The Gaia Ultracool Dwarf Sample. I. Known L and T dwarfs and the first Gaia data release

TL;DR

This paper compiles and analyzes known ultracool dwarfs observed by Gaia DR1, matching them with infrared surveys to improve spectral typing, identify outliers, and discover new candidate systems with common proper motions.

Contribution

It provides the first comprehensive Gaia-based ultracool dwarf sample, combining Gaia and infrared data to refine classifications and identify new binary candidates.

Findings

321 L and T dwarfs observed in Gaia DR1

45% of known LT dwarfs brighter than G=20.3 mag included

15 new candidate common proper motion systems identified

Abstract

We identify and investigate known ultracool stars and brown dwarfs that are being observed or indirectly constrained by the Gaia mission. These objects will be the core of the Gaia ultracool dwarf sample composed of all dwarfs later than M7 that Gaia will provide direct or indirect information on. We match known L and T dwarfs to the Gaia first data release, the Two Micron All Sky Survey and the Wide-field Infrared Survey Explorer AllWISE survey and examine the Gaia and infrared colours, along with proper motions, to improve spectral typing, identify outliers and find mismatches. There are 321 L and T dwarfs observed directly in the Gaia first data release, of which 10 are later than L7. This represents 45 % of all the known LT dwarfs with estimated Gaia G magnitudes brighter than 20.3 mag. We determine proper motions for the 321 objects from Gaia and the Two Micron All Sky Survey positions. Combining the Gaia and infrared magnitudes provides useful diagnostic diagrams for the determination of L and T dwarf physical parameters. We then search the Tycho-Gaia astrometric solution Gaia first data release subset to find any objects with common proper motions to known L and T dwarfs and a high probability of being related. We find 15 new candidate common proper motion systems.

Tables

Table 1. Table 1: Distance limits for L0 to T7 spectral types using Eq.  2 . D = G < 20.3 {}_{G<20.3}= distance limit assumming G < 𝐺 absent G< 20.3 mags and D G<20.7 for G < 𝐺 absent G< 20.7 mags.

L	DG<20.3	DG<20.7	T	DG<20.3	DG<20.7
SpT	(pc)	(pc)	SpT	(pc)	(pc)
L0	69	82	T0	12	14
L1	55	67	T1	12	14
L2	45	54	T2	12	14
L3	36	44	T3	12	14
L4	29	35	T4	11	14
L5	24	29	T5	10	12
L6	19	23	T6	8	10
L7	16	19	T7	6	7
L8	13	15	T8	3	4
L9	10	12	T9	2	2

Table 2. Table 2: The GUCDS input catalogue a .

GUCDS Name $\alpha$ $\delta$ Spectral ${\mu }_{\alpha }\cos \delta$ ${\mu }_{\delta }$ $J$ ID [deg] [deg] type [mas/yr] [mas/yr] [mag] J0000+2554 2MASS J00001354+2554180 0.0564165 +25.905000 T4.52 +6$\pm$19 +130$\pm$22 15.063 J0001+1535 2MASS J00011217+1535355 0.3007080 +15.593194 L42 +150$\pm$19 –169$\pm$19 15.522 J0001–0841 2MASS J00013166-0841234 0.3830416 –8.690806 L1 p(blue)2 +331$\pm$14 –299$\pm$14 15.712 J0002+2454 2MASS J00025097+2454141 0.7123755 +24.903917 L5.52 +2$\pm$23 –36$\pm$29 17.165 J0004-6410 2MASS J00040288-6410358 1.0120000 –64.176611 L1 $\gamma$1 +64$\pm$5 –47$\pm$12 15.786 …

Table 3. Table 3: Input catalogue objects with mismatches in Gaia  DR1.

Name	Gaia DR1 Source ID	$\Delta G$[mag]	Remarks
WISEPA J062720.07-111428.81	3000505938722626560	-1.4	Probable satellite of galaxy USNO-B1 0787-0079432
USco J160714.79-232101.22	6242316978518793856	-1.4	Star or galaxy USNO-B1 0666-0359684
USco J163919.15-253409.93	6046485475060342528	-1.1	Star or galaxy USNO-B1 0644-0382256
WISE J170745.85-174452.54	4135505777467362304	-2.6	In crowded region towards Galactic centre
2MASS 18000116-15592355	4145132555890843520	1.3	In crowded region of low Galactic latitude
WISE J200403.17-263751.76	6850032688873127424	-0.9	Star or galaxy USNO-B1 0633-0938809
WISE J223617.59+510551.94	1988335902592180608	-0.3	Star or galaxy USNO-B1 1410-0455016

Table 4. Table 4: New parameters for the GUCDS-DR1 catalogue. a

Short	Gaia	$\alpha ,\delta$	${\mu }_{\alpha }$cos$\delta$, ${\mu }_{\delta }$	$G$	$\Delta G$	Nobs, N3, Fμ
Name	Source ID	[deg]	[mas/yr]	[mag]	[mag]
J0006-1720	2414607592787544320	1.585253567, -17.347415311	-41$\pm$ 11, 0$\pm$ 12	20.525$\pm$0.043	0.163	97,0,1
J0006-0852	2429054454021227648	1.704590901, -8.880825977	-61$\pm$ 12, -324$\pm$ 12	18.485$\pm$0.008	-0.078	239,0,0
J0006-6436	4900323420040865792	1.742217614, -64.615322431	82$\pm$ 5, -65$\pm$ 13	17.988$\pm$0.011	0.103	145,0,0
J0016-1039	2428008410441149824	4.156258155, -10.653818836	-111$\pm$ 12, -193$\pm$ 12	20.028$\pm$0.022	0.073	185,0,0
J0018-6356	4900453540369793408	4.695160455, -63.938248147	324$\pm$ 6, -381$\pm$ 13	19.774$\pm$0.025	-0.110	108,0,0
…

Table 5. Table 5: The common-distance, common-proper-motion pair candidates identified here. A machine readable version of this table is available online.

Name	$\alpha$	$\delta$	${\mu }_{\alpha }\cos \delta$	${\mu }_{\delta }$	$d$	PDF	Proj. Sep.	SpT
	[deg]	[deg]	[mas yr-1]	[mas yr-1]	[pc]		[103 au]
HIP 10346	33.31768733	-59.56700153	126.5$\pm$ 0.1	-8.3$\pm$ 0.1	58.4$\pm$ 1.8	2.2e-05	400
J0223-5815	35.97766900	-58.25187300	134.0$\pm$10.0	5.0$\pm$19.0	49.0$\pm$10.0
HIP 12158	39.17535990	-3.15589420	323.4$\pm$ 0.1	58.2$\pm$ 0.1	24.1$\pm$ 0.2	3.6e-08	145
J0230-0225	37.66214300	-2.43167270	329.0$\pm$16.8	51.3$\pm$14.9	27.0$\pm$ 6.0
TYC 146-1101-1	97.54755003	0.87648584	68.4$\pm$ 2.9	-105.7$\pm$ 2.4	67.6$\pm$ 1.0	3.3e-06	252
J0626+0029	96.58839400	0.49281806	84.0$\pm$15.0	-92.0$\pm$15.0	67.0$\pm$14.0
HIP 38492	118.24796721	22.55612499	-85.9$\pm$ 0.2	-61.4$\pm$ 0.1	34.2$\pm$ 0.3	3.0e-07	157
J0758+2225	119.62429000	22.42408300	-105.0$\pm$ 8.0	-62.8$\pm$ 8.2	33.0$\pm$ 8.0
TYC 230-109-1	139.99273886	4.99944728	-80.0$\pm$ 1.2	-40.2$\pm$ 1.1	42.2$\pm$ 0.7	1.5e-07	178
J0915+0531	138.93388000	5.51780560	-95.0$\pm$ 5.5	-57.7$\pm$ 4.4	33.0$\pm$ 6.0
TYC 2504-466-1	145.55216670	33.92992123	-103.7$\pm$ 3.8	-67.3$\pm$ 1.6	61.5$\pm$ 1.9	1.1e-06	156
J0939+3412	144.77713000	34.21596100	-107.1$\pm$10.4	-64.3$\pm$12.6	62.0$\pm$12.0
HIP 47704	145.89719170	10.51834984	37.5$\pm$ 0.1	-124.9$\pm$ 0.1	71.7$\pm$ 1.2	2.5e-06	211
J0943+0942	145.95667000	9.70094440	45.4$\pm$10.9	-119.9$\pm$ 8.8	79.0$\pm$15.0
TYC 824-423-1	145.44963031	11.17517624	64.6$\pm$ 0.9	-111.1$\pm$ 0.7	102.8$\pm$ 3.2	8.2e-06	576
J0943+0942	145.95667000	9.70094440	45.4$\pm$10.9	-119.9$\pm$ 8.8	79.0$\pm$15.0
HIP 57734	177.58699124	10.06723706	-89.9$\pm$ 0.1	-16.5$\pm$ 0.0	86.2$\pm$ 2.2	1.2e-06	77
J1150+0949	177.66163000	9.82858330	-107.6$\pm$17.1	-31.9$\pm$ 4.5	60.0$\pm$27.0
HIP 58241	179.18153043	-32.26744260	-178.8$\pm$ 0.6	-7.1$\pm$ 0.3	35.4$\pm$ 0.3	5.7e-06	229
J1154-3400	178.67596000	-34.01084900	-161.0$\pm$13.0	4.0$\pm$15.0	30.0$\pm$ 6.0
HIP 58240	179.17544959	-32.26819149	-172.0$\pm$ 0.4	-8.3$\pm$ 0.3	35.8$\pm$ 0.4	4.0e-06	231
J1154-3400	178.67596000	-34.01084900	-161.0$\pm$13.0	4.0$\pm$15.0	30.0$\pm$ 6.0
HIP 59887	184.22818632	37.48407627	-107.9$\pm$ 0.1	-2.0$\pm$ 0.1	87.8$\pm$ 1.7	9.0e-06	153
J1214+3721	183.64024000	37.35327100	-122.6$\pm$10.6	15.7$\pm$13.4	82.0$\pm$17.0
HIP 62350	191.64195696	11.37849569	-112.4$\pm$ 0.1	-0.5$\pm$ 0.0	61.5$\pm$ 1.9	1.1e-05	286
J1244+1232	191.05429000	12.53363900	-104.8$\pm$ 8.6	4.5$\pm$ 7.3	46.0$\pm$ 8.0
TYC 2587-1547-1	248.21928595	35.07510874	88.0$\pm$ 0.5	-61.8$\pm$ 0.5	34.6$\pm$ 0.3	1.4e-10	2
J1632+3505	248.23375000	35.08545700	91.6$\pm$ 9.7	-65.3$\pm$11.9	37.0$\pm$ 8.0
HIP 101880	309.67956167	-43.73289384	235.1$\pm$ 0.2	-371.5$\pm$ 0.1	51.4$\pm$ 0.6	2.1e-10	270
J2037-4216	309.46379000	-42.27922200	229.0$\pm$10.0	-391.0$\pm$10.0	51.0$\pm$10.0

Table 6. Table 6: Probable late M dwarfs identified in the colour-colour and colour-spectral type diagrams.

Long Name	Opt SpT	Remarks
Short Name	NIR SpT
Plot label	$\Delta G$
SSSPM J0109–4955 J0109-4954 1	M81 L12 -0.45	This object is bluer in all colours than we would expect for an M8 but further spectral observations are needed to clarify its spectral type.
SSSPM J0134–6315 J0133-6314 2	… L02 -1.02	This is the bluest LT dwarf in the majority of colour-colour plots. It has been classified as early as an M5 (Lodieu et al., 2005a) and a X-Shooter spectra is a best fit with an M6 template. The $W3$ magnitude is an upper limit which explain its outlier position in the $W3$ colour combinations.
2MASS 11164800+6037309 J1116+6037 3	L03 … -0.33	This is at the border of the majority of L0 colour loci. Using the best fitting template procedure from Marocco et al. (2015) on its published SDSS spectra we find it is a late M dwarf.
2MASS 12455566+4902105 J1245+4902 4 1	L13 M84 -0.65	This object has been classified as M84 in the infrared and both L13 and M9 (West et al., 2011) in the optical – both from SDSS spectra. It is a borderline M/L object.
2MASS 12504567+4418551 J1250+4418 5	L05 … -0.66	Very blue in many colour-colour plots and comparisons of the SDSS spectra using the procedure from Marocco et al. (2015), the same spectra as used by West et al. (2008) to find L0, we find the object to be a late M dwarf. The $W3$ magnitude is an upper limit and its extreme position in the W3 colour combinations indicates that it is significantly fainter that the published value.
2MASS 12512841+6243108 J1251+6243 6	M8V5 L44 -0.90	This object was listed by as an L44 but erroneously cited as coming from Zhang et al. (2009); West et al. (2008) classified it as an M8V. Our $G$ magnitude would be more consistent with the earlier type and we think this is a case of object mis-identification and the actual object is an M8V.
2MASS 13331284+1509569 J1333+1509 7	L03 M84 -0.26	This object has been classified as M84 in the infrared and both L03 and M9 (West et al., 2011) from the same SDSS spectra in the optical. It is a borderline M/L object. The $W3$ magnitude is an upper limit and it is an outlier in the ALLWISE $W3$ colour combinations so it is probably fainter.

Table 7. Table 7: Mean colours and standard deviation for all spectral type bins from L0 to L5 where there were at least 3 entries.

Optical		N
SpT	[mag]
L0	4.52 $\pm$ 0.19	126
L0.5	4.53 $\pm$ 0.07	7
L1	4.58 $\pm$ 0.13	55
L1.5	4.63 $\pm$ 0.10	14
L2	4.70 $\pm$ 0.15	29
L2.5	4.65 $\pm$ 0.14	8
L3	4.72 $\pm$ 0.17	17
L3.5	4.95 $\pm$ 0.09	5
L4	4.92 $\pm$ 0.14	11
L4.5	5.01 $\pm$ 0.13	3
L5	5.02 $\pm$ 0.17	8

Table 8. Table 8: Objects with H G > 24.0 subscript 𝐻 𝐺 24.0 H_{G}>24.0 .

Long name	Opt SpT	NIR SpT	$H_G$ [mag]
ULAS J033350.84+001406.1	L0 sd1	L0 sd2	24.2
DENIS J081730.0-615520	…	T63	24.9
2MASS 11555389+0559577	…	L7.54	24.0
2MASS 11582077+0435014	L7 sd5	L7 sd5	25.6
SDSSp J120358.19+001550.3	L36	L57	24.4
2MASS 12074717+0244249	L88	T09	24.5
2MASS 12304562+2827583	…	L110	24.3
ULAS J124425.90+102441.9	…	L0.5 sd2	24.2
2MASSW J1411175+393636	L1.511	L1.57	24.1
2MASSW J1515008+484742	L612	L613	25.2
2MASS 22114470+6856262	…	L25	24.0
2MASS 22490917+3205489	L512	…	25.0

Equations

$$G_{\mathrm{pred}}=J-12.63+0.244\mathrm{SpT},$$

$$G_{\mathrm{est}}=J-1.098+0.080\mathrm{SpT}$$

$$H_G=G+5+5\log \mu ,$$

Full text

1 Introduction

Gaia is observing over a billion objects in our Galaxy and is revolutionizing Astronomy in many areas (Gaia Collaboration et al., 2016a). One of these areas is the study of the bottom of the main sequence and beyond. L and T (hereafter LT) dwarfs are very cool faint objects that are either substellar or at the stellar-substellar boundary (Delfosse et al., 1997; Martín et al., 1999; Kirkpatrick et al., 1999; Burgasser et al., 2006a; Dieterich et al., 2014). In the billion-object catalogue of Gaia, there will be direct observations of about a thousand LT dwarfs (Sarro et al., 2013; Smart, 2014). This sample, even though it is relatively small, will be more homogeneous, accurate, complete and larger than the current catalogue of known L and early T dwarfs with measured parallactic distances.

The reason for the relative paucity of LT dwarfs in the Gaia observations is because they emit predominantly in the infrared and are very faint in the Gaia bands (see Fig. 1). However, Gaia will provide a magnitude-limited complete sample of the early LT spectral types in the solar neighbourhood. The nominal Gaia $G$ magnitude limit is 20.7 mag and we expect the mission to be complete to $G=20.3$ mag (Gaia Collaboration et al., 2016b). Internal validation, with models and clusters, finds a completeness of 50% at $G=20.3$ mag for the Gaia first data release (hereafter DR1)111http://gaia.esac.esa.int/documentation/GDR1/. In Table 1 we report the distance limits for L0 to T9 objects with a $G$ limit of 20.3 mag and 20.7 mag using Eq. 2 developed in Section 3.1. In addition to solar-metallicity LT dwarfs, Gaia will also provide a volume limited sample of old thick disk or halo L-type sub-dwarfs, and young LT objects in the solar neighbourhood.

Other LT and even cooler Y dwarfs (Cushing et al., 2011) will be indirectly detected in Gaia observations, for example, as low-mass companions in unresolved binary systems (Pope et al., 2013; Sozzetti et al., 2014; Littlefair et al., 2014; Burgasser et al., 2015; Ma et al., 2016), and as gravitational microlenses (Belokurov & Evans, 2002; Proft et al., 2011; Sahu et al., 2014; Ranc & Cassan, 2014). Gaia will constrain other LT and Y dwarfs in common proper motion (CPM) systems of wide binaries or moving groups where distances and kinematics of the brighter members, visible to Gaia, can be matched to the fainter objects with kinematics found from other surveys.

Ultracool dwarfs (UCDs) are defined as objects later than M7 (see Jones & Steele, 2001). We have begun a systematic project to catalogue and characterise the cooler part of the Gaia Ultracool Dwarf sample (hereafter GUCDS), being all L, T and Y dwarfs that Gaia will directly observe or indirectly constrain. The GUCDS will be the primary sample in the near future to test atmospheric models and evolution scenarios, and to derive fundamental properties of objects at the end of the main sequence.

Here we find the LT dwarfs directly observed by Gaia as isolated objects with an identifiable entry in the Gaia DR1 and we find those LT dwarfs in CPM systems with the Gaia DR1 subset with astrometric solutions. In Section 2 we describe the L, T and Y dwarf input catalogue used to search the Gaia DR1; in Section 3 we describe the production of the GUCDS catalogue of known matched LT dwarfs; in Section 4 we describe the discovery of new CPM candidates; in Section 5 we discuss the two catalogues in various magnitude, colour, and proper motion parameter spaces and in the last section we summarise the results.

2 Catalogue of known L, T and Y dwarfs

2.1 Input catalogue

LT dwarfs seen by Gaia will all be nearby ($d<$ 82 pc; Table 1) and, therefore, have significant proper motions. With this in mind, we used as the starting point for our input catalogue of known late M, L, T and Y dwarfs the online census being kept by J. Gagné222https://jgagneastro.wordpress.com/list-of-ultracool-dwarfs/. This included objects from the Dwarfarchives333http://www.dwarfarchives.org/, the work of Dupuy & Liu (2012), and the PhD thesis catalogue of Mace (2014). To this compilation, we added the objects in Marocco et al. (2015) and Faherty et al. (2016). We did not include the significant number of UCD candidates with photometry-based spectral types (e.g. Folkes et al., 2012; Smith et al., 2014; Skrzypek et al., 2016), since they are mostly too faint for Gaia and do not yet have proper motion estimates.

We confine our sample to all objects that have an optical or infrared spectral type equal to or later than L0 or are young late type M dwarfs that are probable brown dwarfs (e.g. TWA 27 A Gizis (2002)). These objects cover a large age range and include objects in the stellar, brown dwarf and giant-planet regimes. While Gaia is only observing directly a few objects later than L7, we included all published L, T and Y dwarfs, as the same list is used to search for common proper-motion objects in the Gaia DR1. Most UCDs (in particular late-M and early L dwarfs) have been classified using both their optical and near-infrared spectra, leading to two different and sometimes discordant spectral types. When we had to choose a spectral type, for example to calculate spectroscopic distances, when available we adopted optical spectral types for late-M and L dwarfs, since the wealth of spectral lines and bands in the 5 000–10 000 Å wavelength range makes the classification more accurate, while for T dwarfs we use their near-infrared spectral type following similar considerations.

The current version of the input catalogue contains 1885 entries. In Table 2 we list the short name, discovery name, equatorial coordinates, adopted spectral type, proper motions and $J$-band magnitude of the first five UCDs of the list. The full GUCDS input catalogue with references for each variable is available online at **TO BE FILLED BY MNRAS. ** This list will evolve, and be updated and maintained, as part of the GUCDS initiative in the MAIA database (Caballero, 2014).

2.2 Predicted $G$ magnitude

To first estimate a Gaia $G$ magnitude for the input catalogue, we used the procedure developed in Smart (2014). Briefly, we combined the Two Micron All Sky Survey (hereafter 2MASS; Skrutskie et al., 2006) $J$ magnitudes, Sloan Digital Sky Survey (hereafter SDSS; York et al., 2000) colours as a function of spectral type from Table 3 in Hawley et al. (2002), and colour transformations between Gaia photometry and the SDSS system from Jordi (2012) to find a predicted $G$ magnitude. To this table we fitted a simple linear polynomial of predicted $G$ magnitudes as a function of spectral type and $J$ magnitude to obtain:

[TABLE]

where SpT is the numerical representation of the LT types from 70 to 89 equivalent to L0 to T9.

The Jordi (2012) Gaia-to-SDSS transformations were based on main sequence stars in the colour range $g-r$ = (–0.5, 7.0) mag. There will be a systematic error in Eq. 1 due to the difference between M and LT dwarf spectral energy distributions, but we estimated this to be less than 0.2 mag by extrapolating the difference between M giants and dwarfs in the transformation construction. The transformation is imprecise because of the multiple steps, the use of 2MASS magnitudes and this systematic error. However, Eq. 1 was only used to constrain the objects that we search for, so we considered it sufficient. From the input list we searched the Gaia DR1 for all objects with a predicted magnitude $G_{\rm pred}<23$ mag. Since the nominal DR1 limit is $G=20.7$ mag, this allowed for significant random or systematic errors in our relationship and its parameters ($J$, SpT). Of the original 1885 objects, 1317 were brighter than this conservative $G_{\rm pred}<$ 23 mag cut. In the final matched catalog the faintest object had a $G_{\rm pred}=22.6$ mag.

3 Identification of DR1 matches

3.1 Initial matching

Since our input objects generally have high proper motions, and both the ground-based GUCDS input catalogue and the DR1 are of different epochs and with varying completeness, the identification of the LT dwarfs in the DR1 required a careful cross match. For each object, we matched the published position moved to the the DR1 epoch using the proper motions in our input list. We found that 328 of the 1317 UCDs had a DR1 entry within a matching radius of 3 arcsec. We also considered other matching radii both smaller (2 arcsec) and larger (5 arcsec), and found 3 arcsec to be the best compromise between too many false matches and missing true high proper motion objects. Of these, 6 objects had more than one DR1 match within 3 arcsec, and 8 had a non-zero duplicated_source flag in the DR1, which indicates that during the Gaia processing the source at some point was duplicated.

We then determined a new relationship for estimating $G$ magnitudes from 2MASS $J$ magnitudes and tabulated spectral types. We selected the 304  cross-matched L dwarfs that ($i$) had only one DR1 match within 3 arcsec, ($ii$) had a zero Gaia duplicated_source flag, and ($iii$) were earlier than L7. For this sub-sample, using least squared absolute deviation we found the first-order polynomial relationship between the colour $G-J$ and spectral type as:

[TABLE]

valid for SpT = 70 to 77, i.e. L0 to L7.

The colour-spectral type diagram in Fig. 2 illustrates the measured $G$ minus $J$ magnitudes with lines that represent $G_{\rm pred}$ (Eq. 1), and this new robust fit, $G_{\rm est}$ (Eq. 2). The new relation in Eq. 2 is much flatter than Eq. 1. We found 7 objects in that have a measured and estimated $G$ difference, $\Delta G=|G-G_{\rm est}|$, larger than 1 mag. While the underestimation of the $G-J$ for the T6 object indicates extrapolating the fit beyond L7 provides uncertain results, we only used this $\Delta G$ flag as an indicator of possible problems.

Fig. 3 is the distribution of the input catalogue in $G_{\rm est}$ magnitudes using Eq. 2 for the input catalogue and the 328 matched objects with measured $G$ magnitudes. The degree of completeness varies greatly from 0 % in the bright bins below $G=16.5$ mag and the faint bins beyond $G=21$ mag, to over 50 % at $G=19$ mag. The brightest bins have the objects with the highest proper motions so are systematically effected by Gaia observation matching problems (Fabricius et al., 2016). In general, the incompleteness can be attributed to objects that were excluded from DR1, matching problems due to imprecise positions and/or proper motions or mis-classifications in the GUCDS input catalogue leading to over estimated $G_{\rm est}$ magnitudes.

In total, there are 1010 L and 58 T dwarfs brighter than $G_{\rm est}=21.5$ mag, and 543 L and 10 T dwarfs brighter than $G=20.3$ mag. In contrast, Smart (2014) predicted only two T dwarfs to $G=20.3$ mag. The higher number estimated in this work is due to the systematic underestimation of the $G_{\rm pred}$ (Eq. 1) used in Smart (2014) with respect to the $G_{\rm est}$ (Eq. 2). Using a more theoretical approach, Sarro et al (2013) predicted of the order of 10 T dwarfs brighter than $G=20.0$ mag.

In Fig. 4 we plot the sky distribution of all the input catalogue with $G_{\rm est}<$ 21.5 mag using Eq. 2. The region of over density in the northern hemisphere is from the SDSS footprint, and is probably representative of a complete sky (Schmidt et al., 2010). However, the Galactic plane is incomplete, as most of the LT dwarfs discovered to date have been via photometric selection, and the crowding in the plane makes this difficult.

Of the 6 objects that had more than one DR1 entry within 3 arcsec some may be due to binarity or a background object near to the catalogue dwarf, but most are due to multiple entries in the DR1 (see Section 4 in Gaia Collaboration et al., 2016b). It is estimated that the multiple entries in the DR1 catalogue is a few percent (Fabricius et al., 2016), consistent with this finding.

In the GUCDS input catalogue, objects either have published proper motions or we estimated them from the 2MASS and Wide-field Infrared Survey Explorer AllWISE444http://irsa.ipac.caltech.edu/data/download/wise-allwise/ positions (Wright et al., 2010). We compared these input values with a derived proper motion from the difference of the Gaia DR1 and the 2MASS position. When the magnitude of the proper motions differed by more than 20 %, we flagged the object. This resulted in 145 objects being flagged, i.e. $\sim$ 50 %. This high percentage is not unexpected given that both proper motions are of low precision and the parallactic motion of the object is unknown.

3.2 Identifying mismatches

Since Gaia does not produce images (in general), we cannot perform the usual visual confirmation to look for mismatches. We confined our examination for mismatches to the catalogue maps and various flags. For each target we constructed a quality assurance output including: the flags for proper motion and $|G-G_{est}|$ magnitude differences; number of DR1 observations; the 2MASS images; positions and magnitudes from the 2MASS, AllWISE and DR1 catalogues; the input spectral types; parallaxes when published; input comments (e.g. known binarity or subdwarf); literature and calculated proper motions; and plots of the fitted proper motions and sky maps for the field in both 2MASS, AllWISE and DR1 catalogues.

In Fig. 5 we show an example of the sky distribution plots for the field around the T6 J0817-6155 (Artigau et al., 2010). The slight misalignment between the cross and the square in the DR1 panel is due to imprecise starting proper motions from the input catalogue. When needed, we also examined online ground based images of the fields.

We examined all 328 targets to see if any of the candidate were obvious mismatches. In particular, we paid special attention to the objects with large magnitude differences, multiple DR1 entries within 3 arcsec, and the 10 objects later than L7. Of the 328 targets we identified 3 objects that we believe are mismatches and are listed in Table 3. Most of these had $\Delta G>1.0$ mag, proper motion differences larger than 20 %, and/or visual inspection of the field did not allow an unambiguous identification. The Gaia second data release is expected to resolve these ambiguities.

3.3 DR1 multiple matches

There are 6 LT dwarfs with multiple matches within 3 arcsec. Three, J0257-3105 (Kirkpatrick et al., 2008), J0543+6422 (Reid et al., 2008) and J1515+4847 (Wilson et al., 2003), are matched to DR1 entries within only 1 arcsec and the DR1 entries have very similar magnitudes. The matches to these three are probably duplicated DR1 entries (see discussion in Fabricius et al., 2016), and we adopted the DR1 entry with the highest number of observations. The candidate J1203+0015 (Fan et al., 2000) is matched to two entries with significantly different magnitudes (both fainter than the estimated Gaia magnitude) so it is probably either close to a background object, or Gaia has resolved the dwarf into a binary system with a 0.3 arcsec separation. The targets J1606-2219 and J1607-2211 (Lodieu et al., 2007) have fainter detections 2-3 arcsec away, which we believe to be background objects. In these last three cases we adopted the match closest to the predicted LT dwarf position.

3.4 Completeness

If we consider input catalogue objects with $G_{\rm est}<20.3$ mag we find only 45 % in the Gaia DR1. This incompleteness is due primarily to the quality assurance cuts of Gaia which are: $N>5,\epsilon_{i}<20$ mas, and $\sigma_{\rm pos,max}<100$ mas, where $N$ is the number of field-of-view transits used in the solution, $\epsilon_{i}$ is the excess source noise, and $\sigma_{\rm pos,max}$ is the semi-major axis of the error ellipse in position at the reference epoch (from Section 5 in Lindegren et al., 2016). In addition, we required all included objects to have valid photometry. The number of field-of-view transits led to a systematic incompleteness that follows the scanning law and can be seen in the sky plots of Gaia  DR1555http://sci.esa.int/gaia/58209-gaia-s-first-sky-map. Importantly for these objects, the cyclic processing does not yet use internal proper motions to update the position of objects during the matching, so the correct matching of high proper motion objects is deficient (Fabricius et al., 2016). Given the documented incompleteness of 50 % at $G=20.3$666http://gaia.esac.esa.int/documentation/GDR1/ mag, and the very high proper motion of most bright LT dwarfs, we consider the success rate of 45 % to be reasonable. The matching for DR2 will include internal proper motions, so it will not have this deficiency.

3.5 Gaia observed L and T dwarfs Catalogue

We produced a catalogue of the parameters for the 321  L and T dwarfs with a reliable entry in Gaia DR1, which are distributed as shown in Fig. 6. This will be actively updated on-line along with the input catalogue. In Table 4 we report new parameters for the first five objects from this catalogue table with: DR1 positions; calculated proper motions with errors; $G$ magnitudes and errors; number of observations in DR1; Gaia SourceID; $\Delta G=G-G_{\rm est}$; number of DR1 entries within 3 arcsec and a flag that indicates if the calculated proper motion was within 20% of the published or estimated value. The published catalogue also has other literature information such as 2MASS and WISE magnitudes for each entry.

4 Common proper motion LT dwarfs and DR1 stars

4.1 The Tycho-Gaia astrometric subset

The Gaia DR1 included a subset of more than 2 million objects that incorporated earlier positional information to find parallaxes and proper motions called the Tycho-Gaia astrometric solution (TGAS; Michalik et al., 2015; Gaia Collaboration et al., 2016b). We selected CPM pairs by cross-matching our catalogue of known LT dwarfs with the TGAS subset. For the input LT dwarfs we used measured parallaxes from the literature, complemented with spectrophotometric distances estimated using the adopted UCD spectral types and near-infrared magnitudes. To estimate their spectrophotometric distance we used the polynomial relations presented in Dupuy & Liu (2012), with the measured 2MASS $J$ magnitude, and, if not available or the 2MASS value has a bad quality flag (Qflag = U), we use the MKO $J$ magnitude.

4.2 Selection criteria

The starting CPM candidate list was generated from finding all TGAS stars within 2 deg of our input LT dwarfs and applying the following criteria:

• •

$\mu$ $>$ 100 mas yr*-1*

• •

$\Delta\mu_{\alpha}\cos{\delta}<\,$20 mas yr*-1* and $\Delta\mu_{\delta}\,<\,$20 mas yr*-1*

where $\mu$ is the total proper motion, $\Delta\mu_{\alpha}\cos{\delta},\Delta\mu_{\delta}$ are the difference between the proper motion components of the UCD and the TGAS star. All selected TGAS objects are close so we do not need to invoke inference techniques to find distances (e.g. Bailer-Jones, 2015), but use the simple inverse of the parallax as the estimated distance and as its error a proportion equal to the relative error of the parallax. We then calculated a chance alignment probability for each system following the method described in Marocco et al. (2017).

The selection criteria requires the objects to have relatively high proper motions and the probability of having two objects with such high proper motions in a limited area is already small. For each candidate pair we used the sample of all TGAS field stars in a radius of 2 deg from the UCD to determine the distance and proper motion distribution of the field population. The distance and proper motion distribution were treated as a probability density function, which we reconstructed using a kernel density estimation. We then draw 10,000 samples of stars from the reconstructed probability density function, and determined how many “mimics” of our system were generated. We considered as a mimic of our CPM system any star within 3$\sigma$ of the distance and proper motion of our selected primaries. The chance alignment probability was assumed to be the number of mimics divided by 10,000. If this probability was below 6 $\times$10*-5*, equivalent to a 4$\sigma$ level, we consider the pair to be a “robust” common proper motion system. Systems with larger chance alignment probability were ruled out.

4.3 LT dwarfs and Gaia CPM system catalogue

This selection yielded a sample of 32 CPM pair candidates. We compiled a list of known binary and CPM systems by combining the objects and list from the following publications: Mason et al. (2001), Deacon et al. (2014), De Rosa et al. (2014), Dhital et al. (2015), Gauza et al. (2015), Smith et al. (2015), Scholz (2016), Kirkpatrick et al. (2016), Gálvez-Ortiz et al. (2016), and Deacon et al. (2017). Of the 32 CPM pair candidates 17 were previously known and the remaining 15, listed in Table 4.3, are presented here for the first time. The majority of new wide systems presented here are not physically bound pairs, but the low chance alignment probabilities we interpret as an indication of common origin. Intrinsically wide binaries and multiple systems can in fact become unbound due to Galactic tides and close encounters (e.g. Veras, 2016; Elliott & Bayo, 2016), and their ejecta would represent a new, as yet unexplored pool of benchmark systems (Pinfield et al., 2006; Yip et al., 2016). In Figs. 8 and 8 we plot celestial and proper motion distributions of example unbound (J0230-0225) and bound (J1632+3505) CPM pairs discussed later.

One key element in our selection process is the requirement of common distance between the main sequence TGAS star and its potential companion. We show in Figure 9 a comparison between the measured astrometric distance to the primaries in our CPM pairs, against the distance (astrometric or spectrophotometric) to their potential companions. Common distance systems are highlighted in black. Uncertainties on the spectrophotometric distance dominate, and at larger distance this results in a much larger scatter around the one-to-one correspondence line. Pairs that passed our angular separation constraint, but were rejected by the common-distance cut, consist of a foreground UCD matched to a background star. In Figure 10 we plot only those systems that we select as having common distance, with those UCDs with measured parallaxes highlighted in green. As expected, systems with astrometric measurements are much closer to the one-to-one correspondence line than those with spectrophotometric distance estimates only. The UCD spectroscopic distances tend to be underestimated compared to the TGAS parallactic distances, e.g. the UCD is brighter than the spectral type indicates. This is as expected from unresolved binarity or a Malmquist bias like effect as our input sample is probably biased to the brighter examples of a given spectral class bin.