Content uploaded by Donald Canfield
Author content
All content in this area was uploaded by Donald Canfield on Aug 07, 2019
Content may be subject to copyright.
sion line and nonlinear dependence of the intensity are clearly
observed up to a current of about 880 mA, where laser threshold is
reached. Lasing takes place at ,4.4 THz, on the high-energy side of
the luminescence line, probably owing to the reduced waveguide
losses at shorter wavelengths. Single-mode emission is obtained,
which is probably a consequence of the relatively narrow gain
spectrum and the wide Fabry–Perot mode spacing. A high-resol-
ution laser spectrum is shown in the inset; the measured linewidth is
limited by the resolution of the spectrometer (3.75 GHz).
Figure 4 shows the light–current (L–I) and voltage–current (V–I)
characteristics of a representative device. At a heat-sink temperature
of 8 K, the output peak power is estimated to be more than 2 mW,
with a threshold current density of 290 A cm
22
. The latter is a very
small value for quantum-cascade lasers and allows operation at high
duty cycles (up to 10%) even in this large device. We expect that
narrower stripes and appropriate changes in sample processing
would readily lead to continuous-wave operation. The initial high
resistivity in the V–Icharacteristics stems from misalignment of the
sub-bands at low field; at higher fields, the injector miniband lines
up with the second miniband of the following stage and carrier
injection into the upper laser level takes place. The electric field at
threshold is 7.5 kV cm
21
. This is larger than the design value of
3.5 kV cm
21
, probably as a result of the non-negligible contact
resistance. As expected, a negative differential resistance feature
was observed at about 850 A cm
22
.
These experimental results match well the theoretical predictions
of Fig. 1b. The V–Icurve has all the distinctive qualitative features
and the measured current densities are of the same order of the
computed ones, showing that even at these small energies carrier
relaxation and transport are dominated by electron–LO phonon
and electron–electron scattering. The discrepancy is a factor of
about 1.5, possibly related to acoustic phonon or impurity scatter-
ing or to a lower than specified free-carrier density in the sample. In
fact, the simulation indicates that a reduction in the doping density
of the injectors by 25% would lead to a reduction in current density
of 35%. From the theoretical values of population inversion and
confinement factor, we calculate a maximum modal gain of 23 cm
21
(ref. 18). This value compares well with the estimated cavity losses
aWþaM¼ð16 þ4Þcm21¼20 cm21,a
M
being the mirror out-
coupling of the 3.1-mm-long stripe. This consistency is confirmed
by the experimental observation that laser stripes shorter than
1 mm, with corresponding larger mirror losses, do not reach laser
threshold.
We believe that improved design of the active region of our device
(in particular aiming at the reduction of thermal backfilling),
together with optimized fabrication (junction-down mounting,
facet coating, lateral overgrowth), would rapidly lead to continu-
ous-wave emission and to operation at liquid-nitrogen tempera-
tures. The present demonstration of a terahertz quantum-cascade
laser, operating below the LO phonon band, represents a first step
towards the development of widely usable terahertz photonics. A
Received 14 January; accepted 19 February 2002.
1. Miles, R. E., Harrison, P. & Lippens, D. (eds) Terahertz Sources and Systems Vol. 27 (NATO Science
Series II, Kluwer, Dordrecht, 2001).
2. Han, P. Y., Cho, G. C. & Zhang, X.-C. Time-domain transillumination of biological tissue with
terahertz pulses. Opt. Lett. 25, 242–244 (2000).
3. Mittleman, D. M., Jacobsen, R. H. & Nuss, M. C. T-ray imaging. IEEE J. Sel. Top. Quant. Electron. 2,
679–692 (1996).
4. Mittleman, D. M., Hunsche, S., Boivin, L. & Nuss, M. C. T-ray tomography. Opt. Lett. 22, 904–906
(1997).
5. Pavlov, S. G. et al. Stimulated emission from donor transitions in silicon. Phys. Rev. Lett. 84,
5220–5223 (2000).
6. Matsuura, S., Tani, M. & Sakai, K. Generation of coherent terahertz radiation by photomixing in
dipole photoconductive antennas. Appl. Phys. Lett. 70, 559–561 (1997).
7. Hu, B. B., Zhang, X.-C. & Auston, D. H. Terahertz radiation induced by subband-gap femtosecond
optical excitation of GaAs. Phys. Rev. Lett. 67, 2709–2712 (1991).
8. Kersting,R., Unterrainer, K., Strasser, G., Kauffmann,H. F.& Gornik, E. Few-cycle THz emission from
cold plasma oscillations. Phys. Rev. Lett. 79, 3038–3041 (1997).
9. Bru
¨ndermann, E., Chamberlin, D. R. & Haller, E. E. High duty cycle and continuous terahertz
emission from germanium. Appl. Phys. Lett. 76, 2991–2993 (2000).
10. Faist, J. et al. Quantum cascade laser. Science 264, 553–556 (1994).
11. Rochat, M., Faist, J., Beck, M., Oesterle, U. & Ilegems, M. Far-infrared (l¼88
m
m)
electroluminescence in a quantum cascade structure. Appl. Phys. Lett. 73, 3724–3726 (1998).
12. Williams, B. S., Xu, B., Hu, Q. & Melloch, M. R. Narrow-linewidth terahertz intersubband emission
from three-level systems. Appl. Phys. Lett. 75, 2927–2929 (1999).
13. Ulrich, J., Zobl, R., Schrenk, W., Strasser, G. & Unterrainer, K. Terahertz quantum cascade structures:
Intra- versus interwell transition. Appl. Phys. Lett. 76, 1928–1930 (2000).
14. Helm, M., England, P., Colas, E., De Rosa, F. & Allen, S. J. Intersubband emission from semiconductor
superlattices excited by sequential resonant tunnelling. Phys. Rev. Lett. 63, 74–77 (1989).
15. Colombelli, R. et al. Far-infrared surface-plasmon quantum-cascade lasers at 21.5
m
m and 24
m
m
wavelengths. Appl. Phys. Lett. 78, 2620–2622 (2001).
16. Hofstetter, D., Beck, M., Aellen, T. & Faist, J. High-temperature operation of distributed feedback
quantum-cascade lasers at 5.3
m
m. Appl. Phys. Lett. 78, 396–398 (2001).
17. Ko
¨hler, R., Iotti, R. C., Tredicucci, A. & Rossi, F. Design and simulation of terahertz quantum cascade
lasers. Appl. Phys. Lett. 79, 3920–3922 (2001).
18. Tredicucci, A. et al. High performance interminiband quantum cascade lasers with graded
superlattices. Appl. Phys. Lett. 73, 2101–2103 (1998).
19. Iotti, R. C. & Rossi, F. Nature of charge transport in quantum-cascade lasers. Phys. Rev. Lett. 87,
146603-1–146603-4 (2001).
20. Iotti, R. C. & Rossi, F. Carrier thermalization versus phonon-assisted relaxation in quantum-cascade
lasers: A Monte Carlo approach. Appl. Phys. Lett. 78, 2902–2904 (2001).
21. Sirtori, C. et al. Long-wavelength (l,8–11.5
m
m) semiconductor lasers with waveguides based on
surface plasmons. Opt. Lett. 23, 1366–1368 (1998).
22. Rochat, M., Beck, M., Faist, J. & Oesterle, U. Measurement of far-infrared waveguide loss using a
multisection single-pass technique. Appl. Phys. Lett. 78, 1967–1969 (2001).
23. Ko
¨hler, R. et al. High-intensity interminiband terahertz emission from chirped superlattices. Appl.
Phys. Lett. 80, 1867 (2002).
Acknowledgements
We thank S. Dhillon for discussions.This work was supported in part by the European
Commission through the IST Framework V FET project WANTED. R.K. was supported by
the C.N.R.; E.H.L. and A.G.D were supported by Toshiba Research Europe Ltd and The
Royal Society, respectively.
Competing interests statement
The authors declare that they have no competing financial interests.
Correspondence and requests for materials should be addressed to R.K.
(e-mail: koehler@nest.sns.it.).
..............................................................
Ocean productivity before about
1.9 Gyr ago limited by phosphorus
adsorption onto iron oxides
Christian J. Bjerrum*†‡ & Donald E. Canfield*
*Danish Center for Earth System Science, Institute of Biology, University of
Southern Denmark, Campusvej 55, DK-5230 Odense, Denmark
†Danish Center for Earth System Science, Niels Bohr Institute for Astronomy,
Physics and Geophysics, University of Copenhagen, Juliane Maries Vej 30, DK-
2100 Copenhagen, Denmark
.............................................................................................................................................................................
After the evolution of oxygen-producing cyanobacteria at some
time before 2.7 billion years ago
1
, oxygen production on Earth is
thought to have depended on the availability of nutrients in the
oceans, such as phosphorus (in the form of orthophosphate). In
the modern oceans, a significant removal pathway for phos-
phorus occurs by way of its adsorption onto iron oxide depos-
its
2,3
. Such deposits were thought to be more abundant in the past
when, under low sulphate conditions, the formation of large
amounts of iron oxides resulted in the deposition of banded iron
formations
4,5
. Under these circumstances, phosphorus removal
by iron oxide adsorption could have been enhanced. Here we
analyse the phosphorus and iron content of banded iron form-
ations to show that ocean orthophosphate concentrations from
‡ Present address: Geological Institute, University of Copenhagen, Øster Voldgade 10, DK-1350
Copenhagen, Denmark.
letters to nature
NATURE| VOL 417 | 9 MAY 2002 | www.nature.com 159
© 2002 Macmillan Magazines Ltd
3.2 to 1.9 billion years ago (during the Archaean and early
Proterozoic eras) were probably only ,10–25% of present-day
concentrations. We suggest therefore that low phosphorus avail-
ability should have significantly reduced rates of photosynthesis
and carbon burial, thereby reducing the long-term oxygen
production on the early Earth
—
as previously speculated
4
—
and
contributing to the low concentrations of atmospheric oxygen
during the late Archaean and early Proterozoic.
The oxidation of dissolved ferrous iron produces insoluble iron
oxyhydroxides, which strongly adsorb phosphates at pH values of
less than 9 (ref. 6). The extent of phosphorus adsorption can be
expressed as: [P
ads
]¼K
ads
[P
d
][Fe
3þ
], where [P
ads
] is the concen-
tration (
m
M) of phosphate adsorbed onto iron oxides, [P
d
] is the
concentration (
m
M) of dissolved orthophosphate (PO
4
32
) in sol-
ution, [Fe
3þ
] is the concentration (
m
M) of iron oxide particles, and
K
ads
(
m
M
21
) is the adsorption constant. Particles of iron oxide
formed from Fe
2þ
oxidation in submarine hydrothermal systems
adsorb phosphate with a K
ads
value of 0.07 ^0.01
m
M
21
(ref. 7). As
phosphorus adsorbs predictably onto newly formed iron oxide
surfaces, the phosphorus content of ancient sediments rich in
iron oxide indicates, in principle, the phosphorus concentration
of the water from which the oxides formed
7
.
Here we use the phosphorus and Fe
3þ
content of Archaean and
early Proterozoic banded iron formations (BIFs), and the value for
K
ads
determined for modern hydrothermal systems
7
, to estimate the
[P
d
] of contemporaneous sea water:
½Pd¼ð1=KadsÞðPads =Fe3þÞð1Þ
The ratio P
ads
/Fe
3þ
is directly available, in many instances, from BIF
analyses (Fig. 1). An average Archaean and early Proterozoic
phosphorus concentration of 0.15 ^0.15
m
M is calculated (ranging
from 0.03 to 0.29
m
M) for the waters where BIFs deposited (Fig. 1).
Generally lower values of [P
d
] are indicated at 1.9–2.0 Gyr ago,
compared to earlier BIFs.
We have avoided siderite-rich BIFs in the analysis as some contain
unusual
13
C-depleted carbonates (
d
13
C,25‰)
8,9
, indicating the
oxidation of sedimentary organic carbon, possibly by Fe
reduction
10
, and the probable inclusion of original organic P into
the total P pool. This would complicate our calculation. Further-
more, iron oxide (re)crystallization or iron reduction during early
diagenesis could have induced a loss of P from the oxide-rich BIFs.
In modern sediments preserving hydrothermally derived iron
oxides, a loss of up to 50% of the originally adsorbed P is indicated
11
.
A 50% loss in P
ads
would mean that our calculated [P
d
] could be too
low by as much as a factor of two (ref. 11). However, this is probably a
maximum correction, because hydrothermal sediments with higher
iron content, approaching that of BIFs, have experienced ,25%
P loss
11
. Acknowledging that further tests of our approach are
required, we conservatively estimate that BIFs deposited in sea
water containing between, on average, 0.15 and 0.6
m
M dissolved P.
We note that P retention in sediments under anoxic ocean
conditions in the Archaean and early Proterozoic was probably
different from today. This is indicated by the preservation of Fe
oxides in BIFs, which deposited from anoxic ocean waters contain-
ing dissolved Fe
2þ
(ref. 4). Today, ocean anoxia results in a sulphidic
water column which effectively reacts all of the available reactive Fe
to form Fe sulphide minerals
5
. Thus, today, anoxia leads to some P
loss from sediments
12–15
, whereas anoxia during BIF deposition
probably led to P retention on Fe oxides, and as ferrous phosphate
minerals when there was Fe reduction in the sediment. The main
difference between now and then was probably low sulphate
concentrations in Archaean and early Proterozoic oceans, signifi-
cantly reducing rates of sulphate reduction
5
. Even in the modern
world, controlled sediment incubation experiments show that loss
of adsorbed P is minimal under sediment diagenesis with low
Figure 1 Element ratios in BIFs and calculated dissolved phosphate concentrations. Mole
ratios of Fe
3þ
/Fe
tot
(a) and P/Fe
tot
(b) are from the literature (see Supplementary
Information). c, Dissolved phosphate concentrations are calculated from equation (1).
Phosphate loss during early diagenesis can result in up to a factor of two error in P
d
(see
text and ref. 11). Standard deviation for each BIF unit is shown as solid error bars. Dashed
error bars indicate that no standard deviation could be calculated from the literature. In
cases where Fe
3þ
content is not available, we use the mean Fe
3þ
/Fe
tot
ratio of 0.43.
Figure 2 Simplified phosphate and iron cycle model of the Archaean and early
Proterozoic. Dissolved ferrous iron (Fe
2+
) is oxidized at the base of the oxic mixed layer,
leading to iron oxide burial (F
Fe,ox
¼aF
Fe,in
) and BIFs with adsorbed P (F
P,ads
). Organic
matter export production (EP) is limited by upwelled P and leads to the burial of organically
derived phosphate (F
P,org+CFA
). For simplicity, ferrous iron is assumed to be buried
(F
Fe,red
) without associated phosphate. The total reactive iron input, F
Fe,in
, is the sum of
iron from land (dissolved+particulate) and hydrothermal iron.
letters to nature
NATURE| VOL 417 | 9 MAY 2002 | www.nature.com160 © 2002 Macmillan Magazines Ltd
sulphate concentrations
16
. The fact that BIFs are preserved, and
contain phosphate, shows that the usual view of anoxia leading to P
release is probably not valid during the periods in Earth’s history
when BIFs formed.
BIFs deposited in environments ranging from shelf and upper
slope to the abyssal plain
17
. The Barberton BIF, associated with, and
probably related to, local volcanism, was deposited at a water depth
of approximately 900 m, the deepest water yet recognized for BIFs
18
.
The other generally larger BIFs analysed here, lacking an obvious
local volcanic source, deposited on the outer shelf or slope well
below storm wave influence
19
. The Fe-oxide-rich facies of the
Kuruman Iron Formation, a reasonable representative of the large
2.6-Gyr and younger BIFs in the data set, was deposited in deeper
water than the contemporaneous siderite-enriched BIF and shales
19
.
Furthermore, the siderite-enriched BIF formed in water with dis-
solved inorganic carbon highly
13
C-depleted compared to surface
ocean water
8,20
. Thus, the siderite-enriched BIF, and by extension
the Fe-oxide-rich facies, deposited well below a pronounced che-
mocline where the accumulation of phosphate should have
accompanied the steep gradients in isotopic composition of dis-
solved inorganic carbon. Therefore, our calculated phosphate
concentrations characterize a region well below the chemocline
and may represent the deep ocean. Phosphate concentrations of
0.15 to 0.6
m
M are considerably less than the average modern ocean
value of 2.3
m
M.
Low phosphate concentrations in the Archaean and early Proter-
ozoic oceans could have arisen from changes in the weathering flux
of dissolved and reactive phosphate (F
P,in
) to the ocean and/or
significant changes in the ocean phosphate sinks (Fig. 2). We have
no evidence supporting significantly reduced weathering fluxes of
reactive phosphate. Continents may have reached their present size
early in Earth’s history and the intensity of chemical weathering,
controlled by temperature and soil pH, might have been high in the
Archaean and early Proterozoic. Alternatively, and more probably,
low P concentrations originated from a strong P sink owing to
significant adsorption onto, and removal by, iron oxide particles.
We use a simple ocean model to explore the possible influence of
BIF deposition on ocean P concentrations (Fig. 2). P is delivered to
the oceans from rivers (F
P,in
) and is removed by adsorption onto
iron oxides (P
ads
) and as organically derived reactive P (P
orgþCFA
),
which includes organic-bound P and P liberated from organic
matter and mainly reprecipitated into carbonate fluorapatite
(CFA) (refs 12, 13). Of the total reactive iron flux (F
Fe,in
) to the
ocean
5
, only a fraction (a) is buried as Fe oxides with adsorbed P;
the rest is assumed to be buried as ferrous iron phases to which P
does not adsorb (that is, sulphides and siderite). The burial flux of
adsorbed P is then:
FP;ads ¼ðKads½PdÞðaFFe;in Þð2Þ
The burial flux of P
orgþCFA
(designated as F
P,orgþCFA
) depends on
[P
d
], the upwelling velocity, u, of deep water supplying P
d
to the
euphotic zone, and the relationship, g, between the export of
organic P from the surface ocean and its burial
4,21
:
FP;orgþCFA ¼gðu½PdÞ ð3Þ
At steady state FP;in 2ðFP;orgþCFA þFP;adsÞ¼0;allowing a solution
for [P
d
]:
½Pd¼ FP;in
guþaKadsFFe;in ð4Þ
Figure 3 shows [P
d
] contoured as a function of the fraction, a,of
total reactive iron buried with adsorbed P, versus reactive phosphate
input (F
P,in
). [P
d
] increases as F
P,in
increases, and decreases with
increasing a. With today’s range in F
P,in
(ref. 12) and F
Fe,in
(ref. 5)
we find a [P
d
] similar to that obtained in the above BIF analysis if
approximately half of the reactive iron flux is buried with adsorbed
P (0.4 ,a,0.6). Thus, low values of ocean P
d
can be maintained
if about half of F
Fe,in
is removed as freshly precipitated iron oxides.
Such a high Fe
3þ
/Fe
tot
ratio is indicated for BIFs
8,19
(Fig. 1a), and if
this ratio represents the removal of F
Fe,in
in general, then the initial
adsorption of P into BIF explains the low [P
d
] inferred for Archaean
and early Proterozoic sea water.
Low values of ocean [P
d
] would have probably limited rates of
primary production, and by extension, organic carbon burial, and
the input of oxygen to the atmosphere
4,20,22
. The relationship
between [P
d
] and the burial rate of organic carbon (F
C,org
)is
given as the product of equation (4) and the ratio of organic carbon
to organically derived P (h
c:p
¼C
org
/P
orgþCFA
):
FC;org ¼gðu½PdÞhc:p ð5Þ
Figure 3 Modelled mean ocean phosphate concentration [P
d
] from equation (4). a, The
standard case with K
ads
¼0.07
m
M
21
;b, The possible case of a 50% P
ads
loss during
early diagenesis with K
ads
<0.035
m
M
21
. The [P
d
] values (in
m
M) are shown on the
contours as a function of the fraction (a) of total iron buried with adsorbed P, and reactive
phosphate input (F
P,in
). Very low ocean phosphate concentrations occur if a large fraction
of reactive iron input is buried with adsorbed P. Boxed regions correspond to the
diagnosed Archaean conditions with the pre-industrial range of F
P,in
(refs 12, 30) and
present F
Fe,in
(ref. 5). At present P, burial fluxes after early diagenesis are partitioned into:
FP;org ¼3:6£1010 mol P yr21;FP;CFA ¼3:6£1010 mol P yr21and FP;ads ¼
0:7£1010 mol P yr21(refs 12, 30). Part of the F
P,CFA
is originally reprecipitated from P
liberated from decaying organic matter as well as P from reduction of iron oxides.
Assuming, conservatively, that ,50% of the CFA burial flux originates from decay of
organic matter, then FP;orgþCFA ¼5:4£1010 mol P yr21:With a burial efficiency g¼
1:9£1022;the above F
P,org+CFA
is reproduced for the present mean ocean P
concentration (2.3
m
M) and mixing u¼1:26 £1018 lyr
21:At present F
P,org+CFA
is
associated with an organic carbon burial of ,1£10
13
mol P yr
21
(ref. 23), which gives a
C
org
/P
org+CFA
ratio of 185.
letters to nature
NATURE| VOL 417 | 9 MAY 2002 | www.nature.com 161
© 2002 Macmillan Magazines Ltd
With values of g,uand h
c:p
comparable to today, a reduction in [P
d
]
to 10–25% of present-day concentration (as we infer for the late
Archaean and early Proterozoic) implies a 75–90% reduction in the
rate of organic carbon burial. By contrast, if much less P
orgþCFA
was
buried with organic carbon, as has been argued for modern euxinic
sediments
14
(see also in ref. 15), then low [P
d
] would support
substantial organic carbon burial, and oxygen production
4
. Thus,
with h
c:p
<1,000, carbon burial would approach modern values
23
.
However, we believe that such high h
c:p
ratios are probably in-
appropriate for an anoxic Fe-containing Precambrian ocean for two
reasons. First, the high h
c:p
ratios in modern euxinic sediments are
partly a result of P loss
13–15
under high sulphate conditions that were
not present during BIF deposition
24
. Second, even in modern
settings, h
c:p
may be much lower than previously thought
15
, with
ratios of 150–200 recently reported from the euxinic, and rapidly
depositing, Saanich Inlet sediments
13
.Thus,low[P
d
]inlate
Archaean and early Proterozoic oceans should have resulted in
reduced rates of carbon burial. Higher heat flow and tectonic
activity could have accelerated rates of geochemical cycling includ-
ing rates of all of the processes discussed here, although the
quantitative influence of heat flow on geochemical cycling rates is
poorly known
25,26
.
Previous analysis of the marine isotope record of organic and
inorganic carbon suggests that for the Archaean (.2.5 Gyr ago)
around 11% of the total carbon buried in ocean sediments was
removed as organic carbon
27
. By comparison, today, organic
carbon represents 20% of the total carbon removal
27
, or a burial
percentage about twice that in the Archaean. Our results are
generally consistent with this, and suggest reduced rates of organic
carbon burial in the late Archaean and early Proterozoic. Further
modelling studies, and a better resolution of the carbon isotope
record, would better help to establish the concordance between our
reconstruction of late Archaean and early Proterozoic nutrient
chemistry and the carbon burial history as revealed from carbon
isotope studies.
In addition to indicating relatively reduced rates of organic
carbon burial in the Archaean, the isotope record of marine
carbonates demonstrates at least one, and perhaps several, very
large positive isotope excursions between 2.4 and 2.0 Gyr ago (refs
20, 27). These excursions indicate large increases in the relative
proportions of organic carbon burial in ocean sediments, and
possibly also indicate increases in the rate of organic carbon
deposition
27
. Recent compilations of the timing of BIF deposition
indicate a curious lack of BIFs during this same time from 2.4 to
around 2.0 Gyr ago (ref. 28). Furthermore, the isotope record of
sedimentary sulphides
24
shows the first occurrence of highly
34
S-
depleted marine sulphides around 2.4 Gyr ago. The sulphur isotope
record indicates an increase in seawater sulphate concentrations to
,1 mM at around 2.4Gyr ago, and a consequent increase in rates of
sulphate reduction due to more sulphate availability
24
. We propose
that the burial pulse of organic carbon between 2.4 and 2.0 Gyr ago
could have been driven, at least in part, by the P made available by
increasing rates of sulphate reduction. Therefore, the rise in atmos-
pheric oxygen concentrations
4,20
that apparently accompanied the
burial pulse of organic carbon
27
could also, in part, have resulted
from a reduction in the Fe oxide sink and a consequently larger
availability of P.
The decline of BIFs between 2.4 and 2.0 Gyr ago (ref. 28) could
have been a logical consequence of the titration of deep-ocean Fe
with sulphide owing to increasing rates of sulphate reduction
5
. For
some reason not yet clear, BIF deposition, decreased organic carbon
burial, and low P concentrations were reinitiated around 2.0 Gyr
ago, persisting until around 1.8 Gyr ago (Fig. 1). Thereafter, con-
sistent with an earlier proposal
5
, sulphidic conditions reoccurred.
There is no indication of high burial fluxes of organic carbon after
1.8 Gyr ago, even though the iron oxide sink for phosphorus would
have been substantially reduced under sulphidic ocean conditions.
Another nutrient may then have limited ocean primary pro-
ductivity
29
.A
Received 23 August 2001; accepted 26 March 2002.
1. Brocks, J. J.,Logan, G. A., Buick, R. & Summons, R. E. Archean molecular fossils and the early rise of
eukaryotes. Science 285, 1033–1036 (1999).
2. Berner, R. A. Phosphate removal from sea water by adsorption on volcanogenic ferric oxides. Earth
Planet. Sci. Lett. 18, 77–86 (1973).
3. Wheat, C. G., Feely, R. A. & Mottl, M. J. Phosphate removal by oceanic hydrothermal processes:
An update of the phosphorus budget in the oceans. Geochim. Cosmochim. Acta 19, 3593–3608
(1996).
4. Holland, H. D. The Chemical Evolution of the Atmosphere and Oceans (Princeton Univ. Press, New
York, 1984).
5. Canfield, D. E. A new model for Proterozoic ocean chemistry. Nature 396, 450–453 (1998).
6. Stumm, W. & Morgan, J. J. Aquatic Chemistry: An Introduction Emphasizing Chemical Equilibria in
Natural Waters (Wiley & Sons, New York, 1981).
7. Feely,R. A., Trefry, J. H., Lebon, G.T. & German, C. R. The relationship between P/Fe and V/Fe ratios
in hydrothermal precipitates and dissolved phosphate in seawater. Geophys. Res. Lett. 25, 2253–2256
(1998).
8. Beukes, N. J., Klein, C., Kaufman, A. J. & Hayes, J. M. Carbonate petrography, kerogen distribution,
and carbon and oxygen isotope variations in an early Proterozoic transition from limestone to iron-
formation deposition, Transvaal Supergroup, South Africa. Econ. Geol. 85, 663–690 (1990).
9. Kaufman, A. J., Hayes,J. M. & Klein, C. Primary and diagenetic controls of isotopic compositions of
iron-formation carbonates. Geochim. Cosmochim. Acta 54, 3461–3473 (1990).
10. Walker, J. C. G. Suboxic diagenesis in banded iron formations. Nature 309, 340–342 (1984).
11. Schaller, T., Morford, J., Emerson, S. R. & Feely, R. A. Oxyanions in metalliferous sediments: Tracers
for paleoseawater metal concentrations. Geochim. Cosmochim. Acta 63, 2243–2254 (2000).
12. Compton, J. et al. in Marine Authigenesis; From Global to Microbial (eds Glenn, C. R., Prevot, L. L. &
Lucas, J.) 21–33 (Society for Sedimentary Geology, Tulsa, 2000).
13. Filippelli, G. M. Carbon and phosphorus cycling in anoxic sediments of the Saanich Inlet, British
Columbia. Mar. Geol. 174, 307–321 (2001).
14. Van Cappellen, P. & Ingall, E. D. Redox stabilization of the atmosphere and oceans by phosphorus-
limited marine productivity. Science 271, 493–496 (1996).
15. Anderson, L. D., Delaney, M. L. & Faul, K. L. Carbon to phosphorus ratios in sediments: Implications
for nutrient cycling. Glob. Biogeochem. Cycles 15, 65–79 (2001).
16. Roden, E. E. & Edmonds, J. W. Phosphate mobilization in iron-rich anaerobic sediments: Microbial
Fe(III) oxide reduction versus iron-sulfide formation. Archiv. Hydrobiol. 139, 347–378 (1997).
17. Fralick, P. & Barrett, T. J. in Sedimentary Facies Analysis; A Tribute to the Research and Teaching of
Harold G. Reading (ed. Plint, A. G.) 137–156 (Special Publication of the International Association of
Sedimentologists, Oxford, 1995).
18. de Ronde, C. E. J., Channer, D. M., De, R., Faure, K., Bray, C. J.& Spooner, E. T. C. Fluid chemistry of
Archean seafloor hydrothermal vents: Implication for the composition of circa 3.2 Ga seawater.
Geochim. Cosmochim. Acta 61, 4025–4042 (1997).
19. Klein, C. & Beukes, N. J. Geochemistry and sedimentology of a facies transition from limestone to
iron-formation deposition in the early Proterozoic Transvaal Supergroup, South Africa. Econ. Geol.
84, 1733–1774 (1989).
20. Karhu, J. A. & Holland, H. D. Carbon isotopes and the rise of atmospheric oxygen. Geology 24,
867–870 (1996).
21. Shaffer, G. A model of biogeochemical cycling of phosphorous, nitrogen, oxygen, andsulphur in the
ocean: One step toward a global climate model. J. Geophys. Res. 94, 1979–2004 (1989).
22. Lenton, T. M. & Watson, A. J. Redfield revisited 2. What regulates the oxygen content of the
atmosphere? Glob. Biogeochem. Cycles 14, 249–268 (2000).
23. Holland, H. D. The Chemistry of the Atmosphere and Oceans (Wiley, New York, 1978).
24. Canfield, D. E., Habicht, K. S. & Thamdrup, B. The archean sulfur cycle and the early history of
atmospheric oxygen. Science 288, 658–661 (2000).
25. Des Marais, D. J. Isotopic evolution of the biogeochemical carbon cycle during the Proterozoic eon.
Org. Geochem. 27, 185–193 (1997).
26. Sleep, N. H. & Zahnle, K. Carbon dioxide cycling and implications for climate on ancient Earth.
J. Geophys. Res. 106, 1373–1399 (2001).
27. Des Marais, D. J., Strauss, H., Summons, R. E. & Hayes, J. M. Carbon isotope evidence from the
stepwise oxidation of the Proterozoic environment. Nature 359, 605–609 (1992).
28. Isley, A. E. & Abbott, D. H. Plume-related mafic volcanism and the deposition of banded iron
formation. J. Geophys. Res. 104, 15461–15477 (1999).
29. Anbar, A. D. & Knoll, A. H. Trace metal limitation of primary production 1.85–1.25 Ga. (American
Geophysical Union Fall Meeting, 1999); available at http://www.agu.org/meetings/fm99top.html.
30. Berner, E. K. & Berner, R. A. The Global Water Cycle, Geochemistry and Environment (Prentice Hall,
New Jersey, 1994).
Supplementary Information accompanies the paper on Nature’s website
(http://www.nature.com).
Acknowledgements
We thank J. Hayes and T. Lenton for comments and suggestions. This work was funded by
the Danish National Research Foundation (Danmarks Grundforskingsfond) and the
Danish National Science Foundation (SNF).
Competing interests statement
The authors declare that they have no competing financial interests.
Correspondence and requests for materials should be addressed to C.J.B.
(e-mail: cjb@geo.geol.ku.dk).
letters to nature
NATURE| VOL 417 | 9 MAY 2002 | www.nature.com162 © 2002 Macmillan Magazines Ltd