The International Deep Planet Survey I.The frequency of wide-orbit massive planets around A-starsBased on observations collected at the European Southern Observatory, Chile, ESO programs 081.C-0519, 083.C-0706, 084.C-0605, 087.C-0559, 088.C-0477, and at the Gemini North observatory, Gemini programs GN-2007B-Q-59, GN-2008A-Q-77, GN-2008B-Q-64, GN-2009A-Q-80, GN-2009B-Q-93.

The International Deep Planet Survey I.
The frequency of wide-orbit massive planets around A-starsthanks: Based on observations collected at the European Southern Observatory, Chile, ESO programs 081.C-0519, 083.C-0706, 084.C-0605, 087.C-0559, 088.C-0477, and at the Gemini North observatory, Gemini programs GN-2007B-Q-59, GN-2008A-Q-77, GN-2008B-Q-64, GN-2009A-Q-80, GN-2009B-Q-93.

A. Vigan Astrophysics group, School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL, United Kingdom
arthur@astro.ex.ac.uk
   J. Patience Astrophysics group, School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL, United Kingdom
arthur@astro.ex.ac.uk Arizona State University, School of Earth and Space Exploration, PO Box 871404, Tempe, AZ 85287-1404, USA
   C. Marois National Research Council of Canada, 5071 West Saanich Road, Victoria, British Columbia V9E 2E7, Canada    M. Bonavita Department of Astronomy, University of Toronto, Toronto, ON, Canada    R. J. De Rosa Astrophysics group, School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL, United Kingdom
arthur@astro.ex.ac.uk Arizona State University, School of Earth and Space Exploration, PO Box 871404, Tempe, AZ 85287-1404, USA
   B. Macintosh Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, California 94550, USA   
I. Song
University of Georgia, Department of Physics and Astronomy, 240 Physics, Athens, GA 30602, USA
   R. Doyon Département de Physique, Université de Montréal, C.P. 6128, Succursale Centre-ville, Montréal, QC H3C 3J7, Canada    B. Zuckerman Department of Physics and Astronomy, University of California, Los Angeles, CA 90095, USA    D. Lafrenière Département de Physique, Université de Montréal, C.P. 6128, Succursale Centre-ville, Montréal, QC H3C 3J7, Canada    T. Barman Lowell Observatory, 1400 West Mars Hill Road, Flagstaff, AZ 86001, USA
Received 9 February 2012 / Accepted 15 June 2012
Key Words.:
instrumentation: adaptive optics – instrumentation: high angular resolution – methods: observational – methods: statistical – stars: imaging

Breakthrough direct detections of planetary companions orbiting A-type stars confirm the existence of massive planets at relatively large separations, but dedicated surveys are required to estimate the frequency of similar planetary systems. To measure the first estimation of the giant exoplanetary systems frequency at large orbital separation around A-stars, we have conducted a deep-imaging survey of young (8–400 Myr), nearby (19–84 pc) A- and F-stars to search for substellar companions in the 10–300 AU range. The sample of 42 stars combines all A-stars observed in previous AO planet search surveys reported in the literature with new AO observations from VLT/NaCo and Gemini/NIRI. It represents an initial subset of the International Deep Planet Survey (IDPS) sample of stars covering M- to B-stars. The data were obtained with diffraction-limited observations in - and -band combined with angular differential imaging to suppress the speckle noise of the central stars, resulting in typical 5 detection limits in magnitude difference of 12 mag at 1, 14 mag at 2 and 16 mag at 5 which is sufficient to detect massive planets. A detailed statistical analysis of the survey results is performed using Monte Carlo simulations. Considering the planet detections, we estimate the fraction of A-stars having at least one massive planet (3–14 M) in the range 5–320 AU to be inside 5.9–18.8% at 68% confidence, assuming a flat distribution for the mass of the planets. By comparison, the brown dwarf (15–75 M) frequency for the sample is 2.0–8.9% at 68% confidence in the range 5–320 AU. Assuming power law distributions for the mass and semimajor axis of the planet population, the AO data are consistent with a declining number of massive planets with increasing orbital radius which is distinct from the rising slope inferred from radial velocity (RV) surveys around evolved A-stars and suggests that the peak of the massive planet population around A-stars may occur at separations between the ranges probed by existing RV and AO observations. Finally, we report the discovery of three new close M-star companions to HIP 104365 and HIP 42334.

1 Introduction

An extensive population of exoplanets has been discovered down to sub-Jovian masses and at orbital separations below 5 astronomical units (AU) based on large-scale radial velocity (RV) surveys (Mayor & Udry 2008; Marcy et al. 2008) and on transit surveys, which are now identifying hundreds of new candidates (Cabrera et al. 2009; Borucki et al. 2011). These indirectly detected planets provide invaluable information on the distribution of close orbit planets (Cumming et al. 2008) and on their frequency around nearby stars covering a large range of masses (Marcy et al. 2008; Johnson et al. 2010a). Planets at orbital radii larger than 5 AU, comparable to the locations of the giant planets in the Solar System, remain outside the range of detection of these methods, and, consequently, little is known about the extrasolar population of wide orbit planetary systems.

Direct detection with high-contrast imaging provides a method to explore the wide orbit planet population. Several adaptive optics (AO) surveys, concentrating on Solar-type stars, have been conducted to search for low-mass substellar companions around nearby young stars. Because direct imaging needs to overcome the large contrast ratio between the star and a potential substellar companion, the existing surveys were focused on FGKM stars, around which young objects down to a few masses of Jupiter (M) would be detectable at separation larger than a few tens of AUs. The majority of these surveys reported no (Masciadri et al. 2005; Biller et al. 2007; Lafrenière et al. 2007a; Kasper et al. 2007; Leconte et al. 2010; Janson et al. 2011; Delorme et al. 2012) or few (Chauvin et al. 2010) substellar companions to nearby stars and have placed upper limits on the population of massive planets at large orbital separation. Some objects were nonetheless discovered (e.g. Lafrenière et al. 2008; Thalmann et al. 2009; Biller et al. 2010), demonstrating that these objects exist, but are indeed rare (Nielsen & Close 2010).

Recent breakthrough detections around young A-stars – HR 8799 (Marois et al. 2008b, 2010b),  Pictoris (Lagrange et al. 2009, 2010) and Fomalhaut (Kalas et al. 2008; Janson et al. 2012) – have provided new perspectives on the search for companions orbiting more massive stars. The results of RV surveys of the evolved counterparts of A-stars (Johnson et al. 2007) have also provided intriguing results (Johnson et al. 2010b; Bowler et al. 2010a; Johnson et al. 2011), with significant differences in the planet population compared to RV-detected exoplanets around lower mass stars. Statistical analysis of the results clearly shows a higher frequency of planets around more massive stars and larger masses for the detected planets (Johnson et al. 2010a). Moreover, theoretical work on planet formation by core-accretion similarly shows the same kind of correlation between the star and planet mass (Kennedy & Kenyon 2008; Alibert et al. 2011).

Although more technically challenging in terms of observations, these recent results suggest that more massive stars may present more favorable targets in terms of exoplanet detections. And estimations of planet yield show that A-stars will be even more favorable for future large direct imaging surveys (Crepp & Johnson 2011) with dedicated upcoming instruments, such as Gemini/GPI (Macintosh et al. 2008) and VLT/SPHERE (Beuzit et al. 2008). To pursue the possibility of a higher frequency of massive planets around early-type stars, we have constructed a sample of 42 young A–F stars observed at high contrast to be sensitive to massive planets. We report new measurements for 39 stars, and we include 3 A-stars from the literature that have been observed in previous surveys. This survey intends to start defining the population statistics of massive planets and brown dwarfs (BDs) at orbital separations in the tens of AUs from their parent A-stars. This sample is an initial subset of the International Deep Planet Survey (IDPS) spanning M- to B-stars, the results of which will be presented in a forthcoming publication (Galicher et al. 2012, in preparation).

The selection of the target sample is explained in Sect. 2. In Sect. 3 the observing strategy, observations and data reduction steps are detailed. The detection limits of the survey are derived in Sect. 4, and we describe the identification of the candidate companions detected in the data. In Sect. 5 we use the detection limits to perform a statistical analysis of the survey, from which we derive a first estimation of the planetary systems frequency at large orbital radii around A-stars. Finally, we discuss our results and conclusions in Sect. 6.

2 Sample selection

Figure 1: vs. color-magnitude diagram of the A-stars of our target sample. The stars with published ages are plotted with a green circle. Stars without a published age are noted with blue triangles and have been assigned an age of 125 Myr, based on their positions within the region defined by the Pleiades A-stars, which are indicated with orange stars. An empirical linear fit to the Pleiades is also shown. The older population of 625 Myr Hyades A-stars is plotted with blue squares.
Figure 2: Age, distance and spectral type of all the stars in our sample. Table 1 gives the individual properties of each target. The size and color of the symbols correspond to the spectral type of each target: bluer colors for earlier spectral types and redder colors for later spectral types. The points grouped at 125 Myr correspond to stars without a published age, as described in Sect. 2.

The sample includes all A-stars previously observed in high contrast AO imaging surveys sensitive to planets (Nielsen & Close 2010; Chauvin et al. 2010; Janson et al. 2011) combined with the 35 A-stars and 4 F-stars observed for this study, resulting in a total of 42 stars, with 38 A-stars and 4 F-stars. Of the nine A-stars included in previous AO surveys, six were part of our selection (see below); they were re-observed and the new observations provided deeper limits. We used the existing surveys and the new observations to define the target set listed in Table 1, and used the best limit for each target. The sample A-stars span the full A0–A9 spectral type range. Because Fomalhaut was not observed as part of an AO planet search survey, Fomalhaut b (Kalas et al. 2008) was not included in the sample or the subsequent statistical analysis.

In order to define a representative sample of young A-stars, the A-stars newly observed for this study were selected to have positions on the color-magnitude diagram within or below the band associated with A-stars in the Pleiades. For the A-stars in the sample, the positions of the stars on the color-magnitude diagram are shown in Fig. 1. The target ages directly impact the minimum detectable mass, given the monotonic decline in planet brightness with age (e.g. Fortney et al. 2008; Chabrier et al. 2000). The majority of the A-stars – 22 targets – have age estimates from the literature combining several techniques (Rhee et al. 2007; Moór et al. 2011; Zuckerman et al. 2011; Janson et al. 2011; Moór et al. 2006; Chauvin et al. 2010; Zuckerman et al. 2001; Tetzlaff et al. 2011; Stauffer et al. 1995). These values are reported in Table 1 and used in the analysis of mass detection limits. Of the 22 A-stars with literature ages, 10 are members of nearby moving groups or associations of young stars. For the remaining 17 A-stars, the age is based on a comparison of the position of the targets and the best-fit to the dereddened Pleiades A-stars, a population of 125 Myr (Stauffer et al. 1998) A-stars. The individual reddening values were taken from (Breger 1986), and Pleiades A-stars on the photometric binary sequence were excluded from the fit. As shown in Fig. 1, the targets without literature ages are located generally very close to the Pleiades fit and within the range of Pleiades members, so an age of 125 Myr is adopted for these 17 stars. To ensure accurate placement on the color-magnitude diagram the A-stars were limited to stars with Hipparcos parallax uncertainty of 5% (Perryman et al. 1997). A distance cutoff was also imposed to probe angular separations corresponding to orbital separations similar to the outer regions of the Solar System and within the radii of protoplanetary disks imaged in scattered light and mm emission (e.g. Fukagawa et al. 2004; Mannings & Sargent 1997) around Herbig Ae stars, the precursors to massive stars. Both debris disk systems and stars without detected excess emission were included. Of the 38 A-stars in the sample, 14 have excess emission from debris disks, as noted in Table 1.

In addition to the A-stars, 4 early F-stars with F0–F5 spectral types were observed. Since F-stars do not evolve off the Main Sequence as rapidly as A-stars, the color-magnitude diagram is not as a reliable tool for age assessments of young F-stars. Consequently, the F-stars were limited to stars with age estimates from the literature based on space motion linked to nearby moving groups or associations (Moór et al. 2011; Zuckerman et al. 2001; Rhee et al. 2007). Of the 4 F-stars, 2 are debris disk systems. A graphical summary of the sample population incorporating spectral type, age, and distance is given in Fig. 2. In summary, the sample consists of a total of 42 stars, with 38 A-stars and 4 F-stars with a median age of 100 Myr and a median distance of 50 pc; key parameters of the sample are listed in Table 1. All members of the sample have no catalogued visual binary within 5 to ensure a detection limit that is uniform with position angle. The targets with wide (5) common proper motion companions from the WDS catalogue (Mason et al. 2001) or close (0.1) speckle companions (Mason et al. 1999) are noted in Table 1.

3 Observations and data reduction

Data were acquired between 2007 and 2012 with VLT/NaCo and Gemini/NIRI, two instruments located at the focus of 8 m telescopes. NaCo is installed at the ESO Very Large Telescope in Chile. It is the combination of the NAOS adaptive optics system (Rousset et al. 2003) and the CONICA observing camera (Lenzen et al. 2003). It provides diffraction-limited images in the near-infrared on a Aladdin InSb detector array. Our data were acquired using the S13 camera, which provides a spatial sampling of 13 mas/pixel (better than Nyquist) and a field of view (FoV) of . The targets were all observed with the broadband filter (2.00–2.35 m).

NIRI (Hodapp et al. 2003) is installed at the Gemini North telescope in Hawaii. It was used in combination with the ALTAIR AO system (Herriot et al. 2000) to obtain diffraction-limited images on a Aladdin InSb detector array. The f/32 camera provides a sampling of 22 mas/pixel and a FoV of . To improve the Strehl ratio and provide sharper image quality at large distance from the star, ALTAIR was used with the field lens inside the optical path. The targets were observed either with the CH4-short filter (CH4s, 1.54–1.65 m) or the filter (1.95–2.30 m).

Both instruments were used in pupil-stabilized mode (pupil-tracking mode on NaCo) to reduce the number of independently rotating speckle patterns, improve the stability of the point-spread function (PSF), and provide rotation of the FoV, which is then used for proper implementation of angular differential imaging (ADI; Marois et al. 2006). To optimize the detection of faint companions, the science images were saturated by using detector integration times (DIT) between 15 and 30 s in most cases. These DITs allow the detections to be background-limited at large separation, but since our targets are very bright (), the regions up to 0.4–0.8 are in general completely saturated and have been masked in the analysis. Each observing sequence started with the acquisition of an unsaturated PSF at five dither positions, obtained with the ND_short neutral density in NaCo (transmission of %) or a short DIT combined with narrow-band filters in NIRI (-continuum in -band or H in -band). These PSFs were then followed by 100–250 deeply saturated science images. For some observations, sky frames were acquired prior to the science frames following a dithering pattern. Due to restrictions on detector saturation, second epoch data on NaCo were acquired with a much smaller integration time and a larger number of exposures. Some NaCo data was also acquired with a much smaller number of exposures but using co-addition of 5 exposures. All the observations are summarized in Table 2.

HIP HD HR Distance Sp. Type Age Age ref. MG/assoc. IR excess IR excess ref. Binary
(J2000.0) (J2000.0) (pc) (Myr) (mag) (mag) (mag) (mag)
Observed targets
7345 9672 451 01 34 37.8 15 40 34.9 59.4 A1V 20 1 0.07 5.62 5.53 5.46 yes 1
10670 14055 664 02 17 18.9 33 50 49.9 34.4 A1V 100 1 0.02 4.03 3.86 3.96 yes 1
11360 15115 02 26 16.2 06 17 33.2 45.2 F2 12 2 Pictoris 0.40 6.79 5.86 5.98 yes 1
12413 16754 789 02 39 48.0 42 53 30.0 35.7 A1V 30 3 Tuc-Hor 0.06 4.74 4.62 4.46
13141 17848 852 02 49 01.5 62 48 23.5 50.5 A2V 100 1 0.10 5.25 5.16 4.97
14551 19545 943 03 07 50.8 27 49 52.1 54.6 A5V 30 3 Tuc-Hor 0.17 6.18 5.85 5.77
15648 20677 1002 03 21 26.6 43 19 46.7 46.2 A3V 125 11 0.05 4.96 4.86 4.78
16449 21997 1082 03 31 53.6 25 36 50.9 71.9 A3IV/V 30 3 Tuc-Hor 0.12 6.38 6.12 6.10 yes 1
22192 30422 1525 04 46 25.8 28 05 14.8 56.2 A3 125 11 0.19 6.18 5.73 5.72 yes 4
22226 30447 04 46 49.5 26 18 08.8 80.3 F3V 30 2 Columba 0.39 7.85 6.95 6.89
23296 32115 1613 05 00 39.8 02 03 57.7 49.6 A8 125 11 0.29 6.31 5.62 5.58
26309 37286 1915 05 36 10.3 28 42 28.9 52.8 A2 30 3 Tuc-Hor 0.15 6.26 5.94 5.86 yes 3
26624 37594 1940 05 39 31.2 03 33 52.9 42.6 A8V 125 11 0.29 5.99 5.36 5.21
32938 50445 2558 06 51 42.4 36 13 49.0 55.2 A3V 125 11 0.18 5.94 5.54 5.51
34782 55568 2720 07 12 04.1 30 49 16.9 47.6 A8V 125 11 0.27 6.10 5.46 5.39
35567 57969 07 20 23.0 56 17 40.7 71.0 A1V 20 9 0.11 6.57 6.39 6.28
41152 70313 3277 08 23 48.5 53 13 11.0 50.4 A3V 200 1 0.13 5.52 5.29 5.25 yes 1
41307 71155 3314 08 25 39.6 03 54 23.1 37.5 A0V 400 5 -0.01 3.91 4.09 4.08 yes 1
42334 73495 3420 08 37 52.2 26 15 18.0 71.1 A0V 125 11 -0.03 5.24 5.35 5.32
44923 78702 3638 09 09 04.2 18 19 42.8 83.7 A0/A1V 200 6 0.00 5.73 5.64 5.69
53771 95429 4296 11 00 08.3 51 49 04.1 61.1 A3 125 11 0.19 6.15 5.79 5.77 yes/10
57013 101615 4502 11 41 19.8 43 05 44.4 65.5 A0V 125 11 0.04 5.54 5.51 5.44
57328 102124 4515 11 45 17.0 08 15 29.2 37.4 A4V 125 11 0.17 4.84 4.54 4.41
60595 108107 4722 12 25 11.8 11 36 38.1 70.5 A1V 125 11 0.04 5.95 5.91 5.83
61468 109536 4794 12 35 45.5 41 01 19.0 35.5 A7V 100 7 0.22 5.12 4.71 4.57
61960 110411 4828 12 41 53.1 10 14 08.3 36.3 A0V 100 1 0.08 4.88 4.76 4.68 yes 1
62983 112131 4901 12 54 18.7 11 38 54.9 68.5 A2V 125 11 0.07 6.00 5.85 5.83 yes/0.04
66634 119024 5142 13 39 30.4 52 55 16.4 53.6 A3V 125 11 0.11 5.46 5.20 5.17
69713 125161 5350 14 16 09.9 51 22 02.0 29.1 A9V 125 11 0.24 4.75 4.32 4.29 yes/32
69732 125162 5351 14 16 23.0 46 05 17.9 30.4 A0 200 1 0.09 4.18 4.03 3.91 yes 1
78078 142703 5930 15 56 33.4 14 49 46.0 51.1 A2 125 11 0.24 6.11 5.39 5.34
92024 172555 7012 18 45 26.9 64 52 16.5 28.5 A7V 12 8 Pictoris 0.20 4.78 4.25 4.30 yes 1 yes/72
95261 181296 7329 19 22 51.2 54 25 26.2 48.2 A0V 12 8 Pictoris 0.02 5.02 5.15 5.01
99273 191089 20 09 05.2 26 13 26.5 52.2 F5V 12 2 Pictoris 0.48 7.18 6.12 6.08 yes 1
104365 201184 8087 21 08 33.6 21 11 37.2 55.1 A0V 125 11 0.00 5.30 5.33 5.31 yes/70
110935 212728 8547 22 28 37.7 67 29 20.6 43.1 A4V 125 11 0.21 5.56 5.14 5.05
114189 218396 8799 23 07 28.7 21 08 03.3 39.4 A5V 30 3 Columba 0.26 5.97 5.28 5.24 yes 1
115738 220825 8911 23 26 56.0 01 15 20.2 47.1 A0 70 3 AB Dor 0.04 4.95 4.95 4.93
116431 221853 23 35 36.2 08 22 57.4 68.4 F0 100 1 0.41 7.35 6.44 6.40
Additional targets included from previous surveys or discoveries.
27321Source: Bonnefoy et al. (2011). Data from ESO program 284.C-5057, PI A.-M. Lagrange. The data were drawn from the ESO archive and analyzed following the procedure described in Sect. 3. 39060 2020 05 47 17.09 51 03 59.44 19.4 A6V 12 8 Pictoris 0.17 3.86 3.54 3.53 yes 1
61498Source: Chauvin et al. (2010). Detection limits provided by G. Chauvin (private communication). 109573 4796 12 36 01.03 39 52 10.23 72.8 A0 8 10 TWA 0.00 5.78 5.79 5.77 yes 1 yes/8
98495Source: Chauvin et al. (2010). Detection limits provided by G. Chauvin (private communication). 188228 7590 20 00 35.56 72 54 37.82 32.2 A0V 50 7 -0.03 3.95 3.76 3.80
111 \tablebib

(1) Rhee et al. (2007); (2) Moór et al. (2011); (3) Zuckerman et al. (2011); (4) Su et al. (2006); (5) Janson et al. (2011); (6) Moór et al. (2006); (7) Chauvin et al. (2010); (8) Zuckerman et al. (2001); (9) Tetzlaff et al. (2011); (10) Stauffer et al. (1995); (11) see Sect. 2.

Table 1: Target sample and properties.

The data reduction followed standard procedures. For the unsaturated dithered PSF frames, a sky frame was constructed by taking the median of the images. The individual images were then sky subtracted and divided by the flat field before being registered to a common center. The final unsaturated PSF was obtained by median combining the five images. This PSF was used for the photometric calibration and calculation of the detection limits. For the saturated images, a sky frame was subtracted when available, and each image was divided by the flat field. Bad pixels identified from the flat field were replaced by the median of adjacent pixels using a custom IDL routine. Field distortion was taken into account for NIRI since it reaches several pixels in the outer part of the field: all images were corrected for this effect using a custom IDL routine and an optimized distortion solution that will be presented in Galicher et al. (2012, in preparation). For NaCo, the field distortion was found to be negligible ( mas) with the S13 camera in the inner part of the detector (Trippe et al. 2008; Fritz et al. 2010). The center of each saturated image was estimated using the fitting of a Moffat profile (Moffat 1969) on the PSF wings. The saturated region of the images was assigned a zero-weight in the fit to avoid any bias. For NIRI, the center of the saturated images moved by up to 5 pixel on the detector over the full observing sequence. For NaCo, a well-documented problem with the pupil-tracking mode (corrected in October 2011) induced a slow drift of the star during some of the observations. This drift, which could reach several pixels per minute, is proportional to the star elevation. To avoid any important drift, the observing strategy included frequent recentering of the star during the observations and the use short DITs (during the follow-up phase of the survey).

In each observing sequence, a frame selection was performed to reject the worst frames where a sudden increase of seeing or an interruption of the AO loop occurred. The frame selection was based on three parameters: the position of the PSF center in each frame compared to previous and subsequent frames, the overall flux in the frame and a measure of the fraction of flux contained in an annulus covering an area just outside of the saturated region. Frames for which a sudden change of the PSF position or a drop in intensity are observed generally correspond to frames where the AO loop is open, and are thus of poor quality. These frames were inspected visually and in most cases removed from the observing sequence. After the frame selection process, the sequence of images usually contained frames of equivalent quality in terms of AO correction.

For astrometric calibration of the NIRI data, we used archive data for the  Ori C field observed on November 30 2008 with the f/32 camera, filter and without the ALTAIR field lens. We used the coordinates reported by McCaughrean & Stauffer (1994) to determine the mean pixel scale and true North orientation of the NIRI detector. The mean pixel scale and true North orientation measured were respectively  mas/pixel and  deg. For NaCo observations, we used calibration data collected within our observing programs and within the NaCo Large Program collaboration (NaCo-LP222ESO program 184.C-0567, PI J.-L. Beuzit, “Probing the Occurrence of Exoplanets and Brown Dwarfs at Wide Orbits”.; Chauvin 2010). The calibrators used were the  Ori C field and IDS 13022+0107, when the former was not observable. The calibration was performed with the S13 camera and the filter at several epochs between November 2009 and January 2012. The mean pixel scale measured was  mas/pixel, and the true North orientation varied from  deg to  deg between the different epochs. The NaCo PSFs are usually not perfectly centro-symmetric, causing the centers determined on unsaturated and saturated PSFs to be slightly different. The effect has been estimated to 6 mas and constitutes the dominant astrometric error. It was estimated using a dedicated calibration where saturated and non-saturated PSFs were taken alternatively, and their center determined using Moffat or Gaussian profile fitting.

Name Date Instrument Filter Num. of exposures DIT NDIT T FoV rotation Median seeing
(s) (min) (deg) (as)
Observed targets
HIP~7345 2007-09-12 NIRI CH4s 140 30.0 1 70 38.9 0.4
HIP~10670 2007-09-16 NIRI CH4s 140 30.0 1 70 87.5 0.5
2008-10-12 NIRI 25 30.0 1 13 7.5 0.5
HIP~11360 2007-12-30 NIRI CH4s 120 30.0 1 60 68.2 0.5
2008-10-12 NIRI 15 30.0 1 8 6.2 0.4
HIP~12413 2009-12-20 NaCo 125 20.0 1 42 26.0 1.1
2011-11-07 NaCo 2200 1.2 1 44 33.3 1.0
HIP~13141 2009-12-21 NaCo 125 20.0 1 42 15.7 1.0
HIP~14551 2009-12-21 NaCo 125 20.0 1 42 10.1 1.2
HIP~15648 2008-10-17 NIRI 128 30.0 1 64 51.2 0.7
HIP~16449 2007-09-22 NIRI CH4s 98 30.0 1 49 32.8 0.9
2008-10-12 NIRI 15 30.0 1 8 2.6 0.4
HIP~22192 2009-12-20 NaCo 125 20.0 1 42 22.4 1.4
HIP~22226 2008-01-15 NIRI CH4s 120 30.0 1 60 33.4 0.6
2008-11-18 NIRI 26 30.0 1 13 4.8 0.8
HIP~23296 2010-01-03 NIRI 252 15.0 1 63 56.7 0.5
HIP~26309 2009-12-21 NaCo 125 20.0 1 42 102.1 0.8
HIP~26624 2008-11-14 NIRI 160 30.0 1 80 55.8 0.4
HIP~32938 2009-12-20 NaCo 125 20.0 1 42 45.9 0.8
HIP~34782 2009-12-21 NaCo 125 20.0 1 42 77.4 0.8
2012-01-11 NaCo 1000 2.8 1 47 77.9 0.9
HIP~35567 2009-12-20 NaCo 100 20.0 1 33 17.7 1.4
2011-12-18 NaCo 360 7.5 1 45 20.9 0.8
HIP~41152 2010-01-03 NIRI 232 15.0 1 58 38.2 0.6
HIP~41307 2009-12-21 NaCo 100 20.0 1 33 17.9 1.1
HIP~42334 2009-12-20 NaCo 125 20.0 1 42 19.6 1.4
2011-04-27 NaCo 530 2.5 1 22 5.5 1.1
HIP~44923 2008-03-23 NIRI CH4s 120 30.0 1 60 40.8 0.5
HIP~53771 2010-03-06 NaCo 25 20.0 5 42 19.6 1.0
2012-01-12 NaCo 880 4.0 1 59 29.1 0.8
HIP~57013 2010-03-07 NaCo 25 20.0 5 42 29.2 0.6
HIP~57328 2008-03-24 NIRI CH4s 130 30.0 1 65 97.8 0.4
HIP~60595 2010-03-07 NaCo 25 20.0 5 42 30.1 0.5
HIP~61468 2009-06-25 NaCo 125 20.0 1 42 33.4 1.0
HIP~61960 2008-03-21 NIRI CH4s 122 30.0 1 61 99.8 0.4
HIP~62983 2010-03-06 NaCo 25 20.0 5 42 37.8 0.9
HIP~66634 2009-02-03 NIRI 42 30.0 1 21 10.7 0.7
HIP~69713 2008-03-21 NIRI CH4s 140 30.0 1 70 51.2 0.5
HIP~69732 2009-02-02 NIRI 88 30.0 1 44 33.9 0.5
HIP~78078 2009-06-24 NaCo 100 20.0 1 33 40.7 1.4
HIP~92024 2009-06-25 NaCo 230 20.0 1 77 32.0 1.1
HR~7329 2008-08-07 NaCo 27 25.0 2 23 11.0 0.5
HIP~99273 2007-08-10 NIRI CH4s 140 30.0 1 70 34.4 0.5
2008-09-04 NIRI CH4s 130 30.0 1 65 32.2 0.5
HIP~104365 2007-09-18 NIRI CH4s 140 30.0 1 70 45.4 0.5
2008-09-09 NIRI CH4s 20 30.0 1 10 5.2 0.8
HIP~110935 2009-06-24 NaCo 125 20.0 1 42 14.4 1.6
HR~8799 2007-10-17 NIRI CH4s 120 30.0 1 60 172.3 0.4
2008-09-01 NIRI CH4s 70 30.0 1 35 152.4 0.6
2008-09-05 NIRI 60 30.0 1 30 139.1 0.3
2008-10-10 NIRI CH4s 120 30.0 1 60 171.6 0.6
2008-10-14 NIRI 60 12.0 1 12 131.2 0.4
HIP~115738 2008-10-15 NIRI CH4s 120 30.0 1 60 69.4 0.4
HIP~116431 2007-09-12 NIRI CH4s 140 30.0 1 70 72.3 0.4
2008-09-24 NIRI CH4s 60 30.0 1 30 54.9 0.7
Targets from previous imaging surveys
$β$~Pictoris 2010-04-10 NaCo 160 0.15 150 60 21.0 0.8
HIP~61498 2003-06-06 NaCo H 12 3.0 25 15 0.7
HIP~98495 2004-04-27 NaCo H 3.0 20 0.6
Table 2: Summary of the observations.
Figure 3: Example of final image after data reduction and processing with LOCI for target HIP 10670. The central white region corresponds to the saturated area in the science images, which has been masked in the data analysis. The point source 4.7 West of the star is a background object 14.6 mag fainter than the central star.

Finally, the cleaned registered ADI images were processed with an implementation of the LOCI algorithm (Lafrenière et al. 2007b). All the targets were analyzed using a similar set of parameters:  FWHM, for NaCo and for NIRI. For each target, partial subtraction induced by LOCI was measured using fake planets introduced into the data at several position angles and angular separations, in a way similar to that described by Lafrenière et al. (2007b). Temporal smearing induced by the FoV rotation was also measured using fake planets. A typical final reduced image is visible in Fig. 3

4 Results

4.1 Identification of companion candidates

Detection of the candidate companions (CCs) was performed by visual inspection of the final images after LOCI processing and of the signal-to-noise ratio (SNR) maps. The SNR maps were obtained by normalizing each pixel in the images to the noise measured in an annulus of width at the same separation as the pixel from the star. The maps were used to identify CC with a detection level above .

Relative astrometry for each CC was obtained using 2D Gaussian fitting, and the position was refined by maximizing the flux in a aperture moved on a  pixel grid by steps of  pixel around the position found with the Gaussian fitting. The main astrometric error for the NaCo data is the 0.5 pixel uncertainty on the actual center of the PSF in saturated data (see Sect. 3). Other sources of error come from the shift induced by the use of the neutral density (0.1 pixel) or the 2D Gaussian fitting (usually 0.1 pixel). In NIRI data, we assume the dominant error term to be the residual distortion (6 mas) close to the star. At separations larger than 5, the distortion correction improves significantly the astrometric measurements for the CCs but not enough to reach a similar accuracy. For these candidates we have assumed a residual error of 1 pixel (21.4 mas). Although there are some uncertainties induced by the distortion in the NIRI field of view, the precision was sufficient to clearly identify background or comoving objects. An updated distortion map for NIRI will be presented in Galicher et al. (2012, in preparation).

Relative photometry was estimated by aperture photometry in aperture centered on the best estimation of the position. Contrast difference with respect to the star was obtained by dividing the flux of each CC by the flux of the non-saturated PSF measured in an aperture of the same diameter, scaled by the ratio of exposure time and the transmission of the neutral density (NaCo data) or the correction due to the fact that unsaturated observations were not performed in the same filters as saturated observations (NIRI data). This correction factor for NIRI was calculated using slightly saturated images acquired regularly during the ADI sequence with the same filter as the deeply saturated science images. The non-saturated PSF was normalized to these slightly saturated images using the integrated flux in an annulus outside of the saturated part of the images (usually between 10 and 20 pixels of radius). A calibration performed on data saturated and unsaturated with the same filters indicated an error on the flux normalization of 4%. The flux loss induced by LOCI partial subtraction and by the temporal smearing of the CC PSF by field rotation was also included as a correction term. Finally, the photometric error was estimated by including a measure of the local SNR at the separation of each CC and the error on the calibration of the flux normalization between the saturated and unsaturated PSF.

For newly detected stellar companions (see Sect. 4.3.1), where a more robust estimation of the relative astrometry and photometry is useful for the determination of their physical parameters, a different approach was taken: instead of estimating the position and signal level of the CC in the final reduced image, we implemented a routine to subtract a “negative CC” in each raw frames before performing the ADI optimization in a way similar to that described in Marois et al. (2010a). The routine was run multiple times for various insertion position of the negative CC in order to minimize the residual noise in the final image at the expected position of the CC and in the surrounding area. When a position minimizing the residuals was found, the routine was used again for different injection flux for the CC, again trying to minimize the residual noise in the final image. Errors were estimated using variations of the best injection point and flux obtained by measuring the residuals in areas of different sizes around the position of the CC.

A total of 50 CCs were identified above the 5 level around 22 of the observed targets. Targets with CCs at a projected physical separation less than or equal to 320 AU were followed-up by subsequent observations to verify if they are co-moving with the target star. This cut-off was chosen according to the widest likely known companion at the time of the follow-up, 1RXS J160929.1-210524 b, which is at a projected separation of 330 AU (Lafrenière et al. 2008). There is only one target for which no second epoch was obtained although there was one CC within 320 AU: HIP 41307 had a CC at a projected separation of 217 AU but it was confirmed as a background star by Janson et al. (2011). For targets with follow-up observations, the position of the candidates was checked for consistency with either a background or a co-moving object based on the target parallax and proper motion, also taking into account uncertainties on these particular values. Similarly to Chauvin et al. (2010) and Ehrenreich et al. (2010), we performed a probability test on the difference in position and with respect to the star at two epochs. These probability tests along with the proper motion plots in RA/DEC or in separation/position angle for each CC allowed attributing a status of background, comoving or ambiguous to each of them. The properties of all detected CCs are summarised in Appendix A.

4.2 Detection limits

Figure 4: Examples of -band 5 detection limits representing the best (HIP 69732) and worst (HIP 15648) limits, as well as an intermediate case (HIP 26624).The right axis shows the detection limit in masses of Jupiter for the median target of our survey, a 100 Myr-old A3V star at 50 pc, converted using the COND evolutionary models (Baraffe et al. 2003).
Name 0.50 0.60 0.75 1.00 2.00 3.00 4.00 5.00 7.50 10.00
Observed targets
HIP 7345 10. 6 11. 5 12. 3 13. 3 15. 4 16. 3 16. 8 16. 9 17. 1 17. 1
HIP 10670 13. 5 15. 8 16. 7 17. 2 17. 6 18. 0 18. 1
HIP 11360 10. 6 11. 6 12. 2 13. 1 15. 1 16. 0 16. 4 16. 6 16. 7 16. 7
HIP 12413 8. 9 10. 0 11. 4 12. 6 14. 2 14. 5 14. 6 14. 6 13. 6
HIP 13141 9. 2 10. 0 11. 3 12. 7 14. 6 15. 1 15. 3 15. 4 14. 8 14. 4
HIP 14551 8. 5 9. 3 10. 8 12. 3 14. 3 14. 8 14. 9 15. 0 14. 4 14. 2
HIP 15648 9. 1 10. 1 10. 9 12. 3 12. 7 12. 8 12. 8 12. 9 12. 8
HIP 16449 8. 0 8. 9 9. 6 10. 6 12. 5 12. 9 13. 0 13. 1 13. 2 13. 2
HIP 22192 9. 5 10. 2 11. 5 12. 6 14. 2 14. 4 14. 5 14. 5 13. 8 13. 6
HIP 22226 9. 8 10. 7 11. 4 12. 3 14. 3 15. 0 15. 2 15. 4 15. 4 15. 4
HIP 23296 11. 0 11. 9 13. 2 15. 4 16. 4 16. 7 16. 9 16. 9 16. 9
HIP 26309 10. 0 10. 8 11. 9 12. 6 14. 1 14. 4 14. 4 14. 3 13. 2 12. 5
HIP 26624 11. 8 14. 2 15. 3 15. 7 15. 8 15. 9 15. 9
HIP 32938 10. 7 11. 6 12. 5 13. 3 14. 4 14. 5 14. 6 14. 6 13. 7 12. 8
HIP 34782 10. 9 11. 8 12. 7 13. 8 15. 0 15. 3 15. 3 15. 3 14. 3 13. 4
HIP 35567 9. 5 10. 9 12. 2 13. 3 14. 2 14. 4 14. 3 14. 4 13. 5
HIP 41152 11. 2 12. 8 15. 0 16. 2 16. 6 16. 8 17. 0 17. 0
HIP 41307 10. 5 11. 7 13. 1 15. 4 16. 1 16. 4 16. 5 16. 1 15. 7
HIP 42334 9. 7 10. 7 11. 5 12. 7 14. 4 14. 7 14. 9 14. 9 14. 2 13. 9
HIP 44923 10. 4 11. 1 11. 8 14. 2 15. 4 16. 0 16. 2 16. 5 16. 5
HIP 53771 10. 0 11. 2 12. 5 13. 6 14. 7 14. 8 14. 8 14. 8 14. 0
HIP 57013 8. 7 9. 7 10. 8 11. 8 13. 8 14. 4 14. 6 14. 7 14. 4 14. 0
HIP 57328 13. 5 15. 5 16. 7 17. 4 17. 7 18. 1 18. 2
HIP 60595 8. 9 9. 6 10. 7 11. 8 13. 8 14. 3 14. 6 14. 7 14. 0 13. 5
HIP 61468 10. 1 11. 2 12. 4 14. 5 15. 1 15. 2 15. 3 14. 4 15. 7
HIP 61960 12. 7 14. 8 16. 0 16. 7 17. 0 17. 4 17. 4
HIP 62983 8. 6 9. 4 10. 5 11. 6 13. 5 14. 0 14. 1 14. 2 13. 4 13. 3
HIP 66634 9. 1 10. 3 11. 3 13. 6 14. 5 14. 7 14. 8 14. 8 14. 8
HIP 69713 12. 4 14. 7 15. 8 16. 6 16. 9 17. 4 17. 5
HIP 69732 16. 3 17. 4 17. 9 18. 2 18. 4 18. 5
HIP 78078 7. 9 9. 0 10. 2 11. 2 12. 8 13. 2 13. 3 13. 3 12. 3
HIP 92024 10. 8 11. 9 13. 9 14. 5 14. 6 14. 7 9. 7
HR 7329 11. 2 12. 1 13. 1 15. 1 16. 0 16. 5 16. 7
HIP 99273 10. 2 11. 3 12. 2 13. 4 15. 4 16. 1 16. 4 16. 6 16. 7 16. 7
HIP 104365 11. 5 12. 4 13. 6 15. 5 16. 3 16. 9 17. 1 17. 3 17. 4
HIP 110935 8. 5 9. 0 9. 9 11. 2 13. 1 13. 5 13. 6 13. 7 13. 2
HR 8799 12. 0 12. 7 13. 6 15. 6 16. 4 16. 8 17. 0 17. 1 17. 0
HIP 115738 12. 0 12. 9 13. 5 15. 6 16. 6 17. 1 17. 3 17. 5 17. 5
HIP 116431 10. 7 12. 0 12. 9 13. 9 15. 8 16. 3 16. 5 16. 6 16. 7 16. 6
Targets from previous imaging surveys
 Pictoris 10. 3 11. 2 12. 0 13. 5 15. 2 15. 3
HIP 61498 9. 5 10. 1 10. 4 11. 1 13. 4 13. 6 14. 0 14. 1
HIP 98495 10. 6 10. 6 10. 7 13. 2 14. 7 14. 6 14. 6 14. 9
333Values are given in magnitude difference in the filter of the observations: CH4s or for NIRI observations, for NaCo observations.
Table 3: Best 5 detection limit for each target of the sample.

The ultimate contrast reached by our observations is quantified using 5 detection limits obtained by measuring at each angular separation the residual noise in an annulus of width , normalized by the flux of the unsaturated PSF. As mentioned previously, two correction factors were applied to the detection limits. The first one accounts for the flux loss of point sources induced by the LOCI processing. It is calculated by introducing fake companions of known flux at various separation and position angle into the raw data, fully processing the data with LOCI, and measuring the flux of the companions in the final image (see Lafrenière et al. 2007b). This process was repeated 10 times with the companions at various azimuthal positions to obtain the average attenuation as a function of angular separation from the star. The second correction factor takes into account the smearing of the PSF of an object in the field induced by the field rotation during the exposures. The dilution factor was calculated by simulating the smearing of point sources at various separations for every frame. Finally, the detection limits were corrected by dividing them at all separations by the appropriate attenuation for the flux loss and dilution for the smearing.

The best magnitude difference detection limits for angular separations ranging from 0.5 to 10.0 are listed in Table 3. Example limits representative of good, median and poor contrast performance in -band are plotted in Fig. 4. The observations are background-limited outside 5 for all targets. The background limits range between 19 and 23 mag for NIRI in -band, between 17.5 and 22.5 mag for NIRI in -band, and between 16.5 and 20.5 mag for NaCo in -band. These values are within the typical ranges for observations in good conditions with NIRI and NaCo in 0.5 to 1 hour of integration time.

A common concern for the estimation of detection limits in high-contrast data is that of the detection threshold. It is known that speckle noise does not follow Gaussian statistics (Goodman 1968; Fitzgerald & Graham 2006). However, Marois et al. (2008a) have performed a detailed analysis of NIRI data processed with ADI (Marois et al. 2006) and demonstrated that the residual noise follows quasi-Gaussian statistics when a sufficient number of images (i.e. of FoV rotation) are combined. For most targets444The second epochs with NIRI have usually been obtained with much shorter integration times than the first epochs, providing a much smaller FoV rotation. They were only intended to confirm the CCs detected in the first epoch images., we obtained a FoV rotation of at least 20 deg (Table 2) over several tens of images, which should be sufficient to reach quasi-Gaussian noise statistics after LOCI processing. For comparison, previous surveys using some form of ADI observations have adopted 5 or 6 thresholds to derive their detection limits, either assuming Gaussian statistics for the residual noise (Biller et al. 2007; Lafrenière et al. 2007a; Leconte et al. 2010; Janson et al. 2011) or using a higher threshold to account for the non-Gaussianity of the noise (Kasper et al. 2007).

The 5 detection limits were finally converted to absolute magnitudes using the Hipparcos parallax and 2MASS or apparent magnitudes. Since observations were obtained in a filter set different from the 2MASS filters (in particular the CH4s NIRI filter), we checked that for the different spectral types of our targets (from A0V to F5V) there was no significant change in apparent magnitude between the 2MASS filters and the NaCo or NIRI filters. The correction terms were found to be completely negligible ( mag) compared to the other systematic uncertainties in the photometry.

4.3 Companion detections

Figure 5: Offset position (black circles) with uncertainties measured relative to the primary star for the co-moving stellar companions to HIP 42334 (left) and HIP 104365 (right). The variation expected for a background stationary object is also plotted (grey line) with the expected positions of background objects (grey square).

Within the 42-star sample, 5 stars have newly identified or previously reported physically associated companions, and we briefly discuss each case of a stellar or substellar companion in this section. Of the companion candidates with follow-up newly reported in this study, three are confirmed as co-moving, and all but one of the remaining candidates have a 0.01% probability to be comoving. The possible residuals on astrometry are well below the threshold that would change the assessment from background to co-moving. In one case, one of candidate companions to HIP 10670, it cannot be excluded that this object has a proper motion of its own, making it inconsistent with both a stationary background object and a co-moving object. Because it shows a motion much larger than a pixel (13.19 mas on NaCo, 21.38 mas on NIRI) between the two epochs, suggesting a significant motion not compatible with a co-moving object, it is treated as a background object in the statistics.

4.3.1 Stellar companions

Primary name Separation PA CH4s MassThe mass range was estimated using the DUSTY evolutionary models (Chabrier et al. 2000) taking into account the uncertainties on the distance, on the magnitude of the primary, on the contrast of the companion, and on the age of the system.
(as) (deg) (mag) (mag) (M)
HIP 42334 0.07–0.11
HIP 104365 0.09–0.13
0.09–0.14
555
Table 4: Properties of the detected stellar companions.

We have identified three co-moving stellar companions around two stars in our data. Both stars were already known to be in wide multiple systems, but the companions we report were previously unknown due to their large contrast ratio with their parent star. Proper motion plots are given in Fig. 5, and relative positions and fluxes are summarized in Table 4.

HIP 42334 – The star is a 125 Myr old A0V star at 71.07 pc in a double system, with the secondary component at a separation of 20 and a magnitude of 13 in the optical (Mason et al. 2001). A point source  mag fainter in than the primary was detected at a separation of and position angle (PA) of . The star was re-observed because of the presence of other fainter point sources in the field, but only the brightest candidate proved to be co-moving (Fig. 5, left). Comparison to evolutionary models at 125 Myr (Chabrier et al. 2000) place this object at the bottom of the main sequence, consistent with an M6–M8 dwarf.

HIP 104365 – The star is 125 Myr old A0V star at 55.12 pc in a quadruple system (Mason et al. 2001). A very tight pair of companions was detected at a separation of 9.57 and a PA of 14.8°. After one year they both clearly showed a common proper motion with HIP 104365 (Fig. 5, right). Given their projected separation of 9.8 AU and their common proper motion, it is extremely likely that they are gravitationally bound together. Their contrast difference with the star in the CH4s filter makes them compatible with M5–M7 dwarfs.

4.3.2 Substellar companions

Name Orbital sep. Mass References
(AU) (M)
HR 8799 b 68Projected separation. 1,2,3,4
HR 8799 c 38Projected separation. 1,2,3,4
HR 8799 d 24Projected separation. 1,2,3,4
HR 8799 e 15Projected separation. 1,2,3,4
 Pictoris b 8–9 7–12 5,6,7,8,9
HR 7329 B 200Projected separation. 20–50 10
666\tablebib

(1) Marois et al. (2008a); (2) Marois et al. (2010b); (3) Currie et al. (2011); (4) Soummer et al. (2011); (5) Lagrange et al. (2009); (6) Lagrange et al. (2010); (7) Bonnefoy et al. (2011); (8) Lagrange et al. (2012a); (9) Chauvin et al. (2012); (10) Neuhäuser et al. (2011).

Table 5: Properties of the substellar companions from the sample.

Three targets in the sample – HR~8799, $β$~Pic, and HR~7329 – have previously reported substellar companions (Marois et al. 2008b, 2010b; Lagrange et al. 2009; Lowrance et al. 2000; Neuhäuser et al. 2011) including four planets and one brown dwarf. A summary of these objects and their properties is given in Table 5. We note that the observations of these three targets were not unusually sensitive and biased, but have detection limits similar to many other targets in the sample.

HR 8799 – This is the first imaged multiple planet system, with four known planetary-mass companions orbiting the star at 68, 38, 24 and 14.5 AU (Marois et al. 2008b, 2010b). It is an A5V star located at  pc from the Sun, classified as  Doradus and  Bootis. It shows a far-infrared excess emission (Vega-like) consistent with the presence of a warm dust disk within 6–15 AU, a massive cold dust disk within 90–300 AU and a dust halo extending up to 1000 AU (Rhee et al. 2007; Su et al. 2009). Recent estimations of the system age converge towards 30 Myr (Doyon et al. 2010; Zuckerman et al. 2011; Currie et al. 2011), with HR 8799 being a likely member of the Columba Association. This estimation leads to a mass of 5–10 M for HR 8799 cde, and 3.5–7 M for HR 8799 b. This system has been extensively studied in the literature since its discovery and we do not intend to make an exhaustive review of the numerous results. For detailed studies, see for example Bowler et al. (2010b), Currie et al. (2011), Barman et al. (2011a), Galicher et al. (2011) and Soummer et al. (2011).

 Pictoris – This  Myr-old (Zuckerman et al. 2001) A5V star at a distance of  pc (Crifo et al. 1997) has been well-known for two decades since the imaging of a debris disk (Smith & Terrile 1984) with several asymmetries and a warp at 80 AU (Kalas & Jewitt 1995; Mouillet et al. 1997), which could be explained by the presence of a giant planet within 50 AU of the star. Lagrange et al. (2009, 2010) confirmed the presence of a planetary-mass companion with an orbital separation compatible with formation by core-accretion. Using the COND evolutionary models and -band observations, Bonnefoy et al. (2011) infered a mass range of 7–10 M for the planet. Follow-up of the system with imaging (Lagrange et al. 2012a; Chauvin et al. 2012) and RV measurements (Lagrange et al. 2012b) recently allowed an accurate determination of the orbital parameters and the mass of the planet. Using Markov Chain Monte-Carlo methods, Chauvin et al. (2012) measure a probable range of  = 8–15 AU for the semimajor axis, with a low eccentricity  0.16 and a high inclination  = 88.5  1.7 deg. Using a dataset where the planet, the main disk and the warp are detected, (Lagrange et al. 2012a) also confirmed that the planet is located above the midplane of the main disk, making it aligned with the warp. These observations strongly support that the presence of the planet is responsible for the warped morphology of the disk. Finally, using 8 years of RV measurements on the star, Lagrange et al. (2012b) determine an upper limit of 10–12 M for the mass of the planet, assuming a semimajor axis of 8–9 AU.

HR 7329 – The substellar companion to the  Pic moving group member HR 7329 (Zuckerman et al. 2001) was discovered with coronagraphic imaging with the Hubble Space Telescope (Lowrance et al. 2000). It has a magnitude difference of F160W = and separation of (200 AU at 47.7 pc). It was further characterized with spectroscopic observations in the -band (Guenther et al. 2001). The most comprehensive compilation of astrometric and photometric measurements of this system and a high significance confirmation that the companion is physically associated with the primary is given in Neuhäuser et al. (2011). Like the HR 8799 and  Pic systems, HR 7329 has a debris disk (Smith et al. 2009).

5 Statistical analysis

5.1 Context

Using AO instruments on large ground-based telescopes, a number of deep imaging surveys have been carried out to search for massive planets around young nearby solar-type stars (Chauvin et al. 2003; Neuhäuser et al. 2003; Masciadri et al. 2005; Biller et al. 2007; Kasper et al. 2007; Lafrenière et al. 2007a; Chauvin et al. 2010; Delorme et al. 2012), and more recently around much more massive stars (Janson et al. 2011). Despite a large improvement of sensitivity to massive planets in the recent years, due largely to the development of advanced data analysis methods to suppress the speckle noise, none of these surveys have reported the detection of planetary objects potentially formed by core-accretion, i.e. within the circumstellar disk of the primary.

Taking advantage of their null results, the most recent large-scale surveys have placed constraints on the population of long-period massive planets. Using a combination of Monte-Carlo simulations and a Bayesian approach, it is possible to estimate the upper limit of the fraction of stars having substellar companions () within a certain range of orbital separation and mass. Nielsen & Close (2010) performed a detailed statistical analysis using the data for 118 stars from the surveys of Masciadri et al. (2005), Biller et al. (2007) and Lafrenière et al. (2007a), which allowed them to constrain the presence of planets down to 4 M around F–M stars. Assuming a flat distribution for the mass and semimajor axis of planets, they rule out at 95% confidence the presence of planets more massive than 4 M around more than 20% of stars in the range 80–280 AU using the cold-start evolutionary models (see Sect. 5.3 for details on the evolutionary models) and in the range 20–500 AU using the hot-start models, which is in agreement with the values previously reported by Lafrenière et al. (2007a). From the observations of 22 very young G–M targets in the Tucana and  Pic moving groups, Kasper et al. (2007) place an upper limit of 5% for planets above 2–3 M at separations larger than 30 AU. Using a sample of 88 young nearby stars, Chauvin et al. (2010) report with 95% confidence that less than 10% of stars can have planetary-mass companions beyond 40 AU, assuming mass and semimajor axis distributions similar to that reported by Cumming et al. (2008) from RV surveys. Finally, Janson et al. (2011) compared predictions of disk instability models to the observation of 15 B2–A0 stars, and they showed that less than 30% of massive stars can form and retain companions below 100 M within 300 AU at 99% confidence.

Previous surveys have also used their non-detections to investigate with what confidence planet distributions obtained from RV surveys can be extrapolated for planets at much larger orbital radii. This is in particular the case for the mass and period and semimajor axis distributions, which appear to be well fitted by simple power laws (Cumming et al. 2008) up to a certain semimajor axis cutoff. For the power law and planet frequency values reported by (Cumming et al. 2008), Nielsen & Close (2010) place the semimajor axis cutoff within 180 AU at 95% confidence with the cold-start models, and within 65 AU with the hot-start models. Similarly, Kasper et al. (2007) rule out the possibility of a positive power law index for a semimajor axis distribution with a cutoff beyond 30 AU. With their data, Chauvin et al. (2010) explored a wide range of parameters for the power law indices and concluded that the non-detection probability is very sensitive to the semimajor power law index, which means that strong constraints on this parameter can be inferred from observations. For the combinations of parameters reported by Cumming et al. (2008), Chauvin et al. (2010) report a value of the semimajor axis cutoff within 100 AU with 95% confidence.

Although the reported values are not identical due to the different samples and sensitivities, all previous surveys converge toward a similar conclusion, which is that the frequency of planets above 2–4 M around F–M stars is below 10-20% beyond 50 AU at high-confidence level. They also show that the distribution of semimajor axis is consistent with a power law of negative index, i.e. with a decrease of the number of planets with orbital separation.

The present work is similar in its aims to previous surveys, but with two major exceptions: it is focused on A-stars, and for the first time it includes the detections of the planetary system around HR 8799 and the planet around  Pic. A-stars are particularly interesting because RV surveys show a strong correlation between planet mass and stellar mass (Johnson et al. 2007, 2010a). Although this correlation may not hold for planets at large radii, it is tempting to test this hypothesis, especially given the planets recently imaged around young A-type stars (Marois et al. 2008b, 2010b; Lagrange et al. 2009, 2010). It is therefore essential that our statistical analysis take into account the confirmed detections for the estimation of the planet frequency around A-stars.

In the following sections, we present our statistical formalism and several applications of our data to place constraints on the frequency of giant exoplanetary systems at large orbital radii around A-stars, and their possible mass and semimajor axis distributions. We perform the analysis for a total of 42 stars: the 39 that we have observed with NaCo and NIRI, and 3 young A-stars included from the survey of young, nearby austral stars by Chauvin et al. (2010) Pic (using the data presented in Bonnefoy et al. 2011), HIP 61498 and HIP 98495.

5.2 Statistical formalism

Our formalism for the statistical analysis is based on previous works by Carson et al. (2006) and Lafrenière et al. (2007a). We consider the observation of stars enumerated by . We note the fraction of stars that have at least one companion of mass and semimajor axis in the interval , and the probability that such a companion would be detected from our observations. The analysis does not differentiate between a single planet system and a multiple planet system, i.e. represents the frequency of giant planetary systems at large orbital radii. With these notations, the probability of detecting such a companion around star is and the probability of not detecting it is . Denoting the detections made by the observations, such that equals 1 if at least one companion is detected around star and 0 otherwise, the likelihood of the data given is

(1)

The determination of the probability that the fraction of stars having at least one companion is is obtained from Bayes’ theorem:

(2)

where is the a priori probability density of , or prior distribution, and is the probability density of given the observations , or posterior distribution. Since we have no a priori knowledge of the wide-orbit massive planets frequency, we adopt a “maximum ignorance” prior, . This approach is similar to previous studies for deep imaging surveys, allowing a direct comparison of the results.

Given a confidence level (CL) , we can use the posterior distribution to determine a confidence interval (CI) for using

(3)

where and represent the bounding values of , i.e. the minimal and maximal fraction of stars with at least one planetary companion. Previous studies have always assumed a value of since no detection was included. In that case, Eq. 3 becomes an implicit equation on , which can be solved numerically to estimate . In our case, we want to use our detections to infer a value for both and . Following Lafrenière et al. (2007a), an equal-tail CI is found by solving:

(4)
(5)

Again, when a value of is fixed, Eq. 4 and 5 become implicit equations on and that can be solved numerically.

5.3 Monte-Carlo simulations

Figure 6: The detection limit (dashed line) for wide orbit planets from the observations of the target HR 8799 in the NIRI CH4s filter (left) and filter (right) compared to the DUSTY (red line) and COND (black line) evolutionary models (Chabrier et al. 2000; Baraffe et al. 2003). The DUSTY model mass-magnitude grid does not extend to magnitudes as faint as the detection limits. The current estimates of the HR 8799 planet masses (Marois et al. 2010b) are also plotted, and the empirical masses are consistent with the COND magnitudes within the error bars.

The critical step of the statistical analysis is the determination of , the completeness, for all observed targets. This completeness represents the fraction of companions that would have been detected in the interval with the observations. It is directly related to the sensitivity of the observations to the physical parameters of the planets, which in turn is a function of the detection limits, age and distance of each star.

A classical approach to estimate is to use Monte-Carlo (MC) simulations to generate large populations of planets with random physical and orbital parameters, and check their detectability given the sensitivity limits of the survey. For the MC simulations, we used the Multi-purpose Exoplanet Simulation System (MESS, Bonavita et al. 2012), a tool specifically designed for the statistical analysis of exoplanet surveys, which has already been used for the statistical analysis of several surveys (Chauvin et al. 2010; Janson et al. 2011; Delorme et al. 2012). The mass and semimajor axis of the planets are generated either from a linear grid or from specific distributions like power laws extrapolated from RV surveys of close-in planets (e.g. Fischer & Valenti 2005; Cumming et al. 2008; Johnson et al. 2010a). Then for each combination of mass/semimajor axis, a large number of planets are generated by sampling randomly the other orbital parameters: the inclination has a uniform distribution in , the longitude of the ascending node and the argument of periastron have a uniform distribution between 0 and , and the time of passage at periastron has a uniform distribution between 0 and , with the orbital period. The eccentricity distribution is uniform between as suggested by surveys of long period exoplanets (Cumming et al. 2008). However, given the large number of planets generated at each mass/semimajor axis point, the effect of the eccentricity distribution does not impact significantly the results. Once all the parameters are chosen for a planet, its projected separation on the sky plane is calculated taking into account the distance of the star, all the orbital parameters and the date of the observations.

The other important observable needed for comparison to the detection limits is the luminosity of the planet. This quantity is directly related to the planet mass and age, and can be estimated using evolutionary models. The predictions from evolutionary models can vary significantly depending on which model is considered. In particular, the cold-start (Marley et al. 2007; Fortney et al. 2008) models predict much fainter planets at young ages for a given mass than the hot-start models (Chabrier et al. 2000; Baraffe et al. 2003; Burrows et al. 2003). Hot-start models generally use arbitrary initial conditions, independent of the outcome of the formation phase, assuming that they will be “forgotten” after a few millions of years (Baraffe et al. 2002). In contrast, cold-start models have been developed to use initial conditions calculated from a core-accretion mechanism for the formation of the planets. However, the simplified treatment of the core-accretion shock results in significant uncertainties on the predictions of these models (Fortney et al. 2008). Existing observations are more consistent with predictions from the hot-start models rather than the cold-start models (see Janson et al. 2011, Fig. 4), although the much fainter magnitudes of cold-start planets would not have been detectable by most observations. Only hot-start models are considered for this study, since the brightness levels of cold-start planets are too faint for the typical sensitivity of current observations.

More precisely, the predicted magnitude from evolutionary models (Chabrier et al. 2000; Baraffe et al. 2003) – calculated in the CH4s and K’ filters of NIRI or in the Ks filter of NaCo (I. Baraffe and F. Allard, private communication) – are used to estimate the planet magnitudes as a function of mass for the age of each target star. Only two extensive evolutionary model grids are available – DUSTY and COND (Chabrier et al. 2000; Baraffe et al. 2003). Of the two evolutionary models, only the COND models extend to sufficiently faint magnitudes and low planet masses to reach the detection limits of our observations, so the analysis in this paper uses the COND models. As an example, the DUSTY and COND mass-magnitude relations at the age of 30 Myr are shown in Fig. 6 which illustrates the limitation of the DUSTY models compared to the sensitivity limit of the observations. Figure 6 also plots the estimates of the masses of the outer three HR 8799 planets based on dynamical stability limits and evolutionary models (Marois et al. 2010b) and compares the masses to the predictions of the DUSTY and COND models. Within the current capacity to perform empirical comparisons, the -band and -band magnitudes are consistent with the COND models, as the range of estimated masses for the HR 8799 planets encompasses both the predicted CH4s and -band magnitudes for that mass range. Of the 42 stars in the sample, 9 were observed with the CH4s filter and 33 were observed with K-band filter. Spectroscopic and photometric observations of the HR 8799 planets have measured differences from the theoretical DUSTY and COND atmospheres and highlighted the need for updated models incorporating non-equilibrium chemistry and an improved treatment of clouds (Janson et al. 2010; Bowler et al. 2010b; Barman et al. 2011a; Galicher et al. 2011), however grids of models including these effects are not currently available.

In the MC simulations, the masses of the generated planets are converted to absolute magnitudes in the filter matching the observations using the evolutionary models, and the signal of each planet is then compared to the 5 detection limit of the observations at the appropriate angular separation. Since we are using the 1D detection limits, the position angle of the simulated planets is simply ignored. A planet is considered as detected if its observed signal lies above the detection limit at its projected position. The completeness is then calculated as the fraction of detected planets over the number of generated planets.

The two new binaries discovered in the course of our survey, HIP 42334 and HIP 104365, were included into the analysis taking into account the possible disruption induced by the stellar binarity. The limiting values for the semimajor axis of the planets were computed using the analytical expressions of Holman & Wiegert (1999). They define the limiting values that the semimajor axis of a planet can reach and still maintain its orbital stability, as a function of the mass-ratio and orbital elements of the binary. For HIP 42334 and HIP 104365, the ranges of semimajor axes where long-term stability is unlikely are respectively 85–720 AU and 198–1750 AU, so no planets were generated within these intervals.

Finally, for the 13 targets where candidates were detected outside of 320 AU but not followed-up by subsequent observations, planets generated in the MC simulation with a projected separation larger than the closest candidate detected were considered as non-detectable.

5.4 Mean probability of detection

Figure 7: Mean probability of detection of our survey as a function of semimajor axis and planet mass (left) and for specific masses, given above or below each curve in M, as a function of semimajor axis (right). The mean probability of detection is obtained by averaging the completeness maps of all targets of the survey.  Pic b (orange square) and the 4 planets of the HR 8799 system (red circles) have been overlaid for reference using the current best estimates of their masse and separation. Note that the HR 8799 planets are represented at their projected physical separation because the true physical separation is not yet known precisely.

The overall sensitivity of our observations can be estimated in the mass interval 1–14 M and the semimajor axis interval 5–320 AU. For this analysis, we perform MC simulations for each target assuming no particular distribution for the mass and semimajor axis of the simulated planets. At each point on a linear grid of semimajor axis and mass, planets are simulated following the procedure described in Sect. 5.3, and their detectability is tested with the detection limits to calculate a completeness map for each target of the sample. The mean survey detection thresholds are calculated to illustrate the sensitivity levels, but the completeness of each target is used in the statistical analysis rather than an aggregate sample sensitivity.

The mean detection probability of the full survey is estimated by averaging all the completeness maps for each of the 42 targets. Only the deepest observation of each target was used in the MC simulations, since the second epoch observations were generally only deep enough to redetect the candidates. For the particular case of HR 8799, which is the only target with five very deep observations, the completeness maps for the five epochs were averaged together before being included with the maps of the other targets to avoid giving disproportionate weight to that specific target.

The mean probability of detection for the full survey is represented in Fig. 7 with contour plots as a function of mass (1–30 M) and semimajor axis (10–2500 AU), and cuts at specific masses. For a planet of given mass and semimajor axis, Fig. 7 describes the probability that we would have detected it from our survey data. Observing bright A-stars necessarily results in a greater minimum-mass detection limit, due to the increased contrast compared to solar-type stars. This is clearly visible on Fig. 7, which shows that our data is most sensitive above 10 M and at separations of 50 AU or more, reaching the highest probabilities of detection between 75 and 300 AU. This is consistent with the median distance of our sample (50 pc), a saturation radius of 0.3–0.5 for the data, and the fact that the sensitivity limits usually reach a plateau at separations larger than 3–4. The peak sensitivity reaches 45%, 80% and 94% respectively for 5, 10 and 14 M. The overall sensitivity to planets similar to HR 8799 bcd and  Pic is low, with a detection probability ranging from 1% for  Pic b to 30% for HR 8799 b, due to the small number of targets covering the appropriate range of orbital separations. The capacity to detect planets like  Pic b is particularly low, and we consider this complication in the frequency calculation in Sect. 5.5. As expected, the sensitivity of the survey to BDs is much higher, with a detection probability above 50% between 75 and 300 AU.

5.5 Estimation of the planet and brown dwarf systems frequency

In this section we publish the first estimation of the frequency of exoplanetary systems in the mass interval 3–14 M and the semimajor axis interval 5–320 AU for A-stars. For each target in , the individual target completeness map is used to estimate the mean value of over the 3–14 M and 5–320 AU intervals. The values of for each target is then used in Eq. 1 to calculate the likelihood for values of between 0 and 1. Using numerial integration of the likelihood with respect to , the posterior distribution is derived using Eq. 2. Finally, 95% and 68% confidence intervals are calculated using numerical integrations of Eq. 4 and 5.

Figure 8 presents the probability density function of calculated from our observations, using a linear-flat prior, compared to a binomial distribution that represent an ideal case in which the observations would be sensitive to all planets in the intervals 5–320 AU and 3–14 M. We have seen from Sect. 5.4 that our sensitivity to low-mass, close-in planets is limited, so the calculated probability density departs significantly from the ideal binomial distribution. When considering both detections around HR 8799 and  Pic, we estimate the planetary systems frequency around A-stars to at a confidence level of 68% (1). This value relies on the assumption made in the Monte-Carlo simulations (see Sect. 5.3 and 5.4) that the mass distributions of the simulated planets is flat. We investigate constraints on other possible distributions in Sect. 5.6. Although this probability is relatively high, it is a direct consequence of our decreased sensitivity to short-period planets with low mass. Upcoming large surveys with increased sensitivity to planets in 5–50 AU orbits will provide a more accurate estimation of the frequency over the full 5–320 AU range.

We note that the inclusion of  Pic in the sample must be considered with care. Although it was included as part of the previous deep imaging survey by Chauvin et al. (2010), it is important to note that their observations were not sensitive enough to detect the planet. For that particular target, we reanalyzed the data presented by Bonnefoy et al. (2011), which is more sensitive than the data of Chauvin et al. (2010). This could potentially cause a bias, because it is unknown whether there are similarly sensitive unpublished datasets for the other A-stars included from the Chauvin et al. (2010) survey. An additional bias could occur because  Pic may have been selected among all potential A-stars targets due to indicators correlated with the presence of a planet, such as the known warp in the disk. Because of these complications, we also present an estimation of the frequency considering  Pic as a non-detection777The contrast curve of Chauvin et al. (2010) is used for that estimation.. In that case, we estimate the frequency to at 68% confidence level ( at 95% confidence level).

Finally, since our data is very sensitive to the brown dwarf regime, we can make an estimation of the brown dwarf systems frequency around A-stars including the detection of HR 7329 B in the survey. In the intervals of 14–75 M and 5–320 AU, we estimate the frequency to at a confidence level of 68%.

Figure 8: Representation of the probability density of given the observations (posterior distribution) as a function of the planetary systems frequency , in the semimajor axis interval 5–320 AU and mass interval 3–14 M, for two detections around HR 8799 and  Pic (see Sect. 5.5 for details). The posterior distribution is obtained from Eq. 2 assuming a linear-flat prior for the frequency and a flat mass distribution for the planets. An equal-tail CI has been calculated using Eq. 4 and 5 for confidence levels of 68% (light grey area) and 95% (dark grey area). The dashed lines gives the binomial distributions for one or two detections out of 42 targets, which represent the probability density that would be obtained in an ideal case where the observations would be sensitive to all planets in the range 5–320 AU and 3–14 M.

5.6 Constraints on giant planets population distributions

Figure 9: Contour plots showing the confidence with which we can exclude a model with specific parameters for the mass distribution power law (, from -1.5 to 1.5, top to bottom), the semimajor axis distribution power law index (, from -2.5 to 0.2) and the semimajor axis distribution power law cutoff (, from 10 to 320 AU). The hatched grey area corresponds to the values of within the separation of the known planet HR 8799 b (dashed line). Given the existence of the planet at a separation of 68 AU, values of below that separation are de facto excluded. The best fit values of Cumming et al. (2008) for solar-type stars in terms of mass and semimajor axis are and . Bowler et al. (2010a) report a 50% confidence for values of higher than zero.

Another interesting aspect of our sample is the possibility to place some initial constraints on the distributions of mass and semimajor axis for the giant planet population. The approach taken by previous studies was to try extrapolating the distributions coming from RV surveys, which are usually parameterized with power laws. One of the most complete RV study on this topic for FGKM stars is the one performed by Cumming et al. (2008) using the sensitivity limits from the Keck Planet Search. They determine that the planet frequency for solar-types stars is 10.5% for 0.3–10 M planets with periods less than 1826 days, and that the mass and semimajor axis distributions are best modeled with power laws respectively of index888We define the power law indices with respect to linear bins rather than the logarithmic bins of Cumming et al. (2008). Also, our defines the semimajor axis distribution rather than the period distribution. and , i.e. and . For early-type stars, Johnson et al. (2010a) find a positive correlation between planet frequency and primary mass, and determine a frequency of % for stars in the range 1.3–1.9 M (see Table 6 for their measured frequency in different bins of stellar mass). For the distributions of mass and semimajor axis around A-stars, Bowler et al. (2010a) use the detections of their survey to place constraints on the possible values of and . They find that the values of Cumming et al. (2008) are inconsistent with their observations, with a probability below 1%, and consequently they favor positive values for both parameters.

For our study we assume the frequency is known over a certain range of mass and semi-major axis from RV surveys, and we model the mass and semimajor axis distributions as power laws. A cutoff on the semimajor axis distribution () is added to define an upper limit to the separation at which planets are present. To obtain meaningful results it is important to define the range of mass and semimajor axis over which the frequency is known and valid. In our case, we use the value reported by Bowler et al. (2010a) and Johnson et al. (2010a) in their survey of giant planets around old A stars: % measured for planets in the ranges 0.5–14 M and 0.1–3.0 AU (median sensitivity of their RV data).

However, our observations are not sensitive to the same range of mass and semimajor axis as the RV data, so the value of needs to be scaled to be valid inside the range of parameters where we are sensitive, for the planet mass and for the semimajor axis. Once the indices and are chosen, the power laws can be integrated over [0.5,14] M and [0.1,3.0] AU for the RV, and over and for the direct imaging. Finally, the value of is scaled by the ratio of the direct imaging integrated values over the RV integrated values. This normalization ensures that the new frequency, , is valid over and , and that its value over [0.5,14] M and [0.1,3.0] AU is always equal to to match the RV data. In our case, we used the current knowledge of the properties of the detected companions around HR 8799 and  Pic to define the appropriate range of mass and semimajor axis: intervals of [3.5,12] M and [8,68] AU were used to encompass these currently known detections.

For each of the targets in our sample, we simulate different populations of planets with distributions generated on a grid of values for (-1.5 to 1.5 by steps of 0.1), (-2.5 to 0.5 by steps of 0.1) and the semimajor axis distribution cutoff, (10 to 320 AU by steps of 10 AU). For each point of the grid, planets are simulated and the fraction of detectable planets in the intervals [3.5,12] M and [8,68] AU is measured. If we assume that all stars of the sample have one planet in that range, the sum of these fractions over our 42 stars would give the number of detection that would be expected from the observations (taking into account the sensitivity around each target). However, we have assumed that the frequency of planetary systems is equal to , so the sum needs to be multiplied by to obtain the expected number of detections over [3.5,12] M and [8,68] AU, taking into account the known frequency from RV surveys. Finally, we can use Poisson statistics to obtain the probability that this expected number of detections matches our real detections. In Fig. 9 we represent contour plots showing the confidence with which we can exclude different combinations of parameters given the detections in the range [3.5,12] M and [8,68] AU. Before describing the results, we remind that increasing the value of is equivalent to increasing the proportion of high-mass planets, and that increasing the value of is equivalent to increasing the proportion of planets at large orbital separations. We first note that given the detection of the planet HR 8799 b at an orbital separation of 68 AU, we assume that the cutoff for the semimajor axis distribution, , lies at higher orbital separations. In the three contour plots, the combination of parameters that can be excluded with 95% confidence defines a vertical area for  AU. The general trend when increasing , i.e. when increasing the proportion of high-mass planets, is to move the allowed range of parameters towards smaller values for .

The top plot of Fig. 9 is close to the value of reported by Cumming et al. (2008). With this value of , their value of is not in the range of parameters allowed by our observations. When exploring higher values of , as suggested by Bowler et al. (2010a), a value of -0.61 for becomes even less likely given a cutoff of  AU. We also note that Bowler et al. (2010a) report higher confidence for positive values of both and . However, when increasing in our simulations, the positive values of become less likely, suggesting a possible change in the population of exoplanets between the short separation planets detected by RV and the planets at very wide orbital separations around A-stars.

6 Discussion and conlusions

Figure 10: Comparison of the mean probability of detection for the present survey (plain orange contours) with a simulated survey with GPI of the same 42 targets (dashed blue contours). The mean probability of detection is obtained using Monte-Carlo simulations as described in Sect. 5.4. In addition to  Pic b (orange square) and the HR 8799 planets (red circles), the giant planets detected around old A stars by Johnson et al. (2010b, 2011) have been overplotted (purple triangles). The dotted purple line shows the median detection threshold of the Bowler et al. (2010a) RV survey around old A-type stars. The HR 8799 planets are represented at their projected physical separation because the true physical separation is not yet known precisely. Note that Fomalhaut b has not been included due to its uncertain nature (see e.g. Janson et al. 2012).
Sample Host mass Tech. Frequency Sep. range Planet mass limit Reference
(M) (AU) (M)
F5–A0 1.5–3.0 AO 5.9–18.8% 5–320 3–14The detection limit is a function of separation, as detailed in Fig. 7. The frequency estimation is based on a flat distribution for the planets mass. this work
Evolved A 1.3–1.9 RV % 0.1–3 0.2–1.3The detection limit is a function of separation, as reported in Bowler et al. (2010a). Johnson et al. (2010a)
K7–F2 0.7–1.5 AO % 25–856 4 Nielsen & Close (2010)
FGK 0.7–1.3 RV % 0.01–3 0.5–0.9Scaled from the limits in Bowler et al. (2010a) using the fixed RV amplitude cutoff of  m s reported in Johnson et al. (2010a) and the ratio of the average mass of the bin to the average mass of the evolved A-stars, taken to the power of 2/3. Johnson et al. (2010a)
M5–M0 0.2–0.6 AO % 9–207 4 Nielsen & Close (2010)
M 0.1–0.7 RV % 0.01–3 0.1–0.5Scaled from the limits in Bowler et al. (2010a) using the fixed RV amplitude cutoff of  m s reported in Johnson et al. (2010a) and the ratio of the average mass of the bin to the average mass of the evolved A-stars, taken to the power of 2/3. Johnson et al. (2010a)
999
Table 6: Samples and methods comparison.

The present survey results on A-star wide orbit planetary systems frequency and population distribution can be compared and combined with previous results to investigate the effects of host star mass, the predictions of theoretical models, the impact of debris disks, and the current state of the planet population across a range of orbital separations. Table 6 summarizes the frequencies and survey sensitivities reported in both AO (current work and Nielsen & Close 2010, which combines the results of Masciadri et al. 2005, Biller et al. 2007, and Lafrenière et al. 2007a) and RV surveys (Johnson et al. 2010b). These studies were selected for comparison, since each covers a large sample and considers the same 68% confidence interval. For the sample with the most similar target mass range, the evolved A-stars, the AO measurement of the wide orbit planetary systems frequency range of 5.9–18.8% encompasses the close orbit planet frequency of %. Direct imaging searches covering similar orbital separations to this study, but targeting lower mass stars (Nielsen & Close 2010) find an upper limit of 20%, while RV searches for close orbit planets measure a declining frequency for the successively lower mass samples (Johnson et al. 2010a). From the current statistics on the wide orbit imaged planet population, it is not possible to determine how the planetary systems frequency scales with host star mass. Theoretical models have predicted a rising (Kennedy & Kenyon 2008), peaked (Ida & Lin 2005), or declining (Kornet et al. 2006) planet frequency on mass, with models incorporating different treatments of factors such as the location and evolution of the snow line, the initial disk size and its dependence on host star mass, and the orbital migration of planets.

Since the survey was designed to detect planets, the full sample is extremely sensitive to brighter brown dwarf companions in wide orbits (probability of detection % in 75–300 AU, see Fig. 7), enabling an initial assessment of the brown dwarf systems frequency around stars more massive than the Sun. Among the 42 targets, one brown dwarf was previously identified around HR 7329 (Lowrance et al. 2000; Neuhäuser et al. 2011), yielding a brown dwarf systems frequency of % (see Sect. 5.5). This low level of brown dwarf companions is consistent with the % frequency measured for a larger scale survey of young (3–3000 Myr), solar-type (F5–K5) stars covering a wider range of separations (28–1590 AU) with an AO survey (Metchev & Hillenbrand 2009).

Theoretical models of the formation of giant planets by gravitational instability have proposed that the directly imaged planets may be the lowest mass products of the same process that forms binary stars and brown dwarfs (Kratter et al. 2010), and these simulations predict a higher proportion of brown dwarfs than planets. While the statistics of the current study are limited, the similarity of the brown dwarf and planetary systems frequencies suggests that the imaged planets may have formed through a different process such as core accretion (e.g. Pollack et al. 1996). An investigation into the conditions for gravitational instability for the specific case of the HR 8799 planets (Nero & Bjorkman 2009) found that the likelihood of this formation mechanism increased with radius towards the outermost planets. Other possible explanations for the wide orbits of some of the directly imaged planets include scattering to larger orbital radii from interactions in multiple planet systems (e.g. Veras & Armitage 2004), outward migration in a resonance with multiple planets (e.g. Crida et al. 2009), and outward migration through a planet-disk interaction (e.g. Veras & Armitage 2004).

The full sample of 42 young, A- and F-stars includes 17 systems with dusty debris disks sustained by the ongoing collisional grinding of planetesimals into smaller particles (Backman & Paresce 1991) or an event such as a catastrophic collision of planets (Cameron 1997; Melis et al. 2010). The 3 targets in this sample with imaged planets or brown dwarfs all reside in systems encircled by dust disks. Additionally, Fomalhaut (not included in this study) would provide another example of a planet-disk system (Kalas et al. 2008) if the presence of a planet was confirmed (see e.g. Janson et al. 2012). While not all A-stars with debris disks harbor massive giant planets in wide orbits, the frequency of planetary systems appears higher among the targets with excess emission from dust. The combination of resolved disk maps and planet imaging is an especially powerful tool to understand the planet-disk interactions that may sculpt young planetary systems (e.g. Liou & Zook 1999; Kuchner & Holman 2003; Wyatt 2006; Quillen & Faber 2006). The structure of the Fomalhaut disk inner edge has been compared with dynamical models of planets with different masses (Chiang et al. 2009), and the size and shape of the HR 8799 disk have been compared with models of the dynamically cleared zones and orbital migration history (Patience et al. 2011). Advances in disk imaging with ALMA and planet imaging with extreme AO will enable more detailed studies to investigate planet-disk interactions which may generate structures such as asymmetries and clumps.

By combining the results and sensitivities of existing direct imaging A-star planet searches with those of RV programs targeting retired A-stars, a comprehensive summary of the currently known A-star planet population is given in Fig. 10. Analysis of the current data already shows distinct differences. The close orbit A-star planets are best fit by a distribution that is flat or rising with increasing orbital separation (Bowler et al. 2010a) and rising with increasing planet mass (Bowler et al. 2010a; Johnson et al. 2010a). In contrast, the wide orbit planet data, incorporating the outer cutoff implied by the outermost HR 8799 planet, are consistent with a distribution that is declining with increasing orbital separation. This result is independent of whether the distribution of planets is rising, flat, or declining with planet mass, as shown in the three panels of Fig. 9 (the allowed region always has a negative power law index). Thus, the existing surveys have identified the boundaries of planets around A-stars – from 0.6 AU (Bowler et al. 2010a) to 70 AU – and have indicated that the data cannot be fit by a single distribution of properties.

Figure 10 also presents the sensitivity of a simulated survey with the upcoming high-contrast imager Gemini/GPI (Macintosh et al. 2008). Using simulated contrast curves for that instrument, we performed Monte-Carlo simulations similar to the ones described in Sect. 5.4 on the 42 targets of our sample. They show that extreme AO imaging will provide the crucial coverage to connect the close orbit and wide orbit populations and to reveal the full distribution of planets as a function of separation and planet mass. Extreme AO data will determine if the transition from a flat or rising population of close orbit planets to a declining population of wide orbit planets is smooth or discontinuous and may discover a peak in the separation distribution. The total number of planets detected in extreme AO surveys, combined with trends measured for close orbit A-star planets, can be used as a test of formation models (e.g. Crepp & Johnson 2011). Since the upcoming extreme AO instrument include integral field units, it will be possible to investigate the nature of the exoplanet atmospheres in addition to the population statistics. Spectra of the currently imaged planetary mass companions have revealed differences with brown dwarfs of similar temperatures (e.g. Janson et al. 2010; Patience et al. 2010; Barman et al. 2011a, b; Skemer et al. 2011), and upcoming AO observations will further explore the architectures and atmospheres of exoplanets around A-stars.

Acknowledgements.
A. V. and J. P. acknowledge support from a Science and Technology Facilities Council (STFC) grant (ST/H002707/1). J. P. acknowledges support from the Leverhulme Trust through a research project grant (F/00144/BJ). The authors would like to thank G. Chauvin for providing his published detection limits, as well as E. Nielsen, M. Viallet and F. Pont for fruitful discussions on the statistical analysis, and J. Johnson for information on comparisons with RV studies. We thank the ESO and Gemini staff for performing the observations. This research made use of the SIMBAD database, operated at CDS, Strasbourg, France.

References

  • Alibert et al. (2011) Alibert, Y., Mordasini, C., & Benz, W. 2011, A&A, 526, A63+
  • Backman & Paresce (1991) Backman, D. E. & Paresce, F. 1991, Meteoritics, 26, 314
  • Baraffe et al. (2002) Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 2002, A&A, 382, 563
  • Baraffe et al. (2003) Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, A&A, 402, 701
  • Barman et al. (2011a) Barman, T. S., Macintosh, B., Konopacky, Q. M., & Marois, C. 2011a, ApJ, 733, 65
  • Barman et al. (2011b) Barman, T. S., Macintosh, B., Konopacky, Q. M., & Marois, C. 2011b, ApJ, 735, L39
  • Beuzit et al. (2008) Beuzit, J.-L., Feldt, M., Dohlen, K., et al. 2008, in SPIE Conference Series, Vol. 7014
  • Biller et al. (2007) Biller, B. A., Close, L. M., Masciadri, E., et al. 2007, ApJS, 173, 143
  • Biller et al. (2010) Biller, B. A., Liu, M. C., Wahhaj, Z., et al. 2010, ApJ, 720, L82
  • Bonavita et al. (2012) Bonavita, M., Chauvin, G., Desidera, S., et al. 2012, A&A, 537, A67
  • Bonnefoy et al. (2011) Bonnefoy, M., Lagrange, A.-M., Boccaletti, A., et al. 2011, A&A, 528, L15+
  • Borucki et al. (2011) Borucki, W. J., Koch, D. G., Basri, G., et al. 2011, ApJ, 736, 19
  • Bowler et al. (2010a) Bowler, B. P., Johnson, J. A., Marcy, G. W., et al. 2010a, ApJ, 709, 396
  • Bowler et al. (2010b) Bowler, B. P., Liu, M. C., Dupuy, T. J., & Cushing, M. C. 2010b, ApJ, 723, 850
  • Breger (1986) Breger, M. 1986, ApJ, 309, 311
  • Burrows et al. (2003) Burrows, A., Sudarsky, D., & Lunine, J. I. 2003, ApJ, 596, 587
  • Cabrera et al. (2009) Cabrera, J., Fridlund, M., Ollivier, M., et al. 2009, A&A, 506, 501
  • Cameron (1997) Cameron, A. G. W. 1997, Icarus, 126, 126
  • Carson et al. (2006) Carson, J. C., Eikenberry, S. S., Smith, J. J., & Cordes, J. M. 2006, AJ, 132, 1146
  • Chabrier et al. (2000) Chabrier, G., Baraffe, I., Allard, F., & Hauschildt, P. 2000, ApJ, 542, 464
  • Chauvin (2010) Chauvin, G. 2010, in In the Spirit of Lyot 2010
  • Chauvin et al. (2012) Chauvin, G., Lagrange, A.-M., Beust, H., et al. 2012, ArXiv e-prints
  • Chauvin et al. (2010) Chauvin, G., Lagrange, A.-M., Bonavita, M., et al. 2010, A&A, 509, A52+
  • Chauvin et al. (2003) Chauvin, G., Thomson, M., Dumas, C., et al. 2003, A&A, 404, 157
  • Chiang et al. (2009) Chiang, E., Kite, E., Kalas, P., Graham, J. R., & Clampin, M. 2009, ApJ, 693, 734
  • Crepp & Johnson (2011) Crepp, J. R. & Johnson, J. A. 2011, ApJ, 733, 126
  • Crida et al. (2009) Crida, A., Masset, F., & Morbidelli, A. 2009, ApJ, 705, L148
  • Crifo et al. (1997) Crifo, F., Vidal-Madjar, A., Lallement, R., Ferlet, R., & Gerbaldi, M. 1997, A&A, 320, L29
  • Cumming et al. (2008) Cumming, A., Butler, R. P., Marcy, G. W., et al. 2008, PASP, 120, 531
  • Currie et al. (2011) Currie, T., Burrows, A., Itoh, Y., et al. 2011, ApJ, 729, 128
  • Delorme et al. (2012) Delorme, P., Lagrange, A. M., Chauvin, G., et al. 2012, A&A, 539, A72
  • Doyon et al. (2010) Doyon, R., Lafrenière, D., Artigau, E., Malo, L., & Marois, C. 2010, in In the Spirit of Lyot 2010
  • Ehrenreich et al. (2010) Ehrenreich, D., Lagrange, A.-M., Montagnier, G., et al. 2010, A&A, 523, A73
  • Fischer & Valenti (2005) Fischer, D. A. & Valenti, J. 2005, ApJ, 622, 1102
  • Fitzgerald & Graham (2006) Fitzgerald, M. P. & Graham, J. R. 2006, ApJ, 637, 541
  • Fortney et al. (2008) Fortney, J. J., Marley, M. S., Saumon, D., & Lodders, K. 2008, ApJ, 683, 1104
  • Fritz et al. (2010) Fritz, T., Gillessen, S., Trippe, S., et al. 2010, MNRAS, 401, 1177
  • Fukagawa et al. (2004) Fukagawa, M., Hayashi, M., Tamura, M., et al. 2004, ApJ, 605, L53
  • Galicher et al. (2011) Galicher, R., Marois, C., Macintosh, B., Barman, T., & Konopacky, Q. 2011, ApJ, 739, L41
  • Goodman (1968) Goodman, J. W. 1968, Introduction to Fourier Optics (McGraw-Hill, New York)
  • Guenther et al. (2001) Guenther, E. W., Neuhäuser, R., Huélamo, N., Brandner, W., & Alves, J. 2001, A&A, 365, 514
  • Herriot et al. (2000) Herriot, G., Morris, S., Anthony, A., et al. 2000, in SPIE Conference Series, ed. P. L. Wizinowich, Vol. 4007, 115–125
  • Hodapp et al. (2003) Hodapp, K. W., Jensen, J. B., Irwin, E. M., et al. 2003, PASP, 115, 1388
  • Holman & Wiegert (1999) Holman, M. J. & Wiegert, P. A. 1999, AJ, 117, 621
  • Ida & Lin (2005) Ida, S. & Lin, D. N. C. 2005, ApJ, 626, 1045
  • Janson et al. (2010) Janson, M., Bergfors, C., Goto, M., Brandner, W., & Lafrenière, D. 2010, ApJ, 710, L35
  • Janson et al. (2011) Janson, M., Bonavita, M., Klahr, H., et al. 2011, ApJ, 736, 89
  • Janson et al. (2012) Janson, M., Carson, J., Lafreniere, D., et al. 2012, ArXiv e-prints
  • Johnson et al. (2010a) Johnson, J. A., Aller, K. M., Howard, A. W., & Crepp, J. R. 2010a, PASP, 122, 905
  • Johnson et al. (2007) Johnson, J. A., Butler, R. P., Marcy, G. W., et al. 2007, ApJ, 670, 833
  • Johnson et al. (2011) Johnson, J. A., Clanton, C., Howard, A. W., et al. 2011, ApJS, 197, 26
  • Johnson et al. (2010b) Johnson, J. A., Howard, A. W., Bowler, B. P., et al. 2010b, PASP, 122, 701
  • Kalas et al. (2008) Kalas, P., Graham, J. R., Chiang, E., et al. 2008, Science, 322, 1345
  • Kalas & Jewitt (1995) Kalas, P. & Jewitt, D. 1995, AJ, 110, 794
  • Kasper et al. (2007) Kasper, M., Apai, D., Janson, M., & Brandner, W. 2007, A&A, 472, 321
  • Kennedy & Kenyon (2008) Kennedy, G. M. & Kenyon, S. J. 2008, ApJ, 673, 502
  • Kornet et al. (2006) Kornet, K., Wolf, S., & Różyczka, M. 2006, A&A, 458, 661
  • Kratter et al. (2010) Kratter, K. M., Murray-Clay, R. A., & Youdin, A. N. 2010, ApJ, 710, 1375
  • Kuchner & Holman (2003) Kuchner, M. J. & Holman, M. J. 2003, ApJ, 588, 1110
  • Lafrenière et al. (2007a) Lafrenière, D., Doyon, R., Marois, C., et al. 2007a, ApJ, 670, 1367
  • Lafrenière et al. (2008) Lafrenière, D., Jayawardhana, R., & van Kerkwijk, M. H. 2008, ApJ, 689, L153
  • Lafrenière et al. (2007b) Lafrenière, D., Marois, C., Doyon, R., Nadeau, D., & Artigau, É. 2007b, ApJ, 660, 770
  • Lagrange et al. (2012a) Lagrange, A.-M., Boccaletti, A., Milli, J., et al. 2012a, ArXiv e-prints
  • Lagrange et al. (2010) Lagrange, A.-M., Bonnefoy, M., Chauvin, G., et al. 2010, Science, 329, 57
  • Lagrange et al. (2012b) Lagrange, A.-M., De Bondt, K., Meunier, N., et al. 2012b, ArXiv e-prints
  • Lagrange et al. (2009) Lagrange, A.-M., Gratadour, D., Chauvin, G., et al. 2009, A&A, 493, L21
  • Leconte et al. (2010) Leconte, J., Soummer, R., Hinkley, S., et al. 2010, ApJ, 716, 1551
  • Lenzen et al. (2003) Lenzen, R., Hartung, M., Brandner, W., et al. 2003, in SPIE Conference Series, Vol. 4841, 944–952
  • Liou & Zook (1999) Liou, J.-C. & Zook, H. A. 1999, AJ, 118, 580
  • Lowrance et al. (2000) Lowrance, P. J., Schneider, G., Kirkpatrick, J. D., et al. 2000, ApJ, 541, 390
  • Macintosh et al. (2008) Macintosh, B. A., Graham, J. R., Palmer, D. W., et al. 2008, in SPIE Conference Series, Vol. 7015
  • Mannings & Sargent (1997) Mannings, V. & Sargent, A. I. 1997, ApJ, 490, 792
  • Marcy et al. (2008) Marcy, G. W., Butler, R. P., Vogt, S. S., et al. 2008, Physica Scripta Volume T, 130, 014001
  • Marley et al. (2007) Marley, M. S., Fortney, J. J., Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2007, ApJ, 655, 541
  • Marois et al. (2006) Marois, C., Lafrenière, D., Doyon, R., Macintosh, B., & Nadeau, D. 2006, ApJ, 641, 556
  • Marois et al. (2008a) Marois, C., Lafrenière, D., Macintosh, B., & Doyon, R. 2008a, ApJ, 673, 647
  • Marois et al. (2008b) Marois, C., Macintosh, B., Barman, T., et al. 2008b, Science, 322, 1348
  • Marois et al. (2010a) Marois, C., Macintosh, B., & Véran, J.-P. 2010a, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 7736, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series
  • Marois et al. (2010b) Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., & Barman, T. 2010b, Nature, 468, 1080
  • Masciadri et al. (2005) Masciadri, E., Mundt, R., Henning, T., Alvarez, C., & Barrado y Navascués, D. 2005, ApJ, 625, 1004
  • Mason et al. (1999) Mason, B. D., Douglass, G. G., & Hartkopf, W. I. 1999, AJ, 117, 1023
  • Mason et al. (2001) Mason, B. D., Wycoff, G. L., Hartkopf, W. I., Douglass, G. G., & Worley, C. E. 2001, AJ, 122, 3466
  • Mayor & Udry (2008) Mayor, M. & Udry, S. 2008, Physica Scripta Volume T, 130, 014010
  • McCaughrean & Stauffer (1994) McCaughrean, M. J. & Stauffer, J. R. 1994, AJ, 108, 1382
  • Melis et al. (2010) Melis, C., Zuckerman, B., Rhee, J. H., & Song, I. 2010, ApJ, 717, L57
  • Metchev & Hillenbrand (2009) Metchev, S. A. & Hillenbrand, L. A. 2009, ApJS, 181, 62
  • Moffat (1969) Moffat, A. F. J. 1969, A&A, 3, 455
  • Moór et al. (2006) Moór, A., Ábrahám, P., Derekas, A., et al. 2006, ApJ, 644, 525
  • Moór et al. (2011) Moór, A., Ábrahám, P., Juhász, A., et al. 2011, ApJ, 740, L7
  • Mouillet et al. (1997) Mouillet, D., Larwood, J. D., Papaloizou, J. C. B., & Lagrange, A. M. 1997, MNRAS, 292, 896
  • Nero & Bjorkman (2009) Nero, D. & Bjorkman, J. E. 2009, ApJ, 702, L163
  • Neuhäuser et al. (2011) Neuhäuser, R., Ginski, C., Schmidt, T. O. B., & Mugrauer, M. 2011, MNRAS, 416, 1430
  • Neuhäuser et al. (2003) Neuhäuser, R., Guenther, E. W., Alves, J., et al. 2003, Astronomische Nachrichten, 324, 535
  • Nielsen & Close (2010) Nielsen, E. L. & Close, L. M. 2010, ApJ, 717, 878
  • Patience et al. (2011) Patience, J., Bulger, J., King, R. R., et al. 2011, A&A, 531, L17
  • Patience et al. (2010) Patience, J., King, R. R., de Rosa, R. J., & Marois, C. 2010, A&A, 517, A76
  • Perryman et al. (1997) Perryman, M. A. C., Lindegren, L., Kovalevsky, J., et al. 1997, A&A, 323, L49
  • Pollack et al. (1996) Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62
  • Quillen & Faber (2006) Quillen, A. C. & Faber, P. 2006, MNRAS, 373, 1245
  • Rhee et al. (2007) Rhee, J. H., Song, I., Zuckerman, B., & McElwain, M. 2007, ApJ, 660, 1556
  • Rousset et al. (2003) Rousset, G., Lacombe, F., Puget, P., et al. 2003, in SPIE Conference Series, ed. P. L. Wizinowich & D. Bonaccini, Vol. 4839, 140–149
  • Skemer et al. (2011) Skemer, A. J., Close, L. M., Szűcs, L., et al. 2011, ApJ, 732, 107
  • Smith & Terrile (1984) Smith, B. A. & Terrile, R. J. 1984, Science, 226, 1421
  • Smith et al. (2009) Smith, R., Churcher, L. J., Wyatt, M. C., Moerchen, M. M., & Telesco, C. M. 2009, A&A, 493, 299
  • Soummer et al. (2011) Soummer, R., Brendan Hagan, J., Pueyo, L., et al. 2011, ApJ, 741, 55
  • Stauffer et al. (1995) Stauffer, J. R., Hartmann, L. W., & Barrado y Navascues, D. 1995, ApJ, 454, 910
  • Stauffer et al. (1998) Stauffer, J. R., Schultz, G., & Kirkpatrick, J. D. 1998, ApJ, 499, L199
  • Su et al. (2006) Su, K. Y. L., Rieke, G. H., Stansberry, J. A., et al. 2006, ApJ, 653, 675
  • Su et al. (2009) Su, K. Y. L., Rieke, G. H., Stapelfeldt, K. R., et al. 2009, ApJ, 705, 314
  • Tetzlaff et al. (2011) Tetzlaff, N., Neuhäuser, R., & Hohle, M. M. 2011, MNRAS, 410, 190
  • Thalmann et al. (2009) Thalmann, C., Carson, J., Janson, M., et al. 2009, ApJ, 707, L123
  • Trippe et al. (2008) Trippe, S., Gillessen, S., Gerhard, O. E., et al. 2008, A&A, 492, 419
  • Veras & Armitage (2004) Veras, D. & Armitage, P. J. 2004, MNRAS, 347, 613
  • Wyatt (2006) Wyatt, M. C. 2006, ApJ, 639, 1153
  • Zuckerman et al. (2011) Zuckerman, B., Rhee, J. H., Song, I., & Bessell, M. S. 2011, ApJ, 732, 61
  • Zuckerman et al. (2001) Zuckerman, B., Song, I., Bessell, M. S., & Webb, R. A. 2001, ApJ, 562, L87

Appendix A Point sources properties

Table 7 summarizes the properties of all point sources detected with our observations.

Name CC# Date Epoch Instrument Filter Sep. P.A. mag Proj. sep. StatusNote that in general, only candidates with a projected separation larger than 320 AU were followed-up.
(yr) (as) (deg) (mag) (AU)
HIP 104365 0 2007-09-18 2007.71 NIRI CH4s 9.545  0.021 15.27  0.13 7.4  0.1 526.2  1.2 Comoving
HIP 104365 0 2008-09-09 2008.69 NIRI CH4s 9.543  0.021 15.25  0.13 7.4  0.1 526.1  1.2 Comoving
HIP 104365 1 2007-09-18 2007.71 NIRI CH4s 9.595  0.021 14.27  0.13 7.5  0.1 529.0  1.2 Comoving
HIP 104365 1 2008-09-09 2008.69 NIRI CH4s 9.601  0.021 14.20  0.13 7.2  0.1 529.3  1.2 Comoving
HIP 104365 2 2007-09-18 2007.71 NIRI CH4s 13.131  0.021 62.07  0.09 14.9  0.1 723.9  1.2 Undefined
HIP 10670 0 2007-09-16 2007.71 NIRI CH4s 4.650  0.006 253.27  0.07 15.6  0.1 160.1  0.2 Background
HIP 10670 0 2008-10-12 2008.78 NIRI CH4s 4.665  0.006 253.93  0.07 14.4  0.2 160.7  0.2 Background
HIP 10670 1 2007-09-16 2007.71 NIRI CH4s 10.668  0.021 161.63  0.11 13.7  0.1 367.4  0.7 Background
HIP 10670 1 2008-10-12 2008.78 NIRI CH4s 10.599  0.021 161.88  0.12 12.7  0.1 365.0  0.7 Background
HIP 10670 2 2007-09-16 2007.71 NIRI CH4s 13.844  0.021 344.29  0.09 13.5  0.1 476.8  0.7 Ambiguous
HIP 10670 2 2008-10-12 2008.78 NIRI CH4s 13.982  0.021 344.42  0.09 12.4  0.1 481.5  0.7 Ambiguous
HIP 110935 0 2009-06-24 2009.48 NaCo 0.582  0.007 52.74  0.66 9.6  0.7 25.7  0.3 Undefined
HIP 11360 0 2007-12-30 2007.99 NIRI CH4s 12.628  0.021 194.10  0.10 14.5  0.1 571.2  0.9 Background
HIP 11360 0 2008-10-12 2008.78 NIRI CH4s 12.572  0.021 194.76  0.10 12.8  0.3 568.6  0.9 Background
HIP 14551 0 2009-12-21 2009.97 NaCo 6.058  0.007 28.65  0.06 14.9  0.2 331.0  0.4 Undefined
HIP 15648 0 2008-10-17 2008.79 NIRI 11.528  0.021 225.35  0.11 10.4  0.1 532.2  1.0 Undefined
HIP 16449 0 2007-09-22 2007.72 NIRI CH4s 10.193  0.021 25.77  0.12 11.5  0.1 733.3  1.5 Background
HIP 16449 0 2008-10-12 2008.78 NIRI CH4s 10.184  0.021 25.46  0.12 9.2  0.1 732.6  1.5 Background
HIP 22226 0 2008-01-15 2008.04 NIRI CH4s 12.156  0.021 148.90  0.10 14.7  0.5 975.6  1.7 Background
HIP 22226 0 2008-11-18 2008.88 NIRI CH4s 12.114  0.021 148.97  0.10 12.4  0.3 972.3  1.7 Background
HIP 23296 0 2010-01-03 2010.01 NIRI 6.510  0.021 308.63  0.19 13.8  0.1 322.9  1.0 Undefined
HIP 23296 1 2010-01-03 2010.01 NIRI 8.016  0.021 99.07  0.15 15.1  0.1 397.6  1.0 Undefined
HIP 23296 2 2010-01-03 2010.01 NIRI 9.174  0.021 329.02  0.13 7.5  0.1 455.0  1.0 Undefined
HIP 23296 3 2010-01-03 2010.01 NIRI 9.727  0.021 29.27  0.13 15.2  0.1 482.5  1.0 Undefined
HIP 23296 4 2010-01-03 2010.01 NIRI 11.360  0.021 173.73  0.11 13.0  0.1 563.5  1.0 Undefined
HIP 23296 5 2010-01-03 2010.01 NIRI 13.586  0.021 73.30  0.09 12.3  0.1 673.9  1.0 Undefined
HIP 26624 0 2008-11-14 2008.87 NIRI 8.753  0.021 232.98  0.14 11.0  0.1 372.6  0.9 Undefined
HIP 26624 1 2008-11-14 2008.87 NIRI 11.914  0.021 42.62  0.10 13.1  0.1 507.2  0.9 Undefined
HIP 34782 0 2009-12-21 2009.97 NaCo 8.667  0.007 316.66  0.04 14.2  0.3 412.1  0.3 Background
HIP 34782 0 2012-01-11 2012.03 NaCo 8.544  0.007 317.54  0.05 14.0  0.3 406.3  0.3 Background
HIP 34782 1 2012-01-11 2012.03 NaCo 8.552  0.007 145.51  0.05 12.2  0.1 406.6  0.3 Undefined
HIP 35567 0 2009-12-20 2009.97 NaCo 4.193  0.007 7.74  0.09 10.1  0.1 297.8  0.5 Background
HIP 35567 0 2011-12-18 2011.96 NaCo 4.097  0.007 8.35  0.09 8.9  0.1 291.0  0.5 Background
HIP 35567 1 2009-12-20 2009.97 NaCo 4.498  0.007 61.62  0.09 10.2  0.1 319.4  0.5 Background
HIP 35567 1 2011-12-18 2011.96 NaCo 4.467  0.007 63.31  0.09 9.1  0.1 317.2  0.5 Background
HIP 35567 2 2009-12-20 2009.97 NaCo 5.499  0.007 8.61  0.07 8.9  0.1 390.5  0.5 Background
HIP 35567 2 2011-12-18 2011.96 NaCo 5.407  0.007 9.14  0.07 7.5  0.1 384.0  0.5 Background
HIP 41307 0 2009-12-21 2009.97 NaCo 5.792  0.007 86.15  0.07 15.7  0.2 217.3  0.3 BackgroundBackground status confirmed by Janson et al. (2011).
HIP 42334 0 2009-12-20 2009.97 NaCo 2.010  0.007 80.29  0.19 11.8  0.1 142.9  0.5 Background
HIP 42334 0 2011-04-27 2011.32 NaCo 2.047  0.007 80.39  0.19 12.0  0.2 145.5  0.5 Background
HIP 42334 1 2009-12-20 2009.97 NaCo 3.176  0.007 11.52  0.12 8.2  0.1 225.7  0.5 Comoving
HIP 42334 1 2011-04-27 2011.32 NaCo 3.166  0.007 11.51  0.12 8.3  0.1 225.0  0.5 Comoving
HIP 42334 2 2009-12-20 2009.97 NaCo 5.291  0.007 288.05  0.07 12.9  0.1 376.0  0.5 Background
HIP 42334 2 2011-04-27 2011.32 NaCo 5.265  0.007 288.37  0.07 12.9  0.1 374.2  0.5 Background
HIP 42334 3 2009-12-20 2009.97 NaCo 5.616  0.007 290.18  0.07 15.3  0.3 399.1  0.5 Background
HIP 42334 3 2011-04-27 2011.32 NaCo 5.584  0.007 290.52  0.07 14.9  0.6 396.9  0.5 Background
HIP 44923 0 2008-03-23 2008.22 NIRI CH4s 13.860  0.021 236.78  0.09 14.1  0.1 1159.8  1.8 Undefined
HIP 53771 0 2010-03-06 2010.18 NaCo 5.800  0.007 266.75  0.07 13.5  0.3 354.3  0.4 Background
HIP 53771 0 2012-01-12 2012.03 NaCo 5.731  0.007 267.64  0.07 13.3  0.1 350.1  0.4 Background
HIP 53771 1 2010-03-06 2010.18 NaCo 6.120  0.007 251.12  0.06 11.7  0.1 373.9  0.4 Background
HIP 53771 1 2012-01-12 2012.03 NaCo 6.058  0.007 251.73  0.06 11.7  0.1 370.1  0.4 Background
HIP 53771 2 2012-01-12 2012.03 NaCo 8.545  0.007 144.52  0.05 12.5  0.2 522.0  0.4 Undefined
HIP 57328 0 2008-03-24 2008.23 NIRI CH4s 12.505  0.021 352.40  0.10 16.8  0.2 467.8  0.8 Undefined
HIP 61960 0 2008-03-21 2008.22 NIRI CH4s 12.732  0.021 57.65  0.10 15.6  0.2 461.8  0.8 Undefined
HIP 66634 0 2009-02-03 2009.09 NIRI 6.955  0.021 230.69  0.18 14.1  0.2 372.7  1.1 Undefined
HIP 69732 0 2009-02-02 2009.09 NIRI 11.462  0.021 319.19  0.11 12.3  0.1 348.0  0.6 Undefined
HIP 78078 0 2009-06-24 2009.48 NaCo 6.670  0.007 76.68  0.06 13.3  0.2 340.8  0.4 Undefined
HR 7329 0 2008-08-07 2008.60 NaCo 4.185  0.007 161.61  0.09 201.8  0.3 ComovingKnown brown-dwarf companion redetected in our data (Lowrance et al. 2000).
HIP 99273 0 2007-08-10 2007.61 NIRI CH4s 4.894  0.006 140.41  0.07 15.1  0.1 255.6  0.3 Background
HIP 99273 0 2008-09-04 2008.68 NIRI CH4s 4.802  0.006 140.26  0.07 15.3  0.2 250.8  0.3 Background
HIP 99273 1 2007-08-10 2007.61 NIRI CH4s 7.577  0.021 74.29  0.16 16.4  0.3 395.7  1.1 Background
HIP 99273 1 2008-09-04 2008.68 NIRI CH4s 7.566  0.021 73.65  0.16 16.4  0.3 395.1  1.1 Background
HIP 99273 2 2008-09-04 2008.68 NIRI CH4s 8.089  0.021 352.47  0.15 16.1  0.3 422.4  1.1 Background
HIP 99273 2 2007-08-10 2007.61 NIRI CH4s 8.005  0.021 352.70  0.15 16.1  0.2 418.0  1.1 Background
HIP 99273 3 2007-08-10 2007.61 NIRI CH4s 8.917  0.021 304.99  0.14 16.1  0.2 465.6  1.1 Background
HIP 99273 3 2008-09-04 2008.68 NIRI CH4s 8.986  0.021 305.22  0.14 16.1  0.3 469.2  1.1 Background
HIP 99273 4 2007-08-10 2007.61 NIRI CH4s 10.497  0.021 147.75  0.12 13.1  0.1 548.2  1.1 Background
HIP 99273 4 2008-09-04 2008.68 NIRI CH4s 10.407  0.021 147.77  0.12 13.0  0.1 543.5  1.1 Background
HIP 99273 5 2007-08-10 2007.61 NIRI CH4s 10.841  0.021 220.66  0.11 8.8  0.1 566.1  1.1 Background
HIP 99273 5 2008-09-04 2008.68 NIRI CH4s 10.845  0.021 221.04  0.11 8.5  0.1 566.3  1.1 Background
HIP 99273 6 2007-08-10 2007.61 NIRI CH4s 12.318  0.021 75.04  0.10 14.1  0.9 643.2  1.1 Background
HIP 99273 6 2008-09-04 2008.68 NIRI CH4s 12.294  0.021 74.66  0.10 14.5  0.1 642.0  1.1 Background
HR 8799 0 2008-09-05 2008.68 NIRI CH4s 0.630  0.006 197.39  0.55 11.6  0.3 24.8  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 1 2007-10-17 2007.79 NIRI CH4s 0.943  0.006 314.08  0.37 13.0  0.4 37.2  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 1 2008-09-01 2008.67 NIRI CH4s 0.943  0.006 315.37  0.37 12.4  0.3 37.2  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 1 2008-09-05 2008.68 NIRI CH4s 0.945  0.006 315.28  0.36 11.1  0.1 37.2  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 1 2008-10-10 2008.77 NIRI CH4s 0.937  0.006 315.55  0.37 12.5  0.5 36.9  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 1 2008-10-14 2008.79 NIRI CH4s 0.943  0.006 314.82  0.37 10.7  0.2 37.2  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 2 2007-10-17 2007.79 NIRI CH4s 1.721  0.006 60.70  0.20 13.8  0.2 67.8  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 2 2008-09-01 2008.67 NIRI CH4s 1.709  0.006 61.70  0.20 13.5  0.2 67.3  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 2 2008-09-05 2008.68 NIRI CH4s 1.722  0.006 61.11  0.20 12.2  0.1 67.8  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 2 2008-10-10 2008.77 NIRI CH4s 1.768  0.006 60.99  0.20 14.1  0.4 69.7  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HR 8799 2 2008-10-14 2008.79 NIRI CH4s 1.731  0.006 59.41  0.20 11.8  0.1 68.2  0.2 ComovingCo-motion of the HR 8799 planets has been confirmed by multiple authors (e.g. see Marois et al. 2008b, 2010b; Bowler et al. 2010b; Currie et al. 2011).
HIP 116431 0 2007-09-12 2007.70 NIRI CH4s 8.748  0.021 28.89  0.14 15.7  0.2 598.8  1.4 Background
HIP 116431 0 2008-09-24 2008.73 NIRI CH4s 8.761  0.021 28.51  0.14 15.6  0.3 599.7  1.4 Background
Table 7: continued.
101010
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
""
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
   
Add comment
Cancel
Loading ...
102756
This is a comment super asjknd jkasnjk adsnkj
Upvote
Downvote
""
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters
Submit
Cancel

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test
Test description