From apjse@sheffield.ac.uk Tue Mar 25 10:26:54 2008 Date: Tue, 25 Mar 2008 11:26:45 -0500 From: apjse@sheffield.ac.uk To: skinners@origins.colorado.edu Cc: ApJ-MS74302@mss.uchicago.edu, apjse@sheffield.ac.uk Subject: Your ApJ Submission MS# 74302 Dr. Stephen L. Skinner University Campus Box 389 Center Astrophysics and Space Astronomy University of Colorado Boulder, CO 80309 USA Dear Dr. Skinner: Enclosed please find the referee's report on your submission to the ApJ entitled "High-Resolution Chandra X-Ray Imaging and Spectroscopy of the \sigma Orionis Cluster" by Stephen L. Skinner, Kimberly R. Sokal, David H. Cohen, Marc Gagne, Stanley P. Owocki, and Richard D. Townsend ( MS# 74302). When you resubmit the manuscript, please include a detailed cover letter containing the (mandatory) listing of the changes you've made to the text and your responses to the report. Processing of your revised manuscript will be expedited if you make your revisions to the manuscript latex file available for downloading from the ApJ Web Peer Review System (http://mss.uchicago.edu/ApJ/). This version includes your previous submission plus any modifications to latex commands necessary for smooth processing by the ApJ electronic system. The associated PDF file is the version seen by the referee. If you have any questions, feel free to contact me. Best regards, Richard de Grijs, Scientific Editor The Astrophysical Journal Phone: +44 114 222 4524 Fax: +44 114 222 3555 apjse@sheffield.ac.uk ************************************************* Report on MS74302v1 -- High-Resolution Chandra X-ray Imaging and Spectroscopy of the sigma Orionis Cluster This is study of X-ray emission from the sigma Orionis star cluster with Chandra. The most significant impact from this paper comes from the study of the grating spectrum of sigma Ori AB. Although no strict conclusion is reached, this study adds a valuable example of a spectral line study of an OB star. However, some of the conclusions of this paper are limited by a lack of clarity and completeness. A general comment is that some of the paper tends to favor the standard radiative wind shock model, although other models have not been ruled out. For example, sections 5.2 and 5.3 present valid models that, although more speculative, are quite plausible. This study would be more helpful if the data were worked a bit further into physical constraints or suggestions for follow-up (particularly in Sec. 5.2 and 5.3). It seems to me that more should come from such a good set of data. Specific points --------------- Abstract: The second sentence says the observations are capable of "strictly testing" X-ray emission models. However, as discussed in the paper, there are several models that may explain the emission, at least partly because the nature of the star system (double? triple?) is not fully known. As currently written, this study does not make a "strict" test and this sentence should be softened. * We have rewritten the Abstract and no longer use the phrase 'strictly testing'.* Introduction: Page 3, first full paragraph -- References to "Miller et al." and "Cohen et al" are incomplete. * Fixed * Page 3, second full paragraph -- Sentence starting "However, the density...". Using "however" here implies a contradiction of previous point, but it is not clear how that sentence is different from the previous ones. * According to my Webster's 7th New Collegiate Dictionary, 'however' can have several meanings, and does not necessarily imply that what follows constitutes a contradiction of a previous point. As used in our manuscript, the word 'however' has a meaning akin to the following two definitions given ' by Webster's 7th: (i) 'in spite of that', or (ii) 'on the other hand'. We believe the use of 'however' was acceptable, but we have removed it anyway to avoid implying a contradiction. * Page 4, second paragraph -- It is noted that some stars in the sigma Ori cluster are still undergoing accretion. It is known how this would affect the results of this study (however small)? * The effects of accretion on X-ray emission (if any) are controversial and not very well understood. But, there is increasing evidence from studies of several star-forming regions that the X-ray luminosities of accreting classical TTS are slightly suppressed relative to non-accreting weak-lined TTS. We now mention this in the last sentence of this paragraph. * the X-ray Section 3, X-ray Source Identification and Properties: Section 3.1, first paragraph: Is it possible to include an estimate of background X-ray source density? It would be useful to know up front what fraction of sources may be unrelated to the cluster. * We've added a short paragraph (2nd par. of Sec. 3.2) that estimates the number of background sources.* Section 3.3, first paragraph: The typical mean photon energy is given, but this is not very meaningful without considering the detector response. Can you give some mean energy for the ACIS sensitivity (e.g., the mean photon energy expected for a flat spectrum source)? * Mean (or median) photon energy is commonly used in X-ray studies to provide a rough comparison of the relative hardness between sources (see for example, Table 1 of the Chandra COUP study results in Getman et al. 2005, ApJS, 160 353). If the events have been energy-filtered to within a specific energy range (as is the case here), outlier photons with very low or very high energies are removed and the mean and median photon energies are similar. Sometimes a 'hardness ratio' is also quoted, but in our opinion the mean energy is a more intuitive quantity. In order to fully account for the detector response, one needs to extract a spectrum and response matrix file (RMF) for the source. This is only feasible for brighter sources, and we give these results in Table 3. It is not clear what is meant by a 'flat spectrum source'. If this means that the number of photons/sec/cm^(-2)/keV falling on the detector is constant as a function of photon energy (E), then such a spectrum can be modeled as a photon power-law with power-law index alpha = 0, that is A(E) = KE^{-alpha} with alpha = 0 and K a constant. (XSPEC model 'powerlaw'). An XSPEC simulation of such a source convolved with the ACIS-S3 RMF shows that the count rate generally increases with E over the range .5 - 5 keV (except for a small interval between 2 - 2.7 keV), then falls off more slowly between 5-7 keV. Thus, the general effect of the detector response will be to suppress low-energy counts, esp. between 0.5 - 1 keV. If the simulated convolved flat power-law spectrum is plotted as 'counts/keV' versus E (keV) then the mean value of counts/keV lies at around E = 1.6 keV. Since the use of mean E to compare relative hardness between different sources in X-ray observations is common practice, we don't believe it is necessary to get into a distracting discussion of RMF effects in this paper (esp. for hypothetical power-law sources with alpha = 0, which do not resemble real stellar sources).* Section 3.4: This section describes variability from the observation, but could be more complete. For example, it would be useful to have the actual variability probability for sources, as opposed to the "v" given in Table 2. Also, it would be useful to have some discussion of variability in a larger context, perhaps addressing the x-ray variability in comparison with previous observations? * We've added an introductory paragraph to Sec. 3.4 placing variability in context with earlier Orion studies. We've added the KS variability probability to variable sources in Table 2.* Section 3.5: The reddening is described as "normal", "moderately", or "heavy". I normally think of this as the degree of reddening, but this section seems to discuss these as differences in the reddening law. Perhaps I am missing something here. At the least, it should be clarified a bit. It would also be useful to discuss the significance of this, since it appears in the abstract and conclusion sections. * We agree that the above nomenclature is potentially confusing. We have now reworded the 2nd par. of Sec. 3.5 as well as item 1. in the Conclusions (Sec. 6) and the Figure 10 caption to avoid this nomenclature. * Section 3.7.2, Page 11, second full paragraph: The discussion of sigma Ori E suggests that the ratio Lx/L* is unusual. It would be nice to have errors here; they might add up significantly, considering how many pieces go into calculating this ratio. * The dominant uncertainty in L_x for sig Ori E is due to the uncertain distance to the star. The previously derived distance of 640 pc versus the cluster distance of 390 pc introduces an uncertainty of a factor of 2.7 in L_x. For this reason, we give an expression for L_x in the 5th par. of 3.7.2 leaving the distance as a free parameter, then we compute L_x/L_bol for the two different distances of 640 and 390 pc for comparison. Uncertainties in L_x/L_star also arise from the uncertainty in the X-ray flux measurement, the uncertain stellar radius, and some disagreement in the literature as to the value of T_eff. We now give these uncertainties in Sec. 3.7.2 and derive a lower limit for L_x/L_star for the case d = 390 pc. If all of the uncertainties conspire to push the ratio downward, then a lower limit of log L_x/L_star = -7.09 is possible. We now give this value in the text. * Section 3.7.4, Second paragraph: There is discussion here about correlation with a radio source, but it is not clear why it is believed to be associated with IRS1. Also, the upper limit to the radio flux from sigma Ori AB is discussed here for some reason. This would be better in the section on that star. * The association between the radio source and IRS1 is based on the close positional agreement between the two. We have now rewritten the first few paragraphs of Sec. 3.7.4 to clarify this, and we now explicitly state the VLA position of the radio source. We also refer the reader to Fig. 1 of van Loon and Oliveira 2003, which shows the relative positions of IRS1, sigma Ori AB, and the VLA radio source. The upper limit on the radio flux density of sig Ori AB provides additional support beyond that based on positions that the radio source is not sigma Ori AB, but instead associated with IRS1. We thus believe this discussion of the radio upper limit should remain in the section on IRS1 (Sec. 3.7.4). * Third paragraph: Here it says that the spectrum "peaks at kT=1 keV". Should this be "E=1 keV"? Also, the end of the paragraph refers to variability of IRS1, but I could not find this lightcurve plotted or any detail on that analysis. This should be fleshed out a bit more. * We have changed this to 'E = 1 keV'. We also now give more information on the variability of IRS1, and a verbal description of the light curve. The light curve is of rather poor quality (large error bars) and we don't think showing it as an additional figure would add much to the text. The representative light curves we already show in the paper (Figs. 4-8) are of much better quality. * Section 4.4, second paragraph: The parameters phi_c and n_c should be defined. Also, it is unclear what limit this problem is, w.r.t. phi_c and n_c; please elaborate on typical values of these parameters to help the reader understand the physical process better. * We have rewritten the 2nd and 3rd par. of Sec. 4.4 to clarify this, and have added values of phi_c and n_c to Table 5. * Section 5.1, first paragraph: This is a nice derivation of a key result. Section 5.1, last paragraph: This paragraph has an odd tone. The previous paragraph summarizes the fact that the line widths are narrow and that this is consistent with the f/i ratio, if the emission originates close to the star. Yet, the final paragraph starts by restating all of this and calling it a "difficulty" because the best models predict line-driven shocks further from the star. This paragraph could be shortened to by stating simply that if this emission is created by line-driven shocks, then the models need to be modified. Alternatively, start the paragraph by summarizing clearly what the models predict. * We've shortened this paragraph a bit and reworded it, and it is now more to the point without reiterating the previous paragraph.* Section 5.2, third paragraph: The end of this paragraph is not very precise. It says that the wind acceleration zone must be "near the star". However, considering the speed of the wind of 500 km/s, wouldn't this be still further than the standard wind shock region constrained by f/i? What exactly is "near" in this case? * The precise wind acceleration law close to the star is not well-known for hot stars, but the 'beta-law' we give in the 4th par. of Sec. 5.1 is almost always adopted, with beta = 0.5 - 1.0 considered realistic for hot star winds. In their text 'Introduction to Stellar Winds' (pg. 9), Lamers and Cassinelli give beta \approx 0.8 as a typical value for hot star winds. Adopting beta = 0.8 and using the terminal wind speed of 1060 km/s (Table 1), the beta-law give v = 550 km/s at a radius r = 1.79 R_star. This radius is consistent with the upper limit of 1.8 R_star that we get from the Mg XI f/i ratio (Table 5). So, there is no contradiction here. We have added a specific distance at the end of the penultimate sentence in the 3rd par. of Sec. 5.2 to make 'near' more precise. * Section 5.2: The discussion seems unnecessarily dismissive of the MCWS model. It is clear that the model may work for the data and that the magnetic field may remain undetected. Thus, couldn't the argument for this section be turned around to motivate a new search for the magnetic field in sigma Ori AB? * We have reread this section and in our opinion it is not 'unnecessarily dismissive' of the the MCWS model. We believe that we have gone as far with this model as is currently justified, given the lack of a magnetic field detection in sig Ori AB. Hopefully our discussion will give rise to new sensitive searches for a (weak) B-field in this star, and our estimates do provide a rough estimate of the minimum strength of the magnetic field that would be needed for wind confinement to be relevant. This estimate will be useful as a guide for the required sensitivity in future magnetic field detection experiments. Although we do believe that some discussion of the possible relevance of the MCWS model for sig Ori AB is justified given its young age, there is some risk of overemphasizing this model and we are trying to avoid doing this. There are so far no other data that we know of that strongly support a MCWS interpretation. Unlike magnetic rotators such as theta-1 Ori C, the X-ray emission of sig Ori AB is soft (no hard component visible in our spectrum), and there is no indication from H-alpha or UV observations for the characteristic rotational signatures of circumstellar matter. We've tried to strike a balance between mentioning MCWS as a possible alternative to the usual radiative wind shock interpretation without overplaying it, and in our opinion Section 5.2 does just that.* Page 20, third full paragraph: The last sentence has a confusing reference to HD150136. This should either be explained more fully or removed. * If sig Ori A does turn out to be a spectroscopic binary, then the analogy with HD 150136 may be very relevant. Because of this, we are not in favor of removing the discussion of HD 150136. We have now expanded the discussion of the HD 150136 system in the 5th par. of Sec. 5.3, which makes its potential relevance to sig Ori A clearer. * This and the next paragraph: This discussion notes that a close binary system (making sigma Ori AB a triple) could explain the observed X-ray emission. However, the lack of information on the star system limits the ability to draw a conclusion. Can you turn this around and use the observations (line ratio, the lack of modulation, etc.) to place constraints on the orientation of a putative tight companion to sigma Ori A? These specific predictions would help draw more concrete results from the current ambiguity. * Our X-ray observation does not place any tight constraints on the orbit orientation of the putative spectroscopic companion. One might be tempted to assert that absence of any evidence for an X-ray eclipse during our ~1-day observation rules out the possibility of a short ~1-day high-inclination spectroscopic orbit. But even that assertion would not be secure because the presence or absence of an X-ray eclipse depends on the relative sizes of the X-ray emitting regions of the two components, and these can differ substantially from the sizes of the stars themselves (both in coronal sources and in hot-star winds). Even in a short-period spectroscopic colliding wind system, the X-ray emission is likely to be a superposition of an extended colliding wind shock component with the hottest plasma on the line of centers between the two stars, plus contributions from the stars themselves. In fact, it is just this sort of multi-component picture we arrived at to explain the X-ray emission detected in our Chandra observation of the O+O spectroscopic binary HD 150136 (Skinner et al. 2005). It is quite possible that even a short-period high-inclination spectroscopic binary could undergo a physical eclipse without showing an X-ray eclipse. The only constraint that we can realistically place on the spectroscopic orbit from our X-ray data is the separation between the two components. As we have shown in Sec. 5.3, the colliding wind shock interface would need to lie at r = 1.6 R_star from the O-star in order to match the observed X-ray temperature with the maximum expected colliding wind shock temperature. And, this radius is consistent with the upper limit that we infer for the line formation radius of Mg XI, as we state. We don't believe the X-ray data justify going any further than this. Longer X-ray time monitoring over several days might prove interesting, but obtaining Chandra time for such time monitoring would be difficult. And, even if telescope time for several days of X-ray monitoring could be acquired, it would come in bits and pieces because of the strict limits on exposure lengths now being imposed because of Chandra thermal protection degradation. High-resolution spectroscopic monitoring or perhaps adaptive optics imaging or interferometric imaging will be needed to substantiate the existence of a spectroscopic companion and constrain the orbit. * Section 5.4, last two paragraphs: These two paragraphs are a bit confusing. Most of it discusses the possibility of a companion, but it is split between the first paragraph and the second half of the second paragraph. Rewriting may help make it clearer. Also, the coronal emission model is sandwiched between the rest of this, although it may deserve a bit more explanation. * We have rewritten the last par. of Sec. 5.4 and now mention the narrow lines of coronal sources last, and added two new refs. for TTS grating spectra. We think the explanation of the coronal model and why it doesnt jive with our sig Ori AB data is clear. The lack of hot plasma in the sig Ori AB spectrum, no detected variability, and broadened emission lines all argue against X-rays from a late-type coronal companion. We have also rewritten the last two sentences of the 1st par. of Sec. 5.4 to clarify that the model for zeta Ori proposed by Waldron & Cassinelli (2001) is actually a composite model that postulates both wind shocks and magnetic plasma confinement near the star (as opposed to magnetic confinement alone, as our original wording may have implied.) * Figure 17: The error bars overlap, so one can't see how they relate to each other. Please separate them. *Done. We've also added a sentence to the caption noting that a slight offset in T_max was introduced to avoid error bar overlap.* ------------- OTHER CHANGES: 1. Added a new age/distance reference (Mayne & Naylor 2008, MNRAS, in press) to the 6th and 8th paragraphs of Sect. 1. 2. Added a new reference to Donati et al. (2002) in the 7th par. of Sec. 3.7.2 and 2nd par. of Sec. 5.4. (This is the discovery paper reporting the B-field detection on Theta-1 Ori C.) 3. In the 4th par. of Sec. 3.7.4, we now give an expression for the X-ray luminosity of IRS1 with the distance left as a free parameter, as well as the value at d = 352 pc. 4. We have rewritten the penultimate sentence of the 3rd par. of Section 1 to note that the analysis of zeta Ori by Cohen et al. (2006) reached a different conclusion than that of Waldron & Cassinelli (2001).