Journal of Catalysis 276 (2010) 85–91

Contents lists available at ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Kinetic analysis of cluster size dependent activity and selectivity

Dmitry Yu. Murzin ⇑
Åbo Akademi University, Biskopsgatan 8, Turku/Åbo, Finland

a r t i c l e i n f o a b s t r a c t

Article history: The impact of nanoparticle size effects in heterogeneous catalytic kinetics over supported metal catalysts
Received 9 July 2010 is discussed taking into account different activities of edges and terraces. Kinetic equations for the Eley–
Revised 2 September 2010 Rideal mechanism as well as for the two-step sequence were derived, analyzed and compared with
Accepted 3 September 2010
experimental data for ethene hydrogenation. Quantitative description for selectivity in a consecutive
reaction of pyrrole hydrogenation/ring opening as function of the metal cluster size is provided demon-
strating good correspondence between theory and experiments.
Keywords:
Ó 2010 Elsevier Inc. All rights reserved.
Nanocatalysis
Cluster size effect
Kinetics

1. Introduction appropriate reaction conditions, samples with higher dispersion

of Pd showed intrinsically faster reaction rate in hydrogenation
Although the dependence of catalytic activity in transition me- of 1-hexyne, thus the reaction was considered to be size
tal catalysis on changes in the particles size has been a subject of dependent.
long-term research, recent years witnessed a renaissance in the Phenylacetylene hydrogenation over gold catalysts [17] exhib-
interest to catalytic properties of materials in the nanometer range, ited strong dependence of selectivity toward the intermediate
e.g. in the domain between 2 and 20 nm. product – styrene on cluster size. For example at a similar conver-
The ratio between the surface and volume is enlarged during sion level, an increase of the size from 2.5 to 30 nm resulted in a
the reduction in size of nanoparticles, thus the fraction of surface substantial decrease of styrene selectivity from 8% to 0.7%.
atoms is increasing [1]. Therefore, catalytic activity calculated per Similar to hydrogenation of phenylacetylene, in the Fischer–
total amount of catalytic phase usually declines with the increase Tropsch reaction selectivity to an intermediate product – methane
of cluster size as fewer exposed sites are available for catalysis. declined from 30% to 10%, while selectivity for C5+ products
Since changes in the metal particle size lead to alterations of the increased (from 51% to 85%) when the cobalt particle size was
relative ratio between edges, corners and terrace atoms, as well increased from 2.6 to 16 nm [18,19].
as changes in the electronic structures, turnover frequency (TOF), The effect of the nanoparticle size on the selectivity in cyclohex-
defined as the activity per unit of exposed surface, demonstrates ene hydrogenation/dehydrogenation, and hydrogenation of ben-
not only independence on size (structure insensitivity) but also zene and crotonaldehyde was discussed [11,12]. In the later case
structure sensitivity [2–10]. Such dependence of TOF in heteroge- [12], selectivity to crotyl alcohol decreased with the particle size
neous catalytic reactions over metals or metal oxides on the size increase. An interesting example of structure sensitivity is
of clusters (Fig. 1) is currently under intensive investigation. carbon–nitrogen ring opening in pyrrole hydrogenation for plati-
Less attention in the literature was devoted to analysis of selec- num nanoparticles smaller than 2 nm [20].
tivity dependence on the cluster size, although few examples were It should be, however, emphasized that quantitative analysis of
reported [11,12]. For instance in glycerol oxidation [13] under selectivity depending on the particles size is seldom reported,
alkaline conditions, selectivity to the intermediate glycerate in- although few attempts can be found in the literature accounting
creased significantly with the size increase from 3 to 17 nm. Other for kinetic features of the cluster size effect based on a thermody-
types of behavior were also reported in the literature. Selectivity in namic approach [21–24].
hydrogenation of triple bonds (acetylene [14] or 1-hexyne [15]) Speaking in terms of kinetics, let us consider a simple
over Pd was seen to be independent on the cluster size. It should reaction of monomolecular isomerization A ? B which follows a
be, however, noted that according to Anderson et al. [16] under mechanism

Aþ $ A
⇑ Fax: +358 2 215 4479. ð1Þ
E-mail address: dmurzin@abo.fi A ! Bþ

0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.09.003
86 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91

Nomenclature


the surface site hF surface coverage, of A
dcluster cluster size l(r) chemical potential
fterraces, fedges fractions of terraces and edges l(1) chemical potential of a metal particle of an infinite size
DGads,terraces Gibss energy of adsorption on terraces (bulk-like)
DGads,edges Gibss energy of adsorption on edges m the order of a cluster.
KA adsorption constant of A v parameter in Eq. (17)
k rate constant k xI frequencies of steps
Z surface site X the atomic volume of the bulk metal.
a Polanyi parameter
c surface free energy

where  is the surface site, adsorption step is at quasi-equilibrium different from that in a metal particle of an infinite size (bulk-like)
and the surface reaction is irreversible. The rate is expressed l(1) in the following way [21,22,25,26]
K ACA lðrÞ  lð1Þ ¼ 2cX=r ð3Þ
r ¼ khA ¼ k ð2Þ
1 þ K ACA where c is the surface free energy and X is the atomic volume of the
bulk metal. For a liquid, the surface energy excess is then a syno-
where KA is the adsorption constant and C is concentration of A, nym of the surface tension [27], and the later notion should not
respectively. be used for solids according to Kolasinki [27]. The treatment of
In the general case of real surfaces with lateral interactions, the Murzin and Parmon [21–24] assumed that changes of the chemical
rate constant k depends not only on the size of nanocluster r and its potential, when the size of nanocluster is changing, are dependent
morphology, but on surface coverage hF, since lateral interactions on the presence or absence of an adsorbate.
obviously are more prominent at higher coverage. In the current contribution, we would like to consider a situa-
In addition to the rate constant, also equilibrium constant can tion when (Fig. 2) the alterations of the chemical potential of
depend on the cluster size, thus coverage can also change with nanoclusters in comparison with bulk metal are the same indepen-
changes of the cluster size. dent on the surface coverage (e.g. bare surface or occupied with
Mechanistic explanations in the literature, which in principle reactants). Since differences in the activation energy between
should account for variations of rate and equilibrium constants edges and terraces are well recognized [7,28], the focus of the
with the particles size, include among others [5] variations in the paper will be in exploring kinetic consequences of different activ-
electronic properties of small metal clusters, e.g. more narrow ities of edges and terraces in terms of reaction rates and selectivity.
metal valance band, geometric aspects, e.g. changes in the relative
fraction of surface sites depending on the cluster size. Another
2. Geometrical and energetic characteristics of clusters
explanation can include presence of metal species with different
valencies (i.e. Pt4+, Pt2+ and Pt0) and thus different activity, which
For the sake of clarity in the following text, we will consider
relative distribution can alter depending on the cluster size.
only the terraces and edges, as sites with different reactivity. Thus,
In addition, mode of adsorption of mainly complex organic mol-
for adsorption, an expression for the Gibbs energy can be written
ecules can depend on the cluster size, leading also to changes in the
binding energy, which in turn will result in cluster size dependent DGads ¼ DGads;terraces fterraces þ DGads;edges fedges ð4Þ
reaction rates.
Strain or compression was also listed [5] among the possible where DGads,terraces and DGads,edges correspond, respectively, to
reasons for the cluster size effect. The latter concept was very adsorption on terraces and edges, while fterraces, fedges denote frac-
much behind the thermodynamic approach [21,22], which albeit tions of these surface sites, which sum is obviously equal to unity.
being rather formal, was successful in describing experimental
data for several heterogeneous catalytic reactions [23,24]. This
thermodynamic approach relied on the concept of chemical poten-
tial l(r), which in a metal particle with a radius of curvature r is

μr
ΔGr

ΔG∞ μ∞

Fig. 2. Energy diagrams illustrating chemical potential differences upon adsorption.

Fig. 1. Structure sensitivity plots. I, II correspond respectively to structure The chemical potential li(r) corresponds to a cluster radius r while l1 denotes a
sensitivity, while III represents a structure insensitive reaction. bulk value.
 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91 87

In a more general case, also terraces of different type (e.g. 1 0 0, Nedges

fedges ¼ ð10Þ
1 1 0 or 1 1 1 surfaces) could be considered, as well as other type of Nedges þ Nsquare faces þ Ntriangular faces
surface sites; however, in the current contribution, these sites are
Making use of expressions in [37], for cubo-octahedral structures
lumped and we will initially consider that the reactivity of square
Eq. (10) transforms into
and triangular faces in a cubo-octahedron are similar. In the cur-
rent treatment, the edges between terraces of different type (eq. 24ðv  1Þ
fedges ¼ ð11Þ
between (1 0 0) and (1 1 1) and between (1 1 0) and (1 1 1)) are 24ðv  1Þ þ 6ðv  1Þ2 þ 4ðv  1Þðv  2Þ
considered non distinguishable. Consideration of edge atoms
(along with corner atoms) as active sites was done for instance where m is the order of a cluster. This dimensionless value, which
in [6] for the case of CO oxidation on Au. Although for small mol- defines the particles size in terms of successive crust added to a
ecules the edges could be envisaged as active sites, most probably, pre-existing core, should be related to the cluster size. The fraction
for somewhat larger molecules, edge atoms together with terrace of surface atoms in edge sites, excluding those in vertices as a func-
atoms constitute an active site. In the present paper, the later tion of the cluster order, is presented in Fig. 3a. It should be noted
extension is not discussed for the sake of clarity. that the model is not based on a continuously variable particle size.
Eq. (4) can be modified Number of atoms in one edge, which for cubo-octahedral struc-
ture is equal to m  1, can be defined as the length of an edge di-
DGads ¼ DGads;terraces ð1  fedges Þ þ DGads;edges fedges vided by the metal–metal distance. For cubo-octahedral, the
¼ DGads;terraces þ fedges ðDGads;edges  DGads;terraces Þ ð5Þ cluster diameter is twofold the edge length, thus

Making use of the relationship between equilibrium constants and dcluster

m1¼ ð12Þ
Gibbs energy of adsorption one arrives at 2dMetal—metal
DGads;terraces þfedges ðDGads;edges DGads;terraces Þ Taking as an example distance between Pd–Pd atoms equal to
K ads ¼ e RT ð6Þ 0.275 nm [35], the fraction of surface atoms in edge sites as function
Linear free energy (or Brønsted–Evans–Polanyi) relationships are of the cluster size is presented in Fig. 3b. Since the assumed model
widely used in homogeneously and heterogeneously catalyzed is not a continuous one, the values of this fraction correspond to
reactions [29–34] in a form which relates reaction constants k with cluster sizes with a difference between the neighboring values
equilibrium constants K in a series of analogous elementary equal to 2dMetal–metal or 0.55 nm in particular case of Pd. For better
reactions: visibility points, corresponding not to all possible sizes of clusters
are given in Fig. 3b. As can be seen from this figure, the fraction
k ¼ gK a 0<a<1 ð7Þ of atoms in edges can be described in a simplified way as
fedges  1/dcluster when d is given in nm.
where g and a (Polanyi parameter) are constants. Eq. (7) related
It was assumed above that the activity of terraces with different
transition state properties ðDG– Þ and those of the initial and final
coordination numbers is the same. Generally, it could not
state (DG).
be the case, thus it is interesting to analyze the ratio between
Combining Eqs. (6) and (7) an expression for the rate constant of
square and triangular faces as a function of the cluster order.
adsorption easily follows
f(square/triangular+square) = 6(v  1)2/(4(v  1)(v  2) + 6(v  1)2). Such
DGads;terraces þfedges ðDGads;edges DGads;terraces Þ
kads ¼ gea RT
analysis (Fig. 3c) demonstrates that for a cluster order above 3
DGads;terraces afedges ðDGads;edges DGads;terraces Þ
(ca. 1.1 nm) fsquare/(square+triangular) is practically independent on the
¼ gea RT e RT ð8Þ cluster order, thus a direct dependence of the cluster size (nm) into
Eq. (9) could be introduced
which finally takes a form
aðDGads;edges DGads;terraces Þ
afedges ðDGads;edges DGads;terraces Þ 0
0
kads ¼ kads e RT ð9Þ kads ¼ kads e RTdcluster
ð13Þ

In order to progress further, geometrical properties of metal clusters where DGads,terraces defined as DGads,terraces = DGads,square_terracesfsquare/
should be considered. To this end, a shape, which can represent (square+triangular) + DGads,triangular_terracesftriangular/(square+triangular) can be
such a cluster, must be selected. Following [28,35], the cubo- approximately considered as independent on the cluster size.
octahedral geometrical model is a good representation of metal Interestingly enough, the rate constant of adsorption according
clusters often encountered in heterogeneous catalysis. In the stud- to (13) has almost the same dependence on the cluster size as de-
ies of supported Pd particles, the cubo-octahedral shape of the par- rived in [23,24] using a thermodynamic approach. It can also pro-
ticles was fully supported by the high resolution TEM investigation vide an easy mechanistic explanation for both increase or decrease
[35]. Certainly, cubo-octahedral geometrical model is a simplifica- of adsorption rates with cluster size increase, relating such behav-
tion of a true crystal, and it could be argued that this is a statistically ior to Gibbs energy of adsorption on edges and terraces.
improbable occurrence, as most will have a partial outer layer for
which many other types of sites or coordination number will ap- 3. Two-step reaction sequences
pear. Such representation is still probably sufficient for kinetic mod-
eling purposes at least as the first approach. It should be noted that The two-step mechanism with two kinetically significant steps
even encapsulated cubo-octahedral Pt nanoparticles can be pre- [38,39], which implies that one of the several surface intermedi-
pared using for example tetradecyltrimethylammonium bromide ates is the most abundant, while all the others are present on the
as capping agent with a high amount of ideal geometry, as demon- surface at much inferior concentration levels, is widely applied in
strated in [36], where 90% of particles were cubo-octahedra and just the literature to account for heterogeneous catalytic kinetics
10% were of irregular shapes. [40,41]
Expressions for the geometrical characteristics for several
structures, including cubo-octahedral one were presented in [37], 1: Z þ A1 $ ZI þ B1
relating the fractions of edges to the total number of atoms on the
2: ZI þ A2 $ Z þ B2 ð14Þ
surface, which includes besides edges also square and triangular
faces A1 þ A2 $ B1 þ B2
88 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91

(a) The reaction rate for this mechanism is well known [41,42]
0.8 k1 P A1 k2 PA2  k1 PB1 k2 PB2
v ðdÞ ¼ ð15Þ
k1 PA1 þ k2 PA2 þ k1 PB1 þ k2 PB2
fraction of atoms in edges

0.6 where PA1, etc., are partial pressures (for gas-phase reactions) or
concentrations (for liquid-phase reactions), ki – kinetic constants.
The rate constant for the first step of mechanism (14) is
0.4 a1 v

k1 ðdÞ ¼ k1 edcluster ð16Þ

where a1 is the Polanyi parameter for step 1 and v is given by
0.2
ðDGads;edges  DGads;terraces Þ
v¼ ð17Þ
RT
0.0
Similarly, the rate constant for the backward reaction is
ð1a1 Þv
0 5 10 15 20 25 30 k1 ðdÞ ¼ k1 e dcluster ð18Þ
cluster order
Analogously, it holds for the second step

k2 ðdÞ ¼ k2 eð1a2 Þv=dcluster ; k2 ðdÞ ¼ k2 ea2 v=dcluster ð19Þ

(b) Following [40], it can be assumed that a1 = a2 = a leading to
fraction of atoms in edges in cubo-octahedron

ðk1 k2 PA1 P A2  k1 k2 PB1 PB2 Þeð12aÞv=dcluster

v ðdÞ ¼
0.8 ðk1 P A1 þ k2 PB2 Þeav=dcluster þ ðk2 PA2 þ k1 P B1 Þeð1aÞv=dcluster
1/d ð20Þ
and for an irreversible reaction (k2  0)
0.6
x2 eð1aÞv=dcluster
v ðdÞ ¼ ð21Þ
1 þ x2 x
þx1 v=d
1
e cluster
0.4
with xI are frequencies of steps following the notation of Temkin
[41] (the rates of steps (stages) in a particular direction divided
by coverage of reacting in this step surface species, i.e. x1 = k1PA1,
0.2
etc.). Eley–Rideal mechanism is just a special case of Eq. (21) and
can be obtained if the frequency of the second step is assumed to
be much slower than x1. Eq. (21) can be rewritten as
0.0
0 5 10 15 p1 eð1aÞv=dcluster
v ðdÞ ¼ ð22Þ
cluster size, nm 1 þ p2 ev=dcluster
where p2 = (x2 + x1)/x1; p1 = x2.
(c) 1.0 f square/(square+triangular)
Eq. (22) in fact was derived in [24] and compared with experi-
0.9 mental data. Here, we would like to give just one example related
to a case when a maximum in TOF as a function of cluster size was
0.8 reported [43]. Hydrogenation of ethene on nanoscale catalyst par-
ticles at atmospheric conditions was described in [43] with special
0.7 emphasis on the influence of the Pd particle size on the reactivity
at industrially relevant process conditions varying the support
0.6 and the hydrogen to ethene ratio.
TOF values as a function of Pd particle size for Pd/TiO2 catalyst at
0.5 H2 to C2H4 ratios equal to 1:1 and 5:1 were presented for experiments
carried out at atmospheric pressure, 293 K and ethene concentration
0.4 of 8.6  104 mol l1. Experimental data [43] are redrawn in Fig. 4
demonstrating a very clear maximum for the stoichiometric ratio be-
0.3 tween hydrogen and ethene. This figure contains also a comparison
between experimental and calculated data according to Eq. (22), con-
2 4 6 8 10
firming good correspondence between experiments and calculations.
cluster order
No excessive dehydrogenation and coking were mentioned in
Fig. 3. Fraction of the atoms in edges to the sum of atoms in edges and faces as a [43], thus deactivation dependence on the cluster size could be dis-
function of (a) cluster order and (b) cluster size for cubo-octahedral structure; (c) carded. In general, it could be, however, an important issue, mak-
the ratio of the atoms in square terraces to the total number of atoms in ing determinations of TOF challenging.
terraces.
Certainly, in case of ethene hydrogenation, Eq. (22), which is
based on two-step sequence or Eley–Rideal mechanism, might
not be the correct one to account for experimental data, thus phys-
ico-chemical analysis of the values of kinetic parameters might not
where A1, A2 are reactants, B1, and B2 are products, Z is the surface be straightforward. As discussed in [23], Langmuir–Hinschelwood
site and I is an adsorbed intermediate. type of kinetic expressions could equally well explain a maximum
 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91 89

In Eq. (24), step 2 is a linear one. For the purpose of our analysis,
8 steps 1 and 3 are considered as quasi-equilibria. The equilibrium
constant of the first step can be determined from Eqs. (16) and (18)
a1 v

6 k1 edcluster v
K 1 ðdÞ ¼ ð1a1 Þv
¼ K 1 edcluster ð25Þ
k1 e dcluster
TOF, s-1

4 Analogously, for the third step

Chi^2/DoF = 0.08745
R^2 = 0.99347 ð1a3 Þv

P1 0.4954 ± 0.18352 k3 e dcluster v

2 P2 0.00038 ± 0.00032 K 3 ðdÞ ¼ a3 v ¼ K 3 edcluster ð26Þ
χ 22.73389 ± 1.34508
α 0.56109 ± 0.08402 k3 e dcluster

As the overall constant K = K1K2K3 does not depend on the cluster

0
size, it is apparently clear that for the isomerization step 2 in (24)
the equilibrium constant as well as the rate constants in forward
2 4 6 8
and reverse directions do not depend on the cluster size, leading fi-
Cluster size, nm nally to the rate expression when the second step is irreversible
v v
Fig. 4. TOF versus Pd particle radius for Pd/TiO2 at H2:C2H4 = 1:1 in ethene
hydrogenation at 293 K and ethene concentration 8.6  104 mol l1. Calculations
k2 K 1 C A edcluster k2 K 1 C A edcluster
v¼ v v ¼ v ð27Þ
according to Eq. (22). 1 þ K 1 C A edcluster þ K 1 1 þ ðK 1 C A e þ K 1
3 CBe 3 CBÞ
dcluster dcluster

where coverage of species A is defined as

in the TOF dependence on the cluster size as a two-step sequence.
v
Thus, in general evaluation of the cluster size effect should be com- K 1 C A edcluster
bined with a proper kinetic analysis, when a form of a kinetic equa- hA ¼ v ð28Þ
tion is established in a reliable way. In any case, the value of 1 þ ðK 1 C A e þ K 1
3 CBÞ
dcluster

Polanyi parameter was close to 0.5 often observed in heteroge- Analysis of (27) and (28) demonstrates that in principle, the turn-
neous catalytic reactions, while estimated value of v leads to a sub- over frequency can change with the particle size due to changes
stantial difference between adsorption on edges and terraces, e.g. in the coverage, while the rate constant will be cluster size indepen-
DGads,edges  DGads,terraces is above 55 kJ/mol. Even a larger differ- dent. Similar type of behavior was recently reported for Fischer–
ence between step-edge sites and terraces of 116 kJ/mol was re- Tropsch synthesis [19].
ported recently in [28].
Theoretical analysis of Eq. (22) clearly shows that the position
5. Selectivity dependence on the particle size
of the maximum in turnover frequency as a function of cluster size
is
For analysis of selectivity in structure sensitive reactions, it is
v interesting to consider hydrogenation of pyrrole to pyrrolidine
dat max ¼ ap2 ð23Þ
ln 1 a and butylamine

depending on the frequencies of steps e.g. on the ratio of (x2 + x1)

to x1, and thus on the partial pressures of the reactants. According
to experimental data [43], the maximum activity for Pd/TiO2 shifts
from a Pd particle size of 3 nm for stoichiometric conditions to lar-
ger particles sizes of about 4.2 nm for excess of H2.
ð29Þ
It should be noted that although in the analysis of experimental
data from [43] only an isothermal case was considered, in general The first reaction is structure insensitive, while selectivity in
the apparent activation energy can show dependence on the clus- butylamine and pyrrolidine depends on the cluster size.
ter size in agreement with Eq. (22). Such type of experiments were Let us now analyze the consecutive reaction network, which in
for instance performed for the total oxidation of methane over Pt/ general is able to explain selectivity in reactions with a similar
Al2O3 [44] confirming applicability the thermodynamic approach scheme as in pyrrole hydrogenation
[21–24], which resulted in an equation similar to Eq. (22). A recent Nð1Þ Nð2Þ
review of Bond [45] addresses various issues of the utilization of
1: A þ Z ¼ ZA 1 0
kinetics in evaluating mechanisms of heterogeneous catalysis,
stressing also that if the apparent activation energy is not constant 2: ZA þ H2 ! B þ Z 1 0
ð30Þ
over the range of a certain variable, the response to that variable 3: B þ Z ¼ ZB 0 1
will be different at each temperature. 4: ZB þ H2 ! Z þ C 0 1
Nð1Þ A ! B; Nð2Þ B ! C
4. Christiansen sequence
where A, B and C stand for pyrrole, pyrrolidine, n-butylamine,
An extension of the reaction mechanism discussed above is a respectively. On the right-hand side of the equations for the steps,
Christiansen sequence containing a linear step of isomerization the stoichiometric numbers along the routes N(1) are N(2) are given.
in the adsorbed state [46] Steps 1 and 3 are considered at quasi-equilibria, thus allowing to
write equations for equilibrium constants
1:  þA $ A  vA  vB

2:  A $ B K A ðdÞ ¼ K A edcluster ; K B ðdÞ ¼ K B edcluster ð31Þ

ð24Þ
3:  B $  þ B The rate constants of steps 2 and 4, which are considered irrevers-
A$B ible, are respectively
90 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91

k2 ðdÞ ¼ k2 eð1a2 ÞvA =dcluster ; k4 ðdÞ ¼ k4 eð1a4 ÞvB =dcluster ð32Þ 6. Conclusions

The reaction rate for consumption of pyrrole is thus given by The ability to control particle size of metal catalysts in the nano-
 vA
meter range allows examining the issue of structure sensitivity
k2 eð1a2 ÞvA =dcluster K A edcluster C A C H2 with well-defined nanoparticles. Difference in reactivity of edges
v2 ¼ vB ð33Þ and terraces is supposed in the present work to be responsible
 vA
C þK B edcluster
dcluster A
1 þ KAe CB for cluster size effects. The ratio between edges and terraces was
calculated for cubo-octahedral clusters and shown to be approxi-
while the generation rate of butylamine is expressed as mately proportional to the reciprocal value of the cluster diameter.
vB
Kinetic equations were derived for two-step sequence assuming
k4 eð1a4 ÞvB =dcluster K B edcluster C B C H2 one most abundant surface intermediate and for several variations
v4 ¼ vA  vB ð34Þ of this mechanism. The theoretical concept advanced in the pres-
1 þ K A edcluster C A þ K B edcluster C B
ent contribution leads to the same kinetic equations, previously
The selectivity toward the intermediate product B could be written derived [23,24] using a formal thermodynamic approach based
taking into account (33) and (34) on the changes of chemical potential. Comparison between exper-
imental and calculated data performed in [23,24] demonstrated
k4 ea4 vB =dcluster K B C B applicability of the kinetic equations to treat experimental data
SB ¼ 1  ð35Þ showing increase, decrease and even maxima in TOF as a function
k2 ea2 vA =dcluster K A C A
of cluster size. An example of such comparison for hydrogenation
Assuming that the values of Polanyi parameter are close to each of ethene on nanoscale catalyst particles at atmospheric conditions
other Eq. (35) can be simplified resulting in was presented in this paper, showing that a substantial difference
between adsorption on edges and terraces could be anticipated
k4 K B C B a4 ðvB vB Þ=dcluster based on this kinetic analysis. Analysis of a mechanism with a lin-
SB ¼ 1  e ¼ 1  p3 ep4 =dcluster ð36Þ
k2 K A C A ear step of isomerization in the adsorbed state (Christiansen se-
quence) demonstrated that turnover frequency can change with
where p3 stands for k4KBCB/k2KACA, while p4 is equal to a4(vB  vA). the particle size due to changes in the coverage, while the rate con-
Eq. (36) provides a possibility to evaluate dependence of selec- stant is cluster size independent in line with recent experimental
tivity on the particle size. When the difference in adsorption observations for Fischer–Tropsch synthesis on cobalt.
behavior of pyrrole on terraces and edges is not the same as this Finally, a quantitative analysis of selectivity was done for struc-
difference for pyrrolidine, it can be expected that selectivity will ture sensitive ring opening of pyrrole to n-butylamine through for-
depend on the cluster size. Let us consider now the experimental mation of pyrrolidine.
data reported in [20]. Selectivity to the intermediate pyrrolidine
decreased significantly with the size increase from 1 to 3 nm. Acknowledgments
Comparison between experimental and calculated values of
selectivity toward pyrrolidine done with Origin 7.5 software is pre- The author is grateful to Profs. G. Somorjai, E. Iglesia, V. Parmon,
sented in Fig. 5. As can be seen from Fig. 5, the trends in selectivity V. Bukhtiyarov, A. Stakheev for valuable discussions on various
are described in a rather accurate way. As pyrrole to pyrrolidine topics related to cluster size effect.
hydrogenation is structure insensitive, within the framework of
the model, developed here, it means that vA  0. Assuming the References
value of Polanyi parameter equal to 0.5, the value of DGads,edges 
DGads,terraces for pyrrolidine adsorption is close to 2.5 kJ/mol. [1] R. van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.
Finally note that parameters vB and vA depend on temperature, [2] A.T. Bell, Science 299 (2003) 1688.
[3] R. Schlögl, S.B. Abd Hamid, Angew. Chem. Int. Ed. 43 (2004) 1628.
indicating also that temperature dependence of selectivity might [4] R. Narayanan, M.A. El-Sayed, Top. Catal. 47 (2008) 15.
be different depending on the cluster size. [5] F. Klasovsky, P. Claus, in: B. Corain, G. Schmid, N. Toshima (Eds.), Metal
Nanoclusters in Catalysis and Materials Science: The Issue of Size Control,
Elsevier, Amsterdam, 2008, pp. 167–181.
[6] B. Hvolbæk, T.V.W. Janssens, B.S. Clausen, H. Falsig, C.H. Christensen, J.K.
Nørskov, Nanotoday 2 (2007) 17.
1.0 [7] R.A. van Santen, Acc. Chem. Res. 42 (2009) 57.
[8] M. Che, C.O. Bennett, Adv. Catal. 36 (1989) 55.
[9] C.R. Henry, Appl. Surf. Sci. 164 (2000) 252.
0.8 Chi^2/DoF = 0.00204 [10] M. Boudart, Adv. Catal. 20 (1969) 153.
R^2 = 0.91066 [11] G.A. Somorjai, J.Y. Park, Angew. Chem. Int. Ed. 47 (2008) 9161.
[12] G.A. Somorjai, J.Y. Park, Top. Catal. 49 (2008) 126.
Selectivity

P3 1.11261±0.04537
0.6 P4 0.37986±0.06209 [13] N. Dimitratos, J.A. Lopez-Sanchez, D. Lennon, F. Porta, L. Prati, A. Villa, Catal.
Lett. 108 (2006) 147.
pyrrolidine [14] M. Ruta, N. Semagina, L. Kiwi-Minsker, J. Phys. Chem. C 112 (2008) 13635.
0.4 [15] N. Semagina, A. Renken, L. Kiwi-Minsker, J. Phys. Chem. C 111 (2007) 13933.
n-butylamine
butane, ammonia [16] J.A. Anderson, J. Mellor, R.P.K. Wells, J. Catal. 261 (2009) 208.
[17] S.A. Nikolaev, V.V. Smirnov, Gold Bull. 42 (2009) 182;
0.2 S.A. Nikolaev, V.V. Smirnov, Catal. Today 147S (2009) S336.
[18] G.L. Bezemer, J.H. Bitter, H.P.C.E. Kuipers, H. Oosterbeek, J.E. Holewijn, X. Xu, F.
Kapteijn, A.J. Van Dillen, K.P. de Jong, JACS 128 (2006) 3956.
0.0 [19] J.P. den Breejen, P.B. Radstake, G.L. Bezemer, J.H. Bitter, V. Frøseth, A. Holmen,
K.P. de Jong, JACS 131 (2009) 7197.
0 2 4 6 [20] J.N. Kuhn, W. Huang, C.-K. Tsung, Y. Zhang, G.A. Somorjai, JACS 130 (2008)
14026.
cluster size [21] V.N. Parmon, Doklady, Phys. Chem. 413 (2007) 42.
[22] V.N. Parmon, Thermodynamics of Non-Equilibrium Processes with a Particular
Fig. 5. Measured [20] and modeled (Eq. (36)) selectivity toward pyrrolidine in Application to Catalysis, Amsterdam, Elsevier, 2010.
hydrogenation of pyrrole over well-defined platinum nanoparticles prepared using [23] D.Yu. Murzin, Chem. Eng. Sci. 64 (2009) 64.
dendrimer and polymer capping agents and supported onto mesoporous SBA-15 [24] D.Yu. Murzin, J. Mol. Catal. A. Chem. 315 (2010) 226.
silica. [25] S. Parker, C.T. Campbell, Phys. Rev. B 75 (2007) 035430.
 D.Yu. Murzin / Journal of Catalysis 276 (2010) 85–91 91

[26] C.T. Campbell, S.C. Parker, D.E. Starr, Science 298 (2002) 811. [36] K.M. Bratlie, H. Lee, K. Komvopoulos, P. Yang, G.A. Somorjai, Nanoletters 7
[27] K.W. Kolasinki, Surface Science. Fundamentals of Catalysis and Nanoscience, (2007) 3097.
Wiley, Chichester, 2002. [37] J.M. Montejano-Carrizales, F. Aguilera-Granja, J.L. Moran-Lopez, Nanostruct.
[28] R. van Santen, in: A. Cybulski, J.A. Moulijn, A. Stankiewicz (Eds.), Novel Mater. 8 (1997) 269.
Concepts in Catalysis and Chemical Reactors: Improving the Efficiency for the [38] M.I. Temkin, Kinet. Katal. 25 (1984) 299.
Future, Wiley, Weinhem, 2010, pp. 1–30. [39] M. Boudart, K. Tamaru, Catal. Lett. 9 (1991) 15.
[29] N. Brønsted, Chem. Rev. 5 (1928) 231. [40] M. Boudart, Kinetics of Chemical Processes, Prentice-Hall, Englewood Cliffs, NJ,
[30] L.P. Hammett, Physical Organic Chemistry, McGraw-Hill Book Co Inc., New 1968.
York, NY, 1940. [41] M.I. Temkin, Adv. Catal. 28 (1979) 173.
[31] M.G. Evans, M. Polanyi, Trans. Faraday Soc. 34 (1938) 11. [42] D.Yu. Murzin, T. Salmi, Catalytic Kinetics, Elsevier, 2005.
[32] A. Logadottir, T.H. Rod, J.K. Nørskov, B. Hammer, S. Dahl, C.J.H. Jacobsen, J. [43] A. Binder, M. Seipenbusch, M. Muhler, G.J. Kasper, J. Catal. 268 (2009) 150.
Catal. 197 (2001) 229. [44] I.E. Beck, V.I. Bukhtiyarov, I.Yu. Pakharukov, V.I. Zaikovsky, V.V. Kriventsov,
[33] D. Loffreda, F. Delbecq, F. Vigné, P. Sautet, Angew. Chem. Int. Ed. 48 (2009) V.N. Parmon, J. Catal. 268 (2009) 60.
8978. [45] G.C. Bond, Catal. Rev. Sci. Eng. 50 (2008) 532.
[34] R.A. van Santen, M. Neurock, S.G. Shetty, Chem. Rev. 110 (2010) 2005. [46] F.G. Helfferich, Kinetics of homogeneous multistep reactions, in: R.G.
[35] G. Agostini, R. Pellegrini, G. Leofanti, L. Bertinetti, S. Bertarione, E. Groppo, A. Compton, G. Hancock (Eds.), Comprehensive Chemical Kinetics, vol. 38,
Zecchina, C. Lamberti, J. Phys. Chem. C 113 (2009) 10485. Elsevier, Amsterdam, 2001.