1: %%**********************************************************************************
2: %% Martins et al., 2006: IRS16SW: a massive binary in the Galactic Center
3: %% Analysis made at MPE, Garching.
4: %%**********************************************************************************
5:
6: %\documentclass[manuscript]{aastex}
7: %\documentclass[12pt,preprint]{aastex}
8: %\documentclass[preprint2]{aastex}
9: \documentclass{emulateapj}
10:
11: %\usepackage{natbib}
12: %\usepackage[figuresright]{rotating}
13: %\usepackage{graphicx}
14: %\usepackage{subfigure}
15: \bibliographystyle{apj}
16:
17: %-----------------------------------------------------------------------
18:
19: \newcommand{\hei}{He~{\sc i} 2.112 $\mu m$}
20: \newcommand{\brg}{Br$\gamma$}
21: \newcommand{\mum}{\ifmmode \mu m \else $\mu m$\fi}
22:
23: \newcommand{\teff}{\ifmmode T_{\rm eff} \else $T_{\mathrm{eff}}$\fi}
24: \newcommand{\logg}{\ifmmode \log g \else $\log g$\fi}
25: \newcommand{\lL}{\ifmmode \log \frac{L}{L_{\odot}} \else $\log \frac{L}{L_{\odot}}$\fi}
26: \newcommand{\mdot}{$\dot{M}$}
27: \newcommand{\myr}{M$_{\odot}$ yr$^{-1}$}
28: \newcommand{\vsini}{$V$ sin$i$}
29: \newcommand{\vinf}{$v_{\infty}$}
30: \newcommand{\vturb}{v$_{turb}$}
31: \newcommand{\vesc}{v$_{esc}$}
32: \newcommand{\kms}{km s$^{-1}$}
33: \newcommand{\msun}{\ifmmode M_{\odot} \else M$_{\odot}$\fi}
34: \newcommand{\zsun}{\ifmmode Z_{\odot} \else Z$_{\odot}$\fi}
35: \newcommand{\lsun}{\ifmmode L_{\odot} \else L$_{\odot}$\fi}
36: \newcommand{\rsun}{\ifmmode R_{\odot} \else R$_{\odot}$\fi}
37:
38: %-----------------------------------------------------------------------
39:
40: %% You can insert a short comment on the title page using the command below.
41:
42: \slugcomment{}
43:
44:
45: \shorttitle{IRS16SW: a massive binary in the GC}
46: \shortauthors{Martins et al.}
47:
48:
49: \begin{document}
50:
51:
52: \title{GCIRS16SW: a massive eclipsing binary in the Galactic Center \footnote{Based on observations collected at the ESO Very Large Telescope (programs 075.B-0547 and 076.B-0259)}}
53:
54:
55: \author{F. Martins \altaffilmark{1}, S. Trippe \altaffilmark{1}, T. Paumard \altaffilmark{1}, T. Ott \altaffilmark{1}, R. Genzel \altaffilmark{1,2}, G. Rauw \altaffilmark{3}, F. Eisenhauer \altaffilmark{1}, S. Gillessen \altaffilmark{1}, H. Maness \altaffilmark{4}, R. Abuter \altaffilmark{5}}
56:
57: \email{martins@mpe.mpg.de}
58:
59:
60: \altaffiltext{1}{MPE, Postfach 1312, D-85741, Garching, Germany}
61: \altaffiltext{2}{Department of Physics, University of California, CA 94720, Berkeley, USA}
62: \altaffiltext{3}{Institut d'Astrophysique et de G\'eophysique, Universit\'e de Li\`ege, All\'ee du 6 Ao\^ut 17, B\^at. B5c, 4000 Li\`ege, Belgium}
63: \altaffiltext{2}{Department of Astronomy, University of California, CA 94720, Berkeley, USA}
64: \altaffiltext{5}{ESO, Karl-Schwarzschild-Str. 2, D-85748 Garching, Germany}
65:
66: \begin{abstract}
67: We report on the spectroscopic monitoring of GCIRS16SW, an Ofpe/WN9
68: star and LBV candidate in the central parsec of the Galaxy. SINFONI
69: observations show strong daily spectroscopic changes in the K
70: band. Radial velocities are derived from the He~{\sc i} $2.112 \mu m$
71: line complex and vary regularly with a period of 19.45 days,
72: indicating that the star is most likely an eclipsing binary. Under
73: various assumptions, we are able to derive a mass of $\sim$ 50 \msun\
74: for each component.
75: \end{abstract}
76:
77: \keywords{Stars: binaries:eclipsing --- Stars: early-type --- Galaxy: center}
78:
79:
80: %%#####################################################################
81: %%------------------------------- Introduction ---------------------
82:
83:
84: \section{Introduction}
85: \label{intro}
86:
87: The central cluster constitutes one of the largest concentrations of
88: massive stars in the Galaxy \citep{genzel03}. Nearly 100 OB and
89: Wolf-Rayet stars are confined in a compact region of radius $\sim$ 0.5
90: parsec centered on the super-massive black hole associated with the
91: radio source SgrA* \citep{paumard06}. Among this population of young
92: massive stars, six are thought to be Luminous Blue Variables
93: (LBV): IRS16NE, IRS16C, IRS16NW, IRS16SW, IRS33E and IRS34W
94: \citep{paumard04,trippe06}. LBVs are evolved massive stars
95: experiencing strong variability in both photometry and spectroscopy
96: due to their proximity to the Humphreys-Davidson limit \citep{hd94}, a
97: region of the HR diagram where the luminosity of the stars reaches the
98: Eddington luminosity so that instabilities develop in their atmospheres,
99: leading to strong mass ejection and drastic changes in the stellar
100: properties (\teff, radius).
101: The six stars mentioned above are only LBV ``candidates'' (LBVc)
102: since they have not been observed to experience the strong outbursts
103: and photometric changes typical of bona fide LBVs such as $\eta$ Car \citep{dh97}. However, their luminosities and
104: spectra are very similar to stars known to be ``quiescent'' LBV,
105: i.e. stars having experienced an LBV event in the past and being now
106: in a more stable phase. In addition, one of them - IRS34W - has shown
107: photometric variability on timescales of months to years which was
108: interpreted as the formation of dust from material previously ejected
109: by an LBV outburst \citep{trippe06}.
110:
111: Among these six stars, IRS16SW deserves special attention. This star was
112: claimed to be a massive eclipsing binary by \citet{ott99} since its K
113: band magnitude displays regular variations with a periodicity of 9.72
114: days. However, the absence of a second eclipse in the light-curve lead
115: \citet{depoy04} to the conclusion that the binary scenario was not
116: correct, and that IRS16SW was instead a pulsating massive star, a
117: class of star predicted by theory but not observed so far.
118:
119: Here, we present results of the spectroscopic monitoring of IRS16SW
120: revealing periodic variations of radial velocities which are
121: interpreted as the signature of a massive spectroscopic and eclipsing
122: binary .
123:
124:
125: \begin{figure*}
126: \epsscale{0.9}
127: \begin{center}
128: \includegraphics[width=8.5cm,height=6.5cm]{f1a.eps}
129: \includegraphics[width=8.5cm,height=6.5cm]{f1b.eps}
130: \end{center}
131: \caption{Montage of spectra of IRS16SW around the region of He~{\sc i}
132: 2.112 $\mu m$ (left) and \brg\ (right) taken between October
133: 2$^{nd}$ and October 12$^{th}$ 2005. The changes in both the line
134: shape and position are clearly seen. \label{var_lines}}
135: \end{figure*}
136:
137:
138:
139:
140: %%#####################################################################
141: %%------------------------------- Observation ---------------------
142:
143: \section{Observations and data reduction}
144: \label{observ}
145:
146: We used SINFONI \citep{frank03} on the ESO/VLT to obtain spectra of
147: IRS16SW. Observations were carried out under seeing limited
148: conditions and were performed on August 28$^{\rm th}$ - September
149: 1$^{\rm st}$, September 4$^{\rm th}$, October 2$^{\rm nd}$, and
150: October 4$^{\rm th}$ - 12$^{\rm th}$ 2005, and March 18$^{\rm th}$ -
151: 21$^{\rm st}$ 2006. In order to get the best spectral resolution
152: available with SINFONI (R = 4000) we restricted ourselves to the K
153: band. Short exposures ($2 \times 60$ seconds) were sufficient to
154: obtain S/N $\sim$ 30. Data reduction was performed as in
155: \citet{frank05}. The final spectra were subsequently carefully
156: extracted from the ``data cubes'' by selecting a circular aperture
157: (radius of 3 pixels) centered on the star and by subtracting from it
158: an annulus of inner (outer) radius 3 (4) pixels. This procedure
159: allowed a good removal of nebular contamination.
160:
161:
162:
163:
164: %%#####################################################################
165: %%------------------------------- Results --------------------------
166:
167: \section{Results}
168: \label{results}
169:
170: %%------------------------------- Spectroscopy / RV ------------------
171: \subsection{Spectroscopic variability and radial velocities}
172: \label{var_rv}
173:
174:
175: Fig.\ \ref{var_lines} shows the variation of the \hei\ and \brg\ lines
176: with time. It is obvious that not only the line shape
177: but also the position of the centroid varies. In order to test
178: the binary scenario, we have derived radial velocities (RV). For that
179: purpose, we have used the line at 2.112 $\mu m$ since it is less
180: affected by wind emission than other lines and is formed closer to the
181: photosphere, allowing a better estimate of the star's motion. This
182: line is however a blend of at least two He lines, and synthetic
183: spectra computed with atmosphere models reveal that the position of
184: the strongest absorption part of the profile can vary by several 100
185: \kms\ around 2.112 $\mu m$ depending on the stellar and wind
186: properties. We have thus adopted this wavelength as our reference,
187: but we stress that the absolute value of the RV may be systematically
188: shifted compared to the real value due to this choice. In practice,
189: we have measured the position of the maximum absorption trough in the
190: line complex and computed the radial velocity from the wavelength
191: shift compared to the adopted reference wavelength. When the line
192: shows a double peak, we have always measured the position of the
193: deepest absorption part of the profile (which also turned out to
194: always be the bluest). This implicitely assumes that if the star is a
195: binary, this absorption is always produced by the same star. Note
196: that when present, the two absorption peaks are separated by the
197: theoretical spacing between the two Helium lines around 2.112 $\mu m$:
198: 170 \kms. Hence we conclude that the second peak most likely
199: comes from the same star.
200: The derived
201: RV are presented in Fig.\ \ref{fit_vrad} and follow very nicely an almost
202: exact cosine curve (see below). This is a strong indication that
203: IRS16SW is a binary star (and justifies a posteriori our method to
204: derive RVs).
205:
206:
207:
208: A period folding analysis applied to the RV curve gives 19.3 $\pm$ 0.4
209: days, in good agreement with the K band light-curve
210: analysis. \citet{depoy04} derived a period of 9.725 $\pm 0.005$ days
211: but argue that if this light-curve was to be produced by a binary
212: star, the absence of second minimum should point to a system composed
213: of two stars with the same K band luminosity, and consequently to a
214: true period of $2 \times 9.725 = 19.45$ days. A re-analysis of the
215: photometric data of \citet{ott99} in view of these new results
216: confirms that a period of 19.447 $\pm 0.011$ days is indeed present in
217: the period folding diagram \citep[for a description of the method,
218: see][]{ott99}. We conclude that IRS16SW is most likely a single line
219: spectroscopic (SB1) and eclipsing binary with an orbital period of
220: 19.45 days.
221:
222:
223:
224: %%------------------------------- Orbital solution -------------------
225: \subsection{Orbital solution and physical parameters}
226: \label{orb_sol}
227:
228: With the RVs in hand, we have performed an orbital
229: solution for an SB1 binary using the method of \citet{rauw00}, which
230: is based on the Wolfe, Horak \& Storer algorithm \citep{wolfe67}. The
231: resulting parameters are given in Table \ref{tab_param_orbit}. Note
232: the small eccentricity justifying that the fit of the RV curve in
233: Fig.\ \ref{fit_vrad} is almost a cosine. The orbital solution also
234: gives the mass function
235:
236: \begin{equation}
237: f(m) = \frac{(M_{2} \sin i)^{3}}{(M_{1}+M_{2})^{2}} = 10.47 M_{\odot}
238: \label{eq_fm}
239: \end{equation}
240:
241: \noindent from which one can estimate the individual masses of each
242: component ($M_{1}$ and $M_{2}$) if one uses in addition
243: Kepler's third law
244:
245: \begin{equation}
246: M_{1}+M_{2} = \frac{4 \pi^{2} r^{3}}{G P^{2}}
247: \label{kepler3}
248: \end{equation}
249:
250: \noindent where $r$ is the separation between the two stars and $P$
251: the period. However, one needs an estimate of 1) the inclination and
252: 2) the separation. As for the latter, the absence of plateau in the K
253: band light-curve indicates that contact is achieved in the binary: as
254: soon as the primary eclipse ends, the secondary eclipse starts. Hence,
255: one can assume that $r$ is simply the sum of the radii of
256: the two components. The value of $a \sin i$ we find (and $a$ itself,
257: $\sin i$ being close to 1, see below) is similar to the stellar radius
258: of LBVc stars derived by \citet{paco97}, so we assume $r = 2 \times
259: a$. In that configuration, the two stars, which we know have similar
260: K magnitude, have similar radii and rotate around each other, the
261: center of mass being the contact point. To get an estimate of the
262: inclination, we have used the software NIGHTFALL \footnote{software
263: developed by R. Wichmann and freely available at the following URL
264: http://www.hs.uni-hamburg.de/DE/Ins/Per/Wichmann/Nightfall.html} to
265: fit the light-curve (see result in Fig.\ \ref{fit_kcurve}). Assuming a
266: similar effective temperature of 28000 K for both components
267: \citep[see][]{paco97} and a light ratio of one, we
268: obtained a reasonable fit with an inclination $i \sim 70 \deg$. Note
269: that we had to adopt Roche lobe filling factors of 1.3 (the maximum
270: allowed value in NIGHTFALL) to correctly reproduce the light-curve. If
271: filling factors lower than one are used, the light-curve can not be
272: reproduced correctly as the ``peaks'' are too broad \citep[a problem
273: encountered by][]{depoy04}. This confirms that contact is achieved and
274: justifies our assumption that the separation is the sum of the stellar
275: radii.
276:
277: Given the limitations on the light-curve fit and the complications in
278: the physics of the star due to contact (mass transfer, departure from
279: sphericity due to gravitational interaction, hot spot in interaction
280: region), we stress that our estimate of the inclination is only
281: indicative. It is however consistent with the value expected for an
282: eclipsing binary. With $i \sim 70 \deg$, and using Eq.\ \ref{eq_fm}
283: and \ref{kepler3}, we finally derive $M_{1} \sim M_{2} \sim 50
284: M_{\odot}$.
285:
286:
287: \begin{figure}
288: \epsscale{0.9}
289: \begin{center}
290: \includegraphics[width=8.5cm,height=6.5cm]{f2.eps}
291: \end{center}
292: \caption{Radial velocity curve of IRS16SW together with the best
293: orbital solution (for P = 19.45 days, solid line). The typical
294: uncertainty on the radial velocity is $\pm$ 20 \kms. Parameters for
295: the best fit solution are given in Table
296: \ref{tab_param_orbit}. \label{fit_vrad}}
297: \end{figure}
298:
299:
300: \begin{figure}
301: \epsscale{0.9}
302: \begin{center}
303: \includegraphics[width=8.5cm,height=6.5cm]{f3.eps}
304: \end{center}
305: \caption{K band light-curve displayed for a period of 19.45 days,
306: together with the best fit (dashed line). Data are from
307: \citet{ott99}. The two eclipses are similar, indicating that both
308: components of the binary have the same K band magnitude. Note also the
309: absence of plateau which suggests that contact is
310: achieved. \label{fit_kcurve}}
311: \end{figure}
312:
313:
314: \begin{table}
315: \begin{center}
316: \caption{Orbital parameters - semi-major axis, eccentricity, systemic
317: velocity, amplitude, longitude of periastron - as derived from the
318: analysis of the radial velocity curve. \label{tab_param_orbit}}
319: \begin{tabular}{l|ll}
320: \tableline
321: $a sini$ [\rsun] & 66.4 $\pm$ 2.2 &\\
322: $e$ & 0.088 $\pm$ 0.023 &\\
323: $v_{0}$ [\kms] & 459.5 $\pm$ 3.6 &\\
324: $K1$ [\kms] & 173.8 $\pm$ 5.5 &\\
325: $\omega$ [$\deg$] & 334.0 $\pm$ 18.3 &\\
326: \tableline
327: \end{tabular}
328: \end{center}
329: \end{table}
330:
331:
332:
333: %%-------------------------- Spectral disentangling -------------------
334: \subsection{Spectral disentangling}
335: \label{spec_disentang}
336:
337: In order to get more insight into the properties of the components of
338: IRS16SW, we have attempted to disentangle the spectra using the method
339: of \citet{gl06}. In practice, the RVs of the primary are used to
340: evaluate an average spectrum in the primary's rest frame. This
341: provides an approximation of the primary spectrum which is then
342: shifted back into the observer's frame and subtracted from the
343: observed spectrum.
344: The residual is composed of the
345: secondary spectrum from which RVs can be estimated. The whole
346: procedure is then re-started inverting the role of the two components,
347: and is iterated until convergence.
348:
349: Due to the limited number of spectral lines and S/N ratio of our
350: spectra, no solution could be found leaving all parameters free. We
351: therefore decided to freeze the primary RVs at their values determined
352: above. Given the results of the orbital solution, we also decided to
353: set the mass ratio to 1.0, and assumed that both components have the
354: same systemic velocities. In that case, convergence could be achieved
355: and the resulting spectra are shown in Fig.\ \ref{fig_disent}. These
356: spectra should be interpreted with caution. Not only they were
357: obtained under the assumption that the mass ratio is 1.0, but the
358: procedure used also implies that the spectra are free of contamination
359: by wind-wind collision or any other interaction in the contact
360: region. With these restrictions in mind, the main qualitative
361: conclusion we draw is that the spectra of both stars, and consequently
362: their properties, are similar and typical of LBV candidates such as
363: the Pistol star \citep{figer99}.
364:
365: \begin{figure}[t]
366: \epsscale{0.9}
367: \begin{center}
368: \includegraphics[width=8.5cm,height=6.5cm]{f4.eps}
369: \end{center}
370: \caption{Spectra of the primary (top) and secondary (bottom)
371: components of IRS16SW as obtained from the spectral disentangling
372: analysis. These spectra should be
373: interpreted with caution since they were obtained under several
374: assumptions (see text). \label{fig_disent}}
375: \end{figure}
376:
377:
378: %%#####################################################################
379: %%------------------------------- Discussion -----------------------
380: \section{Discussion}
381: \label{disc}
382:
383:
384: %%------------------------ Binary vs Pulsating variable --------------
385: \subsection{Binary versus pulsating variable}
386: \label{disc_bin}
387:
388: \citet{depoy04} argue that IRS16SW was a pulsating massive star based
389: on the absence of second minimum in the light-curve and the difficulty
390: to fit this light-curve in the binary scenario (but see Sect.\
391: \ref{orb_sol}). They compare the observed variation in K magnitude to
392: the predictions of the dynamical models of \citet{dg00} and conclude
393: that there is a reasonable qualitative agreement. However, there are
394: some quantitative discrepancies. First, the amplitude of the variation
395: is much larger than predicted: although \citet{dg00} do not compute K
396: band photometry, one can estimate the variation in this band to be at
397: most 0.2 mag (inspection of their Table 2 reveals that the amplitude
398: of photometric variations decreases with wavelength and is $\lesssim
399: 0.2$ mag in the I band), while we observe 0.55 mag. Second, the
400: period we derive $-$ 19.45 days $-$ is larger than expected in the
401: pulsating scenario \citep[see Table 1 of][]{dg00}.
402:
403: Concerning spectroscopic changes, although in principle one can not
404: completely rule out the possibility that they are due to motions of
405: the atmosphere and fluctuations of the physical parameters (\teff,
406: radius) due to pulsations \citep{dg00}, the timescales are again not
407: consistent: $\sim$ 1 day for pulsations compared to 19.45 days
408: observed. Besides, so far there are no theoretical predictions of
409: spectroscopic changes caused by pulsations to which we could compare
410: our observed spectra. The binary nature of IRS16SW is thus strongly
411: favored.
412:
413: The absence of secondary eclipse in the light-curve is
414: explained by the similar K band magnitude of the two components.
415: This is another indication that both stars are very similar.
416: \citet{depoy04} report the presence of a variation in $H-K$ on a
417: period of 9.725 days, $H-K$ being bluer when the system is brighter. A
418: similar trend was observed in the optical photometry of the
419: massive contact binary V606 Centauri \citep{lorenz99}. We interpret
420: this as a sign of heating in the contact zone, making
421: the spectral energy distribution in this region bluer. Massive
422: binaries are indeed known to produce X-rays through colliding winds,
423: and we might expect the same kind of interaction and heating around
424: the contact region. Again, since this region is seen twice during an
425: orbital revolution, an observed period half the true orbital period is
426: naturally derived from the $H-K$ curve.
427:
428:
429:
430: %%------------------------ LBV & stellar evolution --------------
431: \subsection{Stellar evolution and the LBV phenomenon}
432: \label{disc_evol}
433:
434:
435: Whether or not all massive stars go through the LBV phase is still
436: under debate. \citet{langer94} and \citet{pasquali97} argue that this
437: is the case, while other observational \citep{crowther95} and
438: theoretical \citep{mm05} studies indicate that the most massive stars
439: ($M \gtrsim 60 \msun$) may skip this phase. This is an important issue
440: since although short, the LBV phase is crucial in the mass loss
441: history, and consequently in the subsequent evolution, of massive
442: stars. Recent studies by \citet{smith06} even claim that most of the
443: mass of hot stars is lost during the LBV phase. Here, we provide an
444: accurate measurement of the present mass of a candidate LBV, confirming
445: that a star with an initial mass larger than 50 \msun\ (and likely
446: of the order 60-70 \msun) may become a LBV. This is a strong
447: constraint for evolutionary models.
448:
449: Of course, one could argue that the star's evolution was affected by
450: binarity. However, inspection of the spectral morphology and physical
451: properties \citep[see][]{paco97} of IRS16SW and the other LBVc in the
452: Galactic Center shows similarities. Our monitoring of IRS16SW also
453: includes the other LBVc in the Galactic Center. Except for IRS16NE,
454: none of these stars showed any spectroscopic variation, ruling out the
455: possibility that they are close binaries. IRS16NE showed some RV
456: fluctuations \citep[see also][]{tanner06}, but so far we do not have
457: enough data point to sample an hypothetical RV curve. Hence, the
458: similarity between the spectrum of IRS16SW and those of the other
459: single LBV candidates leads us to the conclusion that binarity has not
460: (yet?) significantly affected the evolution of IRS16SW. Since the mass
461: of the two components is similar, one can speculate that IRS16SW is
462: composed of two stars initially equally massive that have so far
463: evolved in parallel in a detached system, without influencing each
464: other. They may have just entered the LBV phase during which contact
465: was achieved due to their respective expansion. This event probably
466: happened very recently (the LBV phase lasting $\sim 10^{5} yr$) so
467: that the general properties of both components have not yet been
468: affected by mass transfer and binary evolution. Such a scenario is
469: consistent with both stars displaying similar spectra (but see Sect.\
470: \ref{spec_disentang} for caution words).
471:
472:
473: Assuming that IRS16SW was not affected by binary evolution, the
474: properties of the LBV candidate in IRS16SW can thus be used to
475: constrain evolutionary models of single massive stars. In their
476: recent models including rotation, \citet{mm05} stressed that the LBV
477: phase is not systematically reached above 45 \msun. Here, we have an
478: example for which it is the case (under the assumption that IRS16SW
479: will turn $-$ or has already turned in the past $-$ into a genuine
480: LBV).
481:
482:
483:
484:
485:
486:
487: %%#####################################################################
488: \acknowledgments
489:
490: We thank all the ESO staff for their help during observations.
491: FM acknowledges support from the Alexander von Humboldt Foundation.
492:
493:
494: %{\it Facilities:} \facility{ESO/VLT}.
495:
496:
497:
498: %%#####################################################################
499: \begin{thebibliography}{}
500:
501: \bibitem[Crowther et al.(1995)]{crowther95} Crowther, P.~A., et al., 1995, \aap, 293, 427
502: \bibitem[DePoy et al.(2004)]{depoy04} DePoy, D.~L., et al., 2004, \apj, 617, 1127
503: \bibitem[Davidson \& Humphreys(1997)]{dh97} Davidson, K., Humphreys, R.~M., 1997, ARA\&A, 35, 1
504: \bibitem[Dorfi \& Gautschy(2000)]{dg00} Dorfi, E.~A., Gautschy, A., 2000, \apj, 545, 982
505: \bibitem[Eisenhauer et al.(2003)]{frank03} Eisenhauer, F., et al., 2003, The Messenger, 113, 17
506: \bibitem[Eisenhauer et al.(2005)]{frank05} Eisenhauer, F., et al., 2005, \apj, 628, 246
507: \bibitem[Figer et al.(1999)]{figer99} Figer, D.F., et al., 1999. \apj, 506, 384
508: \bibitem[Genzel et al.(2003)]{genzel03} Genzel, R., et al., 2003, \apj, 594, 812
509: \bibitem[Gonz\'alez \& Levato(2006)]{gl06} Gonz\'alez, J.F., Levato, H., 2006, \aap, 448, 283
510: \bibitem[Humphreys \& Davidson(1994)]{hd94} Humphreys, R.~M., Davidson, K., 1994, \pasp, 106, 1025
511: \bibitem[Langer et al.(1994)]{langer94} Langer, N., et al., 1004, \aap, 290, 819
512: \bibitem[Lorenz et al.(1999)]{lorenz99} Lorenz, R., et al., 1999, \aap, 345, 531
513: \bibitem[Meynet \& Maeder(2005)]{mm05} Meynet, G., Maeder, A., 2005, \aap, 429, 581
514: \bibitem[Najarro et al.(1997)]{paco97} Najarro, F., et al., 1997, \aap, 325, 700
515: \bibitem[Ott et al.(1999)]{ott99} Ott, T., Eckart, A., Genzel, R., 1999, \apj, 523, 248
516: \bibitem[Pasquali et al.(1997)]{pasquali97} Pasquali, A., et al., 1997, \apj, 478, 340
517: \bibitem[Paumard et al.(2004)]{paumard04} Paumard, T., et al., 2004, Proc. XXXIXth Rencontres de Moriond-La Thuile, Editions Frontieres, Paris, p.\ 377
518: \bibitem[Paumard et al.(2006)]{paumard06} Paumard, T., et al., 2006, \apj, 643, 1011
519: \bibitem[Rauw et al.(2000)]{rauw00} Rauw, G., et al., 2000, \aap, 360, 1003
520: \bibitem[Smith \& Owocki(2006)]{smith06} Smith, N., Owocki, S.P., 2006, ApJL, in press
521: \bibitem[Tanner et al.(2006)]{tanner06} Tanner, A., et al., 2006, \apj, 641, 891
522: \bibitem[Trippe et al.(2006)]{trippe06} Trippe, S., et al., 2006, \aap, 448, 305
523: \bibitem[Wolfe et al.(1967)]{wolfe67} Wolfe, R.~H.~Jr., et al., 1967, in Modern Astrophysics, Gordon \& Breach (New York), p.\ 251
524: \end{thebibliography}
525:
526:
527: \end{document}
528: