arXiv:0904.3936v1 [astro-ph.EP] 25 Apr 2009
Submitted to ApJ April 24, 2009
A Search for Wide Companions to the Extrasolar Planetary
System HR 8799
Laird M. Close and Jared R. Males
Steward Observatory, University of Arizona, Tucson, AZ 85721
The extrasolar planetary system around HR 8799 is the first multiplanet
system ever imaged. It is also, by a wide margin, the highest mass system with
> 27 Jupiters of planetary mass past 25 AU. This is a remarkable system with
no analogue with any other known planetary system. In the first part of this
paper we investigate the nature of two faint objects imaged near the system.
These objects are considerably fainter (H=20.4, and 21.6 mag) and more distant
(projected separations of 612, and 534 AU) than the three known planetary
companions b, c, and d (68-24 AU). It is possible that these two objects could
be lower mass planets (of mass ∼ 5 and ∼ 3 MJup) that have been scattered
to wider orbits. We make the first direct comparison of newly reduced archival
Gemini adaptive optics images to archival HST/NICMOS images. With nearly a
decade between these epochs we can accurately assess the proper motion nature
of each candidate companion. We find that both objects are unbound to HR
8799 and are background. We estimate that HR 8799 has no companions of
H< 22 from ∼ 5 − 15′′. Any scattered giant planets in the HR 8799 system are
> 600 AU or less than 3 MJup in mass. In the second part of this paper we
carry out a search for wider common proper motion objects. While we identify
no bound companions to HR 8799, our search yields 16 objects within 1 degree in
the NOMAD catalog and POSS DSS images with similar (±20 mas/yr) proper
motions to HR 8799, three of which warrant follow-up observations.
Subject headings: planetary systems, stars: individual: HR 8799
There have been several surveys to directly image extrasolar planets for the ground with
adaptive optics (AO) and from space with HST. Until very recently all surveys returned null
– 2 –
results, and so it was generally assumed that wide, massive, extrasolar planets would be rare
(at least around Sun-like stars) past 20 AU (Lafreniere et al. 2007a; Nielsen et al. 2008 and
references within). However, in November 2008 Marois et al. 2008 announced the discovery
of 3 planets orbiting the A5V star HR 8799, based on Near-IR imaging at the Keck and
Gemini telescopes. Using data from Keck in 2004-2008 and Gemini in 2007-2008, they were
able to establish common-proper motions (assuming the non-common motion component of
∼ 25 mas/yr for b and c was due to orbital motion of the planets around the star). The
three planets, HR 8799b,c, and d orbit at approximately 68, 38 and 24 AU respectively.
Marois et al. 2008 estimate effective temperatures of 870, 1090, and 1090K for the three
planets, and arrive at estimates for mass of 7, 10, and 10 MJup. These estimates are based
on the estimated 30-160 Myr age of HR 8799 and hybrid theoretical cooling tracks (luminosity
vs. age) for giant planets (Marois et al. 2008).
It is possible that additional companions could be discovered during a search at wider
separations. The binary fraction of late A stars is at least 70% (Bate et al. 2009). Recently,
(Veras et al. 2009) have predicted a population of giant planets at large separations (100
AU - 10,000 AU) from stars hosting relatively close-in planets, where the distant objects are
dynamically scattered/pumped to large separations after system assembly. Such effects may
explain the very wide, low mass, companions GQ Lup b at > 100 AU (Neuhauser et al. 2005)
and/or AB Pic B at > 250 AU (Chauvin et al. 2005). There is certainly the possible ex-
istence of low mass, reddened, stellar companions that have not yet been detected. It is
even possible that HR 8799 has a small, common-proper motion group, around it. Given the
unique nature of the HR 8799 system, a search for wide (> 100 AU) companions is important
and motivated this paper.
2.Observations & Reductions
On Oct. 30 1998 UT HST/NICMOS observed HR 8799 with its coronagraph, and 2
candidate companions were identified (Lowrance et al. 2005) in the roll subtracted images.
These faint point sources where reported at 13.7′′and 15.7′′(540 AU and 619 AU) with H mag-
nitudes of 21.6 and 20.4, respectively. These are much wide separations than the 3 confirmed
planets which were not discovered in the NICMOS data at the time (Lafreniere et al. 2009).
Figure 1 shows our own roll subtraction of the otherwise already pipeline reduced NICMOS
data with the 2 objects identified. HR 8799’s high galactic latitude (b = −35o) indicates that
there is a finite chance one (or both) are not background objects (see Fig 2). However, we
need to determine if these are real common proper motion companions or simply background
objects. For lack of a better nomenclature we have elected to call the closest candidate “HR
– 3 –
8799B” and the farther one “HR 8799C”, however, the use of these labels does not imply
they are physical companions.
In October 25, 2007 (Marois et al. 2008) obtained 118x30s Angular Differential Imaging
(ADI; (Marois et al. 2006)) dataset of HR 8799 on the Gemini-North telescope with the
Altair AO system and NIRI NIR camera. Typically an ADI dataset would not be the ideal
method of imaging faint companions at 13.7′′and 15.7′′since there will be a risk of rotational
blurring of the images in the azimuthal direction. However, in this HR 8799 dataset the
integrations of 30s were short enough that only minimal azimuthal blurring occurred –and
only then in the fastest rotating images near transit. Therefore, we were able to create a
new custom “ADI-like” IRAF pipeline to produce “Wide Field” ADI images (WiFi ADI)
to image faint companions at the very edge of ADI datasets.
ADI is wide companions there is no need to use more advanced ADI reductions like LOCI
(Lafreniere et al. 2007b).
Since the focus of WiFi
Our WiFi ADI pipeline is very similar to standard ADI reduction and runs in a standard
IRAF environment. In ADI the telescope rotator is disabled and so the median of the images
gives an estimate of the master PSF without “contamination” from real objects on the sky.
Then one must subtract the master PSF off each individual frame after a cross correlation
alignment of each frame as in Close et al. 2002. Once the frames have the PSF removed they
need to be rotated by the parallactic angle and median combined (the subroutine to calculate
the rotation angle is a custom script developed for Gemini data in Close et al. 2003). One
way that the WiFi ADI pipeline differs from the standard is that it is optimized to preserve
any faint off-axis objects that might fall past the edge of the IR array in the majority of
the individual ADI frames. For example, the code accurately masks many bad pixels in
the corners of the NIRI array and uses a final median combine of all 118 re-rotated (master
PSF subtracted) images with no pixel clipping or rejection (so even the corner pixels of the
individual frames are utilized in the final WiFi image). We also carefully offset each image
by the mode of the outer region of each image, this allows the final median combine to be
most sensitive in the outer regions of the image.
Our WiFi ADI pipeline when used with the NIRI detector with its 0.0219”/pixel scale
is then capable of creating a round WiFi ADI FOV of 31.71′′in diameter when there is at
least 90 degrees of field rotation during the ADI observation, and the object is centered in
the detector. In the case of the Oct. 25 2007 Gemini HR 8799 data of Marois et al. 2008
the above assumptions are all true and our pipeline produced a final image (see Fig. 3) of
all H < 22 objects within ∼ 5 − 15′′of HR 8799.
– 4 –
As is clear from figure 1 and 3 there are two faint objects near HR 8799. Based on de-
tailed search of the literature and the VLT/Gemini/Subaru/HST archives we have concluded
that there has not been any published attempt to recover “B” or “C” until now. HR 8799A
has a total proper motion of 119 mas/year so the recovery of the two candidate companions
at the same separations as in 1988 wrt HR 8799A would be an unambiguous confirmation of
their physical association. These objects are both fainter than the HD8799b-d planets (they
would have masses of 3 and 5 Mjupon the 0.1 Gyr (Baraffe et al. 2002) COND tracks; which
predict reasonable luminosities at these ages for higher mass objects; Close et al. 2007b),
consistent with being scattered by the heavier, close-in, planets.
In Figure 4 and 5 the 2007 positions of these two faint companions are shown. In
both cases the current positions are much closer to the locations calculated for distant
background objects rather than physical companions. Therefore, our astrometry proves that
the NICMOS companions of Lowrance et al. 2005 appear to be faint background objects
unrelated to HR 8799A (see Table 1 for a detailed list of our astrometric measurements).
3.1.Does HR 8799b Show any Parallax Motion of a Background Object?
In the direction of Pegasus most nearby stars appear to be moving towards the East
South East. This is due to the Sun’s motion in the opposite direction wrt the LSR. In fact,
the Sun’s space motion causes a stationary object (wrt LSR) at 39.9 pc to have a measured
proper motion of 95.51 mas/yr to the East and 38.51 mas/yr to the South –based on values
for the Solar motion given by Jaschek & Valbousquet 1993. While it is clear that planets
HR 8799b and HR 8799c have similar proper motions on the sky to HR 8799A (see left side
of Fig. 6), once we subtract the Solar motion there is less agreement (see right side of Fig.
6). However, this lack of common proper motion can be explained by increased velocity due
to orbital motion of b and c around A (Marois et al. 2008).
For the planet HR 8799b (which has the largest timeline of observations), there appears
to be some “scatter” in its measured position from A from the nearly straight line expected
for b’s long period orbit. The exact solution for a stable orbit of the massive planets b, c,
and d is still somewhat uncertain (Fabrycky & Murray-Clay 2008).
In Fig. 7 we consider the question what would this “scatter” resemble if b was actually
a background object at 100 pc that had similar proper motions, by chance, to HR 8799A
(at a distance of 100 pc HR 8799b would roughly have the normal luminosity and colors of
a background L dwarf). In this model b’s position should show a “reverse” parallax w.r.t.
– 5 –
HR 8799A with an amplitude of 60% that of HR 8799A’s parallax. We calculate parallax
of HR8799 in the usual manner (Biller & Close 2007) and then multiply by -0.6 to calculate
In Fig. 7 we show two models for the nature of HR 8799b motion: a background
object at 100 pc; and a simple “linear” planetary arc. The small triangles in Fig. 7 denote
time stamps in the 100 pc model to each observation date over the 9.885 yr period since
the detection of b in a LOCI analysis of the NICMOS dataset by Lafreniere et al. 2009.
We note how neither the slightly curving “linear” orbit or the 100 pc background model
fit all data points simultaneously inside the 1σ errorbars. The “linear” orbit is a simpler
model with a better fit, but it certainly is not perfect since the reduced of χ2
an ∼ 8% chance that it is the correct model –assuming no unknown systematic errors in
the astrometry. On the other hand, a background source at 100 pc model can be rejected
with 99.95% confidence (χ2
fit for b, we cannot reject it (we also do not, at this time, know what orbit to fit it to
(Fabrycky & Murray-Clay 2008)). However, we can reject the hypothesis that this scatter
in HR 8799b’s position is reverse parallax due to it being a background object at 100pc
(assuming, of course, the reported astrometric values and errors are correct).
ν∼ 2 gives
ν∼ 4.4). While there is still significant scatter in the linear orbit
3.2.A Search For Other Common Proper Motion Companions to HR 8799
Even though the “B” and “C” objects are clearly background, there exists the possibility
of other wider companions to HR 8799A that might be found by searching nearby for similar
proper motion. We searched 12875 objects within ±1oof HR 8799 in the the US Naval
Observatory Merged Astrometric Dataset (NOMAD) catalog (Zacharias et al. 2004) with
non-zero proper motions. See Figure 8 for a plot of the resulting proper motion vectors.
Most objects inside 1oof HR 8799 are background and so do not have as large a proper
motion as HR 8799, which is dominated by the Sun’s motion w.r.t the LSR. In Figure 8 we
have highlighted those objects which have proper motion vectors within a circle of radius 20
mas/yr around HR8799’s proper motion. This yields a list of 15 objects, two of which may
warrant follow up observations. We further discuss these objects below.
To increase our sensitivity to fainter objects that might have been missed by previous
surveys we manually examined the Palomar Optical Sky Survey (POSS) POSS1 (1951 August
12 08:12:00 UT) and POSS2 (1991 October 02 05:47:00 UT) Red DSS images to search for
objects within 1 square degree of HR 8799A with similar proper motions. The POSS1 image
has a plate scale of 1.7′′per pixel, whereas POSS2 has 1.0′′per pixel. We first magnified the
POSS1 image to match the POSS2 plate scale. Next 25 background stars were selected, and