Issue |
A&A
Volume 495, Number 2, February IV 2009
|
|
---|---|---|
Page(s) | 523 - 535 | |
Section | Interstellar and circumstellar matter | |
DOI | https://doi.org/10.1051/0004-6361:200811067 | |
Published online | 20 January 2009 |
VLT/NACO coronagraphic observations of fine structures in the disk
of
Pictoris
A. Boccaletti1 - J.-C. Augereau2 - P. Baudoz1 - E. Pantin3 - A.-M. Lagrange2
1 - LESIA, Observatoire de Paris-Meudon, 92195 Meudon, France
2 - Laboratoire d'Astrophysique de Grenoble, Université Joseph Fourier, CNRS, UMR 5571, Grenoble, France
3 - Service d'Astrophysique, CEA Saclay, 91191 Gif-sur-Yvette, France
Received 1 October 2008 / Accepted 19 December 2008
Abstract
Aims. We present ground-based observations of the disk around the A-type star Pictoris to obtain scattered light images at the highest angular resolution (60 mas, equivalent to about 1 AU at the distance of the star) and the highest contrast in the very close environment of the star. The purpose of this program is to perform a close inspection of the inner disk morphology.
Methods. Images were collected with NACO, the AO-assisted near-IR instrument on the VLT (ESO) which includes two types of coronagraphs: classical Lyot masks and phase masks. In this program we took advantage of both types of coronagraphs in two spectral bands, H-band for the Lyot mask and Ks-band for the phase mask. The Lyot mask blocks a large central region around the star (radius <0.35
)
but allows deep integrations and hence good signal-to-noise ratio at large distances, while the phase mask allows imaging at very close separation (down to
in theory) but conversely is more sensitive to residual aberrations. In addition, we simulated an extended object to understand the limitations in deconvolution of coronagraphic images.
Results. The reduced coronagraphic images allow us to carefully measure the structures of the debris disk and reveal a number of asymmetries of which some were not reported before (position, elevation and thickness of the warp). Our analysis also demonstrates the advantage of the phase mask coronagraph to explore the very close environment of stars. In this program, the circumstellar material is visible as close as 0.7
(13.5 AU) owing to the phase mask while the Lyot mask generates artifacts which hamper the detection of the dust at separations closer than 1.2
(23.2 AU). The point source detection limit is compared to recently published observations of a planet candidate. Finally, the simulations show that deconvolution of coronagraphic data may indeed produce artificial patterns within the image of a disk.
Key words: stars: individual:
Pic - stars: pre-main sequence - stars: planetary systems: formation - stars: circumstellar matter - methods: observational - techniques: high angular resolution
1 Introduction
The young (<20 Myr) and nearby A-type star
The discoverers, Smith & Terrile (1984), resolved the disk in the visible at
projected separations larger than 5.2
(
100 AU, adopting the
Hipparcos distance of 19.3 pc), and, extrapolating the observed surface brightness profile to shorter
distances, they proposed that the disk should be optically thick below
15 AU. About ten years latter, Kalas & Jewitt (1995) observed the disk down
to 2.8
(54 AU) and identified morphological and brightness
asymmetries between the north-east and south-west extensions of the
edge-on disk. The WFPC2 instrument onboard the HST was used to observe
the disk as close as 1.5
(29 AU) and confirms the asymmetrical
structure.
Mid-infrared imaging (Lagage & Pantin 1994) of the innermost regions of the disk
(0-80 AU) showed that the inner disk (r<25 AU) is depleted by a factor
of
100 with respect to the regions of peak density (
AU).
Based on new datasets, this depletion was confirmed by Pantin et al. (1997) who proposed
a two-component model to interpret both the visible/near-infrared and mid-infrared
data.
From modeling, Burrows et al. (1995) derived the presence of a relatively
clear zone within 40 AU. In addition, they revealed a peculiar
symmetrical structure, the so-called warp, at a distance <70 AU
which should have disappeared in about 1 Myr. They interpreted this
feature as the gravitational signature of a planetary object which
would also cause the clearing of the inner disk regions. With the
improvement of Adaptive Optics system (AO) on large ground-based
telescopes, the warp was confirmed by Mouillet et al. (1997a) in the
near-IR using the 3.6 m telescope at La Silla (ESO). To account for
the presence of the warp Mouillet et al. (1997b) developed a model
which concluded the presence of a planet on an inclined orbit
(3-5
)
and located in the range 1-20 AU from the star. This
warp was also observed in detail with the HST/STIS instrument at
visible wavelengths (Heap et al. 2000). The small Lyot mask and the
quality of the PSF subtraction process allow the detection of the
disk as close to the star as 0.75
.
According to Heap et al. (2000), the
perturbing planet may have a mass of a few times that of Jupiter if
located closer than 20 AU. The most recent analysis
(Freistetter et al. 2007) based on numerical simulations constrained by
the presence of planetesimal belts inferred from the observed temporal
variations of metallic lines in the
Pictoris spectrum suggests a planet of
2 MJ at 12 AU. Note that high resolution spectra were also used
to search for radial velocity variations of
Pictoris(Galland et al. 2006). The
available data allowed them to detect pulsations (also seen in photometry)
and to constrain the presence of planets with small periods (e.g. a
few days). Longer periods planets are still poorly constrained with
radial velocity data, with for example, at 1 AU, a limit of 9 MJ.
While this paper was being reviewed, a candidate planet was discovered by Lagrange et al. (2008) taking advantage of a reduced contrast in L' band images. With an inferred mass of 8 MJ, at a projected distance of 8 AU it is significantly more massive than the Freistetter et al. (2007) predictions although more measurements would be needed to definitely rule on the mass of this candidate planet and confirm the companionship.
A detailled dynamical model of the Pictoris dust disk was developed by
Augereau et al. (2001) who simulate a disk of planetesimals and dust
particles pertubed by a giant planet on an inclined orbit with
respect to the disk midplane. The Kalas & Jewitt (1995) large scale
asymmetries (at hundreds of AUs) and the warp observed in scattered
light are well reproduced assuming a planet located at 10 AU about
103 times less massive than the central star. The inferred surface
density for the disk corresponds to a broad annulus peaked around
80-110 AU and smoothly declining inward within about 70 AU.
Although the far-infrared thermal emission is correcty predicted by
this model, it is shown that most of the mid-infrared emission is very
likely produced by an additional hot population of small grains close to the star
and not modeled. Recent high-resolution mid-IR images
(Telesco et al. 2005) of the inner disk obtained with Gemini have revealed
the presence of an asymmetric structure located at about 50 AU and
which may correspond to a clump of particles differing in temperature
and/or in size to the other particles in the disk. This observation
supports the commonly shared hypothesis that the disk is replenished
by short-lived dust originating from collisions of planetesimals but
the mechanisms that can produce a clump of that size are still
unclear.
The latest images of the Pictoris disk were obtained by Golimowski et al. (2006) in the
visible with ACS on the HST. Despite the very high sensitivity of the
ACS, the large opaque Lyot mask does not allow them to probe the disk
morphology at an angular separation less than about 1.5
to the star. Using a
specific data reduction process to emphasize high spatial frequency
structures in the disk, they suggest that the warp is in fact the
result of a blend of a main disk with a much fainter, inclined
secondary disk which can be separated from the main one from 80 AU up
to about 150 AU in ACS images. However, the method used to separate
the two disks is arguable (see Sect. 4.1 for a more
detailled discussion) and it is therefore not clear if this secondary
disk is a fully distinct component or originates in the main, twisted
disk as supported by the work of Mouillet et al. (1997b) and
Augereau et al. (2001). Golimowski et al. (2006) also showed an
interesting color dependence of the disk in scattered light from which
they inferred a typical minimum grain size of about
m where
the planetesimals lie (r<120 AU).
Improving the angular resolution in the first few tens of AU from Pictoris is now clearly mandatory to access the planet forming regions.
Optimal conditions are met either with large ground-based telescopes
(providing the wavefront is corrected with high-order AO systems) or
from space with the HST. In this paper, we present new
coronagraphic images obtained with an 8 m telescope at the European
Southern Observatory in the near-IR. Although the sensitivity is not
sufficient to detect giant planets, the coronagraph we used is
potentially sensitive to smaller separations than any previous
observations. In Sect. 2 we describe the observations we
carried out and the data reduction process. The analysis of the disk
in terms of morphology, surface brightness profile, vertical flux and
position of the midplane is presented in Sect. 3. We then discuss the hypothesis of a secondary disk
(Sect. 4.1) and the contamination by PSF structures in
coronagraphic images (Sect. 4.2).
2 Observations and data reduction
2.1 Observing strategy
Coronagraphic observations of



The principle of the FQPM is discussed in Rouan et al. (2000). It uses a
phase shift in the focal plane resulting in a destructive
interference inside the geometric pupil of a point-like object
centered on the mask. A monochromatic FQPM operating at
m
was implemented in NACO on August 2003. Performance assessment of the
FQPM is reported in Boccaletti et al. (2004). This coronagraph has no
obscuration in the focal plane and therefore is able to reach the
theoretical diffraction limit of the telescope. As a drawback, in
actual observations (Strehl ratio <50%) the attenuation of the PSF
peak is less than for a Lyot coronagraph (a factor of
10 instead
of
200 depending on the level of the AO compensation). The
chromatism of the FQPM is not a limitation on NACO as demonstrated in
Boccaletti et al. (2004). Therefore, we decided to take advantage of both
coronagraphs to get (i) a good signal to noise in the disk at large
angular separations using the Lyot coronagraph; and (ii) to
characterize the circumstellar dust at very close separations with the
FQPM.
The reference star HD 45 291 (V=5.97, H=3.84, K=3.77), which was
observed right after the target star, was chosen to have the same
parallactic angle as Pictoris in order to preserve the same pupil
orientation. This way, the diffraction spikes of the reference star
have about the same orientation as those of
Pictoris with respect to the
field of view. The reference star has the same declination but a
difference in right ascension which depends on the integration time
(36 min here). We observed
Pictoris for a total of 800 s in the
H band (Lyot coronagraph), and 700 s with the FQPM in the Ks band (but
with a transmission of only 10% to avoid saturation of individual
frames), and similarly for the reference star. In addition, the
field of view was rotated by 45
when the FQPM is used to place the disk
in a different direction than that of the phase transitions of the phase mask. The pixel sampling is
13 mas/pixel.
![]() |
Figure 1:
Target to reference flux ratios as a function of the
angular separation to the star for the Lyot image ( left panel) and
the FQPM image ( right panel). The red and blue curves are the average values measured in angular sectors of 30 |
Open with DEXTER |
2.2 Data reduction
During the observations, the individual exposures (DIT = 1 s) are combined in long exposure images of 100 s each (NDIT = 100). Cosmetic reduction is applied on each image including flat fielding, bad pixel correction and sky subtraction. Images are then re-centered and co-added independently for the target star and the reference star.
Our procedure to remove the stellar residuals (unattenuated by the
coronagraph) using a reference star is similar for the Lyot and the
FQPM images. As described for example in Augereau et al. (1999) and
Boccaletti et al. (2003), the image of the target is first divided by that
of the reference (re-centered at a subpixel scale) to provide a map of
the intensity ratio. The scaling factor is measured at a radial
distance where the intensity ratio map is almost flat (
on
the Lyot image and
on the FQPM image) and is
azimuthaly averaged in regions free of circumstellar material
(perpendicularly to the disk midplane, in the north-west and south-east directions).
The intensity ratio profiles are displayed in Fig. 1.
The reference star image is then normalized to the scaling factor and subtracted from the target star image. After this process, the Lyot subtracted image still has a diffraction pattern originating from the high frequency zonal polishing defects of the primary mirror (2). Similar patterns are observed in HST images of AU Mic (Krist et al. 2005) for instance. This feature was estimated in several sectors around the star and an azimuthaly averaged pattern was subtracted. However, a residual zonal pattern remains in the subtracted image and produces a modulation of the disk intensity in some particular regions. Nevertheless, the disk is clearly visible all across the field of view. The raw coronagraphic images and the subtracted images are displayed in Figs. 2 and in 3 for the Lyot and the FQPM.
For the photometric calibration, out-of-mask images of the star were collected, but with a different set-up than coronagraphic images. Indeed, these calibration data are obtained with a neutral density filter (attenuation of the intensity by factors of 80 in H-band and 90 in Ks-band) and of a full aperture stop (instead of a 10% undersized unobstructed circular stop). The procedure for photometric calibration of such data is fully described in Boccaletti et al. (2008). A detailed estimation of photometric uncertainties accounting for the differences between out-of-mask and on-mask data leads to error bars of 0.16 mag in both the H and Ks filters. The photometric error budget is actually driven by the flux extraction method of a point source (0.13 mag) rather than the scaling factors. The sources of errors being identical to those in Boccaletti et al. (2008), we will adopt the same photometric uncertainties.
2.3 Point source detection limit
Finally, we measured the point-source detection limits on the images of Figs. 2 and 3. The 5-sigma contrast is calculated as the standard deviation of all pixels located at a given radius and that for each radius. Once converted to absolute magnitude, the result displayed in Fig. 4 shows the detection limit for the H and Ks bands along the disk. Note that the 5-sigma detection level is taken here for convenience but does not rigorously correspond to a confidence level as usually defined for a Gaussian distribution. A detailed study of confidence level in the presence of diffraction residuals is given in Marois et al. (2008).
The FQPM image (Fig. 3) has a significantly worse detection limit in
the background dominated region (>2
)
owing to the 10% transmission.
However, the complementarity is clearly demonstrated in the limit of detection.
The expected absolute magnitudes of giant planets (according to the DUSTY model of Chabrier et al. 2000)
for 5, 8, 10 and 20 MJ were overplotted in Fig. 4.
At 0.4
,
the 4QPM has the ability to pick out an object fainter than 8 MJ while this level of sensitivity is achieved at 0.5
in the Lyot image. The candidate planet reported by Lagrange et al. (2008) in Nov. 2003 data should have been detected at least in the
K band image if located at the same position. However, our observations and those of Lagrange et al. (2008) are separated by nearly 1 year and hence orbital motion is to be expected.
![]() |
Figure 2:
Images of |
Open with DEXTER |
![]() |
Figure 3:
Images of |
Open with DEXTER |
![]() |
Figure 4:
Point source detection limits (expressed in absolute magnitudes) in the
H and Ks bands
as measured respectively on the Lyot ( left panel) and FQPM images
( right panel) along the disk (angular sector of |
Open with DEXTER |
![]() |
Figure 5:
Magnified image of the central 3
|
Open with DEXTER |
The projected distance of the planet candidate was 8 AU ( mas) on
Nov. 2003.
We therefore re-analyze our data to look around this position but found no firm evidence of a companion. However, a suspicious point-like pattern is visible in the Lyot image at
mas i.e.
mas away from the discovery position (Fig. 5). It differs significantly from the Lagrange et al. (2008) observations. This pattern intensity is compatible with the DUSTY model (
). But no counterpart is detectable in the 4QPM image (
)
while the DUSTY model does not predict an equivalent flux in the H and Ks bands. Given the proper motion of
Pictoris (about 82 mas/yr to the north) a background star would appear to move closer to the star between 2003 (Lagrange et al. 2008) and 2004 (this paper). The source, if real in both epochs of data, moves farther away from the star between 2003 and 2004. Despite a lack of constraints on the orbital parameters, a projected orbital motion of 107 mas can be consistent with an actual separation larger than 8 AU. However, the position angle of the suspicious point-like pattern differs by 4.5
with respect to the Lagrange et al. (2008) observations which appears inconsistent with an orbit aligned with the disk. We therefore favor a false positive detection as an interpretation of the presence of this pattern.
Two hypotheses can be drawn from our 4QPM detection limit at Ks:
- 1/
- The physical separation in 2004 is less than 8 AU and then the planet candidate is at an angular separation less than 0.4
which may account for the non detection in our data. In that case the motion is greater than 50 mas;
- 2/
- the physical separation in 2004 is greater or equal to 8 AU and the planet candidate is fainter than the model prediction by almost 1 mag at Ks (
). Parameters of the system (for instance age) must be reviewed and other models could be considered.
![]() |
Figure 6:
Comparison of |
Open with DEXTER |
3 Characterization of the dusty disk
3.1 Morphology of the disk
The






The FQPM image is displayed in the middle panel of
Fig. 6 (a sector of 45
about the midplane was selected to
avoid confusion with instrumental radial features). To better
distinguish the large scale structures from the background noise we
applied a Gaussian filtering to the image shown in
Fig. 3 (right). As explained in Sect. 2, the
signal to noise is worse than in the Lyot image, despite similar
integration times, since the transmission is 10 times lower. The disk
is detected down to a level of 12.5 mag/arcsec2 at a radius of
about 2
.
Although the central field is not obscured as in the Lyot
image, a discontinuity at 0.7
on both sides suggests that the
patterns located below 0.7
are actually diffraction residuals. This
is confirmed by the measurement of the surface brightness in
Sect. 3.2. The warp is obviously not detected due to a lack of
sensitivity at distances larger than 2
.
However, a very pronounced
pattern resembling a knot at about 1
is clearly visible on the SW
side. It was not seen in the Lyot image owing to the presence of a
diffraction spike and its origin remains unexplained.
This feature surprisingly resembles the clump observed by Telesco et al. (2005) in the mid-IR although not located at the same position (52 AU in the SW side), while we are probing closer separations (1
is equivalent to about 20 AU). Other clumps (beyond 50 AU) reported by Wahhaj et al. (2003) and Weinberger et al. (2003) in the mid-IR are interpreted as the projection of rings since they are symmetrical. Such rings are also observed at much larger separations (>500 AU) in the visible by Kalas et al. (2000). However, in our near-IR data, the clump has no symmetrical couterpart in the NE side and the origin must be found elsewhere than in the presence of rings. The refinement of the planet candidate orbit (Lagrange et al. 2008) will certainly allows to perform more accurate dynamical simulations and possibly be helpful for the interpretation of such patterns.
A smooth contour of the Lyot image is overplotted on the FQPM image. The overlap is only partial because the usable field is different for each coronagraph (positions of the spikes and coronagraph signatures are different).
![]() |
Figure 7:
Surface brightness of the disk (in mag/arcsec2) versus
the radial separation for the Lyot ( left) and for the FQPM
( right). The integration times are respectively 800 s in the H band
for the Lyot and 700 s in the Ks band with a transmission of 10%
for the FQPM.
The solid lines show the radial profile of the raw image (before subtraction of a reference).
The dashed and dash-dotted lines stand for the NE and SW sides, averaged over
a 16 |
Open with DEXTER |
3.2 Surface brightness
The surface brightness (SB) of the disk was measured along the NE (










On the NE side, the two profiles are similar within the intensity range (grey shades), while on the SW side a small difference appears near 1.5
.
However, it is difficult to interpret this difference as a color effect
because the coronagraphs have different signatures and the overlap of the detected regions is only partial.
Also, it is not sufficient to constrain the grain size distribution as we did for HD 141 569 in Boccaletti et al. (2003).
In both the Lyot and FQPM cases, the diffraction residuals in the
center can probably be improved with a more careful matching of the
star and reference images during the observation (better centering and
better correction of the low order aberrations are needed). The
advantage of the FQPM over the Lyot coronagraph, although less
sensitive in the background dominated region, is clearly seen
here and allows us to detect the disk at a separation which is almost
half the size than on the Lyot image. This compares to the best
distance reported in the literature and achieved with the HST/STIS
instrument, where the disk is detected down to 0.75
(Heap et al. 2000). However, the exact signal-to-noise ratio at these
distances cannot be estimated on the STIS images because
Heap et al. (2000) do not compare the average brightness of the disk with
the noise level.
![]() |
Figure 8: Left: departure of the disk spine with respect to the midplane as measured on the smoothed Lyot image. Right: thickness of the disk along the midplane. The FWHM of the two sides are overplotted to show the asymmetry. For these 2 plots, parabolic fits were used to derive positions of the asymmetries. Error bars are obtained from the same data with several smoothing factors relevant to the measurement of such spatial scales. |
Open with DEXTER |
3.3 Spine and thickness
The position angle of disk midplane is measured on the Lyot image by a line passing through the star and going from the edge of the NE side of the disk to that of the SW side. The disk PA is found to be




In the inner disk regions, the known warp makes the disk not perfectly
aligned with the midplane, but instead it appears twisted as shown for
instance by Mouillet et al. (1997b) and Heap et al. (2000). Along cuts
orthogonal to the midplane, the distance of the peak brightness
position to the midplane defines the disk spine which is displayed in
the left panel of Fig. 8 for our Lyot image and for a
separation larger than
.
This figure was obtained in two
steps. First, the image of the disk was convolved with a Gaussian
function to account for the sole large-scale structures. Then, we
measured the altitude of the peak brightness with respect to the
midplane as a function of the distance from the star using two
different methods: one consists of directly measuring the distance of
the maximum intensity to the disk midplane, while in the second case,
the spine position is obtained from a Gaussian fitting to the vertical
profile. Both methods yield very similar results.
According to Fig. 8 (left panel), the position of the
warp in the SW and NE sides appears to be not symmetrical with respect
to the star. The warp extends to 3
on both sides but peaks at about 3.3
(65 AU) on the SW side and
4.0
(78 AU) on the NE side as measured with parabolic fits
(overplotted on Fig. 8). This shift cannot be attributed to a
mis-alignment of the star behind the mask, since the centering
accuracy is better than 0.1
.
The difference in the position of the
warp with respect to the star is not reported in previous
observations. It is reminiscent of the disk offset discovered in
Fomalhaut by Kalas et al. (2005) and to a lesser extent in the disk around
HD 141 569 (Boccaletti et al. 2003; Mouillet et al. 2001). The amplitude of
this offset is physically almost identical in the cases of Fomalhaut and
Pictoris (about 15 AU), although the
Pictoris offset might be larger due to
projection effects. It has been suggested that a planetary companion
on an eccentric orbit (e=0.02) could create such asymmetries
(e.g. HR 4796, Wyatt et al. 1999). However, it is noted that the mass of the pertubating planet is not very well constrained since several combinations of mass and semi-major axis are able to reproduce the observations.
The recent detection of a planet around Fomalhaut (Kalas et al. 2008) confirms the relationship between the perturber and the eccentricity of the ring. The most stringent study in the case of
Pictoris is presented by Freistetter et al. (2007). To reproduce the
planetesimal belt positions inferred from high-resolution mid-IR spectroscopy, a system with three planets is necessary, one of them being significantly more massive (
2 +3-0.5 MJ) and located at 12 AU.
A revised value of the
Pictoris age (12 Myr, Zuckerman et al. 2001) together with the reasoning of Heap et al. (2000)
and an averaged position of 71 AU of the warp yield a mass of about 2 MJ, similar to that of Freistetter et al. (2007). Our detection limit presented in
Fig. 4 is able to place an upper mass limit of 5-6 MJ at 10 AU for an hypothetic planet of 10 Myr. This is identical to the upper limit of Freistetter et al. (2007) and therefore not relevant to improve the mass constraint while the planet candidate from Lagrange et al. (2008) is actually more massive (8 MJ).
The altitude of the maximum deviation with respect to the
midplane reaches 0.10
(1.96 AU) in the SW and 0.07
(1.37 AU) in the
NE extension. Heap et al. (2000) also observed a higher altitude in the SW
side although they did not discuss this point. Another deviation
from the midplane is also visible at closer separations
and is located at 1.1
and 1.5
respectively on the SW and NE sides. This
inverse warp is likely an artifact since we do not detect it on the FQPM
image which is more accurate than the Lyot image in this
region. We suspect that the confusion near the center between
circumstellar material and diffraction residuals from the on-axis star
is responsible for such patterns which are enhanced by the use of a
Lyot coronagraph. It might be the same kind of artifact that lead
Heap et al. (2000) and Golimowski et al. (2006) to conclude that the two
extensions in the inner disk do not intersect at the star.
![]() |
Figure 9:
Vertically integrated flux as a function of the angular
separation to the star for the Lyot ( left) and the FQPM
( right). Dotted lines correspond to measurements at 90 |
Open with DEXTER |
The Gaussian fit also yields the full width at half maximum (FWHM)
of the disk along the midplane. The maximum thickness occurs at about
4.5
and 4.0
from the center respectively for the SW and NE sides
(Fig. 8, right). This asymmetry is opposite to that of
the warp position. In addition, the SW side is marginally thinner than
the NE side (at 1-sigma) a characteristic also reported at mid IR wavelengths
(Pantin et al. 1997).
The average thickness is about 0.9
(18 AU), in perfect agreement with Heap et al. (2000) and again with Pantin et al. (1997). There is a consensus for this parameter at visible, near IR and mid IR wavelengths.
In Golimowski et al. (2006), the SW side appears thinner although error bars are missing to confirm this point. However, they claim that
actual thicknesses are 30
less after PSF deconvolution. However,
the narrow HST/ACS PSF (
)
cannot increase the apparent size
of a pattern which is more than 15 times larger. With a simple
numerical simulation, we measured that a disk of which the FWHM is
15 times larger than the PSF size undergoes an increase of only 0.5% in
size once convolved with the PSF. Clearly, the impact of
PSF-deconvolution in coronagraphic imaging is a very critical issue
and will be discussed in Sect. 4.2.
A summary of the asymmetries we measured inside the disk is reported in Table 1.
Table 1:
Registration of asymmetries in the warp of Pictoris.
3.4 Vertically integrated flux and normal optical thickness
We finally estimate the vertically integrated flux which is measured across the disk thickness (z-axis) and shown as a function of the radial distance to the star in Fig. 9. This flux is obtained by integrating the disk surface brightness over a





The disk being seen edge-on, the observed surface brightness profile
at any given projected distance
is the sum of the
brightness contributions, modulated by the scattering phase function,
of all the grains physically located at distances
along the considered line of sight. Deriving the
shape of the surface density profile from an edge-on disk brightness
profile requires the use of a deprojection method, such the one
employed by Augereau & Beust (2006) for AU Mic
or the numerical one described in Pantin et al. (1997) for
PictorisIn this approach, the product of the dust surface density profile
and the
mean scattering cross section of the grains
is
reconstructed from the outer disk edge down to the closest distance.
For the inversion process to work properly, smooth profiles that
extend to large distances where the disk brightness vanishes are
required. It is therefore not well adapted to the FQPM profiles, but
more suited to the Lyot surface brightness profiles which cover a much
larger range of distances. Before performing the inversion, the Lyot
profiles have nevertheless been extrapolated
beyond 140 AU with a r-4.5 radial profile and a Gaussian vertical profile (FWHM(r) taken from
Golimowski et al. 2006), and then smoothed over 10 pixels.
![]() |
Figure 10:
|
Open with DEXTER |
Examples of deprojected normal disk profiles using the inversion
technique of Augereau & Beust (2006) are displayed in
Fig. 10 for the SW and NE sides. The figure shows
the product of the mean scattering cross section of the grains in the
H-band
by the dust disk surface density
for different scattering asymmetry factors |g| between
0 (isotropic scattering) and 0.5. The
profiles are equivalent to normal
dust disk optical thickness profiles multiplied by the mean albedo of
the grains in the H-band. The
Pictoris disk appears 2 to 4 times less
dense at the peak position than the 2.5 times more distant annulus
about the F5/F6 V star HD 181 327 that peaks around
87 AU (Schneider et al. 2006).
Beyond 150 AU, the calculated profiles follow a r-2.5power law as expected from the extrapolated surface brightness
profiles at these distances. The SW
profiles show a peak, whose exact position varies with |g|, but
is situated in the range 80-90 AU. The global ring-shape of the SW
dust disk is consistent with the normalized (to allow comparison)
planetesimal surface density inferred by Augereau et al. (2001)
(gray thick line in Fig. 10). The dust profiles in
Fig. 10 appear more extended than the theoretical
planetesimal belt because the radiation pressure force causes the
grains produced through collisions to populate the outer disk regions
as shown in Augereau et al. (2001). The NE
profiles below 100 AU do not
mirror the SW profiles suggesting a non-axisymmetrical structure. The
NE profiles indeed appear flatter than the SW profiles, and the peak
at around 90 AU only becomes visible for very anisotropic
scatterers (large |g| values). It is also noticeable that the observations
in the mid-infrared deduced surface density profiles (Pantin et al. 1997) are
consistent with |g|=0.5 profiles estimated here.
![]() |
Figure 11:
Figures identical to Fig. 6 ( top and middle) but after deconvolution and
division by the original, non-deconvolved image. Two dashed lines
at |
Open with DEXTER |
Although the exact shape of the disk
cannot be derived uniquely (because of uncertainties in the
observations and in the scattering properties in particular), both the
NE and SW
profiles point toward the
presence of dust material inside the main 80-100 AU planetesimal
belt, as previously suggested for example by mid-IR observations
(Telesco et al. 2005, and ref. therein), and by the present FQPM images. Overall, the results tend to support the idea of a complex,
structured and asymmetric system inside the apparent warp location,
which led Freistetter et al. (2007) to conclude the presence of several
planetary companions.
4 On the warp
The warp of
4.1 The secondary disk
The coronagraphic images presented in Golimowski et al. (2006) were obtained with the HST/ACS instrument and classically subtracted with the image of a reference star. In this paper we adopted a similar approach for the data reduction (Sect. 2.2). This processing is extremely efficient in the case of the HST owing to the stability of the PSF compared to that of ground-based telescopes. Then, the HST/ACS subtracted coronagraphic image was deconvolved with a standard Lucy Richardson algorithm taking a synthetic Tiny Tim (Krist & Hook 2004) image as a PSF.
In classical imaging, the image results from the convolution of the
object by the PSF. Deconvolution assumes that the PSF is stationary (independent
of its position in the field). This is no longer the case in coronagraphy since the image variation is strongly
non-linear especially near the center (Malbet 1996). The term PSF is no longer applicable.
In a coronagraph, the convolution occurs in the pupil plane downstream of
the mask where the pupil is convolved by the Fourier transform of the
mask (Malbet 1996). A PSF defined for each point in the image might
be necessary in this case for deconvolution. Golimowski et al. (2006)
have worked around the problem by avoiding the inner region inside a
1.5
radius, but the effect of deconvolution on coronagraphic image
is not further analyzed. At this stage, the warp is still visible as a
structure in the main disk. To improve the visibility of fine
structures, Golimowski et al. (2006) have divided the deconvolved
with the non-deconvolved subtracted coronagraphic image. This
procedure reinforces the high spatial frequencies of the images (as in
unsharp masking) and reveals the warp as two separate disks. This
hypothetic secondary disk can be distinguished from the main disk
beyond 80 AU. Below that distance, the intrisic disk thickness
prevents any partition of the two disks.
However, this structure is difficult to interpret. It could be a
secondary disk fully separated from the main disk and centered on the
star, or could instead originate from the main disk as for instance
suggested by the inward extrapolation of NE secondary disk which
intercepts the main NE disk at around 30 AU
(Golimowski et al. 2006). The surface brightness profile of the secondary
disk as measured by Golimowski et al. (2006) is very reminiscent of the
brightness profile of the main disk beyond 125 AU. The latter
distance is supposed to represent the outer edge of the main
planetesimal belt from which the dust grains originate that populate
the outer disk regions due to radiation pressure forces
(e.g. Augereau et al. 2001, and ref. therein). Similarly, the secondary
disk could represent the dusty tail of an inclined population of
planetesimals located within 80 AU.
It is noteworthy that the extrapolation inward of the secondary disk surface
brightness measured by Golimowski et al. (2006) would make it
brighter than the main disk below 70 AU, which would make the main
disk brightness profile much steeper than actually observed. The
rather flat surface brightness profile observed below 4
suggests
that the secondary disk brightness profile breaks at around
70-80 AU and becomes flatter inside that distance. The reasoning
developed for the main disk may thus also be valid for the secondary
disk, which would then originate from parent bodies arranged in the
form of an inclined belt with an outer edge close to
70-80 AU. It is remarkable that this distance matches the
maximum extension for the warp produced by the inclined planet in the
models of Mouillet et al. (1997a) and Augereau et al. (2001).
Our reduced coronagraphic images were deconvolved as in
Golimowski et al. (2006) except that we were using an actual PSF image of
Pictoris (out of the coronagraph) and a maximum likelihood algorithm for
the deconvolution. As a result of the deconvolution process, we
marginally detect a radial structure with an inclination of
with respect to the disk midplane and consistent with the secondary
disk, but also one almost symmetrical structure about the disk
midplane at
from the main disk (Fig. 11,
top panel). It is however difficult to measure the radial distance
where these structures originate, although they are consistent with
purely radial patterns
intersecting the star unlike the
secondary disk (at least on the NE side) observed with the ACS
instrument. Here, we interpret the radial patterns revealed by the
deconvolution process as PSF radial patterns instead of real
patterns. These patterns are therefore different to those observed by Golimowski et al. (2006). Our emphasized FQPM image is presented in the bottom panel
of Fig. 11 but does not show any features.
![]() |
Figure 12:
Simulated images of ( from top to bottom), the disk model,
the disk convolved with the PSF, the coronagraphic image
subtracted with that of a reference star and the deconvolved
image.
The FOV is
|
Open with DEXTER |
4.2 Deconvolution of coronagraphic images: simulation
To evaluate the significance and the nature of the PSF patterns that may appear after deconvolution of a coronagraphic image of a circumstellar disk, we produce synthetic HST/ACS-like coronagraphic observations of an edge-on disk that we afterwards deconvolve. Instead of using a Tiny Tim PSF, we produce a synthetic HST PSF image using our own diffraction code that we have developed to model the behavior and performance of coronagraphs, as well as to gauge the potential of several planet finding projects (Boccaletti et al. 2006; Boccaletti 2004; Boccaletti et al. 2005).
We consider a 2.4 m diameter telescope observing at m,
with a bandwidth of 33%, and providing an angular resolution of
0.052
(similar to the ACS parameters). We generate the image of an edge-on circumstellar disk
having a 4.5
diameter extension and a midplane surface brightness
proportional to r-1.5, r being the angular distance to the
star. The disk extends vertically over 0.52
(10 times the
angular resolution) with a Gaussian profile of FWHM = 0.17
(Fig. 12, top panel). The disk model is absolutely free of structures.
Note that some of these parameters are arbitrary and are not intended to exactly model the
Pictoris disk.
This object contains 8542points for which we have to calculate the image through a coronagraph
(a Lyot mask of 1.8
diameter) and for several wavelengths so as to
produce a realistic coronagraphic polychromatic image resembling that
of the ACS instrument. To simulate realistic images, the model
accounts for the phase aberrations using the measured phase map of the
HST (Krist & Burrows 1995), and we assume similar static aberrations on the
reference star but with an additional aberration in the form of a
defocus (4 nm rms) to account for the so-called ``PSF breathing''
(the HST PSF is slightly evolving over time). This amount of defocus is
taken arbitrarily as
we are looking for qualitative results here. Another way to produce differences in the target star and
reference star images would be to apply a different pointing for the
two images with respect to the Lyot mask. The star to disk flux ratio
is also arbitrary and is selected to have the
residuals on the subtracted image of the order of the
disk intensity. We suspect that this is precisely in this regime of
contrast that deconvolution may generate artifacts. The synthetic
images is then deconvolved as in Sect. 4.1.
![]() |
Figure 13:
Comparison of the radial intensity profile ( left) and the
vertical profile of the disk obtained at 1.5
|
Open with DEXTER |
Several images to illustrate the process of image formation are displayed in Fig. 12. The two top panels show the disk model alone (no star) and the disk convolved with the PSF. The two lower panels are coronagraphic images (reference star subtracted) before and after deconvolution. Several patterns like speckles, rings, and radial spikes which were not included in the disk model are seen and are clearly enhanced in the deconvolution process. For instance, the radial spikes mimic the splitting of the disk midplane into two parts but are obviously not real. These spikes are aligned here with the disk to depict a worse case but could be differently oriented.
We used these simulated data to measure the averaged radial profile and the disk
thickness (Fig. 13), before and after deconvolution
of the image. Despite the intensity fluctuations resulting from the
PSF patterns the averaged radial profile is consistent with the
model except near the mask edges. It is interesting to see how the deconvolution
enhances the variations of the intensity profile
(Fig. 13, left) especially at distances closer than 1.0-1.2
.
As a confirmation of the discussion in Sect. 3.3 and contrary to what is proposed in
Golimowski et al. (2006), the disk thickness is not affected by the
deconvolution as long as it is thicker than the angular resolution
(Fig. 13, right). In
the simulation, the FWHM is only 0.17
which is about 3 times the
angular resolution while in the actual data of
Pictoris, the disk thickness
is about 0.87
(Fig. 8) so the PSF blurring should be
even less visible. The vertical cut shown in Fig. 13 (right) is obtained at 1.5
(left side of the image). The shape of the blue line (deconvolved image) changes significantly at closer separations, so the FWHM of the disk must be measured in a region less affected by PSF patterns.
Although the simple simulation presented here reveals the production of artifacts, a more general analysis would be necessary to place quantitative limitations when using deconvolution with coronagraphic images of circumstellar disks. We note that the simulations presented here do not rule out the the reliability of the secondary disk shown in Golimowski et al. (2006). This secondary disk is clearly significant with respect to PSF patterns arising in deconvolution of coronagraphic images. In particular, it is symmetrical and originates in the main disk at a distance where PSF patterns are much less preponderant (see Sect. 4.1). The PSF patterns that may be confused with disks structures like spikes mostly originate from the star.
5 Conclusion
This paper reports on the observation of the




The Lyot mask coronagraph allows a very good characterization of the disk at large distances and especially of the famous warp. We have been able to precisely measure the position, elevation and thickness of the warp and we have shown geometric and photometric particular asymmetries that could eventually been used as inputs to dynamical modeling in order to better constrain the characteristics of the forming planetary system. Some of these asymmetries were reported in various spectral bands (visible and mid IR). The inversion of the surface brightness profiles have confirmed the presence of an asymmetric main planetesimal belt peaking at around 80-100 AU responsible for the dust material beyond that distance. The inversion further supports the presence of significant amounts of additional material inside the main planetesimal belt as suggested by other observations, depicting a complex planetary system.
The FQPM image reveals the presence of very close patterns very similar to what is observed in the mid-IR, a spectral regime that is sensitive to the dust emission. The interpretation of such structures would require further analysis possibly involving dynamical models to understand if the origin is similar to those observed in the mid-IR.
To understand the issue of deconvolution in coronagraphic imaging we have performed numerical simulations to mimic HST Lyot coronagraphic data. Assuming a disk with no structure, this qualitative work shows how the data reduction process may emphasize some PSF structures that reinforce the need for a thorough analytical study for this type of data.
These new images of the Pictoris debris disk presented here demonstrate the interest of ground-based instruments to achieve high angular resolution at longer wavelengths than with the HST and motivate for even more accurate observations. The sub-arcsec region is now receiving most of the attention with the recent discovery of a planet candidate. Constraints on the characteristics of this object and possibly others could be inferred form the spatial distribution of the dust.
Therefore, the precise registration of the inner structures we have performed if complemented with perturbation theory will help to better characterize this planetary system. To confirm the numerical simulations of Freistetter et al. (2007) the study of the disk and its structures near 12 AU is essential. This corresponds precisely to the limit reached in our data (0.62
)
and therefore the motivation is strong to perform more accurate observations. This is becoming possible at the VLT. The coronagraphic suite of NACO now implements other FQPMs tuned for the H and Ks bands with the ability to reach a better attenuation. Additional techniques also have been installed (Angular Differential Imaging, Spectral Differential Imaging) that may facilitate the calibration of the residual starlight and hence the registration of structures even closer to the star.
After more than 20 years of intensive studies, the dusty disk around Pictoris is still the subject of many questions.
Direct imaging observations are complementary to spectroscopy and photometry and are mandatory to better constrain the spatial distribution of the dust, planetesimals and hypothetical planets in order to contribute to the understanding of the planetary formation phase.
Acknowledgements
We are grateful to the ESO staff supporting observations with NACO at the VLT and to the referee for an objective report that help to improve the manuscript. This work also received the support of PHASE, the high angular resolution partnership between ONERA, Observatoire de Paris, CNRS and University Denis Diderot Paris 7.
References
- Augereau, J.-C., & Beust, H. 2006, A&A, 455, 987 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Augereau, J.-C., Lagrange, A.-M., Mouillet, D., & Ménard, F. 1999, A&A, 350, 51 (In the text)
- Augereau, J.-C., Nelson, R. P., Lagrange, A.-M., Papaloizou, J. C. B., & Mouillet, D. 2001, A&A, 370, 447 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Boccaletti, A. 2004, in EAS Publ. Ser., 12, 165
- Boccaletti, A., Augereau, J.-C., Marchis, F., & Hahn, J. 2003, ApJ, 585, 494 [NASA ADS] [CrossRef] (In the text)
- Boccaletti, A., Riaud, P., Baudoz, P., et al. 2004, PASP, 116, 1061 [NASA ADS] [CrossRef] (In the text)
- Boccaletti, A., Baudoz, P., Baudrand, J., Reess, J.-M., & Rouan, D. 2005, Adv. Space Res., 36, 1099 [NASA ADS] [CrossRef]
- Boccaletti, A., Mouillet, D., Fusco, T., et al. 2006, in Direct Imaging of Exoplanets: Science Techniques, ed. C. Aime, & F. Vakili, IAU Coll., 200, 519
- Boccaletti, A., Chauvin, G., Baudoz, P., & Beuzit, J.-L. 2008, A&A, 482, 939 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Burrows, C. J., Krist, J. E., & Stapelfeldt, K. R. 1995, AAS 187, BAAS, 27, 1329 (In the text)
- Chabrier, G., Baraffe, I., Allard, F., & Hauschildt, P. 2000, ApJ, 542, 464 [NASA ADS] [CrossRef] (In the text)
- Freistetter, F., Krivov, A., & Löhne, T. 2007, A&A, 466, 389 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Galland, F., Lagrange, A.-M., Udry, S., et al. 2006, A&A, 447, 355 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Golimowski, D., Ardila, D., Krist, J., et al. 2006, AJ, 131, 3109 [NASA ADS] [CrossRef] (In the text)
- Heap, S. R., Lindler, D. J., Lanz, T. M., et al. 2000, ApJ, 539, 435 [NASA ADS] [CrossRef] (In the text)
- Kalas, P., & Jewitt, D. 1995, AJ, 110, 794 [NASA ADS] [CrossRef] (In the text)
- Kalas, P., Larwood, J., Smith, B. A., & Schultz, A. 2000, ApJ, 530, L133 [NASA ADS] [CrossRef] (In the text)
- Kalas, P., Graham, J. R., & Clampin, M. 2005, Nature, 435, 1067 [NASA ADS] [CrossRef] (In the text)
- Kalas, P., Graham, J. R., Chiang, E., et al. 2008, ArXiv e-prints (In the text)
- Krist, J. E., & Burrows, C. J. 1995, Appl. Opt., 34, 4951 [NASA ADS] [CrossRef] (In the text)
- Krist, J. E., & Hook, R. 2004, The tiny Tim User's Guide, v6.3 (Baltimore STScI) (In the text)
- Krist, J. E., Ardila, D., Golimowski, D. A., et al. 2005, AJ, 129, 1008 [NASA ADS] [CrossRef] (In the text)
- Lagage, P. O., & Pantin, E. 1994, Nature, 369, 628 [NASA ADS] [CrossRef] (In the text)
- Lagrange, A., Gratadour, D., Chauvin, G., et al. 2008, ArXiv e-prints (In the text)
- Malbet, F. 1996, A&AS, 115, 161 [NASA ADS] (In the text)
- Marois, C., Lafrenière, D., Macintosh, B., & Doyon, R. 2008, ApJ, 673, 647 [NASA ADS] [CrossRef] (In the text)
- Mouillet, D., Lagrange, A.-M., Beuzit, J.-L., & Renaud, N. 1997a, A&A, 324, 1083 (In the text)
- Mouillet, D., Larwood, J. D., Papaloizou, J. C. B., & Lagrange, A.-M. 1997b, MNRAS, 292, 1997 (In the text)
- Mouillet, D., Lagrange, A. M., Augereau, J. C., & Ménard, F. 2001, A&A, 372, L61 [NASA ADS] [CrossRef] [EDP Sciences]
- Pantin, E., Lagage, P. O., & Artymowicz, P. 1997, A&A, 327, 1123 (In the text)
- Rouan, D., Riaud, P., Boccaletti, A., Clenet, Y., & Labeyrie, A. 2000, PASP, 112, 1479 [NASA ADS] [CrossRef] (In the text)
- Rousset, G., Lacombe, F., Puget, P., et al. 2003, SPIE, 4839, 140 [NASA ADS] (In the text)
- Schneider, G., Silverstone, M. D., Hines, D. C., et al. 2006, ApJ, 650, 414 [NASA ADS] [CrossRef] (In the text)
- Smith, B., & Terrile, R. 1984, Science, 226, 1421 [NASA ADS] [CrossRef] (In the text)
- Telesco, C. M., Fisher, R. S., Wyatt, M. C., et al. 2005, Nature, 433, 133 [NASA ADS] [CrossRef] (In the text)
- Wahhaj, Z., Koerner, D. W., Ressler, M. E., et al. 2003, ApJ, 584 (In the text)
- Weinberger, A. J., Becklin, E. E., & Zuckerman, B. 2003, ApJ, 584, L33 [NASA ADS] [CrossRef] (In the text)
- Wyatt, M. C., Dermott, S. F., Telesco, C. M., et al. 1999, ApJ, 527, 918 [CrossRef] (In the text)
- Zuckerman, B., Song, I., Bessell, M. S., & Webb, R. A. 2001, ApJ, 562, L87 [NASA ADS] [CrossRef] (In the text)
All Tables
Table 1:
Registration of asymmetries in the warp of Pictoris.
All Figures
![]() |
Figure 1:
Target to reference flux ratios as a function of the
angular separation to the star for the Lyot image ( left panel) and
the FQPM image ( right panel). The red and blue curves are the average values measured in angular sectors of 30 |
Open with DEXTER | |
In the text |
![]() |
Figure 2:
Images of |
Open with DEXTER | |
In the text |
![]() |
Figure 3:
Images of |
Open with DEXTER | |
In the text |
![]() |
Figure 4:
Point source detection limits (expressed in absolute magnitudes) in the
H and Ks bands
as measured respectively on the Lyot ( left panel) and FQPM images
( right panel) along the disk (angular sector of |
Open with DEXTER | |
In the text |
![]() |
Figure 5:
Magnified image of the central 3
|
Open with DEXTER | |
In the text |
![]() |
Figure 6:
Comparison of |
Open with DEXTER | |
In the text |
![]() |
Figure 7:
Surface brightness of the disk (in mag/arcsec2) versus
the radial separation for the Lyot ( left) and for the FQPM
( right). The integration times are respectively 800 s in the H band
for the Lyot and 700 s in the Ks band with a transmission of 10%
for the FQPM.
The solid lines show the radial profile of the raw image (before subtraction of a reference).
The dashed and dash-dotted lines stand for the NE and SW sides, averaged over
a 16 |
Open with DEXTER | |
In the text |
![]() |
Figure 8: Left: departure of the disk spine with respect to the midplane as measured on the smoothed Lyot image. Right: thickness of the disk along the midplane. The FWHM of the two sides are overplotted to show the asymmetry. For these 2 plots, parabolic fits were used to derive positions of the asymmetries. Error bars are obtained from the same data with several smoothing factors relevant to the measurement of such spatial scales. |
Open with DEXTER | |
In the text |
![]() |
Figure 9:
Vertically integrated flux as a function of the angular
separation to the star for the Lyot ( left) and the FQPM
( right). Dotted lines correspond to measurements at 90 |
Open with DEXTER | |
In the text |
![]() |
Figure 10:
|
Open with DEXTER | |
In the text |
![]() |
Figure 11:
Figures identical to Fig. 6 ( top and middle) but after deconvolution and
division by the original, non-deconvolved image. Two dashed lines
at |
Open with DEXTER | |
In the text |
![]() |
Figure 12:
Simulated images of ( from top to bottom), the disk model,
the disk convolved with the PSF, the coronagraphic image
subtracted with that of a reference star and the deconvolved
image.
The FOV is
|
Open with DEXTER | |
In the text |
![]() |
Figure 13:
Comparison of the radial intensity profile ( left) and the
vertical profile of the disk obtained at 1.5
|
Open with DEXTER | |
In the text |
Copyright ESO 2009
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.