An Extended Lyα Outflow from a Radio Galaxy at z = 3.7?

Outflows are believed to be the main driver through which active galactic nuclei (AGN) regulate the evolution of their host galaxies and affect the large-scale environment (AGN feedback), operating mostly at the epoch of the peak of star formation between redshift 2 and 3 (Hopkins 2006; Miley & De Breuck 2008). Outflows, extended on kiloparsec scales, powered by the AGN radiative pressure (quasar feedback mode) and/or radio jets (radio feedback mode), have been resolved in ionized, atomic, and molecular gas both in local (e.g., Rupke & Veilleux 2011; Cicone et al. 2012; Rodriguez Zaurin et al. 2013; Rupke & Veilleux 2013; Speranza et al. 2021; Ramos Almeida et al. 2022) and in high-redshift (radio-loud and radio-quiet) quasars (e.g., Alexander et al. 2010; Nesvadba et al. 2010; Maiolino et al. 2012; Cano-Díaz et al. 2012; Cresci et al. 2015). Characterizing and constraining the effects and extent to which AGN feedback affects star formation is one of the most important problems in modern astrophysics, as it has ties to the biggest questions in galaxy evolution and cosmology.

In this context, high-redshift radio galaxies (HzRGs) are particularly relevant as the nuclear regions are obscured along our line of sight due to a circumnuclear dusty structure (e.g., Antonucci 1993) or by the host galaxy itself. The lower nuclear continuum luminosities allow us to reduce the observational uncertainties produced by the subtraction of a prominent point-spread function (PSF), probing the host and the gas structure down to the limit of the spatial resolution of only a few kiloparsecs.

We started a comprehensive program of observations with the Multi Unit Spectroscopic Explorer (MUSE) at the Very Large Telescope to explore the properties of the ionized gas in HzRGs. In this Letter, we present the results obtained for TN J1049-1258, an HzRG at redshift 3.697 ± 0.004 (Bornancini et al. 2007), located at α = 10:49:06.2, δ = −12:58:19 (J2000). The scale factor at this redshift (assuming an H₀ = 69.6 and Ω_m = 0.286 cosmology) is 7.3 kpc arcsec⁻¹. Its radio flux measured at 74 MHz by the NRAO Very Large Array (VLA) Sky Survey is 5.18 ± 0.56 Jy while its radio spectral index is α = 1.4 (Condon et al. 1998). ⁹ The jet power, P_jet, can be obtained using the correlation between P_jet and radio luminosity, based on the measurement of the mechanical power of radio sources producing cavities in the surrounding hot gas as proposed by Cavagnolo et al. (2010). The radio luminosity of TN J1049-1258 at (rest frame) 327 MHz can be estimated as P₃₂₇ = 2.7 × 10⁴⁴ erg s⁻¹, which yields a jet power of P_jet ∼ 6 × 10⁴⁵ erg s⁻¹. A similar value is obtained by using the relation between radio and jet power from Willott et al. (1999). In the Karl G. Jansky VLA Sky Survey (Lacy et al. 2020), performed at 3 GHz with a resolution of 25, its radio emission is dominated by two radio components separated by 11'' (80 kpc) oriented at the position angle −80°, with the host located close to their midpoint. Its radio luminosity at the rest-frame frequency of 500 MHz is 8.7 × 10²⁸ W Hz⁻¹, making TN J1049-1258 one of the most luminous radio sources known (see Miley & De Breuck 2008). The flux at 3.5 μm, corresponding to the optical emission in the rest frame, measured by Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010) of TN J1049-1258 indicates a luminosity of 2 × 10¹¹ L_⊙, suggesting that its host is already a very massive galaxy.

The observations were carried out with the Multi Unit Spectroscopic Explorer (MUSE) on 2022 January 29 as part of the program ID:108.22FU. Four separate observations were performed, between which the telescope was rotated 90° to reject cosmic rays, for a total of 2800 s of exposure time. The seeing measured from several point sources in the field of view was estimated as 071 ± 004. We used the ESO MUSE pipeline (version 2.8.7) to obtain a fully reduced and calibrated data cube (Weilbacher et al. 2020). We derived an absolute astrometric calibration of the MUSE data by cross-matching visible sources in both MUSE and Panoramic Survey and Rapid Response System (Pan-STARRS; Chambers et al. 2016). Based on the scatter of the sources considered, we estimate the derived astrometry to have an error of ∼02.

After first inspecting the nuclear spectrum on a large aperture around the peak of Lyα emission, we found two emission lines separated by about 50 Å. To fully characterize both the emission line morphology and the gas kinematics, we then performed a two Gaussian component fit in each spaxel (see Figure 1 for the nuclear and off-nuclear spectra). Aiming to increase the signal-to-noise ratio (S/N) per pixel, the cube was spatially rebinned into 2 × 2 pixels, resulting in a 04 × 04 area. The modeling tool from Astropy's Python package (Astropy Collaboration et al. 2013, 2018; Price-Whelan et al. 2022) was used to model and fit both the continuum and emission lines. For the continuum, we used a linear least squares fit, while the emission lines were modeled with a Gaussian using a Levenberg–Marquardt least squares fit.

Zoom In Zoom Out Reset image size

From the fit, we obtained the distribution of total flux and first and second moments for both components, while the uncertainties of the parameters were obtained through a Monte Carlo simulation. Each realization consists of adding flux variations to each spectral pixel, with the value of the variations taken at random from a normal distribution centered at the pixel noise. From each realization, a value for the relevant parameters (line center, amplitude, width) is obtained. After a set number of iterations (100 in our case) we obtain a value distribution for each of these parameters, from which we extract the mean (parameter value) and 1σ standard deviation (parameter uncertainty). The pixel noise was calculated as the 1σ standard deviation of the flux in a 50 nm interval centered on Lyα from a section of empty sky of equal size as the cube extracted for the line analysis (80 × 80 pixels or 16'' × 16''). All emission line maps have been obtained by setting a signal-to-noise threshold at 3.

The continuum map was obtained using multiple intervals within MUSE's wavelength range (475–935 nm, 101–198 nm rest frame) over a total range of 144 nm, selected to avoid both sky lines as well as object emission lines.

We estimated the redshift of TN J1049-1258 from the fit of the main component of Lyα line in the nuclear spectrum, which corresponded to z = 3.697 ± 0.001, the same value obtained by Bornancini et al. (2007). However, it must be noted that the Lyα line peak has been observed to be redshifted from the systemic velocity (measured by the Hα emission) by ∼450 km s⁻¹ on average, and by as much as ∼900 km s⁻¹ (e.g., Steidel et al. 2010).

The emission maps of both components and all relevant velocity and velocity dispersion maps are shown in Figure 2. The velocity field estimated from the host galaxy's Lyα emission presents a large gradient, with a velocity range of ∼900 km s⁻¹ and a velocity dispersion ranging from 500 to 1500 km s⁻¹, that can be ascribed to either rotation or radial motion. While there is a clear spatial correlation between width and relative velocity, interpreting the absolute values of the velocity dispersion is made difficult by the resonant character of Lyα emission. Neutral hydrogen in the line of sight of the emitting gas will result in peaks and troughs, deeply modifying the original line profile depending on the density, distribution, and velocity of the neutral H gas.

Zoom In Zoom Out Reset image size

The redshifted component has an elongated structure extending ∼25, i.e., ∼18.5 kpc, measured as the distance between the two farthest pixels with S/N>3, along the same position angle of the radio structure. Its relative velocity, measured using the line profile in the off-nuclear spectrum (see Figure 1) is v = 2250 ± 60 km s⁻¹ and it remains remarkably similar throughout the spatial extent of the emission zone. In a similar fashion, the velocity dispersion values are within a range of ∼70 km s⁻¹, with this red component being much narrower than the one in the host galaxy; its FWHM measured from the off-nuclear spectrum is σ_RC = 430 ± 20 km s⁻¹.

In the region cospatial to the Lyα red component we also detected emission from the He ii λ1640 line (see Figure 3). This line is blueshifted by 1404 ± 9 km s⁻¹ (value obtained from the off-nuclear spectrum shown in Figure 1) with respect to the Lyα emission. Its velocity and velocity dispersion maps are shown in Figure 2. From this spectrum. we obtain a line ratio Lyα/He ii = 1.5 ± 0.3, which is at the very low end of the values measured in HzRGs (see Villar-Martín et al. 2007).

Zoom In Zoom Out Reset image size

UV continuum emission, within the rest-frame range ∼1000–2000 Å, (see Figure 4) is observed on the west side of the host galaxy, cospatial with the Lyα red component. The source observed ∼5'' south of the radio galaxy (RG from hereon) is a galaxy at z = 0.87, based on the identification of the [O ii]λ λ 3726,3729 doublet as well as Hδ and Hγ in its spectrum. The brightest UV source in the field is another potential companion we named G1, located ∼8'' (∼60 kpc) southeast of the host galaxy. Its Lyα emission was detected at a very similar redshift (see Figure 5, top panel) and its high-redshift identification is also supported by it being a g-dropout based on the photometry from Pan-STARRS (Chambers et al. 2016). ¹⁰

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To the east of the radio galaxy, another patch of Lyα emission can be seen in the top panel of Figure 2. This is associated with a companion galaxy we named G2, whose identification is possible thanks to the presence of Lyα (see Figure 5, bottom panel), He ii λ1640, and C iv λ λ1548, 1550 in its spectrum. Its redshift, z = 3.685 ± 0.002, was estimated from the He ii λ1640 emission, which is a nonresonant line. The presence of G1 and G2 suggests TN J1049-1258 may be harbored within a protocluster.

The main result of these observations is the presence of extended Lyα emission redshifted by ∼2250 km s⁻¹ and we can envisage two possibilities for its origin:

(1) A companion galaxy. The presence of UV continuum emission cospatial with the Lyα region suggests that this might be associated with another galaxy at a similar redshift. This interpretation is also supported by the low velocity dispersion (σ = 430 ± 20 km s⁻¹) of the Lyα emission. The Lyα equivalent width (EW) of 6.6 ± 2.0 Å does not provide useful information as to the origin of the line emission since the EW of Lyman alpha emitters spans a considerably wide range, with measurements ranging from a few angstroms to several hundreds (see Kerutt et al. 2022). However, studies of high-redshift galaxies found that the typical radius of a z ∼4 galaxy is 02, that is, 1.5 kpc (Bouwens et al. 2004; Ferguson et al. 2004). The FWHM of the Lyα red component emission is significantly larger: by fitting its spatial distribution with a Gaussian we obtain a size of 179 ± 014, equivalent to a deconvolved extension of R = 12.0 ± 0.9 kpc. We must note, however, that G1 is also extended, with a deconvolved size of 7.3 ± 0.5 kpc. The strongest argument against this interpretation is the blueshift of ∼1400 km s⁻¹, which the He ii λ1640 line presents with respect to the extended Lyα line, as it greatly exceeds the range of velocity offsets between Lyα and nebular lines observed in high-redshift sources (Steidel et al. 2010), arguing in favor of a different origin.

(2) An outflow of ionized gas. This hypothesis is supported by simulations of the theoretical profile of Lyα (Ahn 2004; Verhamme et al. 2006; Behrens et al. 2014). In particular, the models by Behrens et al. (2014) show that a bipolar expanding shell, a reasonable approximation for a nuclear outflow, can produce a secondary redshifted peak. This unusual line profile is explained by the fact that Lyα is a resonant emission line, subject to the effects of radiative transfer due to the large cross-section of interaction between Lyα photons and neutral hydrogen atoms. As a consequence, the ubiquity of neutral hydrogen not only within galaxies but in the intergalactic medium along our line of sight prevents most, if not all, light blueward of the central Lyα wavelength from emerging and, as is the case here, only the red side is observed. These effects can also explain the large velocity difference between the Lyα and the He ii.

Additionally, the presence of UV continuum aligned with the radio axis is a common feature of HzRGs and has been explained as due to a combination of scattered nuclear light, fast shocks, and jet-induced star formation (see, e.g., McCarthy 1993 or Miley & De Breuck 2008), which also points to an in-host origin for the emission.

With all of these considerations taken into account, we conclude that the outflow hypothesis is the more plausible scenario.

We estimated the mass of the outflow, M_out, from the Hβ luminosity, L_Hβ , using the relation derived by Osterbrock (1989):

$$\begin{eqnarray}&&{M}_{\mathrm{out}}=7.5\,\times {10}^{-3}\left(\displaystyle \frac{{10}^{4}}{{n}_{e}}\displaystyle \frac{{L}_{{\rm{H}}\beta }}{{L}_{\odot }}\right){M}_{\odot },\end{eqnarray} \tag{ 1 }$$

assuming the ratios Lyα/ Hα = 8.7 (Sobral & Matthee 2019) and Hα/Hβ = 2.86, and a gas density of n_e = 200 cm⁻³, which enables a comparison between the energetics of this object and a sample of low redshift radio-loud AGN that utilizes this same value (Speranza et al. 2022).

We derived the following physical quantities (see, e.g., Fiore et al. 2017) for the outflow:

• 1.  
  mass outflow rate: $\dot{M}=3{v}_{\mathrm{out}}\tfrac{{M}_{{\rm{H}}\beta }}{{R}_{\mathrm{out}}}=3.4{M}_{\odot }$ yr⁻¹
• 2.  
  kinetic energy: ${{\rm{E}}}_{\mathrm{kin}}=\tfrac{1}{2}{M}_{{\rm{H}}\beta }{v}_{\mathrm{out}}^{2}={10}^{56.6}$ erg
• 3.  
  kinetic power ${\dot{E}}_{\mathrm{kin}}=\tfrac{1}{2}\dot{M}{v}_{\mathrm{out}}^{2}={10}^{42.4}$ erg s⁻¹

where v_out = 2250 km s⁻¹ and R_out = 18.5 kpc are the outflow velocity and full extension, respectively.

These estimates should be considered as lower limits for two reasons: first of all, we cannot correct for the effects of projection and, in addition, there might be a contribution of collisional excitation to the production of emission lines.

The star formation rate (SFR) can be estimated from the UV and far-infrared emission of the source. While the UV light directly traces the young stellar population, the far-infrared emission is produced by warm dust heated by the UV photons emitted by young stars. We found the UV contribution to the SFR estimate for the RG to be negligible; it is also worth mentioning that UV continuum is subject to various effects that make it a more unreliable tracer at higher redshift (Wilkins et al. 2012).

Conversely, TN J1049-1258 has been detected by pointed far-infrared observations from the Spectral and Photometric Imaging Receiver onboard the Herschel satellite (Griffin et al. 2010; Poglitsch et al. 2010). At 250 and 350 μm, the observed fluxes are 63.0±9.5 and 67.7±11.8 mJy, respectively. These frequencies correspond to ∼50 and 70 μm in the rest frame, which coincides with the peak of the warm dust emission, heated by star-forming regions. Measurements from WISE were obtained on the W1 and W2 passbands (W3 and W4 were upper limits). From a fit to the spectral energy distribution using the WISE and Herschel measurements (see Balmaverde et al. 2016 for details) the derived total far-infrared luminosity is L_{(8−1000μm)}=3.2 × 10¹³ L_⊙, which makes this source a hyperluminous infrared galaxy. It corresponds to a prodigious star formation rate of SFR∼4,500 M_⊙ yr⁻¹, among the highest values ever estimated (see the compilation of SFRs from Lagache et al. 2018). However, due to the low spatial resolving power of Herschel (the Herschel data reduction guide suggested to adopt an aperture photometry radius for the 250 and 350 μm images of 22'' and 30'', respectively), both companions and the z = 0.87 galaxy are included within the aperture, and thus there may be contamination from these sources. Additionally, there might be an AGN contribution. However, the torus emission peaks at 20–30 μm (Nenkova et al. 2008) and, in order to reproduce the far-infrared fluxes, this component would exceed the WISE measurements. We conclude that the AGN can only be a minor contributor to the Herschel fluxes.

Regarding the energetics of the system, the jet power is ∼4 orders of magnitude larger than the outflow kinetic power. Speranza et al. (2021) estimated the outflow kinetic power of a sample of radio galaxies at low redshift: while they measured values of $\dot{{E}_{\mathrm{kin}}}$ similar to that of TN J1049-1258, they found significantly larger values of $\dot{{E}_{\mathrm{kin}}}/{P}_{\mathrm{jet}}$, typically ∼0.1. This suggests that the relativistic jets and the outflow may be decoupled and produced by two independent acceleration mechanisms: the jets might be driven by the magnetic field in the innermost regions of the accretion disk, while the outflow is powered by radiation pressure. Alternatively, the ionized gas traced by the Lyα emission might just be the tip of the iceberg of a larger amount of outflowing gas, because this line is a poor tracer in terms of representative masses and luminosities. Dedicated radiative transfer simulations with geometries and parameters resembling those observed in TN 1049-1258 will be essential to fully model the results of these observations. To obtain a comprehensive view of the energetic of the outflow, we should also consider other ionized gas tracers, e.g., the optical [O iii] line, unaffected by resonance and absorption effects. Moreover, a relevant contribution to the energetic of the outflow is expected to originate from the molecular gas component, which can be studied using lines produced by the CO molecule transitions or the [C ii]158 μm emission.

Another source of uncertainty in the outflow estimates is the ionization mechanisms in play, as collisional excitation may contribute. In fact, the low Lyα/He ii ratio observed in the extended emission is not dissimilar to the one obtained in Scarlata et al. (2009) where, within another high-redshift (z = 2.38) system of Lyman alpha emitters, one of the sources displayed a comparably low ratio (Lyα/He ii = 2.2 ± 1.0) and the mechanism of origin was suspected to be accretion of cold gas onto the dark matter halo, resulting in collisional excitation.

Despite the large amount of energy injected into the host galaxy by the ionized outflow and the relativistic jets, TN J1049-1258 displays globally a very large SFR, although the companion galaxies might also contribute to the Herschel-measured flux. Rest frame far-infrared observations at high spatial resolution, e.g., those that can be produced by facilities such as the Atacama Large Millimeter/submillimeter Array, will be fundamental to obtain a more accurate estimate of the SFR in the host galaxy, and if possible, resolve its spatial distribution in a bid to better understand the interplay between jets, outflows, and gas reservoirs.

The possibility of outflows and more generally, jets, inducing star formation has been studied in both galactic (see Bicknell et al. 2000; Miley & De Breuck 2008; Capetti et al. 2022) and cosmological scales (Gilli et al. 2019), as well as simulations (Gaibler et al. 2012). As such, it is possible that the observed star formation from the RG is either induced or enhanced by the jet. Spatially resolved observations of cold molecular gas could answer questions such as where the star formation is occurring, whether jet-driven shocks are pressurizing gas bubbles in the intergalactic medium and causing fragmentation, or whether the expanding cocoon of the jet is inducing similar phenomena within the host galaxy itself. The cospatiality between the UV continuum and outflow suggests the former, though a mixture of both is not out of the question.

The presence of TN J1049-1258 inside a protocluster is within expectations. Studies have found that up to 75% of powerful HzRGs reside within these structures (Venemans et al. 2006). This is understandable considering that the powerful AGN responsible for this radio emission requires massive galaxies, which, in turn, inhabit rich, dense environments that act as fertile soil for cluster-like structures (Hatch et al. 2014). There is a great deal of relevance in protoclusters within the ΛCDM model, as they represent massive structures hypothetically linked by large dark matter halos: understanding how high-z and nearby galaxy clusters relate to each other is essential in unveiling the behavior and characteristics of dark matter. Their evolution through the cosmic ages as well as their relationship to powerful RLAGN, which likely become the brightest cluster galaxies, are ongoing and widely studied topics within cosmology and high-redshift astrophysics.

Based on observations collected at the European Southern Observatory under ESO programme 108.22FU. CRA acknowledges the projects "Feeding and feedback in active galaxies", with reference PID2019-106027GB-C42, funded by MICINN-AEI/10.13039/501100011033, and 'Quantifying the impact of quasar feedback on galaxy evolution', with reference EUR2020-112266, funded by MICINN-AEI/10.13039/501100011033 and the European Union NextGenerationEU/PRTR.

Facilities: VLT:Yepun - , Herschel - European Space Agency's Herschel space observatory, VLA - Very Large Array, WISE - Wide-field Infrared Survey Explorer.