TWO DISTINCT RED GIANT BRANCHES IN THE GLOBULAR CLUSTER NGC 288

Contrary to the conventional wisdom, observations made during the past decade have revealed that many globular clusters (GCs) possess more than one stellar population. Some of these peculiar GCs, such as ω Cen (Lee et al. 1999; Bedin et al. 2004), M54 (Layden & Sarajedini 1997; Siegel et al. 2007), M22 (Da Costa et al. 2009; Lee et al. 2009a; Marino et al. 2009), NGC 1851 (Han et al. 2009; Lee et al. 2009b; Carretta et al. 2010), Terzan 5 (Ferraro et al. 2009), and NGC 2419 (Cohen et al. 2010; Di Criscienzo et al. 2011), show evidence of supernovae (SNe) enrichment, indicating that they are relics of more massive primeval dwarf galaxies, rather than being normal GCs. For other GCs with multiple populations, such as NGC 2808 (Piotto et al. 2007), NGC 6388 (Moretti et al. 2009), and M4 (Marino et al. 2008), however, the evidence for the discrete distribution of heavy elements as observed in the red giant branch (RGB) of ω Cen is lacking, although spreads in some lighter elements (Carretta et al. 2009, and references therein) and helium (D'Antona et al. 2005; Lee et al. 2005; Piotto et al. 2007; Yoon et al. 2008) are reported. Therefore, the presence of chemical inhomogeneity and multiple populations in these GCs is largely considered to be due to the pollution from the intermediate-mass asymptotic giant branch (AGB) stars and (or) fast-rotating massive stars (Ventura & D'Antona 2008; Decressin et al. 2007), which is expected even in normal GCs.

Because of their important implications on the hierarchical merging paradigm of Galaxy formation, searching for more GCs with dwarf galaxy origin (i.e., with evidence of SNe enrichment) is extremely important. The purpose of this Letter is to report that NGC 288 is also showing a clear split in the RGB from the narrowband calcium photometry. This observation is compared with our stellar population models to argue that the two populations are different in terms of overall metallicity, helium, and age by small amounts.

Our observations in Ca, b, and y passbands were performed using the Cerro Tololo Inter-American Observatory (CTIO) 4 m Blanco telescope on 2009 July 27. The telescope was equipped with eight 2048 × 4096 SITe CCDs, providing a plate scale of 0.27 arcsec pixel⁻¹ and a field of view of 36 × 36 arcmin on the sky. However, since the 4 × 4 inch filters used in our photometry cannot cover the entire field, only four CCD chips located in central region (i.e., chip 2, 3, 6, and 7) covering 2048 × 2500 pixels per chip were used in the final photometry. The total exposure times for the Ca, Strömgren b, and y passbands were 1650, 264, and 132 s, respectively, split into short and long exposures in each band. NGC 288 was placed on chip 6, approximately 3.0 arcmin south and 3.1 arcmin east from the CCD center. The IRAF³ MSCRED Package was used for preprocessing, including bias correction and flat fielding. The brightnesses of objects in NGC 288 were measured with the point-spread function fitting routine DAOPHOT II and ALLFRAME (Stetson 1987, 1994), and aperture corrections were calculated using the DAOGROW (Stetson 1990). Our photometry in the Ca, b, and y passbands were then used to calculate the hk [ = (Ca − b) − (b − y)] index defined by Anthony-Twarog et al. (1991). The Ca filter in the hk index is meant to measure essentially ionized calcium H and K lines, and the hk index is known to be about three times more sensitive to metallicity than the m₁ [ = (v − b) − (b − y)] index is (Twarog & Anthony-Twarog 1995). The same filter set employed in this observation was extensively used by us in our previous investigations of GCs (Rey et al. 2000, 2004; Lee et al. 2009a, 2009b).

Figure 1 shows color–magnitude diagrams (CMDs) of NGC 288 in (b − y, y) and (hk, y) planes. To examine the CMD features more clearly, magnitude error, chi, sharpness, and separation index (Stetson et al. 2003) were used to reject stars with large photometric uncertainty and those affected by blending and adjacent starlight contamination. All of the stars in Figure 1 lie within the chip 6, and therefore our CMDs are not subject to any uncertainty stemming from the possible chip-to-chip variations of the mosaic CCDs. The most remarkable feature of Figure 1 is the presence of two distinct RGBs in the hk versus y CMD. When measured at y = 16.5 mag, the mean separation between the two RGBs is about 0.10 mag in the hk index. The discrete distribution shown in RGB, however, is not apparent in the subgiant branch (SGB). Note also that the RGB split is not shown in the (b − y, y) CMD. This is most likely because the Ca filter in the hk index is much more sensitive to changes in Ca abundance than other color indices like b − y.

Zoom In Zoom Out Reset image size

Given the small foreground reddening value of E(B − V) = 0.03 (Harris 1996)⁴ toward NGC 288, it is very unlikely that the differential reddening caused the double RGBs. Furthermore, in contrast to other color indices, the hk index is known to be insensitive to interstellar reddening, E(hk)/E(B − V) = −0.12 and E(hk)/E(b − y) = −0.16 (Anthony-Twarog et al. 1991). Therefore, if we adopt the reddening of E(B − V) = 0.03, E(hk) = −0.0036 is obtained for NGC 288, which is negligible compared with the separation in hk index (∼0.10 mag) between the two RGBs. Star counts of two subpopulations indicate that the bluer RGB population ("Pop-1") takes about 60% of the whole population, while the redder RGB population ("Pop-2") comprises about 40% of total population, in the magnitude interval y = 16.5 ± 1.5 mag. This ratio is not sensitive to the adopted separation index in our photometry.

In order to better understand the origin of the RGB split in hk index, and to place constraints on the chemical combinations of two subpopulations, we have constructed stellar population models based on the latest version of the Yonsei–Yale (Y²) isochrones (Yi et al. 2008) and horizontal branch (HB) evolutionary tracks (S.-I. Han et al. 2011, in preparation). Readers are referred to Lee et al. (1990, 1994) and Yoon et al. (2008) for the details of our model construction. Figures 2 and 3 present our synthetic CMDs for NGC 288 in (hk, y) and (b − y, y) planes, respectively. Our models were constructed under three different assumptions regarding the chemical enrichment and age spread in NGC 288. First, we assumed that the second generation population (Pop-2; redder RGB) is more enhanced in metallicity and younger (Δ[m/H] ≈ 0.16 dex, Δt ≈ 1.5 Gyr), but not enhanced in helium abundance (hereafter ΔZ + ΔAge model; panel (b) in Figures 2 and 3). These models match well with the observed CMDs from the main sequence (MS) through the RGB in (hk, y) and (b − y, y) planes. Yet, the models fail to reproduce the HB, as the synthetic HBs are too extended in color, including significant numbers of RR Lyraes and red HB stars (HB type⁵ = 0.61), while the observed HB is only mildly extended with mostly blue HB stars (HB type = 0.91). This is because both metal enhancement and younger age in the second population move the HB to red in the CMD (see Lee et al. 1994). Second, we then assumed that both metal and helium abundances are enhanced (Δ[m/H] ≈ 0.16 dex, ΔY ≈ 0.03 dex), but age is constant (hereafter ΔZ + ΔY model; panel (c) in Figures 2 and 3). These models match well with the observed CMDs from the RGB through the HB in (hk, y) and (b − y, y) planes. However, they cannot reproduce the narrow SGB in (hk, y) CMD. Therefore, both ΔZ + ΔAge and ΔZ + ΔY models are in conflict with the observed CMDs of NGC 288.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Finally, we assumed that not only metal and helium abundances are enhanced, but also age is younger in Pop-2 (hereafter ΔZ + ΔY + ΔAge model; panel (d) in Figures 2 and 3). These models are in good agreements with the observations from the MS to the HB. The enhanced metal abundance in Pop-2 makes the RGB split in hk index as observed, while the younger age can explain the narrow and apparently single SGB in (hk, y) CMD. The increase in helium abundance in Pop-2 moves HB bluer (see Lee et al. 2005), almost canceling out the effects by enhanced metallicity and younger age, making the blue HB only mildly extended, with the HB type similar to the observed value (HB type = 0.90). Note that, in our HB simulations, we employ the standard Reimers (1977) mass-loss law and the same mass-loss parameter η for the two subpopulations. Input parameters adopted in our best models (i.e., ΔZ + ΔY + ΔAge model) are listed in Table 1. Our models are computed with the same abundance of [CNONa/Fe] for both Pop-1 and Pop-2. More detailed models including the possible difference in [CNONa/Fe] between the two subpopulations would change the age estimates (see, e.g., Cassisi et al. 2008).

We have shown that the RGB of NGC 288 is split into two distinct sequences. While this is most likely the effect of Ca ii H and K lines, it is important to check whether the CN band at λ ≈ 3870 Å could affect the hk index due to the proximity of the CN band to the blue tail of Ca filter transmission curve. In particular, according to Kayser et al. (2008), lower RGB stars in NGC 288 exhibit CN bimodality that spans about 0.6 dex. In Figure 4(a), we have matched their spectroscopic data with our photometry, where we can see that "CN-strong" stars lie well on the redder RGB sequence, whereas "CN-weak" stars are on the bluer RGB. This further suggests that possible contamination of the CN band to the hk index should be investigated in more detail. It is known that the passband of the narrowband interference filter, such as the Ca filter employed in our photometry, depends on the angle of incidence beam (see Clarke et al. 1975; Lee et al. 2009a). This issue is therefore more relevant when the Ca filter is used with a relatively fast telescope like the prime focus of the CTIO 4 m telescope, where the central wavelength drift of the Ca filter is estimated to be about 15 Å to the shorter wavelength⁶ (see Figure 4(b)).

Zoom In Zoom Out Reset image size

In order to see the influence of the CN band on the hk index in our photometry, we have calculated synthetic spectra using the ATLAS 9 model atmosphere (Castelli & Kurucz 2003) for the RGB star at the magnitude level of HB with T_eff = 4750 K, log g = 2.0 (in cgs unit), v_turb = 2.0 km s⁻¹, and [Fe/H] = −1.6 (see Lee et al. 2009a supplementary information for detail). The typical RGB stars in GCs show an anticorrelation between CN band and CH band strengths and a correlation between CN band and NH band strengths, indicating that the nitrogen controls the CN band strength (Briley & Smith 1993). Therefore, we have compared the synthetic spectra between the stars with normal and enhanced nitrogen abundances (see Figure 4(c)). We obtain Δhk ≈ +0.006 for Δ [N/Fe] = +1.0 dex, which suggests that the influence of the CN band on the hk index would be negligible in our photometry.⁷ Recently, Sbordone et al. (2011) also reached a similar conclusion, finding that CNONa anticorrelations have an effect of at most ∼0.04 mag in the hk color at fixed M_v. Their calculations include not only the direct effects of CN and other absorption bands (NH and CH), but also the effects on the continuum levels in the Ca, b, and y filters.

If the CNONa abundances have only minor effects on the hk index as discussed above, which should be confirmed in the forthcoming works, the RGB split discovered in our hk versus y CMD would indicate a small difference in Ca abundance between the two subpopulations. Since calcium and other heavy elements can only be supplied through SNe explosions, this in turn would suggest that the second generation stars were affected by SNe enrichment, together with the pollution of lighter elements (such as the enhancement of N and the depletion of O) by intermediate-mass AGB stars or fast-rotating massive stars. Spectroscopy of stars in the two distinct RGB sequences is crucial to confirm the small difference in the abundance of heavier elements suggested in our photometry.

We thank the referee for a number of helpful suggestions. Support for this work was provided by the National Research Foundation of Korea to the Center for Galaxy Evolution Research. This material is based upon work supported by AURA through the NSF under AURA Cooperative Agreement AST 0132798, as amended.