A Census of the Extended Neutral Hydrogen Around 18 MHONGOOSE Galaxies
Amy Sardone, D.J. Pisano, N. M. Pingel, Amidou Sorgho, Claude Carignan, W.J.G. de Blok
DDraft version January 21, 2021
Typeset using L A TEX twocolumn style in AASTeX63
A Census of the Extended Neutral Hydrogen Around 18 MHONGOOSE Galaxies
Amy Sardone ,
1, 2
D.J. Pisano,
3, 4, 5
N. M. Pingel ,
6, 3, 4
A. Sorgho , Claude Carignan ,
7, 8 andW. J. G. de Blok
9, 7, 10 Department of Astronomy, The Ohio State University, 140 West 18th Avenue, Columbus, OH 43210, USA Center for Cosmology and Astroparticle Physics, 191 West Woodruff Avenue, Columbus, OH 43210, USA Department of Physics and Astronomy, West Virginia University, Morgantown, WV 26506 Gravitational Wave and Cosmology Center, Chestnut Ridge Research Building, Morgantown, WV 26505 USA Adjunct Astronomer at Green Bank Observatory, Green Bank, WV Research School of Astronomy & Astrophysics, Australian National University, Canberra, ACT 2611, Australia Department of Astronomy, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa Laboratoire de Physique et de Chimie de l’Environnement, Observatoire d’Astrophysique de l’Universit´e Ouaga I Pr Joseph Ki-Zerbo(ODAUO), 03 BP 7021, Ouaga 03, Burkina Faso Netherlands Institute for Radio Astronomy (ASTRON), Oude Hoogeveensedijk 4, 7991 PD Dwingeloo, the Netherlands Kapteyn Astronomical Institute, University of Groningen, PO Box 800, 9700 AV Groningen, The Netherlands (Accepted January 21, 2021)
Submitted to ApJABSTRACTWe present the analysis of the diffuse, low column density Hi environment of 18 MHONGOOSEgalaxies. We obtained deep observations with the Robert C. Byrd Green Bank Telescope, and reacheddown to a 3 σ column density detection limit of N HI = 6 . × cm − over a 20 km s − linewidth. Weanalyze the environment around these galaxies, with a focus on Hi gas that reaches column densitiesbelow N HI = 10 cm − . We calculate the total amount of Hi gas in and around the galaxies revealingthat nearly all of these galaxies contained excess Hi outside of their disks. We quantify the amountof diffuse gas in the maps of each galaxy, defined by Hi gas with column densities below 10 cm − ,and find a large spread in percentages of diffuse gas. However, by binning the percentage of diffuse Hi into quarters, we find that the bin with the largest number of galaxies is the lowest quartile (0 − Hi ). We identified several galaxies which may be undergoing gas accretion onto the galaxy diskusing multiple methods of analysis, including azimuthally averaging column densities beyond the disk,and identifying structure within our integrated intensity (Moment 0) maps. We measured Hi massoutside the disks of most of our galaxies, with rising cumulative flux even at large radii. We also finda strong correlation between the fraction of diffuse gas in a galaxy and its baryonic mass, and testthis correlation using both Spearman and Pearson correlation coefficients. We see evidence of a darkmatter halo mass threshold of M halo ∼ . M (cid:12) in which galaxies with high fractions of diffuse Hi allreside below. It is in this regime in which cold-mode accretion should dominate. Finally, we suggest arotation velocity of v rot ∼
80 km s − as an upper threshold to find diffuse gas-dominated galaxies. Keywords: galaxies: evolution – galaxies: structure – galaxies: accretion – galaxies: spirals INTRODUCTIONThe question of how galaxies get their gas and howthey use this gas to continue to form stars remainsamong the main unanswered questions in astronomy.
Corresponding author: Amy Sardone; NSF Astronomy and As-trophysics Postdoctoral [email protected]
We know that as the star formation rate density overcosmic time has decreased since a redshift of z ∼ ∼ a r X i v : . [ a s t r o - ph . GA ] J a n Sardone et al.
Gas accreting from the IGM flows into the galaxy viadiffuse filamentary structures (Birnboim & Dekel 2003;Katz et al. 2003; Kereˇs et al. 2005). The gas travelsfrom the cosmic web (Bond et al. 1996) through thesefilaments, and into the circumgalactic medium (CGM)of galaxies before falling onto the galaxy disk. The pro-cess of accretion described here typically proceeds viathe hot or the cold mode (Kereˇs et al. 2005), where gasentering the halo either becomes shock heated to thevirial temperature before cooling and condensing ontothe galaxy disk or it flows along cold filaments throughthe CGM and remains cool as is falls onto the disk.While gas in both scenarios is predominantly ionized,the gas accreting via the cold-mode maintains a smallneutral fraction, making detection feasible. Cold accre-tion dominates in the lower mass galaxy regime and fa-vors low galaxy density environments. Observations ofthe CGM and disks of galaxies encompassing a range ofthese parameters will provide insight into cold accretionfrom the IGM and potentially direct detections of thisgas falling onto galaxies.Deep Hi surveys provide the observational link todirect detections of the effects of cold mode accre-tion from the IGM. The MeerKAT Hi Observations ofNearby Galactic Objects; Observing Southern Emitters(MHONGOOSE ; de Blok et al. 2016) survey will pro-vide high spatial resolution, high column density sen-sitivity maps of 30 nearby disk and dwarf galaxies.Most other Hi surveys achieve one or the other, butnot both, with the exception of the IMAGINE sur-vey with ATCA, which still has a much lower angu-lar resolution than MHONGOOSE. HIPASS (Koribalskiet al. 2004) and ALFALFA (Haynes et al. 2011) surveysprovide an extremely large number of Hi detections, al-though neither at high resolution nor with column den-sity sensitivities matching ours. High resolution surveyssuch as THINGS (Walter et al. 2008) and HALOGAS(Heald et al. 2011) (one of the first high resolution sur-veys designed to systematically detect accretion of Hi ),achieve column density sensitivities of ∼ cm − and ∼ cm − , respectively. These levels of column densi-ties enable detection of clumpy Hi that could be missedby a single dish telescope, yet would not be low enoughto detect any extended or more diffuse gas accretingfrom the IGM. The MHONGOOSE survey will use theSouth African MeerKAT radio telescope, a 64-dish pre-cursor to the Square Kilometre Array (SKA), to mapthese 30 galaxies to a 3 σ column density detection limitof 7 . × cm − over a 16 km s − linewidth at an an-gular resolution of 30 (cid:48)(cid:48) . At their poorest angular res-olution, 90 (cid:48)(cid:48) , this column density limit goes down to5 . × cm − . The high resolution maps produced by https://mhongoose.astron.nl MHONGOOSE will further our understanding of howgalaxies get their gas, how galaxies sustain star forma-tion, and how matter that we can detect relates to thedark matter associated with galaxies, influencing galac-tic evolution. Specifically, this survey will help us tounderstand how gas flows in or out of galaxies, the condi-tions that allow the fuelling of star formation, accretionfrom the IGM, and ultimately its connection to the cos-mic web (e.g. Carignan 2016). The first of these dataare presented in de Blok et al. (2020), detailing theirdetection of low column density Hi clouds as well as afilament extending off of ESO 302-G014, which could bethe result of a minor interaction with a dwarf galaxy.Early observations of a subset of MHONGOOSEgalaxies were presented in Sorgho et al. (2019, hereafterS19), using MeerKAT, KAT-7, the seven-dish MeerKATprecursor array, and the Robert C. Byrd Green BankTelescope (GBT). The subset of galaxies mapped withthe GBT are presented here, with the description of theMHONGOOSE sample in Section 2. We describe ourGBT observations and data reduction process in Sec-tion 3. Two galaxies not presented in S19, NGC 1744and NGC 7424, are presented in Section 4. The anal-ysis we performed on all of these galaxies is describedin Section 5. Our results are presented in Section 6.We discuss these results in Section 7 and summarize ourfindings in Section 8. MHONGOOSE GALAXY SAMPLEThe MHONGOOSE sample of galaxies was chosenfrom galaxies having been previously detected in mul-tiple wavelengths ( Hi from HIPASS, H α , optical, in-frared, and ultraviolet from the SINGG and SUNGG(Meurer et al. 2006) surveys). MHONGOOSE galax-ies were primarily selected to cover a large range of Hi masses. A secondary parameter used in the selection,the star formation rate (SFR), enables the separation of Hi coincident with the star forming disk from extrapla-nar Hi (e.g. Marasco et al. (2019)). An equal number ofgalaxies were selected for each of six mass bins within arange of 6 < log (M HI ) <
11 M (cid:12) for a total of 30 galax-ies with edge-on, face-on, and intermediate inclinations,and a wide range of morphologies from dwarf irregularsto grand-design spirals. The sample of galaxies chosenfor this set of observations coincides with the galaxiesfrom the MHONGOOSE sample ( δ < ◦ ) which canbe seen by the GBT, which has a lower limit of declina-tions δ > − ◦ . This narrows the sample to 18 galax-ies. Our GBT-MHONGOOSE sample sacrifices someof the uniformity of the original sample, while continu-ing to span the full range in Hi masses. These galaxiessimilarly span a wide range of stellar masses falling be-tween 6 . < log (M (cid:63) ) < . (cid:12) (Leroy et al. 2019).GBT integrated intensity (moment 0) maps and global Hi profiles for 16 of these sources were presented in S19,with two others, NGC 1744 and NGC 7424, having beenobserved separately and introduced here. BT-MHONGOOSE log ( M HI / M )012345 N u m b e r o f S o u r c e s GBT-MHONGOOSE MHONGOOSE (MeerKAT)
Figure 1.
Histogram of Hi masses in the MHONGOOSEsample. Each bin is separated by color and hatch mark-ing. The hatch markings (both directions) identify the GBT-MHONGOOSE sample and the gray solid represents thefull MHONGOOSE with MeerKAT sample. This histogramdemonstrates the coverage of the Hi mass range in the bothMHONGOOSE samples. Note that the lowest and highestmass bins are larger due to the scarcity of galaxies meetingthe survey criteria in those mass ranges. Each galaxy in the sample presented here, as well asthe full MHONGOOSE sample, has a Galactic latitudeof | b | > ◦ , peak Hi fluxes, as detected in HIPASS, ofgreater than 50 mJy, and Galactic standard of rest ve-locities >
200 km s − . Additionally, each source is belowa declination of δ < ◦ and is within a distance of 30Mpc. The MHONGOOSE sample was chosen to uni-formly cover the range of M HI listed above, and ourGBT sample covers that range nearly uniformly as well.Coverage of the Hi mass range in each bin can be seenin Figure 1. We note that both the lowest and highest Hi mass bins cover a larger mass range than the oth-ers due to the scarcity of galaxies in those mass rangeswhich also meet the survey criteria. Further details onthe MHONGOOSE sample selection can be found in deBlok et al. (2016). OBSERVATIONS AND DATA REDUCTIONObservations for the 16 galaxies presented in S19 werecarried out between August 2016 and January 2017 forproject GBT16B-212. Each of these galaxies were ob-served for 10 hours each, a total of 160 hours, with atheoretical brightness temperature noise sensitivity of ∼
13 mK over 5 . − channels. This noise sen-sitivity corresponds to a column density sensitivity of N HI ∼ cm − over the same channel width. Thetwo additional galaxies presented here (NGC 1744 andNGC 7424) were observed over the course of three GBTobserving semesters (GBT15B-346, GBT16B-408, and GBT17A-478) from 2015 to 2017, for a total of 20 hourson NGC 1744 and 14 hours on NGC 7424. We observedthese two galaxies over a bandwidth of 23.4 MHz, with afrequency resolution of 0.715 kHz. We used the sources3C48 and 3C147, which have stable, well-understoodfluxes, over the course of 12 observing nights to calibrateour data. We smoothed the data using a boxcar functionwith a final smoothed velocity resolution of 5 . − ,or 24.3 kHz. The remainder of the data reduction fol-lowed the same approach presented in S19.For each of our 18 galaxies, we used the GBT’sVErsatile GBT Astronomical Spectrometer (VEGAS)backend in L-band (1.15-1.73 GHz), where the FWHMbeamwidth is 9 . (cid:48) , to map 2 ◦ × ◦ regions around eachsource. Data cubes were made for all 18 sources withpixel sizes of 1 . (cid:48) . The 16 sources from S19 weresmoothed to a velocity resolution of 6 . − .We reached rms noise levels of 6 to 20 mK (3.2 - 10.4mJy) in the cubes, corresponding to a 1 σ column den-sity sensitivity of N HI = 1 . × cm − per 5 . − channel. At the noise level reached in this cube, a de-tectable signal at the 5 σ level over a 20 km s − linewidthis N HI = 2 × cm − . Our 1 σ mass sensitivities extenddown to 1 . × M (cid:12) . PROPERTIES OF GALAXIESData cubes and integrated intensity (Moment 0) mapsfor the 16 sources mentioned above were inspected andsearched for anomalous Hi in S19. In this section wewill present the two additional sources, NGC 1744 andNGC 7424, and discuss their properties. These proper-ties are listed in Table 2.4.1. NGC 1744
NGC 1744 is an inclined (69 . ◦ ) SBcd galaxy. It isone of the more massive galaxies in our sample, whichwe measured the total Hi mass to be 7 . × M (cid:12) ( ± . × ). We measured a total integrated flux of174.7 Jy km/s with an rms noise of σ rms = 3 . . × M (cid:12) . These and other properties of NGC 1744 were tab-ulated in Table 2 and discussed further in Section 5.We detected one additional source in the NGC 1744data cube. Sbc galaxy, ESO 486-G021 (Figure 3), isdetected with an integrated Hi flux density of 24.5 Jykm s − . The systemic velocities measured from the twogalaxies differ by ∼
100 km s − , and while this is lowenough to investigate potential interaction, we find theirphysical separation to be too great to make this likely.The angular separation is 58 (cid:48) , and at a distance of 12Mpc, that becomes a physical separation of ∼
200 kpc. http://leda.univ-lyon1.fr Sardone et al.
Figure 2.
Left.
Integrated intensity (moment 0) map of NGC 1744 with companion, ESO486-21 in upper left. Contours at 10,20, and 40 times the integrated rms noise of 0.15 K. The column density equivalent to these contour levels is N HI = 0.69, 1.3,2.7, and 5 . × cm − . Right.
Integrated intensity (moment 0) map of NGC 7424. Contours at 5, 10, 20, 40, 80, and 160times the integrated rms noise of 0.16 K. The column density equivalent to these contour levels is N HI = 0.7, 1.48, 2.9, 5.9, and11 . × cm − . We calculated the Jacobi radius, the maximum radiusexpected of a central galaxy and companion galaxy sys-tem, with r J = R Sep (cid:18) M Comp M Central (cid:19) / . (1)Here, R Sep is the physical separation between the cen-ters of the two galaxies, M Comp is the dynamical massof the companion galaxy, and M Central is the dynamicalmass of the central galaxy. This gives us a maximumexpected radius of ∼
79 kpc using the dynamical mass,derived from the Hi rotation velocity, of the companionof 2 . × M (cid:12) . It is evident that the 200 kpc projectedseparation is too large to consider plausible interaction.Further, we do not detect any extraplanar or anomalous Hi around NGC 1744 at these levels. A total Hi intensitymap including both NGC 1744 and ESO 486-G021 canbe seen in Figure 2. Spectra of NGC 1744 can be seen inFigure 3. At a mass sensitivity in this cube of 7 . × M (cid:12) , and because we did not detect any anomalous Hi ,we can say that apart from ESO 486-G021 there are noadditional sources with 3 σ Hi mass greater than 7 . × M (cid:12) . 4.2. NGC 7424
NGC 7424 is a mostly face-on SABcd galaxy (de Vau-couleurs et al. 1991), and one of the most Hi massivesources in the survey. It falls into the mass bin: 10 < log (M HI ) <
11 M (cid:12) , one of the more unexplored massregimes in low column density Hi studies. We mea-sured a total Hi mass of 2 . × M (cid:12) ( ± . × )in NGC 7424, where the mass sensitivity in the cube is2 . × M (cid:12) corresponding to a 1 σ column density sen-sitivity of N HI = 9 . × cm − per 5 . − channel.NGC 7424’s global Hi profile and moment map can beseen in Figure 3.We detected two additional galaxies within our mapof NGC 7424: the Sd galaxy ESO 346-G018 with an an-gular separation of 58 (cid:48) , and the SBbc galaxy NGC 7462with an angular separation from NGC 7424 of 64 (cid:48) . How-ever, due to their positions at the edges of our cube, wecan not confidently measure the total Hi . Instead, wecan provide the following lower limits. We measuredan integrated flux of 8.8 Jy km/s for ESO 346-G018,with a linewidth at 20% maximum of 176 km/s arounda systemic velocity of 1875 km/s. NGC 7462 has an inte-grated flux lower limit of 25.0 Jy km/s, a 20% maximumlinewidth of 205 km/s, and a systemic velocity of 1066km/s. The total integrated profiles were created for eachgalaxy and can be seen in Figure 3. No other significantdetections of Hi were seen throughout the cube. ANALYSISEach source in this survey was analyzed by extractinginformation from the data cubes and deriving individualgalaxy properties, followed by creating both masked and
BT-MHONGOOSE
400 600 800 1000 1200 1400
Velocity [km/s] I n t e n s i t y [ K ] NGC1744
400 600 800 1000 1200 1400
Velocity [km/s] I n t e n s i t y [ K ] NGC7424
600 800 1000 1200
Velocity [km/s] I n t e n s i t y [ K ] ESO486-21
400 600 800 1000 1200 1400
Velocity [km/s] I n t e n s i t y [ K ] NGC7462
Velocity [km/s] I n t e n s i t y [ K ] ESO346-18
Figure 3.
Total integrated Hi profiles. Top row: NGC 1744 (left) and NGC 7424 (right). Bottom row: ESO486-21 detectedin the NGC 1744 cube (left), and detected in the NGC 7424 cube is NGC7462 (center) and ESO346-18 (right). unmasked total integrated intensity (moment 0) maps,and measuring the global properties of the galaxy envi-ronments from these images. Moment 0 maps for eachsource can be found in S19. In doing this, we can iden-tify the statistical properties over a wide population ofgalaxies. 5.1. Total integrated Hi flux profiles We created total integrated Hi flux profiles for eachgalaxy using the MIRIAD task mbspect . mbspect makes a measurement of the data cube over a widevelocity range, where channels including emission aremasked, and a first-order polynomial is fitted to thebaseline, removing residual baseline variation. We ob-tain the 1 σ rms noise over emission-free channel ranges, as well as the total integrated flux, and the linewidths ofeach source at 20% (W20) and 50% (W50) of the peakflux value. These measured values are listed in Table 2.Each galaxy’s global line profile can be seen in Figures7 through 23.5.2. Derived galaxy properties
For each data cube, we calculate the per channel 1 σ column density sensitivity. We use measurements from mbspect to derive physical properties of each galaxy inthe sample, such as the total Hi masses, total dynamicalmasses, and neutral gas fractions.Column densities are calculated with N HI = 1 . × (cid:18) T B K (cid:19) (cid:18) dv km s − (cid:19) cm − , (2) Sardone et al. where T B is the brightness temperature in units ofKelvin, and dv is the resolution of our data in km s − .When estimating the detectable column density level inthe cube, we use a 5 σ limit and a minimum linewidth ofa typical Hi detection (20 km s − ).We integrate over the total flux values to estimate the Hi mass of each source, assuming the Hi is optically thin,using: M Hi = 2 . × D (cid:90) v v S ( v ) dv M (cid:12) . (3)Here, the distance, D , is in Mpc, and (cid:82) v v S ( v ) dv isthe total integrated flux over velocities enclosing the Hi profile v to v , and is in units of Jy km s − . Thebrightness temperatures from each cube were divided bythe 1 .
86 K Jy − gain of the GBT to obtain units of Jy.We calculated this gain using an aperture efficiency of ∼ .
65 (Boothroyd et al. 2011) for the GBT at 1420MHz.In order to calculate the total dynamical mass ofeach galaxy, we first calculated the physical Hi diameter, D Hi , using the Broeils & Rhee (1997) scaling relation(Eq. 2.5), which is dependent on the optical diameterof the galaxy, D , measured at the 25 th mag arcsec − isophote in the B-band. We used values for D fromthe Lyon-Meudon Extragalactic Database (LEDA). Weuse the dynamical mass equation: M dyn = 2 . × (cid:18) v rot /sin ( i )km s − (cid:19) (cid:18) r kpc (cid:19) M (cid:12) , (4)where v rot is the rotation velocity in km s − , and i theinclination, also taken from LEDA. These properties ob-tained from NED and LEDA can be seen in Table 1. Weuse our linewidth at 20% maximum to estimate v rot as v rot = W /
2. Dynamical masses are calculated insidethe Hi radius, r in kpc, calculated as D Hi / Hi fraction. This is defined as thefraction of Hi mass in the galaxy and is calculated with f Hi = M HI / M dyn . Each of these derived propertiesare listed in Table 2 for every galaxy in the sample.5.3. GBT beam model
One of the main science goals of this survey is to de-tect low column density Hi around our sample of galax-ies. In order to distinguish between low column density Hi from an extragalactic source and the low level radia-tion entering the sidelobes of the GBT’s main beam, weuse a beam model which measures these sidelobe levelsprecisely, modeling the beam response of an unresolvedpoint source. We adopt the beam model used by Pin-gel et al. (2018). Without this model, emission enteringthe nearest sidelobes from the main beam could be con-fused for low column density emission from our sources.For each source in our sample, we used the GBT beam Table 1.
Sample of MHONGOOSE Galaxies
Source R.A.(J2000) Decl.(J2000) D incl. D log M (cid:63) [h:m:s] [ ◦ : (cid:48) : (cid:48)(cid:48) ] [Mpc] [deg] (cid:48) [M (cid:12) ](1) (2) (3) (4) (5) (6) (7)ESO300G014 03:09:37 -41:01:50 12.9 61.2 4.47 8.72ESO300G016 03:10:10 -40:00:11 9.3 35.6 0.78 ...ESO302G014 03:51:40 -38:27:08 11.7 27.6 1.35 7.79ESO357-G007 03:10:24 -33:09:22 17.8 72.0 1.29 8.31KK98-195 13:21:08 -31:31:45 5.2 55.7 0.45 ...KKS2000-23 11:06:12 -14:24:26 12.7 90.0 0.6 ...NGC1371 03:35:01 -24:56:00 20.4 47.5 4.9 10.7NGC1592 04:29:40 -27:24:31 13.0 64.4 1.02 8.13NGC1744 04:59:57 -26:01:20 10.0 69.9 5.25 9.19NGC3511 11:03:23 -23:05:12 14.2 72.6 6.03 9.63NGC5068 13:18:54 -21:02:21 6.9 30.1 7.41 9.32NGC5170 13:29:48 -17:57:59 28.0 90.0 7.94 10.72NGC5253 13:39:55 -31:38:24 3.0 70.1 5.01 8.57NGC7424 22:57:18 -41:04:14 13.5 32.4 5.01 9.31UGCA015 00:49:49 -21:00:54 3.3 67.4 1.62 6.74UGCA250 11:53:24 -28:33:11 24.4 90.0 3.63 9.74UGCA307 12:53:57 -12:06:21 8.6 62.0 1.82 7.96UGCA320 13:03:16 -17:25:23 7.7 90.0 6.76 8.11 Note —Positions listed in Columns 2-3 and distances in 4 gath-ered from NED. Inclination angles from face-on in columns 5 andthe optical diameter, D , measured at the 25 th mag arcsec − isophote in the B-band values taken from HyperLEDA:http://leda.univ-lyon1.fr. Stellar masses were obtained fromLeroy et al. (2019). model as a template to regrid each data cube to have4 (cid:48)(cid:48) pixels with size 1024 × MIRIAD tasks imgen and regrid .5.4.
Integrated intensity images (Moment 0)
We created integrated intensity images, or Moment 0images, for each source in our sample. An unmaskedimage was made by integrating pixel values over a chan-nel range in the data cube containing emission from thecentral source. We also created masked Moment 0 im-ages with the intention of separating signal from noisewhen searching for low column density Hi . In order todo this, we set to zero the pixels with values below threetimes the noise in that cube, and again integrate over thesame channel range. After integration over the channelrange we selected, we expect the σ rms noise in each map(masked and unmasked) to increase by a factor of: σ N = √ N σ rms (5)
BT-MHONGOOSE N is the number of channels integrated over, and σ rms is the root-mean-squared noise per channel in thedata cube. We calculated a 1 σ noise map for both themasked and unmasked images this way. We then createour maps using a 3 σ cutoff in the same way as Pingelet al. (2018) by creating a signal-to-noise (S/N) mapfor both the masked and unmasked images. The S/Nmap for the unmasked image is made by dividing theunmasked integrated image by the unmasked 1 σ noisemap, while the masked S/N map is created by dividingthe image where pixels below 3 σ rms were set to zero bythe masked 1 σ noise image. From each of these S/Nmaps, we calculated a 3 σ threshold by taking the meanvalue of the pixels with S/N values falling between 2.75and 3.25. We then use this new 3 σ value to mask, or setto zero, the pixels in our masked integrated image thatfall below this value. In doing this, we are confident inour characterization of the noise in our maps.5.5. Cumulative Hi vs. N HI We want to find out how much Hi mass is containedat different column density levels in each galaxy. Indoing this, we will be able to identify the amount oflow column density Hi in the galaxy as a percentage ofthe total Hi mass. Accordingly, in the way described inPingel et al. (2018), we bin the column densities abovethe 3 σ level in our unmasked image and calculate thepercentage of Hi mass that falls into each N HI bin byconverting each pixel in the image to a column densitylevel, and subsequently into an Hi mass. We determinethe percentage of Hi mass at or above a given columndensity bin, and can therefore quantify the percentageof Hi mass below some column density threshold. Eachgalaxy is normalized to that galaxy’s maximum Hi massin the cumulative Hi vs. N HI plots in Figures 7 through23.In addition to the data from each cube, we use thebeam models described above to determine whether eachsource can be reliably analyzed. Each beam model foreach galaxy is scaled to the maximum column densityof the associated GBT image from the data and we an-alyze it in the same way as the GBT images describedin the previous paragraph, calculating the cumulative Hi mass in each N HI bin. In doing this, we can deter-mine if the source is following the response of the GBTbeam or deviating from it. If the source follows thebeam, we can determine that the source is unresolved,and therefore does not satisfactorily fill the beam, caus-ing the Hi column density to spread out over the beamarea appearing as if low column density Hi was detected.To this end, we would omit these unresolved sourcesfrom our statistical analysis of the disk. We distinguishwhich sources are unresolved by looking at the cumula-tive Hi mass ratio of the data versus the model. If thedata follows the beam response of the model, we deter-mine that source to be unresolved. Data points abovethe model is characterized as resolved, and if the error bars fully clear the beam model, we categorize it as verywell resolved.We can infer additional information about the amountof low column density Hi in the galaxy by looking at theshape of the profile in relation to the GBT beam model.If a source contains excess diffuse Hi , we will see a pos-itive deviation in the data at low column densities ascompared to the beam model. Several of the resolvedsources follow the shape of the beam model, and theyflatten out at or above the 3 σ column density thresh-old described in Section 5.3, which is indicative of thescarcity of low column density Hi in that galaxy. If, in-stead, we see an increase at low column densities as com-pared to the beam model, then we know we are detect-ing an excess of low column density Hi within that halo,which could be diffuse gas or clumpy material spreadover the beam. This excess could be the result of anumber of scenarios including detection of a low columndensity companion, extended or clumpy Hi clouds, thepresence of a > σ noise artefact, or the detection ofaccretion of low column density Hi from the CGM. Foreach image showing a higher fraction of low column den-sity Hi , we search for > σ artefacts in order to rule outthat possibility, and look for higher S/N regions in boththe cubes and the integrated images.5.6. Radial N HI We also want to determine how Hi column densities( N HI ) behave as a function of distance from the galaxy.The optimal way of identifying abundances in low col-umn density Hi is to take the average of the Hi columndensity levels in annuli around the galaxy. This allowsus to characterize the azimuthally averaged N HI at vari-ous physical galaxy radii. We define the annuli for eachgalaxy using the masked rather than unmasked imagesso as to reduce the effect of quantifying low level noiseas signal. Each image is regridded to have 1024 × (cid:48)(cid:48) per pixel for the purpose ofprecision when quantities inside each annulus are calcu-lated. The GBT beam’s FWHM of 9 . (cid:48) sets a lower limitto the radius of the smallest annulus to be 4 . (cid:48) , or ∼ N HI within each annulus isthen averaged by summing the values greater than 3 σ and dividing this by the total number of pixels in theannulus. Dividing by the total, rather than the numberof pixels above 3 σ , we run the risk of underestimatingthe average N HI in the annulus, but we are avoiding abias toward higher N HI values which would result fromthe average using only values > σ .Once again, we want to be able to compare the dataagainst the GBT beam model to look for deviation fromthe response of the beam. Each model is scaled to the Sardone et al.
Table 2. Hi Measurements and Derived Properties of MHONGOOSE Galaxies
Source σ rms S Hi V sys W W log N HI 1 σ log N HI 3 σ log M σ log M Hi log M dyn f Hi [mJy] [Jy km s − ] [km s − ] [km s − ] [km s − ] [cm − ] [cm − ] [M (cid:12) ] [M (cid:12) ] [M (cid:12) ](1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)ESO300G014 10.4 33.6 955.0 129.2 145.2 17.35 18.33 6.42 9.12 10.36 0.058ESO300G016 9.1 4.9 710.5 30.1 43.9 17.29 18.27 6.07 8.0 8.78 0.167ESO302G014 9.6 13.7 869.3 65.5 87.1 17.31 18.29 6.29 8.65 9.91 0.055ESO357-G007 4.9 15.0 1118.3 118.4 148.5 17.02 18.0 6.37 9.05 9.91 0.137KK98-195 5.9 8.1 570.6 27.1 42.2 17.1 18.08 5.38 7.71 7.95 0.577KKS2000-23 5.3 14.2 1037.5 80.2 96.8 17.06 18.03 6.11 8.73 9.02 0.515NGC1371 7.6 90.9 1454.3 386.4 403.8 17.21 18.19 6.68 9.95 11.63 0.021NGC1592 4.8 5.8 942.9 55.1 96.2 17.01 17.99 6.08 8.36 9.34 0.104NGC1744 3.2 174.7 743.4 191.1 207.2 16.75 17.82 5.59 9.62 10.56 0.112NGC3511 8.3 74.6 1130.8 272.4 308.7 17.25 18.23 6.4 9.55 11.11 0.028NGC5068 4.8 191.5 667.7 67.9 108.0 17.01 17.99 5.53 9.33 10.53 0.063NGC5170 6.2 106.2 1546.7 504.0 523.0 17.13 18.1 6.87 10.29 11.94 0.023NGC5253 8.3 57.6 404.5 63.4 99.1 17.25 18.23 5.05 8.09 9.38 0.051NGC7424 5.6 297.2 934.9 152.6 170.5 16.99 18.06 6.1 10.11 10.99 0.13UGCA015 5.2 5.2 293.6 25.5 43.2 17.05 18.02 4.93 7.12 8.23 0.078UGCA250 7.6 87.6 1700.6 272.2 289.5 17.21 18.19 6.83 10.09 11.03 0.115UGCA307 5.0 32.0 822.9 68.5 95.2 17.03 18.01 5.74 8.75 9.42 0.211UGCA320 7.8 133.9 749.1 106.3 126.9 17.23 18.2 5.84 9.27 10.08 0.156 Note —(1) Source name. (2) measured rms noise. (3) Total integrated flux. (4) Systemic velocity. (5) Linewidth at 50% maximum. (6) Linewidthat 20% maximum. (7) 1 σ rms column density sensitivity per channel. (8) 3 σ rms column density level over a 20 km s − linewidth. (9) Hi masssensitivity. (10) Hi mass. (11) Dynamical mass. (12) Neutral gas fraction. maximum N HI value in the image, and drops off at largerradii. The galaxies in our sample follow the response ofthe beam at low radii, and begin to flatten out aroundthe 3 σ threshold. Some of the galaxies flatten out abovethis threshold, which could indicate a smooth extent ofthe galaxy rather than an ionization edge at a particularcolumn density level around that galaxy. The 1 σ and 5 σ thresholds are indicated for each galaxy as the dashedhorizontal lines in Figures 7 through 23.We can also infer details about the environments ofthese galaxies by looking at the shape of the radial N HI profile. If the data follows the noise levels out tolarge radii, we can then say that there is no low columndensity Hi surrounding that galaxy for levels at or abovethe rms noise sensitivity reached in that cube. However,several of the galaxies show a positive deviation in theaveraged N HI at these large radii. For each galaxy inwhich this positive deviation is identified, we investi-gate the cube and integrated maps in order to identify asource for the excess Hi . The excess could be due to de-tection of a companion, an extended Hi cloud, a higher σ noise spike, or the accretion of Hi onto the galaxy fromthe CGM. 5.7. Radial Flux
We can trace the flux in each galaxy out to the samephysical radii that we explored in the radial N HI analysisusing the same method to calculate each annuli. Weuse the unmasked images for this analysis as this is thebest way to characterize the total flux of the galaxy atvarious radii, since our unmasked images contain all theflux detected from the source.Since we are measuring the cumulative flux at physicalradii, we should see the profile continue to increase lin-early as the area inside each annulus increases linearly ifthe average N HI described above is flat. If the radial fluxprofile deviates from a positive linear trend, we know tolook for an increase in low column density gas aroundthat galaxy. A dip in the profile is likely indicative ofan unphysical negative feature in the image, causing thecumulative flux to decrease at that radius. If indeedwe see rising cumulative flux levels at increasing radii,these would be consistent with the detection of Hi atlarge impact parameters as is seen in Das et al. (2020) BT-MHONGOOSE Hi is detected at each point along the minor axisout to ∼
120 kpc. RESULTSMost of the galaxies in our sample are moderatelyto well-resolved in the GBT beam, and give us reliablestatistical properties. As described in Section 5.5, a re-solved source is defined by a positive offset of the datafrom the GBT beam model in the cumulative Hi mass vs. N HI plots and in the radially averaged N HI vs. physicalradius plots. The plots in Figures 7 through 23 show,clockwise from top left, the total integrated Hi profilevs. velocity, the cumulative Hi mass vs. N HI , the cu-mulative flux vs. physical radius, and the azimuthallyaveraged N HI levels vs. physical radius. For reference,we have put this analysis for NGC 7424 in Figure 4,while the remainder of the analysis for each galaxy inour sample can be found in the Appendix. The resultsfrom analysis of all of our sources will be described be-low, and they have been grouped by low (Bin 1), inter-mediate (Bins 2-5), and high (Bin 6) Hi mass displayedin Figure 1.6.1. Low M HI ( < log (M HI ) < (cid:12) ) Figure 7 displays a very slight positive offset ofESO 300-16 from the GBT model in the cumulative Hi mass vs. column density, where the lowest columndensity levels deviate in the positive direction. At thesmallest radii seen in the radially averaged N HI plot, wecan see that this source is not well resolved until we movebeyond the disk, where our underestimated N HI valuesare consistent with the noise and begin to rise. The cu-mulative flux continues to rise even at large distancesfrom the disk.We see the response of the data in the plots of KK98-195 (Figure 8) are fairly unremarkable. We see avery slight positive deviation of the cumulative Hi massas compared to the GBT beam model at the low-est N HI level, demonstrating a small fraction (less than10%) of the Hi mass in the galaxy to be at the lowestcolumn densities. The average N HI flattens outside thedisk in the radially averaged N HI plot, and the flux in-creases linearly in the cumulative flux plot.UGCA015 (Figure 9) is ∼ (cid:48) in angular distance fromthe large galaxy, NGC 247, which shows up in our datacube at overlapping velocities. For this reason, we an-alyzed a smaller region around UGCA015 in order toexclude possible emission from NGC 247. We see a pos-itive offset in the Hi mass at the lower column densities.While we were not able to extend out to farther radii,we note that the cumulative flux again does not beginto flatten beyond the disk.Peak column densities in each of these low M HI galaxies sit right around N HI = 10 cm − . Noneof the three galaxies in this mass range display any pos-itive deviation from the noise level in the azimuthallyaveraged N HI plots, but each one shows an increase in the fraction of Hi mass at the lowest column densities.The largest of these fractions comes from UGCA015,where the offset from the beam model deviates signifi-cantly, making up over 20% of the mass fraction fromthe lowest column densities. Each of the radially aver-aged cumulative flux profiles continue to rise as far asour maps reach. This indicates that we have not yetreached the end of the Hi emission in our maps of thesesmallest galaxies.6.2. Intermediate M HI ( < log (M HI ) <
10 M (cid:12) ) Figure 10 displays the properties of ESO 302-14,where we see an increase at the lowest column densi-ties in the cumulative Hi mass plot, flattening outsidethe disk in the radially averaged N HI plot, and increas-ing flux levels in the cumulative flux plot. de Bloket al. (2020) detect a filament extending off the diskof ESO 302-14 as well as an Hi cloud with a peak col-umn density of N HI = 4 × cm − about 30 kpc southof the center of the galaxy. These structures are signifi-cantly smaller than the FWHM beamwidth of the GBTand this emission would be spread out over the beam,diluting the signal.NGC 1592 (Figure 11) has a peak column density levelof just over N HI = 10 cm − . The cumulative flux risesat large radii.The properties of NGC 5253 are shown in Figure 12.There is an increase in the azimuthally averaged columndensity levels around 25 kpc, which is associated withone bright region in the integrated intensity (moment0) image. This bright region has no associated opticalcounterpart, previously identified Hi cloud, nor does thespectra show anything we would consider to be RFI.Thus, we considered it to be a candidate for accretionof Hi onto the galaxy.ESO 357-007 (Figure 13) contains increasing amountsof diffuse Hi beginning around N HI = 10 . cm − , com-prising over 20% of the cumulative Hi mass fraction. Wecan see in the azimuthally averaged N HI plot that thegalaxy does not appear to be resolved in the GBT beamat low physical radii of the galaxy, and displays an in-creasing average N HI at the radii beyond the disk. Thecumulative flux does show a non-linear increase at thelargest physical radii mapped. In the absence of evi-dence of a companion galaxy, or any bright artefact inthe data cube causing the cumulative flux to increase,we are led to believe that this evidence of excess neu-tral hydrogen at large radii could be accretion throughthe CGM onto ESO 357-007. Alternatively, this excessof low column density Hi could be indicative of a largereservoir of Hi sitting in the halo of ESO 357-007.KKS2000-23 shows a significant portion of its Hi massto be made up of low column density gas, as can be seenin the cumulative Hi mass plot in Figure 14. However,inspection of the data cube and the moment map reveala higher level of noise in the image, making it difficultto determine if the low column density gas comes from0 Sardone et al.
Figure 4.
Top left : Total integrated Hi profile. Top right : Cumulative Hi mass. The total Hi mass in the moment 0 map isplotted by the fraction of gas in column density bins, and compared to the GBT beam model, scaled to the peak column density. Lower left : Azimuthally averaged N HI . Column density averaged over annuli extending radially from the center of the galaxyand compared to the GBT beam model. The black dashed line characterizes the 1 σ noise in each annulus, and the blue dot-dashline represents the 5 σ noise in each annulus. Lower right : Cumulative flux. Flux in each of those annuli are summed to obtaina measure of the total flux out to the edge of each map.
BT-MHONGOOSE N HI gas.UGCA307 also shows an increase in low column den-sity gas in Figure 15. However, the azimuthally aver-aged N HI vs. physical radius displays a steep increasebetween 55 kpc and 65 kpc. We inspected the datacube and discovered a small galaxy, which we believe isLCRS B125208.8-112329, residing 72 kpc away in pro-jected distance. This can also be seen in the cumulativeflux vs. physical radius plot, where the flux increasesaround the same physical distance. This would be thefirst velocity information available for this galaxy, witha central velocity of 863 km/s.The spiral galaxy ESO 300-14 can be seen in Figure16. The total integrated flux of ESO 300-14 displaysthe expected double horn feature for an inclined, ro-tating galaxy. It is resolved in the GBT beam, wherethe cumulative flux follows the shape of the GBT beammodel, and is positively offset in relation to that model.The azimuthally averaged column densities vs. physi-cal radius show a rising N HI level outside the disk, andthe cumulative flux at increasing physical radii increaseslinearly.Similarly, the face-on galaxy, NGC 5068, exhibits sim-ilar behavior when comparing these same properties.The cumulative Hi mass plot in Figure 17 shows that ∼
99% of the mass in the galaxy is seen in column den-sities above N HI = 10 cm − . This is substantiated inboth the average N HI and cumulative flux vs. physicalradii plots, where there is no evidence of an increase in Hi gas at large radii.We detected two additional sources in the region closeto UGCA320: UGCA319 at ∼
41 kpc, and a second de-tection around ∼
61 kpc projected distances. The closercompanion, UGCA319 can be seen as a slight bump inthe averaged N HI around ∼
45 kpc in Figure 18. Thesecond Hi detection lacks a known counterpart in Hi ,but may be associated with the small galaxy LEDA886203, the only optical galaxy near this position. Thesetwo companions are likely what is causing the gradualincrease in cumulative Hi flux at radii larger than 40kpc. This would also be the first redshift informationavailable on LEDA 886203 with a central velocity of 727km/s.NGC 1371 is well-resolved in the GBT beam, as seenin the cumulative Hi mass plot in Figure 19. In mostof these plots, NGC 1371 behaves as expected, with theexception of an artefact at ∼
150 kpc as seen in theradially averaged N HI plot.One of our more massive galaxies, NGC 1744, showsno signs of excess amounts of low column density gas asseen in the cumulative Hi mass plot in Figure 20. Wecan see in the radially averaged N HI plot that this sourceis well-resolved, and shows an unusually high average N HI at radii beyond the disk. This could be indicative of smooth accretion of Hi , a reservoir of Hi halo gas, ora higher noise floor than we had estimated.NGC 3511 (Figure 21) shows large deviation at lowcolumn density levels, where the cumulative amount of Hi increases almost linearly through N HI = 10 . cm − .It is resolved in the GBT beam as seen in both the cu-mulative mass plot and the averaged N HI plot. Similarto NGC 1744, this source displays higher N HI values outinto the halo.NGC 5170, in contrast with NGC 1744 which re-sides in the same mass bin, does show an increase to-wards lower column density Hi mass. Figure 22 revealsa sharp change in the slope of the cumulative Hi massat column densities lower than ∼ . cm − . The az-imuthally averaged N HI plot does show an increase inaverage N HI around 175 kpc. Inspection of the moment0 image reveals some bright, filamentary structure at thesame physical distance, which could be accreting Hi gas.Without any indication of a nearby optical counterpartassociated with those bright regions which appear tolack optical counterparts, we consider this a candidatefor accretion from the CGM.6.3. High M HI ( < log (M HI ) <
11 M (cid:12) ) We can see NGC 7424, one of the most Hi massivegalaxies in our sample, in Figure 4. The data in the av-eraged N HI vs. physical radius plot shows slightly higherand gradually increasing values of N HI at radii outsidethe disk. Visual inspection of the image did not revealany regions of particular interest.One of our largest sources, UGCA250 (Figure 23), oc-cupies our highest mass bin: 10 < log (M HI ) < (cid:12) . The cumulative Hi mass increases at low columndensities, and the cumulative flux continues to grow lin-early as the radius increases. Inspection of the datacube revealed a bright, positive stripe across the top ofUGCA250, at the same declination as another brightsource on the edge of our cube, UGCA247. Withinthat stripe is a possible new Hi detection which is co-incident with the only catalogued galaxy within the sizeof the GBT beam at this position, FLASH J115508.00-282045.1 at a radius of ∼
185 kpc. DISCUSSION7.1. Hi in the CGM Our analysis of the cumulative Hi mass allowed us toquantify the total Hi mass in the moment maps. Ourmaps were made at a constant angular size of 2 ◦ × ◦ , sowe do not have a consistent physical region around eachmapped source where initial map sizes range from 105 –977 kpc on one side. The measurement of the Hi mass ismade in the moment map where we can subtract off the Hi mass of the disk, which is derived from the integratedspectra of each galaxy. The total Hi mass in the momentmap less the integrated Hi disk mass from the spectragives us some value for the amount of mass outside thedisk. We refer to this amount as f CGM . This fraction2
Sardone et al. represents the total amount of Hi mass in the CGM inrelation to the Hi disk mass: f CGM = M
CGM / M disk .Out of our 18 galaxies, 16 contained an excess amountof gas outside their disk by 0.02 to 3 times the amountof Hi in the disk. The only two galaxies we did not de-tect Hi in their CGM were NGC 1744 and NGC 7424.We can see this in both the cumulative Hi mass plotsand the cumulative flux plots for both of these galax-ies. In both of these galaxies, the cumulative Hi massstays mostly flat at column densities below its turnover( N HI (cid:46) . cm − ) and the cumulative flux does notrise at large distance from the galaxy. This same flat-tening also occurs with NGC 5068, which contains only2% more Hi outside its disk. This tells us that there isvery little, if any, low column density gas outside thedisks of these three galaxies. We generally find thatmaps covering a smaller physical area contain a smallertotal amount of M HI outside of the disk, and larger phys-ical maps contain a larger total amount of M HI outsideof the disk. This may seem like an intuitive result if Hi permeates the CGM of most galaxies, and yet it hasevaded most observational studies of Hi emission in theCGM of galaxies. This result is consistent with Das et al.(2020) where Hi emission was detected in each measure-ment at increasing distances along the minor axis of thetwo galaxies, NGC 891 and NGC 4565. These detectionscould be the outcome of ubiquitous CGM Hi which, inmapping observations such as this current work, wouldresult in a larger amount of Hi detected over larger phys-ical areas. A study of this type could be improved onwith a consistent physical parameter guiding the size ofthe area observed, such as the virial radius of a galaxy.Mapping the full extent of the halo, as the upcomingParkes-IMAGINE survey (Sardone et al., in prep) willdo, would give insight into the amount of Hi permeatingthe entire CGM of a galaxy.As our physical map sizes are not uniform, we wouldlike to know how the fraction of Hi mass in the CGMwould change, assuming smooth coverage of the gas, ifeach galaxy were mapped precisely to its virial radius,calculated from the virial mass found using the Mosteret al. (2010) stellar-to-halo-mass relation. We show thischange in Figure 5 where our empirical measurementsare shown in the top panel, and a scaled version is shownin the lower panel. We find that as an increasing frac-tion of the halo is probed, and in several cases multipletimes the halo, the fraction of Hi gas in the CGM alsoincreases. The largest value of f CGM comes from themap of NGC 1592. We made measurements over anarea twice the size of its halo and found 1.3 times theamount of Hi disk mass in the surrounding area. How-ever, when scaled the map to cover precisely one haloarea, A map /A vir = 1, this fraction reduces to 0.62 timesthe Hi disk mass. We note that the scaled fractions, f CGM , scaled , reveal an outlier containing three times theamount of Hi mass in its CGM, NGC 5253, which we Figure 5.
Top . The amount of Hi detected in the CGM as afraction of the Hi disk mass. These CGM mass measurementswere taken throughout the map area and are shown relativeto the target galaxy’s virial area. The colorbar identifies thephysical size of the map. The blue dashed line indicates amap area equal to the virial area. Bottom . The fraction ofmass in the CGM in the top plot now scaled such that eachmap area is equal to its virial area. Triangles offset for visualease. earlier identified as a candidate for accretion. We dis-cuss this further in the following sections.7.2.
Fraction of low N HI We define a characteristic column density as low N HI ifit is below a level of 10 cm − . This value is takenfrom the prediction that below ∼ cm − the amountof hydrogen in the neutral phase is truncated, and thegas transitions from mostly neutral to mostly ionized, BT-MHONGOOSE Hi disk. This was first ex-plained by Bochkarev & Siuniaev (1977) and later byMaloney (1993) who suggested that the ionization frac-tion sharply increases at a critical column density ofa few times 10 cm − , irrespective of galaxy mass orhalo parameters. An example of this sharp drop-off in Hi is the KAT-7 observations of M83, which revealeda steep decrease in N HI at the edge of the galaxy disk(Heald et al. 2016). Recently, Bland-Hawthorn et al.(2017) demonstrated that it may be possible to detectlow column density Hi at significantly larger radii thanthe predicted truncation radius from Maloney (1993).Specifically, they showed that the radius at which 50%of the total amount of hydrogen is ionized is significantlylarger. These results seem to be congruent with those ofIanjamasimanana et al. (2018) who investigated radiallyaveraged column density profiles of 17 galaxies in searchof a break at the ∼ cm − level and found no evi-dence for a sharp change anywhere above their detectionlimits.We determined the diffuse neutral fraction for eachgalaxy using (Pingel et al. 2018): f = 1 − M M Hi . (6)In this equation, the diffuse neutral fraction, f , is de-fined by the fraction of Hi below column densities of10 cm − , where M is the Hi mass at column densitylevels N HI ≥ cm − and M Hi is the total Hi massas determined in Section 5.5. Values of f are listed inTable 3.With our f values in hand, we would like to com-pare the diffuse neutral fraction as a function of galaxyparameter, such as galaxy density, baryonic mass, androtation velocity, and others plotted in Figure 6. Wechose these properties in order to make a comparisonwith theoretical predictions on the relationships betweendiffuse gas and cold mode accretion from the IGM.Our diffuse fractions range from 0.05 to 0.93 of thetotal Hi mass measured in each galaxy. However, asthe cumulative flux profile in nearly all of our galax-ies continued to rise even at larger radii, these fractionrepresent a lower limit. The three galaxies, NGC 1744,NGC 7424, and NGC 5068, in which the Hi truncates atsome radius have a very low diffuse fraction and are un-likely to change. Eleven of our galaxies are made up ofless than 50% diffuse gas. These galaxies are generallyour highest Hi mass galaxies, demonstrating a perhapsintuitive correlation between mass and diffuse fraction:higher Hi mass corresponds to a lower fraction of diffuse Hi . This correlation can be seen when we combine thestellar mass and Hi mass, or the baryonic mass, in panel(e) of Figure 6. In panel (a) of Figure 6, we demonstrate the size ofthe galaxies in relation to the GBT 9 . (cid:48) FWHM beamarea, which is illustrated by the dashed vertical line.This helps us determine if a galaxy is large enough tobe considered resolved in the GBT beam, which tells usif the N HI detected within the beam is representative ofthe real N HI or if it is spread out over the beam, dilutingthe physical N HI values of the galaxy. Each galaxy’sarea was determined by the number of pixels within thegalaxy where column density values were greater than1 × cm − . We note that the highest f value, fromESO 300-16, corresponds to the map with the smallestangular area.At the largest distances, the GBT’s beam covers thelargest physical area, and could potentially be affectedby beam dilution producing unphysically high values of f . It is clear in panel (b) that we do not see evidenceof high f values at large distances, or correspondinglylarge physical map areas. At an intermediate distanceof 9.3 Mpc, beam dilution in the map of ESO 300-16 isnot a concern.Another way to look for this bias toward unphysicallow column densities is to plot f against physical area,which we do in panel (c). Bias in this plot would presentas large physical areas with high f values, due to beamdilution. As we see nothing to indicate any such trend,we are confident that our spatial resolution is suitablefor this kind of analysis. We draw attention to the twooutlier galaxies in panel (c): NGC 5170 and UGCA 250.These galaxies make up the largest distances in our sam-ple, are the first and third largest Hi disk masses in oursample, and contain less than 25% diffuse Hi . This isconsistent with the general idea that galaxies with largerpotential wells are more efficient at bringing halo gas tothe disk, leaving less diffuse gas in the CGM. It couldalso be the result of a hot halo, which is more commonaround massive galaxies (Kereˇs et al. 2009), and most ofthe gas has been ionized, resulting in a smaller neutralfraction.In panel (d) we look at the environments in whichthese galaxies live using galaxy number densities fromthe Nearby Galaxy Catalog (Tully 1988). This catalogidentifies the density of galaxies brighter than -16 mag-nitudes measured in the B-band within a Mpc regionof the target source using a 3D-grid at 0.5 Mpc spacing.We were unable to obtain galaxy number densities forfour of our galaxies (ESO300-16, KK98-195, KKS2000-23, and NGC1592), reducing the size of our sample forthis analysis. Regardless, the data we show in panel (d)do not indicate any relation between a galaxy’s diffuse Hi fraction and its density environment.In panel (e) we plot f with baryonic mass for allbut three of our galaxies. Kereˇs et al. (2005) showthat cold mode accretion dominates at baryonic masses(M bary = 1 . · M Hi + M ∗ ) less than M bary = 10 . M (cid:12) . We calculated the baryonic masses using our de-rived M Hi disk value, which was corrected by a factor of4 Sardone et al.
Figure 6.
Comparisons of the diffuse neutral fraction for each GBT-MHONGOOSE galaxy. Circles represent measurementsover an area larger than or equal to the area of the halo. Triangles represent measurements over a region smaller than thehalo and should be interpreted as a lower limit. The three galaxies (NGC 5068, NGC 1744, NGC 7424) whose cumulativeflux profiles flatten at large radii, indicating the Hi has truncated, are shown as unfilled markers. (a) Angular areas calculatedinside region where column densities reach N HI > × cm − . Vertical dashed line represents the angular area of the 9 . (cid:48) FWHM GBT beam. (b) Distance. (c) Physical areas calculated from the N HI = 1 × cm − angular diameter. (d) Galaxydensity. (e) Baryonic mass using M bary = 1 . · M Hi + M ∗ . Vertical dashed line at the cold accretion threshold given inKereˇs et al. (2005). We show a best fit line through the data revealing, rather than a firm threshold for cold accretion, acorrelation between a galaxy’s baryonic mass and its diffuse gas fraction. The strength of this correlation is demonstrated withthe Spearman correlation coefficient and the associated p-value. (f) Dark matter halo mass derived from the Moster et al. (2010)relation. Vertical dashed line at threshold also given in Kereˇs et al. (2005). The shaded region marks a lower threshold showing,based on low N HI measurements from this work, the regime in which cold accretion is more likely. (g) Rotation velocity usingV rot = W /
2. A vertical dashed line at 125 km/s shows the threshold suggested by Kannappan et al. (2013), below whichgalaxies become dominated by their gas. We mark a shaded region, based on this work, demonstrating a lower threshold for adiffuse gas dominated regime below 80 km/s. (h) Specific star formation rate calculated using SFR from Leroy et al. (2019).
BT-MHONGOOSE Hi lie entirely below this threshold, complementingthe picture of diffuse gas from the IGM flowing directlyinto galaxies below this mass threshold. However, ratherthan one mass threshold below which diffuse gas mayflow from the IGM, we see a relation between this galaxymass and the diffuse Hi fraction. The trend we see inthe comparison of f with baryonic mass is quantifiedby performing tests using both the Spearman correla-tion coefficient and the Pearson correlation coefficientresulting in nearly identical values of r S (cid:39) r (cid:39) − . p (cid:39) − . These testsdemonstrate a strong relation between the amount ofdiffuse Hi in a galaxy and that galaxy’s baryonic mass.This observational trend complements the simulationsfrom (Kereˇs et al. 2009) which reveal a smooth trend ofincreasing cold gas fraction as galaxy mass decreases.We show dark matter halo masses in panel (f), de-rived from the Moster et al. (2010) stellar-to-halo-massrelation. In this panel we see that all galaxies withhalo masses below ∼ . M (cid:12) have a spread in diffuse Hi fractions, while above this threshold the diffuse frac-tion remains (cid:46) .
2. This data indicates that there maybe a dark matter halo mass threshold around 10 . M (cid:12) ,above which a galaxy contains less than 25% diffuse Hi .We have overlaid a vertical dashed line at the predictedthreshold where cold accretion should dominate belowM halo = 10 . M (cid:12) given in Kereˇs et al. (2005). Belowthis halo mass threshold, we should find larger diffusefractions associated with those galaxies. The shaded re-gion in panel (f) pushes back this threshold to below ∼ . M (cid:12) .Rotation velocities are plotted against f in panel (g)with a distinction being made at the 125 km s − mark.Kannappan et al. (2013) suggests that this is the thresh-old below which galaxies are gas-dominated. In thisregime, galaxies are refueled and have larger gas frac-tions than those above this threshold. We plot our al-ternative measurement using f , which would estimatea “diffuse” gas richness. Once again, while we have aspread of diffuse fraction values below v rot = 125 km s − ,galaxies with higher f values all lie below this thresh-old supporting the picture of diffuse gas fueling galaxies.We further demonstrate that the data suggests an evenlower threshold where galaxies potentially become dom-inated by their diffuse gas. This is marked in panel (g)by the shaded below a rotation velocity of 80 km/s.The last panel (h) in Figure 6 is of specific star forma-tion rate (sSFR) versus f . We derive sSFR, the starformation rate per unit stellar mass, using the same stel-lar masses mentioned above with star formation ratesderived from UV+IR provided by Leroy et al. (2019).As star formation and galaxy mass is directly related Table 3. Hi Depletion Timescales
Source SFR log M Hi disk τ gasdisk f [M (cid:12) yr − ] [M (cid:12) ] [Gyr](1) (2) (3) (4) (5)E300G014 0.063 9.12 28.41 0.44E300G016 ... 8.0 ... 0.93E302G014 0.038 8.65 15.98 0.64ESO357-G007 0.041 9.05 37.46 0.60KK98-195 ... 7.71 ... 0.72KKS2000-23 ... 8.73 ... 0.72N1371 0.427 9.95 28.41 0.17N1592 0.079 8.36 3.92 0.83N1744 0.195 9.62 29.08 0.09N3511 0.447 9.55 10.8 0.21N5068 0.288 9.33 10.08 0.07N5170 0.589 10.29 45.03 0.14N5253 0.525 8.09 0.32 0.30N7424 0.372 10.11 47.16 0.05UGCA015 0.001 7.12 24.18 0.83UGCA250 0.603 10.09 27.77 0.22UGCA307 0.031 8.75 24.75 0.43UGCA320 0.021 9.27 121.21 0.18 Note —Star formation rates obtained from Leroy et al.(2019). Column 3 lists the Hi mass of the disk. Col-umn 4 lists the gas depletion timescales for the diskgas, including a correction for Helium mass. Col-umn 5 lists the diffuse neutral fraction of gas below N HI < cm − .to cold mode accretion, we want to look for trends insSFR as it relates to both. We note in this panel thatthe highest diffuse fractions tend to fall at the higherend of our sSFR range, which could be contributed toby either inflows into or outflows from the galaxy (e.g.Rubin et al. (2010); Martin et al. (2012); Krumholz &Dekel (2012)). 7.3. Depletion timescales
One method of determining whether nearby galaxiescontain enough gas to continue to fuel star formation isto calculate the depletion timescale of the galaxy, given6
Sardone et al. the gas mass and the SFR. This timescale determineshow long it will take the galaxy to use up all of its gasif all of this gas is converted into stars. Our depletiontimescales are calculated with: τ gasdisk = M gasdisk SF R , (7)where τ gasdisk is the depletion timescale given the gas massfrom the disk, and M gasdisk is the Hi mass from the disk,measured from the total integrated Hi profiles, correctedfor helium with a factor of 1.36. These timescales, listedin Table 3, have a large range of values which fall be-tween 0.32 Gyr and 121 Gyr, with a median depletiontime of 27.8 Gyr. The depletion timescales we findare consistent with the Hi depletion timescales foundby both Bigiel et al. (2010) and Roychowdhury et al.(2014). These results tell us that these galaxies currentlyhave enough gas available to them to sustain star for-mation for well over a Hubble time, with the exceptionof NGC 1592, NGC 3511, NGC 5068, and NGC 5253whose values are less than 13 Gyr. In Section 6.2 wenoted NGC 5253 as a potential source for accretion.NGC 5253 has been shown to have extremely efficientstar formation and streams of metal-enriched dense gasflowing into the starbursting dwarf galaxy (Turner et al.2015). If NGC 5253 is undergoing ongoing accretion,this along with its star formation efficiency could ex-plain the lack of an established reservoir of diffuse Hi ,as well as its shorter depletion timescale. SUMMARYWe analyzed 21 cm data from GBT observations of18 MHONGOOSE galaxies to search for diffuse, lowcolumn density Hi in the CGM of these galaxies. Wereached a mean 1 σ column density sensitivity of N HI =1.3 × cm − per channel, and a 3 σ column densityof N HI = 1 . × cm − over a linewidth of 20 km s − .1. By comparing the total Hi mass in our maps to the Hi measured in just the disk of each galaxy, we de-termined that 16 out of 18 galaxies contained addi-tional Hi mass outside of their disks by an amountof 0.02-3 times as much. This extraplanar or CGM Hi was not found in NGC 1744 or NGC 7424.2. We measured the amount of diffuse Hi within eachmap, defined as Hi gas with column densities be-low N HI = 10 cm − . We compared the amountof diffuse Hi with the total Hi mass in the map. Wefound that the galaxies in our sample were madeup of between 5 - 93% diffuse Hi . If we bin our dif-fuse fraction values, f , by quartiles we find thatmore galaxies fall into the lowest quartile than anyother bin.3. We took measurements of azimuthally averaged Hi column densities around each galaxy extend-ing out to the edges of the uniform noise regions in our maps. In eleven of our galaxies, these av-eraged N HI values remained nearly constant out-side the disk region, showing no excess of Hi inthe averaged annuli throughout the CGM. How-ever, we identified seven galaxies where this col-umn density profile showed some > σ increasein averaged N HI outside of the galaxy disk. Thesegalaxies are KKS2000-23, NGC 1744, NGC 3511,NGC 5170, NGC 7424, UGCA250, and UGCA320.There are a number of reasons to find a high aver-aged N HI value outside of the disk, which includedetection of a companion (e.g. UGCA 320), orfilamentary structure possibly associated with ac-cretion (e.g. NGC 5170).4. We see a rising cumulative flux level at increasingdistances in the maps of 15 of our 18 galaxies. Thisis consistent with our measurement of the Hi massoutside the disk, where we find the same threegalaxies (NGC 1744, NGC 7424, and NGC 5068)do not contain any detectable amount of Hi massbeyond their disks. This excess mass is seen inthe cumulative flux plots where the flux does notflatten at large distances from the galaxy as mightbe expected if the Hi emission drops to zero.5. We compared the fraction of diffuse Hi in eachmap with various environmental properties of ourgalaxies in Figure 6. We look at particular prop-erties such as baryonic mass, dark matter halomass, and rotation velocities which have been as-sociated, either empirically or theoretically, withaccretion of gas onto galaxies. Most notably, wefind a strong correlation between the fraction ofdiffuse gas and the galaxy’s baryonic mass. Usingboth the Spearman and the Pearson correlationcoefficients, we find that these two properties arestrongly correlated, and have a probability of achance correlation of nearly zero.6. Simulations such as seen in Kereˇs et al. (2005) de-termine a dark matter halo mass threshold belowwhich cold-mode accretion is the dominate form ofaccretion of gas from the IGM onto galaxies. Ourresults indicate a lower threshold in the dark mat-ter halo mass of a galaxy below which we find allof our high diffuse Hi fractions. Our dark matterhalo mass threshold lies around ∼ . M (cid:12) . Thistells us that we can expect a galaxy with a darkmatter halo mass above this threshold to containno more than about 25% diffuse gas.7. We demonstrate that galaxies with high dif-fuse Hi fractions all have rotation velocities belowabout 80 km/s. We suggest this as a regime inwhich galaxies become dominated by their diffusegas. BT-MHONGOOSE Hi mass of the disks of each galaxy,along with their star formation rates, to find thedepletion timescales of each galaxy. These de-pletion times have a large spread, but only fourgalaxies have depletion times below a Hubble time.Most galaxies in our sample currently have enough Hi to continue to fuel star formation for well overa Hubble, where the median time is 27 Gyr.The high resolution MHONGOOSE survey withMeerKAT will provide additional insight into the originsof low column density gas in the CGM of these galaxies.Galaxies with diffuse fractions supporting cold mode ac-cretion models in our sample will be better understoodwith higher resolution data. With this single-dish coun-terpart data to the MeerKAT interferometer data, we will have a complete census of the Hi in each of thesegalaxies at all angular scales.ACKNOWLEDGMENTSA.E.S. is supported by an NSF Astronomy and As-trophysics Postdoctoral Fellowship under award AST-1903834. D.J.P. and A.E.S. acknowledge partial supportby NSF CAREER grant AST-1149491. This project hasreceived funding from the European Research Council(ERC) under the European Union’s Horizon 2020 re-search and innovation programme grant agreement no.882793, project name MeerGas. This research made useof the NASA/IPAC Extragalactic Database (NED). TheRobert C. Byrd Green Bank Telescope is operated bythe Green Bank Observatory. The Green Bank Observa-tory is a facility of the National Science Foundation op-erated under cooperative agreement by Associated Uni-versities, Inc.REFERENCES Bigiel, F., Leroy, A., Walter, F., et al. 2010, TheAstronomical Journal, 140, 1194,doi: 10.1088/0004-6256/140/5/1194Birnboim, Y., & Dekel, A. 2003, MNRAS, 345, 349,doi: 10.1046/j.1365-8711.2003.06955.xBland-Hawthorn, J., Maloney, P. R., Stephens, A., Zovaro,A., & Popping, A. 2017, ApJ, 849, 51,doi: 10.3847/1538-4357/aa8f45Bochkarev, N. G., & Siuniaev, R. A. 1977, Soviet Ast., 21,542Bond, J. R., Kofman, L., & Pogosyan, D. 1996, Nature,380, 603, doi: 10.1038/380603a0Boothroyd, A. I., Blagrave, K., Lockman, F. J., et al. 2011,A&A, 536, A81, doi: 10.1051/0004-6361/201117656Broeils, A. H., & Rhee, M.-H. 1997, A&A, 324, 877Carignan, C. 2016, in Astronomical Society of the PacificConference Series, Vol. 502, Frontiers in RadioAstronomy and FAST Early Sciences Symposium 2015,ed. L. Qain & D. Li, 55.https://arxiv.org/abs/1510.03462Das, S., Sardone, A., Leroy, A. K., et al. 2020, ApJ, 898,15, doi: 10.3847/1538-4357/ab97b9de Blok, W. J. G., Adams, E. A. K., Amram, P., et al.2016, in Proceedings of MeerKAT Science: On thePathway to the SKA. 25-27 May, 7.https://arxiv.org/abs/1709.08458de Blok, W. J. G., Athanassoula, E., Bosma, A., et al. 2020,A&A, 643, A147, doi: 10.1051/0004-6361/202038894 de Vaucouleurs, G., de Vaucouleurs, A., Corwin, Herold G.,J., et al. 1991, Third Reference Catalogue of BrightGalaxiesDi Teodoro, E. M., & Fraternali, F. 2014, A&A, 567, A68,doi: 10.1051/0004-6361/201423596Haynes, M. P., Giovanelli, R., Martin, A. M., et al. 2011,AJ, 142, 170, doi: 10.1088/0004-6256/142/5/170Heald, G., J´ozsa, G., Serra, P., et al. 2011, A&A, 526,A118, doi: 10.1051/0004-6361/201015938Heald, G., de Blok, W. J. G., Lucero, D., et al. 2016,MNRAS, 462, 1238, doi: 10.1093/mnras/stw1698Ianjamasimanana, R., Walter, F., de Blok, W. J. G., Heald,G. H., & Brinks, E. 2018, AJ, 155, 233,doi: 10.3847/1538-3881/aabbaaKannappan, S. J., Stark, D. V., Eckert, K. D., et al. 2013,ApJ, 777, 42, doi: 10.1088/0004-637X/777/1/42Katz, N., Keres, D., Dave, R., & Weinberg, D. H. 2003, inAstrophysics and Space Science Library, Vol. 281, TheIGM/Galaxy Connection. The Distribution of Baryons atz=0, ed. J. L. Rosenberg & M. E. Putman, 185,doi: 10.1007/978-94-010-0115-1 34Kereˇs, D., Katz, N., Weinberg, D. H., & Dav´e, R. 2005,MNRAS, 363, 2, doi: 10.1111/j.1365-2966.2005.09451.xKereˇs, D., Katz, N., Fardal, M., Dav´e, R., & Weinberg,D. H. 2009, MNRAS, 395, 160,doi: 10.1111/j.1365-2966.2009.14541.xKoribalski, B. S., Staveley-Smith, L., Kilborn, V. A., et al.2004, AJ, 128, 16, doi: 10.1086/421744Krumholz, M. R., & Dekel, A. 2012, ApJ, 753, 16,doi: 10.1088/0004-637X/753/1/16 Sardone et al.
Leroy, A. K., Sandstrom, K. M., Lang, D., et al. 2019,ApJS, 244, 24, doi: 10.3847/1538-4365/ab3925Madau, P., & Dickinson, M. 2014, ARA&A, 52, 415,doi: 10.1146/annurev-astro-081811-125615Maloney, P. 1993, ApJ, 414, 41, doi: 10.1086/173055Marasco, A., Fraternali, F., Heald, G., et al. 2019, A&A,631, A50, doi: 10.1051/0004-6361/201936338Martin, C. L., Shapley, A. E., Coil, A. L., et al. 2012, ApJ,760, 127, doi: 10.1088/0004-637X/760/2/127Meurer, G. R., Hanish, D. J., Ferguson, H. C., et al. 2006,ApJS, 165, 307, doi: 10.1086/504685Moster, B. P., Somerville, R. S., Maulbetsch, C., et al.2010, ApJ, 710, 903, doi: 10.1088/0004-637X/710/2/903Pingel, N. M., Pisano, D. J., Heald, G., et al. 2018, ApJ,865, 36, doi: 10.3847/1538-4357/aad816 Rhee, J., Lah, P., Briggs, F. H., et al. 2018, MNRAS, 473,1879, doi: 10.1093/mnras/stx2461Roychowdhury, S., Chengalur, J. N., Kaisin, S. S., &Karachentsev, I. D. 2014, Monthly Notices of the RoyalAstronomical Society, 445, 1392,doi: 10.1093/mnras/stu1814Rubin, K. H. R., Weiner, B. J., Koo, D. C., et al. 2010,ApJ, 719, 1503, doi: 10.1088/0004-637X/719/2/1503Sorgho, A., Carignan, C., Pisano, D. J., et al. 2019,MNRAS, 482, 1248, doi: 10.1093/mnras/sty2785Tully, R. B. 1988, Nearby galaxies catalogTurner, J. L., Beck, S. C., Benford, D. J., et al. 2015,Nature, 519, 331, doi: 10.1038/nature14218Walter, F., Brinks, E., de Blok, W. J. G., et al. 2008, AJ,136, 2563, doi: 10.1088/0004-6256/136/6/2563
BT-MHONGOOSE
Top left :Total integrated Hi profile measured used to measure the total integrated Hi mass. Top right : Cumulative Hi mass.The total Hi mass in the moment 0 map is plotted by the fraction of gas in column density bins, and compared tothe GBT beam model, scaled to the peak column density. Lower left : Azimuthally averaged N HI . Column densityaveraged over annuli extending radially from the center of the galaxy and compared to the GBT beam model. Theblack dashed line characterizes the 1 σ noise in each annulus, and the blue dot-dash line represents the 5 σ noise in eachannulus. Lower right : Cumulative flux. Flux in each of those annuli are summed to obtain a measure of the total fluxout to the edge of each map.0
Sardone et al.
Figure 7.
Same as Figure 4 for ESO 300-16.
BT-MHONGOOSE Figure 8.
Same as Figure 4 for KK98-195. Sardone et al.
Figure 9.
Same as Figure 4 for UGCA015.
BT-MHONGOOSE Figure 10.
Same as Figure 4 for ESO 302-14. Sardone et al.
Figure 11.
Same as Figure 4 for NGC 1592.
BT-MHONGOOSE Figure 12.
Same as Figure 4 for NGC 5253. Sardone et al.
Figure 13.
Same as Figure 4 for ESO 357-007.
BT-MHONGOOSE Figure 14.
Same as Figure 4 for KKS2000-23. Sardone et al.
Figure 15.
Same as Figure 4 for UGCA307.
BT-MHONGOOSE Figure 16.
Same as Figure 4 for ESO300-14. Sardone et al.
Figure 17.
Same as Figure 4 for NGC 5068.
BT-MHONGOOSE Figure 18.
Same as Figure 4 UGCA320. Sardone et al.
Figure 19.
Same as Figure 4 for NGC 1371.
BT-MHONGOOSE Figure 20.
Same as Figure 4 for NGC 1744. Sardone et al.
Figure 21.
Same as Figure 4 for NGC 3511.
BT-MHONGOOSE Figure 22.
Same as Figure 4 for NGC 5170. Sardone et al.