Radio Spectral Energy Distribution of Low-$z$ Metal Poor Extreme Starburst Galaxies: Novel insights on the escape of ionizing photons
Extracted source text · arXiv 2503.17327
% aa.dem
% AA vers. 9.1, LaTeX class for Astronomy & Astrophysics
% demonstration file
% (c) EDP Sciences
%-----------------------------------------------------------------------
%
\documentclass{aa}
%\documentclass[referee]{aa} % for a referee version
%\documentclass[onecolumn]{aa} % for a paper on 1 column
%\documentclass[longauth]{aa} % for the long lists of affiliations
%\documentclass[letter]{aa} % for the letters
%\documentclass[bibyear]{aa} % if the references are not structured
% according to the author-year natbib style
%
%\documentclass{aa}
%
\usepackage{graphicx}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage{txfonts}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage[colorlinks = true,
linkcolor = blue,
urlcolor = blue,
citecolor = blue,
anchorcolor = blue]{hyperref}
\usepackage{subfig}
\usepackage{subcaption}
%\usepackage[round,authoryear]{natbib}
%\usepackage{multirow}
% To add links in your PDF file, use the package "hyperref"
% with options according to your LaTeX or PDFLaTeX drivers.
%
\newcommand{\gmrt}{uGMRT}
\newcommand{\vla}{VLA}
\newcommand{\lofar}{LOFAR}
%\newcommand{\yarp}{\textsc{RadioBricks}}
\newcommand{\yarp}{\textsc{YARP}}
\newcommand{\python}{\texttt{Python}}
\newcommand{\aips}{\textsc{AIPS}}
\newcommand{\etal}{et al.\ }
\newcommand{\fesc}{$f_\mathrm{esc}^\mathrm{LyC}$}
%\newcommand{\alphanth}{$\alpha_\mathrm{nth}$}
\newcommand{\alphanth}{\ifmmode \alpha_\mathrm{nth} \else $\alpha_\mathrm{nth}$\fi}
\newcommand{\alphacs}{$\alpha^{\mathrm{3GHz}}_\mathrm{6GHz}$}
\newcommand{\Oratio}{O$_{32}$}
\newcommand{\sfrdensity}{$\Sigma_\mathrm{SFR}$}
\newcommand{\ssfrdensity}{$\Sigma_\mathrm{sSFR}$}
\newcommand{\betatwo}{$\beta_{1550}$}
\newcommand{\hi}{H{\sc i}}
\newcommand{\hii}{H{\sc ii}}
\newcommand{\RCth}{RC$_\mathrm{th}$}
\newcommand{\RCnth}{RC$_\mathrm{nth}$}
\newcommand{\ewhalpha}{EW(H$\alpha$)}
\newcommand{\ewhbeta}{EW(H$\beta$)}
%\newcommand{\fth}{f$_\mathrm{th}$}
\newcommand{\fth}{\ifmmode \mathrm{f}_{1 \mathrm{GHz}}^\mathrm{th} \else f$_{1 \mathrm{GHz}}^\mathrm{th}$\fi}
\newcommand{\nuturn}{$\nu_\mathrm{t}$}
\newcommand{\odepth}{$\tau_\nu^{\mathrm{ff}}$}
\newcommand{\nion}{$N_e$}
%\newcommand{\odepth}{\ifmmode \tau_\nu^{\mathrm{ff}} \else $\tau_\nu^{\mathrm{ff}}$ }
\newcommand{\SFRUV}{SFR$_\mathrm{UV}$}
\newcommand{\SFRHbeta}{SFR$_{\mathrm{H}\beta$}}
\newcommand{\comment}{\color{blue} }
\newcommand{\revtext}[1]{\textbf{#1}}
\newcommand{\SFRRC}[1]{SFR$_{\mathrm{#1}}^{\mathrm{RC}}$}
\newcommand{\mstar}{$\mathrm{M_*}/\mathrm{M_\odot}$}
\newcommand{\rcsed}[1]{radio-SED{#1}}
\newcommand{\xsfg}[1]{xSFG{#1}}
\newcommand{\bbten}{J150934+373146}
\newcommand{\bbthirteen}{J160810+352809}
\newcommand{\sbs}{SBS 0335-052}
\newcommand{\izw}{I Zw 18}
\newcommand{\ebv}{\textit{E(B-V)}}
\newcommand{\hbeta}{H~$\beta$}
\newcommand{\hbetaradio}{ $\mathrm{Flux}^\mathrm{Exp}_{\mathrm{H}\beta}$ }
\newcommand{\specindex}{$\alpha^{\mathrm{low}}_\mathrm{high}$}
%\Lya{} velocity peak separation (\vsep{}
\newcommand{\Lya}{Ly~$\alpha$}
\newcommand{\vsep}{$V_\mathrm{sep}$}
\defcitealias{Hunt04}{H04}
\defcitealias{Sebastian19}{SB19}
%\newcommand{\J1032}{J1032+4919}
%\newcommand{\J1355}{J1355+4651}
%\newcommand{\J0159}{J0159+0751}
%\newcommand{\J2229}{J2229+2725}
%\newcommand{\revtext}[1]{#1} % to produce a clean copy without bold face
%\newcommand{\ebvuv}{}
%\newcommand{\arcsec}{$^{\prime\prime}$}
\begin{document}
%\title{Extreme radio continuum suppression in the most metal deficient compact star forming galaxies }
\title{Radio Spectral Energy Distribution of Low-$z$ Metal Poor Extreme Starburst Galaxies:}
\subtitle{Novel insights on the escape of ionizing photons}
\author{
Omkar Bait
\inst{1,2}
\fnmsep\thanks{O. Bait is the corresponding author}
\and
Daniel Schaerer
\inst{1,3}
\and
Yuri I. Izotov
\inst{4}
\and
Biny Sebastian
\inst{5}
}
%\institute{Observatoire de Gen\`eve, Universit\'e de Gen\`eve, Chemin Pegasi, 1290 Versoix, Switzerland\\
% \email{omkar.bait@unige.ch}
\institute{Observatoire de Gen\`eve, Universit\'e de Gen\`eve, Chemin Pegasi, 1290 Versoix, Switzerland \\
\email{omkar.bait@unige.ch}
\and
SKA Observatory, Jodrell Bank, Lower Withington, Macclesfield, SK11 9FT, UK\\
\email{omkar.bait@skao.int}
\and
CNRS, IRAP, 14 Avenue E. Belin, 31400 Toulouse, France
\and
Bogolyubov Institute for Theoretical Physics, 14-b Metrolohichna str, Kyiv 03143, Ukraine
\and
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA
}
\date{Received; accepted}
% \abstract{}{}{}{}{}
% 5 {} token are mandatory
\abstract
% context heading (optional)
% {} leave it empty if necessary
{Recent optical surveys have found a rare population of low-$z$ ($z \sim 0.01 - 0.06$) extreme star-forming galaxies (\xsfg{s}) that are the most metal-poor galaxies with strong emission lines, extremely high specific star-formation rate and low stellar mass. Using the Hubble Space Telescope Cosmic Origins Spectrograph, it was found that \xsfg{s} are strong Ly~$\alpha$ emitters (LAEs). Their Ly~$\alpha$ properties indirectly suggest that they are also strong LyC leakers. This along with several other global properties makes them similar to the recently found $z > 6$ reionization-era star-forming galaxies using the James Webb Space Telescope. }
% aims heading (mandatory)
{Here we aim to study the radio spectral energy distribution (\rcsed{}) of $8$ \xsfg{s} to understand mechanisms behind the extreme nature of star-formation in these galaxies, particularly their high ionisation state and possible strong LyC leakage. Using \rcsed{} analysis we can constrain the thermal/non-thermal fraction which allows us to study the role of supernova and cosmic-ray feedback in driving the extreme nature of \xsfg{s} and also constrain the density of the interstellar medium (ISM). The radio thermal emission can also be used to study the presence of dust, particularly that which could be missing in the optical.}
% methods heading (mandatory)
{We present new multi-frequency radio continuum (RC) observations using a combination of upgraded Giant Metrewave Radio Telescope (uGMRT), the Karl G. Jansky Very Large Array (VLA) together with archival LOw Frequency ARray (LOFAR) survey data for a few sources. {We performed RC observations using the \gmrt{} at Band-$5$ ($1060 - 1460$ MHz) together with \vla{} S- ($2-4$ GHz), C- ($4-8$ GHz), X- ($8-12$ GHz) and Ku- ($12-18$ GHz) bands.} The VLA S- and C-band observations were performed in B-configuration, whereas the X- and Ku-band observations were performed in C-configuration. Such a multi-configuration setup allows us to study the \rcsed{} in a roughly spatially matched resolution of $3 \sim 2$\arcsec{}. For a few sources we also present LOFAR 120-168 MHz observations at $6$\arcsec{} resolution. Using these RC observations, we map the \rcsed{} of $8$ \xsfg{s} in a wide frequency range from $\sim 1$ GHz (from $150$ MHz for a few sources) - $15$ GHz, up to $\sim2$ orders of magnitude in frequency. }
% results heading (mandatory)
{The \rcsed{} of \xsfg{s} is flat between 6-15 GHz and shows a strong evidence for a turnover at lower frequencies in the range of $2-10$ GHz. They can be well described using a thermally dominated radio spectra with a free-free absorption component with a high emission measure. Thus the star-forming complexes in these galaxies are extremely young (pre-SNe stage; below $\sim 5$ Myrs) and the interstellar medium is very dense. This, along with the overall shape of the \rcsed, suggests a dominance of young massive star clusters (YMCs) in these galaxies. One source, \bbten{}, is an exception to this trend and instead has a steep spectrum below $10$~GHz and shows a break at higher frequencies. A comparison between the radio free-free emission at $10$~GHz with that observed from the \hbeta{} emission line in the optical provides evidence of dust in several \xsfg{s}. In a few cases there is evidence for a small fraction of the ionizing emission of the galaxy being not visible in the Balmer lines, which is possibly due to being deeply embedded in YMCs. We confirm previously found dependence between Lyman continuum escape fraction (\fesc{}), ionisation state and the radio spectral index, particularly for strong leakers.
}
% conclusions heading (optional), leave it empty if necessary
{Our \rcsed{} analysis suggests that a dominance of YMCs in metal poor extreme starburst galaxies leads to their several extreme star-forming properties along with potentially leaking a significant amount of hydrogen ionising photons.
}
\keywords{galaxies: starburst -- Radio continuum: galaxies -- Radio continuum: ISM
}
\maketitle
%
%-------------------------------------------------------------------
\section{Introduction}
\label{sec: introduction}
%start from RC --> describe some recent studies
% RCSED gives insights nsights on dust, SF (thermal), SN, CR and magnetic fields and their dependece on physical properties.
% Here we study rcsed of metal poor galaxies.
% metal poor galaxies + highly star-forming galaxies --> why are they interesting in general
% introduce compact SFGs --> some general interests -> high fesc important for reionization etc
% -> JWST results etc
% -> now come to the radio studies -> highlight your work (Bait+24)
% -> show the Biny suppression of BBs + GPs. Briefly mention what that meant etc
% - > Now motivate your study, xSFGs and end it
Identifying the nature of sources (at redshift $z > 6$) that led to the reionization of the Universe remains a matter of intense debate. Numerous studies, both on the observational and simulation side, suggest that low-mass dwarf starburst galaxies are likely the dominant sources of reionization \citep[e.g.,][]{Ouchi09, Bouwens15, Robertson15, Atek15}. With the advent of the James Webb Space Telescope (JWST), rest-frame optical spectroscopic studies of reionization-era galaxies ($z \sim 6 - 8$) have become possible revealing their key physical properties. It is found that these galaxies generally have low stellar masses ($\log$ \mstar{} $\leq 9$), compact size ($< 1$ kpc), and high specific star-formation rates sSFR of $\sim 100$ Gyr$^{-1}$ \citep{Schaerer22, Rhoads23, Williams23, Endsley23, Fujimoto23_JWST,Mascia24}. Several of these galaxies are extremely metal-poor (XMPs), with gas-phase metallicities $ 12 + \log\mathrm{O/H} < 8.0$ and have hard ionizing radiation \citep[][]{Jones20, Schaerer22, Rhoads23, Langeroodi23, Topping24, Saxena24, Atek24}. And importantly, they are found to be strong \Lya{} emitters (LAEs) and also produce a significant amount of ionizing radiation to easily reionize the Universe \citep{Simmonds23, Simmonds24, Saxena24, Atek24}.
Despite these significant advances in our understanding of high-$z$ galaxies, the physical mechanisms responsible for their extreme star-forming properties and high LyC escape responsible for reionization remain poorly understood due to a lack of multi-wavelength studies, which is extremely difficult with the current generation telescopes. Moreover, due to optically thick inter-galactic medium it is extremely difficult to detect LyC escape from high-$z$ galaxies \citep{Inoue14}. We thus have to rely on the local analogs of such high-$z$ galaxies with similar physical properties, e.g., low-mass, compact sizes, high sSFR, low-metallicity and strong \Lya{} emission etc. \citep[e.g.,][]{cardamone2009, yang2017, Izotov11, Izotov21a}, generally found in the Sloan Digital Sky Survey (SDSS) data. These low-$z$ ($z \sim 0.3$) analogs are known to produce a significant amount of LyC photons \citep{Schaerer16, Izotov21a} and importantly a large fraction of them leak \citep{Borthakur14, Izotov16b, Izotov16c, Izotov18a, Izotov18b-vsep, Izotov21b}. These successes in finding local LyC leakers or emitters (LCEs) have led to the recent systematic study of the Low-$z$ Lyman Continuum Survey \citep[LzLCS;][]{Flury22a} which has revealed interesting relations between the LyC escape fraction (\fesc{}) and several key galaxy physical properties \citep{Flury22b, Saldana-Lopez22, Xu23, Jaskot24-multivariate, Bait24a}.
Recently, a rare population of extreme star-forming galaxies (\xsfg{s}) were discovered with the SDSS which are very compact XMPs having even lower stellar mass ($\log$ \mstar{} $\sim 5 - 8$), high sSFRs ($\sim 300~ $Gyr$^{-1}$) and high [OIII] $\lambda5007$ /[OII] $\lambda3727$ ratios; \Oratio $\geq 5$, with several in the range of 20-50 \citep{Izotov17b, Izotov20-diverseLyA, Izotov24-LyA-metalpoor, Jaskot19}. They show diverse \Lya{} properties with several sources showing narrow \Lya{} peak velocity separation (\vsep{}) \citep{Izotov20-diverseLyA, Izotov24-LyA-metalpoor}. Such a narrow \vsep{} suggests that they are also strong LyC leakers based on the tight relation between \vsep{} and \fesc{} \citep{Verhamme15_vsep, Izotov18b-vsep}. Here we focus on the multi-frequency radio continuum (RC) observations of \xsfg{s} to gain new insights on the physical mechanisms driving their extreme SF and potentially strong \fesc{}.
RC emission from galaxies in the megahertz (MHz)- to gigahertz (GHz) frequency range has provided important insights on the star-formation process \citep{condon1992}. RC emission has been found to be an excellent tracer of extinction-free star-formation rate (SFR) in galaxies \citep[e.g.,][ and references therein]{murphy2011}. The RC emission in normal SF galaxies consists of a combination of a flat thermal emission component (free-free emission with a spectral index of $-0.1$) and a non-thermal emission component (non-thermal spectral index \alphanth{} $\sim -0.8$). The non-thermal component arises from synchrotron emission from cosmic-ray particles (CRs) accelerated by supernova (SNe) under magnetic fields. Thus the non-thermal emission can be used to infer the SNe rate, cosmic-ray (CR) energy distribution and the equipartition magnetic fields. The radio spectral energy distribution (\rcsed{}) of normal SF galaxies shows a power-law behaviour due to a combination of the thermal and non-thermal component. It is generally found that at $\sim 1.4$ GHz the non-thermal component is dominant comprising of almost 90\% of the emission component \citep[e.g.,][]{Niklas97_thermal_frac, tabatabaei2017}. Detailed multi-frequency \rcsed{} analysis of nearby SF galaxies has allowed to constrain the relative fraction of thermal (\fth{}) and non-thermal emission, and \alphanth{} \citep{tabatabaei2017} and its dependence on various galaxy physical parameters (e.g., the SFR surface density, magnetic fields etc.). Additionally in luminous and ultraluminous infrared galaxies (ULIRGs), there can be a turnover at lower frequencies in the \rcsed{} due to free-free absorption (FFA) effects \citep[e.g,][]{Clemens10, Galvin18}. Using a combination of thermal and non-thermal radio emission and FFA model, these studies find a high emission measure (EM) in ULIRGs in the range of $\sim 10^6 - 10^7$ pc cm$^{-6}$ \citep{Clemens10, Galvin18, Dey24_ULIRGS}.
RC studies of dwarf galaxies have revealed different properties compared to massive SF galaxies. The thermal fraction (\fth{}) is higher in dwarf galaxies, particularly in blue compact dwarfs \citep{Klein91, Deeg93, Thuan2004, ramya2011}. A \rcsed{} study of nearby dwarf galaxies from $\sim 0.3-24$ GHz have shown that they cannot be described by a single power-law and instead show a cuttoff/break at higher frequencies due to various physical processes related to the lack of CRs \citep{klein2018}. More relevant to our study here, for example in an extremely metal-poor compact dwarf starburst, SBS 0335-052, the integrated \rcsed{} shows a strong turnover at GHz frequencies, arising due to FFA with a EM $\sim 2 - 8 \times 10^7$ pc cm$^{-6}$ and compact size ($\sim 17$ pc) \citep[][hereafter: \citetalias{Hunt04}]{Hunt04}. Such a high EM is more commonly observed in resolved radio studies of ultra dense \hii{} regions comprising of young deeply embedded star clusters \citep[e.g.,][]{Kobulnicky99, Johnson03}. Previous single frequency (at $1.4$ GHz and $150$ MHz) RC observations of objects similar to \xsfg{s} (e.g., green-peas, blueberries) have found that these systems have lower RC flux than that expected from the standard RC-SFR relation \citep{chakraborti2012, Sebastian19, Borkar24-BB-LOFAR}. However, due to limited frequency coverage it is difficult to discern the exact physical mechanisms responsible for such a suppression. Finally, \citet{Bait24a} in a recent radio follow-up study of a sub-sample of LzLCS sources found that the \fesc{} (and the \Oratio{}, SFR density) depends on the radio spectral index (specifically measured between $3 - 6$ GHz). These results motivate a detailed \rcsed{} study of LCEs to better understand the physical processes governing these relations.
Here, we present new multi-frequency \rcsed{} observations of a sample of eight \xsfg{s} selected from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor} and \citet{Jaskot19}. Our multi-frequency RC observations were conducted using the NRAO\footnote{The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.} Karl G. Jansky Very Large Array (\vla) at 6 (C-band), 3 (S-band), and (a subsample) at 1.5 (L-band) GHz. We also combine new observations with the upgraded Giant Metrewave Radio Telescope (uGMRT) at Band-5 (1060 - 1460 MHz) together with archival LOw Frequency ARray (LOFAR) data at 120-168 MHz data using the LOFAR Two-metre Sky Survey Data-Release 2 \citep[LoTSS DR2;][]{Shimwell17_LoTSS-DR1, Shimwell22_LoTSS-DR2}.
Our paper is organized as follows. In Section \ref{sec: data} we describe our sample, VLA and uGMRT observation setup, data analysis and RC flux density measurements in different radio bands. In Section \ref{sec: results} we first present the observed \rcsed{} of our sample and the various model \rcsed{s} that represent our sample. We then present the dust attenuation measurement using our radio observations. In Section \ref{sec: discussion}, we discuss, 1) the cause(s) of the lack of non-thermal emission in \xsfg{s} based on our well-sampled \rcsed{}, 2) the relationship between \fesc{}, \Oratio{}, metallicity, and radio spectral index for \xsfg{s} and 3) the overall implications of our observed \rcsed{} to better understand the extreme nature of high-$z$ galaxies found using the JWST. We then summarize our main results and conclude in Section \ref{sec: conclusions}.
%Throughout this paper we assume $H_0=70$ km s$^{-1}$\ Mpc$^{-1}$, $\Omega_m=0.3$, and $\Omega_\Lambda=0.7$
%--------------------------------------------------------------------
\section{Data}
\label{sec: data}
In this section, we describe our sample selection and new \vla{} observations at multiple observing bands, namely: the S (2 - 4 GHz), C (4 - 8 GHz), X (8 - 12 GHz) and Ku (12-18 GHz) bands. We support these observations with new low frequency data using the \gmrt{} at Band-5 (1060 - 1460 MHz). We also present archival \lofar{} data (120 - 168 MHz) covering most of our sample. We also describe our calibration and imaging strategy.
\subsection{Sample}
\begin{table*}[ht!]
\caption{Basic properties of our sample of \xsfg{s}.} %\\
\label{table: basic properties}
\centering
\begin{tabular}{ccccccccc}
\hline
Target & RA & DEC & $z$ & $12+\log\mathrm{O/H}$ & $\log$ \mstar{} & \Oratio{}$^a$ & SFR$^b$ & Lit. source \\
& (deg) & (deg) & & & & & (M$_\odot$/yr) & \\
\hline \hline \\
J0159+0751 & $29.96979$ & $+7.86356$ & $0.06105$ & $7.56$ & $8.6$ & $39.0$ & $2.3$ & \citet{Izotov20-diverseLyA} \\
J0811+4730 & $122.96721$ & $+47.50729$ & $0.0444$ & $6.98$ & $5.88$ & $5.9$ & $0.19$ & \citet{Izotov24-LyA-metalpoor} \\
J0820+5431 & $125.08029$ & $+54.52781$ & $0.03851$ & $7.49$ & $6.0$ & $22.0$ & $0.2$ & \citet{Izotov20-diverseLyA}\\
J1032+4919 & $158.23633$ & $+49.32979$ & $0.0442$ & $7.59$ & $6.8$ & $24.0$ & $2.4$ & \citet{Izotov20-diverseLyA} \\
J1355+4651 & $208.85692$ & $+46.86426$ & $0.02811$ & $7.56$ & $5.8$ & $23.0$ & $0.1$ & \citet{Izotov20-diverseLyA} \\
J150934+373146 & $227.39239$ & $+37.52948$ & $0.0325$ & $7.88$ & $8.1$ & $15.1$ & $1.35$ & \citet{Jaskot19} \\
J160810+352809 & $242.04318$ & $+35.46926$ & $0.0327$ & $7.83$ & $7.5$ & $34.9$ & $0.46$ & \citet{Jaskot19} \\
J2229+2725 & $337.38829$ & $+27.42378$ & $0.07622$ & $7.08$ & $6.96$ & $53.0$ & $0.68$ & \citet{Izotov24-LyA-metalpoor} \\
\hline \\
\end{tabular}
\tablefoot{The physical properties presented here are taken from the literature values shown in the last column. \\
$^a$ \Oratio{} $\equiv$ [OIII] $\lambda5007$ /[OII] $\lambda3727$. \\
$^b$ The SFR is derived using the dust-corrected \hbeta{} line using the relation from \citet{Kennicutt98}. For \bbten{} and \bbthirteen{} the H$\alpha$ line was used and the relation from \citet{Kennicutt12}. See the references (last column) for details.
}
\end{table*}
% Now contains the 8 sources which are the high priority sources
Our sample is selected from the recently identified sample of \xsfg{s} at low-$z$, ~$z \sim 0.01 - 0.06$, from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor} and \citet{Jaskot19} \citep[see also][]{yang2017}. All our sources have extremely low-metallicity ($12+\log\mathrm{O/H} \sim 6.98 - 7.88$), low stellar mass ($\log$ (\mstar{}) $\sim 5.8 - 8.6$), very high \Oratio{} ($\gtrsim 15$ to up to $\sim 53$) and high \hbeta{} equivalent width (\ewhbeta{} $\gtrsim 250~\AA$). We observed several of these sources at S- and C-bands using the \vla{} with the B-configuration (PI: Bait; Project Code: 23A-138 \& 24A-103) and at higher frequency bands at X- and Ku with the C-configuration (PI: Bait; Project Code: 24A-103). We also conducted follow-up \gmrt{} Band-5 observations (PI: Bait, Proposal ID: 43\_062 and 45\_075) for our sample of \xsfg{s}.
Here we aim to study the \rcsed{} primarily at GHz frequencies. We thus present a sample of 8 \xsfg{s} which has a seamless RC coverage between $\sim 1 - 12$~GHz (S- to X- band) from the two \vla{} and \gmrt{} programs. Except for one source (J0159+0751), all our sources also have Ku-band ($12 - 18$~GHz) observations (Project Code: 24A-103). For two sources, \bbten{} and \bbthirteen{}, we use the published Band-5 data from \citet{Sebastian19} (\citetalias{Sebastian19} hereafter). J1032+4919 was also observed with the \vla{} at L-band ($1 - 2$~GHz; under Project Code: 23A-138).
%Additionally, for \bbten{}, we present lower-frequency \gmrt{} Band-3 (250 - 500 MHz) observations(PI: Sebastian, Proposal code: 38\_010).
We also present \lofar{} 150 MHz data which covers 7 out of the 8 sources from our sample. The various physical properties (e.g., metallicity, \Oratio{}, \vsep{}, observed and dust extinction corrected \hbeta{} line flux) of the 8 \xsfg{s} studied here are taken from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor} and \citet{Jaskot19} and shown in Table \ref{table: basic properties}.
In the next section, we present details on the new RC observations at multiple \vla{} and \gmrt{} bands together with archival \lofar{} observations. Table \ref{table: obs summary} provides a brief summary of these observations.
\subsection{VLA L-, S-, C-, X- and Ku-band Observations and Data Analysis}
\label{sec: vla data}
We used the default \vla{} setup for our L-, S-, and C-band continuum observations. For the L- and S-bands we used a total bandwidth of 1 and 2 GHz respectively with a 8-bit sampler. The L-band (S-band) was split into one (two) baseband(s) each comprising of 8 $\times$ 128 MHz subbands. For the C-band we used a total bandwidth of 4 GHz with a 3-bit sampler which was split into two basebands each comprising of 16 $\times$ 128 MHz subbands. For each of these cases, the subbands were further split into 64 channels. All of these observations were performed in the B-configuration. This setup is identical to the previously followed setup for RC observations of the LzLCS sources at L-, S- and C-bands from \citet{Bait24a}. We adopted the standard observing strategy used for continuum observations at L-, S- and C-bands which was also followed for the LzLCS sources \citep[see][for details]{Bait24a}. Briefly, after an initial setup, we observed the flux density scale calibrator (for $3 -5$ mins). We then alternative between the phase calibrator ($\sim 1 $min per cycle) and the target ($\sim 5 - 6$ mins per cycle). The S- and C-band observations were conducted within a single scheduling block (SB) lasting approximately $\sim$1~hr. We achieved an on-source integration time of 22 mins each for S- and C-band. For the L-band SBs typically lasted for a duration of $\sim$1~hr and 40 mins, where we achieved an on-source integration time of $\sim 75$~mins.
For the high-frequency X- and Ku-band observations we again used the default \vla{} setup, with a total bandwidth of 4 GHz and 6 GHz respectively with a 3-bit sampler. The total bandwidth for the X-band (Ku-band) was split in 2 (3) subbands each comprising 16 $\times$ 128 MHz subbands. Here as well each subband was further split into 64 channels. These observations were performed in the C-configuration. After an initial setup, we observed the flux density scale calibrator for approximately 5 mins. For the Ku-band an additional scan of $\sim$4 mins for pointing calibration was performed, where the phase calibrator was used as a pointing calibrator. We then alternative between the phase calibrator ($\sim 2$ mins per cycle) and the target ($\sim 8$ mins per cycle). We achieved an on-source integration time of $\sim 15$ mins for both the X- and Ku-band.
For all our \vla{} data reduction, we used the Common Astronomy Software Applications (\textsc{CASA}) data processing software \citep{mcmullin2007casa, CASA}. We followed a similar data reduction procedure as described in \citet{Bait24a}. Briefly, each SB was initially flagged and calibrated using the standard VLA calibration pipeline v6.5.4. We manually inspected the calibrated flux density scale and phase calibrators for each SB. Any leftover RFI and poorly performing antennas were manually flagged. And the \vla{} pipeline was then re-run.
We then extracted the calibrated target and inspected for any leftover RFI in some spectral windows, which was then further flagged. Then we proceeded with the imaging of these data using CASA \textsc{tclean}. For S- and C-bands, since these were observed in the multi-band mode, we first split the target in S- and C-band before imaging \citep[see][]{Bait24a}). For all our imaging, CASA \textsc{tclean} was setup to use the multi-term multi-frequency synthesis (mtmfs) algorithm \citep{Rau11} where the \textsc{nterms} were set to 2 and a \textsc{robust} value of 0.5 was used. The flux density and corresponding error ($\sigma$) was then estimated by manually making elliptical regions around each of the target sources using CASA task \textsc{imview}, and a 2D Gaussian fit was performed using the task \textsc{imfit} \citep[see][for more details on this procedure]{Bait24a}. For one marginally resolved source in our sample, \bbten{} which is also relatively extended in the optical compared to other \xsfg{s}, the flux density and error was measured using PyBDSF \citep[the Python Blob Detector and Source Finder;][]{Mohan15}. Table \ref{table: radio fluxes} presents the flux density, noise ($\sigma$). For non-detections, where sources have a signal-to-noise ratio (S/N) below 3, we present the 3$\sigma$ upper limits as the flux density. Table \ref{table: radio fluxes} also shows the synthesized beam properties (beam major/minor axis and position-angle) from our observations in various bands. Note that for all further analysis on the RC flux density, spectral index and \rcsed{} analysis we add a 5\% calibration error in quadrature \citep[following][]{Bait24a}.
\subsection{\gmrt{} Band-5 Observations and Data Analysis}
\label{sec: gmrt data}
For the \gmrt{} Band-5 (1460 - 1060 MHz) observations, we used a total bandwidth of 400 MHz split in 2048 channels, similar to that used in \citetalias{Sebastian19}.
%For the Band-3 (500 - 300 MHz) observations, we used a total bandwidth of 200 MHz split in 2048 channels.
The data reduction was carried out using the \yarp{}\footnote{The current development version of the \yarp{} package can be found here: \url{https://github.com/omkarbait/yarp/}} package. \yarp{} comprises a collection of high-level \python{} recipes to perform flagging, calibration and imaging which are based on various standard CASA tasks at the lowest-level. The different flagging, calibration and imaging recipes can be combined to design custom radio data reduction pipelines/workflows for both continuum and spectral line data. The \yarp{} package was first used for GMRT \hi{} 21-cm spectral-line data reduction in \citet{Paudel23}, and a detailed paper describing the package is in preparation.
Here we use the \yarp{} package with CASA v6.5.5 to design a custom pipeline for our \gmrt{} Band-5 continuum data reduction. Briefly, in this pipeline the raw \gmrt{} data initially underwent a round of flagging using the CASA task \textsc{tfcrop}. Here any bad/non-working antennas were also flagged which was manually provided to \yarp{}. This data then underwent 5 rounds of calibration (flux, phase and bandpass) and flagging (which used a combination of CASA tasks \textsc{tfcrop, rflag} and \textsc{clip}). The calibrated data for phase calibrator, flux density scale calibrator and target was manually inspected for any leftover RFI. We then extracted the calibrated target and self-calibrated the data. Each self-calibration round includes three rounds of phase-only self-calibration and one round of Amplitude and Phase self-calibration. For a few cases, the residual (data - model) was flagged using CASA task \textsc{rflag} and the flags were applied to the data before proceeding to another round of self-calibration. This final calibrated data was then used for imaging using the {\sc mtmfs} algorithm with \textsc{nterms} set to 2. For all our new Band-5 observations presented here, we do not detect any emission and thus we present the $3\sigma$ upper limits. For two \xsfg{s}, \bbten{} and \bbthirteen{}, the flux density and noise are taken from \citetalias{Sebastian19}. Note that J1032+4919 was also observed by \citetalias{Sebastian19} and was also not detected. Our upper limits for J1032+4919 are consistent with \citetalias{Sebastian19}. Table \ref{table: radio fluxes} presents the flux properties for our sample.
\subsection{Archival LoTSS DR2 data}
\label{sec: lofar data}
We also present archival LoTSS (120-168 MHz) DR2 imaging data \citep{Shimwell22_LoTSS-DR2} for our sample. We download cutouts around each of the sources in our sample using the LoTSS DR2 cutout service \footnote{\url{https://lofar-surveys.org/dr2_release.html}}. We inspected each image to search for detections. Out of the 7 sources which have LoTSS coverage, we detected emission in only one source, \bbten{}. The flux was estimated using PyBDSF. For the rest of the sources we present the $3\sigma$ upper limits (see Table \ref{table: radio fluxes}).
\begin{table}[ht!]
\caption{Brief summary of the new multi-frequency radio observations.} %\\
\label{table: obs summary}
\centering
\begin{tabular}{cccccc}
\hline
Tel. & Band & Cen. Freq.$^a$ & BW$^b$ & TOS$^c$ & Array$^d$ \\
& & (GHz) & (GHz) & (mins) & \\
\hline \hline \\
%\gmrt{} & Band-3 & 0.400 & 0.25 & \textcolor{red}{XX} & Full \\
\gmrt{} & Band-5 & 1.25 & 0.4 & 120 & Full \\
\vla{} & L & 1.5 & 1 & 75 & B \\
\vla{} & S$^e$ & 3.0 & 2 & 20 & B \\
\vla{} & C$^e$ & 6.0 & 4 & 20 & B \\
\vla{} & X & 10.0 & 4 & 15 & C \\
\vla{} & Ku & 15.0 & 6 & 15 & C \\
\hline
\end{tabular}
\tablefoot{$^a$ Central frequency of the observing band. \\
$^b$ Total Bandwidth. \\
$^c$ Time on source (approximate values). \\
$^d$ \vla{} configuration. \\
$^e$The S- and C-band observations were performed simultaneously in a multi-band mode.
}
\end{table}
\begin{sidewaystable*}
\caption{Radio Fluxes of \xsfg{}.} %\\
\label{table: radio fluxes}
%Target & obs_flux_gmrt_L & obs_rms_gmrt_L & obs_flux_vla_L & obs_rms_vla_L & obs_flux_S & obs_rms_S & obs_flux_C & obs_rms_C & obs_flux_X & obs_rms_X & obs_flux_Ku & obs_rms_Ku & obs_flux_LOFAR & obs_rms_LOFAR & Beam Size GMRT Band-5 & Beam Size VLA L & Beam Size S & Beam Size C & Beam Size X & Beam Size Ku & Beam Size LOFAR & GMRT Band 5 P.A. & VLA L P.A. & S P.A. & C P.A. & X P.A. & Ku P.A. & LOFAR P.A. \\
\begin{tabular}{cccccccccc}
\hline
Freq. (GHz) & Details$^{a}$ & J0159+0751 & J0811+4730 & J0820+5431 & J1032+4919 & J1355+4651 & J150934+373146 & J160810+352809 & J2229+2725 \\
\hline \hline \\
%$z^a$ & & $0.06105$ & $0.0444$ & $0.03851$ & $0.0442$ & $0.02811$ & $0.0325$ & $0.0327$ & $0.07622$ \\
%\hline \\
& $S_\nu$ ($\mu$Jy) & $< 64.5$ & $< 58.2$ & $< 69.3$ & $< 48.3$ & $< 46.2$ & $478\pm36.9^b$ & $58\pm17.0^b$ & $< 79.8$ \\
1.25 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $2.3\prime\prime \times 2.1\prime\prime $ & $2.4\prime\prime \times 1.9\prime\prime $ & $5.0\prime\prime \times 2.3\prime\prime $ & $2.4\prime\prime \times 2.2\prime\prime $ & $4.0\prime\prime \times 2.1\prime\prime $ & $3.1\prime\prime \times 2.6\prime\prime $ & $7.3\prime\prime \times 3.3\prime\prime $ & $3.9\prime\prime \times 2.0\prime\prime $ \\
& P.A. & $56.2^\circ$ & $21.1^\circ$ & $85.4^\circ$ & $7.4^\circ$ & $-35.6^\circ$ & $-69.7^\circ$ & $55.4^\circ$ & $44.0^\circ$ \\
\hline \\
& $S_\nu$ ($\mu$Jy) & $--$ & $--$ & $--$ & $< 41.1$ & $--$ & $--$ & $--$ & $--$ \\
1.5 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $--$ & $--$ & $--$ & $3.8\prime\prime \times 3.5\prime\prime $ & $--$ & $--$ & $--$ & $--$ \\
& P.A. & $--$ & $--$ & $--$ & $-40.8^\circ$ & $--$ & $--$ & $--$ & $--$ \\
\hline \\
& $S_\nu$ ($\mu$Jy) & $< 36.6$ & $< 27.3$ & $< 31.8$ & $68.7\pm9.6$ & $< 32.7$ & $330.0\pm39.6$ & $--^{c}$ & $< 29.7$ \\
3.0 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $3.3\prime\prime \times 1.8\prime\prime $ & $2.7\prime\prime \times 1.9\prime\prime $ & $2.0\prime\prime \times 1.7\prime\prime $ & $2.3\prime\prime \times 1.8\prime\prime $ & $1.9\prime\prime \times 1.7\prime\prime $ & $6.5\prime\prime \times 1.7\prime\prime $ & $2.4\prime\prime \times 1.8\prime\prime $ & $2.3\prime\prime \times 1.8\prime\prime $ \\
& P.A. & $55.4^\circ$ & $68.5^\circ$ & $38.6^\circ$ & $-83.8^\circ$ & $49.7^\circ$ & $59.9^\circ$ & $84.3^\circ$ & $71.1^\circ$ \\
\hline \\
& $S_\nu$ ($\mu$Jy)& $34.6\pm6.4$ & $< 16.8$ & $< 18.9$ & $61.9\pm5.7$ & $< 19.8$ & $170.0\pm15.14$ & $46.45\pm4.6$ & $< 18.6$ \\
6.0 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $1.6\prime\prime \times 1.0\prime\prime $ & $1.5\prime\prime \times 1.0\prime\prime $ & $1.2\prime\prime \times 0.9\prime\prime $ & $1.2\prime\prime \times 1.0\prime\prime $ & $1.1\prime\prime \times 0.9\prime\prime $ & $2.4\prime\prime \times 1.0\prime\prime $ & $1.1\prime\prime \times 1.0\prime\prime $ & $1.1\prime\prime \times 1.1\prime\prime $ \\
& P.A. & $50.0^\circ$ & $66.0^\circ$ & $42.4^\circ$ & $-62.0^\circ$ & $43.5^\circ$ & $63.6^\circ$ & $89.7^\circ$ & $57.6^\circ$ \\
\hline \\
& $S_\nu$ ($\mu$Jy)& $36.0\pm6.6$ & $< 18.4$ & $< 39.9$ & $53.9\pm6.8$ & $31.0\pm6.6$ & $210.0\pm12.16$ & $43.5\pm7.6$ & $< 16.7$ \\
10.0 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $4.4\prime\prime \times 1.9\prime\prime $ & $3.4\prime\prime \times 1.9\prime\prime $ & $2.8\prime\prime \times 1.9\prime\prime $ & $3.9\prime\prime \times 1.8\prime\prime $ & $5.5\prime\prime \times 1.8\prime\prime $ & $2.0\prime\prime \times 1.9\prime\prime $ & $2.1\prime\prime \times 1.8\prime\prime $ & $2.7\prime\prime \times 1.9\prime\prime $ \\
& P.A. & $-56.0^\circ$ & $79.4^\circ$ & $-83.4^\circ$ & $-74.6^\circ$ & $64.3^\circ$ & $41.7^\circ$ & $-19.3^\circ$ & $-75.8^\circ$ \\
\hline \\
& $S_\nu$ ($\mu$Jy)& $--$ & $< 17.2$ & $--$ & $69.0\pm5.0$ & $28.3\pm5.4$ & $180.0\pm22.42$ & $45.14\pm6.0$ & $< 13.9$ \\
15.0 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $--$ & $1.6\prime\prime \times 1.3\prime\prime $ & $--$ & $1.4\prime\prime \times 1.2\prime\prime $ & $2.0\prime\prime \times 1.2\prime\prime $ & $3.1\prime\prime \times 1.3\prime\prime $ & $2.5\prime\prime \times 1.3\prime\prime $ & $2.9\prime\prime \times 1.3\prime\prime $ \\
& P.A. & $--$ & $-70.7^\circ$ & $--$ & $7.4^\circ$ & $89.7^\circ$ & $73.1^\circ$ & $71.6^\circ$ & $67.0^\circ$ \\
\hline \\
& $S_\nu$ ($\mu$Jy)& $--$ & $< 532.4$ & $< 169.5$ & $< 200.2$ & $< 181.9$ & $400.68\pm126.6$ & $< 225.8$ & $< 316.0$ \\
0.150 & $\theta_{\mathrm{maj}} \times \theta_{\mathrm{min}}$ & $--$ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ & $6\prime\prime \times 6\prime\prime $ \\
& P.A. & $--$ & $90^\circ$ & $90^\circ$ & $90^\circ$ & $90^\circ$ & $90^\circ$ & $90^\circ$ & $90^\circ$ \\
\hline
\end{tabular}
\tablefoot{Sources with no observations in a specific band are shown with empty values. \\
%$^a$ Redshift values are taken from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor} and \citet{Jaskot19}.
$^a$ Shows the measured flux density ($S_\nu$) and beam parameters for each observing frequency. For non-detections we present the $3\sigma$ upper limits on the flux densities. \\
$^b$ Flux densities and beam parameters taken from \citetalias{Sebastian19}. \\
$^c$ The VLA S-band flux density for this source could not be measured due to strong artefacts from a neighbouring bright source.}
%\end{table*}
\end{sidewaystable*}
%\subsection{Archival LoFAR observations}
%\subsection{uGMRT and VLA Images}
\section{Results}
\label{sec: results}
We present the \rcsed{s} of our sample of \xsfg{s} spanning a wide range of radio frequencies from 1.2 GHz (from 150 MHz for \bbten{}) to up to 15 GHz using a combination of \vla{}, \gmrt{}, and \lofar{} telescopes described in Section \ref{sec: radio-sed}.
\subsection{Radio-SEDs of \xsfg{s}}
\label{sec: radio-sed}
\begin{figure*}[ht!]
\captionsetup[subfigure]{labelformat=empty}
\centering
\subfloat[]{{\includegraphics[width=0.44\hsize]{figures/sed_fits/J0159+0751_FFA_model2_Hunt_MLfit.pdf} }}%
\qquad
\subfloat[]{{\includegraphics[width=0.44\hsize]{figures/sed_fits/J1032+4919_FFA_model2_Hunt_MLfit.pdf} }}%
\qquad
\subfloat[]{{\includegraphics[width=0.44\hsize]{figures/sed_fits/J1355+4651_FFA_model2_Hunt_MLfit.pdf} }}%
\qquad
\subfloat[]{{\includegraphics[width=0.44\hsize]{figures/sed_fits/J160810+352809_fixed_nth__MLfit.pdf} }}%
\qquad
\subfloat[]{{\includegraphics[width=0.54\hsize]{figures/sed_fits/J150934+373146_FFA_model2_Hunt_MLfit.pdf} }}%
%\qquad
\caption{Observed \rcsed{s} of xSFGs for our detection sample. RC detections are shown in blue error bars (wth $1\sigma$ errors), and non-detections ($3\sigma$ upper limits) are shown in red arrows. We overlay the \rcsed{} with different models (FFA and simple) in solid grey line. We also show the thermal (blue dashed line) and non-thermal (red dotted line) component of the \rcsed{} model for each of our sources. Note that the \rcsed{} model parameters are ``indicative'' values and are not derived using a formal fit. J0159+0751 is also compatible with a purely thermal model (simple model with \fth{}$= 1$) shown in green line. For \bbten{} we show the FFA model (dashed line) which explains the low-frequency data points ($<10$ GHz). At high-frequencies (shown in brown) there appears to be break from a single FFA model. Note that here the thermal component of the \rcsed{} model is not visible in the plot as the source is possibly strongly non-thermally dominant. See Sec. \ref{sec: radio-sed} for a discussion on \rcsed{} models.}
\label{fig: rado-sed}
\end{figure*}
%We present the \rcsed{s} of \xsfg{s} in Figure \ref{fig: rado-sed}.
As seen from Figure \ref{fig: rado-sed}, most sources (except J1355+4651) can be described as being flat between 6-15 GHz, with several sources showing evidence for a strong turnover at lower frequencies. The turnover frequency (\nuturn{}) lies in the range of $2-5$ GHz. J1355+4651 has a flat spectrum between $10 - 15$ GHz and shows a turnover close to $10$ GHz. Besides that at lower frequencies ($< 6$ GHz) there is a diversity in the \rcsed{}. For J0159+0751, the 3$\sigma$ upper limit at 3.0 GHz from the \vla{} S-band data suggests either a flat spectrum or possibly a turnover. Two other sources, \bbten{} and \bbthirteen{}, are exceptions to this general feature in the \rcsed{}. \bbthirteen{} has a flat radio spectrum down to 1.2 GHz, with no significant evidence for a turnover in our current data. The \rcsed{} of \bbten{}, on the other hand, has the most complex features in our sample. The spectrum is relatively flat (within the error bars) between 6-15 GHz, and is fairly steep ($\alpha \approx -1.0$) below 6 GHz to up $\sim 1$ GHz. Between 1 GHz - 150 MHz range it shows evidence for a turnover (or flattening).
The presence of a strong turnover at lower frequencies in the radio spectrum of star-forming galaxies can be a signature of free-free absorption (FFA)\footnote{Synchrotron self-absorption (SSA) can also lead to low-frequency turnovers. However, the brightness temperature (T$_b$) of our sources is too small ($\leq 100$ K) for SSA to be significant at GHz frequency range, which will require an extremely high T$_b \approx 10^{12}$ K \citep{Condon_16_essential_RA}.}. The radio spectrum can show such a turnover at GHz frequencies due to a high emission measure (EM), owing to high ionised gas density \citepalias{Hunt04}. \sbs{}, a nearby low-metallicity starburst, also shows such a turnover but at a slightly lower frequency of $\sim 1.5$ GHz, although at higher frequencies it is relatively steeper than \xsfg{s} in our sample \citepalias{Hunt04}.
We particularly, consider two \rcsed{} models,
\begin{itemize}
\item Simple Model:\ Here the total radio emission ($S_{\nu}$) comprises of a mix of a single thermal ($S_{\nu}^{\mathrm{th}}$) and non-thermal component ($S_{\nu}^{\mathrm{nth}}$), described as:
\begin{equation}\label{eq: simple sed model}
S_{\nu} = S_{\nu}^{\mathrm{th}} + S_{\nu}^{\mathrm{nth}}
= A_1\left(\frac{\nu}{\nu_\circ}\right)^{-0.1} + A_2 \left(\frac{\nu}{\nu_\circ}\right)^{\alpha_\mathrm{nth}}.
\end{equation}
Here, $\nu$ is the observed frequency, $\nu_\circ$ is a reference frequency which is fixed to 1 GHz in all our models. And the \alphanth{}\footnote{In the following, we use this sign convention to define the spectral index. Note that some references use the opposite sign convention.} is the non-thermal spectral index, which we fix to a canonical value of $-0.8$ \citep{condon1992, Bait24a}. We then define the thermal fraction at a reference frequency of 1 GHz (\fth{}) as $\fth{} = S_{1 \mathrm{GHz}}^{\mathrm{th}}/S_{1 \mathrm{GHz}}$.
\item FFA Model:\ Here the \rcsed{} shows a strong turnover at lower frequencies due to the effect of FFA arising due to high EM in the ISM. We define the turnover frequency (\nuturn{}) as the frequency at which the free-free absorption optical depth (\odepth{}) becomes 1 \citep[see e.g.,][]{Hunt04, Clemens10, Galvin18}. The EM is thus directly related to \nuturn{}, such that increasing the EM leads to a higher value of \nuturn{}. Specifically, in this work we used the relation from \citetalias{Hunt04} (their Eq. 1) to derive the EM. We model the effect of FFA assuming that the absorbing medium is intermixed with both the thermal and the non-thermal components, following \citepalias[][Model 2]{Hunt04};
\begin{equation}\label{eq: FFA model}
S_{\nu} = \frac{1 - \exp({-\tau_\nu^{\mathrm{ff}}}) }{ \tau_\nu^{\mathrm{ff}} } ~(S_{\nu}^{\mathrm{th}} + S_{\nu}^{\mathrm{nth}} ).
\end{equation}
In this model the free parameters are: a) EM, b) $\theta$ (angular size of the emitting region) and c) $S_{\nu_\circ}^{\mathrm{nth}}$. We also derive the \fth{} for the FFA model following the above definition. In general, \citetalias{Hunt04} introduces two other FFA models depending on the geometry of the absorbing medium. These are the foreground screen absorption model (Model 1) and an absorption screen affecting only the non-thermal component (Model 3). Given our limited number of data points in which we have a detection, it is difficult to perform a formal fit across these models to rule them out and/or to perform a parameter fitting.
%\item Mixed Model:\ In order to describe the complex \rcsed{} behaviour of \bbten{}, we consider a third scenario, which consists of a combination of two FFA components, having two different EMs.
\end{itemize}
Next we describe the individual sources which are best described by one of the above \rcsed{} models. Note that we did not perform a formal fit to the parameters, since given our limited number of data points with detections, we cannot remove the several degeneracies in the parameters. Importantly, to constrain \nuturn{} in the FFA model, we need detections below the turnover, however, currently, we only have upper limits. We thus present the \rcsed{} models with indicative parameters which can overall describe the observed \rcsed{}, wherein there could be some degeneracy between these parameters.
%Only \bbten{} is an exception where we have a good number of data points in a large frequency range.
\begin{itemize}
\item \bbthirteen{}: This source has a flat radio spectrum from 1.2 - 15 GHz, and can thus be easily modelled with a ``Simple Model'' with a higher thermal fraction ($ \fth{} \sim 0.98$) as shown by the grey line in Figure. \ref{fig: rado-sed}. However, given the relatively larger error bar at 1.2 GHz, we cannot completely rule out FFA at low frequencies.
\item J0159+0751, J1032+4919, J1355+4651: These sources have flat spectra at high frequency and the upper limits at low-frequencies suggest a turnover in the spectra due to FFA. In Figure \ref{fig: rado-sed} we overplot the grey lines which show the ``FFA model'' with a high \fth{} $\sim 0.46 - 0.77$, a \nuturn{} in the range of $\sim 2 - 10$ GHz and a typical size of this emitting region in the $\sim 1\arcsec{}$ range. For J0159+0751 the $3\sigma$ upper limit at 3 GHz is in principle also consistent with a thermally dominant spectrum with no FFA (\fth{}$\approx 1$; shown with a green line) similar to \bbthirteen{}. For J1355+4651 since the \nuturn{} is close to 10 GHz, we only have two detections at 10 and 15 GHz. It is possible that there is a larger fraction of non-thermal emission at lower frequencies, which is absorbed due to FFA. Thus, the true \fth{} for this source can be lower than what we have presented. This source has the highest \nuturn{} in our sample and here even with detections at low frequencies, it will be very difficult to constrain \fth{}. The EM in these sources is very high, in the range of $\sim (4 - 80)\times 10^7$ pc cm$^{-6}$ (see Table \ref{table: physical properties}). \sbs{} also has a EM of $\sim (2 - 8) \times 10^7$ pc cm$^{-6}$ which is within this range. This value of EM is $\sim 1 - 3$ orders of magnitude higher than that found in typical nearby star-forming galaxies \citep[e.g.,][]{Chyzy18, Grundy24}.
\item \bbten{}: Lastly \bbten{} has the most complex \rcsed{} in our sample. Here the \rcsed{} below $10$ GHz is consistent with a steep spectrum and a turnover at a relatively lower value of $0.41$ GHz (EM $1.2 \times 10^6$ pc cm$^{-6}$). The \rcsed{} is possibly non-thermally dominant as opposed to the rest of the sample. Peculiarly, the \rcsed{} at high frequencies ($> 10$ GHz) shows a break from the steeply spectrum and rather shows a rising spectrum. Thus it is apparent that the \rcsed{} cannot be modelled by a simple or FFA model. This could require a combination of two FFA components, having two different \nuturn{}, a high value close to $10$ GHz and a lower value of $0.4$ GHz (see also Section \ref{sec: discussion} for a discussion).
\end{itemize}
In Table \ref{table: physical properties} we show the various physical properties with parameters (indicative values) derived from the above described \rcsed{} models.
%In summary, the \rcsed{} of \xsfg{s} are consistent with a high thermal fraction and a high EM compared to normal star-forming galaxies,
%\textcolor{red}{Add some discussion on EM in this section.}
\begin{table*}[ht!]
\centering
\caption{Physical properties of \xsfg{s} from our RC observations.}
\label{table: physical properties}
\begin{tabular}{cccccccc}
\hline \\
Target & \fth$^a$ & EM$^b$ & $N_e^c$ & \hbetaradio{}$^d$ & \specindex$^e$ & \fesc{} (indirect)$^f$ & \rcsed{} Model \\
\hline \hline \\
J0159+0751 & $0.464$ & $18.85$ & $ 660 \pm 300$ & $84.76 \pm 16.26 $ & $0.078^\mathrm{C}_\mathrm{X} \pm 0.021$ & $0.091 \pm 0.03$ & FFA/Simple (\fth{}$\approx 1$) \\
& & & & & & & \\
J1032+4919 & $0.775$ & $3.95$ & $560 \pm 50$ & $130.39 \pm 17.85$ & $-0.150^\mathrm{S}_\mathrm{C} \pm 0.027$ & $--$ & FFA \\
& & & & & & & \\
J1355+4651 & $0.691$ & $80.26$ & $1200 \pm 150$ & $73.33 \pm 16.15 $ & $-0.222^\mathrm{X}_\mathrm{Ku} \pm 0.065$ & $0.169 \pm 0.03$ & FFA \\
& & & & & & & \\
\bbten{} & $0.001$ & $0.12$ & & $501.44 \pm 38.36$ & $-0.958^\mathrm{S}_\mathrm{C} \pm 0.158 $ & $0.034 \pm 0.01$ & FFA$^g$ \\
\bbthirteen{} & $0.984$ & $--$ & & $103.87 \pm 18.88$ & $-0.128^\mathrm{C}_\mathrm{X} \pm 0.027$ & $0.31 \pm 0.13$ & Simple \\ \hline \\
\end{tabular}
\centering
\tablefoot{ The various parameters from the \rcsed{} models presented here are indicative values which can overall describe the observed \rcsed{} and are not derived using a formal fit with a full analysis of parameter degeneracy. \\
$^a$ The thermal fraction is defined at $1$ GHz for reference and is derived from the \rcsed{} models shown in Fig. \ref{fig: rado-sed}, and it not fitted with the data at $\sim 1$ GHz. \\
$^b$In units of $10^7\times~$pc cm$^{-6}$. Here we use the T$_\mathrm{e}$ values from \citet{Izotov20-diverseLyA} and \citet{Izotov24-LyA-metalpoor}. For \bbten{} and \bbthirteen{} we use a fixed value of $2\times 10^4$K.\\
$^c$In units of cm$^{-3}$. Values are taken from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor}. \\
$^d$Values shown are in units of $10^{-16}$ ergs/s/cm$^2$. Derived using the 10 GHz flux density. See Appendix \ref{appendix sec: radio-based Hbeta flux details} for details.\\
$^e$The \vla{} bands used to derive \specindex{} is shown around each value. \\
$^f$\fesc{} values taken from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor} which are indirectly estimated using \vsep{} \citep{Izotov18b-vsep}. For \bbten{} and \bbthirteen{} the \fesc{} are derived here using \vsep{} from \citet{Jaskot19}. \\
$^g$The \rcsed{} shows a deviation from a single FFA model at high-frequencies.
}
\end{table*}
%The presence of such a flat
%\textcolor{red}{Add brief discussion on active, passive starburst etc. YMC, SSC in FFA objects etc}
\subsection{Comparison between H$\beta$ dust attenuation derived using Balmer lines and RC emission}
\label{sec: hbeta flux comparison}
\begin{figure}[ht!]
\resizebox{\hsize}%{!}{\includegraphics{figures/radio_sed_main_text_sources.pdf}}
%{!}{\includegraphics{figures/Hbeta_comparison_10_GHz_E_B_V.pdf}}
{!}{\includegraphics{figures/Hbeta_comparison_10_GHz_E_B_V_singlepanel.pdf}}
\caption{Left-axis: Ratio of the observed \hbeta{} flux density (dust uncorrected) to that expected from the 10 GHz flux density vs. \ebv{}. Right-axis: Same as left axis except that the observed \hbeta{} flux density is corrected for dust. The dashed line connects the points with and without dust correction. Overall we find that in spite of correcting the \hbeta{} flux density for dust extinction using hydrogen Balmer decrement, in a few cases (e.g., J1032+4919 and J1355+4651) it is possible that a small fraction ($\sim20-30\%$) of the ionizing radiation emitted by the galaxy is not visible in the Balmer lines. See Sec. \ref{sec: hbeta flux comparison} for a detailed discussion and caveats. %J0159+0751 is remarkably more dusty than the rest of our sample although it has no signs of hidden dust.
%See Wang19 to check for the relation between AHalpha, EBV and dust correction.
}
\label{fig: hbeta flux comparison}
\end{figure}
Based on the \rcsed{s} presented above, we can see that the RC emission at high frequencies ($\geq 10$ GHz) for most of our sample is dominated by the free-free component. Importantly, RC emission at this frequency is not affected by dust extinction. Since the free-free emission in a galaxy at both the radio and optical wavelengths arise from the same emitting regions, we can use the RC flux density at 10 GHz to independently derive the expected dust-free \hbeta{} flux density (\hbetaradio). In Appendix \ref{appendix sec: radio-based Hbeta flux details}, we describe the details on measuring \hbetaradio{} using RC emission at 10 GHz. Table \ref{table: physical properties} shows the derived \hbetaradio{} for our sample. We can then use the ratio between the observed \hbeta{} (uncorrected for dust) and \hbetaradio{} to measure the total dust-extinction. We use the observed and dust-extinction corrected \hbeta{} flux densities from \citet{Izotov20-diverseLyA} and \citet{Izotov24-LyA-metalpoor} which are derived from the SDSS spectrum. For \bbten{} and \bbthirteen{}, we measured the \hbeta{} line flux density from their SDSS spectrum. We estimated the dust extinction at \hbeta{} ($\mathrm{A}_{\mathrm{H}\beta}$) following the procedure described in \citet{Izotov16a}\footnote{Here the $\mathrm{A}_{\mathrm{H}\beta}$ is derived using $C$(\hbeta{}) which depends on the $\mathrm{A}_\mathrm{V}$ and $\mathrm{R}_{\mathrm{V}}$. Effectively this assumes the \citet{Cardelli89} extinction law which is found to be a better fit for such extreme galaxies \citep{Izotov17}. }, where we use the V-band extinction ($\mathrm{A}_\mathrm{V}$) from \citet{Jaskot19} and assume a $\mathrm{R}_{\mathrm{V} }$ of 2.7 \citep{Izotov17}.
In Figure \ref{fig: hbeta flux comparison} (left-axis), we compare this ratio against \ebv{}\footnote{Here \ebv{} is derived using the $C$(\hbeta{}) values from \citet{Izotov20-diverseLyA} and \citet{Izotov24-LyA-metalpoor}, and using $C$(\hbeta{}) $= 1.47\times$\ebv{}.}. In the right-axis of Figure \ref{fig: hbeta flux comparison}, we compare the ratio between the dust-corrected \hbeta{} flux density, derived using hydrogen Balmer decrement, and \hbetaradio{} with \ebv{}. Using such a ratio we can compare if the dust correction applied to the observed \hbeta{} line is completely accounted for by the Balmer decrement. For the cases when this ratio is below 1, we can interpret this as evidence of a fraction of ionizing radiation emitted by the galaxy being missed in the Balmer lines. Note that for a few sources the uncertainties of \ebv{} are too small and are not apparent on the figure.
From Figure \ref{fig: hbeta flux comparison} (left-axis), we find that many sources in our sample show evidence of dust attenuation at optical wavelengths. It is intriguing to find the presence of dust in such low-mass and low-metallicity systems. We notice that despite the dust-correction from the Balmer decrement to the \hbeta{} line, it is possible that some ionizing radiation emitted by the galaxy is missing from the Balmer lines (e.g., in J1032+4914 and J1355+4651). In the case of J1032+4914 and J1355+4651, which have \ebv{} values close to zero and thus also have a negligible dust correction to the \hbeta{} line, yet they show evidence for such a missing emission using the radio-derived \hbetaradio{}. The amount of ionizing emission missed in these objects is generally small, between $\sim 20 - 30\%$. However in J1355+4651, as described in Section \ref{sec: radio-sed}, it is possible that there is a higher amount of non-thermal emission which is difficult to account for due to non-detections below $10$ GHz. This can reduce the \hbetaradio{} thus making it compatible with the dust-corrected \hbeta{} line flux. In some cases, for example, in \bbten{}, J0820 + 5431 and J0811 + 4730, there is apparently a large fraction of ionizing emission missing from the Balmer lines. As shown in Section \ref{sec: radio-sed}, \bbten{} has a large fraction of non-thermal emission and thus the true \hbetaradio{} is lower than our estimated value. Moreover, \bbten{} has an extended structure, thus we could be missing some \hbeta{} flux from the SDSS spectrum. Both these effects could reduce the fraction of missing ionizing emission in the Balmer lines than our estimate. In J0820+5431 and J0811+4730 due to non-detections in the RC emission, we can only present a lower limit on the total dust extinction in the \hbeta{} line.
J0159+0751 has the highest \ebv{} value in our sample and the ratio between the observed \hbeta{} and \hbetaradio{} flux is also relatively low. Contrary to other sources, both J0159+0751 and \bbthirteen{} have \hbetaradio{} consistent with that derived using the dust-corrected \hbeta{} line within the uncertainties (see Figure \ref{fig: hbeta flux comparison}).
Overall, we observe a large variation in the dust properties of \xsfg{s} with evidence for a small fraction of the ionizing radiation emitted by the galaxy missed in the Balmer lines.
% show the rc sed at different freqs
%Discuss the individual cases
% Evidence for dust very interesting
% UCHII region one cause
% Implications: source of dust in low-metallicity environments??
% This can also contribute the origin of the SFR offsets??
% If also valid for high-z galaxies: JWST NIRSpec observations will miss these dusty objects
% Overall dust geomtery can explain the lack of dust (and excess dust) seen in xSFGs.
% Also try the SFR RC alpha relation from Bait et al. 2024, are the SFRs in the same range?
\section{Discussion}
\label{sec: discussion}
%\subsection{On the lack of non-thermal emission in xSFGs}
\subsection{RC suppression in xSFGs at low frequencies}
\label{sec: non-thermal suppression discussion}
\begin{figure}[ht!]
\resizebox{\hsize}%{!}{\includegraphics{figures/radio_sed_main_text_sources.pdf}}
%{!}{\includegraphics{figures/Hbeta_comparison_10_GHz_E_B_V.pdf}}
{!}{\includegraphics{figures/RC-SFR_3_GHz.pdf}}
\caption{\xsfg{s} on the RC-SFR relation at $3$ GHz. We show the standard total RC-SFR relation \citep{murphy2011} using a $\alphanth{} = -0.8$ with a solid black line. We also show the RC$_\mathrm{th}$-SFR relation with a red solid line. In both these cases we have used a fixed value of T$_e = 20000$ K. Our \xsfg{s} show a significant suppression of RC emission than that expected from the total RC-SFR relation. See Section \ref{sec: non-thermal suppression discussion} for a discussion.
}
\label{fig: RC suppression}
\end{figure}
In Figure \ref{fig: RC suppression}, we plot the observed RC luminosity at $3$ GHz against the SFRs for our sample of \xsfg{s}. Our SFRs are taken from the literature (see Table \ref{table: basic properties}) and are derived using the dust-corrected \hbeta{} and H$\alpha$ line luminosity following the \citep{Kennicutt98, Kennicutt12} relation. For reference, we overplot the RC (total)-SFR relation at $3$ GHz for star-forming galaxies from \citet[][Eq. 15]{murphy2011} assuming a \alphanth{} of -0.8, shown in solid black line. The thermal RC (\RCth{}) - SFR relation from \citet[][Eq. 11]{murphy2011} is shown in solid red line. Note that for \bbthirteen{}, since we were unable to measure the flux density at $3$ GHz (see Table \ref{table: radio fluxes}), we have used the modelled RC flux density at $3$ GHz from Section \ref{sec: radio-sed} and a noise level of $10~ \mu$Jy typically achieved at S-band for our observations. We note that all our \xsfg{s} show suppressed RC emission than that expected from the RC-SFR relation at $3$ GHz, with a few sources even below the \RCth{} - SFR relation. We notice a similar trend in our data at $1.25$ GHz using the \gmrt{} data.
Previous single band radio studies of \xsfg{s} at low-frequencies ($150$ MHz and $\sim1.5$ GHz) had also found that the RC emission in such galaxies is suppressed \citep{chakraborti2012, Sebastian19, Borkar24-BB-LOFAR}, which was indirectly associated with a lack of non-thermal emission. Our finding of a thermally dominant \rcsed{} of xSFGs provides conclusive evidence for the prevalence of such a scenario. Morevoer the effect of FFA at GHz frequencies could further suppress the RC emission. For example, the RC emission in J1032+4919 was found by \citetalias{Sebastian19} to be highly suppressed at $1.2$ GHz compared to other sources in their sample and was the only undetected source in their study. Our \rcsed{} observation suggests that this non-detection is due to FFA at $\sim 2.2$ GHz. In some systems RC emission can get suppressed due to FFA even in the presence of non-thermal emission (e.g., in \bbten{}).
In general, there are several scenarios which can lead to a lack of non-thermal emission: 1) extremely young ages ($< 5$ Myrs) which lack enough SNe \citep[e.g,][]{Hayes07, carilli2008, chakraborti2012, greis2017, Sebastian19}, 2) CR escape \citep{lisenfeld2004,Sebastian19}, 3) strong synchrotron losses due to enhanced magnetic fields \citep{chakraborti2012} and 4) Inverse Compton (IC) losses due to high photon density \citep{condon1992}. The young ages of these galaxies are probably the primary reason for the flat spectrum as based on strong nebular emission lines \citep[e.g., high \ewhbeta{} $>250~\AA$,][]{Izotov20-diverseLyA, Izotov24-LyA-metalpoor, Jaskot19, yang2017}. The rest of the scenarios involving the lack of CRs (IC/syncrotron losses, CR escape etc.) would lead to a significant break at high frequencies \citep{klein2018}. We do not observe such a break in the \rcsed{} of \xsfg{s} in our sample. Indeed, \bbten{} is an exception having the highest non-thermal fraction, which possibly has a more complex formation scenario due to a merger with another dwarf galaxy as argued by \citet{Dutta24}.
%The rest of the scenarios leading to a lack of non-thermal emissions are particularly important in galaxies with small sizes having dense ISM. For example, in sub-kpc size galaxies, CRs can easily escape in an open box model.
%5) cutoff in the CR spectrum \citep{Israel88, klein2018, Bait24a}.
% 1 most likely young ages seen in the optical EW etc. 2, 3, and 4 more likely in dense galaxies. bb13 is flat, and thus could be a genuinly young galaxy. In general klein like modelling helps to distinguish with low-freq data. FFA in Izotov sources makes estimating non-thermal specindex difficult. Thus hard to test several possibilities.
% high-z radio sources could also be flat. IC-CMB losses can further flatten the radio spectra carilli and Murphy papers to cite.
% Then discuss the YMC issue
% high EM --> YMC like environemnts --> dense ISM high O32 ration LyC leakage
% YMCs --> can explain high brightness in ultra-high z galaxies solving the JWST problems
% YMC --> can explain the high SFEs in galaxies. also cite dutta, kanekar HI papers for SFE. And also cite Hunt paper on SBS for SFE. Then discuss high SFE in high-z JWST galaxies
% Discuss if the EM is consistent with Ne and sizes of galaxies, Ne of 1032 is similar to SBS E etc.
In summary, \xsfg{s} lack a significant amount of non-thermal emission as they are young systems lacking a significant population of CRs which are accelerated in SNe explosions. This would suggest that feedback from SNe and CRs is not dominant in these galaxies. Thus radiation feedback might be important in such metal-poor starbursts \citep[e.g.,][]{Jeckmen23-delayedfeedback}.
\subsection{Dependence of LyC escape fraction on radio spectral index}
\label{sec: fesc-alpha}
% Plotting the sample (+LzLCS) on O32-R23 plot color-coded by alpha
\begin{figure}[ht!]
\resizebox{\hsize}%{!}{\includegraphics{figures/radio_sed_main_text_sources.pdf}}
{!}{\includegraphics{figures/total_spectral_index_fesc_lzlcs.pdf}}
\caption{Dependence of \fesc{} on the radio spectral index, \specindex{}. xSFGs from this paper are shown in blue hexagons. The sub-sample of sources with \vla{} RC observations from \citet{Bait24a} are shown in grey circles, along with Haro 11 (red star) and Tol 1247 - 232 (yellow star). Here the \fesc{} is measured directly using the HST/COS observations and the \specindex{} is measured from observations in the C- and S-bands. For our sample of xSFGs the \fesc{} is estimated indirectly using the empirical relation between \vsep{} and \fesc{} from \citet{Izotov18b-vsep}. The \specindex{} is measured at GHz frequencies which show a detection (see text for details). The two horizontal dashed lines demarcate the regions termed as strong-, weak- and non-leakers. The vertical dashed line represents \alphacs{} $= -0.64$, corresponding to Simple Model, with \fth{} $= 0.1$ and \alphanth{} $= -0.8$. }
\label{fig: fesc alpha}
\end{figure}
\citet{Bait24a} using RC observations of a sub-sample of LzLCS sources from \citet{Flury22a}, showed that the radio spectral index measured between the \vla{} C- and S-bands (\alphacs{}) correlates with the observed \fesc{}. We refer to this as the \fesc{}-\specindex{} dependence hereon. Here we measure the radio spectral index (\specindex{}) for our detection sample of xSFGs and compare it with the indirectly estimated \fesc{} in Figure \ref{fig: fesc alpha}. The \specindex{} for different sources is measured at several GHz frequencies which show detection. In particular, we use X- and C-bands for J0159+0751 and \bbthirteen{}, Ku- and X-bands for J1355+4651, and C- and S-bands for J150934+373146. The \fesc{} is indirectly estimated using the empirical relation between the \Lya{} velocity peak separation (\vsep{}) and \fesc{} from \citet{Izotov18b-vsep} and are taken directly from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor}. For \bbten{} and \bbthirteen{}, we use the \vsep{} values from \citet{Jaskot19} to estimate the \fesc{}. The \fesc{} values for LzLCS sources are taken from \citet{Flury22a} and \citet{Saldana-Lopez22} which are measured using the ``UV-fit method'' from HST/COS observations. And the \specindex{} is measured between the \vla{} C- and S-bands taken from \citet{Bait24a}. We also overplot in Figure \ref{fig: fesc alpha} two nearby weak leakers, Haro 11 and Tol 1247 - 232 \citep[see ][for references]{Bait24a}. The two horizontal dashed lines split the sources in three broad categories: strong leakers (\fesc{} > 0.1), weak leakers ( 0.01 < \fesc{} < 0.1), and non-leakers (\fesc{} < 0.01). The vertical dashed line represents the \alphacs{} for a simple radio-SED (Simple Model) with \fth{} $= 0.1$ and \alphanth{} $= -0.8$. We generally refer to sources with \specindex{} $> -0.64$ as flat spectrum and \specindex{} $< -0.64$ as steep spectrum. Note that for J1355+4651, the data point exactly falls on another strong leaker source from the LzLCS sample.
Most xSFGs in our sample, except for \bbten{}, are likely strong LyC leakers and possess a flat radio spectrum, as seen in Figure \ref{fig: fesc alpha}. \bbten{} is the only source which has a steeper radio spectrum and is also a likely weak leaker. Taken together we observe that strong leakers are generally described by a flat radio spectrum at GHz frequencies. On the contrary, not all flat spectrum sources show strong LyC leakage, as we observe a few outliers from the LzLCS sample. Moreover, sources with steep spectrum are either weak- or non-leakers. Quite remarkably that our xSFGs with higher \Oratio{} values are less metal enriched (see Figure \ref{fig: alpha physical props} for a comparison between \Oratio{} and metallicity) and also span a lower stellar mass range (\mstar{} $\leq 10^{8}$) than LzLCS sources, follow the same \fesc{}-\specindex{} dependence. We caution that the \fesc{} for xSFGs is indirectly measured and can be uncertain. Thus, there is tentative evidence that the \fesc{}-\specindex{} dependence holds independently of the stellar mass, metallicity and \Oratio{} ratio for compact star-forming galaxies. Future RC observations particularly focused on strong leakers with direct \fesc{} measurements are required to confirm the universality of the \fesc{}-\specindex{} dependence.
%\textcolor{red}{Should we discuss the issue of dust and fesc, J0159 has higher dust than J1509, yet has higher fesc?}
%\citet{Bait24a} argued that the \fesc{}-\specindex{} correlation could be primarily driven by a combination of the age of the starburst and the density variation of the ISM. Since the radio spectral index is a proxy for the age of the starburst \citep{Cannon04, Hirashita06}, the youngest starburst (ages $< 5$ Myrs) would lack significant non-thermal emission (due to pre-SNe stellar populations) and thus they will have a flat radio spectrum. Moreover, a dense ISM associated with a high EM would lead to an FFA feature at GHz frequencies, even in the presence of small amounts of non-thermal emission, which can flatten the radio spectra \citep[see][ for a discussion on this]{ Hirashita06, Bait24a}. We see direct evidence for a thermally dominated \rcsed{} with FFA at GHz frequencies (see Section \ref{sec: radio-sed}) for our xSFGs (except in \bbten{}) thus supporting this claim. Both these physical conditions of a young starburst and a very dense ISM are conducive to significant leakage of LyC photons (some \textcolor{red}{citations?}).
\citet{Bait24a} argued that the \fesc{}-\specindex{} correlation could be primarily driven by a combination of the age of the starburst and the density variation of the ISM. Since the radio spectral index is a proxy for the age of the starburst \citep{Cannon04, Hirashita06}, the youngest starbursts (ages $< 5$ Myrs) would lack significant non-thermal emission (due to pre-SNe stellar populations) and thus they will have a flat radio spectrum. Additionally, if the source is compact, a large cluster of young stars would be packed in a small region thus boosting the ionized gas density ($N_e \gtrsim 10^3$~cm$^{-3}$). Since the EM is strongly dependent on $N_e$ \footnote{EM $\propto N_e^2$L, where L is the size of the emitting region, usually in the range of 10-100 pc \citep{Hirashita06}.}, this leads to a high EM \citep[in the range of $\sim 10^{7-8}$ pc cm$^{-6}$, see][for a detailed model]{Hirashita06}. Such high EMs are typically a sign of a dominance of young massive star clusters \citep[YMCs; see][for a simulation study]{Inoguchi24_FFA-sims} in the ISM. Thus the radio free-free emission under such conditions would be optically thick at GHz frequencies \citep[e.g.,][]{Johnson03, Hunt04}, further leading to a flat/rising radio spectrum associated with a FFA component \citep[see also ][ for a discussion on this]{ Hirashita06, Bait24a}.
Based on our study here, we observe a direct evidence for such a thermally dominated \rcsed{} which is optically thick at GHz frequencies (see Section \ref{sec: radio-sed}) for our xSFGs (except in \bbten{}). This suggests that strong leakers possess a dominant population of YMCs in their \hii{} regions. YMCs can boost the star-formation efficiency \citep[SFE;][]{Fall10_SFE, Menon24a_bursty_radiation_outlfows} thus leading to a paucity of neutral/ molecular gas thus opening channels for LyC photons to escape \citep{Marques-Chaves24_efficientSFmode}. Such conditions can thus lead to both efficient production and escape of ionizing photons. Overall, this can explain the flat/rising radio spectrum of strong LyC leakers.
%\citet{Hirashita06} propose two modes of star-formation for metal-poor starbursts, defined as active based on the presence of FFA feature at GHz frequencies, and passive based on the absence of it. It is worth noting that strong leakers are characterized by both active (J1032+4919, J1355+4651) and passive (\bbthirteen{}) starbursts. This suggests that multiple physical mechanisms are responsible for strong LyC leakage from galaxies.
%\textcolor{red}{OB: Yuri's comment on J160810 passive starburst to be addressed.}
For relatively older starbursts ($5 - 10$ Myrs), where the radio free-free emission is optically thin and SNe explosions have accelerated the CR electrons producing a significant amount of non-thermal emission, the radio spectra are steeper. Moreover, in a relatively older starburst pressure-driven expansion of the \hii{} region would lead to a lower EM \citep{Hirashita06}, reducing the effect of flattening of the radio spectra due to FFA. Such conditions in the ISM would in general lead to a weaker LyC leakage, as observed in \bbten{} which has a steep radio spectrum, with a \nuturn{} $< 1$~GHz and low \fesc{} ($< 10\%$). Finally, \citet{Farcy22} using radiation-magnetohydrodynamics simulations of disc galaxies find that the CR feedback overall leads to a smooth ISM filling it with denser gas. This overall leads to a lower \fesc{} from galaxies. Thus the mere presence (lack) of CRs in weak/non-leakers (strong leakers) could also drive the \fesc{}-\specindex{} dependence and explain its possible existence over a wide stellar mass range ($\log~$\mstar{} $= {6-11}$). More work is needed on both the observation and modelling front to understand the exact physical mechanism(s) responsible for the \fesc{}-\specindex{} dependence.
%discuss role of CRs in reducing fesc)
%which is an evidence for the presence of CRs,
\subsection{Relations between \alphacs{}, ionization state and metallicity}
\label{sec: alpha ionization}
% mention the case of BB10, high O32 but low LyC escape -- Nakajima et a. 2020 also find this, and argue that viewing angle might be a reason. They also found no other physical parameter distinguishes a leaker from a non-leakers. Contrast that CR could be the main driver.
\begin{figure*}[ht!]
\resizebox{\hsize}%{!}{\includegraphics{figures/radio_sed_main_text_sources.pdf}}
{!}{\includegraphics{figures/xSFGs_alpha_phy_props.pdf}}
\caption{Relations between \Oratio{} ratio and metallicity ($12+\log\mathrm{O/H}$) with radio spectral index for \xsfg{s} (grey triangles). We show the LzLCS sources with RC measurements from \citet{Bait24a} (grey circles). Around each source we draw circles in blue (\fesc{} $> 10\%$), yellow (\fesc{} $= 1 - 10\%$) and red (\fesc{} $< 1\%$). See Figure \ref{fig: fesc alpha} and text for details on how the \fesc{} values are derived.}
\label{fig: alpha physical props}
\end{figure*}
High values of \Oratio{} ratio in galaxies could be associated with low-metallicity, high ionization parameter or the presence of density bounded \hii{} regions \citep{Jaskot-Oey13, Nakajima14-Ionization-Stat}. A high \Oratio{} ratio is found to be a necessary but not a sufficient condition for high \fesc{} in galaxies \citep{Izotov18b-vsep, Jaskot19, Nakajima20, Flury22b}. Thus we investigate if the \fesc{}-\specindex{} dependence could be driven by the dependence of \specindex{} on \Oratio{} or metallicity.
The dependences between \Oratio{}-\specindex{} and metallicity-\specindex{} are presented in the left- and right-panel respectively of Figure \ref{fig: alpha physical props} for xSFGs (in triangles). Such relations were studied by \citet{Bait24a} for a sub-sample of LzLCS sources which are shown in filled circles. This sub-sample lacked sources with very high \Oratio{}, which is covered by our xSFGs sample. We draw coloured circles around these points differentiating them in three categories: strong- (blue circles), weak- (orange circles) and non-leakers (red circles). Note that although J0159+0751 is classified as a weak-leaker, its \fesc{} indirectly derived from the \vsep{} is $\sim 9.1\%$, thus very close to a strong leaker. Overall, we extend the positive correlation between \Oratio{} and \specindex{} found by \citet{Bait24a} for the LzLCS sources, to the highest \Oratio{} ratios ($> 10$). And indeed as previously found the \fesc{} generally increases with increasing \Oratio{} ratio \citep[e.g.,][]{Izotov18b-vsep}. We also observe several outliers to this trend, wherein the flat spectrum sources have a range of \Oratio{} (and \fesc{} values). The dependence of metallicity on \specindex{} is not so tight. We do observe that the high metallicity sources from LzLCS have steep spectra (and low \fesc{}), although the flat radio spectrum sources are a mixed population. Thus as previously discussed, not all flat spectrum radio sources have high \fesc{} (or high \Oratio{}/low-metallicity). Similar to the \Oratio{} ratio, it appears that a flat radio spectrum is a necessary but not a sufficient condition for strong LyC leakage from galaxies.
Based on the finding that not all high \Oratio{} ratio sources have high \fesc{}, several studies have suggested that there could be additional physical properties which govern LyC leakage from galaxies, with the viewing angle being a likely possibility \citep[e.g.,][]{Izotov18b-vsep, Jaskot19, Nakajima20, Katz20_orientation, Bassett19_lack_of_O32_fesc}. In our xSFGs, we have such an example of \bbten{}, which has a high \Oratio{} value ($> 10$), and low metallicity comparable for instance to \bbthirteen{}. Yet, their \fesc{} values differ by an order of magnitude. However, we notice that their \specindex{} values vary considerably and can be used to predict their indirect \fesc{} values better (see Figure \ref{fig: fesc alpha}). On the other side, we also have sources from the LzLCS sample with low \Oratio{} that are strong leakers. Here again, such strong leakers have flat \specindex{}. Thus it appears that strong LyC leaking galaxies possess a flatter radio spectrum. Certainly, more data is needed to confirm if \specindex{} could be a primary parameter governing LyC escape in the strongest leakers. If true, this along with the overall correlation between \specindex{} and \fesc{}, and the fact that the \specindex{} is a spatially averaged property could suggest that at least to a first order the orientation might not be a primary driver of \fesc{} \citep[see ][for a more detailed discussion]{Bait24a}. More importantly, the presence of a flat radio spectrum in strong leakers would suggest a lack of SNe in these systems, and consequently a lack of SNe feedback thus supporting theoretical models on metal-poor starbursts \citep{Jeckmen23-delayedfeedback}. Thus other modes of feedback (e.g., radiation driven feedback) might be more important in leading to strong LyC leakage \citep[see also ][]{Carr24, Flury24_feedback}.
%\textcolor{red}{OB: also mention flat radio spectra would mean the lack of SN, radiation dominated feedback}
In summary, estimating \fesc{} from galaxies is a multi-variate problem \citep[e.g.,][]{Flury22b, Jaskot24-multivariate}, and along with the \Oratio{} ratio, metallicity, dust content and other physical properties, the radio spectral index can play an important role.
%\subsection{Dust build up in metal-poor starbursts}
%\label{sec: dust discussion}
%Generally in more evolved ($12+\log \mathrm{O/H} > 8.0$) dwarf and spiral galaxies, the growth of dust is closely related to the stellar mass (e.g., dust to stellar mass ratio), gas mass (e.g, the dust to gas mass ratio) and metallicity \citep[e.g.,][]{Remy-Ruyer14}. However, in two metal-poor ($12+\log \mathrm{O/H} < 8.0$), low-mass starburst galaxies in the local Universe, \sbs{} and \izw{}, which have properties similar to our sample of \xsfg{s}, the dust content was found to differ by almost two orders of magnitude \citep{Hunt14-SBS-ALMA}. We consider the three most massive galaxies (\mstar{} $> 10^{8}$) in our sample: J0159+0751, \bbten{} and \bbthirteen{}. All three galaxies have similar metallicities in a narrow range of $7.5 < 12+\log\mathrm{O/H} < 7.9$, yet they have very different dust content. J0159+0751 has the highest amount of dust whereas \bbten{} and \bbthirteen{} have negligible amounts of dust. \citet{Schneider16-dust-formation-model-metalpoor} suggested that the dust build-up in such metal-poor galaxies can become very efficient with increasing ISM density. We know that J0159+0751 has a denser ISM compared to \bbten{} and \bbthirteen{} based on the observed EM values and also the higher \nion{} in J0159+0751 \citep{Izotov20-diverseLyA}. This could explain the higher dust content in J0159+0751. On the contrary, J1032+4919 and J1355+4651 also have high \nion{} and EM but they have a relatively low dust content compared to J0159+0751, possibly due to their $1-2$ orders of magnitude lower stellar mass. Moreover we also observe that a small fraction of the dust is hidden from the optical lines. As inferred from the radio-SEDs, if \xsfg{s} are dominated by YMCs, it is possible that the dust is in deeply embedded SF regions which can remain hidden in the optical.
%Overall, using our sample of \xsfg{s} and the two well studied objects in the literature, we find that the dust content of metal-poor starbursts shows a large variety, with a few \xsfg{s} harbouring large amounts of dust. Exactly which parameters lead to such variation in dust content still remains open to investigation.
%\textcolor{red}{ Can clumpy dust explain the issue of hidden dust and dust variations?}
%\textcolor{red}{To be discussed: Issue of hidden dust, where does it arise from? 1. dust in UCHII regions missed in the optical, 2. clumpy dust sitting away from the line of sight? dust covering fraction? Use the Hunt ALMA SBS image to explain the clumpiness point}
% discuss the dust line of sight issue.
%\subsection{Predictions for the SKA and ngVLA}
%\label{sec: ska predictions}
\subsection{Overall implications from the \rcsed{} of xSFGs on reionization and high-$z$ galaxies found with JWST}
\label{sec: rcsed implications}
% thermally dominant --> lack of SN --> implications on feedback Jecman &Oey (but mention the caveats)
The \rcsed{} of several xSFGs is found to be thermally dominated along with a turnover at GHz frequencies owing to FFA. These \rcsed{s} are akin to those found in ultracompact and hyper-compact \hii{} regions in our Galaxy \citep{Wood_UCHII_radio, Yang21_HCHII_radio-sed, Patel24}. The presence of FFA at GHz frequencies is predicted to be a strong observational signature for the presence of YMCs from simulations \citep{Inoguchi24_FFA-sims}.
Thus our study provides a strong evidence for a picture wherein the extreme star-forming complexes in \xsfg{s} are dominated by YMC(s) with pre-SNe stellar population (inferred from their thermal radio spectra). This leads to a high star-formation rate surface density together with a lack of SNe feedback which boosts the star-formation efficiency (SFE) \citep{Fall10_SFE, Menon24a_bursty_radiation_outlfows} and can explain their extreme star-formation properties \citep{Jeckmen23-delayedfeedback, Marques-Chaves24_efficientSFmode}. Such a high SFE can lead to a lack of gas in such galaxies. This can explain the lack of \hi{} detections in low-$z$ green-pea/blueberry galaxies with high \Oratio{} ratio \citep{Kanekar21_HI-GP, Chandola24b} and the spatial offset between \hi{} and peak of the SF, e.g., in \bbten{} \citep{Dutta24}. Together these effects lead to conditions which are suitable for a significant escape of LyC photons \citep{Marques-Chaves24_efficientSFmode, Menon24a_bursty_radiation_outlfows}.
Since xSFGs are excellent analogues of high-$z$ galaxies \citep[e.g.,][]{Schaerer16, Schaerer22, Rhoads23}, this suggests that the very high-$z$ ($z \geq 6$) galaxies found by JWST might also possess a dominant population of YMCs in their \hii{} regions. Indeed several studies find a dominance of such very dense starclusters in high-$z$ JWST galaxies \citep{Vanzella23, Adamo24, Messa24a, Fujimoto24}. Consequently, these galaxies are expected to be strong LyC leakers and major contributors to reionization.
Finally, our \rcsed{} study suggests that future deep and ultra-deep radio surveys targeting high-$z$ JWST galaxies with the SKA and ngVLA need to be sensitive to the free-free emission from these galaxies as they are expected to lack a dominant fraction of non-thermal emission.
%\textcolor{red}{OB: I can also do a quick estimation of the required sensitivity if we know a typical (or a range of) SFR of JWST highz galaxies.}
%We also make some predictions for SKA-MID and ngVLA ultra-deep surveys for detecting such sources at $z > 6$. \textcolor{red}{to be discussed...}.
\section{Conclusions}
\label{sec: conclusions}
We present new radio continuum observations of low-$z$ \xsfg{s} using the \gmrt{}, \vla{} and archival \lofar{} telescope to study their \rcsed{}. \xsfg{s} are a population of rare metal-poor ($ 12+\log\mathrm{O/H} < 8.0$), compact starburst galaxies at low-$z$. Using various indicators they are found to be excellent analogues of high-$z$ ($z > 6$) galaxies found with JWST. We mapped the \rcsed{} in a wide frequency range from $\sim$1 to 15 GHz (with a lower limit of 150 MHz for a few sources), to gain novel insights on the nature of extreme star-formation properties and the escape of LyC from these galaxies.
Our main results can be summarized as follows:
\begin{enumerate}
\item The \rcsed{s} of \xsfg{s} are generally flat between 6-15 GHz and thus are dominated by free-free emission, and showing a strong evidence for FFA with a turnover in the range 2-10 GHz (Figure \ref{fig: rado-sed}). The estimated EM is very high in the range of $\sim 10^{7 - 9}$ pc cm$^{-6}$. This suggests that the starformation is dominated by YMCs and pre-SNe stellar population. This explains the lack of non-thermal emission found in previous studies (Sec. \ref{sec: non-thermal suppression discussion}) and their extreme properties (Sec. \ref{sec: rcsed implications}).
\item \bbten{} has a complex \rcsed{}. It is steep at $<10$ GHz with a turnover at $\sim 400$ MHz. At high frequencies ($>10$ GHz) it shows a break and is flatter compared to that below $10$ GHz (Figure \ref{fig: rado-sed} bottom panel).
\item We compared the free-free emission from 10 GHz radio flux with that expected from the observed optical emission line (\hbeta{}) and find evidence for dust in several \xsfg{s}. In a few \xsfg{s} correcting for this dust using Balmer decrement does not account all the dust, thus suggesting a small amount of ionizing radiation emitted by the galaxy but not visible in the Balmer lines (Figure. \ref{fig: hbeta flux comparison}).
%The presence of dust in such metal-poor young systems highlights a different (probably rapid) mode of dust formation in these galaxies.
\item Using indirect estimates of \fesc{} in \xsfg{s}, we confirmed the relation between \alphacs{} and \fesc{} from \citet{Bait24a}, particularly for strong leakers (Figure. \ref{fig: fesc alpha}). This suggests that most strong leakers lack SNe feedback and their SF is dominated by YMCs. Other modes of feedback, for example radiation driven mode might be more important for LyC leakage.
\item We also extended the previously found empirical relation between \alphacs{} and \Oratio{} ratio to high values of \Oratio{} (Figure \ref{fig: alpha physical props}).
\item Since \xsfg{s} are excellent analogues of high-$z$ galaxies, our study highlights the role of YMCs in explaining several extreme star-formation properties observed in recent JWST observations and can also lead to high LyC escape.
\end{enumerate}
\begin{acknowledgements}
O. Bait (OB) is supported by the {\em AstroSignals} Sinergia Project funded by the Swiss National Science Foundation. OB would like to thank Heidi Medlin for help with the VLA scheduling. We thank the staff of the GMRT who have made these observations possible. GMRT is run by the National Center for Radio Astrophysics (NCRA) of the Tata Institute of Fundamental Research. OB would also like to thank NCRA for hosting him where part of this work was completed. Y.I. acknowledges support from the National Academy of Sciences of Ukraine (Project No. 0123U102248) and from the Simons Foundation.
LOFAR is the Low Frequency Array designed and constructed by ASTRON. It has observing, data processing, and data storage facilities in several countries, which are owned by various parties (each with their own funding sources), and which are collectively operated by the ILT foundation under a joint scientific policy. The ILT resources have benefited from the following recent major funding sources: CNRS-INSU, Observatoire de Paris and Université d'Orléans, France; BMBF, MIWF-NRW, MPG, Germany; Science Foundation Ireland (SFI), Department of Business, Enterprise and Innovation (DBEI), Ireland; NWO, The Netherlands; The Science and Technology Facilities Council, UK; Ministry of Science and Higher Education, Poland; The Istituto Nazionale di Astrofisica (INAF), Italy.
This research made use of the Dutch national e-infrastructure with support of the SURF Cooperative (e-infra 180169) and the LOFAR e-infra group. The Jülich LOFAR Long Term Archive and the German LOFAR network are both coordinated and operated by the Jülich Supercomputing Centre (JSC), and computing resources on the supercomputer JUWELS at JSC were provided by the Gauss Centre for Supercomputing e.V. (grant CHTB00) through the John von Neumann Institute for Computing (NIC).
This research made use of the University of Hertfordshire high-performance computing facility and the LOFAR-UK computing facility located at the University of Hertfordshire and supported by STFC [ST/P000096/1], and of the Italian LOFAR IT computing infrastructure supported and operated by INAF, and by the Physics Department of Turin university (under an agreement with Consorzio Interuniversitario per la Fisica Spaziale) at the C3S Supercomputing Centre, Italy.
\end{acknowledgements}
\bibliographystyle{aa}
\bibliography{references.bib}
\begin{appendix} %First appendix
\section{Deriving the radio-based \hbetaradio{} flux density} \label{appendix sec: radio-based Hbeta flux details}
We derive the expected \hbeta{} line flux density from radio (\hbetaradio{}) following \citetalias{Hunt04} (see their Appendix Eq. A10) which relates the radio-derived thermal flux density and the \hbeta{} line flux density. This relation has been found to be better applicable for low-metallicity \hii{} with high electron temperatures ($> 10,000$ K) which is the case for \xsfg{s}. Here we also assumed a ionized Helium to Hydrogen ratio of 0.08 which is typical of low-metallicty environments \citep{Melnick92, Hunt04}. We used the T$_\mathrm{e}$ values from \citet{Izotov20-diverseLyA, Izotov24-LyA-metalpoor}. For \bbten{} and \bbthirteen{} we used a fixed value of $20,000$ K.
We have used the observed high-frequency \vla{} $10$ GHz (X-band) flux densities to derive \hbetaradio{}. This is because for \xsfg{s} we find that their entire \rcsed{} is significantly thermally dominant, thus further reducing any minor contribution from non-thermal emission. Further, we prefer the X-band since it has the largest coverage for our sample as opposed to the Ku-band ($15$ GHz).
\end{appendix}
\end{document}
gmrt-L & $< 64.5$ & $< 396.0$ & -- & $< 57.3$ & -- & $478\pm36.9$ & $58\pm17.0$ & -- \\
Beam-gmrt-L & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
PA-gmrt-L & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
\\
\hline
\\
vla-L & -- & -- & -- & $< 41.1$ & -- & -- & -- & -- \\
Beam-vla-L & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-vla-L & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
\\
\hline
\\
S & $< 36.6$ & $< 27.3$ & $< 31.8$ & $68.7\pm9.6$ & $< 32.7$ & $330.0\pm39.6$ & -- & $< 29.7$ \\
Beam-S & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-S & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
C & $34.6\pm6.4$ & $< 16.8$ & $< 18.9$ & $61.9\pm5.7$ & $< 19.8$ & $170.0\pm15.14$ & $46.45\pm4.6$ & $< 18.6$ \\
Beam-C & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-C & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
X & $36.0\pm6.6$ & $< 18.4$ & $< 39.9$ & $53.9\pm6.8$ & $31.0\pm6.6$ & $210.0\pm12.16$ & $43.5\pm7.6$ & $< 16.7$ \\
Beam-X & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-X & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
Ku & -- & $< 17.2$ & -- & $69.0\pm5.0$ & $28.3\pm5.4$ & $180.0\pm22.42$ & $45.14\pm6.0$ & $< 13.9$ \\
Beam-Ku & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-Ku & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
LOFAR & -- & $< 532.4$ & $< 169.5$ & $< 200.2$ & $< 181.9$ & $400.68\pm126.6$ & $< 225.8$ & $< 316.0$ \\
Beam-LOFAR & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ & $0.0\prime\prime \times0.0\prime\prime$ \\
Beam-PA-LOFAR & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ & $0.0$ \\
\begin{sidewaystable*}
\caption{Observation details of \xsfg{}.} %\\
\label{table: obs details}
\begin{tabular}{cccccccccc}
\hline
Band & Obs. Details & J0159+0751 & J0811+4730 & J0820+5431 & J1032+4919 & J1355+4651 & J150934+373146 & J160810+352809 & J2229+2725 \\
\hline \hline \\
& Date Observed & & & & & & & & \\
GMRT L & Flux Cal & & & & & & & & \\
& Phase Cal & & & & & & & & \\
& Bandwidth & & & & & & & & \\
\hline \\
& Date Observed & & & & 2024-06-12 & & & & \\
VLA L & Flux Cal & & & & 3C147 & & & & \\
&Phase Cal & & & & J1035+5628 & & & & \\
& Bandwidth & & & & & & & & \\
\hline \\
& Date Observed & & & & 2023-01-14 & & 2024-05-02 & 2024-04-27 & \\
VLA S,C & Flux Cal & & & & 3C147 & & 3C48 & 3C286 & \\
& Phase Cal & & & & J1035+5628 & & J1613+3412 & J1613+3412 & \\
& Bandwidth & & & & & & & & \\
& Date Observed & & 2024-01-26 & 2024-01-26 & & 2024-01-26 & & & \\
VLA X & Flux Cal & 3C48 & 3C48 & 3C48 & & 3C48 & & & \\
& Phase Cal & J0149+0555 & J0808+4950 & J0824+5552 & & J1417+4607 & & & \\
& Bandwidth & & & & & & & & \\
\hline \\
VLA Ku & Flux Cal & & & & & & 3C48 & 3C48 & 3C48 \\
& Phase Cal & & & & & & J1613+3412 & J1613+3412 & J2236+2828 \\
& Bandwidth & & & & & & & & \\
\hline
\end{tabular}
\end{sidewaystable*}
\begin{thebibliography}{103}
\expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
\bibitem[{{Adamo} {et~al.}(2024){Adamo}, {Bradley}, {Vanzella}, {Claeyssens}, {Welch}, {Diego}, {Mahler}, {Oguri}, {Sharon}, {Abdurro'uf}, {Hsiao}, {Xu}, {Messa}, {Lassen}, {Zackrisson}, {Brammer}, {Coe}, {Kokorev}, {Ricotti}, {Zitrin}, {Fujimoto}, {Inoue}, {Resseguier}, {Rigby}, {Jim{\'e}nez-Teja}, {Windhorst}, {Hashimoto}, \& {Tamura}}]{Adamo24}
{Adamo}, A., {Bradley}, L.~D., {Vanzella}, E., {et~al.} 2024, \nat, 632, 513
\bibitem[{{Atek} {et~al.}(2024){Atek}, {Labb{\'e}}, {Furtak}, {Chemerynska}, {Fujimoto}, {Setton}, {Miller}, {Oesch}, {Bezanson}, {Price}, {Dayal}, {Zitrin}, {Kokorev}, {Weaver}, {Brammer}, {Dokkum}, {Williams}, {Cutler}, {Feldmann}, {Fudamoto}, {Greene}, {Leja}, {Maseda}, {Muzzin}, {Pan}, {Papovich}, {Nelson}, {Nanayakkara}, {Stark}, {Stefanon}, {Suess}, {Wang}, \& {Whitaker}}]{Atek24}
{Atek}, H., {Labb{\'e}}, I., {Furtak}, L.~J., {et~al.} 2024, \nat, 626, 975
\bibitem[{{Atek} {et~al.}(2015){Atek}, {Richard}, {Jauzac}, {Kneib}, {Natarajan}, {Limousin}, {Schaerer}, {Jullo}, {Ebeling}, {Egami}, \& {Clement}}]{Atek15}
{Atek}, H., {Richard}, J., {Jauzac}, M., {et~al.} 2015, \apj, 814, 69
\bibitem[{{Bait} {et~al.}(2024){Bait}, {Borthakur}, {Schaerer}, {Momjian}, {Sebastian}, {Saldana-Lopez}, {Flury}, {Chisholm}, {Marques-Chaves}, {Jaskot}, {Ferguson}, {Worseck}, {Ji}, {Komarova}, {Trebitsch}, {Hayes}, {Pentericci}, {Ostlin}, {Thuan}, {Amor{\'\i}n}, {Wang}, {Xu}, \& {Sargent}}]{Bait24a}
{Bait}, O., {Borthakur}, S., {Schaerer}, D., {et~al.} 2024, \aap, 688, A198
\bibitem[{{Bassett} {et~al.}(2019){Bassett}, {Ryan-Weber}, {Cooke}, {Diaz}, {Nanayakkara}, {Yuan}, {Spitler}, {Me{\v{s}}tri{\'c}}, {Garel}, {Sawicki}, {Gwyn}, \& {Golob}}]{Bassett19_lack_of_O32_fesc}
{Bassett}, R., {Ryan-Weber}, E.~V., {Cooke}, J., {et~al.} 2019, \mnras, 483, 5223
\bibitem[{{Borkar} {et~al.}(2024){Borkar}, {Grossov{\'a}}, {Svoboda}, {Moravec}, {Kouroumpatzakis}, {Boorman}, {Adamcov{\'a}}, {Mingo}, \& {Ehle}}]{Borkar24-BB-LOFAR}
{Borkar}, A., {Grossov{\'a}}, R., {Svoboda}, J., {et~al.} 2024, \aap, 687, A137
\bibitem[{{Borthakur} {et~al.}(2014){Borthakur}, {Heckman}, {Leitherer}, \& {Overzier}}]{Borthakur14}
{Borthakur}, S., {Heckman}, T.~M., {Leitherer}, C., \& {Overzier}, R.~A. 2014, Science, 346, 216
\bibitem[{{Bouwens} {et~al.}(2015){Bouwens}, {Illingworth}, {Oesch}, {Caruana}, {Holwerda}, {Smit}, \& {Wilkins}}]{Bouwens15}
{Bouwens}, R.~J., {Illingworth}, G.~D., {Oesch}, P.~A., {et~al.} 2015, \apj, 811, 140
\bibitem[{{Cannon} \& {Skillman}(2004)}]{Cannon04}
{Cannon}, J.~M. \& {Skillman}, E.~D. 2004, \apj, 610, 772
\bibitem[{{Cardamone} {et~al.}(2009){Cardamone}, {Schawinski}, {Sarzi}, {Bamford}, {Bennert}, {Urry}, {Lintott}, {Keel}, {Parejko}, {Nichol}, {Thomas}, {Andreescu}, {Murray}, {Raddick}, {Slosar}, {Szalay}, \& {Vandenberg}}]{cardamone2009}
{Cardamone}, C., {Schawinski}, K., {Sarzi}, M., {et~al.} 2009, \mnras, 399, 1191
\bibitem[{{Cardelli} {et~al.}(1989){Cardelli}, {Clayton}, \& {Mathis}}]{Cardelli89}
{Cardelli}, J.~A., {Clayton}, G.~C., \& {Mathis}, J.~S. 1989, \apj, 345, 245
\bibitem[{{Carilli} {et~al.}(2008){Carilli}, {Lee}, {Capak}, {Schinnerer}, {Lee}, {McCraken}, {Yun}, {Scoville}, {Smol{\v c}i{\'c}}, {Giavalisco}, {Datta}, {Taniguchi}, \& {Urry}}]{carilli2008}
{Carilli}, C.~L., {Lee}, N., {Capak}, P., {et~al.} 2008, \apj, 689, 883
\bibitem[{{Carr} {et~al.}(2024){Carr}, {Cen}, {Scarlata}, {Xu}, {Henry}, {Marques-Chaves}, {Schaerer}, {Amor{\'\i}n}, {Oey}, {Komarova}, {Flury}, {Jaskot}, {Saldana-Lopez}, {Ji}, {Huberty}, {Heckman}, {Ostlin}, {Bait}, {Hayes}, {Thuan}, {Berg}, {Giavalisco}, {Borthakur}, {Chisholm}, {Ferguson}, {Michel-Dansac}, {Verhamme}, \& {Worseck}}]{Carr24}
{Carr}, C.~A., {Cen}, R., {Scarlata}, C., {et~al.} 2024, arXiv e-prints, arXiv:2409.05180
\bibitem[{{Chakraborti} {et~al.}(2012){Chakraborti}, {Yadav}, {Cardamone}, \& {Ray}}]{chakraborti2012}
{Chakraborti}, S., {Yadav}, N., {Cardamone}, C., \& {Ray}, A. 2012, \apjl, 746, L6
\bibitem[{{Chandola} {et~al.}(2024){Chandola}, {Tsai}, {Saikia}, {Li}, {Li}, \& {Ma}}]{Chandola24b}
{Chandola}, Y., {Tsai}, C.-W., {Saikia}, D.~J., {et~al.} 2024, arXiv e-prints, arXiv:2411.13527
\bibitem[{{Chy{\.z}y} {et~al.}(2018){Chy{\.z}y}, {Jurusik}, {Piotrowska}, {Nikiel-Wroczy{\'n}ski}, {Heesen}, {Vacca}, {Nowak}, {Paladino}, {Surma}, {Sridhar}, {Heald}, {Beck}, {Conway}, {Sendlinger}, {Cury{\l}o}, {Mulcahy}, {Broderick}, {Hardcastle}, {Callingham}, {G{\"u}rkan}, {Iacobelli}, {R{\"o}ttgering}, {Adebahr}, {Shulevski}, {Dettmar}, {Breton}, {Clarke}, {Farnes}, {Orr{\'u}}, {Pandey}, {Pandey-Pommier}, {Pizzo}, {Riseley}, {Rowlinson}, {Scaife}, {Stewart}, {van der Horst}, \& {van Weeren}}]{Chyzy18}
{Chy{\.z}y}, K.~T., {Jurusik}, W., {Piotrowska}, J., {et~al.} 2018, \aap, 619, A36
\bibitem[{{Clemens} {et~al.}(2010){Clemens}, {Scaife}, {Vega}, \& {Bressan}}]{Clemens10}
{Clemens}, M.~S., {Scaife}, A., {Vega}, O., \& {Bressan}, A. 2010, \mnras, 405, 887
\bibitem[{{Condon}(1992)}]{condon1992}
{Condon}, J.~J. 1992, \araa, 30, 575
\bibitem[{{Condon} \& {Ransom}(2016)}]{Condon_16_essential_RA}
{Condon}, J.~J. \& {Ransom}, S.~M. 2016, {Essential Radio Astronomy}
\bibitem[{{Deeg} {et~al.}(1993){Deeg}, {Brinks}, {Duric}, {Klein}, \& {Skillman}}]{Deeg93}
{Deeg}, H.-J., {Brinks}, E., {Duric}, N., {Klein}, U., \& {Skillman}, E. 1993, \apj, 410, 626
\bibitem[{{Dey} {et~al.}(2024){Dey}, {Goyal}, {Ma{\l}ek}, \& {D{\'\i}az-Santos}}]{Dey24_ULIRGS}
{Dey}, S., {Goyal}, A., {Ma{\l}ek}, K., \& {D{\'\i}az-Santos}, T. 2024, \apj, 966, 61
\bibitem[{{Dutta} {et~al.}(2024){Dutta}, {Bera}, {Bait}, {Narayan}, {Sebastian}, \& {Vaddi}}]{Dutta24}
{Dutta}, S., {Bera}, A., {Bait}, O., {et~al.} 2024, \mnras [\eprint[arXiv]{2406.08341}]
\bibitem[{{Endsley} {et~al.}(2023){Endsley}, {Stark}, {Whitler}, {Topping}, {Chen}, {Plat}, {Chisholm}, \& {Charlot}}]{Endsley23}
{Endsley}, R., {Stark}, D.~P., {Whitler}, L., {et~al.} 2023, \mnras, 524, 2312
\bibitem[{{Fall} {et~al.}(2010){Fall}, {Krumholz}, \& {Matzner}}]{Fall10_SFE}
{Fall}, S.~M., {Krumholz}, M.~R., \& {Matzner}, C.~D. 2010, \apjl, 710, L142
\bibitem[{{Farcy} {et~al.}(2022){Farcy}, {Rosdahl}, {Dubois}, {Blaizot}, \& {Martin-Alvarez}}]{Farcy22}
{Farcy}, M., {Rosdahl}, J., {Dubois}, Y., {Blaizot}, J., \& {Martin-Alvarez}, S. 2022, \mnras, 513, 5000
\bibitem[{{Flury} {et~al.}(2022{\natexlab{a}}){Flury}, {Jaskot}, {Ferguson}, {Worseck}, {Makan}, {Chisholm}, {Saldana-Lopez}, {Schaerer}, {McCandliss}, {Wang}, {Ford}, {Heckman}, {Ji}, {Giavalisco}, {Amorin}, {Atek}, {Blaizot}, {Borthakur}, {Carr}, {Castellano}, {Cristiani}, {De Barros}, {Dickinson}, {Finkelstein}, {Fleming}, {Fontanot}, {Garel}, {Grazian}, {Hayes}, {Henry}, {Mauerhofer}, {Micheva}, {Oey}, {Ostlin}, {Papovich}, {Pentericci}, {Ravindranath}, {Rosdahl}, {Rutkowski}, {Santini}, {Scarlata}, {Teplitz}, {Thuan}, {Trebitsch}, {Vanzella}, {Verhamme}, \& {Xu}}]{Flury22a}
{Flury}, S.~R., {Jaskot}, A.~E., {Ferguson}, H.~C., {et~al.} 2022{\natexlab{a}}, \apjs, 260, 1
\bibitem[{{Flury} {et~al.}(2022{\natexlab{b}}){Flury}, {Jaskot}, {Ferguson}, {Worseck}, {Makan}, {Chisholm}, {Saldana-Lopez}, {Schaerer}, {McCandliss}, {Xu}, {Wang}, {Oey}, {Ford}, {Heckman}, {Ji}, {Giavalisco}, {Amor{\'\i}n}, {Atek}, {Blaizot}, {Borthakur}, {Carr}, {Castellano}, {De Barros}, {Dickinson}, {Finkelstein}, {Fleming}, {Fontanot}, {Garel}, {Grazian}, {Hayes}, {Henry}, {Mauerhofer}, {Micheva}, {Ostlin}, {Papovich}, {Pentericci}, {Ravindranath}, {Rosdahl}, {Rutkowski}, {Santini}, {Scarlata}, {Teplitz}, {Thuan}, {Trebitsch}, {Vanzella}, \& {Verhamme}}]{Flury22b}
{Flury}, S.~R., {Jaskot}, A.~E., {Ferguson}, H.~C., {et~al.} 2022{\natexlab{b}}, \apj, 930, 126
\bibitem[{{Flury} {et~al.}(2024){Flury}, {Jaskot}, {Saldana-Lopez}, {Oey}, {Chisholm}, {Amor{\'\i}n}, {Bait}, {Borthakur}, {Carr}, {Ferguson}, {Giavalisco}, {Hayes}, {Heckman}, {Henry}, {Ji}, {Komarova}, {Leclercq}, {Le Reste}, {McCandliss}, {Marques-Chaves}, {{\"O}stlin}, {Pentericci}, {Ravindranath}, {Rutkowski}, {Scarlata}, {Schaerer}, {Thuan}, {Trebitsch}, {Vanzella}, {Verhamme}, {Wang}, {Worseck}, \& {Xu}}]{Flury24_feedback}
{Flury}, S.~R., {Jaskot}, A.~E., {Saldana-Lopez}, A., {et~al.} 2024, arXiv e-prints, arXiv:2409.12118
\bibitem[{{Fujimoto} {et~al.}(2023){Fujimoto}, {Arrabal Haro}, {Dickinson}, {Finkelstein}, {Kartaltepe}, {Larson}, {Burgarella}, {Bagley}, {Behroozi}, {Chworowsky}, {Hirschmann}, {Trump}, {Wilkins}, {Yung}, {Koekemoer}, {Papovich}, {Pirzkal}, {Ferguson}, {Fontana}, {Grogin}, {Grazian}, {Kewley}, {Kocevski}, {Lotz}, {Pentericci}, {Ravindranath}, {Somerville}, {Wilkins}, {Amor{\'\i}n}, {Backhaus}, {Calabr{\`o}}, {Casey}, {Cooper}, {Fern{\'a}ndez}, {Franco}, {Giavalisco}, {Hathi}, {Harish}, {Hutchison}, {Iyer}, {Jung}, {Lucas}, \& {Zavala}}]{Fujimoto23_JWST}
{Fujimoto}, S., {Arrabal Haro}, P., {Dickinson}, M., {et~al.} 2023, \apjl, 949, L25
\bibitem[{{Fujimoto} {et~al.}(2024){Fujimoto}, {Ouchi}, {Kohno}, {Valentino}, {Gim{\'e}nez-Arteaga}, {Brammer}, {Furtak}, {Kohandel}, {Oguri}, {Pallottini}, {Richard}, {Zitrin}, {Bauer}, {Boylan-Kolchin}, {Dessauges-Zavadsky}, {Egami}, {Finkelstein}, {Ma}, {Smail}, {Watson}, {Hutchison}, {Rigby}, {Welch}, {Ao}, {Bradley}, {Caminha}, {Caputi}, {Espada}, {Endsley}, {Fudamoto}, {Gonz{\'a}lez-L{\'o}pez}, {Hatsukade}, {Koekemoer}, {Kokorev}, {Laporte}, {Lee}, {Magdis}, {Ono}, {Rizzo}, {Shibuya}, {Shimasaku}, {Sun}, {Toft}, {Umehata}, {Wang}, \& {Yajima}}]{Fujimoto24}
{Fujimoto}, S., {Ouchi}, M., {Kohno}, K., {et~al.} 2024, arXiv e-prints, arXiv:2402.18543
\bibitem[{{Galvin} {et~al.}(2018){Galvin}, {Seymour}, {Marvil}, {Filipovi{\'c}}, {Tothill}, {McDermid}, {Hurley-Walker}, {Hancock}, {Callingham}, {Cook}, {Norris}, {Bell}, {Dwarakanath}, {For}, {Gaensler}, {Hindson}, {Johnston-Hollitt}, {Kapi{\'n}ska}, {Lenc}, {McKinley}, {Morgan}, {Offringa}, {Procopio}, {Staveley-Smith}, {Wayth}, {Wu}, \& {Zheng}}]{Galvin18}
{Galvin}, T.~J., {Seymour}, N., {Marvil}, J., {et~al.} 2018, \mnras, 474, 779
\bibitem[{{Greis} {et~al.}(2017){Greis}, {Stanway}, {Levan}, {Davies}, \& {Eldridge}}]{greis2017}
{Greis}, S.~M.~L., {Stanway}, E.~R., {Levan}, A.~J., {Davies}, L.~J.~M., \& {Eldridge}, J.~J. 2017, \mnras, 470, 489
\bibitem[{{Grundy} {et~al.}(2024){Grundy}, {Seymour}, {Wong}, {Lee-Waddell}, {Galvin}, \& {Cluver}}]{Grundy24}
{Grundy}, J.~A., {Seymour}, N., {Wong}, O.~I., {et~al.} 2024, arXiv e-prints, arXiv:2412.03143
\bibitem[{{Hayes} {et~al.}(2007){Hayes}, {{\"O}stlin}, {Atek}, {Kunth}, {Mas-Hesse}, {Leitherer}, {Jim{\'e}nez-Bail{\'o}n}, \& {Adamo}}]{Hayes07}
{Hayes}, M., {{\"O}stlin}, G., {Atek}, H., {et~al.} 2007, \mnras, 382, 1465
\bibitem[{{Hirashita} \& {Hunt}(2006)}]{Hirashita06}
{Hirashita}, H. \& {Hunt}, L.~K. 2006, \aap, 460, 67
\bibitem[{{Hunt} {et~al.}(2004){Hunt}, {Dyer}, {Thuan}, \& {Ulvestad}}]{Hunt04}
{Hunt}, L.~K., {Dyer}, K.~K., {Thuan}, T.~X., \& {Ulvestad}, J.~S. 2004, \apj, 606, 853
\bibitem[{{Inoguchi} {et~al.}(2024){Inoguchi}, {Hosokawa}, {Fukushima}, {Tanaka}, {Yajima}, \& {Mineshige}}]{Inoguchi24_FFA-sims}
{Inoguchi}, M., {Hosokawa}, T., {Fukushima}, H., {et~al.} 2024, \mnras, 527, 3612
\bibitem[{{Inoue} {et~al.}(2014){Inoue}, {Shimizu}, {Iwata}, \& {Tanaka}}]{Inoue14}
{Inoue}, A.~K., {Shimizu}, I., {Iwata}, I., \& {Tanaka}, M. 2014, \mnras, 442, 1805
\bibitem[{{Izotov} {et~al.}(2016{\natexlab{a}}){Izotov}, {Guseva}, {Fricke}, \& {Henkel}}]{Izotov16c}
{Izotov}, Y.~I., {Guseva}, N.~G., {Fricke}, K.~J., \& {Henkel}, C. 2016{\natexlab{a}}, \mnras, 462, 4427
\bibitem[{{Izotov} {et~al.}(2017{\natexlab{a}}){Izotov}, {Guseva}, {Fricke}, {Henkel}, \& {Schaerer}}]{Izotov17}
{Izotov}, Y.~I., {Guseva}, N.~G., {Fricke}, K.~J., {Henkel}, C., \& {Schaerer}, D. 2017{\natexlab{a}}, \mnras, 467, 4118
\bibitem[{{Izotov} {et~al.}(2021{\natexlab{a}}){Izotov}, {Guseva}, {Fricke}, {Henkel}, {Schaerer}, \& {Thuan}}]{Izotov21a}
{Izotov}, Y.~I., {Guseva}, N.~G., {Fricke}, K.~J., {et~al.} 2021{\natexlab{a}}, \aap, 646, A138
\bibitem[{{Izotov} {et~al.}(2011){Izotov}, {Guseva}, \& {Thuan}}]{Izotov11}
{Izotov}, Y.~I., {Guseva}, N.~G., \& {Thuan}, T.~X. 2011, \apj, 728, 161
\bibitem[{{Izotov} {et~al.}(2016{\natexlab{b}}){Izotov}, {Orlitov{\'a}}, {Schaerer}, {Thuan}, {Verhamme}, {Guseva}, \& {Worseck}}]{Izotov16a}
{Izotov}, Y.~I., {Orlitov{\'a}}, I., {Schaerer}, D., {et~al.} 2016{\natexlab{b}}, \nat, 529, 178
\bibitem[{{Izotov} {et~al.}(2016{\natexlab{c}}){Izotov}, {Schaerer}, {Thuan}, {Worseck}, {Guseva}, {Orlitov{\'a}}, \& {Verhamme}}]{Izotov16b}
{Izotov}, Y.~I., {Schaerer}, D., {Thuan}, T.~X., {et~al.} 2016{\natexlab{c}}, \mnras, 461, 3683
\bibitem[{{Izotov} {et~al.}(2018{\natexlab{a}}){Izotov}, {Schaerer}, {Worseck}, {Guseva}, {Thuan}, {Verhamme}, {Orlitov{\'a}}, \& {Fricke}}]{Izotov18a}
{Izotov}, Y.~I., {Schaerer}, D., {Worseck}, G., {et~al.} 2018{\natexlab{a}}, \mnras, 474, 4514
\bibitem[{{Izotov} {et~al.}(2020){Izotov}, {Schaerer}, {Worseck}, {Verhamme}, {Guseva}, {Thuan}, {Orlitov{\'a}}, \& {Fricke}}]{Izotov20-diverseLyA}
{Izotov}, Y.~I., {Schaerer}, D., {Worseck}, G., {et~al.} 2020, \mnras, 491, 468
\bibitem[{{Izotov} {et~al.}(2017{\natexlab{b}}){Izotov}, {Thuan}, \& {Guseva}}]{Izotov17b}
{Izotov}, Y.~I., {Thuan}, T.~X., \& {Guseva}, N.~G. 2017{\natexlab{b}}, \mnras, 471, 548
\bibitem[{{Izotov} {et~al.}(2024){Izotov}, {Thuan}, {Guseva}, {Schaerer}, {Worseck}, \& {Verhamme}}]{Izotov24-LyA-metalpoor}
{Izotov}, Y.~I., {Thuan}, T.~X., {Guseva}, N.~G., {et~al.} 2024, \mnras, 527, 281
\bibitem[{{Izotov} {et~al.}(2021{\natexlab{b}}){Izotov}, {Worseck}, {Schaerer}, {Guseva}, {Chisholm}, {Thuan}, {Fricke}, \& {Verhamme}}]{Izotov21b}
{Izotov}, Y.~I., {Worseck}, G., {Schaerer}, D., {et~al.} 2021{\natexlab{b}}, \mnras, 503, 1734
\bibitem[{{Izotov} {et~al.}(2018{\natexlab{b}}){Izotov}, {Worseck}, {Schaerer}, {Guseva}, {Thuan}, {Fricke}, \& {Orlitov{\'a}}}]{Izotov18b-vsep}
{Izotov}, Y.~I., {Worseck}, G., {Schaerer}, D., {et~al.} 2018{\natexlab{b}}, \mnras, 478, 4851
\bibitem[{{Jaskot} {et~al.}(2019){Jaskot}, {Dowd}, {Oey}, {Scarlata}, \& {McKinney}}]{Jaskot19}
{Jaskot}, A.~E., {Dowd}, T., {Oey}, M.~S., {Scarlata}, C., \& {McKinney}, J. 2019, \apj, 885, 96
\bibitem[{{Jaskot} \& {Oey}(2013)}]{Jaskot-Oey13}
{Jaskot}, A.~E. \& {Oey}, M.~S. 2013, \apj, 766, 91
\bibitem[{{Jaskot} {et~al.}(2024){Jaskot}, {Silveyra}, {Plantinga}, {Flury}, {Hayes}, {Chisholm}, {Heckman}, {Pentericci}, {Schaerer}, {Trebitsch}, {Verhamme}, {Carr}, {Ferguson}, {Ji}, {Giavalisco}, {Henry}, {Marques-Chaves}, {{\"O}stlin}, {Saldana-Lopez}, {Scarlata}, {Worseck}, \& {Xu}}]{Jaskot24-multivariate}
{Jaskot}, A.~E., {Silveyra}, A.~C., {Plantinga}, A., {et~al.} 2024, arXiv e-prints, arXiv:2406.10171
\bibitem[{{Jecmen} \& {Oey}(2023)}]{Jeckmen23-delayedfeedback}
{Jecmen}, M.~C. \& {Oey}, M.~S. 2023, \apj, 958, 149
\bibitem[{{Johnson} \& {Kobulnicky}(2003)}]{Johnson03}
{Johnson}, K.~E. \& {Kobulnicky}, H.~A. 2003, \apj, 597, 923
\bibitem[{{Jones} {et~al.}(2020){Jones}, {Sanders}, {Roberts-Borsani}, {Ellis}, {Laporte}, {Treu}, \& {Harikane}}]{Jones20}
{Jones}, T., {Sanders}, R., {Roberts-Borsani}, G., {et~al.} 2020, \apj, 903, 150
\bibitem[{{Kanekar} {et~al.}(2021){Kanekar}, {Ghosh}, {Rhoads}, {Malhotra}, {Harish}, {Chengalur}, \& {Jones}}]{Kanekar21_HI-GP}
{Kanekar}, N., {Ghosh}, T., {Rhoads}, J., {et~al.} 2021, \apjl, 913, L15
\bibitem[{{Katz} {et~al.}(2020){Katz}, {{\v{D}}urov{\v{c}}{\'\i}kov{\'a}}, {Kimm}, {Rosdahl}, {Blaizot}, {Haehnelt}, {Devriendt}, {Slyz}, {Ellis}, \& {Laporte}}]{Katz20_orientation}
{Katz}, H., {{\v{D}}urov{\v{c}}{\'\i}kov{\'a}}, D., {Kimm}, T., {et~al.} 2020, \mnras, 498, 164
\bibitem[{{Kennicutt} \& {Evans}(2012)}]{Kennicutt12}
{Kennicutt}, R.~C. \& {Evans}, N.~J. 2012, \araa, 50, 531
\bibitem[{{Kennicutt}(1998)}]{Kennicutt98}
{Kennicutt}, Jr., R.~C. 1998, \araa, 36, 189
\bibitem[{{Klein} {et~al.}(2018){Klein}, {Lisenfeld}, \& {Verley}}]{klein2018}
{Klein}, U., {Lisenfeld}, U., \& {Verley}, S. 2018, \aap, 611, A55
\bibitem[{{Klein} {et~al.}(1991){Klein}, {Weiland}, \& {Brinks}}]{Klein91}
{Klein}, U., {Weiland}, H., \& {Brinks}, E. 1991, \aap, 246, 323
\bibitem[{{Kobulnicky} \& {Johnson}(1999)}]{Kobulnicky99}
{Kobulnicky}, H.~A. \& {Johnson}, K.~E. 1999, \apj, 527, 154
\bibitem[{{Langeroodi} {et~al.}(2023){Langeroodi}, {Hjorth}, {Chen}, {Kelly}, {Williams}, {Lin}, {Scarlata}, {Zitrin}, {Broadhurst}, {Diego}, {Huang}, {Filippenko}, {Foley}, {Jha}, {Koekemoer}, {Oguri}, {Perez-Fournon}, {Pierel}, {Poidevin}, \& {Strolger}}]{Langeroodi23}
{Langeroodi}, D., {Hjorth}, J., {Chen}, W., {et~al.} 2023, \apj, 957, 39
\bibitem[{{Lisenfeld} {et~al.}(2004){Lisenfeld}, {Wilding}, {Pooley}, \& {Alexander}}]{lisenfeld2004}
{Lisenfeld}, U., {Wilding}, T.~W., {Pooley}, G.~G., \& {Alexander}, P. 2004, \mnras, 349, 1335
\bibitem[{{Marques-Chaves} {et~al.}(2024){Marques-Chaves}, {Schaerer}, {Vanzella}, {Verhamme}, {Dessauges-Zavadsky}, {Chisholm}, {Leclercq}, {Upadhyaya}, {Alvarez-Marquez}, {Colina}, {Garel}, \& {Messa}}]{Marques-Chaves24_efficientSFmode}
{Marques-Chaves}, R., {Schaerer}, D., {Vanzella}, E., {et~al.} 2024, arXiv e-prints, arXiv:2407.18804
\bibitem[{{Mascia} {et~al.}(2024){Mascia}, {Pentericci}, {Calabr{\`o}}, {Santini}, {Napolitano}, {Arrabal Haro}, {Castellano}, {Dickinson}, {Ocvirk}, {Lewis}, {Amor{\'\i}n}, {Bagley}, {Bhatawdekar}, {Cleri}, {Costantin}, {Dekel}, {Finkelstein}, {Fontana}, {Giavalisco}, {Grogin}, {Hathi}, {Hirschmann}, {Holwerda}, {Jung}, {Kartaltepe}, {Koekemoer}, {Lucas}, {Papovich}, {P{\'e}rez-Gonz{\'a}lez}, {Pirzkal}, {Trump}, {Wilkins}, \& {Yung}}]{Mascia24}
{Mascia}, S., {Pentericci}, L., {Calabr{\`o}}, A., {et~al.} 2024, \aap, 685, A3
\bibitem[{McMullin {et~al.}(2007)McMullin, Waters, Schiebel, Young, \& Golap}]{mcmullin2007casa}
McMullin, J.~P., Waters, B., Schiebel, D., Young, W., \& Golap, K. 2007, in Astronomical data analysis software and systems XVI, Vol. 376, 127
\bibitem[{{Melnick} {et~al.}(1992){Melnick}, {Heydari-Malayeri}, \& {Leisy}}]{Melnick92}
{Melnick}, J., {Heydari-Malayeri}, M., \& {Leisy}, P. 1992, \aap, 253, 16
\bibitem[{{Menon} {et~al.}(2024){Menon}, {Burkhart}, {Somerville}, {Thompson}, \& {Sternberg}}]{Menon24a_bursty_radiation_outlfows}
{Menon}, S.~H., {Burkhart}, B., {Somerville}, R.~S., {Thompson}, T.~A., \& {Sternberg}, A. 2024, arXiv e-prints, arXiv:2408.14591
\bibitem[{{Messa} {et~al.}(2024){Messa}, {Vanzella}, {Loiacono}, {Bergamini}, {Castellano}, {Sun}, {Willott}, {Windhorst}, {Yan}, {Angora}, {Rosati}, {Adamo}, {Annibali}, {Bolamperti}, {Brada{\v{c}}}, {Bradley}, {Calura}, {Claeyssens}, {Comastri}, {Conselice}, {D'Silva}, {Dickinson}, {Frye}, {Grillo}, {Grogin}, {Gruppioni}, {Koekemoer}, {Meneghetti}, {Me{\v{s}}tri{\'c}}, {Pascale}, {Ravindranath}, {Ricotti}, {Summers}, \& {Zanella}}]{Messa24a}
{Messa}, M., {Vanzella}, E., {Loiacono}, F., {et~al.} 2024, arXiv e-prints, arXiv:2407.20331
\bibitem[{{Mohan} \& {Rafferty}(2015)}]{Mohan15}
{Mohan}, N. \& {Rafferty}, D. 2015, {PyBDSF: Python Blob Detection and Source Finder}, Astrophysics Source Code Library, record ascl:1502.007
\bibitem[{{Murphy} {et~al.}(2011){Murphy}, {Condon}, {Schinnerer}, {Kennicutt}, {Calzetti}, {Armus}, {Helou}, {Turner}, {Aniano}, {Beir{\~a}o}, {Bolatto}, {Brandl}, {Croxall}, {Dale}, {Donovan Meyer}, {Draine}, {Engelbracht}, {Hunt}, {Hao}, {Koda}, {Roussel}, {Skibba}, \& {Smith}}]{murphy2011}
{Murphy}, E.~J., {Condon}, J.~J., {Schinnerer}, E., {et~al.} 2011, \apj, 737, 67
\bibitem[{{Nakajima} {et~al.}(2020){Nakajima}, {Ellis}, {Robertson}, {Tang}, \& {Stark}}]{Nakajima20}
{Nakajima}, K., {Ellis}, R.~S., {Robertson}, B.~E., {Tang}, M., \& {Stark}, D.~P. 2020, \apj, 889, 161
\bibitem[{{Nakajima} \& {Ouchi}(2014)}]{Nakajima14-Ionization-Stat}
{Nakajima}, K. \& {Ouchi}, M. 2014, \mnras, 442, 900
\bibitem[{{Niklas} {et~al.}(1997){Niklas}, {Klein}, \& {Wielebinski}}]{Niklas97_thermal_frac}
{Niklas}, S., {Klein}, U., \& {Wielebinski}, R. 1997, \aap, 322, 19
\bibitem[{{Ouchi} {et~al.}(2009){Ouchi}, {Mobasher}, {Shimasaku}, {Ferguson}, {Fall}, {Ono}, {Kashikawa}, {Morokuma}, {Nakajima}, {Okamura}, {Dickinson}, {Giavalisco}, \& {Ohta}}]{Ouchi09}
{Ouchi}, M., {Mobasher}, B., {Shimasaku}, K., {et~al.} 2009, \apj, 706, 1136
\bibitem[{{Patel} {et~al.}(2024){Patel}, {Urquhart}, {Yang}, {Moore}, {Thompson}, {Menten}, \& {Csengeri}}]{Patel24}
{Patel}, A.~L., {Urquhart}, J.~S., {Yang}, A.~Y., {et~al.} 2024, \mnras, 533, 2005
\bibitem[{{Paudel} {et~al.}(2023){Paudel}, {Yoon}, {Bait}, {Sengupta}, {Zee}, {Chhatkuli}, {Adhikari}, \& {Aryal}}]{Paudel23}
{Paudel}, S., {Yoon}, S.-J., {Bait}, O., {et~al.} 2023, \apjl, 951, L36
\bibitem[{{Ramya} {et~al.}(2011){Ramya}, {Kantharia}, \& {Prabhu}}]{ramya2011}
{Ramya}, S., {Kantharia}, N.~G., \& {Prabhu}, T.~P. 2011, \apj, 728, 124
\bibitem[{{Rau} \& {Cornwell}(2011)}]{Rau11}
{Rau}, U. \& {Cornwell}, T.~J. 2011, \aap, 532, A71
\bibitem[{{Rhoads} {et~al.}(2023){Rhoads}, {Wold}, {Harish}, {Kim}, {Pharo}, {Malhotra}, {Gabrielpillai}, {Jiang}, \& {Yang}}]{Rhoads23}
{Rhoads}, J.~E., {Wold}, I. G.~B., {Harish}, S., {et~al.} 2023, \apjl, 942, L14
\bibitem[{{Robertson} {et~al.}(2015){Robertson}, {Ellis}, {Furlanetto}, \& {Dunlop}}]{Robertson15}
{Robertson}, B.~E., {Ellis}, R.~S., {Furlanetto}, S.~R., \& {Dunlop}, J.~S. 2015, \apjl, 802, L19
\bibitem[{{Saldana-Lopez} {et~al.}(2022){Saldana-Lopez}, {Schaerer}, {Chisholm}, {Flury}, {Jaskot}, {Worseck}, {Makan}, {Gazagnes}, {Mauerhofer}, {Verhamme}, {Amor{\'\i}n}, {Ferguson}, {Giavalisco}, {Grazian}, {Hayes}, {Heckman}, {Henry}, {Ji}, {Marques-Chaves}, {McCandliss}, {Oey}, {{\"O}stlin}, {Pentericci}, {Thuan}, {Trebitsch}, {Vanzella}, \& {Xu}}]{Saldana-Lopez22}
{Saldana-Lopez}, A., {Schaerer}, D., {Chisholm}, J., {et~al.} 2022, \aap, 663, A59
\bibitem[{{Saxena} {et~al.}(2024){Saxena}, {Bunker}, {Jones}, {Stark}, {Cameron}, {Witstok}, {Arribas}, {Baker}, {Baum}, {Bhatawdekar}, {Bowler}, {Boyett}, {Carniani}, {Charlot}, {Chevallard}, {Curti}, {Curtis-Lake}, {Eisenstein}, {Endsley}, {Hainline}, {Helton}, {Johnson}, {Kumari}, {Looser}, {Maiolino}, {Rieke}, {Rix}, {Robertson}, {Sandles}, {Simmonds}, {Smit}, {Tacchella}, {Williams}, {Willmer}, \& {Willott}}]{Saxena24}
{Saxena}, A., {Bunker}, A.~J., {Jones}, G.~C., {et~al.} 2024, \aap, 684, A84
\bibitem[{{Schaerer} {et~al.}(2016){Schaerer}, {Izotov}, {Verhamme}, {Orlitov{\'a}}, {Thuan}, {Worseck}, \& {Guseva}}]{Schaerer16}
{Schaerer}, D., {Izotov}, Y.~I., {Verhamme}, A., {et~al.} 2016, \aap, 591, L8
\bibitem[{{Schaerer} {et~al.}(2022){Schaerer}, {Marques-Chaves}, {Barrufet}, {Oesch}, {Izotov}, {Naidu}, {Guseva}, \& {Brammer}}]{Schaerer22}
{Schaerer}, D., {Marques-Chaves}, R., {Barrufet}, L., {et~al.} 2022, \aap, 665, L4
\bibitem[{{Sebastian} \& {Bait}(2019)}]{Sebastian19}
{Sebastian}, B. \& {Bait}, O. 2019, \apjl, 882, L19
\bibitem[{{Shimwell} {et~al.}(2022){Shimwell}, {Hardcastle}, {Tasse}, {Best}, {R{\"o}ttgering}, {Williams}, {Botteon}, {Drabent}, {Mechev}, {Shulevski}, {van Weeren}, {Bester}, {Br{\"u}ggen}, {Brunetti}, {Callingham}, {Chy{\.z}y}, {Conway}, {Dijkema}, {Duncan}, {de Gasperin}, {Hale}, {Haverkorn}, {Hugo}, {Jackson}, {Mevius}, {Miley}, {Morabito}, {Morganti}, {Offringa}, {Oonk}, {Rafferty}, {Sabater}, {Smith}, {Schwarz}, {Smirnov}, {O'Sullivan}, {Vedantham}, {White}, {Albert}, {Alegre}, {Asabere}, {Bacon}, {Bonafede}, {Bonnassieux}, {Brienza}, {Bilicki}, {Bonato}, {Calistro Rivera}, {Cassano}, {Cochrane}, {Croston}, {Cuciti}, {Dallacasa}, {Danezi}, {Dettmar}, {Di Gennaro}, {Edler}, {En{\ss}lin}, {Emig}, {Franzen}, {Garc{\'\i}a-Vergara}, {Grange}, {G{\"u}rkan}, {Hajduk}, {Heald}, {Heesen}, {Hoang}, {Hoeft}, {Horellou}, {Iacobelli}, {Jamrozy}, {Jeli{\'c}}, {Kondapally}, {Kukreti}, {Kunert-Bajraszewska}, {Magliocchetti}, {Mahatma}, {Ma{\l}ek}, {Mandal}, {Massaro}, {Meyer-Zhao}, {Mingo}, {Mostert}, {Nair},
{Nakoneczny}, {Nikiel-Wroczy{\'n}ski}, {Orr{\'u}}, {Pajdosz-{\'S}mierciak}, {Pasini}, {Prandoni}, {van Piggelen}, {Rajpurohit}, {Retana-Montenegro}, {Riseley}, {Rowlinson}, {Saxena}, {Schrijvers}, {Sweijen}, {Siewert}, {Timmerman}, {Vaccari}, {Vink}, {West}, {Wo{\l}owska}, {Zhang}, \& {Zheng}}]{Shimwell22_LoTSS-DR2}
{Shimwell}, T.~W., {Hardcastle}, M.~J., {Tasse}, C., {et~al.} 2022, \aap, 659, A1
\bibitem[{{Shimwell} {et~al.}(2017){Shimwell}, {R{\"o}ttgering}, {Best}, {Williams}, {Dijkema}, {de Gasperin}, {Hardcastle}, {Heald}, {Hoang}, {Horneffer}, {Intema}, {Mahony}, {Mandal}, {Mechev}, {Morabito}, {Oonk}, {Rafferty}, {Retana-Montenegro}, {Sabater}, {Tasse}, {van Weeren}, {Br{\"u}ggen}, {Brunetti}, {Chy{\.z}y}, {Conway}, {Haverkorn}, {Jackson}, {Jarvis}, {McKean}, {Miley}, {Morganti}, {White}, {Wise}, {van Bemmel}, {Beck}, {Brienza}, {Bonafede}, {Calistro Rivera}, {Cassano}, {Clarke}, {Cseh}, {Deller}, {Drabent}, {van Driel}, {Engels}, {Falcke}, {Ferrari}, {Fr{\"o}hlich}, {Garrett}, {Harwood}, {Heesen}, {Hoeft}, {Horellou}, {Israel}, {Kapi{\'n}ska}, {Kunert-Bajraszewska}, {McKay}, {Mohan}, {Orr{\'u}}, {Pizzo}, {Prandoni}, {Schwarz}, {Shulevski}, {Sipior}, {Smith}, {Sridhar}, {Steinmetz}, {Stroe}, {Varenius}, {van der Werf}, {Zensus}, \& {Zwart}}]{Shimwell17_LoTSS-DR1}
{Shimwell}, T.~W., {R{\"o}ttgering}, H.~J.~A., {Best}, P.~N., {et~al.} 2017, \aap, 598, A104
\bibitem[{{Simmonds} {et~al.}(2024){Simmonds}, {Tacchella}, {Hainline}, {Johnson}, {Pusk{\'a}s}, {Robertson}, {Baker}, {Bhatawdekar}, {Boyett}, {Bunker}, {Cargile}, {Carniani}, {Chevallard}, {Curti}, {Curtis-Lake}, {Ji}, {Jones}, {Kumari}, {Laseter}, {Maiolino}, {Maseda}, {Rinaldi}, {Stoffers}, {{\"U}bler}, {Villanueva}, {Williams}, {Willot}, {Witstok}, \& {Zhu}}]{Simmonds24}
{Simmonds}, C., {Tacchella}, S., {Hainline}, K., {et~al.} 2024, arXiv e-prints, arXiv:2409.01286
\bibitem[{{Simmonds} {et~al.}(2023){Simmonds}, {Tacchella}, {Maseda}, {Williams}, {Baker}, {Witten}, {Johnson}, {Robertson}, {Saxena}, {Sun}, {Witstok}, {Bhatawdekar}, {Boyett}, {Bunker}, {Charlot}, {Curtis-Lake}, {Egami}, {Eisenstein}, {Ji}, {Maiolino}, {Sandles}, {Smit}, {{\"U}bler}, \& {Willott}}]{Simmonds23}
{Simmonds}, C., {Tacchella}, S., {Maseda}, M., {et~al.} 2023, \mnras, 523, 5468
\bibitem[{{Tabatabaei} {et~al.}(2017){Tabatabaei}, {Schinnerer}, {Krause}, {Dumas}, {Meidt}, {Damas-Segovia}, {Beck}, {Murphy}, {Mulcahy}, {Groves}, {Bolatto}, {Dale}, {Galametz}, {Sandstrom}, {Boquien}, {Calzetti}, {Kennicutt}, {Hunt}, {De Looze}, \& {Pellegrini}}]{tabatabaei2017}
{Tabatabaei}, F.~S., {Schinnerer}, E., {Krause}, M., {et~al.} 2017, \apj, 836, 185
\bibitem[{{THE CASA TEAM} {et~al.}(2022){THE CASA TEAM}, {Bean}, {Bhatnagar}, {Castro}, {Donovan Meyer}, {Emonts}, {Garcia}, {Garwood}, {Golap}, {Gonzalez Villalba}, {Harris}, {Hayashi}, {Hoskins}, {Hsieh}, {Jagannathan}, {Kawasaki}, {Keimpema}, {Kettenis}, {Lopez}, {Marvil}, {Masters}, {McNichols}, {Mehringer}, {Miel}, {Moellenbrock}, {Montesino}, {Nakazato}, {Ott}, {Petry}, {Pokorny}, {Raba}, {Rau}, {Schiebel}, {Schweighart}, {Sekhar}, {Shimada}, {Small}, {Steeb}, {Sugimoto}, {Suoranta}, {Tsutsumi}, {van Bemmel}, {Verkouter}, {Wells}, {Xiong}, {Szomoru}, {Griffith}, {Glendenning}, \& {Kern}}]{CASA}
{THE CASA TEAM}, {Bean}, B., {Bhatnagar}, S., {et~al.} 2022, arXiv e-prints, arXiv:2210.02276
\bibitem[{{Thuan} {et~al.}(2004){Thuan}, {Hibbard}, \& {L{\'e}vrier}}]{Thuan2004}
{Thuan}, T.~X., {Hibbard}, J.~E., \& {L{\'e}vrier}, F. 2004, \aj, 128, 617
\bibitem[{{Topping} {et~al.}(2024){Topping}, {Stark}, {Senchyna}, {Plat}, {Zitrin}, {Endsley}, {Charlot}, {Furtak}, {Maseda}, {Smit}, {Mainali}, {Chevallard}, {Molyneux}, \& {Rigby}}]{Topping24}
{Topping}, M.~W., {Stark}, D.~P., {Senchyna}, P., {et~al.} 2024, \mnras, 529, 3301
\bibitem[{{Vanzella} {et~al.}(2023){Vanzella}, {Claeyssens}, {Welch}, {Adamo}, {Coe}, {Diego}, {Mahler}, {Khullar}, {Kokorev}, {Oguri}, {Ravindranath}, {Furtak}, {Hsiao}, {Abdurro'uf}, {Mandelker}, {Brammer}, {Bradley}, {Brada{\v{c}}}, {Conselice}, {Dayal}, {Nonino}, {Andrade-Santos}, {Windhorst}, {Pirzkal}, {Sharon}, {de Mink}, {Fujimoto}, {Zitrin}, {Eldridge}, \& {Norman}}]{Vanzella23}
{Vanzella}, E., {Claeyssens}, A., {Welch}, B., {et~al.} 2023, \apj, 945, 53
\bibitem[{{Verhamme} {et~al.}(2015){Verhamme}, {Orlitov{\'a}}, {Schaerer}, \& {Hayes}}]{Verhamme15_vsep}
{Verhamme}, A., {Orlitov{\'a}}, I., {Schaerer}, D., \& {Hayes}, M. 2015, \aap, 578, A7
\bibitem[{{Williams} {et~al.}(2023){Williams}, {Kelly}, {Chen}, {Brammer}, {Zitrin}, {Treu}, {Scarlata}, {Koekemoer}, {Oguri}, {Lin}, {Diego}, {Nonino}, {Hjorth}, {Langeroodi}, {Broadhurst}, {Rogers}, {Perez-Fournon}, {Foley}, {Jha}, {Filippenko}, {Strolger}, {Pierel}, {Poidevin}, \& {Yang}}]{Williams23}
{Williams}, H., {Kelly}, P.~L., {Chen}, W., {et~al.} 2023, Science, 380, 416
\bibitem[{{Wood} \& {Churchwell}(1989)}]{Wood_UCHII_radio}
{Wood}, D. O.~S. \& {Churchwell}, E. 1989, \apjs, 69, 831
\bibitem[{{Xu} {et~al.}(2023){Xu}, {Henry}, {Heckman}, {Chisholm}, {Marques-Chaves}, {Leclercq}, {Berg}, {Jaskot}, {Schaerer}, {Worseck}, {Amor{\'\i}n}, {Atek}, {Hayes}, {Ji}, {{\"O}stlin}, {Saldana-Lopez}, \& {Thuan}}]{Xu23}
{Xu}, X., {Henry}, A., {Heckman}, T., {et~al.} 2023, \apj, 943, 94
\bibitem[{{Yang} {et~al.}(2021){Yang}, {Urquhart}, {Thompson}, {Menten}, {Wyrowski}, {Brunthaler}, {Tian}, {Rugel}, {Yang}, {Yao}, \& {Mutale}}]{Yang21_HCHII_radio-sed}
{Yang}, A.~Y., {Urquhart}, J.~S., {Thompson}, M.~A., {et~al.} 2021, \aap, 645, A110
\bibitem[{{Yang} {et~al.}(2017){Yang}, {Malhotra}, {Rhoads}, \& {Wang}}]{yang2017}
{Yang}, H., {Malhotra}, S., {Rhoads}, J.~E., \& {Wang}, J. 2017, \apj, 847, 38
\end{thebibliography}