1.
below the È10-Mpc scales calculated here
would tell us about density perturbations in
photons (and baryons) in the early universe,
even at scales smaller than the diffusion scale at
recombination. In this sense, the magnetic field
generated by this mechanism can be regarded
as a fossil of density perturbations in the early
universe, whose signature in photons and
baryons has been lost. Therefore, this result
provides the possibility of probing observations
on how density perturbations in photons have
evolved and been swept away at these small
scales, where no one can, in principle, probe
directly through photons.
The amplitude of the cosmologically gen-
erated magnetic fields is too small to be ob-
served directly through polarization effects or
synchrotron emission. However, magnetic fields
with such small amplitude may be detected by
gamma ray burst observations (25) through the
delay of the arrival time of gamma ray photons
due to the magnetic deflection of high-energy
electrons responsible for such gamma ray
photons (26, 27). We suggest that, because the
weak magnetic fields should inevitably be
generated from cosmological perturbations as
presented here, and because they should exist
all over the universe even in the intercluster
fields, then the weak fields should be detectable
by future high-energy gamma ray experiments,
such as GLAST (the Gamma Ray Large Area
Space Telescope) (28).
Although the power of B increases as k2 on
small scales (Fig. 2), the diffusion due to
Coulomb scattering between electrons and
protons damps the magnetic fields around k È
1012 Mpcj1 E(1 þ z)/(104)^7/4. Therefore, the
energy density in magnetic fields remains finite.
Because magnetic fields at smaller scales result
from density perturbations in the early universe,
one needs to take into consideration the high-
energy effects neglected in the collision term:
e.g., relativistic corrections for the energy of
electrons and weak interactions along with the
Compton scattering. These effects would be-
come important when the temperature of the
universe was above È1 MeV, which corre-
sponds to the comoving wave number È105
Mpcj1. These effects can be safely neglected at
the scales considered here.
References and Notes
1. L. Biermann, A. Schlu¨ter, Phys. Rev. 82, 863 (1951).
2. L. Mestel, I. W. Roxburgh, Astrophys. J. 136, 615 (1962).
3. H. Hanayama et al., Astrophys. J. 633, 941 (2005).
4. L. M. Widrow, Rev. Mod. Phys. 74, 775 (2003).
5. N. Y. Gnedin, A. Ferrara, E. G. Zweibel, Astrophys. J. 539,
505 (2000).
6. R. M. Kulsrud, R. Cen, J. P. Ostriker, D. Ryu, Astrophys. J.
480, 481 (1997).
7. G. Davies, L. M. Widrow, Astrophys. J. 540, 755 (2000).
8. A. H. Guth, Phys. Rev. D 23, 347 (1981).
9. K. Sato, Mon. Not. R. Astron. Soc. 195, 467 (1981).
10. B. Ratra, Astrophys. J. 391, L1 (1992).
11. D. Lemoine, M. Lemoine, Phys. Rev. D 52, 1955 (1995).
12. K. Bamba, J. Yokoyama, Phys. Rev. D 69, 043507 (2004).
13. M. S. Turner, L. M. Widrow, Phys. Rev. D 37, 2743 (1988).
14. C. Caprini, R. Durrer, Phys. Rev. D 65, 023517 (2002).
15. E. R. Harrison, Mon. Not. R. Astron. Soc. 147, 279
(1970).
16. H. Lesch, M. Chiba, Astron. Astrophys. 297, 305 (1995).
17. Z. Berezhiani, A. D. Dolgov, Astropart. Phys. 21, 59 (2004).
18. S. Matarrese, S. Mollerach, A. Notari, A. Riotto, Phys.
Rev. D 71, 043502 (2005).
19. K. Takahashi, K. Ichiki, H. Ohno, H. Hanayama, Phys. Rev.
Lett. 95, 121301 (2005).
20. C. L. Bennett et al., Astrophys. J. 148, (suppl.), 1 (2003).
21. A.-C. Davis, M. Lilley, O. Tornkvist, Phys. Rev. D 60,
021301 (1999).
22. A. A. Starobinsky, Phys. Lett. B 117, 175 (1982).
23. H. Maki, H. Susa, Astrophys. J. 609, 467 (2004).
24. J. Silk, Astrophys. J. 151, 459 (1968).
25. R. Plaga, Nature 374, 430 (1995).
26. E. Waxman, P. Coppi, Astrophys. J. 464, L75 (1996).
27. S. Razzaque, P. Meszaros, B. Zhang, Astrophys. J. 613,
1072 (2004).
28. N. Gehrels, P. Michelson, Astropart. Phys. 11, 277 (1999).
29. K. Yamamoto, N. Sugiyama, H. Sato, Astrophys. J. 501,
442 (1998).
30. K.T. and K.I. are supported by a Grant-in-Aid for the Japan
Society for the Promotion of Science Fellows and are
research fellows of the Japan Society for the Promotion of
Science. N.S. is supported by a Grant-in-Aid for Scientific
Research from the Japanese Ministry of Education
(no. 17540276). This work was supported in part by
the Kavli Institute for Cosmological Physics through
the grant NSF PHY-0114422.
Supporting Online Material
www.sciencemag.org/cgi/content/full/1120690/DC1
SOM Text
References
28 September 2005; accepted 12 December 2005
Published online 5 January 2006;
10.1126/science.1120690
Include this information when citing this paper.
Reductive Cyclotrimerization of Carbon
Monoxide to the Deltate Dianion by
an Organometallic Uranium Complex
Owen T. Summerscales,1
F. Geoffrey N. Cloke,1
* Peter B. Hitchcock,1
Jennifer C. Green,2
Nilay Hazari2
Despite the long history of the Fischer-Tropsch reaction, carbon monoxide has proven remarkably
resistant to selective homologation under mild conditions. Here, we find that an organouranium(III)
complex induces efficient reductive trimerization of carbon monoxide at room temperature and
pressure. The result is a triangular, cyclic C3O3
2–, or deltate, dianion held between two uranium(IV)
units. The bonding within the C3O3
2– unit and its coordination to the two U centers have been
analyzed by x-ray diffraction and density functional theory computational studies, which show a
stabilizing C-C agostic interaction between the C3 core and one U center. Solution nuclear magnetic
resonance studies reveal a rapid equilibration of the deltate unit between the U centers.
C
arbon monoxide, usually obtained
from coal or natural gas, is an im-
portant industrial feedstock for the
production of hydrocarbons and oxygenates,
especially in times of limited crude oil sup-
ply, through the Fischer-Tropsch process.
This process uses CO/H2 as the feedstock
and is catalyzed by both homogeneous and
heterogeneous systems (1). Metal-catalyzed
coupling of CO with unsaturated organic sub-
strates (e.g., hydroformylation catalysis) repre-
sents a further industrially important method
of C-C bond formation (2). The concept of
reductively homologating CO directly to form
larger units of (CO)n is an appealing one;
however, the strength of the CO triple bond
has made this route difficult to achieve effec-
tively. The cyclic, aromatic, oxocarbon anions
CnOn
2– (n 0 3 to 6 in Scheme 1) have fas-
cinated organic chemists for many years (3, 4)
and, in principle, are accessible by such a
route. This would provide a facile synthesis
of fused carbon ring systems as building blocks
for further elaboration to more complex or-
ganic molecules (e.g., possible ring expansion
of A for prostaglandin synthesis).
The formation of salts of the croconate
and rhodizonate dianions—C and D—from the
reduction of CO by molten alkali metals Ewhich
also affords salts of the linear C2O2
2– ethyne-
diolate dianion (5)^ was first reported in the
early part of the 19th century (6), and there
is some nuclear magnetic resonance (NMR)
evidence for formation of trace amounts of
the deltate dianion A in a complex mixture of
products from the reaction of CO with Na-K
alloy in tetrahydrofuran (THF) (7). However,
salts of the trimer A have eluded crystallo-
graphic characterization or selective metal-
mediated synthesis. The squarate dianion B
has been generated by electrochemical meth-
ods using very high pressures of CO (8), and
surface catalysis routes using alkaline earth
oxides yield mixtures of various (CO)n
2– (n 0
2 to 6) species (9).
The idea of using low-valent f-element
complexes, which combine high reduction po-
1
Department of Chemistry, School of Life Sciences, Univer-
sity of Sussex, Brighton BN1 9QJ, UK. 2
Inorganic Chemistry
Laboratory, South Parks Road, Oxford OX1 3QR, UK.
*To whom correspondence should be addressed. E-mail:
f.g.cloke@sussex.ac.uk
REPORTS
www.sciencemag.org SCIENCE VOL 311 10 FEBRUARY 2006 829

2.
tentials with solubilizing ligands, to give iso-
lable (CO)n
2– species was demonstrated by Evans
et al., who reported that ESm(h-Cp*)2(THF)2^
(Cp* is pentamethylcyclopentadienyl) will
react under a 90 pounds per square inch over-
pressure of CO to give a doubly charged
ketene carboxylate unit (O2CCCO)2– bound
to samarium (10). Here, we extend that idea
to selectively synthesize and structurally char-
acterize the deltate dianion A directly from CO
under very mild conditions. Specifically, reduc-
tive cyclotrimerization of CO is induced by an
organometallic U(III) complex.
There is considerable current interest in the
binding of small molecules to U(III) centers,
e.g., CO (11), CO2 (12), and N2 (13). As part
of our continuing investigation into reduction
chemistry by low-valent uranium complexes,
we recently reported the reversible binding and
concomitant double reduction of dinitrogen by
the pentalene complex EU(h-C8H4
.)(h-Cp*)^
(14) E. is 1,4-bis(tri-isopropylsilyl)^. We ex-
plored the reactivity of the related silylated
cyclooctatetraene (COT) complex with CO;
the use of bulky tri-isopropylsilyl substituents
on the COT ring provides steric protection for
reactive metal centers and imparts high crys-
tallinity to the derived metal complexes. The
THF adduct EU(h-COT.)(h-Cp*)(THF)^ 1 was
obtained by a route similar to that employed for
the unsilylated analog EU(h-COT)(h-Cp*)(THF)^
(15); reaction of UI3 with KCp* in THF yielded
EUI2Cp*(THF)3^, which was treated in situ
with 0.8 equivalent of K2ECOT.^ in THF to give
1 as a dark purple, crystalline material in 39%
overall yield after recrystallization from pen-
tane at –50-C (Scheme 2). Compound 1 has
been structurally characterized by x-ray diffrac-
tion (16).
Compound 1 reacts with ambient pressures
(1 bar) of CO in pentane to give the dimeric
U(IV) complex 2 in 40% isolated yield. Crys-
tallization from diethyl ether at –50-C gave
dark red needles of 2 suitable for x-ray diffrac-
tion, which revealed a reductively homologated
CO trimer held between the two uranium(IV)
centers (Scheme 2 and Fig. 1) (17).
The structure shows a cyclic (C3O3) moiety
bound through the oxygen atoms in a bridg-
ing h1:h2 fashion to the two uranium cen-
ters. Metal-ligand bond lengths within the
EU(h-COT.)(h-Cp*)^ fragments reflect the steric
hindrance around the respective uranium cen-
ters, i.e., average U-COT. and U-Cp* ring
carbon distances are slightly longer around
the h2-bound U1 center (2.696 ) and 2.775 ),
respectively) than the h1-bound U2 (2.656 )
and 2.740 )). We propose an oxidation
state of (IV) for both U centers in 2 on the
basis of the density functional theory (DFT)
computational studies below. Oxidation from
U(III) in 1 to U(IV) in 2 is not, however,
reflected in the structural parameters; such
changes were similarly lacking in the conversion
of the pentalene complex EU(h-C8H4
.)(h-Cp*)^ to
EU(h-C8H4
.)(h-Cp*)^2(m-h2:h2-N2) (14), presum-
ably due to steric congestion in the uranium(IV)
dimers.
Because the other bond distances and angles
associated with the two EU(h-COT.)(h-Cp*)^
fragments are unexceptional, we now focus
on the U(C3O3)U core. The (C3O3) unit is planar,
with the two uranium centers lying slightly above
(U2, 0.0906 )) and below (U1, 0.1747 )) this
plane. The h1-uranium-oxygen distance (U2-O3)
of 2.183(3) ) is slightly longer than those
found in U(IV) aryloxides Ae.g., 2.120 ) in
EU(h-Cp*)3(OPh)^Z (Ph, is phenyl) (18), but
U1-O1 E2.516(3) )^ and U1-O2 E2.484(3) )^
are significantly longer. The C-O bond lengths
Ewhich lie between typical values for single
C-O (1.43 )) and double C0O (1.21 )) bonds^
show complementary variations to the U-O
bond lengths: C3-O3 is somewhat longer than
C1-O1 and C2-O2 (Fig. 1).
The triangular C3 skeleton in 2 is also
noticeably distorted (Fig. 1), with two short
C-C bonds and one long one, the latter
showing some interaction with U1. We probed
the bonding in this deltate diuranium com-
plex using DFT. The unsubstituted compound
EU2(h-COT)2(h-Cp)2(m-h1:h2-C3O3)^ II was
used as a model for compound 2. The ex-
perimental coordinates determined from the
solid state structure of 2 were used to de-
scribe the initial geometry of II, and a geom-
etry optimization was performed with no
symmetry constraints. In addition, a fragment
analysis was conducted in which II was
divided into two U(h-COT)(h-Cp) fragments
and a C3O3 fragment. It is often difficult to
achieve self-consistent field (SCF) convergence
in calculations involving open-shell actinides,
because there are a large number of orbitals close
to the Fermi surface of the molecule. In this case,
the geometry optimization was carried out as a
spin-restricted calculation, with electron density
smeared over the frontier orbitals to assist SCF
convergence (16). As a result, the frontier orbitals
have fractional occupancy; however, the molec-
ular geometry is typically not very sensitive to
the exact description of the f electrons (19).
There was good agreement between the
experimental and calculated bond distances
and angles in the U(C3O3)U core of II (table
S1), demonstrating that DFT can accurately
predict the structure of 2. The calculated struc-
ture reproduced the distortions in the C3 core,
with two short C-C bonds (C2-C3, 1.41 );
C1-C3, 1.41 )) and one long bond (C1-C2,
1.47 )). The C3O3 unit was planar, and the
C-O and U-O bond distances followed the
pattern observed in the solid-state structure.
Scheme 1.
Fig. 1. (Left) The mo-
lecular structure of 2
(thermal ellipsoids at
50%). (Right) View
of core structure (dis-
tances in A˚; values in
parentheses are errors
in the last significant
digit).
REPORTS
10 FEBRUARY 2006 VOL 311 SCIENCE www.sciencemag.org830

3.
There were minor differences between the
calculated and experimental U-COT and U-Cp
bond lengths and angles, suggesting that these
interactions are controlled by steric inter-
actions; however, the discrepancy is not suf-
ficiently great to invalidate the model. The
calculations indicate that each U is best de-
scribed as having two electrons localized in 5f
orbitals. Thus, the U configuration is consistent
with U(IV); chemical plausibility and theoret-
ical studies (20) also suggest that the deltate
structure is not likely to be stable in a neutral
state, whereas it is known in a dianionic 2-state.
The COT and Cp ligands bind to the U centers
as expected, with primarily a p interaction be-
tween U and Cp and a d interaction between U
and COT.
Previous work has shown that the highest
occupied molecular orbitals (MOs) of the re-
laxed (C3O3)2– anion (in D3h symmetry) are a
p orbital of a2µ symmetry (which confers aro-
maticity on the dianion) and a degenerate pair
of C-C bonding orbitals (e¶), which have C-O
antibonding character (21). On binding to the
two U fragments, the symmetry of the (C3O3)2–
anion is lowered, but the MOs of the distorted
fragment are clearly related to the MOs of
the relaxed parent fragment in D3h symmetry.
Fragment analysis indicates that four MOs of
the deltate anion have a notable interaction
with the two U-containing fragments (table S2
and fig. S1). The primary interaction involves
donation of electron density from the oxygen
atoms of the (C3O3)2– ligand to the U atoms.
However, a more complex interaction occurs
between the lengthened C-C bond and a U f
orbital. Figure 2 shows a representation of this
parent (C3O3)2– MO, which shows that the
orbital primarily consists of the lone pairs on
O1 and O2 and a bonding interaction between
C1 and C2. The parent orbital is bonding be-
tween C1 and C2 and is antibonding with re-
spect to the lone pairs on both O1 and O2. This
orbital interacts with an f orbital on U1 to form
an MO of II (which is also shown in Fig. 2).
This MO comprises not only a bonding inter-
action between U1 and O1, O2 but also (and of
more importance) a bonding interaction be-
tween U1 and C1, C2. The net effect of this
second interaction is to weaken the C1-C2
bond, because some of the electron density
between C1 and C2 is shared with U1, and to
strengthen the C1-O1 and C2-O2 bonds,
because the antibonding electron density is de-
creased. Thus, a C-C agostic interaction ap-
pears to be responsible for the distortion in the
C3 core observed in 2. The highly nodal char-
acteristics of the f orbital on U enable such an
interaction and, thus, actinides may be par-
ticularly suited to stabilize such CO homologs.
This finding suggests that tuning the ligand
environment around U(III) may give access to
higher homologs of the deltate dianion.
1H NMR solution studies of 2 in d8-toluene
do not reflect the asymmetry present in the
solid-state structure (Fig. 1). Only single en-
vironments for the Cp* ligands and the C8H6
.
ligands are observed at 25-C, indicating a
rapid (milliseconds on the NMR time scale)
fluxional process leading to equilibration of
the deltate unit between U1 and U2 at this
temperature. This finding was confirmed by a
study of the isotopically labeled complex
EU(h-COT.)(h-Cp*)^2 (m-h1 :h2 -1 3 C3 O3 )
2-13CO, prepared by reaction of 1 with 13CO:
2-13CO displays a singlet (n1/2 12 Hz) at d
225 parts per million for the deltate carbons
in the 13C NMR spectrum at 25-C. At –100-C,
both 1H and 13C spectra for 2-13CO display the
expected changes in chemical shifts with tem-
perature for a paramagnet, but, apart from
considerable broadening of the signals, there
is no evidence for freezing out of the fluxional
process at this temperature. Hence, these data
do not allow us to distinguish between the two
likely mechanisms for the equilibration of the
C3O3 unit in 2, i.e., a Bwindshield wiper[ mo-
tion of O1 (or O2) between U1 and U2, versus
a rotational Bpropeller[ mechanism involving
all three oxygen atoms.
References and Notes
1. C. K. Rofer-DePoorter, Chem. Rev. 81, 447 (1981).
2. B. Cornils, W. A. Herrmann, Applied Homogenous
Catalysis with Orgaonmetallic Compounds (VCH, Weinheim,
Germany, 1996).
3. R. West, R. Eggerding, J. Am. Chem. Soc. 98, 3641 (1976).
4. G. Seitz, P. Imming, Chem. Rev. 92, 1227 (1992).
5. W. Buchner, Helv. Chim. Acta 46, 2111 (1963).
6. L. Gmelin, Ann. Phys. Chem. 4, 31 (1825).
7. P. W. Lednor, P. C. Versloot, J. Chem. Soc. Chem.
Commun. 285 (1983).
8. G. Silvestri, S. Gambino, G. Filardo, G. Spadaro,
L. Palmisano, Electrochim. Acta 23, 413 (1978).
9. S. Coluccia, E. Garrone, E. Guglielminotti, A. Zecchina,
J. Chem. Soc., Faraday Trans. 1 77, 1063 (1981).
10. W. J. Evans, J. W. Grate, L. A. Hughes, H. Zhang,
J. L. Atwood, J. Am. Chem. Soc. 107, 3728 (1985).
11. I. Castro-Rodriguez, K. Meyer, J. Am. Chem. Soc. 127,
11242 (2005).
12. I. Castro-Rodriguez, H. Nakai, L. N. Zakharov,
A. L. Rheingold, K. Meyer, Science 305, 1757 (2004).
13. W. J. Evans, S. A. Kozimor, J. W. Ziller, J. Am. Chem. Soc.
125, 14264 (2002).
14. F. G. N. Cloke, P. B. Hitchcock, J. Am. Chem. Soc. 124,
9352 (2002).
15. A. R. Schake et al., Organometallics 12, 1497 (1993).
16. Materials and methods are available as supporting
material on Science Online.
17. Data collection was performed at 173(2) K on an
Enraf-Nonius Kappa charge-coupled device diffractometer
with graphite-monochromated MoKa radiation
(l 0 0.71073 A˚). The molecular structure was solved
by direct methods and refined on F2 by full-matrix
least squares techniques. For 2.Et2O: triclinic, P¯1111 (No. 2), a 0
10.7193(2) A˚, b 0 16.6281(2) A˚, c 0 23.2496(4) A˚,
a 0 99.768(1)-, b 0 95.816(1)-, g 0 91.500(1)-,
V 0 4058.97(11) A˚3, Z 0 2, R1 0 0.034, wR2 0 0.070.
18. M. R. Spirlet, J. Rebizant, C. Apostolidis, G. van den
Bossche, B. Kanellakopolus, Acta Crystallogr., Sect. C,
Struct. Commun. C46, 2318 (1990).
19. F. G. N. Cloke, J. C. Green, N. Kaltsoyannis, Organometallics
23, 832 (2004).
20. P. v. R. Schleyer, K. Najafian, B. Kiran, H. Jiao, J. Org.
Chem. 65, 426 (2000).
21. R. West, D. Eggerding, J. Perkins, D. Handy, E. C. Tuazon,
J. Am. Chem. Soc. 101, 1710 (1979).
22. We thank the Engineering and Physical Sciences
Research Council for financial support of this work.
Metrical data for 1 and 2 are available from the
Cambridge Crystallographic Data Centre under
reference numbers CCDC-292279 and CCDC-292280,
respectively.
Supporting Online Material
www.sciencemag.org/cgi/content/full/311/5762/829/DC1
Materials and Methods
SOM Text
Fig. S1
Tables S1 to S3
References
25 October 2005; accepted 12 December 2005
10.1126/science.1121784
Fig. 2. (A) Key orbital of
the distorted deltate di-
anion involved in an agostic
interaction with the U cen-
ter. Its parentage is one of
the degenerate e¶ C-C bond-
ing orbitals of (C3O3)2–. (B)
MO showing the agostic
interaction between the
deltate dianion and an f
orbital on U(I). Details of
these orbitals are given in
table S2.
Scheme 2.
REPORTS
www.sciencemag.org SCIENCE VOL 311 10 FEBRUARY 2006 831