BANYAN. VI. DISCOVERY OF A COMPANION AT THE BROWN DWARF/PLANET-MASS LIMIT TO A TUCANA–HOROLOGIUM M DWARF

The J-band discovery dataset of J0219–3925 was obtained with SIMON on 2013 November 5 at the CTIO, and follow-up H and K_s-band imaging were obtained on 2014 February 2, with the same instrument and telescope. For all three imaging sequences, we employed a 4-point dither pattern along the corners of a $2\prime \times 2\prime$ square. At each dither position, 15 images (J) and five images (H and K_s) were taken. All sequences used a 30 s per-frame exposure time. The images were sky-subtracted using a sky frame constructed from the median combination of all science images taken on that night within that band. Flat-fielding was performed with a flat constructed from images of a flat screen. The astrometric solution was performed using a cross-match of the 2MASS catalog with field stars. All images were registered and median-combined to produce the final science frame.

The contrast ratio between the two components was determined by performing a PSF fitting of the two J0219–3925 components using two isolated bright and nearby field stars ($\lt 2\buildrel{\,\prime}\over{.} 5$, $J\gt 13.5$) as input PSFs. The magnitude of J0219–3925 B was then determined from the 2MASS magnitude of its host and its contrast ratio within each NIR bandpass. The presence of the companion is unlikely to have significantly affected the 2MASS magnitudes of its host, as it contributes only 2%–4% of the total integrated flux and is resolved (∼2 FWHM). Table 2 summarizes the photometric and astrometric properties of both components of the system.

Table 2.  Properties of the J02193925 System

Short Name	J0219–3925	J0219–3925 B
2MASS name	02192210–3925225
${\mu }_{\alpha }$ a (mas yr−1)	107.37 ± 2.27
${\mu }_{\delta }$ a (mas yr−1)	−34.95 ± 1.65
Statistical distanceb	39.4 ± 2.6 pc
Position angle	173.9 ± 02
Separation	3.96 ± 002
Sky-plane separation	156 ± 10 AU
B	18.50	⋯
${{\rm \Delta }}g$ c	$\gt 5.5$
${{\rm \Delta }}i$ c	6.6
R	15.23	⋯
Id	13.42 ± 0.03	⋯
${J}_{2\mathrm{Mass}}$ e	11.381 ± 0.026	⋯
JMKOf	11.321 ± 0.026	15.54 ± 0.10
${H}_{2\mathrm{Mass}}$ e	10.811 ± 0.027	⋯
HMKOf	10.855 ± 0.027	14.63 ± 0.10
${K}_{2\mathrm{Mass}}$ e	10.404 ± 0.025	⋯
KMKOf	10.444 ± 0.025	13.82 ± 0.10
W1g	10.148 ± 0.023
W2g	9.901 ± 0.020
W3g	9.614 ± 0.037
vrad	10.6 ± 0.7g	⋯
$v\mathrm{sin}i$	6.5 ± 0.4g	⋯
SpT	M6 $\gamma$ h	L4 γ
EW Hα (Å)	−7.02	⋯
EW Li6708 (Å)	639.9	⋯
FeHz	1.068 ± 0.001	1.040 ± 0.002
FeHJ	1.07 ± 0.01	1.20 ± 0.02
Ki 1.169 μm	0.17 ± 0.08	0.6 ± 0.4
Ki 1.177 μm	1.25 ± 0.06	0.71 ± 0.32
Ki 1.244 μm	⋯i	3.0 ± 0.5
Nai 1.138 μm	2.58 ± 0.08	2.4 ± 0.5
H-Cont	0.995 ± 0.001	0.963 ± 0.001
VOz	1.001 ± 0.001	1.372 ± 0.004

Notes.

^aGirard et al. (2011). ^bSee Gagné et al. (2014c) for a discussion on the definition and the accuracy of the statistical distance. ^cOnly contrasts were derived in g and i, and not magnitudes measurements as the images were not taken under photometric conditions. ^dDENIS Consortium (2005). ^eSkrutskie et al. (2006). ^fThe 2MASS to MKO transform has been determined for J0219–3925 from its FIRE spectrum. The companion's contrast ratio has been determined with SIMON that uses MKO filters. ^gCutri et al. (2013). ^hKraus et al. (2014) give a M5.9 spectral type from SED fitting and a spectroscopic spectral type of M4.9. ⁱThe spectrum of J0219–3925 B around the KI 1.244 μm is strongly affected by bad pixels on the science array.

Download table as:  ASCIITypeset image

Upon discovery of a possible faint companion to J0219–3925, we noticed that images of that field usable for science were present in the Gemini science archive (see Figure 1). GMOS-South (i and g band) and Flamingos-II (J band) spectroscopic acquisition images of the host star have been obtained as follow-up observations of the BANYAN All-Sky Survey (BASS; Gagné et al. 2015). The ${{\rm \Delta }}i=6.22$ contrast and non-detection in g (${{\rm \Delta }}g\gt 5.5$) pointed toward an object much redder than the mid-M host, prompting for a dedicated spectroscopic follow-up of the companion. Inspection of the K-band 2MASS images taken in 1999 shows that the companion is marginally detected and provided valuable constraints on the common proper motion of the pair (see Section 4.4 and Table 3).

Zoom In Zoom Out Reset image size

Table 3.  Constraints on Properties from Models

	J0219–3925 A	J0219–3925 B
Mass MJ)	114 ± 13 ${M}_{\mathrm{Jup}}$	13.0 ± 0.7 ${M}_{\mathrm{Jup}}$
Mass (MH)	108 ± 11 ${M}_{\mathrm{Jup}}$	13.9 ± 1.1 ${M}_{\mathrm{Jup}}$
Mass (MK)	115 ± 13 ${M}_{\mathrm{Jup}}$	14.8 ± 1.6 ${M}_{\mathrm{Jup}}$
Mass (mean)	113 ± 12 ${M}_{\mathrm{Jup}}$	13.9 ± 1.1 ${M}_{\mathrm{Jup}}$
q (MJ)	0.113 ± 0.007
q (MH)	0.128 ± 0.006
q (MK)	0.127 ± 0.007
q (mean)	0.123 ± 0.006
T${}_{\mathrm{eff}}$ (MJ)	3070 ± 73 K	1615 ± 41 K
T${}_{\mathrm{eff}}$ (MH)	3047 ± 84 K	1686 ± 43 K
T${}_{\mathrm{eff}}$ (MK)	3074 ± 72 K	1746 ± 49 K
T${}_{\mathrm{eff}}$ (mean)	3064 ± 76 K	1683 ± 43 K
log Luminosity (MJ)	−2.22 ± 0.06 ${L}_{\odot }$	−3.92 ± 0.03 ${L}_{\odot }$
log Luminosity (MH)	−2.26 ± 0.06 ${L}_{\odot }$	−3.84 ± 0.05 ${L}_{\odot }$
log Luminosity (MK)	−2.22 ± 0.06 ${L}_{\odot }$	−3.76 ± 0.06 ${L}_{\odot }$
log Luminosity (mean)	−2.23 ± 0.06 ${L}_{\odot }$	−3.84 ± 0.05 ${L}_{\odot }$
$\mathrm{log}\;g$ (MJ)	4.59 ± 0.06 cm s−2	4.23 ± 0.02 cm s−2
$\mathrm{log}\;g$ (MH)	4.59 ± 0.06 cm s−2	4.24 ± 0.04 cm s−2
$\mathrm{log}\;g$ (MK)	4.59 ± 0.06 cm s−2	4.25 ± 0.06 cm s−2
$\mathrm{log}\;g$ (mean)	4.59 ± 0.06 cm s−2	4.24 ± 0.04 cm s−2
Deuterium (MJ)	0	0.77 ± 0.06
Deuterium (MH)	0	0.68 ± 0.07
Deuterium (MK)	0	0.61 ± 0.09
Deuterium (mean)	0	0.69 ± 0.07
Radius (MJ)	2.75 ± 0.14 R${}_{\mathrm{Jup}}$	1.41 ± 0.02 R${}_{\mathrm{Jup}}$
Radius (MH)	2.68 ± 0.12 R${}_{\mathrm{Jup}}$	1.44 ± 0.04 R${}_{\mathrm{Jup}}$
Radius(MK)	2.76 ± 0.14 R${}_{\mathrm{Jup}}$	1.46 ± 0.04 R${}_{\mathrm{Jup}}$
Radius (mean)	2.73 ± 0.13 R${}_{\mathrm{Jup}}$	1.44 ± 0.03 R${}_{\mathrm{Jup}}$

Download table as:  ASCIITypeset image

On 2014 February 13, we obtained NIR spectroscopic observations with the Folded-Port Infrared Echelette (FIRE; Simcoe et al. 2013) at the Magellan 6.5 m telescope. FIRE was used in its high-resolution echellette mode, providing a spectral resolution $R\sim 5000$ continuously over the entire NIR domain (0.8–2.4 μm). We used a slit width of $0\buildrel{\prime\prime}\over{.} 6$ under a $1\buildrel{\prime\prime}\over{.} 2$ seeing and an airmass of 1.3–1.6 (J0219–3925 B) and 1.8–1.9 (J0219–3925 A). The reference star for both objects was observed at the same airmasses. For J0219–3925 B, we used an ABBA dither pattern for improved sky subtraction on this relatively faint target, while for the brighter J0219–3925 A and reference star, the two exposures were taken with an AB dither pattern.

Data reduction was performed using the standard FIREHOSE pipeline. Flat-field correction was performed using flat frames derived from dome flat images and telluric absorption at the time observation was derived from an A0V star (HD 17683) observation taken immediately before (J0219–3925 B) and after (J0219–3925 A) the science integration. The spectrum of the host star and its companion are compared to field and low-gravity objects of comparable spectral types in Figure 2. The spectrum of J0219–3925 B has an average signal-to-noise ratio of ∼30 when sampled at an R = 800 resolution.

Zoom In Zoom Out Reset image size

J0219–3925 has been identified as a member of the THA by two teams independently. First, it is part of the 129 new late-type (K3 and later) THA members identified by Kraus et al. (2014); their membership being established by Li and RV measurements, but lack parallaxes and are based on spectroscopic distances. Spectroscopic fitting of optical spectrum and overall spectral energy distribution (SED) of J0219–3925 respectively lead to a spectral type estimate of M4.9 ± 1.0 and M5.9 ± 0.3; these values are consistent with the NIR spectral type of M6 γ determined here (see Section 4.3). The radial velocity they measure (10.6 ± 0.7 km s⁻¹) agrees within 0.71 km s⁻¹ with the expected radial velocity of a THA member in that line of sight. This is below the internal dispersion of THA velocities (∼1 km s⁻¹) and well below the cut off in velocity difference between members and non-members in their analysis (±3 km s⁻¹).

The host star J0219–3925 was identified independently as a candidate member of THA as part of the BASS survey. 2MASS and AllWISE photometry and astrometry were used to derive its membership probability to several moving groups in the Solar neighborhood. The BANYAN II tool compare the input parameters to spatial, kinematic, and photometric models of young moving groups and the field population using a naive Bayesian classifier. We refer the reader to Gagné et al. (2014c) for details on this analysis. As J0219–3925 had been identified as a high probability candidate of THA, it was included in our search for distant companions before the publication of the bulk of the survey.The BANYAN II tool gives a P = 99.34% probability that J0219–3925 is a member of THA. When such measurements are available, the BANYAN II tool can use the radial velocity and/or parallax to constrain membership probability further. Using the radial velocity of 10.6 ± 0.7 km s⁻¹ measured by Kraus et al. (2014), the THA membership probability of the host star increases to P = 99.94%.

We can set additional constraints on the age of this object since it displays lithium (Kraus et al. 2014) and typical signs of low-gravity in its NIR spectrum (see left panel of Figure 2 and Section 4.3). Within that spectral type, the presence of lithium (EW $=\;639.9$ mÅ) points to an age below ∼125 Myr. Adding this age constraint in the BANYAN II analysis raises the THA membership probability further (P = 99.93%). In addition to a membership likelihood, the BANYAN analysis provides a kinematic distance to an average precision of $\sim 10\%$ for young moving group members (see Figure 5 in Malo et al. 2013 and Figure 8 in Gagné et al. 2014c). The kinematic distance for J0219–3925 is 39.4 ± 2.6 pc. The 7% uncertainty includes both the contribution from the uncertainties on the proper motion of J0219 and the space velocity scatter of THA members.

As an additional confirmation, we used J0219–3925 apparent I_c and J magnitudes, the amplitude of the proper motion, and the sky position to verify its membership probability in the BANYAN I analysis (Malo et al. 2013). It yields an a priori probability of P = 96.3% of membership in THA. Similarly to the BANYAN II analysis, it is possible to include the radial velocity as an additional constraint in the analysis. When we do so, the probability is increased to ${P}_{\nu }\;=\;99.9\%.$ The statistical distance inferred by this analysis is the same as predicted by BANYAN II.

The Tucana and Horologium associations were discovered independently by Zuckerman & Webb (2000) and Torres et al. (2000). Further studies confirmed that what was initially considered as two associations were more likely to constitute a single young moving group to be called the THA (Zuckerman et al. 2001). THA is part of the so-called Great Austral Young Association (Torres et al. 2008), including the Columba and Carina associations. Several studies (Zuckerman & Song 2004; Torres et al. 2008; Kiss et al. 2011) proposed members of THA based on their galactic space velocity, galactic position, and signs of youth, yielding a sample of members with spectral type between A1V and early-M spawning a distance range of 36–71 pc. Recent studies (Malo et al. 2013; Moór et al. 2013; Rodriguez et al. 2013; Gagné et al. 2014a, 2014b, 2014c; Kraus et al. 2014) proposed more than 200 strong THA candidate members, which still require more robust kinematics to confirm their membership.

The age range of THA was estimated to be 10–40 Myr (Zuckerman & Webb 2000; Zuckerman et al. 2001) based on various age indicators (Hα, lithium, HR diagram). More recently, Kraus et al. (2014) derived an average isochronal age of 30 Myr using a new sample of 142 candidate members combined with the BCAH models (Baraffe et al. 1998).

In addition to the isochronal age, the lithium depletion boundary (LDB) is a key indicator to determine the age of the low-mass star population. This method was used by Kraus et al. (2014) to determine the THA lithium depletion age of 41 ± 2 and 38 ± 2 using the BCAH and D'Antona & Mazzitelli (1997) models, respectively.

This discrepancy between isochronal and LDB ages was already demonstrated by several studies (Song et al. 2002; Yee & Jensen 2010; Binks & Jeffries 2014). Recently, Malo et al. (2014a) have shown that using new Dartmouth Magnetic evolutionary models (Feiden & Chaboyer 2013), the isochronal age is in better agreement with the LDB age for young low-mass stars. In general, isochronal ages are revised upward with the inclusion of magnetic fields. As shown in Malo et al. (2014b), magnetic field strength measurements from high-resolution spectroscopy would further constrain its age. Pending such measurement, we conservatively adopt an age range of 30–40 Myr for the THA association, and thus, for J0219–3925 system.

We used the method of K. Cruz et al. (2015, in preparation; see Cruz & Núñez 2012) to assign a spectral type to both components of J0219–3925. This method consists of a visual comparison with a grid of spectral templates while normalizing each NIR band individually. Inspecting the slope and shape of several features in each band allows to choose a template that best matches the observations. Our sequence of field, intermediate-gravity, and low-gravity templates were build by median-combining various objects within each spectral types and gravity class. The spectra used to build the templates were obtained from Allers & Liu (2013) and the SpeX Prism Spectral Libraries.⁵ Both the visual comparisons of J0219–3015 and b yielded best matches to very-low gravity templates; J0219–3015 and its companion were assigned a spectral type of M6 γ and L4 γ respectively. The gravity classification scheme of Allers & Liu (2013) was subsequently used to confirm that both objects have weaker alkali lines compared to field dwarfs of the same spectral types, and that J0219–3015 B has stronger VO absorption and a triangular-shaped H-band continuum (which was already apparent from the visual comparison). Both objects were consistently categorized as very-low gravity dwarfs by this index-based classification scheme. We show that they both display typical spectroscopic signatures of a low-gravity, including lower-than-normal alkali (K i, Na i, and FeH) equivalent widths. The lower gravity causes a lower pressure in the atmosphere, which decreases both the effects of pressure broadening (responsible for the lower alkali equivalent widths) and collision-induced absorption (CIA) of the H₂ molecule. The triangular H-band continuum is shaped by water absorption. In the case of field brown dwarfs, the H₂ CIA redirects part of the flux to the bluer side, which masks the triangular shape of the H-band continuum (see Rice et al. 2010 for further detail). Figure 3 illustrates the gravity-sensitive spectroscopic indices defined by Allers & Liu (2013) for both components of the J0219–3925 system. Figure 3 illustrates the gravity-sensitive spectroscopic indices defined by Allers & Liu (2013) for both components of the J0219–3925 system; numerical values for these indices are given in Table 2.

Zoom In Zoom Out Reset image size

The intermediate gravity and very-low gravity classifications generally correspond to the β and γ gravity classifications in the optical (Kirkpatrick 2005; Kirkpatrick et al. 2006; Cruz et al. 2009), hence we choose here to use the Greek-letter nomenclature even though our gravity classification was done in the NIR.

Astrometric measurements were performed on GMOS-S i band, F2 J band, and SIMON imaging obtained on 2013 November 5. SIMON imaging obtained in early 2014 were not included as they were taken under poor seeing conditions ($\sim 2\prime\prime$) and do not constrain proper motion. Archival 2MASS imaging was also used; it provides only modest constraints on the PA and separation, but the long time baseline (∼15 years) makes it a useful addition to confirm common proper motion. Astrometric errors were estimated to be at the $\sim 0\buildrel{\prime\prime}\over{.} 03$ level for F2 and GMOS-S data and $\sim 0\buildrel{\prime\prime}\over{.} 1$ for SIMON imaging. Astrometric measurements are shown in Figure 4. The ${\chi }^{2}$ for the co-moving case is 5.9 for 8 degrees of freedom. The co-moving case is equivalent to a $0.9-\sigma$ event in a Gaussian distribution while the background object model would correspond to a $6.0-\sigma$ event. While this demonstrates that J0219–3925 B is comoving to within astrometric uncertainty, any interloper would be an L dwarf at roughly the distance of J0219–3925 with a significant proper motion that could, conceivably, match that of J0219–3925 within astrometric uncertainties. The strongest argument in favor of the two objects forming a physical pair arises from the relative rarity of field L dwarfs per sky surface unit.

Zoom In Zoom Out Reset image size

Finding a low-gravity L dwarf within $4\prime\prime$ of an unrelated young M star is very unlikely. Considering that we started with a sample of 300 young stars, the area within a 2–6'' annulus around these stars covers $5.6\times {10}^{-8}$ of the entire celestial sphere. With an L dwarf spatial density of $\sim 3.8\times {10}^{-3}$ pc⁻³ (Cruz et al. 2007), there should be ∼5500 L dwarfs within 70 pc. The likelihood of finding a coincident L dwarf to a young star in our sample is about $\sim 3\times {10}^{-4}$ without considering the fact that the L dwarf itself also displays a low surface gravity.

Orbital motion of the pair is expected to be close to the detection threshold. With a total mass of ∼0.12 ${M}_{\odot }$ (see Table 3 apj514376t3) and a separation of ∼160 AU, the orbital period is on the order of 6000 years. Assuming a face-on orbit, this leads to an orbital motion of 4 mas yr⁻¹. Orbital velocity is on the order of 800 m s⁻¹ and within reach of upcoming high-resolution infrared spectrographs such as VLT/CRIRES+ (Follert et al. 2014). Both measurements are challenging but possible with existing facilities, but not with the dataset in hand. Proper constraints on the differential motion within the system will allow one to constrain whether the orbit of J0219–3925 B is consistent with a circular orbit or a highly eccentric one. Constraints on orbital eccentricity for a few systems similar to J0219–3925 could set strong limits on the plausible formation mechanisms. These measurements would provide constraints on the J0219–3925 system similar to the ones obtained for a few similar companions in the planetary-mass regime by Ginski et al. (2014). These authors find that fitted orbits suggest that distant companions' motion is consistent with eccentric orbits; whether this holds for the J0219–3925 system remains to be determined.

To estimate fundamental parameters of mass, effective temperature, gravity, and radius for both J0219–3925 and J0219–3925 B we used two distinct approaches with the same set of theoretical models. First, we constrained the bulk properties of J0219–3925 A and J0219–3925 B (mass, effective temperature, surface gravity, radius, and luminosity) by comparing the absolute magnitudes in each NIR bandpass with BT-Settl models⁶ (CIFIST 2011 opacities; Allard 2014) at 30 and 40 Myr. Absolute magnitudes were estimated from the kinematic distance; uncertainties in the kinematic distance were assumed to be 7% (see Section 4.2), leading to a 0.14 mag uncertainty on the absolute magnitudes. Table 3 compiles the value derived for each photometric bandpass, and the mean value for all three bandpasses is used as the best estimate. Upper and lower bounds are averaged for this best estimate, but not divided by the square root of the number of measurements as these are correlated (i.e., same age assumption, same uncertainty on distance). Overall, the mass estimate range of 12–15 ${M}_{\mathrm{Jup}}$ sets the companion at the brown dwarf/planet regime limit. Uncertainties for log g values are small (0.02 and 0.06, respectively) as the surface gravity varies very little with effective temperature for these relevant masses and the main contribution to this uncertainty arises from the uncertainty on the age of the system. The relative uncertainty on the mass ratio (q) is smaller ($\sim 6\%$) than the relative uncertainty on either component's mass ($\sim 10\%$) as the derived masses for both components correlate through the distance and age estimates.

The second approach to constrain the properties of J0219–3925 B has been to perform a ${\chi }^{2}$ fit between the observed and theoretical spectra in three individual bandpasses ($Y+J$, H, and K_s; respectively 1.00–1.33, 1.45–1.81 and 1.95–2.40 μm). For each bandpass, spectra were normalized over part of the domain (horizontal lines in Figure 5). By performing this normalization, we assume no prior knowledge of distance and absolute magnitude. The spectral fitting analysis has been performed by steps of 0.5 dex in $\mathrm{log}\;g$ between 3.5 and 5.5, and over the 1500–1800 K domain. Models at intermediate gravity (4.5) better fit the SED. Within the H band (middle columns), the higher-gravity (5.5) models show the distinctive flattening of the SED, which is the hallmark of old, field, L dwarfs. Within the K band (right), there is a clear shift of the SED peak from ∼2.1 to ∼2.25 μm between $\mathrm{log}\;g=5.5$ and $\mathrm{log}\;g=3.5$, for all temperatures, with 2M0219–3925 B being intermediate between these scenarios. For $Y+J$ and K, the best-fitting model has a temperature of 1700 K and $\mathrm{log}\;g=4.5$. The value derived for H is only marginally different with ${T}_{\mathrm{eff}}=1600$ K and $\mathrm{log}\;g=4.0$. These values are in very good agreement with those derived from photometry alone (see Table 3), with ${T}_{\mathrm{eff}}=1683\pm 43$ K and $\mathrm{log}\;g=4.24\pm 0.04$.

Zoom In Zoom Out Reset image size

Overall, the mass of J0219–3925 B falls squarely at the upper limit of the International Astronomical Union definition for an exoplanet—consisting of an object with a true mass below the limiting mass for thermonuclear fusion of deuterium—and is in orbit around a star, indeed J0219–3925 has a mass estimate above the hydrogen burning limit for plausible ages of THA. Evolution models predict that J0219–3925 B will have retained ∼70 % of its initial deuterium by 40 Myr. As it is expected to have partially burned its deuterium, and considering its brightness, it would constitue a good target to spectroscopically test the brown dwarf/planet boundary, that is expected to be moderately metallicity-dependent (Spiegel et al. 2011) and its exact predicted location varies by ∼1 ${M}_{\mathrm{Jup}}$ depending on atmosphere models (Saumon & Marley 2008). The exact nomenclature for such an object is still to be properly defined, and it joins a number of other objects such as J0103–5515(AB)b and J0122–2439 B (Bowler et al. 2013; Delorme et al. 2013) at the upper limit of planethood. As J0219–3925 B is expected to have burned some of its deuterium and formally lies just above the deuterium-burning limit, we use the "B" designation for stellar and brown dwarf companions rather than the "b" in usage for planetary companions even though it is predicted to be slightly less massive than objects initially described as planetary companions (e.g., AB Pictoris b).