当前位置:首页 >> >>


Janusz Kaluzny Warsaw University Observatory, Al. Ujazdowskie 4, 00-478 Warsaw, Poland e-mail: jka@sirius.astrouw.edu.pl


astro-ph/9505129 27 May 1995

Wojciech Krzeminski Carnegie Observatories, Las Campanas Observatory, Casilla 601, LaSerena, Chile e-mail: wojtek@roses.ctio.noao.edu and Beata Mazur Copernicus Astronomical Center, ul. Bartycka 18, Warsaw, Poland e-mail: batka@camk.edu.pl

BV photometry is presented for 12 RR Lyr variables discovered in the presumably young galactic globular cluster Ruprecht 106. All variables are type RRab, and their periods span a narrow range from 0.574 to 0.652 day. We report also on the discovery of 3 SX Phe variables among the cluster blue stragglers. A likely background RR Lyr variable and two foreground contact binaries were also found in the cluster eld. The reddening of Ruprecht 106 is estimated at E (B V ) = 0:20 based on the (B-V) colors exhibited by the cluster RR Lyr variables at minimum light. Analysis of the period" versus \amplitude" and \period" versus \rise-time" diagrams suggests similar metallicities of Ruprecht 106 and M3. A peak (or bump) is present in the luminosity function of the red giant branch of Ru 106. Its position relative to the horizontal branch is consistent with a cluster metallicity of Fe=H] 1:6.
Subject headings: clusters: globular | stars: variables { color magnitude diagram
1 Based on observations collected at the Las Campanas Observatory of the Carnegie Institution of


1. Introduction
Ruprecht 106 (hereafter Ru 106) belongs to a small group of relatively young galactic globular clusters. Buonanno et al. (1990) and Buonanno et al. (1993) published color-magnitude diagrams of Ru 106 which were used to constrain cluster parameters. They found that the metallicity of Ru 106 is Fe=H] = 1:9 0:2 and that it is 3{5 Gyr younger than other galactic globular clusters with similar metallicities. It has been suggested that Ru 106, as well as some other exceptionally young globulars, may have been captured by the Milky Way from the Magellanic Clouds (Lin & Richer 1992). However, such a hypothesis has been criticized by van den Bergh (1994). In 1991 two of us (WK & BM) used the 2.5-m duPont telescope at Las Campanas Observatory to conduct a preliminary survey for variable blue stragglers in Ru 106. This survey resulted in serendipitious discovery of several candidates for RR Lyr variables. In this paper we present results of follow up observations aimed at a study of the photometric properties of RR Lyr stars in Ru 106. An independent observing run conducted on the duPont telescope was devoted to a detailed study of variable blue stragglers in this cluster. Results of these observations will be published in a complementary paper (Krzeminski et al., in preparation).

2. Observations and data reductions
A eld centered approximately on the cluster center was monitored during 5 nights spanning the period from April 27 to May 01, 1993 (UT). All observations were made using the 1-m Swope telescope at Las Campanas Observatory. A thinned 1024 1024 Tektronix chip with a scale of 0.695 arcsec/pixel, was used as the detector. Preliminary processing of the CCD frames was done with the standard routines in the IRAF-CCDPROC2 package. The at- eld frames were prepared by combining \dome ats" and exposures of the twilight sky. The reduction procedures reduced total instrumental systematics to below 1% for the central 780 780 pixels2 area of the images. Some systematic residual pattern at the 1%-4% level was left near borders of the images, perhaps due to an uneven antire ecting coating. Observations were performed using Johnson B and V lters. The exposure time ranged from 320 to 400 sec for the V-band, and was equal to 500 sec for the B-band. The
of Universities for Research in Astronomy, Inc., under contract with the National Science Foundation.
2 IRAF is distributed by the National Optical Astronomical Observatories, operated by the Association

{3{ observations in both lters were performed on 3 nights out of 5. Each observation in the B lter was followed by two exposures in the V lter. On two nights observations were performed only in the V lter. A full listing of all exposures is given in Table 1 which is available on the CD-ROM supplemented to the Astronomical Journal. Several elds containing standard stars from Landolt (1992) were observed during 3 nights. Independent transformations from the instrumental to the standard BV system were rst derived for these nights. Subsequently we adopted averaged values for the color terms and derived new zero points and extinction coe cients for each night. Photometry of Ru 106 was transformed to the BV system using data obtained on the night of May 1, 1993. The following relations were obtained for that night:

v = const + V 0:019(B V ) 0:168X (b v) = const + 0:913(B V ) 0:085X

(1) (2)

where X is the airmass, and lower-case letters refer to the instrumental magnitudes. In Fig. 1 we show residuals between the standard and the recovered magnitudes and colors for 12 stars from 3 Landolt (1992) elds which were observed immediately before Ru 106 on the night of May 1, 1993. These 3 elds were observed at air masses ranging from 1.17 to 1.86. A total of 232 V and 54 B images were used for photometry of Ru 106. The instrumental photometry was extracted using DAOPHOT/ALLSTAR package (Stetson 1987, 1991). It was found that the point spread function (PSF) showed some positional dependence, and a PSF varying linearly across the image was adopted. The observed eld was rather crowded. Moreover, the analyzed images were poorly sampled due to the large scale of pixels. The FWHM reached 2.0 pixels on the best images. To improve the stability of the derived photometry, we adopted the following procedure. First, we extracted photometry from three V frames, which were judged to be of a particularly high quality. Coordinates of all measured stars were transformed to one common system and a master list containing objects detected on at least two frames was created. A list of stars suitable for calculation of the PSF was also created. Reduction of individual frames started with a derivation of the aperture photometry for the relatively bright stars. These stars were then used to transform the coordinates of stars from the master list to the coordinates of the currently processed frame. Subsequently, aperture photometry was derived for stars from the master list and stars used for determination of the PSF were identi ed. After calculating the PSF for the current frame we ran the ALLSTAR program (Stetson 1987) to obtain pro le photometry for stars from the master list. Such an approach not only saves CPU time but, more importantly, helps to obtain better

{4{ photometry from relatively poor frames a ected by bad seeing or bright sky. A similar procedure was used for the B-band images. Photometry for all frames was transformed to a common instrumental system and nally transformed to the BV system. Data bases for the V and B observations were created and the average magnitudes for stars from the data bases were used to create the color-magnitude diagram (CMD hereafter) for the monitored eld. 5206 and 5208 stars were included in the V and B data bases, respectively.

3. Variable Stars
A search for variables was conducted using the data bases for both lters. For every star which was measured on at least 50% of frames for a given data set, we calculated a 2 statistic. Objects with P ( 2) < 10 4 (eg. Press et al. 1986) were considered candidate variables and their light curves were tested for variability with periods in the range 0.03 to 5 days. To determine the most probable periods we used an aov statistic (Schwarzenberg-Czerny 1989, 1991). To look more e ciently for faint short-period variables we analyzed separately the data from the rst two nights of the run. Photometry obtained on these two nights was particularly deep and accurate due to favorable observing conditions (dark sky and good seeing). In Fig. 2 we present a plot of rms deviation versus the average V magnitude for the light curves which are based on frames taken during the rst two nights of the observing run. The limiting magnitude of the photometry is V 22:3. The median value of rms is about 0.02 at V = 19 and reaches about 0.05 at V = 20:7. We discovered a total of 18 periodic variables located in the eld of Ru 106. Their equatorial and rectangular coordinates are listed in Table 2. The rectangular coordinates correspond to the V-band image which is available on the CD-ROM supplemented to the Astronomical Journal. A transformation from rectangular to equatorial coordinates was derived based on positions of 16 stars from the Guide Star Catalogue (Lasker et al. 1988). Our sample of identi ed variables consists of 13 RR Lyr stars, 2 contact binaries and 3 SX Phe stars. Phased light curves of all variables are shown in Figs. 3 and 4. We did not make any tests to determine the degree of completeness of our photometry of Ru 106. Based on the overall quality of the photometry we are inclined to believe that our sample of RR Lyr variables is complete down to amplitudes of about 0.1 mag. On the other hand, it is likely that we missed some variable blue stragglers. In the following discussion we focus our attention mostly on the RR Lyr variables. Particularly, we use the observed characteristics of these stars to constrain the reddening and metallicity of Ru 106. A dedicated observing session was devoted to study variable blue stragglers hosted by the cluster. Results of that study will be published elsewhere

{5{ (Krzeminski et al., in preparation).

3.1. RR Lyr variables
Table 3 lists the basic characteristics of the light curves of 13 RR Lyr stars discovered in the eld of Ru 106. The mean B and V magnitudes were calculated by numerically integrating the phased light curves after converting them into an intensity scale. Mean values for (B V) were obtained by integrating the color curves. Table 3 also lists the B and V amplitudes of the light curves. V-band amplitudes could be derived for all variables. B-band amplitudes could not be derived for stars V10 and V15 due to incomplete coverage of their light curves. The color at the minimum light is denoted by (B V )min. It was derived by averaging observations during the interval 0:50 < < 0:80 (we follow Sturch's (1966) de nition). The parameter gives the fraction of the period from minimum to maximum light. It was calculated using the V-band light curves. In Fig. 5 we show a section of the color-magnitude diagram for the central part of Ru 106. Variable V2 is located about 1.3 mag below the horizontal branch of the cluster. This suggests that it is a eld object located behind the cluster. Variable V2 was not included in the derivation of cluster reddening and metallicity. The properties of the Ru 106 RR Lyr stars can be used to estimate some basic parameters of the cluster itself. In Fig. 6 we show the positions of the variables on a \period" versus \B amplitude" diagram. Fig. 7 shows \period\ versus \rise-time" . Fiducial relations for RRab stars from globular clusters M3 and M15 are also marked in Figs. 6 and 7. These relations were adopted after Sandage et al. (1981). The \reference clusters" M3 and M15 have, respectively, Fe=H] = 1:66 and 2:15 (Zinn & West 1984). The Ru 106 variables are grouped in Figs. 6 and 7 around relations de ned by the M3 stars. In fact, the majority of Ru 106 stars fall on the metal-rich side of the M3 line indicating that the metallicity of Ru 106 is likely to be Fe=H] 1:66. This is consistent with the results of Da Costa et al. (1992) who obtained Fe=H] = 1:69 0:05 from analysis of the spectra of 7 Ru 106 giants. Sarajedini (1994) derived Fe=H] = 1:61 0:20 from an analysis of the V/V-I CMD of the cluster. Sturch (1966) presented a formula relating the unreddened colors of RRab variables with their periods and metallicities. To calculate the E (B V ) for the Ru 106 variables, we used Sturch's formula in the form given by Walker (1990):

E (B V ) = (B V )min 0:24P 0:056 Fe=H] 0:347:


{6{ This equation holds when the Zinn & West (1984) metallicity scale is used. Individual values of the reddening were calculated for 10 variables after adopting Fe=H] = 1:66. Stars V5 and V9 were dropped from the analysis. For V9 the value of (B V )min is uncertain while E (B V ) derived for V5 deviates signi cantly from the average value obtained for the remaining variables. The derived individual values of E (B V ) range from 0.167 to 0.223 and the average value for the 10 variables is < E (B V ) >= 0:183 0:002. Knowing the reddening we are in position to estimate the cluster metallicity from the color of its red giant branch. Zinn and West (1984) calibrate the intrinsic color of the giant branch at the level of the horizontal branch as: (B V )0;g = 1:16 + 0:23 Fe=H]: (4)

Using the data shown in Fig. 5 we estimate (B V )g = 0:93 0:025. The quoted error includes both uncertainty of the zero-point of the color transformation, and the uncertainty involved in the determination of the ducial relation for the RGB of Ru 106. It is encouraging that the same value of (B V )g comes from the ducial sequence provided by Buonanno et al. (1993). For E (B V ) = 0:18 we obtain (B V )0;g = 0:75 and a cluster metallicity Fe=H] = 1:78 0:11 follows from Eq. 4. It has to be kept in mind that clusters used by Zinn & West to calibrate Eq. 4 show a scatter of = 0:20 around the mean relation. The derived metallicity is by 0.12 lower than the value which was used above to calculate E (B V ) from Eq. 3. In fact Eqs. 3 and 4 can be solved simultaneously yielding a self-consistent values of E (B V ) and Fe=H]. Such a procedure gives E (B V ) = 0:198 0:002 and Fe=H] = 1:87 0:23. Based on the properties of their RR Lyr stars, galactic globular clusters are conventionally classi ed into two groups: Oosterho type I (metal-intermediate objects with < Pab > 0:55 and a predominance of ab over c-type RR Lyr variables) and Oosterho type II (metal-poor objects with < Pab > 0:65 and comparable fractions of c-type and ab-type variables). Recent reviews of this subject and a summary of relevant observational data can be found in Sandage (1993) and in Bono et al. (1994). The average period of the 12 RRab variables discovered in Ru 106 is < Pab >= 0:616 with ab = 0:021. We note that the distribution of periods of the Ru 106 variables is extremely narrow. None of clusters listed in Table 1 of Sandage (1993) has such a small value of ab as Ru 106. Our data also show a complete lack of RRc variables in Ru 106. Only 4 out of 37 clusters listed by Sandage (1993) are devoid of RRc stars. Sandage (1993) devised a method allowing a correctionof the observed mean period values < Pab > to what they would have been if the distribution of variables for a given cluster had been uniform. He obtained:

log < P >= 0:0267(r 1)=(r + 1);


{7{ where r is the ratio of the number of HB stars at the red edge to stars at the blue edge. For Ru106 we may adopt log < P >= 0:027 as its HB lacks any stars on the blue side of the instability strip. The corrected mean period is then log < Pab >corr= 0:237. In Fig. 8 we show relation between corrected average period < Pab > and metallicity which was adopted by Sandage (1993). Positions of clusters used to calibrate that relation are shown together with positions of Ru 106 plotted for Fe=H] = 1:9 0:2 (Buonanno et al. 1993) and for Fe=H] = 1:69 0:05 (Da Costa et al. 1992). The relatively low average period of RRab stars in Ru 106 is consistent with the higher metallicity of the cluster. A formal application of Sandage's calibration:

log < Pab >corr= 0:121

Fe=H] 0:431


leads to Fe=H] = 1:60. When deriving his calibration (Eq. 5 above), Sandage (1993) dropped from consideration clusters with 1:9 < Fe=H] < 1:7. Clusters from this metallicity range tend to have extremely blue horizontal branches. Their RR Lyr variables are on the evolved track on their way to the AGB. Such variables are brighter than the ZAHB and have longer periods than variables on the ZAHB (see also Lee et al. 1990). Clearly this problem is of no concern in the case of Ru 106 as it possesses a red HB.

3.2. Contact binaries and SX Phe stars
Buonanno et al. (1990) noted that Ru 106 hosts a sizeable population of blue stragglers. Two types of variables are known to occur among blue stragglers in globular clusters. These are eclipsing binaries (Hut et al. 1992)) and pulsating stars of SX Phe type (Nemec et al. 1994). Table 4 gives some basic photometric data for the SX Phe stars and contact binaries discovered in the eld of Ru 106. The positions of these variables in the cluster CMD are shown in Fig. 9. Both contact binaries are too bright to be cluster members, and are likely foreground stars. All three SX Phe stars are located in the area of the CMD occupied by cluster blue stragglers and these variables are very likely cluster members. It has to be noted that our photometry of blue stragglers in Ru 106 is rather poor. These stars are relatively faint and most of them are located in the crowded central part of the cluster. The light curves of SX Phe stars obtained in the course of this survey are slightly distorted due to relatively long exposure times as compared with periods of pulsation of the variables. More detailed study of variable blue stragglers in Ru 106 will be published elsewhere (Krzeminski et al. in preparation).

4. The color-magnitude diagram

{8{ Photometry extracted from 232 images in the V band and from 54 images in the B band was averaged to produce a color magnitude diagram (CMD) for the observed eld. Only stars measured on at least 130 V images and on at least 30 B images were retained in the nal CMD. No attempt was made to estimate degree of incompleteness of the photometry. The limiting magnitude of our BV photometry is Vlim 22:2 for stars located on the cluster main sequence. Deeper photometry of Ru 106 was published by Buonanno et al. (1993; Vlim 23:7) and Buonanno et al. (1992; Vlim 22:5). Data presented here cover, however, a much larger eld than in the quoted studies. In Fig. 10 we show the CMD for an 8:6 8:6 arcmin2 eld centered approximately on the cluster center. The eld monitored for variables had a size of 11:6 11:6 arcmin2. However, as we reported in Sec. 2, no precise photometry could be obtained for stars located near the edges of the images. A large fraction of stars in Fig. 10 are eld objects. Despite this complication two interesting features of the CMD can easily be noted. First, the horizontal branch of the cluster lacks any stars located to the blue of the RR Lyr variables. Second, the blue edge of the cluster main sequence is sharply marked below the turno region. This indicates that the metallicity of the background halo population is not lower than the metallicity of Ru 106. Two very blue objects with B V < 0:1 are present in Fig. 10. These stars are located at V 20:3 and V 22:0. If they are cluster members or background objects then their unreddened colors are B V < 0:3 for the adopted E (B V ) = 0:20. The fainter of these blue stars is located at an angular distance r 1:9 arcmin from the cluster center, while the corresponding distance for the brighter star is r 3:4 arcmin. These distances are lower than the tidal radius of the cluster. However, the available data do not allow to reject possibility that both identi ed blue objects are eld stars or unresolved extra-galactic objects. The most remote of the RR Lyr variables belonging to Ru 106 is the star V1, with a projected distance from the cluster center of 4.8 arcmin. This sets a lower limit on the tidal radius of Ru 106. All 3 SX Phe stars reported in this paper are located closer that 2 arcmin from the cluster center. The group of potential blue stragglers visible in the CMD of Ru 106 extends down to V 20:9. To look in more details on the blue stragglers population of the cluster we isolated two groups of stars: those lying within radius R = 2:94 arcmin from the cluster center, and those lying in an outer ring, between 3.31 and 4.43 arcmin. The inner circle and the outer ring cover equal areas on the sky. The CMDs for both selected sub elds are presented in Fig. 11. We limited our attention to stars with V < 21:7 to minimize problems related to the incompleteness of the photometry whose degree depends heavily

{9{ on the distance from the cluster center. As we noted above, the radius of Ru 106 exceeds 4.8 arcmin. Therefore, the CMD for the outer ring is populated by some cluster members. In fact a clearly marked main sequence of Ru 106 is visible on the right panel of Fig. 11. However, at the same time only two blue straggler candidates are visible in the CMD for the outer ring. We used data presented in Fig. 11 to obtain a eld-star corrected CMD for the inner region of Ru 106. The outer ring served as a \comparison eld". For each star from the \comparison eld" a nearest match in the CMD for the inner eld was located. Subsequently, a pair of stars with the lowest separation was removed from both corresponding CMDs. Thi