SCExAO/CHARIS Direct Imaging Discovery of a 20 au Separation, Low-mass Ratio Brown Dwarf Companion to an Accelerating Sun-like Star^*

Facility and now extreme adaptive optics systems have provided the first direct detections of exoplanets around nearby, young stars (Marois et al. 2008, 2010; Kuzuhara et al. 2013; Currie et al. 2014b, 2015; Macintosh et al. 2015; Keppler et al. 2018). Blind direct imaging surveys have provided the first constraints on the frequency of Jovian planets and low-mass brown dwarfs at orbital distances of 10–500 au (Brandt et al. 2014; Nielsen et al. 2019).

Unfortunately, the low yield of blind direct imaging surveys shows that exoplanets detectable with current instruments are rare (e.g., Nielsen et al. 2019), especially around solar- and subsolar-mass stars. The few companions found by these surveys typically exceed five Jovian masses and orbit well beyond separations where the Jovian planet frequency peaks (Fernandes et al. 2019). Improving blind survey yields will only be possible with substantial contrast gains at small separations and/or sensitivity gains at wider separations (Crepp & Johnson 2011). Limited sample sizes and sparse coverage of ages, temperatures, and surface gravities impede our understanding of the atmospheric evolution of gas giant planets.

Targeted searches focused on stars showing evidence of gravitational pulls from massive planets could improve survey yields. While radial-velocity data have been used for sample selection with some success for finding (sub)stellar companions (Crepp et al. 2014), astrometry can also select direct imaging targets, even around stars whose activity and spectral type preclude precise radial-velocity measurements. The Hipparcos-Gaia Catalog of Accelerations (HGCA; Brandt 2018) provides absolute astrometry for 115,000 nearby stars, including those with clear dynamical evidence for unseen massive companions. HGCA-derived accelerations can provide dynamical masses of imaged exoplanets and low-mass brown dwarfs independent of luminosity evolution models and irrespective of stellar age uncertainties (Brandt et al. 2019; Dupuy et al. 2019).

In this Letter, we report the direct imaging discovery of a low-mass substellar companion 20 au from the nearby Sun-like star HD 33632 Aa using the Subaru Coronagraphic Extreme Adaptive Optics Project (SCExAO; Jovanovic et al. 2015b) coupled to the CHARIS integral field spectrograph (Groff et al. 2016) and using the NIRC2 camera on Keck. The HGCA identifies an astrometric acceleration for the primary induced by HD 33632 Ab, providing a dynamical mass of 46 ± 8 M_J for the companion. CHARIS spectra and NIRC2 photometry reveal HD 33632 Ab to be a higher-gravity, higher-mass counterpart to young L/T transition planet-mass objects such as HR 8799 cde with a low eccentricity more typical of directly imaged planets than brown dwarfs.

The HD 33632 primary is a 26.56 pc distant F8V 1.1 M_⊙ main-sequence star (Takeda et al. 2007; Gaia Collaboration et al. 2018). The star has an M-star companion, 2MASS J05131845+3720463, at ρ ∼ 34'' (900 au; Scholz 2016) with the same parallax as HD 33632 Aa and a proper motion difference of ∼3 mas yr⁻¹ (Brandt 2018; Gaia Collaboration et al. 2018). ²⁶

Age estimates for the HD 33632 primary drawn from CaHK measurements (log(${R}_{\mathrm{HK}}^{{\prime} }$) = −4.76 to −4.93) encompass ∼1.2–4.5 Gyr (e.g., Takeda et al. 2007; Pace 2013). The Mamajek & Hillenbrand (2008) activity–rotation–age relation and gyrochronology as implemented in Brandt et al. (2014) favor the low end of this range (∼1–2.5 Gyr). HD 33632 Aa's fractional X-ray luminosity (log(L_X/L_bol = −5.2) overlaps with those for 800 Myr old Hyades stars (Brandt & Huang 2015; Freund et al. 2020). Li abundances are consistent with the Hyades but marginally consistent with 4 Gyr old M67 (Ramírez et al. 2012; Castro et al. 2016). Ages estimated using neutron-capture elements abundances individually (e.g., [Zr/Fe]) or in aggregate (e.g., the average abundances of "heavy" elements Ce, Ba, and La; [Y/Mg]) cluster around 1.5–2.5 Gyr (Spina et al. 2018). Given activity, gyrochronology, and neutron-capture abundance-derived estimates, we adopt a broad age range of ${1.5}_{-0.7}^{+3.0}$ Gyr with a preference for 1.0–2.5 Gyr.

HGCA reveals the HD 33632 primary to have a statistically significant acceleration (χ²  ≈ 116) for a model of constant proper motion, equivalent to 10.5σ with 2 degrees of freedom: evidence for the dynamical pull of a companion. The acceleration is almost 1000 times larger than that expected from the M-dwarf companion. The Lick radial-velocity survey monitored HD 33632 Aa for 11 years (1998 January 18 to 2009 February 1) without finding evidence of a planet or brown dwarf (Fischer et al. 2014): the companion must be widely separated. Following estimates in Brandt et al. (2019), its minimum mass at angles accessible by SCExAO/CHARIS (ρ  ∼  01–105) ranges from a few Jovian masses to over 50 M_J.

Table 1 summarizes our observations with SCExAO/CHARIS and Keck/NIRC2 in 2018 and 2020. The seeing ranged between θ_V  = 04 and 10; conditions were photometric each night. SCExAO achieved a high-fidelity AO correction; we utilized the CHARIS integral field spectrograph (Groff et al. 2016) in low spectral resolution (broadband) mode covering the JHK passbands simultaneously (1.16–2.37 μm at ${ \mathcal R }$ ∼ 18). NIRC2 data were taken in the L_p broadband filter (λ_o  = 3.78 μm) using Keck's facility AO system.

All CHARIS data utilized satellite spots for precise astrometric and spectrophotometric calibration (Jovanovic et al. 2015a). The CHARIS data were taken with a Lyot coronagraph (0139 radius occulting mask); the NIRC2 data were taken behind a Lyot coronagraph with a 02 radius occulting mask. All observations were conducted in pupil-tracking mode, enabling angular differential imaging (ADI; Marois et al. 2006); the CHARIS data also enable spectral differential imaging (SDI; Marois et al. 2000).

We extracted CHARIS data cubes using the standard CHARIS pipeline (Brandt et al. 2017) and performed basic reduction steps including sky subtraction, image registration, and spectrophotometric calibration as in Currie et al. (2018). For spectrophotometric calibration, we adopted a Kurucz stellar atmosphere model appropriate for an F8V star. We reduced the NIRC2 data using the ADI-based pipeline from Currie et al. (2011).

For all data, we used A-LOCI for point-spread function (PSF) subtraction (Currie et al. 2018). Due to a lack of field rotation, we subtracted the PSF for the 2018 CHARIS data in SDI mode only. The 2020 CHARIS data enabled PSF subtraction with ADI, SDI, and ADI+SDI (i.e., performing SDI on the A-LOCI PSF subtraction residuals produced from ADI). In all cases, we used a large rotation gap/radial scaling criterion of δ > 1 λ/D to limit self-subtraction. For the CHARIS SDI reduction (component), we employed a pixel mask over the subtraction zone (Marois et al. 2010; Currie et al. 2015) to significantly reduce biasing of a companion spectrum.

Figure 1 shows the detection of a faint point source, hereafter HD 33632 Ab, ρ ≈ 075 west from the primary in each epoch. The detection's signal-to-noise ratios (S/Ns) in conservative ADI/A-LOCI or SDI/A-LOCI reductions range between 11–22 in the 2018 discovery epoch to 41 for the 2020 August 31 CHARIS broadband follow-up data. CHARIS data do not identify any other companion within ρ ∼ 105. In addition to resolving HD 33632 Ab, NIRC2 data identify a wider separation object at [E, N]'' = [−122, −185] and L_p ∼ 16.3 (not shown). However, 2020 CHARIS data reveal this object to be a background star based on its astrometry and its near-infrared brightness.

Zoom In Zoom Out Reset image size

3.1. Infrared Colors, Spectrum, and Atmosphere of HD 33632 Ab

To derive throughput-corrected spectrophotometry, we forward-modeled point sources at HD 33632 Ab's locations using stored LOCI coefficients (Currie et al. 2018). We focus on ADI-only reductions due to the more straightforward applicability of ADI-reduced data to forward modeling (Pueyo 2016). We adopt our highest-quality data set (2020 August 31) for our analyses. Our results agree with those from other data sets and with SDI reductions.

We extract spectra from the best-fit centroid position of the wavelength-collapsed image. Throughputs typically exceed 75%; because we use ADI only, they are independent of HD 33632 Ab's spectrum. HD 33632 Ab's broadband photometry in standard Maunakea Observatory filters is $J=16.91\pm 0.11$, H = 16.00 ± 0.09, K_s = 15.37 ± 0.09, and L_p = 13.67 ± 0.15 (${\rm{\Delta }}J,H,{K}_{{\rm{s}}}=11.52,10.83,10.21$). Photometry obtained at different epochs or with different reductions agrees with these results to within errors.

HD 33632 Ab lies at the transition from cloudy L dwarfs to (nearly) cloud-free T dwarfs (Burrows et al. 2006). Figure 2 places it on an infrared color–magnitude diagram. The companion's J/J-K_s position is consistent with field L/T objects, while its L_p/H-L_p position agrees with both field objects and the low-gravity, cloudy exoplanets HR 8799 cde (Currie et al. 2011; Barman et al. 2015).

Zoom In Zoom Out Reset image size

The CHARIS spectrum of HD 33632 Ab (Figure 2, bottom-left panel; Table 2) is highly peaked, characteristic of a substellar atmosphere shaped by water opacity. The 2018 CHARIS spectrum is significantly noisier and offset by ∼10% compared to the 2020 epoch data. Otherwise, spectra extracted using different reduction techniques and taken on different epochs agree to within errors. The detection significance of HD 33632 Ab reaches S/N = 20–30 in some channels for the 2020 August 31/ADI-reduced spectrum. However, contributions from uncertainties in the absolute flux calibration (i.e., the satellite spot flux uncertainty) limit the spectrophotometric precision to about 10%.

Table 2. HD 33632 Ab Spectrum

Wavelength (μm)	Fν (mJy)	σ Fν (mJy)	S/N
1.160	0.210	0.029	8.9
1.200	0.249	0.027	12.6
1.241	0.253	0.030	10.7
1.284	0.304	0.028	13.9
1.329	0.225	0.027	10.7
1.375	0.113	0.022	5.5
1.422	0.175	0.024	8.1
1.471	0.195	0.024	9.3
1.522	0.273	0.028	13.0
1.575	0.387	0.037	14.4
1.630	0.505	0.038	21.3
1.686	0.479	0.038	20.8
1.744	0.360	0.037	16.1
1.805	0.209	0.026	10.2
1.867	0.288	0.029	13.0
1.932	0.247	0.028	10.9
1.999	0.312	0.030	14.2
2.068	0.514	0.044	28.0
2.139	0.519	0.035	30.1
2.213	0.445	0.038	22.7
2.290	0.380	0.040	12.7
2.369	0.356	0.060	6.5

Note. Throughput-corrected HD 33632 Ab spectrum extracted from 2020 August 31 data, reduced using ADI/A-LOCI.

Download table as:  ASCIITypeset image

We compared HD 33632 Ab's CHARIS spectrum to objects in the Montreal Spectral Library ²⁷ (e.g., Gagné et al. 2014), considering the impact of spatially and spectrally correlated noise (Greco & Brandt 2016). The spectral covariance at HD 33632 Ab's angular separation in the 2020 August 31 ADI/A-LOCI reduced data is nearly zero for off-diagonal elements indicating largely uncorrelated noise.

The L9.5 field brown dwarfs SIMPJ0956-1447 and SIMPJ0150+3827 provide the best fit to HD 33632 Ab's spectrum (Figure 2, bottom-right panel). While the Montreal library contains few objects later than L5, the ${\chi }^{2}$ distribution for library objects exhibits a local minimum between L6.5 and T0.5. Thus, conservatively we assign HD 33632 Ab a spectral type of L9.5${}_{-3.0}^{+1.0}$. Adopting the mapping between spectral type and temperature from Stephens et al. (2009), the object's effective temperature is ${1300}_{-100}^{+100}$ K: slightly higher than that derived for the inner three HR 8799 planets (Currie et al. 2011; Barman et al. 2015). Adopting the polynomial fit from Golimowski et al. (2004) and a distance of 26.56 pc, HD 33632 Ab's luminosity is ${\mathrm{log}}_{10}(L/{L}_{\odot })=-{4.62}_{-0.08}^{+0.04}$.

3.2. Astrometry and Common Proper Motion

3.3. Orbit and Dynamical Mass

We use the open-source code orvara (Brandt et al. 2020) to fit HD 33632 Ab's mass and orbit using a combination of HGCA measurements, Lick Observatory radial-velocity data (Fischer et al. 2014), and relative astrometry from CHARIS and NIRC2. No other bound companion is detected with CHARIS, Lick data do not identify an inner companion, and the only other object seen with NIRC2 is a background star. Given its separation, the M-dwarf companion contributes <1% of the acceleration seen by HGCA. Therefore, we assume that HD 33632 Ab is solely responsible for the HGCA acceleration. The astrometric acceleration of ∼0.1 mas yr⁻² is more than 1000 times smaller than the apparent acceleration due to parallactic motion: it negligibly affects the Gaia parallax. We fit for a radial-velocity jitter, the six Keplerian orbital elements, and the masses of both components. We adopt a prior on HD 33632 Aa's mass of 1.1 ± 0.1 M_⊙ and a log-uniform prior on the mass of HD 33632 Ab. We analytically marginalize out the nuisance parameters of barycenter proper motion, parallax, and radial-velocity zero-point.

Figure 3 shows our posterior distributions for selected orbital parameters, the primary mass (almost identical to our prior of 1.1 ± 0.1 M_⊙), and HD 33632 Ab's mass. Table 3 lists all priors and posteriors from the fit. The companion has a best-fit semimajor axis of ${21.2}_{-4.5}^{+4.1}$ au with an orbit inclined by $i=39\buildrel{\circ}\over{.} {4}_{-20}^{+8.0}$. The semimajor axis is double-peaked because our relative astrometry spans only a 2 yr baseline, while there is a 25 yr elapse between absolute astrometric measurements: Hipparcos mainly constrains HD 33632 Ab's 1991 position. Continued orbital monitoring will remove the double peak, improving the companion's mass measurement. HD 33632 Ab's median posterior eccentricity is e = 0.19 but the distribution peaks at zero (i.e., a circular orbit): the 68% and 95% upper limits are 0.29 and 0.46, respectively.

Zoom In Zoom Out Reset image size

Table 3. Markov Chain Monte Carlo Orbit Fitting Priors and Results

Parameter	16%/50%/84% Quantiles	95% Confidence Interval	Prior
Fitted Parameters
RV jitter (m s−1)	${2.53}_{-0.18}^{+0.19}$	(2.19,2.91)	log-flat
${M}_{\mathrm{pri}}$ (M⊙)	${1.108}_{-0.093}^{+0.094}$	(0.93,1.29)	Gaussian,1.1 ± 0.1
Msec (MJup)	${46.4}_{-7.5}^{+8.1}$	(32.5,62.8)	1/Msec (log-flat)
Semimajor axis a (au)	${21.2}_{-4.5}^{+4.1}$	(15.2,30.9)	$1/a$ (log-flat)
$\sqrt{e}\sin \omega$ a	${0.05}_{-0.24}^{+0.21}$	(−0.39,0.47)	uniform
$\sqrt{e}\cos \omega$ a	${0.15}_{-0.52}^{+0.37}$	(−0.62,0.64)	uniform
Inclination (°)	${39.4}_{-20}^{+8.0}$	(6.0,52.2)	$\sin i$ (geometric)
PA of the ascending node Ω (°)	${38.2}_{-7.0}^{+7.2}$	(21.3,256)	uniform
Mean longitude at 2010.0 (°)	${21}_{-10}^{+14}$	(10.3,286)	uniform
Parallax (mas)	$37.647\pm 0.071$	(37.52,37.78)	Gaussian,37.646 ± 0.064
Barycenter ${\mu }_{\alpha * }$ (mas yr−1)	−144.90 ± 0.13	(−145.15,−144.66)	uniform
Barycenter ${\mu }_{\delta }$ (mas yr−1)	−135.21 ± 0.32	(−135.8, −134.55)	uniform
Radial-velocity zero-point (m s−1)	$-{129}_{-68}^{+31}$	(−246, 180)	uniform
Derived Parameters
Period (yrs)	91 ± 27	(55, 160)
Argument of periastron ω (°) a	${151}_{-131}^{+155}$	(4, 356)
Eccentricity e	${0.18}_{-0.14}^{+0.20}$	<0.46
Semimajor axis (mas)	${800}_{-170}^{+150}$	(570, 1160)
Periastron time T0 (JD)	${2440900}_{-13100}^{+4500}$	(2404000, 2453500)
Mass ratio	${0.0402}_{-0.0073}^{+0.0078}$	(0.0267, 0.0563)

Note.

^a Orbital parameters listed are of HD 33632 Ab. HD 33632 Aa has ω shifted by 180°.

Download table as:  ASCIITypeset image

HD 33632 Ab has a dynamical mass of ${46.4}_{-7.6}^{+8.1}$ ${M}_{{\rm{J}}}$ and a mass ratio of q ∼ 4.0% ± 0.7%. We may compare this to masses derived by matching its observed luminosity and age with substellar evolutionary models. For its full (preferred) age range of ${1.5}_{-0.7}^{+3.0}$ Gyr (1–2.5 Gyr), the Baraffe et al. (2003) and Burrows et al. (1997) luminosity evolution models imply masses of ${48}_{-9}^{+17}$ M_J (41–60 M_J) and ${55}_{-15}^{+18}$ M_J (45–65 M_J), respectively. These model-dependent masses agree well with the dynamical mass if we adopt our preferred age of 1–2.5 Gyr, but show moderate tension for ages ≳2.5 Gyr.

Our SCExAO/CHARIS and complementary Keck/NIRC2 imaging identify a low-mass-ratio brown dwarf companion at r_proj ∼ 20 au around the main-sequence Sun-like star, HD 33632 Aa. HD 33632 Ab's infrared colors and spectrum are best matched by a field L/T transition object. Combining measurements from Hipparcos, Gaia, and precision radial-velocity with our CHARIS/NIRC2 relative astrometry yields a dynamical mass of ${46.4}_{-7.6}^{+8.1}$ M_J. This is more precise than estimates from luminosity evolution models alone, given age uncertainties, and it will improve with continued astrometric monitoring and future Gaia data releases. Provided that the system age and companion mass are better constrained in future work, HD 33632 Ab adds to the still-small sample of objects enabling us to test substellar evolutionary models across masses and ages.

HD 33632 Ab lies at the field L/T transition tracking the dissipation of clouds/dust in substellar atmospheres. In L_p/H-L_p, its colors overlap with and its temperature is just slightly exceeds those of the young exoplanets HR 8799 cde. HD 33632 Ab represents an older, higher-mass, higher-gravity, and (likely) less cloudy/dusty counterpart to some of the first imaged exoplanets. Thermal infrared follow-up imaging and spectroscopy on the ground and with the James Webb Space Telescope could further clarify how its atmospheric chemistry compares with HR 8799 cde and other companions. Theoretical modeling incorporating these higher-resolution data will quantify constraints on HD 33632 Ab's atmosphere.

HD 33632 Ab's mass, mass ratio, and orbital separation are nearly identical to those of the older, colder brown dwarf companion GJ 758 B (Thalmann et al. 2009; Brandt et al. 2019), which likewise orbits a Sun-like star. Its mass and mass ratio bracket those of other companions to mid-F to early-G stars orbiting at 20–50 au: e.g., the more massive, older T dwarf companion HD 19467 B (Crepp et al. 2014) and (near-)planet-mass companions HR 2562 b (Konopacky et al. 2016) and GJ 504 b (Kuzuhara et al. 2013) with comparable or younger ages.

HD 33632 Ab's eccentricity posterior probability distribution has a 68% upper limit of 0.29. This is smaller than the eccentricities of most brown dwarfs studied in Bowler et al. (2020), who argue that systematically higher eccentricities of brown dwarf-mass companions relative to exoplanets point to a different formation mechanism. Future CHARIS/NIRC2 astrometry for HD 33632 Ab together with more precise HD 33632 Aa astrometry in future Gaia data releases will better constrain the companion's mass and eccentricity.

In addition to HD 33632 Aa, hundreds of nearby stars show evidence for a statistically significant acceleration plausibly caused by an unidentified substellar companion. Upcoming Gaia data releases will identify thousands more accelerating stars. Some will be young enough that extreme AO systems like SCExAO can image self-luminous Jovian exoplanets responsible for the astrometric trends. Our work is a proof in concept that direct imaging searches targeting HGCA-selected stars may identify new substellar and potentially even planetary companions, increasing the yield of discoveries.

We thank Eric Mamajek and William Cochran for helpful comments regarding the HD 33632 system properties. The authors acknowledge the very significant cultural role and reverence that the summit of Maunakea holds within the Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

We acknowledge the critical importance of the current and recent Subaru and Keck Observatory daycrew, technicians, support astronomers, telescope operators, computer support, and office staff employees, especially during the challenging times presented by the COVID-19 pandemic. Their expertise, ingenuity, and dedication is indispensable to the continued successful operation of these observatories.

We thank the Subaru and NASA Keck Time Allocation Committees for their generous support of this program. T.C. was supported by a NASA Senior Postdoctoral Fellowship and NASA/Keck grant LK-2663-948181. T.B. gratefully acknowledges support from the Heising-Simons foundation and from NASA under grant #80NSSC18K0439. M.T. is supported by JSPS KAKENHI grant #18H05442.

The development of SCExAO was supported by JSPS (Grant-in-Aid for Research #23340051, #26220704, and #23103002), Astrobiology Center of NINS, Japan, the Mt Cuba Foundation, and the director's contingency fund at Subaru Telescope. CHARIS was developed under the support by the Grant-in-Aid for Scientific Research on Innovative Areas #2302. Some of the data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation.