A Survey for Planetary Nebulae in M31 Globular Clusters
George H. Jacoby, Robin Ciardullo, Orsola De Marco, Myung Gyoon Lee, Kimberly A. Herrmann, Ho Seong Hwang, Evan Kaplan, James E. Davies
aa r X i v : . [ a s t r o - ph . C O ] M a r A Survey for Planetary Nebulae in M31 Globular Clusters
George H. Jacoby
Giant Magellan Telescope / Carnegie Observatories, 813 Santa Barbara Street, Pasadena,CA 91101 [email protected]
Robin Ciardullo
Department of Astronomy & Astrophysics, The Pennsylvania State University, UniversityPark, PA 16802 [email protected]
Orsola De Marco
Department of Physics & Astronomy, Macquarie University, Sydney, NSW 2109, Australia [email protected]
Myung Gyoon Lee
Astronomy Program, Department of Physics and Astronomy, Seoul National University,Seoul 151-742, Korea [email protected]
Kimberly A. Herrmann
Lowell Observatory, Flagstaff, AZ 86001 [email protected]
Ho Seong Hwang
Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, MA 02138 [email protected]
Evan Kaplan
Vassar College, 124 Raymond Ave., Poughkeepsie, NY 12604 [email protected]
Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, MA 02138 [email protected]
ABSTRACT
We report the results of an [O III] λ R ∼ ∼ . ∼ . λ β ratios ranging from 2 to &
12. We discuss the individualcandidates, their likelihood of cluster membership, and the possibility that theywere formed via binary interactions within the clusters. Our data are consistentwith the suggestion that PN formation within globular clusters correlates withbinary encounter frequency, though, due to the small numbers and large uncer-tainties in the candidate list, this study does not provide sufficient evidence toconfirm the hypothesis.
Subject headings: planetary nebulae: general — globular clusters: general —galaxies: individual (M31) — stars: evolution
1. Introduction
Despite over a century of observations and decades of detailed modeling, the stellarpopulation that forms planetary nebulae (PNe) is still somewhat of a mystery. The tradi-tional theory states that the progenitors of PNe are low- to intermediate-mass single starsat the end of the AGB phase (Shklovski 1956; Abell & Goldreich 1966). This hypothesisexplains the distribution of PNe throughout space, and is responsible for the widely-heldbelief that the Sun will eventually evolve through this easily identifiable nebular phase (e.g.,Abell & Goldreich 1966; Ciardullo et al. 1989; Buzzoni et al. 2006). However, this theorydoes not provide a natural explanation for the non-spherical morphologies observed for the 3 –great majority of PNe, nor their low rate of formation. For these and other inconsistencies(see De Marco 2009), a new paradigm has been developed, wherein most planetary nebulaeare shaped via the interaction of an AGB wind with a binary companion (e.g., Soker 1997).Unfortunately, while the binary interaction model explains some of the anomalies as-sociated with the observed planetary nebula population (Moe & De Marco 2006; De Marco2009; Moe & De Marco 2012), this theory awaits final confirmation: the number of PN cen-tral stars with known binary companions is still relatively small, and programs to detectsuch objects are extremely challenging (De Marco et al. 2011, 2012). Moreover, neither thesingle-star nor binary-star hypothesis can explain the luminosities observed for PNe in theold stellar populations of elliptical galaxies and spiral bulges (Ciardullo et al. 2005), northe invariance of the bright end of the planetary nebula luminosity function (PNLF) acrossstellar populations (Ciardullo et al. 2002). For this, one must invoke yet another formationscenario, wherein significant mass transfer (or a complete stellar merger) occurs prior to theplanetary nebula phase (Ciardullo et al. 2005; Soker 2006).It is difficult to probe the different PN formation scenarios using field stars, since one hasalmost no prior information about the properties of the PN progenitors. However, within starclusters, the situation is different, as both the age and metallicity of the progenitor can beaccessed. Moreover, the low turnoff mass of old globular clusters (GCs) provides a tool withwhich to probe the binary formation scenario directly. Because GCs generally have turnoffmasses less than 1 M ⊙ , any post-AGB core arising from simple single-star stellar evolutionmust have a very small mass, M core . . M ⊙ (Kalirai et al. 2008), and an extremely longevolutionary timescale, t > yr (e.g., Caloi 1989). Such objects cannot make planetarynebulae by themselves, since the mass lost during the AGB phase will disperse long before thecore becomes hot enough to generate ionizing radiation. Any PN detected in these systemsmust therefore come from an alternate evolutionary channel, such as a common-envelopeinteraction or a mass augmentation process (i.e., a stellar merger).Searches for planetary nebulae in Galactic clusters have turned up only a few asso-ciated objects. Although more than a dozen PNe are projected near open clusters, thevast majority are undoubtedly line-of-sight coincidences (Majaess et al. 2007; Parker et al.2011). Similarly, out of 130 Galactic GCs surveyed, only four host PNe: Ps 1 in M15(Pease 1928), GJJC-1 in M22 (Gillett et al. 1986), JaFu1 in Pal 6, and JaFu2 in NGC 6441(Jacoby et al. 1997). Two of these PNe have high mass central stars more appropriate toPNe within open clusters ( ∼ . M ⊙ for Ps 1 and ∼ . M ⊙ for GJJC-1; Bianchi et al.2001; Harrington & Platoglou 1993), while the others have highly non-spherical nebulae(De Marco 2011). (The true nature of GJJC-1 is currently being re-assessed due to its highstellar mass, low nebular mass, and bizarre chemical composition (Jacoby et al. 2013), but 4 –for this paper, we adopt the usual PN classification.) These facts, along with the observationthat three of the four PNe are located in clusters that are rich in X-ray sources, suggest thatinteracting binaries play a role in the formation of cluster PNe (Jacoby et al. 1997).Since the sample of PNe within Galactic globular clusters is small, the significance ofany conclusion based on their properties is low. To better understand the processes thatform PNe within clusters, many more objects are required. For this, we must look to othergalaxies. Unfortunately, while there have been a few isolated associations of [O III] λ ∼ . §
2, we describe our multi-fiber observations and the basic reductionsteps needed to analyze ∼
460 candidate star clusters in M31. In §
3, we describe our searchfor embedded planetary nebulae, and the techniques required to recover objects that aremore than ∼ λ §
4, we describe the individual PNcandidates and their host clusters. Finally, we discuss our results, and show that the numberof PNe recovered within M31’s globular cluster system is roughly consistent with surveys ofMilky Way clusters.
2. Observations and Reductions
On 2008 Oct 25-28 (UT), we targeted 467 candidate star clusters in M31 with the 3.5-mWIYN telescope on Kitt Peak and the Hydra multi-fiber spectrograph. The objects selectedfor study were largely taken from a list of clusters given in the Revised Bologna Catalog(RBC; Galleti et al. 2004, 2007) and supplemented using the X-ray cluster identifications ofFan et al. (2005) and the young cluster candidates of Caldwell et al. (2009). As a control,we also positioned spare fibers on known M31 planetary nebulae taken from the list ofMerrett et al. (2006). These PNe span a range of brightnesses, from 20 . < m < . − . > log F > − . whydra .Since the M31 globular cluster system covers a much larger area than the 1 ◦ field-of-view ofthe WIYN+Hydra instrument (see Figure 1), we began by visualizing the locations of theindividual clusters using the [O III] images of the Local Group Galaxies Survey (Massey et al.2007). To accommodate the bulk of the galaxy’s cluster population, we located our first foursetups on M31’s bulge. Thereafter, we alternated between regions southwest and northeastof the galactic center, each time allowing whydra to identify an optimal field position bysearching a 25 ×
25 grid of space in 0 . ′ ◦ and 30 ◦ ) at each location. Top priority for fiber assignments was always given to previouslyunobserved clusters, with duplicate clusters, field PNe, and blank sky positions assignedlesser precedence. In total, eight different setups were executed over the four nights of theobserving run, and data were acquired for 391 RBC clusters, 64 X-ray clusters, and 12 youngclusters, with 30 of the systems being targeted more than once. An additional 55 of M31’sfield planetary nebulae were also observed. A log of the observations is given in Table 1.To execute these observations, the WIYN+Hydra system was configured to use theinstrument’s array of 3 ′′ diameter blue-sensitive fibers, the WIYN Bench spectrograph, anda 740 lines mm − Volume Phase Holographic (VPH) grating designed to optimize throughputnear 5000 ˚A. The resultant spectra covered the wavelength range between 4400 ˚A and 5450 ˚Aat 1 ˚A resolution and 0.5 ˚A pixel − . Typically, each Hydra setup was observed for 3.5 hr viaa series of seven 30 minute exposures.Our initial reduction procedures were similar to those described in Herrmann & Ciardullo(2009). We began with the tasks within the IRAF ccdred package: the data were trimmedand bias-subtracted via ccdproc , dome flats (typically three per setup) were combined us-ing flatcombine , and CuAr comparison arcs, which bracketed the program exposures, werecombined via imcombine . Next, dohydra within the hydra package was used to combine andlinearize the spectra, with the averaged dome flats serving to define the extraction apertures,and the averaged comparison arcs providing the wavelength calibration. We estimate thesewavelength calibrations to be precise to 0.04 ˚A yielding 1 σ errors of ∼ . − . As weillustrate below, this is small in comparison with the other uncertainties associated with ourradial velocity determinations. IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Asso-ciation of Universities for Research in Astronomy (AURA) under cooperative agreement with the NationalScience Foundation. scombine to combine the extracted spectra from the multipleexposures and then invoked skysub to perform the subtraction. As will be explained in § β and [O III] λ λ λ λ λ ∼
90% that at 5007 ˚A.The procedure for estimating the instrumental response at H β was slightly more compli-cated. We began by identifying 15 bright clusters and comparing their continuum flux near5000 ˚A to that surrounding the H β line. Spectral libraries (e.g., Jacoby et al. 1984) showthat, after applying M31’s foreground reddening ( E ( B − V ) = 0 . β , we simply adoptedthe inverse of the observed λ β continuum ratio. Our estimate of 68% is slightlygreater than the 62% value expected from the Bench spectrograph system and the VPHgrating (Bershady et al. 2008). However, since the efficiency of this grating is only knownto ∼ λ ∼
100 counts, our observations are sensitive to PNe thatare ∼
3. Finding PNe Within Star Clusters
Three criteria must be met before we can claim the detection of a PN candidate withinan M31 star cluster. First, we must identify the presence of emission lines in the spectrumof the cluster. Second, the observed emission lines must have line-ratios consistent withthose expected from a planetary nebula. Because of our limited spectral coverage (chosen toachieve the resolution needed to optimize emission line detections), this criterion is equivalentto requiring that the ratio of [O III] λ β be greater than some threshold (seebelow). Third, the velocity of a PN candidate, as derived from its emission-lines, must beconsistent with that of a star bound to its parent star cluster. Since the escape velocity froma typical M31 cluster is low, the precision of our emission-line and absorption line velocitymeasurements is an important parameter for our selection criteria. To search for PN emission within globular clusters, we began by normalizing each pro-gram spectrum with the continuum command within IRAF. We next divided the spectrainto five classes based on the strengths and widths of the absorption features, thereby effec-tively grouping the clusters by age and metallicity. The highest signal-to-noise spectrum ofeach class was then chosen as a template, and shifted to zero velocity using, as a reference,the absorption lines of H β , the Mgb triplet ( λλ β . Nevertheless, in virtually all cases, the tech-nique worked well around 5007 ˚A, as it effectively suppressed the continuum, allowing usto detect extremely weak emission from [O III]. Moreover, even when H β was poorly sub-tracted, we could still measure the emission-line ratios extremely well, as the underlyingstellar absorption was far broader than the unresolved Balmer emission. 8 – Our next step was to look for evidence of planetary nebula emission. Because ourdata were taken using a multi-fiber spectrograph, rather than traditional slit spectroscopy,local sky subtraction was not possible. Consequently, the emission arising from the diffuseionized gas of M31’s disk could not always be removed cleanly, and, even in the galacticbulge, line contamination was frequently a problem (see Ciardullo et al. 1988, for an imageof M31’s bulge emission). In fact, as summarized in Table 3, roughly one-third of the globularclusters surveyed displayed some evidence of emission, due mostly to M31’s warm interstellarmedium.To guard against this form of false detection, we considered the expected line ratiosof a planetary nebula. Most bright PNe are high excitation objects: in the top 2.5 magof the [O III] PNLF, all planetary nebulae have [O III] λ α ratios greater than 3(i.e., I ( λ /I (H β ) & λ β (Ciardullo et al. 2002; Herrmann & Ciardullo 2009). In the metal-poorenvironments of globular clusters, this lower limit is even more appropriate: of the 11 haloand globular cluster PNe observed by Howard et al. (1997) and Jacoby et al. (1997), allhave an excitation parameter, R = I ( λ /I (H β ) >
2. In contrast, the vast majority ofM31’s H II regions and diffuse ionized emission have H β brighter than [O III] λ R , its ionizing source (either a single Ostar or a very young OB association) must be very hot. Such an object will therefore bevery luminous — more than 100 or 1000 times the brightness of a PN central star — anddetectable either via its blue continuum or its overly bright emission-line luminosity.The only other sources that may have an excitation similar to that of a PN are supernovaremnants (SNRs), supersoft x-ray sources, and symbiotic stars. Supernova remants arerelatively rare (a factor of ∼
10 less numerous than PNe), and those remnants with
R > R ∼ all the bright PNe seen in extragalactic surveys are actually symbiotic stars.But this is hard to prove, and PN surveys in the Milky Way and the LMC find that symbiotic 9 –systems are only a minor contaminant (Viironen et al. 2009; Miszalski et al. 2011). Thus,emission-line sources that have [O III] λ β are muchmore likely to be PNe than H II regions, SNRs, or some other line-emitting object.Another way of testing for unrelated line emission is through the use of direct images.Deep H α and [O III] λ region along M31’s disk areavailable through the Local Group survey program of Massey et al. (2007). These images,which reach point-source flux limits of log F ∼ − . α and log F ∼ − . Even when high-excitation emission was detected in our fibers, its source was not alwaysassociated with the underlying globular cluster. Some of the emission within M31’s disk doeshave relatively high excitation, and there is always the possibility of a chance superpositionof a cluster with a true but unassociated PN. We can quantify the latter likelihood bycomputing the probability that a field planetary nebula would be projected within 1 . ′′ ∼ × L ⊙ of M31’s diffuse luminosity (i.e., exclusive of the globular clusters) is projectedwithin our survey fibers. This number, coupled with M31’s luminosity-specific PN density(Merrett et al. 2006), and the planetary nebula luminosity function (Ciardullo et al. 1989)implies the existence of ∼ . ∼ v , between its emission-line radial velocity and the absorption-line radial velocity of the cluster’s stars should satisfy the criterion ∆ v . σ eff , where σ eff isthe quadrature sum of the system’s internal velocity dispersion and the uncertainty in theradial velocity measurements. The former quantity exists for ∼
60% of the globular clustersin our survey, mostly through the R ∼ ,
000 MMT echelle spectroscopy of Strader et al.(2011). For the remaining old stellar systems, we can estimate the line-of-sight velocity 10 –dispersions through the clusters’ fundamental plane relation (e.g., Djorgovski et al. 1997;Strader et al. 2009, 2011). While we generally do not have access to information about acluster’s size or surface brightness, a projection of the Strader et al. (2011) clusters onto the M − σ plane yields σ (km s − ) ∼ . M T + 0 . M T (1)where M T is the cluster’s absolute T magnitude in the Washington system (Kim et al.2007), and σ is the observed velocity dispersion. Typically, these velocity dispersions spanthe range 3 . σ .
30 km s − , with a median value near ∼ − . Young clusters donot necessarily follow this relation, but from the structural analysis by Barmby et al. (2009),their line-of-sight velocity dispersion should be small, σ . − .The second term which enters into σ eff is that arising from the uncertainty of our velocitymeasurements. This error has two components. The first, which is associated with ourcentroiding of the [O III] λ ∼ − . This is in agreement with the results ofHerrmann & Ciardullo (2009), who used the same telescope and instrument setup to obtain . − precision for faint PNe in distant galaxies.The other component of the error term, that coming from the absorption line measure-ments, is more complex. Almost half of our clusters have high-quality ( σ v . − ) veloc-ity measurements, mostly through the MMT + Hectoechelle observations of Strader et al.(2011). As the left panel of Figure 5 shows, there is no systematic difference between ourmeasurements and those of the Hectoechelle. Six clusters have highly discrepant velocities,as they differ from their Strader et al. (2011) values by more than 3 times the internal er-rors of the measurements. Yet when these objects are removed, the remaining 199 objectshave a mean WIYN+Hydra velocity that is just ∆ V = 0 . − greater than that of theHectoechelle. Since the two sets of observations are on the same system, we can adopt thehigher precision Strader et al. (2011) velocities in our analysis.For most of the remaining clusters, we can use our own velocity measurements, alongwith their associated measurement errors. As the right panel of Figure 5 illustrates, thedispersion between our independent velocity estimates of clusters observed in more than oneHydra setup is consistent with the expectations of internal measurement error. Moreover, itis possible to obtain a quantitative estimate of our uncertainties by comparing our velocitymeasurements (and their errors) to those of the Strader et al. (2011) Hectoechelle data usingthe χ statistic χ = X i ( v − v S ) i ( σ + σ S ) i (2)where v , σ , v S , and σ S represent our velocities and their uncertainties, and those of Strader, 11 –respectively. When the six discrepant systems are removed, the reduced χ for the 199degrees of freedom is 0.96. This strongly suggests that the internal errors of our velocitydeterminations are accurate, and can be adopted as the true uncertainties of our measure-ments.Finally, as Figure 5 shows, the typical error of our absorption line velocities is between15 and 20 km s − . However, a small number of objects have velocity uncertainties thatare significantly greater than this. In a few cases, more accurate velocities are availablefrom the literature, and in those cases, we either adopted the previously measured values, oraveraged our velocities with the published data. The remaining objects, where our velocityuncertainties were greater than 30 km s − , were excluded from the analysis. Those clustersfor which our velocities disagree with previously measured values by more than four timesthe internal errors are given in Table 2.
4. The Clusters Hosting Emission
After forming σ eff , we excluded all PN candidates with [O III] λ σ eff . We note that this criterionmay not remove all the false detections, since young clusters are expected to have kinematicssimilar to that of the underlying disk. However, for the older systems, this should be a veryeffective discriminant. Since the rotation-corrected velocity dispersion of M31’s globularcluster system is ∼
130 km s − (Lee et al. 2008), the chance of finding a superposed disksource with the same velocity as one of our candidate clusters is extremely low.Table 3 lists all the cluster candidates surveyed in this program, along with their Wash-ington system photometry obtained with the KPNO 0.9-m telescope (Kim et al. 2007), theirradial velocities, the equivalent widths of their [O III] λ λ β emission-line ratio, and an age/type classification. For most of the clusters, this clas-sification comes from the analyses of either Caldwell et al. (2009, 2011) or Peacock et al.(2010), and are based on a variety of measurements, including multi-bandpass photometry, HST color-magnitude diagrams, and, in many cases, high signal-to-noise spectra. To inferthe ages of the 12 remaining objects without any age classification, we used the clusters withknown ages as a training set for our photometry. As Figure 6 illustrates, most of the clustersdesignated as “young” are quite blue, with C − T < . M − T < .
7. Conversely,the overwhelming majority of redder clusters are old. Thus for the clusters without a pre-vious age determination, we can use color as a proxy for evolutionary status. We classifyany object redder than C − T = 1 . β emission-linemeasurements are also tabulated. The first three objects give the candidates associatedwith stellar systems confirmed as being old; the next five list candidates in the youngerclusters. Finally, two of our PN candidates are associated with controversial objects, i.e.,cluster candidates which may, in fact, be foreground stars.To estimate the [O III] λ M magnitude (central wavelength 5075 ˚A). By measuring the strength of [O III] λ m = − . − . F (3)and placed on an absolute scale by assuming a distance of 750 kpc (Freedman et al. 2001)and a differential extinction of E ( B − V ) = 0 .
062 (Schlegel et al. 1998). On this scale, thebrightest PNe in M31 attain a luminosity corresponding to M = − . . ′′ ′′ , but JaFu 1 in Pal 6 would be offset by a full 1.8 ′′ . This means that for three outof the four objects, the photometric error due to their position of the PN within the clusterwould be negligible ( <
1% loss through our 3 ′′ fibers), but in 1 ′′ seeing, JaFu 1’s luminositywould be underestimated by a factor of ∼
5. In the absence of high resolution imaging, ourspectroscopic [O III] luminosities are the best that can be achieved for these objects, andlikely represent lower limits to the true [O III] λ Jacoby et al. (1997) argued that the single stars of old globular clusters cannot formPNe due to the time scale of their post-AGB evolution: by the time their cores becomeshot enough for ionization, their ejected gas would have dispersed far into the interstellarmedium. Thus, Jacoby et al. (1997) concluded that PN formation inside globular clustersmust involve binary stars, either through mass transfer, which increases the core mass tothat of a higher mass progenitor, or through a binary interaction which accelerates the speedof post-AGB evolution (Moe & De Marco 2012). More recently, Buell (2012) has suggestedan alternative, wherein the single stars of globular clusters evolve high mass cores by beingenriched in the helium produced by previous generations of star formation in the cluster. Inany scenario, however, central star mass is a critical parameter of the PN system, but onethat is very difficult to determine, even for Galactic PNe.We can, however, place limits on the mass of a PN central star using models of post-AGBevolution. At best, the [O III] λ ∼
10% of theluminosity being emitted from its central star (Dopita et al. 1992; Sch¨onberner et al. 2010).Moreover, for any central star luminosity, there is a minimum core required to generate thatenergy (Vassiliadis & Wood 1994; Bl¨ocker 1995). If we find that this minimum mass is toohigh to be produced by the evolution of a single star of an old stellar population, then wewill have strong evidence for a previous binary interaction. We apply this approach to ourcandidate objects.
B115-G177:
According to Caldwell et al. (2011), this globular cluster is old and metal-rich, with [Fe/H] ∼ +0 . ± .
1. Yet the system harbors an emission-line source that is brightenough to stand out in an Fe5015 versus Fe5270 plot, and hot enough to have an [O III] λ β . To place a lower limit on the luminosityof the exciting source, we can use the fact that no more than ∼
10% of a central star’s totalluminosity is reprocessed into [O III] λ M ∼ − .
2, its exciting star must be at least ∼ L ⊙ and have a mass of at least ∼ .
54 M ⊙ . Although low-mass cores are capable ofgenerating this amount of luminosity, their evolutionary time scale is too slow to produce aplanetary nebula. This is our best candidate for a PN inside an M31 globular cluster andan object formed by binary evolution. It may be coincidental that a rare PN is found ina relatively rare metal-rich, yet old, globular cluster (Woodley et al. 2010), or perhaps the 14 –combination of properties is a clue to PN formation. BH16:
This cluster, classified as old by Strader et al. (2011), possesses the X-ray sourceJ004246.0+411736 (Fan et al. 2005). Our velocity measurement for the system is relativelypoor ( − ±
28 km s − ), due to possible contamination from the underlying galacticbulge, and inconsistent with the − . ± . − Hectoechelle measurement obtainedby Strader et al. (2011). If we adopt the latter value, then the velocity of the superposed[O III] λ − ( ∼ . σ eff ),making an association likely. The emission-line source itself has a high-excitation ( R ∼ L ⊙ . NB89:
The Lick indices (Gonz´alez 1993) of this system imply an age of ∼
10 Gyr and ametallicity of [Z/H] ∼ − . λ ∼ λ β , and its inclusion in our list is partly due to the large ( ∼
30 km s − ) uncertaintyin our estimate of cluster velocity. If, instead of using our own measurement, we adoptthe velocity of − ± − observed by Barmby et al. (2000), then the emission-line’svelocity of − ± − is no longer consistent with it being part of the cluster. Caldwell(priv. comm.) reports that forbidden emission from [O II] and [S II] are strong, furthersuggesting that the emission-lines are interstellar in nature. SK044A:
Caldwell et al. (2009) classify this object as an M31 cluster (of indeterminateage) based on its spectrum and its profile on archival
HST frames. In contrast, Peacock et al.(2010) call the object a star, citing their analysis of UKIRT and SDSS data. We alsohave difficulty classifying the object: although the cluster’s neutral color places it 0 . ± .
056 mag blueward of our “young” versus “old” dividing line, its measured radial velocitydiffers from that of M31’s underlying disk by ∼
100 km s − (Chemin et al. 2009). Furthercomplicating the interpretation is our relatively poor determination for the cluster velocity( − ±
39 km s − ): even when combined with the − ±
46 km s − measurement ofKim et al. (2007), the resultant ±
29 km s − uncertainty still dominates the error budget.Nevertheless, the system is interesting, since its emission line velocity is inconsistent with awarm disk origin, and, with an [O III] λ β ratio of &
12, it has the highest emission-line excitation in our sample. The [O III] λ ∼ L ⊙ . Unfortunately, without better 15 –velocity information, we cannot say for certain whether the observed 48 km s − differencebetween the cluster’s emission lines and absorption features is indicative of an associationor a chance superposition. Caldwell (priv. comm.) notes that [S II] is weak in his spectrum,providing further support for the idea that the emission line is produced in a PN rather thanthe warm interstellar medium. SK051A:
This is our faintest PN candidate, and another object for which we have arelatively large (22 km s − ) measurement error. Nevertheless, our derived cluster velocityis in excellent agreement with that found by Kim et al. (2007), and the source does possesshigh-excitation ( R >
5) [O III] emission at a velocity consistent with both estimates. Thesystem has the colors of an old cluster, but without better data, we cannot confirm itsassociation with the emission line. Peacock et al. (2010) classify the object as a foregroundstar based on its appearance on images from UKIRT and SDSS, and Caldwell (priv. comm.)notes that the object is not resolved on
HST frames. Thus, it is possible that the observedcontinuum is from a point source contaminant, rather than a compact cluster.
B458-D049:
The cluster just barely satisfies our criterion, as the velocity of the [O III] λ − , or 2 . σ eff . It is a youngsystem, as evidenced by the poor results of our template subtraction about the H β absorptionline, and Caldwell et al. (2009) estimate its age at 0.5 Gyr. If the [O III] emission doescome from a planetary, then the physics of single-star stellar evolution implies that thePN is a high core-mass object. Specifically, the relationship between age and turnoff mass(Iben & Laughlin 1989), coupled with the initial mass-final mass relation (Kalirai et al. 2008)yields M core ∼ . M ⊙ . If the emission does come from a cluster PN, then the objecteither evolved from a single massive (young) star and is now well-past its peak [O III] λ ∼ L ⊙ , or it was created through a binary pathway and is likely the result ofcommon envelope evolution. One can sometimes distinguish between these two possibilitiesin the Galaxy where morphology and abundance anomalies provide some discrimination(Miszalski et al. 2013). At the distance of M31, luminosity is the primary indicator; if anobject is very bright, then it likely derives from a single massive star. M040:
Caldwell et al. (2011) re-classified this object as young, though they did notestimate an age. [O III] λ λ Hβ is virtuallyundetectable. The velocity agreement between the set of emission lines and that of theunderlying cluster continuum is not particularly good, between 13 and 28 km s − , dependingon whether we adopt our velocity or that of Caldwell et al. (2011). Given the ∼
30 km s −
16 –uncertainties of both measurements, an association remains a possibility.
C009-LGS04131:
Our velocity for this faint system is poor ( − ±
51 km s − ) but itis in excellent agreement with the value of − ±
32 km s − measured by Caldwell et al.(2009) and with the velocity of its [O III] λ − ± − ). Thecluster itself is young, with an age of ∼ . ∼ . M ⊙ (Iben & Laughlin 1989). Although [O III] λ M ∼ +1 . R >
6. Ifthe exciting source is a planetary, then, based on the turnoff mass, it is either a ∼ . M ⊙ core mass object which has faded substantially since its peak luminosity, or an object thatwas formed through a binary interaction. SK018A:
We observed this young cluster twice, with consistent results (∆ v = 6 . − ).Caldwell et al. (2009) estimate the age of the cluster to be ∼ . ∼ . M ⊙ (Iben & Laughlin 1989) and a PN core mass of ∼ . M ⊙ (Kalirai et al.2008). Like C009-LGS04131, the object is more than ∼ λ β , which implies R > DAO47:
Our velocity, when combined with two other determinations in the literature(Perrett et al. 2002; Caldwell et al. 2009), yields a value that is within 9 km s − of thatmeasured for the [O III] line. The spectrum is relatively noisy, and even after template-subtraction, residual stellar H β absorption is still visible. Nevertheless, [O III] λ β emission line is completely absent. Caldwell et al.(2009) estimate the age of the system to be ∼ . ∼ . M ⊙ (Iben & Laughlin 1989). The initial mass-final mass relation then implies M core ∼ . M ⊙ .
5. Discussion5.1. Expected and Observed Numbers of PN Candidates
In their survey of 130 Milky Way globular clusters, Jacoby et al. (1997) identified fourplanetary nebulae. Since our M31 survey targeted ∼
270 old clusters, a simple scaling ofnumbers suggests that we should have found ∼ ∼ ∼ × L ⊙ of bolometric light, and extended ∼ . ∼ . ∼ . N ( M ) ∝ e . M { − e M ∗ − M ) } (4)then the NGC 4472 observations imply that we should have seen . M V ∼ − .
6, and only M V ∼ − . − .
85 (Buzzoni et al. 2006), this implies thatour survey of young clusters sampled ∼ × L ⊙ of light. From the theory of stellarenergy generation, the bolometric luminosity stellar evolutionary flux for systems with agesbetween ∼ . ∼ ∼ × − stars yr − L − ⊙ (Renzini & Buzzoni 1986). Thus,we might expect these populations to produce one PN every ∼ ,
000 years. Even if thesePNe remained visible for 50,000 yr, that is still not enough time to build up a detectablepopulation. It is therefore likely that the larger velocity errors associated with these fainterclusters, and the kinematic similarity of the clusters and the field exacerbate our ability todiscriminate PN candidates from superposed disk emission. 18 –
Ciardullo et al. (2005) have argued that blue stragglers are responsible for most, if notall, of the bright PNe found in elliptical galaxies. Not only do these objects possess themain-sequence masses needed to build high-mass post-AGB cores, but their evolutionarytimescale, relative to that of PN ( ∼ ), is roughly the same as the relative numbers of thetwo objects. Since the creation of blue stragglers in globular clusters may, in some way, berelated to the rate of stellar encounters, Γ (i.e., Davies et al. 2004; Leigh et al. 2011), thenit is at least possible that the probability of finding a PN would also be proportional to thisfactor. From Verbunt & Hut (1987), this means that N ( P N ) ∝ Γ ∝ ρ r c σ (5)where ρ is the cluster’s central luminosity density, r c is the core radius, and σ is the stellarvelocity dispersion. The structural parameters of Milky Way clusters (Harris 1996, 2010)do seem to support this idea, as two of the systems which host PNe have extremely highencounter rates (NGC 6441 has the second highest rate of all Galactic GCs, and M15’s rateranks as seventh), and three of the four rank in the top half of clusters. Moreover, thisresult is not simply due to the clusters having more stars: if one calculates the mass-specificencounter rates of the Milky Way clusters, then again, three of the four clusters hostingPNe appear in the top half of the list. The statistics of only four objects are poor, but thenumbers do suggest a connection between stellar encounters and PN formation.Unfortunately, our emission-line survey of M31’s globular cluster system does not yetallow us to increase the statistics significantly. Although over 250 of M31’s globular clustershave structural measurements (Barmby et al. 2007; Peacock et al. 2010), only one of theold clusters listed in Table 4 are included in that number. Interestingly, if we assume thatthe clusters are in virial equilibrium (so that we can approximate the encounter rate usingΓ ∝ ρ . r c ), then the cluster in question (B115-G177, which contains our best PN candidate)has a value of Γ that ranks in the top ∼
25% of M31 systems. This again is suggestive, butit cannot be considered definitive until the other systems listed in Table 4 are surveyed.Alternatively, we can attempt to explore the frequency of stellar encounters using X-rayemission as a proxy for encounter rate. Pooley et al. (2003) have shown that in Galacticglobular clusters, there is an excellent correlation between Γ and the number of low-massX-ray binaries (LMXBs). If each LMXB had the same luminosity, we could test the PNbinary-formation hypothesis by searching for a correlation between PN presence and X-raybrightness. Jacoby et al. (1997) did this experiment in the Milky Way, where each individualX-ray source can be identified. This led the authors to suggest that binaries were responsiblefor the cluster planetary nebulae. 19 –In M31, one cannot resolve the individual X-ray sources within each cluster, and count-ing the total X-ray emission from a GC is not the same as measuring its total LMXB pop-ulation. In fact, the large range of luminosities possible for LMXBs makes any connectionbetween X-ray luminosity and binary population tenuous at best. In our survey, ∼
15% ofthe globular clusters have X-ray sources, but only one appears in Table 4 (Stiele et al. 2011).Consequently, X-ray emission in M31 clusters does not appear to present any evidence forthe PN binary formation hypothesis.
Figure 9 displays the [O III] λ For Galactic clusters, searching for PNe is relatively straightforward. As demonstratedby Jacoby et al. (1997), one can use the classic on-band/off-band technique to detect [O III] λ Mimics : Objects other than PNe (e.g.,, H II regions, SNRs, and diffuse emission) canhave similar spectral signatures over the limited wavelength range of the WIYN Bench Spec-trograph and its 740 lines mm VPH grating. This problem is partially technical, as manymodern spectrographs offer more complete spectral coverage at comparable resolution, al-lowing better discrimination against potential mimics. Yet even in the Milky Way, PNclassifications can be controversial (e.g., Viironen et al. 2009; Frew & Parker 2010) and PNcandidates are constantly being re-evaluated. The limited information available on extra-galactic objects only exacerbates this problem.
Spectral resolution and sensitivity:
These instrumental parameters are probably themost critical factors for successfully and definitively finding PNe in extragalactic globularclusters. As described above, our ability to define the velocity (and the velocity dispersion)of the underlying star cluster severely limited our ability to exclude chance superpositions.We were fortunate that the literature provided an excellent source of velocities for many ofour objects. For searches beyond M31, . − velocity measurements will not always beavailable. Similarly, sensitivity becomes increasingly important at larger distances, both fordetermining cluster velocities and for probing the faint end of PNLF. The study in NGC 4472(Peacock et al. 2012) only reached 2.5 mag down the PNLF; had we not gone far beyondthis limit in M31, we likely would not have identified any PN candidates. Finally, resolutionis also a helpful factor in this type of survey, both for the detection of PN emission, andfor eliminating mimics such as SNRs and nova shells, which will have broad emission lines.Again, this is a technical issue where large telescopes can dramatically advance studies likethis one. Spectral aperture (slit width, fiber size):
Our fiber diameter was limited to 3 ′′ , which isa reasonable match to the half-light radius of most M31 clusters. Still, these fibers do notsample all of the cluster light, and outlying PN could be overlooked, either because theyfall entirely outside the fiber, or near the fiber limits, where flux is lost due to the effectsof seeing. One could, of course, choose to use a large fiber or slit size for the observation(though at WIYN, the largest fiber size is 3 ′′ ), but this would reduce the signal-to-noise of themeasurement by admitting more sky and galactic background. This problem is amelioratedwhen going to more distant galaxies, though again at the cost of a higher galactic background 21 –and a loss of sensitivity due to the greater distance. Multiple objects along the line of sight:
When observing a distant galaxy, there is afinite probability that two unrelated objects will fall within the same spectroscopic aperture.This problem becomes worse as the distance increases, as a given fixed aperture represents abroader spatial swath of the galaxy. To compensate for this effect, one needs better velocitymeasurements so that unrelated objects can be discriminated.In the future, searches for PNe in extragalactic globular clusters should be more pro-ductive as many of these challenges can be overcome with technological advancements. Forexample, the M31 problem becomes relatively easy with the high signal-to-noise spectraproduced by an extremely large (25-40 m class) telescope (ELT) equipped with a multiob-ject, dual-channel, medium-resolution spectrograph. Similarly, adaptive optics on ELTs (ornarrow-band filters on the
Hubble Space Telescope could be used to resolve the cluster intostars, allowing the PN candidate to be imaged directly. This technique would not work aswell for systems beyond a few Mpc, but could produce a complete census of cluster PN inthe nearby universe.
6. Conclusions
We have demonstrated that it is possible to identify PN candidates in distant globularclusters using spectroscopy around the [O III] λ Hubble Space Telescope narrow-band images are needed to confirm their existence, especiallyfor those objects within young clusters. However, once confirmed, these targets represent anew source of material for understanding the physics of PN formation, and the chemistry oftheir parent clusters.
Acknowledgments
REFERENCES
Abell, G.O., & Goldreich, P. 1966, PASP, 78, 232Barmby, P., Huchra, J.P., Brodie, J.P., Forbes, D.A., Schroder, L.L., & Grillmair, C.J. 2000,AJ, 119, 727Barmby, P., McLaughlin, D.E., Harris, W.E., Harris, G.L.H., & Forbes, D.A. 2007, AJ, 133,2764Barmby, P., Perina, S., Bellazzini, M., Cohen, J.G., Hodge, P.W., Huchra, J.P., Kissler-Patig,M., Puzia, T.H., & Strader, J. 2009, AJ, 138, 1667Beasley, M.A., Brodie, J.P., Strader, J., Forbes, D.A., Proctor, R.N., Barmby, P., & Huchra,J.P. 2005, AJ, 129, 1412Bergond, G., Zepf, S.E., Romanowsky, A.J., Sharples, R.M., & Rhode, K.L. 2006, A&A,448, 155Bershady, M., et al., 2008, SPIE, 7014, 15Bessell, M.S. 2001, PASP, 113, 66Bianchi, L., Bohlin, R., Catanzaro, G., Ford, H., & Manchado, A. 2001, AJ, 122, 1538Blair, W.P., Kirshner, R.P., & Chevalier, R.A. 1982, ApJ, 254, 50Bl¨ocker, T. 1995, A&A, 299, 755Buell, J.F., 2012, MNRAS, 419, 2867Buzzoni, A., Arnaboldi, M., & Corradi, R.L.M. 2006, MNRAS, 368, 877Caldwell, N., Harding, P., Morrison, H., Rose, J.A., Schiavon, R., & Kriessler, J. 2009, AJ,137, 94Caldwell, N., Schiavon, R., Morrison, H., Rose, J.A., & Harding, P. 2011, AJ, 141, 61Caloi, V. 1989, A&A, 221, 27Chemin, L., Carignan, C., & Foster, T. 2009, ApJ, 705, 1395Chomiuk, L., Strader, J., & Brodie, J.P. 2008, AJ, 136, 234Ciardullo, R., Feldmeier, J.J., Jacoby, G.H., Kuzio de Naray, R., Laychak, M.B., & Durrell,P.R. 2002, ApJ, 577, 31 24 –Ciardullo, R., Rubin, V., Ford, W.K., Jacoby, G.H., & Ford, H.C., 1988, AJ, 95, 438Ciardullo, R., Jacoby, G.H., & Ford, H.C. 1989, ApJ, 344, 715Ciardullo, R., Sigurdsson, S., Feldmeier, J.J., & Jacoby, G.H. 2005, ApJ, 629, 499Davies, M.B., Piotto, G., & de Angeli, F. 2004, MNRAS, 349, 129De Marco, O. 2009, PASP, 121, 316De Marco, O. 2011, in Asymmetrical Planetary Nebulae V, ed. A. Ziljstra, I. McDonald &E. Lagadec, Jodrell Bank Centre for Astrophysics, 251De Marco, O., Jacoby, G.H., Davies, J., Bond, H.E., & Harrington, P. 2011, BAAS, 43, 152De Marco, O., Passy, J-C., Frew, D.J., Moe, M., & Jacoby, G.H., 2012, MNRAS, in pressDjorgovski, S.G., Gal, R.R., McCarthy, J.K., Cohen, J.G., de Carvalho, R.R., Meylan, G.,Bendinelli, O., & Parmeggiani, G. 1997, ApJ, 474, L19Dopita, M.A. Jacoby, G.H., & Vassiliadis, E. 1992, ApJ, 389, 27Fan, Z., Ma, J., Zhou, X., Chen, J., Jiang, Z., & Wu, Z. 2005, PASP, 117, 1236Frankowski, A., & Soker, N. 2009, ApJ, 703, L95Freedman, W.L., Madore, B.F., Gibson, B.K., Ferrarese, L., Kelson, D.D., Sakai, S., Mould,J.R., Kennicutt, R.C. Jr., Ford, H.C., Graham, J.A., Huchra, J.P., Hughes, S.M.G.,Illingworth, G.D., Macri, L.M., & Stetson, P.B. 2001, ApJ, 553, 47Frew, D.J., & Parker, Q.A. 2010, PASA, 27, 129Galarza, V.C., Walterbos, R.A.M., & Braun, R. 1999, AJ, 118, 2775Galleti, S., Federici, L., Bellazzini, M., Fusi Pecci, F., & Macrina, S. 2004, A&A, 416, 917Galleti, S., Federici, L., Bellazzini, M., Buzzoni, A., & Fusi Pecci, F. 2006, A&A, 456, 985Galleti, S., Bellazzini, M., Federici, L., Buzzoni, A., & Fusi Pecci, F. 2007, A&A, 471, 127Gillett, F.C., Neugebauer, G., Emerson, J.P., & Rice, W.L. 1986, ApJ, 300, 722Gonz´alez, J.J., 1993, PhD. thesis, Univ. California, Santa CruzGreenawalt, B., Walterbos, R.A.M., & Braun, R. 1997, ApJ, 483, 666 25 –Harrington, J.P, & Platoglou, G. 1993, ApJ, 411, L103Harris, W.E. 1996, AJ, 112, 1487Harris, W.E. 2010, arXiv:1012.3224Henize, K.G., & Westerlund, B.E. 1963, ApJ, 137, 747Herrmann, K.A., & Ciardullo, R. 2009, ApJ, 703, 894Howard, J.W., Henry, R.B.C., & McCartney S. 1997, MNRAS, 284, 465Iben, I., Jr., & Laughlin, G. 1989, ApJ, 341, 312Jacoby, G.H., De Marco, O., Lotarevich, I., Lahm, L., Bond, H.E., & Harrington, P. 2013,in preparationJacoby, G.H., Morse, J.A., Fullton, L.K., Kwitter, K.B., & Henry, R.B.C. 1997, AJ, 114,2611Jacoby, G.H., Hunter, D.A., & Christian, C.A. 1984, ApJS, 56, 257Kalirai, J.S., Hansen, B.M.S., Kelson, D.D., Reitzel, D.B., Rich, R.M., & Richer, H.B. 2008,ApJ, 676, 594Kent, S.M. 1987, AJ, 93, 816Kim, S.C., Lee, M.G., Geisler, D., Sarajedini, A., Park, H.S., Hwang, H.S., Harris, W.E.,Seguel, J.C., & von Hippel, T. 2007, AJ, 134, 706Larsen, S.S. 2008, A&A, 477, L17Lee, M.G., Hwang, H.S., Kim, S.C., Park, H.S., Geisler, D., Sarajedini, A., & Harris, W.E.2008, ApJ, 674, 886Leigh, N., Sills, A., & Knigge, C. 2011, MNRAS, 416, 1410Majaess, D.J., Turner, D.G., & Lane, D.J. 2007, PASP, 119, 1349Magnier, E.A., Prins, S., van Paradijs, J., Lewin, W.H.G., Supper, R., Hasinger, G., Pietsch,W., & Truemper, J. 1995, A&A, 114, 215Massey, P., McNeill, R.T., Olsen, K.A.G., Hodge, P.W., Blaha, C., Jacoby, G.H., Smith,R.C., & Strong, S.B. 2007, AJ, 134, 2474 26 –McConnachie, A.W., Irwin, M.J., Ferguson, A.M.N., Ibata, R.A., Lewis, G.F., & Tanvir, N.2005, MNRAS, 356, 979Merrett, H.R., Merrifield, M.R., Douglas, N.G., Kuijken, K., Romanowsky, A.J., Napoli-tano, N.R., Arnaboldi, M., Capaccioli, M., Freeman, K.C., Gerhard, O., Coccato,L., Carter, D., Evans, N.W., Wilkinson, M.I., Halliday, C., & Bridges, T.J. 2006,MNRAS, 369, 120Minniti, D., & Rejkuba, M. 2002, ApJ, 575, L59Miszalski, B., Napiwotzki, R., Cioni, M.-R.L., Groenewegen, M.A.T., Oliveira, J.M., &Udalski, A. 2011, A&A, 531, A157Miszalski, B., Boffin, H.M.J., & Corradi, R.L.M., 2013, MNRAS, 428, L39Moe, M., & De Marco, O. 2006, ApJ, 650, 916Moe, M., & De Marco, O. 2012, in I.A.U. Symp. 283, Planetary Nebulae: An Eye to theFuture, ed. A. Manchado, L. Stanghellini, & D. Sch¨onberner (Cambridge), 340Parker, Q.A., Frew, D.J., Miszalski, B., Kovacevic, A.V., Frinchaboy, P.M., Dobbie, P.D., &K¨oppen, J., 2011, MNRAS, 413, 1835Peacock, M.B., Maccarone, T.J., Knigge, C., Kundu, A., Waters, C.Z., Zepf, S.E., & Zurek,D.R. 2010, MNRAS, 402, 803Peacock, M.B., Zepf, S.E., & Maccarone, T.J. 2012, ApJ, 752, 90Pease, F.G. 1928, PASP, 40, 342Perrett, K.M., Bridges, T.J., Hanes, D.A., Irwin, M.J., Brodie, J.P., Carter, D., Huchra,J.P., & Watson, F.G. 2002, AJ, 123, 2490Pietsch, W., Fliri, J., Freyberg, M.J., Greiner, J., Haberl, F., Riffeser, A., & Sala, G. 2005,A&A, 442, 879Pooley, D., Lewin, W.H.,G., Anderson, S.F., Baumgardt, H., Filippenko, A.V., Gaensler,B.M., Homer, L., Hut, P., Kaspi, V.M., Makino, J., Margon, B., McMillan, S., Porte-gies Zwart, S., van der Klis, M., & Verbunt, F. 2003, ApJ, 591, L131Reid, W.A., & Parker, Q.A. 2010, MNRAS, 405, 1349Renzini, A., & Buzzoni, A. 1986, in Spectral Evolution of Galaxies, ed. C. Chiosi & A.Renzini (Dordrecht: Reidel), 195 27 –Remillard, R.A., Rappaport, S., & Macri, L.M. 1995, ApJ, 439, 646Schlegel, D.J., Finkbeiner, D.P., & Davis, M. 1998, ApJ, 500, 525Sch¨onberner, D., Jacob, R., Sandin, C., & Steffen, M. 2010, A&A, 523, A86Shklovskii, I.S. 1956, Astr. Zh., 33, 315Soker, N. 1997, ApJS, 112, 487Soker, N. 2006, ApJ, 640, 966Strader, J., Smith, G.H., Larsen, S., Brodie, J.P., & Huchra, J.P. 2009, AJ, 138, 547Strader, J., Caldwell, N., & Seth, A.C. 2011, AJ, 142, 8Stiele, H., Pietsch, W., Haberl, F., Hatzidimitriou, D., Barnard, R., Williams, B.F., Kong,A.K.H., & Kolb, U. 2011, A&A, 534, 55Vassiliadis, E., & Wood, P.R. 1994, ApJS, 92, 125Verbunt, F., & Hut, P. 1987, in I.A.U. Symp. 125, The Origin and Evolution of NeutronStars, ed. D.J. Helfand & J.-H. Huang (Dordrecht: Reidel), 187Viironen, K., Greimel, R., Corradi, R.L.M., Mampaso, A., Rodr´ıguez, M., Sabin, L.,Delgado-Inglada, G., Drew, J.E., Giammanco, C., Gonz´alez-Solares, E.A., Irwin,M.J., Miszalski, B., Parker, Q.A., Rodr´ıguez-Flores, E.R., & Zijlstra, A. 2009, A&A,504, 291Woodley, K.A., Harris, W.E., Puzia, T.H., Mat´ıas, G., Harris, G.L., & Geisler, D., 2010,ApJ, 708, 1335
This preprint was prepared with the AAS L A TEX macros v5.2.
28 –Fig. 1.— The locations of our WIYN+Hydra fields and the targeted M31 clusters, superposedon a mosaic of [O III] images from Massey et al. (2007). North is up, and east is to the left.Each 1 ◦ colored circle represents a different Hydra setup. 29 –Fig. 2.— A comparison of the absolute [O III] magnitudes of M31 PNe measured byMerrett et al. (2006) to the number of λ ∼ M ∗ = − .
5, which in M31 is equivalent to a monochromatic flux of2 . × − ergs cm − s − . 30 – Template ATemplate AsTemplate AwTemplate B
Template E
Fig. 3.— Spectra of the five template clusters used to suppress the stellar continuum andenhance the visibility of emission lines. The strongest absorption in this part of the spectrumcomes from H β ; most of the others lines are due to iron. The differences between spectrarepresent variations in cluster’s age and metallicity. 31 – B094-G156 Raw SpectrumSky SubtractedDoppler Shifted4800 4900 5000 5100Template Subtracted R e l a t i v e C oun t s Fig. 4.— The spectra of an M31 globular cluster between 4800 ˚A and 5100 ˚A, showing our rawdata, along with the data after sky subtraction, Doppler shifting, and template subtraction.Note that after template subtraction, the [O III] λ ∼ λ . − . Solid points display old(globular) clusters; open circles show younger systems. 33 –Fig. 6.— Washington system photometry of M31 clusters classified by Caldwell et al. (2009,2011), with blue representing clusters with ages t . t ∼
14 Gyr. On the left is a two-color diagram; on the right is a histogram of C − T colors. In the absence of a spectroscopic or HST imaging age designation, we classifysystems with C − T < . C − T > . β is roughly a factor of 1.46 less than that at [O III] λ β . 35 –Fig. 8.— The spectra of 5 M31 young clusters in the wavelength range between 4800 ˚A and5100 ˚A. To enhance the visibility of the emission lines, the spectrum of a template youngcluster been subtracted from each object; in some cases, a mismatch in age has resulted ina poor subtraction around H β . The y-axis represents counts; to convert to relative flux,note that the response at H β is roughly 1.46 times less than that at [O III] λ β . 36 –Fig. 9.— The planetary nebula luminosity function for the bulge of M31 (black pointswith error bars). These data, which extend over nearly 5 magnitudes, represent the deepestM31 PNLF currently available. The solid line shows the model PNLF (an exponential witha bright-end cutoff) that was adopted by Ciardullo et al. (1989). The magnitudes of theclusters PNe are marked: Milky Way PNe on the lower row, and M31 PNe on the upperrow. The red circles, blue triangles, and tan squares represent PN candidates in old confirmedclusters, young clusters, and candidate globular clusters, respectively. 37 –Table 1. Log of Hydra SetupsExposure Times Number of TargetsSetup UT Date (minutes) GCs Young ClustersM31-Y1 2008 Oct 25 7 ×
30 71 2M31-Y2 2008 Oct 25 7 ×
30 68 1M31-Y3 2008 Oct 26 7 ×
30 57 8M31-Y4 2008 Oct 26 7 ×
30 43 18M31-B1 2008 Oct 27 7 ×
30 31 25M31-G1 2008 Oct 27 6 ×
30 37 23M31-B2 2008 Oct 28 7 ×
30 21 26M31-R2 2008 Oct 28 7 ×
30 16 6 38 –Table 2. Globular Clusters with Discrepant VelocitiesCluster T Our v (km s − ) Published v (km s − ) Source a Difference ( σ )B034D 17.69 − ± − ±
25 1 8.1B079D 17.89 − ± − ±
25 1 7.9B104-NB5 16.29 − ±
27 32 . ± . − ± − . ± . − ± − ±
12 3 22.3BH16 17.16 − ± − . ± . − ± − . ± . − ± − ±
54 4 4.5SK021A 18.25 − ± − ±
39 4 5.8SK026A 18.90 − ± − ±
55 4 7.9SK045A 15.50 +46 ± − ±
52 4 11.2SK064A 19.30 − ± − ±
42 4 4.9SK079A 18.02 − ± − ±
26 4 5.3SK094A 18.16 − ± − ±
29 4 4.2SK096A 17.94 − ± − ±
52 4 4.9SK104A 17.37 − ± − ±
17 4 6.6SK106A 18.94 − ± − ±
38 4 14.9 a REFERENCES: (1) Galleti et al. (2006); (2) Strader et al. (2011); (3) Perrett et al.(2002); (4) Kim et al. (2007)
Table 3. Globular Cluster Observations
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesAU010 16.555 0.017 18.325 0.031 17.388 0.019 − . − . − . − . − . − . − . v = 11 . − B008-G060 16.270 0.006 18.077 0.009 17.103 0.005 − . v = 0 . − B009-G061 16.379 0.006 17.720 0.006 17.128 0.005 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . v = 3 . − B024-G082 16.308 0.004 18.045 0.009 17.098 0.005 − . − . − . i B026-G086 16.857 0.009 18.823 0.015 17.768 0.009 − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB026D-V216 17.265 0.016 19.970 0.050 18.322 0.020 − . i B027-G087 15.119 0.002 16.650 0.003 15.886 0.003 − . − . − . − . − . − . − . − . − . − . − . − . − . − . v = 3 . − B038-G098 15.934 0.010 17.635 0.017 16.772 0.011 − . − . − . − . v = 5 . − B041D 17.394 0.036 19.704 0.100 18.578 0.053 − . − . − . i B043-G106 16.630 0.009 17.070 0.005 16.988 0.007 − . − . i B044-G107 16.108 0.011 18.288 0.029 17.075 0.014 − . d B045-G108 17.201 0.004 18.762 0.009 18.029 0.006 − . − . − . − . − . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB051-G114 17.440 0.005 19.391 0.015 18.424 0.008 − . − . − . − . − . v = 0 . − B056-G117 16.603 0.009 18.684 0.017 17.509 0.010 − . − . β B058-G119 14.486 0.002 16.005 0.002 15.232 0.002 − . − . − . − . − . e B064-G125 15.810 0.009 17.234 0.013 16.536 0.009 − . . − . − . − . − . − . − . − . − . − . − . − . − . v = 19 . − B076-G138 16.243 0.013 17.785 0.020 17.065 0.014 − . − . − . − . − . − . v = 21 . − B082-G144 14.617 0.002 17.755 0.007 16.000 0.003 − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB083-G146 16.923 0.009 18.148 0.018 17.422 0.010 − . − . − . v = 8 . − B087 19.996 0.045 21.394 0.087 20.645 0.052 − . − . − . − . − . − . − . − . > . − . − . > . − . − . v = 3 . − B096D 17.592 0.025 19.555 0.036 18.862 0.036 − . − . − . − . − . − . − . − . − . − . − . − . − . v = 8 . − B107-G169 15.263 0.005 17.091 0.011 16.144 0.007 − . − . − . > . − . > . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB112D-M27 18.327 0.041 19.756 0.038 18.929 0.032 − . − . − . − . v = 4 . − B117-G176 15.910 . . . 16.677 . . . (16.669) . . . − . > . − . − . − . − . − . v = 3 . − B125-G183 15.998 0.010 17.361 0.014 16.701 0.010 − . − . − . − . v = 0 . − B129 18.072 0.008 22.425 0.171 19.963 0.026 − . > . − . − . − . − . > . − . − . − . − . − . < . − . − . − . R = 0 . − . − . − . − . − . v = 0 . − B154-G208 16.373 0.014 18.437 0.033 17.225 0.016 − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB156-G211 16.459 0.015 17.902 0.022 17.179 0.016 − . − . v = 5 . − B159 18.674 0.014 20.872 0.051 19.689 0.022 − . − . − . v = 0 . − B162-G216 16.881 0.013 18.868 0.019 17.840 0.015 − . − . . v = 5 . − B165-G218 16.080 0.006 17.321 0.005 16.722 0.005 − . − . − . − . − . v = 2 . − B171-G222 14.756 0.002 16.691 0.003 15.590 0.002 − . − . > . − . − . v = 2 . − B178-G229 14.632 0.002 16.077 0.002 15.326 0.002 − . − . − . v = 0 . − B182-G233 14.895 0.002 16.751 0.003 15.754 0.002 − . v = 1 . − B183-G234 15.478 0.003 17.349 0.005 16.294 0.003 − . − . − . v = 2 . − B187-G237 16.500 0.009 18.485 0.014 17.497 0.011 − . > . − . − . − . − . − . − . v = 2 . − B196D-SH08 (18.667) . . . (20.560) . . . (19.441) . . . . . . . . . Young 3 0.1 0.8B197-G247 17.056 0.015 19.259 0.027 18.069 0.018 − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB199-G248 17.109 0.007 18.442 0.011 17.796 0.009 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . v = 3 . − B220-G275 16.097 0.007 17.530 0.007 16.834 0.006 − . − . − . − . − . − . − . − . v = 0 . − B231-G285 16.722 0.012 18.344 0.013 17.494 0.011 − . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB235-G297 15.777 0.005 17.414 0.006 16.559 0.005 − . − . − . − . − . − . − . − . − . − . f B251 17.194 0.015 19.985 0.050 18.306 0.020 − . i B253 18.269 0.041 19.229 0.030 18.872 0.035 − . − . − . − . . − . − . v = 11 . − B271 17.736 0.029 19.693 0.041 18.706 0.032 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB313-G036 15.773 0.003 17.661 0.006 16.671 0.004 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesB378-G311 17.182 0.011 18.508 0.020 17.917 0.014 − . − . − . − . − . − . − . − . − . − . v = 11 . − B451-D037 17.755 0.018 18.894 0.019 18.316 0.017 − . − . − . − . − . − . − . − . − . − . − . − . > . − . − . − . − . − . − . v = 6 . − Fan 37 15.290 . . . . . . . . . . . . . . . − . − . − . − . . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesM009 17.306 0.016 18.661 0.015 18.019 0.014 − . − . − . − . − . − . − . − . − . − . − . − . > . − . > . − . − . − . − NB63 16.788 0.021 17.955 0.024 17.296 0.018 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . > . v = 6 . − SK019A 18.225 0.029 20.675 0.073 19.283 0.031 . . . . . . Old 4 . . . . . .SK021A 18.251 0.040 19.963 0.054 19.191 0.045 − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesSK022A 18.631 0.025 19.960 0.037 19.391 0.031 − . − . − . − . − . − . > . − . − . − . − . − . − . − . − . − . > . − . − . − . > . − . > . . − . − . − . > . − . − . > . − . − . − . − . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesSK061A 18.754 0.075 19.823 0.050 18.904 0.040 − . − . − . − . − . − . − . i SK068A 19.083 0.041 21.031 0.101 19.866 0.052 − . − . − . i SK071A 18.760 0.073 21.380 0.178 19.778 0.083 − . − . − . − . z = 0 .
074 galaxy h SK076A 18.343 0.051 19.989 0.055 19.266 0.053 − . − . − . i SK079A 18.025 0.038 20.200 0.063 19.179 0.048 − . − . − . − . > . − . − . − . − . − . − . − . − . − . > . − . − . − . Table 3—Continued
Velocity σ V Name T σ T a C a σ C M a σ M (km s − ) (km s − ) Type b Ref b R c EW (˚A) NotesSK095A 18.741 0.044 20.518 0.138 19.859 0.085 − . − . − . − . − . − . − . − . − . − . − . − . − . z = 0 .
017 galaxy h V202 . . . . . . . . . . . . . . . . . . − . − . − . − . a Numbers in parenthesis have been transformed into the Washington system using literature
BV I measurements and the relations of Bessell (2001). b The cluster age/type classification has been adopted from the following references: (1) Strader et al. (2011); (2) Caldwell et al. (2011); (3) Caldwell et al.(2009); (4) Peacock et al. (2010) c Derived ratio of [O III] to H β using the procedures described in § d Classified on the basis of an emission line a 5046 ˚A. e Classified on the basis of a 50 ˚A FWHM logarithmic-profile emission line at 4836 ˚A. f Classified on the basis of weak emission at 4548 ˚A. g Both [O III] λ β are well resolved and have the same double-peaked emission profile. h Classified using the redshifted emission lines of H β and the [O III] doublet. i Classified as a galaxy in the literature, but our spectrum is consistent with the object being a star cluster.
Table 4. Clusters with Candidate Planetary NebulaeCluster Age (Gyr) a M V S/N (5007 ˚A) S/N (H β ) R b EW (˚A) M ∆ v (km s − ) ∆ v/σ eff Globular ClustersB115-G177 14 − .
54 72 16 3.0 2.5 − . − − .
46 8.3 2.4 4.0 0.7 +0 . −
19 2.71NB89 c − .
60 36 12 2.1 1.5 +0 . −
34 1.10Young ClustersB458-D049 0.5 − .
73 6.0 1.6 2.1 0.8 +0 . − .
98 5.5 1.1 4.1 1.6 +1 . − .
58 6.3 . . . > . . − .
09 7.0 . . . > . . − .
52 3.8 . . . 3.5 0.5 +2 . − .
36 15 . . . >
12 1.9 +1 . − .
40 6.5 . . . > . . −
33 1.45Milky Way ClustersPal 6/JaFu1 Old − .
79 . . . . . . 6.3 . . . − . −
25 0.29M15/Ps 1 Old − .
19 . . . . . . 2.0 . . . +0 . − .
63 . . . . . . 3.3 . . . +2 . −
21 1.06M22/GJJC-1 Old − .
50 . . . . . . >
50 . . . +5 . −
16 1.14 a Age estimate of Caldwell et al. (2009, 2011); Objects labeled as “Star” were classified by Peacock et al. (2010). b Ratio of [O III] to H β cc