Radio Morphology of Red Geysers

We use optical data primarily from the recently completed SDSS-IV MaNGA survey (Blanton et al. 2017; Bundy et al. 2015; Drory et al. 2015; Law et al. 2015; Yan et al. 2016; Albareti et al. 2017). MaNGA is an integral-field spectroscopic survey that provides spatially resolved spectroscopy for nearby galaxies (z ∼ 0.03) with an effective spatial resolution of 25 (full width at half maximum; FWHM). The MaNGA survey uses the SDSS 2.5 m telescope in spectroscopic mode (Gunn et al. 2006) and the two dual-channel BOSS spectrographs (Smee et al. 2013) that provide continuous wavelength coverage from the near-UV to the near-IR: 3600–10,000 Å. The spectral resolution varies from R ∼ 1400 at 4000 Å to R ∼ 2600 at 9000 Å. An r-band S/N of 4–8 Å⁻¹ is achieved in the outskirts (i.e., 1 − 2 R_e ) of target galaxies with an integration time of approximately 3 hr. MaNGA has observed more than 10,000 galaxies with $\mathrm{log}\,({M}_{\star }/{M}_{\odot })\geqslant 8$ across ∼2700 deg² over its 6 yr duration. In order to balance radial coverage versus spatial resolution, MaNGA observes two-thirds of its galaxy sample to ∼1.5 R_e and one-third to 2.5 R_e . The MaNGA target selection is described in detail in Wake et al. (2017).

The raw data are processed with the MaNGA Data Reduction Pipeline (DRP, Law et al. 2016). In this work, we use the MaNGA sample and data products drawn from the MaNGA Product Launch-9 (MPL-9) and Data Release 16 (DR16, Ahumada et al. 2020). We use spectral measurements and other analyses carried out by MaNGA Data Analysis Pipeline (DAP), specifically, version 2.3.0. The data we use in this work are based on a DAP analysis of each spaxel in the MaNGA data cubes. The DAP first fits the stellar continuum of each spaxel to determine the stellar kinematics using the penalized pixel-fitting algorithm pPXF (Cappellari & Emsellem 2004; Cappellari 2017) and templates based on the MILES stellar library (Falcón-Barroso et al. 2011). The templates are a hierarchically clustered distillation of the full MILES stellar library into 49 templates. This small set of templates provide statistically equivalent fits to those that use the full library of 985 spectra in the MILES stellar library. The emission-line regions are masked during this fit. The DAP then subtracts the result of the stellar continuum modeling to provide a (nearly) continuum-free spectrum that is used to fit the nebular emission lines. This version of the DAP treated each line independently, fitting each for its flux, Doppler shift, and width, assuming a Gaussian profile shape. The final output from the DAP are gas and stellar kinematics, emission-line properties, and stellar absorption indices. All the spatially resolved 2D maps shown in the paper are outputs from the DAP with a hybrid binning scheme. An overview of the DAP used for DR15 and its products is described by Westfall et al. (2019), and assessments of its emission-line fitting approach are described by Belfiore et al. (2019). All the integrated quantities reported in this paper are S/N weighted average taken over one effective radius.

We use ancillary data drawn from the NASA-Sloan Atlas ¹⁵ (NSA) catalog, which reanalyzes images and derives morphological parameters for local galaxies observed in SDSS imaging. It compiles spectroscopic redshifts, UV photometry (from GALEX, Martin et al. 2005), stellar masses, and structural parameters. We have specifically used spectroscopic redshifts and stellar masses from the NSA catalog. The SFRs are derived from Salim et al. (2016), who used GALEX-SDSS and WISE to perform UV, optical, and IR SED fitting.

The LOFAR Two-meter Sky Survey (LoTSS, Shimwell et al. 2017, 2019) is an ongoing sensitive, high-resolution survey that will cover the whole northern sky with 3168 pointings in the frequency range between 120 and 168 MHz. The LoTSS first data release (DR1, Shimwell et al. 2019) covers 424 square degrees centered on the Hobby Eberly Telescope Dark Energy Experiment (HETDEX; Hill et al. 2008) Spring Field region (R.A. 10h45m00s to 15h30m00s and decl. 45°00'00'' to 57°00'00'') and contains over 300,000 sources with an S/N >5. The median sensitivity is ∼71 μJy/beam, and 95% of the area in the DR1 release has an rms noise level below 150 μJy/beam. The angular resolution is 6'' and the positional accuracy is within 02 for high S/N sources; the positional accuracy increases to ∼05 for the faintest sources with a flux density lower than 0.6 mJy. The source density is a factor of ∼10 higher than the most sensitive existing very wide-area radio-continuum surveys such as the NRAO VLA Sky Survey (NVSS, Condon et al. 1998), Faint Images of the Radio Sky at Twenty-Centimeters (FIRST, Becker et al. 1995), Sydney University Molonglo Sky Survey (SUMSS, Bock et al. 1999; Mauch et al. 2003), and WEsterbork Northern Sky Survey (WENSS, Rengelink et al. 1997).

The second LoTSS data release (DR2), to be released publicly in 2021, consists of two contiguous fields at high Galactic latitude centered around 0h and 13h and covering approximately 5,700 square degrees (Shimwell et al. in preparation). DR2 provides fully calibrated mosaics at the same 6'' resolution as DR1, and images can also be obtained from individual LoTSS pointings, outside the DR2 area. For the red geysers in the existing LoTSS coverage, including fields not part of the DR2 release, we obtain the fluxes and sizes from either (a) the internally released catalog (Shimwell et al. in prep) or (b) similarly generated catalogs for small areas of individual pointings around our target objects. In both cases, the flux-scale correction described by Hardcastle et al. (2021) and Shimwell et al. (in prep) is applied to the data so that flux densities are as close as possible to the flux scale of Roger et al. (1973). The residual flux-scale uncertainty lies between 5 and 10% for the DR2 area (Hardcastle et al. 2021) and is likely to be ∼10% for individual fields.

The total number of red geysers with currently available LoTSS data is 103, which is about 74% of the parent 140 red geyser sample. The list of LOFAR-detected red geysers is presented in Table 1.

The Very Large Array (VLA) Faint Images of the Radio Sky at Twenty Centimeters (FIRST, Becker et al. 1995) survey is a systematic survey over 10,000 square degrees of the North and South Galactic Caps at frequency channels centered at 1.36 GHz and 1.4 GHz. FIRST uses the VLA in B-configuration and achieves an angular resolution of 5'', and the survey is insensitive to structures larger than ∼60'' as it is carried out in the VLA B-configuration. The source-detection threshold is ∼1 mJy, corresponding to a source density of ∼90 sources deg⁻². The astrometric accuracy of each source is 0.5–1'' at the source-detection threshold. Because the FIRST survey area was designed to overlap with the Sloan Digital Sky Survey (SDSS, York et al. 2000; Abazajian et al. 2009), most MaNGA targets have FIRST data coverage. However, the 1 mJy threshold results in nondetections for most MaNGA galaxies. For each pointing center, there are 12 adjacent single-field pointings that are coadded to produce the final FIRST image. Sources are extracted from coadded reduced images and fit by 2D Gaussians to derive the peak flux, integrated-flux densities, and size information (Becker et al. 1995). The current FIRST catalog is accessible from the FIRST search page. The full images are available from ftp://archive.stsci.edu/pub/vla_first/data.

The Very Large Array Sky Survey (VLASS, Lacy et al. 2020) is a community-driven initiative to carry out a synoptic radio-sky survey using the Karl G. Jansky Very Large Array (VLA). VLASS will eventually use ∼5500 hr to cover the whole sky visible at the VLA (δ > − 40 deg), observing a total of 33,885 deg² at an angular resolution of ∼25. The data will be acquired in three epochs and will cover the frequency range 24 GHz in 2 MHz channels, with calibrated polarimetry in Stokes I, Q, and U, providing wideband spectral and polarimetric data for a myriad of targets and source types. The angular resolution is ∼2.5 arcsec, and the survey is insensitive to structures larger than ∼30'' as it is carried out in the VLA B-configuration. The 1σ sensitivity goal for a single pass is 120 μJy, while it is 69 μJy when all three epochs are combined. Thus the VLASS is an all-sky radio-sky survey with a unique combination of high angular resolution, high sensitivity, full linear Stokes polarimetry, time-domain coverage, and wide bandwidth. Observing began in September 2017, and the survey will complete observations in 2024.

The LoTSS DR2 footprint contains 93 of the parent sample of 140 red geyser galaxies. Additionally, 10 more sources are contained within individual LoTSS pointings outside the DR2 area. Thus the total number of red geysers with currently available LOFAR data is 103, which is about 75% of the complete red geyser sample. Thirty-four ± 6 out of those 103 sources (33% ± 5.5%) are found to be radio-detected at the frequencies observed by LOFAR (∼150 MHz), where quoted errors are obtained from standard Poisson statistics.

We cross-matched the FIRST catalog with our red geyser sample with a cross-matching radius of 10''. For visibly extended radio sources in the FIRST image, we extended the cross-matching radius up to a maximum of 1' but restricted the central component of the detected source to lie within <10'' of the galaxy center. We find that 29 ± 5 out of 140 red geyser galaxies are detected at a frequency of 1.4 GHz with a detection fraction of 21% ± 3.5%. This FIRST detection rate is roughly in agreement with our previous work (Roy et al. 2018), which noted a 15% radio-detection rate from the FIRST survey. However, Roy et al. (2018) used a preliminary red geyser sample of 84 sources drawn from an earlier MaNGA data release—MPL-5. The current updated red geyser sample from MPL-9 has increased the total number of red geysers by 56 sources, which results in an additional 17 FIRST detections for this work.

Last, we cross-matched the parent red geyser sample with the VLASS (3 GHz) survey using the same cross-matching radius as the FIRST survey and found 29 radio detections, which is a similar detection rate to FIRST. Thirteen red geysers are detected in LOFAR but not in VLASS and FIRST. On the other hand, eight sources are detected in FIRST and VLASS, but are not within the LoTSS field of view. Twenty-one ± 4 out of 140 red geysers (∼15% ± 4%) have simultaneous radio detections from all three surveys, which are used for calculation of spectral indices (discussed in Section 5.3). Forty-two ± 6 red geysers (30% ± 4%) are detected in at least one of the LoTSS, FIRST, or VLASS surveys, and they are listed in Table 1.

In order to understand how, if at all, the radio-detected red geyser galaxies differ intrinsically from those that are nondetected, we compare the host galaxy properties of the respective samples of interest. Figure 2 shows the histograms of SDSS I-band magnitude (upper left), Hα luminosity (upper right), stellar mass (lower left), and SFR (lower right) of non-radio-detected red geysers (in blue) and 42 radio-detected red geysers (in salmon)—detected in at least one of the LoTSS, FIRST, or VLASS surveys. The SFR estimates are obtained from the Salim et al. (2016) catalog, while the magnitude and the stellar mass reported here are acquired from the NSA catalog. We note that the distributions of I-band magnitude, Hα luminosity, and stellar mass are quite different between the radio-detected and nondetected sample. This is statistically confirmed by a Kolmogorov–Smirnov (KS) test, which rejects the null hypothesis that the radio-detected and nondetected samples show similar distributions. This is shown by extremely small p-values of 1.48 × 10⁻⁶, 1.84 × 10⁻⁶, and 1.05 × 10⁻⁶ for I-band magnitude, Hα luminosity, and stellar mass distributions, respectively. However, for the SFR distributions, we cannot reject the null hypothesis at a level < 1% (p = 0.06), signifying similar distributions of radio and non-radio-detected galaxies. The radio-detected sources are in general brighter in the SDSS I band (mean M_I = − 22.5) and more massive (mean log₁₀ M_⋆ ∼ 11 M_⊙) than the non-radio-detected galaxies (mean M_I = − 21 and log₁₀ M_⋆ ∼ 10.5 M_⊙). The Hα luminosity in the radio-detected sample also tend to be higher on average (mean log₁₀ L_Hα = 40.5 erg s⁻¹) than the non-radio-detected sample (log₁₀ L_Hα = 40 erg s⁻¹), possibly implying that the galaxies with higher amount of ionized gas and thus a more prominent Hα bisymmetric pattern are more likely to have enhanced radio emission. The SFR distribution, however, is unchanged regardless of radio-detection, indicating no underlying correlation between them.

Zoom In Zoom Out Reset image size

Figure 3 shows the radio luminosities at 1.4 GHz versus the observed total [O iii] luminosities extracted from the central 3'' as observed by the SDSS fiber of the 42 radio-detected red geysers (in black). Because LOFAR is most sensitive to fainter and extended radio emissions, it provides a more accurate estimate of the total flux density than FIRST. Hence we convert flux densities (S) measured at 144 MHz from LoTSS into 1.4 GHz for the LOFAR-detected sources in order to compare to existing literature, assuming a spectral index (α) of −0.7 using S_ν ∼ ν^α (Condon et al. 2002). We use FIRST-measured flux densities for the eight sources outside the LoTSS footprint. The radio luminosities thus obtained are compared to the Mullaney et al. (2013) z < 0.2 AGN population, plotted as green circles. The blue line marks the division between radio-loud and radio-quiet sources from Xu et al. (1999). A majority of our radio red geyser sample are classified as radio-quiet according to this definition, with only three galaxies lying in the radio-loud regime. If we assume ${L}_{1.4\mathrm{GHz}}\sim {L}_{[{\rm{O}}\,{\rm\small{}}{iii}]}^{\beta }$ similar to Xu et al. (1999), β ranges from 0.53 to 0.60 in the red geyser sample, with a mean value of 0.56. This is consistent with β ∼ 0.5 as reported in Xu et al. (1999).

Zoom In Zoom Out Reset image size

In addition to using the criterion of Xu et al. (1999), we compute the commonly used radio-loudness parameter R, the ratio of radio to optical brightness (Kellermann et al. 1989), to differentiate between radio-loud and radio-quiet sources in our sample. Similar to Ivezić et al. (2002) and Jarvis et al. (2021), we calculate R using the radio flux density at 1.4 GHz calculated in a similar way as mentioned above, and the SDSS i-band apparent magnitude using the following equation:

$$\begin{eqnarray}&&R=0.4({m}_{i}-t).\end{eqnarray} \tag{ 1 }$$

Here m_i is the Petrosian i-band apparent magnitude from SDSS DR16 (Ahumada et al. 2020). The Petrosian magnitudes used here recover essentially all of the flux of an exponential galaxy profile and about 80% of the flux for a de Vaucouleurs profile. Here, t is the AB radio magnitude, defined as

$$\begin{eqnarray}&&t=-2.5\ \mathrm{log}\left(\displaystyle \frac{{{\rm{S}}}_{1.4\ \mathrm{GHz}}}{3631\mathrm{Jy}}\right),\end{eqnarray} \tag{ 2 }$$

where S_{1.4 GHz} is the radio flux density (in Jy) measured at 1.4 GHz. We find that according to the R parameter criterion, 4 out of 42 radio-detected red geysers would be classified as radio-loud with R > 1. This includes the three galaxies that were identified as radio-loud according to Xu et al. (1999) and a fourth galaxy that was a borderline case—lying just below the division line separating radio-loud and radio-quiet population (see Figure 3). Although the R-parameter criterion is generally implemented in quasars with typical R values for radio-loud sources going up to 2.8 (Ivezić et al. 2002), the R values reported here simply quantify the relative contribution of the radio luminosity over optical light. Ninety-four percent of the radio-detected red geysers are radio-quiet according to both criteria. This is consistent with Roy et al. (2018), who stated that the radio-detected red geysers occupy the low-luminosity end (L_1.4GHz < 10²³ W Hz⁻¹) of the radio population in the MaNGA quiescent galaxy sample.

The next section is dedicated to understanding the dominant mechanism responsible for the observed radio emission in these sources via physically motivated tests.

In Figure 3 we see that 39 out of 42 radio-detected red geysers in our sample would be classified as radio-quiet by the Xu et al. (1999) criterion. An important and significantly challenging follow-up question to address is whether the observed radio emission is associated with the central radio AGN or SF. Although, we have predicted in Roy et al. (2018) that the radio-continuum emission in red geysers is generally associated with central low-luminosity, radiatively inefficient radio-mode AGN, it is important to verify that interpretation in the light of other observations and in our increased sample of 42 radio-detected red geysers. Star-forming galaxies (SFGs) emit at radio wavelengths primarily due to synchrotron emission from shocks associated with supernovae (Klein et al. 2018), and hence their radio luminosity is expected to correlate broadly with the SFR. They generally display a diffuse clumpy radio emission not extending beyond the host galaxy with a steep spectral slope (Webster et al. 2021; Jarvis et al. 2019). On the other hand, radio emissions in radio AGNs are primarily dominated by a jet originating from the central supermassive black hole. Unlike megaparsec-scale radio jets in the centers of massive clusters and giant radio galaxies, the jets in low-luminosity AGN hosts are small scaled and confined near the very central region of the host galaxy (Jarvis et al. 2019; Venturi et al. 2021; Capetti et al. 2020; Webster et al. 2021), which often remains unresolved due to the low spatial resolution of various radio observations. This gives rise to compact or slightly extended radio sources with no visible lobes/ jets, which are hard to distinguish from SFGs.

In order to consider the possibility that SF might give rise to the detected radio emission ( ∼ 10²² W Hz⁻¹, Figure 3), we need to detect a significant amount of SF ( ∼ 1 − 5 M_⊙, Brown et al. 2017) in the red geyser galaxies. If no similar level of SF is detected, we can rule out SF and attribute the observed radio emission to be from the central radio AGN. We consider three diagnostic plots to classify the radio-detected sources as either SFGs or non-SFGs:

• 1.  
  Identification based on WISE colors, particularly in W3W2 (Yan et al. 2013). Star-forming galaxies possess W3 − W2 > 0.3.
• 2.  
  Using emission-line diagnostics, in particular the ratio of [O iii] 5007 and Hβ line fluxes, and that of [N ii] 6584 and Hα (Baldwin et al. 1981), referred to as the BPT method. Galaxies with $\mathrm{log}([O\,{\rm\small{}}{iii}]/H\beta )\lt 0.5$ and $\mathrm{log}([N\,{\rm\small{}}{ii}]/H\alpha )\lt -0.3$ lying below the Kauffmann et al. (2003) curve are SFGs.
• 3.  
  Using the relation between the 4000 Å break strength and radio luminosity per stellar mass (Best et al. 2005), hereafter referred to as the D4000 vs L_rad/M method. Galaxies with D4000 ≤ 1.6 are dominated by young stellar populations and hence constitute SFGs.

Specific WISE mid-infrared colors can be used to separate galaxies with and without SF. Thus, SFGs are separated from the typical hosts of radio AGN in their WISE colors, particularly in W2–W3 (4.6 to 12 micron color, Yan et al. 2013). Figure 4 upper left (panel a) shows a plot of W1W2 versus W2W3 mid-infrared WISE colors for 42 red geysers that are radio-detected in at least one of the LoTSS, VLASS, or FIRST surveys (in black). The background orange contours represent the WISE colors for all galaxies in SDSS-MaNGA DR16 sample. The contour clearly indicates a bimodal distribution in the color space, representing the star-forming and quenched galaxy population. Star-forming galaxies mostly occupy regions with 0.3 < W3 − W2 < 0.9, while the quenched population has − 0.6 < W3 − W2 < 0.0. The radio-detected red geysers lie in the quenched part of the diagram, which confirms the passive nature of these galaxies.

Zoom In Zoom Out Reset image size

Since our red geyser targets possess strong emission lines, a common and useful method to separate SFG from AGN hosts is through the ionization of the gas via a Baldwin-Phillips-Terlevich diagram (BPT, Baldwin et al. 1981). By observing the relative strengths of four emission lines, namely [O iii]/Hβ and [N ii]/Hα, we can separate SFG and AGN-host galaxies based on the hardness of their ionizing spectrum, which in turn drives the relative fluxes of different emission lines. This causes the AGN-host galaxies to occupy a separate region in the diagram from the SF galaxies, with the Kewley et al. (2006) and Kauffmann et al. (2003) demarcation lines in between. Figure 4 upper right (panel b) shows the BPT diagram of all galaxies from MaNGA Data Release 16 in black. The red diamonds indicate the radio-detected red geysers (detected in at least one radio band). The red geysers land in either the LINER or AGN regions of the BPT diagram and show no indication of SF activity. The absence of SF through the BPT diagram provides an useful diagnostic, as this confirms and reiterates the quiescent "red and dead" nature of the galaxies and indicates that the possible source of the radio emission is an AGN.

The D4000 versus L_rad/M method for identifying radio AGN was developed by Best et al. (2005). The parameter D4000 is the strength of the 4000 Å break in the galaxy spectrum, and L_rad/M_⋆ is the ratio of radio luminosity (measured in a specific radio band) to stellar mass. This identification process is constructed on the basis that SFGs with a wide range of SF histories occupy the same region in this plane because both L_rad/M and D4000 depend broadly on the specific SFR of the galaxy. On the other hand, radio-loud AGN have enhanced values of L_rad and are thus separable on this plane. Among the low-luminosity radio sources, low D4000 value would distinguish galaxies with active SF, which would possibly be the dominant cause behind the observed radio emission in these sources. This identification method, later implemented with slight modifications by Kauffmann et al. (2008) and Sabater et al. (2019), has generally been successful, with a few cases of misclassification.

In Figure 4 lower left panel (c), we plot D4000 versus L₁₅₀/M_⋆ for the radio-detected red geysers in black. We use only the 34 LOFAR-detected sources in this analysis, using the flux measurements from the 150 MHz band, which are then compared with existing sources from the literature. We overplot the radio sources from SDSS DR7 from Sabater et al. (2019) in the background. The data points are color-coded in blue circles and red arrows, which represents SFG and radio AGN, respectively, classified using a combination of diagnostics (see Sabater et al. 2019). In general, sources with an average D4000 value exceeding ∼1.7 do not exhibit enough active SF to show substantial radio emission due to supernovae or stellar activity. Hence radio sources with L₁₅₀/M ${}_{\star }\gt 11\,W\,{{Hz}}^{-1}\,{M}_{\odot }^{-1}$ and D4000 > 1.7 are predominantly radio AGN. All the red geysers in our sample land in the radio AGN portion of the diagram, as they have a relatively old stellar population with D4000 exceeding 2.0.

In addition to these three diagnostic plots, we also show the relation between radio luminosity (from LOFAR at ∼150 MHz) and SFR (panel d) for the LOFAR-detected red geysers sample (in magenta). As mentioned previously, the SFR is expected to correlate with radio luminosity in SFGs due to synchrotron emission from supernovae shocks. Gürkan et al. (2018) studied the low-frequency radio luminosity—SFR relation in a large sample of SDSS galaxies and found the best-fit (single power-law) relation to be

$$\begin{eqnarray}&&{\mathrm{log}}_{10}({L}_{150})=1.07\pm 0.01\times {\mathrm{log}}_{10}(\mathrm{SFR})+22.07\pm 0.01.\end{eqnarray} \tag{ 3 }$$

The above relation is shown as a red line in Figure 4. Any galaxy lying above this relation posses an excess amount of radio emission, too high to be produced from the corresponding level of SF. We find that all the red geysers lie above the Gürkan et al. (2018) relation, which indicates that the radio emission is consistent with galaxies hosting radio AGNs.

We primarily use data from the LoTSS survey to analyze the radio morphology of the red geysers for the following reasons. First, the LoTSS data are most sensitive to extended fainter radio emissions among the interferometers used in this work because this survey has shorter interferometric baselines than the VLA surveys. Second, the low-frequency (∼150 MHz) radio-continuum emission reflects the oldest and the lowest energy emission from the plasma, which helps in characterizing the full extent of the structure and enables a robust classification. As mentioned in Section 4.1, 34 out of 103 red geysers with LOFAR observations are detected at 150 MHz. Six of them are unresolved with a deconvolved major axis <5'' (i.e., < 3 kpc at median z = 0.03) and are thus physically contained within the central region of the galaxy. There are 28 LOFAR-detected red geysers that are resolved compared to the beam size. We visually classify the morphology of these 28 sources into five types based on the LOFAR images, roughly following Baldi et al. (2018), Kimball et al. (2011), and Jarvis et al. (2021).

• 1.  
  Compact (C): if the source shows no visibly spatially resolved features (i.e., has the appearance of a single 2D Gaussian) and is constrained within the host galaxy. These sources have sizes larger than the beam size and may be elongated intrinsically, but higher resolution is needed to confirm this.
• 2.  
  Extended jet (E): if the source is visibly spatially extended in one direction, but is composed of one contiguous feature, i.e., one distinct peak in the radio emission.
• 3.  
  Double (D): if the source shows two distinct peaks in the radio emission.
• 4.  
  Triple (T) : if the source shows three distinct peaks in the resolved radio image, typically consisting of two jets and one core.
• 5.  
  Irregular (I): sources generally with one distinct peak in the radio emission, but with unique spatially extended (irregular) radio morphologies that do not fit within the above categories.

The most common morphology is the "compact" class, found in 14 out of 28 sources. The object with MaNGAID 1-256446 in Figure 5 belongs to this class of sources. In five sources, we see a comparatively extended morphology stretching in one direction, which we define as "extended jet". It is to be noted that this term is used to describe the morphology only and might not be physically associated with an AGN jet. Four sources show a double-peaked radio emission (classified under "double"). They generally consist of a radio core and a one-sided bubble/tail that may or may not be directly attached to the central component. MaNGAID 1-23958 (Figure 5) is one such example. In three sources, we see three or more distinct peaks in radio emission. Two of them show a central core emission with large-scale double-sided lobes on either side, extending >40 kpc. The third source shows an extended morphology that consist of several distinct radio peaks with >50σ. These belong to the "triple" category. The galaxy with MaNGAID 1-378770 in Figure 5 is an example with double-sided jets. Finally, MaNGAID 1-188530 represents the irregular morphological class; in this case, the LOFAR image looks quite peculiar, resembling a jellyfish, with a bright central core superimposed on a large-scale structure of diffuse radio emission.

Zoom In Zoom Out Reset image size

Although we perform our morphological classification primarily using LOFAR images, there are eight FIRST-detected red geysers that are not covered by the LOFAR observations. We similarly categorize these sources using FIRST data into the five classes described above. Out of these eight sources, four are unresolved with a deconvolved major axis <25. The rest belong to either the compact or extended-jet classes.

Table 1 shows the classification type of each LOFAR-detected red geyser along with the additional eight FIRST-detected sources that do not have corresponding LOFAR images (indicated by square brackets).

Figure 6 shows the morphological comparison of the LoTSS images (left panel) with images from the FIRST (middle panel) and VLASS surveys (right panel) of two example red geysers out of the five sources from Figure 5. The radio-detected red geysers, depending on the morphological class, exhibit spatially diffuse extended features that are often only visible with LOFAR data. In most cases, the corresponding VLASS and FIRST images look rather compact.

Zoom In Zoom Out Reset image size

The actual physical sizes of the radio sources are independent of the frequency of observations. However, the apparent linear size, as measured from different radio bands—i.e., using FIRST, VLASS, and LoTSS data, can be different (as evident from Figure 6) for several reasons. First, the FIRST/VLASS data are at higher frequencies and hence are less sensitive than LOFAR to structures of typical spectral index, which can lead to a smaller estimated size. Second, FIRST and VLASS lack short interferometric baselines, which makes observations of extended structures above a certain size difficult in these surveys. Surface brightness sensitivity, on the other hand, can limit LOFAR sizes as well. Hence, we estimate sizes from both LoTSS and FIRST surveys, and report the larger of the two as our best estimate of the physical size of the radio source. In almost all cases, the size estimate obtained from LOFAR is greater than that from FIRST by at least a few factors. So in cases where only FIRST data are available, we report the FIRST-measured size as a lower limit. Our method of determination of size depends on the nature of the radio morphology.

For the sources under the label "double" and "triple", which show two and three distinct radio features in LOFAR images, respectively, the linear size in the 150 MHz is calculated as the distance between the peak emission of the two farthest morphological features detected within 10σ contours. For sources having contiguous, extended (E), and irregular morphology (I) with closely blended components, we measure the end-to-end linear size of the radio structure detected within >10σ. For the cases where the source is featureless and has only one primary morphological feature (classified as compact "C"), we use the major-axis size, deconvolved from the beam, as listed in the LOFAR catalog.

Depending on the specific structure of the FIRST image, the linear size in 1.4 GHz is also calculated in a similar way. The FIRST catalog also provides major-axis measurements (FWHM in arcsec) from the elliptical Gaussian model for the source, which are then deconvolved to remove blurring by the elliptical Gaussian point-spread function. As mentioned above, we consider the larger of the measured sizes from LOFAR and FIRST to be the "largest linear size", or simply, the radio size. Table 1 lists the measured sizes of the radio sources along with the 1σ uncertainty. For the compact and unresolved sources, the errors are derived from the respective catalog, which reports the 1σ uncertainty in the deconvolved major axes, derived from the Gaussian models of the sources. For resolved objects showing spatially extended morphologies, we assume the uncertainty to be the linear size (in kpc) corresponding to half the beam width, i.e., 3'' for LOFAR. The uncertainty in size from the FIRST band is obtained from the relation σ_Size = 10''*(1/S/N + 1/75), ¹⁶ where S/N is given by: S/N = (F_peak − 0.25)/RMS. F_peak and rms signifies the peak flux and the root mean square deviation measured from the catalog, respectively.

Table 1 shows the radio sizes along with radio luminosities of the radio-detected red geysers in both the LOFAR and FIRST radio bands. Figure 7 shows the radio luminosity (at 1.4 GHz) versus linear size of the radio-detected red geysers, detected in at least one of the radio bands (as black stars), overplotted along with different classes of radio sources from the literature in different colored contours. For the red geysers with available LoTSS counterpart, we convert the measured 144 MHz flux from LOFAR into 1.4 GHz, assuming a spectral index of −0.7 in order to calculate the desired luminosity (Condon et al. 2002). This is preferred over using the FIRST flux, which misses extended emission. However, for those without LOFAR data, corresponding FIRST luminosities are used. The differently colored contours in the figure, as indicated, represent compact symmetric objects (CSO), gigahertz peaked spectrum (GPS), compact steep spectrum sources (CSS), Fanaroff–Riley class 1 (FRI), Fanaroff–Riley class 2 (FRII), RQQs, and Seyferts. The data for these radio-detected AGNs have been compiled by Jarvis et al. (2019) from a variety of studies of radio AGN population, namely An & Baan (2012), Gallimore et al. (2006), Kukula et al. (1998), Baldi et al. (2018), and Mingo et al. (2019). Most of the radio-detected red geyser sources, marked as black stars, overlap with the LINER-/Seyfert-type classification and with the tail of the distribution of RQQs. There are two sources that lie on the FRI part of the diagram, both of which are categorized as "triple" according to the morphology classification. Our sources are in general consistent with having similar small-scale low-luminosity jetted morphologies as observed in the RQQs in Jarvis et al. (2019) or with FR0 sources, which remain classified as "compact" unless higher-resolution observations are available to resolve the sources, as in Baldi et al. (2015), Capetti et al. (2019), and Hardcastle et al. (2019).

Zoom In Zoom Out Reset image size

In this section, we study the spectral indices for our sample of 21 red geysers that have simultaneous radio detection from LOFAR, FIRST, and VLASS data (see Section 4.1 for more details). The radio-continuum spectrum is generally dominated by nonthermal synchrotron emission with a characteristic power law, S_ν ∝ ν^α , where α is the spectral index, ν is the frequency of radio emission, and S_ν is the flux density measured at frequency ν. In SFGs, there may also be some additional contribution from the thermal bremsstrahlung (free–free) emission (Duric et al. 1998; Gioia et al. 1982), but this is irrelevant here.

The spectral index of a radio galaxy can provide information about the relative contributions of the core and the extended lobed structure in the total radio emission. Core-dominated emission and any compact source typically have a flat (α > − 0.5) spectrum due to the effects of synchrotron self-absorption and free–free absorption (e.g., O'Dea & Saikia 2021). The extended lobes, on the other hand, tend to have steep spectra (α ≤ − 0.5) because the predominant emission mechanism is optically thin synchrotron. Thus, radio sources that are more core-dominated tend to have flatter spectra than those dominated by extended emission. This is confirmed from the observation of a high-core dominance in FR0s, owing to their compact nature, based on high-resolution images (Baldi et al. 2019).

As can be seen in Figure 5 and Table 1, our sources exhibit a range of radio morphologies from LOFAR data. We use the integrated-flux densities and uncertainties from the LoTSS, FIRST, and VLASS catalogs for our analyses (see Table 1 for the flux densities and uncertainties for each source). For sources having more than one component/ region, we visually identify all the individual components detected at a significance of >5σ that are associated with the source and add the flux densities from these components together. We define the radio spectral index, α, using S_ν ∝ ν^α and measure α by fitting the flux densities measured at three different frequencies (1.4 GHz, 3 GHz, and 150 MHz) with this function for each source. The errors are obtained from the uncertainties in flux densities via simple error propagation. Table 1 lists the spectral indices with 1σ uncertainty of the 21 red geyser galaxies.

The red geysers show a large spread in spectral indices, ranging from −1.0 to 0.0 with a median value of −0.62 (see Table. 1 and Figure 8). The extended radio sources, which do not belong to the "compact" morphological class, exhibit steeper spectra on average with a mean spectral index α =− 0.67. The fraction of sources with a steep spectrum (α <− 0.5) is ∼57%, indicating the presence of extended emission or dominance of lobed components in these radio sources. This automatically implies that 43% of our sample of 21 red geysers (i.e., eight sources) have a flatter spectrum, with core-dominated compact structures. This agrees well with our morphological classifications, in that a fairly large fraction of red geysers have compact radio morphologies confined within the host galaxy. The bright nuclear radio component with a moderately flat spectral index (i.e., α > − 0.5) may indicate a contribution from radio emission associated directly with an AGN core/unresolved base of the jet (Padovani 2016).

Zoom In Zoom Out Reset image size

Figure 8 shows the variation of the spectral index with 1.4 GHz flux density as measured from FIRST. Galaxies are color-coded by the morphological class, as given in Table 1. For this figure, the unresolved sources are also color-coded as the "compact" class. All three galaxies classified as "triple" have steep spectral indices, indicating that these sources have more extended lobes with predominantly optically thin synchrotron emission. With an exception of two galaxies, almost all the red geysers that are not compact (any color except red) have a steep spectrum and lie below the dashed line. The two sources showing a flat spectrum might be due to the result of a combination of underlying biases in the measured fluxes and puzzling morphology leading to an incorrect classification. On the other hand, half of the sources labeled "compact" show a flat spectrum, while the other half exhibit a steep spectral index. We do not see any significant correlation of spectral indices with radio flux density. The implications of these results are discussed in Section 7.

Traditionally, for compact radio-loud sources (CSS/GPS), a weak positive trend has been observed between the luminosity of strong emission lines and radio size (O'Dea & Saikia 2021). Larger radio sources are found to be more commonly associated with higher [O iii] luminosities (O'Dea 1998). A similar correlation has also been observed in radio-quiet sources using SDSS-measured [O iii] luminosities (Jarvis et al. 2021). Here we investigate whether a similar correlation exists in red geyser galaxies. Here we use Hα as a tracer of ionized gas instead of the traditional [O iii] line because the characteristic bisymmetric pattern identifying a red geyser is most prominently observed in the spatial distribution of Hα.

Figure 9 shows the Hα luminosity, integrated over one effective radius as observed by MaNGA, versus the linear size of the radio emission of the 42 radio-detected red geysers. The data are color-coded by the radio luminosity at 1.4 GHz. The radio luminosities are derived by converting L_150MHz to L_1.4GHz assuming a spectral index of −0.7, as before. Our data show a positive correlation that is consistent with what has been observed in radio-loud compact sources and in some recently studied radio-quiet sources. Liao & Gu (2020) showed that for the radio-loud population, the [O iii] bright sources (with L[O iii] > 10⁴² erg s⁻¹) have radio sizes >0.75 kpc. This is consistent with our result because the red geysers show radio sizes >10 kpc for similar L_Hα . The existence of this positive correlation in the red geysers is strongly supported by a Spearman rank correlation coefficient of 0.6. Figure 9 also shows that the radio luminosity generally shows a higher value ( ∼ 10²³ W Hz⁻¹) for larger radio size and higher Hα luminosities (i.e., toward the upper right portion of the figure). On the other hand, as L_1.4GHz drops to 10^21.5 W Hz⁻¹ for L_{[O iii ]} < 10⁴⁰ erg s⁻¹, the radio size decreases to <5 kpc. This is expected because of the first-order dependence of the morphology of the radio sources on radio power. Large-scale radio structures are found to be generally more abundant in radio-loud sources, although this correlation is not so apparent in radio-quiet sources (Morganti 2021). In the case of red geysers, this possibly implies that larger radio sources with a greater radio luminosity host a greater amount of ionized gas.

Zoom In Zoom Out Reset image size

In previous studies, similar relations between the strength of an ionized gas outflow tracer and the size of the radio source have been explained by considering the interaction between an embedded, AGN-driven radio jet and the ISM. Regardless of the size and structure of the radio jet—which is unlikely to be resolved owing to spatial resolution in this case—the important parameter to quantify the impact on the surrounding ISM is jet energy. Although there have been several methods to estimate the jet power, the method presented by Willott et al. (1999) based on the synchrotron properties of the radio sources has been shown to be particularly effective. They proposed the following conversion between the jet mechanical energy and the radio luminosity:

$$\begin{eqnarray}&&{E}_{\mathrm{mech}}=2.8\times {10}^{37}{\left(\displaystyle \frac{{L}_{1.4\mathrm{GHz}}}{{10}^{25}\ W\ {{Hz}}^{-1}}\right)}^{0.68}\ W.\end{eqnarray} \tag{ 4 }$$

This expression is used in this work, although there are limitations of this approach, as discussed in Hardcastle & Krause (2013) and Croston et al. (2018). We find that red geysers show a large range in E_mech, spanning three orders of magnitude, within 10^41.5 − 10^44.5 erg s⁻¹. In order to reaffirm the hypothesis that the source of this estimated energy is the central radio AGN, we also calculate the corresponding supermassive black hole mass (SMBH) of our sources. The SMBH mass and the jet mechanical energy are typically connected in an AGN. A radio AGN tends to be more radio loud and is associated with more energy as the black hole becomes more massive (Best et al. 2005). However, these two quantities are not expected to show any correlation for radio sources associated with other astrophysical phenomenon. We estimate the black hole mass, M_BH, using the following relation (McConnell & Ma 2013; Cheung et al. 2016):

$$\begin{eqnarray}&&{\mathrm{log}}_{10}\ ({M}_{\mathrm{BH}}/{\text{}}{M}_{\odot })=8.32+5.64{\mathrm{log}}_{10}[{\sigma }_{\star }/(200\ \mathrm{km}\ {{\rm{s}}}^{-1})],\end{eqnarray} \tag{ 5 }$$

where σ_⋆ is the velocity dispersion of the stars, extracted from the central 2'' radius aperture. Figure 10 shows the jet mechanical energy versus SMBH mass in the radio-detected red geysers, color-coded by their respective morphological class. We find that they are moderately correlated. We fit the data points with a linear function using least-squares optimization technique and find a relation of the form ${\mathrm{log}}_{10}\ ({E}_{\mathrm{mech}})\,={\left[{{log}}_{10}\ ({M}_{\mathrm{BH}})\right]}^{0.39}+38.925$, with a Spearman correlation coefficient (r-value) of 0.4. This positive trend implies that the central AGN, with possibly unresolved small-scale radio jets, is driving the radio emission seen in these galaxies.

Zoom In Zoom Out Reset image size

In order to estimate the possible contribution of this jet energy to the quenching of SF, Figure 11 shows the radio luminosity at 150 MHz versus specific SFR (sSFR) for the LOFAR-detected sources, color-coded by their morphological classification. Similar to Figure 8, the unresolved sources are also indicated as "compact". We find that the radio sources that are noncompact and belong to any of the extended, irregular, double, or triple class, have a much lower sSFR with average log₁₀ sSFR = −13 yr⁻¹, compared to the compact sources with average log₁₀ sSFR = −11.75 yr⁻¹. This implies that the radio sources showing more extended radio morphology are either more effective in quenching or reside in a larger halo with higher stellar mass, bringing the total sSFR down by several factors. A weak negative correlation is also visible between radio luminosity and sSFR, although this apparent trend can be due to the low sample size and driven predominantly by a few large and extended radio sources that are generally more radio-loud on average. This seems to be the case here because the three sources under the "triple" category and one "irregular" source, which are solely responsible for driving the negative trend, have radio luminosity (L_150MHz) at least an order of magnitude more than the average luminosity of the rest of the sources.

Zoom In Zoom Out Reset image size

In addition to the integrated properties discussed above, we now compare the spatially resolved ionized gas flux and kinematic maps with radio-image morphology.

While every galaxy in the red geyser sample shows signatures of ionized gas outflows via extended bisymmetric pattern in the EW map, there is a distinct lack of extended visible radio lobes on a similar scale in these galaxies as observed from a combination of the LoTSS, FIRST, and VLASS survey. As reported in Section 5.1, 14 out of 28 (∼50%) sources showing resolved radio emission in LOFAR observations display a compact radio morphology. Although similar radio-quiet compact sources hosting radio AGN have been observed to host small-scale (∼1 kpc) radio jets in higher spatial resolution (<1'') radio observations in previous studies (Jarvis et al. 2019, 2021; Venturi et al. 2021; Panessa et al. 2019; Baldi et al. 2018; Webster et al. 2021), the presence of resolved radio jets in the red geyser sample is observed to be quite rare with the current 5'' resolution.

For the red geysers belonging to the "compact" or "extended" class, the spatial correlation between ionized wind cone and the radio jets is difficult to infer. However, the sources belonging to the "double" or "triple" category provide the most insight. In two out of three red geysers belonging to the "triple" class (MaNGAID: 1-378770 and 1-595166), the radio lobes align perpendicular to the direction of the ionized wind cone. On the other hand, for the rest of the sources that are resolved but not "compact" (four "double", one "triple" and two "irregular"), the elongation axis in the radio images roughly aligns with the direction of the ionized wind cone.

We choose five prototypical galaxies representing the compact, irregular, double, and triple morphological class to explore the detailed ionized gas-radio interaction using spatially resolved maps.

MaNGAID: 1-245451 in Figure 12 shows an example of a radio-detected red geyser belonging to the "compact" radio morphology class. The three columns in the figure correspond to the SDSS optical image (first row), Hα-EW (second row), and velocity dispersion (third row) extracted from the Hα emission line. The LOFAR radio contours (in green) are overplotted on top of each map. The on-sky diameter of the MaNGA fiber bundles (overplotted as the magenta hexagon in the optical image) generally ranges between 17'' and 32'', corresponding to a physical size of 10–30 kpc at the median redshift of the MaNGA observations (z ∼ 0.03). The bisymmetric pattern in the Hα EW map traces the ionized wind cone. The absence of structures in the radio image makes it hard to associate the radio properties with any specific ionized features. However, we note very high gas dispersion within the inner radio contours detected with >20σ, indicating extreme ionized outflow kinematics there.

Zoom In Zoom Out Reset image size

For the red geyser with MaNGA ID—1-188530 (belonging to the "irregular" morphology class), the LOFAR image has an unusual extended morphology spanning a distance of ∼30 kpc, with a central radio core and a plateau of diffuse emission. Interestingly, the radio structure is spatially extended in the direction of the Hα enhancement in the EW map, similar to the other sources in the "irregular" class. Additionally, we note that the galaxy shows an elongated region of enhanced line width (>200 km s⁻¹), spanning about >7 kpc.

In the red geyser with the MaNGA ID: 1-23958, the radio image shows a one-sided low surface brightness bubble, detached from the central bright core. Similar to the previous example, the direction of the radio bubble roughly aligns with the bisymmetric pattern in the Hα emission-line map. This is the case for all sources classified as "double".

Finally, as already mentioned, double-lobes and distinct jets are observed to be quite rare in the red geyser sample using the current ∼5'' resolution of the radio images. Out of the three sources classified in the "triple" morphological category, two (Figure 12, MaNGA ID: 1-378770 and 1-595166) have radio jets lying perpendicular to the ionized gas traced by the Hα EW map (Figure 13). This is unlike what is observed in the above cases. The gas-velocity dispersion is enhanced perpendicular to the Hα bisymmetric feature in 1-378770, but is aligned in 1-595166. The implications of these findings are discussed in more detail in Section 7.

Zoom In Zoom Out Reset image size

Although the radio emission in radio-quiet sources is often attributed to being dominated by SF processes, red geyser targets have very little SF activity (log SFR < 10⁻²M_⊙ yr⁻¹ Roy et al. 2021b, and Figure 2). Section 4 shows the quiescent nature of these galaxies via WISE infrared colors, confirms the complete lack of SF, and the presence of an old stellar population through the D4000 versus L_rad/M_⋆ method, and it showed the lack of ionization from young stars via the BPT diagram. Indeed, in Cheung et al. (2016), the central radio-continuum emission in the prototypical red geyser was from a low-luminosity radio AGN (L_1.4GHz ∼ 10²¹ W Hz⁻¹) with a low Eddington ratio (λ ∼ 10⁻⁴). Roy et al. (2018) showed that the expected SFR ( ∼ 1 M_⊙ yr⁻¹) derived from the average radio luminosity from the red geyser sample exceeds the observed SFR, derived from ultraviolet to infrared SED fitting, by two to three orders of magnitude. If we perform a similar calculation on our current sample of 29 FIRST-detected sources, we obtain an average radio luminosity L_1.4GHz ∼ 5 × 10²² W Hz⁻¹ (Figure 7). From the best-fit relation between 1.4 GHz radio-continuum luminosity and the Balmer decrement corrected Hα (Brown et al. 2017), we obtain a corresponding Hα luminosity ∼2.5 × 10⁴² erg s⁻¹. Using the known relation between SFR and Hα luminosity (Kennicutt et al. 2009; Brown et al. 2017) assuming a Kroupa IMF (Kroupa & Weidner 2003), we obtain an expected SFR exceeding 5 − 10 M_⊙ yr⁻¹, which is not observed in our galaxy sample. This is further confirmed by Figure 4 (panel d), which shows that our objects lie above the low-frequency radio luminosity—SFR relation from Gürkan et al. (2018). This implies that the observed radio luminosity cannot be explained by the detected very low amount of SF and is consistent with radio emission from central radio AGN.

Another widely discussed source of the radio emission in radio-quiet sources is radiatively driven accretion-disk winds that result in synchrotron-emitting shocks through the interstellar medium (Zakamska & Greene 2014; Nims et al. 2015; Zakamska et al. 2016). However, these are generally associated with quasars with a bolometric luminosity L_AGN ∼ 10⁴⁵ erg s⁻¹, where outflow velocities of ∼ 1000 km s⁻¹ interact with the ISM and can produce radio luminosities similar to our sources (L_{1.4 GHz} ∼ 10²² − 10²³ W Hz⁻¹). However, the red geysers show typical gas velocities ∼ 300 km s⁻¹ (Figure 1) with no signature of quasar-like broad emission lines. A rough estimation of the bolometric radiative luminosity from the [O iii] λ 5007 Å emission-line flux (within the central 2'' radius aperture), using the relation L_rad = 3500L_{[O iii ]} (Heckman et al. 2004), yields L_rad ∼ 10⁴³ erg s⁻¹. We calculate the classical Eddington limit with L_Edd = 3.3 × 10⁴ M_BH, where M_BH has been calculated in Section 6.1 (Equation (5)). Inserting the jet mechanical energy E_mech(Equation (4)), radiative luminosity L_rad, and Eddington luminosity L_Edd, we calculate the Eddington radio λ = (E_mech + L_rad)/L_Edd. Figure 14 shows the distribution of Eddington ratios for the radio red geysers, which spans primarily between ∼ 10⁻⁴–10⁻³. These fairly low Eddington ratios imply that these are radiatively inefficient sources that cannot be formed due to radiative accretion-disk winds.

Zoom In Zoom Out Reset image size

Low-luminosity jets from radio-quiet AGN are the most plausible explanation for the observed low-power radio emission in our red geyser targets. Sufficiently deep and high-resolution radio observations have been able to identify small-scale radio jets in several radio-quiet quasars and Seyfert galaxies (Gallimore et al. 2006; Baldi et al. 2018; Jarvis et al. 2019, 2021; Venturi et al. 2021). Our sample of radio-quiet red geysers has many properties in common with jetted compact radio galaxies, similar to those in Kimball et al. (2011), Baldi et al. (2018), and Jarvis et al. (2019).

At the frequencies of interest in this work, the radio-continuum emission is dominated by nonthermal synchrotron emission. In the absence of a constant energy injection source, the replenishment of fresh electrons ceases and the radio spectrum is dominated by radiative loss. Because the energy-loss rate is directly proportional to the frequency, the energy-loss rate at lower frequencies ( ≤ 1.4GHz) is lower, which enables the original injection index of the electrons to be retained for much longer. Thus, the lower frequency emission generally is more extended, characterizing emission from older plasma where injection took place longer ago. Additionally, LOFAR is more sensitive to extended emission than the FIRST and VLASS surveys. This is consistent with our findings (Figure 6), which shows that the LOFAR-measured sizes are roughly two to three times more spatially extended than the FIRST and VLASS images, sometimes revealing intriguing structures that are not visible in higher frequency images.

We observe a range of radio morphologies from LOFAR observations in our red geyser targets. From Table 1, we see that 16 out of 42 radio-detected sources exhibit a resolved but compact morphology, with the spatial extent ranging between 3-7 kpc in the 150 MHz frequency band. They can be represented by a 2D Gaussian with no particular feature in their radio images. On the other hand, 16 other sources show extended features (≥9 kpc), with contiguous one-sided morphology, bubbles, and double-lobes (belonging to the "extended", "irregular", "double", and "triple" morphology class). The remaining 10 sources are unresolved in the typical resolution of LOFAR and FIRST. Thus, ∼38% of the radio-detected sources show spatially extended features extending to scales of more than 10 kpc. This is consistent with the results of Pierce et al. (2020), who have studied similar moderate-luminosity radio AGN population, although with a greater radio luminosity range (22.5 < L_{1.4 GHz} < 25. ) than the red geysers. However, of those 38%, the 7 galaxies classified as "extended" also do not show any resolved radio jets, although they show elongation in a specific direction, indicating some underlying radio structures remaining unresolved at the current spatial resolution.

To quantitatively compare the radio morphologies of our sample to the traditional radio AGN population, we investigate the radio size versus radio luminosity plane for the red geysers compared to the literature compilation of radio selected AGN from An & Baan (2012), Gallimore et al. (2006), Kukula et al. (1998), Mingo et al. (2019), and Jarvis et al. (2019) in Figure 7. In terms of linear size, most of the red geyser sources are similar to the CSS sources (spanning ∼1–25 kpc in 1.4 GHz radio image), with a few showing even more compact structures with no structures resolvable beyond the nuclear component, similar to the GPS sources (typically <1 kpc, O'Dea 1998). However, unlike the red geysers, the CSS and GPS sources are powerful radio-loud AGN (L_{1.4 GHz} >10²⁵ W Hz⁻¹). Hence, our sources would be excluded from these samples due to a much lower radio luminosity. The red geysers are more aligned with the RQQs from Jarvis et al. (2019) and also from the LINER and Seyfert radio sources from the LeMMINGs survey (Baldi et al. 2018, 2021), according to the radio luminosity- size diagram. A few are consistent with the lowest-luminosity AGN in the sample of Gallimore et al. (2006).

As noted by Jarvis et al. (2019, 2021), objects in the RQQ category possess small-scale jets when observed in higher-resolution (subarcsecond scale) VLA and e-MERLIN images, but exhibit compact or slightly extended kpc-scale structures when observed in low spatial resolution (>3''). They have a similar radio morphology, spectral index, and radio size as the red geysers. Considering the stark similarities of the radio properties of the red geyser sources with the RQQ sample, it seems plausible that our sources also possess small-scale jets that are blurred in the current resolution. These could resemble "frustrated" jets, which are small and contained within the inner 1 kpc central region of the galaxy, occurring due to the surrounding dense environments that does not enable the jet to grow to a large size (van Breugel et al. 1984). Two of our targets overlap with the Fanaroff–Riley class I (FRI; Fanaroff & Riley 1974) galaxies in the luminosity–size plane, but the majority of our targets do not fit within the traditional FRI and FRII radio classifications. Due to the abundance of compact and featureless radio morphology in our red geyser galaxies, these sources can also be classified as FR0 galaxies (e.g., Baldi et al. 2015; Capetti et al. 2020). However, higher spatial resolution data may reveal more complex morphologies, jets, and hot spots on smaller scales (see discussion in Hardcastle & Croston 2020).

We have previously identified kpc-scale outflows in ionized gas in the red geyser galaxies (Cheung et al. 2016; Roy et al. 2021a), which are marked by the bisymmetric extended pattern in the Hα EW map (Figure 1). Our study of red geysers complements several other studies that have aimed to characterize the drivers of ionized outflows by investigating the radio properties of the central radio AGN. For example, there are many spatially resolved studies of multiphase gas outflows driven by local galaxies hosting low-power AGN (Mingozzi et al. 2019; Wylezalek et al. 2020; Venturi et al. 2021). Specifically, Capetti et al. (2019, 2020) have studied FR0 galaxies showing similar compact radio structures detected with LOFAR observations and their effect on the galaxy environments, while Webster et al. (2021) has discovered galaxy-scale jets and their interaction with the interstellar medium of the host galaxy. There have been numerous studies at high redshift as well, but they primarily focus on powerful radio galaxies and luminous AGN (Nesvadba et al. 2017; Circosta et al. 2018; Perna et al. 2015; Zakamska et al. 2016), where the impact of quasars is dominant (Hopkins et al. 2007). The faint radio emission from radio-quiet AGN is difficult to detect, although a growing number of studies are being done in recent times to study the interaction between the radio emission from the AGN and the ionized gas outflows (Al Yazeedi et al. 2021). However, high S/N and high spatial resolution radio data (e.g., Jarvis et al. 2019; Venturi et al. 2018) are required to resolve the small-scale radio jets and morphologies to establish further connection with the ionized outflows.

Figure 9 aims to establish the connection between ionized gas and radio emission in the red geysers via integrated Hα luminosity with radio sizes. We show that our results are consistent with Jarvis et al. (2021) studying similar radio-quiet (L_150MHz < 10²⁴ W Hz⁻¹) compact (generally < 10 kpc) AGN sources from the quasar-feedback survey, which states that ionized gas tracers are correlated with the central radio AGN. Additionally, radio-detected red geysers not classified as "compact", possessing large radio sizes, are shown to have relatively lower sSFR (${\mathrm{log}}_{10}\ {sSFR}\sim -13.0\ {{yr}}^{-1}$) than the compact ones (Figure 11). Assuming these compact objects have similar SF history as the extended, lobed, and irregular ones, large radio sources having lower SFR for a given stellar mass might be implying more efficient quenching due to the presence of jets than the compact ones. The other possibility is that the extended radio sources are generally found in more massive and evolved sources with higher stellar mass, perhaps even massive central galaxies in large halos. Extended or lobed radio sources may be the evolved form of compact FR0 galaxies in such environments. This idea is consistent with the theory that compact FR0 radio galaxies are younger and will eventually evolve to form the traditional FRI or FRII galaxies (O'Dea 1998). Thus the range of radio morphologies observed in the red geysers could represent the various stages of transition of FR0 class of objects to the traditional FRI/FRII sources.

Figures 12 and 13 compare the spatially resolved EW and kinematics of ionized gas, traced by Hα, compared to the distribution of radio emission for five example red geysers belonging to different radio morphological class. We find that for the galaxies in the "irregular" and "double" class, elongation along a specific direction or presence of one-sided bubbles in the radio emission (MaNGAID: 1-188530 and 1-23958, Figure 12) roughly aligns with the ionized gas features, marked by high gas dispersion. Such observations indicate that the large-scale ionized outflows, stretching to >10 kpc, are possibly driven by small-scale radio jets, unresolved at the current LOFAR resolution. Often in these cases, the velocity dispersion map is clumpy and shows high values in distinct parts of the galaxy, and also in some cases, perpendicular to the bisymmetric Hα feature. This is similar to the observation by Venturi et al. (2021), who found that increased line widths were perpendicular to small-scale radio jets. They interpreted this to be due to the low-power jet strongly interacting with the ISM in the galaxy, releasing energy and giving rise to highly turbulent motions in the perpendicular direction. Similar characteristics have been seen in a few other local Seyfert galaxies as well (see Riffel et al. 2014, 2015; Lena et al. 2015; Freitas et al. 2018).

However, among the three rare cases (classified as "triple") where we detect clear evidence of large-scale radio jets and lobes, particularly notable are the galaxies with MaNGAID 1-378770 and 1-595166, where we find that the radio jets, shown by LOFAR radio contours and ionized wind, traced by the Hα emission map, lie perpendicular to each other (Figure 13).

There can be several possible explanations for the perpendicular incidence of the radio jet and ionized broad angled wind.

• 1.  
  Changing orientation of the magnetic field, as proposed by Mehdipour & Costantini (2019).
• 2.  
  The formation of an expanding cocoon structure described in simulations of jet-driven feedback (Begelman et al. 1984).
• 3.  
  A precessing accretion disk due to a misalignment between the orientation of the disk and the spin of the black hole (Riffel et al. 2019).
• 4.  
  Radiation-driven midplane wind (Proga & Kallman 2004).

In a sample of 16 radio-loud Seyfert-1 AGN galaxies, Mehdipour & Costantini (2019) showed an inverse correlation between the column density of the ionized wind and the radio-loudness parameter (R) of the jet observed to be misaligned with the wind. They argued that this indicates a wind-jet bimodality in radio-loud AGNs with the AGN alternating between powering a radio jet and an uncollimated broad wind. They proposed that the magnetic field is the primary driving mechanism for the observed accretion-disk wind and that the change in the magnetic field configuration from toroidal to poloidal causes this switch. This bimodality can explain the low incidence of radio jets and high prevalence of ionized wind in the red geyser sample along with a 90° misalignment between the jet and the wind. However, if the magnetic driving mechanism can work equally efficiently on low-power radio-quiet AGNs is still open to questioning.

Wagner & Bicknell (2011), Wagner et al. (2012), Mukherjee et al. (2016), and Mukherjee et al. (2018) have studied the interaction between radio-jet and multiphase ISM using detailed 3D hydrodynamical simulations. They find that when a radio jet propagates through an inhomogeneous ISM, they not only impact the ISM along the radio-jet axis, but also create a spherical bubble that drives gas clouds outward in all directions. This leads to outflowing gas traveling at a modest speed mostly in the path of least resistance. Although this mechanism cannot lead to extreme velocity outflows escaping the galaxy altogether, it can impart enough energy and turbulence to heat the gas and sufficiently inhibit SF. Because we observe similar modest outflowing gas velocities (∼300 km s⁻¹) accompanied by turbulent high-velocity dispersions (exceeding ∼220 km s⁻¹) in specific regions of the galaxy in red geysers with very little SF activity, this phenomenon might be the primary mechanism to explain the perpendicularity between radio jets and winds observed in the two red geysers.

A third possibility is due to the precession in the accretion disk, as proposed by Riffel et al. (2019). They suggested that if there is a misalignment between the orientation of the accretion disk and the spin of the black hole, it can create a torque leading to a constant precession. This mechanism in turn can lead to a small-scale jet, contained within the central part near the nucleus to be misaligned with the large-scale ionized wind observed in MaNGA. Riffel et al. (2019) have successfully implemented this model to the prototypical red geyser to explain the misalignment observed between the direction of the ionized wind in a small- (1–2 kpc) and large-scale (>10 kpc) spatially resolved Hα map. A similar procedure can possibly explain the large-scale radio jet and ionized wind misalignment, although further work is needed to confirm whether the small-scale precession near the black hole can be implemented in a larger spatial scale >20 kpc.

The final proposed mechanism exploring the radiation-driven midplane disk wind has been proposed by Proga & Kallman (2004). However, as discussed in Section 7.1, the red geysers do not seem to harbor radiatively driven accretion-disk winds owing to low Eddington ratios between 10⁻⁴–10⁻³ and an absence of extreme outflow velocities of ∼1000 km s⁻¹.

Finally, while there is a body of evidence that supports the interpretation of red geyser kinematics as the result of outflowing winds, this interpretation may be wrong in some cases. It is tempting to associate the perpendicular orientation of the major kinematic axis with a diffuse accretion disk, giving rise to extended, bipolar radio jets. While difficult to rule out, this explanation is unlikely because it implies that a black hole accretion disk on subparcsec scales would remain aligned to a galaxy-scale gaseous disk on kpc scales. In simulations and observations, such an alignment is extremely rare.