Discovery of An au-scale Excess in Millimeter Emission from the Protoplanetary Disk around TW Hya

It is widely accepted that protoplanetary disks (PPDs) are the birthplace of planets. Obtaining observational evidence of a forming planet in PPDs is crucial for understanding of the formation and diversity of (exo)planets (Ida & Lin 2004). The detection of a substructure related to a forming planet is a promising way to investigate the planet formation process. Recent high-resolution observations using the Atacama Large Millimeter/submillimeter Array (ALMA) have revealed complex disk substructures, such as multiple, axisymmetric gaps, spiral arms, and lopsided emissions (e.g., DSHARP; Andrews et al. 2018), which are likely related to the planet formation process.

On the other hand, it is theoretically predicted that the forming planet accretes material from a circumplanetary disk (CPD). Such CPDs hold the promise of direct detection with sensitive detectors as a localized small-scale substructure in the PPD, and some theoretical models suggest that CPDs should be detectable at millimeter wavelengths (Zhu et al. 2016). Large-scale nonaxisymmetric substructure in PPDs has been found at millimeter wavelengths for Oph IRS 48 (∼60 au in azimuthal extent; van der Marel et al. 2013), HD 142527 (∼150 au; Casassus et al. 2013; Fukagawa et al. 2013), MWC 758 (∼30 au; Dong et al. 2018), HD 143006 (∼86 au; Pérez et al. 2018), HD 163296 (∼17 au; Isella et al. 2018), and HD 135344B (∼210 au; Cazzoletti et al. 2018). These lopsided substructures are interpreted as the accumulation of dust particles due to a large-scale gas vortex in the PPD. On the other hand, astronomical-unit-scale substructures or nonaxisymmetric components, which are expected to be the signatures of CPDs, have not yet been discovered at these wavelengths (Isella et al. 2014).

TW Hya is the nearest T Tauri star with a distance of 59.5 pc (Gaia Collaboration et al. 2016). The stellar mass is 0.8 M_⊙ and the stellar age is 10 Myr (Andrews et al. 2012). The disk orientation is almost face-on with an inclination of 7° (Qi et al. 2004). High-resolution observations with ALMA have resolved multiple axisymmetric gaps; in particular, deep gaps at ∼25 and 41 au (Andrews et al. 2016; Tsukagoshi et al. 2016; Huang et al. 2018). Evidence for nonaxisymmetry or small-scale substructure has not yet been reported for the PPD around TW Hya.

In this Letter, we report the results of our high-sensitivity observations at Band 6 using ALMA and the first finding of an astronomical-unit-scale substructure at a radius of 52 au in the PPD around TW Hya. We describe our observations and data reduction in Section 2. In Section 3, we show the resulting images of the Band 6 continuum emission, and the finding of the astronomical-unit-scale substructure in the TW Hya disk is presented. In Section 4, we discuss the detailed structure and the expected origin of the substructure.

2.1. Sensitive ALMA Observation at Band 6

2.2. Data Reduction of the Band 7 Archive Data

Figure 1(a) shows the global distribution of the 233 GHz continuum emission. The total flux density from the PPD and the known axisymmetric features, gaps at 25 and 41 au, are consistent with previous works (Andrews et al. 2016; Tsukagoshi et al. 2016; Huang et al. 2018).

Zoom In Zoom Out Reset image size

In the southwest part of the PPD, the resolved excess of emission is discovered at a radius of 52 au, as shown in Figure 1(b). The peak intensity of this feature is measured to be 308.4 μJy beam⁻¹, corresponding to a signal-to-noise ratio of 34. The excess of emission is significant in comparison to the surrounding background emission of the PPD. The excess is clearly detected both in the radial and in the azimuthal profiles as shown in Figure 2. The average of the surrounding emission is measured to be 197.1 μJy beam⁻¹ from position angles of 226°–230° and 244°–248°; hence, the excess is calculated to be 111.5 ± 9.1 μJy beam⁻¹ (∼12σ), i.e., the emission feature is 1.5 times brighter than the surrounding PPD. There is no clear excess of emission above 3σ in this radial region of the PPD except for the emission feature. We also note that no clear gap-like structure is found at 52 au from the central star.

Zoom In Zoom Out Reset image size

Since the emission from the PPD is almost wholly axisymmetric, the emission feature is extracted by subtracting the axisymmetric background emission in the visibility (i.e., u–v) domain. The size, position, and total flux density of the emission feature are measured by model fitting in the u–v domain using the residual visibilities.

The subtraction process works well as shown in Figures 3(a) and (b). To deduce the structure of the emission feature, we performed a least squares fitting to the subtracted visibilities using the uvmodelfit task installed on CASA. A single component described by a Gaussian function was assumed for the model. The modeled visibilities determined by the fitting were converted to an image using CLEAN with the same parameters as for the original image, and the resulting image is shown in Figure 3(c). From the fitting, the peak position of the emission feature is measured to be (Δα, Δδ) = (−726 ± 1, −471 ± 1) mas from the central star, corresponding to (−43.2 ± 0.1, −28.0 ± 0.1) au or a radius of 51.5 ± 0.1 au. The FWHMs of major and minor axes of the fitted Gaussian are measured to be (74.7 ± 3.3) × (16.2 ± 3.2) mas with a position angle of −383 ± 21, corresponding to (4.4 ± 0.2) × (1.0 ± 0.2) au. The total intensity of the fitted Gaussian is 250 ± 5 μJy.

Zoom In Zoom Out Reset image size

The brightness temperature of the emission feature can be converted from the total flux density. Assuming that the millimeter emission of the PPD behind the resolved feature is uniform, the total flux density of the background inside the feature area is estimated to be 122 μJy from the average of the surrounding emission, and thus the summation of the total flux density inside this area is 372 μJy. Therefore, the brightness temperature at the position of the emission feature is converted using the Planck function to be 11.6 K. The brightness temperature is nearly equal to or less than the temperature of the dust disk at 52 au, where ∼14–18 K is expected from the midplane temperature profile (Andrews et al. 2016; Tsukagoshi et al. 2016; Huang et al. 2018), indicating that the emission feature may be partially optically thin.

Assuming optically thin conditions, the mass of the emission feature is estimated from the equation

$$\begin{eqnarray}&&{M}_{\mathrm{dust}}=\displaystyle \frac{{F}_{\nu }{d}^{2}}{{\kappa }_{\nu }B({T}_{\mathrm{dust}})},\end{eqnarray} \tag{ 1 }$$

where F_ν is the integrated flux density, d is the distance to the source, and B(T_dust) is the Planck function for the dust temperature T_dust. We employ a dust mass opacity coefficient κ_ν of 2.3 cm² g⁻¹ at a frequency of 233 GHz, which is determined from κ_ν of 2.8 cm² g⁻¹ at a wavelength of 870 μm (Andrews et al. 2012) and β, the power-law index of κ_ν, of ∼0.5. Assuming T_dust = 18 K (Huang et al. 2018), the total dust mass of the emission feature is (2.83 ± 0.06) × 10⁻² M_⊕, where M_⊕ indicates the Earth mass. It should be noted that the estimated mass depends on the uncertainty of the opacity and could be a lower limit because the emission may not be completely optically thin.

The emission feature can also be seen in a high-resolution 325 GHz ALMA image within close proximity to the location of our 233 GHz emission feature. The 325 GHz continuum image we reconstructed from the archive data is shown in Figures 4(a) and (b), and the subtracted image made by using the same procedure as done for our 233 GHz data is shown in Figures 4(c) and (d). It is clear that there is a local emission peak near the position of the emission feature that we found in the 233 GHz map, while the emission seems to be azimuthally elongated. The positional offset between the emission feature at Band 6 and the residual emission at Band 7 is much less than the proper motion of the TW Hya system. If the excess emission is a background source, the positional offset must be 136 mas for 2 yr in the R.A. direction according to the proper motion of the TW Hya system ((−68.225, −13.934) mas yr⁻¹; Gaia Collaboration et al. 2016). The positional deviation of the residual emission at Band 7 is within only ∼50 mas with respect to the emission feature at Band 6, as shown in Figure 4. Therefore, we conclude that the emission feature we found is situated in the PPD and is likely orbiting the central star. Due to the significant phase noise of the Band 7 data, we conservatively conclude that it is unclear whether there is a positional offset due to the Keplerian motion for 2 yr (see Figure 4(d)).

Zoom In Zoom Out Reset image size

Lastly, we also note that the millimeter flux density fluctuates azimuthally (see the profile at 45 au in Figure 2(b)). The fluctuation might be related to a moving surface brightness asymmetry, which is probably due to disk shadow (Debes et al. 2017), but we do not focus on this structure in this Letter.

We have found a few astronomical-unit-scale, elongated emission feature in the PPD of TW Hya. Similar asymmetric structures have been found in several other PPDs as listed in Section 1, but the feature in the TW Hya disk is the smallest ever discovered.

In many PPDs, the asymmetric features are interpreted as the dust particles trapped in a gas vortex (e.g., Raettig et al. 2015). The morphology of the emission feature we found may also be interpreted along this line. The width of the gas vortex is limited to several times the gas scale height. The radial half width of the emission feature is only ∼0.5 au and is much smaller than the expected gas scale height at this radius, which is ∼4–5 au (Andrews et al. 2012). The emission feature is azimuthally elongated and the ratio of the azimuthal to the radial widths, or aspect ratio, is ∼4. The shape of gas vortices has been investigated by several authors (Lesur & Papaloizou 2009; Richard et al. 2013), and the aspect ratio of stable vortices is of the order of ∼several. The dust particles trapped in a gas vortex are concentrated at the vortex center, but the aspect ratio of the dust distribution is similar to that of gas (Lyra & Lin 2013).

The mean surface density of the emission feature is expected to be ∼2 times higher than that of the surrounding PPD under the assumption that the temperature at 52 au is constant at 18 K (Huang et al. 2018). If the overall dust-to-gas mass ratio of the TW Hya disk is ∼100, the excess emission region may have a dust-to-gas mass ratio of around 50. Such a small overdensity may be realized even with a weak gas vortex.

The gas disk may be full of such "weak vortices" if the PPD is moderately turbulent. A chain of vortices in the same radial location will merge into one vortex (e.g., Ono et al. 2018), but there may be more vortices at different radii. In this sense, the emission feature could be a "tip of the iceberg" of even smaller dust concentrations.

An alternative scenario that may explain why there is only one emission feature is the existence of a planet. If there is a planet accreting gas and dust from the surrounding PPD (e.g., Pollack et al. 1996; Canup & Ward 2002), the temperature and the density may increase locally (e.g., Owen 2014) and a CPD may be formed around the planet. The emission feature may be the remnant of the accretion onto the planet and/or the CPD.

We first make a rough estimate of the planet mass using the radial half width of the emission feature (∼0.5 au), which may be interpreted as the maximum radius of the putative CPD. The size of the CPD is considered to be several times smaller than the Hill radius ${r}_{{\rm{H}}}={r}_{{\rm{p}}}{\left({M}_{\mathrm{pl}}/3{M}_{* }\right)}^{1/3}$, where, r_p is the orbital radius of the planet, M_p is the planet mass, and M_* is the mass of the central star. The exact size of the CPD has recently been a topic of active debate. It has been considered as $\sim {r}_{{\rm{H}}}/3$ (Quillen & Trilling 1998; Ayliffe & Bate 2009) while recent simulations indicate that it may be $\sim {r}_{{\rm{H}}}/10$ or smaller for a low-mass planet like Neptune (e.g., Wang et al. 2014; Ormel et al. 2015; Szulágyi et al. 2018). The size of the emission feature corresponds to ${r}_{{\rm{H}}}/3$ for a Neptune mass planet and r_H/10 for a 30 Neptune mass planet.

Two observational evidences prefer a lower mass planet. First, the existence of a planet more massive than ∼1–2 Jovian mass (=20–40 Neptune mass) at ∼50 au from the central star is ruled out by 3.8 μm (L'-band) observations (Ruane et al. 2017). Second, we do not observe any gap structure at r ∼ 50 au. This requires a low planet mass and/or high viscosity (e.g., Kanagawa et al. 2017). Dipierro & Laibe (2017) discuss that a planet with a mass of $\lesssim \mathrm{several}\times {10}^{-5}\,{M}_{* }$, roughly corresponding to Neptune mass, does not form a significant gap in both the gas and dust distributions, while such low-mass planets may still open a gap in a low viscosity (α ≲ 10⁻⁴) environment (e.g., Dong et al. 2017, 2018). Since only loose upper limits on the α-parameter of viscosity is given observationally ($\alpha \lesssim \mathrm{several}\times {10}^{-3};$ Teague et al. 2016; Flaherty et al. 2018), we consider ∼1 Neptune mass as an upper limit for the planet mass.

However, it should be noted that the observed emission feature may not be fully accounted for by emission from a CPD. We use a simple model by Zhu et al. (2016) to estimate the flux density at millimeter wavelengths from the CPD with a radius of r_H/3 around a Neptune mass planet. Figure 5 shows the flux density of the modeled CPDs for a range of viscous parameters α and mass accretion rates onto the planet $\dot{M}$. If the mass accretion rate is ${10}^{-7}\,{M}_{\mathrm{Nep}}$ per year, the flux density of the CPD is no larger than ∼100 μJy. We consider this as an upper limit because the CPD around a low-mass planet may be much smaller in size, resulting in a much smaller emitting area. This value may be compared with the observed emission within a circle with a diameter of 1 au, which is ∼60 μJy as the total emission is ∼250 μJy within 4.4 × 1.0 au. Therefore, a part of the excess emission may come from the CPD while the entire emission may be accounted for by the overdensity due to a surrounding envelope-like structure of accreting material around the system of the planet and the CPD.

Zoom In Zoom Out Reset image size

In short, in the case of the planet scenario, the allowed planet mass range may be limited to 1 Neptune mass or less. The CPD around the putative planet may account for a part of the excess emission, while it is not possible to explain the entire feature. We note that Dong et al. (2018) suggested a ∼2 Neptune mass planet at 45 au as a cause of gap structures. The putative planet at the location of the emission feature may not coexist with this planet because the orbits are too close together to form a stable system.

With observations at millimeter wavelengths only, it seems difficult to judge whether the hypothetical planet actually exists at the location of the emission feature. The simple CPD model described above suggests that the millimeter emission from the CPD is similar to that of dust concentration within a gas vortex, making it difficult to determine whether the planet exists or not. The most prominent feature of an accreting planet is that the material is heated up to ≳1000 K in the close vicinity of the planet. Therefore, direct imaging observations at infrared wavelengths of the detection of an accretion signature in emission lines may be critical to prove (or rule out) the existence of a planet. Furthermore, high angular resolution observations of molecular line emission may reveal the kinematics of the emission feature, thus providing further evidence for the presence or otherwise of a vortex.

This Letter makes use of the following ALMA data: ADS/JAO.ALMA#2017.1.00520.S and 2016.1.00842.S. We also use the following public ALMA archive data: ADS/JAO.ALMA#2015.1.00308.S, 2015.1.00686.S, 2016.1.00229.S, 2016.1.00311.S, 2016.1.00440.S, 2016.1.00464.S, 2016.1.00629.S, and 2016.1.01495.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. A part of the data analysis was carried out on the common-use data analysis computer system at the Astronomy Data Center of NAOJ. This work was supported by JSPS KAKENHI grant Nos. 17K14244, 17H01103, 18H05441, and 19K03932. T.J.M. thanks STFC for support under grant reference ST/P000321/1. C.W. acknowledges financial support from STFC (grant reference ST/R000549/1) and the University of Leeds.

Facility: ALMA. -

Software: astropy, CASA.