astro-ph0504276/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: % preprint (don't forget to change tab1.tex)%
3: \documentclass[apjl]{emulateapj}
4: %\usepackage{natbib}
5: \usepackage{apjfonts}
6: \usepackage{epsfig}
7: 
8: \bibpunct{(}{)}{;}{a}{}{,}
9: 
10: \begin{document}
11: 
12: \title{IRS 16SW - A New Comoving Group of Young Stars 
13: in the Central Parsec of the Milky Way}
14: 
15: \author{
16: J. R. Lu,
17: A. M. Ghez\altaffilmark{1},
18: S. D. Hornstein,
19: M. Morris,
20: E. E. Becklin
21: }
22: 
23: \affil{UCLA Department of Physics and Astronomy, Los Angeles, CA 90095-1562}
24: \email{jlu, ghez, seth, morris, becklin@astro.ucla.edu}
25: \altaffiltext{1}{UCLA Institute of Geophysics and Planetary Physics, 
26: Los Angeles, CA 90095-1565}
27: 
28: 
29: \begin{abstract}
30: One of the most perplexing problems associated with the
31: supermassive black hole at the center of our Galaxy is
32: the origin of the young stars in its close vicinity.
33: Using proper motion measurements and stellar number density
34: counts based on 9 years of diffraction-limited K(2.2 \micron)-band
35: speckle imaging at the W. M. Keck 10-meter telescopes, 
36: we have identified a new comoving group of stars, which we call the
37: IRS 16SW comoving group, located 1\farcs9 (0.08 pc, in projection)
38: from the central black hole.
39: Four of the five members of this comoving group
40: have been spectroscopically identified as massive young stars,
41: specifically He I emission-line stars and OBN stars.
42: This is the second young comoving 
43: group within the central parsec of the Milky Way to be recognized and is the 
44: closest, by a factor of 2, in projection to the 
45: central black hole. These comoving groups
46: may be the surviving cores of massive infalling star clusters
47: that are undergoing disruption in the strong tidal field
48: of the central supermassive black hole. 
49: \end{abstract}
50: 
51: \keywords{black hole physics -- Galaxy:center -- 
52: infrared:stars -- techniques:high angular resolution}
53: 
54: 
55: \section{Introduction}
56: 
57: The central parsec of our Galaxy harbors not only a supermassive
58: black hole (SBH) of mass 
59: $\sim$3.7 $\times$ 10$^6$ M$_{\sun}$ ({Ghez} {et~al.} 2003, 2005; 
60: {Sch{\" o}del} {et~al.} 2003), 
61: but also a cluster of young, massive stars within the
62: sphere of influence of this black hole. The young stars
63: include $\sim$40 He I emission-line stars 
64: identified from spectra as blue supergiants (Of), 
65: luminous blue variables (LBVs),
66: and Wolf-Rayet (WN/C) stars with masses ranging from 30 to 120 M$_{\sun}$, 
67: ages of 2-7 Myr, and distances limited to 1-10\arcsec ~from the SBH 
68: (e.g. {Allen}, {Hyland}, \& {Hillier} 1990; {Paumard} {et~al.} 2004).
69: Dynamical studies of these He I emission-line stars 
70: used the three-dimensional velocities to show that several, including
71: the brightest members of the IRS 16 complex such as 16C and 16SW,
72: exhibit a coherent, clockwise rotation pattern
73: ({Genzel} {et~al.} 2000)
74: and may all lie on a common orbital plane highly inclined to our line
75: of sight ({Levin} \& {Beloborodov} 2003).
76: In addition to this clockwise plane, 
77: the counter-clockwise orbiting He I emission-line stars
78: are consistent with a second plane oriented nearly face-on to our
79: line of sight ({Genzel} {et~al.} 2003).
80: Within the counter-clockwise plane, IRS 13 has been 
81: identified as a comoving, compact cluster of massive young stars
82: located $\sim$4\arcsec ~from the SBH 
83: ({Maillard} {et~al.} 2004). 
84: 
85: The origin of these young stars is puzzling given that the 
86: local gas densities are orders of magnitude too low to overcome the
87: tidal shear of the SBH and collapse to form stars 
88: ({Sanders} 1992; {Morris} 1993). 
89: Several possible solutions have been proposed
90: including scenarios that enhance the local gas densities such as 
91: a self-gravitating accretion disk 
92: ({Levin} \& {Beloborodov} 2003)
93: or infalling and colliding dense gas clouds 
94: ({Genzel} {et~al.} 2003),
95: both of which allow stars to form {\it in situ}. 
96: However, no such sufficiently dense gas clouds have
97: been observed. Alternatively, stars might form at larger radii where
98: the tidal forces are less extreme and
99: migrate inwards through dynamical friction as part of a
100: massive stellar cluster 
101: ({Gerhard} 2001; {Kim} \& 
102: {Morris} 2003; {Portegies Zwart}, {McMillan}, \&  {Gerhard} 2003;
103: {McMillan}, \&  {Portegies Zwart} 2003). 
104: It has been suggested that for this cluster to migrate
105: to within a few arcseconds of the SBH within the lifetime of the massive
106: young stars observed, the stellar cluster would need to be so dense as
107: to undergo core collapse and contain an intermediate mass black hole 
108: (IMBH; {Hansen} \& {Milosavljevi{\' c}} 2003; 
109: {Kim}, {Figer}, \& {Morris} 2004; {G{\" u}rkan} \& {Rasio} 2005). 
110: Although no clear cut case for an IMBH 
111: has yet emerged, the IRS 13 cluster is  coincident
112: with a bright, discrete X-ray source 
113: ({Baganoff} {et~al.} 2003; 
114: {Muno} {et~al.} 2003) and 
115: may be the remaining core of an infalling cluster harboring such an
116: IMBH ({Maillard} {et~al.} 2004). 
117: 
118: The orbital kinematics of the young stars provides
119: important insight into their origins.
120: In this paper we report the discovery of a new comoving group of stars
121: within the clockwise plane
122: which includes the bright young He I emission-line star, IRS 16SW. 
123: Diffraction-limited observations 
124: and data reduction are described in \S\ref{obsdata}. Section \ref{results}
125: reports the discovery and quantitative significance of the IRS 16SW
126: comoving group. Finally, \S\ref{discussion} addresses the implications 
127: of this result
128: for the proposed theories for the origins of young stars at the Galactic
129: Center and discusses the possibility
130: that this group is the remnant core of a infalling cluster.
131: 
132: 
133: \section{Observations \& Data Analysis} \label{obsdata}
134: 
135: Speckle imaging observations of the Galaxy's central stellar cluster were 
136: taken in the K-band ($\lambda_{o}$=2.2 \micron, 
137: $\Delta\lambda$=0.4 \micron) using the facility near-infrared
138: camera, NIRC ({Matthews} {et~al.} 1996), on the Keck I 10-meter telescope.
139: Data sets taken between 1995 June and 2003 September, detailed in 
140: {Ghez} {et~al.} (1998, 2000, 2005), and new 
141: data sets taken on 2004.34, 2004.56, and 2004.66, consist of $\sim$10,000
142: short (t$_{exp}$ = 0.1 sec) exposures per observing run with a plate
143: scale of 20.40 $\pm$ 0.04 mas/pixel and a 
144: 5\farcs22 $\times$ 5\farcs22
145: field of view. Frames are combined using a 
146: weighted shift-and-add technique (Hornstein et~al. in prep)
147: to form a final high resolution map for each data set 
148: (see {Ghez{ {et~al.} 2005 for details).
149: The data from each run were 
150: also divided into three sub-sets to create ``sub-maps'' used to determine
151: positional and brightness uncertainties.
152: 
153: 
154: \begin{figure*}
155: \epsscale{1.0}
156: \plotone{f1.eps}
157: \caption{
158: {\it (a):} The positions of all stars in our sample (asterisks)
159: overlaid on a map of the two-dimensional velocity dispersion
160: (grayscale). The sizes of the asterisks represent the stars' 
161: 2.2\micron ~brightness. The black region is a minimum in the
162: velocity dispersion, located $\sim$2\arcsec ~from Sgr A* (black cross),
163: and is caused by 5 comoving stars (blue), which define the newly
164: identified IRS 16SW comoving group.
165: {\it (b):} A K[2.2\micron]-band speckle image showing the clustering
166: of bright sources at the position of the IRS 16SW comoving group. 
167: Group members are marked with blue Xs.
168: {\it (Panels):} The proper motions of the IRS 16SW comoving group members.
169: In each  0\farcs1 $\times$ 0\farcs1 panel, the stellar positions
170: are plotted with different years of data labeled with different colors.
171: }
172: \label{fig:vdisp_image}
173: \end{figure*}
174: 
175: Sources were identified using
176: the IDL PSF fitting routine, {\it StarFinder} ({Diolaiti} {et~al.} 2000). 
177: {\it StarFinder} generates a PSF from several bright stars in the field
178: and cross-correlates the resulting PSF with the image. Candidate 
179: sources are those with correlation peaks above 0.7 in the main maps and 
180: above 0.5 in the sub-maps. Only sources detected in all 3 submaps are
181: included in the final source list for each observation. The coordinate
182: system for each list is transformed to a common local reference frame
183: by minimizing the net offsets of all stars as described in 
184: {Ghez} {et~al.} (1998, 2005). Centroiding uncertainties are 
185: $\sim$1 mas, while alignment uncertainties range from
186: $\sim$1-5 mas. The final relative positional
187: uncertainty is the quadrature sum of the centroiding and alignment
188: uncertainties and is $\sim$2 mas for the bright (K$\lesssim$13.5)
189: stars near IRS 16SW. 
190: Proper motions are derived by fitting lines to the 
191: positions as a function of time, weighted by the positional 
192: uncertainties. We conservatively require that only sources detected
193: in 9 or more epochs, out of 22 total epochs,
194: are included in the final sample. 
195: This results in a final sample of 180 stars, which have an 
196: average total proper motion uncertainty of 0.53 mas/yr
197: for all sources located beyond one arcsecond of the central SBH.
198: All proper motions were converted to linear velocities using a 
199: distance of 8 kpc and the uncertainty in this distance is not included
200: in the velocity uncertainties ({Reid} 1993).
201: 
202: \section{Results} \label{results}
203: 
204: A two-dimensional velocity dispersion map of the stars in the sample
205: reveals a minimum located between IRS 16SW and IRS 16SW-E 
206: (Fig.~\ref{fig:vdisp_image}a; grayscale). The velocity dispersion map
207: is produced by calculating, at each position separated by 0\farcs1, 
208: the following quantity for the nearest 6 stars: 
209: $\sigma^2_{intrinsic}$ = $\sigma^2_{measured}$
210: - $\sum_{i=0}^N$ [error$^2$(v$_{x,i}$) + error$^2$(v$_{y,i}$)] / [2*(N-1)],
211: where the first term is the dispersion of the measured proper motions
212: and the second term removes the bias introduced by the uncertainties
213: in the proper motion measurements. The minimum in the velocity dispersion
214: map is insensitive to the number of stars used in the calculation;
215: using the nearest 5 to the nearest 8 stars produces a similar result. 
216: The significance of the velocity dispersion minimum is determined by 
217: comparing it to the velocity dispersion of stars in the sample that are
218: at comparable radii (1\arcsec $\le$ r$_{2D} \le $2\farcs6). Because
219: the young stars are known to show some level of dynamical
220: anisotropy due to coherent
221: rotation about the SBH ({Genzel} {et~al.} 2003),
222: we restrict the comparison sample to known late-type stars 
223: ({Figer} {et~al.} 2003; {Ott} 2003). 
224: The minimum in the velocity dispersion is significantly lower 
225: (4.6$\sigma$) than the field velocity dispersion.
226: 
227: The velocity dispersion minimum arises from a comoving group of stars;
228: to formally define the members of the comoving group, we must
229: first eliminate those stars which appear near the group due to
230: projection effects. Formal membership is determined by
231: considering the difference between the velocity of each individual star
232: and the group's average velocity. Within the region of the velocity
233: dispersion minimum, only 5 stars have velocity offsets
234: that are consistently $\le 2\sigma$. Using only these 5 stars
235: to re-define the values of group velocity and velocity dispersion,
236: we find no additional stars with a total velocity offset less
237: than 3.5$\sigma$ within a 1\farcs1 search radius.
238: We therefore define these 5 stars, which include IRS 16SW, as the 
239: members of the comoving group 
240: (Table \ref{tab:members}, Fig.~\ref{fig:vdisp_image}-Panels).
241: The IRS 16SW comoving group has an average and RMS distance
242: from Sgr A* of 1\farcs92 and 0\farcs43, respectively, 
243: and a velocity dispersion of 
244: 36 $\pm$ 13 km/s in RA and 38 $\pm$ 13 km/s in DEC.
245: 
246: \input{tab1.tex}
247: 
248: There are two additional, independent lines of evidence supporting
249: the existence of this comoving group. First, two of the 
250: group members, IRS 16SW and IRS 16SW-E, have identical
251: radial velocities ({Ott} 2003); the other members, unfortunately,
252: have no measured radial velocities. Second, the stellar number density
253: counts show an enhancement at the position of the IRS 16SW comoving group
254: (see Fig.~\ref{fig:vdisp_image}b). Since the stellar number counts are
255: strongly affected by the varying sensitivity across the field from the
256: halos of bright stars, we consider only stars with 
257: K$<$13.5, corresponding to 99$\%$ completeness, for this analysis.
258: Using an aperture of 0\farcs55, which is the minimum radius necessary
259: to encompass the IRS 16SW comoving group, 
260: we generated a stellar number density map.
261: Positions including the IRS 16SW comoving group members
262: show the highest stellar number density, which corresponds to  
263: 6 sources per aperture. The probability of such an enhancement at this
264: radius over the field's average bright star density, 
265: 2 sources per aperture, is relatively small (0.03).
266: 
267: Members of the IRS 16SW comoving group appear to be young, massive stars.
268: The two brightest members have spectra that classify them as
269: He I emission-line stars identified as massive, young 
270: stars of type Ofpe/LBV or WR ({Krabbe} {et~al.} 1995). 
271: An additional two stars, S2-6 and S2-4, have recently been 
272: proposed, from spectroscopic observations, as
273: OBN stars, which are more luminous, nitrogen-rich OB stars 
274: ({Paumard} {et~al.} 2004).
275: The identification of four of the five members as young, massive stars
276: further supports the existence of this comoving group.
277: The remaining member, S2-5, lacks any spectral measurements to 
278: definitively classify it; 
279: however, its absolute K-magnitude, assuming A$_v$ = 30, 
280: is consistent with either a late-type giant
281: (M3 or M4 III) or a young massive star (WR or OB). 
282: The latter possibility is favored by 
283: narrow-band photometry through a CO feature, 
284: which shows no evidence of the CO absorption characteristic 
285: of late-type giants (Ott 2003). Furthermore, S2-5's dynamical association
286: with other young stars suggests that it is also a young star.
287: 
288: A model-dependent estimate of the IRS 16SW comoving group's three-dimensional 
289: distance from the SBH suggests that the line-of-sight component is
290: negligible.  
291: Specifically, IRS 16SW and IRS 16SW-E are supposed members of
292: the clockwise plane identified by 
293: {Levin} \& {Beloborodov} (2003) and {Genzel} {et~al.} (2003).
294: If we assume that all the members of the comoving group lie on this plane,
295: then their line-of-sight offsets from Sgr A* range from
296: -1\farcs1 to +0\farcs5, with an average values of -0\farcs4 (-0.016 pc).
297: This is small compared
298: to their plane-of-the-sky offsets from Sgr A* 
299: and places the group along the line of nodes where the
300: plane of the sky intersects the orbital plane.  Therefore, in this 
301: model, the physical distance between the IRS 16SW comoving group and 
302: the SBH is $\sim$0.08 pc.
303: 
304: It is highly unlikely that an inclined plane of stars orbiting the 
305: SBH could give rise to apparent comoving groups at the line of
306: nodes.  To quantify the likelihood that observational biases
307: might lead to apparent coherence,
308: Monte Carlo simulations were performed of 1000 stars with the 
309: following assumptions: 1) stars 
310: are distributed with uniform surface density in the plane out to
311: a radius of 7\arcsec, which is the outer edge imposed by
312: {Genzel} {et~al.} (2003), and 2) the stars are in circular orbits. 
313: Results from the simulation show that the line of nodes,
314: where the IRS 16SW comoving group lies,
315: is a point of high velocity dispersion, not low velocity dispersion as 
316: is observed. Given the above assumptions, the probability of finding
317: an apparent comoving group by random chance 
318: projected along the line of nodes within 4\arcsec ~of the SBH
319: is less than 1 in 10$^4$. This suggests that the observed IRS 16SW 
320: comoving group members are dynamically associated with each other 
321: beyond simply belonging to the clockwise orbital plane. 
322: 
323: 
324: \section{Discussion \& Conclusions} \label{discussion}
325: 
326: The IRS 16SW comoving group is now 
327: the second such grouping found within the central parsec of the 
328: SBH. The first, IRS 13, has been proposed as a compact,
329: massive star cluster located 3\farcs6 in projection from Sgr A*
330: and possibly harboring an IMBH
331: ({Maillard} {et~al.} 2004; {Eckart} {et~al.} 2004). 
332: Although the IRS 16SW comoving group is assigned to the clockwise plane and
333: the IRS 13 cluster is assigned to the counter-clockwise plane, 
334: both contain young, massive stars, 
335: indicating that the two groups have comparable ages,
336: strongly suggesting a similar formation mechanism.
337: However, there are distinct differences between the two comoving groups.
338: Unlike the IRS 13 cluster, which
339: has a high stellar number density of $\sim$40-80 objects/arcsec$^{2}$
340: ({Eckart} {et~al.} 2004),
341: the IRS 16SW comoving group shows only a slight stellar number density
342: enhancement ($\sim$7 objects/arcsec$^2$ to a similar 
343: limiting magnitude of $\sim$16.5). Likewise, while the IRS 13 
344: complex sits on a bright complex of dust, ionized gas, and possibly an
345: unresolved population of stars, the IRS 16SW comoving group shows
346: no equivalently bright halo in the K-band. 
347: Any formation scenario
348: for the young, massive stars in the galactic center must explain
349: the presence of both groups.
350: 
351: One proposed {\it in situ} formation mechanism invokes self-gravitating
352: accretion disks present a few million years ago but since consumed 
353: by star formation, accreted onto the SBH, or dissipated via 
354: winds from the young, massive stars 
355: ({Morris} 1993; {Levin} \& {Beloborodov} 2003). The disk
356: would produce stars in circular orbits in a common orbital plane.
357: The low velocity dispersion of the IRS 16SW comoving group along
358: the line of nodes suggests either a non-uniform distribution within
359: the plane, which would indicate that star formation was clumpy,  
360: or non-circular orbits, which would argue against 
361: accretion disk formation scenarios.
362: It remains to be determined whether two accretion disks can assemble,
363: form massive clusters of stars, and dissipate within only a few million 
364: years of each other. 
365: Another {\it in situ} formation
366: scenario, colliding gas clouds, cannot be supported or disputed
367: observationally as no detailed simulations of star formation during
368: such interactions have been performed. 
369: 
370: Alternative scenarios invoke the formation of massive star clusters 
371: at larger radii which spiral in via dynamical friction.
372: As a result of mass segregation and possible core
373: collapse, the surviving core would contain primarily massive stars.
374: If the IRS 16SW comoving group is a bound cluster, then the
375: virial, Bahcall-Tremaine ({Bahcall} \& {Tremaine} 1981), and 
376: Leonard-Merritt ({Leonard} \& {Merritt} 1989) 
377: projected mass estimators yield a cluster mass of 
378: $M\sim10^4 M_{\sun}$, 
379: which is an upper limit if the cluster is not bound. 
380: Similarly, for the simple assumption that the cluster acts as a point mass,
381: observations of the reflex motion of 
382: Sgr A* rule out clusters of $>10^4 M_{\sun}$ at the projected distance
383: of the IRS 16SW comoving group, assuming circular orbits 
384: ({Reid} \& {Brunthaler} 2004). 
385: In order for the IRS 16SW comoving group
386: to remain bound, the required mass density to overcome
387: the tidal forces at 0.08 pc from Sgr A* is 
388: $\sim4\times10^9 M_{\sun}/pc^3$ yielding a cluster mass 
389: lower limit of $M_{lower}\sim10^5 M_{\sun}$ assuming
390: a uniform density cluster for an order of magnitude calculation. 
391: Therefore, if the comoving group is indeed situated only 0.08 pc
392: from Sgr A*, then it is most likely unbound.
393: 
394: If the IRS 16SW comoving group is not a bound cluster, then it is 
395: difficult to understand how such a comoving group can have survived
396: in the extreme tidal field of the supermassive black hole.
397: One possibility is that the group is the dissipating remnant core
398: of a tidally disrupted cluster.
399: {Gerhard} (2001) shows that,
400: for the remaining unbound core of a cluster with 
401: extent $\alpha$r$_t$, where r$_t$ is the tidal
402: radius, the timescale for orbital phases to spread by $\sim\pi$
403: is of order $t_{spread} \sim \frac{t_{orbit}}{\alpha}$. 
404: The circular orbital period at
405: the group's distance from Sgr A* is $<$ 1000 years, suggesting that
406: any clustering should be disrupted within a few thousand years.
407: If the IRS 16SW comoving group is the 
408: remaining core of a massive star cluster, then the disruption may
409: have begun at larger radii ($\sim0.4$ pc), where the furthest
410: young He I emission-line stars are observed, while the core continued to
411: migrate in via dynamical friction ({Gerhard} 2001). Given the range of 
412: orbital timescales, $\sim$10$^3$-10$^4$ yrs for 0.04-0.4 pc, 
413: and the present existence of a comoving
414: group, disruption would have occurred only a few tens of thousands
415: of years ago.  The other young, massive stars that seem to lie
416: on the same orbital plane as the IRS 16SW comoving group may be other
417: members of the disrupted cluster core which are spreading in phase. 
418: IRS 13 and the associated stars in the 
419: second orbital plane would then be a second infalling cluster in an
420: earlier stage of disruption. 
421: Increasing the number of known young stars and measuring the
422: spread in phases would provide a better estimate of the time since
423: disruption. 
424: Simulations of the final disruption of 
425: an infalling cluster are needed to fully test this possibility,
426: and to determine if such a disruption can also give rise to the
427: cusp of OB stars within 0\farcs5 of Sgr A* (e.g. {Ghez} {et~al.} 2003,
428: {Eisenhauer} {et~al.} 2005).
429: The short duration of the final cluster disruption, our
430: observations at such a unique epoch, and the presence of two young
431: comoving groups on different orbital planes suggest that either the 
432: frequency of formation and infall of massive clusters 
433: is high ({Portegies Zwart} {et~al.} 2002) 
434: or, more likely, there was an epoch of triggered star 
435: formation several million years ago ({Figer} {et~al.} 2004) 
436: which produced two massive clusters at similar distances from
437: the central supermassive black hole. 
438: In conclusion, the IRS 16SW comoving group provides further evidence
439: for formation scenarios that involve clustering whether from infalling 
440: massive star clusters or {\it in situ} formation. 
441: 
442: \acknowledgements
443: 
444: Support for this work was provided by NSF grant AST-0406816 and
445: the NSF Science \& Technology Center for AO, managed by UCSC
446: (AST-9876783) and the Packard Foundation. 
447: The W.M.~Keck Observatory is operated as a scientific 
448: partnership among the California Institute of Technology, the University 
449: of California and the National Aeronautics and Space Administration. 
450: The Observatory was made possible by the generous financial support of 
451: the W.M.~Keck Foundation. 
452: 
453: \bibliography{}
454: 
455: \end{document}
456: