Large Pt anomaly in the Greenland ice core points to
a cataclysm at the onset of Younger Dryas
Michail I. Petaev
, Shichun Huang
, Stein B. Jacobsen
, and Alan Zindler
Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138; and
Division of Solar, Stellar, and Planetary Sciences,
Harvard–Smithsonian Center for Astrophysics, Cambridge, MA 02138
Edited by Mark H. Thiemens, Univer sity of California, San Diego, La Jolla, CA, and approved June 26, 2013 (received for review February 28, 2013)
One explanation of the abrupt cooling episode known as the
Younger Dryas (YD) is a cosmic impact or airburst at the YD
boundary (YDB) that triggered cooling and resulted in other
calamities, including the disappearance of the Clovis culture and
the extinction of many large mammal species. We tested the YDB
impact hypothesis by analyzing ice samples from the Greenland Ice
Sheet Project 2 (GISP2) ice core across the Bølling-Allerød/YD
boundary for major and trace elements. We found a large Pt
anomaly at the YDB, not accompanied by a prominent Ir anomaly,
with the Pt/Ir ratios at the Pt peak exceeding those in known
terrestrial and extraterrestrial materials. Whereas the highly frac-
tionated Pt/Ir ratio rules out mantle or chondritic sources of the Pt
anomaly, it does not allow positive identiﬁcation of the source.
Circumstantial evidence such as very high, superchondritic Pt/Al
ratios associated with the Pt anomaly and its timing, different
from other major events recorded on the GISP2 ice core such as
well-understood sulfate spikes caused by volcanic activity and the
ammonium and nitrate spike due to the biomass destruction, hints
for an extraterrestrial source of Pt. Such a source could have been
a highly differentiated object like an Ir-poor iron meteorite that is
unlikely to result in an airburst or trigger wide wildﬁres proposed
by the YDB impact hypothesis.
he Younger Dryas (YD), a millennium-long cooling period
amid postglacial warming well documented in the Greenland
ice cores (e.g., refs. 1, 2), is thought to result from an abrupt
change in atmospheric and oceanic circulation (3). Whether such
a change was caused by a catastrophic event or it is an integral,
although still poorly understood, feature of the deglaciation
process remains unclear (4).
Among testable catastrophic hypotheses, the most popular,
attractive, and long-lasting idea of a sudden discharge of fresh
water from the proglacial Lake Agassiz into the Arctic Ocean
(5–7) eve ntually was found inconsisten t with geomorphological
and chronological data (4, 8). The long-term effect of the pro-
posed “volcanic winter” in the northern hemisphere induced by
the catastrophic eruption of the Laacher See volcano 12,916 cal-
endar years before 1950 (cal BP) (9) is not clear as the Laacher
See tephra, found in many European lacustrine deposits, is absent
in the Greenland ice cores (10). The impact hypothesis (11), once
declared dead (12, 13), recently gained new support from the
discovery of siliceous scoria-like objects (SLOs) with global dis-
tribution, which provide strong evidence for processing at high
temperatures and pressures consistent with a cosmic impact (14).
The ever-controversial impact hypothesis was initially invoked
to explain the disappearance of the Clovis culture and the ex-
tinction of many mammal species, including mammoths, by a
cometary airburst that resulted in massive wildﬁres and, ulti-
mately, the YD cooling. A C-rich layer exposed at many sites in
North America and Europe at or near the YD boundary (YDB
layer), which is enriched in magnetic grains with Ir, magnetic
microspherules, charcoal, soot, carbon spherules, glass-like car-
bon with nanodiamonds, and fullerenes with extraterrestrial
He (11), has been interpreted as documenting this event.
Ammonium and nitrate spikes at the onset of the YD in
Greenland Ice Core Project (GRIP) and Greenland Ice Sheet
Project 2 (GISP2) ice cores, perhaps resulting from biomass
burning, are taken as further support for the impact hypothesis.
Subsequent studies (13, 15, 16) questioned the origins of many
impact markers cited by ref. 11, but the discovery of the SLOs
alongside abundant, compositionally similar microspherules in
three YDB sites in North America and Asia is difﬁcult to explain
by anything other than a cosmic impact. In its latest incarnation,
the impact hypothesis calls for three or more epicenters of an
impact or airburst (14). However, the invoked markers have
never been supported by a clear geochemical impact signature
such as a sharp increase in Ir or other platinum group element
(PGE) concentrations at the YDB.
We have tested the YDB impact hypothesis by measuring
trace and major element concentrations in ice samples from the
GISP2 ice core across the Bølling-Allerød/YD boundary (depth
of 1,709–1,720 m, 12,279–13,064 cal BP) with a spatial resolution
of ∼12.5 cm, corresponding to a time resolution of 2.5–4.6 y (17).
The elemental concentrations in melted ice samples were mea-
sured by inductively coupled plasma mass spectrometry (ICP-MS)
that is known to have possible interferences of LuO and HfO
peaks with Ir and Pt peaks. This issue was resolved by measuring
Lu and Hf oxide formation in well-calibrated standards (details in
Materials and Methods).
The major ﬁnding of this study (Fig. 1 and Table S1) is the lack
of a striking Ir anomaly in the analyzed ice samples. Instead, we
found a large Pt anomaly in the middle of the Bølling-Allerød–
YD transition that is contemporaneous with a sharp drop in the
O values (Fig. 1B and Fig. S1). The Pt peak (Fig. 1) is unlikely
to be a mass-spectrometry artifact because (a) it is smoothly de-
ﬁned by seven ice samples; (b) the HfO interferences (
Pt isotopes, respectively, were
carefully assessed; (c)the
Hf signals in sample 63 with
the highest Pt concentration are at least a factor of 10 lower than
Pt; and (d) there is no linear correlation between
Hf concentrations and Pt concentrations in the ice samples.
The Pt concentrations gradually rise by at least 100-fold over
∼14 y and drop back during the subsequent ∼7 y. The decay of
the Pt signal is consistent with the ∼5-y lifetime of dust in the
stratosphere. The observed gradual ingrowth of the Pt concen-
tration in ice over ∼14 y may suggest multiple injections of
Pt-rich dust into the stratosphere that are expected to result in
a global Pt anomaly.
The Pt anomaly is accompanied by extremely high Pt/Ir and
Pt/Al ratios (Fig. 2), indicative of a highly unusual source of Pt in
the ice. Such a source is unlikely to be laboratory contamination
Author contributions: M.I.P., S.H., S.B.J., and A.Z. designed research, performed research,
analyzed data, and wrote the paper.
The authors declare no conﬂict of interest.
This article is a PNAS Direct Submission.
To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
www.pnas.org/cgi/doi/10.1073/pnas.1303924110 PNAS Early Edition
AND PLANETARY SCIENCES
because (a) all samples deﬁning the anomaly are from the same
ice core samples (Fig. S2) and were collected using the same set
of tools and latex gloves, (b) three peak samples are from a single
continuous chunk, and (c) the samples were randomly dissolved
and analyzed. Contamination during ice coring and subsequent
slicing is also unlikely because PGEs are essentially insoluble in
pure acids, let alone in water (Materials and Methods).
Materials with high Pt/Ir ratios and essentially no Al are
known among magmatic iron meteorites (20, 21). Finding a ter-
restrial Pt-rich and Ir-, Al-poor source is difﬁcult. Most volcanic
rocks have elevated Pt/Ir ratios, although not as high as in iron
meteorites, but Pt/Al ratios are very low (e.g., refs. 22, 23).
Mantle rocks are depleted in Al, but have essentially unfractio-
nated Pt and Ir (24). The only known terrestrial material with Pt/Ir
ratios comparable to those in iron meteorites is the Pt-rich sul-
ﬁdes from the Sudbury Footwall (25). However, both terrestrial
and extraterrestrial high-Pt sources have substantially lower Pt/Ir
ratios than those at the top of the Pt peak, implying either Pt-Ir
fractionation during atmospheric processing of the Pt-rich mate-
rials or multiple injections of materials with different Pt/Ir ratios
not sampled so far.
Thus, the highly fractionated Pt/Ir ratio rules out mantle or chon-
dritic sources of the Pt anomaly (Fig. 2). A further discrimination
between Pt-rich crustal materials like Sudbury Footwall ore (25)
and fractionated extraterrestrial sources such as Ir-poor iron
meteorites like Sikhote-Alin (26) is difﬁcult because of the com-
parable magnitude of the Pt/Ir fractionation in these materials.
Circumstantial evidence hints at a n extraterrestrial source of
Pt, such as very high, superchondritic Pt/Al ratios at the Pt
anomaly and its timing, which is clearly different from other
major events recorded in the GISP2 ice core, i ncluding well-
understood sulfate spikes caused by volcanic activity and the
ammonium and nitrate spikes associated with biomass destruction
Until the question about the nature of Pt-rich material and the
means of its delivery to the ice is resolved, an extraterrestrial
source of Pt appears likely. For example, the Pt anomaly could be
explained by multiple impacts of an iron meteorite like Sikhote-
Alin or Grant (21, 26); the former is a large crater-forming
meteorite s hower. Assuming a global anomaly, the 62.5-cm-
thick ice layer with the average Pt concentration of 30 parts
per trillion (ppt) (Fig. 1) would require an iron meteorite like
Sikhote-Alin of ∼0.8 km in diameter to account for the Pt budget
at the YDB. Because complete disintegration of such a large iron
meteorite during its atmospheric passage seems unlikely, the
event is expected to form a crater of a few kilometers in di-
ameter. No such crater at YDB has been found so far.
The main conclusion of our study is the detection of an un-
usual event during th e Bøl ling-Allerød –YD transition period
that resulted in deposition of a large amount of Pt to the
Greenland ice. The Pt anomaly precedes the ammonium and
nitrate spike in the GISP2 ice core (2) by ∼30 y and, thus, this
event is unlikely to have triggered the biomass burning and de-
struction thought to be responsible for ammonium increase in
the atmosphere and the Greenland ice (11). Although the data
do not allow an unambiguous identi ﬁcation of the Pt source, they
clearly rule out a chondritic origin of Pt. One of the plausible
sources of the Pt spike is a metal impactor with an unusual
Younger Dryas Bolling-Allerod
12800 12850 12900 12950 13000 13050
Age, cal BP
O, per mil
Transit on periodi
Fig. 1. (A and B)Ir(A) and Pt (B) variations in the GISP2 ice samples across
the Bølling-Allerød/YD boundary. Open symbols indicate upper limits. Dur-
ing the transitional period the annual rate of ice accumulation changes from
high to low value s (1). The Ir contents in the ice (A) show some variations
with no clear substantial anomalies. On the contrary, a large Pt anomaly
(B) occurs in the middle of the Bølling-Alle rød –YD transition period. The
anomaly coincides with a sharp drop in the δ
O values in ice (black curve).
O curve is calculated from the data of ref. 18 averaged for 40 cm of
ice thickness. (C) The time of the Pt anomaly differs from those of the vol-
spikes (19) and the Laacher See eruption (LSE) (9) and precedes
the onset of the NH
spike (2) by ∼30 y.
12800 12850 12900 12950 13000 13050
Age, cal BP
Younger Dryas Bolling-Allerod
Cont nental crusti
iTransit on periodi
Cont nental crusti
Fig. 2. (A and B) Pt/Ir (A) and Pt/Al (B) ratios in the GISP2 ice samples across
the Bølling-Allerød/YD boundary vary between the chondritic and conti-
nental crust values. The only exception is the Pt anomaly with both Pt/Ir and
Pt/Al ratios greatly exceeding chondritic and crustal values, with three top
points also exceeding Pt/Ir ratios of low-Ir iron meteorites. Such a large
fractionation of both Pt/Ir and Pt/Al ratios within the anomaly most likely
results from atmospheric processing of the Pt-rich material.
www.pnas.org/cgi/doi/10.1073/pnas.1303924110 Petaev et al.
composition derived from a highly fractionated portion of a
Materials and Methods
Sample Preparation. The 11 ice core pieces (∼2cm× 3cm× 1 m each) pro-
vided by the National Ice Core Laboratory were processed in our clean lab-
oratory at room temperature. The ice “sticks” packed in plastic sleeves
arrived partially broken into smaller pieces. Before sampling, each stick was
carefully aligned in the original sleeve and then split into eight samples,
each ∼12.5 cm long, using a metal bar wrapped ﬁrst in aluminum foil and
then in plastic wrap. Subsequently, each sample was extracted from the
sleeve, carefully washed with ultrapure deionized (DI) water to remove
possible surface contamination, and placed in a tall 300-mL pol ypropylene
container, precleaned in 1:1 HCl and then 1:1 HNO
for several days. Small
ice fragments and water left in a sleeve were composited and used to de-
velop and test analytical protocols. “Composites” from several sleeves were
combined to get total weight comparable to that of the real ice samples. In
total, 88 ice samples (samples 1–88 in Table S1), ranging from 38.42 g to
68.90 g, and 5 composite samples (samples 89–93 in Table S1) were collected
and processed. The empty and ﬁlled polypropylene containers were
weighed with a precision of ±0.01 g.
Ice samples were melted at room temperature in closed containers and the
water was slowly evaporated at 80–90 °C on a hot plate down to 1–2 mL. The
remaining samples were carefully transferred into precleaned (boiled in 1:1
HCl and then 1:1 HNO
) 6-mL, perﬂuoralkoxy (PFA) teﬂon beakers. Each
polypropylene container was then washed at least twice with 1–2mLul-
trapure DI water, and the washes were collected and added to the corre-
sponding PFA beakers.
Sample Dissolution. We attempted to measure the total Ir and Pt concen-
trations in ice, including all solids cemented in ice. Consequently, we applied
a multiple-step dissolution method as documented below.
The 93 samples along with 2 procedure blanks were (i) dried down on
a hot plate, (ii) sealed with 1 mL of a 1:1 mixture of concentrated HF and
and heated at 150 °C in an oven for 1 wk, ( iii ) dried
down again, (iv) redissolved in 0.4 mL of aqua regia (one part concentrated
and three parts concentrated HCl), and (v) ﬁnally diluted with 1 mL of
ultrapure DI water to yield analytical solutions of ∼1.45 g each. Both empty
and ﬁlled PFA beakers were weighed with a precision of ±0.0001 g.
step is aimed at breaking down the silicates, and the aqua
regia step is aimed at transferring all PGEs into solution because it is well
known that Ir and Pt are not stable in diluted HNO
(e.g., ref. 27). The latter
has been conﬁrmed by our tests using the composite samples. Our approach
differs from previous studies such as refs. 28–30, in which ice samples were
melted, preconcentrated by subboiling point evaporation, and added con-
to form 1% HNO
analytical solutions for ICP-MS mea-
surements. Such a treatment is unable to transfer all Ir and Pt into solution.
Our approach is also different from that of ref. 31, which measured PGE
concentrations only in particles in the size range 0.45–20 μm. This explains
why our measurements yield much higher Ir a nd Pt concentrations in ice
compared with those in refs. 28 and 31.
ICP-MS Measurements. All ice samples, in diluted aqua regia matrix, were an-
alyzed in two analytical sessions, using a GV Instruments Platform ICP-MS, ﬁrst
for Ir, Pt, Lu, and Hf with an Apex inlet system to increase sensitivity and reduce
oxide interferences and then for major elements with a normal glass spray
chamber. For Ir-Pt measurement, peaks of
176, 177, 178, 179, 180
Pt were monitored. Because of the special memory effects of
Ir and Pt, diluted aqua regia (2.5 mL concentrated HCl + 7.5 mL concentrated
+ 90 mL H
O) was used as a wash solution. Instrumental blank was also
measured for such a solution. Oxide formation rates of LuO/Lu and HfO/Hf
were determined for a 10 parts per billion (ppb) Lu-Hf standard solution pre-
pared from HPS 1,000-ppm single-element standards before and after ana-
lytical sessions. During all measurements, LuO/Lu was <0.01 and HfO/Hf < 0.05.
Ratios of LuO/Lu (<0.01) and HfO/Hf (<0.05), measured in a Lu-Hf solution
together with ice samples, were used to correct for oxide interferences
respectively. The oxide corr ections were ty pically <5atomic%.Afteroxide
interference correction, the measured Ir and Pt isotopic compositions of ice
samples are within 3% of t hose meas ured on the Ir-Pt standard solution,
showing that our Ir-Pt measurements were not affected by LuO and
HfO interferences .
During our Ir-Pt measurements, the ICP-MS was tuned with a 10-ppb In
solution, and this solution was analyzed every ﬁve samples to monitor the
instrumental sensitivity drift, which was less than 10% during any of our
analytical sessions. A 10-ppb Ir-Pt standard solution, prepared from HPS
1,000-ppm single-element standards, was analyzed at the end of an analytical
session to convert measur ed signal into concentrations. This is specially
designed because of the long memory effects of Ir. For Ir, concentrations
Ir (after oxide corrections) are within 3%, and an
average is used. For Pt, although peak
Pt is free of oxide interference, it
suffers isobaric interference from
Hg. Consequently, we used only
Pt peaks (after oxide corrections). Concentrations obtained using
Pt peaks are within 3% of each other, and an average is used. All
ice samples were analyzed at least twice in a random order, with the aver-
age value s reported in Table S1.
For major element measurements (Al), a Basalt, Hawaiian Volcanic Ob-
servatory (BHVO)-1 standard solution was measured every ﬁve to six samples
to monitor the instrumental sensitivity drift, whi ch was less than 10% during
any of our analytical sessions. This solution was also used for calibration. All
ice samples were analyzed at least twice in random order, with the average
values reported in Table S1.
ACKNOWLEDGMENTS. We thank Prof. Wally Broecker for encouraging this
study, Dr. Mark S . Twickler for providing the GISP2 ice s amples, and Dr.
Robert P. Acker t for inspiring discussions. W e also thank H. J. Melo sh and
two anonymous reviewers for helpful r eviews. T his work was supported by
National Science Foundation Grant AGS-1007367 (to S.B.J.).
1. Alley RB, et al. (1993) Abrupt increase in Greenland snow accumulation at the end of
the Younger Dryas event. Nature 362:527–529.
2. Mayewski PA, et al. (1993) The atmosphere during the younger dryas. Science
3. Mayewski PA, et al. ( 1994) Changes in atmospheric circulation and ocean ice
cover over the North Atlantic during the last 41,000 years. Science 263(515 4):
4. Broecker WS, et al. (2010) Putting the Younger Dryas cold event into context. Quat Sci
5. Rooth C (1982) Hydrology and ocean circulation. Prog Oceanogr 11(2):131–149.
6. Broecker WS, et al. (1989) Routing of meltwater from the Laurentide Ice Sheet during
the Younger Dryas cold episode. Nature 341:318–321.
7. Broecker WS (2006) Geology. Was the Younger Dryas triggered by a ﬂood? Science
8. Lowell TV, et al. (2005) Testing the Lake Agassiz meltwater trigger for the Younger
Dryas. EOS Trans 86:365–373.
9. Baales M, et al. (2002) Impact of the Late Glacial Eruption of the Laacher See Volcano,
Central Rhineland, Germany. Quat Res 58:273–288.
10. Svensson A (2012) The missing tephra horizons in the Greenland ice cores. Quat Int
11. Firestone RB, et al. (2007) Evidence for an extraterrestrial impact 12,900 years ago
that contributed to the megafaunal extinctions and the Younger Dryas cooling. Proc
Natl Acad Sci USA 104(41):16016–16021.
12. Kerr RA (2010) Mammoth-killer impact ﬂunks out. Science 329(5996):1140–1141.
13. Pinter N, et al. (2011) The Younger Dryas impact hypothesis: A requiem. Earth Sci Rev
14. Bunch TE, et al. (2012) Very high-temperature impact melt products as evidence for
cosmic airbursts and impacts 12,900 years ago. Proc Natl Acad Sci USA 109(28):
15. van Hoesel A, et al. (2012) Nanodiamonds and wildﬁre evidence in the Usselo horizon
postdate the Allerod-Youn ger Dryas boundary. Proc Natl Acad Sci USA 109(20):7648–7653.
16. Pigati JS, et al. (2012) Accumulation of impact markers in desert wetlands and im-
plications for the Younger Dryas impact hypothesis. Proc Natl Acad Sci USA 109(19):
17. Meese DA, et al. (1997) The Greenland Ice Sheet Project 2 depth-age scale: Methods
and results. J Geophys Res 102:26411–26423.
18. Stuiver M, Grootes PM (2000) GISP2 oxygen isotope ratios. Quat Res 53:277–283.
19. Zielinski GA, Mayewski PA, Meeker LD, Whitlow S, Twickler MS (1996) A 110,000-yrrecord
of explosive volcanism from the GISP2 (Greenland) ice core. Quat Res 45(2):109–118.
20. Wasson JT (1999) Trapped melt in IIIAB irons; solid/liquid elemental partitioning during
the fractionation of the IIIAB magma. Geochim Cosmochim Acta 63:2875–2889.
21. Wasson JT, Huber H, Malvin DJ (2007) Formation of IIAB iron meteorites. Geochim
Cosmochim Acta 71:760–781.
22. Day JMD (2013) Hotspot volcanism and highly siderophile elements. Chem Geol 341:
23. Zhang Z, Mao J, Mahoney JJ, Wang F, Qu W (2005) Platinum group elements in the
Emeishan large igneous province, SW China: Implications for mantle sources. Geo-
chem J 39:371–382.
24. Maier WD, et al. (2012) The concentration of platinum-group elements and gold in
southern African and Karelian kimberlite-hosted mantle xenoliths: Implications for
the noble metal content of the Earth’s mantle. Chem Geol 302/303:119–135.
25. Mungall JE, Naldrett AJ (2008) Ore deposits of the platinum-group elements. Ele-
Petaev et al. PNAS Early Edition
AND PLANETARY SCIENCES
26. Petaev MI, Jacobsen SB (2004) Differentiation of metal-rich meteoritic parent bodies:
I. Measurements of PGEs, Re, Mo, W, and Au in meteoritic Fe-Ni metal. Meteorit
Planet Sci 39:1685–1697.
27. Greenwood NN, Earnshaw A (1985) Chemistry of the Elements (Pergamon, Oxford),
28. Gabrielli P, et al. (2004) Meteoric smoke fallout over the Holocene epoch revealed by
iridium and platinum in Greenland ice. Nature 432(7020):1011–1014.
29. Gabrielli P, et al. (2006) A climatic control on the accretion of meteoric and super-
chondritic iridium-platinum to the Antarctic ice cap. Earth Planet Sci Lett 250:459–469.
30. Soyol-Erdene TO, Huh Y, Hong S, Hur SD (2011) A 50-year record of platinum, iridium,
and rhodium in Antarctic snow: Volcanic and anthropogenic sources. Environ Sci
31. Karner DB, e t al. (2003) Extraterrestrial accretion from the GISP2 ice core. Geochim
Cosmochim Acta 67:751–763.
www.pnas.org/cgi/doi/10.1073/pnas.1303924110 Petaev et al.
Petaev et al. 10.1073/pnas.1303924110
Fig. S1. The Pt anomaly superimposed on the Greenland Ice Sheet Project 2 (GISP2) oxygen-isotope record (1) of the period around the Allerød/Younger Dryas
boundary. The temperature scale is adapted from refs. 2 and 3, and the ﬁgure is modiﬁed after ref. 4.
Fig. S2. Photomosaic of the ice “stick” F containing anomalously Pt-rich samples identiﬁed by green numbers. The peak Pt concentration is in sample 63. Red
dashes mark sample boundaries; some look nonparallel due to the tilting and displacement of individual images. Green arrows show breaks in the stick. The
original markings and labels are in black.
1. Stuiver M, Grootes PM (2000) GISP2 oxygen isotope ratios. Quat Res 53:277–283.
2. Cuffey KM, et al. (1994) Calibration of the δ
O isotopic paleothermometer for central Greenland, using borehole temperatures. J Glaciol 40:341–349.
3. Cuffey KM, Clow JD (1997) Temperature, accumulation, and ice sheet elevation in central Greenland through the last deglacial transition. JGR 102:26383–26396.
4. Denton GH, et al. (2005) The role of seasonality in abrupt climate change. Quat Sci Rev 24:1159–1182.
Petaev et al. www.pnas.org/cgi/content/short/1303924110 1of3
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720
Ice depth, m
Pt concentrat on in ice, ppti
Fig. S3. Comparison of Pt concentrations in the real (2–88) and composite (89–93) ice samples. Each sample is plotted in the middle of the sampling interval.
The symbol size of real samples corresponds to the sampling depth range. The sampling range of composite samples is shown by horizontal bars. Note that the
real samples were precleaned (Materials and Methods) to remove possible surface contamination, whereas the composite samples were not precleaned. Also,
real and composite samples were analyzed in different analytical sessions. Therefore, a good overall consistency in Pt concentrations between the real and
composite samples provides strong evidence against both external and internal (our laboratory) contamination.
Petaev et al. www.pnas.org/cgi/content/short/1303924110 2of3
Table S1. Concentrations of Ir, Pt, Lu, Hf, and Al in the GISP2 ice core
Sample no. Top Bottom Age,* cal BP Ir, ppt Pt, ppt Lu, ppt Hf, ppt Al, ppb
2 1,719.875 1,719.750 13,059.7 0.022 0.065 0.112 0.361 6.03
10 1,718.875 1,718.750 13,036.5 0.060 0.162 0.189 1.74 19.5
18 1,717.875 1,717.750 13,013.3 0.024 0.065 0.040 0.692 14.9
26 1,716.875 1,716.750 12,993.1 0.038 0.563 0.058 0.596 10.7
34 1,715.875 1,715.750 12,972.1 0.025 0.081 0.067 0.548 6.27
42 1,714.875 1,714.750 12,951.3 0.029 0.038 0.079 0.641 9.98
49 1,714.000 1,713.875 12,933.4 0.025 1.54 0.045 0.838 8.35
50 1,713.875 1,713.750 12,930.3 0.027 0.469 0.033 0.292 8.52
51 1,713.750 1,713.625 12,927.2 0.025 1.63 0.047 0.278 4.84
52 1,713.625 1,713.500 12,924.1 0.027 0.391 0.077 0.406 4.55
53 1,713.500 1,713.375 12,920.9 0.029 0.158 0.032 0.280 5.62
54 1,713.375 1,713.250 12,917.8 0.047 0.447 0.078 0.623 3.02
55 1,713.250 1,713.125 12,914.7 0.031 0.078 0.052 0.202 6.11
56 1,713.125 1,713.000 12,911.6 0.025 1.05 0.040 0.321 11.7
57 1,713.000 1,712.875 12,908.3 0.032 1.67 0.084 0.659 23.2
58 1,712.875 1,712.750 12,904.8 0.035 1.79 0.094 0.596 13.5
59 1,712.750 1,712.625 12,901.3 0.034 1.29 0.056 0.449 16.2
60 1,712.625 1,712.500 12,897.8 0.034 7.76 0.041 0.214 7.59
61 1,712.500 1,712.375 12,894.3 0.035 3.41 0.044 0.559 10.9
62 1,712.375 1,712.250 12,890.8 0.045 27.6 0.157 2.05 12.9
63 1,712.250 1,712.125 12,887.3 0.065 82.2 0.166 1.67 8.18
64 1,712.125 1,712.000 12,883.8 0.081 24.0 0.160 1.80 13.1
65 1,712.000 1,711.875 12,879.9 0.058 0.785 0.118 1.22 19.8
66 1,711.875 1,711.750 12,875.8 0.022 0.132 0.061 0.338 7.29
67 1,711.750 1,711.625 12,871.7 0.038 0.609 0.127 0.864 15.1
68 1,711.625 1,711.500 12,867.6 0.047 0.281 0.154 1.52 14.1
69 1,711.500 1,711.375 12,863.4 0.057 0.227 0.192 1.81 8.27
70 1,711.375 1,711.250 12,859.3 0.031 1.43 0.063 0.353 7.66
71 1,711.250 1,711.125 12,855.2 0.063 0.472 0.136 0.986 12.9
72 1,711.125 1,711.000 12,851.1 0.031 0.474 0.049 0.562 5.64
73 1,711.000 1,710.875 12,846.9 0.031 0.287 0.094 0.773 14.5
74 1,710.875 1,710.750 12,842.8 0.094 0.198 0.230 1.70 22.9
75 1,710.750 1,710.625 12,838.7 0.063 1.66 0.209 1.92 16.7
76 1,710.625 1,710.500 12,834.6 0.062 0.179 0.210 1.48 12.0
77 1,710.500 1,710.375 12,830.4 0.050 0.507 0.149 1.55 13.6
78 1,710.375 1,710.250 12,826.3 0.102 0.134 0.183 1.28 10.8
79 1,710.250 1,710.125 12,822.2 0.103 0.272 0.261 2.70 14.3
80 1,710.125 1,710.000 12,818.1 0.054 0.266 0.190 1.54 14.6
81 1,710.000 1,709.875 12,813.7 0.031 0.149 0.161 0.960 52.3
82 1,709.875 1,709.750 12,809.1 0.046 0.095 0.212 1.44 23.8
83 1,709.750 1,709.625 12,804.4 0.034 0.096 0.209 1.26 34.6
84 1,709.625 1,709.500 12,799.8 0.032 0.118 0.152 0.90 18.9
85 1,709.500 1,709.375 12,795.2 0.051 0.571 0.388 3.21 33.3
86 1,709.375 1,709.250 12,790.6 0.037 0.276 0.151 0.87 26.4
87 1,709.250 1,709.125 12,785.9 0.137 0.493 0.259 1.87 59.1
88 1,709.125 1,709.000 12,781.3 0.051 0.110 0.192 1.43 16.4
1,720 1,716 0.015 0.165
1,716 1,714 0.018 0.204
1,714 1,712 0.115 32.3
1,712 1,710 0.504 1.72
1,710 1,709 0.430 1.98
0.031 0.315 0.003 0.027
Cal BP, calendar years before 1950; ppb, parts per billion; ppt, parts per trillion.
*Age of each sample is calculated for the middle of sampling depth interval by interpolation of tabulated age-
time data from the GISP2 Meese/Sowers Timescale (ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/ greenland/
Composite samples, plotted in Fig. S3.
Sample 95 is the measurement of the entire procedural blank. For any sample with similar or lower reported
concentrations, the reported values represent only upper limits of the true concentrations. The mass-spectro-
metric detection limits are much lower than the reported procedur al blank. The detection limit for Ir and Pt in
the samples is 0.004 ppt.
Petaev et al. www.pnas.org/cgi/content/short/1303924110 3of3