Spectral resolution of fluxional organometallics. The observation and FTIR characterization of all-terminal [Rh4(CO)12]

Article abstract of DOI:10.1039/b500044k

In situ FTIR spectroscopy at 1 cm^-1 resolution was conducted on n-hexane solutions of the bridged [Rh₄(CO)₉-(μ-CO) ₃] in the interval T = 268-288 K and P_T = 0.1-7.0 MPa using either helium or carbon monoxide as dissolved gas. Analysis of the spectral data sets was conducted using band-target entropy minimization (BTEM), in order to recover the pure component spectra. A new spectral pattern was recovered with terminal vibrations at 2075, 2069.8, 2044.6 and 2042 cm ^-1. The new spectrum is consistent with an all-terminal [Rh ₄(CO)₁₂] species with a C_3v anticubeoctahedron structure where 2 different [Rh(CO)₃] moieties exist, although the presence of some T_d structure can not be entirely excluded. The equilibrium between all-terminal [Rh₄(CO)₁₂] and the bridged [Rh₄(CO)₉(μ-CO)₃] was determined in the presence of both helium and CO. The equilibrium constant K_eq = [Rh₄(CO)₁₂]/[Rh₄(CO)₉-(μ-CO) ₃] at 275 K was ca. 0.011 and the determined equilibrium parameters were Δ_rG = 12.63 ± 4.8 kJ mol^-1, Δ_rH = -21.45 ± 2.3 kJ mol^-1 and Δ_rS = -114.3 ± 8.35 J mol^-1 K^-1. The free energy indicates a very small difference between the bridged and terminal geometry, and the lower entropy is consistent with a higher symmetry. This finding helps to address a long-standing issue concerning the existence of various [M₄(CO)₁₂] symmetries. In a more general context, the present study illustrates the considerable utility of quantitative infrared spectroscopy (occurring on a fast vibrational timescale) combined with sophisticated deconvolution techniques in order to resolve systems which have been demonstrated to be fluxional on the NMR timescale. The Royal Society of Chemistry 2005.

Full text of DOI:10.1039/b500044k

View Article Online / Journal Homepage / Table of Contents for this issue
F U L L P A P E R
Spectral resolution of fluxional organometallics. The observation and
FTIR characterization of all-terminal [Rh4(CO)12]
Ayman D. Allian and Marc Garland*
Department of Chemical and Biomolecular Engineering, 4 Engineering Drive 4,
National University of Singapore, Singapore 117576. E-mail: chemvg@nus.edu.sg.;
Fax: +65-6-779-1936; Tel: +65-6-874-6617
Received 6th January 2005, Accepted 13th April 2005
First published as an Advance Article on the web 16th May 2005
In situ FTIR spectroscopy at 1 cm−1 resolution was conducted on n-hexane solutions of the bridged [Rh4(CO)9-
(l-CO)3] in the interval T = 268–288 K and PT = 0.1–7.0 MPa using either helium or carbon monoxide as dissolved
gas. Analysis of the spectral data sets was conducted using band-target entropy minimization (BTEM), in order to
recover the pure component spectra. A new spectral pattern was recovered with terminal vibrations at 2075, 2069.8,
2044.6 and 2042 cm−1. The new spectrum is consistent with an all-terminal [Rh4(CO)12] species with a C3v anticube-
octahedron structure where 2 different [Rh(CO)3] moieties exist, although the presence of some Td structure can not
be entirely excluded. The equilibrium between all-terminal [Rh4(CO)12] and the bridged [Rh4(CO)9(l-CO)3] was
determined in the presence of both helium and CO. The equilibrium constant Keq = [Rh4(CO)12]/[Rh4(CO)9-
(l-CO)3] at 275 K was ca. 0.011 and the determined equilibrium parameters were DrG = 12.63 4.8 kJ mol−1, DrH =
−21.45 2.3 kJ mol−1 and DrS = −114.3 8.35 J mol−1 K−1. The free energy indicates a very small difference
between the bridged and terminal geometry, and the lower entropy is consistent with a higher symmetry. This finding
helps to address a long-standing issue concerning the existence of various [M4(CO)12] symmetries. In a more general
context, the present study illustrates the considerable utility of quantitative infrared spectroscopy (occurring on a fast
vibrational timescale) combined with sophisticated deconvolution techniques in order to resolve systems which have
been demonstrated to be fluxional on the NMR timescale.
Introduction
The Group VIII (Co, Rh, Ir) tetranuclear metal carbonyl clusters
[M4(CO)12] have been the subject of intense research for over
40 years. In particular, their structures and fluxional behavior
in solution have generated considerable discussion.1–7 Three
primary all-terminal geometries [M4(CO)12] obeying the 18e-
rule for closed polyhedra have been discussed, namely the (1) Td
symmetric cubeoctahedral, (2) the T symmetric icosahedron and
(3) the C3v anticubeoctahedron.5,8,9,10 In addition two bridged
species have been discussed, (4) the C3v symmetric icosahedral
[M4(CO)9(l-CO)3] and (5) the D2d symmetric [M4(CO)8(l-CO)4].
For completeness, it can be mentioned that (6) a triply bridged
CO geometry (CO)8(l3-CO)4] has also been suggested as a
possible geometry.4 Fig. 1 provides representations for these
structures.
In the case of cobalt, the X-ray crystal structure shows a
C3v symmetric icosahedral [Co4(CO)9(l-CO)3].11–16 In solution,
both the normal infrared spectra and the 13CO enriched spectra
are consistent with the same geometry.17,18 In addition, the
13C NMR,19 17O NMR20 and the 59Co NMR21–24 solution
spectra are also consistent with the C3v symmetric icosahedral
geometry. The 13CO NMR studies provided clear evidence for
cluster fluxional behavior at higher temperatures particularly
coalescence at ca. 283 K.10,23,25,26 In the case of the iridium X-ray
crystal structure, a complication arises due to the presence of 3
distinct molecules in the unit cell. However, it was concluded that
there is essentially Td symmetric cubeoctahedral [Ir4(CO)12].27,28
The Td symmetry is preserved in solutions as indicated by 13CO
NMR29 and IR spectroscopy.30
The rhodium cluster [Rh4(CO)12] represents a special case
due in part to its highly fluxional behavior. The reddish orange
[Rh(CO)3]n was first reported by Heiber and Lagally,31 and
Chini reported the synthesis and isolation of [Rh4(CO)12].32–34
The X-ray structure has been determined on at least 3 occasions,
Fig. 1 The proposed structures for [M4(CO)12]: (1) Td cubeoctahedron;
and an idealized C3v symmetric icosahedral [Rh4(CO)9(l-CO)3]
(2) T icosahedron; (3) C3v anticubeoctahedron; (4) C3v icosahedron; (5)
has been shown.14,35,36 In solution, the infrared18 and the
D2d with four bridging carbonyls; (6) triply bridged (structures 1–4, 5
and 6 were adapted from ref. 5, 15 and 4 respectively).
Raman spectra37 are consistent with the same geometry at room
T h i s j o u r n a l i s
T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 5
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5
1 9 5 7
©

View Article Online
temperature. In addition, the 13C NMR3,10 and 103Rh solution
spectra38 are also consistent with the bridged C3v symmetric
icosahedral geometry at least in the vicinity of 213 K. However,
at ca. 268 K coalescence occurs3,10,39 indicating equivalence of
all CO ligands on the NMR timescale. This suggests the inter-
conversion: bridged C3v ↔ all-terminal ↔ bridged C3v.1–3,9,10 The
possible all-terminal geometries include Td cubeoctahedron,
T icosahedron and C3v anticubeoctahedron. Arguments exist
to support the T icosahedron geometry as the lowest energy
configuration.40,41
Rhodium and rhodium carbonyl complexes in the presence
of CO, either in matrices or solution, have yielded numerous
mononuclear and polynuclear species. An extensive list is pro-
vided in Table 1. The interconversion between [Rh4(CO)12] and
[Rh2(CO)6(l-CO)2] is extremely rapid,42–44 and the omnipresence
of [Rh6(CO)12(l3-CO)4] is consistently observed even under high
CO pressure.
In the present contribution, the solution chemistry of
[Rh4(CO)12] is re-investigated under various helium or CO
partial pressures in n-hexane as solvent in the interval T =
268–288 K and PT = 0.1–7.0 MPa using FTIR as the spectro-
scopic tool and band-target entropy minimization (BTEM) as
the signal processing algorithm.50,51 BTEM has repeatedly shown
its ability to retrieve pure component spectra with very weak
signal intensities, i.e. [HRh(CO)4].52 Since infrared spectroscopy
has a significantly faster timescale than NMR4 and since these
experiments are conducted above the NMR coalescence temper-
ature, one or more all-terminal geometries should be resolvable,
if the superimposed signals can be adequately deconvoluted.
More generally, the problem of fluxionality on the NMR
timescale and its resolution on a fast infrared timescale can
be illustrated using Scheme 1. Organometallic molecules [MxLn]
with geometries 1 and 2 exist in equilibrium exchange, where
the transition state ‡ separates the two states. If transformation
occurs on a NMR timescale, then the signal(s) are broadened
and are no longer specific to the individual states involved.
In contrast, infrared spectroscopy, occurring on the vibration
timescale, can discern the individual states. Thus infrared spec-
troscopy, coupled with a sophisticated deconvolution technique,
holds considerable promise for resolving spectroscopically sim-
ilar states in fluxional systems.
Results
Spectroscopic aspects
Seven isothermal semibatch reactions were conducted. Five were
conducted in the presence of CO and two were conducted in the
presence of helium. Although the experimental spectral range
was 1000–2500 cm−1, large regions contained no identifiable
changes, and the spectral data was truncated to the region 1800–
2100 cm−1. The 7 semibatch reactions resulted in 475 in situ
spectra. These were consolidated into one matrix A475 × 1501
.
Singular value decomposition (SVD) and BTEM
In order to extract the pure components, the matrix was decon-
voluted via BTEM without any preconditioning. Mathematical
details of BTEM are available elsewhere.50,51 The first step
in BTEM is the singular value decomposition (SVD) of the
observed spectra, A475 × 1501, in order to obtain the right singular
VT matrix whose row vectors contain untangled and abstract
information on the pure component spectra. Indeed, inspection
of the VT vectors reveals the characteristic features of the pure
components spectra present in solution as shown in Fig. 2,
where some particularly noteworthy spectral features have been
marked.
Scheme 1 The transformation of an organometallic [MxLn(1)] , through
a transition state [MxLn]‡, to another organometallic [MxLn(2)] and then
back to its original form [MxLn(1)] . This equilibrated exchange process
is typically said to be fluxional if it occurs on the NMR timescale.
Table 1 List of known unsubstituted neutral rhodium carbonyl species
relevant to in situ spectroscopic studies in matrices and hydrocarbon
solutions
Species
Band/cm−1
Ref.
[Rh(CO)]
[Rh(CO)2]
[Rh(CO)3]
[Rh(CO)4]
2012.8, 2008a
45
45
45
45
46
42
43
44
2014.6a
2018.4a
2019, 2012a
[Rh2(CO)6(l-CO)2]
2060, 2043, 2038, 1852, 1830a
2086, 2061, 1860, 1845b
2084, 2060, 1862, 1847c
2087, 2062, 1852, 1832e
2084.6, 2060.2, 1861.2, 1845.4c This study
[Rh4(CO)9(l-CO)3]
[Rh6(CO)12(l3-CO)4]
2076, 2071, 2045, 1886b
2075, 2070, 2045, 1887c
2076, 2072, 2045, 1885d
2074, 2069, 2043, 1885c
2075.2, 2070, 2044.8, 2042.8,
1885.6c
42
43
44
47
Fig. 2 Several right singular vectors (VT) from the SVD of the
experimental data: a) 1st vector, b) 3rd vector, c) 6th vector, d) 7th vector,
e) 12th vector, f) 70th vector.
This study
2070, 1793e
48
49
2077, 2046, 1805f
The BTEM algorithm was used to target individual spectral
features of interest in VT vectors, and subsequently reconstruct
pure component spectra one-at-a-time. A list of the targeted
spectral features from Fig. 2 used for the BTEM calculations and
the resultant pure components obtained are shown in Table 2.
2075.4, 2044.6, 1818.6c
This study
a Argon matrix. b Liquid paraffin–heptane mixture. c Hexane.
d Dodecane. e Nujol. f Chloroform.
1 9 5 8
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5

View Article Online
Table 2 List of the bands used for BTEM calculations and resultant pure components
Marked spectral
VT vector where marker
is first indicated in Fig. 2
Singular value associated
Spectral marker
positions/cm−1
Recovered species
with VT vector
1
2
3
4
5
6
7
8
9
2025.4, 1896
2075.2, 2070, 2044.8, 1885.6
2096
1844.2
2075, 2069.8
2084.6, 2060.2
1818
2046
2023.6
Hexane-solvent
[Rh4(CO)9(l-CO)3]
CO dissolved gas
Moisture
1
1
1
1
3
6
7
7
7
640.07
640.07
640.07
640.07
29.119
1.8009
1.0877
1.0877
1.0877
0.4840
0.01487
Species X
[Rh2(CO)6(l-CO)2]
[Rh6(CO)12(l3-CO)4]
[Ni(CO)4]
[Fe(CO)5]
Failed attempts
10
11
1883
Mostly white noise
12
70
It is useful to mention that SVD orders the vectors according
to their associated singular values, which represent their contri-
bution to the total variance in the observations. Thus, features
associated with major components having high concentrations
appear in the initial vectors with large singular values, while
the features associated with minor components usually emerge
in the latter vectors with smaller singular values. The ratio of
component signal to random noise is largest in the first vector
and decreases with subsequent vectors. The 70th right singular
vector is predominantly white noise.
The pure component spectra of all the metal carbonyls
observed in this study are shown in Fig. 3. The pure component
spectra, arising from the rhodium carbonyls [Rh4(CO)9(l-CO)3]
and [Rh6(CO)16], possess very high signal to noise ratios.
The pure component spectra from [Rh2(CO)6(l-CO)2] and
the species X, have slightly elevated noise levels. The pure
component spectra from [Ni(CO)4] and [Fe(CO)5], which are
at trace concentrations, have relatively high noise levels. The
wavenumbers for the rhodium carbonyls mentioned above and
determined in this study are presented in Table 1. It can be noted,
that the use of higher resolution has allowed the observation of
the splitting of the broad band of the bridged [Rh4(CO)9(l-CO)3]
at ca. 2043 cm−1 into a sharp band at 2044.8 cm−1 and a shoulder
at 2042.8 cm−1. This result is in agreement with the observed
spectra of [Rh4(CO)9(l-CO)3] made by Bor.18 The observed
wavenumbers for [Ni(CO)4] and [Fe(CO)5] were 2046 and 2023.6,
2000.4 cm−1 respectively, consistent with the literature values.53,54
The impurity [Fe(CO)5] is frequently present in commercial CO
gas stored in steel cylinders, but the present in situ study is the
first in which this group has been able to identify the presence
of [Ni(CO)4]. In addition, with higher resolution, a third peak
for [Rh6(CO)16] at 2045.6 cm−1 is more clearly seen in this study,
compared to previous studies by this group.55 The spectrum of
[Rh6(CO)16] is in agreement with the one obtained by Beck56 and
Weber49 and consistent with the molecular geometry Td.57 No
further pure component metal carbonyl spectra could be readily
obtained.
Signal recovery in this study was high. The signal contribution
of each pure component to the total observed signal is shown
in Table 3. The organometallic species contribute ca. 25% of the
signal in the 5 CO containing experiments, and ca. 40% of the
signal in the 2 He containing experiments. The sum total of all
signals recovered is very close to 100% in both the CO and He
experiments.
Normalized concentrations
Each of the 7 semibatch reactions consisted of 2 additions of
[Rh4(CO)9(l-CO)3] to the system, and many perturbations in
CO or helium (see Experimental section). The first addition
of [Rh4(CO)9(l-CO)3] corresponds to the pressure interval 0.1–
2.0 MPa and the second addition of [Rh4(CO)9(l-CO)3] corre-
sponds to the pressure interval 3.0–7.0 MPa. Fig. 4 provides
the normalized concentrations of all the metal carbonyl species
for these 7 semibatch experiments consisting of 475 spectral
Table 3 Percentage of reconstructed integrated absorbance of each
component compared to the total original experimental data
Signal ratio (%)
Components
CO set
He set
57.41
Hexane
35.64
22.89
39.95
1.002
0.736
0.710
0.375
0.371
0.123
101.7
[Rh4(CO)9(l-CO)3]
Dissolved CO
Moisture
37.74
a
2.134
Species X
1.77
a
[Rh2(CO)6(l-CO)2]
[Rh6(CO)16]
[Ni(CO)4]
[Fe(CO)5]
Total
0.130
a
a
99.23
a Signal undetected.
Fig. 3 Pure component spectra obtained from BTEM analysis.
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5
1 9 5 9

View Article Online
Fig. 4 The normalized metal carbonyl concentrations provided by
BTEM analysis for the 7 semibatch reactions. Note that the mean
concentrations of the species vary over many orders of magnitude,
therefore, one-to-one comparison of the individual scales in the figure is
not possible.
measurements. Since the pure component spectra obtained
from BTEM in Fig. 3 are all of unit height, the calculated
concentrations shown in Fig. 4 are relative and not properly
scaled at this point in the analysis.
Fig. 5 Validation/uniqueness test using TTFA (above). The spectrum
of species X superimposed on that of [Rh4(CO)9(l-CO)3] in the terminal
CO region (middle). Additional comparison to the Td symmetry of
[Ir4(CO)12] (below).
The CO experiments show that (1) the concentrations of X are
fixed ratios of the [Rh4(CO)9(l-CO)3] concentrations for the first
perturbation of [Rh4(CO)9(l-CO)3] up to 2.0 MPa CO but slowly
decrease during the second perturbation of [Rh4(CO)9(l-CO)3]
up to 7.0 MPa CO, (2) the omnipresent but not equilibrated
[Rh6(CO)16] exists in all experiments, and generally increases
with increasing CO pressure (addition of more impurities
promotes formation), (3) the equilibrated dinuclear complex
[Rh2(CO)6(l-CO)2] increases as a function of CO pressure as
anticipated and (4) the impurities [Ni(CO)4] and [Fe(CO)5]
increase with increasing CO pressure.
The helium experiments show that (1) the concentrations of X
are fixed ratio’s of the [Rh4(CO)9(l-CO)3] concentrations even
under variable pressure, (2) the concentration of [Rh6(CO)16]
is non-zero and shows little variation with increasing pressure
during each semi-batch run and (3) the dinuclear complex
[Rh2(CO)6(l-CO)2] is not formed since no excess CO is present.
This last observation is consistent with Whyman’s work with
[Rh4(CO)9(l-CO)3] under nitrogen.42
is superimposed. The 4 bands of X, at 2075, 2069.2, 2044.6,
2042 cm−1 are virtually coincident with the terminal bands of
[Rh4(CO)9(l-CO)3] but much narrower. In addition, the species
X appears to be two super-imposed patterns similar to the all-
terminal [Ir4(CO)12]. The latter possesses Td symmetry where
each of the [Ir(CO)3] moieties are equivalent and have two
vibrational bands.30 Therefore, the new spectrum suggests that
there are two [Rh(CO)3] groups present with different local
environments.
The normalized concentration profiles in Fig. 4, particularly
those under helium and those under CO up to and including
2.0 MPa, indicate that the concentration of the new species
X is a fixed ratio compared to [Rh4(CO)9(l-CO)3] at any
temperature T. This concentration dependence is consistent
with an elemental stoichiometry for the species X identical to
[Rh4(CO)12]. We propose that the species X is in fact, at least
1 isomer of the all-terminal [Rh4(CO)12]. The BTEM recovered
spectrum could represent (A) just one isomer of [Rh4(CO)12]
possessing two different [Rh(CO)3] moieties or (B) more than
one isomer of all-terminal [Rh4(CO)12].
Preliminary interpretation of the new experimental spectrum
Real concentrations and equilibria
As seen from Fig. 3, the species X has no bridging CO. This
spectrum was obtained using 70 basis vectors VT. In order
to check whether any serious signal artifacts were introduced
into the spectral reconstruction, a validation/uniqueness test
in the form of target transform factor analysis (TTFA)58 was
carried out. In this analysis, the BTEM reconstruction of X
is projected onto the reduced vector-space formed from the
first 25 VT vectors. If the resulting vector is very similar to the
tested vector, then the reconstructed spectrum probably does not
contain any serious artifacts. This is indeed the case as seen in
Fig. 5. Furthermore, an expanded view of the terminal region
is provided in Fig. 5 where the spectrum of [Rh4(CO)9(l-CO)3]
The data for the CO series of experiments were regressed to
obtain the absolute (rather than normed) absorptivities for the
4 rhodium carbonyl clusters, namely, [Rh4(CO)9(l-CO)3], all-
terminal [Rh4(CO)12], [Rh2(CO)6(l-CO)2] and [Rh6(CO)12(l3-
CO)4]. The mass balances for rhodium in each semibatch run,
and each perturbation within the run, were used to constrain the
numerical solutions. Details of similar optimization methods to
properly scale absorptivities can be found elsewhere.58
The maximal absorptivities for [Rh4(CO)9(l-CO)3], [Rh4-
(CO)12], [Rh2(CO)6(l-CO)2] and [Rh6(CO)12(l3-CO)4] were;
2.82 × 104 l mol−1 cm−1 (2070 cm−1), 2.85 × 105 l mol−1 cm−1
1 9 6 0
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5

View Article Online
(2069.8 cm−1), 2.72 × 104 l mol−1 cm−1 (2060.2 cm−1) and 1.34 ×
105 l mol−1 cm−1 (2075.4 cm−1) respectively. The absolute values
for the absorptivities of [Rh4(CO)9(l-CO)3] and [Rh2(CO)6(l-
CO)2] determined in this study are only marginally higher (ca.
10%) than those previously determined,43,59 and this can be
traced to the slightly higher resolution used in this study, as
well as the more accurate mass balances.
An independent calibration for [Rh6(CO)12(l3-CO)4] in n-
hexane proved to be very difficult to obtain due to its extreme
low solubility in hexane, but a minimum value of (4.1 0.98) ×
104 l mol−1 cm−1 (2075.4 cm−1) is certain. In addition the pure
component spectra of [Rh6(CO)12(l3-CO)4] obtained confirmed
the presence of the third peak at 2045.6 cm−1.
Integrating the properly scaled absorptivities of [Rh4(CO)9(l-
CO)3] and all-terminal [Rh4(CO)12] over the data channels
used (1501 channels from 1800–2100 cm−1), provided values
of 1.8 × 106 l channel mol−1 cm−1 and 3.7 × 106 l channel
mol−1 cm−1 respectively. Assuming that similar CO force con-
stants are involved, the determined properly scaled absorptivity
for [Rh4(CO)12] cannot be seriously in error.
The maximum mole fractions for [Rh4(CO)9(l-CO)3],
[Rh4(CO)12], [Rh2(CO)6(l-CO)2] and [Rh6(CO)12(l3-CO)4] ob-
served in the entire experimental study were; 6 × 10−5, 7.5 ×
10−7, 5.5 × 10−7 and 6.5 × 10−7 respectively.
The primary rhodium carbonyl cluster equilibrium present in
this study is shown in eqn. (1).
Fig. 7 Estimation of thermodynamic parameters based on the Van’t
[Rh4(CO)12]
ꢀ
ꢁ
Keq1
=
(1)
Hoff equation.
Rh4(CO)9(l-CO)3
terminal CO vibrations should be broad/broader. Third, if
a species [Rhx(CO)yL] were produced in situ from coordina-
tion/reaction of a variable amount of an impurity L (i.e.
from introduced CO) with [Rh4(CO)12], then the concentration
of [Rhx(CO)yL] would be influenced during the 0–3.0 MPa
partial pressure CO experiments, the observed terminal CO
vibrations should be broad/broader, and [Rhx(CO)yL] should
not be identifiable in the helium series of experiments. Since such
correlations/observations are not present in the experiments
performed, the probability that one or more [Rhx(CO)yL]
complexes has lead to misinterpretation of the results seems
extremely remote.
The calculated equilibrium constants are shown in Fig. 6.
Fig. 6 The calculated equilibrium constants for [Rh4(CO)12].
The helium data sets as well as the CO data sets below 3.0 MPa
provided consistent equilibrium constants. The reason for the
variable Keq at P > 3.0 MPa is at the moment not entirely clear
but may be due to significantly changing the bulk dielectric
constant of the liquid phase when more than ca. 3 mole percent
of dissolved CO is present. The lower pressure data was regressed
for Keq(T). A Van’t Hoff plot is provided in Fig. 7.
Discussion
In recent years, indications for the existence of observable
quantities of all-terminal [Rh4(CO)12] have been obtained during
in situ infrared spectroscopic studies of the unmodified rhodium
catalyzed hydroformylation made with 4 cm−1 resolution.51,52
The narrow bands at ca. 2068 and 2076 cm−1 were observed, but
the presence of bands in the region of 2040 cm−1 could not be
conclusively substantiated. In the present study, far more spectra
have been measured, under non-catalytic conditions, with higher
resolution 1 cm−1, and under a much wider range of CO/He
partial pressures. Similar narrow bands are again deconvoluted,
but more importantly, 2 bands at 2044.6 and 2042 cm−1 have been
resolved. The thermodynamic data confirms the stoichiometry
[Rh4(CO)12], and the free energy of reaction indicates a structure
only 12 kJ mol−1 higher than [Rh4(CO)9(l-CO)3].
Concluding remarks on the possible role of impurities
In the present results, the mean concentrations of the new species
[Rh4(CO)12] was estimated as ca. 0.6 × 10−6 mole fraction. Since
impurities are always present in chemical systems, and since the
new species exists at such low concentrations, the possibility
of misinterpretation of the results should be addressed. The
possibility that the new species represents one or more impurities
can be excluded with considerable certainty, using the following
arguments. First, if the spectrum X was associated with an impu-
rity in the starting material i.e. an [Rhx(CO)yL] complex present
in the polycrystalline [Rh3(CO)9(l-CO)3], then its concentration
would not be expected to be a function of temperature, and/or its
concentration would be influenced during the 0–3.0 MPa partial
pressure CO experiments, and finally the observed terminal CO
vibrations should be broad/broader due to lower symmetry (in-
fra vida). Secondly, if a species [Rhx(CO)yL] were produced in situ
from coordination/reaction of a fixed amount of an impurity L
(i.e. from the solvent) with [Rh4(CO)12], then the concentration
of [Rhx(CO)yL] would be influenced during the 0–3.0 MPa
partial pressure CO experiments, and as well, the observed
The spectra of all-terminal [Rh4(CO)12] and variety of structures
As stated previously in the Results section, the present spectrum
of [Rh4(CO)12] arises either from (A) one isomer possessing
two different [Rh(CO)3] groups or (B) more than one isomer
of an all-terminal [Rh4(CO)12] molecule. The line shapes are
considerably narrower than usual for rhodium carbonyl species.
The deconvoluted spectra indicate line widths at half height
of ca. 2 cm−1 for [Rh4(CO)12] compared to ca. 5 cm−1 for
[Rh4(CO)9(l-CO)3] in the terminal CO region. This is consistent
with the fact that higher molecular symmetry leads to more
localized energy levels. This tendency is readily apparent in a
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5
1 9 6 1

View Article Online
variety of systems including (i) [Mo(CO)6] and [Mo(CO)5-
PPh3],60 (ii) [Mn2(CO)10] and [Mn2(CO)9]61 and (iii)
[{(C6H11)3P}2Mo(CO)4] and trans[{(PriO)3P}2Mo(CO)4] where
the “irregular packing of the PrO fragment” in the latter
significantly broadened the IR bands.62
If case (A) is occurring, then there is one isomer possessing
two different [Rh(CO)3] groups. This excludes Td symmetric
[Rh4(CO)12] by itself since only 2 bands will arise. The T-
icosahedron is believed to have the lowest energy of the all the
suggested [M4(CO)12] terminal structures41 but this geometry
is not consistent with 4 observable bands. The C3v anticuboc-
tahedron symmetry was suggested by Johnson, as the most
appropriate intermediate that can explain the total coalescence
of the NMR signal at T = −5 K, and has two different [Rh(CO)3]
groups, i.e. those on the basal plane, a, and the apical group, b,
as shown in Fig. 8. This would be consistent with four observed
bands.
Scheme
2
The transformation of the organometallic cluster
[Rh4(CO)9(l-CO)3], through one or possibly more transition states
[Rh4(CO)12]‡, to [Rh4(CO)12]. This equilibrated exchange process is
known to be fluxional on the NMR timescale at ca. 298 K. The
all-terminal state [Rh4(CO)12] was resolved by infrared spectroscopy at
ca. 298 K.
Bulk physico-chemical effects on equilibria
The helium experiments clearly showed that there is no ob-
servable total pressure effect on the [Rh4(CO)9(l-CO)3] and
[Rh4(CO)12] equilibrium, and the low pressure CO experiments
provide consistent values for the equilibrium constant. However,
the high CO partial pressures indicate a decreasing value for Keq
as a function of PCO, thus, CO concentration in solution. This
effect appears to be associated with a change in a bulk solution
physico-chemical property.
Non-polar hexane has a zero dipole moment,64 and disso-
lution of atomic helium is not expected to change the bulk
dielectric constant since it also has a zero dipole moment.64
However, the dissolution of CO, with a non-zero dipole moment
of 0.1,65 will increase the bulk solution dielectric constant in
a non-negligible manner. In addition, the highly symmetric
species [Rh4(CO)12] should have a vanishingly small overall
dipole moment whereas the same cannot be said of [Rh4(CO)9(l-
CO)3]. Taken together, the above physico-chemical issues argue
for a decreasing Keq1 with increasing CO concentration. As
the solution polarity increases, the least polarized molecule
is less favored thermodynamically. This finding raises issues
about the importance of bulk physico-chemical properties on
organometallic transformations but at the same time, the general
lack of such solution property information.
Fig. 8 The two possible geometrical isomers. In the Td symmetry all
[Rh(CO)3] moieties are similar (a). In contrast, the C3v symmetry’s apical
[Rh(CO)3] has a different local environment (b).
In case (B) the new spectrum arises from the simultaneous
presence of more than one [Rh4(CO)12] cluster. For example,
one possibility is the simultaneous existence of C3v symmetric
[Rh4(CO)12] (major isomer) and some Td symmetric [Rh4(CO)12]
(minor isomer). Both have a common [Rh(CO)3] local symme-
try/environment, a, as shown in Fig. 8. Since the integrated
areas under the respective bands (i.e. 2075 : 2069.2 and 2044.6 :
2042 cm−1) are not exactly 1 : 3 as anticipated if only the C3v
isomer were present, the coexistence of some Td isomer should
not be ruled out. It is worth mentioning that a mixture of
C3v and Td will still give rise to 4 observable bands due to
superpositions of the signals due to the a groups. It can be
noted that the existence of 2 energetically similar clusters is
consistent with remarks by Farrugia63 that “carbon monoxide
may bind to ensembles of metal atoms in energetically similar,
but geometrically distinct modes”.
Expressed in somewhat different form, the BTEM recon-
structed spectrum is not conclusive enough by itself to differenti-
ate between the presence of 1 isomer of [Rh4(CO)12] or the simul-
taneous presence of 2 isomers of [Rh4(CO)12]. Indeed, the BTEM
recovered spectrum could arise if almost-exact collinearity of
2 concentration profiles occurs. NMR data3,10 confirms that
inter-conversion of [Rh4(CO)9(l-CO)3] and [Rh4(CO)12] isomers
is occurring on a fast timescale (ca. 10−3 s), and the present
thermodynamic results confirm that the terminal isomer(s) have
only marginally higher free energy. Therefore, given the small
temperature interval used, little concentration variation might
occur between 2 co-existing isomers and an inseparable super-
position of spectra would result.
[Rh2(CO)6(l-CO)2]. In n-hexane solution, [Rh4(CO)9-
(l-CO)3] is rapidly converted to [Rh2(CO)6(l-CO)2] according
to eqn. (2).
[Rh4(CO)9(l-CO)3] + 4CO ↔ 2[Rh2(CO)6(l-CO)2]
(2)
The equilibrium constant can be written as shown in eqn. (3).
[Rh2(CO) (l-CO )2]
6
2
ꢀ
ꢁ
Keq2
=
(3)
Rh4(CO) (l-CO) [CO]4
9
3
Under the CO partial pressures used in this study, the
maximum conversion observed in this study was only ca. 0.4%.
The equilibrium between [Rh4(CO)9(l-CO)3] and [Rh2(CO)6(l-
CO)2] was also obtained. At low CO pressures, the conversion
is too small to provide meaningful values of Keq. At pressures
of ca. 50–70 bar CO, consistent Keq of ca. (5.21 2) × 10−3 are
obtained. The present value of Keq2 is reasonably consistent with
the literature value provided by Oldani and Bor, and obtained
from data in the pressure range of 100–200 bar.43
Further spectroscopic signal issues
The presence of an observable triply bridged species
[M4(CO)8(l3-CO)4] is not supported by the present data. A triply
bridged CO on a neutral rhodium cluster should give rise to a
spectral feature at ca. 1820 cm−1, as is the case for [Rh6(CO)12(l3-
CO)4]. No new unaccountable spectral features in the region
1800–1840 cm−1 were seen in the right singular vectors.
Given the well known fluxionality of rhodium carbonyl
clusters, and the labile inter-conversion of rhodium carbonyl
species, the possible existence of an all-terminal dinuclear
rhodium carbonyl [Rh2(CO)8] must be considered. This is
particularly true in light of the reported coexistence of both
the bridged and terminal dimers of the phosphite substituted
[Rh2(CO)6{P(OPh3)}2] and [Rh2(CO)7{P(OPh3)}].66 Also, the
analogous cobalt chemistry is well documented by the spectro-
scopic identification of [Co2(CO)8] equilibrated in solutions of
Although not complete, the identification of all-terminal
[Rh4(CO)12] provides some new information on the fluxional
system Scheme 2. Muetterties indicated that the transition
state [Rh4(CO)12]‡ has a free energy 56 kJ mol−1 greater than
[Rh4(CO)9(l-CO)3].4 The present thermodynamic data indicates
that the species [Rh4(CO)12] has a free energy 12 kJ mol−1
greater than [Rh4(CO)9(l-CO)3] or ca. 44 kJ mol−1 lower than
[Rh4(CO)12]‡.
1 9 6 2
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5

View Article Online
[Co2(CO)6(l-CO)2]/aliphatic hydrocarbon.67–69 By comparison
with [HRh(CO)4]52 and [RCORh(CO)4],70 a totally symmetric
vibration in the region of ca. 2110–2130 cm−1 may exist due
to the local symmetry of the moiety [Rh(CO)4] in [Rh2(CO)8].
Extensive analysis of the right singular vectors did not reveal any
new spectral features. Since there are many reasons to presume
the simultaneous co-existence of [Rh2(CO)8] with [Rh2(CO)6(l-
CO)2], the present results suggest that its concentration is simply
too low for detection. Broadly speaking, if the Keq for the latter
species is similar to the observed Keq1 between the tetranuclear
unbridged dimer and the bridged form, then, only parts-per-
billion of the unbridged (1% of the bridged) are expected in
solution.
The vibrational signals of the bridged carbonyls in neutral
rhodium species occur in the interval of ca. 1840–1900 cm−1. The
lower wavenumber region is associated with [Rh2(CO)6(l-CO)2]
while the higher wavenumber region is associated with higher
nuclearity species. As indicated by the SVD results shown in
Fig. 2, there was a rather clear indication that more than one
spectral feature is embedded in the higher wavenumber bridging
CO region of the present experimental data. The most prominent
feature is the band at 1885 cm−1, which arises from [Rh4(CO)9(l-
CO)3], but a very symmetric feature is also apparent at ca.
1883 cm−1 when the observations are untangled (see the 12th
VT vector). This feature is only really apparent in the VT vectors
greater than ca. 10. This suggests a very low mean concentration.
Repeated attempts to reconstruct the whole spectrum associ-
ated with 1883 cm−1 using BTEM failed. Reconstructed spectra
containing the signal at 1883 cm−1 where obtained, however, the
spectra were clearly super-positions of more than one species.
This is evidenced by features associated with n-hexane as well
as other known species. Restricting this discussion to coordi-
nately saturated tetranuclear clusters, the species [Rh4(CO)8(l-
CO)4], [Rh4(CO)8(solvent)(l-CO)3],71 [Rh4(CO)10(l-CO)3] and
[Rh4(CO)12(l-CO)2] are potential candidates. The last two
structures would be analogues to the postulated structures
[Co4(CO)10(l-CO)3] and [Co4(CO)12(l-CO)2] discussed at length
by Bor.72
the reaction system. In any typical experiment at ca. 50 bar
pressure, 5 × 10−5 mol [Rh4(CO)9(l-CO)3], 2 mol of solvent
and ca. 0.2 mol of gas were present. The step-wise increase in
[Rh6(CO)16] concentrations is undoubtedly associated with the
introduction of trace gas-phase impurities and the subsequent
degradation of [Rh4(CO)9(l-CO)3].
Resolution, BTEM and time-scales
This study is the first in which 1 cm−1 resolution and BTEM
have been combined for an organometallic study. The increased
resolution permitted more exact deconvolution of the species
present using BTEM. For example, this is the first study in
which [Ni(CO)4] was successfully deconvoluted in the presence
of rhodium carbonyls. The extreme overlap between the band at
2046 cm−1 for [Ni(CO)4] and that at 2044 cm−1 for [Rh4(CO)12]
is very problematic, especially when the ratio of signal intensity
in that region is of the order of 10%. Also, the resulting BTEM
spectral estimates are, on average, a little narrower. This permits
total signal recovery very close to 100% as shown in Table 3.
One possible drawback for the use of high resolution is the
observation of rotational modes for dissolved CO which can
complicate deconvolution. These modes can be seen in Fig. 2
at ca. 2096 cm−1. At lower resolution, these rotational modes
are not apparent and BTEM deconvolution is more readily
achieved.
Perhaps the most important broad implication of the present
study concerns the combined use of faster-than-NMR spectro-
scopies and BTEM to resolve NMR fluxional systems.
Conclusion
High resolution infrared spectroscopy was combined with the
BTEM deconvolution technique in order to obtain the pure
component spectrum of the all-terminal cluster [Rh4(CO)12].
This was achieved in spite of the fact that there was only ca.
1% conversion of the starting cluster [Rh4(CO)9(l-CO)3], and
that the terminal vibrations of both species are very similar.
The deconvolution was accomplished in the temperature regime
where the system is fluxional on the NMR timescale. The
quantitative high pressure infrared measurements confirmed
the stoichiometry and provided the free energy, enthalpy and
entropy of reaction. The present results have straightforward
general implications for the analysis of fluxional systems,
namely, a fast timescale spectroscopy (infrared) combined with
deconvolution can resolve spectroscopically complex systems.
Impurities
The relative concentration profiles shown in Fig. 4 indicate
that the concentrations of [Ni(CO)4] and [Fe(CO)5] are zero
in the helium experiments and step-wise increasing with CO
perturbations. The observed trends are consistent with the
known problems associated with commercial grade CO stored in
stainless steel cylinders. Although an additional deoxy/zeolite
gas purification system was used (see Experimental section for
further details), minute quantities of these carbonyl impurities
enter the reaction system. By visual inspection, neither [Ni(CO)4]
nor [Fe(CO)5] can be seen in the experimental spectra. Only after
SVD is performed (see Fig. 2) can these species be detected, and
only after BTEM is it possible to confirm their identities. The
SVD results (Fig 2), BTEM results (Fig 3) and the numerical
results (Table 2), namely (1) the presence of spectral features in
VT = 6, 7, (2) the relatively low signal to noise ratio of ca. 15 :
1 and 10 : 1, and (3) the very small integrated signal intensities
confirm the very low concentrations present.
In Fig 4, the concentrations of [Rh6(CO)16] remained es-
sentially constant in the helium experiments, and gradually
increased during the CO semibatch experiments. At the tem-
peratures and conditions used in this study, [Rh6(CO)16] is
not equilibrated with [Rh4(CO)9(l-CO)3]. Indeed, if it were
in equilibrium exchange, the concentrations of [Rh6(CO)16]
would decrease rather than increase with increasing CO. The
observed trends are consistent with the typical ultra-high
purity of commercial helium compared to the typical purity of
commercial CO (see Experimental section for further details).
Although an additional deoxy/zeolite purification system was
used for gases prior to use, minute quantities of impurities enter
Experimental
General information
All solution preparations and transfers were carried out under a
purified argon (99.9995%, Saxol, Singapore) atmosphere using
standard Schlenk techniques.73 The argon was further purified
before use by passing it through a deoxy and zeolite column.
Purified carbon monoxide (Research grade, 99.97%, Linde,
Singapore) and helium (99.9995%, Saxol, Singapore) were also
further purified through a deoxy and zeolite columns before
they were used in the experiments. Purified nitrogen (99.9995%,
Saxol, Singapore) was used to purge the Perkin-Elmer FT-IR
spectrometer system.
[Rh4(CO)9(CO)3] (98%) was purchased from Strem Chemicals
(Newburyport, MA, USA) and was used as obtained. The Puriss
quality n-hexane (99.6%, Fluka AG) was distilled from sodium–
potassium under argon for ca. 5 hours to remove the trace water
and oxygen.
Apparatus
In situ studies were performed in a 1.5 l stainless steel (SS316)
autoclave (Pmax = 22.5 MPa, Buchi-Uster, Switzerland) which
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5
1 9 6 3

View Article Online
was connected to a high pressure infrared cell. The autoclave was
equipped with a packed magnetic stirrer with six-bladed turbines
in both the gas and liquid phases (Autoclave Engineers, Erie, PA)
and had a mantle for heating/cooling. The liquid-phase reaction
mixtures were circulated from the autoclave to-and-from the
high pressure IR cell with a high membrane pump (Model DMK
30, Orlita AG, Geissen, Germany) with a maximum rating of
32.5 MPa and a 3 l h−1 flow rate via jacketed 1/8-inch (SS316)
high pressure tubing (Autoclave Engineers).
A polyscience cryostat Model 9505 was used to keep the entire
system, autoclave, transfer lines and infrared cell, isothermal.
Temperatures were measured at the cryostat, autoclave and IR
cell with PT-100 thermoresisters. The necessary connections
to vacuum and gases were made with 1/4 inch (SS316) high-
pressure tubing (Autoclave Engineers) and 30 MPa manometers
were used for pressure measurements (Keller AG, Winter,
Switzerland). The entire system is gas-tight under vacuum as
well as at 20.0 MPa.
The high-pressure infrared cell was constructed at ETH-
Zurich using SS316 steel and could be heated and cooled.
The CaF2 single crystal windows (Korth Monokristalle, Kiel,
Germany) have dimensions of 40 mm diameter by 15 mm
thickness. Two sets of Viton and silicone gaskets provide the
necessary sealing, and Teflon spacers are used between the
windows.74 This high-pressure infrared cell is situated in a
Perkin-Elmer SPECTRUM 2000 FT-IR spectrometer. The
spectral resolution used in this study was 1 cm−1 at an interval
of 0.2 cm−1 for the range 1000–2500 cm−1. A schematic diagram
of the experimental setup can be found in ref .75
288 K, 283 K, 278 K, 273 K and 268 K. A total of 337 spectra
were collected. The additional two helium semibatches were
performed at temperatures of 293 K and 275 K and 138 spectra
were collected.
Acknowledgements
We would like to thank Prof. Brian Heaton (Liverpool, UK) and
Prof. John Shapley (University of Illinois, USA) for discussions
and Prof. Brian Johnson (Cambridge University, UK) for
correspondence. We also would like to thank Assoc. Prof.
Fan Wai Yip and Chong Thiam Seong at the Department of
Chemistry at NUS for providing the IR spectra of Ir4(CO)12.
References
1 F. A. Cotton, Inorg. Chem., 1966, 5, 1083.
2 F. A. Cotton, L. Kruczynski, J. Lewis, B. L. Shapiro and L. F.
Johnson, J. Am. Chem. Soc., 1972, 94, 6191.
3 J. Evans, B. F. G. Johnson, J. Lewis, J. R. Norton and F. A. Cotton,
J. Chem. Soc., Chem. Commun., 1973, 807.
4 E. Band and E. L. Muetterties, Chem. Rev., 1978, 78, 636.
5 Y. V. Roberts, B. F. Johnson and R. E. Benfield, Inorg. Chim. Acta.,
1995, 229, 221.
6 T. Eguchi and B. Heaton, J. Chem. Soc., Dalton Trans., 1999, 3523.
7 F. A. Cotton, Inorg. Chem., 2002, 41, 643.
8 D. L. Kepert and S. C. Williams, J. Organomet. Chem., 1981, 217,
235.
9 B. F. G. Johnson, J. Chem. Soc., Chem. Commun., 1976, 211.
10 J. Evans, B. F. G. Johnson, J. Lewis, T. E. Matheson and J. R. Norton,
J. Chem. Soc., Dalton Trans., 1978, 626.
11 P. Corradini, J. Chem. Phys., 1959, 31, 1676.
12 P. Corradini and A. Sirigu, Ric. Sci., 1966, 36, 188.
13 C. H. Wei and L. F. Dahl, J. Am. Chem. Soc., 1966, 88, 1821.
14 C. H. Wei, Inorg. Chem., 1969, 8, 2384.
In situ spectroscopic studies
15 F. H. Carre, F. A. Cotton and B. A. Frenz, Inorg. Chem., 1976, 15,
380.
As mentioned earlier, the experimental study consisted of seven
isothermal semibatch reactions, five under CO and two under
helium gas. All of these semibatches were conducted in a
similar manner. Thus, only the detailed description of the first
isothermal batch, temperature 288 K under CO, is shown in
Table 4. The total mass of [Rh4(CO)12] present in any semibatch
experiment was 32–102 mg.
Each of the seven semibatches was initiated as follows. First,
background spectra of the IR sample chamber were recorded.
Then 100 ml n-hexane was transferred under argon to the
autoclave. Under 0.5 MPa of gas, CO or He, infrared spectra of
the n-hexane in the high-pressure cell were recorded. The total
system pressure was raised to 1.0 MPa, and the stirrer and high-
pressure membrane pump were started. After equilibration,
infrared spectra of the gas/n-hexane solution in the high-
pressure cell were recorded.
16 L. J. Farrugia, D. Braga and F. Grepioni, J. Organomet. Chem., 1999,
573, 60.
17 G. Bor, Spectrochim. Acta, 1963, 19, 1209.
18 G. Bor, G. Sbrignadello and K. Noak, Helv. Chim. Acta, 1975, 58,
815.
19 S. Aime, R. Gobetto, D. Osella, L. Milone, G. E. Hawkes and E. W.
Randall, J. Magn. Reson., 1985, 65, 308.
20 S. Aime, D. Osella, L. Milone, G. E. Hawkes and E. W. Randal,
J. Am. Chem. Soc., 1981, 103, 5920.
21 E. A. C. Lucken, K. Noack and D. F. Williams, J. Chem. Soc. A,
1967, 148.
22 H. Haas and R. K. Sheline, J. Inorg. Nucl. Chem., 1967, 29, 693.
23 M. A. Cohen, D. R. Kidd and T. L. Brown, J. Am. Chem. Soc., 1975,
97, 4408.
24 S. Aime, L. Milone, D. Osella and A. Poli, Inorg. Chim. Acta, 1978,
30, 45.
25 J. Evans, B. F. G. Johnson, J. Lewis and T. W. Matheson, J. Am.
Chem. Soc., 1975, 97, 1245.
At the end of the initial steps, a solution of ca. 50 mg
[Rh4(CO)12] dissolved in 50 ml n-hexane was prepared, trans-
ferred to the high-pressure reservoir under argon, pressurized
and then added to the autoclave. After equilibration, infrared
spectra of the [Rh4(CO)12]/dissolved gas/n-hexane solution in
the high-pressure cell were recorded.
26 S. Aime and L. Milone, Prog. Nucl. Magn. Reson. Spectrosc., 1977,
11, 183.
27 M. R. Churchill and J. P. Hutchinson, J. Inorg. Chem., 1978, 12, 3528.
28 G. R. Wilkes, Ph.D. Thesis, University of Wisconsin, Madison, WI,
1965.
29 G. F. Stuntz, Ph. D. Thesis, University of Illinois, Urbana-
Champaign, IL, 1978.
Each semibatch reaction consisted of six perturbations and
the duration of each perturbation was ca. two hours. (I) The
gas pressure was raised to 1.5 and then (II) to 2 MPa. (III) An
additional solution of ca. 50 mg [Rh4(CO)12] dissolved in 50 ml
n-hexane was added to the autoclave. The pressure was then
raised to (IV) 3.0, (V) 5.0 and (VI) 7.0 MPa CO.
Spectra were recorded at 10 min intervals in the range 1000–
2500 cm−1. Each semibatch lasted 14 hours and ca. 70 spectra
were collected. For the five CO semibatches, temperatures were
30 F. Cariati, V. Valenti and G. Zerbi, Inorg. Chim. Acta, 1969, 3, 378.
31 W. Heiber and H. Legally, Z. Anorg. Allg. Chem., 1943, 96, 251.
32 P. Chini and S. Martinengo, Chem. Commun., 1968, 5, 251.
33 P. Chini and S. Martinengo, Inorg. Chim. Acta, 1969, 3, 315.
34 P. Chini and B. T. Heaton, Top. Curr. Chem., 1977, 71, 1.
35 C. H. Wei, G. R. Wilkes and L. F. Dahl, J. Am. Chem. Soc., 1967, 89,
4792.
36 L. J. Farrugia, J. Cluster Sci., 2000, 11, 39.
37 J. A. Creighton and B. T. Heaton, J. Chem. Soc., Dalton Trans., 1981,
1972.
38 B. T. Heaton, L. Strona, R. Della Pergola, L. Garlaschelli, U.
Sartorelli and I. H. Sadler, J. Chem. Soc., Dalton Trans., 1983, 173.
39 L. M. Jackman and F. A. Cotton, Dynamic Nuclear Magnetic
Resonance, Academic Press, New York, NY, 1975.
40 R. E. Benfield and B. F. G. Johnson, J. Chem. Soc., Dalton Trans.,
1978, 1554.
Table 4 Experiment design using CO at 288 K
CO/MPa
1.0
52.5 52.5 52.5 96.5 96.5 96.5 96.5
200 200 200 250 250 250 250
1.5
2.0
2.0
3.0
5.0
7.0
Rh4(CO)12/mg
Hexane/ml
41 J. W. Lauher, J. Am. Chem. Soc., 1986, 108, 1521.
42 R. Whyman, J. Chem. Soc., Dalton Trans., 1972, 1375.
1 9 6 4
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5

View Article Online
43 F. Oldani and G. Bor, J. Organomet. Chem., 1983, 246, 309.
44 J. Vidal and W. E. Walker, Inorg. Chem., 1981, 20, 249.
45 G. A. Ozin and A. J. Hanlan, Inorg. Chem., 1979, 18, 2091.
46 A. J. Hanlan and G. A. Ozin, J. Am. Chem. Soc., 1974, 20, 6324.
47 B. T. Heaton and L. Longhetti, J. Organomet. Chem., 1981, 213, 333.
48 P. J. Chini, Chem. Commun., 1967, 440.
49 W. A. Weber and B. C. Gates, J. Phys. Chem. B, 1997, 101, 10423.
50 W. Chew, E. Widjaja and M. Garland, Organometallics, 2002, 21,
1882.
62 P. S. Braterman, Metal Carbonyl Spectra, Academic Press, London,
1975, pp. 153–15.
63 L. J. Farrugia, J. Chem. Soc., Dalton Trans., 1997, 1783.
64 A. McCellan, Tables of Experimental Dipole Moments, W. H. Free-
man, San Francisco, CA, 1963, vol. 1, pp. 28, 220.
65 J. M. Prausnitz, R. N. Lichtenthaler and E. G. Azevedo, Molecular
Thermodynamics of Fluid-Phase Equilibria, Prentice Hall, New
Jersey, 3rd edn., 1999, p. 62.
66 D. T. Brown, T. E. Eguchi, B. T. Heaton, J. A. Iggo and R. Whyman,
J. Chem. Soc., Dalton Trans., 1991, 677.
67 K. Noack, Spectrochim. Acta, 1963, 19, 1925.
68 K. Noack, Helv. Chim. Acta, 1964, 47, 1064.
69 R. L. Sweany and T. L. Brown, Inorg. Chem., 1977, 16, 415.
70 M. Garland and G. Bor, Inorg. Chem., 1989, 28, 410.
71 D. V. Krupenya, E. O. Danilov, M. A. J. Rodgers and S. P. Tunik,
J. Phys. Chem. A, 2003, 107, 8867.
72 G. Bor, U. K. Dietler and P. Pino, J. Organomet. Chem., 1978, 154,
301.
51 E. Widjaja, C. Li and M. Garland, Organometallics, 2002, 21, 1991.
52 C. Li, E. Widjaja, W. Chew and M. Garland, Angew. Chem., Int. Ed.,
2002, 20, 3785.
53 G. Bouquet and M. Bigorna, Spectrochim. Acta, Part A, 1971, 27,
139.
54 M. Bigorna, J. Organomet. Chem., 1970, 24, 211.
55 W. Chew, E. Widjaja and M. Garland, Organometallics, 2002, 21,
1982.
56 W. Beck and K. Lottes, Chem. Ber., 1961, 94, 2578.
57 E. R. Corey and L. F. Dahl, J. Am. Chem. Soc., 1963, 85, 1203.
58 E. Widjaja, C. Li and M. Garland, J. Catal., 2004, 223, 278.
59 M. Garland and P. Pino, Organometallics, 1991, 10, 1693.
60 K. S. Suslick, E. B. Flint and J. A. Jensen, J. Chem. Educ., 1987, 64,
547.
73 D. F. Shriver and M. A. Drezdzon, The Manipulation of Air-Sensitive
Compounds, Wiley, New York, 1986.
74 R. Whyman, in Laboratory Methods in Vibrational Spectroscopy, ed.
H. A. Willis, J. H. van der Maas and R. G. J. Miller, Wiley, New
York, 3rd edn., 1987, ch. 12.
61 J. C. Owrutsky and A. P. Barnovski, J. Chem. Phys., 1996, 105, 9864.
75 J. Feng and M. Garland, Organometallics, 1999, 18, 417.
D a l t o n T r a n s . , 2 0 0 5 , 1 9 5 7 – 1 9 6 5
1 9 6 5