The Blue Gray Needle: LBT AO Imaging of the HD 15115

Timothy J. Rodigas + Phil Hinz, Andy Skemer, Kate Su, Glenn Schneider, Laird Close, Vanessa Bailey, Thayne Currie & 48 Co-authors… University of Arizona/LBTO/INAF

Rodigas et al. 2012 (ApJ) L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS dence for significant flattening inward toward the (Fig. 3). All Figure 1 shows the PSF-subtracted images of HD 15115 four surface brightness profiles are well represented by a single with HST and with Keck. The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, and is detected from then it resides within a 40 AU radius. This constraint is consistent 0.5 ע optical HST data hasP.A. p 278Њ.5 the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman is detected as far as 7Љ (315 AU) radius. At this radius the &Song2004;Williams&Andrews2006).∼ ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" ronagraph’s 3.0Љ occulting spot. FurtherHD east, 15115’s past the spot and debrisregion where the diskH-band and V-band data overlap (Fig. 3). At

V band (HST) H band (Keck)

5” 5” Kalas et al. 2007

p p Fig. 1.—False-color,• High log scale fractional images of the HDluminosity 15115 disk as originally disk discovered (5x10^-4; using the Moor ACS/HRC F606Wet al. (l c2006)591 nm,Dl 234 nm; left)and confirmed in H band with Keck II adaptive optics (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use gray fields over the occulting bar and a 3.0Љ occulting spot located to the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding HD 15115 itself.• IfVery the HD 15115asymmetric disk were a symmetric structure, then the east side of the disk would have been detected within the rectangular box, shown below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk within a 2Љ radius, with asymmetry becoming more apparent beyond the 2Љ radius. Due to poor observing conditions, the field is contaminated by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand 7 o’clock relative• toDynamical HD 15115). However, interactions whereas the residual with diffraction ISM pattern noiseto south-east? of the telescope rotates (Debes relative to the et sky al. orientation 2008) over a series of exposures, the image of the disk remains fixed and it is confirmed as real. • BLUE colors (V-H) → small dust grains (< ~ 1 µm?) (Kalas et al. 2007, Debes et al. 2008) Rodigas/NCAD/July 20, 2012 L42 DEBES, WEINBERGER, & SONG Vol. 684

Is the disk really that blue/asymmetric? No. 1, 2008 HD 15115 AT 1.1 mm L43

1.1 µm 1” Red

2” Blue

Debes et al. 2008 Fig. 1.—Top: False-color images of the observations of HD 15115 in the F110W filter at two spacecraft orientations. The images are logarithmically scaled. Bottom: False-color, final combined image of the HD 15115 disk, showing the contaminatingRed feature inside A. The dashed line 1”? represents the nominal position angle of the disk at 278.9Њ, while the asterisk represents the position of the star. We have masked out separations in the disk within 0.68Љ, where 3” Blue PSF subtraction residuals dominate. Noticeable are a warp on the western lobe, Fig. 2.—Measurements of the HD 15115 disk midplane PA (top) and FWHM and the asymmetries between the western and eastern lobes. (bottom) as a function of distance for the two lobes of the disk. Squares represent the western lobe, while triangles represent the eastern lobe. The disk emission to measure the location of the midplane (and sharp rise in PA for the western lobe interior 1.7Љ is due to the warp, and the thus the local PA) as well as the FWHM of the disk. We took eastern lobe’s PA beyond 1.8Љ is marginally different than the western lobe. [See the electronic edition of the Journal for a color version of this figure.] an average of the values for both orientations and estimated the uncertaintyFavorable as being half for the differenceNIR betweenimaging the two Fig. 4.—Comparison of the scattering efficiency (or color) vs. wavelength Fig. 3.—Surface brightness profiles of the eastern and western lobes of the 0.23Љ apertures centered on the midplane of the disk. Be- measurements.disk in F110W magnitudes Figure 2 per shows square arcsecond the PA for and a 0.23 FWHMЉ # 0.23Љ ofaperture. the west- of HD 15115’s# disk at 1Љ (45 AU), 2Љ (90 AU), and 3Љ (135 AU). The symbols ernNo lobe corrections of the were disk applied. as a The function symbols are of the radius. same as in Beyond Fig. 2. [See 1.7 theЉ, we are the sametween as in 0.7 Fig.Љ 2.and The disk 1.8 dramaticallyЉ, the disk changes SB color of the over wavelength western lobe falls as 0.06עfrom0.1 the central star. [See theRodigas/NCAD/July electronic edition of the 20, Journal 2012Ϫ3.56 עelectronic edition of the Journal for a color version of this figure.] and distanceϪ1.4 for a colorrr version, of while this figure. at radii] 11.8Љ, the SB drops as . The 0.06ע0.2Њ,whichisconsistent Ϫ2.14עfind a PA for the disk of 278.9Њ .0.5Њ.However,interiorto eastern lobe drops steadily asr from 0.7Љ to 3Љעwith KFG07’s value of 278.5Њ for the H band ,0.02 ע 1.7Љits, we and note Strehl that ratios, the but disk no PA standard increases technique with is decreasing applied to dis- PSF, we found a correction of 1.23 Efficiency0.06, and for vs. F110W Wavelength we ע the study of circumstellar disks. Intuitively, one would expect we found a correction3.3. Scattering of 1.56 .1Њ at that 0.68Љעtanceminor from corrections the star as to a a function maximum of wavelength PA of 290 providedЊ .0.04 ע The changing PA of the midplane is seen at both orientations. found a correction of 1.36 the disk structure did not significantly change between wave- To compareCombining the NICMOS our data and other with measurements, the reported weV and mustH SB data, a Furthermore, we find at 2Љ a FWHM of 0.29Љ 0.04Љ which KFG07 make ameasure correction of for the the slightly scattering different efficiency photometric of the aperture dust around HD ע.lengths and that the disk itself was mostly resolved is consistentestimated aperture with the corrections FWHM using of 0.19 pointЉ source0.1 PSFs.Љ reported Other by sizes of15115 0.23Љ can# 0.23 beЉ constructed.and 0.25Љ # 0.25 TheЉ, measured respectively. SB To at do any radius in ע KFG07.methods The have median included FWHM deconvolution of the of disk the data from (Golimowski 0.68Љ to 3.69Љ this wethe oversampled disk can be our self-consistently model NICMOS images modeled by for a factor composition and et al.Љ 2006), orЉ using apertures that capturem the majority of the of 3 and calculated the predicted difference in SB measure- The FWHM at 1.1 m is broadened by the grain size distribution provided a knowledge of the density . 0.07 ע is 0.26 NICMOSobserved F110W emission PSF, from which the disk has at each a FWHM wavelength of 0.1 (SchneiderЉ,implyingments between our 0.23Љ # 0.23Љ aperture and a 0.25Љ # 0.25Љ et al. 1999; Weinberger et al. 2002; Debes et al. 2008). We distribution of the dust is known (see details in Debes et al. an intrinsic FWHM for the disk midplane of 0.24Љ or 11 AU. one. This multiplicative correction changes with distance from modeled the effect of the different PSFs of NICMOS,∼ ACS, the star,2008). and ranges The from SB 1.2 of to a 0.8. dust disk at a particular radius is Measurementand Keck H-band of the adaptive eastern optics side on of the the measured disk is complicated surface In FigureF f 4(v we)Q plot(L SB//LFn !), as where a functionf(v) of is wavelength the phase for function of the by the contamination of feature A from PSF subtraction. To ! sca ! IR, brightness for the 0.23Љ # 0.23Љ aperture. HD 15115’sdust∝ at two a particular disk lobes scattering at 1Љ,2Љ, and angle, 3Љ. ThisQ observed is the scattering ef- Our approach to calculating the corrections derives from the sca measure the FWHM and PA, we masked out pixels heavily quantity is Qsca. Our 1.1 mm data provide a third point in the 4 ficiency of the dust, andL /L is the ratio of stellar contaminatedreliable theoretical by this PSFs feature. generated We again by Tiny took Tim thefor averageHST as of the measure of∝ HD 15115’s scattering! efficiencyIR when combined to the IR luminosity of the dust distribution. This is difficult twowell sets as of using measurements a Keck H PSF and corresponding estimated the to uncertainties the October as with the 0.55 mm(V)( and 1.65 mmH) measurements in for HD 15115 because the density distribution of the dust is beingobservations half the ofdifference HD 15115 between (M. Fitzgerald the two 2008, measurements. private com- We KFG07. We derive the uncertainties in the scattering efficiency munication). We took the PSFs for each instrument and con- not known through, for example, spatially resolved thermal find no evidence of a warp in the eastern lobe, which has a for 0.55 and 1.65 mm based on Figure 3 in KFG07. Where no volved them with model disks constructed from the measured error barsemission were images. given for However, the 1.65 m multiwavelengthm data in KFG07, imaging we can pro- 1.3Њ. To compare to the PA of theעmedian PA of 96.6Њ power-law slopes and disk FWHMs for each observation. We assumedvide an useful uncertainty compositional of 20%. In the information western lobe even at 2Љ in, we the absence of western side we add 180 to get 276.6 , consistent with no then measured the SB as aЊ function of distanceЊ for the original find thatspatially the disk’s resolved scattering thermal efficiency emission is neutral constraints. out to 1.1 In the case of significantand convolved bowing models in the with large-scale the 0.23 structureЉ # 0.23Љ ofaperture the disk in as is mm butHD then 15115, drops such three that images the spectrum at different is blue wavelengths at 1.65 mm. will be in- F110W and a 0.25Љ # 0.25Љ aperture for V and H and took seen for HD 61005 (Hines et al. 2007). The median value of The spectrumsufficient is the to provide same from a definitive 0.55 to 1.65 picture,mm at 3butЉ. At can 2Љ, constrain cer- -0.06Љ. correction to the ob- the eastern lobe is strongly blue and at 3Љ the blue trend con עthethe eastern ratio as FWHM a single-valued is 0.29Љ multiplicative served SB profiles. We verified that the slope of the surface tinuestain out to compositions. 1.1 mm. KFG07 did not report SB for the H data brightness profiles did not change due to convolution. To es- In order to compare results across different wavelengths, 3.2. Surface Brightness Profiles at 3Љ, but if the spectrum is the same as the inner part of the timate the uncertainty in this correction, we used the standard disk, oneinstruments, would expect and the telescopes, blue trend it to is continue imperative to longer that a careful and Figuredeviation 3 shows of the measurements a comparison between of the 2 1.1Љ andmm 3Љ,theregion SB profiles in wavelengths.consistent treatment of SB measurements is implemented. Ob- where all the observationsϪ2 overlap. For the ACS coronagraphic Finally, interior to about 1.8Љ, we can get a rough sense of DmmF110Warcsec relative to the central star ( F110W p 6.12) servations of disks at different wavelengths can be affected by for the eastern and western lobes of the disk by taking 0.23Љ the near-IRthe respective colors of PSFsthe disk of in telescopes the east and with west. different We took diffraction lim- 4 See http://www.stsci.edu/software/tinytim/. KFG07’s measurements of the two lobes in H at 1Љ and com- λ > 2 µm imaging: what it gets you

560 A. K. Inoue et al. [Vol. 60, calculation, • constrain dust grain size 1 + 2! (esp. large grains) H.!;!" / : (10) ! 1 + 2!p1 !" " The dotted curves are results from numerical simulations; • inform on geometry (large a1+1Dradiative transfer calculation with the variable Eddington factor method was developed by Dullemond et al. grains also asymmetric?) (2002), but extended to treat the isotropic scattering (A. K. Inoue et al., in preparation). We showed the spectra of the single annulus to be for same as the solid curves. In the calcula- • with 3 µm image, constrain tion, we kept ˇ = 0.05, but iteratively solved the vertical hydro- static equilibrium in the annulus for the central , water ice fraction (Inoue et M = 2.5M ,tobeconsistent with the temperature structure fro#mthe radiˇative equilibrium. As shown in both panels of al. 2008, Honda et al. 2009) figure 3, we obtained excellent agreements between the simple analytic models of equation (9) and the numerical simulations. In the NIR around the 3 !m H2Oiceabsorption feature, • probe for Jupiter-mass thestellarspectrum can be expressed by the Rayleigh–Jeans law. Thus, as an approximation, the scattered spectra may be given by ! "2 (optically thick disk), • I obs " (11) higher Strehl → higher S/N " / #s"2 (optically thin disk) , Inoue et al. 2008 ! " Fig. 4. Expected colors of the scattered light as a function of the dust size: (a) K H O for optically thick case, (b) K H O for optically closer to star which are shown as the dashed curves in figure 3 [see also " 2 " 2 equation (13)]. Indeed, we find that this approximation is good thin case, (c) H2 O L0 for optically thick case, and (d) H2 O L0 for optically thin case." In each panel, open symbols are silicate" dust enough to discuss just if the 3 !m H2Oicefeature is observ- and filled symbols are silicate and ice dust. The dotted line is the Rodigas/NCAD/July 20, 2012 able or not. Hereafter, we use this simple expression as the Rayleigh–Jeans law. The zero of the vertical axis refers to the Vega scattered spectra. magnitude. Figure 4 shows two photometric colors expected as a func- tion of the dust grain size: (a) and (b) for K H2 O,and(c) If the dust includes H Oice(filled diamonds andsquares), " 2 and (d) for H2 O L0.Wehaveassumed the usual set of the significant differences appear; for optically thick disks with ice " 2 filter transmission functions, K and L0. This choice of bands (diamonds), K H O is much bluer than that without ice and " 2 is not definitive. We could choose any two bands shortward H O L0 is much redder than that without ice, because of 2 " and longward of the H2 O band. prominent absorption in the H O band. Note that K H O 2 " 2 The H2 O band transmission function is assumed to be rect- becomes blue even for a large (& 1 !m) dust size, which makes angular between 3.02 and 3.16 !m.Themagnitude zero points gray scattering in K L .Thus,wecan conclude that there is " 0 are taken from Cohen et al. (1992) for K and L0.ForH2 O,we H2Oicydust with a large size if we observe neutral K L0 and " assume F".Vega/ = 357Jy at $ = 3.09 !m based on figure 2 blue K H O colors. On the other hand, we can conclude that " 2 in Cohen et al. (1992). Four disk models, optically thick disk there is H2Oiceonly if we observe red H O L ,because the 2 " 0 with silicate (open circles), optically thick disk with silicate red color is not produced without H2Oice. and ice (filled diamonds), optically thin disk with silicate (open For optically thin disks with ice (squares), K H2 O starts triangles), and optically thin disk with silicate and ice (filled from the Rayleigh–Jeans color (i.e., much redder" color than squares), are shown inthepanels. The dotted lines indicate the that without ice) and gradually becomes bluer as the grain size Rayleigh–Jeans law. becomes larger. H O L oppositely starts from a very blue 2 " 0 The scattered-light colors without H2Oice(opencircles and color, and gradually becomes redder as the grain size becomes triangles) are bluer than the Rayleigh–Jeans law for a small larger. Such behavior is causedbytheresonance feature in the 3 (. 1 !m) dust size, and they approach the Rayleigh–Jeans H2 O band seen in the scattering cross section (figure 2d). color as the dust size increases. The observational fact that Interestingly, these differences, caused by the feature in the most of the scattered light from the disks is not bluer than the H2 O band, enable us to classify the disk and dust properties stellar color indicates a large (& 1 !m) dust size (butAUMic; on a two-color diagram, for example K H2 O and K L0, Krist et al. 2005). as shown in figure 5. In this figure, we" find four sequences:" an optically thick disk with silicate (open circles), an 2 We used the machine readable form downloaded from the Subaru/ optically thick disk with silicate and ice (filled diamonds), an CIAO web page; http://subarutelescope.org/Observing/Instruments/CIAO/ optically thin disk with silicate (open triangles), and an camera/sensitivity.htmlh . optically thin disk with silicate and ice (filled squares). Each i 3 In this paper, we do not consider any specific size distribution function of sequence consists of five points corresponding to grain sizes of the dust particle, but consider a “typical” size of the particle. This typical 0.1, 0.3, 1.0, 3.0, and 10 ! .Thelargeasterisk indicates the size can be defined by an average over the size distribution function with m a certain weight, for example, cross section in a certain band. We can location expected fromtheRayleigh–Jeans law. In figure 5, we consider a typical size of the interstellar dust to be 0.1 !m. find five classifications: $ Observations with LBT AO: new Ks and L’ images 4 4 Rodigas et al. Rodigas et al. 2.15 µm (PISCES) 3.8 µm (LBTI/LMIRCam)

2 2 2.5 2.5 2.5 2.5 2.5 2.5

2 2 2 2

2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 1 1 0.5 0.5 0.5 0.5 1 1 0 0 0 0 0.5 0.5 arcseconds arcseconds arcseconds arcseconds −0.5 −0.5 0.5 0.5 −0.5 −0.5

−1 −1 −1 −1 0 0 0 0 −1.5 −1.5 −1.5 −1.5

−0.5 −2 −2 −0.5 −2 −2 −0.5 −0.5 −2.5 −2.5 −2.5 −2.5 −1 −1 −2.5 −2 −1.5 −1 −0.5 −2.50 −0.52 −1.51 −1.51 −0.52 2.50 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 −2.50 0.5−2 −1.51 1.5−1 −0.52 2.50 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(a) (a) (a) (a) The disk is morphologically different between 2 & 4 µm, 8 8 2.5 2.5 2.5 2.5 and between 0.5-1 & 2-4 µm 5 5 7 2 2 7 2 2

1.5 1.5 6 6 1.5 1.5 Rodigas/NCAD/July 20, 2012 4 4

1 1 1 1 5 5

0.5 0.5 0.5 0.5 3 3 4 4 0 0 0 0 3 arcseconds 3 arcseconds arcseconds arcseconds 2 −0.5 −0.5 −0.5 −0.5 2

2 2 −1 −1 −1 −1

1 1 1 −1.5 −1.5 1 −1.5 −1.5

−2 −2 −2 0 0 −2 0 0 −2.5 −2.5 −2.5 −1 −2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS dence for significant flattening inward toward the star (Fig. 3). All Figure 1 shows the PSF-subtracted images of HD 15115 four surface brightness profiles are well represented by a single with HST and with Keck. The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, and is detected from then it resides within a 40 AU radius. This constraint is consistent 0.5 ע optical HST data hasP.A. p 278Њ.5 the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman is detected as far as 7Љ (315 AU) radius. At this radius the &Song2004;Williams&Andrews2006).∼ ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" ronagraph’s 3.0Љ occulting spot. Further east, past the spot and region where the H-band and V-band data overlap (Fig. 3). At

Observations with LBT AO: new Ks and L’ images 4 4 0.6 µm Rodigas et al. Rodigas et al. 2.15 µm (PISCES) 3.8 µm (LBTI/LMIRCam)

2 2 2.5 2.5 2.5 2.5 2.5 2.5

2 2 2 2

2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 1 1 0.5 0.5 0.5 0.5 1 1 0 0 0 0 0.5 0.5 arcseconds arcseconds arcseconds arcseconds −0.5 −0.5 0.5 0.5 −0.5 −0.5

−1 −1 −1 −1 0 0 0 0 −1.5 −1.5 −1.5 −1.5

−0.5 −2 −2 −0.5 −2 −2 −0.5 −0.5 −2.5 −2.5 −2.5 −2.5 −1 −1 −2.5 −2 −1.5 −1 −0.5 −2.50 −0.52 −1.51 −1.51 −0.52 2.50 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 −2.50 0.5−2 −1.51 1.5−1 −0.52 2.50 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(a) (a) (a) (a) The disk is morphologically different between 2 & 4 µm, 8 8 2.5 2.5 2.5 2.5 and between 0.5-1 & 2-4 µm 5 5 7 2 2 7 2 2

1.5 1.5 6 6 1.5 1.5 Rodigas/NCAD/July 20, 2012 4 4

1 1 1 1 5 5 p p 0.5 Fig. 1.—False-color, log scale images of the HD 151150.5 disk as originally discovered using the ACS/HRC F606W (lc 591 nm,Dl 234 nm; left)and 0.5 0.5 3 3 confirmed in H band with Keck II4 adaptive optics4 (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use 0 0 0 0 3 arcseconds 3 arcseconds gray fields overarcseconds the occulting bar and a 3.0Љ occulting spot located toarcseconds the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding −0.5 −0.5 2 2 HD 15115 itself.−0.5 If the HD 15115 disk were a symmetric structure,− then0.5 the east side of the disk would have been detected within the rectangular box, shown 2 2 −1 −1 −1 −1 Љ below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk1 within a 2 radius,1 with asymmetry becoming more apparent beyond 1 −1.5the 2Љ radius.−1.5 Due to poor observing conditions, the1 field−1.5 is contaminated−1.5 by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand −2 −2 −2 0 0 −2 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of0 the telescope0 rotates relative to the sky orientation over a series of −2.5 −2.5 −2.5 −1 −2.5 exposures,−2.5 −2 −1.5 −1 the−0.5 image0 0.5 1 of1.5 the2 disk2.5 remains fixed and−1 it is−2.5 confirmed−2 −1.5 −1 as−0.5 real. 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS dence for significant flattening inward toward the star (Fig. 3). All Figure 1 shows the PSF-subtracted images of HD 15115 four surface brightness profiles are well represented by a single with HST and with Keck. The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, and is detected from then it resides within a 40 AU radius. This constraint is consistent 0.5 ע optical HST data hasP.A. p 278Њ.5 the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman is detected as far as 7Љ (315 AU) radius. At this radius the &Song2004;Williams&Andrews2006).∼ ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" ronagraph’s 3.0Љ occulting spot. Further east, past the spot and region where the H-band and V-band data overlap (Fig. 3). At

Observations with LBT AO: new Ks and L’ images 4 4 0.6 µm 1.1 µmRodigas et al. Rodigas et al. 2.15 µm (PISCES) 3.8 µm (LBTI/LMIRCam)

2 2 2.5 2.5 2.5 2.5 2.5 2.5

2 2 2 2

2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 L42 DEBES, WEINBERGER, & SONG Vol. 684 1 1 0.5 0.5 0.5 0.5 1 1 0 0 0 0 0.5 0.5 arcseconds arcseconds arcseconds arcseconds −0.5 −0.5 0.5 0.5 −0.5 −0.5

−1 −1 −1 −1 0 0 0 0 −1.5 −1.5 −1.5 −1.5

Fig. 1.—Top: False-color images of the observations of HD 15115 in the F110W filter at two spacecraft orientations. The images are logarithmically scaled. Bottom: False-color, final combined image of the HD 15115 disk, −0.5 showing the contaminating feature−0.5 A. The dashed line represents the nominal −2 −2 position angle of the disk at 278.9Њ, while the asterisk represents−2 the position −2 of the star. We have masked out separations in the disk within 0.68Љ, where PSF subtraction residuals dominate. Noticeable are a warp on the western lobe, Fig. 2.—Measurements of the HD 15115 disk midplane PA (top) and FWHM and the asymmetries between the western and eastern lobes. (bottom) as a function of distance for the two lobes of the disk. Squares −0.5 −0.5 represent the western lobe, while triangles represent the eastern lobe. The −2.5 disk emission to measure the location of the−2.5 midplane (and sharp rise in PA for the western lobe interior 1.7Љ is due to the warp, and the −2.5 thus the local PA) as well as the FWHM of the disk. We took eastern lobe’s PA beyond 1.8Љ is marginally different− than2.5 the western lobe. [See the electronic edition of the Journal for a color version of this figure.] −1 an average of the values−1 for both orientations and estimated −2.5 −2 −1.5 −1 −0.5 −2.50 −0.52 −1.51 −1.51 −0.52 2.50 0.5 1 1.5 2 the2.5 uncertainty as being half the difference between− the2.5 two −2 −1.5 −1 −0.5 −2.50 0.5−2 −1.51 1.5−1 −0.52 2.50 0.5 1 1.5 2 2.5 measurements. Figure 2 shows the PA and FWHM of the west- # 0.23Љ apertures centered on the midplane of the disk. Be- arcseconds ern lobe of the disk as a function of radius. Beyond 1.7Љ, we tween 0.7Љ and 1.8Љ, the disk SB of the westernarcseconds lobe falls as The arcseconds . 0.06עwhile at radii 11.8Љ, the SB drops asϪ3.56 ,0.1עarcseconds rrϪ1.4 0.06ע0.2Њ,whichisconsistent Ϫ2.14עfind a PA for the disk of 278.9Њ .0.5Њ.However,interiorto eastern lobe drops steadily asr from 0.7Љ to 3Љעwith KFG07’s value of 278.5Њ 1.7Љ, we note that the disk PA increases with decreasing dis- 1Њ at 0.68Љ. 3.3. Scattering Efficiency vs. Wavelengthעtance from the star to a maximum PA of 290Њ The changing PA of the midplane is seen at both orientations. Combining our data with the reported V and H SB data, a Furthermore, we find at 2Љ a FWHM of 0.29Љ 0.04Љ which measure of the scattering efficiency of the dust around HD ע is consistent with the FWHM of 0.19Љ 0.1Љ reported by can be constructed. The measured SB at any radius(a) in 15115 ע (a) (a) KFG07. The median FWHM of the disk from 0.68Љ to 3.69Љ (a) the disk can be self-consistently modeled for composition and Љ Љ m The FWHM at 1.1 m is broadened by the grain size distribution provided a knowledge of the density . 0.07 ע is 0.26 NICMOS F110W PSF, which has a FWHM of 0.1Љ,implying distribution of the dust is known (see details in Debes et al. an intrinsic FWHM for the disk midplane of 0.24Љ or 11 AU. 2008). The SB of a dust disk at a particular radius is Measurement of the eastern side of the disk∼ is complicated F f(v)Q (L /L ), wheref(v) is the phase function of the by the contamination of feature A from PSF subtraction. To ! sca ! IR dust∝ at a particular scattering angle,Q is the scattering ef- measure the FWHM and PA, we masked out pixels heavily sca ficiency of the dust, andL /L is the ratio of stellar luminosity contaminated by this feature. We again took the average of the ! IR The disk is morphologically differentto the IRbetween luminosity of the dust distribution. This is difficult2 & 4 µm, two sets of measurements and estimated the uncertainties as for HD 15115 because the density distribution of the dust is being half the difference between the two measurements. We 8 8 not known through, for example, spatially resolved thermal find no evidence of a warp in the eastern lobe, which has a emission images. However, multiwavelength imaging can pro- 2.5 2.5 median PA of 96.6 1.3 . To compare to2.5 the PA of the 2.5 Њ vide useful compositional information even in the absence of עЊ western side we add 180 to get 276.6 , consistent with no Њ Њ spatially resolved thermal emission constraints. In the case of and between 0.5-1 & 2-4significant µm bowing in the large-scale structure of the disk as is HD 15115, three images at different wavelengths will be in- 5 5 seen for HD 61005 (Hines et al. 2007). The median value of sufficient to provide a definitive picture, but can constrain cer- 7 the eastern FWHM is 0.29Љ 0.06Љ. tain compositions. 2 2 ע 7 2 2 In order to compare results across different wavelengths, 3.2. Surface Brightness Profiles instruments, and telescopes, it is imperative that a careful and Figure 3 shows a comparison of the 1.1 mm SB profiles in consistent treatment of SB measurements is implemented. Ob- Ϫ2 DmmF110Warcsec relative to the central star ( F110W p 6.12) servations of disks at different wavelengths can be affected by 1.5 1.5 6 for the eastern and western6 lobes of the disk by1.5 taking 0.23Љ the respective PSFs of telescopes with different1.5 diffraction lim- Rodigas/NCAD/July 20, 2012 4 4

1 1 1 1 5 5 p p 0.5 Fig. 1.—False-color, log scale images of the HD 151150.5 disk as originally discovered using the ACS/HRC F606W (lc 591 nm,Dl 234 nm; left)and 0.5 0.5 3 3 confirmed in H band with Keck II4 adaptive optics4 (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use 0 0 0 0 3 arcseconds 3 arcseconds gray fields overarcseconds the occulting bar and a 3.0Љ occulting spot located toarcseconds the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding −0.5 −0.5 2 2 HD 15115 itself.−0.5 If the HD 15115 disk were a symmetric structure,− then0.5 the east side of the disk would have been detected within the rectangular box, shown 2 2 −1 −1 −1 −1 Љ below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk1 within a 2 radius,1 with asymmetry becoming more apparent beyond 1 −1.5the 2Љ radius.−1.5 Due to poor observing conditions, the1 field−1.5 is contaminated−1.5 by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand −2 −2 −2 0 0 −2 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of0 the telescope0 rotates relative to the sky orientation over a series of −2.5 −2.5 −2.5 −1 −2.5 exposures,−2.5 −2 −1.5 −1 the−0.5 image0 0.5 1 of1.5 the2 disk2.5 remains fixed and−1 it is−2.5 confirmed−2 −1.5 −1 as−0.5 real. 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS L86 KALAS, FITZGERALD, & GRAHAM Vol. 661 dence for significant flattening inward toward the star (Fig. 3). All TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Figure 1 showsStellar Properties the PSF-subtractedradius. The appearance images of the disk ofis more HD symmetric 15115 in the four surface brightness profiles are well represented by a single Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). with HSTSpectral typeand...... with F2 Keck. M0 Hipparcos The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an then it resides within a 40 AU radius. This constraint is consistent ע Distance (pc) ...... 44.78ϩ2.22 40.54ϩ4.38 pHipparcos Њ optical HST data hasϪ2.01 P.A.Ϫ3.61 278 .5edge-on orientation0.5 to and the line of is sight. detected The west midplane from is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Hipparcos cluding a characteristic width asymmetry (Kalas & Jewitt 1995; with model fits of the spectral energy distribution that place the 4.46 ע 82.32 1.09 ע m (mas yrϪ1)...... 86.09 the edgea of the occulting spot at 1.5Љ (67 AU) radius to the Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 2.7Љ (554 Tycho-2 AU)the full radius. width at half-maximum The across east the disk midplane midplane at 2Љ dominant emitting dust component at 35 AU radius (Zuckerman ע 12.38Ϫ55.1 1.2 עedge ofmd the(mas yr )...... fieldϪ50.9 at Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 ∽ a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 .(Љ0.5(315 …a AU)ever, the radius. vertical cuts are not At symmetric this about the radius midplane when the &Song2004;Williams&Andrews2006עas Ϫ10.07 2.3 ע is detectedW (km sϪ1)...... as farϪ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively.∼ HWQM north of the west midplane is1.6 (2006) east midplane begins to intersectthan the that of outer the HWQM south portion of the west midplane. of This the width co- The color of the disk may be estimated in the 2.0"" –3.3 "" resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry ronagraph’sthe 2007 January 3.0 dataЉ thereforeocculting yields a detection of spot. the west Furtheris found to be in the east, opposite direction past in the the opposite spot midplane, and region where the H-band and V-band data overlap (Fig. 3). At ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- Observations3. RESULTS with denceLBT for significant AO: flattening inward new toward the starKs (Fig. 3). and All L’ images 4 4 Figure 1 shows the PSF-subtracted images of HD 15115 Rodigasfour surface brightness et al. profilesRodigas are well represented et al. by a single with HST and with Keck. The west0.6 side µm of the disk1.1 in the µmpower-law 1.65 decrease µm with radius. If there is an inner dust depletion, PISCES) and is detected from then it resides within a 40 AU radius. This3.8 constraint µm is (LBTI/LMIRCam) consistent) 0.5 עoptical HST data hasP.A. p2.15278Њ.5 µm the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman 2 2 2.5 2.5 2.5 ∼ 2.5 is detected as far as 7Љ (315 AU) radius. At this radius2.5 the &Song2004;Williams&Andrews2006).2.5 ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" 2 ronagraph’s2 3.0Љ occulting spot. Further east, past the spot and region where the H-band and2 V-band data overlap (Fig. 3).2 At 2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 L42 DEBES, WEINBERGER, & SONG Vol. 684 1 1 0.5 0.5 0.5 0.5 1 1 0 0 0 0 0.5 0.5 arcseconds arcseconds arcseconds arcseconds −0.5 −0.5 0.5 0.5 −0.5 −0.5

−1 −1 −1 −1 0 0 0 0 −1.5 −1.5 −1.5 −1.5

Fig. 1.—Top: False-color images of the observations of HD 15115 in the F110W filter at two spacecraft orientations. The images are logarithmically scaled. Bottom: False-color, final combined image of the HD 15115 disk, −0.5 showing the contaminating feature−0.5 A. The dashed line represents the nominal −2 −2 position angle of the disk at 278.9Њ, while the asterisk represents−2 the position −2 of the star. We have masked out separations in the disk within 0.68Љ, where PSF subtraction residuals dominate. Noticeable are a warp on the western lobe, Fig. 2.—Measurements of the HD 15115 disk midplane PA (top) and FWHM and the asymmetries between the western and eastern lobes. (bottom) as a function of distance for the two lobes of the disk. Squares −0.5 −0.5 represent the western lobe, while triangles represent the eastern lobe. The −2.5 disk emission to measure the location of the−2.5 midplane (and sharp rise in PA for the western lobe interior 1.7Љ is due to the warp, and the −2.5 thus the local PA) as well as the FWHM of the disk. We took eastern lobe’s PA beyond 1.8Љ is marginally different− than2.5 the western lobe. [See the electronic edition of the Journal for a color version of this figure.] −1 an average of the values−1 for both orientations and estimated −2.5 −2 −1.5 −1 −0.5 −2.50 −0.52 −1.51 −1.51 −0.52 2.50 0.5 1 1.5 2 the2.5 uncertainty as being half the difference between− the2.5 two −2 −1.5 −1 −0.5 −2.50 0.5−2 −1.51 1.5−1 −0.52 2.50 0.5 1 1.5 2 2.5 measurements. Figure 2 shows the PA and FWHM of the west- # 0.23Љ apertures centered on the midplane of the disk. Be- tween 0.7Љ and 1.8Љ, the disk SB of the western lobe falls as arcseconds 0.06עarcsecondsϪ3.56 0.1עarcseconds arcseconds ern lobe of the disk as a function of radius. Beyond 1.7Љ, we Ϫ1.4 Fig. l p rr, whileDl at radiip11.8Љ, the SB drops as . The left)and 0.06ע;F606WЊ,whichisconsistent (c 591 nm,234Ϫ2.14 nm 0.2עFalse-color, log scale images of the HD 15115 disk as originally discovered usingfind a PA the for the ACS/HRC disk of 278.9Њ—.1 .0.5Њ.However,interiorto eastern lobe drops steadily asr from 0.7Љ to 3Љעwith KFG07’s value of 278.5Њ confirmed in H band with Keck II adaptive optics (right;2006October26data).Northisup,eastisleft,andthescalebarsspan51.7Љ, we note that the disk PA increases with decreasing dis- Љ. In the HST image we use 1Њ at 0.68Љ. 3.3. Scattering Efficiency vs. Wavelengthעtance from the star to a maximum PA of 290Њ gray fields over the occulting bar and a 3.0Љ occulting spot located to the left of HD 15115,The as changing well PA as of thea graymidplane disk is seen at covering both orientations. PSF-subtraction artifacts surrounding Combining our data with the reported V and H SB data, a Furthermore, we find at 2Љ a FWHM of 0.29Љ 0.04Љ which measure of the scattering efficiency of the dust around HD ע HD 15115 itself. If the HD 15115 disk were a symmetric structure, then the east side of theis consistent disk with would the FWHM have of 0.19 beenЉ 0.1 detectedЉ reported by within the rectangular box, shown can be constructed. The measured SB at any radius(a) in 15115 ע (a) (a) KFG07. The median FWHM of the disk from 0.68Љ to 3.69Љ (a) below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk within a 2Љ radius, with asymmetry becomingthe disk can be self-consistently more apparent modeled for beyond composition and Љ Љ m The FWHM at 1.1 m is broadened by the grain size distribution provided a knowledge of the density . 0.07 ע is 0.26 NICMOS F110W PSF, which has a FWHM of 0.1Љ,implying the 2Љ radius. Due to poor observing conditions, the field is contaminated by residual noise due to the diffraction pattern of thedistribution telescope of the (e.g., dust is known at 2 (see o’cloc details inkand Debes et al. an intrinsic FWHM for the disk midplane of 0.24Љ or 11 AU. 2008). The SB of a dust disk at a particular radius is Measurement of the eastern side of the disk∼ is complicated 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of the telescope rotates relative to the skyF f(v)Q orientation(L /L ), wheref ove(v) isr the a phase series function of of the by the contamination of feature A from PSF subtraction. To ! sca ! IR dust∝ at a particular scattering angle,Q is the scattering ef- exposures, the image of the disk remains fixed and it is confirmed as real. measure the FWHM and PA, we masked out pixels heavily sca ficiency of the dust, andL /L is the ratio of stellar luminosity contaminated by this feature. We again took the average of the ! IR The disk is morphologically differentto the IRbetween luminosity of the dust distribution. This is difficult2 & 4 µm, two sets of measurements and estimated the uncertainties as for HD 15115 because the density distribution of the dust is being half the difference between the two measurements. We 8 8 not known through, for example, spatially resolved thermal find no evidence of a warp in the eastern lobe, which has a emission images. However, multiwavelength imaging can pro- 2.5 2.5 median PA of 96.6 1.3 . To compare to2.5 the PA of the 2.5 Њ vide useful compositional information even in the absence of עЊ western side we add 180 to get 276.6 , consistent with no Њ Њ spatially resolved thermal emission constraints. In the case of and between 0.5-1 & 2-4significant µm bowing in the large-scale structure of the disk as is HD 15115, three images at different wavelengths will be in- 5 5 seen for HD 61005 (Hines et al. 2007). The median value of sufficient to provide a definitive picture, but can constrain cer- 7 the eastern FWHM is 0.29Љ 0.06Љ. tain compositions. 2 2 ע 7 2 2 In order to compare results across different wavelengths, 3.2. Surface Brightness Profiles instruments, and telescopes, it is imperative that a careful and Figure 3 shows a comparison of the 1.1 mm SB profiles in consistent treatment of SB measurements is implemented. Ob- Ϫ2 DmmF110Warcsec relative to the central star ( F110W p 6.12) servations of disks at different wavelengths can be affected by 1.5 1.5 6 for the eastern and western6 lobes of the disk by1.5 taking 0.23Љ the respective PSFs of telescopes with different1.5 diffraction lim- Rodigas/NCAD/July 20, 2012 4 4

1 1 1 1 5 5 p p 0.5 Fig. 1.—False-color, log scale images of the HD 151150.5 disk as originally discovered using the ACS/HRC F606W (lc 591 nm,Dl 234 nm; left)and 0.5 0.5 3 3 confirmed in H band with Keck II4 adaptive optics4 (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use 0 0 0 0 3 arcseconds 3 arcseconds gray fields overarcseconds the occulting bar and a 3.0Љ occulting spot located toarcseconds the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding −0.5 −0.5 2 2 HD 15115 itself.−0.5 If the HD 15115 disk were a symmetric structure,− then0.5 the east side of the disk would have been detected within the rectangular box, shown 2 2 −1 −1 −1 −1 Љ below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk1 within a 2 radius,1 with asymmetry becoming more apparent beyond 1 −1.5the 2Љ radius.−1.5 Due to poor observing conditions, the1 field−1.5 is contaminated−1.5 by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand −2 −2 −2 0 0 −2 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of0 the telescope0 rotates relative to the sky orientation over a series of −2.5 −2.5 −2.5 −1 −2.5 exposures,−2.5 −2 −1.5 −1 the−0.5 image0 0.5 1 of1.5 the2 disk2.5 remains fixed and−1 it is−2.5 confirmed−2 −1.5 −1 as−0.5 real. 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS L86 KALAS, FITZGERALD, & GRAHAM Vol. 661 dence for significant flattening inward toward the star (Fig. 3). All TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Figure 1 showsStellar Properties the PSF-subtractedradius. The appearance images of the disk ofis more HD symmetric 15115 in the four surface brightness profiles are well represented by a single Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). with HSTSpectral typeand...... with F2 Keck. M0 Hipparcos The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an then it resides within a 40 AU radius. This constraint is consistent ע Distance (pc) ...... 44.78ϩ2.22 40.54ϩ4.38 pHipparcos Њ optical HST data hasϪ2.01 P.A.Ϫ3.61 278 .5edge-on orientation0.5 to and the line of is sight. detected The west midplane from is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Hipparcos cluding a characteristic width asymmetry (Kalas & Jewitt 1995; with model fits of the spectral energy distribution that place the 4.46 ע 82.32 1.09 ע m (mas yrϪ1)...... 86.09 the edgea of the occulting spot at 1.5Љ (67 AU) radius to the Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 2.7Љ (554 Tycho-2 AU)the full radius. width at half-maximum The across east the disk midplane midplane at 2Љ dominant emitting dust component at 35 AU radius (Zuckerman ע 12.38Ϫ55.1 1.2 עedge ofmd the(mas yr )...... fieldϪ50.9 at Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 ∽ a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 .(Љ0.5(315 …a AU)ever, the radius. vertical cuts are not At symmetric this about the radius midplane when the &Song2004;Williams&Andrews2006עas Ϫ10.07 2.3 ע is detectedW (km sϪ1)...... as farϪ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively.∼ HWQM north of the west midplane is1.6 (2006) east midplane begins to intersectthan the that of outer the HWQM south portion of the west midplane. of This the width co- The color of the disk may be estimated in the 2.0"" –3.3 "" resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry ronagraph’sthe 2007 January 3.0 dataЉ thereforeocculting yields a detection of spot. the west Furtheris found to be in the east, opposite direction past in the the opposite spot midplane, and region where the H-band and V-band data overlap (Fig. 3). At ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- Observations3. RESULTS with denceLBT for significant AO: flattening inward new toward the starKs (Fig. 3). and All L’ images 4 4 Figure 1 shows the PSF-subtracted images of HD 15115 Rodigasfour surface brightness et al. profilesRodigas are well represented et al. by a single with HST and with Keck. The west0.6 side µm of the disk1.1 in the µmpower-law 1.65 decrease µm with radius. If2.15 there is an µm inner dust depletion, PISCES) and is detected from then it resides within a 40 AU radius. This3.8 constraint µm is (LBTI/LMIRCam) consistent) 0.5 עoptical HST data hasP.A. p2.15278Њ.5 µm the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman 2 2 2.5 2.5 2.5 ∼ 2.5 is detected as far as 7Љ (315 AU) radius. At this radius2.5 the &Song2004;Williams&Andrews2006).2.5 ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" 2 ronagraph’s2 3.0Љ occulting spot. Further east, past the spot and region where the H-band and2 V-band data overlap (Fig. 3).2 At 2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 L42 DEBES, WEINBERGER, & SONG Vol. 684 1 1 0.5 0.5 0.5 0.5 4 Rodigas et al. 1 1

2 2.5 2.5 0 0 0 2.5 0

2 2

2 1.5 1.5 1.5 0.5 0.5 arcseconds arcseconds arcseconds arcseconds

1 1.5 1 −0.5 −0.5 0.5 0.5 −0.5 −0.5 1 0.5 0.5 1 0 0 0.5 arcseconds arcseconds −0.5 0.5 −0.5

−1 −1 −1 −1 −1 −1 0 0 0 −1.5 0 −1.5 −2 −0.5 −2 0 0 −0.5 −2.5 −2.5 −1.5 −1.5 −1 −1.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 −1.50.5 1 1.5 2 2.5 arcseconds arcseconds Fig. 1.—Top: False-color images of the observations of HD 15115 in the F110W filter at two spacecraft orientations.(a) The images are logarithmically (a) scaled. Bottom: False-color, final combined image of the HD 15115 disk, −0.5 showing the contaminating feature−0.5 A. The dashed line represents the nominal −2 −2 position angle of the disk at 278.9Њ, while the asterisk represents−2 the position −2 of the star. We have masked out separations in the disk within 0.68Љ, where8 PSF subtraction2.5 residuals dominate. Noticeable are a warp on the western lobe, Fig. 2.—Measurements2.5 of the HD 15115 disk midplane PA (top) and FWHM and the asymmetries between the western and eastern lobes. 5 −0.5 2 7 (bottom) as a function of2 distance for the two lobes of the disk. Squares −0.5 represent the western lobe, while triangles represent the eastern lobe. The disk emission1.5 to measure the location of the midplane (and6 sharp rise in PA for the western1.5 lobe interior 1.7Љ is due to the warp, and the −2.5 −2.5 4 −2.5 thus the local PA) as well as the FWHM of the disk. We took eastern lobe’s PA beyond 1.8Љ is marginally different− than2.5 the western lobe. 1 [See the electronic edition1 of the Journal for a color version of this figure.] −1 an average of the values−1 for both orientations and estimated 5 the uncertainty0.5 as being half the difference between the two 0.5 3 −2.5 −2 −1.5 −1 −0.5 −2.50 −0.52 −1.51 −1.51 −0.52 2.50 0.5 1 1.5 2 2.5 −2.5 4 −2 −1.5 −1 −0.5 −2.50 0.5−2 −1.51 1.5−1 −0.52 2.50 0.5 1 1.5 2 2.5 measurements. Figure 2 shows the PA and FWHM of the west- # 0.23Љ apertures centered on the midplane of the disk. Be- 0 tween 0.7Љ and 1.8Љ,0 the disk SB of the western lobe falls as arcseconds ern lobe of the disk as a function of radius. Beyond 1.7Љ, we3 arcseconds arcseconds 0.06עarcseconds Ϫ3.56 0.1עarcseconds arcseconds Ϫ1.4 F606WЊ,whichisconsistent (l p 591rr nm,, whileDl at radii−p0.5 11.8234Љ, the nm;SB dropsleft as)and . The 2 0.2עFig. 1.—False-color, log scale images of the HD 15115 disk as originally discovered usingfind a PA the for−0.5 the ACS/HRC disk of 278.9Њ .from 0.7Љ to 3Љ 0.06עc eastern lobe drops steadily asrϪ2.14 0.5Њ.However,interiorto2עwith KFG07’s value of 278.5Њ confirmed in H band with Keck II adaptive optics (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5−1 Љ. In the−1 HST image we use 1.7Љ, we note that the disk PA increases with decreasing dis- 1 1Њ at 0.68Љ1. 3.3. Scattering−1.5 Efficiency vs. Wavelengthעtance from the−1.5 star to a maximum PA of 290Њ gray fields over the occulting bar and a 3.0Љ occulting spot located to the left of HD 15115,The as changing well PA as of thea graymidplane disk is seen at covering both orientations. PSF-subtraction artifacts surrounding −2 0 Combining our data−2 with the reported V and H SB data, a 0.04Љ which 0 ע Furthermore, we find at 2Љ a FWHM of 0.29Љ HD 15115 itself. If the HD 15115 disk were a symmetric structure, then the east side of the disk−2.5 would have been detected withinmeasure the of the rectangular scattering−2.5 efficiency box, of the dust show aroundn HD 0.1Љ reported by−1 ע is consistent with the FWHM of 0.19Љ (a) (a) −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 15115 can be constructed.−2.5 −2 The−1.5 measured−1 −0.5 SB0 at0.5 any1 radius1.5 (a)2 in 2.5 (a) KFG07. The median FWHM ofarcseconds the disk from 0.68Љ to 3.69Љ arcseconds below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk within a 2Љ radius, with asymmetry becomingthe disk can be self-consistently more apparent modeled for beyond composition and Љ Љ m The FWHM at 1.1 m is broadened by the grain size distribution provided a knowledge of the density . 0.07 ע is 0.26 NICMOS F110W PSF, which has a FWHM of 0.1Љ,implying the 2Љ radius. Due to poor observing conditions, the field is contaminated by residual noise due to the diffraction(b) pattern of thedistribution telescope of the (e.g., dust is known at 2 (see o’cloc details(b) inkand Debes et al. an intrinsic FWHM for the disk midplane of 0.24Љ or 11 AU. Fig. 1.— Top : Final PISCES Ks band image of the HD 151152008). TheFig. SB of2.— aTop dust disk at! a particular radius is Measurement of the eastern side of the disk2∼ is complicated :FinalL image of the disk obtained with LMIR- 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise ofdebris the disk, telescope in units of mJy/arcsecond rotates relative, with North-up, to the East-left. skyF f(v)Q orientation(L /L ), wheref ove(v) isr the a phase series2 function of of the by the contamination of feature A from PSF subtraction. To ! scacam,! inIR units of mJy/arcsecond , with North-up, East-left. The The white dot marks the location of the star, and representsdust∝ the at a particularwhite dot scattering marks the angle, locationQ ofis the scatteringstar, and represents ef- the size of exposures, the image of the disk remains fixed and it is confirmed as real. measure the FWHM and PA, we masked out pixels heavily ! sca !! size of a resolution element at Ks band. For display purposesficiency a of thea resolution dust, and elementL /L is at theL ratio. For of display stellar luminosity purposes a 0. 9 radius mask contaminated0!!. 8 radius by mask this hasfeature. been We added again in took post-processing. the average of the The western ! IR The disk is morphologically different2 to the IRbetween luminosityhas been added of the in dust post-processing. distribution. This The is image difficult2 has been& smoothed 4 µm, twoside sets SB of is measurements∼ a magnitude/arcsecond and estimated thebrighter uncertainties than as the eastern by a Gaussian kernel with FWHM = λ/D and binned by a factor for HD 15115 because the density distribution! of the dust is 8 beingside half SB, the as isdifference seen at shorter between wavelengths the two measurements. (Kalas et al. We 2007; Debes of 4 to bring out the disk. At L the disk is mostly symmetric and et al. 2008). Bottom8: SNRE map of the image. The eastern sidenot is known through, for example, spatially resolved thermal find no evidence of a warp in the eastern!! lobe, which has a is equally bright on both sides. These features are nearly opposite 2.5 detected at SNRE ∼ 3outto∼ 2. 1. The2.5 western side of theemission disk images. However, multiwavelength imaging can pro- :1.3Њ. To compare!! to the PA of the to what is seen at Ks band and at2.5 shorter wavelengths. Bottomעmedian PA of 96.6Њ 2.5 is detected at SNRE ∼ 3-8 from ∼ 1-2. 5. vide usefulSNRE compositional map of the information final L! image even in shown the absence above. ofThe disk is detected western side we add 180Њ to get 276.6Њ, consistent with no !! spatially resolvedat SNRE thermal∼ 2.5 emission between constraints.∼ 1-1. 5 on In both the case sides. of and between 0.5-1 & 2-4significant µm bowing in the large-scale structure of the disk as is of data, images were averaged in sets of 4. We registeredHD 15115, three images at different wavelengths will be in- 5 5 seen for HD 61005 (Hines et al. 2007). The median value of sufficient to provide a definitive picture, but can constrain cer- pixelЉ. by calculating the center overwhelm the brightness of any disk structure seen in 0.06 ע thethe eastern images FWHM to is a 0.297 fiducialЉ 7 2 2 of light around the star, excluding the2 saturated pixelstain compositions.the final image. Therefore we2 used only the first half In order to compare results across different wavelengths, ∼ 3.2. Surface Brightness Profiles ∼ ∼ ◦ within λ/D, in the same manner as for the Ks bandinstruments,of and the telescopes, data ( it40 is imperative minutes thatof integration, a careful and with 50 of Figureimages. 3 shows We a performed comparison of similar the 1.1 centermm SB profiles of light in calcula-consistent treatmentparallactic of SB angle measurements rotation). is implemented. Ob- tions on theϪ2 unsaturated images of HD 15115 and found After scaling and subtracting (in the same manner as DmmF110Warcsec relative to the central star ( F110W p 6.12) servations of disks at different wavelengths can be affected by 1.5 1.5 6 forthe the dieasternfference and westernbetween lobes the of measured the disk by1.5 and taking true 0.23 centroidЉ the to respectivefor PSFs the ofKs telescopesband images) with different the1.5 diffraction master PSF lim- from each im- be ∼ 0.2 pixels6 (2 mas). We then calculated the radial age in the first half of the data (240 averaged images;Rodigas/NCAD/July 20, 2012 profile of each image and subtracted it, also in the same 960 total), we rotated each image by its parallactic an- 4 4 manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The 1 the data, we median-combined the first1 half of the data, offset was determined to be 1.81◦ ± 0.0685◦ by calculat- 1 the second half, and the entire set. We saw that though ing the PA of the binary HD 37013,1 which was observed 5 the quality of the5 PSFs was good, the second half of the with LMIRcam the night before. The error was deter- data suffered from bright extended streaks appearing in mined by independently calculating the offset with two Fig. 1.—False-color, log scale images of thethe raw data, HD out to several arcseconds. 15115 These streaks diskdifferent software as routines originally and measuring the difference. discovered using the ACS/HRC F606W (l p 591 nm,Dl p 234 nm; left)and 0.5 0.5 were not caused by the spider arms,0.5 and were not sym- The PSF-subtracted images were0.5 then median-combined. 3 c metric, so they could not be easily masked out and would The final image, after smoothing by a Gaussian kernel 3 confirmed in H band with Keck II4 adaptive optics4 (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use 0 0 0 0 3 arcseconds 3 arcseconds gray fields overarcseconds the occulting bar and a 3.0Љ occulting spot located toarcseconds the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding −0.5 −0.5 2 2 HD 15115 itself.−0.5 If the HD 15115 disk were a symmetric structure,− then0.5 the east side of the disk would have been detected within the rectangular box, shown 2 2 −1 −1 −1 −1 Љ below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk1 within a 2 radius,1 with asymmetry becoming more apparent beyond 1 −1.5the 2Љ radius.−1.5 Due to poor observing conditions, the1 field−1.5 is contaminated−1.5 by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand −2 −2 −2 0 0 −2 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of0 the telescope0 rotates relative to the sky orientation over a series of −2.5 −2.5 −2.5 −1 −2.5 exposures,−2.5 −2 −1.5 −1 the−0.5 image0 0.5 1 of1.5 the2 disk2.5 remains fixed and−1 it is−2.5 confirmed−2 −1.5 −1 as−0.5 real. 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel L86 KALAS, FITZGERALD, & GRAHAM Vol. 661

TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Stellar Properties radius. The appearance of the disk is more symmetric in the Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). Spectral type ...... F2 M0 Hipparcos Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an ϩ2.22 ϩ4.38 Distance (pc) ...... 44.78Ϫ2.01 40.54Ϫ3.61 Hipparcos edge-on orientation to the line of sight. The west midplane is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Ϫ1 cluding a characteristic width asymmetry (Kalas & Jewitt 1995; Hipparcos 4.46 ע 82.32 1.09 ע ma(mas yr )...... 86.09 Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 Tycho-2 the full width at half-maximum across the disk midplane at 2Љ 2.7 ע Ϫ55.1 1.2 ע md (mas yr )...... Ϫ50.9 Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 a ever, the vertical cuts are not symmetric about the midplane when… 0.5 ע Ϫ10.0 2.3 ע W (km sϪ1)...... Ϫ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively. HWQM north of the west midplane is1.6 (2006) than that of the HWQM south of the west midplane. This width resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry the 2007 January data therefore yields a detection of the west is found to be in the opposite direction in the opposite midplane, ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- 3. RESULTS L86 KALAS, FITZGERALD, & GRAHAM Vol. 661 dence for significant flattening inward toward the star (Fig. 3). All TABLE 1 to the edge of the field, no nebulosity is detected 9.0Љ–14.9Љ Figure 1 showsStellar Properties the PSF-subtractedradius. The appearance images of the disk ofis more HD symmetric 15115 in the four surface brightness profiles are well represented by a single Parameter HD 15115 HIP 12545 Reference 2006 October Keck data, which show the disk between 0.7Љ (31 AU) and 2.5Љ (112 AU). with HSTSpectral typeand...... with F2 Keck. M0 Hipparcos The west side of the disk in the power-law decrease with radius. If there is an inner dust depletion, Optical surface brightness contours (Fig. 2) reveal a sharp mV (mag) ...... 6.79 10.28 Hipparcos

Mass (M, ) ...... 1.6 0.5 Cox (2006) midplane morphology for the west extension that indicates an then it resides within a 40 AU radius. This constraint is consistent ע Distance (pc) ...... 44.78ϩ2.22 40.54ϩ4.38 pHipparcos Њ optical HST data hasϪ2.01 P.A.Ϫ3.61 278 .5edge-on orientation0.5 to and the line of is sight. detected The west midplane from is R.A. (ICRS) ...... 02 26 16.2447 02 41 25.89 Hipparcos qualitatively similar to that of b Pic’s northeast midplane, in- Decl. (ICRS) ...... ϩ06 17 33.188 ϩ05 59 18.41 Hipparcos Hipparcos cluding a characteristic width asymmetry (Kalas & Jewitt 1995; with model fits of the spectral energy distribution that place the 4.46 ע 82.32 1.09 ע m (mas yrϪ1)...... 86.09 the edgea of the occulting spot at 1.5Љ (67 AU) radius to the Ϫ1 Hipparcos Golimowski et al. 2006). The northern side of the west midplane 2.45 ע Ϫ55.13 0.71 ע md (mas yr )...... Ϫ50.13 Ϫ1 ,Tycho-2 is more vertically extended than the southern side. For example 4.3 ע 82.3 1.2 ע ma(mas yr )...... 87.1 Ϫ1 2.7Љ (554 Tycho-2 AU)the full radius. width at half-maximum The across east the disk midplane midplane at 2Љ dominant emitting dust component at 35 AU radius (Zuckerman ע 12.38Ϫ55.1 1.2 עedge ofmd the(mas yr )...... fieldϪ50.9 at Ϫ1 a -0.10Љ in both the optical and NIR data. How ע radius is 0.19Љ … 0.5 ע Ϫ14.0 1.9 ע U (km s )...... Ϫ13.2 ∽ a… 0.9 ע Ϫ16.7 1.2 ע V (km sϪ1)...... Ϫ17.8 .(Љ0.5(315 …a AU)ever, the radius. vertical cuts are not At symmetric this about the radius midplane when the &Song2004;Williams&Andrews2006עas Ϫ10.07 2.3 ע is detectedW (km sϪ1)...... as farϪ6.0 measuring the half-width at quarter-maximum (HWQM). The a Galactic kinematics for HD 15115 and HIP 12545 from Moo´r et al. times greater 0.1 ע and Song et al. (2003), respectively.∼ HWQM north of the west midplane is1.6 (2006) east midplane begins to intersectthan the that of outer the HWQM south portion of the west midplane. of This the width co- The color of the disk may be estimated in the 2.0"" –3.3 "" resulting in disk self-subtraction at small radii. Our analysis of asymmetry is confirmed in the Keck data. If the width asymmetry ronagraph’sthe 2007 January 3.0 dataЉ thereforeocculting yields a detection of spot. the west Furtheris found to be in the east, opposite direction past in the the opposite spot midplane, and region where the H-band and V-band data overlap (Fig. 3). At ansa in the region 1.3Љ–3.3Љ radius. The photometry in this then Kalas & Jewitt (1995) refer to such a feature as the butterfly second epoch agrees well with that of the first epoch (on av- asymmetry. The butterfly asymmetry is evident in the mor- erage, the 2007 January disk photometry is 0.13 mag phology of b Pic that Golimowski et al. (2006) recently related arcsecϪ2 fainter than 2006 October), suggesting that our frame to the presence of a second disk midplane tilted relative to the selection technique for the first epoch of cloudy conditions main disk midplane. However, our detection of HD 15115’s east effectively filtered out nonphotometric data. midplane has insufficient signal-to-noise to confirm the presence of a width asymmetry here. We note that none of the surface brightness profiles show evi- Observations3. RESULTS with denceLBT for significant AO: flattening inward new toward the starKs (Fig. 3). and All L’ images 4 4 Figure 1 shows the PSF-subtracted images of HD 15115 Rodigasfour surface brightness et al. profilesRodigas are well represented et al. by a single with HST and with Keck. The west0.6 side µm of the disk1.1 in the µmpower-law 1.65 decrease µm with radius. If2.15 there is an µm inner dust depletion,3.8 µm PISCES) and is detected from then it resides within a 40 AU radius. This3.8 constraint µm is (LBTI/LMIRCam) consistent) 0.5 עoptical HST data hasP.A. p2.15278Њ.5 µm the edge of the occulting spot at 1.5Љ (67 AU) radius to the with model fits of the spectral energy distribution that place the edge of the field at 12.38Љ (554 AU) radius. The east midplane dominant emitting dust component at 35 AU radius (Zuckerman 2 2 2.5 2.5 2.5 ∼ 2.5 is detected as far as 7Љ (315 AU) radius. At this radius2.5 the &Song2004;Williams&Andrews2006).2.5 ∼ east midplane begins to intersect the outer portion of the co- The color of the disk may be estimated in the 2.0"" –3.3 "" 2 ronagraph’s2 3.0Љ occulting spot. Further east, past the spot and region where the H-band and2 V-band data overlap (Fig. 3).2 At 2 2 1.5 1.5 1.5 1.5 1.5 1.5

1 1 1 1.5 1.5 1 L42 DEBES, WEINBERGER, & SONG Vol. 684 1 1 0.5 0.5 0.5 0.5 4 Rodigas4 et al. Rodigas et al. 1 1

2 2 2.5 2.52.5 2.5 0 0 2.5 0 2.5 0

2 2 2 2

2 21.5 1.5 1.5 1.51.5 1.5 0.5 0.5 arcseconds arcseconds arcseconds arcseconds 1 1 1 1.5 1 1.5 −0.5 −0.5 0.5 0.5 −0.5 1 −0.5 1 0.5 0.50.5 0.5 1 1 0 0 0 0 0.5 0.5 arcseconds arcseconds arcseconds arcseconds −0.5 0.5 −0.5−0.5 0.5 −0.5

−1 −1 −1 −1−1 −1 −1 −1 0 00 0 −1.5 0 −1.5−1.5 0 −1.5 −0.5 −2 −2−2 −0.5 −2 0 0 −0.5 −0.5 −2.5 −2.5−2.5 −2.5 −1.5 −1 −1.5 −1 −1.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5−2.5 −2−2 −1.5−1.5 −1−1 −0.5−0.5 00 0.50.5 11 1.51.5 22 2.52.5 −2.5 −2 −1.5 −1 −0.5 0 −1.50.5 1 1.5 2 2.5 arcseconds arcsecondsarcseconds arcseconds Fig. 1.—Top: False-color images of the observations of HD 15115 in the (a) F110W filter at two spacecraft orientations.(a) The(a) images are logarithmically (a) scaled. Bottom: False-color, final combined image of the HD 15115 disk, −0.5 showing the contaminating feature−0.5 A. The dashed line represents the nominal −2 −2 position angle of the disk at 278.9Њ, while the asterisk represents−2 the position −2 8 of the star. We have masked out separations in the disk within 0.68Љ, where8 2.5 2.5 PSF subtraction2.5 residuals dominate. Noticeable are a warp on the western lobe, Fig. 2.—Measurements2.5 of the HD 15115 disk midplane PA (top) and FWHM 5 5 7 and the asymmetries between the western and eastern lobes. −0.5 2 2 2 7 (bottom) as a function of2 distance for the two lobes of the disk. Squares −0.5 represent the western lobe, while triangles represent the eastern lobe. The 1.5 6 disk emission1.51.5 to measure the location of the midplane (and6 sharp rise in PA for the western1.5 lobe interior 1.7Љ is due to the warp, and the −2.5 −2.5 4 4 −2.5 thus the local PA) as well as the FWHM of the disk. We took eastern lobe’s PA beyond 1.8Љ is marginally different− than2.5 the western lobe. 1 1 1 [See the electronic edition1 of the Journal for a color version of this figure.] −1 5 an average of the values−1 for both orientations and estimated 5 0.5 0.5 0.5 3 0.5 3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 4 the uncertainty as being half the difference between− the2.5 two −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 4 # 0.23Љ apertures centered on the midplane of the− disk.2.5 Be- −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 0 measurements.0 0 Figure 2 shows the PA and FWHM of the west- 0 3 tween 0.7Љ and 1.8Љ, the disk SB of the western lobe falls as arcseconds arcseconds ern lobe of thearcseconds disk as a function of radius. Beyond 1.7Љ, we3 arcseconds arcseconds 0.06עarcseconds Ϫ3.56 0.1עarcseconds arcseconds 2 Ϫ1.4 Fig. −0.5 −0.5−0.5 l p rr, whileDl at radii−p0.5 11.8Љ, the SB drops as . The 2 left)and 0.06ע;F606WЊ,whichisconsistent (c 591 nm,234Ϫ2.14 nm 0.2עFalse-color, log scale images of the HD 15115 disk as originally discovered usingfind a PA the for the ACS/HRC disk of 278.9Њ—.1 2 2 eastern lobe drops steadily asr from 0.7Љ to 3Љ. 0.5Њ.However,interiorto −1עwith KFG07’s−1−1 value of 278.5Њ −1 confirmed in H band with Keck II adaptive optics (right;2006October26data).Northisup,eastisleft,andthescalebarsspan51 Љ. In the HST image we use 1.7Љ, we note that the disk PA increases with decreasing dis- 1 1 1Њ at 0.68 PSF-subtractionЉ1. 3.3. Scattering−1.5 artifacts Efficiency vs. surrounding Wavelengthעgray fields over the occulting bar and a 3.0Љ occulting spot−1.5 located to the left of HD 15115,tance as from well the−1.5−1.5 star as to a a graymaximum disk PA of 290 coveringЊ −2 0 The changing−2− PA2 of the midplane is seen at both orientations. −2 00 Combining our data with the reported V and H SB data, a 0.04Љ which 0 ע Furthermore, we find at 2Љ a FWHM of 0.29Љ HD 15115 itself. If the HD 15115 disk were a symmetric−2.5 structure, then the east side of the disk−2.5−2.5 would have been detected withinmeasure the of the rectangular scattering−2.5 efficiency box, of the dust show aroundn HD −1 0.11.5Љ reported2 2.5 by−1 1ע is consistent with −2.5 the−2 − FWHM1.5 −1 −0.5 of 0.190 0.5Љ 2.5 2 1.5 1 0.5 0 −0.5 −1 −1.5 −2 −2.5 (a) −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 (a)2 2.5 (a) arcseconds arcseconds 15115 can be constructed. The measured SB at any radius in (a) KFG07. The median FWHM ofarcseconds the disk from 0.68Љ to 3.69Љ arcseconds below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk within a 2Љ radius, with asymmetry becomingthe disk can be self-consistently more apparent modeled for beyond composition and Љ Љ m The FWHM at 1.1 m is broadened by the grain size distribution provided a knowledge of the density . 0.07 ע is 0.26 (b) NICMOS F110W PSF, which has a FWHM(b) of 0.1Љ,implying the 2Љ radius. Due to poor observing conditions, the field is contaminated by residual noise due to the diffraction(b) pattern of thedistribution telescope of the (e.g., dust is known at 2 (see o’cloc details(b) inkand Debes et al. Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 an intrinsic FWHM for the disk! midplane of 0.24Љ or 11 AU. ! 2 Fig.Fig. 1.— 2.—TopTop::Final Final PISCESL imageKs ofband the∼ disk image obtained of the with HD LMIR-151152008). TheFig. SB of2.— aTop dust:Final disk atL image a particular of theradius disk obtained is with LMIR- 7 o’clock relative to HD 15115). However, whereasdebris the disk, residual in units of mJy/arcseconddiffraction, with pattern North-up, noise East-left. ofMeasurement the telescope of the eastern rotates side of2 the relative disk2 is complicated to the sky orientation over a series2 of debriscam, disk, in units in units of mJy/arcsecond of mJy/arcsecond, with, with North-up, North-up, East-left. East-left. TheF f(v)Q cam,(L /L in) units, where off mJy/arcsecond(v) is the phase, function with North-up, of the East-left. The The white dot marks the location of the star, and represents the by the contamination of feature A from PSF subtraction. To ! sca ! IR Thewhite white dot dot marks marks the locationthe location of the of star, the star,and represents and represents the sizedust∝ the of at a particularwhite dot scattering marks the angle, locationQ ofis the scatteringstar, and represents ef- the size of exposures, the image of the disk remains fixed and itsize is of a confirmed resolution element as at real.Ks band. For display purposes a measurea resolution the FWHM element and at PA,L!. we For masked display out purposes pixels heavilya 0!!. 9 radius mask ! sca !! !! size of a resolution element at Ks band. For display purposesficiency a of thea resolution dust, and elementL /L is at theL ratio. For of display stellar luminosity purposes a 0. 9 radius mask 0. 8 radius mask has been added in post-processing. The western contaminated0!!.has8 radius been by added mask this hasinfeature. post-processing. been We added again in took post-processing. The the image average has of been the The smoothed western ! IR The disk is morphologically∼ 2 different2 to the IRbetween luminosityhas been added of the in dust post-processing. distribution. This The is image difficult2 has been& smoothed 4 µm, side SB is a magnitude/arcsecond brighter than the eastern twosideby setsa SB of Gaussian is measurements∼ a kernelmagnitude/arcsecond with and estimated FWHM = theλbrighter/D uncertainties and binned than as the by eastern a factor by a Gaussian kernel with FWHM = λ/D and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! for HD 15115 because the density distribution! of the dust is sideof 4 SB, to bringas is seen out atthe shorter disk. At wavelengthsL the disk (Kalas is mostly et al. symmetric 2007; Debes and L et al. 2008). Bottom: SNRE8 map of the image. The eastern side is being half the difference between the two measurements. We of 4 to bring out the disk. At the disk is mostly symmetric and !! etis al. equally 2008). brightBottom on8: both SNRE sides. map These of the features image. The are nearly eastern opposite sidenot is knownis through, equallyfor bright example, on both spatially sides. These resolved features thermal are nearly opposite detected at SNRE ∼ 3outto∼ 2. 1. The western side of the disk findto no what evidence is seen ofa at∼ warpKs band in the and∼ eastern!! at shorter lobe, which wavelengths. has a Bottom: 2.5 !! detected at SNRE 3outto! 2. 1. The2.5 western side of theemission disk images.to what However, is seen multiwavelengthat Ks band and imaging at shorter can wavelengths. pro- Bottom: final∼1.3LЊ. Toimage compare∼ shown!! to above. the PA The of disk the is detected ! 2.5 עis detected at SNRE ∼ 3-8 from ∼ 1-2. 5. medianSNRE PA map of 96.6 of theЊ 2.5 is detected at SNRE 3-8 from!! 1-2. 5. vide usefulSNRE compositional map of the information final L image even in shown the absence above. ofThe disk is detected western side we∼ add 180Њ to∼ get 276.6Њ, consistent with no !! at SNRE 2.5 between 1-1. 5 on both sides. spatially resolvedat SNRE thermal∼ 2.5 emission between constraints.∼ 1-1. 5 on In both the case sides. of and between 0.5-1 & 2-4significant µm bowing in the large-scale structure of the disk as is of data, images were averaged in sets of 4. We registered of data, images were averaged in sets of 4. We registeredHD 15115, three images at different wavelengths will be in- 5 5 the images to a fiducial pixel by calculating the center seenoverwhelm for HD 61005 the (Hines brightness et al. 2007). of anyThe median disk structure value of seensufficient in to provide a definitive picture, but can constrain cer- pixelЉ. by calculating the center overwhelm the brightness of any disk structure seen in 0.06 ע thethe eastern images FWHM to is a 0.297 fiducialЉ 7 2 2 of light around the star, excluding the saturated pixels ofthe light final around image. the Therefore star, excluding we used the2 only saturated the first pixelstain half compositions.the final image. Therefore we2 used only the first half ∼ ∼ ∼ ◦ In order to compare results across different wavelengths, within λ/D, in the same manner as for the Ks band of the∼ data3.2. (Surface40 minutes Brightness of Profiles integration, with 50 of ∼ ∼ ◦ within λ/D, in the same manner as for the Ks bandinstruments,of and the telescopes, data ( it40 is imperative minutes thatof integration, a careful and with 50 of images. We performed similar center of light calcula- images.parallactic We angle performed rotation). similar center of light calcula- parallactic angle rotation). tions on the unsaturated images of HD 15115 and found FigureAfter 3 shows scaling a comparison and subtracting of the 1.1 m (inm SB the profiles same in mannerconsistent as treatment of SB measurements is implemented. Ob- Dmmtionsarcsec on theϪ2 relative unsaturated to the central images star of ( HD 15115p 6.12) and foundservations of disksAfter at scaling different and wavelengths subtracting can be affected(in the by same manner as the difference between the measured and true centroid to F110Wfor the Ks band images) the master F110W PSF from each im- 1.5 1.5 ∼ 6 forthe the dieasternfference and westernbetween lobes the of measured the disk by1.5 and taking true 0.23 centroidЉ the to respectivefor PSFs the ofKs telescopesband images) with different the1.5 diffraction master PSF lim- from each im- be 0.2 pixels (2 mas). We then calculated the radial beage∼ in0.2 the pixels first6 (2 half mas). of the We data then (240 calculated averaged the images; radial age in the first half of the data (240 averaged images;Rodigas/NCAD/July 20, 2012 profile of each image and subtracted it, also in the same profile960 total), of each we image rotated and each subtracted image by it, its also parallactic in the same an- 960 total), we rotated each image by its parallactic an- 4 4 manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images.◦ To test the◦ quality of gle plus an offset to obtain North-up, East-left. The the data, we median-combined the first half of the data, offset was determined to be 1.81 ± 0.0685 by calculat- ◦ ◦ 1 the data, we median-combined the first1 half of the data, offset was determined to be 1.81 ± 0.0685 by calculat- 1 the second half, and the entire set. We saw that though theing second the PA half, of the and binary the entire HD 37013, set. We which saw was that observed though ing the PA of the binary HD 37013,1 which was observed the quality of the PSFs5 was good, the second half of the thewith quality LMIRcam of the5 the PSFs night was before. good, the The second error half was of deter- the with LMIRcam the night before. The error was deter- data suffered from bright extended streaks appearing in datamined suff byered independently from bright extendedcalculating streaks the off appearingset with two in mined by independently calculating the offset with two Fig. 1.—False-color, log scalethe raw data, out toimages several arcseconds. These of streaks thethediff rawerent data, software HD out toroutines several and arcseconds. 15115 measuring These the diff streakserence. diskdifferent software as routines originally and measuring the difference. discovered using the ACS/HRC F606W (l p 591 nm,Dl p 234 nm; left)and were not caused by the spider arms, and were not sym- The PSF-subtracted images were then median-combined. 0.5 0.5 were not caused by the spider arms,0.5 and were not sym- The PSF-subtracted images were0.5 then median-combined. 3 c metric, so they could not be easily masked out and would metric,The final so they image, could after not smoothing be easily masked by a Gaussian out and would kernel The final image, after smoothing by a Gaussian kernel 3 confirmed in H band with Keck II4 adaptive optics4 (right;2006October26data).Northisup,eastisleft,andthescalebarsspan5Љ. In the HST image we use 0 0 0 0 3 arcseconds 3 arcseconds gray fields overarcseconds the occulting bar and a 3.0Љ occulting spot located toarcseconds the left of HD 15115, as well as a gray disk covering PSF-subtraction artifacts surrounding −0.5 −0.5 2 2 HD 15115 itself.−0.5 If the HD 15115 disk were a symmetric structure,− then0.5 the east side of the disk would have been detected within the rectangular box, shown 2 2 −1 −1 −1 −1 Љ below the ACS/HRC occulting finger. The NIR data (right) show a more symmetric disk1 within a 2 radius,1 with asymmetry becoming more apparent beyond 1 −1.5the 2Љ radius.−1.5 Due to poor observing conditions, the1 field−1.5 is contaminated−1.5 by residual noise due to the diffraction pattern of the telescope (e.g., at 2 o’clockand −2 −2 −2 0 0 −2 7 o’clock relative to HD 15115). However, whereas the residual diffraction pattern noise of0 the telescope0 rotates relative to the sky orientation over a series of −2.5 −2.5 −2.5 −1 −2.5 exposures,−2.5 −2 −1.5 −1 the−0.5 image0 0.5 1 of1.5 the2 disk2.5 remains fixed and−1 it is−2.5 confirmed−2 −1.5 −1 as−0.5 real. 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds arcseconds arcseconds

(b) (b) (b) (b) Fig. 1.— Top : Final PISCES Ks band image of the HD 15115 ! ! Fig. 1.— Top : Final2 PISCES Ks band image of theFig. HD 2.— 15115Top :FinalFig.L 2.—imageTop of:Final the diskL obtainedimage of with the disk LMIR- obtained with LMIR- debris disk, in units of mJy/arcsecond , with North-up, East-left.2 2 2 debris disk, in units of mJy/arcsecond , with North-up,cam, East-left. in units of mJy/arcsecondcam, in units of, with mJy/arcsecond North-up, East-left., with North-up, The East-left. The The white dot marks the location of the star, and represents the The white dot marks the location of the star, and representswhite dot the marks thewhite location dot marks of the star,the location and represents of the star, the size and of represents the size of size of a resolution element at Ks band. For display purposes a ! ! !! !! !! size of a resolution element at Ks band. For displaya resolutionpurposes a elementa atresolutionL . For displayelement purposes at L . For a 0 display. 9 radius purposes mask a 0. 9 radius mask 0. 8 radius mask has!! been added in post-processing. The western 0. 8 radius mask has2 been added in post-processing.has The been western added in post-processing.has been added in The post-processing. image has been The smoothed image has been smoothed side SB is ∼ a magnitude/arcsecond brighter than the2 eastern side SB is ∼ a magnitude/arcsecond brighter thanby the a Gaussian eastern kernelby with a Gaussian FWHM kernel = λ/D with and FWHM binned = byλ a/D factor and binned by a factor side SB, as is seen at shorter wavelengths (Kalas et al. 2007; Debes ! ! side SB, as is seen at shorter wavelengths (Kalas et al.of 2007; 4 to Debes bring out theof disk. 4 to bring At L outthe the disk disk. is mostly At L symmetricthe disk is and mostly symmetric and et al. 2008). Bottom: SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite et∼ al. 2008).∼Bottom!! : SNRE map of the image. The eastern side is is equally bright on both sides. These features are nearly opposite detected at SNRE 3outto 2. 1.∼ The western∼ side!!. of the disk to what is seen at Ks band and at shorter wavelengths. Bottom: detected∼ at SNRE∼ !! 3outto 2 1. The western side of the disk to what! is seen at Ks band and at shorter wavelengths. Bottom: is detected at SNREis detected3-8 from at SNRE1-2. ∼5. 3-8 from ∼ 1-2!!. 5. SNRE map of the final L image shown above.! The disk is detected ∼ SNRE∼ map!! of the final L image shown above. The disk is detected at SNRE 2.5 betweenat SNRE1-1∼.2.55 on between both sides.∼ 1-1!!. 5 on both sides. of data, imagesof were data, averaged images in were sets averaged of 4. We in registered sets of 4. We registered the images to athe fiducial images pixel to a by fiducial calculating pixel the by calculatingcenter overwhelm the center the brightnessoverwhelm of the any brightness disk structure of any seen disk in structure seen in of light around the star, excluding the saturated pixels the final image. Therefore we used only the first half ∼ of light around the star, excluding the saturated pixels ∼ the final image. Therefore we∼ used◦ only the first half within λ/D,within in the same∼ λ/D, manner in the as same for the mannerKs band as for theofKs theband data ( 40of minutes the data of (∼ integration,40 minutes with of integration,50 of with ∼ 50◦ of images. We performedimages. similarWe performed center of similar light calcula- center of lightparallactic calcula- angleparallactic rotation). angle rotation). tions on the unsaturatedtions on the images unsaturated of HD 15115 images and of found HD 15115 andAfter found scaling andAfter subtracting scaling and (in the subtracting same manner (in the as same manner as the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- ∼ the difference between the measured and true centroid to for the Ks band images) the master PSF from each im- be 0.2 pixelsbe (2∼ mas).0.2 pixels We then (2 mas). calculated We then the radial calculatedage the in radial the firstage half in of the the first data half (240 of averaged the data images; (240 averaged images; profile of each imageprofile and of each subtracted image andit, also subtracted in the same it, also in960 the total), same we rotated960 total), each we image rotated by its each parallactic image by an- its parallactic an- manner as for the Ks band images. To test the quality of gle plus an offset to obtain North-up, East-left. The manner as for the Ks band images. To test the quality of gle plus an offset◦ ± to obtain◦ North-up, East-left. The the data, we median-combinedthe data, we median-combined the first half of the data,first half ofoff theset data, was determinedoffset was to be determined 1.81 0.0685 to be 1.81by calculat-◦ ± 0.0685◦ by calculat- the second half,the and second the entire half, set. and We the saw entire that set. though We saw thating thoughthe PA of theing binary the PA HD of 37013, the binary which HD was 37013, observed which was observed the quality of thethe PSFs quality was of good, the PSFs the second was good, half ofthe the secondwith half of LMIRcam the with the night LMIRcam before. the The night error before. was deter- The error was deter- data suffered fromdata bright suffered extended from bright streaks extended appearing streaks in appearingmined by in independentlymined by calculating independently the o calculatingffset with two the offset with two the raw data, outthe to raw several data, arcseconds. out to several These arcseconds. streaks Thesedifferent streaks softwarediff routineserent software and measuring routines the and di measuringfference. the difference. were not causedwere by the not spider caused arms, by the and spider were arms, not sym- and wereThe not PSF-subtracted sym- The PSF-subtracted images were then images median-combined. were then median-combined. metric, so they couldmetric, not so be they easily could masked not be out easily and masked would outThe and wouldfinal image,The after final smoothing image, after by a smoothing Gaussian kernel by a Gaussian kernel Large Grains in the HD 15115 Debris Disk 9

2009)). Because HD 15115 is moving to the south-east, 1 5 nearly parallel with its disk major axis, one would not 4 0.5 expect the disk to have a bow-shape, especially with an 3 6 Rodigas et al. offset from the star to the north. However, in the HD 0 2 61005 debris disk system, Maness et al. (2009) showed

arcseconds 1 0.3 −0.5 99 that dust grain interactions with the ISM in directions Ks East 0 Ks East 98.5 along the disk midplane could actually create bow-shapes −1 −1 0.25 −2.5 −2 −1.5 −1 −0.5 0 0.598 1 1.5 2 2.5 perpendicular to the disk major axis, roughly reproduc- arcseconds Disk Morphology: an 97.5apparent bow-shapeing that disk’s observed morphology. Similar dynamical 0.2 97 1 5 interactions in the HD 15115 system could be creating 96.5 4 the observed bow-shape, though no explicit model has 0.15 0.5 96 3 tested this hypothesis for this system. FWHM (arcseconds)

Position Angle (degrees) 95.5 0 2 Another perhaps simpler explanation for the bow- 0.1 95

arcseconds 1 shape and the apparent offset is that these are caused by −0.5 94.5 0 geometrical effects. We can reproduce the observed bow- 0.05 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 1.2 1.4 1.6 1.8 2 2.2 2.4 distance from star−1 (arcseconds) distance from star (arcseconds)−1 shape and vertical offset with a simple inclined, ringed −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 6 Rodigas et al. (a) arcseconds (a) disk model. We used the radiative transfer equations de- Disk Position Angle vs. distance from star scribing the intensity of light scattering off dust particles 0.3 99 (a) and made three assumptions about the disk: it is not Ks East 0.26 Ks East Ks West Ks West 286 98.5 ◦ 0.24 exactly edge-on, but rather inclined to 87 ; it has a gap 0.25 98 East West !! 0.22 285 from 0-1. 1(seeSection4.3foradditionaldiscussionof 97.5 280 0.2 284 the possible disk gap); and we set the Henyey-Greenstein 0.2 0.18 97 scattering parameter g =0.5forpredominantlyforward- 0.16 96.5 279 283 scattering grains. This is a reasonable assumption for 0.15 0.14 96 282

FWHM (arcseconds) !

FWHM (arcseconds) large ( 1 µm) grains scattering NIR light (Duchˆene 0.12 278

Position Angle (degrees) 95.5 Position Angle (degrees) et al. 2004). We set the dust grain size uniformly to 3 0.1 0.1 95 281 0.08 94.5 277 µmandsetthediskSBpower-lawindexto-3/2.Fig.7a 280 0.06 shows the rotated model image, along with the rotated 0.05 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 1 1.21.5 1.4 1.6 2 1.8 2 2.5 2.2 2.4 1 1.5 2 2.5 distance from star (arcseconds) distance from276 stardistance (arcseconds) from star (arcseconds) distance from star (arcseconds) final Ks band image. Dashed lines have been inserted to (a) (b) (a) (b) guide the eye to the apparent bow-shape and northern Fig.The 3.— TopPA: Ks forband easternan275 un-bowed disk FWHM as a function edge-on of Fig. disk 4.— Top :should PA of the eastern remain side of the disk ~constant at Ks bandoffset from the star. Fig. 7b shows the PA vs. stellocen- distance from the star. See text for methodology. There is no as a function of distance from the star. See text for methodology. obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PA increases with increasing distance from the star; over the Position Angle (degrees) tric distance for the western side of the disk in the model 0.26 Ks West as above, except for the western side of the disk. The FWHMKs West same stellocentric distances a similar trend was observed at 1.1 286 274 0.24 increases with increasing distance from the star, though this may µm (Debes et al. 2008). Bottom: Same as above,Rodigas/NCAD/July except for theimage, 20, 2012 computed in the same manner as for the real data. be an effect from the data reduction (see Appendix). western side. The PA increases closer to the star, as is seen in the 0.22 285 Debes et al. (2008) 1.1 µmbanddata. The PA increases closer to the star, as is observed in the to avoid any potential wavelength-dependent273 changes in 0.2 2008). The western PA increases closer to the star. This PA. Comparing284 our final PISCES Ks band image with real data (Fig. 4b). !! also agrees with the western PA vs. stellocentric distance 0.18 the Keck image between 1-2 revealed a difference in ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverallWhile the parameters used in our model are not a 0.16 overall PA of 3.9283 ,sowerotatedourPISCESimagebyan272 ◦ ◦ 0.8 1 1.2 1.4 1.6PA (278.51.8 )measuredforthewesternsideofthediskat2 2.2 2.4 2.6 additional 3.9 clockwise. Fig. 1a shows this correctly- !! unique explanation for the observed features of the disk, 0.14 distance from0.6 starµmbetween1.5-12 (arcseconds) (Kalas et al. 2007). rotated image.282 !

FWHM (arcseconds) 0.12 We calculated the PA of the disk at L in the samethe proposition that the effects can be explained by the

To test for anyPosition Angle (degrees) warps in the disk, we calculated the PA manner as for the Ks band image, except we used 3 of the disk at Ks band on both sides of the disk as a func- 0.1 281 pixel(b) by 11 pixel (∼ 0!!. 13 by 0!!. 47) boxes. The L! imagedisk’s geometric orientation in space is attractive because tion of distance from the star, using the disk midplane 0.08 was again binned by a factor of 4. We do not plot PA vs. pixel locations from the Gaussian fits. The errors were it explains the observations without contradicting the 280 Fig. 7.— Top : Rotated modeldistance and from real theKs starband due to disk the poor images. spatial coverage 0.06 measured as the difference between these values and the ◦ The model image has an inclinationof the disk of and, 87 , even scattering after binning, parameter the low SNRE (∼evidence2- supporting the ISM interaction interpretation 1 1.5 2 2.5center of light along1 each 1D row1.5 at each discrete2 pixel 2.5 distance from star (arcseconds) distance from star (arcseconds) 3); however the measured values for both sides of the distance from the star.g =0 The.5, errors 3 µm also grains, included a SB the power-law 1 index of -3/2, and a gap from (Debes et al. 2009). !! "" disk were globally consistent with the values measured pixel = 0. 194 astrometric. centroiding uncertainty. Fig. (b) 0-1 1. The model(b) image has beenat Ks convolvedband (this is with also evident the PSF by examination from of the 4a and Fig. 4b show the Ks band PA of the eastern and ! 4.2. Fig. 3.— Top : Ks band eastern disk FWHM as a function of Fig. 4.— theTop : PAphotometric of the eastern side image of the ofdisk HBC at Ks bandKs 388,band and and GaussianL disk images noise in Fig. has 1abeen and Fig. 2a, Disk Color and Grain Sizes western sides of the disk as a function of distance from 2 distance from the star. See text for methodology. There is no as a function of distance from the star. See text for methodology.respectively). the star, respectively.added. The eastern Units PA are increases in mJy/arcsecond with in- for both images. The white obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PA increasesdot markswith increasing the distance location from of the the star; star over the in both images. The dashed lines To determine the disk color as a function of distance as above, except for the western side of the disk. The FWHMcreasing distancesame stellocentric from the distances star, similar a similar to trend what was is seenobserved at 1.1 3.3. Surface Brightness Profiles ! ! increases with increasing distance from the star, though thisover may the sameµm (Debes spatial etare al. locations meant2008). Bottom at to 1.1: guideµ Samem(Debesetal. as the above, eye except to for the the apparent northern offset of the be an effect from the data reduction (see Appendix). western side. The PA increases closer to the star, as is seen in the from the star, we calculated ∆(Ks - L )=(Ks - L )Disk Debes et al.disk (2008) relative 1.1 µmbanddata. to the star. Bottom: PA vs. stellocentric distance for ! to avoid any potential wavelength-dependent changes in -(Ks - L )Star in the regions of spatial overlap between 2008). Thethe western western PA increases side of closer the to disk the star. in the This model image. The PA increases ! !! PA. Comparing our final PISCES Ks band image with the Ks band and L images (1.1-1. 45). The errors were !! also agrees with the western PA vs. stellocentric distance the Keck image between 1-2 revealed a difference in closer to the star due to the apparent bow-shape, similar to what ! ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverall calculated by summing the individual Ks and L errors overall PA of 3.9 ,sowerotatedourPISCESimagebyan ◦is seen for the real data shown in Fig. 4b. ◦ PA (278.5 )measuredforthewesternsideofthediskat additional 3.9 clockwise. Fig. 1a shows this correctly- 0.6 µmbetween1.5-12!! (Kalas et al. 2007). in quadrature. The data suggest that the eastern side rotated image. We calculated the PA of the disk at L! in the same To test for any warps in the disk, we calculated the PA and 2a), as well as the PA vs. stellocentric distance plots of the disk becomes redder than the western side with manner as for the Ks band image, except we used 3 of the disk at Ks band on both sides of the disk as a func- !! pixel by 11for pixel both (∼ 0!!. 13 sides by 0!!. 47) of boxes. the The diskL! image (Fig. 4a and 4b) suggests that increasing distance from the star. Interior to 1. 3both tion of distance from the star, using the disk midplane was again binned by a factor of 4. We do not plot PA vs. pixel locations from the Gaussian fits. The errors were the debris disk has a bow-like shape and is offset by a few sides of the disk are grey. distance from the star due to the poor spatial coverage measured as the difference between these values and the of the diskAU and, even to after the binning, north the from low SNRE the (∼ star.2- Specifically the offsets To constrain the characteristic dust grain sizes, we center of light along each 1D row at each discrete pixel 3); however the measured!! values for both sides of!! the ! ! distance from the star. The errors also included the 1 are ∼ 0. 012 (0.5 AU) and 0. 04 (2 AU) at L ,fortheeast- compared our ∆(Ks - L )diskcolorstothemodelsof !! disk were globally consistent with the values measured pixel = 0. 194 astrometric centroiding uncertainty. Fig. !! at Ks band (this is also evident by examination of the ∼ 4a and Fig. 4b show the Ks band PA of the eastern and ern and western sides, respectively; and 0. 11 (5 AU) Inoue et al. (2008). These models calculate disk colors Ks band and L! disk!! images in Fig. 1a and Fig. 2a, western sides of the disk as a function of distance from respectively).and 0. 17 (8 AU) at Ks band for the eastern and western in the NIR and assume silicate dust grain composition. the star, respectively. The eastern PA increases with in- sides, respectively. Since the astrometric uncertainty in The modeled grain sizes range from 0.1 µmto10µmand creasing distance from the star, similar to what is seen 3.3. Surface Brightness Profiles!! !! ! over the same spatial locations at 1.1 µm(Debesetal. centroiding is ∼ 0. 002 and 0. 0194 at L and Ks band, assume a single grain size for the dust population. We respectively, it is unlikely that the northern offsets could plot the model disk colors for 1, 3, and 10 µmgrains. be due to centroid error. Bow-like disk morphologies These are shown as the colored lines in Fig. 8. We also have been seen in systems moving in near-perpendicular modeled 0.1 µmand0.3µmgrains,buttheircolorswere directions to their disk major axis positions (e.g., HD too blue to be supported by the data. 61005 (Hines et al. 2007) and HD 32297 (Debes et al. From Fig. 8, we see that 1-10 µm-sized grains are the Can ISM interactions explain the bow?

Rodigas/NCAD/July 20, 2012 Can ISM interactions explain the bow?

324 DEBES, WEINBERGER, & KUCHNERHD 61005 (“the moth”) Vol. 702

Figure 7. Comparison between an F110W image of HD 15117’s disk from Debes et al. 2009 Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly provided by G. Schneider) and a model of a disk interacting with a cloud of (A color version of this figure is available in the online journal.) ISM gas with the parameters mentioned in Section 4.1.3. (A color version of this figure is available in the online journal.) P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe sky is at roughly the same P.A. as the disk. If we again assume 1 the same assumptions we used above as for HD 15115, the a cloud with velocity of 14 km s− on the sky, the minimum 1 relative velocities will range in magnitude from 2 km s− to and maximum possible relative transverse velocity is between 1 1 30 km s− ,withavectorprimarilyalongthex–z plane at an 7and35kms− if the cloud’s velocity vector is parallel or inclination of 68 relative to the x-axis. If icy 0.3 µmgrainsare antiparallel to HD 15115’s . The morphology of ◦ the population we observe around HD 61005, we can derive Dg the disk is consistent with the midplane plowing headlong into 3 under the assumption of an ISM gas density N(H i) 100 cm− . an ISM flow. 18= 1 These two assumptions result in Dg 3.3 10 cm . We assumed that HD 15115 had a stellar mass of 1.5 M and = × − − # We experimented with various relative velocity vectors until that grains were launched from 35 AU, the rough inner radius of we found a qualitative match to the data. A velocity vector that HD 15115’s disk as inferred from its IR excess (Zuckerman & is nearly perpendicular to the disk midplane with a magnitude Song 2004; Williams & Andrews 2006). Experimenting with 1 18 1 of 25 km s− matches the data best, along with an inclination different models showed that Dg 1.3 10− cm− and = × between the observer and the disk of 20◦.Figure8 shows a Rodigas/NCAD/July 20, 2012 arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale midplane and along the observer’s line of sight with a magnitude 1 and stretch. We also see good qualitative agreement with the of 30 km s− best matched the observed data. morphology of this disk as well. Figure 7 shows a comparison of the F110W image of HD 15115 with our disk model on the same size scale. The 5. DISCUSSION eastern lobe (left side of Figure 7)iscontaminatedbyastrong residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the model matches the observation of HD 15115 showing a significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains figure shows that our models predict that the FWHM of the disk midplane should be wider on the truncated side. This aspect For a disk interacting with an atomic ISM cloud with small of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe observations (Debes et al. 2008a). Another proposed scenario radiation pressure of the central star was strong enough to scatter for this morphology is the close flyby of HD 15115 by another small grains away from a circumstellar disk. The distance rav depends on β, v , the initial velocity of a grain at large distances, star (Kalas et al. 2007). However, no definitive candidates for a ∞ flyby have currently been detected (Debes et al. 2008a). and the stellar mass M".

2(β 1)GM" 4.1.3. An ISM Gas Interaction Model for HD 61005 r − . (6) av = v2 The disk observed around HD 61005, the “Moth,” shows a ∞ peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum disk interaction with the gas component of the ISM. The star radius rav that is free from sandblasting from small ISM grains. itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may − 74.53 mas in (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for Can ISM interactions explain the bow?

324 DEBES, WEINBERGER, & KUCHNERHD 61005 (“the moth”) Vol. 702

Figure 7. Comparison between an F110W image of HD 15117’s disk from Debes et al. 2009 Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly provided by G. Schneider)For the and a model moth, of a disk interacting yes with a cloud of (A color version of this figure is available in the online journal.) ISM gas with the parameters mentioned in Section 4.1.3. (A color version of this figure is available in the online journal.) P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe sky is at roughly the same P.A. as the disk. If we again assume 1 the same assumptions we used above as for HD 15115, the a cloud with velocity of 14 km s− on the sky, the minimum 1 relative velocities will range in magnitude from 2 km s− to and maximum possible relative transverse velocity is between 1 1 30 km s− ,withavectorprimarilyalongthex–z plane at an 7and35kms− if the cloud’s velocity vector is parallel or inclination of 68 relative to the x-axis. If icy 0.3 µmgrainsare antiparallel to HD 15115’s proper motion. The morphology of ◦ the population we observe around HD 61005, we can derive Dg the disk is consistent with the midplane plowing headlong into 3 under the assumption of an ISM gas density N(H i) 100 cm− . an ISM flow. 18= 1 These two assumptions result in Dg 3.3 10 cm . We assumed that HD 15115 had a stellar mass of 1.5 M and = × − − # We experimented with various relative velocity vectors until that grains were launched from 35 AU, the rough inner radius of we found a qualitative match to the data. A velocity vector that HD 15115’s disk as inferred from its IR excess (Zuckerman & is nearly perpendicular to the disk midplane with a magnitude Song 2004; Williams & Andrews 2006). Experimenting with 1 18 1 of 25 km s− matches the data best, along with an inclination different models showed that Dg 1.3 10− cm− and = × between the observer and the disk of 20◦.Figure8 shows a Rodigas/NCAD/July 20, 2012 arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale midplane and along the observer’s line of sight with a magnitude 1 and stretch. We also see good qualitative agreement with the of 30 km s− best matched the observed data. morphology of this disk as well. Figure 7 shows a comparison of the F110W image of HD 15115 with our disk model on the same size scale. The 5. DISCUSSION eastern lobe (left side of Figure 7)iscontaminatedbyastrong residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the model matches the observation of HD 15115 showing a significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains figure shows that our models predict that the FWHM of the disk midplane should be wider on the truncated side. This aspect For a disk interacting with an atomic ISM cloud with small of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe observations (Debes et al. 2008a). Another proposed scenario radiation pressure of the central star was strong enough to scatter for this morphology is the close flyby of HD 15115 by another small grains away from a circumstellar disk. The distance rav depends on β, v , the initial velocity of a grain at large distances, star (Kalas et al. 2007). However, no definitive candidates for a ∞ flyby have currently been detected (Debes et al. 2008a). and the stellar mass M".

2(β 1)GM" 4.1.3. An ISM Gas Interaction Model for HD 61005 r − . (6) av = v2 The disk observed around HD 61005, the “Moth,” shows a ∞ peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum disk interaction with the gas component of the ISM. The star radius rav that is free from sandblasting from small ISM grains. itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may − 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for Can ISM interactions explain the bow?

324 DEBES, WEINBERGER, & KUCHNERHD 61005 (“the moth”) Vol. 702 324HD 15115 DEBES, WEINBERGER, & KUCHNER Vol. 702

Figure 7. Comparison between an F110W image of HD 15117’s disk from Debes et al. 2009 Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly Figure 7. Comparison between an F110W image of HD 15117’s disk from provided by G. Schneider) and a model of a disk interacting with a cloud of Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas (A color version of this figure is available in the online journal.) For the moth, yes ISM gas with the parameters mentioned in Section 4.1.3. with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly (A color version of this figure is available in the online journal.) provided by G. Schneider) and a model of a disk interacting with a cloud of (A color version of this figure is available in the online journal.) ISM gas with the parameters mentioned in Section 4.1.3. P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe sky is at roughly the same P.A. as the disk. If we again assume (A color version of this figure is available in the online journal.) 1 the same assumptions we used above as for HD 15115, the a cloud with velocity of 14 km s− on the sky, the minimum 1 P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe relative velocities will range in magnitude from 2 km s− to and maximum possible relative transverse velocity is between 1 sky is at roughly the same P.A. as the disk. If we again assume 1 30 km s− ,withavectorprimarilyalongthex–z plane at an 1 the same assumptions we used above as for HD 15115, the 7and35kms− if the cloud’s velocity vector is parallel or a cloud with velocity of 14 km s on the sky, the minimum 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare − relative velocities will range in magnitude from 2 km s to antiparallel to HD 15115’s proper motion. The morphology of − the population we observe around HD 61005, we can derive Dg and maximum possible relative transverse velocity is between 30 km s 1,withavectorprimarilyalongthex–z plane at an the disk is consistent with the midplane plowing headlong into 3 1 − under the assumption of an ISM gas density N(H i) 100 cm− . 7and35kms− if the cloud’s velocity vector is parallel or an ISM flow. 18= 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare These two assumptions result in Dg 3.3 10− cm− . antiparallel to HD 15115’s proper motion. The morphology of We assumed that HD 15115 had a stellar mass of 1.5 M and = × the population we observe around HD 61005, we can derive Dg # We experimented with various relative velocity vectors until the disk is consistent with the midplane plowing headlong into 3 that grains were launched from 35 AU, the rough inner radius of under the assumption of an ISM gas density N(H i) 100 cm− . we found a qualitative match to the data. A velocity vector that an ISM flow. 18= 1 HD 15115’s disk as inferred from its IR excess (Zuckerman & These two assumptions result in Dg 3.3 10 cm . is nearly perpendicular to the disk midplane with a magnitude We assumed that HD 15115 had a stellar mass of 1.5 M and = × − − Song 2004; Williams & Andrews 2006). Experimenting with of 25 km s 1 matches the data best, along with an inclination # We experimented with various relative velocity vectors until 18 1 − that grains were launched from 35 AU, the rough inner radius of different models showed that Dg 1.3 10− cm− and we found a qualitative match to the data. A velocity vector that = × between the observer and the disk of 20◦.Figure8 shows a Rodigas/NCAD/July 20, 2012 arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115’s disk as inferred from its IR excess (Zuckerman & is nearly perpendicular to the disk midplane with a magnitude midplane and along the observer’s line of sight with a magnitude Song 2004; Williams & Andrews 2006). Experimenting with 1 1 and stretch. We also see good qualitative agreement with the 18 1 of 25 km s− matches the data best, along with an inclination of 30 km s− best matched the observed data. morphology of this disk as well. different models showed that Dg 1.3 10− cm− and between the observer and the disk of 20 .Figure8 shows a Figure 7 shows a comparison of the F110W image of = × ◦ arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115 with our disk model on the same size scale. The 5. DISCUSSION midplane and along the observer’s line of sight with a magnitude eastern lobe (left side of Figure 7)iscontaminatedbyastrong 1 and stretch. We also see good qualitative agreement with the of 30 km s− best matched the observed data. morphology of this disk as well. residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- Figure 7 shows a comparison of the F110W image of et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the model matches the observation of HD 15115 showing a HD 15115 with our disk model on the same size scale. The 5. DISCUSSION significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains eastern lobe (left side of Figure 7)iscontaminatedbyastrong figure shows that our models predict that the FWHM of the disk residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- midplane should be wider on the truncated side. This aspect For a disk interacting with an atomic ISM cloud with small et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe of the model matches the observation of HD 15115 showing a radiation pressure of the central star was strong enough to scatter observations (Debes et al. 2008a). Another proposed scenario significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains for this morphology is the close flyby of HD 15115 by another small grains away from a circumstellar disk. The distance rav depends on β, v , the initial velocity of a grain at large distances, figure shows that our models predict that the FWHM of the disk star (Kalas et al. 2007). However, no definitive candidates for a ∞ For a disk interacting with an atomic ISM cloud with small and the stellar mass M". midplane should be wider on the truncated side. This aspect flyby have currently been detected (Debes et al. 2008a). of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe radiation pressure of the central star was strong enough to scatter 2(β 1)GM" observations (Debes et al. 2008a). Another proposed scenario 4.1.3. An ISM Gas Interaction Model for HD 61005 rav − . (6) small grains away from a circumstellar disk. The distance r = v2 for this morphology is the close flyby of HD 15115 by another av ∞ depends on β, v , the initial velocity of a grain at large distances, The disk observed around HD 61005, the “Moth,” shows a star (Kalas et al. 2007). However, no definitive candidates for a ∞ peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter flyby have currently been detected (Debes et al. 2008a). and the stellar mass M". up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum 2(β 1)GM" disk interaction with the gas component of the ISM. The star radius rav that is free from sandblasting from small ISM grains. 4.1.3. An ISM Gas Interaction Model for HD 61005 r − . (6) itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from av = v2 that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust The disk observed around HD 61005, the “Moth,” shows a ∞ grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some disk interaction with the gas component of the ISM. The star radius r that is free from sandblasting from small ISM grains. The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may av − itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may − 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for Can ISM interactions explain the bow?

324 DEBES, WEINBERGER, & KUCHNERHD 61005 (“the moth”) Vol. 702 324HD 15115 DEBES, WEINBERGER, & KUCHNER Vol. 702

Figure 7. Comparison between an F110W image of HD 15117’s disk from Debes et al. 2009 Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly Figure 7. Comparison between an F110W image of HD 15117’s disk from provided by G. Schneider) and a model of a disk interacting with a cloud of Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas (A color version of this figure is available in the online journal.) For the moth, yes For HD 15115, unlikely ISM gas with the parameters mentioned in Section 4.1.3. with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly (A color version of this figure is available in the online journal.) provided by G. Schneider) and a model of a disk interacting with a cloud of (A color version of this figure is available in the online journal.) ISM gas with the parameters mentioned in Section 4.1.3. P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe sky is at roughly the same P.A. as the disk. If we again assume (A color version of this figure is available in the online journal.) 1 the same assumptions we used above as for HD 15115, the a cloud with velocity of 14 km s− on the sky, the minimum 1 P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe relative velocities will range in magnitude from 2 km s− to and maximum possible relative transverse velocity is between 1 sky is at roughly the same P.A. as the disk. If we again assume 1 30 km s− ,withavectorprimarilyalongthex–z plane at an 1 the same assumptions we used above as for HD 15115, the 7and35kms− if the cloud’s velocity vector is parallel or a cloud with velocity of 14 km s on the sky, the minimum 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare − relative velocities will range in magnitude from 2 km s to antiparallel to HD 15115’s proper motion. The morphology of − the population we observe around HD 61005, we can derive Dg and maximum possible relative transverse velocity is between 30 km s 1,withavectorprimarilyalongthex–z plane at an the disk is consistent with the midplane plowing headlong into 3 1 − under the assumption of an ISM gas density N(H i) 100 cm− . 7and35kms− if the cloud’s velocity vector is parallel or an ISM flow. 18= 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare These two assumptions result in Dg 3.3 10− cm− . antiparallel to HD 15115’s proper motion. The morphology of We assumed that HD 15115 had a stellar mass of 1.5 M and = × the population we observe around HD 61005, we can derive Dg # We experimented with various relative velocity vectors until the disk is consistent with the midplane plowing headlong into 3 that grains were launched from 35 AU, the rough inner radius of under the assumption of an ISM gas density N(H i) 100 cm− . we found a qualitative match to the data. A velocity vector that an ISM flow. 18= 1 HD 15115’s disk as inferred from its IR excess (Zuckerman & These two assumptions result in Dg 3.3 10 cm . is nearly perpendicular to the disk midplane with a magnitude We assumed that HD 15115 had a stellar mass of 1.5 M and = × − − Song 2004; Williams & Andrews 2006). Experimenting with of 25 km s 1 matches the data best, along with an inclination # We experimented with various relative velocity vectors until 18 1 − that grains were launched from 35 AU, the rough inner radius of different models showed that Dg 1.3 10− cm− and we found a qualitative match to the data. A velocity vector that = × between the observer and the disk of 20◦.Figure8 shows a Rodigas/NCAD/July 20, 2012 arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115’s disk as inferred from its IR excess (Zuckerman & is nearly perpendicular to the disk midplane with a magnitude midplane and along the observer’s line of sight with a magnitude Song 2004; Williams & Andrews 2006). Experimenting with 1 1 and stretch. We also see good qualitative agreement with the 18 1 of 25 km s− matches the data best, along with an inclination of 30 km s− best matched the observed data. morphology of this disk as well. different models showed that Dg 1.3 10− cm− and between the observer and the disk of 20 .Figure8 shows a Figure 7 shows a comparison of the F110W image of = × ◦ arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115 with our disk model on the same size scale. The 5. DISCUSSION midplane and along the observer’s line of sight with a magnitude eastern lobe (left side of Figure 7)iscontaminatedbyastrong 1 and stretch. We also see good qualitative agreement with the of 30 km s− best matched the observed data. morphology of this disk as well. residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- Figure 7 shows a comparison of the F110W image of et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the model matches the observation of HD 15115 showing a HD 15115 with our disk model on the same size scale. The 5. DISCUSSION significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains eastern lobe (left side of Figure 7)iscontaminatedbyastrong figure shows that our models predict that the FWHM of the disk residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- midplane should be wider on the truncated side. This aspect For a disk interacting with an atomic ISM cloud with small et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe of the model matches the observation of HD 15115 showing a radiation pressure of the central star was strong enough to scatter observations (Debes et al. 2008a). Another proposed scenario significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains for this morphology is the close flyby of HD 15115 by another small grains away from a circumstellar disk. The distance rav depends on β, v , the initial velocity of a grain at large distances, figure shows that our models predict that the FWHM of the disk star (Kalas et al. 2007). However, no definitive candidates for a ∞ For a disk interacting with an atomic ISM cloud with small and the stellar mass M". midplane should be wider on the truncated side. This aspect flyby have currently been detected (Debes et al. 2008a). of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe radiation pressure of the central star was strong enough to scatter 2(β 1)GM" observations (Debes et al. 2008a). Another proposed scenario 4.1.3. An ISM Gas Interaction Model for HD 61005 rav − . (6) small grains away from a circumstellar disk. The distance r = v2 for this morphology is the close flyby of HD 15115 by another av ∞ depends on β, v , the initial velocity of a grain at large distances, The disk observed around HD 61005, the “Moth,” shows a star (Kalas et al. 2007). However, no definitive candidates for a ∞ peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter flyby have currently been detected (Debes et al. 2008a). and the stellar mass M". up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum 2(β 1)GM" disk interaction with the gas component of the ISM. The star radius rav that is free from sandblasting from small ISM grains. 4.1.3. An ISM Gas Interaction Model for HD 61005 r − . (6) itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from av = v2 that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust The disk observed around HD 61005, the “Moth,” shows a ∞ grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some disk interaction with the gas component of the ISM. The star radius r that is free from sandblasting from small ISM grains. The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may av − itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may − 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for Can ISM interactions explain the bow?

324 DEBES, WEINBERGER, & KUCHNERHD 61005 (“the moth”) Vol. 702 324HD 15115 DEBES, WEINBERGER, & KUCHNER Vol. 702

Figure 7. Comparison between an F110W image of HD 15117’s disk from Debes et al. 2009 Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly Figure 7. Comparison between an F110W image of HD 15117’s disk from provided by G. Schneider) and a model of a disk interacting with a cloud of Debes et al. (2008b) and a model of a disk interacting with a cloud of ISM gas (A color version of this figure is available in the online journal.) For the moth, yes For HD 15115, unlikely ISM gas with the parameters mentioned in Section 4.1.3. with the parameters mentioned in Section 4.1.2. Figure 8. Comparison between an F110W image of HD 61005’s disk (kindly (A color version of this figure is available in the online journal.) provided by G. Schneider) and a model of a disk interacting with a cloud of (A color version of this figure is available in the online journal.) ISM gas with the parameters mentioned in Section 4.1.3. P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe sky is at roughly the same P.A. as the disk. If we again assume (A color version of this figure is available in the online journal.) 1 the same assumptions we used above as for HD 15115, the a cloud with velocity of 14 km s− on the sky, the minimum 1 P.A. of 279 ◦.ThetransversevelocityvectorofHD15115onthe relative velocities will range in magnitude from 2 km s− to and maximum possible relative transverse velocity is between 1 sky is at roughly the same P.A. as the disk. If we again assume 1 30 km s− ,withavectorprimarilyalongthex–z plane at an 1 the same assumptions we used above as for HD 15115, the 7and35kms− if the cloud’s velocity vector is parallel or a cloud with velocity of 14 km s on the sky, the minimum 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare − relative velocities will range in magnitude from 2 km s to antiparallel to HD 15115’s proper motion. The morphology of − the population...But we observe is around good HD 61005, candidate we can derive Dtog explainand maximum east-west possible relative transverseasymmetry velocity is between 30 km s 1,withavectorprimarilyalongthex–z plane at an the disk is consistent with the midplane plowing headlong into 3 1 − under the assumption of an ISM gas density N(H i) 100 cm− . 7and35kms− if the cloud’s velocity vector is parallel or an ISM flow. 18= 1 inclination of 68◦ relative to the x-axis. If icy 0.3 µmgrainsare These two assumptions result in Dg 3.3 10− cm− . antiparallel to HD 15115’s proper motion. The morphology of We assumed that HD 15115 had a stellar mass of 1.5 M and = × the population we observe around HD 61005, we can derive Dg # We experimented with various relative velocity vectors until the disk is consistent with the midplane plowing headlong into 3 that grains were launched from 35 AU, the rough inner radius of under the assumption of an ISM gas density N(H i) 100 cm− . we found a qualitative match to the data. A velocity vector that an ISM flow. 18= 1 HD 15115’s disk as inferred from its IR excess (Zuckerman & These two assumptions result in Dg 3.3 10 cm . is nearly perpendicular to the disk midplane with a magnitude We assumed that HD 15115 had a stellar mass of 1.5 M and = × − − Song 2004; Williams & Andrews 2006). Experimenting with of 25 km s 1 matches the data best, along with an inclination # We experimented with various relative velocity vectors until 18 1 − that grains were launched from 35 AU, the rough inner radius of different models showed that Dg 1.3 10− cm− and we found a qualitative match to the data. A velocity vector that = × between the observer and the disk of 20◦.Figure8 shows a Rodigas/NCAD/July 20, 2012 arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115’s disk as inferred from its IR excess (Zuckerman & is nearly perpendicular to the disk midplane with a magnitude midplane and along the observer’s line of sight with a magnitude Song 2004; Williams & Andrews 2006). Experimenting with 1 1 and stretch. We also see good qualitative agreement with the 18 1 of 25 km s− matches the data best, along with an inclination of 30 km s− best matched the observed data. morphology of this disk as well. different models showed that Dg 1.3 10− cm− and between the observer and the disk of 20 .Figure8 shows a Figure 7 shows a comparison of the F110W image of = × ◦ arelativevelocityvectorwithequalcomponentsalongthe comparison of the model and the data, at the same size scale HD 15115 with our disk model on the same size scale. The 5. DISCUSSION midplane and along the observer’s line of sight with a magnitude eastern lobe (left side of Figure 7)iscontaminatedbyastrong 1 and stretch. We also see good qualitative agreement with the of 30 km s− best matched the observed data. morphology of this disk as well. residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- Figure 7 shows a comparison of the F110W image of et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the model matches the observation of HD 15115 showing a HD 15115 with our disk model on the same size scale. The 5. DISCUSSION significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains eastern lobe (left side of Figure 7)iscontaminatedbyastrong figure shows that our models predict that the FWHM of the disk residual that points roughly northeast in the figure (see Debes In this section, we discuss some details that our ISM interac- midplane should be wider on the truncated side. This aspect For a disk interacting with an atomic ISM cloud with small et al. 2008a). Otherwise, the figure shows that the morphology tion model presently lacks. of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe of the model matches the observation of HD 15115 showing a radiation pressure of the central star was strong enough to scatter observations (Debes et al. 2008a). Another proposed scenario significant truncation of one lobe of the disk. Additionally, the 5.1. ISM Dust Grains for this morphology is the close flyby of HD 15115 by another small grains away from a circumstellar disk. The distance rav depends on β, v , the initial velocity of a grain at large distances, figure shows that our models predict that the FWHM of the disk star (Kalas et al. 2007). However, no definitive candidates for a ∞ For a disk interacting with an atomic ISM cloud with small and the stellar mass M". midplane should be wider on the truncated side. This aspect flyby have currently been detected (Debes et al. 2008a). of the disk morphology is not significantly probed by current grains, AC97 calculated an “avoidance radius,” rav,wherethe radiation pressure of the central star was strong enough to scatter 2(β 1)GM" observations (Debes et al. 2008a). Another proposed scenario 4.1.3. An ISM Gas Interaction Model for HD 61005 rav − . (6) small grains away from a circumstellar disk. The distance r = v2 for this morphology is the close flyby of HD 15115 by another av ∞ depends on β, v , the initial velocity of a grain at large distances, The disk observed around HD 61005, the “Moth,” shows a star (Kalas et al. 2007). However, no definitive candidates for a ∞ peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter flyby have currently been detected (Debes et al. 2008a). and the stellar mass M". up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum 2(β 1)GM" disk interaction with the gas component of the ISM. The star radius rav that is free from sandblasting from small ISM grains. 4.1.3. An ISM Gas Interaction Model for HD 61005 r − . (6) itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from av = v2 that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust The disk observed around HD 61005, the “Moth,” shows a ∞ grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. peculiar disk morphology reminiscent of material being pushed Incoming ISM dust grains pile up at this radius and scatter determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- up out of the disk midplane (H07). This shape might indicate away from the star, leaving a paraboloidal region with minimum from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some disk interaction with the gas component of the ISM. The star radius r that is free from sandblasting from small ISM grains. The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may av − itself is a late G8 dwarf at a distance of 34 pc, and H07 found This boundary marks the separation between where grains from 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an that the surface brightness of the dust is consistent with dust the ISM could have a significant effect and where the ISM dust major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for grains roughly the size of the blow-out radius, 0.3 µm. H07 also does not impact the evolution of dust in a circumstellar disk. determined from HD 61005’s disk SED that the grains originate The color asymmetries observed for HD 32297 could poten- from radii 10 AU from the star. tially be explained by this ISM grain sandblasting. With some The proper∼ motion of HD 61005 is 56.09 mas in R.A., and assumptions about the ISM grains as well as the cloud that may − 74.53 mas in declination (van Leeuwen 2007). If we rotate these be impacting HD 32297’s disk, we can estimate what rav is for velocities to a frame where the x-axis is parallel to the presumed HD 32297. AC97 calculated expected β’s for ISM dust for an major axis of the disk and convert to relative velocities under A5 star such as HD 32297. They calculated a maximum β for Just viewing geometry?

Rodigas/NCAD/July 20, 2012 The Astrophysical Journal,752:57(13pp),2012June10 Rodigas et al.

1 5 2 4 West 0.5 East 3 1

0 2

arcseconds 1 0 10 µm −0.5 3 µm Just viewing geometry?0 1 µm −1 −1 −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds L’) (mags) − Real −2 1 5 (Ks ∆ 4 0.5 −3 3

0 2 −4

arcseconds 1 −0.5 0 −5 −1 −1 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds distance from star (arcseconds)

(a) Figure 8. Disk color vs. distance from the star, expressed as ∆(Ks L#) − = (Ks L )Disk (Ks L )Star.Thisiscalculatedonlywherethediskisdetected − # − − # 280 at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel colors from Inoue et al. (2008) (colored horizontal lines). The data suggest that the eastern side of the disk becomes redder than the western side with increasing 279 distance from the star. 1–10 µmgrainsarethebestfittothewesternside,while 3–10 µmgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe 278 west side of the disk is composed of smaller grains than the east side. (A color version of this figure is available in the online journal.) 277 To constrain the characteristic dust grain sizes, we compared 276 our ∆(Ks L#) disk colors to the models of Inoue et al. (2008). These models− calculate disk colors in the NIR and assume 275 silicate dust grain composition. The modeled grain sizes range

Position Angle (degrees) from 0.1 µmto10µmandassumeasinglegrainsizeforthe 274 dust population. We plot the model disk colors for 1, 3, and 10 µmgrains.TheseareshownasthecoloredlinesinFigure8. 273 We also modeledRodigas/NCAD/July 0.1 µmand0.3 20,µmgrains,buttheircolors 2012 were too blue to be supported by the data. 272 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 From Figure 8,weseethat1–10µm sized grains are the best distance from star (arcseconds) match to the western side of the disk, while 3–10 µmarethe (b) best match to the eastern side. This suggests that the observed Figure 7. (a) Rotated model and real Ks-band disk images. The model image scattered light from the disk is tracing the large parent body dust has an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB ◦ = grains in the disk. The data also may suggest that the west side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been convolved with the PSF− from the photometric image of HBC 388, and Gaussian of the disk is composed of smaller grains than the east side. The 2 noise has been added. Units are in mJy arcsec− for both images. The white blowout grain size for this system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn 2010), assuming a stellar mass of 1.3 M , a luminosity∼ % 3 to guide the eye to the apparent northern offset of the disk relative to the star. of 3.3 L ,andagraindensityintherange1–2.5gcm− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller than% the blowout size would be blown out radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) back to the west, resulting in the western side being dominated by smaller blue-scattering grains. Our observational constraints proposition that the effects can be explained by the disk’s geo- on the dust grain sizes offer some support for these predictions. metric orientation in space is attractive because it explains the observations without contradicting the evidence supporting the 4.3. Does the Disk have a Gap? ISM interaction interpretation (Debes et al. 2009). The SB profiles at Ks band and at L# drop off or flatten out 4.2. Disk Color and Grain Sizes (at low S/N) near 1##.Foranedge-ondiskwithnogap,theSB should continue to increase closer to the star. Because we do To determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star has a gap interior to 1.1. This is consistent with the prediction − # = − # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# of a gap near 1## by Moor´ et al. (2011b), using spectral energy images (1##.1–1##.45). The errors were calculated by summing the distribution (SED) analysis of the system. Though degenerate individual Ks and L# errors in quadrature. The data suggest that with temperature and dust grain size, their best-fit model of the eastern side of the disk becomes redder than the western side HD 15115’s infrared SED yields a two-component disk, with with increasing distance from the star. The western side of the the inner at 4 AU and the outer at 42 AU (both 2AU).Given disk becomes slightly bluer closer to the star. adistancetothestarof45.2pc,theouteredgeofthegapwould±

9 6 Rodigas et al.

0.3 99 Ks East Ks East 98.5

0.25 98

97.5

The Astrophysical Journal0.2 ,752:57(13pp),2012June10 97 Rodigas et al.

96.5 1 5 0.15 96 2

FWHM (arcseconds) 4 West

Position Angle (degrees) 95.5 0.5 East 0.1 3 95 1

0 2 94.5

arcseconds 0.05 1 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 01.2 1.4 1.6 1.8 2 2.2 2.4 10 µm −0.5 distance from star (arcseconds) distance from star (arcseconds) 3 µm Just viewing geometry?0 (a) (a) 1 µm −1 −1 −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds PA vs. radius L’) (mags) 0.26 Real − Ks West Ks West 286 −2 1 5 0.24 (Ks ∆ 0.22 4 285 0.5 −3 0.2 3 284 Real 0.18 0 2 0.16 283 −4

arcseconds 1 −0.5 0.14 282

FWHM (arcseconds) 0.12 0 Position Angle (degrees) −5 −1 0.1 −1 281 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 0.08 arcseconds distance from star (arcseconds) 280 0.06 (a) Figure 8. Disk color vs. distance from the star, expressed as ∆(Ks L#) 1 1.5 2 2.5 1 1.5 2 2.5 − = (Ks L )Disk (Ks L )Star.Thisiscalculatedonlywherethediskisdetected distance from star (arcseconds) − # distance− from− star (arcseconds)# 280 at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel (b) colors from Inoue et(b) al. (2008) (colored horizontal lines). The data suggest that Fig. 3.— Top : Ks band eastern disk FWHM as a function of Fig. 4.—the easternTop : PA side of the of eastern the disk side becomes of the disk redder at Ks thanband the western side with increasing 279 distance from the star. See text for methodology. There is no as a function of distance from the star. See text for methodology. obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PAdistance increases withfrom increasing the star. distance 1–10 µ frommgrainsarethebestfittothewesternside,while the star; over the as above, except for the western side of the disk. The FWHM same stellocentric distances a similar trend was observed at 1.1 increases with increasing distance from the star, though this may µm (Debes3–10 etµ al.mgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe 2008). Bottom: Same as above, except for the 278 be an effect from the data reduction (see Appendix). westernwest side. The side PA of increases the disk closer is composed to the star, of as smaller is seen in grains the than the east side. Debes et al. (2008) 1.1 µmbanddata. to avoid any potential wavelength-dependent changes in (A color version of this figure is available in the online journal.) 2008). The western PA increases closer to the star. This PA. Comparing our final PISCES Ks band image with 277 !! also agrees with the western PA vs. stellocentric distance the Keck image between 1-2 revealed a difference in ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverall overall PA of 3.9 ,sowerotatedourPISCESimagebyan ◦ ◦ PA (278.5 )measuredforthewesternsideofthediskat additional 3.9 clockwise. Fig. 1a shows this correctly- To constrain the characteristic dust grain sizes, we compared 276 0.6 µmbetween1.5-12!! (Kalas et al. 2007). rotated image. We calculatedour ∆(Ks the PAL# of) disk the disk colors at L to! in the the models same of Inoue et al. (2008). To test for any warps in the disk, we calculated the PA manner as for the −Ks band image, except we used 3 of the disk at Ks band on both sides of the disk as a func- These models calculate disk colors in the NIR and assume 275 pixel by 11 pixel (∼ 0!!. 13 by 0!!. 47) boxes. The L! image tion of distance from the star, using the disk midplane silicate dust grain composition. The modeled grain sizes range was again binned by a factor of 4. We do not plot PA vs.

Position Angle (degrees) pixel locations from the Gaussian fits. The errors were distancefrom from 0.1 theµ starmto10 due to theµmandassumeasinglegrainsizeforthe poor spatial coverage 274 measured as the difference between these values and the of the disk and, even after binning, the low SNRE (∼ 2- center of light along each 1D row at each discrete pixel dust population. We plot the model disk colors for 1, 3, and 3); however the measured values for both sides of the distance from the star. The errors also included the 1 !! disk were10 µ globallymgrains.TheseareshownasthecoloredlinesinFigure consistent with the values measured 8. 273 pixel = 0. 194 astrometric centroiding uncertainty. Fig. at Ks band (this is also evident by examination of the 4a and Fig. 4b show the Ks band PA of the eastern and We also modeled 0.1 µmand0.3µmgrains,buttheircolors Ks band and L! disk imagesRodigas/NCAD/July in Fig. 1a and Fig. 20, 2a, 2012 western sides of the disk as a function of distance from respectively).were too blue to be supported by the data. 272 the star, respectively. The eastern PA increases with in- 0.8 1 creasing1.2 distance1.4 1.6 from the1.8 star, similar2 to2.2 what2.4 is seen 2.6 From Figure 8,weseethat1–10µm sized grains are the best distance from star (arcseconds) 3.3. Surface Brightness Profiles over the same spatial locations at 1.1 µm(Debesetal. match to the western side of the disk, while 3–10 µmarethe (b) best match to the eastern side. This suggests that the observed Figure 7. (a) Rotated model and real Ks-band disk images. The model image scattered light from the disk is tracing the large parent body dust has an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB ◦ = grains in the disk. The data also may suggest that the west side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been convolved with the PSF− from the photometric image of HBC 388, and Gaussian of the disk is composed of smaller grains than the east side. The 2 noise has been added. Units are in mJy arcsec− for both images. The white blowout grain size for this system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn 2010), assuming a stellar mass of 1.3 M , a luminosity∼ % 3 to guide the eye to the apparent northern offset of the disk relative to the star. of 3.3 L ,andagraindensityintherange1–2.5gcm− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller than% the blowout size would be blown out radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) back to the west, resulting in the western side being dominated by smaller blue-scattering grains. Our observational constraints proposition that the effects can be explained by the disk’s geo- on the dust grain sizes offer some support for these predictions. metric orientation in space is attractive because it explains the observations without contradicting the evidence supporting the 4.3. Does the Disk have a Gap? ISM interaction interpretation (Debes et al. 2009). The SB profiles at Ks band and at L# drop off or flatten out 4.2. Disk Color and Grain Sizes (at low S/N) near 1##.Foranedge-ondiskwithnogap,theSB should continue to increase closer to the star. Because we do To determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star has a gap interior to 1.1. This is consistent with the prediction − # = − # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# of a gap near 1## by Moor´ et al. (2011b), using spectral energy images (1##.1–1##.45). The errors were calculated by summing the distribution (SED) analysis of the system. Though degenerate individual Ks and L# errors in quadrature. The data suggest that with temperature and dust grain size, their best-fit model of the eastern side of the disk becomes redder than the western side HD 15115’s infrared SED yields a two-component disk, with with increasing distance from the star. The western side of the the inner at 4 AU and the outer at 42 AU (both 2AU).Given disk becomes slightly bluer closer to the star. adistancetothestarof45.2pc,theouteredgeofthegapwould±

9 6 Rodigas et al.

0.3 99 Ks East Ks East 98.5

0.25 98

97.5

The Astrophysical Journal0.2 ,752:57(13pp),2012June10 97 Rodigas et al.

96.5 1 5 0.15 96 2

FWHM (arcseconds) 4 West

Position Angle (degrees) 95.5 0.5 East 0.1 3 95 1

0 2 94.5

arcseconds 0.05 1 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 01.2 1.4 1.6 1.8 2 2.2 2.4 10 µm −0.5 distance from star (arcseconds) distance from star (arcseconds) 3 µm Just viewing geometry?0 (a) (a) 1 µm −1 −1 −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds PA vs. radius L’) (mags) 0.26 Real − Ks West Ks West 286 −2 1 5 0.24 (Ks ∆ 0.22 4 285 0.5 −3 0.2 3 284 Real 0.18 0 2 0.16 283 −4

arcseconds 1 −0.5 0.14 282

FWHM (arcseconds) 0.12 0 Position Angle (degrees) −5 −1 0.1 −1 281 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 0.08 arcseconds distance from star (arcseconds) 280 0.06 (a) Figure 8. Disk color vs. distance from the star, expressed as ∆(Ks L#) The Astrophysical Journal1 ,752:57(13pp),2012June101.5 2 2.5 1 1.5 2 2.5 Rodigas− et= al. (Ks L )Disk (Ks L )Star.Thisiscalculatedonlywherethediskisdetected ○ distance from star (arcseconds) − # distance− from− star (arcseconds)# 280Model (87 inclined ring, forward-scattering grains) at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel 1 (b) 5 colors2 from Inoue et(b) al. (2008) (colored horizontal lines). The data suggest that Fig. 3.— Top : Ks band eastern disk FWHM as a function of Fig. 4.— Top : PA of the eastern side of the disk at Ks band 4 the eastern sideWest of the disk becomes redder than the western side with increasing 279 distance from the star. See text for methodology. There is no as a function of distance from the star. See text for methodology. 0.5 obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PAdistance increases withfrom increasing theEast star. distance 1–10 µ frommgrainsarethebestfittothewesternside,while the star; over the as above, except for the western side of the disk. The FWHM3 same stellocentric1 distances a similar trend was observed at 1.1 increases with increasing distance from the star, though this may µm (Debes3–10 etµ al.mgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe 2008). Bottom: Same as above, except for the 2780 be an effect from the data reduction (see Appendix). 2 westernwest side. The side PA of increases the disk closer is composed to the star, of as smaller is seen in grains the than the east side. Debes et al. (2008) 1.1 µmbanddata. arcseconds to avoid any potential wavelength-dependent changes in1 (A color0 version of this figure is available in the online journal.) 10 µm 2008). The western PA increases closer to the star. This 3 m −0.5 PA. Comparing our final PISCES Ks band image with µ 277 !! 0 also agrees with the western PA vs. stellocentric distance the Keck image between 1-2 revealed a difference in ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverall 1 µm −1 overall PA of 3.9 ,sowerotatedourPISCESimagebyan−1 ◦ −1 −2.5 −2 −1.5 −1 −0.5 ◦ 0 0.5 1 1.5 2 2.5 PA (278.5To)measuredforthewesternsideofthediskat constrain the characteristic dust grain sizes, we compared additional 3.9 clockwise. Fig. 1a shows this correctly- !! 276 arcseconds 0.6 µmbetween1.5-12 (Kalas et al. 2007). rotated image. ourL’) (mags) ∆(Ks L#) disk colors to! the models of Inoue et al. (2008). We calculated− the PA of the disk at L in the same To test for any warps in the disk, we calculated the PA manner as− for2 the −Ks band image, except we used 3 1 5 These models calculate disk colors in the NIR and assume 275 of the disk at Ks band on both sides of the disk as a func- (Ks ∼ !! !! !

pixel by 11∆ pixel ( 0. 13 by 0. 47) boxes. The L image tion of distance from the star, using the disk midplane silicate dust grain composition. The modeled grain sizes range 4 was again binned by a factor of 4. We do not plot PA vs.

Position Angle (degrees) pixel locations from the Gaussian fits. The errors were 0.5 distancefrom from−3 0.1 theµ starmto10 due to theµmandassumeasinglegrainsizeforthe poor spatial coverage 274 measured as the difference between these values and the 3 of the disk and, even after binning, the low SNRE (∼ 2- center of light along each 1D row at each discrete pixel dust population. We plot the model disk colors for 1, 3, and 3); however the measured values for both sides of the 0 distance from the star. The errors also included the 12 !! disk were10 µ globallymgrains.TheseareshownasthecoloredlinesinFigure consistent with the values measured 8. 273 pixel = 0. 194 astrometric centroiding uncertainty. Fig. −4 arcseconds 1 at Ks band (this is also evident by examination of the 4a and Fig. 4b show the Ks band PA of the eastern and We also modeled 0.1 µmand0.3µmgrains,buttheircolors −0.5 Ks band and L! disk imagesRodigas/NCAD/July in Fig. 1a and Fig. 20, 2a, 2012 western sides of the disk as a function of distance from 0 respectively).were too blue to be supported by the data. 272 the star, respectively. The eastern PA increases with in- −5 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 −1 creasing distance from the star, similar to what is seen−1 From3.3. Figure 8,weseethat1–10µm sized grains are the best −2.5 −2 −1.5 −1distance−0.5 from0 star0.5 (arcseconds)1 1.5 2 2.5 Surface Brightnessdistance Profiles from star (arcseconds) over the samearcseconds spatial locations at 1.1 µm(Debesetal. match to the western side of the disk, while 3–10 µmarethe (b) (a) bestFigure match 8. Disk to color the eastern vs. distance side. from This the suggestsstar, expressed that as the∆(Ks observedL#) − = Figure 7. (a) Rotated model and real Ks-band disk images. The model image (Ks L )Disk (Ks L )Star.Thisiscalculatedonlywherethediskisdetected scattered− # light− from− the# disk is tracing the large parent body dust has280 an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel ◦ = grains in the disk. The data also may suggest that the west side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been colors from Inoue et al. (2008) (colored horizontal lines). The data suggest that convolved with the PSF− from the photometric image of HBC 388, and Gaussian ofthe the eastern disk side is of composed the disk becomes of smaller redder grains than the than western the side east with side. increasing The 279 2 noise has been added. Units are in mJy arcsec− for both images. The white blowoutdistance from grain the sizestar. 1–10 forµ thismgrainsarethebestfittothewesternside,while system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn3–10 µmgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe2010), assuming a stellar mass of 1.3 M , a luminosity∼ 278 % 3 to guide the eye to the apparent northern offset of the disk relative to the star. ofwest 3.3 sideL of,andagraindensityintherange1–2.5gcm the disk is composed of smaller grains than the east side.− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller(A color versionthan% the of blowout this figure size is available would in be the blown online out journal.) radially; on the image.277 The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) backTo to constrain the west, the resulting characteristic in the western dust grain side sizes, being we dominated compared 276 by smaller blue-scattering grains. Our observational constraints our ∆(Ks L#) disk colors to the models of Inoue et al. (2008). proposition that the effects can be explained by the disk’s geo- onThese the dust models− grain calculate sizes offer disk some colors support in forthe these NIR predictions. and assume 275 metric orientation in space is attractive because it explains the silicate dust grain composition. The modeled grain sizes range 4.3. Does the Disk have a Gap? observationsPosition Angle (degrees) without contradicting the evidence supporting the from 0.1 µmto10µmandassumeasinglegrainsizeforthe ISM274 interaction interpretation (Debes et al. 2009). dustThe population. SB profiles We at Ks plotband the modeland at L disk# drop colors off or for flatten 1, 3, out and (at10 lowµmgrains.TheseareshownasthecoloredlinesinFigure S/N) near 1 .Foranedge-ondiskwithnogap,theSB8. 273 4.2. Disk Color and Grain Sizes ## shouldWe also continue modeled to 0.1 increaseµmand0.3 closer toµmgrains,buttheircolors the star. Because we do were too blue to be supported by the data. To272 determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star hasFrom a gap Figure interior8,weseethat1–10 to 1.1. This is consistentµm sized with grains the are prediction the best distance− from# = star (arcseconds)− # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# ofmatch a gap to near the 1 western## by Mo sideor´ et of al. the (2011b disk,), while using 3–10 spectralµmarethe energy (b) images (1##.1–1##.45). The errors were calculated by summing the distributionbest match to (SED) the eastern analysis side. of the This system. suggests Though that the degenerate observed Figureindividual 7. (a)Ks Rotatedand modelL# errors and real in quadrature.Ks-band disk The images. data The suggest model image that withscattered temperature light from and the dust disk grain is tracing size, the their large best-fit parent model body dust of has an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB the eastern side of the◦ disk becomes redder= than the western side HDgrains 15115’s in the infrared disk. The SED data yields also may a two-component suggest that the disk, west with side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been convolvedwith increasing with the PSF− distance from the from photometric the star. image The of western HBC 388, side and Gaussian of the theof the inner disk at is 4 composed AU and the of outer smaller at 42 grains AU (both than the2AU).Given east side. The 2 ± noisedisk has becomes been added. slightly Units bluer are in closer mJy arcsec to the− for star. both images. The white adistancetothestarof45.2pc,theouteredgeofthegapwouldblowout grain size for this system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn 2010), assuming a stellar mass of 1.3 M , a luminosity∼ % 3 to guide the eye to the apparent northern offset of the disk relative to the star. 9 of 3.3 L ,andagraindensityintherange1–2.5gcm− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller than% the blowout size would be blown out radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) back to the west, resulting in the western side being dominated by smaller blue-scattering grains. Our observational constraints proposition that the effects can be explained by the disk’s geo- on the dust grain sizes offer some support for these predictions. metric orientation in space is attractive because it explains the observations without contradicting the evidence supporting the 4.3. Does the Disk have a Gap? ISM interaction interpretation (Debes et al. 2009). The SB profiles at Ks band and at L# drop off or flatten out 4.2. Disk Color and Grain Sizes (at low S/N) near 1##.Foranedge-ondiskwithnogap,theSB should continue to increase closer to the star. Because we do To determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star has a gap interior to 1.1. This is consistent with the prediction − # = − # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# of a gap near 1## by Moor´ et al. (2011b), using spectral energy images (1##.1–1##.45). The errors were calculated by summing the distribution (SED) analysis of the system. Though degenerate individual Ks and L# errors in quadrature. The data suggest that with temperature and dust grain size, their best-fit model of the eastern side of the disk becomes redder than the western side HD 15115’s infrared SED yields a two-component disk, with with increasing distance from the star. The western side of the the inner at 4 AU and the outer at 42 AU (both 2AU).Given disk becomes slightly bluer closer to the star. adistancetothestarof45.2pc,theouteredgeofthegapwould±

9 6 Rodigas et al.

0.3 99 Ks East Ks East 98.5

0.25 98

97.5

The Astrophysical Journal0.2 ,752:57(13pp),2012June10 97 Rodigas et al.

96.5 1 5 0.15 96 2

FWHM (arcseconds) 4 West

Position Angle (degrees) 95.5 0.5 East 0.1 3 95 1 0 2 94.5 Large Grains in the HD 15115 Debris Disk 9

arcseconds 0.05 1 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 01.2 1.4 1.6 1.8 2 2.2 2.4 10 µm −0.5 distance from star (arcseconds) distance from star (arcseconds) 2009)). Because HD3 15115 µm is moving to the south-east, Just viewing geometry?0 1 5 nearly parallel with its disk major axis, one would not 4 (a) 0.5 (a) expect the disk to have1 µm a bow-shape, especially with an −1 −1 −1 3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 offset from the star to the north. However, in the HD arcseconds 0 PA vs. radius 2 61005 debris disk system, Maness et al. (2009) showed L’) (mags) arcseconds 1 −0.5 − that dust grain interactions with the ISM in directions 0.26 Real Ks West Ks West 286 −2 0 1 5 along the disk midplane could actually create bow-shapes 0.24 −1 (Ks −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 ∆ perpendicular to the disk major axis, roughly reproduc- arcseconds 0.22 4 285 ing that disk’s observed morphology. Similar dynamical 0.5 1 −3 5 interactions in the HD 15115 system could be creating 0.2 3 Real 284 4 the observed bow-shape, though no explicit model has 0.18 0.5 0 2 3 tested this hypothesis for this system. 0.16 2830 −4 2 Another perhaps simpler explanation for the bow- arcseconds

1 arcseconds 0.14 1 shape and the apparent offset is that these are caused by −0.5 −0.5 282 0 geometrical effects. We can reproduce the observed bow-

FWHM (arcseconds) 0.12 0 Position Angle (degrees) −1 −5 −1 shape and vertical offset with a simple inclined, ringed −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −1 0.1 −1 281 1.1 arcseconds1.15 1.2 1.25 1.3 1.35disk model.1.4 We1.45 used the radiative transfer equations de- −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 0.08 distance from star (arcseconds)scribing the intensity of light scattering off dust particles arcseconds (a) 280 and made three assumptions about the disk: it is not 0.06 ◦ (a) Figure 8. Disk color vs. distance from the star, expressedexactly as edge-on,∆(Ks butL rather#) inclined to 87 ; it has a gap The Astrophysical Journal1 ,752:57(13pp),2012June101.5 2 2.5 1 1.5 2 2.5 from 0-1!!. 1(seeSection4.3foradditionaldiscussionofRodigas− et= al. distance from star (arcseconds) (Ks L#)Diskdistance(Ks from starL (arcseconds)#)Star.Thisiscalculatedonlywherethediskisdetected ○ 280 − − − the possible disk gap); and we set the Henyey-Greenstein 280Model (87 inclined ring, forward-scattering grains) at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel 1 (b) 5 (b) scattering parameter g =0.5forpredominantlyforward- 279colors2 from Inoue et al. (2008) (colored horizontal lines). The data suggest that Fig. 3.— Top : Ks band eastern disk FWHM as a function of Fig. 4.— Top : PA of the eastern side of the disk at Ks band scattering grains. This is a reasonable assumption for 4 the eastern sideWest of the disk becomes redder than the westernlarge side(! 1 withµm) increasing grains scattering NIR light (Duchˆene 279 distance from the star. See text for methodology. There is no as a function278 of distance from the star. See text for methodology. 0.5 obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PAdistance increases withfrom increasing theEast star. distance 1–10 µ frommgrainsarethebestfittothewesternside,while the star; over the et al. 2004). We set the dust grain size uniformly to 3 as above, except for the western side of the disk. The FWHM3 same stellocentric1 distances a similar trend was observed at 1.1 2773–10 µmgrainsarethebestmatchtoeasternside.ThismaysuggestthattheµmandsetthediskSBpower-lawindexto-3/2.Fig.7a increases with increasing distance from the star, though this may µm (Debes et al. 2008). Bottom: Same as above, except for the shows the rotated model image, along with the rotated 2780 be an effect from the data reduction (see Appendix). 2 westernwest side. The side PA of increases the diskModel closer is composed to the star, of as smaller is seen in grains the than the east side. Debes276 et al. (2008) 1.1 µmbanddata. final Ks band image. Dashed lines have been inserted to arcseconds to avoid any potential wavelength-dependent changes in1 (A color0 version of this figure is available in the onlineguide journal.) the eye to the apparent10 µm bow-shape and northern −0.5 2008). The western PA increases closer to the star. This 3 µm PA. Comparing our final PISCES Ks band image with 275 offset from the star. Fig. 7b shows the PA vs. stellocen- 277 !! 0 also agrees with the western PA vs. stellocentric distance

the Keck image between 1-2 revealed a difference in Position Angle (degrees) tric distance for the western side of the disk in the model ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverall 1 µm −1 overall PA of 3.9 ,sowerotatedourPISCESimagebyan−1 274 ◦ −1 image, computed in the same manner as for the real data. −2.5 −2 −1.5 −1 −0.5 ◦ 0 0.5 1 1.5 2 2.5 PA (278.5To)measuredforthewesternsideofthediskat constrain the characteristic dust grain sizes, we compared 276 additional 3.9 clockwise. Fig. 1a shows this correctly- !! The PA increases closer to the star, as is observed in the arcseconds 0.6 µmbetween1.5-12273 (Kalas et al. 2007). rotated image. ourL’) (mags) ∆(Ks L#) disk colors to! the models ofreal Inoue data (Fig. et al. 4b). (2008). We calculated− the PA of the disk at L in the same To test for any warps in the disk, we calculated the PA − While the parameters used in our model are not a manner272 as− for2 the Ks band image, except we used 3 1 5 These0.8 1 models1.2 1.4 calculate1.6 1.8 2 disk2.2 colors2.4 2.6 in the NIR and assume 275 of the disk at Ks band on both sides of the disk as a func- (Ks ∼distance!! from star!! (arcseconds) ! unique explanation for the observed features of the disk,

pixel by 11∆ pixel ( 0. 13 by 0. 47) boxes. The L image tion of distance from the star, using the disk midplane silicate dust grain composition. The modeledthe proposition grain sizes that range the effects can be explained by the 4 was again binned by a factor of 4. We do not plot PA vs.

Position Angle (degrees) pixel locations from the Gaussian fits. The errors were disk’s geometric orientation in space is attractive because 0.5 distancefrom from−3 0.1 theµ starmto10 due(b) to theµmandassumeasinglegrainsizeforthe poor spatial coverage 274 measured as the difference between these values and the it explains the observations without contradicting the 3 ofFig. the 7.— diskTop and,: Rotated even after model binning, and real theKs lowband SNRE disk images. (∼ 2- center of light along each 1D row at each discrete pixel dust population. We plot◦ the model diskevidence colors forsupporting 1, 3, the and ISM interaction interpretation 3);The however model image the has measured an inclination values of 87 for, scatteringboth sides parameter of the 0 distance from the star. The errors also included the 12 g =0.5, 3 µm grains, a SB power-law index of -3/2, and a gap from (Debes et al. 2009). !! disk"" were10 µ globallymgrains.TheseareshownasthecoloredlinesinFigure consistent with the values measured 8. 273 pixel = 0. 194 astrometric centroiding uncertainty. Fig. 0-1. 1. The model−4 image has been convolved with the PSF from arcseconds 1 attheKs photometricband (this image is of also HBC evident 388, and by Gaussian examination noise has of been the 4.2. Disk Color and Grain Sizes 4a and Fig. 4b show the Ks band PA of the eastern and We also modeled2 0.1 µmand0.3µmgrains,buttheircolors −0.5 Ksadded.band Units and areL in! mJy/arcseconddisk imagesRodigas/NCAD/July infor Fig. both 1a images. and TheFig. 20, white 2a, 2012 western sides of the disk as a function of distance from0 dot marks the location of the star in both images. The dashed lines To determine the disk color as a function of distance respectively).were too blue to be supported by the data. ! ! 272 the star, respectively. The eastern PA increases with in- are meant to− guide5 the eye to the apparent northern offset of the from the star, we calculated ∆(Ks - L )=(Ks - L ) 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 Disk −1 creasing distance from the star, similar to what is seen−1 disk relativeFrom to the star. FigureBottom8:,weseethat1–10 PA vs. stellocentric distanceµ form sized-(Ks grains- L!) arein the the best regions of spatial overlap between −2.5 −2 −1.5 −1distance−0.5 from0 star0.5 (arcseconds)1 1.5 2 2.5 3.3. Surface Brightness Profiles Star over the same spatial locations at 1.1 µm(Debesetal. the western side of the disk in the modeldistance image. Thefrom PA star increases (arcseconds)the Ks band and L! images (1.1-1!!. 45). The errors were arcseconds closer tomatch the star due to to the the western apparent bow-shape, side of similar the to disk, what while 3–10 µmarethe ! (b) is seen for the real data shown in Fig. 4b. calculated by summing the individual Ks and L errors (a) bestFigure match 8. Disk to color the eastern vs. distance side. from This the suggestsstar, expressed that as the∆(Ks observedL#) in quadrature. The− data= suggest that the eastern side Figure 7. (a) Rotated model and real Ks-band disk images. The model image and 2a),scattered(Ks as wellL as#)Disk lightthe PA(Ks from vs. stellocentricL the#)Star disk.Thisiscalculatedonlywherethediskisdetected distanceis tracing plots the largeof the parent disk becomes body dust redder than the western side with for both sides− of the disk− (Fig.− 4a and 4b) suggests that increasing distance from the star. Interior to 1!!. 3both has280 an inclination of 87◦,scatteringparameterg 0.5, 3 µmgrains,anSB grainsat both the in theKs band disk. and The at L data#.Toconstraindustgrainsize,wealsoplotmodel also may suggest that the west side power-law index of 3/2, and a gap from 0 to 1. 1.= The model image has been the debriscolors disk from has a Inoue bow-like et al. shape (2008 and) (colored is offset by horizontal a few lines).sides of The the data disk suggest are grey. that − ## AU toof the the north disk from is composedthe star. Specifically of smaller the o grainsffsets thanTo the constrain east side. the characteristic The dust grain sizes, we convolved with the PSF from the photometric image of HBC 388, and Gaussian ∼ the!! eastern side of the!! disk becomes! redder than the western side with increasing! 279 2 are 0. 012 (0.5 AU) and 0. 04 (2 AU) at L ,fortheeast- compared our ∆(Ks - L )diskcolorstothemodelsof noise has been added. Units are in mJy arcsec− for both images. The white ern andblowoutdistance western from sides, grain the respectively; sizestar. 1–10 forµ this andmgrainsarethebestfittothewesternside,while∼ system0!!. 11 (5 AU)is expectedInoue to et al. be (2008).1–3 Theseµm models calculate disk colors dot marks the location of the star in both images. The dashed lines are meant and 0!!. (Hahn173–10 (8 AU)µmgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe2010 at Ks),band assuming for the eastern a stellar and western mass of 1.3in theM NIR, a and luminosity∼ assume silicate dust grain composition. 278 sides, respectively. Since the astrometric uncertainty in The modeled% grain3 sizes range from 0.1 µmto10µmand to guide the eye to the apparent northern offset of the disk relative to the star. ofwest 3.3 sideL of!!,andagraindensityintherange1–2.5gcm the disk is!! composed! of smaller grains than the east side..Grains centroiding is ∼ 0. 002 and 0. 0194 at L and Ks band, assume a single− grain size for the dust population. We (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller(A color versionthan% the of blowout this figure size is available would in be the blown online out journal.) radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, respectively, it is unlikely that the northern offsets could plot the model disk colors for 1, 3, and 10 µmgrains. 277 be dueeastern to centroid side, error. these Bow-like grains disk would morphologies hit the ISMThese and arecould shown be as blown the colored lines in Fig. 8. We also similar to what is seen for the real data shown in Figure 4(b). have been seen in systems moving in near-perpendicular modeled 0.1 µmand0.3µmgrains,buttheircolorswere (A color version of this figure is available in the online journal.) directionsbackTo to to their constrain the disk west, major the resulting axis characteristic positions in the (e.g., western dust HD grain sidetoo sizes, blue being to we be dominated supportedcompared by the data. 276 61005by (Hines smaller et al. 2007)blue-scattering and HD 32297 grains. (Debes Our et al. observationalFrom Fig. constraints 8, we see that 1-10 µm-sized grains are the our ∆(Ks L#) disk colors to the models of Inoue et al. (2008). proposition that the effects can be explained by the disk’s geo- onThese the dust models− grain calculate sizes offer disk some colors support in forthe these NIR predictions. and assume 275 metric orientation in space is attractive because it explains the silicate dust grain composition. The modeled grain sizes range 4.3. Does the Disk have a Gap? observationsPosition Angle (degrees) without contradicting the evidence supporting the from 0.1 µmto10µmandassumeasinglegrainsizeforthe ISM274 interaction interpretation (Debes et al. 2009). dustThe population. SB profiles We at Ks plotband the modeland at L disk# drop colors off or for flatten 1, 3, out and (at10 lowµmgrains.TheseareshownasthecoloredlinesinFigure S/N) near 1 .Foranedge-ondiskwithnogap,theSB8. 273 4.2. Disk Color and Grain Sizes ## shouldWe also continue modeled to 0.1 increaseµmand0.3 closer toµmgrains,buttheircolors the star. Because we do were too blue to be supported by the data. To272 determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star hasFrom a gap Figure interior8,weseethat1–10 to 1.1. This is consistentµm sized with grains the are prediction the best distance− from# = star (arcseconds)− # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# ofmatch a gap to near the 1 western## by Mo sideor´ et of al. the (2011b disk,), while using 3–10 spectralµmarethe energy (b) images (1##.1–1##.45). The errors were calculated by summing the distributionbest match to (SED) the eastern analysis side. of the This system. suggests Though that the degenerate observed Figureindividual 7. (a)Ks Rotatedand modelL# errors and real in quadrature.Ks-band disk The images. data The suggest model image that withscattered temperature light from and the dust disk grain is tracing size, the their large best-fit parent model body dust of has an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB the eastern side of the◦ disk becomes redder= than the western side HDgrains 15115’s in the infrared disk. The SED data yields also may a two-component suggest that the disk, west with side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been convolvedwith increasing with the PSF− distance from the from photometric the star. image The of western HBC 388, side and Gaussian of the theof the inner disk at is 4 composed AU and the of outer smaller at 42 grains AU (both than the2AU).Given east side. The 2 ± noisedisk has becomes been added. slightly Units bluer are in closer mJy arcsec to the− for star. both images. The white adistancetothestarof45.2pc,theouteredgeofthegapwouldblowout grain size for this system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn 2010), assuming a stellar mass of 1.3 M , a luminosity∼ % 3 to guide the eye to the apparent northern offset of the disk relative to the star. 9 of 3.3 L ,andagraindensityintherange1–2.5gcm− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller than% the blowout size would be blown out radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) back to the west, resulting in the western side being dominated by smaller blue-scattering grains. Our observational constraints proposition that the effects can be explained by the disk’s geo- on the dust grain sizes offer some support for these predictions. metric orientation in space is attractive because it explains the observations without contradicting the evidence supporting the 4.3. Does the Disk have a Gap? ISM interaction interpretation (Debes et al. 2009). The SB profiles at Ks band and at L# drop off or flatten out 4.2. Disk Color and Grain Sizes (at low S/N) near 1##.Foranedge-ondiskwithnogap,theSB should continue to increase closer to the star. Because we do To determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star has a gap interior to 1.1. This is consistent with the prediction − # = − # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# of a gap near 1## by Moor´ et al. (2011b), using spectral energy images (1##.1–1##.45). The errors were calculated by summing the distribution (SED) analysis of the system. Though degenerate individual Ks and L# errors in quadrature. The data suggest that with temperature and dust grain size, their best-fit model of the eastern side of the disk becomes redder than the western side HD 15115’s infrared SED yields a two-component disk, with with increasing distance from the star. The western side of the the inner at 4 AU and the outer at 42 AU (both 2AU).Given disk becomes slightly bluer closer to the star. adistancetothestarof45.2pc,theouteredgeofthegapwould±

9 6 Rodigas et al.

0.3 99 Ks East Ks East 98.5

0.25 98

97.5

The Astrophysical Journal0.2 ,752:57(13pp),2012June10 97 Rodigas et al.

96.5 1 5 0.15 96 2

FWHM (arcseconds) 4 West

Position Angle (degrees) 95.5 0.5 East 0.1 3 95 1 0 2 94.5 Large Grains in the HD 15115 Debris Disk 9

arcseconds 0.05 1 94 1 1.2 1.4 1.6 1.8 2 2.2 2.4 1 01.2 1.4 1.6 1.8 2 2.2 2.4 10 µm −0.5 distance from star (arcseconds) distance from star (arcseconds) 2009)). Because HD3 15115 µm is moving to the south-east, Just viewing geometry?0 1 5 nearly parallel with its disk major axis, one would not 4 (a) 0.5 (a) expect the disk to have1 µm a bow-shape, especially with an −1 −1 −1 3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 offset from the star to the north. However, in the HD arcseconds 0 PA vs. radius 2 61005 debris disk system, Maness et al. (2009) showed L’) (mags) arcseconds 1 −0.5 − that dust grain interactions with the ISM in directions 0.26 Real Ks West Ks West 286 −2 0 1 5 along the disk midplane could actually create bow-shapes 0.24 −1 (Ks −1 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 ∆ perpendicular to the disk major axis, roughly reproduc- arcseconds 0.22 4 285 ing that disk’s observed morphology. Similar dynamical 0.5 1 −3 5 interactions in the HD 15115 system could be creating 0.2 3 Real 284 4 the observed bow-shape, though no explicit model has 0.18 0.5 0 2 3 tested this hypothesis for this system. 0.16 2830 −4 2 Another perhaps simpler explanation for the bow- arcseconds

1 arcseconds 0.14 1 shape and the apparent offset is that these are caused by −0.5 −0.5 282 0 geometrical effects. We can reproduce the observed bow-

FWHM (arcseconds) 0.12 0 Position Angle (degrees) −1 −5 −1 shape and vertical offset with a simple inclined, ringed −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −1 0.1 −1 281 1.1 arcseconds1.15 1.2 1.25 1.3 1.35disk model.1.4 We1.45 used the radiative transfer equations de- −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 0.08 distance from star (arcseconds)scribing the intensity of light scattering off dust particles arcseconds (a) 280 and made three assumptions about the disk: it is not 0.06 ◦ (a) Figure 8. Disk color vs. distance from the star, expressedexactly as edge-on,∆(Ks butL rather#) inclined to 87 ; it has a gap The Astrophysical Journal1 ,752:57(13pp),2012June101.5 2 2.5 1 1.5 2 2.5 from 0-1!!. 1(seeSection4.3foradditionaldiscussionofRodigas− et= al. distance from star (arcseconds) (Ks L#)Diskdistance(Ks from starL (arcseconds)#)Star.Thisiscalculatedonlywherethediskisdetected ○ 280 − − − the possible disk gap); and we set the Henyey-Greenstein 280Model (87 inclined ring, forward-scattering grains) at both the Ks band and at L#.Toconstraindustgrainsize,wealsoplotmodel 1 (b) 5 (b) scattering parameter g =0.5forpredominantlyforward- 279colors2 from Inoue et al. (2008) (colored horizontal lines). The data suggest that Fig. 3.— Top : Ks band eastern disk FWHM as a function of Fig. 4.— Top : PA of the eastern side of the disk at Ks band scattering grains. This is a reasonable assumption for 4 the eastern sideWest of the disk becomes redder than the westernlarge side(! 1 withµm) increasing grains scattering NIR light (Duchˆene 279 distance from the star. See text for methodology. There is no as a function278 of distance from the star. See text for methodology. 0.5 obvious trend in FWHM vs. stellocentric distance. Bottom:Same The PAdistance increases withfrom increasing theEast star. distance 1–10 µ frommgrainsarethebestfittothewesternside,while the star; over the et al. 2004). We set the dust grain size uniformly to 3 as above, except for the western side of the disk. The FWHM3 same stellocentric1 distances a similar trend was observed at 1.1 2773–10 µmgrainsarethebestmatchtoeasternside.ThismaysuggestthattheµmandsetthediskSBpower-lawindexto-3/2.Fig.7a increases with increasing distance from the star, though this may µm (Debes et al. 2008). Bottom: Same as above, except for the shows the rotated model image, along with the rotated 2780 be an effect from the data reduction (see Appendix). 2 westernwest side. The side PA of increases the diskModel closer is composed to the star, of as smaller is seen in grains the than the east side. Debes276 et al. (2008) 1.1 µmbanddata. final Ks band image. Dashed lines have been inserted to arcseconds to avoid any potential wavelength-dependent changes in1 (A color0 version of this figure is available in the onlineguide journal.) the eye to the apparent10 µm bow-shape and northern −0.5 2008). The western PA increases closer to the star. This 3 µm PA. Comparing our final PISCES Ks band image with 275 offset from the star. Fig. 7b shows the PA vs. stellocen- 277 !! 0 also agrees with the western PA vs. stellocentric distance

the Keck image between 1-2 revealed a difference in Position Angle (degrees) tric distance for the western side of the disk in the model ◦ seen at 1.1 µm(Debesetal.2008)andwiththeoverall 1 µm −1 overall PA of 3.9 ,sowerotatedourPISCESimagebyan−1 274 ◦ −1 image, computed in the same manner as for the real data. −2.5 −2 −1.5 −1 −0.5 ◦ 0 0.5 1 1.5 2 2.5 PA (278.5To)measuredforthewesternsideofthediskat constrain the characteristic dust grain sizes, we compared 276 additional 3.9 clockwise. Fig. 1a shows this correctly- !! The PA increases closer to the star, as is observed in the arcseconds 0.6 µmbetween1.5-12273 (Kalas et al. 2007). rotated image. ourL’) (mags) ∆(Ks L#) disk colors to! the models ofreal Inoue data (Fig. et al. 4b). (2008). We calculated− the PA of the disk at L in the same To test for any warps in the disk, we calculated the PA − While the parameters used in our model are not a manner272 as− for2 the Ks band image, except we used 3 1 5 These0.8 1 models1.2 1.4 calculate1.6 1.8 2 disk2.2 colors2.4 2.6 in the NIR and assume 275 of the disk at Ks band on both sides of the disk as a func- (Ks ∼distance!! from star!! (arcseconds) ! unique explanation for the observed features of the disk,

pixel by 11∆ pixel ( 0. 13 by 0. 47) boxes. The L image tion of distance from the star, using the disk midplane silicate dust grain composition. The modeledthe proposition grain sizes that range the effects can be explained by the Model reproduces PA well 4 was again binned by a factor of 4. We do not plot PA vs.

Position Angle (degrees) pixel locations from the Gaussian fits. The errors were disk’s geometric orientation in space is attractive because 0.5 distancefrom from−3 0.1 theµ starmto10 due(b) to theµmandassumeasinglegrainsizeforthe poor spatial coverage 274 measured as the difference between these values and the it explains the observations without contradicting the 3 ofFig. the 7.— diskTop and,: Rotated even after model binning, and real theKs lowband SNRE disk images. (∼ 2- (thoughcenter of light not along each unique) 1D row at each discrete pixel dust population. We plot◦ the model diskevidence colors forsupporting 1, 3, the and ISM interaction interpretation 3);The however model image the has measured an inclination values of 87 for, scatteringboth sides parameter of the 0 distance from the star. The errors also included the 12 g =0.5, 3 µm grains, a SB power-law index of -3/2, and a gap from (Debes et al. 2009). !! disk"" were10 µ globallymgrains.TheseareshownasthecoloredlinesinFigure consistent with the values measured 8. 273 pixel = 0. 194 astrometric centroiding uncertainty. Fig. 0-1. 1. The model−4 image has been convolved with the PSF from arcseconds 1 attheKs photometricband (this image is of also HBC evident 388, and by Gaussian examination noise has of been the 4.2. Disk Color and Grain Sizes 4a and Fig. 4b show the Ks band PA of the eastern and We also modeled2 0.1 µmand0.3µmgrains,buttheircolors −0.5 Ksadded.band Units and areL in! mJy/arcseconddisk imagesRodigas/NCAD/July infor Fig. both 1a images. and TheFig. 20, white 2a, 2012 western sides of the disk as a function of distance from0 dot marks the location of the star in both images. The dashed lines To determine the disk color as a function of distance respectively).were too blue to be supported by the data. ! ! 272 the star, respectively. The eastern PA increases with in- are meant to− guide5 the eye to the apparent northern offset of the from the star, we calculated ∆(Ks - L )=(Ks - L ) 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 Disk −1 creasing distance from the star, similar to what is seen−1 disk relativeFrom to the star. FigureBottom8:,weseethat1–10 PA vs. stellocentric distanceµ form sized-(Ks grains- L!) arein the the best regions of spatial overlap between −2.5 −2 −1.5 −1distance−0.5 from0 star0.5 (arcseconds)1 1.5 2 2.5 3.3. Surface Brightness Profiles Star over the same spatial locations at 1.1 µm(Debesetal. the western side of the disk in the modeldistance image. Thefrom PA star increases (arcseconds)the Ks band and L! images (1.1-1!!. 45). The errors were arcseconds closer tomatch the star due to to the the western apparent bow-shape, side of similar the to disk, what while 3–10 µmarethe ! (b) is seen for the real data shown in Fig. 4b. calculated by summing the individual Ks and L errors (a) bestFigure match 8. Disk to color the eastern vs. distance side. from This the suggestsstar, expressed that as the∆(Ks observedL#) in quadrature. The− data= suggest that the eastern side Figure 7. (a) Rotated model and real Ks-band disk images. The model image and 2a),scattered(Ks as wellL as#)Disk lightthe PA(Ks from vs. stellocentricL the#)Star disk.Thisiscalculatedonlywherethediskisdetected distanceis tracing plots the largeof the parent disk becomes body dust redder than the western side with for both sides− of the disk− (Fig.− 4a and 4b) suggests that increasing distance from the star. Interior to 1!!. 3both has280 an inclination of 87◦,scatteringparameterg 0.5, 3 µmgrains,anSB grainsat both the in theKs band disk. and The at L data#.Toconstraindustgrainsize,wealsoplotmodel also may suggest that the west side power-law index of 3/2, and a gap from 0 to 1. 1.= The model image has been the debriscolors disk from has a Inoue bow-like et al. shape (2008 and) (colored is offset by horizontal a few lines).sides of The the data disk suggest are grey. that − ## AU toof the the north disk from is composedthe star. Specifically of smaller the o grainsffsets thanTo the constrain east side. the characteristic The dust grain sizes, we convolved with the PSF from the photometric image of HBC 388, and Gaussian ∼ the!! eastern side of the!! disk becomes! redder than the western side with increasing! 279 2 are 0. 012 (0.5 AU) and 0. 04 (2 AU) at L ,fortheeast- compared our ∆(Ks - L )diskcolorstothemodelsof noise has been added. Units are in mJy arcsec− for both images. The white ern andblowoutdistance western from sides, grain the respectively; sizestar. 1–10 forµ this andmgrainsarethebestfittothewesternside,while∼ system0!!. 11 (5 AU)is expectedInoue to et al. be (2008).1–3 Theseµm models calculate disk colors dot marks the location of the star in both images. The dashed lines are meant and 0!!. (Hahn173–10 (8 AU)µmgrainsarethebestmatchtoeasternside.Thismaysuggestthatthe2010 at Ks),band assuming for the eastern a stellar and western mass of 1.3in theM NIR, a and luminosity∼ assume silicate dust grain composition. 278 sides, respectively. Since the astrometric uncertainty in The modeled% grain3 sizes range from 0.1 µmto10µmand to guide the eye to the apparent northern offset of the disk relative to the star. ofwest 3.3 sideL of!!,andagraindensityintherange1–2.5gcm the disk is!! composed! of smaller grains than the east side..Grains centroiding is ∼ 0. 002 and 0. 0194 at L and Ks band, assume a single− grain size for the dust population. We (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller(A color versionthan% the of blowout this figure size is available would in be the blown online out journal.) radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, respectively, it is unlikely that the northern offsets could plot the model disk colors for 1, 3, and 10 µmgrains. 277 be dueeastern to centroid side, error. these Bow-like grains disk would morphologies hit the ISMThese and arecould shown be as blown the colored lines in Fig. 8. We also similar to what is seen for the real data shown in Figure 4(b). have been seen in systems moving in near-perpendicular modeled 0.1 µmand0.3µmgrains,buttheircolorswere (A color version of this figure is available in the online journal.) directionsbackTo to to their constrain the disk west, major the resulting axis characteristic positions in the (e.g., western dust HD grain sidetoo sizes, blue being to we be dominated supportedcompared by the data. 276 61005by (Hines smaller et al. 2007)blue-scattering and HD 32297 grains. (Debes Our et al. observationalFrom Fig. constraints 8, we see that 1-10 µm-sized grains are the our ∆(Ks L#) disk colors to the models of Inoue et al. (2008). proposition that the effects can be explained by the disk’s geo- onThese the dust models− grain calculate sizes offer disk some colors support in forthe these NIR predictions. and assume 275 metric orientation in space is attractive because it explains the silicate dust grain composition. The modeled grain sizes range 4.3. Does the Disk have a Gap? observationsPosition Angle (degrees) without contradicting the evidence supporting the from 0.1 µmto10µmandassumeasinglegrainsizeforthe ISM274 interaction interpretation (Debes et al. 2009). dustThe population. SB profiles We at Ks plotband the modeland at L disk# drop colors off or for flatten 1, 3, out and (at10 lowµmgrains.TheseareshownasthecoloredlinesinFigure S/N) near 1 .Foranedge-ondiskwithnogap,theSB8. 273 4.2. Disk Color and Grain Sizes ## shouldWe also continue modeled to 0.1 increaseµmand0.3 closer toµmgrains,buttheircolors the star. Because we do were too blue to be supported by the data. To272 determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star hasFrom a gap Figure interior8,weseethat1–10 to 1.1. This is consistentµm sized with grains the are prediction the best distance− from# = star (arcseconds)− # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# ofmatch a gap to near the 1 western## by Mo sideor´ et of al. the (2011b disk,), while using 3–10 spectralµmarethe energy (b) images (1##.1–1##.45). The errors were calculated by summing the distributionbest match to (SED) the eastern analysis side. of the This system. suggests Though that the degenerate observed Figureindividual 7. (a)Ks Rotatedand modelL# errors and real in quadrature.Ks-band disk The images. data The suggest model image that withscattered temperature light from and the dust disk grain is tracing size, the their large best-fit parent model body dust of has an inclination of 87 ,scatteringparameterg 0.5, 3 µmgrains,anSB the eastern side of the◦ disk becomes redder= than the western side HDgrains 15115’s in the infrared disk. The SED data yields also may a two-component suggest that the disk, west with side power-law index of 3/2, and a gap from 0 to 1##. 1. The model image has been convolvedwith increasing with the PSF− distance from the from photometric the star. image The of western HBC 388, side and Gaussian of the theof the inner disk at is 4 composed AU and the of outer smaller at 42 grains AU (both than the2AU).Given east side. The 2 ± noisedisk has becomes been added. slightly Units bluer are in closer mJy arcsec to the− for star. both images. The white adistancetothestarof45.2pc,theouteredgeofthegapwouldblowout grain size for this system is expected to be 1–3 µm dot marks the location of the star in both images. The dashed lines are meant (Hahn 2010), assuming a stellar mass of 1.3 M , a luminosity∼ % 3 to guide the eye to the apparent northern offset of the disk relative to the star. 9 of 3.3 L ,andagraindensityintherange1–2.5gcm− .Grains (b) P.A. vs. stellocentric distance for the western side of the disk in the model smaller than% the blowout size would be blown out radially; on the image. The P.A. increases closer to the star due to the apparent bow shape, similar to what is seen for the real data shown in Figure 4(b). eastern side, these grains would hit the ISM and could be blown (A color version of this figure is available in the online journal.) back to the west, resulting in the western side being dominated by smaller blue-scattering grains. Our observational constraints proposition that the effects can be explained by the disk’s geo- on the dust grain sizes offer some support for these predictions. metric orientation in space is attractive because it explains the observations without contradicting the evidence supporting the 4.3. Does the Disk have a Gap? ISM interaction interpretation (Debes et al. 2009). The SB profiles at Ks band and at L# drop off or flatten out 4.2. Disk Color and Grain Sizes (at low S/N) near 1##.Foranedge-ondiskwithnogap,theSB should continue to increase closer to the star. Because we do To determine the disk color as a function of distance from the not see this in our data, this may be an indication that the disk star, we calculated ∆(Ks L ) (Ks L )Disk (Ks L )Star has a gap interior to 1.1. This is consistent with the prediction − # = − # − − # ∼ ## in the regions of spatial overlap between the Ks band and L# of a gap near 1## by Moor´ et al. (2011b), using spectral energy images (1##.1–1##.45). The errors were calculated by summing the distribution (SED) analysis of the system. Though degenerate individual Ks and L# errors in quadrature. The data suggest that with temperature and dust grain size, their best-fit model of the eastern side of the disk becomes redder than the western side HD 15115’s infrared SED yields a two-component disk, with with increasing distance from the star. The western side of the the inner at 4 AU and the outer at 42 AU (both 2AU).Given disk becomes slightly bluer closer to the star. adistancetothestarof45.2pc,theouteredgeofthegapwould±

9 Large Grains in the HD 15115 Debris Disk 7

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! West!! brighter than East eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! West!! brighter than East eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! West!! brighter than East eastern side at band over the same~ equal stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux Ks East !! Ks West in circular apertures with a radius of 0. 15 centered on 13.5 ) the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- 14 ture size because it maximized the SNR of the disk flux ! ! at L .ThiswasnecessarybecausetheL disk is de- 14.5 tected at low SNRE (∼ 2.5), so using larger apertures incorporated more background noise. We measured the 15 errors as the noise value at each disk midplane location, using the noise images created for the SNRE maps (see 15.5 Fig. 1b and Fig. 2b). We computed the aperture correc- tion at Ks band to be 0.64 by convolving the image of 16 Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unity 16.5 counts/pixel and ∼ equal widthSurface to the observed diskBrightness at Profiles Ks band. The SB as a function of distance from the star 17 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. !! Large Grains in theBetween HD 15115∼ 1.2-2 Debris. 1(wherebothsidesofthedisk Disk 7 (a) are detected), the typical western disk SB is ∼ a For both the Ks band and the L! disk images, wemagnitude/arcsecond ob- 2 brighter than2.15 the µm eastern side. A 3.8 µm 13 11 tained the SB profiles by calculating the median disk fluxsimilar brightness asymmetry was seen at 1.1 µm(DebesKs East !! L’ East in circular apertures with a radius of 0. 15 centeredet on al. 2008). It is also interesting that the west-Ks West L’ West !! ) 13.5 ∼ ) 12 2 the Gaussian-fitted disk midplane location at each dis-ern SB appears to drop off near 1 ,by half a 2 crete pixel distance from the star. We chose this aper-magnitude/arcsecond2 (although the noise is high inte- ture size because it maximized the SNR of the disk fluxrior to this distance).14 We saw this SB reduction in our 13 at L!.ThiswasnecessarybecausetheL! disk isfinal de- ADI image as well as in our final LOCI image (Fig. tected at low SNRE (∼ 2.5), so using larger apertures1a), even after14.5 correction with insertion and recovery of 14 incorporated more background noise. We measuredan the artificial disk. The drop in SB near 1!! was not seen errors as the noise value at each disk midplane location,at shorter wavelengths15 (Kalas et al. 2007; Debes et al. 15 using the noise images created for the SNRE maps (see2008), which raises suspicion that the feature may not Fig. 1b and Fig. 2b). We computed the aperture correc-be real. However15.5 the shorter wavelength results appear 16 tion at Ks band to be 0.64 by convolving the imageto of be limited by PSF residuals interior to 1!!,soitmay Surface Brightness (mags/arcsecond the photometric reference HBC 388 with a bar of unitynot be appropriate16 to compare to these data. Surface Brightness (mags/arcsecond 17 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 18 is shown in Fig. 5a. “Magnitudes” in this paper refers 1 TABLE1.5 1 2 2.5 1 1.1 1.2 1.3 1.4 1.5 band SB power-lawdistance from indices star (arcseconds) distance from star (arcseconds) to Vega magnitudes. Ks Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk Side index(a) (b) ∼ West (1-1!!. 2) 3.48 ± 0.5 are detected), the typical western disk SB is a !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western 2 West (1. 2 1. 8) −4.40 ± 0.24 Ks similar brightness asymmetry was seen at 1.1 µm(Debes !! West!! brighter than East eastern side at band over the same~ equal stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 since this has been corrected for. Bottom: SB profile of the disk ! !! magnitude/arcsecond2 (although the noise is high inte- at L . Both sides of the disk are ∼Rodigas/NCAD/Julyequally bright beyond 20, 2012 1. 2, and just as in the Ks band image there is low SNR evidence for a rior to this distance). We saw this SB reduction in ourTable 1 shows13 the power-law indices measured for the !! final ADI image as well as in our final LOCI image (Fig.eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. 1a), even after correction with insertion and recoverypower-law of indices14 do not agree very well with the indices ! !! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, an artificial disk. The drop in SB near 1 was not seen using the same size aperture as we did for the Ks band at shorter wavelengths (Kalas et al. 2007; Debes etcies al. could be explained15 by the differing spatial coverage of the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 2008), which raises suspicion that the feature may not to increase the signal-to-noise per pixel. We computed be real. However the shorter wavelength results appearalso possible that16 the disk SB falls differently at Ks band ! !! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving to be limited by PSF residuals interior to 1 ,soitmay the photometric image of HD 15115 with a bar of unity at these wavelengthsSurface Brightness (mags/arcsecond would help clarify this matter. We not be appropriate to compare to these data. 17 ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor 18 the Ks band SB profile. Fig. 5b shows the SB profile of TABLE 1 Using the Ks1 band power-laws,1.1 1.2 we can1.3 quantify1.4 the 1.5 ! Ks band SB power-law indices distance from star (arcseconds)!! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! !! !! beyond 1. 2, which is not seen at Ks band. Interestingly, Side index the power-law index for 1. 2 1. 8) −4.40 ± 0.24 This suggests that weKs are seeing a ∼ 2σ reduction in SB !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1. 4) −1.1 ± 0.07 We calculatedcannot simply the SB be explained vs. distance by flux from loss the due starto disk for self-subtraction, the Limits on Planets since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux A SB drop-off? Ks East !! in circular apertures with a radius of 0. 15 centered on Ks West

) 13.5 the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distance from the star. We chose this aper- ture size because it maximized the SNR of the disk flux 14 at L!.ThiswasnecessarybecausetheL! disk is de- Rodigas/NCAD/July 20, 2012 tected at low SNRE (∼ 2.5), so using larger apertures 14.5 incorporated more background noise. We measured the errors as the noise value at each disk midplane location, 15 using the noise images created for the SNRE maps (see Fig. 1b and Fig. 2b). We computed the aperture correc- 15.5 tion at Ks band to be 0.64 by convolving the image of the photometric reference HBC 388 with a bar of unity Surface Brightness (mags/arcsecond 16 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk (a) are detected), the typical western disk SB is ∼ a magnitude/arcsecond2 brighter than the eastern side. A 11 similar brightness asymmetry was seen at 1.1 µm(Debes L’ East et al. 2008). It is also interesting that the west- L’ West !! ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 magnitude/arcsecond2 (although the noise is high inte- rior to this distance). We saw this SB reduction in our 13 final ADI image as well as in our final LOCI image (Fig. 1a), even after correction with insertion and recovery of 14 an artificial disk. The drop in SB near 1!! was not seen at shorter wavelengths (Kalas et al. 2007; Debes et al. 15 2008), which raises suspicion that the feature may not be real. However the shorter wavelength results appear 16 to be limited by PSF residuals interior to 1!!,soitmay not be appropriate to compare to these data. Surface Brightness (mags/arcsecond 17

18 TABLE 1 1 1.1 1.2 1.3 1.4 1.5 Ks band SB power-law indices distance from star (arcseconds) Side index (b) West (1-1!!. 2) 3.48 ± 0.5 !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western West (1. 2 1. 8) −4.40 ± 0.24 Ks !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1!!. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux A SB drop-off? Ks East !! where it should be in circular apertures with a radius of 0. 15 centered on Ks West

) 13.5 the Gaussian-fitted disk midplane location at each dis- where2 it actually is crete pixel distance from the star. We chose this aper- ture size because it maximized the SNR of the disk flux 14 at L!.ThiswasnecessarybecausetheL! disk is de- Rodigas/NCAD/July 20, 2012 tected at low SNRE (∼ 2.5), so using larger apertures 14.5 incorporated more background noise. We measured the errors as the noise value at each disk midplane location, 15 using the noise images created for the SNRE maps (see Fig. 1b and Fig. 2b). We computed the aperture correc- 15.5 tion at Ks band to be 0.64 by convolving the image of the photometric reference HBC 388 with a bar of unity Surface Brightness (mags/arcsecond 16 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk (a) are detected), the typical western disk SB is ∼ a magnitude/arcsecond2 brighter than the eastern side. A 11 similar brightness asymmetry was seen at 1.1 µm(Debes L’ East et al. 2008). It is also interesting that the west- L’ West !! ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 magnitude/arcsecond2 (although the noise is high inte- rior to this distance). We saw this SB reduction in our 13 final ADI image as well as in our final LOCI image (Fig. 1a), even after correction with insertion and recovery of 14 an artificial disk. The drop in SB near 1!! was not seen at shorter wavelengths (Kalas et al. 2007; Debes et al. 15 2008), which raises suspicion that the feature may not be real. However the shorter wavelength results appear 16 to be limited by PSF residuals interior to 1!!,soitmay not be appropriate to compare to these data. Surface Brightness (mags/arcsecond 17

18 TABLE 1 1 1.1 1.2 1.3 1.4 1.5 Ks band SB power-law indices distance from star (arcseconds) Side index (b) West (1-1!!. 2) 3.48 ± 0.5 !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western West (1. 2 1. 8) −4.40 ± 0.24 Ks !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1!!. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

For both the Ks band and the L! disk images, we ob- 13 tained the SB profiles by calculating the median disk flux A SB drop-off? Ks East !! where it should be in circular apertures with a radius of 0. 15 centered on Ks West

) 13.5 the Gaussian-fitted disk midplane location at each dis- where2 it actually is crete pixel distance from the star. We chose this aper- ture size because it maximized the SNR of the disk flux 14 ! ! at L .ThiswasnecessarybecausetheL disk is de- Low-SNR evidence for a gap! Rodigas/NCAD/July 20, 2012 tected at low SNRE (∼ 2.5), so using larger apertures 14.5 incorporated more background noise. We measured the errors as the noise value at each disk midplane location, 15 using the noise images created for the SNRE maps (see Fig. 1b and Fig. 2b). We computed the aperture correc- 15.5 tion at Ks band to be 0.64 by convolving the image of the photometric reference HBC 388 with a bar of unity Surface Brightness (mags/arcsecond 16 counts/pixel and ∼ equal width to the observed disk at

Ks band. The SB as a function of distance from the star 16.5 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes. Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk (a) are detected), the typical western disk SB is ∼ a magnitude/arcsecond2 brighter than the eastern side. A 11 similar brightness asymmetry was seen at 1.1 µm(Debes L’ East et al. 2008). It is also interesting that the west- L’ West !! ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 magnitude/arcsecond2 (although the noise is high inte- rior to this distance). We saw this SB reduction in our 13 final ADI image as well as in our final LOCI image (Fig. 1a), even after correction with insertion and recovery of 14 an artificial disk. The drop in SB near 1!! was not seen at shorter wavelengths (Kalas et al. 2007; Debes et al. 15 2008), which raises suspicion that the feature may not be real. However the shorter wavelength results appear 16 to be limited by PSF residuals interior to 1!!,soitmay not be appropriate to compare to these data. Surface Brightness (mags/arcsecond 17

18 TABLE 1 1 1.1 1.2 1.3 1.4 1.5 Ks band SB power-law indices distance from star (arcseconds) Side index (b) West (1-1!!. 2) 3.48 ± 0.5 !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western West (1. 2 1. 8) −4.40 ± 0.24 Ks !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1!!. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

Large Grains in the HD 15115 Debris Disk 7 For both the• ForKs band an edge-on and the L! disk debris images, disk, we ob- the 13 tained the SB profiles by calculating the median disk flux Ks East !! in circular aperturesSB should with a radius increase of 0. 15 closer centered to on the Ks West

) 13.5 the Gaussian-fitted disk midplane location at each dis- 2 crete pixel distancestar from the star. We chose this aper- ture size because it maximized the SNR of the disk flux 14 at L!.Thiswasnecessarybecausethe• A flattening/drop-offL! disk would is de- tected at low SNRE (∼ 2.5), so using larger apertures 14.5 incorporated morethen background indicate noise. a deficit We measured in the errors as the noise value at each disk midplane location, 15 using the noisereflecting images created material for the SNRE (ie, maps less/no (see Fig. 1b and Fig.dust) 2b). We computed the aperture correc- 15.5 tion at Ks band to be 0.64 by convolving the image of the photometric reference HBC 388 with a bar of unity Surface Brightness (mags/arcsecond 16 counts/pixel• andSED∼ equal suggests width to thea two-component observed disk at

Ks band. The SB as a function of distance from the star 16.5 disk, with a gap near 1” (45 1 1.5 2 2.5 is shown in Fig. 5a. “Magnitudes” in this paper refers distance from star (arcseconds) to Vega magnitudes.AU) (Moor et al. 2011b) Between ∼ 1.2-2!!. 1(wherebothsidesofthedisk (a) are detected), the typical western disk SB is ∼ a magnitude/arcsecond2 brighter than the eastern side. A 11 similar brightness asymmetry was seen at 1.1 µm(Debes L’ East et al. 2008). It is also interesting that the west- L’ West SED + observations!! pointing to a gap near 45 AU ∼ ) 12 ern SB appears to drop off near 1 ,by half a 2 magnitude/arcsecond2 (although the noise is high inte- rior to this distance). We saw this SB reduction in our 13 Rodigas/NCAD/July 20, 2012 final ADI image as well as in our final LOCI image (Fig. 1a), even after correction with insertion and recovery of 14 an artificial disk. The drop in SB near 1!! was not seen at shorter wavelengths (Kalas et al. 2007; Debes et al. 15 2008), which raises suspicion that the feature may not be real. However the shorter wavelength results appear 16 to be limited by PSF residuals interior to 1!!,soitmay not be appropriate to compare to these data. Surface Brightness (mags/arcsecond 17

18 TABLE 1 1 1.1 1.2 1.3 1.4 1.5 Ks band SB power-law indices distance from star (arcseconds) Side index (b) West (1-1!!. 2) 3.48 ± 0.5 !! !! Fig. 5.— Top : SB profile of the disk at Ks band. The western West (1. 2 1. 8) −4.40 ± 0.24 Ks !! !! eastern side at band over the same stellocentric distances. The East (1. 2 1!!. 4) −1.1 ± 0.07 cannot simply be explained by flux loss due to disk self-subtraction, since this has been corrected for. Bottom: SB profile of the disk at L!. Both sides of the disk are ∼ equally bright beyond 1!!. 2, and just as in the Ks band image there is low SNR evidence for a Table 1 shows the power-law indices measured for the !! eastern and western sides of the disk at Ks band. The decrease/flattening in SB near 1-1. 2. power-law indices do not agree very well with the indices ! at 1.1 µm reported by Debes et al. (2008). The discrepan- L image in the same manner as for the Ks band image, cies could be explained by the differing spatial coverage of using the same size aperture as we did for the Ks band the disk at the two wavelengths. On the other hand, it is image. Additionally the image was binned by a factor 4 to increase the signal-to-noise per pixel. We computed also possible that the disk SB falls differently at Ks band ! than at 1.1 µm. Additional high-contrast imaging data the aperture correction at L to be 0.63 by convolving at these wavelengths would help clarify this matter. We the photometric image of HD 15115 with a bar of unity ! counts/pixel and ∼ equal width to the measured disk at do not calculate the power-law indices at L due to the ! lower SNRE and poor spatial coverage of the disk. L .Wecomputedtheerrorsinthesamemannerasfor the Ks band SB profile. Fig. 5b shows the SB profile of Using the Ks band power-laws, we can quantify the ! !! the disk at L .Bothsidesofthediskare∼ equal in SB significance of the SB reduction interior to 1. 1. From !! the power-law index for 1!!. 2

2

1

0 10 µm 3 µm

1 µm −1 L’) (mags) − −2 (Ks Δ

−3

−4 West 8 Rodigas et al. East

−5 We do not detect any high SNR point-sources in any 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 Fig. 9.— Equilibrium disk temperature vs. distance from the distance from star (arcseconds) 0.8star, for several different silicate grain sizes. See text for methodol- of our data reductions that would point to a possible 2.5 ogy. The horizontal dashed lines represent temperature constraints . To set firm constraints on what we could haveFig. 8.— Disk color vs. distance from the star, expressed as 0.7of 179 K and 57 K from Mo´or et al. (2011b). Given that we have ! 2 ! ! ∼ !! detected, we reduced the same data that was used to∆( de-Ks - L )=(Ks - L )Disk -(Ks - L )Star. This is calculated observational evidence for the predicted gap at 1 , we can inde- ! ! pendently constrain the dust grain sizes to be ∼ 3 µm for astro- tect the disk at L with an aggressive LOCI algorithm.only where the disk is1.5 detected at both Ks band and at L .To 0.6 ! constrain dust grain size, we also plot model colors from Inoue nomical silicates. We examined only the L data, as opposed to includ- 1 et al. (2008) (colored horizontal lines). The data suggest that the 0.5emission energy of a particle with adopted dust prop- ing the Ks band data as well, because we had nearlyeastern side of the disk becomes redder than the western side with 0.5 erties for astronomical silicates (Laor & Draine 1993). ! ∼ increasing distance from the star. 1-10 µm grains are the best fit 0.4 double the integration time at L ( 40 minutes vs. 20 Under an optically thin condition, the heating is solely to the western side, while0 3-10 µm grains are the best match to minutes at Ks band) and double the parallactic angle 0.3 ◦ ◦ eastern side. This may suggest that the west side of the disk is from the central star, representative as the best-fit Ku- ∼ ∼ arcseconds rotation ( 40 vs. 20 at Ks band). For thecomposed ag- of smaller− grains0.5 than the east side. rucz model with a stellar temperature of 7000 K and a 0.2 gressive LOCI reduction we required only 0.75 FWHM stellar luminosity of 3.3 L!. best match to the−1 western side of the disk, while 3-10 of parallactic angle rotation between images, and our op- 0.1 µmarethebestmatchtotheeasternside.Thissug- The horizontal dashed lines represent the predicted timization section size (NA)was100.Ourfinalimage(in Limits−1.5 on a potentialdisk planet temperatures ofinside 179 K and 57the K from gap Mo´or et al. gests that the observed scattered light from the disk is 0 units of σ,whereσ was calculated in concentric annuli (2011b). Assuming we have observationally detected the ∼ !! tracing the large parent−2 body dust grains in the disk. the size of the PSF FWHM ( 0. 1)), shown in Fig. 6a, −0.1inner edge of the outer disk component (and hence the The data also may•−2.5 suggestCan use that well-known the west side chaotic of the disk zone equation to"" set limit on planet mass vs. reveals no 5σ point-source detections. gap) at ∼ 1. 1, the parent body dust grain sizes are con- is composed of smallersemimajor−2.5 −2 grains−1.5 −1 than axis−0.5 the0 0.5 east1 side.1.5 2 The2.5 ∼ To ascertain the minimum flux density point-sourceblowout ob- grain size for this system isarcseconds expected to be ∼ 1- strained to be 3 µm. This is consistent with our dust ject we could have detected, we inserted artificial planets grain size estimates from the disk colors. 3 µm(Hahn2010),assumingastellarmassof1.3• M!, into the raw data and re-ran our aggressive LOCI reduc- Combine this with(a) our observational constraints on planets in the systemLarge to Grains set in the HD 15115 Debris Disk 11 aluminosityof3.3L!,andagraindensityintherange 4.4. tion. The artificial planets were bright PSFs obtained1-2.5 g/cm3.Grainssmallerthantheblowoutsizewouldstringent limits Limits on a planet inside the gap from the unsaturated images of HD 15115. We didbe not blown out radially; on the eastern side, these grains We can also calculate an independent estimate of the with SED analysis, this may be a sign that the disk scale down the PSFs in brightness. Instead we simply mass of a planet creating the gap in the disk using the !! would hit the ISM and be blown back to the west, re- 35 has a gap interior to 1 (45 AU). Additional high- 3 M , 10 Myr calculated the SNR for each artificial planet’s locationsulting in the western15.5 side being dominated by smallerJup equation describing the relationship between the width contrast observations at NIR wavelengths with bet- (inside a radius = FWHM aperture) and determined the of the chaotic zone30 M around (1 Gyr) observational an assumed constraint coplanar, low- blue-scattering grains. Our observational constraints on 30 J ter inner working angles would help confirm or dis- 30 M , 1 Gyr flux that would have resulted in a 5σ detection. Thesethe dust grain sizes offer some support for these predic-Jup eccentricity planet and its mass and semimajor axis prove the proposition that the disk has a gap. are plotted in Fig. 6b. We reach a 5σ background limittions. 16 (Malhotra 1998): 25 !

!! 2 M , 10 Myr ) of ∼ 17.8 mags (in 40 mins of integration). At 0. 6, the Jup J 6. The Ks - L disk color is mostly grey for both ≈ 2/7 !! contrast (computed by subtracting the star’s L! magni- 4.3. Does the disk have a gap? ∆a 1.4 ap (Mp/M∗) , (1) sides of the disk between 1.1-1. 45 (50-66 AU). 1-10 16.5 20 tude of 5.763) is ∼ 9.5 mags. These sensitivity and con- " µmgrainsizesarethebestmatchtothewestern The SB profiles at Ks band and at L drop off or flatten where ∆a is the width of the chaotic zone, ap and Mgapp are trast limits show the incredible potential of direct imag- "" disk color. 3-10 µmgrainsizesarethebestmatch out (at low SNR)17 near 1 .Foranedge-ondiskwithno the semimajor15 axis and mass of the planet, respectively, sensitivity (mags) ing with LBTI/LMIRcam. For comparison, with Clio-2 Planet Mass (M for the eastern side. Given the system’s expected gap, the SBσ should continue to increase closer to the and M∗ is the mass of the star. We make the assumption 5 grain blowout size of 1-3 µm, our dust grain size at the 6.5 m MMT, a 5σ background limit of 17.8 magsstar. Because we do not see this in our data, this may be that, from10 our SB profiles, the outer edge of the gap, ! "" "" constraints may support the ISM interaction inter- at L was achieved in 2.5 hours of integration (Rodigasan indication that17.5 the disk has a gap interior to ∼ 1. 1. rgap,isat1. 1(50AU).Since∆a = rgap − ap,wecan "" pretation (Debes et al. 2009), which predicts that et al. 2011). This is consistent with prediction of a gap near 1 by substitute5 this into Eq. 1 and solve for the planet mass small grains would be blown to the western side We also overplot the magnitudes of 1-3 MJ plan-Mo´or et al. (2011b), using spectral energy distribution1 M , 10 Myr as a function of semimajor axis (Fig. 10). 18 Jup 3 M (10 Myr) observational constraint ets (10 Myr old) and, conservatively, the magnitude(SED) of analysis of the system. Though degenerate with The planet’sJ mass approaches zero the closer it orbits of the disk, leaving large, unaffected grains on the 0.5 1 1.5 2 2.5 34 36 38 40 42 44 46 eastern side. SED analysis, combined with our ob- a30MJ brown dwarf (1 Gyr old) from the Baratemperatureffe and dust grain size, their best-fit model of to the disk edge and approachessemimajor axis the (AU) brown dwarf mass !! et al. (2003) COND mass-luminosity models (horizon-HD 15115’s infrared SED yieldsdistance a two-component from star (arcseconds) disk, regime closer to the star. Assuming the gap was created servational evidence for a gap near 1 (45 AU), also ∼ tal dashed lines). Assuming a young stellar age, atwith 5σ the inner at 4 AU and the outer at(b) 42 AU (both ±2 byFig. an object 10.— Masses whose of semimajor a possible object axis =orbiting its projected inside a disk sep- gap predicts a dust grain size of 3 µm. Planet constrained to ~ 3-30 MJ and ~ !!34-45 AU orbit confidence we rule out planets more massive than AU).∼ 1- Given a distance to the star of 45.2! pc, the outer arationwith its outer at the edge epoch at 1. 1 of (50 our AU), observations as a function of and semimajor that the axis, ! !! !! Fig. 6.— Top :Finalimageat"" ± L"",inunitsofσ, computed withcomputed an using Eq. 1 (solid line). The dashed line represents our 7. We do not detect any ! 5σ point-sources at L in- 2MJ outside of 1 ,andoutsideof0. 6wecanruleoutedge of the gap would be at 0. 93 0. 04. This agrees system is young, we can say that the object must be ! a aggressive LOCI reduction. See text for"" methodology. The3M whiteJ (10 Myr old age) observationalRodigas/NCAD/July constraint and assumes 20, 2012 that dicative of planets. We rule out companions more planets more massive than ∼ 3MJ .IfHD15115’sagewell with thedot marks observed the location drop off ofsinSBnear1 the star and the. size of a resolutionfew ele- MJ because we could have detected " 3MJ planets !! ! the planet’s semimajor axis = its projected separation at the epoch massive than ∼ 3M beyond 0. 6ifthestaris We canment independently at L . No high constrain SNR point-sources the parent are body detected. dust Bottom(dashedof:5 ourσ observations. line in Fig. The 10). dash-dot The line planet’s is the same, semimajor except showing axis J is much older, then we can only rule out brown dwarfs ! "" !! !! grain sizesensitivity using estimates curve for of our theL disk’sdata, equilibrium computed by tem- inserting artificialwouldour 30 then MJ (1 be Gyr constrained old age) observational to be between constraint. 0.87-1 If HD(40-45 15115 young and more massive than 30 MJ beyond 0. 7 more massive than ∼ 30 MJ beyond 0. 7. perature asplanets a function into the of rawdistance data and from running the star, our for aggressive differ- LOCIAU). reduc-is old, If then the the system object is creating old, the the object gap must must reside be between less than∼ 34- if the star is old. Independently we constrain the tion. We reach a background limit of ∼ 17.8 mags, and a contrast∼45 AU. If the system is young, the allowed semimajor axis range mass of a single, coplanar, low-eccentricity planet ent grain sizes∼ (Fig. 9). The∼ !! equilibrium dust tempera- 30 MJ and its semimajor axis would be constrained to 4. INTERPRETATIONS of 9.5 mags at 0. 6. The horizontal dashed lines representshrinks the"" to 40-45 AU. ∼ ture was computed by balancing the absorption and the 0.75-1 (34-45 AU). creating the gap in the disk to be afewMJ if it is magnitudes of 10 Myr old and 10 Gyr old planets, from the BaraItffe is certainly possible that we did not detect the pu- 4.1. Disk Structure et al. (2003) COND mass-luminosity models. We rule out planets close to the disk edge, and to be in the brown dwarf ∼ tative planet because it is currently in front of or behind It has been suggested that HD 15115 is interacting with more massive than 3MJ if the star is young and 30 MJ if the regime if its orbit is closer to the star. Assuming star is old. the star. It is also possible that the gap has not been cre- the object’s semimajor axis = its projected sepa- the local (ISM), given its space mo- ated by a planet, but instead arises from other dynamical ration at the epoch of our observations, the planet tion to the south-east nearly parallel with its disk major ! Since the disk is symmetric at L ,whichissensitivetointeractions. Additional high-contrast, high angular res- would be " 3M and orbit between 0.87-1!! (40-45 axis PA (Debes et al. 2009) and the disk asymmetries J large, grey-scattering grains, our data support theolution ISM- imaging of the debris disk would help constrain AU) if the star is young; if the system is old, the seen at shorter wavelengths (Kalas et al. 2007; Debes interaction interpretation. the existence of both the gap and any possible planets. ∼ ∼ 2 object would be less massive than 30 MJ and et al. 2008). The 1magnitude/arcsecond brightness It is also certainly possible that other dynamical ef- orbit between 0.75-1!! (34-45 AU). asymmetry (between 1.2-2!!. 1) and the east-west morpho- fects (e.g., planets or a close passage of a nearby star as 5. SUMMARY logical asymmetry seen in our Ks band data both sup- suggested by Kalas et al. (2007)) are responsible for theWe have presented several intriguing results on the de- port this proposition. These effects could be caused by observed morphological and SB asymmetries. Howeverbris disk surrounding HD 15115. These results are: the eastern side of the disk plowing head-first into the there is currently no supporting evidence for a stellar We thank Piero Salinari for his insight, leadership and ISM. In this case the eastern side would be much more flyby (Kalas et al. 2007; Debes et al. 2008, 2009). Dy-1. We detect east-west asymmetry in the disk mor- persistence which made the development of the LBT affected than the western side, and small grains could be namical modeling of the disk with embedded planets is phology at Ks band, with the western side of the adaptive secondaries possible. We thank the LBTO staff blown out to the west. This could cause the eastern side beyond the scope of this paper, but such modeling would disk being a ∼ magnitude/arcsecond2 brighter than and telescope operators for their hard work facilitating to be more truncated and fainter relative to the western help clarify what exactly is causing the observed asym- the eastern side at the same stellocentric distances. use of the telescope and instruments. We are grateful side, both of which we observe. Any large (> afewµm) metries in the disk. The asymmetry and brightness differences are con- to Elliott Solheid, the lead mechanical engineer on the grains in bound orbits should remain mostly unaffected. Inspection of our Ks band and L! disk images (Fig. 1a sistent with results at shorter wavelengths and lend PISCES project. Roland Sarlot and Andrew Rakich pro- additional evidence to the interpretation that ISM vided support in the PISCES optical design and engineer- interactions are affecting the disk structure (Debes ing, respectively. We acknowledge support for LMIRcam et al. 2009). from the National Science Foundation under grant NSF AST-0705296. We thank A.K. Inoue and M. Honda for 2. At L!,wedetectsymmetricand∼ equally bright sharing their disk modeling data and for helpful discus- disk structure. sions. We thank John Debes, the referee, for helpful discussions and for his careful review of the manuscript. 3. We detect an overall bow-like shape to the disk at ! T.J.R. acknowledges support from the NASA and both Ks band and L ,andthediskappearsoffset Space Science Graduate Fellowship. from the star to the north by a few AU. We are able to explain these observed effects using a model disk with a near edge-on inclination, a gap between 0-1!!. 1, large (3 µm) grains, and mostly forward- scattering grains (g =0.5). 4. The FWHM of the disk at Ks band and at L! is consistent with the disk FWHM at 1.1 µm(Debes et al. 2008) and at 0.6 µm(Kalasetal.2007). 5. The disk SB profile at Ks band shows evidence for a2σ reduction interior to 1!!. 1(50AU).Combined Does10 the disk contain large grains?Rodigas et al.

2

1

0 10 µm 3 µm

1 µm −1 L’) (mags) − −2 (Ks Δ

−3 Gray!

West −4 East

−5 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 Fig. 9.— Equilibrium disk temperature vs. distance from the distance from star (arcseconds) star, for several different silicate grain sizes. See text for methodol- ogy. The horizontal dashed lines represent temperature constraints Fig. 8.— Disk color vs. distance from the star, expressed as of 179 K and 57 K from Mo´or et al. (2011b). Given that we have The disk is ~gray! , contains! 1-10 µm grains,! and shows evidence for ∼ !! ∆(Ks - L )=(Ks - L )Disk -(Ks - L )Star. This is calculated observational evidence for the predicted gap at 1 , we can inde- theonly west where side the disk containing is detected atsmaller both Ks grainsband and than at Lthe!.To east pendently constrain the dust grain sizes to be ∼ 3 µm for astro- constrain dust grain size, we also plot model colors from Inoue nomical silicates. et al. (2008) (colored horizontal lines). The data suggest that the Rodigas/NCAD/Julyemission 20, 2012 energy of a particle with adopted dust prop- eastern side of the disk becomes redder than the western side with erties for astronomical silicates (Laor & Draine 1993). increasing distance from the star. 1-10 µm grains are the best fit to the western side, while 3-10 µm grains are the best match to Under an optically thin condition, the heating is solely eastern side. This may suggest that the west side of the disk is from the central star, representative as the best-fit Ku- composed of smaller grains than the east side. rucz model with a stellar temperature of 7000 K and a best match to the western side of the disk, while 3-10 stellar luminosity of 3.3 L!. µmarethebestmatchtotheeasternside.Thissug- The horizontal dashed lines represent the predicted gests that the observed scattered light from the disk is disk temperatures of 179 K and 57 K from Mo´or et al. tracing the large parent body dust grains in the disk. (2011b). Assuming we have observationally detected the The data also may suggest that the west side of the disk inner edge of the outer disk component (and hence the gap) at ∼ 1"". 1, the parent body dust grain sizes are con- is composed of smaller grains than the east side. The ∼ blowout grain size for this system is expected to be ∼ 1- strained to be 3 µm. This is consistent with our dust grain size estimates from the disk colors. 3 µm(Hahn2010),assumingastellarmassof1.3M!, aluminosityof3.3L!,andagraindensityintherange 4.4. 1-2.5 g/cm3.Grainssmallerthantheblowoutsizewould Limits on a planet inside the gap be blown out radially; on the eastern side, these grains We can also calculate an independent estimate of the would hit the ISM and be blown back to the west, re- mass of a planet creating the gap in the disk using the sulting in the western side being dominated by smaller equation describing the relationship between the width blue-scattering grains. Our observational constraints on of the chaotic zone around an assumed coplanar, low- the dust grain sizes offer some support for these predic- eccentricity planet and its mass and semimajor axis tions. (Malhotra 1998): ≈ 2/7 4.3. Does the disk have a gap? ∆a 1.4 ap (Mp/M∗) , (1) " The SB profiles at Ks band and at L drop off or flatten where ∆a is the width of the chaotic zone, ap and Mp are out (at low SNR) near 1"".Foranedge-ondiskwithno the semimajor axis and mass of the planet, respectively, gap, the SB should continue to increase closer to the and M∗ is the mass of the star. We make the assumption star. Because we do not see this in our data, this may be that, from our SB profiles, the outer edge of the gap, "" "" an indication that the disk has a gap interior to ∼ 1. 1. rgap,isat1. 1(50AU).Since∆a = rgap − ap,wecan This is consistent with prediction of a gap near 1"" by substitute this into Eq. 1 and solve for the planet mass Mo´or et al. (2011b), using spectral energy distribution as a function of semimajor axis (Fig. 10). (SED) analysis of the system. Though degenerate with The planet’s mass approaches zero the closer it orbits temperature and dust grain size, their best-fit model of to the disk edge and approaches the brown dwarf mass HD 15115’s infrared SED yields a two-component disk, regime closer to the star. Assuming the gap was created with the inner at 4 AU and the outer at 42 AU (both ±2 by an object whose semimajor axis = its projected sep- AU). Given a distance to the star of 45.2 pc, the outer aration at the epoch of our observations and that the edge of the gap would be at 0"". 93 ± 0"". 04. This agrees system is young, we can say that the object must be ! a "" well with the observed drop offsinSBnear1 . few MJ because we could have detected " 3MJ planets We can independently constrain the parent body dust (dashed line in Fig. 10). The planet’s semimajor axis grain size using estimates of the disk’s equilibrium tem- would then be constrained to be between 0.87-1"" (40-45 perature as a function of distance from the star, for differ- AU). If the system is old, the object must be less than ent grain sizes (Fig. 9). The equilibrium dust tempera- ∼ 30 MJ and its semimajor axis would be constrained to ture was computed by balancing the absorption and the 0.75-1"" (34-45 AU). Putting it all together • HD 15115’s debris disk consists of both small and large grains • Its color changes from blue to gray/red closer to the star • The disk may have a gap near 45 AU, potentially containing a planet (but none were detected) • The east side of the disk is probably being dynamically affected by the ISM, leaving only large grains on this side • The system is dynamically complex, and should be imaged/modeled again in the future

ISM

Rodigas/NCAD/July 20, 2012 13

A brief note on HD 32297... 15 11

PA vs. distance Fig. 3.— Disk FWHM (left) and position angle vs. angular separation for the NE and SW sides of the disk. The disk narrows at smaller angular separations. The two sides of the disk are offset in position angle by ∼ 3–4◦;thediskcurvestowardsthenorthstartingatr = 0.9"".

A. Boccaletti et al.: Imaging of HD 32297 the separation criterion and the size of the optimization zone.

After each algorithm, the frames are derotated to align northwith N the vertical and a median is taken over to provide the final image. As circumstellar disks are extended objects, these techniques E are not optimized since they favor the detection of unresolved Bowed! sources and apply to different independent zones in the image breaking of the continuous shape of a disk. Nevertheless, they can be very efficient in emphasizing regions with strong gra- dients of similar spatial scales as point sources (rings for in- stance). In the past few , several impressive results were obtained using ADI and revealed the very detailed structures within some debris disks (Buenzli et al. 2010; Thalmann et al. 2011; Lagrange et al. 2012a,b). Self-subtraction is an important issue and is caused by the contribution of the disk itself to the Fig. 7.— Comparisons between the modeled and measured disk position angles (right). The triangle identifies the location of the mm building of the reference image to be subtracted from the data brightness peak from Maness et al. (2008). We use the two-component model with the dust ring centered on the star and with a 10 AU cube. Because of the field rotation, the inner regions of a cir- 0.25" cumstellar disks are more affected by self-subtraction in the ADI offset. processing since they cover smaller angular sectors. Moreover, the self-subtraction of the disk increases as the algorithm abil- ity to suppress speckles improves (LOCI being more aggressive Fig. 2. Upper left 1 1” quadrant of the Ks LOCI image, which than cADI for instance). This is particularly important in our emphasizes the deviation× away from to the midplane (red line, data since the rotation amplitude is less than 25◦.Adetailedanal- PA = 47.4◦). The star is at the lower-right corner. The color bar ysis of the self-subtraction arising when imaging disks withADI indicates the contrast with respect to the star per pixel (which is will be presented in Milli et al. (submitted). not applicable to the display-saturated central residuals). SB vs. distance Angular differential imaging algorithms can produce a wealth of various images across which both the stellar residuals and the signal of the object will be modulated. The adjustment of the parameters relies on the balance between self-subtraction and the separation from the star at which the disk is detected. This qualitative analysis leads us to set the separation criterion for both rADI and LOCI to 1.0 FWHM and 1.5 FWHM for, re- See Currie et al. 2012 (+ spectively, the Ks and H bands,× where FWHM=×65 mas in both filters. The radial width is 2 FWHM for both the LOCI subtrac- tion zones and rADI rings.× As for LOCI, we adopt the geome- try described in Lafreni`ere et al. (2007)wherethezonesaredis- tributed centro-symmetrically. Subtraction zones have a radial to azimuth size factor of unity. The optimization zones start atthe Rodigas, Debes, same radius as the subtraction zones but extend to larger radii with a total area of NA = 300 PSF cores. To build the refer- ence image in each zone, all the× available frames were used and the coefficients were calculated with a singular value decompo- sition. AfirstlookattheresultingimagesshowninFig.1 indicates Fig.Kuchner…) 3. Signal-to-noise map obtained from LOCI images in Ks & Boccaletti et Asymmetry / > σ that (1) no information is retrieved at separations closer than 0.5- (top) and H (bottom) for S N 5 per resolution element. The field of view is 4” 1.5” and the disk is derotated (PA = 47.4 ) 0.6” since stellar residuals still dominate the image (for instance, × ◦ the cADI images display a strong wafflepatternasanimprintof to ensure that the disk is aligned along the horizontal. The outer > . < . the AO system), (2) the H band data are of lower quality than ( 1 5”) and inner ( 0 6”) regions are excluded. The color bar / in Ks (as expected from AO performance, see section 2.1,and indicates the S Nperresolutionelements. ffi al. 2012 for more details & flattening! seeing variations, see Table 1)butLOCIe ciently reduces the stellar residuals at similar radii in both bands, (3) the diskhas about the same thickness through all algorithms especially in olution elements (from 0.6” to 1.0”). The main objective of the Ks band which favors a thin and highly inclined disk, and the paper is to find a disk∼ morphological∼ model that can account (4) the H and Ks morphologies are very similar. The structure for this deviation. In addition, we present signal-to-noise-ratio of stellar residuals changes according to the algorithm, anddif- (S/N) maps in Fig. 3.Thenoiselevelwasestimatedazimuthally fers notably from the well-known speckled pattern, especially on LOCI images (both H and Ks), which are smoothed with a owing to the azimuthal averaging caused by the de-rotation of Gaussian (of FWHM = 65ms) so as to derive the S/Nperreso- the frames. A closer inspection shows that, although the disk lution element. We achieved detection levels from about 10σ to is quite linear, it features a small deviation from the midplane 20σ along the disk midplane. We caution that the noise estima- Fig. 8.— Comparisons between the modeled and measured surface brightness profiles for a dust ring centered on the star (left) and one more evident in LOCI images to the NE side (especially in Ks tion can be biased, especially in the speckle-dominated regime with a 5–10 AU pericenter offset (right; the SW side is 5–10 AU closer). Both models reproduce the wavy SB profile. The offset makes the but also in H). Figure 2 shows the position of the midplane (red close to the star as explained in Marois et al. (2008). In the case disk model qualitatively reproduce the differences in the observed SB peaks/plateausRodigas/NCAD/July at r<1!!,althoughitdegradesthemodel’sfidelity 20, 2012 line) with respect to the spine of the disk. The deviation can be of Fig. 3,thisoccursatdistancescloserthan0.5 0.6”, although Fig. 4.— Surface brightness!! profiles for the two sides of the disk. In agreement with previous work (e.g. Schneider et al. 2005), thedisk ∼ at r>1 ,especiallyontheSWside,whereitissubstantiallyunderl"" "" uminous. "" clearly seen in the middle of this image and spans several res- afewspecklesstillappearfurtheroutassignificantpatterns in exhibits power law breaks at r =1.5 and r =1.1 .Weidentifyastrongjumpinsurfacebrightnessstartingatr ≈ 0.5–0.7 .

4

Fig. 1.— Reduced image (top) and signal-to-noise map (bottom) for ourNIRC2HD32297data.Thecolorbardepictingunitsforthe image are in counts, whereas they range from 0 to 9σ for the signal-to-noise map. The central dark region identifies the coronagraphic spot (r =0.3!!). The panels have the same size scale. Looking ahead

Home

• Ultimately want to directly detect light from another Earth • Learn about other indirectly--by studying the circumstellar environments in which they formed (disk gaps, grain sizes, gas giant planets) • The LBT is pushing us closer Rodigas/NCAD/July 20, 2012 Back-up slides

Rodigas/NCAD/July 20, 2012 12 Proof Rodigasof concept et al.

2.5 5 2.5 5

2 Model disk Ks 2 Ks 4 4 1.5 1.5

1 1 3 3

0.5 0.5

2 2 0 0 arcseconds arcseconds −0.5 −0.5 1 1

−1 −1

0 0 −1.5 −1.5

−2 −2 Expected −1 Observed −1

−2.5 −2.5

−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 arcseconds arcseconds

(a) (b)

1.6 1.5 1.6 1.5 Model disk L’ L’ 1.4 1.4 1 1 1.2 1.2 1 0.5 1 0.5 0.8 0.8

0 0 0.6 0.6 arcseconds arcseconds 0.4 0.4 −0.5 −0.5 0.2 0.2 0 −1 −1 0 −0.2 Expected −0.2 Observed −1.5 −1.5 −0.4

−1.5 −1 −0.5 0 0.5 1 1.5 −1.5 −1 −0.5 0 0.5 1 1.5 arcseconds arcseconds Rodigas/NCAD/July 20, 2012 (c) (d) Fig. 11.— Expected and observed artificial disks at Ks band and L!. Model disks are always vertical. Top left : Ks band image of the artificial disk, showing what the disk should look like if unaltered by the LOCI algorithm reduction. Top right : Ks band image of the recovered model disk. Bottom left: the same as (a) except at L!. Bottom right: the same as (b), except at L!. In all cases, the model disks are recovered, though some self-subtraction is evident.

APPENDIX ACCOUNTING FOR DISK SELF-SUBTRACTION BY LOCI To account for disk self-subtraction by the LOCI algorithm, at each wavelength we inserted an artificial model disk into the raw data, at a PA nearly perpendicular to the known disk, ran the data through the pipeline, and recovered the model disks. The artificial disks were set to SB levels comparable to the real disks at each wavelength, and the widths of the artificial and real disks were comparable. Fig. 11a and Fig. 11b show the expected and recovered model disks at Ks band, respectively; Fig. 11c and Fig. 11d show the same at L!. After recovering the model disks, we compared the observed PA, FWHM, and SB values as a function of distance from the star with the expected PA, FWHM, and SB values. The calculations were performed with identical methods to the real disk data analysis. At both wavelengths, the expected and observed PA values are consistent with each other, therefore no correction was needed (see Fig. 12a and Fig. 12b). At Ks band, an addition to the observed FWHM of ∼ 0!!. 03 was needed to correct the apparent constant offset (see Fig.12c).AtL!,acorrectionwasalso needed, with a value of ∼ 0!!. 11 (see Fig. 12d). Both FWHM correction offsets have been included in the FWHM analysis of the real disk images. To correct for the reduction in disk SB at both wavelengths relative to the expected values, we mapped out the self-subtraction as a function of distance and multiplied this into each recovered disk. The corrected SB values, along with the expected and observed, are shown in Fig. 12e (Ks band) and Fig. 12f (L!). The SB corrections have been included in the SB analysis of the real disk images. Large Grains in the HD 15115 Debris Disk 13

360 1.5 Observed Observed Expected Expected 359.9 1

0.5 359.8

0 359.7 −0.5 359.6 −1

Position Angle (degrees) 359.5 Position Angle (degrees) −1.5

359.4 −2

359.3 −2.5 1 1.2 1.4 1.6 1.8 2 2.2 1 1.1 1.2 1.3 1.4 1.5 distance from star (arcseconds) distance from star (arcseconds)

(a) (b)

0.2 0.25 Observed 0.18 Expected Corrected 0.2 0.16

0.14 0.15

0.12

0.1 0.1 FWHM (arcseconds) FWHM (arcseconds)

0.08 Observed 0.05 0.06 Expected CorrectingCorrected for self-subtraction 0.04 0 0.8 1 1.2 1.4 1.6 1.8 2 2.2 1 1.1 1.2 1.3 1.4 1.5 distance from star (arcseconds) distance from star (arcseconds)

(c) (d) 2.15 µm 3.8 µm 3 0.35 Observed Observed Expected Expected Corrected 0.3 Corrected ) ) 2 2.5 2

0.25

2 0.2

1.5 0.15

0.1 1 Surface Brightness (mJy/arcsecond Surface Brightness (mJy/arcsecond 0.05

0.5 0 1.2 1.4 1.6 1.8 2 2.2 1 1.1 1.2 1.3 1.4 1.5 distance from star (arcseconds) distance from star (arcseconds)

(e) (f) Fig. 12.— Top left : Expected (blue) and observed (red) model disk PA vs. distance from the star at Ks band. No correction is needed. Top right : the same, except at L!. No correction is needed. Middel left: Expected (blue), observed (red), and corrected (green) model disk FWHM vs. distanceThe from drop-off the star at Ks inband. SB An is off setnot correction an effect of ∼ 0!!. 03 from is needed. theMiddle data right reduction: the same, except at L!.Anoffset correction of ∼ 0!!. 11 is needed. Bottom left: Expected (blue), observed (red), and corrected (green) model disk SB vs. distance from the star. Bottom right: the same, except at L!.

Rodigas/NCAD/July 20, 2012 Proof of concept

16

Fig. 10.— Images used to model biasing from LOCI processing. (Left) Ourfinalimagewithafakediskadded.(Right)Ourfinalimage where we add the fake disk to each registered image and processthesetofimageswithourpipeline.Inbothcases,werotatethe image to the parallactic angle of the first image in the sequence (PA ∼ -15.06◦), not to true north as we do in Figure 1.

Rodigas/NCAD/July 20, 2012

Fig. 11.— (Top panels) Surface brightness (left) and disk FWHM (right)vs.angularseparationforthefakedisk.(Bottompanels)Ratio of the “observed” (after processing) and expected surface brightness (left) and disk FWHM (right) vs. angular separation. The dotted lines identify power-law fits to correct our SB and disk FWHM measurements for biasing. The need for LBT AO

• Sky is bright at λ > 2 µm • Thermal noise becomes important • Need large aperture + precise AO • At the LBT (and MMT, and soon Magellan), the secondary is the adaptive mirror!

Rodigas/NCAD/July 20, 2012 Rodigas/NCAD/July 20, 2012 Talk about high Strehl...

Rodigas/NCAD/July 20, 2012 Rodigas/NCAD/July 20, 2012 Rodigas/NCAD/July 20, 2012