Archean molecular fossils and the early rise of eukaryotes. Science

Abstract

Molecular fossils of biological lipids are preserved in 2700-million-year-old shales from the Pilbara Craton, Australia. Sequential extraction of adjacent samples shows that these hydrocarbon biomarkers are indigenous and syngenetic to the Archean shales, greatly extending the known geological range of such molecules. The presence of abundant 2alpha-methylhopanes, which are characteristic of cyanobacteria, indicates that oxygenic photosynthesis evolved well before the atmosphere became oxidizing. The presence of steranes, particularly cholestane and its 28- to 30-carbon analogs, provides persuasive evidence for the existence of eukaryotes 500 million to 1 billion years before the extant fossil record indicates that the lineage arose.

Full text

DOI: 10.1126/science.285.5430.1033
, 1033 (1999); 285Science et al.Jochen J. Brocks,
Eukaryotes
Archean Molecular Fossils and the Early Rise of
www.sciencemag.org (this information is current as of September 1, 2009 ):
The following resources related to this article are available online at
http://www.sciencemag.org/cgi/content/full/285/5430/1033
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/cgi/content/full/285/5430/1033#otherarticles
, 6 of which can be accessed for free: cites 12 articlesThis article
367 article(s) on the ISI Web of Science. cited byThis article has been
http://www.sciencemag.org/cgi/content/full/285/5430/1033#otherarticles
72 articles hosted by HighWire Press; see: cited byThis article has been
http://www.sciencemag.org/cgi/collection/paleo
Paleontology : subject collectionsThis article appears in the following
http://www.sciencemag.org/about/permissions.dtl
in whole or in part can be found at: this article permission to reproduce of this article or about obtaining reprintsInformation about obtaining
registered trademark of AAAS. is aScience1999 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on September 1, 2009 www.sciencemag.orgDownloaded from

Archean Molecular Fossils and
the Early Rise of Eukaryotes
Jochen J. Brocks,
1,2
* Graham A. Logan,
2
Roger Buick,
1
Roger E. Summons
2
Molecular fossils of biological lipids are preserved in 2700-million-year-old
shales from the Pilbara Craton, Australia. Sequential extraction of adjacent
samples shows that these hydrocarbon biomarkers are indigenous and synge-
netic to the Archean shales, greatly extending the known geological range of
such molecules. The presence of abundant 2␣-methylhopanes, which are char-
acteristic of cyanobacteria, indicates that oxygenic photosynthesis evolved well
before the atmosphere became oxidizing. The presence of steranes, particularly
cholestane and its 28- to 30-carbon analogs, provides persuasive evidence for
the existence of eukaryotes 500 million to 1 billion years before the extant fossil
record indicates that the lineage arose.
Microfossils (1), stromatolites (2), and sedi-
mentary carbon isotope ratios (3) all indicate
that microbial organisms inhabited the oceans
in Archean times [⬎2500 million years ago
(Ma)]. But these lines of evidence are not very
informative about what these microbes were or
how they lived. Potentially, a better insight into
primordial biological diversity can be obtained
from molecular fossils derived from cellular
and membrane lipids (“biomarkers”). Although
such soluble hydrocarbons were first extracted
from Archean rocks more than 30 years ago,
their significance was generally discounted af-
ter amino acids of recent origin were found in
the same rocks (4). Prevailing models of ther-
mal maturation dictated that complex hydrocar-
bons should not survive the metamorphism ex-
perienced by all Archean terrains. However,
indications of greater hydrocarbon stability (5)
and observations of oil in Archean fluid inclu-
sions (6) suggest that these maturation models
are unduly pessimistic and that biomarkers
could indeed be preserved in low-grade Arche-
an metasedimentary rocks. Furthermore, sys-
tematic sampling strategies, improved analyti-
cal techniques, and greater geochemical knowl-
edge (7) should make their recognition easier
and their interpretation more rigorous. We now
report molecular fossils in late Archean shales
that have suffered only minimal metamor-
phism. These molecular fossils reveal that the
Archean biota was considerably more complex
than currently recognized and that the domains
Eucarya and Bacteria were already extant.
Samples came from depths of around
700 m in diamond drill core WRL#1, collared
near Wittenoom in the Pilbara Craton of north-
western Australia (8). They represent the
⬃2600-Ma Marra Mamba Formation (9)
(lowermost Hamersley Group), the underly-
ing Roy Hill and Warrie Members of the
⬃2690-Ma Jeerinah Formation (10), and the
⬃2715-Ma Maddina Formation (Fortescue
Group) (9), all of which have only been
metamorphosed to prehnite-pumpellyite fa-
cies in this area (11)(⬃200° to ⬃300°C).
The closely spaced sampling of different
lithologies allowed comparison of rocks with
varying compositions, porosities, and kero-
gen contents but with identical postdeposi-
tional histories. Most analyses were of finely
laminated kerogenous shales from the Roy
Hill Shale and Marra Mamba Formation that
were deposited in a marine continental-slope
environment below storm-wave base under
suboxic conditions (12). Those from the
Marra Mamba Formation were interbedded
with oxide-facies banded iron formation.
Black chert from the Warrie Member, quartz
sandstone and vein dolomite from the Roy
Hill Member, and terrestrial basalt from the
Maddina Formation served as controls for
laboratory procedures.
Although these rocks are well-preserved
by Archean standards, they are nevertheless
1
School of Geosciences, University of Sydney, Sydney,
NSW 2006, Australia.
2
Australian Geological Survey
Organisation (AGSO), Canberra, ACT 2601, Australia.
*To whom correspondence should be addressed. E-
mail: jochen.brocks@agso.gov.au or brocks@es.su.oz.au
Fig. 1. Drill core WRL#1 showing
the gas chromatograms of extract-
able hydrocarbons at different
depths (in meters). The kerogen-
poor samples on the left (white
squares on drill core) yielded low
quantities of extractable organic
matter, whereas the kerogen-rich
shales on the right (black squares
on drill core) yielded relatively
high concentrations.
RESEARCH ARTICLE
www.sciencemag.org SCIENCE VOL 285 13 AUGUST 1999 1033
on September 1, 2009 www.sciencemag.orgDownloaded from

highly mature and therefore contain only
small quantities of extractable hydrocarbons.
Hence, they are especially prone to modern
contamination by petroleum products during
drilling, storage, and analysis. Accordingly,
each sample underwent three solvent rinses to
ensure that extracted bitumen was intimately
associated with the rock and not superficial or
lining cracks. After each rinse was analyzed
chromatographically, the rock was broken
into smaller fragments to expose fresh bed-
ding surfaces, fractures, and fissures. If the
last rinse proved clean, the sample was then
finely ground and exhaustively extracted.
The kerogenous shales yielded ⬎25 ␮gof
saturated hydrocarbons per gram of rock, 50
to 500 times more than the concentrations
obtained from the chert and basalt and 50,000
times greater than laboratory system blanks
(Table 1). The low hydrocarbon contents of
the sandstone and vein dolomite represent
traces of oil expelled from directly adjacent
shales. Because modern petrochemical con-
tamination should produce similar hydrocar-
bon distributions in all samples, the marked
variations in these (Fig. 1) indicate that the
extracted hydrocarbons are indeed indigenous.
There are three potential sources of an-
cient but post-Archean organic contamination:
local subsurface biological activity, ground-
water carrying biolipids from the surface, and
migrating petroleum carrying geolipids. The
first two are easily excluded. Hydrous pyrol-
ysis at 350°C of kerogens isolated from the
shales Roy3 and Mam1 yielded insignificant
quantities of hydrocarbon, as expected from
highly mature Precambrian organic matter
(13). Hydrocarbons extracted from these
kerogens are also highly mature, based on the
saturate and aromatic biomarker parameters
conventionally used to assess thermal evolu-
tion (Table 2). This indicates that they were
generated during the final stages of kerogen
maturation, before or during peak metamor-
phism at 2450 to 2000 Ma (14). Because the
rocks have not experienced a subsequent
thermal event capable of turning immature
biolipids into highly mature geolipids, the ex-
tracted hydrocarbons were not derived from
post-Archean surface or subsurface biota.
Adulteration from migrating Phanerozoic or
Proterozoic petroleum is harder to discount.
The youngest rocks in the depositional basin
belong to the ⬃2400-Ma Turee Creek Group.
Younger petroleum-prone rocks were never de-
posited over the top of the basin in sufficient
thickness for hydrocarbon expulsion to occur.
The WRL#1 drill hole is located centrally with-
in the craton, at least 150 km from the nearest
post-Archean basin, so younger oils migrating
laterally would have to travel long distances
through deformed and metamorphosed Arche-
an rocks without continuous cross-cutting frac-
tures. Thus, the only likely source for adulter-
ating oil is from older, rather than younger,
rocks. Moreover, bedding-parallel permeability
of the black shale Roy3 and pyritic sandstone
Roy4 was below the detection limit of 0.01
millidarcy. Thus, these rocks might have been
sealed to hydrocarbon migration since they
were last metamorphosed at 2450 to 2000 Ma.
Also, higher yields of bitumen from shales than
from kerogen-poor rocks (Table 1, Fig. 1) are
inconsistent with staining by migrated oil,
which would have equally tainted all rocks of
similar permeability. Notably, typical Phanero-
zoic geochemical characters, such as biomark-
ers derived from higher plants, are absent.
However, isoalkanes are abundant, as in many
Proterozoic bitumens (15,16), and the n-alkane
distribution has a diminution in relative abun-
Table 1. Kerogen and bitumen contents of WRL#1 samples. TOC, total organic carbon. n.m., not
measured.
Sample AGSO
number
Depth
(m) Lithology TOC
(%)
Saturates
(␮g/g of rock)
Mam1 9983 665.39 Sideritic brown shale 2.6 28
Roy1 9710 8284 679.39 Pyritic black shale 7.4 36
Roy2 9710 8285 681.90 Pyritic black shale 8.6 26
Roy5 1999 0333 696.92*Pyritic black shale n.m. ⬎25
Roy4 1999 0331 696.94 Pyritic quartz sandstone – 8
Roy6 1999 0332 709.59 Fibrous vein dolomite – 5
Roy3 9984 718.23 Pyritic black shale 9.0 27
War1 9982 777.86 Massive black chert 0.5 0.6
Mad1 9985 780.65 Amygdaloidal basalt – 0.07
Blank – – – – 0.0004†
*A 2-mm-thick layer on top of Roy4. †Referring to 100 g of rock.
Table 2. Abundances and isotopic compositions of kerogens and individual compounds. Biomarker ratios
were measured according to published methods (7, 15). Numbers in subscript refer to the number of
carbon atoms in the biomarker.
Mam1 Roy1 Roy2 Roy5 Roy3
Depth (m) 665.4 679.4 681.9 696.9 718.2
Acyclic hydrocarbons*
Pristane/phytane 1.2 (0.1)†1.3 1.3 1.0 1.3
Pristane/n-C
17
0.2 (0.1) 0.3 0.4 0.5 0.3
Phytane/n-C
18
0.1 (0.1) 0.2 0.3 0.5 0.2
Hopanes‡(ppm)§230 300 75 – 110
C
27
Ts/(C
27
Ts⫹C
27
Tm)㛳0.59 (0.01) 0.57 0.59 0.56 0.58
C
29
Ts/C
29
㛳0.23 (0.04) 0.24 0.16 0.20 0.25
(C
27
Ts⫹C
27
Ts)/C
29
㛳,¶ 1.3 (0.40) 1.1 0.57 1.0 1.82
C
29
/C
30
¶1.3 (0.20) 1.3 2.1 1.4 0.9
C
30
/C
31
¶1.1 (0.16) 1.4 1.5 1.3 1.5
2␣-Methylhopane index ¶,# 14 (1.4) 12 17 12 12
Steranes/Hopanes 㛳,¶,** 0.79 (0.20) 0.95 2.6 0.61 1.1
Steranes (ppm)§180 280 190 – 120
Dia- /Regular 㛳,¶,†† 0.58 (0.10) 0.91 1.33 0.77 0.87
C
29
20S/(20S⫹20R)㛳0.53 (0.04) 0.64 0.77 0.63 0.57
C
29
␣␤␤/(␣␣␣⫹␣␤␤)㛳0.54 (0.02) 0.58 0.63 0.61 0.57
C
27
/(C
27
⫹C
28
⫹C
29
)¶0.52 (0.03) 0.54 0.57 0.50 0.55
C
28
/(C
27
⫹C
28
⫹C
29
)¶0.21 (0.04) 0.19 0.20 0.20 0.20
C
29
/(C
27
⫹C
28
⫹C
29
)¶0.28 (0.04) 0.28 0.23 0.30 0.25
TA-I/(TA-I⫹TA-II) 㛳,‡‡ 0.92 0.94 0.96 ⬎0.90 0.96
Carbon isotopes (␦
13
C per mil PDB)
Kerogen ⫺46.8§§ ⫺39.5 ⫺39.6 – ⫺41.1
n-C
17
⫺26.2㛳㛳 ⫺26.2 ⫺26.0 – ⫺27.0
n-C
18
⫺26.5 ⫺26.1 ⫺27.4 – ⫺26.3
Pristane ⫺28.8 ⫺29.4 ⫺29.8 – ⫺28.9
Phytane ⫺29.6 ⫺29.5 ⫺30.1 – ⫺27.8
*Ratios measured using GC-FID peak areas. †The value in parentheses is the maximum error of the particular ratio
obtained in a series of repeat measurements. ‡C
27
Ts ⫽18␣(H)-22,29,30-trisnorneohopane; C
27
Tm ⫽17␣(H)-
22,29,30-trisnorhopane; C
29
Ts ⫽18␣(H)-30-norneohopane; C
29
⫽17␣,21␤(H)-30-norhopane; C
30
⫽17␣,21␤(H)-
hopane; C
31
⫽22S ⫹22R 17␤,21␣(H)-29-homohopane; 2␣MeC
31
⫽2␣-methyl-17␣,21␤(H)-hopane. §Estimated
concentration of hopanes (C
27
Ts, C
27
Tm, C
29
Ts, C
29
,C
30
,C
31
,2␣-MeC
31
) and steranes (C
27
to C
29
20S ⫹20R
13␤,17␣(H)-diasteranes and C
27
to C
29
20S ⫹20R 5␣,14␣,17␣(H) and 5␣,14␤,17␤(H)-regular steranes) in the saturated
fraction of the bitumen. D
4
-20R-5␣,14␣,17␣(H)-cholestane was used as internal standard. 㛳Indicates a parameter
primarily controlled by degree of thermal maturation. ¶Indicates a parameter primarily controlled by organic matter
sources. #The 2␣-MeC
31
/(2␣MeC
31
⫹C
30
) is expressed as a percent. **Ratio measured using hopanes C
27
Ts,
C
27
Tm, C
29
Ts, C
29
,C
30
,C
31
and the C
27
to C
29
dia- and regular steranes. ††Ratio measured using C
27
to C
29
20S ⫹
20R 13␤,17␣(H)-diasteranes and C
27
to C
29
20S ⫹20R 5␣,14␣,17␣(H) and 5␣,14␤,17␤(H)-regular steranes. ‡‡Tri-
aromatic-sterane ratios measured using SIR m/z231 as defined in (7). §§Carbon isotope data measured by
combustion of isolated kerogens (4). 㛳㛳 Analytical errors for GC-C-IRMS are within 0.4 per mil.
RESEARCH ARTICLE
13 AUGUST 1999 VOL 285 SCIENCE www.sciencemag.org1034
on September 1, 2009 www.sciencemag.orgDownloaded from

dance at the n-C
22
homolog (Fig. 1), a pattern
not observed in younger bitumens.
The strongest argument for hydrocarbon
syngeneity is the variation in biomarker distri-
butions between closely spaced shale samples
(Table 2). The sterane:hopane ratio and 2␣-
methylhopane index are both gauges of differ-
ent biological inputs during sedimentation, as
are the ratios of sterane and hopane homologs.
These differences among biomarker ratios can-
not be explained by water washing, biodegra-
dation, thermal degradation, or geochromatog-
raphy of a homogeneous migrated oil. The ob-
served variation is instead consistent with the
diversity of syngenetic bitumen typically
present within source rock facies.
To provide clues about organic origins,
carbon isotopic ratios were measured for bulk
kerogen and for individual hydrocarbons.
Kerogens are depleted in
13
C as is typical of
rocks of this age, with values of ␦
13
C falling
between ⫺40 and ⫺47 per mil (Table 2). The
␦
13
C values for 12- to 22-carbon (C
12
-C
22
)
n-alkanes range from ⫺26 to ⫺29 per mil.
This enrichment in
13
C relative to kerogen is
very rare in Phanerozoic kerogen-bitumen
pairs but typical of, though more extreme
than, other Precambrian samples (17, 18)—
for example, the 600-Ma Pertatataka Forma-
tion where C
16
-C
20
n-alkanes are up to 6 per
mil heavier than kerogen (19). It indicates
that the organisms forming most of the kero-
gen did not contribute significantly to the
n-alkanes. The isoprenoids pristane and
phytane are depleted in
13
C relative to n-alkyl
C
17
and C
18
compounds by ⬃3 per mil (Ta-
ble 2), as in other analyses of Precambrian
bitumen (18). The n-alkanes are probably the
products of a diverse biota of primary pro-
ducers and heterotrophs. The isoprenoids are
derived from photosynthetic microbes with a
possible contribution from isotopically light
methanogenic Archaea, which are known to
yield only small quantities of pristane and
phytane relative to total biomass and no n-
alkanes when pyrolyzed (20). The contrasting
depletion of
13
C in the kerogen may be at-
tributed to contributions from methanotrophs,
as proposed by Hayes (21). However, n-al-
kanes derived from isotopically depleted
membrane lipids of methanotrophs could not
be identified in our samples. Furthermore, we
cannot discount the possibility that the observed
isotopic pattern could also reflect the products
of an extinct biochemistry.
More information about biological precur-
sors can be obtained from the tetracyclic and
pentacyclic terpane biomarkers in the hydro-
carbon extracts (Fig. 2). Until now, evidence
for cyanobacteria in the Archean has been
equivocal, based on poorly preserved micro-
fossils (22) and on indirect geochemical ar-
guments (23). High concentrations of 2␣-
methylhopanes (Table 2), also observed in
other Precambrian and early Paleozoic sedi-
ments and apparently of cyanobacterial origin
(24), are consistent with the early appearance
of these organisms. Since oxygenic photo-
synthesis is their preferred physiology, met-
abolic oxygen excretion was evidently occur-
ring well before significant oxygen had accu-
mulated in the atmosphere, at about 2000 Ma
(25). The abundance of cyanobacterial bio-
markers in the Marra Mamba Formation, a
unit predominantly consisting of oxide-facies
banded iron formation, suggests that although
Precambrian iron formations could have been
produced by abiotic photochemical processes
(26) or anoxygenic phototrophic bacteria (27 ),
those in the Hamersley Group probably formed
as a result of biogenic oxygen production (28).
Steranes also occur in the extracted bitu-
mens (Fig. 2). They comprise most of the
known C
26
to C
30
pseudohomologs, their aro-
matic counterparts, and A-ring methylated ster-
anes that have been recorded in younger sedi-
ments.The biosynthesis of these sterols is char-
acteristic of eukaryotes. The only prokaryotes
known to synthesize sterols have biosynthetic
pathways that stop short of cholesterol (29)or
exclusively produce C
27
cholestenols (30). Al-
though it is possible that other yet unknown or
extinct prokaryotes also produced sterols, the
wide structural range of steranes, in relative
abundances like those of younger bitumen, is
convincing evidence for the presence of eu-
karyotes in the late Archean. The phylogenetic
position of these eukaryotes remains unclear.
The curtailed n-alkane distribution indicates
that contributions from polymethylenic chains,
derived from algal cell wall components (al-
gaenans), were much lower than observed in
younger marine sediments or were missing. We
conclude that the domain Eucarya first ap-
peared before 2700 Ma and is at least 500 to
1000 My older than indicated by current pale-
ontological data (31). This age should provide a
new calibration point for molecular clocks and
the universal tree of life.
The biomarkers we report are the oldest
known that are demonstrably indigenous and
syngenetic. They are more than a billion years
older than those from the ⬃1640-Ma Barney
Creek Formation (15), previously the oldest
well-characterized molecular fossils. Their dis-
Fig. 2. Distribution of (A) sterane and (B) hopane hydrocarbons from Roy1. The data are MRM
(multiple reaction monitoring) chromatograms obtained by gas chromatography–mass spectrom-
etry (GC-MS) analysis of metastable molecular ions fragmenting to daughter ions at mass-to-
charge ratio (m/z)⫽217 for steranes, m/z⫽191 for hopanes, and m/z⫽205 for methylhopanes.
Chromatograms are identified by carbon numbers, the relative height (abundance) of the most
intense peak in the trace, and the reaction transition.
RESEARCH ARTICLE
www.sciencemag.org SCIENCE VOL 285 13 AUGUST 1999 1035
on September 1, 2009 www.sciencemag.orgDownloaded from

tinctive features are very similar to those ob-
served in the ⬃2500-Ma Mt. McRae Shale, and
their age is supported by more thorough analyt-
ical protocols (24). The discovery and careful
analysis of biomarkers in rocks of still greater
age and of different Archean environments will
potentially offer new insights into early micro-
bial life and its evolution.
References and Notes
1. J. W. Schopf, Science 260, 640 (1993).
2. M. R. Walter, in Earth’s Earliest Biosphere,J.W.
Schopf, Ed. (Princeton Univ. Press, Princeton, NJ,
1983), pp. 187–213.
3. S. J. Mojzsis et al.,Nature 384, 55 (1996).
4. J. M. Hayes, I. R. Kaplan, K. W. Wedeking, in (2), pp.
93–134.
5. F. D. Mango, Nature 352, 146 (1991).
6. A. Dutkiewicz, B. Rasmussen, R. Buick, ibid. 395, 885
(1998).
7. K. E. Peters and J. M. Moldowan, The Biomarker Guide
(Prentice-Hall, Englewood Cliffs, NJ, 1993).
8. R. C. Morris, Precambrian Res. 60, 243 (1993).
9. A. F. Trendall, D. R. Nelson, J. R. de Laeter, S. W.
Hassler, Aust. J. Earth Sci. 45, 137 (1998).
10. N. T. Arndt, D. R. Nelson, W. Compston, A. F. Trendall,
A. M. Thorne, ibid. 38, 261 (1991).
11. R. E. Smith, J. L. Perdrix, T. C. Parks, J. Petrol. 23,75
(1982).
12. J. M. Gressier, thesis, University of Sydney, Sydney,
Australia (1996).
13. T. C. Hoering and V. Navale, Precambrian Res. 34, 247
(1987).
14. D. R. Nelson, A. F. Trendall, J. R. de Laeter, N. J.
Grobler, I. R. Fletcher, ibid. 54, 231 (1992).
15. R. E. Summons, T. G. Powell, C. J. Boreham, Geochim.
Cosmochim. Acta 52, 1747 (1988).
16. R. E. Summons and M. R. Walter, Am. J. Sci. 290A,
212 (1990).
17. T. C. Hoering, Carnegie Inst. Wash. Yearb. 64, 215
(1965); ibid. 65, 365, (1966).
18. G. A. Logan, J. M. Hayes, G. B. Heishima, R. E. Sum-
mons, Nature 376, 53 (1995); G. A. Logan, R. E.
Summons, J. M. Hayes, Geochim. Cosmochim. Acta
61, 5391 (1997).
19. G. A. Logan et al.,Geochim. Cosmochim. Acta,63,
1345 (1999).
20. S. J. Rowland, Org. Geochem. 15, 9 (1990).
21. J. M. Hayes, in Early Life on Earth, Nobel Symposium
No. 84, S. Bengtson, Ed. (Columbia Univ. Press, New
York, 1994), pp. 220–236.
22. J. W. Schopf and B. M. Packer, Science 237, 70 (1987).
23. R. Buick, ibid. 255, 74 (1992).
24. R. E. Summons, L. L. Jahnke, J. M. Hope, G. A. Logan,
Nature, in press.
25. H. D. Holland and N. J. Beukes, Am. J. Sci. 290A,1
(1990); A. H. Knoll and H. D. Holland, in Effects of
Past Global Change on Life, S. M. Stanley, Ed. (Na-
tional Academy Press, Washington, DC, 1995), pp.
21–33.
26. P. S. Braterman, A. G. Cairns-Smith, R. W. Sloper,
Nature 303, 163 (1983).
27. F. Widdel et al.,ibid. 362, 834 (1993).
28. P. Cloud, Science 160, 729 (1968); Econ. Geol. 68,
1135 (1973).
29. G. Ourisson, M. Rohmer, K. Poralla, Annu. Rev. Micro-
biol. 41, 301 (1987).
30. W. Kohl, A. Gloe, H. Reichenbach, J. Gen. Microbiol.
129, 1629 (1983).
31. T.-M. Han and B. Runnegar, Science 257, 232 (1992);
A. H. Knoll, ibid. 256, 622 (1992).
32. Supported by the Studienstiftung des Deutschen
Volkes (J.J.B.) and American Chemical Society Petro-
leum Research Fund (R.B.). We thank J. Gressier and
RioTinto Exploration for samples, the AGSO Isotope
& Organic Geochemistry staff for technical assist-
ance, J. Kamprad for x-ray diffraction analyses, T.
Blake, D. Des Marais, L. L. Jahnke, C. J. Boreham, D. S.
Edwards, T. G. Powell, D. E. Canfield, and M. R. Walter
for advice, and J. M. Hayes, A. Knoll, and an anony-
mous reviewer for their thoughtful comments. G.A.L.
and R.E.S. publish with the permission of the Execu-
tive Director of AGSO.
19 May 1999; accepted 13 July 1999
REPORTS
Josephson Persistent-Current
Qubit
J. E. Mooij,
1,2
* T. P. Orlando,
2
L. Levitov,
3
Lin Tian,
3
Caspar H. van der Wal,
1
Seth Lloyd
4
A qubit was designed that can be fabricated with conventional electron beam
lithography and is suited for integration into a large quantum computer. The
qubit consists of a micrometer-sized loop with three or four Josephson junc-
tions; the two qubit states have persistent currents of opposite direction.
Quantum superpositions of these states are obtained by pulsed microwave
modulation of the enclosed magnetic flux by currents in control lines. A su-
perconducting flux transporter allows for controlled transfer between qubits of
the flux that is generated by the persistent currents, leading to entanglement
of qubit information.
In a quantum computer, information is stored
on quantum variables such as spins, photons, or
atoms (1–3). The elementary unit is a two-state
quantum system called a qubit. Computations
are performed by the creation of quantum su-
perposition states of the qubits and by con-
trolled entanglement of the information on the
qubits. Quantum coherence must be conserved
to a high degree during these operations. For a
quantum computer to be of practical value, the
number of qubits must be at least 10
4
. Qubits
have been implemented in cavity quantum elec-
trodynamics systems (4), ion traps (5), and
nuclear spins of large numbers of identical mol-
ecules (6). Quantum coherence is high in these
systems, but it seems difficult or impossible to
realize the desired high number of interacting
qubits. Solid state circuits lend themselves to
large-scale integration, but the multitude of
quantum degrees of freedom leads in general to
short decoherence times. Proposals have been
put forward for future implementation of qubits
with spins of individual donor atoms in silicon
(7), with spin states in quantum dots (8), and
with d-wave superconductors (9); the technol-
ogy for practical realization still needs to be
developed.
In superconductors, all electrons are con-
densed in the same macrosopic quantum
state, separated by a gap from the many
quasi-particle states. This gap is a measure
for the strength of the superconducting ef-
fects. Superconductors can be weakly cou-
pled with Josephson tunnel junctions (regions
where only a thin oxide separates them). The
coupling energy is given by E
J
(1 ⫺cos ␥),
where the Josephson energy E
J
is proportion-
al to the gap of the superconductors divided
by the normal-state tunnel resistance of the
junction and ␥is the gauge-invariant phase
difference of the order parameters. The cur-
rent through a Josephson junction is equal to
I
o
sin ␥,with I
o
⫽(2e/ប)E
J
, where eis the
electron charge and បis Planck’s constant
divided by 2␲. In a Josephson junction circuit
with small electrical capacitance, the num-
bers of excess Cooper pairs on islands n
i
,n
j
and the phase differences ␥
i,
␥
j
are related as
noncommuting conjugate quantum variables
(10). The Heisenberg uncertainty between
phase and charge and the occurrence of quan-
tum superpositions of charges as well as
phase excitations (vortexlike fluxoids) have
been demonstrated in experiments (11). Co-
herent charge oscillations in a superconduct-
ing quantum box have recently been observed
(12). Qubits for quantum computing based on
charge states have been suggested (13,14).
However, in actual practice, fabricated Jo-
sephson circuits exhibit a high level of static
and dynamic charge noise due to charged
impurities. In contrast, the magnetic back-
ground is clean and stable. Here, we present
the design of a qubit with persistent currents
of opposite sign as its basic states. The qubits
1
Department of Applied Physics and Delft Institute for
Microelectronics and Submicron Technologies, Delft
University of Technology, Post Office Box 5046, 2600
GA Delft, Netherlands.
2
Department of Electrical En-
gineering and Computer Science,
3
Department of
Physics and Center for Materials Science and Engi-
neering,
4
Department of Mechanical Engineering,
Massachusetts Institute of Technology, Cambridge,
MA 02139, USA.
*To whom correspondence should be addressed. E-
mail: mooij@qt.tn.tudelft.nl
RESEARCH ARTICLE
13 AUGUST 1999 VOL 285 SCIENCE www.sciencemag.org1036
on September 1, 2009 www.sciencemag.orgDownloaded from