Revisiting the Evidence for Double Sequences of Blue Straggler Stars in Globular Clusters

Abstract

Blue straggler stars (BSSs) are believed to form through mass transfer in binary systems or stellar collisions. The reported presence of double BSS sequences in some globular clusters (GCs) has been interpreted as evidence that these two formation channels produce distinct sequences in color-magnitude diagram (CMD). We reassess this claim using HST UV Globular Cluster Survey (HUGS) photometry of 56 Galactic GCs. We used the Hartigan Dip Test to test bimodality, and Akaike model comparison to test whether BSS distance distributions are better described by a mixture of two unskewed Gaussians or a skewed unimodal Gaussian model. We find no strong statistical evidence for bimodality; no cluster yields a dip test p-value below 0.15, and Akaike model comparison favors the skewed unimodal model in 94 out of 112 cases. We re-examine NGC 7099 (M30), the prototypical case of a double BSS sequence, using three reductions of HST data. We find bimodality is detected at p = 4 x 10-3 , vs. the originally reported p ~ 10-5 , in the original photometry. The observed uncertainties derived from the subgiant branch widths are comparable to the suggested separation between the proposed BSS sequences, making the detection of statistically significant bimodality challenging. Our results suggest that the dip between two BSS sequences in M30 photometry is a coincidence, and that later bifurcation claims can be explained as skew in the BSS color distribution, rather than two separate distributions.

Figures (9)

• Figure 1: HUGS UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ CMDs of NGC 7078 before (left) and after (right) correcting for differential-reddening Legnardi2023.
• Figure 2: Normalized CMDs with the three different BSS selection regions used in this study. The selected BSSs are shown in blue, with the selection polygon indicated by solid black lines, and other cluster stars are shown in gray. The selection regions are adapted from Raso2017. In NGC 7099, the red boundary separating fainter BSSs from the supra-MS plume follows the original definition by Raso2017. In NGC 6624 and NGC 6388, we adopted two different red boundaries to avoid contamination from MS stars.
• Figure 3: Dip test analysis of the BSS sequence in the UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ HUGS CMDs of NGC 7099 (M30). The red solid line shows the 1 Gyr BaSTI isochrone. Blue stars denote BSSs selected in the UV$_{275}$--U$_{336}$ that also occupy the BSS region in the optical CMD. Magenta stars show the BSSs (selected in UV$_{275}$--U$_{336}$) that do not occupy the BSS region in the optical CMD. Stars surrounded with squares are W UMa-type stars, while those encircled with circles are SX Phe variables, taken from Clement2001. In the V$_{606}$--I$_{814}$ CMD, there is one less SX Phe variable shown, as it didn't pass the optical photometry checks. The gray lines connecting some stars (at various magnitudes) to the 1 Gyr isochrone illustrate their shortest geometric distances in the CMD. The histograms show the distributions of the shortest geometric distances of the BSSs in the UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ CMDs from the 1 Gyr isochrone. The fits of the skewed unimodal Gaussian model and the mixture of two unskewed Gaussian distributions are shown by the orange and green curves, respectively. The Hartigan dip test $p$-values, $\Delta\mathrm{AIC_c}$ and ER values of these distributions are given in the legends, for both UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ BSS distributions.
• Figure 4: Dip test analysis of the BSS sequences in the UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ CMDs of HUGS clusters in which a double BSS sequence has been previously reported (NGC 362, NGC 1261, NGC 6752, and NGC 7078). The red solid line shows the 1 Gyr BaSTI isochrone. Blue stars denote BSSs selected in the UV$_{275}$--U$_{336}$ that also occupy the BSS region in the optical CMD. Magenta stars show the BSSs (selected in UV$_{275}$--U$_{336}$) that do not occupy the BSS region in the optical CMD. Stars surrounded with squares are W UMa-type stars, while those encircled with circles are SX Phe variables, taken from Clement2001. The gray lines connecting some stars (at various magnitudes) to the 1 Gyr isochrone illustrate their shortest geometric distances in the CMD. The histograms show the distributions of the shortest geometric distances of the BSSs in the UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ CMDs from the 1 Gyr isochrone. The fits of the skewed unimodal Gaussian model and the mixture of two unskewed Gaussian distributions are shown by the orange and green curves, respectively. The Hartigan dip test $p$-values, $\Delta\mathrm{AIC_c}$ and ER values of these distributions are given in the legends, for both UV$_{275}$--U$_{336}$ and V$_{606}$--I$_{814}$ BSS distributions.
• Figure 5: Dip test analysis of the BSS sequence in the v$_{555}$--i$_{814}$ HST/WFPC2 CMD of NGC 7099 (M30) from Ferraro2009 (top) and Lugger2007 (bottom). The red solid line shows the 1 Gyr BaSTI isochrone. The BSSs are color-coded as blue and red based on the classification by Ferraro2009. The Lugger2007 CMD shows the same cross-matched BSSs, except for one BSS not found in the Lugger2007 catalog. Stars surrounded with squares are W UMa-type stars, taken from Clement2001. The gray lines connecting some stars (at various magnitudes) to the 1 Gyr isochrone illustrate their shortest geometric distances in the CMD. The histograms show the distributions of the shortest geometric distances of the BSSs from the 1 Gyr isochrone. The fits of the skewed unimodal Gaussian model and the mixture of two unskewed Gaussian distributions are shown by the orange and green curves, respectively. The Hartigan dip test $p$-values, $\Delta\mathrm{AIC_c}$ and ER values of these distributions are given in the legends.