1: \documentclass[]{article}
2: \usepackage{emulateapj}
3: \usepackage{graphicx}
4:
5: \advance \voffset by -0.50cm\relax
6: \def\msun{{\rm ~M}_{\odot}}
7: \def\rsun{{\rm ~R}_{\odot}}
8: \def\myr{{\rm ~Myr}}
9: \def\mdot{\dot M}
10: \def\mpy{{\rm ~M}_{\odot} {\rm ~yr}^{-1}}
11:
12:
13: \begin{document}
14:
15: \title{The most massive progenitors of neutron stars: CXO J164710.2-455216}
16:
17: \author{Krzysztof Belczynski\altaffilmark{1,2,3},
18: Ronald E.\ Taam\altaffilmark{4,5}}
19:
20:
21: \affil{
22: $^{1}$ Los Alamos National Laboratory,
23: CCS2/ISR1 Group, P.O. Box 1663, MS D466,
24: Los Alamos, NM 87545\\
25: $^{2}$ Oppenheimer Fellow\\
26: $^{3}$ New Mexico State University, Dept of Astronomy,
27: 1320 Frenger Mall, Las Cruces, NM 88003\\
28: $^{4}$ Northwestern University, Dept of Physics \& Astronomy,
29: 2145 Sheridan Rd, Evanston, IL 60208\\
30: $^{5}$ ASIAA/National Tsing Hua University - TIARA, Hsinchu, Taiwan\\
31: kbelczyn@nmsu.edu, r-taam@northwestern.edu}
32:
33: \begin{abstract}
34: The evolution leading to the formation of a neutron star in the very young Westerlund 1
35: star cluster is investigated. The turnoff mass has been estimated to be $\sim 35 \msun$,
36: indicating a cluster age $\sim 3-5$ Myr. The brightest X-ray source in the cluster,
37: CXO J164710.2-455216, is a slowly spinning ($10$ s) single neutron star and potentially a
38: magnetar. Since this source was argued to be a member of the cluster, the neutron star
39: progenitor must have been very massive ($M_{\rm zams} \gtrsim 40 \msun$) as noted by Muno
40: et al. (2006). Since such massive stars are generally believed to form black holes (rather
41: than neutron stars), the existence of this object poses a challenge for understanding
42: massive star evolution. We point out while single star progenitors below $M_{\rm zams}
43: \lesssim 20 \msun$ form neutron stars, binary evolution completely changes the progenitor
44: mass range. In particular, we demonstrate that mass loss in Roche lobe overflow enables
45: stars as massive as $50-80 \msun$, under favorable conditions, to form neutron stars. If
46: the very high observed binary fraction of massive stars in Westerlund 1 ($\gtrsim 70\%$) is
47: considered, it is natural that CXO J164710.2-455216 was formed in a binary which was
48: disrupted in a supernova explosion such that it is now found as a single neutron star. Hence,
49: the existence of a neutron star in a given stellar population does not necessarily place
50: stringent constraints on progenitor mass when binary interactions are considered. It is
51: concluded that the existence of a neutron star in Westerlund 1 cluster is fully consistent
52: with the generally accepted framework of stellar evolution.
53: \end{abstract}
54:
55:
56: \keywords{stars: evolution -- binaries: close -- stars: neutron}
57:
58:
59: \section{Introduction}
60:
61: The recent detection of the variable X-ray source, CXO J164710.2-455216, located 1.'6 from
62: the core of the young open cluster, Westerlund 1, by Muno et al (2006) can be considered key to
63: understanding the
64: formation history and nature of compact objects in young massive stellar clusters. CXO
65: J164710.2-455216 is the brightest X-ray source in the cluster, radiating at a luminosity of
66: $\sim 3 \times 10^{33} (D/5 kpc)^2$ ergs s$^{-1}$ in the 0.5-8 keV energy band and
67: characterized by a pulse period of 10.6 s. Based on subsequent XMM-Newton observations,
68: Israel et al. (2007) and Muno et al. (2007) showed that the source increased its luminosity
69: to $\sim 10^{35} (D/5 kpc)^2$ ergs s$^{-1}$ and exhibited a rapid spin down after an outburst,
70: exhibiting properties that are very reminiscent of highly magnetized neutron stars known as
71: magnetars.
72:
73: Previous spectroscopic and photometric observations of Westerlund 1 were carried out by Clark
74: et al. (2005), who showed that the population of massive stars could be understood if the cluster
75: age was $4 \pm 1$ Myr (see also Crowther et al. 2006), a result which is also consistent with
76: a more recent age determination
77: based on an analysis of its intermediate and low mass stellar content (Brandner et al. 2008).
78: For an assumed solar metallicity, a lower limit for the progenitor mass of CXO J164710.2-455216
79: was estimated to be $\sim 35 \msun$. The reconstruction of the evolutionary history of CXO
80: J164710.2-455216 is highly desirable in placing limits on its progenitor, however, it is
81: dependent upon whether it evolved in isolation as a single star or as a member of an interacting
82: binary system. We note that the recent studies by Clark et al. (2008) indicate a high binary
83: fraction, $\gtrsim 70\%$, for the massive Wolf Rayet star population in Westerlund 1 and that a
84: binary evolutionary channel may be very likely.
85:
86: It is generally inferred from core collapse simulations that single stars more massive
87: than $\sim 20-25 \msun$ form black holes (e.g., Fryer 1999). Stars
88: more massive than $\sim 20-25 \msun$ in binaries are also expected to form black holes
89: in order to explain the Galactic black hole transient systems and to satisfy nucleosynthetic
90: constraints (e.g., Maeder 1992; Portegies Zwart, Verbunt \& Ergma 1997; Kobulnicky \& Skillman
91: 1997; Ergma \& van den Heuvel 1998). However, a number of studies have been pointed out
92: that binary interactions
93: can significantly increase the neutron star/black hole mass formation limit. For example,
94: Van den Heuvel \& Habets (1984) studied the binary LMC X-3 containing a neutron star with a
95: massive companion, and found that stars as massive as $40 \msun$ may form neutron stars.
96: Kaper et al. (1995) further increased this limit, estimating a mass of $50 \msun$ based on the
97: analysis of a binary Wray 977, although it was later demonstrated that such a system could be
98: explained with the much lower mass limit of $26 \msun$ (Wellstein \& Langer 1999). However,
99: Wellstein \& Langer (1999) reemphasized that, in general, the critical initial mass limits for
100: neutron star/black hole formation in binary systems may be quite different than for single
101: stars and can be as high as $100 \msun$. We point out that, in this regard, Brown et al.
102: (2001) have noted that the final mass of the compact object may depend upon whether the evolved
103: core was exposed before or after central helium burning. More recently, Fryer et al. (2002)
104: demonstrated that a star of $60 \msun$ that was stripped of its envelope in a binary system
105: can end up as a neutron star or a massive black hole, depending on the adopted wind mass loss
106: rates during the Wolf-Rayet stage. These studies, taken as a whole, indicate that binary
107: interactions can lead to a range of initial stellar masses separating the formation of neutron stars
108: from black holes rather than a single limiting mass. We use this concept to examine the
109: existence of a neutron star in the very young environment of Westerlund 1.
110:
111: In this paper, we report on the formation history of neutron stars from very massive progenitors
112: ($\gtrsim 35 \msun$) both as a single star and as a component of an interacting binary system
113: to determine the conditions under which a neutron star in a young cluster such as Westerlund 1
114: can form. In the next section, we outline the basic assumptions and input parameters of our
115: numerical model as based on the {\tt StarTrack} binary population synthesis code (Belczynski et al.
116: 2008). The numerical results are presented for the binary star and single star evolutionary channels
117: in \S 3. The implications of these results for stellar and binary evolution of
118: massive stars, as applied to Westerlund 1, are discussed in the last section.
119:
120:
121:
122: \section{Population Synthesis Model}
123:
124: Our population synthesis code, {\tt StarTrack}, was initially developed to
125: study double compact object mergers in the context of $\gamma$ ray burst progenitors
126: (Belczynski, Bulik \& Rudak 2002b) and gravitational-wave inspiral sources
127: (Belczynski, Kalogera, \& Bulik 2002a). In recent years {\tt
128: StarTrack} has undergone major updates and revisions in the physical treatment
129: of various binary evolution phases, and especially the mass transfer phases.
130: The new version has already been tested and calibrated against observations and
131: detailed binary mass transfer calculations (Belczynski et al.\ 2008). It has
132: been used in various applications (e.g., Belczynski \& Taam 2004; Belczynski et
133: al.\ 2004; Belczynski, Bulik \& Ruiter 2005; Belczynski et al. 2006; Belczynski
134: et al.\ 2007). The physics updates that are most important for compact object
135: formation and evolution include: a full numerical approach for the orbital evolution
136: due to tidal interactions, calibrated using high mass X-ray binaries and open
137: cluster observations, a detailed treatment of mass transfer episodes fully
138: calibrated against detailed calculations with a stellar evolution code, updated
139: stellar winds for massive stars, and the latest determination of the natal
140: kick velocity distribution for neutron stars (Hobbs et al.\ 2005). For helium
141: star evolution, which is of a crucial importance for the formation of double
142: neutron star binaries (e.g., Ivanova et al.\ 2003; Dewi \& Pols 2003), we
143: have applied a treatment matching closely the results of detailed evolutionary
144: calculations. If the helium star fills its Roche lobe, the systems are examined
145: for the potential development of a dynamical instability, in which case they
146: are evolved through a common envelope (CE) phase, otherwise a highly non-conservative mass
147: transfer ensues. We treat CE events using the energy formalism (Webbink 1984),
148: where the binding energy of the envelope is determined from a set of He star
149: models calculated with the detailed evolutionary code by Ivanova et al.\ (2003).
150: In case the CE is initiated by a star crossing the Hertzsprung gap (HG) we
151: assume a merger and abort further binary evolution. This is due to the fact
152: that there is no clear core-envelope boundary (and no entropy jump as for
153: more evolved stars) in the interior structure of HG donors to facilitate the
154: formation of a remnant binary system. As a consequence, a large decrease in
155: the formation efficiency of close double compact binaries results (Belczynski
156: et al. 2007). Rejuvenation is considered in detail following the method presented
157: by Tout et al. (1997). For a detailed description of the revised code we refer
158: the reader to Belczynski et al.\ (2008).
159:
160:
161: \section{Results}
162:
163: \subsection{Binary star progenitor}
164:
165: We have calculated the evolution of $2 \times 10^6$ massive binaries with our
166: standard evolutionary model (Belczynski et al. 2008) for solar metallicity
167: ($Z=0.02$) and an initial mass function (IMF) with an exponent of $-2.7$ for primaries
168: in the mass range $6 < M_{\rm zams,1} < 150 \msun$. The secondaries ($4 < M_{\rm zams,2} < 150
169: \msun$) were chosen from a flat mass ratio ($q=M_{\rm zams,2}/M_{\rm zams,1}$)
170: distribution. The initial separations were chosen from the distribution flat in
171: logarithm (i.e., $\propto 1/a$) limited by the maximum separation of $10^5
172: \rsun$, while eccentricities were chosen from a thermal-equilibrium distribution
173: ($2e$).
174:
175: As a result we have obtained populations of compact objects, i.e., white
176: dwarfs, neutron stars and black holes, both single and in binaries. Single
177: compact objects were formed upon disruption of a given binary during a supernova
178: explosion either through a natal kick, mass loss or both. We have used the
179: kick distribution presented by Hobbs et al. (2005) represented by a single
180: Maxwellian with $\sigma=265$ km s$^{-1}$. We employed the above distribution
181: in its original form for neutron stars, while the kick velocity was decreased
182: for black holes due to the fall back (e.g., Fryer \& Kalogera 2001).
183:
184: The stellar models employed in our population synthesis (Hurley, Pols \&
185: Tout 2000) result in an age of Westerlund 1 of 5 Myr: this is a main
186: sequence lifetime of a $35 \msun$ star. Actual fits of single stellar models
187: to over 50 massive stars in Westerlund 1 resulted in an age estimate of
188: $4 \pm 1$ Myr (e.g., Clark et al. 2005; Brandner et al. 2007). We adopt an age of
189: 5 Myr, although we note that any estimate based on single star models is highly
190: uncertain in the case of a cluster like Westerlund 1 with the very high content
191: of binaries ($\gtrsim 70\%$; Clark et al. 2008). At 5 Myr our calculations reveal
192: a population of single neutron stars that originate from disrupted binaries.
193:
194: {\em Evolution. }
195: In the following we briefly describe a typical evolution leading to the formation
196: of a very young solitary pulsar that originates from a disrupted binary. The evolution
197: starts on the ZAMS with two massive stars: $M_{\rm zams,1}=65 \msun$
198: and $M_{\rm zams,2}=44 \msun$, on a rather close $a=220 \rsun$ and eccentric
199: orbit: $e=0.65$. The primary initiates Roche lobe overflow (RLOF) at $t=3.3$ Myr while still
200: on the main sequence (main sequence lifetime of an unaffected single star of that mass is
201: $\tau_{\rm ms}=3.9$ Myr, and it forms a He core of $M_{\rm He}=18 \msun$).
202: At the onset of mass transfer (Case A) the orbit is circularized ($a=80 \rsun$, $e=0$) and
203: the component masses (depleted by stellar winds) are $M_{\rm 1}=52 \msun$ and
204: $M_{\rm 2}=41 \msun$. The mass transfer is stable and proceeds on the nuclear
205: timescale of the primary with rates ranging from $2 - 8 \times 10^{-5} \mpy$.
206: We assume non-conservative evolution in which half of the mass lost by primary
207: is accreted onto secondary, while the other half leaves the system, carrying away
208: orbital angular momentum. At the end of the primary main sequence phase we note the mass
209: ratio has reversed ($M_{\rm 1}=31 \msun$ and $M_{\rm 2}=49 \msun$) and the orbit has
210: expanded ($a=100 \rsun$). The significant mass loss (RLOF) on the main sequence leads to
211: the decrease of the central temperature, reducing the rate of hydrogen (H) burning in the core
212: of the primary. As a result, the primary forms a smaller core at the end of the main
213: sequence phase (new $M_{\rm He}=10 \msun$) and, thus, will form a lower mass compact object.
214: Although the lifetime of the primary star increases with the decrease of its core mass,
215: the fact that the mass transfer started late in the main sequence phase does not
216: significantly affect the lifetime of the star (new $\tau_{\rm ms}=4.0$ Myr).
217:
218: The mass transfer continues while the primary is crossing the HG (Case B), transferring mass
219: at a high rate $\sim 10^{-2} \mpy$ such that the entire H-rich envelope is removed and
220: the primary becomes a massive naked helium star with $M_1=10 \msun$. The secondary, on the
221: other hand, remains on the main sequence, but with its mass increased to $M_2=60 \msun$.
222: Mass transfer ceases as massive helium stars are characterized by small radii ($R_1=0.9
223: \rsun$ for the primary). Thereafter, the primary evolves through the helium star phase,
224: losing more mass via stellar winds and finally exploding in a Type Ib supernova. At the
225: time of explosion ($t=4.8 Myr$) the primary has a mass of $M_1=6.6 \msun$ with core mass
226: of $4.9 \msun$. Such a star forms a massive neutron star ($M_{\rm ns}=1.9 \msun$) as little
227: matter is expected to fall back in such a case (e.g., Fryer \& Kalogera 2001). The explosion
228: disrupts the system provided that either a high natal kick or a kick off the orbital plane
229: is applied, releasing the neutron star from the binary. The massive secondary soon ($t=6.1$ Myr)
230: completes its evolution and forms a solitary black hole of $M_{\rm bh}=7.9 \msun$.
231:
232:
233:
234: {\em Physical properties. }
235: In Figure~1 we show the initial mass distribution of very young pulsar progenitors. We note
236: that the majority of the progenitors are massive $M_{\rm zams}=50-80 \msun$ since high initial
237: masses are required in the framework of the above formation channel. Stars must be sufficiently
238: massive to promptly complete their evolution within 5 Myr, but not too massive for otherwise
239: black hole formation occurs.
240:
241: The mass distribution of the very young solitary pulsars formed at $t_{\rm form} \lesssim 5$
242: Myr is also illustrated in Figure~1. As expected these pulsars are very massive: $M_{\rm
243: ns} = 1.8-2.5 \msun$. The high mass limit reflects our assumption on the maximum neutron
244: star mass ($M_{\rm ns,max}=2.5 \msun$), whereas the peak at $M_{\rm ns}=1.8 \msun$
245: originates from a wide range of progenitor masses. Specifically, our model for single
246: stars has approximately a bimodal distribution of neutron stars masses;
247: single stars with $M_{\rm zams}=8-18 \msun$ form $M_{\rm ns}=1.3 \msun$,
248: while stars with $M_{\rm zams}=18-21 \msun$ form $M_{\rm ns}=1.8 \msun$.
249: Such a distribution was adopted following the study of Timmes, Woosley \&
250: Weaver (1996), which showed bimodality in the core mass of massive stars reflecting
251: the differences in the entropy of their core regions. \footnotetext{All
252: details of the adopted scheme for neutron star mass calculation are given
253: in Belczynski et al. (2008).} Therefore, since our scenario preferentially
254: selects the most massive stars that form neutron stars the peak exists
255: at $M_{\rm ns}=1.8 \msun$, while there are virtually no neutron stars with $M_{\rm
256: ns}=1.3 \msun$.
257:
258: In Figure~2 the cumulative distribution of neutron star formation times is
259: displayed with solitary neutron stars originating from binaries denoted
260: by the solid line. Here, the fraction of neutron stars that have formed before
261: a given time is shown. Neutron star formation begins at $t_{\rm form} \sim 4.3$
262: Myr, and is complete by $t_{\rm form} \sim 60$ Myr. The early phase of neutron star formation
263: from very massive stars proceeded through the mass loss associated with the Roche
264: lobe overflow episodes (as described above), while the neutron stars with the
265: longest formation times result from the evolution of intermediate mass stars
266: either through electron capture supernovae (e.g. Podsiadlowski et al. 2004) or via
267: rejuvenation. For the adopted age of Westerlund 1 $t=5$ Myr we find that the
268: fraction of neutron stars formed is $f_{\rm form}=0.001$, implying that for $1000$
269: neutrons formed via disrupted binaries, one would form within the first $5$ Myr. Since
270: {\em (i)} most of the massive stars in Westerlund 1 appear to reside in binaries and
271: {\em (ii)} most of the massive binaries ($\sim 95\%$; Belczynski, Lorimer \& Ridlay 2008)
272: are disrupted in supernovae explosions, one requires at least $1000$ massive stars
273: ($\gtrsim 8 \msun$) in the Westerlund 1 cluster to form one very young pulsar.
274: One should note the steepness of the cumulative curve for the times close to the
275: age of Westerlund 1, where the fraction changes by more than two orders of magnitude
276: from $t_{\rm form}=4.6$ Myr ($f_{\rm form}=0.0001$) to $t_{\rm form}=5.9$
277: ($f_{\rm form}=0.01$).
278:
279: There are about $\sim 10^5$ stars in Westerlund 1 and only stars with masses
280: exceeding $2 \msun$ have had sufficient time to contract onto the main sequence.
281: If we adopt a power-law IMF with an exponent of $-2.7$, there are about $10^4$
282: stars more massive than $\gtrsim 8 \msun$ that can potentially form a neutron star. If
283: these $10^4$ massive stars are required to form the very young pulsar as observed
284: in Westerlund 1 it would yield a fractional efficiency of 0.0001, corresponding
285: to $f_{\rm form}$ for a cluster age of 4.6 Myr. It should be noted that this approximate
286: estimate is not intended to provide a calibrated rate/efficiency of young pulsar
287: formation, but only to illustrate the an order of magnitude consistency of our results
288: with the particular observation of one object.
289:
290: \subsection{Single star progenitor}
291:
292: The possibility that a primordial single star forms a very young pulsar has also been
293: explored. The calculations were performed with the {\tt StarTrack} code for solar
294: metallicity and with standard wind mass loss rates as described in Belczynski et al. (2008).
295: Within the framework of our assumptions regarding the relation between initial and final
296: compact object masses (Belczynski et al. 2008), stars within the mass range of $M_{\rm zams}
297: \sim 8-21 \msun$ can form neutron stars if the maximum neutron star mass is taken to be $2.5
298: \msun$. The minimum formation time of a neutron star is $t_{\rm form}=9.4$ Myr for the most massive
299: progenitor ($\sim 21 \msun$) and is longer by nearly a factor of 2 than the time required
300: to produce a pulsar in Westerlund 1. This is shown in Figure~2 where the cumulative
301: distribution of the neutron star formation times for primordial single stars (dashed line)
302: is also shown.
303:
304: Adopting a higher maximum neutron star mass of $3.0 \msun$ results in a slight decrease of
305: the minimum formation time to $t_{\rm form}=9.0$ Myr, corresponding to the final evolution of a $M_{\rm
306: zams}=21.5 \msun$ star. This is due to the fact that our predicted initial-final mass relation
307: (see Fig.~1 of Belczynski et al. 2008) steeply rises with an initial mass above $M_{\rm zams}=20
308: \msun$ as expected for the increased amount of fall back during compact object formation.
309:
310: A further decrease in the minimum time for formation of a neutron star could be accomplished by
311: increasing the metallicity of the stars in Westerlund 1. For example, given $M_{\rm ns,max}=3.0
312: \msun$ and allowing for a metallicity of $Z=1.5 \times Z_\odot=0.03$ as may be expected for a
313: young cluster, the minimum time would decrease further to $t_{\rm form}=7.7$ Myr
314: ($M_{\rm zams}=24.0 \msun$), but still significantly high for the age of Westerlund 1.
315:
316: The only remaining alternative (within the framework of our adopted stellar models) is to
317: increase the stellar wind mass loss rates. In this case, the mass loss rates may reduce the
318: stellar mass sufficiently that even initially very massive stars form neutron stars rather than
319: black holes. To examine this possibility, we have introduced two scaling factors to multiply our
320: standard wind mass loss rates. Specifically, the factors $f_{\rm wind,H}$ and $f_{\rm wind,He}$
321: are used to scale winds of stars with H-rich envelopes and naked helium stars, respectively. In
322: all the calculations presented herein, $f_{\rm wind,H}=f_{\rm wind,He}=1$.
323: We follow the Hurley et al. (2000) scheme and the choice of wind mass loss rates.
324: In particular, the rates from Kudritzki \& Reimers (1978) and Iben \& Renzini (1993) are
325: used for giant branch and further evolution; Vassiliadis \& Wood (1993) are
326: applied to asymptotic giant branch; Nieuwenhuijzen \& de Jager (1990) are
327: used for massive stars, but modified for a metallicity dependence ($\propto
328: (Z/Z_\odot)^{1/2}$); Hamman \& Koesterke (1998) is employed for Wolf-Rayet-like
329: winds; Humphreys \& Davidson (1994) rates are used for luminous blue
330: variables ($L>6 \times 10^5$ and $R \times \sqrt{L}>10^5$; where $L,\ R$ are
331: luminosity and radius expressed in solar units).
332:
333: In Figure~3 the initial-final mass relation for single star models with various wind mass loss
334: rates for stars with H-rich envelopes: $f_{\rm wind,H}= 1,\ 2,\ 3$ are presented. We have
335: chosen models with high metallicity ($Z=0.03$) and fixed wind mass loss rates for naked helium
336: stars $f_{\rm wind,He}=1$. Since the increased winds reduce the final compact object mass we
337: note the change in mass of the most massive star that can still form neutron star. For $M_{\rm
338: ns,max}=3.0 \msun$ it is found that only stars with $M_{\rm zams} \lesssim 24,\ 40,\ 62 \msun$
339: can form neutron stars for $f_{\rm wind,H}=1,\ 2,\ 3$, respectively. It is also found that only
340: stars more massive than $M_{\rm zams} \gtrsim 44,\ 50,\ 58 \msun$ can form compact objects within 5
341: Myr. Both constraints can only be satisfied for the model with very high winds: $f_{\rm wind,H}=3$.
342: Therefore if single star is to produce a very young pulsar the stellar winds would need to be
343: increased by factor of $\sim 3$ or higher.
344:
345: We find that such a drastic increase is unrealistic as it would prohibit formation of massive
346: black holes over $\sim 4.5 \msun$ for a high metallicity and over $\sim 6.5 \msun$ for solar
347: metallicity. However, a few black holes in Galactic X-ray transient sources have dynamical
348: mass estimates exceeding $10 \msun$ (e.g., Orosz 2003; Casares 2006). In addition, recent
349: observations indicate that stellar black holes can form with even higher masses in low
350: metallicity environments; $\sim 16 \msun$ black hole in M33 (Orosz et al. 2007) and $\sim
351: 23 \msun$ black hole in IC10 (Prestwich et al. 2007; see also Silverman \& Filipenko 2008
352: for an improved orbital estimate). Although it is expected that black holes form with higher
353: mass in lower metallicity environments, it was recently demonstrated that a reduction in
354: the wind loss rate was necessary for the formation of such massive black holes (Bulik,
355: Belczynski \& Prestwich 2008).
356:
357: We have only investigated the increase of winds for stars with H-rich envelopes, as it has been
358: argued that stellar wind mass loss rates for massive helium stars (i.e., Wolf-Rayet stars) are
359: systematically overestimated due to the ``clumpiness'' of mass outflows from these stars (e.g.,
360: Hamann \& Koesterke 1998; Nugis \& Lamers 2000). The reduction due to the clumpiness is already
361: included in our study, but a further decrease (factor of $\sim 2$) was recently suggested by
362: Vanbeveren, Van Bever \& Belkus (2007), thereby, making it more unlikely that enhanced wind
363: loss rates from single stars are responsible for the formation of neutron stars in Westerlund 1.
364:
365: \section{Discussion}
366:
367: A population synthesis study of the massive star population for Westerlund 1 has been carried
368: out to examine the possible formation history its neutron star member, CXO J164710.2-45526.
369: The young age of the cluster ($4 \pm 1$ Myr), corresponding to a star of $35 \msun$ at the
370: turn off of the main sequence places important constraints on its prior history. It is
371: found that an isolated neutron star can be formed as a result of the disruption of a binary
372: system during a supernova phase (within 5 Myr) provided that the initial progenitor system
373: is characterized by primary masses in the range of $50-80 \msun$. On the other hand, only
374: stars of mass $\sim 60 \msun$, for single star models can produce a neutron star within
375: 5 Myr, however, the rate of mass loss via stellar winds must be enhanced to the extent that
376: the formation of black holes with masses exceeding $10 \msun$ becomes problematic.
377:
378: A major feature in the binary scenario of neutron star formation is the significant
379: mass loss taking place in Case A RLOF. In this mass transfer phase, a much smaller He core
380: ($M_{\rm He} \sim 10 \msun$) is formed at the end of the main sequence phase as compared to the
381: isolated evolution of the same star ($M_{\rm He} \sim 20 \msun$). The removal of the H-rich
382: envelope through the ongoing RLOF stage before central He-ignition (Case B) leads to the
383: onset of strong Wolf-Rayet type winds, preventing an increase of the He core mass via H shell
384: burning. Although we employ
385: revised (decreased to account for ``clumpiness'') wind mass loss rates for naked helium
386: stars (e.g., Hamann \& Koesterke 1998), the Wolf-Rayet phase leads to a further loss of $\sim 3
387: \msun$ from the primary star. This is a secondary, but nevertheless important effect
388: leading to the formation of very young neutron stars. An additional factor enhancing the
389: formation of neutron stars from very massive binary components was pointed out by Brown et
390: al. (2001; and references within), who argued that if the H-rich envelope is removed (e.g.,
391: in case B or early case C RLOF) and H shell burning is extinguished before the end of central
392: He burning the resulting Fe core mass is reduced in comparison to the case where the H-envelope
393: is not removed. Hence, the formation of a low mass compact object (i.e., a neutron star rather
394: than a black hole) could be enhanced. This finding was connected to the supply (or lack thereof
395: in the case of the neutron star formation) of fresh helium into the burning core that can alter
396: the amount of carbon and thus the carbon burning lifetimes in this phase of stellar
397: evolution. Since this effect is not modelled in our simulations, our predictions are more
398: conservative. In other words, we could expect even more massive stars (than predicted here) to
399: form neutron stars under the favorable conditions.
400:
401: Due to ejection of matter and natal kicks during a supernova, it is expected that the neutron
402: star will be displaced from its formation site. The projected distance of CXO J164710.2-455216
403: from the cluster center is about 2.3 pc (e.g., Muno et al. 2006). The source could have been
404: formed anywhere in Westerlund 1 and been moving since the time of its formation ($t_{\rm form}$)
405: with velocity ($V_{\rm ns}$) imposed on the neutron star during the supernova explosion.
406: We note that motion of the system prior to the supernova explosion because of interactions with
407: other stars or binaries in the cluster could also contribute,
408: but the former are expected to be significantly larger. In Figure~4 the space velocities
409: of the very young pulsars formed in our simulations under the assumption of a null initial velocity
410: of the binary prior to explosion are shown. The results are presented for two calculations; a
411: reference calculation with the Hobbs et al. (2005) natal kick distribution ($\sigma=265$ km
412: s$^{-1}$), and one obtained with low kicks ($\sigma=133$ km s$^{-1}$). It is found that the
413: average velocities are $V_{\rm ns} = 380,\ 210$ km s$^{-1}$ for the reference and low kick
414: models, respectively. In addition, due to the smaller probability of progenitor binary disruption
415: (and thus formation of a solitary pulsar) the number of young solitary pulsars decreases from
416: $N_{\rm ns} \sim 1200$ for the reference model to $\sim 800$ for the low kick model. Using the
417: velocity distribution with the age of each pulsar (Fig.~2), the average distance traveled since
418: its birth is calculated as $D_{\rm ns} = 90,\ 50$ pc for the reference and low kick models,
419: respectively. This reveals that the majority of the young pulsars is bound to leave the cluster.
420: For a region characterized by 10 pc (comparable to the size of Westerlund 1) an estimate of the
421: probability of finding a pulsar within the cluster yields $\sim 10\%,\ 20\%$ of pulsars
422: for the reference and low kick models, respectively. This
423: result may indicate that the predicted fractional efficiency of forming very young pulsars is
424: lower (by factors of $\sim 5-10$) than presented in Figure~2. The decreased efficiency (even
425: by factor of 10), however, still permits one very young pulsar to be found within the cluster age
426: of 5 Myr.
427:
428:
429: \acknowledgements
430: We would like to thank K.Stepien and N.Langer for useful comments on
431: this study.
432: KB thanks Academia Sinica Institute of Astronomy and Astrophysics in Taipei
433: for hospitality. We acknowledge partial support through NSF Grant number
434: AST-0703960 (RT), and by the Theoretical Institute for
435: Advanced Research in Astrophysics (TIARA) operated under Academia Sinica and
436: the National Science Council Excellence Projects program in Taiwan administered
437: through grant number NSC 96-2752-M-007-007-PAE.
438:
439:
440: \begin{references}
441:
442: \reference{} Belczynski, K., Kalogera, V., \& Bulik, T.\ 2002a, \apj, 572,
443: 407
444: \reference{} Belczynski, K., Bulik, T., \& Rudak, B.\ 2002b, \apj, 571, 394
445: \reference{} Belczynski, K., Kalogera, V., Zezas, A., \& Fabbiano, G.\ 2004,
446: \apj, 601, L147
447: \reference{} Belczynski, K., \& Taam, R.\ 2004, \apj, 616, 1159
448: \reference{} Belczynski, K., Bulik, T., \& Ruiter, A.\ 2005, \apj, 629, 915
449: \reference{} Belczynski, K., Perna, R., Bulik, T., Kalogera, V., Ivanova,
450: N., \& Lamb, D.Q.\ 2006, \apj, 648, 1110
451: \reference{} Belczynski, K., Taam, R., Kalogera, V., Rasio, F., \& Bulik,
452: T.\ 2007, \apj, 662, 504
453: \reference{} Belczynski, K., Kalogera, V., Rasio, F., Taam, R., Zezas, A.,
454: Bulik, T., Maccarone, T., \& Ivanova, N. \ 2008, \apjs, 174, 223
455: \reference{} Belczynski, K., Lorimer, D., \& Ridlay, J.\ 2008, \apj,
456: to be submitted (arXiv:07.......)
457: \reference{} Brandner, W., Clark, J., Stolte, A., Waters, R., Negueruela, I.,
458: \& Goodwin, S.\ 2007, \aap, submitted (arXiv:0711.1624)
459: \reference{} Brown, G., Heger, A., Langer, N., Lee, C., Wellstein, S., \&
460: Bethe, H.\ 2001, New Astronomy, 6, 457
461: \reference{} Bulik, T., Belczynski, K., \& Prestwich, A.\ 2008, \apj,
462: submitted (arXiv:0803:3516)
463: \reference{} Casares, J. 2006, IAU Symposium 238: "Black Holes: From
464: Stars to Galaxies -Across the Range of Masses", in press
465: (astro-ph/0612312)
466: \reference{} Clark, J., Negueruela, I., Crowther, P., \& Goodwin, S.\ 2005,
467: \aap, 434, 949
468: \reference{} Clark, J., Muno, M., Negueruela, I., Dougherty, P., Crowther, P.,
469: Goodwin, S., \& de Grijs, R.\ 2008, \aap, 477, 147
470: \reference{} Crowther, P. A., Hadfield, L. J., Negueruela, I., \& Vacca, W. D. 2006,
471: \mnras, 372, 1407
472: \reference{} Dewi, J., \& Pols, O.\ 2003, \mnras, 344, 629
473: \reference{} Ergma, E., \& van den Heuvel, E.\ 1998, \aap, 331, L29
474: \reference{} Fryer, C.\ 1999, \apj, 522, 413
475: \reference{} Fryer, C., \& Kalogera, V.\ 2001, \apj, 554, 548
476: \reference{} Fryer, C., Heger, A., Langer, N., \& Wellstein, S.\ 2002, \apj,
477: 578, 335
478: \reference{} Hamann W.-R., \& Koesterke L.\ 1998, \aap, 335, 1003
479: \reference{} Hobbs, G., Lorimer, D., Lyne, A., \& Kramer,
480: M.\ 2005, \mnras, 360, 974
481: \reference{} Humphreys R.M., \& Davidson K.\ 1994, PASP, 106, 1025
482: \reference{} Hurley, J.\ R., Pols, O.\ R., \& Tout, C.\ A.\ 2000, \mnras,
483: 315, 543
484: \reference{} Iben I.Jr., \& Renzini A.\ 1983, ARA\&A, 21, 271
485: \reference{} Israel, G. L., Campana, S., Dall'Osso, S., Muno, M. P., Cummings, J.,
486: Perna, R., \& Stella, L. 2007, \apj, 664, 448
487: \reference{} Ivanova, N., Belczynski, K., Kalogera, V., Rasio, F., \&
488: Taam, R. E.\ 2003, \apj, 592, 475
489: \reference{} Kaper, L., Lamers, H., Ruymaekers, E., van den Heuvel, E., \&
490: Zuiderwijk, E.\ 1995, \aap, 300, 446
491: \reference{} Kobulnicky, H. A., \& Skillman, E. D.\ 1997, \apj, 489, 636
492: \reference{} Kudriztki, R., \& Reimers, D.\ 1978, \aap, 70, 22
493: \reference{} Maeder, A.\ 1992, \aap, 264, 1057
494: \reference{} Muno, M., et al.\ 2006, \apj, 636, L41
495: \reference{} Muno, M. P. et al. 2007, \mnras, 378, L44
496: \reference{} Nieuwenhuijzen, H., \& de Jager, C.\ 1990, \aap, 231,
497: 134
498: \reference{} Nugis, T., \& Lamers, H.\ 2000, \aap, 360, 227
499: \reference{} Orosz, J.~A\. 2003, in IAU Symposium, Vol. 212, A Massive Star
500: Odyssey: From Main Sequence to Supernova, ed. K.~van der Hucht,
501: A.~Herrero, \& C.~Esteban, 365
502: \reference{} Orosz, J.~A. et al.\ 2007, \nat, 449, 872
503: \reference{} Podsiadlowski, P., Langer, N., Poelarends, A.J.T., Rappaport,
504: S., Heger, A., \& Pfahl, E.D.\ 2004, \apj, 612, 1044
505: \reference{} Portegies Zwart, S. F., Verbunt, F., \& Ergma, E.\ 1997, \aap, 321, 207
506: \reference{} Prestwich, A., et al.\ 2007, \apj, 669, L21
507: \reference{} Silverman, J., \& Filipenko, A.\ 2008, \apj, submitted
508: (arXiv:0802.2716)
509: \reference{} Timmes, F., Woosley, S., \& Weaver, T.\ 1996, \apj, 457, 834
510: \reference{} Tout, C., Aarseth, S., Pols, O., \& Eggleton, P.\ 1997, \mnras,
511: 291, 732
512: \reference{} Vanbeveren, D., Van Bever, J., \& Belkus, H.\ 2007, \apj, 662, L107
513: \reference{} Van den Heuvel, E., \& Habets, G.\ 1984, Nature, 309, 598
514: \reference{} Vassiliadis E., \& Wood P.R.\ 1993, \apj, 413, 641
515: \reference{} Webbink, R. F.\ 1984, \apj, 277, 355
516: \reference{} Wellstein, S., \& Langer, N.\ 1999, \aap, 350, 148
517: \end{references}
518: \clearpage
519:
520:
521: \begin{figure}
522: \includegraphics[width=1.1\columnwidth]{f1.ps}
523: \caption{
524: {\em Top panel:} Initial (Zero Age Main Sequence) mass of neutron star progenitors as
525: predicted in population synthesis calculation for Westerlund 1 at the age of 5 Myr
526: (solar metallicity, standard winds). All neutron star progenitors originate from binaries
527: that were disrupted in a supernova explosion. Note the very high masses of the progenitors
528: ($M_{\rm zams} \sim 50-80 \msun$). {\em Bottom panel:} Neutron star masses predicted in
529: the same calculation. Note the high mass ($M_{\rm ns} \sim 1.8-2.5 \msun$) of the neutron stars.
530: }
531: \end{figure}
532: \clearpage
533:
534:
535: \begin{figure}
536: \includegraphics[width=1.1\columnwidth]{f2.ps}
537: \caption{
538: Cumulative distribution of formation time of single neutron stars predicted
539: in population synthesis calculation for Westerlund 1.
540: Single neutron stars that originate from disrupted binaries are shown with
541: solid/blue line, while the ones that originate from primordial single stars
542: are shown with dashed/red line.
543: Note that primordial single stars do not form neutron stars below 9 Myr,
544: while a small fraction of neutron stars ($f_{\rm form} \sim 0.001$) form in
545: the first 5 Myr from binary progenitors.
546: }
547: \end{figure}
548: \clearpage
549:
550:
551: \begin{figure}
552: \includegraphics[width=1.1\columnwidth]{f3.ps}
553: \caption{
554: Initial-final mass relation for single star models with various
555: wind mass loss rates for stars with H-rich envelopes: $f_{\rm wind,H}=1,\ 2,\ 3$.
556: All models are presented for high metallicity ($Z=0.03$) and fixed wind mass loss
557: rates for naked helium stars ($f_{\rm wind,He}=1$). We mark the initial mass
558: ($M_{\rm zams}$) range for each curve (shaded) for which formation of a compact
559: object takes shorter than $t_{\rm form}=5$ Myr. Also marked is the final compact
560: object mass ($M_{\rm f}$) range in which it is expected that the transition of a
561: neutron star to black hole formation takes place. Note that only for models with
562: the very high winds ($f_{\rm wind,H}=3$; bottom curve) can neutron stars be formed
563: from single progenitors within $5$ Myr.
564: }
565: \end{figure}
566: \clearpage
567:
568:
569: \begin{figure}
570: \includegraphics[width=1.1\columnwidth]{f4.ps}
571: \caption{
572: Neutron star space velocities for two natal kick models;
573: solid line shows results of our reference calculation with Hobbs et al.
574: (2005) kicks ($\sigma=265$ km s$^{-1}$), while dashed line represents results
575: obtained with low kicks ($\sigma=133$ km s$^{-1}$).
576: Note that not only the average velocity decreases from reference ($V_{\rm ns}
577: \sim 380$ km s$^{-1}$) to low kick model ($V_{\rm ns} \sim 210$ km s$^{-1}$),
578: but also the number of young solitary pulsars ($N_{\rm ns} \sim 1200$) decreases
579: due to the smaller probability of progenitor binary disruption for low kicks
580: ($N_{\rm ns} \sim 800$).
581: }
582: \end{figure}
583:
584:
585:
586: \end{document}
587:
588:
589:
590:
591: