Does the Debris Disk around HD 32297 Contain Cometary Grains?

0.44 MB·

by  Timothy J. Rodigas

Draftversion January16,2014
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
DOES THE DEBRIS DISK AROUND HD 32297 CONTAIN COMETARY GRAINS?*+
Timothy J. Rodigas1,2, John H. Debes3, Philip M. Hinz1, Eric E. Mamajek4, Mark J. Pecaut4,5, Thayne Currie6,
Vanessa Bailey1, Denis Defrere1, Robert J. De Rosa7, John M. Hill8, Jarron Leisenring1, Glenn Schneider1,
Andrew J. Skemer1, Michael Skrutskie9, Vidhya Vaitheeswaran1, Kimberly Ward-Duong7
Draft version January 16, 2014
ABSTRACT
4 WepresentanadaptiveopticsimagingdetectionoftheHD32297debrisdiskatL′(3.8µm)obtained
1 with the LBTI/LMIRcaminfraredinstrumentatthe LBT.The disk is detected atsignal-to-noiseper
0 resolution element (SNRE) 3-7.5 from 0′.′3-1′.′1 (30-120 AU). The disk at L′ is bowed, as was
2 seenatshorterwavelengths.∼Thislikelyind∼icatesthediskisnotperfectlyedge-onandcontainshighly
forward scattering grains. Interior to 50 AU, the surface brightness at L′ rises sharply on both
n
∼
sides of the disk, which was also previously seen at Ks band. This evidence together points to the
a
J disk containing a second inner component located at . 50 AU. Comparing the color of the outer
(50 < r/AU < 120) portion of the disk at L′ with archival HST/NICMOS images of the disk at 1-2
4
µm allows us to test the recently proposed cometary grains model of Donaldson et al. (2013). We
1
findthatthe modelfailstomatchthe disk’ssurfacebrightnessandspectrumsimultaneously(reduced
] chi-square = 17.9). When we modify the density distribution of the model disk, we obtain a better
P overallfit(reducedchi-square=2.9). Thebestfittoallofthedataisapurewatericemodel(reduced
E chi-square = 1.06), but additional resolved imaging at 3.1 µm is necessary to constrain how much (if
. any) water ice exists in the disk, which can then help refine the originally proposed cometary grains
h
model.
p
Subject headings: instrumentation: adaptive optics — techniques: high angular resolution — stars:
-
o individual (HD 32297)— circumstellar matter — planetary systems
r
t
s
a 1. INTRODUCTION sharp edges (Schneider et al. 2009; Chiang et al. 2009),
[ and with their specific dust grain compositions. Since
Debris disks, which are thought to be continually
1 replenished by collisions between large planetesimals manyoutersolarsystembodiescontaincopiousamounts
of water ice and organic materials, finding other debris
v (Wyatt 2008), canpoint to interesting planets in several
disksystemsthatcontainwatericeand/ororganicmate-
3 ways: with warps and/or gaps (Lagrange et al. 2010),
rials (Debes et al. 2008a) would point to planetary sys-
4
tems that might contain the ingredients necessary for
3
*Based on observations made at the Large Binocular Tele- Earth-like life. Therefore constraining the dust grain
3
scope(LBT). TheLBTisaninternationalcollaborationamong
compositions in debris disks is crucial.
. institutionsintheUnitedStates,ItalyandGermany. LBTCor-
1 poration partners are: The University of Arizona on behalf of Narrowandbroadbandscatteredlightimagingisapar-
0 the Arizona university system; Istituto Nazionale di Astrosica, ticularly powerful tool for constraining composition be-
4 Italy; LBT Beteiligungsgesellschaft, Germany, representing the cause it can substitute for spectra that would otherwise
1 Max-Planck Society, the Astrophysical Institute Potsdam, and
betoodifficulttoobtain. Thewavelengthrangebetween
: HeidelbergUniversity; TheOhioState University, andTheRe-
v searchCorporation,onbehalfofTheUniversityofNotreDame, 1-5µm,inparticular,containsstrongabsorptionfeatures
i UniversityofMinnesotaandUniversityofVirginia. forwatericeandorganicsliketholins(bothnear3.1µm;
X +BasedonobservationsmadeusingtheLargeBinocularTele- Inoue et al. (2008); Buratti et al. (2008)). Imaging at
scope Interferometer (LBTI). LBTI is funded by the National
r these wavelengths from the ground using adaptive op-
a Aeronautics and Space Administrationas part of its Exoplanet
Explorationprogram. tics(AO)alsooffershighStrehlratios,allowingformore
1Steward Observatory, The University of Arizona, 933 precisecharacterizationoffaintextendedsourcescloseto
N. Cherry Ave., Tucson, AZ 85721, USA; email: rodi- their host stars.
[email protected]
2Carnegie Postdoctoral Fellow; Department of Terrestrial Obtaining high signal-to-noise (S/N) detections of
Magnetism, Carnegie Institute of Washington, 5241 Broad faint debris disks at these wavelengths from the ground
BranchRoad,NW,Washington, DC20015, USA is challenging due to the bright thermal background of
3Space Telescope Science Institute, Baltimore, MD 21218, Earth’s atmosphere, as well as the warm glowing sur-
USA
4Department of Physics and Astronomy, University of faces in the optical path of the telescope. An AO sys-
Rochester, Rochester,NY14627-0171, USA temthatsuppressesunwantedthermalnoiseisnecessary
5RockhurstUniversity,1100RockhurstRd,KansasCity,MO to overcome these obstacles. The Large Binocular Tele-
64110, USA
scope (LBT), combined with the Large Binocular Tele-
6University of Toronto, 50 St George St., Toronto, ON M5S
1A1,Canada scope Interferometer (LBTI, Hinz et al. (2008)), is one
7School of Earth andSpace Exploration, ArizonaState Uni- suchsystem. TheLBTAOsystem(Esposito et al.2011)
versity,POBox871404, Tempe,AZ85287-1404, USA consistsoftwo secondarymirrors(one for eachprimary)
8Large Binocular Telescope Observatory, University of Ari-
thatcaneachoperatewithupto400modesofcorrection,
zona, Tucson,AZ85721, USA
9University of Virginia, Department of Astronomy, 530 Mc- resultinginveryhighStrehlratios( 70-80%atH band,
∼
CormickRoad,Charlottesville,VA22903, USA 90% at Ks band, and > 90% at longer wavelengths).
∼
2 Rodigas et al.
Thisequatestoveryhigh-contrast,high-sensitivityimag- with the core of the star saturated out to 0′.′1. After fil-
ingcapabilities,allowingdetectionsofplanetsanddebris tering out images taken while the AO loop was open13,
disks that were previously too difficult. thefinaldatasetconsistedof726images,resultingin2.99
HD 32297 is a young A star located 112 pc away hours of continuous integration (not including the min-
(van Leeuwen 2007) surrounded by a bright edge- imum exposure photometric images). Throughout the
on debris disk. The disk has recently been re- observations, which began just after the star’s transit,
solved at Ks band (2.15 µm) by Currie et al. (2012), the FOV rotated by 50.84◦, enabling the star itself to
Boccaletti et al. (2012), and Esposito et al. (2013). The actasthe pointspreadfunction(PSF)referenceforsub-
disk has also been detected in the far-infrared (FIR) traction during data reduction.
by Herschel(Donaldson et al. 2013, hereafter D13), who
2.2. Data reduction
modeledthediskasconsistingofporouscometarygrains
(silicates, carbonaceous material, water ice). Alldatareductiondiscussedbelowwasperformedwith
To further test this model, we have obtained a custom Matlab scripts. We first divided each science
high S/N image of the disk at L′ (3.8 µm) with exposure image by the number of coadds (15) and in-
LBTI/LMIRcam. Using this new image, along with tegration time (0.99 s) to obtain units of counts/s for
archival HST/NICMOS images of the disk at 1-2 each pixel. Next we corrected for bad pixels and sub-
µm from Debes et al. (2009)12, we determine how well tractedoppositechopbeamimagesofthestartoremove
the D13 cometary grains model matches the scattered detector artifacts and the sky background, resulting in
light of the disk at 1-4 µm. flat images with 0 background counts/s, except for
∼
In Section 2 we describe the observations and data the“patchy”regionsofhigherskynoise. We determined
reduction. In Section 3 we present our results on the the sub-pixel location of the star in each sky-subtracted
disk’s surface brightness (SB) from 1-4 µm, analysis of imagebycalculatingthecenteroflightinsidea0′.′5aper-
thedisk’smorphology, andourdetectionlimitsonplan- ture centered on the approximate location of the star.14
ets in the system. In Section 4 we present our modeling Wethenregisteredeachimagesothatthestar’slocation
of the disk. In Section 5 we discuss the implications of was at the exact center of each image. We binned each
our results on the disk’s structure and composition, and image by a factor of 2 to ease the computational load
in Section 6 we summarize and conclude. required in processing 726 images, which is not a prob-
lembecausethe PSFisstilloversampledbymorethana
2. OBSERVATIONSANDDATAREDUCTION factor of 4. To reduce the level of the patchy sky back-
2.1. Observations ground, for each image we subtracted a 15 pixel (0′.′32)
by15pixelboxmedian-smoothedimagefromitself. The
We observed HD 32297 on the night of UT November
unsaturated minimum exposure images were reduced as
4 2012atthe LBT onMt. GrahaminArizona. We used
LBTI/LMIRcam and observed at L′ (3.8 µm). LMIR- detailedabove,exceptforthislaststepofunsharpmask-
ing,sincethe majorityofthe star’sfluxdoes notoverlap
cam has a field of view (FOV) of 11′′ on a side and
a plate scale of 0′.′0107/pixel. Ski∼es were clear during with the backgroundpatches. It is also not necessary to
the observations, and the seeing was 1′.′3 throughout. unsharp mask the unsaturated photometric images be-
∼ cause,as will be discussedin Section A, we compute the
Observations were made in angular differential imaging
fully-corrected disk flux that accounts for the unsharp
mode (ADI; Marois et al. (2006)), and no coronagraphs
maskingandotherbiasesinherentinthe datareduction.
were used. We used the single left primary mirror with
During the observing run, the LMIRcam detector suf-
its deformable secondary mirror operating at 400 modes
fered from “S-shaped” non-linearity (which has since
fortheAOcorrection. Toincreasenoddingefficiency,we
been corrected). This had the effect of artificially in-
choppedaninternalmirror(ratherthannodding the en-
flating raw counts per pixel at long exposures relative
tiretelescope)byseveralarcsecondseveryfewminutesto
to short exposures, complicating photometric compar-
obtainimagesoftheskybackground. Whilethismethod
isons. Using a linearity curve constructed from images
did dramatically increase efficiency (to 85%), neglect-
∼ taken throughout the observing run, we multiplied each
ingtomovethe telescoperesultedinaresidual“patchy”
reducedimage’spixelsbythecorrespondinglinearitycor-
background that remained even after sky subtraction.
rection factors. We verified the effectiveness of the lin-
Thiswasduetothedifferenceinopticalpaththroughthe
earity correction by comparing the total flux within an
instrument, such that the chop positions were not per-
annulus centered on the star between 0′.′1-0′.′2 for the
fectly matched. Ultimately we removed this unwanted
unsaturated (linear) final image and the final (linearity-
background by unsharp-masking all images, which re-
corrected)long exposureimage; the mediancounts/sfor
sults in self-subtraction of the debris disk; we account
the two images agreedto within 4%, verifying the lin-
forandmitigatethiseffectviainsertionofartificialdisks, ∼
earity correction.
which we discuss in Section A.
Weobtained759imagesofHD32297incorrelateddou- 13 Filtering out images of poor quality can be accomplished in
ble sampling (CDS) mode, so that eachimage cube con- several ways: the fits headers contain keywords relating to the
sistedof15coaddedminimumexposure(0.029s)images statusoftheAOloop,whichausercancheckinthedatareduction;
containing the unsaturated star (for photometric com- the raw images themselves can be examined by the user to check
the appearance of the PSF (pointy vs. blurry); or the user can
parison),and 15 coadded 0.99 s science exposure images
checkthelogoftheobservations,whichshoulddenotewhen/where
problemswiththeAOoccurred. Weusedthefirstandthirdoptions
12 While resolved images of the disk have also been obtained tofilteroutopenloopdataonHD32297.
fromthegroundatthesewavelengths,theHSTdataarepreferred 14 Since the PSF is only saturated out to 0′.′1, it contributes
because they do not suffer from the biases inherent in ADI/PCA verylittletothecenteroflightcalculation;wehavedemonstrated
datareduction. sub-pixelaccuracyusingthismethodinRodigasetal.(2012).
3
As a first check on the efficacy of the steps described thestandarddeviationoftheequivalentSBmeasurments
above, we performed classical ADI subtraction by sub- all around the star (excluding the disk). We found that
tracting a median-combined master PSF image from all bothmethodsresultedincomparableerrors,thereforewe
the images, and then derotating the images by their chose to use the second method since this method was
parallactic angle at the time of the exposure. The re- also used for computing the errors on the LBT data.
sulting image revealed edge-on disk structure at the ex-
pected position angle (PA) of 47◦(Debes et al. 2009; 3.2. Surface Brightness Profiles
Mawet et al. 2009). ∼ In addition to the image of the disk at L′, we also re-
To obtain the highest possible S/N detection of analyzed archival, reduced images of the disk at 1.1,
HD 32297’s debris disk, we reduced the images us- 1.6, and 2.05 µm from HST/NICMOS present∼ed in
ing principal component analysis (PCA, Soummer et al. Debes et al. (2009).16 For each image, we rotated the
(2012)). PCA has recently been shown to pro- diskby47◦counterclockwisesothatitsmidplanewasap-
duce equal-to or higher S/N detections of plan- proximately horizontal in the image. We measured the
ets and disks (Thalmann et al. 2013; Bonnefoy et al. SB in all the HST/NICMOS images by taking the me-
2013; Boccaletti et al. 2013; Soummer et al. 2012; dian value in 3 pixel (0′.′2262) by 3 pixel boxes. For the
Meshkat et al.2013)thanLOCI(Lafreni`ere et al.2007), F110W (1.1 µm) images, these boxes were centered on
though this may be a result of LOCI’s tunable pa- the brightest pixel at each specific pixel distance from
rameters not being optimized correctly (C. Marois, pri- the star (0′.′45-1′.′1). For the F160W (1.6 µm) and the
vate communication). We did also reduce the data us- F205W(2.05µm)images,theboxeswerecenteredonthe
ing conventional LOCI algorithms (Rodigas et al. 2012; same pixel locations as were used in the F110W image.
Currie et al.2012;Thalmann et al.2011)butfoundthat Weconvertedthe imagecounts/stomJy/arcsecond2 us-
the gain in computational speed for PCA with no loss ing the reported photometric conversions given on the
in S/N warranted the ultimate preference of PCA over Space Telescope Science Institute website and applied
LOCI for this dataset. the appropriate correction factors to the data (see the
We followed the prescription outlined in Appendix for a detailed description of these correction
Soummer et al. (2012), with the main tunable pa- factors).
rameter being K, the number of modes to use for a For the final L′ image, we calculated the median
given reduction. Increasing K reduces the noise in counts/s in a 5 pixel (0′.′106) by 5 pixel box centered
the final image but also suppresses the flux from the on the brightest pixel at each horizontal distance from
disk; therefore this parameter must be optimized. After 0′.′45-1′.′1 from the star, and divided this number by the
examining the average S/N per resolution element plate scale of the binned images (0′.′0212) squared. We
(SNRE)15 over the disk’s spatial extent for varying used a smaller aperture for the L′ data than for the
values of K, we determined the optimal number of HST/NICMOS data due to the disk appearing thinner
modes to be K = 3 (out of a possible 726). We then (FWHM 0′.′1 λ/D) at L′, and to avoid the large
fed all the images through our PCA pipeline, rotated negative r∼esidual≈s above and below the disk close to the
the images by their corresponding PAs clockwise to star. WeconvertedthesevaluestomJy/arcsecond2using
obtain North-up East-left, and combined the images the total measured flux in the unsaturated photometric
using a mean with sigma clipping. Fig. 1a shows this image of HD 32297 and applied the appropriate correc-
final image, and Fig. 1b shows the correspondingSNRE tion factors to the data (see the Appendix for a detailed
map. We detectthe disk from 0′.′3-1′.′1(30-120AU)at description of these correction factors).
∼
SNRE 3-7.5, which is a significant improvement from Fig. 2 shows the final, corrected SB of the disk at
our pre∼vious L′ imaging of HD 15115 (Rodigas et al. 1-2 µm and at 3.8 µm. The star’s magnitude at 1-2
2012). The detection of the disk at high S/N allows us µm(7.7)andL′ (7.59)hasbeensubtractedfromthedisk
to more precisely measure its SB, which in turn results magnitude/arcsecond2 to yieldthe intrinsicdisk colorat
in better constraints on the composition and size of the each wavelength. We find that the SB profile is asym-
dust grains producing the observed scattered light. metric from 0′.′5-0′.′8 (56-90 AU) at L′, in agreement
∼
3. RESULTS withtheasymmetryatKsbandreportedbyCurrie et al.
(2012) and Esposito et al. (2013). We do not find evi-
3.1. Calculation of Uncertainties dence for asymmetry interior to 0′.′4 (45 AU), however,
For all the images analyzed in this study, the uncer- with no evidence for the bright spot identified as a mm
tainties are dominated by residuals left over from PSF peakbyManess et al.(2008)andalsoseenatKsbandby
subtraction. Specifically, both the HST and LBT data Currie et al.(2012). Exteriorto 0′.′8(90 AU), bothsides
sufferfromazimuthalandradialresidualstructures. For ofthe diskare equalSBatL′. TheHSTdataaregen-
the LBT data, we computed the standard deviation of erally consisten∼t from 1-2 µm, within the uncertainties,
the equivalentSB measurements allaroundthe star(ex- exceptforinteriorto0′.′8(90AU)wherethenortheastern
cluding the disk). For the HST data, we calculated the lobe appears to be brighter than the southwestern lobe.
errorsintwoways: bycomputingtheequivalentSBmea- We fit the SB profiles at each wavelength to functions
surements90◦awayfromtherealdisk;andbycomputing with power-law form (see Table 1). We find that the in-
dicesareallwithin2σ ofeachotherfortheouterportion
15 As in Rodigasetal. (2012), the SNRE map is calculated by of the disk (0′.′5 < r < 1′.′1), with most values between
convolving the final image by a Gaussian with full-width half-
maximum=0′.′094, constructinganoiseimagebycomputingand -1.3 and -1.5. Interior to 0′.′4, the disk is not detected
∼
storing the standard deviation in concentric annuli of width = 1
pixel around the star, then dividing the final Gaussian-smoothed 16 We refer the reader to Debesetal. (2009) for these images
imagebythisnoiseimage. anddescriptionsofhowtheywerereduced.
4 Rodigas et al.

0.8 7
1 1
0.6 6
0.5 0.5 5
0.4
arcseconds 0 0.2 arcseconds 0 4
3
−0.5 −0.5
0 2
−1 −1
1
−0.2
0
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
arcseconds arcseconds
(a) (b)
Fig.1.—Left: FinalreducedL′imageoftheHD32297debrisdisk,inunitsofdetectorcounts/s,withNorth-up,East-left. Thewhitedot
marksthelocationofthestarandrepresentsthesizeofaresolutionelementatL′. A0′.′2radiusmaskhasbeenaddedinpost-processing.
The southwest side of the disk is ∼ 0.5 magnitudes/arcsecond2 brighter than the northeast side from 0′.′5-0′.′8 (56-90 AU), which was
also seen at Ks band (Currieetal. 2012). However there is no brightness asymmetry at the location of the mm peak first identified by
Manessetal. (2008) and later seen at Ks band by Currieetal. (2012). Right: SNRE map of the final image. Both sides of the diskare
detected from∼0′.′3-1′.′1(30-120AU)atSNRE∼3-7.5.
3 3
L’ L’
F110W F110W
F160W F160W
2d)4 northeast F205W 2d)4 southwest F205W
n n
o o
ec ec
arcs5 arcs5
gs/ gs/
a a
m m
∆ess ( 6 ∆ess ( 6
n n
ht ht
Brig7 Brig7
e e
ac ac
Surf8 Surf8
9 9
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
distance from star (arcseconds) distance from star (arcseconds)
(a) (b)
Fig.2.—Left: SBprofilesfortheHD32297debrisdiskforthe1-2µmHST/NICMOSand3.8µmLBTIdataforthenortheasternlobe.
Right: The same except for the southwestern lobe of the disk. Interestingly, the SB asymmetry at 0′.′5-0′.′8 (56-90 AU) first noticed by
Currieetal.(2012)isalsoevidentintheL′data. Inwardof0′.′5,thedisk’sSBdeclinesmoresteeplythanatlargerstellocentricseparations,
and we do not find evidence for an asymmetry near 0′.′4 (45 AU) that was identified as a mm peak by Manessetal. (2008). Exterior to
0′.′8(90AU),theHST1-2µmdataaregenerallyconsistentwiththeL′ datawithintheuncertainties.
at 1-2 µm with HST/NICMOS but is detected at L′. disk relative to the midplane as a function of distance
At these close distances, the disk SB clearly falls faster, from the star. We measure the midplane offset, as op-
declining like r−2.6. We do not compare our reported posed to the disk’s PA, because the former is a more
∼
power-lawindicestoindicesreportedinotherworksmea- intuitive indicator of a bow-shaped disk. The offsets
sured farther from the star because the disk is thought were measured in manners analogous to those described
to havea breakinthe SBdistributionnear110AU(1 ′′; in Rodigas et al. (2012) and Currie et al. (2012). Fig. 3
Boccaletti et al. 2012; Currie et al. 2012). showstheseoffsetsforthenortheasternandsouthwestern
lobes, along with the midplane offsets for the Ks band
3.3. Midplane Offset Measurements data from Currie et al. (2012) for reference. The disk
is clearly bowed at L′, with the offsets increasing closer
Currie et al.(2012)measuredthe PA ofthe HD 32297
to the star on both sides of the disk to a peak value of
debris disk as a function of separation from the star at
0′.′04 (4.5 AU). This is comparable to the peak mid-
Ksbandandfoundthatthe diskwasbowedclosetothe
∼
planeoffsetreportedbyEsposito et al.(2013)andagrees
star. Similar bowing was reported by Boccaletti et al.
with the peak offsets in the Ks band data (Fig. 3c and
(2012) and Esposito et al. (2013). To test whether the
bow shape is seen at L′, we measured the offset of the Fig. 3d).
5
TABLE 1
SurfaceBrightness Profile Power-law Indices
3.8µm 2.05µm 1.6µm 1.1µm
northeastouter(0′.′5<r<1′.′1) -1.5(-2.45,-0.55)∗ -1.83(-2.12,-1.53) -1.56(-1.99,-1.13) -1.33(-1.73,-0.93)
southwestouter(0′.′5<r<1′.′1) -1.3(-1.86,-0.72) -1.56(-1.88,-1.24) -1.55(-1.92,-1.17) -1.32(-1.52,-1.11)
northeast inner(r<0′.′5) -2.45(-3.36,-1.53) – – –
southwest inner(r<0′.′5) -2.72(-10.74,5.3) – – –
∗ Theseandallotherparenthetical valuesdenote95%confidence bounds.
0.1 0.05
0.08
0.04
0.06
midplane offset (arcseconds)−−0000....000024420 northeast midplane offset (arcseconds) 000...0001230 southwest
−0.06
−0.01
−0.08
−0.1 −0.02
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
distance from star (arcseconds) distance from star (arcseconds)
(a) (b)
0.1 0.06
0.09
0.055
0.08
nds)0.07 nds) 0.05
o o
arcsec0.06 arcsec0.045
et (0.05 et ( 0.04
offs offs
ne 0.04 ne 0.035
a a
dpl0.03 dpl
mi mi 0.03
0.02
0.025
0.01
0 0.02
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
distance from star (arcseconds) distance from star (arcseconds)
(c) (d)
Fig.3.—Topleft: Diskmidplaneoffsetasafunctionofdistancefromthestarat3.8µmforthenortheasternlobe. Topright: Thesame
except forthesouthwestern lobeofthedisk. Bottom row: thesameasthetoprow,except theoffsetsaremeasuredontheKsbanddata
fromCurrieetal.(2012). TheoffsetfromthemidplaneincreasesclosertothestaratbothL′ andKsband,indicatingabow-shapeddisk.
Theoffsetspeakat∼0′.′04-0′.′06(4.5-7AU).
3.4. Revised Spectral Classification, Luminosity, Mass, thespectrumtoadensegridofMKstandardstarsusing
and Age of HD 32297 the python tool “sptool”18, we estimate the star’s spec-
tral type to be kA3hA6mA6 V (see Fig. 4a).19 The oft-
Estimating the masses of exoplanets detected via di-
quotedspectraltype forthe starofA0isclearlytoohot.
rectimagingrequiresatmosphericmodels,whichdepend
Basedonthe hydrogentype ofA6, weadoptaneffective
heavilyontheageoftheplanetandthereforeonthehost
temperatureofT =8000 150Konthemoderndwarf
star. The age of HD 32297, like many stars, is poorly eff
±
T vs. spectral type scale from Pecaut & Mamajek
constrained. Estimatesofthe star’sspectraltype, which eff
(2013). Plotting the star’s updated position on an HR
can be used to constrain its age, have ranged from A0
diagram(Fig. 4b),alongwithisochronesfromtheevolu-
(Torres et al. 2006) to A5 (Heckmann 1975). Based on
anarchivalopticalspectrumofthestartakenon2Febru-
18 http://rumtph.org/pecaut/sptool/
ary 2006 with the 300 line grating of the FAST spectro-
19Eachofthelower-caseprefixletterscorrespondstoadifferent
graph on the Tillinghast telescope17, and comparison of
spectraltypingmethod;“k”referstothestar’sspectraltypebased
on Ca K absorption; “h” refers to the star’s hydrogen type; “m”
17seeFASTarchiveathttp://tdc-www.harvard.edu/cgi-bin/arc/fsearcrheferstothestar’smetal-type.
6 Rodigas et al.
In addition to being sensitive to scattered light from
debris disks at L′, imaging at this wavelength is also
particularlysensitive to self-luminous planets, which are
expected to become redder (from 1-5 µm) as they age
and cool (Burrows et al. 2003; Baraffe et al. 2003). Un-
derstandingtheconnectionbetweenmassiveplanetsand
debris disks is critical to constraining planetary system
architectures and consequently planet formation.
By inspection of the SNRE map in Fig. 1b, only one
sourcestandsat5σabovethenoise(thetypicalminimum
thresholdfordetectingimagedexoplanets). Howeverthis
feature, located 1′′ below the star to the southeast, is
∼
not point-like in the final reduced image (Fig. 1a), is
not very symmetrical, and does not persist for different
reduction methods (e.g., varying the number of modes
used in PCA). Therefore we do not treat this as a real
(a)
astronomical source. Other than this feature, there are
no point-source features outside the disk at S/N > 5 in
the final L′ image.
To ascertain what planets we could have detected, we
assume that any planets in the HD 32297 system must
currently reside within the debris disk itself and there-
foreinsertartificialplanetsintothemidplaneofthedisk.
The insertion of artificial sources into the disk is a valid
method because the disk is likely to be optically thin, so
thesignalsfromanyrealembeddedplanetsshouldtravel
toEarthrelativelyunimpeded. Wevarythebrightnesses
ofthe artificialplanetsandre-reducethe datauntil each
is recovered at 5σ confidence. The S/N is calculated
≥
asthepeakpixelvalueintheSNREmapatagivenposi-
tion. Artificialplanetsaremadebyextractingthecentral
0′.′094 (=FWHM at L′) of the unsaturated photometric
imageofHD32297. The reductionpipeline forthe plan-
ets uses the same parameters as were used to detect the
disk at high S/N except that we set K =5.
WhencalculatingtheS/Nsoftherecoveredplanets,we
(b) havetobecarefulabouthowweinterpretthevaluessince
Fig.4.— Top: Visible spectrum of HD 32297, used for deter- signalsfrom the planets lie on top of the signalfrom the
mining its spectral type, which we classify as A6V. Bottom: HD disk. Thereforeanymeasurementofarecoveredplanet’s
32297’s position on an HR diagram with pre-main sequence and S/N must first subtract out the S/N of the recovered
main sequence isochrones from Bressanetal. (2012) for an as-
disk. After doing this, only planets at a contrastlevel of
sumedprotosolarcompositionwithHeliummassfractionY=0.27
and metal mass fraction Z = 0.017. The star is likely older than 2.5 10−5 (11.5 mags) were detected at 5σ confidence
∼15Myrandyoungerthan∼500Myr. Inthisstudy,wetakethe beyo×nd 1′.′25. At a contrast level of 5 10−5, all planets
ageofthestartobe100Myr. with separations 0′.′75 were successf×ully detected, and
at5.9 10−5 allp≥lanets at 0′.′5 weredetected. Within
tionary tracks of Bressan et al. (2012), we estimate that 0′.′5, o×nly planets 100 times≥fainter than the star could
HD32297isolderthan 15Myrandyoungerthan 0.5
∼ ∼ be detected at > 5σ confidence, and even these become
Gyr. The star’s kinematics and the debris disk’s high
highly elongateddue to the increasedself-subtraction so
fractional luminosity both point to a young age for the
close to the star. This self-subtraction could in prin-
star ( 10-20 Myr). However, we conservatively adopt
∼ ciple be removed by reducing the data more carefully
a stellar age of 100 Myr for this study because no plan-
(e.g., including a minimum azimuthal field rotation be-
ets are detected, meriting conservative upper limits on
fore subtracting a PSF image), but such an exhaustive
planet mass.
optimizationof PCAis unnecessarysince it wouldprob-
We estimate HD 32297’s luminosity as log(L/L⊙) = ably not increase contrast levels here by more than a
0.79 0.09 dex. For three sets of Bertelli et al. (2009)
± factor of 10. Fig. 5 shows three example images, along
tracks of varying composition, allowing only for vary-
with their S/N maps, of artificial planets inserted into
ing T and log(L) points that were “physical” (i.e. not
eff the disk. Fig. 6 summarizes all the planet detection re-
below the zero agemain-sequence),andassumingproto-
sults. Adopting a stellar age of 100 Myr and using the
solar composition, we determine the mass of HD 32297
hot-startatmosphericmodelsofBaraffe et al.(2003),we
tqouobteed1.h65igh±er0.m1aMss⊙v.aTluheis(∼is3mMuc⊙h)smthaaltlecrotrhreasnptohnedsoftto- sreuplearoauttiopnlsanet0s′.′m5o(r5e6mAaUs)s,ivaentdhapnla8neMtsJmaotreprmojaescstievde
a star of spectral type A0. than 6 M≥beyond 1.′.′25 (140 AU).
J
∼
3.5. Limits on Planets 4. MODELINGTHEDEBRISDISK’SDUST
7

1 1 1
1 1 1
0.8 0.8 0.8
0.5 0.6 0.5 0.6 0.5 0.6
arcseconds 0 0.4 arcseconds 0 0.4 arcseconds 0 0.4
−0.5 0.2 −0.5 0.2 −0.5 0.2
0 0 0
−1 −1 −1
−0.2 −0.2 −0.2
−1 −0.5 arcse0conds 0.5 1 −1 −0.5 arcse0conds 0.5 1 −1 −0.5 arcse0conds 0.5 1
(a) (b) (c)
9 9 9
8 8 8
1 1 1
7 7 7
0.5 6 0.5 6 0.5 6
arcseconds 0 45 arcseconds 0 45 arcseconds 0 45
−0.5 3 −0.5 3 −0.5 3
2 2 2
−1 −1 −1
1 1 1
−1 −0.5 arcse0conds 0.5 1 0 −1 −0.5 arcse0conds 0.5 1 0 −1 −0.5 arcse0conds 0.5 1 0
(d) (e) (f)
Fig.5.— Artificial planets of varying brightnesses inserted into the HD 32297 debris disk and recovered. The top row consists of the
re-reducedimageswiththeplanetsinserted,andthebottom rowisthecorrespondingS/Nofthedetections (aftersubtractingtheS/Nof
thediskitself). Fromlefttoright,thebrightnessesoftheplanetsincrease,withtheleftmostpanelshowingno5σdetections (planetsthat
are 10−5 times fainter than the star), the middle panel showing detections beyond 0′.′75 (contrast level of 5×10−5), and the rightmost
showingsuccessfuldetections forallplanetsat≥0′.′5(contrastlevelof5.9×10−5).
warmer component is preferred to fit an additional hot
5 component of dust (D13, Currie et al. (2012)), but this
isunobservableatthecurrentinnerworkingangles. Any
6 modelmustbe able to reproducethe SB distributions in
Fig. 2.
7
s) 4.1. Scattered light model of an optically-thin edge-on
g
ma 8 disk
ast ( We construct a model of the disk in a similar fashion
ontr 9 to Currie et al. (2012) (and references therein). An an-
c
σ5 alytical density distribution of dust is generated in a 3
10 dimensional array and sampled in a Monte Carlo fash-
8 MJ, 100 Myr ion with 2 million particles representing a model dust
11 population. Scattering anglesarecalculatedfor the den-
5 MJ, 100 Myr sitydistributionwithagivenPAandinclination. Inputs
12 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 to the model include a cross-sectional averaged asym-
distance from star (arcseconds) metry parameter < g >, which can be used to model
the forward scattering nature of a grain model in a self-
Fig.6.— Limits on the masses of planets that could have been
detectedat5σ confidenceinourL′ dataset(solidblackline). The consistent fashion assuming a specific grain size distri-
dashed lines correspond to the contrast (in mags) 100 Myr old bution (Augereau & Beust 2006; Wolf & Voshchinnikov
planets would have in this system from Baraffeetal. (2003). We 2004):
ruleoutplanetsmoremassivethan8MJ atprojectedseparations
≥0′.′5,andplanetsmoremassivethan∼6MJ beyond1′.′25. amaxn(a)C (a)g(a)da
<g >= amin sca , (1)
R amaxn(a)C (a)da
The high S/N images of HD 32297 debris disk at mul- amin sca
tiple wavelengthsprovideaunique windowinto the dust R
where n is the density of the dust in the disk andC
sca
grain properties within the disk. Under the assumption
is the scattering cross-section. The variable < g > can
that a single population of grains can explain allthe ob-
create an approximate phase function as a function of
servations, we set out to test the recent models of the
scattering angle Φ(θ) for the dust under the assumption
HD 32297 disk from D13, which are constrained by ob-
of a Henyey-Greenstein (HG) functional form:
servationsofthediskatKsband(Boccaletti et al.2012)
along with detailed FIR SED modeling. The primary 1 (1 <g >2)
structure of their modeled debris disk is that of at least Φ(θ)= − . (2)
4π(1+<g >2 2cosθ)1.5
onecomponentwithasharpedgeat110AUandadrop- −
offinsurfacedensitywithincreasingradius. Aninterior, The code can include linear combinations of < g >
8 Rodigas et al.
x 10−3 8x 10−3 x 10−3
9 6
7
8
−2SB / F (arcseconds)diskstar34567 −2SB / F (arcseconds)diskstar23456 −2SB / F (arcseconds)diskstar2345
12 59 AU 1 68 AU 1 76 AU
0 1 1.5 2waveleng2th.5 (µm) 3 3.5 4 0 1 1.5 2waveleng2th.5 (µm) 3 3.5 4 0 1 1.5 2waveleng2th.5 (µm) 3 3.5 4
(a) (b) (c)
x 10−3 5x 10−3
5 4.5
4.5 4
−2SB / F (arcseconds)diskstar123...234555 −2SB / F (arcseconds)diskstar123...23555
1 85 AU 1 93 AU
0.5 0.5
0 1 1.5 2waveleng2th.5 (µm) 3 3.5 4 0 1 1.5 2waveleng2th.5 (µm) 3 3.5 4
(d) (e)
Fig.7.— HD 32297 debris disk spectrum from 59-93 AU, averaged at each distance between the northeast and southwest lobes (black
stars),alongwiththeD13modelspectrum(solidpurplelines),themodifiedD13model(dashedpurplelines),andthepurewatericemodel
(solidblue lines). The horizontal black lines denote the filter widths at each wavelength. The originalD13 model agrees poorlywiththe
disk’sspectrum,withareducedchi-squareof17.9. ThemodifiedD13modelisabettermatch,withareducedchi-squareof2.87. Thebest
matchtothedataisthepurewatericemodel,withareducedchi-squareof1.06.
parameters, which might be appropriate for debris constraints:
disks and can reproduce the phase function that has
b(Ceeunrrifeouentdal.fo2r01t2h;eHzoondgia1c9a8l5d).usTtoicnontvheertsoalnarobssyesrtveemd n(r)=n √2 r −2αout + r −2αin −1/2,
0
disk SB into a mass, one must solve for the combina- (cid:18)(cid:16)110AU(cid:17) (cid:16)110AU(cid:17) (cid:19)
tion of both the phase function of the dust and the size- (4)
averagedcross section of the dust <C >: wheren isthemidplanenumberdensityatthereference
sca 0
distance of 110 AU, and α = 5 and α = 2. We
out in
amax also kept their assumption of <−g > = 0.5, which is
<C >= n(a)C (a)da. (3)
sca sca somewhatdegeneratewiththechoiceofaninteriorsteep
Z
amin power law drop-off in density interior to 110 AU. It is
We calculated scattering cross-sections < C > and alsonotconsistentwithMie calculationsofthe expected
sca
<g >usingtherealandimaginarypartsofthe complex <g >, which is closer to 0.99.
indicesofrefractionforthebest-fittinggrainmodel(pro-
4.2. Comparison to original D13 cometary grains model
vided kindly by J. Donaldson, private communication)
from the code miex, which has been designed specifi- For the D13 model to be accepted as a good match to
cally for fast modeling of debris disks with a size distri- the data, it must reproduce the disk’s SB at all wave-
bution of dust (Wolf & Voshchinnikov 2004; Ertel et al. lengths and distances from the star. This can be mea-
2011). Thegrainmodelusedhere(andbyD13)isa1:2:3 sured by calculating the reduced chi-square. We accom-
mixture of 90% porous silicates, carbon, and water ice plishthisbysummingupthe squareddifferencebetween
grains2.1-1000µminsize;suchcompositionsmayalsobe the model disk SB and the real disk SB at all wave-
appropriate for other debris disks (e.g., Lebreton et al. lengths and at all disk locations, divided by the number
2012). Models were generated for each image (wave- of degreesof freedom and the uncertainties squared. We
length)withtheappropriatepixelscaleandsampledina measured this value to be 17.9, indicating a poor fit. To
similarfashiontoourSBprofiles(Section3.2). Ascaling illustrate,Fig. 7showsthespectrumofthediskat59-93
factorforeachlobeofthediskwascalculatedbyratioing AU (black stars), along with the D13 cometary grains
the models with the observeddisk SB profiles as a func- model (solid purple lines). The SB values for the north-
tion of wavelength. For the disk’s density distribution, east and southwest lobes of the disk have been averaged
we used the best-fitting distribution in Boccaletti et al. at each wavelength because the D13 model is axisym-
(2012), based on analysis of their Ks band disk data, metric.
because D13 also used this model for their geometrical Ingeneral,thismodeloverpredictsthedisk’sSBandis
9
thereforeapooroverallfit. However,at 85-93AU(Fig. were hard spheres 3 µm in size. This model is the best
∼
7d and Fig. 7e), the model does a much better job of fit to all of the available NIR data (Fig. 7; reduced chi-
fitting the data. This makes sense since Boccaletti et al. square = 1.09).
(2012) computed the model disk from images in which If we ignore the requirement that the models must
the disk was only detected beyond 0′.′7 (80 AU). More match the disk’s SB distribution, allowing the model
importantly, the discrepancy between the model fit at spectrato be arbitrarilyscaledto matchthe disk’s spec-
small and large separations demonstrates the need for trum, we can determine which model is the best spectral
resolved imaging with small inner working angles (such fit to the data. To increase the S/N of the data, we nor-
as our L′ image). malized the disk spectra from 59-93 AU at 1.1 µm and
computed the median value at each wavelength. These
4.3. Comparison to modified D13 cometary grains model values are shown in Fig. 8 (black stars). We then scaled
the original D13 model spectrum until the residuals be-
Seeking to obtain an alternate model fit to the data,
tweenthedataandmodelwereminimized. Thisisshown
we modified the density distribution of the original D13
as the solid purple line in Fig. 8. After optimizing the
cometary grains model. This is necessary because the
spectralfit,the D13modelisblueandunderpredictsthe
disk’sactualSB at1-2µmisshallower(lowerSB power-
disk’sSBat3.8µm,whiletherealdisk’sspectrumisgray
laws) than the model disk’s SB. Therefore we modified
withinthe uncertainties. We repeatedthis procedurefor
the model’s dust distribution to better reproduce the
the pure water ice model. The chi-square value for the
data. Specifically we tested different values of α and
out water ice model is 3 lower than the D13 model, in-
αin, picking αout =−2 and αin =5 based on chi-square dicating this model∼is ×still preferred over the cometary
minimization. This modification resulted in a much im-
grains model based on the available data.
proved reduced chi-square value of 2.87. This model is
shown in Fig. 7 as the dashed purple lines.
5. DISCUSSION
4.4. Comparison to pure water ice model 5.1. Disk Structure
Fig. 2 and Fig. 3 allow us to characterize the struc-
ture of HD 32297’s debris disk. The SB asymmetry be-
tween the two sides of the disk from 0′.′5-0′.′8 (56-90AU)
1.1 at L′ (and Ks band from Currie et al. (2012)) suggests
thatthenortheasternlobeofthediskhasadeficitofdust
1
grainshere. Furthermore,byinspectionoftheSBprofile
power-law indices in Table 1, we can see that interior to
Fstar0.9 0′.′5 (56 AU), the disk’s SB rises very sharply, whereas
SB / disk0.8 oisuetxsiadcetltyhtishelobceahtiaovniotrhoenSeBwoinuclrdeaexsepsecmtotroesseleowfolry.aTdhisiks
ed 0.7 with a second interior disk component located close to
z
ali the star. A single-component disk would have a SB pro-
m 0.6
or file that either steadily increased close to the star with
n
0.5 the same power-law index as further from the star, or a
combined
SB profile that flattened out close to the star (e.g., HD
0.4 15115 as in Rodigas et al. 2012). The steep rise in SB
close the star was also seen at Ks band (Currie et al.
0.3
1 1.5 2 2.5 3 3.5 4 2012), corroborating this feature. The inner component
wavelength (µm)
could be located at . 50 AU but needs to be confirmed
Fig.8.— Combined spectrum of the disk from 59-93 AU, ob- with additional imaging that is sensitive to scattered
tainedbynormalizingthespectrafromFig. 7at1.1µmandthen light close to the star.
takingthemedianvalueateachwavelength(blackstars). Thehor- Thisisnotthefirsttimeasecondinnerdiskcomponent
izontalblacklinesdenotethefilterwidthsateachwavelength. The
has been suggested for the HD 32297 debris disk. D13
solidpurplelinecorrespondstotheD13modelspectrum,adjusted
such that the residuals between the data and the model are at a favoredawarminnercomponentfortheirmodeling,and
minimum. The solid blue line corresponds to the pure water ice Fitzgerald et al. (2007) used 11 µm images to suggest
model spectrum, adjusted in the same fashion. After scaling the a population of dust grains that peaks near 60 AU.
D13 model spectrum, the pure water ice model has a chi-square ∼
Such thermally-emitting grains may also be contribut-
value∼3× lower,indicatingabetter spectral fittothe data, but
additional data at 3.1 µm is required to help distinguish between ing to the scattered light that has been detected at 2-4
cometsandpurewater ice. µm. Currie et al. (2012) also preferred multiple belts,
includinga componentat 45AU,to explainthe disk’s
We alsotested a pure waterice model, since the inclu- observed SB and SED. O∼ur L′ data appear to support
sion of water ice significantly improved the model fits to the notion of a second interior belt, but additional data
the FIR SED of the disk (D13). Simpler compositions withverysmallinnerworkingangleswouldhelpvalidate
likeastronomicalsilicatesarenotmodeledbecausethese this explanation.
producedpoorfitstotheFIRSEDdata(D13),andmore The midplane offset profiles show that the disk is
complexmodelsliketholins(Debes et al.2008a)resulted bowed close to the star, agreeing with similar find-
in poor fits to the data . The pure water ice model disk ings at Ks band (Currie et al. 2012; Boccaletti et al.
had the same density distribution as the modified D13 2012; Esposito et al. 2013). We do not explicitly
model (α = 2 and α = 5), except the dust grains modelthisphenomenonbecausewehavealreadyshowed
out in
−
10 Rodigas et al.
that it can be explained (for both HD 15115 and disk’s NIR spectrum, even if such a dust composition
HD 32297) by a highly-inclined ring-like disk consist- is not easily explained. This is why obtaining a disk
ing of forward-scattering grains (Rodigas et al. 2012; detection at 3.1 µm is crucial; it can help distinguish
Currie et al. 2012). betweenvarious dust models that mightfit the available
No high-mass (8 M ) planets at projected separa- NIR data.
J
tions & 56 AU currently reside in the disk based on
our L′ imaging. If such planets exist in this system, 5.3. Limitations
their projected separations must be very small, imply- Whileourpurewatericemodelisthebestmatchtothe
ing either small semimajor axes or an unlucky epoch scatteredlightdata,itmaynotbefortheFIRdata(used
of imaging. The null detection of high-mass planets in by D13). Our model must reproduce all of the available
thissystem,thoughdisappointing,shouldnotbesurpris- photometry at all stellocentric distances to be accepted
ing, since there is now copious evidence that such plan- as valid; therefore our scattered light modeling should
ets are rare at large separations from their stars (e.g., not be considered final. This is especially true because
Wahhaj et al. 2013). While lower-massplanets may still we currently lack data near 3.1 µm, where our preferred
reside in the HD 32297 system, detecting them will re- model has a very deep absorption feature. We stress
quire more advanced imaging capabilities than are cur- that the goal of this study was to test the originally-
rently available. This is further complicated by the like- proposedD13cometarygrainsmodel;ifthatfailedtofit
lihood of such planets residing in the midplane of the the data, the goal was to find a reasonable alternative.
disk,whichifbrightenough,canhideplanets. Residuals Pure water ice is one such alternative composition, and
from PSF subtraction (especially in ADI datasets) can it makes predictions that are testable with future data,
themselves also resemble point sources, and since they which can then help refine the original cometary grains
can be found anywhere in an image (Fig. 1b), includ- model.
ing in the midplane of the disk, distinguishing between We also note that producing scattered light models of
real and artificial planets is difficult for systems like HD debris disks comes with several problems. The scatter-
32297. Future imaging searches for planets around this ing asymmetry parameter, < g >, is not self-consistent
starmustadequatelytakeintoaccounttheeffectsofboth when computed using the Mie formalism. Additionally
the disk itself and the PSF residuals. the commonly assumed single component HG formalism
maynotbeappropriateforthisdisk(Currie et al.2012).
5.2. Dust Grain Composition
Thiswouldimmediatelymakemodelingthedustcompo-
D13 found that comet-like porous grains consisting of sitionmorecomplicatedandbeyondthescopeofthisini-
silicates, carbon, and water ice were the best match to tial effort. Finally the increasing evidence that the disk
the HD 32297 disk SED at wavelengths longer than 25 may have an inner component at . 50 AU further com-
µm. Based on our analysis of the disk’s 1-4 µm SB and plicates the modeling process, since we would then have
spectrum, we cannot lend further evidence to support to consider the possibility of two separate dust popula-
this claim, at least with respect to their original model. tionswithtwodifferentdustcompositions. Atthis time,
The overall agreement between the data and the D13 it is unclear what the underlying distribution of dust is
model is poor (reduced chi-square = 17.9). around HD 32297. Lacking this knowledge, assuming a
This does not necessarily exclude comet-like materials single disk component as we have in this study is a rea-
from being present in the disk. We were able to achieve sonable starting point.
a much better fit simply by altering the dust density
distribution in the original cometary grains model. It 6. SUMMARY
may be that additional tweaking of model parameters We have presented an imaging detection of the HD
will yield an even better fit. Furthermore D13 reported 32297debrisdiskatL′. ThediskisdetectedathighS/N
only one “comet” combination; other combinations with from 0′.′3-1′.′1 (30-120 AU). Based on our L′ imaging,
∼
differingratiosofcarbon/silicates/watericeandperhaps we show that the system does not contain any planets
the dust’s porosity might better match the disk’s spec- more massive than 8 M beyond 0′.′5.
J
trum at λ > 1 µm. Because the spectral coverage is The disk at L′ i∼s bowed, as was seen at Ks band.
currently sparse from 1-4 µm, complex cometary grain This likely indicates that the disk is inclined by a
modeling is beyond the scope of this paper. few degrees from edge-on and contains highly forward-
The best fit to the data is achieved by a pure wa- scatteringgrains(Rodigas et al.2012). TheSB atL′ in-
ter ice model, likely due to the fact that the model is teriorto50AUrisessharply,aswasalsoseenatKsband
gray from 1-4 µm, like the disk. This model has a very (Currie et al. 2012). This evidence together suggests
deep absorption feature near 3.1 µm, evident in Fig. 8. that the disk may contain a second inner component lo-
Thefeatureisalsoevidentinthecometarygrainsmodel, cated at < 50 AU.
though it is much shallower. Therefore a key additional Comparing the disk’s color at 1-4 µm over the outer
test of the disk’s dust grain composition would be ob- portion of the disk (> 50 AU) with the recently pro-
taining very high S/N photometry of the disk using a posedcometarydustgrainmodelofD13 showsthat this
narrowband filter centered around 3.1 µm. This would model is a poor overall fit to the disk’s SB distribution
help determine if the disk’s dust contains any water ice and spectrum (reduced chi-square = 17.9). A modified
(e.g., Honda et al. 2009), which could then be used to version of this model produces a much better overall fit
refine the D13 cometary grains model. to all the data (reduced chi-square = 2.87). The best
Other dust grain compositions not tested in this work fit to the data is achieved with a pure water ice model,
mayalsobettersupporttheavailabledata. Forexample, though this model is not the only possible dust compo-
aperfectlyflat/graymodelspectrummightbetterfitthe sition. Additional imaging of the disk near 3.1 µm can