Hydrogenation of cinnamaldehyde using catalysts prepared from supported palladium phosphine complexes

Article abstract of DOI:10.1006/jcat.1997.1491

The liquid phase hydrogenation of cinnamaldehyde has been studied using catalysts prepared from the binuclear palladium complexes [Pd₂X₄(PR₃)₂] (X=Cl, Br, or I; R=Me, Et, and Pr_i or Bu for X=Br only) on silica, and the results have been compared with the performance of a standard Pd/SiO₂ catalyst prepared from Pd(NO₃)₂ and with that of the unsupported molecular complexes. When X=Br, cinnamaldehyde is rapidly hydrogenated selectively to hydrocinnamaldehyde and no further reaction occurs. When X=I, the reactions are extremely slow but are again limited to hydrocinnamaldehyde formation. The catalysts prepared from the chloride-containing precursors behave differently, and some phenyl propanol is formed along with hydrocinnamaldehyde. However, the reaction mechanisms are different from that which operates with the silica-supported palladium catalyst. The rates of the reactions are dependent on the nature of the phosphine substituent, but the selectivities are dependent primarily on the halide. Hydrogenation reactions on all of the catalysts prepared from the binuclear complexes proceed only after an induction period during which partial reduction and/or surface reconstruction of the supported complexes occur. This contrasts with the performance of the standard Pd/SiO₂ catalysts for cinnamaldehyde hydrogenation, which proceeds without any induction period. The reactions of all of these catalysts are thought to take place on a hydrocarbonaceous overlayer which is formed rapidly from unsaturated aldehydes and alcohols. Hydrogenations of cinnamaldehyde using the unsupported complexes in solution proceed at much lower rates, but reveal the same selectivities observed with their silica-supported counterparts.

Full text of DOI:10.1006/jcat.1997.1491

JOURNAL OF CATALYSIS 166, 89–97 (1997)
ARTICLE NO. CA971491
Hydrogenation of Cinnamaldehyde Using Catalysts Prepared
from Supported Palladium Phosphine Complexes
ꢀ
Graham R. Cairns, Ronald J. Cross, and Diane Stirling
Chemistry Department, University of Glasgow, Glasgow G12 8QQ, United Kingdom
Received February 2, 1996; revised April 22, 1996; accepted October 14, 1996
[Pd2X4(PR3)2] (10) and several other monophosphine–
palladium derivatives which also have a 1 : 1 P to Pd ratio
(6, 11), bis(phosphine) complexes [PdX2(PR3)2] (10, 12),
and complexes of bidentate (13–15) and terdentate (16,
17) phosphorus ligands. In a number of cases, mixtures
of tertiary phosphines with complexes of palladium(II) or
palladium(0) have been employed as homogeneous hydro-
genation catalysts, and while the active species are almost
certainly Pd–PR3 derivatives, their identities remain unre-
solved (18, 19).
The applicability, activity, and selectivity of these cata-
lysts vary considerably. They are believed to operate via
Pd(H) and Pd(substrate) intermediate complexes, and in a
number of cases IR and NMR spectroscopic evidence for
The liquid phase hydrogenation of cinnamaldehyde has been
studied using catalysts prepared from the binuclear palladium com-
plexes [Pd2X4(PR3)2] (X = Cl, Br, or I; R = Me, Et, and Pri or Bu
for X = Br only) on silica, and the results have been compared
with the performance of a standard Pd/SiO2 catalyst prepared from
Pd(NO3)2 and with that of the unsupported molecular complexes.
When X = Br, cinnamaldehyde is rapidly hydrogenated selectively
to hydrocinnamaldehyde and no further reaction occurs. When
X = I, the reactions are extremely slow but are again limited to
hydrocinnamaldehyde formation. The catalysts prepared from the
chloride-containing precursors behave differently, and some phenyl
propanol is formed along with hydrocinnamaldehyde. However, the
reaction mechanisms are different from that which operates with
the silica-supported palladium catalyst. The rates of the reactions
are dependent on the nature of the phosphine substituent, but the
selectivities are dependent primarily on the halide. Hydrogenation
reactions on all of the catalysts prepared from the binuclear com-
plexes proceed only after an induction period during which partial
reduction and/or surface reconstruction of the supported complexes
occur. This contrasts with the performance of the standard Pd/SiO2
catalysts for cinnamaldehyde hydrogenation, which proceeds with-
out any induction period. The reactions of all of these catalysts are
thought to take place on a hydrocarbonaceous overlayer which is
formed rapidly from unsaturated aldehydes and alcohols. Hydro-
genations of cinnamaldehyde using the unsupported complexes in
solution proceed at much lower rates, but reveal the same selec-
c
–
Pd H intermediates has been reported (6, 9, 16, 17). Inter-
estingly, both addition of N-donors to palladium phosphine
catalysts and addition of phosphines to palladium nitrogen
catalysts cause deactivation (20, 21), suggesting that the
availability of coordination sites is of critical importance
(14). In some cases induction periods have been noted (2),
and the identity of the catalysing species is sometimes ob-
scure (22). In addition, all suffer from the usual drawback
of homogeneous catalysis, that of catalyst separation.
A number of attempts have been made to prepare
heterogeneous catalysts by supporting molecular palla-
dium complexes. The use of organic polymer-based benzyl-
diphenylphosphine (23) or isocyanopropylpolyacrylate
(24) as ligands produced heterogeneous analogues of
[PdCl2(PR3)2] and [PdCl2(CNR)2], respectively. In both
cases the polymer-based materials were more active than
their monomeric analogues as hydrogenation catalysts (the
latter comparingwellwith commercialPd/C catalysts) while
retaining many of the desirable features of their homo-
geneous counterparts, such as high selectivity. Likewise,
catalysts prepared by contacting solutions of chloropalla-
dium amine complexes with alumina, silica, MgO, BaSO4,
CaCO3, or zeolites also led to materials with enhanced ac-
tivity (25), which were quite stable to poisons such as H2O,
PPh3, or thiophene and performed well after repeated use.
Treatment of SiO2/hydrosilane with [Pd(PPh3)4] and acetic
acid produced immobilised catalysts which specifically
tivities observed with their silica-supported counterparts.
ꢁ 1996
Academic Press
INTRODUCTION
Many palladium complexes are known to behave as
homogeneous hydrogenation catalysts. These include pal-
ladium(II) compounds with nitrogen donor atoms (1–
–
7), of which ortho-metallated examples with Pd C bonds
appear most versatile, palladium(II) complexes of S,N-
bidentate ligands (8), and palladium(0) diimine compounds
(9). Complexes with phosphine ligands have also been
extensively examined, including some binuclear species
ꢀ
E-mail: D.Stirling@chem.gla.ac.uk.
89
0021-9517/97 $25.00
c
Copyright ꢁ 1997 by Academic Press
All rights of reproduction in any form reserved.

90
CAIRNS, CROSS, AND STIRLING
reduced alkynes to Z-alkenes (26). A third type of Preparation of the Standard Pd/SiO2 Catalysts
“heterogenised” palladium-based molecular catalyst takes
These catalysts were prepared by impregnating Gasil
advantage of the protection offered by intercalating and
35M silica with sufficient [Pd(NO3)2] in water to give 5%
anchoring the complexes into the layers of pillared clays
Pd loading, removing the solvent after 24 h using a rotary
like montmorillonite (27) or into the pores of zeolites (28):
evaporator, and then pumping at high vacuum as before.
both offer enhanced catalyst stability.
The catalysts were calcined in flowing air (60 ml minꢃ1) at
We describe in this work the activities and selectivities
200ꢂC for 3 h and reduced in H2 (60 ml minꢃ1) at 200ꢂC for
of catalysts prepared from [Pd2X4(PR3)2] (X = Cl, Br, or I;
18 h in the reactor prior to the hydrogenation reactions.
R = Me or Et) and [Pd2Br4(PR30 )2] (R0 = Me, Et, Pri, or Bu)
on silica, using the hydrogenation of cinnamaldehyde as a
model reaction.
RESULTS
Preparation of the Catalysts
EXPERIMENTAL
The catalyst precursors were made by impregnating so-
The palladium complexes were prepared by standard
methods and their identities and purities were checked
lutions of the palladium complexes onto the silica to give a
5% w/w loading of palladium. C, H, halogen, and phospho-
rus microanalyses on these yellow or orange materials indi-
cated that the palladium complexes retained their integrity
on the silica. The analytical values for hydrogen were all
about 0.5% higher than calculated, and these discrepancies
were assigned to the presence of surface silanol groups on
the silica.
The catalyst precursors did not change in appearance
upon the introduction of the solvent or the cinnamalde-
hyde substrate, even at temperatures up to 135ꢂC. (A
separate solution reaction between [Pd2Cl4(PMe3)2] and
cinnamaldehyde showed no evidence for reaction, the pal-
ladium complex being recovered intact). The introduction
of hydrogen gas, however, immediately caused the mate-
rials to turn black, even at temperatures as low as 70ꢂC,
producing the active catalysts.
1
by microanalyses, melting points, H and 31P NMR spec-
troscopy, and IR spectroscopy (29 –36). Gasil 35M silica was
supplied by Crosfield and dried at 100ꢂC for 3 h before use.
Gasil 35M has a surface area of 320 m2 gꢃ1, a pore volume
of 1.2 ml gꢃ1 and a mean particle size of 4.0 ꢀm. Gas chro-
matography (GC) analyses were carried out on a Philips
PU4500 gas chromatograph fitted with a Phase Separations
2 m glass column of 5% carbowax 20M on chromosorb W
(mesh size 80–100 ꢀm) and a flame ionisation detector.
Preparation of the Catalyst Precursors
Precursors with a 5% w/w Pd loading were prepared by
stirring a solution of the palladium complex in ethanol or
ethanol/chloroform with Gasil 35M silica for 24 h and then
removing the solvent with a rotary evaporator at room tem-
perature followed by pumping at high vacuum. The stoi-
chiometry of each of these yellow or orange materials was
checked by C, H, halide, and P microanalyses.
Catalytic Activity
The hydrogenation of cinnamaldehyde can give rise to
three likely products, namely hydrocinnamaldehyde, cin-
namyl alcohol, and phenylpropanol. No cinnamyl alcohol,
Catalyst Activation and Hydrogenation
==
Under a helium atmosphere, 0.5 g of the catalyst pre- PhCH CHCH2OH, was detected as a product of any of
cursor and 25 ml of decahydronaphthalene (decalin) were our experiments. Table 1 lists the results of the liquid phase
treated with 0.25 ml of cinnamaldehyde, and the temper- hydrogenations of cinnamaldehyde at 1 atm pressure and
ature was raised to 135ꢂC. Hydrogen gas was bubbled 135ꢂC by catalysts prepared from the supported molecu-
through the liquid at 1 atm pressure at a rate of 70 ml minꢃ1
.
lar complexes [Pd2X4(PR3)2] (X = Cl, Br, or I; R = Me or
The catalyst rapidly turned black. Aliquots of the reaction Et) and [Pd2Br4(PR30 )2] (R0 = Me, Et, Pri, or Bu). All of the
mixture were removed at intervals, and were subjected to results were reproducible. The performance of the stan-
GC analysis after filtration. Additional experiments were dard Pd/SiO2 catalyst is included in the table for compar-
carried out at 95ꢂC in order to followthe activitiesand selec- ison. The results are presented as a percent conversion of
tivities during the initial stages of the reaction. The hydro- cinnamaldehyde and selectivity to either hydrocinnamalde-
genations of hydrocinnamaldehyde (0.25 ml) and cinnamyl hyde or phenylpropanol. The SiO2 itself and [PMe3H]Br on
alcohol at 135ꢂC were also investigated in a similar manner silica were confirmed to be catalytically inactive under the
using selected catalysts.
same conditions.
The hydrogenation of cinnamaldehyde using the homo-
The nature of the halide had the largest effect on both the
geneous analogues of the binuclear palladium complexes activity and the product selectivity in the hydrogenation of
were carried out under identical conditions to those for the cinnamaldehyde over the catalysts prepared from the sup-
heterogeneous catalysts, using the same weights of com- ported halogen-bridged binuclear compounds. The cata-
plexes.
lysts from the chloro-complexes [Pd2Cl4(PR3)2] (R = Me

HYDROGENATION CATALYSTS FROM PALLADIUM COMPLEXES
91
formed in turn from the hydrogenation of the cinnamalde-
hyde substrate. It is of interest to note that the performance
of a catalyst prepared by impregnation of [Pd2Cl4(PMe3)2]
on SiO2 (5% Pd) followed by calcination at 400ꢂC and re-
duction at 200ꢂC before use was indistinguishable from
that of the standard 5% Pd/SiO2 catalyst, indicating that
under those somewhat extreme conditions effectively all
of the halide and phosphine had been removed from the
palladium.
The catalysts prepared from the bromo-bridged com-
pounds, [Pd2Br4(PR3)2], showed remarkable selectivity.
They all hydrogenated cinnamaldehyde rapidly and com-
pletely to hydrocinnamaldehyde and no further reac-
tion then occurred. The iodide-containing catalysts from
[Pd2I4(PR3)2] (R = Me or Et) were equally selective but
acted very slowly, both of them producing only 6% hy-
drocinnamaldehyde over a 48 h time period under the same
conditions. In an extended cinnamaldehyde hydrogena-
tion reaction carried out with the catalyst prepared from
[Pd2I4(PMe3)2]/SiO2, only 10% of the cinnamaldehyde was
hydrogenated after 200 h, hydrocinnamaldehyde again be-
ing the only product formed. This selectivity indicates that
the halide is retained as an integral part of the catalysts.
The activity differences between bromides and iodides may
simply be steric in origin. As with the chloride catalysts, a
marked deactivation was apparent in these catalysts with
repeated use.
Experiments using a catalyst precursor prepared simply
by grinding together sufficient [Pd2Br4(PMe3)2] and silica
to give a 5% Pd loading also produced an active catalyst
that was 100% selective to hydrocinnamaldehyde. Some-
what counterintuitively, these reactions proceeded faster
than those of the catalysts prepared by prior impregnation
of the complex onto the support. These differences pre-
sumably arise from differences in the particle size of the
active palladium-containing species on the surface, since
microanalyses indicated that the complexes retained their
integrity when impregnated onto the supports.
TABLE 1
Hydrogenation of Cinnamaldehyde at 135ꢂC over Supported
Palladium Catalysts
Product Selectivity
Time on Line Conversion
HCa
PPb
Catalyst Precursor
Pd2Cl4(PMe3)2/SiO2
(h)
(% )
(mol% )
(mol% )
1
2.5
4
29
35
53
79
25
31
42
47
4
4
5
24
32
Pd2Br4(PMe3)2/SiO2
Pd2I4(PMe3)2/SiO2
0.5
1
24
27
100
100
27
100
100
0
0
0
1
2.5
4
1
3
4
6
1
3
4
6
0
0
0
0
24
Pd2Cl4(PEt3)2/SiO2
Pd2Br4(PEt3)2/SiO2
0.5
3
24
100
100
100
29
28
23
71
72
77
0.5
3
24
76
100
100
100
76
100
100
100
0
0
0
0
48
Pd2I4(PEt3)2/SiO2
Pd2Br4(PPri3)2/SiO2
48
6
6
0
0.5
48
100
100
100
100
0
0
Pd2Br4(PBun3)2/SiO2
0.5
1
48
96
100
100
96
100
100
0
0
0
Pd(NO3)2/SiO2
0.25
81
100
100
100
100
100
100
100
100
81
99
98
91
78
71
62
16
0
0
1
2
0.5
0.75
1
9
2.5
3
4
24
48
22
29
38
84
100
a HC = hydrocinnamaldehyde.
b PP = phenylpropanol.
Although the nature of the halide present produced the
largest influence on the behaviour of the catalyst, it was
apparent that the tertiary phosphine ligands also played a
critical role. Thus catalysts prepared by grinding together
crystals of PdBr2 with silica (5% Pd loading) or from the
standard Pd/SiO2 catalyst modified by [NBu4]Br produced
phenylpropanol as well as hydrocinnamaldehyde, although
the amount of phenylpropanol was considerably less than
that from the use of the unmodified Pd/SiO2 catalyst. This
suggests that the presence of phosphine is also important
in promoting the selectivity to hydrocinnamaldehyde, and
possibly that intimate contact of the halide with each of the
Pd atoms prior to catalyst formation and use is important.
The nature of the alkyl groups on the tertiary phos-
phines affected the rates but not the selectivities of the
catalystsprepared from the silica-supported [Pd2Br4(PR3)2]
or Et) hydrogenated cinnamaldehyde to both hydrocin-
namaldehyde and phenylpropanol. Conversion to phenyl-
propanol did not appear to take place sequentially from hy-
drocinnamaldehyde, however, since even after 24 h some
hydrocinnamaldehyde remained (Table 1). This conclusion
was supported by independent experiments on hydrocin-
namaldehyde itself. Neither catalyst reduced it to phenyl-
propanol, even after prolonged exposure under these con-
ditions. The catalysts slowly deactivated during use.
This behaviour can be contrasted with that of a standard
5% Pd/SiO2 catalyst prepared by calcination and reduction
of supported [Pd(NO3)2]. Phenylpropanol was formed in
100% yield by hydrogenation of the hydrocinnamaldehyde,

92
CAIRNS, CROSS, AND STIRLING
(Table 1). This may be predominantly a steric effect, since it pense of the hydrocinnamaldehyde. Separate experiments
wasobserved that the rate ofconversion ofcinnamaldehyde confirmed that there was no reaction between hydrocin-
to hydrocinnamaldehyde increased with increasing cone- namaldehyde and hydrogen over the catalyst prepared
angle size of the phosphine (37). Differences in the elec- from [Pd2Cl4(PMe3)2]/SiO2. Interestingly, another separate
tronic effects of this selection of phosphines are small (37). experiment with this catalyst showed that cinnamyl alcohol
Thus, after the hydrogenation reactions had proceeded for was partly hydrogenated to phenylpropanol and partly iso-
30 min, the yields of hydrocinnamaldehyde were 27, 76, 96, merised to hydrocinnamaldehyde.
and 100 mol% for R = Me, Et, PBun3 and PPri3, respectively.
Figure 2 shows the plots of mol fraction against time ob-
A similar promoting effect of the phosphine on the cat- tained for the hydrogenation of cinnamaldehyde at 95ꢂC
alytic activity was observed for catalysts prepared from the over the catalysts prepared from [Pd2Br4(PMe3)2]/SiO2 and
chloride complexes, [Pd2Cl4(PR3)2] (see Table 1).
[Pd2Br4(PEt3)2]/SiO2. Both catalysts remained 100% se-
The activities of selected catalysts prepared from the sup- lective to hydrocinnamaldehyde at the lower temperature.
ported chloro- and bromo-bridged palladium complexes An induction period was again apparent during which the
were determined at the lower temperature of 95ꢂC in ei- conversion to hydrocinnamaldehyde was very slow. This
ther decalin or heptane solvents so that the progress of the was followed by rapid hydrogenation of the unsaturated
reactions could be followed in more detail. These catalytic aldehyde to hydrocinnamaldehyde. The hydrogenation of
runs showed no apparent dependence on the solvent used cinnamaldehyde was faster with the catalyst formed from
and displayed the same trends in selectivity as their higher [Pd2Br4(PEt3)2]/SiO2 than that from [Pd2Br4(PMe3)2]/SiO2.
temperature counterparts, albeit at the expected lower rates Thus, 100% conversion to hydrocinnamaldehyde occurred
of reaction.
after 180 min for the catalyst formed from the triethylphos-
The hydrogenation of cinnamaldehyde over the cata- phine complex, whereas the reaction slowed down con-
lyst prepared from [Pd2Cl4(PEt3)2]/SiO2 is shown in Fig. 1. siderably after 180 min for the catalyst formed from
The reaction had an induction period of about 40 min be- [Pd2Br4(PMe3)2]/SiO2. Even after 72 h on line only 83% of
fore conversion of cinnamaldehyde to hydrocinnamalde- the cinnamaldehyde had been hydrogenated to hydrocin-
hyde at an appreciable rate was observed. Hydrogenation namaldehyde at this temperature and the reaction had ef-
to hydrocinnamaldehyde was then quite rapid until at high fectively stopped.
conversions the process became reactant limited. Phenyl-
Separate experiments in which the hydrogenation of hy-
propanol was not detected until a significant concentration drocinnamaldehyde and cinnamyl alcohol was studied over
of hydrocinnamaldehyde was formed. This phenylpropanol the catalysts prepared from the bromo complexes again
did not appear to originate from the hydrocinnamalde- showed that hydrocinnamaldehyde was not hydrogenated
hyde, however, since the yield did not increase at the ex- over these catalysts whereas cinnamyl alcohol was readily
FIG. 1. Mole fraction vs time in minutes for the catalyst derived from [Pd2Cl4(PEt3)2/SiO2] and used in the hydrogenation of cinnamaldehyde in
decalin at 95ꢂC. Hydrocinnamaldehyde, ; cinnamaldehyde, ᭡; phenylpropanol, ᭹.

HYDROGENATION CATALYSTS FROM PALLADIUM COMPLEXES
93
FIG. 2. Mole fraction vs time in minutes for the hydrogenations of cinnamaldehyde in decalin at 95ꢂC over the catalysts prepared from silica-
supported [Pd2Br4(PMe3)2] (Cat. 1) and [Pd2Br4(PEt3)2] (Cat. 2). Hydrocinnamaldehyde (Cat. 1), ; cinnamaldehyde (Cat. 1), ᭡; hydrocinnamaldehyde
(Cat. 2), ᭹; cinnamaldehyde (Cat. 2), ꢀ.
hydrogenated to phenylpropanol, this time with negligi- at 95ꢂC. With this catalyst no induction period was observed
ble isomerisation to hydrocinnamaldehyde. Thus, while cin- and cinnamaldehyde was rapidly and completely hydro-
namyl alcohol could conceivably be an intermediate in the genated to hydrocinnamaldehyde. No phenylpropanol was
catalysis by the chloro-bridged complexes, it is clearly not detected until virtually all of the cinnamaldehyde had been
produced when the bromo-bridged complexes are used hydrogenated to hydrocinnamaldehyde, suggesting that
since no phenylpropanol is observed among the products.
this reaction followed a sequential pathway in which cin-
Figure 3 shows the results obtained for the hydrogena- namaldehyde is converted to hydrocinnamaldehyde, which
tion of cinnamaldehyde over the standard Pd/SiO2 catalyst is desorbed from the surface before being readsorbed onto
FIG. 3. Mole fraction vs time in minutes for Pd/SiO2 used in the hydrogenation of cinnamaldehyde in decalin at 95ꢂC. Cinnamaldehyde,
hydrocinnamaldehyde, ᭡; phenylpropanol, ᭹.
;

94
CAIRNS, CROSS, AND STIRLING
the (now vacated) same sites and hydrogenated to phenyl- erably after 1 h, but the formation of phenylpropanol from
propanol. Unsaturated alcohols are known to adsorb more cinnamyl alcohol continues strongly for a longer period.
strongly onto supported metals than on saturated aldehy- The slow conversion of hydrocinnamaldehyde to phenyl-
des, in accordance with this interpretation (38).
propanol then ensues.
Independent observations of the hydrogenations of
Finally, cinnamaldehyde hydrogenation reactions were
hydrocinnamaldehyde and cinnamyl alcohol over the also carried out at 135ꢂC using unsupported [Pd2Cl4 ꢄ
standard Pd/SiO2 catalysts shed more light on the pro- (PMe3)2] and [Pd2Br4(PMe3)2], respectively, in decalin as a
cesses involved and on the nature of the catalyst surfaces. solvent. These catalysts turned black when placed in contact
Hydrocinnamaldehyde was converted to phenylpropanol, with hydrogen gas and formed a noncolloidal suspension
but the reaction proceeded much more slowly than either of particles in the solution. Interestingly, the reactions pro-
the reduction of cinnamaldehyde to hydrocinnamaldehyde ceeded with the same selectivities as their silica-supported
under the same conditions or the reduction of hydrocin- counterparts, but with much lower activities despite the ob-
namaldehyde to phenylpropanol on the catalysts which vious heterogeneous nature of the material produced. The
had first reduced cinnamaldehyde to hydrocinnamalde- operation of an induction period with the bromide-derived
hyde. Conversion was complete only after 48 h. Moreover, catalyst wasagain clearlyvisible, but the reaction proceeded
a distinct induction period was apparent before hydrogena- to only 55% in 3 h at 135ꢂC. The progress on the chloride-
tion commenced, even at 135ꢂC. It isclear from Fig. 3that no derived catalyst was so slow even at 135ꢂC that any such
induction period operates even at 95ꢂC when cinnamalde- activation period was difficult to discern; the hydrogena-
hyde is used as substrate.
tion had progressed to only 8% in 3 h. It is of interest to
The hydrogenation reaction of cinnamyl alcohol over note that the catalytic activities of a variety of complexes
the Pd/SiO2 catalyst, on the other hand, proceeds very have been found to be enhanced when they have been
rapidly to produce about 70% of phenylpropanol and “heterogenised” (23–25). In our case, at least, although
30% of hydrocinnamaldehyde. The reaction then contin- the unsupported metal complexes clearly are not acting in
ues much more slowly to convert the hydrocinnamalde- the homogeneous phase they are much less active than the
hyde to phenylpropanol, with phenylpropanol accounting silica-supported materials.
for 85% of the products after 48 h at 135ꢂC. Figure 4 shows
the early stages of this reaction at 95ꢂC. The operation of
DISCUSSION
an induction period is apparent before both the isomerisa-
tion and the hydrogenation reactions simultaneously com-
The catalysts described in this work fall into three dis-
mence. The processes appear to the independent, however, tinct behaviour profiles: (i) the standard palladium/silica
since the formation of hydrocinnamaldehyde slows consid- catalyst; (ii) catalysts prepared from the binuclear bromo-
FIG. 4. Mole fraction vs time in minutes for Pd/SiO2 used in the hydrogenation of cinnamyl alcohol in decalin at 95ꢂC. Cinnamyl alcohol,
hydrocinnamaldehyde, ᭡; phenylpropanol, ᭹.
;

HYDROGENATION CATALYSTS FROM PALLADIUM COMPLEXES
95
and iodo-phosphine complexes; and (iii) those prepared essentially removed from the palladium. A catalyst made
from the chloro analogues. The Pd/SiO2 catalysts are dealt from PdCl2/SiO2 also behaved in the same way.
==
with first. The hydrogenations of cinnamaldehyde, hy-
The preferential hydrogenation of the C C bonds by the
drocinnamaldehyde, and cinnamyl alcohol have each been palladium catalysts is not surprising. Palladium catalysts are
examined over these catalysts. Cinnamaldehyde reacts se- known to selectively adsorb and reduce alkenes (39). There
quentially to form hydrocinnamaldehyde and then phenyl- is also a precedent for the coordination of cinnamalde-
==
propanol, most probably at the same sites, and with no hyde by its C C bond only to platinum(0) in the molecu-
observable induction period even at 95ꢂC. Cinnamyl alco- lar complex [Pt(PhCH CHCHO)(PPh3)2] (41). There are
==
hol reacts after a short induction period (30 min at 95ꢂC) also precedents for the reaction and activation of H2 at Pt0
both by isomerisation to produce hydrocinnamaldehyde (42), and it is almost certain that palladium(0) complexes
and by hydrogenation to produce phenylpropanol. The rate would behave in the same way, as would the palladium cata-
of production of phenylpropanol at the early stages of the lysts in the present study.
hydrogenation is greater than that from the hydrogenation
The second category of catalysts are those produced
of hydrocinnamaldehyde, either in the sequential process by the H2 reduction of [Pd2Br4(PR3)2]. These can be un-
starting from cinnamaldehyde or independently on a fresh supported, supported on silica by impregnation, or simply
catalyst. The rate of production of hydrocinnamaldehyde ground together with silica. The latter method produces
from this substrate decays after about 1 h at 95ꢂC and it the most active catalysts, but the activity of all of them is
is then steadily but slowly converted to phenylpropanol. remarkable in view of the well known property of halides
The hydrogenation of hydrocinnamaldehyde to phenyl- and phosphines to poison catalysts (43). All of the catalysts
propanol over a fresh Pd/SiO2 catalyst requires a long in- in this category selectively hydrogenate cinnamaldehyde to
duction period (nearly 80 min at 135ꢂC) before it slowly hydrocinnamaldehyde in the liquid phase. They require an
proceeds.
The differing induction periods can best be understood activated with prolonged or repeated use. Exposure to air
==
induction period before reactions begin and are slowly de-
in terms of the greater affinity of palladium for C C bonds between the catalytic runs does not affect the rate of the
==
than for C O bonds (38, 39). Thus strong interactions with deactivation process. Although these catalysts do not re-
==
==
PhCH CHCHO and PhCH CHCH2OH can lead to rapid duce hydrocinnamaldehyde, they readily convert cinnamyl
surface activation, perhaps through the formation of a hy- alcohol to phenylpropanol, so the formation of cinnamyl
drocarbonaceous overlayer (40). The active sites formed alcohol followed by isomerisation cannot contribute to the
from the contact of the fresh Pd/SiO2 catalyst and hydrocin- mechanism.
namaldehyde, on the other hand, would be the results of a
Microanalyses of some of the silica-supported catalysts
weaker interaction. They may simply be fewer in number or show that a substantial proportion of the bromine and phos-
of a different type from those produced in the cinnamalde- phorus available in the precursors is retained by the cata-
hyde reaction, catalysing the hydrogenation of hydrocin- lysts, but it is not yet known what form this is in. The
namaldehyde to phenylpropanol but at a lower rate.
halide is presumably in intimate contact with the active
The isomerisation and hydrogenation reactions under- sites, however, since catalysts prepared from the analogous
gone by cinnamyl alcohol appear to proceed at different iodo-complexes show the same selectivity but are much
sites. While the two processes begin after a similar in- less active. There is also some evidence that the organic
duction period, the production of hydrocinnamaldehyde substituents are retained on phosphorus, since the reaction
falls away but formation of phenylpropanol continues. This rates varied in the sequence R = Pri > Bu > Et > Me. This
must be the result of deactivation of the isomerisation would have to be a size effect, since the electronic parame-
sites rather than of rapid hydrogenation of the hydrocin- ters of the four tertiary phosphines differ only slightly (37).
namaldehyde produced, since the hydrogenation of that However, some caution must be exercised in reaching this
hydrocinnamaldehyde continues only at a very slow rate. conclusion, as it is evident that the physical nature of the
The high rate of formation of phenylpropanol from the hy- catalyst precursors also affects the rate. We cannot there-
drogenation of cinnamyl alcohol would rule out the pos- fore entirely rule out the possibility that these, relatively
sibility of the production and subsequent isomerisation of small, activity differences might be assignable to coinciden-
cinnamyl alcohol as an intermediate in the hydrogenation talparticle size effectsinstead. Thiscaveat appearsthe more
of cinnamaldehyde.
relevant when it is realised that the large steric effects ex-
We note at this stage that catalysts formed from the calci- erted by the halide ions operate in the opposite direction,
nation at 200ꢂC of [Pd2Cl4(PEt3)2] on silica followed by re- the larger ions slowing the reactions.
duction performed in the same way for the hydrogenation
It thus seems likely that these catalysts produce a single
of cinnamaldehyde as did the Pd/SiO2 catalysts prepared type of active site, capable of converting cinnamaldehyde
from [Pd(NO3)2] on silica, indicating that under those rig- to hydrocinnamaldehyde or cinnamyl alcohol to phenyl-
orous conditions all the phosphine and chloride had been propanol. Whilst this again demonstrates the preference

96
CAIRNS, CROSS, AND STIRLING
==
of palladium for C C coordination and reduction, these duced. Like the bromide catalysts, those derived from the
sites are clearly different from those of the Pd/SiO2 cata- PEt3 derivative appear to be more active than those from
lysts above, which were also able to convert hydrocin- PMe3.
namaldehyde to phenylpropanol. While the major differ-
We conclude that at least two types of active sites operate
ence in this family of catalysts is the nature of the halide in these chloro catalysts in the presence of cinnamaldehyde.
present, it must also be kept in mind that the presence One type is probably quite analogous to that which oper-
of phosphine is necessary, since catalysts formed from the ates selectively in the bromide and iodide catalysts and is
reduction of PdBr2 on silica, though not identical with responsible for producing most (but not necessarily all) of
those produced from Pd(NO3)2 or PdCl2, do produce some the hydrocinnamaldehyde. A second type produces phenyl-
phenylpropanol as well as hydrocinnamaldehyde. Likewise, propanol. These appear to require a longer induction period
a standard Pd/SiO2 catalyst modified before use by treat- than the first type before becoming active and are deacti-
ment with (Bu4N)Br also produced both phenylpropanol vated more quickly. This second site type could operate in
and hydrocinnamaldehyde.
one of three ways. It may produce phenylpropanol directly
Hydridopalladium complexes are not very robust, and from cinnamaldehyde via a single visit to the catalyst. The
any palladium hydrides formed on treatment of any of the diminished amount of halide in the catalyst might permit
==
==
[Pd2X4(PR3)2]/SiO2 precursors with H2 would be unlikely simultaneous coordination of both C C and C O to adja-
to survive under the conditions of our experiments. Likely cent Pd atoms, both of which could be hydrogenated before
elimination products would be HX and/or [PR3H]+, result- desorption. Second, the second site types may produce cin-
ing in formal reduction of palladium. The diminished levels namyl alcohol, some of which is further (and rapidly) hy-
of halogen and phosphorus in the catalysts are compatible drogenated and some of which is isomerised to hydrocin-
with the formation of Pd0- and/or PdI-containing materials namaldehyde. It is known that chloride ions can promote
by these routes. Some phosphine loss was apparent from all isomerisation reactions at metal surfaces (45). Third, these
of these catalysts, and minor peaks observed in the GC trace sites could be the same as those produced by the adsorption
of the products correspond to the retention times of free of cinnamaldehyde on Pd/SiO2.
R3P. Some of the retained phosphorus could be in the form
In conclusion, we believe the hydrogen reductions of our
of [PR3H]+, shown to be inactive by itself. These catalysts [Pd2X4(PR3)2] compounds form Pd0- and/or PdI-containing
required a lengthy induction period before hydrogenation materials which retain some of their X and P. The liquid-
of cinnamaldehyde commenced. Since this differs from the phase hydrogenation activities of these solid materials are
behaviour of the Pd/SiO2-based material, it is possible that dependent on the presence or absence of a support and on
surface activation or restructuring of a type different from the physical nature of the precursors. It is likely that the
that caused by contact of Pd with cinnamaldehyde is taking hydrogenation of cinnamaldehyde takes place in the pres-
place.
ence of a carbonaceous overlayer formed from adsorbed
The third category of catalysts, those prepared from the cinnamaldehyde in the case of all the catalysts studied. Cin-
chloride complexes [Pd2Cl4(PR3)2], is the most intriguing. namyl alcohol may adsorb similarly to form an overlayer in
Supported or unsupported, the catalysts require an induc- the case of catalysts formed from [Pd2Cl4(PR3)2]/SiO2 and
tion period and produce both hydrocinnamaldehyde and Pd/SiO2 but hydrocinnamaldehyde does not contribute to
phenylpropanol. However, the phenylpropanol does not the activation process since only Pd/SiO2 showed any ac-
appear to come from the hydrocinnamaldehyde. Although tivity in the hydrogenation of hydrocinnamaldehyde and
a substantial amount of hydrocinnamaldehyde forms be- even then the reaction was very slow. The failure to detect
fore any phenylpropanol is observed, the formation of the an induction period for the hydrogenation of cinnamalde-
latter stops before hydrogenation of cinnamaldehyde to hyde over Pd/SiO2 reflects the fact that the carbonaceous
hydrocinnamaldehyde is complete. Fresh catalysts do not overlayer is formed more rapidly on this catalyst than on
reduce hydrocinnamaldehyde to phenylpropanol, though the catalysts formed from [Pd2X4(PR3)2]/SiO2. Partial loss
this is possibly the result of a different surface modification of halide and/or phosphine, or surface restructuring, could
in the absence of cinnamaldehyde. Some of the phospho- occur during the induction periods for the catalysts formed
rus and chloride from the precursor are retained by the from the supported binuclear palladium complexes. Fur-
catalysts, but the amount of chloride retained is less than thermore, it is likely that the catalysts so formed, unlike
the amount of bromide or iodide from the previous cat- the standard Pd/SiO2 catalysts, are continuously changing
egory discussed. This might be expected, as chloride is a during the course of the reactions, since they are slowly de-
poorer nucleophile towards palladium than towards bro- activated during use. The retained halides and phosphines,
mide or iodide (44) and will thus be more easily removed. which control the activities and selectivities of the catalysts,
Cinnamyl alcohol is partially hydrogenated to phenyl- may become labile under reaction conditions and block ac-
propanol, but most is isomerised to hydrocinnamaldehyde tive sites or promote further (and detrimental) surface re-
which, unlike the case of the Pd/SiO2 catalysts, is not re- structuring.

HYDROGENATION CATALYSTS FROM PALLADIUM COMPLEXES
97
Work is underway to characterise the materials at each 23. Bruner, H., and Bailar, J. C., Inorg. Chem. 12, 1465 (1973).
24. Corain, B., Zecca, M., Biffis, A., Lora, S., and Palma, G.,
J. Organometal. Chem. 475, 283 (1994).
stage of their catalytic lifetime in order to understand the
origins of the remarkable specificity displayed by some and
to enable us rationally to modify the catalysts in a suitable
way to extend their working lifetime.
25. Frolov, V. M., Parenago, O. P., Shuikina, L. P., Novikova, A. V.,
Elnatanova, A. I., Cherkashin, G. M., Kliger, E. G., and Mirskaya,
E. Ya, “Proc. 5th International Symposium on Related Homo-
geneous and Heterogeneous Catalysts, 1986” (Ermakov, Yu. I.,
and Likholobov, V. A., Eds.), VNU Science Press, Utrecht, The
Netherlands, p. 587.
ACKNOWLEDGMENT
The authors thank the EPSRC for the award of a studentship (to
G.R.C.).
26. Kini, A. D., Nadkarni, D. V., and Fry, J. L., Tetrahedron Lett. 35, 1507
(1994).
27. Kumar, K. R., Chaudary, B. M., Jamil, Z., and Thygarajan, G., J. Chem.
Soc., Chem. Comm., 130 (1986).
REFERENCES
28. Kowalak, S., Weiss, R. C., and Balkus, K. J., J. Chem. Soc., Chem.
Comm., 57 (1991).
1. Banerjee, T. K., and Saha, C. R., Indian J. Chem. A 19, 964 (1980).
2. Bhattacharya, S., Santra, P. K., and Saha, C. R., Indian J. Chem. A 23, 29. Chatt, J., and Venanzi, L. M., J. Chem. Soc., 2351 (1957).
724 (1984).
30. Evans, J. G., Goggin, P. L., Goodfellow, R. J., and Smith, J. G., J. Chem.
Soc. A, 464 (1968).
31. Goodfellow, R. J., Evans, J. G., Goggin, P. L., and Duddell, D. A.,
J. Chem. Soc. A, 1604 (1968).
32. Duddell, D. A., Evans, J. G., Goggin, P. L., Goodfellow, R. J., Rest,
A. J., and Smith, J. G., J. Chem. Soc. A, 2134 (1969).
33. Grim, S. O., and Keiter, L., Inorg. Chim. Acta 4, 56 (1970).
3. Santra, P. K., and Saha, C. R., Chem. Ind. (London), 713 (1984).
4. Bose, A., and Saha, C. R., Chem. Ind. (London), 199 (1987).
5. Santra, P. K., and Saha, C. R., J. Mol. Catal. 39, 279 (1987).
6. Bose, A., and Saha, C. R., Indian J. Chem. A 29, 461 (1990).
7. Mukherjee, D. K., Palit, B. K., and Saha, C. R., Indian J. Chem. A 31,
242 (1992).
8. Naiini, A. A., Ali, H. M., and Brubaker, C. H., J. Mol. Catal. 67, 47 34. Uson, R., Fornies, J., and Martinez, F., J. Organometal. Chem. 132, 429
(1991).
(1977).
9. van Asselt, R., and Elsevier, C. J., J. Mol. Catal. Lett. 65, 13 (1991).
10. Lindner, E., Speidel, R., Fawzi, R., and Hiller, W., Chem. Ber. 123,
2255 (1990).
11. Strukul, G., and Carturan, G., Inorg. Chim. Acta 35, 99 (1979).
12. Goldsworthy, D. H., Hartley, F. R., and Murray, S. G., J. Mol. Catal.
19, 257, 269 (1983).
13. Cullen, W. R., and Han, N. F., Appl. Organometal. Chem. 1, 1
(1981).
14. Davies, J. A., Hartley, F. R., and Murray, S. G., J. Mol. Catal. 10, 171
(1981); Inorg. Chem. 19, 2299 (1980); J. Chem. Soc. Dalton Trans., 2246
(1980).
35. Fergusson, J. E., and Heveldt, P. F., Inorg. Chim. Acta 31, 145
(1978).
36. Baratta, W., and Pregosin, P. S., Inorg. Chim. Acta 209, 85 (1993).
37. Tolman, C. A., Chem. Rev. 77, 313 (1977); J. Amer. Chem. Soc. 92, 2953
(1970).
38. Hubaut, R., Bonnelle, J. P., and Daage, M., J. Mol. Catal. 55, 170
(1989).
39. Thomson, S. J., and Webb, G., “Heterogeneous Catalysis,” Oliver and
Boyd, Edinburgh, 1967.
40. Webb, G., Catalysis Today 7, 139 (1990); Thomson, S. J., and Webb, G.,
J. Chem. Soc., Chem. Comm., 526 (1976).
15. Stern, E. W., and Maples, P. K., J. Catal. 27, 120 (1972).
16. Reddy, V. V. S., J. Mol. Catal. 45, 73 (1988).
17. Khan, M. M. T., Khan, B. T., and Begum, S., J. Mol. Catal. 45, 305
(1988).
18. Trost, B. M., and Braslau, R., Tetrahedron Lett. 30, 4657 (1989).
19. Harsy, S. G., Tetrahedron 46, 7403 (1990).
20. Mondal, T. K., Banerjee, T. K., and Sen, D., Indian J. Chem. A 19, 846
(1980).
21. Banerjee, T. K., and Sen, D., J. Chem. Technol. Biotechnol. 31, 676
(1981).
22. Ker, J. M., Suckling, C. J., and Bamfield, P., J. Chem. Soc., Perkin
Trans. (1), 887 (1990).
41. Cenini, S., Ugo, R., and La Monica, G., J. Chem. Soc., A, 409 (1971).
42. Yoshida, T., and Otsuka, S., J. Amer. Chem. Soc. 99, 2134 (1977);
Paonessa, R. S., and Trogler, W. C., J. Amer. Chem. Soc. 104,
1138 (1982); Paonessa, R. S., Prinano, A. L., and Trogler, W. C.,
Organometallics 4, 647 (1985).
43. Meith, J. A., and Schwarz, J. A., J. Catal. 118, 218 (1989); Goodman,
D. W., and Houston, J. E., Science 236, 403 (1987).
44. See, e.g., Atwood, J. D., “Inorganic and Organometallic Reaction
Mechanisms,” Brooks/Cole, Monterey, CA, 1985.
45. Gates, B. C., Katzer, J. R., and Schuit, G. C. A., “Chemistry of Catalytic
processes,” McGraw–Hill, New York, 1979.
46. Salmeron, M., and Somorjai, G. A., J. Phys. Chem. 86, 341 (1982).