Liquid-phase hydrogenation of citral over Pt/SiO2 catalysts - I. Temperature effects on activity and selectivity

Article abstract of DOI:10.1006/jcat.1999.2803

Liquid-phase hydrogenation of citral (3,7-dimethyl-2,6-octadienal) over Pt/SiO₂ catalysts was studied under kinetic conditions free of transport limitations and poisoning effects at 298-423 K and 7-21 atm. The rate of citral hydrogenation exhibited a minimum in activity between 298 and 423 K, and this was attributed to the relative activation energies for the decomposition of the unsaturated alcohol and desorption of CO. At low reaction temperatures (298 K), the rate of citral disappearance was low and CO was slowly accumulated under reaction conditions to block active sites and eventually led to a complete loss of activity. At 373 K, alcohol decomposition was more rapid, but the CO desorption rate was significantly enhanced and a pseudo-steady state was readily established which resulted in minimal additional inhibition and conventional Arrhenius behavior with an activation energy of 7 kcal/mole. The reaction kinetics for each hydrogenation reaction in the reaction network were modeled using a Langmuir-Hinshelwood model invoking dissociative adsorption of hydrogen, competitive adsorption between hydrogen and the reactive organic species, and the observed product distributions at each temperature were described very well. Furthermore, deactivation due to decomposition of the unsaturated alcohols geraniol and nerol, along with CO desorption at higher temperatures, was invoked to account for the unusual rate dependence on temperature.

Full text of DOI:10.1006/jcat.1999.2803

Journal of Catalysis 191, 165–180 (2000)
doi:10.1006/jcat.1999.2803, available online at http://www.idealibrary.com on
Liquid-Phase Hydrogenation of Citral over Pt/SiO2 Catalysts
I. Temperature Effects on Activity and Selectivity
Utpal K. Singh and M. Albert Vannice1
Department of Chemical Engineering, Pennsylvania State University, 107 Fens`ke Laboratory, University Park, Pennsylvania 16802-4400
Received September 3, 1999; revised December 16, 1999; accepted December 22, 1999
the specialty and fine chemicals industry where NaBH4 is
used as a stoichiometric reductant (2). Significant focus has
been placed on studying selectivity in hydrogenation reac-
tions of , -unsaturated aldehydes, and several reviews of
this work have recently been published (3–5). Numerous
vapor-phase hydrogenation studies conducted under dif-
ferential conditions to obtain reaction kinetics have proved
valuable in obtaining mechanistic information (6–14); how-
ever, liquid-phase hydrogenation studies reported in the
literature have focused primarily on obtaining a qualita-
tive understanding of the reaction by monitoring the effect
of mono- and bimetallic catalysts on product distribution
(15–17). There is scant literature that examines the influ-
ence of reaction parameters such as temperature, pressure,
and transport limitations on kinetics of liquid-phase hy-
drogenation reactions, and this deficiency was voiced in a
recent review by Gallezot and Richard on hydrogenation
of , -unsaturated aldehydes (3).
The work of Lercher and co-workers represents one of
the few attempts at examining the effect of various reaction
parameters such as temperature, pressure, catalyst weight,
and water concentration on the liquid-phase hydrogena-
tion of , -unsaturated aldehydes (18, 19). Regardless, a
number of questions still remain to be resolved including
the significance of internal diffusion resistance, catalyst poi-
soning, and the role of intermediates during hydrogenation
reactions. One important consideration is that many of the
liquid-phase hydrogenation studies reported in the litera-
ture have been performed after an ex situ reduction of the
catalyst, and proper precautions were not taken to remove
dissolved oxygen from the liquid phase prior to the start
of reaction. This can introduce various changes in the state
of the catalyst and can also influence the reaction kinetics
(20).
Liquid-phase hydrogenation of citral (3,7-dimethyl-2,6-octa-
dienal) over Pt/SiO2 catalysts was studied in the temperature and
pressure ranges 298–423 K and 7–21 atm, respectively. The reac-
tion kinetics were shown to be free of artifacts arising from trans-
port limitations and poisoning effects. The reaction rate in hexane
as the solvent exhibited an activity minimum at 373 K. The ini-
tial turnover frequency for citral disappearance over 1.44% Pt/SiO2
catalyst at 20 atm H2 pressure decreased from 0.19 s 1 at 298 K to
0.02 s 1 at 373 K, but exhibited normal Arrhenius behavior between
373 and 423 K with an activation energy of 7 kcal/mol. Reaction at
298 K produced substantial deactivation, with the rate decreasing
by more than an order of magnitude during the first 4 h of reac-
tion; however, reaction at temperatures greater than 373 K exhibited
negligible deactivation and a constant rate up to citral conversions
greater than 70%. These unusual temperature effects were mod-
eled using Langmuir–Hinshelwood kinetics invoking dissociative
adsorption of hydrogen, competitive adsorption between hydrogen
and the organic compounds, and addition of the second hydrogen
atom to each reactant as the rate-determining step. Decomposition
of the unsaturated alcohol (either geraniol or nerol) was proposed to
occur concurrently with the hydrogenation steps to yield adsorbed
CO and carbonaceous species which cause the deactivation, but at
higher temperatures these species could be removed from the Pt
surface by desorption or rapid hydrogenation, respectively. The ac-
tivity minimum observed in the present study is attributed to the
relative rates of the alcohol decomposition reaction and CO desorp-
tion, with the decomposition reaction having an activation barrier
lower than that for CO desorption.
c
2000 Academic Press
INTRODUCTION
Approximately 50–100 kg of by-product can be produced
for a single kilogram of product in the pharmaceutical and
fine chemicals industry, with much of the by-product being
present as inorganic salts arising from multistep syntheses
using stoichiometric reductants or homogeneous catalysts
(1). Selective hydrogenation of , -unsaturated aldehydes
represents a broad class of industrially relevant reactions in
We have studied liquid-phase hydrogenation of citral
(3,7-dimethyl-2,6-octadienal) over supported Pt catalysts.
Citral has three unsaturated bonds including a conjugated
==
==
==
C
C-C O bond system and an isolated C C bond. The
reaction chemistry of citral hydrogenation is shown in
Fig. 1, which also includes minor side products not routinely
1 To whom correspondence should be addressed.
165
0021-9517/00 $35.00
c
Copyright
2000 by Academic Press
All rights of reproduction in any form reserved.

166
SINGH AND VANNICE
FIG. 1. Reaction chemistry for citral hydrogenation.
presented in the reaction pathways from previous reports
In the present investigation, a thorough, methodical
(5). The two isomers of citral can undergo hydrogenation study of liquid-phase hydrogenation of citral has been per-
==
of the conjugated C O functionality to yield the cis and formed with the aim of obtaining quantitative kinetic data
trans isomers of the unsaturated alcohol (UALC), nerol free of artifacts from transport resistance and catalyst poi-
and geraniol, respectively. Alternatively, both citral isomers soning. Particular emphasis has been placed on understand-
==
can react via the conjugated C C bond or the isolated ing the unusual effect of temperature on reaction rates and
==
C
C bond to yield either the partially saturated aldehyde product distribution.
(PSALD) or the cis and trans isomers of 3,7-dimethyl-2-
octenal (ENAL), respectively. The latter product has not
been confirmed unequivocally, but its presence has been
inferred from mass spectral data and reaction kinetics as
discussed in the subsequent paper (21). UALC and PSALD
EXPERIMENTAL
SiO2 (Davison grade 57, 220 m2/g, 60–100 mesh) was
can undergo further hydrogenation of the remaining un- dried and calcined at 773 K for 4 h prior to impregnation.
saturated bond in the initially conjugated system to yield Three catalysts were prepared with hydrogen hexachloro-
the partially saturated alcohol (PSALC), which can subse- platinate(IV) hydrate (H2PtCl6, Aldrich 99.995% ) as a pre-
quently be hydrogenated to yield the completely saturated cursor using the incipient wetness (IW) technique; i.e., an
product, 3,7-dimethyloctanol (SAT). PSALD can also react aqueous solution of chloroplatinic acid was added dropwise
first via the isolated C C species to yield dihydrocitronellal to a calcined support (2 cm3/g SiO2) and subsequently dried
==
(DCAL) or it can isomerize with ring closure to yield iso- at 393 K overnight. A 3.8% Pt/SiO2 catalyst was prepared
pulegol (IP). ENAL can undergo hydrogenation of either with Pt(NH3)4Cl2 using an ion-exchange method. Hydro-
==
==
the C O bond or the C C bond to yield 3,7-dimethyl-2- gen chemisorption was measured at 300 K in a static volu-
octenol or DCAL, respectively. These two intermediates metric apparatus with a base pressure of 10 7 Torr using a
can further hydrogenate to yield the completely saturated procedure described elsewhere (22). All catalysts were re-
product (SAT). In addition to the above reaction chem- duced at 673 K for 75 min under flowing H2. Although dual
istry, some surface side reactions can also occur, as will be isotherms were obtained, Pt dispersions were calculated us-
described later in this paper.
ing the value of the total H2 uptake isotherm extrapolated

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
167
TABLE 1
H2 Chemisorption on Pt/SiO2 Catalysts at 300 K
H2 uptake ( mol/gcat)
Activity
mol/g cat/min)a
Initial TOF
1
a
Catalyst
(H2)Total
(H2)Rev
HTot/Pt
d (nm)
(
(s
)
2.65% Pt/SiO2
1.44% Pt/SiO2
0.49% Pt/SiO2
3.80% Pt/SiO2
15.0
15.2
5.6
6.1
7.4
3.4
0.22
0.41
0.45
0.66
5.1
2.8
2.5
1.7
164
27
8
0.09
0.017
0.012
0.007
64.1
24.6
53
a Reaction conditions: 373 K, 20 atm H2, 1 M citral in hexane.
to zero pressure and assuming an adsorption stoichiometry extending inside the reactor and collected in a closed, N2-
of unity, i.e., a Had :Pts ratio of 1. purged vessel, and it was then analyzed in a HP 5890 gas
Hydrogenation experiments were conducted in a 100-ml chromatograph equipped with a thermal conductivity de-
EZ-seal autoclave with an automated data acquisition sys- tector and a 10% Carbowax 20M on Supelcoport packed
tem that can monitor H2 uptakes. Details have been pro- column. Initial rates of citral disappearance were evaluated
vided elsewhere (23). The reactor was loaded with 0.5 to using the slope of the linear portion of the temporal citral
1.0 g of catalyst, sealed, and leak tested at 54 atm to ensure conversion profile (for citral conversions less than 20% ).
a tight seal. Helium (MG Industries, 99.999% ) was used as The product selectivity was calculated as follows:
received while H2 (MG Industries, 99.999% ) was further
concentration of species i
purified by passage through a high-pressure Oxytrap (All-
P
Si =
[1]
products concentration of species i
tech). Five pressure/vent cycles were conducted at room
temperature by pressurizing the reactor to 54 atm in He
followed by venting to atmospheric pressure to remove all
traces of oxygen from inside the reactor. The pretreatment
procedure was then initiated by flowing 200 cm3(STP)/min
He over the catalyst while heating the reactor from room
temperature to 673 K in 3 h. The flow was then switched
to 500 cm3(STP)/min hydrogen for 75 min prior to cool-
ing to room temperature and storing overnight under a
static atmosphere of 20 psia H2. This storage procedure
had no effect on catalytic behavior as evidenced by identi-
cal turnover frequencies (TOFs) and product distributions,
within experimental uncertainty, whether the catalyst was
stored overnight or used immediately after the reduction.
The standard reaction conditions were 20 atm hydrogen
pressure using 1 mol citral/liter in hexane with a total reac-
tion volume of 60 ml. Citral (Sigma, 97% ) and hexane were
degassed by sparging with high flow rates of nitrogen for
30 min prior to injection into the reactor. Hexane (Fisher
99.9% saturated C6 hydrocarbons) was fed into the reactor
at reaction temperature and pressure using a high-pressure
syringe pump (ISCO 500D) in a closed system to prevent
any exposure to air. The hexane/catalyst slurry was stirred
at 1000 rpm and allowed to equilibrate for 30 min prior
to the introduction of citral to obtain a total reaction vol-
ume of 60 ml. The pressure was continuously maintained to
within 5% of the set point with a Brooks 5860 pressure con-
troller duringthese semi-batch reactor experiments, and the
pressure decay within the 5% range, i.e., 315–300 psia, was
monitored to obtain the instantaneous rate of H2 uptake
at any time during the reaction. In addition, a 0.5-ml liq-
uid sample was periodically withdrawn through a dip tube
RESULTS
Table 1 gives information about the catalysts and dis-
plays the H2 chemisorption results and pretreatment pro-
cedures for the catalysts used. Two catalysts with similar
dispersions (H/Pt 0.4) but a threefold variation in plat-
inum loading (0.49 and 1.44% Pt/SiO2) were used to apply
the Madon–Boudart test to check for the absence of heat
and mass transport effects (24). The results of the test un-
der standard conditions and 373 or 393 K are shown in
Table 2. The equal TOFs, within experimental uncertainty,
for citral disappearance on the two catalysts at two differ-
ent reaction temperatures verifi the absence of internal and
external heat and mass transfer limitations. In addition, no
catalyst deactivation was observed at these temperatures as
evidenced by a constant hydrogenation rate as citral con-
version varied from 10% to greater than 80% . Similar runs
TABLE 2
Madon–Boudart Test with 0.49% Pt/SiO2
(H/Pt = 0.45) and 1.44% Pt/SiO2 (H/Pt =
0.41) at 373 and 393 K, 20 atm H2, and 1 M
Citral in Hexane
1
TOF (s
)
373 K
393 K
0.49% Pt/SiO2
1.44% Pt/SiO2
0.012
0.017
0.047
0.049

168
SINGH AND VANNICE
the catalyst at 673 K in flowing hydrogen and passivating it
at 300 K with 5% O2 in He using 100 cm3(STP)/min before
its subsequent exposure to air. The catalyst was transferred
into the autoclave and purged first with helium and finally
with hydrogen. The reactor temperature and pressure were
then raised to 373 K and 20 atm H2 at which time 50 cm3
hexane (with no degassing) was added. The hexane/catalyst
slurry was stirred and allowed to equilibrate under H2 for
60 min at the reaction temperature and pressure prior to
the addition of citral (with no degassing) to commence the
reaction. The initial rate of disappearance of citral on the
ex situ-reduced catalyst (TOF = 0.007 s 1) was one-third
that on the in situ-reduced catalyst (TOF = 0.020 s 1),
which suggests either that the catalyst may not be com-
pletely activated following an ex situ reduction procedure
or that low concentrations of water can affect the kinetics.
In addition, a nonzero intercept was observed after either
pretreatment method at 373 K. All results reported in the
present study were obtained following in situ reduction of
the catalyst at 673 K in hydrogen and degassing of citral
and hexane with nitrogen prior to their introduction into
the reactor.
FIG. 2. Temporal H2 uptake profiles during hydrogenation of 1 M
citral in hexane at 298 K and 20 atm H2, over 0.49% Pt/SiO2 (H/Pt = 0.45),
1.44% Pt/SiO2 (H/Pt = 0.41), and 3.8% Pt/SiO2 (H/Pt = 0.67) catalysts.
were conducted at 298 K and standard conditions with a
series of catalysts whose concentration of surface Pt atoms
varied by more than an order of magnitude from 11 mol
Pts/g cat for 0.49% Pt/SiO2 to 128 mol Pts/g cat for 3.8%
Pt/SiO2, and these results are presented in Fig. 2. The sim-
ilar trend in TOF for H2 uptake versus time for the three
catalysts is additional evidence indicating the absence of
artifacts due to transport limitations or poisoning effects
at 298 K and 20 atm H2; the obvious deactivation that is
observed is attributed to a side reaction, as discussed later.
Figures 3a and 3b show the effect of catalyst weight on the
reaction kinetics during hydrogenation under standard con-
ditions at 298 K using 1.44% Pt/SiO2. Despite the eightfold
difference in catalyst mass inside the reactor, the temporal
H2 uptake curves on a TOF basis are identical; however,
substantial differences between the two runs are seen in
Fig. 3b when the H2 uptake is plotted versus citral conver-
sion. Due to the strong deactivation/inhibition behavior,
the reaction with 0.09 g catalyst yielded 6.5% citral conver-
sion after 3 h of reaction as compared with 57% for the run
with 0.81 g catalyst. The identical values during the initial
stages of the reaction for the two runs reinforce the fact
that there is no transport limitation related to the transfer
of H2 from the gas to the liquid phase (25).
Experiments were conducted to obtain the reaction or-
ders in citral concentration and H2 pressure for the initial
Figure 4 illustrates the effect that the pretreatment
procedure, regarding exposure to O2, can have on the tem-
poral conversion profile for the reaction under standard
conditions at 373 K. The in situ pretreatment procedure was
described under Experimental. Briefly, the 1.44% Pt/SiO2
catalyst was reduced in situ at 673 K in flowing hydrogen,
then cooled to reaction temperature in hydrogen. Degassed
hexane was added to the reactor (ensuring the exclusion
of oxygen) and the slurry was allowed to equilibrate for
30 min prior to adding degassed citral to commence the
FIG. 3. Effect ofcatalyst weight on TOF for H2 uptake duringreaction
at 298 K and 20 atm H2 over 1.44% Pt/SiO2 with 1 M citral in hexane
reaction. The ex situ pretreatment consisted of reducing (a) as a function of time and (b) as a function of citral conversion.

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
169
FIG. 4. Effect of pretreatment procedure for reaction over 1.44% Pt/SiO2 catalyst at 373 K, 20 atm H2 pressure, and 1 M citral in hexane. Catalyst
amounts 0.50 and 0.80 g were used for the in situ and ex situ reduced catalysts, respectively.
rate of citral hydrogenation (citral conversions less than at 298 and 373 K. The initial TOF for citral disappearance
20% ) over a 2.65% Pt/SiO2 catalyst at temperatures from is 0.18 s 1 at 298 K and 0.02 s 1 at 373 K, which represents
298 to 423 K, H2 pressures from 7 to 41 atm, and citral con- almost an order of magnitude decrease in the rate even
centrations from 0.97 to 5.9 mole/liter in hexane. The initial though the reaction temperature is 75 K higher. However,
rate of citral hydrogenation was zero order and first order the higher initial rate at 298 K is accompanied by strong
with respect to citral concentration and hydrogen pressure, deactivation, as shown in Fig. 6 where the instantaneous
respectively.
H2 uptake profile is plotted as a function of citral conver-
Figure 5 displays the temporal citral conversion profile sion under standard conditions and 298, 373, and 423 K.
for reaction under standard conditions using 1.44% Pt/SiO2 The instantaneous TOF for H2 uptake at 298 K with 0.8 g
1
of catalyst decreased from 0.5 s during the first 10 min
1
of reaction to 0.05 s after 4 h of reaction time and 50%
citral conversion. Artifacts due to transport resistance and
FIG. 5. Temporal citral conversion profile for reaction over 0.80 g of
1.44% Pt/SiO2 catalyst at 298 and 373 K, 20 atm H2, and 1 M citral in
hexane.
FIG. 6. TOF for H2 uptake as a function of citral conversion for reac-
tion over 1.44% Pt/SiO2 at 298, 373, and 423 K at 20 atm H2 and 1 M citral
in hexane.

170
SINGH AND VANNICE
poisoning were discounted after application of the Madon–
Boudart test, as shown in Fig. 2 and Table 2. Reaction with
0.5 g of 1.44% Pt/SiO2 at 373 K yielded an initial TOF for
H2 uptake that was more than an order of magnitude lower
than that at 298 K, but remained approximately constant as
citral conversion increased. No H2 uptake points were ob-
tained for citralconversionslessthan 5% due to the fact that
this level of citral conversion was obtained during the first
few minutes of reaction, a period too short to allow quan-
titative analysis. The initial reaction rate decreased as the
temperature increased from 298 to 373 K, but conventional
Arrhenius behavior was seen during a further increase from
373 to 423 K with an activation energy of 7 kcal/mol. This
is displayed in Fig. 6 which shows the instantaneous rate of
H2 uptake at 423 K is greater than that at 373 K. Similar
to the reaction at 373 K, and in contrast to that at 298 K,
negligible deactivation was observed at 423 K.
This unusual temperature effect on activity was ac-
companied by a dramatic change in product distribution,
as shown in Figs. 7 and 8 for hydrogenation over 1.44%
Pt/SiO2 under standard conditions. Figures 7a and 8a show
FIG. 8. (a) Concentration profile for reactants and products during
citral hydrogenation over 1.44% Pt/SiO2 at 373 K, 20 atm H2 pressure,
and 1 M citral in hexane. (b) Selectivity to reaction products as a function
of citral conversion under identical reaction conditions.
the concentration profiles for reaction under standard con-
ditions at 298 and 373 K, respectively. Consistent with the
results in Fig. 5, the temporal concentration profile for citral
at 298 K exhibits a sustained exponential-like decrease,
with time, while a linear decrease, i.e., zero-order behavior,
is observed for citral at 373 K (Fig. 8a). The concentrations
of the cis and trans isomers of citral were summed and
plotted in Figs. 7 and 8 since both isomers exhibit identical
concentration dependencies and activation energies on
Pt/SiO2, although the TOF ratio between the E and Z
isomers is 1.5. The selectivity to the UALC increases from
9 to 19% during the first 50% of citral conversion at 298 K,
as contrasted with an increase from 40 to 80% for reaction
at 373 K. The increase in selectivity to UALC at 373 K is
accompanied by a decrease in the selectivity to the PSALD
from approximately 40 to 5% during the first 50% conver-
sion of citral. The selectivity to UALC decreases after 50%
citral conversion at 373 K due to its further hydrogenation
FIG. 7. (a) Concentration profile for reactants and products during
citral hydrogenation over 1.44% Pt/SiO2 at 298 K, 20 atm H2 pressure,
and 1 M citral in hexane. (b) Selectivity to reaction products as a function
of citral conversion under the identical reaction conditions.

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
TABLE 3
171
Effect of Reaction Temperature on Product Selectivity (mol%)
during Reaction over 1.44% Pt/SiO2 at 20 atm H2 with 1 M Citral
in Hexane: 30% Citral Conversion
Temperature (K) UALC IP PSALC PSALD DCAL SAT ENAL
298
373
423
21
72
76
3
5
5
24
3
5
26
14
13
2
0
0
11
6
1
13
0
0
to PSALC. The selectivity to SAT at 373 K decreased as
the citral conversion increased due to the fact that this
product was formed only during the initial stages, and its
concentration stayed constant with time even though the
concentrations of other products increased. Thus, the nu-
merator in Eq. [1] for SAT is constant with increasing citral
conversion while the denominator increases with increasing
citral conversion, resulting in a decrease in the cumulative
selectivity to SAT with increasing citral conversion. For re-
action at 298 K, on the other hand, the selectivity to PSALC
and SAT increased with increasing citral conversion while
the selectivity to IP decreased with increasing conversion.
Table 3 lists the selectivities to the various products at 30%
citral conversion for reaction over 1.44% Pt/SiO2 under
standard conditions and temperatures from 298 to 423 K.
The high initial activity at 373 K during the first 5 min of
reaction was associated primarily with the production of
PSALD, and after this short period, the rate of formation
of PSALD was suppressed while that for UALC increased.
More insight into this unique temperature behavior was
obtained by performing a series of reactions under standard
conditions with 1.44% Pt/SiO2 and changing the tempera-
ture during the course of the reaction. Figures 9a and 9b
depict the respective temporal citral conversion profiles for
experiments in which the reaction was initially conducted
at 373 K for 5 h prior to lowering the temperature to 298 K
and vice versa. Figure 9a shows that decreasing the reac-
tion temperature from 373 K, after 5 h of reaction and 17%
citral conversion, to 298 K resulted in complete loss of ac-
tivity. Figure 9b shows that the TOF at 298 K exhibited
the usual decrease in the rate of citral disappearance and
increasing the reaction temperature to 373 K resulted in
a TOF for citral disappearance identical to that observed
when the reaction was conducted at 373 K for the entire
course of the reaction. The product distribution during the
experiment where the reaction temperature was increased
from 298 to 373 K midway though the reaction is displayed
in Fig. 10. A drop in the selectivity to PSALD, SAT, and
DCAL was accompanied by an increase in selectivity to
UALC and PSALC after increasing the reaction tempera-
ture from 298 K during the first 4 h of reaction to 373 K.
Similar experiments were also conducted with 3.8% Pt/
SiO2 catalyst (H/Pt = 0.67) at temperatures between 373
FIG. 9. (a) Temporal citral conversion profile during reaction over
1.44% Pt/SiO2 (0.80 g) at 20 atm H2 with 1 M citral in hexane (a) initially
at 373 K prior to decreasing the reaction temperature to 298 K and (b)
initially at 298 K prior to increasing to 373 K after 4 h of reaction.
ral conversion profiles for the reaction initially conducted
at 373 K prior to increasing the reaction temperature to
423 K and vice versa. Increasing the reaction temperature
1
from 373 to 423 K yielded a TOF of 0.04 s at 423 K
FIG. 10. Product distribution during the reaction initially at 298 K,
20 atm H2, and with 1 M citral in hexane prior to increasing the reaction
and 423 K, and Figs. 11a and 11b display the temporal cit- temperature to 373 K after 4 h of reaction and 50% citral conversion.

172
SINGH AND VANNICE
ance at 373 and 423 K was retained regardless of the
initial reaction temperature; however, there was no ac-
tivity at 298 K when the initial reaction temperature was
373 K.
DISCUSSION
Influence of Mass Transfer and Poisoning Effects
It is imperative in liquid-phase reactions that heat and
mass transfer effects as well as poisoning be shown to be
absent. The Madon–Boudart test wasconducted to discount
artifacts due to heat and mass transfer limitations and poi-
soning on the measured kinetics of citral hydrogenation. It
offers an efficient and reliable experimental technique for
verifying the absence of heat and mass transfer limitations
because the activity for a reaction is proportional to the
number of active sites which, for a structure-insensitive re-
action in the absence of poisoning effects, is proportional
to the concentration of surface metal atoms (24). Previous
studies had shown that no external mass transport limita-
tions existed at stirring speeds above 500 rpm; all experi-
ments here were conducted at 1000 rpm (23). The appar-
ent structure sensitivity suggested for these types of reac-
tions complicated the use of the Madon–Boudart test (26);
however, this difficulty was overcome by using two cata-
lysts with essentially equal dispersions of 0.41 and 0.45, as
measured by H2 chemisorption at 300 K. These two cata-
lysts yielded similar TOFs at 373 and 393 K within exper-
imental uncertainty (see Table 2), which indicated the ab-
sence of any transport limitations as well as poisoning ef-
fects due to impurities in the liquid phase. Furthermore,
the constant turnover frequency for H2 uptake at 298 K
for catalysts with an order-of-magnitude variation in the
concentration of surface Pt atoms and the identical tem-
poral H2 uptake curves at 298 K for an eightfold change
in catalyst loading are further indications that these ki-
netic data are free of transport limitations and poisoning
effects.
Since citral was used as received, with a purity of 97% ,
one must also be concerned about possible poisoning due
to impurities in the reactant source. However, this was not
a problem because a sixfold variation in citral concentra-
tion did not affect the activity or selectivity versus conver-
sion behavior between 298 and 423 K, as indicated by the
zero-order dependence of the initial rate on citral concen-
tration (21). The difference in conversion between the runs
at 298 K with an eightfold variation in catalyst loading (see
Figs. 3a and 3b) is attributed to a concurrent inhibition by
UALC, one of the intermediate products in the reaction,
and not to poisoning effects. The possible leaching of met-
als from the reactor walls was also considered, but such an
effect was discounted not only because catalyst poisoning
by the leached metals would result in the Madon–Boudart
criterion not being satisfied and the temporal H2 uptake
FIG. 11. (a) Temporal citral conversion profile during reaction over
3.8% Pt/SiO2 (0.30 g) at 20 atm H2 with 1 M citral in hexane (a) initially
at 373 K prior to increasing to 423 K after 4 h of reaction and (b) initially
at 423 K prior to decreasing to 373 K after 4 h of reaction.
which was identical to that obtained when the experiment
was performed at 423 K for the entire course of the reac-
tion. Similarly, decreasing the reaction temperature from
1
423 to 373 K yielded a TOF of 0.007 s at 373 K which
was identical to that obtained when the reaction was con-
ducted entirely at 373 K. The selectivity-versus-citral con-
version profile obtained during these latter temperature
experiments was similar to that exhibited when the reac-
tion was performed in its entirety at each of the initial
temperatures. The results of these experiments involving
increasing and decreasing reaction temperature are sum-
marized in Table 4. The observed rate of citral disappear-
TABLE 4
Relative Rates on 1.44% Pt/SiO2 following a Sequence of Re-
action Temperatures with 1 M Citral in Hexane and 20 atm H2
Pressurea
Tinitial → TFinal
Relative rate
298 K → 373 K
373 K → 298 K
373 K → 423 K
423 K → 373 K
(Rate-373 K)final/(rate-373 K)initial = 1
(Rate-298 K)final/(rate-298 K)initial = 0
(Rate-423 K)final/(rate-423 K)initial = 1
(Rate-373 K)final/(rate-373 K)initial = 1
a Based on Figs. 9 and 11.

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
173
curves for an eightfold variation in catalyst loading not be- did not consider the possible isomerization of unsaturated
ing identical, but also because elemental analysis of hexane, alcohol to saturated aldehyde; therefore, the relative rate
==
used as a solvent for benzene hydrogenation, showed that enhancement for hydrogenation of the C O bond may be
the concentration of any leached metals on either the solid even larger. Considering these arguments, it is surprising
catalyst or in the liquid phase was below the detection limit that, despite the steric constraints associated with the citral
(21, 23). Consequently, all experimental results, combined molecule, a 40% selectivity for hydrogenation of only the
==
with utilization of the Weisz–Prater criterion, consistently conjugated C C bond to yield PSALD and a 15% selec-
==
verified the absence of artifacts due to internal or external tivity for hydrogenation of the trisubstituted, isolated C
C
diffusion limitations or to poisoning.
bond to yield DCAL and SAT can be obtained at low cit-
ral conversions, i.e., less than 20% . The results are even
more striking when one considers that no isomerization of
UALC to PSALD was observed in the present study, even
during hydrogenation of the pure intermediate feeds, i.e.,
geraniol and nerol (the cis and trans isomers of UALC).
These results indicate that the conjugated bonds in citral
participate in adsorption and that steric constraints, while
present, do not play a dominant role in determining the
product distribution over these unmodified clean Pt/SiO2
catalysts. Coordination via the sterically hindered site was
reported in a RAIRS study of crotonaldehyde adsorption
on Pt(111) surfaces (30). However, steric constraints may
come into play during the course of the reaction due to
surface modification (31).
The product distribution during citral hydrogenation was
strongly dependent on the reaction temperature. All results
are under standard conditions unless otherwise noted. Re-
actions conducted at 373 and 423 K gave high selectivities
to the primary products, i.e., UALC and PSALD, whereas
hydrogenation at 298 K yielded numerous secondary prod-
ucts due to readsorption and further reaction of the primary
products. This is evidenced by higher selectivity to PSALC
and SAT at 298 K compared with 373 K (Figs. 7, 8). The
selectivity to PSALD at 298 K was approximately 40% at
20% citral conversion and 30% at 50% citral conversion.
In contrast, reaction at 373 K initially yielded a selectivity
to PSALD of approximately 40% , which decreased to 10%
after 20% citral conversion and to less than 5% after 50%
citral conversion (Fig. 8b). The sharp decrease in the se-
lectivity to PSALD at 373 K was due to its formation only
during the first few minutes of reaction, after which period
its net rate of formation was suppressed and the rate of
UALC formation was enhanced. The selectivity to UALC
at 298 K increased from 9 to 19% as the citral conversion
increased from 20 to 50% ; however, at 373 K the UALC
selectivity increased from 40 to 80% as the citral conver-
sion grew from 5 to 50% . Thus, it appears as though there
is a surface modification that occurs under reaction con-
ditions whose formation is suppressed at 298 K compared
with 373 K.
Effect of Pretreatment Procedure
The results in Fig. 4 clearly show that the pretreatment
procedure can have a pronounced effect on the reaction
rate because the catalyst reduced ex situ exhibited a TOF
that was only one-third of that for the catalyst reduced
in situ; however, the two different pretreatment procedures
did not influence the product distribution. While treatment
at 373 K in flowing H2 is sufficient to reactivate a catalyst
previously reduced at 673 K in hydrogen (27); the pres-
ence of solvent during the reactivation process may influ-
ence the state of the catalyst at the start of the reaction.
Evidence of solvent effects during the reduction process
was obtained by Busser et al., who studied the reduction
of polymer-stabilized Rh particles in various solvents and
found the average metal particle size to vary fourfold de-
pending on the nature of the solvent (28). Solvent effects
during the reactivation process can also influence the cata-
lyst by adsorbing competitively with hydrogen or by solvent
decomposition (29). The lower TOF may also be due to the
smallamountsofoxygen present on the Pt surface and in the
liquid feed that could form water, which may alter or block
part of the catalyst surface. Although the oxygen concen-
tration in the liquid phase is low, Augustine and Tanielyan
have shown that oxygen dissolved in the solvent can modify
the kinetics of ethyl pyruvate hydrogenation (20). There-
fore, any complications that might arise from these effects
were avoided by reducing the catalyst in situ and degassing
citral and hexane with N2 to remove any dissolved oxygen.
Effect of Reaction Temperature on Product Distribution
The conjugated functionality in citral may have consid-
erable steric constraints imposed on its adsorbed state be-
cause of the methyl and long-chain aliphatic substituents
that are not present in smaller molecules such as acrolein
and crotonaldehyde. Steric constraints imposed on the con-
==
==
jugated C C and the trisubstituted isolated C C bonds
should make hydrogenation of these functionalities more
difficult. In fact Beccat et al. have shown by a compari-
son of crotonaldehyde and methylcrotonaldehyde that the
presence of a methyl group in place of a hydrogen on the
Gallezot and co-workers observed similar changes in
product distribution during liquid-phase hydrogenation
of cinnamaldehyde over Pt and Ru catalysts. They sug-
gested that both geometric and electronic effects induced
bycompetitive hydrocinnamaldehyde (saturated aldehdye)
==
conjugated C C functionalitydecreasesthe hydrogenation
==
rate of the C C bond fourfold and increases the rate of hy-
==
drogenation of the C O bond 4-fold (6). These authors

174
SINGH AND VANNICE
adsorption were responsible for the selectivity-versus- in greater, and hence citral conversion is greater, for the
conversion behavior. The geometric effect was proposed reaction with the greater catalyst mass (21). The source of
to be due to the surface ligands imposing constraints on the the inhibition reaction was believed to be a decomposition
adsorption of cinnamaldehyde such that adsorption only of UALC. A number of studies have been reported in the
through the carbonyl bond was allowed, while the elec- literature that have shown the existence of decomposition
tronic effect involved donation of electrons from the lig- reaction for such systems yielding chemisorbed CO and
ands, thus increasing the electron density at the surface carbonaceous species (31–33). Furthermore, the work of
==
and inhibiting adsorption via the conjugated C C bond Lercher and co-workers for liquid-phase hydrogenation of
(31). Lercher and co-workers have suggested that forma- crotonaldehyde indicated that purging a deactivated cata-
tion of butyraldehyde during liquid-phase hydrogenation of lyst with air to oxidize CO to CO2 regenerated the initial
crotonaldehyde is accompanied by deactivation of sites re- hydrogenation activity that was observed prior to deactiva-
sponsible for butyraldehyde formation (19). Thisisin agree- tion (19).
ment with IR studies of 3-methyl-2-butenal which show
that the formation of saturated aldehyde is accompanied Effect of Temperature on Reaction Rate
by subsequent decarbonylation to yield adsorbed CO and
As illustrated in Fig. 6, an activity minimum is observed
carbonaceous deposits on the surface (32). Thus, several
between 298 and 423 K which occurred over the entire
studies have suggested surface modifications occur under
pressure range of 7 to 20 atm. Lercher and co-workers
reaction conditions that can alter selectivity, and a majority
have reported a decreasing reaction rate with increasing
of these studies have indicated that PSALD may be respon-
temperature for crotonaldehyde hydrogenation (19); how-
sible for the modification. However, in view of the results
ever, to our knowledge, an activity minimum for this or any
in our subsequent article, the modification is attributed to
other liquid-phase hydrogenation reaction has not been re-
UALC rather than PSALD (21).
ported in the literature. An activity maximum, on the other
It is worth noting that a number of bimetallic systems
hand, has been frequently reported for vapor-phase hydro-
have been investigated for selective hydrogenation of , -
genation reactions, and it can be readily attributed to the
unsaturated aldehydes, with the purpose of enhancing se-
competition between an increasing rate constant and a de-
lectivity to the UALC, because it was thought that sim-
creasing adsorption equilibrium constant as the tempera-
pler monometallic catalysts such as Pt/SiO2 exhibited a low
ture increases (34, 35). This explanation cannot explain an
selectivity to UALC. However, in view of the results pre-
activity minimum, however. Lercher and colleagues have
sented in this paper, it is apparent that the effect of reaction
suggested that the activity decrease with increasing temper-
parameters on product distribution can be significant and
ature is due to decarbonylation of butyraldehyde occurring
can alter the selectivities as much or more than the addition
concurrently with crotonaldehyde hydrogenation, and they
of a second metal.
speculated that the former reaction may have a higher ac-
tivation energy compared with the latter reaction; thus the
relative inhibition due to site blockage would increase at
Deactivation Process via Inhibition
There is more than an order of magnitude drop in the re- higher reaction temperatures (19). Regardless, this mech-
action rate during the first 4 h of reaction at 298 K, as shown anism alone is not capable of explaining either an activity
in Fig. 3. This deactivation process cannot be attributed to minimum or the differences in deactivation behavior ob-
transport limitations or poisoning effects, as discussed be- served at the different temperatures in the present study.
fore. Figures3a and 3b showthat, despite an eightfold differ-
We propose that the activity minimum is caused by the
ence in catalyst loading, the temporal H2 uptake curves are decomposition of UALC combined with CO desorption.
identical while the H2 uptake-versus-citral conversion pro- Our choices for the reactant that decomposes are nerol
files are significantly different. Such behavior is consistent and geraniol, rather than an aldehyde suggested in previous
with a concurrent side reaction yielding inhibiting products studies (19, 30, 32). This conclusion is based on the results
that accumulate on the catalyst surface on the time scale discussed in the subsequent paper (36). Briefly, these data
of the reaction (21). If the reaction were zero order with indicate zero-order behavior for hydrogenation of pure
no inhibition reaction, then the temporal TOF profile and PSALD at 298 and 373 K with little deactivation, which
the TOF-versus-conversion profile would be identical for contrasts noticeably with the sharp decrease in the rate
the two runs with an eightfold different catalyst loading. observed during UALC hydrogenation at 298 and 373 K.
The identical TOF-versus-time profile in Fig. 3a is exhib- Furthermore, simultaneous hydrogenation of PSALD and
ited because the fraction of sites occupied by the inhibiting UALC resulted in a continuous decrease in the rate of dis-
species is identical for the two runs with different catalyst appearance of PSALD in contrast to the stable rate ob-
loading, and therefore, the turnover frequency is constant tained for hydrogenation of PSALD alone, which strongly
with time. However, due to the inhibition reaction the total suggests a deactivation phenomenon due to UALC (36).
number of sites available for reaction at each point in time Further indirect evidence for UALC as the source of CO

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
175
lies in the fact that citral hydrogenation at 373 K initially oc- and CO desorption. It was implicitly assumed in the model
curs at a very high reaction rate which rapidly decreases to a that the concentrations of products arising from the alcohol
psuedo-steady-state behavior as soon as UALC is formed. decomposition reaction, i.e., organic fragment and CO, are
It is possible that citral decomposition may also play a role very small.
in inhibition and such a reaction cannot be discounted based
Reactions1–5in Fig. 1are the principalonesin our system
on the results presented in the present study; nevertheless, and were modeled using the following five mole balances
a decomposition reaction is believed to occur under the for the species of interest:
present reaction conditions. The CO formed from this re-
dCCITRAL
action begins to desorb at higher reaction temperatures,
=
r1 r2 r3,
[2a]
[2b]
[2c]
[2d]
[2e]
dt
dCENAL
i.e., 373 K and above. TPD studies of aldehyde and alcohol
decomposition on Pd and Rh as well as ketene decompo-
sition on single-crystal Pt surfaces indicate that the decar-
bonylation product, CO, begins to desorb from the surface
at 350 K, with a peak maximum around 450 K (33, 37–44).
At 298 K, CO desorption from Pt essentially does not
occur; therefore, the reaction rate gradually decreases with
time due to buildup of adsorbed CO on the time scale of
the reaction, eventually leading to complete loss of activity
(Figs. 6, 7a, 9a). Increasing the temperature to 373 K results
in the rate enhancement for both the alcoholdecomposition
and CO desorption processes, and a psuedo-steady state is
readily established for the CO coverage on Pt which results
in a stable activity during reaction at 373 K. Furthermore,
normal Arrhenius behavior is observed between 373 and
423 K (Figs. 6, 11) since the desorption rate of CO is high
and a steady-state Pt surface with a lower CO coverage is
again rapidly obtained. Such a mechanism would imply that
the activation barrier for decomposition reaction should be
lower than that for CO desorption. Indeed, BOC theory
estimates of the activation barrier associated with decar-
bonylation of acetyl species, which have been proposed as
intermediates in alcohol decomposition, range from 15 to
26 kcal/mol depending on the mechanism (21, 45), while the
activation barrier for CO desorption from Pt(111) surfaces
has been reported to be approximately 26 kcal/mol (46,
47). The activation barrier estimated for decomposition on
Pt surfaces is consistent with the value of 16 kcal/mol ob-
tained for propenoyl decomposition on a Pd(111) surface
(36). The mechanism of the decomposition reaction is dis-
cussed in greater detail later; however, it is important to
emphasize that the activity minimum can be explained by
the relative activation barriers for alcohol decomposition
and CO desorption.
= r1,
dt
dCUALC
= r2,
dt
dCPSALD
= r3 r5,
= r5,
dt
dCPSALC
dt
where Ci and rj are the concentration of species i and reac-
tion rate of step j in Fig. 1, respectively. Equations [2c] and
[2e] were formulated based on the assumption that hydro-
genation of geraniol to citronellol, i.e., r4, is kinetically in-
significant under citral hydrogenation reaction conditions.
Thiswasverified experimentallyasshown in the subsequent
paper by addition of either geraniol or citronellal to a 1 M
citral in hexane hydrogenation reaction mixture at 373 K.
Addition of geraniol resulted in negligible change in the net
rates of formation and disappearance of geraniol and cit-
ronellol, respectively. Addition of citronellal, on the other
hand, to a similar reaction mixture resulted in a significant
enhancement in the net rate of disappearance of citronellal
and formation of citronellol and isopulegol (36). Therefore,
the primary route to citronellol formation appears to be via
citronellal hydrogenation. Only the organic species listed
above were used in this simplified model because they con-
stitute approximately 95% of the total products (Figs. 7,
8). An identical sequence of elementary steps comprising
a simple Langmuir–Hinshelwood model describes each of
the reactions in steps 1–5, i.e.,
KH
H2 + S
2H-S,
[3a]
[3b]
KC
K1i
Org + S
Org-S,
Kinetic Modeling
Org-S + H-S
OrgH-S + H-S
OrgH-S + S,
OrgH2-S + S,
[3c]
[3d]
The mechanism proposed above would suggest the pres-
ence of a hydrocarbon with one carbon atom less than
UALC, and evidence for this species in trace quantities was
obtained via GC–MS; however, analytical limitations pre-
ki
1/KP
OrgH2-S
OrgH2 + S,
[3e]
vented resolution of this species in our gas chromatograms where Orgisthe organicreactant, i.e., citralin reactions1, 2,
(21). The reaction sequence proposed previouslyand shown and 3 and PSALD in reaction 5. The terms S, Org-S, OrgH-
in Fig. 1 was modeled via a Langmuir–Hinshelwood mech- S, OrgH2-S, and H-S represent vacant active sites, adsorbed
anism for each hydrogenation step along with a sequence organic, adsorbed half-hydrogenated organic species, ad-
of irreversible steps to account for alcohol decomposition sorbed product, and adsorbed hydrogen, respectively.

176
SINGH AND VANNICE
In addition to the above sequence of steps for each hy- that PH remains constant during reaction and is therefore
2
drogenation reaction, it was also assumed that there is a lumped into the constant . Using the site balance along
0
concurrent inhibiting mechanism due to decarbonylation with the assumptions that 2H, 2PSALD, 2UALC, and 2C
of UALC to yield adsorbed CO and carbonaceous deposits 2Citral and 2CO, the following expression for the fractional
as follows:
surface coverage of sites is obtained:
KET
UALC-S + S
Etoxy-S + H-S,
[4a]
2Citral + 2S + 2CO = 1,
[7]
KACT
1
2CO
Etoxy-S + 2S
Acyl-S + S
CO-S
Acyl-S + 2H-S,
CO-S + C0-S,
CO + S,
[4b]
[4c]
[4d]
[4e]
2S =
;
1 + KCitralCCitral
kACT
kD
and the resulting rate expression is
ki0 K1i KOrg KH PH COrg (1 2CO
2
kHC
)
C0-S + H-S
H2C0 + 2S.
2
2
ri =
,
[8]
2
(1 + KCitralCCitral
)
The first step isthe quasi-equilibrated dissociation ofUALC
to yield a surface ethoxide species and adsorbed hydrogen.
The surface ethoxide species then dehydrogenates to form
an acyl species (30), which subsequently undergoes C–C
bond scission to yield adsorbed CO and a carbonaceous
species with one carbon atom less than UALC. The ad-
sorbed CO can desorb and the carbonaceous species can
hydrogenate with rate constants kD and kHC, respectively.
The details of the mechanism and generalizations based on
other metals are discussed later.
Assuming adsorption of the reactants and formation
of the half-hydrogenated species are quasi-equilibrated,
the following respective expressions are obtained for the
surface coverage of hydrogen, the organic, and the half-
hydrogenated state:
where L is incorporated into the rate constant ki0. Fur-
thermore, using the inhibition process controlled by reac-
tions involving UALC decarbonylation and CO desorption,
shown in steps 14a–14e, the following differential equation
for the surface coverage of CO can be obtained after using
step 15d:
d2CO
= kACT2ACYL2S kD 2CO
,
dt
d2CO
dt
k
ACT(KACT KET
K
UALC)CUALC25S
=
kD 2CO , [9]
KH3/2 PH3/223S
2
d2CO
= kC0 O
C
UALC22S kD 2CO
.
dt
Therefore, our kinetic model consists of Eq. [8] and six dif-
ferential equations comprising Eqs. [2a]–[2e] and [9] sub-
ject to the initial conditions that CCitral = 1 M, COrg = 0 for
all reactants other than citral and 2CO = 0. These equa-
tions were solved numerically with six adjustable param-
2H = KH1/2 PH1/22S,
[5a]
[5b]
2
2
2Org = KOrg
C
Org2S,
2H
2OrgH = K1i 2Org
= K1i KOrg K1/2 PH1/2
C
Org2S, [5c]
H2
2
2S
eters, i.e., four lumped constants,
–
1
and 5, and the
3
2ETOXY22S
adsorption equilibrium constant for cit0ral and the appar-
2ACYL = KACT
ent rate constant for CO dissociation, k ACT. The value for
2H2
kD was calculated using results from the surface science
3/2
2
3/2 PH
K
KUALCCUALC2S
ET
literature, i.e., kD = 1013
e
(46). The numbers of
26,000/RT
= KACT
= KA0 CT
K
H2
data points used for fitting purposes at 298, 373, and 423 K
are 30, 44, and 32, respectively. Admittedly, this model re-
quires a large number of adjustable parameters; however,
they are essential to accurately account for the complex
reaction chemistry. Parameter estimation was performed
using the Powell method via a commercially available non-
C
UALC2S,
[5d]
where Ki and Ci are the adsorption equilibrium constant
and concentration of species i, respectively, and K1i and
K0ACT are the equilibrium constant for formation ofthe half-
hydrogenated species and a lumped apparent equilibrium
constant for formation of the acyl species, respectively. If
the addition of the second H atom is assumed to be the
rate-determining step, then the rate of the hydrogenation
reaction in Fig. 1 for component i is
linear regression software package named Scientist (Micro
6
Math Scientific Software) which has a tolerance of 10
.
The optimum fit of the above model to the concentration
profiles under standard conditions and 298, 373, and 423 K
is shown in Fig. 12, and the corresponding parameters are
listed in Table 5. When possible, the thermodynamic consis-
tency of these values was examined, as discussed later. All
the fitting was performed at a reactor pressure of 20 atm;
therefore, the adsorption equilibrium constant and the
ri = Lki K1i KOrg KH PH COrg2S2 = i COrg22S,
[6]
2
2
where the subscript Org represents the organic reactant in
each step and L is the concentration of active sites. Note

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
177
partial pressure of hydrogen were lumped in with other
constants.
Prior to discussing the modeling results it is important
to address the various assumptions involved. The model
was restricted to the consideration of reactions 1–3 and 5
at 298 K because the products of these five reactions con-
stituted 95% of the total products, and at 373 and 423 K
only reactions 2, 3, and 5 were considered because the con-
centration of ENAL was negligible at these temperatures.
Dissociative hydrogen adsorption on Pt is well established
(48, 49). A single type of active site was used, i.e., one that
can adsorb hydrogen as well as the organic molecules; such
an assumption is routinely made for these types of hydro-
genation reactions and is the simplest choice possible. The
first-order dependence on H2 implies a low surface cover-
age of H atoms. The addition of the second H atom as the
rate-determining step (rds) has been proposed previously
for the vapor-phase hydrogenation of ketones and alde-
hydes on metal surfaces (50, 51). This assumption allows
first-order dependence on hydrogen pressure, as observed
experimentally, in contrast to a maximum half-order de-
pendency that is predicted if addition of the first H atom is
assumed to be the rds.
The rate inhibition caused by site blockage due to CO, as
represented by steps 4a–4e, was invoked to explain the un-
usual temperature dependence. The need for such a postu-
late has been discussed previously; therefore, we now con-
centrate on the chemistry involved in these steps. Much
evidence exists that clearly shows that decomposition of
saturated aldehydes and unsaturated alcohols does ocur;
consequently, all these species may undergo a surface de-
composition reaction to different extents under our reac-
tion conditions. However, onlythe decomposition ofUALC
is considered here based on the reasons summarized earlier
and more clearly enumerated in a subsequent paper (21).
The elementary decomposition reactions proposed in the
FIG. 12. Fit of kinetic model to experimental data for reaction over present model are consistent with the work of Barteau and
1.44% Pt/SiO2 at 20 atm H2 with 1 M citral in hexane at (a) 298 K,
(b) 373 K, and (c) 423 K. Symbols represent experimental data while solid
lines represent the model prediction.
co-workers for alcohols on Pd surfaces (52, 53). Recently,
Carlos de Jesus and Zaera have confirmed that decarbony-
lation of acrolein and crotonaldehyde does indeed occur on
TABLE 5
Optimized Parameters for Eqs. [2], [8], and [9]
Ea or 1Ha0ds
Ln(A0) or 1Sa0ds
423 K
373 K
298 K
(kcal/mol)
(cal/mol/K)
1
1
1
1 = k10 K11 KCitral KH PH (h
)
)
)
0.00
1.50
0.28
4.86
4.60
0.00
27.63
8.47
89.35
15.98
62926
20.50
23.66
26.06
105.24
354.87
20.14
4887
2
2
2 = k20 K12 KCitral KH PH (h
2
2
3 = k30 K13 KCitral KH PH (h
2
2
1
5 = k50 K15 KPSALD KH PH (h
)
2
2
K
Citral (liters/mol)
19a
8
26b
36a
21
38b
kCO (liters/mol/h)
138216
1303.00
1
kD (h
)
0.003
a Enthalpies and entropies of adsorption were evaluated at a standard state of 1 atm in the gas phase.
b Value was fixed based on estimates from the literature (46, 47).

178
SINGH AND VANNICE
Pt surfaces to yield CO and volatile carbonaceous species hydrogen bond and 26 kcal/mol for cleavage of the C–H
(31, 54). Unfortunately, similar detailed studies regarding bond in a surface acetyl species to form adsorbed ketene
decomposition pathways of unsaturated alcohols on Pt sur- (21, 46). An alternate reaction pathway was proposed by
faces were not found and, therefore, the justification of the Barteau and co-workers for decomposition of acetaldehyde
present model is based heavily on studies with Pd. TPD on a Pd(110) surface in which the aldehydic hydrogen was
studies of allyl alcohol adsorption on a Rh(111) surface in- cleaved to form a surface acetyl species which subsequently
dicate desorption of CO, hydrogen, and allyl alcohol, and decomposed to CO and CH3 without going through the
no C2 hydrocarbons could be detected (42). In contrast, ketene intermediate (41). The apparent structure sensitiv-
similar studies of Pd(110) yielded CO, hydrogen, ethylene, ity of the decomposition reaction does not allow us to say
propanal, ethane, and allyl alcohol (33), while studies with a unequivocally which reaction occurs; therefore, for the sake
Pd(111) surface identified hydrogen, CO, propylene, ethy- of simplicity the acyl species has been assumed to decar-
lene, and water, which is indicative of C–O bond cleavage bonylate to yield CO and an alkyl species. The activation
and suggestive of an apparent structure sensitivity for alco- barrier for this step on Pt(111) surfaces was estimated from
hol decomposition on Pd surfaces (37, 52). In view of these bond-order conservation theory to be 15 kcal/mol (21). The
results, one must be careful about making broad generaliza- observed minimum in activitycan, in principle, be explained
tions regarding details of the decomposition reaction from using either of the two mechanisms proposed by Barteau
one metal to another. Furthermore, Rendulic and Sexton and co-workers as long as each activation barrier is lower
compared their dehydrogenation results under UHV con- than that for CO desorption. It is also possible that the
ditions with those on a supported Pt catalyst and concluded apparent structure sensitivity seen during decomposition
that extreme caution must be used in comparing data from of aliphatic aldehydes and alcohols on Pd surfaces may be
UHV conditions with those relevant at higher pressures related to the apparent structure sensitivity observed for
(55). Nevertheless, it is apparent that such side reactions citral hydrogenation during reaction at 373 K over Pt/SiO2
do occur on Group VIII metal surfaces and the general catalysts as shown in Table 1 (41, 52, 57).
results can be used to explain the unusual temperature be-
The reverse of the acyl decomposition reaction is simi-
havior. Therefore, for the purposes of the present study the lar to the hydroformylation reaction; thus, acyl decompo-
decomposition pathway of the unsaturated alcohol on Pt sition can, in principle, be a reversible reaction. However,
catalysts was assumed to be similar to that on Pd surfaces. this step was chosen to be irreversible since it is expected
Alcohol decomposition on Pd surfaces is believed to occur that the alkyl species should rapidly hydrogenate and the
via an aldehyde intermediate to yield a surface acyl species very low CO pressure would make the reverse reaction rate
which subsequently undergoes C–C bond scission to yield extremely slow. Evidence in the surface science literature
CO and carbonaceous species (53, 56); consequently, a sim- suggests high selectivity toward hydrogenation of the car-
ilar mechanism is proposed for UALC decomposition in bonaceous deposits to form saturated and unsatured alkyl
the present study. This raises the question that if the unsat- species rather than the formation of alkylidyne species on
urated alcohol decomposition occurs via an aldehyde inter- the surface. Shekhar and Barteau have reported a selectiv-
mediate then what precludes the initial unsaturated alde- ity of 88% for C2 hydrocarbon products during decomposi-
hyde i.e., citral, from also undergoing a decarbonylation tion of allyl alcohol in the presence of hydrogen, as opposed
reaction. This issue is addressed in the subsequent paper to 67% in the absence of hydrogen. In both cases C2 species
(36); however, at this point citral decarbonylation cannot desorbed below 300 K (33); therefore, it is logical to assume
be discounted although it is our assertion that the unsat- that the surface coverage of these species is negligible.
urated alcohol forms a precursor prior to decarbonylation
The kinetic model just discussed adequately describes
to yield CO and carbonaceous species. Furthermore, this the experimental data and is consistent with observations
precursor is speculated to be an acyl surface intermediate reported by numerous workers. Furthermore, the fitted pa-
based on the surface science studies of Davis and Barteau rameters yield thermodynamically consistent parameters
(56).
when such evaluation is possible. The thermodynamic con-
Davis and Barteau studied aldehyde decomposition on straints on the equilibrium adsorption constants have been
Pd(111) and concluded that decomposition of an acyl described before along with the proper choice of standard
species via a ketene intermediate was the rate-determining states (58, 59). Only the adsorption equilibrium constants
step rather than cleavage of the aldehydic C–H bond, and for citral and UALC could be evaluated because the other
the activation energy for the former step was estimated to parameters were coupled in such a manner as to disallow
be 16 kcal/mol during decomposition of propenoyl (39). extraction of the equilibrium constants. Table 5 lists the fit-
This is in general agreement with the trends for activa- ted values of the adjustable parameters and the thermody-
tion barriers obtained from bond-order conservation the- namic parameters obtained from the Arrhenius plots. The
ory for a Pt(111) surface which indicates activation barri- enthalpies and entropies of adsorption have been corrected
ers of 13 kcal/mol for cleavage of the aldehydic carbon– to a standard state of 1 atm in the gas phase by a procedure

LIQUID-PHASE HYDROGENATION OF CITRAL OVER Pt/SiO2, I
179
of activity. At higher reaction temperatures, i.e., 373 K, al-
cohol decomposition is more rapid, but the CO desorption
rate is significantly enhanced and a psuedo-steady state is
readily established which results in minimal additional in-
hibition and conventional Arrhenius behavior with an acti-
vation energy of 7 kcal/mol. The reaction kinetics for each
hydrogenation reaction in the reaction network were mod-
eled using a Langmuir–Hinshelwood model invoking dis-
sociative adsorption of hydrogen, competitive adsorption
between hydrogen and the reactive organic species, and
addition of a second H atom as the rate-determining step,
and the observed product distributions at each temperature
were described very well. Furthermore, deactivation due
to decomposition of the unsaturated alcohols geraniol and
nerol, along with CO desorption at higher temperatures,
was invoked to account for the unusual rate dependence
on temperature.
FIG. 13. Prediction of CO surface coverage as a function of citral
conversion for reaction at 20 atm H2 with 1 M citral in hexane at 298, 373,
and 423 K using Eqs. [2], [8], and [9] with parameters in Table 5.
ACKNOWLEDGMENT
described earlier (23). Due to the large number of fitting
parameters in the model, only a limited analysis of the re-
sulting fitting parameters is possible. The enthalpy of citral
adsorption citral was evaluated to be 19 kcal/mol, which
is quite reasonable, and the entropy loss of 36 e.u. during
citral adsorption was well below the absolute entropy of 83
e.u. calculated for citral in the gas phase at 298 K (21, 58,
59, 60).
This study was supported by the DOE, Division of Basic Energy Sci-
ences, under Grant DE-FE02-84ER13276.
REFERENCES
1. Sheldon, R. A., J. Mol. Catal. A 107, 75 (1996).
2. Mercadante, L., Neri, G., Milone, C., Donato, A., and Galvagno, S.,
J. Mol. Catal 105, 93 (1996).
3. Gallezot, P., and Richard, D., Catal. Rev. Sci. Eng. 40, 81 (1998).
4. Ponec, V., Appl. Catal. A 149, 27 (1997).
The model described above was used to predict the sur-
face coverage of CO under standard reaction conditions at
298, 373, and 423 K, and the results are shown in Fig. 13.
It should be stressed that only relative comparison of the
surface coverage at different reaction temperatures should
be undertaken. The surface coverage of CO at 298 K starts
out slowly but eventually reaches 0.9 at 50% citral con-
version. At 10% citral conversion, the surface coverage of
CO at 373 K is an order of magnitude higher than that at
298 K, which is consistent with our assertion that the lower
reaction rate at 373 K is due to blockage of active sites by
CO. Furthermore, at 10% conversion the surface coverage
of CO at 423 K is lower than that at 373 K because of the
enhanced CO desorption rate at this temperature, and con-
ventional Arrhenius behavior is observed above 373 K.
5. Kulson, P., and Cerveny, L., Appl. Catal. A 128, 13 (1995).
6. Beccat, P., Bertolini, J. C., Gauthier, Y., Massardier, J., and Ruiz, P.,
J. Catal. 126, 451 (1990).
7. Vannice, M. A., and Sen, B., J. Catal. 115, 65 (1989).
8. Patil, A., Banares, M. A., Lei, X., Fehlner, T. P., and Wolf, E., J. Catal.
159, 458 (1996).
9. Birchem, T., Pradier, C. M., Berthier, Y., and Cordier, G., J. Catal. 146,
503 (1994).
10. Coq, B., Figueras, F., Geneste, P., Moreau, C., Moreau, P., and
Warawdekar, M., J. Mol. Catal. 78, 211 (1993).
11. Marinelli, T. B. L. W., Nabuurs, S., and Ponec, V., J. Catal. 151, 431
(1995).
12. Marinelli, T. B. L. W., and Ponec, V., J. Catal. 156, 51 (1995).
13. Touroude, R., J. Catal. 65, 110 (1980).
14. Birchem, T., Pradier, C. M., Berthier, Y., and Cordier, G., J. Catal. 161,
68 (1996).
15. Galvagno, S., Milone, C., Donato, A., Neri, G., and Pietropaolo, R.,
Catal. Lett. 17, 55 (1993).
16. Neri, G., Mercadantge, L., Milone, C., Pietropaolo, R., and Galvagno,
S., J. Mol. Catal. A 108, 41 (1996).
SUMMARY
17. Neri, G., Milone, C., Donato, A., Mercadante, L., and Visco, A. M.,
J. Chem. Technol. Biotechnol. 60, 83 (1994).
18. Englisch, M., Ranade, V. S., and Lercher, J. A., J. Mol. Catal. 121, 69
(1997).
19. Englisch, M., Ranade, V. S., and Lercher, J. A., Appl. Catal. A 163, 111
(1997).
20. Augustine, R. L., and Tanielyan, S. K., J. Mol. Catal. A 118, 79 (1997).
21. Singh, U. K., Ph.D. thesis, Pennsylvania State University, in progress.
22. Palmer, M. B., and Vannice, M. A., J. Chem. Technol. Biotechnol. 30,
205 (1980).
Citral hydrogenation was studied under kinetic condi-
tions free of transport limitations and poisoning effects.
The rate of citral hydrogenation exhibited a minimum in
activity between 298 and 423 K, and this is attributed to the
relative activation energies for the decomposition of the
unsaturated alcohol and desorption of CO. At low reaction
temperatures, i.e., 298 K, the rate of citral disappearance is
low and CO slowly accumulates under reaction conditions
to block active sites and eventually lead to a complete loss
23. Singh, U. K., and Vannice, M. A., AIChE J. 45, 1059 (1995).

180
SINGH AND VANNICE
24. Madon, R. J., and Boudart, M., Ind. Eng. Chem. Fundam. 21, 438 42. Brown, N. F., and Barteau, M., J. Am. Chem. Soc. 114, 4258 (1992).
(1982).
43. Radloff, P. I., Mitchell, G. E., Greenlief, C. M., and White, J. M., Surf.
Sci. 183, 377 (1987).
44. Mitchell, G. E., Radloff, P. L., Greenlief, C. M., Henderson, M. A., and
White, J. M., Surf. Sci. 183, 403 (1987).
25. Satterfield, C. N., “Mass Transfer in Heterogeneous Catalysis,” Robert
E. Krieger, New York, 1981.
26. Englisch, M., Jentys, A., and Lercher, J. A., J. Catal. 166, 25 (1997).
27. Coq, B., Crabb, E., Warawdekar, M., Bond, G. C., Slaa, J. C., Galvagno,
45. Shustorovich, Adv. Catal. 37, 101 (1990).
S., Mercadante, L., Ruiz, J. G., and Sierra, M. C. S., J. Mol. Catal. 92, 46. McCabe, R. W., and Schmidt, L. D., Surf. Sci. 66, 101 (1977).
107 (1994).
47. Ertl, G., Neumann, M., and Streit, K. M., Surf. Sci. 64, 393 (1977).
48. Aben, P. C., vanDer Eizk, H., and Oelderik, in “Proceedings, 5th Inter-
national Congress on Catalysis, Palm Beach, 1972” (J. W. Hightower,
Ed.), p. 717. North Holland, Amsterdam, 1973.
28. Busser, G. W., van Ommen, J. G., and Lercher, J. A., “Advanced Cata-
lysts and Nanostructured Materials.” Academic Press, San Diego,
1996.
29. Mallat, T., Bodner, B., Borszeky, K., and Baiker, A., J. Catal. 168, 183 49. Tsuchiya, S., Amensmiya, Y., and Cvetanovic, R. J., J. Catal. 19, 215
(1997).
(1970).
30. Carlos de Jesus, J., and Zaera, F., Surf. Sci. 430, 99 (1999).
31. Gallezot, P., Giroir-Fendler, A., and Richard, D., in “Chemical Indus-
try 47” (Catalysis of Organic Reactions) (W. E. Pascoe, Ed.), 1992.
32. Waghray, A., and Blackmond, D. G., J. Phys. Chem. 97, 6002 (1993).
33. Shekhar, R., and Barateau, M., Surf. Sci. 319, 298 (1994).
34. Chou, P., and Vannice, M. A., J. Catal. 107, 129 (1987).
35. Chou, P., and Vannice, M. A., J. Catal. 107, 140 (1987).
36. Singh, U. K., Sysak, M., and Vannice, M. A., J. Catal. 190, (2000).
37. Davis, J. L., and Barteau, M., J. Mol. Catal. 77, 109 (1992).
38. Brown, N. F., and Barteau, M., Langmuir 8, 862 (1992).
39. Davis, J. L., and Barteau, M., J. Am. Chem. Soc. 111, 1782 (1989).
40. Houtman, C. J., and Barteau, M., J. Catal. 130, 528 (1991).
41. Shekhar, R., Barteau, M., Plank, R., and Vohs, J., J. Phys. Chem. 101,
7939 (1997).
50. Sen, B., and Vannice, M. A., J. Catal. 113, 52 (1988).
51. Dandekar, A. B., Ph.D. thesis, Pennsylvania State University, 1998.
52. Shekhar, R., and Barteau, M., Catal. Lett. 31, 221 (1995).
53. Davis, J. L., and Barteau, M., Surf. Sci. 187, 387 (1987).
54. Carlos deJesus, J., and Zaera, F., J. Mol. Catal. A 138, 237 (1999).
55. Rendulic, K. D., and Sexton, B. A., J. Catal. 78, 126 (1982).
56. Davis, J. L., and Barteau, M. Surf. Sci. 235, 235 (1990).
57. Singh, U. K., and Vannice, M. A., in “Proceedings, 12th International
Congress on Catalysis,” in press.
58. Boudart, M., AIChE J. 18, 465 (1972).
59. Vannice, M. A., Hyun, S. H., Kalpakci, B., and Liauh, W. C., J. Catal.
56, 358 (1979).
60. Reid, R. C., Prausnitz, J. M., and Sherwood, T. K., “The Properties of
Gases and Liquids.” McGraw–Hill, New York, 1977.