A Uniform Retrieval Analysis of Ultracool Dwarfs. III. Properties of Y-Dwarfs

TL;DR

This study conducts a uniform atmospheric retrieval analysis on 14 Y and T-dwarfs, revealing insights into their temperature structures, molecular abundances, and alkali metal depletion, with implications for understanding substellar atmospheres and future JWST observations.

Contribution

It provides the first comprehensive retrieval analysis of Y-dwarfs, confirming alkali metal depletion trends and linking them to atmospheric chemistry and color transitions.

Findings

Temperature structures are consistent with radiative-convective equilibrium.

Water and methane abundances match chemical equilibrium predictions.

Alkali metals sodium and potassium are increasingly depleted with decreasing temperature.

Abstract

Ultra-cool brown dwarfs offer a unique window into understanding substellar atmospheric physics and chemistry. Their strong molecular absorption bands at infrared wavelengths, Jupiter-like radii, cool temperatures, and lack of complicating stellar irradiation, make them ideal test-beds for understanding Jovian-like atmospheres. Here we report the findings of a uniform atmospheric retrieval analysis on a set of 14 Y and T-dwarfs observed with the Hubble Space Telescope Wide Field Camera 3 instrument. From our retrieval analysis, we find the temperature-structures to be largely consistent with radiative-convective equilibrium in most objects. We also determine the abundances of water, methane, and ammonia and upper limits on the alkali metals sodium and potassium. The constraints on water and methane are consistent with predictions from chemical equilibrium models, while those of ammonia may be affected by vertical disequilibrium mixing, consistent with previous works. Our key result stems from the constraints on the alkali metal abundances where we find their continued depletion with decreasing effective temperature, consistent with the trend identified in a previous retrieval analysis on a sample of slightly warmer late T-dwarfs in Line et al. (2017). These constraints show that the previously observed Y-J color trend across the T/Y transition is most likely due to the depletion of these metals in accordance with predictions from equilibrium condensate rainout chemistry. Finally, we simulate future James Webb Space Telescope observations of ultra-cool dwarfs and find that the NIRSpec PRISM offers the best chance at developing high-precision constraints on fundamental atmospheric characteristics.

Tables

Table 1. Table 1: 31 Free-Parameters in Our Retrieval Model

Parameter	Description
log(fi)	7 log(constant-with-altitude
	Volume Mixing Ratios)
log(g)	log(GM/R2) [cm s-2]
(R/D)2	radius-to-distance scale [RJup/pc]
T(P)	temperature at 15 pressure levels [K]
$\Delta \lambda$	wavelength calibration uncertainty [nm]
$b$	errorbar inflation exponent
	(Part I, Equation 3)
$\gamma ,\beta$	TP-profile smoothing hyperparameters
	(Part I, Table 2/Equation 5)
${\kappa }_{P_0},P_0,\alpha$	Cloud profile parameters
	(Part II, Equation 1)

Table 2. Table 2: Basic photometric properties of our sample.

WISE/AllWISE Name	Spec. Type	YMKO111Synthetic Photometry from Schneider et al. (2015). [mag]	JMKO1 [mag]	HMKO1 [mag]	Dist. [pc]
WISEA J032504.52-504403.0	T8	19.980$\pm$0.027	18.935$\pm$0.024	19.423$\pm$0.027	27.2$\pm$2.2222Distances from Kirkpatrick et al. (2019).
WISEA J040443.50-642030.0	T9	20.328$\pm$0.032	19.647$\pm$0.025	19.970$\pm$0.033	21.9$\pm$1.42
WISEA J221216.27-693121.6	T9	20.282$\pm$0.023	19.737$\pm$0.024	20.225$\pm$0.036	12.2$\pm$0.42
WISEA J033515.07+431044.7	T9	20.166$\pm$0.029	19.467$\pm$0.023	19.938$\pm$0.031	13.9$\pm$0.52
WISEA J094306.00+360723.3	T9.5	…	19.766$\pm$0.025	20.315$\pm$0.038	10.7$\pm$0.32
WISEA J154214.00+223005.2	T9.5	20.461$\pm$0.028	19.937$\pm$0.026	20.520$\pm$0.045	11.6$\pm$0.62
WISEA J041022.75+150247.9	Y0	…	19.325$\pm$0.024	19.897$\pm$0.038	6.52$\pm$0.17333Distances from Martin et al. (2018).
WISEA J073444.03-715743.8	Y0	20.870$\pm$0.041	20.354$\pm$0.029	21.069$\pm$0.071	15.1$\pm$1.23
WISEA J173835.52+273258.8	Y0	…	19.546$\pm$0.023	20.246$\pm$0.031	7.34$\pm$0.223
WISEA J205628.88+145953.6	Y0	…	19.129$\pm$0.022	19.643$\pm$0.026	7.23$\pm$0.203
WISEA J222055.34-362817.5	Y0	20.899$\pm$0.034	20.447$\pm$0.025	20.858$\pm$0.035	11.9$\pm$0.753
WISEA J163940.84-684739.4	Y0 Pec.	20.833$\pm$0.023	20.626$\pm$0.023	20.764$\pm$0.029	4.39$\pm$0.103
WISEA J140518.32+553421.3	Y0.5	21.33$\pm$0.057	21.061$\pm$0.035	21.501$\pm$0.073	6.76$\pm$0.493
WISE J154151.65-225024.9	Y1	20.461$\pm$0.028	19.934$\pm$0.026	20.520$\pm$0.045	5.98$\pm$0.143

Table 3. Table 3: Retrieved & Derived Evolutionary Parameters.

WISE/ALLWISE Name	Spec. Type	Teff [K]	log(g) [cgs]	$R$ [RJup]	Mass [MJup]	Priors
WISEA J032504.52-504403.0	T8	664${}_{-26}{}^{}{}_{}{}^{+34}$	4.97${}_{-0.30}{}^{}{}_{}{}^{+0.19}$	1.08${}_{-0.11}{}^{}{}_{}{}^{+0.11}$	44${}_{-23}{}^{}{}_{}{}^{+24}$	Free
		660${}_{-24}{}^{}{}_{}{}^{+29}$	5.06${}_{-0.36}{}^{}{}_{}{}^{+0.27}$	1.10${}_{-0.10}{}^{}{}_{}{}^{+0.11}$	56${}_{-32}{}^{}{}_{}{}^{+46}$	Constrained
WISEA J040443.50-642030.0	T9	646${}_{-32}{}^{}{}_{}{}^{+38}$	5.27${}_{-0.30}{}^{}{}_{}{}^{+0.18}$	0.78${}_{-0.06}{}^{}{}_{}{}^{+0.06}$	47${}_{-24}{}^{}{}_{}{}^{+22}$	Free
		639${}_{-30}{}^{}{}_{}{}^{+37}$	5.20${}_{-0.43}{}^{}{}_{}{}^{+0.22}$	0.81${}_{-0.06}{}^{}{}_{}{}^{+0.06}$	42${}_{-25}{}^{}{}_{}{}^{+24}$	Constrained
WISEA J221216.27-693121.6	T9	555${}_{-25}{}^{}{}_{}{}^{+27}$	5.88${}_{-0.35}{}^{}{}_{}{}^{+0.08}$	0.47${}_{-0.03}{}^{}{}_{}{}^{+0.05}$	69${}_{-34}{}^{}{}_{}{}^{+8}$	Free
		540${}_{-32}{}^{}{}_{}{}^{+40}$	5.25${}_{-0.29}{}^{}{}_{}{}^{+0.16}$	0.71${}_{-0.02}{}^{}{}_{}{}^{+0.02}$	36${}_{-17}{}^{}{}_{}{}^{+16}$	Constrained
WISEA J033515.07+431044.7	T9	484${}_{-24}{}^{}{}_{}{}^{+25}$	4.87${}_{-0.22}{}^{}{}_{}{}^{+0.22}$	0.87${}_{-0.06}{}^{}{}_{}{}^{+0.06}$	23${}_{-10}{}^{}{}_{}{}^{+15}$	Free
		483${}_{-25}{}^{}{}_{}{}^{+24}$	4.87${}_{-0.21}{}^{}{}_{}{}^{+0.23}$	0.88${}_{-0.06}{}^{}{}_{}{}^{+0.06}$	23${}_{-9}{}^{}{}_{}{}^{+14}$	Constrained
WISEA J094306.00+360723.3	T9.5	494${}_{-31}{}^{}{}_{}{}^{+33}$	4.89${}_{-0.31}{}^{}{}_{}{}^{+0.30}$	0.70${}_{-0.07}{}^{}{}_{}{}^{+0.07}$	15${}_{-8}{}^{}{}_{}{}^{+15}$	Free
		494${}_{-36}{}^{}{}_{}{}^{+36}$	4.86${}_{-0.30}{}^{}{}_{}{}^{+0.30}$	0.75${}_{-0.06}{}^{}{}_{}{}^{+0.07}$	16${}_{-8}{}^{}{}_{}{}^{+17}$	Constrained
WISEA J154214.00+223005.2	T9.5	488${}_{-30}{}^{}{}_{}{}^{+60}$	5.16${}_{-0.18}{}^{}{}_{}{}^{+0.15}$	0.61${}_{-0.05}{}^{}{}_{}{}^{+0.05}$	21${}_{-8}{}^{}{}_{}{}^{+11}$	Free
		484${}_{-26}{}^{}{}_{}{}^{+39}$	5.07${}_{-0.26}{}^{}{}_{}{}^{+0.16}$	0.71${}_{-0.04}{}^{}{}_{}{}^{+0.05}$	23${}_{-10}{}^{}{}_{}{}^{+12}$	Constrained
WISEA J041022.75+150247.9	Y0	530${}_{-54}{}^{}{}_{}{}^{+90}$	5.30${}_{-0.32}{}^{}{}_{}{}^{+0.23}$	0.73${}_{-0.08}{}^{}{}_{}{}^{+0.10}$	43${}_{-21}{}^{}{}_{}{}^{+24}$	Free
		529${}_{-86}{}^{}{}_{}{}^{+83}$	5.06${}_{-0.59}{}^{}{}_{}{}^{+0.29}$	0.75${}_{-0.04}{}^{}{}_{}{}^{+0.07}$	27${}_{-13}{}^{}{}_{}{}^{+24}$	Constrained
WISEA J073444.03-715743.8	Y0	467${}_{-37}{}^{}{}_{}{}^{+51}$	5.39${}_{-0.28}{}^{}{}_{}{}^{+0.17}$	0.71${}_{-0.07}{}^{}{}_{}{}^{+0.07}$	50${}_{-23}{}^{}{}_{}{}^{+21}$	Free
		456${}_{-34}{}^{}{}_{}{}^{+50}$	5.24${}_{-0.33}{}^{}{}_{}{}^{+0.18}$	0.77${}_{-0.05}{}^{}{}_{}{}^{+0.77}$	42${}_{-22}{}^{}{}_{}{}^{+22}$	Constrained
WISEA J173835.52+273258.8	Y0	371${}_{-29}{}^{}{}_{}{}^{+27}$	5.43${}_{-0.17}{}^{}{}_{}{}^{+0.13}$	0.71${}_{-0.05}{}^{}{}_{}{}^{+0.05}$	59${}_{-22}{}^{}{}_{}{}^{+15}$	Free
		371${}_{-30}{}^{}{}_{}{}^{+33}$	5.20${}_{-0.29}{}^{}{}_{}{}^{+0.2}$	0.73${}_{-0.03}{}^{}{}_{}{}^{+0.04}$	34${}_{-17}{}^{}{}_{}{}^{+20}$	Constrained
WISEA J205628.88+145953.6	Y0	493${}_{-42}{}^{}{}_{}{}^{+40}$	4.95${}_{-0.35}{}^{}{}_{}{}^{+0.31}$	0.67${}_{-0.05}{}^{}{}_{}{}^{+0.07}$	16${}_{-8}{}^{}{}_{}{}^{+15}$	Free
		485${}_{-38}{}^{}{}_{}{}^{+38}$	4.93${}_{-0.29}{}^{}{}_{}{}^{+0.25}$	0.72${}_{-0.02}{}^{}{}_{}{}^{+0.03}$	18${}_{-9}{}^{}{}_{}{}^{+14}$	Constrained
WISEA J222055.34-362817.5	Y0	444${}_{-33}{}^{}{}_{}{}^{+74}$	5.09${}_{-0.23}{}^{}{}_{}{}^{+0.24}$	0.72${}_{-0.07}{}^{}{}_{}{}^{+0.08}$	26${}_{-9}{}^{}{}_{}{}^{+21}$	Free
		449${}_{-35}{}^{}{}_{}{}^{+57}$	5.07${}_{-0.26}{}^{}{}_{}{}^{+0.16}$	0.74${}_{-0.07}{}^{}{}_{}{}^{+0.07}$	26${}_{-12}{}^{}{}_{}{}^{+13}$	Constrained
WISEA J163940.84-684739.4	Y0 Pec.	654${}_{-38}{}^{}{}_{}{}^{+16}$	4.35${}_{-0.08}{}^{}{}_{}{}^{+0.09}$	0.40${}_{-0.02}{}^{}{}_{}{}^{+0.03}$	1.5${}_{-0.3}{}^{}{}_{}{}^{+0.3}$	Free
		-	-	-	-	Constrained555Retrieval model could not converge upon a physically realistic TP profile
WISEA J140518.32+553421.3	Y0.5	327${}_{-44}{}^{}{}_{}{}^{+83}$	4.39${}_{-0.31}{}^{}{}_{}{}^{+0.28}$	0.66${}_{-0.09}{}^{}{}_{}{}^{+0.12}$	4.4${}_{-2.1}{}^{}{}_{}{}^{+3.9}$	Free
		338${}_{-58}{}^{}{}_{}{}^{+98}$	3.89${}_{-0.15}{}^{}{}_{}{}^{+0.22}$	0.75${}_{-0.06}{}^{}{}_{}{}^{+0.08}$	1.7${}_{-0.5}{}^{}{}_{}{}^{+1.4}$	Constrained
WISE J154151.65-225024.9	Y1	323${}_{-42}{}^{}{}_{}{}^{+81}$	5.06${}_{-0.48}{}^{}{}_{}{}^{+0.50}$	0.33${}_{-0.05}{}^{}{}_{}{}^{+0.07}$	5.4${}_{-3.4}{}^{}{}_{}{}^{+9.7}$	Free
		389${}_{-87}{}^{}{}_{}{}^{+85}$	3.91${}_{-0.19}{}^{}{}_{}{}^{+0.33}$	0.72${}_{-0.02}{}^{}{}_{}{}^{+0.04}$	1.7${}_{-0.6}{}^{}{}_{}{}^{+2.0}$	Constrained

Table 4. Table 4: Retrieved Atmospheric Abundances

WISE/ALLWISE Name	Spec.	H2O666All abundances are reported as the log of the volume mixing ratio log(VMR) where the remainder of the gas is taken to be H2-He at a fixed solar ratio.	CH46	CO6777All of these measurements represent 3$\sigma$ upper limits (see text).	CO267	C/O888These are not relative to solar. Solar [C/O] is 0.55 in this table.	H2S67	NH36	Na+K6
	Type
WISEA J032504.52-504403.0	T8	-3.31${}_{-0.13}{}^{}{}_{}{}^{+0.12}$	-3.05${}_{-0.16}{}^{}{}_{}{}^{+0.11}$	-4.1	-3.7	0.99$\pm 0.18$	-5.0	-4.49${}_{-0.18}{}^{}{}_{}{}^{+0.12}$	-5.52${}_{-0.07}{}^{}{}_{}{}^{+0.09}$
WISEA J040443.50-642030.0	T9	-3.01${}_{-0.13}{}^{}{}_{}{}^{+0.11}$	-2.74${}_{-0.16}{}^{}{}_{}{}^{+0.10}$	-3.0	-3.3	1.02$\pm$0.16	-5.0	-4.63${}_{-0.20}{}^{}{}_{}{}^{+0.13}$	-6.07
WISEA J221216.27-693121.6	T9	-2.59${}_{-0.18}{}^{}{}_{}{}^{+0.07}$	-2.56${}_{-0.16}{}^{}{}_{}{}^{+0.05}$	-2.9	-3.3	0.79$\pm$0.09	-6.8	-4.05${}_{-0.13}{}^{}{}_{}{}^{+0.08}$	-5.07
WISEA J033515.07+431044.7	T9	-3.35${}_{-0.09}{}^{}{}_{}{}^{+0.09}$	-3.48${}_{-0.11}{}^{}{}_{}{}^{+0.11}$	-3.8	-3.9	0.57$\pm$0.07	-5.3	-4.78${}_{-0.12}{}^{}{}_{}{}^{+0.13}$	-5.97${}_{-0.10}{}^{}{}_{}{}^{+0.07}$
WISEA J094306.00+360723.3	T9.5	-3.35${}_{-0.15}{}^{}{}_{}{}^{+0.14}$	-3.13${}_{-0.15}{}^{}{}_{}{}^{+0.17}$	-3.3	-3.2	1.22$\pm$0.25	-5.0	-4.46${}_{-0.16}{}^{}{}_{}{}^{+0.16}$	-5.27
WISEA J154214.00+223005.2	T9.5	-3.04${}_{-0.08}{}^{}{}_{}{}^{+0.09}$	-2.92${}_{-0.10}{}^{}{}_{}{}^{+0.08}$	-4.2	-4.3	0.95$\pm$0.15	-6.0	-4.32${}_{-0.12}{}^{}{}_{}{}^{+0.10}$	-6.77
WISEA J041022.75+150247.9	Y0	-2.90${}_{-0.15}{}^{}{}_{}{}^{+0.13}$	-2.63${}_{-0.19}{}^{}{}_{}{}^{+0.17}$	-3.3	-4.1	1.09$\pm$0.30	-4.3	-4.11${}_{-0.19}{}^{}{}_{}{}^{+0.15}$	-5.07
WISEA J073444.03-715743.8	Y0	-2.91${}_{-0.15}{}^{}{}_{}{}^{+0.12}$	-2.77${}_{-0.14}{}^{}{}_{}{}^{+0.09}$	-3.4	-3.7	0.78$\pm$0.16	-6.0	-4.29${}_{-0.14}{}^{}{}_{}{}^{+0.10}$	-6.07
WISEA J173835.52+273258.8	Y0	-2.87${}_{-0.08}{}^{}{}_{}{}^{+0.08}$	-2.75${}_{-0.10}{}^{}{}_{}{}^{+0.12}$	-3.3	-4.1	0.79$\pm$0.23	-5.0	-4.21${}_{-0.09}{}^{}{}_{}{}^{+0.10}$	-5.27
WISEA J205628.88+145953.6	Y0	-3.18${}_{-0.15}{}^{}{}_{}{}^{+0.16}$	-2.89${}_{-0.17}{}^{}{}_{}{}^{+0.18}$	-4.2	-4.4	1.10$\pm$0.27	-5.0	-4.44${}_{-0.17}{}^{}{}_{}{}^{+0.17}$	-5.57
WISEA J222055.34-362817.5	Y0	-3.04${}_{-0.10}{}^{}{}_{}{}^{+0.11}$	-3.00${}_{-0.12}{}^{}{}_{}{}^{+0.11}$	-4.2	-4.3	0.62$\pm$0.10	-5.8	-4.19${}_{-0.10}{}^{}{}_{}{}^{+0.08}$	-6.87
WISEA J163940.84-684739.4	Y0Pec.	-3.32${}_{-0.04}{}^{}{}_{}{}^{+0.04}$	-3.42${}_{-0.04}{}^{}{}_{}{}^{+0.05}$	-4.3	-4.6	0.46$\pm$0.06	-6.3	-4.72${}_{-0.04}{}^{}{}_{}{}^{+0.05}$	-7.07
WISEA J140518.32+553421.3	Y0.5	-3.24${}_{-0.13}{}^{}{}_{}{}^{+0.15}$	-3.33${}_{-0.16}{}^{}{}_{}{}^{+0.14}$	-3.6	-3.8	0.46$\pm$0.10	-5.0	-4.84${}_{-0.16}{}^{}{}_{}{}^{+0.14}$	-6.07
WISE J154151.65-225024.9	Y1	-2.68${}_{-0.24}{}^{}{}_{}{}^{+0.26}$	-2.80${}_{-0.26}{}^{}{}_{}{}^{+0.26}$	-3.5	-3.6	0.45$\pm$0.17	-5.0	-4.43${}_{-0.23}{}^{}{}_{}{}^{+0.21}$	-6.47

Table 5. Table 5: ScCHIMERA Grid Model Ranges and Step Sizes

Parameter	Range	Step Size
Teff [K]	300$-$950	50
log(g) [cgs]	3.0$-$5.5	0.5
$[$M/H$]$	-1$-$1	0.5
C/O	0.1$-$0.7	0.2
	0.7$-$0.9	0.05
log(Kzz)	2$-$8	2

Full text

††journal: ApJ††software: python2.7, matplotlib, CHIMERA (Line et al., 2013), ScCHIMERA (Piskorz et al., 2018; Bonnefoy et al., 2018), emcee (Foreman-Mackey et al., 2013), corner.py

1 Introduction

Brown dwarfs have solicited intriguing questions since their discovery several decades ago (Becklin & Zuckerman, 1988; Rebolo et al., 1995; Oppenheimer et al., 1995). While not being massive enough to to fuse hydrogen into helium (Hayashi & Nakano, 1963; Kumar, 1963), they were still too massive to be considered as “traditional” planets following the roughly 13MJup definition based on the fusion of deuterium (Shu, 1977; Saumon et al., 1996). More recently there have been arguments that formation pathways, rather than mass limits, are more useful when defining the difference between brown dwarfs and planets (Boss, 2001; Bate et al., 2003). This has placed the study of brown dwarfs at an interesting crossroads between planetary science and stellar astrophysics. Efforts to understand the physics of brown dwarfs have thus pulled methodologies from both fields in order to measure the physical characteristics and understand the evolution of these objects (for a review, Marley & Robinson, 2015).

Motivation for studying the atmospheres of brown dwarfs is two-fold. First, brown dwarfs do not have a stable internal energy source, and thus their evolution is highly dependent upon their initial formation mass (e.g. Baraffe et al., 2003) and specific physical/chemical structure of their atmosphere (e.g. Saumon & Marley, 2008). Secondly, brown dwarfs offer the chance to study planetary-like atmospheric conditions, while not having to include the complication of an irradiating host star. Understanding the physical and chemical mechanisms at work in cooler brown dwarf atmospheres thus provides constraints on both their evolution, and the characteristics of planetary-like atmospheres.

The bulk properties of field brown dwarfs (mass, radius, Teff, etc.) have been well studied over the past several decades (for a review, Burrows et al., 2001). With cool effective temperatures ($200K\lesssim$ Teff $\lesssim 3000K$) over photospheric pressures ($300\lesssim$ P $\lesssim 10^{-4}$ bar), their thermal emission predominately radiates in the near-to-mid infrared, with their spectra being sculpted by strong molecular and atomic opacities of species such as: hydrogen and helium (H2/He), water (H2O), methane (CH4), ammonia (NH3), and alkali metals such as potassium (K) and sodium (Na) for the coolest objects to carbon monoxide and dioxide (CO,CO2), H2O, H2/He, and metal hydrides and oxides for the hottest (Fegley & Lodders, 1996; Lodders & Fegley, 2002; Lodders, 2003). The precise molecular and cloud compositions, and their evolution with temperature, give rise to spectral signatures which define the L-T-Y spectroscopic classes (Oppenheimer et al., 1995; Kirkpatrick et al., 1999; Cushing et al., 2005; Kirkpatrick, 2005; Cushing et al., 2011).

While empirical approaches exist (e.g. Cruz et al., 2009; Filippazzo et al., 2015), the primary method of choice for inferring atmospheric properties relies upon detailed comparisons between theoretical models and the observed spectra. This often takes the form of pre-computing a large grid of theoretical spectra across a range of key physical parameters (Allard et al., 1996, 2012; Marley et al., 1996; Tsuji et al., 1996). Most commonly these grids include effective temperature and gravity, but more recently have been modified to include variable cloud models (Ackerman & Marley, 2001; Marley et al., 2002), eddy diffusion within the atmosphere (Saumon et al., 2006), rainout of specific condensates, and varying metallicity and carbon-to-oxygen (C/O) ratios (Marley et al., 2017; Mollière et al., 2017; Samland et al., 2017). These large grid models are then interpolated between grid points and fit via standard maximum likelihood comparisons (e.g. Cushing et al., 2008) or modern MCMC methods (e.g. Madhusudhan & Seager, 2011; Rice et al., 2015; Mann et al., 2015; Samland et al., 2017).

Though this grid modeling approach provides a useful baseline for beginning the analysis of infrared spectra, it has been shown to fail to accurately reproduce key spectral features, and often provides poor fits to the data (e.g. Leggett et al., 2017). For example, Patience et al. (2012) has demonstrated that grid models from different groups cannot reproduce statistically similar results for the same dataset of young brown dwarf companions. These inconsistencies between grid model fitting and the observational data suggest that not all of the possible atmospheric physics and chemistry is being taken into account within the established grid models. Despite this, a more recent effort in Baudino et al. (2017) found greater consistency between several widely-used grid models, though outstanding issues in abundance determinations still remain. These inconsistencies motivate the need for a new methodology to compliment the grid-modeling approach to reach a more complete understanding of brown dwarf spectra.

Realizing the limitations of the grid modeling approach, Line et al. (2015, 2017) (hereafter Parts I & II) applied well established atmospheric retrieval (Twomey et al., 1977; Fletcher et al., 2007; Lee et al., 2012; Line et al., 2012; Benneke & Seager, 2012) tools to the problem by performing a uniform retrieval analysis on a sample of late-T dwarfs. In Part I, the authors were able to validate their model on two benchmark T-dwarfs by showing that the overall retrieved abundances and C/O ratios were consistent with the objects’ stellar companion. With the larger sample (11 T7-T8 targets) in Part II, they found a strong depletion of the combined Na+K abundances with decreasing Teff. This had long been a theoretical expectation from rainout chemistry (Fegley & Lodders, 1994), and hypothesized from trends of near-infrared colors (Marley et al., 2002; Leggett et al., 2010; Liu et al., 2012; Lodieu et al., 2013), but the measured abundance depletion had never been directly detected. These investigations demonstrate that the retrieval method as applied to brown dwarf atmospheres is able to constrain key atmospheric properties often overlooked in traditional methods.

Our primary goals in this work, Part III, are to both expand the previously analyzed dataset into the cooler, early-Y dwarf (Y0-Y1) regime to see if the trends identified in Part II continue to cooler temperatures, and to test the various model assumptions made in Parts I & II. This is accomplished by performing a retrieval analysis on a set of objects from Schneider et al. (2015), which contains near-IR (0.9-1.7$\mu m$, Y,J,H band) spectra of 6 late-T and 16 early-Y dwarfs using Hubble Space Telescope’s Wide Field Camera 3 (HST,WFC3).

In Section 2 we briefly outline the methods of our atmospheric retrieval model. Section 3 discusses the dataset from WFC3 and the history of our targets. Constraints on the temperature structure (Section 4.1), evolutionary parameters (Section 4.2), and chemical abundances (Section 4.3) are then discussed. We also perform a comparison of our retrieval method with a recently published grid model in Section 4.4. In Section 5 we predict how well the future James Webb Space Telescope (JWST) will be able to improve our constraints. Finally, we list our primary conclusions in Section 6.