**5. Binarity**

One of the most heavily debated questions in massive star research is the issue of binarity. Observations have shown that a significant but still contested fraction of massive stars are found in binary systems. Studies of un-evolved massive stars typically find an observed binary fraction of 30–35% for O-type stars in relatively short period (less than ∼100 days) systems [124,125]. When long-period systems are included, this percentage approaches 70% or higher [126,127]. This question of binarity also extends to WRs. Methods range from light curve analysis, searching for spectral signatures (such as radial velocity variations), and the presence of X-ray emission. As discussed earlier, the galaxies of the Local Group provide an excellent test-bed for such studies as we are able to determine a complete sample of WRs with which to study the binary fraction.

Over the decades, many papers have attempted to tackle the issue of binarity head-on. In 1981, Massey and Conti found that the fraction of Galactic WR stars that were close WR+O star systems was ∼25%, and thus the total fraction must be <50% when the issue of compact companions were included [128]. In 2001, van der Hucht compiled an updated list of WRs in the Galaxy bringing the total up to 227 [19]. They found that the binary fraction of observed and probable binaries was around 40%. Foellmi et al. published papers in 2003 looking at the Magellanic Clouds finding close binary fractions of 40% in the SMC and 30% in the LMC [129,130]. More recently, in 2014, Neugent et al. obtained multi-epoch spectra of nearly all of the WRs in M31 and M33 and searched for short period binary systems by observing radial velocity variations within the prominent emission and hydrogen absorption lines. Such hydrogen lines tend to sugges<sup>t</sup> the presence of an O-type star companion (with the notable exceptions being the WN3/O3s, and some hydrogen-rich WRs found in the Galaxy and in the SMC) [131]. This study found that ∼30% of the WRs within M31 and M33 were in short-period binary systems. They additionally found that there was no correlation between binarity and metallicity. Thus, overall, the close binary fraction of WRs appears to be around 30–40% within all metallicity cases, similar to what is observed for O-type stars. (The exact definition of "close" is a debatable one, but we use here a "spectroscopist's definition", corresponding to detection of orbital motions on the order of several 10s of km s<sup>−</sup>1, corresponding to periods of order 100 days or less for massive stars.)

One further way of searching for WR binaries is through the presence of hard X-ray emission. Most single WRs show soft X-ray emission produced by the winds of the single stars. However, in WR binaries, harder, more luminous X-ray emission forms due to the macroscopic shock interactions between the winds in a binary bound system [132,133]. Such X-ray signatures have been found in a few known binary WRs. One of the most extreme such examples is Mk 34 located in the rich OB association of 30 Doradus in the LMC. It has been classified as a WN5ha and is thought to have a (disputably) high mass of 380 *M* as derived through spectroscopic analysis [134], but see also [135]. Garofali et al. additionally found a candidate colliding wind binary (WC + O star) in M31 that is located in the dense HII region NGC 604. It is not nearly as bright as Mk 34, but it still shows X-ray emission as discovered by *Chandra* [136]. While searching for X-ray emission is not the most prominent way of detecting WR binaries, it is more frequently being used as a method of determining binarity.

As one of our good friend and colleague often reminds us, "One can never prove any star is *not* a binary." That said, another colleague has noted that the presence of a companion star often makes itself known in the spectrum, albeit in subtle ways.

In single star evolution, the type of WR is heavily influenced by the metallicity of the gas out of which the star formed. As discussed in the introduction, WN stars that show the hydrogen burning byproducts will appear before WC stars which show the helium burning byproducts. Thus, in a low metallicity environment, one expects to find fewer WCs than in a high metallicity environment. However, once binary evolution is considered, this metallicity dependence decreases because the stripping is being done by Roche-lobe overflow instead of metal-driven stellar winds. Thus, one test of binarity is to look for an excess of WCs in an environment—or, even more compelling, is to identify the even more evolved WOs (oxygen-rich WRs) in low metallicity environments. There are two prime examples of such stars that were most likely created through binary evolution. The first is the WO star in the SMC. As discussed earlier, there are only 12 known WRs in the SMC (a low metallicity environment of 0.25× solar) and 11 of them are WNs, as expected. However, the 12th one is a WO that should only form in a high metallicity environment [4,137]. There is an additional example of a WO forming in the low metallicity environment of IC1613 [138], which has a metallicity of ∼0.15× solar [139]. Although evolution to the WO stage is not expected by even the most massive single stars in low metallicity environments, models that include binary evolution do predict WOs in low metallicity environments [5]. These two stars are thus examples of WRs likely forming through binary evolution; undoubtedly, there are many more.

While many studies have shown the close binary fraction to be around 30–40%, the actual value is still hotly debated. Proponents of binary evolution argue that the currently single WR stars were once multiple, but their companions have merged. There is little evidence, however, to support this conjecture. There is additionally the question of whether the WRs that formed from binary evolution began with initial masses grea<sup>t</sup> enough to sugges<sup>t</sup> that they would have become WRs anyway and the binary mechanism simply sped up the process. Thus, it is possible that the importance of binary evolution may be somewhat overstated, even if the fraction of WRs in binary systems is higher than currently observed.
