1: \documentclass[12pt,preprint]{aastex}
2: %\documentstyle[emulateapj,apjfonts,epsf,versioncopy]{article}
3: %\def\versionnumber{9} % <-- here the version number
4: \def\psr{PSR~J1740$-$5340}
5: \def\com{COM~J1740$-$5340}
6: \newcommand{\gras}[1]{\mbox{\boldmath $#1$}}
7: \def\etal{{et al.~}}
8: \def\ltsima{$\; \buildrel < \over \sim \;$}
9: \def\gtsima{$\; \buildrel > \over \sim \;$}
10: \def\lsim{\lower.5ex\hbox{\ltsima}}
11: \def\gsim{\lower.5ex\hbox{\gtsima}}
12: \def\lapp{\ifmmode\stackrel{<}{_{\sim}}\else$\stackrel{<}{_{\sim}}$\fi}
13: \def\gapp{\ifmmode\stackrel{>}{_{\sim}}\else$\stackrel{<}{_{\sim}}$\fi}
14:
15: \newdimen\minuswidth %define @ width of minus sign for tables
16: \setbox0=\hbox{$-$}
17: \minuswidth=\wd0
18: \catcode`@=\active
19: \def@{\kern\minuswidth}
20: \newdimen\digitwidth %define ! a one digit width for tables
21: \setbox0=\hbox{\rm0}
22: \digitwidth=\wd0
23: \catcode`!=\active
24: \def!{\kern\digitwidth}
25:
26: %\shorttitle{Phase resolved spectroscopy of COM~J1740$-$5340}
27: %\shortauthors{Ferraro et al.}
28:
29: \begin{document}
30:
31: \title{The complex H-$\alpha$ line profile of
32: the bright companion to PSR J1740-5340 in NGC 6397.
33: \footnote{Based on observations collected at the
34: European Southern Observatory, Chile, proposal 69.D-0264}
35: }
36:
37: \author{
38: E. Sabbi\footnote{Dipartimento di Astronomia Universit\`a
39: di Bologna, via Ranzani 1, I--40127 Bologna, Italy,
40: sabbi@bo.astro.it, ferraro@bo.astro.it},
41: R. Gratton
42: \footnote{INAF--Osservatorio Astronomico di Padova,
43: vicolo dell'Osservatorio 5, I--35122 Padova, Italy,
44: gratton@pd.astro.it},
45: F.R. Ferraro\altaffilmark{2},
46: A. Bragaglia\footnote{INAF--Osservatorio Astronomico di Bologna, via
47: Ranzani 1, I--40127 Bologna, Italy, angela@bo.astro.it},
48: A. Possenti\footnote{INAF--Osservatorio Astronomico di Cagliari,
49: Loc. Poggio dei Pini, Strada 54, I--09012 Capoterra, Italy,
50: possenti@ca.astro.it},
51: N. D'Amico\footnote{INAF--Osservatorio Astronomico di Cagliari,
52: Loc. Poggio dei Pini, Strada 54, I--09012 Capoterra, Italy,
53: \& Dipartimento di Fisica, Universit\`a di Cagliari,
54: Strada Provinciale Monserratu--Sestu, km 0.700, I--09042 Monserrato, Italy,
55: damico@ca.astro.it},
56: F. Camilo\footnote{Columbia Astrophysics Laboratory, Columbia
57: University, 550 West 120th Street, New York, NY 10027,
58: fernando@astro.columbia.edu}
59: }
60: \medskip
61:
62:
63: \begin{abstract}
64: We present a detailed study of the H$\alpha$ and He~{\sc i} spectral
65: features of \com~(the companion to \psr~in the Galactic Globular
66: Cluster NGC~6397), exploiting a series of high resolution spectra
67: obtained at different orbital phases. The H$\alpha$ absorption line
68: shows a complex two-component structure, revealing that optically thin
69: hydrogen gas resides outside the Roche lobe of \com. The line
70: morphology precludes the existence of any residual disk around the
71: millisecond pulsar, and suggests the presence of a stream of material
72: going from the companion toward the neutron star. This material never
73: reaches the neutron star surface, being driven back by the pulsar
74: radiation far beyond \com. By analyzing the He~{\sc i} absorption
75: lines as a function of orbital phase, we infer the presence of an
76: overheated longitudinal strip (about 150 times narrower than it is
77: long) on the \com~surface facing the radio pulsar.
78: \end{abstract}
79:
80: \keywords{Globular clusters: individual (NGC~6397)
81: --- stars: individual (COM~J1740$-$5340)
82: --- pulsars: individual (PSR~J1740$-$5340)
83: --- techniques: spectroscopic}
84:
85: \section{Introduction}
86: \label{intro}
87:
88: The nearby Galactic globular cluster NGC~6397 hosts an eclipsing binary
89: millisecond pulsar (\psr; D'Amico et al. 2001a), whose companion (\com) has
90: been optically identified (Ferraro et al. 2001) with a bright (V$\sim$16.6)
91: variable star which has completely filled its Roche lobe (Ferraro et al.
92: 2003). Radio (D'Amico et al. 2001b) and X-ray observations (Grindlay et al.
93: 2001, 2002) show evidence of interaction between the millisecond pulsar (MSP)
94: and mass released from the companion. Interestingly, despite the large pulsar
95: rotational energy loss, no photometric signature of heating can be found in the
96: light curves (Kaluzny et al. 2003, Orosz \& van~Kerkwijk 2003), which are
97: instead dominated by ellipsoidal variations (Orosz \& van~Kerkwijk 2003). The
98: combination of these observational facts represents a fascinating mosaic that
99: has triggered a wealth of hypotheses about the origin and evolution of this
100: binary system (see Possenti 2002; Orosz \& van Kerkwijk 2003; Grindlay et al.
101: 2002 for a review).
102:
103: In the framework of a coordinated spectro-photometric project devoted
104: to investigate the nature of this peculiar object, we recently
105: reported on preliminary results from Very Large Telescope UVES high
106: resolution observations: we measured the mass ratio of the system
107: ($q=M_{\rm PSR}/M_{\rm COM}=5.85\pm 0.13$) with unprecedented
108: precision for a non-relativistic binary pulsar system and confirmed
109: the cool effective temperature ($\sim$5,500 K) of \com~(Ferraro et
110: al. 2003). A radial velocity $V_{\rm bary}=17.7\pm 2.3$ km s$^{-1}$
111: was also derived for the binary barycenter, fully compatible with the
112: systemic radial motion of the cluster (Harris 1996).
113:
114: Here we focus on a detailed analysis of the H$\alpha$ and He~{\sc i}
115: lines: the complex structure of the former is used to investigate the
116: distribution of matter lost by \com, whereas the study of the He~{\sc
117: i} feature aims at confirming the hypothesis that a narrow, hot strip
118: (T$_{eff}>$10,000~K) is located on the pulsar companion surface
119: (Ferraro et al. 2003), and at estimating the area of the heated
120: region.
121:
122: \section{The H$\alpha$ region}
123: \label{Halfa}
124:
125: At radio wavelengths, \psr~displays eclipses for more than 40\% of its orbital
126: period, and evident irregularities of the signal at all orbital phases. The
127: latter characteristic was ascribed by D'Amico et al. (2001b) to the motion of
128: the MSP within a large envelope of matter released by the secondary. This
129: hypothesis finds additional support in the modulated X-ray emission detected by
130: {\it Chandra} (Grindlay et al. 2001, 2002), which may be due to the interaction
131: between the MSP wind and the stellar matter. Furthermore, the anomalous
132: position of \com~in the (V, H$\alpha-$R) color-magnitude diagram (see Fig.~1 of
133: Ferraro et al. 2001 and Taylor et al. 2001) suggests a H$\alpha$ excess:
134: this is a unique feature for a MSP companion and could suggests either
135: RS~CVn--like activity in the companion atmosphere (Orosz \& Van Kerwijk 2003)
136: or the presence of significant interactions between the energetic pulsar flux
137: and the matter lost by \com.
138:
139: Motivated by these observations, we have exploited high quality, phase
140: resolved spectroscopic data to carefully look at the H$\alpha$ line
141: region. A set of 8 high resolution spectra of \com~was obtained with
142: the {\it Ultraviolet-Visual Echelle Spectrograph} (UVES) mounted on
143: the ESO Very Large Telescope on Cerro Paranal (Chile) in 2002
144: June--July. The observational strategy and the data are described in
145: Ferraro et al. (2003; see their Table~1). The system hosting \psr~is
146: located in a crowded region of the cluster; thus, in order to avoid
147: possible contamination from nearby stars, we have here used only
148: spectra acquired in the best seeing conditions, which correspond to
149: orbital phases $\phi \sim$ 0.02, 0.56, 0.78, and 0.36, close to
150: quadratures, opposition and conjunction, respectively ($\phi=0.25$
151: when \com~is at the minimum distance from the observer).
152:
153: As can be seen in Figure 1 (right panel), \com~spectra clearly show
154: H$\alpha$ absorption lines, but the line profiles appear complex
155: and require a detailed analysis. To perform this, the
156: \com~spectra must be compared with a reference template. We took
157: advantage of the availability of spectra observed with a similar UVES
158: configuration in a project devoted to study the chemical abundance in
159: NGC~6397 (Gratton et al. 2001). In particular the high quality spectra
160: of three subgiants (namely \# 669, 793 and 206810 in Gratton et al.)
161: were averaged and convolved with a rotation profile of $V_{\rm
162: rot}=49.6$ km s$^{-1}$ in order to account for the inferred rotation
163: velocity of \com~(Ferraro et al. 2003). The resulting spectrum was
164: then normalized to the \com~continuum at each orbital phase and used
165: as a template. The temperatures of the three subgiants are fully
166: comparable with that derived for \com~($\sim$ 5,500 K, Ferraro et
167: al. 2003; Orosz \& van Kerkwijk 2003) and their overall chemical
168: composition is found compatible with that observed in \com~(although
169: with a few notable exceptions, as will be discussed in Sabbi et
170: al. 2003, in preparation). In fact, the agreement between the template
171: and the \com~spectra in the H$\gamma$ region is excellent (see the
172: left panel of Fig.~1), supporting the similarity of \com~with
173: respect to a normal subgiant branch star of NGC~6397 and hinting at
174: the power of the differential analysis.
175:
176: By using this method, we can easily pinpoint the anomalies of the
177: \com~spectra in the H$\alpha$ region (right panel of Fig.~1): {\it
178: (i)} the line shallowness, which suggests a certain amount of
179: H$\alpha$ emission, and {\it (ii)} the complex structures in the line
180: wings. In order to better analyze these features, we subtracted the
181: template spectrum from the \com~one. The result for the spectrum taken
182: at $\phi=0.56$ is presented in Figure~2, whereas Figure~3 displays the
183: changing structure of the H$\alpha$ emission line along the orbit. At
184: least two different H$\alpha$ emission components can be
185: distinguished. The first one appears as a bright, narrow and symmetric
186: line, clearly visible as the main peak at $\lambda_p\sim 6567$ {\AA}
187: in Figure~2. At all orbital phases the narrow peak is always easily
188: recognizable (see Fig.~3). The second component is much fainter,
189: broader and asymmetric, characterized by a large bump at
190: $\lambda<\lambda_p$ and a long tail at $\lambda>\lambda_p$ in
191: Figure~2. It covers a large velocity range: in the highest quality
192: available spectra (taken at phase $\phi=0.56$), its peak displays a
193: slow radial velocity of $V_{\rm b}\sim 18$ km$^{-1}$ (in the observer
194: frame, see Fig.~2), while the terminal velocity of the tail $V_{\rm
195: t}$ is about 20 times higher. The overall shape of this broad second
196: component appears (specularly reflected with respect to the shape
197: observed at $\phi=0.56$) also at the opposite quadrature (panels $d$
198: and $d\arcmin$ of Fig.~3), whereas near opposition and conjunction
199: only the weakest tail is distinguishable (panels $a$, $a\arcmin$, $c$
200: and $c\arcmin$ of Fig.~3).
201:
202: Since the differential analysis clearly reveals H$\alpha$ emission
203: while excluding any significant H$\gamma$ emission (Fig.~1, left
204: panel), we can conclude that there is a large decrement in Balmer line
205: intensity and hence that the gas emitting the H$\alpha$ photons must
206: be optically thin.
207:
208: \section{Helium lines}
209: \label{He}
210:
211: As first reported by Ferraro et al. (2003), strong He~{\sc i} absorption lines
212: are present in most of the observed spectra, implying that a region of the
213: \com~surface has a high temperature, T$_{eff}>$10,000 K. This feature was
214: unexpected since the other temperature indicators (e.g., photometric colors and
215: H$\alpha$ wings) suggest that the surface temperature is $\sim$ 5,500 K
216: (Ferraro et al. 2001; Orosz \& van Kerkwijk 2003). On the basis of a
217: qualitative inspection of the He~{\sc i} line widths, Ferraro et al. (2003)
218: proposed the existence of a hot strip along the \com~surface.
219:
220: In the aim of better investigating this hypothesis, we carefully compared
221: the shape and the broadening profile of the He~{\sc i} lines with
222: those of other \com~spectral lines. In particular we have built 4
223: empirical rotation line profiles based on many lines in the blue part
224: of the spectrum, and convolved these profiles with that of the He~{\sc
225: i} lines. The comparison with the observed spectrum (Fig.~4) shows
226: that the empirical profiles can nicely reproduce both the Doppler
227: shift and the rotational profiles of the observed He~{\sc i} lines at
228: all the orbital phases. As a consequence, the hot region must be
229: longitudinally extended across the stellar surface. Moreover, at
230: phase $\phi$=0.36 (i.e. when the surface of the companion facing the
231: pulsar is almost invisible to the observer) the He~{\sc i} line
232: appears much fainter than at the other orbital phases.
233: %and its width is smaller than the empirical rotation profile.
234: We can conclude that the hot region does not encircle the entire star, but is
235: primarily located on the companion hemisphere visible from the pulsar.
236:
237: One can estimate the projected area of such a heated region. For this purpose,
238: we assumed a simple model where the atmosphere of the star is composed of two
239: (horizontal) regions: one hot, with temperature roughly similar to that of a B
240: star; and one cool, with the temperature of a G8 star. The ratio between the
241: fluxes emitted by these two regions can be estimated by comparing the residual
242: emission at the center of the Ca~{\sc i}~K line at 3933 \AA~(EW = 4.1\AA)
243: where only light from the cool region is seen and at the center of the He~{\sc
244: i} line at 5878 \AA~(EW = 0.19\AA) where we assume to be observing only
245: light from the hot region. The flux ratios emitted at these two wavelengths by
246: stars of spectral type B7 (T$_{\rm eff}=15,000$ K) and G8 (T$_{\rm eff}=5,500$
247: K) were taken from Gunn \& Stryker (1983). Comparing these ratios with the
248: \com~spectrum, suggests that the heated region should cover less than 1\% of
249: the stellar surface.
250:
251: \section{Discussion}
252:
253: \subsection{H$\alpha$ lines}
254: \label{disH}
255:
256: Figure~3 (left panel) shows the subtracted spectra corrected for the Doppler
257: shift due to the orbital motion of \com~around the MSP: since at each orbital
258: phase the main peak has always the same radial velocity as \com, we can safely
259: conclude that the brightest peak is due to gas emitted by a hot and optically
260: thin (i.e. rarefied) gaseous region close to the star surface,
261: %namely the stellar chromosphere.
262: as often observed in the atmospherically active binaries (e.g. Montes et
263: al. 1997,2000 for a review).
264:
265: The evident asymmetric shape (Fig.~2) of the secondary broad component
266: of the H$\alpha$ emission precludes an origin in a Keplerian disk
267: surrounding \psr. Instead, the synchronous rotation --- modulated by
268: the radial orbital velocity around the center of mass of the system
269: --- of the broad and the narrow components (particularly evident when
270: comparing panels {\it b'} and {\it d'} of Fig.~3) is suggestive of a
271: lump of matter protruding from \com~and roughly orbiting as a rigid
272: body with the companion star. In this picture, the broad peak of the
273: fainter component (corresponding in quadrature to a radial velocity
274: $(V_{\rm b}-V_{\rm bary})\sim 0$ with respect to the center of mass of
275: the binary) would represent hydrogen gas flowing from the Roche lobe
276: of \com~toward the center of mass of the system (located at $\sim
277: 1~{\rm R_\odot}$ from the pulsar), where it is stopped and then swept
278: away by the pulsar energetic flux of radiation and/or accelerated
279: particles (a similar {\it radio-ejection}
280: mechanism has been postulated for this binary system by Burderi, D'Antona \&
281: Burgay 2002). The swept material could form a wide (cometary-like?)
282: tail possibly winding around the binary. The portion of the gaseous
283: tail which is closest to \com~should follow almost rigidly the orbital
284: motion of the companion star, always pointing roughly in the opposite
285: direction (with respect to \com) of the MSP. If it produces the
286: declining end of the broad line component (see Fig. 2), the terminal
287: radial velocity seen at quadratures ($V_{\rm t}\sim 300-400$ km
288: s$^{-1}$) would imply that the swept gas maintains a high enough
289: temperature for emitting H$\alpha$ photons up to $R_{\rm t}=(V_{\rm
290: t}P_{\rm orb}/2\pi-a_{\perp}q)/\sin i \sim (5.2~{\rm R_\odot})/\sin i$
291: outside the orbit of \com, where $P_{\rm orb}$ and $a_{\perp}$ are the
292: orbital period and the projected semimajor axis of the pulsar orbit
293: (D'Amico et al. 2001b), whereas $q$ and $i (\sim 50~\deg$; Ferraro et
294: al. 2003) are the mass ratio and the orbital inclination.
295:
296: Interestingly, $R_{\rm t}$ is of the same order of the radius of the
297: region (concentric with the companion star) causing the long eclipses
298: of \psr~at 1.4 GHz (D'Amico et al. 2001b). Thus the gas from which the
299: H$\alpha$ broad emission originates may be responsible also for the
300: radio eclipses of \psr~(at 1.4 GHz) between orbital phases 0.05
301: and 0.45. The matter blown away from the binary at larger distances
302: than $R_{\rm t}$ is probably cooler and not completely ionized. It
303: would be undetectable at optical wavelengths, but could cause the
304: striking irregularities displayed by the radio signal of \psr~
305: %even at orbital phases far from the nominal eclipse region
306: at nearly all orbital phases, even far from the nominal eclipse region
307: (D'Amico et al. 2001b).
308:
309: \subsection{He~{\sc i} lines}
310:
311: In order to understand the origin of the He~{\sc i} lines, we looked
312: for other systems that show physical and/or spectral
313: similarities. AM~Her binaries are a possibility, since they contain a
314: WD and a MS star, and do not have an accreting disc, since the
315: magnetic field stops its formation; but the spectra of these stars are
316: dominated by high excitation emission lines such as He~{\sc ii} (e.g.,
317: Schwope et al. 1997), while we do not find these features in our
318: spectra. The Herbig~Ae/Be stars, young intermediate mass stars, have
319: similar spectra which show H$\alpha$ emission and He {\sc i}
320: absorption lines (see Waters \& Waelkens 1998 for a review). But even
321: if the origin of the H$\alpha$ lines is similar, the T$_{\rm eff}$ of
322: \com~ is too low to justify the helium. We have to consider other
323: mechanisms to justify the helium presence in our spectra.
324:
325: The detailed inspection of the rotation profile and of the orbital
326: modulation of the He~{\sc i} absorption lines strongly supports the
327: existence of a hot ``barbecue-like'' strip{\footnote{A suggestive
328: definition proposed by Josh Grindlay (private communication).}}
329: elongated across the stellar photosphere facing the pulsar. Assuming
330: for simplicity that a rectangular ribbon encircles $\sim$75\% of the
331: star (in order to accomplish the observed weakening of the He~{\sc i}
332: absorption line near only one of the four orbital epochs) and has an
333: area $\lapp 1$ \% of the star photosphere (as estimated in
334: \S\ref{He}), the ribbon's length would be $\sim 7~{\rm R_\odot}$ and
335: its width $\lapp 0.04~{\rm R_\odot}.$
336:
337: What can cause such a ``stretched feature''? The high temperatures
338: required for producing the He~{\sc i} absorption line could be due to
339: a shocked filament standing, e.g., at the boundary between the
340: companion surface and the gas swept back toward the companion by the
341: pulsar flux. However it seems difficult to maintain a hot filament of
342: gas confined within a few hundredths of ${\rm R_\odot}.$ Another
343: intriguing possibility is that the heated strip is due to the effects
344: of the pulsar irradiation on the companion surface. In this case, the
345: shape of the heated region can be accounted for only by a highly
346: anisotropic pulsar emission pattern confined (at least in the
347: direction orthogonal to the binary orbital plane) within a
348: surprisingly small angle of $\sim 0.4$ degrees. Since such a narrow
349: pencil (or highly flattened fan) beam would probably illuminate the
350: \com~ surface only for a portion of the orbit, the tiny width of the
351: heated region would be preserved if the cooling time of the irradiated
352: gas is much shorter than the orbital timescale $\sim 1$ day. Under
353: basic assumptions (as supposing a perfect gas and an atmospheric
354: density of $10^{-7}$ g cm$^{-3}$), we estimate that only a few hundreds
355: seconds are necessary to cool the gas from 10,000 K to 5,500 K.
356:
357: Finally we note that the occurrence of highly anisotropic pulsar
358: emission would explain why irradiation does not leave any detectable
359: imprint on the photometric curves of \com~presented by Orosz \& van
360: Kerkwijk (2003) and by Kaluzny et al. (2003), without the need for
361: invoking any reduction of the intrinsic spin period derivative $\dot P$ of
362: \psr~with respect to the value calculated using radio timing data
363: (D'Amico et al. 2001b).
364:
365: \acknowledgements{\small We thank the referee Peter Edmonds
366: for his helpful
367: suggestions, which improved the presentation.
368: Financial support to this research has been
369: provided by the Agenzia Spaziale Italiana (ASI) and the {\it Ministero
370: dell'Istruzione, dell'Universit\`a e della Ricerca} (MIUR). FC is
371: supported by the US NSF and NASA.}
372:
373: \begin{thebibliography}{}
374:
375: \bibitem[Burderi, D'Antona \& Burgay 2002]{bdb02}
376: Burderi, L., D'Antona. F. \& Burgay, M. 2002, ApJ, 574, 325
377:
378: \bibitem[D'Amico et al. 2001a]{dlm+01}
379: D'Amico, N., Lyne, A. G., Manchester, R. N., Possenti, A. \& Camilo, F. 2001a,
380: ApJ, 548, L171
381:
382: \bibitem[D'Amico et al. 2001b]{dpm+01}
383: D'Amico, N., Possenti, A., Manchester, R. N., Sarkissian, J., Lyne, A. G.\&
384: Camilo, F. 2001b, ApJ, 561, L89
385:
386: \bibitem[Ferraro et al. 2001]{fpds01}
387: Ferraro, F.~R., Possenti, A., D'Amico, N. \& Sabbi, E. 2001, ApJ, 561, L93
388:
389: \bibitem[Ferraro et al. 2003]{fpds03}
390: Ferraro, F.~R., Sabbi, E., Gratton, R., Possenti, A., D'Amico, N.,
391: Bragaglia, \& A., Camilo, F. 2003, ApJ, 584, L13
392:
393: \bibitem[Gratton et al. 2001]{g+01}
394: Gratton, R. et al. 2001, A\&A, 369, 87
395:
396: \bibitem[Grindlay et al. 2001]{ghemc01}
397: Grindlay, J.~E., Heinke, C.~O., Edmonds, P.~D., Murray, S., \& Cool, A. 2001,
398: ApJ, 563, L53
399:
400: \bibitem[Grindlay et al. 2002]{grindlay02}
401: Grindlay, J.~E., Camilo, F., Heinke, C.~O., Edmonds, P.~D.,
402: Cohn, H., \& Lugger, P. 2002, ApJ, 581, 470
403:
404: \bibitem[]{}
405: Gunn, J.~E. \& Stryker, L.~L. 1983, ApJS, 52, 121
406:
407: \bibitem[Harris 1996]{h96}
408: Harris, W~E. 1996, AJ, 112, 1487
409:
410: \bibitem[Kaluzny et al. 2003]{kaluzny03}
411: Kaluzny, J., Rucinski, S.~M., Thompson, I.~B. 2003, AJ, 125, 1546
412:
413: \bibitem[Montes et al. 2000]{M00}
414: Montes, D., Fern\'andez--Figueroa, M.~S., De Castro, E., Cornide, M.,
415: Latorre, A., \& Sanz--Forcada, J. 2000, A\&AS, 146, 103
416:
417:
418: \bibitem[Montes et al. 1997]{M97}
419: Montes, D., Fern\'andez--Figueroa, M.~S., De Castro, E., \& Sanz--Forcada,
420: J. 1997, A\&AS, 125, 263
421:
422: \bibitem[Orosz \& van Kerkwijk 2003]{ov03}
423: Orosz, J.~A., \& van Kerkwijk, M.~H. 2003, A\&A, 397, 237
424:
425: \bibitem[Possenti 2002]{p02}
426: Possenti, A. 2002, Neutron Stars Pulsars and Supernova Remnants,
427: eds. W. Becker, H. Lesch \& J. Tr\"umper, MPE Report 278, p. 183
428:
429: \bibitem[Schowpe, Mantel \& Horne 1997]{smh97}
430: Schwope, A.~D., Mantel, K.~H., \& Horne, K. 1997, A\&, 319, 894
431:
432: \bibitem[Taylor et al. 2001]{t2001}
433: Taylor, J.~M., Grindlay, J.~E., Edmonds, P.~D., Cool, A.~M. 2001, ApJ, 553,
434: L169
435:
436: \bibitem[Waters \& Waelkens 1998]{ww98}
437: Waters, L.~B.~F. \& Waelkens, C. 1998, ARAA, 36, 233
438:
439: \end{thebibliography}
440:
441: \clearpage
442:
443: \begin{figure}
444: \plotone{f1.ps}
445: \caption {
446: The observed spectra ({\it light solid lines}) of \com~in the H$\gamma$ ({\it
447: left panel}) and H$\alpha$ ({\it right panel}) regions at two orbital phases
448: ($\phi$=0.02 and $\phi$=0.56, respectively. We have applied the
449: same convention of D'Amico et al. 2001b, in which \com~is nearest to the
450: observer at $\phi$=0.25). Template spectra (heavy solid lines), obtained by
451: averaging the spectra of three normal subgiants of NGC~6397 (Gratton et al.
452: 2001), are overplotted for comparison: note the excellent agreement between the
453: spectra near H$\gamma$; conversely the H$\alpha$ absorption in \com~is clearly
454: shallower and shows anomalous wings, suggesting the existence of a complex
455: emitting structure.} \end{figure}
456:
457: \clearpage
458:
459: \begin{figure}
460: \plotone{f2.ps}
461: \caption{\label{Hemisdet}
462: Subtracted spectrum of \com~(see text for description of the adopted
463: procedure) in the H$\alpha$ region at orbital phase $\phi=0.56$. No
464: wavelength shift has been applied to the spectrum. The {\it vertical dashed
465: line} marks the H$\alpha$ line rest wavelength and the {\it horizontal dashed
466: line} is the continuum level. Two components can be seen: the narrower and
467: brighter (modeled by the {\it heavy dashed line}) is emitted by the stellar
468: chromosphere (see \S\ref{disH}) while the broader and fainter (modeled by the
469: {\it heavy solid line}) shows a peak (here named $V_b$) at smaller velocities
470: than the star, but extending farther, with the velocity of its tail ($V_t$)
471: greater than 300 km s$^{-1}$.}
472: \end{figure}
473:
474: \clearpage
475:
476: \begin{figure}
477: \plotone{f3.ps}
478: \caption{\label{Hemis1}
479: Subtracted spectra of \com~at 4 orbital phases. In the {\it left panel} the
480: Doppler shifts corresponding to the orbital motion of \com~(Ferraro et al.
481: 2003) are applied, while in the {\it right panel} a Doppler shift
482: corresponding to the center of gravity radial velocity has been applied to
483: each spectrum. The spectra are arbitrarily shifted in intensity for the sake
484: of clarity. The {\it horizontal dashed lines} are the continuum levels, while
485: {\it the vertical dashed lines} represent the H$\alpha$ line rest wavelength.
486: Note that in the left panel the main peak, at each phase, has the same radial
487: velocity as \com, denoting its chromospheric origin.}
488: \end{figure}
489:
490: \clearpage
491:
492: \begin{figure}
493: \plotone{f4.ps}
494: \caption{\label{He-line}
495: The heavy solid lines overplotted on the observed spectra are
496: empirical profiles obtained by combining the typical rotation
497: broadening profiles with the He {\sc i} line intensity. An arbitrary
498: shift in intensity is applied to the spectra for the sake of clarity.}
499: \end{figure}
500:
501: \clearpage
502:
503:
504: \end{document}
505: