Research Article

Observation of the 60Fe nucleosynthesis-clock isotope in galactic cosmic rays

See allHide authors and affiliations

Science  06 May 2016:
Vol. 352, Issue 6286, pp. 677-680
DOI: 10.1126/science.aad6004

Cosmic rays from a nearby supernova

Supernova explosions produce unstable isotopes, spreading them through space in the form of cosmic rays. Binns et al. used NASA's Advanced Composition Explorer spacecraft to search for previously undetected traces of 60Fe in cosmic rays passing through the solar system. Seventeen years of observations detected just 15 60Fe nuclei—a small but statistically significant number. Because 60Fe is radioactive, with a half-life of 2.6 million years, these nuclei must have formed relatively recently in a nearby supernova. The most likely candidates are massive stars in the Scorpius-Centaurus association.

Science, this issue p. 677


Iron-60 (60Fe) is a radioactive isotope in cosmic rays that serves as a clock to infer an upper limit on the time between nucleosynthesis and acceleration. We have used the ACE-CRIS instrument to collect 3.55 × 105 iron nuclei, with energies ~195 to ~500 mega–electron volts per nucleon, of which we identify 15 60Fe nuclei. The 60Fe/56Fe source ratio is (7.5 ± 2.9) × 10−5. The detection of supernova-produced 60Fe in cosmic rays implies that the time required for acceleration and transport to Earth does not greatly exceed the 60Fe half-life of 2.6 million years and that the 60Fe source distance does not greatly exceed the distance cosmic rays can diffuse over this time, ⪍1 kiloparsec. A natural place for 60Fe origin is in nearby clusters of massive stars.

Signature of recent nucleosynthesis

The radioactive isotope 60Fe [which decays by β decay with a half-life of 2.62 × 106 years (1)] is expected to be synthesized and ejected into space by supernovae, and thus could be present in galactic cosmic rays (GCRs) near Earth, depending upon the time elapsed since nucleosynthesis and the distance of the supernovae. 60Fe is believed to be produced primarily in core-collapse supernovae of massive stars with mass M > ~10 solar masses (M), which occur mostly in associations of massive stars (OB associations). It is the only primary radioactive isotope with atomic number Z ≤ 30 [with the exception of 59Ni, for which only an upper limit is available (2)] produced with a half-life long enough to potentially survive the time interval between nucleosynthesis and detection at Earth. (Primary cosmic rays are those that are synthesized at the GCR source, as opposed to secondary cosmic rays, which are produced by nuclear interactions in the interstellar medium.) 60Fe is difficult to measure with present-day instruments because of its expected extreme rarity, based on nucleosynthesis calculations for supernovae (3, 4). The detection of 60Fe in cosmic rays would be a clear sign of recent, nearby nucleosynthesis. The long period of data collection (17 years) achieved by the Cosmic Ray Isotope Spectrometer (CRIS) aboard NASA’s Advanced Composition Explorer (ACE) (5), the excellent mass and charge resolution of the CRIS instrument, and its capability for background rejection have enabled us to detect 60Fe.

60Fe has been detected in other samples of matter. Measurements of diffuse γ-rays from the interstellar medium (ISM) by the spectrometer on the International Gamma-Ray Astrophysics Laboratory (INTEGRAL) spacecraft have revealed line emission at 1173 and 1333 keV from 60Co, the daughter product of 60Fe decay, clear evidence that “nucleosynthesis is ongoing in the galaxy” (6). As expected, this emission is diffuse instead of point-like, since the 60Fe lifetime is sufficiently long to allow it to diffuse over distances that are large compared to the size of a supernova remnant. This is one of many strong connections between γ-ray astronomy and direct cosmic-ray studies (7).

Deep-sea manganese crusts from two different locations have also been found to harbor elevated 60Fe levels (8, 9). Analysis of crust layers using accelerator mass spectrometry showed significant increases in the 60Fe/Fe ratio 2.8 million years (My) ago, “compatible with deposition of supernova ejecta at a distance of a few tens of pc” (9). The measurement was verified by an independent analysis (10), although these investigators did not find a corresponding increase in a marine-sediment sample [see (10, 11) for discussion]. We note that the manganese crust studies (810) used outdated half-lives for 60Fe and 10Be—1.49 and 1.51 My , respectively—instead of the currently accepted 2.62 and 1.387 My (1). Using these recent lifetimes, it has been estimated that the peak in the 60Fe/Fe ratio (9) as a function of depth corresponds to an age of 2.2 My (11). Lunar surface samples also show elevated 60Fe/Fe ratios consistent with supernova debris arriving on the Moon ~2 My ago (12, 13). These deep-sea manganese crust and lunar surface observations were compared with expectations from possible stellar sources (11) and found to be consistent with an origin in core-collapse supernovae, but inconsistent with Type Ia supernovae, which produce orders of magnitude less 60Fe.

Cosmic-ray 60Fe detection

The CRIS instrument was launched on ACE in 1997 and has operated continuously since that time except for intense solar-active periods, lasting for a few days each, when large fluxes of solar energetic particles exceeded the CRIS trigger rate capability. The instrument (Fig. 1A) uses a scintillating fiber hodoscope to determine particle trajectory and four stacks of silicon solid-state detectors to measure the energy loss (ΔE) and the total energy of cosmic-ray nuclei stopping in a detector stack. These measurements are used to identify particle charge, mass, and energy per nucleon (5). Data illustrating this method are shown in Fig. 1B, where we plot the energy loss (ΔE) versus the residual energy of nuclei stopping within a silicon detector. The elements are clearly separated into bands, and within an element band, subbands corresponding to the element’s isotopes are evident.

Fig. 1 CRIS instrument and data sample.

(A) A cross-section drawing (not to scale) of the CRIS instrument is shown. There are four x,y planes of fibers; the top plane provides a trigger signal and the next three planes (x1y1,x2y2,x3y3) provide the trajectory. Beneath the fibers are four stacks of 15, 3-mm-thick silicon detectors (two stacks are visible in this side view). The central active regions of the silicon detectors are shown in blue, and guard rings used to veto side-exiting particles are in green. The detectors are grouped so that up to nine energy-loss signals (E1 to E9) are obtained for each particle entering one of the detector stacks. We use the dE/dx versus residual energy technique to determine the atomic number (Z), mass (A), and energy (E), of each cosmic ray. (B) Cross-plot of the sum of scaled energy losses of cosmic rays in detectors E1, E2, and E3 on the y axis versus the scaled energy loss in E4 on the x axis for a sample of particles stopping in E4, with both energies scaled by (sec θ)–1/1.7, where θ is the angle of incidence through the instrument. Clear “bands” are seen for each element extending from calcium through nickel. Within some of the element bands, traces for individual isotopes of that element can be seen.

Figure 2A is a mass histogram of the observed iron nuclei that entered the CRIS instrument through the scintillating optical fibers, penetrated the silicon detectors E1 through E3, and stopped in silicon detectors E4 through E8. These data were collected for 6142 days from 4 December 1997 to 27 September 2014. They have been selected for consistency among mass calculations using different detector combinations to reject nuclei that interacted within the instrument and other spurious events, as well as selections related to event quality (14).

Fig. 2 Mass histograms of the observed iron and cobalt nuclei.

(A) The mass histogram of iron nuclei detected during the first 17 years in orbit is plotted. Clear peaks are seen for masses 54, 55, 56, and 58 amu, with a shoulder at mass 57 amu. Centered at 60 amu are 15 events that we identify as the very rare radioactive 60Fe nuclei. There are 2.95 × 105 events in the 56Fe peak. From these data, we obtain an 60Fe/56Fe ratio of (4.6 ± 1.7) × 10−5 near Earth and (7.5 ± 2.9) × 10−5 at the acceleration source. (B) The mass histogram of cobalt isotopes from the same data set are plotted. Note that 59Co in (B) has roughly the same number of events as are in the 58Fe peak in (A). There is only a single event spaced two mass units to the right of 59Co, whereas there are 15 events in the location of 60Fe, which is two mass units from 58Fe. This is a strong argument that most of the 15 nuclei identified as 60Fe are really 60Fe, and not a tail of the 58Fe distribution.

Abundance of 60Fe

Clear peaks are seen for masses from 54 to 58 amu (atomic mass units), with the exception of 57 amu, which is a shoulder on the 56 amu distribution. To the right of the 58 amu peak are 15 60Fe nuclei collected over this period, clearly separated from 58Fe and located where 60Fe should fall. These 15 events have a mean mass estimate of 60.04 amu and a standard deviation from the mean of 0.28 ± 0.05 amu, consistent with the 0.245 ± 0.001 amu measured for 56Fe. The number of 56Fe nuclei in this plot is 2.95 × 105, obtained from a Gaussian fit, and the total number of iron nuclei is 3.55 × 105. We have performed a number of tests to confirm that the 15 events are not a tail on the much more abundant 58Fe distribution and have verified that in all respects tested, these events are not unusual (supplementary text and figs. S2 to S6). The strongest argument that they are 60Fe nuclei and not spillover from 58Fe is shown in Fig. 2B, which is a histogram of the neighboring cobalt (Z = 27) isotopes from the same data set. Note that 59Co has roughly the same number of events as the 58Fe peak. To the right of 59Co, there is only a single event spaced by 2 amu. Similarly, we should expect only ~1 58Fe event to lie in the neighborhood of 60Fe, but we see 15. It is also unlikely that these are spillover from 56Fe since there is a clear separation between 58Fe and the events that we identify as 60Fe. Therefore, we conclude that at most, ~1 ± 1 of the 15 nuclei identified as 60Fe could be spillover from lighter iron isotopes (supplementary text and fig. S1). Using the cosmic-ray propagation model described in (15), we have calculated the expected intensity ratio between secondary 60Fe produced by fragmentation of heavier nuclei during interstellar transport and total Fe to be 2 × 10−6, from which we conclude that only ~1 of the observed 60Fe could be secondary. Thus, we have observed 13 ± 3.9 (statistical) ± 1 (systematic) = 13 ± 4.9 primary 60Fe nuclei, giving measured ratios for 60Fe/56Fe and 60Fe/Fe of (4.4 ± 1.7) × 10−5 and (3.7 ± 1.4) × 10−5, respectively.

These measured ratios need two small corrections. (i) Our data analysis eliminates particles that interact in the detector system. The cross-section for interaction of 60Fe is slightly larger than that for 56Fe, resulting in a correction factor of 1.009. (ii) Our analysis selects 60Fe and 56Fe with the same interval of depth of penetration in the detector. As a result, the energy interval for 60Fe is smaller by a factor 0.960 than the interval for 56Fe. Also, the 60Fe energies per nucleon are slightly lower, although this has a very small effect because the Fe spectrum is quite flat at these energies (15). The resulting energy-interval correction is a factor of 1.042. Taken together, these require that the ratios be multiplied by a factor 1.051 (supplementary text). Thus, the corrected ratios near Earth are 60Fe/56Fe = (4.6 ± 1.7) × 10−5 and 60Fe/Fe = (3.9 ± 1.4) × 10−5.

Using a leaky-box model, which accounts for radioactive decay of 60Fe, nuclear interactions as the cosmic rays travel through space, and leakage out of the galaxy, these observations imply 60Fe/56Fe and 60Fe/Fe ratios at the cosmic-ray acceleration source of (7.5 ± 2.9) × 10−5 and (6.2 ± 2.4) × 10−5, respectively (supplementary text).

Evidence of recent nucleosynthesis

Massive stars that undergo core-collapse exist primarily in localized groups called OB associations. The stars within these associations (or major subassociations) typically form at about the same time (within ~ 1 My), and associations have lifetimes of ~30 to 40 My (16). By that time, the stars with sufficient mass (>~10 M) have undergone core collapse. As the massive stars expel mass through high-velocity winds and supernova ejecta, superbubbles populated with the stellar wind material and ejecta are formed around the association (16). Calculations by Woosley and Heger (W&H) (3) and by Chieffi and Limongi (C&L) (4) estimate the production of 60Fe and 56Fe (outflow from stellar winds and supernova ejecta) from stars with initial masses from 12 to 120 M. Nearly all 60Fe is in the supernova ejecta, not in the wind material. We have combined these results for a range of stellar masses with the lifetimes of the progenitor stars to determine the expected 60Fe/56Fe ratio in an OB association as a function of time. In Fig. 3, we plot the cumulative 60Fe/56Fe ratio of material injected into a superbubble as a function of time since the formation of the association at T = 0. We then derived the time that must have elapsed for this ratio to decrease to the value inferred from our cosmic-ray measurements due to the decay of the 60Fe. Several different methods in which we compare our measured ratio with the modeling results lead us to conclude that an upper limit on the time between nucleosynthesis and acceleration is a few million years (supplementary text).

Fig. 3 Comparison of the 60Fe/56Fe ratio derived from measurements with models of supernova ejecta and stellar wind outflow in OB associations.

Models of massive-star ejecta have been used to estimate the 60Fe/56Fe ratio evolution as a function of OB association age, including stellar lifetimes and radioactive decay of the ejected 60Fe. The blue dotted curve shows the ratio for nonrotating stars using yields calculated by (3) (W&H); the red-dashed and black solid curves are for nonrotating and rotating stars, respectively, using yields calculated by (4) (C&L). The black dashed line is our estimated ratio (RA) at the acceleration source derived from our measurements. Assuming that the composition at the acceleration source is 20% massive-star material and 80% normal interstellar medium material (supplementary materials), we obtain the massive-star production ratio (RMS) (red dashed line). The shaded areas indicate uncertainties (not including the massive-star mix fraction uncertainty). We have compared the maximum predicted ratio (age ~3.4 My) with our measured ratio and estimate an upper limit to the time between nucleosynthesis and acceleration of several million years. Combining this with the previous measurement of 59Ni (2), we conclude that the time between nucleosynthesis and acceleration is 105 years < T < several My.

In previous work, on the basis of ACE-CRIS measurements of GCR 59Ni and 59Co, it was concluded that there is at least a 105-year delay between nucleosynthesis and cosmic-ray acceleration (2). Putting that lower limit together with this 60Fe observation, we obtain a mean time (T) between nucleosynthesis and acceleration of 105 years < T < several My. This is quite consistent with what is expected in an OB association since a typical time between supernovae is ~1 My (16). In addition, the time delay between nucleosynthesis and acceleration indicates that the nuclei synthesized in a supernova are not accelerated by that supernova, but require at least one more nearby supernova to accelerate the material, on a time scale sufficiently short so that a substantial fraction of the 60Fe has not decayed. The natural place for two or more nearby supernovae to occur within a few million years of each other is in OB associations. Thus, our observation of 60Fe lends support to the emerging model of cosmic-ray origin in OB associations (16, 17).

Using the mean lifetime of 60Fe, we can also estimate the distance to the associations contributing 60Fe GCRs at Earth. Assuming a diffusive propagation model, cosmic rays originate within a volume with radius L = (Dγτ)1/2 surrounding the solar system, where γ is the Lorentz factor and τ is the effective cosmic-ray lifetime. Assuming a diffusion coefficient of D = 3.5 × 1028 cm2/s (18), and using γ and τ calculated for 56Fe and 60Fe (supplementary materials), we find that L56 = 790 pc and L60 = 620 pc. Because the volume scales as the cube of these diffusion lengths, the volume contributing to 60Fe is only about half of that contributing to 56Fe. There are >20 OB associations or major association subgroups that have been identified within 620 pc of the Sun, including the very large and nearby (<150 pc) Sco-Cen association subgroups—Upper Scorpius (with 83 OB stars), Upper Centaurus Lupus (with 134 OB stars), and Lower Centaurus Crux (with 97 OB stars) (19)—and the Orion OB1 association, modeled by (20) (with ~70 OB stars). These are very likely major contributors to the 60Fe we have detected, owing to their size and proximity.

We can draw the model-independent conclusion that the detection of the radioactive supernova product 60Fe surviving in cosmic rays implies that the time required for acceleration and transport to Earth does not greatly exceed the 60Fe half-life of 2.62 My. Our distance from the source of this nuclide cannot greatly exceed the distance that cosmic rays can diffuse over this time scale, which is ⪍1 kpc.

Note added in proof: Additional detections of 60Fe in deep-sea crusts in all major oceans of the world have recently been reported (21), strengthening the conclusions reached in (9). Also, additional detections of 60Fe in lunar samples (22)strengthen the conclusion reached in (12, 13)


Materials and Methods

Supplementary Text

Figs. S1 to S6

References (2333)


  1. Materials and methods are available as supplementary materials on Science Online.
Acknowledgments: This work was supported by NASA grants NNX08Al11G and NNX13AH66G for work performed at the California Institute of Technology, the Jet Propulsion Laboratory, and Washington University in St. Louis. Work done at Goddard Space Flight Center was funded by NASA through the ACE Project. We thank S. Woosley and J. Brown at the University of California Santa Cruz for providing modeling calculation data and for discussions about uncertainties in their calculations. We thank A. Chieffi at the Institute for Space Astrophysics and Planetology, Rome, Italy, for discussions on the uncertainties in the C&L calculations. Accelerator testing of the CRIS detectors was made possible by N. Anantaraman, R. Ronningen, and the staff at the National Superconducting Cyclotron Laboratory at Michigan State University, while H. Specht, D. Schardt, and the staff of the GSI heavy-ion accelerator in Darmstadt, Germany made possible the heavy-ion calibrations of the completed CRIS instrument. The data used are archived at NASA’s Space Physics Data Facility ( as data set ac_h2_cris and can be retrieved from this site by direct download or through the SPDF’s CDAWeb data service.

Stay Connected to Science

Navigate This Article