Jet Feedback in Star-Forming Galaxies

Abstract

In this paper, I review our understanding of how jet feedback works in star-forming galaxies. There are some interesting differences to radiative feedback from Active Galactic Nuclei (AGN). Jets act on galaxy haloes as well as on dense gas, for example in regularly rotating discs, where they can suppress star formation (particularly in the centre, negative feedback), but also enhance it (positive feedback). Jet feedback may produce turbulent, multi-phase gas structures where shocks contribute to the ionisation and is observed in connection with galactic outflows. The exact driving mechanism of these outflows is still unclear, but may be a combination of effects linked to star formation, jet-induced turbulence and radiative AGN feedback. Supermassive black holes in any galaxy can produce jets. Preferential radio detections in more massive galaxies can be explained with different conditions in the circumgalactic medium and, correspondingly, different jet–environment interactions.

1. Introduction

Understanding the effect of feedback via Active Galactic Nuclei (AGN) on the evolution of galaxies is a major challenge in astrophysics. Arguments for an important role of the central supermassive black hole (SMBH) include the correlation between the SMBH mass M and the mass and velocity dispersion $\sigma$ of the spheroidal component of the host galaxy [1,2,3,4,5,6,7,8,9], which can be interpreted as the outbursts of black holes of a certain size only being able to quench star formation in a given galaxy [10,11]. Recent simulations explore this in much detail, also finding ways of co-evolution of black holes and galaxies that do not involve AGN feedback [12], but typically do require AGN feedback, at least for massive galaxies $\gtrsim {10}^{10}{M}_{\odot }$, e.g., [13,14,15]. Another argument is the need for additional feedback beyond the one from star formation to limit galaxy growth to observed masses in cosmological models [16,17,18,19,20,21,22]. AGN, including radio AGN, while rare among the entire population of galaxies, are found frequently enough with sufficient power to make the scenario of AGN feedback energetically plausible [23,24,25,26,27]. This evidence includes radio AGN in dwarf galaxies [28]. Evidence for the effect of radio jets on hot gas around galaxies has been seen directly with X-ray imaging [29,30,31] and is implied from the formation of fat radio lobes e.g., [32,33].

AGN feedback via jets, sometimes called radio mode feedback, has some important physical differences compared to the quasar mode, which is linked to the radiative output of the AGN. In the quasar mode, relatively dense gas around the AGN is accelerated via radiation pressure e.g., [34,35,36,37,38,39], subject to instabilities [40,41,42,43,44,45], gravity and interaction with the environment e.g., [46,47]. The radiation force acts mainly in the ionisation cones and declines away from the SMBH as $r^{-2}$. It is hence strongest in the centre of the galaxy. Jets can interact directly with dense gas in a galaxy to a varying degree, depending on their inclination. After breakout from any dense interstellar medium (ISM) they can then channel the feedback energy far away from the AGN, so that it can interact for example with circumgalactic gas, or, if the galaxy is a member of a cluster, the intracluster medium, with indirect effects on the host, and potentially other galaxies [48]. The physics of jet feedback is thus different from the one of radiative AGN.

This review is structured from small to large scales. After a general introduction to jet physics and the physics of the jet–environment interaction (Section 2), I first review jet feedback in the central kpc of a galaxy (Section 3), and then discuss the scale of an entire galaxy including jet-induced star formation (Section 4). Section 5 covers emission line lobes and the alignment effect, which have been connected to outflows and jet-induced star formation outside the host galaxies. Another feature of high-redshift radio galaxies are associated absorption line systems, which have been interpreted as halo shock waves from starburst winds preceding the jet event (Section 6). The contribution of jets to the heating of galactic gas halos and thus prevention of gas cooling and condensation in the host galaxy is discussed in Section 7.

2. Jet–Environment Interaction in General

A jet channels energy away from an AGN and transfers it to gas phases of interest when and where it interacts with them. Evidence for this jet–environment interaction is seen directly in the hotspots of extragalactic radio sources e.g., [49,50,51]. There is good evidence that AGN jets are generally relativistic on parsec scales [52,53,54,55,56,57], hence supersonic with respect to their ambient medium. Jets are likely initially accelerated as well as collimated by magnetic fields e.g., [58,59]. Hydrodynamic interaction with the ambient medium takes over the collimation at some radius, possibly even on the parsec scale [60,61]. The ambient gas pressure then drives a recollimation shock into the beam, which leads to a stable, cylindrical jet, if the opening angle from the magnetic collimation phase was not too large [62,63]. Unless there is significant entrainment, which would result in the jet slowing down, becoming unstable and disrupt [64,65], the jet velocity will remain essentially constant up to the hotspot. Hotspots are therefore best interpreted as shocks where the high bulk velocities are isotropised and electrons are accelerated to high Lorentz factors, such that they become observable via synchrotron radiation e.g., [66,67]. Hotspots are high-pressure regions that will expand and inflate the radio lobes, whenever the jet density is significantly below the ambient density [68,69]. All three components, jets, hotspots and lobes, are clearly visible on radio images of a large number of extragalactic radio sources, for a review see [31]. Radio lobes are initially strongly overpressured and drive a fairly spherical shock wave into the ambient medium around the radio source [33,70]. The shock’s structure then elongates as the sideways Mach number drops [71]. Many observed radio galaxies are close to pressure equilibrium sideways and have very weak shocks around them, except near the hotspots [72,73]. An overview sketch can be found in Figure 1. Jets interact with different components of the host galaxy ISM in different ways, which I address in turn in the following.

3. The Central Kiloparsec

The ISM in the central parts of galaxies can be complex, for example shaped by bars, if present e.g., [74]. The gas dynamics can lead to suppressed star formation in regions of low gas density or shear, or strong star formation in regions of gas accumulation. In the Milky Way, an example for a barred galaxy, the gas density is approximately constant with radius down to the bar end at a few kpc, then dips by a factor of approximately two and then peaks up strongly in the central kpc, the so-called central molecular zone [75]. This region probably undergoes a cyclic behaviour with periods of re-filling with gas channeled down by the bar, and repeated starbursts [76].

The central kpc is of particular significance for the jet–environment interaction in galaxies, because the size is comparable to the scale height of the neutral ISM, typically a few hundred parsecs [77,78,79,80]. Jets are expected to interact strongly with the clumpy ISM in galaxies while contained in it. This is well-studied in nearby jetted Seyfert galaxies. MRK 78 is an excellent example [81], where the knotty structure in radio continuum as well as optical emission lines suggest strong and localised interaction between the jets and the clumpy ISM. Other examples include NGC 4151 [82,83] and NGC 5929 [84,85]. Sources of these sizes in general are well-known from radio observations as peaked spectrum sources and compact symmetric objects, with common spectral features that can be well-explained by absorption from a dense ISM [86,87].

Jets interacting with the clumpy media likely present in the centres of many star-forming galaxies have been modelled in hydrodynamic simulations e.g., [88,89,90,91,92,93,94]. The strong interaction with clumps is also seen in such simulations. In general, the relativistic jet plasma is efficiently isotropised by jet–cloud interactions. The shocked high-pressure plasma drives a shock wave through the ISM. The result is a complex combination of cloud acceleration, ablation and compression, both triggering and suppressing star formation. If the jet is directed into the disc of a star-forming galaxy, the interaction will lead to a general expansion of the ISM with strong kinematic perturbations of the gas, and this is in good agreement with observations of line kinematics in at least the well-studied case of IC 5063 [95]. The energy transfer from the jet to cloud kinetic energy is of the order of 20–30 per cent in this phase [89,94]. The momentum transfer for any given direction exceeds unity, because the pressure of the shock-heated jet and ambient plasma produces additional momentum compare also [96], the mechanical advantage. The different directions cancel, of course, if one takes the sum, to conserve momentum overall. This phase takes about ${10}^5$–${10}^6$ yrs for jets with typical energy fluxes (${10}^{43-46}$ erg/s), possibly longer, if the jet is oriented in the disc plane [94,95]. After the jet has broken out of the dense ISM of a galaxy, the centre is predicted to have a significantly reduced gas density e.g., [97], perhaps with the exception of very central and dense components. A shock is driven from the centre into the outer parts of the ISM that may enhance star formation, and some gas is set up with enhanced kinematics to form an outflow, possibly observable in optical emission lines. The latter has nicely been demonstrated in 3D simulations by Meenakshi et al. (2022) [94]. Because the interaction with the clumpy interstellar medium is stochastic, the jets will in general have acquired a substantial asymmetry, with one jet likely being significantly longer than the counterjet [98].

On much smaller scales of the order of the scale height of the dense molecular gas $\approx 50$ pc, e.g., [99], radio lobes could drive star-forming shock waves that may even produce high-velocity stars [100].

This general picture is consistent with the rapid decline of star formation inferred in the centre of some post-starburst galaxies e.g., [101,102]. The general quenching of galaxies may, however, not be strongly influenced by jet feedback, at least at low redshift, because radio AGNs appear typically only about 1.5 Gyr after the peak of a star formation episode [103,104].

4. Small Radio Galaxies and Jet-Induced Star Formation

This section is about radio galaxies with radio sizes comparable to the diameters of gaseous galactic discs (tens of kpc). Once the jets have left the dense ISM of the host galaxies they are expected to form some kind of radio lobe. How prominent they are and how well they are observable differs with generally more prominent lobes in more massive galaxies [105]. Synchrotron emission increases with both magnetic and relativistic electron pressure. Denser environments will generally provide more resistance and thus lead to higher pressure, and hence more radio emission. While the intracluster medium in galaxy clusters can reach $>{10}^{-2}$ cm${}^{-3}$, the circumgalactic gas around the Milky Way, with a stellar mass $M_{\ast }\approx 6\times {10}^{10}{M}_{\odot }$ [106,107], is already at $n\approx {10}^{-4}$ cm${}^{-3}$ [108]. For smaller galaxies, the virial temperature of the halo, which is the expected temperature scale for the gas, drops below ${10}^6$ K. The volumetric cooling rate is proportional to $\Lambda n^2$, with the cooling function $\Lambda$ being a function of the temperature and, in general, also metallicity. Hence, any hydrostatic halo would have to be much more tenuous to prevent pressure loss from cooling for details [105]. Such halos are easily pushed to inflow or outflow states.

Radio lobes in the circumgalactic medium are nevertheless known. A likely example is the $\approx 8$ kpc sized pairs of the Fermi bubbles in the Milky Way e.g., [109,110,111,112], though the radio emission was actually not observed in this case, probably due to our special location. The nearby Circinus galaxy seems to be a similar case [113], with lobe sizes of the order of one kpc. Similar detections in X-rays and gamma rays include M31 [114], NGC 891 [115], and NGC 3079 [116]. More radio detections can be expected with the SKA [117]. The luminous infrared galaxy IC 2497 (Hanny’s Voorwerp) at redshift of $z=0.05$ is one of a few spiral galaxies, with likely jet-related radio emission in the circumstellar gas tens of kpc from the nucleus, [118], where the radio emission is only seen on the side of the galaxy that seems to contain the denser gas structures. Nesvadba et al. (2021) [119] presented an analysis of the nearby massive spiral galaxy J2345-0449, which is also associated with radio lobes of 1.6 Mpc diameter. The galaxy features a massive molecular gas ring that is obviously kinematically impacted by the jet at two opposite interaction points.

Nesvadba et al. (2021) [119] reported an unusually low star formation rate and suggested the jet impact to stir the molecular gas ring and thus prevent star formation. The width of the observed ring of 24 kpc makes this appear difficult with the so far discussed mechanisms, but the authors also explain the difficulties with other, non-jet related mechanisms that might suppress star formation in this molecular ring. Many galaxy-sized radio sources are known, for example, from the 3CRR and LOFAR surveys [26], and also at high redshift $z\gtrsim 2$, e.g., [120].

When the radio source expands into the galactic halo, a radio lobe may form, if the circumgalactic gas is denser than the jet e.g., [63,68,70]. The point of highest pressure in the radio lobe is the hotspot at its tip. Therefore, there is generally a backflow in the radio lobes towards the galaxy. Gaibler et al. (2012) [121] found, in 3D hydrodynamical simulations of powerful jets developing lobes in the ISM, that the pressure of the lobes compressed the clumpy ISM in the simulated galaxy (compare Figure 2). They concluded that this likely led to star formation enhanced by a factor of a few propagating outwards from the galactic centre, as the radio lobes grow. Follow-up studies found that this phase of enhanced star formation may last for much longer than the active time of the radio source, and thus lead to more rapid gas exhaustion [122].

Isolated gas clouds in the circumgalactic gas that are overrun by the jet’s bow shock will experience compression, and plausibly form stars [123]. If they are also overrun by the radio lobes, they will also experience shear, be torn apart, and mix with the lobe material, thus suppressing star formation [124,125]. Regions of well-developed turbulence in the radio lobes can also enhance drop-out of cold gas, and thus, star formation [126]. As shown in 3D high-resolution zoom simulations by Krause (2008) [127], it depends sensitively on the initial conditions, such as on the initial mass loading of the radio lobes, if the cold gas mass increases or decreases. This source also derives a power estimate for emission lines that cold gas emits in a multiphase turbulence situation, such as in a radio lobe due to shock heating and mixing. The simulations gave a cooling power of dense gas as a proxy for emission line luminosity of roughly 1 erg/s for every ${10}^{12}$ erg of turbulent kinetic energy. Meenakshi et al. (2022) [94] used a more sophisticated emission model to predict the emission for, specifically, the O III luminosity in a region of clumpy ISM impacted by a jet, and found ${10}^{43}$ erg/s peak flux for simulations with about 100 times more powerful jets. With a simulation time of $\approx 1$ Myr and conversion of about 10 percent of their energy flux into kinetic energy of dense gas, this translates to a peak efficiency of 1 erg/s in O III luminosity for ${10}^{14}$ erg of kinetic energy in dense gas. Since much of the emission may be from other lines, particularly hydrogen, the order of magnitude comparison seems to be in reasonable agreement. Multiphase turbulence in radio lobes can thus sustain a certain level of dense emission line clouds, possibly leading to the alignment effect, which is discussed in the next section.

5. Emission Line Lobes and Alignment Effect

The combination of optical and radio observations that constitute the alignment effect has been reviewed, for example, in the Refs. [126,128,129]. Good examples are found in the Refs. [130,131,132]. Optical, nebular emission, is aligned with, and in many well-resolved cases “inside”, that is close to the host galaxy on the same axis or co-spatial with the radio lobes (compare Figure 3). The observational characteristics of these sources are well-explored. Smaller radio sources (≲100 kpc) have more prominent emission line lobes with large bulk outflow speeds, comparable to the escape velocity of the host galaxy, and velocity widths often in excess of 1000 km/s [133]. Their emission line ratios are more consistent with shock ionisation rather than photo-ionisation by a hidden quasar (as is the case for the larger sources). The picture is consistent with a turbulent outflow, where the emission line gas might be stirred up by strong interaction of the jets with the ISM in the very early phases [97] and interaction of the turbulent backflow in the radio lobes in young (≲100 kpc) sources [96]. Turbulence would then have decayed in the larger scale sources, which would also lower the emission line power due to shock ionisation (compare above), such that photoionisation dominates. Another important effect is the detachment of the radio lobes due to buoyancy. In small sources, the lobes likely extend back to the host galaxy and join there, which, due to synchrotrons ageing backwards from the hotspots, can only be observed at low frequencies [134]. This can also, for example, be seen from the low-frequency radio images and associated X-ray cavities in the nearby radio galaxy Cygnus A [135,136], where the gas content and star formation rate of about 10 $M_{\odot }$/yr [137] is probably too low, given the high mass of the eliptical galaxy, to produce prominent emission line lobes. At around 100 kpc, powerful radio sources are expected to come into a pressure equilibrium and the lobes detach from the host galaxy, moving outwards along the radio axis [72]. The entrained gas is expected to be carried along with the radio lobes, especially later, when the source is switched off and the lobes keep buoyantly rising away from the host galaxy [138]. The surrounding hot atmosphere then flows back in where the radio lobes have left. Such dense gas is a major obstacle for galactic scale gas outflows [139,140,141]. It is hence very possible that emission line lobes constitute one distinct episode of a gaseous outflow in a massive galaxy: whatever mechanism ejects the dense ISM does so when the low-density radio lobes surround the galaxy. Then the radio lobes take this gas with them, when they rise away from the galaxy. This can be a major gas ejection mechanism, with derived outflowing gas masses of up to the order of ${10}^{10}{M}_{\odot }$ [130].

In this picture, it is probably not surprising that stars may form in such emission line lobes, when the turbulence in the lobes settles down, for example, when the lobes grow and elongate in later phases, or when the driving power of the jet declines, perhaps temporarily, for some reason. A young stellar population has been observed unambiguously with spectral features including typical absorption lines in the redshift $z=3.8$ radio galaxy 4C41.17 [142,143].

Generally similar physics seems to be at play for lower power jet events [144], and the emission line region in the X-ray cavities of the nearby radio galaxy 3CR 196.1 could also be a similar phenomenon [145].

6. Associated Absorption Line Systems

At the higher redshifts ($z\gtrsim 2$), emission line halos become well-observable in Ly $\alpha$. A typical feature of small (≲50 kpc) radio galaxies is that one or more narrower absorption lines are seen against the Ly $\alpha$ emission [120,146]. The absorption occurs much more frequently than expected from the Ly $\alpha$ forest. Some absorbers were confirmed at high spectral resolution [147], and integral field spectroscopy demonstrated a coherent absorption system across the Ly $\alpha$-emitting regions of several galaxies [148,149,150].

Different ideas for the physical interpretation of these absorption systems have been reviewed by Krause (2005) [151]. Their small velocity width suggests that they are probably thin shells around the host galaxies, rather than extended parts of the halo. The absorbers are probably partially photoionised [152,153] and their low emission suggests that the power of the process driving them into the halo is probably equally low. This makes models that use the jet as a driver of the shell [154] more difficult. A consistent scenario was developed and supported by hydrodynamic simulations in Ref. [151]. In the halos around massive high-redshift galaxies, one expects tenuous gas with a lower temperature and cooling time than in the intracluster or intragroup medium at a lower redshift, where many sources that have been observed in some detail are found. Bow shocks driven by any star-formation-related galactic wind will hence efficiently cool and sweep up the halo gas similar to the well-known snow-plough phase for stellar winds. When a jet then hits this galactic wind shell, it is stopped for some time until the radio lobes fill the space inside the wind shell. The enhanced pressure then accelerates the wind shell, which subsequently fragments via the Rayleigh–Taylor instability. The shell gas mixes turbulently into the radio lobes while falling into the galaxy. Radio sources that are greater than the wind shells will thus experience no more associated absorption. The model is further corroborated by the discovery of similar absorption systems around high-redshift galaxies with no jets [155,156,157,158]. Such shells could also form a significant amount of stars, probably in massive clusters [154].

7. Indirect Feedback via Halo Heating

A hydrostatic, though probably tenuous gas halo may be expected in galaxies with stellar masses above $\approx {10}^9{M}_{\odot }$ [105,159]. Once the jet escapes the dense interstellar medium of the host galaxy, it may heat the halo gas and thus prevent it from cooling and subsequent condensation in the given galaxy where it could then have formed stars. Strong, close to isotropic shocks will heat a small, inner part of the halo gas, before the shock strength declines e.g., [70,72]. Changing jet direction, perhaps due to mergers of supermassive black holes after galaxy mergers, may plausibly help heating halos of gas-poor (elliptical) galaxies or the intracluster medium [160,161,162]. In a star-forming and therefore gas-rich galaxy, much of the jet duty cycle compare, e.g., [26,163] might be spent interacting with the ISM, rather than with the halo, if the jet axis is substantially misaligned with the minor axis of the galaxy. However, once a jet escapes the dense interstellar medium of the host galaxy, it may contribute to heating gaseous halos around star-forming galaxies in a similar way as known from the intracluster medium, including via rising bubbles and sound waves triggered by the dynamics inside radio lobes e.g., [164,165,166]. Taking into account the impact of jets on the surrounding gas, Rouf et al. (2017) [167] are able to explain the mass and radio luminosity functions of massive galaxies in a semi-analytical cosmological model.

Extended gas halos have been detected in Ly$\alpha$ around high-redshift radio galaxies beyond the radio structures [168,169]. The gas halos show quiescent kinematics with a rotating component. They have been interpreted in terms of clumps of cold gas possibly ionised by a hidden quasar in such radio galaxies. Lyman $\alpha$ emission is, however, also expected from hydrostatic halo gas at somewhat lower temperatures than the X-ray clusters known from lower redshifts. High-redshift radio galaxies are massive galaxies that often go through episodes of strong star formation [170,171]. It is at least plausible that cooling of halo gas strongly contributes to fuelling their star formation. The extended radio sources seen in these galaxies are evidence that halo heating is taking place in these galaxies, likely cutting off their gas supplies to at least some degree.

8. Summary

The interaction of jets with the ISM in their host galaxies is rich in interesting processes. Informed by a host of new, excellent observations and significant simulation efforts, the last few decades have seen a consistent gain in our understanding of jet feedback. We now start to be in a position where we can put together the pieces of the jigsaw and tell a coherent story. I summarise the scenario developed in this review below and in the sketch in Figure 4.

Jet feedback in star-forming galaxies produces interesting effects on all scales. This is because the jet plasma is efficiently shocked and isotropised by its interaction with molecular and other clouds. Dense gas on scales of tens of parsecs around a supermassive black hole may be compressed and accelerated, such that high-velocity stars are formed (Section 3). Complex interaction in the central kiloparsec leads to triggered star formation and outflows from this central zone (Section 3). Inclined radio sources can, of course, have more impact. The combination of central pressure and radio lobes surrounding the galaxy should then enhance star formation in the rest of the galaxy. Enhanced stellar feedback and the removal of dense halo gas may enhance gaseous outflows from those outer parts of the galaxy, such that eventually, most of the ISM takes part in the outflow. Emission line lobes around smaller radio galaxies could be the observational manifestation of this. (Section 5). Even more emission line gases would be produced if galactic wind shells produced by a snow plough effect in a gaseous halo with the right conditions is overrun and turbulently entrained by the radio lobes (Section 6). The latter seems to happen frequently for massive high-redshift radio galaxies, but the general process is also found at a lower redshift for weaker radio sources. Eventually, radio lobes detach from the host galaxy, taking their load of metal-enriched gas with them to further spread it into the intergalactic medium. Approximately hydrostatic gas halos are probably present around galaxies above some mass. The hot halo has been clearly detected in the Milky Way. Jets likely contribute to heating these halos, and thus reduce the amount of fuel for star formation in the host galaxy.