Nickel particle size effects in catalytic hydrogenation and hydrodechlorination: Phenolic transformations over nickel/silica

Article abstract of DOI:10.1039/b212407f

The gas phase hydrodechlorination of 2,4-dichlorophenol (2,4-DCP) and 2-chlorophenol (2-CP) has been studied over three series of Ni/SiO₂ catalysts prepared by three different methods, i.e. (i) impregnation with nickel nitrate, (ii) impregnation with nickel ethanediamine and (iii) deposition-precipitation. This has led to catalysts with a wide range of Ni loadings (0.7-20.9% w/w Ni) and an average Ni particle size from 1.4 to 16.8 nm. The chloroarene was continuously fed to the reactor as a methanolic or aqueous solution (with subsequent vaporisation) where consistently higher hydrodechlorination activities were associated with the aqueous feed due to methanol/support interactions that served to inhibit the surface hydrogenolytic reaction. Hydrodechlorination of 2-CP yielded the fully dechlorinated aromatic product (phenol) and HCl while 2,4-DCP also yielded the partially dechlorinated (2-CP) aromatic product. There was no detectable aromatic ring reduction and the hydrodechlorination kinetics can be represented by a pseudo-first order approximation. Structure sensitivity has been conclusively established where the specific Cl removal was consistently greater with an increase in the average Ni particle diameter. The specific hydrodechlorination rate for 2,4-DCP was lower than that for 2-CP, diagnostic of a deactivating electronic effect associated with the Cl substituent(s) and this was manifested by higher apparent activation energies (ΔE_a) for the 2,4-DCP reaction; the ΔE_a values were unaffected by catalyst preparation/activation/Ni loading. The direct hydrogenation of phenol (as feed) was largely insensitive to Ni particle size albeit there was a discernible decrease in specific phenol consumption over average Ni diameters less than ca. 3 nm. Hydrogenation selectivity remained unchanged and cyclohexanol was the predominant product irrespective of Ni loading or synthesis route.

Full text of DOI:10.1039/b212407f

View Article Online / Journal Homepage / Table of Contents for this issue
Nickel particle size effects in catalytic hydrogenation and
hydrodechlorination: phenolic transformations over nickel/silica
a
Gonzalo Pina, Catherine Louis and Mark A. Keane*
b
a
a
Department of Chemical and Materials Engineering, University of Kentucky,
Lexington KY 40506-0046, USA. E-mail: makeane@engr.uky.edu; Tel: +1 859 2575857
Laboratoire de R e´ activit e´ de Surface, UMR 7609 CNRS, Universit e´ Pierre et Marie Curie,
b
4
place Jussieu, 75252 Paris Cedex 05, France
Received 17th December 2002, Accepted 5th March 2003
First published as an Advance Article on the web 21st March 2003
The gas phase hydrodechlorination of 2,4-dichlorophenol (2,4-DCP) and 2-chlorophenol (2-CP) has been
studied over three series of Ni/SiO catalysts prepared by three different methods, i.e. (i) impregnation with
2
nickel nitrate, (ii) impregnation with nickel ethanediamine and (iii) deposition-precipitation. This has led to
catalysts with a wide range of Ni loadings (0.7–20.9% w/w Ni) and an average Ni particle size from 1.4 to 16.8
nm. The chloroarene was continuously fed to the reactor as a methanolic or aqueous solution (with subsequent
vaporisation) where consistently higher hydrodechlorination activities were associated with the aqueous
feed due to methanol/support interactions that served to inhibit the surface hydrogenolytic reaction.
Hydrodechlorination of 2-CP yielded the fully dechlorinated aromatic product (phenol) and HCl while 2,4-
DCP also yielded the partially dechlorinated (2-CP) aromatic product. There was no detectable aromatic ring
reduction and the hydrodechlorination kinetics can be represented by a pseudo-first order approximation.
Structure sensitivity has been conclusively established where the specific Cl removal was consistently greater
with an increase in the average Ni particle diameter. The specific hydrodechlorination rate for 2,4-DCP was
lower than that for 2-CP, diagnostic of a deactivating electronic effect associated with the Cl substituent(s) and
a a
this was manifested by higher apparent activation energies (DE ) for the 2,4-DCP reaction; the DE values were
unaffected by catalyst preparation/activation/Ni loading. The direct hydrogenation of phenol (as feed) was
largely insensitive to Ni particle size albeit there was a discernible decrease in specific phenol consumption over
average Ni diameters less than ca. 3 nm. Hydrogenation selectivity remained unchanged and cyclohexanol was
the predominant product irrespective of Ni loading or synthesis route.
the supported metal particle, notably particle size and electron
density. In this paper we examine the action of Ni/SiO (with
2
Introduction
1
7
The performance of solid transition metal catalysts is deter-
mined by the availability and reactivity of surface metal sites
where the smaller the metal particle, the greater is the surface:
varying Ni loading and sample preparation) in promoting (a)
the hydrodechlorination of 2,4-dichlorophenol (2,4-DCP)
and 2-chlorophenol (2-CP) and (b) the hydrogenation of phe-
nol. The role of the surface active metal centre in the catalytic
transformation of haloarenes is still open to question. We have
chosen 2,4-DCP and 2-CP as model reactants that are also
1
,2
volume ratio. The generation of ‘‘small’’ metal particles is
facilitated through the use of a support to which the particles
are more or less firmly anchored, and on which they are effec-
tively separated from each other. There is, however, a powerful
thermodynamic driving force towards particle aggregation and
growth (sintering). The mean diameter of the supported metal
particle is influenced by a number of interrelated variables that
1
8
established sources of environmental pollution. Catalytic
hydrodechlorination represents a progressive non-destructive
environmental remediation strategy where the chloroarene is
transformed into recyclable products in a closed system with
limited toxic emissions. The gas phase hydrodechlorination
of mono-chlorophenols has been reported over Ni-Mo/
3
–8
5,6,9
include synthesis method,
loading, prereduction treatment
tions.
nature of the support,
metal
and reduction condi-
2
2,10
2
,11
19,20
21
22
The performance of supported metals can be gov-
Al
2
O
3
,
multiple Cl atoms from an aromatic host has been studied to
Ni/Al
2
O
3
and Ni/SiO
2
while the removal of
erned to varying degrees by the combined influence of
particle morphology, dispersion and electronic properties. Cat-
alysed reactions have been classified as being either structure-
sensitive or structure-insensitive, an attribute that is entirely
dependent on the nature of the metal-reaction system under
2
3,24
a lesser extent.
structure sensitivity or insensitivity in hydrodechlorination
There is no general consensus regarding
2
5
systems. However, Juszczyk and co-workers have noted a
higher turnover frequency of CF CFCl and CCl for larger
Pd particles supported on Al and attributed this to an
ensemble effect. The hydrogenation of phenol has been
3
2
2 2
F
1
consideration. Supported metal particles with diameters from
2 3
O
less than 1 nm up to 10 nm can display quite distinct catalytic
activities/selectivities associated with a switch from atomic to
bulk metal character.
2
6–29
reported for a range of nickel based catalysts; catalytic
activity and selectivity (cyclohexanone vs. cyclohexanol pro-
duction) over supported metals has been found to depend on
1
2
Supported nickel catalysts, the focus of this work, have been
widely used to promote hydrogen addition/scission reac-
3
0–33
31,32,34
metal loading,
inclusion of promoters (alkali metals, alkaline earth metals
surface acidity/basicity
and the
1
3–16
tions. The hydrogenation and hydrogenolysis of aromatic
compounds has been shown to be dependent on the nature of
2
8,32
and lanthanides).
The impact of Ni particle size (supported
DOI: 10.1039/b212407f
1
924
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931
This journal is # The Owner Societies 2003

View Article Online
ꢀ
1
3
ꢀ1
on SiO
2
) on the activity/selectivity associated with these two
min in 25 cm min 5% v/v H
to the reaction temperature.
2
/Ar to 973 K, then cooled
related hydrogenation and hydrogenolysis (hydrodechlorina-
tion) reactions is assessed in this report. This study represents
a collaboration between two groups that have been involved in
Samples for off-line analysis were cooled (in He) and passi-
vated in a 1% v/v O /He mixture at room temperature. The
2
fine tuning the preparation and reduction of Ni/SiO
with the goal of controlling metal dispersion characteris-
tics: this report addresses the catalytic implications
of the synthesis/reduction strategies.
2
catalysts
latter step provided a protective oxide layer over the Ni surface
that prevented bulk oxidation upon exposure to the atmo-
sphere. High resolution transmission electron microscopy
(HRTEM) analysis was carried out using a JEOL 2000FX
TEM/STEM equipped with a UTW energy dispersive X-ray
1
0,11,35–40
(
2
EDX) detector and operated at an accelerating voltage of
00 kV. The specimens were prepared by ultrasonic dispersion
Experimental
in 2-butanol, evaporating a drop of the resultant suspension
onto a holey carbon support grid. Selected area electron dif-
fraction (SAED) confirmed that the Ni component was present
in the metallic form and not as an oxide. The Ni particle size
distribution profiles (and mean Ni sizes) presented in this study
are based on a measurement of up to 1100 individual particles.
Catalyst preparation, reduction and characterisation
Impregnation with nickel nitrate (INinitrate). A range of Ni
impregnated samples (labeled INinitrate -I to -V, where 0.7–
9
.6% w/w Ni) was prepared by standard incipient wetness
impregnation where a 2-butanolic Ni(NO solution was
added dropwise at 353 K to the silica substrate (Sigma-
2
ꢀ1
3
)
2
The specific Ni surface area (SNi , m g(Ni) ) was calculated
from the relationship
42
2
ꢀ1
3
ꢀ1
Aldrich, SBET ¼ 192 m
g
, pore volume ¼ 1.73 cm
g ,
˚
6
average pore diameter ¼ 198 A) with constant agitation (500
SNi ¼
ð1Þ
4
1
rpm). The Ni content was determined (to within ꢁ2%) by
rdNi
atomic absorption spectrophotometry (VarianSpectra AA-
where dNi is the mean particle size and r is the Ni specific mass.
Analysis of the catalysts after reaction did not reveal any
appreciable changes to the average Ni particle size.
10); samples for analysis were digested in HF (37% conc.) at
ambient temperature. The impregnated samples were dried
ꢀ
1
overnight at 383 K and activated by heating (10 K min , con-
trolled using a Eurotherm 91 e temperature programmer) in
3
0 cm min (monitored using a Humonics Model 529 digital
ꢀ1
Catalytic reactor system
6
flow meter) dry H (99.999% v/v) to 723 K, maintaining this
2
All the catalytic reactions were carried out under atmospheric
pressure in situ (after reduction) in a fixed bed vertically
mounted continuous flow glass reactor (i.d. ¼ 15 mm) over
the temperature range 408 K ꢂ T ꢂ 483 K. The catalyst was
supported on a glass frit and a layer of glass beads above
the catalyst bed served as a pre-heating zone that ensured
the reactants reached the reaction temperature before contact-
ing the catalyst. The reactor temperature was monitored by
a thermocouple inserted in a thermowell within the catalyst
bed; the reactor temperature was constant to within ꢁ1 K. A
Model 100 (kd Scientific) microprocessor controlled infusion
pump was used to deliver the organic feed (as aqueous or
methanolic solutions) via a glass/Teflon air-tight syringe and
Teflon line at a fixed, calibrated rate and the vapour was
carried through the catalyst bed in a stream of dry purified
temperature for 6 h, with a subsequent cooling to the reaction
temperature.
Deposition-precipitation (DPNi). A 24% w/w sample was
prepared by deposition-precipitation as described else-
1
1,37–39
where.
This synthesis route involves the precipitation
of a nickel(II) phase onto the silica support by basification of
a nickel salt solution/silica suspension via the decomposition
of urea. A porous silica Spherosil XOA400 (Rh oˆ ne-Poulenc,
2
ꢀ1
France, purity > 99.5%, SBET ¼ 356 m g , pore volume ¼
3
ꢀ1
˚
1
as support. A 500 cm aqueous solution containing nickel
.25 cm g , average pore diameter ¼ 80 A) was employed
3
ꢀ
3
nitrate (0.14 mol dm ), urea (0.42 mol dm ) and nitric acid
ꢀ3
ꢀ
3
0.02 mol dm ) was added to a vessel containing 3.8 g of silica
(
and thermostated at 363 K. The suspension (7.6 g(silica) dm
2
7
ꢀ3
hydrogen ( > 99.999%). Phenol, 2-chlorophenol (2-CP) and
)
2
9
,4-dichlorophenol (2,4-DCP) served as feedstock (Aldrich,
9 + %) where the inlet molar flow rate was in the overall range
was stirred magnetically for 4 h, then cooled to 300 K and fil-
tered. The resultant solid was washed three times by stirring
for 10 min at 323–333 K in 800 cm distilled water with subse-
ꢀ6
ꢀ2
ꢀ1
3
1 ꢃ 10 –5 ꢃ 10 mol h ; all the reactants were used without
further purification. Hydrogen was maintained in at least 70 or
1
quent filtration. The sample was then dried in air at 363 K for
2
5 times excess relative to stoichiometric quantities for the con-
4 h. Chemical analysis was performed by inductive coupling
version of 2,4-DCP/2-CP and phenol, respectively. In a series
of blank tests, the passage of each reactant in a stream of
hydrogen through the empty reactor, i.e. in the absence of cat-
alyst, did not result in any detectable conversion. Moreover,
reaction of the chloroarenes (as a methanolic or aqueous feed)
over silica alone was not accompanied by any detectable
hydrodechlorination and/or hydrogenation: the supported
Ni phase is essential for hydrogen addition/scission. The cata-
lytic measurements were made at W/F values in the range
plasma at the CNRS Center of Chemical Analysis (Vernaison,
France). The catalyst precursor was activated by heating (7.5
K min ) in 5% v/v H /Ar (10, 50 and 100 cm min , labeled
DPNi-II, -III and -IV, repectively) to 1173 K, cooling to the
reaction temperature.
ꢀ
1
3
ꢀ1
2
A
1.5% w/w Ni loading on SiO
2
(Sigma-Aldrich,
2
ꢀ1
3
ꢀ1
SBET ¼ 192 m g , pore volume ¼ 1.73 cm g , average pore
˚
diameter ¼ 198 A) was also prepared by deposition-precipita-
tion (labeled DPNi-I) and reduced using the same procedure as
that employed for the INinitrate samples: full preparation con-
ditions and reduction details/characterization are available
ꢀ
1
2
2–2072 g mol h where W is the weight of activated catalyst
and F is the inlet molar flow rate of aromatic. Each catalytic
run was repeated (up to six times) using different samples from
the same batch of catalyst and the measured rates did
not deviate by more than ꢁ7%.
3
5
elsewhere. The nickel nitrate used for INinitrate and DPNi
preparation was supplied by Aldrich (purity > 99%).
The reactor effluent was frozen in a liquid nitrogen trap
for subsequent analysis which was made using a Perkin-
Elmer Auto System XL chromatograph equipped with a
Impregnation with nickel ethanediamine (INi ). A 5.5, 10.8
en
and 20.9% w/w Ni loading (labeled INien-I, -II and -III,
respectively) was generated by impregnation of Spherosil
split/splitless injector and
a flame ionization detector,
XOA400 with aqueous solutions of Ni(en)
by addition of ethanediamine to an aqueous solution of nickel
nitrate (3 en per Ni). The precursors were reduced at 7.5 K
3
(NO
3
)
2
obtained
employing a DB-1 (50 m ꢃ 0.20 mm i.d., 0.33 mm) capillary
column (J&W Scientific). The overall level of hydrogena-
tion/hydrodechlorination was converted to mol% conversion
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931
1925

View Article Online
Table 1 Nickel content and the average Ni particle size (dNi) associated with each catalyst under the stated reduction conditions
Catalyst
% Ni w/w
Reduction conditions
dNi/nm
ꢀ
1
3
10 K min in 60 cm min
ꢀ1
ꢀ1
ꢀ1
ꢀ1
ꢀ1
ꢀ1
INinitrate-I
INinitrate-II
INinitrate-III
INinitrate-IV
INinitrate-V
DPNi-I
0.7
2.8
4.5
6.2
9.6
1.5
24
H
2
H
2
H
2
H
2
H
2
H
2
to 723 K (t ¼ 16 h)
to 723 K (t ¼ 16 h)
to 723 K (t ¼ 16 h)
to 723 K (t ¼ 16 h)
to 723 K (t ¼ 16 h)
to 723 K (t ¼ 16 h)
4.5
6.7
8.1
9.8
ꢀ
1
3
10 K min in 60 cm min
ꢀ
1
3
10 K min in 60 cm min
ꢀ
1
3
10 K min in 60 cm min
ꢀ
1
3
10 K min in 60 cm min
16.8
1.4
7.8
4.2
3.4
2.6
2.2
2.7
ꢀ
ꢀ
1
3
10 K min in 60 cm min
1
3
ꢀ1
ꢀ1
DPNi-II
7.5 K min in 10 cm min 5% v/v H
2
/Ar to 1173 K
/Ar to 1173 K
7.5 K min in 100 cm min 5% v/v H /Ar to 1173 K
ꢀ
1
3
DPNi-III
DPNi-IV
INien-I
24
24
7.5 K min in 50 cm min 5% v/v H
2
ꢀ ꢀ1
1
3
2
ꢀ
1
3
ꢀ1
5.5
10.8
20.9
7.5 K min in 25 cm min 5% v/v H
2
2
2
/Ar to 973 K
/Ar to 973 K
/Ar to 973 K
ꢀ ꢀ1
1
3
INien-II
7.5 K min in 25 cm min 5% v/v H
ꢀ
1
3
7.5 K min in 25 cm min 5% v/v H
ꢀ1
INien-III
using detailed calibration plots for each feedstock. Quantita-
tive analysis was based on relative peak area with acetone as
solvent where analytical repeatability was better than ꢁ0.4%
and the detection limit typically corresponded to a feedstock
conversion less than 0.1 mol%. Qualitative analysis for the
presence of chlorine gas proved negative in every instance
and HCl was the only inorganic product. HCl quantitative
analysis (to ensure a chlorine mass balance) was performed
by passing the effluent gas through an aqueous NaOH (3.5–
Results and discussion
The Ni content of the catalyst precursors and the average Ni
particle diameter are given in Table 1 wherein the reduction
conditions are also summarised. The HRTEM derived Ni
particle size distribution is illustrated in Fig. 1, taking
INinitrate-III, INinitrate-V, DPNi-III and INien-III as representa-
tive catalysts. The INinitrate samples are characterised by a
broader size distribution when compared with either the DPNi
4
3
ꢀ
3
ꢀ3
8
.0 ꢃ 10 mol dm , kept under constant agitation at ꢄ 300
or INien systems. It is well known that Ni/SiO
impregnation exhibits weak interaction with the substrate
2
prepared by
rpm) trap and monitoring continuously the pH change by
means of a Hanna HI Programmable Printing pH Bench-
Meter. The concentration of HCl generated was also measured
by potentiometric analysis of the scrubbing solution using a
3
5
and a greater tendency for Ni agglomeration/growth at ele-
vated temperatures, resulting in larger Ni particles under the
stated reduction conditions. The size distribution is clearly
wider for the INinitrate samples, especially in the case of the
higher loaded sample, leading to larger average particle sizes
3
Metrohm (Model 728) Autotitrys where AgNO served as
titrant with a combined Ag electrode; chlorine mass balance
was complete to better than ꢁ10%. All the data presented here
have been generated in the absence of any significant catalyst
deactivation.
3 3 2
(Table 1). In contrast, the use of Ni(en) (NO ) as the precur-
II
sor for impregnation, where Ni ions are coordinated with
strong chelating ligands, generates appreciably smaller Ni par-
ticles after reduction when compared with the Ni(NO ) pre-
cursor. Moreover, Ni particle size does not vary with Ni
loading in the case of the INien system. In the temperature pro-
grammed reduction (TPR) of INien , a single composite TPR
peak (at ca. 573 K) was recorded due to a concomitant nickel
4
3
The catalytic system has been shown to operate with neg-
ligible diffusion retardation of the reaction rate; effectiveness
factor (Z) > 0.99 at 573 K. Heat transport effects can also be
disregarded; the temperature differential between the catalyst
particles and bulk fluid phase was <1 K. It is generally true
that laboratory scale fixed-bed reactors approximate plug flow
3 2
0
reduction and hydrogenolysis of ethanediamine by the Ni that
is produced. The decomposition of ethanediamine may lead
4
4
36
pattern. In this case, the ratio of reactor diameter to catalyst
particle diameter ¼ 90, exceeding the lower limit of 10 set by
II
to the formation of an intermediate grafted Ni species, i.e.
4
5
Froment and Bischoff, satisfying the application of plug-flow
nickel that interacts strongly with the support, that inhibits
nickel migration and the formation of large particles. Ni/
SiO2 preparation by deposition-precipitation (DP) has been
shown elsewhere to generate smaller average Ni particle
conditions. The degree of hydrodechlorination (x ) is given by
Cl
½
Clꢅin ꢀ ½Clꢅ
out
xCl ¼
ð2Þ
½
Clꢅ
in
ꢀ
3
where [Cl] represents concentration (mol dm ); ‘in’ and ‘out’
refer to the inlet and outlet reactor streams, respectively. The
selectivity (as a percentage) with which 2-CP (S2-CP) is pro-
duced from 2,4-DCP can be calculated from
½
2-CPꢅ
out
S2 CPð%Þ ¼
ꢃ 100
ð3Þ
½
2; 4-DCPꢅ ꢀ ½2; 4-DCPꢅout
in
In this paper the composition of the product mixture is
expressed in terms of molar fractions (a) where, again taking
the production of 2-CP from 2,4-DCP
n2-CP
a2-CP ¼
ð4Þ
P
n2;4-DCP þ n
products
and n2,4-DCP and nSproducts (n2,4-DCPout +nSproducts ¼ n2,4-DCPin)
denote the number of moles in the effluent stream. The
system of differential equations that describes the reaction
kinetics was solved using Maple V (Maplesoft, Waterloo,
Canada), the commercial mathematical package.
Fig. 1 Nickel particle size distributions for the activated (a) INien-III
(
open bars), (b) DPNi-III (downward hatched bars), (c) INinitrate-III
(solid bars) and (d) INinitrate-V (upward hatched bars).
1
926
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931

View Article Online
3 2
diameters when compared with the less controlled Ni(NO )
3
5,37,38,46
impregnation route.
formation of a nickel phyllosilicate phase that is responsible
The DP synthesis results in the
3
for enhanced Ni dispersion. As demonstrated previously,
9
11
the ultimate Ni particle size is dependent on reduction condi-
tions. An increase in the reductant flow rate (from 10 to 100
3
ꢀ1
cm min ) lowers the reduction temperature of the supported
nickel phyllosilicate phase, facilitating the elimination of the
water associated with the samples and produced during reduc-
tion, which serves to limit Ni particle growth.
The gas phase hydrodechlorination of 2,4-DCP and 2-CP
over the range of catalysts listed in Table 1 yielded either the
partially (in the case of 2,4-DCP) or fully dechlorinated aro-
matic (phenol), i.e. the reaction was 100% selective in terms
of dechlorination. It is noteworthy that the transformation
of 2,4-DCP yielded 2-CP as the only intermediate product
and there was no evidence of any 4-CP in the effluent gas
stream. The absence of 4-CP is suggestive of the involvement
of steric constraints where hindrance due to the proximity of
the –OH group to the ortho-positioned Cl substituent renders
it less susceptible to cleavage than the para-positioned Cl. In
addition, the aromatic ring remained intact and there were
no detectable products (cyclohexanone, cyclohexanol or
benzene/cyclohexane) of phenol hydrogenation or hydro-
dehydroxylation. In contrast, the use of Pd- and Rh-based
catalysts to promote aromatic hydrodechlorination has been
shown to result in further hydrogenation of the dechlorinated
Fig. 2 Pseudo-first order relationships for the hydrodechlorination
of aqueous (open symbols) and methanolic (solid symbols) solutions
of 2-CP over INinitrate-I (X,S), INinitrate-III (K,L) and INinitrate-V
(N,/) at 423 K.
sion of aqueous and methanolic 2-CP solutions over three
representative INinitrate catalysts are shown in Fig. 2. Each line
passes through the origin, diagnostic of adherence to pseudo-
first order kinetics. The associated pseudo-first order rate con-
stants generated for a methanolic 2-CP or 2,4-DCP feed were
consistently lower than those delivered for the aqueous feed, as
was the degree of full dechlorination of 2,4-DCP (reflected by
the Sphenol values): see Table 2. The latter suggests some inhi-
biting effect by methanol that serves to limit chloroarene
dechlorination. Indeed, there was evidence of significant ani-
sole formation in the hydrogen treatment of methanolic solu-
tions of phenol (as feed), formed through a catalyzed
methylation step. The involvement of methanol in secondary
4
7,48
aromatic.
2
In the chloroarene/Ni/SiO interaction(s), the
aromatic ring resonance stability must not be disrupted to
any significant extent and the aromatic nucleus did not
undergo hydrogenation. We have opted for a standard
pseudo-first order kinetic treatment to probe the effect of Ni
particle size on the dechlorination rate, an approach that has
2
4,49–51
been applied successfully in a number of related studies.
Plug-flow operation, under steady state where hydrogen
was maintained far in excess, yields the following applicable
reactor/kinetic expression
reactions during gas phase aromatic hydrogenation over Ni/
53
2
SiO has been demonstrated elsewhere and attributed to
ꢀ
ꢁ
the catalytic action of the silica support where methanol is acti-
vated via the oxygen lone pair interactions. Methanol/support
interactions must serve to inhibit the hydrodechlorination step.
The lower dechlorination rate associated with the di-chlori-
nated reactant (see Table 2), regardless of the solvent, is indi-
cative of an electrophilic mechanism where the presence of
the second Cl substituent deactivates the ring for electro-
philic attack via an inductive effect that decreases the electron
density of the aromatic ring, i.e. destabilises the cationic tran-
W
lnð1 ꢀ xClÞꢀ ¼ k
1
ð5Þ
FCl
ꢀ
1
where FCl is the inlet molar chlorine feed rate (mol h ), xCl is
the fractional dechlorination of the feed over a given weight
W) of catalyst; the parameter W/FCl (units, g h mol(Cl) )
has the physical significance of contact time. From a consid-
eration of gas phase reaction equilibrium constants, the
dechlorination step(s) under the stated reaction conditions
can be taken to be irreversible. The linear relationships
ꢀ
1
(
5
2
sition state. Similar electrophilic effects have been reported
in related heterogeneous catalytic dechlorinations.
ꢀ
1
20–22
between ln(1 ꢀ xCl
)
and W/FCl , generated for the conver-
The
Table 2 Pseudo-first order hydrodechlorination rate constants (k) for the treatment of methanolic and aqueous solutions of 2-CP and 2,4-DCP
ꢀ1
and phenol selectivities from 2,4-DCP (W/FCl ¼ 694 g molꢀ
1
h
) over representative Ni/SiO
2
catalysts: T ¼ 423 K; reduction conditions are
summarised in Table 1
5 ꢀ1 ꢀ1
0 k/mol(Cl) g h
1
2,4-DCP feed
Methanolic solution
2-CP
Aqueous solution
2-CP
Sphenol (%)
Catalyst
2,4-DCP
2,4-DCP
Methanolic solution
Aqueous solution
INinitrate-I
INinitrate-III
INinitrate-V
DPNi-I
15.3
17.6
27.1
14.2
11.1
12.9
12.6
9.4
5.4
7.1
10.1
3.4
5.0
4.0
4.6
3.2
3.4
3.9
20.6
23.6
40.8
20.3
16.2
17.4
19.5
14.9
15.8
17.8
7.7
8.7
14.5
5.7
8.1
6.3
7.4
4.8
5.9
6.2
31
43
55
35
41
40
41
40
36
36
44
54
63
42
51
48
49
50
48
44
DPNi-II
DPNi-III
DPNi-IV
INien-I
INien-II
11.3
12.1
INien-III
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931
1927

View Article Online
Table 3 Apparent activation energies (with 95% confidence limits)
associated with the hydrodechlorination of 2-CP and 2,4-DCP over
2
selected Ni/SiO catalysts
ꢀ1
DE /kJ mol
a
Catalyst
2-CP
2,4-DCP
INinitrate-I
DPNi-II
INien-II
52.4 ꢁ 2.7
54.4 ꢁ 1.1
55.5 ꢁ 2.4
68.1 ꢁ 1.9
65.4 ꢁ 2.2
65.6 ꢁ 2.4
2
-CP and 2,4-DCP over three selected Ni/SiO
2
catalysts aris-
ing from three different preparation methods, with different
Ni loadings and different particle sizes, is shown in Fig. 4 as
a
apparent Arrhenius plots. The higher DE values for the 2,4-
DCP reaction (Table 3) are consistent with a deactivating effect
of the second Cl substituent that serves to raise the energy bar-
Fig. 3 Specific 2-CP (open symbols) and 2,4-DCP (solid symbols)
hydrodechlorination rate constant (k) as a function of the average
Ni particle diameter (dNi) for reaction over INinitrate (X,S), DPNi
rier for hydrodechlorination. It should be noted that the DE
a
(
N,/) and INien (K,L): T ¼ 423 K; reaction data refer to aqueous
values did not deviate to any significant extent (within the
stated 95% confidence limits) for the reaction of either 2-CP
or 2,4-DCP over the three catalysts and the values generated
in this study are in accordance with those reported previously
for mono- and di-chlorophenol dechlorination over supported
solutions.
hydrodechlorination rate constants were converted to specific
values using eqn. (1) and are plotted as a function of average
Ni particle diameter in Fig. 3 where the degree of chlorine
removal from the dichlorinated arene was consistently lower.
It is immediately evident that the specific Cl removal rates
increase with increasing Ni particle diameter, with almost
two orders of magnitude difference in the specific rates. These
results clearly establish structure sensitivity for aromatic
hydrodechlorination over Ni dispersed on silica. The enhance-
ment in dechlorination with increasing diameter of Ni particles
2
2,52
Ni.
The hydrodechlorination of polychlorinated aromatics has
been shown elsewhere to generate a partially dechlorinated
2
4,52,54–56
product stream.
The dechlorination of 2,4-DCP can
proceed through the series/parallel scheme, shown in Fig. 5,
where 2-CP serves as a reactive intermediate and phenol is
the ultimate dechlorination product. Applying pseudo-first
order conditions at quasi-steady state for irreversible dechlor-
ination steps, three differential equations describe the system
supported on a common SiO substrate is certainly diagnostic
2
ꢀ
ꢁ
of some ensemble effect, as was noted for CFC transforma-
tions over Pd/Al O . Moreover, the support can have a sig-
d½2; 4-DCPꢅ
dW
2
5
F2;4-DCP
¼ ꢀðk1 þ k3Þ½2; 4-DCPꢅ
ð6Þ
ð7Þ
ð8Þ
2
3
nificant influence on metal site catalytic performance through
electronic interactions. Support-induced electronic effects
ꢀ
ꢁ
1
7
d½2-CPꢅ
dW
F2;4-DCP
¼ k1½2; 4-DCPꢅ ꢀ k2½2-CPꢅ
are considered to be negligible for larger (ca. 10–40 nm) metal-
lic particles. Smaller supported transition metal particles (< 3
1
ꢀ
ꢁ
d½phenolꢅ
nm) have a tendency to remain electron deficient and there are
many instances in the literature where reactivity is strongly
influenced by the electron density of a highly dispersed metal
F2;4-DCP
¼ k3½2; 4-DCPꢅ þ k2½2-CPꢅ
dW
Integration of these rate equations provides the following
dependence of product mol fraction (a) on contact time (W/
2
phase. Given the electronic-sensitive nature of hydrogenolysis
reactions, the latter must have some bearing on 2-CP/2,4-DCP
hydrodechlorination. An increase in reaction temperature
F2,4-DCP
)
ꢂ
ꢃ
W
(from 408 to 483 K) served to increase the hydrodechlorination
rate and the temperature dependence for the conversion of
a2;4-DCP ¼ exp ꢀðk1 þ k3Þ
ð9Þ
F2;4-DCP
ꢀ
ꢁꢀ
ꢂ
ꢃ
k1
W
a2-CP ¼
exp ꢀðk1 þ k3Þ
k2 ꢀ ðk1 þ k3Þ
F2;4-DCP
ꢂ
ꢃꢁ
W
9
ꢀ exp ꢀk2
ð10Þ
ð11Þ
F2;4-DCP
aphenol ¼ 1 ꢀ a2;4ꢀDCP ꢀ a2ꢀCP
Taking the experimentally determined a values at various
contact times, the resultant non-linear equations were
1solved to yield values for the stepwise rate constants. The
applicability of this scheme can be assessed in Fig. 6 wherein
Fig. 4 Apparent Arrhenius plots for the hydrodechlorination of 2-
CP over INinitrate-I (S), DPNi-II (/) and INien-II (L) and the hydro-
dechlorination of 2,4-DCP over INinitrate-I (X), DPNi-II (N) and
INien-II (K): reaction data refer to aqueous solutions.
Fig. 5 Series/parallel reaction scheme for the dechlorination of
2,4-DCP to phenol.
1
928
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931

View Article Online
Fig. 8 Variation of 2-CP selectivity (S2-CP) with fractional 2,4-DCP
conversion (x2,4-DCP) for the hydrodechlorination of aqueous (open
symbols) and methanolic (solid symbols) solutions over INinitrate-I
Fig. 6 Experimental (symbols) and calculated (lines: based on a ser-
ies/parallel scheme) 2-CP product molar compositions (a2-CP) resulting
from the hydrodechlorination of aqueous solution of 2,4-DCP over
INinitrate-V (S), DPNi-IV (/) and INien-I (L) at 423 K.
(X, S), INinitrate-III (K,L) and INinitrate-V (N,/) at 423 K.
particles used to promote dechlorination of a methanolic feed.
The selectivity/activity profiles converge to a common relation-
ship at x2,4-DCP > 0.2 and phenol formation predominated.
As an addendum to the study of hydrodechlorination struc-
ture sensitivity, we also considered the effect of Ni loading and
preparation on phenol hydrogenation activity. Given the
appreciable catalysed methylation step (noted above), the phe-
nol hydrogenation measurements were limited to an aqueous
feedstock. Moreover, at T > 473 K there was significant ben-
zene formation via a catalysed hydrodehydroxylation. An
assessment of the intrinsic hydrogenation activity of the Ni/
representative experimental and calculated product composi-
tions are presented: the agreement is uniformly good. The
value of the rate constant ratio k /k signifies the relative
1 3
importance of the two possible hydrodechlorination pathways,
stepwise (responsible for selectivity in terms of 2-CP) and con-
certed (one-step reaction), respectively. The higher specific
activity associated with the larger Ni particles (Fig. 3) is man-
1 3
ifested by the appreciably lower k /k values where a complete
dechlorination is promoted to a far greater extent, as shown in
Fig. 7. The removal of both chlorines in one step (concerted
pathway) can occur via a concomitant attack of surface hydro-
gen at both C–Cl bonds with a subsequent desorption of the
aromatic product. The results generated in this study suggest
that this concerted Cl removal is facilitated by larger sup-
ported Ni particles. The relationship between 2-CP selectivity
2
SiO catalysts was accordingly limited to an aqueous phenol
feed (with vaporisation prior to reaction) at T < 473 K. While
there were no observed phenol hydrogenation by-products
during the hydrogen treatment of 2-CP or 2,4-DCP, Ni/SiO2
was active in the hydrogenation of phenol as feedstock. The
latter suggests quite different surface requirements for the
direct hydrogenation of phenol compared with an ‘‘indirect’’
hydrogenation of any phenol that is generated via chlorophe-
(S2-CP) and overall 2,4-DCP conversion (x2,4-DCP) is shown
in Fig. 8 for three INinitrate catalysts (I, III and V) with differ-
ent Ni loadings and particle sizes (Table 1). An increase in con-
version was accompanied by a decrease in S2-CP as is to be
expected where 2-CP is a reactive intermediate in a series/par-
allel scheme. At x2,4-DCP values lower than ca. 0.2, the activity/
selectivity relationship is sensitive to both Ni particle size and
the nature of the solvent where higher S2-CP values (at a given
57
nol hydrodechlorination. Indeed, we have shown elsewhere
that the reactive hydrogen associated with Ni/SiO that parti-
2
cipates in hydrogen scission is quite distinct from that involved
in hydrogen addition with the possible involvement of spillover
hydrogen (on the support and metal/support interface) in the
former case. It is instructive to note that the reduced Ni/SiO2
samples (active for direct phenol hydrogenation), once con-
tacted with 2-CP or 2,4-DCP (where Cl/Ni mol ratio > 1),
only exhibited a residual phenol hydrogenation activity
x2,4-DCP) are associated with smaller average Ni diameter
(xphenol < 0.05) on a re-introduction of the same phenol feed.
This effect was observed regardless of Ni loading or method
of preparation. The surface conditions which prevail during
hydrodechlorination involve an appreciable residual HCl com-
4
3,58
ponent on the catalyst surface.
This surface HCl compo-
nent does not induce any short term loss of HDC activity
but serves to severely inhibit phenol hydrogenation. Work
is on-going to probe the nature of surface Cl/Ni interactions
that poison hydrogenation sites while hydrogenolysis is
unaffected.
Pseudo-first order kinetics were also applicable to the phenol
consumption raw data and specific rate constants delivered by
a range of Ni/SiO catalysts are recorded in Table 4. Cyclo-
2
hexanone and cyclohexanol were the only observed reaction
products in the gas phase reaction of aqueous phenol solutions
at 423 K. A sequential hydrogenation of phenol has been
Fig. 7 Effect of the average Ni particle size (dNi) on the extent of
stepwise (k ) vs. concerted (k ) Cl removal in the hydrodechlorination
2
6,27,59,60
noted in a number of instances
with cyclohexanone
1
3
acting as a reactive intermediate and cyclohexanol is the ulti-
mate product. Phenol hydrogenation did not exhibit the same
of aqueous solutions of 2,4-DCP over INinitrate (S), DPNi (/) and
INien (L) at 423 K.
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931
1929

View Article Online
Table 4 Specific phenol hydrogenation pseudo first order rate con-
stant (k) and associated selectivity with respect to cyclohexanone pro-
catalysts with 2-CP or 2,4-DCP (with associated hydrodechlor-
ination and HCl formation) serves to severely inhibit phenol
hydrogenation although the hydrodechlorination reaction is
unaffected. The results point to different active site require-
ments for aromatic hydrogenation and hydrodechlorination
reactions.
duction (S
K; reduction conditions are summarised in Table 1
C
=
O
) for reaction over selected Ni/SiO
2
catalysts: T ¼ 423
5
0 k mol h
ꢀ1
ꢀ1 ꢀ1
)
1
(m g(Ni)
2
Catalyst
C O
=
S (%)
INinitrate-I
INinitrate-III
INinitrate-IV
DPNi-I
11.4
14.1
15.2
5.6
2.2
4.0
5.4
2.8
4.2
3.4
4.0
4.4
5.1
3.6
Acknowledgements
The assistance of E.-J. Shin, C. Menini and A. Dozier with cat-
alyst characterization/kinetic measurements is acknowledged.
This work was supported in part by the National Science
Foundation through Grant CTS-0218591.
DPNi-II
DPNi-III
DPNi-IV
INien-I
12.6
11.7
10.5
6.8
INien-II
6.4
7.9
INien-III
References
degree of structure sensitivity that was evident for hydro-
dechlorination but there was a discernible decrease in specific
rate for reaction over smaller average Ni particles (<ca. 3
nm for the DPNi-I and INien samples). However, reaction
selectivity was essentially insensitive to Ni loading/dispersion
and cyclohexanol was the predominant product (Table 4). This
supports the findings given above, i.e. the active sites that par-
ticipate in aromatic hydrogenation differ from those involved
in aromatic hydrodechlorination.
1
2
3
G. C. Bond, Chem. Soc. Rev., 1991, 20, 441.
M. Che and C. O. Bennett, Adv. Catal., 1989, 36, 55.
R. Takahashi, S. Sato, T. Sodesawa, M. Kato, S. Takenaka and
S. Yoshida, J. Catal., 2001, 204, 259.
K. Hadjivanov, M. Mihaylov, D. Klissurski, P. Stepanov,
N. Abadjieva, E. Vassileva and L. Mintchev, J. Catal., 1999,
4
1
85, 314.
5
6
7
8
9
M. Kermarec, J.-Y. Carriat, P. Burattin, M. Che and A.
Decarreau, J. Phys. Chem., 1994, 98, 12 008.
A. Gil, A. Dias, L. M. Gandia and M. Montes, Appl. Catal. A,
1
994, 109, 167.
P. Lesage, O. Clause, P. Moral, D. Didillon, S. P. Candy and
J. M. Basset, J. Catal., 1995, 155, 238.
J. T. Richardson, R. J. Dubus, J. G. Crump, P. Desai, U.
Osterwalder and T. S. Cale, Stud. Surf. Sci. Catal., 1982, 16, 311.
J. Loosdrecht, A. M. van de Kraan, A. J. van der Dillen and J. W.
Geus, J. Catal., 1997, 170, 217.
Conclusions
The data presented in this paper support the following conclu-
sions.
(
i) Average Ni particle size associated with Ni/SiO
4% w/w Ni) prepared by impregnation with nickel nitrate,
nickel ethanediamine and by deposition-precipitation spanned
the range 1.4–16.8 nm. Ni/SiO , irrespective of Ni loading,
2
(0.7–
10 C. Louis, Z. X. Cheng and M. Che, J. Phys. Chem., 1993, 97,
703.
P. Burattin, M. Che and C. Louis, J. Phys. Chem. B, 2000, 104,
0 482.
5
2
1
1
2
1
2
1
R. Burch, in Catalysis, ed. G. C. Bond and G. Webb, The Royal
Society of Chemistry, London, 1985, p. 149.
preparation and reduction conditions is 100% selective in pro-
moting mono- and di-chlorophenol hydrodechlorination.
13 M. A. Vannice, Catal. Rev.-Sci. Eng., 1976, 14, 153.
14 M. A. Keane, J. Catal., 1997, 166, 347.
(
ii) Chlorine removal from 2,4-DCP generated a mixture of
1
5
A. Stanislaus and B. H. Cooper, Catal. Rev.-Sci. Eng., 1994,
6, 75.
M. A. Keane and P. M. Patterson, Ind. Eng. Chem. Res., 1999, 38,
295.
partially dechlorinated 2-CP as well as the fully dechlorinated
phenol where the absence of any significant 4-CP formation
can be attributed to steric effects. Fractional Cl removal from
3
1
6
1
2
,4-DCP was lower than that for 2-CP and the associated
17 A. Yu. Stakheev and L. M. Kustov, Appl. Catal. A, 1999, 188, 3.
18 J. K. Fawell and S. Hunt, Environmental Toxicology, Organic Pol-
lutants, Ellis Horwood, Chichester, 1988.
apparent activation energy was appreciably higher, suggesting
that the electron withdrawing properties of chlorine renders
the total Cl complement less susceptible to hydrogen cleavage.
The DE values obtained for 2,4-DCP hydrodechlorination do
not depend on catalyst preparation/reduction procedure.
(
19
20
21
D. I. Kim and D. T. Allen, Ind. Eng. Chem. Res., 1997, 36, 3019.
C. Chon and D. T. Allen, AIChE J., 1991, 37, 1730.
A. R. Suzdorf, S. V. Morozov, N. N. Anshits, S. I. Tsiganova,
S. I. and A. G. Anshits, Catal. Lett., 1994, 29, 49.
E.-J. Shin and M. A. Keane, Appl. Catal. B, 1998, 18, 241.
a
iii) Hydrodechlorination performance was affected by the
2
2
nature of the feedstock solvent in that the dechlorination of
a methanolic 2-CP/2,4-DCP feed was significantly lower than
that of an aqueous feed under identical reaction conditions, an
effect that is attributed to the methanol/support interaction.
23 E.-J. Shin and M. A. Keane, Catal. Lett., 1999, 58, 141.
24 E.-J. Shin and M. A. Keane, J. Chem. Technol. Biotechnol., 2000,
7
5, 159.
W. Juszczyk, A. Mallinowski and Z. Karpinski, Appl. Catal. A,
998, 166, 311.
2
5
1
(
iv) We have adopted the notion of structure sensitivity in
2
2
6
7
E.-J. Shin and M. A. Keane, Ind. Eng. Chem. Res., 2000, 39, 883.
E.-J. Shin and M. A. Keane, J. Catal., 1998, 173, 450.
the broad sense where metal particle size effects can be linked
to the metal ensemble size with the associated intrinsic electro-
nic properties and any perturbation(s) due to metal/support
interactions. The hydrodechlorination of 2-CP and 2,4-DCP
over Ni/SiO can be explicitly designated as structure sensitive
2
where the specific rate and degree of complete dechlorination
28 V. Vishwanathan, N. Mahata and M. A. Keane, React. Kinet.
Catal. Lett., 2001, 72, 297.
2
3
3
9
0
1
S. Narayanan and G. Sreekanth, Catal. Lett., 1993, 51, 449.
S. Narayanan and K. Krishna, Appl. Catal. A, 1996, 147, L253.
N. Mahata, K. V. Raghavan, V. Vishwanathan, C. Park and
M. A. Keane, Phys. Chem. Chem. Phys., 2001, 3, 2712.
G. Neri, A. M. Visco, A. Donato, C. Milone, M. Malentacchi and
G. Gubitosa, Appl. Catal. A, 1994, 110, 49.
(
in the case of 2,4-DCP) are elevated with increasing average
Ni particle diameter.
v) Such a distinct structure sensitivity does not extend to the
3
2
(
33 S. T. Srinivas, L. J. Lakshmi and P. K. Rao, Appl. Catal. A, 1994,
10, 167.
1
hydrogenation of phenol over the same catalysts although
there is a decided reduction in phenol consumption over smal-
ler Ni particles (<ca. 3 nm average diameter): selectivity is
insensitive to Ni loading/particle size and complete hydro-
genation to cyclohexanol predominates. A pre-contact of the
3
4
A. K. Talukdar, K. G. Bhattacharyya and S. Sivasanker, Appl.
Catal. A, 1993, 96, 229.
3
3
5
6
M. A. Keane, Can. J. Chem., 1994, 72, 372.
Z. X. Cheng, C. Louis and M. Che, Stud. Surf. Sci. Catal., 1995,
91, 1027.
1
930
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931

View Article Online
37
38
39
P. Burattin, M. Che and C. Louis, J. Phys. Chem. B, 1997, 101,
060.
P. Burattin, M. Che and C. Louis, J. Phys. Chem. B, 1998, 102,
722.
P. Burattin, M. Che and C. Louis, J. Phys. Chem. B, 1999, 103,
171.
48 P. Bodnariuk, B. Coq, G. Ferrat and F. Figueras, J. Catal., 1989,
116, 459.
49 J. Frimmel and M. Zdra zˇ il, J. Catal., 1997, 167, 286.
50 B. F. Hagh and D. T. Allen, Chem. Eng. Sci., 1990, 45, 2695.
51 F. Gioia, E. J. Gallagher and V. Famiglietti, J. Hazard. Mater.,
1994, 38, 277.
7
2
6
4
4
0
1
B. Coughlan and M. A. Keane, Zeolites, 1991, 11, 2.
C. Park and M. A. Keane, J. Colloid Interface Sci., 2002,
250, 37.
M. A. Armendia, V. Borau, I. M. Garcia, C. Jimenez, F. Lafont,
A. Marinas, J. M. Marinas and F. J. Urbano, J. Catal., 1999,
52 E.-J. Shin and M. A. Keane, Chem. Eng. Sci., 1999, 54, 1109.
53 M. A. Keane, J. Mol. Catal. A: Chem., 1999, 138, 197.
54 B. F. Hagh and D. Allen, AIChE J., 1990, 36, 773.
55 V. A. Yakovlev, V. V. Terskikh, V. I. Simagina and V. A.
Likholobov, J. Mol. Catal. A: Chem., 2000, 153, 231.
56 F. Murena, V. Famiglietti and F. Gioia, Environ. Prog., 1993,
12, 233.
57 E.-J. Shin, A. Spiller, G. Tavoularis and M. A. Keane, Phys.
Chem. Chem. Phys., 1999, 1, 3173.
58 C. Park, C. Menini, J.-L. Valverde and M. A. Keane, J. Catal.,
2002, 211, 451.
4
2
1
87, 392.
G. Tavoularis and M. A. Keane, J. Chem. Technol. Biotechnol.,
999, 74, 60.
4
4
4
3
4
5
1
O. Levenspeil, Chemical Reaction Engineering, John Wiley, New
York, 1972.
G. Froment and K. B. Bischoff, Chemical Reactor Analysis and
Design, John Wiley, New York, 1990.
J. W. E. Coenen, Appl. Catal., 1991, 75, 193.
B. Coq, G. Ferrat and F. Figueras, J. Catal., 1986, 101, 434.
59 Y. Z. Chen, C. W. Liaw and L. I. Lee, Appl. Catal A, 1999, 177, 1.
60 S. G. Shore, E. Ding, C. Park and M. A. Keane, Catal. Commun.,
2002, 3, 77.
4
4
6
7
Phys. Chem. Chem. Phys., 2003, 5, 1924–1931
1931