This arXiv preprint (arXiv:2604.13154, submitted April 2026) from Zirui Chen and collaborators is not yet peer-reviewed. It uses controlled 3D hydrodynamic simulations—comparing planar turbulent radiative mixing layers against turbulent-box setups under identical microphysical conditions (same cooling curves, thermal conduction, and turbulent driving)—to demonstrate that gas geometry, not just microphysics, sets the temperature probability distribution function (PDF) in multi-phase media. The methodology involves a limited set of high-resolution numerical experiments rather than a large statistical sample; limitations include idealized setups that omit magnetic fields, self-gravity, and active star formation, restricting direct applicability to real galactic environments.

Previous CGM literature has leaned heavily on analytic planar mixing-layer models, assuming somewhat universal temperature PDFs that successfully match certain line ratios and phase fractions. ISM studies, by contrast, routinely produce broad PDFs in turbulent boxes but have struggled to provide clean theoretical grounding. What much of this prior work missed or oversimplified is the morphological transition: cold gas in turbulent conditions does not remain in static sheets. Instead, it forms clumps whose surface-area-to-volume ratios increase with temperature until interfaces percolate into connected warm sheets. The PDF can be decomposed as the product of isosurface area (set by large-scale geometry) and layer thickness (controlled by cooling and conduction, well-captured by existing mixing-layer theory). This geometric effect produces the broad intermediate-temperature mass fractions long seen in simulations but never fully explained.

Synthesizing this with related work clarifies the advance. Tumlinson et al. (2017, arXiv:1704.00418) documented an unexpectedly large reservoir of O VI in the low-redshift CGM using Hubble's COS-Halos survey—ions that exist primarily at intermediate temperatures (~10^5.5 K). Standard mixing-layer models underpredict this reservoir; the clumpy-to-sheet transition identified by Chen et al. naturally supplies more gas at these temperatures. Similarly, Fielding et al. (2020, arXiv:2006.00005) modeled radiative turbulent mixing layers and emphasized microphysical regulation of cooling lengths. The new preprint builds directly on that foundation but reveals its geometric blind spot: when turbulence is allowed to sculpt 3D morphology, the resulting PDFs diverge dramatically from planar predictions.

A third connection appears in jellyfish galaxies, where ram-pressure stripping creates tails with tight X-ray to Hα correlations. These observations have resisted simple mixing-layer interpretations; the broad temperature PDFs arising from percolating clump interfaces offer a coherent explanation for co-spatial hot and warm gas.

The implications reach cosmology's foundational processes. Modern galaxy-formation simulations (IllustrisTNG, EAGLE, FIRE) rely on sub-grid prescriptions for feedback, cooling, and star formation. If temperature structure—and therefore cooling rates and observational tracers—is geometry-dependent, many predicted quantities (stellar masses, metal distributions, outflow velocities) may carry systematic errors. Thermally unstable gas at ~10^4–10^5 K in the ISM, long puzzling because isobaric cooling theory predicts rapid phase transitions, can persist when turbulent geometry continuously replenishes intermediate-temperature interfaces.

Thus the preprint reframes a fundamental control knob in multi-phase astrophysics. Galaxy evolution, star formation efficiency, and feedback loops—all pillars of the current cosmological paradigm—are sensitive to whether gas lives in isolated clumps or extended sheets. Future observational campaigns with JWST, ELT, and SKA can test these geometric signatures through spatially resolved line-ratio mapping. The work supplies both a diagnostic and a modeling imperative: resolve or parameterize gas morphology, or risk mischaracterizing the thermal backbone of cosmic structure formation.