Observing multiple stellar populations with FORS2@VLT
Aims: We present the photometric analysis of the external regions of three Galactic Globular Clusters: NGC 6121, NGC 6397 and NGC 6752. The main goal is the characterization of the multiple stellar populations along the main sequence (MS) and the study of the radial trend of the different populations hosted by the target clusters.
Methods:The data have been collected using FORS2 mounted at the ESO/VLT@UT1 telescope in filters. From these data sets we extracted high-accuracy photometry and constructed color-magnitude diagrams. We exploit appropriate combination of colors and magnitudes which are powerful tools to identify multiple stellar populations, like versus and versus CMDs.
confirm previous findings of
a split MS in NGC 6752 and NGC 6121. Apart from the extreme case of
Centauri, this is the first detection of multiple MS
from ground-based photometry. For NGC 6752 and NGC 6121 we compare the number
ratio of the blue MS to the red MS in the cluster outskirts with
the fraction of first and second generation stars measured in the
central regions. There is no evidence for significant radial
The MS of NGC 6397 is consistent with a simple stellar population. We propose that the lack of multiple sequences is due both to observational errors and to the limited sensitivity of photometry to multiple stellar populations in metal-poor GCs.
Finally, we compute the helium abundance for the stellar populations hosted by NGC 6121 and NGC 6752, finding a mild () difference between stars in the two sequences.
Over the last years, the discovery that the color-magnitude diagrams (CMDs) of many globular clusters (GCs) are made of multiple sequences has provided overwhelming proof that these stellar systems have experienced a complex star-formation history. The evidence that GCs host multiple stellar populations has reawakened the interest on these objects both from the observational and the theoretical point of view.
Multiple sequences have been observed over all the CMD, from the main sequence (MS, e.g. 2007ApJ...661L..53P) through the sub-giant branch (SGB, e.g. 2012ApJ...760...39P) and from the SGB to the red-giant branch (RGB, e.g. 2008A&A...490..625M) and even in the white-dwarf cooling sequence (2013ApJ...769L..32B).
Multiple populations along the RGB have been widely studied in a large number of GCs (e.g. 2008ApJ...684.1159Y, 2009Natur.462..480L, 2013MNRAS.431.2126M) by using photometry from both ground-based facilities and from the Hubble Space Telescope (HST). In contrast, with the remarkable exception of Centauri (2007ApJ...654..915S, 2009A&A...507.1393B), the investigation of multiple MSs has been carried out with HST only (e.g. 2004ApJ...605L.125B, 2007ApJ...661L..53P, 2012ApJ...745...27M and references therein, 2013ApJ...765...32B).
In this paper we will exploit the FOcal Reducer and low dispersion Spectrograph 2 (FORS2), mounted at the Very Large Telescope (VLT) of the European Southern Observatory (ESO) to obtain accurate U,B,V,I photometry of MS stars in the outskirts of three nearby GCs, namely NGC 6121 (M 4), NGC 6397, and NGC 6752, and study their stellar populations.
The paper is organized as follows: in Sect. 2 we provide an overview on the three GCs studied in this paper. The observations and the data reduction are described in Sect. 3. The CMDs are analyzed in Sect. 4, where we also show evidence of bimodal MSs for stars in the outskirts of NGC 6121 and NGC 6752 and calculate the fraction of stars in each MS. In Sect. 5 we study the radial distribution of stellar populations in NGC 6752 and NGC 6121. In Sect. 6 we estimate the helium difference between the two stellar populations of NGC 6121 and NGC 6752. A summary follows in Sect. 7.
2 Properties of the target GCs
Multiple stellar populations have been widely studied in the three GCs analyzed in this paper. In this section we summarize the observational scenario and provide useful information to interpret our observations.
2.1 Ngc 6121
NGC 6121 is the closest GC ( kpc) and has intermediate metal abundance ([Fe/H]=1.16 1996AJ....112.1487H, 2010 edition).
The RGB stars of this cluster exhibit a large spread in the abundance distribution of some light-elements such as C, N, O, Na and Al (1986A&A...169..208G; 1990AJ....100.1561B; 1992ApJ...395L..95D; 2005PASP..117..895S). There is evidence of a CN bimodality distribution and Na-O anticorrelation (e.g. 1981ApJ...248..177N; 1999AJ....118.1273I).
The distribution of sodium and oxygen is also bimodal. Sodium-rich (oxygen-poor) stars define a red sequence along the RGB in the versus CMD, while Na-poor stars populates a bluer RGB sequence (2008A&A...490..625M). Further evidence of multiple sequences along the RGB of NGC 6121 are provided by 2009Natur.462..480L and 2013MNRAS.431.2126M. NGC 6121 has a bimodal HB, populated both on the blue and red side of the instability strip. The HB morphology of this cluster is closely connected with multiple stellar populations, indeed blue HB stars are all Na-rich and O-poor (hence belong to the second stellar generation) while red-HB stars have the same chemical composition as first-generation ones (2011ApJ...730L..16M).
2.2 Ngc 6397
Located at a distance of kpc, NGC 6397 is a very metal-poor GC (; 1996AJ....112.1487H, 2010 edition).
In late 1970s, 1979ApJ...229..604B have already demonstrated that the RGB stars of this cluster show a spread in light-element abundance. NGC 6397 exhibits modest star-to-star variations of oxygen and sodium and a mild Na-O anti-correlation (e.g. 2002AJ....123.3277R, 2009A&A...505..117C). Similarly to NGC 6121 the distribution of sodium and oxygen is bimodal, and the groups of Na-rich (O-poor) and Na-poor (O-rich) stars populate two distinct RGBs in the Strömgren versus index diagram (2011A&A...527A.148L).
The MS of NGC 6397 is also bimodal, but the small color separation between the two MSs can be detected only when appropriate filters (like the F225W, F336W from HST/WFC3) are used. Observations of the double MSs from multi-wavelength HST photometry have been interpreted with two stellar populations with different light-element abundance and a modest helium variation of 0.01 ( 2010A&A...511A..70D, 2012ApJ...745...27M).
2.3 Ngc 6752
NGC 6752, is a nearby metal-poor GC ( kpc, ; 1996AJ....112.1487H, 2010 edition).
Since the 1980s, spectroscopic data reported ‘anomalies’ in the light-element abundances of the RGB stars of this cluster (1981ApJ...244..205N, 1981ApJ...245L..79C). More recent works confirm star-to-star light-element variations in NGC 6752 (2002A&A...385L..14G; 2003A&A...402..985Y; 2008ApJ...684.1159Y; 2013MNRAS.434.3542Y; 2005A&A...433..597C), O-Na, Mg-Al, C-N (anti-)correlations for both unevolved (2001A&A...369...87G; 2010A&A...524L...2S; 2010A&A...524A..44P) and RGB stars (2005A&A...438..875Y; 2007A&A...464..927C; 2012ApJ...750L..14C) in close analogy with what was observed in most Galactic GCs (see e.g. 2002AJ....123.3277R, 2009A&A...505..117C and references therein). In particular, there are three main groups of stars with different Na, O, N, Al, that populates three different RGBs when appropriate indices are used (like the and Stromgren indices or the visual index, 2008ApJ...684.1159Y; 2003A&A...402..985Y, 2012ApJ...750L..14C, 2013MNRAS.431.2126M).
As shown by 2013ApJ...767..120M (hereafter Mi13), the CMD of NGC 6752 is made of three distinct sequences that can be followed continuously from the MS to the SGB and from the SGB to the RGB. These sequences correspond to three stellar populations with different light-element and helium abundance.
3 Observations and data reduction
For this work we used binned images taken with the ESO/FORS2 ( MIT CCDs) mounted at the VLT, using the standard resolution collimator. With this configuration, the field of view of FORS2 is reduced to by the MOS unit in the focal plane and the pixel scale is . The dithered images of NGC 6121, NGC 6397 and NGC 6752 were acquired using u_HIGH, b_HIGH, v_HIGH and I_Bessel broad band filters between April 14, 2012 and July 23, 2012. A detailed log of observations is reported in Table 1. Figure 1 shows the combined field of view for each cluster.
For the data reduction we used a modified version of the software described in 2006A&A...454.1029A. Briefly, for each image we obtained a grid of 18 spatially varying empirical point spread functions (PSFs, an array of PSFs for each chip of FORS2) using the most isolated, bright and not saturated stars. In this way, to each pixel of the image corresponds a PSF that is a bi-linear interpolation of the closest four PSFs of the grid. This makes it possible to measure star positions and fluxes in each individual exposure using an appropriate PSF and to obtain a catalog of stars for each frame. We registered, for each cluster and for each filter, all star positions and magnitudes of each catalog into a common frame (master-frame) using linear transformations. The final result is a list of stars (master list) for each cluster. For each filter and each star measured in NGC 6121, in Fig. 2 we plot the rms of the photometric residual and of the position as a function of the mean magnitude. In the case of NGC 6752 and NGC 6397 the distributions are similar. As required by the referee, we specify here that we used magnitude to express the luminosities of stars in this paper.
We noted that all the CMDs of the three GCs showed unusually spread out sequences (see panel (a) of Fig. 3 for an example of the versus CMD of NGC 6397). Part of this spread is due to differential reddening, but this is not the only cause. In fact, we found that, selecting stars in different regions of the master list, we obtained shifted MSs. As an example, in panel (d) of Fig. 3 we show the variation of the color for NGC 6397. This plot shows that there is an important gradient of along the x-axis. We selected in this plot two groups of stars: the stars with and that with . We plotted these two sub-samples in the versus CMDs, respectively in black crosses () and red circles (): panel (a) of Fig. 3 shows the result. The two groups form two shifted MSs. This effect is present in all the CMD of the three GCs, even if with different extent levels.
2007ASPC..364..113F showed that there is an illumination gradient in the FORS2 flats produced by the twilight sky. This gradient changes with time and with the position of the Sun relative to the pointing of the telescope and could produce a photometric zero point variation across the FORS2 detectors.
The illumination gradient in the flat fields could contribute to the observed enlargement of the CMDs.
We obtained star fluxes using local sky values, and therefore it is expected that these systematic effects are negligible. If the gradient in the flat-field images is not properly removed during the pre-reduction procedure, the pixel quantum efficiency correction will be wrong. The consequence is that the luminosity of a star measured in a given location of the CCD will be underestimated (or overestimated) with respect to the luminosity of the same star measured in another location of the CCD.
Using the measured star positions and fluxes, we performed a correction in a similar way to what described by 2009A&A...493..959B. It is a self-consistent auto-calibration of the illumination map and takes advantage of the fact that the images are well dithered.
For each cluster and for each filter, the best image (characterized by lower airmass and best seeing) is defined as reference frame. We considered the measured raw magnitude of each star in each image . Using common stars between the image and the reference frame we computed the average magnitude shift:
where M is the number of stars in common between the reference frame and the single image . For each star, that is centered in a different pixel in each dithered frame, we computed its average magnitude in the reference system:
where is the number of images in which the star appears. Then we computed the residual:
We divided each FORS2 chip in a spatial grid of boxes, and, for each box, we computed the average of the residuals from the stars located in that region in each single image. This provides a first spatial correction to our photometry. To obtain the best correction, we iterate until the residual average became smaller than 1 mmag. To guarantee convergence we applied, for each star, half of the correction calculated in each box. Moreover, to obtain the best correction at any location of the camera, we computed a bi-linear interpolation of the closest 4 grid points. At the edges of the detectors the correction is less efficient, because the corresponding grid-points have been moved toward the external borders of the grid to allow the bi-linear interpolation to be computed all across the CCDs. In panel (e) of Fig. 3 we show our final correction grid for the v_HIGH filter.
The final correction grids are different for each filter and for each set of images. In particular, the patterns are different from filter to filter, as well as the size of the zero point variations. This quantity also changes using different data-sets. The maximum amplitudes of our corrections are tabulated in Table 2.
|NGC 6121||NGC 6397||NGC 6752|
We corrected the spread of the CMD due to differential reddening using the procedure as described by 2012A&A...540A..16M. Briefly, we defined a fiducial line for the MS of the cluster. Then, for each star, we considered a set of neighbors (usually 30, selected anew for each filter combination) and estimated the median offset relative to the fiducial sequence. These systematic color and magnitude offsets, measured along the reddening line, represent an estimate of the local differential reddening. With this procedure we also mitigated the photometric zero-point residuals left by the illumination correction (especially close to the corners of the field of view). Panel (c) of Fig. 3 shows the CMD after all the corrections are applied.
The photometric calibration of FORS2 data for Johnson and Cousins bands was obtained using the photometric Secondary Standards star catalog by 2000PASP..112..925S. We matched our final catalogs to the Stetson standard ones, and derived calibration equations by means of least squares fitting of straight lines using magnitudes and colors.
3.1 Proper motions
Since NGC 6121 (l,b=35097,1597) and NGC 6397 (l,b=33817;1196) are projected at low Galactic latitude, their CMDs are both dramatically contaminated by Disk and Bulge stars, contrary to NGC 6752 (l,b=33649;2563) that presents low field contamination. The average proper motions of NGC 6121, NGC 6397 and NGC 6752 strongly differ from that of these field stars (e.g. 2003AJ....126..247B, 2006A&A...456..517M). Therefore, to minimize the contamination from field stars we identified cluster members on the basis of stellar proper motions.
In order to get information on the cluster membership, we estimated the displacement between the stellar positions obtained from our FORS2 data-set and those in the ground-based data taken from the image archive maintained by 2000PASP..112..925S and also used in 2013MNRAS.431.2126M. These observations include images from different observing runs with the Max Planck 2.2m telescope, the CTIO 4m, 1.5m and 0.9m telescopes and the Dutch 0.9m telescope on La Silla. To obtain the displacement, we used six-parameter local transformations based on a sample of likely cluster members, in close analogy to what was done by 2003AJ....126..247B and 2006A&A...454.1029A to calculate stellar proper motions.
Results are shown in Fig. 4. The figure shows the versus CMDs for NGC 6121, NGC 6397 and NGC 6752. The insets show the vector-point diagrams of the stellar displacements for the same stars shown in the CMDs: the cluster-field separation is evident. Likely cluster members are plotted in black both in the CMDs and in the vector-point diagrams, in gray the rejected stars.
4 The CMDs of the three GCs
Previous studies on multiple stellar populations have demonstrated that the color is very efficient in detecting multiple RGBs (see 2008A&A...490..625M and 2010ApJ...709.1183M for the cases of NGC 6121 and NGC 6752), and multiple MSs (see 2012A&A...540A..16M, Mi13 for the cases of NGC 6397 and NGC 6752). As discussed by 2011A&A...534A...9S, CNO abundance variations affect wavelengths shorter than nm owing to the rise of molecular absorption bands in cooler atmospheres. The consequences are that the CMDs in filters show enlarged sequences, mainly due to variations in the N abundance, with the largest variations affecting the RGB and the lower MS.
Motivated by these results, we started our analysis from the versus CMDs shown in Fig. 5; the inset of each panel is a zoom in of the upper MS, between and magnitudes below the turn off.
A visual inspection at these CMDs reveals that the color broadening of MS stars in both NGC 6121 and NGC 6752 is larger than that of NGC 6397. A small fraction of MS stars in NGC 6121 and NGC 6752 defines an additional sequence on the blue side of the most-populous MS.
To investigate whether the widening of the MSs is due to the presence of multiple populations we identified in the versus CMD of each cluster two groups of red-MS (rMS) and blue-MS (bMS) stars, as shown in panels (a) of Fig. 6. We colored the two MSs in red and blue respectively, and these colors are used consistently hereafter. In the case of NGC 6121 and NGC 6752, where there is some hint of a split MS, we identified by eye the fiducial that divide the rMS and bMS. In the case of NGC 6397 we have considered as rMS (or bMS), the stars that are redder (or bluer) than the MS fiducial line. Each inset shows the color distribution of for the two MSs, where is obtained by subtracting the color of the fiducial that divide the two MSs to the color of the rMS and bMS stars. In the cases of NGC 6121 and NGC 6752 the color distributions show a double peak that could be due to the presence of two populations; we fitted them with a sum of Gaussians (in red and blue respectively for the rMS and bMS). We applied a moving box procedure to further verify that the distribution of in the case of NGC6121 is bimodal. We changed the binsize, ranging from 0.005 to 0.02 (approximately the error in color) with steps of 0.001. Furthermore, from our dataset we determined a kernel-density distribution by assuming a Gaussian kernel with 0.02 mag. I all cases, we consistently found that the distribution can only be reproduced by two Gaussians.
A multiple sequence in NGC 6752 had already been identified by Mi13 using HST data. Very recently, we had the first F275W, F336W, F438W WFC3 images of NGC 6121 from the HST GO-13297 UV Large Legacy Program (P.I. Piotto). Even a preliminary reduction of the data shows a clear separation of the MS into two branches in the F438W vs F336W-F438W CMD, fully confirming what we anticipate here in the equivalent, groundbased vs diagram of Fig. 6.
In the case of NGC 6397, it is possible to fit the distribution with a single Gaussian.
As an additional check for the presence of multiple populations, we investigated whether the widening of the MSs of all the GCs is intrinsic or if it is entirely due to photometric errors. We have compared two CMDs, versus with versus , obtained using independent datasets. We considered the rMS and bMS defined previously: if the color spread is entirely due to photometric errors, a star which is red (or blue) in the versus CMD will have the same chance of being either red or blue in the vs . By contrast, the fact that the two sequences identified in the first CMD have systematically different colors in the second one, would be a proof that the color broadening of the MS is intrinsic. In panels (b) of Fig. 6 we plotted the rMS and bMS in versus CMDs. The fact that the rMS stars of both NGC 6121 and NGC 6752 have, on average, different than bMS stars demonstrates that the color broadening of their MS in the versus CMDs is intrinsic. This is the first evidence that the MS of NGC 6121 is not consistent with a simple stellar population. In the case of NGC 6397 rMS and bMS stars share almost the same thus suggesting that most of the colors broadening is due to photometric errors.
As last test, we plotted the two MSs in the versus CMD. The index, which is defined as the color difference , is a very efficient tool to identify multiple sequences in GCs. Indeed it maximizes the color separation between the stellar populations that is due to both helium and light-element variations (2013MNRAS.431.2126M). Panels (c) of Fig. 6 confirm the previous results: rMS and bMS of NGC 6121 and NGC 6752 are well defined in the versus CMDs, but this is less evident for NGC 6397.
As a last prof, we plotted in Fig. 7 the versus diagrams for each cluster: in red and in blue are plotted the rMS and bMS defined previously. The figure shows that for both NGC 6121 and NGC 6752 the two MSs are well defined, while for NGC 6397 the rMS and bMS stars are mixed
Figure 8 shows the versus diagram for the three GCs studied in this paper. In their analysis of multiple stellar populations in 22 GCs, 2013MNRAS.431.2126M found that all the analyzed clusters show a multimodal or spread RGB in the versus diagram, and the value of each star depends on its light-element abundance. The -index width of the RGB () correlates with the cluster metallicity, with the more metal rich GCs having also the largest values of . In order to compare the MSs of the three GCs studied in this paper, we introduce a quantity, , which is akin of , but is indicative of the -index broadening of the MS. The procedure to determine is illustrated in Fig. 9 for NGC 6121 and is the same for the other clusters. We have considered the magnitude of the TOs as magnitude of reference: in the case of NGC 6397, for NGC 6752, and for NGC 6121. Panel (a) of Fig. 9 shows the versus diagram for NGC 6121 in a range of magnitudes from to . In this range of magnitudes, we obtained the fiducial line for the MSs computing the -clipped median of the color in interval of 0.35 mag and interpolated these points with a spline. In our analysis we used only MS stars with where the MS split is visible for both NGC 6121 and NGC 6752. This magnitude interval is delimited by the two dashed lines of Fig. 9a. The thick line is the fiducial in the considered magnitude interval. The verticalized versus diagrams is plotted in panel (b) of Fig. 9, while panel (c) shows the histogram distribution of . The MS width, , is defined as the extension of the histogram and is obtained by rejecting the 5% of the reddest and the bluest stars on the extreme sides. To account for photometric error, we have subtracted from the observed the average error in in the same magnitude interval, i.e. .
We found that the most metal-rich GC, NGC 6121, exhibits the largest
index width for MS stars ().
The spread in is smaller in the case of NGC 6752 () and drops down to in the most metal-poor GC NGC 6397. To estimate the statistical uncertainty in measuring , we used the bootstrap resampling of the data to generate 10000 samples drawn from the original data sets. We computed the standard deviation from the mean of the simulated and adopted this as uncertainty of the observed .
These findings make it tempting to speculate that the index width of the MS could be correlated with the cluster metallicity, in close analogy with what observed for RGB stars. An analysis of a large sample of GCs is mandatory to infer any conclusion on the relation between and [Fe/H].
Mi13 have identified three stellar populations in NGC 6752 that they have named as ‘a’, ‘b’, and ‘c’. Population ‘a’ has a chemical composition similar to field halo stars of the same metallicity, population ‘c’ is enhanced in sodium and nitrogen, depleted in carbon and oxygen and enhanced in helium (0.03), while population ‘b’ has an intermediate chemical composition between ‘a’ and ‘c’ and is slightly helium enhanced (0.01). However, the MSs of populations ‘b’ and ‘c’ are nearly coincident in the versus CMD, while population ‘a’ stars have bluer colors (Mi13, see their Fig. 8). The three MSs exhibit a similar behavior also in the and colors. Since these colors are similar to , the less populous MS identified in this paper should correspond to the population ‘a’ identified by Mi13, while the rMS hosts both population ‘b’ and population ‘c’ stars.
4.1 The fraction of rMS and bMS in NGC 6121 and NGC 6752
In order to measure the fraction of stars in each MS we followed the procedure illustrated in Fig. 10 for NGC 6752, which we already used in several previous papers (e.g. 2007ApJ...661L..53P, Mi13).
Panel (a) shows the versus CMD of the MS stars in the magnitude interval , where the MS split is most evident. We verticalized the selected MS by subtracting the color of the stars to the color of the fiducial line of the rMS, obtaining . The fiducial line is obtained by hand selecting the stars of the rMS, dividing them in bins of magnitude, computing the median colors of the stars within each bin and interpolating these median points with a spline. The verticalized versus diagram is plotted in panel (b).
The color distribution of the stars for three magnitude intervals is shown in panels (c). Each histogram clearly shows two peaks and has been simultaneously fitted with a double Gaussian, whose single components are shown in blue and in red for the bMS and the rMS, respectively.
For each magnitude interval, from the area under the Gaussians we infer the fraction of bMS and rMS stars. The errors () associated to the fraction of stars are estimated as , where is derived from binomial statistics and is the uncertainty introduced by the histogram binning and is derived as in 2014A&A...563A..80L. Briefly, we have derived times the population ratio as described above, but by varying the binning and starting/ending point in the histogram. We assumed as the rms scatter of these determinations.
We computed the weighted mean of the bMS and rMS fractions of the three magnitude intervals, using as weight . In the case of versus , we obtained that the rMS and bMS contain respectively and of MS stars. In panels (d), (e) and (f) of Fig. 10 we applied the same procedure in the versus using the stars with . We obtained that the blue MS contains of the total number of MS stars, and the red MS is made of the remaining stars. We also calculated the weighted mean of the results of the two CMDs of NGC 6752, obtaining that in the rMS there are the of stars, and in the bMS the remaining .
We have already demonstrated that the distribution, in the versus CMD of NGC 6121, shows a double peak, proving the presence of multiple populations (see panel a of Fig. 6). We performed a detailed analysis of the MS of NGC 6121, applying the same procedure described for NGC 6752. The procedure and the results are shown in Fig. 11. We find from the analysis of the versus CMD that the bMS contains of MS stars, and the rMS includes the remaining . In the case of the versus diagram we infer that rMS and bMS contain and of the total number of MS stars, respectively. We computed the fraction of rMS and bMS for NGC6121 using different histogram binsizes and changing starting/ending points. We used binsizes with values between 0.005 and 0.025 (larger than color error), starting points between -0.5 and -0.15 and ending points between 0.15 and 0.5. I all cases, the resulting fraction of stars are in agreement with the values we quote above within the errors The results from the two CMDs imply that in the rMS and bMS there are, respectively, and of MS stars.
5 The radial distribution of stellar populations in NGC 6752 and NGC 6121
The analysis of the radial distribution of rMS and bMS stars in NGC 6121 and NGC 6752 is an important ingredient to shed light on the formation and the evolution of multiple stellar populations in these GCs. Indeed, theoretical models predict that, when the GC forms, second-generation stars should be more centrally concentrated than first-generation ones, and many GCs could still keep memory of the primordial radial distribution of their stellar populations (e.g. 2008MNRAS.391..825D, 2011MNRAS.412.2241B, 2013MNRAS.429.1913V).
The radial distribution of stellar populations in NGC 6752 is still controversial. 2011A&A...527L...9K determined wide-field multi-band photometry of NGC 6752 and studied the distribution of its stellar populations across the field of view. They have concluded that there is a strong difference in the radial distribution between the populations of RGB stars that are bluer (bRGB) and redder (rRGB) in color, and obtained similar findings from the study of the SGB. Specifically, at a radial distance close to the half-mass radius (; 1996AJ....112.1487H, 2010 edition) the fraction of rRGB stars abruptly decreases. These results are in disagreement with the conclusions by Mi13 who showed that the three stellar populations identified in their paper share almost the same radial distribution. Kravtsov and collaborators analyzed stars with a radial distance from the center of NGC 6752 out to , while the study by Mi13 is limited to the innermost arcmin. In this paper we extend the analysis to larger radii111We assume that stars in the fields of NGC6752 and NGC6121 are representative of stellar populations at the studied radial distance; we are not able to investigate any dependency on the angular position using the dataset presented in this work..
As already mentioned in Sect. 4, we suggest that the bMS of NGC 6752 corresponds to the population ‘a’ identified by Mi13, while the most populous rMS hosts both population ‘b’ and population ‘c’ stars of Mi13. For this reason, in this section, we rename the bMS in MSa and the rMS in MSbc.
In order to investigate the radial distribution of stellar populations within the field of view analyzed in this paper, we divided the catalog of NGC 6752 stars into two groups at different radial distance from the cluster center, each containing almost the same total number of stars.
The inner sample of stars (inner field) lies between and from the cluster center. The outer group of stars (outer field) is between and from the center. We estimated the fraction of stars in each group by following the same procedure described in Sect. 4.1.
The results are illustrated in Fig. 12. In the left panels we show the verticalized versus and the versus diagrams for stars in the inner field. In this region the MSa contains and the MSbc hosts the remaining of the total number of MS stars. In the outer field (right panels of Fig. 12) the MSa and the MSbc are made of the and of MS stars, respectively. We conclude that there is no evidence for a gradient within the field of view studied in this paper.
To further investigate the radial distribution of stellar populations in NGC 6752 we compare the results obtained in this paper for stars with distance from the cluster center larger than arcmin and the fraction of stars that have been estimated by Mi13 in the internal regions by using the same method.
Since the MSbc contains both populations ‘b’ and population ‘c’ stars, we have added together the fractions of population ‘b’ () and population ‘c’ stars () listed by 2013ApJ...767..120M and calculated the fraction of stars in these two populations: . As aforementioned in Sect. 4, we further compare the fractions of population ‘a’ stars by Mi13, with the fractions of MSa stars derived in this paper. The values of and are listed in Tab. 3.
Results are shown in Fig 13, where the top panels show the distribution of the fraction of population ‘a’ (in blue) and the fraction of population ‘b’+‘c’ (in red) as a function of the radial distance from the cluster center, while the bottom panels show the radial trend of the ratio between the fraction of population ‘a’ and the fraction of population ‘b’+‘c’. In the left panels we show both the above described distributions considering single radial intervals for each set of data, while in the right panels we divided each radial interval in different bins. Our findings suggest that there is no evidence for a radial gradient among population ‘a’ and population ‘b’+‘c’ of NGC 6752.
In order to investigate the radial distribution of stellar populations in NGC 6121, we divided the field of view analyzed in this paper into two regions, with radial distance from the cluster center (inner field) and (outer field). Each region contains almost the same number of stars. We determined the fraction of rMS and bMS stars by following the same recipe described in detail for NGC 6752. The results are shown in Fig. 14. We found that in the inner field the fraction of bMS is and the fraction of rMS is . For the outer field we obtain that the bMS and the rMS contains respectively the and the of the total number of the considered MS. Also for NGC 6121 we found no evidence of population gradients.
6 The helium content of stellar populations in NGC 6121 and NGC 6752.
The ultraviolet pass-band is very efficient to separate multiple sequences due to its sensitivity to difference in C, N, O abundance (2008A&A...490..625M, 2011A&A...534A...9S). In contrast, and colors are marginally affected by light-element variations, but are very sensitive to the helium abundance of the stellar populations (e.g. 2002A&A...395...69D, 2007ApJ...661L..53P, 2011A&A...534A...9S, 2012AJ....144....5K, 2013MmSAI..84...91C), thus providing us with an efficient tool to infer the helium content.
6.1 Ngc 6121
The procedure to estimate the average helium difference between bMS and rMS stars for NGC 6121 is illustrated in Fig. 15 and is already used in several papers by our group (Mi13, 2012ApJ...745...27M; 2012ApJ...744...58M). Since we have already extracted the stellar populations in NGC 6121 by using the versus CMD of Fig. 11, we can now follow them in any other CMD. By combining photometry in four filters, we can construct three CMDs with versus (), where . The fiducial lines of bMS and rMS in these CMDs are plotted in the upper panels of Fig. 15. rMS is redder than bMS in color, whereas it is bluer in and colors.
We measured the color distance between the two MSs at a reference magnitude (), and repeated this procedure for =17.3, 17.5, 17.7, 17.9, and 18.1 (corresponding to the magnitude interval where the two MS separations is maximal, cf Fig. 6). The color difference () is plotted in the lower panel of Fig. 15 as a function of the central wavelength of the filter (gray dots), for the case of =17.7.
We estimated effective temperatures () and gravities () at different = for the two MS stars and for the different helium contents by using BaSTI isochrones (2004ApJ...612..168P; 2009ApJ...697..275P).
We assumed a primordial helium abundance for the bMS, , and used for the rMS different helium content, with varying from 0.248 to 0.400 in steps of =0.001. To account for the appropriate chemical composition of the two stellar populations of NGC 6121 we assumed for the bMS and the rMS the abundances of C, N, O, Mg, Al, and Na as measured for first and second-generation RGB stars listed by 2008A&A...490..625M.
We used the ATLAS12 program and the SYNTHE code (2005MSAIS...8...14K,2005MSAIS...8...25C, 2007IAUS..239...71S) to generate synthetic spectra for the adopted chemical compositions, from 2,500 Å to 10,000 Å. Synthetic spectra have been integrated over the transmission curves of the filters, and, we calculated the color difference for each value of helium of our grid.
The best-fitting model is determined by means of chi-square minimization. Since the magnitude is strongly affected by the abundance of light elements we used and colors only to estimate . The helium difference corresponding to the best-fit models are listed in Table 4 for each value of .
We derived that the rMS is slightly helium enhanced with respect to the bMS (which has ), with an average helium abundance of This is the internal error estimated as the rms scatter of the independent measurements divided by the square root of . Results are shown in Fig. 15 for the case of =17.7, where we represented the synthetic colors corresponding to the best-fitting model as red asterisks.
For completeness we also calculated synthetic colors of two MS stars with the and the same chemical composition (same abundance of light elements). We assumed for bMS primordial helium and for rMS the helium abundance of the best-fitting model. Results are represented as blue squares in Fig. 15 and confirm that the abundance of light elements assumed in the model does not affect our conclusion on the helium abundance of the two MSs, which are based on the optical colors. Instead the different CNO content strongly affect the band.
In principle, the He content of stellar populations in GCs can also be estimated using He lines in HB star spectra (e.g. the HeI line at line, 2009A&A...499..755V; 2012ApJ...748...62V; 2014MNRAS.437.1609M ). However, spectroscopic measurement of He in GC stars has many limitations. First of all, He can only be measured for stars in a very limited temperature interval ( K). In fact, stars with K are not sufficiently hot to form He lines, while stars bluer than the Grundahl jump (1999ApJ...524..242G, K) are affected by He sedimentation and metal levitation which alter the original surface abundance. The HB of NGC 6121 is populated both on the red and the blue side of the RR Lyrae instability strip. Spectroscopic investigation by 2011ApJ...730L..16M reveals that the blue HB is made of second population Na-rich and O-poor stars, while red HB stars belong to the first population. In NGC 6121, the HB segment with K corresponds to the blue HB, and therefore it only provides partial information only. In this cluster (as in many others) it is not possible to spectroscopically measure the He content of the first population.
2012ApJ...748...62V have used the HeI line at to estimate the helium content of six blue-HB stars in the blue HB of NGC 6121. All of them are second-population stars. They derived a mean value of =0.290.01 (random) 0.01 (systematic) and conclude that second-population stars would be enhanced in helium by 0.04-0.05 dex. This estimate of the He content has been made by assuming LTE approximation. However, the HeI line at is affected by NLTE effect, which can cause an error in the estimate as large as =0.10 (see 2014MNRAS.437.1609M for the case of NGC 2808). Appropriate NLTE analysis is required to infer reliable He abundances from spectroscopy of HB stars in NGC 6121. In contrast, the He difference between red- and blue-MS stars in NGC 6121 comes from the colors of the fiducial lines, which have small color uncertainties.
6.2 Ngc 6752
We followed the same procedure to estimate the average helium difference between MSa and MSbc stars. We measured the color distance between the two fiducial lines of MSa and MSbc in the versus CMDs (Fig. 16), where , at reference magnitudes and . The color difference at is plotted in the bottom panel of Fig. 16 as a function of the central wavelength of the filter.
We used BaSTI isochrones (2004ApJ...612..168P; 2009ApJ...697..275P) to estimate and at different .
We assumed that MSa has primordial helium abundance, , and varied the helium content of the MSbc between 0.248 and 0.400 in steps of . We assumed for the MSa the same C, N, O, Mg, Al and Na abundances of the population ‘a’ of Mi13; for the chemical composition of the MSbc we considered the average of the abundances of the population ‘b’ and ‘c’ listed by Mi13.
As mentioned above, we obtained synthetic spectra for the adopted chemical compositions, integrated them over the transmission curves of the , , , filters, and computed the helium difference using the best-fitting model.
We obtained that the MSbc is helium enhanced with respect to the MSa of . As for NGC 6121, since the magnitude is affected by the abundance of light elements, we used and colors only to estimate . Note that the abundance of light elements assumed in the model does not affect our conclusion on the helium abundance of the two MSs, as already proved in the case of NGC 6121.
6.3 Relation between HB morphology and Helium abundance
In their work, 2014ApJ...785...21M have sought correlations between HB morphology indicators and physical and morphological GC parameters. Among these parameters there is also the maximum helium difference between stellar populations hosted by GCs.
They introduced two different parameters to describe the HB morphology: , that is the color difference between the RGB and the coolest border the HB, and , that is the color extension of the HB (for more details, see Fig. 1 of 2014ApJ...785...21M).
They divided the sample of 74 GCs in three groups: in the first group, G1, there are GCs with [Fe/H]; the second group, G2, includes GCs with [Fe/H] and ; the third group, G3 contains GCs with .
They found a tight correlation between and the maximum internal helium difference (, measured on the MS) for the group G2+G3 (see Fig. 8 of their paper).
In our work we add two more points to their data-set, the Helium difference between the two populations of NGC 6752 and NGC 6121, as computed in this work. In the case of NGC 6752, the added point constitutes a lower limit because the Helium difference between Pop and Pop, (Pop-Pop), is the average value between (Pop-Pop) and (Pop-Pop).
The result is in Fig. 17: in black there are the points of 2014ApJ...785...21M and in grey the points added in this work. In analogy to the work of 2014ApJ...785...21M, the crosses refer to the G1 GCs, triangles to G2 group and dots to G3 clusters. Our data points confirm the tight correlation between L2 and . We found a Spearman’s rank correlation coefficient (to be compared to found by 2014ApJ...785...21M), with (the uncertainty in is estimated by means of bootstrapping statistic, as in 2014ApJ...785...21M).
This result is a further proof that the helium-enhanced stellar populations are likely related to the HB extension, as predicted by theory.
The photometric analysis of ESO/FORS2 data of the external regions of the three nearby Galactic GCs NGC 6121 (M 4), NGC 6752 and NGC 6397 has confirmed that the first two GCs host multiple stellar populations. Indeed, the versus and versus CMDs of NGC 6752 and NGC 6121 show a split of the MS in two components. Excluding the unique case of Cen, this is the first time that a split of the MS is observed using ground-based facilities.
The multiple stellar populations of NGC 6397 was investigated by 2012ApJ...745...27M using HST data. They found two stellar populations characterized by a modest helium variation . Unfortunately, in this work, it was not possible to analyze these populations, because of the size of our photometric errors is comparable to the small color separation between the MSs.
Using HST data, Mi13 have already demonstrated that NGC 6752 host three stellar populations. They computed the radial trend of the ratio between the number of stars of different populations out a radial distance from the center of . Because of larger photometric errors, we have resolved only two MSs. Comparing them with the work of Mi13, we found that the less populous MS corresponds to their population ‘a’, while the most populous MS hosts both their populations ‘b’ and ‘c’. In average we found that the MSa contains about 26% of the total number of stars and the MSbc host about 74% of the MS stars. The most straightforward interpretation is that the MSa is formed by stars of the first generation with chemical abundances similar to that of the Galactic halo field stars with the same metallicity; the MSbc hosts stars of second generation, formed out of material processed through first-generations stars. This population is characterized by stars enhanced in helium, with . Our measurement of the helium enhancement is in agreement with the average of the populations ‘b’ () and ‘c’ () obtained by Mi13. We extended the study of the radial trend of the populations of NGC 6752 to more external regions, confirming the results of Mi13, of a flat distribution. Therefore we cannot confirm the results by 2011A&A...527L...9K and 2014ApJ...783...56K; they found that the two populations show a strong gradient at a radial distance close to the half-mass radius.
In a recent work on NGC 6121, 2014MNRAS.439.1588M investigate the bottom of the MS of this cluster using HST near-infrared photometry. They found that the MS splits into two sequences below the MS knee. In particular they identified two MSs: a MS that contains of stars and MS formed by the remaining . They show that the split of the MS is mainly due to the effect of molecules, present in the atmospheres of M-dwarfs, on their near-infrared color, and that it is possible to associate the MS to a first generation of stars and the MS to a second one. 2008A&A...490..625M, analyzing spectra of RGB stars, found that of stars are Na-rich and O-poor and the remaining have chemical abundances similar to those of Halo-field stars with the same metallicity. All these results are in agreement with what we have obtained in this work: the MS of NGC 6121 splits in the versus and versus CMDs. We found two MSs: a less populous MS that contains of MS stars and which constitutes the first generation of stars and a more populous second generation MS that contains of stars. 2012ApJ...748...62V, using spectroscopic measurements of blue HB (bHB) stars, obtained that the difference in helium abundance between these stars and the red HB (rHB) stars is . A spectroscopic analysis of 2011ApJ...730L..16M revealed that the rHB stars have solar-scaled [Na/Fe], while bHB stars are Na enhanced. In contrast to the results of 2012ApJ...748...62V, a lower constraint to the level of He enhancement is set by 2014ApJ...782...85V, founding a maximum between bHB and rHB stars. Analyzing how the two MS of NGC 6121 behave in different CMDs, we computed the helium abundance difference between them. Our result is , in agreement with that obtained by 2012ApJ...748...62V. Also in the case of NGC 6121, we did not find evidence of changes in the fraction of bMS/rMS stars in the radial range between .
2014ApJ...785...21M found a correlation between the HB morphological parameter and the maximum helium difference among stellar populations in GCs. Using the helium abundances computed in this work for NGC 6121 and NGC 6752, we confirm this correlation and the theoretical indications that helium enhanced stellar populations are responsible of the HB extension.