The Ionized Warped Disk and Disk Wind of the Massive Protostar Monoceros R2-IRS2 Seen with ALMA

The formation processes of massive stars (with masses ≥8 M_⊙) are under debate. Massive stars could form either by the monolithic collapse of a turbulent molecular core (McKee & Tan 2002), by the competitive accretion of core materials (Bonnell & Bate 2006), or by Bondi–Hoyle accretion onto the central star (Keto 2007). In all these theories, gas accretion occurs through a circumstellar disk coupled with an expanding wind/outflow (which removes angular momentum), in a scaled-up version of low-mass star formation. High angular-resolution interferometry has unveiled several cases of circumstellar disks around massive protostars such as Cepheus A HW2 (Patel et al. 2005; Jiménez-Serra et al. 2007, 2009), IRAS 20126+4104 (Cesaroni et al. 2014; Chen et al. 2016), G17.64+0.16 (Maud et al. 2019), or GGD 27-MM1 (Añez-López et al. 2020). However, the level of detail in these observations only goes down to 40 au (Maud et al. 2019; Añez-López et al. 2020), insufficient to probe the innermost structure of these disks. Hints of holes have recently been reported (Maud et al. 2019), but it remains unknown whether warps exist within these disks as found for their low-mass counterparts (Sakai et al. 2019; Kraus et al. 2020; Sai et al. 2020). The launching process of winds/outflows has not been witnessed either during the protostellar phase of massive stars.⁴

Hydrogen radio recombination lines (RRLs) probe the kinematics of the ionized gas in ultracompact (UC) H ii regions during the process of massive star formation (Churchwell 2002). About 30% of these regions show broad RRLs with line widths ≥60–80 km s⁻¹, which exceed those produced by just thermal broadening (FWHM ∼ 30 km s⁻¹; Churchwell 2002). These regions show elongated morphologies and thus broad RRL emission likely arises from ionized winds (Jaffe & Martín-Pintado 1999).

Non-LTE effects such as maser amplification are expected in RRLs. These effects are pronounced in transitions at submillimeter wavelengths when electron densities reach values n_e ∼ 10⁶–10⁸ cm⁻³ (Figure 5 in Strelnitski et al. 1996). However, RRL masers are rare. Only a few objects have been detected so far, such as MWC349A (Martín-Pintado et al. 1989a; Báez-Rubio et al. 2013), eta Carinae (Cox et al. 1995), or the evolved B[e]-type massive star MWC922 (Sánchez Contreras et al. 2019). Thanks to their brightness, the kinematics of the innermost ionized regions toward these evolved objects can be measured with accuracies down to a few astronomical units (Weintroub et al. 2008; Martín-Pintado et al. 2011; Báez-Rubio et al. 2013; Zhang et al. 2017). This has revealed that they not only present circumstellar disks rotating following a Keplerian law, but also that ionized winds are launched from their disks at radii ≲25–30 au (Martín-Pintado et al. 2011; Sánchez Contreras et al. 2019). Unfortunately, similar studies do not exist for massive protostellar objects.

Monoceros R2 (hereafter MonR2) is a massive star-forming cluster located at a distance of 893 pc (Dzib et al. 2016). It hosts a blister-type H ii region and a cluster of IR sources (Massi et al. 1985; Wood & Churchwell 1989; Carpenter et al. 1997; Treviño-Morales et al. 2019). Among them, IRS2 is a compact and massive young stellar object with a luminosity ∼0.5–1 × 10⁴ L_⊙ (Howard et al. 1994; Álvarez et al. 2004). High angular-resolution observations carried out with the Submillimeter Array (SMA) revealed that MonR2-IRS2 is a UC H ii region with its RRLs at λ ≤ 0.85 mm experiencing maser amplification. However, these observations were unable to resolve the internal structure of the source (Jiménez-Serra et al. 2013).

We present high angular-resolution observations (80 mas × 60 mas, 70 au × 50 au) of the H21α RRL carried out with the Atacama Large Millimeter/submillimeter Array (ALMA) toward the MonR2-IRS2 UC H ii region. Our H21α images show that the ionized gas in MonR2-IRS2 is distributed in a warped Keplerian disk and an ionized wind that is launched at radii ∼11 au.

The H21α RRL at 662.40416 GHz was observed on 2015 September 25 with ALMA in Band 9 (project 2012.1.00522.S) using the 12 m array (baselines from 43 m to 2.27 km). Observations were performed in dual polarization mode using a spectral bandwidth of 1.875 GHz and a spectral resolution of 976 kHz (0.44 km s⁻¹). As calibrators, J0522−3627 was used for bandpass calibration, J0423−013 for flux calibration, and J0607−0834 for phase and amplitude calibration.

The calibrated data set was obtained by running the original pipeline reduction scripts, and the data were additionally self-calibrated using the CASA (the Common Astronomy Software Applications) package version 5.0.0. The MonR2-IRS2 spectra are clean from molecular emission, thus the 0.4 mm continuum map was generated using the CASA task uvcontsub across 5.61 GHz of line-free bandwidth. The angular resolution of the 0.4 mm continuum image was 60 mas × 50 mas, P.A. = 78° and the rms noise level was 1.9 mJy beam⁻¹. The H21α image was obtained from the continuum-subtracted data cube after smoothing the velocity resolution to 1 km s⁻¹. The beam and rms noise level of the H21α image were 80 mas × 60 mas, P.A. = −88°, and 30 mJy beam⁻¹ in 1 km s⁻¹-channels, respectively.

3.1. 0.4 mm Continuum Emission

Figure 1 (left panel) reports the 0.4 mm continuum map obtained toward MonR2-IRS2 (black contours and grayscale). The source is barely resolved at the resolution of ALMA (deconvolved size 62 mas × 43 mas, or 55 au × 38 au). Its morphology clearly deviates from Gaussianity at the low-emission level (≤12σ, with 1σ = 1.9 mJy beam⁻¹), likely due to the presence of a wind (Section 4). By fitting a 2D Gaussian to the continuum emission, the peak coordinates of MonR2-IRS2 are α(J2000) = 06^h07^m458034 (±00008), δ(J2000) = −06°22'535155 (±00005), which are consistent with the previously measured values (Jiménez-Serra et al. 2013).

Zoom In Zoom Out Reset image size

The integrated continuum flux at 0.4 mm is (345 ± 10) mJy. This value is larger than that expected from the spectral index of α = −0.16 ± 0.2 measured by the SMA (with S ∝ ν^α; Jiménez-Serra et al. 2013).⁵ The free–free contribution is 130 ± 30 mJy; thus while the continuum emission at λ ≥ 0.85 mm is entirely due to ionized gas, the λ = 0.4 mm continuum flux is dominated by dust. The mass for the neutral disk can then be estimated using a dust flux of 215 ± 32 mJy and by assuming dust temperatures T_dust = 500–1200 K (Howard et al. 1994), a gas-to-dust mass ratio R = 100, and a dust opacity κ_dust = 25.8 cm² g⁻¹ (consistent with H₂ gas densities of 10⁸ cm⁻³ and no ices; Ossenkopf & Henning 1994). The neutral disk mass is ∼2–4 × 10⁻⁴ M_⊙, with a 15% uncertainty.

3.2. H21α Maser Emission

In Figure 2, we present the H21α spectrum integrated over the MonR2-IRS2 continuum source (Figure 1). Consistent with the H30α and H26α lines measured with the SMA (Jiménez-Serra et al. 2013), the H21α transition shows two emission peaks shifted by ±22 km s⁻¹ with respect to MonR2-IRS2's radial velocity at 10 km s⁻¹ (Torrelles et al. 1983). The intensity of the two peaks is about the same (∼2.8 Jy) and a factor of 10 higher than that of the H30α RRL (∼0.28 Jy beam⁻¹; Table 2 in Jiménez-Serra et al. 2013).⁶ This intensity increase is expected since the line-to-continuum flux ratio (LTCR)⁷ of RRLs under LTE conditions and optically thin emission is known to increase with frequency as ν^1.1. The LTCR is predicted to be ∼206 km s⁻¹ for the H21α line. However, the measured one is ∼1130 km s⁻¹, which clearly exceeds the LTCR predicted under LTE. This, together with our RRL radiative transfer calculations of Section 4, confirm that the RRLs at λ ≤ 0.85 mm toward MonR2-IRS2 are masers.

Zoom In Zoom Out Reset image size

Figure 1 reports the integrated intensity maps for the H21α blueshifted and redshifted peaks from −20 to −5 km s⁻¹ and from 25 to 40 km s⁻¹, respectively (middle panel), and for the high-velocity blue and red wings (from −35 to −29 km s⁻¹ and from 48 to 50 km s⁻¹ respectively; right panel). The H21α maps of the emission peaks follow the morphology of the continuum (left panel), but their maxima are displaced by ∼002 (∼18 au) from the continuum peak. While the redshifted emission peaks toward the northwest of MonR2-IRS2, the blueshifted part peaks toward the southeast. As shown in Section 4, this emission is well reproduced by a Keplerian ionized disk. For the high-velocity wings, the H21α line follows a similar kinematic trend (redshifted gas in the northwest, blueshifted emission in the southeast), but its morphology appears slightly elongated in the direction perpendicular to the disk (see the dashed line in Figure 1, right panel), as if an ionized wind were launched from the disk (Sections 3.3 and 4).

3.3. H21α Centroid Map

The bright emission of the H21α masers allows us to investigate the kinematics and internal structure of MonR2-IRS2 with an exquisite accuracy. For this, we have constructed the centroid map of the H21α emission in 5 km s⁻¹ channels using 2D Gaussians (Figure 3, black crosses and filled circles). The typical errors in the centroid positions (error bars in Figure 3) are 1–2 mas (0.9–1.8 au) for velocities between −20 and 40 km s⁻¹. Note that positional errors are also a function of the errors in the bandpass calibration (Zhang et al. 2017). However, the phase noise in the ALMA bandpass data was ≤1°, which implies a negligible effect in the positional errors of the H21α centroids.

Zoom In Zoom Out Reset image size

Figure 3 shows that for velocities between −8 and 32 km s⁻¹, the H21α centroids are distributed along an almost perfect line in the northwest–southeast direction. The centroids, however, depart from this line at high velocities, especially for the blueshifted gas (≤−8 km s⁻¹), although some of the associated positional errors are large. This suggests the presence of a second kinematic component. As shown in Section 4, this configuration is consistent with the presence of an almost edge-on ionized disk, and a disk wind.

Similar configurations to the RRL centroid map obtained for MonR2-IRS2 have been observed for MWC349A (Martín-Pintado et al. 2011; Báez-Rubio et al. 2013; Zhang et al. 2017) and MWC922 (Sánchez Contreras et al. 2019). This configuration has been modeled assuming that the massive star is surrounded by a (partly) ionized Keplerian disk and an ionized wind. We have thus adopted a similar physical structure to model the H21α centroid map derived for MonR2-IRS2. The set of input parameters that best fit our ALMA observations is described below (see also Table 1).

Table 1.  Input Parameters of the Best Fit of the MonR2-IRS2 H21α Data

Parameter	Value
Central mass, M*	15 M⊙
Density distributiona, ${N}_{{\rm{e}}}(r,\theta )$	5.36 × 107r−2 $\exp \left[\left(\theta -{\theta }_{{\rm{a}}}\right)/20\right]$ cm−3
Inner radius, rmin	13 au
Double-cones semi-opening, θa	57°
Electron temperature, Te	5850 K
Inclination angle, θi	−25°
Opening angle of the ionized disk, θd	65
Outflow terminal velocity, v0	10 km s−1
Outflow turbulent velocity, vtu	5 km s−1
Mass-loss rate, $\dot{M}$	2.8 × 10−7 M⊙ yr−1
Mass of the ionized disk, md	2.4 × 10−5 M⊙

Note.

^ar is in units of $15$ au and angles are measured relative to the polar axis; θ and θ_a, are in degrees.

Download table as:  ASCIITypeset image

In our model, the ionized wind lies inside a double cone with a semi-opening angle of 57° (the neutral disk), while the ionized disk layer is contained within an opening angle of 65 on the surface of the double cone (Figure 4). The ionized wind expands radially at a constant velocity, v₀, while the circumstellar disk rotates following a Keplerian law with the velocity rotation component being added to the expansion velocity of the wind (Martín-Pintado et al. 2011; Báez-Rubio et al. 2013). The inclination angle is measured from the direction perpendicular to the disk and is −25°.

Zoom In Zoom Out Reset image size

We have used the non-LTE 3D radiative transfer code MORELI (MOdel for REcombination LInes; Báez-Rubio et al. 2013) and the departure coefficients of hydrogen, b_n and β_n, calculated by Walmsley (1990). The velocity separation between the H21α emission peaks (blueshifted and redshifted; Figure 2) constrains the central mass of the MonR2-IRS2 source to 15 M_⊙. This mass corresponds to a B0-type star on the zero-age main sequence, which is consistent with the luminosity of MonR2-IRS2 (∼0.5–1 × 10⁴ L_⊙; Howard et al. 1994).

The radial density distribution of electrons within the ionized double cone, N_e, follows a r⁻² law with an inner radius of 13 au (Table 1). We choose a r⁻² law because it fits nicely the 1.3, 0.85 and 0.4 mm free–free continuum fluxes of MonR2-IRS2, and it allows us to simultaneously model the disk and the expanding wind in a simple way. The predicted mass-loss rate for MonR2-IRS2 is 2.8 × 10⁻⁷ M_⊙ yr⁻¹, and the mass of the ionized disk is 2.4 × 10⁻⁵ M_⊙, a factor ≥10 lower than the neutral disk mass (∼2–4 × 10⁻⁴ M_⊙; Section 3.1).

The electron temperature, T_e, that best fits our ALMA data is 5850 K. This value is lower than those typically measured in H ii regions (∼8000 K). However, note that T_e can only be constrained using RRL thermal emission. The expanding velocity of the ionized wind is 10 km s⁻¹, with a turbulent component of 5 km s⁻¹. Expansion velocities ≥20 km s⁻¹ are ruled out because the predicted H21α line widths in the wings would be too large. As shown below, the absence of an expanding wind cannot explain the departure of the H21α centroids from the disk plane observed at high velocities (especially blueshifted ones).

Our model fits very well the H21α RRL profile measured toward MonR2-IRS2 (red line, Figure 2). A comparison between the LTE and non-LTE cases⁸ reveals that non-LTE effects need to be invoked to explain the H21α double-peaked profile. For the H30α and H26α RRLs, the model predicts peak intensities a factor of ∼2 higher than observed. However, we note that RRL masers are highly variable (Thum et al. 1992).

The H21α centroid distribution predicted by our model also nicely matches that measured with ALMA (color lines and filled squares; Figure 3). The Keplerian component reproduces fairly well the kinematics of H21α emission along the disk plane. The observations, however, reveal small deviations with respect to the model at velocities 2, 12, and 17 km s⁻¹ (such deviations are larger than the centroids positional errors), as if the disk were warped. Disk warping may appear in systems presenting a secondary object such as a stellar companion or a massive planet (Nealon et al. 2018; Kraus et al. 2020). This would be consistent with the fact that the modeled electron density distribution of MonR2-IRS2 requires an internal hole of radius 13 au. This was already noted by Jiménez-Serra et al. (2013) since, unlike MWC349A, the free–free continuum emission of MonR2-IRS2 at λ ≤ 1.3 mm is optically thin. Alternatively, the warping of the disk could be produced by anisotropic accretion of gas (Sakai et al. 2019) or by a misalignment between the disk rotation axis and the magnetic field direction (Ciardi & Hennebelle 2010).

From Figure 3, we also find that the ionized gas at high velocities (mostly blueshifted) departs from the disk plane at distances ∼12 mas from the disk center. This departure can only be explained by the presence of an ionized wind launching at radii ∼11 au. Indeed, the same model but without a wind, does not reproduce such deviations (black line in Figure 3). As found toward other sources (MWC349A or MWC922; Martín-Pintado et al. 2011; Báez-Rubio et al. 2013; Zhang et al. 2017; Sánchez Contreras et al. 2019), launching radii of ∼10 au are consistent with magnetically-regulated disk wind models (Blandford & Payne 1982) rather than with X-wind theory (Shu et al. 1994). Disk UV photoevaporation cannot be responsible either for the formation of the disk wind in MonR2-IRS2 because its gravitational radius, r_g (the distance at which photoevaporating disk winds are launched; Hollenbach et al. 1994), is ∼100 au, i.e., 10 times larger than that observed in MonR2-IRS2.

Alternatively, the departure of the high-velocity centroid positions could be due to a second warp in the outer disk of MonR2-IRS2. However, as shown by Nealon et al. (2018), the disk twist produced by a companion such as a massive planet can go up to 60°. This value is smaller than the putative disk twist observed for MonR2-IRS2 (>90°, Figure 3).

Finally, note that the model used here is different from the one proposed by Jiménez-Serra et al. (2013). The previous model was motivated by the asymmetry of the H26α RRL detected toward MonR2-IRS2 with the SMA. Such an asymmetry is not observed in the H21α line (Section 3.2), which suggests that the H26α maser may be variable. Variability is common in RRL masers since their line profiles can change on timescales as short as 30 days (see MWC349A; Martín-Pintado et al. 1989b; Thum et al. 1992). Future ALMA observations will reveal whether or not the H26α line asymmetry detected with the SMA was transitory.

Sections 3 and 4 show that RRL masers are powerful tools to unveil the kinematics and physical structure of the innermost regions around massive stars. Indeed, RRL masers are very sensitive to the electron density distribution, geometry, electron temperature, and kinematics of the ionized gas (Strelnitski et al. 1996; Báez-Rubio et al. 2013, 2014). Regardless of their evolutionary stage, all RRL maser objects firmly detected to date (MWC349A, MWC922 and MonR2-IRS2; Martín-Pintado et al. 2011; Báez-Rubio et al. 2013; Sánchez Contreras et al. 2019, and this work) present the same physical structure for the ionized gas: a circumstellar disk rotating in a Keplerian fashion, and an expanding ionized wind launched at distances of tens of astronomical units that follows the same Keplerian rotation. This disk+wind geometry may thus be the optimized configuration for RRL masers to form.

We thank the anonymous referee for the constructive comments. This Letter makes use of the ALMA data ADS/JAO.ALMA 2012.1.00522.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), NSC and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO, and NAOJ. I.J.-S., A.B.-R., and J.M.-P. acknowledge support from the Spanish FEDER (project number ESP2017-86582-C4-1-R) and the State Research Agency (AEI; project number PID2019-105552RB-C41). V.M.R. has received funding from the EU Horizon 2020 program under the Marie Skłodowska-Curie grant agreement No. 664931. We thank A. Richards for the help provided during the calibration/imaging of the data.