A Comparison of Circumgalactic MgII Absorption between the TNG50 Simulation and the MEGAFLOW Survey
Daniel DeFelippis, Nicolas F. Bouché, Shy Genel, Greg L. Bryan, Dylan Nelson, Federico Marinacci, Lars Hernquist
DDraft version February 18, 2021
Typeset using L A TEX twocolumn style in AASTeX63
A Comparison of Circumgalactic Mg ii Absorption between the TNG50 Simulation and theMEGAFLOW Survey
Daniel DeFelippis , Nicolas F. Bouch´e , Shy Genel ,
3, 4
Greg L. Bryan ,
1, 3
Dylan Nelson , Federico Marinacci , and Lars Hernquist Department of Astronomy, Columbia University, 550 West 120th Street, New York, NY 10027, USA Univ Lyon, Univ Lyon1, Ens de Lyon, CNRS, Centre de Recherche Astrophysique de Lyon UMR5574, F-69230 Saint-Genis-Laval,France Center for Computational Astrophysics, Flatiron Institute, 162 Fifth Avenue, New York, NY 10010, USA Columbia Astrophysics Laboratory, Columbia University, 550 West 120th Street, New York, NY 10027, USA Universit¨at Heidelberg, Zentrum f¨ur Astronomie, Institut f¨ur theoretische Astrophysik, Albert-Ueberle-Str. 2, 69120 Heidelberg, Germany Department of Physics and Astronomy “Augusto Righi”, University of Bologna, Via Gobetti 93/2, I-40129, Bologna, Italy Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA
Submitted to ApJABSTRACTThe circumgalactic medium (CGM) contains information on gas flows around galaxies, such asaccretion and supernova-driven winds, which are difficult to constrain from observations alone. Here,we use the high-resolution TNG50 cosmological magneto-hydrodynamical simulation to study theproperties and kinematics of the CGM around star-forming galaxies in 10 . − M (cid:12) halos at z (cid:39) ii absorption lines, which we generate by post-processing halos to account forphotoionization in the presence of a UV background. We find that the Mg ii gas is a very good tracerof the cold CGM, which is accreting inwards at an inflow velocity of ∼
50 km s − . For sightlinesaligned with the galaxy’s major axis, we find that Mg ii absorption lines are kinematically shifteddue to the cold CGM’s significant corotation at speeds up to 50% of the virial velocity for impactparameters up to 60 kpc. We compare mock Mg ii spectra to observations from the MusE GAs FLowand Wind (MEGAFLOW) survey of strong Mg ii absorbers ( EW ˚ A > . ii spectra reflect the diversity of observedkinematics and EWs from MEGAFLOW, even though the sightlines probe a very small fraction ofthe CGM. Mg ii absorption in higher-mass halos is stronger and broader than in lower-mass halos buthas qualitatively similar kinematics. The median specific angular momentum of the Mg ii CGM gas inTNG50 is very similar to that of the entire CGM and only differs from non-CGM components of thehalo by normalization factors of (cid:46)
Keywords:
Galaxy formation (595) – Galaxy dynamics (591) – Galaxy kinematics (602) – Galaxystructure (622) – Circumgalactic medium (1879) – Hydrodynamical simulations (767) INTRODUCTIONThe accretion of gas onto disk galaxies is a funda-mental part of galaxy formation and evolution, as gaswithin disks is continually used to form stars and musttherefore be regularly replenished (e.g., Putman 2017).
Corresponding author: Daniel [email protected]
All such gas, whether pristine gas from cosmological in-flows or recycled gas in the process of reaccreting, mustpass through the local environment surrounding galax-ies, often called the circumgalactic medium (CGM). TheCGM might contain a substantial amount of angularmomentum as shown by many studies of galaxy simu-lations (e.g. Stewart et al. 2011; Danovich et al. 2015;DeFelippis et al. 2020). As the gas accretes onto thegalaxy, the angular momentum will flow inwards too, a r X i v : . [ a s t r o - ph . GA ] F e b DeFelippis et al. meaning the CGM is a source not just of the mass ofthe disk, but its angular momentum as well.Not all gas surrounding galaxies is inflowing though:the CGM also contains outflowing gas ejected from thegalaxy by feedback from supernovae and active galacticnuclei (AGN), which is capable of affecting the way inwhich CGM gas eventually joins the galaxy (DeFelippiset al. 2017). All of these physical processes occur con-currently and result in a multiphase environment shownin observations to have complex kinematics (see Tum-linson et al. 2017, and references therein).A large number of recent observations of the CGMhave been accomplished through absorption line stud-ies of background quasars through dedicated surveys.For instance, some surveys are constructed by cross-correlating quasar absorption lines with spectroscopicredshift surveys such as the Keck Baryonic StructureSurvey (KBSS: Rudie et al. 2012; Turner et al. 2014) orwith photometric surveys like SDSS (Lan & Mo 2018;Lan 2020). Other CGM surveys attempt to either matchindividual absorption lines to known galaxies (i.e. are“galaxy-selected”), like the COS-Halos (e.g. Tumlinsonet al. 2011; Werk et al. 2013; Burchett et al. 2019)and the low-redshift Keck surveys conducted at KeckObservatory (Ho et al. 2017; Martin et al. 2019), ormatch galaxies near known absorbers (i.e. “absorber-selected”) such as the MusE GAs FLOw and Wind sur-vey (MEGAFLOW: Schroetter et al. 2016, 2019, 2021;Wendt et al. 2021; Zabl et al. 2019, 2020, 2021).The Mg ii ion has been a focus of many recent surveysincluding the Mg ii Absorber-Galaxy Catalog (MAGI-ICAT: e.g. Nielsen et al. 2013, 2015), the MUSE Analy-sis of Gas around Galaxies Survey (MAGG: Dutta et al.2020), the PRIsm MUlti-object Survey (PRIMUS: Coilet al. 2011; Rubin et al. 2018), and the aforementionedMEGAFLOW survey. These studies belong to a longhistory of Mg ii λ ii has also been seen in emission inextended structures around the galaxy and in the CGM(e.g. Rubin et al. 2011; Rickards Vaught et al. 2019;Rupke et al. 2019; Burchett et al. 2020; Zabl et al. 2021).Along with this wealth of Mg ii observations, re-searchers in recent years have found Mg ii kinematicsto be correlated over large spatial scales. In particu-lar, both Bordoloi et al. (2011) and Bouch´e et al. (2012)found a strong dependence of Mg ii absorption with az-imuthal angle: specifically, more absorption near φ = 0 ◦ and 90 ◦ and a lack of absorption near 45 ◦ . This type ofabsorption distribution is generally interpreted as bipo- lar outflows along the minor axis and inflows along themajor axis. In this context, both galaxy-selected (e.g.Ho et al. 2017; Martin et al. 2019) and absorption-selected Mg ii studies (e.g. Bouch´e et al. 2013, 2016; Zablet al. 2019) have given support to the interpretation ofaccretion of gas from the CGM onto the galaxy. TheseMg ii studies show that when sightlines are located nearthe major axis of the galaxy there are clear signaturesof corotating cold gas with respect to the galaxy kine-matics.However, despite such extensive observational data,developing a general understanding of cold gas in theCGM from the Mg ii line alone remains difficult due tothe limited spatial information provided by the obser-vational technique (though IFU mapping of lensed arcsin e.g. Lopez et al. 2020 and Mortensen et al. 2020 canimprove this in the future), as well as the fact that Mg ii gas may not be representative of the entire cold phase ofthe CGM. To study more physically fundamental prop-erties of the CGM, it is therefore necessary to turn togalaxy simulations.In cosmological simulations (see Vogelsberger et al.2020 for a review), the CGM has been notoriously diffi-cult to model accurately due to the need to resolve verysmall structures (e.g. Hummels et al. 2019; Peeples et al.2019; Suresh et al. 2019; Corlies et al. 2020). Nonethe-less, the CGM has been shown to preferentially alignwith and rotate in the same direction of the galaxy, espe-cially near the galaxy’s major axis (Stewart et al. 2013,2017; Ho et al. 2019; DeFelippis et al. 2020), which isqualitatively consistent with observations in the samespatial region of the CGM (e.g. Zabl et al. 2019). How-ever, this general qualitative agreement between simula-tions and observations is difficult to put on firm groundsquantitatively due to the inherent differences betweenobservations and simulations.In this paper, we analyze a set of halos from theTNG50 simulation (Nelson et al. 2019; Pillepich et al.2019) using the Trident tool (Hummels et al. 2017) tomodel the ionization state of the CGM and then per-form a quantitative comparison of the kinematics of thecool ( T (cid:46) × K) CGM traced by Mg ii to major-axis sightlines from the MEGAFLOW survey (Zabl et al.2019) while attempting to match the observational se-lection criteria as described in Section 2. We note thatour comparison to MEGAFLOW galaxies with stellarmasses M ∗ ∼ M (cid:12) is complementary to that of Nel-son et al. (2020) who study the origins of cold CGM gasof very massive galaxies ( M ∗ > M (cid:12) ).The paper is organized as follows. In Section 2, wedescribe the TNG50 simulation and MEGAFLOW sam-ple used in the comparison, and we outline the analysis ircumgalactic Mg ii in TNG and MEGAFLOW METHODS2.1.
Simulations
We utilize the TNG50 simulation (Nelson et al. 2019;Pillepich et al. 2019), the highest resolution versionof the IllustrisTNG simulation suite (Marinacci et al.2018; Naiman et al. 2018; Nelson et al. 2018; Pillepichet al. 2018; Springel et al. 2018), which is itself basedon the original Illustris simulation (Vogelsberger et al.2014a,b). TNG50 evolves a periodic ≈ (52 Mpc) boxfrom cosmological initial conditions to z = 0 with themoving-mesh code Arepo (Springel 2010; Weinbergeret al. 2020). It has a baryonic mass resolution of ∼ . × M (cid:12) per cell, which is a factor of ≈
16 betterthan the resolution of TNG100. We discuss the effect ofsimulation resolution on our results later in Section 3.2.2.
Observational Data
The MEGAFLOW survey (Bouch´e et al., in prep.)consists of a sample of 79 Mg ii λλ , ii ab-sorbers from the Zhu & M´enard (2013) SDSS catalogin the redshift range 0 . < z < . ii ] λλ , − ∼ . − . ii absorber-galaxy pairswhose quasar location is positioned within 35 ◦ of themajor axis of the host galaxy (Zabl et al. 2019). Thissubset consists of nine absorber-galaxy pairs with red-shifts 0 . < z < . b ) rangingfrom 13 −
65 kpc with a mean of ≈
34 kpc. Zabl et al.(2019) found that the Mg ii gas in these absorbers showa strong preference for corotation with their correspond-ing host galaxies.The galaxies in Zabl et al. (2019) are both fairlyisolated by having at most one companion within100 kpc, and star-forming with [O ii ] fluxes f O ii > × − erg s − cm − , i.e. star-formation rates (cid:38) M (cid:12) yr − . The galaxies have a mean stellar mass vir [M ]2.01.51.00.50.00.51.01.52.0 l o g S F R [ M y r ] l o g M * [ M ] Figure 1.
Star formation rate of the central galaxy vs. halomass for all TNG50 halos between 10 M (cid:12) and 10 M (cid:12) at z = 1. Each point is colored by the stellar mass of the halo’scentral galaxy. Two thick vertical lines demarcate the halomass range of the fiducial sample. M ∗ ≈ M (cid:12) and halo masses M vir ranging from ≈ . − . M (cid:12) , where M vir is defined from the stel-lar mass-halo mass relation from Behroozi et al. (2010).As Zabl et al. (2019) show, these halo masses generallymatch the Bryan & Norman (1998) definition of M vir .2.3. Sample selection and forward-modeling
Figure 1 shows the central galaxies’ instantaneous starformation rates (SFR) and stellar masses of all TNG50halos in and around the mass range of interest. Since weaim to compare the Mg ii absorption properties of mockline-of-sight (LOS) observations through TNG50 halosto those of major-axis sight-lines of the MEGAFLOWsurvey, we first select a sample of simulated halos at z =1 in the mass range 10 . M (cid:12) < M halo < M (cid:12) usingthe Bryan & Norman (1998) definition for M halo , whichresults in a sample of 495 halos. In the remainder of thispaper, we will refer to this sub-sample as the “fiducial”sample. The chosen redshift is typical for the Zabl et al.(2019) sample and the halo mass range covers the typicalinferred virial masses of their halos. Nearly all of thehalos in our fiducial sample host central galaxies withSFR (cid:38) (cid:12) yr − and stellar masses of ∼ M (cid:12) ,which is consistent with the MEGAFLOW sub-sampleas described in Section 2.2.For each halo, we adjust all velocities to be in thecenter-of-mass frame of the stars in the central galaxy,and we rotate it so that the stellar specific angular mo-mentum of the central galaxy points in the + z -direction(the x and y directions are both arbitrary). With this ge-ometry we then define a sightline in the x − z plane by theimpact parameter b , the azimuthal angle α , and the in-clination angle i , where b is the projected distance fromthe center of the galaxy in the y − z plane (i.e. “sky”- DeFelippis et al. plane), α is the angle above the rotational plane of thegalaxy, and i is the angle of the sightline with respect tothe sky-plane. In this setup, edge-on and face-on viewshave i = 90 ◦ and i = 0 ◦ respectively (see Figure 1 ofZabl et al. 2019 for a sketch of the geometry describedhere). In order to mimic the observations of Zabl et al.(2019), we select sightlines through each halo at valuesof b ranging from 15 kpc to 60 kpc, α = 5 ◦ and 25 ◦ , andat i = 60 ◦ , representing the average inclination angle ofa random sightline.In order to generate observations of our TNG50 sam-ple, we use the Trident package (Hummels et al. 2017),which calculates ionization parameters for outputs ofgalaxy simulations using properties of the simulated gascells and
Cloudy (Ferland et al. 2013) ionization tables.These tables take as input the gas temperature, density,metallicity, and cosmological redshift of each gas cell andprovide ionization fractions and number densities of de-sired ions. We make use of the current developmentversion of
Trident (v1.3), which itself depends on thecurrent development version of yt (v4.0). In this pa-per, we use a set of ion tables created assuming col-lisional ionization equilibrium, photoionization from aFaucher-Gigu`ere et al. (2009) UV background, and self-shielding of neutral hydrogen (for details see Emericket al. 2019 and Li et al. 2021), as this was the backgroundradiation model used to evolve the TNG50 simulation.We also use the elemental abundance of magnesium ineach gas cell tracked by the simulation rather than as-suming a constant solar abundance pattern throughoutthe halo to achieve greater self-consistency with TNG50.We note, however, that our results are not particularlysensitive to either of these choices.Since our focus is on the Mg ii λ ii mass probability density. From this plot,it is clear that Mg ii is mostly formed from the coldest( (cid:46) . K) and densest ( (cid:38) .
01 cm − ) gas in the halo,though some Mg ii mass exists at a larger range of tem-peratures and densities. However, contours showing thetotal gas mass demonstrate that despite this large rangein temperature and density, essentially none of the dif-fuse “hot” phase, comparable in mass to the cold phase,contributes to Mg ii absorption. We also note here thatfor this analysis we are excluding star-forming gas asits temperature and density are defined using an effec-tive equation of state (Springel & Hernquist 2003) and trident-project.org yt-project.org ]345678 l o g T [ K ] M g II p r o b a b ili t y d e x Figure 2.
Temperature-number density phase diagram ofa single TNG50 halo at z = 1, colored by the Mg ii massprobability density per dex . Contours show the distributionof all gas mass in the halo. are therefore not analogous to the physical properties ofnon-star-forming gas. Properly modeling the physicalproperties of the star-forming gas (see Ramos Padillaet al. 2021 for an example of this technique) introducesa level of complexity not necessary for this analysis: wefind that our results are not affected by the exclusionof this gas since our sightlines through the CGM rarelyintersect any star-forming gas cells as most of them arewithin the galactic disk. RESULTSWe first present in Section 3.1 the results of di-rectly comparing the Mg ii properties of TNG50 andMEGAFLOW using the analysis described in Section 2.Then, we further analyze the 3D kinematic propertiesof the Mg ii bearing gas from TNG50 in Section 3.2 andconsider evolution of Mg ii absorption properties withhalo mass and simulation resolution in Section 3.3.3.1. Comparing TNG50 to MEGAFLOW
In Figure 3, we show Mg ii column density maps of aselection of TNG50 halos drawn from our fiducial sampleat z = 1. The halos are aligned so that the angular mo-mentum vector of the stars in the central galaxy pointsalong the vertical axis: thus, the view is edge-on. Thestrongest Mg ii columns are found within and very closeto the galaxy, demarcated by a red circle with a radiusof twice the galaxy’s stellar half-mass radius (the samedefinition used in DeFelippis et al. 2020). Beyond thegalaxy, Mg ii gas consistently appears to both surroundthe galactic disk and be very clumpy, but the amountand morphology of such gas varies greatly. In particular, ircumgalactic Mg ii in TNG and MEGAFLOW halo 225
50 kpc l o g M g II c o l u m n d e n s i t y [ c m ] halo 265 halo 271 halo 340 Figure 3. Mg ii column density maps of four TNG50 halos from the fiducial halo mass bin of 10 . M (cid:12) < M halo < M (cid:12) at z = 1, aligned so the angular momentum vector of the stars in the central galaxy points along the vertical axis (i.e. edge-on).The lower limit of the colorbar is chosen to approximate observational detection limits. The red circle in each panel is centeredon the galaxy and has a radius of twice the galaxy’s stellar half-mass radius, and the blue scale-bar shows a distance of 50 kpcon the maps. The complexity and diversity of Mg ii structure in the CGM of similar-mass halos is evident even in this smallsample. there is significant variation with azimuthal angle: thehighest Mg ii columns generally appear in the plane ofrotation, but strong columns can occur above and be-low the disk as well, such as in halo 265 (the bottomleft panel of Figure 3). P´eroux et al. (2020) found theCGM gas metallicity to vary with azimuthal angle, butinterestingly, they found gas near the major axis to havelower than average metallicity in the halo, indicatingthat large Mg ii columns do not necessarily correspondto metal-enriched gas. High Mg ii columns are much lesscommon in the outer halo ( r (cid:38)
50 kpc), but the pres-ence of satellite galaxies can populate that region withMg ii gas, shown most clearly in halo 340 (the bottomright panel of Figure 3). Within our fiducial sample, it is evident that the dis-tribution of Mg ii varies drastically, presumably due todifferent halo formation histories. Sightlines throughdifferent halos will therefore likely produce different ab-sorption profiles even for sightlines with identical ge-ometries. This highlights the necessity of calculatingpopulation averages of Mg ii properties from TNG50 tocompare to MEGAFLOW.We begin such a comparison with Figure 4, whichshows the average strength of Mg ii absorption, repre-sented as the rest-frame equivalent width (EW ) as afunction of impact parameter ( b ) for our fiducial sam-ple. In this plot, we make an important distinction be-tween the entire fiducial sample, shown in black, and the DeFelippis et al. subset of “strong absorbers” in red. We define strongabsorbers as sightlines through a halo that produce anabsorption spectrum with EW > . ≈ b ≤
20 kpc to a factor of ≈
30 at60 kpc. If, instead, we compare the average EW of thestrong absorber subset (EW > . b ≥
40 kpc,but it is still as much as a factor of ≈ b ≤
20 kpc. However, the limitedsize and large scatter of the MEGAFLOW points fromZabl et al. (2019) make it difficult to assess the preciselevel of disagreement with TNG50. Sightlines at α = 5 ◦ (solid) and α = 25 ◦ (dotted) produce essentially identi-cal equivalent widths over both the entire fiducial sampleand the subset of strong absorbers.The blue lines in Figure 4 show the fraction of allsightlines that host strong absorbers as a function of im-pact parameter. At sightlines very close to the galaxy( b = 15 kpc), strong absorbers are common and in factrepresent a majority of all halos. However, by b = 20 kpcthe strong absorber fraction drops below 50%, and atthe largest impact parameters shown, the fraction isonly ≈ α = 5 ◦ compared to α = 25 ◦ , which can beunderstood by noting that the sightlines with smaller α pass through the disk midplane closer to the galaxy’scenter, where gas is generally denser. However, this dif-ference in strong absorber fraction does not affect themeasured equivalent widths, indicating that the TNG50halos’ agreement with MEGAFLOW for sightlines nearthe galaxies’ major axes is not subject to the precisegeometries of the sightlines.Having established the degree of consistency of Mg ii equivalent widths, we now examine kinematic signaturesof Mg ii along sightlines in TNG50 and compare themto MEGAFLOW. In Figure 5, we explicitly draw theconnection between the Mg ii gas cells that contributeto the column densities seen in Figure 3 and the veloc-ity spectrum created from a subset of those cells thatintersect a sightline through the halo. In each row, weshow two orientations of one of the four halos from Fig-ure 3 overlaid with a sightline with b = 30 kpc, α = 5 ◦ ,and i = 60 ◦ , and the Mg ii velocity spectrum generated
15 20 30 40 50 60 70b [kpc]0.050.10.20.5123 E W [ Å ] halo [M ] < 12 = 5EW > 0.5 Å= 25megaflow 0.00.20.40.60.81.0 s t r o n g a b s o r b e r f r a c t i o n Figure 4.
Mean Mg ii equivalent widths of halos in our fidu-cial sample vs. the impact parameter of sightlines throughthose halos. Black and red lines and corresponding shadedregions show mean and ± σ scatter of all halos and the sub-set of strong absorbers (EW > . α = 5 ◦ and 25 ◦ areshown with solid and dotted lines respectively. Observationsof individual accretion systems from Zabl et al. (2019) areshown as green squares. The fraction of strong absorbersas a function of impact parameter (blue) is shown with theright vertical axis. from that sightline. From these few examples it is clearthat the gas producing the Mg ii absorption is generallynot distributed uniformly along any sightline: it is usu-ally concentrated in discrete clumps in regions of thesightline nearest to the galaxy. This is seen clearly inrows one, two, and four of Figure 5, where the majorityof gas cells have positive LOS velocities (i.e. corotatingwith the galaxy) and produce distinct kinematic com-ponents in the spectrum that are often saturated.It is also notable that by comparing the spectra aloneit is possible to distinguish morphological differences inthe Mg ii distribution between halos. The first two ha-los, for example, have a prominent Mg ii disk that bothspectra reveal to be primarily corotating. The halo inrow three, however, does not have such a clear disk,and the spectrum is instead composed of a cluster ofcounter-rotating gas cells significantly above the planeof the galaxy. The halo in row four has a spectrumwith substantial corotating and counter-rotating com-ponents, which imply Mg ii structure in between the or-dered halos (rows one and two) and disordered ones (rowthree). With this small sample, we have demonstratedthat the velocity spectrum, despite being composed of avery small fraction of all of the Mg ii gas, is capable ofreflecting the potential diversity of Mg ii gas kinemat-ics in halos of similar mass, but is also fairly consistentbetween halos with similar morphologies. Later in the ircumgalactic Mg ii in TNG and MEGAFLOW x y halo 225 x z l o g M g II c o l u m n d e n s i t y [ c m ] vir sin(i))0.00.20.40.60.81.0 b = 30, = 5 , i = 60 x y halo 265 x z l o g M g II c o l u m n d e n s i t y [ c m ] vir sin(i))0.00.20.40.60.81.0 b = 30, = 5 , i = 60 x y halo 271 x z l o g M g II c o l u m n d e n s i t y [ c m ] vir sin(i))0.00.20.40.60.81.0 b = 30, = 5 , i = 60 x y halo 340 x z l o g M g II c o l u m n d e n s i t y [ c m ] vir sin(i))0.00.20.40.60.81.0 b = 30, = 5 , i = 60 Figure 5.
Each row contains two Mg ii column density maps of a halo from Figure 3 projected along the vertical (left) anda horizontal (middle) axis. A sightline at b = 30 kpc, α = 5 ◦ , and i = 60 ◦ is overlaid along with gas cells that intersectthat sightline and have a Mg ii column density of at least 10 cm − , which accounts for >
95% of the Mg ii mass along thosesightlines. The Mg ii gas cells and the resulting flux-normalized velocity spectrum (right) are colored by the velocity along theline of sight normalized by V vir sin( i ), where V vir is the virial velocity of the halo. Dashed circles show twice each galaxy’s stellarhalf-mass radius. DeFelippis et al. paper, we consider whether the Mg ii gas reflects thekinematics of other components of the CGM.From these results, we now compare stacked spectrafrom the fiducial sample to the stacked spectra presentedin Zabl et al. (2019). Figure 6 shows stacked spectrafor the entire TNG50 fiducial sample (black), TNG50strong absorbers (red), and the absorbers from Zablet al. (2019) (green). The two panels correspond to twodifferent impact parameters which allows a comparisonbetween absorbers nearer to a galaxy and farther froma galaxy. In the left panel, showing stacked spectra atsmall impact parameters, there is a very clear kinematicpicture. The strong absorber spectrum from TNG50 issymmetric and centered at ∼ V vir /
2, as is the spectrumof Zabl et al. (2019). Thus, qualitatively, strong Mg ii absorbers as a population generally have LOS velocitiesin the same direction as their corresponding galaxies’rotations. One slight difference with the stacked spec-tra for strong absorbers is that the TNG50 spectrum(red) is somewhat shallower than the observed spectrum(green). However, there is essentially no difference be-tween TNG50 spectra generated from sightlines at thetwo azimuthal angles α = 5 ◦ (solid line) and 25 ◦ (dottedline).In Figure 6 (left), the only difference between the fullfiducial spectrum and the strong absorber-only spec-trum is the depth, indicating that, as a population,strong absorbers are not kinematically distinct from ab-sorbers in general at this impact parameter. The precisereason for the discrepancy in the depth is difficult to de-termine, but it may be sensitive to certain parametersin the TNG physics model (e.g. metal-loading of out-flows from supernovae). However, it could also be an ef-fect of simulation resolution (see Section 3.3). So, whileTNG50 potentially slightly underproduces the observedamount of Mg ii gas at 20 kpc, it does possess averagekinematics that are consistent with observations of thesame region of the CGM.Figure 6 (right) compares the stacked spectra at alarger impact parameter ( b = 40 kpc). The strongabsorbers from TNG50 and MEGAFLOW (Zabl et al.2019) are both shallower, wider, no longer symmetric,and significantly noisier, though both are still approxi-mately centered around V vir /
2. At this impact param-eter, the depths of the simulated strong absorber andobserved spectra are consistent with each other. How-ever, strong absorbers no longer kinematically resemblethe full fiducial sample: in addition to being much rarerat 40 kpc than at 20 kpc, the strong absorbers havelarger positive velocities, indicating that Mg ii in thisregion is tracing atypically faster moving gas. As wasthe case at 20 kpc, the difference in the spectra between the two azimuth angles is minor. We also note here, butdo not show, that the shapes and depths of individualspectra from Zabl et al. (2019) match quite well withparticular individual spectra from the much larger fidu-cial sample from TNG50 (examples of individual spectrafrom TNG50 are shown in Figure 5).3.2.
3D Kinematics of Mg ii in TNG50 In this section, we characterize the three-dimensionalkinematics of the Mg ii gas in TNG50 and its relation tothe observed quantities we discussed in Section 3.1. Weshow average velocity profiles of the halos in the fiducialsample in Figure 7. The top panel shows the azimuthalvelocity component ( v φ ) in spherical coordinates as afunction of radius. We divide gas into cold and hotcomponents based on a temperature threshold of 3 × K, which is chosen to separate the cold and hotclusters seen in Figure 2, although the profiles are notsensitive to the precise choice of temperature threshold.To understand the relationship of the hot and cold gasto Mg ii -bearing material we also show the Mg ii mass-weighted profiles.First, we see that the Mg ii gas and the cold gashave nearly identical v φ profiles throughout the halo.In the innermost regions of the CGM, the cold gas hasan azimuthal velocity of ≈ . V vir which decreases to ≈ . V vir in the outermost region. Furthermore, thoughnot explicitly shown, most of the cold and Mg ii gas massis closer to the major rather than the minor axis becausethe all- α profiles are much more similar to the α < ◦ (dashed) profiles than the α > ◦ (dotted) profiles.Hot gas has lower azimuthal velocities at all radii anda slightly shallower slope to its profile, but is otherwisequalitatively similar to the cold and Mg ii gas. This re-lationship between hot and cold gas is consistent withsimilar measurements of v φ made from TNG100 in De-Felippis et al. (2020).In the radial velocity profiles (Figure 7, bottom), wesee a gulf between the velocities of the hot and cold gasdevelop within ∼
80 kpc where the overall averaged ve-locity of the gas indicates an inflow at ∼ . V vir . Movingtowards smaller radii, the cold gas inflow velocities be-come larger, while hot gas inflow velocities decrease andthen switch to a net outflow within ∼
40 kpc. The Mg ii gas still traces the cold gas and reaches typical inflowingvelocities of ∼
50 km s − around 30 kpc. However, inthe outermost regions of the halo, the radial velocitiesof all components of the gas converge. The geometryof accretion and outflows is evident from this panel aswell: hot gas has especially large mean outflowing veloc-ities for α > ◦ while cold gas in the same region has amean inflowing velocity in the inner halo and nearly no ircumgalactic Mg ii in TNG and MEGAFLOW vir sin(i))0.20.40.60.81.0 b = 20 kpc all halosEW > 0.5Åmegaflow b < 30 kpc 2.0 1.5 1.0 0.5 0.0 0.5 1.0 1.5 2.0v/(v vir sin(i))0.20.40.60.81.0 b = 40 kpc = 5= 25megaflow b > 30 kpc Figure 6.
The stacked Mg ii velocity spectra for the full fiducial TNG50 sample (black) and the subset of strong absorbers(red) for sightlines with α = 5 ◦ (solid) and 25 ◦ (dotted), and b = 20 kpc (left) and 40 kpc (right). Spectra are normalized by V vir sin( i ), where V vir is the halo’s virial velocity. The green line in each panel is the stacked spectrum of the 4 smallest (left)and largest (right) impact parameters from Zabl et al. (2019), and the green shaded region is an estimate of the error frombootstrapping. net radial motion in the outer halo. Most of the cold andMg ii gas mass is moving towards the galaxy in regionssurrounding the major axis out to a substantial fractionof the virial radius. It is also clear that kinematically,Mg ii gas in TNG50 is nearly identical to a simple cuton temperature and so is an excellent tracer of the kine-matics of cold CGM gas. In the context of Section 3.1,these results indicate that mock Mg ii spectra are repre-sentative of the entire cold phase of the CGM along thesame sightlines.Finally, we examine the 3D velocities of the Mg ii gasalong our sightlines. In Figure 8, we plot stacked spectrafor Mg ii using the three spherical velocity componentsindividually ( r , θ , and φ ), and compare those to thespectrum generated with the full velocity of our fiducialsample of halos. Both the r and θ component spectraare centered at 0 km s − , indicating that over the entiresample they do not contribute any net velocity shift tothe gas along the sightlines. The spectrum of the φ component is remarkably similar to the spectrum of theentire velocity, both in terms of velocity shift and width.This means that for our fiducial sample, the shape of thestacked velocity spectrum along sightlines is completelydetermined by only the φ (i.e. rotational) component ofthe velocity along those sightlines.3.3. Effects of halo mass and resolution on Mg ii inTNG50 We now describe how our main results vary with halomass and mass resolution. To study the effect of halomass, we consider two mass bins containing halos fromTNG50 with 10 M (cid:12) < M halo < . M (cid:12) and 10 M (cid:12) < M halo < . M (cid:12) at z = 1, which areabove and below the fiducial mass range and contain1130 and 167 halos respectively. As in Section 3.1 wecalculate Mg ii equivalent widths and generate velocityspectra which we show in Figure 9. For easier compari-son, we also show the TNG50 fiducial sample.As shown in the left panel of Figure 9, at a given im-pact parameter, the shape of the equivalent width distri-bution changes with halo mass: lower halo masses (cyan)are much more likely to host weak or non-absorbers thanhigher halo masses (magenta), and they are much lesslikely to host strong absorbers. We find this trend tohold at all impact parameters studied in this paper. Wecan see the effect on observability with the vertical linesin this panel, which show the mean equivalent widths ofthe strong absorbers in each mass bin. Typical strongabsorbers in the fiducial sample have only slightly largerequivalent widths than those those at lower halo masses,but are substantially weaker than the strong absorbersat higher halo masses. At larger impact parameters,the mean equivalent widths of all strong absorbers is ≈ . (cid:38) M (cid:12) .Also shown in the left panel of Figure 9 is the equiv-alent width distribution of 4315 halos with the samemass as the fiducial sample from the TNG100 simula-tion, which has a lower baryonic mass resolution thanTNG50 by a factor of ∼
16. Decreasing the simulation0
DeFelippis et al. v / V v i r
20 40 60 80 100250255075100125 p h i v e l o c i t y [ k m / s ] Mg IIT < 3 × 10 KT > 3 × 10 K 0.60.40.20.00.20.40.6 v r / V v i r
20 40 60 80 100r [kpc]7550250255075 r a d i a l v e l o c i t y [ k m / s ] all > 45 < 45 Figure 7.
Mean mass-weighted velocity profiles of thespherical phi-component ( v φ , top) and r-component ( v r , bot-tom) for cold gas (blue), hot gas (red) and Mg ii gas (black)in spherical bins. Velocity is given in km s − and as a frac-tion of the virial velocity. A temperature of 3 × K isused to separate “cold” and “hot” gas. Profiles are shownfor gas in the entire halo (solid), gas with α > ◦ (dotted),and gas with α < ◦ (dashed). Shaded regions show the ± σ -scatter of the solid lines and are of similar size for allprofiles. resolution lowers equivalent widths overall and steepensthe distribution in the same way as decreasing the halomass does, but the effect is weaker. The mean equiva-lent width of strong absorbers is largely unaffected bythe change in resolution.In the right panel of Figure 9 we examine the effectof halo mass and resolution on the observed Mg ii spec-trum of strong absorbers. We note that the spectra ofthe entire samples, as in Figure 6, have the same shapeand center as their corresponding strong absorber sub-set, but are substantially shallower. We also plot thereal velocity rather than the normalized velocity to em-phasize the difference in equivalent widths, which canbe more easily read off.We see that the fiducial and lower-mass bins haveremarkably similar spectra: they are both symmetricand centered at moderate positive velocities. The spec-trum of the higher-mass bin is markedly different: it vir sin(i))0.40.50.60.70.80.91.0 b = 20, = 5 , i = 60 v tot vv r v Figure 8.
Stacked Mg ii velocity spectra for the full fiducialTNG50 sample at a single sightline. The contributions ofthe three spherical components of velocity – v r (dotted red), v φ (dashed green) and v θ (dot-dashed blue) – are shown, aswell as the spectrum created from the total velocity (solidblack). is much broader, asymmetric, and centered at a signifi-cantly higher velocity. It still, however, shows a prefer-ence for Mg ii gas to be corotating. We note that the dif-ference between Figure 9 as shown and the correspond-ing velocity-normalized spectrum (not shown) is that thenormalized higher-mass spectrum is compressed slightlyand therefore appears more similar to the normalizedfiducial spectrum. Additionally, while the lower-massand fiducial spectra are both centered at ≈ . V vir , thehigher-mass spectra is peaked at ≈ V vir . Higher halomasses ( (cid:38) M (cid:12) ) thus have substantially more Mg ii absorption and more complex kinematic signatures thanfor the halo masses of the fiducial sample and lower.Finally, we consider the difference that resolutionmakes in the Mg ii absorption spectrum. As was thecase with equivalent widths, the difference caused byresolution is smaller than the difference caused by ei-ther increasing or decreasing the halo mass. Apart froma slight change in the depth of the spectrum, the kine-matic properties of strong absorbers in TNG are essen-tially resolution independent (see solid vs. dotted curvesin Figure 9 for TNG50 and TNG100 respectively). Theeffect of increasing the resolution of the simulation istherefore primarily to increase the occurrence of strongabsorbers at a given halo mass. DISCUSSION4.1.
The Role of Mg ii in TNG We consider here the ramifications of the detailedanalysis of Mg ii in TNG from Section 3. In Figure 7,we found that Mg ii gas is very well approximated by a ircumgalactic Mg ii in TNG and MEGAFLOW [Å]10 b = 20 kpc, = 5 , i = 60
200 100 0 100 200v [km/s]0.20.30.40.50.60.70.80.91.0 b = 20 kpc, = 5 , i = 60
TNG50 fiducialTNG100 fiducialTNG50 lower massTNG50 higher mass
Figure 9.
Left: rest-frame equivalent width distribution of the TNG50 fiducial sample (solid black), lower-mass halos with10 M (cid:12) < M halo < . M (cid:12) (cyan), higher-mass halos with 10 M (cid:12) < M halo < . M (cid:12) (magenta) and the same masshalos from TNG100 (dotted black) at the same sightline of b = 20 kpc, α = 5 ◦ , and i = 60 ◦ . The mean EW of the strongabsorbers in each halo mass bin is shown with a translucent vertical line of the same color. Right: Stacked velocity spectra ofthe same halo samples with velocities in km s − . simple temperature cut. Therefore, we expect the angu-lar momentum of cold gas in the CGM of TNG galaxiesshould be very similar to that of Mg ii . DeFelippis et al.(2020) found cold CGM gas in halos of this mass rangeand redshift to have higher angular momentum whensurrounding high-angular momentum galaxies, meaningMg ii is likely tracing high-angular momentum gas in theCGM of these halos. As the velocity spectrum’s centerand shape is almost completely set by the rotationalvelocity component (see Figure 8), it should thereforebe possible to use Mg ii velocity spectra from sightlinesnear the major axis to estimate the angular momentumof cold gas in the CGM.In Section 3.3 we examined possible halo mass and res-olution dependencies of our results with two main goalsin mind: to establish any broad effects of the TNG feed-back model on Mg ii , and to determine to what extentthe cosmological simulation can capture Mg ii kinemat-ics. Feedback is known to be important for regulatinggas flows into, out of, and around galaxies, and thereforecould have observable signatures in the Mg ii spectra, es-pecially at different halo masses. The results of the halomass analysis suggest that for halos with masses between10 M (cid:12) and 10 M (cid:12) , the physical mechanisms affect-ing their CGM are similar enough to result in Mg ii spec-tra that essentially scale with the halo’s virial velocity.This is presumably because feedback from supernovaeis the dominant form of feedback that affects the CGMfor all halo masses below ∼ M (cid:12) and produces Mg ii gas with similar kinematic signatures. For halos above10 M (cid:12) however, Mg ii gas has stronger overall absorp-tion, as reflected by their flatter EW distribution, and substantially larger velocities and velocity dispersions,as reflected by their very broad velocity spectra. Thisis likely due to the dominant form of feedback switch-ing from stars to AGN around this halo mass. However,within the higher-mass sample, halos with larger blackhole masses do not themselves have broader Mg ii spec-tra, so there is probably a combination of effects whichresult in a noticeable difference in the properties of thespectrum at higher masses.Nelson et al. (2020) has recently used TNG50 to studythe origin of cold Mg ii gas in the CGM of very massive( M ∗ (cid:38) M (cid:12) ) galaxies and found structures of size afew × pc that are sufficient to explain the observedcovering fractions and LOS kinematics. They also notethat while some fundamental properties like the num-ber of cold gas clouds present in halos are not con-verged at TNG50’s resolution, the total cold gas massof such halos is converged in TNG50. This supportsour findings that our kinematic results do not qualita-tively change even going from TNG50 to TNG100, afactor of ∼
16 in mass resolution (Figure 9), becausethe majority of the Mg ii mass is already in the halo byTNG50’s resolution. We expect higher resolution simu-lations to produce more strong absorbers at a given halomass but the rotation of Mg ii near the major axis ap-pears to be a resolution-independent aspect of the CGMfor MEGAFLOW analogs in the TNG simulations.Finally, in Figure 10, we show the specific angularmomentum ( j ) of different halo components as a func-tion of stellar mass of their central galaxies, with thegoal of contextualizing the angular momentum of Mg ii gas (black line) in the CGM in relation to the rest of2 DeFelippis et al. stars l o g j [ k m / s × k p c ] cold CGMhot CGMMgII CGM starsdark matter megaflowfiducialsample Figure 10.
Median specific angular momentum vs. galacticstellar mass for the cold (blue), hot (red), and Mg ii (black)CGM as defined in Figure 7, as well as the dark matterhalo (dotted orange) and the stellar component of the galaxy(purple) at z = 1. Unlike previous figures, medians are cal-culated using a sample of all halos containing central galaxieswith stellar masses 10 M (cid:12) < M ∗ < M (cid:12) . Shaded re-gions show the 16th and 84th percentiles of the distributionsof the Mg ii gas (black), which is similar in size to all com-ponents except dark matter (orange) which has noticeablylarger scatter. Black points show the Mg ii specific angularmomentum of the halo mass-selected fiducial sample whichis biased towards higher j for M ∗ (cid:46) . . Green squaresshow estimations for the specific angular momentum of theMEGAFLOW absorbers using inferred rotational velocitiesfrom Zabl et al. (2019). the gas in the CGM as well as to the other compo-nents of the halo. The slope of this j − M ∗ relation forthe stellar component of galaxies (purple line) is ∼ . j than that of the dark matter by ∼ . ii traces the angular momentum of the both thecold and hot components of the CGM quite well.Also shown in Figure 10 are two sets of points repre-senting Mg ii gas in individual halos: the fiducial samplein black and the Zabl et al. (2019) sample in green, forwhich j was estimated using their derived rotational ve-locities. The two are not directly comparable since thepoints from Zabl et al. (2019) represent Mg ii gas along a single sightline, yet they are still able to reproduce thescatter in this relation found in TNG50, though they aresomewhat biased towards higher j . This bias is likelydue to the selection in Zabl et al. (2019) of strong Mg ii absorption near the major axis, which is where high- j cold gas tends to reside in the CGM as shown in De-Felippis et al. (2020). Nevertheless, from Figure 10 wecan conclude that estimations of the angular momen-tum content of the CGM provided by single sightlinesof Mg ii can get within ∼ . Comparisons to Recent Work
We now highlight results from previous work on Mg ii absorption in observations and simulations in the con-text of our results. Observations of Mg ii using sightlinesnear the major axis of galaxies have generally found thatgas is corotating with the galaxy both for small impactparameters of <
15 kpc (e.g. Bouch´e et al. 2016) andlarge impact parameters of >
50 kpc (e.g. Martin et al.2019). Using a lensed system, Lopez et al. (2020) ob-served multiple sightlines of the same CGM and mea-sured a decreasing Mg ii rotation curve that is quali-tatively similar to Figure 7. However, their absorptiondata only goes out to ≈
30 kpc. Our work suggests Mg ii rotation curves should continue to decrease to at least100 kpc, though based on the maps in Figure 3 the Mg ii column densities at those distances are significantly be-low current observational limits.While this paper is focused on Mg ii gas near the ma-jor axis, there are also recent results suggesting Mg ii outflows along the minor axis of galaxies with velocities >
100 km s − (e.g. Schroetter et al. 2019; Zabl et al.2020). It is worth noting though that Mortensen et al.(2020) found a lensed system with Mg ii on the geomet-ric minor axis of the absorber galaxy with LOS velocities <
100 km s − and a large velocity dispersion, indicatingthat the kinematics of Mg ii outflows may vary signifi-cantly. We showed in Figure 7 that the TNG fiducialsample does not display large mean radial velocities forMg ii gas in regions of the CGM around the minor axis.However, we find (but do not show) that there are Mg ii outflows with velocities upwards of ∼
50 km s − that arepresent throughout the CGM, especially near the minoraxis, but at all radii there is less outflowing Mg ii gasthan inflowing Mg ii gas. It is therefore likely that haloswith very fast Mg ii outflows in their CGM are rare inTNG50, but properly analyzing such kinematic signa-tures and comparing to observations of Mg ii outflows isbeyond the scope of this paper.Ho et al. (2020) recently studied similar aspects ofMg ii absorption in the EAGLE simulation at z ≈ . ircumgalactic Mg ii in TNG and MEGAFLOW ii structure aroundstar-forming galaxies as well as a lower detection frac-tion of Mg ii near the minor axis. They also find thathigher-mass galaxies host detectable (i.e. above a fixedcolumn density) Mg ii structures out to larger distancesin the CGM, which we indirectly show with the EWdistributions in Figure 9, where higher-mass halos havemore strong absorbers. SUMMARYWe have simulated Mg ii absorption in the CGM ofhalos from TNG50 comparable to those observed in theZabl et al. (2019) subset of MEGAFLOW and comparedabsorption and kinematic properties of the two samples.We also examined the 3D kinematics of the Mg ii inTNG50. Our conclusions are as follows:1. The equivalent widths of absorber-selected halos(i.e. strong absorbers) from TNG50 match reason-ably well with the equivalent widths of major-axissightlines from Zabl et al. (2019) (Figure 4).2. A majority of halos are strong absorbers at thesmallest impact parameter studied (15 kpc), butthe strong absorber fraction drops quickly as afunction of distance (Figure 4).3. The stacked velocity spectra of TNG50 strong ab-sorbers match the stacked spectra of Zabl et al.(2019) very well, thus supporting the physicalinterpretation of corotation both below 30 kpc,where the spectra are strongly peaked near ∼ . V vir and symmetric, and above 30 kpc, wherethe spectra are similarly peaked but are much nois-ier, broader, and asymmetric (Figure 6).4. In TNG50, Mg ii gas has velocity profiles nearlyidentical to gas below a temperature cutoff of3 × K, meaning Mg ii absorption is a goodproxy for cold gas kinematics in general. There issubstantial rotation and typical inflow velocities of ∼
50 km s − out to ∼
50 kpc in the CGM (Figure7).5. The radial and polar velocity components bythemselves do not cause any net velocity shift inthe stacked spectrum, which implies that Mg ii ab-sorption kinematics alone cannot be used to mea-sure typical inflow speeds of rotating gas in theCGM. (Figure 8). 6. Mg ii absorption strengths and spectra arestronger and broader for halos more massive thanthe fiducial sample of 10 . − M (cid:12) halos butdo not change very much for halos less massivethan the fiducial sample. Lowering the resolutionfrom TNG50 to TNG100 only modestly changesany of the Mg ii kinematic properties (Figure 9).7. The median specific angular momentum of theMg ii component of the CGM as a function ofgalactic stellar mass is very similar to that of bothcold and hot CGM gas, and it is larger than that ofthe dark matter halo and the stars in the galaxyby ∼ . ∼ . ii from the Zabl et al. (2019) data are also reason-ably close to the values from TNG50 to within afactor of ∼ . ii observations from TNG50 generates absorption spectrathat are comparable to real data. In particular, our re-sults are consistent with the emerging picture of rotatingMg ii gas found in observations and also other simula-tions. In future work, we plan to widen our investiga-tion to include other ions that trace warmer and morediffuse gas, as well as follow gas at particular redshiftsbackwards and forwards through time to determine thestability of various ion structures and their role in trans-porting angular momentum to or from the galaxy.ACKNOWLEDGMENTSWe thank Johannes Zabl, Edouard Tollet, JoakimRosdahl and Jeremy Blaizot for insightful and use-ful discussions, as well as Cameron Hummels for as-sistance with Trident . D.D. acknowledges supportfrom the Chateaubriand Fellowship of the Office forScience & Technology of the Embassy of France inthe United States. N.B. acknowledges funding supportfrom the French Agence National de Recherche (ANR)grant ‘3DGasFlows’ (ANR-17-CE31-0017). Flatiron In-stitute is supported by the Simons Foundation. GLB ac-knowledges financial support from the NSF (grant AST-1615955, OAC-1835509) and computing support fromNSF XSEDE. F.M. acknowledges support through theProgram “Rita Levi Montalcini” of the Italian MUR.
Software:
Trident (Hummels et al. 2017), yt (Turket al. 2011), Cloudy (Ferland et al. 2013),
NumPy (vander Walt et al. 2011),
Matplotlib (Hunter 2007), and
IPython (Perez & Granger 2007)4
DeFelippis et al.
REFERENCES
Bacon, R., Bauer, S., Boehm, P., et al. 2006, in Society ofPhoto-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 6269, Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series, ed.I. S. McLean & M. Iye, 62690J, doi: 10.1117/12.669772Behroozi, P. S., Conroy, C., & Wechsler, R. H. 2010, ApJ,717, 379, doi: 10.1088/0004-637X/717/1/379Bergeron, J., & Boiss´e, P. 1991, A&A, 243, 344Bergeron, J., Cristiani, S., & Shaver, P. A. 1992, A&A, 257,417Bordoloi, R., Lilly, S. J., Knobel, C., et al. 2011, ApJ, 743,10, doi: 10.1088/0004-637X/743/1/10Bouch´e, N., Hohensee, W., Vargas, R., et al. 2012,MNRAS, 426, 801, doi: 10.1111/j.1365-2966.2012.21114.xBouch´e, N., Murphy, M. T., Kacprzak, G. G., et al. 2013,Science, 341, 50, doi: 10.1126/science.1234209Bouch´e, N., Finley, H., Schroetter, I., et al. 2016, ApJ, 820,121, doi: 10.3847/0004-637X/820/2/121Bryan, G. L., & Norman, M. L. 1998, ApJ, 495, 80,doi: 10.1086/305262Burchett, J. N., Rubin, K. H. R., Prochaska, J. X., et al.2020, arXiv e-prints, arXiv:2005.03017.https://arxiv.org/abs/2005.03017Burchett, J. N., Tripp, T. M., Prochaska, J. X., et al. 2019,ApJL, 877, L20, doi: 10.3847/2041-8213/ab1f7fCoil, A. L., Blanton, M. R., Burles, S. M., et al. 2011, ApJ,741, 8, doi: 10.1088/0004-637X/741/1/8Corlies, L., Peeples, M. S., Tumlinson, J., et al. 2020, ApJ,896, 125, doi: 10.3847/1538-4357/ab9310Danovich, M., Dekel, A., Hahn, O., Ceverino, D., &Primack, J. 2015, MNRAS, 449, 2087,doi: 10.1093/mnras/stv270DeFelippis, D., Genel, S., Bryan, G. L., & Fall, S. M. 2017,ApJ, 841, 16, doi: 10.3847/1538-4357/aa6dfcDeFelippis, D., Genel, S., Bryan, G. L., et al. 2020, ApJ,895, 17, doi: 10.3847/1538-4357/ab8a4aDutta, R., Fumagalli, M., Fossati, M., et al. 2020, MNRAS,499, 5022, doi: 10.1093/mnras/staa3147Emerick, A., Bryan, G. L., & Mac Low, M.-M. 2019,MNRAS, 482, 1304, doi: 10.1093/mnras/sty2689Fall, S. M., & Romanowsky, A. J. 2013, ApJL, 769, L26,doi: 10.1088/2041-8205/769/2/L26Faucher-Gigu`ere, C.-A., Lidz, A., Zaldarriaga, M., &Hernquist, L. 2009, ApJ, 703, 1416,doi: 10.1088/0004-637X/703/2/1416Ferland, G. J., Porter, R. L., van Hoof, P. A. M., et al.2013, RMxAA, 49, 137. https://arxiv.org/abs/1302.4485 Ho, S. H., Martin, C. L., Kacprzak, G. G., & Churchill,C. W. 2017, ApJ, 835, 267,doi: 10.3847/1538-4357/835/2/267Ho, S. H., Martin, C. L., & Schaye, J. 2020, ApJ, 904, 76,doi: 10.3847/1538-4357/abbe88Ho, S. H., Martin, C. L., & Turner, M. L. 2019, ApJ, 875,54, doi: 10.3847/1538-4357/ab0ec2Hummels, C. B., Smith, B. D., & Silvia, D. W. 2017, ApJ,847, 59, doi: 10.3847/1538-4357/aa7e2dHummels, C. B., Smith, B. D., Hopkins, P. F., et al. 2019,ApJ, 882, 156, doi: 10.3847/1538-4357/ab378fHunter, J. D. 2007, Computing in Science Engineering, 9,90, doi: 10.1109/MCSE.2007.55Lan, T.-W. 2020, ApJ, 897, 97,doi: 10.3847/1538-4357/ab989aLan, T.-W., & Mo, H. 2018, ApJ, 866, 36,doi: 10.3847/1538-4357/aadc08Li, F., Rahman, M., Murray, N., et al. 2021, MNRAS, 500,1038, doi: 10.1093/mnras/staa3322Lopez, S., Tejos, N., Barrientos, L. F., et al. 2020, MNRAS,491, 4442, doi: 10.1093/mnras/stz3183Marinacci, F., Vogelsberger, M., Pakmor, R., et al. 2018,MNRAS, 480, 5113, doi: 10.1093/mnras/sty2206Martin, C. L., Ho, S. H., Kacprzak, G. G., & Churchill,C. W. 2019, ApJ, 878, 84, doi: 10.3847/1538-4357/ab18acMortensen, K., Keerthi Vasan G., C., Jones, T., et al. 2020,arXiv e-prints, arXiv:2006.00006.https://arxiv.org/abs/2006.00006Naiman, J. P., Pillepich, A., Springel, V., et al. 2018,MNRAS, 477, 1206, doi: 10.1093/mnras/sty618Nelson, D., Pillepich, A., Springel, V., et al. 2018, MNRAS,475, 624, doi: 10.1093/mnras/stx3040—. 2019, MNRAS, 490, 3234, doi: 10.1093/mnras/stz2306Nelson, D., Sharma, P., Pillepich, A., et al. 2020, MNRAS,498, 2391, doi: 10.1093/mnras/staa2419Nielsen, N. M., Churchill, C. W., Kacprzak, G. G., &Murphy, M. T. 2013, ApJ, 776, 114,doi: 10.1088/0004-637X/776/2/114Nielsen, N. M., Churchill, C. W., Kacprzak, G. G., Murphy,M. T., & Evans, J. L. 2015, ApJ, 812, 83,doi: 10.1088/0004-637X/812/1/83Peeples, M. S., Corlies, L., Tumlinson, J., et al. 2019, ApJ,873, 129, doi: 10.3847/1538-4357/ab0654Perez, F., & Granger, B. E. 2007, Computing in ScienceEngineering, 9, 21, doi: 10.1109/MCSE.2007.53P´eroux, C., Nelson, D., van de Voort, F., et al. 2020,MNRAS, 499, 2462, doi: 10.1093/mnras/staa2888Pillepich, A., Nelson, D., Hernquist, L., et al. 2018,MNRAS, 475, 648, doi: 10.1093/mnras/stx3112 ircumgalactic Mg ii in TNG and MEGAFLOW15