Multiwavelength investigation of extended green object G19.88-0.53: Revealing a protocluster

Low-frequency radio emission associated with G19.88-0.53 is studied with the upgraded Giant Meterwave Radio Telescope (uGMRT), located near Pune, India. Observations for G19.88-0.53 were carried out in Band 4 ($550-850$ MHz) and Band 5 ($1000-1450$ MHz). Observation and map details are listed in Table 1. The data were calibrated incorporating spectral index and spectral curvature across the wide band using the Astronomical Image Processing System (AIPS). Data reduction is carried out along the lines discussed in Issac et al. (2019). We refer the reader to this paper for an elaborate discussion of GMRT configuration and data reduction steps. In case of G19.88-0.53, calibrated and channel averaged data are split into sub-bands of bandwidth $\sim 32$ and $\sim 36$ MHz in Band 4 and Band 5, respectively. This is done with the envisaged goal of obtaining in-band spectral index of the radio source. We retain six clean sub-bands in Band 4 and three in Band 5, with central frequencies 651.4, 682.5, 714.1, 745.9, 777.1, and 808.3 MHz and 1320.7, 1355.8, and 1391.6 MHz, respectively. Continuum maps for each sub-band were generated separately by limiting the uv coverage to a common range of $\rm 0.15-36\,K\lambda$ with beams matched to 11.5 arcsec$\times$7.6 arcsec, the largest beam among the six sub-bands. Similarly, for Band 5 the uv range is fixed to $\rm 0.17-100\,K\lambda$ and beams matched to 4.3 arcsec$\times$2.5 arcsec. As discussed in Issac et al. (2019), scaling factors are estimated to range between 1.65 to 2.10 and 1.22 to 1.24 within Band 4 and Band 5, respectively. Using these, the primary beam corrected and self-calibrated sub-band maps are scaled to account for variations in the system temperatures of the antennas at the GMRT while observing targets located in the Galactic plane. .

We carried out the near-infrared (NIR) spectroscopic observations towards G19.88-0.53 with the 3.8-m United Kingdom Infrared Telescope (UKIRT), Hawaii. Observations were obtained on 2 April 2015 with the UKIRT $\rm 1-5\,\mu m$ Imager Spectrometer (UIST, Ramsay Howat et al. 2004) using the HK grism set-up. The slit ($\sim$0.48 arcsec wide and 120 arcsec long) was oriented at an angle 88°east of north and centered at ($\rm\alpha_{J2000}=18^{h}29^{m}14.7^{s},\delta_{J2000}=-11\degree 50\arcmin 16.7\arcsec$) as shown in Fig. 3. The exposure time per frame was 120s and the total on-source integration time was 720s. SAO 142372, an A5 IV/V type star, was also observed, with the same instrument settings as for the target, for telluric and instrumental corrections. Following the standard procedures explained in Issac et al. (2019), UKIRT pipeline ORAC-DR and Starlink packages FIGARO and KAPPA were used for data reduction, spectra extraction and wavelength calibration. As the sky conditions during our observations were not photometric, the final spectra are not flux calibrated.

To understand the physical properties of the cold dust emission associated with G19.88-0.53, we use Herschel far-infrared (FIR) archival data. The Herschel Space Observatory houses a 3.5-m telescope with instruments covering the spectral range of $\rm 55-671\,\mu m$ (Pilbratt et al., 2010). For our study, we have utilized observations carried out with the Photodetector Array Camera and Spectrometer (PACS, Poglitsch et al. 2010) and Spectral and Photometric Imaging Receiver (SPIRE, Griffin et al. 2010). The $\rm 70-500\,\mu m$ images retrieved were observed as a part of the Herschel infrared Galactic plane Survey (Hi-GAL, Molinari et al. 2010). The images have resolutions of 5, 13, 18.1, 24.9 and 36.4 $\arcsec$ at 70, 160, 250, 300, and 500 $\rm\mu m$, respectively.

APEX+Planck database provides 870 $\rm\mu m$ images. These are generated by combining 870 $\rm\mu m$ LABOCA bolometer array data from the APEX Telescope Large Area Survey of the Galaxy (ATLASGAL, Schuller et al. 2009) and 850 $\rm\mu m$ data from Planck/HFI. The combined data-set offers improved spatial dynamic range of Galactic cold dust emission (Csengeri et al., 2016). The retrieved APEX+Planck image of the region associated with G19.88-0.53 has pixel size of 3.4$\arcsec$ and angular resolution of 21$\arcsec$.

Atacama Large Millimeter Array (ALMA) archival data at 2.7 mm (Band 3) and 870 $\rm\mu m$ (Band 7) are used to study the G19.88-0.53 complex. Observations were made during 2017-2018 (PI: S. Leurini #2017.1.00377.S). The high- and medium-resolution ALMA observations were obtained with the 12-m array in the FDM Spectral Mode and the low-resolution observations used the 7-m array in the ACA Spectral Mode. The high resolution 2.7 mm continuum map, with a beam size of 0.46 arcsec$\times$0.28 arcsec, is used to identify the compact dust cores associated with G19.88-0.53. Along with the continuum observations, data cubes in both the bands are also retrieved. Three data sets with resolutions 16.9 arcsec$\times$8.0 arcsec, 2.6 arcsec$\times$2.2 arcsec, and 0.46 arcsec$\times$0.28 arcsec and velocity resolution of $\rm 0.67\,km\,s^{-1}$ in Band 3 ($\rm 84-116\,GHz$) and two data sets with resolutions 5.2 arcsec$\times$2.6 arcsec and 0.67 arcsec$\times$0.47 arcsec and velocity resolution of $\rm 0.86\,km\,s^{-1}$ in Band 7 ($\rm 275-373\,GHz$) are used. In Band 3, rms of 4.2, 3.6 and 17 mJy are achieved for high, medium and low resolution observations, respectively. In Band 7, rms of 3.5 and 17 mJy are achieved for high and low resolution observations, respectively. Analysis of molecular line data is carried out using CLASS90 (Continuum Line Analysis Single-dish Software), a GILDAS²²2http://www.iram.fr/IRAMFR/GILDAS software (Grenoble Image and Line Data Analysis Software).

Using the uGMRT, we probe the low-frequency domain of the ionized emission associated with G19.88-0.53. In Band 4, none of the sub-band images have detection above the $3\sigma$ level while faint, elongated emission is seen in the Band 5 maps. The deep uGMRT maps in Band 5 with a rms level of $\rm\sim 45\,\mu Jy\,beam^{-1}$ enable the detection of weak ionized emission. In comparison, there is no emission detected in the VLA 20 cm map from MAGPIS. The sub-band image at 1391.6 MHz is shown in Fig. 2. This map clearly reveals the presence of an elongated, two-component emission feature in the east-west direction. The two distinct and compact radio components are labelled R1 and R2. The contours of the 4.5 $\rm\mu m$ emission plotted show that the radio emission is confined to the central part of the EGO.

The radio emission associated with G19.88-0.53 has been studied by Zapata et al. (2006), Rosero et al. (2016), and Rosero et al. (2019) using VLA. Based on observations at 3.6 cm (8.5 GHz), 1.3 cm (22.5 GHz) and 7 mm (43.3 GHz), Zapata et al. (2006) classify this source as a ‘triple stellar system’ and identify three radio components (IRAS 18264-1152a, b, and c), the positions of which are shown in Fig. 2. In the recent paper, Rosero et al. (2016) present high-sensitivity, sub-arcsecond resolution observations at 6 cm (4.9 and 7.4 GHz) and 1.3 cm (20.9 and 25.5 GHz), where twelve compact radio sources are detected towards G19.88-0.53. Eight among them, sources $\rm A-H$, lie within the $3\sigma$ level of the 4.5 $\rm\mu m$ emission and are marked Fig. 2. The component R1 is co-spatial (within $\sim 1$ arcsec) with the radio sources, F and IRAS 18264-1152b. Component R2 has similar positional match with the radio source, A. The position of the MYSO located at the 4.5 $\rm\mu m$ emission peak (De Buizer & Vacca, 2010) also agrees well with the coordinates of R2. However, this component is not detected in the VLA maps presented by Zapata et al. (2006).

The position of the components and the peak flux densities are listed in Table 2. Component R1 is barely resolved and component R2 is unresolved in the uGMRT maps. It is thus difficult to get a reliable estimate of the integrated flux density for these unresolved sources with weak emission (Urquhart et al., 2009; Rosero et al., 2016). Hence, in the estimation of the spectral index we use the peak flux density. Following the discussion in Urquhart et al. (2009), the upper limits to the deconvolved sizes of R1 and R2 are taken to be half the FWHM of the restored beam size. This gives a value of 2.1 arcsec$\times$1.3 arcsec for the components.

As propounded by several authors (Caratti o Garatti et al., 2015, and references therein), shock-excited lines of H₂ and [Feii] and broad bandheads of CO are the spectral carriers responsible for the enhanced 4.5 $\rm\mu m$ emission of EGOs. Of the few spectroscopic studies of EGOs (e.g. De Buizer & Vacca, 2010; Caratti o Garatti et al., 2015; Onaka et al., 2016; Issac et al., 2019), De Buizer & Vacca (2010) focussed on MIR spectroscopy of two EGOs including G19.88-0.53 using the NIRI instrument on the Gemini North telescope. The spectra extracted towards the three associated knots of G19.88-0.53 show the presence of the shock-excited $\rm 0-0~{}S(9)$ line of H₂ at 4.695 $\rm\mu m$. Moreover no continuum emission was detected towards these knots. In our study, we investigate the NIR regime to identify shock indicators and confirm the association of G19.88-0.53 with protostellar outflows.

The H₂ line image towards G19.88-0.53, presented in Varricatt et al. (2010) reveals the presence of a bipolar outflow in the east-west direction which is consistent with CO outflow detected by Beuther et al. (2002b). Same results were obtained by Lee et al. (2012) using the UWISH2 survey and they enlist G19.88-0.53 as an EGO with a large-scale (angular scale of $\sim$ 1.6 arcmin) bipolar outflow. In Fig. 3 we display the continuum-subtracted H₂ line image from Varricatt et al. (2010) where the orientation of the slit position for UIST observations is marked. The identified apertures, K1, K2, and K3, over which the spectra are extracted are also highlighted in the figure. Molecular hydrogen objects (MHOs) from the catalogue of molecular hydrogen emission-line objects³³3http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/511/A24 (Davis et al., 2010) are also indicated. MHO 2245 and MHO 2246 are new additions to the catalogue by Lee et al. (2012) and Ioannidis & Froebrich (2012), respectively. Identified locations and extent of H₂ knots from Froebrich et al. (2011) are also shown in the figure. Apertures K1 and K2 probe the knots in MHO 2204 and K3 samples the central knot in MHO 2203. The extracted spectra are shown in Fig. 4 where we see strong detections of several H₂ lines. The lines detected and their corresponding wavelengths are listed in Table 3.

The H₂ lines detected in the spectra could have either a thermal or a non-thermal origin. Jets and outflows that are heated to typically few 1000 K can give rise to thermal emission from shock fronts. Whereas, non-thermal emission is ascribed to fluorescence produced by non-ionizing UV photons. Shock excited emission arises from low levels of excitation, whereas UV fluorescence populates both high and low-v states (Davis et al., 2003; Caratti o Garatti et al., 2015). Our spectra do not show the presence of transitions with v $\geq$ 6. In addition, the ratio of $\rm 1-0\,S(1)/2-1\,S(1)$ is estimated to be $\sim 8:1$. These suggest the origin of the detected H₂ lines to be due to thermal excitation. However, the absence of relevant fluorescent H₂ emission in G19.88-0.53 can also be attributed to high extinction that screens the UV photons emanating from the central star. Nonetheless, morphological resemblance of the extended H₂ emission knots to a bipolar jet and the association with an outflow source supports the shock excitation scenario. The shock-excited origin of the detected $\rm H_{2}$ lines confirms that the transitions of this molecule within the IRAC 4.5 $\rm\mu m$ band are among the spectral carriers responsible for the enhanced ‘green’ emission that classifies G19.88-0.53 as an EGO.

IRAC bands trace the thermal emission associated with the warm dust component and emission from the polycyclic aromatic hydrocarbons excited by the UV photons in the photodissociation regions. However, emission from the stellar photosphere dominates the shorter wavelength bands (3.6 and 4.5 um) (Watson et al., 2008). In addition to this, the 4.5 $\rm\mu m$ band includes emission from shock-excited H₂ lines and CO (v$=1-0$) band at 4.6 $\rm\mu m$ with contribution from the Br$\alpha$ and Pf$\beta$ lines. The spatial distribution of dust emission towards G19.88-0.53 is shown in Fig. 5(a), which is a colour composite image generated using the IRAC 8.0, 4.5, and 3.6 $\rm\mu m$ bands. The APEX+Planck 870 $\rm\mu m$ contours are overlaid along with the contours of the 4.5 $\rm\mu m$ emission. A clump is clearly identified at 870 $\rm\mu m$. Warm dust sampled at 8.0 $\rm\mu m$ is seen to be extended towards the south and west of the clump and, at 870 $\rm\mu m$, a clearly discernible large-scale filamentary structure is visible with the clump located within it.

Molecular line observations are tools for understanding the kinematics and chemical environment of a molecular cloud and and its stage in the evolutionary sequence. In this work, we use archival data from ALMA to study the emission from molecular lines in the region associated with G19.88-0.53.

Given that G19.88-0.53 is classified as an EGO, and hence likely associated with a jet/outflow, we are prompted to investigate the nature of the associated radio emission. As seen in Fig. 2(b), the ionized emission associated with G19.88-0.53 exhibits a linear morphology consisting of the components R1 and R2 in the east-west direction. The radio emission from both the components is weak ($\rm\sim 0.2\,mJy$) and they are located towards the centroid of a large scale ($\sim$ 1 arcmin) molecular outflow (Fig. 15a, Section 4.3.2). These traits are in agreement with the typical observational features expected in ionized jets (Anglada, 1996; Rodriguez, 1997; Purser et al., 2016; Rosero et al., 2019; Obonyo et al., 2019).

The spectral index is a crucial parameter that helps in understanding the nature of the radiation mechanism. This is defined as $S_{\nu}\propto\nu^{\alpha}$, where, $S_{\nu}$ is the flux density at frequency $\nu$. As discussed in Section 3.1, G19.88-0.53 is not detected in Band 4. Further, weak emission in Band 5 would render the in-band spectral index estimation highly unreliable as has also been discussed by Rosero et al. (2016). Emission at 1391.6 MHz encompasses the component IRAS 18264-1152b. This makes it possible to use this uGMRT sub-band and VLA maps from Zapata et al. (2006) to derive the spectral index of R1. An accurate determination of the spectral index requires same spatial scales to be probed at the frequencies used. The synthesized beam sizes of uGMRT 1391.6 MHz and VLA 8 GHz (3.6 cm) maps are nearly identical indicating similar uv coverage and hence similar spatial scales probed. This is not the case for the 1.3 cm data. For this, we convolve the map to the match the resolution of the other two frequencies. The radio SED is shown in Fig. 11. For 651.4 MHz, we obtain an upper limit for the flux density (rms of the map) which is shown in the plot. For the VLA maps, we use the peak flux density as well. The fit to the radio SED of component R1 (IRAS 18264-1152b) yields a spectral index value of $0.58\pm 0.19$. that corroborates well with radio continuum emission originating due to the thermal free-free emission from an ionized collimated stellar wind (Rosero et al., 2019, and references therein). For component R2, similar analysis of spectral index estimation is not possible given that this component is not detected in the VLA maps of Zapata et al. (2006).

The spectral index estimation of R1 obtained from the combination of uGMRT and VLA maps of Zapata et al. (2006) is in excellent agreement with value of $0.6\pm 0.1$ derived by Rosero et al. (2016) using flux densities at four central frequencies (4.9 and 7.4 GHz at 6 cm and 20.9 and 25.5 GHz at 1.3 cm). In comparison, Zapata et al. (2006) obtain a lower value of $0.27\pm 0.06$. The 3.6 cm flux density of IRAS 18264-1152b used by these authors would also include emission from IRAS 18264-1152a and IRAS 18264-1152c thus rendering a shallower slope. Based on the weak emission, elongated morphology, association with molecular outflows and rising spectral index values, the radio emission favours shock ionization over the HII region picture. Hence, Rosero et al. (2019) list source F (R1) as a ionized jet/wind candidate. Inference of a likely thermal jet or a partially optically thick HII region is drawn by Zapata et al. (2006) as well. The identification of R1 as a thermal jet is well supported by the detection of several shock-excited H₂ lines in the UKIRT-NIR spectra of G19.88-0.53 presented in Section 3.2. To the best of our knowledge, this is the first detection of the ionized jet associated with G19.88-0.53 at the lower frequency regime of uGMRT. Similar result is obtained by our group for EGO G12.42+0.50 (Issac et al., 2019). The authors propose the coexistence of a thermal jet and a UC Hii region driven by the same MYSO to explain the observed radio, mm, and IR emission for G12.42+0.50.

EGO G19.88-0.53 has been proposed as an outflow powered by a MYSO (De Buizer & Vacca, 2010) and a multiple outflow source (Varricatt et al., 2010). Several studies (e.g. Cyganowski, Brogan & Hunter, 2007; Cyganowski et al., 2011) have shown that MYSOs are actually protoclusters harbouring protostars with different masses and in various stages of evolution. Given the predilection of high-mass stars to form in clustered environments, it is likely that EGOs, in fact harbour protoclusters rather than single high-mass stars. Based on SOFIA FORCAST imaging and archival infrared data of twelve EGOs, Towner et al. (2019) suggest the number of massive sources per EGO to be between 0.9 to 1.9. Similar inferences have been drawn from studies towards other identified EGOs (e.g. Brogan et al., 2011; Cyganowski et al., 2012).

The 870 $\rm\mu m$ clump cocoons the EGO G19.88-0.53 where the extended 4.5 $\rm\mu m$ emission lies towards the centre as is seen in Figure 5(a). In Fig. 12, we present the various radio and mm sources associated with this EGO. The 2.7 mm dust emission, with the cluster of detected cores lie deeply embedded within the 4.5 $\rm\mu m$ emission. Portion of the $3\sigma$ contour of the 4.5 $\rm\mu m$ emission is seen in the region displayed in the figure. Within the dust emission lie the two radio components R1 and R2 mapped at 1391.6 MHz with uGMRT. As discussed in the previous section, R1 (VLA source F) is most likely an ionized thermal jet. Regarding component R2 (VLA source A) and the other six compact VLA sources (B, C, D, E, G, and H) located within the EGO, Rosero et al. (2016) discuss the likelihood of them being Hii regions around rapidly accreting massive stars that quench the UV photons thus giving rise to weak radio sources. They also keep alive the debate of these being ionized jets instead. Zapata et al. (2006) also suggest that the component IRAS 18264-1152c (VLA source E) is likely to be an optically thick Hii region. These authors detect an additional component IRAS 18264-1152a of similar nature taking the number of distinct radio sources to be nine. Furthermore, from the analysis of ALMA 2.7 mm high-resolution continuum map, we discover six compact cores located within the EGO. Table 7 lists the radio and mm sources and their cross-identification.

Given the detection of several radio components and mm cores, the picture of a protocluster with multiple (eleven) members is clearly perceivable. The observational signatures of the members show clear evidence of different evolutionary phases. However, high-sensitivity radio surveys of ionized jets have shown association with extended lobes and string of radio knots (e.g. Garay et al., 2003; Guzmán, Garay & Brooks, 2010; Purser et al., 2016; Hofner et al., 2017). From the alignment of the radio sources B, C and F, and the spectral index of F, one can visualize a similar scenario. It is thus plausible that we are probing a string of knots of a single ionized jet driven by R1/MM2/F. Similarly, the components A and G could be part of this jet system though the possibility of another jet driven by R2/MYSO/A cannot be ruled out. If these radio components are indeed knots/lobes of the identified jet-systems, then the number of distinct members of the detected protocluster would be less than eleven. Further studies are required to interpret the exact nature the compact radio sources.