CGM/IGM Feedback Physics Weekly: July 20-27, 2026

Five qualifying arXiv preprints submitted July 20–23 examine feedback through different parts of the gas cycle: a simulated quasar reshaping the CGM at z ≳ 6, O VI absorption across the starburst-to-quiescent transition, neutral outflows at z > 3, and two tests of whether current AGN prescriptions can match both halo gas and galaxy populations. Read together, they point to a practical modeling constraint: the observable imprint depends on where energy couples, which gas phase records it, and how long that phase survives. The integrated energy budget does not determine those observables by itself.

No qualifying standalone GitHub release or public astronomy dataset release was verified for this window, so those categories remain empty.

What it does: The GOLIATH survey studies the multiphase CGM of massive blue starburst and post-starburst galaxies at mean redshift about 0.43, using O VI absorption in the inner halo at R/Rvir ≤ 0.6. 1

Method: The analysis compares O VI column densities, line widths, masses, and covering fractions across star-forming and quiescent systems. The sample is aimed at the warm-hot CGM rather than the full halo, so the spatial selection is part of the result.

Finding: In the log(M★/M☉) = [11, 12) bin, star-forming galaxies have roughly three times the O VI column density of quiescent galaxies, about 1.5 dex more O VI mass (log(MO VI/M☉) ≈ 7.3 versus 5.8), and a 62.5% versus 24% covering-fraction contrast above log N(O VI) = 14. The inferred cooling time is only about 10–100 Myr. 1

Why it matters: A short cooling time makes O VI a clock as well as a temperature tracer. Sustaining the observed reservoir requires continued replenishment by active feedback, which is a stronger constraint on duty cycle and coupling than a one-time energy injection would be.

Caveat: The feedback interpretation is consistent with the O VI line widths and columns, not a direct causal measurement. Residual O VI in quiescent galaxies may come from ambient gas near the hot end of the cooling curve, and the result is limited to massive systems and the inner CGM. 1

What it does: Tortora et al. use zoom-in simulations of a massive protocluster at z ~ 6, comparing the fiducial FABLE model with variants that seed black holes earlier and allow mildly super-Eddington accretion. A new ray-tracing calculation follows the resulting emission. 2

Method: The simulations follow the host ISM and CGM through a strong quasar-feedback or “blow-out” phase, then use radiative transfer to predict neutral-hydrogen structure and Hα nebulae.

Finding: In the modeled system, strong feedback opens escape channels for ionizing radiation and drives metal-enriched outflows that produce cold, fast, neutral clumps. The covering fraction of neutral hydrogen increases, while the size and luminosity of the CGM Hα nebula grow with feedback strength and fall with obscuration. 2

Why it matters: This paper gives observers a linked set of targets rather than a single feedback proxy: obscuration, neutral-hydrogen covering fraction, and extended Hα emission respond together in the simulation. It is a useful bridge between subgrid quasar physics and the multiphase CGM JWST can probe.

Caveat: The result comes from one massive z ~ 6 protocluster setup and specific changes to black-hole seeding and accretion. It predicts CGM signatures in that environment; it does not establish a universal quasar-outflow response or a direct IGM measurement. 2

What it does: Sapori et al. analyze 811 galaxies at z > 3 in low-resolution JWST/NIRSpec PRISM spectra, using Na I D absorption to search for neutral outflows during the first two billion years of cosmic history. 3

Method: Prospector SED fits provide physical properties and star-formation histories. The authors subtract the fitted stellar continuum, measure Na I D equivalent widths, and stack subsamples before estimating mass outflow rates under conservative assumptions.

Finding: Na I D excess absorption appears in 20 galaxies, an overall detection fraction of about 2.5%. The fraction rises to about 11% among the most massive galaxies and about 43% among massive quenched systems. Equivalent widths span 4–16 Å, and the inferred mass outflow rates are 4–12 M☉ yr⁻¹; for about 30% of detections, the inferred rate exceeds the current star-formation rate. 3

Why it matters: The mass and quiescence dependence places neutral gas expulsion in the same early epoch where massive galaxies are assembling and shutting down. It supplies a high-redshift observational target for the clumpy-CGM predictions in the quasar simulation, without requiring the two studies to be treated as the same population.

Caveat: PRISM’s R ~ 100 resolution supplies no direct outflow kinematics. The outflow interpretation is based on Na I D equivalent widths and comparison with prior medium-resolution observations, so the velocities and rates remain model-dependent rather than directly measured. 3

What it does: Suresh and Blanton compare radiative-mode AGN predictions from EAGLE, SIMBA, and TNG100 with observational constraints on the fraction of galaxies hosting a radiative AGN above an Eddington ratio of 10⁻³. They also compare EAGLE’s combined radiative-plus-radio feedback energy with observational estimates. 4

Method: The comparison maps the AGN-host fraction as a function of stellar mass and specific star-formation rate. It then tests whether the simulated energy accounting agrees with observed total feedback after adopting stated bolometric and jet-power conversion factors.

Finding: None of the three simulations reproduces the observed radiative-AGN demographic trends even qualitatively. For EAGLE with a constant coupling efficiency of 0.15, the total feedback energy is broadly consistent with observations under the adopted conversions, but the simulated Eddington ratios are about ten times higher than observed. 4

Why it matters: This is the cleanest warning in the set against treating an integrated energy budget as a sufficient validation target. A model can land near the right total while distributing activity across the wrong host population or accretion states, changing when and where the CGM receives energy.

Caveat: EAGLE does not explicitly separate radio and radiative modes, and the comparison depends on the adopted conversion factors. The authors therefore call for more detailed comparison before drawing firm physical conclusions. 4

What it does: Mazengo et al. compare Magneticum, IllustrisTNG, and SIMBA with spatially resolved MaNGA data and hot-gas fraction–halo-mass constraints from eROSITA and Sunyaev–Zeldovich measurements. 5

Method: The comparison places galaxy properties and halo thermodynamics on the same calibration target instead of checking one population statistic at a time.

Finding: Matching the observed hot-gas fraction–halo-mass relation requires strong AGN feedback in the comparison. Magneticum and SIMBA match that relation but produce too many quenched galaxies; IllustrisTNG’s weaker AGN feedback gives more realistic star-forming fractions but overpredicts hot gas in massive groups and poor clusters. 5

Why it matters: The result turns “feedback strength” into a multidimensional calibration problem. Total energy is only one axis; timing, location, and coupling to surrounding gas determine whether the same prescription regulates galaxies and the halo reservoir at once.

Caveat: The comparison is based on three existing simulation suites and the observational constraints named in the abstract. It identifies an unresolved tradeoff rather than a unique replacement prescription; the authors point to a companion study on whether the required feedback strengths inevitably overquench galaxies. 5

Start with GOLIATH if the goal is a direct CGM constraint: its inner-halo contrast and 10–100 Myr cooling time make the phase and timescale explicit. Pair the two AGN-model papers next; one tests host-population statistics and energy accounting, while the other tests galaxy quenching against halo gas fractions. The quasar simulation then supplies a high-redshift prediction for how those coupling choices may appear in neutral clumps and Hα, while the JWST study offers an observational route to early neutral outflows but leaves kinematics unresolved.

The combined lesson is deliberately narrower than a field-wide trend. These five papers support a chain that feedback models must preserve: coupling determines phase and location; cooling and replenishment determine the lifetime of the tracer; the tracer then constrains the model. Matching only the total injected energy can miss the observable physics.