THE COLDEST BROWN DWARF (OR FREE-FLOATING PLANET)?: THE Y DWARF WISE 1828+2650

WISE 1828+2650 was observed with HST in the WFC3-IR F140W filter in conjunction with a grism spectrum which was the basis of this object's identification as a ⩾Y1 dwarf by (Cushing et al. 2011; Figure 1). The data consist of two pairs of individual 78 s exposures taken at positions offset by (Δα, Δδ) = 082 and 567 to reduce the effects of cosmic rays and the undersampling of the individual frames. The pipeline corrects for the geometric distortion of the WFC3-IR camera to a level estimated to be ∼5 mas (Kozhurina-Platais et al. 2009) which is of the same order or less than the extraction uncertainties of the faint target. Over 200 sources were extracted from the Space Telescope Science Institute's (STScI) "MultiDrizzle" mosaic using the Gaussian-fitting IDL FIND routine to determine centroid positions and the APER routine⁸ with a 3 pixel radius for photometric measurements. The FWHM for these undersampled data is 026 consistent with STScI analyses (Kozhurina-Platais et al. 2009). The data were calibrated using the zero point for Vega magnitudes, ZP(Vega) = 25.1845 mag,⁹ from the WFC3 Handbook (Rajan 2010). A strong source denoted Source A appears to be slightly extended and is located roughly 5'' W and 3'' S of WISE 1828+2650. This object is seen in the Keck, HST, and Spitzer images and will be used as a local astrometric reference in subsequent analysis. As discussed below, the extracted position of Source A is consistent between the Keck and HST measurements to within 10 mas which is small compared to the other uncertainties.

Zoom In Zoom Out Reset image size

WISE 1828+2650 was observed in the H band using the Keck-2 telescope (2010 July) using NIRC2 with the laser guide star adaptive optics (AO) system (Wizinowich et al. 2006; van Dam et al. 2006) and a tip-tilt star located 50'' away. The wide-field camera (40 mas pixel⁻¹ scale) was used as a compromise between good spatial resolution and covering enough field of view (40'') to ensure an adequate number of reference stars for astrometry. A sequence of images with a total integration time of 1080 s was obtained at airmass of 1.0–1.1 with 2''–10'' offsets in right ascension and declination (Kirkpatrick et al. 2011). The individual images were sky-subtracted with a sky frame created by the median of the science frames and flat-fielded with a dome flat using standard and custom IDL routines. Individual images were "de-warped" to account for optical distortion in the NIRC2 camera as described below. The reduced images were shifted to align stars on a common, larger grid and the median average of overlapping pixels was computed to make the final mosaic. The target was detected at a relatively low S/N of ∼5. As reported in Kirkpatrick et al. (2012), the source was also weakly detected in the J band with NIRC2 at this epoch (Table 2).

The source was re-observed in 2011 (Figure 2) and twice in 2012 following the same procedure but for longer total integration times (3960, 2880, and 2880 s, respectively), yielding S/N  ∼  10 and with smaller offsets (2''–4'') to reduce the effects of optical distortion on the final image. The airmass for the 2011 October observations ranged from 1.1 to 1.4 with a mean value of 1.25. In 2012 July, the airmass values ranged between 1.04 and 1.15 with a mean value of 1.09. In 2012 October, the seeing and photometric conditions were relatively poor with an average airmass of 1.2. The typical point-spread function (PSF) in the final mosaicked images from the first three epochs varies slightly across the field, but at the center of the field close to WISE 1828+2650, the FWHM of the AO-corrected core is ∼3–4 pixels, or ∼120–160 mas. In 2012 October, the image quality was about a factor of two worse as reflected in the poorer astrometric solution described below. Sources were extracted using the IDL APER routine with an aperture of 5 pixel radius. To calibrate these data photometrically we used observations of the WISE 1828+2650 field obtained with the Wide Field Infrared Camera (WIRC; Wilson et al. 2003) at the Palomar 5 m telescope with a MKO-NIR H filter (Tokunaga et al. 2002). The 87 field of this camera was sufficient to detect ≈350 2MASS stars surrounding WISE 1828+2650 as well as six stars seen in the Keck field. While the Palomar image was not deep enough to detect WISE 1828+2650 itself, we were able to calibrate the Keck field using the calibrated WIRC magnitudes of the six stars in common. Combined with the 2MASS calibration uncertainties (Carpenter 2001), the overall calibration uncertainty is ∼5% which is small compared to the uncertainty in the average magnitude for the WISE target itself (∼9%). There is no evidence for variability between the different epochs. No photometry was attempted for 2012 October.

Zoom In Zoom Out Reset image size

The source positions obtained from the Keck images were corrected for the effects of differential refraction relative to the center of the field using meteorological conditions available at the Canada–France–Hawaii telescope weather archive¹⁰ to determine the index of refraction corrected for wavelength, local temperature, atmospheric pressure and relative humidity (Lang 1983) and standard formulae (Stone 1996). As discussed in the Appendix, for the small field of view of the NIRC2 images and the relatively low airmasses under consideration here (<1.5), the first order differential corrections we do apply to the source positions are small, <10 mas across the ±20'' field and proportionately less at smaller separations.

The effects of optical distortion in the wide-field NIRC2 camera were corrected using a distortion map derived by comparing Keck data of the globular cluster M15 taken immediately after the WISE 1828+2650 observations in 2011 and 2012 July to HST/WFC3 observations of M15 (Anderson et al. 2008; Yelda et al. 2010). Details of this distortion mapping are described in the Appendix but the correction amounts to <1 pixel (40 mas) across most of the array and up to 2 pixels at the edges of the array. At the array center where the WISE 1828+2650 observations were made, the maximum uncorrected distortion for separations up to 5'' is <20 mas. After our correction procedure the residual distortion errors are less than 10 mas over the entire field. As a test of our ability to co-register data from individual Keck images, we examined the positional dispersions of 15 sources in each apparition with respect to their average positions. For the first three Keck epochs the average dispersion in right ascension and declination is ±6 mas. Due to poor conditions, the last epoch was not considered in this analysis. The ability to register these images to an external reference, i.e., the sources in the HST image, is discussed below (Section 3).

The Spitzer Space Telescope observed WISE 1828+2650 using Director's Discretionary Time and at three subsequent general observer epochs. The IRAC camera (Fazio et al. 2004) was used in its full array mode to make observations at 3.5 (Channel-1) and 4.5 μm (Channel-2) in the first two epochs and only at 4.5 μm in the subsequent two epochs. We analyzed post-BCD mosaics from the Spitzer Science Center (SSC) to make photometric and astrometric measurements, extracting sources using a 3 pixel radius aperture and normalized using SSC-recommended aperture corrections.¹¹ WISE 1828+2650 was detected at both 3.5 μm (S/N∼10) and at 4.5 μm (S/N > 40; Table 2). Source A is seen as a weak but definite detection with about one-sixth the brightness of WISE 1828+2650 (S/N = 9). We fitted the pixel data for these two objects simultaneously using Gaussian profiles to determine accurately the source centroids. The fitting uncertainties for the two objects in the four Spitzer observations are 10–20 mas for WISE 1828+2650 and 120–130 mas for Source A (Table 6). There was no evidence for variability in the brightness of WISE 1820+2650 between the four epochs at the <6% level. Since Source A appears extended at shorter wavelengths, there is a concern that position of Source A might shift with wavelength. There is no evidence for this effect in our data. The Spitzer position for Source A is self-consistent within 15 mas over four epochs and consistent with the Keck and HST positions to within (Δα, Δδ) = (60,10) mas which is small relative to the fitting uncertainties for Source A of 120 mas.

The IRAC camera has significant optical distortions over its entire field that are corrected in the SSC pipeline to an average residual value of 100 mas over the whole field.¹² We have used ∼20 HST stars observed within ±90'' of WISE 1828+2650 as probes of the residual distortion as described above for the Keck observations. As a test of our ability to co-register data from individual Spitzer images, we examined the positional dispersions of 22 sources in each apparition with respect to their average positions. The average dispersion in the four Spitzer epochs is ±60 mas in right ascension and declination. The ability to register these images to an external reference, i.e., the sources in the HST image, is discussed below (Section 3). These values are less than the quoted 100 mas uncertainty from the SSC, in part because we are confining our observations to the central part of the IRAC array. The total uncertainties in Table 5 reflect the systematic uncertainties combined with a brightness-dependent term for WISE 1828. In the case where we consider only offsets from WISE 1828+2650 to Source A, the distortion effects will be considerably smaller given the ∼10 pixel distance between the two objects. From examination of the polynomial expression to the distortion correction given by the SSC, we estimate the distortion component of the uncertainty in the separation between the two sources to be <10 mas, negligible compared to the uncertainties due to the source fitting (Table 6).

Table 5 lists the positions of WISE 1828+2650 on the HST/2MASS astrometric grid while Table 6 lists the offsets of WISE 1828+2650 relative to source A. Both sets of right ascension and declination data were fitted to a model incorporating proper motion and parallax (Smart 1977; Green 1985):

$$\begin{eqnarray*} \alpha ^\prime &\equiv & \alpha _0+\mu _\alpha (t-T_0)/\cos (\delta ^\prime)\\ \delta ^\prime &\equiv & \delta _0+\mu _\delta (t-T_0)\qquad \hspace{124.5pt} \mathrm{(1)}\\ \alpha (t)&=& \alpha ^\prime +\pi (X(t) \sin \,\alpha ^\prime - Y(t) \cos \,\alpha ^\prime )/\cos \,\delta ^\prime \\ \delta (t)&=& \delta ^\prime +\pi (X(t) \cos \,\alpha ^\prime \sin \,\delta ^\prime + Y(t) \sin \,\alpha ^\prime \sin \,\delta ^\prime\\ && -Z(t) \cos \,\delta ^\prime ),\qquad \hspace{122pt} \mathrm{(2)} \\ \end{eqnarray*}$$

where (α₀, δ₀) are the source position for equinox and epoch T₀ = J2000.0, μ_{α, δ} are proper motion in the two coordinates in '' yr⁻¹, and π is the annual parallax. The coefficients X(t), Y(t), and Z(t) are the rectangular coordinates of a terrestrial or Earth-orbiting telescope (WISE, Keck, HST) or Spitzer in its Earth-trailing orbit as seen from the Sun. Values of X, Y, Z for the terrestrial or Earth-orbiting observatories are taken from the IDL ASTRO routine XYZ. Spitzer values of X, Y, Z are obtained from the image headers provided by the SSC. Equations (1) and (2) are solved simultaneously using the Mathematica routine NonLinearModelFit incorporating appropriate uncertainties for each data point. Figure 3 shows the fit of the data to this model. Figure 4 shows the fit to parallactic ellipse with the effects of proper motion removed.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We developed two solutions, one using best estimates of the time-varying positions of WISE 1828+2650 in the HST/2MASS astrometric frame (Method-1; Table 5) and the other using offsets between the source and the nearby object denoted Source A (Method-2; Table 6). The Method-2 position offsets use only the rotation angle and plate scale derived from the full astrometric solution, and should thus be relatively immune to residual distortion and refraction effects that apply across the larger field. On the other hand, the only Method-2 observable is the offset which is compromised by Source A being both slightly extended in all the images and blended with WISE 1828+2650 in the Spitzer images.

The two solutions give consistent results (Table 7), yielding parallaxes of 103 ± 16 mas and 79 ± 12 mas and corresponding to distances of 9.7^+1.6_−1.1 pc (Method-1) and 12.6^+2.2_−1.9 pc (Method-2). To generate our final kinematic estimates, we combined and fitted simultaneously the Method-1 and Method-2 data sets, Method-(1+2). While the input imaging data are the same in both cases, the sources of uncertainty are different and, furthermore, the blending in the Method-2 offsets lowers the significance of the Spitzer results and thereby halves the number of epochs with precise measurements. Figure 5 shows histograms for the parallax and distance solutions from Method-(1+2) resulting from a 10,000-iteration Monte Carlo calculation wherein the source positions were varied according to their Gaussian uncertainties and solutions recorded at each step. While the parallax distribution is consistent with the nominal average value and its associated uncertainties, the distance distribution shows a tail to greater distances expected due to its inverse relationship to the parallax. We derived the asymmetric 1σ distance uncertainties from the Cumulative Distribution Function of the errors (Figure 5(b)). Our best estimate of the distance to WISE 1828+2650 is 11.2^+1.3_−1.0 pc with a χ² = 36.3 with 31 degrees of freedom. A model of proper motion plus a non-zero parallax provides a much better fit to the data than a proper motion-only model which yields χ² = 135 with 32 degrees of freedom for the Method-(1+2) data set. On the other hand, there is no evidence that a more complex model, e.g., one including the presence of a perturbing second body in the system, is required.

Zoom In Zoom Out Reset image size

Table 7. Parallax and Proper Motion

Parameter	Method-1a	Method-2b	Method-(1+2)b
μα (''yr−1)	0.969 ± 0.017	0.936 ± 0.015	0.954 ± 0.011
μδ (''yr−1)	0.157 ± 0.019	0.152 ± 0.017	0.153 ± 0.0125
Parallax, π ('')	0.103 ± 0.016	0.079 ± 0.012	0.090 ± 0.0095
Distance (pc)	9.7+1.6−1.1	12.6+2.2−1.9	11.2+1.3−1.0
αJ2000.0c	30344 ± 0077	30375 ± 0012	30359 ± 0009
δJ2000.0c	36120 ± 018	36200 ± 021	36278 ± 0147
χ2 (with no. of degrees of freedom)	18.6  (13)	12.7  (13)	36.3 (31)
Adopted distanced (pc)	11.2+1.3−1.0
Tangential velocityd, vtan (km s−1)	45+15−10

Notes. ^aBased on "absolute coordinates"; ^bbased on offsets from Source A; ^cwith respect to 18^h28^m and 26°50'. The random uncertainty does not include ∼80 mas uncertainty in absolute 2MASS reference frame; ^dfrom the combination of Method-1 and Method-2.

Download table as:  ASCIITypeset image

Finally, we note that the parallax determination here is a relative one with respect to nearby references stars (not galaxies, except for Source A). Thus the motions of the reference frame can introduce a bias into the parallax determination. Since this bias is at the level of 0.5–2 mas (Dupuy & Liu 2012), we can neglect this correction relative to the 100 mas signal of WISE 1828+2650.

WISE 1828+2650 is one of the coolest (latest) Y dwarfs yet identified. The photometry of WISE 1828+2650 covers the 1–5 μm range and can compared against the predictions of brown dwarf atmospheric and evolutionary models (Baraffe et al. 2003). Figure 6 shows a number of models fitted to the observations as described below. Unfortunately, none of these models provide a good fit to the full range of the data. The COND models pass through either the two short wavelength points or the 3–5 μm points, but not both. The over- or underestimates are significant, up to factors of five, which suggests that the models have omitted one or more important physical processes in the atmosphere, e.g., non-equilibrium chemistry or new absorbers, that dramatically affect the SED. Marley et al. (2010) highlight the importance of clouds in the L/T dwarf transition which greatly change the near-IR (J, H, and K bands) appearance of these warmer objects. The onset of still lower temperature condensates, e.g., Na₂S producing sulfide clouds (Morley et al. 2012), may play a role in the appearance of the Y dwarfs (see however Burrows et al. 2003).

Zoom In Zoom Out Reset image size

While acknowledging the failings of the models in this uncharted range of effective temperature, we will use them to make preliminary estimates of the properties of WISE 1828+2650. We take advantage of our knowledge of the distance to the source and use absolute magnitudes rather than relative fluxes to constrain further the evolutionary status of the object.

Magnitudes corresponding to the dust-free COND models (Baraffe et al. 2003) were calculated using bandpasses matched to HST, WISE, Spitzer, and ground-based filters. Models were first selected to match either the absolute H magnitude M_H = 22.21^+0.25_−0.22 mag or M_[4.5] = 14.42^+0.24_−0.20 mag where the uncertainty in the absolute magnitude is dominated by the distance uncertainty. Selecting models matching either absolute magnitude led to two classes of object. Matching the H-band absolute magnitude leads to atmospheres with a narrow range of effective temperatures (T_eff = 275 ± 40 K) but with a broad variation of ages and masses from (0.5 M_Jup, 0.05 Gyr) to (8 M_Jup, 10 Gyr). Matching the [4.5 μm] absolute magnitude leads to hotter, more massive objects (T_eff = 450 ± 40 K) with a range of ages and masses from (2 M_Jup, 0.1 Gyr) to (30 M_Jup, 10 Gyr). Matching the absolute magnitudes in F140W, H, and the two Spitzer bands simultaneously, each with an uncertainty of ∼0.25 mag, yields a best fitting model with (0.1 M_Jup, 0.1 Gyr), T_eff = 285 K, and a χ² = 95 for 2 degrees of freedom (four data points and two parameters). The variation of the quality of fit as a function of mass and age (Figure 7) shows a minimum for a young, low-mass object, but a broad valley tending to older, more massive objects. We emphasize the poor quality of the model fits by pointing out that the contour values in Figure 7 show the logarithm of the χ² value.

Zoom In Zoom Out Reset image size

Finally, we examined the possibility that the SED could be the result of a combination of two coeval bodies in a binary system (Leggett et al. 2013). Fitting the four absolute magnitudes yields a pair of equal-mass objects with (0.5 M_Jup, 0.05 Gyr), T_eff = 285 K, and χ² = 55. However, since none of these models provides a particularly good fit to the data across the entire 1–5 μm range, the estimated physical parameters should be taken with a grain of salt. What all the solutions have in common is a low T_eff ∼ 250–450 K and a mass range of 0.5–20 M_Jup depending on the unknown age of the system (Section 5.2).

Other models lead to a similar range of predicted properties for WISE 1828+2650 depending on the wavelength used for the comparison and the assumed mass and/or age. The Marley et al. (2010) and Burrows et al. (2004) models also suggest an effective temperature ∼300–400 K which can be associated with a broad range of masses and ages with no single model providing a good simultaneous fit to all of the 1–5 μm observations. Spiegel & Burrows (2012) calculate models for both "hot start" and "cold start" formation scenarios (Marley et al. 2007) for ages less than 0.1 Gyr. Matching our data to the predicted absolute H-band models for either the "cold start" or "hot start" cases yields 1 M_Jup objects with ages of 0.02–0.03 Gyr. Matching the predicted M-band models (only an approximation to our Spitzer filter) yields objects in the range (1–5 M_Jup, 0.007–0.1 Gyr). Differences between the COND models and the Spiegel & Burrows models are relatively minor in the area of age overlap.

The nature of WISE 1828+2650 is ambiguous for two reasons. First, the failure of the models to reproduce successfully the SED and absolute magnitudes makes it hard to draw any quantitative conclusions. Second, even with improved models, e.g., ones including clouds or new low-temperature absorbers, there would remain a family of models with roughly similar T_eff that are degenerate in mass and age.

WISE 1828+2650 is similar in appearance to the brown dwarf companion to the ∼2 Gyr old white dwarf WD 0806-661 (Luhman et al. 2012) located 19.2 pc away. The twofold greater distance to this system has so far prevented a detection of WD 0806-661 B in the 1–2 μm region. As a result, we do not yet have a comparable SED, but the near- to mid-IR colors are comparably red with J − [4.5] = 9.25 ± 0.35 (Kirkpatrick et al. 2012) for the WISE object and J − [4.5] > 7 for WD 0806-661 B. The 3–5 μm colors are also similar, [3.6] − [4.5] = 2.77  ±  0.15 mag for WD 0806-661 B and [3.6] − [4.5] = 2.81  ±  0.06 mag for WISE 1828+2650, although neither of these objects are as red as other Y dwarfs. With an age estimate for the white dwarf it is possible to narrow the mass range for its companion based on its colors. Luhman et al. (2012) suggest a mass of 6–9 M_Jup which is consistent with our estimates for WISE 1828+2650 in that age range.

An important peculiarity of WISE 1828+2650 is its bright absolute magnitude, M_W2 = 14.15^+0.25_−0.21 mag for its presumed Y2 spectral type. Compared with WD 0806-661 B with an absolute magnitude of M_[4.5] = 15.5 mag, WISE 1828+2650 is >1 mag more luminous. More troubling, however, is the fact that compared with other Y dwarfs with reasonably secure parallaxes, WISE 1828+2650 is >2 mag brighter than other Y dwarfs in M_W2 and, to a lesser extent, brighter than some of the earlier Y dwarfs in M_H. Figure 8, which illustrates this problem, is an updated version of the absolute magnitude versus spectral type figures appearing in Kirkpatrick et al. (2012). This new figure uses recently published Y dwarf parallaxes by Marsh et al. (2013) and Tinney et al. (2012). The locations of WISE 1828+2650 and WD 0806-661 B are also shown. The spectral type of WD 0806-661 B has been estimated from its effective temperature since no spectrum is available and only an upper limit is available for the H-band magnitude. However, in both cases it is apparent that the two objects are similarly deviant from the trend set by earlier Y objects.

Zoom In Zoom Out Reset image size

A simple explanation of at least part of the ∼1.5 mag difference in absolute M_[4.5] between WISE 1828+2650 and WD 0806-661 B is that WISE 1828+2650 is an unresolved, equal mass binary which would increase the total system brightness by 0.7 mag, close to the difference with WD 0806-661B. Any such companion to WISE 1828+2650 would have to have an orbit <0.5 AU (<005) to be consistent with the HST and Keck data and if present would eventually manifest itself in either the astrometric data or future spectroscopic data. No such evidence for binarity yet exists. In any event, such an explanation does not account for the significant deviation from the trends established for earlier spectral types. Some other explanation must be sought for the differences in absolute magnitude at 3–4 μm.

We cannot eliminate solutions corresponding to young, low-mass objects for WISE 1828+2650. Of the approximately 250 low-mass objects (M, L, and T spectral types) within 10 pc of the Sun there are at least three M stars younger than ∼100 Myr: AU Mic (9.9 pc; Kalas et al. 2004), GJ 896 B (6.3 pc; Riedel et al. 2011), and most recently AP Col (4.6 pc; Henry et al. 2006; Riedel et al. 2011). A number of brown dwarfs have been claimed to be young based on near-IR spectra which indicate low surface gravity. Examples include the L dwarf 2MASS J01415823−4633574 (Kirkpatrick et al. 2006; Patience et al. 2012) and the nearby, 7.5 pc, L5 brown dwarf 2MASS J035523+113337 (Faherty et al. 2013). If these 2MASS objects are indeed that young, then their masses might be as low as 10 M_Jup. As noted above, if WISE 1828+2650 were as young as 0.1 Gyr, its mass would be only 1–2 M_Jup. A young, brown dwarf of this age is certainly a possibility, albeit one with a low probability (∼1%) based on the statistics of nearby low-mass objects, i.e., three young M stars plus one young L dwarf out of 250 low-mass objects.

The simplest metric for assessing the age of WISE 1829+2650 is the directly measured tangential velocity which comes from the proper motion and distance: v_tan = 4.74 μ/Π = 51 ± 5 km s⁻¹, where μ is the total proper motion and Π the parallax (Smart 1977). This value falls in the middle of the distribution of tangential velocities measured for nearby (<20 pc) L and T brown dwarfs (see Figure 7, Faherty et al. 2009). These velocities are consistent with models of disk population with ages of 3–8 Gyr. This range narrows to 2–4 Gyr if the highest velocity objects (v_tan > 100 km s⁻¹ are excluded. Within the age range and using fits to the COND models described above, we estimate that the mass of WISE 1828+2650 is ∼3–6 M_Jup.

A number of authors have used kinematic properties to associate brown dwarfs with nearby young clusters, but the results to date are only suggestive (Faherty et al. 2013). For example, might WISE 1828+2650 be a young object that could be associated with one of the nearby young moving groups, e.g., AB Dor, β Pic, or Tucana/Horlorgium (Zuckerman & Song 2004; Torres et al. 2008) or the Argus/IC 2391 group as in the case of AP Col? These groups are located at distances of 20–100 pc with ages ranging from 8 to 50 Myr. We evaluated possible galactic velocity vectors for WISE 1828+2650, (U, V, W), and their associated uncertainties σ_{U, V, W}, from its position, parallax, and proper motion vector determined here for a range of the unknown V_rad from −100 to 100 km s⁻¹. We calculated the difference between cluster (U_cl, V_cl, W_cl) values and possible values for WISE 1828 as a function of V_rad:

$$\begin{eqnarray*} && \hbox{$ \Delta {\it UVW}(V_{{\rm rad}})$} = \\ \ms && \quad \hbox{$ \sqrt{(U_{{\rm WISE}}(V_{{\rm rad}})-U_{{\rm cl}})^2+(V_{{\rm WISE}}(V_{{\rm rad}})-V_{{\rm cl}})^2 +(V_{{\rm WISE}}(V_{{\rm rad}})-V_{\rm cl})^2}.$}\\ && \qquad \hspace{214pt} \mathrm{(3)} \end{eqnarray*}$$

For V_rad between −100 and 100 km s⁻¹ there is no good match in (U, V, W) space for any of the six moving groups with ages from 8 to 50 Myr (Table 8; Zuckerman et al. 2004; Riedel et al. 2011). The smallest separation is ΔUVW ∼ 50 km s⁻¹. While it is impossible to rule out the possibility that WISE 1828+2650 is a young straggler ejected from one of these groups, it is more likely that WISE 1828+2650 is an older, higher mass object.

The suggestion that a population of free-floating planets may be responsible for short-duration microlensing events, $t\sim \sqrt{M/M_{{\rm Jup}}}$ days ∼0.3–3 days for M = 0.1–10 M_Jup (Sumi et al. 2011), immediately leads to the conjecture that objects like WISE 1828+2650 might be members of such a population. Sumi et al. (2011) calculate that fitting the distribution of short-duration microlensing events requires a population of ∼1–2 M_Jup objects roughly 1.3–1.9× as numerous as all main-sequence stars (0.08–1 M_☉). Quanz et al. (2012) argue that imaging surveys may be starting to reveal a significant number of bound planets located at 10 s of AU from low-mass stars which would decrease the need for a large population of free-floating, planetary mass objects.

The RECONS database (Henry et al. 2006) lists 300 main-sequence stars (M-G) within 10 pc, or 0.072 object pc⁻³. Estimates of the entire population of Y dwarfs down to the coolest objects are premature but we note that WISE 1828+2650's distance implies a minimum local density of 1.7 ± 0.5 × 10⁻⁴ pc⁻³. For WISE 1828+2650 to represent the tip of the microlensing iceberg, the population of similar objects would have to rise very sharply below the few M_Jup mass of this object. Between 0.1 and 10 M_Jup where the Sumi et al. (2011) data are sensitive (Figure 9 of Sumi et al. 2011), the mass distribution function, dN/DM∝M^−α, would have to be quite steep. Estimating the mass of WISE 1828+2650 using a uniform distribution between 3 and 6 M_Jup yields α = 2.2 ± 0.2 to produce enough extremely low mass dwarfs (0.1–10 M_Jup) equal in number to the main-sequence stars within 10 pc. This is dramatically steeper than the flat brown dwarf index, α ∼ 0 (Figure 14 in Kirkpatrick et al. 2012), and steeper than the mass distribution of planets determined by radial velocity studies, α = 1.5 (Cumming et al. 2008; Howard et al. 2010).

To date, 14 "free-floating" Y dwarfs have been identified (Cushing et al. 2011; Kirkpatrick et al. 2012; Tinney et al. 2012). Several other WISE-selected Y dwarf candidates, such as those presented in Griffith et al. (2012), remain but lack spectroscopic confirmation. All are on active parallax and spectroscopic classification campaigns, so it will soon be known if other WISE 1828+2650 and WD 806-661 B analogs exist. Even if, for example, the local density of objects as cool as WISE 1828+2650 eventually proves to be five times higher than estimated solely on the basis of this one object, then the power-law index required to account for the microlensing results would still be >1.5. In the absence of convincing evidence for an upturn in the mass distribution, the most conservative interpretation is that WISE 1828+2650 sits at the low-mass end of the brown dwarf distribution rather than at the start of a much larger population of free-floating "planets."

A less conservative interpretation of these results is also possible. As Luhman et al. (2011) themselves suggest WD 0806-661 B could have formed as a ∼6–9 M_Jup object in a protoplanetary disk around its ∼2 M_☉ primary, WD 0806-661 A. Only later, after the primary shed mass during its evolution off the main sequence, would WD 0806-661 B have migrated into a larger orbit, currently of radius ∼2500 AU. The formation mechanism between the WISE Y0 and Y1 discoveries, which presumably formed via cloud fragmentation as do normal stars, and WB 0806-661 B would thus be different. Compositional differences should be seen between the two types of objects, given that chemical differentiation is expected in a protoplanetary disk and not in a star formation cloud. Jupiter's enhanced metallicity is an obvious example of this difference (Wong et al. 2004; Alibert et al. 2005). Objects formed via core accretion may also have a pronounced core of heavy elements as inferred for transiting planets (Rogers & Seager 2010) which might manifest itself in significant ways. Such differences between these two classes could possibly account for the discrepancies noted in the absolute magnitude versus spectral type diagrams of Figure 8.

Taking this one step further, the similarity between WD 0806-661 B and WISE 1828+2650 may be a direct reflection of the fact that both objects were birthed via the same process. In this case, WISE 1828+2650 would be a "free-floating" planet that, given our mass estimate of ∼3–6 M_Jup, would fall at the high-mass end of the rogue planet population proposed by Sumi et al. (2011). In this case, the poor agreement with theoretical spectra may be a consequence of a poor match between the true elemental abundances or interior structure of WISE 1828+2650 and those in the model assumptions. If the space density of truly "free-floating" planets measured by Sumi et al. (2011) is correct, then there should be colder examples (<300 K) at much smaller distances (≪10 pc) than WISE 1828+2650 itself. Some of these may yet be discoverable at lower S/Ns in the WISE data.