Intermediate-Mass Mergers: A New Scenario for Several FS CMa Stars

Abstract

We summarise the properties and nature of a peculiar group of B-type stars called FS CMa stars. These stars show the B[e] phenomenon, i.e., their spectra exhibit both forbidden emission lines and infrared excess. Such properties point to an extended circumstellar gas and dust component. Although the phenomenon has been explained in most B[e] stars, the origin and nature of FS CMa stars is disputed. Here, we focus on the merger hypothesis, for which evidence has recently been discovered.

1. Introduction

B[e] stars are B-type stars with forbidden emission lines and infrared excess. The name was introduced by Conti at IAU Symposium 70 (1975). Lamers et al. [1] found stars with these properties among supergiants (sgB[e] stars), pre-main sequence Herbig Ae/Be stars (HAeB[e] stars), compact planetary nebulae (cPNB[e]), and symbiotic stars (SymB[e] stars). For more details about these interesting objects, we refer to several very good reviews [2,3,4,5,6,7].

Here, we focus on a fifth group of B[e] stars introduced later by Miroshnichenko [8]. He noticed that the B[e] stars that remained unclassified have similar properties. These stars do not fit into any of the four original groups, or may be in more than one of them. However, some of them were not classified because of a lack of good data. This is why the list of FS CMa stars also contains some symbiotics, supergiants of lower luminosity, young planetary nebulae, or Be/X-ray binaries.

2. Observed Properties

Despite the uncertain classification, their spectral properties are unique. Based on new data, we were able to specify the following properties and behaviour:

2.1. Forbidden Emission Lines

The presence of forbidden emission lines of neutral and singly ionized atoms is one of the criteria defining the FS CMa group [8]. The oxygen doublet [O i] $\lambda \lambda$ 6300, 6364 Å is always present. Especially important are lines of [S ii] $\lambda \lambda$ 6716, 6731 Å and [N ii] $\lambda \lambda$ 6548, 6583 Å that allow the usage of nebular diagnostics, e.g., Baldwin, Phillips & Terlevich (BPT) diagrams [9]. Forbidden emission lines are very narrow (FWHM ≈ 50 km s⁻¹), symmetric, and sometimes double-peaked. They are used as tracers of the conditions in the disk. However, this has to be carried out with caution, since stimulated emission plays a key role here.

2.2. Dust Emission

Sheikina at al. [10] found that IR excess of some unclassified B[e] stars steeply decreases longward of 25 μm. Later, Miroshnichenko [8] used this property as one of the key characteristics of FS CMa stars. The episodes of dust creation have been observed in some stars, e.g., in FS CMa [11,12], MWC 349 [13] and MWC 342 [14]. The dust region is very clumpy. Sometimes, huge clouds form and may cover a star, strongly affecting the photometry. Such events have been observed in, e.g., FS CMa [12].

Miroshnichenko [15] detected a weak silicate emission in most FS CMa stars. However, the dusty disk has a more complicated structure. Gauba et al. [16] found that the inner part of the disk around MWC 939 is composed of graphite grains, while the outer part is composed of silicates. Even more complex is the dust disk around HD 50138 [17]. Close to the star, the disk is composed of forsterite. In farther regions, there is forsterite mixed with enstatite and the most distant parts contain amorphous silicates.

2.3. UV Radiation and Resonance Lines

UV radiation plays a key role in the spectrum formation of FS CMa stars and affects their entire spectrum and its variability. The UV radiation passing through the circumstellar matter is strongly absorbed in spectral lines of iron group elements, creating the so-called iron curtain. The absorption may be strong enough to reduce the UV flux by about an order of magnitude compared to a classical Be star of the same temperature [18]. Since the density in the circumstellar envelope is sufficiently low to enable the non-local thermodynamic equilibrium regime, this absorbed energy is partially responsible for the emission of spectral lines of these species in the visible and IR regions. The redistribution of the energy in the circumstellar matter is also detectable in the Johnson U-band photometric filter [19]. This way, it is very easy to study its variability, e.g., in MWC 342 [19].

The circumstellar matter is not static and it may be the case that, due to the Doppler shift, a photon from a spectral line of one element is absorbed in a spectral line of another element. This effect is especially strong in resonance lines because they have a very high transition probability. The temperature and velocity of the matter around FS CMa stars arrange the coincidence of L$\beta$ and resonance line of neutral oxygen 1026 Å. This way, the variability of the Balmer lines and forbidden oxygen lines is coupled, even if their line-formation region is different.

Almost every resonance line is in the UV region. However, Ca ii H and K lines in the near-UV region are observable with ground-based telescopes and Na i D1 and D2 fall into the visible region. This allows the study of their variability. Most FS CMa stars show symmetric emission of these lines; nevertheless, asymmetric emission, P Cygni profiles, inverse P Cygni profiles as well as rarely observed absorption (distinct from interstellar) have also been detected. The absorption, if detected, is very variable. The pure emission lines are stable on time scales of weeks [12,20,21,22]; however, they do show changes over a longer period of time [23]. The resonance lines of Ca ii and Na i occasionally show discrete absorption components. They were reported by Babcock [24] for FS CMa and Pogodin [20] for HD 50138. We also detected these discrete absorption lines in MWC 342, AS 225, AS 174, HD 328990, and HD 50138 [23].

2.4. Hydrogen Lines

FS CMa stars are typified by the strong emission of the Balmer lines. For example, the equivalent width of the H$\alpha$ line may reach even $-400$ Å. The higher the member of the Balmer series, the weaker and narrower the line is. However, this is just the radiative transfer effect that every emission line star shows. More important is the large number of Balmer and Paschen lines. One may usually count around forty members. They set the limits of the density and turbulent velocity using the Inglis–Teller formula [25]. The density and also turbulent velocity is very low in these outer parts, e.g., for FS CMa are the limits of the ion density ≈$1\times {10}^{11}$ cm⁻³ and turbulent velocity $v_{\mathrm{turb}}<40$ km/s [26].

The H$\alpha$ line is usually double-peaked with a weaker blue component. To date, the blue peak has been found to be more intense than the red one only once in HD 50138 [27]. H$\alpha$ lines of lower intensities may show weak absorption wings. The line profile changes on scales longer than a few days. However, very frequently, humps that move across the line are detected.

2.5. Molecules

Although the effective temperatures of the central stars are too high to allow the presence of molecules, the inner parts of the disk serve as an effective shield against the stellar UV radiation and prevent the dissociation of molecules. The most frequently detected molecule is CO. That is natural, because this molecule has high binding energy. Besides CO, SiO, TiO, ${\mathrm{H}}_3^{+}$, and PAHs were also found. For a recent summary of detected molecules, we refer to table A.5 of Korčáková [23].

2.6. Magnetic Field

A magnetic field has been detected only in one FS CMa star, namely IRAS 17449+2320 [28]. This star is also the first magnetic B[e] star. The magnetic field strength is comparable to that in the strongest magnetic Ap stars. The values of the magnetic field modulus (around 5 kG) as well as longitudinal magnetic field strength vary with the rotation period of 36.12 days [29].

2.7. Variability

The spectral lines vary on many time scales. Sometimes it is even better to talk about irregular behaviour. The absorption lines of helium and metal lines show night-to-night variations. Their line profile may show extreme changes from pure absorption to pure emission, through the P Cygni profile as well as inverse P Cygni profile, or absorption with emission wings. The H$\alpha$ line and pure emission lines of metals show multiperiodic behaviour on scales of weeks and months. The behaviour of the H$\alpha$ and He i 5876 Å lines is described very well by Pogodin [20] for HD 50138. Forbidden emission lines are the most stable, but they also change on scales of months and years [27,30,31,32]. To such a complex variability contribute temperature, density, and motion in the atmosphere and the disk, and also non-LTE effects, multiple scattering, and coincidence of wavelengths of resonance lines in the circumstellar gas. In other words, strong absorption in the UV and variable UV continuum affect the spectral lines in the visible and the IR.

Along with the spectral lines, the continuum varies as well [31]. As a consequence, the intensity of all spectral lines decreases or increases. Depending on the continuum source, the red and blue part of the spectra may behave differently [28]. The changes of the continuum are caused mainly by obscuring of the dust with different densities and dust reflection.

2.8. Stellar Masses

The stellar masses for individual stars are determined based on stellar evolutionary tracks in the Hertzsprung–Russell diagram (HRD). The obtained value is affected by the uncertainty of stellar parameters and assumptions used for the calculation of the evolutionary tracks. Both these effects are very large in the case of FS CMa stars. Due to the circumstellar matter, the standard methods of the determination of effective temperature usually fails. Sometimes the guess based on the presence of different ionisation stages of individual elements provides an even more accurate value than the fit by synthetic spectra. Nevertheless, Miroshnichenko et al. [33] obtained the first HRD. The second problem lies in the choice of evolutionary tracks. Among FS CMa stars, some strongly interacting binaries, or even mergers, are hidden. For mergers, only a few evolutionary models exist.

Dynamical masses from binary motion provide a more accurate method of mass determination. Barsukova et al. [34] derived the mass of the primary between 10 and 12 ${\mathrm{M}}_{\odot }$ and a secondary between 0.98 and 1.77 ${\mathrm{M}}_{\odot }$ for CI Cam based on the orbital solution with the inclination angle determined from interferometry [35]. However, the question is whether CI Cam is a typical member of the FS CMa group, since it has been classified as Be/X-ray binary [36,37] or B[e] supergiant [38]. Miroshnichenko et al. [32] determined the mass function for MWC 728 ($2.3\times {10}^{-2}$). A more exotic system is AS 386, where the primary star has mass of around 7 ${\mathrm{M}}_{\odot }$ and the secondary is very likely a black hole [39]. The companions with similar masses between 6 and 8 ${\mathrm{M}}_{\odot }$ has a system HD 327083 [40]. The mass of the primary of MWC 645 was determined to be around 7 ${\mathrm{M}}_{\odot }$ and its secondary around 2.8 ${\mathrm{M}}_{\odot }$ [41]. Recently, the masses of the system were found for the most brightest FS CMa star, 3 Pup. The primary star has a mass around 8 ${\mathrm{M}}_{\odot }$ and the secondary 0.75 ${\mathrm{M}}_{\odot }$ [42]. For more details, we refer to Miroshnichenko et al. [43], who present an HRD that also includes some questionable binary systems.

3. Nature of FS CMa Stars

The precise determination of parameters of the central stars is difficult because of the surrounding gas and dust. We do not have reliable information for some of these stars. Therefore, the usage of standard techniques is limited and many observations may be easily misinterpreted.

3.1. Herbig Ae/Be Stars

Spectra of FS CMa stars are almost identical to Herbig Ae/Be stars (for recent review of Herbig Ae/Be stars see [44]). However, the FS CMa stars are far from star forming regions [8], a key element of HAeBe classification.

3.2. Post-AGB Stars

FS CMa spectra are also similar to post-AGB stars. Several FS CMa stars are in the list of post-AGB stars, e.g., Hen3-938, Hen3-847, or MWC 939. Miroshnichenko [45] summarized the properties of FS CMa stars and found the differences between these two groups. Post-AGB stars with masses larger than 5 ${\mathrm{M}}_{\odot }$ have the peak of the IR excess at wavelengths longer than 30 μm, while FS CMa stars from 10 to 30 μm, i.e., the dust is warmer and hence closer to a star in FS CMa objects. Moreover, massive post-AGB stars evolve so quickly that the temperature changes should be observed on decadal timescales. Post-AGB stars of lower masses, around 1 ${\mathrm{M}}_{\odot }$, show weak emission lines. However, this is the case of a single star, as already mentioned by Miroshnichenko [45]. Post-AGB binaries also have dust close to the star. Its emission peak is around 10 μm [46]. However, almost all post-AGB binaries have effective temperatures that are too low to be B- or early A-type stars [47]. Nevertheless, increasing number of observations and more precise models may lead to re-classification of some objects from these two groups.

3.3. Classical Be Stars

Even if it may seem that FS CMa stars are only an extreme case of classical Be stars, there are significant differences. Classical Be stars (recent review [48]) do not have dust surrounding them, and if they do, it is only a small amount. Moreover, classical Be stars are fast rotators, while FS CMa stars are slow rotators ([23], Table A.7). Therefore, the mechanism of the creation of their gaseous disk must be different.

3.4. Binarity

Binarity is currently an often-discussed scenario [49,50]. It provides the simplest explanation of the large amount of circumstellar gas and dust. Dust is particularly problematic, since it needs special conditions for its creation. A detailed discussion of dust formation is provided by Korčáková [23]. Here, we briefly summarise the important points in favour of and against the current binarity hypothesis.

3.4.1. Mass-Loss Rate ($\dot{M}$)

The large amount of circumstellar matter is the strongest argument for the binarity of all FS CMa stars. $\dot{M}$ has been calculated only for HD 87643 [51], AS 78 [52], and IRAS 00470+6429 [53]. The derived values of $\dot{M}$ from $2.5\times {10}^{-7}$ to $1.5\times {10}^{-6}{\mathrm{M}}_{\odot }\times {\mathrm{yr}}^{-1}$, are not possible to reach by the radiatively driven wind of a main-sequence star of such a mass. All three models assumed a $\beta$ velocity law or a similar one, i.e., monotonously increasing velocity until the limiting value of the terminal velocity. However, an expanding decelerating layer in MWC 342 [31] was later detected. The presence of moving layers was explained by MHD simulations of Moranchel Basurto et al. [54,55]. If the matter is accumulated around the star, the approximation of the freely expanding matter gives a significant overestimate of $\dot{M}$.

3.4.2. Detection of the Secondary

The detection of a companion has been reported in many FS CMa stars. The binary hypothesis is supported by the fact that the fraction of binaries among early B-type stars is about 95% [56]. However, this ratio is only about 30% for late-type main-sequence stars [57] and this is the temperature range of most FS CMa stars. Indeed, evidence of a secondary star has been found only in about one-third of FS CMa stars [49]. The detection of a secondary star may be explained by another phenomenon, low-quality data, or misinterpretation of the observations in many cases. For example, spectroastrometric observations indicate a secondary in HD 50138, FS CMa and HD 85567 [58]. However, the spectroastrometric signal is sensitive to any deviation from spherical symmetry [59,60]. The non-uniform disk discovered interferometrically in HD 50138 [61] mimics a companion in this case. Unambiguous evidence for a secondary would be provided by radial velocity measurements. However, sufficient data are only available for MWC 623 [30], MWC 342 [31], MWC 728 [32], HD 50138 [27], AS 386 [39], IRAS 07080+0605 [62], and IRAS 17449+2320 [28]. Regular periodicity has been found only in AS 386 and MWC 728. Other stars (in addition to those discussed here) also show radial velocity changes. However, those variations are generally quasiperiodic and are not connected with orbital motion, or it is not possible to reject the quasiperiodicity based on several measurements [23].

3.4.3. Central Quasiemission Peak

A small emission peak at the system velocity sometimes appears in the H$\alpha$ line, similar to the classical Be stars. This emission is taken as further support of the binarity, because it is assumed that it originates near the Lagrangian point L1 [62]. However, Hanuschik [63] showed that it is only a radiative transfer effect. The emission is formed through the circumstellar disk that is optically thick in the line and optically thin in the continuum. In the case of FS CMa stars, there is an even simpler explanation: a moving hump revealing inhomogenities in the disk is detected accidentally at zero velocity.

3.4.4. Spectrum of a Hot and Cold Component

The spectrum of MWC 623 is a composition of B4III and K2II (or K2Ib) spectral type [64]. This is a definite proof of it being a binary system, however in ordinary stars. FS CMa stars are surrounded by geometrically and optically thick disks. The cold component spectrum may be formed through the disk. This explanation is supported by a modelling of variability of bisectors of the H$\alpha$ line in MWC 623 [65]. The orbital motion alone is not able to describe the changes. The radial velocities support this idea [30]. The lithium resonance doublet, which is the strongest in MWC 623, is likely formed in a similar manner.

3.4.5. Lithium Lines

About half of FS CMa stars show the lithium resonance doublet at 6708 Å [66]. This line should not be observed in hot B-type stars because lithium is ionised in their hot atmospheres (the ionization potential of Li i is about 5.4 eV). Therefore, the presence of Li i lines has been taken as the evidence of a secondary. However, FS CMa stars have an extended circumstellar region. The lower-temperature circumstellar gas causes the additional absorption. In support of this idea is the appearance of a sharp absorption in He i line 5876 Å in stars with Li i lines [66]. This sharp absorption overlapping the emission is likely created through the circumstellar disk. There are no radiative transfer calculations of these objects that are sufficiently accurate to confirm this guess. However, a similar behaviour is also shown by some Herbig Ae/Be stars. Based on polarimetric measurements, Alecian et al. [67] noticed that lithium lines in these Herbig Ae/Be stars do not originate in the atmosphere of the secondary.

As we can see, many effects and phenomena play here an important role and it is necessary to follow the original papers with the observations.

3.5. Mergers

There is a group of FS CMa objects whose observed properties cannot be explained by known stellar types. Based on the latest observations and numerical simulations, a merger scenario becomes very promising. We summarize the points in favour of a merger idea in the following section.

4. Post-Mergers Among FS CMa Stars

First, it is necessary to note here that the post-merger origin is not an explanation for every FS CMa star. The definition of FS CMa stars (see the introduction) is not a sufficiently sharp criterion and the current list of FS CMa stars contains some interacting binaries, young planetary nebulae, symbiotic stars, Herbig Ae/Be stars, post-AGB stars, as well as supergiants.

The idea of mergers among FS CMa stars is relatively old. However, it has not been discussed in the literature because of the lack of evidence. The first who opened the discussion was de la Fuente et al. [68]. They discovered two FS CMa stars in the central parts of two clusters. They explained this observation as the consequence of the capture of two stars in dense regions. However, as shown by Dvořáková et al. [69], the mergers may be found with almost the same probability in the central as well as outer parts of clusters. Still, there is no strong and straightforward evidence, but rather a combination of many observations and simulations that all come together to create a consistent picture:

4.1. Magnetic Field

The strength of the magnetic field found in IRAS 17449+2320 is comparable with the strongest magnetic Ap stars. Such a strong magnetic field may be generated by the mixing during the merger process [70]. Moreover, the spectropolarimetry of IRAS 17449+2320 shows that the longitudinal magnetic field strength and magnetic modulus changes with a period which corresponds to the rotation period rather than the orbital one [29].

4.2. Slow Rotators

The magnetic field generated during the merger and the later outflow in the form of a wind slows down the newly born star very effectively, e.g., [70]. Indeed, FS CMa stars are rather slow rotators ([23], Table A.7). Some of the published values of the rotation velocity are very likely overestimated. This is the case for stars with a magnetic field that causes the line split or broadening for less sensitive lines or a weak field. To obtain a real value of $vsini$, synthetic spectra with the magnetic field must be used.

4.3. Space Velocities

Most FS CMa stars have a large space velocity [69]. Their peculiar W component is especially important for us. It indicates that the star was very likely kicked from the birth cluster due to an accidental interaction with another star in the cluster. If it is a binary, it usually has a short semi-major axis and a large eccentricity. Consequently, the tidal forces circularise the orbit. As a consequence of the momentum conservation, the semi-major axis shrinks and binary may merge. Let us highlight at this point that binary components may merge not only in triple systems, or during the common envelope phase, but it may also be a natural consequence of the dynamical evolution of the birth cluster. Dynamical processes may also lead to mergers in clusters [69], i.e., the large space velocity is in favour of a merger; however, a small space velocity does not reject the merger origin.

4.4. Li Overabundance

Despite the high temperature of B-type stars, about half of FS CMa stars show lithium lines (see the previous section). Lithium resonance doublet is very strong in some of these stars, which points to a lithium overabundance in these objects. Lithium overabundace has been found in about half of red novae [71], a nova-like event which occurs during a merger.

4.5. Hertzsprung–Russell Diagram (HRD)

Most of FS CMa stars are located around the terminal-age main sequence in the HRD, see, e.g., [72]. According to Schneider et al. calculations [70] of the evolution of a merger, the newly born star becomes a slow rotator around the terminal-age main sequence (TAMS).

4.6. N-Body Simulations

The N-body simulations of the dynamical evolution of 510 clusters analysed by Dvořáková et al. [69] show that about 1.7% of all stars are involved in mergers. About 50% of all mergers are of the spectral type B. This is a natural consequence of the initial mass function, initial distribution of the binary parameters and large range of masses of B-type stars (see Figure 1). We also provide a more detailed

Table 1. The ratio of mergers of individual spectral types from Dvořáková et al. [69].

Spectral Type	Stars at the Beginning [%]	Stars Involved in Mergers [%]	Merger Products [%]	Merger Ratio [%]
O	0.12	3.32	3.46	26.22
B	3.53	48.15	50.48	12.54
A	2.77	11.49	15.18	4.80
F	3.28	8.95	7.11	1.90
G	4.05	6.53	3.88	0.84
K	15.29	10.71	10.29	0.59
M	70.97	10.82	8.81	0.11

Let us follow the dynamical evolution of stars of the spectral type B. According to the initial mass function, about 3.53% of all stars are of the spectral type B at the beginning of the simulation. 48.15% of stars used for the merger creation are of the spectral type B. This causes that 50.48% of all merger products become finally B-type stars (highlighted in red). This also means a high percentage of mergers among B-type stars (12.54%). Note, that the highest fraction of mergers among the given spectral type is in O stars. However, there are not so many O-type stars; therefore, the highest chance to find a post-merger is among B-type stars.

. The table as well as the figure show the results of simulations over Hubble time. To obtain the current observed ratio of mergers in the sky, it is necessary to include the lifetime of individual spectral types and galactic synthesis (Dvořáková et al., in preparation). Note that these simulations very likely underestimated the number of mergers, because the dynamical evolution was calculated only for single stars and binaries. More hierarchical systems may also lead to the merger. The detailed description of the evolution of the triple systems is provided by Toonen et al. [73].

4.7. MHD Simulations

Moranchel Basurto et al. [54,55] performed the first 2.5D MHD simulations of a post-merger based on parameters determined from the observations of FS CMa stars. They explored different scenarios considering the low density of a corona, sub-Keplerian disk rotating around the star, and a strongly magnetized non-rotating and rotating star with a dipolar configuration. They found that due to the magnetic field, low density of the disk and the relative rotation speed between the disk and the star, the disk structure changes extremely (Figure 2, upper panel). The matter concentrations and gaps are created and the disk becomes significantly thicker. The accretion may occur through a thin stream in the equatorial region. It may also be accompanied by a stream following the magnetic field lines, so-called funnel effect. However, this funnel effect becomes unstable under some conditions. The matter may be captured by the magnetic field above the star. After some time, it may either be ejected from the system or fall onto the stellar surface. The magnetic field lines become extremely twisted in a very short time. This configuration is highly unstable and leads to the magnetic reconnection. The released energy heats the matter in the corona which is fed by the wind from the disk, a jet originated at the inner edge of the disk or by a hot plasmoid (Figure 2, bottom panel).

These MHD simulations finally explain most of the behaviour of FS CMa stars, including moving decelerating layers [31] or detection of weak Raman lines in some of these stars.

5. Role of FS CMa Post-Mergers

The simulations of a merger evolution by Schneider et al. [70] show that a star is overluminous before it reaches the main sequence. That points to the possibility that lower massive mergers of B spectral type become magnetic main-sequence A-type stars. This transition phase offers us a unique possibility to test models of the stellar structure and evolution.

As the star evolves on the main sequence, the strength of the magnetic field may decrease due to Ohmic decay and/or flux freezing. As the main-sequence lifetime is shorter for more massive stars, the more massive FS CMa mergers with originally an extremely strong magnetic field may still have a measurable magnetic field at the end of the main-sequence (for the effect of Ohmic decay, see observation surveys of Ap stars, [74]). Hence, more massive magnetic FS CMa mergers may offer the explanation of a recently discovered young strongly magnetic low-mass white dwarf [75].

Before the merger, during it, and also after the merger, a significant amount of the matter is released, allowing the creation of molecules and dust. Part of this material enriches the interstellar matter (Dvořáková et al., in preparation). The intergalactic medium is also slightly affected, because some binaries are kicked out of the birth cluster with a high speed [69].

6. Concluding Remarks

The observed properties of FS CMa stars point to an extended region of circumstellar gas and dust the inner parts of which are very dynamic. These properties may correspond to many different classes of objects. The list of FS CMa stars contains interacting binaries, young planetary nebulae, symbiotic stars, low-luminosity supergiants, or post-mergers. Up to now, only one post-merger has been identified: IRAS 17449+2320 [28]. However, it is reasonable to assume that more post-mergers are hidden in this group. This scenario is supported by MHD simulations conducted by Moranchel Basurto et al. [54,55] that explain the unusual behaviour of several FS CMa stars, e.g., moving decelerating layers, Raman lines in some objects, or discrete absorption components of resonance lines. A strong support to start the systematic search of post-mergers among the FS CMa group is provided by the N-body simulations of Dvořáková et al. [69]. They found that a significant amount of all stars (1.7%) are involved in mergers and about 50% of mergers are of spectral type B.

The determination of the nature of the individual objects is very problematic. The observations as well as the results of the analysis may be easily misinterpreted due to the circumstellar gas and dust. However, FS CMa post-mergers may finally explain long-standing issues connected to the evolution of magnetic stars. The FS CMa stars with lower masses are potentially the progenitors of some magnetic main-sequence A-type stars, since post-mergers are overluminous compared to the stars in the main sequence. Mergers also enrich the interstellar matter. Their contribution will likely be very small compared to those of supernovae or AGB stars; however, a significant percentage of pre-merger binaries have space velocities large enough to leave the galaxy and slightly affect the intergalactic matter. Since several FS CMa stars are strongly interacting binaries that exhibit dust creation, this group offers us a unique possibility to identify the whole evolutionary sequence of mergers.