The Second Discovery from the COCONUTS Program: A Cold Wide-orbit Exoplanet around a Young Field M Dwarf at 10.9 pc

Direct imaging of exoplanets provides a valuable window into the atmospheres, formation, and evolution of gas-giant planets (e.g., Bowler 2016). In addition to photometry and spectroscopy of the exoplanets themselves, characterization of the host stars is also critical in such investigations, as system properties such as birth environment, stellar insolation, and host mass, age, and metallicity are central to understanding the past, present, and future of the exoplanets. Exoplanetary systems with wide-separation planets are particularly useful, given the relative ease with which both the host star and the exoplanets can be characterized, in comparison to smaller-separation exoplanets where the host's starlight can impede direct observations. As part of our ongoing COol Companions ON Ultrawide orbiTS (COCONUTS) program to find wide separations planetary-mass and substellar companions to stars (e.g., Zhang et al. 2020a), we have established the physical association of the active M dwarf L 34-26 (PM J07492−7642, TYC 9381-1809-1; hereafter COCONUTS-2A) with the T9 dwarf WISEPA J075108.79−763449.6 (hereafter COCONUTS-2b). This Letter presents characterization of this system, with astrometric, spectrophotometric, and physical properties summarized in Table 1.

Table 1. Properties of COCONUTS-2

		COCONUTS-2A			COCONUTS-2b
Properties		Value	Ref.		Value	Ref.
Spectral Type		M3	Torr06		T9	Kirk11
Age (Myr)		150−800	This Work		⋯	⋯
[Fe/H] (dex)		0.00 ± 0.08	Hojj19		⋯	⋯
Astrometry and Kinematics
R.A., Decl. (epoch J2000; hms, dms)		07:49:12.68, −76:42:06.72	Gaia16,20		07:51:08.81, −76:34:49.43	Kirk19
μα cosδ, μδ (mas/yr)		−102.15 ± 0.02, −192.92 ± 0.02	Gaia16,20		−104.8 ± 2.8, −189.7 ± 4.5	Kirk19
Parallax (mas)		91.826 ± 0.019	Gaia16,20		97.9 ± 6.7	Kirk19
Tangential Velocity (km/s)		11.276 ± 0.002	This Work		10.5 ± 0.7	This Work
Radial Velocity (km/s)		1.19 ± 0.61	Schn19		⋯	⋯
Position Angle (east of north; deg)		⋯	⋯		42.9	This Work
Projected Separation		⋯	⋯		594'' (6471au)	This Work
Spectrophotometric Properties
V (mag)		11.276 ± 0.065	Kira12		⋯	⋯
Gaia DR2 G (mag)		10.146 ± 0.002	Gaia16,18		⋯	⋯
Gaia DR2 BP (mag)		11.574 ± 0.005	Gaia16,18		⋯	⋯
Gaia DR2 RP (mag)		9.009 ± 0.003	Gaia16,18		⋯	⋯
2MASS J (mag)		7.406 ± 0.021	Cutr03		⋯	⋯
2MASS H (mag)		6.860 ± 0.030	Cutr03		⋯	⋯
2MASS K (mag)		6.579 ± 0.018	Cutr03		⋯	⋯
MKO Y (mag)		⋯	⋯		20.020 ± 0.100	Legg15
MKO J (mag)		⋯	⋯		19.342 ± 0.048	Kirk11
MKO H (mag)		⋯	⋯		19.680 ± 0.130	Legg15
MKO K (mag)		⋯	⋯		20.030 ± 0.200	Legg15
W1 (mag)		6.490 ± 0.081 a	Cutr14		16.946 ± 0.066 a	Cutr14
W2 (mag)		6.244 ± 0.027	Cutr14		14.530 ± 0.036 a	Cutr14
Spitzer/IRAC [3.6] (mag)		⋯	⋯		14.416 ± 0.036	Kirk11
Spitzer/IRAC [4.5] (mag)		⋯	⋯		14.620 ± 0.020	Kirk11
EWLi λ6708 (Å)		<0.05	Schn19		⋯	⋯
EWHα (Å)		2.3–8.0	⋆ b		⋯	⋯
Physical Properties
$\mathrm{log}({L}_{H\alpha }/{L}_{\mathrm{bol}})$ (dex)		−3.74 ± 0.13	This Work		⋯	⋯
$\mathrm{log}({L}_{{\rm{X}}})$ (dex)		28.92 ± 0.02	Schn19		⋯	⋯
Prot (days)		2.83 ± 0.28	This Work		⋯	⋯
$\mathrm{log}({L}_{\mathrm{bol}}/{L}_{\odot })$ (dex)		−1.732 ± 0.011	This Work		−6.384 ± 0.028	This Work
Teff (K)		3406 ± 69	Gaid14		hot-start: 434 ± 9 (ATMO), 429 ± 9 (Sonora-Bobcat)	This Work
					cold-start: 431 ± 9 (Sonora-Bobcat)	This Work
log(g) (dex)		4.83 ± 0.03	This Work		hot-start: ${4.11}_{-0.18}^{+0.11}$ (ATMO), ${4.08}_{-0.18}^{+0.12}$ (Sonora-Bobcat)	This Work
					cold-start: ${4.11}_{-0.19}^{+0.12}$ (Sonora-Bobcat)	This Work
R (A: R⊙; b: RJup)		0.388 ± 0.011	This Work		hot-start: 1.11 ± 0.03 (ATMO), ${1.13}_{-0.03}^{+0.04}$ (Sonora-Bobcat)	This Work
					cold-start: ${1.12}_{-0.03}^{+0.04}$ (Sonora-Bobcat)	This Work
M (A: M⊙; b: MJup)		0.37 ± 0.02	This Work		hot-start: ${6.3}_{-1.9}^{+1.5}$ (ATMO), ${6.1}_{-1.8}^{+1.5}$ (Sonora-Bobcat)	This Work
					cold-start: ${6.5}_{-2.0}^{+1.7}$ (Sonora-Bobcat), ${11.6}_{-2.8}^{+1.9}$ (Spie12)	This Work

Notes.

^a AllWISE photometry might be contaminated by the scattered light halo from a nearby bright source. ^b EW_Hα was measured as 2.3–8.0 Å in literature (e.g., Torres et al. 2006; Gaidos et al. 2014; Schneider et al. 2019) with a median of 4.1 Å.

References—Bail21: Bailer-Jones et al. (2021), Cutr03: Cutri et al. (2003), Cutr14: Cutri (2014), Gaia16: Gaia Collaboration et al. (2016), Gaia18: Gaia Collaboration et al. (2018), Gaia20: Gaia Collaboration et al. (2020), Gaid14: Gaidos et al. (2014), Hojj19: Hojjatpanah et al. (2019), Kirk11: Kirkpatrick et al. (2011), Kira12: Kiraga (2012), Kirk19: Kirkpatrick et al. (2019), Legg15: Leggett et al. (2015), Spie12: Spiegel & Burrows (2012), Schn19: Schneider et al. (2019), Torr06: Torres et al. (2006).

Download table as:  ASCIITypeset image

As part of the COCONUTS program, we examined astrometry of 440 free-floating T5−Y1 dwarfs in the UltracoolSheet ⁵ to search for their co-moving primary stars from Gaia Early Data Release 3 (EDR3; Gaia Collaboration et al. 2016, 2020). We identify co-moving pairs if: (1) the projected separation between the primary star and the companion is <10⁴ au; (2) the difference between the two components' parallaxes are within 30% of the primary star's parallax; and (3) the vector difference between the two components' proper motions are within 30% of the primary star's total proper motion (e.g., Dupuy & Liu 2012). The COCONUTS-2 system was identified in this process.

COCONUTS-2A has a parallax = 91.826 ± 0.019 mas (=10.888 ± 0.002 pc; Bailer-Jones et al. 2021) and a proper motion of (μ_α cosδ, μ_δ ) = (−102.15 ± 0.02, −192.92 ± 0.02) mas yr⁻¹ from Gaia EDR3, measured from 25 visibility periods with a renormalized unit weight error of 1.07. COCONUTS-2b has a parallax = 97.9 ± 6.7 mas ( = 10.2 ± 0.7 pc) and a proper motion of $({\mu }_{\alpha }\cos \delta ,{\mu }_{\delta })=(-104.8\pm 2.8,-189.7\pm 4.5)$ mas yr⁻¹ as measured by Kirkpatrick et al. 2019 using 18 epochs of ground-based astrometry spanning 5.3 yr. The physical association between the A and b components is clearly indicated by their common distances and proper motions (Figure 1), with a consistency of 0.98σ and 1.07σ, respectively. The angular separation between these two components is 594'', corresponding to a projected separation of 6471 au at the primary star's distance.

Zoom In Zoom Out Reset image size

To estimate the chance alignment probability of the COCONUTS-2 system, we simulate the kinematics of field ultracool dwarfs and identify synthetic objects that would be identified as COCONUTS-2A's companions by our aforementioned criteria. We generate 10⁷ ensembles of ultracool dwarfs for this test. In each ensemble, we assume objects are uniformly distributed within a 20 pc-radius sphere centered on Earth with a space density of 1.33 × 10⁻² pc⁻³ (measured for T5−Y1 dwarfs by Kirkpatrick et al. 2021), leading to random Galactic XYZ positions for a total of 446 synthetic objects. We generate random UVW space motions of these objects by assuming 3D normal distributions with the mean and standard deviation of U = 2.7 ± 31.6 km s⁻¹, V = −5.1 ± 27.0 km s⁻¹, W = 1.1 ± 16.5 km s⁻¹ as determined by Burgasser et al. (2015) for M and L dwarfs within 20 pc. We then convert these synthetic XYZUVW into equatorial coordinates, proper motions, parallaxes, and radial velocities, finding that 24 synthetic objects would be identified as a companion of COCONUTS-2A among all 10⁷ ensembles. Given that our full search examined 440 T5−Y1 dwarfs, the probability that our search would result in a single chance alignment "discovery" like COCONUTS-2b is thus 24 × 10⁻⁷ × 440 = 10⁻³.

3.1. Physical Properties

3.2. Age

We estimate the star's age using spectroscopic, activity, rotation, kinematic, and photometric properties.

Lithium. Li λ6708 absorption is absent in high-resolution spectra (R ∼ 35000–50000) with an upper limit on the equivalent width (EW) <50 mÅ (Schneider et al. 2019). Assuming 99% of the object's initial Li abundance is depleted, the Chabrier & Baraffe (1997) models predict COCONUTS-2A is likely >30 Myr, given its M3 spectral type and K_S -band absolute magnitude.

Alkali Lines. The Na I (λ8183/8195) and K I (λ7665/7699) doublets are sensitive to surface gravity and thus a possible youth indicator for M dwarfs. Riedel et al. (2014) measured an Na I index = 1.18 and EW_K7699 = 1.0 Å for COCONUTS-2A, which suggest an age of ≳100 Myr given the star's V − K_S = 4.70 ± 0.07 mag, though Reidel et al. advised caution in the interpretation of these lines at V − K_S < 5.

Hα Emission. COCONUTS-2A exhibits strong Hα emission, with EW_Hα = 2.3–8.0 Å from the literature (Table 1). Adopting EW_Hα = 4.1 ± 1.2 Å based on the median and 16th to 84th percentiles, we compute ${\mathrm{log}}_{10}({L}_{{\rm{H}}\alpha }/{L}_{\mathrm{bol}})\,={\mathrm{log}}_{10}({\mathrm{EW}}_{{\rm{H}}\alpha }\times \chi )\,=-3.74\pm 0.13$ dex, with χ = (4.4872 ± 0.4967) × 10⁻⁵ from Douglas et al. (2014). To estimate the stellar age, we adapt the Kiman et al. (2021) Hα activity–age relation into a Bayesian framework. We use an Markov Chain Monte Carlo (MCMC) process to derive the stellar age (t) posterior using a log-likelihood function:

$$\begin{eqnarray}&&\mathrm{ln}{ \mathcal L }(t)=-\displaystyle \frac{{\left[\mathrm{log}{\left({L}_{{\rm{H}}\alpha }/{L}_{\mathrm{bol}}\right)}_{\mathrm{model}}-\mathrm{log}{\left({L}_{{\rm{H}}\alpha }/{L}_{\mathrm{bol}}\right)}_{\mathrm{obs}}\right]}^{2}}{2\times ({\sigma }_{\mathrm{obs}}^{2}+{\sigma }_{V}^{2})},\end{eqnarray} \tag{ 1 }$$

where $\mathrm{log}{({L}_{{\rm{H}}\alpha }/{L}_{\mathrm{bol}})}_{\mathrm{obs}}=-3.74$ dex and σ_obs = 0.13 dex. We use the best-fit model parameters from Kiman et al. (2021) to compute $\mathrm{log}{({L}_{{\rm{H}}\alpha }/{L}_{\mathrm{bol}})}_{\mathrm{model}}$ as a function of t, with σ_V = 0.22 dex. Assuming a uniform prior of [1.5 Myr, 10 Gyr] in t, we run emcee (Foreman-Mackey et al. 2013) with 20 walkers and 5000 iterations (with the first 100 "burn-in" iterations excluded) and obtain $t={750}_{-500}^{+800}$ Myr. This age is consistent with the dynamical model-based activity lifetime of <2 ± 0.5 Gyr for M3 dwarfs (West et al. 2008).

X-ray and UV Emission. Schneider et al. (2019) measured an X-ray luminosity of L_X = 28.92 ± 0.02 dex using the second ROSAT all-sky survey. This L_X is similar to those of ONC (≈1 Myr), Pleiades (≈112 Myr), and Hyades (≈750 Myr) members (e.g., Preibisch & Feigelson 2005) but much larger than field-age dwarfs. We derive $\mathrm{log}{R}_{{\rm{X}}}=\mathrm{log}({L}_{{\rm{X}}}/{L}_{\mathrm{bol}})=-2.93\pm 0.02$ dex, which falls in the saturation regime of the X-ray activity–age relation. We also compute the ratio between X-ray and Two Micron All-Sky Survey (2MASS) J-band fluxes, $\mathrm{log}({f}_{{\rm{X}}}/{f}_{J})\,=\mathrm{log}({f}_{{\rm{X}}})+0.4J+6.3=-1.97\pm 0.02$ dex, which is consistent with those of Pleiades members and higher than most Hyades members (e.g., Shkolnik et al. 2009). Schneider et al. (2019) also converted COCONUTS-2A's Galaxy Evolution Explorer (GALEX) UV photometry to flux ratios of $\mathrm{log}({f}_{\mathrm{NUV}}/{f}_{J})=-3.94\pm 0.02$ dex and $\mathrm{log}({f}_{\mathrm{FUV}}/{f}_{J})\,=-4.64\pm 0.06$ dex, which we find are higher than field dwarfs with comparable stellar masses but similar to members of the Hyades and younger associations (see Figures 3–4 of Schneider & Shkolnik 2018). The X-ray and UV activity thus suggests an age of ≲750 Myr.

Rotation. We used a Lomb–Scargle periodogram to estimate the star's rotation period, based on TESS data from sectors 3, 4, 6, 7, 10, 11, 12, 13, 27, 30, 33, 34, and 36, spanning a range of 920 days. We queried the two-minute cadence, SPOC PDCSAP light curves from Mikulski Archive for Space Telescopes (MAST) using lightkurve (Lightkurve Collaboration et al. 2018). We found a rotation period of P_rot = 2.8299 ± 0.0025 (rand) ± 0.2830 (sys) days, where the random uncertainty was estimated by fitting a Gaussian to the main periodogram peak using least-squares minimization. We assumed a 10% systematic uncertainty to account for the possibility of surface differential rotation. The peak false alarm probability was vanishingly small, and the only other significant peak was the half-period alias, which we ruled out upon visual inspection of the light curve. The light curve amplitude was 2.2% ± 0.3%, which we obtained by finding the difference between the 95th and 5th percentiles of flux for each sector and computing the median over all sectors. Our TESS-based rotation period is also consistent with the periods of 2.827 and 2.829 days measured using the All Sky Automated Survey (ASAS) and All Sky Automated Survey for SuperNovae (ASAS-SN) photometry by Kiraga (2012) and Jayasinghe et al. (2019), respectively. The primary star's P_rot lines up with members of Pleiades and younger associations (≲60 Myr) and lies at the short-period end of Praesepe members (∼800 Myr; see Rebull et al. 2018). Therefore, we draw a rotation-based age estimate of ≲800 Myr.

Kinematics. Based on Gaia EDR3 astrometry and the radial velocity measured by Schneider et al. (2019), we find that COCONUTS-2A is not associated with any young moving group using BANYAN Σ (Gagné et al. 2018). The low tangential velocity (11.276 ± 0.002 km s⁻¹) of the star is consistent with being a young field object. However, the object's space motion UVW = (+6.4 ± 0.2, +4.5 ± 0.5, −8.2 ± 0.2) km s⁻¹ is outside the Zuckerman & Song (2004) "good box" of young stars, with −15 ≤ U ≤ 0 km s⁻¹, −34 ≤ V ≤ −10 km s⁻¹, and −20 ≤ W ≤ +3 km s⁻¹, suggesting an age of ≳150 Myr.

HR Diagram Position. Pre-main-sequence M dwarfs with comparable masses as COCONUTS-2A are still in the process of contraction over their first ∼100 Myr, and thus should have brighter absolute magnitudes than field-age objects. We find Gaia photometry of COCONUTS-2A lines up with the main sequence, as well as members of AB Doradus (≈149 Myr) and older groups (e.g., Kiman et al. 2021), suggesting an age of ≳150 Myr.

Altogether, we adopt an age estimate of 150–800 Myr for the COCONUTS-2 system.

COCONUTS-2b was initially found as a field dwarf by Kirkpatrick et al. (2011), who assigned a T9 spectral type based on low-signal-to-noise ratio (S/N) near-infrared spectra. Leggett et al. (2017) suggested that this object is likely a subdwarf by comparing its [3.6]—[4.5] and J_MKO − [4.5] colors with trends in the multi-metallicity Tremblin et al. (2015) model atmospheres (though changes in these models' adiabatic indices can also affect these colors and the models do not accurately reproduce these colors for most late-T and Y field dwarfs—see Figure 8 of Leggett et al. 2017). Kirkpatrick et al. (2019) found that this T9 dwarf has fainter absolute magnitudes in H, W1, and [3.6] bands than other field dwarfs with similar [3.6]–[4.5] colors but also noted such peculiarity is not a hallmark of subdwarfs. As shown here, COCONUTS-2b is in fact a bound companion to a young solar-metallicity star, eliminating the possibility of it being a subdwarf.

To infer the physical properties of COCONUTS-2b, we first derive its bolometric luminosity using the Dupuy & Kraus (2013) super-magnitude method as updated by W. Best et al. (2021, in preparation). Using polynomials determined from the Sonora−Bobcat model atmospheres (Marley et al. 2021), we combine and convert the object's absolute magnitudes in J_MKO, H_MKO, [3.6], and [4.5] bands into log(L_bol/L_⊙) = −6.384 ± 0.028 dex. To derive the companion's physical properties (e.g., effective temperature T_eff, surface gravity log(g), radius R, and mass M), we then use our measured bolometric luminosity (assumed to follow a normal distribution) and its host star's age (assumed to follow a uniform distribution within 150–800 Myr) to interpolate both the Sonora−Bobcat (Marley et al. 2021) and the ATMO 2020 (Phillips et al. 2020) evolutionary models. Both these models adopt hot-start initial conditions and assume objects form with high initial entropy without any accretion. For ATMO 2020, we use all three available versions: chemical-equilibrium models (CEQ), and non-equilibrium models (CNEQ) with two levels of eddy mixing (K_zz ; CNEQ weak, and CNEQ strong). All these models are for solar-metallicity, cloud-free atmospheres. The Sonora−Bobcat and ATMO CEQ models assume rainout chemical equilibrium. We find that the physical properties derived from all these evolutionary models are consistent within 1σ, with the ATMO 2020 models giving T_eff = 434 ± 9 K, log(g) = ${4.11}_{-0.18}^{+0.11}$ dex, R = 1.11 ± 0.03 R_Jup, and M = ${6.3}_{-1.9}^{+1.5}$ M_Jup. At such surface gravity, the ATMO CNEQ weak and strong models correspond to $\mathrm{log}{K}_{{zz}}\approx 5$ and 7, respectively (see Figure 1 of Phillips et al. 2020).

For COCONUTS-2b, hot-start initial conditions are vastly more likely than a cold-start via core accretion, as a finely-tuned dynamical kick would be needed to remove it from its starting point in the protoplanetary disk of COCONUTS-2A to its present location but not eject it. Regardless, for a point of comparison, we use the cold-start version of the Sonora−Bobcat models (Marley et al. 2021; also see Section 5.5 of Nielsen et al. 2019) to characterize COCONUTS-2b and find the derived physical properties (e.g., M = ${6.5}_{-2.0}^{+1.7}$ M_Jup) are all consistent with hot-start model predictions within 1σ (Table 1). We also derive COCONUTS-2b's properties from the cold-start models of Spiegel & Burrows (2012) using the same method as T. J. Dupuy et al. (2021, in preparation) did for 51 Eri b. The Spiegel & Burrows (2012) models cover the most tabulated masses (1–15 M_Jup) and ages (1 Myr to 1 Gyr) of any cold-start models, and they also provide a wide range of initial specific entropy (S_init) values. The online Spiegel & Burrows (2012) model data ⁷ include a model spectrum at each grid point, which we numerically integrate to compute L_bol at each mass and age. Using their cloud-free, solar-metallicity evolutionary models, we find $M={11.6}_{-2.8}^{+1.9}$ M_Jup from the coldest-start initial conditions (S_init = 8.0 k_B baryon⁻¹), which is significantly higher than from hot-start Sonora−Bobcat and ATMO 2020 models, as well as the cold-start Sonora models.

Compared to (older) field T9 dwarfs (e.g., Dupuy & Kraus 2013; Kirkpatrick et al. 2021), COCONUTS-2b has 0.5–1.5 mag fainter absolute magnitudes in Y/J/H/K/[4.5] bands, a similar M_[3.6], ≈0.3 mag bluer [3.6]–[4.5] color, and ≈0.5 mag redder J−[4.5] and H−[4.5] colors. Also, COCONUTS-2b has 0.2–0.6 dex fainter L_bol, 100 − 250 K cooler T_eff, and ≈1.0 dex lower log(g) than those of T9 field dwarfs as studied by Dupuy & Kraus (2013) using the Baraffe et al. (2003) evolutionary models with an assumed age of 5 Gyr. These might suggest a surface-gravity dependence of ultracool dwarf properties at T/Y transition, similar to the well-known phenomenon for the L/T transition (e.g., Metchev & Hillenbrand 2006; Liu et al. 2016). All these peculiarities of COCONUTS-2b are similar to those of the T6.5 imaged exoplanet 51 Eri b, but they are not seen for the T8 wide-orbit planetary-mass companion Ross 458C. High-quality spectroscopy and a more accurate spectral type of COCONUTS-2b would help investigate whether its properties are indeed abnormal compared to the field population, given that its current T9 spectral type is based on relatively low-S/N spectra (Kirkpatrick et al. 2011).

To investigate the atmospheric processes of COCONUTS-2b, we compare its infrared photometry with the predicted photometry by hot-start cloudless Sonora−Bobcat and ATMO 2020 models given this planet's bolometric luminosity and age (Figure 2). We also compare with the Morley et al. (2012) equilibrium-chemistry models with optically thin sulfide and salt clouds, which might provide a better match to late-T dwarfs' spectrophotometry than cloudless models (e.g., Zhang et al. 2020b, 2021b). We find the near-infrared photometry of COCONUTS-2b can be best explained by cloudy models with condensate sedimentation efficiencies of f_sed = 2 or 3. All cloudless models predict too blue near-infrared colors, although the ATMO non-equilibrium models perform significantly better than equilibrium models in predicting both ${M}_{{J}_{\mathrm{MKO}}}$ and J_MKO − H_MKO of COCONUTS-2b. When mid-infrared magnitudes are considered, we find the photometry of COCONUTS-2b might be better explained by ATMO non-equilibrium models than cloudy models, with the latter models significantly off the field sequence of T/Y dwarfs (also see discussion in Leggett et al. 2017). COCONUTS-2b might have both clouds and non-equilibrium processes in its atmosphere, which is not surprising given its cold effective temperature. The co-existence of these atmospheric processes have also been suggested for Ross 458C based on spectroscopic analysis (e.g., Zhang et al. 2020b).

Zoom In Zoom Out Reset image size

The COCONUTS-2 system has a mass ratio of ${0.016}_{-0.005}^{+0.004}$ based on the hot-start model-predicted mass of COCONUTS-2b (Figure 3). This ratio, combined with the companion's low mass of ${6.3}_{-1.9}^{+1.5}$ M_Jup, satisfy the IAU working definition of an exoplanet (which suggests an upper limit of ≈0.04 in mass ratio). ⁸ At 10.9 pc, COCONUTS-2b is the nearest imaged exoplanet to Earth to date. Given its projected separation of 6471 au and T_eff ≈ 434 K, COCONUTS-2b is the second widest (after TYC 9486-927-1 b at a projected separation of 6900au; Deacon et al. 2016) and the second coldest (after WD 0806−661B with T_eff ≈ 328 K; Luhman et al. 2011) imaged exoplanet discovered so far.

Zoom In Zoom Out Reset image size

To convert COCONUTS-2b's projected separation into a semimajor axis, we use the scaling factor of ${1.16}_{-0.32}^{+0.81}$ from Dupuy & Liu (2011). We assume this factor follows a distribution composed of two half-Gaussians joined at 1.16, with a standard deviation of 0.81 and 0.32 toward higher and lower values (truncated at 0), respectively, leading to a semimajor axis of ${7506}_{-2060}^{+5205}$ au. Combining this value with the primary star's mass, we derive a very long orbital period of ${1.1}_{-0.4}^{+1.3}$ Myr for COCONUTS-2b. We also compute the binding energy of the COCONUTS-2Ab system as $({4.7}_{-1.9}^{+2.9})\times {10}^{39}$ erg by using COCONUTS-2b's hot-start model-derived mass, which is the lowest among all host systems of imaged exoplanets.

In addition, we note that COCONUTS-2A is not the brightest star in the night sky of COCONUTS-2b, especially given the wide orbital separation and the low luminosity of the primary star. Over time, the brightest star in the sky likely varies as the system drifts relative to bright stars in the solar neighborhood from the perspective of COCONUTS-2b. The star Canopus matches the flux of the host star COCONUTS-2A at 500 nm and exceeds its output for wavelengths bluer than ∼500 nm. A human observer, if present, would likely perceive Canopus as the brightest source in the sky.

Figure 4 compares the bolometric luminosity and age of COCONUTS-2b to previously known L6−Y1 planetary-mass and substellar benchmarks from Zhang et al. (2020a) and Zhang et al. (2021a), as well as the hot-start and cold-start Sonora−Bobcat (Marley et al. 2021) and the coldest-start Spiegel & Burrows (2012) evolutionary models. Similar to 51 Eri b (Macintosh et al. 2015), COCONUTS-2b has a sufficiently low luminosity to be consistent with the cold-start process that may form gas-giant (exo)planets. However, given its wide orbital separation, COCONUTS-2b probably formed in situ, like components in stellar binaries via the gravitational collapse of molecular cloud. In future work, high-quality spectroscopic follow-up and photometric monitoring of COCONUTS-2b can probe its atmosphere's composition (e.g., C/O ratio) and thermal structure, with the benefit of metallicity and abundance constraints provided from analysis of the host star. This will allow us to investigate the formation mechanism of the COCONUTS-2Ab system and study the T/Y transition of self-luminous exoplanetary atmospheres at young ages for the first time.

Zoom In Zoom Out Reset image size

We thank Jennifer van Saders for helpful discussions about stellar rotation and Eric Mamajek for helpful comments. This work has benefited from The UltracoolSheet at http://bit.ly/UltracoolSheet, maintained by Will Best, Trent Dupuy, Michael Liu, Rob Siverd, and Zhoujian Zhang, and developed from compilations by Dupuy & Liu (2012); Dupuy & Kraus (2013); Liu et al. (2016); Best et al. (2018), and Best et al. (2020a). This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This paper includes data collected by the TESS mission, which are publicly available from the Mikulski Archive for Space Telescopes (MAST). Funding for the TESS mission is provided by NASA's Science Mission directorate. This research has made use of the Exoplanet Follow-up Observation Program website, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program. Z.Z. and M.C.L. acknowledge support from the National Science Foundation through grant AST-151833. This work is funded in part by the Gordon and Betty Moore Foundation through grant GBMF8550 to M.C.L.

Software: emcee (Foreman-Mackey et al. 2013).