Radio continuum and OH line emission of high-z OH megamaser galaxies

Based on a sample of 30 OHM galaxies with redshift ranges from z$\geq$0.15 to the highest 0.2644 from Zhang et al. (2014), we have found archive VLA data for OH line and radio continuum observations, which are available for 12 and 21 sources, respectively (see Table 1 and 4). The high-resolution VLA observations of the OH line emission are mainly from two VLA projects, AD462 and AD483 (PI: Darling, J.). The details about these projects for OH line and radio continuum emission are listed in Table 4. Furthermore, we randomly selected 12 OHM galaxies for pilot observations with the FAST telescope to study the possible variable properties of the OH line profiles. The observational and derived properties of the selected sample are present in Table 1.

We reduced the historical VLA data with standard procedures using the NRAO Astronomical Image Processing System (AIPS), including initial data flagging, amplitude, phase, calibrations, and the correction for bandpass. We also shifted the radial velocity to correct the effects of the Earth’s rotation and its motion within the Solar System and towards the Local Standard of Rest. The EVLA data for the continuum emission were performed using the Common Astronomy Software Applications (CASA McMullin et al., 2007) package pipeline (version 5.4.1). We imported all the calibrated data into the DIFMAP package Shepherd (1997) to obtain the continuum and OH line channel images. The continuum images of the OH line projects were produced using line-free channels, and we made the OH line channel images after the subtraction of the clean model obtained from the radio continuum emission.

We have done pilot FAST observations of 12 high-z OHMs and listed the observational parameters in Table 2. We use the central beam (M01) of the 19-beam receiver with ON/OFF mode. We record the data every 1.00663296 s and inject a 10 K continuum signal using a noise diode every 32.21225472 s. Our data reduction procedures are as follows:

First, we convert the spectral signals into units of K using the nearest noise diode from the data. And we have adopted the average value of about 11.5 K for the injected signal, which is the mean value in ”median_20190115.Tcal-results.HI_w.high.sav” measured by the FAST group. Second, we further converted the signals into units of Jy with the equation: Gain=$\sim$16.0 K/Jy, which is from the fundamental performance parameters of the FAST telescope when zenith angle $\theta_{ZA}$¡26.4 ^o (available in the webpage ¹¹1https://fast.bao.ac.cn/cms/article/97/). Third, we generated the spectra of target-ON and target-OFF by combining the first two polarizations and then did ON–OFF subtraction (see Fig. 3). Because the baseline of the ON-OFF spectra still contains ripple caused by standing waves, we did the baseline fitting with equation

, where ’i’ ranges from 1 to 6 for different sources, and finally did the baseline subtraction and got the spectrum for each target as present in Fig. 1 and 4. The frequency interference (RFI) affected channels were labeled in yellow colors, which were identified by visually comparing the ON and OFF spectra, where both the on and off spectra show bumps higher than the typical waves (see Fig. 1, 3 and 4).

We have detected OH line emission from 11 of the 12 sources using the VLA archival data. We generated the OH spectra (see Fig. 1) by integrating the flux densities from a 3 $\sigma$ region defined from the combined line emission images as shown in Fig. 5. The OH emission images all show core-dominated structures. We fitted these images with one gaussian component, and the fitted component sizes are smaller than one beam (see Table 4). As shown in Fig. 1, the OH line profiles indicated that we had recovered all the OH line flux found from single-dish Arecibo observations. These results suggest that the dominant OH emissions are compact (see Fig. 5) and distributed in a region with sizes less than the arcsecond scale.

We present the OH spectra from pilot FAST observations for 12 sources (see Fig. 1 and 4). The OH profiles from FAST observations agree with the results from previous Arecibo observations in the literature (Darling & Giovanelli, 2000, 2001, 2002a) or archive VLA data investigated in this work. Six sources are likely showing OHM emission variability for their narrow line components (see Fig. 1), including three known variable sources (02524+2046, 12032+1707 and 21272+2514, see Darling & Giovanelli, 2002b; Darling, 2007).

The OH satellite lines in OHM galaxies are rarely detected and we also tried to search for the OH satellite lines (OH 1612 and 1720 MHz lines) based on our pilot observations. The binned spectra of each source with a resolution of about 10 km s^-1 are shown in Fig. 6. We have presented an upper limit of the two OH satellite lines for ten sources (see Table 2), which are lower than the values measured by McBride et al. (2013). Our results agree with the view from McBride et al. (2013) that there is no evidence for a significant population of strong satellite line emitters among OHMs.

The arcsecond-scale radio structure of OHM galaxies is available for 21 sources (see Table 4). The radio images for sources with arcsecond-scale VLA line observations and extended structures are presented in Fig. 2, 5 and 7). We found that the radio continuum emission of these sources shows no clear extended structure from the VLASS survey and L-band VLA-A observations (see Fig. 5). Five sources show possible extended structure at sub-arcsec scale from C or X band VLA data (see Fig. 2), which might be caused by possible jet or starburst regions. However, all the radio continuum emissions can be fitted with one gaussian component with a size smaller than the beam FWHM (see Table 4).

Because the majority of these galaxies are likely showing multi-nuclei from optical images (see Fig. 5), there is a possibility that the broad OH line profiles of these galaxies are caused by the combination of maser emission from multi-masing-nuclei (see notes of each source in Sect. 4.7). We have presented the arcsec-scale structure of OH line emission for 11 of the 12 sources with available VLA-A observations. Our results have recovered all the OH maser emissions detected by a single-dish telescope in the literature or our new FAST observations. It indicated that the majority of the fitted component sizes of the OH emission are much smaller than the beam size of VLA-A ($\sim$ 1.4”), and the actual physical sizes should be smaller than the kpc-scale.

It is also possible that these OH masing regions contain two nuclei within 1 kpc (e.g., the separation of the two nuclei of the well-known OHM galaxy Arp 220 is about 0.4 kpc), which needs higher resolution observations of these sources. However, VLBI observations of four galaxies in this sample are available in the literature, showing that the compact OH maser emission is distributed in hundreds pc scale (IRAS 02524, 08201, 12032, 14070, see Section 4.7 for the notes of each source and references therein). It means that other factors should cause the broad OH profiles, including orbiting disk, inflows/outflows, or particular merging stage (see Pihlström et al., 2005).

Generally, variability in OHMs provides additional evidence that narrow lines correspond to compact masing regions as it is believed to be caused by interstellar scintillation (Darling & Giovanelli, 2002b). Harvey-Smith et al. (2016) and McBride et al. (2015) have found the existence of intrinsic variability in amplitudes of OH features related to the masing region rather than that produced by interstellar scintillation. And one characteristic of the inherent variability caused by masing clouds is a significant decrease in OH line flux density during more than ten years (Harvey-Smith et al., 2016; McBride et al., 2015). Nearly all the OHM galaxies in this study were first detected by Arecibo OH surveys 20 years ago. However, our new FAST observations find that the OH flux densities are consistent with previous observations, likely showing no similar cases caused by intrinsic changes. Because the intrinsic variability is unclear, decreasing OH flux densities might only respond to the compact masers found in VLBI observations (Harvey-Smith et al., 2016), which needs high-resolution interferometric observations to study these variabilities further.

The VLA and FAST observations have shown that the OH profiles of six sources are variable. The results confirm the arguments from Darling (2007) that IRAS 12032+1707 shows variability in at least two line components (Darling, 2007). Two other known variable sources in OH line profiles (Darling, 2007), IRAS 02524+2046 and IRAS 21272+2514, also show variabilities by comparing the multi-epoch observations. We also found that three new sources, including IRAS 08201+2801, IRAS 23028+0725, and IRAS 11028+3130, were also variable in their OH profiles. Three variable sources are available with VLBI observations for OH line emission, and all show compact structures of OH emission (IRAS 02524, IRAS 08201, and IRAS 12032, see notes of each source in Sect. 4.7). So, the OH emission variability is consistent with the small size of masing clouds. These six sources detected with OH line variability show this sample’s highest line flux densities. Other sources with weak OH emission might need higher sensitivity observations to catch their variabilities.

We have presented the radio structure of continuum emission for 21 sources and OH line emission for 11 sources. We found that our selected sample’s arcsecond-scale radio continuum and OH line emission show no significant extended structures. The fitted image component size is smaller than the beam size (see Table 4). We estimated the relative positions between OH line and continuum peak positions ($\delta_{lc}$) (see Table 1). The $\delta_{lc}$ ranges from 0.06 to 0.58 arcsec and are all smaller than the 3 $\sigma$ of the uncertainties. It means that $\delta_{lc}$ are not significant, which is similar to the results for low-redshift OHM galaxies (see Lo, 2005, and references therein).

Generally, the FIR-radio (q)-ratio, brightness temperature ($T_{b}$), and spectral index are three independent radio parameters for classifying the radio activity in the nuclei. Based on these three parameters, Baan & Klöckner (2006) defined an activity factor $\beta_{c}$ to provide a consistent picture of the radio nature of the nuclei. The OHMs and candidates in Baan & Klöckner (2006) were detected before the Arecibo OH surveys (Darling & Giovanelli, 2000, 2001, 2002a), which are nearly all known OHMs at low redshift (z¡0.1). We selected the low-redshift OHMs from Baan & Klöckner (2006) based on their OH luminosity available in Zhang et al. (2014), including 34 low-redshift OHMs (30 sources with z¡0.1 and four sources with z¡0.136 ) as the control sample.

We have calculated the radio parameters for our selected high-redshift OHMs following the methods in Baan & Klöckner (2006) and reported the results in Table 1. We performed the two-sided Kolmogorov–Smirnov (K–S) statistical test to compare the properties of the low and high redshift sample; the results are presented in Table 3. We can see that the $L_{OH}$, $L_{1.4GHz}$, and $L_{IR}$ of the two samples are significantly different, with p-values much more minor than 0.01. However, the p-values for q(1.4), q(4.8), $\beta_{c}$ and $T_{b}$ are all much higher than 0.01, which indicates that the two samples do not show significant evidence of being drawn from different distributions for these radio parameters.

The radio continuum emission of OHMs shows extended emission might be related to a central disk, possible jet structures, or multiple nuclei structures. Because most OHMs show compact radio emission for our selected high-z samples and low-z samples from Baan & Klöckner (2006), we only compare the sources with large extended structures. Baan & Klöckner (2006) have measured the physical scale of four sources with significant extensions, which are about 3.0, 4.3, 6.4, and 16.8 kpc. About five sources in this work show clear structures ( see Fig. 2). And the scales (the distance between peak centers or peak center to the farthest edge measured from 3$\sigma$ radio contour images) are about 3.5, 8.6, 2.3 1.8 (21.8), 2.4 kpc (see Fig. 2 and Notes of each source in Sect. 4.7). Then, there also seem to be no significant differences in the scale of extended structures between the low and high redshift OHMs in our selected sample.

Zhang et al. (2014) have found that the AGN fraction (based on empirical criteria from Wide-Field Infrared Survey Explorer (WISE) color) is similar for both OHM and non-OHM samples with a value of 40%. Similarly, Darling & Giovanelli (2006) also found no distinguishing spectral properties to differentiate OHMs from nonmasing mergers using optical spectrophotometry studies. Based on the optical classification listed in Table 1, about 71% of sources (15) can be classified as AGN or AGN candidates. We also found that there are 62% of the low-z sample (21 of 34 OHMs) are AGN or AGN candidates. Although the AGN rate is slightly higher in the high-redshift sources, the AGN fraction of the low and high-redshift sample seems to have no significant differences if considering the uncertainties ($\sim$ 16-19% derived based on the 95% confidence level of a binomial proportion). Sotnikova et al. (2022) showed no significant correlation between $L_{OH}$ with spectral indexes of the radio continuum emission. We further found that there are no significant differences in the classification of radio properties between the low and high-redshift OHM galaxies. We noted that the $L_{OH}$ of our selected high-redshift sample is, on average, much higher than the low-redshift sample. These results show that the nuclei’s classifications of the power source do not significantly affect the luminosity of detected OH line emission.

Furthermore, high-resolution VLBI observations are available for only several sources in this sample: IRAS 08201+2801, IRAS 10339+1548, IRAS 12032+1707, IRAS 02524+2046, IRAS 14070+0525, and all these sources show no radio continuum emission above a 3$\sigma$ level (see section 4.7 and references therein). Farrah et al. (2003) show that the radio flux correlates with the starburst luminosity and the gas masses rather than the AGN in the nuclear regions governs the total luminosities of ULIRGs. It means that the origin of the radio continuum emission might be from starburst rather than the central AGN, which is consistent with the view from Rovilos et al. (2003) that the existence of an AGN in OHM galaxies may well be a random phenomenon not associated with their masing properties. High-resolution optical imaging and spectroscopy observations of OHM galaxies have also shown the presence of plenty of star-forming regions in the nuclei of these galaxies (see Hekatelyne et al., 2020; Migenes et al., 2011). These results support a starburst origin for the OHMs, and the OH masing regions are largely believed to be star-forming (see Pérez-Torres et al., 2021; Wu et al., 2022). Because of the existence of star-forming regions, the signature of AGN in OHM galaxies may be lost when using standard diagnostic diagrams (see Migenes et al., 2011). Meanwhile, an ongoing project using high-resolution, optical–IR multi-wavelength imaging and spectroscopy analysis has studied five from a sample of 15 OHM galaxies known to be dominated by star formation. However, the results showed the presence of a previously unknown AGN in four sources, which support the hypothesis that OHM galaxies harbor faint recent triggered AGN (see Hekatelyne et al., 2020, and references theirin). Although compact radio jets might be with much shorter lifetimes than the triggered AGN (see Emonts et al., 2006; Evans et al., 1999), further high-sensitivity radio interferometry and VLBI observations of a large number of OHM galaxies can also be useful in finding evidence of radio AGN and their connections with the megamaser emission (Peng et al., 2020).

Because the OHM emission region of our sample is compact at the kpc-scale, it is also possible that there are multi-masing regions or nuclei at the sub-kpc scale similar to Arp 220 (the separation of two cores is about 0.4 kpc Rovilos et al., 2003). However, we do not find evident two isolated components from sub-arcsecond scale radio images of 12 galaxies at C and X band (see Fig. 2 and Table. 4). Because of the association of OH emission and radio continuum, likely, the OH emission may also be compact at sub-arcsecond scale observations. However, OH line observations at these resolutions are needed to further confirm the OH emission structure.

We have randomly selected 12 high-redshift sources for Pilot FAST observations from 30 known high-z OHM galaxies. We have performed a K-S test that shows no significant differences between our chosen sample and the rest 18 high-redshift OHM galaxies, including OH line peak, FWHM, and $L_{OH}$. The 12 sources for our pilot observations might give us clues about the properties of a complete sample of high-redshift OHM galaxies available in the literature: 1) The detection of intrinsic change of masing regions-related variability might need observations of a much larger sample or blind surveys because the detection rate should be lower than 1/10. 2) The detection rate of variability caused by ISS is about 1/3, and the actual detection rate might be higher than this value if we set a flux density lower limit. It supports the view that the OHM emission clouds are small; the actual time scales of varying will need shorter spacing observations to detect. 3) The OH 1665 MHz line existed for all the sources. However, over 2/3 of these sources show broad multi-peak OH 1667 MHz profiles blended with the OH 1665 MHz line. The detection of two OH satellite lines needs significantly lower sensitivity observations of these OHM galaxies.