Aqueous phase hydrogenation of substituted phenyls over carbon nanofibre and activated carbon supported Pd

Article abstract of DOI:10.1016/j.jcat.2009.11.028

The hydrogenation of aromatic acids and other substituted phenyls has been studied for two palladium-based catalysts; one supported on carbon nanofibres (CNFs) and the other on a steam-activated carbon. The reactions were conducted in both aqueous solution and aprotic organic solvents. The major product over both catalysts was the same irrespective of the substrate indicating that support characteristics and Pd dispersion play at most only a minor role in defining the reaction pathway. The key factor in determining the major reaction product resulting from either preferential hydrogenation of the aromatic ring, or reaction of the external functional group, was the extent to which the external group interacted with water molecules which acted in some cases to protect the external function from interaction with the metal surface and induced selective reduction of the aromatic ring.

Full text of DOI:10.1016/j.jcat.2009.11.028

Journal of Catalysis 270 (2010) 9–15
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Aqueous phase hydrogenation of substituted phenyls over carbon nanofibre
and activated carbon supported Pd
c
c
*
J.A. Anderson , A. Athawale, F.E. Imrie, F.-M. M Kenna, A. M Cue, D. Molyneux, K. Power,
M. Shand, R.P.K. Wells
Surface Chemistry and Catalysis Group, Department of Chemistry, University of Aberdeen, Meston Walk, Old Aberdeen AB24 3UE, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
The hydrogenation of aromatic acids and other substituted phenyls has been studied for two palladium-
based catalysts; one supported on carbon nanofibres (CNFs) and the other on a steam-activated carbon.
The reactions were conducted in both aqueous solution and aprotic organic solvents. The major product
over both catalysts was the same irrespective of the substrate indicating that support characteristics and
Pd dispersion play at most only a minor role in defining the reaction pathway. The key factor in determin-
ing the major reaction product resulting from either preferential hydrogenation of the aromatic ring, or
reaction of the external functional group, was the extent to which the external group interacted with
water molecules which acted in some cases to protect the external function from interaction with the
metal surface and induced selective reduction of the aromatic ring.
Received 2 October 2009
Revised 30 October 2009
Accepted 30 November 2009
Available online 28 December 2009
Keywords:
Pd catalysts
Nanofibres
Aqueous hydrogenation
Substituted aromatics
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
taining functionalities, complete miscibility with water is seldom
achieved [8] thus limiting these types of catalyst for water-based
The need to produce more environmentally acceptable chemical
processes has led to increased interest in replacing stoichiometric
reagents with reusable heterogeneous catalysts. The replacement
of volatile organic compounds (VOCs) by solvents with lower va-
pour pressures or by liquids which are deemed to be environmen-
tally benign is also desirable. The use of water as a solvent has
received attention [1–3] and there are numerous cases in which
additional benefits, such as increased reaction rates have been
found when water replaces organic solvents particularly in Diels
Alder and coupling reactions [4,5]. A less well-documented case
is where the addition of water to hexahydrobenzoic acid as a sol-
vent, lead to enhanced rates of benzoic acid hydrogenation over
Pd, Rh and Ru catalysts [6]. Henry’s constant for hydrogen in water
(7.8 Â 10À4 mol LÀ1 atmÀ1 at 298 K) permits hydrogenation reac-
tions to proceed at moderate hydrogen partial pressures. However,
the rates of reaction may be relatively slow and, potentially,
strongly influenced by gas–liquid mass transfer [7].
Palladium is often the metal of choice in selective hydrogena-
tion reactions and this is often partnered by carbon as a support
due to its high surface area and low relative cost along with the
ease by which the supported metal may be recovered. Although
the relative hydrophilicity of a carbon may be increased by treat-
ment of the surface by reagents such as nitric acid and hydrogen
peroxide which lead to increased density of surface oxygen con-
hydrogenation and other reactions. Oxidation of carbon nanofibres
allows the active phase of the precursor salt to be applied from an
aqueous solution [9]. There are different views as to the potential
impact of such functionalities in selective hydrogenation reactions
on activated carbons [10–12] whereas in the case of carbon nano-
fibres, these oxygen functionalities have been shown to play a key
role in influencing activity and selectivity in the liquid phase
hydrogenation of cinnamaldehyde [9]. In an earlier communication
[13], we reported on the use of nitric acid treated carbon nanofi-
bres (CNFs) as a support for Pd which lead to the formation of a
material which showed high miscibility with water, and permitted
the hydrogenation of water-soluble aromatic acids to proceed with
high selectivity to the corresponding cyclohexane carboxylic acid.
Pd is generally regarded as a poor catalyst for the hydrogenation
of most aromatic rings at low temperatures [14], and, in general,
aromatic ring hydrogenation requires more severe conditions than
those required to hydrogenate other functional groups [6,15]. Car-
boxylic acid groups are difficult to reduce [16,17] but may be con-
verted to aldehydes over oxides at high temperatures and
pressures [16–18]. Current methods for ring hydrogenation of phe-
nyl carboxylic acids include the use of Na–K alloys [19]. Alterna-
tives involve the use of higher temperatures which enable the
reaction of molten phase benzoic acid [20] and there are indica-
tions that this reaction may also be performed over carbon-sup-
ported precious metals in supercritical CO2 [21]. In this paper, Pd
supported on carbon nanofibres are compared with catalysts based
on an acid treated activated carbon to determine whether the high
* Corresponding author. Fax: +44 1224 272921.
E-mail address: j.anderson@abdn.ac.uk (J.A. Anderson).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.11.028

10
J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
Table 1
yields to cyclohexane carboxylic acids in an aqueous phase hydro-
genation [13] were specific to the nanofibre-supported material
and also whether preferential ring hydrogenation occurred when
substrate molecules containing more readily reducible external po-
lar functional groups were employed.
Characteristics of the prepared catalysts.
Sample
BET
Micropore
Micropore
volume
Metal
dispersion
(%)
(m2 gÀ1
)
area (m2 gÀ1
)
(cm3 gÀ1
)
Pd/AC
Pd/CNF
Rh/CNF
1187
185
187
189
–
–
0.06738
–
–
7.15
12.7
29.9
2. Experimental
Carbon nanofibres (CNFs) were prepared by Grupo Antolin
using the floating catalyst technique. Pretreatment of the as-re-
ceived nanofibres involved treating each gram in 100 cm3 of
65 wt% HNO3 at 343 K in a round-bottomed flask fitted with a re-
flux condenser. The resultant material was washed with distilled
water until the washings gave a pH of 5–6. The sample was then
dried overnight at 373 K. The resultant material had an elemental
oxygen content of 6.55%. Pd/CNF catalyst was prepared by combin-
ing the nitric acid pretreated nanofibres with a volume of an aque-
ous solution of Pd(NO3)2 equivalent of the adsorption capacity of
the support and containing a predetermined quantity of salt to pre-
pare a 5% loading of Pd on reduction. The resultant slurry was dried
in air at 373 K for 24 h and then stored. Some tests experiments
were also conducted on a 5% Rh/CNF catalyst which had been pre-
pared from Rh(NO3)3 solution but otherwise was prepared in an
identical fashion to the Pd/CNF.
The activated carbon-based catalyst was prepared as 5% palla-
dium (Pd/AC) loaded on the carbon support (Johnson Matthey, Acid
Treated Carbon) by the wet impregnation method. A pre-weighed
amount of carbon (9.5 g) was wetted by slow addition of ca.
100 ml of water followed by dropwise addition of 249 ml of
0.018 M Pd(NO3)2 solution. After complete addition of solution,
the whole mixture was stirred for a further 16 h followed by rotary
evaporation and drying at 110 °C. The sample was then treated at
523 K in N2 for 1 h to decompose the precursor. A number of exper-
iments were also conducted over a commercial 5% Pt/charcoal (Pt/
AC) sample (Johnson Matthey, Ref. 99905).
Prior to use in reaction or characterisation, all the catalysts were
activated by exposure to a 1:1 H2/N2 mixture (50 cm3 minÀ1) at
423 K for 1 h followed by cooling to 298 K under nitrogen.
Catalytic reactions using 50, 100 or 200 mg of catalyst were car-
ried out using a 0.01 M aqueous solution of the aromatic acid except
in the case of cinnamic (0.003 M) acids. High pressure reactions
were performed at 15 Bar H2 pressure in a 270 cm3 stainless steel
autoclave over 24 h at 358 K with temperature maintained constant
by mounting the autoclave in a pre-heated silicone oil bath. A head
assembly (Ken Kimble) fitted with pressure gauges, relief valve and
magnetically driven stirrer was employed. Reactions at 1 Bar pres-
sure were performed by continuously bubbling hydrogen through
the solution in a three-necked flask with a stirrer arrangement.
The reaction temperature was maintained at 358 K using an oil bath
with a provision for heating and stirring. A series of reactions were
also conducted using cyclohexane as a solvent but otherwise retain-
ing all the other parameters. In the case of the mixed solvent reac-
tion, which involved various compositions of dioxane and water,
these were conducted at 333 K but using otherwise equivalent con-
ditions. Products of reaction were identified by GC–MS and concen-
trations were determined by the use of calibration standards using
either GC or HPLC.
support materials, the surface area of the Pd/AC sample was ca.
six times greater than the Pd/CNF sample and the former contained
a proportion (ca. 16%) of the surface area in micropores. TPR pro-
files showed that Pd precursor in the Pd/CNF was reduced upon
contact with hydrogen at room temperature as indicated by a sin-
gle, negative feature at ca. 353 K resulting from Pd hydride decom-
position [22]. Pd/CNF prepared from Na2PdCl4 on the other hand,
gave a TPR profile with a peak at ca. 400 K due to the reduction
of Pd2+ to Pd0 [23]. TEM and CO chemisorption measurements indi-
cate that a particle size of ca. 9 nm (12.7% dispersion) was preva-
lent. Dried Pd/AC gave a TPD in nitrogen plot which displayed an
intense signal at 523 K indicative of decomposition of the precur-
sor salt and hence the sample was pretreated at 523 K in N2 for
1 h before use. Subsequent TPR profiles displayed a negative fea-
ture due to Pd hydride decomposition [22] along with positive fea-
tures above 623 K which can be attributed to the release of
methane following hydrogenation of the support material. Pd/AC
catalyst reduced at 423 K gave a CO uptake of 16.82 l
mol gÀ1
which, assuming a Pd:CO ratio of 2:1 [24], corresponds with a dis-
persion of 7.15% and an average particle size of 16 nm.
The Rh/CNF gave a single reduction feature in the TPR profile
with an integrated intensity which accounted for a complete
reduction. Sample which had been reduced at 423 K gave a CO up-
take which equated to a dispersion of 29.9% assuming a 2:1 Rh:CO
stoichiometry and which leads to an estimated particle size of
3.8 nm. The much smaller estimated particle size for Rh/CNF com-
pared with Pd/CNF of similar loading was supported by the ease by
which metal crystallites were observed by TEM of the latter but not
of the former.
3.2. Catalytic tests
Pd-based samples were firstly compared in the hydrogenation
of benzoic acid. Cyclohexane carboxylic acid was the only product
detected for both Pd-based samples (Table 2) although neither
sample completely converted all the starting acid after 24 h using
15 Bar hydrogen at 358 K. The conversions equate to average turn-
over frequencies (TOFs) of 2.83 and 5.70 hÀ1 for Pd/CNF and Pd/AC,
respectively. Under otherwise identical conditions, the Pd/AC sam-
ple converted only 34% benzoic acid when the hydrogen pressure
was reduced to 1 atm (equivalent to an average TOF of 2.1 hÀ1
)
although the high selectivity to cyclohexane carboxylic acid was
maintained (Table 2).
The conversion of benzoic acid in an aqueous solution at 333 K
over the Pt/AC was 83% (Fig. 1). However, when the reaction was
Table 2
Conversion of benzoic acid to cyclohexane carboxylic acid over 24 h over the two
catalysts at 358 K.
3. Results
Catalyst
15 Bar hydrogen
Conversion
1 Bar hydrogen
Conversion
34
Selectivity
Selectivity
100
3.1. Characterisation
Pd/ACa
92
80
100
100
Pd/CNFb
Table 1 contains the relevant surface area and pore characteris-
tics of the two, as-prepared Pd-containing samples. Consistent
with the expectations based on the data for the original starting
Catalyst (100 mg) using 50 cm3 benzoic acid solution.
Catalysts (50 mg) using 25 cm3 benzoic acid solution.
a
b

J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
11
reaction period. The least reactive of the three, the benzamide,
was still converted to a greater extent (92%) than the benzoic acid
(80%). As observed for the acid, the main product of benzamide
reaction involved hydrogenation of the aromatic ring, although
with a selectivity of 96.1% in the case of the amide as opposed to
100% for the acid. Recycle experiments using Pd/CNF gave 91.3%
conversion and 96.1% selectivity to the cyclohexane carboxamide.
Both the benzaldehyde and acetophenone underwent hydrogena-
tion of the carbonyl functional groups which was not observed in
cases of the acid or amide. However, products originating from ring
hydrogenation were still the dominant feature with 59.7% of the
products of acetophenone, and 83.2% in the case of benzaldehyde
resulting from ring hydrogenation.
On the other hand, Rh/CNF catalyst converted 100% of the ace-
tophenone and produced two major products, which resulted from
hydrogenation of the aromatic ring (Table 3). Cyclohexylethanol
was the principal product with a selectivity of 85.4% while cyclo-
hexylethanone was present at 10.3% of the total products. Other
minor products were detected. The results suggest that Rh prefer-
entially hydrogenated the aromatic ring of acetophenone initially
followed by partial hydrogenation of the external carbonyl
function.
To determine the extent to which additional reducible func-
tional groups influenced the balance between ring versus external
functional group hydrogenation, and other factors such as the
length of carbon chain between the ring and the acid functionality
were important, a number of substrates were selected and com-
pared with the data for benzoic acid (Table 4). Increasing the
length of the carbon chain separating the acid functionality from
the phenyl group by the use of phenylacetic acid did not diminish
the excellent selectivity to ring hydrogenated product although the
conversion for both the Pd/AC and Pd/CNF samples was diminished
with respect to benzoic acid. The impact of increasing the chain
80
60
40
20
0
0
20
40
60
80
100
% Dioxane in water
Fig. 1. Conversion of benzoic acid as a function of solvent composition over Pt/AC at
333 K.
performed in dioxane under similar conditions, conversion was
only 3%. In both cases, cyclohexane carboxylic acid was the only
reaction product. As the proportions of dioxane and water were
varied, the extent of reaction was modified with water-rich solu-
tions showing the higher levels of conversion (Fig. 1).
To determine whether the high selectivity to the hydrogenated
ring was specific to substrates which included an acid functional
group, a number of other functionalised phenyls were studied (Ta-
ble 3). All molecules, as with the benzoic acid, contained an exter-
nal carbonyl functional group. In the case of the Pd/CNF, all
substrates were more reactive than the benzoic acid with benzal-
dehyde and acetophenone being 100% converted during the 24-h
Table 3
Conversion and selectivity to major products of the aqueous phase hydrogenation at 15 Bar H2 and 358 K of several carbonyl containing substituted aromatic compounds.
Catalyst
Substrate (conversion)
Ring hydrogenated
product (selectivity)
Functional group hydrogenated
product (selectivity)
O
H
HO
OH
O
+
Benzaldehyde
Benzyl alcohol
Cyclohexyl
methanal
Cyclohexyl
methanol
Pd/AC (%)
Pd/CNF (%)
100
100
85.0
83.2
15.0
16.8
O
O
HO
OH
+
Cyclohexyl
ethanone
Cyclohexyl
ethanol
Acetophenone
Phenyl-1-ethanol Ethylbenzene
Pd/AC (%)
Pd/CNF (%)
Rh/CNF (%)
100
100
100
65
59.7
95.4
35.0
40.3
4.6
O
NH2
O
NH2
Benzamide
Cyclohexane carboxamide
Pd/AC (%)
Pd/CNF (%)
100
92
100
96.1

12
J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
Table 4
Conversion and selectivity of the hydrogenation at 15 Bar H2 and 358 K of several aromatic acids containing additional functional groups.
Catalyst
Substrate (conversion)
OH
Primary product (selectivity)
OH
Other product(s)
None
O
O
Benzoic acid
Cyclohexanecarboxylic
acid
Pd/AC (%)
Pd/CNF (%)
92
80
100
100
None
O
O
OH
OH
Phenyl acetic acid
Cyclohexaneacetic acid
Pd/AC (%)
Pd/CNF (%)
81
65
100
100
O
HO
OH
HO
O
Cyclohexylethanol
Cyclohexylpropanoic acid
Cinnamic acid
Pd/AC (%)
Pd/CNF (%)
100
100
100
97
0
3
O
OH
O
OH
OH
Cyclohexanol
Cyclohexanecarboxylic
acid
OH
4-hydroxy benzoic
acid
Pd/AC (%)
Pd/CNF (%)
100
93
75
51
0
3
length further, but also adding an olefinic functionality to the chain
separating the acid group and the aromatic ring was tested by the
use of cinnamic acid (Table 4). The results indicate high activity of
the catalysts for this substrate (with respect to benzoic acid) where
the starting concentration was three times that of the benzoic acid.
However, the olefinic double bond was also hydrogenated in the
principal product obtained. The high conversion level obtained
lead to high selectivity to cyclohexylpropanoic acid indicative that
both olefinic and phenyl rings had undergone hydrogenation.
Experiments conducted for various time periods less that 24 h
showed that 65% of the starting cinnamic acid had undergone reac-
tion in the first 4 h and that the principal intermediate formed (to-
wards the formation of cyclohexylpropanoic acid) still retained the
aromatic ring. This indicates that the olefinic function of the cin-
namic acid underwent initial, more rapid reaction, and that this
was followed by ring hydrogenation showing that the acid group
again was preserved at the expense of the ring, consistent with
the findings for the other aromatic acids studied (Table 4). Aro-
matic acids, which also contained phenolic functions, were studied
using 4-hydroxy benzoic acid as an example. The results at the
high conversions obtained (Table 4) indicate that the phenyl unit
had been reduced through to the corresponding cyclohexyl rather
than cyclohexanone product but again, with retention of the acid
functionality. Only in the case of the Pd/CNF catalysts was there
indication of acid group hydrogenation to generate minor quanti-
ties of cyclohexanol.
In order to test the impact of solvent choice on selectivity for
the substrates containing external functional groups, the hydroge-
nation of acetophenone and benzaldehyde and benzamide were
studied using cyclohexane as a solvent. In the case of benzamide,
the extent of hydrogenation was limited and only the product of
ring hydrogenation was obtained. However, in the cases of the
benzaldehyde and acetophenone, with readily reducible external
carbonyl functional group [14], no ring hydrogenation products
were detected (Table 5), despite the high conversion levels ob-
tained and the only reaction products were the result of partial
or full hydrogenation of the external carbonyl function.
4. Discussion
The use of carbon nanofibre-based catalysts has received signif-
icant attention in the recent years with one key advantage over

J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
13
Table 5
Conversion and selectivity in the hydrogenation of several carbonyl containing substituted aromatic compounds at 15 Bar H2 and 358 K using cyclohexane as a solvent.
Catalyst
Substrate (conversion)
Ring hydrogenated products (selectivity)
Functional group hydrogenated products (selectivity)
O
H
HO
O
HO
+
Benzaldehyde
Cyclohexyl
methanal
Cyclohexyl
methanol
Benzyl alcohol Toluene
Pd/AC (%)
Pd/CNF (%)
100
100
0
0
100
93
O
O
HO
OH
+
Cyclohexyl
ethanone
Cyclohexyl
ethanol
Acetophenone
Phenyl-1-ethanol Ethylbenzene
Pd/AC (%)
Pd/CNF (%)
100
100
0
0
100
100
O
NH2
O
NH2
Benzamide
Cyclohexane carboxamide
Pd/AC (%)
Pd/CNF (%)
10
10
100
100
activated carbons being the absence of micropores (Table 1) which
removes the potential disadvantages resulting from diffusion lim-
itations and secondary reactions [12] which often lead to poorer
selectivity to the desired product. It is noted that the activated car-
bon selected for comparison here had a relatively low proportion of
its total surface area (ca. 16%) in micropores and this value is often
considerably higher. Other advantages of the CNF have been made
regarding selectivity induced through functionalisation of the
nanofibre surfaces by oxidation [9,13,25]. The results obtained
here, by comparing selectivities in the hydrogenation of function-
alised aromatics, provide no evidence (Tables 3–5) for modified
behaviour over carbon nanofibres compared with the activated
carbon when used to support the same metal (i.e. Pd). Differences
in dispersion (Table 1) also led to minimal differences in selectivity
although differences in turn-over frequencies were noted, with the
poorer dispersed Pd on the activated carbon support (Table 1) giv-
ing the greater TOF for benzoic acid hydrogenation. As the only
reaction product observed was cyclohexane carboxylic acid, which
resulted from ring hydrogenation which may be assumed to re-
quire a flat (parallel) lying ring requiring an ensemble of metal
atoms (Fig. 2), it can be assumed that smaller crystallites on higher
dispersed samples contain arrangements of atoms at steps and
edges which are unable to adsorb the aromatic ring in such a
favourable orientation.
Replacement of palladium by rhodium as the active metal on
the carbon nanofibres led (Table 3) to a predictable [6,12] en-
hanced selectivity to products involving hydrogenation of the aro-
matic ring with hydrogenation of the external carbonyl group of
acetophenone occurring as a secondary (consecutive) reaction (Ta-
ble 3). Palladium on the other hand, gave a mixture of products for
reaction with acetophenone which resulted from either initial ring
hydrogenation or carbonyl group hydrogenation when the reaction
was conducted in water (Table 3). However, as will be discussed in
detail below, this was a considerable shift in selectivity from when
Fig. 2. Representation showing the role of the water in solvating the external, polar
functional group and orientating the aromatic ring over the Pd surface in a manner
which facilitates ring hydrogenation.
the reaction was conducted in cyclohexane where no ring hydroge-
nation products were observed (Table 5). A similar shift towards
ring hydrogenation was observed using benzaldehyde as
substrate.
a
The use of solvents in liquid-phase hydrogenation reactions to
influence selectivity in substrates with multiple functional groups
has been reported. In molecules containing both olefinic and car-
bonyl groupings, polar solvents appear to activate hydrogenation
of the C@O whereas non-polar solvents favoured C@C hydrogena-
tion [26,27]. In the case of molecules containing both aromatic and
carbonyl functions, choice of solvent with increasing dielectric
constant, led to an increase in reaction rate (conversion) of aceto-
phenone over Pt–Ni, and with reduced selectivity to the 1-phenyl

14
J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
ethanol. However, no ring hydrogenation was observed and the
loss of selectivity was mainly a consequence of formation of ether
and methylene groups at the external function [28]. Reactions in
this study performed in aprotic solvents over Pd/CNF and Pd/AC
(Table 5) gave only phenyl 1-ethanol and ethyl benzene consistent
with the results using n-hexane as a solvent [28]. The results using
water, with a dielectric constant of 80, higher than any solvent em-
ployed by Malyala in their studies of acetophenone [28], show that
the selectivity towards products arising from the hydrogenation of
external functional group dropped from 100% (Table 5) to 40.3%
(Table 3) with the remaining products resulting from hydrogena-
tion of the aromatic ring. Similarly, the hydrogenation of benzalde-
hyde which produced only benzyl alcohol and toluene in aprotic
solvent (Table 5) produced mainly products arising from ring
hydrogenation when the reaction was performed in water (Table
3).
In addition to solvation effects, limiting interaction between the
hydrogenating Pd surface and the polar, reducible functional
group, high selectivities towards ring hydrogenation partially arise
due to enhance reaction rates of ring hydrogenation in water com-
pared with aprotic solvents. The observation of enhanced reaction
rates of benzoic acid over Ru/C and Rh/C by addition to water was
made by Rylander [6] and is entirely consistent with observations
made here (Fig. 1) where the conversion of benzoic acid in water,
dioxane and mixtures of the two, showed reaction rates to be
greater in water. Note that the differences cannot be attributed
to hydrogen availability since the mole fraction of hydrogen in
water is approximately an order of magnitude less than it is in
dioxane. (Mole fraction in water at 15 Bar (assuming Henry’s law
holds) is 1.965 Â 10À4 whereas the value can be calculated as
3.585 Â 10À3 for dioxane. Value for water at 333 K quoted in [29]
and calculated at 333 K for dioxane using the relationship,
lnx = À5.7347–8.6743/(T/100 K) [29].)
(1 1 1) when it is completely solvated by water [33]. It is not clear
at this stage whether it is as benzoic acid or as the benzoate anion
that the species undergoing hydrogenation is present.
Although the strong solvation of the external polar functional
group in water was sufficient to orientate the molecule in such a
manner that preferred adsorption mode was through the aromatic
ring (Fig. 2), leading to reduced selectivities to the functional group
hydrogenated product which had been favoured in aprotic solvent,
external groups such as the olefinic unit in cinnamic acid, which
are readily hydrogenated over Pd catalysts [14], did not appear
to receive significant protection.
5. Conclusions
Aromatic compounds with highly polar functional groups such
as acids and amides undergo preferred ring hydrogenation where
the rate of ring hydrogenation is enhanced compared to reaction
in aprotic solvents due to a combination of strong solvation of
the polar side group and solvent-assisted orientation of the ring
on the metal surface which favoured hydrogenation. The nature
of the carbon support has little influence in this reaction. Less polar
functional groups which were more prone to hydrogenation in
aprotic solvents were less susceptible to reaction in water, which
again enhanced the relative rate of hydrogenation of the aromatic
ring, leading to considerable differences in selectivity in the two
types of solvent.
Acknowledgments
We thank the Commonwealth Scholarship Commission for a
fellowship (to A.A.) and the Carnegie Trust for the Universities of
Scotland for summer studentships awards (to A.Mc, F.E.I. and
M.S.). We also greatly appreciate initial discussions with Prof. An-
gel Linares-Solano, Carbon Materials and Environmental Research
Group, Departamento de Quimica Inorganica, Universidad de Ali-
cante, Spain regarding the use of pretreated carbon nanotubes
and for supplying these for preliminary investigations.
Increased rate of ring hydrogenation can be attributed to
favourable orientation of the aromatic ring on the metal surface
brought about by solvation of the polar functional group, and ori-
entation of the ring due to the local structure of water in the prox-
imity of the surface. The orientation of the molecule via solvation
effects is known and may be exemplified by the case of methanol
on Pt(1 1 1) where in the vapour phase, the methanol interacts
with the surface though the oxygen atom, whereas on the intro-
duction of water, methanol twists through 90° to allow the CH
bond to interact with the surface [30]. This is driven by a desire
of the molecule to maximise methanol–water interactions of the
polar groups. Similar scenarios leading to preferential ring down
adsorption (Fig. 2) would explain the selectivity effects observed
leading to increased ring hydrogenation relative to function group
hydrogenation in water (Table 3) in cases where the functional
group undergoes facile hydrogenation in aprotic solvents (Table
5). However, the enhanced reaction rate (increased conversion)
in water compared to aprotic solvent of molecules containing func-
tional groups not readily hydrogenated such as benzamide (Tables
3 and 5) and benzoic acid (Fig. 1) suggests that additional factors
are involved. On surfaces such as Pt, Pd and Rh employed here,
water is known to form ordered overlayers as a consequence of
the ability of water to wet the metal surfaces as predicted by the
hydrophilicity or wettability parameter, w, as described by Meng
References
[1] U.M. Lindström (Ed.), Organic Reactions in Water, Blackwell Publishing,
Singapore, 2007.
[2] G.-J. ten Brink, I.W.C.E. Arends, R.A. Sheldon, Science 287 (2000) 1637.
[3] V.L. Budarin, J.H. Clark, R. Luque, D.J. Macquarrie, Chem. Commun. (2007) 634.
[4] R. Breslow, D. Rideout, J. Am. Chem. Soc. 102 (1980) 7816.
[5] S. Otto, J.B.F.N. Engberts, Pure Appl. Chem. 72 (2000) 1365.
[6] P.N. Rylander, Catalytic Hydrogenation over Platinum Metals, Academic Press,
New York, 1967.
[7] Y. Sun, R.N. Landau, J. Wang, C. LeBlond, D.G. Blackmond, J. Am. Chem. Soc. 118
(1996) 1348.
[8] B.K. Pradhan, N.K. Sandle, Carbon 37 (1999) 1323.
[9] M.L. Toebes, F.F. Prinsloo, J.H. Bitter, A.J. van Dillen, K.P. de Jong, J. Catal. 214
(2003) 78.
[10] F. Coloma, A. Sepulveda-Escribano, J.L.G. Fierro, F. Rodríguez-Reinoso, Appl.
Catal. A 150 (1995) 165.
[11] B. Bachiller-Baeza, A. Guerrero-Ruiz, I. Rodríguez-Ramos, Appl. Catal. A 192
(2000) 289.
[12] D. Jin Suh, T.-J. Park, S.-K. Ihm, Ind. Eng. Chem. Res. 31 (1992) 1850.
[13] J.A. Anderson, F.-M. McKenna, A. Linares-Solano, R.P.K. Wells, Catal. Lett. 119
(2007) 16.
[14] H.-U. Blaser, A. Indolese, A. Schnyder, H. Steiner, M. Studer, J. Mol. Catal. A 173
(2001) 3.
et al. [31] w = E
HB/DEads, where the numerator denotes the
[15] B. Chen, U. Dingerdissen, J.G.E. Krauter, H.G.J. Lansink Rotgerink, K. Möbus, D.J.
Ostgard, P. Panster, T.H. Riermeier, S. Seebald, T. Tacke, H. Trauthwein, Appl.
Catal. A 280 (2005) 17.
[16] R.L. Augustine, Catal. Rev. Sci. Eng. 13 (1976) 285.
[17] E.J. Grootendorst, R. Pestman, R.M. Koster, V. Ponec, J. Catal. 148 (1994) 261.
[18] T. Yokoyama, N. Yamagata, Appl. Catal. A 221 (2001) 227.
[19] D. Gaude, R.L. Goaller, J.-L. Luche, J.-L. Pierre, Tetrahedron Lett. (1984) 5897.
[20] Snia Viscosa, Italian Patent, IE1122161/1968.
water–water hydrogen bond strengths and the denominator de-
notes the calculated adsorption energy for water on the different
metals. Although such a semi-rigid layer at the metal surface
may impose a barrier to the solute molecule in terms of transport,
this layer may also enhance reactivity by localising (and orientat-
ing) the surface species close to the metal and stabilising partial
charges in the transition or product states [32]. For example, the
acetate anion is stabilised by an additional 57 kJ molÀ1 on Pd
[21] H. Wang, F. Zhao, Int. J. Mol. Sci. 8 (2007) 628.
[22] J. Sá, G.D. Arteaga, R.A. Daley, J. Bernardi, J.A. Anderson, J. Phys. Chem. 110
(2006) 17090.

J.A. Anderson et al. / Journal of Catalysis 270 (2010) 9–15
15
[23] C.D. Taboada, J. Batista, A. Pintar, J. Levec, Appl. Catal. B 89 (2009) 375.
[24] J.A. Anderson, M. Fernandez-Garcia, A. Martinez Arias, in: J.A. Anderson, M.
Fernandez-Garcia (Eds.), Supported Metals in Catalysis, ICP, London, 2004, p.
123.
[28] R.V. Malyala, C.V. Rode, M. Arai, S.G. Hegde, R.V. Chaudhari, Appl. Catal. A 193
(2000) 71.
[29] C.L. Young (Ed.), Hydrogen and Deuterium, Solubility Series, vol. 5/6. IUPAC,
Pergamon, 1981.
[25] H. Markus, A.J. Plomp, P.M. Arvela, J.H. Bitter, D.Y. Murzin, Catal. Lett. 113
(2007) 141.
[26] H. Yamada, S. Goto, J. Chem. Eng. Jpn. 36 (2003) 586.
[27] I. Kun, G. Szöllösi, M. Bartok, J. Mol. Catal. 169 (2001) 235.
[30] T.R. Mattson, S.J. Paddison, Surf. Sci. 544 (2003) L697.
[31] S. Meng, E.G. Wang, S. Gao, J. Chem. Phys. 119 (2003) 7617.
[32] C.D. Taylor, M. Neurock, Curr. Opin. Solid State Mater. Sci. 9 (2005) 49.
[33] S.K. Desai, V. Pallassana, M. Neurock, J. Phys Chem. B 105 (2001) 9171.