The hydrogenation of o-, m-, and p-xylene over Ni/SiO2

Article abstract of DOI:10.1006/jcat.1997.1527

The gas phase hydrogenation of o-, m-, and p-xylene was studied over a Ni/SiO₂ catalyst prepared by homogeneous precipita tion/deposition. The hydrogenation of each xylene yielded steroisomeric product mixtures of the saturated dimethylcyclohexane. The stereospecificity of the reaction is related to the nature of the reactant/catalyst interaction which governs the mode of addition of hydrogen to the carbons bearing the methyl substituents. The appearance of a common well-defined reversible maximum (T_max) in the rate vs. temperature plots is reported and discussed. Reaction orders with respect to xylene partial pressures are plotted as a function of reaction temperature. The temperature dependence of the rate constants was fitted to an Arrhenius equation and generated positive (T ≤ T_max) and negative (T ≥ T_max) activation energies. The derivation of true activation energies and heats of adsorption from the kinetic data is presented. Turnover frequencies at a particular temperature decreased in the order p-xylene > m-xylene > o-xylene. The respective roles of steric and electronic effects in determining the strength of adsorption and surface reactivity are discussed. A compensation effect, which is established for the experimentally determined or apparent kinetic parameters, is attributed to variations in the temperature dependencies of the surface concentration of the reactive species.

Full text of DOI:10.1006/jcat.1997.1527

JOURNAL OF CATALYSIS 166, 347–355 (1997)
ARTICLE NO. CA971527
The Hydrogenation of o-, m-, and p-Xylene over Ni/SiO2
Mark A. Keane
Department of Chemical Engineering, The University of Leeds, Leeds LS2 9JT, United Kingdom
Received July 31, 1996; revised November 4, 1996; accepted November 6, 1996
of the xylenes over diverse catalyst systems have been re-
The gas phase hydrogenation of o-, m-, and p-xylene was stud- ported. However, the xylene hydrogenation activity has,
ied over a Ni/SiO2 catalyst prepared by homogeneous precipita- in the main, been presented as a percentage conversion
tion/deposition. The hydrogenation of each xylene yielded steroiso-
meric product mixtures of the saturated dimethylcyclohexane. The
stereospecificity of the reaction is related to the nature of the re-
actant/catalyst interaction which governs the mode of addition of
hydrogen to the carbons bearing the methyl substituents. The ap-
catalysts. The results presented in this paper are a natu-
pearance of a common well-defined reversible maximum (Tmax) in
the rate vs. temperature plots is reported and discussed. Reaction
orders with respect to xylene partial pressures are plotted as a func-
tion of reaction temperature. The temperature dependence of the
or a rate relative to the rate of benzene hydrogenation.
Rahaman and Vannice (22) have recently provided the first
steady-state specific gas phase activities and activation en-
ergies for a series of unsupported and supported palladium
ral continuation of earlier investigations of benzene and
toluene hydrogenation over supported nickel systems (23–
2
6) and represent, to the best of the author’s knowledge,
rate constants was fitted to an Arrhenius equation and generated the only steady-state kinetic study of o-, m-, and p-xylene
positive (T � Tmax) and negative (T � Tmax) activation energies. The hydrogenation on Ni/SiO2. The objective of this report is
derivation of true activation energies and heats of adsorption from to establish the role of structural isomerism in determining
the kinetic data is presented. Turnover frequencies at a particular
temperature decreased in the order p-xylene > m-xylene > o-xylene.
The respective roles of steric and electronic effects in determining
activity and selectivity in catalytic aromatic hydrogenation
systems.
the strength of adsorption and surface reactivity are discussed. A
compensation effect, which is established for the experimentally de-
METHODS
termined or apparent kinetic parameters, is attributed to variations
in the temperature dependencies of the surface concentration of the
The catalyst wasprepared bythe homogeneousprecipita-
tion/deposition of nickel onto a nonporous microspheroidal
reactive species.
�c 1997 Academic Press
2
� 1
Cab–O–Sil 5M silica of surface area 194 m g as has been
described in detail elsewhere (27). The catalyst precursor
has a nickel loading of 1.5% w/w and low water content
(
<1% w/w). The hydrated catalyst precursor was reduced,
INTRODUCTION
3
� 1
without a precalcination step, by heating in a 150 cm min
�
1
The heterogeneous hydrogenation of aromatic com- stream of dry hydrogen (99.9% ) at a fixed rate of 5 K min
pounds is normally carried out with platinum, rhodium, to a finaltemperature of723 � 1K which wasmaintained for
ruthenium, palladium, or nickel based catalysts, in many 18 h. Temperature-programmed reduction studies revealed
0
cases at elevated temperatures and/or pressures (1). In view a complete reduction of the nickel content to Ni (28). The
of the stringent environmental regulations governing the nickel metal dispersion, reproducible to better than � 3% ,
aromatic level in diesel fuels, aromatic hydrogenation has expressed as (Nisurface/Nitotal) � 100% equals 73% . The sup-
become a key upgrading parameter in processing middle ported nickel particle size is estimated to be 1.4 nm from
distillates (2). Taking an overview of the various compila- the relationship (29) d = 101/D, where D is the dispersion
tions of the literature on this topic (1–8), it is readily ap- and d represents the surface weighted average crystallite
parent that benzene has been the overwhelming choice as diameter assuming (spherical) particles � 1 nm to be 100%
2
the model aromatic feedstock. The rate of aromatic ring dispersed. A value of 0.633 nm for the average surface
2
� 1
hydrogenation is however influenced by both steric and nickel atom (30) yields a surface area of 74 m g . Re-
electronic factors (1–4). In general, hydrogenation rates peated use of the catalyst did not result in any change in
decrease with ring substitution by alkyl groups unless the the dispersion or dimensions of the supported nickel crys-
substituents introduce exceptional strain in which case the tallites.
strained aromatic systems undergo facile saturation (3).
All the catalytic reactions were carried out under at-
The liquid (9–14) and gas (2, 15–22) phase hydrogenation mospheric pressure in a fixed-bed glass reactor over the
347
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

3
48
MARK A. KEANE
temperature range 393 K � T � 523 K. Details of the cata-
lytic reactor, procedure, and analysis are available else-
where (23). All the activity data presented in this paper
were obtained at steady state and are the average of at
least six separate test samples. The kinetic measurements
were made using catalyst meshed in the 125–150 �m range
3
� 1
at a space velocity of 2 � 10 h (STP) and in the W/F
�
1
range 42–112 g mol h, where W is the weight of activated
catalyst and F is the flow rate of aromatic. Mol% conver-
sion was kept below 20% by varying W in order to minimize
heat and transfer effects. The reactor has been shown previ-
ously (23) to operate in the absence of appreciable diffusion
limitations under the stated experimental conditions. The
reaction order with respect to the three aromatics was mea-
sured at a constant hydrogen pressure of 0.94 atm, where
the pressure of each aromatic was varied in the range 0.01–
0
.06 atm, using nitrogen as the makeup gas. Molar selectiv-
FIG. 1. The temperature dependence of the turnover frequency of
o-xylene (� ), m-xylene, (᭝) and p-xylene (᭺).
ity in terms of product x is defined as mx/mtot � 100, where
mtot is the total number of moles of product. The o-, m-,
and p-xylene (Aldrich, 98+% ) reactants were thoroughly
degassed by purging with purified helium and were stored
over activated molecular seive type 5 A.
(
22) noted that the rate of hydrogenation of o-xylene over
palladium was four to six times lower than that of m-xylene
and the latter hydrogenated at comparable or slightly lower
rates than p-xylene. The relative TOF for o-: m-: p-xylene
at representative temperatures are given in Table 1. While
the relative specific rates vary appreciably with tempera-
RESULTS AND DISCUSSION
The hydrogenation of o-, m-, and p-xylene over Ni/SiO2 ture, the differential observed between the three xylenes is
yielded 1,2-, 1,3-, and 1,4-dimethylcyclohexane (DMC), re- not as great as that reported for palladium (22). Benzene
spectively, as the only hydrogenated products. The satu- is known to be adsorbed on nickel via �-bond interactions
rated cylcloparaffin products were present as a mixture of and it is now generally accepted that the presence of methyl
the cis and trans forms. An isomerization of o-xylene to substituents on the benzene ring stabilizes the adsorbed
m-xylene, which has been reported for palladium systems �-complex with the resultant introduction of a higher en-
(
22), or a partial hydrogenation to dimethylcyclohexenes, ergy barrier for aromatic ring hydrogenation (32–34). From
as has been observed for liquid phase reactions (9, 11), was a consideration of the observed sequence of activities, it
not detected in this study. Passage of each aromatic in a may be concluded that the adsorbed o-xylene �-complex
stream of hydrogen over the silica support did not result is the most stable or least reactive of the three xylenes.
in any measurable degree of hydrogenation. At tempera- It has been proposed that the stability of �-complexes in-
tures in excess of 523 K each xylene feed was observed to creases with decreasing ionization potential (21). However,
undergo partial hydrogenolysis with the result that this in- the ionization potential (35) of p-xylene (8.44 eV) is lower
vestigation has been limited to temperatures � 523 K. The
effect of temperature on the turnover frequency (TOF) or
the number of xylene molecules converted per metal site
per second is illustrated in Fig. 1. The only available steady
state TOFs in the literature refer to palladium systems at
TABLE 1
The Temperature Dependence of the Relative
Turnover Frequencies of the Three Xylenes
4
13 K (22) and these are 1.3–59 times greater the values
Relative TOF
generated in this study. It is however well established that
the aromatic ring hydrogenation activity of palladium is
greater than that of nickel (3, 31). The addition of methyl
group(s) to the aromatic ring has been found to lower the
rate of ring hydrogenation and the ease of reduction has
been shown (3, 4, 13, 22) to decrease in the order ben-
zene > toluene > xylene. The TOFs of the three xylenes
recorded in this study are very close but do increase in
the order o-xylene < m-xylene < p-xylene, over the entire
temperature range that was studied. Rahaman and Vannice
T (K)
o-Xylene : m-Xylene : p-Xylene
3
4
4
4
93
08
28
43
1 : 1.19 : 1.34
1 : 1.14 : 1.26
1 : 1.05 : 1.12
1 : 1.02 : 1.05
1 : 1.08 : 1.12
1 : 1.22 : 1.33
1 : 1.23 : 1.33
1 : 1.24 : 1.36
1 : 1.30 : 1.49
458
473
4
5
5
88
03
18

�
HYDROGENATION OF o-, m-, AND p-XYLENE
349
than that of o- (8.56 eV) and m- (8.58 eV) xylene which are namic and diffusion effects, it is reasonable to also attribute
essentially the same. The differences in activity exhibited the occurrence of the common Tmax for the conversion the
by the three xylenes is not due, therefore, primarily to an three xylenes over Ni/SiO2 to adsorption phenomena.
electronic effect. An alternative explanation which invokes
The kinetics of the reaction may be represented by the
the contribution of steric constraints due to the neighboring power equation
methyl groups will be supported by subsequent kinetic data.
A feature of the three plots shown in Fig. 1 is the presence
of a common temperature related activity maximum (Tmax)
m
n
TOF = kP P ,
[1]
x
H2
at 453 K. The source of such a maximum may be arrived where k is the rate constant, Px and PH the partial pressure
2
at by considering the obvious causes. Each maximum could of the particular xylene reactant and hydrogen and m and
be traversed from either the high or low temperature side n the orders of the reaction with respect to the aromatic
without any loss in activity which, in effect, negates catalyst and hydrogen partial pressures, respectively. The reaction
deactivation asa possible cause. Diffusion effectscan be dis- orders with respect to xylene were determined by means of
counted as the catalytic reactor is known (23) to operate in logarithmic plots where at constant reaction temperature
the differential mode and the degree of hydrogenation was and PH2
far from equilibrium conversions. Nickel crystallite growth
during catalysis can, likewise, be excluded. The possible
contribution of the reverse, dehydrogenation, reaction to
the observed maxima in hydrogenation rates was investi- and plots of log TOF against log Px yield m. The orders
gated by passing each of the saturated cycloparaffins over with respect to xylene partial pressure derived from the
the activated catalyst. There was no measurable dehydro- experimental data are plotted as a function of tempera-
genation of the cycloparaffin feed at T � 473 K. The exis- ture in Fig. 2. Representative logarithmic plots depicting
tence ofa temperature dependent benzene (24) and toluene the variation of TOF with the partial pressure of m-xylene
�
�
n
H2
log TOF = log kP
+ m log Px .
[2]
(
25) hydrogenation maximum has been ascribed in the case at selected temperatures are shown in the inset to Fig. 2.
of Ni/Y zeolites to a decrease in the surface coverage by the The value of m increased from <0.1 to 0.44 as the tem-
aromatic with increasing temperature which at some point perature was raised from 393 to 523 K. The author could
results in a decrease in the reaction probability. In the ab- find no documented temperature dependence of reaction
sence of catalyst deactivation, secondary reactions, changes orders in xylene hydrogenation reactions on comparable
in active site distribution, product inhibition and thermody- nickel/amorphous carrier systems. The near zero order
FIG. 2. Dependence of the reaction order (m) with respect to the partial pressure of o-xylene ( ), m-xylene (᭡), and p-xylene (᭹). (Inset) TOF
of m-xylene as a function of partial pressure at 403 K (� ), 453 K (᭡), and 503 K (᭹).

3
50
MARK A. KEANE
FIG. 3. Variation with temperature of P mx , where x represents
o-xylene (� ), m-xylene (᭡), and p-xylene (᭹): Px = 0.04 atm.
FIG. 4. Apparent Arrhenius plots for the hydrogenation of o-xylene
(� ), m-xylene (᭡), and p-xylene (᭹).
dependence on all three aromatics observed in this study Arrhenius relationships are shown in Fig. 4. The computed
where m < 0.1 at T � 433 K is consistent with a high surface activation energies (with 95% confidence limits) and the
coverage by the aromatic. The increase in the reaction or- associated temperature independent preexponential factor
der with increasing temperature suggests that the surface (A), often referred to as the reaction frequency factor, are
coverage is lower at higher temperatures. The values of m given in Table 2 where
are very close for the three xylene reactants but the experi-
mental values at a particular temperature do exhibit a defi-
nite increase in magnitude (where T > 430 K), in the order
k = Aapp exp(� Eapp/RT ).
[4]
o-xylene < m-xylene < p-xylene. In a bimolecular surface The activation energy values for the three xylenes are very
reaction, the reaction rate is proportional to the surface similar and there is some overlap within the 95% confidence
coverage (�) of both reactants, i.e.,
band. Nevertheless, at T � 453 K, there is a discernible trend
of increasing values in going from the p- to m- to o- forms
[3] which agrees with the observed sequence of TOFs. Steady-
TOF = k�x �H2 .
�
1
state activation energies (31–82 kJ mol ) generated from
m
x
From Eqs. [1] and [3] it follows that �x ∝ P and the varia- a range of palladium catalysts (22) encompass the values
tion of the surface coverage by the xylene reactants may be reported in this paper, while the values quoted for o- and
�
1
approximated by the temperature dependence of the exper- p-xylene reduction (39–59 kJ mol ) over Pt, Pd, and Pt/Pd
m
x
imentally determined P values and these relationships are supported (on alumina) systems using a pulse reactor (18)
plotted in Fig. 3. The influence of increasing temperature on are appreciably lower. The temperature dependence of the
m
Px and by inference on the fractional surface coverage is rate constants at T � Tmax also obey the Arrhenius relation-
considerable. Under the stated experimentalconditions, the ship and yield the negative activation energies which are
reaction order with respect to the hydrogen concentration
(
n) increased with temperature from 0.7 to 2.3. However,
TABLE 2
n
the values of P have been shown to be largely unaffected
H2
Kinetic Data Obtained from the Arrhenius Relationships
Shown in Fig. 4
by temperature and to be independent of the nature of the
aromatic (23), an observation which is taken to be indica-
tive of a noncompetitive adsorption of the reactants during
catalysis. The temperature dependence of TOF shown in
Fig. 1 can then be considered to essentially mirror the com-
bined effect of an increase in the rate of hydrogenation on
the catalyst surface and the accompanying decrease in the
concentration of surface reactive aromatic species which
ultimately results in the generation of a Tmax.
� 1
Eappa (kJ mol� 1
)
Xylene feed
T range (K)
Aapp (s
)
6
o-Xylene
393–453
4 � 10
73.4 � 3.8
�
9
4
53–523
393-453
53–523
393–453
53–523
2 � 10
� 60.5 � 3.0
6
m-Xylene
p-Xylene
1 � 10
68.1 � 3.4
�
9
8
4
6 � 10
� 56.1 � 3.3
5
4 � 10
3 � 10
64.9 � 2.6
�
4
� 49.6 � 3.2
�
1
The rate constants (k, units of s ) for the three reac-
a
tions were calculated using Eq. [1] and the experimental
With 95% confidence limits.

HYDROGENATION OF o-, m-, AND p-XYLENE
351
included in Table 2. In heterogeneously catalyzed systems,
TABLE 3
the activation energies obtained from the relationship be-
tween the experimentally obtained rate constants and tem-
perature are generally accepted to be apparent activation
energies as the surface coverage by the reactive species does Xylene feed
not remain constant throughout the temperature range
Values of Eapp, 1Hads, and Etrue with 95% Confidence Limits
Obtained from the Arrhenius Relationships shown in Fig. 4
Eapp (kJ mol 1)
�
1Hads (kJ mol
� 1
)
� 1
Etrue (kJ mol )
o-Xylene
studied. Consequently, apparent values of A(Aapp) and
m-Xylene
73.4 � 3.8
68.1 � 3.4
64.9 � 2.6
� 133.9 � 4.8
� 124.2 � 4.7
� 114.5 � 4.1
207.3 � 6.1
192.3 � 5.7
179.4 � 4.7
E(Eapp) are obtained by application of the Arrhenius equa-
p-Xylene
tion to the overall rate data and these parameters do not
directly refer to the rate-limiting surface bond redistribu-
tion step. The true activation energy (Etrue) is obtained by
adding the enthalpy of adsorption as a positive quantity to
the apparent activation energy (36, 37)
–
and C CH3 bonds bend away from the surface. In the case
of xylene adsorption both methyl groups must be bent away
from the surface in order to relieve the steric repulsion. The
repulsive potential is greatest in the case of neighbouring
[5] or ortho-methyl groups with the result that o-xylene has
to overcome the largest barrier and the adsorption energy
Etrue = Eapp � 1Hads.
The process of chemisorption also has an activation energy must be greatest as is the case with the inferred values given
(
Eads)
in Table 3. The aromatic ring distortion which allows the
methyl substituents to point away from the plane of the ring
must then increase in going from p- to m- to o-xylene. An
kads = � Z exp(� Eads/RT ),
[6]
where kads is the rate coefficient for chemisorption, � is increase in the energy of interaction with the surface is ac-
the sticking probability or fraction of collisions that leads companied by a decrease in reactivity with the consequent
to adsorption, and Z is the number of collisions per unit sequence of TOFs shown in Fig. 1 and Table 1.
area per unit time. This equation, based on collision the-
It is often observed (12, 42, 43), in the case of either the
ory, is analogous to Eq. [4]. Desorption is an activated pro- same reaction conducted over a series of different catalysts
cess since the minimum Edes is that equal to 1Hads, where or a series of related reactions conducted over the same
(
12, 38, 39)
catalyst, that a relationship exists between Eapp and Aapp
which takes the form
Edes � Eads = 1Hads.
[7]
ln Aapp = eEapp + B.
[8]
Each of the xylene hydrogenations reported in this study
where conducted in the absence of (a) diffusion/mass trans-
port effects, (b) rate inhibition by the products, and (c) sec-
ondary reactions and, as such, may be considered to be sur-
face controlled where the overall reaction rate is governed
by the surface concentration of the reactive species. The
activation energies obtained over the temperature ranges
corresponding to increasing and decreasing rate constants
are, therefore, essentially equivalent to Eads and Edes, re-
spectively. Taking each set of experimentally evaluated ac-
tivation energies, values of 1Hads and Etrue for the three
cases were obtained from a combination of Eqs. [5] and [7]
and are quoted with 95% confidence limits in Table 3. The
order of increasing Etrue is the same as that of Eapp and there
is a definite increase in 1Hads values in going from p- to m-
to o- substitution. The author could find no spectroscopic
study of o-xylene adsorption on metals. Nevertheless, the
strength of xylene adsorption has been considered to be
greater than that of benzene as a result of the stronger do-
nation from the �-electron cloud to the metal due to the
presence of the methyl groups (40, 41); i.e., the effect is
electronic in nature. Differences in 1Hads within the fam-
The above relationship is termed the compensation effect
CE), where e and B are compensation factors. The effect
(
is aptly named as an increase in ln Aapp is offset or com-
pensated by an increase in E which results in a decrease in
reaction rate. The preexponential factors (Aapp) obtained in
this study are plotted against the corresponding apparent
activation energies in Fig. 5. It can be seen that the linear
relationship (correlation coefficient >0.999) holds true for
the three sets of positive and negative activation energies
where e = 0.264 � 0.007 and B = � 4.199 � 0.046. Previously
reported data for the hydrogenation of benzene (24) and
toluene (25) over Ni/K–Y and benzene and toluene over
Ni/SiO2 (23) are also included in Fig. 5 and are likewise
consistent with Eq. [8]. A true compensation effect is es-
tablished if a single point of concurrence appears in the Ar-
rhenius plots (12, 42, 45). The temperature corresponding
to such a point, the isokinetic temperature (Tiso) is obtained
from
Tiso
=
1 ,
eR
[9]
ily of xylenes can however result from steric effects. Minot where e is the slope of the compensation plot and R is
and Gallezot, in a theoretical investigation (29), found that the gas constant; the calculated value of Tiso is 456 � 8 K.
toluene adsorbs parallel to the active surface and the C–H The experimentally determined rate constants at 453 K

3
52
MARK A. KEANE
FIG. 6. The temperature dependence of ln �Z for the hydrogenation
FIG. 5. Compensation plot for the hydrogenation of o-xylene (� , � ), of o-xylene (� ), m-xylene (᭡), and p-xylene (᭹).
m-xylene (᭡, ᭝) and p-xylene (᭹, ᭺) over Ni/SiO2 where the equation of
the fitted line is ln Aapp = (0.264 � 0.007)Eapp � 4.199 � 0.046. Note: ln Aapp
vs. Eapp data for the hydrogenation of benzene (᭜, ᭛) and toluene (᭢, ᭞)
three plots generated using calculated values of Etrue are
over Ni/SiO2 (23) and the hydrogenation of benzene (ଙ, ) (24) and shown in Fig. 7. Such plots illustrate the dependence on
toluene (
,
) (25) over Ni/K–Y.
temperature of the rate constant for the true surface re-
action and unlike the apparent Arrhenius relationships do
not possess a single point of concurrence. It may be con-
cluded that the compensation effect is a direct result of ap-
parent kinetic measurements which yield composite kinetic
parameters.
The hydrogenation of each xylene yielded stereoisomeric
mixtures where the ratio of isomers in the product was con-
stant at a particular temperature over the range of aromatic
concentrations that were studied. The variation, with tem-
perature, of the rate constant for hydrogenation to both cis
and trans products is illustrated in Fig. 8 and the selectivity
in terms of trans formation at representative temperatures
yield the ratios ko-xylene : km-xylene : kp-xylene = 1 : 1.01 : 1.02.
The agreement between the calculated and experimen-
tal Tiso values is good and confirms the attribution of a
compensation effect to the hydrogenation of the three
xylenes over Ni/SiO2. The actual source of the compen-
sation effect is still a matter of some debate and the
validity of the effect as a kinetic observation and not
merely an experimental artifact remains open to question.
The kinetic investigation which yielded this compensa-
tion effect was conducted under experimental conditions
where the enthalpy of the adsorption term is important;
i.e., the CE is only obeyed by the experimentally deter-
mined or apparent A or E values. The true preexponen-
tial factor, corresponding to the true activation energy,
must take account of both the probability of collision
between the two reactants and the change in the surface
concentration with temperature. A modified Arrhenius
equation, based on collision theory
k = � Z exp(� Etrue/RT )
[10]
describes the true temperature dependence of the rate con-
stant, where Etrue refers to the rate determining step and the
term �Z makes due allowance for variations with temper-
ature in the probability of surface collisions. Values of �Z
were calculated for each temperature using Eq. [10] and the
relationship between �Z and T is illustrated in Fig. 6. The
decrease in collision probability with temperature, depicted
in the three plots, isdiagnosticofa decreasingconcentration
of xylene on the catalyst surface. A plot of ln (k/�Z) versus
FIG. 7. True Arrhenius plots for the hydrogenation of o-xylene (� ),
1
/T gives a slope of � Etrue/R according to Eq. [10] and the m-xylene (᭡), and p-xylene (᭹).

HYDROGENATION OF o-, m-, AND p-XYLENE
353
hydrogenation of dialkylbenzenes has largely been the cis-
dialkylcyclohexane (6, 17, 18, 46, 47). An increase in the pro-
portion of trans product from 1,2- and 1,4-dialkylbenzene
with increasing temperature has been recorded (6, 12, 14).
Hydrogen pressure and the size of the alkyl substituent
have also been identified as factors which influence product
distribution (9). In addition, the nature of the catalyst has
been recognized as having a bearing on the stereoselective
process (46). In this study, the hydrogenation of o-xylene
preferably yielded cis- 1,2-DMC at T < 428 K, while the
proportion of the trans product increased at higher temper-
atures to a maximum selectivity of ca. 63% which was essen-
tially constant at 463 K � T � 523 K. In contrast, m-xylene
was preferentially hydrogenated to trans-1,3-DMC at a per-
centage selectivity (ca. 63% ) which was constant over the
entire remperature range that was studied. The reduction of
p-xylene produced a higher concentration of cis-1,4-DMC
but the fraction of trans-1,4-DMC increased slightly with
increasing temperature to give virtually equimolar mix-
tures at T > 433 K. Each pair of cis/trans 1,2-, 1,3-, and
1
,4-DMC diastereomers exist in two interconvertible chair
conformations. In the case of the 1,2- and 1,4- derivatives,
the trans conformation is the more stable, whereas with
the 1,3- derivative the reverse is true (48). The product
distribution generated over Ni/SiO2 clearly departs from
FIG. 8. The effect of temperature on the rate constant for the forma-
tion of the trans (open symbols) and cis (closed symbols) saturated product the stereochemical equilibria. The observed selectivities, in
from (a) o-xylene (� , � ), (b) m-xylene (᭡, ᭝), and (c) p-xylene (᭹, ᭺).
turn, differ to a greater degree from the reported liquid
phase data but direct comparisons are not very meaningful.
The source of steroisomeric differentiation in a continu-
ous flow gas phase system may well differ from that which
controls stereoselectivity in a batch liquid phase process.
While the interaction between substrate and catalyst has
been proposed (9, 14) as the major factor that determines
stereospecificity in the liquid phase, desorption of the di-
ene has been shown to occur (9) and this intermediate can
undergo a second hydrogenation step which may involve
a different substrate/catalyst interaction. The liquid phase
reaction is conducted in an organic medium that may also
influence the stereospecific step. In the gas phase hydro-
genation of the xylenes the stereochemistry of reduction
must be determined at the adsorption step and the nature
of the steric interaction between the reactant(s) and the
catalyst surface determines the ultimate isomeric form of
the saturated product. The gas phase hydrogenation of ben-
zene has been viewed in the literature as occurring via the
sequential addition of hydrogen atoms to the adsorbed aro-
matic where no rate determining step is apparent (49, 50)
or where the addition of the first (51, 52), the second (22) or
the sixth (51) hydrogen atom is the slow step. As the same
compensation plot can be applied to benzene and methyl-
benzene hydrogenation (Fig. 5), the transformation of each
aromatic may be tacitly assumed to proceed via the same
rate demanding step. The destabilization of the resonance
energy that is initiated by the addition of the first hydrogen
is recorded in Table 4. Practical interest in such selective
hydrogenations is limited but the reactions may be consid-
ered as the simplest models to probe the effects of reaction
variables on the stereochemistry of substituted aromatic re-
duction. The majority of the available literature (6, 9, 12,
1
4) in which the relative cis/trans product ratios are docu-
mented have dealt with reactions conducted in the liquid
phase and the available selectivity data for gas phase hy-
drogenation (17–19) are scant. The principal product of the
TABLE 4
The Effect of Temperature on the Selectivity to trans Di-
methylcyclohexane formation for the Three Xylenes
Strans (% )
T (K)
o-Xylene
m-Xylene
p-Xylene
3
4
4
4
4
4
4
5
5
93
08
28
43
58
73
88
03
18
40.3
43.7
49.2
52.4
59.5
62.9
63.5
63.2
63.6
63.3
63.3
63.4
63.2
63.4
63.4
63.4
63.3
63.3
41.0
42.9
45.1
45.8
45.9
45.9
46.0
45.9
46.1

3
54
MARK A. KEANE
should be very energetically demanding and is proposed to uct mixtures were � 1.2 greater than in the actual sat-
be the slow step. The observed increase in activation energy urated feedstock which is considerably lower than the
with increasing resonance energy of the aromatic nucleus is increase by a factor of 2.6 induced by temperature vari-
consistent with such a mechanism. The turnover frequency ations in the direct hydrogenation of o-xylene. Cat-
is influenced by the concentration of the surface reactive alytic epimerization may be discounted as a principal
aromatic (23) and once equilibrium adsorption has been contributory factor in determining stereodifferentiation
attained the rate of the overall surface reaction is governed in this particular study. The hydrogenation of m-xylene
by the rate of the slow step; the nature of this step is not generated trans-1,3-DMC, the thermodynamically less sta-
controlled by adsorption phenomena. The true Arrhenius ble stereoisomer, as the favored product which in com-
plots shown in Fig. 7 illustrate the temperature dependence mon with cis-1,2-DMC, the initially preferred product from
of the surface reaction given in Eq. [3]. The direction of ad- o-xylene, has one methyl group equatorially and one axi-
dition ofhydrogen to the two carbonswhich bear the methyl ally oriented with respect to the plane of the ring. How-
group is controlled by the geometry of adsorbed moiety. It is ever, the ratio of cis/trans-1,3-DMC is independent of
assumed that the xylenes adsorb with the aromatic ring par- temperature, as shown in Table 4, which suggests that a
allel to the surface and both methyl groups directed away decrease in surface constraints did not result in a rearrange-
from the surface in order to relieve the steric repulsion. At ment of the adsorbed species to the more stable conforma-
�
1
the lowest reaction temperature, o-xylene is preferably hy- tion. The cis isomer of 1,3-DMC is 7.5 kJ mol more stable
�
1
drogenated to cis-1,2-DMC which isthe thermodynamically than the trans form compared with the 11.3 kJ mol dif-
less stable of the two possible stereoisomers. Under these ferential between trans and cis 1,2-DMC (47). It is there-
reaction conditions o-xylene is most strongly held at the sur- fore proposed that in the stepwise addition of hydrogen to
face and the crowding effects due to neighboring adsorbed m-xylene the possible reduction of van der Waals strain
molecules constrain the motion of each adsorbed species. In is not great enough to induce geometrical rearrangement
such a constrained system one of the hydrogen atoms that is of the adsorbed species as the reaction temperature is in-
added to a carbon which is bonded to a methyl group lies in creased. In the case of p-xylene, steric hindrance in the
the plane of the ring, while the other lies axially to the ring mode of adsorption is not as severe with the result that
and one of the methyl groups is now oriented equatorially, the product mixture approaches equimolarity. Neverthe-
the other axially. The adsorbed molecule therefore adopts a less, the cis product is preferred where, once again, one
conformation which differs from the conformation of low- hydrogen is added axially and one equatorially to the car-
est energy in the isolated molecule. From adsorption studies bonsbearingthe methylgroups. An increase in temperature
of benzene and toluene and their saturated analogues (39, favours, albeit marginally, the more stable trans isomer but
�
1
5
3), it can be assumed that 1,2-DMC is not as strongly held as with 1,3-DMC the difference in energy (7.5 kJ mol )
on the surface as o-xylene and desorbs rapidly in the hydro- between the two isomeric forms does not appear to be suf-
gen gas stream in the stereoisomeric form that is generated ficient to result in an appreciable temperature effect.
as a result of the surface interactions. The more stable trans
isomer is also produced at a comparable rate and is iso-
lated as the principal product at T > 433 K. An increase
in reaction temperature is accompanied by a decrease in
CONCLUSIONS
The TOFs of the three xylenes increase, at steady state
the surface concentration of the active aromatic (as may in the absence of diffusion constraints, catalyst deactiva-
be inferred from Fig. 6) and the geometrical constraints are tion and secondary conversions, in the order o-xylene <
somewhat relieved. It is proposed that the weakening of the m-xylene < p-xylene in the temperature range 393 K �
surface interaction and decreased crowding with increas- T � 523 K. The specific rate of hydrogenation of each xylene
ing temperature allows the rearrangement of atoms not di- passes through a maximum at a common temperature Tmax
rectly involved in the binding process with a consequent in- (453 K) and this is attributed to a critical loss of the reactive
crease in the proportion of trans product where both methyl aromatic species from the surface. The reaction order with
groups are equatorial to the plane of the ring and van der respect to each xylene isomer increases with temperature
Waals strain is reduced. Such a switch in stereoselectivity from 0 to 0.44 where the value of the partial pressures raised
could also be due to a metal catalyzed epimerization or to the particular reaction order is considered to reflect the
conformational inversion of one asymmetric carbon atom changes in the fractional surface coverage. The variation
(
14, 54, 55). Known mixtures of trans/cis 1,2-DMC (mo- of rate constant with temperature yields both positive and
lar ratios in the range 0.17–0.58) were passed, in a stream negative activation energies. Taking the reaction energies
of hydrogen, using the same experimental conditions to be equivalent to the chemisorption energies, values of
as employed in hydrogenation step through an empty re- Etrue and 1Hads can be evaluated where both parameters
actor, over the silica support and over activated Ni/SiO2 increase in the order p-xylene < m-xylene < o-xylene. As-
at 393 K � T � 523 K. The trans/cis ratios in the prod- suming the xylenes adsorb with the aromatic ring parallel

HYDROGENATION OF o-, m-, AND p-XYLENE
355
to and both methyl groups directed away from the surface, 20. V o¨ lter, J., J. Catal. 3, 297 (1964).
2
2
2
1. V o¨ lter, J., Hermann, M., and Heise, K., J. Catal. 12, 307 (1968).
2. Rahaman, M. V., and Vannice, M. A., J. Catal. 127, 251, 267 (1991).
3. Keane, M. A., and Patterson, P. M., J. Chem. Soc. Faraday Trans. 92,
413 (1996).
4. Coughlan, B., and Keane, M. A., Zeolites 11, 12 (1991).
the repulsive potential is greatest in the case of neighboring
or o-methyl groups with the result that the consequent ad-
sorption energy is greatest. Hydrogenation yields stereoiso-
meric product mixtures where, due to steric constraints, one
1
2
hydrogen is added axially and one equatorially to the two 25. Coughlan, B., and Keane, M. A., Catal. Lett. 5, 101 (1990).
2
2
2
2
6. Keane, M. A., Ind. J. Technol. 30, 51 (1992).
7. Keane, M. A., and Webb, G., J. Catal. 136, 1 (1992).
8. Keane, M. A., Can. J. Chem. 72, 372 (1994).
carbons bearing a methyl group. The above orientation is
favored over the entire temperature interval for the hydro-
genation of m- and p-xylene, whereas at T > 433 K, in the
case of o-xylene the atoms not directly involved in the bind-
9. Smith, J. S., Thrower, P. A., and Vannice, M. A., J. Catal. 68, 270
(
1981).
ing process are rearranged to relieve the van der Waal’s 30. Coenen, J. W. E., Appl. Catal. 75, 193 (1991).
3
3
3
3
3
1. Greenfield, H., Ann. N.Y. Acad. Sci. 214, 233 (1973).
2. Minot, C., and Gallezot, P., J. Catal. 123, 341 (1990).
3. Candy, J., and Foilloux, P., J. Catal. 38, 110 (1975).
4. Renouprez, A. J., and Clagnet, G., J. Catal. 74, 296 (1982).
5. Dean, J. A., “Handbook of Organic Chemistry.” McGraw–Hill, New
York, 1987.
strain and preferably generate the more stable trans iso-
mer. A compensation effect has been observed for the re-
lationship between ln Aapp and Eapp with a Tiso = 456 � 8 K
and it is concluded that Aapp and Eapp are composite terms
which incorporate contributions due to the temperature
dependence of the surface concentration of the reactive
species.
3
3
6. Martin, G. A., J. Catal. 60, 345 (1979).
7. Frennet, A., Lienard, G., Crucq, A., and Degols, L., J. Catal. 53, 150
(
1978).
3
3
8. Ashmore, P. G., “Catalysis and Inhibition of Chemical Reactions.”
Butterworths, London, 1963.
9. Adamson, A. W., “Physical Chemistry of Surfaces,” 5th ed. Wiley, New
York, 1990.
REFERENCES
1
. Kieboom, A. P. G., and van Rantwijk, F., “Hydrogenation and Hy-
drogenolysis in Synthetic Organic Chemistry.” Delft Univ. Press, Delft, 40. Abon, M., Bertolini, J. C., Bily, J., Massardier, J., and Tardy, B., Surf.
1
977.
Sci. 162, 395 (1985).
2
3
. Stanislaus, A., and Cooper, B. H., Catal. Rev. Sci. Eng. 36, 75 (1994).
. Rylander, P. N., “Hydrogenation Methods.” Academic Press, London,
41. Tsai, M. C., and Muetterties, E. L., J. Am. Chem. Soc. 104, 2543
(1992).
1
985.
42. Galwey, A. K., Adv. Catal. 26, 247 (1977).
43. Galwey, A. K., and Brown, M. E., J. Catal. 60, 335 (1979).
�
ˇ
4
5
. Cerven y´ , L., and Ru zˇ i cˇ ka, V., Catal Rev. Sci. Eng. 24, 503 (1982).
. Augustine, R. L., “Catalytic Hydrogenation.” Dekker, New York, 44. Agrawal, R. K., J. Thermal Anal. 31, 73 (1986); 35, 73 (1989).
975. 45. Exner, O., Collect. Czech. Commun. 38, 1425 (1973).
. Rylander, P. N., “CatalyticHydrogenation in OrganicSyntheses.” Aca- 46. Bartok, M., Czombos, J., Felfoldi, K., Gera, L., Gondos, Gy.,
1
6
7
8
9
demic Press, London, 1979.
. Freifelder, M., “Catalytic Hydrogenation in Organic Synthesis, Pro-
cedures and Commentary.” Wiley, New York, 1978.
Molmar, A., Northeisz, F., Palinko, I., Wittmann, Gy., and Zsigmond,
A. G., “Stereochemistry of Heterogeneous Metal Catalysis.” Wiley,
New York, 1985.
ˇ
. “Catalytic Hydrogenation” (L. Cerven y´ , Ed.), Stud. Surf. Sci. Catal., 47. Siegal, S., Smith, G. V., Dmuchovsky, B., Dubell, D., and Halpern, W.,
Vol. 27. Elsevier, Amsterdam, 1986.
J. Am. Chem. Soc. 84, 3136 (1962).
. Siegel, S., Adv. Catal. 16, 123 (1966).
48. Kalsi, P. S., “Stereochemistry, Conformation and Mechanism.” Wiley
Eastern Ltd., New Delhi, 1990.
1
1
1
1
1
1
0. Rader, Ch. P., and Smith, H. A., J. Am. Chem. Soc. 84, 1443 (1962).
1. Hartog, F., and Zwietering, P., J. Catal. 2, 79 (1963).
49. Snagovskii, Yu., Lyubarskii, G. D., and Ostrovskii, G. M., Kinet. Katal.
7, 232 (1966).
50. van Meerten, R. Z. C., and Coenen, J. W. E., J. Catal. 46, 13 (1977).
51. Prasad, K. H. V., Prasad, K. B. S., Muallikarjunan, M. M., and
Vaidyeswaren, R., J. Catal. 84, 65 (1983).
52. Zrn cˇ evi c´ , S., and Ru sˇ i c´ , D., Chem. Sci. Eng. 43, 763 (1988).
53. Coughlan, B., and Keane, M. A., J. Chem. Soc. Faraday Trans. 86, 3961
(1990).
54. Barbier, J., Morales, A., and Maurel, R., Nouv. J. Chim. 4, 223
(1980).
2. Bond, G. C., “Catalysis by Metals.” Academic Press, London, 1962.
3. Wauquier, B., and Jungers, J. C., Bull. Soc. Chim. France 1280 (1957).
4. Burwell, R. L., Jr., Chem. Rev. 57, 895 (1957).
5. Leitz, G., and V o¨ lter, J., in “Mechanism of Hydrocarbon Reactions.”
p. 151. Elsevier, Amsterdam, 1975.
1
1
6. Dufresne, P., Bigeard, P. H., and Billon, A., Catal. Today 1, 367 (1987).
7. Armendia, N. A., Borau, V., Jimenez, C., and Marinas, J. M., Bull. Soc.
Chim. Belg. 91, 7743 (1982).
1
1
8. Gomez, R., Del Angelo, G., and Corro, G., Nouv. Chem. 4, 219 (1980).
9. Vinnegra, M., Cordoba, G., and Gomez, R., J. Mol. Catal. 58, 107
55. Quinn, H. A., Graham, J. H., McKervey, M. A., and Rooney, J. J.,
J. Catal. 22, 136 (1976).
(
1990).