ABUNDANCES OF RED GIANTS IN OLD OPEN CLUSTERS. II. BERKELEY 17

The Astronomical Journal, 129:2725 2730, 2005 June # 2005. The American Astronomical Society. All rights reserved. Printed in U.S.A. A ABUNDANCES OF RED GIANTS IN OLD OPEN CLUSTERS. II. BERKELEY 17 Eileen D. Friel 1,2 Division of Astronomical Sciences, National Science Foundation, Arlington, VA 22230; efriel@nsf.gov Heather R. Jacobson Department of Astronomy, Indiana University, Bloomington, IN 47405; hrj@astro.indiana.edu and Catherine A. Pilachowski Department of Astronomy, Indiana University, Bloomington, IN 47405; catyp@astro.indiana.edu Receivved 2005 January 3; accepted 2005 February 28 ABSTRACT We present an analysis of high-dispersion echelle spectra of three giant stars in the 10 Gyr open cluster Berkeley17 obtained with the KPNO 4 m Telescope. Abundances were determined relative to the bright, somewhat metal-poor disk giant Arcturus. Be 17 is found to have a mean ½Fe/HŠ ¼ 0:10 0:09 (rms). Oxygen abundances, determined from the forbidden [O i] lines, and the -elements Mg, Ca, and Ti show scaled solar abundance ratios, while the Si abundance is slightly enhanced. The odd-z elements Na and Al are significantly enhanced relative to scaled solar abundances. These abundance patterns are similar to those seen in other old open clusters and suggest that the Galactic disk was enriched to solar abundance levels at very early times. Key words: Galaxy: abundances open clusters and associations: individual (Berkeley 17) stars: abundances Online material: machine-readable table 1. INTRODUCTION This second paper presents the results for the oldest open cluster known, Berkeley 17. The paper is organized as follows: Open clusters are critical probes of the Galactic disk. They have Section 2 discusses the observations and the properties of the been used to trace star formation, spiral structure, Galactic abundance gradients, and age distributions. While most open clusters cluster; x 3 describes the abundance analysis; x 4 provides a discussion of the results; and in x 5 we summarize the results and are no more than a few hundred million years old, a small number manage to survive to great ages, some reaching ages of 8 10 Gyr. offer some conclusions. These old open clusters offer a special opportunity to examine the early evolution of the Galactic disk. Their ages are more readily 2. OBSERVATIONAL DATA determined than individual field stars, so the cluster population 2.1. Properties of Berkeley 17 has the potential to serve as a time line for study of the evolution of properties of the Galactic disk. Be 17 is among the oldest, if not the oldest, of the open clusters There have been relatively few high-resolution studies of known. It is located at a distance of 2.7 kpc in the direction older open clusters, and these have concentrated primarily on l ¼ 176 and b ¼ 3N6, placing it 11.2 kpc from the Galactic the light element Li and on Fe, C, and N (Brown 1987; Friel & center and 170 pc below the plane. Boesgaard 1990). Only a handful of studies have explored the Various photometric studies (Janes & Phelps 1994; Kaluzny oxygen or -element abundances, and even fewer the heavier 1994; Phelps 1997; Carraro et al. 1999a) have arrived at age r- ands-process elements (Gratton & Contarini 1994; Brown et al. determinations that range from 7 to 12 Gyr. Based on the photometry of Phelps et al. (1994), Janes & Phelps (1994) used the 1996; Tautvaisiene et al. 2000; Bragaglia et al. 2001; Peterson & Green 1998; Pasquini et al. 2004). morphology of the color-magnitude diagram (CMD) to determine We have begun a study of abundances of red giants in the oldest of the open clusters. Our goals have been to refine the Fe an age of 12.6 Gyr on their morphological age indicator scale, making Be 17 the oldest open cluster known and giving it an age abundances and to concentrate on the oxygen and -elements that overlaps those of globular clusters on the same age scale. because of their importance in relating the old open clusters to They also found a reddening E(B V ) of 0.72 on the basis of a the disk field and halo populations and the impact they may have comparison of the observed color of the red giant clump to a fiducial intrinsic value. on cluster age determinations. The first paper in this series presented results for Cr 261 (Friel et al. 2003), an 8 Gyr old southern Kaluzny (1994) carried out a deep BVI photometric study of hemisphere cluster. Be 17 and analyzed the CMD by comparison to other old open clusters. The conclusions he could draw were limited by lack of information on the metallicity of the cluster, but the CMD morphology indicated that E(V I ) 0:7 and (m M ) v 15:0if 1 Any opinions, findings, and conclusions and recommendations expressed in this material are those of the author and do not necessarily reflect the views of Be 17 has the same age as NGC 6791. The relative color difference between the positions of the turnoff and red giant branch the National Science Foundation ( NSF). 2 Visiting Astronomer, Kitt Peak National Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA), Inc., under cooperative agreement with the NSF. NGC 6791. indicated that Be 17 is slightly older and/or more metal-poor than 2725

2726 FRIEL, JACOBSON, & PILACHOWSKI Vol. 129 TABLE 1 Berkeley 17 Stars Observed ID a ID b V V I V K J K UT Date S/N V rad (km s 1 ) 265... 117 14.91 1.88 4.270 1.036 1996 Dec 1, 2 80 74.0 362... 82 14.55 1.79...... 1996 Nov 29, Dec 1 100 43.8 569... 4607 13.89 1.94 4.506 1.088 1996 Nov 28 75 74.3 1035... 79 14.49 2.10 4.802 1.161 1996 Dec 2 80 72.8 a Identification from Phelps et al. (1994). b Identification used in WEBDA ( http://obswww.unige.ch/webda). Phelps (1997) studied the cluster in BVI and derived reddenings of E(B V ) ¼ 0:52 0:68 or E(V I ) ¼ 0:61 0:71, a distance modulus of (m M ) 0 ¼ 12:15 0:10, and an age of 10 13 Gyr by comparing the cluster to stellar evolutionary tracks. Carraro et al. (1999a) analyzed the photometry of Kaluzny (1994) and Phelps (1997) with new theoretical models to conclude that good fits to the data were possible for the parameters E(B V ) ¼ 0:55 0:67, E(V I ) ¼ 0:64 0:78, (m M ) 0 ¼ 12:13 0:07, and an age of 9 1 Gyr. Be 17 was also studied in JK photometry by Carraro et al. (1999b), who derived a similar age and distance with a reddening of E(J K ) ¼ 0:3 0:1, which transforms to E(B V ) ¼ 0:58. Although these ages are somewhat smaller than that derived by Phelps, it is clear that Be 17 is among the oldest, if not the oldest, open cluster known in the Galaxy. Only Collinder 261and NGC 6791 have similarly large ages, with determinations that range from 7 to 11 Gyr. The cluster has also been the subject of low-resolution spectroscopic observations by Friel et al. (2002). They determined overall metallicities from spectroscopic indices measuring primarily Fe and Fe-peak blends. They observed 25 stars in the field of Be 17 with colors appropriate for red giants in the cluster. On the basis of their radial velocities, which are accurate to only 10 km s 1, they concluded that 13 of the stars were members or likely members of the cluster; these stars gave a mean cluster velocity of 84 11 km s 1. They also derived a mean cluster metallicity of ½Fe/HŠ ¼ 0:33 0:12 (s.d.) assuming a cluster reddening of E(B V ) ¼ 0:58. The Friel et al. (2002) study is the source of targets for the echelle observations discussed here. Four stars were selected from the survey that had velocities and overall metallicities close to the cluster mean. The photometric properties of these stars are summarized in Table 1. 2.2. Description of Data Observations of the four target stars in the field of Be 17 were obtained with the echelle spectrograph on the 4 m telescope at KPNO on the nights of 1996 November 28 through December 2 (UT). The data have a resolution of R 25;000 and a typical signal-to-noise ratio (S/N) of 80 or higher. Details of the observations and published photometry for the stars are given in Table 1. Conditions through the run varied, and some observations were taken through cloud. In all cases, multiple exposures were taken, typically of 30 40 minutes each, and combined with filtering to minimize contamination by cosmic rays. Data were reduced using standard procedures with IRAF 3 (Image Reduction and Analysis Facility). Processing included 3 IRAF is distributed by the National Optical Astronomy Observatory, which is operated by AURA, Inc., under cooperative agreement with the NSF. bias subtraction, division by flat field, and extraction of the individual echelle orders, which were then dispersion corrected and continuum fitted. Along with the cluster observations, two kinds of standard, calibrating stars were observed. Bright radial velocity standards were observed throughout the run, and rapidly rotating, hot stars were observed near the position of the cluster before and after the cluster observations so that the extent of the terrestrial contamination could be assessed and corrected for, particularly in the region of the forbidden [O i] lines at 6300 and 6363 8. 2.3. Stellar Radial Velocities Radial velocities for the Be 17 stars were measured using IRAF RVCOR software, which uses cross-correlation techniques similar to those of Tonry & Davis (1979). After normalization and apodization of the spectrum, the linear wavelength scale is transformed to a log scale, a cross correlation of the spectrum with a template is performed, and a Gaussian is fitted to the upper 50% of the cross-correlation peak. Four radial velocity standards of similar spectral type to the cluster giants were used as templates for cross correlation. Final stellar velocities given in Table 1 are averages of cross-correlation results against several radial velocity standards and from echelle orders that were free of telluric absorption. The typical scatter about the measurements over all orders of a given template is 0.2 0.4 km s 1, and for a given star against several different templates is no more than 0.4 km s 1. Star 362, with a velocity that deviates by 30 km s 1 from the other stars, is not a member of the cluster. The average velocity from the remaining three stars is 73:7 0:8 (s.d.) km s 1. This value is consistent with, but improves on, the value of 84 11 km s 1 from the low-resolution data of Friel et al. (2002). 3. ABUNDANCE ANALYSIS 3.1. Equivalent Widths We have followed the same procedures and methods for abundance analysis and the measurement of equivalent widths as for Cr 261. Here we give only a brief description of the method; further details can be found in our previous paper (Friel et al. 2003). Equivalent widths for the Cr 261 stars were measured from the extracted spectra using the interactive routines in IRAF and fitting Gaussian profiles to each line. The line list we adopted for our study of Cr 261 was already selected to avoid strongly blended lines and we use the same line list here. In the study of Cr 261, we applied an upper limit of 150 m8, and we do the same in our analysis of Be 17 stars. However, we provide measurements for several lines that are only slightly above this limit or are from elements, such as Na, that have very few lines.

No. 6, 2005 ABUNDANCES IN BE 17 2727 We estimate the measurement uncertainty to be approximately 5 6 m8 on the basis of a star-to-star comparison of equivalent widths and a comparison of independent measurements by two of us (E. D. F. and H. R. J.). 3.2. Atmospheric Parameters Initial estimates of atmospheric parameters T eff and log g were obtained from the available photometric data of Phelps et al. (1994), Phelps (1997), Kaluzny (1994), and the Two Micron All Sky Survey (2MASS) catalog. The work of Kaluzny and Phelps provides VI data for all stars, but an examination of their results indicates that the Phelps et al. (1994) photometry is offset from the Kaluzny data by 0.21 mag in Vand 0.06 mag in V I in the sense of being fainter and redder than the Kaluzny values. Phelps (1997) also finds such an offset, and his newer photometry agrees with Kaluzny. However, none of the Be 17 stars we observed were measured in the Phelps (1997) paper. Consequently, we have adopted the Kaluzny (1994) values when available and transformed the Phelps et al. (1994) data by the above offsets when they were the only data available (for star 569). All the Be 17 stars were also observed by 2MASS, and Table 1 includes the J and K magnitudes obtained from the 2MASS catalog. The reddening to Be 17 is large and uncertain, with estimates from CMD analysis ranging from 0.52 to 0.68 in E(B V ) and from 0.61 to 0.78 in E(V I ). Because it was not clear that any part of this range was to be preferred, we began our analysis by looking at the temperatures inferred from the extremes of this range. Adopting the Cardelli et al. (1989) reddening relationships, we explored several color-temperature relationships using dereddened (V I ), (J K ), and (V K ) colors. The temperature calibrations of Cohen et al. (1978), Di Benedetto (1998), and Alonso et al. (1999) were used, as well as the empirical (V I ) calibration based on M67 used in Paper I for comparison. We found the (J K ) and (V K ) colors to give consistent temperatures from all calibrations. The (V I ) colors yielded temperatures that differed significantly from those obtained from the IR colors and between different calibrations. Consequently, we adopted temperatures that were the median of those given by the (V K ) and (J K ) colors. The effective reddening that would yield these temperatures is E(B V ) ¼ 0:60. In all cases, however, the final temperature was found through the spectroscopic abundance analysis, by ensuring that there were no trends in the Fe i abundance with excitation potential (EP). We found it unnecessary to alter the temperatures for stars 265 and 569 from our initial values. For star 1035, the temperature deduced from photometry was 150 K lower than that determined through the spectroscopic analysis. We note that the reddening to Be 17 is large, and it would be quite possible for it to be variable across the cluster field. A change of no more than 0.1 mag from our effective adopted reddening value would lead to a temperature difference of this magnitude. It seems reasonable to think Star TABLE 2 Adopted Atmospheric Parameters T eff (K) log g (dex) v t (km s 1 ) Be 17 265... 4490 2.0 1.2 Be 17 569... 4300 1.5 1.5 Be 17 1035... 4300 1.7 1.5 Arcturus a... 4300 1.5 1.5 a Parameters for T eff and log g from Peterson et al. (1993). Element TABLE 3 Equivalent Widths in Berkeley 17 Stars k (8) EP logg f EW 265 (m8) EW 569 (m8) EW 1035 (m8) Na i... 5682.65 2.10 0.83 160.0 173.4 176.5 Na i... 5688.21 2.10 0.72 152.4 161.5 167.5 Na i... 6154.23 2.10 1.80 86.2 97.3 106.7 Na i... 6160.75 2.10 1.58 112.2 142.1 127.8 Mg i... 5711.09 4.33 1.81 138.8 159.3 151.7 Note. Table 3 is published in its entirety in the electronic edition of the Astronomical Journal. A portion is shown here for guidance regarding its form and content. that intrinsic reddening variations in the cluster could contribute to this apparent discrepancy. A variation in reddening is also not ruled out by the cluster color-magnitude diagram, which shows a fairly broad main sequence, although the presence of binaries and photometric errors would contribute to this scatter as well. However, given that our estimated uncertainty in the adopted stellar temperatures is 100 K, we caution against using the spectroscopic temperatures to draw strong conclusions about the reddening to the cluster. Initial values of surface gravity were determined from the star s position in the CMD, assuming a true distance modulus of 12.15 for Be 17, which is obtained by both Phelps (1997) and Carraro et al. (1999a), a turnoff mass of 1.0 M, and bolometric corrections from Cohen et al. (1978) based on our original estimates of temperature. Again, however, these values were adjusted by satisfying the ionization equilibrium during the abundance analysis. Changes in log g of no more than 0.4 dex from the initial values were necessary, in all cases to lower values. Initial values of microturbulence were those used in our study of Cr 261 for stars of similar temperature and gravity. For stars 265 and 569, the initial values of microturbulence did not require alteration in the spectroscopic analysis. For star 1035, the microturbulent velocity was modified, since the temperature was adjusted to remove trends in equivalent width; the final value adopted was identical to that found for other Be 17 and Cr 261 stars of similar temperature. The final derived stellar parameters are listed in Table 2. We estimate the uncertainties in the spectroscopically determined stellar parameters to be 100 K in T eff, 0.2 in log g, and0.2 km s 1 in microturbulent velocity. Our analysis was made by reference to Arcturus, a red giant of temperature very similar to that of the Be 17 stars. As in our study of Cr 261, we adopted the stellar parameters for Arcturus from the study by Peterson et al. (1993). 3.3. Atomic Parameters and Abundance Determinations We adopt the same line list as that used in our previous work on Cr 261, where it is described in detail. All g f-values are derived by reference to Arcturus, adopting the abundances and atmospheric parameters determined in Peterson et al. (1993). The Arcturus abundances are based in turn on the solar abundances of Grevesse & Anders (1989). The full line list and measured equivalent widths in the Be 17 stars are given in Table 3. The LTE abundance program MOOG (Sneden 1973) was used to determine abundances. Model atmospheres were interpolated from the grid of Bell et al. (1976) for the preliminary T eff and log g values estimated from photometry for Be 17. From these initial calculations, the effective temperature, log g values, and microturbulent velocity were altered by small amounts

2728 FRIEL, JACOBSON, & PILACHOWSKI Vol. 129 TABLE 4 Mean Stellar Abundances Arcturus a Be 17 265 Be 17 569 Be 17 1035 Element A [X/H] [X/H] N [ X/Fe] [X/H] N [X/Fe] [ X/H] N [X/Fe] Fe i... 7.02 0.5 0.00 0.14 43... 0.13 0.16 39... 0.18 0.14 40... Fe ii... 7.02 0.5 +0.02 0.20 9... 0.11 0.13 9... 0.19 0.16 9... Na i b... 6.13 0.2 +0.24 0.12 4 +0.24 +0.23 0.23 4 +0.36 +0.20 0.13 4 +0.38 Na i c... 6.13 0.2 +0.28 0.13 2 +0.28 +0.33 0.30 2 +0.46 +0.28 0.05 2 +0.46 Mg i... 7.48 0.1 +0.02 0.03 3 +0.02 +0.06 0.25 2 +0.19 0.04 0.07 2 +0.14 Al i... 6.37 0.1 +0.17 0.04 3 +0.17 +0.11 0.10 3 +0.24 +0.16 0.08 3 +0.34 Si i... 7.45 0.1 +0.26 0.14 8 +0.26 +0.22 0.15 8 +0.35 +0.12 0.11 8 +0.30 Ca i... 6.16 0.2 0.03 0.08 6 0.03 0.21 0.07 4 0.08 0.20 0.09 4 0.02 Ti i... 4.79 0.2 0.18... 1 0.18 0.23... 1 0.10 0.20... 1 0.02 Ti ii... 4.79 0.2 +0.08... 1 +0.08 0.01... 1 +0.12 +0.04... 1 +0.22 Cr i... 5.17 0.5 +0.15... 1 +0.15........................ Ni i... 5.75 0.5 +0.01 0.09 4 +0.01 0.20 0.07 4 0.07 0.07 0.06 4 +0.11 O i... 8.83 0.1 +0.02 0.07 2 +0.02 0.18 0.07 2 0.05 0.13 0.14 2 +0.05 a Abundances from Peterson et al. (1993). b Abundances computed from all lines. c Abundances computed from lines with equivalent widths less than 150 m8. in an iterative fashion as necessary to remove trends in EP, ionization equilibrium, or equivalent width, respectively, for the Fe i lines. Table 4 presents the abundances obtained for the Be 17 stars, as well as the reference abundances of Peterson et al. (1993) used for Arcturus. (We note that there is a typographic error for the abundance of Mg in the similar table in Friel et al. [2003] for the Cr 261 stars. That error is corrected here.) Figure 1 shows a representative portion of the spectrum of star 569 in Be 17 compared to that of Arcturus. The Arcturus spectrum has been smoothed to match the resolution of the Be 17 observations. Measured lines for several species are labeled. Sensitivities of the derived abundances to errors in each of the atmospheric parameters are given in Table 5 of Paper I. 3.4. Oxygen Synthesis The oxygen abundances were determined by spectral synthesis of the forbidden lines at 6300.3 and 6363.8 8. Spectra for Be 17 stars were divided by the spectrum of a hot, rapidly rotating star taken near the time of observation to correct for possible telluric contamination before the comparison with the synthesized region was made. Spectrum synthesis was performed using the LTE code MOOG, using the stellar parameters derived above and a line list provided by C. Sneden (2003, private communication). Because of the formation of CO, the oxygen abundances we determine depend on the carbon abundances assumed in the spectrum synthesis. As we did in the analysis of Cr 261 stars, we make the assumption that [C/Fe] for these stars is the same as that for Arcturus: ½C/FeŠ¼ 0:06 as found by Smith et al. (2002). The oxygen abundances were evaluated by comparing the observed spectra to synthetic spectra computed with oxygen abundances varying by 0.5 dex about the scaled solar abundance. The final adopted abundances for the Be 17 stars are given in Table 4. An example of such a comparison is shown in Figure 2, where the spectrum of Be 17 569 is plotted along with the spectral synthesis in the region of the k6300 line. We estimate the uncertainty in [O/H] to be 0.15 dex for each line from the syntheses. The uncertainty in the carbon abundance has little effect on the oxygen determination. A difference of 0.3 dex in the carbon abundance from the assumed value of ½C/FeŠ¼ 0:06 would lead to a difference of at most 0.1 dex in oxygen. Variation in the nitrogen abundance by 0.3 dex similarly changes the estimated TABLE 5 Mean Cluster Abundances: Berkeley 17 Element [X/H] [X/Fe] Fe i... 0.10 0.09...... Fe ii... 0.09 0.11...... Na i a... +0.22 0.02 +0.33 0.08 Na i b... +0.30 0.03 +0.40 0.10 Mg i... +0.01 0.05 +0.12 0.09 Al i... +0.15 0.03 +0.25 0.09 Si i... +0.20 0.07 +0.30 0.05 Ca i... 0.15 0.10 0.04 0.03 Ti i... 0.20 0.03 0.10 0.08 Ti ii... +0.04 0.05 +0.14 0.07 Cr i... +0.15... +0.15... Ni i... 0.09 0.11 +0.02 0.09 O i... 0.10 0.10 0.00 0.05 Fig. 1. Region of the spectrum of star 569 in Be 17 (solid line)comparedto that of Arcturus (dotted line, from Hinkle et al.2000). a Abundances determined from all lines available. b Abundances determined from lines with equivalent widths less than 150 m8.

No. 6, 2005 ABUNDANCES IN BE 17 2729 Fig. 2. Comparison of the observed spectrum of star 569 in Be 17 with the result of spectral synthesis in the region of the [O i] k6300 line. The observed spectrum is indicated by the squares. The synthesis is indicated for several oxygen abundances with the dotted line denoting the scaled solar abundance and the solid and the dashed lines denoting oxygen abundances decreased by 0.5 dex and enhanced by 0.5 dex, respectively. oxygen abundance by no more than 0.1 dex. The effects of the neighboring CN and Ni features were also evaluated and found to make no significant contribution to the strength of the oxygen feature. 4. DISCUSSION The average cluster abundances for Be 17 are shown in Table 5, which gives the mean values of [X/H] and [X/Fe] averaged over the three stars in the sample. The third and fifth columns give the 1 dispersions about the mean. In the case of the [X/Fe] values, these are computed from the individual stellar abundance ratios rather than using the mean [Fe/H] for the cluster. 4.1. Iron We find an average cluster iron abundance of ½Fe/HŠ ¼ 0:10 0:09 (s.d.). This value is higher than that from the low-resolution spectroscopy of Friel et al. (2002), who found 0:33 0:13, although just within the combined uncertainties in the determinations. We note that the low-resolution results are sensitive to reddening; an increase in the reddening will yield higher metallicities (0.1 dex increase for each 0.1 mag). The somewhat higher reddening implied by the spectroscopic temperature for star 1035 suggests that some of this difference in metallicity may be due to a too low reddening assumed for some stars. 4.2. Oxygen We find good agreement in the determinations of oxygen abundance among the three Be 17 stars, all showing the solar ratio, with mean ½O/FeŠ¼0:00:05 formally but with measurement uncertainties of 0.15 dex. Most other open clusters also show oxygen abundances in the solar ratio, but this result for Be 17 is particularly interesting because of its large age. A solar-like abundance was also found in Cr 261 and in NGC 6791, the two other open clusters with ages of 7 10 Gyr. Together, they indicate that the products from Type Ia supernovae had contributed significantly to the gas of the Galactic disk already 10 billion years ago. The most notable exceptions to solar-like oxygen abundance are the outer disk clusters Be 29 and Saurer 1, which are found to have enhanced oxygen (Carraro et al. 2004). Their special location in the disk, at galactocentric distances greater than19 kpc, suggests that they may have experienced a different nucleosynthetic history than the majority of normal open clusters that are located within 13 kpc of the Galactic center. 4.3. The -Elements Mg, Si, Ca, and Ti 4.3. The -Elements Mg, Si, Ca, and Ti The open clustersoverall show scaledsolar -element ratios, with a suggestion that [Si/Fe] may be enhanced by 0.1 dex (Friel et al. 2003; Friel 2005). The three Be 17 stars continue this behavior. The cluster mean, averaged over the three stars, is ½/FeŠ¼þ0:10 0:04. This value is largely driven by the [Si/Fe] value of +0:30 0:05 (s.d.). The other elements Mg, Ca, and Ti are all in the solar ratios within the uncertainty. We note that the Ti abundance is particularly uncertain, since it rests on only one line each of Ti i and Ti ii in each star, which show differences of typically 0.2 dex. We attribute this discrepancy to observational error and do not treat it as a sign of errors in the ionization equilibrium. The excellent agreement of the Fe i and Fe ii lines, which rest on measures of roughly 40 and nine lines, respectively, is much more secure. We have taken the average of the Ti i and Ti ii abundances as representative of the overall Ti abundance in Be 17. Overall, the behavior of the Mg, Ca, and Ti abundances is consistent with the solar oxygen ratio, since the -elements and oxygen are thought to originate in massive stars that undergo Type II supernova explosions. However, the enhanced Si abundance possibly indicates a distinct nucleosynthetic site. In fact, although these elements are the product of massive star nucleosynthesis, they are not all the product of the same burning cycles (Woosley & Weaver 1995). Mg is the product of hydrostatic C and Ne burning, while Si and Ca are products of both hydrostatic and explosive oxygen burning that are sensitive to a variety of factors that may vary from star to star. Ti is due mostly to explosive oxygen burning. It is then not surprising that there are variations in the behavior of individual -elements. Perhaps more surprising is the uniformity of behavior in clusters that span such a large range of position and age in the disk. 4.4. Na and Al The Be 17 stars show enhanced values of Na and Al, with cluster averages of ½Al/FeŠ ¼þ0:25 0:09 (rms) and ½Na/FeŠ ¼ þ0:3 to +0.4, depending on which lines are used. Four Na lines are available in our spectra, although two of them yield measurements greater than 150 m8 in all the Be 17 stars. Using all lines available, we obtain [Na/Fe] values of +0.2 to +0.4 for the Be 17 stars. If the strong lines kk5682 and 5688 are omitted, the average [Na/Fe] value increases to +0.3 to +0.45, with a cluster average of +0:40 0:10. Since lines this strong were not used in the analysis of Cr 261 stars, for consistency, we adopt this later value. Enhanced Na and Al are seen in a number of other open clusters. The [Na/Fe] ratio is high in the old clusters NGC 6791 (+0.4) and Cr 261 (+0.48), as well as in several other clusters of intermediate age (e.g. NGC 6819; Bragaglia et al. 2001; Friel 2005). The [Na/Fe] ratio also shows much larger dispersion among open clusters than do the -elements. Enhanced Na and Al are often seen in globular cluster giants, where they are generally correlated with O and Mg depletion. These abundance patterns have been interpreted as a sign of the mixing of processed material to the stellar surface, although Na-O anticorrelations are also seen in turnoff and early red giant branch stars (Gratton et al. 2001), suggesting a primordial origin. The star-to-star abundance variation among the Be 17 stars is

2730 FRIEL, JACOBSON, & PILACHOWSKI very small and shows none of these correlations: nor did the evolved stars observed in Cr 261. However, recent results on abundances in both giants and dwarfs in the open cluster IC 4651 (Pasquini et al. 2004) show Na abundances more than 0.2 dex higher in the giants than in the dwarfs. They interpret this effect as being due to the dredge-up of Na in the giants, since the magnitude of enhancement is consistent with predictions for stars of masses found in this 1 Gyr old cluster. However, the [Na/Fe] enhancements are also predicted to decrease with decreasing mass (El Eid & Champagne 1995) and the higher Na enhancements seen in the lower mass stars of older clusters like Be 17 and Cr 261 are not consistent with this picture. It would be particularly interesting to determine Na and Al abundances in the unevolved stars in these old clusters and to explore other signatures of mixing. 4.5. Cr and Ni Only one line of Cr was measured in the Cr 261 sample, and only that in star 265 was below 150 m8. As a result, the mean cluster abundance for Cr rests on only one line and is quite uncertain. The Ni abundance determination, resting on four lines in each star, is more secure and shows that the Be 17 stars have the solar ratio of [Ni/Fe], with a mean value of +0:02 0:09. 5. CONCLUSIONS We have used high-dispersion echelle spectroscopy to determine abundances of Fe, O, and the -elements Mg, Si, Ca, and Ti for three evolved stars in the old open cluster Be 17. The value of [Fe/H] found, 0:10 0:09, is somewhat higher than determinations based on low-resolution spectroscopy but within the combined observational errors. The oxygen abundance is in the solar ratio, as are the abundances of most of the -elements. The [Si/Fe] ratio is high, with a cluster average of +0.3 dex. The Na and Al abundances are enhanced over solar by 0.3 0.4 dex. These abundances are similar to those found in many other open clusters. It is particularly interesting, however, that they are found in Be 17, with its age of 10 Gyr. Combined with results for NGC 6791 and Cr 261, also of great age, these oxygen and -element abundances suggest that the Galactic disk, at least in the regions populated by these clusters, is enriched very early on by the products of Type Ia supernovae that reduce the originally high element-to-fe ratios in the oldest populations of the Galaxy to those of the sun. There are suggestions, however, that clusters in the very outermost disk, at galactocentric distances greater than 15 kpc, are enhanced in oxygen and -elements (Carraro et al. 2004). Whether these abundances are tracing a general change in the star formation and nucleosynthetic history of the outer disk or revealing the origin of the clusters themselves awaits larger samples in the outer disk, which are fortunately on their way in studies by ourselves and others. The initial stages of this research were supported by grants from the NSF (AST-9300391 and AST-9315300 to E. D. F.). C. A. P. gratefully acknowledges the support of the Daniel Kirkwood Research Fund at Indiana University. This publication makes use of data products from the 2MASS, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by NASA and the NSF. Alonso, A., Arribas, S., & Martinez-Roger, C. 1999, A&AS, 140, 261 Bell, R., Eriksson, K., Gustafsson, B., & Nordlund, A. 1976, A&AS, 23, 37 Bragaglia, A., et al. 2001, AJ, 121, 327 Brown, J. A. 1987, ApJ, 317, 701 Brown, J. A., Wallerstein, G., Geisler, D., & Oke, J. B 1996, AJ, 112, 1551 Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 Carraro, G., Bresolin, F., Villanova, S., Matteucci, F., Patat, F., & Romaniello, M. 2004, AJ, 128, 1676 Carraro, G., Girardi, L., & Chiosi, C. 1999a, MNRAS, 309, 430 Carraro, G., Vallenari, A., Girardi, L., & Richichi, A. 1999b, A&A, 343, 825 Cohen, J. G., Persson, S. E., & Frogel, J. A. 1978, ApJ, 222, 165 Di Benedetto, G. P. 1998, A&A, 339, 858 El Eid, M. F., & Champagne, A. E. 1995, ApJ, 451, 298 Friel, E. D. 2005, in Chemical Abundances and Mixing in Stars in the Milky Way and its Satellites, ed. L. Pasquini & S. Randich (Berlin: Springer), in press Friel, E. D., & Boesgaard, A. M. 1990, ApJ, 351, 480 Friel, E. D., Jacobson, H. R., Barrett, E., Fullton, L., Balachandran, S., & Pilachowski, C. A. 2003, AJ, 126, 2372 (Paper I) Friel, E. D., Janes, K. A., Tavarez, M., Scott, J. E., Katsanis, R., Lotz, J., Hong, L., & Miller, N. 2002, AJ, 124, 2693 REFERENCES Gratton, R., & Contarini, G. 1994, A&A, 283, 911 Gratton, R., et al. 2001, A&A, 369, 87 Grevesse, N., & Anders, E. 1989, in AIP Conf. Proc. 183, Cosmic Abundances of Matter, ed. C. J. Waddington (New York: AIP), 1 Hinkle, K., Wallace, L., Valenti, J., & Harmer, D. 2000, Visible and Near-Infrared Atlas of the Arcturus Spectrum (San Francisco: ASP) Janes, K. A., & Phelps, R. L. 1994, AJ, 108, 1773 Kaluzny, J. 1994, Acta Astron., 44, 247 Pasquini, L., Randich, S., Zoccali, M., Hill, V., Charbonnel, C., & Nordstrom, B. 2004, A&A, 424, 951 Peterson, R. C., Dalle Ore, C. M., & Kurucz, R. L. 1993, ApJ, 404, 333 Peterson, R. C., & Green, E. M. 1998, ApJ, 502, L39 Phelps, R. L. 1997, ApJ, 483, 826 Phelps, R. L., Janes, K. A., & Montgomery, K. A. 1994, AJ, 107, 1079 Smith, V. V., et al. 2002, AJ, 124, 3241 Sneden, C. A. 1973, Ph.D. thesis, Univ. Texas, Austin Tautvaisiene, G., Edvardsson, B., Touminen, I., & Ilyin, I. 2000, A&A, 360, 499 Tonry, J., & Davis, M. 1979, AJ, 84, 1511 Woosley, S. E., & Weaver, T. A. 1995, ApJS, 101, 181