CLUSTERED STAR FORMATION AND OUTFLOWS IN AFGL 2591

We conducted VLBA⁵ observations in the K band to study the 6₁₆ − 5₂₃ H₂O maser emission (rest frequency 22.235079 GHz) toward AFGL 2591. The VLBA observations were scheduled under program BM272H at four epochs: 2008 November 10, 2009 May 6, 13, and November 13. While these observations were optimized to measure the annual parallax and Galactic proper motion of the source (Rygl et al. 2011), in this paper we focus on the internal kinematics of the water maser emission.

We performed phase-referencing observations by fast switching between the maser target and the ICRF calibrator J2007+4029. That allowed us to determine the water masers absolute position within an accuracy of ±1 mas. Two fringe finders (3C345 and 3C454.3) were observed for bandpass, single-band delay, and instrumental phase-offset calibration. Further details about these observations as well as the distance measurement to AFGL 2591 can be found in Rygl et al. (2011). We employed four adjacent intermediate frequencies (IFs) of 8 MHz bandwidth in dual circular polarization, each one split into 256 spectral channels. This receiver setup provided both a large-enough bandwidth to increase the signal-to-noise ratio of the weak continuum calibrators and a 0.42 km s⁻¹ channel width to adequately sample the maser lines. The third IF centered on an LSR velocity (V_LSR) of −5.0 km s⁻¹ covered the previous detections of water maser emission in the region (Trinidad et al. 2003). The data were processed with the VLBA correlation facility in Socorro (New Mexico) using an averaging time of about 0.9 s, which limited the instantaneous field of view of the interferometer to about 2'' (i.e., without significant amplitude losses). Data were reduced with the NRAO Astronomical Image Processing System (AIPS) following the procedure described in Reid et al. (2009).

The natural CLEAN beam was an elliptical Gaussian with an FWHM size of 0.80 mas × 0.40 mas at a position angle (P.A.) of −149 (east of north), with small variations from epoch to epoch. In each observing epoch, with an on-source integration time of about 2 hr, the effective rms noise level on the channel maps was about 0.01 Jy beam⁻¹. The total-power spectrum of the 22.2 GHz masers toward AFGL 2591 is shown in Figure 1.

Zoom In Zoom Out Reset image size

We retrieved VLA A-configuration observations toward AFGL 2591 at 1.3 and 0.7 cm from the VLA⁶ archive for the purposes of a detailed comparison with the high-resolution VLBA maser data. These observations were scheduled on 1999 June 29 (Trinidad et al. 2003) and 2002 March 23 (van der Tak & Menten 2005), respectively. We recalibrated these data with the NRAO AIPS software package using standard procedures and cross-calibration between the strong H₂O maser lines and the continuum at 1.3 cm (e.g., Reid & Menten 1990). For the 1.3 cm data, we cleaned the VLA maps with a circular restoring beam, equal to the natural-beam minor axis for a "ROBUST 0" weighting (Table 1).

We imaged a range of LSR velocities between −40 and +25 km s⁻¹ with a field of view of 14 × 14 about the VLA-3 radio source. These limits include the emission previously reported by Trinidad et al. (2003, their Tables 3 and 4) toward VLA-3. In the following, the term feature refers to spots spatially overlapping in contiguous velocity channels, that we treat as an individual masing cloud (see Sanna et al. 2010a for a detailed discussion).

We identified 80 distinct water maser features distributed within an area of about 077 × 048 (∼2500 AU × 1500 AU). They are mainly grouped in five clusters labeled in Table 2, northwest (NW), middle-west (MW), northeast (NE), middle-east (ME), and southeast (SE), with respect to Figure 2(a). The properties of individual features are presented in Table 2 and show peak brightness ranging from 22 Jy beam⁻¹ to a detection limit of 0.05 Jy beam⁻¹ (5σ). Water maser features redshifted with respect to the V_sys (of −5.7 km s⁻¹) belong to the MW cluster, whereas the blueshifted emission is clustered toward the SE. The full spread in LSR velocities ranges from −0.4 km s⁻¹ for the most redshifted feature (13) to −34 km s⁻¹ for the most blueshifted one (77). The angular distribution and full-space motions of the water masers (i.e., line-of-sight velocities plus proper motions) are plotted in Figure 2(a). With a time baseline of 1 yr, we measured relative proper motions of individual features with an uncertainty less than 3 km s⁻¹. Positions and velocities in Table 2 are relative to the compact feature 4 which was stable in the spectral domain over the 1 yr observations. The magnitude of relative proper motions ranges from 5.0 km s⁻¹ for feature 17 to 56 km s⁻¹ for feature 54. Relying on the symmetry of the maser spatial distribution, we have corrected the proper motions by the average velocity of all features of V_x = 8.8 ± 0.4 and V_y = 9.0 ± 0.4, to approximate the actual motions with respect to the center of expansion.

Zoom In Zoom Out Reset image size

Table 2. Parameters of VLBA 22.2 GHz Water Maser Features

Feature	Detection	VLSR	Fpeak	Δx	Δy	Vx	Vy
	(epochs)	(km s−1)	(Jy beam−1)	(mas)	(mas)	(km s−1)	(km s−1)
Northwest Cluster
1	1	−5.00	14.87	−39.72 ± 0.11	122.69 ± 0.05	⋅⋅⋅	⋅⋅⋅
2	1, 2, 3, 4	−5.42	14.17	−78.51 ± 0.06	125.13 ± 0.05	4.0 ± 2.9	14.2 ± 2.3
3	1	−5.00	0.55	−46.13 ± 0.08	124.11 ± 0.05	⋅⋅⋅	⋅⋅⋅
Middle-west Cluster
4	1, 2, 3, 4	−11.74	20.60	0 ± 0.06	0 ± 0.09	⋅⋅⋅	⋅⋅⋅
5	1, 2, 3, 4	−5.42	13.41	−4.07 ± 0.12	4.16 ± 0.06	4.3 ± 3.2	14.6 ± 2.1
6	1, 2, 3, 4	−6.68	8.29	0.02 ± 0.05	0.47 ± 0.07	6.9 ± 1.9	3.4 ± 2.2
7	1, 2, 3	−7.10	2.98	−12.47 ± 0.07	0.59 ± 0.06	−14.4 ± 3.3	−0.3 ± 4.1
8	1, 2, 3, 4	−4.57	1.60	−3.91 ± 0.07	4.18 ± 0.07	8.5 ± 2.1	15.0 ± 2.2
9	1	−1.20	0.24	−0.37 ± 0.06	1.79 ± 0.08	⋅⋅⋅	⋅⋅⋅
10	4	−2.89	0.24	−16.96 ± 0.08	1.19 ± 0.12	⋅⋅⋅	⋅⋅⋅
11	1	−4.15	0.18	−11.78 ± 0.05	0.10 ± 0.06	⋅⋅⋅	⋅⋅⋅
12	1	−4.58	0.17	−3.04 ± 0.05	3.99 ± 0.06	⋅⋅⋅	⋅⋅⋅
13	4	−0.36	0.14	−14.65 ± 0.07	−0.97 ± 0.07	⋅⋅⋅	⋅⋅⋅
14	4	−6.26	0.09	−16.26 ± 0.07	1.51 ± 0.10	⋅⋅⋅	⋅⋅⋅
15	1	−7.94	0.09	−13.20 ± 0.06	1.38 ± 0.08	⋅⋅⋅	⋅⋅⋅
16	1	−1.20	0.06	−10.40 ± 0.07	−1.00 ± 0.06	⋅⋅⋅	⋅⋅⋅
Northeast Cluster
17	1, 2, 3, 4	−5.00	22.25	651.05 ± 0.06	305.69 ± 0.06	−2.5 ± 1.8	4.4 ± 2.1
18	1, 2, 3, 4	−9.21	6.01	650.18 ± 0.05	317.22 ± 0.06	−7.5 ± 1.7	13.2 ± 2.0
19	1, 2, 3, 4	−12.58	5.44	615.75 ± 0.06	295.33 ± 0.06	−7.5 ± 2.0	4.3 ± 2.5
20	1, 2, 3, 4	−7.53	4.52	618.75 ± 0.10	208.02 ± 0.10	0.3 ± 2.2	−12.5 ± 2.4
21	1, 2, 3, 4	−8.79	2.89	624.70 ± 0.06	312.39 ± 0.09	−13.2 ± 2.1	2.8 ± 2.1
22	1, 2, 3, 4	−8.79	2.29	623.68 ± 0.11	312.34 ± 0.05	−5.5 ± 1.7	5.2 ± 2.0
23	4	−9.21	2.03	645.05 ± 0.06	316.82 ± 0.06	⋅⋅⋅	⋅⋅⋅
24	1, 2, 3	−8.37	2.01	655.31 ± 0.10	317.05 ± 0.07	11.3 ± 4.2	16.0 ± 4.3
25	2, 3	−8.37	1.85	654.88 ± 0.08	317.44 ± 0.06	⋅⋅⋅	⋅⋅⋅
26	1, 2, 3, 4	−7.53	1.60	608.52 ± 0.06	207.16 ± 0.06	−14.3 ± 1.8	−11.3 ± 2.2
27a	1, 2, 3, 4	−8.37	1.35	620.27 ± 0.19	311.32 ± 0.16	⋅⋅⋅	⋅⋅⋅
28	1, 2, 3, 4	−7.11	0.69	625.15 ± 0.05	215.20 ± 0.05	8.5 ± 1.8	−1.7 ± 2.1
29a	1, 2, 3	−6.69	0.53	625.44 ± 0.05	215.62 ± 0.06	⋅⋅⋅	⋅⋅⋅
30	1, 2, 3, 4	−5.00	0.52	648.84 ± 0.08	306.30 ± 0.06	−4.0 ± 2.0	3.1 ± 2.0
31a	1, 2, 3, 4	−9.21	0.47	641.48 ± 0.09	315.73 ± 0.05	⋅⋅⋅	⋅⋅⋅
32	1, 2, 3, 4	−8.27	0.42	631.55 ± 0.05	314.48 ± 0.06	−8.5 ± 1.8	5.4 ± 2.2
33	1, 2, 3, 4	−9.63	0.40	625.42 ± 0.09	312.92 ± 0.10	−9.7 ± 2.1	6.4 ± 2.2
34	1, 2, 3	−8.37	0.39	653.85 ± 0.06	317.29 ± 0.24	14.1 ± 3.3	14.0 ± 8.7
35	1, 2, 3, 4	−7.53	0.39	594.08 ± 0.05	223.24 ± 0.13	−14.4 ± 1.8	8.5 ± 2.2
36	2, 3, 4	−8.37	0.31	657.75 ± 0.06	309.98 ± 0.06	24.1 ± 3.0	5.7 ± 3.2
37	1, 2, 3	−6.26	0.24	658.35 ± 0.07	312.14 ± 0.09	18.0 ± 3.3	−11.0 ± 4.7
38	1, 2, 3, 4	−6.69	0.23	688.86 ± 0.08	308.38 ± 0.14	1.4 ± 1.8	9.0 ± 2.2
39a	1, 2, 3	−6.69	0.22	685.88 ± 0.05	311.26 ± 0.05	⋅⋅⋅	⋅⋅⋅
40	1, 2, 3, 4	−6.69	0.19	626.85 ± 0.05	221.95 ± 0.06	5.8 ± 1.8	−0.2 ± 2.1
41	2, 3, 4	−7.53	0.19	620.88 ± 0.25	311.72 ± 0.08	−6.1 ± 3.7	0.5 ± 3.0
42	1	−7.95	0.16	614.48 ± 0.05	206.15 ± 0.05	⋅⋅⋅	⋅⋅⋅
43	4	−7.95	0.13	592.54 ± 0.08	219.73 ± 0.14	⋅⋅⋅	⋅⋅⋅
44	4	−7.53	0.10	604.34 ± 0.08	206.63 ± 0.07	⋅⋅⋅	⋅⋅⋅
Middle-east Cluster
45	1, 2, 3, 4	−4.57	3.14	567.48 ± 0.06	80.27 ± 0.10	9.7 ± 2.2	−16.9 ± 3.5
46	1, 2, 3, 4	−5.00	0.58	568.34 ± 0.06	82.61 ± 0.11	15.5 ± 1.9	−3.8 ± 2.7
47	1, 2, 3	−5.84	0.28	565.23 ± 0.06	76.66 ± 0.06	3.0 ± 3.3	−10.8 ± 4.2
Southeast Cluster
48	2, 3, 4	−24.38	4.83	626.73 ± 0.05	−153.13 ± 0.05	−7.2 ± 3.2	−31.9 ± 3.9
49	2, 3	−27.75	4.37	621.15 ± 0.07	−164.35 ± 0.06	⋅⋅⋅	⋅⋅⋅
50	1	−16.38	4.32	623.54 ± 0.12	−136.49 ± 0.07	⋅⋅⋅	⋅⋅⋅
51	2, 3, 4	−19.32	2.05	624.58 ± 0.08	−137.92 ± 0.11	−7.7 ± 3.4	−30.7 ± 3.8
52	2, 3, 4	−19.75	1.71	620.92 ± 0.15	−134.25 ± 0.11	−1.9 ± 4.4	36.4 ± 3.8
53	2, 3	−18.90	1.59	621.43 ± 0.07	−134.56 ± 0.06	⋅⋅⋅	⋅⋅⋅
54	2, 3, 4	−25.65	1.52	626.89 ± 0.06	−152.60 ± 0.09	−15.7 ± 3.1	−54.0 ± 3.3
55	2, 3	−17.22	1.23	623.17 ± 0.06	−135.73 ± 0.06	⋅⋅⋅	⋅⋅⋅
56a	2, 3, 4	−18.06	0.97	626.23 ± 0.08	−143.15 ± 0.09	⋅⋅⋅	⋅⋅⋅
57	2, 3, 4	−20.59	0.94	624.09 ± 0.10	−137.34 ± 0.10	−10.5 ± 3.9	−34.2 ± 4.4
58	1, 2, 3, 4	−14.27	0.82	612.29 ± 0.23	−115.46 ± 0.10	−6.8 ± 3.6	10.1 ± 2.9
59	2, 3, 4	−29.44	0.79	624.39 ± 0.06	−159.05 ± 0.07	−5.5 ± 3.3	14.0 ± 3.6
60	1	−21.01	0.70	631.75 ± 0.14	−144.95 ± 0.05	⋅⋅⋅	⋅⋅⋅
61	1, 2, 3	−30.70	0.61	625.12 ± 0.06	−153.96 ± 0.07	20.7 ± 3.6	−38.4 ± 4.4
62	1	−20.17	0.50	623.75 ± 0.07	−137.19 ± 0.08	⋅⋅⋅	⋅⋅⋅
63	2, 3, 4	−21.01	0.45	631.22 ± 0.13	−145.12 ± 0.05	24.0 ± 5.2	1.6 ± 3.4
64	2, 3, 4	−23.96	0.41	623.51 ± 0.06	−136.87 ± 0.07	0.4 ± 3.2	−25.7 ± 3.3
65	1	−29.44	0.38	625.51 ± 0.06	−153.29 ± 0.06	⋅⋅⋅	⋅⋅⋅
66	2, 3	−15.53	0.36	626.12 ± 0.05	−141.52 ± 0.10	⋅⋅⋅	⋅⋅⋅
67	1	−22.28	0.31	631.30 ± 0.07	−144.40 ± 0.06	⋅⋅⋅	⋅⋅⋅
68	1, 2, 3	−14.69	0.24	611.01 ± 0.13	−114.73 ± 0.12	−6.9 ± 4.9	13.0 ± 5.4
69	2, 3, 4	−21.85	0.21	630.94 ± 0.05	−144.92 ± 0.05	12.7 ± 4.0	−2.7 ± 3.5
70	1	−22.70	0.20	631.24 ± 0.05	−143.37 ± 0.06	⋅⋅⋅	⋅⋅⋅
71	1	−15.53	0.18	626.92 ± 0.06	−143.72 ± 0.07	⋅⋅⋅	⋅⋅⋅
72	2, 3, 4	−22.27	0.18	630.73 ± 0.08	−144.46 ± 0.10	11.2 ± 3.8	−3.2 ± 4.3
73	4	−21.01	0.16	623.82 ± 0.08	−136.98 ± 0.09	⋅⋅⋅	⋅⋅⋅
74	4	−17.22	0.14	624.97 ± 0.13	−140.05 ± 0.09	⋅⋅⋅	⋅⋅⋅
75	4	−23.96	0.10	626.35 ± 0.06	−151.82 ± 0.07	⋅⋅⋅	⋅⋅⋅
76	4	−25.22	0.09	626.43 ± 0.07	−152.47 ± 0.10	⋅⋅⋅	⋅⋅⋅
77	2, 3	−33.65	0.09	624.83 ± 0.06	−155.15 ± 0.07	⋅⋅⋅	⋅⋅⋅
78	1	−13.01	0.07	616.11 ± 0.07	−118.00 ± 0.09	⋅⋅⋅	⋅⋅⋅
79	4	−22.28	0.06	623.72 ± 0.06	−141.92 ± 0.10	⋅⋅⋅	⋅⋅⋅
80	1	−26.91	0.05	625.20 ± 0.20	−159.19 ± 0.65	⋅⋅⋅	⋅⋅⋅
Reference Feature 4: absolute position on 2009 May 6
		R.A. (J2000)		Decl. (J2000)
		$20^{{\rm h}}29^{{\rm m}}24\mbox{$.\!\!^{\mathrm s}$}8228$		40°11'19593

Notes. For each identified feature belonging to a given cluster of maser emission, the label (given in Column 1) increases with decreasing brightness and the different epochs of detection are reported in Column 2. Columns 3 and 4 report the LSR velocity and brightness of the brightest spot of each feature, observed at the epoch underlined in Column 2. Columns 5 and 6 give the position offset at the first epoch of detection relative to the feature 4 (centered on the phase-reference maser channel 145) in the east and north directions, respectively. The uncertainties give the intensity-weighted standard deviation of the spots distribution within a feature and was combined in quadrature with a conservative 50 μas uncertainty. This 50 μas plateau accounts for deviations from a Gaussian fit assumption. The absolute position of the reference maser feature is reported at the bottom of the table and is accurate to within ±1 mas. Columns 7 and 8 report the projected components of the feature proper motion relative to the feature 4, along the east and north directions, respectively. A proper motion of 1 mas yr⁻¹ corresponds to 15.6 km s⁻¹ at a distance of 3.3 kpc. ^aFeatures with too uncertain proper motions (>70%)

Download table as:  ASCIITypeset images: 1 2

In Figures 2(b), (c), and (f), we zoom in on the three water maser clusters labeled NE, MW, and SE, showing a symmetric expansion away from their centroids. The NE and SE clusters are particularly interesting due to their full kinematics and are discussed in detail in Sections 4.1 and 4.2, respectively.

In Figures 2(d) and (e), we overlay the radio continuum emission described in Section 3.2 with the water maser emission from the second-epoch observations. This maser emission was summed over the full velocity range of the SE cluster (from V_LSR of −34 to −14 km s⁻¹). Its projected shape along the plane of the sky delineates a straight "V" (hereafter the H₂O V-shape) with a full opening angle of 110° ± 10° and a P.A. of 256°. The variation of LSR velocities along this V-shape is quite regular, ranging from a maximum of −13 km s⁻¹ (feature 78) at the northern side of the V-shape to a minimum of about −33 km s⁻¹ (feature 77) toward the southern side.

Our measurements of the 1.3 and 0.7 cm continuum emissions are summarized in Table 1 and plotted in Figures 2(d) and (e). Published VLA C-configuration observations at 0.7 cm show that significant emission is resolved out with the VLA A-configuration, missing more than half of the flux measured in the compact configuration (van der Tak & Menten 2005). The relative position of the 1.3 cm continuum with respect to the reference maser feature in Table 2 is obtained by matching the overall water maser distribution between the VLBA and VLA-A maps. We note that the north-south spread of the maser emission overlaid on the radio continuum (∼50 mas) have remained almost stationary over about 10 years (cf. Trinidad et al. 2003).

The peak position of the 0.7 cm continuum is assumed to be aligned with the 1.3 cm peak. The three-lobed geometry of the 7 mm emission follows the 1.3 cm elongation toward the east and it branches off toward the west matching the H₂O V-shape (Table 1). While this matching is particularly striking, we note that more sensitive, Q-band, observations are mandatory to confirm the spatial structure of the 0.7 cm emission. Because of the stability of the water maser distribution, but also the similarities between the morphology of the water maser emission and the 0.7 cm continuum (Figure 2(e)), we estimate a relative position accuracy between the radio continuum and water masers of ±001.

Previous observations spanning a broadband spectrum between 6 cm and 1.3 mm show that the spectral index (S∝ν^γ) for wavelengths longer than 7 mm is dominated by free–free emission (γ ∼ 0.9), whereas at shorter wavelengths the flux density is mostly due to dust emission (γ > 2; van der Tak & Menten 2005 and reference therein).

In this section, we discuss two basic characteristics observed in the NE water maser cluster (Figure 2(b)). First, we detected two arc-like filaments of water maser emission elongated in the NE–SW direction with no emission in between them. Second, these arc-like filaments expand from each other with proper motions increasingly aligned with the NE–SW direction as one approaches this direction. Bipolar distributions of water masers with expanding proper motions have been reported in a large number of star-forming regions (e.g., Gwinn et al. 1992; Sanna et al. 2010b; Moscadelli et al. 2011). These detections were interpreted as footprints of the interaction between collimated jets and/or wide-angle winds from YSOs with the surrounding molecular envelope. Following this evidence, we want to show that the overall kinematics of the NE water maser cluster can be explained by assuming that a single object, inbetween the maser emission, emits a high-velocity, bipolar jet along the NE–SW direction (Figures 2(b) and 3(a)). At first, we will compare the velocity profile of the water masers along the arc-like structures with the model of jet-driven outflows presented in Ostriker et al. (2001) and Lee et al. (2001). The jet model predicts that the interaction between a bow shock and the ambient material gives an upper and lower limit to the velocity of the processed material. This processed material is a mixture of material that was already part of the bow shock and material incorporated from the ambient medium as the shock front expands. The shock front delineates an arc-like structure (as do the masers) and the velocity component along the jet axis decreases roughly as r^−2/3, where r is the offset distance from the head of the jet (e.g., Figure 2 in Ostriker et al. 2001).

Zoom In Zoom Out Reset image size

The similarities of the spatial elongation and velocities of the water masers would be naturally explained if the NE–SW direction is associated with an outflow in this direction. In Figure 3(a), we define an elliptical symmetry that matches to first order both the width and the NE–SW elongation of the water maser distribution with a simple central symmetry (major axis and eccentricity on top of Figure 3(a)). This elliptical pattern allows us to define a new reference frame with origin at the center of the ellipse (Δx = 625 mas and Δy = 260 mas with respect to the reference feature in Table 2) and x, y axes along the minor (R) and major (z) axes of the ellipse, respectively. This reference frame is tilted by 22° to the east of the north direction. Due to the small dispersion of LSR velocities (1.5 km s⁻¹) centered at about −7.9 km s⁻¹, we also assume that all the maser features lie along the same plane. We want to study the velocity profile of the shock fronts traced by the water maser proper motions with respect to the new z–R reference frame. In the following, we use the same terminology of Ostriker et al. (2001) and Lee et al. (2001) in order to facilitate the comparison with their jet model.

In Figure 3(b), we report the behavior of the velocity components (V_z) of the water masers along the major axis of the ellipse (z). The idea that the z-direction is a preferred outflow direction is supported by the velocity profile in Figure 3(b), which resembles that expected for a ballistic bow-shock produced by a jet along the z-direction (cf. Figure 4 of Lee et al. 2001). For such a bow-shock model velocities are constrained in Figure 3(b) by two solid lines and the expected gas velocity should lie between them. These limits are derived from Equations (22), (18), and (20) of Ostriker et al. (2001) and define the bow shock profile z = f(R), the mean velocity of the mixing of jet and ambient material v'_z (solid line) in a shell, and the velocity of newly shocked ambient material u'_z (dashed steeper line), respectively. These equations are parameterized by the jet radius R_j, the velocity of the bow shock v_s, and the ejection velocity βc_s of the material processed by the working surface, expressed as a multiple of the isothermal sound speed (Ostriker et al. 2001). For the radius of the high-velocity "core" of the jet (R_j), we set the minimum projected distance to the R-axis measured with the masers of 1.8 mas (≈6 AU). This value implies a width-to-length ratio of the jet from the center of the ellipse of 0.05, consistent with that found in the literature for H₂ jets at typical scales of 10³–10⁴ AU (e.g., HH 212 in Zinnecker et al. 1998). We stress that R_j is the only constraint we put on the model, whereas we consider the best value of the ratio βc_s/v_s = 0.5 and β = 4.1 inferred from the simulations by Lee et al. (2001). By setting the tip of the bow shock at the maximum maser projection along z (the "0" in Ostriker et al. 2001), we find a good agreement of the measured maser velocities (black squares in Figure 3(b)) with the jet-driven outflow model. We note that the maser velocities are closer to the ambient material velocity (dashed line) in agreement with simulations by Lee et al. (2001, their Figure 4). This behavior may reflect an incomplete mixing of material already in the bow shock shell and material added from the ambient medium. The βc_s/v_s ratio implies a bow shock speed of about 66 km s⁻¹ for a sound speed of 8 km s⁻¹ at the cooling cutoff of 10⁴ K (see Ostriker et al. 2001). On the contrary, in the alternative scenario of a wide-angle wind the V_z motions would increase linearly with z, that is not observed (e.g., Lee et al. 2001 their Figure 9). This suggests that the maser cloudlets are dragged along the z-direction by a jet-driven shock component.

On the other hand, the jet-driven outflow model predicts an upper limit to the transverse velocity component V_R with respect to the jet direction of the order of the sound speed (∼8 km s⁻¹; see Figure 3 in Ostriker et al. 2001 and Figure 4 in Lee et al. 2001). This limit is intimately related to the assumption that the transverse radial forces are dominated by thermal pressure gradients that act at the head of the jet. In Figure 3(c), we measure the greatest deviations from the jet model far from the head of the jet, at an inclination of about 20° from the jet axis. This behavior may suggest that, away from the head of the jet, thermal pressure gradients no longer dominate the expansion. One possibility would be that a low-velocity (<10 km s⁻¹) wind with a wide opening angle (ca. 20°) acts with the jet simultaneously. In the wind-driven outflow model, the expansion of the outflow away from the central axis is driven by the ram pressure of a wide-angle wind (Shu et al. 1991): the V_R velocity is maximum far from the tip of the outflow shell and goes to zero at its tip (Figure 9 in Lee et al. 2001). The simultaneous presence of the two types of outflow mechanisms would agree with the hypothesis that any outflow contains a wide-angle wind at some level, in order to explain typical width-to-length ratio of about 1: 10, much higher than in a "pure jet model" (e.g., Arce et al. 2007).

Finally, if the single jet-driven model is correct, we can also get an estimate of the dynamical age of the jet-event exciting the water masers. The velocity of the jet (v_j) is related to the bow shock speed (v_s) by the ratio of the jet-to-ambient density, where v_s approximates v_j if the jet density exceeds the ambient density (see Ostriker et al. 2001). By considering the lower limit v_j ≈ v_s, the inferred value of v_s = 66 km s⁻¹, and the value of the higher maser projection along z, we obtain a dynamical age of t = z_maj/v_j⋍14 yr. On the one hand, this dynamical age is consistent with other estimates based on the expanding motions of masers around forming YSOs (e.g., Torrelles et al. 2003; Moscadelli et al. 2007). On the other hand, given how unlikely it would be to observe such a short-lived event, it strongly suggests that we are tracing a recurrent phenomenon of ejection of matter from an YSO. That would favor a pulsed-jet paradigm against a steady-jet one as supported, for instance, by the recurrent knots and bow shocks observed in the prototypical Herbig–Haro object 212 (Zinnecker et al. 1998). This interpretation naturally explains also the eastern water masers in Figure 3(a) (features 38, 39 in Table 2) and the masers at about 150 mas to the south of the NE cluster (the ME cluster in Table 2). These masers would belong to an older jet emission from the same object.

In this section, we discuss three characteristics observed in the SE water maser cluster: (1) the projected V-shape of the water maser distribution on the plane of the sky, "pointing" toward the peak of the radio continuum emission (Figures 2(d) and (e)); (2) the water maser proper motions that mainly expand along the V-shape (Figure 2(a)); (3) the line-of-sight velocity gradient observed through the V-shape, with velocities increasing to the south (Figure 2(f)). In the following, we relate these properties with the large-scale outflow and the nature of the radio continuum emission and suggest that the H₂O masers trace the edges of the outflow cavity associated with a massive YSO (see Figure 4(a)).

Zoom In Zoom Out Reset image size

The H₂O emission from the SE cluster is blueshifted with respect to the overall velocity of the cloud (−5.7 km s⁻¹) and covers a range of line-of-sight velocities consistent with the small-scale, blueshifted wing of the CO(3–2) outflow (between −10 and −40 km s⁻¹; Hasegawa & Mitchell 1995). The V-shape observed in the water maser emission opens toward the west, where a limb-brightened cavity was previously outlined as NIR loops (Tamura et al. 1991; Minchin et al. 1991; Preibisch et al. 2003). These loops were explained as wind-bubbles, which were created as the blueshifted (western) side of the outflow pushed against the environment and swept up ambient material (Preibisch et al. 2003). We note that the wide opening angle of the NIR loops (>100°) is consistent with that of the V-shape (110°). Actually, if the loops outlined the outflow cavity, one might expect the water masers along the outflow walls, where shocks are more likely. The H₂O V-shape opens toward a P.A. = 256°in good agreement with the direction of the western side of the outflow, which has been detected in a number of tracers: (1) the NIR bright loops, with a 262° ⩽ P.A. ⩽ 256°, surrounding the outflow cavity with a symmetry axis at a P.A. of 259° (e.g., Preibisch et al. 2003, see their Figure 3); (2) the blueshifted CO lobe with a P.A. of about 260° (e.g., Hasegawa & Mitchell 1995, see their Figure 4); (3) the western, aligned, Herbig–Haro objects with a 258° ⩽ P.A. ⩽ 261° (Poetzel et al. 1992) and the hot H₂ knots in the same direction (Tamura & Yamashita 1992); (4) the elongated 1.3 cm continuum with a P.A. of 268° and a spectral index consistent with an ionized wind (Figure 2(d)). If water masers would belong to the blueshifted outflow lobe, and might possibly amplify the continuum background at 1.3 cm, it would naturally explain the non-detection of redshifted maser emission, that would belong to the receding outflow lobe. Also, since the outflow axis is close to the line-of-sight (within 30°–45°; Minchin et al. 1991; Hasegawa & Mitchell 1995; van der Tak et al. 1999), this scenario agrees with the direction of the maser proper motions, that are mainly expanding parallel to the V-shape with a strong component toward the observer (Figure 4). The velocity gradient observed across the maser V-shape (Figure 2(f)) might be reproduced with a slow precession of the outflow axis about the mean direction Z_out in Figure 4 (as speculated at first by Trinidad et al. 2003).

The strong water masers and the radio continuum emission from VLA-3 (Figures 2(d) and (e)) suggest an association with a massive YSO (van der Tak & Menten 2005). If the free–free emission longward of 7 mm wavelength originates from an ionized wind, we can estimate a lower limit to the mass-loss rate ($\dot{M}_w$) of the wind and put a constraint to the nature of the driving source. We apply Equation (4) of Rodriguez & Canto (1983) and use the flux at 1.3 cm to estimate the mass-loss rate in the assumption of a spherical, isothermal, completely photoionized, uniform wind:

$$\begin{eqnarray} \biggr (\frac{ \dot{M}_w}{10^{-5}\, M_{\odot }\,\rm yr^{-1}}\biggr) & \approx & 1.7 \biggr (\frac{S_{\nu }}{\rm 1.6\, mJy}\biggr)^{3/4} \biggr (\frac{\nu }{\rm 22.2\, GHz}\biggr)^{-0.45} \nonumber \\ & & \cdot\,\, \biggr (\frac{d}{\rm 3.3\, kpc}\biggr)^{3/2}\biggr (\frac{v_w}{\rm 10^3\, km\, s^{-1}}\biggr), \end{eqnarray} \tag{ 1 }$$

where v_w is the velocity of the wind. We have considered the accurate value of the recently measured distance to AFGL 2591 and a wind velocity of several 10² km s⁻¹, reported in both the $\rm [S\,\mathsc{ii}]$ lines by Poetzel et al. (1992) and the IR ¹²CO broad wings by van der Tak et al. (1999). The assumption of a spherical wind is justified by the wide opening angle observed both with the maser emission and the NIR loops (>100°). Note that Trinidad et al. (2003) used the lower flux at 3.6 cm to estimate the mass-loss rate, based on their model with a dusty disk that would have affected the flux at 1.3 cm. Still, at the new measured distance of 3.3 kpc, their modeled flux would decrease by a factor 10 and no longer be applicable, while the spectral index for wavelengths longer than 0.7 cm suggests an origin in a ionized wind (van der Tak & Menten 2005). The mass-loss rate from Equation (1) implies a photoionizing ZAMS star of type O9, with a rate of ionizing photons (N_L) of about 1.2 × 10⁴⁸ s⁻¹ (Rodriguez & Canto 1983). We compare this value with the estimates of mass and Lyman continuum flux derived by assuming a single star dominating the IR luminosity of the region. Following van der Tak & Menten (2005), we used the theoretical HR diagram by Maeder & Meynet (1989) to estimate the mass and effective temperature of a star with an IR luminosity of 2 × 10⁵ L_☉; then, we used the stellar atmosphere models by Schaerer & de Koter (1997) to convert effective temperatures to Lyman fluxes. These new values of M and N_L are about 38 M_☉ and 2 × 10⁴⁹ s⁻¹, respectively. The estimates of Lyman flux obtained independently from the radio continuum emission and the IR luminosity of the region agree within an order of magnitude. Since the former one is a lower limit derived close to the source whereas the later one gives an upper limit to the number of Lyman photons emitted over a larger field of view, this calculation provides evidence that a single massive object dominates the IR energetics of the region (according to the high-resolution NIR image of Preibisch et al. 2003).

We finally collect the information from the water masers and the radio continuum emission about the ongoing star formation in the region. Taking into account a possible bias of detecting only a small number of (bright) sources, there is evidence of at least two clustering scales for star formation in the region. One on a linear scale of about 0.1 pc on the sky, from continuum sources VLA-1, VLA-2, and VLA-3 detected at first by Campbell (1984, named by Trinidad et al. 2003). The other scale is an order of magnitude smaller and comes from our observation of at least three centers of expansion that pinpoint two other centers of star formation (the NE and MW cluster), separated by about 2000 AU on the sky from the VLA-3 radio source. Note that the NW maser cluster might be either associated with the MW cluster or trace a further young object. Our analysis in Section 4.2 shows that VLA-3 is the most massive (≲ 38 M_☉) and young object in the field dominating the IR luminosity. Several lines of arguments from the spatial scale of the maser distribution also strengthen the association between VLA-3 and the large-scale dynamics of the molecular envelope, as traced by both jet "footprints" (HH objects and H₂ knots) and the wide-angle outflow seen at the sub-parsec scale (∼0.5 pc from VLA-3). The other two objects indicated by water maser emission in Figure 2 (the NE and MW cluster) were not detected at cm and mm wavelengths. We give an upper limit to their flux density at 1.3 cm of about 0.5 mJy (3σ). On the one hand, by assuming this radio continuum would come from a photoionized stellar wind, we can put an upper limit on the spectral type of the exciting YSOs. Following Equation (1) of Rodriguez & Canto (1983) which gives the rate of ionizing photons required to fully ionize a constant velocity wind, we get a value of N_L ≲ 10⁴⁷ s⁻¹ that would imply a ZAMS star with a spectral type later than B0. On the other hand, the high, isotropic, maser-luminosity of the NE and MW objects (2.5 × 10⁻⁵ L_☉ and 3.3 × 10⁻⁵ L_☉, respectively) would suggest they are not even low-mass YSOs but possibly late B-type stars, by comparison with the SE cluster luminosity (1.2 × 10⁻⁵ L_☉) and the H₂O luminosity usually associated with low-mass stars (<10⁻⁷ L_☉; e.g., Furuya et al. 2003 and their erratum).⁷ In the formulation of Reid & Moran (1988), the maser luminosity is a consequence of the dissipation in the ambient cloud of the mechanical energy of the outflow from a young star, and thus could be related to the mass of the driving source.