Technical Comments

Response to Comment on "Determining Chondritic Impactor Size from the Marine Osmium Isotope Record"

See allHide authors and affiliations

Science  29 Aug 2008:
Vol. 321, Issue 5893, pp. 1158
DOI: 10.1126/science.1159234


Morgan argues that excursions in the marine Os record are of little value for estimating impactor size. This claim is based on computer simulations of the formation of the Chicxulub crater and distribution of the ejecta, which are difficult to validate. More important, by narrowly focusing on the Cretaceous-Tertiary event Morgan's comment misses the broader implications of our study.

We used the marine osmium isotope record to determine the size of the Late Eocene and Cretaceous-Tertiary (K-T) boundary chondritic projectiles (1). These estimates require knowledge of the fraction of projectile-derived Os that dissolves in seawater. Morgan asserts that most (>85%) of the Os carried by the Chicxulub projectile remains close to the impact site (2, 3), whereas our projectile size estimates assume that between 30 and 100% of the projectile-derived Os is vaporized and dissolved in seawater. Morgan seems to assert that her model results are accurate and that our assumptions are incorrect. We disagree and argue that Os isotope data from the rock record can be used to test our assumptions directly.

Our report (1) explicitly acknowledged that computer simulations of impact crater formation yield projectile size estimates substantially larger than our estimates based on Os isotopes: “If these larger projectile-size estimates are correct, this implies that only 2 to 7% of the Os carried by the K-T and Late Eocene Impact (LEI) projectiles dissolved in seawater” (1). The discrepancy between our smaller projectile size estimates and those derived from computer simulation could be eliminated by assuming that only a small fraction of the projectile-derived Os vaporizes and dissolves in seawater, for example, by using the 0.14 factor suggested by Morgan and Artemieva's simulations (3). We are not inclined to favor this interpretation because available K-T boundary data do not provide evidence that >85% of the projectile Ir remains unrecognized somewhere close to the impact site. As summarized in (4), Ir concentrations in K-T horizons do not vary systematically with distance from the Chicxulub crater. The lower Ir concentration (<1 part per billion) inside the Gulf of Mexico indicates dilution over thicker sedimentary sequences, but when integrated, this latter Ir flux is directly comparable to the Ir flux reported in distal sites. More recent results from the Chicxulub Drilling Project have revealed low (5, 6) and heterogeneously distributed (7) meteoritic enrichments of siderophile elements in the impactites. Therefore, we did not assume in our calculations that most of the K-T boundary Ir and Os accumulated close to the site of impact.

Morgan's comment (2) seems to suggest that Os and Ir are likely to behave similarly during projectile vaporization and subsequent condensation, leaving some readers with the impression that Os isotopes and Ir are redundant tracers of impact events. They are not. First, as emphasized in (1), the Os isotope excursion in the deep-sea sediment record provides a globally integrated measure of how much impact-derived Os dissolved in seawater. In contrast, the Ir fluence approach includes data from all depositional settings, marine and terrestrial, without regard to the chemical form of the Ir or how it was delivered to any particular site. Second, although both Os and Ir are regarded as refractory siderophile elements, their chemical behavior in the vapor phase can differ considerably. Under sufficiently oxidizing conditions, the stable form of Os is OsO4, a volatile compound with a boiling point of up to 130°C (8). Iridium does not form an analogous volatile oxide, although it can be volatile in the presence of high concentrations of halides (8). Thus, during impact events we suspect that there is substantial potential for Ir, and particularly Os, to escape early condensation in the vapor plume and become more widely dispersed than the bulk of the projectile. Additionally, our data across the LEI horizon at Ocean Drilling Program site 1090 in the Southern Ocean fit quite well to a simple one-box ocean model that assumes the entire inventory of Os and Ir in the sediment passed through the aqueous phase and was scavenged from seawater. If only a small fraction of these elements was vaporized and dissolved in seawater, we would expect the Os and Ir in the sediment to be carried mainly by particles that condensed from the vapor plume, thus conforming to a physical mixing model. The fact that they do not is another observation that is difficult to reconcile with Morgan's model results.

Finally, our data show that the mass of soluble impact-derived Os scales with crater size for the two largest known Cenozoic impact craters (Popigai Crater and Chicxulub Crater). These data demonstrate a qualitative sensitivity of the marine Os isotope record to projectile size. If the soluble fraction of projectile-derived Os is in the range of 30 to 100% as we have assumed, then the marine Os isotope record should allow detection of smaller impact events recorded by numerous known impact craters, as well as impact events that are as yet unrecognized. If the soluble fraction of projectile-derived Os is small and highly variable, then the magnitude of Os isotope excursion will not vary with crater size in a consistent manner and no excursions will be associated with smaller impact craters. In closing, we favor empirically testing our assumptions through further study of the marine Os isotope record rather than assuming they are incorrect based on model results from a single large impact event. This approach may lead to convergence in our understanding of impact events between geochemical approaches and computer modeling of impact dynamics.

References and Notes

View Abstract

Navigate This Article