Stung by a WASP
Here’s an example of something that’s so very, very frustrating.
Basically it’s an object that’s been spotted in a transit survey. As such, it has a robust radius and a reasonably-robust mass estimate. It is in mutual orbit with a reasonably well-behaved F9-type star. That is to say, a spectral type where it’s possible to do age determinations and metallicity estimations, at a level of passable precision. Now, as presumably star and brown dwarf companion must have formed together, we’d expect a common composition. (Or at least, we can reasonably assume a common composition.)
Now, brown dwarfs have some joys but also some nuisances. The major nuisance is that there are a lot of degenerate properties – that is, features of the spectrum that can be affected in the same way by several different processes. Obviously, this makes distinguishing between them really, really difficult.
As an example, composition – the fraction of heavy elements (metallicity) will change the spectrum of a brown dwarf, because different compositions absorb light differently. One such effect is called ‘collisionally-induced hydrogen absorption’. I shan’t go into too much detail (unless anyone wants me to), but suffice to say that you get more of it in lower-metallicity atmospheres. But, there’s a problem – what’s creating it isn’t directly the metal content, but rather the effects that has on the pressure of the atmosphere. And high surface gravity can also increase the pressure. So you can certainly find brown dwarfs that show CIA H2, but you’re never entirely sure how much of what you’re seeing is down to high gravity and how much to low metallicity.
This is rather annoying, to say the least.
Now, you can possibly tease out some more information by comparing object spectra to theoretical models. Models, in this case, are spectra that have been calculated assuming different sets of temperature, gravity, metallicity and so on. Except, obviously, the problem here is that purely-theoretical models are no better than the assumptions that go into them. There is a phrase in computer science, called JIJO, ‘Junk In = Junk Out’. Ideally, the way we make sure that the models do actually mirror reality is by comparing them against real objects – only the spectra are degenerate! So, for instance, if you’re trying to constrain metallicity effects, it helps to be sure that the features you’re looking at are created by metallicity, not surface gravity.
Except, of course, you never know that for sure.
Now, one possible way around this is by looking at not one but every brown dwarf (of a given spectral type). You could perhaps create an ‘average’ spectrum, out of all of those dozens of others. At this point, you make the assumption that this spectrum conforms to the galactic averages for metallicity and surface gravity etc., and as long as you know what those averages are, you can now go about constraining the models. And, you can take at least a punt at those numbers based on arguments from stellar evolution and (I suppose) population studies of stars.
Problem is, of course, those numbers will still come with health warnings. Also, there are still fewer than a thousand or so confirmed brown dwarfs (and most of those are L dwarfs – it’s much worse for T dwarfs). Compared to the estimated ~100 billion or so actual stars in the galaxy, this number isn’t sounding so statistically-robust, is it? And that number will be biased toward the brighter, easier-to-detect objects, as well. So, in fact, we’re already sliding back toward square one. We’re trying to analyse the population as a whole using objects that we already know are probably abnormal.
Bugger. This isn’t looking so good, is it?
But, there is another approach. It’s called binary benchmarks. Basically, I mentioned common composition in binary systems. Suppose you have a star and a brown dwarf. Say the star has solar metallicity and an age of 5 billion years. Then you can reasonably assume that the brown dwarf has the same too. Suddenly, the only free parameter that’s left is surface gravity! The other problems disappear almost like the morning dew.
Binary benchmarks are incredibly important for brown dwarf research.
This is why objects found in transit surveys are so frustrating. Take the one linked to above. It could almost be the perfect benchmark. A robust age constraint, a radius, a mass, a robust metallicity – we can calculate surface gravity directly, there’s not even any need to use the models there! So far, so wonderful.
Except there’s one final, maddening problem.
The irritating little bugger has parked itself in a 0.05 AU orbit around a main sequence star. The star’s age is believed to be around 1-2 billion years. Now, brown dwarfs cool and fade as they age. Assuming it’s behaving exactly like the Tucson group models, then for this mass you would expect a luminosity of about 0.03% of the Sun by now. (This is based on a plot in the book in front of me, and a quick bit of pencil-and-rulering. Don’t quote me on these numbers, they’re very probably wrong!)
0.03%. That’s not much.
And to make it worse, it’s less than 0.1 AU from a star that is hotter and brighter than our Sun. Or, to put it another way, there seems to be no hope in hell of getting out a spectrum for this object. The brown dwarf will be pretty much entirely lost in the glare. It’d be like looking for a candle next to an search-and-rescue floodlight. And it’s the spectrum that we need for model-constraints. And it’s the spectrum that we have little hope of extracting.
Now, obviously, the numbers above are probably a bit off. Nonetheless, they give you an idea of the size of the problem. And it’s a real pity, because otherwise, transit-searched objects would be an ideal group for use in brown dwarf research.