DISCOVERY OF WHITE DWARFS IN THE GLOBULAR CLUSTERS M13 AND M22 USING HST ACS PHOTOMETRIC DATA

Abstract

A search for hot and bright white dwarfs (WDs) in the Milky Way globular clusters M13 (NGC 6205) and M22 (NGC 6656) is carried out using the deep and homogeneous V I photometric catalog of Anderson et al. and and Sarajedini et al., based on data taken with the ACS/WFC aboard the Hubble Space Telescope (HST). V versus V − I color-magnitude diagrams (CMDs) of M13 and M22 are constructed and numerous spurious detections are rejected according to their photometric quality parameters qfit(V ) and qfit(I). In the case of M13, further radial restriction is applied to reject central stars with higher photometric errors due to central crowding. From each resultant V versus V −I CMD, sixteen and thirteen WD candidates are identified in M13 and M22, respectively. They are identified as stellar objects in the accompanying ACS/WFC images and are found to be randomly distributed across the central regions of M13 and M22. Their positions in the CMDs are in the bright part of the DA WD cooling sequences indicating that they are true WDs. In order to confirm their nature, follow-up spectroscopic observations are needed.

1. INTRODUCTION

White dwarfs (WDs) are degenerate end products of normal stars undergoing stable nuclear burning in their interiors whose masses are below ∼ 8–9 M⊙ (Fontaine et al. 2001; Kalirai et al. 2009). Therefore, ∼98% of normal stars end their lives as WDs (Kalirai et al. 2009). The search for WDs in Milky Way globular clusters (GCs) began in 1978 by Richer (1978) in the GC NGC 6752, and was continued by Ortolani & Rosino (1987) in the GC ω Centauri (NGC 5139) and Richer & Fahlman (1988) in the GC M71 (NGC 6838). However, clear identifications of WDs in Galactic GCs were first made by Paresce et al. (1995b) in the GC NGC 6397, Paresce et al. (1995a) in the GC 47 Tucanae (NGC 104), Elson et al. (1995) in ω Cen, and most clearly Richer et al. (1995) in the GC M4 (NGC 6121) using the Hubble Space Telescope (HST).

Renzini et al. (1996) derived a distance to the GC NGC 6752 with the HST as (m − M)0 = 13.05 ± 0.1 mag using the WDs in that GC. Richer et al. (1997) analyzed in great detail 258 WDs in the GC M4 and derived the WD cooling age. Piotto & Zoccali (1999) derived WD cooling sequences (WDCSs) for the GCs M10 (NGC 6254), M22 (NGC 6656), and M55 (NGC 6809). Moehler et al. (2000) obtained spectra of five WDs in NGC 6397 and found that the spectral types of four of them are hydrogen-rich DA. Hansen et al. (2002) and Richer et al. (2002) identified the end of WDCS of M4, and Hansen et al. (2002) derived the WD cooling age of M4 as 12.7 ± 0.7 Gyr. Layden et al. (2005) discovered WDs in the GC M5 (NGC 5904) and used them to estimate the distance to M5. Monelli et al. (2005) identified the WDCS of ω Cen, with more than 2000 WDs discovered. Richer et al. (2006) and Hansen et al. (2007) identified the end of WDCS of NGC 6397, and Hansen et al. (2007) derived the WD cooling age of NGC 6397 as 11.47 ± 0.47 Gyr. Kalirai et al. (2012) identified the end of the WDCS of 47 Tuc and Hansen et al. (2013) found that the WD cooling age is younger than that of NGC 6397 (9.9 ± 0.7 Gyr compared to 11.7 ± 0.3 Gyr). Richer et al. (2013) compared the WDCSs of 47 Tuc and NGC 6397 and found that the WDCSs of 47 Tuc and NGC 6397 coincide very well despite the large metallicity difference between them. In contrast, due to this metallicity difference, the main sequences of 47 Tuc and NGC 6397 do not coincide when the color-magnitude diagrams (CMDs) are placed on the same absolute magnitude and reddening-corrected color plane. Bellini et al. (2013) found a double WDCS in ω Cen and explained it in terms of mass and helium abundance differences of WDs. According to Moehler et al. (2004) and Davis et al. (2009), the spectral types of the brightWDs inM4, NGC 6752, and NGC 6397 are all hydrogen-rich DA. Up to now, about ten Galactic GCs have been reported to haveWDs as their members (e.g., NGC 104, 5139, 5904, 6121, 6254, 6397, 6656, 6752, and 6809).

The usefulness ofWDs in the studies of Galactic GCs can be summarized as follows. First, the WDs can be used to derive the WD cooling ages of the GCs independent of stellar evolution theory (Hansen et al. 2002, 2004, 2007, 2013; García-Berro et al. 2014) like in the Galactic open clusters (Richer et al. 1998; Bellini et al. 2010; García-Berro et al. 2010; Bedin et al. 2015) and the Galactic disk (Oswalt et al. 1996; Leggett et al. 1998; Knox et al. 1999). Second, due to GC WDs being at the same distance, they can be used to derive the distance to a GC independently of the mainsequence fitting method or distances that depend on the horizontal-branch (HB) evolution theory (Renzini et al. 1996; Layden et al. 2005). Third, the WDs can be used to derive the initial-final mass relation (IFMR) of the normal stars that end their lives as WDs (Weidemann & Koester 1983; Weidemann 1987; Koester & Reimers 1996; Lee & Sung 1996; Kalirai et al. 2008, 2009; Williams et al. 2009). The IFMR can be used to derive the maximum mass of the WD progenitors (Weidemann & Koester 1983; Lee & Sung 1996; Williams et al. 2009), the amount of mass lost during the final stages of the progenitors stellar evolution, and to reconstruct the initial mass function of a given GC. Fourth, the mass of young WDs in GCs can be used to measure the age of the Galactic halo through the mass-age relation of the newly formed WDs and the Galactic halo (Kalirai 2012). Lastly, the GC WDs can be used to probe stellar evolution theory (e.g., Calamida et al. 2008).

WDs can be classified into DA and DB WDs spectroscopically. The DA WDs have pure hydrogen atmospheres in their envelopes above the core and helium mantle, while the DB WD atmospheres have pure helium (for more detailed spectral classification, see Kleinman et al. 2013). WDs can be further classified into helium-core WDs, carbon-oxygen core WDs, and oxygen-neon core WDs, according to their core types. The composition of the WD cores are determined by the masses and evolutionary paths of their progenitors. The WD core types can be inferred from their masses, determined by spectroscopic methods or gravitational redshifts (e.g., Reid 1996). The majority of WDs have carbon-oxygen cores (for more detailed information about helium-core WDs, see Serenelli et al. 2002; Althaus et al. 2009; Strickler et al. 2009). The progenitors of the helium-coreWDs do not undergo core helium flashes after the red giant branch evolution due to particular evolutionary circumstances, e.g., mass-loss episodes during the binary evolution.

M13 (NGC 6205; R.A.(J2000.0) = 16h 41m 41.24s; Dec.(J2000.0) = +36° 27′ 35.5′′) is a metal-intermediate ([Fe/H] = −1.53) and large Milky Way GC whose distance from the Sun and from the Galactic plane is R⊙ = 7.1 kpc and Z = 4.7 kpc, respectively (Harris 1996, 2010 edition). Its interstellar reddening and apparent distance modulus have been estimated to be E(B −V ) = 0.02 mag and (m−M)V = 14.33 mag, respectively (Harris 1996, 2010 edition).

M22 (R.A.(J2000.0) = 18h 36m 23.94s; Dec.(J2000.0) = −23° 54′ 17.1′′) is a slightly metal-poor ([Fe/H] = −1.70) and nearby large GC in the Galaxy whose distance from the Sun and from the Galactic plane is R⊙ = 3.2 kpc and Z = −0.4 kpc, respectively (Harris 1996, 2010 edition). Its interstellar reddening and apparent distance modulus have been estimated to be E(B −V ) = 0.34 mag and (m − M)V = 13.60 mag, respectively (Harris 1996, 2010 edition).

A search for WDs in Galactic GCs has been undertaken using the deep and homogeneous photometric data of Anderson et al. (2008) and Sarajedini et al. (2007), taken with the ACS/WFC (Advanced Camera for Surveys/Wide Field Channel) aboard the HST. The data is freely available in the public domain. In this way, the feasibility to discover WDs in nearby Galactic GCs using modest size ground-based telescopes (e.g., Mochejska et al. 2002) has been explored. In this study, we report the discovery of hot and bright WDs in the Galactic GCs M13 and M22. This study was originally motivated by evidence seen in the V versus B−V photometric data of M13 in Table 3 of Richer & Fahlman (1986), which were not shown entirely in Figure 2 of that paper. In the case of M22, despite being the fourth closest GC to the Sun following M4, NGC 6397, and NGC 6544, its WDs were not studied in great detail except by Piotto & Zoccali (1999). In contrast, the WDs of M4 and NGC 6397 were studied in great detail. In this context, a search for WDs in M13 and M22 was carried out, resulting in the unexpected discovery of the hot and brightWDs in the central regions of M13 and M22.

Section 2 presents the observational data and analysis and Section 3 reports the characteristics of the CMDs and WD candidates of M13 and M22. Section 4 presents a brief summary of the present study.

 

2. OBSERVATIONAL DATA AND ANALYSIS

We use the photometric catalog of Anderson et al. (2008) and Sarajedini et al. (2007) and accompanying long-exposure HST ACS/WFC images in the F606W band of 65 Galactic GCs, now in the public domain through the STScIMAST archive1. In addition to these data, new data of six outer Galactic GCs by Dotter et al. (2011) are also available in the MAST archive. All the photometric data and accompanying ACS/WFC images of 71 Galactic GCs were downloaded to construct V versus V − I CMDs and to reject spurious detections in the CMDs. All the data were taken with the HST ACS/WFC in F606W and F814W bands and were analyzed by Sarajedini et al. (2007), Anderson et al. (2008), and Dotter et al. (2011). Brief observational information of M13 and M22 is summarized in Table 1.

Table 1Log of Observations for M13 and M22

As Anderson et al. (2008) argued in their study, the photometric data used here contain spurious detections that have to be properly rejected before any rigorous scientific analysis is undertaken. For the present study, rejection criteria that are simple and easy but applicable to various GCs were applied. The rejection criteria were tested for the GCs NGC 1851, M53 (NGC 5024), M4, M92 (NGC 6341), and M54 (NGC 6715), which have various degrees of stellar crowding in their centers, using routines in IRAF, to confirm that photometric data represent real stars in the accompanying images. For these procedures, only the catalog of Anderson et al. (2008) and Sarajedini et al. (2007) was used. The accompanying images in the F606W band were used for the confirmation of the positions of the real stars and the extraction of the pixel information for drawing the profiles of the celestial objects. For the confirmation of the positions of the real stars, the IRAF TVMARK routine was used and the IRAF IMEXAMINE routine was used for drawing the profiles of the celestial objects in order to determine whether a given object is a real star or not. The IRAF IMEXAMINE routine is most appropriate for the seeing conditions of the HST among similar routines in IRAF. However, due to severe crowding, the confirmation process for the real stars was carried out in severalmagnitude intervals. Especially, M53 and M92 were used as testbed cases for defining the rejection criteria. The process of selecting the rejection criteria was a simple but time-consuming job.

To reject spurious detections in M13 and M22, detections with photometric quality parameters qfit(V ) > 0.30 and qfit(I) > 0.30 (or qV > 0.30 and qI > 0.30 according to Anderson et al. 2008) were removed from the photometric data of each cluster for fainter regions in the CMDs. Detections with qfit(V ) = 2.50 and qfit(I) = 2.50 were retained since they were saturated bright stars in the long-exposure frames (smaller photometric quality parameter represents better photometric quality). Given that photometric errors, σV and σI , have severe redundancy with the photometric quality parameters, the photometric error criteria were not used to reject spurious detections in M13 and M22. In the case of M13, in addition to the spurious detection procedure, central stars with cluster-centric distances r < 0.62′ were removed from the photometric data, due to their large photometric errors. These criteria were originally devised to measure the relative ages of M53 and M92 using the V I photometric data of Anderson et al. (2008) and Sarajedini et al. (2007) and to independently verify the result of the relative age measurements of M53 and M92 using the BV photometric data of the Bohyunsan Optical Astronomy Observatory 1.8 m telescope (Cho et al. 2015). Further photometric studies using these photometric data and based on these criteria will be published in the future (e.g., D.-H. Cho et al. 2015, in preparation).

 

3. CMDS AND WD CANDIDATES

Color-magnitude data for M13 and M22 are taken from the V versus V −I photometric data after the rejection of spurious detections as described in Section 2 in the Johnson-Cousins photometric system defined by Landolt (1992). V versus V − I photometric data were acquired from the mF606W and mF814W Vega magnitude system using the transformation equations of Sirianni et al. (2005) and the photometric zero points of Mack et al. (2007). The resultant V versus V − I CMDs of M13 and M22 are shown in Figure 1. In Figure 2, I versus V −I CMDs of M13 and M22 are shown for reference. In Figure 1a, the total number of stars is 97579 and only stars with cluster-centric distances r ≥ 0.62′ are shown. In Figure 1b, the total number of stars is 74783 and all stars with cluster-centric distances r ≥ 0.00′ are shown. Small open circles with a small dot represent sixteen and thirteen WD candidates in M13 and M22, respectively in Figures 1a and 1b.

Figure 1.V vs. V −I CMDs of M13 (left) and M22 (right). In each figure, small open circles with a dot are WD candidates. In (a), only the stars satisfying the condition r ≥ 0.62′ are shown while in (b) all the stars are shown. For M13, the total number of stars is 97579 and for M22, the total number of stars is 74783. In both figures, the closed circles with error bars indicate the mean V magnitude (< V >) and mean V − I color error (< σ(V −I) >) of each V magnitude interval. In the upper magnitude intervals, most mean V − I color errors are smaller than the size of the circles.

Figure 2.Same as Figure 1, but in the I vs. V − I planes. In both figures, the closed circles with error bars indicate the mean I magnitude (< I >) and mean V − I color error (< σ(V −I) >) of each I magnitude interval.

The WD candidates in Figures 1 and 2 were identified as point sources in the accompanying ACS/WFC images using IRAF routines. The WD candidates of M13 are more scattered in the CMDs than those of M22. Some WD candidates in both GCs are isolated while others have brighter nearby optical companions. Generally, the WD candidates of M22 are more isolated than those of M13. Therefore, partial reasons why the WD candidates of M13 are more scattered in the CMDs than those of M22 in Figures 1 and 2 might be as follows. First, the photometric errors due to stellar crowding may be larger in the M13 WD candidates than those of M22. Second, as the average number of photometric measurements is larger for M22WD candidates, their magnitudes and colors may be determined more accurately than those of M13. Coordinates, magnitudes, colors, and their errors for all WD candidates are summarized in Tables 2 and 3.

Table 2More detailed information on the WD candidates in M13 can be found in the original photometric catalog of Anderson et al. (2008) and Sarajedini et al. (2007). a Star number in the original photometric catalog of Anderson et al. (2008) and Sarajedini et al. (2007). b This WD candidate appears to have unusually large σV and σ(V −I), and its σI is 0.051 mag, and is a real star according to the present investigation.

Table 3More detailed information on the WD candidates in M22 can be found in the original photometric catalog of Anderson et al. (2008) and Sarajedini et al. (2007). a Star number in the original photometric catalog of Anderson et al. (2008) and Sarajedini et al. (2007).

The M22 WD candidates shown in Figure 1b are not included in the WD region of M22 in Figure 2 of Piotto & Zoccali (1999). According to Heyl et al. (2015), bright WDs are born in the central regions of GCs and diffuse out into the outer regions by gravitational relaxation and cool down. Since the V versus V − I CMD of the present study is based on the wider field of view ACS/WFC at the very center of M22, these brightWDs can be detected. This was not the case for Piotto & Zoccali (1999), who covered the outer 4.5′ region from the center of M22 using the smaller field of view HST WFPC2 (Wide Field Planetary Camera 2). According to previous studies of the WDs in Galactic GCs, WDs start to appear in the magnitude range MV ≈ 9.5–10.0 mag (or V ≈ 23.0–23.5 mag at the distance modulus of M22) with a cooling age of ≈10 Myr in the outer regions of the GCs. As a result, these bright WDs could not be found in the V versus V − I CMD of Piotto & Zoccali (1999).

In order to estimate the MV and (V − I)0 distributions of the WD candidates of M13 and M22, the V versus V − I CMDs were transformed into the MV versus (V −I)0 CMDs. For this, the apparent distance modulus (m −M)V and the interstellar reddening in V − I, E(V − I) must be known. Here, we adopt the interstellar reddening relations of Dean et al. (1978), Bessell & Brett (1988), and Cardelli et al. (1989) (E(V −I) = 1.250E(B−V )), and the (m−M)V and E(B−V ) values of Harris (1996, 2010 edition). The adopted E(V − I) values are 0.025 mag for M13 and 0.425 mag for M22. The resultant MV versus (V −I)0 CMDs are presented in Figure 3. Absolute magnitude distributions of the M13 and M22 WD candidates are MV ≈ 7.0–9.5 mag and MV ≈ 7.5–9.5 mag, respectively. Mean reddeningcorrected V − I colors are < (V − I)0 > ≈ −0.30 mag for both GCs. According to Table 3 of Bessell et al. (1998), (V − I)0 = −0.30 mag corresponds roughly to Teff ≈ 31,000 K. Thus, mean Teff of the WD candidates is around 31,000 K. Compared to WDs in other GCs, the WD candidates of M13 and M22 are very hot and luminous, similar to the case of ω Cen (Monelli et al. 2005; Calamida et al. 2008). The positions of the WD candidates in M13 and M22 are in the bright part of the DA WD cooling curve with mean mass < > ≃ 0.53 M⊙ as determined by Moehler et al. (2004) for the bright WDs in NGC 6752 and Kalirai et al. (2009) for the bright WDs in M4.

Figure 3.MV vs. (V − I)0 CMDs of M13 and M22. In each figure, small open circles with a dot are the WD candidates discussed in this paper. Only the stars satisfying the condition, r ≥ 0.62′ , are shown for M13 and all the stars are shown for M22. In each figure, the solid line is the DA WD cooling track of Salaris et al. (2010) with = 0.54 M⊙ and the dotted line is the DB WD cooling track of Salaris et al. (2010) with = 0.54 M⊙.

According to Moehler et al. (2004), Davis et al. (2009), and Williams et al. (2009), the spectral types of most brightWDs in Galactic open clusters and GCs are hydrogen-rich DA except for some DB WDs found in open clusters. The WDCSs in the outer regions of GCs start to appear in the MV ≈ 9.5–10.0 mag according to previous studies on GC WDs. In this sense, the WDs in M13 and M22 of the present study are bright (MV ≈ 7.0–9.5 mag). Therefore, we assume that the WDs in M13 and M22 presented in Figures 1–3 are bright DA WDs. Moreover, the WD sequences in Figure 3 are consistent with the cooling track of the = 0.54 M⊙ DA WD by Salaris et al. (2010).

In order to investigate the spatial distribution of the WD candidates in the observed regions of M13 and M22, the M13 and M22 star map are presented in Figures 4 and 5. Because of severe crowding, only stars brighter than V = 20 mag and V = 19 mag are shown in Figures 4 and 5, respectively. In each figure, thick solid circles denote the positions of the WD candidates. According to Figures 4 and 5, the WDs are randomly distributed across the observed regions of M13 and M22. This indicates that these WDs are not spurious detections such as image artifacts or diffraction spikes of bright stars, but real stars.

Figure 4.Observed region of M13. Only stars brighter than V = 20 mag are shown. Thick solid circles show the positions of the WD candidates. North is upward, and East is to the left.

Figure 5.Observed region of M22. Only stars brighter than V = 19 mag are shown. Thick solid circles show the positions of the WD candidates. North is upward, and East is to the left.

 

4. SUMMARY AND CONCLUSIONS

Sixteen WD candidates in the GC M13 and thirteen WD candidates in the GC M22 were discovered using the HST/ACS photometric data of Anderson et al. (2008) and Sarajedini et al. (2007), specifically through the application of a newly devised spurious detection rejection criteria. They were identified as point sources in the accompanying HST/ACS images and their threedimensional profiles were also investigated using IRAF routines. Absolute magnitudes of the WD candidates of M13 are MV ≈ 7.0–9.5 mag and those of M22 are MV ≈ 7.5–9.5 mag. Mean intrinsic colors of the WD candidates of M13 and M22 are < (V −I)0 > ≈ −0.30 mag and correspond roughly to Teff ≈ 31,000 K. Therefore, the discovered WD candidates are luminous and very hot. They are randomly distributed across the observed regions of M13 and M22 suggesting that they are not image artifacts but rather true detections. Given their implied young age, they will be helpful to examine the formation mechanism of WDs in GCs and to investigate the characteristics of the bright WDs in GCs. However, in order to derive their fundamental characteristics, additional follow-up spectroscopic observations are required. Their fainter counterparts, as in other GCs (e.g., 47 Tuc, ω Cen, M4, and NGC 6397), will be discovered in the future using photometric observations with high performance telescopes. However, due to the extreme crowding in the center of M13 and M22, follow-up spectroscopic observations may need the resolving and light-gathering power of next generation extremely large telescopes, such as the GMT (Giant Magellan Telescope). We plan to further exploit the photometric data used in this paper to discover more hot and luminous WD candidates in nearby Galactic GCs. The results of these studies will be published in future papers.

The search for WDs in Galactic GCs using groundbased telescopes (mainly 4 m class) since 1978 has not been very successful until HST discovered WDs in Galactic GCs in 1995. In the present study, even the discovery of bright WDs in the nearby GC M13 with (m−M)V = 14.33 mag requires the use of photometric observations with a limiting magnitude V ≈ 25 mag. However, photometric observations with deeper exposure times using the 4 m class telescopes (or larger monolithic mirror telescopes for both imaging and spectroscopy) with high performance and larger and more sensitive CCDs (e.g., single 4096 × 4096 CCD and widely used wide-field imagers) will discover moreWDs in the outskirts of nearby GCs with apparent distance modulus (m−M)V ≈ 14.50 mag or less (R⊙≲8 kpc).