Where do the Solar System's heavy elements come from? We know that many of the elements heavier than iron must have been formed in supernovae; but exactly how? Taken together, three new papers (one on page 261 of this issue1, and two to appear in the Astrophysical Journal2,3) imply that two different types of supernova are responsible for elements in different mass regimes.
It is now some four decades since Suess and Urey4, largely working from analyses of meteorites, assembled a table of abundances of the elements that enabled nuclear physicists to identify the principal processes in stellar interiors that had produced those elements. The actual nuclear physics of those processes became reasonably well understood within the following few years. However, it has taken considerably longer for us to understand the astrophysical environments within stellar interiors in which many of those processes take place. The process whose environment has proved most elusive has been the r-process, in which neutron capture takes place on a very rapid timescale.
In the evolution of a star considerably more massive than the Sun, nuclear fusion reactions build toward products in which the binding energy per nucleon becomes maximized. This produces an abundance peak at 56Fe. Making nuclides much heavier than this involves neutron capture, which generally produces nuclides on the neutron-rich side of the region of stable elements, known as the valley of beta stability. Only in the violence of a supernova explosion can the nuclides on the neutron-deficient side be produced, primarily by losing nucleons in photodisintegration. It is possible to distinguish these processes by examining the abundances of nuclei in the Solar System, particularly of stable isobars, in which more than one stable nucleus exists at a given mass number.
There are three basic nuclear processes that produce heavy elements: the s-, r- and p-processes, standing for slow, rapid and proton-rich. The abundances produced by each of these5 are shown in Fig. 1.
The s-process probably takes place in red giant stars that are burning helium, with a given nucleus capturing a neutron perhaps every few thousand years. But only fractions of a second separate successive photodisintegrations in the p-process or neutron captures in the r-process, implying that these processes must occur in supernova explosions. So somedetails of supernova explosions can be deduced from isotopic analysis of meteoritic material — in particular, interstellar grains that have undergone relatively little thermal processing during and since the formation of the Solar System.
One important recent discovery was that there are two different r-processes6: one responsible for the r-process nuclides up to about A = 140 (where A is the mass number), the other for nuclides above A = 140. The discovery stemmed from extinct radionuclides, which lived long enough to survive with measurable abundances from the time of their production to their injection into the forming Solar System, but not long enough to be measurably present now. They are detected through excesses of the daughter elements that result from their decay.
The abundance of 182Hf in the early Solar System is consistent6 with the continuous production of the actinides in the Galaxy, with mixing to maintain a roughly constant abundance level on a timescale consistent with the mean life of 182Hf — about 107 years. But trouble arises from two lighter radionuclides, 107Pd and 129I. If these were made by the same r-process that produced the actinides, then they would be some two orders of magnitude more abundant than they are. So there must have been an interval of ~108 years between the last r-process production of 107Pd and 129I and their injection into the forming Solar System.
This is the basis for the claim that there are two distinct r-processes, with yields confined above and below about A = 140. The high-mass r-process must operate more frequently; a typical point in the interstellar medium receives products from it at roughly 107-year intervals, but the low-mass-number r-process is much rarer, giving that point a contribution only every 108 years or so.
The newest results, which both confirm this idea and suggest how the two processes arise, come from interstellar diamond grains extracted from meteorites, and the anomalous isotopic abundances of their xenon and tellurium. These diamonds are tiny — only about one in a million contains a xenon atom, so the measurements must be made on bulk samples.
Figure 2 shows the overabundances in the principal xenon component contained in the diamonds (called XeHL). This pattern persists over a wide range of conditions of gas extraction from different samples at different temperatures7, indicating a common astrophysical source. Most striking are the high abundances of mass numbers 134 and 136 (r-process products), and 124 and 126 (p-process products). It has been suggested8 that this XeHL pattern is produced in a supernova environment by some kind of separation between those xenon isotopes with shorter-lived precursors and those with longer-lived precursors. But a new paper contradicts this idea.
Instead of xenon, Richter et al.1 look at the tellurium in interstellar diamonds. It has large overabundances only at mass numbers 128 and 130, both r-process isotopes whose precursors live about an hour. The tiny overabundance (<4 × 10−5) for the A = 120 p-process product, for example, is in striking contrast to the xenon p-process products shown in Fig. 2. The time delay in forming A = 120 is also only about an hour, and the supernova p-process should9 have a yield at A = 120 comparable to those at A = 124 and 126 for xenon. So a separation based on precursor lifetimes can't explain the tellurium data. What can? To address this, we will have to look more closely at the extreme physical conditions occurring in supernovae.
The r-process is thought to involve neutrino-driven winds in the surface layers of a newly formed neutron star. A neutron star must be formed during the collapse process that produces a supernova, to supply the observed amount of energy. Enough energy is released so that the thermal energy, E = kT(where k is Boltzmann's constant, and the temperature T is usually measured in MeV) is large compared with the rest-mass energy of the electron, and enough to fill the available phase space of quantum states with neutrinos. Inside the neutron star, the six flavours of neutrino — electron, muon and tau, and their corresponding antiparticles — interchange rapidly among themselves and with the radiation field.
The neutrinos leak out of the neutron star in seconds. Even when they are near the surface layers of the neutron star, however, the neutrinos continue to interchange with one another and with electrons, positrons and photons. This heats the surface matter, blowing it off as a dense, neutron-rich wind. As it expands and cools, nuclear reactions take place that build up seed nuclei, and the r-process takes place on these seed nuclei as they cool further.
But this general description of the r-process can be applied to two very different cases involving high and low entropy — corresponding to high and low thermal energy.
In the hot, high-entropy case, the core of a massive star collapses to form a neutron star (and possibly further to form a black hole). For several years this has been the standard setting for the r-process, but simulations show that to match Solar System r-process abundances demands very high entropies indeed, corresponding to entropies of at least 400 (in units of Boltzmann's constant per baryon). Theory balks at this: it appears to be impossible10,11 to reach an entropy as high as 200. Perhaps the theorists should concentrate less on increasing the numbers of neutrons and look at reducing the numbers of seed nuclei, so that the existing neutrons can carry the r-process build-up to higher mass number.
Nevertheless, neutron stars with massive stellar precursors seem essential to understand both the r-process and the p-process that produced the XeHL in Fig. 2. The p-process is thought to operate mainly in the oxygen-neon envelope of such a massive star9. Can the high-entropy picture then be adapted to provide two distinct r-processes? One suggestion2 is that the formation of a neutron star as an end point might generate enough neutrons to drive the r-process up into the actinide region, whereas if the endpoint is instead a black hole, then the neutrons would be prematurely cut off and only the low-mass-number range of products would be produced. Although this is an attractive idea, it doesn't predict the tellurium results, in which there is no trace of a p-process yield. As the supernova shock would be launched before the black hole ended neutron production, the p-process products should be there.
But a low-entropy r-process has now been discovered3 that could give products primarily in the lower range of mass numbers. Many stars of 6-12 solar masses exist in binary systems. At some point, these stars expand, transfer most of their hydrogen to their companions, and become white dwarfs. Eventually they accrete material back from their companions, and some of them grow in mass until the Chandrasekhar limit is reached, at which point the star collapses to form a neutron star. This is accretion-induced collapse (AIC), and results in so-called type-1a supernovae.
The entropy per particle in AIC is only about 3-5. As an entropy of at least 8 would be needed to completely break up the nuclei into nucleons, the star must retain very large numbers of seed nuclei. So the ratio of neutrons to seed nuclei would be relatively small, and this would favour a final distribution of r-process products in the lower range of mass numbers. Further, the lack of a massive envelope means that AIC events do not lead to a p-process, and so could fit the tellurium results.
How were these isotopes implanted in the interstellar diamonds? Industrial diamonds are often produced by chemical vapour deposition in a cooling flow; and presumably something similar happens when a supernova shock wave encounters a carbonaceous layer. In a massive supernova, such a layer would be in the outer envelope of the star, whereas in an AIC supernova the carbon is likely to be in the material from the companion star flowing onto the white dwarf in the binary pair. But this requires further exploration.
Nevertheless, a consistent picture has emerged in which rare, low-entropy AIC supernovae inject the lighter r-process isotopes, and more common core-collapse supernovae take over for isotopes heavier than about A = 134.
Richter, S., Ott, U. & Begemann, F. Nature 391, 261–263 (1998).
Qian, Y.-Z., Vogel, P. & Wasserburg, G. J. Astrophys. J. (in the press).
Baron, E., Cowan, J. J., Rogers, T. & Gutierrez, K. Astrophys. J. (submitted).
Suess, H. E. & Urey, H. C. Rev. Mod. Phys. 28, 53–74 (1956).
Cameron, A. G. W. in Essays in Nuclear Astrophysics (eds Barnes, C. A., Clayton, D. D. & Schramm, D. N.) 23-43 (Cambridge Univ. Press, 1982).
Wasserburg, G. J., Busso, M. & Gallino, R. Astrophys. J. 466, L109-L113 (1996).
Huss, G. R. & Lewis, R. S. Meteoritics 29, 791–810 (1994).
Ott, U. Astrophys. J. 463, 344–348 (1996).
Rayet, M.et al. Astron. Astrophys. 298, 517–527 (1995).
Qian, Y.-Z. & Woosley, S. E. Astrophys. J. 471, 331–351 (1996).
Hoffman, R. D., Woosley, S. E. & Qian, Y.-Z. Astrophys. J. 482, 951–962 (1997).
About this article
Chemical Geology (2000)
Science Progress (1999)
Publications of the Astronomical Society of the Pacific (1999)