Chemical Engineering Science 221 (2020) 115695

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: www.elsevier.com/locate/ces

Modelling of a microreactor for the partial oxidation of 1-butanol on a

titania supported gold catalyst
Yaseen Khan a,⇑,1, Teuvo Kilpiö b,1, Minna Marin a, Vincenzo Russo c, Juha Lehtonen d, Reetta Karinen a,
Tapio Salmi b
a
Aalto University, School of Chemical Engineering, P.O. Box 16100, FI-00076, Finland
b
Åbo Akademi University, Department of Chemical Engineering, Laboratory of Industrial Chemistry and Reaction Engineering, FI-20500 Turku/Åbo, Finland
c
University of Naples, ‘‘Federico II”. Chemical Sciences Department, IT-80126 Naples, Italy
d
VTT Technical Research Centre of Finland, P.O.Box 1000, FI-02044 VTT, Finland

h i g h l i g h t s

n-Butyraldehyde was oxidized to products on a Au/TiO2 catalyst.

A coated-wall microreactor was used in the experiments.

The reactant conversion and the product yields were described with a kinetic model.

The work shows that microreactors are strong tools in the investigation of heterogeneous gas-phase kinetics.

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

Article history: Microreactors can provide more uniform operation conditions than conventional reactors due to the
Received 29 September 2019 characteristic advantages related to mass and heat transfer. In our earlier research, a catalyst testing
Received in revised form 27 January 2020 microreactor was used to study the gas-phase partial oxidation of 1-butanol to n-butyraldehyde and
Accepted 28 March 2020
subsequently a Au/TiO2 catalyst was selected for kinetic experiments and mathematical modelling in this
Available online 2 April 2020
work. Power law and mechanistic kinetic models were applied in parameter estimation using a
1D-pseudo-homogenous plug-flow reactor (PFR) model implemented in Matlab. In addition, a 2D model
Keywords:
including dynamic mass balances for both the gas and solid catalyst phases with power law kinetics was
Partial oxidation
1-butanol
implemented by using the process modelling tool gPROMS. The estimated activation energies and
Microreactor reaction rate constants were in physically meaningful order of magnitude. The activation energy for
Gas phase process the partial oxidation of 1-butanol to n-butyraldehyde was found to be 77 kJ/mol. The predicted reactor
Heterogeneous catalysis performance results from the numerical simulations were in agreement with the experimental observa-
Gold tions. The presented modelling approach can be used with more advanced mechanistic models involving
Titania surface species for the partial oxidation of 1-butanol. The model could in future be extended to optimize
Mathematical modelling the yield of n-butyraldehyde for further process development.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction of such a bio-resource, and it can be obtained, for example, from

acetone-butanol-ethanol (ABE) fermentation (Chen et al., 2014).
In the low carbon world of future, fossil resources are left Especially in regions with abundant biomass resources, bio-1-
underground and carbon-based products such as organic chemi- butanol is a potential source for chemical intermediates and fuel
cals are produced from aboveground carbon resources such as components. Valuable intermediates can be obtained by partial
waste streams from biomass. Bio-based 1-butanol is an example oxidation of 1-butanol in the presence of a solid catalyst. Oxidation
reactions are of industrial importance in the synthesis of various
aldehydes, ketones, and acids from the corresponding primary
⇑ Corresponding author. and secondary alcohols.
E-mail addresses: [email protected] (Y. Khan), [email protected] (T. Kilpiö), Aldehydes, in general, are important intermediates for the syn-
[email protected] (M. Marin), [email protected] (V. Russo), [email protected] thesis of fine chemicals and pharmaceuticals. n-Butyraldehyde is
(J. Lehtonen), [email protected] (R. Karinen), [email protected] (T. Salmi).
1 an important intermediate for the production of chemicals and fuel
Both authors contributed equally to this work.

https://doi.org/10.1016/j.ces.2020.115695
0009-2509/Ó 2020 Elsevier Ltd. All rights reserved.
2 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695

Nomenclature

List of symbols Pe Peclet number [–]

Ac cross sectional area of the channel [m2] Qp objective function [–]
BuOH 1-butanol R inner reactor radius [m]
Buald n-butyraldehyde RW reactor radius [m]
C i;FEED concentration of component i in inlet gas stream r radial location [m]
[mol/m3] rp radial location within the washcoated layer [m]
C i;S local concentration of component i in solid [mol/m3] rj reaction rate of reaction j [mol/(m3∙s)]
C i;G concentration of component i in gas [mol/m3] s residence time [s]
Deff ;i effective diffusivity of component i [m2/s] t time [s]
Dz axial dispersion coefficient [m2/s] uG superficial velocity of the gas [m/s]
Eaj activation energy of reaction j [J/mol] mi;j stoichiometric coefficient of i in reaction j [–]
i component i X1 molar fraction of component i [–]
j reaction j ycal model value of component i [–]
kj kinetic constant of reaction j yexp experimental value of component i [–]
kj,ref kinetic constant of reaction j calculated at 573 K. z axial location [m]
L reactor length [m] ep porosity [–]
n_ i molar flow of the component i [mol/s] qB catalyst bulk density [kg/m3]
P pressure [Pa] n dimensionless washcoat layer [–]
Pbutanol butanol partial pressure [kPa]

components, such as acetal, 2-ethylhexanol, 2-ethylhexanoic acid, 2016). Particularly for rapid heterogeneously catalyzed gas-phase
polyvinyl butyral, and trimethylolpropane (Mascal, 2012). Cur- reactions, microreactors coated with thin catalyst layers (40 mm
rently, butyraldehyde is produced industrially by the hydroformy- or less) serve as reliable and efficient tools for intrinsic kinetic
lation of propene in the presence of homogeneous catalysts, such studies (Russo et al., 2015; Kolb and Hessel, 2004; Görke et al.,
as RhH(CO)(PR3)3, RhH(CO)4, and CoH(CO)4 (Knözinger, 2009). Pro- 2009). Due to rapid heat transfer, isothermal conditions can be
ducing n-butyraldehyde from bio-based 1-butanol by partial gas- approached in microreactors (Walter et al., 2005). Often, kinetic
phase oxidation provides another alternative bio-based reaction modelling applying microreactors is performed assuming laminar
route for this widely used component. or plug flow conditions using a pseudo-homogenoeus phase mod-
The molar alcohol-to-oxygen ratio in the feed stream is an elling approach (Kolb and Hessel, 2004; Walter et al., 2005; Kiwi-
important process parameter to avoid total oxidation and to favour Minsker and Renken, 2005; Schmidt et al., 2015; Khan et al.,
partial oxidation to desired products. The partial oxidation reaction 2015). However, more advanced dynamic 2D models have recently
of 1-butanol to n-butyraldehyde proceeds ideally according to the been discussed and reported, accounting for both the gas and solid
following overall reaction phase events as well as incorporating axial dispersion effect (Russo
et al., 2015).
2C4 H10 O + O2 ! C4 H8 O + 2H2 O ð1Þ
Rapidly running robust and reliable kinetic and reactor models
In a recent study, we investigated monometallic gold (Au) and are highly valuable for composite reaction systems as multiple par-
palladium (Pd) as well as bimetallic Au-Pd nanoparticles supported allel and consecutive reactions proceed simultaneously and the
on titania (TiO2) for the partial oxidation of 1-butanol in gas phase number of adjustable kinetic parameters becomes high. The mod-
and found that the Au nanoparticle catalyst supported on titania elling approach with advanced dynamic 2D models suits particu-
(Au/TiO2) exhibited the best performance for high yield towards larly well to modelling by gPROMS software, because gPROMs
n-butyraldehyde (Khan et al., 2018). The main products for the par- provides the solvers and solution methods (e.g. numerical method
tial oxidation of 1-butanol on the Au/TiO2 catalyst were n- of lines for the solution of partial differential equations and maxi-
butyraldehyde, water, butenes, CO2 as well as propene and CO mum likelihood method for the estimation of kinetic parameters)
from subsequent retro-hydroformylation of n-butyraldehyde. Fur- as rapidly running standard options. A sophisticated, yet classic
thermore, other products formed in trace amounts were butyric way to generate rapidly running, robust, and reliable reactor mod-
acid, n-butylbutyrate, 4-heptanone, dibutyl ether and butane. The els is to simplify the fluid flow description to a minimum. For a
reaction system for the partial oxidation of n-butanol in liquid microreactor application, this can be made by either assuming a
and gas phases has been reported by other authors as well laminar flow profile in the reactor channel, which might well pre-
(Requies et al., 2012; Gandarias et al., 2016; Gandarias et al., vail due to the small hydraulic diameters of the flow channels or
2015; Biella and Rossi, 2003). using dispersion terms combined with a relevant correlation for
A conventional way to carry out heterogeneously catalyzed gas- the pressure drop. Models of varying degree of sophistication for
phase reactions is to use a fixed bed reactor filled with catalyst par- microreactors can be found in textbooks (Keil, 2007; Kolb and
ticles. In recent years, more advanced reactor technologies have Hessel, 2004). In our study, the latter approach, taking into account
been introduced, such as monoliths, solid foams, catalyst fibres the fluid flow effects was selected to be applied for 1-butanol par-
and microreactors. In addition to possible industrial applications, tial oxidation. The selected modelling approach was well in line
microreactors have been developed as tools to conduct very pre- with the one successfully applied on another case study, ethylene
cise kinetic studies to generate intrinsic kinetic data by utilizing oxidation to ethylene oxide (Russo et al., 2015). However, the cur-
the main advantages of microreactors: well-defined flow pattern rent case, partial oxidation of 1-butanol, is a clearly more compli-
and rapid mass and heat transfer (Salmi et al., 2011; Keil, 2007; cated reaction system as alcohols with longer carbon chain lengths
Aroh and Jensen, 2018; Schwolow et al., 2014; Moore et al., produce a larger number of products than shorter alcohols. This
 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695 3

becomes evident, for example, when comparing the reaction sys- mixer (Supplementary Information). Both the reactant and product
tem of 1-butanol (Khan et al., 2018) with nine overall reactions gases were analyzed with online Fourier Transform InfraRed (FTIR)
producing n-butyraldehyde, water, butenes, CO, CO2, C3H6 and spectrometer (Gasmet instruments CR 5000). The experimental
C3H8 to that of ethanol (Kilpiö et al., 2017), where three reactions procedure has been already elaborated in detail in another related
have been observed producing acetaldehyde, acetic acid, and ethyl study (Khan et al., 2018).
acetate. The modelling of a microreactor for heterogeneously cat-
alyzed gas-phase partial oxidation of 1-butanol on titania sup- 2.2. Catalyst preparation and characterization
ported gold catalyst, to best of our knowledge, has not been
reported in the literature prior to this work. The reactor plates were coated with nano-TiO2 supported
In this work, the partial oxidation reactions of 1-butanol with monometallic Au catalyst. The details of the catalyst preparation,
air in the gas phase using a microreactor coated by Au/TiO2 cat- coating procedure and characterization are provided in another
alyst was studied. Two kinetic models, a mechanistic model related microreactor study (Khan et al., 2018). Based on pre-
based on presumed surface reaction steps and a power-law screening experiments, a 0.6 wt% Au/TiO2 catalyst was selected
model based on the law of mass action, were compared applying for this work to generate kinetic data. The thickness of the catalyst
the pseudo-homogeneous modelling approach. Finally, the layer was 17 ± 7 mm and the average Au nanoparticle diameter, for
numerical values of the kinetic parameters were estimated by coated and calcined at 450 °C, was 3.6 nm (Khan et al., 2018). All
using an advanced reactor model incorporating the fluid flow the kinetic experiments used in this work were produced using
properties. A generic modelling approach based on dynamic the same pair of Au/TiO2 coated microreactor plates. Thus, in the
mass balances of each compound in gas and solid phases was repeat experiments, besides same experimental conditions, the
applied to simulate and explain the behavior of this challenging coated plates were also the same (Khan et al., 2018).
multiple reaction system. The kinetic data for the partial oxidation of 1-butanol was
obtained at different temperatures, in a range between 130 and
2. Experimental equipment and procedures 350 °C, and by changing the inlet concentrations of 1-butanol
[Cin(C4H9OH)] and air [Cin(O2)] in stoichiometric ratios and the flow
2.1. Experimental arrangement of inert nitrogen to change the flow velocity (uG) inside the reactor.
Each experiment was conducted at five different temperatures
The partial oxidation of 1-butanol with air was studied in a cus- (130, 165, 250, 300 and 350 °C). The experimental parameters
tomized catalyst testing microreactor (CTMR) presented in Fig. 1, are collected in Table 1.
manufactured by Fraunhofer ICT-IMM, Mainz, Germany. The struc- The results of the experiments presented in our previous study
tured reactor plates which were coated with the catalyst (Sec- (Khan et al., 2018) for the 0.6 wt% Au/TiO2 (experiment 4 in
tion 2.2) were loaded inside the reactor. Each plate had 13 Table 1), were repeated again (two experiments) in this study.
channels with the dimensions of 1000 mm in width, 150 mm in The results were repeatable and within experimental error of ±2
depth and 75 mm in length. The catalyst plate geometry for the percentage points.
catalyst coating was selected in other related studies (Khan et al.,
2018; Nummipuro, 2015). Two plates were placed inside the reac- 2.3. Pseudo-homogenous plug flow model (1D)
tor with the channels facing each other so that the combined pair
of plates created a microreactor. A steady state 1D-pseudo-homogeneous plug flow model was
1-Butanol (VWR, 99.9%) was fed with a Gilson 307 pump to a used for the initial parameter estimation to compare the mechanis-
tubular heat transfer micro device (THTMD) acting as a microstruc- tic model and power law model approaches and finally to compare
tured evaporator by Fraunhofer ICT-IMM, Mainz, Germany. The selected power law model with kinetic modelling approach apply-
1-butanol vapors were then mixed with nitrogen (AGA, 5.0) and ing a 2D reactor model. The pseudo-homogeneous gas phase mass
synthetic air (AGA, 5.0) in a desired ratio using an in house built balance can be written as follows,

Fig. 1. Flowsheet of the equipment used for 1-butanol partial oxidation reaction with online FTIR.
4 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695

Table 1
Experimental matrix for 1-butanol partial oxidation reaction in CTMR for partial pressure of 1-butanol of 13.5 kPa (experiments 1–4) and 18 kPa (experiments 5–8).

EXP T [°C] Pbutanol [kPa] Cin,butanol (T = 130 °C) [mol/m3] Cin,O2 (T = 130 °C) [mol/m3] uG (T = 130 °C) [m/s]
1 130–350 13.5 3.79 2.10 2.47
2 130–350 13.5 3.99 2.08 1.60
3 130–350 13.5 3.63 2.10 1.25
4 130–350 13.5 3.44 2.10 0.94
5 130–350 18.0 5.28 2.76 1.98
6 130–350 18.0 5.20 2.79 1.63
7 130–350 18.0 5.40 2.74 1.21
8 130–350 18.0 4.79 2.45 0.94

dn_ i expressions to describe the physical properties of the components

¼ r i q B AC ð2Þ
dx and their mixtures under the process conditions.
The 1D pseudo-homogeneous plug flow model was solved using The microreactor used was a multiple-channel unit; the chan-
MATLABÒ software. The Nelder-Mead simplex algorithm was used nels were semi-cylindrical, and in this modelling study, the
for the minimization of the residual sum of squares (RSS). The ordi- hydraulic diameter approach was used, which is a standard
nary differential equations (ODEs) were solved by ODE15s solver. practice.
The objective function (Qp) was minimized by using non-linear The model presumed the reaction to progress within the
regression, washcoat layer on evenly distributed active sites of the catalyst
(Au/TiO2). Due to the high flow velocity and similar observations
Xnm  2
Q p ¼ min yexp  ycal : ð3Þ made in a corresponding reaction and reactor systems, the fluid–
i¼1
solid external mass transport resistance was assumed to be negli-
The rate constants ðk) were calculated at the average tempera- gible (Khan et al., 2015; Schmidt et al., 2013). Both the individual

ture (T ) using the temperature mean-centring approach to sup- reactions of each component and their internal diffusion steps
press the correlations between the Arrhenius parameters, were incorporated in the washcoat model (Salmi et al., 2013;
  Hayes et al., 1999). This kind of modelling approach has been suc-
 Ea 1 1
T  cessfully applied even for three-phase systems (Russo et al., 2015).
k ¼ k exp R T : ð4Þ
The approach relies heavily on mass balances, and it was combined
In addition, particle swarm algorithm was also used to cross with a simplified, well-accepted and powerful interpretation of the
check the minima. fluid flow by means of the dispersion concept. Based on the com-
parison of kinetic models applying the 1D pseudo-homogeneous
2.4. Two-dimensional dynamic model (2-D) reactor model, the power law kinetic model was chosen for the
studies with the 2D reactor model discussed in Section 2.4.2.
The modelling approach presented in Fig. 2 was selected to Another challenge was to find and use realistic physical properties
describe the microreactor. As discussed previously, this approach for the pure components and the gas mixture, as discussed in
was well in line with the ethylene oxidation case study (Russo Section 2.4.4.
et al., 2015) but the reaction system in the current case, the partial
oxidation of 1-butanol, was more complicated. Thus, applying the 2.4.1. Numerical approach
method required incorporating the expressions which describe The model equations were solved in the gPROMSÒ environment,
the case-specific multiple reaction kinetics as well as the which provides an inbuilt numerical method of lines solver for the
simultaneous solution of partial differential equations (PDEs)
(Schiesser, 1991). Some other potential modelling alternatives
and approaches that have been used to take the fluid flow effects
more punctually into account were summarized and discussed
by Russo et al. (Russo et al., 2015). However, due to the flexibility
and ease in the application of the model equations in the gPROMSÒ
environment, the discussed method was selected. The model is
based on two important distinctions. The first is the simple fluid
flow description and the second is the formulation of partial differ-
ential equations (PDEs), which describe the dynamic mass balances
allowing the solution to be performed by applying the well-
established and reliable numerical methods of lines. The simulta-
neous solution of the complete, perfectly defined set of equations
(PDEs, ODEs and algebraic equations) could then be obtained using
this method in the gPROMS ModelBuilder Software.

2.4.2. Parameter estimation

The generated reactor model with kinetic expressions was well
capable of providing the local concentrations of all the major com-
ponents within the microreactor and also within the washcoat
Fig. 2. Reactor model description for the gas phase and the solid phase, adapted
from Russo et al. (2015). The highlighted (grey) parts represent the catalyst layer
layer, where all the catalytic reactions were assumed to take place.
(solid phase). The empty space in between represents the free channel for the fluid The performance variables used as the target variables in the
(gas phase). parameter estimation were the experimental conversion of
 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695 5

1-butanol as well as the yield of the products. A modified Arrhe- The feed concentrations were used for both the gas and solid
nius law (Eq. (4)) was used to improve the parameter identification phases at the inlet. Both concentrations and gas velocity values
(Russo et al., 2015). The objective function (Eq. (3)) was the least were calculated at the temperature of each experiment assuming
square deviation between the model results and experimental the ideal gas law. As the initial condition, the feed concentrations
observations. The experimental observations were the 1-butanol were assumed to prevail all over the reactor space. As a standard
conversion and the yields of various product compounds. procedure, the Danckwerts closed boundary condition, the first
axial derivatives of concentrations were all set to reach zero at
the reactor outlet. At the gas–solid interface, the boundary condi-
2.4.3. Physical properties
tions implied that the mass fluxes from the gas phase were set
The physical properties for gaseous components (molecular dif-
equal to the fluxes entering into the solid washcoat.
fusivities, effective diffusivities, and axial dispersion coefficient)
The rate expressions were based upon power law kinetics as
were obtained by following the procedure given by Poling et al.
described in Table 3. The rate constants ðk) were calculated at aver-
(2000), Salmi et al. (2013) and by the recommendations of Russo 
et al. (2015). Briefly, the Wilke-Lee correlation was used to predict age temperature (T ) using the temperature mean-centring
the individual molecular diffusivities. The effective diffusivities approach to suppress the correlations between the Arrhenius
were estimated by adjusting the molecular diffusivity in a classical parameters (Eq. (4)).
way by particle porosity and tortuosity (getting roughly 1/10 the
value). The influence of the axial dispersion in the microreactor 3. Modelling results and discussion
channels was checked by a parametric investigation (Fig. 6). A
Péclet number of 100 was demonstrated to be sufficient enough 3.1. Reaction network and derivation of reaction kinetics based on
to exclude the axial dispersion dominance, mainly due to the high power law
gas velocity used to conduct the experiments.
The presumed reaction scheme for the partial oxidation of
1-butanol is displayed in Table 3. The partial oxidation with a high
2.4.4. Mass balance equations
yield to n-butyraldehyde was targeted, but all parallel and consec-
The collection of the mass balance equations is provided in
utive reactions could not be suppressed. The total oxidation reac-
Table 2. The reactions were assumed to progress only in the wash-
tions of 1-butanol and n-butyraldehyde were the most
coat catalyst layer at evenly distributed catalytically active sites.
exothermic reactions observed in the experiments, but the total
The concentrations were calculated both in the bulk phase of the
oxidation reactions were minimized by selecting the operation
gas and within the locations of the thin washcoat layer. In case
conditions and a selective catalyst. After surveying the experimen-
of rapid reactions, concentration gradients evolve within the cata-
tal data in combination with the reaction scheme presented in our
lyst layer due to diffusion limitations. The balance equations for
earlier study (Khan et al., 2018), four reactions were selected as the
gas and solid phases are summarized in Table 2. Finally, the com-
most significant and representative ones within the range of the
plete set of the boundary conditions are listed for this set of partial
operation conditions of kinetic experiments. The main reaction
differential equations.
taking place is the partial oxidation of 1-butanol to n-
A dynamic modelling approach was selected; the main benefit
butyraldehyde (1). Besides the main reaction, the considered side
of this approach is that it provides the opportunity to give an
reactions are the 1-butanol dehydration to butene (2) and the total
own initialization for all the state variables, which helps to achieve
oxidation of 1-butanol to carbon dioxide and water (3). The unde-
the final solution with a more reliable convergence. The mass bal-
sired subsequent reaction considered for n-butyraldehyde is the
ance equations for the gas phase involved terms describing accu-
retro-hydroformylation (4). The rate equations based on the law
mulation, convection, and dispersion, whereas the chemical
of mass action (power law model) of the applied reactions are
reactions were assumed to progress in the catalytic layer only.
given in Table 3.
The mass fluxes at the interface between the gas and solid catalyst
layer were specified as one of the boundary conditions. Energy bal-
ances were not considered here because microreactors offer supe- 3.2. Derivation of the kinetic model based on the assumptions of
rior specific surface areas for heat transfer (Kursawe, 2009). The surface reactions and surface species
temperature control sensors were inserted at two ends of the
microreactor which measured and controlled the temperature con- There are no studies published on the surface reactions of the
firming isothermal operation. 1-butanol gas-phase oxidation on noble metal catalysts. Therefore,
existing information on the oxidation processes of other corre-
sponding alcohols were utilized in the derivation of the kinetic
model based on adsorbed surface species and elementary reaction
Table 2 steps. Holz et al. (2014) reported that ethanol is adsorbed on the
Mass balance equations and boundary conditions.
Au/TiO2 surface predominantly as ethoxy species in the tempera-
PDE system – Mass balance equations ture range of our kinetic experiments. This was confirmed by sur-
GAS @C i;G @C i;G @ 2 C i;G face mechanistic studies on ethanol oxidation on gold
@t ¼ uG @z þ Dz @z2
  P
SOLID @C i;S D
¼  eeffp ;i
2
@ C i;S @C
þ rsp @ri;Sp þ e1p mij rj
@t @rp 2
j

Boundary conditions Table 3

GAS SOLID Reaction network and rate expressions including reaction orders.
 
z=0 C i;G z¼0 ¼ C i;Feed C i;S z¼0 ¼ C i;Feed J Reaction Rate expression (rj)
 
z=L @C i;G  @C i;S 
@z  ¼0 @z  ¼0 1 C4H9OH + 0.5 O2 ? C4H8O + H2O r 1 ¼ k1 C 2butaol;S C O2 ;S
z¼L z¼L

rp = R – @C i;G @ 2 C i;G @C i;S  2 C4H9OH ? C4H8 + H2O r 2 ¼ k2 C butanol;S
uG @z ¼ Dz @z2
þ Deff ;i RWsR @rp r ¼R
 p 3 C4H9OH + 6 O2 ? 5 H2O + 4 CO2 r 3 ¼ k3 C butanol;S C O2 ;S
rp = RW – @C i;S 
@r p r ¼R ¼0 4 C4H8O ? C3H6 + CO + H2 r 4 ¼ k4 C aldehyde;S
p W
6 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695

Table 4 reaction steps, reactions 4a and 4b in Table 4, being rate determin-

Reaction steps for n-butyraldehyde and 1-butene formation from 1-butanol. The rate ing steps and assuming other steps to be rapid steps in quasi-
determining steps are denoted for the formation of n-butyraldehyde (RDS1) and 1-
butene (RDS2).
equilibria. The final rate equations for the two overall reactions
are presented in Eqs. (5) and (6). The complete derivation of final
1 BuOH + * , BuOH* rate equations is presented in the Supplementary Information.
2 O2 + 2* , 2O*
3 BuOH* + O* , Butoxy* + OH*
For two other rate equations (full oxidation and retro-
4a Butoxy* + * , Buald*+H* RDS1 hydroformylation) power law equations introduced in Table 3
4b Butoxy* , Butene + OH* RDS2 were assumed due to their low significance for the reaction system.
5 OH* + H* , H2O* + *
6 H2O* , H2O + * 0 cBuOH c0:25
O
7 Buald* , Buald + *
k4a c0:5
2

H O
8 OH* + OH* , H2O* + O* RODH ¼  2
2 ð5Þ
1 þ K BuOH cBuOH þ K 0:5 c
O2 O2
0:5
þ K 0H2 O cH2 O þ K 0Buald cBuald

0 cBuOH c0:25
O
k4b c0:5
2

nanoparticles by Gong and Mullins (2008). Based on this observa- RDHD ¼  2 H O

 ð6Þ
tion, it can be assumed that 1-butanol is mainly adsorbed as 1 þ K BuOH cBuOH þ K 0:5 c
O2 O2
0:5
þ K 0H2 O cH2 O þ K 0Buald cBuald
butoxy species under our experimental conditions. Galvanin et al.
(2018) compared different surface reaction mechanisms for the Galvanin et al. (2018) proposed that the denominator in the cor-
oxidation of benzyl alcohol on AuPd/TiO2 catalyst and performed responding rate equations for the benzyl alcohol oxidation reaction
model discrimination based on kinetic parameter estimation. They system can be neglected due to constant total coverage of the sur-
proposed that the alcohol is first adsorbed molecularly on the cat- face species thus leading to a constant coverage of vacant sites.
alyst surface after which the alkoxy species are formed through a However, this hypothesis was not considered in our study and it
reaction of adsorbed alcohol and dissociatively adsorbed oxygen. was shown that the values of different adsorption equilibrium con-
These surface steps for butoxy species formation from 1-butanol stants (KBuOH, KO2, KH2O, KBuald) were relatively well identified.
are depicted in Fig. 3. However, in order to avoid over-parametrization of the system,
Holz et al. (2014) reported oxidative dehydrogenation (ODH) these adsorption equilibrium constants were assumed to be tem-
being the prevailing mechanism for the formation of aldehyde perature independent in the parameter estimation.
from the corresponding alcohol (ethanol). Therefore, of the mech-
anisms proposed by Galvanin et al. (2018), ODH was chosen for the 3.3. Comparison of the power law model and the surface kinetics
mechanism to describe the oxidation of 1-butanol to n- model with the 1D pseudo-homogeneus reactor model
butyraldehyde. Holz et al. (2014) also reported dehydration of
ethoxy species being main pathway for the formation of corre- The parameters obtained as well as the parameter estimation
sponding alkene i.e., ethene. Correspondingly, the dehydration of statistics with the power law model (PL) and mechanistic model
butoxy species (DHD) was selected to be the reaction pathway are given in Table 5.
for the formation of 1-butene. The reaction steps for the formation The R2 values of 99.0% and 99.4% were obtained for the power
of n-butyraldehyde and 1-butene from 1-butanol are displayed in law and mechanistic models, respectively. A better fit of the model
Table 4. (higher R2) value was obtained by using the mechanistic model as
The rate equations for the 1-butanol oxidation to revealed by Table 5. All the parameters in the mechanistic model
n-butyraldehyde and 1-butene were derived assuming the surface were relatively well identified including the adsorption

Table 5
Estimated parameter values for PL and mechanistic models using 1D-pseudohomogenous reactor model. Reference temperature was 240 °C. M: correlation matrix.

Parameter Value Error M Ea1 Ea2 Ea3 Ea4 k1,ref k2,ref k3,ref k4,ref KC4H9OH KO2 KH2O kC4H8O
Power law model
Ea1 [J/mol] 77.4103 7.0% Ea1 1.0
Ea2 [J/mol] 18.8103 15.3% Ea2 0.6 1.0
Ea3 [J/mol] 74.5103 26.2% Ea3 0.3 0.1 1.0
Ea4 [J/mol] 46.8103 29.1% Ea4 0.2 0.2 0.1 1.0
k1,ref [m3/(mols)] 8.8103 8.3% k1,ref 0.2 0.1 0.1 0.1 1.0
k2,ref [1/s] 1.7102 12.1% k2,ref 0.2 0.5 0.1 0.1 0.3 1.0
k3,ref [m3/(mols)] 1.2103 29.1% k3,ref 0.0 0.1 0.6 0.1 0.4 0.1 1.0
k4,ref [1/s] 3.7102 20.8% k4,ref 0.2 0.2 0.1 0.8 0.1 0.1 0.1 1.0
Mechanistic model
Ea1 [J/mol] 77.1103 6.9% Ea1 1.0
Ea2 [J/mol] 47.4103 8.9% Ea2 0.3 1.0
Ea3 [J/mol] 81.7103 18.8% Ea3 0.2 0.01 1.0
Ea4 [J/mol] 40.4103 18.5% Ea4 0.2 0.2 0.01 1.0
k1,ref [m3/(mols)] 3.6101 20.8% k1,ref 0.01 0.18 0.12 0.2 1.0
k2,ref [1/s] 3.6102 11.1% k2,ref 0.3 0.16 0.2 0.3 0.4 1.0
k3,ref [m3/(mols)] 8.1104 21.3% k3,ref 0.04 0.1 0.36 0.1 0.1 0.1 1.0
k4,ref [1/s] 2.4102 11.2% k4,ref 0.5 0.31 0.03 0.5 0.4 0.4 0.1 1.0
KC4H9OH 0.5 51.5% KB 0.14 0.03 0.05 0.1 0.5 0.1 0.1 0.1 1.0
KO2 0.83 41.6% KO 0.2 0.1 0.03 0.2 0.2 0.1 0.4 0.03 0.3 1.0
KH2O 1.19 33.1% KW 0.13 0.2 0.01 0.0 0.3 0.3 0.1 0.0 0.2 0.1 1.0
KC4H8O 4.5 24.6% KAl 0.3 0.44 0.036 0.2 0.2 0.4 0.2 0.4 0.1 0.1 0.5 1.0
 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695 7

Fig. 3. Surface reaction steps for the formation of butoxy species.

equilibrium constants. On the other hand, the variances of the

adsorption parameters were higher for the mechanistic model,
which could be due to the higher number of parameters compared
to the power law model.
The activation energy of the partial oxidation of 1-butanol was
determined to ca. 77 kJ/mol for both the kinetic models. The
activation energy of the 1-butanol partial oxidation on Au/TiO2
catalyst has not been reported in literature prior to this work, as
per our current knowledge. Jiang et al. (2012), have summarized
that the activation energies of aliphatic alcohols including 1-
butanol are between the region of 40–125 kJ/mol. Hence the
obtained activation energy in this work is in a reasonable order
of magnitude.
The parity plots of the power-law and mechanistic rate
equations with the 1D-pseudo-homogeneous reactor model are
presented in Fig. 4a and b. Based on these plots, it is evident that
a slightly better fit was obtained by mechanistic model. However,
due to minor difference in the fits of the model and the
simpler form of the model, power-law model was chosen for the
further parameter estimation and simulations with the 2D
Fig. 4. Overall parity plot (a) PL model and (b) mechanistic model with 1D-
dynamic model. pseudohomogenous reactor model, obtained considering all the available experi-
mental data.

3.4. Comparison of 1D-pseudohomogenous model and 2D dynamic

model with the power-law kinetic model

The parity plots of the parameter estimation with 2D dynamic 0.20

model applying the power-law kinetics is displayed in Fig. 5 and
the numerical results of the parameter estimation are given in
Table 6, along with statistical information. Comparison of the par- 0.15
ity plots (Figs. 4a and 5) and the parameter values (Tables 5 and
6) reveal that the identifiability is within the same region for the
xi,calc [-]

main reaction, i.e. the partial oxidation of 1-butanol. The variation

in the activation energy for the reaction, ca. 13 kJ/mol, could be 0.10
explained by the included dispersion effects in the fluid domain
and the radial diffusion in the catalyst domain, which might have
induced mass transfer limitations. The gPROMS 2D model 0.05
provided activation energies of ca. 150 kJ/mol for retro-
hydroformylation and total oxidation reactions compared to the
1D-pseudo-homogenous model, 45 and 75 kJ/mol respectively. ±15%
Consequently, the effect is reflected in the values of the rate con- 0.00
0.00 0.05 0.10 0.15 0.20
stants, which indicates that the identifiability of these parameters
could be further improved. However, the confidence intervals xi,exp [-]
with both models are acceptable, and the parameters had a low
correlation only. The model predictions show a window of error Fig. 5. Overall parity plot of 2D dynamic model obtained considering all the
of roughly 15%. available experimental data.
8 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695

Table 6
Estimated parameter values for the reactions using the 2D dynamic model. The reference temperature was 240 °C. C.I.: 95% confidence intervals. M: correlation matrix.

Parameter Value 95% M Ea1 Ea2 Ea3 Ea4 k1,ref k2,ref k3,ref k4,ref
3 2
Ea1 [J/mol] 63.710 1.810 Ea1 1.0
Ea2 [J/mol] 19.4103 6.6102 Ea2 0.3 1.0
Ea3 [J/mol] 149103 3.42102 Ea3 0.1 0.1 1.0
Ea4 [J/mol] 150103 4.52102 Ea4 0.2 0.1 0.1 1.0
k1,ref [m3/(mols)] 2.73 2.18102 k1,ref 0.5 0.2 0.5 0.3 1.0
k2,ref [1/s] 4.02 6.36102 k2,ref 0.1 0.1 0.2 0.2 0.4 1.0
k3,ref [m3/(mols)] 7.75102 8.2104 k3,ref 0.4 0.1 0.1 0.4 0.4 0.3 1.0
k4,ref [1/s] 1.50 2.52102 k4,ref 0.1 0.3 0.3 0.1 0.1 0.1 0.1 1.0

Fig. 6. The conversion of 1-butanol and the yields of the main products detected at the reactor outlet plotted against the reactor temperature for experiments 1–4 (a–d)
conducted at 13.5 kPa and experiments 5–8 (e–h) conducted at 18 kPa, Table 1. The symbols represent the experimental data, while the continuous lines represent the
predicted profiles from the 2D dynamic model in gPROMS using the power-law kinetics.

3.5. Comparison of experimental data and modelling results and total oxidation. Increasing the residence time increase for both
the partial pressures, also increased the yields of CO, CO2, and C3H6.
The evolution of the experimental conversion of the reactant The 1D-pseudo-homogeneous model was able to reproduce the
and yields of main products with respect to the reaction tempera- data well and it predicts satisfactorily all the trends of the compo-
ture are given in Fig. 6 for the 2D dynamic model and Figs. 7 and 8 nent concentration changes as a function of temperature. However,
for the 1D-pseudo-homogenous model applying the power-law the 2D dynamic model solved in gPROMS tends to overpredict the
model and the mechanistic model, respectively. extent of retro-hydroformylation reaction beyond 300 °C, com-
The main products observed at reaction conditions were n- pared to the 1D-pseudohomogenous model. The reasons for these
butyraldehyde, water, CO, CO2, and C3H6. The products detected deviations are related to variations in the obtained kinetic
in trace amounts were n-butylbutyrate and n-butanoic acid. The parameters.
trend for the product distribution with increasing temperatures
was presented in our recent study for the residence time of 3.6. Microreactor simulations with the 2D dynamic model
0.137 s and at the partial pressure of 13.5 kPa (Khan et al., 2018).
Similarly, till 250 °C, n-butyraldehyde is the main product but as In case of 2D dynamic model all the computations were con-
the temperature is increased to 300 and 350 °C, the amounts of ducted by considering the dispersion effects using the Péclet (Pe)
CO2, CO and C3H6 increase. The change in the product distribution number of 100, which is a typical order of magnitude value for a
indicated subsequent reactions through retro-hydroformylation microreactor applied in this study. A sensitivity analysis was
 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695 9

Fig. 7. The conversion of 1-butanol and the yields of the main products detected at the reactor outlet plotted against the reactor temperature for experiments 1–4 (a–d)
conducted at 13.5 kPa and experiments 5–8 (e–h) conducted at 18 kPa, Table 1. The symbols represent the experimental data, while the continuous lines represent the
predicted profiles from the 1D-pseudo-homogenous model in Matlab using the power-law kinetics.

performed checking the influence of the Péclet number on the nano-Au/TiO2 coating on the walls of the reactor channels. A quan-
aldehyde molar fraction, fixing the experimental conditions of titative explanation of reaction kinetics was targeted by the
experiment 2 (Table 1) and using different Pe-values of 10, 200 detailed modelling study and the experimental data were success-
and 500, Fig. 9. It is evident that the parameter becomes influenced fully reproduced by both models. The estimated kinetic parameters
only at low Péclet numbers, which are values out of the range were in a meaningful order of magnitude. The activation energy of
when operating with microreactors. the 1-butanol partial oxidation to n-butyraldehyde on the surface
For the 2D dynamic model, the concentration gradients for of Au/TiO2 was 77 kJ/mol. The simple power-law kinetic model
butanol and aldehyde were calculated within the washcoat layer was sufficient to explain the experimental observations and it
displaed in Fig. 10. The effective diffusion coefficients were evalu- was chosen for further studies to compare one-dimensional
ated according to a conservative criterion using the value of 1/10 of pseudo-homogeneous reactor model and the two-dimensional
molecular diffusivities. The 1-butanol concentration profile shows dynamic model.
a flattened profile indicating that at high 1-butanol concentrations The successful implementation of the modelling strategy pro-
the internal mass transfer resistance is negligible. In case of n- vides the evidence that besides the robust two-dimensional
butyraldehyde, the concentration profile gradient is visible along dynamic dispersion model, a simplified one-dimensional pseudo-
the catalyst layer. Hence, including the internal mass transfer resis- homogenous plug flow reactor model is sufficient to describe the
tance is necessary to obtain detailed information about the reac- microreactor performance. Both the models were equally good in
tion kinetics in microreactors for rapid reactions as in the explaining the experimental data but it can be stated that more
present case. realistic description of the microreactor system is obtained by the
two-dimensional model. For instance, including the radial diffusion
4. Conclusions effect inside the catalyst layer using the two-dimensional model is
beneficial to account for the mass transfer of the species within the
Power-law and surface kinetic modelling approaches were porous catalyst layer. Furthermore, the model could be extended
applied to the partial oxidation of gaseous 1-butanol to for studying the process optimization by adjusting other key
n-buryraldehyde on a gold catalyst. The experimental data parameters, such as the catalyst layer thickness to reveal the corre-
were produced by using a structured microreactor with a sponding effects on the reactant conversion and the product yields.
10 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695

Fig. 8. Conversion of 1-butanol and the yields of the main products detected at the reactor outlet plotted against the reactor temperature for experiments 1–4 (a–d)
conducted at 13.5 kPa and experiments 5–8 (e–h) conducted at 18 kPa, Table 1. The symbols represent the experimental data, while the continuous lines represent the
calculated profiles from the 1D-pseudo-homogenous model in Matlab using the mechanistic model.

Fig. 9. The molar fraction of the aldehyde as a function of the reactor temperature Fig. 10. Dimensionless concentration profiles along the dimensionless washcoat
computed fixing different Pe numbers in the 2D dynamic model. layer with 2D dynamic model.
 Y. Khan et al. / Chemical Engineering Science 221 (2020) 115695 11

Declaration of Competing Interest Hayes, R.E., Awdry, S., Kolaczkowski, S.T., 1999. Catalytic combustion of methane in
a monolith washcoat: Effect of water inhibition on the effectiveness factor. Can.
J. Chem. Eng. 77, 688–697. https://doi.org/10.1002/cjce.5450770410.
The authors declare that they have no known competing finan- Holz, M.C., Tölle, K., Muhler, M., 2014. Gas-phase oxidation of ethanol over Au/TiO2
cial interests or personal relationships that could have appeared catalysts to probe metal–support interactions. Catal. Sci. Technol. 4, 3495–3504.
https://doi.org/10.1039/C4CY00493K.
to influence the work reported in this paper.
Jiang, B., Chang, R., Hou, Y., Lin, Y., 2012. Kinetics of n-butanol partial oxidation to
butyraldehyde over Lanthanum–transition metal perovskites. Ind. Eng. Chem.
Res. 51, 13993–13998. https://doi.org/10.1021/ie3017175.
Acknowledgments
Keil, F.J., 2007. Modelling of Process Intensification. Wiley-VCH Verlag GmbH & Co.
KGaA, Weinheim, Germany. https://doi.org/10.1002/9783527610600.
Academy of Finland (MICATOX Project and Academy Professor Khan, Y., Marin, M., Karinen, R., Lehtonen, J., Kanervo, J., 2015. 1-Butanol
Grant of Prof. Tapio Salmi), Aalto University Department of Chem- dehydration in microchannel reactor: Kinetics and reactor modelling. Chem.
Eng. Sci. 137, 740–751. https://doi.org/10.1016/j.ces.2015.07.026.
ical and Metallurgical Engineering, and the University of Naples Khan, Y., Marin, M., Viinikainen, T., Lehtonen, J., Puurunen, R.L., Karinen, R., 2018.
‘‘Federico II” are acknowledged for the financial support. Process Structured microreactor with gold and palladium on titania: Active, regenerable
System Enterprise (PSE) is acknowledged for providing licenses and durable catalyst coatings for the gas-phase partial oxidation of 1-butanol.
Appl. Catal. A Gen. 562, 173–183. https://doi.org/10.1016/j.apcata.2018.06.010.
to enable this study. Aalto research infrastructure (Bioeconomy) Kilpiö, T., Behravesh, E., Russo, V., Eränen, K., Salmi, T., 2017. Physical modelling of
was used to complete the work. The work is a part of the activities the laboratory-scale packed bed reactor for partial gas-phase oxidation of
of Process Chemistry Centre (PCC), a center of Excellence financed alcohol using gold nanoparticles as the heterogeneous catalyst. Chem. Eng. Res.
Des. 117, 448–459. https://doi.org/10.1016/j.cherd.2016.10.026.
by Åbo Akademi University. Mr. Seppo Jääskeläinen and Mr. Pekka Kiwi-Minsker, L., Renken, A., 2005. Microstructured reactors for catalytic reactions.
Koivulaakso from Aalto University School of Chemical Engineering Catal. Today 110, 2–14. https://doi.org/10.1016/j.cattod.2005.09.011.
workshop are thanked for assembling the microreactor system, Knözinger, H., 2009. Heterogeneous catalysis and solid catalysts. Ullmann’s Encycl.
Ind. Chem. https://doi.org/10.1002/14356007.
troubleshooting and helping with designing of the mixer. M.Sc. Kolb, G., Hessel, V., 2004. Micro-structured reactors for gas phase reactions. Chem.
Janne-Joonas Tiitinen and Dr. Tiia Viinikainen are thanked for Eng. J. 98, 1–38. https://doi.org/10.1016/j.cej.2003.10.005.
assistance in the microreactor experiments. Prof. Riikka L. Puu- Kursawe, A., 2009. Partial Oxidation of Ethene to Ethylene Oxide in Microchannel
Reactors. Technische Universität Chemnitz.
runen is acknowledged for her continued support and valuable
Mascal, M., 2012. Chemicals from biobutanol: technologies and markets. Biofuels
insights for the preparation of the manuscript. Dr. Cesar Araujo is Bioprod. Biorefin. 6, 483–493. https://doi.org/10.1002/bbb.1328.
thanked for assistance and valuable discussions for the Matlab Moore, J.S., Smith, D., Jensen, K.F., 2016. Kinetics analysis and automated online
code. screening of aminocarbonylation of aryl halides in flow. React. Chem. Eng., 272–
279 https://doi.org/10.1039/c6re00007j.
Nummipuro, S., 2015. Coating Micro Reactor Plates with a Catalyst. School of
Chemical Technology, Aalto University https://aaltodoc.aalto.fi/handle/
Appendix A. Supplementary material 123456789/15545.
Poling, B.E., Prausnitz, J.M., O’Connell, J.P., 2000. The Properties of Gases and Liquids.
Supplementary data to this article can be found online at McGraw-Hill Professional, London, U.K..
Requies, J., Güemez, M.B., Iriondo, A., Barrio, V.L., Cambra, J.F., Arias, P.L., 2012. Bio
https://doi.org/10.1016/j.ces.2020.115695. n-butanol partial oxidation to butyraldehyde in gas phase on supported Ru and
Cu catalysts. Catal. Lett. 142, 417–426. https://doi.org/10.1007/s10562-012-
0793-5.
References
Russo, V., Kilpiö, T., Di Serio, M., Tesser, R., Santacesaria, E., Murzin, D.Y., Salmi, T.,
2015. Dynamic non-isothermal trickle bed reactor with both internal diffusion
Aroh, K.C., Jensen, K.F., 2018. Efficient kinetic experiments in continuous flow and heat conduction: Sugar hydrogenation as a case study. Chem. Eng. Res. Des.
microreactors. React. Chem. Eng. 3, 94–101. https://doi.org/10.1039/ 102, 171–185. https://doi.org/10.1016/j.cherd.2015.06.011.
c7re00163k. Russo, V., Kilpiö, T., Hernandez Carucci, J., Di Serio, M., Salmi, T.O., 2015. Modelling
Biella, S., Rossi, M., 2003. Gas phase oxidation of alcohols to aldehydes or ketones of microreactors for ethylene epoxidation and total oxidation. Chem. Eng. Sci.
catalysed by supported gold. Chem. Commun., 378–379 https://doi.org/ 134, 563–571. https://doi.org/10.1016/j.ces.2015.05.019.
10.1039/b210506c. Salmi, T., Hernández Carucci, J., Roche, M., Eränen, K., Wärnå, J., Murzin, D., 2013.
Chen, C., Wang, L., Xiao, G., Liu, Y., Xiao, Z., Deng, Q., Yao, P., 2014. Continuous Microreactors as tools in kinetic investigations: Ethylene oxide formation on
acetone-butanol-ethanol (ABE) fermentation and gas production under slight silver catalyst. Chem. Eng. Sci. 87, 306–314. https://doi.org/10.1016/j.
pressure in a membrane bioreactor. Bioresour. Technol. 163, 6–11. https://doi. ces.2012.10.016.
org/10.1016/j.biortech.2014.04.004. Salmi, T.O., Mikkola, J.-P., Wärnå, J.P., 2011. Chemical Reaction Engineering and
Galvanin, F., Sankar, M., Cattaneo, S., Bethell, D., Dua, V., Hutchings, G.J., Gavriilidis, Reactor Technology. CRC Press.
A., 2018. On the development of kinetic models for solvent-free benzyl alcohol Schiesser, W., 1991. The Numerical Method of Lines: Integration of Partial
oxidation over a gold-palladium catalyst. Chem. Eng. J. 342, 196–210. https:// Differential Equations. Academic Press.
doi.org/10.1016/j.cej.2017.11.165. Schmidt, S.A., Kumar, N., Reinsdorf, A., Eränen, K., Wärnå, J., Murzin, D.Y., Salmi, T.,
Gandarias, I., Miedziak, P.J., Nowicka, E., Douthwaite, M., Morgan, D.J., Hutchings, G. 2013. Methyl chloride synthesis over Al2O3 catalyst coated microstructured
J., Taylor, S.H., 2015. Selective oxidation of n-butanol using gold-palladium reactor—Thermodynamics, kinetics and mass transfer. Chem. Eng. Sci. 95, 232–
supported nanoparticles under base-free conditions. ChemSusChem. 8, 473– 245. https://doi.org/10.1016/j.ces.2013.03.040.
480. https://doi.org/10.1002/cssc.201403190. Schmidt, S.A., Balme, Q., Gemo, N., Kumar, N., Eränen, K., Murzin, D.Y., Salmi, T.,
Gandarias, I., Nowicka, E., May, B.J., Alghareed, S., Armstrong, R.D., Miedziak, P.J., 2015. Kinetics of ethanol hydrochlorination over c-Al2O3 in a microstructured
Taylor, S.H., 2016. The selective oxidation of n -butanol to butyraldehyde by reactor. Chem. Eng. Sci. 134, 681–693. https://doi.org/10.1016/j.
oxygen using stable Pt-based nanoparticulate catalysts: an efficient route for ces.2015.05.050.
upgrading aqueous biobutanol. Catal. Sci. Technol. 6, 4201–4209. https://doi. Schwolow, S., Heikenwälder, B., Abahmane, L., Kockmann, N., Röder, T., 2014.
org/10.1039/C5CY01726B. Kinetic and scale-up investigations of a michael addition in microreactors. Org.
Gong, J., Mullins, C.B., 2008. Selective oxidation of ethanol to acetaldehyde on gold. Process Res. Dev. 18, 1535–1544. https://doi.org/10.1021/op5002758.
J. Am. Chem. Soc. 130, 16458–16459. https://doi.org/10.1021/ja805303s. Walter, S., Malmberg, S., Schmidt, B., Liauw, M.A., 2005. Mass transfer limitations in
Görke, O., Pfeifer, P., Schubert, K., 2009. Kinetic study of ethanol reforming in a microchannel reactors. Catal. Today. 110, 15–25. https://doi.org/10.1016/
microreactor. Appl. Catal. A Gen. 360, 232–241. https://doi.org/10.1016/j. j.cattod.2005.09.019.
apcata.2009.03.026.