THE SEEDS DIRECT IMAGING SURVEY FOR PLANETS AND SCATTERED DUST EMISSION IN DEBRIS DISK SYSTEMS

The close circumstellar environment around mature (post-T Tauri and Herbig Ae/Be) stars has traditionally been difficult to study directly, due to the strong flux from the star itself, which drowns out the light of its physical surroundings over a wide range of wavelengths. However, developments in high-contrast and high-resolution instruments and techniques have made this environment increasingly accessible to detailed study in recent years. Several surveys have been performed (e.g., Kasper et al. 2007; Lafrenière et al. 2007a; Rameau et al. 2013) and a number of extrasolar planets have been imaged (e.g., Marois et al. 2010; Lagrange et al. 2010; Carson et al. 2013), and while the most extreme debris disk systems have been possible to image for some time (e.g., Smith et al. 1984), the sample of spatially resolved disks is presently growing rapidly, both in thermal (e.g., Greaves et al. 2005; Wilner et al. 2011; Acke et al. 2012) and scattered radiation (e.g., Krist et al. 2005; Kalas et al. 2005; Buenzli et al. 2010). Nonetheless, most planets and disks are still discovered only indirectly, through stellar radial velocity or transits in the case of planets (e.g., Mayor & Queloz 1995; Borucki et al. 2011) and through infrared excess in the case of disks (e.g., Beichman et al. 2006; Su et al. 2006).

The Strategic Exploration of Exoplanets and Disks with Subaru (SEEDS; Tamura 2009) is a large-scale survey using adaptive optics (AO) assisted high-contrast imaging for studying planets and disks, from primordial and transitional systems (e.g., Kusakabe et al. 2012; Muto et al. 2012; Grady et al. 2013) to mature systems. A sub-survey of this larger effort concerns the study of debris disk systems. This study has several purposes, including: (1) searching for direct light from debris disks, in the sense of acquiring spatially resolved images of disks that have previously only been identified from infrared excess, (2) searching for planets in systems with known debris disks, and (3) studying interactions and correlations between planets and debris disks. Interestingly, many of the recently imaged planets coincide with debris disks (Marois et al. 2008, 2010; Lagrange et al. 2009). Many disks also have morphological indications of the presence of dynamical influence from planets in the system (e.g., Hines et al. 2007; Buenzli et al. 2010; Thalmann et al. 2011; Currie et al. 2012b; Quanz 2013), such as eccentric gaps with sharp inner boundaries or apparently resonant dust concentrations (e.g., Quillen & Thorndike 2002; Quillen 2006), although alternative mechanisms have been suggested (e.g., Jalali & Tremaine 2012; Lyra & Kuchner 2012). Thus, stars hosting debris disks are promising targets for imaging of massive exoplanets.

In previous publications, we have presented two results from the debris disk survey, in the form of spatially resolved disks around HR 4796 A (Thalmann et al. 2011) and HIP 79977 (Thalmann et al. 2013). Here, we will summarize the results from the rest of the survey so far, including images of spatially resolved disks and detection limits for planets which are interpreted in the context of the disk architecture in the system, and which form part of the basis for a statistical study that is presently in progress (T. D. Brandt et al., in preparation). In the following, we first describe the target selection in Section 2 and the observations and data reduction in Section 3, followed by a presentation of the results in Section 4. We discuss and summarize our results in Section 5.

A master list of targets was compiled from a wide range of literature sources identifying debris disk host stars based on infrared excess as measured by telescopes such as IRAS and Spitzer (e.g., Rieke et al. 2005; Rhee et al. 2007; Trilling et al. 2008; Plavchan et al. 2009). Targets for specific SEEDS runs were then selected continuously from this list, prioritized on the basis of disk properties (fractional luminosity and predicted angular separation) and possible planet properties (ease of detection, based on proximity and youth, as well as stellar mass assuming a constant typical planet–star mass ratio). Special emphasis was placed on cold disks, characterized by the presence of dust at large physical separations but indications of gaps or cavities at smaller separations. Such gaps could be caused by planets (see, e.g., Apai et al. 2008 and references therein.), which could in turn be observable in high-contrast images. A few warm disks however were also observed—these could have planets at larger separations, and the disk in such systems should be highly luminous at small separations, where HiCIAO performs the most competitively. Some high-profile planet-search targets were purposefully omitted—these are cases where specialized deep observations have been performed in dedicated studies, upon which it would be difficult or impossible to improve in a general survey with a 1 hr observation in the H band. In particular, this is true for the targets Eri and Vega (Janson et al. 2008; Heinze et al. 2008). The special case of Fomalhaut (e.g., Kalas et al. 2008; Janson et al. 2012; Currie et al. 2012a; Galicher et al. 2013; Kenworthy et al. 2013) was also omitted for this reason. Histograms for the spectral type, distance, and age of the targets are shown in Figure 1.

Zoom In Zoom Out Reset image size

The observations were carried out as part of the SEEDS program at the Subaru telescope, using the HiCIAO camera (Tamura et al. 2006; Hodapp et al. 2008) with the AO188 AO system (Hayano et al. 2008). The bulk of observations were taken throughout 2011 and 2012, with some observations also taken in 2009 and 2010 (see Table 1). No mask was used, but the detector was instead allowed to saturate at the point spread function (PSF) core, typically out to a radius of 03. All observations made use of the angular differential imaging (ADI) technique (Marois et al. 2006) with the pupil fixed on the detector, and were performed using the H-band filter, with a central wavelength of 1.65 μm and a bandwidth of 0.29 μm. In most cases, the instrument was set to direct imaging, but in a few cases, the polarimetric differential imaging (PDI) mode was used, in which the beam is split into two orthogonal polarization states using a Wollaston prism, with each corresponding image mapped onto one half of the detector. In those cases, the results presented here are based on separate reductions of each polarization state, which we then average together. We do not include any PDI reductions in this study. The typical telescope time spent on a target was ∼1 hr including overheads.

The ADI reductions were uniformly performed using the ACORNS-ADI pipeline (Brandt et al. 2013), with the same procedure as given in the ACORNS paper. As a brief summary, the data were destriped,²⁹ flat fielded, and corrected for field distortion. Relative centroiding was done using PSF fitting on non-saturated parts of the PSF, and absolute centering was based on visual inspection with a ∼0.5 pixel precision. PSF subtraction was performed with a Locally Optimized Combination of Images (LOCI) based scheme (Lafrenière et al. 2007b). As LOCI parameters, we used a PSF FWHM of 6 pixels, an angular protection zone of 0.7 FWHM, and 200 PSF footprint optimization regions. Individual PSF-subtracted frames were de-rotated and combined using a trimmed mean approach to produce the final image (see Figure 2 for an example). For each final image of a given target, a signal-to-noise (S/N) map was produced by dividing the signal at all positions by the local noise (calculated in an annulus at the corresponding separation). In this process, we include a correction for the signal attenuation imposed by the LOCI algorithm. The S/N map provides a data format in which point sources can be easily identified and in which it can be determined whether or not they are statistically significant. Detection limits for a 5.5σ criterion were produced by normalizing the radial noise profiles by the primary brightness, which was determined from non-saturated exposures acquired before and after each ADI sequence.

Zoom In Zoom Out Reset image size

In cases where candidates were present in the HiCIAO images and the targets had been previously observed with AO-assisted imagers, we analyzed the archival images using similar procedures as above, but adapted to the respective telescopes and instruments. Images from Gemini/NIRI (Hodapp et al. 2003), Gemini/NICI (e.g., Artigau et al. 2008), Keck/NIRC2 (e.g., McLean & Sprayberry 2003), Subaru/IRCS (Kobayashi et al. 2000) and HST/NICMOS (e.g., Schultz et al. 2003) were used for this purpose, saving several hours of Subaru telescope time that would otherwise have been necessary for executing follow-up observations in those cases, and thus demonstrating the broad utility of archiving data from large telescopes.

4.1. General Results

As can be generally expected in a survey of this kind, many faint point sources are detected in the images, the majority of which are physically unrelated background stars. Due to the fact that the contaminant fraction increases rapidly with angular separation from the parent star, small angular separations have been prioritized for follow-up. Companion candidates that were detected in the data inside of 5'' and with a >5.5σ significance were checked for common proper motion by either using archival data when available, or second epoch observations over a ∼1 yr baseline (see Figure 3 for an example). No substellar companions have been verified so far among the targets. One target remains for which candidates inside of the priority region have not yet been followed up; HD 162917 was observed in late 2012 and will be re-observed at a later stage. Given the low galactic latitude of this target, the candidates are likely to be background stars. The point sources are listed in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Properties of the Imaged Point Sources

HD ID	CC	ΔH	Δ R.A.	Δ Decl.	Epoch
		(mag)	('')	('')
HD 15745	1	11.1 ± 0.1	−1.85 ± 0.01	−0.63 ± 0.01	2011 Sep 6
HD 60737	1	10.3 ± 0.1	6.29 ± 0.01	−3.02 ± 0.01	2012 Jan 1
HD 69830	1	13.4 ± 0.1	−5.73 ± 0.01	−3.91 ± 0.01	2010 Jan 23
HD 70573	1	13.8 ± 0.2	2.61 ± 0.01	−2.24 ± 0.01	2011 Jan 30
HD 73350	1	11.7 ± 0.1	2.90 ± 0.01	5.23 ± 0.01	2011 Dec 30
HD 73752	1	1.2 ± 0.1	0.67 ± 0.01	0.80 ± 0.01	2011 Mar 25
HD 73752	2	13.7 ± 0.1	−4.50 ± 0.01	6.02 ± 0.01	2011 Mar 25
HD 88215	1	14.5 ± 0.1	−7.47 ± 0.01	−0.89 ± 0.01	2011 Dec 31
HD 104860	1	12.1 ± 0.1	−3.10 ± 0.01	−0.55 ± 0.01	2012 Apr 11
HD 106591	1	15.1 ± 0.1	3.22 ± 0.01	−1.25 ± 0.01	2010 Jan 25
HD 106591	1	15.1 ± 0.1	3.08 ± 0.01	−1.30 ± 0.01	2011 Jan 28
HD 106591	2	15.9 ± 0.1	1.26 ± 0.01	−5.57 ± 0.01	2010 Jan 25
HD 106591	2	16.1 ± 0.2	1.06 ± 0.01	−5.59 ± 0.01	2011 Jan 28
HD 107146	1	14.9 ± 0.1	−3.69 ± 0.01	−5.07 ± 0.01	2011 Mar 25
HD 113337	1	13.6 ± 0.1	4.88 ± 0.01	−0.10 ± 0.01	2011 May 21
HD 113337	1	13.4 ± 0.1	4.97 ± 0.01	−0.13 ± 0.01	2012 Feb 27
HD 128311	1	12.4 ± 0.1	4.33 ± 0.01	−6.38 ± 0.01	2012 Feb 27
HD 141569	1	2.2 ± 0.1	−5.50 ± 0.01	5.23 ± 0.01	2011 Mar 26
HD 161868	1	14.2 ± 0.1	−6.10 ± 0.01	−0.10 ± 0.01	2012 Jul 11
HD 161868	2	14.8 ± 0.1	6.05 ± 0.01	3.89 ± 0.01	2012 Jul 11
HD 162917	1	12.0 ± 0.1	2.46 ± 0.01	−1.67 ± 0.01	2012 Jul 9
HD 162917	2	12.5 ± 0.1	−2.73 ± 0.01	2.06 ± 0.01	2012 Jul 9
HD 162917	3	12.8 ± 0.1	0.35 ± 0.01	−4.41 ± 0.01	2012 Jul 9
HD 175742	1	10.6 ± 0.1	1.72 ± 0.01	1.97 ± 0.01	2011 May 23
HD 175742	1	10.8 ± 0.1	1.59 ± 0.01	2.24 ± 0.01	2012 May 11
HD 183324	1	13.7 ± 0.1	−0.73 ± 0.01	1.71 ± 0.01	2012 Jul 10
HD 183324	2	14.6 ± 0.1	3.25 ± 0.01	1.36 ± 0.01	2012 Jul 10
HD 183324	3	14.6 ± 0.1	3.40 ± 0.01	−1.16 ± 0.01	2012 Jul 10
HD 183324	4	13.6 ± 0.1	1.49 ± 0.01	−4.29 ± 0.01	2012 Jul 10
HD 183324	5	13.9 ± 0.1	−4.44 ± 0.01	−3.15 ± 0.01	2012 Jul 10
HD 183324	6	15.2 ± 0.2	5.00 ± 0.01	4.43 ± 0.01	2012 Jul 10
HD 192263	1	13.6 ± 0.1	−4.41 ± 0.01	−5.83 ± 0.01	2012 May 14
HD 206860	1	15.3 ± 0.2	1.69 ± 0.01	2.45 ± 0.01	2011 Aug 3
HD 281691	1	1.7 ± 0.1	4.33 ± 0.01	5.22 ± 0.01	2012 Nov 7

Download table as:  ASCIITypeset image

In some cases, the debris disk itself could be spatially resolved in our images. Two of these detections have been analyzed in particular detail and published separately: HR 4796 A (Thalmann et al. 2011) and HIP 79977 (Thalmann et al. 2013). Three other targets for which secure disk detections could be made are HD 15115, AU Mic, and HD 141569. These cases are discussed in the individual notes below. The disk detection space of our survey has a very good complementarity to that of the Hubble Space Telescope (HST). HST is able to observe at visible wavelengths with exquisite sensitivity and has a PSF that is unaffected by the atmosphere, which means that it can observe faint and smooth disk emission. Such emission is much more difficult to observe from the ground, since our near-infrared observations are more sensitivity limited, in addition to the fact that PSF variations due to varying seeing are very similar in their characteristics to smooth disk material. In addition, this study has made use of ADI, which benefits the detection of sharp features in the disk while strongly self-subtracting smooth emission, particularly if it is azimuthally symmetric. On the other hand, the high contrast and spatial resolution of HiCIAO allows for detection of disks and disk features at small angular separations, where HST is unable to provide a comparable performance. Hence, we are unable to detect large-scale, smooth, and low-inclination structures such as the second ring of the HD 141569 disk, but can provide novel results on small-scale, sharp and high-inclination features such as the inner region of the HIP 79977 disk.

The contrast performances for point sources are shown in Figure 4 and listed in Table 3. The achievable contrast is largely dependent on the field rotation during an observation. This is due to the connection between field rotation and ADI performance. An increased total field rotation benefits ADI because it maximizes the number of reference frames in which the planet signature is sufficiently separated from its location in the target frame to be useful, which helps as long as the number of reference frames does not become so large that the LOCI optimization becomes overconstrained. In our reductions, we avoid this overconstraining by limiting the number of reference frames for each target frame to ∼80, uniformly spread across the observing sequence, which we find to produce roughly optimal performance. The contrast depends not only on the total field rotation, but also on the rotation rate. This is caused by the fact that frames taken over a small time span tend to correlate better than frames taken over larger time spans. A larger rotation rate thus allows for the use of reference frames that are better correlated with the target frame. The rotation rate that can be acquired depends on the declination of the target—a minimal |δ − l| provides a maximal rotation rate, where δ is the declination and l the latitude of the telescope.

Zoom In Zoom Out Reset image size

Table 3. Contrast at a Range of Angular Separations

HD ID	Ep.	025	05	075	10	15	20	30	50
		(mag)	(mag)	(mag)	(mag)	(mag)	(mag)	(mag)	(mag)
HD 377	1	...	...	9.6	11.3	12.9	13.9	14.1	14.3
HD 7590	1	...	8.8	10.7	12.4	14.0	14.6	15.0	15.2
HD 8907	1	7.8	9.5	11.5	13.1	14.7	15.3	15.6	15.7
HD 9672	1	...	8.7	10.3	11.6	13.2	13.9	14.4	14.6
HD 10008	1	7.0	8.8	10.6	12.1	13.8	14.4	14.7	14.9
HD 12039	1	...	8.7	10.7	12.1	13.6	14.1	14.4	14.6
HD 15115	1	5.8	7.8	9.6	11.0	11.9	12.2	12.3	12.4
HD 15115	2	7.1	9.3	11.4	12.9	14.0	14.3	...	...
HD 15745	1	7.6	9.2	11.2	12.7	14.1	14.5	14.9	15.0
HD 17925	1	...	8.8	10.8	12.3	14.2	15.1	15.6	15.8
HD 25457	1	...	9.4	11.2	12.9	14.3	14.8	15.3	15.4
HD 281691	1	7.4	9.7	11.0	11.9	12.2	12.4	12.5	12.5
HD 31295	1	5.0	6.5	7.7	8.8	10.6	11.8	12.6	12.8
HD 40136	1	...	10.4	12.2	13.7	15.4	16.1	16.9	17.1
HD 60737	1	...	8.7	10.8	12.3	13.9	14.4	14.8	14.9
HD 69830	1	...	8.7	10.5	11.9	13.7	14.6	14.9	15.1
HD 70573	1	6.5	8.6	10.6	12.0	13.3	13.6	13.9	14.0
HD 73350	1	7.7	9.0	10.7	12.1	13.7	14.6	15.1	15.3
HD 73752	1	...	6.4	6.9	6.8	10.4	12.5	13.8	14.2
HD 72905	1	...	9.0	10.9	12.4	14.1	14.7	15.1	15.2
HD 76151	1	...	8.9	10.6	12.0	13.6	14.3	14.6	14.7
HD 88215	1	...	8.8	11.2	12.7	14.4	15.2	15.6	15.8
HD 91312	1	...	9.3	11.3	12.8	14.6	15.3	15.6	15.7
HD 92945	1	...	7.2	9.0	10.3	12.0	12.9	13.3	13.5
HD 102647	1	...	...	11.8	13.6	15.4	16.3	16.9	17.1
HD 104860	1	...	10.3	12.2	13.6	14.8	15.2	15.4	15.7
HD 106591	1	...	8.9	11.5	13.1	15.1	16.0	16.4	16.5
HD 106591	2	...	9.5	11.4	12.9	15.1	15.8	16.3	16.5
HD 107146	1	6.9	8.8	10.8	12.4	13.8	14.4	...	...
HD 107146	2	7.5	9.1	11.0	12.5	14.2	14.8	15.0	15.1
HD 109085	1	...	...	10.8	12.3	14.1	15.0	15.4	15.8
HD 109573	1	...	9.8	11.5	12.4	14.6	15.2	15.6	15.7
HD 110411	1	...	8.8	10.7	12.1	14.2	15.0	15.4	15.5
HD 112429	1	...	9.4	11.3	13.0	14.6	15.1	15.4	15.6
HD 113337	1	...	9.6	11.4	12.9	14.5	15.2	15.5	15.7
HD 113337	2	...	9.8	11.4	12.9	14.5	15.2	15.7	15.7
HD 125162	1	...	7.9	9.6	11.1	13.3	14.3	15.0	15.2
HD 127821	1	...	8.6	10.0	11.4	13.2	14.0	14.5	14.6
HD 128167	1	...	10.1	12.0	13.6	15.2	16.0	16.4	16.5
HD 128311	1	...	9.5	11.4	12.7	14.4	15.0	15.5	15.6
HD 135599	1	8.4	10.7	12.5	13.6	14.9	15.3	15.6	15.6
HD 135599	2	...	10.4	12.6	13.9	15.3	15.7	16.1	16.2
HD 139006	1	...	9.9	12.1	13.7	15.6	16.4	16.9	...
HD 139664	1	...	8.4	10.0	11.5	13.4	14.4	15.2	15.5
HD 141569	1	8.7	9.9	11.9	13.0	14.0	14.4	14.6	14.7
HD 146897	1	...	9.4	11.2	12.2	13.2	13.5	13.7	13.8
HD 146897	2	...	8.4	10.5	11.6	12.7	13.1	13.2	...
HD 152598	1	...	9.6	11.7	13.3	14.6	15.0	15.3	15.4
HD 161868	1	...	9.1	11.0	12.6	14.2	15.0	15.4	15.6
HD 162917	1	...	8.8	10.7	12.1	13.8	14.5	14.8	15.1
HD 175742	1	7.8	10.2	12.1	13.5	14.6	15.0	15.3	15.4
HD 175742	2	...	10.3	12.4	13.8	14.9	15.3	15.5	15.7
HD 183324	1	...	9.7	11.6	13.0	14.3	14.8	15.2	15.3
HD 192263	1	...	10.2	12.1	13.4	14.7	15.1	15.3	15.5
HD 197481	1	...	8.5	11.1	12.5	14.2	14.8	15.5	...
HD 206860	1	...	10.0	12.0	13.5	14.9	15.5	15.7	15.8
HD 207129	1	...	9.2	11.1	12.6	14.2	14.9	15.3	15.4

Download table as:  ASCIITypeset image

4.2. Individual Targets

Below, we list individual notes concerning the results on different targets in the survey, such as detections of disks or point sources, as well as other details from the scientific literature that are relevant for the context.

HD 15115 (HIP 11360). This star has a known debris disk that has been spatially resolved at several near-infrared wavelengths (e.g., Kalas et al. 2007b; Debes et al. 2008; Rodigas et al. 2012). We also detect the disk in the HiCIAO data (see Figure 5), but at limited S/N which does not improve on the results in previous studies.

Zoom In Zoom Out Reset image size

HD 15745 (HIP 11847). The debris disk around HD 15745 has been spatially resolved in HST observations (Kalas et al. 2007a), but is not visible in the HiCIAO images due to its smooth and azimuthally extended features. A candidate companion was seen at Δα = −185 and Δδ = −063 with HiCIAO. The point source is faint but visible in the archival HST images from 2004, where it has Δα = −165 and Δδ = −089, demonstrating that it is a background contaminant. There is also an intermediate epoch available from Keck in 2007, with the point source located at Δα = −172 and Δδ = −075, further confirming this conclusion.

HD 60737 (HIP 37170). The field of HD 60737 is empty except for a point source at Δα = 629 and Δδ = −302, which has already been identified as a background star by Metchev & Hillenbrand (2009).

HD 69830 (HIP 40693, LHS 245). This system is notable for its planetary system which contains three known planets so far, all with Neptune-like masses (Lovis et al. 2006). It also hosts a warm debris disk (Beichman et al. 2006), which has been resolved with interferometry (Smith et al. 2009). Our images do not reveal the disk, and due to the probably quite old age of the system (approximately 6 Gyr; Mamajek & Hillenbrand 2008) and small physical scale (∼1–2 AU) of the dust location, no stringent constraints can be drawn regarding planets near the disk edge from the imaging. We do detect a point source at Δα = −573 and Δδ = −391. This candidate is visible in an archival HST image from 2007 with Δα = −524 and Δδ = −594, hence it is a physically unrelated background star.

HD 70573 (V748 Hya). There is an object at Δα = 261 and Δδ = −224 in the HiCIAO images. Although the source appears somewhat extended, we nonetheless examined archival data to test its nature. This turned up the object in archival NICI images, where it is located at Δα = 247 and Δδ = −238, indeed, implying non-common proper motion. This star has a planet candidate from radial velocity measurements, at a semi-major axis of 1.8 AU (Setiawan et al. 2007).

HD 73350 (HIP 42333, V401 Hya). There is a point source at Δα = 290 and Δδ = 523 in the HiCIAO data. It is considered of low priority due to its relatively large separation from the primary.

HD 73752 (HIP 42430, LHS 5139). A known binary (e.g., Mason et al. 2001), the location of the secondary relative to the primary in the HiCIAO images is Δα = 067 and Δδ = 080. There is another possible candidate in the image at Δα = −450 and Δδ = 602, but it is just at the edge of the detector, hence it is considered of low priority.

HD 88215 (HIP 49809, HR 3991). An extended source is present at Δα = −747 and Δδ = −089, which is probably a background galaxy.

HD 92945 (HIP 52462, V419 Hya). The debris disk around HD 92945 has been recently spatially resolved with HST (Golimowski et al. 2011). It is not visible in the HiCIAO images.

HD 104860 (HIP 58876). The only point source in the field of HD 104860 is located at Δα = −310 and Δδ = 055, and has already been identified as a background star in Metchev & Hillenbrand (2009).

HD 106591 (HIP 59774, δ Uma). This star was observed in two separate epochs. Two point sources are present in the images. The brighter of the candidates resides at Δα = 322 and Δδ = −125 in the first epoch and Δα = 308 and Δδ = −130 in the second epoch. The fainter one is located at Δα = 126 and Δδ = −557 in the first epoch and Δα = 106 and Δδ = −559 in the second epoch. Neither is therefore physically bound to HD 106591. The brighter candidate however displays a peculiar astrometric behavior, with a deviation of close to 100 mas from the trajectory of a static background star over a baseline of one year. This could imply that it is a field brown dwarf at a similar distance as HD 106591, or otherwise that it is a distant background star with an anomalously high proper motion.

HD 107146 (HIP 60074, NLTT 30317). The debris disk around HD 107146 has been spatially resolved in the past (e.g., Ardila et al. 2004; Ertel et al. 2011), but since it is smooth and has a nearly face-on orientation, it is not visible in the HiCIAO images. An object is visible at Δα = −369 and Δδ = −507, which has been classified as a background galaxy in Ertel et al. (2011).

HD 109573 (HR 4796 A, HIP 61498). As described in Thalmann et al. (2011), we have spatially resolved the disk in this system using ADI, which enabled us to confirm and strengthen conclusions from previous studies of the system (e.g., Schneider et al. 1999, 2009), such as the fact that the disk has a non-zero eccentricity. As is also shown in Thalmann et al. (2011), a planet near the gap edge (coplanar with the disk) would have been detectable at a mass of ∼3 M_Jup at maximum projected separation, but at minimum projected separation the upper limit is much softer (∼17 M_Jup) due to the relatively high inclination of the target.

HD 113337 (HIP 63584, HR 4934). There are two epochs of observation available for HD 113337 from HiCIAO, due to the presence of a companion candidate in the data. The candidate has Δα = 488 and Δδ = −010 in the first epoch and Δα = 497 and Δδ = −013 in the second epoch, which demonstrates that it is a background star. Furthermore, archival data from NIRC2 in 2010 places the candidate at Δα = 474 and Δδ = −007, further strengthening this conclusion.

HD 128311 (HIP 71395, HN Boo). The single candidate that can be seen in the HiCIAO field at Δα = 433 and Δδ = −638 has been established as a background star in Heinze et al. (2010).

HD 139664 (HIP 76829, NLTT 40843). A spatially resolved scattered light HST image of the debris disk around HD 139664 exists (Kalas et al. 2006). Although the disk has a high inclination, it appears that it was too faint to be detectable in the HiCIAO images.

HD 141569 (HIP 77542). Despite the fact that the disk around HD 141569 is smooth and has a relatively low inclination, it is nonetheless visible in our HiCIAO images (see Figure 6) due to the high surface brightness. As expected, the S/N is lower than in HST images of the target (Clampin et al. 2003). For point sources on the other hand, HiCIAO provides strong limits, with sensitivity down to 1 M_jup planets in the sensitivity-limited region. The already known binary companion (Weinberger et al. 2000) is present toward the edge of the field of view.

Zoom In Zoom Out Reset image size

HD 146897 (HIP 79977). This USco (Upper Scorpius OB association) member has a debris disk (Chen et al. 2006), which was spatially resolved for the first time with HiCIAO, as we reported in Thalmann et al. (2013). The disk has an inner gap within ∼40 AU (Chen et al. 2011), but owing to the large distance of 123 pc to the target (van Leeuwen 2007), the gap itself cannot be confidently distinguished in the existing data, and any giant planet that might be responsible for the gap would have been easily missed, particularly since the disk orientation is close to edge-on.

HD 161868 (HIP 87108, γ Oph). The star γ Oph is in a relatively crowded field with several background stars, although all are outside of 5'' separation. The HiCIAO image is relatively shallow and does not reveal as many candidates as archival NICI data from 2009. However, there are two candidates that overlap between the two data sets. One candidate has Δα = −610 and Δδ = −010 in the HiCIAO data and Δα = −618 and Δδ = −029 in the NICI data, and the other has Δα = 605 and Δδ = 389 in the HiCIAO data and Δα = 595 and Δδ = 363 in the NICI data. As expected from the large separations, both candidates are background stars.

HD 175742 (HIP 92919, V775 Her). There is an object in the field which, judging by its morphology, is probably a close background binary star. It has been observed in two HiCIAO epochs with Δα = 172 and Δδ = 197 in the first epoch and Δα = 159 and Δδ = 224 in the second, confirming its physically unrelated status. In addition, there is an archival epoch from NIRC2 in 2010 where the candidate is located at Δα = 190 and Δδ = 181, further strengthening this conclusion.

HD 183324 (HIP 95793, V1431 Aql). The brightest and closest companion candidate to HD 183324 has been observed several times with 8 m class telescopes. In the HiCIAO data, it is located at Δα = −073 and Δδ = 171. In archival H-band Keck/NIRC2 images from 2010, it is located at Δα = −073 and Δδ = 163. The motion clearly demonstrates that the candidate is a physically unrelated background object. There are also three other candidates inside of 5'' in the data: one in the northeast located at Δα = 325 and Δδ = 136 in the HiCIAO image and Δα = 324 and Δδ = 129 in the Keck image, one in the northwest located at Δα = 340 and Δδ = −116 in the HiCIAO image and Δα = 340 and Δδ = −120 in the Keck image, and one toward the south located at Δα = 149 and Δδ = −429 in the HiCIAO image and Δα = 150 and Δδ = −435 in the Keck image. Hence, all of these are also physically unrelated to the target star.

HD 192263 (HIP 99711, V1703 Aql). Aside from its debris disk, HD 192263 also hosts a planet candidate detected through radial velocity (e.g., Santos et al. 2003). In Chauvin et al. (2006), it is mentioned that several candidates have been discovered and confirmed to be background stars in NACO images of HD 192263. We observe one of these objects within the HiCIAO field of view at Δα = −441 and Δδ = −583, and otherwise no new objects.

HD 197481 (HIP 102409, AU Mic). Best known as AU Mic, this star has a well known debris disk which shows up clearly in our data (see Figure 7). The field of view is smaller than for most stars in our sample, due to the PDI setting that was used for this observation (see Section 3).

Zoom In Zoom Out Reset image size

HD 206860 (HIP 107350, HN Peg). A candidate is visible at Δα = 169 and Δδ = 245. We retrieved the companion in NIRI data from 2006, where the candidate is located at Δα = 291 and Δδ = 193, consistent with a background star.

HD 207129 (HIP 107649, NLTT 52100). Spatially resolved images in scattered light of HD 207129 have been acquired with HST (Krist et al. 2010). It is spatially extended and very faint, and hence, as expected, it is invisible in the HiCIAO images.

HD 281691 (V1197 Tau). We observe a previously known companion in the HiCIAO images at Δα = 433 and Δδ = 522, which was first discovered by Köhler & Leinert (1998) and has been confirmed by Metchev & Hillenbrand (2009).

Both the β Pic and HR 8799 systems have debris disks with gaps or cavities in them, and directly imaged planets that are consistent with being responsible for carving these features. Given that we are sensitive to similar mass planets in our observations, and given that we have constraints on the semi-major axis space where the gaps originate from the spectral energy distributions (SEDs) of the targets, it is possible to address to which extent similarly massive planets are responsible for debris disk gaps in general. Given the many caveats involved in such a study, however, such an analysis should be treated with caution.

One primary issue in the analysis is the uncertainty in the location of the gap. It is possible to constrain the spatial distribution of the circumstellar dust from the SED by constraining the temperature, but since only a very limited number of data points are available in general, there are ambiguities between the location and the radiative properties of the dust. In this study, we adopt values of a_dust from the literature based on the global assumption that the dust grains emit like blackbodies. How the resulting physical separation relates to the semi-major axis of a given hypothetical shepherding planet in the system is another complex uncertainty. Here, we simply take the a_dust itself to represent the physical scale around which we wish to evaluate the presence or absence of a planet; the motivation being that the dominating disk flux should arise close to the inner edge (since that is where the dust is hottest and, in general, most dense) and that the planet responsible for carving the gap should be close to the edge. This is not necessarily relevant if, for instance, there are multiple planets responsible for the gap. With regards to planet detectability near the gap, the gap locations adopted here are probably very conservative, as can be seen in Booth et al. (2013). In all cases studied by Booth et al. (2013) where the real gap location could be observed, the real location is never smaller than the blackbody prediction, but is often larger by a factor two.

We derive mass detection limits from the contrast curves using COND- and DUSTY-based evolutionary models (Chabrier et al. 2000; Allard et al. 2001; Baraffe et al. 2003) and the age limits in Table 4 (see Figure 8). COND was used whenever the predicted temperature was below 1700 K, and DUSTY when it was above this limit. These "hot-start" models may overpredict the brightness for a given mass and age if the initial entropy is lower than assumed in those models (see, e.g., Spiegel & Burrows 2012). However, the exoplanets that have been discovered to date are consistent with hot-start conditions and exclude at least the coldest ranges of initial conditions (e.g., Janson et al. 2011; Bonnefoy et al. 2013; Marleau & Cumming 2013). Furthermore, the absence of heating from deuterium burning in the COND/DUSTY models may conversely underpredict the brightness for a given mass and age (Mollière & Mordasini 2012). Nonetheless, the uncertainties in mass–luminosity relationships is a further uncertainty that should be kept in mind.

Zoom In Zoom Out Reset image size

In order to put the issue of gap-opening super-Jupiters in a statistical context, we evaluate the probability that planets with masses similar to that of β Pic b of ∼10 M_jup (Bonnefoy et al. 2013 and T. Currie et al. in preparation) would be detectable near the gap in each observed system. This is done by calculating the full projected separation distribution corresponding to a semi-major axis of a_dust for random orbital orientations and a uniform eccentricity distribution between 0.0 and 0.6 (Janson et al. 2011; Bonavita et al. 2012). The fraction of 10 M_jup planets that are detectable in a given system is denoted f_y for the lower limit of the age of the star and f_o for the upper limit. The individual values of f_y and f_o are listed in Table 4. In some of the systems, such a planet is simply not detectable (0% in both f_y and f_o), while in the best cases the fraction is close to 100%. From the collection of these values and the fact that no planets were detected in the sample, we can estimate an upper limit on the frequency of planets with equal or higher mass than β Pic b near the estimated gap edge, using Bayes theorem following the procedure in Janson et al. (2011). As a result, we find that at 95% confidence, <15.2% of the stars host such planets in the extreme case where the younger age limit is adopted in all cases, and <30.1% in the opposite case where the upper limits are adopted. In other words, if giant planets are a dominant cause of gaps in debris disks, then the majority of them must be less massive than β Pic b. In either case, it implies that β Pic b is probably in the upper mass range of any gap-causing planets that may exist.

An illustration of typical mass detection limits around a_dust for the individual stars is shown in Figure 9, where the detectable mass is evaluated at α_dust = a_dust/d/1.26, which represents the average angular separation of a planet with semi-major axis a_dust for random orbital orientations (Fischer & Marcy 1992). Here, d denotes the distance to the target. The mass limits in our survey are contrast-limited rather than sensitivity-limited, hence it would be possible to substantially enhance the limits with upcoming extreme AO-assisted instruments such as SPHERE, GPI, or CHARIS (Beuzit et al. 2008; Macintosh et al. 2008; Peters et al. 2012). These facilities may thus be able to detect a large number of gap-opening super-Jupiters if they are relatively common, or otherwise put yet more stringent limits on their presence and properties.

Zoom In Zoom Out Reset image size

In this study, we have presented high-contrast imaging of a sample of 50 stars primarily in the G–A-type range with known infrared excess due to debris disks, using the HiCIAO camera at the Subaru telescope. Targets were particularly selected if they had excess only at long wavelengths, implying cold debris disks with an inner gap, possibly carved out by massive planets within the disk. The targets were observed both in order to attempt to spatially resolve the disk, as well as to try to detect the putative planets that may be responsible for the disk morphology. No planets were discovered, despite the fact that β Pic b-like planets (∼10 M_jup) could have been detected near the estimated gap edges in many cases. This led to an upper limit of 15%–30% on the frequency of such planets, implying that if planets are a general cause of the commonly existing gaps in debris disk systems, then they must generally be lower in mass than β Pic b. Five debris disks have been spatially resolved during the survey, two of which have already been presented in previous publications (Thalmann et al. 2011, 2013). Future studies with upcoming instrumentation will be able to put yet more stringent constraints on planet occurrences in debris disk systems, by probing down to smaller planetary masses and smaller semi-major axes, and thus may conclusively address whether the gaps in debris disks are typically caused by planets, or whether other mechanisms dominate the disk architecture.

Support for this work was provided by NASA through Hubble Fellowship grant HF-51290.01 awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. J.C. was supported by NSF award 1009203. Archival data from the Subaru, Gemini, Keck and Hubble telescopes have been used as part of this study. We acknowledge the cultural significance of Mauna Kea to the indigenous population of Hawaii. This study made use of the CDS services SIMBAD and VizieR, as well as the SAO/NASA ADS service.