Вы находитесь на странице: 1из 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/244108095

Simplified kinetic models of methanol oxidation on silver

Article  in  Applied Catalysis A General · August 2005


DOI: 10.1016/j.apcata.2005.05.004

CITATIONS READS

24 354

4 authors, including:

Anders Andreasen
Ramboll Oil & Gas
43 PUBLICATIONS   967 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimization Methods for process facilities View project

All content following this page was uploaded by Anders Andreasen on 26 October 2017.

The user has requested enhancement of the downloaded file.


Risø National Laboratory

Postprint

Materials Research Department


Year 2005
Paper www.risoe.dk/rispubl/art/2006_20.pdf
dx.doi.org/10.1016/j.apcata.2005.05.004

Simplifed kinetic models of methanol oxidation on silver

A. Andreasen, H. Lynggaard, C. Stegelmann, and P. Stoltze

Department of Chemistry and Applied Engineering Science, Aalborg


University, 6700 Esbjerg, Denmark
Materials Research Department, Risø National Laboratory, DK-4000
Roskilde, Denmark
Interdisciplinary Research Center for Catalysis, Department of Chemical
Engineering
Technical University of Denmark, DK-2800 Lyngby, Denmark

Required publisher statement


© copyright (2005) Elsevier
Simplified kinetic models of methanol oxidation on silver

A. Andreasena,b,∗, H. Lynggaard, C. Stegelmann†, and P. Stoltze

Department of Chemistry and Applied Engineering Science, Aalborg University, 6700 Esbjerg, Denmark

a Materials Research Department, Risø National Laboratory, DK-4000 Roskilde, Denmark

b Interdisciplinary Research Center for Catalysis, Department of Chemical Engineering

Technical University of Denmark, DK-2800 Lyngby, Denmark

Abstract

Recently the authors developed a microkinetic model of methanol oxidation on silver (Andreasen
et al. Surf. Sci. 544 (2003) 5-23 ). The model successfully explain both surface science experiments
and kinetic experiments at industrial conditions applying physically realistic parameters. Unfortu-
nately the rate expression based on this microkinetic model is complex and impractical to apply for
reactor engineering purposes. In this paper the rate expression is simplified by a number of approx-
imations to make it suitable for practical applications without loosing significant accuracy. This is
achieved by converting from a statistical thermodynamic description to a classical thermodynamic
description resulting in a reduction of model parameters from more than 100 to 26. Further, other
assumptions are applied including the Most Abundant Reaction Intermediate (MARI) approximation
and the Irreversible Step (IS) approximation. This leads to further reduction in model complexity
and in number of parameters. Computations verify that the performance of the model is preserved
despite the reduction of the vast amount of parameters.

Keywords: Methanol oxidation, Formaldehyde synthesis, Microkinetic modeling, Catalysis, Silver,


Quasi-Equilibrium approximation, MARI approximation, IS approximation.

∗ Corresponding author.
Email address: anders.andreasen@risoe.dk (A. Andreasen).
Telephone: +45 46 77 57 76 (Office).
Fax: +45 46 77 57 58 (Department).
† Present address: Rambøll Oil & Gas, Willemoesgade 2, DK-6700 Esbjerg, Denmark

1
1 Introduction

The partial oxidation of methanol to formaldehyde on silver is an important industrial process due to
the versatility of formaldehyde as an intermediate in chemical synthesis [1]. The process is carried out
at ∼900 K and atmospheric pressure, the feed consist of a fuel-rich mixture of methanol and air (BASF
process). At typical reaction conditions the selectivity towards formaldehyde is approximately 90 % and
the conversion of oxygen approach 100 % [1, 2]. Steam is added to increase selectivity and heat transport
[1, 3, 4].
Recently the authors developed a microkinetic model describing both oxidation of methanol to
formaldehyde and the further oxidation of formaldehyde to CO2 [5]. The microkinetic model is based
on the mechanism depicted in Table 1 derived by Wachs and Madix [6] from various surface science
experiments.
The mechanism is very simple compared to other mechanisms suggested in the literature: it is a
Langmuir-Hinshelwood mechanism, only one type of atomic oxygen is active in the oxidations and selec-
tivity is described by a simple consecutive reaction path. Still the microkinetic model is able to explain
both surface science experiments and kinetic experiments close to industrial reaction conditions applying
physically realistic model parameters. The model successfully explains experimentally observed reaction
orders, selectivity as a function of conversion and temperature, apparent activation energy of methanol
oxidation as a function of temperature, and the choice of industrial reaction conditions [5, 7]. See ref.
[2, 8, 9, 10, 11, 12] for experimental observations. For detailed discussion of the model in relation to other
suggested mechanisms in the literature we refer to [5, 13] and references therein.
Unfortunately the model is very complex and contains more than 100 parameters due to a detailed
statistical thermodynamic description of gases and intermediates in the model. This does not prompt the
use of the model in practical reactor engineering such as describing macrokinetic data, detailed reactor
simulations involving physical transport processes, non-ideal flow patterns, reaction heats, advanced
reactor configurations etc. However it is possible to simplify the microkinetic model without loosing
information because most of the parameters are uncritical.
In this paper we derive a number of rate expression to the full microkinetic model and compare the
outcome of these models to the full quasi equilibrium solution. The first rate expression (RATE I) corre-
sponds to the full quasi equilibrium solution except that the microscopic thermodynamic description using
statistical thermodynamics is replaced by a macroscopic thermodynamic description using enthalpies and
entropies for each elementary step thus leading to a sixfold reduction in the number of model parameters.
In the second rate expression (RATE II) the Most Abundant Reaction Intermediate (MARI) approxi-

2
Table 1: Reaction mechanism used for the microkinetic model. The * signifies a surface site and X∗ is
an adsorbed species.

CH3 OH(g) + ∗ *
) CH3 OH∗ (step 1)

O2 (g) + ∗ *
) O2 ∗ (step 2)

O2 ∗ +∗ *
) 2O∗ (step 3)

2CH3 OH ∗ +O∗ *
) 2CH3 O ∗ +H2 O∗ (step 4)

CH3 O ∗ +∗ *
) H2 CO ∗ +H ∗ (slow) (step 5)

H2 CO∗ *
) H2 CO(g) + ∗ (step 6)

2H∗ *
) H2 (g) + 2∗ (step 7)

H2 O∗ *
) H2 O(g) + ∗ (step 8)

H2 CO ∗ +O∗ *
) H ∗ +HCOO∗ (step 9)

HCOO ∗ +∗ *
) H ∗ +CO2 ∗ (slow) (step 10)

CO2 ∗ *
) CO2 (g) + ∗ (step 11)

mation is applied. The third rate expression (RATE III) uses the irreversible step approximation (IS).
Finally we make the assumption that the coverage of free sites is practically one and end up with a
power-law type of expression (RATE IV).
Parameters are supplied for all the rate expression, the assumptions behind these parameters are
discussed in our original paper [5].

2 Kinetic models

In this section we deduce different kinetic expressions based on the mechanism, cf. Table 1. However
first the mathematics and assumptions of the full quasi equilibrium solution is reviewed.

2.1 Full microkinetic model

The microkinetic model is based on the mechanism deduced from the UHV work of Wachs and Madix
on Ag(110) [6] depicted in Table 1. From Table 1 the two following overall gas phase reaction may be

3
Table 2: Rate and equilibrium equations for the kinetic model. Ki is the equilibrium constant for step i,
ki is the rate constant, ri is the reaction rate of the rate limiting step i, θX∗ is the coverage of species X∗,
pi is the partial pressure of i and p is the thermodynamic reference pressure. All equilibrium constants
are dimensionless and rate constants are given in s−1 .
pCH3 OH
θCH3 OH∗ = K1 p θ∗
pO 2
θ O2 ∗ = K2 p θ ∗
1 1
  21
pO 2
θO∗ = K22 K32 p θ∗
1 1 1 1
  41  − 12
pO 2 pCH3 OH pH 2 O
θCH3 O∗ = K1 K24 K34 K42 K82 p p p θ∗
k5
r5 = k5 θCH3 O∗ θ∗ − K5 θH2 CO∗ θH∗

1 pH2 CO
θH2 CO∗ = K 6 p θ ∗
  12
1 pH 2
θH∗ = K7 − 2 p θ∗
1 pH 2 O
θH2 O∗ = K 8 p θ ∗
1 1 1
  21  − 12
pO 2 pH2 CO pH 2
θHCOO∗ = K22 K32 K6−1 K72 K9 p · p p θ∗
k10
r10 = k10 θHCOO∗ θ∗ − K10 θCO2 ∗ θH∗

1 pCO2
θCO2 ∗ = K11 p θ∗

θ∗ = 1 − θCH3 OH∗ − θO2 ∗ − θO∗ −θCH3 O∗ − θH2 CO∗ − θH∗

−θHCOO∗ − θCO2 ∗ − θH2 O∗

derived:
1 1 1
CH3 OH + O2 *
) H2 CO + H2 + H2 O (1)
4 2 2

1
H2 CO + O2 *
) CO2 + H2 (2)
2
i.e. the process can be divided into the oxidation of methanol to formaldehyde and the further oxida-
tion of formed formaldehyde to carbon dioxide. The mechanism is analyzed as a Langmuir-Hinshelwood
mechanism applying the Quasi-Equilibrium Approximation assuming that step 5 and step 10 are rate
limiting steps. From these assumptions rates and coverages of surface species may be derived, Table 2.
The equilibrium constants in the original model is described by statistical thermodynamics resulting in
more than 149 parameters and the rate constants of two rate limiting steps (step 5 and 10) are described

4
by Arrhenius expressions resulting in further 4 parameters. All of these parameters are assigned physically
realistic parameters and most of them have been deduced from surface science experiments. However a
few critical parameters have been fitted to kinetic experiments.
Even though the microkinetic model is very detailed and complex it still rely on a number of approx-
imations and simplifications. The full microkinetic model is thus not an exact model and may need to be
improved in order to explain future experiments. However the model is the most simple and consistent
model available in the literature today.
From Table 2 we may deduce the rate laws for methanol (r5 ) and formaldehyde oxidation (r10 )
respectively, resulting in:

  41  − 12
pCH3 OH p O2 p H2 O
r5 = k 5 KA (1 − βI ) θ∗2 (3)
p p p
 − 12  1
pH2 CO p H2 p O2 2
r10 = k10 KB (1 − βII )θ∗2 (4)
p p p

where
   12   12
pH2 CO pH 2 pH 2 O
p p p
βI =    41 (5)
pCH3 OH pO 2
p p Kg,I
1 1 1
KA = K1 (K2 K3 ) 4 K42 K82 (6)
1 1 1 1
Kg,I = K1 (K2 K3 ) K4 K5 K6 K7 K8
4 2 2 2
(7)
  
pCO2 pH 2
p p
βII =    21 (8)
pH2 CO pO 2
p p Kg,II
p
KB = K2 K3 K7 K6−1 K9 (9)
p
Kg,II = K2 K3 K6−1 K7 K9 K10 K11 (10)
  21
pCH3 OH p O2 1 1 p O2
θ∗−1 = 1 + K1 + K 2 + K 2
2 K 3
2

p p p
  41  − 12
1 1 1 1 p O2 pCH3 OH pH2 O
+K1 K24 K34 K42 K82
p p p
 1
1 pH2 CO − 12 p H2 2 1 p H2 O
+ + K 7 +
K6 p p K8 p
 1  − 21
1 1
−1
1 pO2 2 pH2 CO pH2
+K2 K3 K6 K7 K9
2 2 2

p p p
1 pCO2
+ (11)
K11 p

5
β is the approach to equilibrium defined as the reaction quotient divided by the overall equilibrium
constant Kg . The rate of methanol oxidation is r5 , the rate of formaldehyde oxidation is r10 and the
r5 −r10
selectivity towards formaldehyde is r5 .

The unit of reaction rate derived in this paper is turnover frequency i.e. the number of reactions
per site per second. Hence in order to apply the models the site density of the silver catalyst has to be
estimated. In the microkinetic model it is assumed that each active site consist of two Ag surface atoms
on the Ag(111) surface i.e. a site density of 6.9·1018 sites/m2 [5, 14]. Another definition of active sites
means that the prefactors of steps 5 and 10 will change as they are inverse proportional.

2.2 Change of thermodynamic description (RATE I)

In the first rate expression the statistical thermodynamics of gases and intermediates i.e. translational,
rotational and vibrational degrees of freedom is substituted with a macroscopic thermodynamic descrip-
tion. The rate is still described by the full quasi equilibrium description as the original model. This is
performed in order to check the influence of the thermodynamic description on predicted rate of reaction.
The macroscopic thermodynamic description consists of enthalpy ∆Hr and entropy ∆Sr changes for each
elementary reaction step r. The enthalpy of individual gas phase molecules and adsorbates is found from
the partition function Q

 
d ln Qi
Hi = RT 2 (12)
dT P

The equilibrium constant of the elementary reaction r is given by:


Y
Kr = Qνi i (13)
i

where νi is the stoichiometric coefficient of species i. The equilibrium constant may also be expressed
using the macroscopic properties

   
∆Sr ∆Hr
Kr = exp exp − (14)
R RT
Thus ∆Sr may be found by rearranging Eq. 14. Values of ∆Sr and ∆Hr is calculated at T=750 K
and the resulting parameters are summarized in Table 3. This leads to a reduction of 149 thermodynamic
parameters in the original description to 22.
As outlined in the previous section the overall gas phase equilibrium constants of methanol oxidation
(Eq. 1) and formaldehyde oxidation (Eq. 2) may be calculated from the equilibrium constants of the
individual reaction steps. Using the parameters given in Table 3 we get a value of K g,I = 3.57 · 109

6
Table 3: Thermodynamic parameters expressed in terms of ∆Hr and ∆Sr for the calculation of equilib-
rium constants Kr for elementary reaction step r. Parameters are calculated from statistical thermody-
namics at T=750 K using the parameters given in ref. [5].

Step r ∆Hr [kJ/mol] ∆Sr [J/(mol K)]

1 -32.955 -82.317

2 -37.735 -120.50

3 -83.069 -44.969

4 42.843 -89.129

5 -52.865 71.413

6 19.324 75.059

7 51.250 132.15

8 47.478 103.27

9 -26.153 -78.897

10 -238.98 92.782

11 15.082 51.745

Table 4: Arrhenius parameters for the two slow steps. A is the prefactor and H ‡ is the activation enthalpy.
Parameters are from ref. [5].

Reaction, i Ai Hi‡

5 4.2 · 1010 s−1 60 kJ mol−1

10 9.8 · 109 s−1 77 kJ mol−1

and Kg,II = 8.16 · 1050 at 298 K. Using tabulated values of molecular standard formation enthalpies and
standard formation entropies [15] we get Kg,I = 3.62 · 109 and Kg,II = 1.40 · 1051 at 298 K. Thus, the
thermodynamic parameters used in the microkinetic model gives a satisfactory description of the overall
gas phase equilibrium. Arrhenius parameters for step 5 and 10 are depicted in Table 4.

7
Table 5: Lumped parameters for RATE II, III and IV.

Equilibrium constants Thermodynamic parameters

Lumped Origin ∆S [J/(mol K)] ∆H [kJ/mol]


1 1 1
KA K1 (K2 K3 ) 4 K42 K82 -117 -18

KB K2 K3 K7 K6−1 K9 -171 -80

KC K2 K3 -165 -121
1 1 1 1
Kg,I K1 (K2 K3 ) 4 K42 K5 K6 K7 2 K82 96 -26

Kg,II K2 K3 K6−1 K7 K9 K10 K11 40 -278

2.3 MARI description (RATE II)

It was previously found that O* is the most abundant reaction intermediate (MARI) at all temperatures
[5]. The number of free sites increases with the decrease in O* coverage. Besides O* all intermediates
have a very low coverage except formate that may have a significant coverage at low temperatures. Thus
it may be reasonable to apply the MARI approximation above the temperatures where formate has any
influence:

θ∗ = 1 − θO∗ (15)
1
=  12 (16)
1 1

pO 2
1 + K2 K32 2
p

The rate expressions for step 5 and 10 are then simplified to:

  14  − 21
pO 2 pCH3 OH pH 2 O
k 5 KA p p p (1 − βI )
r5 =   21 2 (17)
1

pO 2
1 + KC 2
p
  21  − 12
pO 2 pH2 CO pH 2
k10 KB p p p (1 − βII )
r10 =   1 2 (18)
1

pO 2 2
1 + KC 2
p

(19)

8
where the values of lumped constants KA , KB and KC are given in Table 5. We note that no
information about r5 or r10 is lost by this lumping and the lumped equilibrium constants KA , KB , and
KC still contain a physical meaning. KA is the equilibrium constant for the formation of CH3 O∗ and
water from methanol and oxygen

1 1
CH3 OH(g) + O2 (g) + ∗ *
) CH3 O ∗ + H2 O(g) (20)
4 2

KB is the equilibrium constant for the formation of HCOO∗ species and hydrogen from formaldehyde
and oxygen
1 1
H2 CO(g) + O2 (g) + ∗ *
) HCOO ∗ + H2 (g) (21)
2 2
Thus the equilibrium constants KA and KB describes the formation of the reactant species of the two
rate limiting steps. Similarly KC describes the dissociation of oxygen.

O2 (g) + 2∗ *
) 2O∗ (22)

The implication of the above description is that the microkinetic model is insensitive to the choice
of many parameters. For instance the stability of methanol (enthalpy and entropy of adsorption) can
be chosen arbitrary, as long as the resulting methanol coverage remains low, since it cancels out in the
lumping procedure. This also applies to formaldehyde, hydrogen and water.
It is believed by many that the MARI always participates in the rate limiting step. Even though this
may be true in many cases this specific case demonstrates that this is not true in general. It should be
stressed that there is nothing exotic about the model of this paper to result in this departure from a rule
of thumb.

2.4 Irreversible step model (RATE III)

To reduce the MARI model further we note that the values of Kg,I and Kg,II are very large and it
would therefore seem reasonable to view methanol and formaldehyde oxidation as irreversible. Thus the
irreversible step approximation (IS) is applied by setting βI and βII equal to zero. The rate expression
for step 5 and 10 then becomes:

  41  − 12
pO 2 pCH3 OH pH 2 O
k 5 KA p p p
r5 =  1 2
 (23)
1

pO 2 2
1 + KC 2
p

9
 12  1
pH 2 − 2
 
p pH2 CO
k10 KB pO 2 p p
r10 =   1 2 (24)
1

pO 2 2
1 + K C p
2

where the values of lumped constants are given in Table 5.

2.5 Clean surface limit (RATE IV)

In the limit θ∗ → 1 i.e. where the coverage of surface species is negligible the model can be reduced to a
power law type rate expressions by setting the denominator in Eq. 23 and 24 equal to one:

 1  − 12
pO2 4 pCH3 OH pH2 O
r5 = k 5 KA (25)
p p p
  12  − 12
p O2 pH2 CO pH2
r10 = k10 KB (26)
p p p
In this model the description of changes in surface coverage of MARI is lost. However this description
is expected to be reasonable at industrial conditions where the coverage of free sites is close to unity [5].

3 Results and Discussion

In the following we test the developed rate expression by comparing simulated conversions and selectivity
with the original model. All simulations are made in the steady-state isothermal plug-flow reactor model
with the reaction conditions listed in Table 6.
In Figure 1 conversion and selectivity have been calculated with RATE I and compared to calculations
with the original model. As shown in the figure both models give more or less the same results. Thus, no
significant information is lost when substituting the statistical thermodynamic description with classical
thermodynamics.

Table 6: Operating conditions used in reactor simulations. Total reactor pressure is 1.13 · 10 5 Pa. Inlet
mole fraction of nitrogen xN2 = 1 − xCH3 OH − xO2 . Site density 1.15·10−5 mol/kg. W/F = 1 (kg s)/mol.

Variable Parameter range

xCH3 OH [%] 5, 10, 15, 20, 25

xO2 [%] 2.5, 5, 7.5, 10, 12.5

T [K] 600, 750, 900

10
100

80 Selectivity [%]
Conversion [%]

60
RATE I

40

20

0
0 20 40 60 80 100
Full microkinetic model

Figure 1: RATE I predictions of conversion of methanol (in % ) and selectivity towards formaldehyde (in
%) plotted against predictions by the full microkinetic model.

The same comparison is made with the other three rate expressions. This gives identical results for
RATE II and RATE III as shown in Figure 1 and the figures are therefore not shown. Instead we use
calculated average deviations in conversion of methanol and selectivity towards formaldehyde in order
to compare all models with the full microkinetic model. The deviation in conversion of methanol, X, is
calculated for each temperature and for each rate expression i by

1 X (Xmkm − XRAT E i )
Dev(i)% = · 100% (27)
N Xmkm

where the Xmkm and XRAT Ei is the conversion calculated with the full microkinetic model and the con-
version calculated by rate expression i, respectively. The summation is over all parameter combinations
of methanol and oxygen inlet mole fractions as shown in Table 6 and N refers to the total number of pa-
rameter combinations. In order to calculate the average deviations in selectivity, conversion of methanol
X in the above equation is replaced by the selectivity towards formaldehyde, S. The calculated deviations
for all rate expression are summarized in Table 7
From the table we note that although RATE I and the full microkinetic model give identical results
according to Figure 1 there is a small difference in both calculated conversions and selectivities. This
difference has its origin in the temperature dependency of the thermodynamic parameters which is not

11
reproduced exactly by RATE I. However this difference is insignificant as shown in Figure 1.
By introducing the MARI approximation (RATE II) the average deviation is more or less unaffected.
This means that the MARI approximation is valid and can be introduced without reducing the predictive
capability of the microkinetic model.
Interestingly, introducing the IS approximation (RATE III) we find no increase in calculated average
deviations compared with MARI model. This is quite interesting since it implies that methanol oxidation
on silver is by no means subject to thermodynamical limitations viz. the approach to equilibrium is close
to zero even with conversions approaching 100%.
Applying a power law rate equation (RATE IV) results in a pronounced increase in the average
deviations at 600 and 750 K whereas at 900 K the deviations are more or less unchanged. To visualize
how severe an error the power law model introduces a graphical representation is made, Figure 2. From
the figure it is noted that at 900 K the results are close to the original rate expression. At 750 K the results
are still close to the original model but with a significant scatter. At 600 K there is no agreement between
power law and the full model. The reason for this behavior can be found in the fact that the surface
is virtually free of adsorbates at high temperature which leads to θ∗ ≈ 1. The lower the temperature
the lower coverage of free sites leading to an increased error. The O* coverage is in the range 0.62 to
0.77 for the conditions listed in table 6, which means that the denominator of eq. 23 and 24 is in the
range 6.6-21 as compared to the value of unity assumed in eq. 25 and 26. The model performance at
low temperatures can be improved significantly by applying reaction orders fitted to steady-state kinetic
data at low temperatures. However, the price being reduced performance at high temperatures.
From the comparisons above it is observed that a power law rate equation may be sufficient to explain

Table 7: Deviation (%) in conversions and selectivities for each reduced rate expression at different
temperatures. Conversion, X, and selectivities, S, are listed as X/S pairs.

Temperature

Rate expression 600 K 750 K 900 K

I 2.4/-0.4 2.3/-0.6 -0.9/0.3

II 2.8/-0.4 2.7/-0.6 -0.7/0.3

III 2.8/-0.4 2.7/-0.6 -0.7/0.3

IV 409/-17 19/-3.3 0.7/-0.2

12
100
Conversion T = 600 K
Conversion T = 750 K
80 Conversion T = 900 K
Selectivity T = 600 K
Selectivity T = 750 K
Selectivity T = 900 K
60
RATE IV

40

20

0
0 20 40 60 80 100
Full microkinetic model

Figure 2: RATE IV predictions of conversion (in %) of methanol and selectivity towards formaldehyde
(in %) plotted against predictions by the full microkinetic model.

methanol oxidation over silver at industrial conditions (T=900 K). However a kinetic rate equation based
on the IS approximation is the simplest sensible rate equation giving satisfactory predictions at the
investigated reaction conditions.
To verify that the derived kinetic models still captures the most important experimentally observed
trends the selectivity towards formaldehyde versus conversion and temperature is simulated with the
irreversible step model (RATE III). Selectivity is a very important feature of methanol oxidation on silver
and it has been examined in great detail experimentally. The results are shown in Figure 3. From Figure
3 it is evident that the selectivity towards formaldehyde decrease at increasing conversion/residence time
at all applied temperatures. This is in agreement with experimental observations [8, 12, 16]. Simulations
show that the kinetic model with selectivity limitations due to a consecutive reaction pathway, explains
the experimental observations well. The observed selectivity vs. conversion trend on Figure 3 can be
accounted for by the increasing amount of formaldehyde formed at higher conversions. Increasing the
amount of formaldehyde leads to an increased rate of oxidation of formaldehyde. This is a classical effect
of a consecutive reaction path [17, 18].
Figure 3 also shows that the selectivity towards formaldehyde generally is favored by increasing tem-

13
Selectivity H2CO [%]

100
90
80
70
100 60
90 50
80 40
70 30
60 20
50
40 10
30 900
20 870
10 840
810
780
750
720 Temperature [K]
0 10 690
20 30 660
4050 60
70 630
Conversion CH3OH [%]

Figure 3: Calculated selectivity as function of conversion and temperature. Total reactor pressure is
1.13 · 105 Pa. Inlet mole fractions of methanol and oxygen is xCH3 OH = 0.0982 and xO2 =0.143. Inlet
mole fraction of nitrogen xN2 = 0.7588. Site density 1.15·10−5 mol/kg. W/F = 0.00017 to 1086 (kg
s)/mol.

peratures. This is consistent with the experimental observations of increased selectivity with increasing
temperature [2, 10, 11, 12, 16, 19, 20, 21]. It might seem surprising that the selectivity increase with
temperature despite the fact that the activation enthalpy for step 5 is smaller than the for step 10 (cf.
table 4). However, the selectivity is determined by the apparent activation enthalpies, H iapp , which are
calculated according to the definition

d ln ri,+
Hiapp = RT 2 (28)
dT

where ri,+ is the forward rate of step i. Inserting eq. 23 and 24, respectively into the above equation, we
get the following apparent activation enthalpies

H5app = H5‡ + ∆HA − θO∗ ∆HC (29)

app ‡
H10 = H10 + ∆HB − θO∗ ∆HC (30)

Inserting values for H5‡ , H10



, ∆HA , and ∆HB from table 3 and 4, we find that the apparent activation

14
enthalpy for step 5 is always 45 kJ/mol higher than for step 10. Thus, selectivity will increase with
increasing temperatures due to a higher temperature sensitivity of step 5.
Hence the first three rate expressions as well as the original model are able to capture the important
selectivity trends of methanol oxidation. The models also captures reaction orders, apparent activation
energies etc. [5].

4 Summary

In this paper simplified rate expressions for methanol oxidation on silver has been derived from a mi-
crokinetic model. These rate expressions include a recalculation of the thermodynamics, the MARI and
IS approximations, and approximation of a clean surface. The first three rate expressions capture all
relevant experiments almost as accurate as the original model and captures important kinetic trends
of methanol oxidation and selectivity. Due to their simplicity the models will be practical for reactor
engineering purposes.
At industrial conditions the rate can be described by a power law expression, which is due to a very
low coverage of intermediates. However at lower temperatures the simplest possible model comes from a
combination of the MARI and IS approximations. This indicates that that partial oxidation of methanol
to formaldehyde is controlled by kinetics and not equilibrium.

References
[1] C. N. Satterfield, Heterogeneous Catalysis, 2nd Edition, McGraw-Hill, 1991.

[2] A. Nagy, G. Mestl, Appl. Catal. A 188 (1999) 337–353.

[3] Ullmann’s Encyclopedia of Industrial Chemistry, 5th Edition, Vol. A11, VCH Verlagsgesellschaft, 1988.

[4] M. V. Twigg, Catalyst Handbook, Manson Publishing, 1996.

[5] A. Andreasen, H. Lynggaard, C. Stegelmann, P. Stoltze, Surf. Sci. 544 (2003) 5–23.

[6] I. E. Wachs, R. J. Madix, Surf. Sci. 76 (1978) 531–558.

[7] H. Lynggaard, A. Andreasen, C. Stegelmann, P. Stoltze, Prog. Surf. Sci. 77 (2004) 71–137.

[8] D. A. Robb, P. Harriott, J. Catal. 35 (1974) 176–183.

[9] S. K. Bhattacharyya, N. K. Nag, N. D. Ganguly, J. Catal. 23 (1971) 158–167.

[10] V. N. Gavrilin, B. I. Popov, Kinet. Catal. (USSR) 6 (1965) 799.

[11] A. E. Obraztsov, B. I. Popov, M. P. Shashalevich, L. P. Taraban, E. A. Pronina, A. G. Shevchenko, Kinet.Catal.


(USSR) 12 (1971) 970–.

[12] M. Qian, M. Liauw, G. Emig, Appl. Catal. A 238 (2003) 211–222.

15
[13] I. E. Wachs, Surf. Sci. 544 (2003) 1–4.

[14] C. T. Campbell, J. Catal. 94 (1985) 436–444.

[15] D. R. Lide (Ed.), Handbook of Chemistry and Physics, 78th Edition, CRC Press LLC, 1997.

[16] L. Leffert, J. G. van Ommen, J. R. H. Ross, Appl. Catal. 23 (1986) 385–402.

[17] G. F. Froment, K. B. Bischoff, Chemical Reactor Analysis and Design, 2nd Edition, John Wiley & Sons, 1990.

[18] L. D. Schmidt, The Engineering of Chemical Reactions, Oxford University Press, 1998.

[19] G. I. N. Waterhouse, G. A. Bowmaker, J. B. Metson, Appl. Catal. A 265 (2004) 85–101.

[20] A. Nagy, G. Mestl, T. Rühle, G. Weinberg, R. Schögl, J. Catal. 179 (1998) 548–559.

[21] C.-B. Wang, G. Deo, I. E. Wachs, J. Phys. Chem. B 103 (1999) 5645–5656.

16

View publication stats

Вам также может понравиться