ARTICLES Emplacement of -Tertiary Boundary from Walter Alvarez,* Philippe Claeys,t Susan W. Kieffer Observations on shocked quartz in Cretaceous-Tertiary (K-T) boundary sediments get rock and with the crystallization age of compellingly tied to Chicxulub crater raise three problems. First, in North America zircons from Chicxulub melt rock (17). It shocked quartz occurs above the main K-T ejecta layer. Second, shocked quartz is now seems clear that both layers were pro- more abundant west than east of Chicxulub. Third, shocked quartz reached distances duced by the Chicxulub impact and that requiring initial velocities up to 8 kilometers per second, corresponding to shock the shocked quartz and other minerals orig- pressures that would produce melt, not the moderate-pressure shock lamellae ob- inated from the basement granite. Iridium served. Shock devolatilization and the expansion of carbon dioxide and water from and quartz are believed to come from dif- impacted wet carbonate, producing a warm, accelerating fireball after the initial hot ferent sources: vaporized and un- fireball of silicate vapor, may explain all three problems. melted basement rock, respectively. Prob- lem 1 is thus, how did the shocked quartz and the iridium end up together in a sepa- rate layer, above the layer ofkaolinitic clay? In wells and outcrops of uninterrupted ma- (12) dating precisely from the The geographic distribution of K-T rine sedimentary rocks outside of North K-T boundary at 65.0 Ma (13) and sur- shocked quartz is not fully known, but it America, the K-T boundary is marked by a rounded out to -4000-km radius by proxi- appears to be much more abundant and single 1- to 10-mm clay layer often contain- mal ejecta at the biostratigraphic K-T slightly coarser grained at longitudes west of ing anomalous iridium and altered impact boundary (5, 7, 8) has strongly confirmed Chicxulub. Many sites in the western inte- spherules, widely interpreted as evidence for the general validity of the impact theory for rior of North America are rich in coarse the impact of a large comet or asteroid (1) the K-T mass extinction. Problems of detail shocked quartz grains (up to 0.64 mm) (2- at the time of the K-T mass extinction of remain, however, including three problems 5). In addition, shocked quartz grains are on July 27, 2009 organisms 65 million ago (Ma). A concerning the shocked quartz grains: (i) abundant and fairly coarse in all seven drill more complicated K-T boundary stratigra- their vertical distribution in the double K-T holes on the Pacific plate in which the K-T phy occurs in nonmarine sediments from layer of North America and their admixture boundary has been found (18, 19). This New Mexico, United States, to Alberta, with the iridium, (ii) the asymmetry of their presents a sharp contrast with sites in Eu- Canada, where the boundary interval be- geographic distribution about the Chicxu- rope, Africa, and Asia, where shocked quartz gins with a bed of kaolinitic clay, which lub crater, and (iii) their occurrence at great has been difficult to find (3, 4, 5, 20). sometimes containes goyazite spherules and distances from the crater. Problem 2 is thus, why is there an asymmetry is typically about 1 cm thick, overlain by a In North America, shocked quartz grains in the abundance of shocked quartz? layer 1 mm to a few millimeters thick that is are virtually absent from the kaolinitic clay K-T shocked quartz grains are found as www.sciencemag.org rich in shocked mineral grains, particularly layer, except where apparently carried far as 10,000 km from Chicxulub crater, at quartz (2, 3). The kaolinitic clay and downward by biogenic disturbance. The Ocean Drilling Project site 596 in the goyazite spherules are probably alteration clay layer and the quartz-bearing layer are southwest Pacific (18, 19). Ballistic trans- products of glassy ejecta (4-6) like that still sharply separated, and carbonized remnants port from Chicxulub to sites in the south- preserved in a few sites around the Gulf of of vegetation in the lower layer seem not to west Pacific requires a launch velocity of 7 Mexico (5, 7, 8). Shocked quartz is formed be present in the overlying quartz layer. to 8 km/s, corresponding to shock pressures under pressures of a few tens of This observation led to the no- that would anneal or melt the de- dynamic previously quartz, Downloaded from gigapascals, depending on the target com- tion that the layers were produced by two stroying the deformation lamellae that pro- position. An iridium anomaly is found in impact events at least one growing season vide the evidence for shock (21). Problem 3 samples taken from the shocked quartz layer apart; the vegetation traces in the lower is thus, how did these shocked quartz grains (9) and is attributed to the vaporization of layer were interpreted as roots of plants that achieve the velocities needed to reach dis- the impacting meteorite because of pres- grew before the overlying layer of ejecta was tal sites without melting? sures of many hundreds of gigapascals deposited (14). The lower layer was attrib- As shown below, the first two problems caused by a high-velocity impact. Some uted to the Chicxulub impact and the upper can be resolved if the shocked quartz trav- authors refer to the lower and upper layers, layer to Manson crater in (14) until eled on steep, high-velocity ballistic trajec- respectively, as the "melt-ejecta" and "fire- the following isotopic and age evidence tories. However, this requirement cannot be ball" layers (10). eliminated Manson as a K-T candidate cra- easily satisfied within the known constraints Recognition of the Chicxulub structure ter: (i) Sr, 0, and Nd isotopic measure- of impact cratering in dry silicate rocks: (11) in the Yucatan subsurface as a giant ments showed the K-T impact glass to be shocked quartz features are produced at pres- indistinguishable from Chicxulub melt sures of a few tens of gigapascals, and the W. Alvarez and P. Claeys are in the Department of Geol- rocks but very different from Manson melt particle velocities obtained are less than 4 ogy and Geophysics, University of California, Berkeley, rocks (15). (ii) The Manson impact was km/s on release from these pressures. Particle CA 94720-4767, USA, and Osservatorio Geologico di to shock Coldigioco, 62020 Frontale di Apiro (MC), Italy. S. W. dated as being earlier than the K-T bound- velocities of 7 to 8 km/s correspond Kieffer is in the Department of Geological Sciences, Uni- ary, at 73.8 + 0.3 Ma (16). (iii) Shocked pressures of 50 to 100 GPa, and the product versity of British Columbia, Vancouver, BC V6T 1Z4, zircons from the shocked have is melt, not moderately shocked quartz Canada. quartz layer crystallization ages much younger than the grains. To explain this contradiction, we *To whom correspondence should be addressed. basement rock at Manson but that of CO2 and tPresent address: Institut for Mineralogie, Museum fur compatible propose expansion H20 Naturkunde der Humboldt-Universitat zu Berlin, Invali- with the granitic Pan African basement vapor released from volatile target rocks denstraBe 43, D-10115 Berlin, Germany. thought to characterize the Chicxulub tar- would accelerate the quartz grains from sub- 930 SCIENCE · VOL. 269 · 18 AUGUST 1995 Itllll.IOI(IPB.PWI.PPslP1IIPB.PII. jacent moderately shocked granite to high site once they leave the atmosphere. from Chicxulub on trajectories steeper velocity. This process provides a mechanism Earth's rotation has two interesting ef- than about 65°. to address all three problems described. fects on these orbits. The first effect is that Developing this concept, we propose The present study is relevant not only to the semimajor axis of the elliptical orbit that the double layer of ejecta in the the specific case of the K-T boundary but depends only on the launch velocity in the western interior of North America reflects also to the understanding of impact process- inertial reference frame (25); to find this two different launch mechanisms during es in general. Large impact craters are com- velocity, the eastward rotational velocity of the cratering event. It has been proposed mon on rocky bodies in the solar system Earth (0.463 km/s at the equator) is added that the clay in the lower layer was formed with the exception of Earth, where craters to the target-frame launch velocity of east- by the alteration of glassy ejecta launched are relatively rare because they are erased by bound particles, but subtracted from that of as part of the ejecta curtain (4-6). The rapid geological processes. Comparison of westbound particles. This has little effect on ejecta curtain observed in hypervelocity lunar, martian, and venusian crater mor- the semimajor axis of slow particles, but at impact experiments is an outward-expand- phologies, as well as field and theoretical launch velocities from about 8 or 9 km/s up ing, downward-pointing cone inclined studies of terrestrial craters (22) suggest that to escape velocity (11.2 km/s), eastbound -45° to the horizontal (26) and repre- volatiles in the target body may significant- particles go higher and stay up longer than sents the coherent front of solid and melt ly influence the impact processes and prod- the corresponding westbound ones. The ejecta particles on independent trajecto- ucts. The Chicxulub crater is buried, so it is second effect is that, because Earth rotates ries, launched at elevation angles .45° inaccessible but uncommonly well pre- beneath in-flight ejecta, the reimpact site is [figure 17 in (27); figure 6.4 in (28)]. served. Study of this large, young terrestrial at the same latitude but displaced to the Because a 45° elevation angle yields the crater will help clarify the processes in- west, as compared with a nonrotating Earth. longest ballistic range for a given launch volved in comet and asteroid impact. More- This effect is small for slow ejecta, but velocity, this material will be the first to over, this was an unusual impact because of important for fast ejecta. arrive at a given site (Fig. 2). For example, the combined carbonate and granite target Because of these effects, there is a for- to reach the K-T site at Clear Creek, lithologies, which would have generated bidden zone east of the impact site that Colorado, ballistic ejecta launched from large amounts of CO2 and H2O. Kieffer and cannot be reached by eastbound ejecta Chicxulub at an angle of 45° above the Simonds (22) and O'Keefe and Ahrens (23) unless the launch velocity is high and the horizon requires an initial velocity of 4.4 have given general consideration to the role elevation angle low (Fig. 1). All of Europe km/s and has a travel time of 14 min. In of COz and H)O vapor in impact and the and Africa is in the forbidden zone for contrast, if the shocked quartz at Clear on July 27, 2009 impact cratering process; we now apply ejecta launched at 70°, and all except the Creek was launched at an elevation angle of these considerations to the specific case of extreme western margin is forbidden at 70°, as suggested by the global distribution, the Chicxulub crater. 60°. At 50°, the forbidden zone is reduced its initial velocity was 5.7 km/s and its to a small area around India and eastern travel time was 28 min. Ejecta-curtain par- Global-Scale Ballistics Africa. The known distribution of K-T ticles launched at 30° to 45° have travel of the Ejecta shocked quartz would be well explained if times to Clear Creek of about 10 to 15 min, most shocked quartz grains were launched whereas steeper ejecta particles, launched at We suggest that both the distribution of the shocked quartz and its occurrence in a sep- www.sciencemag.org arate layer can be explained by transport of the quartz grains on ballistic trajectories different from those of the glassy ejecta which altered to form the kaolinitic clay layer. We have calculated the reimpact loci of ballistic ejecta from Chicxulub as a func- 0 tion of the and elevation of 0)

velocity angle 0 Downloaded from 0n launch, taken around a 360° range of 0 0 launch azimuths (24). Except when inter- 0,0 acting with the atmosphere during launch and re-entry, ballistic ejecta particles fol- low elliptical orbits with one focus at Earth's center, when plotted in inertial coordinates, until they reimpact Earth's surface. We ignore atmospheric interac- tions during re-entry, considering them to -180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180 have only a minor effect on the deposition Longitude (degrees) point of sand-sized grains. However, set- Fig. 1. Reimpact patterns of ballistic ejecta launched at Chicxulub shown on a map of continental tling times through the atmosphere may positions at K-T boundary time (the dashed line bounds pre-K-T Pacific plate that has not subsequently complicate our conclusions and need fur- been subducted) (35). These patterns ignore atmospheric effects, which occur only at the times of launch ther consideration when more data on and re-entry. For a launch angle of 70° above the horizon, thin lines show reimpact loci for launch particle size and atmospheric conditions velocities at 1 -km/s increments. Heavy lines show the limit of the forbidden zone at 70° as well as for 60° after impact are available. Interactions and 50° launch angles; in the latter two cases reimpact loci are omitted to avoid clutter (36). Solid squares launch are and are dis- mark sites with coarse shocked quartz (grains >250 pim in diameter); circles mark sites with abundant during important fine shocked quartz; diamonds mark sites with rare, fine shocked quartz; and crosses mark sites where cussed below, but the size of the expanding shocked quartz has been reported, but information is insufficient to determine its abundance. Schultz has ball of gas around the impact site is small suggested that the abundance of shocked quartz in the U.S. western interior K-T sites reflects a compared with long-range trajectories, so low-angle oblique impact toward the northwest (37). Alternatively, we propose that the asymmetrical the particles can be treated as if launched distribution of shocked quartz grains, heavily concentrated west of the impact site, can be explained if into ballistic trajectories from the impact most grains were launched at an angle steeper than -65°. SCIENCE · VOL. 269 · 18 AUGUST 1995 931 .tt.tirolrrss6. 65° to 80°, have travel times of 23 to 63 We assume that a stony meteorite 10 km tion wave that releases the shocked mete- min. Thus, all the ejecta-curtain material in diameter traveling at 24.6 km/s strikes a orite, and eventually the shocked target would have arrived at Clear Creek before region with a 3-km-thick layer of wet car- rock, back toward ambient pressure (stage any of the steep ejecta began to fall (Fig. 2). bonate overlying a granitic basement (33). 2b and Fig. 3, B and C). By the time the For a more northern site, like Brownie During stage 2 and the early parts of stages rarefaction reaches the meteorite-target in- Butte, Montana, the corresponding travel 3 and 4, the shock Hugoniots and high- terface, the meteorite has reached the end times are 13 to 18 min for ejecta curtain pressure parts of the release adiabats are of its penetration path at 13 km. The me- particles and >33 min for steep ejecta. relatively independent of rock type, and for teorite and a closely adjacent mass of rock This travel time sorting may explain the these stages the shock equation-of-state of roughly equivalent mass are vaporized double layer of the western interior, but not properties used are those of diabase for the and begin ascending in a hot fireball. if the organic traces in the lower layer rep- stony meteorite and those of granite for the Although some energy is released along resent the roots of plants that grew between target rock. The properties of volatile rocks the whole penetration path, and although two falls, which would thus be separated by become important later in the cratering free-surface effects on the meteorite and at least one growing season. We suggest event and are considered separately below. target rock are important in detail (23, 32), instead that the organic traces may mark The meteorite is assumed to have a density to first order the process can be modeled by the stems of plants that burned during the of 3000 kg/m3, with a corresponding mass of examining the decay of the peak-pressure arrival of the particles of the lower layer, 1.6 x 1015 kg and kinetic energy of 4.8 X isobars radially from a maximum value of heated by their atmospheric re-entry (28, 1023 joules (114 X 106 Mton). 660 GPa centered in a mass of material at 29), and that had fully burned (30) and On contact (stage 1), a shock wave is the depth of penetration (Fig. 3, B and C). collapsed before the fall of the second layer. propagated into the target at nearly 20 km/s, Because the depth of penetration is only shocking both the meteorite and the target about one meteorite diameter, the shock Model of the Cratering Event up to a pressure of 660 GPa (Fig. 3A). For waves penetrating downward have a dif- the first 0.5 s (stage 2), while the shock wave ferent decay pattern than those moving A cratering model is required that explains travels to the back of the meteorite, the toward, and reflecting from, the surface the high velocities, near-vertical trajecto- meteorite penetrates through the carbonate (22, 23, 30, 32). This causes radial differ- ries, and relatively cool temperatures need- and into the granite. The shock wave that is ences in peak pressure which, combined ed for the shocked quartz to be carried far to simultaneously traveling into the ground ac- with the different lithologies along differ- the west and not annealed. To examine the celerates material downward or radially away ent paths, influences the state of shocked on July 27, 2009 cratering process, we used the semianalytic from the meteorite, but not yet upward. material ejected. model of Kieffer and Simonds (22), in When the shock wave reflects from the In dry silicate rocks, the products of which the processes that occur during an back of the meteorite, it becomes a rarefac- shock decompression vary with pressure in are subdivided into seven suc- cessive, sometimes overlapping, stages: stage 1, initial contact; stage 2, compression Ballistics Stratigraphy and release of the meteorite; stage 3, rar- efaction and attenuation in the target; stage .2 t www.sciencemag.org 4, excavation and flow within the crater; 0 stage 5, ejecta launch and fallback; stage 6, mechanical modification; and stage 7, hy- -13 drothermal and chemical alteration. In this article, we consider primarily 0 -10 5. For each the 5 2 Cu stages through stage, S conservation laws and thermo- Sc appropriate -5 dynamic properties (or approximations) Downloaded from were solved to yield properties such as I.- depth of penetration, duration of each stage, peak shock pressure in the meteorite o -0 of and ground, and attenuation peak pres- 0, sure as a function of distance from the 0 meteorite at its point of maximum penetra- I tion. this model is Although simplistic 0 10 20 30 40 50 60 70 80 9""'Forbidden zone" when compared with computer simulations (ejecta from Chicxulub with launch angles steeper such as those of O'Keefe and Ahrens (23) Launch angle~~Launch(degresangl(degrees aboeabove horthanhorizontal) 81° cannot reach Clear Creek, CO) and Roddy et al. and Vickery and Melosh Fig. 2. Calculated history of ejecta arrival at Clear Creek, Colorado (37° 05.26' N, 104° 31.33' W), (30), it has the advantage of providing a compared with the fine-scale stratigraphy of the K-T boundary at the Clear Creek North site of Izett (3), relatively intuitive overview of the whole ignoring atmospheric effects during launch and re-entry. The boundary clay layer is made largely of cratering process. The results (31) are in kaolinitic clay, probably resulting from alteration of glassy ejecta (4-6). If launched in an ejecta curtain like reasonable semiquantitative agreement with that observed in experiments many orders of magnitude smaller, these particles would have had launch these earlier models and are in angles of 30° to 45° and would have reached this site in 10 to 15 min, with launch and reimpact velocities particularly close to 4.5 km/s. The overlying layer is rich in shocked quartz. If these grains traveled on trajectories good agreement with the model of Pope et steeper than -65°, as inferred from the geographic distribution of shocked quartz (Fig. 1), they would al. (32) who did a similar calculation based have had initial velocities >5.2 km/s and would have arrived >23 min after the impact. Launch velocities on anhydrite thermodynamics in order to this high would implyenergies sufficient to melt these obviously unmelted grains (21), so we infer that they estimate climate effects. The sensitivity of were accelerated in an expanding "warm fireball" of CO2 and H20O vapor. Wispy carbonized remains the model to impact velocity and to mete- within the boundary claystone, previously interpreted as the roots of plants that grew in an interval of -1 orite and target composition is discussed in between deposition of the two layers, may instead represent the stems of plants ignited by infrared detail by Kieffer and Simonds (22). heat from re-entry of the early ejecta and covered a few minutes later by the shocked quartz. 932 SCIENCE * VOL. 269 * 18 AUGUST 1995 IBlb.L.eB.UIIIIIIIIIIIIPI. -.lsllllCrcil Ir.L-----aL3L.sa·lr---- -r.·------IIII.II.LI-.-.101111111(-..

the following sequence: vaporized meteorite launched at an angle of -45° and at veloc- tially break down into CaO and CO2 on and vaporized rock (shocked to pressures ities of a few kilometers per second. It would decompression from -45 GPa and com- over 100 GPa), melted rock (shocked to be expected that the shocked quartz pro- pletely break down if shocked to over --70 pressures >50 GPa roughly), highly and duced deep under the impact point proxi- GPa. The pressure range for anhydrite moderately shocked rock (including mal to the melt zone would largely be en- breakdown is similar, although the degree of shocked quartz grains and shocked feldspars trained in the ejecta curtain (22). However, equilibrium attained in any of these devola- at pressures of a few tens of gigapascals), if 4.5 km/s is the maximum launch velocity tilization reactions is a subject of much weakly shocked, and then fractured rock. that will not anneal or melt shock features controversy. In our simple model, 70 GPa is These are the products that we would ex- (21), then shocked grains will travel no reached at a radius of - 11 km (surface) and pect to form sequentially and temporally as farther than the Gulf Coast of the United 45 GPa at about 12 km. Water in pores and the shock wave decays down and out into States (24); this agrees with the presence of cracks will vaporize if shocked to over 10 the granite. For example, in this model, the rare shocked quartz in ejecta-curtain (lower GPa, reached at a surface radius of 18 km in peak pressure of 660 GPa is attained at 13 layer) deposits in the Gulf of Mexico K-T this model. km depth, and 15 GPa at a depth of 30 km. sections (8) and the absence of shocked Material in the carbonate layer out to a Thus, we expect shocked quartz to be pro- quartz in the lower K-T layer in the western radius of at least 18 km will partially or even duced under the meteorite site at depth. United States (3-5) (Fig. 1). totally vaporize, releasing a mixture of A hot fireball is formed from vaporized In the granite above the depth of pen- steam, COz, oxides, and fragments of the material surrounding the penetration cavity etration we would expect the same se- carbonate. This is a very large amount of (Fig. 3, B and C). The rise of this vapor from quence of shocked products as produced vapor. If we consider only an annulus of the impact site may be coupled with other below the meteorite. The relative abun- carbonate between 11- and 18-km radius, atmospheric phenomena such as atmospher- dances will be different than below the shocked to between 70 and 10 GPa, the ic preheating during entry of the meteorite meteorite because they are influenced not volume of material is 2000 km3. At a den- and explosions if the meteorite had partly only by radial pressure decay but by free- sity of 2500 kg/m3 (purposely reduced from disintegrated within the atmosphere (34). surface and entry path effects. We esti- the density used for average shock wave Complicated shock waves travel in the air, mate that at a radial distance of --10 km properties of the target to allow for some both from the initial meteorite entry and from the center of impact, shock pressures porosity and water content), the mass from the rising and decompressing fireball. are a few tens of gigapascals, and lamellae would be 5 x 1015 kg, approximately three This hot vaporized material is rich in mete- and other deformation features should be times the mass of the impacting meteorite. on July 27, 2009 oritic components and in vaporized compo- produced in the granite. A typical energy for vaporizing this material nents from near the penetration cavity, car- The wet carbonate cover, however, has a is about 1013 ergs/kg. If only 10% of the bonate and silicate in this case. vastly different behavior in these pressure material vaporized, this would require input The strongly, moderately, and some of ranges. Wet carbonate can produce CO2 of 3 x 1027 ergs, or about 105 Mton. This the weakly shocked material surrounding and H20 vapor at relatively low pressure; energy would be deposited in the vapor on the impact site and lining the walls of the devolatilization of incorporated anhydrite a time scale of a few seconds. expanding transient cavity is turned upward would also contribute volatiles over approx- We suggest that the CO2 and H2O vapor and outward by rarefaction waves. This flow imately the same pressure range. Although from this devolatilized carbonate zone be- develops into the ejecta curtain formed by data are sparse and somewhat inconsistent, tween --10 and 20 km (rounded off) as- www.sciencemag.org the coherent front of melt and solid ejecta in general it is agreed that carbonates par- cended as a "warm fireball" which dragged

B C,, D f shocks-I .fireballl Shocked " ~ 0(D0,m atmosphere E ~ E 'CDO~~20 2404kkm ~.{~2 < ' Hot fireball co^- o ^ Shocked *7Bl'/ffli^E7y~lj20 40 km S ^W y> latmosphere 0 Downloaded from a ~~Warm 20o',~_v y~gSS^B»B ~~ot fireball Gt40/km 40 km

(A) At time4 t 0.5 s a strong shock(S) moves down and out 40

gthe of the meteoritei the atmosphere. At t intoentry3the target rocks. Airthroughshocks have been formed(B)during km ?'w 1 s, taken when the rarefaction from the back surface of the meteorite has just reached the meteorite-target interface./ This rarefaction causes expansion and reverse flow of the S§Shocked meteorite vaporized meteorite and silicate out of the penetration cav- WetW carbonate20 ity. (C) At t - 2 s, rarefaction (R) from free surfaces is now Grnt Shocked carbonate eating back in toward the shock front (S), decompressing '-j shocked granite the materials from peak pressures and turning the flow up Shce grait and out of the crater. (D) A few seconds after the impact, the ejecta curtain of melted, moderately shocked, and unshocked material is developing beneath a fireball doublet. The doublet consists of a hot f fireball of and and entrained rock fireball of rock vapor and a warm CO H2O vapor fragments. t-~ a few seconds

SCIENCE * VOL. 269 * 18 AUGUST 1995 933 .lllgsrrPssrr.li·I.8par..

with it not only shocked fragments from the Prediction of a Third Layer valuable analog for impact studies on both carbonate from which it originated but also dry planets (moon, Mercury) and volatile- moderately shocked fragments from the The K-T impact bed in the western United containing planets (Mars, Venus). subjacent granite near the contact. We did States contains two clearly recognizable lay- not attempt to estimate the velocity of this ers. We interpret the lower layer as repre- REFERENCES AND NOTES curtain fireball with our model because by the time senting the ejecta and the upper 1. L. W. Alvarez, W. Alvarez, F. Asaro, H. V. Michel, it formed, the atmosphere was highly al- layer as derived from the warm fireball. The Science 208, 1095 (1980); J. Smit and J. Hertogen, tered by previous shocks, and details of iridium anomaly is found in samples taken Nature 285, 198 (1980); F. T. Kyte, Z. Zhou, J. T. and free surfaces have from the The carrier for iridium Wasson, ibid. 288,651 (1980); J. Smit and G. Klaver, cratering geometry upper layer. ibid. 292, 47 (1981); A. Montanari et al., Geology 12, become important. Experimental data on has never been recognized in any K-T sites, 696 (1984). dry and wet, nonporous and porous, media and it probably occurs in extremely fine 2. B. F. Bohor, E. E. Foord, P. J. Modreski, D. M. Triple- show that the maximum launch Iridium and come from dif- horn, Science 224, 867 (1984); G. A. Izett and C. L. velocity particles. quartz Pillmore, Eos 66, 1149 (1985); C. L. Pillmore and R. obtainable from shock pressures of 30 GPa ferent sources: vaporized meteorite and un- M. Flores, Geol. Soc. Am. Spec. Pap. 209, 111 is 7 km/s, somewhat under our desired value melted basement, respectively. The iridium (1987). of 8 is to have been as a 3. G. A. Izett, Geol. Soc. Am. Spec. Pap. 249,1 (1990). km/s (21). likely deposited separate 4. B. F. Bohor, ibid. 247, 335 (1990). Buoyancy effects and atmospheric-scale veneer on the top of this upper layer be- 5. C. Koeberl and H. Sigurdsson, Geochim. Cosmo- pressure gradients may be equally impor- cause of the slow settling of fine particles. chim. Acta 56, 2113(1992). tant. Jones and Kodis (34) have suggested Iridium does occur in the finest material at 6. B. F. Bohor, B. P. Glass, W. J. Betterton, Abstr. in Lunar Planet. Sci. Conf. 24, 145 (1993); R. M. Pol- that if the energy deposited suddenly in a the top of the marine K-T boundary beds lastro and B. F. Bohor, Clays Clay Miner. 41, 7 compressed gas exceeds - 12,000 Mton, a the deep Gulf of Mexico (8). We therefore (1993). unconfined the will 2) that even resolution 7. G. A. Izett, J. Geophys. Res. 96, 20879 (1991); F. fireball, by atmosphere, predict (Fig. higher Science 1690 should reveal a J.-M. R. Maurrasse and G. Sen, 252, rise, passing the point of neutral buoyancy iridiumstratigraphy triple (1991); H. Sigurdsson et al., Nature 349,482 (1991); where pressure and density become equal to boundary layer at sites in the western Unit- J. Smit et al., Proc. Lunar Planet. Sci. Conf. 22, 87 atmospheric pressure and density, before it ed States, with the three layers related to (1992); C. Koeberl, Geochim. Cosmochim. Acta 56, For three distinct mechanisms for 4319 (1992). decompresses completely. comparison launching 8. W. Alvarez et al., Geology 20, 697 (1992); J. Smit et we estimated above that the warm fireball ejecta during the impact event: (i) An ex- al., ibid., p. 99; P. Claeys, W. Alvarez, J. Smit, A. R. had -105 Mton of energy, more than suf- tremely hot fireball consisting of vaporized Hildebrand, A. Montanari, Abstr. Lunar Planet. Sci. ficient for this unconfined rise. In meteorite and carbonate and sil- Conf. 24,297 (1993); J. Smit, T. B. Roep, W. Alvarez, on July 27, 2009 buoyant target-rock P. Claeys, A. Montanari, Geology 22, 953 (1994). this case, the fireball continues accelerating icate is launched first, but because of the 9. C. J. Orth, M. Attrep Jr., L. R. Quitana, Geol. Soc. upward along the decreasing atmospheric fine grain size of the material it carries, this Am. Spec. Pap. 247, 45 (1990). We that the is the last to settle. (ii) A -45° 10. A. R. Hildebrand and W. V. Boynton, paper present- pressure gradient. hypothesize portion ed at the 2nd Conference on Global Catastrophes in volume of shocked carbonate and aqueous ejecta curtain carries melted and shocked Earth History, Snowbird, UT, 20 to 23 October 1988, gases produced in the zone peripheral to the solid rock fragments that are emplaced first Lunar Planet. Inst. Contrib. No. 673 (1988), p. 78; B. hot fireball ascended with fireball- because of their lower, more direct F. Bohor and W. J. Betterton, Abstr. Lunar and Plan- rapidly trajec- Conf. B. F. Bohor and B. P. warm of et. Sci. 21,107 (1990); like buoyancy dynamics and that this warm tories. (iii) A quite distinct fireball Glass, Meteoritics 30, 182 (1995). fireball accelerated the shocked and steam accelerates moderately 11. A. T. and A. Hernandez 0., Bol. Asoc. Mex.

granite CO2 Cornejo www.sciencemag.org fragments to velocities of 8 to 11 km/s and shocked granite fragments into ballistic tra- Geol. Pet. 2, 453 (1950). than about and 12. G. T. Penfield and A. Camargo Z., Soc. Explor. Geo- into nearly vertical paths (Fig. 3D). As the jectories steeper 65°, they phys. Tech. Program, Abstr. Biogr. 51,37 (1981); A. warm fireball expanded, the solid particles arrive after the ejecta curtain material but R. Hildebrand et al., Geology 19, 867 (1991); A. decoupled from the gas and followed ballis- before the hot-fireball iridium. Camargo Z. and G. Suarez R., Bol. Asoc. Mex. of Geof's. Explor. 34, 1 (1994). tic trajectories to their sites deposition. 13. C. C. Swisher III et al., Science 257, 954 (1992); V. L. In summary, during the release phase, we Conclusion and Applicability Sharpton et al., Nature 359, 819 (1992). propose that three relatively distinct flow 14. G. A. Izett, Eos 72,278 (1991); E. M. Shoemaker and to occur G. A. Izett, Abstr. Lunar Planet. Sci. Conf. 23, 1293 fields a hot an In order address the that Downloaded from developed: fireball, ejecta processes (1992). curtain of melt plus strongly and moderately during planetary-scale impacts, this study 15. J. D. Blum etal., Nature 364, 325 (1993). shocked solid ejecta, and a separate vapor- has involved field, laboratory, and theoret- 16. G. A. Izett, W. A. Cobban, J. D. Obradovich, M. J. mixture which we have termed a ical aspects. Because of the sheer scale of Kunk, Science 262, 729 (1993). particle 17. T. E. Krogh, S. L. Kamo, B. F. Bohor, Earth Planet. "warm fireball." The warm fireball is at the process, no single technique is likely to Sci. Lett. 119, 425 (1993); W. R. Premo and G. A. least conceptually distinct from the initial be sufficient. As sophisticated as numerical Izett, Abstr. Lunar Planet. Sci. Conf. 24,1171(1993); fireball produced by vaporization of the me- modeling of the theoretical concepts has T. E. Krogh, S. L. Kamo, V. L. Sharpton, L. E. Marin, of the material A. R. Hildebrand, Nature 366, 731 (1993); S. L. teorite and of the carbonate and silicate become, the complexities Kamo and T. E. Krogh, Geology 23, 281 (1995). target rock directly beneath the impact properties of geologic, planetary, and mete- 18. F. T. Kyte, J. A. Bostwick, L. Zhou, paper presented The time of oritic and processes cannot be at the conference on New Developments Regarding point. origin, pressure-temper- compositions in Earth His- and as- addressed. are in- the KT Event and Other Catastrophes ature histories, chemical content, Laboratory experiments tory (Lunar and Planetary Institute, Houston, TX, 9 to cent dynamics of these two kinds of gaseous herently limited in scale and duration, but 12 February 1994), Lunar Planet. Inst. Contrib. No. material are different. We refer to the dou- they provide valuable insights about pro- 825 (1994), pp. 64-65. doublet" and material Field 19. J. A. Bostwick and F. T. Kyte, in Proceedings ofthe ble gaseous regime as a "fireball cesses, rates, properties. Conference on New Developments Regarding the (Fig. 3D). We cannot say whether the dou- observations are limited by exposure and KT Boundary and Other Catastrophes in Earth His- blet merges or separates with time because interpretation, but they are global in scale. tory, G. Ryder, S. Gartner, D. Fastovsky, Eds. (Geo- of the In this we have a model logical Society of America, Boulder, CO, in press). our model does not allow calculation study, proposed 20. B. F. Bohor and G. A. Izett, Abstr. Lunar Planet. Sci. detailed time history of fireball-doublet be- based on all three techniques of geologic Conf. 17, 68 (abstr.) (1986); A. Montanari, J. Sedi- havior. Detailed studies of the vertical dis- analysis and have proposed some tests to ment. Petrol. 61, 315 (1990). Shocked quartz abun- tribution of iridium and other meteoritic further work. We have that dance determinations made in different laboratories guide proposed may not be strictly comparable. The difference in components compared with that of the a thin veneer (3 km) of volatile sediments shocked quartz abundance between the Pacific and shocked quartz may allow testing of this strongly influenced global ejecta distribu- European sites has only recently emerged through idea, as described below. tion. If true, then Chicxulub will provide a the work of Bostwick and Kyte (18, 19). It will be

934 SCIENCE · VOL. 269 * 18 AUGUST 1995 .i.33.f..a.:urrelrs

important to have sites from the two areas compared meteorite radius ranged from a minimum of 2.6 (dia- farther west, but not farther east, so these are firm in the same laboratory. base, carbonate, granite) to 3.2 (permafrost), to a limits on the west edge of the forbidden zone for each 21. Deformation features are directly correlated with maximum of 4.3 (liquid water). elevation angle. Cusps in the boundaries of the for- peak shock pressure. Numerous studies show that 32. K. 0. Pope, K. H. Baines, A. C. Ocampo, B. A. Ivanov, bidden zones are an artifact produced byjoining seg- moderate shock features (reduced refractive index, Earth Planet. Sci. Lett. 128 (N3-4), 719 (1994). ments of reimpact loci lines at 1 -km/s increments. lamellar features) form at pressures between 20 and 33. E. Lbpez Ramos, Geol. Mex. 3, 277 (1981). The 37. P. H. Schultz, in (18), pp. 105-106. 40 GPa (depending on the rock type), and shock- "rocas volcanicas" mentioned on p. 277 are proba- 38. We thank J. Bostwick and F. T. Kyte for discussion fused quartz at pressures above 50 GPa. In turn, bly the impact melt rocks of the Chicxulub crater, not and for a preprint of their work on the K-T boundary peak shock pressure is directly related to the particle basement rock. of the Pacific plate and D. Rowley for plate-rotation velocity attained behind the shock. To a very good 34. E. M. Jones and J. W. Kodis, Geol. Soc. Am. Spec. parameters. Comments by R. Jeanloz and C. Koe- approximation in dry rocks, the velocity is doubled by Pap. 190, 175(1982). berl on an earlier version of this manuscript are great- the passing rarefaction [R. G. McQueen, S. P. 35. Continental positions calculated with the 66-Ma ly appreciated. Supported by the National Aeronau- Marsh, J. N. Fritz, J. Geophys. Res. 72, 4999 rotation parameters of the University of Chicago tics and Space Administration (NASA) grant NAGW- (1967)]. For a peak shock pressure of 20 GPa, the Paleogeographic Atlas Project, D. Rowley, person- 3008 and NSF grant EAR-91-05297 to W.A., NASA measured shock velocities in 12 representative al communication. grant NAGW-1740 to S.W.K., and University of Cal- rocks are 0.8 to 1.4 km/s. The rarefaction velocities 36. At velocities greater than 10 km/s, ejecta can travel ifornia Berkeley Esper A. Larsen Jr. Research Fund. are then in the range of 1.6 to 2.8 km/s. Ahrens and Rosenberg [in of Natural Min- erals, B. French and N. Short, Eds. (Mono, Balti- more, MD, 1968), p. 59] extended the range of pres- sures to 45 GPa and measured all rarefaction (free Resonance Light Scattering: surface) velocities to be less than 4.5 km/s. We therefore take 4.5 km/s as an upper limit on the velocity that could be attained by moderately A New Technique for Studying shocked quartz grains launched by a simple crater- ing event in dry rocks. At higher pressures, higher rarefaction velocities are obtained, but the features Chromophore Aggregation of moderately shocked quartz are not preserved be- cause the quartz is annealed or fused. Target rock texture (porosity in this case) and wetness also affect Robert F. Pasternack and Peter J. Collings shock metamorphic features and rarefaction veloci- ties [S. W. Kieffer, J. Geophys. Res. 76, 5449 (1971)]. A significant effect for this discussion is that Light scattering experiments are usually performed at wavelengths away from absorp- porosity and volatile content both increase the pos- tion bands, but for species that aggregate, enhancements in light scattering of several sible rarefaction velocities. For example, dry soil orders of can be observed at characteristic of these shocked to 30 GPa releases magnitude wavelengths species. (-porous quartz soil) on July 27, 2009 with a velocity of -5.7 km/s, higher than that mea- Resonance light scattering is shown to be a sensitive and selective method for studying sured for single crystal or nonporous polycrystalline electronically coupled chromophore arrays. The approach is illustrated with several quartz. Wet soil shocked to the same pressure re- drawn from and chlorin The under- leases with a particle velocity of -7.3 km/s [G. D. examples porphyrin chemistry. physical principles Anderson et al., Technical Report AFWL- TR-65-146 lying resonance light scattering are discussed, and the advantages and limitations of (SRI International, Menlo Park, CA, 1966)]. Thus, an the technique are reviewed. impact into porous, water-saturated carbonates will produce products with significantly higher velocities than a similar impact into a dry, nonporous silicate target. 22. S. W. Kieffer and C. H. Simonds, Rev. Geophys. We have recently reported (1) on a new encouraged the development of experimen- Space Phys. 18, 143 (1980). resonance called resonance tal of and www.sciencemag.org 23. J. D. O'Keefe and T. J. Ahrens, Geol. Soc. Am. technique light techniques capable detecting Spec. Pap. 190,103 (1982); Nature 338,247 (1989). scattering (RLS) that is both extremely sen- characterizing these assemblies. Under- 24. W. Alvarez, in (19). The necessary equations were sitive and selective in probing chromophore standing the chemical, biological, and phar- modified from those given by A. Dobrovolskis [Icarus in a number of different of a 47, 203 (1981)]. His dimensionless parameters were aggregation sys- macological activity complex system converted to Earth's case as follows: 1 time unit = tems. The theory behind this technique is requires knowledge of the state of molecular 806.3 s, 1 length unit = 6371 km, and 1 mass unit = not new, and in fact RLS has been tried in aggregation of the system's components. Re- 5.973 x 1027 g. Time-of-flight calculations follow the past for purposes other than studying search issues that involve relations between those described by A. D. Dubyago [The Determina- In those cases, the the of and the tion of Orbits (Macmillan, New York, 1961)]. aggregation. technique properties complex systems Downloaded from 25. The semimajor axis, a = 1/(2 - v2) in dimensionless was only marginally successful. However, formation of large aggregates include (i) the units, where v is the launch velocity. we have found that in aggregation experi- organization of chlorophyll and other pig- 26. P. H. Schultz and D. E. Gault, Geol. Soc. Am. Spec. RLS not meets and ments in in which a Pap. 190, 153(1982). ments, only sensitivity chlorosomes, particular 27. V. R. Oberbeck, Rev. Geophys. Space Phys. 13,337 selectivity criteria but offers the additional molecular assembly is required for efficient (1975). benefits of simplicity and versatility. In this photosynthesis (2); (ii) the relation of pho- 28. H. J. Melosh, Impact Cratering: A Geologic Process article, we describe the basic physics behind tosensitization efficiency to the extent of (Univ. of Arizona Press, Tucson, 1988) some research on its of the active at can- 29. H. J. Melosh, N. M. Schneider, K. J. Zahnle, D. RLS, survey ongoing aggregation component Latham, Nature 343, 251 (1990). refinement and application to different sys- cer cells, which must be resolved to enable 30. D. J. Roddy et al., Int. J. Impact Eng. 5, 525 (1987); tems, and discuss a number of areas in rational design of alternative reagents in A. M. Vickery and H. J. Melosh, Geol. Soc. Am. which RLS make contribu- cancer (3); and (iii) Spec. Pap. 247, 289 (1990). may important photodynamic therapy 31. The sensitivity of the impact cratering model to vary- tions in the near future. the state of aggregation of lipids and drugs in ing meteorite and target compositions and to impact liposomes, which has been shown in certain velocity can be found in (22), especially tables 2 to 4, Experimental Approaches to cases to be a crucial factor in their effective- and figures 6 and 7. During the compression stage of ness in disease an impact, peak shock pressure depends on the Studying Aggregation treating (4). Although light material properties of the meteorite and target and scattering experiments are frequently used to on the impact velocity. We ran a range of simulations The recent flurry of research activity on study such problems, in these three exam- for impact of a stony meteorite into targets of dia- assemblies and their conventional would base, permafrost, carbonate, granite, and water supramolecular applica- ples light scattering [properties all given in (22)]. Peak pressures ranged tion in the construction of nanodevices has likely not yield useful data because of the from a minimum of 380 GPa (ice target) to 555 GPa background signals provided by the medium; (permafrost), to 657 GPa (granite), to 705 GPa (dia- R. F. Pasternack is in the Department of Chemistry and P. not the existence but the of base). In all cases, energy was partitioned about half J. Collings is in the Department of Physics and Astrono- only identity into the target and half into the meteorite during the my at Swarthmore College, 500 College Avenue, Swarth- the scatterer must be determined. compression stage. The ratio of penetration depth to more, PA 19081, USA. Even though light scattering experiments SCIENCE * VOL. 269 · 18 AUGUST 1995 935