Starburst-induced Gas–Star Kinematic Misalignment

A kinematic misalignment of the stellar and gas components is a phenomenon observed in a significant fraction of galaxies. However, the underlying physical mechanisms are not well understood. A commonly proposed scenario for the formation of a misaligned component requires any preexisting gas disk to be removed, via flybys or ejective feedback from an active galactic nucleus. In this Letter, we study the evolution of a Milky Way mass galaxy in the FIREbox cosmological volume that displays a thin, counterrotating gas disk with respect to its stellar component at low redshift. In contrast to scenarios involving gas ejection, we find that preexisting gas is mainly removed via the conversion into stars in a central starburst, triggered by a merging satellite galaxy. The newly accreted, counterrotating gas eventually settles into a kinematically misaligned disk. About 4% (8 out of 182) of FIREbox galaxies with stellar masses larger than 5 × 109 M ⊙ at z = 0 exhibit gas–star kinematic misalignment. In all cases, we identify central starburst-driven depletion as the main reason for the removal of the preexisting corotating gas component, with no need for feedback from, e.g., a central active black hole. However, during the starburst, the gas is funneled toward the central regions, likely enhancing black hole activity. By comparing the fraction of misaligned discs between FIREbox and other simulations and observations, we conclude that this channel might have a non-negligible role in inducing kinematic misalignment in galaxies.


Introduction
The kinematic properties of galaxies, such as the distribution and motion of stars and gas, provide important insights into their formation and evolution (e.g., Somerville & Davé 2015).In general, the stellar component of galaxies is expected to inherit the kinematic properties of the gas out of which it forms, which are in turn set by those of their host dark matter halo (e.g., Hoyle 1951;Peebles 1969).Theoretical models also predict that the axisymmetric or triaxial potential of the stars exerts a torque on the gaseous component that leads to kinematic relaxation, i.e., the alignment of the average angular momenta of both components (Tohline et al. 1982;Lake & Norman 1983).Thus, as a first approximation, both components (stars and gas) are expected to be substantially aligned.However, observations reveal that about 10% (30%) of latetype (early-type) galaxies exhibit either a gaseous or stellar component that is counterrotating with respect to a co-spatial stellar component (e.g., Galletta 1987;Bertola et al. 1992;Rubin et al. 1992;Merrifield & Kuijken 1994;Ciri et al. 1995;Bertola et al. 1996;Kuijken et al. 1996;Pizzella et al. 2004;Sil'chenko et al. 2009;Coccato et al. 2011;Pizzella et al. 2014Pizzella et al. , 2018;;Sil'chenko et al. 2019;Proshina et al. 2020).
Galaxies with misaligned components are also predicted by cosmological galaxy formation simulations (Duckworth et al. 2020a;Khim et al. 2020;Khoperskov et al. 2021;Koudmani et al. 2021) but with large variations, especially in the fraction of galaxies exhibiting gas-star kinematic misalignment (KM).This fraction ranges between about 0.7% and 14%, based on the misalignment angle between the angular momenta of gas and stars in central galaxies with stellar mass M å  10 9 M e at redshift z = 0 (e.g., Velliscig et al. 2015;Starkenburg et al. 2019;Casanueva et al. 2022).Furthermore, galaxies with gasstar counterrotation likely exhibit also a kinematically misaligned stellar component and vice versa, suggesting a common origin scenario (Katkov et al. 2013;Khoperskov et al. 2021;Bao et al. 2022).
In order for a galaxy to display KM, the new counterrotating gas disk has to replace any extended preexisting corotating gas disk.The latter must thus be removed, e.g., via ejective feedback from an active galactic nucleus (AGN; e.g., Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.Starkenburg et al. 2019) or tidally stripped from nearby satellites (e.g., Starkenburg et al. 2019;Khoperskov et al. 2021).Stellar feedback-driven outflows could also play a role in expelling the preexisting gas (e.g., Muratov et al. 2015;Pandya et al. 2021); however, their role in relation to KM has received little attention.An alternative possibility to AGN/ stellar feedback or tidal stripping is that the preexisting gas is depleted via star formation.The relative role of different mechanisms in effectively removing the preexisting gas has not been explored.
A correlation between AGN activity and the emergence of a counterrotating gaseous component has been recently observed (Raimundo et al. 2023) and has also been predicted by cosmological and idealized simulations (e.g., Capelo & Dotti 2017;Starkenburg et al. 2019;Duckworth et al. 2020b).However, whether or not this AGN-KM correlation is indicative of AGN feedback expelling gas is still an open question (see, e.g., Khoperskov et al. 2021).
In this work, we consider a sample of galaxies from the FIREbox cosmological simulation (Feldmann et al. 2023) that exhibits KM at z = 0. FIREbox simulates galaxy evolution with a high dynamic range and implements an accurate treatment of the interstellar medium physics by means of the FIRE-2 models (Hopkins et al. 2018).FIREbox does not implement feedback from active black holes, making it an ideal numerical experiment to explore the occurrence of KM in the absence of this feedback mechanism.In this Letter, we will mainly focus on a single galaxy of approximate Milky Way mass at z = 0.This galaxy displays a well-defined, counterrotating gaseous disk, representing the most dramatic example of KM in FIREbox.We aim at understanding the physical processes responsible for removing any possible preexisting corotating gas disk, promoting the formation of a misaligned component.
FIREbox is run with the meshless-finite-mass code GIZMO (Hopkins 2015), with FIRE-2 sub-grid physics (Hopkins et al. 2018).Gravity is calculated with a modified version of the parallelization and tree gravity solver of GADGET-3 (Springel 2005) that allows for adaptive force softening.Hydrodynamics is solved with the meshless-finite-mass method introduced in Hopkins (2015).The FIRE-2 model includes gas cooling down to 10 K, which naturally results in a multiphase interstellar medium (ISM).Star formation occurs stochastically in dense (n > 300 cm −3 ), self-gravitating, Jeans-unstable, and self-shielded gas, with a 100% efficiency per free-fall time.Stellar feedback channels include energy, momentum, mass, and metal injections from Type II and Type Ia supernovae and stellar winds from OB and AGB stars.Radiative feedback models account for photoionization and photoelectric heating and radiation pressure from young stars, using the Locally Extincted Background Radiation in Optically Thin Networks approximation (Hopkins et al. 2012).Feedback from supermassive black holes is not included.
The FIRE-2 model has been validated in several studies on galaxy formation and evolution across wide ranges of stellar masses and numerical resolutions (Wetzel et al. 2016;Ma et al. 2018aMa et al. , 2018b;;Hopkins et al. 2018).FIREbox reproduces many of the observed galaxy properties (see Feldmann et al. 2023) and has been analyzed in a number of recent studies (e.g., Bernardini et al. 2022;Moreno et al. 2022;Rohr et al. 2022;Gensior et al. 2023aGensior et al. , 2023b;;Cenci et al. 2023).
The analysis presented here is based on the fiducial FIREbox hydrodynamical simulation (N b = 1024 3 and N DM = 1024 3 ), with a mass resolution of m b = 6.3 × 10 4 M e for baryons and m DM = 3.3 × 10 5 M e for dark matter.The force resolution is 12 pc (physical, up to z = 9; comoving for z > 9) for stars and 80 pc for dark matter.The force softening of gas particles is adaptive (and related to their smoothing length) down to a minimum of 1.5 pc in the dense ISM.

Definitions and Selection
To identify dark matter haloes we employ the AMIGA Halo Finder7 (Gill et al. 2004;Knollmann & Knebe 2009).The halo radius R vir is defined based on the virial overdensity criterion, so that the halo virial mass is where r z m ( ) is the critical density at a given redshift, is the overdensity parameter, and W L z ( ), W z m ( ) are the cosmological density parameters at redshift z (Bryan & Norman 1998).We define the total galaxy radius to be 10% of the halo virial radius (Price et al. 2017).
For each particle in a given galaxy, we define the circularity parameter as in Abadi et al. (2003): where j z is the projection of the specific angular momentum along the total specific angular momentum of the stars in the galaxy (representing the z-axis in the reference frame we choose) and j E max ( ) is the maximum specific angular momentum allowed on a circular orbit with the same specific binding energy E. Particles with ε > 0 (ε < 0) are on corotating (counterrotating) orbits with respect to the average rotation axis of the stellar component.In the following, we will refer to the orientation of orbits with respect to the rotation axis of the stars.The average circularity, 〈ε〉, computed as the massweighted arithmetic average of the circularity parameter of all the considered particles, gives a first estimate of the importance of the disk component and on the direction of rotation.For instance, 〈ε〉 ∼ 1 (〈ε〉 ∼ −1) indicates the presence a prominent corotating (counterrotating) disk component, while 〈ε〉 ∼ 0 is related to the presence of either a dominant bulge component or equally important counter-and corotating components.A kinematic (corotating) disk component is commonly identified with the set of particles with a circularity parameter ε ε th , typically with ε th = 0.7 (e.g., Aumer et al. 2013).Similarly, we measure the contribution of the counterrotating gas disk component as the mass fraction of gas particles in the galaxy with ε −ε th .
We use the average circularity of all gas particles in the galaxy to determine if the galaxy shows KM.About 4.4% (8 out of 182) of all z = 0 FIREbox galaxies with M å 5 × 10 9 M e have 〈ε〉 0. This fraction becomes about 1.1% (2 out of 182) if we are more restrictive with our definition and demand 〈ε〉 −0.7.Evaluating the degree of misalignment using the angle between the total angular momenta of the gas and stellar components gives similar results, and it does not affect our conclusions.For comparison, about 5.4% (10 out of 182) of z = 0 FIREbox galaxies with M å 5 × 10 9 M e display a misalignment angle between the gas and star's angular momenta θ 90°.This fraction reduces to about 1.7% (3 out of 182) for θ 150°.
In this work, we focus on studying a specific Milky Way mass galaxy at z = 0, which represents the most dramatic case of KM in FIREbox.This galaxy displays a distribution of circularity parameters at z = 0 that is consistent with having a well-defined, counterrotating gas disk, with average circularity 〈ε〉  −0.9.On the other hand, its stellar component has an 〈ε〉 ∼ 0.1.

The Two Phases of the Kinematic Misalignment
Figure 1 shows the most dramatic example of KM from FIREbox taking place in the galaxy that we will hereafter label as H22.At z = 0, H22 has virial mass M vir ∼ 1.2 × 10 12 M e , stellar mass M å ∼ 6.3 × 10 10 M e , and total gas mass M gas ∼ 1.8 × 10 10 M e within its size of 0.1 R vir ∼ 27.6 kpc at z = 0. Close to the time of the KM event, H22 accretes gas mainly by tidal stripping from a gas-rich satellite galaxy.At z ∼ 0.7, the satellite was at its last apocenter before merging, at about 100 kpc from the center of H22.At that time, the satellite has stellar mass M å ∼ 10 9 M e and total gas mass M gas ∼ 2 × 10 9 M e , both computed within a characteristic size of about 6 kpc (of the order twice its stellar half-mass radius).
The left panels in Figure 1 show the temporal evolution of the average circularity, 〈ε〉, for the gas (top panel) and stars (bottom panel) in H22.We compute 〈ε〉 within 0.1 R vir and within fixed apertures of 10 kpc and 5 kpc.In the ∼800 Myr between redshift z ∼ 0.35 and 0.45, 〈ε〉 of gas drops and switches signs from ∼0.8 to ∼ −0.75 (independent of the considered aperture).The average circularity then remains approximately constant at 〈ε〉 ∼ −0.75 until z = 0.2, after which it decreases to ∼ − 0.9 by z = 0. We will hereafter refer to the transition period of time between z ∼ 0.35 and 0.45 as the KM event.
The 〈ε〉 of the stars within 0.1 R vir and 10 kpc increases from ∼0.3 to ∼0.42, from redshift z = 0.5 to z ∼ 0.45, then steadily decreases down to ∼0.1 at z = 0.These results are consistent with a steady counterrotating gas disk fueling the star formation in the galaxy until z ∼ 0.
The right panels in Figure 1 show the temporal evolution of the total star formation rate (SFR; averaged over 100 Myr) and total gas mass (M gas ), within 0.1 R vir , 10 kpc, and 5 kpc.In the first ∼600 Myr during the misalignment event (z ∼ 0.45-0.38), the SFR steadily increases, changing by ∼0.2 dex, then decreasing by a factor ∼1 dex in the following ∼200 Myr (until z ∼ 0.35).The change in SFR over time is independent of the chosen aperture, implying that the bulk of the star formation is always occurring within 5 kpc.Until z = 0, the galaxy maintains an approximately constant SFR, within ∼0.3 dex.The total gas mass stays approximately constant (within ∼0.1 dex) from z ∼ 0.45-0.4,and then decreases by ∼0.6 (∼1) dex until z ∼ 0.35, within 0.1 R vir (5 kpc).After z ∼ 0.35, the total gas mass within 0.1 R vir gradually increases by ∼0.8 dex until z = 0, while the total gas mass within smaller radii increases by only ∼0.5 dex until z = 0.
To summarize, the misalignment event consists of mainly two steps.In the first phase, the SFR increases at constant total gas mass, and the average gas (stellar) circularity decreases (increases).During the second phase the gas mass drops drastically, followed by a drop in SFR about 200 Myr later, and the average circularity of both the gas and stars decreases until complete KM is reached.

Depletion of the Preexisting Gas and New Accretion
To study the fate of the corotating gas, prior to the emergence of the kinematically misaligned disk, we trace the properties of the gas particles constituting the preexisting gas reservoir through time in the simulation.
Figure 2 shows the evolution of the total gas content in H22, during z = 0.3-0.45 when the KM occurs.Particles are colorcoded based on whether they constitute the preexisting gas reservoir, i.e., whether they are within 10 kpc at z = 0.45.By z ∼ 0.35 the gas content of H22 is almost entirely replaced by the gas that has been inflowing onto the galaxy from a merging satellite.Furthermore, we show that most of the stars formed out of the preexisting gas reside within the galaxy at z = 0.3 and turned into stars, rather than expelled/stripped from the galaxy.At z = 0, the stars formed out of the preexisting gas disk are found in a thicker stellar structure in the innermost 2 kpc of H22 (see, e.g., Yu et al. 2021, 2023, andreferences therein), while most of the new stars at z = 0 are forming in the more extended counterrotating gaseous disk.
Figure 3 shows how the baryons constituting the preexisting gas reservoir within 10 kpc at z = 0.45 (i.e., the preexisting baryons) progressively move to smaller galactocentric radii and form stars.About 80% of the the preexisting baryons in H22 are still found within the galaxy at z ∼ 0.3, especially within ∼2-5 kpc.About 95% of the preexisting baryons turn into stars already by z = 0.35.Furthermore, approximately 80% of them retain a circularity parameter ε > 0, implying that they mainly contribute to the corotating component, although not necessarily to the disk.Furthermore, we find that about 45% (10%) of preexisting baryons have ε >0.7 (ε < − 0.7) at z = 0.3 and therefore contribute to the corotating (counterrotating) stellar disk.By z = 0, the preexisting baryons are still found in H22 and have a more bulge-like kinematic morphology, based on the distribution of their circularities, with only 20% of them contributing to the (either co-or counterrotating) stellar disk component.

Summary and Discussion
In this Letter, we studied the origin of the KM between the gas and stellar components of z = 0 galaxies in the FIREbox cosmological volume (Feldmann et al. 2023).Specifically, we Figure 2. Face-on and edge-on projections (upper and lower panels, respectively) of H22 in four snapshots around the period of KM and at z = 0.The gas is colorcoded according to whether it is within (purple) or outside (blue) the 10 kpc from the center of the galaxy at z = 0.45, when the preexisting gas disk is still in place.The projections are rotated in the x y , ( )-plane such that the x-axis is aligned with the principal inertia axis associated with the largest moment.The preexisting gas is depleted by z ∼ 0.35, and the counterrotating gas accreted from the merging satellite settles into a misaligned disk by z ∼ 0.3.Stars formed out of the preexisting gas (orange points) remain within the galaxy.At z = 0, the stars formed from the preexisting baryons are still found in the central regions of H22.
aimed at understanding the mechanisms causing the depletion of the preexisting gas reservoir and thus allowing for the formation of a stable, counterrotating gas disk.
Our main finding is that the preexisting gas disk that was corotating with respect to stars at z = 0.45 is almost entirely (∼95%) converted into stars by z = 0.3 (in about 800 Myr; see Figure 3, left and middle panels), rather than being ejected.This compaction-driven starburst (see, e.g., Dekel & Burkert 2014;Zolotov et al. 2015;Tacchella et al. 2016;Cenci et al. 2023) is triggered by instabilities induced by a nearby gas-rich merging satellite.
In this case study, KM occurs in essentially two phases.In the first phase, torques arising from the interaction destabilize the corotating gas disk against global gravitational collapse, driving a gas compaction event and increasing the SFR (see, e.g., Barnes & Hernquist 1991;Mihos & Hernquist 1994, 1996;Hopkins et al. 2006;Di Matteo et al. 2007;Cox et al. 2008;Renaud et al. 2014;Moreno et al. 2015;Hopkins et al. 2018;Renaud et al. 2019;Moreno et al. 2021;Segovia Otero et al. 2022;Cenci et al. 2023).As a result of corotating gas being converted into stars, the average circularity of the stellar component increases, while the average circularity for the gas starts decreasing.In the second phase, gas is rapidly depleted and converted into stars, and the newly accreted gas from the merging satellite gradually forms a star-forming counterrotating disk with respect to stars.Consequently, the average circularity of both the stellar and gas components decreases.
About 80% of the stars formed out of the preexisting gas retain a corotating orbital configuration by the end of the KM period.The newly accreted counterrotating gas affects the orbits of the star-forming, preexisting gas, increasing the fraction of preexisting baryons on counterrotating orbits, especially in the second phase of KM.
A starburst-induced KM may occur repeatedly during the lifetime of galaxies and its impact is possibly imprinted in the distributions of ages and metallicities for stars in the corotating and counterrotating components, as reported by observations (e.g., Pizzella et al. 2014;Bassett et al. 2017;Nedelchev et al. 2019;Katkov et al. 2023;Zinchenko 2023).This naturally depends on the origin of the newly accreted gas, and it is therefore not trivial to pinpoint when KM happened.Compared to a KM induced by AGN feedback, the drop in SFR and total gas mass in the galaxy happens on a longer timescale (of the order of the gas depletion time) in the case of a starburstinduced KM.Furthermore, we showed that most of the stars formed during KM would reside in the central region of the galaxy, as opposed to what was expected if the preexisting gas was expelled from the galaxy by AGN feedback.High spectral and spatial resolution data from, e.g., Multi Unit Spectroscopic Explorer (Bacon et al. 2010) and James Webb Space Telescope (Gardner et al. 2006) will enable a more complete analysis of the kinematics, metallicities, and ages of stars in galaxies with a counterrotating component, helping to shed light on the exact mechanism that leads to KM.
Our starburst scenario promotes central gas inflow, potentially fueling the central black hole.The AGN-KM connection might thus arise from central gas compaction events, inducing both the AGN activity and removal of the preexisting gas disk.This picture is consistent with recent observations of a large fraction of galaxies exhibiting both kinematically misaligned components and AGN (e.g., Raimundo et al. 2023).It also aligns with theoretical work arguing that gas removal by AGN feedback is subdominant compared to other processes such as, e.g., tidal stripping (e.g., Khoperskov et al. 2021).
About 4.4% of z = 0 galaxies in FIREbox displays gas-star KM, comparable to those reported in previous numerical works (e.g., Velliscig et al. 2015;Starkenburg et al. 2019;Casanueva et al. 2022).Furthermore, Beom et al. (2022) recently found that about 2% of edge-on galaxies in the MaNGA survey (Abdurro'uf et al. 2022) exhibit a visual misalignment angle between their gas and stellar component of more than 150°.These results are in agreement with what we found using similar selection criteria (see Section 3).Therefore, we conclude that the channel we propose plays a potentially a important role in driving KM in the absence of AGN feedback.
We showed that merger-induced starbursts may be a novel mechanism to promote gas-star KM in galaxies.However, its imprint on the galaxy properties might not be easily disentangled from that of scenarios involving black hole physics.Further work with larger samples of galaxies displaying KM is thus required to quantify the relative importance of this channel and its implications for galaxy evolution.Swiss National Science Foundation (grant No PP00P2_194814).E.C., R.F., L.B., and M.B. acknowledge financial support from the Swiss National Science Foundation (grant No 200021_ 188552).J.G. gratefully acknowledges financial support from the Swiss National Science Foundation (grant No CRSII5_193826).J.M. is funded by the Hirsch Foundation.The FIREbox simulation was supported in part by computing allocations at the Swiss National Supercomputing Centre (CSCS) under project IDs s697, s698, and uzh18.This work made use of infrastructure services provided by S3IT (www.s3it.uzh.ch), the Service and Support for Science IT team at the University of Zurich.All plots were created with the MATPLO- TLIB library for visualization with Python (Hunter 2007).

Figure 1 .
Figure 1.Left panels: temporal evolution of the average circularity parameter for the gas (top panel) and stars (bottom panel) in H22, computed with respect to the average angular momentum of the stellar component, within different apertures (0.1 R vir , 10 kpc, and 5 kpc).The gray bands, at z = 0.45-0.4and z = 0.4-0.35,highlight approximately the time period during which the gas-star kinematic misalignment (KM) event occurs.Right panels: temporal evolution of the star formation rate (SFR; averaged over 100 Myr; top panel) and total gas mass (M gas ; bottom panel) in H22, within different apertures (0.1 R vir , 10 kpc, and 5 kpc).The gas disk is corotating with the stellar component at redshift z = 0.5 and experiences KM at z ∼ 0.45-0.35that coincides with a large reduction of both the SFR and total gas mass.

Figure 3 .
Figure3.Evolution of the baryons forming the preexisting gas disk in H22 at z = 0.45 (i.e., the preexisting baryons).Left panel: evolution of the fraction of preexisting baryons that formed stars (circles), which is still within the initial aperture of 10 kpc (triangles) and which has a circularity parameter ε > 0 (i.e., on corotating orbits with respect to the stellar component of the galaxy).Middle panel: mass-weighted probability density function (PDF) of the galactocentric radii of the preexisting baryons.Right panel: mass-weighted PDF of the circularity parameter of preexisting baryons with respect to stars, within 10 kpc.The color code refers to redshift.About 70% of the preexisting baryons is still within the galaxy by z = 0.3 and is concentrated within 2-5 kpc.About 95% of the preexisting baryons is converted into stars already by z ∼ 0.35.About 80% of the preexisting baryons retain a circularity parameter ε > 0.