1: \documentclass[preprint,proceedings]{rmaa}
2: \usepackage{paralist}
3: \usepackage{psfrag,color}
4:
5: % This is only used for making the header for the preprint version
6: \SetYear{2004}
7: \SetConfTitle{Gravitational Collapse: From Massive Stars to Planets}
8:
9: % Please use mixed case here, since this title gets propagated onto
10: % the web page, ADS entry, etc.
11: \title{Photon Bubbles in Young Massive Stars}
12:
13: % For the conference proceedings, the author affiliations should be
14: % subscripted, using \altaffil and/or \altaffilmark + \altaffiltext
15: % Note that \altaffilmark goes after a comma and that `and' is spelt
16: % out.
17: \author{
18: N. J. Turner,\altaffilmark{1}
19: H. W. Yorke,\altaffilmark{1}
20: A. Socrates,\altaffilmark{2} and
21: O. M. Blaes\altaffilmark{2}
22: }
23:
24: \altaffiltext{1}{Jet Propulsion Laboratory, California Institute of
25: Technology.}
26: \altaffiltext{2}{University of California at Santa Barbara.}
27:
28: % Authors for running headers - surnames only, et al. if more than 3.
29: \shortauthor{Turner et al.}
30: % Title for running header
31: \shorttitle{Photon Bubbles in Young Massive Stars}
32:
33: % Full postal addresses (in alphabetical surname order!)
34: % plus email addresses in parentheses.
35: \fulladdresses{
36: % Formatted in list environment, so each group is an \item
37: \item N. J. Turner and H. W. Yorke: MS 169-506, Jet Propulsion
38: Laboratory, California Institute of Technology, Pasadena CA 91109,
39: USA; \email{neal.turner@jpl.nasa.gov}.
40: \item O. M. Blaes and A. Socrates: Physics Department, University of
41: California, Santa Barbara CA 93106, USA.
42: }
43:
44: % List of authors used to construct table of contents
45: \listofauthors{N. J. Turner, H. W. Yorke, A. Socrates \& O. M. Blaes}
46: % Each author in Surname, Initials format, used in generating Author
47: % Index entries.
48: \indexauthor{Turner, N. J.}
49: \indexauthor{Yorke, H. W.}
50: \indexauthor{Socrates, A.}
51: \indexauthor{Blaes, O. M.}
52:
53: % English abstract
54: \abstract{Spectroscopic studies indicate that gas in the photospheres
55: of young O stars moves at speeds up to the sound speed. We show,
56: using two-dimensional radiation MHD calculations and results from a
57: local linear analysis, that the motions may be due to photon bubble
58: instability if young O stars have magnetic fields. }
59:
60: % Spanish abstract - leave blank and it will be translated by the
61: % editors.
62: \resumen{ }
63:
64: % Keywords must be from the standard list and in alphabetical order.
65: \addkeyword{stars: early-type}
66: \addkeyword{stars: formation}
67:
68: \def\msol{\rm\,M_\odot}
69: \def\divv{{\bf\nabla\cdot v}}
70:
71: \begin{document}
72: \maketitle
73:
74: \section{Introduction}
75:
76: Young high-mass stars may trigger or terminate nearby star and planet
77: formation through ionizing radiation and winds launched from their
78: surfaces. The properties of the radiation and winds depend on the
79: structure of the stellar surface layers. Stars of more than 15 Solar
80: masses reach the main sequence while still accreting material
81: \citep{ys02}. High-mass main sequence stars are thought to have
82: stably stratified outer layers, yet gas motions are present in the
83: atmospheres of these stars.
84:
85: Absorption lines formed in the photospheres of O stars are broader
86: than expected based on the temperature, pressure and stellar rotation.
87: If the rotation axes are randomly oriented with respect to our line of
88: sight, some stars will be viewed pole-on and show no rotational
89: broadening. However samples of O stars appear to contain no such
90: pole-on cases \citep{s56,ce77,p96,hs97}. The minimum apparent
91: equatorial rotation speed is a function of spectral type and
92: luminosity. Among main-sequence O stars in the Small Magellanic Cloud
93: cluster NGC 346, the excess broadening of UV metal lines ranges from
94: 25 km s$^{-1}$ at O2, similar to the speed of sound at the
95: photosphere, to 5 km s$^{-1}$ at O9.5 \citep{bl03}.
96:
97: Processes that might be responsible for the additional line
98: broadening include
99: \begin{enumerate}
100: \item {\em Vertical velocity gradients in the stellar wind
101: acceleration region} \citep{k92}: Minimum line widths are similar in
102: stars with and without strong outflows. Also, stellar atmosphere plus
103: wind calculations including a variation in velocity through the
104: acceleration region produce photospheric lines similar to those in
105: hydrostatic calculations \citep{bl03}.
106: \item {\em Global non-radial pulsations:} Low levels of spectral line
107: shape variations in main-sequence O stars \citep{fg96} suggest that
108: any global non-radial modes are either of low amplitude or high order.
109: An unresolved issue is how such modes might be excited.
110: \item {\em Strange modes:} Instability occurs only for masses greater
111: than $80\msol$ in zero-age main sequence stars with metallicity
112: $Z=0.02$ \citep{gk93}, while excess photospheric line widths are
113: observed in main-sequence stars with masses down to $20\msol$.
114: \item {\em Shaviv modes:} The instabilities are found only at
115: luminosities greater than half the Eddington value \citep{s01}, while
116: excess line broadening occurs in stars with one-tenth the Eddington
117: luminosity.
118: \end{enumerate}
119:
120: Here we explore the possibility that a fifth mechanism, photon bubble
121: instability \citep{a92,g98}, may lead to small-scale motions in the
122: surface layers of O stars. A general local WKB analysis of linear
123: disturbances in optically-thick radiating atmospheres indicates
124: instability under a wide range of conditions \citep{bs03}. Photon
125: bubbles are driven by the radiative flux
126: (Figure~\ref{fig:howitworks}), and grow if (1) Rosseland mean optical
127: depth per wavelength is greater than unity, (2) a magnetic field is
128: present, and (3) the radiative flux exceeds a critical value that in
129: radiation-supported atmospheres is approximately the gas sound speed
130: times the sum of gas and photon energy densities. The surface layers
131: of O stars generally satisfy the first and third criteria, but until
132: recently were thought to be unmagnetized. However $\theta^1$~Ori~C,
133: the illuminating star of the Orion Nebula and one of the nearest young
134: O stars, shows polarimetric variations in photospheric lines with
135: rotational phase, consistent with a 1.1~kG dipole field inclined
136: $42^\circ$ from the rotation axis \citep{db02}. Also, narrow X-ray
137: emission lines indicate some gas near the star moves more slowly than
138: the wind. The line ratios show that most of the X-ray flux is thermal
139: emission from plasma hotter than 15 million~K, too hot to arise in
140: shocks in the wind, and as hot as coronae in some magnetically active
141: lower-mass stars \citep{sc03}. Overall, there is good evidence for a
142: magnetic field in $\theta^1$~Ori~C.
143:
144: \begin{figure}[t!]
145: \vspace*{-6mm} \hspace*{-3mm}
146: \includegraphics[width=0.95\columnwidth]{f1.eps} \vspace*{-1mm}
147: \caption{How photon bubbles work. {\em Top:} A patch of atmosphere
148: (gray square) is initially in hydrostatic balance. When density is
149: reduced slightly in a small elongated region, photons diffuse more
150: easily up the long axis. The perturbed radiation flux exerts an
151: extra force with a component along the magnetic field. Gas is
152: driven along the field to the right, out of the region of low
153: density, and the perturbation grows. {\em Bottom:} As fresh gas
154: enters the region of low density from the left, the pattern
155: propagates to the left at about the gas sound speed. }
156: \label{fig:howitworks}
157: \end{figure}
158:
159: Photon bubbles, unlike strange modes, can grow in scattering
160: atmospheres. They differ from Shaviv instabilities in that they
161: require magnetic fields and may grow when the luminosity is
162: substantially less than the Eddington value.
163:
164: \section{Methods}
165:
166: A small patch of the surface layers of the star is followed in
167: Cartesian coordinates $(x, y, z)$. The $x$-axis is horizontal, the
168: $z$-axis vertical, and symmetry is assumed along the $y$-axis. The
169: domain is $10^{10}$~cm on a side, or about 1.5\% of the radius of
170: $\theta^1$ Ori C, and divided into $32\times 32$ zones. The top of
171: the domain is placed at the photosphere. The side boundaries are
172: periodic, and the upper and lower boundaries are transparent walls
173: through which radiation may pass, but gas and magnetic fields do not.
174: The temperature of the lower boundary is fixed at its initial value,
175: while the radiative flux is continuous across the upper boundary. The
176: equations of radiation MHD,
177: \begin{equation}\label{eqn:cty}
178: {D\rho\over D t}+\rho\divv=0,
179: \end{equation}
180: \begin{equation}\label{eqn:eomg}
181: \rho{D{\bf v}\over D t} = -{\bf\nabla}p
182: + {1\over 4\pi}({\bf\nabla\times B}){\bf\times B}
183: + {\chi_F\rho\over c}{\bf F} - \rho g {\bf\hat z},
184: \end{equation}
185: \begin{equation}\label{eqn:eoeg}
186: \rho{D\over D t}\left({e\over\rho}\right) =
187: - p\divv - \kappa_P\rho(4\pi B - c E),
188: \end{equation}
189: \begin{equation}\label{eqn:eoer}
190: \rho{D\over D t}\left({E\over\rho}\right) =
191: - {\bf\nabla\cdot F} - {\bf\nabla v}:\mathrm{P}
192: + \kappa_P\rho(4\pi B - c E),
193: \end{equation}
194: and
195: \begin{equation}\label{eqn:dbdt}
196: {\partial{\bf B}\over\partial t} = {\bf\nabla\times}({\bf v\times B}),
197: \end{equation}
198: \citep{mm84,sm92,bs03} are solved using the ZEUS code
199: \citep{sn92a,sn92b} with its flux-limited radiation diffusion module
200: \citep{ts01}. A uniform gravity $g$ is included. Rotation is
201: ignored, as photon bubbles grow to saturation within a typical stellar
202: rotation period. Emission proportional to the blackbody rate $B$ is
203: assumed in equations~\ref{eqn:eoeg} and~\ref{eqn:eoer}. Temperatures
204: are calculated using a mean mass per particle of $0.6$ protons. The
205: total flux-mean opacity $\chi_F$ includes contributions from electron
206: scattering 0.34 cm$^2$ g$^{-1}$ and bound-free and free-free processes
207: $2\times 10^{52} \rho^{4.5}e^{-3.5}$ cm$^2$ g$^{-1}$, where $e$ is the
208: gas internal energy per unit volume and $\rho$ is the mass density.
209: The Planck-mean absorption opacity $\kappa_P$ due to the bound-free
210: and free-free effects is $7.4\times 10^{53} \rho^{4.5}e^{-3.5}$ cm$^2$
211: g$^{-1}$. The radiative flux ${\bf F} =
212: -{c\Lambda\over\chi_F\rho}{\bf\nabla}E$ is calculated using a limiter
213: $\Lambda$ that equals $\frac{1}{3}$ in optically-thick regions, and is
214: reduced in regions of low optical depth so that the flux does not
215: exceed the product of the radiation energy density $E$ and light speed
216: $c$ \citep{lp81}. The set of equations is closed using an ideal gas
217: equation of state $p=(\gamma-1)e$ with $\gamma=5/3$.
218:
219: Initial conditions are constructed by integrating the equations of
220: hydrostatic equilibrium and radiative flux conservation from the
221: photosphere down into the interior. Three initial states are
222: considered, spanning the range of effective temperatures found in O
223: stars. The photospheric temperatures in Kelvins and densities in g
224: cm$^{-3}$ are (A) $50\,000, 1.8\times 10^{-9}$, (B) $40\,000,
225: 2.2\times 10^{-9}$, and (C) $30\,000, 3.0\times 10^{-9}$. Surface
226: gravity $10^{3.8}$ cm s$^{-2}$ is used in all cases. The Eddington
227: ratios between the last two terms in equation~\ref{eqn:eomg} at the
228: photospheres are (A) $0.65$, (B) $0.28$, and (C) $0.11$. We study the
229: growth of photon bubbles at the stellar magnetic equator by beginning
230: with a uniform, horizontal magnetic field of 1~kG. Radiation, gas and
231: magnetic pressures at the photosphere are in ratios (A) $0.40:0.31:1$,
232: (B) $0.16:0.31:1$, and (C) $0.05:0.31:1$. Random zone-to-zone initial
233: density perturbations with amplitude 0.1\% are added only in the
234: middle half of the domain height, to reduce any effects of the
235: boundaries on the linear growth.
236:
237: \section{Results}
238:
239: Each calculation passes through three stages. First, the
240: perturbations cause small damped vertical oscillations with period
241: equal to the gas sound crossing time and amplitude 0.1 km s$^{-1}$.
242: Next, linear eigenmodes appear and grow exponentially. The
243: fastest-growing modes are gas sound waves with wavefronts inclined 30
244: to 45 degrees from vertical, and group velocity roughly parallel to
245: the magnetic field. Finally, the growth saturates and amplitudes vary
246: little over many gas sound crossing times.
247:
248: \subsection{Linear Growth}
249:
250: The time histories of the horizontal kinetic energies are shown in
251: Figure~\ref{fig:time}. Linear growth rates increase with photospheric
252: temperature. In the $50\,000$~K case (A), the fastest mode has 3
253: wavelengths in the domain width and 2 in the height, corresponding to
254: wavelength $2.8\times 10^9$ cm and wavevector $34^\circ$ from
255: horizontal. The power in the mode, measured using the Fourier
256: transform of the horizontal velocity pattern, grows $e$-fold every
257: 3000~seconds. The growth rate is $2/3$ that obtained by solving the
258: linear dispersion relation, Blaes \& Socrates (2003) equation~49, for
259: conditions at the center of the computational domain. The slower
260: growth in the simulation may be due to the modest grid resolution.
261:
262: \subsection{Saturation}
263:
264: The disturbances reach larger amplitudes in the cases with hotter
265: photospheres. The largest horizontal velocities are $30$ km~s$^{-1}$
266: in the $50\,000$~K case (A), $12$ km~s$^{-1}$ in the $40\,000$~K case
267: (B), and $6$ km~s$^{-1}$ in the $30\,000$~K case (C). In the hottest
268: case (A), the speeds are similar to the isothermal gas sound speed of
269: $26$ km~s$^{-1}$ at the photosphere, and weak shocks are present. The
270: structure resembles that proposed by Begelman (2001), but with crossed
271: shocks due to the initial horizontal symmetry. Gas oscillates back
272: and forth along magnetic field lines, experiencing repeated
273: compression and expansion. Densities vary about their initial values
274: by up to a factor three, while radiation energy remains almost uniform
275: due to rapid diffusion (Figure~\ref{fig:saturated}). Radiation
276: escapes more easily through the inhomogeneous atmosphere than through
277: the original hydrostatic structure, and the time-averaged radiative
278: flux is 10\% greater than initially. The density variations in the
279: $40\,000$~K case (B) are 50\%, and in the $30\,000$~K case (C) are
280: 20\%. The magnetic fields bend $30^\circ$ up and down in the hottest
281: case, and remain close to horizontal in the two cooler cases.
282:
283: \begin{figure}[t!]
284: \hspace*{1mm} \includegraphics[width=0.91\columnwidth]{f2.eps}
285: \caption{Domain-averaged kinetic energies in horizontal motions
286: versus time, in the calculations with photospheric temperatures
287: $50\,000$ (top), $40\,000$ (middle), and $30\,000$~K (bottom). The
288: upper horizontal line indicates the energy density in the initial
289: magnetic field. } \label{fig:time}
290: \end{figure}
291:
292: \begin{figure}[t!]
293: \hspace*{12mm} \includegraphics[width=0.67\columnwidth]{f3.eps}
294: \caption{Snapshot of saturated photon bubble instability in the
295: calculation with photospheric temperature $50\,000$~K, at
296: $56\,000$~seconds. The initial photosphere lies along the top
297: boundary, and the domain height and width are $10^{10}$~cm. {\em
298: Top:} Density on a gray scale logarithmic between $1.2\times
299: 10^{-9}$ (black) and $1.3\times 10^{-8}$ g cm$^{-3}$ (white).
300: Velocities are shown by arrows, the longest corresponding to 29 km
301: s$^{-1}$. {\em Bottom:} Radiation pressure on a gray scale linear
302: between $6\,600$ and $210\,000$ dyn cm$^{-2}$. Magnetic fields are
303: shown by arrows, the longest 1500~G. } \label{fig:saturated}
304: \end{figure}
305:
306: \section{Conclusions}
307:
308: Photon bubble instability is present in the surface layers of
309: magnetized main-sequence O stars according to the results of a local
310: linear WKB analysis. We use two-dimensional radiation MHD
311: calculations with horizontal magnetic fields and closed vertical
312: boundaries to show that the instability can lead to small-scale
313: movements of gas in the surface layers, with velocities about equal to
314: the observed linewidth excesses. In a calculation with parameters
315: appropriate for an early O star, photon bubbles result in horizontal
316: density variations near the photosphere. Radiation diffuses more
317: rapidly through the inhomogeneities than through the hydrostatic
318: atmosphere with the same column depth, and the radiative flux is 10\%
319: greater than that calculated assuming hydrostatic equilibrium. An
320: enhanced flux
321: may affect measurements of stellar parameters. Issues remaining to be
322: resolved
323: \linebreak \adjustfinalcols
324: include the dependence on the local magnetic field
325: orientation and the shapes of the spectral lines expected from
326: emission integrated over the stellar disk.
327:
328: Photon bubbles reaching amplitudes similar to those in the hottest
329: case we considered may eject blobs of gas through the photosphere, and
330: could lead to density variations in the winds of O stars. The motions
331: might transfer energy into an initially weak magnetic field.
332: Additional radiation MHD calculations indicate the saturation
333: amplitude generally increases with decreasing surface gravity, so
334: photon bubbles may reach larger amplitudes in O giants and
335: supergiants. Solutions of the dispersion relation show that photon
336: bubbles may be present also in accretion disks around young massive
337: stars.
338:
339: \acknowledgements
340:
341: N. J. T. was supported by a National Research Council fellowship. The
342: work was carried out in part at the Jet Propulsion Laboratory,
343: California Institute of Technology.
344:
345: \begin{thebibliography}
346: \bibitem[Arons(1992)]{a92} Arons J. 1992, ApJ, 388, 561
347: \bibitem[Begelman(2001)]{b01} Begelman M. C. 2001, ApJ, 551, 897
348: \bibitem[Blaes \& Socrates(2003)]{bs03} Blaes O. \& Socrates A. 2003,
349: ApJ, 596, 509
350: \bibitem[Bouret et al.(2003)]{bl03} Bouret J.-C. et al.
351: %, Lanz T., Hillier D. J., Heap S. R., Hubeny I., Lennon D. J.,
352: %Smith L. J. \& Evans, C. J.
353: 2003, ApJ, 595, 1182
354: \bibitem[Conti \& Ebbets(1977)]{ce77} Conti P. S. \& Ebbets D. 1977,
355: ApJ, 213, 438
356: \bibitem[Donati et al.(2002)]{db02} Donati J.-F., Babel J., Harries
357: T. J., Howarth I. D., Petit P. \& Semel, M. 2002, MNRAS, 333, 55
358: \bibitem[Fullerton et al.(1996)]{fg96} Fullerton A. W., Gies
359: D. R. \& Bolton C. T. 1996, ApJS, 103, 475
360: \bibitem[Gammie(1998)]{g98} Gammie C. 1998, MNRAS, 297, 929
361: \bibitem[Glatzel \& Kiriakidis(1993)]{gk93} Glatzel W. \& Kiriakidis
362: M. 1993, MNRAS, 262, 85
363: \bibitem[Howarth et al.(1997)]{hs97} Howarth I. D., Seibert K. W.,
364: Hussain G. A. J. \& Prinja R. K. 1997, MNRAS, 284, 265
365: \bibitem[Kudritzki(1992)]{k92} Kudritzki R. P. 1992, A\&A, 266, 395
366: \bibitem[Levermore \& Pomraning(1981)]{lp81} Levermore C. D. \&
367: Pomraning G. C. 1981, \apj, 248, 321
368: \bibitem[Mihalas \& Mihalas(1984)]{mm84} Mihalas D., \& Mihalas
369: B. W. 1984, Foundations of Radiation Hydrodynamics (Oxford: Oxford
370: Univ.\ Press)
371: \bibitem[Penny(1996)]{p96} Penny L. R. 1996, ApJ, 463, 737
372: \bibitem[Schulz et al.(2003)]{sc03} Schulz N. S., Canizares C.,
373: Huenemoerder D., \& Tibbets K. 2003, ApJ, 595, 365
374: \bibitem[Shaviv(2001)]{s01} Shaviv N. J. 2001, ApJ, 549, 1093
375: \bibitem[Slettebak(1956)]{s56} Slettebak A. 1956, ApJ, 124, 173
376: \bibitem[Stone \& Norman(1992a)]{sn92a} Stone J. M., \& Norman
377: M. L. 1992a, \apjs, 80, 753
378: \bibitem[Stone \& Norman(1992b)]{sn92b} Stone J. M., \& Norman
379: M. L. 1992b, \apjs, 80, 791
380: \bibitem[Stone, Mihalas \& Norman(1992)]{sm92} Stone J. M., Mihalas
381: D., \& Norman M. L. 1992, \apjs, 80, 819
382: \bibitem[Turner \& Stone(2001)]{ts01} Turner N. J. \& Stone
383: J. M. 2001, \apjs, 135, 95
384: \bibitem[Yorke \& Sonnhalter(2002)]{ys02} Yorke H. W. \& Sonnhalter
385: C. 2002, ApJ, 569, 846
386: \end{thebibliography}
387:
388: \end{document}
389: