Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
From: BXSVR0::OWOCKI 12-JAN-2001 10:16:41.12 To: DCOHEN CC: IGNACE,OWOCKI Subj: inquiry on X-ray line formation -- recombination or collisional excitation?? David, Attached below are some emails with Rico on the issue of whether X-ray lines might be optically thick to line-scattering. At the end is a postscript file of notes Rico composed on the matter. A key question in my eyes regards the formation mechanism of the observed X-ray lines. If they are formed by recombination, then it seems likely the absorption opacity won't be too important. If formed by collisional excitation from a ground state, it seems it might be. What do you know about the formation mechanism for these observed lines. Do you think they could ever be optically thick to line-absorption and scattering?? If so, then I think it is still possible to generalize our line emission analysis to account for the Sobolev escape probability in the angular redistribution. Rico says he's done some preliminary work that suggests including line optical depth could make the observed profiles more symmetric. But as you can see below, I'm don't really understand how that could happen. Anyway, I do agree this is something that needs to be considered, and so I would appreciate your input, particularly on the issue of line formation. Regards, Stan -----Attachment: email exchange with Rico From: SMTP%"[email protected]" 11-JAN-2001 12:43:26.11 To: [email protected] CC: Subj: Re: good news on my x-ray profile paper OK. I am glad yours is submitted, so that now I can reference your paper for doing the vlaw effects. By the way, a little more food for thought, one of the points raised by the original referee was the question of the line optical depth for strong permitted lines. I have done a quickie calculation. One can construct a kind of upper limit to the line optical depth for any given ion species using Sobolev theory. It helps to think about the integrated line optical depth T which relates to the column density for the atoms in question. The interesting things to fold in are (a) an exospheric approx, since line emission mostly comes exterior to r_1(E) and (b) a kind of filling factor related quantity C_x = N_x / N_w, the ratio of the hot gas column density to the wind column density. So C_x is not a well known quantity, but you can estimate it. Making the assumptions that the oscillator strength f=1, the ionization fraction for the ion in question q=1, then one can place a limit on T for elements of different abundances, and lines at different energies. I find that T << 1 for B stars (owing to low Mdot) and WR stars (owing to large r_1), but can be of order unity for O stars. And so far all of the chandra/xmm stuff we have seen is for O stars. So I began to think about potential optical depth effects for these profiles. What we have is collisional excitation (lets ignore any recombination contribution for the moment) followed by radiative decay. The star as an X-ray source can be ignored. So following the expressions in Mihalas, the source function looks like S = eps * B / (eps + beta - beta*eps) where eps must depend on density and beta is the escape probability. For tauL the Sobolev optical depth, the intesity at any impact parameter is I = S (1-exp(-tauL)) * exp(-tauC) for tauC the continuum optical depth. So in thin parts the line flux goes like emissivity times volume, but in thick parts, is is the source function times effective area. So I considered v ~ r, which is especially simple for Sobolev, just as a toy model. Sure enough, I got profiles that are reasonably symmetric, with slight blueshifts of the peak line emission. There is still some asymmetry between red and blue wings, as there must, but optical depth effects do introduce a new 'knob', and the lines being shown in papers are presumably the strongest ones. I haven't explored the parameter space very much. Anyhow, this could be something that one or both of us might pursue. Rico ================== RFC 822 Headers ================== Rico, This is a very interesting thought. I'll have to think a bit more about your upper limit approach and estimate. But at least from a first glance I'm very surprised, even perplexed, about your initial result that with optically thick line effects you were able to get more symmetric profiles. By my understanding, line optical depth effects might influence the details of the *angular* emission, through the escape probability. But it should only broaden the emission by the Doppler width assumed in your Sobolev treatment. If this is anything close to the ion thermal speed for even 10^7 K gas, for most ions it should be of order 100 km/s or less. This is much smaller than the ca. 1000 km/s directed motion of the wind, and so should not contribute much to the overall wind-broadened profile. Since the wind-expansion emission thus still has inherently much more optical depth for red-shifted gas emitted in the backside hemisphere, it remains hard to get much red-side emission. So for me the key question is, in your supposedly more symmetric emission, where does the red-side emission come from?? So for now I suggest we continue this dialog with a general aim of determing whether such optical effects could indeed be important for providing symmetric, broad profiles. If that somehow turns out to be the case, it would really be quite significant and I would be very much interested in being involved in any paper that demonstrates this. On the other hand, if you feel you already have done enough analysis to make a convincing case for this on your own, I'd be happy to provide external critique of your claims. Stan From: SMTP%"[email protected]" 11-JAN-2001 15:08:15.32 To: [email protected] CC: Subj: Re: good news on my x-ray profile paper Good thoughts. And no, I don't think that I have convinced myself of anything at this point, only that maybe that optical depth effects should be looked at. I would be more than happy to do this together. I really have only glanced at this. My code might even have a bug. Also my choice of v ~ r was just to be quick about it. To make it at all interesting, one has to choose rmax to be something a little large, but that is almost always too large to represent a realistic wind vlaw (I should think). For example the fwhm of my profiles are fairly narrow. I suspect that a beta law would broaden the line cores a bit. (Indeed, perhaps my claim at 'more symmetric' is just a contrast problem - the low wings gives the illusion of greater symmetry.) Anyhow, I think it is worth looking at, and I thought it would interest you. Just to get things rolling, let me send you my wee (and slightly terse) notes on the line optical depth as a ps file (below). Let me know your thoughts. (Things are pretty hectic here. Classes are about to start, and my wife is expecting this month, so we could have a newcomer any time, literally. If I start falling to answer any of your emails, that will be why!) Cheers, Rico From: BXSVR0::OWOCKI 11-JAN-2001 17:41:41.43 To: SMTP%"[email protected]" CC: OWOCKI Subj: Re: good news on my x-ray profile paper Rico, I've been thinking a bit more about your line-optical depth estimate. One question I have regards the line formation process, namely whether it is by recombination or collisional excitation. Your upper limit estimate assumes it is primarily the latter, but I though I understood from discussions with David that is was more the latter. If so, then the line opacity for absorption of X-ray line photons should be very small, and the whole process is essentially really a locally optical thin emission, subject then to a global continuum absorption, of course. Would you mind if I forwarded your emails to David and asked him to comment on this question of recombination vs. collisional excitation? Regards, Stan From: SMTP%"[email protected]" 11-JAN-2001 17:53:15.07 To: [email protected] CC: Subj: Re: good news on my x-ray profile paper >Rico, > >I've been thinking a bit more about your line-optical depth >estimate. One question I have regards the line formation >process, namely whether it is by recombination or collisional >excitation. Your upper limit estimate assumes it is primarily >the latter, but I though I understood from discussions with >David that is was more the latter. If so, then the line >opacity for absorption of X-ray line photons should >be very small, and the whole process is essentially really >a locally optical thin emission, subject then to a global >continuum absorption, of course. You refer to the 'latter' twice. RICO, YES I MEANT THE FORMER "LATTER" TO BE "FORMER". :) >Would you mind if I forwarded your emails to David and asked >him to comment on this question of recombination vs. collisional >excitation? > >Regards, > >Stan Sure, you can forward to David. I think the idea is that collisional excitation creates X-ray photons. But once created, do they freely stream away without interacting with atoms (i.e., of the highly ionized species). In that sense, you care about the resonance line scattering within the isovelocity zone. The referee pointed out that a line like CIV can have an optical depth of many thousands or even a million. Although there is much less X-ray gas, it is not a million times less than the normal wind. The question is what is the optical depth of strong permitted lines to X-ray photons that are created via collisional excitation followed by radiative decay? Rico --- CUT HERE TO EXTRACT POSTSCRIPT NOTES ---%!PS-Adobe-2.0 %%Creator: dvips 5.58 Copyright 1986, 1994 Radical Eye Software %%Title: opdepth.dvi %%CreationDate: Thu Jan 11 12:29:54 2001 From: To: CC: Subj: BXSVR0::OWOCKI 15-JAN-2001 10:55:53.20 SMTP%"[email protected]" DCOHEN,OWOCKI RE: correction to my previous comment Rico, Thanks for the quick response and query on my optical depth TeX notes. Here is my response. > Hi Stan. Thanks for you notes. Just a question and comment. > The fact that you composed notes for optical depth effects suggests > that you believe such effects might exist for the stars under > consideration. Would that be an accurate inference? Did Dave give you > any feedback on our previous email discussions? Yes, in a phone conversation we had on Friday, David said he believed the observed X-ray lines were formed by collisonal excitation, not recombination. Generally they are indeed resonance lines, so there should be a substantial number of atoms in the ground level. So in principle there could be cases when $\tau_\mu > 1$. I thus went back to my class notes to figure out what the likely outcome might be for a case of mixed scattering and thermal emission, parameterized in the usual two-level atom formalism by the collisional destruction parameter $\epsilon$. > In eqn II-2, is epsilon << beta really a good approx? If tau_mu >> 1 > for all mu, then > beta ~ (1/4pi) * Int dOmega (1/tau_mu) > However, epsilon is a collisional de-excitation parameter, normalized > to A_ul, I think. And since A_ul is probably the largest rate in this > problem, I should think that epsilon << 1 is good. Perhaps the source > function would be better represented as > S = epsilon*B / (epsilon + beta) > which only slightly tweeks your derivation, since for thin lines, beta >> eps, > and eqn II-8 still holds. Well even in a dense atmosphere for so-called collisionally dominated lines, e.g. Ca K and Mg k in the sun, this parameter is typically of order $\epsilon \sim 10^{-4}$, so I would imagine that in a stellar wind it would be truly tiny, of order $\epsilon \sim 10^{-8}$. Note in fact that if it were of order unity, you would recover LTE, and thus have an X-ray flux that would be on the order an X-ray photosphere with temperature of $T \sim 10^7-10^8$ K, which is enourmous, with a bolometric flux that is $(T/T_\odot)^4 \sim 10^16$ times larger than the solar surface! So clearly we can't be anywhere near LTE, and $\epsilon$ must be very small. Indeed, it is what gives the second factor of density in the standard $\rho^2$ emission picture. Anyway, your estimates for $\beta$ are correct, but your expression retaining $\epsilon$ term in the denominator suggests there could be cases for which $\epsilon \tau_\mu \ge 1$, and I doubt there are any lines that are that optically thick. In fact, I believe the case $\tau_\mu > 1$ also gives you something close to LTE. Thanks for your quick feedback. A much more subtle question that I glossed over a bit in my derivation is the way I have to move the $\Phi$ function off the exponential $e^{-\tau_\mu \Phi (x)$ and make instead the Heaviside step function multiply the full factor $1-e^{-\tau_\mu}$. I think what I did is formally ok, but I'm still puzzling over what pieces of physics I'm throwing away in this approximation. I think it only has to do with the specifics of the resonance zone, i.e. that the step actually occurs at around a thermal speed on the blue-edge of the local resonance, not at line-center. But I still need to ponder this a bit more. Keep those comments coming... Stan P.S. Just as I was about to send this reply, I got your further email: > Oops, what I meant for eqn II-7 and II-8 is that for thin lines, the > square brackets is still unity even if you use S = eps*beta/(eps+beta). > But now I guess eqn II-8 could end up different. Yea, see above. -Stan From: To: CC: Subj: BXSVR0::OWOCKI DCOHEN OWOCKI FYI, Rico's reply... 15-JAN-2001 12:15:19.41 From: SMTP%"[email protected]" 15-JAN-2001 11:19:57.61 To: [email protected] CC: Subj: Re: correction to my previous comment OK, I didn't have a good feel for just how small epsilon would be, and I agree that even for relatively thick lines (and we are probably talking tau<10 for these X-ray lines, since the X-ray gas is trace compared to the normal wind component), that beta >> epsilon should be good. Fair enough. I am happy to see that Dave thinks collisional excitation dominates recombination. In the revision of my paper, I had to respond to the question of optical depth effects, and I made a few brief comments in the revised paper. No space for a derivation (!), so I 'made speculation' and commented that it should be checked more rigorously. I knew that coll excit was important, but when you suggested that recomb might be dominant in some instances, my initial reaction was 'Oh crap' - very professorial! :) No baby yet. I guess our false alarm last weekend made us think that it could be any moment, but a whole week has passed. Still, Teresa (my wife) has been having some strong contractions, and is dilated about 3 cm (10 cm is the magic number, in case you don't remember), so it could come any time now. Even though we have been through this before, it is still exciting. Rico ================== RFC 822 Headers ================== Return-Path: [email protected] Received: from vesta.physics.uiowa.edu (128.255.34.140) by bxsvr0.bartol.udel.edu (V5.0A-1, OpenVMS V7.2-1 Alpha); Mon, 15 Jan 2001 11:19:56 -0500 Received: from homer.physics.uiowa.edu (homer [128.255.34.147]) by vesta.physics.uiowa.edu (8.8.8+Sun/8.8.8) with ESMTP id KAA05290 for <[email protected]>; Mon, 15 Jan 2001 10:19:54 -0600 (CST) Received: (from ri@localhost) by homer.physics.uiowa.edu (8.8.8+Sun/8.8.8) id KAA20741 for [email protected]; Mon, 15 Jan 2001 10:19:52 -0600 (CST) Date: Mon, 15 Jan 2001 10:19:52 -0600 (CST) From: Richard Ignace <[email protected]> Message-Id: <[email protected]> To: [email protected] Subject: Re: correction to my previous comment From: SMTP%"[email protected]" 15-JAN-2001 14:41:37.10 To: [email protected], [email protected] CC: Subj: Re: correction to my previous comment Hi, guys, Sorry to be out of touch over the weekend. I'm in a frenzy of class preparation now, but I promise to look over the notes and emails carefully in the next few days (I have looked them over quickly). Let me just provide a bit of information, which is basically a recap of the things I mentioned to Stan on the phone on Friday. The traditional view is that stellar X-rays come from 'coronal' plasmas. In this context, coronal means: - ionization/excitation equilibrium (i.e. steady-state ionization fractions and level pops) - ionization balance is set by collisional ionization and radiative recombination (note: inverse processes); And dielectronic recombination/autoionization are sometimes included, and can definitely be important - level populations are set by collisional excitation from the ground state and spontaneous emission (note: even though RR sets the ionization balance, recombination cascades are generally assumed (and simulations benchmarking both astrophysical observations and laboratory experiments show) that this makes a negligible contribution to the level populations within a given ionization state) -- implications/assumptions: densities are too low for collisional de-excitation or collisional excitation from excited states to be important; however, this assumption is not strictly a good one: metastable levels do get populated, and collisions (and even some radiative transitions in the UV that affect X-ray lines -- see Wayne and Joe's recent paper on zeta Ori) can be important between excited levels -- this can affect details of individual line ratios, etc., but not the overall spectrum. And for resonance lines, which are the most likely to be optically thick, of course, this is not a concern. - thermal radiation from the hot plasma itself is unimportant (i.e. no upward radiative transitions -- this is NOT true in situations where there's an external source of X-rays, which means anything that's accretion powered) -- but if some resonance lines are significantly optically thick, then this assumption would be wrong, and if this investigation shows this to be the case, then it may have some interesting effects on the X-ray spectrum generally (if I'm thinking correctly about this, it seems to me that by definition an optically thick resonance transition has significant radiative excitation and thus the population of the upper level can be boosted significantly over the coronal approximation, and excition from the upper level of a resonance transition then becomes more of a possibility -- perhaps leading to unexpectedly strong lines elsewhere in the spectrum; This would be akin to the Gabler and Gabler photon pumping mechanism in hot star atmospheres). So -- in the coronal approx -- you can picture a god-given ionization balance (well, at least a Arnaud and Raymond- or Dopita and Sutherland-given ionization balance), and then for any specific ion just think about collisional excitation from the ground state followed by spontaneous emission. Because of the low densities of these X-ray emitting plasmas, the ground states tend to be quite overpopulated w.r.t. LTE (i.e. big departure coefficients for the ground state). So, things are pretty conducive to high optical depths (except, as you mentioned, for the fact that we're talking about a small fraction of the available plasma being hot). In the cool star/coronal community, people sometimes worry a little bit about optically thick lines. The common wisdom is that the very thickest lines in the strongest sources are only marginally optically thick. But emission measures for hot stars can be somewhat bigger than for cool stars. Of course the big difference between the two situations is that in cool stars the plasma is assumed to be stationary, so the relevant length to put in the optical depth integral is the physical dimension of the coronal loop, rather than the Sobolev length. One interesting thing to look at might be the emission measure to column density ratio, because the emission measure is an easily observed quantity. The more you spread out the hot gas (i.e. the lower the density), the more column depth and thus opacity, you're going to get for a given emission measure (right?). Indications from recent observations (as well as old observations) are that the plasma is distributed over many stellar radii (though, again, bulk motions and the related Doppler shifting should help reduce the Sobolev optical depths). Finally, in the cool star community, resonance scattering is not thought to be that important, even if it does occur, because of the assumption of a static plasma and the fact that scattering conserves photons. The only time people worry about it is when the resonance scattering might happen very close to the photosphere so that back-scattered photons are destroyed by the stellar core. You probably knew or could surmise the important aspects of all this already, but that's my two cents for now. Good luck with the baby, Rico and congratulations on getting the paper accepted. You produce babies faster than I produce publications! David ================== RFC 822 Headers ================== Return-Path: [email protected] Received: from hven.dupont01.swarthmore.edu (130.58.80.90) by bxsvr2.bartol.udel.edu (V5.0A-1, OpenVMS V7.2-1 Alpha); Mon, 15 Jan 2001 14:41:36 -0500 Received: (qmail 30286 invoked from network); 15 Jan 2001 19:40:02 -0000 Received: from d55.dupont01.swarthmore.edu (HELO hven.swarthmore.edu) (130.58.80.55) by hven.dupont01.swarthmore.edu with SMTP; 15 Jan 2001 19:40:02 -0000 Message-ID: <[email protected]> Date: Mon, 15 Jan 2001 14:37:56 -0500 From: "David H. Cohen" <[email protected]> Reply-To: [email protected] X-Mailer: Mozilla 4.74 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: [email protected], [email protected] Subject: Re: correction to my previous comment References: <[email protected]> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: BXSVR0::OWOCKI 22-FEB-2001 22:05:10.25 To: KEN,DCOHEN,VIKRAM,ASIFUDEL CC: OWOCKI Subj: Penn State "core-dump"... David, Ken, Vikram, Asif: Just got back from PSU, fighting the snow. I thought I'd write a quick core dump while my experience is still fresh in my memory. I was invited by Richard Wade, who I met at the white dwarf conference in Delaware last June. Richard works of white dwarfs, cataclysmic variables, and dwarf novae. He's recently written several papers with Raman on wind variability in CV disk winds; see A&A for 2000, with Prinja as lead author. I gave my colloquium on "Winds that Sail on Starlight" in the Astronomy and Astrophysics department. It seemed well-received, and I had several people, especially grad students and post-docs, stay on to ask questions after the talk. I also had a few quite interesting one-on-one interactions. Unfortunately, the people on the Chandra ACIS (Advanced CCD Imaging Spectrometer) had a group meeting that conflicted with my talk, so I didn't get much interaction with them, with a couple of exceptions. After my talk, I spoke with two Chandra post-doc/RS's, Yoshi Maeda and Leisa Townsley. Yoshi told me about a particular wide high-mass boundary image taken with ACIS, which separated out the 2000 AU sepsaration between an early O (O4 I, I think) from a B0.5 V star. The surprise here was that essentially ALL the X-rays were coming from the O, and Yoshi estimated that the B0.5 star was a couple orders of mag down from a Lx~10^-7 Lbol law. It seemed peculiar to me, because that's the exact same spectral type as Tau Sco, n'est-ce-pas, yet is seems as much under luminous in X-rays as Tau Sco is overluminous (and hard). Leisa told me about a particular bright X-ray source in 30 Dor (I think), which had a quite hard, > 3 kev, spectrum. She didn't know off-hand the Lx/Lbol, but she said it was by far the brightest X-ray source in that quite crowded field. The advantage of this instrument is that it can resolve out most of that very crowded group (which, as I recall, was mistaken in pre-HST days for a supermassive star, n'est-ce-pas?). Any, I opined to her that I didn't think intrinsic X-ray emission from the wind could make that big and hard an X-ray flux, and we discussed whether it might be a binary, even interacting bubbles from close group of stars. It might be a case to look at with Vikram in interacting wind simulations. We had a group of about 10 for dinner, including 3 fac, 1 post-doc, and 6 grad. students. This a.m. I asked to speak with Niel Brandt, who is an asst. prof on ACIS team. In my pre-visit web browse, I noted he had recently discovered, apparently for the first time, P-Cygni profiles in X-rays. This was from an X-ray binary in Circinius, about 6 kpc toward galactic center, so it can hardly be observed in other wavebands, though it shines through nicely in hard X-rays. Because of this obscuration, it's not even clear what kind of binary it is, though apparently people lean toward low-mass with RLOF, instead of HMXRB with wind accretion. I haven't had a chance to look at the paper, though apparently the first one (of two) is now in print. In my discussions, we were wondering whether it is possible to have an X-ray line-driven wind. It seems the X-ray luminosity in this accretion disk is estimated to be near the Edd. limit all by itself. Apparently, Danielle Progra has already latched onto this idea, though I've not yet had time look at the reference. I gave Brandt a copy of the X-ray profile paper with Dave, and told him that, in principle, it should be possible to generate X-ray P-Cygni profiles for toy models. One of the great puzzles in this system is that the X-rays go from P-Cygni to pure-emission and back over time scales of less than a day, sometimes only an hour or so. This is much shorter than the inferred orbital period of ca. 16 days. They don't see any clear evidence of narrow components within the P-Cygni, at least down to the spectral resolution, which was a good as ~200 km/s for the lowest energy lines, about 0.5 kev, as I recall; in velocity units, this degrades in proportion to energy for high energy lines. Brandt also showed me some X-ray spectra of AGN/QSOs, non-BAL as I recall. Here the S/N and spectral resolution is not as good. So what he does is bin together several lines from the same species, or with similar ionization energies. In these composites, he does see some lines with P-Cygni shape, so I guess this means that QSO's can also have X-ray P-Cygni outflows, perhaps again even driven by X-rays. Another thing he mentioned is that the X-ray brightness of quasars anti-correlates with equivelent width of aborption features like CIV. This is perhaps not too surprising, since lots of X-rays would fry CIV, but it is also in general agreement with the Murray & Chiang model of disk winds in QSOs. Indeed, much of the above anti-correlation is attributed to the fact that BAL QSOs tend to have much lower Lx. After this, I talked with a couple of students working on QSOs. Rajib Ganguley told me about narrow absorption features that he sees at velocities ranging upto to 60,000 km/s!, but with widths < 500 km/s. I think these were HST observations of stuff like CIV and NV, for BAL QSOs. We discussed how these might be blobs formed in an unsteady disk wind, much as in the Proga et al. simulations. I also talked with Brandt's student, Sarah Gallagher. She has some broad band X-ray spectra of Seyfert galaxies and QSO's. Apparently, there is some kind of Seyfert (type II??) in which the central X-ray source is occulted, but can be still seen a high energies (many kev?) by Compton scattering off of some kind of galactic ring. Apart from the low overall X-ray brightness, another signature in this case is a relatively flat spectrum (index 0.2 vs. the usual 1.5-2 or so). As I understood it, Sarah had also some QSO (maybe BAL) Chandra data in which the X-rays were likewise reduced with a flat spectrum, but in this case there is no "ring". So the question was whether this could be backreflection of X-rays off the other of side of the accretion disk wind. We tried to estimate how big a disk Mdot you would need to keep the disk optically thick to Compton scattering over a large enough opening angle to scatter enough X-rays. She's going to work on this and get back in touch with me. Finally, I had an interesting chat with an asst. prof named Mike Eracleous. He works on AGNs and GRB; in fact, we had to cut short out disucssion because earlier that morning he got a beeper alarm that GRB had gone off, and had already an optical counterpart. He was going to try to see if they could observe it that night with this new big telescope they have in Texas, the "Hobby-Eberly telescope". Apart from this, we did also have some interesting discussion about AGN/QSO disk winds. He seemed to indicate that this Murray-Chiang model was having quite an impact in the QSO field, though it also had its critics. From his point of view, he said it fit many of the observational contraints quite nicely. Towards the end, we had an interesting discussion about what this model would predict for how line widths might vary with ionization stage. That is, would it be like WR winds, where the highest ionization stage are relatively narrow, since they're formed in the inner wind, or would perhaps higher stages come from a more inner region of the disk where the escape speed is higher, and thus makes a faster speed wind. All in all, a stimulating day and a half. It's a lot of work and a bit disruptive to make such visits, but it can also help stir up the idea machine. Anyway, I thought some of these intereractions might be things that we could discuss further. Clearly, David and I could talk more about the X-ray observation stuff, both for QSO's and the stars. Ken, I thought you might want to think more about this issue of whether X-rays themselves could DRIVEN a wind through scattering in X-ray lines. Maybe we can both take a look at what Progra has been up to in this and other aspects of disk winds. (BTW, I heard from Mike that Progra will likely go to JILA, and so not take the post-doc position in Pittsburgh with Hillier and Turnshek, on that NSF grant that I was a co-I on.) Finally, it might be good for Vikram and me to talk about the 30 Doradus stuff in the context of X-rays from interacting wind bubble models. That's all for now. Keep in touch. Stan P.S. The web line for Brandt's X-ray P-Cygni stuff is: http://www.science.psu.edu/alert/Brandt11-2000.htm The other people can be accessed via the PSU A&A department personnel page: www.astro.psu.edu/deptinfo/faculty.html From: SMTP%"[email protected]" 23-FEB-2001 13:34:21.50 To: [email protected] CC: Subj: Re: Penn State "core-dump"... Stan, Thanks for the information about your very productive and interesting sounding trip to Happy Valley (as I think they call Penn State). A couple of quick things, with more to follow later: - Brandt is not the only one who's observed P Cygni profiles in the X-ray with Chandra. I've put a poster from Hawaii in the mail to you... - I've put the submitted version of Joe's paper on the Chandra spectrum of zeta Pup in ftp://hven.swarthmore.edu/pub/outgoing/stan/zpup15feb.ps Please let me know when you've downloaded it, I'd like to delete it from there. I was waiting till I had a chance to read it over one last time (minor corrections before submission allegedly) before I gave it to you, but I've been too busy for even that. - Duane Liedahl, who's worked on that Vela X-1 X-ray analysis involving clumped winds, wants to come visit us sometime to talk about clump physics and clump evolution/survival. We should talk about this sometime. OK, I've got a student in the office...talk to you soon. David ================== RFC 822 Headers ================== Return-Path: [email protected] Received: from hven.dupont01.swarthmore.edu (130.58.80.90) by bxsvr2.bartol.udel.edu (V5.0A-1, OpenVMS V7.2-1 Alpha); Fri, 23 Feb 2001 13:34:21 -0500 Received: (qmail 22627 invoked from network); 23 Feb 2001 18:33:47 -0000 Received: from d55.dupont01.swarthmore.edu (HELO hven.swarthmore.edu) (130.58.80.55) by hven.dupont01.swarthmore.edu with SMTP; 23 Feb 2001 18:33:47 -0000 Message-ID: <[email protected]> Date: Fri, 23 Feb 2001 13:30:25 -0500 From: "David H. Cohen" <[email protected]> Reply-To: [email protected] X-Mailer: Mozilla 4.74 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: [email protected] Subject: Re: Penn State "core-dump"... References: <[email protected]> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: BXSVR0::OWOCKI 26-JUN-2001 17:49:21.09 To: DCOHEN CC: OWOCKI Subj: toe-step concern... David, I've had an interesting email exchange with Ken, attached below. The bottom line is that it seems the new X-ray profile paper including optical depth effects actually has Rico as first author, with Ken saying his participation was modest. For me, this puts a bit different spin on the whole thing. I had basically given Ken the green light to run with the ball as he saw fit on any problem we may have had conversations about. The idea is that he needs to establish his independence from me as part of his upcoming tenure. But as you may recall, Rico and I had several quite lengthy exchanges about line optical depth effects, and I even sent him some pretty extensive notes. In case you're interested in looking these over, I've put on my ftp site two files containing the full collection: ftp.bartol.udel.edu/Owocki/xray/ignace_xray_emails1.jan01 ftp.bartol.udel.edu/Owocki/xray/ignace_xray_emails2.feb01 The first file contains emails dealing with our respective papers from last winter. The second file deals specifiicaly with exchanges we had about line optical depth effects. Obviously, I'm a bit perturbed that Rico would run with the ball in this way without even apparently considering the toe-step issue. To me this makes it pretty hard to trust him with future email exchanges of ideas. Moreover, given that we even offered him co-authorship on our paper on the pure possibility of toe-stepping, I'm really quite flabbergasted that he didn't consider this as an issue for his paper. I'd be interested in getting your perpective on this. I've just gotten an email from Rico with the ps file of the paper attached. I've not yet had time to read it or compare the method with ideas laid out in our email exchange. Maybe it'll turn out to be completely independent, but given the closeness in subject matter, that seems a priori unlikely. I'll send this ps file along in next email. Perhaps we can chat about this when you visit here later this week. Speaking of which, I have indeed decided to buy a Platinum Powerbook. It should arrive tomorrow. So you don't need to bring yours, but if you can do, maybe we can try to mate them to save $ for the next generation of laptops! :) Let me know your visit plans as they develop. Stan -------- #498 26-JUN-2001 16:28:33.03 GAYLEY From: SMTP%"[email protected]" To: [email protected] CC: Subj: Re: recommend you read Achim Feldmeier's habilitation thesis; file on my ftp site... Thanks Stan, the unpacked version worked fine. I have no idea what my system didn't like about the gzipped version, but you know how these things go. Also, I think I might get full funding from NSF also! If so, I'm convinced that I was the absolute last proposal in the queue to get funded. So I got lucky, but I'll take it, what with tenure review right around the corner. Truly this was a banner year for hot stars, with you getting your B-field modeling at last, Nordsieck getting the Hanle effect at last, and me getting the WR line-bunch stuff, an idea that I couldn't even sell my own university on for internal grants. I guess the lesson is, you can be in an unfashionable area, but you have to do your homework and above all have a lot of patience! On the science side, I may or may not have seen Dave Cohen in Boston, who may or may not have been participating in the Chandra review. If this meeting occurred, I'm sure I would have told him about the recent wrinkle that Rico and I have added to the X-ray profiles. We just submitted a short analysis of what happens if the lines are optically thick, and lo and behold, they get a lot more symmetric if the emission comes from the region of constant expansion. You can quickly convince yourself this will be true if you note that the Sobolev emission will be primarily in the azimuthal direction, favoring a generally Gaussian shape to the whole profile. You can get it even more symmetric if you consider re-absorption within the Sobolev resonance zone itself, since then wind absorption plays a less important role. With negligible wind re-absorption, the profile from an optically thick resonance zone is simply a semi-circle. However, none of the profiles we get look exactly like Joe's observations, so there are still a lot of issues about the geometry of the emission and the wind absorption to play with. Overall though, I would argue that optical depth effects may go a long way toward explaining profile symmetry. Now if we could just understand all this f/i/r stuff... -Ken #499 26-JUN-2001 16:42:45.15 GAYLEY From: BXSVR0::OWOCKI To: SMTP%"[email protected]" CC: OWOCKI Subj: Re: recommend you read Achim Feldmeier's habilitation thesis; file on my ftp site... Ken, Glad the unpacked file was ok. It's distressing that a simple thing like gzipping can go awry. One would thing that this would be pretty standard nowadays. Congrats on the NSF success! When it rains it pours, I guess... The timing wrt tenure review seems very fortunate indeed. It does seem that persistence and patience can pay off. On science, I may or may not have had a similar conversation with David about Chandra etc. Your X-ray results sound interesting indeed. If you've already submitted a paper, might I get a preprint?? Or would you rather wait till it comes back from referee?? I wouldn't worry too much about the fir stuff. If you look at the Zeta Ori and Zeta Pup papers from Joe C. and Wayne, I believe there's only one fir ratio that seems at odds with wind formation. And these are generally the highest energy transitions, which I believe are deexcited by relative far UV photons, i.e. in Ly continuum, where the intensities could well be off by a few 10% percent. That could be enough to lift the formation radius above the photosphere. I'm off to Europe soon, visiting UCL, Potsdam, Holland, and Belgium from July 10-Aug 4, also attending IAU meeting on Pulsation in Belgium. So I gotta get back to work on talk prep. Keep in touch. Stan #500 26-JUN-2001 17:04:52.38 GAYLEY From: SMTP%"[email protected]" To: [email protected] CC: [email protected] Subj: Re: recommend you read Achim Feldmeier's habilitation thesis; file on my ftp site... Rico is the first author, so I'll forward him your preprint request, though I'm sure he'll be happy to forward it to you. We'd be pleased to have your comments, it's likely they would be more beneficial than the review process itself! All I really did was the self-absorption within the resonance zone, and some prose-tightening. I think it's quite interesting that we now have three universal profile shapes to play with for constant expansion: very thin, very thick, and lots of local re-absorption. Of course all of this applies only to very idealized emission geometries. I think the next place to go is to look at how various geometries of both emission and wind absorption (i.e., shells, blobs, filaments, etc.) will alter the profile characteristics. If I had had the recent opportunity to look at Chandra proposals, I would have been struck by how glibly observers think that they will translate observed spectra into meaningful shock physics. The reality is, that capability really does not currently exist in any kind of quantitative way, beyond flux ratios and rough conclusions about temperatures and degrees of ionization. I've never even seen an argument that equilibrium models should be expected to apply, and I think the presence of lots of diffuse emission argues for the importance of heat conduction, an issue we've wondered about in the past. If optically thick emission is included, then it complicates things even more, but also it gives an additional handle on the escape physics and hence the actual unresolved structure. So it might actually be helpful rather than just another complication to look at optically thick X-ray line profiles, if they indeed exist. -Ken From: BXSVR0::OWOCKI 26-JUN-2001 18:26:11.77 To: DCOHEN CC: OWOCKI Subj: toe-step concern decreasing.. David, I've only had a chance to glance over Rico and Ken's paper, but it's pretty interesting. There's quite a lot new in there that was not part of our email dicussion per se, so my level of peevishness is decreasing. But I still don't fully understand the result physically. It seems that continuum optical depth effects should still attenuate red-side of profile. The Sobolev escape just means that line photons prefer to escape out the side, and so should appear almost unshifted, i.e. near line center. I still need to understand how the red emission near v_infty can be so high... Anyway, it seems like a good paper, and the results seem almost analytic, so it should be possible to figure out what's going on. Stan From: SMTP%"[email protected]" 26-JUN-2001 23:34:49.15 To: [email protected] CC: Subj: Re: toe-step concern decreasing.. Stan, I printed out Rico's paper, and I still have your and Rico's emails from last winter sitting face-up on a shelf, as if I'm about to do something with them. I haven't read it yet, but printing it out is a good first step. I was thinking of coming down on Friday morning, but going back to PA at 2:00. Would that work OK for you (meeting at Bartol at 9:30, or something)? Should we ask Marc Gagne to join us? I gave you the Noesys disk, right? Could I temporarily borrow it back? David [email protected] wrote: > David, > > I've only had a chance to glance over Rico and Ken's > paper, but it's pretty interesting. There's quite a > lot new in there that was not part of our email > dicussion per se, so my level of peevishness is > decreasing. > > But I still don't fully understand the result > physically. It seems that continuum optical depth > effects should still attenuate red-side of profile. > The Sobolev escape just means that line photons > prefer to escape out the side, and so should appear > almost unshifted, i.e. near line center. I still need > to understand how the red emission near v_infty can > be so high... > > Anyway, it seems like a good paper, and the results > seem almost analytic, so it should be possible to > figure out what's going on. > > Stan ================== RFC 822 Headers ================== Return-Path: [email protected] Received: from unknown.hostname (130.58.80.90) by bxsvr1.bartol.udel.edu (V5.0A-1, OpenVMS V7.2-1 Alpha); Tue, 26 Jun 2001 23:34:47 -0500 Received: (qmail 3550 invoked from network); 27 Jun 2001 03:34:03 -0000 Received: from ppp64.remote.swarthmore.edu (HELO hven.swarthmore.edu) (130.58.24.64) by hven.dupont01.swarthmore.edu with SMTP; 27 Jun 2001 03:34:03 -0000 Message-ID: <[email protected]> Date: Tue, 26 Jun 2001 23:33:35 -0400 From: "David H. Cohen" <[email protected]> Reply-To: [email protected] X-Mailer: Mozilla 4.73C-CCK-MCD {C-UDP; EBM-APPLE} (Macintosh; U; PPC) X-Accept-Language: en MIME-Version: 1.0 To: [email protected] Subject: Re: toe-step concern decreasing.. References: <[email protected]> Content-Type: text/plain; charset=us-ascii; x-mac-type="54455854"; x-mac-creator="4D4F5353" Content-Transfer-Encoding: 7bit From: BXSVR0::OWOCKI 27-JUN-2001 10:00:17.85 To: SMTP%"[email protected]" CC: OWOCKI Subj: Re: toe-step concern decreasing.. David, > I printed out Rico's paper, and I still have your and Rico's emails from > last winter sitting face-up on a shelf, as if I'm about to do something > with them. I haven't read it yet, but printing it out is a good first > step. I read it over last night. It's actually a bit sloppy and awkward in many places (for example in their eqn. (31), the \sin \theta and \theta are upside down); but I now think the basic results are probably right. For me a key point is that the optically thick Sobolev escape tends to increase horizontal over vertical emission, and this tends to make the profile **narrower** and somewhat more symmetric. This was the essential fault in my earlier reasoning, namely I kept wanting a mechanism to make the profiles **broader** and more symmetric. In Rico and Ken's paper, the broadening comes entirely from assuming the profile forms at the terminal speed, so there's plenty of that. But then the line optical thickness transforms the profile from a flattop to a more Gaussian form, since relatively more of the emission is coming from the side lobes with small l.o.s. velocity. So I guess I have to give Rico credit for following through on actually computing the line profiles. We had all the tools to do so, but I got distracted and never did. I'm kind of kicking myself... The good news is that Rico and Ken's analysis is again unnecessarily restricted to constant velocity, in order to get fully analytic profiles. It is relatively straightforward to extend our more general beta-law analysis to the case of optically thick lines. In the next email I'll forward some initial TeX notes on how to do this. Since I'm busy preparing for the Europe trip, I don't have time right now to code this option into the Mathematica notebook, but that should be pretty easy. Thus, it should be possible to reproduce all the plots of our paper for the optically thick line case. This potentially could make a good, quick, follow-up paper. The treatment of the general case of marginally thick lines is a bit trickier, since the angle escape integral cannot generally be done analytically then. We'll have to think about whether it would be worthwhile to examine this "transitional" case. > I was thinking of coming down on Friday morning, but going back to PA at > 2:00. Would that work OK for you (meeting at Bartol at 9:30, or > something)? That'd be fine with me. > Should we ask Marc Gagne to join us? Sure. > I gave you the Noesys disk, right? Could I temporarily borrow it back? Sure. See you Friday a.m. Stan From: BXSVR0::OWOCKI 27-JUN-2001 10:07:08.80 To: DCOHEN CC: OWOCKI Subj: thick_em_mod.tex ; TeX notes on modifying our beta-law emission analysis for optically thick lines... % % thick_em_mod.tex % % TeX notes on modifying Owocki & Cohen (2001) X-ray line-emission % analysis to include case of optically thick lines. % % initial version of June 27, 2001 \def\blankline{\par\vskip \baselineskip} \headline={\hss {S. Owocki - Draft notes 6/27/01}} \centerline{\bf Optically thick line-emission for beta velocity law} \blankline It is a relatively straightforward to convert the line-emission analysis in our Owocki and Cohen (2001) paper to the case of optically thick lines. The central modification needed is to put the Sobolev angle escape into the general emission integral, simply by multiplying the emissivity, e.g. in eqn. (7) of our paper, by the angle factor $$ { p (\mu ) \over \beta } \eqno(1) $$ where the angle-dependent Sobolev escape probability is given by $$ p (\mu) = { 1 - e^{-\tau_\mu} \over \tau_\mu }, \eqno(2) $$ with it's angle average given by $$ \beta = {1 \over 2 } \int_{-1}^1 d\mu p(\mu) . \eqno(3) $$ Recall here that $\tau_\mu = \tau_0/(1+\sigma \mu^2 $, with $\sigma = (r/v)(dv/dr) - 1$ and $\tau_0 = \kappa v_{th} \rho r/v$. For an optically thin line, this factor simply approaches unity, but in the optically thick limit, it becomes $$ { p (\mu) \over \beta } \rightarrow { 1 + \sigma \mu^2 \over 1 + \sigma/3} ~~ ; ~~ \tau_\mu \gg 1 \eqno(4) $$ For Rico and Ken's case of constant flow speed v, $\sigma = -1$, and so the angle factor reduces to $$ { p (\mu) \over \beta } = {3 \over 2} (1 - \mu^2) . \eqno(5) $$ Moreover, in such a constant speed flow, the Doppler-shifted wavelength (in units of $v_\infty$) depends simply on this angle, $x = -\mu$. As such, the overall profile simply gets multiplied by a new factor of $1-x^2$, which thus makes it much more "rounded", with greater emission at line center, $x = 0$. In a more general beta velocity law, we simply need to evaluate $\sigma $ as a function of radius. In term of the inverse radius coordinate $u = R_\ast/r$, this becomes $$ \sigma (u) = \beta u (1-u) - 1 . \eqno(6) $$ [Note in particular, for constant speed ($\beta=0$), we recover $\sigma = -1$.] After angle integration over the delta function (our eq. (8)) converts $\mu \rightarrow -x/(1-u)^\beta$, we find that we just multiply the integrand in our eqn. (9) by the factor $$ { 1 + \sigma (u) x^2/(1-u)^{2 \beta} \over 1 + \sigma (u)/3} . \eqno(7) $$ The more general case when a line is only marginally optically thick, or goes from optically thick in the inner wind to thin in the outer wind, is rather messier to deal with, essentially because the $\beta$ integral is harder to evaluate analytically (maybe even impossible in general -I have to look at this...). But both the pure thin and pure thick limits are quite easy to compute, e.g. using the Mathematica analysis with a single numerical integral over radius. \bye