Download Infrared Views of the TW Hya Disk

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Astronomical spectroscopy wikipedia , lookup

Cygnus X-1 wikipedia , lookup

Accretion disk wikipedia , lookup

Transcript
Infrared V iew s of the T W H ya D isk
A .J.W einberger1;2,E.E.Becklin1,G .Schneider3,E.I.C hiang4,P.J.Low rance1,M .
Silverstone3,B.Zuckerm an1,D .C .H ines3,and B.A .Sm ith5
arXiv:astro-ph/0110342v1 15 Oct 2001
A B ST R A C T
T he face-on disk around T W H ya is im aged in scattered light at 1.1 and
1.6 m using the coronagraph in the N ear Infrared C am era and M ultiO bject
Spectrom eteraboard theH ubbleSpaceTelescope.Stellarlightscattered from the
optically thick dustdisk isseen from 20{230 A U .T he surface brightnessdeclines
asa powerlaw ofr 2:6 0:1 between 45 and 150 A U .T hescattering pro leindicates
thatthedisk is ared,notgeom etrically at.T hedisk,w hile spatially unresolved
in therm al radiation at 12 and 18 m in observations from the W . M . K eck
O bservatory,show s am orphous and crystalline silicate em ission in its spectrum .
A disk w ith silicate grainsofa 1 m icron in size in itssurface layerscan explain
thecolorofthescattered lightand theshapeofthem id-infrared spectrum . M uch
largergrainsin the disk interiorare necessary to tthe m illim eter-wave spectral
energy distribution, and hence grain grow th from an original interstellar size
population m ay have occurred.
Subject headings: stars:circum stellar m atter,individual(T W H ya) { infrared radiation
1. Introduction
T W H ya (= C D {34 7151,H = 7.6 m ag) is a star w ith a large H em ission equivalent
w idth, visual variability and location at 23 galactic latitude, and it was thus labeled as
1
D ivision of A stronom y,
lin,low rance,ben@ astro.ucla.edu
U C LA , B ox
951562,
Los
A ngeles,
CA
90095-1562,
beck-
2
D epartm entofTerrestrialM agnetism ,C arnegieInstitution ofW ashington,5241 B road B ranch R d.N W ,
W ashington,D C 20015,weinberger@ dtm .ciw .edu
3
Steward O bservatory, U niversity of A rizona, 933 N . C herry A ve., Tucson, A Z 85721, gschneider,m urray,dhines@ as.arizona.edu
4
D epartm entofA stronom y,U niversity ofC alifornia atB erkeley,601 C am pbellH all,B erkeley,C A 94720,
echiang@ astron.berkeley.edu
5
Institute for A stronom y,U niversity ofH awaii,H onolulu,H I,brad@ m ahina.hawaii.edu
{2{
\peculiar" by H enize (1976). R ucinski & K rautter (1983) identi ed it as a spectral type
K 7Ve T Tauristarfarfrom any m olecularcloud (i.e.\isolated"). T W H ya displaysclassical
signsofbeing surrounded by an accretion disk,including variable H em ission (M uzerolle et
al.2000) and excess ux over photospheric in the ultraviolet and near-infrared. In addition
it has disk signatures in the form offar-infrared (R ucinski1985),subm illim eter continuum
(W eintraub et al.1989) and C O line em ission (Zuckerm an,Forveille,& K astner 1995). A ll
ofthese disk indicators are large for a star as old as T W H ya,w hich has an age of 8 M yr
determ ined from pre-m ain sequence evolutionary tracks (W ebb et al.1999). Zuckerm an et
al. inferred a face-on view ing geom etry forthe disk based on the sm allC O line w idths they
observed. R ecent im aging observations do indeed reveala circular disk. Scattered radiation
from the disk has been im aged at visible and near-infrared wavelengths (K rist et al. 2000;
Trilling et al.2001),and circum stellar dust em ission has been im aged at a wavelength of7
m m using the V LA (W ilner et al.2000).
T he H ipparcos m ission m easured the distance to T W H ya as 56 7 pc (W ichm ann et
al.1998).A n eponym ousassociation of 20 starsystem shasbeen identi ed from x-ray and
properm otion surveys w ithin 10 and 10 pc ofT W H ya (de la R eza etal.1989;W ebb etal.
1999;Sterzik etal.1999;Zuckerm an etal.2001).T he approxim ately coevalm em bers ofthe
T W H ya A ssociation are am ongst the products ofrecent star form ation closest to the Sun.
H owever,only three other m em bers have m easurable far-infrared excesses from IR A S:H R
4796A ,H D 98800,and H en3-600 (LIR /L?= 0.01,0.19,and 0.21,respectively). T he optical
depth around T W H ya (LIR /L?= 0.25) is the largest and corresponds to the reprocessing
expected for a optically thick disk (A dam s,Lada,& Shu 1987).
T hestateofdisksofinterm ediateage( 107 yr)hasnotbeen wellstudied.T hestructure
ofthe youngest disks (< 106 yr),hundreds to thousands ofA U in radius around em bedded
protostars have been copiously studied w ith m illim eter interferom etry (Beckw ith & Sargent
1996,e.g.) and visualand infrared im aging (Burrow s et al.1996;Padgett et al.1999,e.g.).
T here are only a few exam ples ofolderlate-type stars surrounded by disks w hich have been
spatially resolved in the near-infrared,m ost ofw hich are in binary system s (R oddier et al.
1996;K oresko 1998,e.g.). A tages& 107 yr,m ostknow n disks orbitearly type starsbecause
only in these system s do the disks scatter enough photons to be detectable (e.g. Pic
(Sm ith & Terrile 1984)). T W H ya,thanks to its proxim ity to the Sun and its age,is thus
an excellent laboratory forstudying a disk ofinterm ediate age around a low m assstar. T his
age range is thought to be criticalin the evolution ofthe planetesim als that form the cores
ofgas giant and terrestrialplanets (W eidenschilling & C uzzi1993).
{3{
2. O bservations and D ata A nalysis
2.1. N IC M O S
T W H ya wasobserved attwo di erentwavelength bandsw ith theN earInfrared C am era
and M ultiO bject Spectrom eter (N IC M O S) aboard the H ubble Space Telescope. T he star
was placed in the coronagraphic hole in cam era 2 ( 0.076 00 pixel1 , 1900 square eld of
view ). T he observationalstrategy was the sam e for both { a totalof1216 s ofintegration
taken in three non-destructive m ultiple-readout (M U LT IA C C U M ) sequences at each oftwo
telescopeorientations,rolled 7 w ith respectto each otheraboutthetargetaxis.To m inim ize
tim e-dependent pointspread function (PSF) variations,the rolled im age sets were obtained
w ithin 40 m inutes in a single targetvisibility period. T he rstobservation wasm ade w ith
the F160W lter( central= 1.59 m ,FW H M = 0.40 m )on 1998 A ugust16 and the second was
m ade w ith the F110W lter ( central= 1.10 m ,FW H M = 0.59 m ) on 1998 N ovem ber 25. A
sum m ary ofthe observations is presented in Table 1.
Prior to each set of coronagraphic observations, two 0.86 s target acquisition fram es
were obtained w ith the F171M ( central= 1.72 m , FW H M = 0.07 m ) lter. C ontem poraneous lam p ats and backgrounds were obtained at F160W for the purpose oflocating the
coronagraphic hole and enabling good at- elding near the hole. Specialcoronagraphic at
elds ateach band were created by m odifying a standard high signalto noise ratio on-orbit
reference at eld. In a 16 16 pixelregion around and including the coronagraphic hole,
the pixelsensitivities were m easured in the F160W internal at eld (hole- nding) im age,
scaled to F110W ifnecessary,and replaced in the reference at.
Table 1. Log ofN IC M O S O bservations in Program s 7226 and 7233
Filter
T W H ya visit
F160W
F160W
F110W
F110W
7
8
91
92
a
D ate
T im e (U T )
PSF Star
V isit
1998 A ug 16
1998 A ug 16
1998 N ov 25
1998 N ov 25
4:05{4:19
4:35{4:48
20:15{20:27
20:42{20:55
H R 8721
H R 8721
1
Eri
1
Eri
81a
81a
80
81
D ate
1998 A ug 11
1998 A ug 11
1998 N ov 14
1998 N ov 14
H R 8721 drifts behind the coronagraph in this visit due to the loss oflock on one guide
star. W e used only the rst 240 s ofintegration in w hich the drift was m inim al. In this
tim e,the SN R ofH R 8721 is 1.5 tim es that ofT W H ya.
{4{
T he M U LT IA C C U M data sets were processed w ith \nicred" (M acLeod 1997). D ark
fram esprepared by the N IC M O S instrum entde nition team were used to subtractthe dark
currentand correctthe detectorshading. A fterbiassubtraction,linearization and at elding,know n bad pixelswere replaced w ith a G aussian weighted average ofneighboring pixels,
and the three M U LT IA C C U M im agesfrom each spacecraftorientation were m edianed. T he
bestavailable photom etric calibration wasapplied to the nalim ages(F110W :1 A D U s 1 =
2.031 10 6 Jy and,referenced to Vega,0 m ag = 1775 Jy;F160W :1 A D U s 1 = 2.190 10 6
Jy and 0 m ag = 1083 Jy). T he uncertainty in this absolute photom etric calibration is 2%
(M .R ieke,private com m unication).
In the reduced and calibrated N IC M O S im ages,scattered and di racted light from the
star can be m uch higher than the disk ux, so the instrum entalPSF from an unresolved
source m ust be subtracted. T he sam e observing strategy,including the telescope roll,was
used in our G T O program for 14 other stars at F110W and 77 at F160W .T hese were all
processed in the sam e way as the T W H ya im ages,and we use targets w hich do not show
extended disk em ission as a library of possible PSF stars. T he detailed character of the
N IC M O S PSF changes w ith tim e prim arily due to therm alvariations in the H ST optical
assem bly (K ulkarnietal.2000)and sm allshiftsofthe cam era cold m ask (K ristetal.1998);
hence som e PSF star subtractions produce m uch lower residuals than others (Schneider et
al. 2001). T he software ID P3 (Lytle et al.1999) utilizing cubic convolution interpolation
was used to register coronagraphic PSFs to both orientations ofthe T W H ya observations.
O nly stars brighter than T W H ya were considered,so the signalto noise ratio in the PSFs
were greater than that ofT W H ya at every radius,and the PSF subtraction residuals were
dom inated by system atic uncertainties. To m inim ize the e ect ofthe cold m ask shift,we
considered PSFs taken w ithin one m onth ofT W H ya. Five F160W PSFs and one F110W
PSF m et allcriteria.
T he ne structure ofthe broad band PSF isa ected by the spectralenergy distribution
ofthe source. W hen subtracting a reference point source,therefore,m ism atches in e ective
tem perature can result in increases in the subtraction residuals. A t F160W ,w here several
possible PSF observations existed, the chosen PSF is the K 4V star H R 8721 (= G L 879)
w hich is quite close to T W H ya (K 7) in spectraltype. A t F110W ,however, w here m any
fewer PSFs were taken,the spectraltype ofthe chosen PSF, 1 Eridani,is F4V .W e tested
thatthee ectofthecolorm ism atch issm allin two ways.First,wesubtracted coronagraphic
PSFs ofdi erent colors and exam ined the residuals;two exam ples are plotted in Figure 1.
A lthough color m ism atch does increase the residuals, it does not introduce a false excess
that could m im ic a disk,nor does it signi cantly a ect the shape ofthe observed T W H ya
disk. Second,we created synthetic PSFs ofdi erent spectraltypes w ith T iny T im (K rist
& H ook 1997). Since T iny T im m odels di raction and not scattering from the coronagraph
{5{
or other elem ents in the opticalpath,its PSFs isolate the e ects ofcolor alone. T he T iny
T im subtraction residualswere signi cantonly w ithin 0.600 ofthe starand were sm allerthan
those from realPSF subtractions.
T he dom inant source ofphotom etric uncertainty in the PSF-subtracted im ages results
from uncertainty in how to appropriately scale the PSF to the brightnessofT W H ya. T here
are two possible m ethods to determ ine the scaling factor: a) the use ofa prioriphotom etry
ofthe two stars and b) the use ofm easurem ents m ade on the subtracted im age,including
nulling the ux in the di raction spikes and dem anding thatthe ux approach zero atlarge
radiifrom the star. T he form er su ers from the fact that no im ages ofthe stars are taken
in the broad-band lters used forthe coronagraphic im aging,because the stars saturate the
detector in the m inim um integration tim e,so stellar m odels m ust be em ployed forthe color
transform ations from the acquisition lters. T he latter su ers from the fact that the disk
contam inates the m easurem ents.
O ur PSF scalings were determ ined by m inim izing the ux in the subtracted di raction
spikes outside the region ofapparent disk ux. W e scale the PSF im ages in A D U s 1 for 1
Eri(F110W )by 0.0099 and G l879 (F160W )by 0.031.T he uncertainty in thisscaling is3%
based on the variation between the four di raction spikes. T his uncertainty in the scaling
results in approxim ately a 10% uncertainty in the disk photom etry in each band.
2.2. K eck
Im ages and spectra ofT W H ya were taken w ith the facility Long W avelength Spectrograph (LW S) (Jones & Puetter 1993) on the 10 m K eck I telescope on U T 11 D ecem ber
2000 and 4{5 February 2001.D uring allthree nights,the weatherwasphotom etric w ith low
watervaporopticaldepth.T he im ageswere lim ited by seeing w ith FW H M = 0.0057 and 0.0050
at 11.7 m and 17.9 m respectively. LW S uses a 128 128 pixelBoeing Si:A s detector,
and hasa plate scale of0.08 arcsec pixel1 ,resulting in a focal-plane eld ofview of10:2400
square. It also provides a dispersion of 0.037 m pixel 1 ; w hen com bined w ith a 6 pixel
(0.0048) slit,the resolution is R
120. W hen viewed through the long slit,vignetting lim its
00
the eld ofview to 8.5 . O bservationaldetails are found in Table 2.
In spectroscopic m ode,data were taken w ith the telescope secondary chopped in the
sam e direction asthe nod w ith a frequency of5 H z and am plitude of1000.O nly theon-source
chop fellon the detector. In prelim inary data analysis,the im ages were double di erenced
to subtract the therm albackground and bad pixels were corrected by interpolation. T he
infrared bright star H R 4532 (11.7 m ux density = 53.6 Jy) was observed just after T W
{6{
H ya to provide atm ospheric calibration.
In im aging m ode,data were obtained by chopping the secondary at 5 H z and nodding
after 20 s.In som e cases,the data were taken by chopping and nodding 5",so fourim ages
appearon thedetector;in othersthe throw was10" so oneim age appearson thedetector. In
eithercase,every double di erenced (i.e.fully background subtracted)im age wascentroided
and recentered to rem ove telescope drift.
T he spectra of T W H ya and of the atm ospheric calibrator were extracted by tting
G aussians in the spatialdirection to the ux atevery spectralpixel. T he uncertainty in the
ux at every wavelength was estim ated from the noise in the sky on either spatialside of
the long slit im age. T he location of the Telluric O 3 line at 9.6 m was m easured in the
sky spectra associated w ith both T W H ya and the standard,from w hich it was determ ined
that sm all unrepeatabilities in the grating locator m echanism introduced a sm all shift in
the centralwavelength between the two. Before division ofthe T W H ya spectrum by the
standard spectrum ,the standard spectrum wasshifted in wavelength by 2.5 pixelsto m atch
that ofT W H ya. A fter division,the resultant spectrum was m ultiplied by a blackbody at
the tem perature ofthe standard star (3100 K ) in order to recover the intrinsic ux distribution. T he location ofthe Telluric ozone line was m easured from a cross correlation w ith
atm ospheric data from the N ationalSolar O bservatory sunspot atlas (W allace,Livingston
& Bernath 1994) and used for wavelength calibration.
Table 2. Log ofM id-Infrared O bservations
M ode
Im aging
O bject1
T W H ya
T W H ya
PSF
T W H ya
PSF
Spectroscopy T W H ya
A tm os. C al.
1
Filter center
(m)
Filter w idth
(m)
D ate
Integration
T im e (s)
A irm ass
11.7
11.7
11.7
17.9
17.9
10.5
10.5
1.0
1.0
1.0
2.0
2.0
4.9
4.9
11 D ec 2000
04 Feb 2001
04 Feb 2001
05 Feb 2001
04 Feb 2001
11 D ec 2000
11 D ec 2000
120
216
216
809
216
408
96
1.7
1.7
1.6
1.8
1.8
1.7
1.5
H R 4532 was used as the PSF and atm ospheric calibrator. T he bright star H R 3748 (11.7 m
ux density = 53.6 Jy) was observed for photom etric calibration.
{7{
3. R esults
3.1. R e ected Light
T he N IC M O S PSF subtracted im ages of T W H ya at F110W (1.1 m ) and F160W
(1.6 m ) are show n on a natural logarithm ic stretch in Figure 2. T he disk becom es visible just outside the coronagraph and continues out to a radius of400 ( 230 A U ).T he noise
com puted in the background of the im ages is 21.1 and 21.2 m ag arcsec 2 at F160W and
F110W respectively. In the disk, the noise is dom inated by subtraction residuals. T he
spatialresolution ofthese im ages is 0.0015 at F160W and 0.0012 at F110W .
T he di raction spikesfrom the stellarcoresare particularly sensitive to tim e-dependent
instrum ent and telescope variations and never subtract out perfectly. W here possible,the
pixels corrupted by the spikes in im ages from one telescope orientation were replaced by
uncorrupted pixelsfrom the otherorientation. T hisresultsin the tapered appearance ofthe
spikes in the nalim ages. T he spikes were not m asked in the central0.009, because even
though they are quite noisy,the very bright inner disk ofT W H ya can stillbe signi cantly
detected above them in the region from 0.0038 (just outside the edge ofthe coronagraphic
hole) to 0.009.
T he totaldisk ux densities were m easured in an annulusbetween radiiof0.0038 and 400
(22 228 A U ) to be 17.4 1.8 m Jy (12.5 0.1 m ag) at F110W and 21.6 2.2 m Jy (11.7
0.1 m ag) at F160W .A t both wavelengths,the pixels m asked by the di raction spikes were
replaced by theaveragevalueofthe ux density attheirradiiand included in thephotom etry
reported above. T he quoted uncertainties are not dom inated by photon counting statistics
but rather by the system atic uncertainty in how to scale the PSF stars to T W H ya. T he
m easured ratios of scattered to stellar light are 0.024 and 0.021 at F110W and F160W
respectively. T he surface brightness peaks at a radius of 0.005 (29 A U ) at 4.5 1.2 m Jy
arcsec 2 atF110W and 5.7 1.4 m Jy arcsec 2 atF160W and declines sm oothly w ith radius.
To m easure any ellipticity in the disk,the radiiofseven independentisophotalcontours
were calculated asa function ofazim uthalangle in the F160W im age.Fora m ore signi cant
result,the contours were norm alized to a radius ofone and averaged in sixteen azim uthal
bins. T he average isophote was t w ith an ellipse using eccentricity (i.e. inclination) and
position angle as two free param eters. T he resulting best t had e= 0.025 0.014. Ifthis
ellipticity isinterpreted asdueto theinclination ofan intrinsically circulardisk,them easured
inclination is ofm arginalsigni cance,but a robust 3 upper lim it to the inclination is 4
from face-on.
T he azim uthally averaged disk surface brightnesses at both wavelengths are show n in
{8{
Figure 3. Between radiiof40 A U and 150 A U ,the disk at both wavelengths is well t by
a single power law ,r 2:6 0:1 (the solid lines in Figure 3). H owever,also seen at F160W is a
\w iggle" in the surface brightness centered at 1.005 (85 A U ).Beyond 150 A U ,the disk falls
o m ore sharply. W ithin 40 A U ,it appears to atten.
To obtain thecolorofthedisk,theim agesatF110W and F160W wereratioed.Sincethe
im ages in the two lters were taken at quite di erent telescope orientations,the di raction
spikes corrupt50% ofthe ratioed im age. T here are no signi cant gradients in the colorasa
function ofradius. T he F110W -F160W coloraveraged over the pixels free from di raction
spike contam ination is 0.77 0.10 m ag.
SinceT W H ya wasneverobserved w ith theN IC M O S F110W and F160W ltersw ithout
the coronagraph,the color ofthe star at these two N IC M O S bands was estim ated based on
a stellar photosphere m odel by K urucz (1993). T he F171M m agnitude of the star was
m easured from the acquisition im ages to be 7.39
0.04 m ag from both the A ugust and
N ovem ber 1998 data. T he U BV R I variability ofT W H ya,w hich am ounts to 0.5 m ag at
V and decreases toward the red,has been exam ined in detailand show n to have an overall
period of1{2 days w ith m uch shorter episodes ofrapid change (R ucinski& K rautter 1983;
M ekkaden 1998;H erbst& K oret1988).T he starbecom esbluerby up to 20% asitbrightens.
T he near-infrared variability hasnotbeen thoroughly studied,butitslevelisexpected to be
low ifthe underlying cause ofthe variation is from hot-spots. T he high levelofconsistency
in the photom etry ofouracquisition im agesfrom the two datesim pliesthatthe stardid not
vary at 1.7 m between A ugust and N ovem ber. W e assum e the star was sim ilarly stable at
1.1 and 1.6 m .
A K urucz stellar atm osphere m odel w ith T eff = 3925 K ,log(g)= 4, and log(Z)= 0 was
used to extrapolate the F171M m agnitude to F110W and F160W giving 8.47 and 7.55 m ag
respectively. T he sam e m odelgives H = 7.53 and J= 8.35 m ag (in the C IT system ) w hich are
10% brighterthan thevaluesgiven by R ucinski& K rautter(1983),w hich had an uncertainty
of5{10% .T he inferred J-H color,0.82,isidenticalto thatobtained by R ucinski& K rautter
(1983),so we believe the m odelextrapolation to the N IC M O S lters to be accurate. T he
F110W F160W stellarcoloris thus 0.92 m ag,w hile the colorofthe disk is 0.77 0.10 m ag,
so the disk color appears to be m arginally (1.3 ) blue.
U sing H ST /W FPC 2,K rist et al. (2000),also found a slightly blue color for the disk
at visualwavelengths (0.6 and 0.8 m ) in im ages m ade w hen the star was as bright as it
haseverbeen recorded. T he bestestim ate ofcolorshould com e from the largestwavelength
range,i.e. 0.6{1.6 m . O ver the disk region between 0.5 and 3 00,F606W F160W = 2.6
0.2. For T W H ya,F606W F160W = 2.5,so the disk appears to scatter neutrally over this
1 m wavelength baseline.
{9{
3.2. T herm al E m ission
T he radialpro les ofT W H ya in therm alem ission at 11.7 and 17.9 m are show n in
Figure 4 com pared w ith im ages ofthe PSF star. T W H ya appears to be the sam e size as
the PSF,i.e. unresolved,at both wavelengths. Integrated ux densities were m easured in a
synthetic aperture ofradius 2.004,and were 0.72 0.04 Jy at 11.7 m and 1.45 0.08 Jy at
17.9 m . T he m easurem ent uncertainties re ect both statisticaland photom etric calibration
contributions. T hese ux densitiesagreevery wellw ith those given in the IR A S FaintSource
C atalog (1990)at12 and 25 m interpolated to ourwavelengths ofobservation. T hus,allof
the m id-infrared ux in the IR A S beam arises from quite close to the star. U sing a K urucz
m odelof a K 7 photosphere (see Figure 5) to predict the L-band ux and then assum ing
R ayleigh-Jeans fall-o from 3.5 to 18 m ,we determ ine the stellar ux density at 11.7 and
18.9 m as 31 and 13 m Jy,respectively. T he m id-infrared color tem perature ofthe disk is
214 K from our m easurem ents.
T he 8{13 m spectrum of T W H ya is show n in Figure 6. T he 11.7 m ux density,
above,was used to norm alize the spectrum ,since the fullw idth ofthat lter iscontained in
the spectralrange. T here is a broad hum p in this region that,sim ilarly to Sitko,R ussell,
& Lynch (2000),we attribute to em ission by a com bination ofam orphous and crystalline
silicates.
3.3. D etection Lim its for C om panions
M ultiplicity in the T W H ya association as a w hole is & 50% ,including stellar and substellar objects (Zuckerm an et al.2001),although T W H ya itselfhas no know n com panion.
A ttheyoung ageofT W H ya any substellarcom panionswould bequitebrightin theinfrared,
so our N IC M O S im ages provide a sensitive way to search for such objects.
W e subtracted the two orientations ofour F160W data. W hile the stellar PSF,the instrum ental scattering function, and detector artifacts rotate w ith the aperture, any real
features in the unocculted area of the detector are una ected by a change in the telescope/cam era orientation. Subtraction ofthe these two im ages has been show n to signi cantly reduce residualPSF background light(Schneideretal.1998).Furtherm ore,since the
T W H ya disk is alm ost face-on,it nulls out nearly perfectly in the roll-subtraction,leaving
higher noise but no excess ux. In our data,the rollwas only 7 ,so point sources in the
rollsubtraction are separated by m ore than their FW H M for radiigreater than 1.200. For
sm aller radii,the roll-subtraction was not usefulfor point source detection and we used the
PSF subtracted im ages discussed already. In both regions, we assessed the observability
{ 10 {
ofpoint sources by planting T iny T im PSFs and m easuring the m agnitude that could be
recovered at > 3 signi cance. N o point sources were found w ithin 3.4 00 ofT W H ya. T he
F160W detection lim its are show n in Figure 7.
4. D iscussion
4.1. D isk M orphology
G iven the large fraction (0.25) ofthe stellar lum inosity re-radiated in the m id and farinfrared,the disk m ust be optically thick. For single scattering by a geom etrically at and
optically thick disk,the ux density persquarearcsecwould beproportionalto r 3 (W hitney
& H artm ann 1992).T hisissom ew hatsteeperthan the r 2:6 we tin the 40{150 A U annular
region and suggests thatthe disk is instead ared. Perhaps the strongest argum ent in favor
ofdisk aring is that the re ected light surface brightness throughout this annulus is m uch
larger than predicted by a at disk m odel.
A warm optically thick disk com posed ofwellm ixed dustand gasw illnaturally are as
a consequence ofverticalhydrostatic equilibrium (K enyon & H artm ann 1987). Ifdust and
gas are wellm ixed and in interstellar proportions,then the height at w hich stellar photons
are scattered is roughly proportionalto the verticalpressure scale height. In this case,the
scattering height scales asr ,w here
1:2{1:4 (C hiang & G oldreich 1997;D ’A lessio et al.
1998;C hiang et al.2001). Such a well-m ixed, ared disk has a scattering brightness pro le
that scales approxim ately as r 2 (W hitney & H artm ann 1992),signi cantly shallower than
the average T W H ya pro le at disk radiigreater than 40 A U .T he discrepancy could re ect
vertical settling of dust w ith respect to the gas such that the actualscattering height at
these radiiis characterized by < 1.
T he F160W surface brightness pro le between 80 and 130 A U scales like r 2 . T his
annulus corresponds to the \zone 3" de ned by K rist et al. (2000) (Figure 8),w here the
visualscattered light also attens out in their im ages. T he appearance ofthe sam e feature
in data from two di erent instrum ents lends credibility to the idea that this feature is real.
Such a variation in the slope ofthe surface brightness pro le m ight re ect an undulation or
ripple in the height ofthe disk surface;at 40 A U . r . 80 A U ,the disk surface m ight be
concave dow n,w hile at 80 A U . r . 130 A U ,the disk m ight be concave up. D istortions
of the disk m ay be caused by instabilities a icting passively heated disks (C hiang 2000;
D ullem ond 2000) or by dynam icale ects created by a m assive body orbiting in the disk.
T he detection lim it on a m assive body at 100 A U from the star is 3M Jup. Beyond 150 A U ,
the surface brightness pro le is signi cantly steeper than r 3 . T he outer disk is probably
{ 11 {
com pletely shadowed.
T he face-on geom etry of the disk and the average N IC M O S data power law for the
disk surface brightness are consistent w ith the W FPC 2 data ofK rist et al. (2000). T heir
average power law in the sam e 40 { 150 A U region (see Figure 8) is -2.4 at F606W (0.6
m ) and -2.6 at F814W (0.8 m ). In addition,the geom etry ofthe disk seen in both H ST
data sets is consistent w ith the generalm orphology reported by Trilling et al.(2001) from
ground-based observations. H owever, there are notable di erences between the N IC M O S
and Trilling et al. data. A t their innerm ost detectable radius of0.0094,they report a peak
surface brightness at H -band of6.8 1.1 m Jy arcsec 2 . A t this sam e radius,w hich is well
outside our coronagraph,and using the F160W lter w hich is very sim ilar to H -band,we
m easure 2.5 0.4 m Jy arcsec 2 . In addition,they m easured a totaldisk ux density in the
0.0094{400 region of21.6 3.5 m Jy,w hereas we m easure 12.0 1.2 m Jy in thatsam e region.
T he Trilling et al. surface brightness power law index, -3.3 0.3, is discrepant by 2.5
w ith the N IC M O S index of-2.6 0.1. System atic uncertainties in their ground-based PSF
subtraction under conditions of0.008 seeing m ay be responsible for the discrepancies.
T helargeopticaldepth ofthedisk m akesitdi cultto usethescattered lightto estim ate
the density of grains. A t any given position in the disk, visible and near-infrared stellar
radiation penetratesto an opticaldepth ofone.In contrast,the disk ispresum ably optically
thin at m illim eter wavelengths, so m illim eter im ages of the dust em ission trace the true
surface density.
4.2. C om position
To elucidate the dust com position, we com bine inform ation from the spectralenergy
distribution,m id-infrared im aging,and m id-infrared spectroscopy.
4.2.1. SED Fitting com pared to M id-infrared Im aging
A sim plem odelofthedisk using grainsthatabsorb and em itlike blackbodiesin therm al
equilibrium w ith a star oflum inosity 0.3 L predicts a tem perature of214 K ,the 11.7 to
17.9 m color tem perature, at a distance ofonly 0.9 A U (0.0004). If the entire disk could
be described by such grains,it would therefore not be surprising that it appears point-like.
In m ore sophisticated m odels ofpassive circum stellar disks,C hiang & G oldreich (1997)and
C hiang et al.(2001) predict the m id-infrared ux em itted by a disk as a function ofradius.
In their two-layer disk m odels,the hotter disk surface layer is heated by direct exposure to
{ 12 {
starlight,w hile the cooler disk interior is heated by radiation from the surface. To com pare
predictions of their m odels w ith our m id-infrared im ages, we m odeled T W H ya after the
prescription of C hiang et al. (2001) w hich accounts explicitly for grain size distributions
and laboratory-based silicate and water ice opacities. O ur procedure was rst to select disk
param eters so as to reproduce the observed SED only. In tting the disk param eters, no
regard is given to any im aging data. O nly after a disk m odel is chosen that provides a
satisfactory t to the SED do we derive surface brightness pro les at 11.7 m and 17.9 m
based on our tted m odeldisk. T hese pro les are convolved w ith our respective PSFs and
com pared to our im ages ofT W H ya.
Figure 5 displays the observed and tted SED s ofT W H ya. M odelinput param eters
are given in Table 3.Foradditionaldetailsofthe m odel,see C hiang etal.(2001).T he SED
m odelpredicts that 99.8% ofthe 12 m ux density and 94.6% ofthe 18 m ux density
arise w ithin a radiusof9 A U (2 pixels)ofthe star.A sisshow n in Figure 4,atthe resolution
ofthe K eck Telescope,no extended em ission is predicted by the m odel. T his is consistent
w ith the observations in w hich no extension is observed.
O utside 0.2 (2)A U ,silicate coresare m antled by waterice in the disk interior(surface).
T he rem arkably large grains in the disk interior,consisting ofm illim eter{centim eter sized
cores, and the large total condensable m ass in the disk of 0.0014 M = 470 M , were
necessary to t the m illim eter-wave SED m easured by W eintraub et al.(1989) and W ilner
et al. (2000). T he m easurem ent at the particularly long wavelength of 7 m m constrains
the grain size distribution in the disk interior to be signi cantly atter than the standard
interstellar law of dN =dr / r 3:5 ,and to possess an upper size cut-o of 6 m m . H ad we
Table 3. Param eters ofT W H ya D isk M odel
Param eter
m axim um grain radius in surface
m axim um grain radius in interior
grain core size distribution (dN =dr)
in disk interior
in disk surface
dust surface m ass density
inner disk radius
outer disk radius
Value
1 m
6000 m
/ r1
/ r 3:5
10 (r/A U ) 1 g cm
0.05 A U
200 A U
2
{ 13 {
em ployed a grain size distribution that was m uch steeper than r 1 and that had an upper
size cut-o sm aller than r = 6 m m , the m odel uxes between 1:3 and 7 m m would have
been too low com pared to the m easurem ents. O ur tted grain size distributions in the disk
interiorare subjectto the well-know n degeneracy between grain size and disk surface density
(see,e.g.,C hiang et al.(2001));reducing the grain sizes or steepening the size distribution
increases the tted dust surface m ass density. H owever, increasing the tted dust surface
density would threaten to m ake the disk gravitationally unstable,ifgasand dustare present
in interstellarproportionsin T W H ya. W e draw the m odel-dependent conclusion thatgrain
grow th has indeed occurred in the T W H ya disk.
For the particular grain m ixture ofm illim eter to centim eter-sized olivine spheres m antled w ith water ice em ployed by our m odel, 1:1m m ;dust = 0:84cm 2 g 1 and 7m m ;dust =
0:14cm 2 g 1 . A som ew hat lower dust m ass of 130 M was derived by W ilner et al.(2000)
and Trilling et al.(2001) using the sam e observationaldata. T he di erence arises largely
because their assum ed dust opacities were 3 tim es greater than ours (we have m ultiplied
theiropacitiesby 100 to isolate the dustcom ponent). T heirvalueswere based on the H ildebrand (1983) m easurem ent of 250 m ;dust = 10 cm 2 g 1 in the dark m olecular cloud N G C
7023 and extrapolated to longerwavelengthsw ith ;dust / ,w here istaken to be 1.0 by
W ilner et al. and 0.9 by Trilling et al. T he assum ed sim ilarity between dust in N G C 7023
and dust in T W H ya has no direct observationalevidence.
Pollack etal.(1994)showed thata single powerlaw description ofthe opacity isinaccurate forrealdustm aterials.T hey com puted opacitiesfora variety ofgrain species,including
water ice,silicates,organics,iron,and troilite. Fortheir\C om posite (50% )" grains,extrapolated values of 7m m ;dust range between 0.03 and 0.06 cm 2 g 1 ,assum ing thatthe values of
thatthey com pute between 650 m and 2.3 m m can beextended to 7 m m .T hese opacities
are a factorof 3 sm allerthan the onesem ployed in ourdisk m odel t,and a factorof5{10
sm aller than the ones adopted by W ilner et al.(2000)and Trilling et al.(2001).
K rist et al. (2000) m odelonly 33 M ofISM -like dust to reproduce their observed
surface brightness pro les in scattered light at = 0.6 and 0.8 m . T he pro les depend
m ore strongly on the distribution ofdust at high altitude above the m idplane than on the
distribution ofdust in the m idplane w here m ost ofthe m ass probably resides. K rist et al.
assum e a G aussian vertical density pro le for m icron-sized dust. H owever, substantially
m ore dust could reside at the m idplane of their disk m odel w ithout detracting from the
goodness oftheir surface brightness ts. T hus,their 33 M could easily represent a lower
lim it on the actualdust m ass.
W e conclude that the T W H ya disk probably contains severalhundred Earth m asses
ofcondensed silicates and ices. T his dust m ass is severaltim es larger than the typicaldust
{ 14 {
m asses estim ated for T Tauristar disks in the Taurus and O phiuchus star form ing regions
on the basis of1.3 m m dust continuum observations (O sterloh & Beckw ith 1995;A ndre &
3:5
M ontm erle 1994). Large,centim eter sized,grains were necessary to t the
slope of
the m illim eter-wave SED .A grain size distribution w ith sm aller grainsis characterized by a
4 5
signi cantly steeper slope,
,and is di cult to reconcile w ith the observations. T he
presence ofthese very large particles is good evidence for grain grow th w ithin the disk.
Totalm ass estim ates for T W H ya’s disk in the literature vary by another two orders
ofm agnitude depending on the assum ed gas:dust ratio. T he standard interstellar gas:dust
m assratio of100:1 m ay be incorrectforan olderdisk like thataround T W H ya. K astneret
al.(1997) nd only 11 M in H 2 gasbased on 13C O observationsand a standard interstellar
H 2/C O ratio. T he low m ass accretion rate estim ated by M uzerolle et al.(2000),an order
ofm agnitude less than for typicalT Tauristars,also suggests that the gas in this system
hasbeen depleted. T he gas-to-dustm ass ratio doesnotform ally enter into ourm odelSED tted disk;only enough gas and/or verticalcirculation ofgas is assum ed to lift dust grains
3{5 vertical scale heights above the m idplane to com prise the ared, superheated disk
atm osphere. Ifwe assum e the prim ordialT W H ya disk had the canonical100:1 gas to dust
ratio,then it once contained 0.14 M . T his early disk w ith 25% ofthe star’s m ass would
have been m arginally gravitationally stable.
4.2.2. M id-infrared spectrum
T he broad silicate em ission feature cannot be explained by am orphous silicates alone
since they have a pronounced peak at9.6 m and fallo rapidly to longerwavelengths. O ur
spectrum agrees very wellw ith the published result ofSitko,R ussell,& Lynch (2000),w ho
note the near absence of a peak at 11.2 m , w ithin the uncertainties of their data. O ur
spectrum hints at a peak at 11.2 m ; furtherm ore,to create the fullw idth ofthe feature
requires substantial ux longward of 10 m . C rystalline species are probably responsible.
D ata ofthis spectralresolution are inadequate to constrain the exact crystalline structure,
however. M agnesium and iron-rich speciesallhave peaksin the 10 { 11.3 m region (Jageret
al.1998). Silicate em ission features becom e less prom inent as the grains get larger,so that
by a size of 5 m theirm id-infrared spectrum isnearly at (Skinner,Barlow & Justtanont
1992).T he T W H ya spectrum suggests that its em itting grains m ust therefore be . 5 m in
size.
{ 15 {
5. C onclusions
T W H ya is surrounded by a optically thick dust disk w hich m ust be ared to account
for the large apparent size of the disk seen in scattered light and for its therm alSED .A
ripple in the surface brightness at85 A U and the steep decline in surface brightness beyond
150 A U could be due to therm alinstabilities or dynam icale ects. H owever,no com panion
to T W H ya is detected dow n to a m ass of 10 M Jup at distances greater than 50 A U from
the star.
A lthough T W H ya lookslike a classicalT Tauristarin term sofitsH em ission,itsdisk
show ssignsofevolution.To tthelargesubm illim eterand m illim eter ux densitiesm easured
from T W H ya,ourSED m odelsrequirea grain sizedistribution in thedisk interiordom inated
by large,m illim eter-centim eter sized,icy grains. Large grains,w hich dom inate the m ass of
the disk,presum ably grew from coagulation ofsm allergrainsin the disk. T he initialm assof
T W H ya’sdisk in gasand dustm ay beseveraltim esthetypicalT Tauristardisk m ass,w hich
m ay help account for its long, 8 M yr, lifetim e. T he therm alem ission rem ains spatially
unresolved at the resolution ofthe K eck Telescope,in accord w ith theoreticalexpectations.
W e thank John K ristand M ike Sitko forproviding theirdata in digitalform ,ouranonym ous referee for insightfulcom m ents on the m anuscript, and the K eck sta for help w ith
LW S.T hiswork wassupported by N A SA grantN A G 5-3042 to the N IC M O S ID T and by a
H ubble Fellow ship grant to EIC .T his paper is based on observations w ith the N A SA /ESA
H ubbleSpaceTelescope,obtained attheSpaceTelescopeScienceInstitute,w hich isoperated
by the A ssociation ofU niversities for R esearch in A stronom y, Inc. under N A SA contract
N A S5-26555. T his paper is also based on observations at the W . M . K eck O bservatory,
w hich is operated as a scienti c partnership between the C alifornia Institute ofTechnology,
the U niversity ofC alifornia,and the N ationalA eronautics and Space A dm inistration and
was m ade possible by the generous nancialsupport ofthe W .M .K eck Foundation.
R E FE R E N C E S
A dam s,F.C .,Lada,C .J.,& Shu,F.H .1987,A pJ,312,788
A ndre,P.& M ontm erle,T .1994,A pJ,420,837
Beckw ith,S.V .W .& Sargent,A .I.1996,N ature,383,139
Burrow s,A .,M arley,M .,H ubbard,W .B.,Lunine,J.I.,G uillot,T .,Saum on,D .,Freedm an,
R .,Sudarsky,D .,& Sharp,C .1997,A pJ,491,856
{ 16 {
Burrow s,C .J.et al.1996,A pJ,473,437
C hiang,E.I.& G oldreich,P.1997,A pJ,490,368
C hiang,E.I.2000,Ph.D .T hesis,C alifornia Institute ofTechnology
C hiang,E.I.,Joung,M .K .,C reech-Eakm an,M .J.,Q i,C .,K essler,J.E.,Blake,G .A .,&
van D ishoeck,E.F.2001,A pJ,547,1077
D ’A lessio,P.,C anto,J.,C alvet,N .,& Lizano,S.1998,A pJ,500,411
de la R eza,R .,Torres,C .A .O .,Q uast,G .,C astilho,B.V .,& V ieira,G .L.1989,A pJ343,L61
D ullem ond,C .P.2000,A & A ,361,L17
H enize,K .G .1976,A pJS,30,491
H erbst,W .& K oret,D .L.1988,A J,96,1949
H ildebrand,R .H .1983,Q JR A S,24,267
Jager,C .,M olster,F.J.,D orschner,J.,H enning,T .,M utschke,H .,& W aters,L.B.F.M .
1998,A & A ,339,904
Jones,B.,& Puetter,R .1993,in Proc.SPIE vol.1946,610
K astner,J.H .,Zuckerm an,B.,W eintraub,D .A .,& Forveille,T .1997,Science,277,67
K enyon,S.J.& H artm ann,L.1987,A pJ,323,714
K oresko,C .D .1998,A pJ,507,145
K rist, J. E. & H ook, R ., N . 1997, in T he 1997 H ST C alibration W orkshop w ith a N ew
G eneration ofInstrum ents,p.192
K rist,J.E.,G olim bow ski,D .A .,Schroeder,G .,& H enry,T .J.1998,PA SP,110,1046
K rist,J.E.,Stapelfeldt,K .R .,M enard,F. Padgett,D .L.& Burrow s,C .J.2000,A pJ,538,
793
K ulkarni, V .P., H ill, J.M ., Schneider, G ., W eym ann, R .J., Storrie-Lom bardi, L.J., R ieke,
M .J.,T hom pson,R .I.,& Januzzi,B.2000,A pJ,536,36
K urucz,R .1993,C D -RO M N o.13,G SFC
{ 17 {
Lytle,D .,Stobie,E.,Ferro,A .,& Barg,I.1999,in A SP C onf.Ser.172,A stronom icalD ata
A nalysisSoftware and System s8,ed.D .M .M ehringer,R .L.Plante,& R .A .R oberts
(San Francisco: A SP),445
M acLeod,B.,A .1997,H ST C alibration W orkshop Proceedings,ed.S.C asertano
M ekkaden,M .V .1998,A & A ,340,135
IR A S Faint Source C atalog Ver.2.0.1990,ed.M oshir,M .et al.
M uzerolle,J.,C alvet,N .,Brice~
no,C .,H artm ann,L.,& H illenbrand,L.2000,A pJ,535,L47
O sterloh,M .& Beckw ith,S.V .W .1995,A pJ,439,288
Padgett,D .L.,Brandner,W .,Stapelfeldt,K .R .,Strom ,S.E.,Terebey,S.,& K oerner,D .
1999,A J,117,1490
Pollack,J.B.,H ollenbach,D .,Beckw ith,S.,Sim onelli,D .P.,R oush,T .,& Fong,W .1994,
A pJ,421,615
R oddier,C .,R oddier,F.,N orthcott,M .J.,G raves,J.E.,& Jim ,K .1996,A pJ,463,326
R ucinski,S.M .& K rautter,J.1983,A & A ,323,L49
R ucinski,S.M .1985,A J,90,2321
Schneider, G .,T hom pson,R .I.,Sm ith,B.A .and Terrile,R .J,1998,Space Telescopes and
Instrum entation V ., at A stronom ical Telescopes and Instrum entation, (SPIE Proceedings Series),3356,222.
Schneider,G .,Becklin,E.E.,Sm ith,B.A .,W einberger,A .J.,Silverstone,M .& H ines,D .
C .2001,A J,121,525
Skinner,C .J.,Barlow ,M .J.& Justtanont,K .1992,M N R A S,255,31P
Sm ith,B.A .and Terrile,R .J.,1984,Science,226,1421
Sitko,M .,R ussell,& Lynch.2000,A J,120,2609
Sterzik,M .F.,A lcala,J.M .,C ovino,E.,& Petr,M .G .1999,A & A ,346,L41
Trilling,D .E.,K oerner,D .W .,Barnes,J.W .,Ftaclas,C .,& Brow n,R .H .2001,A pJ,552,
L151
{ 18 {
W allace,L.,Livingston W .,& Bernath.,P.1994,N .S.O .TechnicalR eport # 1994-01 (available at ftp://argo.tuc.noao.edu/pub/atlas/spot2atl/)
W ebb,R .A .,Zuckerm an,B.,Platais,I.,Patience,J.,W hite,R .J.,Schwartz,M .J.,M cC arthy,C .1999,A pJ,512,L63
W eidenschilling,S.J.& C uzzi,J.N .1983,Protostars and Planets III,ed.,E.H .Levy & J.
I.Lunine (U .A Z Press),1031
W eintraub D .A .,SandellG .,& D uncan W .D .1989,A pJ,340,69
W hitney,B.A .& H artm ann,L.1992,A pJ,395,529
W ichm ann,R .,Bastian,U .,K rautter,J.,Jankovics,I.,& R ucinski,S.M .1998,M N R A S301,
39
W ilner,D .J.,H o,P.T .P.,K astner,J.H .& R odriguez,L.F.2000,A pJ,534,L101
Zuckerm an,B.,Forveille,T .,& K astner,J.H .1995,N ature,373,494
Zuckerm an,B.,W ebb,R .A .,Schwartz,M .,& Becklin,E.E.2001,A pJ,549,L233
T his preprint was prepared w ith the A A S LATEX m acros v5.0.
{ 19 {
Fig. 1.| (a) C om parison ofT W H ya w ith PSF nullat F160W .T he spectraltypes ofT W
H ya (K 7) and its PSF (G L 879;K 5) are quite close. To illustrate the e ect ofsubracting a
PSF ofa di erent spectraltype,the nullisthe resultofsubtracting an F3-type PSF from a
K 7-type PSF.(b) C om parison ofT W H ya and a PSF N ullat F110W .T he spectraltype of
the PSF, 1 Eriis F5,so color residuals should not be greater than the leveldem onstrated
in the PSF nullofparta.T he PSF nullin this case isof 1 Erim inus an A 1-type PSF.For
allthese subtractions,com parions w ith T iny T im show thatthe residuals are dom inated by
tim e-variable e ects and not by color m ism atch.
.
{ 20 {
Fig. 2.| Left: PSF subtracted, roll-com bined im age of T W H ya at F110W ;Right: PSF
subtracted,roll-com bined im age ofT W H ya atF160W ,both show n w ith a naturallogarithm ic stretch and on the sam e spatialscale and orientation.T hisstretch isem ployed since the
disk falls o as a power law and so it highlights the outer disk. T he size (radius 0.003) and
location ofthe coronagraph are show n w ith the centralgray circle. T he di raction spikes
have been m asked in regions w here their noise is greater than the disk signal. T hey have a
tapered shape because pixels obscured atone telescope orientation were replaced w ith those
not obscured at the other angle.
{ 21 {
Fig. 3.| Surface brightness ofT W H ya at both 1.1 m and 1.6 m as a function ofradius.
T he black lines show a power law ofr 2:6 t between 40 and 150 A U and extrapolated to
other radii. T he error bars on each point represent the standard deviation ofallthe pixels
at the given radius in an annulus one pixelw ide.
{ 22 {
Fig. 4.| In the left panels,the azim uthalpro les ofour K eck point spread function star
11.7 (top row )and 17.9 m (bottom row )are com pared to pro lesofourim agesofT W H ya.
In the rightpanels,the m odeldescribed in x4.2.1 convolved w ith ourPSF iscom pared to the
PSF.A t each radius,the error bars show the standard deviation about the m ean value of
pixels in a one pixelw ide annulus. T he PSF is evidently oversam pled at these wavelengths,
and the rst A iry m inim um can just be seen in the 17.9 m pro le at 0.0048. T W H ya is
unresolved (left). T his result is consistent w ith the m odel(right).
{ 23 {
Fig. 5.| M easured spectralenergy distribution for T W H ya com pared to that ofa K urucz
m odelstellarphotosphere w ith T eff = 3925 K ,log(g)= 4.0,and Solarm etallicity (dashed line)
and the x4.2.1 disk m odel(dotted line). T here is no therm alexcess at wavelengths 3 m ,
so dust does not extend to w ithin 0.05 A U .
{ 24 {
Fig. 6.| T he 8{12.5 m spectrum ofT W H ya w ith resolution =
120. T he em ission
in this region show s the character ofam orphous silicates (peaking at 9.6 m ) m ixed w ith
crystalline silicates (w ith peaks at 11.2 m ). A round 9.7 m ,the sm alldepression in ux
is not realbut is caused by im perfect calibration of the tim e variable atm ospheric O zone
em ission. T he overallnorm alization is set to m ake the 11.7 m ux equalthat m easured in
the im aging. T he overallshape agrees wellw ith that observed by Sitko,R ussell,& Lynch
(2000) but the evidence for crystalline com ponents is even m ore com pelling in the atness
ofthe spectrum longward of10 m .
{ 25 {
Fig.7.| T he lim iting F160W absolute m agnitude ofa detectable pointsource asa function
ofdistance from T W H ya. T he m asses as a function ofabsolute m agnitude are based on an
age of10 M yrand cooling m odelsofBurrow setal.(1997)and are accurate to 0.3 m ag (A .
Burrow s,personalcom m unication). A t the radius of 85 A U w here the dip in disk surface
brightness occurs,the lim itis 3M Jup. N o pointsource wasdetected w ithin 400 ofT W H ya.
{ 26 {
12
14
16
18
20
22
0.6
0.8
1
2
4
Fig. 8.| Pro le ofT W H ya disk observed w ith N IC M O S com pared to that observed w ith
W FPC 2 by K ristetal. (2000).Both setsofobservationsshow a sm alldepression in surface
brightness at 1.005 as wellas the sam e overallslope ofthe surface brightness.