aa r X i v : . [ a s t r o - ph . GA ] F e b Astronomy & Astrophysicsmanuscript no. Barstng6 © ESO 2021February 8, 2021
Bar-like galaxies in IllustrisTNG
Ewa L. Łokas
Nicolaus Copernicus Astronomical Center, Polish Academy of Sciences, Bartycka 18, 00-716 Warsaw, Polande-mail: [email protected]
February 8, 2021
ABSTRACT
We study a sample of bar-like galaxies in the Illustris TNG100 simulation, in which almost the whole stellar component is in theform of a prolate spheroid. The sample is di ff erent from the late-type barred galaxies studied before. In addition to the requirementof a high enough stellar mass and resolution, the 277 galaxies were selected based on the single condition of a low enough ratioof the intermediate to long axis of the stellar component. We followed the mass and shape evolution of the galaxies as well as theirinteractions with other objects and divided them into three classes based on the origin of the bar and the subsequent history. In galaxiesof class A (comprising 28% of the sample), the bar was induced by an interaction with a larger object, most often a cluster or groupcentral galaxy, and the galaxies were heavily stripped of dark matter and gas. In classes B and C (27% and 45% of the sample,respectively) the bars were induced by a merger or a passing satellite, or they were formed by disk instability. Class B galaxies werethen partially stripped of mass, while those of class C evolved without strong interactions, thus retaining their dark matter and gas inthe outskirts. We illustrate the properties of the di ff erent classes with three representative examples of individual galaxies. In spite ofthe di ff erent evolutionary histories, the bars are remarkably similar in strength, length, and formation times. The gas fraction in thebaryonic component within two stellar half-mass radii at the time of bar formation is always below 0.4 and usually very low, whichconfirms in the cosmological context the validity of this threshold, which has previously been identified in controlled simulations.Observational counterparts of these objects can be found among early-type fast rotators, S0 galaxies, or red spirals with bars. Key words. galaxies: evolution – galaxies: interactions – galaxies: kinematics and dynamics – galaxies: structure – galaxies: clusters:general
1. Introduction
The origin of galaxy morphology to some extent remains anopen question. This is especially true for barred galaxies orfor those with an elongated stellar component. In general, late-type spiral galaxies with central bars are classified as barred(Buta et al. 2015), but early-type objects with similarly elon-gated components exist as well (Baillard et al. 2011). At leasttwo di ff erent formation mechanism for bars have been identi-fied. They can be created either by an inherent instability of thinenough disks (Hohl 1971; Ostriker & Peebles 1973), moderatedby the presence of dark matter halos (Athanassoula 2003), orthrough tidal interactions with other galaxies (Noguchi 1987;Gerin et al. 1990; Miwa & Noguchi 1998; Berentzen et al. 2004;Lang et al. 2014; Łokas et al. 2014; Łokas 2018) or clusters(Mastropietro et al. 2005; Łokas et al. 2016). The latter mech-anism tends to produce bar-like rather than barred galaxies inthe sense that almost the whole stellar component is transformedfrom a disk to a prolate spheroid, and the outer stars may in ad-dition be tidally stripped.The mechanism of tidal bar formation has so far mostly beeninvestigated in detail in controlled simulations of tidal interac-tions between galaxies and of galaxies orbiting a cluster. It hasbeen demonstrated that the mechanism is extremely e ff ective ininducing the bar-like shape when the pericenter of the galaxyorbit in a cluster or an impact parameter of a fly-by encounterwith another galaxy is small enough (Łokas et al. 2014, 2016;Łokas 2018; Gajda et al. 2017). The tidal force exerted by thecompanion is then su ffi ciently strong to temporarily distort thedisk and trigger the permanent change of stellar orbits into more radial shapes, thus leading to the formation of a stable bar. Ifthe outer part of the disk is additionally stripped and lost, anelongated stellar component remains. Further evolution and sub-sequent tidal shocks at the following pericenters may shorten thelength of the bar to make it more spherical. Because in the clus-ter environment the gas is also lost due to ram-pressure stripping,the remains appear like a red elliptical galaxy.This scenario has recently been placed in the cosmologicalcontext (Łokas 2020b) by using the IllustrisTNG simulationsof galaxy formation (Springel et al. 2018; Marinacci et al. 2018;Naiman et al. 2018; Nelson et al. 2018; Pillepich et al. 2018).These simulations allow following the evolution of dark matterand baryons in a su ffi ciently large region and with good enoughresolution to study properties of single galaxies. It has beendemonstrated that the simulations are able to reproduce di ff erentobserved properties of galaxies, in particular, their morphologies(Nelson et al. 2018; Genel et al. 2018; Rodriguez-Gomez et al.2019).Łokas (2020b) demonstrated that the evolution of galaxies ina cluster leads to the formation of tidally induced bars. A fewconvincing examples of galaxies were presented in which theformation of a bar coincides with a pericenter passage on an or-bit around the cluster, and in another few galaxies, the bars mighthave been enhanced by the interaction with the cluster. In addi-tion, these galaxies were ram-pressure stripped of their gas andhad a substantially reduced dark matter fraction compared to ob-jects that interacted less strongly. However, some galaxies withvery similar properties were also indentified that had never inter-acted with the cluster. One such case has recently been describedin Łokas (2020c). Here the bar was induced by a merger and a Article number, page 1 of 10 & Aproofs: manuscript no. Barstng6 selected galaxiesremaining galaxies0.0 0.2 0.4 0.6 0.8 1.00.00.20.40.60.81.0 b / a c / a / a T / a f / a f / a T T Fig. 1.
Selection of the galaxy sample. The six panels show the properties of the selected galaxies in di ff erent parameter combinations: the axisratios b / a and c / a , the triaxiality parameter T , and the rotation parameter f . Black points correspond to the selected galaxies, and the green pointsshow the remaining galaxies of the total of 6507 objects. passing satellite, and the gas and dark matter were stripped by asubsequent interaction with a group of galaxies. This shows thatbar-like objects can originate in a variety of scenarios.In this work we study the whole population of bar-like galax-ies in the Illustris TNG100 simulation and investigate their pos-sible formation scenarios. We show that the galaxies have a va-riety of properties. Some galaxies are dominated by dark mat-ter and still contain gas; these have not interacted strongly withany neighbors in the past. At the same time, they are not barsembedded in disks; these have been studied previously using Il-lustris and IllustrisTNG simulations (Peschken & Łokas 2019;Rosas-Guevara et al. 2020; Zhou et al. 2020; Zhao et al. 2020).In order to distinguish them from the previously studied objects,we refer to these galaxies as bar-like rather than barred. In sec-tion 2 we describe our sample of selected galaxies. In section 3we discuss their evolutionary histories, dividing them into threesubsamples, and we present three fiducial examples in more de-tail. Section 4 describes the properties of the bars in detail, andthe discussion follows in section 5.
2. Sample selection
For the purpose of this work, we used the publicly available sim-ulation data from IllustrisTNG described by Nelson et al. (2019).In order to have a su ffi cient number of objects in the sample andenough resolution in each object, we chose the TNG100 run.We selected the galaxies at z = M ⊙ ,which translates into about 10 stellar particles per object andthus makes the morphological analysis possible. This require- ment is met by 6507 objects in the final output of the IllustrisTNG100 simulation. Next, we imposed a single condition: theintermediate to longest axis ratio b / a of the stellar componenthad to be lower than 0.6. For these values we used (and re-produced) the measurements based on the mass tensor of b / a within two stellar half-mass radii, 2 r / , provided by the Illustristeam in the Supplementary Data Catalogs of stellar circularities,angular momenta, and axis ratios and calculated as describedin Genel et al. (2015). The axis ratios were estimated from theeigenvalues of the mass tensor of the stellar mass obtained byaligning each galaxy with its principal axes and calculating threecomponents ( i = M i = ( Σ j m j r j , i / Σ j m j ) / , where j enu-merates over stellar particles, r j , i is the distance of stellar particle j in the i -axis from the center of the galaxy, and m j is its mass.The eigenvalues were sorted so that M < M < M , whichmeans that the shortest to longest axis ratio is c / a = M / M ,while the intermediate to longest axis ratio is b / a = M / M .The sample of galaxies with b / a < . c / a < b / a ,where a , b , and c are the longest, intermediate, and shortest axis,respectively. The properties of the selected galaxies in compari-son to the whole sample are shown in Fig. 1. In the six panels ofthe figure we plot the positions of the galaxies in di ff erent planesof parameters: the axis ratios b / a , c / a , the triaxiality parame-ter T = [1 − ( b / a ) ] / [1 − ( c / a ) ], and the rotation parameter, f . The rotation parameter is defined as the fractional mass of The catalogs are available from the webpage https: // / data / docs / specifications / all stars with circularity parameter ǫ > .
7, where ǫ = J z / J ( E ) , and J z is the specific angular momentum of the star along theangular momentum of the galaxy, while J ( E ) is the maximumangular momentum of the stellar particles at positions between50 before and 50 after the particle in question in a list where thestellar particles are sorted by their binding energy (Genel et al.2015).In the three upper panels of Fig. 1 the axis ratio b / a lies inthe horizontal axis. The panels illustrate the selection with thesimple cuto ff at b / a < .
6. In the lower three panels the posi-tions of the selected galaxies are non-trivial, and all the selectedgalaxies have T > .
74 and f < . , which means that they areindeed prolate and do not rotate fast. We note that T > / ffi cient to apply this condition for the sample selectionbecause galaxies with high triaxiality can still have high valuesof the axis ratios and therefore can be close to spherical. We alsonote that f < . f g = M g / ( M g + M ∗ ). The solid black lines indicate where thedark masses are equal to stellar, M dm = M ∗ (lower line) or tentimes higher, M dm = M ∗ (upper line), as well as a generalexpectation from the abundance matching for central galaxies(Behroozi et al. 2013). The dark masses are higher than the stel-lar masses for most galaxies in the sample, although a substan-tial number of galaxies (34) contains less dark matter than stars.All the galaxies with M dm < M ∗ contain very little or no gas,while galaxies above the threshold, with M dm > M ∗ , typicallycontain a significant amount of gas, with the highest gas fractionof 0.87. Overall, 151 out of 277 (55%) contain gas at present,and the remaining 126 galaxies have no gas at all. We note thatthe gas distribution is more extended than the stars, and in partic-ular, the gas fraction within two stellar half-mass radii at presentis much lower, always below 0.12, and higher than 0.01 only forfive galaxies.In the lower panel of Fig. 2 we plot the ratio of the presentdark matter mass that is bound to the galaxy to the maximumdark matter mass that the galaxy possessed throughout its his-tory, M dm ( z = / M dm , max , versus the present stellar mass, M ∗ ( z = M dm ( z = / M dm , max = , whichmeans that they reach their maximum dark masses at present andare therefore still accreting mass from their neighborhood. M d m = M * M d m = M * A M * ( z = ) [ M ⊙ ] M d m ( z = ) [ M ⊙ ] * ( z = ) [ M ⊙ ] M d m ( z = ) / M d m , m a x Fig. 2.
Masses of the galaxies in the sample. Upper panel: Dark mass vs.stellar mass of the galaxies at present. The black lines mark the relationsof M dm = M ∗ , M dm = M ∗ and the prediction from the abundancematching (AM). Lower panel: Ratio of the present dark mass to themaximum dark mass as a function of stellar mass. In both panels thepoints are colored according to the galaxy gas fraction.
3. Evolutionary histories with three fiducialexamples
The di ff erent mass and gas content properties of the galaxiessuggest that their evolutionary histories were varied. To explorethem, we traced the evolution of the di ff erent mass componentsand other properties in time for each galaxy. The properties thatprovide most information on the formation of the bar-like shapeinclude the evolution of the axis ratios and the triaxiality pa-rameter as well as the rotation parameter discussed in the pre-vious section. In addition to these, we also calculated the evo-lution of the commonly used measure of the strength of thebar (Athanassoula & Sellwood 1986; Athanassoula & Misiriotis2002; Athanassoula et al. 2013) in the form of the m = A ( R ) = | Σ j m j exp(2 i θ j ) | / Σ j m j , where θ j is the azimuthal angle Article number, page 3 of 10 & Aproofs: manuscript no. Barstng6
Table 1.
Median parameter values for di ff erent samples of bar-like galaxies. Sample
N M dm / M ∗ r / [kpc] g − r [mag] f max A , max R max [kpc] R max / r / t f [Gyr] f g ( < r / ) at t f Total 277 10.9 2.60 0.79 0.60 0.61 2.37 1.11 6.82 0.030A 77 1.4 2.53 0.81 0.56 0.58 2.15 1.16 7.13 0.016B 74 5.7 2.60 0.79 0.60 0.61 2.39 1.14 6.21 0.076C 126 35.4 2.65 0.78 0.61 0.62 2.48 1.05 6.91 0.027of the j th star, m j is its mass, and the sum goes up to the numberof particles in a given radial bin. The measurements were firstmade for stars within two stellar half-mass radii, 2 r / . The for-mation of the bar-like shape should be visible as a decrease in b / a , increase in T , and decrease in rotation parameter f (due tothe transition from circular to more radial orbits in the bar). Themost direct signature of the bar formation is the increase in barmode A , however.In addition to tracing the evolution of these parameters, wealso verified whether the formation of the bar could be due tosome external factor, such as a merger or tidal interaction withanother object. For this purpose we used the SubLink mergertrees of the subhalos (Rodriguez-Gomez et al. 2015) that are pro-vided together with the IllustrisTNG data, searched for neigh-bors within 500 kpc of a given galaxy and determined the rela-tive distance between the galaxy and the neighbor as a functionof time. This analysis, and the inspection of evolutionary histo-ries such as those shown in Fig. 3 below for three fiducial ex-amples, led us to conclude that the sample of bar-like galaxiescan be divided into three subsamples based on their evolutionand the origin of the bar-like shape. In the following we refer tothese samples as classes A, B, and C.We assigned galaxies to class A (comprising 77 objects) ifthe transition from an oblate to a prolate shape (i.e., the forma-tion of the bar, see section 4) coincides with a pericenter pas-sage around a more massive companion, which suggests thattheir bars were tidally induced by an interaction with a largergalaxy, more often a central galaxy of a cluster or group (64 ob-jects) than another galaxy (13 objects). Such galaxies lost mostof their dark matter mass and all the gas, and they contain no gasat present (except for two objects with very low gas fractions).The galaxies were assigned to class B (74 objects) if theirbar had formed di ff erently, by mergers or passing satellites, orjust by a bar instability, but they interacted later with a moremassive object and lost a significant amount of dark mass (morethan 50% of the maximum dark mass). During such interactions,these objects are also often stripped of their gas so that most ofthem (51 out of 74) are completely gas-free at present, while theremainder retain small amounts of gas (with gas fractions up to0.16).The galaxies were included in class C (126 objects) if theydid not show any signatures of strong interactions with moremassive objects and did not lose more than 50% of the maxi-mum dark mass. In contrast to members of classes A and B, theytypically grow in mass and continue to merge with small satel-lites until the present time. In these objects the bars form as inclass B by mergers or passing satellites or through a bar instabil-ity of the disk. All the objects of this class retain some gas untilthe end, with gas fractions between 0.01 and 0.87.Classes A and C can be understood as opposite in terms ofthe strength of the interactions they experienced and their e ff ecton morphology. In the case of class A, the interactions were es-pecially strong and resulted in the morphological transformationfrom a disk to a prolate spheroid in addition to the very strongmass loss. Class C galaxies instead evolved rather in isolation, steadily growing in mass or experiencing only very mild tidale ff ects from more massive structures, and their morphologicaltransformation was not induced by such e ff ects. The galaxies ofclass B are intermediate between A and C in the sense that theirtidal interactions were significant in terms of mass loss, but notdirectly related to the morphological transformation. This pic-ture is confirmed by verifying the classification of the galaxiesas centrals or satellites given in the Subfind catalogs of the simu-lation. While almost all galaxies of classes A and B (except one)are classified as satellites, most objects of class C (110 out of126, or 87%) are centrals.Examples of the evolution of galaxies belonging to samplesA, B, and C are shown in the columns of Fig. 3. The galaxieswere named using their identification numbers in the subhalocatalogs and selected as the most convincing and clear represen-tative cases of a given class. The first two rows of the panels plotthe evolution of the total masses in three di ff erent components(dark, stellar, and gas) in di ff erent mass scales. The median val-ues of the ratio of dark to stellar masses at the end of evolutionfor di ff erent samples are listed in the third column of Table 1.The next two rows of panels in Fig. 3 display the evolution ofthe shape parameters b / a , c / a and T as well as the rotation pa-rameter f and the bar strength A . The last row of the panelsshows the distance of the galaxy from the neighbors that moststrongly a ff ected its evolution.Galaxy ID76122 (left column of Fig. 3), classified as belong-ing to subsample A, grew in mass until about 6 Gyr, the time offirst pericenter passage around the massive galaxy ID76086, acentral of a cluster, which caused strong mass stripping and in-duced the bar, as indicated by a sudden growth of the triaxiality T and the bar mode A , as well as by a decrease in b / a and ro-tation f around that time. The second tight pericenter at about10 Gyr led to further mass loss but left the bar-like shape intact,while the amount of rotation was further decreased. The gas islost completely at 9.2 Gyr, when the galaxy approached the sec-ond pericenter. The third pericenter is more distant and has littlee ff ect on the galaxy properties. The final dark mass of the galaxyis only 2.4 times higher than its stellar mass.The bar in galaxy ID225157 (middle column of Fig. 3), be-longing to class B, was most probably induced by a merger witha subhalo with ten times lower mass that occurred at 6 Gyr. Itlater entered an orbit around the larger galaxy ID225156 (thesecond largest in the cluster), and the first percenter was at 8.5Gyr, although it was not very tight. This and the second, tighterpericenter caused enough stripping to rid the galaxy of most ofits dark matter and gas. Its gas fraction at the end is 0.14 andthe dark matter mass is 6.4 times higher than the stellar mass.Another example of this class of objects, ID44, was discussed indetail in Łokas (2020c).The bar in the galaxy ID457361 (right column of Fig. 3),of class C, was most probably induced by an orbiting satellitewith a ten times lower mass that merged with it at about 8 Gyr.This galaxy did not experience any significant interactions withlarger objects during its lifetime and grew steadily in mass untilthe present time by accretion of small subhalos. At the end of the Article number, page 4 of 10wa L. Łokas: Bar-like galaxies in IllustrisTNG dark matterstarsgasID761220 2 4 6 8 10 12 140.00.10.20.30.40.50.60.7 time [ Gyr ] M [ M ⊙ ] [ Gyr ] M [ M ⊙ ] b / a c / a T0 2 4 6 8 10 12 140.00.20.40.60.81.0 time [ Gyr ] b / a , c / a , T f A [ Gyr ] f , A ID76122 - ID760860 2 4 6 8 10 12 140.00.51.01.52.0 time [ Gyr ] d [ M p c ] ID2251570 2 4 6 8 10 12 140.00.51.01.52.0 time [ Gyr ] M [ M ⊙ ] [ Gyr ] M [ M ⊙ ] [ Gyr ] b / a , c / a , T [ Gyr ] f , A ID225157 - ID2251560 2 4 6 8 10 12 140.00.20.40.60.8 time [ Gyr ] d [ M p c ] ID4573610 2 4 6 8 10 12 140.00.51.01.52.0 time [ Gyr ] M [ M ⊙ ] [ Gyr ] M [ M ⊙ ] [ Gyr ] b / a , c / a , T [ Gyr ] f , A [ Gyr ] d [ M p c ] Fig. 3.
Evolution of galaxies belonging to classes A, B, and C. The columns present the results for di ff erent galaxies, ID76122 (class A), ID225157(class B), and ID457361 (class C). Upper row: Evolution of the total dark, stellar, and gas mass shown with the blue, red, and green lines,respectively. Second row: Same masses, but the vertical scale is adjusted to the stellar and gas mass. Third row: Evolution of three structuralproperties of the galaxies, the axis ratios b / a (blue line) and c / a (red), and the triaxiality parameter T (green). Fourth row: Rotation measure interms of the fractional mass of stars on circular orbits f (blue) and the bar strength A (red). Fifth row: Distance of the galaxy from a more massiveobject (blue) or a merging satellite (red). evolution, its dark matter mass is almost 40 times higher than thestellar mass, and its gas fraction is as high as 0.52. We summarize the properties of the galaxies in the sample,emphasizing the di ff erences between classes A, B, and C, inFig. 4. The upper panel of the figure shows the distributions of Article number, page 5 of 10 & Aproofs: manuscript no. Barstng6
TotalABC - dm / M * ) N / [ kpc ] N - r [ mag ] N max N Fig. 4.
Distributions of di ff erent properties for the whole sample of bar-like galaxies (gray shaded histograms) and for classes A, B, and C (col-ored histograms). the present dark to stellar masses M dm / M ∗ , for which the medi-ans are given in the third column of Table 1. As expected fromthe adopted criteria, the galaxies of class A are not strongly dom-inated by dark matter, and all have M dm / M ∗ <
10, while those ofclass C are strongly dominated, with M dm / M ∗ >
10. The galax-ies of class B form a wider, intermediate distribution. The secondpanel of the figure shows the distributions of the stellar half-massradius, r / . Although the mass loss in stars is not very strongeven in class A, the galaxies of classes A, B, and C tend to haveprogressively larger radii, reflecting the fact that galaxies of class A are more tidally truncated. This is confirmed by the higher me-dian values of this parameter in the fourth column of Table 1.Another important property of the galaxies that was not yetdiscussed is their color, which we measure as the g − r parameterof a subhalo (in terms of SDSS-like filters). This is shown inthe third panel. Because g − r = g − r > .
6, andless so for class C than for class A. (Class C contains only sixgalaxies and class B one galaxy with g − r < .
6, not shown inthe plot.) The median color values are given in the fifth columnof Table 1. In line with the colors, we find most of the galaxies tobe quenched, with nonzero star formation rates only for 7 objectsin class B and 32 in class C (and values above 3 M ⊙ yr − onlyfor one object in class B and one in class C).Finally, in order to confirm that the bar-like galaxies of thesample indeed formed from disks, either through their inherentinstability or as a result of interactions, we calculated the maxi-mum values of the rotation parameter f the galaxy acquired dur-ing this history. For this purpose we used the evolutionary his-tories of this parameter, such as those shown in the fourth rowof panels of Fig. 3 for individual galaxies. Similarly to the casesshown in the figure, maximum values are always higher than thepresent values (by a factor of 1.5 to 19) and lie in the range 0.36-0.78, which means that the galaxies were indeed disks dominatedby rotation at some point in their history before transforming intobar-like shapes with more radial motion. The median maximumvalues (see the sixth column of Table 1) increase from class Ato C, which means that strong tidal interactions are apparentlyable to induce bars even in less regularly rotating disks. We notethat the decrease in rotation parameter f is a good indicator of amore general transformation of galaxies from disks to spheroids,resulting in particular from the evolution in clusters, as discussedin detail by Joshi et al. (2020).
4. Properties of the bars
In this section we study the properties of the bars in the galaxysample, starting with the three fiducial examples discussed inthe previous section. Figure 5 shows the projections of the stel-lar density in the face-on view for the three galaxies ID76122,ID225157, and ID457361. The images show that they indeedhave bar-like shapes with pronounced elongations in the cen-tral parts. Moreover, the bars have various detailed density dis-tributions, from uniformly elongated (ID76122), to distributionswith two density maxima (ID225157) and with bulge-like cen-ters and elongated outer parts (ID457361) similar to barlenses,that is, lens-like shapes discussed by Athanassoula et al. (2015)and Salo & Laurikainen (2017).The bar-like character of the stellar surface density distribu-tion is further quantitatively confirmed by calculating the profilesof the bar mode A ( R ) as a function of the cylindrical radius, R . Examples of such profiles for galaxies ID76122, ID225157,and ID457361 are shown in Fig. 6. Their shapes are typical ofbars, increasing up to a maximum A , max at R max and then againdecreasing to zero. The value of A , max can be used as a mea-sure of the bar strength, in addition to the global values of A within two stellar half-mass radii plotted in the fourth row ofpanels in Fig. 3. The length of the bar can be estimated as theradius R where A ( R ) drops to half the maximum value. How-ever, because the A ( R ) profiles are approximately symmetricaround R max and the profiles are typically much more noisy at R > R max than near the center (because of a small number of starsin the outer bins), it is convenient to use the values of 2 R max as a Article number, page 6 of 10wa L. Łokas: Bar-like galaxies in IllustrisTNG - - - - - - [ kpc ] y [ kp c ] ID76122 - - - - - - [ kpc ] y [ kp c ] ID225157 - - - - - - [ kpc ] y [ kp c ] ID457361
Fig. 5.
Surface density distribution of the stellar components of galaxies ID76122, ID225157, and ID457361 of class A, B, and C, respectively, inthe face-on view at the present time. The surface density, Σ , is normalized to the central maximum value in each case, and the contours are equallyspaced in log Σ . t = GyrID76122ID225157ID4573610 2 4 6 8 10 120.00.20.40.60.8 R [ kpc ] A ( R ) Fig. 6.
Profiles of bar mode, A ( R ) , for galaxies ID76122, ID225157,and ID457361 of class A, B, and C, respectively, at the present time.Measurements were carried out in bins of ∆ R = . good approximation of the bar length. For the galaxies ID76122,ID225157, and ID457361, the bar lengths are thus about 5-6 kpc.A few galaxies of class A are at present at the pericentersof their tight orbits around more massive objects and thus inter-act very strongly. The outer parts of these galaxies are tidallyextended, which manifests itself in a secondary increase in the A ( R ) profiles at larger R . For such objects the values of A , max and R max can still be reliably measured, except for one object(ID60783), which has a monotonically increasing A ( R ) . For thisgalaxy we performed the measurements using the penultimatesimulation output, where the A ( R ) is well behaved.The evolution of the A ( R ) profiles in time for galaxiesID76122, ID225157, and ID457361 is shown in Fig. 7 in thecolor-coded form. After they are triggered, the bars grow steadilyin strength and length. For ID76122 (the upper panel of Fig. 7)we note an increase in A ( R ) at larger R at about 6.3 Gyr, cor-responding to the first pericenter passage of this galaxy arounda larger host, which induced the bar. An indication of this in-crease is also visible at the time of the second pericenter, at 9.8 s(cid:0)(cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6) A (cid:10)(cid:11)(cid:12)(cid:13)(cid:14) t(cid:15)(cid:16)(cid:17) [ G(cid:18)(cid:19) ] R [ kp c ] I(cid:20)(cid:21)(cid:22)(cid:23)(cid:24)(cid:25)(cid:26) (cid:27) (cid:28) (cid:29)(cid:30) (cid:31) !" [ *+, ] R [ kp c ] -./023579 : ; <= >? @ABCD EFHJ [ KLM ] R [ kp c ] I D Fig. 7.
Evolution of the profiles of the bar mode, A ( R ), over time forgalaxies ID76122, ID225157, and ID457361 of class A, B, and C, re-spectively (from top to bottom). Gyr. Such e ff ects are easier to see in higher resolution, controlled Article number, page 7 of 10 & Aproofs: manuscript no. Barstng6 simulations of the evolution of galaxies in clusters (Łokas et al.2016), however.We measured the values of A , max and R max for all galaxiesin the sample. The distributions of these parameters are shownin the two upper histograms of Fig. 8 for the whole sample andfor classes A, B, and C. Because we chose strongly elongatedgalaxies initially using the criterion of axis ratio b / a < .
6, thevalues of A , max are rather high, characteristic of strong bars, butstill have a wide distribution. The values obviously correlate withthe alternative measure of bar strength A , an average bar modewithin two stellar half-mass radii used in Fig. 3, and another pos-sible measure of ellipticity, 1 − b / a , although the scatter is large.The seventh and eighth column of Table 1 list the median val-ues of A , max and R max for the whole sample and subsamples A,B, and C, showing that bars are slightly stronger and longer be-tween subsamples A to C.The third histogram of Fig. 8 shows the distribution of theratio R max / r / , where r / is, as before, the half-mass radius ofthe stellar component of the galaxy. The values of r / are alsosystematically higher from A to C (see Fig. 4 and the fourth col-umn of the table) because the galaxies of classes B and C arenot as strongly tidally truncated as those of A, or they are not atall truncated. However, the median values of R max / r / decreaseslightly with class, but are always of about unity (the ninth col-umn of Table 1). This means that the length of the bar, estimatedas 2 R max , is typically about 2 r / .Next, we estimated the time of the formation of the bar. Forthis purpose we used the measurements of A within two stel-lar half-mass radii as a function of time, examples of which areshown in the fourth row of panels of Fig. 3. Inspection of theseresults for many galaxies reveals that although at early times A typically varies strongly, once it crosses the threshold of 0.2, itremains above this value, and the final values for all galaxies arein the range 0.26 to 0.54. The bar formation times were thus es-timated by determining the last simulation output when A wasabove 0.2 before it dropped below this value when going back intime from the present. The distribution of the bar formation timesestimated in this way is shown in the fourth histogram of Fig. 8.The distribution is very wide; some bars formed very early andsome only recently. The typical formation time is of about halfthe age of the Universe. The median values of this parameter arelisted in the penultimate column of Table 1 and show that theformation times are earliest for galaxies of class B. This may berelated to the possibility that many of these bars were formed bymergers, which are more frequent at earlier times.Finally, we calculated the gas fractions of the galaxies within2 r / at the time of bar formation. As stated above, the lengthsof the bars are typically of this size, therefore the gas content inthis region is more important than the overall gas distribution. Aswe noted before, in the galaxies that contain gas, it tends to bemore extended than the stellar component, but this external gasis expected to have little e ff ect on the formation of the bar inside2 r / . The distribution of these inner gas fractions is shown inthe last histogram of Fig. 8. Interestingly, most of the galaxieshad very little gas in this region at the time of bar formation. Inparticular, only six galaxies had a gas fraction between 0.3 and0.41, and none had a higher value. The median values of thisparameter are listed in the last column of Table 1. We note thatthe value is highest for the galaxies of class B, which is probablyrelated to the fact that their bar formation times are earlier, sothat the galaxies were naturally more gas rich. TNOPQ A RS UVWX YZ[\ ]^_‘ abcd ef A ghijkl mno pqr uvw xyz{|} R ~(cid:127)(cid:128) [ kpc ] (cid:129) (cid:130)(cid:131)(cid:132) (cid:133)(cid:134)(cid:135) (cid:136)(cid:137)(cid:138) (cid:139)(cid:140)(cid:141) (cid:142)(cid:143)(cid:144) (cid:145)(cid:146)(cid:147)(cid:148) R (cid:149)(cid:150)(cid:151) / r / (cid:152) (cid:153) (cid:154) (cid:155)
10 120102030 (cid:156)(cid:157)(cid:158)(cid:159) bar formation (cid:160)¡¢£ [ Gyr ] ⁄ ¥ƒ§ ¤'“«‹› fifl(cid:176) –†‡ fraction within 2 r / at bar formation time N Fig. 8.
Distributions of di ff erent properties of the bars for the wholesample of bar-like galaxies (gray shaded histograms) and for classes A,B, and C (colored histograms).Article number, page 8 of 10wa L. Łokas: Bar-like galaxies in IllustrisTNG
5. Discussion
We studied the properties of well-resolved bar-like galaxies inthe Illustris TNG100 simulation, selected by the single criterionof low enough axis ratio of the stellar component, b / a < . ff erences in their evolutionary histories. The bars in galaxiesof class A were induced by an interaction with a larger object,and the galaxies were strongly stripped of dark matter and gas.The galaxies of classes B and C contain bars that were inducedby a merger or a passing satellite, or that were formed by diskinstability. Class B galaxies were then partially stripped of mass,while those of class C evolved without strong interactions andpreserved most of their dark matter and gas in the outskirts. Thisclassification is in line with the characterization of the galax-ies as centrals and satellites because almost all the galaxies ofclasses A and B are found to be satellites, while those of class Care mostly centrals.Although the bars of galaxies belonging to these sampleswere created through di ff erent mechanisms, their properties areremarkably similar in terms of strength, length, and formationtimes. As expected, the bar lengths correlate well with the stellarhalf-mass radii of the galaxies, but otherwise, we find no corre-lations between di ff erent bar properties. For example, there is nocorrelation between the bar strength and the following parame-ters: bar length, stellar mass, formation time, and the gas fractionwithin 2 r / at the formation time.One of the most interesting results of this study is the de-termination of the upper limit on the gas fraction within 2 r / at the time of bar formation. In order for the bar formationto be possible, this gas fraction apparently cannot exceed 0.4.This is consistent with earlier studies using controlled simu-lations (Shlosman & Noguchi 1993; Athanassoula et al. 2013),although here it was confirmed in the full cosmological context.The exact value of the threshold in particular agrees very wellwith our recent study (Łokas 2020a), where we performed con-trolled simulations of bar formation in Milky Way-like galaxieswith di ff erent gas fractions between 0 and 0.4 with a step of 0.1.It was found that only galaxies with gas fractions 0-0.3 formedbars, while those of 0.4 and higher values did not. These sim-ulations only used a single set of galaxy structural parameters,and the gas distribution followed that of the stars. However, theresults from IllustrisTNG presented here place them in the cos-mological context and validate the threshold of 0.4 as a generalthreshold because it appears to be obeyed by a variety of galax-ies with di ff erent structural parameters that are present in Illus-trisTNG.The bar-like galaxies discussed in this study are dif-ferent from barred galaxies identified earlier using Illus-tris and IllustrisTNG simulations (Peschken & Łokas 2019;Rosas-Guevara et al. 2020; Zhou et al. 2020; Zhao et al. 2020).Previous studies focused on bars in late-type fast-rotating disksrather than on the early-type elongated objects we found. Thebar-like galaxies of this study typically contain little gas, andmost of the stars contribute to the elongated part rather than adisk. Although some streaming motion remains in these objects,as is natural for bars, the orbits of stars are elongated rather thancircular, and most of the galaxies have low rotation parameters.The sample of galaxies presented here is also di ff erent from thesample of prolate galaxies in the Illustris simulation discussedby Li et al. (2018), where the selected objects were more spher- ical and bars were explicitly rejected. The sample is also verydi ff erent from the prolate rotators discussed by Ebrová & Łokas(2017), some of which had prolate shapes.The bar-like galaxies can have very di ff erent properties interms of their dark matter content. This reflects their distinctevolutionary paths. The bars that were formed by tidal inter-actions with a more massive galaxy (class A) are also heavilystripped of their dark matter and gas. About half of this class ofobjects has less dark matter than stars and thus can be classifiedas examples of galaxies lacking dark matter recently discussedby van Dokkum et al. (2018, 2019). The scenario of tidal strip-ping, which a ff ects dark matter more than the stars, would beparticularly viable for galaxies residing in groups, as is the casefor those observed. The bars in the intermediate class B formedthrough a di ff erent channel (often by mergers), but the galaxieslater interacted with a more massive object and lost some of themass. The most numerous class C evolved undisturbed, form-ing the bar in isolation or in small mergers and retained a highdark matter content. This class often also contains a significantamount of gas, but mostly in the outer parts, outside the bar.It is interesting to ask what fraction of the bar-like galax-ies discussed here formed by interactions. Unfortunately, thisquantity is very di ffi cult to estimate. Only for class A, whichcomprises 77 out of 277 galaxies (28% of the sample), can webe reasonably certain that the bar was induced by tidal interac-tions because its formation coincides with a pericenter passagearound a more massive object, and in this configuration, the tidalforces are particularly strong. For many of the class B and C ob-jects we were able to identify significant mergers coinciding withthe morphological transformation that probably caused enoughdistortion in the disk to induce a bar. However, the history ofmany C class galaxies is very quiescent, and only small subha-los were accreted continuously, and it is impossible to determinewhether any of them contributed to the formation of the bar.Because the formation of the bar is likely a stochastic process(Sellwood & Debattista 2009), even a small perturbation may beenough to trigger it.We may wonder what the counterparts of these bar-likegalaxies are among real objects. Some of them, mostly thoseof class A, could be indistinguishable from elliptical galax-ies. Depending on the resolution of the observations and theline of sight, they would be classified as fast or slow rotators(Emsellem et al. 2007). The detection of a significant rotationsignal is most probable in the case of the end-on view becausethe streaming motion of the stars in the bar is strongest in thiscase. When viewed edge-on, they may look like disks as a re-sult of their elongation and rotation. Because they lack disksand spiral arms, they would not in general be seen as typicalbarred galaxies. They could be classified as S0s when seen faceon, however. Good examples of such objects are PGC 31198(NGC 3266) and PGC 41302 (NGC 4479) from the EFIGI cata-log of nearby galaxies, shown in fig. 15 of Baillard et al. (2011).NGC 4479 is a member of the Virgo cluster, therefore its barmay have been tidally induced by the interaction with the cen-tral galaxy. Galaxies of class C bear some similarity to red spirals(Guo et al. 2020), many of which possess large bars and gas onlyin the outskirts, as well as to the blue compact dwarf NGC 2915(Meurer et al. 1996), which has a bar-like stellar component andan extended gaseous disk with almost no stars.Pulsoni et al. (2020) recently studied the sample of early-type galaxies (ETG) in IllustrisTNG and identified among thema subsample of bar-like galaxies with a stellar mass axis ratio b / a < . r / . This subsam-ple overlaps our sample to some extent. Our sample was selected Article number, page 9 of 10 & Aproofs: manuscript no. Barstng6 by the same condition, but applied to stars within 2 r / . In par-ticular, we find that out of 277 galaxies in our sample 201 (73%)also have b / a < . r / . After measuring the kinematicproperties of their bar-like galaxies, Pulsoni et al. (2020) con-cluded that many of them are elongated slow rotators that are notfound in samples of real ETGs, nor in the original Illustris simu-lation. They suggested that the simulated bar-like galaxies in Il-lustrisTNG are probably failed disks resulting from the changedgalaxy formation model in TNG with respect to Illustris.We confirmed that the criterion of b / a < . r / indeed selects only 35 objects in the Illustris-1 simulation, buta comparison similar to the one shown in Fig. 1 reveals thatIllustris-1 galaxies are typically thicker and thus less suscepti-ble to bar instability. In general, the TNG galaxies agree bet-ter with observational properties (Nelson et al. 2018), exceptfor overquenching (Angthopo et al. 2020; Sherman et al. 2020),which is most probably due to too strong kinetic feedback fromactive galactic nuclei, combined with high thresholds for theseed black hole mass and halo mass. This may be of concernin the case of galaxies of our class C, which may rid themselvesof the gas too e ffi ciently and form bars too easily. It should notbe a problem for galaxies of classes A and B, however, for whichthe quenching is mostly environmental and the tidal interactionsand ram-pressure stripping of the gas depend very little on thedetails of the model. These galaxies are clear examples of barsthat were tidally induced or whose growth was aided by the lowgas and dark matter content caused by interactions.While the simulations are certainly not yet perfect, some im-provements may also be needed on the observational side. Theclassification of the galaxies in terms of their shapes includesdeprojection, which is a highly degenerate procedure for triaxialsystems, and in contrast to common belief, the intrinsic shapesof galaxies are not well known. For example, Weijmans et al.(2014) noted that for slow rotators from the Atlas3D project theprolate shapes provide a very good fit when there is no intrinsickinematic misalignment, which is the case of bar-like galaxies.Bassett & Foster (2019) warned that assuming kinematic mis-alignment, as is typically done for triaxial systems, is not a re-liable approach because there are prolate objects without mis-alignment. It is therefore possible that more bar-like galaxies willbe identified in observations when the methods of inferring theintrinsic shapes of early-type galaxies improve. Acknowledgements.
I am grateful to the anonymous referee for useful com-ments, to Gerhardt Meurer for letting me know about NGC 2915 and to theIllustrisTNG team for making their simulations publicly available.
References