Journal Club – When White Dwarfs Collide


According to Wikipedia, a journal club is a group of individuals who meet regularly to critically evaluate recent articles in scientific literature. Being Universe Today if we occasionally stray into critically evaluating each other’s critical evaluations, that’s OK too. And of course, the first rule of Journal Club is… don’t talk about Journal Club.

So, without further ado – today’s scheduled-for-demolition journal article is about the ongoing problem of figuring out what events precede a Type 1a supernova.

Today’s article:
Dan et al How the merger of two white dwarfs depends on their mass ratio: orbital stability and detonations at contact.

There is growing interest about the nature of the events that precede Type 1a supernovae. We are confident that the progenitor stars of Type 1a supernovae are white dwarfs – but these stars have generally very long lives, making it difficult to identify stars that are potentially on the brink of exploding.

We are also confident that something happens to cause a white dwarf to accumulate extra mass until it reached its Chandrasekhar limit (around 1.4 solar masses, depending on the star’s spin).

For a long time, it had been assumed that a Type 1a supernova probably arose from a binary star system with a white dwarf and another star that had just evolved into a red giant, its outer layers swelling out into the gravitational influence of the white dwarf star, This new material was accreted onto the white dwarf until it hit its Chandrasekhar limit – and then kabloowie.

However, the white-dwarf-red-giant-binary hypothesis is currently falling out of favour. It has always had the problem that any Type 1 supernovae has, by definition, almost no hydrogen absorption lines in its light spectrum – which makes sense for a Type 1a supernovae arising from a hydrogen-expended white dwarf – but then what happened to the new material supposedly donated by a red giant partner (which should have been mostly hydrogen)?

Also, the recently discovered Type 1a SN2011fe was observed just as its explosion was commencing, allowing constraints to be placed on the nature of its progenitor system. Apparently there is no way the system could have included something as big as a red giant and so the next most likely cause is the merging (or collision) of two white dwarfs.

Other modelling research has also concluded that the two white dwarf merger scenario maybe statistically more likely to take place than the red giant accretion scenario – since the latter requires a lot of Goldilocks parameters (where everything has to be just right for a Type 1a to eventuate).

This latest paper expands the possible scenarios under which a two white dwarf merger could produce a Type 1a supernovae – and finds a surprising number of variations with respect to mass, chemistry and the orbital proximities of each star. Of course, it is just modelling but it does challenge the current assertion at the relevant Wikipedia entry that white dwarf mergers are a second possible, but much less likely, mechanism for Type 1a supernovae formation.

So – comments? Anyone want to defend the old red-giant-white-dwarf scenario? Does computer modelling count as a form of evidence? Want to suggest an article for the next edition of Journal Club?

30 Replies to “Journal Club – When White Dwarfs Collide”

  1. I am only vaguely interested in the physics, as it effects cosmology a bit. (But not too much, IIRC UT articles on the topic.)*

    But the science background interests. Yes, I would say that anything predictive, testable, is evidence, up to and including ad hocs that only describes the results without a theoretical connection. If nothing else it can be used, as here, to put likelihoods against each other.

    However, this is prone to error the more the model is removed from theory.

    What is done here is relatively sound though, is my personal guess. Stars like to form binaries, and my impression (which could easily be wrong) is that these systems are weighted towards equal massed stars. Those could evolve to white dwarf binaries.

    And likewise it was my impression (again not literature based) that nearby stars, especially evolving and so mass changing stars, collide frequently enough.

    If so, there is some loose connection with the basic theory. And if it is indeed a scenario based model, it gets another connection as it attempts to exhaust the possibilities that a theory would encompass.
    * My suggestion for journal club post is a late/latest review on this, because it would fit nicely.

    1. My sketchy understanding of history suggests that Aristotelian geocentric models of the solar system (with epicycles etc) actually fit early observational data better than early heliocentric models – until Kepler showed planetary orbits were elliptical rather than circular. An exemplar of the complexities of working with models I suppose.

      Yes – these authors didn’t seem to think this new approach was a major problem for SN1ae as standard candles (though they didn’t really explain why – and nope I haven’t come across a good review on all that. Suggestions?)

      1. Looking at historical case studies would fudge the issue. They used other ideas of science for one. And in this case there were many outside concerns and iteration.

        But it is a fair example in a modern setting, what a basic model without looking at what gravity or the helio- vs geocentric choice entails.

        In modeling, deterministic or statistic, using more degrees of freedom will make for a better fit. Those models are most times rejected based on parsimony. So Keplerian types of model wins.

        [As an aside I can’t say anything on the helio- vs geocentric issue of data fit, since I don’t know anything about it. You often hear this claim, but it is never referenced.

        Scouring Wikipedia it is apparent that there were never a comparison between similar models. At best Ptolemy’s ‘equant’ epicycles were compared with Copernicus’ ‘true’ epicycles.

        As I said, is is better to stay away from historical data digging. Especially here, this has all the signs of a folk tale.]

        No suggestions on the review, unfortunately.

  2. From the perspective of cosmology this is actually unwelcome. A single white dwarf star that siphons off mass from a companion star reaches a precise tipping point for the explosion. A merger of white dwarfs would similarly explode as the coalesced mass is beyond the Chandrasekhar limit, but could be above it by a significant percentage of a solar mass. This constitutes “noise in the detector” which needs to be filtered out.


    1. There have been reported super-Chandrasekhar mass Type1as – e.g. – the key indicator for it having more than 1.4 solar masses being Nickel-56 apparently.

      A fast spinning WD should also has a higher Chandrasekhar mass limit than a slow spinning one.

      On the other hand, modelling suggests that too much mass gives a WD star too much binding energy to allow a supernova to take place (i.e. it contains the explosion and just swells in size). So there is still only a small window whereby a SN1a is possible via collision or mass accretion.

      So the variability is a bit vexing – but I don’t think it’s enough to seriously challenge the current cosmological model.

      1. Thanks for the reminder! Supernova modeling isn’t a cut & dry issue [room for those errors] and is to my knowledge not a settled area. But at least there is hope for the standard candles in this.

      2. The Chandrasekhar limit is the first order model which has spherical symmetry with an S-mode. It was always obvious that angular momentum would influence things, and the collapse is now dipolar or has a dominant P-mode. I am not sure how much this perturbs the standard S-limit. The coalescence of two WDs will have higher order modes and could involve a net mass considerably above 1.4M_{sol}. Maybe if the mass is much larger than 1.4M_{sol} the energy of the C-O fusion to Ni-56 goes into the production of e-e^+ pairs and the energy is consumed in generating more degrees of freedom. The energy might then be more slowly released.


      3. To be honest, i cant see how a merger scenario can result in anything but a superrotating primary with an effective chandrasekhar limit significantly above the 1.4M limit.

        Given time and an accretion disk it can shed through magnetic interaction some angular momentum and initiate the collapse, but then the collapse would immediately be countered by a spinup. That could indicate that the ignition happens while the WD is semi-stable by rotation.

  3. The secret in all of this is the number of white dwarfs very near the Chandresaker limit of 1.41 to 1.45 solar masses. Progenitors of SNI must be able to trip this limit to cause the collapse in the first place, My understanding is that white dwarf merging easily accounts for exceeding the limit, however, the issue is getting two such stars close enough (losing their gravitational energy) is the bigger problem to solve.
    I do think the red giant/ white dwarf is still viable, though the range of tipping point is much narrower. I.e. A couple of hundredths of solar masses. Mass transfer of material from one ordinary star to a white dwarf is also possible, but it too assumes the white dwarf is on the tipping point.

    Between all the scenarios there must be various sub-catergories of SNI progenitors.

    Another point not usually mentioned is the changes occurring in white dwarfs as they very slowly cool down. No doubt there is still some thermal energy radiating from the core that is decreases the density slightly, which prevents the collapse. Here the tipping point is where the cooling star minutely compresses enough to collapse Other factors might be the composition of the white dwarf atmosphere and inner stellar core. One idea that goes back to the 1970s is that violent starquakes make be enough to form an energy cascade throughout the white dwarf, where the limit is just overwhelmed, starting the point of no return collapse. This races through the white dwarf in an instant forming SN Type 1. (Here you don’t need any companion star.)

    Another simple possibility I have only very rarely heard of is a collision of a sizeable planetary body with a white dwarf near its tipping point. Collapse could come by increasing the mass just enough to cause the collapse (or by the subsequent violent release energy that then the collapse.)

    I cannot think we are missing something important in our knowledge here. Until we know something more of the structures of white dwarfs and its exotic state of matter, it will continue to elude us.

    1. No doubt there is still some thermal energy radiating from the core that is decreases the density slightly, which prevents the collapse

      Degenerate pressure do not depend on temperature, only on density (and on the properties of the particles causing the degenerate pressure, in this case electrons).

      A cool white dwarf have the same density as a hot one, up to the temperature where thermal gas pressure or radiation pressure again becomes dominant. So cooling off in this respect will not make the white dwarf shrink.

      1. Saying; “A cool white dwarf have the same density as a hot one,…”

        Sorry absolutely not. (What a totally crazy notion! Are you kidding me with a subtle little joke here?)

        Let see. Ffor example, Table 1 pg. 695 in the Chabrier 1993 paper “Quantum effects in dense Coulumbic matter – Application to the cooling of white dwarfs”

        If this were so, then how do you explain that white dwarfs are assumed to have a ‘crust’ and crystallisation when the temperature drops below a certain temperature?

        In fact, it is very simply shown that the opacity of the material of white dwarf material (negating cystallised white dwarfs) follows the very basic Kramer’s law; κ = κ0 . ρ . T^-3.5

        Where κ is the opacity, ρ is the density, and T is the temperature.

        From this hydrostatic and radiative equilibrium to be maintained, this density must change depending on the depth inside the white dwarf, (It also changes and is dependant on the composition of the material against depth. I.e. κ(ρ,T)

        Were this NOT true, we would be unable to calculate the cooling times of white dwarfs (and even neutron stars.)

        Also Ni-56 is mostly found in SNII not SN I. There are no remnants in SNI’s!!

        [I think you confusing all this with the theory of neutron stars!]

      2. Yes, the surface of the white dwarf is not degenerate, you are correct about that. But overall the WD is in a degenerate state, and the surface temperature has exceedingly small overall effect. Unless the surface is millions of degrees hot, and i think you can agree the surface cools below that very quickly.

        Also Ni-56 is mostly found in SNII not SN I. There are no remnants in SNI’s!!

        The lightcurve of SN Ia is mainly powered by the decay of large amounts of Ni-56, you can even see references to that within the comments of this article. Or as it says on wikipedia (for easy reference)

        Type Ia supernovae follow a characteristic light curve—the graph of luminosity as a function of time—after the explosion. This luminosity is generated by the radioactive decay of nickel-56 through cobalt-56 to iron-56.[

        I didnt say there is a remnant in normal SN-Ia (there isnt), but there is a theoretical class of supernova where the collapsing WD contains a oxygen-neon-magnesium core. And this type do, if they happen, create a neutron star in the explosion. They would be rare, if they even exist, but would technically be a white dwarf collapsing.

        You may need to check your references.

      3. “Yes, the surface of the white dwarf is not degenerate, you are correct about that. But overall the WD is in a degenerate state, and the surface temperature has exceedingly small overall effect. Unless the surface is millions of degrees hot, and i think you can agree the surface cools below that very quickly.”

        Good. Excellent. You confirm my main point. You agree it shows that your statement; “A cool white dwarf have the same density as a hot one,…” is quite wrong!

        My point is that densities vary enormous from the smallest to the largest white dwarfs.

      4. Good. Excellent. You confirm my main point. You agree it shows that your statement; “A cool white dwarf have the same density as a hot one,…” is quite wrong

        A thin shell on the surface have very little effect when some 99% of the total mass is in a degenerate state. No, i am not wrong.

        And are you still claiming SNIa does not produce large amounts of Ni56?

        The reactive pressure in degenerate matter are not affected by the temperature, only a function of density.

        The large range of densities in white dwarfs are because they are degenerate, not because of temperature. When you increase the mass-load on degenerate matter, it shrinks, which cause a large increase in density. The heaviest white dwarfs are the one that are smallest in size, and temperature is not a factor.

      5. Fermions occupy states differently than bosons. The occupation number of fermions in the ith state with energy E_i is given by Fermi-Dirac (FD) statistics

        = 1/[e^{i(E_i – ?)/kT} + 1]

        for ? the chemical potential. The chemical potential at zero temperature is the maximum energy the fermion can have, which defines a Fermi surface. For the temperature T such that kT = ?/x, this occupation function in the limit x — > ? with constant ? approaches a step function for = 1 with e_i/? < 1 and = 0 with e_i/? > 1. For x small, this function becomes more Boltzmann-like appearing as an exponential decaying function of e_i/?.

        There are of course complications with white dwarf stars, but generally after collapse the temperature of the WD is extreme, around a trillion K. Of course the pressures are enormous as well so there is a large ?, but the statistics deviate from the step function. This will particularly be the case for matter near the surface, and WDs have a sort of atmosphere as well. As the WD cools, it becomes less white and a cold WD becomes a black dwarf. The electrons all line up in their proper FD states which are unique and the equation of state approaches a condition where the occupation number for the ith state fits in the step function.


      6. Yes ofc, once the collapse have been initiated the temperatures will skyrocket.

        But i was referring to the stable conditions. A 1Ma old WD would have settled into a stable degenerate state, with the internal pressure dominated by degeneracy. And even though the surface is hot by comparison to most normal stars, the WD would be cold compared to any situation where the degeneracy could be lifted.

      7. Wait wait.

        A white dwarf cannot be a trillion K hot!

        At a trillion K all nuclei are quickly broken down by disociation caused by gamma rays (this actually happens around 6 billion K, and this process consumes energy being the initiator of a collapse into a neutron star). A trillion K hot might be applicable for a freshly created neutron star, but not to a WD.

        Do you mean Billion K hot? During the actual SN yes, should even be hotter than that since Ni56 is produced.

      8. Yes of course, it is a billion K, not a trillion. A trillion K would convert nuclei into a proton-neutron gas.


  4. Won’t orbital dynamics provide some evidence of binary mergers, after all binary systems do not just smash head on (although that must happen sometimes), they dance around each other as their orbits decay. I’ve not worked on dual-body tidal effects but I would assume that the pull of the other body would counter the star’s own internal gravitic pressure which would mean that the volume nearest the other star would be drawn in a cone of matter that graduates from degenerate matter near the core to non-degenerate matter of lessening density out to the lagrange point and then back to the other star.

    It’s a fair assumption that each star is rotating so would a volume of the star be dynamically fluctuating between degenerate matter and non-degenarate matter and moving towards and back from the lagrange point as it rotates? Surely this dynamic would give rise to some unique observable dynamic phenomenon. I’m not knowledgable enough to work out what it might be but maybe it would show up in the low level the X-Ray emissions or in the gravitic red-shift of the light?

    Also what would happen inside this cone of non-degenerate matter, if Carbon Oxygen fusion reiginted then each star would gets progressively denser as the fused material falls back on rotation which would cause an exponential decay of their orbits which perhaps is why it all happens so suddenly.

    1. It’s a fair assumption that each star is rotating so would a volume of the star be dynamically fluctuating between degenerate matter and non-degenarate matter and moving towards and back from the lagrange point as it rotates?

      This scenario will produce a very strong tidal lock, both stars will ‘face’ the other with the same side.

      Other than that, yes the interaction becomes very complex.

      1. Don’t they have a lot of rotational energy, though? That is the energy source of a pulsar, right? Developing that tidal lock must mean shedding a lot of energy, fairly quickly. Depending on the timescale, that could have observable effects.

        Or maybe I’m confused and the magnetic field of a white dwarf actually damps the rotation down to nothing, so tides only have to spin it back up so rotation and revolution periods match.

      2. You are correct, the rotational energy must be shed.

        But the timeframe for the white dwarfs to approach for the merger is many millions, sometimes even billions, of years. And the tidal strenght here is many-many orders of magnitude higher than it is within the Earth-Moon system, or any other tidally locked system in the solarsystem. Imagine the tides if the Moon was orbiting only a few Earth-radi or less, and each of the earth and the moon was weighing like the sun.

        That energy is converted to heat by the friction within each respective body, and to some extent also transferred into orbital changes which eventually are lost anyway due to atmospheric braking and gravitational radiation.

        I am sure magnetic fields also have something to say in this, but without specific examples on strengths its hard to determine to what degree, and should be simulated in detail for a better view. A few millions or even billions of years of close interaction surely will have a large effect.

        What happens very close before the actual merger is very complex and requires simulations to represent properly. But up to the point of actual deteriation of one or both of the white dwarfs into a stream of matter, the tidal lock would be extremely powerfull, and keep both facing eachother with the same side, libration effects present.

        It should also be noted that at the time when one of the white dwarfs starts to disrupt, the loss of mass over to the companion will make the less massive white dwarf _expand_. This is the nature of degenerate matter, the less massive it becomes, the larger the physical size, the direct opposite of matter under more normal pressures. The more mass removed, the faster and easier it become to remove more mass.

    2. All good suggestions – but white dwarf binaries are expected to take an awful long time to evolve to the brink of SN1a-ness. Hence finding a candidate system from which you could pull all these pre-explosion orbital dynamics data is not currently possible.

      Not sure about a WD “dynamically fluctuating between degenerate matter and non-degenarate matter and moving towards and back from the lagrange point as it rotates.”

      Isn’t this just the old thought experiment of what would happen if the Sun suddenly became a black hole possesing the same mass – but several orders of magnitude more density. And the answer is that we would all die from lack of sunlight, but the Earth’s orbit would not change significantly.

  5. We have been offered a choice of a white dwarf companion or a red giant. Are these all the choices? The white dwarf may be sitting in a planetary nebula. The pretty spherical planetary nebulas will probably never come back to the parent star, but some of the more disturbed shapes may leak material back again, or have it pushed back by neighbouring stars.

    If something is feeding the white dwarf extra matter, then it will probably have an accretion disk. For most viewing directions, the disc will not obscure the star and supernova directly. The paper showed that a merger between two white dwarfs where on was twice the size of the other still gave a structure with a recognisable accretion disk even if it only survived for an orbit or so: you had to get a very close match in the star sizes to get the symmetric pattern with a central blob and two outliers. I suspect this may be rather rare.

    Would we see absorption lines if the white dwarf was absorbing a slow trickle of hydrogen or carbon? Would a small supply of new, light materials get absorbed in the bulk, or would it always spread along the chromosphere and give a spectral signature?

    1. I think it is because the spectral signature can resolve white dwarfs capturing comets and other dust, that non-star material is discounted.

  6. Hmm, a standard candle relying on random sized WD crashing into each other which I would think would initiate random sized explosions therefore not standard candle. Also, the age question comes into play here. How likely are binary white dwarves to orbitally decay to the point where they no longer stay apart? Isn’t that a multi billion year scenario? I’m glad the binary WD/RG scenario is losing credibility as I never liked it but I think some of the suggestions here of odd happenings within WD’s as they age or compress is probably the more likely answer.

    1. White dwarf stars have a bit of atmosphere. Two dwarf stars in a mutual orbit may, if they are close enough, exchange their atmospheres a bit. White dwarfs are extraordinarily hot, so their atmospheres are boiling off as the surface evaporates a bit to replace it. This means the two dwarf stars are orbiting in a medium which drags their orbits. There is a general relativistic contribution in the production of gravity waves, but this is likely a far weaker process than drag.

      Neutron stars are much more compact and do not have quite the extensive atmosphere of a white dwarf. In this case the orbit of two neutron stars will decay by the emission of gravity waves and not drag.


  7. If white-dwarf’s require a sudden increase of energy, in order to go super nova, it might just be from a more distant source, or sources, than generally considered. Black holes are not fully understood, but it is possible that in our fractured and expanding universe that some of their energy could be transfered through folded space to re-emerge at points of attraction. If the cosmic influence of white-dwarfs cause them to be “magnets” for this type of energy transfer, then the sudden overload and its result would go a long way toward explaining the phenomena and maintaining universal equilibrium.

  8. I have heard very little on the subject, so I do not have a certain point of view on this point, but if to put it as a simple question whether computer modelling counts as evidence or not, the answer is evident – absolutely not. And I want to thank you for the article you gave the link to – it was rather curious to read.

    pdf editor

Comments are closed.