Could binaries grow the globular cluster family tree?

Title: Stellar collisions in globular clusters: the origin of multiple stellar populations

Authors: Valery Kravtsov, Sami Dib, Francisco A Calderón, and José Antonio Belinchón

First Author’s Institution: Sternberg Astronomical Institute, Lomonosov Moscow State University, Moscow, Russia

Status: Published in Monthly Notices of the Royal Astronomical Society [open access]

Our Milky Way is home to dozens of globular clusters (GCs): massive, tightly-bound groups of thousands (or even millions!) of stars. For centuries, astronomers thought GCs were the poster children for simple stellar populations (SSPs), groups of stars that all formed at the same time from the same molecular cloud. But the real GC family tree has proven to be much more complicated.

Because they were born together, the stars in an SSP share the same age and metallicity. They still span a range of masses, so some stars evolve faster than others. But if you plot an SSP on the Hertzsprung-Russell diagram, the famous plot used to separate stars by evolutionary state, you should see that all the stars fall along the same narrow track (Figure 1).

The earliest H-R diagrams of globular clusters (dating back to the 1910s) looked a lot like SSPs. But more accurate measurements have revealed that GC stars often fall along multiple evolutionary tracks (Figure 1), and follow-up spectroscopy has confirmed that the stars associated with these tracks have different chemical compositions. Somehow, most GCs are home to multiple generations of stars, but the how remains a mystery. 

This “multiple population problem” has received a lot of attention in recent years, with astronomers proposing all sorts of different mechanisms to form additional stellar generations. Stellar rotation, dark matter, globular cluster mergers, and supernova-driven star formation have all been considered, but nothing has been able to fully explain our observations. Hoping for better luck, the authors of today’s paper propose yet another theory: binary stellar mergers, AKA making new stars by smashing together old ones!

Mysterious mergers

We’ve known for a while that binary stars frequently interact with each other. Not only do they influence each other’s motion through good ol’ gravity, but they can also transfer mass back and forth if the conditions are right. In some cases, we think this mass transfer becomes unstable, leading to a mysterious phase of stellar evolution that could result in a stellar merger: two stars combining into one more massive star.

This process is VERY difficult to model, and even harder to observe given that an ongoing merger would probably just look like one really bright star. As a result, our understanding of mass transfer and stellar mergers is pretty incomplete. But that hasn’t stopped astronomers from identifying strange stars that seem to have resulted from the merger process. Blue stragglers, for example, are stars that look like massive (i.e. blue) MS stars on the H-R diagram, but appear in clusters where the massive stars have already gone on to become red giants. At first, astronomers thought that these stars were somehow much younger than the rest of the population. Now, they think that blue stragglers might be the outcome of stellar mergers, with two smaller stars combining to produce the large, unusually blue star we observe today.

As promising as this might sound, we’ve only ever observed a handful of blue stragglers in a given GC. So isn’t it crazy to think that mergers alone could create an entire generation of new stars? Maybe not! Globular clusters are extremely dense stellar environments, which makes it a lot more likely for stars to run into each other. The only question is whether or not these collisions happen often enough to explain the populations that we see. 

To the catalogs!

To test their theory, the authors gathered publicly available data on stellar populations in globular clusters. They started from the catalog by Milone et al. 2017 (M17), which used Hubble Space Telescope data to derive several useful quantities for 54 Milky Way GCs, including N_G1/N_tot. Here, N_G1 is the number of first-generation (G1) red giant branch (RGB) stars in a given cluster, and N_tot is the total number of RGB stars in the same cluster. (The catalog focuses on RGB stars because they’re brighter and easier to detect than smaller stars, but they should be representative of the total stellar population!) This quantity shows at a glance what fraction of a GC’s stars are first-generation versus later-generation, enabling comparisons between clusters.

For 35 out of the 54 GCs from M17, the authors retrieve mass function (MF) slopes from Ebrahimi et al. 2020 and Sollima & Baumgardt 2017. The MF describes the distribution of stellar masses in a given GC. Since this relation is usually a simple power-law, the MF slope is a quick way of quantifying how many stars have low versus high masses. A steep slope would mean that more stars have lower masses and fewer stars have higher masses, while a shallower slope would mean the opposite. If binary star mergers are driving the creation of later-generation stars, the MF slope should get shallower over time as smaller stars are merged to create bigger ones. 

For 51 out of the 54 GCs from M17, the authors also retrieve stellar encounter rates from Bahramian et al. 2013. These rates, denoted with Γ, quantify how common interactions between two stars in a given cluster are. Γ increases with stellar density, since having more stars in the same volume makes encounters more likely. It also decreases with stellar velocity dispersion, since faster-moving stars are less likely to encounter each other. Because binary star mergers are a type of stellar encounter, Γ should be directly related to how many later-generation stars mergers can realistically produce.

Promising patterns 

Armed with all of this data, the authors were finally ready to put their theory to the test. First, they plotted Γ versus N_G1/N_tot (Figure 2, panel A), revealing a strong negative correlation between the two variables. GCs with a higher fraction of G1 stars had lower stellar encounter rates (and vice versa). The authors suggest that this could be due to stellar mergers acting to decrease the number of G1 stars by combining them to produce later-generation stars. Because GCs with higher Γ should also have higher stellar merger rates, the observed anticorrelation aligns with the predictions of the authors’ theory.

Next, the authors plotted MF slope (α) versus N_G1/N_tot (Figure 2, panel B). They found a negative correlation between the two variables, meaning that GCs with a higher fraction of G1 stars have a steeper MF. As the fraction of G1 stars decreases, the MF slope gets shallower, matching the predictions of the authors’ theory. However, the scatter in the relationship is quite high. Dividing their sample up into three groups based on GC mass (Figure 2, panel C), the authors found that the correlation is more significant for the most massive GCs. This suggests that stellar mergers are not the only factor influencing the MF slope; lower-mass clusters might also be losing mass in other ways (which is its own area of study). Luckily, this result doesn’t rule out the stellar merger theory – but it does indicate that more work is needed to understand all the variables at play.

Avoiding common pitfalls

The authors wrap up their paper by discussing how their stellar mergers theory addresses two of the biggest unknowns of the multiple population problem: mass-budget balancing and chemical evolution. Most competing theories are based on the assumption that later-generation stars form from gas ejected by the first-generation stars. This would explain why the different generations have different chemical compositions, as each new generation would be forming from gas that had been enriched in heavy elements by the previous generations. However, these theories often fail to explain how the first-generation stars could produce enough enriched gas to form as many later-generation stars as we observe. This is the so-called mass budget problem, and it’s one of the most pressing issues that new theories must address.

The authors point out that if later-generation stars are formed primarily through stellar mergers, the mass budget problem is no longer a problem, since no gas is needed to form new stars. They construct a toy model to support this conclusion, showing that merging pairs of low-mass main sequence stars can produce N_G1/N_tot and MF slope values consistent with those observed, depending on the choice of a merger rate f_ext (Figure 3).

Since we have no idea how a stellar merger actually proceeds, the question of chemical evolution is a bit trickier. Observations show that later-generation stars often have the same amount of iron, but different amounts of light elements (like C, N, O, Na, Mg, and Al) when compared to first-generation stars. Hypothetically, a merger could boost the central temperature of the new star enough to form additional light elements. And the authors point out that stars produced by mergers are expected to rotate rapidly right after their formation, which could help bring these elements to the stellar surface. But it’s clear that our understanding of the merger process will need to advance significantly before we can confirm or reject this theory.

Despite the uncertainties, the stellar mergers model is a promising addition to the long list of theories competing to explain the multiple population problem. Soon enough, we might find that binary stars are the fertilizer that grows the globular cluster family tree! 

Astrobite edited by Lucas Brown.

Featured image credit: Xavier Dengra (tree); NASA, ESA, and the Hubble Heritage (STScI/AURA)-ESA/Hubble Collaboration (star cluster); NASA/AEI/ZIB/M. Koppitz and L. Rezzolla (binary star)