Free Access
Issue
A&A
Volume 565, May 2014
Article Number L4
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201423839
Published online 05 May 2014

© ESO, 2014

1. Introduction

HD 95086 b is a directly imaged planet (5 ± 2 MJ, aproj = 55.7 ± 2.5 AU) discovered by Rameau et al. (2013a) in L (3.8 μm) with VLT/NaCo (Lenzen et al. 2003; Rousset et al. 2003) orbiting the young A8 star HD 95086 (M ~ 1.6 M), a member of the Lower Centaurus Crux subgroup (17 ± 4 Myr, Pecaut et al. 2012; Meshkat et al. 2013). Additional L images taken later in 2013 confirmed that the object is comoving with its star (Rameau et al. 2013b).

NaCo Ks (2.18 μm) and NICI (Chun et al. 2008) H-band (1.65 μm) observations failed to reveal the planet (Rameau et al. 2013a; Meshkat et al. 2013). However, 5σ lower limits of Ks-L′ = 1.2 ± 0.5 mag and HL′ = 3.1 ± 0.5 mag suggest that the planet may have extremely red colors, similar to the young planets HR 8799 bcde and 2M 1207 b (Chauvin et al. 2004; Marois et al. 2008, 2010a), which have very dusty/cloudy atmospheres (Barman et al. 2011; Currie et al. 2011). Higher contrast near-IR data able to detect HD 95086 b can provide better comparisons with these objects and better constrain its atmosphere.

In this Letter, we present detections of HD 95086 b with the recently installed Gemini Planet Imager (GPI, Macintosh et al. 2014) on Gemini South from public data as a part of GPI commissioning observations (Perrin et al. 2014). The data (acquired and reduced by the GPI team), their analysis, and the detections are presented in Sect. 2. In Sect. 3, we combine GPI H and K1 photometry with NaCo L photometry to constrain the physical properties of HD 95086 b.

2. Observations and data reduction

The GPI is a new instrument for imaging and characterizing planets around young nearby bright stars, combining an extreme adaptive optics system, coronagraphs, and an integrated field spectrograph (IFS). The IFS platescale is 14.3 ± 0.3 mas px-1 for a 2.8′′ field-of-view (FOV) and the true North position angle is given within 1 deg1.

Table 1

Observing log of HD 95086 with GPI.

The HD 95086 spectral data were obtained at H (1.5 − 1.8 μm, R = 44 − 49) and K1 (1.9 − 2.19 μm, R = 62 − 70) in 2013 December using apodized Lyot coronagraphs (Table 1) and angular differential imaging (ADI, Marois et al. 2006a). Conditions were good: 0.43′′ and 0.6′′ DIMM seeing, air masses of 1.32 and 1.34, and coherence times of 19 ms and 17 ms, respectively. The GPI commissioning team used their pipeline for bad-pixel removal, destriping, non-linearity and persistence corrections, flat-fielding, wavelength calibration, and converting the data into spectral data cubes. We used the data cubes relying on the GPI pipeline quality. The data are made of 21 and 17 spectral cubes at H and K1 bands, respectively, consisting of 37 spectral channels each.

To further process the data, we registered each slice of the spectral cubes using the barycenter of the four satellite spots (attenuated replica of the central star PSF induced by a grid placed in a pupil plane, Marois et al. 2006b). Then, we minimized the speckle noise in each slice using independent pipelines each adopting various methods (Marois et al. 2006a; Lafrenière et al. 2007; Lagrange et al. 2010; Boccaletti et al. 2012; Chauvin et al. 2012; Soummer et al. 2012; Currie et al. 2013; Marois et al. 2014) used for ADI and spectral differential imaging (SSDI, Racine et al. 1999). Finally, all slices were mean-combined to yield an integrated broad-band image to maximize the signal-to-noise ratio (S/N) of any off-axis source. Binning images in wavelength and suppressing the speckles (ADI), or suppressing the speckles in each spectral channel (ADI/ADI+SSDI) and binning images give similar results, and all our pipelines recover HD 95086 b, which is the sole bright spot at the expected separation. Thus, we provide the first detections at H and K1 bands (Fig. 1) with an S/N of ~3–4 and 5–6, respectively. The central bright speckles are masked up to 500 mas. These are the first detections of HD 95086 b with an instrument that is not NaCo/VLT. No spectrum can be extracted given the low S/N.

thumbnail Fig. 1

Final images of the HD 95086 system at H (top) and K1 (bottom) bands from two of our pipelines. The planet (arrow) is detected in all images. The bright speckles are masked up to 500 mas from the central star.

Open with DEXTER

To estimate the planet flux and position, we needed unsaturated GPI PSFs. As GPI cannot acquire off-axis observations of the star, we calibrated photometry and astrometry using the satellite spots, which are expected to have same shape and brightness for a given filter. In the laboratory the spot-to-central-star flux ratios were 2.035 × 10-4 (9.23 mag) and 2.695 × 10-4 (8.92 mag) at H and K1 bands. To check these values, we compared H and K photometry of HD 8049 B (VLT/ NACO-SINFONI, Zurlo et al. 2013) with our measurements derived from public GPI HD 8049 data. Assuming that the object is not photometrically variable with time and considering the laboratory spot contrasts, GPI and VLT photometry are consistent within ϵ1 = 0.2 mag, which we take as the error on the ratios. From these ratios, we assessed biases induced by our processing by injecting fake point-sources (i.e., unsaturated PSFs) into the data before applying speckle-suppression techniques (Lagrange et al. 2010; Marois et al. 2010b; Chauvin et al. 2012; Galicher et al. 2012). We obtained templates of the planet image. Adjusting the flux of the templates, we found the planet photometry and the fitting error ϵ2, which depends on the detection quality. ϵ2 is 0.8 mag and 0.3 mag at H and K1. Finally, we estimated the variation ϵ3 of stellar flux over the sequence with the variation of spot flux. ϵ3 is 0.2 mag and 0.3 mag over the H and K1 sequences including the variations between spots. The resulting photometric error is the quadratic error, which is dominated by the low S/N at H and is a mix of all errors at K1.

For the astrometric error, we considered uncertainties in the centroiding accuracy of individual slices (0.3 pixel), the plate scale (0.02 pixel), the planet template fit (0.7 pixel at H, 0.5 pixel at K1), and the North position angle (1 deg). The error is dominated by the low S/N of the detections and the generic GPI calibrations. The current precision is good enough to assess the comoving status of the companion (Fig. 2). We tried to use the astrometric standard HD 8049 B in GPI data to better constrain the North orientation. We did not succeed because of the high orbital motion of HD 8049 B and because there is no contemporary observation from other instruments.

thumbnail Fig. 2

HD 95086 b positions from its star in RA (Δα) and Dec (Δδ). GPI and NaCo measurements are marked in blue and expected positions of a background object in yellow.

Open with DEXTER

Final measurements are presented in Table 2. We include revised 2012 NaCo L photometry obtained by 1) better calibrating the planet signal (as in Currie et al. 2013) and 2) precisely deriving the L neutral density (ND) filter throughput (used to flux-calibrate HD 95086) by comparing ND and unsaturated β Pic data.

3. Characterization

Absolute magnitudes were derived from the contrast ratios (Table 2): MH = 15.29 ± 0.91 mag, MK1 = 14.11 ± 0.51 mag, and ML = 11.44 ± 0.22 mag using the 2MASS and WISE W1 (Cutri et al. 2003, 2012) photometry of the star2.

Combining the H band GPI data with revised NaCo L data, we compared the L/HL color–magnitude diagram position of HD 95086 b with that of young companions, field dwarfs (Leggett et al. 2010, 2013), and LYON evolutionary tracks (Chabrier et al. 2000; Baraffe et al. 2003) generated for the GPI/NaCo passbands3.

thumbnail Fig. 3

Color–magnitude diagram using the new H-band photometry of HD 95086 b (yellow star) and data from Bonnefoy et al. (2013), Bonnefoy et al. (2014), and Currie et al. (2014).

Open with DEXTER

Table 2

HD 95086 b photometry and astrometry at H and K1 (GPI data) and L (Rameau et al. 2013a,b, and revision*).

We converted the GPI measurements into H photometry by applying correction factors derived from published spectra, the filter transmissions, and a spectrum of Vega. HD 95086 b lies at the LT transition in this diagram, similar to other young (8–30 Myr) planets like HR 8799 cde (Marois et al. 2008, 2010a) and 2M1207 b (Chauvin et al. 2004). Its red HL color compared with the sequence of field dwarf objects (Leggett et al. 2010, 2013) suggests a high content of photospheric dust (Barman et al. 2011; Currie et al. 2011), owing to reduced surface gravity (e.g. Fig. 11 of Marley et al. 2012).

We built the 1.5–4.8 μm spectral energy distribution (SED) of the planet following Bonnefoy et al. (2013) by combining the GPI photometry with the L one. The normalized SED (at L) is best compatible with the young exoplanets HR 8799 bcde and 2M1207 b, but is redder. Its colors are also ~1 mag redder than those of the benchmark dusty L6.5L7.5 field dwarf 2MASS J22443167+2043433 (Dahn et al. 2002; Stephens et al. 2009).

We also compared the SED of HD 95086 b with the predictions from grids of synthetic spectra for BT-SETTL (Allard et al. 2012) and LESIA atmospheric models (Baudino et al. 2014). Each synthetic SED was normalized to that of HD 95086 b at L. The BT-SETTL grid covers 400K ≤ Teff ≤ 3500K with 50 to 100 K increments, −0.5 ≤ logg ≤ 6.0 dex with 0.5 dex increments, and M/H = 0.0 or +0.5 dex. The BT-SETTL models that reproduce the photometry of HD 95086 b have 600K ≤ Teff ≤ 1500K and 3.5dex ≤ log g ≤ 4.5dex. The three LESIA grids assume 700K ≤ Teff ≤ 2100K, 2.1 ≤ logg ≤ 4.5 dex, and solar abundances: one without clouds and two with clouds of Fe and Mg2SiO4 particles. For each LESIA model, we selected the planet radius that minimizes χ2 between the observed and calculated apparent magnitudes. We only kept models with a radius in a realistic range derived from evolution models (0.6 to 2 Jupiter radii, Mordasini et al. 2012). All LESIA models that reproduced the HD 950866 b photometry have 900K ≤ Teff ≤ 1500K and 2.1dex ≤ logg ≤ 4.5dex.

The planet mass cannot be derived from the atmosphere models, and evolutionary models are needed. Comparing the planet’s L luminosity with hot-start DUSTY and COND models for an age of 17 ± 4 Myr, we find a planet mass of M = 5± 2 MJup (Table 3). We did not use the H and K1 photometries because they are poorly reproduced by the models for an object at the LT transition (larger uncertainties than for L). The models predict Teff and log g, in agreement with those derived from the SED fit.

Table 3

Physical parameters predicted by hot-start evolutionary models for the observed absolute magnitudes.

Alternatively, we used the warm-start models (Spiegel & Burrows 2012) to account for possible different initial conditions for the planet (parameterized by the initial entropy between 8 and 13 kB/baryon). The models assume solar metallicity and atmospheres enriched by a factor of 3 with/without dust clouds as boundary conditions. Synthetic SEDs are generated from predicted spectra of planets4. For the full range of initial entropies we considered, models assuming masses of 4 − 14 MJup match the SED of HD 95086 b (Fig. 4). For much of this range (Sinit = 9.5 − 13) a mass of 4 MJup is favored.

thumbnail Fig. 4

Combination of initial entropies (Sinit) and masses (shaded areas) for which the planet 1.6–4.8 μm photometries are reproduced by the warm-start models of Spiegel & Burrows (2012) within 1σ. Three boundary conditions are considered: with (hy) and without (cf) cloudy atmospheres, at solar (1s) and 3x solar (3s) metallicity. Initial entropies for the cold-start (filled circles) and hot-start (open circles) models of Marley et al. (2007) are overlaid.

Open with DEXTER

4. Conclusions

We reported the near-IR detections of HD 95086 b from GPI public commissioning data. We confirmed that the companion is comoving with HD 95086 and derived the first estimates of its magnitudes with respect to its star: H = 13.1 ± 0.9 and K1 = 12.1 ± 0.5.

While the mid-IR luminosity of HD 95086 b is best consistent with an LT transition object, it has redder near-IR colors than other young, imaged planet-mass companions. This is consistent with a very dusty and low surface gravity atmosphere.

Comparison with atmosphere models provided 600K ≤ Teff ≤ 1500K and 2.1dex ≤ log g ≤ 4.5dex. Evolutionary models are consistent with a mass of 5 ± 2 MJup. However, the models are affected by systematic errors that are difficult to quantify because of the lack of young objects at the LT transition.

More higher-precision spectroscopic and photometric data for HD 95086 b are required to refine the planet properties.


2

Correction factors from the GPI/NaCo and 2MASS/WISE photometry, derived from the spectrum of an A7III star in the Pickles et al. (1998) library, are negligible.

Acknowledgments

We thank the consortium who built the GPI instrument and the data analysis team for developing reduction tools. We are grateful to Dave Spiegel and Adam Burrows for making the warm-start models publicly available. J.R., M.B., G.C., and A.M.L. acknowledge financial support from the French National Research Agency (ANR) through project grant ANR10-BLANC0504-01. This research has benefitted from the SpeX Prism Spectral Libraries, maintained by Adam Burgasser at http://pono.ucsd.edu/~adam/browndwarfs/spexprism. J.L.B.’s PhD is funded by the LabEx “Exploration Spatiale des Environnements Planétaires” (ESEP) # 2011-LABX-030. T.C. is supported by a McLean Postdoctoral Fellowship.

References

All Tables

Table 1

Observing log of HD 95086 with GPI.

Table 2

HD 95086 b photometry and astrometry at H and K1 (GPI data) and L (Rameau et al. 2013a,b, and revision*).

Table 3

Physical parameters predicted by hot-start evolutionary models for the observed absolute magnitudes.

All Figures

thumbnail Fig. 1

Final images of the HD 95086 system at H (top) and K1 (bottom) bands from two of our pipelines. The planet (arrow) is detected in all images. The bright speckles are masked up to 500 mas from the central star.

Open with DEXTER
In the text
thumbnail Fig. 2

HD 95086 b positions from its star in RA (Δα) and Dec (Δδ). GPI and NaCo measurements are marked in blue and expected positions of a background object in yellow.

Open with DEXTER
In the text
thumbnail Fig. 3

Color–magnitude diagram using the new H-band photometry of HD 95086 b (yellow star) and data from Bonnefoy et al. (2013), Bonnefoy et al. (2014), and Currie et al. (2014).

Open with DEXTER
In the text
thumbnail Fig. 4

Combination of initial entropies (Sinit) and masses (shaded areas) for which the planet 1.6–4.8 μm photometries are reproduced by the warm-start models of Spiegel & Burrows (2012) within 1σ. Three boundary conditions are considered: with (hy) and without (cf) cloudy atmospheres, at solar (1s) and 3x solar (3s) metallicity. Initial entropies for the cold-start (filled circles) and hot-start (open circles) models of Marley et al. (2007) are overlaid.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.