PLANETS AROUND LOW-MASS STARS (PALMS). I. A SUBSTELLAR COMPANION TO THE YOUNG M DWARF 1RXS J235133.3+312720^*

The overall strategy of our survey consists of vetting close visual binaries from our sample and obtaining deep imaging of single young M dwarfs. Binaries are removed for several reasons: (1) moderate-separation binaries (≲40 AU) are likely disruptive to giant planet formation as a result of rapid disk dispersal (Kraus et al. 2012); (2) for giant planets forming in circumbinary disks, very tight binaries can be approximated as single point masses made of the total mass of the binary pair, and therefore cannot be included in our final statistical analysis which is focused on low stellar masses; and (3) wavefront correction in AO systems is generally optimized for single point sources, so binaries will tend to reduce image quality and lower Strehl ratios. Our deep coronagraphic imaging of each target is conducted in angular differential imaging (ADI) mode (e.g., Liu 2004; Marois et al. 2006); these data will be discussed in a future publication.

We imaged 1RXS J2351+3127 AB with Keck-II/Near Infrared Camera 2 (NIRC2) coupled with natural guide star (NGS) AO (Wizinowich et al. 2000) on 2011 June 21 UT and 2011 November 15 UT. The narrow camera setting was used for both epochs, resulting in a field of view of 102 × 102. Our 2011 June data were obtained at an airmass of 1.17 with the Mauna Kea Observatory H-band filter (Simons & Tokunaga 2002; Tokunaga & Vacca 2005). We first acquired short unsaturated images of the primary (reading out the central 192 × 248 pixels instead of the full 1024 × 1024 array) for photometric calibration and to check for binarity. We then obtained five 10 s images using the full array with 1RXS J2351+3127 A placed behind the 06 diameter translucent focal plane mask; 1RXS J2351+3127 B is clearly visible in individual frames (Figure 2). Conditions were photometric and the Differential Imaging Motion Monitor on Canada–France–Hawaii Telescope (CFHT) reported 04 seeing during the observations. Second epoch imaging in 2011 November was performed in the K'-band filter at an airmass of 1.05. Nine short (0.2 s) non-coronagraphic frames of 1RXS J2351+3127 AB were acquired during excellent (∼05) conditions. Our imaging data are summarized in Table 1.

Zoom In Zoom Out Reset image size

Cosmic ray and bad pixel removal, dark subtraction, and flat fielding were performed on each image. The coronagraphic frames suffer from dust features present on the focal plane slide which contains the occulting spot. This slide may not return to exactly the same position with each new setup, so we created two flat frames to remove normal detector and optical inhomogeneities as well as residual dust features from this nonstatic optical component. We obtained dome flats with and without the mask, then generated a "coronagraph" flat by dividing the normal dome flat. After masking out the occulting spot from the coronagraph flat (that is, the region centered on the coronagraph spot was set to unity), we matched the flat to the image through cross correlation, then divided it into the image. Optical distortions were corrected using the distortion solution made available by the Keck Observatory, which was developed by B. Cameron and is accurate to ∼0.2–0.3 pixels across the entire image (Yelda et al. 2010).

Our H-band coronagraphic data preceded deep ADI (which will be presented in a future publication) so the image rotator was turned off for the observations (i.e., the rotator was in "vertical angle mode"), causing 1RXS J2351+3127 B to rotate about the primary by 034 during the short sequence. The images were registered by fitting a two-dimensional elliptical Gaussian to the primary star behind the coronagraph and derotated to a common position angle (PA) using a cubic convolution interpolation. North alignment was performed using keywords stored in the FITS header, taking into account the offset between the AO and NIRC2 detector (+07) and the sky orientation on the detector of +0252 ± 0009 derived by Yelda et al. (2010).

Astrometry was measured for each image using centroid positions of 1RXS J2351+3127 A and B and the NIRC2 plate scale of 9.952 ± 0.002 mas pixel⁻¹ from Yelda et al. (2010). Errors in the separation and PA were computed from Monte Carlo realizations, taking into account uncertainty in the north angle alignment, plate scale, centroid position (assumed to be 0.1 pixel), and distortion solution (0.3 pixel). The adopted values for each epoch are weighted averages from the individual frames. (The uncertainties in separation and PA from individual images are roughly 4 mas and 01, respectively.) Our measurements are summarized in Table 2.

We computed flux ratios for our H-band data using the short images of the primary and the longer 10 s coronagraphic frames where the companion is visible. We performed aperture photometry at the centroided positions of both components using an aperture radius of 10 pixel and annular sky subtraction, arriving at an H-band flux ratio 5.68 ± 0.04 mag and a K'-band flux ratio of 5.04 ± 0.05 mag.

We obtained a low-resolution 0.8–2.5 μm spectrum of 1RXS J2351+3127 B with IRTF/SpeX (Rayner et al. 2003) in prism mode on 2011 October 14 UT. The 03 slit was used (2 pixels per resolution element), resulting in an average resolving power (R ≡ λ/Δλ) of ∼250 across the spectrum. Atmospheric conditions were good during the observations (DIMM on CFHT reported 05 seeing) with some light cirrus. The slit was oriented perpendicular to the binary PA; although this differed from the parallactic angle, differential chromatic refraction was negligible because of the low airmass (sec z = 1.1). A total of 12 minutes of data were obtained by nodding along the slit in an ABBA pattern (Table 1). Immediately afterward we observed the A0V standard HD 222749 for telluric correction at a similar airmass (sec z = 1.2) and position on the sky. Internal flats and arc frames were taken for flat fielding and wavelength calibration. The spectra were extracted, median combined, and corrected for telluric features using the IDL package Spextool (Cushing et al. 2004; Vacca et al. 2003). The median signal-to-noise ratio (S/N) per pixel between 0.8–2.5 μm is 74 and reaches over 130 in the J band.

The spectrum of 1RXS J2351+3127 B suffers from significant contamination from the primary at λ  ≲ 1.2 μm, which was evident in the pair-subtracted images. However, the smaller point-spread function (PSF) FWHM at longer wavelengths resulted in a better separation at H and K bands, so contamination should be negligible in those regions. We flux calibrated the spectrum using the K-band photometry from Keck and the conversion to the MKO system described in Section 3.2.

We observed 1RXS J2351+3127 AB with the IRTF/SpeX spectrograph in short cross-dispersed mode (SXD) on 2011 December 2 UT. Conditions were excellent with DIMM on CFHT reporting 04 seeing. We used the 05 slit (R ∼ 1200) oriented at the binary PA so that both the primary and companion were observed. The observations were taken in an ABBA pattern over an airmass range from 1.02 to 1.06. We obtained twenty-four 120 s exposures, resulting in a total on-source integration time of 48 minutes. The A0V star 7 Tri was then targeted for telluric correction, and calibration frames were taken at the same telescope position.

The data were reduced with Spextool. It was clear from the collapsed spatial profiles of each order that the wing of the PSF from the primary overlapped with the companion. As in the prism data, the system was better separated at H and K because of the smaller FWHM, but contamination became progressively worse at ≲ 1.1 μm. As a result, we processed the spectra of the companion without optimal extraction and sky subtraction, and we limited the extraction to orders 3, 4, and 5 (K, H, and J bands, respectively, or 1.1–2.5 μm). The SXD spectrum appears to be slightly redder than the prism spectrum, which may indicate there is less contamination. (The synthetic J−K (MKO) colors of the SXD and prism spectra are 1.26 mag and 1.17 mag.) The fourth order of the companion fell on a region of the detector where a faint ghost exists. Consequently, a large artifact was present in the reduced spectrum between 1.71–1.75 μm, so we removed this spectral region. The extraction of the primary star and the standard were performed the usual way with Spextool using optimal extraction and background subtraction. The median S/N per pixel of the companion spectrum is 26 and is over 40 in H and K.

We obtained additional relative photometry of 1RXS J2351+3127 AB with the guider camera on IRTF/SpeX on 2011 Dec 2 UT. We imaged the system in the Y,⁸ J, H, and K_S filters in nodded patterns. The details of our observations are listed in Table 1. Each frame was corrected for bad pixels and cosmic rays, pair subtracted, and divided by a flat frame created from the science data after masking out the system. The images were then registered and stacked to extract relative photometry. The companion is clearly visible in all filters and sits in the wing of the primary's PSF (Figure 3). We performed relative photometry by modeling each PSF as the sum of three elliptical Gaussians as described in Liu et al. (2010a). Uncertainties were computed by inserting and extracting artificial companions at same separation as the real object but different PAs. The artificial companion was created by scaling the primary star to the same brightness as the real companion. The resulting photometry is presented in Table 3, where quoted uncertainties represent the standard deviation of several hundred realizations of this process. Our IRTF K-band photometry disagrees with our Keck measurements by ∼4σ. Since the PSFs overlap in our IRTF data this discrepancy is probably a result of a slight systematic error or underestimated measurement uncertainty in our IRTF photometry. We emphasize that while our IRTF data were obtained in excellent seeing conditions, our photometry and spectroscopy of the companion may contain minor systematic errors as a result of its close proximity to the primary star. Our Keck data taken with AO are expected to be much more reliable.

Zoom In Zoom Out Reset image size

On 2011 December 26 UT we obtained Keck-II NGS-AO 1.18–1.35 μm spectroscopy of 1RXS J2351+3127 B with the OH-Suppressing Infrared Imaging Spectrograph (OSIRIS; Larkin et al. 2006). Unlike our IRTF/SpeX data, the combination of AO and an integral field unit enables resolved spectroscopy of the companion without contamination from the primary star. We obtained three nodded pairs with exposures of 300 s per position with the Jbb filter and the 50 mas plate scale, totaling 30 minutes of on-source data (Table 1). The resulting field of view was 08 × 32 and the resolving power was ∼3800. Immediately following our science observations, we targeted the A0V star HD 78215 to correct for telluric features.

The data were reduced with the OSIRIS data reduction pipeline⁹ and the latest rectification matrix made available by the Keck Observatory. Nodded pairs were used for mutual sky subtraction. The spectra were extracted from the data cubes using 3 pixel (150 mas) circular apertures. The individual spectra were first scaled to a median-combined spectrum, then deviant pixels were removed with a sigma-clipping algorithm at each wavelength using a 3σ threshold, and finally the spectra were combined by computing mean and standard errors at each wavelength. Telluric correction was performed with the xtellcor_general routine in the Spextool spectroscopic reduction package. At its native resolution the final spectrum has a median S/N per pixel of ∼15.

We use the sky coordinates, distance estimate (50 ± 10 pc, see Section 3.2), and proper motion of 1RXS J2351+3127 A along with first epoch astrometry of the candidate companion to predict the separation and PA (or, equivalently, change in right ascension and declination) of a distant background object over time. The results are shown in Figure 4, with shaded errors at each epoch incorporating uncertainties in distance, proper motion, and first epoch astrometry. Our second epoch Keck astrometry is consistent with our first epoch measurements (within 3σ) and rules out the background hypothesis at 7σ, proving the companion is comoving and very likely gravitationally bound to the primary. The second epoch separation is indistinguishable from the expected separation of a background object, but the PA of a background object differs from the second epoch measurement by 099 ±013. Most of the uncertainty in this value is from the error in the background model PA at that epoch rather than the measured PA of the companion.

Zoom In Zoom Out Reset image size

There are several distance estimates to the primary star 1RXS J2351+3127 A in the literature. Reid et al. (2007) use absolute magnitude–optical band strength index relations (M_J-TiO5 and M_J-CaH2) to arrive at an estimated M_J of 7.08 ± 0.34 mag and a corresponding distance of 35.0 ± 5.6 pc. Riaz et al. (2006) use a slightly different M_J-TiO5 relation and arrive at a distance of 50 pc (we calculate an uncertainty of 20 pc using the quoted rms from their relation). Both estimates make use of field relations and assume the star is single. Recently, Schlieder et al. (2012) estimated a kinematic distance of 41.3 ± 3.2 pc for 1RXS J2351+3127 A using the technique described in Lépine & Simon (2009), which assumes it is a member of the AB Dor YMG (see Section 3.3.3).

The late-type companion can also be used to obtain an independent distance estimate for the system. The most extensive compilation of field M, L, and T dwarfs with parallaxes was recently assembled by Dupuy & Liu (2012). They measure an $M_{K_\mathrm{MKO}}$ value of 10.46 ± 0.15 for L0 objects, which is the spectral type we adopt for 1RXS J2351+3127 B (Section 3.4). Our K' contrast measurement must first be converted to K_MKO to use this relation. The synthetic K_MKO − K' and K_S − K' colors from our SXD spectrum of 1RXS J2351+3127 A are <0.02 mag, so we assume K_MKO ∼ K' ∼ K_S for the primary. Our prism spectrum of the companion yields a K_MKO − K' color of −0.09 mag. Taking this into account and using the Two Micron All Sky Survey (2MASS) magnitude of the primary gives a K_MKO-band magnitude of 13.92 ± 0.05 mag and a distance of 49 ± 5 pc.

If the system is young then field relations will underestimate the objects' luminosities and distances. Assuming instead an age comparable to the Pleiades (∼120 Myr), we can make use of photometry of known late-type Pleiades members to infer an empirical SpT–absolute magnitude relation for young objects. Isolating Pleiades members from Bihain et al. (2006) with spectral types between M9 and L1 and using a cluster distance of 133 pc (Soderblom et al. 2005) yields an average $M_{K_\mathrm{MKO}}$ value of 10.0 ± 0.6 mag and a slightly larger distance of 60 ± 17 pc for 1RXS J2351+3127 B. Altogether we adopt a distance of 50 ± 10 pc, with a younger age favoring a larger distance.

3.3.1. X-Ray Activity

Low-mass stars have long been known to exhibit age–rotation–activity relationships spanning their pre-main-sequence and main-sequence lifetimes (Skumanich 1972). They are born with high angular momenta and fast rotation rates, which induce strong magnetic fields and result in active coronal and chromospheric emission (e.g., Feigelson & Montmerle 1999; West et al. 2008). Angular momentum loss through stellar winds slows rotation and diminishes magnetic field strengths over time (Feigelson et al. 2004; Barnes & Kim 2010; Reiners & Mohanty 2012). Observationally, this manifests as an evolution of rotation rates (e.g., Irwin et al. 2011) and X-ray activity (e.g., Preibisch & Feigelson 2005); both are calibrated to young coeval clusters and old field stars but have considerable dispersion for a given age and stellar mass.

1RXS J2351+3127 A was first identified in a large spectroscopic survey by Riaz et al. (2006) to locate new nearby M dwarfs using a combination of 2MASS and the ROSAT All Sky Survey catalogs. They found a high fractional X-ray luminosity of log L_X/L_Bol = −3.02, which corresponds to the saturation limit for low-mass stars (e.g., Delfosse et al. 1998; Pizzolato et al. 2003; Wright et al. 2011). How does this compare to typical values for young clusters? Preibisch & Feigelson (2005) present a comprehensive analysis of the evolution of X-ray activity for various stellar mass bins between the ages of ∼1 Myr and several Gyr. Fractional X-ray luminosities decline over time, but less precipitously for low-mass stars compared to solar-type stars. The median values for 0.1–0.5 M_☉ stars only decrease by ∼0.4 dex from the 1–10 Myr clusters to Hyades ages (∼625 Myr), whereas 0.9–1.2 M_☉ stars vary by ∼1.7 dex over the same time frame. Both stellar mass bins show a subsequent drop of ∼1.3 dex from the Hyades to the field age. The fractional X-ray luminosity for 1RXS J2351+3127 A is well above typical values of even the youngest clusters. The cumulative distributions for low-mass stars (0.1–0.5 M_☉) from Preibisch & Feigelson (2005) indicate that even the most active tail of field objects never reaches values of −3.0, although a more precise age determination is difficult using log L_X/L_Bol alone since the distributions for young clusters overlap.

The cumulative distributions of X-ray luminosities (log L_X) from Preibisch & Feigelson (2005) are somewhat less degenerate than for fractional X-ray luminosities. Using the count rate to flux conversion factor from Fleming et al. (1995) and assuming a distance of 50 ± 10 pc yields log L_X = 29.3 ± 0.2 erg s⁻¹ for 1RXS J2351+3127 A. In Figure 5 (left panel) we compare this to the distribution of luminosities for various populations from Preibisch & Feigelson (2005). 1RXS J2351+3127 A is consistent with the ONC, Pleiades, and the most X-ray luminous members of the Hyades, but is inconsistent with field objects.

Zoom In Zoom Out Reset image size

The ROSAT Position Sensitive Proportional Counter (PSPC) instrument had modest energy resolution, enabling a rough measurement of the X-ray spectrum using hardness ratios as defined in Voges et al. (1999). Stellar X-ray emission softens with age as a result of decreasing absorption by foreground gas at very early ages (≲10 Myr; Neuhäuser et al. 1995) and an evolving coronal gas temperature at intermediate and old ages (∼10 Myr–10 Gyr; Kastner et al. 2003), so hardness ratios can be used as a crude indicator of age. For a comparison sample we searched the RASS Bright Source Catalog (Voges et al. 1999) for X-ray counterparts within 40'' of stars in two age bins: Upper Scorpius members (∼5 Myr) from Preibisch & Mamajek (2008); and AB Dor, β Pic, Columba, Tuc-Hor, and TWA YMG members (∼10–100 Myr) from Torres et al. (2008). In addition, we use the NEXXUS 2 catalog (updated from Schmitt & Liefke 2004; C. Liefke 2012, private communication) for an older sample (∼1–10 Gyr) of field stars with RASS detections, limiting distances to <15 pc. Figure 6 shows the resulting HR1/HR2 distributions; USco members have very hard HR1 values near 1.0, the intermediate age sample has HR1 values between ∼–0.4 and +0.3, and field M dwarfs have values between ∼–0.5 and 0.0. The hardness ratio of 1RXS J2351+3127 A (HR1 = −0.33 ± 0.16) is consistent with both YMG members and M dwarfs in the field.

Zoom In Zoom Out Reset image size

We also compute X-ray luminosities for the USco, YMG, and field populations and show the log L_X versus HR1 distributions in Figure 5 (right panel). Note that not all of the queried objects had X-ray counterparts, and since the RASS is a flux-limited survey the samples and distributions for each age group represent the most X-ray luminous members of each population. Deeper pointed observations of young compact clusters show a much larger spread in X-ray luminosities at a given age, spanning, for example, almost three orders of magnitude for the ∼1 Myr ONC cluster (Preibisch & Feigelson 2005). (These results also suggest that all-sky searches which require X-ray detections to identify nearby young stars are missing a large fraction of young X-ray-faint members.) The X-ray luminosity of 1RXS J2351+3127 A appears to be inconsistent with field M dwarfs, agreeing better with YMG members which have ages of ∼10–100 Myr.

Shkolnik et al. (2009) and Schlieder et al. (2012) compute fractional X-ray fluxes for their samples of X-ray selected targets and derive values of log F_X/F_J = −2.23 and log $F_X/F_{K_S}$ = −2.06, respectively, for 1RXS J2351+3127 A. Both values lie near the saturation limit for early-M dwarfs and are comparable to YMG members with ages less than the Pleiades (∼120 Myr). Shkolnik et al. (2009) obtained two epochs of high-resolution spectroscopy of 1RXS J2351+3127 A and rule out spectroscopic binarity as the source of increased activity. They also found a CaH band strength suggesting low surface gravity but did not detect Li, leading to a likely age of 20–150 Myr.

3.3.2. UV Activity

3.3.3. Kinematics

Both Schlieder et al. (2012) and Shkolnik et al. (2012) independently find that 1RXS J2351+3127 A is a likely member of the AB Dor YMG. Schlieder et al. use the method of Lépine & Simon (2009) to compute a kinematic distance and predict an RV assuming cluster membership. Their kinematic distance for 1RXS J2351+3127 A (41.3 ± 3.2 pc) matches our photometric distance of the system (50 ± 10 pc), and their predicted RV (–14.0 ± 1.3 km s⁻¹) is in excellent agreement with the measured value of −13.5 ± 0.6 km s⁻¹ by Shkolnik et al. The RV of 1RXS J2351+3127 A was measured from two high-resolution spectra (λ/Δλ ∼ 68,000) obtained by Shkolnik et al. (2009) on 2006 August 14 UT and 2006 October 5 UT with the Échelle SpectroPolarimetric Device for the Observation of Stars at the CFHT. Details about the measurements will appear in Shkolnik et al. (2012). In brief, the spectra were cross correlated with an RV standard with a similar spectral type and the RV was derived from Gaussian fits to the cross-correlation functions.

UVWXYZ values for 1RXS J2351+3127 A with respect to the Sun's space motion and position are listed in Table 4. Given the large range of photometric distances for the system, we also compute space motions for distances of 35, 45, 55, and 65 pc, which are plotted in Figure 7 relative to YMGs from Torres et al. (2008). For distances between ∼35 and 50 pc, the kinematics of 1RXS J2351+3127 A agree well with the AB Dor group and sit in the cluster center at ∼45 pc. Distances larger than 55 pc are discrepant with the cluster. Although membership is highly probable, a parallax for the system and verification using cluster convergence-point methods (e.g., Torres et al. 2006; Mamajek 2005; Galli et al. 2012) is essential for unambiguous association.

Zoom In Zoom Out Reset image size

Table 4. Properties of 1RXS J2351+3127 AB

Property	Primary	Secondary
Age (Myr)	50–150a
dphot (pc)	50 ± 10
Proj. sep. ('')	2.386 ± 0.002
Proj. sep. (AU)	119 ± 24
μαcos δ (mas yr−1)	105.9 ± 3.5b
μδ (mas yr−1)	−81.8 ± 5.3b
RV (km s−1)	−13.5 ± 0.6c
U (km s−1)	−10.5 ± 3.0
V (km s−1)	−29.2 ± 3.7
W (km s−1)	−15.4 ± 4.6
X (pc)	−13.6 ± 2.7
Y (pc)	41.2 ± 8.3
Z (pc)	−24.8 ± 5.0
log(LX/LBol)	−3.02d
log(LBol/L☉)	−1.37 ± 0.19	−3.6 ± 0.2
Spectral type	M2.0 ± 0.5e	L0+2−1
Mass	0.45 ± 0.05 M☉	32 ± 6 MJup

Notes. UVWXYZ values are based on the photometric distance estimate. U is positive toward the galactic center. ^aAssumes the system is a member of the AB Dor YMG. ^bUCAC-3; Zacharias et al. (2010). ^cShkolnik et al. (2012). ^dRiaz et al. (2006). ^eShkolnik et al. (2009).

Download table as:  ASCIITypeset image

3.3.4. Age Summary

3.4.1. Low-resolution Prism Spectrum

Our 0.8–2.5 μm SpeX prism spectrum of 1RXS J2351+3127 B shows typical features of late-M- and early-L-type objects (i.e., deep 1.5 and 1.9 μm steam bands; strong 2.3 μm CO absorption; and (blended) Na i, K i, and FeH features in the J band; Cushing et al. 2005) despite clear and significant contamination from the primary star at λ ≲1.2 μm (see Section 2.3). Here we attempt to use the uncontaminated (or minimally contaminated) regions of the spectrum for the purposes of spectral classification. Our comparison spectra come from the SpeX Prism Spectral Library.¹⁰ We tried fitting several spectral regions of 1RXS J2351+3127 B to a sample of 618 published M, L, and T dwarf spectra which were also obtained with IRTF/SpeX in prism mode. The χ² statistic was used as a goodness-of-fit metric, and we ignored measurement uncertainties in the library spectra.

The best-fitting object to the entire 1.20–2.45 μm region is the intermediate-age M9.0 (optical spectral type) brown dwarf LP 944-20 (Figure 8). Optical spectroscopy of LP 944-20 by Tinney (1998) revealed Li absorption, indicating it is a young substellar object (see also Pavlenko et al. 2007). Follow-up studies uncovered evidence for X-ray flaring (Rutledge et al. 2000), quiescent radio emission (Berger et al. 2001), possible photometric variability (Tinney & Tolley 1999), and optical, but not infrared, periodic RV variations (Martín et al. 2006). All of this indicates that LP 944-20 harbors an unusually strong magnetic field for its spectral type. Ribas (2003) found that LP 944-20 is a likely member of the Castor YMG (which includes the well-studied stars Vega and Fomalhaut) based on its space motion. Age estimates for the Castor YMG range from ∼200–400 Myr (Barrado y Navascués 1998; Torres & Ribas 2002; Ribas 2003), i.e., intermediate in age between the Pleiades (125 Myr) and the Hyades (625 Myr).

Zoom In Zoom Out Reset image size

The near-infrared prism spectrum of LP 944-20 itself (Burgasser et al. 2008) reveals only subtle hints of youth. One of the most discriminating features exhibited by young brown dwarfs at low spectral resolution is a triangular-shaped H band (e.g., Lucas et al. 2001; Allers et al. 2007). This feature is readily seen in Figure 8, which shows the spectrum of LP 944-20 compared to the L0 optical standard 2MASS J0345432+254023 (Burgasser & McElwain 2006) and the young M9 TWA brown dwarf 2MASS J11395113–3159214 (Looper et al. 2007). LP 944-20 has an intermediate H-band shape reflecting its adolescent age. This subtle spectral peculiarity is shared by 1RXS J2351+3127 B and suggests a comparable age to LP 944-20 based on this morphology alone. (The 1.4–1.8 μm region of 1RXS J2351+3127 B, which samples the H₂O band depth and the entire H-band region, is also best fit by the spectrum of LP 944-20.) However, since the spectral properties of brown dwarfs at intermediate ages are not well calibrated, it is unclear how quickly these features evolve and therefore what robust age constraints can be obtained from this method.

Separate fits to individual bands produce best-fit spectral types of M9–L1.5. The 1.15–1.34 μm region yields the L0 (optical type: West et al. 2008)/M9 (NIR type: Kirkpatrick et al. 2010) object 2MASS J12490872+4157286. The best match to the 1.50–1.80 μm region is the L1.5 (NIR type: Kirkpatrick et al. 2010) object 2MASS J01472702+4731142. The 2.0–2.45 μm region is best fit by the L0 (optical type: Wilson et al. 2001) object HD 89744 B from Burgasser et al. (2008).¹¹ Finally, we note that the gravity-independent classification index from Allers et al. (2007) yields a spectral type of M8.4^+1.1_−1.0, which is consistent with our previous estimates.

3.4.2. Moderate-resolution SXD and OSIRIS Spectra

In Figure 9 we compare our SXD spectrum to late-M and early-L dwarfs from the IRTF Spectral Library (Cushing et al. 2005). 1RXS J2351+3127 B most closely resembles the L0.5 and L1 field objects. A comparison of individual bands to the templates likewise suggests M9–L1. Like our prism spectrum, there is some contamination from the primary at shorter wavelengths, which results in a slightly bluer spectrum and artificially diminished line strengths. The H and K bands were less contaminated by the primary in the raw SXD data.

Zoom In Zoom Out Reset image size

On the other hand, our OSIRIS spectrum was obtained with AO and the companion was well separated from the primary. The J-band spectrum is shown in Figure 10 relative to M8–L3 field templates. The best overall match is the L2 template, although L1–L3 objects are also good fits. The depth of the 1.25 μm K i gravity-sensitive lines (e.g., McGovern et al. 2004; Kirkpatrick et al. 2006; Allers et al. 2009) do not appear particularly shallow relative to the field objects, which excludes very young ages (≲10 Myr). Recently, Geißler et al. (2012) presented NIR spectroscopy of a new M8 ± 1 companion to the Pleiades member H ii 1348. Their OSIRIS J-band spectrum also lacks shallow alkali lines, which suggests that any impact on these lines from low surface gravity occurs at younger ages, at least at this temperature. Altogether we adopt a spectral type of L0⁺²₋₁ for 1RXS J2351+3127 B. The asymmetric error bars reflect the slightly earlier type suggested by the prism and SXD spectra compared to the OSIRIS spectrum.

Zoom In Zoom Out Reset image size

The mass of 1RXS J2351+3127 B can be estimated from its luminosity and age using substellar cooling models. Assuming a distance of 50 ± 10 pc, the K-band bolometric correction from Golimowski et al. (2004) yields a luminosity of log L/L_☉ = −3.6 ± 0.2 (the uncertainty incorporates intrinsic scatter in the relation, photometric errors, and a spectral type accuracy of one subtype).¹² The age estimate of the system depends on whether it is an AB Dor member, in which case the age of ∼50–150 Myr (see Section 4) can be leveraged from the entire cluster. If it is not a member then the constraints are much poorer, but probably lie between 50–500 Myr based on the high-energy emission from the primary. We arrive at masses of 32 ± 6 M_Jup and 50 ± 11 M_Jup for age ranges of 50–150 Myr and 50–500 Myr, respectively, based on a grid of finely interpolated evolutionary models from Burrows et al. (1997).¹³ Figure 11 shows the influence of the distance estimate on the inferred mass of 1RXS J2351+3127 B, which lies between ∼25–40 M_Jup for distances of 35–65 pc assuming an age of 50–150 Myr.

Zoom In Zoom Out Reset image size

The mass of the primary can be obtained from stellar evolutionary models. 1RXS J2351+3127 A has a spectral type of M2.0 ± 0.5 which corresponds to an effective temperature of ∼3520 K (Drilling & Landolt 2000). At 100 Myr (500 Myr), the models of Baraffe et al. (1998) give a mass of 0.45 M_☉ (0.40 M_☉) for this temperature. Since the mass is only weakly dependent on age we adopt a value of 0.45 ± 0.05 M_☉ for 1RXS J2351+3127 A. Note that we have made use of solar metallicity evolutionary models here (with Y = 0.275 and L_mix = H_P); a non-solar composition and any intrinsic errors in the evolutionary models will result in systematic errors in the inferred mass. We also compute the luminosity of the primary star using the H-band bolometric correction from Casagrande et al. (2008) and the system's photometric distance, which yields log L/L_☉ = −1.37 ± 0.19.