THE SPECTRAL ENERGY DISTRIBUTION OF THE COLDEST KNOWN BROWN DWARF^*

We have observed WISE 0855–0714 in six optical and near-IR filters to better characterize its SED and in a mid-IR band at 4.5 μm with Spitzer to provide additional astrometry for refining its parallax. We also have measured photometry from previous images in a second Spitzer band at 3.6 μm. To illustrate how the bandpasses of these filters align with the expected spectral features of WISE 0855–0714, we plot in Figure 1 the filter transmission profiles and an example of a model spectrum for a 250 K brown dwarf at a distance of 2 pc (Morley et al. 2014b). When possible, we selected filters that primarily encompassed continuum emission rather than deep absorption bands, as in the case of F127M on HST and the CH₄ continuum filter on VLT. For reference, we include in Figure 1 the YJHK filters from the Mauna Kea Observatories photometric system (Tokunaga et al. 2002).

Zoom In Zoom Out Reset image size

WISE 0855–0714 has been observed on multiple occasions by Spitzer's Infrared Array Camera (IRAC; Fazio et al. 2004). It contains two 256 × 256 arrays that are currently operating, each with a plate scale of 1.2'' pixel⁻¹ and a field of view of 52 × 52. The two arrays cover adjacent areas on the sky in different filters that are centered at 3.6 and 4.5 μm (denoted as [3.6] and [4.5]). Point sources in these bands exhibit FWHM = 1.7''.

All IRAC observations of WISE 0855–0714 are listed in Table 1. Luhman (2014) obtained the data on the first two dates, which were in 2013 and 2014. On the first date, WISE 0855–0714 was imaged in both [3.6] and [4.5]. The primary purpose of the second observation was the measurement of astrometry for WISE 0855–0714, so it was observed only in the band in which it is brightest, namely, [4.5]. The next two epochs of data were collected in 2014 by Luhman & Esplin (2014), which were used to measure additional astrometry. The second of those observations consisted of a large map (840'' × 840'') rather than a single field of view of the camera. Melso et al. (2015) performed a second map of the same size in 2015, corresponding to the fifth epoch overall. They searched for companions to WISE 0855–0714 in those two maps based on common proper motions. During all of the observations of WISE 0855–0714 in [4.5], images of a flanking field were taken in [3.6]. As a result, when the two maps were obtained in [4.5], the flanking fields in [3.6] produced maps that encompassed WISE 0855–0714. To continue the astrometric monitoring, we have obtained images in [4.5] on two additional dates in 2015. T. Esplin et al. (2016, in preparation) also have performed continuous imaging of WISE 0855–0714 in [3.6] and [4.5] during two 23 hr periods to characterize its variability.

We have measured astrometry for WISE 0855–0714 from all epochs of [4.5] images except the two photometric monitoring campaigns. During the monitoring for a given filter, WISE 0855–0714 was placed in one corner of the array and held at a fixed location without dithering, which optimizes photometric precision over the course of an observation but produces larger astrometric errors than the standard observing strategy of dithering the target near the center of the array. To measure astrometry for WISE 0855–0714 from the other seven epochs, we applied the methods described in Luhman & Esplin (2014) and the distortion corrections from Esplin & Luhman (2016). The resulting astrometric measurements are presented in Table 2.

We also have measured photometry for WISE 0855–0714 in [3.6] and [4.5] from all of the data except the monitoring campaigns and the second epoch in [3.6]. The latter set of data was excluded because WISE 0855–0714 was slightly blended with a brighter star. WISE 0855–0714 is much brighter in [4.5], so its photometry was unaffected by the neighboring star. For the monitoring campaigns, we have adopted the mean photometry measured by T. Esplin et al. (2016, in preparation). We have measured photometry from the other data with the methods from Luhman et al. (2012). The resulting photometric data are listed in Table 3.

Table 3.  Photometry of WISE J085510.83–071442.5

Band	Magnitude	Reference
i'	>27.2a	1
z'	>24.3a	2
m850	${26.85}_{-0.44}^{+0.31}$	1
Y	>24.4a	3
m105	27.33 ± 0.19	1
m110	26.71 ± 0.19	1
m110	26.47 ± 0.13	1
m110	26.00 ± 0.12	1
m127	24.52 ± 0.12	1
m127	24.49 ± 0.11	1
m127	24.36 ± 0.09	1
J	${25.0}_{-0.53}^{+0.33}$	4
CH4 cont	23.2 ± 0.2	1
H	>22.7a	5
Ks	>18.6a	6
W1	17.82 ± 0.33	5
W2	14.02 ± 0.05	5
[3.6]	17.44 ± 0.05	1
[3.6]	17.30 ± 0.05	1
[3.6]	17.34 ± 0.02	7
[3.6]	17.28 ± 0.02	7
[4.5]	13.88 ± 0.02	1
[4.5]	13.90 ± 0.02	1
[4.5]	13.92 ± 0.02	1
[4.5]	13.93 ± 0.02	1
[4.5]	13.86 ± 0.02	1
[4.5]	13.82 ± 0.02	1
[4.5]	13.84 ± 0.02	7
[4.5]	13.86 ± 0.02	1
[4.5]	13.80 ± 0.02	7

Note. All data are Vega magnitudes. For bands with multiple measurements, the data are listed in the order of the dates of observations from Table 1, with the exception of the second epoch in [3.6], for which a measurement is not presented because of blending with another star.

^aS/N < 3.

References. (1) This work; (2) Kopytova et al. 2014; (3) Beamin et al. 2014; (4) Faherty et al. 2014; (5) Wright et al. 2014; (6) VISTA Hemisphere Survey; (7) T. Esplin 2016, in preparation.

Download table as:  ASCIITypeset image

We obtained images of WISE 0855–0714 with the High Acuity Wide-field K-band Imager (HAWK-I) on the Unit Telescope 4 of the VLT. The camera contains four 2048 × 2048 HAWAII-2RG arrays and has a plate scale of 0.106'' pixel⁻¹ (Kissler-Patig et al. 2008). For these observations, we selected a medium-band filter that encompasses the H-band continuum that separates two bands of CH₄ and H₂O absorption (see Figure 1). Images of WISE 0855–0714 in that filter were taken during portions of six nights across a period of nearly 2 months (see Table 1). The total exposure time was 6.4 hr. Based on its proper motion and parallax (Section 3.1), WISE 0855–0714 moved ∼1'' between the first and last nights. Therefore, when registering and combining the individual frames from the six nights, we applied offsets to 3'' × 3'' sections of the images surrounding the expected positions of WISE 0855–0714 that compensated for the proper and parallactic motions. Point sources in the final combined image exhibited FWHM ∼ 0.45''. We aligned the world coordinate system (WCS) of the image to that of the IRAC images using offsets in right ascension, declination, and rotation that were derived from sources detected in both sets of data.

A source appears at the expected position of WISE 0855–0714 in the reduced image. A section of the HAWK-I image surrounding that position is shown in Figure 2. The image has been smoothed to a lower resolution of 0.7'' to facilitate visual identification of the detection. We conclude that this source is WISE 0855–0714 since no (stationary) object is detected at that location in the HST images, which are deeper than the HAWK-I data. We derived the flux calibration with H-band photometry from the Point Source Catalog of the Two-Micron All Sky Survey (2MASS; Skrutskie et al. 2006) for sources in the image under the assumption that they have ${m}_{\mathrm{CH}4}-H\sim 0$. Aperture photometry was then measured for WISE 0855–0714 using an aperture radius of 5 pixels and radii of 7 and 11 pixels for the inner and outer boundaries of the sky annulus, respectively. The photometry for WISE 0855–0714 is in Table 3. We have not used the HAWK-I image to measure astrometry for WISE 0855–0714 because the errors would be larger than those from the IRAC and HST images, which offer higher a signal-to-noise ratio (S/N) and higher resolution, respectively.

Zoom In Zoom Out Reset image size

At the time of the planning of our initial observations with HST, WISE 0855–0714 had been detected only at mid-IR wavelengths. Therefore, we began by observing it in the filter on HST that appeared to offer the greatest sensitivity to cold brown dwarfs, F110W on Wide Field Camera 3 (WFC3; Kimble et al. 2008). The same approach was taken by Luhman et al. (2014) in seeking the first near-IR detection of the brown dwarf WD 0806-661 B. After detecting WISE 0855–0714 in F110W, we pursued imaging with HST in three additional bands that are aligned with the wavelengths at which much of the near-IR flux is predicted to emerge, consisting of F850LP on the Advanced Camera for Surveys (ACS) and F105W and F127M on WFC3 (see Figure 1).

We observed WISE 0855–0714 with the IR channel of WFC3 and the Wide Field Channel (WFC) of ACS. WFC3/IR contains a 1024 × 1024 HgCdTe array in which the pixels have dimensions of ∼0.135'' × 0.121''. ACS/WFC contains two 2048 × 4096 SITe CCD arrays with plate scales of ∼0.05'' pixel⁻¹. For each of the four filters that we selected, WISE 0855–0714 was observed during six orbits that were divided into three two-orbit visits. In a given orbit, one exposure was taken at each position in a three-point dither pattern. The dither patterns in the two orbits in each visit were offset by 3.5 pixels along the x-axis of the array. The position of WISE 0855–0714 predicted by its proper motion and parallax was placed at the IR and WFC1-CTE apertures in WFC3 and ACS, respectively. The WFC1-CTE aperture was selected because it is near one of the readout amplifiers, which minimizes photometric losses due to imperfect charge transfer efficiency. The dates and exposure times for the visits are listed in Table 1.

The WFC3 and ACS images were registered and combined using the tasks tweakreg and astrodrizzle within the DrizzlePac software package. We adopted drop sizes of 0.85 native pixels and resampled plate scales of 0.065'' pixel⁻¹ and 0.035'' pixel⁻¹ for WFC3 and ACS, respectively. For F110W and F127M, we combined the six exposures within each two-orbit visit, resulting in one reduced image for each of the three visits. WISE 0855–0714 is well detected in each of those images, as shown in Figures 2 and 3. Because the F127M visits spanned only 1 week, the movement of WISE 0855–0714 among those visits was very small. As a result, we have included an image from only one of the three F127M visits in Figure 3. Because the S/N of WISE 0855–0714 is low in F850LP and F105W, we combined all exposures from the three visits for each of those filters. As done with the HAWK-I data, we corrected for the expected motion of WISE 0855–0714 among the visits within a small area (1.3'' × 1.3'') surrounding its expected location when registering and combining the images in F850LP and F105W. The reduced images for those filters are shown in Figure 3. WISE 0855–0714 is clearly detected in F105W, but only a marginal detection (S/N ∼ 3) is present in F850LP. Because the F850LP data were taken within a few days of two of the F127M visits, the expected position of the brown dwarf in F850LP is known precisely, and that position does coincide with the weak source indicated in Figure 3.

Zoom In Zoom Out Reset image size

To measure astrometry for WISE 0855–0714, we began by aligning the WCS of the image from the first F110W visit to that of the HAWK-I image, which was aligned to IRAC (Section 2.3). We then aligned the WCSs of the other WFC3 and ACS images to the new WCS for the first F110W visit. The astrometry for WISE 0855–0714 from each of the visits in F110W and F127M is provided in Table 2. We do not report astrometry from the F850LP and F105W images because of the large errors that result from the low S/N. As with the IRAC astrometry (see Luhman & Esplin 2014), we have estimated the astrometric errors based on the differences in right ascension and declination between different visits for stars with similar magnitudes to WISE 0855–0714.

Aperture photometry was measured for WISE 0855–0714 from each of the reduced images using an aperture radius of 4 pixels and radii of 4 and 10 pixels for the inner and outer boundaries of the sky annulus, respectively. For the WFC3 images, we measured aperture corrections of 0.097 (F105W), 0.097 (F110W), and 0.125 mag (F127M) between those apertures and radii of 0.4'' using bright stars in the images. We then applied those corrections and the zero-point Vega magnitudes of 25.4523 (F105W), 25.8829 (F110W), and 23.4932 (F127M) for 0.4'' apertures³ to the photometry of WISE 0855–0714. To calibrate the F850LP photometry, we computed the zero-point STMAG magnitude for a 5.5'' aperture from the image header keyword photflam and converted it to the Vega system with the transformation from Sirianni et al. (2005), arriving at a value of 24.316. Because the aperture correction in F850LP is significantly larger for redder objects, we used the software package synphot to estimate a correction of 0.68 mag between the aperture applied to WISE 0855–0714 and a 5.5'' aperture. Drizzled images can contain correlated noise, which would lead to an underestimate of the photometric errors. Therefore, to estimate reliable values for the errors, we created separate versions of the reduced images that used drop sizes of 0.1 native pixels, which should minimize the correlated noise. We then adopted the errors produced by aperture photometry on those images. The photometry in F850LP, F105W, F110W, and F127M is presented in Table 3, where separate measurements are reported for each of the three visits in F110W and F127M.

We obtained images of WISE 0855–0714 in the i' filter with Gemini Multi-Object Spectrograph (GMOS) at the Gemini South telescope. We originally proposed to conduct these observations in the z' filter, but we selected i' after deeper imaging in F850LP (≈z') with HST was approved. GMOS contains three 2048 × 4096 Hamamatsu CCD arrays that have plate scales of 0.08'' pixel⁻¹. WISE 0855–0714 was observed during portions of four nights that spanned nearly 2 months with a total exposure time of 6.8 hr (see Table 1). The FWHM of point sources in the images ranged from 0.4'' to 0.8''. As done with other data, when registering and combining the individual frames, we compensated for the parallactic and proper motions of WISE 0855–0714 within a 3'' × 3'' section surrounding its expected location. It was not detected in the final combined image (see Figure 3). In Table 3, we provide the magnitude limit in i' that corresponds to S/N = 3.

The proper motion and parallax of WISE 0855–0714 have been previously measured by Luhman (2014), Wright et al. (2014), and Luhman & Esplin (2014). For the most recent measurements, Luhman & Esplin (2014) combined four epochs of astrometry from Spitzer with three epochs from WISE during its initial survey and after reactivation (NEOWISE; Mainzer et al. 2014) that were derived by Wright et al. (2014). Luhman & Esplin (2014) arrived at a proper motion of (μ_α cos δ, μ_δ) = (−8.10 ± 0.02, 0.70 ± 0.02'' yr⁻¹) and a parallax of 0.433±0.015''. We can further refine those parameters with the new astrometry from our IRAC and HST images. To do that, we employ the two epochs of astrometry from WISE (Wright et al. 2014) after the adjustments by Luhman & Esplin (2014), the astrometry that we have measured from all seven epochs of Spitzer [4.5] images in which WISE 0855–0714 was near the center of the array (i.e., excluding the photometric monitoring campaigns), and our astrometry from the six HST visits in F127M and F110W. Those data are compiled in Table 2. As mentioned in Sections 2.2 and 2.3, we have not measured astrometry from the F850LP, F105W, and HAWK-I images because the errors would be much larger than those for the other data that we are utilizing. We also exclude the NEOWISE astrometry for the same reason. Although the WISE data have fairly large errors, we retain them because they significantly extend the baseline of the astrometry.

We performed least-squares fitting of proper and parallactic motion to the 15 epochs of astrometry for WISE 0855–0714 in Table 2 with the IDL program MPFIT. The reduced χ² for the resulting fit was 0.4. To verify the errors produced by the fitting, we created 10,000 sets of astrometry that consisted of the sum of the measured astrometry and Gaussian noise. We then fitted parallactic and proper motion to each set. The resulting standard deviations of μ_α, μ_δ, and parallax were similar to the errors from MPFIT. The proper motion and parallax are presented in Table 4. They are consistent with all of the previous estimates. In Figure 4, we plot the relative coordinates among the 15 epochs after subtraction of the best-fit proper motion.

Zoom In Zoom Out Reset image size

We wish to compare the available photometry for WISE 0855–0714 to the predictions from models of the atmospheres and interiors of brown dwarfs. A comparison of a given magnitude or color for a brown dwarf to the model predictions can serve as a test of the validity of the model if the defining properties of the brown dwarf are assumed, or it can constrain the properties of the brown dwarf if the model photometry is assumed to be accurate. If several photometric measurements (or a spectrum) are available, then one can pursue both a test of the models and a constraint on the brown dwarf's properties by checking whether one suite of models (e.g., cloudless with equilibrium chemistry) produces a clearly superior fit, and then adopting the properties of the best-fit model from that suite. Such comparisons of data and models can be performed for a single object or for a population. In this section, we summarize previous comparisons of photometry for WISE 0855–0714 to model predictions, compare it and other known Y dwarfs to the models on color–magnitude diagrams (CMDs), and perform a comparison to the full SED that we have measured for WISE 0855–0714.

3.2.1. Previous Studies of WISE 0855–0714

3.2.2. Color–Magnitude Diagrams

We can use CMDs to place the photometry of WISE 0855–0714 in the context of data for other brown dwarfs and to compare trends within this population to model predictions. Because few data are available for Y dwarfs in optical bands (Lodieu et al. 2013; Kopytova et al. 2014; Leggett et al. 2015), we have not constructed CMDs with i' and F850LP (≈z'). Previous data for F105W, F110W, F127M, and the CH₄ continuum filter are also limited, but most known Y dwarfs have been observed in the overlapping bands of Y, J, and H (see Figure 1). Therefore, we select the latter three bands and the two Spitzer filters for our CMDs. We plot M_4.5 on the vertical axis of each diagram because it encompasses less absorption and exhibits higher S/N for Y dwarfs than the other filters, and because it captures most of the flux of Y dwarfs at <5 μm.

In Figure 5, we show CMDs with colors that extend from Y/J/H/[3.6] to [4.5]. We also plot CMDs with Y − J and J − H in Figure 6. To place WISE 0855–0714 in these CMDs, we have combined our measurements of m₁₀₅, m₁₂₇, and m_CH4 with $Y-{m}_{105}$, $J-{m}_{127}$, and $H-{m}_{{CH}4}$ as predicted by the models at the value of M_4.5 for WISE 0855–0714. The three suites of models that we consider (Section 3.2.1) produce similar values of $J-{m}_{127}$ (∼0.89) and $H-{m}_{{CH}4}$ (∼0.85), but the predicted $Y-{m}_{105}$ ranges from −0.25 to −0.9. We adopted $Y-{m}_{105}=-0.6$ for plotting WISE 0855–0714 in Figure 5. Its position relative to the models in the $Y-[4.5]$ CMD does not change significantly if a different value is adopted given that the model suites span more than 3 mag in $Y-[4.5]$ at the magnitude of WISE 0855–0714. For m₁₂₇ and [4.5], we have adopted the means of the multiple measurements that are available. WISE 0855–0714 is plotted with an error bar in $J-[4.5]$ that corresponds to the range of the three m₁₂₇ measurements. For [3.6]–[4.5], we have adopted the mean color that was measured in the four epochs with [3.6] data. Because model colors have been used to convert from m₁₀₅/m₁₂₇/m_CH4 to Y/J/H for WISE 0855–0714, the CMDs are technically showing the positions of WISE 0855–0714 relative to the models in ${m}_{105}/{m}_{127}/{m}_{{CH}4}-[4.5]$ and the positions of other Y dwarfs relative to the models in $Y/J/H-[4.5]$.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the CMDs, we have plotted a sample of T dwarfs (Dupuy & Liu 2012, references therein) and all known Y dwarfs that have measurements of parallaxes and photometry in relevant bands (Cushing et al. 2011, 2014; Kirkpatrick et al. 2012, 2013; Luhman et al. 2012, 2014; Tinney et al. 2012, 2014; Beichman et al. 2013, 2014; Dupuy & Kraus 2013; Leggett et al. 2013, 2015, 2016; Schneider et al. 2015). For the Y dwarf WISE J035000.32–565830.2 (hereafter WISE 0350–5658), the parallax of 0.291 ± 0.50'' from Marsh et al. (2013) places it at a discrepant location in CMDs relative to other Y dwarfs (Leggett et al. 2015, 2016). Meanwhile, four known Y dwarfs lack previous parallax measurements, consisting of WISE J030449.03−270508.3 (Pinfield et al. 2014), WISE J082507.35+280548.5 (hereafter WISE 0825+2805), WISE J120604.38+840110.6 (hereafter WISE 1206+8401), and WISE J235402.77+024015.0 (Schneider et al. 2015). A significant amount of multi-epoch Spitzer data is now publicly available for WISE 0350–5658, WISE 0825+2805, and WISE 1206+8401. To resolve the discrepancy in the CMD locations of WISE 0350–5658 and to allow the addition of WISE 0825+2805 and WISE 1206+8401 to the CMDs, we have measured their parallaxes from the Spitzer data with the same methods that were applied to WISE 0855–0714, arriving at 0.184 ± 0.01'', 0.158 ± 0.007'', and 0.085 ± 0.007'', respectively. It is likely that these values will be soon superseded by measurements that include the Spitzer images that are not yet publicly available, as well as astrometry from other telescopes. Using our new parallaxes, all three of these objects exhibit locations in the CMDs that are consistent with the sequences formed by other Y dwarfs.

For comparison to the data in the CMDs, we have included magnitudes and colors predicted by three suites of models that were mentioned in Section 3.2.1: cloudless with chemical equilibrium, cloudless with nonequilibrium chemistry, and 50% cloud coverage with chemical equilibrium (Saumon & Marley 2008; Morley et al. 2012, 2014b; Saumon et al. 2012). The ages of the Y dwarfs in the CMDs are unknown, with the exception of WD 0806-661 B (2 ± 0.5 Gyr; Luhman et al. 2012), so we show model isochrones for ages of 1, 3, and 10 Gyr, which span the ages of most stars in the solar neighborhood. The models with equilibrium and nonequilibrium chemistry are plotted for temperatures of <450 and <350 K, respectively.

In Figure 5, the sequences formed by known Y dwarfs are qualitatively similar among the CMDs with $Y-[4.5]$, $J-[4.5]$, and $H-[4.5]$. The CMD with [3.6]–[4.5] contains the most well-defined sequence because [3.6] and [4.5] usually offer the most accurate photometry for Y dwarfs from available bands. Because of their large scatter, the sequences of Y dwarfs brighter than WISE 0855–0714 in $Y/J/H-[4.5]$ do not tightly constrain the models, but those sequences are in rough agreement with all three model suites, agreeing somewhat better with the cloudless/chemical equilibrium models in $Y-[4.5]$ and $J-[4.5]$. All of the models are much redder than the data in [3.6]–[4.5], as found in a number of previous studies (Leggett et al. 2010; Beichman et al. 2014; Luhman et al. 2014). It is unfortunate that the CMD with the best-defined Y dwarf sequence contains a color that is especially difficult for the models to reproduce. The differences between the three suites of models in the CMDs of Y/J/H − [4.5] increase at fainter magnitudes, so WISE 0855–0714 offers the greatest potential for discriminating among those models. However, the suite of models that best matches the position of WISE 0855–0714 changes from one CMD to the next: cloudless/chemical equilibrium in $Y-[4.5]$, cloudless/nonequilibrium chemistry in $J-[4.5]$, and cloudy in $H-[4.5]$. For the $J-[4.5]$ CMD, Luhman & Esplin (2014) arrived at a similar result using the J-band measurement from Faherty et al. (2014). We note that the overall agreement between the Y dwarf sequences and the model isochrones is not improved by replacing [4.5] with a different band on the vertical axes in the CMDs.

The Y − J CMD in Figure 6 exhibits a well-defined sequence of Y dwarfs plus two blue outliers, WISE 0350–5658 and WISE 182831.08+265037.8. The sequence of late T and Y dwarfs becomes bluer at fainter magnitudes, as found in many previous studies (Liu et al. 2012; Dupuy & Kraus 2013; Leggett et al. 2013, 2015, 2016; Morley et al. 2014b; Schneider et al. 2015). That trend has been attributed to the depletion of neutral alkalis into gases and solids (Liu et al. 2012), which would reduce the absorption from pressure-broadened alkali lines at red optical wavelengths (Marley et al. 2002; Burrows et al. 2003). The cloudless models with chemical equilibrium reproduce the Y dwarf sequence in Y − J, while the isochrones from the other two suites of models are too red (Liu et al. 2012; Leggett et al. 2013, 2015; Morley et al. 2014b). Tremblin et al. (2015) found that their cloudless models produced similar Y − J colors for equilibrium and nonequilibrium chemistry, both of which matched the data for Y dwarfs. Whereas Y dwarfs have previously exhibited bluer Y − J at fainter magnitudes, our data imply a redder color for WISE 0855–0714. A similar result is found when WISE 0855–0714 is compared to other Y dwarfs in terms of ${m}_{105}-J$ using F105W photometry from Schneider et al. (2015). As shown in Figure 6, all three sets of models do predict a shift of the Y dwarf sequence back to redder values of Y − J for objects as faint as WISE 0855–0714.

As in the Y − J CMD, a clear extension of the T dwarf sequence is evident among the Y dwarfs in the J − H CMD, except with more discrepant objects in the latter. The blue outliers are WISE J053516.80–750024 and WISE J014656.66+423410.0, and the reddest outlier is WISE 182831.08+265037.8. Previous studies have compared similar J − H CMDs (often with M_J) to model predictions (Morley et al. 2012, 2014b; Leggett et al. 2013, 2015, 2016; Marsh et al. 2013; Beichman et al. 2014), finding that the cloudless and partly cloudy models of Saumon et al. (2012) and Morley et al. (2012, 2014b) tend to produce colors that are too blue (see also Figure 6). Those models agree with the observed colors only if the surface gravity is increased to a value that becomes unphysical for cooler Y dwarfs (log g ≳ 5) and the cloud coverage is very large (Morley et al. 2014b). The cloudless nonequilibrium models in Figure 6 are also too blue, although those from Tremblin et al. (2015) are red enough to match the J − H data (Leggett et al. 2016). Previous studies have shown that Y dwarfs start to become redder in J − H at fainter magnitudes (Leggett et al. 2015, 2016; Schneider et al. 2015), and we find that WISE 0855–0714 continues that trend, which agrees with the model predictions (see Figure 6).

3.2.3. Spectral Energy Distribution

In Figure 7, we have constructed an SED for WISE 0855–0714 from our photometric measurements in i', F850LP, F105W, F110W, F127M, the CH₄ continuum filter, [3.6], and [4.5]. The other available data in Table 3 are omitted from the SED because they are superseded by our more sensitive photometry in similar bands or they are not deep enough to provide a useful constraint on the models (K_s). As in Figure 1, we include in Figure 7 an example of a model spectrum of a cold brown dwarf (Morley et al. 2014b) to indicate the wavelengths of major spectral features relative to the bands of our photometry.

Zoom In Zoom Out Reset image size

For each of the three model suites considered in this work and for each of the ages of 1, 3, and 10 Gyr, we have identified the model that has the same value of M_4.5 as WISE 0855–0714 because this band encompasses most of the flux at <5 μm. For each of those nine models, we then computed the difference between the observed and predicted absolute magnitudes for each of the bands in the SED for WISE 0855–0714. In Figure 7, we plot the resulting differences for the three model suites at 3 Gyr. The temperatures for those best-fit models are 237, 244, an 249 K for cloudless/chemical equilibrium, cloudless/nonequilibrium chemistry, and 50% cloudy, respectively. The other two ages have been omitted because, for a given model suite, most of the magnitude differences for 1, 3, and 10 Gyr span a modest range on the scale of Figure 7 (≲0.5 mag). Positive and negative deviations correspond to predicted fluxes that are brighter and fainter than the data, respectively.

For all of the model suites considered, the predicted SEDs exhibit large deviations from the observed SED of WISE 0855–0714. No single model provides a clearly superior match to the data. When the models are selected to match the object at [4.5] as we have done, they are too faint in H and [3.6]. That is a reflection of the fact that the models are too red in $H-[4.5]$ and [3.6]–[4.5], as found in the CMDs. The deviation in H is smallest for the cloudy models, which is also evident from the $H-[4.5]$ CMD in Figure 5. However, the two suites of cloudless models agree better with the SED of WISE 0855–0714 in F850LP through F127M. The cloudy models are too bright by 1.5–2.5 mag in those bands, where the deviation becomes progressively larger at shorter wavelengths.

If one selected a model that matched WISE 0855–0714 in a different band than [4.5], all of the deviations in Figure 7 would shift vertically in the same direction so that the new normalization band exhibited zero deviation. In addition, because the temperature of that model would differ from that derived by fitting to [4.5] and because the near-IR fluxes are more sensitive to temperature than the flux in [4.5], the shifts of the deviations would be larger in the near-IR bands. As with the CMDs, adopting a different band than [4.5] in this analysis does not improve the agreement between the observed SED and the model predictions.

Among the images in which WISE 0855–0714 has been detected, those from HST offer the highest resolution and therefore can detect a companion at the smallest separations. The F110W and F127M images have provided the highest S/N for WISE 0855–0714 from among the four filters with which it was observed by HST. The S/N in each of those filters is similar to that of the brown dwarf WD 0806-661 B in F110W images from Luhman et al. (2014). Thus, the constraints on the presence of a companion to WISE 0855–0714 are similar to those derived from the images of WD 0806-661 B, which were capable of detecting ∼80% of companions with Δm₁₁₀ ≲ 0.7 mag at separations beyond 0.13'' (≳0.3 AU for WISE 0855–0714).

Because of improvements in the accuracies of parallaxes for Y dwarfs (Dupuy & Kraus 2013; Beichman et al. 2014; Tinney et al. 2014; Section 3.1), the Y dwarf sequence in the [3.6]–[4.5] CMD (Figure 5) is now sufficiently narrow and well defined that one could attempt to identify unresolved binaries via their elevated positions relative to the sequence. That method is not yet applicable to WISE 0855–0714 since no other Y dwarfs have been found near its absolute magnitude.