Control of catalytic debenzylation and dehalogenation reactions during liquid-phase reduction by H2

Article abstract of DOI:10.1016/j.jcat.2005.11.017

Heterogeneous catalysis, which is being used more frequently in the fine chemicals industry, does not always provide the desired selectivity because various functional groups become labile to reduction by hydrogen on a noble metal under similar reaction conditions. To study the selectivity of debenzylation versus dechlorination in an aromatic compound, the kinetics of the reduction of 4-chloro-N,N-dibenzylaniline on supported Pd catalysts in a buffered solvent under hydrogen reduction conditions were investigated. The turnover frequency on a Pd/C catalyst was 2-40 times higher than that on Pd dispersed on SiO₂, TiO₂, and Al₂O₃. The effects of hydrogen pressure, organic substrate concentration, temperature, solvent, and catalyst support on the selectivity were not great at a pH of 5.4, whereas the effect of acid or base modifiers was very significant. Only dechlorination reactions occurred under basic conditions, and the use of an acid confirmed a preferential promotion of debenzylation reactions; consequently, as the pH decreased from 12 to 0.1, the selectivity for dechlorination decreased from essentially 100% to almost 0. Using a triethylamine/acetic acid buffer system to control pH, it was determined that the inflection point in selectivity corresponds to a pH value equal to the pKa of the benzyl-protected amine (4.5 in this case). Consequently, one important finding is that the dechlorination reaction can occur at high rates even under acidic conditions, as long as the pH of the reaction mixture is greater than the pKa of the protected amine. Furthermore, generation of an acidic product like HCl during a dechlorination reaction can decrease the system pH and markedly alter selectivity unless a buffer is present. For reactions carried out at a pH of 5.4, the initial rate of organic substrate consumption was modeled by a classical Langmuir-Hinshelwood sequence, with the assumption of two types of active sites, one site to adsorb H atoms and the other to adsorb the organic substrate, yielding a slightly better fit to the data. The Weisz-Prater criterion was applied to verify the absence of mass transfer effects.

Full text of DOI:10.1016/j.jcat.2005.11.017

Journal of Catalysis 237 (2006) 349–358
www.elsevier.com/locate/jcat
Control of catalytic debenzylation and dehalogenation reactions during
liquid-phase reduction by H2
Adrian David, M. Albert Vannice ∗
Department of Chemical Engineering, Pennsylvania State University, University Park, PA 16802, USA
Received 8 September 2005; accepted 10 November 2005
Abstract
Heterogeneous catalysis, which is being used more frequently in the fine chemicals industry, does not always provide the desired selectivity
because various functional groups become labile to reduction by hydrogen on a noble metal under similar reaction conditions. To study the
selectivity of debenzylation versus dechlorination in an aromatic compound, the kinetics of the reduction of 4-chloro-N,N-dibenzylaniline on
supported Pd catalysts in a buffered solvent under hydrogen reduction conditions were investigated. The turnover frequency on a Pd/C catalyst
was 2–40 times higher than that on Pd dispersed on SiO , TiO , and Al O . The effects of hydrogen pressure, organic substrate concentration,
2
2
2 3
temperature, solvent, and catalyst support on the selectivity were not great at a pH of 5.4, whereas the effect of acid or base modifiers was
very significant. Only dechlorination reactions occurred under basic conditions, and the use of an acid confirmed a preferential promotion of
debenzylation reactions; consequently, as the pH decreased from 12 to 0.1, the selectivity for dechlorination decreased from essentially 100% to
almost 0. Using a triethylamine/acetic acid buffer system to control pH, it was determined that the inflection point in selectivity corresponds to a
pH value equal to the pK of the benzyl-protected amine (4.5 in this case). Consequently, one important finding is that the dechlorination reaction
a
can occur at high rates even under acidic conditions, as long as the pH of the reaction mixture is greater than the pK of the protected amine.
a
Furthermore, generation of an acidic product like HCl during a dechlorination reaction can decrease the system pH and markedly alter selectivity
unless a buffer is present. For reactions carried out at a pH of 5.4, the initial rate of organic substrate consumption was modeled by a classical
Langmuir–Hinshelwood sequence, with the assumption of two types of active sites, one site to adsorb H atoms and the other to adsorb the organic
substrate, yielding a slightly better fit to the data. The Weisz–Prater criterion was applied to verify the absence of mass transfer effects.
2005 Elsevier Inc. All rights reserved.
Keywords: Debenzylation; Dechlorination; Chemoselectivity; Pd/C; Reaction kinetics; Effect of pH
1. Introduction
transformation sequence under conditions that will not affect
other groups in the substrate.
For example, free amines contain a –N group that is labile
under a variety of reaction conditions and thus needs to be
protected to ensure its integrity over the course of a synthesis
involving multiple steps. Benzyl groups are often the protective
group of choice for amines, and at the end of the transforma-
tion, the free amine groups can be recreated by removing the
benzyl protective group from the N atom by various reactions
[1–6]. N-debenzylation under catalytic reduction conditions us-
ing H2 [6] is an attractive alternative solution when the use of
strong acids like p-toluenesulfonic acid [2] or strong bases like
Li di-isopropylamide [3] is not desirable; however, catalytic re-
duction conditions can also lead to undesired side reactions,
such as dehalogenation and double-bond reduction [7,8].
The manipulation of functional groups is essential in multi-
step organic syntheses. To avoid side reactions during a chem-
ical transformation involving a multifunctional organic com-
pound, some functionalities may need to be protected while
carrying out reactions directed at other functional groups in the
substrate (reactant). Such a protective group must attach specif-
ically to the functional group for which it is intended, must be
inert under the reaction conditions directed at other function-
alities, and must be relatively easy to detach at the end of the
*
Corresponding author.
E-mail address: mavche@engr.psu.edu (M.A. Vannice).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.11.017
2005 Elsevier Inc. All rights reserved.

350
A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
Fig. 1. Reaction network for the reduction of 4-chloro-N,N-dibenzylaniline (CNNDBA) to aniline.
Selectivity control is also important because in some in-
only makes it difficult to determine the effect of other reaction
parameters on selectivity, but also complicates a precise study
of the effect of pH on these reactions. To overcome this diffi-
culty, we used buffered systems to maintain different pH levels
and examine the selectivity in greater detail. We also investi-
gated the influence of several solvents and of different supports
for palladium.
stances the dehalogenation reaction is desired and debenzyla-
tion must be suppressed. Studer and Blaser pointed out that
“little is known about the selective removal of benzyl groups
in molecules containing aromatic halogens, and reports of ben-
zylated anilines are scarce and often incomplete” [9]. Thus
they used 4-chloro-N,N-dibenzylaniline—a compound subject
to both N-debenzylation and dehalogenation—to examine the
factors determining selectivity to debenzylation products [9].
Reduction under one atmosphere of H2 in the presence of a
Pd/C catalyst results in the reaction network illustrated in Fig. 1.
Studer and Blaser demonstrated that debenzylation is favored
in the presence of acids, whereas dehalogenation is favored in
the presence of bases. The choice of catalyst had no signifi-
cant effect on selectivity, but carbon-supported Pd was found to
be more effective for debenzylation compared with oxidic Pd
or Pd(OH)2. Polar solvents appeared to favor dehalogenation,
whereas nonpolar solvents seemed to favor debenzylation [9].
Our study was undertaken to obtain additional knowledge
about the factors that channel the reaction pathway to either
debenzylation or dehalogenation, with an emphasis on the role
of pH. The substrate used by Studer and Blaser (4-chloro-
N,N-dibenzylaniline [CNNDBA]) was again chosen as a model
compound. In their original work, Studer and Blaser stopped
short of developing a kinetic model based on reaction rates
measured at different temperatures, pressures, and reactant con-
centrations; therefore, another objective of our study was to
acquire appropriate rate data over a range of reaction condi-
tions to enable us to propose a kinetic model to describe the
disappearance of the CNNDBA substrate. It is well known that
acidic conditions promote debenzylation [9–11], whereas ba-
sic conditions inhibit debenzylation and favor dechlorination
[9,12,13]. A protonated nitrogen atom is positively charged and
electron-attracting, and thus it is expected to behave differently
than a neutral amine nitrogen atom, which can act as an electron
donor. When starting with a neutral reaction mixture (pH = 7),
the pH can drop rapidly if the parallel dehalogenation reaction
is significant, because this reaction generates HCl in situ, which
can then increase the rate of debenzylation. This process not
2. Experimental
2.1. Materials
Methanol (99.8%), tetrahydrofuran (99.9%, inhibitor-free),
acetic acid (glacial, 99.8%), triethylamine (99.5%), concen-
trated sulfuric acid (98%), hydrochloric acid (37% in wa-
ter), chloroform (99.8%, stabilized with ethanol), sodium car-
bonate (powder, 99.5%), sodium sulfate (granular, 99+%),
lithium hydroxide monohydrate (98+%), aniline (99%), 4-
chloroaniline (99+%), benzyl chloride (99+%), tetradecane
(99+%), and aniline (99%) were all purchased from Aldrich.
N-benzylaniline (98%, GC) and N,N-dibenzylaniline (99%,
GC) were purchased from TCI America. The 5% Pd/C cata-
lyst was purchased from Johnson Matthey (A102038-5, ∼50%
wet). Its mean particle diameter was 30 µm and the pore size
distribution was reported to be <2 nm, 0.3 ml/g; 2–10 nm,
0.2 ml/g; 10–20 nm, 0.1 ml/g; 20–30 nm, 0.05 ml/g; 30–
100 nm, 0.1 ml/g; 100–500 nm, 0.35 ml/g; and 500–1000 nm,
0.3 ml/g. The Pd/Al2
O3, Pd/SiO2, and Pd/TiO2 catalysts were
prepared in our laboratory as described previously [14,15]. Hy-
drogen (99.999%) and argon (99.999%) were purchased from
MG Industries. The CNNDBA substrate was prepared from 4-
chloroaniline and benzyl chloride according to a procedure de-
veloped by Blaser and Studer [9], as described previously [14].
Characterization of the white CNNDBA precipitate gave a
melting point of 377–378 K (377–378 K [9]), and its 1H NMR
spectrum in acetone d-6 gave the following chemical shifts (δ,
in ppm): 4.72 (s,4H), 6.71 (d,2H), 7.08 (d,2H), and 7.22–7.36
(m,10H).

A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
351
2.2. Reaction apparatus
was produced; therefore, even though the H2SO4/acetic acid
mixture was not actually buffering at a pH of 0.1, the pH of
the reaction mixture did not drift during the hydrogen addi-
tion.
Reduction of CNNDBA in H2 was conducted in a 500-ml
jacketed glass autoclave (Ace Glass) provided with a turbine ag-
itator operated with an electrical stirrer motor (IKA-Eurostar),
a thermometer, a pressure gauge, and a sampler [14]. The re-
actor temperature was controlled by a water bath provided
with a recirculation pump (Fisher Scientific). The reactor was
connected to a house vacuum system and to pressure regula-
tors attached to hydrogen and argon cylinders. The gases were
passed through a drying column (Supelco) before entering a gas
flow meter (model GFM 171 TOT 1-10, Aalborg Instruments)
equipped with a flow totalizer.
Each sample was transferred via a filter-pipette to a vial
containing 10 ml of chloroform and 10 ml of 5% aqueous
sodium carbonate and shaken vigorously, and then the lay-
ers were allowed to settle. About 3 ml of the bottom chlo-
roform phase was pipetted out, diluted 50% with chloroform,
and dried over sodium sulfate before being injected directly
into the GC column. (Important note: If the CNNDBA sub-
strate and the nitrogen-containing products are not fully de-
protonated by, e.g., treating the samples with sodium carbon-
ate, then the results of the GC analyses are inaccurate because
the protonated fraction (RNH3+, R2NH2+, R3NH+) will not
elute through the column, due to the strong interaction of the
ionic compounds with the highly polar support.) An internal
standard method using tetradecane was used to determine a
response factor (RFi), for each component i in the reaction
network. The response factors relative to the tetradecane inter-
nal standard (RFIS = 1) were as follows: RFCNNDBA = 1.33
for 4-chloro-N,N-dibenzylaniline, RFNNDBA = 1.03 for N,N-
2.3. Reaction procedures
The reduction experiments were carried out in a semibatch
mode at constant pressure and temperature in the jacketed Ace
Glass autoclave. The solvent containing CNNDBA was charged
to the reactor, and the powder catalyst was subsequently added
under mild agitation. The system was sealed and placed un-
der an inert atmosphere by three consecutive cycles of evac-
uation followed by pressurization with Ar to 240 kPa. Then
hydrogen was introduced, and the reactor was pressurized three
times to 240 kPa in the absence of agitation. (Because of sup-
pressed mass transfer into the solvent when the stirrer was
idle, no hydrogen uptake was observed.) The system was then
brought to the desired H2 reaction pressure and allowed to sta-
bilize until the gas flow meter indicated no flow. The reaction
was started by rapidly raising the agitation rate to 700 RPM
(zero-time point), and the desired reaction temperature was
maintained using the water bath. H2 was flowed through the
gas meter and totalizer to maintain a constant H2 pressure
and to quantify consumption. Between four and six 3-ml sam-
ples were removed at different intervals, with the time and the
total hydrogen uptake recorded when each sample was with-
drawn.
dibenzylaniline, RFNBA = 1.11 for N-benzylaniline, RFCA
=
1.66 for 4-chloroaniline, and RFA = 1.21 for aniline. The re-
sponse factor for 4-chloro-N-benzylaniline (CNBA), which was
not available from a commercial source, was calculated theoret-
ically as 1.31 based on its molecular weight [16]. Programming
details with the H-P 5890 Series II chromatograph fitted with a
Rhino capillary column (15 m × 0.025 mm × 1 µm film thick-
ness) are provided elsewhere [14]. Mass balances (µmol CN-
ꢀ
NDBA reacted vs.
5%.
µmol products formed) agreed to within
2.4. Catalyst characterization
The concentration of surface Pd (Pds) atoms (i.e., the ac-
tive component in each catalyst) must be known not only to
calculate a turnover frequency (TOF; molecules reacted per
second per surface metal atom), but also to calculate the dis-
persion and average crystallite size of Pd. The Pds atoms were
counted both by hydrogen chemisorption using the method
of Benson et al. [17] and by carbon monoxide chemisorption
using the method of Yates and Sinfelt [18] in an experimen-
tal system described previously [19]. The 5% Pd/C catalyst,
which was used without pretreatment in the reduction of CN-
NDBA, was dried directly in the chemisorption cell at 393 K
under high vacuum (ca. 10−7 kPa) before CO isotherms were
obtained for total and reversible uptakes. To check the mea-
surement of Pds atoms, irreversible H2 adsorption at 298 K
was determined after a pretreatment consisting of 2 h at 573 K
under 50 cc/min of flowing H2, followed by a 30-min flush
with helium and then evacuation to 10−7 kPa. XRD line-
broadening measurements were also conducted with the 5%
Pd/C catalyst to calculate a volume-weighted average crystal-
lite size.
Because of the low solubility of the reactant in many or-
ganic solvents, reactions were conducted in mixtures contain-
ing tetrahydrofuran (THF), which is the best solvent for CN-
NDBA. The effects of solvent, catalyst support, catalyst pre-
treatment, and buffer pH on the reaction rate and selectivity
were then examined, and rate data (concentration vs. time)
were collected at different H2 pressures (115, 156, 194, and
263 kPa), temperatures (278, 298 and 330 K), and CNNDBA
starting concentrations (9.5, 19.0, 28.3, and 37.8 µmol/ml).
To minimize and control the effect of changing pH, reduc-
tion experiments were conducted in a THF solvent buffered
at pH levels of 12.0, 6.0, 5.2, 4.7, 3.5, and 0.1, as measured
by a pH meter (Corning model 450). All buffers with the ex-
ception of the pH = 0.1 system were prepared from triethyl-
amine and acetic acid such that they maintained the reaction
mixture within a narrow pH range ( 0.25 pH units) through-
out the course of the reaction [14]. The system with a pH
of 0.1 was prepared by adding a mixture of acetic acid and
H2SO4 to a solution of CNNDBA in THF. At low pH levels,
the dechlorination reaction was totally suppressed and no HCl

352
A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
Table 1
sumption along with the corresponding TOFs based on Pds
atoms obtained from the CO chemisorption data. Except for the
5% Pd/C catalyst, all catalysts used for the runs summarized
in Table 2 were pretreated in H2 at 573 K for 2 h, evacu-
ated, and then passivated in air at 300 K to create a layer of
chemisorbed oxygen before being used in the reactor. The low
rates with the oxide-supported catalysts are not due to mass
transport limitations, because the Weisz–Prater (W-P) numbers
were much smaller than those for the Pd/C catalyst, as dis-
cussed later. It is more likely that the low activities are due to
the absence of an in situ high-temperature reduction of the pas-
sivated catalysts before their use. At this pH, a low selectivity
for debenzylation is expected, and the uncertainty limits in the
pH control could account for much of the variations in selectiv-
ity because of the strong dependence on pH in this region (see
Fig. 5).
Irreversible uptakes on and dispersions for Pd catalysts
Catalyst Irreversible chemisorption
Dispersion
CO/Pd
µmol /g
µmol /g
H/Pd
cat
cat
H2
CO
2.0% Pd/Al O
1.8% Pd/TiO
2.0% Pd/SiO
78.2
46.7
15.2
43.9
43.4
24.9
11.4
27.9
0.42
0.28
0.081
0.093
0.13
0.46
0.29
0.12
0.12
0.10
2
3
2
2
2
5.0% Pd/SiO
5.0% Pd/C
a
a
b
61.4
23.9
As received after evacuation and drying.
After pretreatment in H at 573 K and evacuation.
b
2
Table 2
Initial rates, TOFs and selectivities for CNNDBA reduction in hydrogen over
Pd catalysts
Catalyst
Initial activity for
CNNDBA disappearance
Selectivity
(%)
a
−1
)
Rate (µmol/(s g ))
TOF (s
cat
3.3. Effect of solvent
2.0% Pd/Al O
0.41
2.37
1.26
2.62
11.2
0.005
0.051
0.082
0.060
0.182
14
12
20
26
7
2
3
1.8% Pd/TiO
2
2
2
It has been established that the presence of acids promotes
debenzylation, whereas the presence of bases favors dechlo-
rination [9,11,12]. Indeed, Fig. 2 indicates that the reaction
generates almost exclusively debenzylation products (CNBA
and CA) when sulfuric acid is present, and this composition
profile replicates the results reported for a Pd/C catalyst (ex-
periment #19) very nicely [9], with the only difference in re-
action conditions being a H2 pressure of 156 kPa instead of
101 kPa. The solubility of CNNDBA depends significantly
on the nature of the solvent, with the best solvents being
polar nonprotic solvents like tetrahydrofuran (THF) and ace-
tone, whereas polar protic solvents like methanol and acetic
acid provide only limited solubility. The availability of trans-
ferable protons in the liquid reaction mixture clearly influ-
ences the product distribution during CNNDBA reduction, as
shown in Figs. 2a–d, and an abundance of transferable pro-
tons also enhances the initial rates and TOFs, as shown in Ta-
ble 3.
2.0% Pd/SiO
5.0% Pd/SiO
b
5.0% Pd/C
Reaction conditions: 298 K, THF solvent buffered at pH = 5.4, p = 194 kPa,
H
2
28.7 µmol CNNDBA/ml initially.
a
Based on CO chemisorption.
Before reduction.
b
3. Results
3.1. Catalyst characterization
The dispersion (D = Pds/Pdtotal) of each Pd catalyst, as
determined by chemisorption measurements assuming adsorp-
tion ratios of CO/Pds = H/Pds = 1, is listed in Table 1. The
CO/Pdtotal ratio of 0.13 for the untreated 5% Pd/C catalyst is in
good agreement with the dispersion of 0.10 obtained from hy-
drogen chemisorption on the pretreated catalyst; therefore, the
irreversible CO uptake of 61.4 µmol/g cat was used to calculate
the TOF for all CNNDBA reduction reactions on the 5% Pd/C
catalyst. The surface-weighted average crystallite size for the
5% Pd/C catalysts, based on CO adsorption and the relationship
3.4. Effect of pH
Studer and Blaser showed that acids promote the deben-
zylation pathway, whereas bases promote the dechlorination
pathway [9]. When starting with a reaction mixture that con-
tains neither acidic nor basic modifiers, the overall reaction rate
and the selectivity can change continually because the slightly
basic starting mixture (the amine functionality of CNNDBA
has basic character) becomes acidic due to HCl produced by
dechlorination; thus the product of the latter pathway becomes
a “promoter” for the debenzylation reaction. To examine the ef-
fect of pH, runs were conducted with 100 µmol CNNDBA/ml at
298 K and 156 kPa H2 in THF at buffered pH levels of 0.1, 3.5,
4.7, 5.2, 6.0, and 12.0. Figs. 3a–f illustrate the changes in prod-
uct distribution going from a reaction mixture at a pH of 0.1 to
one at a pH of 12. It can be seen from Figs. 3a–f that at lower pH
levels, the debenzylation pathway is predominant, whereas at
higher pH levels, dechlorination is favored. Table 4 summarizes
the initial rates of CNNDBA conversion for reaction mixtures
¯
¯
ds = 1.13/D, was ds = 8.7 nm, whereas a volume-weighted
average crystallite size of 17 nm was obtained from XRD line-
broadening measurements.
3.2. Effect of catalyst support
Reduction reactions of CNNDBA in hydrogen were car-
ried out over the five catalysts under a standard set of reaction
conditions, which are given in Table 2 along with initial ac-
tivity and selectivity. The selectivity (S) for debenzylation is
defined as S = rate of debenzylation/(rates of debenzylation +
dechlorination). To counter the effect of changing pH on the
overall reaction rate due to the formation of HCl from the
dechlorination reaction, this comparison was conducted within
a narrow pH range (5.4 0.25) using a triethylamine/acetic
acid buffer. Table 2 shows the initial rates of CNNDBA con-

A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
353
buffered at different pH values, based on CNNDBA concentra-
tions, and Fig. 4 illustrates the dependence of the initial rate of
CNNDBA reduction on the pH of the buffered reaction mixture.
A significant change in slope is apparent at a pH just under 5,
and increased acidity below this point markedly enhances the
rate of debenzylation. Table 5 gives the initial debenzylation
rate (to CNBA, rB) versus dechlorination rate (to NNDBA, rC)
for reaction mixtures buffered at different pH levels, based on
product concentrations. Fig. 5 shows that the pH controls the
reaction pathway (rB or rC) and, at a pH of around 4.5, the se-
lectivity for debenzylation equals that for dechlorination.
3.5. Effect of temperature, substrate concentration, and
hydrogen pressure
The effects of temperature, initial CNNDBA concentration,
and hydrogen pressure on the initial rate of CNNDBA disap-
pearance and the debenzylation selectivity were studied during
CNNDBA reduction over 5% Pd/C in THF solutions under
hydrogen, with solutions buffered at a pH of 5.4 using triethyl-
amine/acetic acid. Initial rates and TOFs for CNNDBA reduc-
tion were determined at 278, 298, and 330 K, and the selectivity
to debenzylation was calculated, as given in Table 6. These rates
give an activation energy of 34.2 kJ/mol [14]. The effect of
the starting concentration (9.5–39.9 µmol/ml) of the organic
substrate on the initial rate of CNNDBA reduction and product
selectivity was studied at 156 kPa H2 and 298 K; the results are
given in Table 7. A reaction order of 0.46 on CNNDBA concen-
tration is obtained from these rates [14]. Reduction reactions at
298 K, an initial CNNDBA concentration of 28.3 µmol/ml, and
a constant H2 pressure of either 115, 156, 194, or 263 kPa were
conducted to determine the effect of the latter on the initial rate
of CNNDBA disappearance and the debenzylation selectivity.
The results are given in Table 8, from which a reaction order on
H2 of 0.67 was determined [14].
4. Discussion
With the exception of the article by Studer and Blaser [9],
there are few detailed accounts of the removal of a benzyl
group from an N atom (N-debenzylation) in the presence of
an acid; however, hydrochloride salts of protected amines have
been reported to undergo debenzylation [20]. In regard to de-
halogenation, there are references that cite the presence of
acids as either favorable [21] or unfavorable [22], whereas it
is well referenced that bases catalyze dehalogenation [23,24].
Bases also help preserve benzyl protective groups when the
reduction of other groups in the molecule is desired [25,26].
Although the effects of acids and bases on N-debenzylation
and dehalogenation reactions have been examined before, the
effect of pH over a wide range has not been reported to
Fig. 2. Temporal composition profile during CNNDBA reduction on 5% Pd/C
at 298 K and 156 kPa H : (a) in CH OH with 120 µmol CNNDBA/ml and
2
3
120 µmol H SO /ml; (b) in acetic acid with 33 µmol CNNDBA/ml; (c) in
2
4
THF + H O (4:1 by wt) with 95 µmol CNNDBA/ml; (d) in THF + CH OH
2
3
(7:3 by wt) with 95 µmol CNNDBA/ml.
Table 3
Initial reaction rates and TOFs for CNNDBA reduction at 298 K Under 156 kPa H
2
Solvent
Modifier
Starting CNNDBA
concentration (µmol/ml)
Initial rate of CNNDBA
disappearance (µmol/(s g ))
TOF for CNNDBA
−1
(s
)
cat
Methanol
Acetic acid
THF/water (4:1 by wt)
THF/methanol (7:3 by wt)
Sulfuric acid (120 µmol/ml)
120
33
95
499
219
99.1
70.0
8.12
3.56
1.61
1.14
None
None
None
95

354
A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
Fig. 3. Reduction of CNNDBA (100 µmol/ml) at 298 K on 5% Pd/C in THF solvent (156 kPa H ): (a) pH = 0.1; (b) pH = 3.5; (c) pH = 4.7; (d) pH = 5.2; (e)
2
pH = 6.0; (f) pH = 12.
Table 4
Initial rates of CNNDBA reduction over 5% Pd/C as a function of the pH of the
reaction mixture
pH
0.1
Initial rate (µmol CNNDBA/(s g ))
cat
282
3.5
4.7
5.2
6.0
12.0
67.4
18.5
9.00
13.3
10.9
Reaction conditions: 298 K, 156 kPa H , 100 µmol CNNDBA/ml initially, THF
2
solvent with buffer.
date; consequently, one goal of the present study was to de-
termine the variation in the reduction rate of CNNDBA and
the selectivity to debenzylation as a function of pH. If the
reduction is carried out in the absence of an acid or base
modifier, then the pH of the solvent will change over the
course of the reaction as HCl is formed from the compet-
itive parallel dechlorination reactions. The dechlorination of
CNNDBA predominates initially (favored by the mild basic-
ity of the tertiary amine group in the molecule), but as the HCl
byproduct increases in the system, debenzylation selectivity in-
creases.
Fig. 4. Initial rate of CNNDBA reduction on 5% Pd vs. pH of the buffered
reaction mixture. (Reaction conditions: 298 K, 156 pKa H , 100 µmol
2
CNNDBA/ml initially.)
For the reductive dehalogenation of a generic organic mole-
cule, a base modifier acts favorably by neutralizing the HCl
byproduct that otherwise could deactivate the catalyst and also
by providing the additional polarity necessary for the cleav-
age of the C-halogen bond. In the case of derivatives of 4-
chloroaniline (such as CNNDBA), a base modifier also keeps

A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
355
Table 6
Effect of temperature on the initial rates of CNNDBA disappearance and deben-
zylation selectivity (to CNBA)
T
Initial rate of
CNNDBA disappearance
TOF
Debenzylation
selectivity
(%)
−1
(K)
(s
)
(µmol/(s g ))
cat
278
298
330
3.62
9.95
23.6
0.059
0.162
0.385
6.8
8.1
11.3
Reaction conditions: HF solvent buffered at pH = 5.4, 156 kPa H and
2
28.3 µmol CNNDBA/ml initially.
Fig. 5. Plot of selectivity to CNBA (debenzylation, r ) and to NNDBA (dechlo-
B
Table 7
rination, r ) over 5% Pd/C as a function of the measured pH of the buffered
C
Effect of CNNDBA concentration on initial rate and debenzylation selectivity
(to CNBA)
reaction mixture. (For reaction conditions, see Fig. 4.)
CNNDBA
concentration
(µmol/ml)
Initial rate of
CNNDBA disappearance
Debenzylation
selectivity
(%)
the nitrogen unprotonated over the course of the reaction. If
a base were not present, then the amine group would become
protonated as HCl formed, and thus would convert from an
electron-donating group to an electron-withdrawing group, as
indicated in Fig. 6. Having an electron-donating group (like a
free amine) in the ortho or para position facilitates removal
of the halogen anion, whereas an electron-withdrawing group
(like a protonated amine) impedes displacement of the halogen.
Nitrogen-containing bases have been shown to inhibit deben-
zylation reactions [27,28], and CNNDBA itself is a nitrogen-
containing base with the potential to self-inhibit debenzylation;
however, the experimental results suggest that the protonated
amine does not act as an inhibitor, thus allowing reductive
debenzylation to occur.
In our study, the Pd dispersed on oxidic supports gave lower
TOFs than Pd on carbon, as shown in Table 2. Studer and Blaser
also found Pd/C to be the catalyst of choice for debenzyla-
tion reactions [9], but it is unclear at this time why the carbon
support allowed a higher specific activity. The selectivity to
debenzylation is low at a pH of 5.4 and shows a 2-fold varia-
tion among the oxide-supported Pd catalysts, but it is noticeably
lower with the Pd/C catalyst. Different catalytic behavior of
Pd/C compared to Pd/oxide has been previously observed for
CO hydrogenation [29]. Studer and Blaser concluded that the
selectivity to dehalogenation for CNNDBA on Pd catalysts in-
creases with the dielectric constant of the solvent, because polar
protic solvents particularly favored the dechlorination reaction,
a fact corroborated by Mukhopadhyay et al. [30]. Acetic acid
was a notable exception, which favored the debenzylation route,
(µmol/(s g ))
cat
9.54
19.1
28.3
37.9
7.12
9.42
9.95
14.4
2.1
5.0
8.1
7.3
Reaction conditions: 298 K, 156 kPa H , THF solvent buffered at pH = 5.4.
2
Table 8
Effect of H pressure on initial rate and debenzylation selectivity (to CNBA)
2
p
Initial rate of
CNNDBA disappearance
(µmol/(s g ))
Debenzylation
selectivity
(%)
(kPa)
cat
115
156
194
263
8.22
9.95
11.2
14.1
6.9
8.1
8.8
9.6
Reaction conditions: 298 K, 28.3 µmol CNNNDBA/ml initially, THF solvent
buffered at pH = 5.4.
as did other acid modifiers, and this pattern was confirmed in
this study (see Fig. 2).
The modifier (acid or base) can interact not only with the
organic reagent, but also with the catalyst surface. Activated
carbon can contain surface groups that incorporate heteroatoms,
and these groups can have acidic, neutral, or basic character,
thus providing an overall amphoteric character to the carbon
support [31]. Both negatively and positively charged surface
sites can exist, depending on the pH. For example, at a basic
pH, the sites containing deprotonated carboxylic groups will
attract a layer of cations from the reaction medium, whereas at
Table 5
Initial rates of Debenzylation and Dechlorination and Selectivity during CNNDBA Reduction over 5% Pd/C as a function of pH
pH
Initial rate of CNBA formation (debenzylation)
Initial rate of NNDBA formation (dechlorination)
Debenzylation
selectivity (%)
−1
−1
r
B
(µmol/(s g ))
TOF (s
)
r
C
(µmol/(s g ))
TOF (s
)
cat
cat
0.1
3.5
4.7
5.2
6.0
287
63
3.5
0.70
0.05
0.02
4.7
0.98
0.057
0.011
0.0008
0.0003
1.0
0.4
16.1
8.9
13.8
12.5
0.016
0.007
0.25
0.14
0.22
0.20
99.6
99.4
17.9
7.3
0.34
0.16
12.0
Reaction conditions: 298 K, 156 kPa H , 100 µmol CNNDBA/ml initially, THF solvent with buffer.
2

356
A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
correlations gave a reaction order of 0.45 for CNNDBA and a
reaction order of 0.67 on H2 pressure.
The kinetic data in Tables 4–8 were verified to be free of
mass transfer effects in the following manner. Rates of H2 up-
take did not change above 500 rpm; therefore, all reactions were
conducted at 700 rpm to guarantee the absence of interphase
(external) mass transport limitations [36]. To ascertain that pore
diffusion had no significant effect on rates, the W-P criterion
was applied to this liquid-phase reaction over the most active
5% Pd/C catalyst [14]. The bulk diffusivity of H2 in THF was
calculated to be 5.5 × 10−5 cm2/s at 298 K [37], and a Henry’s
law calculation gave a solubility of 5.7 µmol H2/cm3 THF at
156 kPa [14]. A different equation was used to calculate a bulk
diffusivity of 2.0 × 10−5 cm2/s for CNNDBA in THF [36,38].
The chemisorption and XRD line-broadening measurements in-
dicated 9- to 17-nm Pd crystallites, so it is reasonable that these
Pd particles reside in pores that are ca. 10 nm in diameter or
larger. Using 10-nm pores and Ternan’s equation [36,39], an ef-
fective diffusivity in the pores of 6 × 10−6 cm2/s is estimated.
With these parameters, the dimensionless W-P criterion (NW-P),
Fig. 6. Free amine groups are electron-donating, whereas protonated amine
groups are electron-withdrawing groups.
an acidic pH, the oxygenated surface groups will attract anions
from solution. During CNNDBA reduction under acidic condi-
tions, the positively charged amine should diffuse to Pd active
sites with no significant interaction between carbon sites of op-
posite electrical charge.
Because of the pH dependence, any kinetic data for rates
and selectivity of debenzylation versus dehalogenation should
be determined in buffered systems. In this study, the choice
of triethylamine/acetic acid provided a buffer that gave good
control in the pH range of 3.5–12. To add one data point at
a pH near 0, a mixture of acetic acid and sulfuric acid was
used. Fig. 4 clearly shows that at a pH of 0–4.5, the initial rate
of CNNDBA consumption decreased markedly with increas-
ing pH; however, for reactions taking place in buffers with a
pH > 4.5, the reaction rate did not change. The selectivity to
debenzylation was essentially 100% in the pH range of 0–3.5
(Fig. 5), but it dropped sharply at higher pH levels and was
totally suppressed at a pH of 6. Fig. 5 shows that the two selec-
tivity curves intersect at a pH near 4.5, the same pH at which
there is an inflection point in the overall rate of reaction (see
Fig. 4). This inflection point corresponds approximately to the
pKa of the organic reagent, where 50% of the CNNDBA is pro-
tonated and 50% is in free-base form [32]. Fig. 5 also illustrates
that dechlorination can be the predominant reaction, even under
acidic conditions, as long as the pH of the buffer is higher than
the pKa of the amine group (i.e., in a pH range of 4.5–7). The
finding that dehalogenation reactions can be catalyzed under
acidic conditions is significant for three reasons: (i) It dispels
previous conclusions that the presence of a base modifier is re-
quire for a dehalogenation reaction to occur; (ii) it shows that
debenzylation reactions can be selectively carried out at higher
pH levels than previously suggested; and (iii) it offers a guide-
line for obtaining high selectivity for either dehalogenation or
debenzylation under mild conditions, which may help preserve
functionalities in the molecule that might otherwise be affected
by strong acidic or basic conditions.
The selectivity to debenzylation was low (2–11%) in THF at
a buffered pH of 5.4, but it increased with an increase in the hy-
drogen pressure, the reaction temperature, or the concentration
of the organic substrate (see Tables 6–8). The apparent acti-
vation energy of 34.2 kJ/mol for the rate of CNNDBA disap-
pearance related primarily to the hydrodechlorination reaction,
which is the dominant reaction at this pH. No comparable val-
ues for an aromatic could be found for comparison, but C–Cl
bond energies for aromatics are similar to, or 20–25% higher
than, those for aliphatics, and apparent activation energies for
the vapor-phase hydrodechlorination of CF3CFCl2 have been
reported to be 52 kJ/mol on Pd/Al2O3 [33] and 75–130 kJ/mol
on Pd/C for removal of the initial Cl atom [34,35]. The initial
rate data obtained at 298 K and a pH of 5.4 under different re-
actant concentrations were first fitted to a power rate law; these
ꢀRp2
CsDeff
NW-P
=
,
(1)
where ꢀ is the observed rate for reactant i per volume of cata-
lyst, Rp is the average catalyst particle radius (15 µm), Cs is the
concentration of reactant i at the catalyst surface, and Deff is
the effective diffusivity of reactant i in the liquid solvent within
the pores, gave values of 0.04 for CNNDBA and 0.02 for H2
at the highest rate in Tables 7 and 8 of 14.4 µmol/(s g). These
values are well below 0.3, indicating no significant intraphase
(internal) mass transfer limitations [36]. Even the highest rate
in Table 4 gave W-P numbers of 0.2–0.3.
In regard to modeling, two forms of a classical Langmuir–
Hinshelwood (L-H) mechanism were found to fit the data; one
sequence assumed a single type of active site, whereas the sec-
ond sequence used a similar L-H catalytic cycle but invoked two
types of active sites, one site to adsorb and dissociate H2 and the
other to adsorb the organic reactant. The former type would cor-
respond to 3-fold and 4-fold hollow sites for H atoms, and the
latter type would represent on-top sites comprised of an ensem-
ble of one or more Pd atoms. Because the latter model provided
a slightly better fit of the data, only it is discussed herein. In
either case, quasi-equilibrated adsorption of H2 and CNNDBA
was assumed with the rate-determining step (RDS) being the
addition of the second H atom. The proposed catalytic cycle is
shown below, where Li represents the leaving group (i.e., LB for
benzyl and LC for Cl) and CNNꢁ is either N,N-dibenzylaniline
(NNDBA) or chloro-N-benzylaniline (CNBA):
K
H
2
H2 + 2S1
CNNDBA + S2
2H–S1,
KCNNDBA
(2)
(3)
CNNDBA–S2,
Ki
CNNDBA–S2 + H–S1
CNNDBAH–S2 + S1,
(4)
(5)
ki
CNNDBAH–S2 + H–S1 → CNNꢁ–S2 + LiH–S1,
1/KCNN
ꢁ
CNNꢁ–S2
CNNꢁ + S2,
(6)

A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
357
LiH–S1
LiH + S1,
(7)
where S1 and S2 are the two types of active sites.
The fractional surface coverages for the species involved in
the quasi-equilibrated steps preceding the RDS are
Θ1(H) = K1/2PH1/2Θ1(S)
(8)
H2
2
and
Θ2(CNNDBA) = KCNNDBA
CCNNDBAΘ(S),
(9)
where the adsorbed species is in parentheses, and the subscript
S indicates a vacant site. From the RDS [Eq. (5)], the rate for
either CNNDBA debenzylation (i = B) or dechlorination (i =
C) is given by:
ri = L1L2kiKiKCNNDBAKH CCNNDBAPH Θ1(S)Θ2(S)
,
(10)
2
2
where ki is the rate constant for the respective RDS and L1
and L2 are the concentrations of the two types of active sites,
S1 and S2, respectively. Assuming that the most abundant sur-
face intermediate on the type 1 or type 2 sites is a H atom or a
CNNDBA molecule, respectively, the site balances for the two
types of sites are then
Θ1(S) + Θ1(H) = 1
and
(11)
Θ2(S) + Θ2(CNNDBA) = 1.
(12)
Fig. 7. Fit of the 2-site LH model to the initial rate dependencies for CNNDBA
The absence of any form of adsorbed HCl in the site balance
is justified by the presence of the buffer; that is, the triethyl-
amine removes HCl by forming a chloride salt, which essen-
tially will not dissociate in the absence of water. A combination
of Eqs. (8)–(12) yields the final rate expression for CNNDBA
disappearance,
reduction at 298 K: (a) rate vs. CNNDBA concentration, PH = 156 kPa;
2
(b) rate vs. H pressure, CCNNDBA = 28.3 µmol/ml initially.
2
Table 9
Optimized parameters for Eq. (13) and reaction orders provided by the two-site
L-H model (298 K and pH = 5.4)
ꢁ
Equation
k (mol/
K
K
Reaction
Reaction
kꢁKCNNDBAKH CCNNDBAPH
H
CNNDBA
2
2
2
i
(s g ))
(1/kPa) (L/mol)
order in H order in
cat
r = rB + rC =
, (13)
2
(1 + KCNNDBACCNNDBA)(1 + K1/2PH1/2
)
CNNDBA
0.48
0.45
H2
2
−5
Two-site LH model 1.45 × 10
Power-law
0.021 58.4
0.67
0.67
where k1ꢁ = L1L2 (kBKB +kCKC). However, it is readily appar-
ent in Tables 6–8 that kC ꢂ kB; therefore, Eq. (13) is primarily a
representation of the dechlorination rate. Fig. 7 illustrates the fit
of the two-site L-H model to the initial rate data for CNNDBA
reduction using the nonlinear regression software package Sci-
entist (Micromath Scientific Software) for optimization. The
optimum parameter values at 298 K are listed in Table 9. Com-
pared with the one-site L-H model, the two-site L-H model
provides reaction orders for CNNDBA and H2 slightly closer
to those values obtained from the regression of the experi-
mental data (Table 9), which gave a regression coefficient of
R = 0.995 [14].
–
–
–
proposed by Eqs. (2)–(7). However, dissociated adsorption
of halogenated organic compounds, particularly halogenated
paraffins, has also been proposed, with the hydrogen then re-
acting with the halide to remove it from the surface [42,43]. If
steps 3–5 were altered to represent this change, then the result-
ing rate expression would be the same as Eq. (13), except that it
would contain a concentration term for the hydrogenated leav-
ing group, such as HCl. However, because of the buffer, this
term would remain constant (and very small), and the resulting
rate equation would be indistinguishable from Eq. (13) as this
term became incorporated into kꢁ. Consequently, we can draw
no detailed conclusions regarding the reaction mechanism from
our kinetic results. For reductive dehalogenation of aliphatic
chlorofluorocarbons on Pd/C, the reaction orders observed by
Ribeiro and coworkers [34,35] were 0.15–1 for the organic sub-
strate and 0.4–0.6 for hydrogen, values similar to those obtained
here for dehalogenation of an aromatic compound. These lat-
ter two studies, which were in unbuffered vapor-phase systems,
Little literature exists on the mechanistic aspects of re-
ductive debenzylation reactions, and, although dehalogenation
reactions are more frequently referenced, no generally ac-
cepted mechanism has yet emerged. Most proposed mecha-
nisms have involved dissociated hydrogen in either hydride or
radical form reacting with the adsorbed organic species [34,35,
40–43]. Some interpretations involve the addition of a H atom
(or a hydride) to the C–Li bond in an aromatic molecule to fa-
cilitate its rupture [40,41]. This would correspond to the model

358
A. David, M.A. Vannice / Journal of Catalysis 237 (2006) 349–358
also reported a near first-order negative dependence on the HCl
byproduct, and the authors made an effort to correct their TOF
values for the varying HCl concentrations [34]. Their rate ex-
pression would also be consistent with the alternative reaction
model discussed above.
Acknowledgment
This study was sponsored by funding from Bristol–Myers
Squibb.
References
5. Summary
[1] A.A. Haddach, A. Kelleman, M.V. Deaton-Rewolinski, Tetrahedron
Lett. 43 (2002) 399.
[2] C.-Y. Chern, Y.-P. Huang, M.W. Kan, Tetrahedron Lett. 44 (2003) 1039.
[3] H. Suzuki, A. Tsukuda, M. Kondo, M. Aizawa, Y. Senoo, M. Nakajima, T.
Watanabe, Y. Yokoyama, Y. Murakami, Tetrahedron Lett. 36 (1995) 1671.
[4] J. Kroutil, T. Trnka, M. Ceerny, Org. Lett. 2 (2000) 1681.
[5] S.D. Bull, S.G. Davies, G. Fenton, A.W. Mulvaney, R.S. Prasad, A.D.
Smith, J. Chem. Soc., Perkin Trans. I (2000) 3765.
[6] P. Rylander, Catalytic Hydrogenation in Organic Synthesis, Academic
Press, New York, 1979.
[7] Y. Ambroise, C. Mioskowski, G. Djega-Mariadassou, B. Rousseau, J. Org.
Chem. 65 (2000) 7183.
[8] J.S. Bajwa, J. Slade, O. Repic, Tetrahedron Lett. 41 (2000) 6025.
[9] M. Studer, H.-U. Blaser, J. Mol. Catal. A 112 (1996) 437.
[10] M. Freifelder, Practical Catalytic Hydrogenation, Wiley, New York, 1971.
[11] H.-U. Blaser, C. Malan, B. Pugin, F. Spindler, H. Steiner, M. Studer, Adv.
Synth. Catal. 345 (2003) 103.
[12] S. Zinovyev, A. Perosa, S. Yufit, P. Tundo, J. Catal. 211 (2002) 347.
[13] T. Mallat, A. Baiker, Appl. Catal. A 200 (2000) 3.
[14] A. David, MS Thesis, The Pennsylvania State University, 2004.
[15] P. Chou, M.A. Vannice, J. Catal. 104 (1987) 17.
[16] A.D. Jorgensen, K.C. Picel, V.C. Stamoudis, Anal. Chem. 62 (1990) 683.
[17] J.E. Benson, H.S. Hwang, M. Boudart, J. Catal. 30 (1973) 146.
[18] D.J.C. Yates, J. Sinfelt, J. Catal. 8 (1967) 348.
[19] J.J. Venter, A.A. Chen, J. Phillips, M.A. Vannice, J. Catal. 119 (1989) 451.
[20] P.W. Ernardt, Synth. Commun. 13 (3) (1983) 103.
[21] J.L. Benitez, G. Del Angel, J. Mol. Catal. A 165 (2001) 9.
[22] J.B. Hook, G.A. Gramiccioni, E.N. Balko, Appl. Catal. B 1 (1992) 285.
[23] M.A. Aramendia, V. Boráu, C. Jiménez, F. Lafont, A. Marinas, J.M. Mari-
nas, F.J. Urbano, J. Mol. Catal. A 184 (2002) 237.
The activity and selectivity for debenzylation versus dechlo-
rination for the reduction of CNNDBA in H2 over Pd dispersed
on various supports was investigated. During the reduction of
CNNDBA dissolved in a THF/water solvent, the debenzylation
selectivity increased over time because the dechlorination reac-
tion, which was initially predominant, produced HCl, a mod-
ifier that suppresses the latter reaction and enhances debenzy-
lation. Consequently, reduction experiments were carried out in
buffered solvents to control and examine the effect of pH on ac-
tivity and selectivity. The results clearly indicated that dechlo-
rination reactions were completely suppressed at a pH < 3,
debenzylation reactions were suppressed at a pH > 6, and the
selectivity for debenzylation equaled that for dechlorination at
a pH of ∼4.5, which corresponds to the pKa of the CNNDBA
substrate. This behavior is interpreted as follows. At a pH equal
to its pKa, 50% of the CNNDBA is protonated and 50% is
unprotonated at the tertiary amine group; thus, at a pH lower
than the kPa, CNNDBA exists primarily in a protonated form,
whereas at a pH higher than the pKa, CNNDBA exists primarily
in an unprotonated form. The protonated amine is an electron-
withdrawing group that inhibits, via resonance, the cleavage of
the halogen group during the dechlorination reaction, whereas
the unprotonated amine, which is an electron-donating group,
facilitates dechlorination under H2. Thus, it was determined
that the selectivity to debenzylation depends primarily on the
pH of the reaction mixture; however, the presence of a base is
not required for the dechlorination reaction to proceed. In fact,
with CNNDBA, dechlorination occurs at a high rate even un-
der acidic conditions as long as the pH of the reaction mixture
is greater than the pKa of the protected amine.
[24] S. Mukhopadhyay, S. Ratner, A. Spernat, N. Quafisheh, Y. Sasson, Org.
Proc. Res. Dev. 6 (2002) 297.
[25] K. Hirota, K. Hattori, H. Sajiki, J. Org. Chem. 63 (1998) 7990.
[26] J. Li, S. Wang, G.A. Crispino, K. Tenhuisen, A. Singh, J.A. Crosso, Tetra-
hedron Lett. 44 (2003) 4041.
[27] K. Hirota, H. Sajiki, Tetrahedron 54 (1998) 13981.
[28] T. Mallat, A. Baiker, Appl. Catal. A 200 (2000) 3.
[29] M.A. Vannice, N. Krishnankutty, J. Catal. 155 (1999) 327.
[30] S. Mukhopadhyay, G.K. Gandi, S.B. Chandalia, Org. Proc. Res. Dev. 3
(1999) 201.
In systems buffered at a pH of 5.4, the selectivity to deben-
zylation was low and only modestly enhanced by increases in
temperature, hydrogen pressure, or initial CNNDBA concen-
tration. The effect of the catalyst support on selectivity was not
pronounced, but a much higher TOF was observed for reac-
tions catalyzed by Pd dispersed on carbon compared with Pd
dispersed on oxidic supports such as Al2O3, TiO2, and SiO2.
The initial rates of organic substrate reduction at a pH of 5.4
were modeled using a L–H sequence invoking two types of ac-
tive sites, one site that dissociatively adsorbs H2 and another
that adsorbs the organic substrate. This model fit the data well
and could provide the observed reaction orders of ∼1/2 for H2
and ∼2/3 for CNNDBA. The W-P criterion was applied to this
liquid-phase reaction to verify that the kinetic data used for
modeling were obtained in the kinetic regime.
[31] M.L. Toebes, J.A. van Dillen, K.P. de Jong, J. Mol. Catal. A 173 (2001)
75.
[32] M. Oki, M. Ohira, Y. Yoshioka, T. Morita, H. Kihara, N. Nakamura, Bull.
Chem. Soc. Jpn. 57 (1984) 2224.
[33] Z. Karpinski, K. Early, J.D. d’Irti, J. Catal. 164 (1996) 378.
[34] C.D. Thompson, R.M. Rioux, N. Chen, F.H. Ribeiro, J. Phys. Chem. B 104
(2000) 3067.
[35] N. Chen, PhD Thesis, Worcester Polytechnic Institute (2003).
[36] M.A. Vannice, Kinetic of Catalytic Reactions, Springer–Kluwer, NY,
2005, Chap. 4.
[37] C.R. Wilke, P. Chang, AIChE J. 1 (1955) 264.
[38] C.J. King, L. Hsueh, K. Mao, J. Chem. Eng. Data 10 (1965) 348.
[39] M. Ternan, Can. J. Chem. Eng. 65 (1987) 244.
[40] A. Finiels, P. Geneste, C. Moreau, J. Mol. Catal. A 107 (1996) 385.
[41] C. Moreau, J. Joffre, C. Saenz, J.C. Alfonso, J.-L. Porefaix, J. Mol. Catal.
A 161 (2000) 141.
[42] B. Coq, G. Ferrat, F. Figueras, J. Catal. 101 (1986) 434.
[43] F.J. Urbano, J. Marinas, J. Mol. Catal. A 173 (2001) 329.