10448 Ind. Eng. Chem. Res.

2009, 48, 10448–10455

Modeling of the Kinetics for Methanol Synthesis using Cu/ZnO/Al2O3/ZrO2

Catalyst: Influence of Carbon Dioxide during Hydrogenation
Hye-Won Lim,† Myung-June Park,*,† Suk-Hwan Kang,‡,§ Ho-Jeong Chae,‡ Jong Wook Bae,‡ and
Ki-Won Jun*,‡
Department of Chemical Engineering, Ajou UniVersity, Suwon 443-749, Korea, and Petroleum Displacement
Technology Research Center, Korea Research Institute of Chemical Technology, Daejeon 305-600, Korea

A kinetic model for methanol (MeOH) synthesis over Cu/ZnO/Al2O3/ZrO2 catalyst has been developed and selected
to evaluate the effect of carbon dioxide on the reaction rates due to its high activity and stability. Detailed kinetic
mechanism, on the basis of different sites on Cu for the adsorption of carbon monoxide and carbon dioxide, is
applied, and the water-gas shift (WGS) reaction is included in order to provide the relationship between the
hydrogenations of carbon monoxide and carbon dioxide. Parameter estimation results show that, among 48 reaction
rates from different combinations of rate determining steps (RDSs) in each reaction, the surface reaction of a
methoxy species, the hydrogenation of a formate intermediate HCO2, and the formation of a formate intermediate
are the RDS for CO and CO2 hydrogenations and the WGS reaction, respectively. It is shown that the CO2
hydrogenation rate is much lower than the CO hydrogenation rate, and this affects the methanol production rate.
However, carbon dioxide decreases the WGS reaction rate, which prevents methanol from converting to dimethyl
ether, a byproduct. In such a way, a small fraction of carbon dioxide accelerates the production of methanol indirectly
within a limited range, showing a threshold value of the CO2 fraction for the maximum methanol synthesis.

1. Introduction dual site mechanism for the dissociative adsorption of the

hydrogen molecule on ZnO while carbon monoxide is adsorbed
Owing to high octane number, methanol (MeOH) is consid-
on the site of monovalent copper. Similarly, the kinetics for
ered as an excellent alternative energy resource. The high octane
the synthesis of methanol from carbon dioxide by a Cu-based
number ensures good antiknock performance, in addition to high
catalyst has been reported more than two decades ago.6,7
volatility, denser fuel-air charge, and excellent lean burn
According to the reports, the kinetic mechanism for the methanol
properties.1 Methanol is an excellent fuel in its own right, and
synthesis may involve both CO and CO2. Klier and co-workers8
it can also be blended with gasoline, although it has half the
assumed the adsorption of reactants, and they evaluated the
volumetric energy density relative to gasoline or diesel.2 It was
effect of carbon dioxide on the catalytic synthesis of methanol
also found that methanol can be conveniently converted into
over the copper-zinc oxide catalysts. They also proposed
ethylene or propylene in the MTO (methanol-to-olefins) process,
several mechanisms: (1) H2, CO, and CO2 were assumed to
and in turn, these olefins can be used to produce hydrocarbon
adsorb on the catalyst competitively (single site mechanism),
fuels and their products.3 In addition, recycling excess carbon
(2) CO and CO2 were assumed to adsorb on the catalyst
dioxide from industrial gases and the atmosphere, by chemical
competitively while hydrogen adsorbs on the different site from
reduction of CO2 with hydrogen to produce MeOH, will mitigate
CO and CO2 (dual site mechanism), (3) the adsorption site for
a major manmade cause of global warming.2 For this reason, a
CO and H2 are different and CO2 competes for both the CO
lot of research work is ongoing for its mass production and the
sites and the hydrogen sites. For more detailed elementary steps,
design of commercial processes. As a catalyst for methanol
it was shown that carbon monoxide is adsorbed in the form of
synthesis, Cu, Zn, Cr, and Pd are mostly used to prevent the
a carbonyl species, and a formate species is a usual intermediate
production of hydrocarbons and maximize the selectivity as well
for methanol synthesis and water gas shift reaction.9 There are
as the yield. Among these, Cu/ZnO catalyst is well-known for
many reports introducing the adsorption of CO and CO2 on two
high activity and selectivity for methanol synthesis reaction.
different sites.8,10,11 However, since their work ignores the
Support such as Al2O3 can further increase the activity and
water-gas shift (WGS) reaction, the relationship between CO
selectivity. Furthermore, the promoter such as Zr is known to
and CO2 (e.g., synergetic effect of CO2) on the methanol
enhance the copper dispersion and catalytic activity on methanol
synthesis reaction rate is not clearly understood.
synthesis catalysts, and its increased dispersion is due to the
As shown by Klier and co-workers,8 the copper is reduced
geometric effect of promoter on the ZrO2-CuO interaction
or oxidized by the adsorption of CO and CO2, respectively, and
originated from oxygen vacancies of ZrO2.4
this phenomenon suggests different adsorption sites for CO and
Although many kinetic mechanisms have been suggested for
CO2. It is worth noting that the published reports present
the synthesis of methanol and water, the exact mechanism is
conflicting viewpoints on the identity of the active sites on Cu-
not clearly discussed. Herman and co-workers5 proposed the
containing catalysts for the methanol synthesis. Most of them
* To whom all correspondence should be addressed. Tel.: +82-31- have considered CO hydrogenation to be the most significant
219-2383. Fax: +82-31-219-1612. E-mail: mjpark@ajou.ac.kr (M.-J.P.). reaction for methanol production. A possible reason for the
Tel.: +82-42-860-7671. Fax: +82-42-860-7388. E-mail: kwjun@ seemingly contradictory results in the literature is that CO2
krict.re.kr (K.-W. Jun). hydrogenation also occurs to some extent under a certain
†
Ajou University.
‡
Korea Research Institute of Chemical Technology. reaction condition. Perhaps, two different sites are necessary
§
Current address: Plant Engineering Center, Institute for Advances for methanol synthesis, and if that is the case, the apparent
Engineering (IAE), Suwon 443-749, Korea. disagreement over the identity of the active site might be solved.
10.1021/ie901081f CCC: $40.75  2009 American Chemical Society
Published on Web 09/16/2009
 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009 10449

Figure 1. Schematic diagram of the lab-scale reactor.

Therefore, the kinetics of methanol synthesis by CO and CO2 to elucidate the particle size of CuO (before reaction) and
hydrogenation is taken into account in the present study. metallic Cu (after reaction) with their crystalline phases.
A kinetic mechanism where CO and CO2 are assumed to 2.2. Activity Test. Catalytic activity tests were carried out
adsorb on different sites of Cu is suggested for the development in an isothermal tubular fixed bed reactor (10.2 mm I.D.) with
of hydrogenation reaction rates, while hydrogen is supposed to a catalyst weight of 1.0 g. The schematic diagram of the lab-
be adsorbed on the site of ZnO. In addition, the WGS reaction12 scale reactor system is shown in Figure 1. Prior to the reaction,
is considered which may relate CO hydrogenation to CO2 the catalyst was reduced for 4 h at 523 K using 5% H2 balanced
hydrogenation. Since there are several elementary steps for each with nitrogen. After reduction, the synthesis gas (H2/CO ) 2)
overall reaction, various combinations of rate determining steps was fed into the reactor. The reaction conditions included the
(RDSs) are developed, and then, the parameters are estimated composition change in feed gas (CO, CO2, and H2), T )
using experimental data to find the best fit model to evaluate 503-553 K, P ) 5.0 MPa and space velocity (SV) )
the detailed mechanism for methanol synthesis. It should be 2000-6000 mL/gcat · h. The above range of reaction variables
noted that the mechanism for the production of dimethyl ether were appropriately selected to show a significant variation of
(DME) is also included in the reaction set, since the experimental catalytic activity under different reaction conditions. The
data in the present study suggest the augmentation of additional catalytic activity at steady state after 15 h on stream without
reactions other than the three main reactions. These details are showing significant catalyst deactivation was selected to conduct
explained in the section 3. further reaction kinetics and reactor modeling. The products
were analyzed with an online gas chromatograph (Donam model
2. Experimental Section DS 6200) using thermal conductivity detector (TCD) to analyze
Ar, CO, and CO2 with a carbosphere packed column and flame
2.1. Catalyst Preparation and Characterization. The cata- ionized detector (FID) for hydrocarbons such as DME and
lysts were synthesized by the coprecipitation method using an MeOH and hydrocarbons with a GS-Q capillary column.
aqueous solution containing copper acetate, zinc acetate, 2.3. Characteristics of Cu/ZnO/Al2O3/ZrO2 Catalyst
aluminum nitrate, and zirconium oxide nitrate of required and Its Catalytic Performance. To select an appropriate
quantities (with the weight ratio of CuO/ZnO/Al2O3/ZrO2 ) catalyst for reaction kinetics and reactor modeling, a catalytic
61.5/31.5/3.3/3.7). The precipitant, Na2CO3, was dissolved in activity test on two different methanol synthesis catalysts, Cu/
deionized water and two separate solutions of precipitant and ZnO/Al2O3 and Cu/ZnO/Al2O3/ZrO2, was carried out under the
metal precursors were simultaneously mixed with a controlled following reaction conditions: T ) 523 K, P ) 5.0 MPa, and
feeding rate of around 10 mL/min at 343 K. The final pH of SV ) 4000 mL/gcat · h. The CO conversion and MeOH selectiv-
the solution was maintained at around 7. The precipitate was ity on Cu/ZnO/Al2O3/ZrO2 catalyst are higher than those of Cu/
further aged for 3 h at 343 K and then washed with deionized ZnO/Al2O3 catalyst as shown in Table 1. The high catalytic
water of 2 L. The dried power was calcined at 573 K for 5 h. performance is mainly attributed to the presence of smaller
Furthermore, catalytic activity on the well-known Cu/ZnO/Al2O3 metallic copper particles and higher surface area around 121.9
methanol synthesis catalyst with composition of CuO/ZnO/Al2O3 m2/g with the help of Zr promoter during the preparation step.
) 61.5/31.5/7.0 was compared, and finally, Cu/ZnO/Al2O3/ZrO2 The presence of small particle size around 9.8 nm on Cu/
catalyst was selected due to its high activity and stability instead ZnO/Al2O3/ZrO2 catalyst compared to that of 11.7 nm on Cu/
of Cu/ZnO/Al2O3 catalyst. The BET surface area and average ZnO/Al2O3 catalyst reveals a high metallic surface area of
pore size of Cu/ZnO/Al2O3 and Cu/ZnO/Al2O3/ZrO2 catalysts copper, resulting in high catalytic performance such as CO
were determined by N2-physisorption method using a Mi- conversion around 35.8% and MeOH selectivity around 94.1%.
cromeritics ASAP2400 apparatus at liquid-N2 temperature Therefore, Cu/ZnO/Al2O3/ZrO2 catalyst was finally selected to
(-196 °C). The powder X-ray diffraction (XRD) patterns were carry out further reaction kinetics and reactor modeling in the
obtained with a Rigaku diffractometer using Cu KR radiation present manuscript.
10450 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009

Table 1. Catalytic Performance and Physicochemical Properties of Cu/ZnO/Al2O3 and Cu/ZnO/Al2O3/ZrO2 Catalysta
surface area average pore particle size particle size CO conversion selectivity [mol %]
catalyst [m2/g]b size [nm]b of CuOc of Cuc [mol %] (MeOH/DME/HCd)
Cu/ZnO/Al2O3 106.9 27.9 11.6 11.7 26.8 93.5/2.7/3.8
Cu/ZnO/Al2O3/ZrO2 121.9 16.1 10.4 9.8 35.8 94.1/1.9/4.0
a
The catalytic activity measurement was performed under the following reaction conditions: T ) 523 K, P ) 5.0 MPa, and SV ) 4000 mL/gcat · h.
b
BET surface area and average pore size of the calcined catalysts were determined by N2-physisorption. c The particle size of copper oxide (before
reaction) and metallic copper (after reaction) was calculated from the values of full width at half-maximum (fwhm) of XRD diffraction peaks at 2θ )
35.5 and 43.3 respectively with the help of Sherrer’s equation. d The byproducts of hydrocarbons (HC) mainly include C1-C4.

3. Reaction Kinetics and Reactor Modeling Table 2. Elementary Reactions for Cu/ZnO/Al2O3/ZrO2 Catalyzed
Methanol Synthesis
3.1. Reaction Rates. Methanol synthesis reaction is com-
posed of three main reactions: (1) the hydrogenation of carbon H2 + 2s2 a 2H · s2
monoxide, (2) hydrogenation of carbon dioxide, (3) water-gas CO + s1 a CO · s1
shift reaction. In addition to these three main reactions, a side adsorption CO2 + s3 a CO2 · s3
reaction for the synthesis of dimethyl ether (DME) from H2O + s2 a H2O · s2
methanol is considered in this study.
CO + 2H2 a CH3OH (1) surface reaction elementary steps
(A) CO
CO2 + H2 a CO + H2O (2) hydrogenation Step 1: CO · s1 + H · s2 a HCO · s1 + s2
reaction13 Step 2: HCO · s1 + H · s2 a H2CO · s1 + s2
H2CO · s1 + H · s2 a H3CO · s1 + s2
CO2 + 3H2 a CH3OH + H2O (3) Step 3:
Step 4: H3CO · s1 + H · s2 a CH3OH + s1 + s2
2CH3OH a CH3OCH3 + H2O (4)
(B) water-gas
shift Step 1: CO2 · s3 + H · s2 a HCO2 · s3 + s2
Reaction 4 is included in this study, because the experimental reaction12,16 Step 2: HCO2 · s3 + H · s2 a CO · s3 + H2O · s2
results for CO2 conversion shows negative values under some
conditions. In the case that reaction 4 is excluded, CO2 is (C) CO2
generated by the WGS reaction (the reverse reaction of 2) whose hydrogenation Step 1: CO2 · s3 + H · s2 a HCO2 · s3 + s2
rate is dependent on the amount of H2O produced by the CO2 reaction14,15 Step 2: HCO2 · s3 + H · s2 a H2CO2 · s3 + s2
hydrogenation (reaction 3). As a result, the amount of CO2 Step 3: H2CO2 · s3 + H · s2 a H3CO2 · s3 + s2
generation cannot be larger than the amount consumed in Step 4: H3CO2 · s3 + H · s2 a H2CO · s3 + H2O · s2
reaction 3, corresponding to non-negative values of conversion. Step 5: H2CO · s3 + H · s2 a H3CO · s3 + s2
Step 6: H3CO · s3 + H · s2 a CH3OH + s3 + s2
Therefore, the additional reaction for the production of water
is required to explain the experimental observation. The
methanol selectivity data also shows values lower than 90%,
which supports the production of a product other than MeOH (C), respectively, are determined by fitting the experimental data
(data not shown). available in the literature.19 Figure 2 shows the comparison
The elementary reactions for the hydrogenation of CO and between experimental data and calculated values.
CO2, and the reverse water-gas shift (RWGS) reaction are listed
in Table 2. It is worth noting that there are two different sites 9.8438 × 104
ln KPA ) - 29.07 (5)
(s1 and s3) for the adsorption of CO and CO2, respectively.5 RT
The symbol “s1” represents Cu1+ while Cu0 is denoted by the
symbol “s3”. -4.3939 × 104
ln KPB ) + 5.639 (6)
For CO hydrogenation, the most strongly supported mecha- RT
nism consists of successive additions of adsorbed hydrogen
atoms to an adsorbed carbon monoxide molecule13 while the KPC ) KPA × KPB (7)
synthesis of methanol from carbon dioxide occurs via a formate
species (HCO2) adsorbed on copper.14,15 There is general where R represents the gas constant in Joules per mole kelvin.
agreement that hydrogen and water adsorb on Zn sites.16 Since The equilibrium constant for DME production is available in
the adsorption of H2 is rapid,10,17 it is assumed that the the literature.20
concentrations of ZnO sites and adsorbed hydrogen on ZnO
sites remain constant during synthesis.
The reaction is assumed to be reversible since the concentra- KDME ) 0.106 exp ( 2.1858 × 104
RT ) (8)
tion of MeOH is high, and the Langmuir-Hinshelwood model
is applied for the reaction rates. The assumption of rapid 3.2. Reactor Model. To consider the effect of external mass
equilibrium is applied to the adsorption steps, and the reaction transfer, the speed of agitation is varied in a batch reactor (the
rates are developed when each step is assumed to be a rate gas flow rate in a fixed-bed reactor). Since the size of the catalyst
determining step (cf. Table 3). Therefore, 48 combinations of particle is very small and both the flow rate and pressure are
reaction rates are obtained for methanol synthesis. For the relatively high, no external mass transfer resistance is assumed.21
production of DME, the reaction rate available in the literature18 The occurrence of any internal pore diffusion limitation is
is used. determined on the basis of the Weisz-Prater criterion, where
Equilibrium constants, KPA, KPB, and KPC for CO hydrogena- the dimensionless Weisz-Prater parameter (CWP) is calculated
tion (A), reverse WGS reaction (B), and CO2 hydrogenation as follows:
 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009 10451

-rA,obsFcRc2 resulting in a concentration gradient from the catalyst surface

CWP ) (9) to its pores.22,23
DeffCA
The effective diffusivity for multicomponents mixture is
calculated using the following equation:24
where -rA,obs is the rate of the reaction in moles per kilogram
of catalyst per second and Fc is the catalyst density in
kilograms per cubic meter. The symbols Rc, Deff, and CA N
Cj
represent the effective radius of the catalyst (the ratio of Deff )
1
DKe i
+ ∑D e
(10)
catalyst pellet volume to the surface area of catalyst) in j)1 i,jCt
j*i
meters, the effective diffusivity in squared meters per second,
and the limiting reactant concentration in the mixture in moles
per cubic centimeter, respectively. If the CWP is much greater where Knudsen diffusion coefficient (DeKi)24 and binary diffusion coefficient
than one, a strong resistance to internal pore diffusion exists, e 25
(Di,j) are calculated as follows:

Table 3. Reaction Rates for Methanol Synthesis Reaction and DME Production
reaction RDS rate equations

kAKCOKH20.5(PCOPH22 - PCH3OH /KPA)/PH21.5

CO hydrogenation A1 rA )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)

kAKCOKH2KCH,CO(PCOPH22 - PCH3OH /KPA)/PH2

A2 rA )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)

kAKCOKH21.5KCH,CO(PCOPH22 - PCH3OH /KPA)/PH20.5

A3 rA )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)

kAKCOKH22KCH,CO(PCOPH22 - PCH3OH /KPA)

A4 rA )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)

kBKCO2KH20.5(PCO2PH2 - PCOPH2O /KPB)/PH20.5

WGS reaction B1 rB )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kBKCO2KH2KCH,WGS(PCO2PH2 - PCOPH2O /KPB)

B2 rB )
(1 + KCOPCO)(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH20.5(PCO2PH23 - PCH3OHPH2O /KPC)/PH22.5

CO2 hydrogenation C1 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH2KCH,CO2(PCO2PH23 - PCH3OHPH2O /KPC)/PH22

C2 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH21.5KCH,CO2(PCO2PH23 - PCH3OHPH2O /KPC)/PH21.5

C3 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH22KCH,CO2(PCO2PH23 - PCH3OHPH2O /KPC)/PH21

C4 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH22.5KCH,CO2(PCO2PH23 - PCH3OHPH2O /KPC)/PH20.5

C5 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kCKCO2KH23KCH,CO2(PCO2PH23 - PCH3OHPH2O /KPC)

C6 rC )
(1 + KH20.5PH20.5 + KH2OPH2O)(1 + KCO2PCO2)

kDMEKCH3OH2(CCH3OH2 - ((CH2OCDME)/KP,DME))
DME production18 rDME )
(1 + 2√KCH3OHCCH3OH + KH2OCH2O)4
10452 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009

4 8RT Table 4. Physical Properties and Reactor Specifications
DKe i ) rpore (11) parameter value unit
3 πMi
bulk gas
kg/m3
10-4T1.75(Mi-1)1/2
0.498
density (Fg)
Dei,j ) (12) bulk pellet 700-900 kg/m3
Patm( ∑V i
1/3
+ ∑V j
1/3 2
) density (FB)
catalyst weight 0.001 kg
tube diameter (D) 0.0127 (1/2 in) m
The values of diffusion volume (V) are available in the heat of reaction CO hydrogenation -98438
literature.25 Since the values of CWP for all experimental (Hrxn) CO2 hydrogenation -54931 kJ/mol
conditions are calculated lower than 0.01, it is assumed that RWGS 43466
there exists no internal diffusion limitation.
Since no mass transfer resistance is assumed and the L-to-D where XCO and XCO2 represent the CO conversion and CO2
ratio of a lab-scale fixed-bed reactor used in the experiments is conversion, respectively. Subscripts “calc” and “exp” mean the
very large, a pseudohomogeneous plug flow model is used to calculated and experimental values, respectively, and wi is a
simulate the reactor as follows: weighting factor.
When the temperature dependence of kinetic parameters is
Mass balance considered, the strong correlation between pre-exponential factor
R N and the activation energy leads to a highly nonlinear problem,
dci
-us
dz
+ FB ∑
Ri,j ) 0 (13) and thus, the following form is used in the estimation procedure
j)1 to reduce the temperature dependency:

Energy balance

usCp
dFgT
) FB
NR

∑
(-∆Hi)jRj +
4U
(T - T) (14)
[ ( )]
kl(T) ) kl,0 exp -
El 1
R T
-
1
T0
(17)

dz Dt W

[ ( )]
j)1
-∆Hi 1 1
Ki(T) ) Ki,0 exp - (18)
Boundary conditions R T T0
at z ) 0, ci ) ci,0 ; T ) Tin (15)
The constants at the reference temperature, activation ener-
Here, us, FB, Fg, Rj, and Ri,j represent the linear velocity, bulk gies, and enthalpies are scaled in the form of yscaled ) yactual/
and gas densities, reaction rate and stoichiometric coefficient, yscaling-factor in order to reduce the statistical correlation and to
respectively, and other symbols are defined in the Nomenclature. prevent the ill-conditioning problems resulting from the differ-
Heat capacity (Cp), calculated using the values of each ences of the orders of magnitude between parameters.
component (Cp,i) as a function of temperature, and reaction The objective function, eq 16, is minimized using the
enthalpy for the reaction j (∆µHj) are assumed to be constant. lsqcurVefit subroutine in Matlab (The MathWorks, Inc.) where
The values of Cp,i and ∆µHj are available in a process simulator the Levenberg-Marquardt method is applied, and the numerical
(UniSim Design Suite, Honeywell Inc.). Other physical proper- integration of eqs 13 and 14, which is required during each
ties and reactor specifications are listed in Table 4. iterative step in the nonlinear regression, is performed by the
3.3. Parameter Estimation. To estimate kinetic parameters, ordinary differential equation (ODE) solver, ode23s imple-
conversions of CO and CO2 are taken into account, and thus, mented in Matlab.
the following objective function is defined:

{(
4. Results and Discusssion
Fobj )
1
2 ∑ w1
XCO,calc - XCO,exp 2
XCO,exp
+ ) Since each experimental condition produces 2 data points (i.e.,
CO conversion and CO2 conversion) and the number of

(
XCO2,calc - XCO2,exp
)}
2 experimental condition is 28, a total of 56 points are used in
w2 (16) the nonlinear data-fitting procedure. Detailed experimental
XCO2,exp
conditions are listed in Table 5.
The estimation results for 48 sets of RDS combinations are
summarized in Table 6. According to the parametric sensitivity
analysis, only 6 (when the B1 is applied) or 7 (RDS for the
WGS is assumed to be B2) influences the reaction rates
significantly, and thus, these parameters are estimated while the
other parameters are based on the values from the literature.11,12
The weighting factors are specified as 1 and 1.2 for CO and
CO2 conversion, respectively, by trial and error to improve the
estimation performance for CO2 conversion. The total average
error is defined as ∑i|(yexp,i - ycalc,i)/yexp,i|(100/N) is calculated
for the results of CO and CO2 conversions separately. Since
one of the aims of parameters estimation is to find the reaction
mechanism which best describes the kinetics of the system over
the wide range of conditions, the standard deviation of the
individual error of each experimental run around the total
Figure 2. Comparison of equilibrium reaction constants between experi- average error (STD) is also calculated as STD ) [(individ error
mental data and calculated values. - t avg error)2/(N - 1)]1/2.
 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009 10453
Table 5. Experimental Conditions Table 7. Estimated Parameters for the RDS Combination of 412
(Model No. 38)
feed composition [mol %]
estimated valuesa
number temperature [K] SV [mL/g · h] CO CO2 H2
1
2
3
523 4000 2
2
2
1
4
2
7
16
10
kA ) 1.16 × 10-9 exp - [ 7.01 × 103 1
R T(-
1
523 )]
4 2 0 4

[ )]
5 2 0.5 5.5
2.70 × 102 1 1
6
7
2
2
1
1
5
8
kB ) 2.82 × 10-5 exp -
R T(-
523
8 2 0.5 3.5
9 2 1.5 6.5
10
11
2
2
0.5
1.5
6.5
9.5
kC ) 1.15 × 10-6 exp - [ 1.19 × 102 1
R T(-
1
523 )]
12 0 1 3
13 2000 2 1 7
14
15
16
503
6000
4000
2
2
2
1
1
0.5
7
7
5.5
kDME ) 2.51 × 1011 exp - [ 6.45 × 105 1
R T
- (1
523 )]
17 2 1.5 8.5

[ )]
18 2 1 5
9.93 × 103 1 1
19
20 2000
2
2
1
1
8
7
KCO ) 4.96 × 10-8 exp
R T(-
523
21 6000 2 1 7
22 553 4000 2 1 7
23
24
2
2
0.5
1.5
5.5
8.5
KCH3OH ) 1.41 × 10-3 exp [ 6.05 × 103 1
R T(-
1
523 )]
25 2 1 5
26 2 1 8
27 2000 2 1 7 a
R is 8.314 J/mol · K, and T is in kelvin.
28 6000 2 1 7
rate determining steps for CO and CO2 hydrogenation are A4
In order to guarantee the balanced estimation between CO (surface reaction of a methoxy species H3CO · s1 and H · s2)9 and
and CO2 conversion, the selection criteria has been applied to C2 (hydrogenation of a formate intermediate HCO2),14 respec-
the error and standard deviation of each conversion; the tively. In addition, model no. 23 is a little bit worse for the
combination is selected if the values of the errors and the STD’s prediction of CO2 conversion since it cannot predict the negative
of CO and CO2 conversion are lower than 17%, 60%, 10%, CO2 conversion which is observed in the experimental data
and 30%, respectively. In Table 6, the selected RDS combina- under some conditions. Therefore, it is assumed that these
tions are highlighted with bold text, among which only two experimental results are also effective for the case of different
combinations have three bold cells: RDS combinations of 225 adsorption sites for CO and CO2 in the present study, and thus,
(A2-B2-C5, model no. 23) and 412 (A4-B1-C2, model no. 38). model no. 38 is selected for the best fit model and the values
Although the estimation result for the combination of 225 of estimated parameters are listed in Table 7.
(model no. 23) is as good as that for the model no. 38 (the It is worth noting that the formation of a formate intermediate
combination of 412), there is experimental evidence that the (B1 in Table 2) is determined as the RDS for the WGS reaction,

Table 6. Total Average Errors and Standard Deviations from the Estimation Results
error STD error STD
no. RDS CO CO2 CO CO2 no. RDS CO CO2 CO CO2
1 111 22.47 61.21 11.55 49.59 25 311 18.95 103.44 9.89 89.86
2 112 82.25 74.06 38.87 70.48 26 312 19.35 59.76 9.95 43.98
3 113 80.20 68.72 35.63 39.87 27 313 18.54 62.76 11.93 46.71
4 114 77.05 289.20 35.73 260.85 28 314 19.37 61.24 9.96 42.30
5 115 78.48 273.21 36.92 257.49 29 315 19.35 59.67 9.95 44.09
6 116 80.61 281.18 38.71 369.79 30 316 18.70 59.20 12.01 43.20
7 121 82.02 72.81 37.46 60.73 31 321 42.06 66.53 23.98 46.00
8 122 83.66 91.68 38.08 28.63 32 322 14.65 62.61 10.34 44.68
9 123 83.64 92.42 38.04 26.03 33 323 42.08 92.01 24.35 20.29
10 124 83.02 91.17 37.69 22.01 34 324 38.00 93.79 23.57 20.29
11 125 42.54 91.89 29.73 21.29 35 325 56.70 92.64 27.41 21.68
12 126 83.26 92.63 37.62 24.59 36 326 55.60 93.65 27.24 21.26
13 211 71.97 111.83 16.77 100.82 37 411 15.68 77.31 12.93 71.88
14 212 74.52 63.86 16.65 49.28 38 412 16.72 58.98 9.27 46.26
15 213 80.06 68.18 35.53 40.99 39 413 51.71 60.70 15.84 45.19
16 214 15.02 59.48 10.92 41.64 40 414 51.76 61.03 15.83 44.05
17 215 74.52 63.71 16.64 48.89 41 415 16.03 64.25 10.03 50.94
18 216 18.84 58.67 10.72 43.61 42 416 51.72 60.67 15.84 45.07
19 221 16.68 65.08 11.07 47.49 43 421 14.57 64.34 11.67 46.23
20 222 15.33 59.40 10.33 40.75 44 422 13.47 72.73 8.63 31.34
21 223 18.21 61.84 10.41 46.05 45 423 15.44 60.27 9.11 41.40
22 224 17.06 58.99 10.08 39.91 46 424 14.20 85.13 10.85 21.69
23 225 15.41 59.43 9.50 43.52 47 425 12.73 59.59 10.35 40.64
24 226 14.65 62.87 11.11 38.10 48 426 13.03 59.31 11.71 40.75
10454 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009

As a result, the DME production rate is high. As the CO2

fraction increases, the WGS reaction rate is decreased and, in
turn, the increase of water prohibits the DME production rate
by the increase in the reverse reaction, leading to the increase
of MeOH concentration. However, since the CO hydrogenation
reaction rate decreases much faster than the WGS reaction with
the increasing CO2 fraction (cf. Figure 4c), the MeOH concen-
tration decreases after a threshold value of CO2 fraction.
Consequently, it is summarized that the existence of CO2
increases the production of methanol to some extent, via the
decrease in WGS reaction, but the effect is limited to the small
fraction of CO2. Between 0 and about 0.25 CO2 fraction, the
Figure 3. Comparison of (a) CO and (b) CO2 conversions between MeOH concentration is higher than the case for no CO2, and
experimental and simulation data using model no. 38. the maximum MeOH concentration is accomplished when the
fraction is about 10%. It is worth noting that this value is based
on the kinetics only. If other factors such as transport resistance
in the reactor and a deactivation mechanism are considered, the
optimal value in the sense of process operation might be slightly
different from our value.29
Although the DME production affects the concentration of
MeOH, methanol selectivity is very high (close to one), which
means that the DME production is not dominant in the whole
reactions. Figure 4d shows that the contribution of CO2
hydrogenation is very low, and thus, one can infer that the CO2
gives rise to the synergetic effect on methanol production not
directly from the CO2 hydrogenation but indirectly through the
WGS reaction. The low reaction rate for CO2 hydrogenation is
also observed in Figure 4f, where the CO2 conversion shows
highly negative values due to the high WGS reaction rate (CO2
generation) and the low CO2 hydrogenation rate (CO2 consump-
tion). It is reported that the high concentration of CO2 diminishes
the methanol synthesis reaction rate,8 and this result supports
the low CO2 hydrogenation rate.
It is worth noting that, since the relationship between CO
Figure 4. Effect of the CO2 fraction on (a) the concentrations of MeOH and CO2 hydrogenation reactions may depend on the number
and DME; (b) MeOH selectivity; cumulative reaction rates of (c) CO of s1 and s3 sites, the different amount of ZrO2 which is known
hydrogenations and DME production, (d) CO2 hydrogenations, (e) reverse to affect the number of Cu0 and Cu1+ sites may change the
water-gas shift reaction, and (f) CO and CO2 conversions at the exit of degree of the relationship. Therefore, it might be necessary to
the reactor.
find a different set of kinetic parameters with respect to the
and this result is also reported by Peppley and co-workers in amount of ZrO2 used in the catalyst preparation.
their modeling study.26 However, Graaf and co-workers12
assumed that B2 is the RDS for their kinetic study. 5. Conclusions
Figure 3 shows the comparison of CO and CO2 conversions
between experimental and simulation data. It is observed that A mechanism for the adsorption of CO and CO2 on the
the CO conversions are in good agreement, while the CO2 different site of Cu-based catalyst for methanol synthesis has
conversion shows deviation from the experimental data, prob- been suggested and total of 48 reaction rates have been
ably resulting from the unknown mechanism. developed on the basis of different combinations of RDS for
One of the interesting features for the MeOH synthesis by CO and CO2 hydrogenations and WGS reaction. Kinetic
both CO and CO2 is that the production rate is rapidly increased parameters estimation shows that the combination of the surface
when even a small fraction of CO2 is included in the feed. This reaction of a methoxy species, the hydrogenation of a formate
feature is reported in several literature works,8,27,28 but there is intermediate HCO2, and the formation of a formate intermediate
no clear kinetics-based explanation of how the CO2 fraction for CO and CO2 hydrogenations and WGS reaction, respectively,
influences the methanol synthesis reaction rates. In order to is the best fit model, and the results are supported by
explain this phenomenon, the reaction rates for each hydrogena- experimental evidence in the literature. Further analysis has
tion reaction and WGS reaction are examined with a variety of implied that the existence of CO2 in the reactants increases the
CO2 fractions in the feed. Figure 4a clearly shows the increase methanol concentration at the exit of the reactor in a synergetic
of MeOH concentration at the exit of the reactor when the CO2 manner, that is, a small fraction of CO2 prevents the conversion
fraction is increased from 0, and the concentration of methanol of MeOH to DME indirectly, via the WGS reaction. Although
is decreased after CO2 fraction gets higher than the threshold there are several reports discussing the effect of CO2 on the
value. This feature is contributed to the decrease of DME MeOH synthesis reaction rates, it is the first time that the kinetic
production with the increasing CO2 fraction. As shown in Figure mechanism for the relationship between CO and CO2 hydro-
4e, when the CO2 fraction is very low or zero, the WGS reaction genation reactions is suggested. In conclusion, the kinetic model
(corresponding to the negative values of RWGS reaction) rate proposed in this study clearly proves the effect of carbon dioxide
is high or highest which results in the consumption of water. on the methanol synthesis and it is expected that the optimal
 Ind. Eng. Chem. Res., Vol. 48, No. 23, 2009 10455
fraction can be calculated using the developed model in Literature Cited
industrial processes.
(1) Chen, Z.; Yao, M.; Zhen, Z.; Zhang, Q. Energy Fuels 2009, 23,
2719–2730.
(2) Olah, G. A. Angew. Chem., Int. Ed. 2005, 44, 2636–2639.
Acknowledgment (3) Olah, G. A.; Doggweiler, H.; Felberg, J. D.; Frohlich, S.; Gridinia,
M. J.; Karpeles, R.; Keumi, T.; Inaba, S.; Ip, W. M.; Lammertsma, K.;
The authors would like to acknowledge the financial support Salem, G.; Tabor, D. C. J. Am. Chem. Soc. 1984, 106, 2143–2149.
of Ajou University (Contract No. 20071030) and the Ministry (4) Toyir, J.; de la Piscina, P. R.; Fierro, J. L. G.; Homs, N. Appl. Catal.,
of Knowledge Economy of Republic of Korea (2007-E-1D25- B 2001, 34, 255–266.
P-02-0-00). (5) Herman, R. G.; Klier, K.; Simmons, G. W.; Finn, B. P.; Bulko, J. B.
J. Catal. 1979, 56, 407–429.
(6) Chinchen, G. C.; Denny, P. S.; Parger, D. G.; Spenser, M. S.; Whan,
Nomenclature D. A. Appl. Catal. 1987a, 30, 333–338.
(7) Denise, B.; Sneeden, R. P. A. J. Mol. Catal. 1982, 17, 359–366.
CA ) concentration [mol /cm3] (8) Klier, K.; Chatikavanij, V.; Herman, R. G.; Simmons, G. W. J. Catal.
Cp ) heat capacity [kJ/kg · K] 1982, 74, 343–360.
CWP ) Weisz-Prater parameter (9) Edwards, J. F.; Schrader, G. L. J. Phys. Chem. 1984, 88, 5624–
5627.
D ) tube diameter [m] (10) McNeil, M. A.; Schack, C. J.; Rinker, R. G. Appl. Catal. 1989, 50,
Deff ) effective diffusivity [m2/s] 265–285.
e
Di,j ) binary diffusion coefficient for component i and j [cm2/s] (11) Coteron, A.; Hayhurst, A. N. Chem. Eng. Sci. 1994, 49, 209–221.
DKi ) Knudsen diffusion coefficient [cm2/s]
e (12) Graaf, G. H.; Stamhuis, E. J.; Beenackers, A. A. C. M. Chem. Eng.
Sci. 1988, 43, 3185–3195.
El ) activation energy [J/gmol · K] (13) Chinchen, G. C.; Hay, C. M.; Vandervell, H. D.; Waugh, K. C. J.
Fobj ) objective function Catal. 1987b, 103, 79–86.
Hj ) reaction enthalpy [kJ/mol] (14) Tagawa, T.; Plerzier, G.; Amenomiya, Y. Appl. Catal. 1985, 18,
kl,0 ) forward reaction rate constant (l ) A, B, C, DME), [gmol/ 285–293.
gcat · s] (15) Bowker, M.; Hadden, R. A.; Houghton, H.; Hyland, J. N. K.;
Waugh, C. J. Catal. 1988, 109, 263–273.
Ki ) species adsorption equilibrium constants (16) Dennison, P. R.; Packer, K. J.; Spencer, M. S. J. Chem. Soc.
KP,I ) reaction equilibrium constants Faraday Trans. I 1989, 85, 3537–3560.
Mi ) molecular weight [g/gmol] (17) Kung, H. H. Catal. ReV. Sci. Eng. 1980, 22, 235–259.
N ) total number of data points (18) Ng, K. L.; Chadwick, D.; Toseland, B. A. Chem. Eng. Sci. 1999,
54, 3587–3592.
Pi ) pressure [Pa] (19) Graaf, G. H.; Sijtsema, P. J. J. M.; Stamhuis, E. J.; Joosten, G. E. H.
R ) gas constant (R ) 8.314) [J/gmol · K] Chem. Eng. Sci. 1986, 41, 2883–2890.
rl ) reaction rate [mol/kgcat · s] (20) Mizsey, P.; Newson, E.; Truong, T.; Hottinger, P. Appl. Catal. A:
rA,obs ) observed rate [mol/kgcat · s] Gen. 2001, 213, 233–237.
Rc ) catalyst radius [m] (21) Chae, H. J.; Choo, S. T.; Choi, H.; Nam, I. S. Ind. Eng. Chem.
Res. 2000, 39, 1159–1170.
Rj ) reaction rate [mol/kgcat · s] (22) Fogler, H. S. Elements of Chemical Reaction Engineering; Prentice-
rpore ) mean pore radius [m] Hall: New Jersey, 1999.
STD ) standard deviation (23) Akpan, E.; Sun, Y.; Kumar, P.; Ibrahim, H.; Aboudheir, A.; Idem,
T ) temperature [K] R. Chem. Eng. Sci. 2007, 62, 4012–4024.
(24) Lommerts, B. J.; Graaf, G. H.; Beenackers, A. A. C. M. Chem.
Tw ) wall temperature [K] Eng. Sci. 2000, 55, 5589–5598.
us ) linear velocity [m/s] (25) Fuller, E. N.; Schettler, P. D.; Gidding, J. C. Ind. Eng. Chem. 1966,
Vi ) diffusion volume [cm3/mol] 58, 19–27.
w ) weighting factor (26) Peppley, B. A.; Amohlett, J. C.; Kearns, L. M.; Mann, R. F. Appl.
X ) conversion Catal. A: Gen. 1999, 179, 21–29.
(27) Monnier, J. R.; Apai, G.; Hanrahan, M. J. J. Catal. 1984, 88, 523–
Greek Letters 525.
(28) Vanden Bussche, K. M.; Froment, G. F. J. Catal. 1996, 161, 1–
Fc ) catalyst density [kg/m3] 10.
FB ) bulk densities [kg/m3] (29) Zahedi, G.; Elkamel, A.; Lohi, A. Energy Fuels 2007, 21, 2977–
Fg ) gas densities [kg/m3] 2983.
Ri,j ) stoichiometric coefficient
ReceiVed for reView July 6, 2009
Subscripts ReVised manuscript receiVed August 26, 2009
Accepted August 27, 2009
calc ) calculated value
exp ) experimental data IE901081F