A Reanalysis of the Fundamental Parameters and Age of TRAPPIST-1

We present the distance-calibrated spectral energy distribution (SED) of TRAPPIST-1 using a new medium resolution (R~6000) near-infrared FIRE spectrum and its Gaia parallax. We report an updated bolometric luminosity (Lbol) of -3.216+/-0.016, along with semi-empirical fundamental parameters: effective temperature Teff=2628+/-42 K, mass=90+/-8 MJup, radius=1.16+/-0.03 RJup, and logg=5.21+/-0.06 dex. It's kinematics point toward an older age while spectral indices indicate youth therefore, we compare the overall SED and near-infrared bands of TRAPPIST-1 to field-age, low-gravity, and low-metallicity dwarfs of similar Teff and Lbol. We find field dwarfs of similar Teff and Lbol best fit the overall and band-by-band features of TRAPPIST-1. Additionally, we present new Allers&Liu 2013 spectral indices for the SpeX SXD and FIRE spectra of TRAPPIST-1, both classifying it as intermediate gravity. Examining Teff, Lbol, and absolute JHKW1W2 magnitudes versus optical spectral type places TRAPPIST-1 in an ambiguous location containing both field- and intermediate-gravity sources. Kinematics place TRAPPIST-1 within a subpopulation of intermediate-gravity sources lacking bonafide membership in a moving group with higher tangential and UVW velocities. We conclude that TRAPPIST-1 is a field-age source with subtle spectral features reminiscent of a low surface gravity object. To resolve the cause of TRAPPIST-1's intermediate gravity indicators we speculate two avenues which might be correlated to inflate the radius: (1) magnetic activity or (2) tidal interactions from planets. We find the M8 dwarf LHS 132 is an excellent match to TRAPPIST-1's spectral peculiarities along with the M9 beta dwarf 2MASS J10220489+0200477, the L1 beta 2MASS J10224821+5825453, and the L0 beta 2MASS J23224684-3133231 which have distinct kinematics making all three intriguing targets for future exoplanet studies.


Introduction
The majority of stars in our galaxy are low mass, M dwarfs being the most numerous with the longest main-sequence lifetime (Bochanski et al. 2010). Their low mass and abundance in the solar neighborhood make M dwarfs favorable targets for exoplanet observations. Their small radii enable easier detection of Earth-sized planets using transit and radial velocity methods; therefore they are prime targets when searching for rocky planets within a star's habitable zone.
With numerous searches for exoplanets-such as Kepler (aimed at detecting planets around Sun-like stars Borucki et al. 2010) and TESS (an all-sky survey searching for planets smaller than Neptune around nearby stars, Ricker et al. 2015)-understanding stellar properties of M dwarfs as exoplanet host stars is extremely pertinent to understanding planet habitability. Kane et al. (2016) found 40% of all Kepler planet candidates, (1) with radii less than 2 R ⊕ and (2) lying within an optimistically sized habitable zone, orbit stars cooler than 4000 K. This is despite cool dwarfs being less than 5% of initial Kepler targets (Batalha et al. 2010). Mulders et al. (2015) and Gaidos et al. (2016) determined planets with radii of 1-3 R ⊕ occur two to four times more often around M dwarfs than around FGK stars. Furthermore, Dressing & Charbonneau (2015) estimate the frequency of habitable Earths around M dwarfs to be 2.5±0.2 planets. Such objects would have radii of 1-4 R ⊕ and periods shorter than 200 days. They calculate an occurrence rate of -+ 0.56 0.05 0.06 Earth-sized planets with periods shorter than 50 days and -+ 0.46 0.05 0.07 super-Earths (1-1.5 R ⊕ ) with periods shorter than 50 days per early-type M dwarf. Ballard (2019) predicts that TESS will detect 900±350 planets around 715±255 M dwarfs spectral typed M1V-M4V.
Four nearby mid-to late-type M dwarfs with habitable zone planets are 2MASS J23062928−0502285 (hereafter TRAP-PIST-1), Proxima Centauri, LHS 1140, and Teegarden's Star. TRAPPIST-1, a M7.5 dwarf at a distance of ∼12.4 parsecs, hosts a system of seven rocky Earth-sized exoplanets (Gillon et al. 2016(Gillon et al. , 2017 with four lying in the habitable zone. Proxima Centauri, a moderately active M5.5 dwarf, is our nearest stellar neighbor 1.295 parsecs away. It hosts an Earthsized planet (1.3 M ⊕ ) that could have liquid water on its surface (Anglada-Escudé et al. 2016). LHS 1140, a metal-poor M4.5 dwarf older than 5 Gyr and about 15 parsecs away, hosts a Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
super-Earth (∼7 M ⊕ ) and an Earth-sized planet (∼2 M ⊕ ) with an Earth-like composition (Dittmann et al. 2017;Ment et al. 2019). Teegarden's Star, an old magnetically quiet M6.5 dwarf located 3.8 parsecs away, was classified as intermediate gravity by Gagné et al. (2015a) and hosts two Earth-sized planets, both within the conservative habitable zone (Zechmeister et al. 2019). Having the most precise stellar parameters for these systems is critical for understanding planet habitability because stellar size and temperature drive the habitable zone boundaries.
While most M dwarfs are stars, late-type M dwarfs can be either stars or brown dwarfs depending on age. Brown dwarfs are low-mass, low-temperature objects unable to sustain stable hydrogen burning in their cores and thus cool throughout their lifetime. With masses <75 M Jup , brown dwarfs lie between the boundary of low-mass stars and planets (Saumon et al. 1996;Chabrier & Baraffe 1997). They typically fall into three main age subpopulations: field dwarfs, low surface gravity dwarfs, and subdwarfs. Field dwarfs anchor the brown dwarf spectral classification system, while the low-gravity and subdwarfs show differences in their observed spectra deviating them from the field classification. Red infrared colors, enhanced metal oxide in the optical, and weak alkali lines differentiate lowgravity dwarfs from the field sources (Kirkpatrick et al. 2006(Kirkpatrick et al. , 2010Cruz et al. 2009;Allers et al. 2010). High-likelihood or candidate membership in young nearby moving groups has been seen for many low-gravity sources (Liu et al. 2013(Liu et al. , 2016Faherty et al. 2016;Kellogg et al. 2016;Schneider et al. 2016;Gagné et al. 2017). Low-gravity dwarfs are further classified into two gravity groups-(1) very low gravity, designated by γ in the optical or by VL-G in the infrared, and (2) intermediate gravity, designated by β in the optical or INT-G in the infrared. Subdwarfs are low-luminosity metal-poor sources that exhibit blue near-infrared (NIR) colors (Burgasser et al. 2003), kinematics consistent with halo membership (Burgasser et al. 2008b;Dahn et al. 2008;Cushing et al. 2009), enhanced metalhydride absorption bands along with weak or absent metal oxides, and enhanced collision-induced H 2 absorption (Burgasser et al. 2003 and references therein).
In this paper we examine the fundamental parameters of TRAPPIST-1 to determine whether it is a typical field M dwarf host star or more akin to low-gravity or subdwarf equivalents. Previously published data on TRAPPIST-1 are presented in Section 2. New Folded-port InfraRed Echellette (FIRE) and SpeX SXD spectra observations are discussed in Section 3. Section 4 presents the motivation of our chosen samples and analysis process. Section 5 discusses how we derive the fundamental parameters of TRAPPIST-1 using its distanced-calibrated spectral energy distribution (SED) and the Filippazzo et al. (2015) method. Section 6 discusses comparative samples for TRAPPIST-1. Sections 7 and 8 examine the full SED of TRAPPIST-1 and the Y, J, H, and K NIR spectra of comparative objects. Section 9 examines whether an extreme subsolar metallicity for TRAPPIST-1 might explain anomalous spectral features. Section 10 makes concluding remarks on the age of TRAPPIST-1 after examining all samples. Section 11 examines LHS 132 and other sources from Burgasser & Mamajek (2017), comparing their overall SEDs to TRAPPIST-1. Lastly, Section 12 presents Allers & Liu (2013b) indices and gravity classifications for the entire comparative sample, absolute magnitude and fundamental parameters versus spectral type comparisons, and a comparison of the kinematics of TRAPPIST-1 to β gravity sources. We also present possible reasons for the gravity classification that TRAPPIST-1 receives.

The Discovery of an M Dwarf
TRAPPIST-1 was discovered by Gizis et al. (2000) as part of a search using optical and NIR sky survey data from the Second Palomar Sky Survey (POSS-II) and the Two Micron All Sky Survey (2MASS). Gizis et al. (2000) spectral typed TRAPPIST-1 as an M7.5 based on its optical spectrum taken by the Ritchey-Chrétien (RC) spectrograph at Kitt Peak on the 4 m telescope. Additional optical spectra of TRAPPIST-1 are presented in Cruz et al. (2007), Schmidt et al. (2007), Reiners & Basri (2009), and Burgasser et al. (2015). NIR spectra are presented in Tanner et al. (2012), Bardalez Gagliuffi et al. (2014, Cruz et al. (2018), and this paper (FIRE). Thus, there are currently five optical and four NIR spectra of TRAPPIST-1.
Gaia DR2 (Gaia Collaboration et al. 2016Lindegren et al. 2018) provides the most precise proper motion and parallax values for TRAPPIST-1, although it was also observed by numerous other surveys (Costa et al. 2006;Schmidt et al. 2007;Weinberger et al. 2016;Boss et al. 2017;Van Grootel et al. 2018). Radial velocity measurements of TRAPPIST-1 were reported in Reiners & Basri (2009), Tanner et al. (2012 (using the NIRSPEC spectrum), and Burgasser et al. (2015) (using the MagE spectrum). In this paper we present updated UVW velocities using updated position and parallax values from Gaia DR2 paired with the Tanner et al. (2012) radial velocity.
Bolometric luminosity, effective temperature, radius, mass, gravity, and age for TRAPPIST-1 were initially determined in Filippazzo et al. (2015) via distance-scaled SED fitting. After the discovery of TRAPPIST-1 as an exoplanet host star with seven rocky planets, Gillon et al. (2017) presented fundamental parameters from a Markov chain Monte Carlo analysis with a priori knowledge of stellar properties from Filippazzo et al. (2015). Using their improved parallax measurement, Van Grootel et al. (2018) determined a bolometric luminosity almost two times as precise as the Filippazzo et al. (2015) value and used that to determine more accurate values of T eff , radius, and mass for TRAPPIST-1. Values for published data from the literature are listed in Table 1.

Discovery as an Exoplanet Host Star
TRAPPIST (the TRansiting Planet and PlanetIsimals Small Telescope) monitored TRAPPIST-1 in the NIR from mid 2015 September to the end of 2015 December. Follow-up photometry in the optical on the Himalayan Chandra 2 m Telescope and in the NIR with the Very Large Telescope and UK Infrared Telescope helped to confirm signatures of three exoplanets: TRAPPIST-1b, TRAPPIST-1c, and TRAPPIST-1d (Gillon et al. 2016). TRAPPIST-1d was later identified by follow-up observations to be the signature of four planets: TRAPPIST-1d, TRAPPIST-1e, TRAPPIST-1f, and TRAPPIST-1g, along with the discovery of TRAPPIST-1h (Gillon et al. 2017). All seven of the planets have Earth-size radii, ranging from 0.755 to 1.127 R Earth , with four of the planets, TRAPPIST-1d, -1e, -1f, and -1g, lying in the habitable zone (Gillon et al. 2017). These planets are in a zone where temperatures are cool enough to potentially have long-lived liquid water present on the surface.

FIRE
We used the FIRE (Simcoe et al. 2013) spectrograph on the 6.5 m Baade Magellan telescope to obtain NIR spectra of TRAPPIST-1. Observations were taken on 2017 July 28 using echellette mode and the 0 45 slit, high gain (1.2 e-/DN), and sample-up-the-ramp readout covering the full 0.8-2.5 μm band. Each exposure was 600 s long with ABBA nodding and was bracketed by quartz lamp, ThAr lamp, and telluric standards. Airmass ranged from 1.7 to 1.0 and seeing from 0.7 to 1.1 over the night. The night sky emission lines were significantly larger than the internal Th-Ar comparison lines, even though the system was in best focus; this suggests that FIRE was not properly collimated at the time of observation. The result is that the resolving power we obtained was λ/Δλ∼6000, not the R∼8000 that should have been obtained. This causes lines in our data to be shallower but wider with the equivalent widths preserved. All FIRE exposures of TRAPPIST-1 obtained on 2017 UT July 27 outside of transit were combined and reduced with the Interactive Data Language (IDL) FireHose v2 package 9 (Bochanski et al. 2009;Gagné et al. 2015b), as described in Gagné et al. (2015a).

SpeX
The following new SpeX spectra were obtained for objects we use in a comparative analysis to TRAPPIST-1.

SXD
A SpeX SXD spectrum was obtained for 2MASS J06085283 −2753583 on 2007 November 13 across the 0.7-2.55 μm region. The 0 5 slit was used, providing a resolving power of λ/Δλ∼1200. We obtained two 200 s exposures and four 300 s exposures for a total integration time of 20 minutes using ABBA nodding. The slit was aligned to the parallactic angle, and we observed at an airmass of1.49. The spectrum was telluric corrected and flux calibrated using the spectrum of the A0V standard HD52487 taken at a similar airmass. Internal flat-field and Ar arc lamp exposures were taken for pixel response and wavelength calibration. The data were then reduced using standard procedures and the SpeXtool Package (Cushing et al. 2004).

Is TRAPPIST-1 Young?
Burgasser & Mamajek (2017) noted that TRAPPIST-1 exhibited weaker FeH absorption and a more triangular H band, features that are typically associated with youth (see Faherty et al. 2016 for details). Using the Allers & Liu (2013b) surface gravity indices, they determined the low-resolution SpeX spectrum of TRAPPIST-1 displayed signs of an intermediate-gravity object, suggesting a young age. However, from examination of the kinematics and the lack of enhanced VO absorption in the SpeX prism spectrum, Burgasser & Mamajek (2017) concluded that the β gravity classification may have arisen from other physical factors and thus is unrelated to youth. Indeed, all previous studies (Gizis et al. 2000;Faherty et al. 2009;Filippazzo et al. 2015) indicated that TRAPPIST-1 was a field object.
The low-gravity indicators in the spectrum of TRAPPIST-1 motivate the remainder of the paper, where we create comparison subsamples to examine whether youth or some other physical factor drives the observed parameters. Throughout our comparisons we examine whether the overall SED shape or specific features in the new FIRE medium-resolution NIR spectrum of TRAPPIST-1 show signs of low surface  Notes. WISE = Wide-field Infrared Survey Explorer. a Not in LSR frame. b We account for gravitational reddening assuming 0.5 km s −1 . References.
(1) Cutri et al. gravity. We also determine gravity indices for all objects in our sample as another aspect of our comparison to TRAPPIST-1, which is discussed in depth in Section 12.1.

Fundamental Parameters of TRAPPIST-1
Fundamental parameters for TRAPPIST-1 were determined from its distance-calibrated SED using the technique of Filippazzo et al. (2015). Parameter values were determined using SEDkit, 10 which requires spectra, photometry, and a parallax to create the distance-scaled SED and to determine the bolometric luminosity. The spectra, photometry, and parallax used in the generation of the SED of TRAPPIST-1 can be found in Table 1 and Tables 6-10 in the Appendix.
Using the optical and NIR spectra, we first construct a composite spectrum of TRAPPIST-1. The overlapping region from 0.8305 to 0.95 μm (shown as a fuzzy blue-green line in Figure 1) was combined as an average. The composite spectrum is then scaled to the absolute magnitudes in each filter. We do not create synthetic magnitudes, those determined through empirical relations in the absence of photometric data as in Filippazzo et al. (2015); instead we only use observed photometric data. The SED of TRAPPIST-1 is shown in Figure 1, with the various components labeled.
The bolometric luminosity (L bol ) was determined by integrating under the distance-calibrated SED from 0 to 1000 μm, using a distance of 12.43±0.02 pc. To obtain a radius estimate and the effective temperature (T eff ), we used the DMEstar models (Feiden & Chaboyer 2012) to extend the brown dwarf Saumon & Marley (2008) hybrid cloud evolutionary models into the low-mass stellar range. They were connected via a cubic spline interpolation in the regions with no evolutionary model coverage. The same was done using the Chabrier et al. (2000) evolutionary models. For all three models we use an age range of 0.5-10 Gyr, corresponding to the field ultracool dwarf age range (Filippazzo et al. 2015). The final radius range was set as the maximum and minimum from all model predictions as done in Filippazzo et al. (2015). The effective temperature was calculated using the inferred radius along with the bolometric luminosity using the Stefan-Boltzmann law. The uncertainty on the T eff comes primarily from the uncertainty in the age of the system (leading to the range of radii) and the SED flux measurement. However, as noted in Dupuy & Kraus (2013) slight differences in radii do not have a large effect on the calculated T eff . The range of masses was determined using the Saumon & Marley (2008) and Chabrier et al. (2000) evolutionary models.
Using this approach, we derived the following parameters assuming a field age: L bol =−3.216±0.016, T eff =2628± 42 K, R=1.16±0.03 R Jup , M=90±8 M Jup , log g=5.21 ±0.06 dex. Given the speculation on the age of TRAPPIST-1, we also repeat this process and assume an age range of 0-0.5 Gyr to address the intermediate-gravity classification. The fundamental parameters derived for TRAPPIST-1 for both age assumptions are also listed in Table 2, which compares our values to literature values. Table 2 contains our calculated fundamental parameters assuming two age ranges (1) 0.5-10 Gyr, the field-age constraint from Filippazzo et al. (2015); and (2) <0.5 Gyr, to address the intermediate-gravity classification. Also listed in Table 2

Sample Selection and Properties
To determine whether TRAPPIST-1 exhibits similar or different features compared with the average field M dwarf, we constructed three comparative samples.
1. Sample 1 assumes TRAPPIST-1 is a field-age (0.5-10 Gyr) source and contains comparative field, very low-gravity (γ), and old dwarfs with similar effective temperatures (within 1σ of TRAPPIST-1) and/or bolometric luminosity (within up to 2.5σ) of TRAPPIST-1. 2. Sample 2 assumes an age of <0.5 Gyr for TRAPPIST-1 and contains field, very low-gravity (γ), and old dwarfs with similar effective temperatures within 1σ. 3. Sample 3 assumes TRAPPIST-1 is field age and contains subdwarfs of varying T eff and L bol (sdM7 and later) with medium-resolution data (R∼5000).
The effective temperature samples compare the overall SED shape and spectral features of TRAPPIST-1 to objects with similar atmospheric chemistry. The bolometric luminosity sample examines how the flux of TRAPPIST-1 is redistributed across different wavelengths compared with comparable L bol sources. The subdwarf sample explores the Na I and K I doublets in the J band to see whether TRAPPIST-1 exhibits similarities with these metal-poor old objects.
Objects in each comparative sample were chosen from (1) Filippazzo et al. (2015), which examined a large sample of field and low-gravity objects, (2) Faherty et al. (2016), which examined a large sample of low-gravity sources, or (3) Gonzales et al. (2018), which examined subdwarfs later than sdM6 with parallax measurements. The bolometric luminosities in each sample were empirically derived, while their effective temperatures were semiempirically derived using radii from evolutionary models depending on the age. Ages were determined by updated membership in moving groups (J. K. Faherty et al., in preparation) for young sources from Faherty et al. (2016). For field sources from Filippazzo et al. (2015) we use their field dwarf age range of 0.5-10 Gyr, and for subdwarfs from Gonzales et al. (2018) we use their subdwarf age range of 5-10 Gyr. Exact evolutionary models used for radii values are listed in Filippazzo et al. (2015), Faherty et al. (2016), and Gonzales et al. (2018).
In order to resolve features in a band-by-band comparative analysis of the spectra, we required objects to have mediumresolution NIR data (λ/Δλ>1000 at J band). Field dwarf comparison SEDs chosen from Filippazzo et al. (2015) were reconstructed with the same data used in that work, with the exception of replacing their low-resolution SpeX data with medium-resolution NIR data in this work. Low-gravity dwarf comparison SEDs from Faherty et al. (2016) were constructed with new NIR FIRE data or rereduced spectra. Subdwarf comparison SEDs were constructed with the same data from Gonzales et al. (2018), with the exception of GJ 660.1B, which was not part of that sample; SSSPM J1013−1356, which only uses the NIR spectrum in this work; and LSR 2036+5059, which includes synthetic WISE photometry points for proper appending of the Rayleigh-Jeans tail.
All SEDs constructed for this paper used measured photometric values alone, with the exceptions listed above. We did not incorporate synthetic photometry as was done by Filippazzo et al. (2015) and Faherty et al. (2016). We also used updated parallax measurements from Gaia DR2 for objects chosen from Filippazzo et al. (2015) and Faherty et al. (2016). Therefore our values differ slightly but fall within ∼1σ for all values except bolometric luminosity.
Sample 1 contains five field sources that have similar T eff and L bol , three sources with similar T eff , and three with similar L bol , for a total of eight sources when examining T eff or L bol . Sample 2 contains eight sources-three field dwarfs, three subdwarfs, and two low-gravity sources. Sample 3 contains the five subdwarfs from Gonzales et al. (2018), which have medium-resolution J-band data. Details for all comparative objects are listed in Table 3. The corresponding parallaxes, photometry, and spectra used in the creation of the SEDs are listed in Tables 6-10 in the Appendix. Table 4 lists the derived fundamental parameters for TRAPPIST-1 along with the comparative objects.
7. Spectral Analysis for Sample 1: Assuming a Field Age for TRAPPIST-1

Full SED Comparisons
Under the assumption that TRAPPIST-1 is field age, we present the overall SEDs for objects of equivalent temperature Notes. The effective temperatures in this paper and Filippazzo et al. (2015) are determined from measured L bol combined with an assumed age and radii ranges obtained from evolutionary models. Values not noted are either not used or derived in that work. a Value from Gillon et al. (2016), where the constraint of >0.5 Gyr is from Filippazzo et al. (2015). b Parallax from Costa et al. (2006). Notes. a We refer to any objects classified as VL-G as γ and INT-G as β sources throughout this work when making reference to the NIR spectral type. b Also part of the Youth Assumption sample. c Spectroscopic binary. and/or bolometric luminosity to that of TRAPPIST-1 in Figures 2 and 3. The sample is composed of three age subsets: field, low-gravity (young), and subdwarfs (old). All objects of similar T eff are within 1σ of TRAPPIST-1 and of similar L bol are within 2.5σ. Figure 2(a) compares the overall SED assuming an older age for TRAPPIST-1 with field dwarfs of similar T eff and L bol across 0.3-31 μm. All comparative field dwarfs in the sample are within one spectral type of TRAPPIST-1. From 0.65 to 1 μm there is a spread in the SEDs (zoomed version in Figure 3(a)); however, this spread does not appear to correspond to temperature or bolometric luminosity of the objects. There appears to be a tighter temperature-dependent sequence from 1.28 to 1.8 μm, which can be seen in Figure 3(b) most clearly from the start of the H band out to 1.70 μm. Other than the large spread in the red optical, the overall SED shape of TRAPPIST-1 is similar to all sources in this subsample. Figure 2(b) compares TRAPPIST-1 to very low-gravity (γ) dwarfs of similar T eff sample across the 0.5-23 μm region. The low-gravity comparison objects are not members of known moving groups so we cannot give a definitive age bracket for them; however, their spectra show strong indications of youth such as triangular H band and weakened FeH (see Faherty et al. 2016;J. K. Faherty et al., in preparation). From ∼0.80 to 1 μm, we see TRAPPIST-1 directly on top of J0608−2753; however, from J band onward all low-gravity sources are much brighter than TRAPPIST-1.
The SEDs of low-gravity objects with equivalent L bol (within 2.5σ) of TRAPPIST-1 across the 0.5-14 μm region are shown in Figure 2 Faherty et al., in preparation). From 0.5 μm to the start of the J band, TRAPPIST-1 is brighter than all low-gravity sources, while in the J and H bands J0443+0002 and TRAPPIST-1 are of similar brightness. Beyond W1 TRAPPIST-1 is fainter than the lowgravity sources.
Overall under the assumption of field age for TRAPPIST-1, comparing the shape to field, low-gravity, and subdwarfs leads to the strong conclusion that this is a field source.

Band-by-band Analysis
While the overall SED comparisons give a general examination of where flux is matched between sources, the subtle but  Note.The effective temperature is determined based off age estimates and evolutionary models. significant feature detail is lost. Detailed line analysis can be used to tease out signatures of gravity or metallicity; therefore in this subsection we compare NIR band-by-band features.

Similar T eff
Figure 4(a) shows the 0.95-1.10 μm Y-band data with the FeH, VO, and H 2 O features labeled. The spectra for the Y band were normalized by averaging the flux taken across the relatively featureless 0.98-0.988 μm region. From first glance, we see that TRAPPIST-1 appears to be most similar in shape in the Y band to the low-gravity dwarf J0436−4114 and the field dwarf vB 10. The slope from 0.95 to 0.99 μm of TRAPPIST-1 is most similar to those of the field dwarfs; however, it also matches the slope of the low-gravity dwarf J0436−4114. The Wing-Ford FeH band head of TRAPPIST-1 appears to be of similar depth to J0436−4114, but only slightly shallower than vB 8 and vB 10. From 1 to 1.05 μm the spectrum of TRAPPIST-1 overlaps that of J0436−4114 but shares the shape of the field dwarfs more than that of the low-gravity dwarfs. There is an indication of slight VO absorption in the spectrum of TRAPPIST-1, similar in depth to the low-gravity dwarf J0436−4114. From all of these features we therefore conclude the TRAPPIST-1 Y-band spectrum exhibits a blend of both field and low-gravity dwarf features. Figure 4(b) shows the 1.12-1.35 μm J-band data with FeH and H 2 O molecular features, as well as the Na I and K I alkali doublets labeled. The J-band spectra were normalized by the average flux over the featureless 1.29-1.31 μm region. The depth of the Na I and K I doublets of TRAPPIST-1 are slightly deeper than the field dwarfs and much deeper than the lowgravity dwarfs. The FeH absorption of TRAPPIST-1 is similar to the field dwarfs, while beyond the 1.25 μm K I doublet the shape is similar for all objects in the sample. TRAPPIST-1 has some J-band spectral features similar to field dwarfs, while others differ from both the field and low-gravity dwarfs.
Figure 4(c) shows the 1.42-1.80 μm H-band data with the FeH and H 2 O molecular features labeled. The H-band spectra were normalized by the average flux over the featureless 1.5-1.52 μm region. The H-band shape of TRAPPIST-1 is similar to the field objects; however, its overall shape is also similar to J0436−4114 but slightly less triangular. There are no features in the H-band spectrum of TRAPPIST-1 that match those of the low-gravity dwarf J0608−2753. In the H band, TRAPPIST-1 is more like a field object.  K-band spectra were normalized by the average flux over the 2.16-2.20 μm region due to the relatively flat spectral region. TRAPPIST-1 has a visible Na I doublet like the field dwarfs, while the low-gravity dwarfs do not display this feature significantly or at all. The depth of TRAPPIST-1ʼs CO lines are similar to the field dwarfs, while the low-gravity dwarfs have shallower CO absorption lines. Thus in the K band TRAPPIST-1 best matches the field dwarfs.

Similar L bol
In the Y band, Figure 5(a), the depth of the Wing-Ford FeH band head for TRAPPIST-1 is similar to all comparative sources except J0320+1854 and J1207−3900, which are deeper than TRAPPIST-1. The shape of the spectrum carved out by the longer FeH band (∼1-1.04 μm) for TRAPPIST-1 is most similar first to J0443+0002 and second to the field dwarfs.
The Na I and K I doublets of TRAPPIST-1 appear to be deeper than all comparative objects in Figure 5(b). Interestingly, the low-gravity dwarfs J0518−2756 and J1207−3900 have relativity deep alkali lines for low-gravity sources. The shape of the FeH feature, particularly near 1.20 μm, for TRAPPIST-1 is shallower than the low-gravity dwarfs, more similar to the field dwarfs. Between the second K I and the H 2 O band we see that J0518−2756 and J1207−3900 slope slightly redward, while TRAPPIST-1 has a flat slope like the other field objects and J0443+0002. Thus in the J band we see TRAPPIST-1 showing a hybrid of features both similar to field and low-gravity dwarfs like J0443+0002. Figure 5(c) shows the H band, where the low-gravity dwarfs are more triangular in shape compared with TRAPPIST-1. Again, we see that the H band of TRAPPIST-1 clearly resembles that of a field object. The overall K band shape of TRAPPIST-1, in Figure 5(d), is similar to the field dwarfs and J0443+0002. TRAPPIST-1 shows Na I absorption and CO depths matching the field dwarfs.
When compared with sources of similar T eff or L bol , the band-by-band fits show that under the assumption of field-age, TRAPPIST-1 exhibits a blend of field and low-gravity spectral features.
8. Spectral Analysis for Sample 2: Assuming an Age of <0.5 Gyr for TRAPPIST-1 Despite the above conclusion that the overall SED of TRAPPIST-1 is well fit as a field source, the SpeX SXD, prism, and FIRE spectra all show subtle signatures deviating from normal leading to an intermediate-gravity classification. If we assume that TRAPPIST-1 is not a field-age source due to its β gravity classification, then we should treat it as we have other β gravity sources and assume an age range of <0.5 Gyr. With this age assumption, the T eff comparison sample we presented in Section 7 would no longer be suitable. Here we present the new temperature samples of field-age, young, and old age sources compared with TRAPPIST-1 assuming a younger age leading to a cooler temperature and larger radius.

Full SEDs
Figures 6(a)-(c) compare TRAPPIST-1 to field-age, lowgravity, and subdwarfs of similar temperatures over the 0.5-13 μm range. Compared with the field-age sources in Figure 6(a), TRAPPIST-1 is brighter than all sources across the entire range, having only a similar brightness from 0.63 to 0.72 μm. In Figure 6(b), TRAPPIST-1 overlaps with the comparative sample from ∼0.6 to 1 μm; however, beyond 1 μm TRAPPIST-1 is fainter. This trend is similar to what is seen in Figure 2(b), showing that no matter the assumed age, TRAPPIST-1 does not resemble the very low-gravity sources. TRAPPIST-1 overlaps with the comparative sample in Figure 6(c) from ∼0.6 to 1 μm; however, spectral features in this region are not well fit by the subdwarfs. Beyond ∼1 μm, TRAPPIST-1 is brighter than the comparative sample; however, it displays a more triangular H band similar to GJ 660.1B. Again, as seen with the low-gravity sources, no matter the assumed age range, TRAPPIST-1ʼs overall SED is poorly fit by subdwarfs.

Band-by-band Analysis
Assuming a younger age for TRAPPIST-1, we take a closer look at the comparative sample described in detail in Tables 3  and 4. At the younger age, TRAPPIST-1 is 319 K cooler, hence the sample changes from an assumed field age. All panels in Figure 7 show how TRAPPIST-1 fits to field and low-gravity equivalent sources in the Y, J, H, and K bands.
In Figure 7(a) TRAPPIST-1 is best fit by the field dwarf DENIS-P J1048.0−3956 (hereafter J1048−3956), from 0.95 to 0.99 μm, while from 0.99 to 1.06 μm TRAPPIST-1 is best fit by low-gravity dwarf 2MASS J20004841−7523070 (hereafter J2000−7523). The majority of the Y band is poorly fit by both comparative field dwarfs and the low-gravity dwarf J0608 −2753. While TRAPPIST-1 has regions that are matched well to J1048−3956 and J2000−7523, the field and low-gravity dwarfs when assuming an older age more closely match TRAPPIST-1ʼs spectrum in the Y band (see Section 7.2).
TRAPPIST-1 shares alkali depth and spectral shape features most similar to the field dwarfs in Figure 7(b), an H-band shape similar to the field dwarfs in Figure 7(c), a Na I depth similar to J0853−0356, and an overall K-band shape similar to both field dwarfs in Figure 7(d). These band-by-band similarities to the field dwarfs clearly anchor the classification of TRAPPIST-1 as a field-age source.

Sample 3: Could TRAPPIST-1ʼs Spectral Features Be
Low Metallicity Mimicking Low Gravity?
The low-metallicity d/sd GJ 660.1B discussed in Aganze et al. (2016) also showed signatures of youth similar to TRAPPIST-1 with its triangular H band. Previous work by Kirkpatrick et al. (2010) also saw this feature in the spectra of late-M dwarfs with high proper motion and with no evidence of youth. Aganze et al. (2016) measured the Allers & Liu (2013b) indices for GJ 660.1B assuming two different spectral types, M7 and M9.5. When typed as an M9.5, two of the indices received intermediate-gravity scores, while when typed as M7 (the final decided-upon spectral type), all scores were field gravity. Thus Aganze et al. (2016) (and references therein) state that additional opacity from the 1.55-1.6 μm FeH absorption band and stronger H 2 O, due to reduced condensate opacity of low-metallicity subdwarfs, may help shape the H-band continuum of mild subdwarfs and therefore potentially skew gravity index-based classifications.
Since the H-band shape of TRAPPIST-1 is similar to GJ 660.1B, we investigate whether the intermediate-gravity classification could be due to a low metallicity. However, there is no medium-resolution spectral data of GJ 660.1B or J1013−1356 and J1444−2019, two of the subdwarfs from the field and younger age assumption samples; therefore, we  instead compare the J-band Na I and K I doublets of TRAPPIST-1 to subdwarfs with medium-resolution spectra from Gonzales et al. (2018). Figure 8 displays these sources in a decreasing effective temperature sequence. Both lines in the Na I doublet of TRAPPIST-1 in Figure 8(a) appear to be of similar depth, unlike the subdwarfs, and are deeper than the subdwarf doublets. In Figure 8(b) the 1.17 μm K I doublet of TRAPPIST-1 is narrower than the subdwarfs and has a depth in between that of the binary LSR J1610−0040 and SDSS J125637.13−022452.4 (hereafter J1256−0224), while the 1.25 μm K I doublet is deeper and narrower than the subdwarfs. Thus the J-band alkali lines of TRAPPIST-1 are not similar to those of subdwarfs. In conclusion, while low metallicity may mimic some low-gravity features, we find no evidence for that in the case of TRAPPIST-1.

Final Thoughts on TRAPPIST-1ʼs Age
From comparing the overall SEDs in the various samples, we see that TRAPPIST-1 most resembles those of the field dwarfs no matter what age we assume. When comparing the NIR band-by-band fits, TRAPPIST-1 is most similar to the field sources in some areas, while it has some similarities to aspects of the low-gravity sources. From our examination with the subdwarfs in the J band, we see no evidence for the low-gravity spectral features to be caused by low metallicity. Thus we conclude that the spectral features of TRAPPIST-1 are a blend of the field dwarfs of equivalent T eff and L bol and that of low-gravity sources of similar L bol . TRAPPIST-1 is likely a field-age source with these spectral features originating from some cause other than youth.

Examining LHS 132 as an Ideal Comparative Source
When determining the age of TRAPPIST-1, Burgasser & Mamajek (2017) Filippazzo et al. 2015). We have created full SEDs of these sources to compare to TRAPPIST-1 in more detail than done in Burgasser & Mamajek (2017).
In Figure 9 we see that LHS 132 fits the overall SED shape of TRAPPIST-1 very well, while J2352−1100 and J2341 −1133 both have significantly more flux over the entire region. LHS 132 has a similar T eff and L bol as TRAPPIST-1, while J2352−1100 and J2341−1133are much hotter, brighter, and have larger radii. Looking at the SEDs alone, J2352−1100 and J2341−1133are both poor comparison sources to TRAPPIST-1 since these objects are fundamentally different, while LHS 132 is still a fair comparative source. LHS 132 also receives an intermediate-gravity classification, which is discussed further in Section 12.1. Therefore, LHS 132 is a target to explore in more Figure 8. Comparison of gravity-sensitive spectral lines of Na I and K I in the J band of TRAPPIST-1 to subdwarfs with medium-resolution data from Gonzales et al. (2018). All spectra were resampled to the same dispersion relation using a wavelength-dependent Gaussian convolution. Spectra are normalized by the average flux taken across 1.29-1.31 μm. (a) 1.14 μm Na I doublet. (b) 1.17 and 1.25 μm K I doublets.
detail to see whether other aspects match those of TRAP-PIST-1.

Comparison of the Allers & Liu (2013b) Gravity Indices
Because of the intermediate-gravity (INT-G or β) classification that Burgasser & Mamajek (2017) found for TRAPPIST-1 when using the Allers & Liu (2013b) gravity-sensitive indices, we calculated the indices for our entire comparative sample using our modified version of the ALLERS13_INDEX IDL code on both low-and medium-resolution spectra when available. The index values and final gravity scores are listed in Tables 11 and 12 in the Appendix. The equivalent-width measurements for the medium-resolution gravity score are listed in Table 13 in the Appendix. All objects receive the same gravity class using both the low-and medium-resolution indices, with the exception of vB 10, which receives a FLD-G classification with low resolution but a β classification with medium resolution due to the FeH J score. We note that while our FeH J score is a 2, the FeH J feature does not appear to look different from the other field sources in our sample. The FeH J score from Martin et al. (2017) was a 0; therefore we may be getting a spurious measurement.
The best-fit source from the Burgasser & Mamajek (2017) sample, LHS 132, also receives an INT-G gravity classification. LHS 132 and TRAPPIST-1 have similar radii, mass, log g, T eff , and L bol as well as the same scores for each of the gravity indices. Therefore, whatever physical factor is causing TRAPPIST-1 to receive a β classification may also be the same for LHS 132.
All subdwarfs receive a β gravity class, with the exception of the spectroscopic binary J1610−0400, which received a FLD-G classification, and LHS377, which did not receive a gravity class due to a lack of measurement for the FEH z index. Therefore as stated in Allers 2017), there is some aspect of the spectrum that fools the indices into classifying objects with older ages as β, which could be due to metallicity. However, as stated in Burgasser & Mamajek (2017) and from our comparison to subdwarfs, TRAPPIST-1 is not low metallicity nor does it show J-band features similar to subdwarfs, and thus this is unlikely to be the cause of the β classification for TRAPPIST-1.
Figures 10(a)-(c) show the low-resolution spectrum index scores for the comparative sample and TRAPPIST-1 with the field dwarf polynomial and the dividing line between a score of 1 or 2 from Allers & Liu (2013b). A score of 1 indicates intermediate gravity, while a score of 2 indicates low gravity for that index. Looking at the low-resolution gravity scores, TRAPPIST-1 received scores of 1, indicating intermediate gravity, in all indices using the prism and SXD data. The FIRE data, however, received a score of 0 in the H-cont index, indicating field gravity, and a score of 1 in the other indices. All FeH z and K I J index measurements (see Figures 10(a) and (b) for TRAPPIST-1), lie in the intermediate-gravity region as defined by Allers & Liu (2013b). For the H-cont index (Figure 10(c)) we see that the FIRE measurement lies in the field dwarf region, while both SpeX measurements lie in the intermediate-gravity region.
Medium-resolution gravity scores of TRAPPIST-1, reveal a difference in the K I line scores between the SXD and FIRE spectra. The K I 1.169, 1.17, and 1.253μm equivalent widths are plotted for all sources in our sample with medium resolution in Figure 11 along with the equivalent widths of all sources in the Martin et al. (2017) sample. We do not show the K I 1.224 μm equivalent-width plot, since as shown by Allers & Liu (2013b) and Martin et al. (2017) there is no visible trend. All three K I lines for the SXD spectrum of TRAPPIST-1 received a score of 0, while the FIRE spectrum K I 1.169 μm and 1.253 μm lines received scores of 1 and the K I 1.177 μm line received score of 0. The K I 1.169 μm and 1.253 μm equivalent-width measurements for the FIRE spectrum lie just below the field sources from Martin et al. (2017) and not far from the corresponding SXD measurement. As supported by our band-by-band analysis in Sections 7.2 and 8.2, TRAPPIST-1 appears to have low surface gravity features despite our conclusion that its overall SED is best fit by a field age.

Comparison to Trends with Spectral Type
To further examine the signatures of youth seen in the NIR spectrum of TRAPPIST-1, we place it in context with fundamental parameters of field sources from Filippazzo et al. (2015), low-gravity sources from Faherty et al. (2016), and subdwarfs from Gonzales et al. (2018). Figure 12 shows  the comparison of L bol versus spectral type, Figure 13 shows the comparison of T eff versus spectral type, and Figures 14 and 15 compare absolute magnitudes versus spectral type for the J, H, K, W1, and W2 bands.
As shown in Gonzales et al. (2018), all sources are mixed when comparing L bol versus optical spectral type in Figure 12. TRAPPIST-1 lands in an area where there are other field sources; however, there is no visible trend for where M7.5 β sources should be located on the diagram due to only one M7.5 β source other than TRAPPIST-1 plotted. Faherty et al. (2016) found that β gravity sources in their sample that were not members of known moving groups fell along the field sequence, and thus not all late M dwarf β sources are young. Figure 13 compares T eff versus optical spectral type with TRAPPIST-1, again landing in an ambiguous location. TRAPPIST-1 lies near the field dwarfs but is surrounded by low-gravity sources as well. As seen in Faherty et al. (2016) the low-gravity and field dwarf polynomials overlap in the M dwarf region, right where TRAPPIST-1 is located. Figures 14 and 15 compare the absolute magnitudes in the J, H, K, W1, and W2 bands of field, low-gravity, and subdwarfs. The M dwarf β gravity sources lie in the same location as the field sources or just slightly above in the J band and begin to move slightly higher than the field sequence by the K band. By W1 and W2, the β sources are further above the field sequence. TRAPPIST-1 remains within the field sequence from J through W2 and does not appear to move in brightness like the other β sources. This further supports the idea that TRAPPIST-1 is a field source requiring a different physical explanation for its low surface gravity features.  Teegarden's Star,and LHS 132. Objects in this work labeled as "not in group" or "nonmembers" are sources from Faherty et al. (2016) that could be ambiguous members, candidate members, or true nonmembers of any currently known group, thus objects that are not bona fide members.
In Figure 16 we see all γ sources, whether in known moving groups or not, clustered in parameter space that corresponds to the thin disk, while the β sources are found across the thin and well into the thick disk region. Thus the β sources with total velocities greater than 50 km s −1 may not truly be young but display signatures of youth for some reason that is unaccounted for. TRAPPIST-1, LHS 132, and Teegarden's Star lie in the thick disk region, along with two intermediate-gravity  Table 5. Since Teegarden's Star also lies in the same region as TRAPPIST-1 and has recently been found to host at least two planets (Zechmeister et al. 2019), we suggest that one idea for the low-gravity features may be the tug of planets on their host star. Consequently, LHS 132, J1022 +0200, and J1022+5825 may be ideal targets for M dwarf planet searches. Figure 17 shows the distribution of tangential velocities for low-gravity sources from Faherty et al. (2016), with their updated membership and V tan values from J. K. Faherty et al. (in preparation). In this sample, any source classified as β gravity in either the optical or NIR receives a classification of β (i.e., M8 in the optical, but M8 β in the NIR, is designated as a β in this sample). The same follows for a source with a γ classification. For sources that received a β classification in one regime but a γ in another, we choose the more extreme gravity classification. Figure 17(a) shows the distribution of tangential velocities for γ gravity sources that are members (on the right) or nonmembers (on the left) of known moving groups. We see that members and nonmembers have similar distributions of V tan , both peaking in the 20-25 km s −1 range. The nonmember distribution is not significantly different from the member distribution. Figure 17(b) shows the distribution for β gravity sources that are members and nonmembers of known moving groups. Figure 12. Optical spectral type vs. L bol for subdwarfs (blue), field objects (gray), and low-gravity objects (red and orange). Field objects come from Filippazzo et al. (2015), low-gravity objects are from Faherty et al. (2016), and subdwarfs are from Gonzales et al. (2018) with updates to sources in this paper. TRAPPIST-1 is shown as a black star. Unlike with the γ gravity sources, we see that β sources in moving groups have V tan ranging from 0 to 35 kms −1 , whereas nonmember sources have a larger range of V tan . However, the bulk of β gravity nonmembers fall in the range seen for member sources. Our calculated V tan for TRAPPIST-1 places it outside of the bulk velocity region for β gravity sources. We also calculated V tan for LHS 132, which places it in the same region as TRAPPIST-1 and the four sources plotted, which are Teegarden's Star, J1022+0200, J1022+5825, and J2322 −3133. However, J0033−1521 lies in the bulk region. Therefore, TRAPPIST-1, along with Teegarden's Star, J1022 +0200, J1022+5825, J2322−3133, and LHS 132, is kinematically distinct from the other suspected young M dwarfs.

Speculation on the β Gravity Class for TRAPPIST-1
The β classification of TRAPPIST-1 could be due to radius inflation. Two possible causes of this could be (1) magnetic activity and/or (2) tidal interactions of the planets with the star. In the case of the former, Chabrier et al. (2007) state that theoretical models for sources with M=0.08 M e show that radii can range from 0.10 to 0.14 R e for black spot coverage of up to 50%, thus offering one pathway toward a radius variation that would mimic a young M dwarf that had not contracted yet. Luger et al. (2017) and Vida et al. (2017) examined the K2 light curve and found evidence of cool, stable magnetic spots on TRAPPIST-1. However, most recently Morris et al. (2018) examined the Spitzer 3.5 and 4.5 μm light curves and found no signature of cool spots, leaving this line of evidence inconclusive for magnetic influence. The second possible cause, tidal interactions of the planets with the star, could be the cause of the classification for both TRAPPIST-1 and  Teegarden's Star. While it is beyond the scope of this work to examine the full influence of the planets on their host star, we suggest that an excellent test for this theory would be to look for planets around LHS 132, the other source that matches many of TRAPPIST-1ʼs features, as well as J1022+0200 and J1022+5825, which are kinemtically distinct from other β sources.

Conclusions
In this work we present a distance-calibrated SED of TRAPPIST-1 using a new NIR FIRE spectrum and a new parallax from the Gaia DR2 data release. With our new distance-calibrated SED we compare TRAPPIST-1 to objects of similar effective temperature and/or bolometric luminosity for young, old, and field-age sources considering an age for TRAPPIST-1 of either 0.5-10 Gyr or <0.5 Gyr. The J-band Na I and K I lines of TRAPPIST-1 were compared with those of the subdwarfs with medium-resolution data from Gonzales et al. (2018). We also looked at TRAPPIST-1 related to objects from Burgasser & Mamajek (2017). We present updated or new fundamental parameters for our comparative sample using Gaia parallaxes and Pan-STARRS photometry when available.
Using our derived fundamental parameters we find field dwarfs of similar T eff and L bol and LHS 132, an M8 dwarf classified as intermediate gravity, best fit the SED shape of TRAPPIST-1. From our band-by-band comparisons, TRAP-PIST-1 exhibits a blend of field and young spectral features.
We measure the Allers & Liu (2013b) indices for TRAPPIST-1, along with our entire comparative sample. TRAPPIST-1 receives a β gravity classification when using three different spectra, indicating it might be young. Examining spectral indices versus spectral type, TRAPPIST-1 lies in the β gravity space, while when looking at equivalent width versus spectral type, TRAPPIST-1 falls with the field sources.
In an effort to better understand the β gravity population and TRAPPIST-1, we plot L bol and T eff versus optical spectral type as well as J, H, K, W1, and W2 absolute magnitudes versus spectral type. We find TRAPPIST-1 lies in an area that has both field and β sources when examining L bol and T eff versus optical spectral type and absolute magnitudes versus optical spectral type.
We present updated UVW velocities for TRAPPIST-1 using the new Gaia astrometry and compare its kinematics to β and γ sources, which are confirmed members or not bona fide members of known moving groups, Teegarden's Star, and LHS 132. TRAPPIST-1, along with Teegarden's Star, LHS 132, J1022+0200, J1022+5825, and J2322−3133, falls within a subpopulation of β gravity sources that are not bona fide members of known moving groups and have higher UVW and tangential velocities. Lastly, we present two possible causes for the β classification of TRAPPIST-1. First, TRAPPIST-1 may have significant magnetic influence as observationally examined by observing cool stable spots. At present there is contradictory information in the literature on this topic. Luger et al. (2017) and Vida et al. (2017) found evidence for spots, while Morris et al. (2018) did not, leaving this line of explanation inconclusive. Our second proposed explanation could be related to tidal interactions with planets, and we suggest LHS 132, J1022+0200, J1022+5825, and J2322−3133 might be excellent targets for exoplanet campaigns given their similarities to both TRAPPIST-1 and Teegarden's Star.  (Nissen 2004;Bensby et al. 2014). None of the UVW velocities in this plot are with respect to LSR. Note that nonmembers in this work are objects that are lacking confirmation as bona fide moving group members. They may have ambiguous kinematics or candidate kinematics, or they might be conclusively nonmembers of any currently known group. See Faherty et al. (2016) for details.  We thank the Magellan telescope operators for their help in collecting FIRE spectra. We thank D. Bardalez Gagliuffi and C. Galindo for their help in obtaining the SpeX spectra of J0608 −2753. This research was supported by the NSF under grant No. AST-1614527 and grant No. AST-1313278, as well as by NASA under Kepler grant No. 80NSSC19K0106. E.G. thanks the LSSTC Data Science Fellowship Program, which is funded by LSSTC, NSF Cybertraining grant No. 1829740, the Brinson Foundation, and the Moore Foundation; her participation in the program has benefited this work. J.T. acknowledges support for this work provided by NASA through Hubble Fellowship grant HST-HF2-51399.001 awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS5-26555. This research has made use of the BDNYC Data Archive, an open access repository of M, L, T, and Y dwarf astrometry, photometry, and spectra. This paper includes data gathered with the 6.5 m Magellan telescopes located at Las Campanas Observatory, Chile. 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. This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www. cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC,https://www.cosmos.esa.int/ web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.

Appendix Photometry and Spectra Tables for SEDs
Photometry and spectra references for sample SEDs are given in Tables 6-13. Figure 17. Histograms of tangential velocities for low-gravity sources in moving groups and not members of groups from J. K. Faherty et al. (in preparation). Members of known moving groups are shown on the left, while nonmembers are on the right. (a) γ sources. (b) β sources. Nonmembership is defined in the same way as in Figure 16. In the right panel of (b), TRAPPIST-1 would lie in the long tail with a tangential velocity of 61.9±0.10 km s −1 . Note.Photometric points with uncertainties greater than 0.5 mag were excluded from the construction of the SED.  Note.Photometric points with uncertainties greater than 0.5 mag were excluded from the construction of the SED.  Note.Photometric points with uncertainties greater than 0.5 magnitudes were excluded from the construction of the SED.

24
The Astrophysical Journal, 886:131 (30pp), 2019 December 1 Note.Photometric points with uncertainties greater than 0.5 magnitudes were excluded from the construction of the SED. a Magnitudes were estimated from 2MASS K s to properly append the Rayleigh-Jeans tail. References.
(1) Cutri et al.    Note.When determining the gravity scores, the literature NIR spectral type was used. In cases where there is no NIR spectral type, we used the optical spectral type. Half spectral types were rounded to the nearest whole type. Gravity scores are listed for each index in order, where scores are as follows: 0-field gravity (FLD-G),  Note.Gravity scores are listed for each index in order as follows: FeH (score based on the FeH z and FeH J scores), VO z , alkali lines scores (combination of the K I 1.169, 1.17, and 1.253 equivalent-width scores), and H-cont. The appropriate combinations of scores are used to get the Allers & Liu (2013b) gravity class designations and follow the same median scores needed as in Table 11. For M8 dwarfs the VO z value has no index score and thus is labeled as "n." For medium-resolution data the FeH indices are combined to get a final FeH score. Scores correspond to gravities as follows: 0-field gravity (FLD-G), 1-intermediate gravity (INT-G), 2-low gravity (VL-G). Again in this paper we use β and INT-G interchangeably as well as γ and VL-G; however, we choose to list the final gravity class in this table following the Allers & Liu (2013b) notation.
References. Spectrum references are the same as those listed in Table 11. Note.Equivalent-width measurements of the Na I 1.138 μm line and the K I 1.169 μm, 1.177 μ, and 1.244 μm lines. Table 10 of Allers & Liu (2013b) shows the equivalent-width measurements cutoff translate to a score of a 0 or 1 for the alkali lines.