Fuel 84 (2005) 917–926

www.fuelfirst.com

Water gas shift reaction kinetics in Fischer–Tropsch synthesis

over an industrial Fe–Mn catalyst
Bo-Tao Tenga,b, Jie Changa,b, Jun Yanga,c, Gang Wanga,b, Cheng-Hua Zhanga,b,
Yuan-Yuan Xua, Hong-Wei Xianga, Yong-Wang Lia,*
a
Group of Catalytic Kinetics and Theoretical Modeling, State Key Laboratory of Coal Conversion, Institute of Coal Chemistry,
Chinese Academy of Sciences, Taiyuan 030001, People’s Republic of China
b
Graduate School of the Chinese Academy of Sciences, Beijing 100039, People’s Republic of China
c
Department of Chemistry, Jinan University, Guangzhou 510632, People’s Republic of China

Received 1 June 2004; received in revised form 13 September 2004; accepted 15 December 2004
Available online 21 January 2005

Abstract
The kinetics of water gas shift (WGS) reaction over an Fe–Mn catalyst under Fischer–Tropsch synthesis (FTS) reaction conditions is
studied in a spinning basket reactor. Experimental conditions are varied as follows: temperature of 533–573 K, reactor pressure of
10.0–26.5 bar, H2/CO feed ratio of 0.66–2.0 and space velocity of 0.66–2.65!10K3 Nm3 kgK1 cat s
K1
. By separately fitting WGS kinetics
parameters with experimental data, which is possible in the spinning basket reactor with neglecting concentration and temperature gradients,
different kinetics models of WGS are derived and discriminated on the basis of four sets of WGS elementary reactions. Kinetics experimental
results show that the WGS reaction under FTS reaction conditions is far from equilibrium. Two types of WGS mechanisms are investigated.
One is the formate mechanism, and the other is the direct oxidation mechanism. It is found that the formate mechanism is better in fitting
experimental data than the direct oxidation mechanism over the Fe–Mn catalyst under the FTS reaction conditions. The optimized kinetics
model with formate intermediate dissociation as the rate-determining step (RDS) can fit the WGS experimental results well. The simplified
WGS kinetics model can easily be used for industrial modeling applications.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Kinetics; Water gas shift reaction; Fischer–Tropsch synthesis

1. Introduction
With the depleting resource of crude oil all over the
world, the Fischer–Tropsch synthesis (FTS) process, which
converts syngas to mainly straight chain hydrocarbons with
wide carbon number distribution, becomes a promising
route to meet the continuously increasing demand for liquid
fuels and chemical feedstocks conventionally obtained from
crude oil [1–3]. The syngas feedstock for FTS processes can
be produced from all kinds of carbon containing resources,
namely coal, biomass and carbon-containing wastes through
gasification, and natural gas through partial oxidation. For
the coal-derived syngas, the iron-based catalysts have
* Corresponding author. Tel.: C86 351 4130 337; fax: C86 351
4050 320.
E-mail address: ywl@sxicc.ac.cn (Y.-W. Li).
0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2004.12.007
normally been used for FTS due to the low cost and high
activity for both FTS and water gas shift (WGS) reactions
[4]. The sufficient WGS activity enables to use the coal
derived syngas with a low H2/CO ratio.
The mechanism and kinetics of the individual WGS
reaction has systemically and extensively been studied
using many methods [5–10]. An excellent review for the
WGS mechanism and kinetics over iron-based, copper-
based and cobalt molybdenum-based catalysts was pub-
lished by Newsome [11] in 1980. However, the kinetics
study of WGS under FTS reaction conditions receives
relatively little attention [12–20]. This is partially due to the
fact that the WGS kinetics in a FTS process becomes
difficult to address because of the accompanying FTS
reactions, which remarkably increases the complexity in the
discrimination of WGS kinetics models as well as in the
WGS kinetics parameter estimation.
918 B.-T. Teng et al. / Fuel 84 (2005) 917–926

Nomenclature
Ev activation energy for WGS reaction, kJ molK1
Fobj,W objective function of WGS reaction
kW4 WGS rate constant of the forward reaction in
WGSI RDS4, molK1 kgK1 sK1 barK1
kW4,0 preexponential factor of rate constant of CO2
formation, molK1 kgK1 sK1 barK1
kWK4 WGS rate constant of the reversible reaction in
WGSI RDS4, molK1 kgK1 sK1 barK1
KW1 equilibrium constant of CO adsorption elemen-
tary step in WGSI RDS4, barK1
KW2 equilibrium constant of H2O dissociation
elementary step in WGSI RDS4, barK1
KW3 equilibrium constant of surface reaction of
[COS] with [OHS] in WGSI RDS4
KW5 equilibrium constant of H2 desorption elemen-
tary step in WGSI RDS4, bar
KP equilibrium constant of WGS reaction
mi,exp experimental molar flow rate, mol sK1
mi,cal calculated molar flow rate, mol sK1
MARR mean absolute relative residual
Nexp the number of experimental points
NRS the number of responses for FTS
Pi partial pressure of component i, bar

Models describing the complex FTS reaction have reactions for the WGS reaction are derived in this work, and
recently been developed by simultaneously fitting both are listed in Table 1.
WGS and FTS kinetics with experimental data [17–20]. For the derivation of the rate expressions, the WGS
This conventional treatment is systematic. However, our reaction and the FTS reaction (hydrocarbon formation) are
experiences in the kinetics development of FTS have assumed to proceed on different active sites [22–24], and
indicated that the complex FTS kinetics can also bring one rate-determining step (RDS) is assumed in the sequence
about uncertainties to the ‘simple’ WGS kinetics as long as of WGS elementary reactions. Corresponding to the
the two coupled reaction routes (WGS and hydrocarbon mechanisms listed in Table 1, WGSI RDS4 means that the
formation) are simultaneously treated typically over iron reaction mechanism is WGSI, and the RDS is the 4th
FTS catalysts. The coupling nature cannot be separated in elementary reaction step. The remaining steps can be
the kinetics study using a fixed-bed reactor because the considered to be at quasi-equilibrium. The rate of formation
reaction environment is different along its axis. However, in of CO2 can be written as:
a continuous spinning basket reactor, the reaction environ-
ment is uniform all over the reactor. Thus in a continuous RCO2 Z kW4 ½COOHS K kWK4 PCO2 ½HS (1)
spinning basket reactor, the WGS kinetics model discrimi-
nation and parameter estimation can be separated from the From the elementary reaction step listed in Table 1, the
FTS kinetics models in a FTS process. intermediates can be expressed as follows:
The objective of this work is to systematically establish ½COS Z KW1 PCO ½S (2)
and discriminate Langmuir–Hinshelwood–Hougen–Watson
(LHHW) kinetics models for the WGS reaction on the basis pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
of possible detailed mechanisms over an industrial Fe–Mn ½HS Z PH2 =KW5 ½S (3)
catalyst under FTS reaction conditions. By using the
experimental data from a continuous spinning basket ½OHS Z Kw2 PH2 O ½S2 =½HS
reactor, a set of WGS kinetics models are estimated and
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
discriminated separately. Z Kw2 PH2 O ½S= PH2 =Kw5 (4)

2. Kinetics models

The formate intermediate and direct oxidation mechan-

isms for the WGS reaction are proposed in literature [21],
and are shown in Fig. 1. For the formate intermediate
mechanism, the formate species is formed through the
reaction between carbon monoxide and a hydroxyl species
or water. The hydroxyl intermediate is formed via the
decomposition of water. Another mechanism is the direct
oxidation mechanism via a regenerate or redox mechanism
to form adsorbed or desorbed CO2. On the basis of the Fig. 1. Water gas shift reaction via the formate species (a) and the direct
mechanisms shown in Fig. 1, four sets of elementary oxidation mechanism (b).
 B.-T. Teng et al. / Fuel 84 (2005) 917–926
Table 1
Elementary reaction steps for WGS reaction
Reaction step Elementary reaction
WGS I 1 COCSZCOS
2 H2OC2SZOHSCHS
3 COSCOHSZCOOHSCS
4 COOHSZCO2CHS
5 2HSZH2C2S
WGS II 1 COCSZCOS
2 H2OCSZH2OS
3 COSCH2OSZ COOHSCHS
4 COOHSZCO2CHS
5 2HSZH2C2S
WGS III 1 COCSZCOS
2 H2OC2SZOHSCHS
3 OHSCSZOSCHS
4 COSCOSZCO2C2S
5 2HSZH2C2S
WGS IV 1 COCSZCOS
2 H2OC2 SZOHSCHS
3 OHSCSZOSCHS
4 COSCOSZCOOSCS
5 COOSZCO2CS
6 2HSZH2C2S
½COOHS ZKW3 ½COS½OHS=½S
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z KW1 KW2 KW3 PH2 O PCO ½S= PH2 =KW5
Normalizing the concentrations of all the intermediates
on the catalyst surface leads to:
½S C ½HS C ½COS C ½OHS C ½COOHS Z 1
Combining Eqs. (2)–(6), the concentration of free site [S]
can thus be expressed as follows:
1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW5 C KW1 PCO C KW2 PH2 O = PH2 =KW5 C KW1 KW2 KW3 PCO PH2 O = PH2 =KW5
Finally, the rate expression of CO2 formation can be
expressed as:
RCO2Z kW4 ½COOHS K kWK4 PCO2 ½HS
Z ðkW4 KW1 KW2 KW3 PCO PH2 O
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
K kWK4 PH2 PCO2 =KW5 Þ½S= PH2 =KW5
According to the chemical reaction equilibrium of the
WGS reaction, the backward rate constant kWK4 can be
expressed as follows:
kWK4 Z kW4 KW1 KW2 KW3 KW5 =KP
The equilibrium constant KP of the WGS reaction is
calculated using the following relation [17]
5078:00
lnKp ¼ K 5:90 þ 13:96 !10K4 T K 27:59 !10K8 T 2
T 4.19 g. The catalyst is diluted using 30 cm3 inert silica
(10)
919
By different assumptions of RDS for the elementary
reactions listed in Table 1, other rate expressions of CO2 can
be derived, and the results are listed in Appendix.
3. Experimental
3.1. Experimental setup
The kinetics experiments are carried out with a
continuous spinning basket reactor (stainless steel, HZ
0.081 m, D0Z0.052 m, DiZ0.046 m) system shown in
Fig. 2, consisting of a feed section and a reactor section. The
CO and H2 streams pass through an activated charcoal trap
and a silica gel molecular sieve trap to remove water and
other impurities. Before entering the reactor, the flow rates
of CO (O99.99%) and H2 (O99.99%) are controlled by two
Brooks 5850E mass flow controllers. The outlet of the
reactor is connected with a hot trap (393 K) and then an ice
trap (273 K) at the system pressure. After these product
collectors, the pressure is released through a backpressure
regulator. Liquid products are accumulated in hot and ice
trap for a typical sampling period of 11–12 h. The flow rate
of tail gas is measured by a wet test gas meter.
(5) Blank experiments show that the spinning basket reactor
charged with inert silica sand without the catalyst has no
conversion of syngas.
A stabilization period of 200 h is conducted under the
reaction conditions, and then the kinetics measurement is
(6)
carried out. After the process conditions are changed, at
least 12 h is used for the system stabilization before a new
mass balance period. A total of 22 sets of experimental data
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (7)
are obtained for detailed kinetics study. The experimental
operations together with the kinetics results are listed in
Table 2. The carbon, hydrogen and oxygen material
balances of all the data in steady-state reactions are between
98 and 102%.
3.2. Catalyst and reduction
(8)
The Fe–Mn catalyst used in the kinetics study is
prepared by the Institute of Coal Chemistry (ICC),
Chinese Academy of Sciences. The preparation method
has been patented [25]. The performance of the catalyst
(9) under different reaction conditions was discussed by Ji
et al. [26] in a fixed-bed reactor and Bai et al. [27] in a
slurry reactor. The fresh catalyst is crushed and sieved to
particles with the diameter of 0.25–0.36 mm (40–60
ASTM mesh). The weight of the catalyst loaded is
sand with the same mesh size range.
920 B.-T. Teng et al. / Fuel 84 (2005) 917–926

Fig. 2. Experimental set-up.

Table 2
Experimental operation conditions and reaction results

No. T (K) P (bar) H2/CO ratio Fin (ml/min) Reaction Fexit (ml/ Oil (g) Wax (g) Water (g)
time (h) min)
1 533 15.8 1.00 167.2 23.95 147.9 2.57 16.65 2.49
2 533 20.5 1.51 208.2 11.87 175.8 2.72 5.78 1.77
3 533 25.1 2.05 250.2 12.00 210.4 1.7 4.85 4.77
4 543 16.5 0.67 166.4 19.08 146.2 3.11 8.15 0.81
5 543 16.3 1.01 249.6 12.33 236.4 1.76 7.03 1.84
6 543 11.0 0.99 167.2 12.18 160.3 1.78 3.46 1.09
7 543 26.0 1.50 333.7 11.98 278.0 4.01 6.83 5.13
8 543 22.0 1.99 333.5 12.32 291.5 3.75 3.13 6.0
9 553 16.7 1.01 249.6 20.38 214.5 6.58 6.83 4.65
10 553 21.5 0.68 416.8 12.75 404.6 3.62 9.36 2.4
11 553 26.1 1.00 333.0 12.00 267.0 6.53 8.4 3.69
12 553 12.0 1.49 167.0 12.19 140.3 2.9 3.06 1.4
13 553 16.5 2.02 250.2 12.32 215.1 2.27 2.13 3.84
14 563 16.7 1.00 333.0 23.32 287.9 11 5.44 6.49
15 563 16.8 1.50 333.7 12.58 289.4 4.24 1.63 5.19
16 563 20.3 1.00 416.4 11.67 364.1 6.7 3.28 4.55
17 563 25.7 0.68 416.8 11.87 345.9 6.88 11.45 3.01
18 563 11.7 2.02 250.2 12.22 226.0 2.64 2.07 3.56
19 573 17.5 1.00 416.4 21.03 359.3 13.57 4.92 7.74
20 573 19.8 1.00 499.8 12.32 429.5 8.75 2.49 4.95
21 573 20.9 1.50 583.7 12.20 524.1 7.13 1.23 10.66
22 573 25.5 1.99 666.6 11.88 597.9 6.2 1.21 14.42
 B.-T. Teng et al. / Fuel 84 (2005) 917–926
The catalyst is in situ reduced using syngas with
H2/COZ1.0, gas hourly space velocity 1000 hK1 at 3.0–
4.0 bar, and 1800 rpm. The temperature programmed
reduction is performed by heating the reactor system from
room temperature to 393 K in 2 h and leveling off for 2 h,
heating to 543 K at the rate of 10 K/h and to 553 K at the
rate of 2 K/h and then keeping this temperature for 48 h until
the accomplishment of the catalyst reduction process.
The reactor temperature is lowered to 493 K, and operation
conditions are adjusted to the desired values for the FTS
reaction.
3.3. Product analysis
The FTS products are separated into three portions: gas
phase, liquid phase (from the ice trap) and wax phase (from
the hot trap). Both the purified syngas and the tail gas are
analyzed on a gas chromatography (GC). H2, O2, N2, CH4
and CO are separated by using a 13! molecular sieve
packed column (1.5 m!3 mm i.d., Ar carrier flow) and
detected with a thermal conductivity detector (TCD). C1–C8
hydrocarbons in the gas phase are analyzed on C022 =CK22
(170–250 mm) packed column (7.2 m!3.2 mm i.d.), flame
ionization detector (FID), N2 carrier. CO2 is separated on a
silica gel packed column (1 m!3 mm i.d., H2 carrier) and
detected with TCD and quantified by using external
standard method. The oil product from the cold trap is
separated on a 60 m!0.25 mm (i.d.) OV-101 capillary
column (N2 carrier, FID). The temperature is raised from
333 K (maintained for 20 min) to 563 K at the rate of
3 K/min. The wax product from the hot trap is firstly
dissolved in CS2 (0.5–1.0 mass %) and then analyzed on
OV-101 capillary column, FID, N2 carrier with temperature
programmed (1 K/min) from 343 to 563 K.
3.4. Estimation of mass transfer limitations
The external mass transfer limitation is investigated by
comparing the CO conversions under different stirring
speeds of the reactor. The experimental results are shown in
Fig. 3. According to Fig. 3, it can be seen that the external
diffusion resistance can be eliminated at 900 rpm at 533 K,
1200 rpm at 553 K and 1500 rpm at 573 K. Apparently the
stirring speed needed to eliminate the external mass transfer
limitation correspondingly increases with the increase of the
reaction temperature. This is due to the fact that, the relative
rate of external mass transfer versus the reaction rate
decreases with the increase of temperature. Therefore, the
corresponding stirring speed should ensure that the
experimental data measured are in the kinetically limited
regime. In our experiments, all the experiments are carried
out at 1800 rpm which is safe to eliminate the external mass
transfer limitations for all kinetic conditions.
Our extensive experimental results proved that the
particle diameter used in this experiment is safe for negligible
intraparticle diffusion limitations [20]. The Weisz–Prater
921
Fig. 3. The effect of stirring speed on the CO conversion (reaction
conditions: 16.0 bar, H2/COZ1.0, 0.66!10K3 Nm3 kgK1
cat s
K1
for 533 K
3
and 16.0 bar, H2/COZ1.0, 1.32!10 Nm kgcat s for 553 and 573 K).
K3 K1 K1
number is estimated to be smaller than 0.1 (based on
conservatively low estimation of the bulk diffusivity for CO
in the effective diffusivity of 1.0!10K5 cm2 sK1 [28]),
which also ensures the absence of the internal mass transfer
limitations of all the experimental data. The kinetics
parameters optimization later in this paper also indicates
that the experimental results are free from the external and
internal mass transfer limitations and in the kinetically
limited regime.
4. Results and discussion
The overall reactions in a simplified form for the FTS
reaction system can be written as follows:
Paraffin formation :
ð2n C 1ÞH2 C nCO/ Cn H2nC2 C nH2 O ð11Þ
Olefin formation : 2nH2 C nCO/ Cn H2n C nH2 O (12)
Water gas shift reaction : H2 O C CO/ CO2 C H2 (13)
The hydrocarbon formation and WGS reactions are
generally believed as two sets of parallel reactions, which
proceed on two different active sites and are coupled
through water formed in hydrocarbon formation reactions.
By separately fitting the WGS kinetics parameters with
experimental data, a set of WGS kinetics models are
discriminated. The objective function for the optimization
of the WGS kinetics is defined as
Nexp  
X mi;exp K mi;cal 2
Fobj;W Z ; (14)
iZ1
mi;exp
where m i,exp and m i,cal represent the experimental
and calculated CO2 molar flow rate of ith experimental
data, Nexp represents the number of experimental points.
922 B.-T. Teng et al. / Fuel 84 (2005) 917–926

The mean absolute relative residual (MARR) between

experimental and calculated molar flow rate of CO2 is
defined as:
Nexp
X jmi;exp K mi;cal j 1
MARR Z $ $100% (15)
iZ1
mi;exp Nexp

The kinetics parameter estimation for a WGS kinetics

model is to find the minimum of the objective function.
In order to optimize the kinetics parameters globally, the
parameters of the various rival models are scanned in a first
step using the genetic algorithm (GA) approach [29], and
then using the Levenberg–Marquardt (LM) algorithm to
make refined optimization. The quality of the fitting
between the calculated and experimental data, the statistical
Fig. 4. Comparison of experimental ðPCO2 PH2 =PCO PH2 O Þ of WGS with the tests and the physicochemical constraints are used to
equilibrium KP. evaluate the significance of models (F-test) and kinetics
parameters (t-test).
Table 3 It is shown in Fig. 4 that the experimental
Different assumptions of RDS for different elementary reactions and
MARR for different kinetics models PCO2 PH2
WGS reaction Rate-deter- MARR PCO PH2 O
mechanism mining step (%)
WGS I 1 COCSZCOS 1 23.44
values over the wide reaction conditions in this kinetics study
2 H2OC2SZOHSCHS 2 20.52 are much less than the equilibrium values KP. This indicates
3 COSCOHSZ 3 12.65 that the WGS reaction under FTS reaction conditions is far
COOHSCS from equilibrium, which is also observed by others [27,30].
4 COOHSZCO2CHS 4 7.85 However, the individual WGS reaction over industrial iron-
5 2HSZH2C2S 5 19.44
based or copper-based catalysts is generally under the
WGS II 1 COCSZCOS
2 H2OCSZH2OS chemical equilibrium control regime [11].
3 COSCH2O SZ 3 15.76 Twelve kinetics models for WGS reaction are estimated,
COOHSCHS and the MARRs for different kinetics models are listed in
4 COOHSZCO2CHS 4 7.89 Table 3.
5 2HSZH2C2S
It is found from Table 3 that the best WGS kinetics
WGS III 1 COCSZCOS
model obtained from the formate mechanism (WGSI,
2 H2OC2SZOHSCHS
3 OHSCSZOSCHS 3 15.36 WGSII) is better in fitting experimental data than the best
4 COSCOSZCO2C2S 4 11.89 model from the direct oxidation mechanism (WGSIII,
5 2HSZH2C2S WGSIV). A possible explanation is that the dissociation
WGS IV 1 COCSZCOS of hydroxyl intermediate to adsorbed oxygen and hydrogen
2 H2OC2 SZOHSCHS is not energetically favorable under FTS reaction con-
3 OHSCSZOSCHS 3 15.61
4 COSCOSZCOOSCS 4 12.11
ditions. This is also supported by quantum calculation on
5 COOSZCO2CS 5 11.34 transition metals that the hydroxyl dissociation is energe-
6 2HSZH2C2S tically unfavorable with a relatively high activation barrier
[31]. In situ infrared spectroscopy has confirmed

Table 4
Values of the parameters for kinetics model of WGSI RDS4 and its reduced kinetics model

WGSI RDS4 Simplified WGSI RDS4

Parameter Value Unit t-Value Parameter Value Unit t-Value
5 5
Kw4,0 (7.03G0.04)!10 mol kg s
K1 K1
bar K
509.95 Kw4,0 (7.10G0.03)!10 mol kg s
K1 K1
bar K1
615.95
1

Ev 81.64G2.37 kJ molK1 88.93 Ev 81.96G1.95 kJ molK1 108.84

KW1 (1.00G0.06)!10K1 barK1 44.44 KW1 (2.65G0.04)!10K2 barK2 153.52
KW2 (7.47G0.23)!10K2 barK1 85.46 KW3 (2.21G0.05)!104 bar 120.67
KW3 37.59G2.84 34.24 F-value 313.68
KW5 460.40G4.85 bar 245.03
F-value 185.90

Statistical results: F0.01(6, 16)Z4.20, and tabulated t0.01(16)Z2.58 [39].

 B.-T. Teng et al. / Fuel 84 (2005) 917–926 923

The MARR of kinetics model of WGSII RDS4 is much

Fig. 5. Comparison of the calculated and experimental CO2 mole flow rates.
the existence of formate species on the iron-based catalyst
[32] and on Fe 100 surface [33]. In situ Fourier transform
infrared spectroscopy (FTIR) in the diffuse reflectance
mode (DRIFTS) has also detected the formate species over
some transition metals including iron [34]. We can therefore
conclude that the formate mechanism is more feasible than
the direct oxidation mechanism over the Fe–Mn catalyst
under the FTS reaction conditions.
According to the MARR analysis on the formate
mechanism WGSI listed in Table 3, only kinetics model
derived from RDS4 is with a MARR less than 10%. The best
WGS kinetics model of WGSI RDS4 is consistent with the
conclusion by Wang et al. [19] and Yang et al. [20] with the
same reaction mechanism and RDS obtained by much more
effort for fitting the FTS and WGS kinetics models
simultaneously. It is also found from Table 3 that the
elementary reactions of CO adsorption, H2O dissociation
and H2 formation are not the RDS in the WGS reaction
under the FTS reaction conditions due to the large
deviations of these models from experimental data.
similar to that of WGSI RDS4 because the two kinetics
models have the similar expression form in spite of different
sets of elementary reaction steps. This indicates that the
kinetics method cannot discriminate whether water reacts as
associative or dissociative form in the surface reaction if the
RDS is the dissociation of formate intermediate to CO2. This
conclusion is also supported by Dry [35]. It has been pointed
out that different surface chemical reactions may lead to the
same kinetics expression and kinetics studies cannot
unambiguously ‘prove’ a proposed mechanism. To dis-
criminate whether water reacts as associative or dissociative
form, some characterization over the catalyst and quantum
chemical calculation [36] or microkinetics analysis of every
elementary reaction step [37] are necessary. However, this
type of work is out of our scope.
The parameter values of the kinetics model of WGSI
RDS4 are listed in Table 4. It is found that the present model
is statistically significant as indicated by F-test results, and
all kinetics parameters are in the confidence intervals above
99% as indicated by t-test results [38,39]. The estimated
activation energy is comparable with the values reported in
literature [12–20]. The high activation energy of WGS,
81.64 kJ molK1, also suggests that the mass transfer
limitations are eliminated and the kinetics experiments are
in the kinetically limited regime.
The comparison of the predicted and experimental CO2
mole flow rates is shown in Fig. 5. The predicted CO2 flow
rates by WGSI RDS4 fit the experimental values well. The
MARR of kinetics model WGSI RDS4 is much less than
that of kinetics models developed in literature as shown in
Table 5.
On the basis of the WGSI RDS4 kinetics model, a
simplified form of the kinetics model can be obtained by
reducing the original model within a two-step mechanism
theory [40,41]. The reduced elementary reactions are listed
in Table 6.

Table 5
Summary of rate expressions for WGS reaction under FTS reaction conditions in literature and MARR calculated in this work

Kinetic expression Refer- Reactor Operation conditions MARR

ence (%)
RWGS Z kW PCO [12,13] Iron catalyst 52.16
RWGS Z kW ðPH2 O PCO K [14] Berty reactor Prec.iron catalyst at TZ538–590 K, PZ10–25 bar, H2/CO feed 50.90
PCO2 PH2 =KP Þ ratioZ1.47–4.04 and space velocity 300–800 hK1
kW ðPH2 O PCOKPCO2 PH2 =KP Þ [18] Slurry tank Fe–Cu–K catalyst at TZ493–523 K, PZ7.9–29.6 bar, H2/CO feed 11.95
RWGS Z PH2 PCOCaPH2 O
reactor ratioZ0.67–1.06 and space velocity 1.0–4.0L(STP) gK1
cat h
K1

kW ðPH2 O PCOKPCO2 PH2 =KP Þ 19.36

RWGS Z PCOCaPH2 OCbPCO2
kW ðPH2 O PCO =P0:5
H KPCO2 PH =KP Þ
0:5
[17] Fixed bed Fe–Cu–K catalyst at TZ523–623 K, PZ6–21 bar, H2/CO feed 25.65
RWGS Z 2
ð1CaPH2 O =P0:5 2
2
H Þ ratioZ3.0–6.0 and space velocity 0.36–2.43 NL kgK1
cat s
K1
2
kW ðPH2 O PCOKPCO2 PH2 =KP Þ [15] Spinning basket Fe–Cu–K–SiO2 catalyst at TZ523 K, PZ8.0–40 bar, H2/CO feed 14.8
RWGS Z ð1CK1 PCOCK3 PH2 O Þ2
reactor ratioZ0.25–4.0, space velocityZ0.5–2.0!10K3 Nm3 kgK1 cat s
K1

kW ðPH2 O PCOKPCO2 PH2 =KP Þ 14.3

RWGS Z P0:5 2
H ð1CK1 PCOCK3 PH2 O Þ
2
k ðP PCOKPCO2 PH2 =KP Þ [19] Fixed bed Fe–Cu–K catalyst at TZ493–542 K, PZ10.9–30.9 bar, H2/CO 13.5
RWGS Z W HP20:5O CK
v PCO PH2 O
H 2 feed ratioZ0.98–2.99, and space velocity of 4000–10,000 hK1
924 B.-T. Teng et al. / Fuel 84 (2005) 917–926

Table 6 1. Kinetics experimental results show that the WGS reaction

Simplified elementary reactions from the kinetics model of WGSI RDS4
Reaction step Elementary reaction
1 COCH2OC2SZCOOHSCHS
2 COOHSZCO2CHS
3 2HSZH2C2S
Using the same method, the kinetics expression for the
reduced two-step mechanism model is obtained:
RCO2Z kW2 ½COOHS K kWK2 PCO2 ½HS
Z ðkW2 KW1 KW3 PCO PH2 O K kWK2 PH2 PCO2 Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!½S= KW3 PH2
1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW3 C KW1 PCO PH2 O = PH2 =KW3
The MARR of the reduced kinetics model is 8.48%,
which also fits the WGS experimental data fairly well. The
parameter values of the simplified kinetics model of WGSI
RDS4 are also listed in Table 4. The simplified WGS
kinetics model can easily be used in the industrial reactor
simulation, design or optimization.
under FTS reaction conditions is far from equilibrium over
the wide reaction conditions.
2. The formate mechanism is more feasible than the direct
oxidation mechanism over the Fe–Mn catalyst under the
FTS reaction conditions.
3. This kinetics method can discriminate that CO adsorption,
water dissociation and hydrogen formation are not the RDS
in the WGS reaction under the frame of the formate
intermediate mechanism.
4. This kinetics method cannot discriminate whether water
reacts as associative or dissociative form in the surface
reaction if the RDS is the dissociation of formate
intermediate to CO2.
(16)
5. The estimated activation energy of the WGS kinetics
model is comparable with the values reported in literature.
(17) Thepredicted CO2 formation rates fit the experimental data
well.
6. The optimal kinetics model of the formate intermedi-
ate dissociation as RDS can be reduced to a two-step
mechanism model. It also fits the WGS experimental
data well and can be easily used for industrial
application.

5. Conclusion Acknowledgements

Experiments for the kinetics of WGS and FTS reactions
over an Fe–Mn catalyst are carried out over a wide range of
industrially relevant reaction conditions. By separately fitting
WGS kinetics parameters with experimental data, detailed
kinetics models of four sets of elementary reactions are
derived and discriminated. The following conclusions can be
drawn:
Financial support from the Chinese Academy of Sciences
(Project No. KGCX1-SW-02), National Ministry of Science
and Technology of China via 863 plan (Project No.
2001AA523010), Shanxi Natural Science Foundation
(20031032), and National Natural Sciences Foundation of
China (Project No. 20473111, 20590361) are gratefully
acknowledged.

Appendix. Rate expressions of the different models of WGS

Kinetics model: WGSI RDS1

ðkW1 KW2 KW3 KW4 KW5 PH2 O PCO K kWK1 PCO2 PH2 Þ½S
RCO2 Z kW1 PCO ½S K kWK1 ½COSZ
KW2 KW3 KW4 KW5 PH2 O

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW5 C PCO2 PH2 =ðKW2 KW3 KW4 KW5 PH2 O Þ C KW2 PH2 O = PH2 =KW5 C PCO2 PH2 =KW5 =KW4
Kinetics model: WGSI RDS2
RCO2 Z kW2 PH2 O ½S2 K kWK2 ½OHS½HS Z ðkW2 KW1 KW3 KW4 PCO PH2 O K kWK2 PCO2 PH2 =KW5 Þ½S2 =ðKW1 KW3 KW4 PCO Þ

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW5 C KW1 PCO C PCO2 PH2 =KW5 =ðKW1 KW3 KW4 PCO Þ C PCO2 PH2 =KW5 =KW4
Kinetics model: WGSI RDS3
ðkW3 KW1 KW2 KW4 KW5 PCO PH2 O K kWK3 PH2 PCO2 Þ½S2
RCO2Z kW3 ½COS½OHS K kWK3 ½COOHS½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
KW5 PH2 KW4
 B.-T. Teng et al. / Fuel 84 (2005) 917–926 925

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW5 C KW1 PCO C KW2 PH2 O = PH2 =KW5 C PCO2 PH2 =KW5 =KW4
Kinetics model: WGSI RDS4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RCO2Z kW4 ½COOHS K kWK4 PCO2 ½HS Z ðkW4 KW1 KW2 KW3 PCO PH2 O K kWK4 PH2 PCO2 =KW5 Þ½S= PH2 =KW5

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW5 C KW1 PCO C KW2 PH2 O = PH2 =KW5 C KW1 KW2 KW3 PCO PH2 O = PH2 =KW5
Kinetics model: WGSI RDS5
RCO2 Z kW5 ½HS2 K kWK5 PH2 ½S2Z ðkW5 KW1 KW2 KW3 KW4 PCO PH2 O K kWK5 PH2 PCO2 Þ½S2 =PCO2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½S Z 1=ð1 C KW1 KW2 KW3 KW4 PCO PH2 O = PCO2 C KW1 PCO C KW1 KW2 KW3 PCO PH2 O PCO2 =KW4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
C KW2 PH2 O PCO2 = KW1 KW3 KW4 PCO Þ
Kinetics model: WGSII RDS3
RCO2Z kW3 ½COS½H2 OS K kWK3 ½COOHS½HS Z ðkW3 KW1 KW2 KW4 PH2 O PCO K kWK3 PCO2 PH2 =KW5 Þ½S2 =KW4

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW5 C KW1 PCO C KW2 PH2 O C PCO2 PH2 =KW5 =KW4
Kinetics model: WGSII RDS4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RCO2Z kW4 ½COOHS K kWK4 PCO2 ½HS Z ðkW4 KW1 KW2 KW3 PH2 O PCO K kWK4 PH2 PCO2 =KW5 Þ½S= PH2 =KW5

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW5 C KW1 PCO C KW2 PH2 O C KW1 KW2 KW3 PCO PH2 O = PH2 =KW5
Kinetics model: WGSIII RDS3
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RCO2 Z kW3 ½OHS½S K kWK3 ½HS½OS Z ðkW3 KW1 KW2 KW4 KW5 PH2 O PCO K kWK3 PCO2 PH2 Þ½S2 =ðKW1 KW4 PCO KW5 PH2 Þ

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW5 C KW1 PCO C KW2 PH2 O = PH2 =KW5 C PCO2 =ðKW1 KW4 PCO Þ
Kinetics model: WGSIII RDS4
RCO2 Z kW4 ½COS½OS K kWK4 PCO2 ½S2 Z ðkW4 KW1 KW2 KW3 KW5 PH2 O PCO K kWK4 PCO2 PH2 Þ½S2 =PH2

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW5 C KW1 PCO C KW2 PH2 O = PH2 =KW5 C KW2 KW3 KW5 PH2 O =PH2
Kinetics model: WGSIV RDS3
RCO2 Z kW3 ½OHS½S K kWK3 ½HS½OS
 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
Z ðkW3 KW1 KW2 KW4 KW5 KW6 PCO PH2 O K kWK3 PH2 PCO2 Þ½S2 = KW1 KW4 KW5 KW6 PH2 PCO

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW6 C KW1 PCO C KW2 PH2 O = PH2 =KW6 C PCO2 =ðKW1 KW4 KW5 PCO Þ C PCO2 =KW5
Kinetics model: WGSIV RDS4
RCO2 Z kW4 ½COS½OS K kWK4 ½COOS½S Z ðkW4 KW1 KW2 KW3 KW5 KW6 PCO PH2 O K kWK4 PH2 PCO2 Þ½S2 =ðKW5 PH2 Þ

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1C PH2 =KW6 C KW1 PCO C KW2 PH2 O = PH2 =KW6 C KW2 KW3 KW6 PH2 O =PH2 C PCO2 =KW5
926 B.-T. Teng et al. / Fuel 84 (2005) 917–926

Kinetics model: WGSIV RDS5

RCO2 Z kW5 ½COOS K kWK5 PCO2 ½S Z ðkW5 KW1 KW2 KW3 KW4 KW6 PCO PH2 O K kWK5 PH2 PCO2 Þ½S=PH2

1
½S Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 C PH2 =KW6 C KW1 PCO C KW2 PH2 O = PH2 =KW6 C ð1 C KW1 KW4 PCO ÞKW2 KW3 KW6 =PH2

References [23] Rao KRPM, Huggins FE, Mahajan V, Huffman GP, Rao VUS,
Bhatt BL, et al. Top Catal 1995;2:71–8.
[1] Dry ME. Appl Catal A 1999;189(2):185–90. [24] Zhang HB, Schrader GL. J Catal 1985;95:325–32.
[2] Schulz H. Appl Catal A 1999;186(1–2):3–12. [25] Zhong B, Wang Q, Peng SY. Chinese Patent ZL95106156.9.
[3] Dry ME. Catal Today 2002;71(3–4):227–41. [26] Ji YY, Xiang HW, Yang JL, Xu YY, Li YW, Zhong B. Appl Catal A
[4] Davis BH. Catal Today 2003;84(1–2):83–98. 2001;214:77–86.
[5] Rethwisch DG, Dumesic JA. J Catal 1986;101:35–42. [27] Bai L, Xiang HW, Li YW, Han YZ, Zhong B. Fuel 2002;81:1577–81.
[6] Ovesen CV, Stoltze P, Norsko JK, Campbell CT. J Catal 1992;134: [28] Zennaro R, Tagliabue M, Bartholomew CH. Catal Today 2000;58:
445–68. 309–19.
[7] Callaghan C, Fishtik I, Datta R, Carpenter M, Chmielewski M, [29] Han RF, Zhang YK, Wang YN, Xu YY, Li YW. J Fuel Chem Technol
Lugo A. Surf Sci 2003;541:21–30. [China] 2001;29(4):371.
[8] Fishtik I, Datta R. Surf Sci 2002;512:229–54. [30] Wu BS, Bai L, Xiang HW, Li YW, Zhang ZX, Zhong B. Fuel 2004;
[9] Ovesen CV, Clausen BS, Hammershøi BS, Steffensen G, Askgaard T, 83:205–12.
Chorkendorff I, et al. J Catal 1996;158:170–80. [31] Michaelides A, Hu P. J Am Chem Soc 2001;123(18):4235–42.
[10] Wang GC, Jiang L, Zhou YH, Cai ZS, Pan YM, Zhao XZ, et al. J Mol [32] Perrichon V, Pijolat M, Primet M. J Mol Catal 1984;25:207–17.
Struct: Theochem 2003;634:23–30. [33] Hung WH, Bernasek SL. Surf Sci 1996;346:165–88.
[11] Newsome DS. T Catal Rev, Sci Engng 1980;21:275–318. [34] Jacobs G, Chenu E, Patterson PM, Williams L, Sparks D, Thomas G,
[12] Dry ME. Ind Eng Chem Prod Res Dev 1976;15:282–6. et al. Appl Catal A 2004;258:203–14.
[13] Feimer JL, Silveston PL, Hudgins RR. Ind Eng Chem Prod Res Dev [35] Dry ME. Appl Catal A 1996;138:319–44.
1981;20:609–15. [36] Linda J, Randall B, Snurr Q. Appl Catal A 2000;200:23–46.
[14] Shen WJ, Zhou JL, Zhang BJ. J Nat Gas Chem 1994;4:385–400. [37] Dumesic JA, Rudd DF, Aparicio LM, Rekoske JE, Trevino AA. The
[15] van der Laan GP, Beenackers AACM. Appl Catal A 2000;193:39–53. microkinetics of heterogeneous catalysis. Washington, DC: American
[16] Lox ES, Froment GF. Ind Eng Chem Res 1993;32:61–70. Chemical Society; 1993.
[17] Lox ES, Froment GF. Ind Eng Chem Res 1993;32:71.
[38] Froment GF, Bischoff KB. Chemical reactor analysis and design. 2nd
[18] Zimmerman WH, Bukur DB. Can J Chem Eng 1990;68:292–301.
ed. New York: Wiley; 1990 p. 94.
[19] Wang YN, Ma WP, Lu YJ, Yang J, Xu YY, Xiang HW, et al. Fuel
[39] Box GEP, Hunter WG. Statistics for experiments: an introduction to
2003;82:195–213.
design, data analysis, and model building. New York: Wiley; 1978 p.
[20] Yang J, Liu Y, Chang J, Wang YN, Bai L, Xu YY, et al. Ind Eng
631.
Chem Res 2003;42:5066–90.
[40] Hosten LH, Froment GF. In: Anderson JR, Boudart M, editors.
[21] van der Laan GP, Beenackers AACM. Catal Rev, Sci Eng 1999;41(3–
4):255–318. Catalysis science and technology, vol. 2. Berlin: Springer; 1981.
[22] Lox ES, Marin GB, de Graeve E, Bussier P. Appl Catal A 1988;40: [41] Satterfield CN. Heterogeneous catalysis in industrial practice. 2nd ed.
197–218. New York: McGraw Hill; 1991.