Near-IR Observations of the Young Star [BHB2007]-1: A Substellar Companion Opening the Gap in the Disk

The NACO images reveal the existence of four stars in the field of view (see Figure 1). The astrometric and photometric properties of these four objects are analyzed in Sections 3.2 and 3.3, respectively. The closer object to the SE of [BHB2007]-1 appears for the first time in the early third Gaia release (EDR3; Gaia Collaboration et al. 2021) and has been labeled as VLT 171104.07-272258.8 in Figure 1, following IAU standards. On the other hand, the known object [BHB2007]-2 is resolved here as a binary system, to our knowledge, for the first time. The less luminous star to the west of the binary system is named [BHB2007]-2B. The NIR images also reveal the existence of an inclined disk around [BHB2007]-1, of extended filaments, and of a possible substellar object near [BHB2007]-1 that are analyzed in Sections 3.4 and 3.5. In the K_S band a faint emission perpendicular to the disk plane is detected (see Figure 2, lower left panel). If the feature is real, and not residual light from the spiders, it might be related to a jet. The jet may be evidence that the protostar is actively accreting. Indeed, from the JHK spectra presented in Covey et al. (2010), both Brγ and Paβ lines, indicators of accretion, are seen in emission. The NB_1.64 image is centered at the [Fe ii] 1.64 μm line, which is a tracer of jets in young objects. However, the bad quality of our NACO observations does not allow us to detect any emission from the potential jet.

In Gaia DR2 (Gaia Collaboration et al. 2018), the only star in our full field of view (FoV) with astrometric measurements available was [BHB2007]-2, for which a parallax of π = 5.9225 mas and proper motions of ${\mu }_{{\rm{R}}.{\rm{A}}.}$ = −1.411 mas yr⁻¹ and ${\mu }_{\mathrm{decl}.}$ = −21.629 mas yr⁻¹ were measured. In Gaia EDR3, the parallax measurement for the source [BHB2007]-1 is also given for the first time, π = 4.30 ± 0.52 mas. Proper motions are ${\mu }_{{\rm{R}}.{\rm{A}}.}$ = −3.67 mas yr⁻¹ and ${\mu }_{\mathrm{decl}.}$ = −17.94 mas yr⁻¹. For [BHB2007]-2 the parallax was updated to π = 6.02 ± 0.31 mas, ${\mu }_{{\rm{R}}.{\rm{A}}.}$ = 2.04 mas yr⁻¹, and ${\mu }_{\mathrm{decl}.}$ = −21.95 mas yr⁻¹.

The EDR3 parallax of [BHB2007]-1 (translating into a distance of 232.6 pc) would place the source far behind the Pipe nebulae (at 163 pc; Dzib et al. 2018). However, a few cases of erroneous parallax measurements from Gaia have been reported for young stars (e.g., Mesa et al. 2019; Garufi et al. 2019). The Gaia astrometric excess noise is relatively large (∼4 mas). We checked for the fidelity of the EDR3 astrometric solutions for BHB2007-1 (source ID 4108624199978985984) and BHB2007-2 (source ID 4108624195634062464) by querying the neural network classifier recently made available by Rybizki et al. (2021). The astrometric fidelities are 0.9995 and 0.05053 for BHB2007-2 and BHB2007-1, respectively, meaning that the astrometric solution of the latter source is most likely spurious. Furthermore, the systemic velocity inferred for [BHB2007]-1 by Alves et al. (2020) from CO millimeter lines (3.6 km s⁻¹) matches the systemic velocity of the other members of the Barnard 59 region. At 232 pc the disk and gap sizes resolved in the ALMA dust maps would be considerably larger than typical values for protoplanetary disks (van der Marel et al. 2019). Therefore, we are inclined to consider the Gaia EDR3 parallax erroneous and assume a distance of 166 pc, like [BHB2007]-2.

To assess whether VLT 171104.07-272258.8 (for which no Gaia parallax is available) is physically bound to the system, we exploited the NACO images from two different epochs (∼4 years apart). For both images, we measured the accurate position of the three systems: the central star [BHB2007]-1, VLT 171104.07-272258.8, and [BHB2007]-2 AB, treated for this calculation as a bound binary system for the very close separation of the two components. The positions and relative motion of VLT 171104.07-272258.8, [BHB2007]-2 A, and [BHB2007]-2 B with respect to [BHB2007]-1, which is in the center of the detector in the two images, are listed in Table 1. Given the differences in R.A. and decl. we can say that VLT 171104.07-272258.8 is not comoving with either of the two other systems visible in the detector. If bound, assuming a circular orbit, VLT 171104.07-272258.8 would orbit around [BHB2007]-2 on a period of ∼1700 yr, so we can easily exclude that the difference in R.A. and decl. on a 4 yr baseline is due to orbital motion. On the other hand, [BHB2007]-2 A and B are a physically bound system as they show the same differences with respect to [BHB2007]-1, inside the error bar. Note that one of the images used for this measurement is the narrowband filter NB_1.64, which has a very poor quality, and [BHB2007]-2 B has a low signal-to-noise ratio (S/N). It is then difficult to infer any orbital motion between [BHB2007]-2 A and B, due to the great error bars. This binary is not comoving to [BHB2007]-1; they may have different proper motions (∼10 mas yr⁻¹) if they are located at similar distance from us. The ALMA continuum and molecular maps (CO, C¹⁸O, and H₂CO isotopologues) also do not show another source within the NACO FoV, indicating that both BHB2007-2AB and VLT 171104.07-272258.8 are more evolved than BHB2007-1. In the analysis we also calculated the photometry in the K_S band of BHB2007-2AB and VLT 171104.07-272258.8, listed in Table 2.

Table 2. Apparent Magnitudes of the Components Measured with Respect to the Central Source [BHB2007]-1

Object	J	H	KS	$L^{\prime}$
VLT 171104.07-272258.8	11.95	10.23	9.27	10.71
BHB2007-2 A	10.13	8.80	8.18	9.40
BHB2007-2 B	11.34	10.72	10.08	10.97

Download table as:  ASCIITypeset image

Here we calculate the stellar properties of [BHB2007]-1 and -2 based on the Gaia EDR3 parallax (see Section 3.2) and the notion that [BHB2007]-2 is a binary system. We collected the B-to-W4-band photometry of −1 and −2 from Vizier. We investigated the ASAS-SN light curves (Kochanek et al. 2017) finding that the visual brightness of both stars is only moderately variable ($\lesssim 1$ mag over years). We adopted a Phoenix model of the stellar photosphere (Hauschildt et al. 1999) with the effective temperature constrained by Covey et al. (2010) (${T}_{\mathrm{eff}}=4060$ K and ${T}_{\mathrm{eff}}=3346$ K, for −1 and −2, respectively, with typical uncertainties of ±300 K) and visual extinction ${A}_{{\rm{V}}}$ calculated from the observed colors between the R, I, and J bands, and stellar luminosity L_* from the integrated photospheric model. For [BHB2007]-1, this yielded ${A}_{{\rm{V}}}=\,1.9\pm 1.0$ mag and ${L}_{* }=0.63\pm 0.21\ {L}_{\odot }$. However, the edge-on geometry of the disk (see Section 3.4) may in principle yield an underestimation of the luminosity. In fact, stars with very inclined disks appear on average older (since less luminous) than the other stars (see Garufi et al. 2018).

As for [BHB2007]-2, we first had to consider the visually triple nature of the star (see Section 3.1). All three stars sit within the 2MASS beam, meaning that the observed photometry is a contribution of all of them. We computed the actual J-, H-, ${K}_{S}$-, and W1-band magnitudes through the relative flux ratios measured in our NACO images. We then constrained visual extinction and stellar luminosity of [BHB2007]-2A assuming that the ${T}_{\mathrm{eff}}$ constrained by Covey et al. (2010) refers to this star since it is significantly brighter than the other two. We found that the ${A}_{{\rm{V}}}$ is comparable with that of [BHB2007]-1 (1.5 ± 0.4 mag) while the luminosity is $0.58\pm 0.16\ {L}_{\odot }$.

Finally, we constrained stellar mass M_* and age t of the two sources through multiple sets of pre-main-sequence tracks (Baraffe, MIST, and Parsec; Baraffe et al. 2015; Dotter 2016; Bressan et al. 2012). Bearing in mind the assumptions and the uncertainties described above, we found ${M}_{* }=0.65\mbox{--}0.85\ {M}_{\odot }$ and $t=0.9\mbox{--}5.7$ Myr for [BHB2007]-1, and ${M}_{* }=0.24\mbox{--}0.30\ {M}_{\odot }$ and $t=0.1\mbox{--}0.8\,\mathrm{Myr}$ for [BHB2007]-2. The stellar mass of [BHB2007]-1 is significantly smaller than what was constrained by Alves et al. (2020) from the position–velocity diagram of the CO emission (2.23 ${M}_{\odot }$).

Such a large discrepancy cannot be explained by an erroneous estimate of the stellar luminosity alone since the evolutionary tracks corresponding to stars with ${T}_{\mathrm{eff}}=4000$ K are nearly vertical, making the stellar mass only marginally impacted by the luminosity. A miscalculation of the effective temperature is also an unlikely explanation. In fact, a ${T}_{\mathrm{eff}}\gt 5000$ K would be necessary to reconcile photometric and dynamical mass. Covey et al. (2010) obtained their estimate of 4060 K from near-IR spectral indices of good-quality spectra making such a mismatch very implausible. A possible, and yet speculative, solution could be that the central star is in reality an unresolved binary with nearly equal-mass components (see, e.g., AK Sco; Andersen et al. 1989; Czekala et al. 2015). In such a configuration, the observed flux would be distributed to the two objects. Each component would have the same mass constrained for a single source because stellar tracks are vertical. The system would be calculated to be older than 5 Myr but the aforementioned underestimation of the stellar luminosity due to the disk inclination could place the system back to the young age of the Pipe Nebula.

In Figures 2 and 3, the light scattered off by the disk around [BHB2007]-1 is visible in spatial correspondence with the millimeter continuum emission presented by Alves et al. (2020). The scattered light is detected in each of the four wave bands and is always stronger on the south side. To obtain a meaningful measurement of the disk brightness, we computed the disk-to-stellar light contrast $\phi ={F}_{\mathrm{disk}}\times 4\pi {r}^{2}/{F}_{* }$, where ${F}_{\mathrm{disk}}$ is the flux measured from the PSF-subtracted image (see Section 2), r is the distance from the central star, and F_* is the stellar flux measured from the innermost resolution element of the star in the original intensity images (see Appendix B of Garufi et al. 2017 for details). This procedure reveals that the disk brightness decreases with increasing the wavelength (spanning on the bright side from 17.2% to 6.0%; see Table 3). Nonetheless, the brightness ratio between the southern and northern sides is constant (with an average ratio of 5.6). The implications of this finding are discussed in Section 4.

Table 3. Disk Brightness Relative to the Stellar Brightness

Wave Band	South (%)	North (%)	Ratio
J	17.2 ± 1.6	3.2 ± 0.3	${5.3}_{-0.8}^{+1.2}$
H	12.4 ± 1.2	2.1 ± 0.2	${5.9}_{-1.0}^{+1.1}$
KS	9.3 ± 0.9	1.6 ± 0.2	${5.8}_{-1.1}^{+1.5}$
$L^{\prime}$	6.0 ± 0.7	1.1 ± 0.1	${5.4}_{-1.0}^{+1.3}$

Note. For the disk north and south sides, the maximum disk-to-stellar light contrast (see Section 3.4) is shown. Errors are obtained from the standard deviation around the brightest pixel.

Download table as:  ASCIITypeset image

Besides its brightness, the disk looks similar across all images. It appears with a near edge-on inclination and is evidently flared. The outermost signal is detected at ∼10, which is comparable with the disk extent constrained from the millimeter continuum emission by Alves et al. (2020). On the other hand, our images do not reveal any disk cavity down to the NACO inner working angle of 012 (∼20 au), in contrast with the millimeter cavity of ∼70 au inferred by Alves et al. (2020, see Figure 2 therein). This discrepancy is routinely observed in protoplanetary disks and is further discussed in Section 4.

Zoom In Zoom Out Reset image size

Since the disk is close to edge-on, we are able to measure the height of the scattering surface as a function of radius, for each observed wavelength. We do this by calculating the distance between the peak of the disk signal and the disk midplane. Here, we assume that the midplane is given by the disk PA of $12.5^\circ$, as determined from the ALMA continuum by Alves et al. (2020). We trace the disk surface by fitting a Gaussian as a function of height, for each pixel along the disk PA. We only considered the brighter (south) side of the disk in the calculation. The height profile is then deprojected by the disk inclination by dividing by $\sin (i)$, where $i=75^\circ$ as reported in Alves et al. (2020). The extracted heights as a function of radius are shown in Figure 4. The height profiles can be well approximated by a power law in radius $h(R)\propto {R}^{1+\alpha }$. Assuming that the height of the scattering surface can be taken as an approximation to the vertical scale-height of the disk, the α index would be similar to the disk flaring. A simple fit to the H-band profile yields an α value of ∼1.2. The α value is lower ($\alpha \sim 1$) for J and Ks.

Zoom In Zoom Out Reset image size

In the south of the object ∼4'' below the NACO brightness peak a faint filament is seen in the H-band filter with higher S/N with respect to the other filters, which can be compatible to the signal seen in the ¹²CO by ALMA (see Figure 3 with the overlap of the two images). The confirmation that the filament seen in the NIR is not a NACO artifact came with the Gaia DR3, where signal is detected at the same position with a magnitude of G = 20.77. The filament may also connect the system with [BHB2007]-2, which shares the same distance. Indeed the filament seems to connect the two young protostars.

The $L^{\prime}$ image, which traces thermal emission as well as scattered light signal, shows a bright peak close to the gap, in the south side of the disk. The blob, which appears in the southwest of the primary in Figure 5, is located at the position of the millimeter dust gap. To calculate its position we first had to subtract the contribution from the bright primary star (as described in Section 2). After removing the contribution of the main star it is possible to distinguish the circumstellar disk from the PSF of the companion. Although, some residual light from the circumstellar disk may still be present at the location of the PSF, and the following assumption for its mass is then the upper limit, as its brightness can be overestimated. The candidate companion is located at a distance of 53 au (032), at a position angle of $186^\circ$. This companion, if real, might be responsible for the presence of the gap itself. The flux ratio in $L^{\prime}$ between the object and the primary star is 1.4 × 10⁻², which for the evolutionary models corresponds to an object in a range of 37–47 M_Jup, which is within the mass range estimated by Alves et al. (2020). For that assumption we used the range of values of 0.9–5.7 Myr for the age of the system as presented in Section 3.3, a value of 7.93 (derived from the WISE W1 data) for the $L^{\prime}$ magnitude of the star, and 166 pc as the distance of the system, which may be slightly different from the real value. The models implied are COND and DUSTY from Allard et al. (2001), and we also explored the BEX models (Linder et al. 2019).

Zoom In Zoom Out Reset image size

The companion is visible in the $L^{\prime}$ image only, as in the other filters at the shortest wavelengths the contribution of the disk at close separations is higher than the thermal emission of the companion itself (see Table 3). For example, in the K_s band the same companion would have a contrast of 5.4 mag, and the disk has a comparable brightness at the location of the companion. For each band we checked that the $L^{\prime}$ detection is consistent with the upper limit of the mass. A similar case is the substellar companion around HD 169142 presented in Reggiani et al. (2014) and Biller et al. (2014), detected in the $L^{\prime}$ band but its signal is not recovered at the shortest wavelengths (e.g., Ligi et al. 2018; Gratton et al. 2019). It is also possible that the $L^{\prime}$ emission comes from a circumplanetary disk of a still forming protoplanet. To explore this scenario, Hα observations of the source might detect emission from both the primary and the substellar companion.

A comparison with the VLA 14 mm map presented in Alves et al. (2020) shows faint emission at the location of the candidate companion (contours in Figure 5). This 14 mm detection can be nonthermal emission from a brown dwarf companion. The nature of this detection can only be confirmed with a follow-up measurement. Future observations with NIRC2 at Keck, for example, could test whether the object is bound to the system.