Could the unseen companion of HR 6819 be a close binary system?

Question

Recently there was an announcement of a black hole candidate in the HR 6819 system (ESO Instrument Finds Closest Black Hole to Earth). In the proposed system, the black hole forms a close binary with the B3III primary of the system, and a Be star is located on a wide outer orbit forming a hierarchical triple:

 ┌ B3III star
┌┤
│└ black hole
┤
│
└ Be star

Reading through the research paper Rivinius et al., 2020, it is mentioned that there is no X-ray emission detected from the system indicating that it is either not accreting significant quantities of material, or that it may not be a black hole. The paper notes that the companion cannot be a single main-sequence star (which they estimate would be of spectral type B7 or earlier) because this would appear in the spectral lines, which given the mass leads to the inference that the companion is a black hole as it is too massive to be a neutron star or white dwarf.

As far as I can see, I don't see any mention of the possibility that the companion could itself be a close stellar binary, i.e. the following architecture:

 ┌ B3III star
 │
┌┤
││┌ close binary, star 1
│└┤
│ └ close binary, star 2
┤
│
│
│
└ Be star

A binary of near-equal-mass stars would have lower luminosity than a single star of the same total mass, which might help it avoid detection.

I'm wondering if the configuration would be possible given the short period (40 days) of the companion's orbit, but maybe the data is sufficient to rule this out.

Answer 1

The absolute minimum mass of the unseen companion is $4.2 M_{\odot}$ (and my reading is that this really is an unlikely minimum, requiring an abnormally faint primary for its spectral type and an orbital inclination of 90 degrees, but let's go with it) and the argument is that a main sequence star of this mass would make its presence known in the spectrum. But let's suppose that instead of a $4.2M_{\odot}$ main sequence star, we had a pair of A stars with $2.1M_{\odot}$ each, in a tight binary. The luminosity of ths combined binary would be less than 25% that of a $4.2M_{\odot}$ single star, so might have gone undetected.

The orbital period of the B3 giant primary with the putative black hole is 40 days. Assuming the minimum mass of the system as a whole is $9.2M_{\odot}$ (for an inclination of 90 degrees), then Kepler's third law yields a binary separation of 0.48 AU.

The radius of a $2.1M_{\odot}$ A star (like Sirius A) is about $1.7R_{\odot}$, so the minimum separation of a pair of such objects in a binary system is about $3.4R_{\odot}$, or 0.016 AU.

As a rule of thumb, as long as the separation of the closely orbiting stars is less than about ten times that of the wider binary then that ought to be stable to tidal disruption (though it also depends a bit on the masses too).

There are examples of systems where there is a wide binary, where one of the components is itself a triple featuring another close binary, in the catalog of binary stars by Tokovinin (1997). These would be multiples with a level 112 in his notation and there are a dozen examples in this catalogue. Perhaps the most famous is Tau CMa, which consists of an O supergiant in a wide, just resolved, 250 year orbit with another bright B-star. However, the O-type star is in a 154 d binary system with a pair of eclipsing B-stars that orbit each other every 1.28 d.

So I think you have a legitimate case that this may not be what the authors claim.

Edit: I was interested enough by this to email Rivinius and put this scenario to him. The reply raised one or two interesting objections.

Firstly, he suggests that the reduction in luminosity of the secondary (compared with a main sequence single star of double the mass) might not render it invisible in the spectrum. The reason is that the A-type stars have narrower (metallic) lines that could be observed than the B-type giant. He cites the star 66 Oph as an example where this has been done. I am not totally convinced by this, or at least not convinced that it has been demonstrated.

The second point is better. If the inclination is anywhere near edge-on (as required for the unseen companion to be at its minimum mass), then in a close binary you would expect eclipses or at least significant ellipsoidal modulation due to distortion from spherical symmetry. Using Spica as an example (a 4d binary) Rivinius argues that even with dilution that this modulation would have been picked up in the SMEI and TESS light curves discussed in the paper.

If the binary were more face-on to reduce this ellipsoidal modulation, then of course the companion mass would increase dramatically from the 4.2 solar mass minimum, making even a binary companion impossible to hide.

So my conclusion changes. I'm reasonably convinced by the second argument - that a close binary companion would be revealed through ellipsoidal modulations.