Dwarf merging makes for an explosive combo

By Phil Plait | February 18, 2010 7:22 am

Type Ia supernovae are very important exploding stars. It’s thought that this particular type of supernova has a very special property: they all explode with about the same energy. This makes them very valuable, because it means that if you can simply measure how bright they appear to be, you can figure out how far away they are. It’s like seeing headlights on the highway; dim ones are far away, and bright ones are close.

hst_sn1994dOf course, in reality, it’s not that easy. But after a Herculean effort, astronomers in the late 1990s figured they had been able to account for any small differences in brightness and could use these stars as "standard candles", benchmarks to calculate cosmic distances. Because they’re so bright, they make great milestones because they can be seen pretty much all the way to the edge of the observable Universe.

The thing is, it’s not clear how a type Ia actually forms. There are two models, both involving white dwarfs. These are the ultradense remnants of dead stars, the exposed cores of stars after they shed their outer layers. The Sun will one day be a white dwarf (in about 6 – 7 billion years, so don’t hold your breath). Because of complicated quantum physics, it turns out that white dwarfs can only have so much mass; if they exceed about 1.4 times the mass of the Sun they can collapse, either forming an even denser neutron star, or exploding as a supernova.


The first model of a Type Ia is a white dwarf orbiting a star like the Sun. The intense gravity of the dwarf draws material off the normal star, a process called accretion. The matter piles up, the mass limit is exceeded, and BANG! Supernova.

Well, it’s a lot more complicated than that, but close enough.

The second idea is that you have two white dwarfs orbiting each other. Over time they spiral in (this time due to relativistic effects called gravitational waves), get too close together, merge, and BANG! Supernova.

Astronomers have always assumed that the accretion scenario is the far more common of the two, because it takes a long time for two dwarfs to merge, whereas accretion can happen easily if a dwarf happens to be paired up with a normal star (which should be pretty common). But how do you tell which is which?

It turns out that the two different scenarios leading up to the explosion have two very different effects: accretion makes a lot of X-rays, while a merger does not. So astronomers did what you’d expect: they pointed the Chandra X-Ray Observatory at a bunch of galaxies and observed supernovae. What they found was pretty surprising: the amount of X-rays from Type Ia supernovae in nearby galaxies was 30 – 50 times lower than what would be expected from accretion. In other words, their observations strongly favor the idea that it’s the merger of white dwarfs that cause Type Ia supernovae.

Well! I was pretty surprised to hear that. Like other astronomers, I figured it was accretion that was the culprit. Now mind you, there are some caveats here. They observed elliptical galaxies, which tend to have an older population than spirals, so you might see more mergers than accretions. Also, it’s possible things were different in the past, and when we observe very distant galaxies were seeing them as they were billions of years ago.

But still, you just don’t expect to see what the astronomers saw, so it seems to me like they’re on to something here.

This has some interesting ramifications. It certainly affects a lot of fields of astronomy, like how binary stars form and change over time. But it may also affect cosmology, the study of the birth, evolution, and eventual fate of the Universe itself. If Type Ias are caused by a different scenario than previously thought, could it mean that our measurements of the distant Universe are wrong?

I asked this question specifically at the Chandra press conference, and was told that the two different scenarios produce explosions with pretty much the same energy, so this may only affect cosmological measurements a small amount. However, right now our theoretical models of the merger scenario are still pretty rough, so it’s unclear if the peak brightnesses of the two models are the same.

This may affect our measurements of dark energy, the mysterious pressure that seems to be accelerating the expansion of the Universe. My gut reaction is that this won’t matter a huge amount, since we have lots of independent ways of measuring dark energy, and they all appear to be in rough agreement. But this means we have one more thing to take into account in those measurements. And it may prove to be useful; if we can distinguish between the two supernova generators, our measurements will get that much more accurate.

I have to say I’m pleased with this; I studied supernovae in college and grad school, eventually studying one for my PhD (though it was of an entirely different flavor from this kind). I remember reading a long technical paper about the different Type Ia scenarios back then: it’s been a mystery for a long, long time. But with perseverance, the right equipment, and more than a touch of cleverness, we’re a big step closer to figuring this all out!

Related posts:
Fireworks and Pinwheels (an overview of Type Ia supernovae)
Dark Energy site open for business (explaining dark energy)
The Universe is 13.73 +/- .12 billion years old
What astronomers do (about the discovery of dark energy)
The cosmological not-so-constant

Image credits: NASA, ESA, The Hubble Key Project Team, and The High-Z Supernova Search Team, and NASA/CXC/M.Weiss (adapted a bit by The Bad Astronomer)

CATEGORIZED UNDER: Astronomy, Science

Comments (39)


    Dr. Phil Plait:

    If Type Ias are caused by a different scenario than previously thought, could it mean that our measurements of the distant Universe are wrong?

    I’ll wager that those damn creationists will take that quote out of context!

  2. Lee

    Thanks for posting this Phil! I had not yet seen this — do you know any paper titles on this yet, or was it just a NASA press release?

  3. I don’t have access to the papers, and probably wouldn’t understand them if I did, but I do have a question:

    The key characteristics of type Ia supernovae are that the luminosity and light-curve are consistent. Because the light-curve is consistent we are able to distinguish type Ia from other types and because the luminosity is consistent we can determine how far away it is by relating it’s apparent magnitude to the expected magnitude.

    The accretion model explained this very well. Material is added to a white dwarf that is below a critical mass. When enough material is added that the star tips over the critical mass the energy released in resulting explosion is very predictable and consistent.

    Since two merging white dwarfs can have masses that vary by a large amount (so long as they are each below the critical mass) how does the proposed theory explain the consistency seen in type Ia supernovae? Are the authors proposing that all white dwarfs have the same mass?

  4. AliCali

    There’s a question that’s bugged me for a while. Dr. Plait says, “The first model of a Type Ia is a white dwarf orbiting a star like the Sun. The intense gravity of the dwarf draws material off the normal star…”

    If the white dwarf has more mass than the companion star, would the companion star orbit the white dwarf? Doesn’t the lower mass object orbit the higher mass object?


    @ AliCali,

    Err… no. The white dwarf is more dense than its companion star, but it does not have a higher mass than the companion star; for instance, lead is over 11 times more dense than water, but 10 kg of water will have 10 times more mass than 1 kg of lead.

    So, the white dwarf is orbiting the larger companion star.

  6. I understand where AliCali is coming from. If both stars formed at the same time, then the one with higher mass is the one that would become a white dwarf first, having burned through its fuel faster. However, stars also eject a great deal of mass when they go through the red giant phase, so the white dwarf does not contain the entire mass of the original star.

  7. Lee

    @ AliCali

    The companion star could have more or less mass than the white dwarf. The force of gravity always acts equally on both objects, so they orbit around their common center of mass. We don’t notice the Sun’s movement due to the gravity of Earth because the Sun is so much more massive than Earth that this motion is very small.

  8. Chris

    For those interested in the “complicated quantum physics” that Phil was afraid to explain

  9. Scott B

    @ AliCali

    Just to add to what Lee said, here’s a picture that shows the movement of the sun from 1945 – 1998 compared to the size of the sun. As you can see, the influence of the planets has a significant impact considering how much more massive the Sun is.


    So with two stars, unless one is much more massive than the other, they will both be orbiting a center of mass that is in between the two stars rather than one appearing to orbit around the other.

  10. AliCali

    Thanks for the responses.

    So the dwarf is more dense, but with less mass. So material is going from the main sequence star onto the dwarf even though the dwarf is orbiting the main sequence star. This is because the dwarf is more dense.

    So am I correct in thinking that the overall gravity/mass of the main sequence star is higher, but since it’s less dense, the edge of the main sequence star feels less gravity from its own star and more gravity from the orbiting dwarf, since the dwarf’s gravity is more concentrated in one spot?

    Naked Bunny brought up something as well. Why did the orbiting star become a white dwarf first? I understand that at the end of a star’s life, it sheds a lot of mass. Oh, maybe the white dwarf wasn’t originally the orbiting star, but when it lost its mass, the dynamics changed, and the common center of mass shifted toward the still-main-sequence star?

    Jeez, I should really take an astronomy course. I originally got into this just to look at the stars and pretty pictures of them, but the physics is fun, too.

  11. DougA

    Don’t worry, constraints on Dark Energy are safe even if there are some problems with the Supernovae data. Measurements from massive, x-ray luminous galaxy clusters + cosmic microwave background + hubble constant measurements currently detect dark energy to the same significance as the supernovae data. And besides, there are other problems with the supernovae data — until recently, the supernovae teams had discrepant results, primarily based on how they handled dust.

  12. Unfortunately, I have already bumped into the limit of my knowledge, and I know how poor a guide intuition can be.

    I did find this paper interesting, albeit hard for the bunny to follow at times. (Note: PDF file.) It gives you a better idea of the complexity of the situation, and I suspect Phil saying “the white dwarf orbiting the companion star” is just shorthand for a wide variety of stellar combinations.

  13. Doug

    @AliCali – actually both stars are orbiting the center of mass of the system (the point where the mass of each star times the distance it is from the center of mass it the same for both stars). If one star is much more massive, the center of mass will be within that massive star, but for more equal mass systems both points will have the focal point in their orbit be a piece of empty space.

    The white dwarf is usually less massive than the companion, but because of the 1/r^2 gravitational force law, where will be a region around the white dwarf where it exerts a greater gravitational force than it’s more massive companion. When the companion star enters its giant phase, the outer edge of it can pass through his boundary, which allows the white dwarf to start stealing material by gravity – http://en.wikipedia.org/wiki/Roche_lobe has a basic description of how this happens.

  14. DougA

    We can make an analogy to the Earth & the moon try to to see this better. As the Earth-Moon system orbits, the moon causes tides in the Earth’s oceans — really the moon is tugging on the nearest ocean the most, the earth as a rigid body second-most, and the far ocean the least. If the moon was much more massive, then as you correctly said, the moon will essentially rip off water from the Earth. The technical terms for this are the ‘Roche limit’ and ‘Roche Lobe.’ http://en.wikipedia.org/wiki/Roche_limit http://en.wikipedia.org/wiki/Roche_lobe

    Remember that both stars orbit each other. Their distance may change over the life of the system, especially as one star explodes and ejects mass. So before the explosion, the two stars are happily orbiting each other, then after the explosion, the orbit changes such that gas is stripped off.

    As to why one star turned into a dwarf first? Main sequence stars in binaries may be born with different masses — and therefore different ages. Its just luck of the draw.

  15. Minos

    In the mess that is my feed reader, I thought at first that this title was from one of the World of Warcraft blogs I read. Fortunately, that sort of dwarf doesn’t tend to cause supernovas.

  16. kevbo


    I think Dr. Plait could recommend a book about this.

  17. @ #1
    Hehe, yea, I expect them to quote it as:
    -Phil plait: ”our measurements of the distant Universe are wrong’
    or something similar.

    Stories like this are what makes science cool.
    How observations are made and hypothesises are formed. Tests are conducted and we form a theory. Then many years later we conduct further tests with better equipment and our knowledge about the subject grows and evolves.
    Even after having a theory for so many years, we still test our knowledge and are willing to learn from new observations. Amazing.

  18. This is definitely an interesting result, and one that will require us to revisit our assumptions about the nature of thermonuclear supernova progenitors. I’m not prepared to go as far as the authors are in their conclusions, namely that accreting white dwarfs can’t be the cause of most Ia SNe. In their paper, they assumed a single X-Ray luminosity for an accreting binary system, based on fusing H to He on the surface of the star via the proton-proton chain, assuming (as far as I can tell) that all that energy comes out in X-Rays. This may or may not be the case. In numerical simulations of accreting white dwarf systems, the X-Ray luminosity can vary greatly, depending on all sorts of factors.

    There are a lot of problems with the double degenerate scenario as well. In particular, two white dwarfs at a reasonable binary separation will take an awfully long time to radiate enough energy away in gravitational waves to get close enough to merge, in many cases longer than the age of the universe (not to mention the time it takes for them to become white dwarfs in the first place). Models of this scenario are unfavorable to the idea that this can be a common occurrence in the universe. But in light of these findings, these models will certainly have to be revisited!

    This paper is an important first step in what I’m sure will be many studies like this to come. It will definitely force us to re-evaluate just how common accretion scenarios are, and how well we really understand them (hint: not very well, as this paper shows!). But we’re not quite ready to rule out horses in favor of zebras just yet…

  19. Luke

    For those still a little confused about the mass transfer mechanism, look up Roche lobes. The important distinction is, for a given point between the two objects, which object dominates the gravitational field at that point.

  20. Chris (8): I’m curious about your phrasing. I’m “afraid” to talk about it? Nope. It wasn’t necessary to the post and I didn’t want to distract from the main point.

  21. Ryan The Biologist

    So this might sound like a strange question, Phil, but how can we be sure that all these type Ia supernovae explode with the same energy if we are “not clear how type Ia actually forms”? If we knew exactly how they formed, I could buy into the idea that we could predict that would they all have the same energy and be able to base astronomical measurements on that. But if we do not have their formation clearly known (especially if we are still making discoveries as significant as this), doesn’t the assumption that all type Ia supernovae explode with similar energy seem somewhat premature? And further, doesn’t it then follow that using them to gauge astronomical distances might also be premature?

  22. Jason Dick

    Like NoAstronomer in #3, I’m extremely skeptical of the merger scenario. The extreme consistency of brightness that we do see in Type IA supernovae seems to make it extraordinarily unlikely that they are primarily the result of mergers (though there has been at least one discovered with brightness much higher than normal Type IA’s, despite having a similar spectrum: that was probably a merger).

    My suspicion would be that we’re not seeing the X-rays because there is some other process at work that prevents us from seeing them. In other words, I think the smart money is that this is telling us something interesting about the accretion scenario, as opposed to saying that it’s not accretion at all.

  23. > do you know any paper titles on this yet

    I do: The paper is in Nature today, which is subscription only – but as good astronomers do these days, the authors have put the complete paper online for free at the famous ‘pre’print server immediately after formal publication.

  24. Ray Wagner

    The inner core of a red giant is essentially a white dwarf. If all the matter above the hydrogen-burning shell atop the core is stripped off through the giant’s Roche lobe into an accretion disk that is eventually dissipated, we will be left with a two-white dwarf binary that would fit the merger scenario. We already know that accretion disks seem to result in a lot of mass loss perpendicular to the disk — we see jets aligned with the poles in a lot of situations. So, we may have learned something interesting about the mass loss and disk behavior. Perhaps, the disk manages (usually) to eject enough matter from the binary system to prevent the first white dwarf from accreting enough to pass the Chandrasekhar limit. This mass loss also may accelerate the closure of the binary separation. So maybe we are left with a scenario in which the accretion disk comes first, and ultimately leads to the merger of two white dwarfs: the original plus the red giant core. We need to identify why and how the disk and jet phenomenon ejects just the right amount of mass to keep the original white dwarf from overloading.

  25. AliCali

    Thanks, everyone, for the explanations. I understood how there is a center of mass that both objects orbit, just like how the Sun is tugged by the planets. I wasn’t clear that the dwarf has less mass but more density, so it is orbiting other star more than vice versa. It was that language I wanted to understand.

    What really brought it home is that the main sequence star will undergo expansion, and then the edge enters into the dwarf’s dominion and then the material gets stripped away.

    At first, I didn’t care about Roche’s ears, but now I see that ‘lobes’ refer to something else, so now I’ll go learn more about that.

  26. The calibration of Type Ia supernovae as standard candles is done on local supernovae in many different stellar environments. So even if there are many types of progenitors (double-degenerate, single-degenerate, etc.), we at least know that phenomenologically all these local Type Ia supernovae follow the same calibration. The bigger uncertainty is whether supernovae at high redshift are qualitatively different from local supernovae. We’re reasonably certain that they’re not, because we have probed so many different types of environments locally, but the possibility still remains.

    The upshot of all of this is that their usefulness as standard candles has already been established, independent of what is making the explosion. Even if we constrain what the predominant Ia progenitor is, it’s probably not going to change our conclusions about dark energy.

  27. Brian Too

    Isn’t the precise nature of the triggering condition much less important than the determination that at least one of the bodies was initially a white dwarf?

    My understanding was that the condition upon which the Type Ia supernova forms was a catastrophic collapse of a white dwarf into a neutron star. You have a white dwarf, it accumulates too much mass and exceeds the atomic pressure limit, causing all those atoms to collapse into neutrons.

    Whether the forming event is an accretion disk or a neutron star merger, wouldn’t this be a wash in terms of the brightness of the Type Ia? You did say that the accretion disk puts out a lot of X-Rays, but wouldn’t this be a relatively small output when compared to the supernova event?

  28. Cindy


    I did my Ph.D. thesis on these type of binaries (called cataclysmic variables or cataclysmic binaries). Previous posters are wrong. The white dwarf is the more massive star of the pair (as well as being more dense) as it became a red giant earlier (and probably engulfed the companion, dragging it closer). The companion is typically a K or M type main sequence star (smaller than the Sun).

    The orbital periods of these binaries are typically 1 – 2 hours or greater than 3 hours, so they are quite close together. The above links about Roche lobe is correct, as that is how gas is pulled off the companion. Basically on the surface of the companion star, the gravitational force from the white dwarf is stronger than gravitational force of the rest of the companion star, so the outer layers are literally pulled off.

    You can read more on Wikipedia: http://en.wikipedia.org/wiki/Cataclysmic_variable_star

  29. mfumbesi

    That was beautiful (not that I celebrate the death of stars, it just the description and the science, ahhh poetry, poetry). Thank you for the explanation.

  30. Murff

    I’ll take this time to thank Phil for his blog. I love this stuff, and he always makes it a good read.


    @ Cindy, and AliCali,

    Reviewing the (brief) answer I gave in my post above, I now realize that I should have been more specific, but I was in a rush. I had based my answer on the Sirius binary system, in which the two stars orbit around their common centre of gravity at a distance of about 20 astronomical units (approximately the distance between the Sun and Uranus) from each other over a period of 49.9 years. Sirius A, the brighter component, is a main sequence star of spectral type A1V, with a mass of about 2.1 times that of the Sun; whereas Sirius B, a white dwarf, is 0.98 solar masses, which is nearly double the average of 0.5 to 0.6 solar masses for white dwarfs, but packs that mass into a volume approximately the size of the Earth!

    The age of the Sirius binary system has been estimated at around 230 million years. Since a white dwarf only forms after the star has evolved from the main sequence and then passed through a red giant stage, it is estimated that when Sirius B was less than half its current age, approximately 120 million years ago, the original star had a mass of about 5 solar masses and was a type B4-5 star when it still was on the main sequence before eventually shedding most of its mass after passing through the red giant stage; it may also have enriched the metallicity of its companion as a result.

    It is predicted that Sirius A will have completely exhausted the store of hydrogen at its core within a billion years of its formation. At this point, it will also pass through a red giant stage, then eventually shed its outer layers and settle down to become a white dwarf; however, I do not think that this binary star pair are close enough for mass transfer to occur resulting in a nova or Type Ia supernova. (Just as well; otherwise, it might be curtains for all of us!)

  32. Messier Tidy Upper

    Awesome post BA – loved it. ūüėÄ
    Wish I’d checked in here & read about / posted on this sooner.

    Also surprised to hear this as I too thought accretion was the main cause for type Ia Supernovae & not merger.

    @ 31. IVAN3MAN AT LARGE :

    Could there have been some mass transfer between the then Red giant Pup and Sirius A back when the originally B4 ( ūüėČ ) star was evolving towards white dwarf-dom? Could this have effected the “Dogstar” (Sirius A) and could we learn anything of that somehow? Does anyone know?

    Also the “Little Dogstar” Procyon is coincidentally quite a similiar system & situation only with stars that are lower in spectral type (F5 V-IV & white dwarf) & thus slightly dimmer and slightly further away. (11 ly vs 9.) The two Procyonese stars are closer together (15 AU vs 19 AU) and thus harder to split too. See Kaler’s Stars site’s Procyon page :


    Wonder if this finding means that these two (very remotely concievably) potentially dangerous supernova candidates

    HR 8210 : http://jumk.de/astronomie/special-stars/hr-8210.shtml

    & T Pyxidis : http://blogs.discovermagazine.com/badastronomy/2010/01/07/no-a-nearby-supernova-wont-wipe-us-out/

    are safer & less dangerous than thought then? :-)


    Messier Tidy Upper:

    Could there have been some mass transfer between the then Red giant Pup and Sirius A back when the originally B4 star was evolving towards white dwarf-dom?

    As I’ve already mentioned, at the end of the second paragraph of my post (#31) above, Sirius B — while it passed through the red giant stage — may have enriched the metallicity of its companion.

    Could this have effected the ‚ÄúDogstar‚ÄĚ (Sirius A) and could we learn anything of that somehow?

    According to the Wikipedia (Sirius) article, the spectrum of Sirius A shows deep metallic lines, indicating an enhancement in elements heavier than helium, such as iron — 316% greater proportion of iron to hydrogen in its atmosphere than that in the Sun’s atmosphere.

  34. Messier Tidy Upper

    @ ^ IVAN3MAN AT LARGE : Ah yes. Thanks. That’s cool – first I’d heard of that & yes, sorry, I missed seeing that in your earlier comment. :-)


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


See More

Collapse bottom bar