Reply to the Referee’s Report for ApJ/300877/ART We would like to thank the referee for his/her thorough review of our manuscript. The paper is much improved after his/her suggested revisions. 2/25/09 2:14 PM Referee report - ApJ/300877/ART Page 1 of 3 http://authors.iop.org/atom/electref.nsf/ViewRefereeReport/8700F3787B15CB3D802575680055858C?OpenDocume nt Referee report Article ID: ApJ/300877/ART Title: Abundant Circumstellar Silica Dust and SiO Gas Created by a Hypervelocity Collision in the ~12 Myr HD172555 System Edited Feb 25 "Abundant Circumstellar silica dust and SiO gas created by a hyper-velocity collision in the ~12 Myr HD172555 system" by C.M. Lisse et al. This paper reports on the analysis of the Spitzer/IRS spectrum of the young (1.2 10^7 yr) main sequence star HD172555. This star has a quite peculiar thermal emission spectrum of circumstellar dust in that the shape of the feature at about 10 micron peaks sharply at 9.3 micron and shows additional structure at shorter wavelengths. The 10 micron spectrum is found to be very similar to that of the T Tauri star FN Tau, and somewhat similar to that of HD23514, a young (10^8 yr) main sequence star. The authors model the thermal spectrum and find that the peak is best represented by glassy silica (tektite and obsidian) and that the structure at shorter wavelengths can be fitted by gas-phase SiO2. Amorphous silicates are not seen. Fitting a power-law for the grain-size distribution and setting the maximum size of the grains to a millimeter they obtain a mass in dust of 10^21 kg. For a larger upper limit of 100m this increases to 10^22 kg. The minimum size of the grains adopted in their model is not mentioned. The temperature of the grains is warm (~200-335 K). The origin of the silica dust and molecules is explained by a collision between two large bodies, hundreds of kilometers in size, impacting at a high relative velocity (dv >= 10 km/sec) and that is shattered and partly vaporized in the event. Subsequently, silica may have condensed from the melt and from gas phase SiO. A high impact velocity is thought to be needed in order to process silicates into silica. In comparing this event to solar-system events, they conclude that it not similar to the collision event in which the Earth's moon was formed (to low dv), but more like an impact that stripped Mercury of its surface layers (the chemical composition of the surface layers of Mercury seems comparable to what is derived for the dust around HD172555). An alternative hypothesis, that is not excluded by the authors, is that the dust is produced by a stirring-up of a planetisimal belt, for instance due to planetary migration, causing (multiple?) collisions of large bodies at high velocities. The event must have happened quite recently (< 0.1 Myr) as otherwise the small silica particles would have been swept out of the system by radiation pressure or accreted onto the star because of the Poynting-Robertson and stellar wind drag. ---major comments 1. Regarding the quality of the data. It seems to me that up to 6 micron the spectrum is very noisy. Molecular SiO2 produces a sharp drop from 5-6,5 micron, and the spectrum appears to show this. However, in view of the noise I doubt whether one can really conclude this? Please explain. The error bars on the data, as displayed, are 2-sigma. Looking at the data, the rather noisy part appears to our eyes to be in the 5-6 um region. Above 6 um the SNR is better than 5 per spectral element, and with the numerous spectral elements, and a careful chi-squared fitting, the modeling is robust. The fundamental of SiO2 emission occurs at 8 um, and has a sharp dropoff to the blue side (in the 7.5-8 um region), and then essentially no emission strength in the 6-7.5 um region. It is this spectral behavior that is driving the fits. The referee is correct in stating that the 5-6 um behavior of the data, while consistent with the spectral model, is also noisy and not conclusive. 2. For the assumptions on the modeling of the thermal spectrum the authors refer to other papers of the same first author. I have checked some of these paper, but could not find a detailed discussion of essential assumptions. Information needs to be added regarding: Our 2008 HD113766a paper, section 3, goes into some detail about the basic tenets of the modeling, the adoption of 7 different classes of materials (olivines, pyroxenes, water ice/gas, amorphous carbon, carbonates, PAHs, metal sulfides, and phyllosilicates, representing the majority dominant dust species for the most abundant atomic species H, C,N,O, Si, Mg, Fe, S). We had been asked by the reviewer of that paper to include more detail of the modeling, and we thought that that would be enough to use in reference in this work. Since many of the referee’s questions relate to the modeling, we have decided instead to add to the HD172555 manuscript the detailed discussion of the spectral modeling. Following the results of the Deep Impact and STARDUST experiments, this section explicitly lists the species in the model, the particle size range applicable to the fitting (0.1 – 1000 um), how each component is fitted independently to the spectrum, so that the dust is modeled not as aggregate particles, but as a mixture of optically thin fine particles (it is always important to remember that we would not be seeing any spectral features unless there is a large preponderance of micron sized particles in the observed dust – larger particles are optically thick and show little spectral behavior above blackbody), and how we have done the minimal amount of modeling consistent with recent sample return and remote sensing results. - The best fit presented in Table 2 consists of quite many components (11!). Are all these species really needed? What if one is excluded; does this result in a significantly poorer fit? (So, please perform an f-test like analysis to show that all these components are really needed). The fits are redone without each component and the chi-squared re-checked to see if the goodness of fit has changed significantly (column “Model chi-squared if not included” in Table 2). We do not list species for which the model chi-squared does not change significantly as having been detected, although they may not be ruled out, either.) - It is not clear to me that for all of their components the authors use measurements (of powders) to obtain extinction characteristics, or whether they derive these from measurements of optical constants for some of the species. Please explain. In case of the former, what l do they assume outside of the wavelength measured? In case of the atter, what do the authors assume for the shape of the grains? We use laboratory thermal emission spectra. As noted by the referee, Mie theory and its assumption of spherical grain structure is not very applicable to the case of randomly oriented astrophysical grains produced by a number of different kinetic and thermodynamic processes, and then evolved by collisions, shocks, etc. We do not assume any spectral behavior outside the wavelength measured. 2/25/09 2:14 PM Referee report - ApJ/300877/ART Page 2 of 3 http://authors.iop.org/atom/electref.nsf/ViewRefereeReport/8700F3787B15CB3D802575680055858C?OpenDocume nt Homogeneous spheres? Or some other (shape) distribution. This may be relevant as homogeneous spheres produce a peak-location that is always bluer than less symmetric particles. To illustrate a possible implication, Chen et al. 2006 use solid cristobalite spheres. If Chen et al. would use less symmetric particles the peak of their cristobalite grains would shift to the red, perhaps producing a good fit without the need for silica. Agreed, and this is why we do not use Mie theory! - The authors do not assume the grains to be part of aggregates (and representing the monomers in the aggregate), in which case it would be very likely that the small and large monomers are in thermal contact. If the origin of the bulk of the grains released in the presumed collision is formed by shattering it seems likely that the dust is in aggregates, and should be represented by the same temperature (distribution). In this case, can the spectrum be fitted? Our model assumes a run of temperature with particle size. The largest particles are optically thick, and behave like blackbodies, and thus are set to LTE. We allow the temperature of the smallest (0.1um) particles of each component to vary independently, and interpolate between the 2 extremes for intermediate particle sizes. In doing this, we found that the best fit model had the temperature of all components roughly equivalent – except for the very absorbing small amorphous carbon particles, which always “run” hot, and the gigantic blackbody particles, which are always at LTE. If not, it seems (again within the collision hypothesis) that the origin of the bulk of the grains is through condensation of the melt or from the gas phase as in this case, I can imagine that individual components are more likely to be formed. So, add a short discussion of chemically homogeneous particles vs. chemically in-homogeneous aggregates. - A remark about nomenclature (though I'm not an expert in this!). The fact that your 'silica' is mixed with significant amounts of Al2O3 (essentially causing the pure silica peak at about 8.5 micron to shift to ~9.7 micron) to me seems to imply this is actually a glassy silicate (i.e. not a glassy silica). We disagree – only one of the two ‘silica’ components has any appreciable alumina component (the tektite), and this at the 14% level. We have amorphous silicates of both olivine and pyroxene composition in the model, and they look spectrally nothing like the amorphous silicate species. The main peak for crystalline quartz is at ~9 um, and the main emission peak we are seeing in our data is at 9.3 um. Alumina has emission features in the 11 – 20 um region, not the 8 – 10 um region, that are much more muted than the silica/silicate features, and which would not shift the emission peak to ~9.7 um. An appropriate terminology for the material may be the catch-all “amorphous silicate of silica composition”, except that silicas and silicates are normally clearly distinguished in the geological literature. Since the stoichiometry of the tektite and obsidian material is close to the SiO2, and not the SiO3 or SiO4 of the silicates, and the spectral signature of the material is clearly highly unusual for any silicate, but nor for the silicas, we find the amorphous silica terminology to be the best. 3. The discussion takes up almost half of the paper. The authors must make this more concise, in any case by removing repetitive statements. In my view section 4.4 and 4.5 can be joined and shortened considerably. ---Minor Comments The English can often be improved. I will give some examples, but these are only meant as an illustration. A thorough 'clean up' of the paper is needed. Introduction - If HD172555 and beta Pic have the same spectral type and class, and are both at a small distance, then indeed their B-V must be the same. We prefer to turn the argument around, and the fact that the B-V is the same is a confirmation of the assignment of HD172555 to the Beta Pic moving group. - ..HD172555 should also have a very rich circumstellar disk. Rich in what sense? We have changed the text from “By implication, HD172555 should also have a rich circumstellar disk. “ to read “By implication, HD172555 should also have an extensive circumstellar disk.” - Provide a reference for '...the estimated age for the initiation of planet formation" We have added in the references Wetherill 1990; Yin et al. 2002; Chambers 2004 to bolster this point. Section 2 - In finding the best Kurucz model its irrelevant to assume an age (12 Myr). The sentence containing the Kurucz discussion has been recast as “The stellar photospheric flux was then estimated by minimum χ2 fitting of a Kurucz stellar atmosphere models to optical and 2MASS (0.3 – 3 um) archival photometry1. See URL http:// http://nsted.ipac.caltech.edu/” - Don't use phrases such as "we were amazed" Agreed. We have changed this sentence to read “On the other hand, a literature search of possible gas species produced a possible match with the fundamental ro-vibrational linear stretch complex of the SiO molecule – it has P, Q, and R branches at 7.5 - 10 um, centered at ~8 um.” - Alpha Tau, an O-star on the asymptotic giant branch. Alpha Tau is a K5III. We meant to say that it is an oxygen star, not an O star as in OBAFGKM. This has been clarified by re-writing the sentence as follows :” Examination of the transition moment calculations of Drira et al. (1997) and the IRAS/LRS and ISO-SWS measurements of Alpha Tau, a oxygenrich K-star on the asymptotic giant branch with a known SiO absorption feature (Cohen and Davies 1995) demonstrated a possible fit to the residuals.” Section 3 - "The minerals disproportionate into SiO and metal" I think you mean decompose or break-up While disproportionate is a correct chemical term, denoting non-stoichiometric evaporation, we like decompose better for the target audience, and have replaced ‘disproportionate’ with ‘decompose’. Thanks! - Again: "SiO gas if disproportionating olivines..." We have re-written this sentence to read “Despite detailed searches, there are no obvious features of FeO or MgO mineral species (expected in addition to SiO gas if olivine or pyroxene pyrolitic decomposition has occurred).” - Mention minimum adopted grain size in 3.2. How sensitive is the derived mass for a_min? The minimum grain size, particles of 0.1 µm in radius, is now explicitly mentioned: “The best-fit model size distribution for the HD172555 circumstellar dust producing the sharp silica feature is dn/da = a-3.95±0.10, 0.1 < a < 1000 µm, more small particle dominated than a purely collisional equilibrium distribution of dn/da = a-3.50 (Dohnanyi 1969). “ The derived total mass is not sensitive to the minimum grain size at all, the vast majority of the mass (but not the surface area) resides in the largest particles. - "... the mass in these particles is not trivial." do you mean insignificant? We have changed this sentence to read “We note that given that the relative surface area in this population is significant, the relative amount of mass is likely to be important as well.” - Explain P-R The sentence where P-R drag is first mentioned now reads “(Here recent is defined as within the condensation lifetime of SiO gas at 5.8 AU from a L* = 9.5 A5 star, and within the Poynting-Robertson (P-R) drag lifetime and radiation pressure blowout time for dust a few microns in radius (Burns et al. 1979, Chen et al. 2006), or within a fraction of 1 Myr.)” - Remove "In terms of the thermal temperature ... is ~206 K" as it doesn't really add anything (Don't you mean radiative equilibrium is LTE?). We have substantially re-written this section to read “This technique has worked well in comparison to imaging measures of the dust location. E.g., we placed in the location of the dominant emitting dust from the inner cavity wall of the HD100546 disk at 13 AU (Lisse et al. 2007a), and STIS 1-D spectroscopic measurements find a cavity for this disk of radius 13.2 AU across (Grady et al. 2001, 2005). Our best-fit model for the HD69830 circumstellar dust ring puts the dust at 1 AU from the K0V primary (Lisse et al. 2007b), while recent Gemini AO measurements have shown that any warm dust in the system resides at < 2 AU from the primary (Beichman et al. private communication 2007). For the HD172555 circumstellar material, scaling from the T T1ejecta = 340K for the hottest Deep Impact dust at 1.51 AU and L* = 1.0, we find for the smallest grains with Tmax = 335 K that rdust = 5.8 ± 0.6 AU. LTE for grains larger than 100 µm in radius at this distance is ~206 K. This agrees well with Wyatt et al. 2007b's Spitzer 24/70 µm based estimate of 4-5 AU for the location of the dust with respect to the HD172555 primary, and with our finding in the best fit of an associated reservoir of ~200 K blackbody-like dust with no spectral signature - there are large pieces of dust associated with the very fine grained material producing the sharp spectral features. A location of 5.8 AU for the circumstellar material is also consistent with the location of the innermost dust belt surrounding b Pic (6.4 AU from the star), as determined by Okamato et al. (2004). The equivalent location in our solar system is at ~1.9 AU from the Sun, or between the orbit of Mars and the inner edge of the asteroid belt.” Section 4 2/25/09 2:14 PM Referee report - ApJ/300877/ART Page 3 of 3 http://authors.iop.org/atom/electref.nsf/ViewRefereeReport/8700F3787B15CB3D802575680055858C?OpenDocume nt - "an asteroid of radius >= 440 km radius asteroid" remove 'radius asteroid' Done. - 'oligarch' sized planetesimal. This is to my knowledge not a common phrase. If I'm right, please remove it from the text (also in the abstract) as it took me until page 18 before I understood its meaning. The term “oligarch” is becoming more and more common in the terrestrial planet formation community. It is denotes the few planetisimals that win the accretion game enough to rise to Ceres to Mars size, and come to dominate the behavior of all planetisimals in the latter part of the planet formation process. We prefer to keep this terminology. "undense circumstellar torus" undense is not an English word. we estimate the time scale for SiO recondensation to occur via 2-body collisions in an circumstellar torus in the molecular flow regime to be ≤ 0.1 Myr.” We have removed the term “undense” and changed the text to read “…