HUBBLE SPACE TELESCOPE ADVANCED CAMERA FOR SURVEYS CORONAGRAPHIC
IMAGING OF THE AU MICROSCOPII DEBRIS DISK
John E. Krist,
1
D. R. Ardila,
2
D. A. Golimowski,
2
M. Clampin,
3
H. C. Ford,
2
G. D. Illingworth,
4
G. F. Hartig,
1
F. Bartko,
5
N. Benı´tez,
2
J. P. Blakeslee,
2
R. J. Bouwens,
4
L. D. Bradley,
2
T. J. Broadhurst,
6
R. A. Brown,
1
C. J. Burrows,
1
E. S. Cheng,
7
N. J. G. Cross,
2
R. Demarco,
2
P. D. Feldman,
2
M. Franx,
8
T. Goto,
2
C. Gronwall,
9
B. Holden,
4
N. Homeier,
2
L. Infante,
10
R. A. Kimble,
3
M. P. Lesser,
11
A. R. Martel,
2
S. Mei,
2
F. Menanteau,
2
G. R. Meurer,
2
G. K. Miley,
8
V. Motta,
10
M. Postman,
1
P. Rosati,
12
M. Sirianni,
2
W. B. Sparks,
1
H. D. Tran,
13
Z. I. Tsvetanov,
2
R. L. White,
1
and W. Zheng
2
Receivved 2004 September 10; accepted 2004 October 20
ABSTRACT
We present Hubble Space Telescope Advanced Camera for Surveys multicolor coronagraphic images of the
recently discovered edge-on debris dis k around the nearby (10 pc) M dwarf AU Microscopii. The disk is seen
between r ¼ 0B75 and 15
00
(7.5–150 AU ) from the star. It has a thin midplane w ith a projected FWHM thickness of
2.5–3.5 AU within r < 50 AU of the star that increases to 6.5–9 AU at r 75 AU. The disk’s radial brightness
profile is generally flat for r < 15 AU, then decreases gradually (I / r
1:8
)outtor 43 AU, beyond which it falls
rapidly (I / r
4:7
). Within 50 AU the midplane is straight and aligned with the star, and beyond that it deviates by
3
, resulting in a bowed appearance that was also seen in ground-based images. Three-dimensional modeling of
the d isk shows that the inner region (r < 50 AU) is inclined to the line of sight by less than 1
and the outer disk by
3
. The inclination of the outer disk and moderate forward scattering ( g 0:4) can explain the apparent bow. The
intrinsic, deprojected FWHM thickness is 1.5–10 AU, increasing with radius. The models indicate th at th e d isk is
clear of dust within 12 AU of the star, in general agreement with the previous prediction of 17 AU based on the
infrared spectral energy distribution. The disk is blue, being 60% brighter at B than I relative to the star. One
possible explanation for this is that there is a surplus of very small grains compared with other imaged debris disks
that have more neutral or red colors. This may be due to the low radiation pressure exerted by the late-type star.
Observations at two epochs show that an extended source se en along the midplane is a background galaxy.
Key words: circumstellar matter — stars: individual (AU Microscopii)
1. INTROD UCTION
Collisions of planetesimals around a star produce dust grains,
creating a debris disk that can be detected in thermal emission
and scattered light. Without the continual replenishment of
these grains, most of the disk will disappear because of radia-
tion and wind pressure, P oyn ting-Robertson d rag, grain coag-
ulation, and tidal interactions with planets. The lifetime o f such
disks is not fully understood, and it may depend strongly on the
luminosity of the star.
Although numerous detections an d images of optically thick
accretion disks have been made, only a few debris disks have
been resolved. The current inventory of imaged debris disks is
largely derived from IRAS-measured infrared excesses of stars.
This list is biased toward stars of spectral types A–F because
their luminosities are high en ough to heat a deb ris disk to an
IRAS-detectable level. Mo re luminous stars would blow away
such disks, whereas cooler stars cannot heat them enough to be
detected. Most of the re solved debris disks surround A stars
( Pic, Vega, Fomalhaut, HD 141569A, and HR 4796), and
only a few have been seen in scattered light. The sensitivity lim-
its of IRAS leave uncertain the frequency of debris disks around
later type stars. The disk around Eridani, a K2 V star with one
of the largest IRAS-measured stellar excesses, was detected by
IRAS only because of its very close proximity (3.2 pc) to us. The
Spitzer Space Telescope should provide a much more compre-
hensive catalog of disk candidates around later type stars.
There are only two M dwarfs, AU Microscopii and Hen
3-600 (Song et al. 2002), not associated with molecular clouds
that have IRAS-measured infrared excesses. AU Mic (HD197481,
GJ 803) is a M1 Ve flare star and a BY Dra type variable (V
max
¼
8:59, V 0:15; Cutispoto et al. 2003). Barrado y Navascue
´
s
et al. (1999) identified AU Mic as part of the Pic moving
group, an association of nearby (10–50 pc) young stars. It is at a
Hipparcos-measured distance of 9.94 pc (Perryman et al. 1997)
and is about 10 Myr old (Zuckerman et al. 2001). Its excess im-
plies that AU Mic has a substantial amount of circumstellar ma-
terial. These characteristics led the Hubble Space Telescope (HST )
1
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD
21218.
2
Department of Physics and Astronomy, Johns Hopkins University, 3400
North Charles Street, Baltimore, MD 21218.
3
NASA Goddard Space Flight Center, Code 681, Greenbelt, MD 20771.
4
UCO/Lick Observatory, University of California, Santa Cruz, CA 95064.
5
Bartko Science and Technology, 14520 Akron Street, Brighton, CO 80602.
6
Racah Institute of Physics, Hebrew University, Jerusalem 91904, Israel.
7
Conceptual Analytics, LLC, 8209 Woburn Abbey Road, Glenn Dale, MD
20769.
8
Leiden Observatory, Postbus 9513, NL-2300 RA Leiden, Netherlands.
9
Department of Astronomy and Astrophysics, Pennsylvania State Univer-
sity, 525 Davey Laboratory, University Park, PA 16802.
10
Departmento de Astronomı´a y Astrofı´sica, Pontificia Universidad Cato
´
lica
de Chile, Casilla 306, Santiago 22, Chile.
11
Steward Observatory, University of Arizona, Tucson, AZ 85721.
12
European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748
Garching, Germany.
13
W. M. Keck Observatory, 65-1120 Mamalahoa Highway, Kamuela, HI
96743.
1008
The Astronomical Journal , 129:1008–1017, 2005 February
# 2005. The American Astronomical Society. All rights reserved. Printed in U.S.A.
Advanced Camera for Surveys (ACS) Investigation Definition
Team (IDT) to include the star in its circumstellar disk imaging
program.
A disk around AU Mic was recently imaged from the ground
using coronagra phs (Kalas et al. 2004, hereafter KLM04; Liu
2004, hereafter L04). These images show a disk t hat at first
glance resembles that of Pic. It is viewed nearly edge-on and
has a radius of at least 21
00
(210 AU). The Keck ada ptive optics
image o f L04 also shows s mall-scale variations in the midplane
that might be due to localized d ensity enhancements caused b y
unseen substellar companions that perturb the disk. So far, im-
aging within 1B5 of the star has not bee n possible from the
ground. The high resolution of A CS and its ability to see closer
to the star can provide more detailed views of this disk.
2. OBSERVATIONS, PHOTOMETRY, AND PROCESSING
2.1. Observvations and Calibration
The ACS IDT observations of AU Mic utilized the corona-
graphic mode of the ACS High Resolution Camera ( HRC),
which has a pixel scale of 25 mas pixel
1
and a corona graph ic
field point-spread function (PSF) FWHM of 72 mas at I,63mas
at V, and 50 mas at B.Ther ¼ 0B 9 occulting spot was used for
all observations. The coronagraph suppresses the diffraction
pattern of the central star, reducing the surface brightness of
its wings by 7 times. The remaining flux from the occulted
source is dominated by scattered light from the telescope optics
for r > 1B5 and by light diffracted by the occulting spot for
r < 1B5. The latter appears as concentric ring s around the image
of the spot. Because of the stability of HST, most of the residual
light can be subtracted using an i mage of another star.
Since AU Mic was on the ACS IDT target list before dis-
covery of its disk , the team’s standard observation strategy for
suspected disks was used: initial imaging in one filter to confirm
the existence of a disk, followed later with multicolor imaging.
The ACS expos ures are listed in Table 1. The first images (HST
program 9987) were taken on 2004 April 3 in filter F606 W
(wide V-band) and the second se t (progra m 10330) on 2004
July 24 in F606W, F435W (ACS B), and F814W (ACS wide I ).
The telescope orientations were specified to place the disk diag-
onally on the detector, and the orientations of the two sets of
images were separated by 90
. Exposures of HD 216149 (V ¼
5:4) in the same filters were taken after AU Mic at both epochs for
use in PSF subtraction. This star was chosen because it is bright, is
of a similar color as AU Mic, and is near AU Mic on the sky so that
any attitude-dependent focus variations could be minimized. In
addition to the coronagraphic exposures, short, direct (noncoro-
nagraphic) images were taken to provide photometry of the stars.
All of the images were calibrated by the HST pipeline. To
account for vignetting around the occulting spot, the corona-
graphic images were manually divided by the spot flats avail-
able from the ACS reference files page at the Space Telescope
Science Institute (STScI) Web site. Because the spot varies in
position over time, each spot flat was shifted to the appropriate
position as listed on the ACS Web page. The duplicate expo-
sures were combined with cosmic-ray rejection. At this stage,
the images w ere not corrected for geometric distortion.
2.2. Stellar Photometry
Photometric measurements of the two stars are required to
properly normalize the HD 2161 49 images pr ior to su btraction
from thos e of AU Mic. The AU Mic fluxes were measured from
the short, noncoronagraphic images u sing circular apertures
with radii of 13 (F435W), 25 (F606W), and 30 ( F814W) pixels,
increasing in size to account for the larger PSFs at longer wave-
lengths. The short exposures of HD 216149 were saturated. As
reported by Gilliland (2004),
14
the well depth of an HRC pixel is
160,000 e
, which is within the 16 bit range of the electroni cs
when a gain of 4 e
per data unit is used. Saturating t he pixel
does not saturate the analog-to-digital converter, so flux is not
lost when a pixel saturates but instead overflows into an adjacent
pixel. As Gilliland (2004) showed, in this case the counts are
conserved, and summation within an aperture that has a radius
that encompasses all of the saturated pixels should provide the
same count rate as in an unsaturated image. The fluxes for HD
216149 in the three filters were measured using 30 pixel radius
apertures, which include all of the saturated pixels.
The measured photometry for both stars were corrected
to eff ectively infinite apertures using conversion coeffic ients
derived from encircled energy c urve s of well-exposed, h igh–
dynamic-range ima ges of other stars. B ecause the AU Mic an d
HD 216149 measurements were made using images that had not
been corrected for ACS geometrical distortion, compensation for
pixel area distortion was made using the pixel area maps pro-
vided by STScI (Pavlovsky et al. 2004).
15
The instrumental fluxes
were converted to standard BVI
C
magnitudes (Table 2) using the
STSDAS SYNPHOT synthetic photometry program assuming
M1 V (for AU Mic) and M0 III (for HD 216149) spectra.
2.3. PSF Subtraction
The HD 216149 coronagraphic images were normalized to
match the measured AU Mic fluxes . They were then iterativ ely
TABLE 1
AU Microscopii and Reference PSF Exposures
Star Date Filter
Exposure
(s) Gain
a
Type
b
AU Mic ........... 2004 Apr 3 F606W 2 ; 0.1 4 D
3 ; 750 2 C
2 ; 1400 2 C
2004 Jul 24 F435W 1 ; 0.1 4 D
1 ; 60 2 D
4 ; 1279.5 2 C
F606W 1 ; 0.1 4 D
1 ; 60 2 D
2 ; 915 2 C
F814W 1 ; 0.1 4 D
1 ; 60 2 D
3 ; 853 2 C
HD 216149 ..... 2004 Apr 3 F606W 2 ; 0.1 4 D
2 ; 90 2 C
8 ; 225 2 C
2004 Jul 24 F435W 1 ; 0.1 4 D
2 ; 52D
2 ; 475 2 C
F606W 1 ; 0.1 4 D
2 ; 52D
2 ; 145 2 C
F814W 1 ; 0.1 4 D
2 ; 52D
2 ; 145 2 C
a
In electrons per data number.
b
D = direct (noncoronagr aphic), and C = coronagraphic.
14
ACS Instrument Science Report ISR 04-01 (Gilliland 2004), available at
http://www.stsci.edu/ hst /acs/documents/isrs.
15
ACS Data Handbook (Pavlovsky et al. 2004), available at http://www.
stsci.edu/hst/acs/documents.
THE AU MIC DEBRIS DISK 1009
shifted by subpixel amoun ts via cubic convolution interpola-
tion and subtracted from the AU Mic images until the residual
scattered-light patterns were visibly minimized. This procedure
appears to align the two stars to within 0.05 pix els, a ided by
the high spatial frequency streaks in the residual pattern, which
can be seen radia ting fro m t he star in Figures 1c–1f. After sub-
traction, the images were corrected for geometric distortion.
The diff raction patte rns inside and around the edge of the oc-
culting spot were not well subtracted. Their shapes and inten-
sities are quite sensitive to the alignment of the star behind the
spot,whichatworstdifferedby16masbetweenAUMicand
the reference PSF. They are also sensitive to focus changes a nd
small difference s between the spectra of the stars over a filter ’s
bandpass. Their residuals, which appear as alternating positive
and negative rings, are the dominant source of error for r < 1B5,
where they create localized uncertainties of 30% on 0B1scales
in the disk brightness in the F 606W images. Although the re
are oscillations within and around the spot in both F606W im-
ages, they appear to have mean residuals n ear zero. However, in
F814W the spot interior ap pears oversubtracted, and in F435W
it is undersubtracted, probably be cause o f focus mismatches.
Further out, the background is dominated by scatter-subtraction
residuals, which cause pixel-to-pixe l errors of less than 10% .
In addition to the coronagraphic images, we also subtracted
the longest noncoronagraphic exposure of HD 216149 from
TABLE 2
AU Microscopii and Reference PSF Photometry
Star Date Passband Magnitude
a
AU Mic ...................... 2004 Apr 3 V 8.63 0.03
2004 Jul 24 B 9.96 0.05
V 8.64 0.03
I
C
6.60 0.03
HD 216149 ................ 2004 Apr 3 V 5.46 0.03
2004 Jul 24 B 6.71 0.05
V 5.46 0.03
I
C
4.05 0.03
a
Stated error includes estimate of magnitude system transformati on error.
a) AU Mic F606W
b) HD 216149 F606W
c) F606W (PSF Subtracted)
d) F606W (PSF Subtracted)
e) F435W (PSF Subtracted)
f) F814W (PSF Subtracted)
0" 10"
E
N
Fig. 1.—First-epoch ACS F606W coronagraphic images of AU Mic and the reference PSF star HD 216149, all displayed with logarithmic intensity scaling.
Image (d ) is smoothed and shown using a stronger stretch to reveal the disk at greater heights and structure associated with the galaxy superposed on the northwest
extension. A background star can be seen below the southeast extension.
KRIST ET AL.1010 Vol. 129
that of AU Mic. Although this clearly revealed the disk, it d id
not allow detection of it within 1
00
of the star because of high
subtraction residuals, providing no improvement o ver the co-
ronagraphic subtractions. We also used the two F606W corona-
graphic images of AU Mic to subtract eac h other, but because of
differences in the spot p ositions between the two epochs, th e
subtraction residuals were larger than those when HD 216149
was used.
3. RESULTS
The AU Mic disk stands out again st the background PSF
structure in the ACS coronagraphic images (Fig. 1a)evenbe-
fore PS F subtraction. In the first-epoch F606W image (the deep-
est e xposure), the northwestern side is detected out to the edge
of the detector, 15B25 (150 AU ) from the star, and the south-
eastern side to 14
00
(140 AU ). The disk can be seen to a
height of 2B5 (25 AU) ab ove the midplane. The observed radial
and height extents in this image are sensitivity-limited. The
interior of the occulting spot is filled with light that was not
blocked by the spot (which is located in the aberrated HST
beam) and was afterward modified by the correc tive A CS op -
tics. An image of the star at the spot’s center allows accurate
measurement of the stellar p osition. An extended so urce, which
we show to be a back ground galaxy, can be seen superposed on
the northwest midpla ne, 9B6 fr om the star. A point source is a lso
seen 1
00
southwest of the disk midplane 13 B7southeastfrom
the star. A few background galaxies are also seen throughout
the field.
The P SF-subtra cted image more clearly re veals the d isk
(Figs. 1c–1f ). Because the disk is fairly bright close to the star
and t he corrective optics modify the aberrated light that passe s
by the occulter, the disk can actually be seen inside the spot to
within 0B75 of the star. The residual levels indicate that using
the c oronagrap h with PSF subtraction improves the disk-to-
background contrast by 150 times compared with dire ct im-
aging without subtraction. Significant residuals are present in a
bar of instrumentally scattered light that extends from the upper
left to the lower right of the coronagraphic field, bu t these ar e
positioned away from the disk. The F606W subtractions from
each epoch are of comparable quality, whereas the F814W sub-
traction has large residuals near and within the spot, p robabl y
because of focu s differences between the two i mages. The re-
siduals in the F435W image are more uniform than in the other
two filters, but because AU Mic is much fainter in that pass-
band, the effects of electronic noise are more significant.
The extended source detection limits in the subtracted images
were estimated by adding 1
00
; 1
00
uniform-intensity squares to
thedataat3
00
and 9
00
from the star perpendicular to the circum-
stellar disk. The intensities of the squares were adjusted until
they could no longer be visually detected. T he de rived visual
detection limits are B ¼ 23:3 and 23.5 m ag arcsec
2
(F435W),
V ¼ 22:6 and 24.0 (F606W), and I ¼ 20:4 and 21.9 ( F814W )
for r ¼ 3
00
and 9
00
, respectively, with estimated errors of 0.2 mag
arcsec
2
.
3.1. Disk Morphologgy
Analysis of the disk morphology is largely confined to the
first-epoch F606W observation, since it was the fi rst and deep-
est exposure. As is shown in x 3. 3, there are no significant dif-
ferences in the disk m orphology among the three filters.
To allow accurate measurement of the disk orientation, sim-
ple spatial filtering was applied to highlight the midplane. The
image w as subtracted by a copy of itself that was smoothed by
a3pixel; 3 pixel me dian filter. This removed the extended
vertical wings while preserving the sharp peak o f the midplane
seen within 5
00
of the star ( Fig. 2). A line fitted to the resulting
image shows that the i nner (r < 5
00
) midplane is oriented at
P:A: ¼ 128N6 0N2. There are sm all (40 mas) oscillations in
the midplane position about this line, which passes directly
through the s tar. Superposing this lin e o n the original image
highlights the deviatio n of the outer (r > 5
00
) midpla ne from
the inner (Fig. 3). At large angles, the disk looks ‘‘bowed,’’ an
effect noted in the ground-ba sed images and also seen in the
Pic disk (Kalas & Jewitt 1995). The apparent outer disk
midplane measured by eye at r ¼ 12B5 is oriented along P:A: ¼
311N1 0N7 on the northwest side and P:A: ¼ 125N6 0N7on
1"
Fig. 2.—First- epoch ACS F606W coronagraphic PSF-subtracted image of AU Mic after subtraction of a median-smoothed copy of itself, highlighting the higher
spatial frequency structures, notably the flat, sharp disk midplane (other features are nois e and PSF subtraction residuals). Inverse intensity to and the same orientation as
Fig. 1.
E
N
0" 5"
Fig. 3.—First-epoch ACS F606W coronagraphic image of the AU Mic disk normali zed by the radial brightness profile power laws de scribed in the paper and
displayed in inverse intensity. Data beyond 7B5 have been median-smoothed. The ‘‘bowing’’ can be seen in the outer disk as an upward deviation from the horizontal
line defined by the inner midplane.
THE AU MIC DEBRIS DISK 1011No. 2, 2005
the southeast. These angles are consistent with those reported
by KLM04 and L04.
Radial surfa ce brightness profiles along the disk midplane,
accounting f or the bow, are shown in Figure 4. Both sides of the
disk a ppe ar to have very similar profiles, although the north-
west side is brighter beyond 10
00
(100 AU) by 2 times relative
to the southeast. The mean midplane radial brightness profile
can be reasonably divided into three zones described by different
power laws. The inner zone (r < 1B5) is nearly flat (I / r
0:3
).
The middle zone (1B5 < r < 4B3) has a moderate radial decrease
in brightness (I / r
1:8
), which is slightly steeper than that mea-
sured by L04 (from r
1.0
to r
1.4
). The outer zone (r > 4B3)
drops off rapidly (I / r
4:7
), which is steeper than the r
3.8
measured by KLM04 between 6
00
and 16
00
but is consistent with
the r
4:40:3
reported by L04.
Localized deviations from the large-scale brightness profile
are present. There are 20% dips at r ¼ 2B3 (23 AU) on both
sides that a ppe ar to be genuine and not PSF subtraction arti-
facts. Between 4B3and5B3 the sou theast side is 25% brighter
than the northwest. Within 1B5 the subtraction residuals are too
large to identify any localized asymmetries of less than 30%. To
highlight the small-scale variations, each row in the image
parallel to the midplane was divided by a smooth function, a
fourth-order polynomial fit derived from the mean midplan e
radial brightness profile. T he resulting image (Fig. 5) empha-
sizes the deviations and can be compared with Figure 3 of L04.
The dip at 23 AU is clear, altho ugh any features closer to the star
than it a re suspec t. There are correlatio ns between feature s in
these images and those seen in the L04 data (L04 labeling con-
vention used; distances in parenthe ses are the centers of the fea-
tures as seen in the L04 image): (A) a ‘‘clump’’ at 26 AU (25 AU),
(B) a dip at 29 AU (29 AU), and (C) a clump at 33 AU (31 AU).
The relatively large discrepancy in the position of feature C be-
tween the HST and L04 data may be due to subtraction errors in
either or both data sets. We note that the positions of the features
appear the same in the HST images at both epochs. At 37 AU on
the northwest side there is a marginal enhancement that may cor-
respond to L04’s clump D, whereas at the same radius in the
southeast side there is a slight dip. Beyond that radius the north-
west side smoothly declines, but there is one more extended
clump at 48 AU in the southeast.
The inner disk’s vertical brightn ess profile ( Fig. 6) shows a
sharp midplane with extended wings. It can be reasonably well
characterized for disk radii less than 5
00
by a Lorentzian profile:
I(z) ¼ 1=(1 þ z
2
=h
2
), where 2h is the profile FWHM and z is
the height above the midplane. Note that we find this particular
shape to be a convenient description of the disk profile and do
not assert any physical sig nificance to it. The profiles are gen-
erally symmetric about the midplane for r < 5
00
, but beyond this
the disk is brighter on the northeast side of the midp lane. T he
outer disk profile is more rounded, lacking a sharp midplane.
The FWHM at each radius was measured by fitting a
Lorentzian profile to a vertical cross-section of the image after
smoothing by 5 pixels along the direction o f the midplan e. As
shown in Figure 7, the inner disk is rather flat, but the outer disk
begins to rapidly broaden bey ond 5
00
.TheFWHMforr < 5
00
(50 AU ) is 0B25–0B35 (2.5–3.5 AU). At r ¼ 7B5 (75 AU) the
FWHM on the southeast side is 0B9(9AU)and0B65 (6.5 AU)
on the northwest. The change in FWHM along the southeast side
can be approximately described by two power laws: FWHM /
r
0:07
for 1B5 < r < 4B6andr
2.4
for 4B6 < r < 7B5. The equivalent
relations for the northwest side are FWHM / r
0:07
for 1B5–3B0
and r
1.0
for 3B0–7B5. These apparent vertical profiles represent the
projection of the optically thin disk along the line of sight con-
volved with the instrumental PSF.
3.2. Disk Modelingg
Derivation of the physical distribution and properties of the
dust in the disk directly from an image is prohibited by the
integrated effects of forward scattering and density variations
seen alo ng the line of sight. They can b e estimated from three-
dimensional scattering models optimized to match the observed
image. We have attempted to derive a reasonable set of model
parameters that match the major features of the disk in the
1 10
Radius (arcsec)
24
22
20
18
16
V mag arcsec
-2
Northwest
Southeast
r
-1.8
r
-0.3
r
-4.7
Fig. 4.—Radial surface brightness profiles following the AU Mic disk mid-
plane (including the bow) measured over 0B25 ; 0B25 subapertures in the first-
epoch ACS F606W coronagraphic image. Power-law profiles are superposed.
0 102030405060708090100
0
0
AU
AU
E
N
E
N
NW side
SE side
A B C
D
Fig. 5.—AU Mic disk ACS F606W image divided by a smooth function fitted to the midplane to highlight localized variations. The southeast side has bee n
flippe d about the star for easier comparison.
KRIST ET AL.1012 Vol. 129
