Imaging a Central Ionized Component, a Narrow Ring, and the CO Snowline in the Multigapped Disk of HD 169142

As can be seen in Figure 1, our 7 mm image shows a narrow ring of emission with a radius of ∼25 au. For the width of the ring we estimate a deconvolved FWHM of ∼8 au, measured along the E–W direction, where the synthesized beam size is smaller. The ring in our images coincides quite well with the ring imaged in previous near-IR (Quanz et al. 2013; Monnier et al. 2017), 7 mm (Osorio et al. 2014), and 1.3 mm (Fedele et al. 2017) observations, but our images reveal additional details.

We have estimated the position of the center of the ring by fitting an ellipse with its major axis along the position angle of the disk, as estimated from previous molecular observations (PA = 5°; Raman et al. 2006), to the ring image, excluding the regions where two dips and a knot of emission are found (see below). From this fit we estimate that the center of the ring in the 7 mm image (epoch ∼ 2014.17) is located at α(J2000) = 18^h24^m2977798 ± 000018, δ(J2000) = −29°46'498673 ± 00028, which coincides within ∼4 ± 3 mas with the position of the star HD 169142 given in the Gaia catalog, after correcting for proper motions (Gaia Collaboration et al. 2016). This result indicates that the ring is very well centered on the star, and that the alignment of the data to make the final image and the quality of its astrometry are very good. From our fit we also estimate a radius of the ring of 0214 ± 0004 (25.0 ± 0.5 au), which is slightly larger than the radius of ∼019 (∼22 au) estimated from the near-IR scattered-light images (Quanz et al. 2013). This suggests that the scattered-light emission arises mainly from the inner rim or wall of the ring, whereas the 7 mm emission traces mainly the surface density of the large dust grains that peaks at slightly larger radii.

The intensity of the ring in our 7 mm image is significantly asymmetric in azimuth, showing a knot of emission ∼4 σ above the average intensity of the ring at PA ≃ −40°, in agreement with the previous results of Osorio et al. (2014), who noted this azimuthal asymmetry from 7 mm data of lower angular resolution. We think that this knot represents a real azimuthal asymmetry since an intensity enhancement appears both in the images made from the new A configuration data alone and in the lower angular resolution maps reported in Osorio et al. (2014). The ALMA 1.3 mm images also show a hint of this asymmetry (Fedele et al. 2017), although at a lower significance, probably due to the higher optical depth of the ring at shorter wavelengths, or an accumulation of the large dust grains preferentially traced at 7 mm.

On the other hand, our new 7 mm image also shows significant decreases of intensity or dips in the ring at PA ≃ 0° and PA ≃ −170°. As shown by Osorio et al. (2014), an elongated beam can produce depressions of emission along the direction of the major axis of the beam in an axisymmetric ring. Additionally, noise fluctuations can produce spurious clumpy structures during the deconvolution process. In order to check whether the substructure detected in our images is real or due to these spurious effects, we obtained simulated images with random thermal noise using the model presented by Osorio et al. (2014) and the SIMOBSERVE task in CASA. These simulations showed that, for certain noise structures, intensity depressions similar to the ones in our observations could be formed in the narrow ring, suggesting that the observed dips in our 7 mm observations could be produced because of the combined effect of noise fluctuations and the elongated beam. Nevertheless, none of our simulations produced a bright knot similar to the one detected in our 7 mm observations, indicating that it is tracing a real azimuthal asymmetry in the ring.

Polarized scattered-light images at H band show an axisymmetric ring with only a possible dip at a PA ≃ 80° (Quanz et al. 2013), where no significant drop of emission is seen in our images. The fact that the knot of emission in the ring is not present at near-IR wavelengths indicates that it is probably produced by azimuthal asymmetries in the density near the disk midplane, to which our 7 mm images are more sensitive, without affecting significantly the distribution of small dust grains in the disk atmosphere, which are traced by the scattered-light images.

Azimuthal asymmetries in the large dust grain distribution are expected to be produced by tidal interactions between a forming planet and the disk (e.g., Baruteau et al. 2014). Hydrodynamic simulations show that planets can create relatively large cavities with vortices at their outer edges. These vortices, in turn, are able to trap the large dust grains in their pressure maxima, producing lopsided asymmetries at millimeter wavelengths (Birnstiel et al. 2013; Zhu & Stone 2014). Thus, the interaction between the disk of HD 169142 and the possible forming planets at the inner (Biller et al. 2014; Reggiani et al. 2014) and second gaps could be responsible for the observed nonaxisymmetric structure.

Our 7 mm, 9 mm, and 3 cm images have revealed the presence of compact emission inside the bright emission ring, originating near its center (Figures 1 and 2). The emission at 7 mm is slightly extended along the E–W direction, with its intensity peak at a projected distance of ∼3 ± 1 au (∼0025 ± 0010) from the central star toward the W direction. The quoted uncertainty is probably only a lower limit, corresponding to the formal error in the position of the emission peak relative to the center of the ring, estimated as ∼0.5(θ/S/N) (Reid et al. 1988), assuming an unresolved source (θ = 01) and an S/N of 5. Further observations are needed to confirm the reality and origin of this possible displacement (see below).

The insufficient sensitivity and angular resolution of the images at 9 mm and 3 cm, respectively, make it difficult to estimate the morphology of the central emission observed at these wavelengths. This central radio source could not be detected in previous observations by Osorio et al. (2014) toward HD 169142 at 7 mm, due to the lack of angular resolution. We do not identify radio emission associated with the IR source detected by Biller et al. (2014) and Reggiani et al. (2014) located ∼016 (∼19 au) north from the central star.

In principle, both dust and ionized gas could be contributing to the radio emission at the innermost regions of transitional disks. To our knowledge, very few transitional disks have been observed with high enough angular resolution to detect and resolve compact emission at millimeter and centimeter wavelengths inside their central cavities or gaps. LkCa 15 (Isella et al. 2014) and AB Aur (Rodríguez et al. 2014) were imaged with the VLA at 7 mm and 3 cm, respectively, whereas TW Hya was observed with ALMA at 0.87 mm (Andrews et al. 2016). The emission in TW Hya has been attributed to inner residual dust located close to the star, although it has also been recently suggested that a contribution of free–free emission from photoionized gas could be present (Ercolano et al. 2017). The morphology and spectral index of the emission in AB Aur indicates that it is associated with free–free emission from an accretion-driven jet. In LkCa 15, however, the lack of observations at other wavelengths makes it impossible to distinguish between a dust or a free–free origin for the emission. In the following we discuss the origin of the observed compact central radio source in HD 169142.

The near-IR excess in the SED of HD 169142 indicates that its disk is a pre-transitional disk, with a hot dust component located very close to the star. Osorio et al. (2014) modeled the broadband SED and the 7 mm images of HD 169142 and found that an inner disk of 0.6 au in radius, together with its inner wall at the dust sublimation radius (∼0.2 au), could fit the near-IR emission of HD 169142. According to their model, this inner dust component would produce only ∼9, ∼5, and <1 μJy of emission at 7 mm, 9 mm, and 3 cm, respectively. These values are much lower than the observed 74 ± 15 μJy, 45 ± 14 μJy, and ∼20 ± 5 μJy at these wavelengths for the central radio source. From a power-law fit to these observed values of the flux density (S_ν ∝ ν^α), we estimate a spectral index α = 0.82 ± 0.17 (see Figure 3), which is too low to correspond to dust thermal emission (dust thermal emission presents α ≥ 2). Therefore, both the flux density and spectral index of the central radio source indicate that its emission is mainly dominated by partially optically thick free–free emission from ionized gas.

Zoom In Zoom Out Reset image size

Gas near young stellar objects has been found to be ionized by two main mechanisms: shocks in accretion-driven jets (Anglada et al. 2015), and photoionization due to the high-energy radiation from the central star (Alexander et al. 2014). Since accretion and ejection of material are correlated (Cabrit 2007), low mass-accretion rate objects, such as transitional disks, are expected to present relatively weak radio jets. However, recent studies with the VLA have shown that radio jets in this type of source can produce free–free emission at levels that are detectable with the improved sensitivity of the VLA (Rodríguez et al. 2014). In particular, Macías et al. (2016) presented VLA observations at 3 cm toward the transitional disk of GM Aur, revealing resolved free–free emission from a radio jet and from a photoevaporating disk, showing that both mechanisms can contribute at the same level to the total free–free emission.

The central radio source detected at 7 mm in HD 169142 presents a slight elongation with a PA ≃ −80°, which is consistent with the position angle of the disk rotation axis (PA ≃ −85°), as shown by molecular line observations (Raman et al. 2006). This suggests that the central radio source in our images could be tracing an accretion-driven radio jet. We can obtain a rough estimate of the free–free emission of an accretion-driven jet in HD 169142 with the empirical correlation between the radio luminosity of a source, S_νd², and its outflow momentum rate, ${\dot{P}}_{\mathrm{out}}$ (Anglada 1995; Anglada et al. 2015): ${({\dot{P}}_{\mathrm{out}}/{M}_{\odot }{\mathrm{yr}}^{-1}\mathrm{km}{{\rm{s}}}^{-1})={10}^{-2.5\pm 0.3}({S}_{\nu }{d}^{2}/\mathrm{mJy}{\mathrm{kpc}}^{2})}^{1.1\pm 0.2}$. Wagner et al. (2015) measured a mass accretion rate onto the star HD 169142 of ${\dot{M}}_{\mathrm{acc}}\simeq (1.5\mbox{--}2.7)\times {10}^{-9}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$. Assuming a ratio between mass-loss rate in jets and mass accretion rate ${\dot{M}}_{\mathrm{out}}/{\dot{M}}_{\mathrm{acc}}\simeq 0.1$ (Cabrit 2007), we estimate that the outflow momentum rate should be ${\dot{P}}_{\mathrm{out}}\simeq {10}^{-7}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}\,\mathrm{km}\,{{\rm{s}}}^{-1}$. Therefore, the correlation would predict a flux density at 3 cm of S_ν ≃ 5 μJy, which is lower than our estimated flux density of 20 ± 5 μJy for the central radio source. Even though this estimate has large uncertainties, it suggests that an additional mechanism other than shocks associated with the outflow could contribute to the ionization of this jet.

Another possibility is that the observed central radio source was originated as a result of photoionization by high-energy radiation from the central star. Extreme-UV (EUV) and, to a lesser extent, X-ray radiation impinging on the inner disk can ionize its surface (Clarke et al. 2001; Gorti et al. 2009; Owen et al. 2010) and contribute to the observed free–free emission. An inhomogeneous inner disk could lead to an inhomogeneous irradiation of its surface, which could result in the asymmetric morphology of the central radio source observed at 7 mm. Additionally, the ejected gas in the accretion-driven jet could also be significantly photoionized by the high-energy radiation emitted by the star (Hollenbach & Gorti 2009).

Finally, the peak of emission of the central radio source could be tracing the position of an independent object at a radius of ∼3 au from the central star. Given the proximity to the central star, the dynamical timescales involved would be short. An orbiting object at a radius of ≲3 au would have an orbital period ≲4 yr and would show detectable orbital proper motions in a few months. On the other hand, knots of emission in a radio jet are expected to be ejected with velocities of ∼300 km s⁻¹. Given the small inclination angle of the disk (i = 13°; Raman et al. 2006), after projection on the plane of the sky, this would result in proper motions of ∼012 yr⁻¹ away from the central star along PA ≃ −80°. Future observations should reveal detectable variations and/or proper motions in the observed emission that will allow us to discriminate whether it traces material ejected from the central star (jet) or orbiting around it (disk or independent object).

In order to improve the S/N of the detected intensity, we have obtained the averaged radial intensity profiles of the 7 and 9 mm images (see Figure 4). These profiles were produced by averaging the intensity within concentric elliptical rings, matching the inclination and position angle of the disk major axis determined by previous molecular line observations (i = 13° and PA = 5°; Raman et al. 2006). The width of the concentric rings was set to the size of the beam, since the rms noise at spatial scales smaller than a beam is not independent.

Zoom In Zoom Out Reset image size

Besides the inner gap (hereafter G1), two other gaps at radii ∼038 (∼45 au; G2) and ∼073 (∼85 au; G3) are revealed at both 7 and 9 mm. The inner (G1) and second gaps (G2) were already detected at near-IR wavelengths (with G1 at radii ≲019 and G2 extending from ∼028 to ∼048; Quanz et al. 2013; Momose et al. 2015), at 7 mm (Osorio et al. 2014), and at 1.3 mm (Fedele et al. 2017). Our observations detect the G1 and G2 gaps at the same radii as precedent studies, while the G3 gap is reported here for the first time.

We point out that a sharp cutoff in the uv coverage of the observations can result in the creation of a spurious annular structure during the deconvolution process. However, this artifact would appear at different radii in the images at different frequencies, whereas the G3 gap appears at the same radius in our observations at 7 and 9 mm. Additionally, our simulated images (see Section 3.1) do not show the presence of spurious features mimicking the G3 gap when an actual gap is not included in the model. Therefore, we conclude that the G3 gap is probably a real annular gap in the disk of HD 169142. We note, however, that the G3 gap is not visible in the ALMA 1.3 mm images (Fedele et al. 2017). This could indicate that the G3 gap is more prominent in the distribution of the centimeter-sized dust grains traced by our VLA observations.

As mentioned above, different studies have proposed that G1 and G2 are probably formed because of dynamical interactions between the disk and forming planets within each gap (Osorio et al. 2014; Reggiani et al. 2014; Momose et al. 2015; Wagner et al. 2015; Fedele et al. 2017). However, the origin of the third gap (G3) is more difficult to understand. Detection of gaps at such large distances, like G3, is very difficult. So far, similar gaps have only been detected in the protoplanetary disks around HL Tau (ALMA Partnership et al. 2015), TW Hya (Andrews et al. 2016), and HD 163296 (Isella et al. 2016), through recent extremely high angular resolution ALMA observations. The density in the disk midplane at such large distances is probably too low to create a planet responsible for clearing the observed gap. Alternatively, the MRI in magnetized disks can produce pressure bumps at the outer edges of the dead zones in the disk, which can in turn trap the large dust grains and create gaps in the millimeter emission of the disk (Flock et al. 2015). However, dead zones are expected to be closer to the central star (at radii ∼50 au), so models do not predict gaps as far as the observed G3 gap in HD 169142.

Another possible origin for G3 would be grain growth and/or an increase in the solids' surface density close to condensation fronts (i.e., snowlines) of volatiles in the disk. This process has been suggested as the one responsible for some of the rings and gaps that were detected by ALMA in the disk around the younger T Tauri star HL Tau (ALMA Partnership et al. 2015; Zhang et al. 2015; Okuzumi et al. 2016). Models and laboratory experiments suggest that dust grains can grow significantly when surrounded by an icy mantle, which would form on the surface of the grains beyond these snowlines (Ros & Johansen 2013; Testi et al. 2014, and references therein). In addition, it has been suggested that the enhanced surface density of solids beyond the snowline can produce viscosity gradients and pressure bumps because of a reduction in the column depth of the MRI-active layer. These pressure bumps could in turn trap the large dust grains (Kretke & Lin 2007), although more recent studies have found that, at least for the water snowline, which is located at much smaller radii, the viscosity gradient would need to be unrealistically high to be able to form a dust trap (Bitsch et al. 2014).

Based on the composition of comets, the most abundant volatiles in protoplanetary disks are thought to be water, CO, and CO₂. These molecules have, for typical disk midplane densities, condensation temperatures of 128–155 K, 23–28 K, and 60–72 K, respectively (Zhang et al. 2015). Comparing these temperatures with the midplane temperatures obtained from the model presented by Osorio et al. (2014), we find that the water and CO₂ snowlines would fall inside G1, while the CO snowline would be located at a distance of 90–130 au.⁸ The lower end of the range of radii for the CO snowline is coincident with the outer edge of G3, favoring grain growth and an increase in the solids' surface density close to the CO snowline as a possible mechanism to explain the origin of the G3 gap.

An independent measurement of the position of the CO snowline in protoplanetary disks can be obtained through observations of molecules whose chemistry is sensitive to the CO (Qi et al. 2013). One of these molecules is the DCO⁺, which is expected to form mainly in the regions of protoplanetary disks where gas-phase CO and low temperatures (T < 30 K) coexist. Because of this, the DCO⁺ emission has been used as a tracer of the CO freezeout in disks, showing a ring-like morphology peaking at just a slightly smaller radius than the CO snowline (Mathews et al. 2013; Öberg et al. 2015). However, recent studies suggest that, in some cases, the DCO⁺ molecule might have a more complex relationship with the CO snowline, and that optically thin CO isotopologues such as C¹⁸O could represent better tracers by showing a decrease of emission at the radius of the snowline (Qi et al. 2015; Huang et al. 2017).

In order to estimate the position of the CO snowline in HD 169142, we have analyzed ALMA archival observations of both the DCO⁺(3–2) and C¹⁸O(2–1) molecular transitions. An image of the velocity-integrated DCO⁺(3–2) emission is shown in Figure 5, whereas the C¹⁸O(2–1) images are reported by Fedele et al. (2017). We have obtained averaged radial profiles of the velocity-integrated emission for both transitions following the same procedures as for the continuum emission (see Figure 6). The image of the DCO⁺(3–2) emission shows a ring morphology (Figure 5) with radius ∼080 (∼95 au), as is clearly seen in the radial profile shown in Figure 6. Additionally, the C¹⁸O(2–1) emission shows a significant change of slope in its radial profile at ∼085 (∼100 au), indicating a decrease of the gas-phase CO beyond this radius. Thus, both molecular transitions indicate that the CO snowline is located at ∼100 au in the disk of HD 169142, very close to the G3 gap and in the range of distances estimated with the Osorio et al. (2014) model.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

These results suggest that the G3 gap might be produced because of an accumulation of large dust grains at the position of the CO snowline. Large grains could grow and accumulate close to the CO snowline, creating a ring of emission at long wavelengths next to an inner zone with a lower density of large grains, resulting in the observed gap in the 7 mm emission of the disk. As noted above, the 1.3 mm emission of the disk does not show the presence of the G3 gap (Fedele et al. 2017), which could indicate that the larger (∼centimeter-sized) dust grains are being trapped in the CO snowline (Kretke & Lin 2007), whereas the smaller (∼submillimeter/millimeter-sized) grains traced by the 1.3 mm emission have continued their inward migration. Further studies, including both modeling and observations, are needed to fully understand the possible changes in the disk appearance with the observing wavelength.

Therefore, even though we cannot completely discard a magnetically induced or planet-induced origin, we favor dust growth and accumulation of large dust grains close to the CO snowline as the most likely mechanism responsible for the proposed G3 gap in HD 169142.