Molecules with ALMA at Planet-forming Scales (MAPS). XVI. Characterizing the Impact of the Molecular Wind on the Evolution of the HD 163296 System

The data presented here were collected as part of the ALMA Large Program, The Chemistry of Planet Formation (2018.1.01055.L), with co-PIs K. I. Öberg, Y. Aikawa, E. A. Bergin, V. V. Guzmán, and C. Walsh. This paper is focused on the HD 163296 disk and the CO line data only. These select CO isotopologues and transitions used in this paper are listed in Table 1. For full details on the program and the data reduction process please see the overview paper (Öberg et al. 2021). We first checked for the presence of the blueshifted wind as detected in Klaassen et al. (2013) by inspecting the data in CASA using plotms. Figure 1 presents the uv amplitude of the data as a function of velocity averaged over the four shortest baselines (19, 28, 29, and 32 m). The channel widths chosen for the J = 2 − 1 and J = 1 − 0 lines are 0.2 km s⁻¹ and 0.5 km s⁻¹, respectively. Full information on the data calibration can be found in Öberg et al. (2021). The blueshifted wind is most apparent in the ¹²CO J = 2 − 1 line from ≈ −14 to −12 km s⁻¹ local standard of rest kinematic frame (LSRK; highlighted in the left-hand gray box in Figure 1). This feature is also detected in the ¹³CO lines but is undetected in the C¹⁸O lines (not shown in Figure 1). The emission from ≈13 to 15 km s⁻¹ LSRK is from a bright background cloud in the Galactic center, situated at a velocity far in the line wings of the emission from HD 163296 itself. HD 163296 is located at low Galactic latitude (l = 724°, b = +149) and the 1.2 m CO telescope survey by Dame et al. (2001) reveals bright ¹²CO J = 1 − 0 emission at the velocities indicated in the MAPS data.

Zoom In Zoom Out Reset image size

Table 1. Observations of CO Isotopologues toward the HD 163296 System

Molecule/Transition	Frequency	Eup	Beam (PA)	δv	rms	rmspbcorr	Peak Flux
	(GHz)	(K)		(km s−1)	(mJy beam−1)	(mJy beam−1)	(mJy beam−1)
12CO J = 2 − 1	230.538000	16.6	030 × 025 (85°)	0.2	1.12	3.05	12.4
12CO J = 2 − 1	230.538000	16.6	10 × 085 (85°)	0.5	2.20	5.00	78.4
13CO J = 2 − 1	220.398684	15.9	10 × 085 (66°)	0.5	1.70	4.45	40.2
13CO J = 1 − 0	110.201354	5.3	10 × 085 (68°)	0.5	1.12	2.95	24.0

Note. Line frequencies and upper energy levels are taken from the Cologne Database for Molecular Spectroscopy (Müller et al. 2005).

Download table as:  ASCIITypeset image

As the line images generated using the MAPS imaging pipeline described in Czekala et al. (2021) cover the velocity range of the disk emission only, custom images were generated for this study. As a starting point, the tCLEAN optimized imaging parameters described in Czekala et al. (2021) were used. The shortest baseline (19 m) was excluded from the imaging because when included, it generated significant striping in the images that is characteristic of larger-scale emission not resolved by the interferometer. At Band 6, the primary beam is ≈38'' in diameter and is ≈76'' at Band 3. As noted in Huang et al. (2021) the maximum recoverable scale at Band 6 is ≈12''. Excluding the 19 m baseline results in a maximum recoverable scale of ≈11''. The spatial extent of the wind is ≈10'', therefore spatial filtering should not be a significant issue as this is less than the maximum recoverable scale. From the ¹³CO, the wind traced in the J = 2 − 1 and J = 1 − 0 transitions is approximately the same spatial extent. Since the Band 3 data has a much larger maximum recoverable scale, we do not appear to be missing more extended emission, but we cannot confirm this without shorter-baseline/single-dish data.

For a Keplerian disk the position–velocity pattern of the line emission is well characterized and therefore we can use an analytical clean mask calculated using the stellar mass and position and inclination angles of the disk. The velocity structure of the wind does not yield to a similar simple parameterization and therefore the clean mask was hand drawn. All images were generated with a Briggs robust parameter of 0.5 (Briggs 1995). As detailed thoroughly in Czekala et al. (2021), we applied a correction to the CLEANed channel maps. This is done because the units of the residuals are in janksy per dirty beam whereas the CLEAN model is in janksy per clean beam. When the dirty beam is non-Gaussian these units are no longer equivalent. This was first outlined in Jorsater & van Moorsel (1995). This so-called JvM correction is a rescaling of the residuals that are added to the CLEAN model in the final stage of the CLEAN pipeline. The primary beam correction was then applied to the resulting images.

The ¹²CO J = 2 − 1 line was imaged at a velocity resolution of 0.2 km s⁻¹, and to improve image quality and signal-to-noise, a uv taper to force a 030 beam major axis was used, resulting in a synthesized beam size of 030 × 025 (85°), which is ≈30 × 25 au. In order to have matching resolution data between the Band 6 and Band 3 ¹³CO J = 2 − 1 and J = 1 − 0 data, the lines were both imaged with the same uv tapers at the same velocity resolution. The beam size was chosen to give the best compromise between signal-to-noise and spatial resolution. A uv taper was used to force a beam major axis of 100, which resulted in a synthesized beam size of 100 × 085 (68°), which is ≈100 × 85 au with a velocity resolution of 0.5 km s⁻¹. The ¹²CO J = 2 − 1 transition was also reimaged with same uv taper as the ¹³CO J = 2 − 1 transition to allow for a direct comparison of the fluxes between emission from the two isotopologues. The rms noise from the line-free channels of the image cubes, before and after the primary beam correction, and the peak values of the emission for all lines, are listed in Table 1. The resulting channel maps for ¹²CO J = 2 − 1 are shown in Figure 7 in the Appendix.

The integrated intensity maps were made by summing over all of the emission in channels with >3σ emission, where σ is the rms noise, before primary beam correction but after the JvM correction. For the ¹²CO J = 2 − 1 line this covered a velocity range of −14.4 to −11.4 km s⁻¹ LSRK (16 channels), and for the ¹³CO lines this was −13.5 to −11.5 km s⁻¹ LSRK (five channels) for the J = 1 − 0 transitions and −13.5 to −12.0 km s⁻¹ LSRK (four channels) for the J = 2 − 1 transition. Figure 2 presents the integrated intensity map and the intensity-weighted velocity map. The latter was generated over the same channels as the integrated intensity map but with a 4σ noise clip applied to the channel map. Figure 3 presents the integrated intensity maps for the ¹³CO J = 1 − 0 and J = 2 − 1 lines and the intensity-weighted velocity map. The latter was generated over the same number of channels as the integrated intensity map but with a 4σ noise clip applied to the channel map. Also shown in Figures 2 and 3 are the brightness temperature maps for ¹²CO and both of the ¹³CO lines. These were generated from the peak intensity (moment 8) maps using the imagecube.jybeam_to_Tb_RJ and imagecube.jybeam_to_Tb functions in the gofish package (Teague 2019). ²⁶

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The integrated intensity maps presented in Figure 2 show that the width of the emission along the axis perpendicular to the jet varies. The emission close to the disk is ≈500 au wide, then this narrows to ≈100 au and then widens again. This is also seen in the channel maps shown in Figure 7 in the Appendix. We recover the double-corkscrew structure in the ¹²CO J = 2 − 1 gas kinematics in the intensity-weighted velocity maps as first reported by Klaassen et al. (2013). The same kinematic structure is seen in the two ¹³CO lines (see Figure 3).

The high-velocity outflow, HH 409, from the HD 163296 system is periodic and asymmetric (Wassell et al. 2006; Ellerbroek et al. 2014). The jet from the near side of the disk has three knots (A, A2, A3) and a bow shock (H). Figure 3 shows the ¹³CO moment maps with the well-constrained axis of the blueshifted jet and the estimated locations of the knots at the time of our observations. The proper motion of the blueshifted knots is 0.49 ± 0.01 yr⁻¹ (Wassell et al. 2006). This means that the knots will have moved ≈325 since the data presented in Klaassen et al. (2013) was taken. In Klaassen et al. (2013) the ¹²CO J = 3 − 2 emission was associated with the location of knot A3. In our data A3 is the only knot that is potentially spatially associated. A3 is ≈20 from the brightest points in the ¹³CO brightness temperature maps. Knots A and A2 are still within the primary beam of the Band 6 data but no significant emission is detected at these locations. Beyond the Band 6 primary beam at 280 there is ¹³CO J = 1 − 0 emission associated with the bow shock H. To determine the current distance of the bow shock (≈2725), the proper motion was assumed to be the same as that for the knots.

To further investigate the kinematics of the wind we make position–velocity (PV) diagrams. We take cuts perpendicular to the axis of the jet and average over one beam. We present Figure 4 as an example case in the main text and the rest are shown in the Appendix in Figure 11. The transverse velocity gradient shown in the moment 1 maps are recovered in the PV diagrams and are in the form of a coherent tilt, highlighted by the red line, that is suggestive of rotation for projected distances, z ≲ 600 au. Quantitatively, the observed tilt in the PV diagram at z = 480 au has a spatial extent and velocity range of Δr ∼ 4 and ΔV ∼ 1.6 km s⁻¹, respectively. Interpreting this tilt as rotation, this gives a rotation velocity of ${V}_{\phi }\equiv \tfrac{{\rm{\Delta }}V}{2\sin i}=1.1$ km s⁻¹ and a specific angular momentum of j ≡ V_ϕ Δr/2 = 220 km s⁻¹ au where i = 467 is the inclination of the flow with respect to the line of sight.

Zoom In Zoom Out Reset image size

From our detection of ¹³CO, we find that the ¹²CO J = 2 − 1 line has to be optically thick as the peak flux ratio of the ¹²CO J = 2 − 1 and ¹³CO J = 2 − 1 lines is ≈2 in the channel maps (see Table 1). This result is in contrast to previous work. In Klaassen et al. (2013) the CO column density (N(CO), cm⁻²) and kinetic temperature (T_Kin, K) of the wind were determined from the ¹²CO J = 3 − 2 and J = 2 − 1 lines. Their analysis focused on a single knot of emission where both lines were detected and here both lines were assumed to be optically thin with T_Kin = 960 K and N(CO) = 1 × 10¹⁵ cm⁻².

Because the molecular wind is traced in both of the ¹³CO J = 2 − 1 and J = 1 − 0 lines, this ratio will provide additional constraints on the column density and excitation conditions. The ratio of the peak brightness temperature maps as shown in Figure 3 was calculated, resulting in a J = 2 − 1/J = 1 − 0 ratio of 0.39 where 1.13 K and 2.88 K are the peak values for each line, respectively. To determine the kinetic temperature, T_Kin, we used the non-LTE radiative transfer code RADEX (van der Tak et al. 2007). RADEX calculates the brightness temperature, T_b , under the Rayleigh–Jeans approximation; therefore, the output is expected to be consistent with the maps generated in Section 2.2. We first calculate T_b for both lines over a grid in temperature (T_Kin from 5 to 1000 K in steps of 5 K) and an assumed a gas density of n_H = 1.9 × 10³ cm⁻³. This density is the same as that adopted in Klaassen et al. (2013) that is based on an independent measurement of the n_H volume density in the knot (1.9 × 10³ cm⁻³; Wassell et al. 2006). This is not necessarily the n_H density in the wind. However, because of the low (≈10³ cm⁻³) critical densities of the low-lying rotational transitions of CO, the hydrogen gas density is not a key parameter in the fit. We fix the line width to 1 km s⁻¹ and first use the same ¹²CO column density as Klaassen et al. (2013) reduced by a factor of 70 for ¹³CO to account for the elemental ¹²C/¹³C ratio. Under these conditions, the brightness temperatures for the lines are a factor of 100 too low when compared to the observations. The results from the above RADEX models are shown in the Appendix in Figure 12. The ¹³CO column density was gradually increased (in steps of 1 × 10¹⁵ cm⁻² when the values were close) until the observed brightness temperatures were reproduced at a temperature consistent with the line ratio. This was done over a finer temperature grid of a smaller range (T_Kin from 5 to 100 K in steps of 0.5 K). This is achieved with a ¹³CO column density of 9.0 ± 1.0 × 10¹⁵ cm⁻² at a kinetic temperature of 7.5 K. The brightness temperatures, line ratio, and optical depths are presented in Figure 5. This is a considerably lower temperature than Klaassen et al. (2013), who found 960 K. However, our data clearly rule this out as the ¹²CO brightness temperature at peak is approximately the same as our temperature derived from the line ratios. The largest error on the derived column density depends on the assumed line width, e.g., with a line width of 0.5 km s⁻¹ the required N(¹³CO) to match the observations is 2 times lower than calculated with 1.0 km s⁻¹. An additional caveat is that we assume both the 2 − 1 and 1 − 0 lines have the same emitting area. This is reasonable given the data in hand, and the lines are close in excitation so should be tracing similar regions of the outflow.

Zoom In Zoom Out Reset image size

The dynamical timescale, i.e., the time since the launching event, can be estimated from the velocity of the gas and the spatial extent of the emission. This assumes that the flow is perpendicular to the ecliptic plane of the disk. For the rotating wind, the deprojected length of the flow is ≈1400 au (assuming an inclination angle of 467; Huang et al. 2018). The deprojected velocity is −28 km s⁻¹, where the projected velocity is taken as the average velocity of the wind, −12.9 km s⁻¹ LSRK, corrected for the velocity of the source (+5.76 km s⁻¹; Teague et al. 2019). The dynamical timescale, t_dyn, is therefore ≈240 yr. We can do the same calculation for the knot detected in ¹³CO J = 1 − 0 line emission associated with the bow shock H and this results in t_dyn ≈ 500 yr.

The mass in the wind can be estimated from the ¹³CO column density required in the RADEX models in Section 3.2. When assuming a ¹²CO/¹³CO abundance ratio of 70 the column density of ¹²CO in the wind is 6.3 × 10¹⁷ cm⁻². This can be considered a lower limit since isotope-selective photodissociation preferentially destroys ¹³CO over ¹²CO in the case where ¹²CO is self-shielding and ¹³CO is not (Visser et al. 2009). If active, this would increase the ¹²CO/¹³CO ratio in the material in the surface of the disk that is carried away by the wind, or indeed within the wind itself. This can then be converted to a total gas column by assuming a CO/H₂ ratio of 10⁻⁴ and a mean molecular mass of 2.4m_H, where m_H is the atomic mass of hydrogen. This mass density (2.54 × 10⁻² g cm⁻²) can be used to calculate a total gas mass by multiplying by the projected area of the wind. For simplicity we take this to be a rectangle with length 1000 au and width 400 au (measured approximately from the PV diagrams, see Figure 11). This results in a total gas mass of M_w = 1.0 × 10⁻³ M_⊙ (or 1 M_Jup). This is 0.07% of the total HD 163296 disk gas mass as determined in Calahan et al. (2021). The mass-loss rate, ${\dot{M}}_{w}$, of the wind can then be determined from M_w /t_dyn and is 4.8 × 10⁻⁶ M_⊙ yr⁻¹. The ejection-to-accretion ratio is ${f}_{M}={\dot{M}}_{w}/{\dot{M}}_{\mathrm{acc}}\simeq 5\mbox{--}50$, with the uncertainty being driven by the error uncertainty on the measurement of the absolute accretion rate and its intrinsic variability (Ellerbroek et al. 2014; Wichittanakom et al. 2020). The ratio of the mass-loss rate from the optical jet compared to the accretion rate is much lower, ≃0.01–0.1 (Ellerbroek et al. 2014). Such a large mass flux is in line with previous results obtained in the rotating molecular flows from sources of different ages and masses. For example, Louvet et al. (2018), Zhang et al. (2019), and de Valon et al. (2020) derive mass-loss rates of 45, >10, and 35 times larger than the mass-loss rate of the optical jets in these systems. Regardless of the origin of the mass-loss process, our data confirm that rotating CO outflows can extract a significant mass fraction of the accreted flow.

4.1.1. Is the Outflow Entrained Envelope Material?

4.1.2. Is the Outflow a Photoevaporative Wind?

4.1.3. Is the Outflow an MHD Disk Wind?

The MHD disk-wind model is the most compelling scenario to account for the high ejection-to-accretion mass ratio and the kinematic structure of the CO outflow. In fact, recent global numerical simulations of weakly ionized disks, including the heating of the disk atmosphere, predicts mass-loss rates of about, or larger than, the stellar accretion rate; this is lower than our estimates (Bai 2017; Béthune et al. 2017; Wang et al. 2019).

The radial region of the disk from which the wind originates can be determined from the specific angular momentum and the axial velocity estimated in the wind. Assuming a steady, axisymmetric, and dynamically cold (negligible enthalpy) MHD disk wind, the launch radius ϖ₀ can be deduced from the Anderson et al. (2003) formula

$$\begin{eqnarray}&&\begin{array}{r}\begin{array}{l}{\varpi }_{0}\approx 13\,{\rm{a}}{\rm{u}}{\left(\displaystyle \frac{{\varpi }_{{\rm{\infty }}}}{100\,{\rm{a}}{\rm{u}}}\right)}^{2/3}{\left(\displaystyle \frac{{v}_{\phi ,{\rm{\infty }}}}{5\,{{\rm{k}}{\rm{m}}\,{\rm{s}}}^{-1}}\right)}^{2/3}\\ \,\times \,{\left(\displaystyle \frac{{v}_{p,{\rm{\infty }}}}{30\,{{\rm{k}}{\rm{m}}\,{\rm{s}}}^{-1}}\right)}^{-4/3}{\left(\displaystyle \frac{{M}_{\ast }}{1{M}_{\odot }}\right)}^{1/3},\end{array}\end{array}\end{eqnarray} \tag{ 1 }$$

where ϖ_∞ is the radius of the rotating wind, v_ϕ,∞ is the rotation velocity, v_p,∞ is the poloidal flow speed, and M_* is the mass of the central star which is taken as 2.0 M_⊙ (e.g., Isella et al. 2018). From the specific angular momentum and the axial velocity estimated in Section 3.2, we deduce ϖ₀ = 4 au. This is a representative radius because, as we discuss in Section 4.2, the wind may be perturbed by the jet. This shows that the wind is coming from the inner ≈10 au region of the disk. This result is in line with other studies of rotating outflows that find a launching radius between 1 and 40 au (e.g., Launhardt et al. 2009; Bjerkeli et al. 2016; Hirota et al. 2017; Tabone et al. 2017; Louvet et al. 2018; Zhang et al. 2018; de Valon et al. 2020).

Another key parameter of the MHD disk wind is the magnetic lever arm parameter λ. This is the ratio between the specific angular momentum in the wind to that at the launching point,

$$\begin{eqnarray}&&\lambda =\displaystyle \frac{{\varpi }_{\infty }{v}_{\phi }}{{v}_{{\rm{K}}}({\varpi }_{0}){\varpi }_{0}},\end{eqnarray} \tag{ 2 }$$

where v_K(ϖ₀) is the Keplerian velocity at the launch point (we neglect here the angular momentum stored in the form of magnetic torsion, see Ferreira et al. 2006). λ can be considered as the efficiency of the wind to drive accretion: the higher the value of λ, the less mass is required to be launched to extract a given amount of angular momentum. The deduced magnetic lever arm parameter at z = 480 au is λ = 2.3. Therefore, the gas would increase its specific angular momentum by about a factor two in the wind via magnetic acceleration. Our estimated value lies between than that of DG Tau B and HH 30 (λ ≃ 1.6, Class I and II objects respectively, Louvet et al. 2018; de Valon et al. 2020) and that derived in HH 212 (λ = 5.5, Class 0, Tabone et al. 2017). These relatively low values of λ, together with the high mass-loss rate is well in line with recent numerical simulations that include the heating of the disk and a relatively low magnetization of the disk (plasma parameter β ∼ 10³–10⁵). Interestingly the λ parameter for the jet is constrained to ≤14 by Ellerbroek et al. (2014) and on small scales (<0.13 au) Brγ emission arises from a hundreds of kilometers per second potential MHD-driven wind (Garcia Lopez et al. 2015) that has been modeled with λ = 3. This shows a change in efficiency of angular momentum extraction from the very inner regions of the system out to a few astronomical units.

The presence of a knotty optical jet recently imaged by the Very Large Telescope's Multi Unit Spectroscopic Explorer, the kinematical properties of the outflow, and the low temperature suggest that the pulsating jet may be interacting and compressing the wind. Figure 9 shows the ¹²CO J = 2 − 1 channel maps with the recent observations of [S ii] 673 nm and Hα from Xie et al. (2020) overlaid.

First of all, the transverse PV diagrams of CO exhibit a bar-like morphology with a straight velocity gradient (red line, Figure 4). This contrasts with synthetic PV diagrams of stationary MHD disk winds that show slower material off the jet axis and faster material closer to the axis (Pesenti et al. 2004; Tabone et al. 2020). An extended MHD disk wind is indeed made of nested streamlines with axial and rotation velocities roughly scaling with the Keplerian velocity at their footpoints. The inner collimated streamlines are therefore faster than the outer wide-angle streamlines. Therefore, the CO emission in HD 163296 traces a rotating thin shell of gas with a limited velocity shear rather than onion-like winds as expected by a stationary MHD-wind model. Such spatio-kinematical patterns have already been unveiled by ALMA toward the T Tauri outflow HH 30 (Louvet et al. 2018) and the massive Orion I Source (Hirota et al. 2017).

The relatively low temperature of the outflow (≈7 K) is also in tension with MHD disk-wind models. Panoglou et al. (2012) show that ion–neutral friction during the MHD acceleration can indeed heat up the gas launched from a few astronomical units up to few thousand kelvin for high values of λ (∼10) values. Wang et al. (2019) computed models that are more in line with the properties of the HD 163296 wind (λ ≈ 2) and find temperatures of about 100 K, an order of magnitude larger than the value derived in HD 163296.

We suggest that the appearance of the outflow as a thin shell of cold gas points toward the presence of shocks in the disk wind. These shocks could be driven by the fast pulsating jet of HD 162396 embedded in the CO outflow. In fact, the interaction between a fast pulsating jet and a slower disk wind results in the formation of a thin shell of gas swept up by the jet's bow shocks (Tabone et al. 2018). Most striking, in Figure 9 over velocity channels from −12.8 to −11.80 km s⁻¹ LSRK, the wind is spatially offset but running parallel to the optical jet. Interestingly, our CO maps, as well as the CO(J = 3 − 2) maps presented in Klaassen et al. (2013), do show these bow-shock structures associated with the jet. The low temperature of CO would then be the result of the compression of the wind by the shock. Further modeling is required to determine under what conditions (density, shock velocity, magnetization), the gas can cool down to ∼7 K in such a short period of time (t_dyn = 500 yr).

One should then keep in mind that if the slow wind is perturbed by the fast jet, the Anderson et al. (2003) formula used above to derive the launching radius is not strictly applicable (De Colle et al. 2016). MHD simulations including a pulsating jet are needed to study the biases in the derivation of the launching radius from the observed rotation signatures.

The high-velocity jet is dust free, but it is possible that the observed near infrared excess and optical variability from HD 163296 is due to dust clouds entrained in the molecular wind that are launched at the same time as the jet ejection events as discussed in Ellerbroek et al. (2014) and Rich et al. (2020). Dust, in particular <0.1 μm sized grains, can also can be transported in MHD winds from the inner regions and then can fall back onto the disk at larger radii (Giacalone et al. 2019). This means that there is potentially mass loss of both gas and dust from the inner planet-forming zone of the disk. The dusty wind can also shield the disk material from far-UV radiation (e.g., Panoglou et al. 2012). This could affect the UV-driven chemistry in the upper layers of the disk.

The relative impact of the wind on disk accretion can be estimated by computing the fraction of disk angular momentum extracted vertically by the MHD disk wind, f_J , across the launching region of the wind (see Equation (36) from Tabone et al. 2020),

$$\begin{eqnarray}&&{f}_{J}=\displaystyle \frac{\lambda }{1+\tfrac{{ln}({\varpi }_{\mathrm{out}}/{\varpi }_{\mathrm{in}})}{2{ln}(1+{f}_{M})}},\end{eqnarray} \tag{ 3 }$$

where ϖ_in and ϖ_out are the inner and outer radius of the launch region, f_M is the ratio of the mass-loss rate of the wind over the mass accretion rate onto the star, and λ is the magnetic lever arm parameter. In principle, f_J is between 0 and 1, where 1 means all the accretion though the disk is driven by the MHD disk wind. One of the main caveats in the application of this formula to the flow of HD 163296 is the radial extent of the wind-launching region ϖ_out/ϖ_in. As mentioned above, the flow appears as a thin shell of gas, suggesting a narrow launching region. However, an extended disk wind might lose its onion-like structure if different streamlines are mixed during the interaction of the wind with the jet's bow shock. In other words, the radial extent of the launching region remains highly uncertain and might be larger if the wind has been perturbed by the jet. For a relatively high value for ϖ_out/ϖ_in = 10, the fraction is f_J > 1.0. A narrower wind-launching region would increase f_J . Thus, the angular momentum flux extracted by the wind is at least enough required to sustain accretion through an extended disk region around r ≃ 4 au. We conclude that turbulence is not necessarily required to account for accretion across this specific region of the disk, which is in line with the low value for α turbulence measured in the line-emitting regions, and modeled across the disk (Flaherty et al. 2015, 2017; Liu et al. 2018).

We find that the launch radius of the HD 163296 disk wind is ≈4 au, thus giving key insight into the physical process occurring in the disk on spatial scales not probed with other MAPS data. The cartoon in Figure 6 summarizes all the processes occurring in the inner 10 au of the disk. Because the wind transports mass in the disk both inwards (accretion) and vertically outwards (wind mass-loss rate), disk models with MHD-driven accretion show significantly different, much flatter, surface density profiles in the inner disk than typical MRI turbulence models (e.g., Bai 2016; Suzuki et al. 2016). Additionally, low turbulence will reduce the level of vertical mixing in the disk. This will impact the amount of volatile sequestration in this region of the disk and as a result will have important effects on the C/O ratio of gas and ice available to be accreted by forming planets here (e.g., Semenov & Wiebe 2011; Krijt et al. 2016).

Zoom In Zoom Out Reset image size

If the current mass-loss rate of the HD 163296 system is sustained (mass accretion rate and disk wind mass-loss rate), then it will only take ≈0.02 Myr for the disk to be drained of mass. This is ≈40× quicker than when just considering mass loss via accretion. In reality this process may take longer due to trapping of disk material in the dust rings or a decline of the mass-loss rate, but nevertheless the presence of a disk wind reduces the lifetime of the gas disk and thus the time available for giant planets to accrete their atmospheres. Although CO gas is now detected in a few debris disks the gas masses are very low, on the order of a lunar mass, and therefore would not contribute much to the atmosphere of a giant planet (e.g., Hughes et al. 2018).

MHD-wind-driven disk evolution has been shown to alter the migration direction and timescale of planets in gas-rich disks when compared to viscous accretion models. Type 1 migration is suppressed with the migration timescale increasing to 1 Myr in the region where the wind is active (Ogihara et al. 2018). This mitigates the problem of rapid inward migration and loss of ∼Mars-sized bodies before the dissipation of the gas disk. Kimmig et al. (2020) show that models of disks with MHD-wind-driven accretion can also lead to Type III outward migration for Saturn-to-Jupiter mass bodies. Some of their models have a λ of 2.25, similar to what we derive for HD 163296. In their models they have a parameter called b that is akin to a viscosity parameter. Using our HD 163296 model parameters from Calahan et al. (2021) and Equations (4) and 6 in Kimmig et al. (2020), we find that within < 50 au, our model has a value of b that is between their 10⁻⁴ and 10⁻³ models. With b = 10⁻⁴, inward migration is seen for both Saturn- and Jupiter-mass planets, and in the case of 10⁻³ this leads to outward migration of Saturn-mass planets and periodic inward and outward migration of Jupiter-mass planets.

The periodic ejection events traced by the knots in the jet bring into question the stability of the mass reservoir in the inner disk. The different mechanisms powering the ejection of material in the innermost disk via the HH 409 outflow have been heavily discussed in the literature (see Ellerbroek et al. 2014; Rich et al. 2020, and references therein). One option is the interaction of disk material with a planet. The ≈16 yr periodicity of the ejection events corresponds to a Keplerian orbital radius of ≈8 au, which is close to the position of a gap in millimeter dust at 10 au (Huang et al. 2018). Another is that the accretion variability, and also the variability in the ejected material traced via the optical jet, is due to gravitational instabilities in the inner disk (e.g., Vorobyov 2009). Regardless of the mechanism behind the ejection events, the HD 163296 star is still actively accreting and the disk evolution continues in parallel with the presence of already formed (or forming) planets (Pinte et al. 2018; Teague et al. 2018). The inner mass reservoir where planet formation is expected to be most efficient is not stable and the mass-loss rate of the wind is high and potential dispersal timescale quick. This will have significant effects on the formation and evolution of planets in the system.