are in the range log(vout/km s−1) ∼ 1. p. 6
Where the paper says it

Following Rupke et al. (2005), we compute the outflow velocities directly from the spectra, as: vout = |vbroad − vnarrow| + 2σbroad,corr, (8) where vbroad and vnarrow are the velocity peaks of the broad and narrow components, respectively, and σbroad,corr is the velocity dispersion of the broad component deconvolved by the instrumental resolution.

8−3.2, similar to what found in local dwarf sources (e. p. 6
Where the paper says it

The estimated outflow velocities are in the range log(vout/km s−1 ) ∼ 1. 2, similar to what found in local dwarf sources (e.

g., Marasco et al. p. 0
Where the paper says it

, McCormick et al. 2018; Marasco et al.

2023) Following Carniani et al. p. 6
Where the paper says it

, Marasco et al. Following Carniani et al.

(2024), we estimated the mass of the outflowing gas from both [O III] and Hα, respectively as Mout,[O III] = 0. p. 6
Where the paper says it

Following Carniani et al. (2024), we estimated the mass of the outflowing gas from both [O III] and Hα, respectively as: Mout,[O III] = 0.

8 × 108 Lbroad[O III]1044 erg/s ZoutZ⊙! p. 6
Where the paper says it

(2024), we estimated the mass of the outflowing gas from both [O III] and Hα, respectively as: Mout,[O III] = 0. 8 × 108   L broad [O III] 1044 erg/s   Zout Z⊙ !

−1nout500 cm−3−1M⊙, (9) and Mout,Hα = 0. p. 6
Where the paper says it

Following Carniani et al. (2024), we estimated the mass of the outflowing gas from both [O III] and Hα, respectively as: Mout,[O III] = 0.

8 × 105 LbroadHα1040 erg/snout100 cm−3−1M⊙, (10) We found outflow efficiencies in the range ∼ 0. p. 7
Where the paper says it

4 the ionized outflow rate as a function of the SFR for the individual outflow candidates, as well as the parameter space covered by the outflows’ mass loading factors. We found outflow efficiencies in the range ∼ 0.

001−1, with a median value η ∼ 0. p. 7
Where the paper says it

We found outflow efficiencies in the range ∼ 0. 001−1, with a median value η ∼ 0.

01, in agreement with findings from smaller samples of low-mass nearby and more distant sources (e. p. 7
Where the paper says it

001−1, with a median value η ∼ 0. 01, in agreement with findings from smaller samples of low-mass nearby and more distant sources (e.

g., Concas et al. p. 1
Where the paper says it

2020; Calabrò et al. 2022; Concas et al.

p. 1
Where the paper says it

2020; Calabrò et al. 2022; Concas et al.

2022; Marasco et al. p. 7
Where the paper says it

, Concas et al. 2022; Marasco et al.

p. 7
Where the paper says it

, Concas et al. 2022; Marasco et al.

2023; Hagedorn et al. p. 1
Where the paper says it

2023; Carniani et al. 2024; Hagedorn et al.

p. 1
Where the paper says it

2023; Carniani et al. 2024; Hagedorn et al.

2026)

Escape into the Circumgalactic Medium and Intergalactic Medium

To investigate whether the ionized outflows in our galaxies are able to escape their gravitational potential enriching the CGM (or possibly even the IGM), we compare their velocities with the escape velocities associated with the galactic disk and darkmatter halo We find that a large fraction of our galaxies have outflow velocities comparable to or larger than vesc,disk, with the fraction satisfying vout > vesc,disk increasing from ∼ 78% for SFGs to ∼ 88% for both composite systems and AGNs This suggests that the high-velocity ionized component can frequently escape the central galaxy and reach the CGM However, only ∼ 11% of star-forming galaxies have vout > vesc,halo, compared to ∼ 56% and ∼ 68% of composite systems and AGNs This suggests that starformation-driven outflows predominantly redistribute ionized gas into the CGM, while AGN activity can more efficiently accelerate gas to velocities capable of escaping the halo and potentially enriching the IGM

Comparison with Previous Studies and Simulations

We show in Fig. 6 the variation of η with the stellar mass, and compare our measurements with previous observational and theoretical studies covering a similar parameter space Our results are broadly consistent with recent studies of ionized outflows in both local and z ∼ 2 dwarf galaxies (e. p. 8

Where the paper says it

Comparison with previous studies We show in Fig. 6 the variation of η with the stellar mass, and compare our measurements with previous observational and theoretical studies covering a similar parameter space.

g., Concas et al. 2022; Marasco et al. 2023; Hagedorn et al. 2026) We find that the mass-loading factor decreases with increasing stellar mass, qualitatively reproducing the behaviour predicted by cosmological simulations The discrepancy between our results and cosmological simulations could be in principle alleviated if considering systematic uncertainties in the observational estimates of η

Comparison with the Average Star-Forming Population

The analysis presented so far is based on individually detected outflows and therefore preferentially selects galaxies in which the broad component is sufficiently prominent to be identified in individual DESI spectra The resulting average mass-loading factors are shown as large diamonds in Fig.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Statistical analysis of ionized outflows in low-mass galaxies can be performed using DESI DR1 data to identify approximately 5800 dwarf galaxies exhibiting robust signatures of ionized outflows using [O III] and Hα emission lines.

  2. The improved AI system can derive mass-loading factors spanning eta equals 0.001 to 1, with a median eta around 0.01, confirming previous findings from smaller datasets at similar stellar masses while noting these values are one to two dex lower than predicted by cosmological simulations.

  3. The system can predict that the ionized phase traces only a minor fraction of the total outflow mass budget based on current DESI DR1 results, suggesting that ionized gas outflows play a significant role in regulating star formation and the baryon cycle of low-mass galaxies.

  4. The AI can model how mass-loading factors decrease with stellar mass, broadly reproducing simulation trends despite a strong normalization offset, and identify that enhanced star formation and burstiness play an important role in regulating feedback efficiency and shaping galactic winds.

  5. The system can compare individual outflow detections with stacked spectra of the general low-mass star-forming galaxy population from DESI to show that mass-loading factors inferred from stacks are systematically lower than those of individual detections, but this difference decreases toward larger offsets above the star-forming main sequence.

  6. The AI can establish a positive correlation between mass-loading factor and specific star formation rate, suggesting that enhanced star formation and burstiness play an important role in regulating feedback efficiency and shaping galactic winds.

  7. The system can differentiate outflow properties based on galaxy classification, finding that AGNs drive stronger outflows with larger mass-loading factors and mass rates per fixed SFR as compared to SFGs, as indicated by the slope of the log(Mout out) equals a times log(SFR) plus b relation being flatter and higher for AGNs.

  8. The AI can determine that a large fraction of galaxies host high-velocity ionized gas capable of reaching the circumgalactic medium (CGM), ranging from about 78% of star-forming galaxies to about 88% of composite systems and AGNs, indicating that the high-velocity ionized component in a substantially larger fraction of composite and AGN systems reaches velocities sufficient to escape the halo and potentially enrich the IGM.

  9. The system can identify that while "ionized outflows can efficiently transport material from the galaxy into the CGM, most of the ionized gas associated with star-formation-driven outflows is expected to remain gravitationally bound to the halo and may eventually be recycled onto the galaxy."

Abstract

We present a statistical analysis of ionized gas outflows in a large sample of low-mass (log(M*/M) 10) galaxies at z<0.45 from the first data release of the Dark Energy Spectroscopic Instrument (DESI DR1). Using [OIII] and H α emission lines, we identify about5800 dwarf galaxies with robust signatures of ionized outflows. Most are star-forming systems above the star-forming main sequence, with a smaller fraction of composites and AGN hosts. The derived mass-loading factors span η=0.001-1, with a median η about0.01, consistent with previous studies at similar stellar masses. These values are 1 -- 2 dex lower than predicted by cosmological simulations, suggesting that the ionized phase traces only a minor fraction of the total outflow mass. Outflow velocities are sufficient to enrich the circumgalactic medium in 80% of cases, while escape into the intergalactic medium appears efficient mainly in composite and AGN hosts. We find that mass-loading factors decrease with stellar mass, broadly reproducing simulated trends despite the normalization offset. Comparing individual detections with stacked spectra of the general low-mass star-forming population, we find systematically lower mass-loading factors in the stacks, although the difference decreases toward larger offsets above the star-forming main sequence. Together with a positive correlation between mass-loading factor and specific star-formation rate, this suggests that enhanced star formation and burstiness regulate feedback efficiency and shape galactic winds. Our results highlight the role of ionized outflows in the baryon cycle of low-mass galaxies and the need for multiphase studies to fully understand feedback in galaxy evolution.

Sources

Related papers