Mid-infrared Studies of HD 113766 and HD 172555: Assessing Variability in the Terrestrial Zone of Young Exoplanetary Systems
Kate Y. L. Su, George H. Rieke, Carl Melis, Alan P. Jackson, Paul S. Smith, Huan Y. A. Meng, Andras Gaspar
DDraft version June 12, 2020
Typeset using L A TEX twocolumn style in AASTeX63
Mid-infrared Studies of HD 113766 and HD 172555: Assessing Variability in the Terrestrial Zone ofYoung Exoplanetary Systems
Kate Y. L. Su, George H. Rieke, Carl Melis, Alan P. Jackson,
3, 4
Paul S. Smith, Huan Y. A. Meng, andAndr´as G´asp´ar Steward Observatory, University of Arizona, 933 N Cherry Avenue, Tucson, AZ 85721, USA Center for Astrophysics and Space Sciences, University of California, San Diego, CA 92093, USA Centre for Planetary Sciences, University of Toronto at Scarborough, 1265 Military Trail, Toronto, ON M1C 1A4, Canada School of Earth and Space Exploration, Arizona State University, 781 E. Terrace Mall, Tempe, AZ 85287, USA (Received May 4, 2020; Accepted June 11, 2020)
ABSTRACTWe present multi-epoch infrared photometry and spectroscopy obtained with warm
Spitzer , Subaruand SOFIA to assess variability for the young ( ∼
20 Myr) and dusty debris systems around HD 172555and HD 113766A. No variations (within 0.5%) were found for the former at either 3.6 or 4.5 µ m,while significant non-periodic variations (peak-to-peak of ∼ Spitzer
IRS spectra taken in 2004, multi-epoch mid-infraredspectra reveal no change in either the shape of the prominent 10 µ m solid-state features or the overallflux levels (no more than 20%) for both systems, corroborating that the population of sub- µ m-sizedgrains that produce the pronounced solid-state features is stable over a decadal timescale. We suggestthat these sub- µ m-sized grains were initially generated in an optically thick clump of debris of mm-sized vapor condensates resulting from a recent violent impact between large asteroidal or planetarybodies. Because of the shielding from the stellar photons provided by this clump, intense collisionsled to an over-production of fine grains that would otherwise be ejected from the system by radiationpressure. As the clump is sheared by its orbital motion and becomes optically thin, a population ofvery fine grains could remain in stable orbits until Poynting–Robertson drag slowly spirals them intothe star. We further suggest that the 3–5 µ m disk variation around HD 113766A is consistent with aclump/arc of such fine grains on a modestly eccentric orbit in its terrestrial zone. Keywords: circumstellar matter – infrared: planetary systems – stars: individual (HD 113766A, HD172555) – infrared: planetary systems INTRODUCTIONSmall bodies in our solar system like asteroids,Kuiper-belt objects and comets are the leftover andfragments of planetesimals that failed to form planets.These bodies are known to be present around stars fromthe detection of circumstellar dust known as a debrisdisks, which are created as these planetesimals are de-stroyed. Debris disks offer the opportunity to charac-terize planetary systems and their evolution from whenthey emerge from protoplanetary disks into old age (Wy-att 2008; Krivov 2010; Matthews et al. 2014; Hughes
Corresponding author: Kate [email protected] et al. 2018). These gas-poor disks are composed ofdust grains ranging upward in size from (cid:46) µ m, whichare continually replenished by sublimation and collisionsof parent bodies as the byproduct of planet formation.Sensitive infrared surveys with space telescopes (e.g., IRAS , Spitzer , Herschel , and
WISE ) have identifiedthousands of debris disks around mature stars throughmeasurement of the infrared signal when this dust isheated. These observations trace a pattern of develop-ment thought to be similar to that of the solar system(Chen, Su, & Xu 2020). Common features, such as theco-existence of warm ( ∼
150 K) and cold ( ∼
50 K) dust,suggest underlying order in debris disk structures andilluminate various processes about the formation andevolution of exoplanetary systems (Su & Rieke 2014). a r X i v : . [ a s t r o - ph . E P ] J un K. Su et al.
A small percentage of stars just beyond the epoch ofnatal gas-rich disk dispersal show extremely large in-frared excesses (Balog et al. 2009; Kennedy & Wyatt2013; Meng et al. 2017). These systems have hot ( (cid:38) µ m-sized grains. HD 113766 is a bi-nary system with a projected separation of 1 . (cid:48)(cid:48) . (cid:48)(cid:48) ∼ V , indicates it is F6V (Chenet al. 2011). This star belongs to the Lower Centaurus-Crux subgroups of Scorpius-Centaurus OB associationand hence is ∼
17 Myr of age (Pecaut et al. 2012). HD172555 is an A7V star, in the ∼
20 Myr-old β Pic mov-ing group (Mamajek & Bell 2014). It is also a binary,but the low-mass companion is at least 2000 au away(Torres et al. 2006) and has no infrared excess (Rebullet al. 2008; Riviere-Marichalar et al. 2014). Althoughboth systems are at a similar age, HD 113766A is muchdustier than HD 172555, with an infrared fractional lu-minosity ( f d = L IR /L ∗ ) of ∼ × − for the former and ∼ × − for the latter (Mittal et al. 2015).Observations used in this work are reported in Sec-tion 2, including new photometry obtained during thewarm Spitzer mission and mid-infrared spectroscopy ob-tained with Stratospheric Observatory for Infrared As-tronomy (SOFIA) and Subaru. We also review all exist-ing infrared observations (published and unpublished)to assess disk variability. We find significant variationsin the hot dust around HD 113766A. However, we findthat the flux from HD 172555 is dominated by the out-put of the stellar photosphere at 3.6 and 4.5 µ m, whichshows no variations in our data. In Section 3, we discussthe disk structures inferred from spectral energy distri-bution (SED) models and spatially resolved images, toanalyze the implications of the observed variability. Theconclusions are given in Section 4. OBSERVATIONS & RESULTS2.1.
Warm Spitzer/IRAC
Multiple IRAC 3.6 and 4.5 µ m observations were ob-tained during the Spitzer warm mission, resulting in atotal of 11 Astronomical Observation Requests (AORs)for HD 172555 from PID 90250 (PI: Stapelfeldt) and11093 (PI: Su), and 67 AORs for HD 113766 from PID11093. Observations for HD 172555 were performed us-ing sub-array mode to avoid saturation, using a frametime of 0.02 s in a four-point Gaussian dither patternwith medium scale. Only 3.6 µ m data were obtainedunder PID 90250, but both the 3.6 and 4.5 µ m bandswere used under PID 11093. For HD 113766, the full ar-ray mode with a frame time of 0.4 s and 10-point cyclingdither pattern in medium scale were used. Both objectshave two Spitzer visibility windows per year with eachhaving a length of ∼ ± ± ± Time [BMJD - 57120] (day)300350400450500550 T C ( K ) E x ce ss F l ux ( m J y ) µ m3.6 µ m Figure 1.
Time-series
Spitzer observation for the HD 113766 system where the upper panel shows the excess disk emissionat 3.6 and 4.5 µ m and the lower panel shows the derived color temperature. In the upper panel, the colored horizontal linesrepresent the average disk values over the four years of Spitzer data. In the lower panel, the average color temperatures (396 ±
24K and 448 ±
23 K), using the segments of two-year
Spitzer data are shown as the two horizontal dashed lines. There is a 2 σ difference in the long-term observed color temperature (details see Section 2.1). individual visibility windows. Details about the obser-vations (AOR Keys, observed date and time) are givenin Tables A1 and A2.These data were first processed by the Spitzer
Sci-ence Center with the IRAC pipeline S19.2.0. We thenperformed aperture photometry on the basic calibrateddata (BCD) images following the procedure outlined inMeng et al. (2015) for both full and sub-array data.The resultant photometry is also given in Tables A1 andA2. The
Spitzer photometry includes contributions fromboth the star and the dust emission around it.For HD 172555, we used an aperture 3 pixels in ra-dius and a sky annulus between radii of 12 and 20 pixels(pixel scale of 1 . (cid:48)(cid:48)
22) with aperture correction factors of1.112 and 1.113 for 3.6 and 4.5 µ m, respectively. Novariability was found within 0.5% rms in either bandbetween 2013 and 2017. The measurements are alsoconsistent with the expected stellar photosphere (i.e.,no infrared excess within a few percent of the stellarphotospheric model at these wavelengths). Limited bythe optical and near-infrared measurements, the pho-tospheric prediction using stellar models is typically asgood as a few percent. We therefore verified the non-infrared excess by comparing signals in the two IRACbands. Since any residual dust is likely to be at a tem-perature of ∼ µ m band than the 3.6 µ m band. However, allowing for the expected color dif-ference between A0V and A7V, the observed fluxes in those bands agree with expectations for the stellar pho-tosphere alone (Bohlin et al. 2011) to within 1.5%.For simplicity, when we refer to the HD 113766 systemhereafter, we mean the planetary system around the pri-mary. Because the binary is not completely resolved inthe IRAC observations, we used an aperture of 5 pixels(6 . (cid:48)(cid:48)
1) with a sky annulus of 12–20 pixels (14 . (cid:48)(cid:48) . (cid:48)(cid:48)
5) tomeasure the photometry and ensure the flux from bothstars is within the aperture. As shown in Table A1, thereis ∼ (cid:46)
2% repeatability is found atboth bands. We conclude that the data on HD 113766have ∼
1% repeatability, similar to other high signal-to-noise measurements obtained in the same program (e.g.,Su et al. 2019). This suggests that the 5–8% variabil-ity is significant. The optical photometry (in both V and g bands) from the ASAS-SN network (Kochaneket al. 2017) suggests that the unresolved binary has acombined stellar output stable within 1%, indicatingthat the variability comes from the disk around the pri-mary. We determined the combined stellar output tobe 728 and 456 mJy in the 3.6 and 4.5 µ m bands, re-spectively (see Appendix B for details). The time-seriesdisk fluxes (at both bands) after subtracting the stellar K. Su et al. components and the associated color temperatures fromthe flux ratio of the two bands are shown in Figure 1.Assuming a standard color between V and the IRACbands, the two binary components should be of equalbrightness for those bands. As a result, correcting forthe assumed non-variable emission of component B, theintrinsic variability is about twice that observed, i.e., ∼ Spitzer data show that thedust emission in the HD 113766 system exhibits non-periodic variability at both 3.6 and 4.5 µ m with peak-to-peak changes at 10–15% levels (relative to the fluxfrom Component A). There is no strong trend in theoverall flux evolution, except that the data taken in thefirst (last) two years are systematically lower (higher)than the four-year average values ( F IRE, . =64.7 mJyand F IRE, . =213.3 mJy) as shown in the upper panelof Figure 1. This weak trend is also seen in the colortemperature: an average of 396 ±
24 K is derived usingthe data in the first two years, compared with an aver-age of 448 ±
23 K in the last two years. There is a 2 σ positive correlation between the long-term disk flux andcolor temperature. This suggests that the disk flux vari-ation might be due to changes in dust temperature (i.e.,location) if the dust emission is optically thin.2.2. SOFIA/FORCAST
SOFIA/FORCAST (Herter et al. 2012) observationsof HD 172555 were carried out in SOFIA Cycle 4 (Pro-gram 04 0015, PI: Su) on 2016 July 20. Data wereobtained in: (1) the grism mode using G111 (cover-ing 8.4–13.7 µ m with a resolution of R=130–260) andG227 (17.6–27.7 µ m, R=110–120), both using the 4 . (cid:48)(cid:48) λ eff =11.1 µ m, ∆ λ =0.95 µ m) and F242 ( λ eff =24.2 µ m, ∆ λ =2.9 µ m) filters in the 2-point chopping configuration and achop throw of 45 (cid:48)(cid:48) . Observations for HD 113766 werecarried out in SOFIA Cycle 5 (Program 05 0019, PI:Su) on 2017 August 3 using similar instrumental set-tings. Details about the observations are given in TableC3.All SOFIA data were processed by the SOFIA Sci-ence Center with the pipeline ”FORCAST REDUX”ver. 1 2 0 for basic reduction. We used the pipeline-produced science-ready, Level-3 data for further analy-sis. For imaging data, the Level-3 products are nod-subtracted, merged and flux-calibrated; for spectro- Following the same procedure as in Su et al. (2019), the uncer-tainties in the color temperatures of the excesses includes 1.5% ofthe stellar output. Because the IRAC photometry include bothbinary components, the representative errors are over estimated.
Table 1.
FORCAST Imaging and Spectroscopic Photometry a HD F111 F242 G111 G227172555 1.45 ± ± ± ± ± ± ± ± a All fluxes are given in units of Jy. ∼ ± ± µ m ozone band. Because telluric calibrators werenot observed at the same time when the target was ob-
10 15 20 25wavelength ( µ m)0.00.51.01.52.02.53.0 f l ux ( J y ) telluric transmission x 2 FORCAST raw spectrumFORCAST smoothed spectrumHD 172555
10 15 20 25wavelength ( µ m)0.00.51.01.52.02.53.03.5 f l ux ( J y ) telluric transmission x 3 FORCAST raw spectrumFORCAST smoothed spectrumHD 113766
Figure 2.
SOFIA spectra of the HD 172555 (left) and HD 113766 (right) systems. The FORCAST Level-3 raw spectrum isshown as the thin green line, overlaid with smoothed one (red dots, details see Section 2.2). The thin grey line represents theestimated telluric transmission associated with the data. The data points are discarded when the transmission is below 70%(particularly the 9.6 µ m region due to atmospheric ozone band). The purple open circles mark the synthesized flux densityusing the transmission curves (dashed blue lines) of F111 and F242 filters. The photometry obtained in the F111 and F242images was shown as blue open squares. All error bars are 1 σ uncertainty. served (i.e., at similar flight latitude and airmass), ac-curate corrections for such absorption are not possible.We excluded the data points for further analysis wherethe estimated telluric transmission (see the grey curvesin Figure 2) is below 70%. Another flux uncertainty inthe grism spectra comes when the source is not well cen-tered in the slit during the observation. The main pur-pose of the imaging data is to evaluate this uncertainty.As shown in Table C3, the imaging data were takenon the same flight as the spectroscopic data, except forHD 113766 where the F242 image was taken three daysearlier than the G227 spectrum. We estimated the aver-aged flux density and its associated uncertainty by sum-ming over the filter transmission curves (also shown inFigure 2). The resultant synthesized photometry us-ing the spectra is also shown in Table 1. For imagingdata, the photometry accuracy ranges from as good as ∼
7% in the F111 filter, to much poorer ( ∼
20% up to50%) in the F242 filter. The spectroscopic photometricaccuracy is generally in the ∼ ± σ , so no slit loss correction is appliedto the FORCAST spectra. We further determined therepeatability of FORCAST grism spectra by assessingmultiyear, archival data of bright flux calibrators. Ageneral repeatability of 6% and 10% was estimated forthe G111 and G227 grism modes (for details, see Ap-pendix C). In summary, the absolute flux uncertaintyassociated with the FORCAST grism spectra is in the6–10% range.2.3. Spitzer/MIPS Photometry & SED-mode data
During the
Spitzer cryogenic mission, multiple obser-vations were obtained with the MIPS instrument forboth systems. For HD 172555, there are three data setsavailable: two of them were in photometry mode withthe 24 and 70 µ m channels, and one in the MIPS-SEDmode. The first set of photometry was taken on 2004Apr 06 (AOR Key 3723776, PID 10) and the data werepublished in Chen et al. (2006); the second set was takenon 2008 May 18 (AOR Key 25964288, PID 50316). TheMIPS-SED mode data were taken in 2007 Oct 27 (AORKey 21942528, PID 40679). For HD 113766, two setsof MIPS data were obtained. The MIPS photometry(AOR Key 4789760, PID 84) was taken on 2004 Feb 23,and published in Chen et al. (2006). The other set wastaken in the MIPS-SED mode (AOR Key 13625856, PID241) on 2005 Aug 2.For consistency, we re-reduced all the photometry us-ing an in-house MIPS data pipeline. The images of thesources are point-like in both bands; therefore, we ex-tracted the final values using PSF fitting (for details,see Sierchio et al. 2014). For HD 172555, the MIPSphotometry in 2004 is: F = 865 . ± .
15 mJy and F = 226 . ± . µ m, respectively.The MIPS photometry in 2008 is: F = 865 . ± . F = 226 . ± . µ m,respectively. Herschel /PACS photometry was also ob-tained at 70 µ m in 2010 (see Section 2.6 and Table 2),with a result of 217 . ± .
27 mJy. That is, there isno flux variation among 2004, 2008, and 2010 withinthe instrumental calibration levels (1% and 5% at 24and 70 µ m, respectively). For HD 113766 (for both Aand B components), the MIPS photometry from 2004 is: F = 1459 . ± . F = 388 . ± . K. Su et al. µ m)0.00.51.01.52.02.5 f l ux ( J y )
14 16 18 20 22 240.70.80.91.01.11.21.3
Figure 3.
Spitzer
IRS spectrum of the HD 172555 system.The published 2004 IRS spectrum is shown as small dotswith different colors representing different orders. The 2007LL2 spectrum is shown as thick green line (covering 14–21.5 µ m). The 2004 high-resolution model data were scaled tomatch the 2007 LL2 data and joined smoothly with the 2004low-resolution data. A smoothed version of the 2004 spec-trum is shown as the dashed black line. For comparison,the synthesized MIPS 24 photometry is 0.969 Jy (thick bluebar) before scaling, but 0.878 Jy (blue open square) afterscaling. The average MIPS 24 µ m photometry, 0.866 ± et al. (2011). The Herschel /PACS 70 µ m measurement(Table 2) from 2011 is 414 . ± .
62 mJy, again consis-tent within the expected errors.The MIPS-SED mode provides a low-resolution( R =15–25) spectrum from 55 to 95 µ m, which over-laps with the MIPS 70 µ m channel and provides ad-ditional spectral slope information on the disk SEDs.We extracted the raw MIPS-SED mode data from the Spitzer archive and reduced them with the MIPS instru-ment team in-house pipeline (details see Su et al. 2015).Synthesized photometry for the MIPS 70 µ m channelwas also computed using these MIPS-SED mode spec-tra. For HD 172555, the 2007 synthesized photometry is252 ±
25 mJy, consistent with the MIPS 70 µ m photom-etry obtained in both 2004 and 2008. For HD 113766,the 2005 MIPS-SED synthesized photometry is 371 ± µ m photometry takenin 2004. In summary, there is no flux variation (within1% and 5% of the measurements) using the MIPS instru-ments. These data will be discussed together in Section3.1. 2.4. Spitzer/IRS
Both systems have
Spitzer
IRS observations. For theHD 113766 system, the measurements were publishedby Chen et al. (2006) and Lisse et al. (2008) along withdetailed mineralogical models. We simply used the pub- lished spectrum for later analysis. For the HD 172555system, there were two sets of IRS observations. Thefirst one was taken on 2004 Mar 22 using the IRS Map-ping mode (AOR Key 3563264, PID 2) with a combi-nation of low- and high-resolution modules. This dataset has been published by Chen et al. (2006) and Lisseet al. (2009). The second set was taken in 2007 Nov06 using IRS Staring mode (AOR Key 24368384, PID1446) only in the second order low-resolution module(LL2). We extracted this LL2 spectrum from the CAS-SIS project (Lebouteiller et al. 2011) using the optimalextraction. The two spectra (published and unpublishedLL2) are shown in Figure 3. It appears that there is aflux jump between the low- and high-resolution part ofthe 2004 spectrum, while the 2007 LL2 spectrum joinsmore smoothly with the 2004 SL data. Given that thereis no flux variation in the MIPS photometry within 1%at 24 µ m (Section 2.3), the flux discrepancy between theIRS spectra taken in 2004 and 2007 is most likely dueto calibration issues between different modes. We scaledthe 2004 high-resolution mode data to match the 2007LL2 data (SH module multiplying by 1.04 and LH mod-ule multiplying by 0.91), joining them smoothly withthe 2004 SL data. We further smoothed the re-scaled2004 high-resolution spectrum to R=60 to match thelow-resolution mode; the final 2004 re-scaled, smoothedspectrum is also shown in Figure 3.Because the IRS spectrum covers the entire band ofthe MIPS 24 µ m channel, we further compute the syn-thesized MIPS 24 photometry using the published andre-scaled 2004 spectra. The derived MIPS 24 µ m pho-tometry is 0.969 and 0.878 Jy, before and after re-scaling, respectively. The re-scaled synthesized photom-etry is within 1.4% of the MIPS 24 µ m photometry,consistent within a few percent as expected for the ab-solute flux calibration established across all Spitzer in-struments (Rieke et al. 2008; Bohlin et al. 2011). Asimilar comparison is done for the HD 113766 systemusing the IRS spectrum taken in 2004 March. TheIRS synthesized MIPS 24 µ m photometry is 1464 mJy,matched well with the MIPS 24 µ m photometry ( F =1459 . ± . Subaru/COMICS
COMICS is the Cooled Mid-Infrared Camera andSpectrometer mounted on the Subaru 8.2 m telescope atMauna Kea (Kataza et al. 2000; Okamoto et al. 2003).We obtained COMICS observations of HD 113766 on2017 January 15. To complement these data and furthersearch for changes in the solid-state features, we alsoretrieved archival COMICS data of HD 113766, takenon 2006 January 13, from SMOKA (Baba et al. 2002). µ m)0.51.01.52.02.53.03.5 f l ux ( J y ) Subaru 2006
Subaru 2017 Spitzer 2004
Figure 4.
Subaru/COMICS spectra of HD 113766 takenin 2006 (pink color) and 2017 (blue color). The error barsshow 1 σ uncertainties, excluding the 10% flux calibrationerror. Dots represent the smoothed spectra in comparisonto the 2004 Spitzer spectrum (dashed green line). The twoCOMICS spectra are very similar in shape, and in overallflux (orange line is the 2006 spectrum shifted down by 5%).
Observations were taken in low-resolution, N-band spec-troscopic mode with a slit width of 0 . (cid:48)(cid:48) ∼ (cid:48)(cid:48) in 2017, and with a slit width of 0 . (cid:48)(cid:48) ∼ (cid:48)(cid:48) in 2006. All data weretaken in various short integration times on multiple (2–30) chop pairs depending on the desired S/N.Data reduction follows standard high thermal back-ground techniques and is executed through in-house IDLroutines. First, each chop pair was differenced and ex-amined for quality. We discarded the chop pairs show-ing strong residual structures due to either rapidly vary-ing background conditions or the source not reasonablycentered on the slit. The chop pairs were then recti-fied along the spatial axis such that the night sky emis-sion lines run vertically in the array coordinate. A sec-ond pass of background removal was done by subtract-ing the median value of background-only pixels alongthe spatial axis at each dispersion pixel location. Fullyprocessed chop pairs were then median-combined andthe positive and negative spectral beams were extractedvia a straight aperture sum. Uncertainties on spec-tral samples were calculated by determining the rms ofbackground-only pixels and summing that in quadraturefor the number of pixels in the aperture and the Poissonnoise on the total summed flux in the aperture.HD 102964 (K3III) and HR 5029 (K1III) were usedas calibrators in 2006 and 2017, respectively. Assumingthe calibrators are Rayleigh-Jean-like in 8–13 µ m, tel-luric correction was performed by dividing each sciencespectrum by a calibrator spectrum shifted in wavelengthto provide the best cancellation of the strong ∼ µ mozone feature. The wavelength calibration was deter- mined using a low-order polynomial fit to the position ofknown bright sky emission lines. Individual calibratedscience target spectra were then scaled and combinedvia weighted mean. Flux calibration was performed us-ing the photometry measured from the images of thetarget and calibrators. We adopted the WISE
W3 mea-surements for the calibrators, and scaled them to thewavelength of the N-band at which the images weretaken. For the 2006 data, both target and calibratorhad imaging-only observations. For the 2017 data, noseparate imaging-only observation was taken, so we hadto rely on the target spectral acquisition images wherethe star was well-displaced from the slit. The overallflux calibration is accurate at the ∼
10% level.Final spectra showing the 10- µ m range covered byCOMICS are shown in Figure 4. For easy comparison,the calibrated spectra were both smoothed to R=100after rejecting points with a S/N less than 12. Theshape of the 10 µ m feature is very similar between thetwo spectra taken 11 years apart. The overall flux alsoagrees well after shifting the 2006 spectrum down by 5%(within the 10% flux uncertainty). Furthermore, no sig-nificant difference is found between both the COMICSspectra and 2004 Spitzer
IRS spectrum (shown as thedashed green line in Figure 4).2.6.
Other Ancillary Data
In addition to the data described in the previous sub-sections, we also collect all available infrared and mil-limeter photometry for both systems. Although bothsystems are saturated in the two shortest
WISE bands(W1 and W2), the data from the two longer bands(W3 and W4) are not affected. We adopt the mea-surements given in the ALLWISE catalog (Cutri & etal. 2014). We also include
AKARI /IRC and
IRAS mea-surements . For IRAS, we adopt the measurements fromthe Faint Source Catalog (FSC). Herschel /PACS obser-vations have been reported in the literature for bothsystems (Riviere-Marichalar et al. 2014; Olofsson et al.2013), but those measurements were based on an earlyreduction pipeline and calibration. Since both sourcesare not resolved by
Herschel , we adopt the values fromthe
Herschel
Point Source catalog (Marton et al. 2017)where all available data were combined and processedwith the final data calibration and pipeline. Table 2 listsall the valid observations along with the years they wereobtained. For the measurement uncertainty, we usedboth the quoted statistical error and the rms of the blanksky region added in quadrature. Finally, both systems These sources are not detected by
AKARI /FIS nor IRAS 100 µ m. K. Su et al.
Table 2.
SED Measurements λ eff F tot F IRE
Note F tot F IRE
Note( µ m) (mJy) (mJy) (mJy) (mJy)HD 172555 HD 1137668.23 1450.97 ± ± AKARI , 2006 1308.00 ± ± AKARI , 200611.56 1124.87 ± ± WISE , 2010 1260.96 ± ± WISE , 201012.00 1520.00 ± ± IRAS , 1983 1580.00 ± ± IRAS , 198317.61 921.00 ± ± AKARI , 2006 1428.00 ± ± AKARI , 200622.09 955.79 ± ± WISE , 2010 1665.57 ± ± WISE , 201023.67 865.90 ± † ± Spitzer , 2004–2008 1459.00 ± ± † Spitzer , 200425.00 1090.00 ± ± IRAS , 1983 1800.00 ± ± IRAS , 198360.00 306.00 ± ± IRAS , 1983 622.00 ± ± IRAS , 198370.00 217.30 ± ± Herschel , 2010 414.80 ± ± Herschel , 201171.42 226.78 ± † ± Spitzer , 2004–2008 388.20 ± ± † Spitzer , 2004–2008100.00 103.00 ± ± Herschel , 2011 222.50 ± ± Herschel , 2012160.00 59.21 ± ± Herschel , 2010–2011 96.13 ± ± Herschel , 2011–20121300.00 0.11 ± ± ± ± Note —Photometry is given as the total (star+disk) flux (F tot ) and the excess flux (F
IRE ) after photospheric subtraction.The note column gives the source of photometry and the rough date that it was taken. † The MIPS photometry uncertaintyis limited by the calibration (1% and 5% at 24 and 70 µ m, respectively), the reported uncertainty includes the calibrationerror added in quadrature.
10 100 1000wavelength ( µ m)10 −1 e x ce ss f l ux ( m J y ) IRASSpitzer/MIPSAKARI/IRCALLWISEHerschel/PACSALMASpitzer/IRS & MIPS−SEDTwo Modified BlackbodiesSpitzer/IRAC
10 100 1000wavelength ( µ m)10 −1 e x ce ss f l ux ( m J y ) IRASSpitzer/MIPSAKARI/IRCALLWISEHerschel/PACSALMASpitzer/IRS & MIPS−SEDModified Blackbody
Figure 5.
Spectral energy distribution of the debris emission in the HD 113766 (left) and HD 172555 (right) systems. Observeddata are shown as various color symbols given on the plot. The associated error bars are the 1 σ uncertainty, except for theIRAC measurements for HD 113766A where the error bars reflect the range of variations. For the HD 113766 system, the diskcontinuum (dark grey, long-dashed line) can be described as the the sum of two modified blackbody functions (light grey, dashedlines) for the inner and outer parts, while only a single such function is needed for the HD 172555 system. have ALMA Band 6 (1.3 mm) observations (Lieman-Sifry et al. 2016 for HD 113766, and Matr`a et al. in prep.for HD 172555). We further discuss the millimeter ob-servations in Section 3.1. The collected measurementsare listed in Table 2 along with the excess emission atthose wavelengths after subtracting the stellar compo-nent. The disk SEDs (i.e., excess emission) are shownin Figure 5.2.7. Variability in Mid-infrared Spectroscopy
Because
Spitzer
IRS spectra cover several mid-infraredbands in various infrared space missions, one can, inprinciple, use synthesized photometry to search for vari-ability (similar to the comparison between MIPS 24 µ mphotometry and IRS synthesized photometry presentedin Section 2.4). However, given the different strategiesin absolute flux calibration, it is not straight forward tocompare photometry directly in high fidelity.Here we complete the search for variations by fo-cusing on the mid-infrared spectroscopy. For the HD113766 system, as has been discussed by Olofsson et al.(2013), the shape of the 10 µ m feature remains simi-lar when comparing the mid-infrared spectra obtainedby VLT/VISIR in 2009 and 2012 to the 2004 Spitzer spectrum; however, the overall level is hard to quantifydue to the difficulty in correcting for atmospheric ab-sorption. Figure 6a shows the comparison of the spec-tra taken with TIMMI2 at ESO in 2002 (Sch¨utz et al.2005),
Spitzer in 2004, Subaru in 2006 and SOFIA in2017. Overall, the shape of the 10 µ m feature is verysimilar with a subtle (up to 15%, ∼ σ ) difference inthe both blue- and red-side of the feature for all fourspectra. In the 20 µ m region, the overall flux level inthe 2017 SOFIA spectrum is consistently lower than the2004 Spitzer spectrum by ∼
10% (see the flux ratio inthe bottom panel), but still within 1 σ . Some of thelarge difference in flux occurs at the wavelengths wherethe atmospheric transmission is the worst. We do notethat the blue and red sides of the 10 µ m feature coverthe wavelength region where the features of sub- µ m-sized silica grains are present (Koike et al. 2013), andthe nominal atmospheric transmission in this region isabove 90%. The subtle difference in the 10 µ m shapemight be due to a change in the amount of silica grains,but this cannot be confirmed because of the calibrationuncertainty.We observe a similar behavior for the mid-infraredspectra in the HD 172555 system (Figure 6b). No changein the 20 µ m region within 10% is detected between Spitzer µ m region shows a subtle change near the wave-length region at which the silica features are present; however, the tentative change (less than 3 σ ) cannot beconfirmed. In summary, no change (more than 20%) isdetected in the mid-infrared spectra over ∼
15 year spanfor both systems. Variations of (cid:46)
10% might be present(particularly in the region of silica features); however,they cannot be confirmed due to calibration uncertain-ties across telescopes/instruments. DISCUSSION3.1.
Disk Structures Inferred from SED Models andSpatially Resolved Images
There are various SED models for both systems inthe literature. Depending whether the fits include theprominent solid-state features and/or excess emissionlong-ward of 100 µ m, the derived dust temperaturesand the number of components in the models vary fromstudy to study. For example, excluding the infraredbands that cover the dust features ( ∼ µ m), we de-termine that the disk emission in HD 172555 requiresonly one modified blackbody ( T d ∼
250 K) to fit the diskSED, while the one in HD 113766 requires two ( T d ∼ (Figure 5). When primar-ily fitting the solid-state features presented in the IRSspectra, 335 K and 490 K are derived for HD 172555 andHD 113766, respectively (Lisse et al. 2009, 2008). In thelatter case, an additional ∼
75 K component is requiredto fit the 70 µ m excess, suggesting two separate compo-nents in the HD 113766 disk (Lisse et al. 2008). How totranslate these dust temperatures to physical distancefrom the star depends significantly on the underlyinggrain properties, particularly the minimum grain sizesin the debris (Thebault & Kral 2019).High angular resolution imaging studies have alsobeen performed for both systems from the ground, pro-viding direct constraints on the location of the dustemission. Smith et al. (2012) presented a comprehen-sive mid-infrared study of these two systems includinginterferometric, spectroscopic and narrow-band imagingobservations obtained with the VLT. For the HD 113766system, Olofsson et al. (2013) confirmed that the diskhas two distinct components by simultaneously fittingthe imaging data from the VLT and the well-sampled in-frared disk SED, including the solid-state features. With The modified black body is a function of wavelength λ givenby B ν ( λ, T d ) for λ < λ and α ( λ o /λ ) β B ν ( λ, T d ) for λ > λ ,where B ν is the Planck function, and the temperature T d andscaling factor α , λ and β are free parameters. This function iswidely adopted for fitting debris disk SEDs to account for thefact that the emission comes from grains in all sizes, and thatgrains will not act as perfect black bodies, in particular emittinginefficiently at wavelengths longer than their own size; λ and β are thus related to the grain size distribution. K. Su et al. f l ux ( J y ) Subaru COMICS spectrum 2006ESO TIMMI2 spectrum 2002Spitzer IRS spectrum 2004SOFIA FORCAST spectrum 201710 15 20 25 λ ( µ m)0.81.01.2 f l ux r a ti o telluric transmission f l ux ( J y ) ESO TIMMI2 spectrum 2002Spitzer IRS spectrum 2004SOFIA FORCAST spectrum 201610 15 20 25 λ ( µ m)0.81.01.2 f l ux r a ti o telluric transmission Figure 6.
Comparison of mid-infrared spectra obtained over ∼
15 year span: left for the HD113766 system and right for theHD 172555 system. The upper panel of the plots shows the spectra in flux level, while the lower panel shows the flux ratiorelative to the
Spitzer spectrum taken in 2004. The grey bars show the 1 σ uncertainty for the SOFIA spectrum (6% for G111and 10% for G227). In the 10 µ m region covering the prominent solid-state features from sub- µ m grains, no significant (morethan 20%) change is found. Variation at (cid:46)
10% level might be present, but cannot be confirmed due to uncertainties acrossdifferent telescopes/instruments. the pronounced dust features and the mid-infrared inter-ferometric measurements, the inner disk in HD 113766is relatively constrained to radii of 0.5–1.2 au (includ-ing 1 σ margin) from the star. The outer disk is esti-mated to lie at radii of ∼ . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48)
81, suggesting the millimeter emissionfrom the system is extended beyond 44 au from the star.Because the bulk of the disk emission peaks at mid-infrared wavelengths (see left panel of Figure 5), it isdifficult to reconcile the large difference in the disk sizesmeasured at mid-infrared and millimeter if both comefrom the same outer disk component. A SED modelusing the parameters derived by Olofsson et al. (2013)and a power-law extrapolation using the measured far-infrared fluxes both predict a total 1.24 mm flux thatis smaller than the ALMA value by a factor of 2. Thissuggests that either the ALMA detection comes from adifferent cold component that predominantly emits inthe millimeter wavelengths or that the ALMA detectionis contaminated by background galaxies.For the HD 172555 system, mid-infrared interferomet-ric observations completely resolve out the disk emis-sion, suggesting the bulk of debris emission is outside of0.5 au from the star (Smith et al. 2012). The disk wasfirst resolved in the thermal infrared (at ∼ µ m, Smithet al. 2012) and later in polarized scattered light (Engleret al. 2018). Mid-infrared imaging suggests that the diskis inclined by ∼ ◦ from face-on. Assuming a Gaussian-ring-like model, the bulk of disk emission at 20 µ m ispeaked at 7.7 ± ∼ ∼ ∼ Spitzer
IRAC and IRS 10–20 µ m wavelengths, andan outer, ∼ ∼ ∼ Implications
As has been discussed in the literature, the amountof small (sub- µ m to µ m size) grains that produce theprominent solid-state features must be stable over adecadal timescale, as reinforced by the new Subaruand SOFIA observations, so that the shape and rela-tive strength of the 10 µ m feature (within ∼
10% level)remains the same. The relative stability implies thatthe dust production and loss rates are balanced. The1production of small grains in typical debris systemsmainly comes from the collisional cascades within theswarm of particles, while the loss mechanism is dom-inated by collisional destruction followed by radiationpressure blowout for stars as luminous as these two. Al-though the exact size of grains that are subject to radia-tion blowout depends sensitively on the properties of thestars (mass and luminosity) and dust grains (composi-tion, sizes and porosity), the blowout limit is around ∼ µ m in both systems if one assumes compact, silicate-likecomposition (which generally gives the smallest blowoutsizes). Arnold et al. (2019) concluded that the blowoutsize could be an order of magnitude larger when mix-ing different kinds of compositions and porosity. How-ever, the compositions used in that study are materialscommonly found in solar system interplanetary parti-cles; minor minerals such as crystalline silicates (pro-ducers of sharp solid-state features in debris disks) arenot included.We first review whether the dust in these systemscan be produced in typical collisional cascades. Weadopted the analytical formula and criterion in Wyattet al. (2007) for the assessment. Due to the necessaryassumptions and associated uncertainties, the adoptedmodel requires a high threshold (a factor of 1000) todifferentiate systems that are most likely in a transientstate from the ones that are consistent with steady-statecollisional cascades. For the HD 172555 system, the fol-lowing parameters were used: 1.8 M (cid:12) , 9 L (cid:12) and 20 Myrfor the star with a planetesimal belt at r ∼ f max )is 2 × − for a typical collisional cascade system. Theobserved infrared fractional luminosity is only 3–4 timeslarger than this value, i.e., within the uncertainties thedust in the HD 172555 can be produced by a massiveplanetesimal belt. For the HD 113766 system, 1.5 M (cid:12) , 4 L (cid:12) and 20 Myr were used for the star, and planetesimalbelts at r ∼ r ∼
10 au were assumed. The corre-sponding f max is 3 × − and 6 × − for the inner andouter planetesimal belts. The observed fractional lumi-nosity is ∼ × − for the inner belt, and ∼ × − forthe outer one (Olofsson et al. 2013). That is, the dustin the outer belt can arise in a conventional collisionalcascade, but that in the inner belt is transient (i.e., theobserved amount is ∼ µ m-sized grains ( ∼ g) and their short life-times ( ∼ . A commonquestion in the literature has been ”How did such tinygrains get created in the first place?” The concern isthat it may be difficult to create such abundant sub- µ m-sized grains in a typical, optically-thin, collisional-cascade debris system because of the short lifetime toradiation pressure blowout for their parent ∼ µ m-sizedgrains. However, Thebault & Kral (2019) have calcu-lated new models of collisional cascades and suggestthat sub- µ m grains below the blowout size can accumu-late in sufficient number potentially to resolve this co-nundrum for bright disks around early-type stars (suchas HD 172555). Their model has a number of uncer-tainties, such as operating only in 1-D, having to makeextrapolations of the strength law toward the sub- µ mregime, etc. It would also make it puzzling that we donot observe such prominent solid-state features towardthe majority of the bright disks around early-type starswith similar fractional excesses; HIP 73145 (an A2V starwith f d ∼ × − ) and HIP 77315 (an A0V star with f d ∼ × − ) from Ballering et al. (2017) are twosuch examples. Furthermore, Thebault & Kral (2019)conclude that massive debris systems around solar-likestars are not sufficiently effective at preserving grainssmaller than the blowout size to result in pronouncedsolid-state features in the mid-infrared. Further work isneeded to test their suggestion.Although such a steady-state solution might be attrac-tive in non-evolving systems, it is less convincing for thesystems with disk variability, such as HD 113766. Therecent CO gas detection (L. Matr`a, priv. comm.) andthe large amounts of atomic species (Riviere-Marichalaret al. 2014; Kiefer et al. 2014; Grady et al. 2018) inthe HD 172555 system also argue for a transient phe-nomenon. In these active systems, an optically-thickcloud of debris produced by a violent impact betweentwo bodies with sizes of large asteroids or proto-planetsmight be a better explanation for the presence of abun-dant sub- µ m grains. The presence of such optically-thick debris clouds is inferred to explain the complexdisk light curves observed in some extremely dusty,young debris disks (Meng et al. 2014; Su et al. 2019).Given the mass and confined volume resulting in a largeimpact, the impact-produced cloud is very likely to beoptically thick initially before thinning by Keplerianshear. Such an environment provides perfect conditions See Appendix C for the discussion about BD+20 307 K. Su et al. to generate over-abundant small grains because stellarphotons can only efficiently remove the small grains atthe surface of the cloud. Once the impact-producedcloud becomes optically thin, the small grains that aresubject to radiation pressure blowout would quickly beremoved from the system, likely resulting in a rapid dropin the system’s flux due to the decrease of dust tempera-tures, as has been observed in the ID8 system (Su et al.2019). However, there may be sufficient time to gen-erate even smaller grains within the cloud that are notsubject to radiation pressure blowout, and are revealedonly after the cloud becomes optically thin. This mightbe the case for the HD 172555 system where Johnsonet al. (2012) suggested that tiny, 0.02 µ m-sized obsid-ian grains are not subject to radiation blowout. Sim-ilarly, 0.1 µ m-sized silicate grains in the HD 113766system are expected to be stable due to the same rea-son. In this scenario, the prominent infrared featureswould remain stable for a long time. The timescale onwhich we could expect them to change would then be setby the (much longer) Poynting–Robertson (P–R) dragtimescale of these fine grains, which is ∼ × yrfor the HD 113766 inner component and ∼ × yr for the HD 172555 system .For the HD 113766 system, there appears to be a con-flict between the variability from warm Spitzer obser-vations and the stable solid-state features in the mid-infrared. We note that the level of variability observedat 3.6 and 4.5 µ m is consistent with that of the possiblechanges in the mid-infrared spectra from Spitzer , Subaruand SOFIA if the dust emission is optically thin. If somepart of the 3.6/4.5 µ m emission is optically thick, onewould not expect the IRAC flux to track the optically-thin solid-state features in the mid-infrared. It is alsolikely that the inner disk component has a backgroundpopulation of (cid:38) km-sized planetesimals, creating addi-tional low-level variations traced by the high-precisionIRAC photometry. The tentative (2 σ ) positive correla-tion between the long-term disk flux and color temper-ature (see Section 2.1) might be due to a change in dustlocation under the optically thin assumption. A 13%change in the color temperature is then translated to a27% change in location ( T d ∼ / √ r ). A dust clump/arcin a modestly eccentric ( e ∼ The P–R timescale is formulated as τ PR = cr / (4 GM ∗ β ) ∼
800 yr ( r/ au) ( M (cid:12) /M ∗ )(0 . /β ) where c is the speed of light, G is the gravitational constant, and M ∗ is the stellar mass. Weassume β =0.5 to calculate the P–R timescale as a minimal value. tent with the hypothesis of a stable population of sub- µ m grains responsible for the pronounced solid-state fea-tures. Future observations in the range of 3–5 µ m canfurther test the hypothesis of an eccentric clump/arc inthe HD 113766 inner component.3.3. Can Dust Composition alone Tell us How SmallGrains are Generated in Young Dusty DebrisSystems?
The infrared wavelength range contains the fundamen-tal bending, stretching and skeleton modes of solid-statespecies. Consequently, infrared spectroscopy of molecu-lar clouds and protoplanetary disks (van Dishoeck 2004;Sargent et al. 2009) provides crucial information on thesizes and composition of dust grains, which are thebuilding blocks embedded in (cid:38) km-sized planetesimalswithin a planetary system. We can learn about thesebuilding blocks through infrared spectroscopy, whenthey are released by planetesimals shattered to formsecond-generation dust debris. In addition to coagu-lation, dust grains in the protoplanetary disks have alsobeen through various heating/cooling processes so thatcrystalline forms of silicates (such as olivine, pyroxeneand silica) are commonly found in the more evolvedstages of planet-forming disks (Sargent et al. 2009),while such crystalline grains are largely absent in theinterstellar medium (Kemper et al. 2004). Therefore,it should not be a surprise to find crystalline silicatesin debris disks if these processed grains are stored inplanetesimals. Nonetheless,
Spitzer /IRS studies of de-bris systems reveal that the majority of these disks havea featureless dust continuum in the mid-infrared (Chenet al. 2005; Mittal et al. 2015). This is understandablegiven that the typical blowout size is ∼ µ m in these sys-tems, and that grains larger than a few µ m have weakand broad features.In contrast, conspicuous crystalline solid-state fea-tures in the mid-infrared are characteristic of extremedebris systems (Song et al. 2005; Rhee et al. 2008; Lisseet al. 2008; Weinberger et al. 2011; Fujiwara et al. 2012;Olofsson et al. 2012, 2013; Lisse et al. 2020), producedby warm sub- µ m-sized silicates similar to the ones com-monly found in Herbig Ae/Be (Bouwman et al. 2001)and T-Tauri disks (Kessler-Silacci et al. 2006; Sargentet al. 2009). Despite different modeling approachesfrom study to study, the strength of the features typ-ically requires ∼ − g of sub- µ m-sized grains in op-tically thin environments. These features are similarto laboratory-measured features from meteoric, terres-trial crustal, and mantle material measured in powderform, implying that the debris dust in these young sys-tems underwent various degrees of shock and/or high3temperature events (Morlok et al. 2014; de Vries et al.2018). Although such connections in dust mineralogybetween debris disks and highly altered/processed ma-terial have been found in the solar system, similar linksseem to be present in the protoplanetary stage of evo-lution when gas was still present. Therefore, identifyingthe dust species alone does not tell us when these alteredgrains were formed either in the recent or far past. How-ever, given the issues of retaining such a large amount ofthese sub- µ m-sized grains (as discussed in the previoussection), it becomes clear that they have to be generatedin the recent past.Another indication of recent violent events in debrisdisks is the detection of freshly condensed sub- µ m-sizedsilica smokes (Rhee et al. 2008; Lisse et al. 2009; Fu-jiwara et al. 2012). Indeed, this is thought to be theprocess that formed the first sub- µ m-sized solids in thesolar nebula, which is also verified by the laboratory ex-periments (see the book by Lauretta & McSween 2006).However, this kind of condensation took place in a low-pressure environment, ∼ − –10 − bar in the early so-lar nebula (Ebel 2006). Such an environment is verydifferent from the high pressure (and temperature) con-dition within a vaporizing collision between two largebodies (Johnson & Melosh 2012). The sizes of the va-por condensates generated in a violent collision betweentwo large planetesimals depend sensitively on the colli-sional conditions, but are nearly all larger than 10 µ m forimpactors larger than 10 km (Johnson & Melosh 2012).Roughly mm-sized impact glasses (spherules and shards)are commonly found in the sample returns from theApollo mission (see the review by Zellner 2019). Mostimportantly, these impact-produced glasses will not giverise to the prominent solid-state features displayed inyoung dusty debris disks, unless they are broken up intomuch smaller grains.Recently, extreme space weathering has been proposedto break up larger grains and to explain the ”unique”dust mineralogy in the young ( ∼
11 Myr, Pecaut et al.2012), dusty ( f d ∼ × − , Mittal et al. 2015) HD145263 system (Lisse et al. 2020). A key ingredient tothe space weathering model is super massive flares orintense X-ray/UV radiation from the young star, a con-dition that is applicable to the AU Mic system (Mac-Gregor et al. 2020) and many other young late K andM-type stars. However, there is no detectable solid-statefeature in the AU Mic system (see the IRS spectra pro-vided by CASSIS project, http://cassis.sirtf.com/) andpronounced solid-state features are rarely seen in late-type disks (Lawler et al. 2009; Mittal et al. 2015). Al-though the energetic stellar photons might be efficientlyaltering the surface composition of large (a few 100 µ m) grains as discussed by Lisse et al. (2020), they mostlikely vaporize small µ m-sized grains as suggested byOsten et al. (2013). Therefore, this mechanism cannotexplain the amount of sub- µ m-sized grains required toproduce the solid-state features observed in this system.Using the same analytical estimate discussed inthe previous section for HD 145263, we found f max ∼ × − , 2.6 × − , and 6.7 × − for a plan-etesimal belt at r ∼
1, 2 and 3 au assuming the samestellar parameters as in HD 113766A and an age of 11Myr. The dust level observed in the HD 145263 systemis on the borderline between transient dust and colli-sional cascade dust produced in a massive planetesimalbelt. Given the amount of sub- µ m-sized grains ( ∼ g, Lisse et al. 2020) and the short lifetime of their par-ent µ m-sized grains in optically thin collisional cascades,we suggest that the fine dust in the HD 145263 system,similar to the two systems we discussed in this paper,was created by transient events. A large collision, in-volving large-sized ( (cid:38)
100 km) asteroids that facilitatedthe over production of sub- µ m-sized grains within anoptically thick impact-produced debris clump, is a bet-ter scenario to explain the distinctive solid-state featuresobserved in these young systems. CONCLUSIONWe present multi-epoch, infrared photometric andspectroscopic data for two young ( ∼
20 Myr), extremelydusty debris systems around HD 113766A and HD172555. High precision (S/N > µ m data were obtained dur-ing Spitzer warm mission to assess the disk variability.We found no variability for HD 172555 within 0.5% ofthe average fluxes in either bands with the data takenbetween 2013 and 2017. These measurements are con-sistent with the expected photospheric values to within1.5% (i.e, no infrared excess). For the HD 113766 sys-tem, on the contrary, non-periodic variability was de-tected at ∼ σ positive correlation between the long-term diskflux and color temperature, suggesting that the varia-tion might be due to changes in dust temperature (i.e.,location) if the dust emission is optically thin.Low-resolution, mid-infrared spectra obtained withSOFIA/FORCAST and Subaru/COMICS are presentedin this study. FORCAST grism spectra covering 8.4– We refer it as borderline because only the smallest (1-au) plan-etesimal belt location meets the criterion, by more than a factorof 1000, as in a transient state while others are only larger by afew 100. K. Su et al. µ m (G111 mode) and 17.6–27.7 µ m (G227 mode)were obtained in 2016 and 2017 for HD 172555 and HD113766, respectively. Using the similar grism data frombright flux standards available in the SOFIA archive,we further determined an overall 1 σ of 6% (for G111mode) and of 10% (for G227) uncertainties associatedwith these SOFIA spectra, including the flux calibrationand instrumental repeatability. Additionally, COMICS8–13 µ m spectra taken in 2006 and 2017 were presentedfor the HD 113766 system. We find the shape of the10 µ m feature and the overall flux level (within 10%)are very similar between the two spectra taken 11 yearsapart.We also present multi-epoch Spitzer
MIPS data ob-tained during the cryogenic mission including photome-try at both 24 and 70 µ m channels and MIPS-SED modedata. No flux variation (within 1% at MIPS 24 µ m and5% at MIPS 70 µ m) was found for HD 172555 using datafrom 2004 to 2008 and HD 113766 using data from 2004to 2005. Combined with other multi-wavelength ancil-lary data, these data were used to discuss the disk struc-ture in each of the systems. The HD 113766 system hasa two-component disk with an inner ∼ ∼ µ m region, and an outer ∼ ∼ ∼
15 year span for bothsystems. Using the high-quality
Spitzer
IRS spectrumas a reference, we found no significant (more than 20%)change in either the shape of the prominent 10- µ m solid-state feature or the overall mid-infrared flux levels forboth systems. Variations of (cid:46)
10% might be presenton the blue and red sides of the 10 µ m feature wherethe dust features from sub- µ m-sized silica are generallyfound. However, such subtle changes cannot be con-firmed due to calibration uncertainties, and need to beconfirmed with future observations obtained with a morestable instrument such as JWST/MIRI.The amount of sub- µ m-sized grains required to pro-duce the prominent 10 µ m feature is on the order of10 g for both systems (Lisse et al. 2009; Olofsson et al.2013). Adopting the analytical calculation and criterionset by Wyatt et al. (2007), we verified that the amountof sub- µ m-sized grains in the inner component of theHD 113766 system is unlikely to be produced in a typi-cal collisional cascade system, while its outer componentand the HD 172555 disk are consistent with steady-statecollisional cascades. For massive debris disks around early-type stars (such as HD 172555), new collisionalcascade calculations by Thebault & Kral (2019) sug-gest that a sufficient amount of sub- µ m-sized grains canaccumulate and produce stable and pronounced 10 µ mfeatures. Nevertheless, some other ingredients might bemissing in their model because the majority of early-type massive disks show featureless dust continua in themid-infrared.Extreme space weathering has been proposed to ex-plain the unusual dust mineralogy in the HD 145263system (Lisse et al. 2020), where the host star is verysimilar to HD 113766A (both early F-type and young).We do not think space weathering is a viable explanationfor the presence of abundant sub- µ m-sized grains. En-ergetic photons from super massive flares around youngsolar-like stars are most likely to destroy/vaporize smalldust grains present in the disk. Although intense spaceweathering might be able to alter the surface composi-tion of large grains and preserve the usual compositionas they broke up in subsequent collisions, this mecha-nism cannot explain the fact that the majority of the de-bris systems around young, active stars show featurelessmid-infrared spectra. We also note that sub- µ m-sizedcrystalline silicates and silica are commonly found inplanet forming disks when gas was still present (Bouw-man et al. 2001; Kessler-Silacci et al. 2006; Sargent et al.2009). Therefore, dust species alone do not inform uswhen these highly processed grains were formed. Theycould be formed in the early protoplanetary disk stage inwhich those altered grains were stored in planetesimalsand subsequent collisional grinding among planetesimalsreleased them into the circumstellar environment. Theycould also be formed in the recent past when amorphousfine grains experienced high temperature events such asviolent impacts involving large asteroidal or planetarybodies.Finally, we suggest that disk variability might be auseful signpost to reveal highly altered grains generatedin the recent past for young and dusty exoplanetary sys-tems. In this case, such as the inner component of theHD 113766 system, the abundant sub- µ m-sized grainsmight be due to the intense collisions among their par-ent mm-sized vapor condensates produced in an initiallyoptically thick, impact-produced debris clump. Such anoptically thick environment would easily lead to an overproduction of fine grains that are smaller than the typi-cal blowout size in the center of the clump because stel-lar photons cannot easily penetrate. Once the impact-produced clump becomes optically thin as stretched byKeplerian shear, the grains subject to radiation pressureblowout (just slightly smaller than the blowout size) arequickly removed from the system, likely resulting in a5rapid flux drop in the system’s infrared output (a phe-nomenon that has been seen in the ID8 system, Su et al.2019). If the impact-produced cloud has enough time inthe optically thick condition to allow for the produc-tion of even smaller grains that are not subject to radia-tion blowout ( ∼ µ m glassy silica in HD 172555 and ∼ µ m crystalline silicates in HD 113766A), these finegrains would be stable after the clump/arc becomes opti-cally thin as suggested by Johnson et al. (2012). If thereis no additional input for the sub- µ m-size grains such asadditional large impacts, the lifetime of these featuresshould be unchanged within the P–R timescale, i.e., be-yond the human lifetime for these two systems. Nev-ertheless, for the systems with such features located onless than a 0.2 au around 1-2 M (cid:12) stars, changes withina couple of decades are possible. For the HD 113766system, the long-term flux trend and the positive corre-lation between disk flux and color temperature observedby Spitzer are consistent with the presence of such anoptically thin clump/arc on a modestly eccentric orbitin the inner component. The lifetime of such an impact-produced clump/arc depends sensitively on the impactcondition and subsequent collisional evolution and theasymmetric phase can last for a few hundred orbital evo-lutions (Jackson et al. 2014; Kral et al. 2015).
Facilities:
Spitzer (IRAC, IRS, MIPS), SOFIA(FORCAST), Subaru (COMICS) ACKNOWLEDGMENTSThis work is based on observations made with the
Spitzer
Space Telescope, which is operated by the JetPropulsion Laboratory, California Institute of Technol-ogy, and made with the NASA/DLR Stratospheric Ob-servatory for Infrared Astronomy (
SOFIA ). SOFIA isjointly operated by the Universities Space Research As-sociation, Inc. (USRA), under NASA contract NAS2-97001, and the Deutsches SOFIA Institut (DSI) underDLR contract 50 OK 0901 to the University of Stuttgart.This work has made use of data from the Euro-pean Space Agency (ESA) mission
Gaia
Gaia
Gaia
Multilateral Agreement. Financial support for thiswork was provided by NASA through award A. WARM IRAC PHOTOMETRY FOR THE HD113766 AND HD 172555 PHOTOMETRYThe measured fluxes using the
Spitzer
IRAC observa-tions described in Section 2.1 are given in Tables A1 andA2 for the HD 113766 and HD 172555 systems, respec-tively. For HD 113766, the quoted fluxes are for boththe A and B components. As discussed in Section 2.1,only the A component has an infrared excess indicativeof circumstellar dust. We predicted the combined stellaroutput from both components in Appendix B and usedit to derive the excess fluxes of HD 113766A at both3.6 and 4.5 µ m (also given in Table A1). The uncer-tainty associated with the excess fluxes include 1.5% ofthe combined stellar output. B. STELLAR PROPERTIES FOR THE HD 113766SYSTEMWe used the photometry that contains both stars(Johnson UVB and 2MASS JHK) to estimate the com-bined stellar output in the IRAC wavelengths. A Ku-rucz model with a temperature of 6750 K provides agood fit to the combined photometry, and results a to-tal luminosity of 9.4 L (cid:12) at a distance of 111 pc (GaiaCollaboration et al. 2016, 2018). The derived combinedluminosity is within ∼
10% of the combined value derivedby Pecaut et al. (2012) after correcting for the adopteddistance. The expected photospheric flux is 728 mJy atIRAC 3.6 µ m band, and 456 at IRAC 4.5 µ m band.6 K. Su et al.
Table A1.
The IRAC fluxes of the HD 113766 systemAOR Key BMJD . F . E . exeF . exeE . BMJD . F . eF . exeF . exeE . (day) (mJy) (mJy) (mJy) (mJy) (day) (mJy) (mJy) (mJy) (mJy)53463040 57128.539790 788.99 3.67 61.09 11.52 57128.538250 676.37 1.80 220.82 7.0753462784 57133.418070 775.56 4.26 47.66 11.72 57133.416540 676.96 1.96 221.41 7.1153462528 57138.283870 788.66 3.29 60.76 11.40 57138.282350 673.29 2.00 217.74 7.1253462016 57142.636470 772.22 4.17 44.32 11.69 57142.634950 673.03 1.41 217.49 6.9853461504 57147.970320 779.14 4.33 51.24 11.74 57147.968810 670.18 1.89 214.63 7.0953460992 57153.371310 777.25 4.44 49.35 11.79 57153.369770 671.43 1.60 215.88 7.02 Note — F and E are the flux and uncertainty including the star, while exeF and exeE are the excess quantities excludingthe star. This table is published in its entirety in the machine-readable format. A portion is shown here for guidanceregarding its form and content. Table A2.
The IRAC fluxes of the HD 172555 systemAOR Key BMJD . F . E . BMJD . F . eF . (day) (mJy) (mJy) (day) (mJy) (mJy)48318464 56455.24044 5316.18 0.85 · · · · · · · · · Note —The average flux densities are 5340 ±
20 mJy and 3496 ± µ m bands, respectively.C. DETAILS ABOUT THE SOFIA/FORCASTOBSERVATIONS AND ASSOCIATEDUNCERTAINTYAs described in Section 2.2, we used the Level-3 dataproducts provided by the SOFIA Science Center for fur-ther analysis. Table C3 gives the details about theSOFIA/FORCAST observations presented in this work,including the time of the observation, flight altitude, in-strumental grism/filter, and total integration time. Toassess the flux calibration and repeatability in the FOR-CAST data, we also made used of multiyear, archivaldata for bright calibrators: α Tau, α Boo, β And, and σ Lib. We used the imaging data to determine the appro-priate aperture correction factor. A total of 206 ”cali- brated” Level-3 images was used for the F111 filter withdata taken from 2013 to 2019, and a total of 22 imagestaken in 2016-2019 was used for the F242 filter. Theaperture correction factor for a specific setting in theaperture was determined by referencing the total fluxderived using the nominal aperture setting [12,15,25](i.e., a radius of 12 pixels and sky annulus of 15–25 pix-els). For an aperture setting of [8,8,12], the aperturecorrection factor is 1.073 ± ± ± ± α Tau has more than a dozenG111 Level-3 (”combspec”) data points taken in 2016–2018, σ Lib has a handful of G227 data taken in 2014–2015, and the other two stars have a few spectra avail-able in the archive. Note that these calibration datawere taken at different flights, altitudes and slit widths(2 . (cid:48)(cid:48) . (cid:48)(cid:48) σ ) for the G111 and G227 grism spectrais not significant. Recently, a flux increased ( ∼ ± Spitzer and 2015 SOFIAspectrum (G111 mode). Taking the uncertainties in ab-solute flux calibration, stability of the instruments, andslit loss and aperture corrections, the reported uncer-tainty is slightly under estimated, suggesting the tenta-tive flux increase is at ∼ σ levels.REFERENCES Arnold, J. A., Weinberger, A. J., Videen, G., & Zubko,E. S. 2019, AJ, 157, 157Baba, H., Yasuda, N., Ichikawa, S.-I., et al. 2002, Report ofthe National Astronomical Observatory of Japan, 6, 23Ballering, N. P., Rieke, G. H., Su, K. Y. L., & G´asp´ar, A.2017, ApJ, 845, 120Balog, Z., Kiss, L. L., Vink´o, J., et al. 2009, ApJ, 698, 1989Bohlin, R. C., Gordon, K. D., Rieke, G. H., et al. 2011, AJ,141, 173Bouwman, J., Meeus, G., de Koter, A., et al. 2001, A&A,375, 950Chen, C. H., Mamajek, E. E., Bitner, M. A., et al. 2011,ApJ, 738, 122Chen, C. H., Su, K. Y. L., & Xu, S. 2020, NatureAstronomy, 4, 328Chen, C. H., Patten, B. M., Werner, M. W., et al. 2005,ApJ, 634, 1372Chen, C. H., Sargent, B. A., Bohac, C., et al. 2006, ApJs,166, 351Cutri, R. M., & et al. 2014, VizieR Online Data Catalog,II/328de Vries, B. L., Skogby, H., Waters, L. B. F. M., & Min, M.2018, Icarus, 307, 400Ebel, D. S. 2006, Condensation of Rocky Material inAstrophysical Environments, ed. D. S. Lauretta & H. Y.McSween, 253Engler, N., Schmid, H. M., Quanz, S. P., Avenhaus, H., &Bazzon, A. 2018, A&A, 618, A151Fabricius, C., & Makarov, V. V. 2000, A&A, 356, 141Fujiwara, H., Onaka, T., Yamashita, T., et al. 2012, ApJL,749, L29Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al.2016, A&A, 595, A1 Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.2018, A&A, 616, A1Grady, C. A., Brown, A., Welsh, B., et al. 2018, AJ, 155,242Herter, T. L., Adams, J. D., De Buizer, J. M., et al. 2012,ApJL, 749, L18Holden, F. 1976, PASP, 88, 52Hughes, A. M., Duchˆene, G., & Matthews, B. C. 2018,ARA&A, 56, 541Jackson, A. P., & Wyatt, M. C. 2012, MNRAS, 425, 657Jackson, A. P., Wyatt, M. C., Bonsor, A., & Veras, D.2014, MNRAS, 440, 3757Johnson, B. C., & Melosh, H. J. 2012, Icarus, 217, 416Johnson, B. C., Lisse, C. M., Chen, C. H., et al. 2012, ApJ,761, 45Kataza, H., Okamoto, Y., Takubo, S., et al. 2000, Societyof Photo-Optical Instrumentation Engineers (SPIE)Conference Series, Vol. 4008, COMICS: the cooledmid-infrared camera and spectrometer for the Subarutelescope, ed. M. Iye & A. F. Moorwood, 1144–1152Kemper, F., Vriend, W. J., & Tielens, A. G. G. M. 2004,ApJ, 609, 826Kennedy, G. M., & Wyatt, M. C. 2013, MNRAS, 433, 2334Kessler-Silacci, J., Augereau, J.-C., Dullemond, C. P., et al.2006, ApJ, 639, 275Kiefer, F., Lecavelier des Etangs, A., Augereau, J. C., et al.2014, A&A, 561, L10Kochanek, C. S., Shappee, B. J., Stanek, K. Z., et al. 2017,PASP, 129, 104502Koike, C., Noguchi, R., Chihara, H., et al. 2013, ApJ, 778,60Kral, Q., Th´ebault, P., Augereau, J.-C., Boccaletti, A., &Charnoz, S. 2015, A&A, 573, A39 K. Su et al.
Table C3.
SOFIA ObservationsLevel-3 a F111 2016 Jul 20 14:50:58.4 43017–43018 55.30199 F111 2016 Jul 20 14:53:04.3 43017–43018 55.30200 F111 2016 Jul 20 14:54:11.2 43018–43019 55.30201-0338 G227 2016 Jul 20 14:55:54.3 43017–43031 4496.70287-0329 G111 2016 Jul 20 16:23:39.7 43017–43031 2010.5HD 1137660084-0089 F111 2017 Aug 03 09:52:55.1 38015–38022 167.20090-0091 F242 2017 Aug 03 09:58:33.9 38010–38020 44.90092 F111 2017 Aug 03 10:00:31.0 38014–38019 18.60193 F111 2017 Aug 03 10:00:57.4 38015–38016 18.60096-0127 G111 2017 Aug 03 10:02:54.7 38013–39002 1560.50010 F111 2017 Aug 06 09:16:30.4 37085–37079 18.90011 F111 2017 Aug 06 09:16:56.4 37068–37079 18.90014-0077 G227 2017 Aug 06 09:19:02.3 37110–38095 3098.8 a No source was found; not used for further analysis.Krivov, A. V. 2010, Research in Astronomy andAstrophysics, 10, 383Lauretta, D. S., & McSween, H. Y. 2006, Meteorites andthe Early Solar System IILawler, S. M., Beichman, C. A., Bryden, G., et al. 2009,ApJ, 705, 89Lebouteiller, V., Barry, D. J., Spoon, H. W. W., et al.2011, ApJS, 196, 8Lieman-Sifry, J., Hughes, A. M., Carpenter, J. M., et al.2016, ApJ, 828, 25Liou, J. C., & Malhotra, R. 1997, Science, 275, 375Lisse, C. M., Chen, C. H., Wyatt, M. C., & Morlok, A.2008, ApJ, 673, 1106Lisse, C. M., Chen, C. H., Wyatt, M. C., et al. 2009, ApJ,701, 2019Lisse, C. M., Meng, H. Y. A., Sitko, M. L., et al. 2020,arXiv e-prints, arXiv:2003.06870MacGregor, A. M., Osten, R. A., & Hughes, A. M. 2020,ApJ, 891, 80Mamajek, E. E., & Bell, C. P. M. 2014, MNRAS, 445, 2169 Marton, G., Calzoletti, L., Perez Garcia, A. M., et al. 2017,arXiv e-prints, arXiv:1705.05693Matthews, B. C., Krivov, A. V., Wyatt, M. C., Bryden, G.,& Eiroa, C. 2014, in Protostars and Planets VI, ed.H. Beuther, R. S. Klessen, C. P. Dullemond, &T. Henning, 521Melis, C. 2016, in IAU Symposium, Vol. 314, Young Stars& Planets Near the Sun, ed. J. H. Kastner,B. Stelzer, & S. A. Metchev, 241–246Melis, C., Zuckerman, B., Rhee, J. H., & Song, I. 2010,ApJL, 717, L57Melis, C., Zuckerman, B., Rhee, J. H., et al. 2012, Nature,487, 74—. 2013, ApJ, 778, 12Meng, H. Y. A., Rieke, G. H., Su, K. Y. L., & G´asp´ar, A.2017, ApJ, 836, 34Meng, H. Y. A., Su, K. Y. L., Rieke, G. H., et al. 2014,Science, 345, 1032—. 2015, ApJ, 805, 779