Professional Documents
Culture Documents
Abstract
In the present paper the relation existing between chemical kinetics, reaction mechanisms and reactor modelling is brie¯y
examined by reporting examples of both homogeneous and heterogeneous catalytic reactions. The scope is to give an
introductory insight into the more general topic of ``Kinetics and Modelling of Catalytic reactions from the laboratory to
industrial reactor''. # 1999 Elsevier Science B.V. All rights reserved.
0920-5861/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 9 9 ) 0 0 0 6 9 - 3
114 E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123
products obtained and the reagents disappeared, at portion of the sectioned tube. It results therefore
different times, in a reactor normally operating
dci
in ideal conditions. Such conditions could be related ri F
PFR reactor: (4)
to: dW
1. the thermal behaviour of the reactor (isothermal, Data about reaction rate can be collected indiffer-
adiabatic, etc.), ently by using one of the above mentioned reactors
2. the operating conditions (batch, semibatch, con- and interpreted to determine a kinetic law equation of
tinuous), and the form:
3. the regime of the flow for continuous reactors ri f
composition; temperature; (5)
(CSTRcontinuous stirred tank reactor,
PFRplug flow reactor). characteristic for a given reaction in the presence of a
well-defined catalyst. An equation of this type is
The measurement of the rate of a reaction corresponds
fundamental for reactor modelling and simulation
to the evaluation of the performance of the reactor
and is very important for the interpretation of the
where the reaction occurs and requires the solution of
catalytic mechanisms. The determination of the cor-
the related material balance equations. When consid-
rect form of this equation is the main goal of the
ering, for example, an isothermal reactor, the material
kinetic study of a reaction.
balance equation referred to the ith component will be:
When the reactor is not isothermal, together with
rate of i flow in rate of i flow out the material balance an energy balance must be con-
sidered, the form of which is similar to that of the
i produced or consumed i accumulation:
material balance.
The balance can be expressed in moles or mass The reaction rates can be related, according to the
according to the convenience, considering the stoi- convenience, to the amount of catalyst or to the
chiometry of the reaction. When there is no change in volume of the reactor. The composition of the reacting
the number of the moles of products with respect to the mixtures can usually be expressed in terms of con-
reactants and, hence, in the volumetric flow rates, it is centrations or partial pressures. Often, adimensional
possible to write, for example terms such as the conversion of the reactant or the
yields of the products are conveniently introduced in
dci the kinetic law expression. Dimensional units of the
Fcin out
i Fci Wri ; (1)
dt kinetic parameters will be adjusted, accordingly.
the rate of formation or disappearing of ``i'' expressed The experimental evaluation of reaction rates is
as (moles/time mass of catalyst) is, therefore usually performed in laboratory reactors in chemical
regime conditions, that is, in conditions not affected by
F ÿ in 1 dci
diffusional limitations; otherwise also the in¯uence of
ri ci ÿ cout
i ÿ : (2)
W W dt diffusion must be taken into account.
In the case of a continuous well stirred reactor,
operating in steady state conditions, the term of accu-
mulation is null. On the contrary, for a batch reactor, 3. Relation between kinetic law equation
the first term is null and consequently and reaction mechanism
1 dci
ri ÿ
batch reactor; 3.1. Scheme of the reaction, single reaction,
W dt
elementary reaction step, stoichiometry,
F in
ri
c ÿ cout
i
CSTR reactor: (3) reaction mechanism
W i
For a tubular plug flow reactor (PFR), the reaction Two reactants can frequently give rise to more than
rate changes along the catalytic bed; therefore, the one reaction. A kinetic study requires to recognise,
material balance must be applied to an infinitesimal ®rst of all, how many reactions occur in a reacting
E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123 115
system, that is, to identify the reaction pattern and the 3.2. Determination of the kinetic law equation
stoichiometry of each reaction of the scheme. through the assumption of two useful
A complex reaction scheme can be seen as a hypotheses
combination of simpler schemes in which equilibrium
and competitive or consecutive reactions appear, Kinetic data collected, for example, in an isother-
while the simpler scheme is made of combinations mal reactor for a generic reaction of the type
of single reactions. Any single reaction occurs with a AB!CD (8)
proper mechanism to which a kinetic equation corre-
can be interpreted by initially considering a power law
sponds. The reaction mechanism is a sequence of
kinetic equation of the type
different steps called ``elementary reactions'' often
involving unstable and undetectable molecular spe- r kpA pB p
C pD ; (9)
cies. The kinetic law equation can easily be written for
where , ,
, , the apparent reaction orders, will be
each elementary step, because it can directly be
small whole, fractional, positive, negative or null
derived from the stoichiometry of the reaction. As a
numbers. The apparent reaction orders and the kinetic
matter of fact, the ``order of the reaction'', with
constant can be determined by the mathematical
respect to each reactant, which is the power to be
regression analysis of the experimental kinetic data.
given to the reactant concentration in the kinetic law
The values found for , ,
, and give useful
equation, coincides with the stoichiometry coef®cient,
suggestions on the form that should be assigned to
as in the following examples:
the true kinetic law equation. For this purpose, a
A ! B C r k A Unimolecular process (6) reasonable mechanism must be formulated in which
the products are formed as a consequence of a se-
2A ! B C r k A2 Bimolecular process quence of elementary reaction steps. The de®nition of
the mechanism can normally be made by the following
A B ! C D r k AB Bimolecular process suggestions derived from literature for both the spe-
ci®c reaction studied and similar reaction. Then, two
A 2B ! C D r k AB2 Trimolecular process
useful hypotheses can be introduced to evaluate the
Elementary reactions have normally a very simple kinetic law equation on a theoretical basis:
stoichiometry. A trimolecular process, for example, 1. the hypothesis of the existence of a rate-control-
has a very low probability to occur, as it requires the ling step;
simultaneous collision of three molecules. 2. the assumption of the stationary state approxima-
On the other hand, our objective is to ®nd a kinetic tion for the unstable intermediate appearing in the
law equation being a function of measurable proper- mechanism.
ties, able to reproduce the evolution with time of
reactants and products appearing in the experimen- When one step, in a sequence of different elementary
tally observed stoichiometry. In the reaction of ammo- reactions, is very slow, the rate of the entire process
nia synthesis, for example can be considered equal to the slow step and this one
becomes the rate determing step. For example, in the
N2 3H2 , 2NH3 (7) case of the reaction
the stoichiometry of the reaction suggests only that C2 H4 12O2 ! CH3 CHO (10)
hydrogen will disappear three times faster than performed through the Wacker process, that is,
nitrogen and that ammonia will be formed with a through the following cycle of reactions
rate twice that of disappearing nitrogen. That is,
r ÿrN2 ÿrH2 =3 rNH3 =2. Pd C2 H4 H2 O ! Pd0 CH3 CHO 2H
The problem is to give a mathematical form to the (11)
rate law and this information cannot be derived from 0
Pd 2Cu
! Pd
2Cu
(12)
the stoichiometric equation. It can be derived only
from the mechanism of the reaction. 4Cu O2 4H ! 4Cu 2H2 O (13)
116 E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123
the following mechanism has been proposed However, the enthalpy change has often an opposite
fast sign with respect to the activation energy, that is
PdCl ÿ
4 C2 H4 $PdCl3 C2 H4 Cl
ÿ
(14) always a positive value. Therefore, in this case, also
the activation energy will be apparent.
fast
PdCl3 C2 Hÿ
4 H2 O $PdCl2
H2 OC2 H4 Cl
ÿ
When the rate of a reaction is conditioned by two or
more elementary steps, the objective of de®ning the
(15)
rate law can be achieved conveniently by introducing
PdCl2
H2 OC2 H4 H2 O the hypothesis of the steady state condition, according
fast to which, excluding the initial conditions, the rate of
$PdCl2
OHC2 Hÿ
4 H3 O
(16)
formation of a de®nite specie is equal to the rate of its
PdCl2
OHC2 Hÿ ÿ disappearing. For example, the mechanism of the
4 ! HOCH2 CH2 PdCl Cl
reaction:
(17)
HOCH2 CH2 PdCl ! CH3 CHO Pd HCl (18) CH3 OH 12O2 ! HCHO H2 O (23)
By assuming reaction (17) as the rate determining catalysed by solid iron molybdate can be written as:
step, we can write k1
CH3 OH S-ox ! HCHO H2 O S ÿ Red (24)
r k PdCl2
OHC2 Hÿ
4 (19)
k2
S-Red 12O2 ! S-Ox (25)
but the concentration of PdCl2
OHC2 Hÿ 4 is not
easily measurable and we need to find a correlation
S-Ox and S-Red being the oxidised and reduced
among this concentration and those of other more
catalytic sites, respectively.
easily measurable species. For this purpose, we can
The reaction rate is that of the formation of the
suppose that the steps preceding the rate determining
product, therefore
one have approximately reached the equilibrium and
we can write consequently r k1 pCH3 OH ox : (26)
Ke1 PdCl3 C2 Hÿ ÿ ÿ
4 Cl =PdCl4 C2 H4 (20) We need again to ®nd an expression giving the
ÿ ÿ
Ke2 PdCl2
H2 OC2 H4 Cl =PdCl3 C2 H4 H2 O concentration of the undetectable species ox (fraction
of oxidised sites) and to do this we assume this species
Ke3 PdCl2
OHC2 Hÿ
4 H3 O =PdCl2
H2 OC2 H4 reacting with the same rate of formation; therefore
H2 O
1=2
dred =dt dox =dt r25 ÿ r24 k2 pO2 red
by rearranging PdCl2
OHC2 Hÿ
4 in relation (19) we
obtain ÿ k1 PCH3 OH ox 0: (27)
3.3. Kinetics and mechanism for homogeneous 3. the property of showing a coordination number
reactions always greater than the oxidation one, favouring
the contact between potentially reactive molecules
In the previous section, we have seen the strict bonded to the same metal atom.
correlation existing between kinetics and reaction
Considering that a transition metal tends towards an
mechanisms. The two aspects should be considered
external electronic configuration with 18 electrons
always together. In the last decades, the possible
(saturated configuration), the mechanism of the reac-
reaction mechanisms have been classi®ed in a rela-
tions catalysed by transition metal complexes can be
tively small number of categories. An example of such
considered as a sequence of many possible elementary
classi®cation is reported in Table 1. This classi®cation
steps such as those reported in Table 2. In said reac-
is based on the simple observation that a reaction
tions either the oxidation or the coordination number
occurs with the cleavage of a chemical bond and the
can change and, consequently, the spatial geometry of
formation of another one.
the metal complex can change too. Many industrial
This can be achieved only by two ways: the homo-
processes of great practical importance occur in the
lytic and the heterolytic mechanisms, the ®rst invol-
homogeneous phase and are catalysed by transition
ving the formation of free radicals and the second the
metal complexes, such as, for example, olefine hydro-
formation of ionic intermediates. Then, different reac-
formylation, methanol carbonylation, ethylene oxida-
tion paths are possible for each of the mentioned
tion to acetaldehyde (Wacker process), etc. We report
mechanisms, as shown in tab.1. The role of catalyst
here, as an example, the catalytic cycle (an impressive
in a reaction is to favour the movement of the electrons
way to represent the mechanism of the reaction) of
involved in both the rupture and the formation of
hydroformylation catalysed by the Wilkinson com-
bonds. When the catalysts are complexes of transition
plex HRh(CO)L3 [13,14]:
metals, other possible mechanisms must be consid- The mechanism of hydroformylation catalyzed by
ered, because transition elements have the following the Wilkinson complex.
prerogatives: The insertion step is normally considered as rate
determining and on this basis the kinetic equation can
1. the possibility to change their oxidation state, be derived.
2. the possibility to give and bonds with d and f An example of the opportunity of interpreting the
internal orbitals, and kinetic law through the study of the reaction mechan-
118 E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123
Table 1
Classification of the mechanisms for a homogeneous reaction
by the formation of a new bond, but in this case, the r k pA pB will be respectively: 0<<1 and <0. This
molecules are previously chemisorbed on the catalyst kinetic model is currently known as Langmuir±Hin-
surface. The reactions occurring between the mole- shelwood mechanism. If adsorption or desorption
cules and the coordinatively unsaturated atoms called rates are determining steps, it is easy to derive there-
``active sites'' existing on the solid surface (chemi- from the corresponding kinetic equations which are,
sorption) can be considered elementary steps of the respectively:
reaction mechanism.
pB pB pB
The approximations introduced in the previous r rA kA pA ÿ 1
Adsorption;
Kp Kp bB
sections, that is, the ``rate determining step'' and the
``steady state'', can still be employed to construct the (43)
rate law equation. Let us consider, for example, the pB
r rB kB Ks pA ÿ pA
1 Ks
Desorption;
simplest reaction occurring in the heterogeneous Kp
phase (44)
A ! B: (37) where Kp being the equilibrium constant of the reac-
We can suggest a reasonable mechanism of the type tion A,B in the gas phase and Ks the equilibrium
A V , A Adsorption reaction of A; (38) constant of the surface reaction A,B; this last can
be set in a group with the other kinetic constants to
A , B Surface reaction; (39) have the smallest number of parameters to be eval-
B , B V Desorption reaction of B; (40) uated. We report here, as an example, the procedure
followed to obtain the kinetic law Eq. (43) valid when
where V being the fraction of empty sites and A and
adsorption is the rate determining step. The rates
B the fraction of sites occupied by A and B, respec-
corresponding to adsorption, surface reaction and
tively (Adsorption Coverage Degree). Therefore,
desorption, respectively, are
ABV1. The surface reaction may be char-
acterised by a sequence of many elementary steps A B
r r A k A pA v ÿ ; rs k s A ÿ ;
instead of only one as indicated here, but again we can bA Ks
solve the problem by considering only the rate deter-
pB v
mining step. For a slow surface reaction, the adsorp- rB kB B ÿ : (45)
bB
tion and desorption steps can be considered nearly at
equilibrium and the chemisorption equilibrium can Super®cial and desorption reactions can be consid-
well be described by the Langmuir isotherm, accord- ered nearly at equilibrium, that is rsrB0. As a
ing to which consequence:
bA pA bB pB rB rs pB B V pB
A ; B : 0 and B V ; A :
1 bA pA b B pB 1 bA pA bB pB kB K s bB Ks bB Ks
(41) (46)
The reaction rate, that is the rate of the slow step,
By substituting in the expression, we obtain
will be
kbA pA pB pB
r rs kA : (42) r A k A V pA ÿ kA V pA ÿ ;
1 bA pA bB pB bB Ks bA Kp
A ®rst observation about this kinetic law equation is (47)
that A and B compete in the adsorption on the empty being KPbBKsbA the equilibrium constant of the
sites and the Langmuir isotherm is able to account for reaction A,B in the gas phase. Remembering that
this competition. Consequently, both A and B have ABV1,
deactivating effects on the reaction, appearing at the
denominator and their apparent reaction order in an 1
V : (48)
experimentally determined ``power law'' of the type
1
pB =Kp
pB =bB
120 E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123
Substituting V in the relation (47), we obtain the On the contrary, in the ``redox'' or ``Mars±Krevelen''
already reported ®nal expression (43). mechanism the ``steady state'' hypothesis is normally
When the reaction considered is more complicated adopted, as we have seen in the above described
than that reported, A!B, say for example, example (23)±(28).
AB,CD (49) Reconsidering that example, if one or more species
are adsorbed on the catalytic sites, we get
the approach is similar and for a surface reaction oxredocc1, the mechanism becomes a mixed
controlling rate, we obtain one (Redox±Langmuir±Hinshelwood) and the reac-
r rs kA A B ÿ kÿr c B ; tion rate expression will be
C D k 1 pA
r rs kr A A ÿ : (50) r1 ÿ P : (56)
Ks 1
k1 =k2
pA =PnO2 1 i bi pi
c D bC pc bB pD
Considering that : Ks All the considerations reported in this section start
A B bA pA bB pB from the assumption that the catalytic sites on the
bC bD surface are uniform, that is, the enthalpy change for
Kp and that : i
bA bB the adsorption of a molecule is the same for all the
b i pi types of sites. In practice, the enthalpy change, as a
P (51)
1 j bj pj function of the surface coverage degree, normally
shows a declining trend. The conclusion is that sur-
it results faces are not homogeneous and sites are not uniform.
bA bB
pA pB ÿ
pC pD =Kp Another constraint of the Langmuir adsorption model
r kr : (52) that could be not respected in the actual case is that
1 bA pA bB pB bC pC bD pD 2
molecules should not have lateral interactions. Despite
In practice, some terms of the denominator are often these adverse observations, the Langmuir±Hinshel-
very small and can be neglected and the relations can wood model is almost universally applied and with
greatly be simpli®ed. few exceptions it performs quite well. This is due to
When two sites intervene in the adsorption of a the fact that normally only a small portion of the
molecule before the occurrence of the surface reaction catalytic sites are active in the adopted conditions and
and the adsorption on the second adjacent site is a rate these working sites can be considered uniform and not
determining step, we have interacting.
A ! A2 : (53) When the Langmuir isotherm is not suitable to
describe the adsorption of a molecule on a solid
The reaction rate in this case is surface, other isotherms can be introduced. A detailed
bA pA description of the non-uniformity of the catalytic sites
r kr A V k A
1 ÿ A ; r kr : can be made by introducing a distribution function of
1 bA pA 2
the heats of adsorption (Qads) such that (Qads) dQads
(54) gives the fraction of sites having adsorption heat in the
This mechanism is called ``dual site'' mechanism. range Qads and QadsdQads and
Another possibility is that a chemisorbed molecule QZads2
reacts with a molecule coming from the ¯uid phase
(``Rideal-Eley mechanism''), that is,
Qads dQads 1: (57)
Qads1
A B ! products in this case : r kr A pB
bA pA pB Different forms can be assigned to the function
kr : (55) (Qads), for example (Qads)constant, that is, the
1 bA pA
sites with different heat of adsorption have the same
In all the mechanisms considered we have intro- probability to exist. The resulting isotherm is the
duced the approximation of the rate controlling step. Temkin one and it can be used only for the inter-
E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123 121
mediate adsorption coverage degree, losing the phy- are necessary to accurately simulate the reaction
sical meaning at the low and high coverage, respec- occurring together with catalyst poisoning.
tively.
If we assume
Qads eQads , i.e. a Gaussian 4. Physical meaning and constraints of kinetic
distribution, the Freundlich isotherm can be derived,- parameters
which is useful at low and intermediate coverage
degrees. Again losing the physical meaning for the The kinetic parameters appearing in the kinetic law
high adsorption coverage degree. The non-uniformity equations must have a physical meaning. For example,
of the catalytic sites has other consequences on the the preexponential factor of a true kinetic constant,
kinetic behaviour of the catalysts being responsible that is, the kinetic constant of an elementary step
for the correlation often observed between the two performed in the gas phase should be lower than
kinetic parameters: the preexponential factor and the collision frequency of the molecules.
activation energy, respectively (compensation effect Similarly, for a gas±solid catalytic process, the
or theta rule). The correlation observed is an almost same parameter should be lower than the collision
linear trend for log Ad (the logarithm of the preexpo- frequency of the reacting molecules on the surface.
nential factor) and E (the activation energy) for dif- The activation energy must be a positive number
ferent catalysts of the same kind on the same reaction. greater than 5 kcal/mol and lower than 50 kcal/mol,
This correlation is due to the fact that, according to that is in the order of magnitude of the energy neces-
the operating conditions, a small number of very sary for the chemical bond breaking.
active sites of low activation energies or alternatively Adsorption parameters are equilibrium parameters
a large number of less active sites characterised by and,consequently, we can write
high activation energies can be involved.
b exp
Sads =Rexp
ÿHads =RT: (58)
Another consequence of surface dishomogeneity is
the change of selectivity with the dispersion of The constraints of these parameters are quite rigid.
the active part of the catalyst on an inert support of The absolute value of Sads, for example, must be
a high interphase area. The atoms of a dispersed greater than the translational entropy of the adsorbed
catalyst, for example of a dispersed metal, can be molecule and lower than the total entropy
geometrically located at one vertex, edge, plane, at one
Strans < Sads < Stot (59)
step, in a hole with different characteristics of unsa-
turation in the coordination. The number of each being
type of atom can strongly change with dispersion. Strans
3=2R lnM
5=2R lnP ÿ 2:2298
e:u::
When a reaction requires a particular type of site, The enthalpy change in adsorption is usually a
the rate of this reaction changes with the dispersion negative number falling in the range
(structure sensitive-reactions or demanding-reac-
ÿ50:000 < Hads < ÿ5:000:
tions). On the contrary, reactions not requiring a
particular degree of unsaturation and of geo- Another important constraint is imposed by the
metrical disposition of sites (structure-insensitive or thermodynamic equilibrium. The direct and inverse
facile-reactions) do not change activities with disper- reactions must be kinetically congruent. This means
sion. that when for example the kinetic law equation for a
Other kinetic aspects peculiar to heterogeneous direct reaction shows fractional orders, the reverse
catalysis are deactivation phenomena, normally occur- reaction will have fractional order too, and by assum-
ring by poisoning or sintering. Empirical correlations ing r0 the expression of the equilibrium constant
are commonly used to describe the deactivating effects must be obtained.
of both poisoning and sintering, when these phenom- This fact can usefully be exploited in kinetic stu-
ena are very slow compared with the contact time of dies, because it is possible to study the kinetics of a
the reactants in the reactor. When poisoning occurs reverse reaction, when this is easier, in order to de®ne
very fast and the catalyst must be regenerated before a the kinetic law equation of the corresponding direct
new cycle of utilisation, more rigorous kinetic models reaction.
122 E. Santacesaria et al. / Catalysis Today 52 (1999) 113±123
Table 3
Scheme of the phases occurring in the kinetic study of a reaction