Submitted to ApJL Expansion Velocities and Core Masses of Bright Planetary Nebulae in the Virgo Cluster Magda Arnaboldi1,2, Michelle Doherty1, Ortwin Gerhard3, Robin Ciardullo4 8 0 5 6 7 8 J. Alfonso L. Aguerri , John J. Feldmeier , Kenneth C. Freeman , George H. Jacoby 0 2 n ABSTRACT a J 7 The line-of-sight velocities and [OIII] 5007 ˚A expansion velocities are mea- ] sured for 11 planetary nebulae (PNs) in the Virgo cluster core, at 15 Mpc dis- h p tance, with the FLAMES spectrograph on the ESO VLT. These PNs are located - o about halfway between the two giant ellipticals M87 and M86. From the [OIII] tr 5007 ˚A line profile widths, the average half-width at half maximum expansion s a velocity for this sample of 11 PNs is v¯ = 16.5 kms−1 (RMS = 2.6 kms−1). HWHM [ We use the PN subsample bound to M87 to remove the distance uncertainties, 1 v and the resulting [OIII] 5007 ˚A luminosities to derive the central star masses. 0 We find these masses to be at least 0.6 M and obtain PN observable life times ⊙ 8 0 t < 2000 yrs, which imply that the bright PNs detected in the Virgo cluster PN 1 core are compact, high density nebulae. We finally discuss several scenarios for . 1 explaining the high central star masses in these bright M87 halo PNs. 0 8 0 Subject headings: Planetary nebulae: general. Galaxies: M87, distance and : v redshift, halos. Galaxies: clusters: Virgo cluster i X r a 1European Southern Observatory, Garching, Germany; [email protected],[email protected] 2INAF, Observatory of Turin, Turin, Italy 3 Max-Planck-Institut fu¨r extraterrestrische Physik, Garching, Germany; [email protected] 4Dept. of Astronomy and Astrophysics, Pennsylvania State University, University park, PA; [email protected] 5Instituto de Astrofisica de Canarias, Tenerife, Spain; [email protected] 6Dept. of Physics and Astronomy, Youngstown State University, Youngstown, OH; [email protected] 7Mount Stromlo Observatory, Research School of Astronomy and Astrophysics, ACT, Aus- tralia;[email protected] 8 WYIN Observatory,Tucson, AZ; [email protected] – 2 – 1. Introduction Since the discovery of free-floating intracluster planetary nebulae (ICPNs) in the Virgo cluster (Arnaboldi et al. 1996), extensive imaging and spectroscopic observations were car- ried out to determine their projected phase space distribution and the fraction of diffuse cluster light not bound to Virgo galaxies. To enlarge the sample of more than 40 ICPN line-of-sight (LOS) velocities available from Arnaboldi et al. (2003, 2004), we have obtained new PN spectra with FLAMES at the ESO VLT (Doherty et al. 2008). The high spectral resolution of the new data allows us for the first time to measure expansion velocities for the [OIII] nebula shells of 11 PNs in the Virgo cluster core, nearly half way between M87 and M86. The expansion velocity of a planetary nebula (PN) is one of the most impor- tant parameters determining its evolution, but currently it is known only for a few hundred GalacticPNs, mostly bright objects inthe Milky WayDisk (Gesicki & Zijlstra2000), andfor a few tens in the Magellanic Clouds and M31. When interpreted using dynamically evolving nebular models (e.g. Sch¨onberner et al. 2005), PN shell expansion velocities provide reliable extimates of PN dynamical ages and distance estimates for Galactic PNs. In this letter, we give a brief summary of our Observations in Section 2. In Section 3, we present the PN [OIII] 5007 ˚A half-width half maximum velocity measurements v , HWHM and m(5007) magnitudes as defined by Jacoby (1989). In Section 4 we build the PN lumi- nosity function (PNLF) for the spectroscopically confirmed PN sample in Virgo and for the subsample bound to the M87 halo. In Section 5, we then estimate the outer radii of the M87 halo PNs from the observed PN expansion velocities and their visibility lifetimes. We finally discuss in Section 6 the remaining distance uncertainties, the shape of the PNLF in the M87 halo, and the mechanisms that may be responsible for the large core masses of the brightest PNs in the halo of M87. 2. Observations Data were acquired in service mode (22 hrs, 076.B-0086 PI: M. Arnaboldi), with the FLAMES spectrograph at UT2 on VLT in the GIRAFFE+MEDUSA configuration. We used the high-resolution grism HR 504.8, covering a wavelength range of 250 ˚A centered on 5036 ˚A and a spectral resolution of ∼ 20000. The redshifted [OIII] emissions of PNs in the Virgo cluster core fall near the center of the grism transmission. With this setup, the instrumental broadening of the arc lines is FWHM = 0.29 ˚A or 17 kms−1, and the error on the wavelength measurements is 0.0025 ˚A or 150 ms−1 (Royer et al. 2002). A total of three FLAMES plate configurations were produced for the Virgo core fields F4 and F7 surveyed by Feldmeier et al. (2003). The exposure times were based on the signal-to-noise – 3 – ratio (S/N) estimate for detecting the [OIII] 5007 ˚A line flux of 4.2 × 10−17ergs cm−2 s−1, i.e., m(5007) = 27.2, with S/N ∼ 5. The data reduction was carried out with the GIRAFFE pipeline, for the CCD prereduction, fiber identification, wavelength calibration, geometric distortion corrections, co-addition, and extraction of the final one-dimensional spectra. Data analysis and the velocity measurements are presented in Doherty et al. (2008). A total of 12 PNs were confirmed spectroscopically. In the extracted PN spectra, the [OIII] 5007 ˚A emissions of 11 PNs have measured FWHM between 0.4 and 0.7 ˚A; the comparison with the arclines’ FWHM (0.29 ˚A) indicates that these [OIII] emission lines are resolved. Monte Carlo simulations give an error on the line width measurements of 0.0025˚A for a S/N per pixel of ∼ 15 (Royer et al. 2002). 3. PN expansion velocities The spectroscopic expansion velocity measured from the [OIII] 5007˚A line profile ap- 1 pears to be representative for the material velocities associated with the PN bright rim , and is systematically lower than the expansion speed of the PN shell’s outer radius, according to 1D hydrodynamical simulations (Sch¨onberner et al. 2005). The shell expansion velocities are associated with fainter structures at larger radii, which have been measured only for Galactic PNs (Corradi et al. 2007). From the FLAMES PN spectra, we are able to measure the spectroscopic expansion velocities from the resolved [OIII] 5007˚A emissions for 11 PNs in the Doherty et al. (2008) sample; high S/N spectra for 4 PNs are shown in Fig. 1. We use the “half-width at half maximum” of the [OIII] 5007 ˚A line, corrected for the instrumental half width, v , as HWHM measurement of the spectroscopic expansion velocity for the PN bright rim. According to Sch¨onberner et al. (2005) and Corradi et al. (2007), these v measurements must be HWHM multiplied by a factor of about two to get an estimate of the PN’s true expansion velocity. When it is possible, we also measure the spectroscopic expansion velocities for [OIII] 4959 ˚A. When double peaks or secondary peaks are present in the resolved line profiles, their counts are consistent with the noise. The average [OIII] spectroscopic expansion velocity for this PN sample is v¯ = 16.5 kms−1, with an RMS dispersion of 2.6 kms−1. Both values HWHM are significantly smaller than those determined for samples of PNs observed in the Galactic bulge (Gesicki & Zijlstra 2000). The Virgo PNs might represent the [OIII] brightest part of the Galactic bulge population, or they might come from an entirely different population; this issue requires further work. 1The bright rim is the thin shell well behind the outer shock, enclosing the wind-blown cavity of a PN. – 4 – Based on LOS velocities, we divide the current dataset in two subsamples: 5 PNs are associated with the halo of M87, i.e. belong to a narrow velocity peak in the LOS velocity distribution (LOSVD) centered at the M87 systemic velocity, with RMS = 78 kms−1 (Doherty et al. 2008), and 6 are “free-flying” cluster PNs, with 300 − 1500 kms−1 velocity differences. The distributions of the v measurements for the two subsamples HWHM are similar within the limited statistics, see Fig. 2; the KS-test gives only a 55% probability that the two subsamples are different. The v vs. m(5007) plot in the top panel of HWHM Fig. 2 shows that the PNs within 0.7 mags of the bright cut-off have v < 20 kms−1. HWHM This is in agreement with the predictions of the spatially integrated line profile v vs. HWWM m(5007) computed with the nebular 1D hydrodynamics code of Sch¨onberner et al. (2007, and in prep.) for central star masses in the range 0.696−0.625 M . ⊙ When a PN’s distance isknown, its expansion velocity andsize canbe used todetermine its dynamical age. This age, in turn, can be combined with information about the central star’s effective temperature to yield an estimate of the core mass (Gesicki & Zijlstra 2007). In our case we cannot measure the outer radii for the Virgo PNs, as they are unresolved at a distance of 15 Mpc. Our approach is then the reverse: we shall estimate the central star masses from the [OIII] 5007 ˚A fluxes, derive the PN lifetimes t from central star evolu- PN tionary tracks and nebular 1-D hydrodynamical models, and infer their physical dimensions as r = v ×t , where v = 2×v (Schoenberner et al., in prep.). PN exp PN exp HWHM 4. The M87 halo PNLF The PNLF for the spectroscopically confirmed PNs in the Virgo core region around M87 is shown in Figure 3. The brightest PNs have m(5007) in the range 25.7−26.5. PNs with similar magnitudes were also detected by Ciardullo et al. (1998) in the outer regions of M87. Based on a sample of 338 PNs, Ciardullo et al. (1998) determined a brightening of the PNLF cutoff of 0.37 mag for the PN subsample at R > 4′ with respect to the sample inside R ≤ 4′. They interpreted the brightening of the PNLF in the M87 outer halo as due to a population of Virgo ICPNs filling the elongated volume of the Virgo cluster (up to 4 Mpc along the LOS). The number of foreground PNs would be roughly proportional to the area of the field, and therefore be largest in the outer regions of the surveyed field. In our spectroscopic PN sample we can bypass the distance ambiguity by selecting the 2 subsample of 14 PNs bound to M87. These PNs are associated with a narrow peak in the LOSVD centred at the systemic velocity of M87, see Fig. 3. Their average velocity 2 5 PNs from the Doherty et al. (2008) sample and 9 from the F3 field from Arnaboldi et al. (2004) – 5 – v = 1306 kms−1 and the RMS = 117 kms−1. The empirical PNLF for the M87 halo LOS PNs is shown in Fig. 3: their m(5007) magnitudes are in the range 26.2−27.2, indicating a slightly brighter cut-off than for the PN population in the central R ≤ 4′ region of M87. However we can now say that this is an intrinsic property of the PN halo population, as we now know that these 14 PNs are all at the distance of M87 within ∼ 100 kpc (see also §6). 5. PN core masses, visibility time scales and outer radius From the measured m(5007) of the M87 halo PNs, we estimate the central star total luminosity. Both models and observations indicate that no more than 10% of the central star’stotalluminositycomesoutinthisline(Jacoby1989). Thereforetheintrinsicluminosity of the post-AGB star that powers an m(5007) = 26.2 PN at 14.5 Mpc must be > 6930 L ⊙ and, based on the evolutionary tracks of Bl¨ocker (1995), we thus obtain a central core mass larger than 0.6 M . Using a 1-D hydrodynamics code and the stellar evolution tracks of ⊙ (Bl¨ocker 1995), Sch¨onberner et al. (2007) computed the nebular physical parameters and evolution of m(5007) as function of nebular age, from near the AGB phase to the white dwarf cooling tracks. The nebular tracks of Sch¨onberner et al. (2007, Fig. 15) that reach the brighest 0.5 mag of the PNLF, and have a central core mass of > 0.60M as the Virgo ⊙ PNs, have very short visibility lifetimes, t < 2.0×103 yrs. From our measurements and PN v = 2×v , we can then determine the outer radii r of the brightest PNs in the exp HWHM PN M87 halo to be ∼ 0.07 pc. The nebular shells of these PNs are compact and similar to those observed for the brightest (logL ≥ 3.8) Galactic Bulge PNs. 6. Discussion Distance uncertainties and PNLF brightening - Webrieflyconsider thequestion whether the bright PNe in the velocity range bound to M87 could be foreground objects. The mean heliocentric velocity of the Virgo cluster is < v > = 1050 ± 35 kms−1 (Binggeli et al. ⊙ VC 1993). M87 is redshifted by about 300 kms−1 (vM87 = 1307 kms−1) with respect to the Virgo cluster mean velocity, and is falling into Virgo from in front (Binggeli et al. 1993) towards M86 (Doherty et al. 2008). PNs at the cutoff m∗ = 26.33 with apparent magnitudes 26.2-26.0 would be between ∼ 1 − 3 Mpc in front of M87. Diffuse light stars at these locations would be in the infall region towards Virgo where the infall velocities are of order 1000 kms−1 and vary rapidly with distance (Mohayee & Tully 2005, and references therein). On the other hand, radial – 6 – velocities of ∼ 1300 kms−1 would be expected for foreground objects ∼ 6 Mpc in front of the Virgo core. It is thus very unlikely that a distribution of foreground PNe should be observed in our fields at exactly the systemic velocity of M87, with a dispersion of only ∼ 100 kms−1, and less likely still that this population should be projected onto the M87 halo inside 150 kpc but not be observed in the adjacent Virgo core region covered by our data. While the Virgo cluster has a significant depth (∼ ±2Mpc), it is unlikely that the diffuse light and ICPN distribution are similarly elongated. The search for ICPNs outside the Virgo cluster core has given negative results (Castro et al. 2008, in prep.), and the deep photometry of Mihos et al. (2005) shows that the diffuse light is mostly associated with the giant elliptical galaxies, M87, M86 and M84, whereas its surface brightness decreases sharply at larger radii. Observations of the diffuse light in z = 0.25 galaxy clusters also show that it is more centrally concentrated than the cluster galaxies (Zibetti et al. 2005), in agreement with cosmological simulation of cluster formation (Murante et al. 2007). Thus we can safely conclude that the selected M87 halo PNs are bound to M87, and that their bright [OIII] magnitudes are an intrinsinc property of this PN population. Mechanism leading to large core masses in the M87 halo PNs - The initial mass-final mass relation (IFMR) for solar metallicity stars predicts that the PN progenitors with ∼ 2.2 M give finalcore masses of0.62 M (see Ciardullo et al.2005;Buzzoni et al.2006). Turnoff ⊙ ⊙ masses of ∼ 2 M belong to populations with ages ∼ 1 Gyr (Iben & Laughlin 1989). The ⊙ question is whether such populations exist in the M87 halo or in the Virgo diffuse stellar 2 light. Observations intheV,Ibands ofthestellar populationina 11.39arcmin fieldhalfway between M87andM86werecarriedoutwiththeAdvancedCameraforSurvey (ACS) andthe HST.This locationfallswithin theF4VirgofieldofFeldmeier et al.(2003) andis included in the area surveyed by Doherty et al. (2008). At this location, Williams et al. (2007) detected some ∼ 5300 intracluster red giant branch stars (IRGB); from the color magnitude diagram (CMD), they estimated the age and metallicity distribution of the parent stellar population. In this region, 70% - 80% of the Virgo IRGB are old (> 10 Gyr), and span a wide range of metallicities (−2.3 < [M/H] < 0.0), with a mean value of [M/H] ∼ −1.0. From the number of PNs within 0.5 mag of the PNLF cut-off, N0.5, we can determine how much luminosity would be present in an intermediate age population and compare it with the measured surface photometry of Mihos et al. (2005) in the surveyed region and the Williams et al. (2007) fit to the IRGB CMD. If we assume the analytical formula of Ciardullo et al. (1989) for the PNLF, then N0.5 = NPN/100, where NPN is the total number of PNs associated with the luminosity of the parent stellar population. The luminosity- specific PN number α = N /L is given by Buzzoni et al. (2006) for stellar populations PN gal with different ages and metallicities, and calibrated using the PN population in the Local – 7 – Group, Leo group, and the Virgo and Fornax clusters. The maximal theoretical value of α which gives the largest PN population for a given luminosity is α = 1PN × (1.85 × max 106L )−1. This theoretical maximal value is independent of metallicity, and also provides an ⊙ upper limit to the observed α for the PN population in different galaxy types (Buzzoni et al. 2006). IntheM87halo,wehaveN0.5 = 9±3,whichgivesatotalpopulationofNPN = 900anda minimal bolometric luminosity of the parent population of L = N /α = 1.66×109L gal PN max ⊙ 2 over a total surveyed area of 570 arcmin . We can thus derive a lower limit to the mean surface brightness in the V band, µ = 28.6 mag arcsec−2, for a possible intermediate age V parent population. Comparing with µ = 28.3 measured by Mihos et al. (2005) at the V position of the Williams et al. (2007) field, we find that, to justify the number of bright PNs, at least 50% of the stars in this field would need to come from a 1 Gyr population. However, the upper limit to the contribution of a younger (< 10 Gyr) component to the IRGB stars given by Williams et al. (2007) is 30% - 20% . We must then conclude that there are not enough intermediate age stars in the Williams et al. (2007) field to justify the observed N0.5. Whatarethepossiblealternatives, ifanintermediateagepopulationisnotpresent? The Williams et al. (2007) results indicate that the Virgo core population of stars is dominated by low metallicity stars ([M/H]≤ −1) with ages > 10 Gyr; thus we may argue that the halo of M87 may be also metal poor. In this case, the PNLF might have a bright cutoff that is different from that of a metal rich population. However Dopita et al. (1992), and PNLF observations in metal-poor galaxies (Ciardullo et al. 2002) show that metal poor systems have values of M∗ that are fainter than those of their metal-rich counterparts. Also the Oxygen abundance measurements by M´endez et al. (2005) of the brightest extragalactic PNs in NGC 4697 indicate near solar metallicities for these stars. Therefore PN evolution from a metal poor population is unlikely to be a viable explanation for the observed large core masses in the M87 halo PN. The evolution of single stars from a 10 Gyr old population leads to central star masses in the range 0.52 <M< 0.55 M (Buzzoni et al. 2006), which cannot ⊙ supply the [OIII] 5007˚A flux required at the PNLF bright cut-off of M∗ = −4.48. This led Ciardullo et al. (2005) to propose an alternative form of evolution, i.e close binaries and blue stragglers stars, as the likely progenitors of [OIII] - bright PNs in a 10 Gyr old stellar population. This evolutionary channel seems better in agreement with the observations for thebrightestPNsintheM87halo,thaneitherthe∼ 1Gyr-oldorthemetal-poorprogenitors. – 8 – 7. Conclusions We have measured the nebular [OIII] 5007˚A spectroscopic expansion velocities for 11 PNs in the Virgo cluster core, at 15 Mpc distance, with the FLAMES spectrograph on the ESO VLT. Based on the [OIII] line profile width, the average spectroscopic expansion velocity for this sample is v¯ = 16.5 kms−1 (RMS = 2.6 kms−1), which is in agreement HWHM with the predictions of dynamically evolving nebular models for high density nebulae close to their maximal m(5007) emission. Large central star masses M > 0.6 M are inferred CS ⊙ from the bright measured [OIII] luminosities and the known distance for the PNs bound to the M87 halo. From the large central star masses and the measured v , we derive HWHM short PN visibility times and small nebular outer radii, ∼ 0.07 pc. The PNs in the M87 halo have large central star masses, are compact and their nebula shells may be similar to those observed for the brightest Galactic Bulge PNs. Three mechanisms are reviewed as possible explanation for the large core masses of the M87 halo PNs: intermediate age population, metallicity effects, and blue stragglers, with the latter being the most likely, given the old age and the low metallicities of the IRGB stars in the Virgo core (Williams et al. 2007). We thank the referee, Dr Detlef Sch¨onberner, for sharing his results before publication and for his constructive comments. We thank Nando Patat and Marina Rejkuba for support. Facilities: ESO VLT, HST (ACS). REFERENCES Arnaboldi, M., et al. 1996, ApJ, 472, 145 Arnaboldi, M., et al. 2003, AJ, 125, 514 Arnaboldi, M., et al. 2004, ApJ, 614, L33 Binggeli, B., Popescu, C.C., Tammann, G.A. 1993, A&AS, 98, 275 Bl¨ocker, T. 1995, A&A, 299, 755 Buzzoni, A., Arnaboldi, M., & Corradi, R.L.M. 2006, MNRAS, 368, 877 Corradi, R.L.M., Steffen, M., Sch¨onberner, Jacob, R. 2007, A&A, 474, 529 Ciardullo, R., et al. 1989, ApJ, 339, 53 Ciardullo, R., et al. 1998, ApJ, 492, 62 – 9 – Ciardullo, R., et al. 2002, ApJ, 577, 31 Ciardullo, R., et al. 2005, ApJ, 629, 499 Doherty, M., et al. 2008, A&A, Submitted Dopita, M.A. et al. 1992, ApJ, 389, 27 Feldmeier, J.J., et al. 2003, ApJS, 145, 65 Gesicki, K, & Zijlstra, A.A. 2000, A&A, 358, 1058 Gesicki, K, & Zijlstra, A.A. 2007, A&A, 467, L29 Iben, I. Jr., & Laughlin, G. 1989, ApJ, 341, 312 Jacoby, G., 1989, ApJ, 339, 39 M´endez, R.H., et al. 2005, ApJ, 627, 767 Mihos, J.C., et al. 2005, ApJ, 631, L41 Murante, G., et al. 2007, MNRAS, 377, 2 Mohayee, R., & Tully, R.B. 2005, ApJ, 635, L113 Royer, F., et al. 2002, Proc. SPIE, 4847, 184 Sch¨onberner, D., Jacob, R., Steffen, M. 2005, A&A, 441, 573 Sch¨onberner, D., Jacob, R., Steffen, M., Sandin, C. 2007, A&A, 473, 467 Zibetti, S., White, S.D.M., Schneider, D.M., & Brinkmann, J. 2005, MNRAS, 358, 949 Williams, B.F., et al. 2007, ApJ, 654, 835 This preprint was prepared with the AAS LATEX macros v5.2. – 10 – Fig. 1.— Resolved [OIII] 5007 ˚A emission lines of 4 spectroscopically confirmed PNs.