Where Are the Missing Black Holes in the Milky Way?

[Paul Colby]

Aren’t you assuming a 2.2 solar mass star will retain 2.2 solar masses in its later stages? Not a bad assumption, I admit, but an assumption nonetheless.

 

[Ken G]

Well, I only need a factor of 11,780. Seriously, stars can gain and lose mass. One of the first things I looked at with the telescope I bought a while back, was the ring nebula. A star that burped out a small fraction of its mass not that long ago. There are stars that do this periodically. Can one rule out from observation such slowish mass ejections as a means of avoiding BH at end of life? Along this line, do we see as many neutron stars as we’d expect?

You are exactly correct. As you saw from information above, a black hole is created when the core is above somewhere between 2.2 and 3 solar masses (no one knows the maximum neutron star mass). So one might expect any star above 3 solar masses to make a black hole. But that's not true, they don't even supernova at all, and everything below maybe 8 solar masses (Betelgeuse) creates a white dwarf instead of going supernova. Why is that?

It is, as you say, that the star loses its outer envelope before it can be added to the white dwarf. A white dwarf can only go up to 1.4 solar masses, so you see a lot is sneaking away: an 8 solar mass star leaves behind only about a 1.4 solar mass core. So some 6.6 solar masses, the majority of the stellar mass, leaks out before the star can go supernova, just as you suggest.

However, this fact is already included in the estimates @Vanadium 50 is using. He is not taking every star above 3 solar masses, but rather every one above 10 or 12 solar masses, to be the ones that turn into black holes. But I think the resolution to the paradox is a selection effect. To know a black hole is there, we need to see it, and generally that means it has to be in a close binary. That doesn't sound like a serious limitation, lots of massive stars start out with close companions, the problem is the close companion has to still be there after the supernova. It's not that the supernova destroys the companion, a star is a tough nut, it's that it unbinds the companion, which becomes a "runaway star." The result is you don't get to see the black hole.

Now, there is still a kind of special dance a star can do such that the black hole sticks around, and this is why we see binaries with black holes in them. What happens is, the star that will become the black hole has its envelope expand and some of it is lost and some transfers to the companion, leaving a stripped core that is still going to go supernova even though it is way below 8 solar masses (it is the mass of the core that counts, not the mass of the whole star). So a star like that could explode and not lose its companion (because by then the companion has more mass than it does, so it's really the companion that's holding it, not the other way around). These kinds of systems in their presupernova state were predicted to happen but have only recently started being seen, so a full accounting of how many there are has not been done.

The upshot of all this is, the resolution is probably (it seems to me) a combination of two factors. The first is, most of the time it is actually still pretty hard to see a black hole even in a binary, but it does seem like a much more important effect must be that it is very very hard to make a black hole and keep a close companion around, because the creation of black hole disrupts the binary you thought you were going to use to be able to detect it. But I don't say there isn't a need to count the systems that should lead to black hole binaries, and connect that to the black hole binaries, every time we start encountering things we haven't seen we always get surprised by something we didn't know was happening. (An example is supernova "kicks," it was always assumed a supernova would be spherically symmetric so would not kick the star that was blowing up, but nature abhors spherical symmetry, and so the black hole can get a pretty good kick and who knows how much that might help it get away from its companion.)

Finally, we can note that a good way to find planets is to look for radial velocity variations of the star, so the more that goes on, the more we might expect to come across invisible companions like black holes. The problem is, if the separation is wide enough that we haven't seen it yet, it means the period is very long, so we may have to watch for a while, many decades perhaps. Could take some persistence.

 

[Vanadium 50]

Aren’t you assuming a 2.2 solar mass star will retain 2.2 solar masses in its later stages?

No, I am assuming a 10 solar mass star will leave behind a 2.2 solar mass core. I believe I wrote that,

 

[snorkack]

Note that about half a percent of the visible stars are SN progenitors. If you want to start playing with that, you need to keep to the constraint of how many progenitors we see. You could even turn the question around - given how few BHs we see, why do we see so many progenitors?

But maybe it is easier to not bother deriving "progenitors"? Start straight from SN?
In the last millennium, there have been 6 visible SN. 4 within Milky Way (1054, 1181, 1572, 1604 - note that 1006 is no longer in the last millennium!) and 2 without (1885 in Andromeda nebula and 1987 in Big Magellanic Cloud). None of them had a visible progenitor (that of SN1987 was +12).
What was the outcome? Three possible ones: whole star blown up, neutron star or black hole. Crab formed a neutron star. What were the results of the other 5?
How often does a visible supernova form a black hole? Do any black holes form otherwise than by supernova?How many black holes formed in the last 10 Gy should we be seeing? Before 10 Gy, Milky Way must have had starburst and formed black holes (and neutron stars!) at rates higher than present.

 

[Ken G]

I think if we do what @snorkack suggests, the numbers come out pretty consistent with what @Vanadium 50 got. It's often said the MW has about 1 SN per century, roughly equal numbers of core collapse and type Ia, so over 10^10 years, we can expect something like 10^8 black holes, or maybe it's only 10^7 but it's not less. So the point still holds, there should be a whole lot of BHs out there we have not detected, and whatever is making them hard to detect must be the fate of the vast majority of them, so that's why I think they must be losing their close companions. There are various ways to make that happen, but it has to be something that happens an awful lot, so it must be something that seems almost inevitable. So that could well be that the mass lost by the SN that makes the BH unbinds the binary, and we are left to explain the rare situations where that doesn't happen, allowing us to see any of the BHs.

 

[snorkack]

Ask it this way: out of the 6 visible supernovae of last millennium, SN1054 formed a neutron star. What did the other 5 do? Can we name any black hole we observe as formed by a historic supernova?

 

[Ken G]

Ask it this way: out of the 6 visible supernovae of last millennium, SN1054 formed a neutron star. What did the other 5 do? Can we name any black hole we observe as formed by a historic supernova?

Yes that's a good way to get insight. We can even venture out of the Milky Way and look at SN 1987a, where JWST was able to find the neutron star. I'm not sure how the neutron star shows up in infrared light, but in any event, it has been seen. So we're getting pretty good at linking SNe to NSs if they make them, but you want to consider the ones expected to make BHs and ask if we have any way to detect those. I think we already know the answer will be no, but the exact reasons for this is what we are discussing.

 

[Vanadium 50]

There are two possibilities:

1. We make fewer than we think.
2. We detect fewer than we think.

That's it.

Now, the thread was restarted by someone who didn't like my estimation of #1. He has not shared a calculation, but there is a big difference between one supernova every 30 years and one every 600,000 years.

We can look at nearby neutron stars. Inside the radius of the nearest black hole candidate, there are about a dozen neutron star candidates. About half are nearby isolated X-ray sources, and are dimming as they cool, and the other half are mixes of pulsars and SN remnants thought to contain relatively quiet compact objects. All are very young - say ten million years. So there should be a few thousand of them at least as close as the nearest BH.

That more or less agrees with the estimation in the OP.

Do we see neutron stars in binary systems? All the time. There are thousands of millisecond pulsars. There are hundreds in nearby globulars. Most companions are red dwarfs, partly because they are by far the most common star, and because globulars are Pop II stars - i.e. old - and red dwarfs live a very long time. Look in a place full of sold red stars and what do you fine? Old, red stars.

So where are they? To paraphrase @Ken G , you need a process that hides 99.9% of them, but does not hide 100% of them. Further, whatever this process is, it needs to be an order of magnitude better at hiding black holes than neutron stars.

 

[snorkack]

There are two possibilities:

1. We make fewer than we think.
2. We detect fewer than we think.

That's it.

Now, the thread was restarted by someone who didn't like my estimation of #1. He has not shared a calculation, but there is a big difference between one supernova every 30 years and one every 600,000 years.

So, what is the observed branching ratio of supernovae for results - neutron star, nothing, black hole?

 

[Ken G]

I think it is not terribly surprising that BHs are found much less often than NSs, because NSs have their own beams that you only have to be in the path of to see. So NSs don't need to be in binaries for us to discover them (and the Wiki on NSs says only 5 percent are in binaries, and it's not clear how many of those we would have known are NSs had we not seen their beams).

But it is interesting that 5 percent of the NSs have managed to be formed without losing their companion, so we might imagine some similar percentage might be true for black holes. So loss of companion can probably only explain 1/20 of BHs being detected, not 1/1000. But maybe if NSs where not beaming, we would miss a lot of that 1/20 that are in binaries (and all the ones that are not).

Possible numbers here are that something like 1/20 compact objects keep their companions, and only some 1/20 of those can be detected based on their binary interaction, meaning that you only see 1/400 (roughly) of compact objects if you have no direct means of detecting them other than binary interaction. That could be roughly what is going on for BHs, though I admit only being able to detect 1/20 that are in a binary seems surprisingly low.

Since comparison with NSs seems useful, another way to frame the question of the thread is, pretend we have no radio antennas to see pulsars, so NSs are just as hard to find as BHs. What fraction of the NSs we now know about would we still know about?

 

[Vanadium 50]

because NSs have their own beams that you only have to be in the path of to see.

But not for very long. A few million years and off they go. Seven of the nearby ones don't pulse at all - they are just weak x-rays sources. Once their surface cools, on a similar timescale, they can only be detected gravitationally or by accretion x-rays, and in both cases, it should be easier to detect BH's.

pretend we have no radio antennas to see pulsars, so NSs are just as hard to find as BHs. What fraction of the NSs we now know about would we still know about?

Many.

Of the nearby ones I mentioned, seven are radio quiet x-ray sources. If you look at Chandra data, globular clusters have many x-ray sources, not all of which are neutron stars, to be sure, but a lot of them are. These are all in the outer part of the cluster; there are too many in the core to separate one from its neighbors.

There are some nice pictures from Chandra on this, and nicer ones with composites with optical pictures of the globular. Since globulars are gas-poor, these are probably oversampling neutron stars in binary systems.

 

[Ken G]

But not for very long. A few million years and off they go. Seven of the nearby ones don't pulse at all - they are just weak x-rays sources. Once their surface cools, on a similar timescale, they can only be detected gravitationally or by accretion x-rays, and in both cases, it should be easier to detect BH's.

OK that's a good point, gravitational detectability can last billions of years, so one can ask how many NSs are known by their pulsing, and how many are known gravitationally. Since the Wiki says only 1/20 of known pulsars are in binaries, we see that many more of them are known by their pulses than by their gravitational effects. The question is then, of those 1/20 that are in binaries, how many of them do we know only from their gravitational effects? We might end up deciding that without pulsars, we'd only know some 1/100 or less of the NSs we currently know about. That's not so far from the shortfall of BHs you are pointing out. So it still sounds like the combination of how few are in binaries, and how hard they are to detect in binaries, accounts for how few BHs we know about.

Many.

Of the nearby ones I mentioned, seven are radio quiet x-ray sources. If you look at Chandra data, globular clusters have many x-ray sources, not all of which are neutron stars, to be sure, but a lot of them are. These are all in the outer part of the cluster; there are too many in the core to separate one from its neighbors.

There are some nice pictures from Chandra on this, and nicer ones with composites with optical pictures of the globular. Since globulars are gas-poor, these are probably oversampling neutron stars in binary systems.

Perhaps there is a problem with the Wiki claim that only 1/20 of NSs are in binaries, they might not be including this very large number of Xray sources that are not specifically identified as NSs but which probably are. However, if we include those, why can't many of them be BHs?

 

[Vanadium 50]

I am not confident of the Wikipedia claim. Apart from the general issue of a wide range of validity, I don't think we know that number well. Consider this - if we only detected millisecond pulsars, we would conclude that all neutron stars are in binaries.

Looking at some globular cluster data, it looks like the 1/20 number is more like 1/10 or higher. Is this sample less biased? I don't know - it is certainly differently biased. And, as has been said, most companions are red dwarfs (for obvious reasons) and most of the searches thus far for unseen companions have (again, for obvious reasons) concentrated on brighter stars.

The nearest black hole orbits (or rather, is orbited by) a G main-sequence star. So a few thousand times brighter than a red dwarf.

 

[Ken G]

I am not confident of the Wikipedia claim. Apart from the general issue of a wide range of validity, I don't think we know that number well. Consider this - if we only detected millisecond pulsars, we would conclude that all neutron stars are in binaries.

Yes, binaries do allow conditions to keep the pulsars shining, making them more visible as with black holes. But I do think it's true that most pulsars are not in binaries (consider for example those two most famous pulsars, Vela and Crab). So there is an advantage for seeing pulsars that black holes don't share.

Looking at some globular cluster data, it looks like the 1/20 number is more like 1/10 or higher. Is this sample less biased? I don't know - it is certainly differently biased. And, as has been said, most companions are red dwarfs (for obvious reasons) and most of the searches thus far for unseen companions have (again, for obvious reasons) concentrated on brighter stars.

The nearest black hole orbits (or rather, is orbited by) a G main-sequence star. So a few thousand times brighter than a red dwarf.

There can be some mass selection effects. Massive stars tend to orbit other massive stars, since angular momentum storage is more or less the purpose of forming in binaries, so it's most efficient to split the mass up more or less equally. Still, you point out that NSs often have red dwarf companions, which surprises me for a different reason. A low mass companion should be lost when the supernova occurs, since the NS progenitor is losing most of its mass. So how do those NSs end up with red dwarfs bound to them?

 

[Vanadium 50]

So how do those NSs end up with red dwarfs bound to them?

I do not know.

These are from globular cluster data, and that may be relevant. Since gravitational capture requires at least three bodies, the greater star density in globulars may play a part. Or maybe it has nothing to do with that. It sounds unlikely, but other features of globulars - low metals, low gas, and such sound even less likely.

One thing I do strongly suspect is that whatever the reason is, it does not depend on the remnant being a black hole or neutron star.

 

[Ken G]

I do not know.

These are from globular cluster data, and that may be relevant. Since gravitational capture requires at least three bodies, the greater star density in globulars may play a part. Or maybe it has nothing to do with that. It sounds unlikely, but other features of globulars - low metals, low gas, and such sound even less likely.

One thing I do strongly suspect is that whatever the reason is, it does not depend on the remnant being a black hole or neutron star.

I agree that whether it has 2 solar masses or 3 shouldn't matter much, but that's why I'm wondering why those Xray sources in the globulars couldn't be a whole lot of BHs along with even more NSs. If so, the question is not so much where are the BHs, it is where are the BHs and NSs that we see in globulars that we don't see nearer to us in the galactic disk. So maybe it does have to do with gravitational capture in high density star clusters, which lights up the accretion. But I don't know how often one expects a wandering BH to grab a low mass companion.

 

[Vanadium 50]

But I don't know how often one expects a wandering BH to grab a low mass companion.

Neither do I. It surely happens more in globulars than it the local galactic neighborhood, but that doesn't say anything about how often it happens. After all, one in a million is larger than one in a billion, but that doesn't make it large.

I would expect it to be harder to see compact objects via accretion in globulars since they contain little gas. So we are likely undercounting them.

 

[mfb]

Sleeping giant surprises Gaia scientists
A 33 solar mass black hole (Gaia BH3) 1900 light years away. It's not the nearest one, but it's an interesting object - the heaviest stellar-mass black hole known in our galaxy. It's orbited by a star at a distance of 16 AU. So far away that the black hole is very quiet, and you need a precision astrometry satellite to find it via the motion of the star. The next data release should contain many more objects:

The next release of Gaia data promises to be a goldmine for the study of binary systems and the discovery of more dormant black holes in our galaxy. “We have been working extremely hard to improve the way we process specific datasets compared to the previous data release (DR3), so we expect to uncover many more black holes in DR4,” concludes Berry Holl of the University of Geneva, in Switzerland, member of the Gaia collaboration.