Near-Infrared Spectroscopy of the Y0 WISEP J173835.52+273258.9 and the Y1 WISE J035000.32-565830.2: the Importance of Non-Equilibrium Chemistry
S. K. Leggett, P. Tremblin, D. Saumon, M. S. Marley, Caroline V. Morley, D. S. Amundsen, I. Baraffe, G. Chabrier
aa r X i v : . [ a s t r o - ph . S R ] M a r Near-Infrared Spectroscopy of the Y0 WISEPJ173835.52 + − S. K. Leggett [email protected] P. Tremblin , D. Saumon M. S. Marley Caroline V. Morley D. S. Amundsen I. Baraffe , G. Chabrier , ABSTRACT
We present new near-infrared spectra, obtained at Gemini Observa-tory, for two Y dwarfs: WISE J035000.32 − R = 540 spectrum was ob-tained for W0350, covering 1 . < λ µ m < .
7, and a cross-dispersed GNIRS R = 2800 spectrum was obtained for W1738, covering 0.993 – 1.087 µ m, 1.191 Gemini Observatory, Northern Operations Center, 670 N. A’ohoku Place, Hilo, HI 96720, USA Maison de la Simulation, CEA-CNRS-INRIA-UPS-UVSQ, USR 3441, Centre d’´etude de Saclay, F-91191Gif-Sur-Yvette, France Astrophysics Group, University of Exeter, EX4 4QL Exeter, UK Los Alamos National Laboratory, PO Box 1663, MS F663, Los Alamos, NM 87545, USA NASA Ames Research Center, Mail Stop 245-3, Moffett Field, CA 94035, USA Department of Astronomy and Astrophysics, University of California, Santa Cruz, CA 95064, USA Ecole Normale Sup´erieure de Lyon, CRAL, UMR CNRS 5574, F-69364 Lyon Cedex 07, France µ m, 1.589 – 1.631 µ m, and 1.985 – 2.175 µ m, in four orders. We alsopresent revised Y J H photometry for W1738, using new NIRI Y and J imaging,and a re-analysis of the previously published NIRI H band images. We comparethese data, together with previously published data for late-T and Y dwarfs, tocloud-free models of solar metallicity, calculated both in chemical equilibriumand with disequilibrium driven by vertical transport. We find that for the Ydwarfs the non-equilibrium models reproduce the near-infrared data better thanthe equilibrium models. The remaining discrepancies suggest that fine-tuningthe CH /CO and NH /N balance is needed. Improved trigonometric parallaxeswould improve the analysis. Despite the uncertainties and discrepancies, themodels reproduce the observed near-infrared spectra well. We find that for theY0, W1738, T eff = 425 ±
25K and log g = 4 . ± .
25, and for the Y1, W0350, T eff = 350 ±
25K and log g = 4 . ± .
25. W1738 may be metal-rich. Basedon evolutionary models, these temperatures and gravities correspond to a massrange for both Y dwarfs of 3 – 9 Jupiter masses, with W0350 being a cooler,slightly older, version of W1738; the age of W0350 is 0.3 – 3 Gyr, and the age ofW1738 is 0.15 – 1 Gyr.
Subject headings: molecular processes, stars: brown dwarfs, stars: atmospheres,stars: individual (WISE J035000.32 −
1. Introduction
The atmospheres of giant gaseous planets and brown dwarfs (objects with a mass belowthat required for stable nuclear fusion, mass .
80 Jupiter masses) are molecule-rich andchemically complex. The deep atmosphere is fully convective; there can be detached con-vection zones above the radiative-convective boundary, if the pressure and composition aresuch that there is strong absorption at the wavelength typical of the flux being emitted atthe temperature of that particular layer. The reader is referred to the review by Marley &Robinson (2015) for further discussion. The brown dwarf atmospheres are turbulent, andchemical species are mixed vertically through the atmosphere. Mixing occurs in the con-vective zones, but may also occur in the nominally quiescent radiative zone by processessuch as gravity waves (e.g. Freytag et al. 2010). If mixing occurs faster than local chemicalreactions can return the species to local equilibrium, then abundances can be very differentfrom those expected for a gas in equilibrium. Species whose abundances are significantlyimpacted by mixing in brown dwarf atmospheres are CH , CO, CO , N and NH (e.g. Noll,Geballe & Marley 1997, Saumon et al. 2000, Golimowski et al. 2004, Leggett et al. 2007, 3 –Visscher & Moses 2011, Zahnle & Marley 2014).Also, various species condense, forming cloud decks in the photosphere. For L dwarfswith effective temperature 1300 . T eff K . . T eff K . T eff decreases further, the next species to condense are calculatedto be H O for T eff ≈
350 K and NH for T eff ≈
200 K (Burrows et al. 2003, Morley et al.2014).It had been hoped that the chemistry of the atmospheres of very cold brown dwarfs,those with 400 . T eff K . T eff ≤
500 K, and these havebeen classified as Y dwarfs (e.g. Kirkpatrick et al. 2012, Leggett et al. 2013). It turns outthat modeling their atmospheres is not simple (e.g. Leggett et al. 2015, hereafter L15).All but one of the known Y dwarfs have been found using the
Wide-field Infrared SurveyExplorer ( WISE ; Wright et al. 2010) by : Cushing et al. (2011, 2014); Kirkpatrick etal. (2012); Liu et al. (2012); Luhman (2014); Pinfield et al. (2014); Schneider et al.(2015); Tinney et al. (2012). The remaining object was discovered by Luhman, Burgasser& Bochanski (2011) as a companion to a white dwarf, using images from the Infrared ArrayCamera onboard the
Spitzer Space Telescope (Werner et al. 2004). L15 studied the propertiesof seventeen Y dwarfs using near-infrared photometry and spectroscopy, together with
WISE and
Spitzer mid-infrared photometry. The observations were compared to spectra and colorsgenerated from model atmospheres with a variety of cloud cover — cloud-free models fromSaumon et al. (2012, hereafter S12), models with homogeneous layers of chloride and sulphideclouds from Morley et al. (2012), and patchy cloud models from Morley et al. (2014). Themodels include updated opacities for NH and pressure-induced H . It was found that themodels qualitatively reproduced the trends seen in the observed colors, and that the cloudlayers are thin to non-existent for these brown dwarfs with T eff ≈
400 K. However the modelfluxes were a factor of two low at the Y , H , K , [3.6], and W3(12 µ m) bands. The models usedin L15 all assumed equilibrium chemistry, and it was suggested that much of the discrepancycould be resolved by significantly reducing the NH abundance, perhaps by vertical mixing.In this work we compare observed Y dwarf photometry and spectroscopy to models byTremblin et al. (2015, hereafter T15) which include non-equilibrium chemistry, as well asan updated line list for CH absorption. We also present new near-infrared spectroscopy for 4 –two Y dwarfs, and revised near-infrared photometry for one of these.
2. Observations2.1. WISE J035000.32 − The discovery of WISE J035000.32 − . < λ µ m < . . ′′
72 slit. The resulting resolvingpower is R = 540.On 2014 November 7, fifty two 300 s frames were obtained over an approximately fivehour period in photometric conditions and 0 . ′′ ± ′′ along the slit. On 2014 December 4 and 2015 January 3 sixteen andfourteen frames, respectively, were obtained in similar conditions with the same instrumentconfiguration. The data from 2015 January 3 were taken at a significantly higher airmass(1.5 – 1.9, compared to 1.2 – 1.3 on the earlier two nights), and were not combined with theother datasets due to the lower signal to noise ratio (S/N). The 68 frames from Novemberand December were reduced in a standard way using calibration lamps on the telescopefor flat fielding and wavelength calibration. The 68 300 s images were combined using the gemcombine IRAF routine, giving 5.7 hours on this source.Bright F3 and F7 dwarf stars were observed on each night, to remove telluric absorptionfeatures and flux calibrate the spectra. The stars used on 2014 November 7 were HD 13517and HD 30526, and on 2014 December 4 HD 36636. Template spectra for these spectraltypes were obtained from the spectral library of Rayner et al. (2009), and used to deter-mine a one-dimensional sensitivity function. A one-dimensional spectrum for the target wasextracted from the combined image using the trace of the standard stars for reference. Thesensitivity spectrum was then used to correct the shape of the target spectrum, and a finalflux calibration was done on the target spectrum using the observed J band photometry forthe source ( Y band coverage was incomplete, and H band was noisy). Figure 1 shows ourspectrum smoothed with a 3 pixel boxcar. The uncertainty in the spectrum was determinedby the sky noise, and is shown in Figure 1; S/N across the J band peak is ∼
10, while acrossthe H band it is ∼
4. 5 –Schneider et al. (2015) used the
Hubble Space Telescope (HST)
Wide Field Camera 3to obtain near-infrared slitless grism spectroscopy of W0350. The G141 grism was used,covering 1.10 – 1.70 µ m at R ∼ HST spectrum, where the latter has been multiplied by a factor of 1.07 to bring it intoagreement with our J band photometry. The agreement is good across the J band butnot as good across the H band, where the flux peaks differ by ∼ H band photometry and comparing it to the measured value suggests that our spectrum istoo bright, while the Schneider spectrum is too faint. The discrepancy is likely due to thefaintness of the source and the resulting low S/N, although variability cannot be excluded.The compilation of photometric variability of brown dwarfs by Crossfield (2014) shows thatT dwarfs can be variable at the ∼
10% level. We compare our spectrum to models in § + The discovery of WISEP J173835.52+273258.9 (W1738) was published by Cushing etal. (2011), who classified it as a Y0 dwarf. Table 1 gives photometry and astrometry for thistarget, with source references.The
Y J H photometry for W1738 presented in Table 1 differs from that published byL15. As part of the Gemini North program GN-2013A-Q-21, W1738 was observed for severalhours with the Gemini near infrared imager (NIRI; Hodapp et al. 2003) in Y and J in asearch for variability. The variability result will be published elsewhere; here we presentimproved values of Y and J . The J band result is significantly different from that previouslypublished. We re-examined the H band photometry obtained on the same night as thepreviously published data, which was a night of poor seeing. Re-reducing the earlier H banddata set paying closer attention to instances of very poor seeing, and therefore detections oflow significance, produces the revised H value given in Table 1. The new J and H valuesare now in better agreement (within 2 σ ) with the synthetic values determined by Schneideret al. (2015) from slitless HST spectra.We obtained spectroscopy for W1738 on Gemini North, via program GN-2014A-Q-64,using the Gemini near-infrared spectrograph (GNIRS; Elias et al. 2006). GNIRS was used incross-dispersed mode with the 111 l/mm grating, the short camera and the 0 . ′′
675 slit, giving R = 2800. A central wavelength of 1.56 µ m resulted in wavelength coverage for orders 3 to6 of 1.985 – 2.175 µ m, 1.589 – 1.631 µ m, 1.191 – 1.305 µ m and 0.993 – 1.087 µ m. Thissetting nicely sampled the flux peaks (Figure 2).A total of 75 300 s frames were obtained over five nights: 2014 March 17 and 22, 2014 6 –May 18, 2014 July 11 and 13. Data were typically taken through thin clouds with 0 . ′′ ′′ along the slit. The Gemini IRAFroutines are not designed for this higher resolution cross-dispersed mode and so reductionwas carried out manually. AB image pairs were subtracted from each other to form an imagewith a positive and negative spectrum. Pattern noise artefacts were then removed using a python script designed for the purpose. Each image resulting from a subtracted pair wasvisually inspected, and some removed because of guider issues or low signal when at highairmass. To form the final coadded image, a total of 36 images with positive and negativespectral traces were combined, for a total of six hours on target. The IRAF apall routinewas used to extract spectra for each order, using the locations of the standard star spectralorders as a reference. One-dimensional multi-order spectra were extracted from the flatfield and arc images obtained from the calibration lamp on the telescope using the locationsof the science apertures as references. The one-dimensional multi-order spectra for scienceand standards were divided by the corresponding one-dimensional flat field spectrum andthe positive and negative spectra for each order were then averaged, after multiplying thenegative spectra by −
1. The uncertainty in the spectrum was estimated from the differencebetween the positive and negative spectra. A wavelength solution was determined graphicallyfrom the arc spectrum and applied.The standard stars observed to remove telluric absorption features and flux calibratethe spectra were HD 149803, an F7V star, and HD 173494, an F6V star. Template spectrafor these spectral types were obtained from the spectral library of Rayner et al. (2009), andused to determine a one-dimensional sensitivity function. The sensitivity spectrum was thenused to correct the shape of the target spectrum, in each order.Cushing et al. (2011) present an
HST
Wide Field Camera 3 G141 grism spectrum ofW1738. We flux calibrated this spectrum using our revised J and H photometry, and scaledour higher resolution spectrum to match. We could not calibrate our spectrum directlybecause the spectral orders do not completely span the filter bandpasses. Figure 2 showsboth our Gemini spectrum and the Cushing et al. HST spectrum. Our spectrum has beensmoothed with a 9 pixel boxcar. The uncertainty in the spectrum is also shown; S/N acrossthe
Y J H band peaks is ∼
10, while across the K band it is ∼
5. We compare our spectrumto models in §
3. Comparison of Models and Data3.1. Models
In this work we use S12 and T15 cloud-free models only. Although chloride and sul-phide clouds are important for T dwarfs with T eff as low as 600 K (Morley et al. 2012),our focus is the 400 K Y dwarfs and it has been shown that cloud-free atmospheres repro-duce observations better than existing cloudy models for such objects (L15). We also useT15 models without the ad hoc modifications to the pressure-temperature gradient theyconsidered, although such modifications can improve the agreement with near-infrared datafor mid-type T dwarfs. The modifications were motivated by the possibility of atmosphericfingering convection induced by species condensation, which is ignored here.The known Y dwarfs are of necessity solar neighborhood objects, due to their instrinsicfaintness. The majority of M dwarf stars in the local neighborhood have near-solar metallicityand age (e.g. Burgasser et al. 2015, Terrien et al. 2015) and the same is likely to be true ofthe Y dwarfs. We restrict the models to surface gravities given by log g = 4.0, 4.5 and 4.8because evolutionary models show that these values correspond to an age range of 0.4 to 10Gyr at T eff ≈
400 K (Saumon & Marley 2008). The corresponding mass range is around 5to 20 Jupiter masses, based on the evolutionary models.Our analysis also uses T15 models whose nitrogen and carbon chemistry is driven outof equilibrium by vertical mixing which is parametrized with an eddy diffusion coefficient K zz cm s − . The departure of the nitrogen chemistry from equilibrium abundances is quiteinsensitive at these temperatures to the value of K zz , however the carbon chemistry remainssensitive to mixing (Zahnle & Marley 2014). The T15 models show that, at T eff = 400 K,models with K zz = 10 cm s − produce a ∼
30% stronger CO absorption at λ ≈ . µ mthan those with K zz = 10 cm s − , as CO is enhanced at the expense of CH . We usedthe well-studied T eff = 600 K T dwarf ULAS J003402.77 − Spitzer mid-infrared spectrum and, combined with other data and an accurate trigonometricparallax, its properties are well determined (Leggett et al. 2009, Smart et al. 2010). Wefind that the T15 models with appropriate T eff and log g (550 – 600 K, 4.5 dex) give a[4.5] magnitude that is too faint by 0.4 magnitudes if K zz = 10 cm s − , but is within 0.1magnitudes of the observed value if K zz = 10 cm s − . Mixing may vary from object toobject, and depend on T eff , log g or metallicity. Previous studies have determined K zz valuesof 10 – 10 cm s − for L and T dwarfs (Geballe et al. 2009; Leggett et al. 2007, 2008, 2010;Stephens et al. 2009) and for Jupiter K zz ≈ cm s − (Lewis & Fegley 1983). Hence avalue of K zz = 10 cm s − is reasonable for Y dwarfs and we adopt that value here. 8 –The T15 models used here contain some improvements over those described in theT15 publication. The thermochemical data for H were updated to be compatible with theJANAF database (Chase 1998) and the Saumon, Chabrier & van Horn (1995) equation ofstate. For the disequilibrium models, the pressure-temperature profile was re-converged morefrequently to ensure that the total flux is consistent with the effective temperature. Finally,the post-processed low-resolution spectra were computed with the correlated– k method andthe high-resolution spectra were computed using line-by-line opacities at a resolution of atleast 1 cm − (a higher resolution than used by T15, see also Amundsen et al. 2014).Apart from inclusion of non-equilibrium chemistry, the S12 and T15 models differ dueto the inclusion by T15 of an updated CH line list (Yurchenko & Tennyson 2014), and theomission by T15 of PH . Also, although condensation is included in the T15 models, theremoval of the condensed species from the local gas is not. At these temperatures rain-outspecies are not important opacity sources, however their inclusion in the gas may change theatmospheric opacities. For example not removing the condensed Fe allows it to react withH S to form FeS, removing the H S absorption (Morley et al. 2014, Marley & Robinson2015). The two model sets also use different solar abundances (Lodders 2003 for S12, Caffauet al. 2011 for T15), and different treatments of line broadening.Figure 3 compares T eff = 400 K and log g = 4.0 solar metallicity models from S12 andT15. The top panel compares S12 and T15 equilibrium chemistry models, and the lowerpanel compares T15 equilibrium and non-equilibrium models. Filter bandpasses are shownfor reference. The reader is referred to L15 (their Figure 5), and Figures shown later in thispaper, for identification of the species causing the pronounced absorption bands in thesespectra at this temperature. The dominant opacity sources are H , H O, CH , and NH ;CO and PH may be important at λ ∼ µ m.Comparison of the S12 and T15 spectra for equilibrium chemistry and T eff = 400 K(Figure 3 top panel), shows that the S12 spectrum is brighter at 0 . < λ µ m < . . < λ µ m < .
58 and 8 . < λ µ m < . . < λ µ m < .
62, by factorsup to 1.5 — 2.0; at other wavelengths the spectra are very similar. We suspect that thedifference at λ ∼ . µ m is due to the omission of PH by T15 (see L15, Figure 5, middlepanel). The differences at other wavelengths may be due to the use of the new CH line listby T15. At these low temperatures the near-infrared spectrum is very sensitive to opacitychanges, as demonstrated by the changes seen when the new H and NH opacities wereincorporated into the S12 models (see S12 Figure 7, bottom panel).Comparison of T15 spectra with and without non-equilibrium chemistry for T eff = 400K (Figure 3 bottom panel) shows that the H band is much brighter in the non-equilibriumcase, as are the 3 µ m and 10 µ m regions. This is because the abundances of CH and NH . Because CO is enhanced, the 4 . µ m region is fainterin the non-equilibrium case. The blue wings of the Y and K bands are brighter because ofthe reduction in NH . The cause of the decrease in flux near the peak of the K band inthe non-equilibrium case is not clear – there are no opacities in this region that should beenhanced by mixing. It may be that the changes introduced into the atmosphere, by mixing,redistributes the flux to regions previously impacted by CH or NH absorption. Althoughnot easily seen in Figure 3, there is a similar decrease in flux near the Y band peak in thenon-equilibrium model. We discuss this further in Section 3.3.We compare the models to data in the following sections. Figures 4 and 5 present near-infrared color-magnitude and color-color diagrams for late-T and Y dwarfs, where observational data are compared to cloud-free solar metallicity S12and T15 model sequences. Data sources are this work and L15, as well as earlier publicationsreferenced in L15. Brown dwarfs with M J >
19 are labelled in Figure 4, as well as knownbinary systems where one component has M J >
19; the Appendix gives full names anddiscovery references for these objects. Filter bandpasses are shown in Figure 3 and in Figure6 below.Figure 4 shows M J as a function of J − H (left) and J − [4.5] (right). S12 and T15equilibrium sequences are shown for log g = 4.5, and T15 non-equilibrium models are shownfor log g = 4.0, 4.5 and 4.8, as identified in the legends. Comparing the T15 equilibriumand non-equilibrium models in each panel shows that the brightening of the H band andthe reduction in the [4.5] flux in the non-equilibrium models (see Figure 3) improves theagreement with the observations, particularly for T eff >
450 K.Figure 5 shows various color-color plots where observations are compared to T15 log g = 4.0, 4.5 and 4.8 non-equilibrium model sequences, and the T15 log g = 4.5 equilibriumsequence. As previously mentioned, the agreement at J − H is much improved with theinclusion of non-equilibrium chemistry, and the agreement at [3.6] − [4.5] is also improvedfor the T dwarfs with T eff >
500 K. The Y − J and J − K colors do not support onetreatment of the chemistry over the other, at least for the current models. There are largediscrepancies in the J − K and [3.6] − [4.5] colors which get worse for redder J − [4.5], orlower temperatures. Both K and [3.6] are too faint in the T15 models (also in the S12 models,see L15). Very little flux is emitted at 2 < λ µ m < and CH are importantopacity sources (e.g. L15), and it is possible that adjusting the carbon and nitrogen mixingcan address the problem. The discrepancies are discussed further below, where we comparethe models to near-infrared spectra.Figure 5 indicates that the modeled Y − J and J − K colors are sensitive to gravity for J − [4.5] >
5, or T eff <
450 K (Figure 4). To explore this further, Figure 6 shows three T15synthetic near-infrared spectra for T eff = 400 K and K zz = 10 cm s − . The models differeither in gravity or metallicity, as indicated by the legend. We calculated a very small numberof non-solar-metallicity models, motivated by the observed dispersion in the observationaldata in Figures 4 and 5, and the knowledge that metallicity does impact the spectral energydistribution of brown dwarfs (see e.g. Burningham et al. 2013). A more complete study ofthe effect of metallicity will be done elsewhere.Figure 6 shows that a decrease in gravity or an increase in metallicity increases theflux emerging at K . This is a well-known effect which is due to the relative importanceof pressure-induced H opacity which increases with gravity and decreases with metallicity(e.g. Borysow et al. 1997, S12). Based on these models, a change in metallicity of 0.2 dexhas a much larger impact on the K magnitude than a change in gravity of 0.5 dex. Figure6 also shows that the flux emerging from the blue half of the Y band is sensitive to gravityand metallicity. Here the changes go in the opposite sense, so that an increase in metallicityor decrease in gravity reduces the emerging flux. Also, the gravity change of 0.5 dex has alarger impact than the metallicity change of 0.2 dex on the Y magnitude, according to thesemodels. The primary opacity source at λ ≈ µ m is H O, which, based on the small changesseen in the wings of the J flux peak, does not appear to be sensitive to these parameters.We will explore nature of the Y band sensitivity to gravity in future work. In this section we compare near-infrared spectra of three Y dwarfs to T15 syntheticspectra. The model to observation comparison is done by eye, and the K band is neglectedbecause of the inadequacies in the models in this region (Figure 5). The Y dwarfs are theY0–0.5 dwarf WISEPC J121756.91+162640.2B (W1217B; Kirkpatrick et al. 2012, Liu etal. 2012), the Y0 dwarf W1738 and the Y1 dwarf W0350. We use the T eff values indicatedby the J − [4.5] color (Figure 4) as the starting point for spectral fits, as this mimics thespectral shape from 1 µ m to 5 µ m. Most of the energy is emitted in the 5 µ m window 11 –for the late-T and Y dwarfs (e.g. Morley et al. 2012, 2014), and therefore it is important toinclude this region in any temperature estimate. Figure 4 shows that we expect T eff valuesof 400 – 450 K for W1217B and W1738, and 325 – 375 K for W0350. The model spectraused in the comparison are either in chemical equilibrium or have log K zz = 6; have surfacegravities given by log g = 4.0, 4.5 and 4.8; and have T eff (K) = 325, 350, 375, 400, 425 and450. For each Y dwarf, then, we have model spectra with three values of T eff , and for eachof these we have three values of log g . We compared the nine model spectra to the observednear-infrared spectrum, where we scale the flux of the model spectra (generated for the Ydwarf surface) by the square of the Y dwarf radius and the inverse-square of the distance tothe Y dwarf. The Y dwarf radius for any T eff and g combination is obtained from Saumon& Marley (2008) evolutionary models. The distances to W1217B and W1738 are taken frompublished values of trigonometric parallax, although we allowed adjustments in the overallbrightness of the model spectra corresponding to the 1 σ quoted uncertainties in the parallaxmeasurements. For W0350 only a preliminary parallax measurement is available, and we findthat a very large adjustment is needed, outside of the quoted uncertainties; we discuss thisin § ∼
10% level is possible(Crossfield 2014). We neglect any change in near-infrared spectral shape due to variability,assuming that that any wavelength-dependence is not significant for the cloud-free Y dwarfs.
Figure 7 shows the spectrum of the Y0–0.5 dwarf W1217B presented by Leggett et al.(2014). Principal absorbers are indicated; for more detailed identifications of features innear-infrared spectra of cool brown dwarfs the reader is referred to Bochanski et al. (2011)and Canty et al. (2015). Leggett et al. studied the properties of the W1217AB system usingcoevality and luminosity arguments for the binary, as well as comparisons of the photometryand spectroscopy of the components to Morley et al. (2012) equilibrium models. The authorsdetermined T eff = 450 K, log g = 4.8, a possibly sub-solar metallicity, and thin to no clouds.T15 fit the same observed spectrum using a model with T eff = 425 K, log g = 4.0 and log K zz = 8. Once the evolutionary radius is used the T15-selected model spectrum would betoo bright, however that could be compensated by adjusting the distance to the brown dwarf,which is quite uncertain.We find that the current suite of T15 models shows that the best fit is given by the T eff = 450 K, log g = 4.5 and log K zz = 6 model spectrum, although this requires a distanceof 11.3 pc, compared to the measured value of 10 . +1 . − . pc (Dupuy & Kraus 2013). The 12 –second-best fit is the T eff = 450 K, log g = 4.8 and log K zz = 6 model (Leggett et al. 2014also find T eff = 450 K and log g = 4.8). This model spectrum would require no adjustmentof the distance, however the fit is poorer at Y and K , and similar at J and H , as shown inFigure 7.Figure 7 also shows the best fitting equilibrium chemistry model, and demonstratesthe impact of mixing in the near-infrared more clearly than Figure 3. Including mixinggreatly improves the fit in the region of the strong NH absorption at λ ≈ . µ m and λ ≈ . µ m. However the equilibrium model better reproduces the data at 1 . . λ µ m . .
09 and 2 . . λ µ m . . T eff = 450 K, log g = 4.5 and log K zz = 6 spectrum gives a superiorfit to the relative heights of the Y , J and H flux peaks compared to the previously publishedfits by T15 and Leggett et al. (2014). Figure 8 shows our best fit to the new R = 2800 spectrum presented here for the Y0dwarf W1738. The best fit in this case is provided by the T eff = 425 K, log g = 4.0 and log K zz = 6 model spectrum, with no adjustment needed to the measured distance of 7 . ± . K , we compared the observations to a small numberof super-solar metallicity models, and found that a model with [m/H] = +0 . T eff = 400 K,log g = 4.0 and log K zz = 6 gives almost as good a fit, however the shape of the Y flux peakis poorer, as shown in Figure 8. This slightly cooler, higher metallicity model matches thedata better if the distance is reduced to the low end of the range measured by Beichmannet al..Figure 8 also compares the observations to the best fitting equilibrium chemistry modelwith T eff = 450 K and log g = 4.5. In this case the match is improved if the distance inincreased towards the high end of the measured range. The expected equilibrium-chemistryproblem of overly strong NH absorption λ ≈ . µ m and λ ≈ . µ m is seen, and alsothe relative Y J H flux peaks are not as well-reproduced by this model. 13 –
Figure 9 shows our best fit to the new R = 540 spectrum presented here for the Y1 dwarfW0350. The apparent drop in observed flux at 1 . . λ µ m . .
07 should be confirmed bynew observations (flux calibration of spectra at the extremes of the wavelength range can beprone to error due to the rapidly changing instrument sensitivity function). If real, it maybe an indicator of water clouds (e.g. Morley et al. 2014, their Figure 10), or it may providean additional constraint on gravity or metallicity (Figure 6).For this brown dwarf, the preliminary trigonometric parallax published by Marsh etal. (2013) implies an unrealistically faint absolute magnitude of M [4 . = 16 .
9, which wouldmake W0350 similar in luminosity to the extreme dwarf W0855 while J − [4.5] is more than3.5 magnitudes bluer (see Figure 4). In matching the spectra, we start with the modelset constrained in temperature and gravity as described in § § . +0 . − . pc. An error in the preliminary Marsh et al. results, especially in theharder to measure smaller-parallax greater-distance direction, would not be surprising. Forcomparison, Beichman et al. (2014) revise the distance for another Y dwarf in their sample(WISE J041022.71+150248.4) from 4 . +1 . − . to 6 . ± . T eff = 350 K, log g = 4.0 and log K zz = 6 model spectrum.At a higher gravity of log g = 4.5 the fit is almost as good, but the H flux peak appears tohave an excess of flux at λ > . µ m, as shown in Figure 9. The fits imply a distance of6.3 pc for the preferred model, and 5.3 pc for the second model. In the absolute magnitudediagrams in Figure 4 we use our distance of 6.3 pc.Figure 9 also shows the best fitting equilibrium chemistry model with T eff = 350 K andlog g = 4.5. This fit implies a distance of 5.3 pc, as found for the non-equilibrium modelwith the same values of T eff and g . There is again evidence that the equilibrium model is toofaint in the blue wing of the H -band, suggesting that even at the low temperature of T eff =350 K mixing of N and NH is significant. Although the non-equilibrium chemistry model fits to the spectra of the three Y dwarfsare remarkably good in some regions, especially the J band, systematic offsets can be seenin Figures 7, 8 and 9. In the Y band, the model flux at 1 . . λ µ m . .
04 is low, as isthe flux at 1 . . λ µ m . .
09. In the H band, the model flux at 1 . . λ µ m . . . . λ µ m . .
60 is high. In particular, there is a 14 –strong absorption feature observed at λ ≈ . µ m that is not seen in the models (Figures7, 8, 9). In the K band the model flux at 2 . . λ µ m . .
18 is low by a factor of 2 – 3.The photometric comparison presented above indicates that the model flux is also too lowat 3 < λ µ m < and CH absorptionis important. The strong absorption at λ ≈ . µ m looks similar to the feature seen inthe spectrum of the 500 K brown dwarf UGPS J072227.51054031.2, and identified by Cantyet al. (2015) as a combination of both NH and CH absorption. Figures 3, 7 and 8 showthat there is a decrease in flux at the Y and K flux peaks when mixing is included in themodels, the cause of which may be simple flux redistribution ( § J − [4.5] colors for these twoobjects are about 0.4 magnitudes larger than observed (Figure 4). Considering the uncer-tainties in the fits and the remaining deficiences in the models, the agreement is reasonable.Table 1 gives the values of T eff and log g for W0350 and W1738 based on the model fitsto the new spectra presented here, with the corresponding mass and age from evolutionarymodels. Note that the lower gravity we find for W1217B compared to Leggett et al. (2014)implies a younger age for the system of 1 – 5 Gyr and a lower mass for W1217B of 10 – 15M Jupiter , compared to the 4 – 8 Gyr and 20 – 24 M
Jupiter found by those authors.
4. Estimated Physical Parameters for W0350 and W1738
Evolutionary models show that mass is tightly constrained by the surface gravity forsolar-neighborhood objects at these temperatures (see e.g. Saumon & Marley 2008, theirFigure 4, and Allard et al. 1996, Marley et al. 1996, Burrows et al. 1997). Changinglog g by 0.5 dex changes the [4.5] flux by a factor ≈ .
25, which is larger than or similarto the likely uncertainty in the model. However our experiments using the
Y J H peakstogether in a spectral by-eye comparison shows that gravity can be further constrained to0.25 dex for these Y dwarfs (e.g. Figure 7). The fits, together with evolutionary models,then imply a mass of 5 +4 − Jupiter masses for both W0350 and W1738. Once the models aremore robust, the Y and K colors show promise for constraining log g (Figures 5 and 6),although metallicity variations will need to also be considered. 15 –Given the range in colors seen for different gravities and different input physics (S12vs. T15) in Figure 4, the uncertainty in our derived temperature is <
50 K. Current modelsshow that a change in 25 K at these temperatures changes the flux in the J band by a factor & . T eff of ∼ ±
25 K. This is supported by the range in parameters for thepreferred and runner up models described in the previous section. Note that exploration ofa limited set of non-solar metallicity models in the previous section showed that a change in[m/H] of 0.2 dex changed the derived T eff by 25 K, consistent with the adopted uncertainty(Figure 9).Gravity and temperature together constrain age (see e.g. Saumon & Marley 2008,their Figure 4). Taking into account uncertainties of 25 K and 0.25 dex in T eff and log g respectively, W0350 has an age of 0.3 – 3 Gyr and W1738 has an age of 0.15 – 1 Gyr.Interestingly, W0350 seems to be a cooler, older version of W1738, as both have a massaround 5 Jupiter masses. Both W0350 and W1738 have low tangential velocities of around20 km s − , based on the proper motions of Marsh et al. (2013) and Beichman et al. (2014),and the distance used here for W0350 and that determined by Beichman et al. for W1738.This is consistent with thin disk kinematics (e.g. Dupuy & Liu 2012) and an age <
5. Conclusion
As expected, models which include vertical mixing and the resulting non-equilibriumchemistry reproduce observations of Y dwarfs better than those which do not include mixing.Hydrogen, carbon, nitrogen and oxygen chemistry shapes the spectral energy distributions ofthese cold brown dwarfs. Vertical mixing in the atmosphere impacts the chemistry of carbonand nitrogen, and the remaining systematic discrepancies between the data and models thatwe have identified, where both NH and CH are important, could be addressed by fine-tuning the mixing. The kinetics of nitrogen species are uncertain, and Moses (2014) showsthat different treatments of quenching do result in significantly different NH abundances;this will be an avenue of future exploration. Mixing and non-equilibrium chemistry is alsoexpected to increase the abundance of PH and HCN in cool atmospheres (Zahnle & Marley2014, Sousa-Silva et al. 2015), which has not been taken into account in current models, andwhich should be included in future models.The model comparisons show that for T eff ≈
400 K Y dwarfs, the atmospheric chemistryneeds to change such that, while conserving flux: 16 – • the CH absorption is decreased at 2 < λ µ m < λ = 4 . µ m • the CH and NH absorption at λ ≈ . µ m is increased without increasing theabsorption elsewhere in the near-infrared • the NH absorption at λ ≈ . µ m is decreased • the flux at λ ≈ . µ m and λ ≈ . µ m is increased without strengthening the CH or NH absorption, except at λ ≈ . µ mIt may be easiest to achieve this by applying a retrieval analysis such as done recently forlate-T dwarfs by Line et al. (2015). On the other hand, less than 10% of the total flux isemitted at λ < . µ m in these objects, all in the Wein tail of the Planck function. This makesthe near-infrared spectrum very sensitive to details of the opacity and chemical abundances.A more robust analysis would be possible with spectroscopic data beyond 3 µ m.New or improved trigonometric parallaxes for the Y dwarfs would be valuable. Lessuncertain distances would allow tighter constraints on gravity and temperature when fittingspectra as we did here, incorporating the brown dwarf radius and distance and requiringthat the absolute flux levels be consistent.Despite the remaining discrepancies and uncertainties, the quality of the models andthe observational data are quite impressive, and the fits that we show here to J bandspectra in particular, are remarkably good. We find that the isolated Y dwarfs WISEJ035000.32 − +4 − Jupitermasses, and their ages are 0.3 – 3 Gyr and 0.15 – 1 Gyr respectively. This is consistent withthe low tangential velocity of around 20 km s − measured for both brown dwarfs and putsthem well within the commonly accepted mass range of planets.Based on observations obtained at the Gemini Observatory, which is operated by theAssociation of Universities for Research in Astronomy, Inc., under a cooperative agreementwith the NSF on behalf of the Gemini partnership: the National Science Foundation (UnitedStates), the Science and Technology Facilities Council (United Kingdom), the National Re-search Council (Canada), CONICYT (Chile), the Australian Research Council (Australia),Minist´erio da Ciˆencia, Tecnologia e Inova¸c˜ao (Brazil) and Ministerio de Ciencia, Tecnolog´ıae Innovaci´on Productiva (Argentina). S. L.’s research is supported by Gemini Observatory.D.S.’ work was supported in part by NASA grant NNH12AT89I from Astrophysics Theory.I. B.’s work is supported by the European Research Council through grant ERC-AdG No. 17 –320478-TOFU.This publication makes use of data products from the Wide-field Infrared Sur-vey Explorer, which is a joint project of the University of California, Los Angeles, and theJet Propulsion Laboratory/California Institute of Technology, funded by the National Aero-nautics and Space Administration. This research has made use of the NASA/ IPAC InfraredScience Archive, which is operated by the Jet Propulsion Laboratory, California Institute ofTechnology, under contract with the National Aeronautics and Space Administration. A. Brown Dwarf Identifications
Table 2 lists the full name of the T and Y dwarfs identified in Figure 4, together withdiscovery references.
REFERENCES
Ackerman, A. S. & Marley, M. S. 2001, ApJ, 556, 872Allard, F., Hauschildt, P. H., Baraffe, I. & Chabrier, G. 1996, ApJ, 465, L123Amundsen, D. S., Baraffe, I., Tremblin, P., Manners, J., Hayek, W., Mayne, N. J. & Acreman,D. M. 2014, A&A, 564, 59Beichman, C. et al. 2014, ApJ, 783, 68Bochanski J. J., Burgasser A. J., Simcoe R. A. & West A. A. 2011, AJ, 142, 169Borysow, A., Jorgensen, U. G. & Zheng, C. 1997, A&A, 324, 185Brook, C. B. et al. 2012, MNRAS, 426, 690Burgasser, A. J., Logsdon, S. E., Gagne, J., Bochanski, J. J., Faherty, J. K., West, A. A.,Mamjek, E. E., Schmidt, S. J. & Cruz, K. L. 2015, ApJS, 220, 18Burningham, B. et al. 2013, MNRAS, 433, 457Burrows, A., Marley, M. S., Hubbard, W. B., Lunine, J. I., Guillot, T., Saumon, D., Freed-man, R., Sudarsky, D. & Sharp, C. 1997, ApJ, 491, 856Burrows, A., Sudarsky, D. & Lunine, J. I. 2003, ApJ, 596, 587Caffau E., Ludwig H. G., Steffen M., Freytag B. & Bonifacio P. 2011,
SoPh , 268, 255 18 –Canty, J. I. et al. 2015, MNRAS, 450, 454Chase, M. W. 1998, NIST-JANAF Thermochemical Tables, 4th edition, J. Phys. Chem.Ref. Data Monograph No. 9Crossfield, I. J. M. 2014, A&A, 566, A130Cushing, M. C. et al. 2011, ApJ, 743, 50Cushing, M. C., Kirkpatrick, J. D., Gelino, C. R., Mace, G. N., Skrutskie, M. F. & Gould,A. 2014, AJ, 147, 113Delorme, P. et al. 2010, A&A, 518, 39Dupuy, T. J. & Kraus, A. L. 2013,
Science , 341, 1492Dupuy, T. J. & Liu, M. C. 2012, ApJS, 201, 19upuy, T. J., Liu, M. C. & Leggett, S. K. 2015, ApJ, 803, 102Eikenberry, S. et al. 2004, SPIE 5492, 1196Elias, J. H, Joyce, R. R., Liang, M., Muller, G. P., Hileman, E. A., & George, J. R., inGround-based and Airborne Instrumentation for Astronomy, eds. I. S. McLean andI. Masanori 2006, SPIE, 6269, 138Faherty, J. K.; Beletsky, Y., Burgasser, A. J., Tinney, C., Osip, D. J., Filippazzo, J. C. &Simcoe, R. A. 2014, ApJ, 790, 90Freytag, B., Allard, F., Ludwig, H.-G., Homeier, D. & Steffen, M. 2010, A&A, 513, 19Geballe, T. R., Saumon, D., Golimowski, D. A., Leggett, S. K., Marley, M. S. & Noll, K. S.2009, ApJ, 695, 844Golimowski, D. A. et al. 2004, AJ, 127, 3516Hargreaves, R. J., Beale, C. A., Michaux, L., Irfan, M. & Bernath, P. F. 2012, ApJ, 757, 46Helling, Ch., Oevermann, M., Luttke, M. J. H., Klein, R. & Sedlmayr, E. 2001, A&A, 376,194Hodapp, K. W. et al. 2003, PASP, 115, 1388Kirkpatrick, J. D. et al. 2012, ApJ, 753, 156 19 –Kirkpatrick, J. D. et al. 2013, ApJ, 776, 128Knapp, G.R. et al. 2004, AJ, 127, 3553Leggett, S. K. et al. 2007, ApJ, 655, 1079Leggett, S. K. et al. 2008, ApJ, 682, 1256Leggett, S. K. et al. 2009, ApJ, 695, 1517Leggett, S. K., Saumon, D., Burningham, B., Cushing, M. C., Marley, M. S. & Pinfield, D.J. 2010, ApJ, 720, 252Leggett, S. K. et al. 2012, ApJ, 748, 74Leggett, S. K. et al. 2013, ApJ, 763, 130Leggett, S. K. et al. 2014, ApJ, 780, 62Leggett, S. K., Morley, C. V., Marley, M. S. & Saumon, D. 2015, ApJ, 799, 37 (L15)Lewis, J. S. & Fegley, M. B. 1983, Space Science Reviews, 39, 163Line, M. R., Teske, J., Burningham, B., Fortney, J. & Marley, M. S. 2015, ApJ, submittedLiu, M. C. et al. 2011, ApJ, 740, 108Liu, M. C., Dupuy, T. J., Bowler, B. P., Leggett, S. K. & Best, W. M. J. 2012, ApJ, 758,57Lodders, K. 2003, ApJ, 591, 1220Luhman, K. L. 2014, ApJ, 786, L18Luhman, K. L., Burgasser, A. J. & Bochanski, J. J. 2011, ApJ, 730, L9Marley, M. S. & Robinson, T. D. 2015, ARA&A, arXiv:1410.6512Marley, M. S., Saumon, D., Guillot, T., Freedman, R. S., Hubbard, W. B., Burrows, A. &Lunine, J. I. 1996,
Science , 272, 1919Marley, M. S., Saumon, D., Cushing, M., Ackerman, A. S., Fortney, J. J. & Freedman, R.2012, ApJ, 754, 135Marsh K. A. et al. 2013, ApJ, 762, 119 20 –Morley, C. V., Fortney, J. J., Marley, M. S., Visscher, C., Saumon, D. & Leggett, S. K. 2012,ApJ, 756, 172Morley, C. V., Marley, M. S.,Fortney, J. J., Lupu, R., Saumon, D., Greene, T. & Lodders,K. 2014, ApJ, 787, 78Moses, J. I. 2014, Philosophical Transactions of the Royal Society A: Mathematical, Physicaland Engineering Sciences, 372, 20130073Noll, K. S., Geballe, T. R. & Marley, M. S. 1997, ApJ, 489, L87Pinfield, D. J. et al. 2014, MNRAS, 444, 1931Radigan, J., Jayawardhana, R., Lafreniere, D., Artigau, E., Marley, M. S. & Saumon, D.2012, ApJ, 750, 105Rayner, J. T., Cushing, M. C. & Vacca, W. D. 2009, ApJS, 185, 289Saumon, D., Geballe, T. R., Leggett, S. K., Marley, M. S., Freedman, R. S., Lodders, K.,Fegley, B., Jr. & Sengupta, S. K. 2000, ApJ, 541, 374Saumon, D., Marley, M. S., Cushing, M. C., Leggett, S. K., Roellig, T. L., Lodders, K. &Freedman, R. S. 2006, ApJ, 647, 552Saumon D. et al. 2012, ApJ, 750, 74 (S12)Saumon, D., Chabrier, G. & van Horn, H. M. 1995, ApJS, 99, 713Saumon, D. & Marley, M. S. 2008, ApJ, 689, 1327Schneider A. C. et al. 2015, ApJ, 804, 92Smart, R. L. et al. 2010, A&A, 511, 30Sousa-Silva, C., Al-Refaie, A. F., Tennyson, J. & Yurchenko, S. N. 2015, MNRAS, 446, 2337Stephens D. S. et al. 2009, ApJ, 702, 154Terrien, R. C., Mahadevan, S., Deshpande, R. & Bender, C. F. 2015, ApJS, 220, 16Tinney C. G., Faherty, J. K., Kirkpatrick, J. D., Wright, E. L., Gelino, C. R.; Cushing, M.C.; Griffith, R. L. & Salter, G. 2012, ApJ, 759, 60Tremblin, P., Amundsen, D. S., Mourier, P., Baraffe, I., Chabrier, G., Drummond, B.,Homeier, D. & Venot, O. 2015, ApJ, 804, L17 (T15) 21 –Tsuji, T., Ohnaka, K., Aoki, W. & Nakajima, T. 1996, A&A, 308Visscher, C. & Moses, J. I. 2011, ApJ, 738, 72Warrn, S. J. et al. 2007, MNRAS, 381, 1400Werner, M. W. et al. 2004, ApJS, 154, 1Wright, E. L. et al. 2010, AJ, 140, 1868Yurchenko S. N. & Tennyson J. 2014, MNRAS, 440, 1649Zahnle, K. J. & Marley, M. S. 2014, ApJ, 797, 41
This preprint was prepared with the AAS L A TEX macros v5.2.
22 –Fig. 1.— The black line is the spectrum of WISE J035000.32 − HST . The gray line is the uncertainty in the Geminiobservational data. 23 –Fig. 2.— The black line is the spectrum of WISEP J173835.52+273258.9 obtained usingGemini and presented in this work, and the green line is a lower resolution spectrum ob-tained by Cushing et al. (2011) using
HST . The gray line is the uncertainty in the Geminiobservational data. 24 –Fig. 3.— Comparison of Saumon et al. (2012) and Tremblin et al. (2015) cloud-free modelspectra for T eff =400 K, log g =4.0, solar metallicity atmospheres. The spectra have beensmoothed to R ∼ W ISE filter passbands (upper panel) and IRAC
Spitzer passbands (lower panel).The upper panel compares equilibrium chemistry models and the lower panel comparesequilibrium and non-equilibrium models. Model parameters are given in the legends. Seetext for discussion. 25 –Fig. 4.— M J as a function of J − H and J − [4.5]. S12 and T15 equilibrium sequences areshown for log g = 4.5, and T15 non-equilibrium models are shown for log g = 4.0, 4.5 and4.8, as indicated in the legends. Color-coded T eff values for the models are shown along theaxes. Data points are colors and magnitudes for late-T and Y dwarfs (triangles and circles re-spectively), using the MKO near-infrared system and the IRAC Vega-based system. Sourcesof photometry and parallax are this work and as referenced in L15. The Y dwarf W0350is plotted using the photometric parallax determined here (see § M J agreeswith the models by definition. The close T/Y binary systems CFBD1458AB, W1217AB andW0146AB have resolved near-infrared photometry but not mid-infrared. Error bars alongthe x axes are omitted for the T dwarfs, for clarity. Full names for the labelled sources aregiven in the Appendix. 26 –Fig. 5.— Color-color plots comparing observations to T15 log g = 4.0 (orange line), log g = 4.5 (red line) and log g = 4.8 (dark red line) non-equilibrium model sequences. A T15equilibrium model sequence with log g = 4.5 (cyan line) is also shown. Symbols are as inFigure 4. Not shown, for clarity, is the extreme dwarf W0855; this dwarf has no published Y , H or K values, and J − [4.5] = 11 . ± . − [4.5] = 3 . ± .
05. 27 –Fig. 6.— Synthetic T15 near-infrared spectra for T eff = 400 K and K zz = 10 cm s − . Themodels differ either in gravity or metallicity, as indicated by the legend. The spectra havebeen scaled to match at the J band peak. MKO near-infrared filter profiles are shown. 28 –Fig. 7.— Comparison of the observed WISEPC J121756.91+162640.2B spectrum (black,Leggett et al. 2014, with 3-pixel boxcar smoothing) to T15 solar metallicity synthetic spectrasmoothed to match the plotted data resolution: T eff = 450K log g = 4 . K zz = 6 (darkred); T eff = 450K log g = 4 . K zz = 6 (red); and equilibrium chemistry T eff = 450Klog g = 4 . . +1 . − . pc (Dupuy & Kraus 2013); for the equilibriumand non-equilibrium T eff = 450K log g = 4 . T eff = 400K log g = 4 . K zz = 6 [m/H] = +0 . T eff = 425K log g = 4 . K zz = 6 [m/H] = +0 . T eff = 450K log g = 4 . . ± . T eff = 400K log g = 4 . K zz = 6 [m/H] = +0 . T eff = 450K,log g = 4 . . − T eff = 350Klog g = 4 . K zz = 6 (red); T eff = 350K log g = 4 . K zz = 6 (orange); and equi-librium chemistry T eff = 350K log g = 4 . T eff = 350K log g = 4 . T eff = 350K log g = 4 . − M − m (err) 2.32(0.37) a M13 0.54(0.17) B14 Y MKO (err) 21.62(0.12) L15 19.79(0.07) this work J MKO (err) 22.09(0.12) L15 19.63(0.05) this work H MKO (err) 22.51(0.20) L15 20.24(0.08) this work K MKO (err) · · · · · · µ m) IRAC µ m) IRAC µ m) W ISE · · · · · · µ m) W ISE µ m) W ISE T eff K 350 ±
25 this work 425 ±
25 this worklog g cm s − ± ± a This preliminary parallax results in an unrealistically faint absolute magnitude;the spectral type and model fits shown here suggest that M − m = 1 . ± . Short Name Full Name Discovery ReferenceW0146AB WISE J014656.66+423410.0 Kirkpatrick et al. 2012; Dupuy, Liu & Leggett 2015W0350 WISE J035000.32 − − − − − − − − − − −−