A Novel Anionic Gold-Indium Cluster Compound: Synthesis and Molecular and Electronic Structure

Article abstract of DOI:10.1021/ic970725c

The insertion of InBr into the Au - Br bond of [(Ph₃P)AuBr] in tetrahydrofuran (thf) in the presence of [(CH₂-PPh₂)₂] (dppe) leads to the formation of an orange complex [(dppe)₂Au]⁺[(dppe)₂Au₃In ₃Br₇(thf)]^-, 2. Analytical, spectroscopic, and X-ray structural investigations showed that this product is an anionic analogue of a neutral chloride complex [(dppe)₂Au₃In₃Cl₆(thf)₃], 1, prepared recently. Both complexes have an Au₃In₃ cluster core of approximate C_2v symmetry with one extremely short Au-Au bond [Au1-Au3 2.575(1) ?] as part of a quasilinear array P1- Au1- Au3-P4, suggesting the presence of a bis(phosphine) complex of the neutral Au₂ molecule as part of the cluster. The third gold atom (Au2) is then assigned oxidation state +1. To gain deeper insight into the structure and bonding of this novel class of gold cluster compounds, regarding mainly the peculiar cluster geometry, the charge distribution, and the oxidation states, a series of scalar relativistic all-electron density functional (DF) calculations on model systems has been performed. As a model for 1, the neutral cluster {Au₃(PH₃)₄[InCl₂(H ₂O)]₃} was studied. For the examination of the geometry of complexes 1 and 2, the cluster Au₃(PH₃),₄I₃ has been considered as a further simplified model, where iodine replaces the InX₂(thf) units. Experimental and calculated cluster geometries agree satisfactorily, and the formal oxidation states of the gold atoms (0 for Au1 and Au3, +1 for Au2) could be confirmed, but for the In centers no interpretable differences of the Mulliken charges were found.

Full text of DOI:10.1021/ic970725c

Inorg. Chem. 1997, 36, 5699-5705
5699
A Novel Anionic Gold-Indium Cluster Compound: Synthesis and Molecular and
Electronic Structure
Franc¸ois P. Gabba1ı,† Sai-Cheong Chung,‡ Annette Schier,† Sven Kru1ger,‡
Notker Ro1sch,*,‡ and Hubert Schmidbaur*,†
Anorganisch-chemisches Institut and Lehrstuhl fu¨r Theoretische Chemie der Technischen Universita¨t
Mu¨nchen, Lichtenbergstrasse 4, D-85747 Garching, Germany
ReceiVed June 12, 1997X
The insertion of InBr into the Au-Br bond of [(Ph3P)AuBr] in tetrahydrofuran (thf) in the presence of [(CH2-
PPh2)2] (dppe) leads to the formation of an orange complex [(dppe)2Au]+[(dppe)2Au3In3Br7(thf)]-, 2. Analytical,
spectroscopic, and X-ray structural investigations showed that this product is an anionic analogue of a neutral
chloride complex [(dppe)2Au3In3Cl6(thf)3], 1, prepared recently. Both complexes have an Au3In3 cluster core of
approximate C2V symmetry with one extremely short Au-Au bond [Au1-Au3 2.575(1) Å] as part of a quasi-
linear array P1-Au1-Au3-P4, suggesting the presence of a bis(phosphine) complex of the neutral Au2 molecule
as part of the cluster. The third gold atom (Au2) is then assigned oxidation state +1. To gain deeper insight into
the structure and bonding of this novel class of gold cluster compounds, regarding mainly the peculiar cluster
geometry, the charge distribution, and the oxidation states, a series of scalar relativistic all-electron density functional
(DF) calculations on model systems has been performed. As a model for 1, the neutral cluster {Au3(PH3)4[InCl2-
(H2O)]3} was studied. For the examination of the geometry of complexes 1 and 2, the cluster Au3(PH3)4I3 has
been considered as a further simplified model, where iodine replaces the InX2(thf) units. Experimental and
calculated cluster geometries agree satisfactorily, and the formal oxidation states of the gold atoms (0 for Au1
and Au3, +1 for Au2) could be confirmed, but for the In centers no interpretable differences of the Mulliken
charges were found.
1. Introduction
systems with few heteroatoms which are not too different in
their element characteristics from the parent gold component.7
The cluster chemistry of gold has been an extremely active
area of research for more than two decades. The emerging rules
of stoichiometry, structure, and bonding of homoatomic gold
clusters in particular are providing a consistent and transparent
taxonomy of a large variety of cluster species.1 The field of
heteroatomic gold clusters is far less developed,2 and only a
small number of metals have been considered as components
of mixed-metal aggregates.3 While at least a few of the late
transition metals, namely, those neighboring gold in the periodic
table, have been included in the more recent investigations in
several laboratories,3,4 there is a complete lack of information
on discrete gold clusters containing the more electropositive
transition or main group metals.5
A deeper knowledge of such clusters is desirable, however,
for a number of reasons: (1) Mixed-metal clusters of well-
defined stoichiometry and structure are important model systems
for bulk alloys and their surfaces. (2) Clusters composed of
metals with strongly different atomic radii and electronegativities
are expected to show tunable and thus more selective reactivity
in stoichiometric and catalytic reactions. (3) The theory of
chemical bonding in hetero clusters and nanoparticles is only
emerging,6-9and it has to date been developed mainly for binary
(4) Depending on the electronegativity of the admixed metal
components, gold can be assigned not only low, nonintegral
positive oxidation states but even nonintegral oxidation states
approaching negative unity as in aurides of the type Cs+Au-.
“Reversing the polarity” of gold is thus not an unreasonable
concept for cluster chemistry, as it has already been demon-
strated for bulk Zintl phases containing gold.5
Following up a program on gold clusters with interstitial non-
metals (B, C, N, O, Cl and their homologues),10 a systematic
study has therefore recently been initiated to probe gold cluster
formation with electropositive main group metals. In the present
report some results on the binary Au/In system are presented.
A preliminary account has been published,11 the contents of
which regarding preparation and crystal structure is not dupli-
cated here. The results are included in the discussion, however,
since, as an important extension of the previous11 and present
experimental work, we recently also turned to theoretical studies
of these important prototypes. Density functional calculations
including relativistic effects have been employed for studying
a number of pertinent problems and have proven extremely
useful for describing structure and bonding in gold hetero
clusters.12,13
† Anorganisch-chemisches Institut.
(6) Schmid, G. Clusters and Colloids; VCH: Weinheim, 1994.
(7) (a) Mingos, D. M. P.; Johnson, R. J. Struct. Bonding 1987, 68, 29.
(b) Mingos, D. M. P. J. Cluster Sci. 1992, 3, 397.
(8) Mingos, D. M. P.; Wales, D. J. Introduction to Cluster Chemistry;
Prentice-Hall: Englewood Cliffs, NJ, 1990.
(9) Heiz, U.; Vayloyan, A.; Schumacher, E.; Yeretzian, C.; Stener, M.;
Gisdakis, P.; Ro¨sch, N. J. Chem. Phys. 1996, 105, 5574.
(10) (a) Schmidbaur, H. Chem. Soc. ReV. 1995, 24, 391. (b) Blumenthal,
A.; Beruda, H.; Schmidbaur, H. J. Chem. Soc., Chem. Commun. 1993,
1005.
‡ Lehrstuhl fu¨r Theoretische Chemie.
X Abstract published in AdVance ACS Abstracts, November 1, 1997.
(1) Hall, K. P.; Mingos, D. M. P. Prog. Inorg. Chem. 1984, 32, 237.
(2) Mingos, D. M. P.; Watson, M. J. AdV. Inorg. Chem. 1992, 39, 327.
(3) (a) Teo, B. K.; Zhang, H. Coord. Chem. ReV. 1995, 143, 611. (b)
Pignolet, L. H.; Aubart, M. A.; Craighead, K. L.; Gould, R. A. T.;
Krogstad, D. A.; Wiley, J. S. Coord. Chem. ReV. 1995, 143, 219.
(4) (a) Steggerda, J. J. Comments Inorg. Chem. 1992, 11, 113. (b) Stra¨hle,
J. J. Organomet. Chem. 1995, 488, 15.
(5) (a) Schmidbaur, H. Gold Bull. 1990, 23, 11. (b) Mingos, D. M. P.;
Powell, H. R.; Stolberg, T. L. Transition Met. Chem. 1992, 17, 334.
(11) Gabbai, F. P.; Schier, A.; Schmidbaur, H. Inorg. Chem. 1995, 34,
3844.
S0020-1669(97)00725-8 CCC: $14.00 © 1997 American Chemical Society

5700 Inorganic Chemistry, Vol. 36, No. 25, 1997
Gabba¨ı et al.
of Gaussian-type orbitals density functional (LCGTO-DF) method.15,16
The local spin density approximation (LSDA) for the exchange-
correlation functional17 was applied during the self-consistency cycles.
This model functional often overestimates ligand binding energies, yet
it is known to yield accurate geometry and vibrational data.18 We
refrained from using a more accurate, yet computationally more
demanding gradient-corrected functional since in the discussion of
energetic aspects we are mainly interested in a comparison of trends.
The Gaussian-type molecular orbital basis sets for the atoms Au, P,
O, and H were taken as in previous investigations on other gold
compounds.19,20 Listing the number of exponents of the uncontracted
bases in parentheses and the size of the contracted ones in brackets,
the basis sets used are as follows: Au(21s,17p,11d,7f) f [11s,10p,-
7d,3f], P(12s,9p,1d) f [5s,4p,1d], O(9s,5p,1d) f [4s,3p,1d], and
H(6s,1p) f [4s,1p]. Basis sets of comparable quality were chosen for
In, I, and Cl: In(18s,14p,8d) f [7s,5p,4d],21,22 I(17s,14p,8d) f
[7s,5p,4d],22 and Cl(12s,9p,1d) f [5s,4p,1d].23,24 These atomic basis
sets were contracted in a generalized fashion using atomic LDA
eigenvectors. The charge density and the exchange-correlation potential
were represented by auxiliary Gaussian-type basis sets.15 The corre-
sponding s- and d(r2)-type functions were generated in a standard
fashion.15 For the heavier atoms, Au, In, I, and Cl, only every second
d(r2)-type function was used. In addition, a set of three p-type and
three d-type polarization exponents was employed for each atomic
center, except for O and H, where only p-type polarization functions
were taken into account. The p-exponents were 0.1, 0.4375, and 1.562;
the d-exponents are scaled by a factor of 2. The auxiliary basis sets
were left uncontracted.
The theoretical treatment of large heavy-metal clusters is a
challenging task, but it has been demonstrated that calculations
on suitable models of reduced complexity make the problems
tractable, yet also lead to meaningful fundamental conclu-
sions.13,14 This strategy has also been successfully used in the
present study by introducing simple pseudoligands with frontier
orbital characteristics and electron counts similar to those of
the ligands in the synthesized systems.
2. Experimental Section
2.1. General Procedures. All experiments were carried out under
an atmosphere of dry, purified nitrogen. Glassware was dried and filled
with nitrogen, and solvents were distilled and kept under nitrogen.
NMR: Jeol GX 400; TMS and 85% aqueous H3PO4 as external
standards for 1H and 31P NMR, respectively. MS: Finnigan MAT 90.
Microanalyses: In-house analyzers (by combustion). InBr was prepared
by melting together 2 equiv of indium powder with 1 equiv of InBr3
under vacuum at 375 °C and purified through sublimation under the
same conditions. Ph3PAuBr was prepared by the reaction of NaBr with
Ph3PAuCl in the two-phase system CHCl3/H2O and purified by
recrystallization from a CHCl3/hexane solution.
2.2. Preparation of [(dppe)2Au]+[(dppe)2Au3In3Br7(thf)]- (2).
InBr (0.10 g, 0.51 mmol), Ph3PAuBr (0.27 g, 0.5 mmol), and dppe
(0.20 g, 0.5 mmol) were mixed as solids and cooled to -78 °C. Thf
(5 mL) was then added to the mixture. With stirring, the reaction
mixture was allowed to reach room temperature, during which time a
yellowish precipitate had formed. After 3 h at 20 °C the precipitate
was isolated through filtration. Addition of CH2Cl2 (5 mL) to the
precipitate followed by filtration yielded an orange solution. A layer
of hexane (5 mL) was slowly allowed to diffuse into this CH2Cl2
solution at -25 °C, which resulted in the precipitation of 0.10 g (24%
yield based on gold) of an orange crystalline compound 2 (mp 105° C
dec). Elemental anal. for C108H104Au4Br7In3OP8. Calcd: C, 38.6; H,
3.1. Found: C, 38.0; H, 3.4. 31P NMR {1H} (109.3 MHz) δ (ppm):
18.0 (4P, [(dppe)2Au]+), 31.3 (2P), 56.3 (2P, [(dppe)2Au3In3Br7(thf)]-).
1H NMR (399.8 MHz) δ (ppm): 1.76, 3.54 (br, 8H, thf); 2.43 (br, 8H,
CH2-[(dppe)2Au]+), 3.10 (4H), 3.42 (4H, br, CH2-[(dppe)2Au3In3Br7-
(thf)]-), 6.96-7.67 and 8.05 (m, 80H, Ph-CH). MS (FAB): m/z 993
([(dppe)2Au]+, 100).
3. Preparation and Properties of a Novel Anionic
Gold/Indium Cluster Complex
Preliminary work has shown that insertion of indium(I)
halides [InX] into the Au-X bonds of gold(I) halide complexes
[LAuX] can give ready access to mixed-metal mixed-valent Au/
In cluster species.11 Thus treatment of [InCl] with [Ph3PAuCl]
in the presence of [Ph2PCH2CH2PPh2] (dppe) in tetrahydrofuran
(thf) affords the cluster [(dppe)2Au3In3Cl6(thf)3], 1 (eq 1). In
3[(Ph3P)AuCl] + 3[InCl] + 2[dppe] + 3[thf] f
2.3. X-ray Crystallography. A specimen of suitable quality and
size (0.1 × 0.25 × 0.40 mm) was mounted in a glass capillary and
used for measurements of precise cell constants and intensity data
collection. Diffraction measurements were made on an Enraf-Nonius
CAD-4 diffractometer using graphite-monochromated Mo KR radiation
(λ ) 0.710 73 Å) with the ω scan mode at -68 °C. An Lp correction
was applied, and intensity data were corrected for decay (-14.1%) and
absorption effects (ψ-scans, Tmin/Tmax ) 0.449/0.999). The structure
was solved by direct methods (SHELXTL-PLUS) and completed by
full-matrix least-squares techniques (SHELXL-93).
[(dppe) Au In Cl (thf) ]
3[Ph3P] +
(1)
2
3
3
6
3
1
attempts to prepare the analogous bromide compound, we now
reacted [InBr] with [Ph3PAuBr] and [dppe] in approximately
equimolar quantities in thf at -78 °C. A yellow precipitate
formed, which could be crystallized from dichloromethane/
hexane at -25 °C (24% yield of orange crystals, mp 105 °C
dec). Solutions in CD2Cl2 show three resonances of the relative
intensities 4:2:2 in the 31P{1H} NMR spectrum (δ ) 18.0, 31.3,
and 56.3 ppm). In the 1H NMR spectra there are the multiplet
signals for metal-coordinated [thf] at δ ) 1.76 and 3.54 ppm
and broad resonances for three types of CH2 groups at δ )
2.43 (8H), 3.10 (4H), and 3.42 ppm (4H), complemented by an
extended multiplet of aryl resonances δ ) 6.96-8.05 ppm (m,
80H). In FAB mass spectra the prominent peak for m/z ) 993
C108H104Au4Br7In3OP8: M ) 3357.37, monoclinic, a ) 12.681(1)
Å, b ) 20.139(1) Å, c ) 44.182(4) Å, â ) 91.06(1)°, space group
P21/c (No. 14), Z ) 4, Dc ) 1.977 g cm-3, F(000) ) 6352 e, µ(Mo
KR) ) 84.2 cm-1; 16 424 intensity data were measured up to (sin
θ/λ)max ) 0.59 Å-1, of which 16 349 independent structure factors were
used for refinement. All non-H atoms were refined with anisotropic
displacement parameters. All H atoms were placed in idealized
calculated positions and allowed to ride on their carbon atoms with
fixed isotropic contributions [U(iso)fix ) 1.5Ueq(C) for all methylene and
2
0.08 for all phenyl H atoms]. The function minimized was {[Σw(Fo
(15) Dunlap, B. I.; Ro¨sch, N. AdV. Quantum Chem. 1990, 21, 317.
(16) Ro¨sch, N.; Kru¨ger, S.; Mayer, M.; Nasluzov, V. A. In Recent
DeVelopments and Applications of Modern Density Functional Theory;
Seminario, J. M., Ed.; Elsevier: Amsterdam, 1996; p 497.
(17) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
(18) Ziegler, T. Chem. ReV. 1991, 91, 651.
(19) Ha¨berlen, O. D.; Ro¨sch, N. J. Phys. Chem. 1993, 97, 4970.
(20) Chung, S.-C.; Kru¨ger, S.; Schmidbaur, H.; Ro¨sch, N. Inorg. Chem.
1996, 35, 5387.
2
2
2
- Fc2)2]/Σ[w(Fo )2]}1/2, with w ) 1/σ2(Fo )+(ap)2 + (bp), p ) (Fo
+
2Fc2)/3, and a ) 0.0855, b ) 241.16. The final Rw and R1 [based on
Σ(||Fo| - |Fc||)/Σ|Fo|] were 0.1621 and 0.0665, respectively, for 988
refined parameters.
2.4. Computational Details. Calculations were performed in an
all-electron fashion by means of the scalar relativistic linear combination
(12) Go¨rling, A.; Ro¨sch, N.; Ellis, D. E.; Schmidbaur, H. Inorg. Chem.
1991, 30, 3986.
(13) Ha¨berlen, O. D.; Schmidbaur, H.; Ro¨sch, N. J. Am. Chem. Soc. 1994,
116, 8241.
(14) Ro¨sch, N.; Go¨rling, A.; Ellis, D. E.; Schmidbaur, H. Angew. Chem.,
Int. Ed. Engl. 1989, 28, 1357.
(21) Huzinaga, S. J. Chem. Phys. 1979, 71, 1980.
(22) Poirier, R.; Kari, R.; Cszimadia, I. G. Handbook of Gaussian Basis
Sets; Elsevier: New York, 1985.
(23) Veillard, A. Theor. Chim. Acta 1968, 12, 405.
(24) Huzinaga, S. Gaussian Basis Sets for Molecular Calculations;
Elsevier: New York, 1984.

A Novel Anionic Gold-Indium Cluster Compound
Inorganic Chemistry, Vol. 36, No. 25, 1997 5701
Figure 1. Molecular structure of [(dppe)2Au]+ with atomic numbering
(hydrogen atoms omitted for clarity). Selected bond lengths (Å) and
angles (deg): Au4-P5 2.391(5), Au4-P6 2.410(6), Au4-P7
2.399(5), Au4-P8 2.403(5); P5-Au4-P6 85.4(2), P5-Au4-P7
129.5(2), P5-Au4-P8 122.6(2), P6-Au4-P7 118.7(2), P6-Au4-
P8 118.7(2), P7-Au4-P8 85.9(2).
Figure 2. Molecular structure of [(dppe)2Au3In3Br7(thf)]- with atomic
numbering (hydrogen atoms omitted for clarity). Selected bond lengths
(Å) and angles (deg): Au1-Au2 2.860(1), Au1-Au3 2.575(1), Au2-
Au3 2.858(1), Au1-In1 2.828(2), Au1-In2 2.950(2), Au1-In3
2.761(2), Au1-P1 2.300(5), Au2-In1 2.973(1), Au2-In2 2.889(2),
Au2-P2 2.377(5), Au2-P3 2.381(5), Au3-In1 2.932(2), Au3-In2
2.922(2), Au3-In3 2.826(2), Au3-P4 2.288(5), In1-Br1 2.594(3),
In1-Br2 2.579(3), In1-O1 2.297(13), In2-Br3 2.579(3), In2-Br4
2.582(3), In2-Br7 2.763(3), In3-Br5 2.613(3), In3-Br6 2.606(3),
In3-Br7 2.785(3); Au2-Au1-Au3 63.18(3), Au1-Au2-Au3
53.52(2), Au1-Au3-Au2 63.29(3), P1-Au1-Au3 173.47(14),
P4-Au3-Au1 178.5(3), In2-Br7-In3 85.68(8).
suggests the presence of [(dppe)2Au]+ as the cationic compo-
1
nent, which agrees well with the 31P and H resonances in the
NMR spectra at δ ) 18.0 and 2.43 ppm, respectively. The
limited elemental analysis data are in agreement with the
complex formula [C108H104Au4Br7In3OP8] finally set up after
the solution of the X-ray crystal structure (below). This formula
can be reduced to the ionic components [(dppe)2Au]+[(dppe)2-
Au3In3Br7(thf)]-, which accounts for the analytical and spec-
troscopic data. Equation 2 is tentatively proposed for the
formation of the unexpected product. The stoichiometry of the
4[(Ph3P)AuBr] + 3[InBr] + 4[dppe] + [thf] f
4[Ph3P] + [(dppe)2Au]+[(dppe)2Au3In3Br7(thf)]-
(2)
2
anion of compound 2 is related to the composition of the neutral
chloride analogue (1, above) in that it features an additional
bromide anion replacing two thf donor molecules. This
introduction of an anion at the cluster then requires the presence
of a cation in the form of [(dppe)2Au]+.
4. Crystal and Molecular Structure of the Cluster
Compound
Figure 3. View of the structure of [(dppe)2Au3In3Br7(thf)]- showing
the bridging of the shortest edge of the gold triangle Au1-Au3 by In3
as well as the bridging two-coordinate bromine atom Br7 between In2
and In3.
Compound 2 crystallizes (from dichloromethane/hexane at
-25 °C) as orange monoclinic needles (space group P21/c, Z
) 4). The crystal lattice is composed of independent
[(dppe)2Au]+ cations and [(dppe)2Au3In3Br7(thf)]- cluster anions
with no unusually close interionic contacts. There are no solvent
molecules present in the interstices between the two sorts of
ionic components. Neither of the ions has a crystallographically
imposed element of symmetry, but the core of the cation
approaches quite closely the symmetry of point group S4 (Figure
1), while the core of the anion obeys mirror symmetry very
closely [the mirror plane being defined by Au2, In1, In2, Br7,
O1, and In3 (Figure 2)].
center) is rather flexible and adjusts very easily to a changing
crystal environment.
The [Au3In3] metal core of the anion closely resembles that
of the neutral molecule 1. Isosceles triangles of three gold and
three indium atoms are at right angles to each other, leading to
an arrangement where two of the indium atoms (In1/2) are
capping the gold triangle on both faces, while the third indium
atom (In3) is bridging the shortest edge of the gold triangle
(Au1-Au3, Figure 3). The two long edges, Au1-Au2 and
Au2-Au3, are spanned by the dppe ligands.
All three indium atoms bear two terminal bromine atoms each
(Br1/2, Br3/4, and Br5/6, respectively), but In1 is further
coordinated to one thf molecule, while In2 and In3 are bridged
by a two-coordinate bromine atom (Br7). This ligand distribu-
tion makes the two sides of the basic gold triangle different,
thus contrasting with the situation in compound 1, where the
molecular cluster core obeys C2V point group symmetry (Figure
4).11 Consistent with the situation in molecule 1, however, all
There is precedent for the structure of the [(dppe)2Au]+
cation, which has been determined also as a component of other
salts at least three times.14,25,26 A brief comparison of essential
differences of structural details shows that the conformation of
the two five-membered rings (with the gold atom as a spiro
(25) Harker, C. S. W.; Tekink, R. T.; Whitehouse, M. W. Inorg. Chim.
Acta 1991, 181, 23.
(26) Berners-Price, S. J.; Mazid, M. A.; Sadler, P. J. J. Chem. Soc., Dalton
Trans. 1984, 969.

5702 Inorganic Chemistry, Vol. 36, No. 25, 1997
Gabba¨ı et al.
(InCl2(thf))3 (Figure 4) was examined as an example. The
ligands had to be simplified to some extent to render the
computations tractable. For this purpose, the thf groups were
modeled by water ligands, which exhibit qualitatively the same
frontier orbitals at the oxygen center bound to In. The dppe
ligands used in the experiment were approximated by simple
phosphines, PH3. As shown in a previous study, PH3 is a
satisfactory model for larger phosphine ligands like PMe3 or
PPh3, in particular with respect to structural aspects.19 On the
other hand, one should keep in mind that PH3 is a weaker
electron donor; thus, binding energies and charge separations
will likely be underestimated.19 (On the other hand, the LDA
approximation used often tends to overestimate binding ener-
gies.) The symmetry of the resulting model cluster Au3(PH3)4-
(InCl2H2O)3 has been idealized to Cs, the mirror plane being
defined by the three In atoms. We derived averaged bond
lengths and angles from the experimental data, treating the two
µ3-In groups (InCl2H2O) as equivalent. The geometrical
parameters employed for the central Au-In cluster core are as
follows: d(Au1-Au3) ) 2.562 Å, d(Au1/3-Au2) ) 2.935 Å,
d(Au1/3-In1) ) 2.846 Å, d(Au2-In1) ) 2.976 Å, and d(Au1/
3-In3) ) 2.809 Å. For the InCl2H2O groups we used d(In1-
Cl) ) 2.442 Å, d(In3-Cl) ) 2.453 Å, d(In1-O) ) 2.311 Å,
d(In3-O) ) 2.336 Å, (Cl-In1-Cl) ) 97.95°, (Cl-In3-
Cl) ) 96.70°, (Cl-In1-O) ) 89.48°, (Cl-In3-O) )
86.75°, (Au1/3-In1-O) ) 99.03°, and (Au1/3-In3-O) )
116.50°.
The parameters related to the phosphine ligands are d(Au1-
P1) ) 2.306 Å, d(Au2-P2) ) 2.394 Å, (In1-Au2-P2,3) )
107.45°, and (In1-Au1/3-P1,4) ) 114.85°. For the phos-
phine and water ligands, the following internal geometry
parameters were used: d(P-H) ) 1.415 Å, (H-P-H) )
93.3°, d(O-H) ) 0.96 Å, and (H-O-H) ) 104.5°.29
5.2. Charge Distribution. The charge distribution of the
model cluster compound Au3(PH3)4(InCl2H2O)3 has been ex-
amined in a Mulliken analysis, and the results have been
evaluated by comparison of Mulliken charges of various subunits
of this system. Mulliken charges are known to have some
serious deficiencies, in particular when interpreted in an absolute
fashion. However, in a comparative study like the one carried
out here, they have often proven quite useful. The following
discussion will confirm this rather general experience.
Figure 4. Molecular structure of (dppe)2Au3(InCl2thf)3 (hydrogen
atoms omitted for clarity).11
three indium vertices of the cluster have three external electron
pair donor ligands providing a total of six ligand electrons.
The most intriguing structural details of clusters 1 and 2 are
the short distances between atoms Au1 and Au3 of only 2.562-
(1)/2.575(1) Å, which suggest particularly strong metal-metal
interactions. These gold atoms bear only one phosphine donor
ligand each, and there is conspicuous linearity of the axis P1-
Au1-Au3-P4, reminiscent of the structure proposed for as yet
unconfirmed phosphine complexes [L-Au-Au-L] of the
dinuclear gold molecule Au2.5 It is therefore tempting to assign
an oxidation state of 0 to the gold atoms Au1 and Au3, but +1
to Au2. The indium atoms In1 and In2 are then to be classified
as +2, and In3 as +1.11
The gold atom Au2 bears two phosphine donor units, and its
distances from Au1 and Au3 are much longer and in a standard
range for many other gold clusters.
In summary, therefore, the structures of the Au3In3 clusters
in the chloride and bromide complexes 1 and 2 are surprisingly
similar, suggesting a pronounced local minimum of energy for
this type of stoichiometry and organization in gold/indium
cluster chemistry. It should be pointed out that there is no
straightforward way of assigning an electron count or oxidation
state to the four types of metal atoms (two types of gold and
two types of indium atoms) using standard qualitative rules of
cluster chemistry. A theoretical study of the electronic structure
of this prototype seemed therefore highly desirable.
In Figure 5 the effect of the In groups L ) (InCl2H2O) on
the central gold triangle is shown. The L groups, which exhibit
similar geometries independent of coordination, act as electron-
withdrawing moieties since in the isolated group In is in the
formal oxidation state +2 (Mulliken charges: In +0.61 au, Cl
-0.37 au, H2O +0.13 au). As gold is a rather electronegative
metal, the charge transfer to the L groups is limited. Addition
of µ2-L or µ3-L groups induces a charge of about 0.1 au on the
gold triangle, which is not enhanced by the concerted effect of
both types of ligand groups. The charge distribution in the Au3
moiety changes due to the addition of L moieties (Figure 5):
while Au1/3 are positively and Au2 is negatively charged in
Au3, the charges change signs in Au3L3. Although differently
coordinated, the µ2- and µ3-L groups exhibit a very similar
internal charge distribution. All In atoms carry a charge of about
0.5 au (Figure 5).
5. Theoretical Investigations
To gain deeper insight into the structure and bonding of this
novel class of gold cluster compounds, a series of density
functional (DF) calculations on model systems has been carried
out. An accurate scalar relativistic DF approach16,27 has been
employed to account for the fact that a theoretical description
of gold chemistry has to properly consider both relativistic and
correlation effects.16,28 The main objectives of our theoretical
studies were a clarification of the charge distribution in the
cluster compounds and support for the preliminary assignment
of oxidation states to the various atoms in the Au3In3 metal
core; furthermore, we aimed at a rationalization of the peculiar
geometry of the central Au3 unit of these Au/In clusters. Finally,
we have also examined how strongly the phosphine and the
(InCl2(thf)) moieties are bound to the cluster core.
The addition of electron-donating phosphine ligands has, not
unexpectedly, a considerable effect on the charge distribution
in the cluster (Figure 6). Phosphine ligation reverses the sign
of charge separation in the bare cluster Au3 and enhances the
donating ability of the Au3 moiety. This leads to larger negative
5.1. Model System. To examine the charge distribution of
the Au/In clusters, which all exhibit a similar geometry in their
central Au3In3 unit, the more symmetric compound (dppe)2Au3-
(27) Ha¨berlen, O. D.; Ro¨sch, N. Chem. Phys. Lett. 1992, 199, 491.
(28) Pyykko¨, P. Chem. ReV. 1988, 88, 563.
(29) Weast, R. C. CRC Handbook of Chemistry and Physics, 61st ed.; CRC
Press, Boca Raton, 1980.

A Novel Anionic Gold-Indium Cluster Compound
Inorganic Chemistry, Vol. 36, No. 25, 1997 5703
Table 1. Binding Energies De (in kcal/mol) of Phosphine Ligandsa
in Au3(PH3)4Ln, n ) 0-3, L ) (InCl2H2O)
system
P1/4H3
P2/3H3
avb
Au3(PH3)4
41.5
48.0
51.4
56.5
30.7
21.4
29.5
32.3
28.6
32.7
38.7
42.9
Au3(PH3)4L
Au3(PH3)4L2
Au3(PH3)4L3
a For the definition of P1/4H3 and P2/3H3 see Figure 4. b Average
binding energy determined from the reaction Au3(PH3)4Ln f Au3Ln +
4PH3.
Table 2. Average Binding Energiesa De (in kcal/mol) of Phosphine
Ligands in Au3(PH3)4Lnm+, n ) 0-2, m ) 0-2, L ) (InCl2H2O)
system
m ) 0
m ) 1
m ) 2
Au3(PH3)4
Au3(PH3)4L
Au3(PH3)4L2
28.6
32.7
38.7
47.0
45.4
51.7
69.4
62.3
60.9
a Average binding energy determined from the reaction Au3(PH3)4Lnm+
m+
f Au3Ln + 4PH3.
Figure 5. Mulliken population analysis of the clusters Au3Ln, n )
0-3, L ) InCl2H2O, at the experimental geometry. The Au3 subunit is
depicted as a solid triangle. Detailed charge distributions of the L groups
of the cluster Au3L3 are given in boxes. The reaction arrows intercon-
necting the different species are labeled with the binding energies of
the newly attached L groups in kcal/mol.
Table 3. Optimized Au-Au distances (in Å) in the Unit Au3, with
and without Phosphine Ligands and for Singly Ionized Speciesa
system
Au3
Au1-Au3
Au1/3-Au2
2.52
2.53
2.56
2.60
2.59
2.64
2.79
2.56
2.60
2.72
Au3(PH3)4
+
Au3
+
+
Au3(PH3)3
Au3(PH3)4
a Atoms numbered according to Figure 4.
au). The Mulliken charges of +0.19 au for Au2 and -0.03 au
for Au1/3 in Au4(PH3)4L3 provide further support for the earlier
assignment of oxidation states +1 and 0,11 respectively, which
was based on geometry arguments as well as on Mo¨ssbauer
spectroscopy investigations. Surprisingly, no significant dif-
ferences were found between the Mulliken charges of the two
types of In ligand moieties. Thus, on the basis of the present
results, we prefer to assign an intermediate formal oxidation
state between +1 and +2 to both of these In units. We are
unable to confirm the tentative interpretation11 of µ2-In as In(+1)
and µ3-In as In(+2) suggested by the different coordination of
these two groups.
5.3. Binding Energies. To examine the binding of the
different ligand moieties of the Au/In model cluster and their
mutual influence, a set of binding energies has been calculated.
As these data have been obtained without taking into account
geometry relaxation of the various moieties involved, they
should only be taken as qualitative measures for comparative
purposes. In Figures 5 and 6 the binding energies of the In
groups L ) (InCl2H2O) to the cluster core are given. For both
cluster models, with and without phosphine ligands, the µ2-L
group is bound more strongly than the µ3-L groups. Surpris-
ingly, the bonding of the µ2- and µ3-L groups is mutually
independent in the sense that the binding energy of µ2-L is
virtually the same if it is attached to Au3 or to Au3(µ3-L)2 (Figure
5) and Vice Versa. The same observation can be made for the
phosphine-ligated models, see Figure 6. In line with this
observation, inspection of the one-electron orbitals of Au3-
(PH3)4L3 shows no mixing between contributions derived from
µ2-In and µ3-In. Due to phosphine ligation the binding energy
of the L groups to the metal cluster core increases, by 16.6 kcal/
mol for µ2-L and by 20.3 kcal/mol for µ3-L.
Figure 6. Mulliken population analysis of the phosphine-ligated
clusters Au3(PH3)4Ln, n ) 0-3, L ) InCl2H2O, at the experimental
geometry. The Au3 subunit is depicted as a solid triangle. The phosphine
ligands are abbreviated by P. Detailed charge distributions of the L
groups of the cluster Au3L3 are given in boxes. The reaction arrows
interconnecting the different species are labeled with the binding
energies of the newly attached L groups in kcal/mol.
charges of the L groups. Both types, µ2-L and µ3-L, exhibit a
charge of about 0.3 au if they are added separately. In the
complete cluster their charge is slightly smaller, -0.29 au for
µ2-L and -0.23 au for µ3-L (Figure 6). The increased oxidation
of the gold triangle as a consequence of L group addition is
nicely reflected by the Mulliken charges of about -0.2 au for
Au3(PH3)4, 0.0 au for Au3(PH3)4(µ2-L) and Au3(PH3)4(µ3-L)2,
and 0.1 au in the full model cluster. A comparison of the
population data for clusters with and without phosphine ligands
shows that the electronic charge donated by the phosphines
(-0.64 au) is mainly taken up by the Cl atoms in the In groups
(-0.48 au), to a smaller extent only by the In atoms (-0.15
This stronger bonding can be attributed to the enhanced
donating ability of Au3(PH3)4 in comparison to Au3. On the
other hand, it is well-known that phosphine ligation enhances
the reactivity of gold since it opens the d shell, facilitating s-d

5704 Inorganic Chemistry, Vol. 36, No. 25, 1997
Gabba¨ı et al.
Table 4. Optimized Bond Lengths (in Å) of the Clusters Au3(PH3)4(I)n, n ) 0-3a
system
Au1-Au3
Au1/3-Au2
Au1-µ2-I
Au1-µ3-I
Au2-µ3-I
Au3(PH3)4(µ2-I)
Au3(PH3)4(µ3-I)2
2.56
2.57
2.57
2.75
2.90
2.90
2.88
3.13
3.04
2.90
2.86
b
Au3(PH3)4I3
2.88
exptlb
1
2
2.56
2.58
2.94
2.86
2.81
2.79
2.85
2.91
2.98
2.93
a Atoms are numbered according Figure 4. b Only the µ3-I atoms have been relaxed; the Au-Au distances have been taken from Au3(PH3)4L2,
and the Au-µ2-I distance has been taken from Au3(PH3)4L. c Shown are the corresponding experimental Au-Au and Au-In distances, averaged
according to the idealization of C2V symmetry.
hybridization.12,13,30-31 This effect may be exemplified by the
effective configuration of Au2. In Au3 a configuration of s1.20
p0.11 d9.80 is calculated, which changes to s1.01 p0.40 d9.57 in Au3-
(PH3)4, in good agreement with theoretical findings for other
gold compounds containing AuPH3 units.13,20,32
Since phosphine ligands play a unique role in gold chem-
istry,5a,10a a detailed discussion of their binding energies is
worthwile. In Table 1 the binding energies of the two
distinguishable phosphine groups, P1/4H3 and P2/3H3, as well
as the average phosphine binding energy, calculated from the
reaction Au3(PH3)4Ln f Au3Ln + 4PH3, are shown for different
cluster subunits. The binding energy for P2/3H3, which has to
share a gold atom with its symmetry partner, is always found
to be lower than for the P1/4H3 groups, which are singly
coordinated to a Au center.
assumed oxidation state of the Au atoms in the full cluster
compound (section 4).
5.4. Geometry Optimization. Since the two very different
Au-Au distances found in both Au/In clusters, about 2.6 and
2.9 Å, are the most striking geometric features of these
compounds, a series of geometry optimizations on various
subunits of a model compound was undertaken. To limit the
computational effort, the model cluster was further simplified
by replacing the complete In groups (including Cl and H2O
ligands) by iodine centers, resulting in the model cluster Au3-
(PH3)4I3. This model should apply equally well to both cluster
compounds 1 and 2. The latter cluster newly synthesized in
the present work differs from 1 only by the ligands attached to
the In centers. Modeling the In groups by iodine centers is
motivated by two observations.
Not unexpectedly for a donating ligand, the phosphine binding
energy increases with growing oxidation of the gold centers:
As an element of the fifth row, iodine exhibits approximately
the same radial extent of the valence orbitals and thus the same
atomic size as indium. Furthermore, the number of seven
valence electrons corresponds to the formal valence electron
count for In in the group (InCl2(thf)) of 1. The geometry
optimization was limited to the distances Au-Au and Au-I,
imposing a C2V symmetry constraint. The Au-P bond lengths
and orientations and the internal degrees of freedom of the
phosphine ligands were kept fixed at the averaged experimental
geometry (see above). The optimizations were carried out in a
cyclic manner, optimizing each degree of freedom separately
until the distances varied were stable to 0.002 Å. To demon-
strate systematically the effect of the oxidation state and the
different building blocks of the Au/In clusters on the geometry
of the central gold triangle, we start the discussion with the
bare Au3 cluster. The moiety Au3 exhibits a Jahn-Teller
distortion D3h f C2V, as expected from a considerartion of the
Au 6s valence orbitals. The bonding orbital a1 is more strongly
localized on Au1/3, and the singly occupied e-type MO (D3h
symmetry) splits into the totally symmetric HOMO 2a1 and the
LUMO b1. The 2a1 MO is bonding between Au1 and Au3, but
antibonding with respect to Au2. The difference in bond lengths
amounts to 0.11 Å (Table 3), which is considerably smaller than
in the Au/In clusters. On the other hand, Au3+, which
corresponds to the cluster core if the formal oxidation states of
the Au atoms are taken literally, forms an equilateral triangle
(Table 3).
It is well-known from previous studies on gold clusters13 that
phosphine ligation elongates Au-Au distances, but only to a
small amount. This effect is demonstrated here by the C3v
symmetric cluster [Au3(PH3)]+. Compared to Au3+, the Au-
Au bond length increases by 0.04 Å (Table 3). A much stronger
geometrical effect is observed when four phosphine ligands are
added to the Au3 unit according to their orientation in the Au/
In clusters 1 and 2. For Au3 the effect of the Jahn-Teller
distortion is considerably enhanced, mainly affecting the weaker
bonds of Au2. In the case of Au3+, the different number of
phosphine ligands attached to the gold atoms induces an
isosceles shape (Table 3). The Au-Au bond elongation of the
13,19
The phosphine-Au bond becomes stronger with increasing
number of L groups attached to the Au3 cluster core (Table 1).
(For P2/3H3 a single exception is found for Au3(PH3)4 which
may be an artifact due to the unrelaxed geometries employed.)
The average phosphine-Au binding energies exhibit this trend,
too. They are calculated to be always slightly smaller (by 2-8
kcal/mol) than the averaged binding energies of the individual
phosphine groups.
To study the effect of charge on the phosphine binding energy
directly, the subclusters Au3(PH3)4Ln (n ) 0-2) have been
ionized in a computer experiment (Table 2). As the average
binding energies shown in Table 2 demonstrate, again an
increase of the phosphine bond strength with growing cluster
charge is found. Interestingly, the trend to larger phosphine
binding energies with growing number of L groups, which has
been observed for the neutral clusters, is reversed for the 2-fold-
ionized species. This effect can be rationalized by observing
that the strengthening influence of a positive charge on the
Au-P bond decreases with increasing coordination of the Au
atoms. While the average phosphine binding energy increases
by 41 kcal/mol due to double ionization of Au3(PH3)4, the same
amount of positive charge results for Au3(PH3)L2 in a bond
strengthening of only 22 kcal/mol. This counteraction of the
effect of positive cluster charging and coordination may prevent
the Au/In clusters from further oxidation of the Au atoms
without decomposition. Comparison of the phosphine binding
energies provides further confirmation of the above assignments
of the formal Au oxidation states. The average phosphine
binding energy is 42.9 kcal/mol in Au3(PH3)4L3, where the Au3-
(PH3)4 subunit carries a Mulliken charge of +0.77 au; this value
compares favorably with the corresponding value of 47.0 kcal/
mol for Au3(PH3)4+, where the positive charge models the
(30) Mingos, D. M. P. J. Chem. Soc., Dalton Trans. 1976, 1164.
(31) De Kock, R. L.; Baerends, E. J.; Boerrigter, P. M.; Hengelmolen, R.
J. Am. Chem. Soc. 1984, 106, 3387.
(32) Pyykko¨, P.; Angermaier, K.; Assmann, B.; Schmidbaur, H. J. Chem.
Soc., Chem. Commun. 1995, 1889.

A Novel Anionic Gold-Indium Cluster Compound
Inorganic Chemistry, Vol. 36, No. 25, 1997 5705
2-fold-ligated Au atom can be rationalized by the fact that the
electron donation of phosphine ligands pushes the highest
occupied orbital of the gold atoms to higher energy.30 In our
study, the HOMO of Au(PH3)2 is found to lie 22 kcal/mol higher
in energy than the HOMO of Au(PH3). Consequently, the
bonding a1 MO of Au3+ becomes more localized on Au1/3 and
the Au-Au bonds to Au2 increase. The same argument applies
in the case of neutral Au3(PH3)4.
leading to a more positive charge of the Au3(PH3)4 unit, in
agreement with the trend derived from the comparison with our
model calculations.
Given the simplicity of our model cluster, it does not come
as a surprise that the calculated Au-I distances do not resemble
the measured Au-In bond lengths too closely (Table 4).
Although a reoptimization of µ3-I in Au3(PH3)4I3 showed a
considerable relaxation of the bonds involved, no closer
agreement with experiment was achieved. Thus, while the
model chosen has been demonstrated to be successful for
simulating and rationalizing the geometry of the central Au3
unit, a quantitative comparison of metal-ligand bond lengths
Au-I and Au-In is beyond its scope.
5.5. Summary. Scalar relativistic all-electron density
functional calculations on model compounds of the Au/In
clusters have been performed, aiming at a clarification of the
oxidation states of the Au and In atoms as well as at a
rationalization of the observed considerable differences in the
Au-Au bond lengths.
The two different oxidation states previously assigned to the
Au atoms were found plausible as judged by a Mulliken
population analysis. For the In atoms, on the other hand, no
interpretable differences of Mulliken charges were obtained and
we resorted to assigning the same intermediate oxidation state
to both of them.
We were able to trace the origin of the two very different
Au-Au bond lengths in the Au/In clusters by means of a series
of geometry optimizations of subunits of the model compound
Au3(PH3)4I3. Two effects are involved: A charge imbalance
is induced on the central Au triangle due to the 2-fold phosphine
ligation of one of the Au atoms. The resulting weaker bond of
the 2-fold-ligated metal atom is further reduced in strength by
the electron-withdrawing µ3-In groups, leading to a final
difference of about 0.3 Å between the two Au-Au distances.
Furthermore, the deviations in the Au-Au bond lengths between
the Cl- and Br-ligated Au/In cluster species were rationalized
by a more positive charge of the Au3(PH3)4 unit in agreement
with a corresponding trend for these bond lengths calculated
for the isolated neutral and singly ionized moiety Au3(PH3)4.
The difference between the two intergold distances amounts
to 0.26 Å in Au3(PH3)4 and 0.13 Å in Au3(PH3)4+. The latter
difference in the Au-Au distances is considerably smaller than
in the parent cluster compounds (1, 0.37 Å; 2, 0.29 Å). Thus,
besides the oxidation of one Au atom, an additional effect of
the In groups is to be expected. The results of geometry
optimizations including iodine atoms as models of the In groups
are given in Table 4. The moiety µ2-I, which acts mainly on
the stronger bond Au1-Au3, has only a minor influence on
the intergold bond lengths. Compared to Au3(PH3)4 the shorter
bond is elongated by 0.03 Å while the longer one shrinks by
0.04 Å. As could have been expected, due to the addition of a
single electron-withdrawing group, a geometry of the gold
+
triangle intermediate between Au3(PH3)4 and Au3(PH3)4 is
adopted (cf. Table 3). A considerably stronger change in
geometry is observed when two µ3-I atoms are added. This
interaction affects predominantly the weak Au1/3-Au2 bond,
which increases by 0.11 Å. We refrained from an optimization
of all 5 degrees of freedom of the complete model Au3(PH3)4I3.
Since the effect of µ2-I on the intergold distances has been
shown to be moderate, we fixed the intergold distances
according to the model Au3(PH3)4(µ3-I)2 and reoptimized only
the position of the µ3-I atoms. Overall, a fairly good estimate
of the geometry of the central Au3 units of the Au/In clusters
was achieved: The difference of the intergold distances is
calculated to be 0.33 Å, which compares favorably with the
experimental results, 0.37 Å for 1 and 0.29 Å for 2. The
absolute values of both distances lie between the values found
experimentally for the compounds 1 and 2 (Table 4).
On the basis of results of these model calculations we may
also rationalize the differences of the Au-Au distances in the
two Au/In clusters 1 and 2. The Au1-Au3 bond in 2 is slightly
longer and the bonds to Au2 are shorter than in 1 (Table 4).
Qualitatively the same trend is observed as a result of ionizing
Au3(PH3)4. Thus, a more positive charge is suggested for the
central Au3(PH3)4 unit in 2 as compared to its Cl-ligated
analogue. A closer look at the differences between the two
clusters reveals that this suggestion is quite plausible. Besides
the substitution of Cl by Br in 2, two thf groups are replaced
by a bridging Br. The electron-withdrawing power of the
additional Br atom is balanced by an extra electron, yielding
an anionic Au/In cluster. Although Br is slightly less elec-
tronegative than Cl, a lower electronic charge has to be attributed
to the Au3 unit in 2 than in 1, since the electron count of two
In atoms is lowered by 2 due to the substitution of the donating
thf groups by a Br- anion. Thus, the electron-withdrawing
capability of the corresponding In groups should be enhanced,
Acknowledgment. This work was supported by Deutsche
Forschungsgemeinschaft and Fonds der Chemischen Industrie.
F.P.G. thanks the Alexander von Humboldt Foundation for a
research fellowship. S.-C.C. is grateful to the Deutscher Aka-
demischer Austauschdienst for a graduate student scholarship.
Support by Degussa AG and Heraeus GmbH through the
donation of chemicals is acknowledged. Mr. J. Riede is thanked
for collecting the X-ray data sets and Katrin Albert for assistance
in the preparation of some of the figures.
Supporting Information Available: Tables listing detailed crystal-
lographic data, atomic positional parameters, and bond lengths and
angles (17 pages). Ordering information is given on any current
masthead page.
IC970725C