Noble gas - Actinide complexes of the CUO molecule with multiple Ar, Kr, and Xe atoms in noble-gas matrices

Article abstract of DOI:10.1021/ja027819s

Laser-ablated U atoms react with CO in excess argon to produce CUO, which is trapped in a triplet state in solid argon at 7 K, based on agreement between observed and relativistic density functional theory (DFT) calculated isotopic frequencies ¹²C¹⁶C, ¹³C¹⁶O, ¹²C¹⁸O). This observation contrasts a recent neon matrix investigation, which trapped CUO in a linear singlet state calculated to be about 1 kcal/mol lower in energy. Experiments with krypton and xenon give results analogous to those with argon. Similar work with dilute Kr and Xe in argon finds small frequency shifts in new four-band progressions for CUO in the same triplet states trapped in solid argon and provides evidence for four distinct CUO(Ar)_4-n(Ng)_n (Ng = Kr, Xe, n = 1, 2, 3, 4) complexes for each Ng. DFT calculations show that successively higher Ng complexes are responsible for the observed frequency progressions. This work provides the first evidence for noble gas - actinide complexes, and the first example of neutral complexes with four noble gas atoms bonded to one metal center.

Full text of DOI:10.1021/ja027819s

Published on Web 02/12/2003
Noble Gas-Actinide Complexes of the CUO Molecule with
Multiple Ar, Kr, and Xe Atoms in Noble-Gas Matrices
Lester Andrews*,† Binyong Liang,† Jun Li,‡ and Bruce E. Bursten*,§
Contribution from the Department of Chemistry, UniVersity of Virginia,
CharlottesVille, Virginia 22904-4319, William R. Wiley EnVironmental Molecular Sciences
Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99352, and the
Department of Chemistry, The Ohio State UniVersity, Columbus, Ohio 43210
Received July 23, 2002 ; E-mail: lsa@virginia.edu
Abstract: Laser-ablated U atoms react with CO in excess argon to produce CUO, which is trapped in a
triplet state in solid argon at 7 K, based on agreement between observed and relativistic density functional
theory (DFT) calculated isotopic frequencies (12C16O, 13C16O, 12C18O). This observation contrasts a recent
neon matrix investigation, which trapped CUO in a linear singlet state calculated to be about 1 kcal/mol
lower in energy. Experiments with krypton and xenon give results analogous to those with argon. Similar
work with dilute Kr and Xe in argon finds small frequency shifts in new four-band progressions for CUO in
the same triplet states trapped in solid argon and provides evidence for four distinct CUO(Ar)4-n(Ng)n (Ng
) Kr, Xe, n ) 1, 2, 3, 4) complexes for each Ng. DFT calculations show that successively higher Ng
complexes are responsible for the observed frequency progressions. This work provides the first evidence
for noble gas-actinide complexes, and the first example of neutral complexes with four noble gas atoms
bonded to one metal center.
carbonyl complexes.10 The argon matrix experiments were
repeated under conditions where UO2 was virtually absent, and
Introduction
New paradigms in the basic structure and bonding of small
actinide-containing molecules have been discovered by com-
bining laser-ablation matrix-infrared experiments with relativistic
density functional calculations. Such collaborations have recently
identified uranium hydrides, oxides, nitroxides, and carbide-
oxides,1-5 and the thorium analogues.6-8 In particular, recent
reactions of laser-ablated U with CO in excess neon and
supporting electronic structure calculations found the linear
singlet CUO molecule to be the major primary reaction product
followed by linear triplet OUCCO as the major secondary
reaction product.5 Earlier argon matrix U + CO studies gave a
different major primary reaction product9 and the CUO iden-
tification in the argon matrix was subsequently questioned5
owing to the presence of UO and UO2, which can also form
the same argon matrix-isolated CUO species was observed. DFT
calculations show that this triplet CUO species is about one
kcal/mol higher than singlet CUO in the gas phase, but the
matrix experiments trap the triplet species in argon at 7 K and
the singlet in neon at 4 K.11 This rare “ground-state reversal”
requires closely spaced low-energy states, which are provided
by the actinide CUO molecule, and a specific interaction with
argon, which is provided by CUO(Ar)n complexes.12 Analogous
CUO(Kr)n and CUO(Xe)n complexes are formed competitively
with 1% Kr and Xe in argon, which suggests the formation of
distinct CUO(Ng)n complexes as is confirmed by DFT calcula-
tions to be reported here.
There are few examples of neutral argon complexes, namely
ArW(CO)5,13-15 ArBeO,16,17 and ArMX (M ) Cu, Ag, Au; X
) F, Cl),18-20 and one neutral compound HArF with a strongly
covalent (H-Ar)+ bond21,22 and a seminal position in noble gas
† Department of Chemistry, University of Virginia.
‡ William R. Wiley Environmental Molecular Sciences Laboratory,
Pacific Northwest National Laboratory.
§ Department of Chemistry, The Ohio State University.
(1) Souter, P. F.; Kushto, G. P.; Andrews, L.; Neurock, M. J. Am. Chem. Soc.
1997, 119, 1682.
(10) Andrews, L.; Zhou, M. F.; Liang, B.; Li, J.; Bursten, B. E. J. Am. Chem.
Soc. 2000, 122, 11 440 (U and Th + CO2).
(11) Andrews, L.; Liang, B.; Li, J.; Bursten, B. E. Angew. Chem., Int. Ed. Engl.
2000, 39, 4565.
(12) Li, J.; Bursten, B. E.; Liang, B.; Andrews, L. Science 2002, 295, 2242.
(13) Wells, J. R.;. Weitz, E. J. Am. Chem. Soc. 1992, 114, 2783.
(14) Sun, X.-Z.; George, M. W.; Kazarian, S. G.; Nikiforov, S. M.; Poliakoff,
M. J. Am. Chem. Soc. 1996, 118, 10 525.
(15) Ehlers, A. W.; Frenking, G.; Baerends, E. J. Organometallics 1997, 16,
4896.
(16) Thompson, C. A.; Andrews, L. J. Am. Chem. Soc. 1994, 116, 423.
(17) Veldkamp, A.; Frenking, G. Chem. Phys. Lett. 1994, 226, 11.
(18) Evans, C. J.; Lesarri, A.; Gerry, M. C. L. J. Am. Chem. Soc. 2000, 122,
6100.
(19) Evans, C. J.; Gerry, M. C. L. J. Chem. Phys. 2000, 112, 9363.
(20) Evans, C. J.; Gerry, M. C. L. J. Chem. Phys 2000, 112, 1321.
(21) Khriachtchev, L.; Pettersson, M.; Runeberg, N.; Lundell, J.; Ra¨sa¨nen, M.
Nature 2000, 406, 874.
(2) Zhou, M. F.; Andrews, L.; Ismail, N.; Marsden, C. J. Phys. Chem. A 2000,
104, 5495.
(3) Kushto, G. P.; Souter, P. F.; Andrews, L.; Neurock, M. J. Chem. Phys.
1997, 106, 5894.
(4) Zhou, M. F.; Andrews, L. J. Chem. Phys. 1999, 111, 11 044 (U and Th+NO
in neon).
(5) Zhou, M. F.; Andrews, L.; Li, J.; Bursten, B. E. J. Am. Chem. Soc. 1999,
121, 9712 (U + CO in neon).
(6) Souter, P. F.; Kushto, G. P.; Andrews, L.; Neurock, M. J. Phys. Chem. A
1997, 101, 1287.
(7) Kushto, G. P.; Andrews, L. J. Phys. Chem. A 1999, 103, 4836.
(8) (a) Zhou, M. F.; Andrews, L.; Li, J.; Bursten, B. E. J. Am. Chem. Soc.
1999, 121, 12188. (b) Li, J.; Bursten, B. E.; Zhou, M. F.; Andrews, L.
Inorg. Chem. 2001, 40, 5448 (Th + CO).
(9) Tague, T. J., Jr.; Andrews, L.; Hunt, R. D. J. Phys. Chem. 1993, 97, 10 920.
9
3126
J. AM. CHEM. SOC. 2003, 125, 3126-3139
10.1021/ja027819s CCC: $25.00 © 2003 American Chemical Society

Noble Gas−Actinide Complexes
A R T I C L E S
chemistry.23 Hence, the observation and characterization of
distinct CUO(Ng) and CUO(Ng)n actinide-argon complexes is
an important new addition to noble-gas and actinide chemistry.
Experimental and Computational Methods
The experiment for laser ablation and matrix isolation spectroscopy
has been described in detail previously.24 Briefly, the Nd:YAG laser
fundamental (1064 nm, 10 Hz repetition rate with 10 ns pulse width)
was focused on the rotating metal uranium or thorium target (Oak Ridge
National Laboratory) using low energy (1-5 mJ/pulse) such that a target
plume was barely observed. Laser-ablated metal atoms were co-
deposited with carbon monoxide (0.2 to 0.4%) in excess argon, krypton,
xenon, or argon doped with Kr or Xe onto a 7 K CsI cryogenic window
at 2-4 mmol/h for one h. Carbon monoxide (Matheson), 13C16O and
12C18O (Cambridge Isotopic Laboratories), and mixtures were used in
different experiments. Infrared spectra were recorded at 0.5 cm-1
resolution on a Nicolet 550 spectrometer with 0.1 cm-1 resolution using
a HgCdTe detector. Matrix samples were annealed at different
temperatures, and selected samples were subjected to broadband
photolysis by a medium-pressure mercury arc (Philips, 175W, globe
removed, 240-700 nm). Additional experiments were done with CBr4
added to the sample at 20% of CO concentration to trap laser-ablated
electrons and to affect the product chemistry.25
Relativistic density functional theory (DFT) calculations have been
performed using the Amsterdam Density Functional (ADF 2000) code,26
with the inclusion of the generalized gradient approach of Perdew and
Wang (PW91).27 This PW91 exchange-correlation functional has
recently been shown to be a more reliable choice than other functionals
for systems with weak interactions.28 The [1s2] cores for C, O, and Ne,
[1s2-2p6] core for Ar, [1s2-3d10] core for Kr, [1s2-4d10] core for Xe,
and [1s2-5d10] core for U were frozen. Slater-type-orbital (STO) basis
sets of triple-ú quality were used for the valence orbitals of all atoms
with d- and f-type polarization functions for the C, O, Ne, Ar, Kr, Xe,
and 2s2p2d2f diffuse functions for the noble-gas elements. Numerical
integration accuracy of INTEGRATION ) 10.0 was used throughout.
The structures of the calculated species were fully optimized with the
inclusion of scalar (mass-velocity and Darwin) relativistic effects, which
were treated within the Pauli formalism via the quasi-relativistic
method.29 Vibrational frequencies were determined via numerical
evaluation of the second-order derivatives of the total energies. Owing
to the use of very large basis sets, the basis-set-superposition-error is
expected to be quite small. Further computational details have been
described elsewhere.30
Figure 1. Infrared spectra in the 2050-1700 cm-1 region for laser-ablated
U atoms co-deposited with 0.3% CO in argon at 7 K: (a) sample deposited
for 70 min, (b) after annealing to 30 K, (c) after λ > 240 nm photolysis for
15 min, (d) after annealing to 40 K, (e) after λ > 240 nm photolysis for 15
min, and (f) after annealing to 43 K.
comparison to analogous neon matrix absorptions, and DFT
calculations. Similar U + CO experiments in krypton, in xenon,
and in argon doped with Kr and Xe, and complementary thorium
investigations will be compared. We will first show by
comparison of frequencies that most of the products are the same
in solid neon and argon but that CUO is unique by virtue of its
energetically near-degenerate singlet and triplet states and a
strong interaction with Ar atoms in the triplet CUO(Ar)n
complex. We will then examine the bonding and structures of
CUO(Ng)n (n ) 1-5) complexes using relativistic DFT
calculations.
OUCCO. The OUCCO product is common to both neon and
argon matrix experiments as demonstrated by a comparison of
the observed and DFT calculated frequencies.5 The strongest
OUCCO absorption at 2027.8 cm-1 in solid argon increases on
annealing and photolysis (Figure 1), shifts to 1963.7 cm-1 with
13CO, and gives a 2027.8, 2018.2, 1973.8, 1963.7 cm-1 1:1:1:1
quartet with 12CO + 13CO (Table 1). This band shifts to 2008.8
cm-1 with C18O and gives a 2027.8, 2008.8 cm-1 1:1 doublet
with C16O + C18O. The 12CO/13CO isotopic frequency ratio
(1.0326) and the C16O/C18O ratio (1.0095) show that this is
primarily a carbon motion. The mixed carbon isotopic quartet
and oxygen isotopic doublet demonstrate that two inequivalent
carbon atoms and one oxygen atom participate in this antisym-
metric C-C-O stretching mode. The 825.2 cm-1 band (and
822.2 cm-1 site) are associated with the 2027.8 cm-1 band on
photolysis and annealing. The 825.2 cm-1 band shifts to 781.1
cm-1 with C18O and the 16/18 ratio (1.0565) and absence of a
13CO shift are characteristic of a U-O stretching mode. A weak
1352.8 cm-1 band appears with the 2027.8 and 825.2 cm-1
bands on annealing and photolysis, and the isotopic shifts are
appropriate for a symmetric C-C-O stretching mode.
Results
The products of the reaction of laser-ablated uranium atoms
with carbon monoxide in solid argon will be identified from
the effects of isotopic substitution in their infrared spectra,
(22) Runeberg, N.; Pettersson, M.; Khriachtchev, L.; Lundell, J.; Ra¨sa¨nen, M.
J. Chem. Phys. 2001, 114, 836.
(23) Pyykko¨, P. Science 2000, 290, 117.
(24) (a) Burkholder, T. R.; Andrews, L. J. Chem. Phys. 1991, 95, 8697. (b)
Hassanzadeh, P.; Andrews, L. J. Phys. Chem. 1992, 96, 9177.
(25) (a) Zhou, M.; Andrews, L. J. Am. Chem. Soc. 1998, 120, 11 499. (b) Zhou,
M. F.; Andrews, L. J. Phys. Chem. A. 1999, 103, 2066.
(26) ADF 2000.02, SCM, Theoretical Chemistry, Vrije Universiteit, Amsterdam,
The Netherlands (http://www.scm.com). (a) te Velde, G.; Bickelhaupt, F.
M.; van Gisbergen, S. J. A.; Fonseca Guerra, C.; Baerends, E. J.; Snijders,
J. G.; Ziegler, T. J. Comput. Chem. 2001, 22, 931. (b) Fonseca Guerra, C.;
Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99,
391.
(27) (a) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244. (b) Perdew, J.
P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh,
D. J.; Foilhais, C. Phys. ReV. B 1992, 46, 6671.
(28) (a) Wesolowski, T. A.; Parisel, O.; Ellinger, Y.; Weber, J. J. Phys. Chem.
A 1997, 101, 7818. (b) Zhang, Y.; Pan, W.; Yang, W. J. Chem. Phys. 1997,
107, 7921.
The 2027.8, 1352.8, and 825.2 cm-1 argon-matrix bands
correspond to bands observed in the neon matrix at 2051.5,
1361.8, and 841.0 cm-1. These bands have the same isotopic
behavior in both matrices, and correspond reasonably well to
the DFT-calculated vibrations at 2125, 1393, and 897 cm-1 for
the linear triplet OUCCO molecule.5 Thus, it appears that this
secondary product is formed by the combination of CO and
CUO in both Ne and Ar matrices. Interestingly, however, in
(29) Ziegler, T.; Baerends, E. J.; Snijders, J. G.; Ravenek, W. J. Phys. Chem.
1989, 93, 3050.
(30) Li, J.; Bursten, B. E. J. Am. Chem. Soc. 1997, 119, 9021.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3127

A R T I C L E S
Andrews et al.
Table 1. Infrared Absorptions (cm-1) Observed from
Co-deposition of Laser-Ablated U Atoms with CO in Excess Argon
at 7 K
16
16
12C O/
12C O/
16
12C O
16
13C O
18
12C O
16
13C O
18
12C O
assignment
OUCCO
3355.0
2138.2
2034.7
2027.8a
2015.3
1972.4
1961.3
1932.6
1894.2
1872.1
1832.6b
1818.4
1766.0
1720.9c
1515.5d
1352.8
924.2
2091.1
1970.7
1963.7
1954.4
1930.2
1919.5
1890.5
1850.7
1831.8
1793.6
1779.1
1728.2
1683.2
1482.2
1316.7
891.9
889.3
908.6
836.8
840.8
834.1
825.2
822.2
819.6
2087.1
2015.6
2008.8
1994.7
1924.2
1913.2
1887.4
1850.4
1828.9
1788.6
1775.3
1723.7
1680.2
1479.6
1320.2
923.3
920.3
863.1
844.7
796.4
790.4
781.1
778.3
775.8
1.0225
1.0325
1.0326
1.0312
1.0219
1.0218
1.0219
1.0235
1.0220
1.0217
1.0221
1.0219
1.0224
1.0225
1.0274
1.0362
1.0358
1.0000
1.0188
1.0000
1.0000
1.0000
1.0000
1.0000
1.0196
1.0016
1.002
1.0225
1.0095
1.0095
1.0100
1.0251
1.0251
1.0239
1.0237
1.0236
1.0246
1.0243
1.0245
1.0236
1.0243
1.0247
1.0010
1.0009
1.0527
1.0092
1.0558
1.0553
1.0565
1.0564
1.0565
1.0469
1.0558
1.057
CO
OUCCO site
OUCCO
OUCCO(CO)x
U(CO)6 site
U(CO)6
U(CO)x
(UCO)
U(CO)x
U(CO)2
U(CO)x
U(CO)x
U(CO)x
(CO)2
Figure 2. Infrared spectra in the 1060-760 cm-1 region for laser-ablated
U atoms co-deposited with 0.3% CO in argon at 7 K: (a) sample deposited
for 70 min, (b) after annealing to 30 K, (c) after λ > 240 nm photolysis for
15 min, (d) after annealing to 40 K, (e) after λ > 240 nm photolysis for 15
min, and (f) after annealing to 43 K.
-
OUCCO
CUO-
921.1
CUO- site
(C2)UO2
CUO(Ar)x
OU(CO)x
OU(CO)x
OUCCO site
OUCCO
UO
908.6e
852.5
840.8
834.1
825.2
822.2
819.6
804.3
798.1
796 sh
788.8
796.8
794 sh
768.3
755.9
753 sh
CUO(Ar)x
CUO-
CUO- site
a Quartet 2027.8, 2018.2, 1973.8, 1963.9 cm-1 with 12CO+13CO and
doublet 2027.8, 2008.8 cm-1 with C16O+C18O. b Triplet at 1832.6, 1820.1,
1793.6 cm-1 with 12CO+13CO, and at 1832.6, 1818.7, 1788.6 cm-1 with
C16O+C18O. c Triplet at 1720.9, 1697.7, 1683.2 cm-1 with 12CO+13CO and
at 1720.9, 1695.4, 1680.2 cm-1 with C16O+C18O. d Triplet at 1515.5,
1497.4, 1482.2 cm-1 with 12CO+13CO. e Triplet at 908.6, 892.0, 863.1 cm-1
with C16O+C18O.
the Ar experiments there is no analogue to the 1047.3 cm-1
band of CUO that was observed in Ne. The 1000 cm-1 region
of the products of U + CO in Ar shows only a trace 1050.8
cm-1 absorption due to UN2.31
Figure 3. Infrared spectra in the 940-740 cm-1 region for laser-ablated
U atoms co-deposited with isotopic CO in argon: (a) 0.3% 12CO after
deposition and λ > 240 nm photolysis, (b) 0.2% 13CO after deposition and
λ > 240 nm photolysis, (c) 0.3% C18O after deposition and λ > 240 nm
photolysis, (d) 0.2% 12CO + 0.2% 13CO after deposition, and (e) 0.2%
C16O + 0.2% C18O after deposition.
CUO-. The broad weak bands at 924.2 and 798.1 cm-1
mimic neon-matrix bands at 929.3 and 803.3 cm-1 in their
isotopic shifts and frequency ratios (Table 1). The 924.2 cm-1
band is predominantly a U-C vibrational mode whereas the
798.1 cm-1 band is mainly a U-O mode. Both bands appear
as doublets in mixed isotopic precursor experiments (Figure 3),
so one C and O are involved in argon-matrix product as was
the case in neon. Both bands are destroyed by λ > 470 nm
photolysis. Experiments with electron-trapping reagents also
provide strong evidence that these bands are due to an anionic
species. Normally, we employ CCl4 dopant to capture electrons
and to prevent the formation of product anions.2,12,13 However,
inasmuch as CCl4 and its photolysis products mask the 925 and
800 cm-1 regions, we used CBr4 for these studies because the
calculated for bent CUO- at 996 and 822 cm-1. The 924.2 and
798.1 cm-1 bands in the argon matrix are therefore assigned to
the same bent CUO- species with further support by marked
decrease in the presence of a strong electron-trapping molecule.
This assignment leads to an intriguing question: Where in the
spectrum is the CUO species that must capture electrons to form
CUO-?
Triplet CUO in Argon. The strongest bands observed
following the reaction of U with CO in argon are those at 852.5
and 804.3 cm-1. This result is in agreement with our previous
experiments in an argon matrix,9 but these bands are substan-
tially different from those at 1047.3 and 872.2 cm-1, which
dominate the neon-matrix results and were assigned to linear,
singlet CUO.5 Nevertheless, several aspects of the argon-matrix
experiments make for a compelling case that the 852.5 and 804.3
cm-1 bands are due to the CUO moiety. First, we note the virtual
absence of UO and UO2 from the present infrared spectra, so
atomic uranium is the primary agent for reaction with CO.
Indeed, the experiments with isotopomers of CO lead to mixed
isotopic doublets for the 852.5 and 804.3 cm-1 bands [Figure
3d,e], demonstrating that single C and O atoms are involved in
+
CBr4, CBr3, and CBr3 absorptions appear at lower frequen-
cies.32 Doping the argon with CBr4 reduces the intensities of
the 924.2 and 798.1 cm-1 bands more than 10-fold relative to
the 852.5 and 804.3 cm-1 absorptions. The (CO)2 anion also
-
failed to appear in the presence of CBr4.13,14 We previously
assigned the neon-matrix bands to the CUO- anion on the basis
of their isotopic shifts and agreement with DFT frequencies
(31) Hunt, R. D.; Yustein, J. T.; Andrews, L. J. Chem. Phys. 1993, 98, 6070.
(32) Prochaska, F. T.; Andrews, L. J. Chem. Phys. 1977, 67, 1091.
9
3128 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
Table 2. Observed and Calculated Frequencies (cm-1) and
Isotopic Frequency Ratios for the Bond Stretching Modes of CUO
and CUO(Ng) Complexes
these two vibrational modes. Note the detailed annealing and
photolysis spectra in Figure 2, which clearly associates these
two bands with a common molecule. On annealing to 25 K,
the two bands sharpen but the integrated intensities slightly
decrease. The bands increase in concert on photolysis with λ >
470, 290, and 240 nm radiation in different experiments. Just
as found for linear singlet CUO absorbing at 1047.3 and 872.2
cm-1 in neon,5 the present CUO species does not form from
cold U and CO, but laser-ablated U or mercury arc irradiated
U react with CO to form the new species absorbing at 852.5
and 804.3 cm-1 in solid argon.
Typical neon-to-argon red shifts for small heavy metal species
include 7.4 cm-1 for HfO,33 8.3 cm-1 for ThO,4,8,34,35 23.9 and
12.5 cm-1 for NThO,4,35 and 19.0 and 11.7 cm-1 for bent triplet
CThO (see below).8 The 194.8 and 67.9 cm-1 neon-to-argon
red shifts observed here for CUO are far too large for a solvent
polarizability effect normally known as the “matrix shift.”36
Hence, we must find another explanation for the markedly
different CUO frequencies.
experimental
12C/13C
calculated (relativistic DFT)
νobs
16O/18O
νcalc
12C/13C
16O/18O
Ne matrix
1.0020
1.0361
CUO (1Σ+)
1.0023
1.0357
872.2
1047.3
1.0554
1.0010
874
1049
1.0549
1.0019
CUO (3Φ)
1.0127
1.0260
843
902
1.0514
1.0052
Ar matrix
1.0196
1.0188
CUO(Ar) (3A′′)
1.0125
804.3
852.5
1.0469
1.0092
834
887
1.0520
1.0045
1.0260
CUO(Kr) (3A′′)
1.0140
803.1
851.0
832
885
1.0512
1.0076
1.0247
CUO(Xe) (3A′′)
1.0155
801.3
848.0
1.0202
1.0181
1.0466
1.0095
830
879
1.0506
1.0061
1.0233
Relativistic DFT calculations were used to characterize
another CUO species that might be compatible with the above
observations. Because CUO only has three atoms, it can only
have four different isomers, i.e., CUO, UCO, UOC, and (U-
η2-CO). Our calculations indicate UCO, UOC, and the side-
bound (U-η2-CO) organoactinide complexes in triplet and
quintet states are much higher in energy than singlet linear CUO.
These molecules also have very different vibrational spectra than
observed in solid argon. Therefore, the spectra are due to the
CUO molecule in a different electronic state. Our calculations
indicate that the CUO molecule is unique in that it has a low-
lying triplet state only about 1 kcal/mol higher in energy than
the 1Σ+ ground state of linear CUO. This triplet state has
stretching frequencies calculated at 902 cm-1 (311 km/mol
intensity) and 843 cm-1 (115 km/mol). These calculated
frequencies are 5.8% and 4.8% higher than those of the argon-
matrix product bands and match the observed 3/1 relative
intensity ratio quite well. The calculated pure singlet state CUO
frequencies are each 0.2% higher than the neon matrix observa-
tions, which is in much better agreement, but the matrix
interaction and shift are typically greater for argon than for
neon.36 As will be detailed later in this paper, a specific
interaction of Ar atoms in the matrix with the CUO molecule
is sufficient to make the triplet form of a complex of CUO with
Ar atoms (hereafter denoted CUO(Ar)n) lower than that for
singlet CUO(Ar)n.11 To understand the basis for this unusual
“matrix interaction” (or “ground-state reversal”), additional DFT
calculations were performed on discrete molecular complexes
of CUO with a varying number of Ar atoms. For the 3A′′ state
of the “mono-argon adduct” CUO(Ar), the calculated C-U and
U-O stretching frequencies are 887 cm-1 (260 km/mol) and
834 cm-1 (137 km/mol). These calculated frequencies, which
are respectively 15 and 9 cm-1 lower than those of isolated
triplet CUO, are closer to the observed bands. The complexation
of additional Ar atoms to CUO leads to a further decrease in
the calculated frequencies; for the triplet C4V CUO(Ar)4 complex,
which will be discussed later, the frequencies are reduced
another 12 and 8 cm-1 more, respectively, and are closer still
to the observed bands.
In addition to 12C16O frequencies, the 13C16O and 12C18O
isotopic frequencies were also calculated as a diagnostic for the
normal mode description. As can be seen in Table 2, the
calculated vs observed isotopic frequency ratios are in very near
agreement ((0.4 cm-1 discrepancy for the major shifts of 36.5
and 45.8 cm-1) for the singlet linear species, but agreement for
the triplet is not as good (the major calculated shifts of 15.7
and 36.0 cm-1 are too high by 7 and 3 cm-1, respectively).
The calculated relative intensities are qualitatively correct for
the upper and lower triplet CUO bands: the observed integrated
intensity ratios 3.2/1.0 vs 1.8/1.0 (calcd) for 12C16O, 12/1 vs
22/1 (calcd) for 13C16O, and 1.1/1.0 vs 1.8/1.0 (calcd) for 12C18O.
The computed displacement coordinates show that the
calculated 887 cm-1 band of triplet CUO(Ar) is an antisym-
metric stretching mode with 2.5 times more carbon than oxygen
displacement and the calculated 834 cm-1 band is a symmetric
stretching mode with 1.7 times more oxygen than carbon motion.
The observed isotopic shifts demonstrate that there is consider-
ably more mode mixing in the triplet CUO(Ar)n complex, where
the two bond stretching frequencies are much closer (48.2 cm-1
separation) than for singlet CUO (175.1 cm-1 separation). For
singlet CUO in neon, the 1047.3 cm-1 band shifted 36.5 cm-1
with 13CO and 1.0 cm-1 with C18O, and the 872.2 cm-1
absorption shifted 1.7 cm-1 with 13CO and 45.8 cm-1 with C18O.
In contrast for triplet CUO in argon, the 852.5 cm-1 band shifted
15.7 cm-1 with 13CO and 7.8 cm-1 with C18O, and the 804.3
cm-1 absorption shifted 15.5 cm-1 with 13CO and 36.0 cm-1
with C18O. In addition, because the argon atoms interact most
strongly with the U and C atoms and the U-C bond order is
reduced in triplet CUO, the U-C stretching frequency is affected
much more than the U-O frequency relative to the CUO
molecule isolated in solid neon.
(C2)UO2. Ultraviolet photolysis in the neon matrix experi-
ments produced the lowest energy UC2O2 isomer, (η2-C2)UO2,
which was observed from its UO2 stretching modes at 922.1
and 843.2 cm-1.5 The higher-energy mode is red-shifted in solid
argon to 908.6 cm-1, which is a typical matrix shift. We were
unable to observe the lower-energy O-U-O stretching mode
(33) Chertihin, G. V.; Andrews, L. J. Phys. Chem. 1995, 99, 6356 and
unpublished results on HfO in solid neon at 965.7 cm-1
.
(34) Gablenick, S. D.; Reedy, G. T.; Chasanov, M. G. J. Chem. Phys. 1974,
60, 1167 (ThO in argon).
(35) Kushto, G. P.; Andrews, L. J. Phys. Chem. A 1999, 103, 4836 (ThO, NThO
in argon).
(36) Jacox, M. E. Chem. Phys. 1994, 189, 149.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3129

A R T I C L E S
Andrews et al.
Figure 5. Infrared spectra in the 860-780 cm-1 region for CUO formed
by laser-ablated U and CO reaction in excess argon: (a) 0.3% CO in pure
argon after sample deposition at 7 K, (b) 0.2% CO, 1% Xe in argon after
deposition at 7K, (c) after annealing to 40 K, (d) 0.2% CO, 2% Xe in argon
after deposition at 7 K, (e) after annealing to 40 K, (f) 0.2% CO, 3% Xe in
argon after deposition at 7 K, (g) after annealing to 40 K, (h) after annealing
to 45 K, and (i) after annealing to 50 K.
Figure 4. Infrared spectra in the 860-780 cm-1 region for CUO formed
by laser-ablated U and CO reaction in excess argon: (a) 0.3% CO in pure
argon after sample deposition at 7 K, (b) 0.3% CO, 2% Kr in argon after
deposition at 7 K, (c) after annealing to 30 K, (d) after full-arc photolysis,
(e) after annealing to 45 K, (f) after annealing to 48 K, (g) after annealing
to 50 K (intensity × 4), (h) after annealing to 52 K (intensity × 4), and (i)
0.4% CO in pure krypton after deposition at 7 K.
Table 3. Comparison of Neon, Argon, and Krypton Matrix
Frequencies (cm-1) Observed for Major Uranium and Thorium
Carbon Monoxide Reaction Products
in the argon matrix. The involvement of two CO submolecules
in the reaction and two equivalent O atoms in the product is
confirmed by the 908.6, 892.0, 863.1 cm-1 1:2:1 triplet observed
with a C16O/C18O mixture.
molecule
OUCCO
neon
argon
krypton
xenon
2051.5
841.0
2027.8
825.2
2021.4
826.7
2012.1
814.3
U(CO)2
U(CO)6
UCO
1840.2
(1959.5)
(1917.8)
889.5
1047.3
872.2
1832.6
1961.3
(1894.2)
819.6
(1824.3)
1964.5
(1886.2)
819.0
U(CO)x. The present argon-matrix experiments add some
information on uranium carbonyl complexes. The very strong
1961.3-1919.5 cm-1 and 1961.3-1913.2 cm-1 doublets for
mixed isotopic precursors after extensive annealing are in accord
with those expected for the triply degenerate mode for an
octahedral complex.37 We therefore propose that this final
product after annealing is U(CO)6, in agreement with previous
authors.38 The observed 1832.6 cm-1 band in argon is just below
the 1840.2 cm-1 band observed in neon. We previously proposed
that this band is due to the symmetric C-O stretching mode of
U(CO)2 in solid neon, and the triplet mixed isotopic spectra
support a like assignment in solid argon. The weak 1894.2-
1850.7 cm-1 doublet is probably due to UCO.5 In 12CO2/13CO2
experiments,10 a major product identified as O2UCO absorbs
at 1893.4-1851.0 cm-1 so this product must be considered here;
however, the virtual absence of UO2 casts doubt on the latter
possibility.
U + CO Reaction Products in Argon Doped with Kr and
Xe. To explore further the interactions of noble-gas atoms with
CUO, additional experiments were done in which the argon
carrier gas is doped with Kr or Xe. For example, an experiment
was done in which laser-ablated U atoms were generated with
0.3% CO and 1% Kr in argon. The same products observed in
pure Ar were observed as described above. In addition, new
features were observed near the 852.5 and 804.3 cm-1 bands
assigned to CUO(Ar)n. In particular, new bands appeared at
850.9 and 803.1 cm-1, and very weak features appeared at 849.3
and 801.8 cm-1 on annealing. Another experiment with 2% Kr
in argon and more extensive annealing yields the spectra shown
in Figure 4. This figure also shows spectra in pure argon and
in pure krypton [Figure 4a,i], which form the boundaries for
UO
CUO (1Σ+)
CUO(Ng)n (3A′′)
OThCCO
852.5
804.3
2024.0
804.2
842.3
797.1
2020.2
803.8
829.8
789.2
2048.6
822.5
Th(CO)2
1827.7
1775.6
1999.7
1817.5
887.1
1787.6
1716.6
1980.5
1800.8
876.5
1780.4
1704.1
1976.7
1792.2
874.2
Th(CO)6
ThCO
ThO
CThO(Ng)n (3A′)
812.2
793.2
784.6
617.7
606.0
596.1
mixed complexes. Upon annealing, two progressions of new
bands below the pure-argon bands evolve at 850.9, 849.3, 847.6,
846.0 cm-1 and at 803.1, 801.8, 800.6, 799.4 cm-1 [Figure 4b-
h]. After annealing to 52 K, the last bands in the progressions
become stronger than the initial pure argon bands at 852.5 and
804.3 cm-1 and the band positions approach those of the major
product in pure krypton at 842.3 and 797.1 cm-1
.
Five U + CO experiments were done with Xe in argon, and
Figure 5 illustrates representative spectra starting with the pure
argon reference [Figure 5a]. With 0.2% CO and 1% Xe, distinct
new absorptions were observed at 848.0 and 801.3 cm-1, and
new bands at 843.5 and 798.3 cm-1 appeared on annealing
[Figure 5b,c]. With 2% Xe in argon, the latter bands were much
stronger and new bands appeared at 839.4, 835.4 cm-1 and at
795.5, 792.6 cm-1. A sequence of annealing and λ > 240 nm
irradiation cycles increased the intensities of the new features
until they ultimately reach the total intensity observed in pure
argon [Figure 5d,e]. The observation that only 2% Xe in argon
is enough to cause half of the total absorbance to be due to the
(37) Darling, J. H.; Ogden, J. S. J. Chem. Soc., Dalton Trans. 1972, 2496.
(38) Slater, J. L.; Sheline, R. K.; Lin, K. C.; Weltner, W., Jr. J. Chem. Phys.
1971, 55, 5129.
9
3130 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
Table 4. Infrared Absorptions (cm-1) for CUO in Various Solid Noble-Gas Environments
Ng
U−C modea
U−O modea
Ne
Ar
Kr
Xe
1047.3
852.5
842.3
829.8b
872.2
804.3
797.1
789.2b
Ne, 1% Ar
Ar, 1% Kr
1047.3,857.2,854.3
872.2,808.3,806.4
852.5,850.9,849.3,847.6,846.0c
[844.7, 843.1, 841.5, 839.8, 838.2]d
852.5, 848.0, 843.6, 839.4, 835.4e
[844.7, 840.0, 835.6, 831.4]d
[836.8, 832.9, 829.1, 825.2, 822.0]f
842.3, 839.7, 837.2, 834.7, 832.2
804.3,803.1,801.8,800.6,799.4c
[768.3, 767.2, 766.0, 764.8, 763.7]d
804.3, 801.3, 798.4, 795.5, 792.6e
[768.3, 765.6, 763.0, 760.3]d
[788.8, 786.8, 785.4]f
Ar, 1% Xe
Kr, 2% Xe
797.1, 795.4, 793.8, 792.1, 790.5
a Dominant internal coordinate in the stretching mode. b Major site in pure xenon. c Last two bands observed with Ar, 2% Kr. d Absorptions for 12C18O.
e Last band observed with Ar, 2% Xe; latter bands stronger with Ar, 3% Xe. f Absorptions for 13C16O.
new features makes it evident that the Xe atoms interact with
the product molecule preferentially over the Ar atoms. With
3% Xe in argon, the satellite progressions are much stronger,
and after 50 K annealing the final 835.4 and 792.6 cm-1 bands
dominate the product spectrum.
Table 4 summarizes the new product absorptions, and
highlights several trends. First, from the spectra of CUO in pure
Ar, Kr, or Xe, we see that the vibrational bands of the CUO(Ng)n
species show substantial red-shifts from Ar to Kr to Xe: Relative
to the 852.5 and 804.3 cm-1 bands of CUO(Ar)n, the bands of
CUO(Kr)n are red-shifted by 10.2 and 7.2 cm-1, respectively,
and those of CUO(Xe)n are shifted by 22.7 and 15.1 cm-1
.
Second, the experiments in which argon is doped with Kr or
Xe show that the first CUO(Kr) and CUO(Xe) species in solid
argon show small red-shifts (1.5 and 1.2 cm-1 for Kr; 4.5 and
3.0 cm-1 for Xe) compared to CUO(Ar)n. Third, the four-band
progressions below the bands of CUO(Ar)n when Kr and Xe
are doped into argon strongly suggest the successive formation
of distinct CUO(Ng)n (Ng ) Kr, Xe; n ) 1, 2, 3, 4) complexes.
The observation of four distinct Kr and four distinct Xe complex
species provides strong evidence that four heavy Ng (Ar, Kr,
Xe) atoms complex with U in the intimate shell around triplet
CUO. These results suggest that the progressions result from
the successive replacement of coordinated Ar atoms with Kr or
Xe atoms, so we propose the formation of a series of six-
coordinate CUO(Ar)4-n(Ng)n (Ng ) Kr, Xe; n ) 1, 2, 3, 4)
complexes.
The 13C16O and 12C18O isotopic frequencies show that the
normal modes of CUO(Ar)3(Xe) are almost the same as for
CUO(Ar)n. The 12C16O/13C16O and 12C16O/12C18O frequency
ratios for the 848.0 cm-1 (1.0181 and 1.0202) and 801.3 cm-1
(1.009 52 and 1.0466) bands can be compared with the values
for CUO(Ar)n in Table 1. Isotopic frequency ratios for the 843.6
and 798.4 cm-1 bands of CUO(Ar)2(Xe)2 (1.0175, not observed
and 1.009 57, 1.0464) and for the 839.4 and 795.5 cm-1 bands
CUO(Ar)(Xe)3 (1.0172, not observed and 1.009 62, 1.0463)
continue the slight trend of decreasing carbon and increasing
oxygen participation in the upper (mostly U-C) mode and
increasing carbon and decreasing oxygen participation in the
lower (mostly U-O) mode with increasing number of Xe atoms.
U + CO Reaction Products in Solid Krypton and Xenon.
When laser-ablated uranium atoms were reacted with 0.4% CO
in krypton, the major product on deposition has absorptions at
842.3 and 797.1 cm-1. These bands were unchanged on
annealing to 25 and 35 K, but weak new bands at 1886.2,
1824.3, and 1758.2 cm-1 increased in intensity. Photolysis (λ
Figure 6. Infrared spectra in the 860-760 cm-1 region for laser-ablated
U atoms co-deposited with 0.4% 12CO + 0.4% 13CO in krypton at 7 K:
(a) sample deposited for 40 min, (b) after annealing to 25 K, (c) after
annealing to 35 K, (d) after λ > 470 nm photolysis for 30 min, and (e)
after annealing to 40 K.
> 470 nm) reduced the latter bands, tripled the former bands
and increased the intensity of bands at 2021.4 and 826.7 cm-1
10-fold. Further annealing to 40 and 50 K reduced the intensities
of all these bands and produced a very strong 1964.5 cm-1
absorption. Table 3 compares the frequencies observed in the
krypton matrix, which are shifted only slightly from those in
the argon matrix.
Figure 6 shows the spectra that result from the reaction of U
atoms with a 12CO/13CO mixture in krypton. The doublets at
842.3-827.9 cm-1 and 797.1-780.6 cm-1 confirm that this
product requires a single CO molecule; it is clearly the krypton-
matrix counterpart of CUO in solid argon. The 12CO/13CO ratios,
1.0174 and 1.0210, are almost the same as the argon matrix
values. Annealing had no effect but photolysis (λ > 470 nm)
again tripled these bands in concert. The 2021.4 cm-1 band gave
a quartet absorption with additional peaks at 2011.5, 1968.3,
and 1957.0 cm-1. The 1824.3 cm-1 band is probably due to
U(CO)2, and the 1758.2 cm-1 absorption arises from a higher
carbonyl. After annealing a strong, broad doublet was ob-
served at 1964.5-1924.5 cm-1, consistent with the presence
of U(CO)6.
Experiments were also performed in which U atoms reacted
with 0.2-0.5% CO in pure xenon. The CUO molecule appar-
ently can sit in two different sites within the matrix, with
absorptions at 834.6 and 797.2 cm-1 (Site 1) and at 829.8 and
789.2 cm-1 (Site 2). Annealing the matrix favors the peaks for
Site 2. On the basis of the results in the Xe-doped Ar and Kr
matrices, it seems likely that the dominant species is CUO(Xe)4.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3131

A R T I C L E S
Andrews et al.
Table 5. Infrared Absorptions (cm-1) from Co-deposition of
Laser-Ablated Th Atoms with CO in Excess Argon at 7 K
16
16
12C O/
12C O/
16
12C O
16
13C O
18
12C O
16
13C O
18
12C O
assignment
2034.5
2024.0
2023.0
1974.2
1972.3
1935.2
1889.4
1872.7
1806.1
1800.8
1797.6
1787.6
1775.1
1724.3
1716.7
1515.5
1342.2
878.8
2016.3
2005.4
2004.5
1925.8
1924.1
1.0090
1.0093
1.0092
1.0251
1.0251
OThCCO site
OThCCO site
OThCCO
Th(CO)6
Th(CO)6 site
?
1959.8a
1959.1b
1932.8
1930.8
1.0328
1.0326
1.0214
1.0215
?
?
1833.8
1766.5
1761.3
1758.2
1748.0c
1736.8
1687.3e
1678.9g
1482.3
1.0212
1.0224
1.0224
1.0224
1.0227
1.0221
1.0219
1.0225
1.0224
1763.4
1758.6
1755.5
1746.0d
1732.1
1682.3f
1676.4h
1479.6
1.0242
1.0240
1.0240
1.0238
1.0248
1.0250
1.0240
1.0243
ThCO site
ThCO site
ThCO
Th(CO)2 sym
Th(CO)2 site
Th(CO)2 site
Th(CO)2 antisym
Figure 7. Infrared spectra in the 860-780 cm-1 region for CUO formed
by laser-ablated U and CO reaction in mixed noble gas samples: (a) 0.2%
CO, 3% Xe in argon after deposition at 7 K, photolysis and annealing to
30 K (b) after annealing to 48 K, (c) 0.5% CO, 2% Kr in argon after
deposition at 7 K, photolysis and annealing to 43 K, (d) after annealing to
52 K, (e) 0.2% CO, 3% Xe in krypton after deposition at 7 K, photolysis
and annealing to 40 K, and (f) after annealing to 60 K.
-
(CO)2
OThCCO
ThO site
ThO
OThCCO
CThO
CThO-
?
878.8
876.4
804.2
792.9
748.2
729.1
583.8
832.0
829.7
761.3
751.2
709.4
692.9
605.7
1.0000
1.0000
1.0000
1.0003
1.0008
1.0004
1.0379
1.0563
1.0563
1.0564
1.0559
1.0555
1.0527
1.0003
876.4
804.2
793.2
748.8
729.4
606.0
Note in Table 4 that there is a gradual matrix-induced red-shift
of the peaks assigned to CUO(Xe)4 when produced via Xe
doping in Ar (835.4 and 792.6 cm-1), Xe doping in Kr (832.2
and 790.5 cm-1) and in pure Xe (829.8 and 789.2 cm-1).
U + CO Reaction Products in Krypton Doped with Xe.
Several experiments were done with U atoms, CO and 2-3%
Xe in krypton. In addition to the bands observed in pure krypton
at 842.3 and 797.1 cm-1, new four-band progressions at 839.7,
837.2, 834.7, 832.2 and 795.4, 793.8, 792.1, 790.5 cm-1
increased on successive annealing cycles. These progressions
are due to CUO(Kr)4-n(Xe)n complexes. Figure 7 contrasts later
annealing cycles for Xe in krypton with spectra for Kr in argon
and Xe in argon. In each case we see progressions that show
the successive replacement of lighter noble-gas atoms with
heavier ones upon annealing.
CThO
a Quartet at 2024.0, 2014.0, 1970.3, 1959.8 cm-1 in 12CO + 13CO.
b Quartet at 2023.0, 2013.1, 1969.8, 1959.1 cm-1 in 12CO + 13CO. c Triplet
at 1787.6, middle band (split into 1773.6, 1772.1), 1748.0 cm-1 in 12CO +
d
13CO. Triplet at 1787.5, middle band (split into 1773.1, 1771.5), 1746.0
cm-1 in C16O + C18O. e Triplet at 1724.3, 1700.9, 1687.3 cm-1 in 12CO +
13CO. f Triplet at 1724.3, 1697.8, 1682.3 cm-1 in C16O + C18O. g Triplet
at 1716.7, middle band (split into 1693.5, 1691.9), 1678.9 cm-1 in 12CO +
13CO. h Triplet at 1716.7, middle band (split into 1691.7, 1690.0), 1746.0
cm-1 in C16O + C18O.
Complementary Th + CO Species. We have previously
reported properties of the CThO molecule, which results from
the reaction of laser-ablated Th atoms with CO in neon.8 Like
CUO, CThO is the result of the insertion of an actinide atom
into the CO triple bond. However, because Th has two fewer
valence electrons than U, CThO has significant electronic and
structural differences with CUO. For the present studies, we
have prepared CThO in solid argon for comparison with the
results in a neon matrix. Table 5 lists the observed frequencies
of CThO in solid argon. Figure 8 shows the infrared spectra of
the products of the reaction between Th atoms and 12CO/13CO
and C16O/C18O mixtures. The strong 793.2-751.2 cm-1 doublet
observed with the C16O/C18O mixture indicates a single Th-O
stretching mode [Figure 8d,e,f]. Likewise, the medium intensity
606.0-583.8 cm-1 doublet observed with the 12CO/13CO
mixture is characteristic of a single Th-C stretch [Figure 8a,b,c].
These bands are associated with the same molecule by their
growth in concert on λ > 470 and 290 nm irradiation and their
decreases on annealing reacting with CO to form OThCCO.
The 793.2 and 606.0 cm-1 bands exhibit almost the same C16O/
C18O (1.0559) and 12CO/13CO (1.0380) isotopic frequency ratios,
respectively, as the 812.2 and 617.7 cm-1 bands in the neon
matrix bands (1.0559 and 1.0382). We previously assigned the
bands in the neon matrix to CThO and used DFT to predict
Figure 8. Infrared spectra in the 830-570 cm-1 region for laser-ablated
Th atoms co-deposited with isotopic CO in argon: (a) 0.2% 12CO + 0.2%
13CO after deposition for 70 min, (b) after λ > 470 nm photolysis for 15
min, (c) after annealing to 25 K, (d) 0.2% C16O + 0.2% C18O after
deposition for 70 min, (e) λ > 470 nm photolysis for 15 min, and (f) after
annealing to 25 K.
that it is a bent triplet molecule with calculated vibrational
frequencies of 811 and 621 cm-1.8 The 19.0 and 11.7 neon-to-
argon matrix redshifts are reasonable for this molecule,36 and
we conclude that the same bent triplet CThO molecule is
observed in solid argon as well as in solid neon. Indeed our
calculations found the singlet state to be 6.6 kcal/mol higher,8a
too high for matrix interactions to reverse the ground state.
The 2024.0 and 804.1 cm-1 absorptions increase together on
photolysis and annealing and are due to the OThCCO molecule,
9
3132 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
which was observed at 2048.6 and 822.5 cm-1 in solid neon.
This molecule is predicted by DFT to be a bent singlet molecule
fact significantly short, which is consistent with the spatially
more accessible 6d orbitals in Th atom. We also note that the
12CO, 13CO components for both bands are split about 1.5 cm-1
,
with intense stretching modes at 2082 and 807 cm-1 8b
.
With
the 12CO/13CO mixture, the upper band forms a 2024.0, 2014.0,
1970.3, 1959.8 cm-1 quartet characteristic of a vibration
involving two inequivalent carbon atoms, and with the C16O/
C18O mixture a 2024.0, 2005.4 cm-1 doublet indicates a single
oxygen atom. The 12CO/13CO and C16O/C18O ratios (1.0328 and
1.0093) are consistent with an antisymmetric C-C-O stretching
mode; further, these isotopic frequency ratios are nearly the same
as found for OThCCO in solid neon.8b Likewise the lower band
forms a 804.1, 761.2 cm-1 doublet for a single oxygen atom
and the C16O/C18O ratio 1.0564 indicates a Th-O stretching
mode. The 24.6 and 18.4 cm-1 neon-to-argon matrix shifts are
reasonable for OThCCO (compare 23.7 and 15.5 cm-1 for
OUCCO) and indicate that the same bent singlet OThCCO
molecule is trapped in solid neon and solid argon.
A weak 748.7 cm-1 band that results from the reaction of
Th with CO was destroyed by λ > 470 nm photolysis. This
band shifted to 709.9 cm-1 with C18O (C16O/C18O ratio 1.0546),
which is appropriate for a Th-O stretching mode. The 748.7
cm-1 band is in very good agreement with the 761.7 cm-1
observation of CThO- in solid neon and the computed Th-O
frequency of 762 cm-1 for the 2A′ ground state.8b We conclude
that CThO- is being produced in solid argon as in solid neon,
with a reasonable 13.0 cm-1 redshift upon changing from a neon
to an argon matrix.
which suggests a slight inequivalence in the two carbonyl
subunits. Finally, annealing the solid argon to 35-40 K produces
a very strong 1980.5 cm-1 absorption, which forms the strong
1980.5, 1936.3 cm-1 doublet with the 12CO/13CO mixture that
is expected for Th(CO)6. The shift from the 1999.7 cm-1 band
of Th(CO)6 in neon8b is appropriate for a usual matrix effect.
To examine further the matrix effects on the Th + CO
chemistry, laser-ablated thorium was also reacted with 0.4%
CO in krypton. The results are summarized in Table 3 and show
that most of the same species are formed. As in argon, CThO
was the major product observed on deposition, with slightly
redshifted bands in krypton at 784.6 and 596.1 cm-1. Annealing
to 25 and 30 K markedly increased a weak absorption at 1792.2
cm-1, which is assigned to ThCO. This species disappeared on
full-arc photolysis, whereas the 784.6, 596.0 cm-1 bands and
weak 2020.2, 803.8 cm-1 absorptions for OThCCO increased.
Annealing to 35 and 45 K regenerated the weak 1792.2 cm-1
band, sharpened the 784.6 and 596.0 cm-1 bands, increased the
bands at 2020.2 and 803.8 cm-1, and produced a strong 1976.7
cm-1 absorption for Th(CO)6. An experiment with 0.4% 12CO
+ 0.4% 13CO in krypton gave the major CThO product at 784.6
cm-1 with a weak doublet at 596.1-574.3 cm-1. The 12CO/
13CO ratio, 1.0380, is unchanged from the argon matrix ratio.
A ThCO doublet at 1792.2-1753.0 cm-1 increased on 25 K
annealing and decreased on λ > 470 nm photolysis. A sharp
OThCCO quartet observed at 2020.2, 2009.5, 1967.0, 1955.4
cm-1 increased on λ > 470 nm photolysis along with an 803.8
cm-1 band. New 2/1 relative intensity bands observed at 1780.3
and 1704.1 cm-1 decrease slightly on λ > 470 nm photolysis,
shift to 1741.0 and 1666.8 cm-1 with 13CO, and show new
1765.9 and 1681.2 cm-1 intermediate components with 12CO
The previous neon-matrix experiments provide clear evidence
for ThCO with a strong 1817.5 cm-1 absorption. This result
was corroborated by DFT calculations, which predict a 1790
3
cm-1 fundamental for Σ- ThCO.8 The present argon matrix
experiments reveal this band split at 1800.8, 1797.8 cm-1 with
the doublet mixed isotopic structure indicative of a single
carbonyl ligand. In solid argon as in neon,8 λ > 470 nm
photolysis isomerizes ThCO to CThO. Again, the 16.7-19.7
cm-1 neon-to-argon shift is reasonable.36
+
13CO. These bands are appropriate for Th(CO)2 in solid
krypton. Final annealing produced a very strong broad 1976.7-
1934.5 cm-1 doublet for Th(CO)6.
The CO stretching modes of the Th(CO)2 molecule were
found in solid neon at 1827.7 and 1775.6 cm-1. Analogous
bands are observed in solid argon at 1787.6 and 1716.7 cm-1
with a 5/2 intensity ratio. As in the neon matrix, photolysis leads
to disappearance of these bands and the appearance of OThCCO.
Further, the isotopic structures and ratios are comparable to those
observed for the bands of Th(CO)2 in the neon matrix.8b We
therefore assign the 1787.6 and 1716.7 cm-1 bands to the
symmetric and antisymmetric C-O stretching modes for
Th(CO)2 solid argon. As found in solid neon, the symmetric
mode is more intense for this acute M(CO)2 species.8b In this
Discussion
Reactions and Products in Different Noble-Gas Matrices.
The experimental observations reported here and earlier5,11,12
indicate that laser-ablated U atoms with excess energy or
photoexcited U atoms react with CO in noble-gas carriers to
form CUO, which is trapped in the solid noble-gas matrix. In
excess argon, krypton, or xenon, CUO is trapped in a triplet
state in the matrix (eq 1). These results contrast those in solid
neon, in which CUO is trapped as a singlet molecule, although
doping of neon with 1% Ar is sufficient to trap CUO in its
triplet state.12 The insertion of a metal atom into the robust triple
bond of CO is unique and has only been observed for the
actinides U and Th5,8 and for niobium upon photoisomerization
of NbCO.39 In argon and krypton the CUO molecule captures
an electron to form the same bent doublet CUO- anion that
was observed in solid neon (eq 2). Our relativistic DFT
calculations find eq 2 to be exothermic by 24 kcal/mol, so λ >
470 nm radiation is more than capable of causing electron
photodetachment. CUO in argon and krypton matrices also
reacts with CO to produce the linear triplet OUCCO molecule,
case, however, the neon-to-argon shifts of 40.1 and 58.9 cm-1
,
respectively, are larger than those observed for the other ThCO
species. We have previously predicted that Th(CO)2 has an
unusual structure with an acute 50° C-Th-C bond angle that
fosters a direct interaction between the carbon atoms. This
structure leaves a large coordination vacancy on the side of the
Th atom opposite the two carbonyl ligands. It may be the case
that Ar atoms from the matrix are providing a nonnegligible
interaction with the Th atom, causing the large matrix shifts.
Indeed, our calculations indicate that the Ar atoms can bind to
Th at a distance of 3.28 Å in ArnTh(CO)2 (n ) 2, 4). When
considering that the covalent radius is almost 0.2 Å larger for
Th atom than for U atom, the optimized Th-Ar distance is in
(39) Zhou, M. F.; Andrews, L. J. Phys. Chem. A 1999, 103, 7785 (Nb, Ta +
CO).
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3133

A R T I C L E S
Andrews et al.
a reaction also observed in solid neon (eq 3). We calculate that
this reaction is exothermic by 69 kcal/mol
U* + CO
U* + CO
9
Ar8 [(U)(CO)]*
Kr8 [(U)(CO)]*
Xe8 [(U)(CO)]*
9
Ar8 CUO(Ar)n (3A′′) (1a)
Kr8 CUO(Kr)n (3A′′) (1b)
Xe8 CUO(Xe)n (3A′′) (1c)
∆E ) -24 kcal/mol (2)
9
9
U* + CO
9
9
Figure 9. Structures of singlet and triplet CThO(Ar) and CUO(Ar)
calculated by relativistic DFT (atoms: black C, blue Th, U, red O, yellow
Ar).
CUO + e- f CUO-
CUO + CO f OUCCO ∆E ) -69 kcal/mol (3)
evidence has also been presented for UO2 and for FeCO trapped
in different electronic states in solid neon and argon.2,40
Electronic transitions are typically red-shifted in solid argon
relative to solid neon owing to a stronger interaction between
the excited-state molecule and the matrix. For example, the near-
ultraviolet transitions for TaO are red-shifted 200-300 cm-1
from neon to argon.41 In the present case, it would be intriguing
to observe the T r S electronic transition of CUO in solid neon
or the S r T electronic transition of CUO(Ar)n in argon.
Unfortunately, these electronic transitions, which have Φ
symmetry in the C∞V single-group and Φ + ∆ + Γ symmetries
in the C∞V* double-group, are not electric-dipole allowed. As a
result, no such bands were found in the 4000-450 cm-1 range.
In contrast to the results for CUO, all of the thorium-
containing species observed here exhibit typical matrix shifts
from neon to argon to krypton with virtually unchanged isotopic
frequency ratios. These results indicate that the same electronic
state of each species is trapped in the three matrices. In the
case of isolated CThO, we calculate that the lowest singlet state
The major physical differences of interest here among the series
of noble gases are their polarizabilities and freezing points,
which determine the inertness and rigidity of the solids formed
as matrix hosts. The initial product yield is higher in neon
because of increased reagent diffusion and reaction during the
slower condensation of neon at 4 K than of argon and krypton
at 7 K. In contrast, more diffusion can take place in annealed
argon and krypton due to the wider temperature range of the
solid and the greater interatomic separations in the solid. Hence,
the formation of U(CO)6 is greatly enhanced on annealing in
solid argon (1961.3 cm-1), in agreement with previous work-
ers,38 and in solid krypton (1964.5 cm-1) relative to the results
in solid neon.
The matrix photochemistry is slightly different in the different
noble-gas matrices. The yield of the UV photolysis products
OUCCO and (C2)UO2 is higher in solid neon because the neon
matrix is more UV transparent, but OUCCO is clearly formed
in solid argon and krypton as well from the isomerism of
U(CO)2. No (C2)UO2 was produced in solid krypton. The
CUO- anion is photodetached by λ > 470 nm radiation in solid
argon whereas λ > 380 nm photolysis was required in solid
neon.
The different noble gases lead to small differences in the
vibrational frequencies for OUCCO (2051.5, 1361.8, 841.0 cm-1
in neon; 2027.8, 1352.8, 825.5 cm-1 in argon; 2021.4, 826.7
cm-1 in krypton), CUO- (929.3, 803.3 cm-1 in neon; 924.2,
798.1 cm-1 in argon) and (C2)UO2 (922.1 cm-1 in neon; 908.5
cm-1 in argon). These are all typical matrix shifts that can be
ascribed to the different polarizabilities of the matrix atoms.36
Furthermore, the isotopic frequency ratios are essentially
unchanged so the normal modes are not altered by the matrix
cage. The results for CUO are in marked contrast to those for
the other molecules; however, the frequencies observed in solid
argon (852.5, 804.3 cm-1) and krypton (842.3, 797.1 cm-1) are
very similar but differ substantially from those observed in solid
neon (1047.3, 872.2 cm-1). In addition, the isotopic frequency
ratios in argon and krypton are markedly different from those
in neon, indicating different normal modes and hence different
electronic states of CUO as discussed above, are trapped in neon
and the heavier noble gas solids. Our DFT calculations on the
isolated CUO molecule find a triplet state about 1 kcal/mol
above the singlet ground state,5 but we have also pointed out
that complexation of CUO with noble-gas atoms preferentially
lowers the triplet state more than the singlet state.12 We will
discuss this effect in greater detail below. It is noteworthy that
3
is 6.6 kcal/mol higher in energy than the A′ ground state,8a
and the bent triplet is apparently the ground state in solid neon,
argon, and krypton. Indeed, when an Ar atom is included in
the calculation, the corresponding triplet of CThO(Ar) is still
3.8 kcal/mol more stable than its singlet counterpart; Ar-Th
bonding is not sufficient to alter the ground state of CThO.
Interestingly, the CThO(Ar) molecule possesses shorter Th-
Ar distances (3.13 and 3.02 Å) and larger CThO-Ar binding
energies (4.7 and 7.6 kcal/mol for the triplet and singlet) than
the CUO(Ar) counterparts, consistent with the more diffuse 6d
orbitals in Th than in U. The structures of the 3A′ and 1A′ states
of CThO(Ar) are shown in Figure 9a,b.
Structure and Bonding of CUO(Ng)n (n > 1) Complexes.
The energetic closeness of the singlet and triplet states of CUO,
and the fact that the presence of Ar, Kr, or Xe atoms is sufficient
to reduce and eventually reverse the relative energies between
these states, are the electronic effects that allowed us to uncover
the remarkable interactions between the CUO molecule and
noble-gas atoms via vibrational spectroscopy. To understand
better the state energetics and the interactions between noble-
gas atoms and the CUO molecule, we have performed relativistic
DFT calculations using very diffuse basis sets to improve the
performance for noble-gas systems.42,43
We have previously suggested that the interaction of CUO
with a single Ar atom will stabilize triplet CUO(Ar) more than
(40) Zhou, M.; Andrews, L. J. Chem. Phys. 1999, 110, 10370 (Fe + CO).
(41) Weltner, W., Jr.; McLeod, D., Jr. J. Chem. Phys. 1965, 42, 882 (Table 5).
(42) Frenking, G.; Cremer, D. Struct. Bonding (Berlin) 1990, 73 17.
9
3134 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
for the singlet form of the molecule.12 The experimental results
presented here provide indisputable evidence for the binding
of multiple noble-gas atoms to the CUO molecule, making it
unlikely that isolated “mono-noble-gas” CUO(Ng) species exist
in the heavier noble gas solids. Nevertheless, the results we
obtain on the series of CUO(Ng) molecules are instructive
inasmuch as they show the formation of U-Ng bonds and the
change in the ground-state multiplicity upon this bonding. We
will therefore begin with a discussion of the bonding in
CUO(Ng) model complexes, and will follow that with a
description of the bonding in the CUO(Ng)n complexes.
the triplet state more than the singlet state, which causes the
triplet state of CUO(Ar) to be energetically close to the singlet
state.
There are two primary reasons that the triplet state of CUO
interacts more strongly with an Ar atom than does the singlet
state, both of which arise because the primary mode of the
interaction of the Ar atom with CUO is donation from a filled
orbital of Ar atom into an empty primarily 6d orbital localized
on the U atom. First, because the U-C distance is longer in
triplet CUO, the Ar atom experiences less repulsion from the
U and C atoms in the triplet state than in the singlet state. As
a consequence, the Ar atom is able to approach the U atom
more closely, as is clear from the U-Ar distances in the
converged structures of triplet (U-Ar ) 3.16 Å) and singlet
(U-Ar ) 3.24 Å CUO(Ar) (Figure 9c,d). Indeed, geometry
optimizations of the CUO(Ar) complex indicate that the U-Ar
distance decreases further by 0.05 Å when the second electron
is transferred from the U-C σ-bonding orbital to the U 5f
orbitals. Second, the transfer of an electron from a U-C bonding
orbital to a U 5f orbital affects a partial reduction of the U atom.
The expansion of the U 6d orbitals upon this reduction leads to
greater overlap and a stronger interaction between the U and
Ar atoms. Thus, the U-Ar bond is essentially the result of a
Lewis acid-base interaction that is facilitated by the longer
U-C bond and the more diffuse U atom in the triplet state. It
is remarkable that this weak interaction is sufficient to reverse
the energies of the triplet and singlet states when enough Ar
atoms are bonded.
Calculations on CUO(Ar) were carried out with the CUO
molecule in its 1Σ+ (1A′) ground state and its low-lying 3Φ (3A′′)
excited state, where the Greek state designations are for the
linear (C∞V symmetry) geometry of CUO whereas the designa-
tions in parentheses are appropriate for Cs symmetry. We find
that CUO(Ar) is a planar molecule with a 3A′′ ground state and
a bent CUO fragment ( CUO ) 166.5°) bound to the Ar atom
primarily through the U atom (Figure 9c). At the DFT level in
the absence of spin-orbit coupling, the 3A′′ state of CUO(Ar)
is calculated to be slightly lower than the 1A′ state. The U-Ar
distance in the converged triplet structure is 3.16 Å, which is
smaller than the sum of the crystal radii of the two elements
(3.25 Å),44 but much longer than the 2.49 Å length45 of the
U-Cl chemical bond in UCl6. Thus, we indeed have an
equatorial complex, rather than a chemical bond.
For both isolated CUO and CUO(Ar), the triplet state
corresponds to the transfer of an electron from an orbital that
has significant U-C σ bonding character to a nonbonding U
5f orbital. As a consequence, the U-C bond is longer and
weaker in the triplet states of CUO and CUO(Ar) than in the
singlet state. The U-C bond is nearly 0.1 Å longer in triplet
CUO (U-C ) 1.846 Å) than in singlet CUO (U-C ) 1.747
Å). The U-C bonds in both states of CUO(Ar) are slightly
longer than those in the corresponding state of CUO and show
the same magnitude of lengthening from singlet to triplet (U-C
in singlet CUO(Ar) ) 1.752 Å; U-C in triplet CUO(Ar) )
1.858 Å). For both CUO and CUO(Ar), the U-O bond length
is only slightly affected by the change from singlet to triplet.
In CUO, for example, the calculated U-O bond length is 1.804
Å in the singlet state and 1.818 Å in the triplet state.
The fact that the bond between U and Ar in CUO(Ar) is a
Lewis acid-base interaction led us to predict that the bonding
would be even stronger for heavier, more polarizable noble
gases. Indeed, our calculations on CUO(Kr) and CUO(Xe) find
that the U-Ng bond strength increases steadily from Ar to Kr
to Xe, as gauged by the calculated dissociation energy for the
following reaction
CUO(Ng) (3A′′) f CUO (3Φ) + Ng (1S)
(4)
We find that the calculated dissociation energies of CUO(Ng)
(3A′′) for eq 4 are 3.7, 4.9, and 7.3 kcal/mol for Ar, Kr, and
Xe, respectively, corresponding to CUO(1Σ+)-Ng(1S) binding
energies of 3.2, 4.4, and 6.8 kcal/mol. We also find that the
stronger U-Ng interactions for Kr and Xe lead to greater
stabilization of the triplet state of CUO(Ng) relative to the singlet
state.
Structure and Bonding of CUO(Ng)4 Complexes. In
matrices with 1-2% Kr in argon, 1-3% Xe in argon, and 2%
Xe in krypton the bands due to CUO are split into new four-
band progressions suggesting that at least four Ng atoms can
complex to CUO depending on the experimental conditions.
The involvement of multiple noble-gas atoms has precedent in
other systems: Xenon atoms appear to solvate HXeBr and
HXeCl in neon matrices even with small amounts of Xe
depending on concentration and annealing history.46 To explore
further the bonding of multiple noble-gas atoms to CUO, we
calculated structures of CUO(Ng)n (Ng ) Ar, Kr, Xe; n ) 2-6)
complexes. Figure 10 presents plots of the total Ng binding
energy as a function of n for the triplet CUO(Ng)n complexes,
The fact that the singlet and triplet states of CUO are so close
in energy is a consequence of a delicate balance between
opposing effects, much as is the case for high- and low-spin
transition metal complexes. In the closed-shell singlet state, the
U-C bonding is significantly stronger than in the triplet state,
wherein one of the U-C bonding electrons has been transferred
to a nonbonding U 5f orbital. This effect is countered by
increased exchange stabilization and decreased Coulomb repul-
sion in the open-shell triplet state. In isolated CUO, the increased
strength of the U-C bonding dominates, and the singlet state
is the ground state. When Ar atoms are present, however, the
interaction of the Ar atoms with the CUO molecule stabilizes
(43) Lundell, J.; Chaban, G. M.; Gerber, R. B. J. Phys. Chem. A 2000, 104,
7944.
(44) Our preliminary calculations using the much more accurate coupled-cluster
CCSD(T) method predict similar U-Ar distances of 3.30 and 3.21 Å for
the singlet and triplet states, respectively. For the radii, see Winter, M. J.
WebElements [http://www.webelements.com/] (WebElements Ltd., Shef-
field, UK, 2001).
(46) Lorenz, M.; Ra¨sa¨nen, M.; Bondybey, V. E. J. Phys. Chem. A 2000, 104,
(45) Schreckenbach, G. Inorg. Chem. 2000, 39, 1265.
3770.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3135

A R T I C L E S
Andrews et al.
the primary coordination sphere of CUO. In support of this
notion, preliminary calculations on the total binding energies
of CUO(Kr)(Ar)n and CUO(Xe)(Ar)n indicate that n ) 3 is the
optimum value, i.e., when a single Kr or Xe atom is coordinated
to CUO, the U atom prefers a six-coordinate octahedral
arrangement of bonded atoms. The experiments using ma-
trixes of dilute Kr in Ar or dilute Xe in Ar suggest the se-
quential complexation of one, two, three, or four Kr or Xe atoms,
where the Kr or Xe atoms apparently displace Ar atoms. We
therefore propose that the progressions in Figures 4 and 5 are
due to the sequential formation of the six-coordinate complexes
CUO(Ar)4-n(Kr)n and CUO(Ar)4-n(Xe)n (n ) 1, 2, 3, 4) upon
annealing of the matrix. The experimental data suggest that the
U-C and U-O stretching frequencies are perturbed more by
the replacement of Ar with Xe than by Kr. The U-C and U-O
frequencies decrease by roughly 1.6 and 1.2 cm-1, respectively,
upon replacement of Ar by Kr, and by 4 and 3 cm-1
respectively, upon replacement of Ar by Xe (Table 4).
,
Table 6 presents the calculated binding energies, geometries,
and vibrational frequencies of the CUO(Ar)4-n(Xe)n (n ) 1, 2,
3, 4) complexes. Results for CUO(Ar)4,5 and CUO(Kr)4 are
included for completeness. Several aspects of the calculated
properties provide strong support for the notion of successive
substitution of Ar by Xe in these complexes. First, the complexes
are all predicted to have a triplet ground state (ignoring spin-
orbit effects) and therefore the vibrational frequencies are
distinctly different than for CUO in solid Ne, which has a singlet
ground state.5,11,12 Second, with one exception, the total binding
energy increases monotonically with increasing Xe substitution.
The calculations on CUO(Ar)2(Xe)2 were not well-behaved with
this basis set, once again due to poor fitting in the Coulomb
potentials, and we suspect that the calculated binding energy
of 26.6 kcal/mol for trans-CUO(Ar)2(Xe)2 is somewhat too high.
Finally, the calculated frequencies of the CUO(Ar)4-n(Xe)n
series exhibit the monotonic red-shift upon successive replace-
ment of Ar atoms by Xe atoms. The calculated red-shifts are
somewhat smaller than those observed experimentally: The
calculated U-C and U-O stretching frequencies shift 11 and
5 cm-1 lower from CUO(Ar)4 to CUO(Xe)4 vs experimental
shifts of 17 and 12 cm-1 from CUO(Ar)n to CUO(Xe)4. The
calculated U-C and U-O frequencies are roughly 30 to 35
and 20 to 25 cm-1 higher than those observed experimentally,
respectively, which may be due in part to solvation effects of
the argon matrix and the omission of spin-orbit coupling
effects. Nevertheless, the calculated frequencies and predicted
red-shifts are in satisfyingly good accord with the experimental
observations.
Figure 10. Binding energies for CUO(Ar)n, CUO(Kr)n, and CUO(Xe)n in
triplet states.
which were calculated as the energy change for the following
process
CUO(Ng)n f CUO (1Σ+) + n Ng (1S)
(5)
For the CUO(Ar)n complexes, the magnitude of the total
binding energy increases monotonically through n ) 5. When
n ) 6, however, the magnitude of the binding energy decreases,
indicating an overall repulsion for binding the sixth Ar atom.
Thus, these results suggest that, in the gas phase, CUO(Ar)5
represents the optimum coordination of Ar atoms around the
CUO molecule. However, although we predict that five Ar
atoms can bind to CUO in the gas phase, we believe that the
octahedral symmetry expected in the face-centered-cubic solid
argon lattice and the effect of solvation by the next layer of
argon atoms could favor CUO(Ar)4 in pure solid argon. At
present, we do not have experimental evidence to determine
whether CUO in solid argon exists preferentially as CUO(Ar)4
or CUO(Ar)5, so we will denote the CUO in pure argon as
CUO(Ar)n where n ) 4 or 5.
For the larger noble gas atoms Kr and Xe, our calculations
indicate that the optimum structures are CUO(Kr)5 and CUO(Xe)4,
respectively. The prediction of five-coordination of Kr is curious
and is not supported by the experimental data or by a simple
geometric model where the Ng atoms are separated by more
than twice their van der Waals radii when at typical U-Ng
bond lengths. Thus, five Ar atoms in a regular pentagonal
arrangement saturate the equatorial plane around CUO; when
the Ar atoms are separated by twice the van der Waals radius
(3.76 Å),44 the U-Ar distance is 3.20 Å, which is very near
the optimized U-Ar distances computed here. The same simple
model predicts that no more than four Kr or Xe atoms can
surround the CUO molecule, which is consistent with the DFT
binding-energy calculations for Xe but not for Kr. Because our
calculations with diffuse functions on the energy of CUO(Kr)5
have large error due to poor fitting of the Coulomb potentials,
it is possible that the DFT result predicting CUO(Kr)5 is
incorrect, and that a maximum of four Kr or Xe atoms occupy
The significance of second-shell solvation effects may be
indicated by small changes in the CUO frequencies of analogous
complexes when the host matrix is changed. For example,
CUO(Xe)4 in argon red-shifts 3.2 and 2.1 cm-1 in krypton and
5.6 and 3.4 cm-1 in pure xenon. In addition, the CUO(Kr)4
complex in argon red shifts 3.7 and 2.3 when pure krypton is
used. Figure 7 shows the small secondary matrix solvent shifts
for CUO(Xe)4 in the argon matrix as compared to krypton.
The experimental observations combined with relativistic
DFT calculations provide strong evidence that the bands at
848.0, 843.6, 839.4, and 835.4 cm-1 are due to CUO(Ar)3(Xe),
CUO(Ar)2(Xe)2, CUO(Ar)(Xe)3, and CUO(Xe)4, respectively.
Figure 11 illustrates the calculated structures of these com-
9
3136 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
Table 6. Optimized Geometries, Binding Energies, and Calculated Stretching Frequencies of CUO(Ar)4-n(Xe)n (n ) 0, 1, 2, 3, 4)
Complexesa
state
E
bind
U−C
U−O
U−Ar
U−Xe(Kr)
ν(U−C)
ν(U−O)
CUO(Ar)4
CUO(Kr)4
CUO(Ar)3(Xe)
3B2
3B2
3A”
15.4
18.3
17.7
1.868
1.873
1.872
1.843
1.839
1.842
3.185 × 4
-
-
881
876
877
824
822
824
3.258 × 4
3.214 × 2
3.360
3.228
b
CUO(Ar)2(Xe)2
CUO(Ar)(Xe)3
3B2
26.6
23.2
1.882
1.877
1.846
1.838
3.118 × 2
3.145
3.379 × 2
3.358 × 2
3.574
875
873
823
820
3A”
CUO(Xe)4
CUO(Ar)5
3B2
28.4
17.9
1.875
1.868
1.833
1.841
-
3.369 × 4
870
880
819
822
3A′′
3.22-3.26
-
a The binding energies are in kcal/mol, the bond lengths in Å, and the frequencies in cm-1. All quantities are calculated using the Vdiff basis sets with
2f diffuse functions deleted. b Only the trans isomer of CUO(Ar)2(Xe)2 converged with this basis set.
A closer examination of the electronic structure of CUO(Ng)4
reveals the nature of the U-Ng bonding in these new octahedral
complexes. We will discuss the bonding in CUO(Ar)4 to show
the major interactions that occur, and use those results to explain
the changes that occur upon substitution of Ar by a heavier Kr
or Xe atom. The U-Ar interactions in CUO(Ar)4 are similar to
those discussed earlier for CUO(Ar), although the higher
symmetry of the larger complex makes the discussion somewhat
clearer. Our calculations predict that ground-state CUO(Ar)4 is
a triplet C4V molecule with a linear CUO fragment and
essentially octahedral coordination about the U atom (Figure
11). The Ar atoms are slightly displaced toward the C atom
( C-U-Ar ) 84.6°), much as the Ar atom in CUO(Ar) is
closer to the C atom than to the O atom (Figure 9).
The weak interactions between the Ar atoms and the U atom
of CUO involve σ-donation from an Ar lone pair into empty
primarily U-based orbitals. Although in principle Ar could act
as a π donor toward the U atom as well, we see no evidence
for any significant Ar-U π donation. Under C4V symmetry, the
Ar σ-donor lone pairs lead to symmetry-adapted linear combi-
nations of a1, b2, and e symmetry (in our coordinate system,
the Ar atoms lie along the lines y ) (x). These Ar group
orbitals can, in principle, interact with the vacant U 7s orbital
(a1), U 7p orbitals (a1 + e), U 6d orbitals (a1 + b2 + e), or U
5f orbitals (a1 + b2 + e). In general, ligand donation to actinide
atoms preferentially involves the U 6d orbitals.47 Indeed, we
find that the most significant Ar-to-U donation occurs in a
bonding molecular orbital of b2 symmetry that involves donation
from the filled Ar 3pσ orbitals into the empty 6dxy orbital. This
MO is composed of roughly 96% Ar 3p and 4% U 6d character.
It is therefore best described as a weak Lewis acid-base
interaction between the Ar 3p (Lewis base) and U 6d (Lewis
acid) orbitals. Some other MOs of triplet CUO(Ar)4 involve
1-2% uranium-based orbitals, and represent additional Ar-to-U
donation. As is pointed out in uranyl bonding, the semi-core
6p orbitals can have bonding interaction with the Ar 2s and 2p
orbitals, which gives rise to the so-called “U 6p core hole”.48
Figure 11. Structures for CUO(Ar)4-n(Xe)n, n ) 0, 1, 2, 3, 4.
plexes, along with that of CUO(Ar)4 for comparison. We
calculate the total CUO-Xe binding energy in CUO(Xe)4 as a
substantial 28.4 kcal/mol, leading to an average U-Xe bond
energy of 7.1 kcal/mol. The observations for CUO with dilute
Kr in solid argon give parallel results, with the formation of
the Kr complexes CUO(Ar)4-n(Kr)n (n ) 1, 2, 3, 4). Likewise
CUO in the presence of Xe in krypton gives analogous
CUO(Kr)4-n(Xe)n complexes. As expected, the red-shifts of the
stretching frequencies induced upon substitution of Ar with Kr
and Kr with Xe are smaller than those upon substitution of Ar
with Xe, but the sum of the former two is almost equal to the
latter red-shifts (peak separations in four-band progressions).
As noted earlier, both the experiments and the calculations
indicate a small red-shift of the CUO modes upon substitution
of a lighter noble-gas atom with a heavier one. We also note
slight changes in the isotopic frequency ratios for the 848.0 and
801.3 cm-1 bands of CUO(Ar)3(Xe) in solid argon relative to
those of CUO(Ar)n (Table 2). For the lower-energy band, the
12CO/13CO and C16O/C18O ratios for CUO(Ar)3(Xe) respectively
increase and decrease slightly relative to CUO(Ar)n. The
behavior of the isotopic ratios for the higher-energy band is
opposite that of the lower band: the 12CO/13CO and C16O/C18O
ratios decrease and increase, respectively. Thus, for each of the
bands, the larger ratio decreases slightly and the smaller ratio
increases slightly upon substitution, implying that substitution
by Xe induces greater mode mixing between the “pure” U-C
and U-O stretching modes. A similar trend is found for the
CUO(Kr) species in solid argon. Both of these trends have been
confirmed by the calculated DFT normal modes.
Mulliken population analysis49 of the converged electronic
structures can provide an indication of the changes that occur
when four Ar atoms bond to CUO. However, the very diffuse
basis functions that are used in these calculations cause a well-
known breakdown of the Mulliken population model, which
assumes that basis functions maximize closest to the atom on
which they are centered. To examine meaningful Mulliken
(47) Pepper, M.; Bursten, B. E. Chem. ReV. 1991, 91, 719.
(48) Larsson, S.; Pyykko¨, P. Chem Phys. 1986, 101, 355.
(49) Mulliken, R. S. J. Chem. Phys. 1955, 23, 1833.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3137

A R T I C L E S
Andrews et al.
populations, we have performed additional calculations at the
converged geometry using a reduced basis set in which the
diffuse functions have been removed. Using this reduced basis,
The new CUO(Ar)4-n(Ng)n complexes reported here are the
only known examples of neutral metal complexes involving four
noble gas atoms. Of these new complexes, CUO(Xe)4 exhibits
the strongest U-Xe bonding, and it is thus interesting to
compare the properties of CUO(Xe)4 with those of other metal-
Xe compounds. The calculated U-Xe distance of 3.37 Å in
3
the Mulliken charge on the U atom in Φ CUO is +1.43.
Binding four Ar atoms to form 3B2 CUO(Ar)4 reduces the
Mulliken charge on U to +1.31. Consistent with the discussion
above, the largest change in the U Mulliken populations occurs
in the 6d orbitals. The total 6d orbital population in triplet CUO
is 1.73, whereas in triplet CUO(Ar)4 it is 1.83, which accounts
for 83% of the charge reduction. The Mulliken charge on the
four Ar atoms in triplet CUO(Ar)4 is +0.16, an indication that
each Ar atom is donating about 0.04 of an electron to the CUO
molecule.
CUO(Xe)4 is, of course, significantly longer than the experi-
2+ 51
mental Au-Xe distance of 2.74 Å in AuXe4
;
the Au-Xe
bonding in this dicationic complex (as well as that in AuXe+
and AuXe2+) is strongly enhanced by the charge-induced dipole
interactions involving the Au2+ cation.52,53 We must compare
CUO(Xe)4 with the experimentally observed neutral transition
metal complexes XeM(CO)5 (M ) Cr, Mo, W). The calculated
M-Xe distances in these carbonyl complexes, 2.99 Å (Cr), 3.10
Å (Mo), and 3.06 Å (W),15 are comparable to the U-Xe
distances predicted here, especially when the larger size of U
is considered. The measured M-Xe dissociation energies in the
XeM(CO)5 complexes, 9.0 (Cr), 8.0 (Mo), and 8.2 (W) kcal/
mol,13,14 are comparable with those computed here for U-Xe.
The present spectra clearly show that four Ng atoms are
involved in the intimate coordination sphere around uranium.
This work provides conclusive evidence for distinct
CUO(Ar)4-n(Ng)n (Ng ) Kr, Xe; n ) 1, 2, 3, 4) and
CUO(Kr)4-n(Xe)n complexes, and the first characterization of
neutral complexes involving four noble-gas atoms on one metal
center.
As noted above, triplet CUO(Ar)4 is about 4.5 kcal/mol below
the singlet form of the molecule. The reasons that the triplet
form of the molecule is favored are entirely analogous to our
discussion above about CUO(Ar). The calculated U-C bond
length in triplet CUO(Ar)4 (1.868 Å) is more than 0.1 Å longer
than that in the singlet form of the molecule (1.757 Å), thus
allowing a closer approach of the Ar atoms: We calculate that
the U-Ar bond length in triplet CUO(Ar)4 (3.185 Å) is more
than 0.1 Å shorter than that in the singlet molecule (3.289 Å).
Also, as in CUO(Ar), the expansion of the U 6d orbitals upon
reduction of U facilitates stronger Ar-to-U donation. The Lewis
acid-base description of U-Ng bonding leads to the prediction
that heavier, more polarizable noble-gas atoms should bind more
strongly to CUO than do lighter ones. That notion is consistent
with both our experimental and computational results. In fact,
the calculated U-Xe interaction is an order-of-magnitude larger
than the van der Waals interaction in Xe2 (0.56 kcal/mol).50
We have seen that Kr and Xe atoms readily replace Ar in the
coordination sphere about CUO, and Xe will replace Kr. The
calculated binding energies for the CUO(Ng)4 species are
consistent with these observations: The binding energies per
Ng atom are 3.9, 4.6, and 7.1 kcal/mol for CUO(Ar)4,
CUO(Kr)4, and CUO(Xe)4, respectively. Thus, it appears that
the substitution chemistry of noble gases on CUO is largely
driven by enthalpic factors related to the strengths of the U-Ng
bonds.
The observation that the U-C and U-O stretching frequen-
cies slightly red-shift on substitution of a lighter noble-gas atom
with a heavier one is largely tied to electronic structure as well,
rather than the mass effects of the heavier noble-gas atoms. The
heavier noble-gas atoms are slightly better electron donors to
CUO than are Ar atoms and thus make the uranium atom slightly
more electron rich. Indeed, in CUO(Xe)4 the four Xe atoms
donate 0.21e, whereas the U charge is 0.08e more comparing
to that in CUO(Ar)4. Although the U-C and U-O bonds have
large covalent components, they also depend on significant
donation from the formal C4- and O2- to the U atom. When
Kr or Xe atoms are bonded to CUO, the donation from C and
O is slightly suppressed, leading to slightly weaker U-C and
U-O bonds and the resultant red-shift. This inductive influence
of the noble-gas atoms on the U-C and U-O stretching
frequencies is largely driven by the number of each type of
noble-gas atom bonded to CUO, and is not expected to be
significantly affected by stereochemistry. We are therefore not
surprised that we do not find experimental evidence for both
the cis and trans isomers of CUO(Ar)2(Xe)2.
Conclusions
Laser-ablated U atoms react with CO in excess argon to
produce CUO, which is trapped as a triplet-state complex in
solid argon at 7 K, based on agreement between observed and
relativistic density functional theory calculated isotopic frequen-
cies (12C16O, 13C16O, 12C18O). This contrasts the neon matrix
environment, which trapped CUO in a linear singlet state
calculated to be about 1 kcal/mol lower in energy. The linear
triplet OUCCO and bent CUO- species are trapped in both
matrices with typical matrix shifts in their infrared spectra, but
CUO fundamentals are displaced 195 and 68 cm-1 from neon
to argon. In addition to large neon-to-argon shifts, the isotopic
frequency ratios (and normal modes) of CUO in neon and the
CUO(Ar)x complex are substantially different. Similar experi-
ments in krypton matrices give small shifts for CUO and
OUCCO in the same states trapped in solid argon. In contrast,
laser-ablated Th atoms react with CO to form CThO, which is
trapped in the same bent triplet state in solid neon, argon and
krypton.
Investigations with 2% Kr and 3% Xe in argon and 3% Xe
in krypton give small shifts for CUO(Ng)n in the same triplet
state trapped in solid argon. The Kr and Xe doped experiments
give additional weak absorptions that evolve on annealing for
three higher CUO(Kr)n and CUO(Xe)n complexes described as
CUOAr4-n(Kr)n, CUOAr4-n(Xe)n, and CUO(Kr)4-n(Xe)n com-
plexes. Relativistic density functional theory calculations show
that distinct CUO(Ng)n complexes are responsible for these
trends in frequencies and estimate the CUO-Ng dissociation
energies as 3.7, 4.9 and 7.3 kcal/mol for Ar, Kr, and Xe,
respectively.
(51) Seidel, S.; Seppelt, K. Science 2000, 290, 117.
(52) Pyykko¨, P. J. Am. Chem. Soc. 1995, 117, 2067.
(53) Schro¨der, D.; Schwarz, H.; Hrusak, J.; Pyykko¨, P. Inorg. Chem. 1998, 37,
624.
(50) Freeman, D. E.; Yoshino, K.; Tanaka, Y. J. Chem. Phys. 1974, 61, 4880.
9
3138 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Noble Gas−Actinide Complexes
A R T I C L E S
Acknowledgment. We acknowledge support for this research
from the NSF (CHE 00-78836 to L.A.) and the Division of
Chemical Sciences, Geosciences, and Biosciences of the U.S.
Department of Energy’s Office of Basic Energy Sciences (DE-
FG02-01ER15135 to B.E.B.). We thank A. W. Ehlers for
providing the diffuse basis sets of Xe. We are grateful to the
Ohio Supercomputer Center for generous grants of computer
time. This research was performed in part using the Molecular
Science Computing Facility (MSCF) in the William R. Wiley
Environmental Molecular Sciences Laboratory, a national
scientific user facility sponsored by the U.S. Department of
Energy’s Office of Biological and Environmental Research and
located at the Pacific Northwest National Laboratory. Pacific
Northwest is operated for the Department of Energy by Battelle.
JA027819S
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 3139