Heavy elements unveil the non primordial origin of the giant HI ring in Leo
Edvige Corbelli, Giovanni Cresci, Filippo Mannucci, David Thilker, Giacomo Venturi
DDraft version January 27, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Heavy elements unveil the non primordial origin of the giant HI ring in Leo
Edvige Corbelli, Giovanni Cresci, Filippo Mannucci, David Thilker, and Giacomo Venturi
3, 1 INAF-Osservatorio di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, MD, USA Instituto de Astrof´ısica, Facultad de F´ısica, Pontificia Universidad Cat´olica de Chile, Casilla 306, Santiago 22, Chile (Accepted by ApJ Letters)
ABSTRACTThe origin and fate of the most extended extragalactic neutral cloud known in the Local Universe,the Leo ring, is still debated 38 years after its discovery. Its existence is alternatively attributed toleftover primordial gas with some low level of metal pollution versus enriched gas stripped during agalaxy-galaxy encounter. Taking advantage of MUSE (Multi Unit Spectroscopic Explorer) operatingat the VLT, we performed optical integral field spectroscopy of 3 HI clumps in the Leo ring whereultraviolet continuum emission has been found. We detected, for the first time, ionized hydrogen inthe ring and identify 4 nebular regions powered by massive stars. These nebulae show several metallines ([OIII], [NII], [SII]) which allowed reliable measures of metallicities, found to be close to or abovethe solar value (0.8 ≤ Z/Z (cid:12) ≤ Keywords:
Galaxy groups ; Intergalactic clouds ; HII regions ; Chemical abundances INTRODUCTIONThe serendipitous discovery of an optically dark HIcloud in the M 96 galaxy group (Schneider et al. 1983),part of the Leo I group, has since then triggered a lot ofdiscussion on the origin and survival of the most mas-sive and extended intergalactic neutral cloud known inthe local Universe ( D ≤
20 Mpc). With an extension ofabout 200 kpc and an HI mass M HI (cid:39) × M (cid:12) , thecloud has a ring-like shape orbiting the galaxies M 105and NGC 3384 (Schneider 1985), and it is known alsoas the Leo ring. As opposed to tidal streams, the mainbody of the Leo ring is isolated, more distant than 3 op-tical radii from any luminous galaxy. The ring is muchlarger than any known ring galaxy (Ghosh & Mapelli2008). The collisional ring of NGC 5291 ( D (cid:39)
50 Mpc)(Longmore et al. 1979; Boquien et al. 2007), of similar
Corresponding author: Edvige [email protected] extent, is vigorously forming stars, as many other colli-sional rings. The Leo ring is much more quiescent andfor many years since its discovery has been detected onlyvia HI emission. Lacking a pervasive optical counterpart(Pierce & Tully 1985; Kibblewhite et al. 1985; Watkinset al. 2014) it has been proposed as a candidate primor-dial cloud (Schneider et al. 1989; Sil’chenko et al. 2003)dating to the time of the Leo I group formation.The bulk of the HI gas in the ring is on the south andwest side, especially between M 96 (to the south) andNGC 3384/M 105 (at the ring center, see Figure 1). In-termediate resolution VLA maps of this region with aneffective beam of 45 (cid:48)(cid:48) revealed the presence of gas clumps(Schneider et al. 1986), some of which appear as distinctvirialized entities and have masses up to 3.5 × M (cid:12) .The position angle of the clump major axes and their ve-locity field suggest some internal rotation with a possibledisk-like geometry and gas densities similar to those ofthe interstellar medium. Distinct cloudlets are found inthe extension pointing south, towards M 96. Detectionof GALEX UV-continuum light in the direction of a few a r X i v : . [ a s t r o - ph . GA ] J a n Corbelli et al.
HI clumps of the ring, suggested star formation activitybetween 0.1 and 1 Gyr ago (Thilker et al. 2009). How-ever, most of the gas mass is not forming massive starstoday since there has been no confirmed diffuse H α emis-sion (Reynolds et al. 1986; Donahue et al. 1995) or COdetection from a pervasive population of giant molecularcomplexes (Schneider et al. 1989).A low level of metal enrichment, inferred fromGALEX-UV and optical colors, favoured the primordialorigin hypothesis. This was supported a few years laterby the detection of weak metal absorption lines in thespectra of 3 background QSOs, 2 of which have sight-lines close or within low HI column density contours ofthe ring (Rosenberg et al. 2014). The low metallicity,estimated between 2% - 16% solar for Si/H, C/H andN/H, has however large uncertainties due to ionisationcorrections. Confusion with emission from the MilkyWay in the QSO’s spectra does not allow to measure HIcolumn densities along the sightlines. This is inferredfrom large scale HI gas maps, and gas substructures onsmall scales can alter the estimated abundance ratios.The Leo ring has also been considered as a possi-ble product of a gas-sweeping head-on collision (Rood& Williams 1985), involving group members such asNGC 3384 and M 96 (Michel-Dansac et al. 2010) or alow surface brightness galaxy colliding with the group(Bekki et al. 2005). A tentative detection of dust emis-sion at 8 µ m in one HI clump (Clump1) (Bot et al. 2009)also supports the pre-enrichment scenario. A direct andreliable measurement of high metallicity gas associatedto the very weak stellar counterpart can give the con-clusive signature of a ring made of pre-enriched gas.In this Letter we present the first detection of nebularregions in the Leo ring. In Section 2 we describe inte-gral field optical spectroscopy of 3 fields in the ring andestimate metal abundances from emission lines in starforming regions. The local metal production and theimplications for the origin of the Leo ring are discussedin the last Section. In a companion paper (Corbelli et al.2021)(hereafter Paper II) we analyse star formation andthe stellar population in and around the detected nebu-lae using GALEX and HST images. THE DISCOVERY OF NEBULAR REGIONSAND THEIR CHEMICAL ABUNDANCESWe assume a distance to the Leo ring of 10 Mpc, as forM 96 and M 105. This implies that an angular separationof 1 (cid:48)(cid:48) corresponds to a spatial scale of 48.5 pc.2.1.
The data
Between December 2019 and March 2020 we have ob-served three 1 × regions in the Leo ring using the integral field spectrograph MUSE (Multi Unit Spec-troscopic Explorer) mounted on the ESO Very LargeTelescope (VLT). The locations of MUSE fields areshown in red in the left panel of Figure 1 overlaid on theSDSS optical image of the M 96 group and on the VLAHI contours of the ring. The fields have been centeredat 3 HI peak locations, Clump1, Clump2 and Clump2E,two in the main body of the ring and one in the filamentconnecting the ring to M 96. They cover completely theultraviolet-bright regions of Clump1 and Clump2E listedby Thilker et al. (2009). The southernmost side of theUV emission in Clump2 is at the border of the MUSEfield.The final cube for each region is the result of two ob-serving blocks, one totalling 960 s and the other 1920 s.The observing blocks are a combination of two and four480 s exposures, respectively, which were rotated anddithered from each other in order to provide a uniformcoverage of the field and to limit systematics. Dedi-cated offset sky exposure of 100 s each were acquiredevery two object exposures. The reduction of the rawdata was performed with the ESO MUSE pipeline (Weil-bacher et al. 2020), which includes the standard proce-dures of bias subtraction, flat fielding, wavelength cal-ibration, flux calibration, sky subtraction and the finalcube reconstruction by the spatial arrangement of theindividual slits of the image slicers. For Clump1 we didnot employ the dedicated sky observations for the skysubtraction, since these were giving strong sky residu-als, especially around the H α line. We thus extractedthe sky spectrum to be subtracted from within the sci-ence cube, by selecting the portions of the FoV free ofsource emission. This allowed to remove the problem-atic sky residuals because the sky spectrum obtainedfrom within the science FoV is simultaneous with thescience spectra. The final dataset comprises 3 datacubes, one per clump, covering a FoV slightly largerthan 1 arcmin . Each spectrum spans the wavelengthrange 4600 - 9350 ˚A, with a typical spectral resolutionbetween 1770 at 4800 ˚A and 3590 at 9300 ˚A. The spatialresolution given by the seeing is of the order of 1 (cid:48)(cid:48) .2.2. HII regions in the ring
We analyse spectral data at the observed spectral andspatial resolution searching for H α emission at the ve-locities of the HI gas in the ring i.e. between 860 and1060 km s − . We detect hydrogen and some collisionallyexcited metal lines in three distinct regions of Clump1(C1a, C1b, C1c) and in two regions of Clump2E (C2Ea,C2Eb). Figure 1 shows the GALEX-FUV continuumand the H α emission in the three MUSE fields. TheFUV emission in Clump2E seems more extended than eavy elements in Leo ring Figure 1. In left panel the HI contours of the Leo ring are overlayed to the optical image of the M96 group (SDSS color image).In magenta the Arecibo contour at N HI = 2 × cm − , in yellow the VLA HI contours of the southern part of the ringSchneider et al. (1986). Crosses indicates background QSOs (Rosenberg et al. 2014), red squares the HI clumps observed withMUSE: Clump1(C1), Clump2(C2) and Clump2E(C2E). The 1 arcmin angular size of the MUSE field is shown in the bottomleft corner. In the right panels the MUSE H α images of Clump1 and of Clump2E show the 5 nebular regions detected. Thecorresponding GALEX-FUV continuum emission is displayed to the left of the H α images. The southernmost FUV source inthe field of Clump1 is a background galaxy. the HII region in H α and suggests a non coeval popu-lation or the presence of some low mass stellar clusterlacking massive stars. No nebular lines are detected inthe field covering Clump2. This clump is the reddestof the three clumps observed, having the largest valuesof UV and optical colours (Thilker et al. 2009; Watkinset al. 2014).The four regions listed in Table 1 are HII regions as-sociated with recent star formation events according totheir line ratios (Kauffmann et al. 2003; Sanders et al.2012) and to the underlying stellar population (see Pa-per II). The data relative to the faintest nebula detected, C1c, is presented and discussed in Paper II because emis-sion line ratios and [OIII]5007 luminosity are consistentwith the object being a Planetary Nebula whose metal-licity is unconstrained due to undetected lines. We givein Table 1 the central coordinates of the HII regions andthe mean recession velocities of identified optical lines.These are consistent with the 21-cm line velocities of theHI gas (Schneider et al. 1986). We fit Gaussian profilesto emission lines whose peaks are well above 3 σ in circu-lar apertures with radius 1.2 (cid:48)(cid:48) , comparable to the seeing.With these apertures we sample more than one third ofthe region total H α luminosity and achieve good signal- Corbelli et al.
Table 1.
HII region coordinates, chemical abundance and extinction. Extinction corrected total H α luminosities are computedusing circular apertures with radius R maxap .Source RA DEC V hel A V Z/Z (cid:12) R maxap A RmaxHα log L Hα km s − mag arcsec mag erg s − C1a 10:47:47.93 12:11:31.9 994 ± +0 . − . +0 . − . ± +0 . − . +1 . − . ± +0 . − . +0 . − . ±
21 8.82 +0 . − . .... 1.35 3.0 .... 35.85 Table 2.
Integrated emission for Gaussian fits to nebular lines with R ap =1.2 (cid:48)(cid:48) . Upper limits are 3 σ values, flux units are10 − erg s − cm − .Source H β [OIII]5007 [NII]6548 H α [NII]6583 [SII]6716/ [SII]6731 FWHM b,r [˚A]C1a 1.89 ± ± < .
46 7.89 ± ± ± ± ± < . < .
49 3.17 ± ± < . < .
40 1.9,2.2C2Ea 7.97 ± .
41 1.34 ± ± .
31 26.57 ± ± ± ± < . < . < .
76 2.25 ± ± < . < .
34 ...,2.1 to-noise (S/N > σ values fornon-detected lines, inferred using the rms of the spectraat the expected wavelength and a typical full spectralextent of the line. For the brightest HII regions we de-tected strong metal lines, such as [O iii ]5007, [N ii ]6583,[S ii ]6716,6731 which can be used to compute reliablemetallicities. 2.3. Chemical abundances
For the four HII regions in Table 1 we compute thegas-phase metal abundances using the strong-line cali-bration in Curti et al. (2020). All the available emis-sion lines and the upper limits to the undetected linesare used to measure metallicity and dust extinction ina two-parameter minimization routine which also esti-mates the uncertainties on these two parameters. Theresulting metal abundances are displayed in Figure 2and in Table 1. In Figure 2 we show the 1- σ confi-dence levels in the oxygen abundance-visual extinctionplane and the best fitting values of chemical abundancesalong the calibration curves for the strong line ratios.Line ratios for all the HII regions are well-reproducedby close to solar metallicities and moderate visual dustextinctions. For C2Eb extinction cannot be constrained.Metallicities in Clump1 are slightly below solar, those inClump 2E are above solar. The HII regions in Clump 1have lower SII/H α line ratios than predicted by Curtiet al. (2020). This is also found in outer disk HII regions (Vilchez & Esteban 1996) and it is likely due to a highionisation parameter driven by a low density interstel-lar medium (Dopita et al. 2013). The mass fraction ofmetals with respect to solar, Z/Z (cid:12) , ranges between 0.79and 1.41 (assuming solar distribution of heavy elementsand Z (cid:12) =0.0142 (Asplund et al. 2009)).Using wide apertures, as listed in column (8) of Ta-ble 1 and chosen to include most of H α emission withno overlap, we derive the HII region total H α luminosi-ties, L Hα . These are given in column (10) already cor-rected for extinction when this can be estimated fromthe Balmer decrement in these apertures (column (9)).Luminosities are high enough to require the presence ofvery massive and young stars, especially for C1a andC2Ea. The local production rate of ionizing photonsby hot stars might be higher than what can be inferredusing L Hα if some photons leak out or are directly ab-sorbed by dust in the nebula.The HII regions in the [OIII]/H β versus [NII]/H α plane, known as the BPT diagram (Baldwin et al. 1981),are consistent with data from young HII regions in galax-ies (Kauffmann et al. 2003; Sanders et al. 2012) andtheir metallicities are in agreement with those predictedby photoionisation models of HII regions (Dopita et al.2013; Byler et al. 2017). Line ratios observed in Clump1are also consistent with the distribution of the recentevolution models of HII regions in gas clouds (Pellegriniet al. 2020), available only for solar metallicity. Thesepredict an age of about 5 Myrs for C1a. Line ratios forC2Ea instead fall outside the area where solar metallic-ity HII regions are found, in agreement with the higherthan solar metallicity we infer for this clump. A very eavy elements in Leo ring A V C1aC1bC2EaC2Eb 12+log(O/H) l og ( H / H ) -1.5-1.0-0.5 l og ([ N II] / H ) -1.2-1.0-0.8-0.6-0.4 l og ([ S II] + / H ) -1.0-0.50.00.5 l og ([ O III] / H ) Figure 2.
Dust visual extinction and gas-phase metallicityfor each HII region, color-coded as in the legend. In the toppanel we show the best fitting values (cross) and 1- σ confi-dence levels of A V ; the dotted line shows the solar metallicity(12+log(O/H)=8.69). The four bottom panels refer to rele-vant strong-line ratios. Diamonds show the observed valuesof the line ratio plotted at the best-fitting value of metallic-ity. The H α /H β ratio is computed for Case B recombinationwith uncertainties due to the unknown temperature, and cir-cles showing extinction-corrected values. The solid curves inthe lower three panels trace the calibrations from Curti et al.(2020), with the relative uncertainties. young age is recovered in Paper II for this HII regionthrough a multiwavelength analysis. IN SITU METAL ENRICHMENT AND THERING ORIGINWe compute the maximum mass fraction of metalswhich could conceivably be produced in situ, f maxZ ,given the observed metal abundances, Z obs , and the lim-iting blue magnitudes of the Leo ring, µ B . For this ex-treme local enrichment scenario we assume that all starshave formed in the ring and use the instantaneous burstor continuous star formation models of Starburst99 (Lei-therer et al. 1999) in addition to population synthesismodels of Bruzual & Charlot (2003) for an initial burstwith an exponential decay ( τ = 1 Gyr). At each timestep we compute the B − V color and the maximumstellar surface mass density which corresponds to thelimiting values of µ B . This stellar density gives themaximum mass fraction of metals produced locally. Inorder to maximise the local metal production we con-sider a closed box model with no inflows or outflows forwhich a simple equation relates the stellar yields to theincrease in metallicity since star formation has switchedon (Searle & Sargent 1972): f maxZ = ZZ obs = y Z Z obs ln( Σ g Σ g ) = y Z Z obs ln(1 + Σ ∗ Σ g ) (1)where Z and Σ ∗ are the abundance of metals by massand the stellar mass surface density produced in situ.The gas mass surface density at the present time andat the time of the Leo ring formation are Σ g and Σ g respectively. The total net yields y Z refers to the massof all heavy elements produced and injected into theinterstellar medium by a stellar population to the rateof mass locked up into low mass stars and stellar rem-nants. There are several factors that can affects theyields: the upper end of the Initial Mass Function (here-after IMF), massive star evolution and ejecta models,metallicity. Since the pioneer work of Searle & Sar-gent (1972) several papers have analysed these depen-dencies (e.g. Maeder 1992; Meynet & Maeder 2002; Ro-mano et al. 2010; Vincenzo et al. 2016). Following theresults of Romano et al. (2010); Vincenzo et al. (2016)we consider negligible the metallicity dependence on theyields and consider the Chabrier IMF i.e. an IMF witha Salpeter slope from 1 M (cid:12) up to its high mass endat 100 M (cid:12) and a Chabrier-lognormal slope from 0.1 to1 M (cid:12) (Salpeter 1955; Chabrier 2003). This IMF hasa total yield y Z =0.06, the highest amongst commonlyconsidered IMF (Vincenzo et al. 2016).To maximize the associated fraction of metals pro-duced locally, f maxZ , we consider zero extinction and Corbelli et al. the best fitted metallicities for C1a and C2Ea minus3 times their dispersion, i.e. Z obs =0.6 and 1.32 Z (cid:12) for Clump1 and Clump2E respectively. A very largefraction of the HI rich ring area corresponding to theVLA coverage of Schneider et al. (1986) has been sur-veyed deeply in the optical B band (Pierce & Tully1985; Watkins et al. 2014). For a very diffuse perva-sive population throughout the Leo ring the survey re-sults give µ B ≥
30 mag arcsec − . For optical emissionin less extended regions as the MUSE fields, or equiva-lently at the VLA HI map spatial resolution (Schnei-der et al. 1986), and following the results of Mihoset al. (2018) we can use the more conservative upperlimit µ B ≥
29 mag arcsec − . Given the optical colors B − V =0.0 ± . B − V =0.1 ± . B − V ≤ B − V ≤ (cid:48)(cid:48) radius is Σ g =3.1 and 0.8 M (cid:12) pc − inClump1 and Clump2E respectively. We compute fromthe models the stellar mass surface density correspond-ing to µ B = 29 mag arcsec − at each time, and f Z withthe above values of Σ g , Z obs and y Z , using equation (1).The value of f maxZ will be f Z at the maximum value of B − V for each clump. In Figure 3 we show f Z for thethree models as a function of time and of B − V . Thedashed line indicates the limiting value of B − V . A dot-ted line has been placed at the value of f maxZ i.e. wherethe limiting colors intersect the models which producesthe highest mass of metals.For Clump1 a starburst 500 Myrs ago that slowly de-cays with time gives the highest possible local metal pro-duction with f maxZ = 3% and Σ ∗ =0.01 M (cid:12) pc − . ForClump2E both an instantaneous burst 500 Myrs ago ora continuous star formation since 2 Gyr ago gives themaximum value of f maxZ = 17% with Σ ∗ =0.04 M (cid:12) pc − .We conclude that the fraction of metals produced locallyis too small to be compatible with a scenario of a pri-mordial metal poor ring enriched in situ. The ring musthave formed out of metal rich gas, with chemical abun-dances above 0.5 Z (cid:12) , mostly polluted while residing ina galaxy and then dispersed into space.We underlines that all models predicts a small fractionof metals produced in situ and that the ones that max-imise f Z are not necessarely the best fitted models to theunderlying stellar population. These will be examined inPaper II. The apparent discrepancy between our resultsand the lower abundances inferred by QSO’s absorptionlines can be resolved if hydrogen column densities alongsightlines to nearby QSOs are lower than those used inthe analysis of Rosenberg et al. (2014) and estimatedfrom HI emission averaged over a large beam. The most Figure 3.
The mass fraction of metals f Z produced in situfor an instantaneous burst (blue lines), a burst with expo-nential decay (red lines), and a continuous star formationmodel (green lines) as a function of optical colors (left pan-els) and time (right panels). Each model for both Clump1(upper panels) and Clump2E (lower panels) is normalised asto produce an apparent magnitude µ B = 29 at any time af-ter star formation switches on. The dashed lines indicate themaximum B − V optical color of the clumps, and the dottedline f maxZ , the highest value of f Z compatible with B − V . discrepant abundance with respect to the nearly solarabundances we infer for the ring is for carbon towardthe southernmost QSO: -1.7 ≤ [C/H]/[C/H] (cid:12) ≤ − .
1. Iffuture measures of the HI column density towards theQSO’s sightline confirm the low metal abundances, thesecan be used to investigate chemical inhomogeneities dueto a mix of metal rich gas with local intragroup metalpoor gas in the ring outskirts.We summarise that our finding has confirmed spec-troscopically the association between stellar complexesdetected in the UV-continuum and the high column den-sity gas (Thilker et al. 2009). The detected H α emissionimplies a sporadic presence of a much younger and mas-sive stellar population then estimated previously (seePaper II for more details). For the first time we havedetected gaseous nebulae in the ring with chemical abun-dances close to or above solar which conflict with theprimordial origin hypothesis of the Leo ring. A sce-nario of pre-enrichment, where the gas has been pollutedby the metals produced in a galaxy and subsequentlytidally stripped and placed in ring-like shape, is in agree-ment with the data presented in this Letter. This pic- eavy elements in Leo ring M L W W Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009,ARA&A, 47, 481,doi: 10.1146/annurev.astro.46.060407.145222Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,PASP, 93, 5, doi: 10.1086/130766Bekki, K., Koribalski, B. S., Ryder, S. D., & Couch, W. J.2005, MNRAS, 357, L21,doi: 10.1111/j.1745-3933.2005.08625.xBot, C., Helou, G., Latter, W. B., et al. 2009, AJ, 138, 452,doi: 10.1088/0004-6256/138/2/452Boquien, M., Duc, P. A., Braine, J., et al. 2007, A&A, 467,93Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000,doi: 10.1046/j.1365-8711.2003.06897.xByler, N., Dalcanton, J. J., Conroy, C., & Johnson, B. D.2017, ApJ, 840, 44, doi: 10.3847/1538-4357/aa6c66Chabrier, G. 2003, PASP, 115, 763, doi: 10.1086/376392Corbelli, E., Mannucci, F., Thilker, D., Cresci, G. &Venturi, G. submitted to A&A(Paper II)Curti, M., Mannucci, F., Cresci, G., & Maiolino, R. 2020,MNRAS, 491, 944, doi: 10.1093/mnras/stz2910Donahue, M., Aldering, G., & Stocke, J. T. 1995, ApJL,450, L45, doi: 10.1086/316771Dopita, M. A., Sutherland, R. S., Nicholls, D. C., Kewley,L. J., & Vogt, F. P. A. 2013, ApJS, 208, 10,doi: 10.1088/0067-0049/208/1/10Ghosh, K. K., & Mapelli, M. 2008, MNRAS, 386, L38,doi: 10.1111/j.1745-3933.2008.00456.xKauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003,MNRAS, 346, 1055,doi: 10.1111/j.1365-2966.2003.07154.x Kibblewhite, E. J., Cawson, M. G. M., Disney, M. J., &Phillipps, S. 1985, MNRAS, 213, 111,doi: 10.1093/mnras/213.2.111Leitherer, C., Schaerer, D., Goldader, J. D., et al. 1999,ApJS, 123, 3, doi: 10.1086/313233Longmore, A. J., Hawarden, T. G., Cannon, R. D., et al.1979, MNRAS, 188, 285Maeder, A. 1992, A&A, 264, 105Meynet, G., & Maeder, A. 2002, A&A, 390, 561,doi: 10.1051/0004-6361:20020755Michel-Dansac, L., Duc, P.-A., Bournaud, F., et al. 2010,ApJL, 717, L143, doi: 10.1088/2041-8205/717/2/L143Mihos, J. C., Carr, C. T., Watkins, A. E., Oosterloo, T., &Harding, P. 2018, ApJL, 863, L7,doi: 10.3847/2041-8213/aad62eOey, M. S., & Kennicutt, R. C., J. 1993, ApJ, 411, 137,doi: 10.1086/172814Pellegrini, E. W., Rahner, D., Reissl, S., et al. 2020,MNRAS, 496, 339, doi: 10.1093/mnras/staa1473Pierce, M. J., & Tully, R. B. 1985, AJ, 90, 450,doi: 10.1086/113750Reynolds, R. J., Magee, K., Roesler, F. L., Scherb, F., &Harlander, J. 1986, ApJL, 309, L9, doi: 10.1086/184750Romano, D., Karakas, A. I., Tosi, M., & Matteucci, F.2010, A&A, 522, A32, doi: 10.1051/0004-6361/201014483Rood, H. J., & Williams, B. A. 1985, ApJ, 288, 535,doi: 10.1086/162819Rosenberg, J. L., Haislmaier, K., Giroux, M. L., Keeney,B. A., & Schneider, S. E. 2014, ApJ, 790, 64,doi: 10.1088/0004-637X/790/1/64Salpeter, E. E. 1955, ApJ, 121, 161, doi: 10.1086/145971