The competitive and non-competitive hydroformylation of conjugated dienes starting with tetrarhodium dodecacarbonyl. An in-situ high-pressure infrared spectroscopic study

Article abstract of DOI:10.1016/S0022-328X(00)00369-7

It is well known that the liquid-phase homogeneous unmodified rhodium catalysed hydroformylation of alkenes is poisoned by the presence of trace quantities of conjugated dienes. Nevertheless, some hydroformylation of conjugated dienes is possible with unmodified rhodium, and this reaction is in general slower than alkene hydroformylations at comparable reaction conditions. In the present contribution, we examined (A) the catalytic behaviour of alkenes in the presence of trace conjugated diene impurities and (B) the catalytic behaviour of a variety of dienes using Rh₄(CO)₁₂ in n-hexane solvent at 293 K under 1.0-4.0 MPa CO and 0.5-2.0 MPa H₂. The analytic method was in-situ high-pressure infrared spectroscopy. It was observed that (I) in the hydroformylation of poisoned alkenes, most of the rhodium reacts with the trace quantity of conjugated dienes and not the alkenes in this competitive situation and (II) the metal carbonyl spectra of the hydroformylation of a variety of dienes are very similar. The primary absorbance maxima observed in the hydroformylations of conjugated dienes occur at circa 2109, 2091, 2087, 2064, 2049, 2037, 2030, 2020, 2012, 1999, and 1990 cm^-1. Given the known chemistry Of Rh₄(CO)₁₂ under syngas, and the very well documented chemistry of Rh₄(CO)₁₂ under alkene hydroformylation conditions, the lack of bridging carbonyls in the present experiments strongly suggested that the new infrared vibrations are due to mononuclear rhodium species. Preliminary analysis suggests the presence of at least three new species. In particular, the formation of observable η³ allyl rhodium tricarbonyl species, σ allyl rhodium tetracarbonyl species and even acyl rhodium tetracarbonyl species RCORh(CO)₄ (R = alkenyl and/or formylalkyl) seems probable. Characteristic wavenumbers of 2108, 2064, 2037, 2020 and 1700 cm^-1 are tentatively assigned to the latter. The reduced hydroformylation activity in the competitive hydroformylation of alkenes arises due to the much higher affinity of rhodium complexes for conjugated dienes than for alkenes under otherwise similar reaction conditions.

Full text of DOI:10.1016/S0022-328X(00)00369-7

www.elsevier.nl/locate/jorganchem
Journal of Organometallic Chemistry 608 (2000) 76–85
The competitive and non-competitive hydroformylation of
conjugated dienes starting with tetrarhodium dodecacarbonyl. An
in-situ high-pressure infrared spectroscopic studyꢀ
Guowei Liu a, Marc Garland b,*
a Laboratorium fu¨r Technische Chemie, Uni6ersita¨tstrasse 6, ETH-Zu¨rich, Eidgeno¨ssische Technische Hochschule, CH-8092 Zurich, Switzerland
b Department of Chemical and En6ironmental Engineering, National Uni6ersity of Singapore, Engineering Dri6e 4, Singapore 117576, Singapore
Received 13 January 2000; received in revised form 28 April 2000; accepted 20 May 2000
Abstract
It is well known that the liquid-phase homogeneous unmodified rhodium catalysed hydroformylation of alkenes is poisoned by
the presence of trace quantities of conjugated dienes. Nevertheless, some hydroformylation of conjugated dienes is possible with
unmodified rhodium, and this reaction is in general slower than alkene hydroformylations at comparable reaction conditions. In
the present contribution, we examined (A) the catalytic behaviour of alkenes in the presence of trace conjugated diene impurities
and (B) the catalytic behaviour of a variety of dienes using Rh4(CO)12 in n-hexane solvent at 293 K under 1.0–4.0 MPa CO and
0.5–2.0 MPa H2. The analytic method was in-situ high-pressure infrared spectroscopy. It was observed that (I) in the
hydroformylation of poisoned alkenes, most of the rhodium reacts with the trace quantity of conjugated dienes and not the
alkenes in this competitive situation and (II) the metal carbonyl spectra of the hydroformylation of a variety of dienes are very
similar. The primary absorbance maxima observed in the hydroformylations of conjugated dienes occur at circa 2109, 2091, 2087,
2064, 2049, 2037, 2030, 2020, 2012, 1999, and 1990 cm−1. Given the known chemistry of Rh4(CO)12 under syngas, and the very
well documented chemistry of Rh4(CO)12 under alkene hydroformylation conditions, the lack of bridging carbonyls in the present
experiments strongly suggested that the new infrared vibrations are due to mononuclear rhodium species. Preliminary analysis
suggests the presence of at least three new species. In particular, the formation of observable h3 allyl rhodium tricarbonyl species,
s allyl rhodium tetracarbonyl species and even acyl rhodium tetracarbonyl species RCORh(CO)4 (R=alkenyl and/or formylalkyl)
seems probable. Characteristic wavenumbers of 2108, 2064, 2037, 2020 and 1700 cm−1 are tentatively assigned to the latter. The
reduced hydroformylation activity in the competitive hydroformylation of alkenes arises due to the much higher affinity of
rhodium complexes for conjugated dienes than for alkenes under otherwise similar reaction conditions. © 2000 Elsevier Science
S.A. All rights reserved.
Keywords: Tetrarhodium dodecacarbonyl; Hydroformylation; Dienes; Deactivation; Rhodium complexes
formylation [3], the chemistry associated with the trans-
formation of rhodium carbonyls [4], and in-situ
1. Introduction
spectroscopic/kinetic studies of alkene hydroformyla-
tions and the identification of RCORh(CO)4 [5], have
The unmodified rhodium catalysed hydroformylation
reaction has been known for almost 50 years [1], and a
added considerably to our knowledge of this reaction.
few early reviews contain much of the knowledge con-
However, a few large gaps in the understanding of the
cerning reactivity and substrate dependences [2]. In
hydroformylation reaction still exist. For example, it is
addition, a host of early kinetic work on alkene hydro-
widely known that the unmodified rhodium catalysed
hydroformylation is very sensitive to a number of im-
ꢀ Primary experimental research was performed at ETH-Zurich
purities including trace quantities of alkynes and dienes
1992–1995. Analysis was performed at the National University of
[6]. Recently, in-situ spectroscopic studies showed that
Singapore 1998.
in the presence of trace alkynes, Rh4(CO)12 is quantita-
tively transformed to dinuclear alkyne rhodium car-
* Corresponding author. Fax: +65-779-1936.
E-mail address: chemvg@nus.edu.sg (M. Garland).
0022-328X/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S0022-328X(00)00369-7

G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
77
the presence of Rh4(CO)12 in n-hexane as solvent at
293–298 K at circa 1.0–4.0 MPa CO and 0.5–2.0 MPa
H2: in both a non-competitive situation and competitive
situation with alkenes. Both non-cyclic dienes such as
2-methyl-1,3-butadiene and 2,3-dimethyl-1,3-butadiene
and cyclic dienes such as 1,3-cyclohexadiene and 1,3-cy-
clooctadiene were used. The primary objective of the
study was to obtain in-situ IR spectroscopic evidence
for the existence of new rhodium complex(es) under
hydroformylation conditions when dienes are present.
The second objective was to understand better the role
of dienes in the poisoning of alkene hydroformylations.
The present contribution attempts to expand on the
very limited body of knowledge concerning the homo-
geneous catalytic hydroformylation of dienes in the
presence of unmodified rhodium carbonyls.
bonyl complexes under hydroformylation conditions,
thereby preventing the availability of rhodium for
alkene hydroformylation [7]. In the present contribu-
tion, we extend our investigation to conjugated dienes.
The homogeneous hydroformylations of conjugated
dienes were first extensively investigated using soluble
cobalt complexes by Natta [8] and Adkins [9]. Later
stoichiometric studies showed that cobalt carbonyl hy-
dride readily attacked conjugated dienes, leading to the
formation of h3 allyl cobalt tricarbonyl species as well
as s allyl cobalt tetracarbonyl species under stoichio-
metric conditions [10–13]. In-situ IR spectroscopic
studies of 1,3-butadiene were undertaken by Mirbach
and the corresponding (h3-C4H7)Co(CO)3 was observed
[14]. Other direct in-situ spectroscopic support for h3
allyl cobalt species comes from King and Tanaka who
prepared (h3-C3H5)Co(CO)3 and observed its intercon-
version to CH2ꢀCH–CH2–Co(CO)4 [15]. Together,
these studies lend strong support for the presence of
both h3 allyl cobalt species as well as s allyl cobalt
species under catalytic conditions [16]. It can be noted
that vibrational and NMR studies have been conducted
on (h3-C4H7)Co(CO)3 [17,18].
The cobalt catalysed hydroformylation of conjugated
dienes normally leads to the formation of saturated
mono-aldehydes and not the twice hydroformylated
products [2b,19]. A variety of conjugated dienes have
been hydroformylated, including 1,3-butadiene, iso-
prene, 2,3-dimethylbutadiene, 3-methyl-1,3-pentadiene,
cyclopentadiene, vinylcyclohexene, and 1,5-cyclooctadi-
ene [9,20,21].
In the case of rhodium catalysed hydroformylation of
dienes, numerous substrates have been studied with [22]
and without phosphine ligands present [23]. The un-
modified rhodium catalysed reaction exhibits the same
primary characteristics as the cobalt catalysed hydro-
formylation of dienes, namely, a strong tendency for
the production of the saturated monoaldehydes and not
the twice hydroformylated products [2b]. However, the
addition of phosphine ligands substantially changes the
reaction characteristics and leads to twice hydroformy-
lated products.
2. Experimental
2.1. General information
All solution preparations were carried out under
argon (99.999% Pan Gas AG, Luzern, Switzerland or
99.999% Saxol Singapore) using standard Schlenk tech-
niques [24]. The argon was further purified prior to use
by passage through a column containing 100 g reduced
BTS-catalyst (Fluka AG Buchs, Switzerland) and 100 g
,
of 4 A molecular sieve to adsorb trace oxygen and
water respectively. All reactions were carried out under
carbon monoxide (99.997% Messner Griesheim GmbH,
Germany) and hydrogen (99.999% Pan Gas AG,
Luzern, Switzerland, or 99.999% Saxol, Singapore) af-
ter further purification through de-oxy and zeolite
columns.
The precious metal complex Rh4(CO)12, with stated
purity of 98% min, was obtained from Strem Chemicals
SA (Bischheim, France or Newbury Port, MA) and was
used without further purification, although trace quan-
tities of the high nuclearity cluster Rh6(CO)16 are virtu-
ally always present. The complex Rh4(CO)12 is known
to be oxygen, water and light sensitive [25]. The n-hex-
ane solvent (stated purity \99.6%, Fluka AG) was
refluxed over sodium potassium alloy under argon.
Weights were measured with a precision of 90.1 mg.
Volumes were measured with a precision of 90.045 ml.
Further microanalytic techniques were not employed
[26]. All conjugated dienes used in this study were of
the highest quality commercially obtainable, usually
99.0+% (Fluka AG Switzerland, Merck Germany) and
the cis oct-4-ene was obtained from ChemSampCo
Ohio. Concerning further purification, the conjugated
dienes were simply degassed before use and the cis
oct-4-ene was dried over CaH2 under nitrogen.
The problem of organic product selectivity in the
rhodium catalysed hydroformylation of conjugated di-
enes is illustrated by Eq. (1) where 1,3-cyclohexadiene is
the substrate.
(1)
In the present contribution we investigate the hydro-
formylation activity of a variety of conjugated dienes in

78
G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
2.2. Equipment
sure cell were recorded. The total system pressure was
raised to 1.0–4.0 MPa CO, and the stirrer and high-
pressure membrane pump were started. After vapor–
liquid equilibration, IR spectra of the alkene–
diene–n-hexane–CO solution in the high-pressure cell
were recorded. A solution of 100 mg Rh4(CO)12 dis-
solved in 100 ml n-hexane was prepared, transferred to
the high-pressure reservoir under argon, pressured with
CO and then added to the autoclave. IR spectra of the
alkene–diene–n-hexane–CO–Rh4(CO)12 solution in
the high pressure cell were recorded. 0.5–2.0 MPa
hydrogen was then added. Spectra were recorded at 30
min intervals in the range 1600–2200 cm−1. A consid-
erable number of spectral subtractions were performed
on each reaction spectrum in order to subtract the
absorbance of the n-hexane solvent, dissolved CO,
alkene, diene, and the absorbance of the Rh4(CO)12.
At 2.0 MPa partial pressures, the solubility of CO
under these reaction conditions was approximately
0.033 mol fraction, and the solubility of H2 was ap-
proximately 0.018 mol fraction [9]. Consequently, a
typical experiment used circa 1.51 mol n-hexane, 0.05
mol CO and 0.03 mol H2. Since the amounts of
Rh4(CO)12 used in the experiments were circa 100 mg,
and since the concentration of alkenes/dienes was low,
the corresponding in-situ measured initial concentra-
tions of Rh4(CO)12 in the experiments were circa 8×
10−5 mol fraction.
In-situ spectroscopic studies were performed in a 1.5
l stainless steel (SS316) autoclave (Bu¨chi-Uster, Switzer-
land) which was connected to a high-pressure IR cell.
The autoclave (Pmax=22.5 MPa) was equipped with a
packed magnetic stirrer with six-bladed turbines in both
the gas and liquid phases (Autoclave Engineers, Erie,
PA) and was constructed with a heating/cooling
mantle. A high-pressure 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, was used to
circulate the n-hexane solutions from the autoclave to
the high pressure IR cell and back to the autoclave via
jacketed 1/8 inch (SS316) high-pressure tubing (Auto-
clave Engineers). The entire system, consisting of auto-
clave, pump, transfer lines and IR cell, was cooled
using a Lauda RX20 cryostat or a Polyscience cryostat
model 9505 and could be maintained isothermal (DT:
0.5°C) at 298–313°C. Temperature measurements were
made at the cryostat, autoclave and IR cell with PT-100
thermoresistors. The necessary connections to vacuum
and gases were made with 1/4 inch (SS316) high-pres-
sure tubing (Autoclave Engineers) and 1.0, 5.0, 10.0
piezocrystals were used for pressure measurements
(Keller AG Winterthur, Switzerland). The entire system
was gas tight under vacuum as well as at 20.0 MPa, the
maximum operating pressure.
The high-pressure IR cell was constructed at the
ETH-Zu¨rich of SS316 steel and could be heated and
cooled. The CaF2 single crystal windows (Korth
Monokristalle, Kiel, Germany) had dimensions of 40
mm diameter by 15 mm thickness. Two sets of Viton
and silicone gaskets provided sealing, and Teflon spac-
ers were used between the windows. The construction
of the flow through cell [27] is a variation on a design
due to Noack [28], and differs in some respects from
other high-pressure IR cells described in the literature
(for a review, see Whyman [29]). The high-pressure cell
was situated in a Perkin–Elmer 983 IR spectrometer.
The resolution was set to 4 cm−1 for all spectroscopic
measurements. A schematic diagram of the experimen-
tal setup can be found in ref. [11]. In the competitive
experiments with cis oct-4-ene poisoned by dienes, a
Perkin–Elmer system 2000 FTIR spectrometer was
used.
3. Results
3.1. Various dienes
In this comparative study, 1,3-cyclooctadiene, 1,3-cy-
clohexadiene, 2-methyl-1,3-butadiene and 2,3-dimethyl-
1,3-butadiene were used. These dienes were used under
a wide range of partial pressures in order to identify
reaction conditions leading to the highest conversion of
Rh4(CO)12, and/or the most interesting metal carbonyl
spectra. After a few hours of reaction time, and in most
cases, significant conversions of Rh4(CO)12 were
achieved.
Fig. 1 shows representative spectra for the hydro-
formylation of the above mentioned substrates. First, it
can be noted that the characteristic wavenumbers for
the metal carbonyl spectra are rather similar, although
the relative magnitudes are not. Secondly, the appear-
ance of aldehyde is clearly shown at circa 1730 cm−1 in
the spectra of 1,3-cyclooctadiene, 1,3-cyclohexadiene
and 2,3-dimethyl-1,3-butadiene. Thirdly, only 2-methyl-
1,3-butadiene shows little if any hydroformylation
product. (The reason for this anomalous lack of activity
is not clear.) However, close inspection shows a small
vibration at circa 1703 cm−1 which is probably due to
the presence of a mononuclear rhodium acyl complex
2.3. In-situ spectroscopic studies
All experiments were performed in a similar manner.
First, single beam background spectra of the IR sample
chamber were recorded. Then, either (A) 10 ml of cis
oct-4-ene or (B) 1–10 ml diene, was dissolved in 150 ml
n-hexane and this solution was transferred under argon
to the autoclave. Under 0.2 MPa CO pressure, IR
spectra of the alkene–diene solution in the high-pres-

G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
79
Fig. 1. In-situ high-pressure IR spectra of the reaction of Rh4(CO)12
under syngas at 293 K and in the presence of conjugated dienes.
Total reaction time is a few hours. Hexane and dissolved CO subtrac-
tion has been performed.
Fig. 2. Deconvolution of 2,3-dimethyl-1,3-butadiene spectrum in the
metal–carbonyl interval. Top curves represent original spectrum and
best fit. Bottom curve represents the deconvolution.
since the wavenumber is consistent with our previous
observations of the acyl moiety in RCORh(CO)4 [5].
Table 1 allows a closer inspection of the metal carbonyl
region. As seen, there is considerable correspondence in
the wavenumber maxima.
experiments, and new spectral characteristics did not
appear.
3.2. Details of 2,3-dimethyl-1,3-butadiene
Other conjugated dienes were used as well at 293 K.
These were 2,3-diphenyl-1,3-butadiene, and 1,4-
diphenyl-1,3-butadiene. In these experiments, significant
conversion of Rh4(CO)12 did not take place during the
Deconvolution of the hydroformylation spectrum of
2,3-dimethyl-1,3-butadiene was performed. This is
shown in Fig. 2.
Table 1
The wavenumbers of the conjugated diene complexesa
Conjugated diene
Wave number (cm−1
)
2,3-Dimethyl-1,3-butadiene
2-Methyl-1,3-butadiene
1,3-Cyclohexadiene
2108.8
2090.8
2091.1
2073.9
2074.5
2074.4
2074.4
2068.4
2066.7
2062.1
2064.6
2064.6
2063.7
2049.7
2051.1
2049.1
2049.4
2087.9
2087.5
1,3-Cyclooctadiene
2108.9
2036.3
2103.0
2091.1
2,3-Dimethyl-1,3-butadiene
2-Methyl-1,3-butadiene
1,3-Cyclohexadiene
2029.2
2029.2
2030.0
2020.4
2020.5
2011.6
2012.9
1998.6
1999.7
1988.4
1991.0
1989.9
1991.7
1965.3
1960.2
2008.3
1,3-Cyclooctadiene
2038.1
2018.5
2012.5
2003.6
a Different fonts represent groups of bands which appear to be related to one and the same species.

80
G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
this would lead to a broad CꢀC vibration, and this is
consistent with the broad feature at 1645.3 cm−1
In the high wavenumber region, more band maxima
.
can be identified. In particular, these are 2108.8, 2090.8,
2073.9, 2068.4, 2062.1, 2049.7, 2036.3, 2029.2, 2020.4,
2011.6, 1998.6 and 1988.4 cm−1. From our previous
in-situ spectroscopic studies, it is possible immediately
to assign the band at 2073.9 cm−1 to the ubiquitous
Rh6(CO)16. Furthermore, the bands at 2108.8, 2062.1,
2036.3 and 2020.4 cm−1 have the correct positions and
intensities for mononuclear acyl rhodium tetracar-
bonyls.
3.3. Different reaction conditions for 2-methyl-1,3-
butadiene
The reaction of 2-methyl-1,3-butadiene with
Rh4(CO)12 was studied at five different partial pressures
of CO and H2. One spectrum from each reaction condi-
tion is shown in Fig. 3. The characteristic wavenumbers
can be found in Table 1.
In the low wavenumber region, the primary spectral
characteristics appear at 1818.4, 1787.2, 1736.4, 1730.5,
1698.3, 1697.0 and a wide band at 1645.3 cm−1. Again,
the band at 1818.4 cm−1 is clearly due to Rh6(CO)16,
and the band at 1787.2 cm−1 is the unsubtracted diene.
This leaves four or five new bands which must be
explained. The bands at 1736.4 and 1730.5 cm−1 sug-
gest two spectroscopically distinct aldehydes, and the
bands at 1698.3 and 1697.0 cm−1 suggest two spectro-
scopically distinct acyl groups. If a conjugated CꢀC/
CꢀO system arises as an intermediate or final product,
Inspection of the spectra indicates that the CO par-
tial pressure plays a significant role, whereas the partial
pressure of H2 hardly affects the spectral characteristics
(and distribution of species). In particular, it seems that
the bands at 2051, 2029 and 1991 cm−1 move together
with approximate intensities of 0.8:0.1:1, and increase
with decreasing CO partial pressure. This suggests that
these bands probably arise from a rhodium tricarbonyl
species. In addition there are three bands at 2000, 2013
and 2065 cm−1 which move together with approximate
intensities of 1:0.9:0.7.
The very weak band seen at 1965 cm−1 in all spectra
is circa 25 cm−1 lower than the prominent vibration at
1991 cm−1. This very weak signal is most likely due to
the isotopomers, i.e. the 1.1% natural abundance of
13CO [30].
Finally, in the interest of completeness, it can be
noted that the spectra of the hydroformylation experi-
ments with the other conjugated dienes also showed
strong CO and H2 dependences, i.e. the distribution of
intermediates was strongly effected.
3.4. Decon6olution of the metal–carbonyl region
The deconvolution of the metal–carbonyl interval
(from the hydroformylation of 2-methyl-1,3-butadiene)
is shown in Fig. 4. The characteristic vibrations after
deconvolution are 2091.1, 2074.5, 2064.6, 2051.1,
2029.2, 2020.5, 2012.9, 2007.1, 1999.7, 1991.0, 1965.3
cm−1. The band at 2020.5 cm−1 was obscured before
deconvolution. The band at 2074.5 cm−1 can be as-
signed to Rh6(CO)16.
3.5. Competiti6e hydroformylation of cis oct-4-ene
Hydroformylation of cis oct-4-ene was performed at
4.0 MPa CO and 2.0 MPa H2. The in-situ high-pressure
IR spectra in the region 2200–1600 cm−1 were not very
similar to other alkene hydroformylations performed in
this laboratory [5]. Instead of observing only four new
metal–carbonyl vibrations at approximately 2108,
2061, 2037 and 2019 cm−1 for a symmetric internal
linear alkene [5h,31], about nine new vibrations ap-
peared. These new vibrations occurred at 2107.3,
2095.3, 2073.8 [Rh6(CO)16], 2067.0, 2050.4, 2039.0,
2022.9, 1990.5 and 1966.5 cm−1. A typical spectrum in
Fig. 3. The reaction of Rh4(CO)12 with 2-methyl-1,3-butadiene under
varying CO and H2 partial pressures. A, 2.0 MPa CO, 0.5 MPa H2.
B, 2.0 MPa CO, 1.0 MPa H2. C, 2.0 MPa CO, 2.0 MPa H2. D, 1.0
MPa CO, 1.0 MPa H2. E, 4.0 MPa CO, 1.0 MPa H2.

G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
81
indicated that there were probably trace quantities of
unconjugated dienes in the ‘cleaned’ cis oct-4-ene which
could not be removed by maleic anhydride treatment,
and which then underwent double-bond migrations un-
der hydroformylation conditions, ultimately to give
conjugated dienes and the characteristic ‘rhodium car-
bonyl/diene’ IR spectra.
4. Discussion and conclusions
4.1. Cluster fragmentation: presence of mononuclear
complexes
The instability of rhodium carbonyl clusters cannot
be overemphasized. Indeed, under CO or syngas,
Rh4(CO)12 readily fragements to observable quantities
of Rh2(CO)8 and Rh6(CO)16 [4]. Furthermore, in the
presence of both syngas and alkenes, mononuclear acyl
rhodium tetracarbonyls are readily formed [5]. These
observations are entirely consistent with the known
Fig. 4. Deconvolution of 2-methyl-1,3-butadiene product in the
metal–carbonyl interval. Top curves represent original spectrum and
best fit. Bottom curve represents the deconvolution.
the high wavenumber region is shown in Fig. 5. It can
be noted that a spectroscopically asymmetric aldehyde
product appeared in the low wavenumber region at
1733.6 cm−1. This is indicative of more than one
aldehyde product. The hydroformylation of very pure
oct-4-ene at these conditions leads to only one aldehyde
[5h].
Since the presence of diene impurity was assumed,
the cis oct-4-ene was analysed over a 10 m capillary
column containing HP series 5, 30m fused silica. In
addition to the cis oct-4-ene elution band, a few small
long retention time bands were also seen. As a similar
problem was previously encountered in the case of
cyclohexene hydroformylations [5d], the cis oct-4-ene
was treated with maleic anhydride under reflux condi-
tions. Periodic GC analysis confirmed that one of the
long retention time elution bands could be eliminated.
The ‘cleaned’ cis oct-4-ene was then used in hydro-
formylation experiments. Although the initial reaction
period proceeded in a normal manner, new bands again
appeared in the spectra at longer reaction times. This
Fig. 5. In-situ high-pressure high-wavenumber IR spectra of the
reaction of Rh4(CO)12 under syngas at 298 K with cis oct-4-ene
poisoned by trace conjugated diene. Typical band positions: 2107.3,
2095.3, 2073.8 [Rh6(CO)16], 2067.0, 2050.4, 2039.0, 2022.9, 1990.5
and 1966.5 cm−1
.

82
G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
Fig. 6. Probable new observable mononuclear rhodium complexes in the hydroformylation of conjugated dienes.
thermochemistry of rhodium. Indeed, the enthalpy of
formation of a rhodium–rhodium bond is only 93 kJ
mol−1 whereas, the enthalpies of rhodium–carbon or
rhodium–carbonyl bonds are in excess of 155 kJ mol−1
[32]. Consequently, cluster fragmentation of Rh4(CO)12
to mononuclear species in the presence of appropriate
organic ligands under catalytic conditions is expected
under a wide range of reaction conditions.
product of the catalysis, the catalytic cycle can be
illustrated as shown in Fig. 7. It should be noted that
the stereochemistry, particularly of the hydride species,
may be different from than that shown.
4.3. Comparison to methylene cyclopropane
hydroformylation
In the present case of the hydroformylation of conju-
gated dienes, cluster fragmentation of Rh4(CO)12 to
lower nuclearity complexes, and specifically new
mononuclear complexes, can be expected. Under syn-
gas, Rh4(CO)12 will undergo cluster opening and/or
fragmentation, and hydrogen activation will occur. The
subsequent reaction of rhodium carbonyl hydrides,
most likely hydrido rhodium tricarbonyl and/or hy-
drido rhodium tetracarbonyl, with conjugated dienes
can be readily assumed.
Recently, we reported the hydroformylations of a
homologous series of methylene cycloalkanes [5h]. Al-
though the high molecular weight members of the series
underwent straightforward hydroformylation, unusual
chemistry occurred when using methylene cyclo-
propane. Instead of obtaining just the corresponding
acyl rhodium tetracarbonyl (R=alkyl) at 2110, 2064,
2038 and 2020 cm−1 [5h], many new bands arose. The
final spectrum after deconvolution gave absorbance
4.2. The non-competiti6e situation
The in-situ IR spectra taken during the hydroformy-
lations of the conjugated dienes indicated the presence
of at least three new rhodium complexes. A preliminary
analysis indicates that the following three sets of vibra-
tions belong together: (A) 2051, 2029 and 1991 cm−1
;
(B) 2000, 2013 and 2065 cm−1; (C) 2108, 2061, 2037
and 2019 cm−1. The latter set clearly indicates the
presence of a mononuclear acyl rhodium tetracarbonyl
species, although it remains unclear whether this ob-
servable acyl complex RCORh(CO)4 is derived from a
R=alkenyl moiety or a R=formylalkyl moiety (or is a
mixture of both). The initial assumption would be that
the complex is an (R=alkenyl) acyl rhodium tetracar-
bonyl, however, the possibility of an (R=formylalkyl)
acyl rhodium tetracarbonyl cannot be ruled out.
Considering the related cobalt carbonyl chemistry as
outlined in Section 1, it is very likely that h3 allyl
rhodium tricarbonyl species, s allyl rhodium tetracar-
bonyl species and (R=alkenyl and/or formylalkyl) acyl
rhodium tetracarbonyl are formed in the present
rhodium carbonyl/diene system in observable quanti-
ties. Fig. 6 illustrates the probable three types of new
observable complexes present in this study.
Fig. 7. Proposed unicyclic homogeneous rhodium catalytic cycle for
the hydroformylation of conjugated dienes under low conversion,
with presence of h3 allyl rhodium tricarbonyl species, s allyl rhodium
tetracarbonyl species and acyl rhodium tetracarbonyl species (R=
alkenyl).
If CO insertion occurs in the s allyl rhodium te-
tracarbonyl, and an unsaturated aldehyde is the final

G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
83
maxima at 2111.9, 2104.1, 2094.2, 2073.8 [Rh6(CO)16)],
2066.9, 2053.6, 2046.6, 2040.9, 2034.0, 2019.7, 2016.0,
Indeed, since these spectra consist of three or more new
species, the determination of the number of observable
species, the number of observable reactions, and the
reconstruction of each of the pure component spectra
would afford considerably more insight into the homo-
geneous catalysis of conjugated dienes. However, as the
spectra indicate, there are subtle changes in the band
positions and subtle changes in the band shapes in-
duced by the pressure and composition differences, and
these problems introduce considerable mathematical
complications.
Recently, we introduced an extensive theory for ob-
taining the number of observable species and the num-
ber of observable reactions in an arbitrary
homogeneous reaction system, from the set of in-situ
IR spectra alone, even in situations where the decadic
absorptivities of the solutes are not constant [33]. This
theory is imbedded in a statistical setting, and permits
the analysis of IR spectroscopic data where band posi-
tions and band shapes change. Once the number of
statistically significant species is known (goal I), recon-
struction of all the individual solute spectra can be
initiated (goal II). In fact, it is possible to reconstruct
highly overlapping solute spectra in situations where
the experimentalist has no a priori information on the
individual components [34,35]. In addition, an efficient
algorithm has been developed to search for all the pure
component spectra in systems involving a large number
of species (with no a priori pure component informa-
tion) [36], and to deal with various related issues.
Finally, given the results of goals I and II, and data on
the initial moles of reactants in the system, it is possible
to reconstruct the reaction stoichiometry of the entire
reaction network (goal III).
2003.9 and 1994.5 cm−1
.
The bands at 2094.2, 2066.9, 2053.6, 2034.0, 2019.7,
2016.0, 2003.9, and 1994.5 cm−1 are very similar to the
spectral characteristics seen in the present study using
conjugated dienes. As we suggested elsewhere [5h], we
believe that attack at the a carbon atom leads to the
formation of a classic acyl rhodium tetracarbonyl under
hydroformylation conditions, but attack at the b car-
bon atom leads to ring cleavage. Ring cleavage, plus a
hydrogen migration would lead to the conversion of
methylene cyclopropane to 1,3-butadiene. Therefore,
the formation of a conjugated diene/rhodium complex
is certainly reasonable in the case of the hydroformyla-
tion of methylene cyclopropane.
4.4. The competiti6e situation
It was shown in the alkene hydroformylation experi-
ments, that the presence of trace quantities of conju-
gated dienes drastically altered the expected chemistry
and kinetics of the catalytic system. In the case of cis
oct-4-ene tainted by conjugated dienes, in-situ IR spec-
tra clearly showed that the expected acyl rhodium
tetracarbonyl (R=alkyl) vibrations were strongly sup-
pressed and the presence of numerous new vibrations
occurred. Thus trace quantities of conjugated dienes
successfully compete against an excess of alkenes for
the rhodium in the system and consequently altered the
hydroformylations. The affinity of conjugated dienes
for rhodium is dramatically higher than the affinity
exhibited by alkenes.
4.5. Additional remark concerning competiti6e
cyclohexene hydroformylation
Presently, we are in the final stages of completing the
development of the set of algorithms I–III. Once a
sufficient number of tests have been successfully per-
formed on synthetic spectroscopic data, it is our inten-
tion to re-investigate the hydroformylation of
2-methyl-1,3-butadiene. Such results can be expected in
a year or two.
As mentioned, the hydroformylation of cyclohexene
has been studied by this group on multiple occasions
[5d,5g]. The resulting acyl rhodium tetracarbonyl (R=
alkyl) has absorbance maxima at 2109, 2063, 2037 and
2019 cm−1. The presence of conjugated dienes was also
a problem in these experiments. However, under exten-
sive reflux with maleic anhydride, the cyclohexene could
be obtained in a very pure form. The experiments
conducted with trace quantities of 1,3-cyclohexadiene
gave a distinct pattern with new absorbance maxima
2086, 2062, 2046, 2024, 2006, and 1986 cm−1. It ap-
pears that cis oct-4-ene is considerably more difficult to
purify than cyclohexene.
4.7. Chemoselecti6ity and possible caging effects in
dual-cycle systems
In the simplest case, solvent caging is exemplified by
the photodissociation of I2 in solution [37]. Upon UV
irradiation, I2 dissociates to create a caged pair. The
iodine atoms will repeatedly collide, perhaps as many as
1000 times, before they undergo recombination to
molecular iodine or successfully escape from their sol-
vent cavity. In fact, the quantum efficiency for generat-
ing a separated iodine atom pair is very low, due to the
large probability of recombination compared to solvent
cage escape. Recently, we have been concerned with a
more complicated scenario, namely the possible effect
4.6. Chemometric analysis and ad6anced decon6olution
The set of 2-methyl-1,3-butadiene reaction spectra
(Fig. 3), obtained under different reaction conditions,
suggests an unusual opportunity for further analysis.

84
G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
[2] (a) L. Marko, in: R. Ugo (Ed.), Aspects of Homogeneous
Catalysis, vol. II, Reidel, Dordrecht, 1974. (b) P. Pino, F.
Placenti, M. Bianchi, in: I. Wender, P. Pino (Eds.), Organic
Syntheses via Metal Carbonyls, vol. II, Wiley, New York, 1977.
[3] (a) B. Heil, L. Marko, Chem. Ber. 101 (1968) 2209. (b) B. Heil,
L. Marko, Chem. Ber. 102 (1969) 2238. (c) G. Csontos, B. Heil,
L. Marko, Ann. NY Acad. Sci. 239 (1974) 47. (d) S. Toros,
Magy. Kem. Lapja 29 (1974) 543; Chem. Abstr. 83 (1975) 177,
870. (e) B. Heil, L. Marko, G. Bor, Chem. Ber. 104 (1971) 3418.
(f) P. Chini, S. Martinengo, G. Garlaschelli, J. Chem. Soc.
Chem. Commun. (1972) 709. (g) P. Pino, F. Oldani, G. Con-
siglio, J. Organomet. Chem. 250 (1983) 491. (h) Y.B. Kagan,
E.V. Slivinskil, V.I. Kurkin, G.A. Korneeva, R.A. Aranovich,
I.G. Fal’kov, N. Rzhevskaya, S.M. Loktev, Neftekhimiya 25
(1985) 791; Chem. Abstr. 104 (1986) 56974d.
[4] (a) J.L. Vidal, W.E. Walker, Inorg. Chem. 20 (1981) 249. (b) R.
Whyman, J. Chem. Soc. Dalton Trans. (1972) 1375. (c) F.
Oldani, G. Bor, J. Organomet. Chem. 246 (1983) 309. (d) P.
Chini, B.T. Heaton, Topics Curr. Chem. 71 (1977) 1. (e) P.
Chini, B.T. Heaton, Inorg. Synth. 20 (1980) 209. (f) P. Chini, J.
Chem. Soc. Chem. Commun. (1967) 440. (g) P. Chini, Inorg.
Synth. 16 (1976) 49.
of solvent caging on selectivity [38], particularly
chemoselectivity in coupled dual-cycle homogeneous
catalytic systems [34b].
Coupled dual-cycle chemoselective systems are quite
common in homogeneous catalysis. These include most,
but not all, catalytic systems capable of performing
both the hydrogenation and hydroformylation of di-
enes. In the present case of diene hydrogenation/hydro-
formylation, two distinct dual-cycle systems A and B
may be involved. It is important to recognize that in
both systems, the products of the first cycle are the also
the initial reactants for the second cycle. Thus, the
cycles are coupled by a material constraint.
Case A. The first cycle involves the reaction of a
rhodium carbonyl hydride with diene to give a rhodium
carbonyl hydride and an alkene as products. The sec-
ond cycle involves the reaction of a rhodium carbonyl
hydride with an alkene to give a rhodium carbonyl
hydride and a saturated aldehyde as products. In other
words, the first cycle involves hydrogenation and the
second involves hydroformylation.
Case B. The first cycle involves the reaction of a
rhodium carbonyl hydride with diene to give a rhodium
carbonyl hydride and an unsaturated aldehyde as prod-
ucts. The second cycle involves the reaction of a
rhodium carbonyl hydride with an unsaturated alde-
hyde to give a rhodium carbonyl hydride and a satu-
rated aldehyde as products. In other words, the first
cycle involves hydroformylation and the second in-
volves hydrogenation.
[5] (a) M. Garland, Dissertation 8585, ETH-Zurich, 1988. (b) M.
Garland, G. Bor, Inorg. Chem. 28 (1989) 410. (c) M. Garland, P.
Pino, Organometallics 10 (1991) 1693. (d) Ch. Fyhr, M. Gar-
land, Organometallics, 12 (1993) 1753. (e) J. Feng, M. Garland,
Organometallics 18 (1999) 417. (f) M. Garland, Organometallics
12 (1993) 535. (g) J. Feng, M. Garland, Organometallics 18
(1999) 1542. (h) G. Liu, R. Volken, M. Garland, Organometal-
lics 18 (1999) 3429.
[6] B. Cornils, in: J. Falbe (Ed.), New Syntheses with Carbon
Monoxide, Springer, Berlin, 1980.
[7] G. Liu, M. Garland, Organometallics 18 (1999) 3457.
[8] G. Natta, Beatti Chim. Ind. (Milano) 27 (1945) 84.
[9] H. Adkins, J.L.R. Williams, J. Org. Chem. 17 (1952) 980.
[10] C.L. Aldridge, H.B. Jonassen, E. Pulkinnen, Chem. Ind. (Lon-
don) (1960) 374.
Since the products for the first cycle are the initial
reactants for the second cycle, for both scenarios A and
B, a solvent cage effect could influence the time-depen-
dent distribution of the organic products. Indeed, if a
strong solvent cage effect exists for scenario A, little
free alkene would be observed. Similarly, if a strong
solvent cage effect exists for scenario B, little free
unsaturated aldehyde would be observed. In fact, there
is rather strong spectroscopic evidence that a solvent
cage effect on chemoselectivity is occurring in the
present experiments. The spectra indicate that there is
little if any free alkene and/or unsaturated aldehyde in
the reaction mixture during the catalysis. It should be
noted that free alkene and/or unsaturated aldehyde
should accumulate if there isn’t as cage effect, since
they are relatively unreactive compared to the conju-
gated dienes.
[11] F. Ungvary, L. Marko, Acta Chim. Hung. 62 (1969) 425.
[12] A.J. Rest, D.L. Taylor, J. Chem. Soc. Dalton. Trans. (1983) 369.
[13] Y.-W. Ge, F. Penig, P.R. Sharp, J. Am. Chem. Soc. 112 (1990)
2632.
[14] M.F. Mirbach, Trans. Met. Chem. 9 (1984) 465.
[15] R.B. King, K. Tanaka, J. Indian Chem. Soc. 109 (1982) 124.
[16] J. Falbe, N. Huppes, Brennstoff-Chem. 47 (1966) 314.
[17] D.C. Andrews, G. Davidson, J. Chem. Soc. Dalton Trans.
(1972) 1381.
[18] D.W. Moore, H.B. Jonassen, T.B. Joyner, A.J. Bertrand, Chem.
Indust. (1960) 1304.
[19] R.L. Pruett, Adv. Organomet. Chem. 17 (1979) 1.
[20] W.F. Gresham, R.E. Brooks, W.M. Bruner, U.S. Patent
2,437,600, 1948.
[21] Inventa, AG, French Patent 1,371,085, 1964.
[22] (a) B. Fell, W. Bull, J. Hagen, Chem. Ztg. 99 (1975) 485. (b) H.
Bahrmann, B. Fell, J. Mol. Catal. (1980) 3329. (c) B. Fell, H.
Bahrmann, J. Mol. Catal. (1977) 210. (d) B. Fell, W. Ruplius,
Tetrahedron Lett. (1969) 2721. (e) A. Spencer, J. Organomet.
Chem. 124 (1977) 85. (f) C. Botteghi, M. Branca, A. Saba, J.
Organomet. Chem. 332 (1987) 229.
[23] (a) M. Monikawa, Bull. Chem. Soc. Jpn. 3 (1964) 79. (b) R.
Kummer, F.J. Weiss, Symposium on Rhodium in Homogeneous
Catalysis, Veszprem, 1978, p. 87.
[24] D.F. Shriver, M.A. Drezdzon, The Manipulation of Air-Sensi-
tive Compounds, Wiley, New York, 1986.
References
[1] (a) G. Schiller (Chem. Verwertungsges. Oberhausen) German
Patent 965,605, 1956; Chem. Abstr. 53 (1959) 11226. (b) V.L.
Hughes (Esso Res. Eng. Co), U.S. Patent 2,880,241, 1959; Chem.
Abstr. 53 (1959) 14938. (c) V.L. Hughes (Esso Res. Eng. Co),
British Patent 801,734, 1958; Chem. Abstr. 53 (1959) 7014.
[25] P. Chini, B.T. Heaton, Topics Curr. Chem. 71 (1977) 1.
[26] (a) N.D. Cheronis, Micro and Semimicro Methods, Interscience,
New York, 1954. (b) S. Maj, V. Horak, Microscale Manipula-
tions in Chemistry, Wiley, New York, 1976. (c) P.L. Kirk,
Quantitative Ultramicro Analysis, Wiley, New York, 1954.

G. Liu, M. Garland / Journal of Organometallic Chemistry 608 (2000) 76–85
85
[27] U.K. Dietler, Dissertation 5428, ETH-Zurich, 1974.
[28] K. Noack, Spectrochim. Acta 24A (1968) 1917.
[29] R. Whyman, in: H.A. Willis, J.H. van der Maas, R.G.J. Miller
(Eds.), Laboratory Methods in Vibrational Spectroscopy, Wiley,
New York, 3rd ed., 1987, Chap. 12.
[30] (a) G. Bor, Inorg. Chim. Acta 1 (1967) 81. (b) G. Bor, Chem.
Commun. (1969) 641. (c) G. Bor, C. Jung, Inorg. Chim. Acta 3
(1969) 69. (d) P.S. Braterman, R.W. Harrill, H.D. Kaesz, J. Am.
Chem. Soc. 89 (1967) 2851. (e) F.A. Cotton, A. Musco, G.
Yagupsky, Inorg. Chem. 6 (1967) 1357. (f) R.W., Harrill, H.D.
Kaesz, J. Am. Chem. Soc. 90 (1968) 1449. (g) H.D. Kaesz, R.
Bau, D. Hendrickson, J.M. Smith, J. Am. Chem. Soc. 90 (1968)
1449. (h) J. Lewis, A.R. Manning, J.R. Miller, J. Chem. Soc. A
(1966) 843. (i) A. Loutellier, M. Bigorgne, Bull. Soc. Chim. Fr.
(1965) 3186. (j) K. Noack, J. Organomet. Chem. 12 (1968) 181.
(k) A. Poletti, R. Catiotti, A. Foffani, Inorg. Chim. Acta 2
(1968) 157.
[33] M. Garland, E. Visser, P. Terweisch, D.W.T. Rippin, Anal.
Chim. Acta 351 (1997) 337.
[34] (a) Y. Zeng, M. Garland, Anal. Chim. Acta 359 (1998) 303. (b)
Y. Zeng, On the implementation of the integral absorbance
method to exploratory chemometric studies, M.Eng. Thesis,
National University of Singapore, 1997.
[35] Y. Pan, L. Susithra, M. Garland, J. Chemometrics 14 (2000) 63.
[36] Y. Pan, The reconstruction of pure component spectra in ex-
ploratory chemometric studies. Entropy minimizations and large
scale problems, M.Eng. Thesis, National University of Singa-
pore, 1999.
[37] (a) J. Franck, E. Rabinowitch, Trans. Faraday Soc. 30 (1934)
120. (b) E. Rabinowitch, W.C. Wood, Trans. Faraday Soc. 32
(1936) 1381. (c) T.J. Chuang, G.W. Hoffmann, K.B. Eisenthal,
Chem. Phys. Lett. 25 (1974) 201. (d) B. Otto, J. Schroeder, J.
Troe, J. Chem. Phys. 81 (1984) 202. (e) Q. Liu, J.K. Wang, A.H.
Zewail, Nature 364 (1993) 427. (f) C. Wan, M. Gupta, J.S.
Baskin, Z.H. Kim, A.H. Zewail, J. Chem. Phys. 106 (1997) 4353.
[38] C.M. Lee, Simulations of the cage effect in homogeneous cata-
lytic systems, B. Eng. Thesis, National University of Singapore,
1998.
[31] G. Liu, The detailed kinetics of unmodified rhodium catalyzed
hydroformylation of cyclooctene, M.Eng. Thesis, National Uni-
versity of Singapore, 2000.
[32] J.A. Conners, Topics Curr. Chem. 71 (1977) 71.
.