Searching for monster stars

In 2010, Professor Paul Crowther and his team discovered the most massive known single star, R136a1.

Can we start by asking what first drew your attention to the R136 star cluster?

Well, it’s probably the prime target for anyone looking for the most massive stars – it’s the most obvious place to look really because it’s the most massive young star cluster in our part of the universe. It’s about the same size as the famous Orion Nebula, but while that’s got a couple of thousand stars, R136 probably contains 100,000 stars or more, if you could see them all. It’s been known about for a long time, but the exciting thing is that now, with Hubble and large ground-based telescopes, we can resolve separate stars and look at them individually.

As I understand it, there’s a really tight knot of stars at the cluster’s centre, called R136a?

Yes – and originally there were claims that R136a was a single supermassive star thousands of times more massive than the Sun. But about 25 years ago astronomers confirmed that it was actually a cluster and now, thanks to technological advances, we can finally analyse the individual stars within it. When our team looked at it with the European Southern Observatory’s Very Large Telescope in Chile, we were actually looking for binary stars, hoping we could use them to measure the masses of stars directly. We didn’t find any binaries, but we did find that the individual stars in the cluster, and the brightest one in particular, are far more exceptional than anyone had thought.

So a binary system would have let you measure the mass of its stars directly – but how do you work out the mass of a giant single star like R136a1?

The first thing you do is work out the star’s luminosity, but that’s a problem in itself. If you’re looking at a yellow star with the same surface temperature as the Sun, then it’s fairly straightforward – you’re seeing most of the radiation in the optical and can work out the total energy output quite easily. Red stars such as cool supergiants emit only a tiny fraction of their energy as visible light, but you can still measure them in the infrared, where most of the energy is coming out. The challenge with hot stars like R136a1 is that the energy’s coming out in the ultraviolet, at wavelengths that get soaked up by the interstellar medium on their way to Earth. We can’t measure the star’s peak energy directly, so we have to infer it in other ways, through other features of its light.

But even once you’ve got an idea of the star’s temperature and its overall luminosity, you still have to go an extra leg to get a mass. Fortunately on the main sequence there’s a clean relationship – the more luminous a star, the more massive it is. So for R136a1, where we came up with a luminosity not far off 10 million times that of the Sun, we asked our colleagues to work out evolutionary models for what the expected mass would be. That’s how we arrived at the figure of 265 solar masses, and the star probably started its life even bigger.

And is there any way to check that theoretical result?

Well, the problem is that you’re relying on one method to get a temperature, another to get a mass, and so on. For me as a sceptical scientist, that’s all a bit dubious – the figures are an interesting possibility, but not really backed up by enough evidence to prove it.

So what we did was go looking for another example of a similar star to prove the technique. Ideally we were looking for a star in a close eclipsing binary system [where the two stars regularly pass in front of each other as seen from Earth], which would let us work out the mass independently. We eventually found just such an object in a cluster called NGC 3603, about 25,000 light years from Earth. That’s now the most massive star system to be confirmed through the laws of orbital motion – it’s got two stars in an orbit of about four days, with masses of 120 and 90 Suns. Once we’d got those robust numbers for that system, we used them to test our temperature and luminosity-based method, and we got basically the right answer. So that was a sanity check – if it works for that object, there’s no reason why the method, and the final result, shouldn’t be correct for R136a1.

Like this post? Please share to your friends: