A highly collimated jet from the Red Square nebula MWC 922

Abstract

Deep, narrow-band H α and 6584 Å [N ii] CCD images of the peculiar, infrared excess B[e] star MWC 922 reveal a collimated, bipolar jet orthogonal to the previously detected extended nebula. The jet consists of a pair of ∼0.15 pc segments on either side of MWC 922 separated by gaps. The most distant jet segments disappear ∼0.6 pc from the star. The north-west beam points to a faint emission-line feature 1.65 pc from MWC 922 that may be a terminal bow shock where the jet rams the ambient medium. The narrow opening-angle of the jet combined with an estimated internal sound speed of ∼10 km s⁻¹ implies a jet speed ∼500 km s⁻¹. The previously detected nebula extends up to 0.6 pc to the south-west of MWC 922 at right angles to the jet and appears to be an extension of the compact, edge-on disc surrounding the star. It points towards the H ii region Messier 16 located ∼1° (∼30 pc in projection) to the south-west. This nebula and jet appear to be externally ionized by the ambient Lyman continuum radiation field and have electron densities of n_e ∼ 50–100 cm⁻³. The south-west nebula and jet have similar surface brightness in H α and [N ii]. Faint 70 μm emission traces the south-west ejecta that likely originates from ∼50 K dust embedded in the photoionized plasma which may shadow the dimmer ejecta north-east of MWC 922. MWC 922 may be a massive member of the Serpens OB1 or OB2 associations surrounding Messier 16 and Sh2-54.

stars: AGB and post-AGB, stars: individual (MWC 922), ISM: jets and outflows

1 INTRODUCTION

MWC 922 (IRAS 18184-1302), also known as the Red Square nebula (RSN) based on its remarkable X-shaped near-infrared (near-IR) morphology (Tuthill & Lloyd 2007), is a dust/gas-enshrouded emission-line star with strong infrared excess at mid-IR wavelengths (see Fig. 1). The spectrum of MWC 922 is dominated by strong emission in hydrogen recombination lines, a forest of fainter [Fe i] and [Fe ii] lines, molecular emission and absorption features, and some diffuse interstellar bands (Wehres et al. 2017).

Figure 1.

The RSN showing an H-band adaptive optics image combining data from the Palomar Observatory 200-inch telescope, which was published in Tuthill & Lloyd (2007) with adaptive optics data from the Keck Observatory 10 m. North is up and east is to the left. This image was featured on APOD in 2016 January 31. See https://apod.nasa.gov/apod/ap160131.html. Image credit: Peter Tuthill, Sydney University Physics Dept., and the Palomar and W.M. Keck observatories.

Lamers et al. (1998) classified B[e] stars into various subgroups, but catalogued MWC 922 as an unclassified ‘uncl B[e]’ star owing to its peculiar atomic, ionic, and molecular spectral features. Some B[e] stars are young and closely associated with their parent molecular clouds, and others appear to be main-sequence multiples experiencing mass transfer on to a companion, while the rest appear to be post-main-sequence objects. Miroshnichenko (2007) clarified the nature of the ‘uncl B[e]’ stars, and renamed members of this class as ‘FS CMa’ stars. FS CMa stars are binary systems that are currently undergoing, or have recently undergone a phase of rapid mass exchange, strong mass-loss, and dust formation. However, the evolutionary stage of MWC 922 remains unclear.

Rodríguez, Báez-Rubio & Miroshnichenko (2012) detected a bright, compact 3.6 cm radio source having dimensions of 0.18 arcsec × 0.20 arcsec in the centre of the RSN. Sánchez Contreras et al. (2017) detected radio recombination lines H30 α, H31 α at λ = 3 mm and H39 α, H41 α at λ  = 1 mm along with several H radio-recombination lines in the β and γ series. From the velocity centroids, Sánchez Contreras et al. (2017) measured the LSR radial velocity of MWC922 to be V_LSR  = + 32.5 ± 0.4 km s⁻¹. The radial velocity combined with the Galactic rotation curve implies D  = 1.7 to 3 kpc if located at the near distance (Sánchez Contreras et al. 2017).

The bright, inner nebulosity surrounding MWC 922, which is about 10 arcsec in diameter, is similar to the well-studied proto-planetary nebula (pPN) known as the Red Rectangle (Bujarrabal et al. 2016). However, the symmetry of the RSN indicates that its disc and the biconical outflow cavity-oriented orthogonal to the disc are seen nearly edge-on. Indeed, adaptive optics images in the near-IR (Fig. 1) show a nearly edge-on equatorial dust lane with a major axis at position angle PA  = 46°.

Marston & McCollum (2008) detected an extended and elongated H α and [N ii] nebula associated with the RSN oriented north-east–south-west, with a major-axis dimension of at least 110 arcsec and a minor-axis dimension of about 10 arcsec to 20 arcsec. The brightest portion of this nebula is located south-west of MWC 922 and points towards Messier 16 located about 1° (∼30 pc in projection) to the south-west at PA ∼ 217°.

If MWC 922 is related to the Messier 16 star-forming complex it may be a member of the Serpens OB1 association. If so, it is likely to have a distance between ∼ 1.7 and 2.0 kpc. Guarcello et al. (2007) investigated the stellar population associated with Messier 16 for which a distance of 1.75 kpc was found. Thus, a value of 1.7 pc is adopted for the distance of MWC 922 in this study.

The radial velocity of molecular gas in the vicinity of Messier 16 ranges from V_LSR  = 17–29 km s⁻¹, about ∼10 km s⁻¹ lower than the gas associated with MWC 922 (Pound 1998; Nishimura et al. 2017). A projected separation of 30 pc between MWC 922 and the core of Messier 16 and a radial velocity difference of 10 km s⁻¹ either implies that the two objects are unrelated, or that MWC 922 is a low-velocity runaway star from Serpens OB1.

The Serpens OB2 association associated with the H ii region Sh2-54 and NGC 6604 located about 1.5° to the north-west of MWC 922 is also at a similar distance (Davidge & Forbes 1988; Reipurth 2008). The combined light from massive stars in Serpens OB1 and OB2 may be responsible for the extended diffuse H α emission in the region and for external ionization of the nebulosity and jets associated with MWC 922.

The spectral energy distribution (SED) of MWC 922 (Sánchez Contreras et al. 2017) implies a luminosity of L ∼ 1.8 × 10⁴ L_⊙ at D  = 1.7 kpc after correction for foreground extinction, estimated to be between A_V  = 1.7–2.7 mag.

2 OBSERVATIONS

Several R ≈ 300 spectra covering 2500  Å-wide spectral windows around H α and H β were acquired with the double imaging spectrograph (DIS) on the 3.5 m telescope at the Apache Point Observatory (APO) located near Sunspot, New Mexico, USA in 2012 August 19. A grating with a dispersion of 2.31 Å per pixel and a spatial scale of 0.4 arcsec per pixel was used. For three 600 s duration exposures, the 5 arcmin by 1.5 arcsec wide long-slit was centred on MWC922 and oriented at PA  = 45° along the major axis of the elongated nebula surrounding the star. A pair of 600 s exposures was also obtained at an offset position 5 arcsec north-west at the same position angle. These spectra show that the nebula extends over 1 arcmin south-west from MWC922 and is nearly equally bright in H α and the 6584 Å [N ii] emission line. Consequently, we used the APO 3.5 m f/10 telescope to acquire narrow-band images that isolated the H α, [N ii], and the red [S ii] doublet. An additional pair of R ≈ 3000 spectra, each covering 1160 Å spectral windows around H α and 1240 Å around H β, were obtained with DIS using exposure times of 600 s in 2017 March 22. The scale of these spectra is 0.58 Å per pixel in the red and 0.74 Å in the blue.

Narrow-band images were obtained between 2016 July 4 and 2018 May 13 using the ARCTIC CCD camera on the APO 3.5 m telescope. ARCTIC uses a backside-illuminated 4096 × 4096 pixel chip with 15 μm pixel pitch, an image scale of 0.114 arcsec per pixel in unbinned mode, and a 7.8 arcmin ×7.8 arcmin view of view. To better match to the native seeing at APO, the observations reported here used 3 × 3 binning to provide a scale of 0.344 arcsec per pixel, implying that there are N_pix  = 8.427 pixels per square-arc-second. Table 1 summarizes the observation dates and exposure times. Images were obtained using narrow-band filters having bandpasses of 30 Å centred on the 6563 Å H α and 6584 Å [N ii] emission lines (UNM657/30, UNM659/30) and a 100 Å bandpass filter centred on the red λλ6717/6731Å [S ii] doublet (CU S ii) with exposure times ranging from 300 to 900 s per frame. The images were processed in the standard fashion using bias, dark, and twilight flat frames. Between 2016 and 2018, a pattern of elevated dark current became apparent in the images. While the 2016 data could be processed by subtraction of a median-filtered stack of bias frames, the 2018 image required subtraction of dark frames to remove these artefacts. The dark-subtracted images were flat-fielded using twilight flats. The processed, individual frames were de-distorted using 2MASS coordinates of stars in the field. Final images in each filter were combined using a median combination of the registered frames.

Table 1.

Observation summary.

Date	Filter	Exposure time
2012 Aug. 19	DIS, R = 300	3 × 600; 2 × 600
2017 Mar 22	DIS, R = 1200	2 × 600
2016 July 04	UNM657/3	3 × 300
	UNM659/3	3 × 300; 3 × 900
2016 July 09	UNM659/3	4 × 600
	CU673/10	3 × 300
2016 Sept. 04	UNM659/3	4 × 600
2016 Sept. 05	UNM659/3	5 × 900
	UNM657/3	9 × 300
2018 May 13	UNM657/3	3 × 300 s
	UNM659/3	6 × 300 s

Approximate photometric calibration is based on the SDSS r-filter magnitudes of non-variable (at a few per cent level), unsaturated, in-field stars identified in the Pan-STARRS catalogue. For both the UNM657/30 and UNM659/30 filters, the zero-point (ZPT; defined as the magnitude of a star that produces 1 adu per second per pixel if all the starlight were concentrated into one pixel) is ZPT  = 20.93 ± 0.1. For the 100 Å bandpass [S ii] filter, ZPT = 22.23 ± 0.1. These values were used to convert counts in the images to flux densities in a given photometry-aperture (F_ν in Jy and erg s⁻¹ cm⁻² Hz⁻¹) using the flux density of Vega (approximately 3174 Jy at the H α wavelength of 6563 Å and 3085 Jy at the mean wavelength of the red [S ii] doublet, 6724 Å). Background-subtracted emission-line surface brightness (SB) values (erg s⁻¹ cm⁻² arcsec⁻²) are obtained by multiplication of the flux density by the bandpasses of the filters (∼30Å for H α and [N ii] and ∼100Å for [S ii]) and division by the areas (in square arc-seconds) of the photometry apertures. Fluxes and SB measurements were converted into estimates of the emission measures (EMs), electron densities, mass-loss rates, momentum injection rates, and mechanical luminosities as discussed below.

The images shown in Figs 2–5 have been converted to SB units (erg s⁻¹ cm⁻² arcsec⁻²) and adjusted to remove added flux from scattered light, light pollution, and airglow. The Southern H-alpha Sky Survey Atlas (SHASSA; Gaustad et al. 2001) was used to estimate the amount of added flux in our images. After conversion, the data values in each pixel give the observed SB in erg s⁻¹ cm⁻² arcsec⁻².

Figure 2.

The RSN showing the [N ii] emission using a logarithmic display. The displayed SB ranges from 0 to |$\rm 1 \times 10^{-15}$| (erg s⁻¹ cm⁻² arcsec⁻²). Labels indicate the various features discussed in the text and listed in Table 2.

The conversion from data values to SB units is achieved by multiplication by the following conversion constants: 2.814 × 10⁻¹⁶ N_pix/τ_exp for H α and [N ii] and 8.499 × 10⁻¹⁷ N_pix/τ_exp for [S ii] where τ_exp is the effective exposure time in seconds (300 s for H α, 900 s for [N ii], and 600 s for [S ii] in the median combined final images).

The added foreground flux is taken to be the difference between the continuum-subtracted SHASSA H α SB and the diffuse H α SB in our ARCTIC H α image. This SB value is subtracted from the final H α image. The ARCTIC [N ii] and [S ii] images are corrected for the added foreground by assuming that the background emission in these emission lines is 0.3 and 0.12 times the H α SB as was determined from our DIS spectra.

The SB scales in Figs 2–5 assume A_V  = 0 mag of interstellar extinction. However, for the analysis below, it is assumed that the extinction to RSN is A_V  = 2 mag. It is evident from the H α image that the extinction across the field is highly structured and variable. Thus, the physical parameters are uncertain by at least a factor of two.

3 RESULTS

The area-integrated flux ratio |$I_{[\mathrm{ N}{\mathrm{\small II}}]} / I_{\mathrm{ H} \alpha }$| is 0.8 ± 0.1 (in the DIS spectra) in a region extending south-west of MWC 922 from 20 arcsec (0.16 pc) to 60 arcsec (0.5 pc) from the star. At positions away from the RSN, |$I_{[\mathrm{ N}{\mathrm{\small II}}]} / I_{\mathrm{ H} \alpha }$| is about 0.3 ± 0.1 everywhere. Thus, the [N ii] emission is enhanced relative to the H α light measured at locations where the slit crossed the RSN. As discussed below, unlike in [N ii], there is no enhancement in the [S ii] emission from the RSN. The λλ6717/6731 line ratio is around 1.3 ± 0.1 in the same DIS aperture in which the [N ii]/H α ratio is enhanced. This is close to the low-density limit for the [S ii] line ratio and implies a density n_e <240 cm⁻³. The mean [S ii] doublet ratio from the background nebula is also near the low-density limit, ∼1.3 ± 0.1, implying a mean density <240 cm⁻³.

Figs 2 and 3 show the outer structure of the RSN in unprecedented detail in [N ii]. Fig. 2 shows the [N ii] image labelled with the various components discussed in the text. Fig. 4 shows the RSN in [N ii] in a deep cut, Fig. 5 shows the RSN in H α, and Fig. 6 shows the same field in [S ii]. The brightest part of the nebula consists of an approximately 2 arcmin-long north-east−south-west-oriented ‘bar’ of emission seen in both H α and 6584 Å [N ii] emission (‘SW disc’). This is the feature detected by Marston & McCollum (2008). The south-west portion facing Messier 16 is much brighter than the north-east portion and is split by a dark band of obscuration that tapers from a width of about 5 arcsec near MWC 922 to about 2 arcsec at its south-western edge located about 60 arcsec from MWC 922. The north-west-facing surface of this band is brighter and thicker in both H α and [N ii] than its south-east-facing surface. These two surfaces merge into a dim, luminous extension which bends due west about 70 arcsec of MWC 922 and can be traced for another 19 arcsec in this direction as a bent ‘tail’ (‘SW tail’). As discussed below, the dark band likely traces a nearly edge-on disc shed by MWC 922 whose surface layers are externally ionized by the ambient radiation field.

Figure 3.

The RSN showing [N ii] emission using a logarithmic display and shown without labels to highlight the brighter components of the extended MWC922 nebula. The displayed SB ranges from 0 to |$\rm 1 \times 10^{-15}$| (erg s⁻¹ cm⁻² arcsec⁻²).

Figure 4.

The RSN showing the [N ii] emission using a linear display emphasizing the fainter components of the extended MWC922 nebula. The displayed SB ranges from 0 to |$\rm 3 \times 10^{-16}$| (erg s⁻¹ cm⁻² arcsec⁻²).

Figure 5.

The RSN showing the H α emission using a logarithmic display showing the full range of nebular emission. The displayed SB ranges from 0 to |$\rm 2 \times 10^{-15}$| (erg s⁻¹ cm⁻² arcsec⁻²).

Figure 6.

The RSN showing the [S ii] emission using a logarithmic display. The displayed SB ranges from 0 to |$\rm 1 \times 10^{-15}$| (erg s⁻¹ cm⁻² arcsec⁻²).

Towards the north-east, in the direction facing away from Messier 16, the emission is much dimmer (‘NE disc’). It is possible that the ‘SW disc’ and ‘SW tail’ discussed above shield the ‘NE disc’. Interior to the ‘SW disc’ and ‘NE disc’ but outside the RSN, there is a dim, roughly hexagonal region approximately 40 arcsec in diameter (‘Hexagon’ in Fig. 2). The hexagon is centred on the Red Square whose bright emission forms a 11 arcsec diameter square centred on MWC 922 (shown in Fig. 1). The outer edges of the hexagon towards the north and east are slightly brighter than the interior or the south-east and north-west-facing edges.

There is a highly collimated, segmented, bipolar jet emerging from MWC 922 along a south-east-north-west axis at PA ≈ 134° (towards the south-east) and ≈316° (towards the north-west). The jet is oriented perpendicular to the edge-on disc whose disc plane has position angle 46° (Tuthill & Lloyd 2007) and the extended nebulae extending towards the south-west in the RSN to within 2°. The innermost segments, NW1 and SE1, emerge from the RSN and can be traced for about 18 arcsec and 16 arcsec from MWC 922. These segments become progressively dimmer with distance from the star. Beyond these inner jet segments, there is a gap extending from 17 arcsec to about 40 arcsec from the star towards both the south-east and north-west. The jet becomes visible again from 40 arcsec to 56 arcsec from MWC 922 towards the south-east (jet segments ‘SE2’). A corresponding segment extends from 44 arcsec to 68 arcsec towards the north-west (‘NW2’). The most distant parts of the jet are marginally resolved with an observed width of 1 arcsec to 2 arcsec. Deconvolution of the ∼1 arcsec seeing disc from the 2 arcsec width about 1 arcmin from MWC 922 results in an estimate of the jet opening angle, about 1.5 to 1.7°. These jet segments have similar structure and SB in the H α and [N ii] images and are not seen at our level of sensitivity in [S ii]. However, the base of the jet is faintly detected as a filament extending up to 15 arcsec from MWC 922 towards both the north-west and south-east with a SB of about 10⁻¹⁷ erg s⁻¹ cm⁻² arcsec⁻² in [S ii] (Fig. 6).

The most distant feature potentially associated with the jet from MWC 922 is a faint 1 arcsec by 5 arcsec emission-line feature visible in both H α and [N ii] but not in [S ii] dubbed ‘NW shock’. ‘NW shock’ is located 194 arcsec north-west from MWC 922 and its centroid is at position angle PA  = 316° with respect to the source star. It lies on the north-west jet axis, and its long dimension is at right angles to the jet. The angular diameter of this feature as seen from MWC 922 is about 3°. The complete absence in our [S ii] images that were obtained with a filter having over three times the bandpass of the H α and [N ii] filters makes it unlikely to be a chance superposition of unresolved, faint stellar images. Thus, the NW shock is a pure emission-line object and, given its location on the north-west jet axis, may be a part of the MWC 922 outflow. If this is a bow shock where the jet impacts slower moving or stationary material, sideways splashing of plasma entering the shock may explain its angular size as seen from the jet source. The H α divided by [N ii] flux ratio is about 2.5, lower than the values in the various parts of the RSN and close to typical H ii region values. No counterpart is seen towards the south-east axis.

Table 2 lists some of the observed and derived characteristics of features described in this section and marked in Fig. 2. The contrast between the jet segments, the terminal bow shock, and the diffuse background that fills the images is greatest in the [N ii] image, making the jet most visible there. The I(H α)/I([N ii]) flux-ratio is around 1 ± 0.3 in the jet segments. In the north-east–south-west RSN, this ratio is around 1.35 ± 0.1, much higher than in H ii regions. On the other hand, the I(H α)/I([S ii]) ratio in the RSN is around 8.4 ± 0.5, similar to the values in H ii regions. The jet and terminal bow are not seen in the [S ii] image, indicating that the flux in H α is at least about an order-of-magnitude larger than in [S ii].

Table 2.

MWC922 dimensions, SB, and derived parameters.

Name	Da	PAb	Rc	t|$^d_{\mathrm{ dyn}}$|	SBe	EM(0)f	EM(2)g	neh	Li	Comments
	( arcsec)	(°)	(pc)	(yr)	(e)	(cm6 pc)	(cm6 pc)	(cm−3)	( arcsec)
SE1	0–18	134	0–0.15	290	0.17	11	46	36–74	1	From [N ii]
Gap SE1	18–40	–	0.15–0.34	645	–	–	–	–	–	Gap
SE2	40–56	134	0.34–0.48	903	0.37	18	77	38–78	1.5	From H α
					0.27	18	74	37–76	’	From [N ii]
RSN	11–11	–	0.09–0.09	–	–	–	–	–	–	RSN
NW1	0–16	316	0–0.14	258	0.62	40	69	69–141	1	From [N ii]
Gap NW1	17–44	316	0.14–0.37	709	–	–	–	–	–	Gap
NW2	44–68	316	0.37–0.58	1095	0.35	17	71	37–75	1.5	From H α
					0.23	15	63	34–70	’	From [N ii]
NW shock	194	316	1.65	3142	1.0	51	216	35–71	5	From H α
					0.40	26	109	25–51		From [N ii]
Hexagon	21–21	–	0.18	–	1.5	72	306	20–40	23	From H α
			0.18	–	1.5	72	304	19–40		From [N ii]
NE ‘disc’	10–33	44	0.28	–	1.4	69	291	16–34	30	From H α
				–	1.05	68	287	16 –34		From [N ii]
SW ‘disc’	19–71	224	0.60	–	6.3	305	1287	35–71	30	From H α
				–	5.1	331	1392	36–74		From [N ii]
SW ‘tail’	71–90	∼260	0.60–0.79	–	2.0	97	409	25–50	19	From H α
				–	1.11	72	304	21–43		From [N ii]

Notes. For features only seen in [N ii], electron density was estimated from the [N ii] SB multiplied by 1.33, the measured H α/ [N ii] intensity ratio. See text for details.

^aD is the projected distance from MWC 922 in arcsec.

^bPA is the position angle of the feature measured from north to east.

^cR is the projected distance from MWC 922 in parsecs assuming a distance of 1.7 kpc.

^dt|$^d_{\mathrm{ dyn}}$| = R/V_jet is the dynamical time assuming a jet velocity, V_jet  = 500 km s⁻¹.

^eSB is the H α or [N ii] SB given in units of 10⁻¹⁶ erg s⁻¹ cm⁻² arcsec⁻².

^fEmission measure assuming A_V  = 0 mag.

^gEmission measure assuming A_V  = 2 mag.

^hElectron density range for A_V  = 0 to 2 mag, assuming a line-of-sight depth L from column i.

ⁱThe assumed line-of-sight depth of the feature expressed as the width of a feature in arcseconds on the images.

Table 2 gives two sets of entries for most features with the first being measured on the H α image and the second estimated from the [N ii] image. Some features such as the inner jet segments SE1 and NW1 are only clearly seen in [N ii], likely because the background from the extended H α emission is so bright. The sixth column (superscript e) in Table 2 gives the measured SB values for H α and [N ii] for various components of the RSN. Note that these are differential measurements in which an OFF-source value is subtracted from the ON-source value using the same aperture. The [N ii] SB values are multiplied by 1.33, the measured mean H α/[N ii] SB ratio, to estimate the expected H α SB values. These values are then used to compute emission measures and electron densities under two assumptions about foreground extinction; A_V  = 0 or A_V  = 2 mag. For the electron density estimation, it is assumed that the line-of-sight (LOS) depth, L, of the emission region is comparable to the jet width for the jet segments, or the maximum length of the emission for the extended RSN ‘disc’ segments. The value used for the path-length L is given in column i in Table 2.

4 DISCUSSION

In this section, the observations are used to deduce the jet velocity and mechanical parameters followed by a discussion of the likely properties of the excretion disc surrounding MWC 922, and the relationship to the Serpens OB1 or OB2 associations and the major star-forming complexes Messier 16 and Sh2-54.

4.1 Jet physical properties determined from H α emission

The H α SB of the jet segments can be estimated by using the photometric ZPT of the images. A rectangular aperture surrounding each region of interest is used to measure the total counts on the labelled jet segments of the MWC 922 system (ON). The sky-level in the surrounding region is measured using the same aperture (OFF). The difference in counts, ON–OFF, is used to compute the total magnitude of an extended nebular feature

\begin{eqnarray*} m_{\nu } = -2.5 \, \mathrm{ log}_{10} [ (\mathrm{ ON}-\mathrm{ OFF}) / t_{\mathrm{ exp}}] + \mathrm{ ZPT}, \end{eqnarray*}

where t_exp is the effective exposure time (ZPT here is in units of counts per second for the telescope and CCD).

The flux in the aperture from each region of interest is then

\begin{eqnarray*} F = 1.0 \times 10^{-23} F_{\nu }(\mathrm{ Vega}) \Delta \nu \, e^{\tau _{\nu }} \, 10^{-m_{\nu } / 2.5} \, (\mathrm{ erg}\, \mathrm{ s}^{-1}\, \mathrm{ cm}^{-2}), \end{eqnarray*}

where F_ν(Vega) is the flux density of Vega in the photometric system in which the stellar magnitudes of in-field standards were measured (SDSS r or Johnson R) in Janskys (Jy). Δν is the bandpass of the filter (30 Å or 100 Å in these observations) expressed in Hz units. τ_ν is the optical depth of foreground dust, τ_ν = A_ν/1.086, where A_ν is determined from the interstellar medium (ISM) reddening curve at ν = c/λ, the frequency of the H α emission line. Fitting the standard reddening curve for R_V ≈ 3.1 between λ  = 5500 Å at which the extinction A_V is usually quoted and the 6563 Å wavelength of H α implies a wavelength dependence of A_λ/A_V proportional to λ^−1.4 implying that A_H α = 0.78 A_V.

The SB of an extended nebular emission feature is then

\begin{eqnarray*} \mathrm{ SB} = F / A \, (\mathrm{ erg}\, \mathrm{ s}^{-1}\, \mathrm{ cm}^{-2}\, \mathrm{ arcsec}^{-2}), \end{eqnarray*}

where A is the projected area of the measurement box enclosing the radiating feature measured in square arcseconds.

The emission measure is related to the SB by

\begin{eqnarray*} \mathrm{ EM} = n_\mathrm{ e}^2 L_{\mathrm{ pc}} = 4.86 \times 10^{17} \mathrm{ SB} \times T_4^{0.9} \, (\mathrm{ cm}^{-6} \mathrm{ pc}) \end{eqnarray*}

where n_e is the electron density (in cm⁻³) and L_pc is the depth of the emitting region along the LOS in units of parsecs (pc). Here, T₄ is the plasma temperature in units of 10⁴ K (Haffner, Reynolds & Tufte 1998). Under the assumption that the LOS depth is identical to the spatial extent of the emission region on the plane of the sky, this formula can be inverted to solve for the electron density of the radiating plasma, assuming it is fully ionized:

\begin{eqnarray*} n_\mathrm{ e} = (\mathrm{ EM} / L_{\mathrm{ pc}})^{1/2} \, (\mathrm{ cm}^{-3}). \end{eqnarray*}

Under these assumptions, the mass per unit area of the emission region M/A, and for material moving with a velocity V, the mass-loss rate |$\skew4\dot{M}$| associated with the flow, the momentum injection rate |$\skew5\dot{P}$|⁠, and the mechanical luminosity (kinetic energy per second) |$\skew5\dot{E}$| can be estimated.

Given the low SB of the MWC 922 jet, we have no direct measure of its radial velocity or internal velocity dispersion. Furthermore, given the detected disc orientation that is nearly edge-on, we expect the predominant motion of the jet to be in the plane of the sky. However, since this is the first detection of the jet and the jet is too faint to have been detected on previous images, no proper motion measurements are available.

Therefore, we estimate the jet velocity from the opening angle of the north-west jet as seen from MWC 922. Photoionized hydrogen-dominated plasmas tend to be thermostated to temperatures in the range of 5000–10 000 K when a near-Solar abundance of heavy elements are present. Emission lines of relatively abundant trace ions and neutrals such as [O i], [O ii], [O iii], and [S ii] that have ∼2 eV transitions tend to set the temperature in this range. The internal sound speed in such a plasma, c_s = (kT/μm_H)^1/2, is around 5.4–11.0 km s⁻¹ for a mean molecular weight μ = 1.4 for mostly neutral atomic gas at T  = 5000 K, to μ = 0.7 for fully ionized plasma at 10⁴ K.

Freely expanding thermal jets unconfined by the pressure of the ambient medium or magnetic fields will expand at the Mach angle given by

\begin{eqnarray*} \theta _{\mathrm{ Mach}} \approx 2 c_\mathrm{ s} / V_{\mathrm{ jet}}. \end{eqnarray*}

Using the estimated opening angle of the jet discussed above of 1.5–1.7°, combined with the sound speed estimated from an assumed temperature of 5000–10 000 K implies a jet velocity of V_jet  = 360–840 km s⁻¹. Given an estimate for the jet velocity, V_jet, the jet mass-loss rate is

\begin{eqnarray*} \skew4\dot{M} = \mu m_\mathrm{ H} n_e V_{\mathrm{ jet}} \pi r^2 _{\mathrm{ jet}}, \end{eqnarray*}

where r_jet is the jet radius which, assuming cylindrical symmetry, is given by L_pc/2, and V_jet is the estimated jet velocity. The momentum injection rate and kinetic energy injection rate (mechanical luminosity) are given by

\begin{eqnarray*} \skew5\dot{P} = \skew4\dot{M} V_{\mathrm{ jet}} \end{eqnarray*}

\begin{eqnarray*} \skew5\dot{E} = (1/2) \skew4\dot{M} V^2_{\mathrm{ jet}}. \end{eqnarray*}

The emission measure, EM, the assumed LOS path-length, and resulting estimates for the electron densities are given in Table 2. For an electron density ranging from 50 to 100 cm⁻³, an estimated jet speed of 500 km s⁻¹, mean jet diameter of 1 arcsec (= 1700 au), and A_V  = 2 mag of foreground extinction, the mass-loss rate, momentum injection rate, and mechanical luminosity of the jet segments are |$\skew4\dot{M} \sim 5 - 10 \times 10^{-8}$| M_⊙ yr⁻¹, |$\skew5\dot{P} \sim 2.5 - 5 \times 10^{-5}$| M_⊙ km s⁻¹ yr⁻¹, and |$\skew5\dot{E} \sim 4 - 8 \times 10^{33}$| erg s⁻¹  = 1–2 L_⊙, respectively.

MWC 922 exhibits the characteristics of both an evolved red giant or supergiant and a hot star. Tuthill & Lloyd (2007) quote a spectral type of B3 to B6. The strong IR-excess, however, indicates the presence of dust typically found in an evolved object. The detection of a circumstellar disc and a highly collimated jet emerging at right angles to the disc provides indirect evidence that MWC 922 is a symbiotic binary star in which mass-transfer on to a hot companion is occurring via Roche lobe overflow. Such a process can produce both an accretion disc accompanied by jet production as well as an expanding excretion disc.

The above mass-loss rate in the jet is about 2 mag lower than the rate of |$\skew4\dot{M} \sim 2 \times 10^{-6}$| M_⊙ yr|$^{-1} (V_{\mathrm{ exp}} / 5\, \mathrm{ km}\, \mathrm{ s}^{-1})$| estimated by Sánchez Contreras et al. (2017) from the ionized component within a few hundred astronomical unit of MWC 922. Sculpting of the inner RSN likely requires such a high rate, which may also be responsible for the extended RSN and the ‘hexagon’ structure. The highly collimated jet likely represents the fast, inner core of the much slower wide-angle outflow emerging orthogonal to the orbit plane or the suspected binary that lies in the centre of the Red Square. The wide-angle flow is likely to have a much lower outflow speed and higher mass-loss rate than the jet. Sánchez Contreras et al. (2017) claim a 5 km s⁻¹ expansion speed for the ionized flow at its ∼150 au outer radius.

For a jet velocity of V_jet = 500 km s⁻¹, the time taken to cross a distance, D, is the dynamical time given by t_dyn = D/V_jet. This quantity is tabulated in Table 2 for each jet segment and gap and the suspected terminal bow shock. The time-scales for the ejecta in the disc-plane are more difficult to ascertain since there is no reliable measure of the current or past wind velocity. Using V ∼5 km s⁻¹ for the free-expansion speed for a dense wind implies a dynamical time-scale of ∼10⁵ yr for the ∼0.5 pc extension of the ‘SW disc’, ∼1.5 × 10⁴ yr for the ∼0.09 pc radius ‘hexagon’, and ∼0.8 × 10⁴ yr for the bright, inner RSN.

Assuming full ionization as the jet emerges from MWC 922, the recombination time, t_rec = 1/n_eα_B, and recombination distance, L_rec ∼ V_jett_rec = V_jet/n_eα_B, can be checked against the distance (from MWC 922) of the most distant part of the jet, about ∼0.6 pc (α_B ≈ 2.6 × 10⁻¹³|$\rm cm^3\,s^{-1}$| is the case-B recombination coefficient for hydrogen at 10⁴ K). Taking a mean jet velocity of V_jet  = 500 km s⁻¹ implies an electron density (assuming full ionization near the base of the jet) of n_e ∼ 50 cm⁻³ which implies a recombination time-scale of order t_rec ∼ 2400 yr. At this jet speed, and in the absence of an external UV radiation field, the beam would recombine at a distance L_rec ∼ 1.2 pc. Thus, it is possible that the large gap between the north-west end of jet segment NW2 and the suspected NW shock is primarily due to recombination of the jet beam. However, the H α and [N ii] emission from the disc surface implies the presence of an external ionizing radiation field.

Because of thermal expansion of the jet beam at the Mach angle described above, its density, emission measure, and therefore the H α SB, are expected to decline with increasing distance from the source. The density of a steady, constant velocity jet spreading into a cone with a constant opening angle is expected to decline as d⁻², where d is the distance from the point of origin. The SB in recombination lines such as H α scales as the emission measure, EM. Since the path-length through the emission region, L, will be proportional to d for a constant jet opening angle, the SB is expected to decline as d⁻³. Thus, jet beam spreading is likely to be a better explanation for the disappearance of the MWC 922 jet beyond the segments NW2 and SE2.

4.2 The extended RSN: properties determined from H α emission

The H α SB of the SW disc region ranges up to I(H α) ≈ 6 × 10⁻¹⁶ erg s⁻¹ cm⁻² arcsec⁻², which implies an emission measure of EM ≈ 1300 cm⁻⁶ pc. The peak occurs about 44 arcsec from MWC 922 (∼0.36 pc). If the H α emission traces the externally photoionized surface of an extended excretion disc, the LOS path-length is likely to be comparable to the radial distance from the star. For a foreground extinction of A_V  = 2 mag, n_e ≈ (EM/L)^1/2 = 70 cm⁻³. For A_V  = 0 mag, n_e ≈ 36 cm⁻³.

The gross asymmetry of the north-east–south-west-oriented RSN that points within a few degrees of the nearby H ii region Messier 16 suggests another possibility. Perhaps MWC 922 was dynamically ejected from the cluster NGC 6611 inside Messier 16. The radial velocity of MWC 922 is about 10 km s⁻¹ higher than the mean radial velocity of the molecular gas associated with Messier 16. No proper motion measurements are available in Gaia Data Release 2 (DR2) for MWC 922, likely because the core is a spatially extended source at visual wavelengths.

Assuming that MWC 922 has a ∼10 km s⁻¹ proper motion in the plane of the sky, it would take about 3 Myr to traverse the projected distance of 30 pc from the location of NGC 6611 to the present location of MWC 922. If this star were moving towards the north-east, the bright south-west nebula could be a tail of gas and dust shed by the motion of a mass-losing star through the ISM. In this model, the south-west nebula is likely to be a cylinder with an LOS thickness comparable to its projected width on the plane of the sky. Using an LOS dimension ranging from 3 arcsec to 10 arcsec implies an electron density n_e ∼ 120–220 cm⁻³. In this interpretation, the ‘hexagon’ surrounding the inner RSN could be a bow shock generated as a slow wind from MWC 922 encounters the ISM.

Such a motion transverse to the jet would imply some degree of jet bending. For a proper motion of 10 km s⁻¹ and a jet velocity of 500 km s⁻¹, the jet might be bent back by ISM interactions by an angle of order 1° that cannot be ruled out by the images. The strongest arguments against such a model are that it requires a stronger Lyman continuum ionizing flux to illuminate the RSN than the edge-on disc model, and that there is no direct evidence for a proper motion of MWC 922.

4.3 The extended RSN: mid-IR emission

MWC 922 is a bright and compact source in the mid-IR. In the Herschel Galactic plane survey, Hi-GAL data (Molinari et al. 2010, 2016), a faint tail of 70 μm dust continuum emission extends to the south-west for about 40 arcsec and coincides with the ‘SW disc’ in our images. The peak 70 μm SB is about 168 MJy sr⁻¹; the mean SB in a 29 arcsec by 12 arcsec box is ∼92 MJy sr⁻¹ corresponding to a total flux density of 0.75 Jy after subtraction of a background using the same aperture above and below the SW disc. The aperture only encloses the outer parts of the disc from ∼23 arcsec to ∼52 arcsec from MWC 922 because the intense infrared light from the star swamps the inner disc in the Hi-GAL data. Although it is possible that this feature is an artefact associated with the extremely bright 70 μm emission from MWC 922 itself, we consider this unlikely for two reasons: First, no other similarly bright compact source in the Hi-GAL data shows such a feature. Secondly, the 70 μm feature coincides in size, shape, and orientation with the brightest H α and [N ii] emission from the SW disc. The disc can also be faintly seen at 170 μm against a very confusing and complex background. Photometry, after background subtraction using the same aperture, gives a total flux density of 0.52 Jy. At 250 μm the flux is less than 16 Jy and is dominated by the complex background of the inner Galactic plane.

Fitting the 70–170 μm flux ratio by a single-temperature Planck function implies a dust temperature of 54 K. Dust heated by illumination from MWC 922 at this distance, or by the ambient radiation field, is expected to be colder than 30 K. The high temperature could be an indication that the grain population radiating at 70 μm consists of very small grains (VSGs). However, it is more likely that the grains are heated by electron impact and UV associated with the ionized surface layers of the disc traced by H α and [N ii] emission. Such warm dust with temperatures in the range 50 to over 100 K is commonly seen as diffuse 24 μm emission closely associated with H ii regions. However, in the Spitzer Space Observatory 24 μm images, the region of extended H α emission south-west of MWC 922 is flooded by the large flux from the star.

Dust mass in the measurement box can be estimated from

\begin{eqnarray*} M = {{1.0 \times 10^{-23} X_{\mathrm{ dust}} S_{\nu } D^2} \over {\kappa _{\nu } B_{\nu } (T)}}, \end{eqnarray*}

where S_ν is the background-subtracted flux density in the measurement area in Janskys, X_dust is the gas-to-dust ratio by mass (for the computations, we assume X_dust  = 100), D is the distance in centimeters, κ_ν is the dust opacity at the measurement frequency, and B_ν(T) is the Planck function at dust temperature T. The values of κ_ν were interpolated from the tables in Ossenkopf & Henning (1994) for MRN distribution of ‘naked’ grains [column 1 in table 1 in Ossenkopf & Henning (1994)] to frequencies corresponding to wavelengths of 70, 170, and 250 μm, the effective central wavelengths of the filters on-board the Herschel Space Observatory. The dust opacities used here are 61.25, 14.03, and 8.0 cm² g⁻¹ at 70, 170, and 250 μm, respectively.

The total hydrogen mass (assuming standard ISM values for the gas:dust ratio) in the 348 square arcsec measurement box that appears to be clear of the bright emission from MWC 922 and which coincides with the south-west extension to the RSN is estimated ∼6 × 10⁻⁴ M_⊙ at 70 μm, with at least a factor of about 2 uncertainty. The mass upper limits set by the complex background at 170 and 250 μm are <2 × 10⁻³ and <0.17 M_⊙, respectively. The 70 μm measurement (and possibly the 170 μm) is likely to be a severe lower bound to the mass of the extended SW disc in the RSN since the observed dust emission likely only traces a small, unusually warn sub-population of dust grains. An alternative very crude estimate for the ‘SW disc’ feature can be obtained from the H α emission. Assuming that this emission traces a r_disc  = 0.5 pc outer radius disc with a thickness of h = 0.05 pc, comparable to the thickness of the H α emission, and that this disc has a uniform density throughout, M = πr²hμm_Hn ∼ 0.1 M_⊙ for n = 50 cm⁻³, the lower bound on the electron density derived from H α emission, presumably located at the irradiated disc surface. This estimate is also likely to be a severe lower bound.

4.4 Where does the ionizing radiation come from?

If the north-east–south-west extension of the RSN is indeed a half-parsec radius excretion disc shed by MWC 922, the emission measure of order EM ∼1300 implies a disc-surface electron density of around 100 cm⁻³. The absence of an intensity gradient away from MWC 922 indicates that the ionizing radiation does not come from that star; it must come from the general environment. At a projected distance of ∼30 pc, the NGC 6611 cluster in Messier 16 by itself is unlikely to provide the UV radiation. Even in the absence of intervening dust, the flux fails by about an order-of-magnitude to explain the observed emission measure. It is more likely that OB stars in the extended Serpens OB1 and possibly the Serpens OB2 associations, some of which may be closer to MWC 922, provide the needed ionization. The H α images show that the entire field of view is filled with diffuse H α emission that is probably ionized by the same stars.

The extended H α emission over the entire ARCTIC field of view indicates the presence of ionizing UV in the environment. The long-slit spectra of the red [S ii] doublet was used to estimate the mean electron density using the intensity of the 6717 Å line divided by the intensity of the 6731Å line. The result is that the diffuse background is in the low-density limit of the [S ii] line ratio, implying has n_e < 240 cm⁻³.

Fig. 7 shows a continuum-subtracted H α image taken from SHASSA (Gaustad et al. 2001). This image shows a bridge of emission filling the space between Messier 16 and Sh2-54 with the RSN (MWC 922) located along the eastern rim of this feature. With a mean SB of 5 × 10⁻¹⁶erg s⁻¹ cm⁻² arcsec⁻² and a radius of ∼30 pc, the mean electron density assuming a uniform-density H ii region and A_V  = 0 mag is n_e ∼ 3 cm⁻³. For A_V  = 2 mag, n_e ∼ 6 cm⁻³. The density and radius implies that the Lyman continuum luminosity of ionizing photons is L(LyC) ∼1–4 × 10⁴⁹ photons s⁻¹, which could be supplied by one or two late-O stars, or a dozen early B stars.

Figure 7.

A SHASSA continuum subtracted H α image showing the environment of the RSN. The H ii regions Messier 16 and Sh2-54 are indicated along with the RSN. The SB scales rom of 0 to |$\rm 4 \times 10^{-14}$| (erg s⁻¹ cm⁻² arcsec⁻²).

The low H α SB, large spatial extent, and an electron density around ∼200 cm⁻³ would indicate that most of the emission likely arrises from a relatively thin layer |$L = \mathrm{ EM} / n^2_\mathrm{ e} \sim$|0.01 to 0.1 pc in depth somewhere along the LOS. These layers likely trace ionization fronts at the interface between an extended, extremely low-density H ii region and surrounding denser clouds. Ionization is likely powered by a distributed population of massive stars located outside the bright H ii regions, Messier 16, Sh2-54, and NGC 6604, and these stars may supply the radiation field illuminating MWC 922.

5 CONCLUSIONS

Deep, narrow-band images of the RSN and its source star, MWC 922, reveal a highly collimated and segmented, parsec-scale jet-oriented orthogonal to the previously identified emission-line nebula that can be traced towards the south-west. The structure of this externally irradiated jet is used to estimate its speed, found to be around ∼500 km s⁻¹. The H α emission measure is combined with the width of the jet to estimate its electron density that is found to be about n_e ∼ 50–100 cm⁻³. These parameters are used to estimate the mass-loss rate, momentum injection rate, and energy injection rate of the jet segments. These values are |$\skew4\dot{M} \sim 5 - 10 \times 10^{-8}$| M_⊙ yr⁻¹, |$\skew5\dot{P} \sim 2.5 - 5 \times 10^{-5}$| M_⊙ km s⁻¹ yr⁻¹, and |$\skew5\dot{E} \sim 1 - 2$| L_⊙.

The new images and spectra show that [N ii] emission line at 6584 Å is enhanced in the RSN. The jet segments are most visible in [N ii], and [N ii] is nearly as bright as H α in the ∼0.6 pc-long south-west-oriented ‘tail’ of extended emission that points directly towards Messier 16, located ∼30 pc to the south-west in projection. There is no counterpart to this feature towards the north-east. However, the images do show a faint, hexagon-shaped nebula surrounding the bright inner Red Square.

Two possible models for the south-west-facing nebula are considered. It might be a large excretion disc or stream of ejecta shed by MWC 922 that is preferentially illuminated and ionized form the direction of Messier 16. Because of its orientation, the south-west part shadows the north-east part. Faint, 70 μm emission traces warm dust at the surface. Alternatively, MWC 922 may have been ejected from near Messier 16. In this model, the south-west component of the RSN is a tail of ejecta left behind the star as mass lost from the star interacts with the ISM through which it moves. If the south-west feature is an externally irradiated disc, the H α emission provides a lower bound on its mass of order ∼0.1 M_⊙.

The RSN and its jet appears to be externally ionized. The surrounding ISM is illuminated by diffuse background Lyman continuum. The extended background emission and ionization of the RSN required a Lyman continuum photon luminosity of order L(LyC) ∼1–4 × 10⁴⁹ H-ionizing photons per second at a distance of about 30 pc from the RSN. Such a radiation field could be produced by a distributed population of about a dozen early B stars, or one or two late-O stars at about 30 pc distance from the RSN.

ACKNOWLEDGEMENTS

The work presented here is based on observations obtained with the Apache Point Observatory 3.5-m telescope, which is owned and operated by the Astrophysical Research Consortium. We thank the Apache Point Observatory Observing Specialists for their assistance during the observations. We acknowledge use of the Southern H-Alpha Sky Survey Atlas (SHASSA), which is supported by the National Science Foundation (Gaustad et al. 2001). We thank John (Jack) Faulhaber for useful discussions and comments on the manuscript. We thank the referee for constructive comments that improved the manuscript.