INDICATIONS OF WATER CLOUDS IN THE COLDEST KNOWN BROWN DWARF^*

Brown dwarfs are substellar mass objects with effective temperatures (T_eff) ranging from stellar-like at the high end (∼3000 K) to planet-like at the low-end. Many brown dwarfs are found in isolation (making direct observations and spectra obtainable), and they can share T_eff s, luminosities, ages, and/or masses with directly imaged exoplanets. Therefore, they are a gateway population to characterization studies (e.g., Delorme et al. 2012; Faherty et al. 2013b, 2013a; Kirkpatrick et al. 2012; Liu et al. 2013; Beichman et al. 2014).

NASA's Wide-Field Infrared Explorer (WISE) mission launched in late 2009, revolutionized our understanding of the coldest brown dwarfs (Wright et al. 2010). Follow-up of WISE-selected targets led to the discovery of nearly 20 objects in the T_eff ∼ 300–500 K temperature range and the definition of a new spectral subtype—the "Y" dwarfs (Cushing et al. 2011, 2014; Kirkpatrick et al. 2011, 2012, 2013; Tinney et al. 2012; Burgasser et al. 2011). These cold brown dwarfs have distinctly lower luminosity (Dupuy & Kraus 2013; Tinney et al. 2014) from warmer "T" dwarf sources that have been studied for the past 15 yr (e.g., Lucas et al. 2011; Burningham et al. 2011a, 2011b).

Y dwarf temperatures are more similar to those of the cold, gas-giant exoplanets known to orbit solar-aged stars than are warmer brown dwarfs: the temperatures and atmospheres of which have more in common with low-mass stars. Their spectra are characterized by remarkably strong CH4 absorption, resulting in J and H band peaks that are narrower than those for warmer T dwarfs (Cushing et al. 2011). Studies of the population of these extremely cold brown dwarfs leads to a measurement of both the ability of star formation processes to make brown dwarfs at the lowest masses, and the mass cut-off for the star formation process itself. Studies of their individual spectral energy distributions reveal the complex chemical processes that dominate low-temperature atmospheres, directly applicable to interpreting exoplanet data.

Recently, Luhman (2014b) discovered the coldest brown dwarf yet detected (T_eff ∼ 225–260 K) as the fourth closest system to the Sun. The source, designated WISE J085510.83-071442.5 (hereafter W0855), is important for studies of cold, gas-giant-planet-like atmospheres. Notably, it allows us to characterize an important opacity source for studies of exoplanets—the clouds—at even lower temperatures than possible in work to date.

In this Letter, we present a deep J3 band image of W0855. A comparison of our photometry with recent model atmosphere predictions from Morley et al. (2014) provides evidence for sulfide and water ice clouds. Section 2 details the background data on W0855. Section 3 describes the data acquired for this Letter and Section 4 investigates the position of W0855 on color–magnitude diagrams. We present the conclusions in Section 5.

In Luhman (2014a) and Kirkpatrick et al. (2014), the high proper motion of the source W0855 was noted between the 2010 May and November WISE images. However, at the time of the original detection by WISE, it was not realized that the photometry was contaminated by a cluster of background sources which skewed the photometry and astrometry (Luhman 2014b, Wright et al. 2014; and see the area surrounding the "WISE Positions" on Figure 1). Luhman (2014b) obtained Spitzer Space Telescope [3.6] and [4.5] μm photometry for W0855 and verified the motion. The astrometry placed W0855 among the four closest systems to the Sun with a parallax of 0.454 ± 0045 (or ∼2.2 pc) and a proper motion of 8.1 ± 01 yr⁻¹ (the third largest proper motion ever detected).

Zoom In Zoom Out Reset image size

Using the absolute magnitude at 4.5 μm and the (J - [4.5]) color limit, Luhman (2014b) compared to model predictions and estimated the T_eff for W0855 to be 225–260 K, making this the coldest compact source ever detected outside our own solar system.

Wright et al. (2014) recently reported new, uncontaminated WISE photometry with NEOWISE-R⁸ that is in far better agreement with the expected "Y" spectral type for this source than the first detections indicated. Wright et al. (2014) also refined the parallax and proper motion of W0855 using a new epoch, reporting π = 0.448 ±  0033 and μ =  8.08 ± 005 yr⁻¹.

Over a given age range of 1–10 Gyr (covering all possible ages for a thin/thick disk Milky Way object), the mass of this "brown dwarf" is 3–10 M_Jup making it a truly remarkable bridge between the lowest mass objects formed by star formation processes and the giant planets we are only beginning to obtain directly imaged information on. Indeed, it easily crosses into the "planetary mass regime" classically placed at 13 M_Jup (Burrows et al. 1997; see discussion in Luhman 2014b).

The majority of brown dwarfs are characterized in the near-infrared and all Y dwarfs have been classified using near-infrared spectroscopy. As follow-up spectroscopy for W0855 is extremely difficult given its cold temperature and low luminosity, near-infrared photometry is required to place it in context with the known brown dwarf population. Moreover, Morley et al. (2014) predict that water condenses in the upper atmospheres of objects with temperatures <450 K, leading to prominent features notable in the J band. As W0855 only had a lower limit on the J band photometry from Luhman (2014b), there was excellent motivation for obtaining a deep image of W0855 in the near-infrared.

3.1. FourStar Image

W0855 was observed with the FourStar infrared mosaic camera mounted on the 6.5 m Magellan Baade telescope at Las Campanas Observatory, Chile (Persson et al. 2013) as part of an ongoing parallax program (e.g., Tinney et al. 2012, 2014; Kirkpatrick et al. 2013). The instrument uses four 2048×2048 Teledyne HAWAII-2RG arrays that produce a 109×109 field of view at a plate scale of 0159 pixel⁻¹.

W0855 was observed at the start of each night on 2014 May 12–14 (UT) for 2–3 hr with the narrowband J3 filter. As demonstrated in Tinney et al. (2012), the J3 filter encompasses the J-band opacity hole for Y dwarfs. At the same time, its narrow wavelength coverage minimizes signal from the sky, making it an optimal choice for imaging the coldest brown dwarfs.

Total integration times collected during the imaging sequence for W0855 varied as did the seeing. As we were observing a target at 8 hr and −7° from Chile in May, we only had the start of the evening before the source fell to an unreasonably high airmass. The first night had the poorest conditions with an average seeing of 09– 10. The second and third nights varied but shared an average seeing of 04– 06.

Our imaging strategy employed a 20 s exposure at J3 with 6 co-adds and an 11 point dither pattern randomly distributed within a 15'' box. We collected 50 images on the first night, 61 images on the second, and 40 images on the third. Data reduction was done using the FourStar pipeline (A. J. Monson et al., in preparation). Raw frames were linearized, dark subtracted, and flat-fielded for each of the four arrays. Then, a first-pass local background was created for each frame using the adjacent nine in the following manner. (1) Each nearby frame had a low-order background term fit and subtracted to remove gradients, then the resultant images were averaged with 50% of the high values clipped to remove sources. (2) The backgrounds which were removed were averaged together without clipping but weighted by the differential time from the desired frame. (3) The two backgrounds (low-order and high-order) were added together to construct the first-pass background for each frame.

The background subtracted images were source-extracted with the Astromatic.net program Sextractor, fit with a World Coordinate System (WCS) grounded with matches to the Two Micron All-Sky Survey (2MASS; Cutri et al. 2003) using Scamp and then re-sampled using Swarp to construct a first-pass mosaic image. Point sources were then detected and individual masks updated for each original input frame.

A second-pass sky subtraction step proceeded like the first except instead of clipping the upper 50% of pixel values, objects were explicitly masked from each frame. A final mosaic was created from all three nights by weighing each exposure by:

$$\begin{equation} {\rm weight} = \frac{1e7}{({\rm fwhm}^{2} \times {\rm stdev}^{2} \times {\rm flxscale})}, \end{equation} \tag{ 1 }$$

where 1e7 is an arbitrary constant, fwhm is the seeing in arcseconds, stdev is the standard deviation from the background counts for each frame, and flxscale is a differential zero point term that takes into account slight airmass or weather related changes in image depth.

The total exposure time accumulated over the three nights was 5.03 hr, however the weighting strategy resulted in a final mosaic at an exposure time of 3.5 hr with a seeing of 05. Consequently, nearly all of the first night's data did not contribute to the final mosaic due to the poor conditions.

To calibrate the photometry in our image, we followed the same prescription described in Tinney et al. (2012). Within the final image, we reached a 5σ aperture limit at J3 of ∼23.8 or J_MKO ∼ 24.0 when the photometric relations from Tinney et al. (2014) are applied (J_MKO−J3 = 0.20 ± 0.03 for Y dwarfs). Figure 1 shows an ∼45'' × 25'' region of the final mosaic containing the previous WISE positions, the Luhman (2014b) Spitzer positions, the NEOWISE-R position from Wright et al. (2014) and our expected 2014 May 14 position.

3.2. The 2.6σ Detection

We investigated the expected position of W0855 in our images with great care to give a thorough limit on the detection. For this analysis, we focused only on a ∼30 square portion of the final mosaic centered on W0855. We first found all sources in a Gaussian smoothed image (with σ equal to the FWHM) with a signal to noise ratio (S/N) > 5σ. We then masked those out using a radius set by where a Gaussian would fall below 10% of the noise floor. We then conducted aperture photometry on the unsmoothed, expected position of W0855 as well as 1000 random places in the image (avoiding the masks) using a radius of 0.6731 × FWHM—based on the optimum size for sources whose point response function profiles can be approximated as Gaussian.⁹

The estimated position for W0855 on 2014 May 14 was deduced from the astrometry and NEOWISE-R position in the Wright et al. (2014) paper taken just 10 days prior to the FourStar image. We estimated a 1 pixel uncertainty by applying the FourStar plate-scale to the Wright et al. (2014) astrometric uncertainties added in quadrature with the WCS uncertainty (01 based on the fit to 2MASS positions). Aperture photometry was conducted at positions of x ± 1 pixel, y ± 1 pixel from the expected position. This accounts for the astrometric uncertainty and pinpoints the exact location of maximum flux associated with W0855. The estimated position of W0855 using the propagated NEOWISE-R position is (R.A. = 133.7861633, decl. = −7.2442513, EPOCH = 56792.0) whereas the 2.6σ position appears at (R.A. = 133.7861696, decl. = −7.2442643, EPOCH = 56792.0). Therefore, our detected FourStar position is <0.3 pixel from the expectation or consistent within the astrometric uncertainty. We note that the difference between the expected and measured position for W0855 is primarily in declination. The same procedure was used for the randomly chosen background pixels, therefore regardless of the 1 pixel buffer, the detections and consequent aperture comparisons are robust.

The cumulative and probability distribution functions for a detection at the position of W0855 are shown in Figure 2. We find that the W0855 aperture is brighter than ∼98% of the 1000 random apertures or equivalent to a 2.6σ detection. The uncertainty for this low-signal source is asymmetric and we measure J3 = 24.8$^{+0.53}_{-0.35}$—or J_MKO = 25.0$^{+0.53}_{-0.35}$ when the photometric relations from Tinney et al. 2014 are applied. This value is consistent with the 5σ J_MKO > 24.0 limit on our photometry.¹⁰

Zoom In Zoom Out Reset image size

Color–magnitude diagrams are a tool for investigating atmospheric properties of the brown dwarf population as well as testing model predictions (e.g., Tinney et al. 2003, 2014, Vrba et al. 2004; Dupuy & Liu 2012; Faherty et al. 2012, 2013b; Patten et al. 2006; Leggett et al. 2010, 2013). For warm brown dwarfs—primarily L dwarfs—iron and silicate grain clouds form thick dust layers and largely shape observable properties (e.g., Lodders 2003; Helling & Woitke 2006). As objects cool into the T dwarf phase, L-dwarf clouds rapidly clear and models predict that less refractory sulfide and salt clouds condense in the photospheres (e.g., Lodders 1999; Morley et al. 2012). For the coldest brown dwarfs, the Y dwarfs at T_eff < ∼450 K (see Figure 3), Morley et al. (2014) predict that water condenses in the upper atmosphere to form ice clouds. At T_eff < 350–375 K, Morley et al. (2014) also show that these water clouds become optically thick, scatter light at optical wavelengths through the J band and absorb in the infrared with prominent features. As such, infrared color–magnitude diagrams are particularly telling about the condensate species present in the photospheres of cold brown dwarfs.

Zoom In Zoom Out Reset image size

Using our new photometry (both the 5σ limit and 2.6σ detection), we look for evidence of water clouds in W0855. Figure 4 shows the J_MKO − W2 versus M_W2 for late-type T and Y dwarfs (Dupuy & Kraus 2013; Tinney et al. 2012; Beichman et al. 2014; Marsh et al. 2013). As stated above, Morley et al. (2014) recently published a set of atmosphere models applicable for the coldest brown dwarfs discovered (200 K < T_eff < 450 K) that include the influence of water clouds in addition to the sulfide clouds described in Morley et al. (2012; the model is a hybrid with 50% cloudy and 50% clear). We have overplotted both of these model predictions on Figure 4. The sulfide cloud model predictions that include Na₂S, MnS, ZnS, Cr, and KCl condensate clouds and are applicable for objects in the range (400 K < T_eff < 1300 K) are shown at log g = 5.0 for differing sedimentation parameters (f_sed = 5 to f_sed   = 2 as first defined in Ackerman & Marley 2001). The water cloud models are shown at f_sed = 5 for log g =  3.0 to log g = 5.0 with the applicable T_eff for the given position labeled on Figure 4. The clear (or cloudless) model predictions from Saumon et al. (2012) are also shown for log g = 3.0 to log g = 5.0, although those show little diversity for differing gravity values. At T_eff < 350 K, the clear model predictions begin to diverge from those that include water clouds, giving a lever for differentiating between the two opacity realms. We note that all models displayed are in chemical equilibrium for solar metallicity. When the range of these parameters are fully incorporated into the models (e.g., non-solar metallicity, non-equilibrium chemistry), predictions may vary (see, for example, Burningham et al. 2013; Hubeny & Burrows 2007).

Zoom In Zoom Out Reset image size

The previous near infrared limit of W0855 of J_MKO > 23.0 was not sufficiently deep enough to discriminate between model predictions (Luhman 2014b). Our new 5σ limit shifts the source 1 mag fainter and into the blue end of atmosphere model predictions that include water clouds. The 2.6σ detection of J_MKO = 25.0$^{+0.53}_{-0.35}$ places W0855 2.7σ from the clear model and squarely within sulfide and water cloud predictions for an object at 225 < T_eff< 250 K. The large uncertainty is inclusive of all gravity predictions, however, given the large tangential velocity (v_tan  =  85 ±  9 km⁻¹) of this source makes a young age unlikely (e.g., Faherty et al. 2009).

The recently discovered brown dwarf WISE J085510.83-071442.5 (W0855) is important for cold atmosphere studies and a gateway object for directly imaged gas-giant characterization programs. In this work, we present a deep near-infrared J3 band image for the source. We find a lower limit on its photometry of J3 > 23.8 which we convert to J_MKO > 24.0 at 5σ and a 2.6σ detection of J3 = 24.8$^{+0.53}_{-0.35}$ which we convert to J_MKO = 25.0$^{+0.53}_{-0.35}$. With J_MKO − W2 = 10.984$^{+0.53}_{-0.35}$ at 2.6σ—or equivalently an upper limit of J_MKO − W2 > 9.984 at 5σ—W0855 is the reddest brown dwarf yet characterized.

Investigating the position of W0855 on the J_MKO − W2 versus M_W2 color–magnitude diagram reveals that this is the first compact source outside our solar system to have evidence for water ice clouds. The 5σ limit reported herein is 1 mag deeper than previous works and places W0855 at the blue end of the water cloud model predictions from Morley et al. (2014). The 2.6σ detection places W0855 2.7σ from the cloudless models and squarely along the water cloud model predictions for a source at 225 K < T_eff < 250 K. A deeper near-infrared image coupled with advancements in the models to include non-equilibrium chemistry and varying metallicity at the temperature of W0855, will refine this result at a higher confidence level.

This publication uses data gathered with the 6.5 m Magellan Telescopes located at Las Campanas Observatory, Chile. We thank the operators, J. Araya and G. Martin, and the engineer D. Ossip for assistance in acquiring data. The authors thank E. Wright and D. Saumon for useful comments on the manuscript. Australian access to the Magellan Telescopes was supported through the National Collaborative Research Infrastructure and Collaborative Research Infrastructure Strategies of the Australian Federal Government. This research was supported by Australian Research Council grants DP0774000 and DP130102695. This publication makes use of data products from WISE, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory (JPL)/California Institute of Technology (Caltech), funded by the National Aeronautics and Space Administration (NASA). This research has made use of the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.