IRS 13N: a new comoving group of sources at the Galactic Center
AAstronomy & Astrophysics manuscript no. ms˙IRS13N˙4 c (cid:13)
ESO 2018October 25, 2018
IRS 13N: a new comoving group of sources at the GalacticCenter
K. Muˇzi´c , , R. Sch¨odel , A. Eckart , , L. Meyer , A. Zensus ,
1) I. Physikalisches Institut, Universit¨at zu K¨oln, Z¨ulpicher Str. 77, 50937 K¨oln, Germany2) Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69, 53121 Bonn, Germany3) Instituto de Astrof´ısica de Andaluc´ıa, Camino Bajo de Hutor 50, 18008 Granada, Spain4) University of California, Division of Astronomy and Astrophysics, Los Angeles, CA 90095-4705e-mail: [email protected]
Received / Accepted
ABSTRACT
Context.
The Galactic Center IRS 13E cluster is located ∼ Aims.
We present the first proper motion measurements of IRS 13N members, and additionally give proper motions of four ofIRS 13E stars resolved in the L’-band.
Methods.
The L’-band (3.8 µ m) observations have been carried out using the NACO adaptive optics system at the ESO VLT.Proper motions have been obtained by linear fitting of the stellar positions extracted by StarFinder as a function of time,weighted by positional uncertainties. Results.
We show that six of seven resolved northern sources show a common proper motion, thus revealing a new comovinggroup of stars in the central half parsec of the Milky Way. The common proper motions of IRS 13E and IRS 13N clusters aresignificantly ( > σ ) different. We also performed a fitting of the positional data for those stars onto Keplerian orbits, assumingSgrA* as the center of the orbit. Our results favor the very young stars hypothesis. Key words.
Galaxy:center – infrared:stars
1. Introduction
The central half parsec of the Milky Way is host to asurprisingly high number of massive young stars (see e. g.Paumard et al. 2006; Ghez et al. 2005), organized in atleast one disk-like structure of clockwise rotating stars(CWS; Genzel et al. 2003; Levin & Beloborodov 2003;Paumard et al. 2006). Paumard et al. (2006) also proposethe existence of a second, less populated disk of counter-clockwise rotating stars (CCWS). The mechanism respon-sible for the presence of young stars in the strong tidalfield of the super-massive black hole (SMBH) at the po-sition of SgrA* is not clear. Two most prominent sce-narios include star formation “in-situ” (in an accretiondisk; Levin & Beloborodov 2003; Nayakshin et al. 2006a),and the in-spiral of a massive stellar cluster formed at asafe distance of 5-30 pc from the Galactic Center (GC;Gerhard 2001; McMillan & Portegies Zwart 2003; Kimet al. 2004; Portegies Zwart et al. 2006). At this point, the former scenario seems to be favored by number of au-thors (Nayakshin & Sunyaev 2005; Nayakshin et al. 2006a;Paumard et al. 2006). Also, recent results by Stolte et al.(2007) definitively rule out the possibility that the Archescluster could migrate inwards and fuel the young stellarpopulation in the GC.A special challenge and a good observational test forboth the above mentioned scenarios is provided by theexistence of the IRS 13 group of sources. IRS 13E (lo-cated ∼
3” west and ∼ a r X i v : . [ a s t r o - ph ] F e b K. Muˇzi´c et. al.: IRS13N comoving group
For both of the above mentioned scenarios of star for-mation at the GC there are several issues when dealingwith IRS 13E. In principle, such a cluster could have beenformed in an accretion disk (Milosavljevi´c & Loeb 2004;Nayakshin & Cuadra 2005). However, in numerical simula-tions of star forming disks, fragmentation of a disk cannotproduce such a dominant feature (Nayakshin et al. 2007).In light of the cluster in-fall scenario, an intermediate-mass black hole (IMBH) was proposed to reside in thecenter of the cluster (Maillard et al. 2004). The existenceof an IMBH makes the process of cluster in-spiral muchmore efficient in terms of increasing dynamical frictionand thus allowing the most massive stars that reside inthe center of a very massive ( > M (cid:12) ) stellar cluster toreach the central parsec of the Galaxy within their life-times (Hansen & Milosavljevi´c 2003; Berukoff & Hansen2006; Portegies Zwart et al. 2006; but see Kim et al. 2004for a characterization of the problems with this hypoth-esis). However, the presence of the IMBH in IRS 13E isdisputed. Sch¨odel et al. (2005) analyze the velocity disper-sion of cluster stars and derive that the mass of such anobject should be at least 7000 M (cid:12) , making its existenceimplausible. Both the X-ray (Baganoff et al. 2003) andradio (Zhao & Goss 1998) source at the position of IRS13E can be explained by colliding winds of high-mass los-ing stars (Coker et al. 2002; Zhao & Goss 1998), withoutthe need for any unusual object. Paumard et al. (2006)suggest that, in the case that IRS 13E is associated withthe Bar of the mini-spiral, the mass of the stellar contentwould be high enough to keep the cluster from disruption.A second comoving group of stars, called IRS 16SWand located 1.9” in projection from SgrA*, was reportedby Lu et al. (2005). Four of the five members of theIRS 16SW comoving group are spectroscopically identi-fied as young massive stars, indicating ages comparableto those of the IRS 13E cluster. However, unlike IRS 13E,IRS 16SW shows only a slight stellar number density en-hancement.Approximately 0.5” north of IRS 13E, a small clus-ter of unusually red compact sources has been reported(IRS 13N; Eckart et al. 2004). Eight sources have been re-solved and labeled α through η as shown in Figs. 1 and 2of Eckart et al. (2004). A strong infrared excess is due tothe emission of warm (T ∼ Table 1.
Proper motions of IRS 13N sources observed inL’-band. name ∆ α a ∆ δ a v R.A. b v Dec. b α -2.70 -1.48 44 ± ± β -2.89 -1.25 - 6 ±
20 228 ± γ -3.10 -0.99 -114 ±
14 298 ± δ -2.92 -0.85 -33 ±
13 248 ± (cid:15) -2.88 -1.02 -58 ±
15 306 ± ζ -3.16 -0.82 -134 ± ± η -3.11 -0.66 -91 ±
22 323 ± c -3.01 -0.93 -73 ± ± a relative to SgrA*, in arcseconds b all velocities are in km s − ; the uncertaintiesrepresent the 1 σ uncertainty of the linear fit c average proper motion of six comoving sources β - η Table 2.
Proper motions of IRS 13E sources observed inL’-band. name ∆ α a ∆ δ a v R.A. b v Dec. b E1 -2.95 -1.64 -107 ±
11 -118 ± ± ± ± ± ±
45 140 ± c -3.17 -1.61 -167 ±
12 37 ± d -3.24 -1.60 -187 ±
15 88 ± a relative to SgrA*, in arcseconds b all velocities are in km s − ; the uncertaintiesrepresent the 1 σ uncertainty of the linear fit c average proper motion of four IRS 13E sources d average proper motion of IRS 13E without E1 motions were derived using L’-band NACO/VLT images .
2. Observations and Data Reduction
The L’ (3.8 µ m) images were taken with theNAOS/CONICA adaptive optics assisted im-ager/spectrometer (Lenzen et al. 1998; Rousset et al.1998; Brandner et al. 2002) at the UT4 (Yepun) atthe ESO VLT. The data set includes images from 6epochs (2002.66, 2003.36, 2004.32, 2005.36, 2006.41 and2007.39) with a resolution of ∼
100 mas and a pixel scaleof 27 mas/pixel. Data reduction (bad pixel correction, skysubtraction, flat field correction) and formation of finalmosaics was performed using the DPUSER software forastronomical image analysis (T. Ott; see also Eckart &Duhoux 1990). All the images were deconvolved using the Based on observations collected at the European SouthernObservatory, Chile. Muˇzi´c et. al.: IRS13N comoving group 3
Fig. 1.
Identification and proper motions of stars in IRS 13 complex. Four arrows are shown for each source, indicatingthe ± σ uncertainty of the value and direction of its proper motion.linear Wiener filter. The absolute positions of sources inour AO images were derived by comparison to the VLApositions of IRS 10EE, 28, 9, 12N, 17, 7 and 15NE asgiven by Reid et al. (2003).Stellar positions were extracted with StarFinder(Diolaiti et al. 2000) and transformed into the commoncoordinate system with the aid of 19 reference stars. Thepositions of the reference stars were corrected for the stel-lar proper motions as derived from the K S -band images(see Muˇzi´c et al. 2007). In Muˇzi´c et al. (2007) we haveshown that the lower resolution L’-band data can be reli-ably used to obtain proper motions.Proper motions were derived by linear fitting of po-sitions as a function of time, weighted by the positionaluncertainties. Both the error of the transformation to thereference frame and the error of the stellar position fittingcontribute to the uncertainties.We assume the distance to the GC to be 7.6 kpc(Eisenhauer et al. 2005).
3. Results
In tables 1 and 2 we list proper motions of IRS 13Nand IRS 13E sources identified in L’-band, respectively. InFig. 1 we show proper motions of all stars superposed onthe L’-band image. Stars β through η show similar propermotions, revealing a new comoving group of sources at theGC. As shown in Eckart et al. (2004), only η can also bedetected in the K-band. Paumard et al. (2006) give the fullvelocity information for this star: v R.A. = (-52 ± − , v Dec. = (257 ± − and v r = (40 ± − . The au-thors identify it as a member of a possibly existing secondstellar disk (CCWS). Concerning IRS 13E, the obtained proper motion forE1 shows the biggest discrepancy with respect to previ-ously published data (Paumard et al. 2006; Sch¨odel et al.2005). It also deviates significantly from the proper mo-tions of the other three stars. It was already noted thatE1 shows the largest deviation from the systemic propermotion of the cluster (Sch¨odel et al. 2005). Paumard et al.(2006) argue that this star is possibly not bound to thecluster. This is supported by our result.It is important to note that, in addition to a very densestellar population in the region, a large amount of diffusedust emission detected at 3.8 µ m makes the astrometrymore challenging to perform than at shorter wavelengths.For this reason, StarFinder was not able to resolve allthe previously reported K-band sources in the region. E3is known to be at least a double source, while here wepresent it as a single one. E4 is located close to E3 butis much fainter in the L’-band. Therefore it could not beclearly identified by StarFinder. However, we were ableto resolve the faint source E5 in all epochs and presentthe first proper motion measurement of this star. E5 wasproposed to be a dusty WR star (Maillard et al. 2004),and it seems to exhibit a proper motion consistent withthose of E2 and E3. This implies that it may be a clustermember.From Tables 1 and 2 one can see that the averageproper motions of the two clusters are significantly ( > σ )different. Thus, it seems that two clusters do not belongto the same system. In order to analyze our measurements we attempted tofit the positional data of IRS 13N stars showing similar
K. Muˇzi´c et. al.: IRS13N comoving group
Fig. 2. ( a ) the best fit orbital solutions for five IRS 13N stars ( γ – η ); ( b ) orbital solutions for a single plane ( i =24 o ,Ω=180 o ), for stars γ – η ; ( c ) chosen 1 σ orbits for η , with the best fit orbit colored black; NOTES: part of the orbitin front of the plane of the sky is colored red; crosses mark present positions of the stars; axes show the offset fromSgrA*.proper motions (stars β to η ) to Keplerian orbits, assum-ing that the gravitational potential is dominated by theSMBH at the position of SgrA*. For details of the fittingprocedure and definition of parameters see e.g. Eisenhaueret al. (2003). We assumed the distance and the mass ofSgrA* to be 7.6 kpc and 3.6 × M (cid:12) , respectively. The ve-locity of SgrA* was assumed to be zero (Reid & Brunthaler2004). Under the assumption that the stars are really onbound orbits around SgrA*, it is clear that our data coveronly a small part of such an orbit. We also lack the radialvelocity information for all of the stars. As a consequence,the χ minimization gives a very wide span of orbital pa-rameters for which χ <χ min +1, even when fixing mostof the free parameters (the mass of and the distance toSgrA*, as well as the position and the velocity of the cen-ter of the orbit). Therefore we repeated the orbital fittingfor each star introducing the radial velocity data as givenin Paumard et al. (2006) for η . We varied the inclinationangle i between 0 o and 90 o in steps of 1 o , Ω (position angleof the ascending node) between 0 o and 180 o in steps of 20 o and ω (longitude of periastron) between 0 o and 360 o witha step size of 20 o . The results are given in Table 3. We listthe best fit parameters i and Ω, and give the 1 σ range forthe inclination. We also list the range in eccentricities ( e )and semi-major axes ( a ) resulting from the fits for whichthere is some combination of angles ( i , Ω, ω ) that givesa fit better than 1 σ ( ). For all the stars except ζ , thebest fit reduced χ value ranges between 0.6 and 1.5. For ζ the best fit is obtained with χ =3.9, probably implyingthat either positional uncertainties are underestimated, orthat assuming the radial velocity of η is a poor assump-tion. In the last three lines of Table 3 we give the best-fitparameters for the Keplerian orbits fitted onto the aver-age positions of the stars within the IRS13N and IRS13Ecomoving groups. For IRS 13E we perform the fitting pro- a fit with χ <χ min +1, for a corresponding star. For aconvenience, we refer to all orbits that satisfy this condition as1 σ orbits. cedure twice: first including the star E1 and afterwardsconsidering it a non-member.Paumard et al. (2006) conclude that IRS 13E is on afairly eccentric orbit, with e (cid:38) β , the range ineccentricities of 1 σ orbits is restricted to fairly low values( < η . Note that the corresponding semi-major axesalso result fall within a restricted range of values. In con-trast, both parameters ( e and a ) are poorly constrainedfor the star β . The size of the best fit orbit of β is an or-der of magnitude higher than those of the other five stars.Since the orbit of β is radically different from orbits of allthe other stars, we note that it probably should not beregarded as part of the co-moving group. Therefore we donot show its orbit in Fig. 2. Since η is the only star withknown radial velocity, we took its orbital parameters andattempted to fit other stars onto the same orbit. We notethat it is not possible to fit all other stars onto exactly thesame orbit (same i , Ω and ω ). Therefore we fix i and Ωand let ω vary. This confines the stars to the same plane,within which they can still have different orbits. For threestars ( γ , (cid:15) , ζ ) there exist 1 σ orbits with given i and Ω,the star δ can be reasonably (but still not within 1 σ ) fit-ted, and β fails completely to be fitted onto this plane(Fig. 2 b ). To get a feeling about the uncertainties of ouranalysis, we plot several 1 σ orbits for the same star ( η ) inFig. 2 c .The implications of these results are further discussedin Section 4.2.
4. Discussion
Paumard et al. (2004) report the morphology and ra-dial velocities of the mini-spiral gas based on He I andBr γ observations. At the position of the IRS 13 sources,the Northern Arm and the Bar of the mini-spiral areoverlapping along the line of sight, with the Bar being . Muˇzi´c et. al.: IRS13N comoving group 5 Table 3.
Results of the Keplerian orbit fitting name i ( o ) a Ω( o ) a i ( o ) b e c a ( (cid:48)(cid:48) ) c β
79 180 53 - 80 0 - 0.9 4.0 - 180.0 γ
19 180 4 - 67 0.17 - 0.21 2.6 - 4.0 δ
48 180 39 - 71 0.15 - 0.4 2.7 - 15.0 (cid:15)
33 180 17 - 56 0.1 - 0.3 2.5 - 7.2 ζ
18 180 0 - 30 0.1 - 0.3 3.2 - 4.7 η
24 180 1 - 45 0.1 - 0.3 2.8 - 5.7IRS 13N d
28 180 14 - 49 0.1 - 0.2 2.8 - 5.7IRS 13E d , e
83 100 80 - 85 0.5 - 0.9 12.5 - 145.0IRS 13E d , f
70 120 14 - 42; 58 - 80 0.5 - 0.8 2.3 - 66.0 a best fit parameters ( χ = χ min ) b all inclination values for which χ <χ min +1 c calculated for ( i , Ω, ω ) combinations for which χ <χ min +1 d average orbit e including E1 f not including E1 placed further away from the observer. The authors sug-gest that IRS 13E must be either embedded in the Bar, orvery close to it. Also, the radial velocities of the IRS 13Estars are similar to the radial velocities of the Bar mate-rial at this position (Paumard et al. 2004). Spectroscopicobservations of the entire IRS 13 region (Moultaka et al.2005) indicate a close spatial correlation between the stel-lar sources and the surrounding material. Higher waterice and hydrocarbon absorptions in the northern part ofIRS 13, as well as a redder continuum emission, suggestthat the IRS 13N sources are more embedded in the ISMthan the rest of the complex. The radial velocity of η isconsistent with the radial velocity of the Bar at this po-sition (between 0 and 50 kms − , Paumard et al. 2004). Ifthe northern sources that show similar proper motions alsohave correlated radial velocities, it would be reasonable tosuggest that the entire complex is indeed associated withthe Bar. According to our orbital fits, IRS 13N is locateda few equivalent arcseconds behind Sgr A*, which nicelyagree with the view of Paumard et al. (2004) about theBar being located somewhat behind Sgr A*. This wouldimply that the IRS 13E and IRS 13N stellar complexes arespatially close. The existence of two comoving clusters,each moving in a completely different direction, poses alot of challenges for star formation scenarios at the GC. As can be seen from the orbital analysis, the best fit orbitsfor IRS 13N stars have rather different orbital parameters.Observing six stars on different orbits exactly at the mo-ment when they appear to be very close in projection andto have similar proper motions is not very probable. Thereis no single orbit onto which all of the stars would fit witha good χ , but at least four of the stars can be confined tothe same plane. In this case the stars are again on differ-ent orbits and consequently have different orbital periods, spanning the range from 1000 to 3000 years. This againimplies that the present arrangement is temporary. Also,it is curious that the moment at which we observe them co-incides exactly with the presence of dense gas at their po-sition. Interestingly, their most likely common plane seemsto coincide with the plane of the counter-clockwise mov-ing stars (CCWS; Paumard et al. 2006). The existence ofthe CCWS disk is still a matter of debate. According toPaumard et al. (2006), the disk is sparsely populated (12stars), with η belonging to it. The assumption that allIRS 13N stars belong to the CCWS significantly increasesthe population of the disk, and at the same time weakensthe claim that the CCWS is essentially in non-circular mo-tion. In the analysis of Paumard et al. (2006), seven out oftwelve CCWS stars are on eccentric orbits ( e> σ orbits for one star (Fig. 2 c ) showsthat the uncertainty of semi-major axes ( a ), as well asthat of orbital periods ( P ), is of the same order as the dif-ferences between a and P of each star. Therefore, at thispoint we still cannot exclude that IRS 13N stars are indeedorbiting together. In the following we consider the possi-bility that the cluster is bound. We calculate the velocitydispersion of stars as σ = σ pm − N (cid:88) i =1 [ error ( v x,i ) + error ( v y,i )] / [2( N − σ pm is the dispersion of the measured proper mo-tions and the second term removes the influence of theproper motion measurement uncertainties.The mass estimate from the velocity dispersion ofstars gives ∼ (cid:12) , an unrealistically high mass forsuch a cluster. For a comparison, we note that Paumardet al. (2006) estimate the total stellar mass of IRS 13Eto be ∼ (cid:12) . We note that even if the uncertaintiesin proper motion measurements were underestimated, the K. Muˇzi´c et. al.: IRS13N comoving group final mass of the cluster calculated from the velocity dis-persion would remain high. For instance, a hypotheticalunderestimation of our uncertainties of 30% would resultin ∼
8% lower velocity dispersion and the mass of 2700 M (cid:12) .The Hill radius r Hill ≈ a (1 − e ) (cid:16) m M (cid:17) for a cluster of m = 3300 M (cid:12) , with a = (2.8 - 5.7)” and e = 0.1 - 0.3 is r Hill ≈ ∼ r Hill ∼ a and minimum e , the minimum mass needed tokeep the cluster from disruption would be 1250M (cid:12) .The relaxation time for a system of N stars with the meanmass m ∗ is given by t relax ≈ N ln Λ × t cross where t cross = R / v and Λ= R v /Gm ∗ ≈ N (Binney &Tremaine 1987). Also, it is important to note that tidal in-teractions tend to shorten relaxation timescales. A clusterwith the IRS 13N velocity dispersion would have a t relax of the order of the IRS 13N orbital period only if it wasconsisted of 500-1000 stars. For m ∼ (cid:12) contained in500-1000 main sequence stars we get that stars should beas bright as m K =17-18 (assuming the extinction A K ≈ (cid:12) , we get m K =19-21, exactlyat the limit of NACO K-band data. Therefore it seemsthat the system could hardly be bound and survive for asignificant time in the orbit around SgrA*.If IRS 13N stars are not bound, the observed veloc-ity dispersion probably indicates that the cluster is in theprocess of dissolution. The velocity dispersion will sub-sequently diminish the stellar surface number density ofthe cluster, which would within only a few hundred yearsreach the background value given by Sch¨odel et al. (2007). The detailed discussion about the nature of IRS 13Nsources is presented in Eckart et al. (2004). Among otherpossibilities, the authors argue that infrared excess sourcesIRS 13N could represent objects that have colors andluminosities consistent with YSOs and Herbig Ae/Bestars. Dusty envelopes that usually surround those objectswould then give rise to the observed strong infrared excess.It is striking that six of the sources that are very closein projection, also show very similar proper motion val-ues. In the previous section we argue that IRS 13N couldhardly survive in the present arrangement for a significantamount of time. The indication of a dynamically youngstellar system concurs with the Eckart et al. (2004) hy-pothesis of IRS 13N stars being extremely young objects.But why then does this object exsist at all? Timescales forits disruption seem to be much shorter than any timescaleswithin which stars normally form. Here we must take into account that the Galactic Center is an extremely com-plex environment and that any star formation process thattakes place there is completely different from the “nor-mal” star formation via the fragmentation of a molecu-lar cloud. Timescales for star formation could be signifi-cantly shorter. The free-fall time in a collapsing cloud atthe distance of the young stars in the central half parsecaround Sg rA* is of the order of 100-1500 yr. Also, simu-lations of star formation in a gaseous disk around Sgr A*by Nayakshin et al. (2007) indicate that the timescales forstar formation are fairly short, of the order of several thou-sand years. This is comparable to the orbital timescale ofIRS 13N.A serious obstacle for forming stars at the GC is thatthe gas densities in the central parsec are way too low forthe gas to be able to form stars. Even the highest densityestimates for the circumnuclear disk (CND; Christopheret al. 2005) are several orders of magnitude lower than re-quired to fulfill the criterion for Jeans instability. However,with the aid of shocks and collisions, a small clump of gasin-falling to the center could be highly compressed andstar formation within it triggered. As a matter of fact,the mini-spiral is a short-lived feature ( t dyn ∼ yr) andis in-falling right now. Furthermore, the material of themini-spiral is susceptible to shocks, as discussed by e.g.Muˇzi´c et al. (2007). A possibility of forming stars in smallassociations has never been discussed so far for this re-gion. Star formation in small groups definitely would notbe able to account for the young co-eval stellar populationin the central parsec, but could, in general, be possible andresult in small associations like IRS 13N.However, it is clear that we still must be very carefulwhen drawing a conclusion that IRS 13N cluster containsvery young stars. In case that the cluster is indeed a partof the CCWS disk and that the stars were formed simul-taneously with other CCWS stars, then they must be sev-eral Myr old. The fact that they are apparently stronglyextincted may then indicate that they are similar to themini-spiral bow-shock sources (Tanner et al. 2005), butmany times less massive and luminous (see discussion inEckart et al. 2004). The question remains: why are theyclustered?
5. Conclusions
In this paper we present the proper motion measurementsof IRS 13N, a newly discovered co-moving group of redsources at the GC. We discuss the boundness of a clusterand a possibility that these stars are the youngest stellarobject ever observed in this environment. Ongoing starformation in the GC would have significant implicationson our understanding of this particular environment, andstar formation in galactic nuclei in general. It has beenassumed that the star formation in the GC occurred intwo star bursts, ∼
100 Myr and ∼ . Muˇzi´c et. al.: IRS13N comoving group 7 would also mean that star formation in the GC must notnecessarily be related to star bursts and that, in fact, acontinuous star formation could go on.The question of the real nature of the IRS 13N clusterremains open. More imaging and especially high resolutionspectroscopic data that could reveal features attributed toYSOs will be necessary to confirm our hypothesis that thestars are extremely young. Acknowledgements.
We thank the referee, Dr. M. Morris,for comments and suggestions that improved this work.Part of this work was supported by the
DeutscheF orschungsgemeinschaft (DFG) via SFB 494. K. Muˇzi´c andL. Meyer were supported for this research through a stipendfrom the International Max Planck Research School (IMPRS)for Radio and Infrared Astronomy at the Universities of Bonnand Cologne.