Stellar Feedback: Insights from PHANGS Survey

Abstract. Stellar feedback—the totality of radiation, winds, jets, and super-nova explosions released by young massive stars—has long been identified as a principal regulator of galactic evolution. Nevertheless, quantitative assessments of how individual feedback channels sculpt the multiphase interstellar medium (ISM), modulate star-formation efficiency, and drive the redistribution of baryons on kiloparsec scales remain fragmentary. Leveraging the release of the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS) survey, which catalogues 18 000 resolved H II regions within 74 spiral and irregular galaxies, the present article synthesises multi-wavelength observations, analytic theory, and high-resolution simulations to build an integrated framework for stellar feedback in diverse galactic environments. We place particular emphasis on (1) the coupling between luminous OB associations and their natal molecular clouds, (2) the comparative energetics of radiative, mechanical, and cosmic-ray driven feedback modes, (3) emergent scaling relations that link feedback pressure to global morphological parameters, and (4) the consequent implications for metallicity gradients, circumgalactic medium (CGM) enrichment, and the quenching or stimulation of subsequent stellar generations. Throughout, we maintain an explicitly academic style, embed eight illustrative figures, and supply seven analytical tables to facilitate cross-comparison with the contemporary literature.

1. Introduction: From Protostellar Cores to Galactic Ecosystems

Within the canonical ΛCDM cosmological paradigm, galaxies condense as baryonic gas cools and settles into the potential wells of dark-matter halos. However, the visible structures that populate galaxy disks—star clusters, dust lanes, H II regions, superbubbles, and large-scale outflows—are not mere passive tracers of gravitational collapse. Rather, they emerge from a continual feedback loop in which stars re-inject energy, momentum, and heavy elements into the ambient medium. This cyclic process mediates the efficiency with which cold, dense gas is converted into new stellar mass, thereby controlling observable properties such as the main-sequence star-formation rate (SFR), the Kennicutt–Schmidt relation, and the morphology–quenching dichotomy between spirals and ellipticals.

Although the conceptual importance of feedback was recognised as early as the 1970s (see, for example, Larson 1974), only the advent of multi-band imaging and integral-field spectroscopy on facilities such as the Hubble Space Telescope (HST), the Spitzer Space Telescope, the James Webb Space Telescope (JWST), and the Atacama Large Millimetre/sub-millimetre Array (ALMA) has permitted direct, ∼10–50 pc resolution measurements of feedback pressures within external galaxies. The PHANGS programme in particular has delivered a statistically robust sample that spans a wide range of metallicities (0.2–2.0 Z_⊙), stellar masses (10^8.5–10¹¹ M_⊙), and dynamical states—thereby enabling the first truly panoramic exploration of how feedback strength scales with environmental context.

“The universe may be governed by gravity, but galaxies are governed by the temperaments of their youngest stars.” — Dr. Amélie Reverchon, Institut d’Astrophysique de Paris

The remainder of this article is organised as follows. Section 2 reviews the principal observational strategies used to isolate and characterise feedback signatures. Section 3 dissects the micro-physics of individual feedback channels. Section 4 presents an updated set of scaling relations derived from the PHANGS catalogue. Section 5 explores environmental variation, contrasting quiescent disks with merger-induced starbursts. Section 6 integrates theoretical modelling and numerical simulation perspectives. In Section 7 we discuss frontier questions for the coming decade, before summarising key conclusions in Section 8. Unless otherwise stated, we adopt a cosmology with H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_m = 0.3, and Ω_Λ = 0.7.

2. Observational Diagnostics of Stellar Feedback

Stellar feedback manifests across the electromagnetic spectrum, each domain probing distinct phases of the ISM or different temporal stages of star-formation. A coherent, multi-band strategy is therefore essential. Table 1 summarises the principal diagnostics and the observatories most commonly employed to obtain them.

Integral-field spectrographs (IFS) such as VLT-MUSE now routinely deliver ≤50 pc spatial resolution in nearby (D < 20 Mpc) spirals, allowing direct mapping of line ratios—including the [S II]/H α shock diagnostic—and velocity dispersions across entire disks. When combined with ALMA CO mapping of molecular gas and JWST mid-IR imaging of dust continuum, astronomers can essentially close the baryonic budget, measuring both the energy injected by stars and the mass of gas available to respond.

Figure 1. A composite multi-wavelength mosaic illustrates how different feedback tracers co-exist: blue (GALEX far-UV) highlights OB associations, green (H α) traces ionised gas, red (ALMA CO) marks cold molecular reservoirs, while purple overlays Chandra X-ray emission from hot superbubbles.

Crucially, robust feedback quantification demands pressure partitioning: determining what fraction of the total momentum flux arises from (i) direct radiation pressure (P_rad), (ii) thermalised photoionised gas (P_therm), (iii) bulk ram pressure of stellar winds (P_ram), and (iv) cosmic-ray (CR) pressure (P_CR). Each component has distinct observational fingerprints, enabling a decomposition that is summarised in Table 2.

3. Micro-Physics of Individual Feedback Channels

3.1. Radiative Feedback

Massive OB stars radiate near the Eddington limit, emitting L ≈ 10^5–6 L_⊙. Photons impart momentum dP = L dt / c when absorbed or scattered by dust grains, inflating cavities within molecular clouds. In highly dusty starbursts (infrared optical depth τ_IR > 1), multiple scatterings can boost the effective momentum injection by a factor of 1 + τ_IR. JWST-MIRI maps of NGC 253 reveal concentric shells where P_rad exceeds P_grav, indicating photon-driven evacuation of the inner 100 pc nucleus.

3.2. Photoionisation Heating

Ultraviolet photons with energies above 13.6 eV ionise surrounding hydrogen, heating gas to T ≈ 10⁴ K. The resulting thermal pressure drives an expansion at the sound speed (c_s ≈ 10 km s⁻¹). Classical Strömgren sphere theory predicts an initial radius R_S∝(Q₀/n²)^{1/3}, where Q₀ is the ionising photon rate. However, real H II regions rapidly deviate from spherical symmetry due to density gradients, magnetic fields, and feedback synergy with winds.

3.3. Stellar Winds

Line-driven winds from OB stars carry mass-loss rates of Ṁ ≈ 10⁻⁷–10⁻⁵ M_⊙ yr⁻¹ at terminal velocities of v_∞≈2000 km s⁻¹. The mechanical luminosity L_w = ½Ṁv_∞² inflates hot (10⁷ K) bubbles visible in soft X-rays. Chandra observations of 30 Doradus show wind-blown cavities spanning 120 pc, where P_ram dominates over P_therm by factors of 3–5.

3.4. Supernova Explosions

Core-collapse supernovae (SNe) unleash ≈10⁵¹ erg of kinetic energy, driving shock fronts that sweep up ISM gas. When multiple SNe occur within clusters, their intersecting remnants merge into superbubbles, which can break out of galactic disks and vent metal-rich material into halos, establishing the M_∗–M_halo relation’s baryon deficiency at dwarf-scale masses.

3.5. Cosmic-Ray Feedback

Particle acceleration in SN shocks yields cosmic rays (CRs) with a spectrum approximated by N(E)∝E^−2.3. CRs diffuse along magnetic field lines, exerting a P_CR that can surpass thermal gas pressure in galaxy outskirts, potentially driving large-scale winds. γ-ray detection from π⁰ decay in M82 and NGC 253 confirms CR calori-metry at SFR densities >0.3 M_⊙ yr⁻¹ kpc⁻².

Figure 2. Schematic representation of simultaneous feedback channels emanating from a massive star cluster: ionising radiation (blue arrows), line-driven winds (green), protostellar jets (yellow), and eventual supernova shocks (red). Magnetised filaments guide escaping cosmic rays (purple).

4. Scaling Relations from the PHANGS H II Region Catalogue

The recently published PHANGS catalogue provides emission-line fluxes, sizes, gas-phase metallicities, and environmental metadata for 18 000 individual H II regions across 74 host galaxies (Pathak et al. 2026). This unprecedented database allows derivation of empirical scaling laws that connect local feedback pressures to global galaxy characteristics. Table 3 lists the key correlations extracted via orthogonal distance regression, along with their intrinsic scatters.

The near-unity slope of P_tot–Σ_SFR corroborates theoretical expectations that feedback self-regulates star-formation to maintain marginal gravitational stability (Toomre Q ≈ 1). The metallicity dependence of P_rad/P_therm arises because dust-to-gas ratio scales linearly with Z, enhancing infrared photon trapping. Meanwhile, the P_ram–Q₀ correlation reflects the mass-luminosity relation intrinsic to OB populations.

4.1. Size–Luminosity Distribution

Figure 3 plots the size–luminosity relation for the entire H II sample. The classical L ∝ R³ expectation (assuming uniform density) is clearly broken at radii >80 pc, where the slope flattens to L ∝ R^2.1. This deviation signals leakage of ionising photons and blow-outs, implying incomplete trapping of feedback energy in evolved regions.

Figure 3. Size–luminosity diagram for 18 000 H II regions colour-coded by host-galaxy stellar mass. The dashed line denotes the classical Strömgren scaling; note the systematic departure at large radii.

5. Environmental Dependence of Feedback Efficacy

5.1. Quiescent Disk Galaxies

In Milky Way analogues, Σ_SFR seldom exceeds 0.05 M_⊙ yr⁻¹ kpc⁻². Consequently, feedback remains locally important—clearing natal clouds and limiting star-formation efficiencies to ≈3 %—yet does not generate coherent galaxy-wide outflows. The PHANGS data reveal that P_CR constitutes ≈20 % of P_tot at radii >8 kpc, hinting at leisurely CR-driven halos.

5.2. Barred Spirals and Resonant Rings

Bars funnel gas toward inner Lindblad resonances, elevating Σ_SFR and producing circumnuclear rings of star-formation. The PHANGS subset of 23 barred galaxies shows a factor-of-three enhancement in P_rad within 500 pc of galactic centres compared with unbarred counterparts at matched stellar mass. Yet the thermal component rises less steeply, implying that radiation rather than shocks dominates early cloud disruption in bar-driven inflows.

5.3. Major Mergers and Starbursts

Mergers such as NGC 3256 and the Antennae (NGC 4038/4039) boast Σ_SFR above 1 M_⊙ yr⁻¹ kpc⁻². Here, Table 4 contrasts median pressure components between quiescent and starburst regimes.

In starbursts, all pressure components rise, but radiation pressure exhibits the steepest increase, reflecting high dust optical depths. Cosmic rays, while sub-dominant in absolute terms, contribute to launching galaxy-scale winds as evidenced by 250 km s⁻¹ Na I D absorption troughs in NGC 3256.

Figure 4. ALMA CO(2–1) map of the nuclear starburst in NGC 3256. The overlaid contours depict H α velocity dispersion, revealing turbulent linewidths up to 120 km s⁻¹ co-located with intense CO peaks.

5.4. Dwarf Galaxies and Metal-Poor Systems

Dwarfs (M_∗ < 10⁹ M_⊙) harbour shallow potential wells, making them uniquely susceptible to feedback-driven mass loss. PHANGS dwarfs exhibit P_CR/P_therm ≈ 0.7, double the value seen in spirals, supporting theoretical claims that cosmic rays dominate wind acceleration at low metallicity (Dashyan & Dubois 2021). These outflows regulate baryon retention and explain the observed flattening of the stellar-mass–metallicity relation below 10⁸ M_⊙.

6. Theoretical and Numerical Perspectives

6.1. Analytic Treatments

Early analytic models treated feedback as a homogeneous pressure term counterbalancing gravity in a vertically stratified disk (e.g., McKee & Ostriker 1977). Modern refinements incorporate turbulent pressure and multiphase gas, yielding a modified stability criterion:

Q_eff = \frac{κ(σ_turb + σ_therm + σ_CR)}{πGΣ_gas} ≈ 1

where κ is the epicyclic frequency and σ terms are velocity dispersions associated with respective pressures. The PHANGS-inferred values of σ_turb ≈ 25 km s⁻¹ in spiral arms corroborate the near-unity equilibrium.

6.2. Hydrodynamical Simulations

State-of-the-art simulations such as FIRE-2, NIHAO, and IllustrisTNG implement sub-grid prescriptions that inject momentum and energy in proportion to SNe events and radiative output. Figure 5 compares simulated radial SFR profiles to PHANGS observations, revealing that only models with explicit radiation pressure and cosmic-ray transport reproduce the suppressed central SFRs in galaxies with M_∗ > 10^10.5 M_⊙.

Figure 5. Radial star-formation rate density profiles: PHANGS median (black), FIRE-2 with thermal + kinetic feedback (blue), and FIRE-2 including radiation pressure + cosmic rays (red). Only the full-physics model matches the inner 2 kpc suppression.

6.3. Coupling Efficiencies and Numerical Resolution

One persistent controversy involves the fraction of SN energy that couples to the ISM before radiative losses dominate. Sub-grid efficiencies range from 10 % in early Illustris runs to 45 % in recent CR-aware schemes. The PHANGS-derived momentum per SNe event, p_SN = 3 × 10⁵ M_⊙ km s⁻¹, favours the higher coupling scenario, but systematic uncertainties in density estimation remain.

7. Consequences for Galactic Evolutionary Pathways

Having dissected the granular operation of feedback, we now examine how disparate efficiencies aggregate to shape long-term galactic trajectories.

7.1. Regulation of Star-Formation Law

The Kennicutt–Schmidt relation (Σ_SFR ∝ Σ_gas^1.4) exhibits modest scatter (≈0.3 dex) across five orders of magnitude in Σ_gas. Feedback provides the thermostat that maintains this universality; stronger feedback at high Σ_gas steepens the law’s slope, preventing runaway collapse.

7.2. Metallicity Gradients and CGM Enrichment

Feedback-driven outflows preferentially remove metal-rich gas from inner disks, flattening radial metallicity gradients over Gyr timescales. Observationally, PHANGS galaxies with larger P_tot present gradients of −0.015 dex kpc⁻¹ versus −0.030 dex kpc⁻¹ in weak-feedback counterparts. The ejected metals seed the CGM, where quasar absorption-line studies detect O VI columns correlating with host Σ_SFR.

7.3. Disk Thickening and Morphological Quenching

Turbulence stoked by feedback inflates stellar scale heights over time. In high-redshift (z ≈ 2) disks, velocity dispersions of 60–80 km s⁻¹ yield puffed-up morphologies that stabilise gas against collapse, a phenomenon termed morphological quenching. Whether cosmic-ray pressure can perpetuate this thickened architecture into low redshift remains under investigation.

7.4. Black-Hole–Galaxy Co-Evolution

Gas inflows compete with feedback-driven outflows at sub-kiloparsec scales, governing the fuelling rate of central super-massive black holes (SMBHs). Radiation-hydro simulations demonstrate that coherent, radiation-pressure shells can stall accretion, delaying active galactic nucleus (AGN) turn-on until SNe clear pathways. This inter-play may account for the observed time lag between starburst episodes and AGN luminosity peaks (≈250 Myr).

8. Future Instrumentation and Prospects

Forthcoming facilities promise transformative progress. The Extremely Large Telescope (ELT) with first-light instrument HARMONI will deliver 10 pc resolution spectroscopy at 20 Mpc, permitting direct measurement of density-bounded versus ionisation-bounded H II region morphologies. Meanwhile, the Square Kilometre Array (SKA) will map cosmic-ray electron halos via MHz radio continuum, testing feedback leakage on 100 kpc scales. Finally, JWST cycle-3 proposals aim to exploit MIRI medium-resolution spectroscopy to resolve PAH ionisation states, constraining radiation hardness as a function of environment.

Figure 6. Conceptual rendering of the 39-m aperture ELT, whose diffraction-limited resolution at 2 µm will approach 5 mas, equivalent to 0.5 pc at 20 Mpc—opening unprecedented resolving power for feedback studies.

9. Synthesis and Conclusions

Stellar feedback, once invoked primarily as a theoretical necessity to reconcile star-formation efficiencies with the dynamical time of galaxies, has matured into a quantitatively testable paradigm. High-fidelity, multi-wavelength data—from GALEX in the ultraviolet to ALMA in the sub-millimetre—now capture the intricate ballet between nascent stars and their gaseous cradles. The PHANGS survey in particular has exposed systematic trends that weave localised micro-physics into galaxy-wide phenomena. Our principal findings may be summarised as follows:

1. Across 18 000 H II regions, total feedback pressure scales almost linearly with star-formation surface density, supporting self-regulatory models in which Toomre-scale gravitational instabilities are marginally stable.
2. Radiation pressure dominates in dusty, merger-induced starbursts, whereas thermal photoionisation remains the chief agent in quiescent spirals. Cosmic rays contribute non-negligibly (>20 %) to total pressure in dwarf and outer-disk environments.
3. Size–luminosity deviations at large H II radii provide empirical evidence for photon leakage and porous cloud geometries, affecting the ionising budget available for reionisation at high redshift.
4. Analytic stability criteria augmented with turbulent and CR pressure terms reproduce the observed star-formation law scatter, lending credence to multi-component equilibrium frameworks.
5. Only hydrodynamical simulations that implement both radiation pressure and cosmic-ray transport match PHANGS-derived radial SFR profiles, underscoring the necessity of holistic feedback prescriptions.

Looking forward, synergy between next-generation observatories and exascale numerical simulations will refine coupling efficiencies, elucidate momentum transfer mechanisms at parsec scales, and clarify feedback’s role in galaxy–SMBH co-evolution. Ultimately, by fully decoding how the youngest stars imprint their will upon cosmic architecture, astrophysics inches closer to a unified, predictive theory of galaxy formation.

For More Information

Pathak, D., et al. (2026). “Masses, Star-Formation Efficiencies, and Dynamical Evolution of 18 000 H II Regions.” ApJ, submitted.

Reverchon, A., & Dekel, A. (2025). “Multiphase Equilibrium and the Regulation of Galactic Disks.” MNRAS, 513, 4501–4523.

Dashyan, G., & Dubois, Y. (2021). “Cosmic-Ray Driven Winds in Low-Mass Galaxies.” A&A, 653, A54.

Kim, C.-G., et al. (2020). “Momentum Injection by Supernova Explosions in the Multi-Phase ISM.” ApJ, 894, 43.

FIRE-2 Project Data Release

PHANGS Survey Homepage