A recent paper by astrophysicists Narciso Benítez and Jesús Maíz-Apellániz at Johns Hopkins University, and biologist Matilde Cañelles at the National Institutes of Health presents evidence that the Scorpius–Centaurus association of hot young stars within just a few hundred light-years of us has produced about 20 supernova explosions within the past 10 million years. 1 That’s far more than our local neighborhood’s fair share, considering how very rare supernovae are. Among the 1011 stars in a galaxy like the Milky Way, there are only a handful of supernovae per century.
But supernovae are not always isolated, random events. They often cluster in time and space, particularly in so-called OB associations like Sco–Cen. These are loose groupings of typically a few hundred O and B stars. (O and B are spectral classifications for the two classes of the most massive, and therefore hottest and shortest-lived, stars.)
What lends the new paper more than passing interest for non-astronomers is its conclusion that parts of Sco–Cen wandered so close to us a few million years ago that a number of its supernovae may have left noticeable physical and biological traces on Earth. Benítez and company extrapolated the positions of Sco–Cen’s present population of about 150 stars back in time as far as 10 million years and estimated the rate at which the association was producing supernovae during that period. Then they made the case that material ejected from about a dozen of its supernovae can account for the high levels of the unstable iron isotope 60Fe discovered three years ago 2 in two layers of ocean-floor crust gradually laid down during the past 6 million years.
More speculatively, the paper suggests that, when Sco–Cen was at its closest about 2 million years ago, one of these supernovae—perhaps only 120 light-years from Earth—may have damaged Earth’s ozone layer enough to cause the rather abrupt extinction of many bivalve species in tropical and temperate seas at the boundary between the Pliocene and Pleistocene epochs.
Hipparcos
Speculation about what a nearby supernova might do to the ozone layer and life on Earth goes back to a 1974 paper by Malvin Ruderman at Columbia University. But nowadays the speculators have access to a valuable resource that didn’t exist before 1997: the Hipparcos catalogs (see Physics Today, Physics Today 0031-9228 51 6 1998 38 https://doi.org/10.1063/1.882271 June 1998, page 38 ). The Hipparcos astrometry satellite, launched in 1989, has provided parallax distance measurements to hundreds of thousands of stars within 500 light-years of Earth. Furthermore, when the Hipparcos data are analyzed together with ground-based observations, they yield accurate determinations of the motion and spectral classification of many of these stars.
Distance is, of course, a key determinant of the damage a supernova might do. If one wants to calculate how close a star has come to Earth in the past 10 million years, one needs to know its present position and velocity. And the spectral classification of a star reveals its approximate mass, lifespan, and ultimate fate.
Sco–Cen, the nearest of all OB associations, is in fact a loose alliance of three subgroups, ranging in age from about 14 million to 5 or 6 million years. In any one subgroup, all the stars are assumed to have been born more or less simultaneously, presumably in the same molecular cloud. The paper of Benítez and company leans heavily on a 2001 paper by co-author Maíz-Apellániz. 3 From the present positions and velocities of the individual Sco–Cen stars, Maíz-Apellániz calculated their trajectories back to the time, some 10 million years ago, when supernovae would first have appeared in the oldest subgroup of Sco–Cen.
A star’s fate is determined primarily by its mass. The spectral classes O and B are subdivided into numbered subclasses by stellar mass and surface temperature. The O stars are the most massive and shortest lived. They all eventually explode as core-collapse supernovae, as do B stars heavier than about 9 solar masses. A 10-solar-mass star explodes after about 10 million years.
Availing himself of carefully vetted membership lists of the Sco–Cen subgroups derived from the Hipparcos catalogs by Tim de Zeeuw and coworkers at the Leiden Observatory, 4 Maíz-Appelániz was able to estimate the number of stars in each Sco–Cen subgroup that have already exploded, by comparing the present spectral distributions with an astrophysical model of the mass distribution with which an OB subgroup would have been born. He concluded that each subgroup, after a 4-million-year gestation period, produced roughly one supernova every million years.
The Local Bubble
The Solar System sits inside an interstellar cavity of hot, low-density gas, some 400 light-years wide, called the Local Bubble. The principal result of Maíz-Apellániz’s Sco–Cen paper was to strengthen the recent argument, by Randall Smith (Harvard University) and Don Cox (University of Wisconsin), 5 that the Local Bubble was excavated by a number of nearby supernovae over the past 10 million years.
Figure 1 shows the migration of the centers of the three Sco–Cen subgroups since they first began producing supernovae. 1 Our closest encounter, according to Maíz-Apellániz’s calculation, was with the subgroup Lower Centaurus Crux (LCC), about 2.5 million years ago. The mean radius of LCC is about 90 light-years. So when the center of LCC was barely 300 light-years from us, a star two standard deviations from its center could have exploded within 120 light-years of Earth.
Varying distance between the Sun and the centers of the three Sco–Cen subgroups of hot young stars, calculated back to the time when each subgroup began producing supernovae. The closest approach, by Lower Centaurus Crux, occurred about 2.5 million years ago.
Varying distance between the Sun and the centers of the three Sco–Cen subgroups of hot young stars, calculated back to the time when each subgroup began producing supernovae. The closest approach, by Lower Centaurus Crux, occurred about 2.5 million years ago.
Iron-60, with a half-life of 1.5 × 106 years, is expected in the ejecta of core-collapse supernovae. The strongest direct evidence for Maíz-Appelániz’s estimate of a steady diet of supernovae from Sco–Cen comes from the radiological analysis of several layers of ferromanganese crust on the floor of the South Pacific by Klaus Knie (Technical University of Munich) and coworkers in Germany. They interpreted the unusually high concentrations of 60Fe they found in a layer gradually deposited over 2 million years, starting 6 million years ago, to be evidence of a nearby supernova. 2 For the youngest layer, however, laid down over the past 2.8 million years, Knie’s group tentatively attributed the measured 60Fe excess to radioactive iron from the solar neighborhood.
Armed with Maíz-Apellániz’s calculation of Sco–Cen’s wanderings and its supernova output, Benítez and company come to a different conclusion for the youngest layer, which started forming when LCC was approaching its close encounter. Figure 2 compares the measured 60Fe abundances for that layer with their estimate of deposition from the ejecta of Sco–Cen supernovae over the last 2.8 million years. The calculated supernova contribution assumes that the three Sco–Cen subgroups have produced a total of about eight supernovae during that period, at a mean distance of 400 light-years. Looking back over the entire 10 million years during which Sco–Cen has been active, Benítez and company conclude that supernovae from the association have ejected enough 60Fe to account for the concentrations found by the German group in all the crustal layers it measured.
Average deposition rate of iron-60, measured (blue) in a layer of ocean-floor crust that began forming 2.8 million years ago, is compared with a calculation (red) of what one would expect from the ejecta of supernovae in the nearby Sco–Cen association of hot, young stars over that same period. The dashed line is the estimated background from nonsupernova sources.
Average deposition rate of iron-60, measured (blue) in a layer of ocean-floor crust that began forming 2.8 million years ago, is compared with a calculation (red) of what one would expect from the ejecta of supernovae in the nearby Sco–Cen association of hot, young stars over that same period. The dashed line is the estimated background from nonsupernova sources.
The ozone layer
To examine the intriguing possibility that a supernova might have caused the bivalve extinction 2 million years ago, Benítez and coauthors concentrated on LCC and considered the effect on Earth’s protective ozone layer of a supernova as close as 120 million light-years. Examining such scenarios in 1995, John Ellis (CERN) and David Schramm (University of Chicago) pointed out that charged cosmic rays are the only emissions from a supernova more than about 10 light-years away that could do serious damage to the biosphere. 6 An increased flux of cosmic rays in the upper atmosphere, they noted, would have sped up the production of nitrogen oxides that catalytically destroy ozone and thus could have endangered sea-surface plankton and those that feed on it. It was they who suggesting looking for 60Fe as evidence of nearby supernovae.
Benítez and company estimate that the rate of energy deposition in the upper atmosphere over a 10-year period by cosmic rays from a supernova 120 light-years away might be as much as half a milliwatt per square meter. That would be enough, they argue, to temporarily deplete the ozone layer in the tropics by about 20% and by as much as 60% at high latitudes. Ellis and Schramm’s estimate of the cosmic-ray flux from a supernova 30 light-years away was only twice as big as that of Benítez and company for a supernova four times as distant.
Why only twice as big and not 16 times? Whereas Ellis and Schramm had assumed that the charged cosmic rays are dispersed on the way here by a typically random interstellar magnetic field, the Benítez calculation invokes the much weaker and more coherent magnetic field one would expect for the Local Bubble, if it was indeed swept out by supernovae. Neil Gehrels (NASA’s Goddard Space Flight Center) tells us that a new model calculation, soon to be reported by his group, concludes that a supernova much farther away than 30 light-years is unlikely to have done much damage to the ozone layer. But like Ellis and Schramm, the Goddard calculation makes no special assumption about the local magnetic field.
A minor extinction
“The increased solar ultraviolet from our estimate of the damage to the ozone layer,” Cañelles told us, “would have caused, at most, a minor extinction,” nothing like the worldwide mass extinction in which the dinosaurs vanished. Because ultraviolet radiation doesn’t penetrate very deep into water, the primary damage would probably have been to photosynthesizing phytoplankton near the surface. A significant reduction of phytoplankton abundance, Cañelles and coathors argue, might well have worked its way up the food chain to the filter-feeding bivalves whose extinction in the Western Atlantic around the time of the Pliocene–Pleistocene boundary roughly 2 million years ago is well documented in the fossil record.
When this regional extinction was first documented in the 1980s, the most widely discussed explanation was the cooling associated with the onset of glaciation that marked the start of the Pleistocene. “But in recent years it has become rather clear that cooling doesn’t suffice to explain the decrease in planktonic productivity that appears to have been the proximate cause,” says Warren Allmon (Cornell University and the Paleontological Research Institution). “By contrast, the abrupt increase of ultraviolet in the supernova hypothesis is quite consistent with what we know about this regional extinction.”
In the supernova scenario, the greatest damage to the ozone layer occurs at high latitudes. Nonetheless, it’s the tropical species that would have suffered worst, because solar radiation is so much more direct in the tropics. To test their admittedly speculative extinction hypothesis, Benítez and coauthors look forward to a more finely time-resolved examination of crustal 60Fe soon to be reported by Knie’s group, in hopes of pinning down the times and distances of individual nearby supernovae.
