1: \documentclass[12pt,preprint]{aastex}
2:
3: %% manuscript produces a one-column, double-spaced document:
4: %%\documentclass[12pt,manuscript]{aastex}
5:
6: \newcommand{\cmcubed}{\hbox{${\rm cm^{\rm -3}}$}}
7: \newcommand{\kmps}{\hbox{${\rm km~s^{-1}}$}}
8: \newcommand{\Rsun}{\hbox{${R_{\sun}}$}}
9: \newcommand{\Msun}{\ensuremath{M_{\sun}}}
10: \newcommand{\Mdot}{\hbox{${M}\kern-0.6em^{^{\bullet}}$}}
11: \newcommand{\MLrate}{\hbox{$\Msun~{\rm yr}^{-1}$}}
12: \newcommand{\Ro}{\hbox{${R_{*}}$}}
13: \newcommand{\Rhox}{\hbox{$\int$\kern-0.05em$\rho$dx}}
14: \newcommand{\Texc}{\ensuremath{T_{\rm exc}}}
15: \newcommand{\Telect}{\ensuremath{T_{\rm e}}}
16: \newcommand{\nelect}{\ensuremath{n_{\rm e}}}
17: \newcommand{\nion}{\ensuremath{n_{\rm i}}}
18: \newcommand{\nhyd}{\ensuremath{n_{\rm H}}}
19: \newcommand\threehalf{\slantfrac{3}{2}}
20: %%\newcommand{\A{1}}{\hbox{A$_{\rm #1}$}
21: \newcommand{\Hyd}[1]{\hbox{\ion{H}{#1}}}
22: \newcommand{\C}[1]{\hbox{\ion{C}{#1}}}
23: \newcommand{\N}[1]{\hbox{\ion{N}{#1}}}
24: \newcommand{\Oxy}[1]{\hbox{\ion{O}{#1}}}
25: \newcommand{\Al}[1]{\hbox{\ion{Al}{#1}}}
26: \newcommand{\Ca}[1]{\hbox{\ion{Ca}{#1}}}
27: \newcommand{\Mg}[1]{\hbox{\ion{Mg}{#1}}}
28: \newcommand{\Fe}[1]{\hbox{\ion{Fe}{#1}}}
29: \newcommand{\Si}[1]{\hbox{\ion{Si}{#1}}}
30: \newcommand{\pubdao}{Publ.\ Dominion Astrophys.\ Obs.\ Victoria}
31: %%\notetoeditor{}
32:
33:
34: \shorttitle{31~Cyg and Single Stars}
35: \shortauthors{Eaton}
36:
37:
38: \begin{document}
39:
40:
41: \title{A MODEL FOR THE CHROMOSPHERE/WIND OF 31~CYGNI \\
42: AND ITS IMPLICATIONS FOR SINGLE STARS}
43:
44: \author{Joel A. Eaton}
45: \affil{Center of Excellence in Information Systems, \\
46: Tennessee State University,\\
47: Nashville, TN}
48:
49: \email{eaton@donne.tsuniv.edu}
50:
51:
52: \begin{abstract}
53: I develop a detailed empirical model for the chromosphere and wind of 31~Cyg
54: based on a previously published analysis of {\it IUE} spectra from the 1993
55: eclipse and on the thermodynamics of how the wind must be driven. I then use
56: this model to interpret observations of single supergiant stars and to assess
57: the evidence that their winds are fundamentally different from those of
58: supergiants in the binary systems. This model naturally predicts a certain
59: level of clumping of the gas to balance the pressure that drives the wind.
60: It also predicts that anisotropic turbulence, such as would result from
61: transverse displacements of Alfven waves directed along radial magnetic flux
62: lines, would not give the roughly Gaussian profiles of emission lines seen
63: in cool giant stars. Furthermore, it implies that \C2] may not tell us much
64: at all about general conditions in chromospheres. Finally, I speculate that
65: chaotic magnetic fields, in dynamical equilibrium with the gas of the wind,
66: are the actual driving mechanism.
67: \end{abstract}
68:
69: \keywords{stars: winds --- stars: chromospheres}
70:
71:
72: \section{INTRODUCTION}
73:
74: It's no great stretch to assume that all cool stars (later than mid-F spectral
75: type) have chromospheres and, likely, winds as well. These are regions above the
76: photosphere too hot to be heated radiatively; rather, they must be heated by magnetic
77: processes or by shocks. Most analyses such stars' chromospheres and winds are based
78: on interpreting the strengths and shapes of emission lines in their spectra. The
79: Sun is the only one for which we can actually map out the chromosphere in any
80: appreciable spatial detail. All the others are unresolved, which limits us to
81: interpreting global properties of the emitting gas in terms of models without any
82: real spatial information (zero-dimensional). There are a number of standard
83: diagnostics used in this work. For cool giant and supergiant stars, we have the
84: \C2] intersystem multiplet at 2325 \AA\ (Stencel et al.\ 1981), the \Al2] multiplet
85: at 2669 \AA\ (e.g., Judge 1986a, 1986b; Harper 1992), the \Mg2\ $h$ and $k$ lines
86: (e.g., Stencel et al.\ 1980; Milkey et al.\ 1975), an abundance of \Fe2\ lines in
87: the UV (ultraviolet) affected by scattering in winds (Judge \& Jordan 1991),
88: H$\alpha$ (Johnson \& Klinglesmith 1965; Mallik 1993), and the \Hyd1\ Lyman lines
89: (Landsman \& Simon 1993; Haisch et al.\ 1977). Many of lines involved fall in the
90: UV because of the temperatures and densities obtaining in these chromospheres, so
91: our knowledge of cool giants' chromospheres necessarily has developed through
92: observations from {\it IUE}, the {\it International Ultraviolet Explorer} satellite
93: (e.g., Judge 1988) and more recently {\it HST}, the {\it Hubble Space Telescope}
94: (e.g., Carpenter et al.\ cited below). Other observations of these chromospheres
95: involve measuring radio thermal emission from their winds (Drake \& Linsky 1986;
96: Harper et al.\ 2001).
97:
98: An alternative, complementary approach to mapping the properties of chromospheres
99: solely by interpreting observations of the emission lines of single stars is to use
100: the $\zeta$~Aurigae binaries (Wright 1959, 1970; Wilson \& Abt 1954) to probe their
101: chromospheres and winds in greater spatial detail (legitimately at least
102: one-dimensionally). These systems consist of a cool supergiant paired with a B dwarf,
103: the prototype system $\zeta$~Aur consisting of K4~Ib and B5~V components. Absorptions
104: in the spectrum of the B star by the gas in the wind and chromosphere of the K star give
105: us a spatial probe of the K star's wind. Again, the telling absorptions are in the
106: UV, so most of our knowledge of these stars has come from {\it IUE} (e.g.,
107: Eaton 1996) with incremental improvements from {\it HST} (e.g., Baade et al.\ 1996).
108: More recently Harper et al.\ (2005) have used radio observations of $\zeta$~Aur
109: to verify the atmospheric structure they derive from UV lines.
110:
111: The purpose of this paper is to develop a detailed physical model for the chromosphere/wind
112: of 31~Cyg, to test it, and then use it to interpret observations of emission lines of single
113: stars. There is a commonly held belief, even prejudice, that the chromospheres/winds of
114: single stars are different than the those of binary components in that winds of single stars
115: accelerate much more rapidly than winds of the binaries and reach lower terminal velocities.
116: This paper assesses whether the observations of single stars might be consistent with the
117: slower acceleration of binary components. I shall begin with a review of the picture
118: developed for single stars through traditional analyses (\S\ 2), then discuss the model
119: for cool supergiants developed from observations of the $\zeta$~Aur binary 31~Cyg (\S\ 3),
120: testing it through comparisons with emission lines and radio continuum emission. I then
121: use this model to interpret the measurements for single stars (\S\ 4) and give some
122: speculations about wind acceleration (\S\ 5). My approach follows the radically different
123: viewpoint that pressures supporting and accelerating the outer atmosphere are a
124: {\it microscopic} phenomenon, operating on scales smaller than we resolve with our
125: observations, and are not simply the manifestation of {\it global} waves.
126:
127:
128: \section{THE GENERAL PICTURE FOR SINGLE STARS}
129:
130: Our observational understanding of single stars consists of several types of data.
131: (1) relative and absolute fluxes of a wide variety of UV and optical chromospheric
132: emission lines, (2) profiles of optically-thin emission lines, reflecting the kinematics of
133: gas forming them, (3) velocity structure of optically thick emission lines, reflecting bulk
134: flows in the chromospheres/winds, (4) line strengths of density-sensitive multiplets, namely
135: \C2] UV0.01 $\lambda$2325; and (5) strengths of fluorescent lines, reflecting mean intensities
136: in some of the deeper parts of the chromosphere. Many of these results come from {\it HST}
137: spectra discussed in observational papers by K.G. Carpenter and his collaborators (Carpenter
138: et al.\ 1991,1995,1997,1999; Robinson et al\ 1998); the more imaginative interpretations
139: of them seem to have come mostly from P.G.\ Judge (e.g., Judge 1994).
140:
141: The important results of these papers are (1) measurements of electron density in some part
142: of the emitting region of the chromosphere, (2) turbulent velocities from optically-thin
143: emission lines, (3) possible evidence for multiple components in the profiles of \C2], but
144: not other optically-thin lines, as well as in the strengths of fluorescent molecular lines,
145: (4) some indication of the temperatures in the chromospheres/winds from excitation of \Fe2\
146: and relative strengths of various emission lines, (5) detection of slight amounts of more
147: highly ionized, possibly hotter, gas (e.g., \C4\ emission), and (6) evidence for atmospheric
148: expansion and a crude measurement of the density-velocity profile from self-reversed \Fe2\
149: lines and \Mg2\ $h$ and $k$.
150:
151: \paragraph{Chromospheric Velocity Structure:} Many of the weaker intrinsic lines of common
152: elements are weak enough in chromospheric spectra as to be optically thin. They should be
153: formed in essentially nebular conditions, with every excitation giving a potentially
154: observable photon, so that their line profiles would reflect the kinematics and ionization
155: structure of the emitting gas. The superior resolution and signal to noise of {\it HST}
156: have made it possible to measure their profiles reliably. The turbulence measured from
157: such an optically-thin line can be conveniently characterized by a velocity, $v_{0}$=FWHM/1.67,
158: the parameter in the Gaussian part of the velocity distribution [exp$^{-(v/v_{0})^{2})}$],
159: which Judge (1986a, 1986b) called $b$. Carpenter et al.\ fit Gaussians to such lines in a
160: variety of cool stars. We may estimate the turbulence by averaging their results for
161: roughly nine lines between 1900 and 2850 \AA, with the \C2] multiplet excluded, which give
162: FWHM = 18.2 \kmps\ ($\alpha$~Tau), 29.9 ($\lambda$~Vel), 27.2 ($\alpha$~Ori), and 23.6
163: ($\gamma$~Cru). I have corrected these values for the 13--15 \kmps\ resolution of the
164: {\it GHRS} spectrograph which Carpenter et al.\ mention occasionally in passing, except in
165: the cases they explicitly stated that they had corrected their published values. These
166: FWHM's correspond to turbulent velocities in the range $v_{0}$ = 10.9--17.9 \kmps\ for the
167: emitting gas. This is well supersonic for the temperatures expected in these chromospheres
168: ($\lesssim$ 10,000~K).
169:
170: The most perplexing result of such kinematic analyses is the existence of broad wings and a
171: $\sim$4 \kmps\ redshift of the \C2] lines in most of these stars. These may well not be
172: formed under similar conditions as most other weak lines, and, in fact, those other lines
173: (such as fluorescent \Fe1\ and weak lines of \Fe2\ and \Al2]) do not show these phenomena.
174: This discrepancy implies some sort of anisotropic turbulence seen only in \C2] or, perhaps,
175: multiple components in the chromosphere, again not seen in the weak non-\C2] lines. Single
176: Gaussians fit to the \C2] profiles in the various stars give values of FWHM = 24 \kmps\
177: ($\alpha$~Tau), 28 ($\gamma$~Dra), 36 ($\lambda$~Vel), 35 ($\alpha$~Ori), and 30 ($\gamma$~Cru),
178: not corrected for the point spread function of the spectrograph. For $\gamma$~Cru a two-component
179: fit gives FWHM=27/42 for core/wings; for $\alpha$~Ori, FWHM=19/48. For $\lambda$~Vel, the star
180: most like 31~Cyg, we have a characteristic velocity of 21.6 \kmps, presumably uncorrected for
181: the resolution of the spectrograph, or 19.9 \kmps\ corrected; again, supersonic for the expected
182: chromospheric temperatures. Of particular significance would be any differences of width in
183: these optically-thin lines; if the \C2] lines were formed primarily in a different region
184: from the others, they would not necessarily have the same broadening. Indeed, the \C2]
185: lines are marginally broader than the other lines in $\alpha$~Tau ($\Delta$FWHM = 1.6 \kmps),
186: $\lambda$~Vel (3.4 \kmps), and $\gamma$~Cru (2.7 \kmps). These differences must at least
187: partially reflect the non-Gaussian profiles of \C2].
188:
189: \paragraph{Electron Densities:} These come primarily from the density-sensitive line ratios
190: of the \C2] (Stencel et al.\ 1981). Electron densities derived from the \C2] line ratios in
191: {\it HST} and {\it IUE} spectra of typical stars are near 10$^{9}$ \cmcubed. It is important
192: to remember that these values apply only to the parts of the chromosphere/wind where the photons
193: are formed. Because C is theoretically expected to be mostly neutral in the deeper reaches
194: of the chromosphere, one might expect that the measured electron densities would apply only
195: to the {\it warmer} parts of the chromosphere/wind. The \C2] lines also give limits on the
196: optical depths of chromospheric gas.
197:
198: Judge (1994) analyzed the variations of very good observations of $\alpha$~Tau with Doppler
199: shift to derive electron density as a function of velocity. Again, his results were perplexing.
200: They gave higher electron densities, by a factor of four, for positive velocities (away from us;
201: ostensibly toward the star) than for negative velocities. Furthermore, the emission in $\alpha$~Tau
202: was shifted to the red by 4 \kmps, which Judge interprets as a downflow in the denser regions
203: of the chromosphere. A redshift of 2--4 \kmps\ seems to be ubiquitous in the cool giants and
204: supergiants (Judge \& Carpenter 1998). On reflection, that the measured electron density is so
205: similar in all these stars, in spite of variations in mass-loss rate of several orders of
206: magnitude (see Harper 1996), is probably just as strange as the line shapes.
207:
208: \paragraph{Line Strengths:} Relative strengths of optically-thin lines give a measure
209: of the mass, temperature, and electron density of emitting gas averaged over the whole
210: chromosphere. They may be interpreted with appropriate empirical or semi-empirical models
211: as Judge (1986a, 1986b) did for several bright stars. This sort of analysis provides
212: global constraints on the integral
213: \begin{equation}
214: f_{\rm line} \sim \int {\nion}{\nelect} exp(-\chi_{\rm exc}/k{\Telect})dV. \\
215: \end{equation}
216: Fluxes for characteristic lines come from {\it IUE} spectra and seem to be roughly consistent
217: from star to star among the cool giants. Such collisionally excited lines as Si~II] UV0.01,
218: \Al2] UV1, and \Mg2\ $h$ and $k$ have consistent ratios for cool giants (Judge \& Jordan 1991).
219: \C2] UV0.01 is not necessarily proportional to the other collisionally excited lines
220: (Judge et al. 1992); however, it is roughly equal in strength to \Al2] for most of these
221: stars. For the values given by Judge \& Jordan, we can summarize these line strengths
222: as in Table 1. For the windy (non-coronal) giants, the flux in \C2] $\lambda$2325.4 seems to
223: be roughly equal to flux in \Al2] $\lambda$2669. We have also listed strengths for three
224: $\zeta$~Aur binaries as best we can judge them. The \Al2] fluxes for these binaries (Eaton 1992)
225: are relatively well determined. The \C2] fluxes, on the other hand, are very poorly measured
226: and in a noisy part of the {\it IUE} spectrum. There are no spectra for 31~Cyg exposed long
227: enough to detect this line, and it is hard to differentiate it from the noise in the spectra
228: for $\zeta$~Aur and 32~Cyg in which Schr\"oder et al.\ (1988) claimed to detect it, although
229: Harper et al.\ (2005) {\it may} have detected the strongest component with {\it HST}. I have
230: given what I consider the best measurements possible for these spectra, but in the following
231: analysis, I shall assume f($\lambda$2325.4) $\approx$ f($\lambda$2669) in all the windy giants.
232: The fluorescent \Fe1\ UV44 lines can be detected in all three classical $\zeta$~Aur systems
233: (e.g., Bauer \& Stencel 1989), but these two lines, especially $\lambda$2844, are highly
234: blended with other features.
235:
236: Judge (1986a, 1986b; Judge \& Jordan 1991) has made simple empirical models for three cool giants,
237: finding the emission is probably excited in a gas near 7000 K and an electron density of 10$^9$
238: (an assumed {\it global} value of \nelect\ from \C2] UV0.01). Carpenter et al.\ (1999) estimated
239: that the \Fe2\ lines observed in the wind of $\lambda$~Vel are scattered by gas at $\sim$ 6000 K
240: from the relative strengths of various multiplets. Other estimates of the electron temperatures
241: of single stars come from adjusting semi-empirical models (e.g., Kelch et al.\ 1978) to give
242: observed line profiles for charactistic emission lines and H$\alpha$ absorption.
243:
244: \paragraph{Fluorescent Lines:} The classical fluorescent lines of K giants are \Fe1\ UV44
245: excited by the \Mg2\ $k$ line. These lines are optically thick and must be formed in the deeper,
246: denser parts of the chromosphere for there to be any neutral iron to scatter them (Harper
247: 1990). Other fluorescent lines are scattered by molecular species (McMurray et al.\ 1999;
248: McMurray \& Jordan 2000). These molecular species must be present in some part of the chromosphere,
249: but the eclipses of $\zeta$~Aur binaries generally do not detect them, probably because there is
250: plenty of cold dense gas in the inner parts of the chromosphere to hide their absorptions in the
251: competing atomic features. McMurray et al.\ have argued that the strengths of fluorescent H$_2$
252: and CO lines, which they could not reproduce with their non-dimensional semi-empirical model of
253: $\alpha$~Tau, require a multi-component atmosphere, possibly with shocks deep in the atmosphere
254: to generate enough Ly$\alpha$ flux there to excite these molecules radiatively.
255:
256: \paragraph{More Highly Ionized Species:} Long exposures in the UV have detected lines of
257: highly ionized species even in the windy giants. These include intersystem lines of \Si3\ and
258: \C3\ (e.g., Carpenter et al.\ 1999) and the resonance doublet of \C4\ (e.g., Robinson et al.\
259: 1998). In the Sun, such highly ionized species emit lines in the transition region at $\sim$
260: 50,000~K. In the windy giants, this interpretation is problematic because there is no evidence
261: for the coronae that create the transition regions in the dwarfs.
262:
263: \paragraph{Expansion-Velocity Structure of the Wind:} This is very difficult to get from
264: observations of single stars. There is ample evidence of winds in the P-Cyg profiles of \Mg2\
265: and from the asymmetric, variable profiles of H$\alpha$ in cool supergiants (Zarro \& Rodgers
266: 1983; Mallik 1993). Harper (1996) summarized analyses of wind profiles derived from single
267: stars; Harper et al.\ (2005) have discussed this in somewhat more detail.
268: Much of the evidence comes from P-Cyg profiles in the UV. \Fe2\ lines of increasing
269: intinsic strength show increasingly negative velocities of their self reversals, attributable
270: to wind acceleration, in a number of cool supergiants, notably $\lambda$~Vel (Carpenter et al.\
271: 1999), $\gamma$~Cru (Carpenter et al.\ (1995), $\alpha$~Ori (Carpenter \& Robinson 1997), and
272: both $\alpha$~Tau and $\gamma$~Dra (Robinson et al.\ (1998). Of these, $\lambda$~Vel had the
273: most extensive coverage of the accelerating wind, with centers of the shell lines shifted
274: to $\sim$ $-$32 \kmps\ with respect to the star, and Carpenter and Robinson have interpreted
275: the data for these stars to infer mass-loss rates, terminal velocities, level of turbulence,
276: and density structure. Such analyses give winds accelerating much faster than those of the
277: $\zeta$~Aur binaries or, for that matter, of $single$ stars as deduced from radio observations
278: (Harper et al.\ 2001, 2005; Carpenter et al.\ 1999). In fact, the analysis of thermal radio emission
279: has become a fruitful technique for deducing the wind structure of single windy giants (e.g.,
280: Drake \& Linsky 1986; Harper et al.\ 2001).
281:
282:
283: \section{THE MODEL FOR 31~CYG}
284:
285: Although optical spectra provided many insights into the nature of the extended atmospheres
286: of the cool supergiant components of $\zeta$~Aur systems, the field was essentially dormant
287: from the mid 1950's until {\it IUE} revitalized it with panchromatic UV spectra. I have
288: discussed the results of such studies in a review at the Cool Stars 9 meeting (Eaton 1996).
289: I shall summarize them here as follows: First, the new UV observations recorded absorptions
290: from most of the important species expected to exist in chromospheres and winds of cool giant
291: stars. In contrast, optical spectra give very few of these species. Furthermore these UV
292: absorptions are often intrinsically strong lines that can be detected at great heights above
293: the stellar surface, giving us the ability to probe winds to much greater height. For the
294: first time we could use the wings of Ly$\alpha$ to measure hydrogen column densities directly
295: for many lines of sight through the wind/chromosphere. The many lines of \Fe2, likewise, gave
296: measurements of excitation temperature (T$_{\rm exc}$ $\sim$ 5,000--12,500 K) and kinematics
297: of the wind throughout much of the wind and upper chromosphere.
298:
299: Some of the results from {\it IUE} are conventional while others are surprising. Strengths and
300: shapes of lines from different ionization stages of metals, most importantly iron, confirm the
301: expectation that the metals are mostly {\it singly-ionized} throughout the chromospheres and winds
302: of these stars. Likewise, the detection of the wings of Ly$\alpha$ at height in chromospheres and
303: rough agreement of the mass column densities derived with those from \Fe2, means that {\it H
304: is primarily neutral} throughout those parts of the wind we can sample. This contradicts the
305: predictions of semi-empirical models in which H becomes completely ionized over the
306: first several scale heights of the chromosphere, giving roughly a constant electron density in the
307: line-emitting regions in spite of a marked decrease of total density (Judge 1990, p.\ 290; \S\
308: 3.2 below). This obsevation that H remains neutral thus places limits on permitted kinetic
309: temperatures and local electron densities in the gas. Furthermore, wind models for 31~Cyg give us
310: some direct insight into the turbulence in chromospheres. Single-component models of the gas with
311: no expansion effects, i.e., the sort of analyses used by Wilson \& Abt, find Doppler widths of the
312: order of 20 \kmps, decidedly supersonic for gas with kinetic temperatures below 10,000~K.
313: {\it IUE} observations show that at least some of this spread is caused by the differential
314: expansion of the gas along the line of sight and need not be attributed to local turbulence. In
315: fact, the {\it IUE} observations for 31~Cyg require a turbulence $\lesssim$ 15 \kmps.
316:
317: Even in the classical optical analyses, the ionization of metals was lower than expected for a
318: uniformly distributed gas and thereby implied clumping for the gas to achive enough electron
319: density to maintain an observable population of trace neutral species, such as \Fe1. UV
320: observations confirm this result. Ionization throughout the wind is lower than expected from
321: simplistic calculations of ionization equilibrium and implies clumping in the range 10--30$\times$
322: to achieve the inferred electron densities (n$_{e}$ $\sim$ 0.2--1.5 x 10$^9$ \cmcubed\ in the inner
323: 150 $\Rsun$ of the chromosphere). This is a complication well beyond most semi-empirical models
324: of chromospheres.
325:
326: We can use an idealized description of the measured physical properties of the gas
327: in the chromosphere and wind of 31~Cyg to test ideas about the structure of
328: chromospheres and, ultimately, about wind mechanisms. These measurements are based
329: on the most extensive set of observations ever obtained for a $\zeta$~Aur binary
330: (Eaton \& Bell 1994). I have chosen to concentrate on 31~Cyg over the years because
331: it has a much longer period than the other two classical $\zeta$~Aur binaries ($\zeta$~Aur
332: and 32~Cyg). This gives it a greater separation and less interaction between the wind
333: and B star, a big advantage in many analyses. The wind-scattered emission lines seen in
334: total eclipse, for instance, are weaker than in the other two stars (Eaton 1992). I am
335: adopting the following properties for the 31~Cyg system: $D$ = 473 pc; $R_{\rm K}$ = 197
336: $\Rsun$, $M_{\rm K}$ = 11.7 $\Msun$, $M_{\rm B}$ = 7.1 $\Msun$, $v_\infty$ = 90 \kmps,
337: $\Mdot$\ = 3.0$\times$10$^{-8}$ \MLrate\ (Eaton 1993c; Eaton \& Bell 1994). We should
338: note that, in observing the more complicated $\zeta$~Aur, Baade et al.\ (1996) probably
339: had a better case than 31~Cyg for applying their wind-scattering analysis. Furthermore,
340: inasmuch as the models for 31~Cyg and $\zeta$~Aur give essentially the same results in
341: terms of velocity structure/extension of the wind and wind temperatures, we can be
342: confident that the results for either of them should be readily applicable to other stars.
343:
344: Table 2 gives the details of the empirical chromospheric/wind structure I am using
345: to test wind models. Quantities listed are (1) radius (distance from center of the
346: star in $\Rsun$), (2) expansion velocity, (3) excitation temperature, (4) a temperature
347: to drive the wind thermally, (5) a clumping factor, CF, giving the inverse of the fraction
348: of space actually filled with matter, (6) log($n_{\rm H}$), the total hydrogen density,
349: (7) log(\nelect), the local electron density from an assumed constant ionization of H
350: (3\%) and amount of clumping, (8) log(\nelect) for the variable ionization developed in
351: \S\ 3.2 and with the assumed clumping, but limited to 10\% in the outer chromosphere,
352: and (9) $v_{\rm equ}$, a velocity derived from equipartition between mass motion and internal
353: thermal energy of the gas (see \S\ 5.2).
354:
355: Although the electron densities measured from photoionization balance imply significant
356: clumping of the gas, they are rather crude and may be systematically wrong. An independent
357: way to estimate this degree of clumping is to look at the difference between the
358: pressures\footnote{Strictly speaking, the wind is driven through the pressure {\it gradient},
359: which appears in the hydrodynamical momentum equation (\ensuremath{F = ma}), but a given
360: pressure distribution, the higher level abstraction I am specifying, implies a pressure
361: gradient, even in the simplistic case of an isothermal wind.} required to drive the observed
362: wind and the gas pressures that would be available in the wind if it were not clumped. This
363: approach works because the gas pressures ($\sim$ $\rho$T$_{\rm gas}$/$\mu$) would be in
364: equilibrium with whatever pressure drives the wind. We do this by determining a temperature
365: structure to drive the observed wind {\it thermally}, as though these stars had a coronal,
366: i.e., generalized Parker-type, wind (see Lamers \& Cassinelli 1999, Chapters 4\&5), then
367: comparing those temperatures to the observed temperatures, point by point, assuming the
368: excitation temperatures we have measured approximate the electron temperature. This last
369: assumption seems reasonable because the excitation temperatures are similar to temperatures
370: derived for semi-empirical chromospheric models of single stars, at least in the inner parts
371: of the wind, but we will test it in later sections. To reiterate, if the gas is clumped,
372: thereby bifurcating into dense clumps and an unspecified interclump medium, there must be a
373: pressure in the interclump medium maintaining the clumping. We will assume that this is the
374: pressure driving the wind and that it somehow breaks the wind up into microscopic clumps and
375: maintains them. This is fundamentally different from the approach of traditional wave models
376: in which the gas is uniform and the waves act on it continuously.
377:
378: The temperature structure to produce the pressure gradient required to drive the wind of 31~Cyg
379: thermally, which we have derived from a thermal-wind model, is given roughly by the equations
380: \begin{equation}
381: T_{\rm therm} = 20,000 K + 75,000 K (r - \Ro)/\Ro\ \hskip30pt for \ \ (r-\Ro) < 150 \Rsun, \eqnum{2a} \\
382: \end{equation}
383: and \\
384: \begin{equation}
385: T_{\rm therm} = 95,000 K - 75,000 K (r - 3\Ro)/(15\Ro) \hskip30pt for \ \ (r-\Ro) > 2\Ro, \eqnum{2b} \\
386: \end{equation}
387: \setcounter{equation}{2}
388: where $T_{\rm therm}$ is thermal (electron) temperature for a uniformly distributed gas,
389: r is the radius, and $\Ro$=197 $\Rsun$\ is the photospheric radius of the star.
390: I don't claim that this structure is a rigorous determination of the thermal-pressure profile,
391: but that it is adequate for giving a good idea of the pressures required throughout the
392: wind. Figure 1 shows the velocity structure calculated for this adopted temperature
393: structure of Eq.\ 2; Figure 2, the amount of clumping implied.
394:
395: A way to test this sort of model is to calculate the emission expected from it. I have
396: done this in two ways, first, with a spherically symmetrical model that calculates the line
397: strengths, line profiles, and chromospheric mass column densities specifically for 31~Cyg.
398: The second way uses a traditional plane-parallel model for $\zeta$~Aur I developed in
399: 1992--1995 with PANDORA (Vernazza et al.\ 1973, Avrett \& Loeser 1992) to explore the
400: ionization structure and effect of clumping in these stars.
401:
402: \subsection{Tests of a Spherically Symmetrical Model for 31~Cyg}
403:
404: The spherical model for 31~Cyg has the distribution of gas given in Table 2. The velocity
405: structure and distribution of mass density (given as n$_{\rm H}$=$\rho$/1.4m$_{\rm H}$) come
406: from fitting spectra of shell lines in atmospheric-eclipse spectra; they should be fairly reliable.
407: Excitation temperature comes from fitting the excitation of \Fe2\ in these atmospheric-eclipse
408: spectra; they should be reliably measured but subject to uncertainty about their meaning.
409: Electron densities depend on level of ionization of H and degree of clumping. We can
410: make educated speculations about these properties as follows. Since H seems
411: to be neutral observationally, we might expect the ionization to be $\lesssim$ 10\%.
412: We will assume it constant with height, following the thoughts of Judge (1990, \S\ IIIb),
413: and of the order of 1--5\%, and adopt a value of 3\% for the sake of a first-order model.
414: In that case, the gas must be clumped even to approach the canonical \C2] electron density
415: almost anywhere in the chromosphere.
416:
417: \subsubsection{Line Emission}
418:
419: For optically-thin lines, we can calculate the emission with the standard physics given
420: by Osterbrock (1974), for example, with assumptions about ionization and chemical
421: abundances. In this approximation, emissivity [ergs~cm$^{-3}$s$^{-1}$] is just
422: \begin{equation}
423: \epsilon_{\nu} = 8.63\times10^{-6} \Omega{_{1,2}} n{_i} n{_e} exp(-\chi/kT) T^{-0.5} \omega{_1}^{-1} \chi \\
424: \end{equation}
425: where n${_i}$ is the density of emitting ions, $\Omega$ is the collision strength, $\chi$ is
426: the excitation energy of the upper level, and $\omega{_1}$ is the statistical weight of the
427: lower level. In this approximation, emission is proportional to collision strength, and we
428: have incorporated collision strengths for \C2] $\lambda$2325 ($\Omega$ = 0.830 and 1.66, for
429: transitions out of the two ground-state levels, with half the emission going into the dominant
430: 2325.4\AA\ line) from Blum \& Pradhan (1992) and for \Al2] $\lambda$2669 ($\Omega$ = 3.3)
431: from Tayal et al.\ (1984). Abundances of C and Al were $A_{\rm C}/A_{\rm H}$ = 2.0
432: $\times$ 10$^{-4}$ and $A_{\rm Al}/A_{\rm H}$ = 3.0 $\times$ 10$^{-6}$ (Eaton \& Bell 1994,
433: \S\ 2.1). To determine the total emission in an optically thin line as a function of radius
434: on the sky, we sum the emissivity along a ray through the atmosphere, with integration limits
435: determined by whether or not it intersects the star. To get a profile for the line, we define
436: a velocity scale and map the emissivity profile of the gas at each point in the atmosphere onto
437: it, with allowance for the atmospheric expansion, $v_{\rm exp}$, an arbitrary (systematic)
438: velocity along the line from the center of the star, $v_{\rm syst}$, and a spectrum of Gaussian
439: turbulence, $v_0$.
440:
441: The first thing we note from these models is the effect of the extended spherical atmosphere
442: on the emission strength and line profiles. Optically thin lines formed in such a structure
443: will be highly {\it limb brightened} and will combine contributions from the face of the star
444: and from a halo beyond its limb in similar proportions. Figure 3a illustrates this effect for
445: our 31~Cyg model for an emitter assumed to exist in the same ionization stage throughout the
446: wind (\nelect\ as in Table 2, Column 7). The emission peaks somewhat more to the limb than mass
447: because electron density drops rapidly with height in these atmospheres, even with the assumed
448: clumping. An emitter that is depleted in the inner chromosphere, as \C2] may be, would peak even
449: further beyond the limb. This is shown in the figure as a dashed line calculated for emission
450: only at temperatures in the model above 5200~K. Even so, most of the emission would come from
451: within the first $\lesssim$ 0.15~$\Ro$ of the chromosphere/wind. Strong resonance lines, on
452: the other hand, should be {\it limb darkened}. Scattering in an extended atmosphere would give
453: a highly forward-peaked source function (e.g., Cassinelli \& Hummer 1971), the reason for the
454: core-halo profile of the discs of WR stars, and, for resonance lines in chromospheres, the
455: large optical depths mean the escape probability for photons migrating through the damping
456: wings would be much smaller for the radial direction than for other lines of sight.
457:
458: Simple calculations like those in Figure 3a cannot represent the conditions in actual chromospheres
459: because they give {\it fluxes} much higher than observed. Table 1 gives fluxes calculated for three
460: cases, (1) uniform 3\% ionization of H, (2) variable ionization of H, limited to 10\%, and
461: (3) variable ionization of H, limited to 3\%. The model with uniform H ionization
462: arbitrarily set at 3\% gives an \Al2] flux too high by a factor of ten and a \C2] flux at least as
463: bad. The model with variable ionization (Table 2, Column 8), on the other hand, gives \Al2] flux
464: high by $\sim$50\% and \C2] high by slightly more, but the high electron density in the outer
465: winds of these models, combined with single ionization of metals, gives much larger line broadening
466: from differential expansion of the atmosphere. These models also have problems with electron density.
467: If we average \nelect\ over emission of \C2], the values are either too high, as for the calculation for
468: the unrealistic uniform 3\% ionization of H, or somewhat low, as for the realistic ionization of
469: H limited to 3\% or even 10\%. Figure 3b shows the intensity profiles for the model with variable
470: ionization. The effect of the 10\% ionization in the wind is obvious, with much of the emission in
471: the wind coming from the hot ionized outer parts. These parts, however, have electron densities much
472: too low (7.7 in the log) to give the \C2] line ratios, at least the ones seen in single stars, and
473: they produce \C2] profiles much broader than observed. They are clearly inappropriate for single
474: stars unless C becomes {\it doubly ionized} in the outer wind or unless the temperature in the outer
475: wind is well below the measured excitation temperatures.
476:
477:
478: \subsubsection{Radio Continuum Emission}
479:
480: A further test of these spherical models comes from the radio continuum observations of Drake
481: \& Linsky (1986) and Drake et al.\ (1987), who detected a number of bright cool supergiants
482: at the 0.1 mJy level (or 10$^{-27}$ ergs~cm$^{-2}$s$^{-1}$Hz$^{-1}$). Their measurement for
483: 31~Cyg (0.36 $\pm$ 0.07 mJy) gives a check on the consistency of the electron densities and
484: temperatures we are assuming. The basic idea (see Harper et al.\ 2005) is to integrate the
485: radio source function, $S_\nu$ = $B_\nu$, along rays through the wind and then sum the resulting
486: intensities over the stellar disc to get a kind of luminosity [ergs~s$^{-1}$~ster$^{-1}$Hz$^{-1}$].
487: This quantity would then be converted to flux at the Earth by multiplying it by the solid angle
488: of 1~cm$^2$ at the star's distance, $D$. The dominant free-free opacity from electron-proton
489: interactions, corrected for stimulated emission, is just
490: \begin{equation}
491: \kappa_{\nu} = 0.212 n{_p} n{_e} T_{e}^{-1.35} \nu^{-2.10} \hskip20pt \\
492: \end{equation}
493: (Harper et al. 2005, Eqn.\ 3), while the generally smaller contribution from neutral-hydrogen--electron
494: interactions (H$^-$) is roughly
495: \begin{equation}
496: \kappa_{\nu} = 6.2\times10^{-35} n{_H} n{_e} (\lambda/1 cm){^2} T_{e} \hskip20pt \\
497: \end{equation}
498: (Allen 1973, \S\ 42 evaluated at 8000 K). For the values of \Telect,CF,\nelect\ in Table 2
499: (Column 8 for \nelect), we calculate a flux at the Earth of 0.04 mJy at 6.2 cm, only about 10\%
500: of the amount measured by Drake et al. Figure 4 shows the calculated intensity profile, a slightly
501: limb-brightened disc with a radius of $\sim$ 3.8 $\Ro$. This failure to reproduce the
502: flux seems to be a common problem in reconciling the radio emission of cool giants with their
503: UV spectra. Carpenter et al.\ (1999), for instance, were unable to reproduce the 3.5-cm
504: flux of $\lambda$~Vel with the rapid wind acceleration they found from UV lines. Harper et al.\
505: (2005), likewise, were unable to reproduce the radio spectrum of $\zeta$~Aur with their model and
506: found it hard even to get the flux level.
507:
508: To get emission as great as is observed for 31~Cyg, the outer wind must be essentially wholly
509: ionized, as the large mass-loss rate of Drake et al.\ implies. If we let the outer wind become
510: mostly ionized, we can at least approach the flux level observed. For instance, with H in the
511: envelope 75\% ionized above 8000~K, we get a flux of 0.22 mJy; for 100\% ionization, 0.28 mJy.
512: Is this level of ionization reasonable? It seems consistent with observations of $\zeta$~Aur
513: stars, especially 31~Cyg, but the evidence for single stars is ambiguous. Such high ionization
514: should reveal itself through absorptions of {\it doubly} ionized species in spectra of $\zeta$~Aur
515: binaries and possibly through emission lines from these species in both the binaries and single
516: supergiants. The broadening of these lines would be a measurement of the terminal velocity and
517: turbulence of the outer wind. Let us look at the observations of these species in some actual
518: stars. There is a weak emission feature at the position of \Si3] $\lambda$1892 in the eclipse
519: spectra of 31~Cyg ({\it IUE} images SWP47335 and SWP47336). In addition, strong emission lines
520: of higher ionization are seen in eclipse in all three classical $\zeta$~Aur binaries. Multiplets
521: \Al3\ UV1 and \Fe3\ UV34 in 31~Cyg have P-Cyg profiles (Bauer \& Stencel 1989, Fig.\ 3) and are so
522: strong they must be formed by scattering of light from the B star in the wind. The line widths
523: at the bases of their profiles are roughly 170 \kmps, about what one would expect for a wind at
524: terminal velocity. Eaton \& Bell found that these absorptions are formed primarily at velocities
525: {\it toward the B star} and likely result from ionization of the outer wind by the B star, not by
526: intrinsic radiation of the supergiant. Thus a large fraction of the outer wind is ionized
527: {\it in the 31~Cyg system}. Since H requires less energy to ionize than the 19~eV to ionize
528: C$^+$ out of its metastable level, H is likely ionized, as well, in these regions of 31~Cyg.
529:
530: Single stars may have lower ionization in their outer winds than the binaries with similar mass-loss
531: rates, although there is evidence in P-Cyg profiles of wind lines that some of the metals are doubly
532: ionized in many of them (see \S 4.2 below). Carpenter et al.\ (1999) detected lines of \C3] and
533: \Si3] in $\lambda$~Vel with about the right broadening and strength to be formed in an ionized wind.
534: However, they interpreted these lines as an indication of gas at $\sim$ 50,000~K, as though high
535: ionization necessarily means high temperature. The problem with wind formation is that the computed
536: line profiles are wrong. The calculated profiles are essentially square, having a central dip
537: reflecting the absence of gas at low expansion velocities, while the observed profiles, although
538: somewhat noisy, seem to have a central peak as though formed at least partly in turbulent gas at
539: low expansion velocity. Therefore the {\it emission} lines of single stars unfortunately tell us
540: nothing about the ionization of their outer winds.
541:
542: \subsection{Calculations with PANDORA}
543:
544: PANDORA is useful for exploring non-traditional models of chromospheres because it lets one
545: specify an arbitrary distribution of turbulence to increase the scale height in hydrostatic
546: equilibrium and to specify an arbitrary distribution of added electron density that can be used
547: to simulate clumping. I have made exploratory calculations for three cases: (1) a chromosphere
548: for $\epsilon$~Gem (G8~Ib) based on the model of Basri et al.\ (1981), (2) a generalized model
549: of a $\zeta$~Aur binary ($\Ro$=150 $\Rsun$) with temperatures and scale heights based on my analyses
550: of $\zeta$~Aur (Eaton 1993a) and 32~Cyg (Eaton 1993b), and (3) models for $\alpha$~Tau to explore
551: excitation of H$\alpha$ in cool giants (Eaton 1995). In the paper about H$\alpha$, I explored
552: the conditions necessary for exciting H$\alpha$ and \C2] in semi-empirical hydrostatic models,
553: finding that a certain amount of clumping of the warmer gas was required to strengthen H$\alpha$
554: and to have electron densities high enough to give reasonable \C2] line ratios.
555:
556: These semi-empirical models can give us some insight into how the gas might be ionized at
557: various depths in actual chromospheres. All the models, regardless of assumptions about
558: scale height, clumping, and temperature, predict species such as Al will be singly ionized
559: throughout the whole emitting atmosphere. For H and C, the calculated ionization increases
560: with height. Figure 5 shows this effect for H. In it, I have
561: parameterized height with the electron temperature, since, by assumption, temperature
562: increases monotonically with height above a temperature minimum in all such semi-empirical
563: models. The calculations show that log(n$_{\rm e}$/n$_{\rm H}$) increases from a minimum
564: of $-$4.0 at $\sim$ 2600~K, determined by ionization of the metals included in the calculation,
565: to 0.0 (complete ionization of H) at $\sim$ 9400~K. For C, complete ionization occurs
566: near 5000~K in these models. The actual models have scale heights somewhat smaller than
567: that of 31~Cyg, so the level of ionization could be higher in 31~Cyg.
568:
569: \subsection{A Further Question about the 31~Cyg Model}
570:
571:
572: One possible error in Eaton \& Bell's analysis of 31~Cyg is the determination of the temperature
573: from excitation of \Fe2. We have assumed that the excitation temperatures measured are {\it bona
574: fide} electron temperatures of the gas. This might be the case, in that most of these lines arise
575: from excited metastable levels which would probably be in thermal equilibrium with the electrons,
576: at least in the denser parts of the wind (e.g., Judge et al.\ 1992). Harper et al.\ (2005) argued
577: this point explicitly, while others deriving temperatures for $\zeta$~Aur binaries (such as Eaton
578: \& Bell) have implicitly assumed it. However, the excitation temperatures in the outer parts of
579: the winds may well be greater than the thermal temperatures, since these regions are bathed in the
580: radiation of a B star, and since the electron densities expected in these zones are lower than the
581: critical density ($\sim$ 10$^6$ per Judge et al.\ 1992) for radiative processes to become important.
582: In the model for 31~Cyg, these temperatures have very little effect on the emission at any wavelength,
583: since the densities are very low.
584:
585: Contrariwise, all of these stars have rather high excitation temperatures in their outer
586: chromospheres, regardless of the effective temperature of the B companion. We see this in Figure 6
587: which gives \Texc\ as a function of tangential mass column density through the chromosphere. We
588: see that the temperature rises to about 8500~K where the Balmer lines become optically thin. This
589: agrees roughly with theoretical calculations for supergiants.
590:
591:
592: \section{IMPLICATIONS FOR {\it SINGLE} STARS}
593:
594: Here I shall attempt to fit the rich
595: lore of space-dimensionless analyses of line emission and absorption of single stars
596: into the context of our knowledge of the legitimately one- to two-dimensional knowledge
597: of $\zeta$~Aur binaries. We might expect the wind structure of the binaries to be
598: essentially the same as that of single stars because the strengths of intrinsic
599: emissions formed in the wind and chromosphere of $\zeta$~Aur binaries are similar to
600: those of single stars (Schr\"oder et al.\ 1988; Eaton 1992; Harper et al.\ 2005, Fig.\ 6),
601: and the H$\alpha$ profiles, and their variation, seem to be the same as well (Eaton 1995;
602: Eaton \& Henry 1996).
603:
604: \subsection{The Reality of Semi-Empirical Models}
605:
606: Most detailed analyses of cool stars' chromospheres are based on semi-empirical models in which
607: one posits a temperature rise through the outer atmosphere, calculates a density structure from
608: hydrostatic equilibrium and radiative transfer, and then calculates the emergent spectrum from the
609: physical conditions derived (e.g., Kelch et al.\ 1978; Basri et al.\ 1981; McMurray 1999). These
610: models are inspired by the solar chromosphere, in which there is a temperature rise from a minimum,
611: determined by the location where non-radiative heating begins to dominate heating by photospheric
612: radiation, to a point where the chromosphere merges with a transition region of rapidly increasing
613: temperature, heated by downward conduction from a corona. The increase in temperature with height
614: follows naturally from observations of the Sun and from the physical consideration that temperature
615: should rise with decreasing density to keep emissivity from falling precipitately with height per
616: Equation (1). Of course, there might be other ways to organize a chromosphere. For example, the
617: material at different temperatures might be more intimately mixed throughout the chromosphere, as
618: McMurry et al.\ (1999) contemplate to some degree. Harper et al.\ (2001) likewise have speculated
619: about a truely radical reinterpretation of $\alpha$~Ori in which the chromospheric line emission
620: comes entirely from hot inclusions in a generally cool neutral wind, although Harper et al.\ (2005)
621: did not attempt to apply this radical approach to $\zeta$~Aur. Indeed, temperature measurements
622: for $\zeta$~Aur binaries, at least to first order, confirm some sort of general temperature rise
623: with height. Figure 6 shows the relation between excitation temperature and tangential mass column
624: density through the atmospheres for four stars. The temperature rise is obvious. While this
625: evidence does not preclude mixing some hot gas with the bulk of the warm gas throughout the
626: chromosphere, it shows that the bulk of the gas behaves sort of like the gas in these classical
627: semi-empirical models.
628:
629: Another fundamental property of semi-empirical models is the density structure. In the calculated
630: models, it usually comes from hydrostatic equilibrium between gravity and thermal motions of the
631: emitting gas ($T$ $<$ 10,000~K). This is problematic in that there are obviously other sources of
632: momentum flux in a typical chromosphere, such as the turbulence we see in profiles of emission lines,
633: which will extend the atmosphere, changing its mass and electron density structure. Furthermore, the
634: mere existence of turbulence implies some sort of clumping of the gas, which would necessarily change
635: the local electron densities and ionization structure. Moreover, such effects would change the
636: transfer of radiation and escape of photons from the chromosphere.
637:
638: A third property of many semi-empirical models is a precipitate temperature rise to coronal
639: temperatures defining the top of the chromosphere. This is not a necessary feature, especially
640: in the windy giants. Nevertheless, McMurry (1999) used such a rise for $\alpha$~Tau to fit the
641: Ly$\alpha$ profile and calculate emission from highly ionized species. However, the observations
642: of $\zeta$~Aur binaries do not require such a rise and may not even allow it; the profiles of \C3]
643: and \Si3] seem to require formation at least partly in the inner chromosphere/wind, and the
644: arguments of McMurry et al.\ (1999, 2000) suggest there are other places to excite \C4.
645:
646: \subsection{Terminal Velocities of the Winds}
647:
648: One of the basic tenets of our understanding of the windy giants is the idea that their terminal
649: velocities are much lower than the surface escape velocity (e.g., Hartmann \& McGregor 1980;
650: Judge 1992; Harper 1996). These terminal velocities are probably not as well known as we think
651: they are, and they may well be {\it much} higher than generally thought, especially in the normal
652: K giants like $\alpha$~Tau, as Judge (1992) argued in a provocative paper about wind energetics.
653: This is hardly a new idea (see, e.g., Judge 1992; Ahmad \& Stencel 1988; Kuin \& Ahmad 1989), but
654: it is certainly worth discussing further in the context of supposed differences between single stars
655: and binary components. For the classical $\zeta$~Aur binaries, we base the terminal velocity on
656: measurements of shell lines seen at all phases (e.g., Eaton 1993c). We should be suspicious of low
657: terminal velocities since the metals in the outer parts of the winds might well be doubly ionized,
658: as they are in calculations of Harper et al.\ (2005, Fig.\ 5) for $\zeta$~Aur and \S\ 3.2 above
659: for single stars. Furthermore, the recombination time for H, given the electron densities in
660: the outer parts of our model ($\sim$ 10$^{3-5}$ \cmcubed), is of the order of 1--100 yr.
661: Semi-empirical models for single stars also become fully ionized, at least in H, in these
662: regions, if that actually means anything. So we really don't know what these winds are doing in a
663: vast volume of space before they recombine and form the shell lines. Occasionally more of the
664: velocity structure may reveal itself through abnormally dense winds, as in AL~Vel in 1992 (Eaton
665: 1994, Fig.\ 8) and in $\lambda$~Vel in 1990 (Mullan et al.\ 1998). Furthermore, we should realize
666: that the winds must be sweeping up interstellar gas in these shells (cf. Wareing, Zijlstra, \&
667: O'Brien 2007). This means that the shell velocities are, if anything, likely to be {\it lower}
668: than the true terminal velocities.
669:
670: Carpenter et al.\ (1999) admit that we probably do not see much of the velocity structure in the
671: traditional shell lines in many stars, while they contend that they have seen it all in $\lambda$~Vel.
672: However, it is not clear from the line profiles of \Mg2, \Oxy1\ UV2, and \Fe2\ UV1 that one sees it
673: even in that case. The velocity structure for $\lambda$~Vel in 1994 seems to be ionization-bound,
674: with the maximum wind velocity sampled limited by ionization of Fe$^+$ to Fe$^{+2}$. In an
675: ionization-bound wind, the relatively high density behind the ionization front would allow somewhat
676: weaker lines to become thick at their normal rate, while the strongest lines would quickly saturate
677: at the velocity of the ionization front, little more than the velocity of those somewhat weaker
678: lines, to give the sort of levelling off seen in the highest measured velocities. The edge velocities
679: for all these strong lines in 1994 are around 50 \kmps, much greater than the supposed terminal
680: velocity of 33 \kmps. Furthermore, in 1990 the star showed absorptions to at least 80 \kmps\
681: (Mullan et al.\ 1998), which places its terminal velocity close to what we think 31~Cyg has, if the
682: increase really did come from lower ionization in the outer wind. Also, in comparing $\alpha$~Tau and
683: $\gamma$~Dra, Robinson et al.\ (1998) found terminal velocities of 30 and 67 \kmps, respectively,
684: for stars that otherwise seem to have similar atmospheric structure. All this evidence suggests
685: terminal velocities several times as large as they often seem, with the {\it apparent} terminal
686: velocity dependent on just how much material is being ejected into the flow at a given time.
687:
688: There is another problem in interpreting these shell lines (cf.\ edge velocities of $\lambda$~Vel)
689: that would lead to errors in the terminal velocity. Carpenter et al.\ have interpreted the
690: widths of Gaussians fit to the shell lines as a measure of {\it turbulence}. This could conceivably
691: be the case for \Fe2, but it is clearly inappropriate for at least some lines, for instance, for \Mg2\
692: in $\gamma$~Cru (Carpenter et al.\ 1995; Fig.\ 10), where \Mg2\ seems to have a significantly lower
693: expansion velocity than even moderate \Fe2\ lines. Such corrections for wind turbulence seem wrong,
694: from both observational and theoretical considerations. Observationally, in models for scattering
695: in shells (e.g., Baade et al.\ 1996), the turbulence drops with height. Theoretically, one would
696: expect the turbulent energy to go into accelerating or heating the wind and be essentially damped
697: out by the time the wind reaches its terminal velocity.
698:
699:
700: \subsection{Turbulence of the Chromospheric Gas}
701:
702: To investigate the nature of the turbulence, we may calculate the profiles of broadening for
703: various velocity distributions for the model of 31~Cyg. The best fits seem to be for {\it isotropic}
704: turbulence. In Figure 7, the dashed line shows the profile for isotropic turbulence giving roughly
705: the line broadening of single supergiants. A Gaussian fits it quite well, as expected for the
706: assumptions in the calculation. The solid curve, on the other hand, shows the effect of imposing
707: a $v_{\rm syst}$=10~\kmps\ downward velocity on this same turbulence. The profile is still Gaussian
708: at the level of the plot, but it is shifted to the red (3.8 \kmps) and broadened slightly more
709: ($\Delta$FWHM = 1.0 \kmps) by the variable projection of the systematic velocity into the line of sight.
710:
711: Calculations for {\it anisotropic} turbulence all give non-Gaussian profiles to some extent.
712: Some of these are shown in Figure 8. Radial turbulence is especially bad in this regard
713: (Figs.\ 8a,c) because the strongest emitting regions are near the stellar limb, where the
714: turbulent motions would be mostly perpendicular to the line of sight. Tangential turbulence is
715: considerably better but still decidedly non-Gaussian (Figs.\ 8b,d), with enhanced wings not seen
716: in profiles of actual stars (see, e.g., Carpenter et al.\ 1991, Fig.\ 3b). Figures 8c and 8d,
717: however, show that {\it elliptical} anisotropic turbulence, in which one component is much larger
718: than the other, is considerably closer to the observed Gaussian shape, and the departures from
719: the observed Gaussian profiles would be somewhat reduced by convolving the calculated profiles
720: with a Gaussian point-spread function for a spectrograph, which I have not done. However,
721: these profiles would still have overly broad wings not seen in actual stars. I chose the nature
722: of elliptical turbulence to reflect an isotropic turbulence of 5 \kmps\ from perhaps thermal
723: motion and 20 \kmps\ of either radial or tangential macro/micro-turbulent motions.
724:
725: Carpenter \& Robinson (1997) and Robinson et al.\ (1998) have cleverly fit the broadened wings of
726: \C2] lines with an anisotropic turbulence somehow averaged over a uniform disc, but that model is
727: clearly inappropriate in light of the large limb effects predicted by $\zeta$~Aur binaries (Figure 3).
728: In fact, a calculation with our model, but limited to gas within the limb of the star, gives the awful
729: non-Gaussian profiles seen in Figure 8e.
730:
731: Given the distribution of mass in the model and the resulting limb brightening, the Gaussian
732: profiles of intrinsic lines imply a fairly {\it isotropic} turbulence. They certainly do not
733: allow mostly {\it radial} turbulence, and they make it unlikely that the turbulence is
734: perponderantly tangential.
735:
736:
737: \subsection{Electron Density and Interpretation of the \C2] Multiplet}
738:
739: We have hardly any actual knowledge of the electron densities in the winds/chromospheres of any
740: of these stars. What we do know is how dense the regions emitting the bulk of \C2] are.
741: Everything else must somehow come from a {\it model} calculation.
742:
743: One of the triumphs of applying physics to cool stars is the use of \C2] to derive precise
744: electron densities for a large number of objects. The kinematic analysis of Judge (1994),
745: however, undermines the importance of this result. The redshift he detects in most stars
746: is difficult to reconcile with other aspects of UV analyses, since it requires a
747: $\sim$ 10~\kmps\ downdraft of all the emitting gas or something more extreme if only part
748: of the gas is participating. It would be convenient if this redshift were the result of a
749: systematic error in the energy levels, and this may ultimately prove to be the cause of it.
750: The energies of the upper levels of \C2] are quoted by Moore (1970) to the nearest 0.1 cm$^{-1}$,
751: about 1 \kmps, while the wavelengths are somehow known to one more significant figure. Likewise,
752: the enhanced electron densities derived for the red side of these profiles seem to manifest
753: themselves in only one of the line ratios, as though it might be affected by an unrecognized blend,
754: but the dominant 2325.4\AA\ line would have to be the one affected.
755:
756: However, if we accept the reality of the shift and enhancement, \C2] must be telling us something
757: profound about wind acceleration. C is two orders of magnitude more plentiful than Al, yet lines
758: of the two are roughly the same strength in cool giants. Very little of the C, therefore, may be
759: emitting \C2], and it must be produced under rather special conditions to give the observed
760: multiplet ratios of single stars. The broad wings of \C2] can be fit by emission in the
761: expanding gas of the upper chromosphere/lower wind with the broadening produced by differential
762: projection of expansion into the line of sight. However, for that approach to work, the
763: inner regions of the chromosphere must be neutral in C, even more so than implied by ionization
764: balance in semi-empirical models such as those of \S\ 3.2 or measured in 31~Cyg. Figure 8f shows
765: such a calculation in which the gas below 5200~K emits no \C2]. This calculation gives a reasonable
766: flux, but it does not fit any of the other properties of \C2]. The average electron density in the
767: emitting gas (5$\times$10$^{7}$) is nowhere near the canonical 10$^{9}$, {\it even with the clumping}.
768: The width of the profile is somewhat more than expected for a real star (FWHM = 35 vs.\ 30 \kmps\
769: for a weak line like \Al2]), the peak is shifted blueward by $-$0.8 \kmps, and the wings are probably
770: too weak relative to the core. Clearly a more radical departure from the expected ionization/velocity
771: structure is required.
772:
773: Looking at this more radically, we can get an idea of the conditions required {\it in the 31~Cyg
774: model} by extending our assumptions about ionization. Given the velocity structure, observationally
775: all the C would be inert below 7000--8000~K to weight the emission at significant expansion velocity
776: enough to give the extended wings (giving the same sort of effect as the anisotropic turbulence
777: dismissed in \S\ 4.3), with the bulk of emission coming from locally dense, probably rather cool
778: regions at depth that may have peculiar velocities. Again, this means that C would be {\it much
779: less ionized} in these atmospheres than predicted by semi-empirical models, with the \C2] emission
780: coming from only a moderately small fraction of the gas. However, the similarity of Doppler widths
781: of \C2] and other optically-thin lines such as \Si2] and \Al2] suggests that both are formed in gas
782: with similar turbulence, perhaps in similar parts of the atmosphere. A possible source of the redshift
783: of the line core is preferential excitation in the downward-facing parts of downward-moving elements
784: in the turbulence distribution. This fraction of the gas would have the greatest relative speed
785: with respect to the outward momentum source and might well be expected to be subjected to the most
786: violent accelerations. Figures 9 show two experiments in simulating this effect. In Figure 9a the
787: emission comes from two sources, all the gas above 8500~K emitting with the velocity structure of
788: Table 2 and 30~\% of the gas at temperatures below 6000~K with a 10~\kmps\ downdraft. Figure 9b
789: shows the emission for a model in which the gas above 8000~K is emitting with no downdraft and only
790: the cooler gas with log(\nelect) $\geq$\ 8.6 is emitting with a 10~\kmps\ downdraft. Both
791: calculations give reasonable fluxes and electron densities averaged over the emitting gas (3.5 and
792: 5.5 $\times$10$^{8}$, respectively) high enough to begin to approximate the observed global values.
793: Thus calculations with the most realistic chromospheric model we can muster are at least
794: {\it consistent} with Judge's suggestions about \C2].
795:
796: This analysis has necessarily been speculative, although it is much better constrained by actual
797: observations than any alternative analysis, e.g., one based on a traditional semi-empirical
798: model, could have been. I think it is fair to conclude that (1) the broad wings of \C2] likely
799: come from formation in the lower wind (the broadening coming from differential expansion) and
800: (2) C may not be ionized in the way expected from the standard non-LTE calculations of
801: semi-empirical atmospheres but may be subject to some other source of ionization associated with
802: the driving mechanism that picks out only part of the radial component of the turbulence
803: distribution. In all these calculations the wings of the \C2] line are broadened by formation
804: in gas with significant expansion velocity projected into the line of sight. The asymmetry
805: favoring the blue wing over the red results from having some of the redshifted gas blocked by
806: the star. This effect is seen in actual stars, e.g., $\alpha$~Tau (Carpenter et al.\ 1991,
807: Fig.\ 3) and $\gamma$~Cru (Carpenter et al.\ 1995, Fig.\ 2). Furthermore, as a caveat, we should
808: note (1) that C would be depleated by the first dredgeup (e.g., Luck \& Lambert 1985), although
809: this effect is presumably included in my adopted C abundance, and (2) that the observations of
810: $\zeta$~Aur binaries do not seem to show mostly neutral C at depth. One should also keep in
811: mind that \C2] may eventually destroy a basic assumption of this paper, and practically
812: every other one written about chromospheres and winds of cool giants, that the chromosphere
813: and wind are one continuous structure spread evenly over the face of the star.
814:
815: \subsection{More Highly Ionized Species}
816:
817: The emissions of highly ionized species such as \C3] and \C4\ do not fit readily into
818: our model for 31~Cyg. The level of ionization required is incompatable with the
819: temperatures and densities measured in the outer winds of $\zeta$~Aur binaries as well as
820: with the level of ionization actually measured in these winds. However, the {\it clumping}
821: of the gas throughout the chromosphere/wind gives us ample opportunity to incorporate more
822: highly ionized material deep in the chromosphere and possible ways to excite it. We have
823: assumed that the gas is confined to blobs that fill $\sim$ 10\% of space, surrounded
824: essentially by a vacuum filled with an, as yet unspecified, source of pressure. The energy
825: densities required to drive the wind through gradients of this pressure are certainly
826: sufficiently high to allow higher ionization than the dominant levels we see in the shell lines.
827:
828: \subsection{Velocity Structure from Fe II Self Reversals}
829:
830: We may use our model for 31~Cyg to test the techniques used to deduce wind structure from
831: shell lines in single stars. The problem here is that such analyses for single stars give
832: winds accelerating much more rapidly than those in $\zeta$~Aur binary components. I have
833: used my 31~Cyg model to predict what the velocities of the shell components of strong \Fe2\
834: lines in $\lambda$~Vel might be if it had the same velocity structure as 31~Cyg. This is
835: the star with the best evidence for wind acceleration in \Fe2\ self reversams, for which
836: Carpenter et al.\ derived a wind with $\beta$=0.9, $v_{\infty}$=29--33 \kmps, and \Mdot =
837: 3$\times$10$^{-9}$ \MLrate. In this simulation, I calculated the line-center radial optical
838: depth into the atmosphere for the \Fe2\ and \Mg2\ lines measured by Carpenter et al.\ and
839: determined the position and expansion velocity for $\tau$=2/3, as though the intrinsic line
840: emission core comes from the center of the disc. Table 3 gives the results, and Figure 10
841: displays them. These calculations assume the constraints of Table 2, with turbulence from
842: Column 9. For the mass-loss rate of 31~Cyg, the lines become optically thick very high in
843: the wind, absent ionization of Fe$^{+}$ to Fe$^{+2}$ (Table 3, Column 5), but if we reduce
844: the mass-loss rate somewhat to 9$\times$10$^{-9}$ \MLrate, to partially reflect Carpenter's
845: lower value, and allow for ionization of Fe to Fe$^{+2}$ for \nelect\ $<$ 2$\times$10$^6$
846: (Table 2, Column 8), we get velocities in much better agreement with observed values, as
847: seen in Figure 10. Unfortunately we cannot apply this sort of analysis directly to the
848: self-reversals of \Fe2\ lines in 31~Cyg because the emision is formed not in the inner
849: chromosphere, as in single stars, but by scattering light from the B star in a huge shell
850: surrounding the binary system.
851:
852: These results show that we can actually reproduce the sort of observations taken to
853: indicate a rapid acceleration of the chromosphere in a single star with a model
854: accelerating much more slowly. The secret for doing this is the realization that
855: the observed absorption is limited by ionization of the outer wind as discussed in
856: \S\ 4.2. Therefore, there is really no compelling reason to believe the wind
857: acceleration in the single windy giants is necessarily any faster than in binary
858: components.
859:
860: There is evidence in occultations, as well as in the H$\alpha$ profiles, that
861: chromospheres/winds of single stars are much more extended than commonly thought.
862: An occultation in H$\alpha$ emission for the single M supergiant 119 Tau at M2 Ib
863: (White et al.\ 1982) and an image of $\alpha$~Ori at M2 Iab (Hebden et al.\ 1987)
864: both showed H$\alpha$ emission, hence significant optical depths in the Lyman lines,
865: out to at least twice the photospheric radii. Also, all of these cool supergiants
866: show asymmetric H$\alpha$ profiles indicating the line core is formed in the wind
867: at least some of the time (Mallik 1993). Furthermore, in their model of the wind of
868: $\alpha$~Ori, Harper et al.\ (2001) find a velocity/density structure giving even slower
869: acceleration than we find in 31~Cyg or they found (Harper et al.\ 2005) in $\zeta$~Aur.
870:
871:
872: \section{DISCUSSION}
873:
874: The importance of $\zeta$~Aur binaries is that they tell us just where the bulk of the gas
875: is in the inner few stellar radii of the stars' chromospheres or winds. We can measure
876: column densities directly and get at least a reasonable idea of how dense and hot it is
877: throughout this critical zone where most of the energy is injected into a star's wind.
878: Furthermore, with models like those of \S 3.2, we know that this gas must be emitting
879: the intrinsic lines that we would easily see absent the B companions. The structure is
880: therefore constrained in ways it simply cannot be in any analysis of a single star. I have
881: subjected the model for 31~Cyg, and by extension models for other $\zeta$~Aur binaries, to
882: tests based on various sorts of emission from their chromospheres/winds. That such a model
883: reproduces the intrinsic \C2], \Al2], and thermal radio emission to $\sim$ 50\% with a
884: minimum of fiddling is an excellent test of the reality of the structure derived.
885: Furthermore, the clumping inferred for such models provides an excellent opportunity (n.b.,
886: this really is fiddling) to accomodate tenuous material emitting lines of the more highly
887: ionized species. This agreement is reassuring, inasmuch as most of the properties of the
888: model are constrained by {\it observations}. The chief uncertainties involve ionization of
889: H throughout the atmosphere, which is {\it not} measured well observationally, and ionization
890: of H and the metals in the outer wind (\Telect\ $>$ 9,000K) where many of the self reversals
891: of strong lines would probably be formed.
892:
893: \subsection{Alfven-Wave Models}
894:
895: There are several fairly extensive investigations of the conditions required for driving winds
896: by Alfven waves that provide potentially testable predictions (Hartmann \& McGregor 1980;
897: Hartmann \& Avrett 1981; Holzen, Fl\aa, \& Leer 1983; Kuin \& Ahmad 1989). For the most part
898: these suffer from limitations of assuming the gas is effectively coupled to the fields (i.e.,
899: fully ionized), from not exploring the mechanisms of transfer of momentum from wave to gas
900: realistically (although Holzer et al.\ did begin this process), and therefore from relying on
901: somewhat nonphysical theories of how the waves are damped.
902:
903: There are a couple of ways of applying a wave model for driving a wind. In the more fundamental
904: approach, one would use theories of how such waves are generated and how they interact with partially
905: ionized media to predict the structure and properties of winds theoretically. Holzer et al.\
906: (1983, \S\ VI) actually calculated some models of this sort. Unfortunately, the theories required
907: to do that are difficult to apply, and a knowledge of magnetic fields
908: of actual stars that would support such waves is lacking. The alternative is to form a semi-empirical
909: model for wave-driven winds, somewhat like the chromospheric models I have been discussing, and determine
910: what the properites it must have to fit the observed structure of a wind (viz., mass-loss rate,
911: terminal velocity, velocity-density structure). A good example of this second approach is the analyses
912: of $\zeta$~Aur binaries by Kuin \& Ahmad. Their models find that damping of the wave amplitude must
913: decrease with height to fit observed velocity profiles, as one might well expect theoretically to keep
914: the terminal velocity consistent with observation. Their models also give predictions of the level
915: of turbulence in the chromosphere/wind by associating the lateral displacements of gas by such waves
916: with the observed Doppler widths of gas in the atmospheres of these stars. Predicted Doppler widths,
917: both by Kuin \& Ahmad and Hartmann \& Avrett seem to be larger than the observed turbulence, both
918: in shell absorptions in $\zeta$~Aur binaries and in the optically-thin emission lines of single windy
919: giants. Furthermore, since Alfven waves are transverse, the ``turbulence" would be anisotropic
920: to first order. This prediction is at odds with the observation of isotropic turbulence, although
921: it is based on the idealization of radial magnetic fields, while the actual topology might well be
922: more complicated (see, e.g., Cassinelli et al.\ 1995).
923:
924: Kuin \& Ahmad found from their semi-empirical models that the damping length for Alfven waves
925: must increase with height. The most convincing mechanism for transferring energy from the wave
926: to the gas, i.e., damping it, is so-called ion-neutral friction in which there is a phase lag
927: between the wave's transferring momentum to ions and the ions' subsequently transferring it to
928: neutrals. Hartmann \& McGregor discussed this mechanism, although they had no way of applying
929: it {\it a priori}. For waves with long periods, the transfer can be so rapid that the neutrals
930: are effectively bound to the ions through elastic collisions, and there is little dissipation
931: of wave energy or transfer of momentum. For high frequency, on the other hand, the neutrals
932: cannot respond fast enough to the passage of a wave to partake in its displacements, and the ions
933: just stir them up and dissipate wave motion as heat. Since the dissipated wave energy would go
934: primarily into heating, this mechanism would be better for heating the chromosphere/wind than
935: driving its outflow. Most of the wave energy would be available for heating the wind but not
936: for driving it, since the measured temperatures in these winds are much too low to drive them
937: by thermal expansion (cf. \S\ 3). Presumably the {\it momentum} of the wave would be
938: transferred into wind motion. However, since the ratio of energy to momentum ($E/p$) goes as
939: 1/$v$, with the Alfven speed generally larger than either a thermal or turbulent velocity,
940: philosophically an attractive driving mechanism would transfer its energy to heat or some other
941: mass motion as an intermediate stage. This is why it is so hard to drive winds with radiation
942: pressure ($E/p$ $\sim$ 1/$c$). If all the waves had the same frequency, we would expect the
943: momentum to be deposited in a narrow range of density, hence height, contrary to measurements
944: of the acceleration of actual winds.
945:
946: The mechanism of ion-neutral friction would imply a damping length that decreased with height
947: contrary to Kuin \& Ahmad's result. However, it also implies that waves of different frequencies
948: would be absorbed at different height and, therefore, that the {\it spectrum} of Alfven waves would
949: determine the velocity profile of the wind. At this point our knowledge of the photospheric
950: motions that might be exciting Alfven waves and the spectrum they would produce seems too
951: sketchy to make any testable predictions about a wind's velocity structure.
952:
953: \subsection{An Alternative Wind Mechanism}
954:
955: Let us now take the liberty to speculate about a different way of driving the mass loss of these
956: windy giants. We have developed here a picture of what conditions are required to drive the wind
957: of one particularly well observed wind structure. Pressures required are an order of magnitude
958: greater than those of the implied density/temperature structure. That the microturbulence
959: required to fit line shapes and widths seems to be rather isotropic means that we likely are
960: not simply seeing the effects of globally organized Alfven waves passing through the gas, as
961: proposed by Hartmann \& McGregor (1980) and implied by models of Airapetian et al.\ (2000),
962: for example. There may be another way of supporting a wind with magentic fields, namely using
963: {\it chaotic fields} emerging from the star and diffusing through the gas into space to drag the
964: gas along with it and away from the star. Some form of this idea was implicit in our previous
965: musings about the wind structure of 31~Cyg (Eaton \& Bell 1994, \S\ 6), and Mullan et al.\
966: (1998) may have waved at it in passing. This very speculative picture is fundamentally different
967: from the Alfven model in that the magnetic field being lost can constitute a moderate amount of
968: luminosity. In the standard Alfven model, gas is driven $\sim$ radially from the star along
969: magnetic flux lines anchored permanently in the stellar surface. Here, magnetic flux would be
970: lost at $\sim$ the same rate as the gas and constitute a significant component of the energy loss
971: in the wind through its adiabatic expansion. Such chaotic field would give a much more isotropic
972: pressure, which would impress itself on the random velocities of the gas. We may estimate the effect
973: by assuming the pressures driving the wind are in equipartition with the kinetic energy of random
974: motions (turbulence) in the gas. For this condition the driving energy is \ensuremath{\onehalf{NkT}}
975: per degree of freedom (three of them), $N$ being the number of particles in a random blob of mass $M$.
976: The kinetic energy of the blob is \ensuremath{\onehalf{Mv_{\rm equ}^2}} (per degree of freedom), so
977: that equipartition gives \ensuremath{Mv_{\rm equ}^2} = $NkT$, or
978: \begin{equation}
979: v_{\rm equ} = (k/<m>)^{0.5} T_{\rm therm}^{0.5} = 0.08 T_{\rm therm}^{0.5} \\
980: \end{equation}
981: where we have taken the average mass per particle to be 1.3$m_{\rm H}$. This is to within a factor
982: of $\surd\gamma$ of the sound velocity, but for our elevated artificial temperature. For values
983: of the driving temperature, $T_{\rm therm}$ (Table 2, Column 4), we get the turbulent velocities given
984: in Table 2, Column 9. These values are comparable to the line-of-sight random velocities measured in cool
985: (super)giants, and this fact argues that the driving force must be able to produce the random motions
986: observed. It is unlikely, therefore, to be global Alfven waves. Of course, the pressure of these
987: turbulent motions is itself a major source of momentum and energy in extending the atmosphere and
988: driving the wind.
989:
990: Energy input determines the magnetic fields required in this model, since the magnetic energy density
991: must be greater than the energy per unit volume required to lift the mass out of the potential
992: of the star, B$^2$/8$\pi$ $>$ G$M_{\rm K}$$\rho$/\Ro. For our model, this leads to a field strength
993: of 25 Gauss. If we assume the energy loss is double the potential energy from kinetic energy of
994: the wind and emission from it, the field strength increases only to 35 Gauss.
995:
996: This kind of driving has the advantage over Alfven waves of being able to begin to explain the
997: variation of mass loss from star to star in a way related to stellar structure and evolution.
998: In our simulation of driving with thermal profile, the expansion-velocity structure is determined
999: by the temperature (i.e., energy-density) profile,
1000: while the mass-loss rate is arbitrary, determined by supplying enough energy at some \Mdot\ to
1001: maintain the energy-density profile. With magnetic-flux emergence as the driving mechanism, the
1002: magnetic energy corresponds to the infusion of thermal energy in the coronal model; mass loss
1003: rate, therefore, is proportional to the rate at which magnetic field emerges. This is very
1004: attractive in that there are indications that the winds of the giants vary in response to changes
1005: that can re-excite dynamos in their cores. See Mullan \& MacDonald (2003) for changes
1006: of mundane giants, and recall the shells episodically thrown off by pulsing AGB stars. Alfven
1007: waves, on the other hand, would most likely be excited by convection, therefore be proportional to
1008: luminosity, and be little affected by the strength of the passive magnetic fields serving as
1009: their medium.
1010:
1011: As an alternative to chaotic magnetic fields filling the voids in chromospheric gas, we may imagine
1012: Alfven waves trapped in the cavities between ionized blobs. These waves would have speeds approaching
1013: the speed of light as the density dropped, and would be reflected off the blobs if their frequencies
1014: were below the cyclotron frequency. The critical frequency would rise as the material became more
1015: highly compressed (denser). Eaton \& Bell actually had this mechanism in mind as the driving pressure
1016: in such atmospheres. There must be a rich optics of these Alfven waves waiting to be discovered.
1017:
1018: \section{SUMMARY}
1019:
1020: I have constructed a model for the chromosphere and wind of 31~Cyg which is based on measurements
1021: of physical properties in the outer atmospheres of 31~Cug and other classical $\zeta$~Aur binaries.
1022: It goes beyond other such models in that it derives the poorly understood turbulence and clumping
1023: of the gas from the pressures driving the wind's expansion. It predicts emission of optically-thin
1024: lines and microwave continuum to within 50~\%\ of observed values, excellent agreement in the
1025: circumstances. In this model, the momentum flux required to drive the wind determines the
1026: stratification of the chromosphere where intrinsic lines would be formed. That momentum flux
1027: gives much lower densities than the stratification from thermal momentum flux alone, with the
1028: consequence that the gas must be clumped to produce the observed flux in intrinsic lines. We
1029: also find that this model can reproduce most of the properties of single stars' chromospheric
1030: spectra and argue that the evidence for fundamental differences between single stars and these
1031: binary components is rather weak.
1032:
1033: We must keep in mind, however, that there are some inconsistencies in this picture. The
1034: ionization of C, for instance, is a problem. It cannot be mostly singly ionized, as it
1035: seems to be observationally, without giving fluxes much larger than observed, and the model
1036: for 31~Cyg does not predict the redshifts seen in single supergiants. \C2] multiplet ratios
1037: from the model likewise would not predict the large global electron densities found from
1038: \C2] in single stars. However, there are also inconsistencies in the interpretation of \C2]
1039: in the single stars themselves, as I discuss in \S\ 4.4.
1040:
1041: \acknowledgements
1042: I dedicate this paper to John S.\ Mathis who seemed to have the visceral intuitive grasp of physics
1043: required to address the sort of problems I have discused in this paper. If I'd had the sense
1044: to take his advice to work in this area when I was in graduate school, I probably would have
1045: written it years ago. This research has been supported by NSF through grant HRD-9706268 and
1046: NASA through grant NCC5-511 to TSU.
1047:
1048:
1049: \begin{thebibliography}{}
1050: \bibitem[]{}Ahmad, I. A., \& Stencel, R. E. 1988,
1051: ``The stellar wind velocity function for red supergiants determined in eclipsing binaries,"
1052: \apj, 329, 797
1053: \bibitem[]{}Airapetian, V. S., Ofman, L., Robinson, R. D., Carpenter, K., \& Davila, J. 2000,
1054: ``Winds from Luminous Late-Type Stars. I. The Effects of Non-Linear Alfven Waves,"
1055: \apj, 528, 965
1056: \bibitem[]{}Allen, C. W. 1973, Astrophysical Quantities, 3rd Ed. (London: Athlone)
1057: \bibitem[]{}Avrett, E. H., \& Loeser, R. 1992,
1058: ``The Pandora Atmosphere Program,"
1059: in ASP Conf. Ser. 26, Seventh Cambridge Workshop on Cool Stars, Stellar Systems, and
1060: the Sun, ed. M.S. Giampapa \& J.A. Bookbinder, (San Francisco: ASP), 489
1061: \bibitem[]{}Baade, R., Kirsch, T., Reimers, D., Toussaint, F., Bennett, P. D., Brown, A., \&
1062: Harper, G.M. 1996,
1063: ``The Wind Outflow of $\zeta$~Aurigae: A Model Revision Using Hubble Space Telescope Spectra,"
1064: \apj, 466, 979
1065: \bibitem[]{}Basri, G. S., Linsky, J. L., \& Eriksson, K. 1981,
1066: ``Outer Atmospheres of cool Stars VIII. IUE Observations and Models for the Supergiant Stars
1067: $\beta$~Draconis, $\epsilon$~Geminorun, and $\alpha$~Orionis,"
1068: \apj, 251, 162
1069: \bibitem[]{}Bauer, W. H., \& Stencel, R. E. 1989,
1070: ``Line Identifications in the Ultraviolet Spectrum of 31 Cygni,"
1071: \apjs, 69, 667
1072: \bibitem[]{}Blum, R. D., \& Pradhan, A. K. 1992,
1073: ``Rate Coefficients for the Excitation of Infrared and Ultraviolet Lines in C II, N III, and O IV,"
1074: \apjs, 80, 425
1075: \bibitem[]{}Carpenter, K. G., \& Robinson, R. D. 1997,
1076: ``Plasma Flows and Turbulence in the Outer Atmosphere of $\alpha$~Orionis (M2 Iab),",
1077: \apj, 479, 970.
1078: \bibitem[]{}Carpenter, K. G., Robinson, R. D., Harper, G. M., Bennett, P. D., Brown, A., \& Mullan,
1079: D. J. 1999,
1080: ``The Outer Atmosphere and Wind of the Nearby K Supergiant $\lambda$~Velorum,"
1081: \apj, 521, 382.
1082: \bibitem[]{}Carpenter, K. G., Robinson, R. D., \& Judge, P. G. 1995,
1083: ``Flow and Turbulent Velocities in the Outer Atmosphere of $\gamma$~Crucis (M3.4 III),"
1084: \apj, 444, 424
1085: \bibitem[]{}Carpenter, K. G., Robinson, R. D., Wahlgren, G. M., Ake, T. B., Ebbets, D. C., \&
1086: Walter, F. M. 1991,
1087: ``The Chromosphere of $\alpha$~Tauri,"
1088: \apjl, 377, L48
1089: \bibitem[]{}Cassinelli, J. P., \& Hummer, D. G. 1971,
1090: ``Radiative Transfer in Spherically Symmetric Systems--II The Non-Conservative Case and Linearly
1091: Polarized Radiation,''
1092: \mnras, 153, 9
1093: \bibitem[]{}Cassinelli, J. P., Ignace, A., \& Bjorkman, J. E. 1995,
1094: ``Winds from Rotating Wolf-Rayet Stars: The Wind-Compressed Zone Model,"
1095: in IAU Symp. 163, Wolf-Rayet Stars: Binaries, Colliding
1096: Winds, Evolution, ed. K.A. van der Hucht \& P.M. Williams (Dordrecht: Kluwer), 191
1097: \bibitem[]{}Drake, S. A., Brown, A., \& Reimers, D. 1987,
1098: ``Radio continuum emission from the ionized stellar winds of the cool supergiants in $\zeta$ Aurigae-like systems,"
1099: in Fifth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, ed. J.L. Linsky \& R.E.
1100: Stencel (Berlin: Springer), 322
1101: \bibitem[]{}Drake, S. A., \& Linsky, J. L. 1986,
1102: ``Radio Continuum Emission from Winds, Chromospheres, and Coronae of Cool Giants and Supergiants,"
1103: \aj, 91, 602
1104: \bibitem[]{}Eaton, J. A. 1992,
1105: ``The Intrinsic Lines of $\zeta$~Aurigae Binaries,"
1106: \mnras, 258, 473
1107: \bibitem[]{}Eaton, J. A. 1993a,
1108: ``On the Chromospheric Structure of $\zeta$~Aurigae,"
1109: \apj, 404, 305
1110: \bibitem[]{}Eaton, J. A. 1993b,
1111: ``On the Chromosphere of 32 Cygni,"
1112: \aj, 105, 1525
1113: \bibitem[]{}Eaton, J. A. 1993c,
1114: ``31 Cygni: The B Star and the Wind,"
1115: \aj, 106, 2081
1116: \bibitem[]{}Eaton, J. A. 1994,
1117: ``A Deep Atmospheric Eclipse of AL Velorum,"
1118: \aj, 107, 729
1119: \bibitem[]{}Eaton, J. A. 1995,
1120: ``H-Alpha Measures for Cool Giants,"
1121: \aj, 109, 1797
1122: \bibitem[]{}Eaton, J. A. 1996,
1123: ``Chromospheres and Winds of $\zeta$~Aurigae Binaries,"
1124: in ASP Conf. Ser. 109, Ninth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun,
1125: ed. R. Pallavicini \& A.K. Dupree, (San Francisco: ASP), 503
1126: \bibitem[]{}Eaton, J. A., \& Bell, C. 1994,
1127: ``The 1992/93 Eclipse of 31 Cygni,"
1128: \aj, 108, 2276
1129: \bibitem[]{}Eaton, J. A., \& Henry, G. W. 1996,
1130: ``Chromospheric Variation in Cool Supergiant Stars,"
1131: in IAU Symp. 176, Stellar Surface Structure, ed. K.G. Strassmeier \& J.L. Linsky (Dordrecht:
1132: Reidel), 415
1133: \bibitem[]{}Hartmann, L., \& Avrett, E. 1981,
1134: ``Predictions of Wave-Driven Wind Models,"
1135: SAO Spec. Rept., 392, 197
1136: \bibitem[]{}Hartmann, L. W., \& McGregor, K. B. 1980,
1137: ``Momentum and energy deposition in late-type stellar atmospheres and winds,"
1138: \apj, 242, 260
1139: \bibitem[]{}Haisch, B. M., Linsky, J. L., Weinstein, A., \& Shine, R. A. 1977,
1140: ``Analysis of the Chromospheric Spectrum of O I in Arcturus,"
1141: \apj, 214, 785
1142: \bibitem[]{}Harper, G. M. 1990,
1143: ``Fe I - Mg II K line fluorescence in K giant and bright giant stars,"
1144: \mnras, 243, 381
1145: \bibitem[]{}Harper, G. M. 1992,
1146: ``The Outer Atmospheres of the `hybrid' bright giants: the chromospheres of $\alpha$~TrA (K4 II),
1147: $\iota$~Aur (K3 II), $\gamma$~Aql (K3 II), and $\theta$~Her (K1 II),"
1148: \mnras, 256, 37
1149: \bibitem[]{}Harper, G. M. 1996,
1150: ``Mass-loss and Winds from Cool Stars,"
1151: in ASP Conf. Ser. 109, Ninth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun,
1152: ed. R. Pallavicini \& A.K. Dupree, (San Francisco: ASP), 481
1153: \bibitem[]{}Harper, G. M., Brown, A., Bennett, P. D., Baade, R., Walder, R., \& Hummel, C. A. 2005,
1154: ``VLA Observations of zeta Aurigae: Confirmation of the Slow Acceleration Wind Density Structure,"
1155: \aj, 129, 1018
1156: \bibitem[]{}Harper, G. M., Brown, A., \& Lim, J. 2001,
1157: ``A Spatially Resolved, Semiempirical Model for the Extended Atmosphere of $\alpha$~Orionis (M2 Iab),"
1158: \apj, 551, 1073
1159: \bibitem[]{}Hebden, J. C., Eckart, A., \& Hege, E. K. 1987,
1160: ``The H-alpha Chromosphere of $\alpha$~Orionis,"
1161: \apj, 314, 690
1162: \bibitem[]{}Holzer, T. E., Fl\aa, T., \& Leer, E. 1983,
1163: ``Alfven Waves in Stallar Winds,"
1164: \apj, 275, 808
1165: \bibitem[]{}Johnson, H. R., \& Klinglesmith, D. A. 1965,
1166: ``On the Coupled Line-Transfer Problem for Hydrogen,"
1167: in The Formation of Spectrum Lines, Proc. Second Harvard-Smithsonian Conference on
1168: Stellar Atmospheres, ed. E.H. Avrett, O.J. Gingerich, \& C.A. Whitney, SAO Spec. Rept. No. 167, 221
1169: \bibitem[]{}Judge, P. G. 1986a,
1170: ``Constraints on the Outer Atmospheric Structure of Late-Type Giant Stars with IUE: Method and
1171: Application to Arcturus ($\alpha$ Boo K2 III),"
1172: \mnras, 221, 119
1173: \bibitem[]{}Judge, P. G. 1986b,
1174: ``Constraints on the Outer Atmospheric Structure of Late-Type Giant Stars with IUE: Application
1175: to $\alpha$ Tau (K5 III) and $\beta$ Gru (M5 III),"
1176: \mnras, 223, 239
1177: \bibitem[]{}Judge, P. G. 1988,
1178: ``Spectroscopy of Cool Stars from IUE Data,"
1179: in IAU Symp. 132, The Impact of Very High S/N Spectroscopy on Stellar Physics, ed. G. Cayrel
1180: de Strobel \& M. Spite, 163
1181: \bibitem[]{}Judge, P. G. 1990,
1182: ``On the Interpretation of Chromospheric Emission Lines,"
1183: \apj, 348, 279
1184: \bibitem[]{}Judge, P. G. 1992,
1185: ``Energetics of Stellar Winds,"
1186: in ASP Conf. Ser. 26, Seventh Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun,
1187: ed. M.S. Giampapa \& J.A. Bookbinder, (San Francisco: ASP), 403
1188: \bibitem[]{}Judge, P. G. 1994,
1189: ``The `Monochromatic Density Diagnostic' Technique: First Detection of Multiple Density Components
1190: in the Chromosphere of $\alpha$ Tauri,"
1191: \apj, 430, 351
1192: \bibitem[]{}Judge, P. G., \& Carpenter, K. G. 1998,
1193: ``On Chromospheric Heating Mechanisms of `Basal Flux' Stars,"
1194: \apj, 494, 828
1195: \bibitem[]{}Judge, P. G., \& Jordan, C. 1991,
1196: ``Fe II emission lines. I - Chromospheric spectra of red giants"
1197: \apjs, 77, 75
1198: \bibitem[]{}Judge, P. G., Jordan, C., \& Feldman, U. 1992,
1199: ``Fe II emission lines. II - Excitation mechanisms in cool stars"
1200: \apj, 384, 613
1201: \bibitem[]{}Kelch, W. L., Chang, S.-H., Furenlid, I., Linsky, J. L., Basri, G. S., Chiu, H.-Y.,
1202: \& Maran, S. P. 1978,
1203: ``Stellar model chromospheres. VII - Capella (G5 III), Pollux (K0 III), and Aldebaran (K5 III),"
1204: \apj, 220, 962
1205: \bibitem[]{}Kuin, N. P. M., \& Ahmad, I. A. 1989,
1206: ``A Semi-Empirical Model for the Red Supergiant's Wind in $\zeta$~Aurigae Systems,"
1207: \apj, 344, 856
1208: \bibitem[]{}Lamers, H.J.G.L.M., \& Cassinelli, J. P. 1999, Introduction to Stellar Winds, (Cambridge
1209: Univ.\ Press)
1210: \bibitem[]{}Landsman, W., \& Simon, T. 1993,
1211: ``A Catalogue of Lyman Alpha Fluxes,"
1212: \apj 408, 305
1213: \bibitem[]{}Luck, R. E., \& Lambert, D. L. 1985,
1214: ``Carbon, Nitrogen, and Oxygen in Intermediate-Mass Supergiants: Is Oxygen Underabundant?,"
1215: \apj, 298, 782
1216: \bibitem[]{}Mallik, S. V. 1993,
1217: ``CCD observations of the H-alpha line in late G and K supergiants and their interpretation,"
1218: \apj, 402, 303
1219: \bibitem[]{}McMurray, A. D. 1999,
1220: ``The outer atmosphere of $\alpha$ Tauri - I. A new chromospheric mode,"
1221: \mnras, 302, 37
1222: \bibitem[]{}McMurray, A. D., \& Jordan, C. 2000,
1223: ``The outer atmosphere of $\alpha$ Tauri - III. Inhomogeneities deduced from cold CO fluorescence,"
1224: \mnras, 313, 423
1225: \bibitem[]{}McMurray, A. D., Jordan, C., \& Carpenter, K. G. 1999,
1226: ``The outer atmosphere of $\alpha$ Tauri - II. Fluorescent lines,"
1227: \mnras, 302, 48
1228: \bibitem[]{}Milkey, R. W., Ayres, T. R., \& Shine, R. A. 1975,
1229: ``Resonance line transfer with partial redistribution. III MG II resonance lines in solar-type stars,"
1230: \apj, 197, 143
1231: \bibitem[]{}Moore, C. E. 1970,
1232: ``Selected Tables of Atomic Spectra C~I -- C~VI,"
1233: NSRDS-NBS 3, Sect.\ 3
1234: \bibitem[]{}Mullan, D. J., Carpenter, K. G. \& Robinson, R. D. 1998,
1235: ``Large Variations in the Wind of Single Giants: $\lambda$ Velorum and $\gamma$ Crucis,"
1236: \apj, 495, 927.
1237: \bibitem[]{}Mullan, D. J., \& MacDonald, J. 2003,
1238: ``Onset of mass Loss in Red Giants: Association with an Evolutionary Event,"
1239: \apj, 591, 1203
1240: \bibitem[]{}Osterbrock, D. E. 1962,
1241: ``The Escape of Resonance-Line Radiation from an Optically Thick Nebula,"
1242: \apj, 135, 195
1243: \bibitem[]{}Osterbrock, D. E. 1974, Astrophysics of Gaseous Nebulae, (San Francisco: Freeman)
1244: \bibitem[]{}Robinson, R. D., Carpenter, K. G. \& Brown, A. 1998,
1245: ``A Comparison of the K5 III Stars $\alpha$ Tauri and $\gamma$ Draconis,"
1246: \apj, 503, 396
1247: \bibitem[]{}Schr\"oder, K.-P., Reimers, D., Carpenter, K. G., \& Brown, A. 1988,
1248: ``What does \C2 $\lambda$2325 \AA\ emission tell us about chrompospheres of red supergiants?
1249: A critical test using $\zeta$ Aurigae-type K supergiants,"
1250: \aap, 202, 136
1251: \bibitem[]{}Stencel, R. E., Linsky, J. L., Mullan, D. J., Basri, G. S., \& Worden, S. P. 1980.
1252: ``The outer atmospheres of cool stars. VII - High resolution, absolute flux profiles of the
1253: Mg~II h and k lines in stars of spectral types F8 to M5,"
1254: \apjs, 44, 383
1255: \bibitem[]{}Stencel, R. F., Linsky, J. L., Brown, A., Jordan, C., Carpenter, K. G., Wing, R. F., \&
1256: Czyzak, S. 1981,
1257: ``Density sensitive C II lines in cool stars of low gravity,"
1258: \mnras, 196, 47
1259: \bibitem[]{}Tayal, S. S., Burke, P. G., \& Kingston, A. E. 1984,
1260: ``Electron impact excitation of intercombination transitions in Al II,"
1261: J.Phys.B, 17, 3847
1262: \bibitem[]{}Unno, W. 1952, \pasj, 4, 100
1263: \bibitem[]{}Vernazza, J. A., Avrett, E. H., \& Loeser, R. 1973,
1264: ``Structure of the Solar Chromosphere. Basic Computations and Summary of the Result,"
1265: \apj, 184, 605
1266: \bibitem[]{}Wareing, C. J., Zijlstra, A. A., \& O'Brien, T. J. 2007,
1267: ``Vortices in the Wakes of Asymptotic Giant Branch Stars,"
1268: \apjl, 660, L129
1269: \bibitem[]{}White, N. M., Kreidel, T. J., \& Goldberg, L. 1982,
1270: ``An Occultation Diameter in H-Alpha Light,"
1271: \apj, 254, 670
1272: \bibitem[]{}Wilson, O. C., \& Abt, H. A. 1954,
1273: ``Chromospheric Structure of the K-Type Component of Zeta Aurigae,"
1274: \apjs, 1, 1
1275: \bibitem[]{}Wright, K. O. 1959,
1276: ``The Inner Chromosphere of the K-Type Component of 31 Cygni as Observed at the 1951 Eclipse,"
1277: \pubdao, 11, 77
1278: \bibitem[]{}Wright, K. O. 1970,
1279: ``The Zeta Aurigae Stars,"
1280: Vistas in Astr., 12, 147
1281: \bibitem[]{}Zarro, D. M., \& Rodgers, A. W. 1983,
1282: ``Stellar chromospheres - H-alpha and Ca II K profiles,"
1283: \apjs, 53, 815
1284: \end{thebibliography}
1285:
1286:
1287: \clearpage
1288:
1289:
1290: %%Table 1--Line Strengths
1291:
1292:
1293: \begin{deluxetable}{cccccccc}
1294: \tabletypesize{\scriptsize}
1295: \tablecaption{Line Strengths in Cool Giants and Models}
1296: \tablewidth{0pt}
1297: \tablehead{
1298: \colhead{Star} & \colhead{Spectral} & \colhead{$V$} &
1299: \colhead{f(Ly$\alpha$)} & \colhead{f(2325)} & \colhead{f(2669)} &
1300: \colhead{EW(H$\alpha$)} & \colhead{log(\nelect)}\\
1301: \colhead{} & \colhead{Type} & \colhead{$(B-V)$} &
1302: \colhead{f$_{line}$/f$_V$} & \colhead{f$_{line}$/f$_V$} &
1303: \colhead{f$_{line}$/f$_V$} & \colhead{f$_{line}$/f$_V$}& }
1304: \startdata
1305: $\alpha$ Boo & K1 III & $-$0.04 & $\geq$1.4$\times$10$^{3}$ & 35.7 & 83.1 & 1.12 &9.7\\
1306: & & 1.23 & $\geq$3.7$\times$10$^{-2}$& 9.4$\times$10$^{-4}$& 2.2$\times$10$^{-3}$& & \\
1307: $\alpha$ Tau & K4 III & 0.85 & $\geq$340 & 31.3 & 30.7 & 1.12 &9.0\\
1308: & & 1.44 & $\geq$2,0$\times$10$^{-2}$& 1.9$\times$10$^{-3}$& 1.88$\times$10$^{-3}$& & \\
1309: $\beta$ Gru & M5 III & 2.13 & $>$50 & 31.7 & 28.8 & \nodata &8.5\\
1310: & & 1.57 & $>$9,7$\times$10$^{-3}$ & 6.2$\times$10$^{-3}$& 5.6$\times$10$^{-3}$& & \\
1311: $\lambda$ Vel& K5 Ib-II& 2.21 & \nodata & \nodata & 12 & 1.52 &8.9\\
1312: & & & & & 2.5$\times$10$^{-3}$& & \\
1313: 31 Cyg & K4 I & 3.79 & \nodata & $<$5 & 2.3 & 1.50 &\nodata\\
1314: & & 1.28 & &$<$5$\times$10$^{-3}$& 2.1$\times$10$^{-3}$& & \\
1315: 32 Cyg & K4 I & 3.98 & \nodata &$\leq$5$\pm$100\% & 3.4 & 1.75 &\nodata\\
1316: & & 1.52 & &$\leq$5.3$\times$10$^{-3}$& 3.6$\times$10$^{-3}$& & \\
1317: $\zeta$ Aur & K4 Ib & 3.79 & \nodata &$\leq$4$\pm$100\% & 2.2 & 1.57 &\nodata\\
1318: & & 1.22 & &$\leq$3.7$\times$10$^{-3}$& 2.0$\times$10$^{-3}$& & \\
1319: Model 1 & & & \nodata & 56 & 28 & \nodata &9.8\\
1320: (3\% ioniz) & & & & & & & \\
1321: Model 2 & & & \nodata & 9.1 & 3.6 & \nodata &8.5\\
1322: (var $<$10\%)& & & & & & & \\
1323: Model 3 & & & \nodata & 6.7 & 3.0 & \nodata &8.6\\
1324: (var $<$3\%) & & & & & & & \\
1325: \enddata
1326: \tablecomments{~Line fluxes at the Earth are in 10$^{-13}$ ergs cm$^{-2}$s$^{-1}$.
1327: f(line)/f$_V$ is in \AA\ and assumes f$_V$ = 3.65$\times$10$^{-9}$
1328: ergs~cm$^{-2}$s$^{-1}$\AA$^{-1}$ at $V$ = 0.0. Values of log(\nelect) in Col.\ 8
1329: come from C~II] multiplet ratios as noted in the text.}
1330: \end{deluxetable}
1331:
1332:
1333:
1334: \clearpage
1335:
1336:
1337: %%Table 2--Model for 31 Cyg
1338:
1339:
1340: \begin{deluxetable}{crrcrcccc}
1341: \tabletypesize{\tiny}
1342: \tablecaption{Details of Model for 31 Cyg}
1343: \tablewidth{0pt}
1344: \tablehead{
1345: \colhead{R} & \colhead{$v_{\rm exp}$} & \colhead{$T_{\rm exc}$} & \colhead{$T_{\rm therm}$} &
1346: \colhead{CF}& \colhead{log($n_{\rm H}$)}& \colhead{log(\nelect)} & \colhead{log(\nelect)} &
1347: \colhead{$v_{\rm equ}$} \\
1348: \colhead{} & \colhead{} & \colhead{} & \colhead{} & \colhead{} & \colhead{} &
1349: \colhead{(3\%)} & \colhead{(var)\tablenotemark{a}} & \colhead{} \\
1350: \colhead{(1)} & \colhead{(2)} & \colhead{(3)} & \colhead{(4)} & \colhead{(5)} & \colhead{(6)} &
1351: \colhead{(7)} & \colhead{(8)} & \colhead{(9)}
1352: }
1353: \startdata
1354: 7000.0 & 83.8 & 12500. & 20000. & 1.60 & 4.53 & 3.21 & 3.73 & 11.3 \\
1355: 6200.5 & 83.0 & 12500. & 20000. & 1.60 & 4.64 & 3.32 & 3.84 & 11.3 \\
1356: 5466.3 & 82.1 & 12500. & 20000. & 1.60 & 4.75 & 3.43 & 3.95 & 11.3 \\
1357: 4794.5 & 81.0 & 12500. & 20000. & 1.60 & 4.87 & 3.55 & 4.07 & 11.3 \\
1358: 3897.8 & 79.1 & 12500. & 20000. & 1.60 & 5.06 & 3.74 & 4.26 & 11.3 \\
1359: 3369.9 & 77.4 & 12500. & 24468. & 1.96 & 5.20 & 3.96 & 4.49 & 12.5 \\
1360: 3126.0 & 76.5 & 12500. & 30660. & 2.45 & 5.27 & 4.13 & 4.66 & 14.0 \\
1361: 2894.9 & 75.5 & 12500. & 36526. & 2.92 & 5.34 & 4.28 & 4.80 & 15.2 \\
1362: 2676.2 & 74.3 & 12500. & 42075. & 3.37 & 5.41 & 4.42 & 4.94 & 16.3 \\
1363: 2469.7 & 73.1 & 12500. & 47316. & 3.79 & 5.49 & 4.55 & 5.07 & 17.3 \\
1364: 2275.0 & 71.8 & 12500. & 52258. & 4.18 & 5.57 & 4.67 & 5.19 & 18.2 \\
1365: 2091.8 & 70.3 & 12500. & 56909. & 4.55 & 5.65 & 4.79 & 5.31 & 19.0 \\
1366: 1919.7 & 68.7 & 12500. & 61278. & 4.90 & 5.74 & 4.90 & 5.43 & 19.7 \\
1367: 1758.3 & 66.9 & 12500. & 65374. & 5.23 & 5.82 & 5.02 & 5.54 & 20.4 \\
1368: 1607.3 & 64.9 & 12500. & 69206. & 5.54 & 5.92 & 5.14 & 5.66 & 21.0 \\
1369: 1466.4 & 62.8 & 12500. & 72783. & 5.82 & 6.01 & 5.25 & 5.77 & 21.5 \\
1370: 1335.2 & 60.4 & 12500. & 76113. & 6.09 & 6.11 & 5.37 & 5.89 & 22.0 \\
1371: 1213.3 & 57.8 & 12500. & 79205. & 6.34 & 6.21 & 5.49 & 6.01 & 22.4 \\
1372: 1100.5 & 55.0 & 12306. & 82068. & 6.67 & 6.32 & 5.62 & 6.14 & 22.8 \\
1373: 996.4 & 51.9 & 11902. & 84712. & 7.12 & 6.43 & 5.76 & 6.28 & 23.2 \\
1374: 900.6 & 48.5 & 11508. & 87143. & 7.57 & 6.54 & 5.90 & 6.42 & 23.5 \\
1375: 812.7 & 45.0 & 11103. & 89373. & 8.05 & 6.67 & 6.05 & 6.57 & 23.8 \\
1376: 732.5 & 41.1 & 10709. & 91408. & 8.54 & 6.80 & 6.20 & 6.73 & 24.1 \\
1377: 659.6 & 37.1 & 10308. & 93259. & 9.05 & 6.93 & 6.37 & 6.89 & 24.3 \\
1378: 593.6 & 32.8 & 9910. & 94934. & 9.58 & 7.08 & 6.53 & 7.06 & 24.6 \\
1379: 534.2 & 28.5 & 9515. & 95000. & 9.98 & 7.23 & 6.71 & 7.23 & 24.6 \\
1380: 481.1 & 24.1 & 9133. & 95000. & 10.40 & 7.39 & 6.89 & 7.41 & 24.6 \\
1381: 433.8 & 19.8 & 8755. & 95000. & 10.85 & 7.57 & 7.08 & 7.60 & 24.6 \\
1382: 392.2 & 15.7 & 8388. & 95000. & 11.33 & 7.76 & 7.29 & 7.81 & 24.6 \\
1383: 355.7 & 12.0 & 7193. & 95000. & 13.21 & 7.96 & 7.56 & 7.78 & 24.6 \\
1384: 324.04 & 8.66 & 6830. & 83520. & 12.23 & 8.18 & 7.75 & 7.76 & 23.0 \\
1385: 296.92 & 5.92 & 6492. & 69960. & 10.78 & 8.42 & 7.93 & 7.74 & 21.1 \\
1386: 273.96 & 3.77 & 6178. & 58482. & 9.47 & 8.69 & 8.14 & 7.77 & 19.3 \\
1387: 254.82 & 2.36 & 5883. & 48912. & 8.31 & 8.95 & 8.35 & 7.81 & 17.6 \\
1388: 239.15 & 1.44 & 5625. & 41077. & 7.30 & 9.23 & 8.57 & 7.87 & 16.2 \\
1389: 226.61 & 0.81 & 5410. & 34803. & 6.43 & 9.52 & 8.80 & 7.98 & 14.9 \\
1390: 216.83 & 0.42 & 5244. & 29917. & 5.70 & 9.84 & 9.08 & 8.16 & 13.8 \\
1391: 209.49 & 0.20 & 5105. & 26245. & 5.14 & 10.20 & 9.39 & 8.39 & 12.9 \\
1392: 204.23 & 0.11 & 4998. & 23614. & 4.72 & 10.48 & 9.63 & 8.57 & 12.2 \\
1393: 200.70 & 0.06 & 4930. & 21850. & 4.43 & 10.75 & 9.87 & 8.77 & 11.8 \\
1394: 198.56 & 0.04 & 4905. & 20781. & 4.24 & 10.94 & 10.04 & 8.92 & 11.5 \\
1395: 197.46 & 0.04 & 4905. & 20231. & 4.12 & 10.98 & 10.07 & 8.95 & 11.3 \\
1396: 197.06 & 0.04 & 4905. & 20029. & 4.08 & 10.98 & 10.07 & 8.95 & 11.3 \\
1397: 197.00 & 0.04 & 4905. & 20000. & 4.08 & 10.98 & 10.07 & 8.95 & 11.3 \\
1398: \enddata
1399: \tablenotetext{a}{Ionization as in Figure 5 but limted to 10\% in
1400: the outer chromosphere.}
1401: %% Note to Table 2: For 31 Cyg, we assume D=473 pc; R = 197 \Rsun, M$_K$ = 11.7 \Msun, M$_B$ = 7.1 \Msun, etc.
1402: %% Mdot = 3.0E-8 & \Msun~s$^{-1}$
1403: \end{deluxetable}
1404:
1405: \vfill
1406:
1407:
1408: \clearpage
1409:
1410:
1411: %%Table 3--Expansion Velocties for Shell Lines
1412:
1413:
1414: \begin{deluxetable}{lcccccc}
1415: \tabletypesize{\scriptsize}
1416: \tablecaption{Calculated Expansion Velocties for Shell Lines}
1417: \tablewidth{0pt}
1418: \tablehead{
1419: \colhead{Multiplet} & \colhead{$\lambda$} & \colhead{log(LSF)} &
1420: \colhead{$v_{\lambda Vel}^{\rm obs}$} & \colhead{$v_{\rm 31 Cyg}^{\rm calc}$} &
1421: \colhead{$v_{\lambda Vel}^{\rm calc}$}\\
1422: \colhead{} & \colhead{(\AA)} & \colhead{@6000K} & \colhead{(\kmps)} & \colhead{(\kmps)} &
1423: \colhead{(\kmps)}}
1424: \startdata
1425: Fe II UV1 & 2599.40& 0.86& $-$35.0& $-$83.4& $-$38.0 \\
1426: Mg II UV1 & 2795.52& 0.66& $-$31.6& $-$83.8& $-$38.0 \\
1427: Mg II UV1 & 2802.70& 0.36& $-$33.1& $-$83.7& $-$38.0 \\
1428: Fe II UV1 & 2598.37& 0.35& $-$33.4& $-$82.4& $-$34.7 \\
1429: Fe II UV1 & 2585.88& 0.30& $-$34.7& $-$82.2& $-$33.9 \\
1430: Fe II UV1 & 2607.09& 0.25& $-$34.1& $-$82.0& $-$33.9 \\
1431: Fe II UV3 & 2332.80& 0.19& $-$32.2& $-$81.8& $-$33.0 \\
1432: Fe II UV62 & 2755.73& 0.04& $-$32.0& $-$81.0& $-$34.7 \\
1433: Fe II UV3 & 2364.83& $-$0.01& $-$30.0& $-$80.6& $-$30.5 \\
1434: Fe II UV1 & 2625.66& $-$0.05& $-$33.7& $-$80.3& $-$30.1 \\
1435: Fe II UV32 & 2739.55& $-$0.05& $-$27.2& $-$80.3& $-$33.9 \\
1436: Fe II UV3 & 2338.01& $-$0.06& $-$32.0& $-$80.3& $-$30.5 \\
1437: Fe II UV1 & 2617.62& $-$0.16& $-$33.4& $-$79.4& $-$28.4 \\
1438: Fe II UV35 & 2362.02& $-$0.62& $-$17.8& $-$69.8& $-$20.3 \\
1439: Fe II UV35 & 2331.30& $-$0.75& $-$29.0& $-$64.0& $-$16.6 \\
1440: Fe II UV35 & 2366.59& $-$0.84& $-$13.5& $-$64.0& $-$15.4 \\
1441: Fe II UV64 & 2593.72& $-$0.91& $-$34.0& $-$59.2& $-$19.4 \\
1442: Fe II UV64 & 2591.52& $-$0.93& $-$16.5& $-$53.9& $-$18.2 \\
1443: Fe II UV32 & 2736.97& $-$0.99& $-$16.4& $-$50.1& $-$17.0 \\
1444: Fe II UV35 & 2354.89& $-$0.99& $-$15.9& $-$49.8& $-$11.8 \\
1445: Fe II UV63 & 2761.81& $-$1.37& $-$7.9& $-$25.8& $-$7.4 \\
1446: Fe II UV63 & 2772.72& $-$1.48&($-$0.8)& $-$19.8& $-$5.4 \\
1447: Fe II UV260 & 2741.40& $-$2.86&($-$2.5)& $-$0.5& 0.0 \\
1448: Fe II UV32 & 2732.41& $-$3.11& +2.7& $-$0.2& 0.0 \\
1449: Fe II UV32 & 2759.34& $-$3.18&($-$2.4)& $-$0.1& 0.0 \\
1450: \enddata
1451: \end{deluxetable}
1452:
1453:
1454:
1455:
1456: \clearpage
1457:
1458:
1459: %%Figure 1--Coronal wind model
1460: %%This figure comes from winds/src/31cygmod-fig01.ps
1461:
1462: \begin{figure}
1463: \epsscale{0.65}
1464: \plotone{f1.eps}
1465: \caption{Coronal wind model to fit the derived velocity structure of 31~Cyg.
1466: The dotted line is the velocity profile derived for 31~Cyg by Eaton \& Bell (1994),
1467: while the solid curve is the velocity profile calculated with the temperature
1468: profile of Equations 2. Other curves are the local escape velocity (dashed)
1469: and sound speed for the velocities in Equations 2 (dot-dashed). The circle
1470: shows the sonic point in this model.\label{fig1}}
1471: \end{figure}
1472:
1473:
1474: %%Figure 2--Inferred clumping
1475: %%This figure comes from winds/analysis/clumping2.ps
1476:
1477: \begin{figure}
1478: \epsscale{0.65}
1479: \plotone{f2.eps}
1480: \caption{Inferred clumping for the 31~Cyg model. Solid curve is the calculated
1481: clumping factor, CF, and the dashed line is the equipartition velocity defined
1482: by Equation 7. The atmospheric expansion velocity, $v_{\rm exp}$ is shown for comparison.
1483: \label{fig2}}
1484: \end{figure}
1485:
1486:
1487: %%Figure 3--Profile for optically-thin lines
1488: %%The panels in this figures here come from winds/analysis/31cygmod-fig03a.ps and
1489: %%winds/analysis/31cygmod-fig03b.ps
1490:
1491: \begin{figure}
1492: \begin{center}
1493: \epsscale{1.05}
1494: \plottwo{f3a.eps}{f3b.eps}
1495: \end{center}
1496: \caption{Calculated intensity of emission over the stellar disc for optically-thin
1497: emission lines. (left) This figure assumes the emitter is in the same ionization stage
1498: throughout the chromosphere, \nelect\ in Column 7 of Table2. (right) This figure is for
1499: the more realistic level of ionization (\nelect\ in Column 8 of Table2).
1500: \label{fig3}}
1501: \end{figure}
1502:
1503:
1504: %%Figure 4--Profile of RADIO emission
1505: %%This figure comes from winds/analysis/radio.radio-emis2.ps
1506:
1507: \begin{figure}
1508: \epsscale{0.75}
1509: \plotone{f4.eps}
1510: \caption{Calculated intensity of 6.17-cm radio emission over the stellar disc for two
1511: models: dashed curve for the model with variable ionization limited to 10\% (\nelect\ in
1512: Column 8 of Table 2) and solid curve for the same model with 75\% ionization of H
1513: in the outer atmosphere. This latter distribution would project a disc of 38 mas and
1514: have a spectral index $\alpha$ = 1.08, comparable to the few values measured by
1515: Drake \& Linsky (1986).
1516: \label{fig4}}
1517: \end{figure}
1518:
1519:
1520: %%Figure 5--Ionization from PANDORA
1521: %%This figure comes from winds/pandora/nenh-raw.ps
1522:
1523: \begin{figure}
1524: \epsscale{0.55}
1525: \plotone{f5.eps}
1526: \caption{Calculated ionization in three semi-empirical models for cool giant stars.
1527: The graph gives the ratio of electron density to total H density as a function of
1528: temperature. At low temperature, where ionization of H is low, the electron density
1529: is dominated by metals. At high temperature H becomes completely ionized, in these
1530: models, if not in actual stars. The region in which most of the chromospheric lines
1531: are emitted has a H ionization of a few percent. The solid curve shows the ralation
1532: I have adopted to relate electron density to excitation temperature in the calculations
1533: with variable ionization of H. Discrete symbols represent calculations for three models:
1534: dots, $\alpha$~Tau (Eaton 1995: Table 4, Model t8), circles, $\epsilon$~Gem, and
1535: asterisks, $\zeta$~Aur.
1536: \label{fig5}}
1537: \end{figure}
1538:
1539: %% at_t8.top (dots), eps_prd.top (circles), and za_t2.top (asterisks * 3)
1540:
1541:
1542: %%Figure 6--Excitation Temperatures
1543: %%This figure comes from winds/iue/texc-rhox.ps
1544:
1545: \begin{figure}
1546: \epsscale{0.55}
1547: \plotone{f6.eps}
1548: \caption{Excitation temperature, \Texc, vs.\ tangential mass column density, \Rhox,
1549: as measured in four $\zeta$~Aur systems. Plusses are for $\zeta$~Aur, asterisks
1550: for 32~Cyg, circles for 31~Cyg, and dots for 22~Vul.
1551: \label{fig6}}
1552: \end{figure}
1553:
1554:
1555: %%Figure 7--Profiles for two ions
1556: %%The two panels in this figures here come from winds/analysis/uniform
1557:
1558: \begin{figure}
1559: \begin{center}
1560: \epsscale{0.85}
1561: \plotone{f7.eps}
1562: \end{center}
1563: \caption{Line profiles for the spherical model for two typical ions. The dashed curve
1564: shows the calculated profile for a line like \Al2] excited throughout the chromosphere with
1565: an isotropic turbulence of 18 \kmps\ superimposed on the expansion of the chromosphere/wind.
1566: This profile is fit to within the resolution of the plot by a Gaussian with FWHM = 30.3 \kmps.
1567: The solid curve shows the effect of superimposing a global 10~\kmps\ downdraft on this profile.
1568: It is again fit with a Gaussian (no \C2]-like wings) but shifted 3.8 \kmps\ to the red and
1569: broadened slightly to FWHM = 31.3 \kmps.
1570: \label{fig7}}
1571: \end{figure}
1572:
1573:
1574: %%Figure 8--Effect of turbulence on profiles
1575: %%Panels a--e come from winds/analysis/anisotropic; panel f from winds/analysis/c2
1576:
1577: \begin{figure}
1578: \begin{center}
1579: \epsscale{0.80}
1580: \plottwo{f8a.eps}{f8b.eps}
1581: \plottwo{f8c.eps}{f8d.eps}
1582: \plottwo{f8e.eps}{f8f.eps}
1583: \end{center}
1584: \caption{\scriptsize The effects of various types of anisotropic turbulence on calculated line
1585: profiles. In these figures, the solid curve is the calculated profile, and the dashed curve is
1586: a Gaussian fit. For the more extreme profiles to the left, this Gaussian is fit to the whole
1587: profile; for the subtler profiles to the right, the Gaussian is fit to inner part
1588: (Intensity$\geq$0.5). Panel a) shows the really awful effect of purely {\it radial} turbulence,
1589: while Panel b) shows the more subtle effect of purely {\it tangential} turbulence. Panels c) and
1590: d) show, respectively, the effect of 20 \kmps\ of radial or tangential turbulence combined with
1591: 5 \kmps\ of isotropic turbulence. Panel e) is an experiment with confining radial and tangential
1592: turbulence to only the disc. The radial distribution is noticably different than Panel a) because
1593: most of the emission in Panel a) comes from {\it beyond} the edge of the disc. Panel f) shows the
1594: effect of isotropic turbulence with only the gas hotter than 5200~K emitting.
1595: \label{fig8}}
1596: \end{figure}
1597:
1598:
1599: %%Figure 9--Profiles pf C II]
1600: %%The two figures here come from winds/analysis/composite
1601:
1602: \begin{figure}
1603: \begin{center}
1604: \epsscale{1.00}
1605: \plottwo{f9a.eps}{f9b.eps}
1606: \end{center}
1607: \caption{An attempt to fit the peculiar shapes and shifts of \C2] emission in windy giants.
1608: Calculated profiles are the solid curves; Gaussians fit to the inner parts of these profiles
1609: (I$>$0.5) are dashed curves. In both cases gas in the deeper parts of the chromosphere
1610: (cooler than some threshold value) was given a 10~\kmps\ downward velocity.
1611: Panel a) at left shows the effect of suppressing all the emission at temperatures between
1612: 6000~K and 8500~K while letting 30\% of the gas below 6000~K emit. Panel b) at right shows
1613: the effect suppressing all emission from gas cooler than 8000~K except for gas with electron
1614: densities above log(\nelect) $\geq$\ 8.6. In both cases extra broadening in the wings
1615: comes from emission from gas with significant expansion velocity projected into the line
1616: of sight.
1617: \label{fig9}}
1618: \end{figure}
1619:
1620:
1621: %%Figure 10--Shell velocities in lam Vel
1622: %%This figure comes from winds/analysis/fe2
1623:
1624: \begin{figure}
1625: \epsscale{0.75}
1626: \plotone{f10.eps}
1627: \caption{A comparison of shell velocities for $\lambda$~Vel with calculations for the
1628: 31~Cyg model under assumptions about mass loss and ionization. Measured values come
1629: from Carpenter et al.\ (1999, Table 3). The circles are two lines for which the measured
1630: values seemed discrepant in Carpenter's paper, one even falling beyond the limits of
1631: his Fig.\ 9. Calculated velocities assume \Mdot = 9$\times$10$^{-9}$ \MLrate and that
1632: Fe is doubly ionized at \nelect\ $<$ 2$\times$10$^6$.
1633: \label{fig10}}
1634: \end{figure}
1635:
1636:
1637: \end{document}
1638: