A SEARCH FOR HIGH PROPER MOTION T DWARFS WITH Pan-STARRS1 + 2MASS + WISE

The Pan-STARRS1 (PS1) survey (Kaiser et al. 2010) represents the first of a new generation of multi-epoch digital sky surveys. In an effort to identify nearby T dwarfs and rare ultracool dwarfs of interest, we are conducting a proper motion search that combines the first epoch of PS1 data with the Two Micron All Sky Survey (2MASS) with initial results from PS1 commissioning data in Deacon et al. (2011). In this Letter, we present the discovery of a very high proper motion late-T dwarf, PSO J043.5395+02.3995 (hereinafter PSO J043.5+02). This object has independently been identified by Scholz et al. (2011) in a search for bright, very red objects in the WISE Preliminary Data Release with proper motions of ≳03 year⁻¹. They estimated a spectral type of T8–T10 based on the (W1 − W2) color, a photometric distance of 5.5^+2.3_{− 1.6} pc, and possible thick disk membership. Our near-IR spectroscopy gives a spectral type of T8, a photometric distance of 7.2 ± 0.7 pc, and probable thin disk membership.

The PS1 survey is obtaining multi-epoch imaging in five optical bands (g_P1, r_P1, i_P1, z_P1, y_P1) with a 1.8 m wide-field telescope on Haleakala, Maui. Images are processed nightly through the Image Processing Pipeline (IPP; Magnier 2006, 2007; Magnier et al. 2008). Photometry is on the AB magnitude scale, and the first generation of astrometry is tied to 2MASS, with limiting uncertainties of ≈70 mas. Although the PS1 filter system (Stubbs et al. 2010) is very similar to the Sloan Digital Sky Survey (SDSS; Fukugita et al. 1996), there are differences. Most notable are the facts that the z_P1 filter is cutoff at 840 nm unlike the long-pass z_SDSS filter, and SDSS has no corresponding y_P1 band (λ_C = 990 nm, Δλ = 70 nm), which exploits the excellent red sensitivity of the PS1 detectors.⁸

We use data from the 3π Survey, the most time intensive of the several PS1 surveys. This survey covers the 3π steradians north of declination −30°, with each filter observed for a total of six epochs over three years, and with two exposures at each epoch spaced ∼30 minutes apart to identify solar system objects. We used the the IPP software known as Desktop Virtual Observatory (DVO; Magnier 2006) to manage the large numbers of PS1 detections and enable large-area cross-correlations (∼10¹⁰ objects). We ingested 2MASS into the DVO database to mine the two data sets. At the time we did this (2011 February), PS1 had not yet completed its first sweep of the sky. Since the filters for the 3π Survey are observed weeks to months apart to optimize sky brightness conditions and parallax factors, we maximized our search area by focusing on the y_P1 data alone, rather than employing multi-color criteria covering a smaller patchwork on the sky.

Our search method is described in detail by Deacon et al. (2011). In brief, we identified proper motion candidates by matching PS1 y_P1 detections with 2MASS J-band detections, excluding galactic latitudes of |b| < 10° and allowing PS1–2MASS separations of ⩽28''. The PS1 data were taken from 2009 June to 2011 January, during commissioning and the first year of operations; as 2MASS was in operation from 1997 to 2001, our search has a mean time baseline of 11 years, leading to a maximum proper motion of ≈27 year⁻¹. We used generous PS1–2MASS color cuts to remove M dwarfs and some instrumental artifacts. False associations, whereby matches between PS1 and 2MASS detections in fact are two distinct stationary objects, were then screened using USNO-B (Monet et al. 2003) and SuperCOSMOS (Hambly et al. 2001), to depths of R < 20.5 and I < 19 mag. (A separate check of objects with I = 16–19 mag counterparts found no strong T dwarf candidates.) Our search reaches magnitudes of J ≈ 16.6 mag and y_P1 ≈ 19.6 mag, with 2MASS setting the search depth for T dwarfs (e.g., Figure 10 of Dupuy & Liu 2009). Given our limiting magnitudes, our culling process is inevitably incomplete. Indeed, the number of candidates increases at larger proper motions as expected due to the remaining mispairs.

Our core PS1+2MASS search program focuses on objects with ≲08 year⁻¹, where the false alarm rate is tractable. However, the release of the WISE Preliminary Data Release (Wright et al. 2010) in 2011 May opened the door to mining higher proper motions, by adding unique color information and an epoch of astrometry from the first half of 2010. We cross-matched our 21,070 high proper motion candidates against the WISE data at the PS1 position and excluded objects with a WISE source at the 2MASS position, returning 8140 matches. For the 131 objects with (W1 − W2) > 0.7 mag, corresponding to spectral type T0 and later, we screened the sample visually to excise image artifacts (using data from all the input surveys) and remaining faint mispairs (using the DSS R-band images, as bright early-T dwarfs might have counterparts on the I-band plates). For the remaining 58 objects, the separations between the PS1 and WISE positions showed the expected bimodal distribution of small separations (true matches) and large separations (false matches), with a minimum around 4'' Thus, we required the two positions to agree within 4'', leaving 49 objects.

The final sample is illustrated in Figure 1. Two objects stand out as being very red in (W1 − W2) and having high proper motions. (Objects with proper motions of <05 year⁻¹ will be discussed in a future paper.) One is the T7.5 dwarf GJ 570D (Burgasser et al. 2000). The other, PSO J043.5395+02.3995 (WISEP J025409.45+022359.1; Figure 2), was previously unknown at the time.⁹ We identified a z_SDSS-only counterpart in the latest SDSS data release (DR8; Aihara et al. 2011). This object's absence in earlier releases explains why it was not found in past SDSS searches. Similarly, its faint J-band magnitude excluded it from previous 2MASS searches.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

At the time of our search, about 8200 deg² had y_P1 data and were in the WISE Preliminary Data Release, with about 7800 deg² of the area also having z_P1 data. (This includes the 74% fill factor for individual PS1 exposures due to gaps between CCDs, masked pixels around bright stars, and bad regions of the detectors.) To check our completeness, we examined the known T dwarfs from Dwarfarchives. Out of the 16 objects that are present in WISE, bright enough to be in 2MASS, and residing within the area imaged by PS-1, we retrieved 12 of them, with the rest being culled during the filtering process described above, e.g., by being too close to a USNO-B source. Thus, our estimated completeness is 75%. The other high proper motion object found by Scholz et al. (2011), the T9–T10 dwarf WISEP J1741+2533, was not selected as its sky position did not have y_P1 imaging when we did our data mining.

We obtained low-resolution (R ≈ 100) spectra of PSO J043.5+02 on 2011 June 25 and July 21 UT from NASA's Infrared Telescope Facility located on Mauna Kea, HI. The two spectra agree well, but the July one had a higher signal-to-noise ratio (S/N) so we use it here. Conditions were photometric with average seeing. We used the near-IR spectrograph SpeX (Rayner et al. 1998) in prism mode with the 08 slit, obtaining 0.8–2.5 μm spectra in a single order. The total on-source integration time was 16 minutes. We observed the A0V star HD 18571 contemporaneously with PSO J043.5+02 for telluric calibration. All spectra were reduced using version 3.4 of the SpeXtool software package (Vacca et al. 2003; Cushing et al. 2004).

The spectrum shows the strong H₂O and CH₄ absorption characteristic of late-T dwarfs (Figure 3). We classified PSO J043.5+02 using the five spectral indices from Burgasser et al. (2006b), with spectral types assigned using the polynomial fits from Burgasser (2007). Excluding the saturated CH₄-K index, the average spectral type from the indices was T8.0 with an rms of 0.14 subclasses. We also computed the W_J and NH₃-H indices of Warren et al. (2007) and Delorme et al. (2008), respectively, and found values similar to known T8 dwarfs (e.g., Burningham et al. 2010).

Zoom In Zoom Out Reset image size

Table 1. Measurements of PSO J043.5+02

Property	Measurement
Astrometrya (Equinox 2000)
2MASS R.A., decl. (ep 2000.727)	02:54:07.886, +02:23:56.35
SDSS R.A., decl. (ep 2008.723)	02:54:09.242, +02:23:58.32
PS1 yP1 R.A., decl. (ep 2010.651)	02:54:09.575, +02:23:58.68
PS1 yP1 R.A., decl. (ep 2010.651)	02:54:09.579, +02:23:58.56
PS1 zP1 R.A., decl. (ep 2010.673)	02:54:09.578, +02:23:58.69
PS1 zP1 R.A., decl. (ep 2010.673)	02:54:09.581, +02:23:58.56
PS1 yP1 R.A., decl. (ep 2011.064)	02:54:09.621, +02:23:58.72
PS1 yP1 R.A., decl. (ep 2011.064)	02:54:09.633, +02:23:58.83
WISE R.A., decl. (ep 2010.075)	02:54:09.449, +02:23:58.56
Proper motion amplitude μ (''year−1)	2.559 ± 0.011
Proper motion P.A. (deg)	84.82 ± 0.25
Parallax π (mas)	171 ± 45
Photometry
SDSS z (AB mag)	19.86 ± 0.07
PS1 zP1 (AB mag)	21.25 ± 0.06b
PS1 yP1 (AB mag)	19.11 ± 0.04b
2MASS J (mag)	16.56 ± 0.16
2MASS H (mag)	15.88 ± 0.20
2MASS $\mbox{$K_S$}$ (mag)	>16.0 (2σ)
WISE W1 (mag)	15.74 ± 0.07
WISE W2 (mag)	12.71 ± 0.03
WISE W3 (mag)	11.04 ± 0.13
WISE W4 (mag)	>9.1 (2σ)
Spectrophotometryc
MKO (synth) J (mag)	16.14 ± 0.12
MKO (synth) H (mag)	16.51 ± 0.12
MKO (synth) K (mag)	16.84 ± 0.12
MKO (synth) (J − H) (mag)	−0.368 ± 0.002
MKO (synth) (H − K) (mag)	−0.336 ± 0.005
MKO (synth) (J − K) (mag)	−0.704 ± 0.005
2MASS (synth) J (mag)	16.43 ± 0.12
2MASS (synth) H (mag)	16.47 ± 0.12
2MASS (synth) $\mbox{$K_S$}$ (mag)	16.69 ± 0.12
2MASS (synth) (J − H) (mag)	−0.044 ± 0.002
2MASS (synth) $(H-\mbox{$K_S$})$ (mag)	−0.213 ± 0.005
2MASS (synth) $(J-\mbox{$K_S$})$ (mag)	−0.257 ± 0.004
H2O-J	0.039 (T8.2)
CH4-J	0.192 (T7.9)
H2O-H	0.175 (T8.0)
CH4-H	0.121 (T7.9)
CH4-K	0.059 (>T6)
WJ	0.321 (T8)
NH3-H	0.625
K/J	0.121
Spectral type	T8
Photometric distance (pc)	7.2 ± 0.7

Notes. ^aThe SDSS, PS1, and WISE astrometry has been tied to common system, using 2MASS as the absolute reference (Section 4). ^bAverage of multi-epoch photometry. ^cBroadband photometry was synthesized from our flux-calibrated spectrum, with errors derived from the 2MASS J- and H-band photometry and the spectrum's uncertainties. The errors in the synthesized colors are smaller than for the magnitudes, because the colors incorporate only the spectrum's uncertainties. Note that the synthesized (J − H) 2MASS color (−0.044 ± 0.002 mag) is consistent with the T8 spectral type, but different than the 0.7 ± 0.3 mag from the 2MASS catalog, likely due to underestimated errors in the 2MASS photometry.

Download table as:  ASCIITypeset image

In addition, we visually classified PSO J043.5+02 by comparing with SpeX prism spectra of late-T dwarf spectral standards. Following the prescription of Burgasser et al. (2006b), the depth of the H₂O and CH₄ absorption bands were examined, normalizing the spectra of PSO J043.5+02 and the standards at their J-, H-, and K-band peaks. The agreement with the T8 standard 2MASS J0415−0935 is excellent, with PSO J043.5+02 showing stronger absorption than the T7.5 dwarf GJ 570D. The T8.5 objects from Delorme et al. (2008) and Burningham et al. (2008) have deeper H₂O absorption and narrower J- and H-band continuua than PSO J043.5+02. Thus, the indices and visual typing both give a spectral type of T8.

We fit the solar metallicity BT-Settl-2010 model atmospheres (Allard et al. 2010) to our spectrum. The models span $\mbox{$T_{\rm eff}$}=500\hbox{--}1500$ K ($\Delta \mbox{$T_{\rm eff}$}=100$ K) and $\mbox{$\log (g)$}=4.0\hbox{--}5.5$ (cgs; $\Delta \mbox{$\log (g)$}=0.5$). We first flux-calibrated our spectrum to the weighted average of the 2MASS J- and H-band photometry (which is only S/N ≈ 5–6 in each filter). Following Bowler et al. (2009) and Cushing et al. (2008), we used a Monte Carlo approach to fit the 0.8–2.4 μm data, excluding the 1.60–1.65 μm region because of the known incompleteness of the methane line list. For each Monte Carlo trial, we changed the flux calibration and spectrum assuming normal distributions for the photometric and spectroscopic uncertainties, respectively. We then fit the model atmospheres to each artificial spectrum and repeated the process 10³ times.

The best-fitting model has $\mbox{$T_{\rm eff}$}=800$ K and $\mbox{$\log (g)$}=4.0$ (Figure 3), comparable to atmosphere fitting results for other comparably late-type objects (e.g., Saumon et al. 2007; Leggett et al. 2007; Liu et al. 2011). The broadband photometry agrees well with the model, except for W2 where the model is a factor of 2.92 ± 0.35 fainter. The mid-IR colors of late-T dwarfs show significant scatter with spectral type (Figure 3), due to non-equilibrium CO/CH₄ chemistry, metallicity, and surface gravity effects (e.g., Leggett et al. 2010), which are not fully incorporated into current models. This empirical scatter also explains the spectral type of T8–T10 inferred by Scholz et al. (2011) based on the mid-IR colors, compared to the actual type of T8.

Based on an unweighted fit to 19 objects from L5–T10 with parallaxes and WISE data, we derive a third-order polynomial for the W2 band absolute magnitude M(W2) = ∑_i a_i (SpT)ⁱ, where a_i = { − 1.075, 1.580, −0.06780, 0.001066} and SpT = 15 for L5, = 20 for T0, etc. The rms scatter about the fit is only 0.18 mag. This gives a photometric distance of 7.2 ± 0.7 pc.¹⁰

The 256 year⁻¹ proper motion of PSO J043.5+02 is exceptional. Among field T dwarfs, the only object with a larger value is the T7.5 dwarf 2MASS J1114−26 (305 year⁻¹; Tinney et al. 2005), which Leggett et al. (2007) characterize as slightly metal-poor ([M/H] ≈ −0.3 dex). Our photometric distance of 7.2 ± 0.7 pc leads to a tangential velocity of 87 ± 8 km s⁻¹. Following Dupuy et al. (2009), we assess the galactic membership of PSO J043.5+02 using an oversampled simulation of the stars within 30 pc from the Besançon model (Robin et al. 2003). The model provides V_tan distributions for thin disk, thick disk, and halo stars. We compute the membership probabilities for PSO J043.5+02 from the fraction of its nearest neighbors that belong to each model population. For V_tan= 87 km s⁻¹, we find membership probabilities in the thin/thick disk of 0.87/0.13, with <0.01 probability of belonging to the halo. Thus, even though the V_tan of PSO J043.5+02 is large, it is merely on the cusp of thin/thick disk membership, because of the many more thin disk stars (a divide of 0.973/0.026/0.002 between thin disk/thick disk/halo members in this model). Thin disk membership is preferred, even accounting for the measurement uncertainty; at the ±1σ limits of V_tan (79 and 95 km s⁻¹), the thin/thick disk probabilities are 0.90/0.10 and 0.77/0.23, respectively.¹¹

The near-IR spectrum indicates a metallicity comparable to other field objects (Figure 3). PSO J043.5+02 is very similar to the T8 dwarf 2MASS J0415−09, which is inferred to have [Fe/H] = 0.0–0.3 and $\mbox{$\log (g)$}=5.0\hbox{--}5.4$ (Saumon et al. 2007). A solar metallicity is also signalled by the Y-band continuum shape, which is metallicity-sensitive (Burgasser et al. 2006a; Leggett et al. 2007). PSO J043.5+02 shows a more pointed shape than the T7.5 dwarf 2MASS J1114−26, which is inferred to be sub-solar ([Fe/H] = −0.3; Leggett et al. 2007). Finally, the weaker K-band flux of PSO J043.5+02 compared to the T8 dwarf Ross 458C (Goldman et al. 2010; Scholz 2010) further indicates that PSO J043.5+02 has a metallicity and surface gravity (age) typical of field objects. The youth (150–800 Myr) and metal-richness ([Fe/H] = +0.2–0.3) of Ross 458C reduce the effect of collision-induced H₂O opacity, leading to brighter K-band emission (e.g., Liu et al. 2007).

Our model atmosphere fitting gives a low surface gravity (log (g) = 4.0); however, the result is ill-constrained, as seen by the χ² surface (Figure 3). Following Warren et al. (2007), a better constraint comes from comparing PSO J043.5+02 to other late-T dwarfs using the W_J and K/J indices, guided by model atmosphere predictions and empirical benchmarks. In such a representation (e.g., Figure 5 of Liu et al. 2011), PSO J043.5+02 is inferred to have a relatively high surface gravity ($\mbox{$\log (g)$}\approx 5.0\hbox{--}5.5$) based on its large K/J value, consistent with an old (≳Gyr) age.

PS1 3π Survey z_P1 and y_P1 imaging are scheduled to maximize the parallax factor between epochs, with the goal of deriving parallaxes over the entire survey. Given the proxmity of PSO J043.5+02, a preliminary result is possible even with only a five-month PS1 baseline (three epochs from 2010.65 to 2011.06, with two exposures per epoch), by adding astrometry from 2MASS (2000.73), SDSS (2008.72), and WISE (2010.08). Altogether, these data span a 10 year baseline. For WISE, there are known ∼05 systematic errors in the Preliminary Release Source Catalog, so we constructed our own astrometric catalog for 1 deg² around PSO J043.5+02 from the Preliminary Single Exposure (L1b) Working Database, which we found to be unaffected by such problems. We cross-matched individual WISE detections and used the weighted average and standard error for positions and uncertainties. For 2MASS and SDSS, we assumed relative positional uncertainties of 100 mas and 70 mas, respectively, based on our experience with matching these catalogs to high-precision (≲ 10 mas) relative astrometry from our parallax program on the Canada–France–Hawaii Telescope (Dupuy 2010). We then combined all nine astrometric data points (2MASS, SDSS, WISE, and 6 × PS1), allowing for small shifts and a full linear transformation. We used a Markov Chain Monte Carlo method to determine the posterior distributions of the proper motion and parallax. The best solution had χ² = 9.9 (13 degrees of freedom), validating our astrometric errors, a parallax of 171 ± 45 mas (Figure 4), and a proper motion of μ = 2559 ± 0011 year⁻¹ and P.A. = 8482 ± 025. This includes a (negligible) correction from relative to absolute parallax of 1.34 ± 0.15 mas computed using the Besançon model as described in Dupuy (2010).

Zoom In Zoom Out Reset image size

For the median V_tan of 30 km s⁻¹ of ultracool dwarfs within 20 pc (Faherty et al. 2009), our proper motion range of 05–27 year⁻¹ corresponds to distances of 12 pc and 2 pc. (Objects with higher tangential velocities of course can be selected at even larger distances.) This approximate kinematic limit is well matched to the J ≈ 16.5 mag limit for our search: the three T7.5 dwarfs with 2MASS photometry and parallaxes have M_J = 15.93, 15.65, and 16.47 mag (HD 3651B, 2MASS J1217−03, and GJ 570D, respectively) and the T8 dwarf 2MASS J0415−09 has M_J = 16.90 mag, meaning our search has a limiting distance of ≈12 pc for T8 dwarfs (9 pc for the 2MASS full-sky completeness of J = 15.8). The limiting distance drops precipitously for later types: the T10 dwarf UGPS J0722−05 (Lucas et al. 2010) at 4 pc has J = 16.49 ± 0.13 mag, right at the nominal 2MASS limit.

The modest number of candidates from our search suggests that the immediate (∼10 pc) solar neighborhood does not contain a large reservoir of undiscovered T dwarfs down to about T8. We found 7–10 candidates (depending on the cutoff used for the PS1–WISE separation) with proper motions of 05–27 year⁻¹ and colors of (W1 − W2) > 0.7, including three previously known objects. In comparison, Dwarfarchives.org lists 30 objects in the same proper motion range and with spectral types of T0–T8 (ignoring two tight companions found by adaptive optics imaging). Since our PS1+2MASS+WISE search covered ≈20% of the sky, a rough estimate based on the known census predicts we should identify about six objects, consistent with our findings.

The Pan-STARRS1 surveys have been made possible by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the institutions of the Pan-STARRS1 Science Consortium (http://www.ps1sc.org), and NASA. Our research has employed the WISE, SDSS-III, and 2MASS data products; NASA's Astrophysical Data System; the SIMBAD database; the M, L, and T dwarf compendium housed at DwarfArchives.org; and the SpeX Prism Spectral Libraries. This research was supported by NSF Grants AST-0507833 and AST09-09222 (awarded to MCL), AST-0709460 (awarded to EAM), AFRL Cooperative Agreement FA9451-06-2-0338, and DFG-Sonderforschungsbereich 881 "The Milky Way." Finally, the authors wish to recognize the very significant cultural role that the summit of Mauna Kea has always had within the indigenous Hawaiian community. We are most fortunate to conduct observations from this mountain.

Facilities: IRTF (SpeX) - Infrared Telescope Facility, PS1 (GPC1) - Panoramic Survey Telescope and Rapid Response System Telescope #1 (Pan-STARRS)