Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Chemical Engineering and Processing 48 (2009) 837–845

Contents lists available at ScienceDirect

Chemical Engineering and Processing:


Process Intensification
journal homepage: www.elsevier.com/locate/cep

Modelling of enantioselective and racemic hydrogenation of ethyl pyruvate on a


Pt/Al2 O3 catalyst in the presence of microwave irradiation
B. Toukoniitty a,∗ , J. Wärnå a , J.-P. Mikkola a , M. Helle b , H. Saxén b , D.Yu. Murzin a , T. Salmi a
a
Åbo Akademi University, Process Chemistry Centre, Laboratory of Industrial Chemistry, 20500 Turku/Åbo, Finland
b
Åbo Akademi University, Heat Engineering Laboratory, 20500 Turku/Åbo, Finland

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

Article history: A kinetic model was developed for racemic and enantioselective ethyl pyruvate hydrogenation based on
Received 13 September 2007 selective and unselective reaction routes in the presence of a solid catalyst (Pt/Al2 O3 ), a catalyst modifier
Received in revised form 21 October 2008 (cinchonidine) and microwave irradiation. The temperature of the catalyst bed was estimated with the
Accepted 23 October 2008
aid of a mathematical model for heat transfer in a packed bed as well as with energy balances for solid
Available online 7 November 2008
and liquid phases. The microwave reactor was modelled with mass and energy balances, which were
decoupled, since the heat effect of the hydrogenation process was negligible compared to that of the
Keywords:
microwave source. The Langmuir–Hinshelwood type of approach was used for the kinetic model, which
Modelling
Hydrogenation
was incorporated in a non-steady state plug flow reactor model. The proposed models described well the
Ethyl pyruvate observed hydrogenation kinetics and the temperature behaviour of the microwave reactor.
Microwave irradiation © 2008 Elsevier B.V. All rights reserved.

1. Introduction on the catalyst surface thus creating hot spots. For example, Pt
sites on SiO2 and ␣-Al2 O3 supports can be selectively heated, since
Heating of reactive systems by microwave irradiation has been microwaves do not efficiently couple with the transparent ceramic
intensively investigated during the last twenty years. Several ben- supports, but may strongly couple with metal particles on the sup-
eficial effects under microwave heating have been reported, for port surface because of their high electrical conductivity. This offers
instance, increased reaction rate and selectivity for various kinds an opportunity to selectively heat the metal particles on the sup-
of reactions. However, the mechanisms of these effects are not yet port surface to temperatures higher than the support matrix [7].
fully understood. In many heterogeneously catalyzed liquid–solid Unfortunately, to get a direct experimental proof of this effect is
and gas–solid systems, significant differences between the rates extremely difficult, since temperature measurement of individual
of conventionally and microwave-heated reactions have been active sites in nanoscale is beyond current experimental capabili-
observed [1,2]. This phenomenon has been explained in terms of ties. However, the temperature of reactive sites has been calculated
microwave effects. Microwave effects are obtained due to the direct by Chemat et al. [8] to be 9–18 K above the bulk temperature. Fur-
interaction of microwave field and irradiated material and thus thermore, the improvement of the reaction rate explained in terms
directly causing temperature fields when adsorbed in the irra- of higher temperatures at active centres on the catalyst surface has
diated media. These effects cannot be obtained by conventional been claimed by several authors [9,10]. On the other hand, some
heating since in case of conventional heating the heating is caused scientists argue that tiny metallic particles typical for supported
by convection and conduction and not by dielectric loss as in case metal catalysts cannot achieve a temperature significantly exceed-
of microwave heating. Microwave effects are proposed to include ing that of the catalyst support because of energy transport at this
superheating and selective heating and formation of hot spots. small scale [11]. Nevertheless, the studies of Thomas et al. [7,12]
However, not all heterogeneous catalytic reactions can be accel- suggest that selective heating of active sites may be enhanced by
erated by microwave irradiation as discussed in references [3–6]. heating with higher frequency microwaves, and by using catalysts
The advantage of microwave irradiation for heterogeneously with large metal cluster sizes.
catalyzed systems is that microwaves do not substantially heat Hot spots are not always exclusively localised on active sites, but
up adsorbed organic layers, but interact directly with metal sites also involve the support material. Temperature measurements of
catalyst beds are nowadays feasible by means of optical and infrared
pyrometers. However, the temperature of the spatial hot spot on
∗ Corresponding author. Tel.: +358 2 215 3248; fax: +358 2 215 4479. the support is different from the average temperature of the cata-
E-mail address: btoukoni@abo.fi (B. Toukoniitty). lyst bed. Zhang et al. [10,13] have approximated the temperature of

0255-2701/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.cep.2008.10.007
838 B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845

Fig. 1. Reaction scheme of ethyl pyruvate hydrogenation.

spatial hot spots on a MoS2 /Al2 O3 catalyst to be 100–150 K above into the reactor (quartz tube) and activated prior the reaction in situ
the average temperatures in bulk catalyst material. at 290 ◦ C under H2 flow (1 bar) for 2 h under conventional heat-
Selective heating and superheating of the catalyst bed can in ing (heating jacket). Catalyst preparation is described elsewhere
some cases have a beneficial effect on heterogeneous catalytic reac- [28]. After cooling the catalyst to 25 ◦ C, the reactant ethyl pyruvate
tions in terms of both rate acceleration and selectivity enhancement (≥97%, Fluka) (A), catalyst modifier (−)-cinchonidine (CD) (≥98%,
[14,15]. This makes ethyl pyruvate hydrogenation (Fig. 1) an inter- Fluka) and solvent toluene (>99.5%, Baker) or ethanol (99.5%, Altia)
esting model reaction for investigation of microwave irradiation, were introduced into the loop reactor set-up. Racemic hydrogena-
since ethyl pyruvate hydrogenation leads to two optical isomers, tion experiments were conducted in the absence of the catalyst
R- and S-ethyl lactate. Furthermore, hydrogenation of ␣-ketoesters modifier (CD). The concentrations of ethyl pyruvate and CD were 0.1
is well studied under conventional heating [16–18], which enables and 6.2 × 10−4 mol dm−3 , respectively. Ethyl pyruvate was distilled
an interesting comparison of the results obtained with microwave in vacuo prior to the reaction. The liquid flow rate in the loop was
and conventional heating. 1.2 dm3 min−1 and the reaction temperatures were 23 ◦ C in toluene
Modelling of racemic hydrogenation leading to fifty-fifty prod- and 36 ◦ C in ethanol. The reactor (quartz tube) was placed into
uct mixture is a relatively simple task compared to enantioselective the microwave cavity (magnetron) and reaction was commenced.
hydrogenation. Enantioselective hydrogenation of ␣-ketoesters on Samples (∼1000 ␮l) were withdrawn and analyzed with a gas chro-
cinchona-modified Pt catalysts has been intensively investigated matograph equipped with a ␤-dex 225 (Supelco) chiral column,
and both Langmuir–Hinshelwood [19,20] and Eley–Rideal type analyses details described in [28]. The enantiomeric excess of the
[21] mechanisms have been proposed for this system. The major product, (R)-ethyl lactate (Fig. 1) is defined as
obstacle in the derivation of kinetic models is the lack of precise
R−S
information about the interaction between the modifier, the reac- ee = 100% (1)
tant and the catalyst surface. Different hypotheses concerning the R+S
active sites have been proposed, as enantiodifferentiaton proceeds. where R and S denote the concentrations of (R)- and (S)-ethyl lac-
Blaser et al. [19] have considered only one kind of active sites, where tate, respectively.
hydrogen, ethyl pyruvate and modifier can competitively absorb. An The experiments were performed in the flexible single-mode
alternative approach to modelling is a two-site, two-step mecha- loop reactor (Fig. 2) equipped with microwave and conventional
nism, according to which both modified and non-modified sites are
considered [22].
In recent years, computer simulation has become a more com-
mon tool in the design of microwave applicators [23,24]. The most
popular numerical schemes for this kind of studies are the Finite
Difference in the Time Domain (FDTD), the Finite Element Method
(FEM) and the Finite Integration Technique (FIT). However, to best
of our knowledge there are only few previous studies in the field of
reaction kinetic modelling in the presence of microwave irradiation
[25–27].

2. Experimental

Hydrogenation reactions were carried out in a single-mode


microwave loop reactor (Fig. 2) operating at 7 bar. 5 g of Pt/Al2 O3
catalyst (3 wt.% Pt determined by directly coupled plasma (DCP) Fig. 2. Schematic drawing of single-mode microwave loop reactor set-up (MJ: metal
method), alumina spheres (La Roche A-201, d ∼ 2.2 mm) was placed jacket with N2 flow, FC: loop filling cup, OF: optic fibre).
B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845 839

heating, necessary control units for accurate temperature adjust- where Q̇MW is the microwave effect, t the time increment, A
ment as well as optical fibre technology (Fiso Technologies Inc.) for the catalyst outer surface area, h the heat transfer coefficient, mS
temperature measurement. The microwave power input delivered the mass increment of the solid catalyst, cVS the heat capacity of the
to the system was varied between 100 and 700 W. The majority solid material, TS and TL temperatures of solid (catalyst) and liquid
of the experiments were carried out with power input of 300 W. (solvent), respectively. A and m can be related to the volume
The microwave power inputs corresponds to microwave power increment as follows
levels specified by equipment supplier. The experiments carried
out under conventional (conductive–convective) heating were per- A = aV (3)
formed in the same reactor set-up. In these experiments microwave mS = VBS (4)
cavity (magnetron) was removed and loop reactor was heated by
using a heating band. The details of the procedure and the reactor where a is the outer surface area-to-volume ratio and BS is bulk
system are provided in the previous work of our group [28]. density of the solid material. By inserting Eqs. (3) and (4) into Eq.
(2) we obtain

3. Results and discussion Q̇MW  TS


+ (−Hrj )Rj BS = ha(TS − TL ) + cVS BS (5)
V t
3.1. Estimation of the catalyst temperature For the solid material, the heat capacities cVS and cPS can be
set equal. Thus we use cPS for the sake of convenience. By let-
The knowledge of the catalyst temperature during the course of ting the volume and time increment to diminish, i.e. V → 0,
reaction is essential for the understanding of the microwave effect t → 0, and assuming a constant microwave effect along the
being present. In our reactor system, it was not possible to directly bed (Q̇MW /V ) = qMW , the energy balance for the solid phase
measure the catalyst temperature in the presence of microwave becomes
irradiation. However, the temperature profiles of the bulk liquid ⎛ ⎞
before the microwave cavity, in the cavity and after the cavity were dTS 1 
logged by means optic fibre technology equipment and thermocou- = ⎝qMW + (−Hrj )Rj BS − ha(TS − TL )⎠ (6)
dt cPS BS
ples (Fig. 3). The temperature of the Pt/Al2 O3 catalyst in reactions, j
where toluene was used as a solvent was estimated with the aid of
The contribution of the chemical reactions to 
the energy balance
a mathematical model for heat transfer in a packed bed and energy
turned out to be negligible and thus the term j
(−Hrj )BS was
balances for solid and liquid phase.
discarded. Then the final form of the energy balance for the solid
phase can be thus written as
3.1.1. Energy balances
The energy balances for solid and liquid phases for plug flow dTS 1
= (qMW − ha(TS − TL )) (7)
(PFR) and continuous stirred tank reactor (CSTR) models are derived dt cPS S
in sequel. The following basic assumption are applied: microwave
3.1.1.2. Energy balance for the liquid phase (plug flow reactor model,
energy is absorbed exclusively by the solid material, but not by the
PFR). By assuming plug flow model for the liquid phase, the energy
liquid phase and the energy balance of the gas phase is discarded
balance can be expressed as
because of the heat capacity of gas is much less than that of liquid.
h(TS − TL )A t = ṁL cPL TL t + cVL mL TL (8)
3.1.1.1. Energy balance for the solid phase. An infinitesimal volume where ṁL and cPL denote the liquid mass flow and the heat capac-
element (V) in the microwave reactor is considered to derive the ity of the liquid, respectively. By inserting Eq. (3) into the energy
energy balances. The energy balance for the solid phase can be balance, approximating cVL ≈ cPL and letting V → 0, t → 0 we
expressed as obtain
 dTL 1
 dTL

Q̇MW t + (−Hrj )Rj mS t = ha(TS − TL ) − ṁL cPL (9)
dt cPL L εL dV
= h(TS − TL )A t + cVS mS TS (2) Furthermore, the volume increment V is defined by

V = A l (10)

A
where l denotes to the length coordinate and is the cross section
of the reactor.
The volumetric flow rate (V̇L ) of the liquid is expressed by

V̇L = wL A (11)

where wL is the supercritical liquid velocity. By inserting Eqs. (10)


and (11) into Eq. (9), we obtain
dTL 1
 ha(T − T ) dTL

S L
= − wL (12)
dt εL cPL L dl
Finally, the energy balance for the liquid phase according to the
plug flow reactor can be expressed as
dTL 1
 ha(T − T ) dTL

S L
= − (13)
dt εL cPL L dL

Fig. 3. Temperatures before cavity (T1 ), in the cavity (optic fibre) and after cavity where L = L/wL and corresponds to the liquid residence time in
(T2 ). the catalyst bed and εL is the liquid hold-up.
840 B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845

3.1.1.3. Energy balance for the liquid phase (continuous stirred tank The model precisely estimated the solvent (toluene, liquid) tem-
reactor model, CSTR). The energy balance for the continuous stirred perature (Fig. 4) as well as the catalyst (Pt/Al2 O3 , solid) temperature
tank reactor model can be obtained from Eq. (9) by integrating during the course of the reaction, which is illustrated by Fig. 4.
over the entire reactor volume. The integration is straight forward,
since the reactor content is assumed to be uniform. Furthermore, 3.1.2. Heat transfer in packed bed
if the temperature dependence of the heat capacity is assumed to An alternative approach is to roughly estimate the catalyst tem-
be negligible, we obtain by using analogous assumptions to those perature by assuming steady state in Eq. (7), thus
introduced in the previous section
qMW − ha(TS − TL ) = 0 (16)
ṁL cPL (TL − T0L ) dTL
h(TS − TL ) = + cVL L εL (14) from which TS can be solved as a function of qMW and TL .
V dt
The heat transfer coefficient between solid and fluid (h) was
where V is the reactor volume. By assuming that cVL ≈ cPL and calculated by using standard equations for heat transfer in packed
ṁL /V = L /L we get beds [30]. The catalyst support temperature was estimated to be
dTL 1
 ha(T − T ) TL − T0L
only 2.2 ◦ C above the bulk liquid temperature (h = 5455 W/m2 K), in
S L
= − (15) case that bubble formation was neglected. Assuming interference
dt εL cPL L L
of gas bubbles between solid and liquid heat transfer, the catalyst
where T0L is the liquid temperature at the reactor entrance, before temperature was approximated to be 4.0 ◦ C higher than the bulk
the cavity. liquid (h = 4572 W/m2 K). This estimation is in agreement with the
The PFR and CSTR models consist of hyperbolic partial differ- previous evaluation (Section 3.1.1) that the catalyst particles are
ential equations (PDEs) and ordinary differential equations (ODEs), not significantly heated above the bulk liquid temperature as the
respectively. The PDEs were converted to ODEs by discretizing the reaction is carried out in toluene.
spatial coordinate in Eq. (12) with backward differences. The ODE One of the reasons for the moderate increase of the Pt/Al2 O3
solver was operated under a parameter estimator implemented in catalyst temperature above the bulk liquid is the inability of Al2 O3
software MODEST [29]. (tan ı = 0.006604) to efficiently couple with microwave irradiation.
Interference of gas bubbles between the solid and liquid heat The dielectric loss tangent defines the ability of a material to con-
transfer was assumed and it was incorporated into the heat transfer vert electromagnetic energy into heat, at a certain frequency and
coefficients. It turned out that a substantial amount of the dielectric temperature
heat was lost in the reactor system due to the inability of the reac- ε
tion medium to efficiently couple with the microwave irradiation tan ı = (17)
ε
and thus the real amount of the microwave heat effect (qMW ) deliv-
where ε describes the ability of a molecule to be polarized by
ered to the reaction system was significantly less than microwave
electric field and ε is the dielectric loss, which describes the effi-
power input transmitted into the system. This observation was
ciency with which the energy of the electromagnetic irradiation is
implemented in our model, by introducing a factor qMW = ˛qMW
converted into heat.
where qMW is the microwave input into the system and qMW is the
However, a decrease of the specific pore volume was observed
real amount absorbed by the system empirical correlations factor.
for Pt/Al2 O3 catalyst treated in the presence of microwave irra-
For parameter estimation use the energy balance Eqs. (7) and
diation under a high microwave input (0.48 cm3 /g) compared to
(15) were written in the form dTS /dt = p1 (˛qMW − p3 (TS − TL )),
the fresh catalyst (0.75 cm3 /g) and the catalyst treated at a low
dTL /dt = (1/εL )(p3 (TS − TL ) − p4 (TL − T0 )). The estimated temper-
microwave power input (0.80 cm3 /g). This might be caused by the
ature increase between the solid catalyst and the liquid was about
catalyst surface restructuring at a higher intensity of the electro-
4 ◦ C (Fig. 4). The lumped parameters p1 , p3 , and p4 were estimated
magnetic field as demonstrated by scanning electron micrographs
and the following parameter values were obtained: p1 = 13,073,
(SEM, Fig. 5).
p3 = 0.0067 min−1 , p4 = 0.087 min−1 and ˛ (1 K W−1 min−1 ) is used
An appearance of hot spots on active sites can be eliminated
for unit conversion.
due to insufficiently large metal particles (<100 nm), since the tiny
metallic particles cannot be heated to temperatures significantly
higher than those of the support matrix [7]. Additionally, Perry
et al. have proved for CO oxidation that Pd metallic particles are
not significantly hotter than the Al2 O3 support and they excluded
the possibility to induce temperature gradients between metal-
lic particles (1–100 nm) and their surroundings in a conventional
supported metal catalyst by using microwave energy [31].

3.2. Mass balances

For parameter estimation purposes, mass balances for the com-


ponents in the microwave reactor are needed. Since the reaction
rate and selectivity remained unaffected by the mode of heating and
no dramatic temperature gradients between the catalyst bed and
bulk liquid were established, a conventional approach to reactor
model was applied.
In case of experiments carried out in the presence of the catalyst
modifier, diffusion-limited kinetics was considered and the effec-
tiveness factor (eff ) was included into the model. The hydrogen
Fig. 4. The temperature estimation of the solid catalyst (Pt/Al2 O3 ) and liquid sol-
vent (toluene, in the catalyst bed) during the course of reaction under microwave mass balance was considered by using hydrogen solubility values
irradiation—experiments ( ) and model fit (—). found from literature [32] as initial concentrations for hydrogen.
B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845 841

Fig. 5. SEM image of fresh (a) and microwave treated (b) Pt/Al2 O3 catalyst.

Due to the large initial liquid amount, the liquid volume did not con- The influence of the mode of heating on the catalyst activity was
siderably decrease by sampling and thus the changes in the liquid investigated in the absence of the catalyst modifier at 23 ◦ C (liquid
volume-to-catalyst ratio were not taken into account. bulk temperature, 300 W) and 7 bar. The overall hydrogenation rate
Gas–liquid mass transfer limitations were not included in the was not affected by the way of heating (Table 1) and practically
model since the reaction was rather slow and catalyst amount com- equal initial rates were observed for conventional and microwave
pared to liquid volume was low. Therefore the liquid phase was heating.
assumed to be saturated with hydrogen meaning constant hydro- The effect of solvent polarity on the enantioselective hydrogena-
gen concentration. tion of ethyl pyruvate was studied in ethanol and toluene. In the
Consequently, the mass balances for the organic compounds case of non-polar toluene, which is microwave transparent, no dif-
according to the plug flow model become ferences in the overall reaction rate and enantioselectivity were
dci
 dci
observed in the presence and absence of microwave irradiation
= ε−1
L
eff,i B ri − (18) (Table 1). These observations could be explained by the inability
dt dL
of Pt/Al2 O3 catalyst to efficiently absorb microwaves and thus be
and for the continuous stirred tank reactor model we have analo- significantly heated above the bulk liquid temperature. In the case
gously of ethanol, the reaction rate remained unaffected by the mode of
dci
 c − c0i
heating. However, the enantiomeric excess decreased more than
= ε−1
L
eff,i B ri − i (19) 20% under microwave exposure (Table 1). This was probably due
dt L
to local superheating of polar ethanol in the cavity. High reaction
The generation rates (ri ) are obtained from the stoichiometry temperatures are not feasible for cinchona-alkaloid modified sys-
 tems, as the enantioselectivity is known to dramatically decrease at
ri = ij Rj . . . (20)
higher temperatures (<50 ◦ C) [33,34]. The decrease of enantioselec-
where ij denotes the stoichiometric coefficient of component i in tivity at higher temperatures is attributed to the desorption of the
reaction j and Rj is the rate of reaction j. modifier and temperature induced changes in the adsorption mode
of modifier. A more detailed discussion of these results is provided
3.3. Qualitative kinetics in a previous work of our group [28].
The size of the catalyst particle significantly influences overall
In the following, the qualitative aspects of ethyl pyruvate (A) reaction rates and selectivities in three-phase catalytic reactions.
hydrogenation kinetics on Pt/Al2 O3 are presented. The principal The mass transfer effect inside the catalyst particles was deter-
reaction scheme for hydrogenation of A is displayed in Fig. 1. A mined by estimating the effectiveness factors from the initial
typical kinetic behaviour in the presence of the catalyst modifier reaction rate [35]. First order kinetics with respect to reactant
(CD) is illustrated in Fig. 6. concentration (ethyl pyruvate) was assumed. In toluene (during
racemic reaction), the effectiveness factor was calculated to be
about eff,H2 ≈ 0.9, indicating that diffusion of hydrogen inside
the catalyst pores does not considerably affect the reaction rate.
However, in the case of toluene (enantioselective conditions) and
ethanol (enantioselective conditions), eff,H2 was estimated to be
0.75 and thus notifying that the processes are affected by internal
diffusion resistance.

Table 1
The solvent effect on the initial reaction rate and enantiomeric excess of the product
(ee) over Pt/Al2 O3 (ee at 50% conversion).

Solvent Initial reaction rate [mol/l gcat min] ee [%]

CH MW CH MW

Toluene 1.2 × 10−4 1.3 × 10−4 0 0


Toluene + CD 3.9 × 10−4 3.6 × 10−4 73 74
Fig. 6. Typical enantioselective hydrogenation kinetics of A at 23 ◦ C and 7 bar under
Ethanol + CD 2.3 × 10−4 2.6 × 10−4 53 31
MW irradiation ( , ethyl pyruvate; , (R)-ethyl lactate; and
, (S)-ethyl lactate).
842 B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845

Adsorption quasi-equlibria are written for substrate (A), the


modifier (M) and the substrate–modifier (AM) complex as follows,
cA∗
KA = (31)
cA c∗
cM ∗
KM = (32)
cM c∗
cAM ∗
KAM = (33)
cA cM ∗
The concentrations cA∗ , cM ∗ and cAM ∗ are solved from the above
equations and inserted in the total balance of organic adsorbates,

cA∗ + cAM ∗ + cM ∗ + c∗ = c0 (34)


Fig. 7. A model for formation of (R)- and (S)-ethyl lactate in the presence and in the where c0 is the total concentration of sites available for organic
absence of catalyst modifier. adsorption. The concentration of vacant sites (c* ) is solved by insert-
ing Eqs. (31)–(33) into Eq. (34),
3.4. Reaction mechanism and rate equations c0
c∗ = (35)
1 + KA cA + KM cM + KAM KM cA cM
The kinetic model of ethyl pyruvate hydrogenation is based on
the reaction scheme presented in Fig. 7. A general treatment is pro- The concentrations of the organic surface species (A* , M* , AM* )
vided in the forthcoming sections, incorporating both racemic and are obtained from the quasi-equlibria and taking Eq. (35) into
enantioselective hydrogenation steps. account the result becomes
KA cA c0
3.4.1. Generation rates cA∗ = (36)
D
From the reaction scheme displayed in Fig. 7, the generation
KM cM c0
rates of the organic components can be written as follows; cM ∗ = (37)
D
rA = −(Rmod + Runmod ) (21) KAM KM cA cM c0
cAM ∗ = (38)
rB = Rmod + 0.5Runmod (22) D
where the adsorption term is described by
rC = 0.5Runmod (23)
D = 1 + KA cA + KM cM + KAM KM cA cM (39)
The rate of hydrogen consumption is given by
The quasi-equilibrium hypothesis for hydrogen leads to the expres-
rH2 = −(Rmod + Runmod ) (24)
sion
The detailed reaction mechanism will be considered in next section. ∗  n
(cH )
2 /n
KH2 = (40)
3.4.2. Surface reaction steps cH2 c∗n
The following assumptions were taken for the surface reaction The total balance for hydrogen adsorption site is
mechanism. The organic reactant and modifier adsorb competi-
∗ 
tively on same surface sites (*), but the reactant adsorbs also on c∗ + cH = c0 (41)
2 /n
modified site (M* ) originating the enantiodifferentation. Adsorp-
tion of reaction products is neglected and adsorption of the organics Combination of Eqs. (40) and (41) gives c*
and hydrogen is presumed to be essentially of non-competitive c0
nature, because of the size difference of hydrogen and the organic c∗ = (42)
1 + (KH2 cH2 )1/n
molecules involved. Furthermore, the surface hydrogenation steps
are assumed to be rate-limiting, while the adsorption steps are and finally the concentration of adsorbed hydrogen,
rapid and the quasi-equilibrium hypothesis can be applied on them.
The adsorption steps of the organics are thus written as ∗ (KH2 cH2 )1/n c0
cH = (43)
2 /n D
A + ∗ ↔ A∗ (25)
where D is equal to
M + ∗ ↔ M∗ (26)
D = 1 + (KH2 + cH2 )1/n (44)
A + M ∗ ↔ AM ∗ (27)
Now the rates of racemic (unmod) and enantioselective (mod)
AM ∗ + 2H ∗ hydrogen (ads) → B + 2∗ + M ∗ (28) hydrogenation are obtained from the surface reaction steps accord-
A∗ + 2H ∗ hydrogen (ads) → 0.5B + 0.5C + 2∗ + ∗ (29) ingly,

where * denotes a vacant site for organic adsorption. Hydrogen kunmod c0 c0 KA cA (KH2 cH2 )1/n
Runmod = (45)
adsorption is a controversial issue, since both molecular and atomic DD
adsorption is principally possible. In general, hydrogen adsorption
kmod c0 c0 KAM KM cA cM (KH2 cH2 )1/n
step can be written as Rmod = (46)
DD
∗
H2 + n ≤ nH2/n∗ (30)
The following merged parameters are introduced,
where n = 1 corresponds to non-dissociative and n = 2 to dissociative  1/n
kunmod = kunmod c0 c0 KA KH (47)
adsorption and * denotes the vacant site for hydrogen adsorption. 2
B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845 843

Fig. 8. Ethyl pyruvate hydrogenation over Pt/Al2 O3 catalyst at 7 bar of H2 —experiments ( ) and model fit (—).
844 B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845

Table 2
The results of parameter estimations. Ethyl pyruvate hydrogenation over Pt/Al2 O3 catalyst (CH: conventional heating, CD: cinchonidine).

Parameter

kmod kunmod

CH (36 ◦ C), ethanol + CD 0.356 × 10−4 ± 0.350 × 10−5 0.975 × 10−3 ± 0.129 × 10−3
300 W (36 ◦ C), ethanol + CD 0.491 × 10−3 ± 0.133 × 10−3 0.372 × 10−3 ± 0.680 × 10−4
CH (23 ◦ C), toluene + CD 0.815 × 10−2 ± 0.108 × 10−2 0.167 × 10−2 ± 0.232 × 10−3
300 W (23 ◦ C), toluene + CD+ 0.116 × 10−2 ± 0.230 × 10−3 0.232 × 10−3 ± 0.728 × 10−4
100 W (23 ◦ C), toluene – 0.159 × 10−3 ± 0.115 × 10−4
300 W (23 ◦ C), toluene – 0.849 × 10−4 ± 0.162 × 10−5
700–500 W (28 ◦ C), toluene – 0.662 × 10−4 ± 0.386 × 10−5
500 W (28 ◦ C), toluene – 0.744 × 10−4 ± 0.534 × 10−5

 1/n
kmod = kmod c0 c0 KAM KM KH (48) stants kmod and kunmod for different experimental conditions are
2
listed in Table 2.
Now the rate expression can be written in a compact form
1/n
4. Conclusions

kunmod cA cH
Runmod = 2
(49)
 c c )(1 + (K c )1/n )
(1 + KA cA + KM cM + KAM Enantioselective and racemic hydrogenation of ethyl pyruvate
A M H2 H2
on the Pt/Al2 O3 catalyst was studied under microwave irradiation.
 1/n The effect of microwave transparent and absorbing solvents on
kmod cA cM cH
Rmod = 2
(50) the reaction rate and the ee was investigated. In case of toluene,
 c c )(1 + (K c )1/n )
(1 + KA cA + KM cM + KAM A M H2 H2 which is microwave transparent, no differences in the reaction rate
where KAM = KAM KM . and enantioselectivity were observed between dielectric and con-
The derived model is valid for both cases, i.e. in the presence ventional heating. In ethanol, the ee dramatically decreased in the
and absence of the modifier: in the absence of modifier, cM = 0 presence of microwaves. However, the reaction rate remained unaf-
and rmod = 0 in the above equations. The role of hydrogen was not fected by the mode of heating.
thoroughly investigated in the present study. Furthermore, the con- The catalyst temperature during the course of the reaction was
centration of dissolved hydrogen can be replaced by its partial roughly estimated with the aid of a mathematical model for heat
pressure in the gas phase, since Henry’s law is valid for sparingly transfer in a packed bed and was found to be only 4.0 ◦ C above the
soluble gases. Thus the term in Eq. (50) bulk liquid. This observation was confirmed by solving the energy
balances for solid and liquid phases. Possible explanation for obser-
1/n vations such a low temperature gradients between the catalyst bed
cH
2
and liquid bulk temperature is the inability of the catalyst to effi-
1 + (KH2 cH2 )1/n ciently absorb microwave heating. This can be caused by the fact
was replaced by simple expression p˛ , where ˛ is an exponent that catalyst surface is coated by metal particles (Pt) and tends to
H2
reflect microwave irradiation. The catalyst inability to efficiently
consistent with experimental data.
couple with microwave irradiation might explain that no differ-
ences in the reaction rate and enantioselectivity for microwave and
3.5. Estimation of kinetic parameters conventional heating were observed.
The kinetic model based on selective and unselective reac-
The following objective function was minimized during the tion routes in the hydrogenation of the reactant was fitted to the
parameter estimation, experimental data. This model gave a sufficient description of the
 experimentally observed kinetics, including both the effects of dif-
Q = (ci,t exp − ci,t model )2 (51)
ferent solvents and different modes of heating.
i i

where ci,t exp is the experimentally recorded concentration of com- Acknowledgements


pound i at the reaction time t and ci,t model is the concentration
predicted from the model. The reaction rates are calculated accord- This work is part of the activities at the Åbo Akademi Process
ing to the solid catalyst temperature TS. The reactor models were Chemistry Centre (PCC) within the Finnish Centre of Excellence
solved numerically using the software MODEST [29]. The objec- Programmes (2005–2010) by the Academy of Finland. Financial
tive function was minimized by using the Levenberg–Marquardt support from Finnish Graduate School in Chemical Engineering
method [36]. The results from the model simulations along with (GSCE) is gratefully acknowledged.
experimental data are presented in Fig. 8. The figures show that
the model is able to describe the kinetics of hydrogenation as well Appendix A. Nomenclature
as the distribution of enantiomers in different solvents, under dif-
ferent modes of heating and different microwave power inputs.
Our model also predicts the decrease of ee carried out in the pres- a outer surface area-to-volume ratio
ence of microwaves compared to conventionally heated condition A catalyst outer surface area
(Fig. 8). The decline of the ee under the influence of electromagnetic A reactor cross section
irradiation is probably caused by solvent (ethanol) superheating in c concentration
the cavity. The parameter estimation was performed for all experi- cP , cPV heat capacities at constant pressure (P) and volume (V)
ments. The values of adsorption constants were estimated through CD (−)-cinchonidine (catalyst modifier)
numerical data fitting, however, the terms which include them D adsorption term in rate expression
were found to be negligible in comparison with unity. The con- h heat transfer coefficient
B. Toukoniitty et al. / Chemical Engineering and Processing 48 (2009) 837–845 845

k rate constant [11] W.L. Holstein, M. Boudart, The temperature difference between a supported
K equilibrium constant catalyst particle and its support during exothermic and endothermic catalytic
reactions, Lat. Am. J. Chem. Eng. Appl. Chem. 13 (1983) 107–119.
l length coordinate [12] J.R. Thomas, J.F. Faucher, Thermal modelling of microwave heated fluidized and
L catalyst bed length packed bed catalytic reactors, J. Microwave Power EE 35 (2000) 165–175.
ṁ mass flow rate [13] X. Zhang, D.O. Hayward, D.M. Mingos, Effects of microwave dielectric heating
on heterogeneous catalysis, Catal. Lett. 88 (2003) 33–38.
m mass increment [14] F. Langa, P. De La Cruz, A. De La Hoz, A. Diaz-Ortiz, E. Diez-Barra, Microwave
n number of adsorption sites irradiation: more than just method for accelerating reactions, Contemp. Org.
qMW microwave effect-to-reactor volume Synth. 4 (5) (1997) 373–386.
[15] I. Almena, A. Diaz-Ortiz, E. Diez-Barra, A. De La Hoz, A. Loupy, Solvent-free ben-
Q̇MW microwave effect zylations of 2-pyridone. Regiospecific N- or C-alkylation, Chem. Lett. 5 (1996)
r generation rate 333–334.
R reaction rate [16] D.Yu. Murzin, P. Mäki-Arvela, E. Toukoniitty, T. Salmi, Asymmetric hetero-
geneous catalysis: science and engineering, Catal. Rev.: Sci. Eng. 47 (2005)
t time increment
175–256.
T temperature [17] M. Studer, H.-U. Blaser, C. Exner, Enantioselective hydrogenation using hetero-
V reactor volume geneous modified catalysts: an update, Adv. Synth. Catal. 345 (2003) 45–65.
V̇ volumetric flow rate [18] T. Bürgi, A. Baiker, Heterogeneous enantioselective hydrogenation over cin-
chona alkaloid modified platinum: mechanistic insights into a complex
w velocity reaction, Acc. Chem. Res. 37 (2004) 909–917.
[19] H.-U. Blaser, H.-P. Jalett, M. Garland, M. Studer, H. Thies, A. Wirth-Tijani, Kinetic
studies of the enantioselective hydrogenation of ethyl pyruvate catalyzed by a
Greek letters cinchona modified Pt/Al2 O3 catalyst, J. Catal. 174 (1998) 282–294.
* adsorption site [20] J. Wang, C. LeBlond, C. Orella, Y. Sun, J. Bradley, D. Blackmond, in: H.-U. Blaser,
˛ exponent constant A. Baiker, R. Prins (Eds.), Heterogeneous Catalysis and Fine Chemicals, vol. IV,
Elsevier, Amsterdam, Stud. Surf. Sci. Catal. 108, 1997, p. 183.
εL liquid hold-up
[21] J. Margitfalvi, M. Hegedüs, E. Tfirst, Enantioselective hydrogenation of
eff effectiveness factor ␣-keto esters over cinchona-Pt/Al2 O3 catalyst. Kinetic evidence for the
 stoichiometric coefficient substrate–modifier interaction in the liquid phase, Tetrahedron: Asymmetry
 density 7 (1996) 571–580.
[22] E. Toukoniitty, P. Mäki-Arvela, J. Wärnå, T. Salmi, Modeling of the enantioselec-
 residence time tive hydrogenation of 1-phenyl-1,2-propanedione over Pt/Al2 O3 catalyst, Catal.
Today 66 (2001) 411–417.
[23] M. Sundberg, P.O. Risman, P.-S. Kildal, T. Ohlson, Analysis and design of indus-
Subscripts and superscripts
trial microwave ovens using the finite difference time domain method, J.
BL bulk liquid Microwave Power EE 31 (3) (1996) 142–157.
BS bulk solid [24] A. Parini, G. Bellanca, L. Guasina, Possible solutions to problems in simulating
huge microwave applicators, in: Proceedings of the 10th International Confer-
i component index
ence Microwave and High Frequency Heating, Modena, Italy, 12–15 September,
j reaction step 2005, pp. 432–435.
L liquid [25] G. Roussy, S. Hilaire, J.M. Thiébaut, G. Maire, F. Garin, S. Ringler, Perma-
S solid nent change of catalytic properties induced by microwave activation on 0.3%
Pt/Al2 O3 (EuroPt-3) and 0.3% Pt–0.3% Re/Al2 O3 (EuroPt-4), Appl. Catal. A: Gen.
156 (1997) 167–180.
References [26] S. Ringler, P. Girard, G. Maire, S. Hilaire, G. Roussy, F. Garin, Mechanistic studies
of NOx reduction reactions under oxidative atmosphere on alumina supported
0.2 wt% platinum catalyst treated under microwave (Part II), Appl. Catal. B:
[1] M. Hájek, Microwave catalysis in organic synthesis, in: A. Loupy (Ed.),
Environ. 20 (1999) 219–233.
Microwaves in Organic Synthesis, Wiley-VCH Verlag GmbH & Co. KGaA, Wein-
[27] G. Pipus, I. Plazl, T. Koloini, Esterification of benzoic acid in microwave turbular
heim, 2002, p. 345.
flow reactor, Chem. Eng. J. 76 (2000) 239–245.
[2] B. Toukoniitty, J.-P. Mikkola, D.Yu. Murzin, T. Salmi, Utilization of electromag-
[28] B. Toukoniitty, O. Roche, J.-P. Mikkola, E. Toukoniitty, F. Klingstedt, K. Eränen, T.
netic and acoustic irradiation in enhancing heterogeneous catalytic reactions
Salmi, D.Yu. Murzin, Ethyl pyruvate hydrogenation under microwave irradia-
(a review), Appl. Catal. A 279 (2005) 1–22.
tion, Chem. Eng. J. 126 (2–3) (2007) 126–132.
[3] B. Toukoniitty, J.-P. Mikkola, K. Eränen, T. Salmi, D.Yu. Murzin, Esterification of
[29] H. Haario, MODEST User’s Manual, Profmath Oy, Helsinki, 1994.
propionic acid under microwave irradiation over an ion exchange resin, Catal.
[30] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, John Wiley & Sons,
Today 100 (2005) 431–435.
Inc., New York, 1960.
[4] R. Laurent, A. Laporterie, J. Dubac, J. Berlan, S. Lefeuvre, M. Audhuy, Specific
[31] W.L. Perry, D.W. Cooke, J.D. Katz, Adsorbed species and reaction rates for
activation by microwaves: myth or reality? J. Org. Chem. 57 (1992) 7099–7102.
NO–CO–O2 over Rh(1 1 1), Catal. Lett. 47 (1) (1997) 5–15.
[5] A. Wali, S.M. Pillai, S. Satish, Heterogeneous Pd catalysts and microwave irradi-
[32] P. Fogg, W. Gerrad, Solubility of Gases in Liquids, Wiley, New York, 1991.
ation in heck arylation, React. Kinetic. Catal. Lett. 60 (1997) 189–194.
[33] R.L. Augustine, S.K. Tanielyan, L.K. Doyle, Enantioselective heterogeneous catal-
[6] I. Plazl, S. Leskovšek, T. Koloini, Hydrolysis of sucrose by conventional and
ysis I. A working model for the catalyst modifier substrate interactions in chiral
microwave heating in stirred tank reactor, Chem. Eng. J. 59 (1995) 253–257.
pyruvate hydrogenations, Tetrahedron: Asymmetry 4 (1993) 1803–1827.
[7] J.R. Thomas, Particle size effect in microwave-enhanced catalysis, Catal. Lett.
[34] I.M. Sutherland, A. Ibbotson, R.B. Moyes, P.B. Wells, Enantioselective hydro-
49 (1997) 137–141.
genation I. Surface conditions during methyl pyruvate hydrogenation catalyzed
[8] F. Chemat, D.C. Esveld, M. Poux, J.L. di Martino, The role of selective heat-
by cinchonidine-modified platinum/silica (EUROPT-1), J. Catal. 125 (1990) 77–
ing in the microwave activation of heterogeneous catalysis reactions using a
88.
continuous microwave reactor, J. Microwave Power EE 33 (1998) 88–94.
[35] J. Hájek, D.Yu. Murzin, Liquid-phase hydrogenation of cinnamaldehyde over a
[9] B.H. Bartley, Dual function catalysis with the metal at a higher temperature
Ru-Sn sol–gel catalyst. 1. Evaluation of mass transfer via combined experimen-
than the support, J. Catal. 40 (1975) 86–93.
tal theoretical approach, Ind. Eng. Chem. Res. 43 (2004) 2030–2038.
[10] X.L. Zhang, D.O. Hayward, D.M.P. Mingos, Apparent equilibrium shifts and hot-
[36] D.W. Marquardt, An algorithm for least squares estimation on nonlinear param-
spot formation for catalytic reactions induced by microwave dielectric heating,
eters, J. Soc. Ind. Appl. Math. 11 (2) (1963) 431–441.
Chem. Commun. (1999) 975–976.

You might also like