HUNTING THE COOLEST DWARFS: METHODS AND EARLY RESULTS

The unique spectroscopic properties of the lowest luminosity brown dwarfs led to the introduction of two new spectral classes, L dwarfs and T dwarfs (Kirkpatrick 2005). Some wide-field searches for low temperature T dwarfs and Y dwarfs are currently in progress, and several objects with T ∼ 500–700 K have been discovered (Warren et al. 2007; Delorme et al. 2008, 2010; Burningham et al. 2009; Eisenhardt et al. 2010; Lucas et al. 2010; Leggett et al. 2010; Mainzer et al. 2011; Burgasser et al. 2011). Although not fully characterized, the recent discovery of two cool brown dwarfs may indeed have been the first examples of the Y type (Liu et al. 2011; Luhman et al. 2011; Rodriguez et al. 2011b). Additionally, seven ultracool brown dwarfs have recently been discovered with the Wide-field Infrared Survey Explorer, six of which are confirmed spectroscopically to have spectral types of Y0 or later (Cushing et al. 2011). Characterizing these Y dwarfs will be challenging because it is very difficult to obtain reliable physical parameters such as age, distance, and metallicity, for such objects. Lowest luminosity brown dwarfs discovered as companions of stars, however, will have better constrained physical parameters since they can be inferred from the host star. For this reason, any found companion can be an important benchmark for use in the study of substellar objects. To this end we have initiated an imaging survey to identify the coolest substellar objects as companions to nearby stars.

There will be a delay between formation of a substellar mass companion and when it has cooled to 500 K. The lower the mass, the faster 500 K can be reached. Thus, it is important to consider the masses of the lowest mass objects formed in binary systems and accessible to imaging searches. Zuckerman & Song (2009) show that the current distribution of separations and masses of imaged companions with the least mass can be accounted for by using the minimum Jeans mass fragmentation of an interstellar cloud description from Low & Lynden-Bell (1976). They show that for a fragment to split, it must have at least twice the minimum fragment mass in a typical dark molecular cloud (derived by Low & Lynden-Bell to be about seven times the mass of Jupiter). Zuckerman & Song (2009) conclude that characteristic separations of imaged companions with the lowest masses (typically hundreds of AU with masses of 10–20 M_jup) are consistent with the Jeans fragmentation model. Consequently, both observations and theory suggest that wide separation secondaries will almost always have masses ≳ 10 M_jup. We assume that companions to stars will have masses ⩾ 15 M_jup.

According to Baraffe et al. (2003), I. Baraffe et al. (2006, private communication), and Burrows et al. (2003), the time needed for a 15 Jupiter mass object to cool to 500 K is ∼2 Gyr. The relatively small number of imaged companions with masses below ∼15 M_jup implies that the detection of a Y-type companion with an age < 2 Gyr will be very rare. Therefore, relatively older systems will make better targets to search for cool companions. Brown dwarf secondaries having masses larger than 15 M_jup are more common than the lowest mass companions listed in Table 1 of Zuckerman & Song (2009) (see Table 2 of the same paper). These companions would have to be around stars older than 2 Gyr in order to cool to the 500 K effective temperature that might herald the onset of the Y class. According to the models of Baraffe et al. (2003), even brown dwarfs as old as 7 Gyr would have to be less massive than ∼25 M_jup to have cooled to 500 K. Therefore, Y dwarfs themselves must be relatively low mass.

We are interested in probing for Y-type dwarfs which are born with separations from their host star in the hundreds to thousands of AU range. It is for the above reasons that the targets selected for our proper motion companion search are those nearby stars with ages > 2 Gyr. Our current list of 50 targets consists of stars from the Gliese catalog because these stars tend to be old (≫100 Myr; Song et al. 2003). Stars with the highest proper motions (total proper motion ⩾650 mas yr⁻¹) were selected from this catalog because of instrumental considerations (see Section 4). Each potential target was cross-matched with the ROSAT All-Sky Survey (RASS) with a search radius of 2' to check for X-ray emission as a possible indicator of age. Of the targets in our sample, seven were found to have been detected in X-rays. The X-ray luminosity measured for each still indicates an age at least as old as that of the Hyades (age ∼ 600 Myr). X-ray luminosity upper limits were calculated for the remainder of our targets. Detections and upper limits for our targets are shown in Figure 1. Additionally, space motions were calculated to verify that they do not fall within the young star UVW defined by Zuckerman & Song (2004). These regions are areas of UVW space inhabited by young (age ≲ 100 Myr), nearby stars. UVW space motions for our targets are plotted in Figure 2, which shows that the space motions of our targets are inconsistent with the space motions of nearby young stars. Each target was also checked for any measure of binarity to exclude those with known companions.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Our target list and current observation status are given in Table 1. Parallax measurements are from the Gliese catalog. Right ascension, declination, and J-magnitudes are from the Two Micron All Sky Survey (2MASS). Spectral type and proper motion are from the SIMBAD astronomical database. Distributions of proper motion and spectral type are shown in Figure 3. After the target list was created, we then checked each target for Galaxy Evolution Explorer(GALEX) near-ultraviolet emission as a possible additional indicator of age (Shkolnik et al. 2011; Rodriguez et al. 2011a). Twenty-eight of our targets had corresponding GALEX matches. Comparison with young stars (Zuckerman & Song 2004), Hyades members (obtained from the WEBDA open cluster database), and the rest of the Gliese catalog again indicate that our sample consists of an older population (Figure 4).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Observations for this survey began in 2007 August and we are still actively acquiring second epoch data for selected stars. This survey is being conducted with the Gemini camera (McLean et al. 1993) located on the Shane 3 m Telescope at the Lick Observatory, Mt. Hamilton, CA. The Lick Gemini camera is unique for our purposes, with its combination of a wide field of view (3' × 3') and a coronagraphic spot. Our observations of main-sequence F-, G-, K-, and M-type stars employ this 5'' coronagraphic spot to best suppress scattered light and obtain maximum sensitivity in the full field of view. These observations yield a radial field of view of ∼90'' and allow us to probe separations out to 900 AU at 10 pc and 1800 AU at 20 pc (distances to our targets range from 10 to 25 pc). While the smallest separations we can probe depend on the apparent brightness of the target star, the typical smallest separations to detect a source at 3σ above the background are ∼10''.

Based on T-dwarf spectra and theoretical models, the very low temperature objects we are seeking are anticipated to be more easily detected at J-band rather than K-band. Therefore, all targets are being observed in Gemini J-band imaging. With 30 minutes of on-source integration time per object, we can (in good conditions) achieve a limiting J-band magnitude of 20 at the 6σ detection level. Based on the Baraffe et al. (2003) models, this magnitude limit is sufficient to detect companions as old as 5 Gyr with temperatures down to 500 K out to 15 pc, and companions with temperatures down to 600 K out to 25 pc.

We search at two epochs for comoving companions to old, high proper motion, main-sequence stars. Through the end of 2009, we have obtained first epoch images for 50 main-sequence stars with spectral types from late-F to mid-M; second epoch imaging to similar depths have been obtained for 41 of these stars. All target images were reduced with custom in-house IDL software routines. Science frames were reduced by performing sky subtraction, flat division, bad pixel masking, shifting, and averaging. Source extraction for this project was done using the IRAF routine DAOFIND.

Background 2MASS sources in each field can be used to apply a world coordinate system (wcs) to each image. This is accomplished by matching pixel coordinates with 2MASS coordinates and using the IRAF task CCMAP. Once the transformation is found, a wcs can be applied to the image with the iraf task CCSETWCS. Once a wcs is applied to an image, measuring separations and position angles can be performed with the IDL routines gcirc.pro and posang.pro, both part of the IDL Astronomy User's Library. This process also allows us to calculate plate scale information for each image. We measure a GEMINI pixel scale of 070 pixel⁻¹ for these observations.

4.1. Astrometry

For each observation, we calculate a 90% magnitude completeness limit. Using these limits, known parallax measurements, and the "COND" evolutionary models from Baraffe et al. (2003), we estimate the minimum mass and effective temperature of a companion that could be detected in each image. The mass and corresponding effective temperature sensitivity limits for each target that has two observed epochs of good quality are shown in Figure 5 (assuming an age of 2 Gyr). The average temperature sensitivity limit for these pairs of images is ∼600 K. This corresponds to an average mass sensitivity limit of 0.018 solar masses (∼19 Jupiter masses) for an estimated age of ∼2 Gyr. One of the coolest dwarfs with a measured parallax is the T9+ dwarf UGPS 0722-05 discovered by Lucas et al. (2010). UGPS 0722-05 has an absolute J-band magnitude of 18.5 ± 0.2. By comparing this with our magnitude completeness limits, we would be able to detect an object as faint or fainter than UGPS 0722-05 around 20 of our targets (40%).

Zoom In Zoom Out Reset image size

The capability of GEOMAP to identify faint, comoving sources with only one year time baseline between epochs is demonstrated with our imaging of the GJ 1263 (Wolf 940; Burningham et al. 2009) wide binary system (M4 + T8.5). We first observed GJ 1263 on 2008 August 10 and again on 2009 August 28. Its comoving companion was identified by plotting positional offsets of surrounding objects (Figure 6) in the method described above. The 1σ uncertainty in position for background objects of median 2MASS J-mag = 16.5 is ∼0.10 pixels. With a total offset of ∼1.213 pixels (the total proper motion of GJ 1263 is ∼970 mas yr⁻¹), the candidate companion is ∼12σ away from the center of all background source residuals. Derived properties of GJ 1263B are displayed in Table 2. Absolute J magnitudes, separations, and projected separations are in good agreement with those quoted in Burningham et al. (2009).

Zoom In Zoom Out Reset image size

The first new discovery of our survey is a comoving companion to the star GJ 660.1. We observed GJ 660.1 on 2008 August 12 and again on 2009 August 30. GJ 660.1 is an M0 star located at a distance of 20.0^+1.6_{− 1.3} pc (van Leeuwen 2007) with proper motion in right ascension of 189 mas yr⁻¹ and −695 mas yr⁻¹ in declination. GJ 660.1 was detected in 2MASS with J-band magnitude of 8.66 ± 0.03. Radial velocity measurements (Reid et al. 1995) and the proper motions mentioned above indicate Galactic UVW space motions (U = 0.5 ± 2.1 km s⁻¹, V = −52.1 ± 3.0 km s⁻¹, W = −60.3 ± 3.5 km s⁻¹) inconsistent with any nearby young stellar associations.⁴ The large negative V and large absolute value of W, along with the low position of GJ 660.1A on an H-R diagram and a null RASS X-ray detection imply that the star is older than ∼2 Gyr.

Measured residuals for the companion and background sources are shown in Figure 6. The 1σ uncertainty in position for background objects is ∼0.091 pixels. With a total offset of ∼1.274 pixels, the companion to GJ 660.1 is ∼14σ away from the centroid of all background source offsets. Properties derived for this companion are shown in Table 2. This companion was detected by the 2MASS along with several background sources in our images. Residuals from this earlier 2MASS image (epoch 1999 March 31) confirm this companion at the ∼35σ level. Based on absolute measured and 2MASS magnitudes found for the companion and a comparison to the magnitudes of known dwarfs (using DwarfArchives.org), we estimate a spectral type of M9 ± 2. Using the models of Baraffe et al. (2003), the mass of this companion is determined to be between 0.075 and 0.080 M_☉ for an estimated age between 1 and 5 Gyr. This companion will be discussed in a future publication in more detail.

We report the methods and early results of a Lick/GEMINI survey of 50 old nearby stars for cool companions. We illustrate the capabilities of our observing strategy via detection of the previously known M4 + T8.5 binary system GJ 1263AB. In addition to this rediscovery, we discovered a late-M dwarf companion to GJ 660.1. By comparing position residuals versus those of stationary background objects, we conclude that GJ 660.1B is comoving with GJ 660.1A. We also show detectable lower limits of effective temperature and mass for possible companions of our targets. Based on our detection limits, we will be able to detect companions as faint or fainter than the T9+ dwarf UGPS 0722-05 for 40% of our targets. These early results show the effectiveness of this method.

This research was supported in part by a NASA grant to UGA and UCLA. This research has benefited from the M, L, and T dwarf compendium house at DwarfArchives.org and maintained by Chris Gelino, Davy Kirkpatrick, and Adam Burgasser. This research has made use of the WEBDA database, operated at the Institute for Astronomy of the University of Vienna. This research has made use of the SIMBAD database and VisieR catalog access tool, operated at CDS, Strasbourg, France. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. C.M. acknowledges support from the National Science Foundation under award no. AST-1003318. We thank Adam J. Burgasser for useful discussion.