Você está na página 1de 17

W C

AA
PT
PA
LL
E
IY
DSS
I
A: GENERAL
ELSEVIER Applied Catalysis A: General 154 (1997) 155-171

Kinetic study of the reverse water-gas shift


reaction over CuO/ZnO/A1203 catalysts
M.J.L. Gin4s, A.J. Marchi, C.R. Apestegufa*
Instituto de lnvestigaciones en Cat61isis y Petroqulmica (INCAPE), UNL-CONICET Santiago del Estero 2654,
(3000), Santa Fe, Argentina

Received 17 June 1996; received in revised form 1 October 1996; accepted 3 October 1996

Abstract

The kinetics of the reverse water-gas shift (RWGS) reaction over CuO/ZnO/AI203 catalysts was
studied by use of CO2/H2 cycles, hydrogen chemisorption and catalytic tests performed in both
differential and integral plug flow reactors. The effect of the reactant composition on the reaction
rate was specifically studied by changing the/~H2//~CO2 ratio between 9.0 and 0.3. It was found that
different reagents become rate limiting depending upon pressure. While in a H2-rich region the rate
increases strongly with CO2 partial pressure and is zero order in hydrogen, under low/~H2//~co,.
ratios the reaction is less active and is strongly positive order in hydrogen and low order in carbon
dioxide. The experimental data were modeled by considering that the reaction proceeds through a
surface redox mechanism, copper being the active metal. A good agreement between experimental
and calculated data was obtained by assuming that in the redox mechanism either the dissociative
CO2 adsorption (H2-rich region) or both the CO2 dissociation and the water formation (H2-1ean
region) determine the rate of the overall reaction. Based on previous studies performed on copper
crystal surfaces, such a change in kinetics may be explained by assuming that under H2-rich
atmosphere a surface structural or phase transition occurs involving a change in reactivity with
respect to CO2 dissociation.

Keywords: Reverse water-gas shift reaction; Kinetics and mechanism;CuO/ZnO/AI203catalysts

1. Introduction

The water-gas shift (WGS) reaction (CO + H 2 0 --~ C O 2 -~- H2) is used
for ammonia synthesis, to adjust the H2/CO2 ratio in the product gas from the

* Corresponding author. Tel.: +54 42 555279; Fax: +54 42 531068/553727; E-mail:capesteg@fiqus.unl.edu.ar

0926-860X/97/$17.00 © 1997 Publishedby Elsevier Science B.V. All rights reserved.


PII S0926-860X(96)00369-9
156 M.J,L. Gin,s et aL /Applied Catalysis A: General 154 (1997) 155-171

steam reforming of hydrocarbons and to detoxify town gas. Ternary CuO/ZnO/


A1203 mixed oxides are preferentially used to catalyze this reaction at low
temperatures (443-523 K) and low concentrations of carbon monoxide [1,2].
The same catalysts are highly effective for methanol synthesis and the
reverse water-gas shift (RWGS) reaction. Although there are general agree-
ment that copper provides the active sites for catalysis [3-5], the mechanism
of the WGS reaction over Cu-based catalysts is still not certain. Whereas
several authors have postulated a Langmuir-Hinshelwood mechanism
involving a surface formate intermediate [6-9], a number of recent published
papers support the interpretation that the WGS reaction over Cu-based catalysts
occurs via an oxidative-reductive or "surface redox" mechanism [10-13]. By
reasoning that the rate of the RWGS reaction at low conversion and high
hydrogen partial pressure should be limited by dissociative CO2 adsorption if
the surface redox mechanism is correct, several authors have studied the kinetics
of the RWGS reaction [6,10,14-16]. However, Spencer [17] has recently
reported that the RWGS reaction is surprisingly fast over several metallic and
oxidic catalysts. By comparing previous experimental results dealing with the
forward and the reverse water-gas shift reactions, this author noted that the
observed values of activation energy of the RWGS reaction were inconsistent
with the values calculated from the standard enthalpy change and the activation
energies of the forward water-gas shift reaction. Spencer [17] suggested three
possible causes for the apparent failure of the principle of microscopic
reversibility: (i) there is a different mechanism for the reaction under forward
and reverse conditions away from equilibrium; (ii) there is a change in the
rate-determining step for the reaction under forward and reverse conditions
away from equilibrium; (iii) the reaction is not occurring on a common catalyst
and no simple relation between activation energies can be expected. Regarding the
last explanation, the author remarked that very different extents of surface
hydration can be expected under forward and reverse conditions because of the
very different partial pressures of water for small conversions. Ernst et al. [ 14] have
proposed that a change in the active phase of Cu-based catalysts can also be
expected when the catalyst is exposed to atmospheres containing very different
P°n2//~co 2 ratios. They found that over a clean Cu(ll0) single-crystal model
catalyst the reaction orders with respect to hydrogen and carbon monoxide
depend strongly on the/~2/P°co2 ratio and postulated that this change in kinetics
may be attributed to a hydrogen-induced surface phase transition that affects the
reaction rate.
In this paper we have studied the kinetics of the reverse water-gas shift
reaction over commercial and laboratory CuO/ZnO/AI203 catalysts by
changing the/~n2//~co2 ratio between 9.0 and 0.3. The aim was to establish the
effect the reactant composition has on the kinetics of the RWGS reaction over
typical commercial catalysts and to examine the implications for the reaction
mechanism.
M.J.L Gin& et al./Applied Catalysis A: General 154 (1997) 155-171 157

2. Experimental

2.1. Catalyst preparation and characterization

Two temary CuO/ZnO/A1203 (samples L-1 and L-2) and one binary CuO/A1203
(sample L-3) samples were prepared by coprecipitation of an aqueous solution
containing the metal nitrates (solution A) with a basic solution of sodium carbonate
(solution B), according to the method described in reference [18]. The two
solutions were added to 500 cm 3 of distilled water kept at 333 K contained in
a 2 1 beaker. Solution A was added by a pump at 1 cm 3 min- ~ and the simultaneous
addition of solution B was controlled so that the pH of the well stirred mixture was
maintained at 7. After solution A was exhausted, the resulting precipitate was
filtered, washed with hot water and dried at 363 K for 12 h. The precursor was then
decomposed in nitrogen for 8 h at 673 K. A commercial CuO/ZnO/A1203 catalyst
provided by ICI company and designated here as sample C- 1 was also used. All the
catalysts contained a similar Cu loading, between 30 and 35 wt %. Two samples of
alumina (Cyanamid Ketjen CK-300) and ZnO (Aldrich, 99.9%), respectively, were
employed as reference samples.
The metallic copper dispersions (Dcu = exposed Cu atoms/total Cu atoms) were
measured using the reactive chemisorption of N20 and frontal chromatography,
based on the method developed by Chinchen et al. [19]. The reactor was a 0.95 cm
O.D. stainless-steel tube of approximately 10 cm in length and was mounted inside
the chromatograph oven. Samples were reduced overnight in a 5% H2/He mixture
at 503 K and then purged in helium to sweep any adsorbed hydrogen on the
catalyst. Then a 4% N20/He mixture was passed over the catalyst at 333 K and the
length of the N2 plateau was determined using a thermal conductivity detector. The
length of the plateau corresponds to the N20 uptake and the metallic copper
dispersion was calculated assuming a molar stoichiometry N20/Cus--0.5, where
Cus implies copper atom on surface.
Powder X-ray diffraction patterns were collected on a Rich-Seifert diffract-
ometer using nickel filtered Cu Kct radiation. The chemical composition of the
samples was measured by atomic absorption spectrometry using a Perkin Elmer
3110 spectrophotometer. BET surface areas (Sg) were measured by N2 adsorption
at 77 K in a Micromeritics Accusorb 2100 sorptometer. The main characteristics of
the samples are given in Table 1.

2.2. Hydrogen chemisorption

The hydrogen uptake on the samples was measured by volumetry at 503 K in a


conventional vacuum apparatus equipped with a MKS Baratron pressure gauge.
Prior to performing gas chemisorption experiments, samples were reduced in a 5%
H2/N2 mixture for 4 h at 503 K. The pressure range of isotherms was 0-100 torr
158 M.J.L. Gin,s et aL/Applied Catalysis A: General 154 (1997) 155-171

Table 1
Main characteristics of the catalysts used in this work

Sample Sg (m2 g-l) Cu loading Cu/Zn (Cu+Zn)/A1 Identified compounds D¢u


(wt.%) (XRD) (%)

L-1 55 34.1 1.1 3.0 CuO, ZnO, ZnAlzO4(t) a 4.8


L-2 52 33.8 1.0 4.2 CuO, ZnO, ZnA1204 3.1
L-3 110 30.3 o~ 0.4 CuO 0.8
C-I 42 32.5 0.8 4.5 CuO, ZnO 5.0
A1203 160 . . . . .
ZnO 38 - - - ZnO

a (t) indicates that the crystalline phase was detected in trace amounts,

and the saturation value at 60 torr was used as a measure of the total hydrogen
uptake on the sample.

2.3. C02/H 2 cycles

Samples were submitted to successive CO2/H2 cycles and characterized by


employing thermogravimetric and chromatographic techniques. Thermogravi-
metric analysis (TGA) were carried out at 1 atm and 503 K with a Cahn 2000
electrobalance. Approximately 80 mg of sample was placed in a glass container
suspended vertically and reduced in a 5% Hz]N2 mixture at 503 K for 4 h. After
cooling the reactor down to room temperature, the sample was heated in nitrogen at
5 K min -a up to 503 K. Then the nitrogen flow was switched to a 5% CO2/N2
mixture and the weight change was registered until a constant weight was verified.
After purging the system with nitrogen, a 5% H2/N2 was passed over the samples
until steady state conditions were reached again. A similar procedure was
employed for studying the CO2/I--I 2 cycles by chromatographic techniques. The
experiments were performed in a fixed-bed flow reactor at 1 atm and 503 K.
Typical catalyst loadings were of 750 mg of sample. The evolved gases were
analyzed using a gas chromatograph equipped with a thermal conductivity detector
and a Porapak QS column.

2.4. Catalytic tests

The RWGS reaction was carried out at atmospheric pressure in a fixed-bed flow
reactor. The C O 2 / H 2 / N 2 reactant mixture was fed to the reactor by means of
electronic mass flow controllers. The catalytic bed dimensions were 0.8 cm ID and
0.2-0.3 cm high. The catalyst bed temperature was controlled to within 1 K. The
catalytic tests were performed in the 493-523 K temperature range using contact
time (W/F°co2) values between 2 and 40 g h mo1-1. Catalyst extrudates were
crushed and 0.2-0.5 g of the 0.35-0.42 mm fraction was separated and loaded to
the reactor. It was verified that using such a particle size range the reaction was
kinetically controlled under the operating conditions employed. Prior to catalytic
M.J.L. Ginis et al./Applied Catalysis A: General 154 (1997) 155-171 159

tests, the samples were first heated in N2 up to 353 K to remove water and then
were activated in a 5% H2/N2 reduction mixture at 503 K for 4 h. In an earlier study
[20], we found that such a reduction treatment reduces completely the CuO phase
without causing any sintering of the resulting metallic copper crystallites. The
reaction products were analyzed by on-line chromatography using a Hewlett
Packard 5890-11 chromatograph, equipped with a Carbosieve S-II column and a
thermal conductivity detector. The output signal was acquired and processed using
a Chrompack data station. Catalyst activity was measured by following the
concentration changes of CO2 and CO in the reactor exit gas. The presence of
water in the products was verified but not quantified. No methanol was detected in
the effluent. Carbon balances were closer to ~:5%. Data were collected every
30 min. Catalytic runs were performed at different W/F°co2 values so that the
catalytic reactor was operated either as a differential or an integral plug flow
reactor. The CO2 conversion (X) was always lower than 4% when the reaction was
carried out in the differential reactor, whereas the X value varied between 2 and
20% in the integral reactor. No deactivation was observed during the runs. Initial
tests repeated 8 times with catalyst C-1 lead to an estimated accuracy of +4% in the
activity measurement.

3. Results

3.1. C02/H2 cycles


Samples were submitted to several consecutive CO2/H2 cycles at 503 K. Fig. 1
illustrates the results obtained by thermogravimetry when two CO2/H2 cycles were
performed on sample L-2. Steady-state conditions were reached after about 5 rain
of introducing either CO2 or H2 gases. In the first cycle, the weight gain measured
under the CO2-containing atmosphere was practically equal to the weight lost in

0 H2

E
Q
0')
C

co2L
i,,,,"
0
W
W
m
=z /C02
v t
5 10 15 20

T i m e (rain)

Fig. l. CO2/I-I 2 cycles studied by thermogravimetry. Catalyst L-2, T=503 K, P = I atm.


160 M.J.L. Gin,s et al./Applied Catalysis A: General 154 (1997) 155-171

Table 2
Thermogravimetric results in CO2/H2 cycles and hydrogen chemisorption at 503 K

Sample Thermogravimetry H2 uptake H2 coverage


0amol g cat- l ) OH (H/Cus)
A W (g/ g c., ) O/Cus
L- 1 0.111 0.44 43.80 0.34
L-2 0.106 0.42 23.08 0.28
L-3 0.103 0.4l 5.58 0.29
C-1 0.118 0.47 39.08 0.32
A1203 - - 0.32 -
ZnO - - 0.51 -

the subsequent treatment with the H2/N 2 mixture so that the initial weight of the
sample was recovered at the end of the cycle. On the other hand, the weight
changes measured in the second cycle following the successive addition of carbon
dioxide and hydrogen were practically the same as compared to those correspond-
ingly obtained in the first cycle. Similar qualitative behavior was determined for
samples L-1, L-3, and C-1, although it was observed that the period length needed
for reaching steady-state conditions increased with increasing copper surface
areas. Quantitative results are given in Table 2, where AW (g/gcu~) is the total
weight increase measured under the CO2-containing atmosphere referred to the
amount of copper on the surface. It is observed that AW values between 100 and
120 g/gcu~ were obtained for the samples used in this work.
The gases evolved during the successive CO2/H 2 cycles were analyzed by
chromatography. In all the cases, only carbon monoxide was formed when the
CO2/N2 mixture was fed over the sample, whereas water was the only product
detected in the effluent following the consecutive exposition of the sample to a H2
stream. In Fig. 2 we have represented the conversion to carbon monoxide (Xco)
and to water (Xrho) as a function of time during the first CO2/H2 cycle performed
on sample L-2. The addition of the COzfN 2 mixture caused the immediate

~ O-Xco
/ - X.o

gli
0 40 80 120 160 200
Time (min)

Fig. 2. Evolved gases during CO2/H2 cycles. Catalyst L-2, T--503 K, P = 1 atm.
M.J.L. Gin's et al./Applied Catalysis A: General 154 (1997) 155-171 161

formation of carbon monoxide. The conversion to CO reached a maximum and


then diminished continuously until the complete disappearance of carbon mon-
oxide in the exit gases in about 100 min. After purging with nitrogen, the 5% H2/N2
mixture was admitted to the reactor. Similarly to the evolved CO trace, the XH20
value diminished with time after reaching a maximum ab-initio of the run. A
similar qualitative behaviour was observed for samples L-l, L-3, and C-1.

3.2. Hydrogen chemisorption

The hydrogen uptakes at 503 K measured on ternary, binary, and reference


samples are presented in Table 2. It is observed that the hydrogen uptake on both
alumina and zinc oxide was negligible. We then assumed that the hydrogen
chemisorption on Cu-based samples takes place essentially on the metal copper
surface. Thus, from the hydrogen uptake values given in column 4, and by
assuming that hydrogen chemisorbs on copper dissociatively with a stoichiometry
H/Cus=l, we calculated the hydrogen coverage (0n) on the samples. In all the
cases, a value of OH ~ 0.30 atom H/atom Cus was obtained (Table 2).

I
I
I

7.5

0
01
J[" 6.5
m
0
E
L.o

t-
"i" 5.5

I
!

4,5 [ I__
0.0 1,0 2.0 3 .0 4.0

-In Pco (atm)


Fig. 3. Dependence of the rate of the RWGS reaction upon CO2 partial pressure at two fixed H2 pressures. (IS])
/~n2 = 0.2 atm; (@) P°n2 = 0.3 atm. Catalyst C-l, T=503 K, P = l atm.
162 M.J.L Ginds et al./Applied Catalysis A: General 154 (1997) 155-171

3.3. Catalytic testing


The activity tests were performed using catalyst C-1. The experimental data
obtained with the differential reactor were interpreted by considering a power-law
rate equation:

ro = k(P°o2)~(P%) ~ (1)

Figs. 3 and 4 illustrate the results obtained at 503 K. In Fig. 3, initial rate r 0 (mo]
CO2 h - l gcat.- 1) was represented as a function of/~co2 in a logarithmic plot for two
values of P°H2 (0.2 and 0.3 atm) whereas Fig. 4 shows the In ro vs. In/~n2 plot for
/~co2= 0.1 atm. The results presented in Fig. 3 and Fig. 4 allowed us to distinguish
between two regions separated by a/~n2//~co2 value of approximately 3. While in
the region of /~n2//~co2 > 3 the rate did not vary with ~n2 (Fig. 4), for
POH2//~CO2 < 3 the ro value depended on partial pressures of both H2 and CO2
(Fig. 3). Parameters k, a, and fl of Eq. (1) were determined graphically in both
regions. For POH~//~CO2> 3, the rate constant and the orders with respect to CO2
and H2 were k ~ 0.037 mol h -1 gcat - 1 atm -(~+~), a ~ 1.1 and/3 ~ 0, respectively,
whereas values ofk _-_-0.015 mol h -1 gcat - 1 atm -(~+13), a ~ 0.3 and/3 __--0.8 were
determined in the P%/P~o2 < 3 region.

7.5

BI []
m
O 6.5
E
0 0
O
PH/Pco= >
c

..~.,~/ PH,/P;o, < 3


J /
I
/
I
/
J [
5.E
0.0 1.0 2.0 3.0

-In P'H~(atm)
Fig. 4. Dependence of the rate of the RWGS reaction upon H 2 partial pressure at a fixed CO2 pressure. Catalyst
C-l,/~co2 =0.1 atm, T=503 K, P = I atm.
M.J.L Gin& et al./Applied Catalysis A: General 154 (1997) 155-171 163

50 513 K 523 K
503 K
493 K
'40

~ " 30

~2o
lO

0.00 0.05 0.10 0.15 0.20

Conversion X(%)

Fig. 5. Integral plug flow reactor data for catalyst C-1 Points, experimental results; solid lines, model
predictions. P= 1 atm,/~2/P°co2 =6,

In order to obtain more insight on the mechanism and kinetics of the RWGS
reaction, several additional catalytic runs were carried out in an integral reactor.
The experiments were performed at 1 atm in the 493-523 K range using different
W/l~co 2 values and a/~n2//~co2 ratio of 6. The experimental data are shown in
Fig. 5, where the CO2 conversion was plotted as a function of contact time.

4. Discussion

Although few papers have been published on the mechanism of the reverse
water-gas shift reaction, the information is coincident in that the reaction proceeds
through the same surface redox mechanism used to describe the kinetics of the
forward WGS reaction, the metallic copper being the active sites [ 10,14,15]. This
is a regenerative mechanism whereby copper is successively oxidised and reduced
by CO2 and H2 according to the following simplified scheme:
C02(g ) -+- 2CU~s) ~ CO(g) -4- Cu20(s ) (2)

H2(g) q- Cu20(s ) ~ HzO(g) q- 2Cu~s) (3)


Our results obtained from the CO2/H2 cycles experiments support this mechanism.
In fact, the evolved gas analysis results (Fig. 2) showed that upon CO2 addition to
the reactor only CO was formed, which is in agreement with reaction (2), whereas
the subsequent addition of hydrogen produced only water as it is precisely
predicted by reaction (3). On the other hand, quantitative results obtained by
gravimetry (Fig. 1 and Table 2) allowed us to calculate the weight changes
following the consecutive addition of carbon dioxide and hydrogen. It was found
that the total weight gain measured following the addition of the CO2-containing
164 M.J.L. Ginds et al./Applied Catalysis A: General 154 (1997) 155-171

atmosphere corresponded to a O/Cus atomic ratio of about 0.4-0.5, which is close


to the value expected according to reaction (2) for the saturation coverage of
oxygen adatoms produced by CO2 adsorption (O/Cus---0.5). Besides, such a mass
gain was practically the same as the mass lost by the catalyst during the
consecutive addition of hydrogen, and this result is in agreement with the overall
redox mechanism represented by reactions (2) and (3). Thus, on the basis of
previous literature and our own results, the present kinetics of the RWGS reaction
will be discussed in terms of the surface redox mechanism. Such a mechanism may
be represented by the following elementary steps:

CO2(g) ÷ 2L ~ CO.L + O.L

CO.L ~ C O ( g ) + L
H2(g) + 2L ,~ 2H.L
2H.L + O.L ~ H20.L + 2L
HzO.L ~ HzO(g)+ L
where L represents the vacant active sites. We assume that the sites are identical
and that there is no interaction between adsorbed species.
Previous work [21] on adsorption/desorption equilibria for both carbon mon-
oxide and water on copper have shown that these two molecules have high sticking
probabilities, very low dissociation probabilities and small heats of adsorption. We
can reasonably assume then that the coverage of these adsorbates is negligible at
493-523 K and at the partial pressures used in our RWGS reaction. Nakamura et al.
[12] calculated coverages for carbon monoxide and water of 1.7% and 0.1% on
Cu(ll0) at the RWGS reaction conditions. On the other hand, and since the
dissociative chemisorption of hydrogen on clean, polycrystalline copper surfaces is
highly activated [22], this reaction has been often omitted in the step-wise
mechanism of the forward WGS reaction. However, several papers have recently
reported that the dissociative chemisorption of hydrogen on supported copper
catalysts is rather a facile process [23,24]. Our results regarding the hydrogen
chemisorption on ternary Cu/ZnO/AI203 and binary Cu/AlzO3 samples showed
that hydrogen is effectively chemisorbed on copper at 503 K, the hydrogen
saturation coverage at 60 torr being approximately 30% (Table 2). This value
is in line with that obtained by Ernst et al. [14] over clean Cu(110). These authors
determined activation energies of 14.3 kcal tool -1 and 13.0 kcal mo1-1 for the
adsorption and desorption of hydrogen, respectively, and calculated that the
hydrogen coverage would be about 20% of saturation at PH2=100 torr and
T=673 K. Salmi and Hakkarainen [9] measured the chemisorption isotherms of
hydrogen on commercial Cu/ZnO/A1203 catalysts at 76 torr and 523 K and found
that the amount of adsorbed hydrogen was 7% of the BET monolayer capacity.
This value was about one order of magnitude greater than the amount of carbon
monoxide adsorbed under the same operating conditions. Also, our results showed
M.J.L. Ginds et al./Applied Catalysis A: General 154 (1997) 155-171 165

that the hydrogen chemisorption on both alumina and zinc oxide samples at 503 K
was negligible (Table 2), thereby indicating that the gas was selectively adsorbed
on the metallic copper phase. Although it has been suggested that the dissociative
chemisorption of hydrogen may occur on the zinc oxide phase followed by reverse
spillover on copper, our results indicate that the hydrogen chemisorption on
supported copper catalysts cannot be attributed to support effects. A similar result
was obtained by Bennett et al. [24]. These authors studied the adsorption of
hydrogen on a variety of oxide supported copper catalysts by in situ NMR and
concluded that the nature of the support was not critical in enhancing the
dissociative chemisorption of hydrogen.
Based upon the arguments presented in the above discussion, the following
simplified mechanism is considered to account for the RWGS reaction. The
nomenclature used is given in the Appendix.
CO2(g ) Jr- L ~ CO(g) -+- O.L r = k l P c o 2 C L - k'IPcoCoL (4)

c2
H2(g ) + 2L ~ 2H.L K 2 -- C2LPH2

PH20 C3
2H.L + O.L ~- H20(g) + 3L K3 -- C2HLCO~
L (6)

In a previous work, Nakamura et al. [25] found that the rate of the dissociative
adsorption of carbon dioxide on Cu(ll0) is similar to the rate of the RWGS
reaction, which suggests that the CO2 dissociation is the rate-determining step in
the RWGS mechanism. Based on this result, and in order to model our results
obtained with the integral reactor under a Hz-rich atmosphere, we assumed that
step (4) is rate determining in the reaction sequence represented by steps (4) to (6).
By employing Langmuir-Hinshelwood kinetics, we deduced the following rate
equation:

klLo~co2 [P°H2(1-X) 2 ~c-~X2]


= (7)
P°a(1 - X) + x/~/~u21.5(1 - X) '5 + K2K3
/~CO2X

where X is the conversion of CO2 and Lo = (CL + COL + CHD is the concentration
of active sites of fresh catalyst. Thus, and by considering that the design equation of
an isothermal integral reactor is:

FO
CO2
-- fax
--
F
(8)
166 M.J.L. Ginds et al./Applied Catalysis A: General 154 (1997) 155-171

Table 3
Kinetic parameters of the redox mechanism determined by modeling experimental data obtained with the
integral plug flow reactor.Catalyst C-l, P = I atm; P°n2//~co2=6
Temperature (K) kiLo K2 (atm 1) ro × 103 ~'0x 103
(mol h i gcat-I atm-2) (mol h -1 gcat-1) (tool h l gcu i)

493 0.02004-0.0015 0.060-1-0.008 2.4 7.17


503 0.04874-0.0044 0.051-1-0.004 5.8 17.7
5l3 0.07114-0.0070 0.041 -t-0.007 8.6 26.3
523 0.12874-0.0077 0.0394-0.002 15.5 47.8

we used Eqs. (7) and (8) for modeling the experimental data. The model parameter
estimation was performed by using a least-squares regression which minimizes the
objective function F = Z [ ( W / F ° o 2 ) e x p - (W/l~co2)calc] 2. AS it is shown in Fig. 5,
a good agreement was obtained between the experimental values and the model
predictions. The values of parameters K2 and klL0 obtained at four temperatures are
given in Table 3. From Eq. (7) we deduced the expression of initial rate ro:

ro --
kl P°o2 (9)

and by introducing the values of K2 and k l / ~ of Table 3 in Eq. (9) we calculated the
corresponding ro values (Table 3). The r0 value determined at 503 K was 5.8 x
10 .3 mol h - 1 gcat-1 which is similar to that measured in the differential reactor
under the same operating conditions (ro =4.5 x 10 - 3 m o l h -1 gcat-1). V a n
Herwijnen and de Jong [6] found a rate of about 2.7 x 10 .3 mol h -1 gcat - 1 at
508 K and/~H2/P~o2 = 6 for the RWGS reaction over a binary CuO/ZnO catalyst
containing 32.2% CuO. Stirling et al. [16] measured the activity of the RWGS
reaction over monophasic CuxZn(1_x)O solid solutions containing between 0.7 and
5.8% tool Cu in the 495-560 K temperature range and at POH~//~CO2 = 11. They
tested eight samples and found that the specific initial rate ~0 (tool h - 1 gcu- 1) was
between 2 x 10 - 3 and 15 x 10 . 3 mol h - 1 gcu -1 for all the samples at 523 K.
These values compare well with the specific reaction rate determined here at the
same temperature and pO/P0co ~ = 6 (Table 3). On the other hand, we plotted the
ln(klLo) values obtained from Table 3 as a function of 1/T for calculating the
apparent activation energy of the RWGS reaction via an Arrhenius-type function
(the plot is not shown here). From the slope of the resulting linear plot we obtained
Ea =24+1.9 kcal mo1-1, which is consistent with those values found by other
authors (18-27 kcal mo1-1) [6,14,17]. Overall, we observe that the kinetic values
determined in the present paper with the integral reactor for a/~H2//~CO2 = 6 ratio
compare favourably with those reported in previous work on copper-based
catalysts under hydrogen-rich atmospheres. Moreover, the experimental data
obtained with the differential reactor showed that in the POH2//~CO2> 3 region,
the apparent orders with respect to H2 and CO2 are about 0 and 1, respectively.
M.J.L. Ginds et al./Applied Catalysis A: General 154 (1997) 155-171 167

These are precisely the approximate orders predicted by Eq. (9) since according to
the K2 value determined using the integral reactor (Table 3), 1 >> x/K2P°m and the
initial rate expression reduces to ro ~ kiLo/~co2" In summary, our v.---resultsobtained
with both differential and integral plug flow reactors show that the kinetics of the
RWGS reaction for P°2/l~co 2 > 3 is well represented by the redox mechanism
described by reactions (4) through (6), the carbon dioxide dissociation being the
rate-limiting step.
However, the experimental data showed that the RWGS kinetics changes when
H2-1ean atmospheres are used. In fact, when the/~H2/P°co2 ratio was lower than
approximately 3 the reaction was dependent on partial pressures of both the H2 (fl
- 0.8) and the CO2 (a ~ 0.3) and the rate constant was clearly lower than that
obtained under H2-rich atmospheres (Table 3). A simple interpretation of these
results is that when OH decreases below a certain critical value the active surface
phase become less active and, as a consequence, the rate-limiting step in the redox
mechanism changes. To test this interpretation, we have employed a quasi-
stationary state approximation for obtaining the rate of the overall reaction
proceeding along the single path of the sequence represented by reactions (4)
through (6). Besides, and taking into account that in our H2-1ean conditions both
PH2 and/(2 are very small, we omitted step (5) in considering our kinetics. The
RWGS reaction mechanism results therefore represented by a sequence of two
reversible elementary steps:
f
CO2(g) + L ~ CO(g) + O.L rl = klPco2CL - k l P c o C o L (10)
H2(g) + O.L ~- HzO(g) + L r2 = k4PH:CoL - k'4PH2oCL (1 1)
Since the quasi-stationary state approximation implies that r -- rl = r2, from Eqs.
(10) and (11) we obtain:
CL (kl k4Pco2PH2 - k'1kt4PcoPH~o )
r = (12)
k4PH2 + klPco
Eq. (12) can be written as:
,k4Lo(P o P - P o/K)
r~
klPco2 + k4PH2 + (k] + k])Pco
1 to co2 [eo2(1 - x) 2 _ X2 /Kl
= , co2(1 - X) + ka H (l - X) + (Z, + (13)

where Lo is now Lo = (CL + COL). From Eq. (13) we obtain the initial reaction rate:
k,LoP°o21~H2
(14)
ro = (kl/k4)~c02 -[- ~H2

If kl<<ka, Eq. (14) reduces to r0 ~ klLoP°co2, which is consistent with the rate
found under H2-rich conditions by assuming that the dissociative adsorption of
168 M.J.L. Gin& et al./Applied Catalysis A: General 154 (1997) 155-171

CO2 was rate-determining. Eq. (14) can be rearranged as

k4 POH2 "1
ro =k4LoP°2 -+-k~I--S~I
P°-- (15)
1 k4 ~= J

and further simplified by the use of the approximation of the term ax/(1 +ax) by the
term bx n where b and n are constants with n constrained to 0< n <l. Thus Eq. (15)
becomes

\--~H2~n
}
ro = b k 4 Z o e ° 2 ( P ~ O z
(16)

which provides a power rate law expression of the RWGS rate:


ro = k" ( l~c% )n(P°2) l-" (17)
Eq. (17) predicts an opposing relationship between the CO2 and H2 orders of
reactions. To test the validity of Eq. (17), we can use the form of Eq. (17) to
calculate the order in H2 given the experimental order in CO2 (n=oL=0.3). The
hydrogen order of reaction is (i-n)--0.7, which is close to that obtained experi-
mentally (/3 ~ 0.8).
In brief, our results show that the kinetics of the RWGS reaction over Cu/ZnO/
A1203 catalysts is complicated under our reaction conditions in that different
reagents become rate-limiting depending upon the reactant pressure. The reaction
shifts from nearly first order in carbon dioxide to first order in hydrogen as the
/~H2//~CO2 ratio decreases. Among the few publications on the RWGS reaction on
Cu-based catalysts only van Herwijnen and de Jong [6] have studied the kinetics on
commercial CuO/ZnO/A1203 catalysts by varying the reactant composition in a
wide range. In accordance with our results, they found that as P ° 2//~co= increases
the rate which is initially controlled by the/~co2 value becomes limited by the
hydrogen partial pressure. Ernst et al. [14], who studied the kinetics of the RWGS
reaction over Cu(110) single-crystals model catalysts between 573 and 723 K, also
found that, whereas in a H2-rich region the rate increases strongly with CO2 partial
pressure and is independent on the hydrogen pressure, at low POH2//~CO2 ratio the
rate is strongly positive order in hydrogen and low order in carbon dioxide. To
explain these results, and based on that high hydrogen coverage may cause the
well-known (1 × 2) missing-row reconstruction on metallic copper [26], Ernst et
al. [14] postulated that under Hz-rich atmospheres a surface structural or phase
transition occurs involving a change in reactivity with respect to carbon dioxide.
Surface phase transitions are known to cause changes in reactivity [27,28]. A
similar explanation may be invoked to interpret the catalytic behaviour observed
here over Cu/ZnO/A1203 samples. We found that the hydrogen coverage must be a
substantial fraction of the monolayer at 503 K under H2-rich atmospheres
M.J.L. Gin,s et al./Applied Catalysis A: General 154 (1997) 155-171 169

(Table 3). Since the heats of adsorption of CO2, HzO, and CO on metallic copper
are all rather small [29], the coverages of these molecules at 503 K must be
negligible. The only remaining species in the mechanism which could be at high
coverage is O.L, which is expected not to inhibit H2 adsorption up to a oxygen
coverage of about 40% [30]. Several studies performed over copper crystal
surfaces have reported reconstruction of the copper surface under H2-rich atmo-
spheres. Rieder and Stocker [31] have found that room temperature adsorption of
hydrogen on Cu(ll0) causes a subsurface reconstruction of the copper atoms.
Recent studies using NMR [32] confirmed the existence of this subsurface
hydrogen, which is observed only when copper is covered with an overlayer of
oxygen. The presence of subsurface hydrogen which is trapped by the recon-
structed copper surface brought about by the oxygen overlayer has also been
postulated by Waugh [33]. This fully reconstructed copper surface at high
P°2/P°o2 ratios would be more reactive to dissociative CO2 adsorption. Our
results suggest that the POH2//~CO2ratio needed for achieving the critical hydrogen
coverage to trigger the surface reconstruction from the less active to more active
phase would be >3, but this transition might obviously be temperature dependent.
To summarize, starting at high PH2/Pco2
0 0 ratios the reaction proceeds on a high
coverage of hydrogen and the rate is limited by the rate of dissociative CO2
adsorption. As/~co2 increases the rate increases, till it is first order in CO2, and the
hydrogen coverage drops below some critical OH where the copper surface
reconstructs to a less active phase. The rate constant diminishes and the reaction
is now limited by the H2 adsorption becoming strongly positive order in hydrogen.
These changes in surface reactivity and in reagents which are rate-limiting are
interpreted in the formulation of our redox mechanism by considering that either
the dissociative CO2 adsorption (high OH) or both the CO2 dissociation and the
water formation (low OH) determine the rate of the overall reaction.

Acknowledgements

Support of this work by the Consejo Nacional de Investigaciones Cientificas y


Trcnicas (CONICET), Argentina, and the Universidad Nacional del Litoral, Santa
Fe, Argentina, is gratefully acknowledged.

Appendix

Nomenclature

CL = concentration of vacant active sites (mol g-l)


CiL = concentration of active sites covered by species i (mol g 1)
Ea = apparent activation energy (kcal mol -l)
F°o2 = feed rate of CO2 (mol h -1)
170 M.J.L. Gin's et al./Applied Catalysis A: General 154 (1997) 155-171

k = apparent rate constant of the RWGS reaction (mol h-~ gcat - 1 atm -(~+~))
kl, k4 = forward reaction constants of reactions (4) and (11), respectively
(a!m -1, h -1)
kl, k4 = reverse reaction constants of reactions (4) and (11), respectively
(atm -I h -1)
K = equilibrium constant of the RWGS reaction (dimensionless)
K2 = equilibrium constant of the dissociative hydrogen adsorption (atm -1)
K3 = equilibrium constant defined in Eq. (6) (atm)
L = vacant active sites
L0 = concentration of active sites of fresh catalysts (mol g-t)
r = reaction rate (mol h - 1 gcat-1)
r0 = initial reaction rate (mol h - 1 gcat-1)
~0 = specific initial reaction rate (mol h -1 gcu -1)
Pco2, PIJ2, Pco -----partial pressures of carbon dioxide, carbon monoxide and
hydrogen, respectively (atm)
/~co2, P°n2 --- partial pressures of carbon dioxide and hydrogen in the feed,
respectively (atm)
W = catalyst loading (g)
X = conversion of carbon dioxide
Xco,XIJ2O --- conversion to carbon monoxide and water, respectively
o~, /3 = reaction orders with respect to carbon dioxide and hydrogen,
respectively

References

[1] D.S. Newsome, Catal. Rev. Sci. Eng., 21 (1980) 275.


[2] P.N. Hawker, Hydroc. Process., 61 (1982) 183.
[3] C.T. Campbell and K.A. Daube, J. Catal., 104 (1987) 109.
[4] G. Petrini and F. Garbassi, J. Catal., 90 (1984) 113.
[5] M.J.L. Gin,s, N. Amadeo, M. Laborde and C.R. Apestegufa, Appl. Catal. A, 131 (1995) 283.
[6] T. van Herwijnen and W.A. de Jong, J. Catal., 63 (1980) 83.
[7] G.A. Vedage, R. Pitchai, R.G. Herman, K. Klier, Proc. 8th International Congress on Catalysis, Berlin,
1984, Vol. 2,, Verlag Chimie, Weinheim, 1984, p. 47.
[8] T. Takagua, G. Pleizer and Y. Amenomiya, Appl. Catal., 18 (1985) 285.
[9] T. Salmi and R. Hakkarainer, Appl. Catal., 49 (1989) 285.
[10] R.A. Hadden, H.D. Vandervell, K.C. Waugh, G. Webb, in M.J. Phillips, M. Ternan (Eds.), Proc. 9th Int.
Cong. Catal, Calgary, 1988, Vol.4, The Chemical Institute of Canada, Ottawa, 1988, p. 1835.
[11] G.C. Chinchen and M.S. Spencer, J. Catal., 112 (1988) 325.
[12] J. Nakamura, J.M. Campbell and C.T. Campbell, J. Chem. Soc. Faraday Trans., 86 (1990) 2725.
[13] C.V. Ovesen, P. Stoltze, J.K. Norksov and C.T. Campbell, J. Catal., 134 (1992) 445.
[14] K.H. Ernst, C.T. Campbell and G. Moretti, J. Catal., 134 (1992) 66.
[15] S. Fujita, M. Usui and N. Takezawa, J. Catal., 134 (1992) 220.
[16] D. Stirling, ES. Stone, M.S. Spencer, in L. Guczi, F. Solymosi, P. T&tnyi (Eds.), Part B, Proc. 10th Int.
Cong. Catal., Budapest, 1992, Elsevier/Akad~mia Kiad6, Amsterdam/Budapest, 1993, p. 1507.
[17] M.S. Spencer, Catal. Lett., 32 (1995) 9.
[18] A.J. Marchi, J.I. Di Cosimo and C.R. Apestegufa, Catal. Today, 15 (1992) 383.
[19] G.C. Chinchen, C.M. Hay, H.D. Vandervell and K.C. Waugh, J. Catal., 103 (1987) 79.
M.ZL. Gin,s et al./Applied Catalysis A: General 154 (1997) 155-171 171

[20] M.J.L. Gin6s and C.R. Apesteguia, Latinam. Appl. Res., 25(4) (1995) 215.
[21] C.T. Campbell and K.A. Daube, J. Catal., 104 (1987) 109.
[22] J. Pritchard, T. Catterick and R.K. Gupta, Surf. Sci., 53 (1975) 1.
[23] J.M. Campbell and C.T. Campbell, Surf. Sci., 259 (1991) 1.
[24] A. Bennett, J.B. Cobb, B.T. Heaton, J.A. Iggo, G. Chinchen, Proceedings Europacat-II Congress,
Maastricht, 1995, p. 691.
[25] J. Nakamura, J.A. Rodriguez and C.T. Campbell, J. Phys. Condens. Matter, 1 (1989) 149.
[26] R. Spitzl, H. Niehus, B. Poelsema and G. Comsam, Surf. Sci., 239 (1990) 243.
[27] N. Eiswirth and G. ErR, Surf. Sci., 177 (1986) 90.
[28] D. Lackey, J. Schott and J.K. Dass, J. Electron Spectrosc. Relat. Phemom., 54/55 (1990) 649.
[29] W.D. Clendening, J.A. Rodriguez, J.M. Campbell and C.T. Campbell, Surf. Sci., 216 (1989) 429.
[30] B.E. Haydon and C.L.A. Lamont, J. Phys. Condensed Matter, 1 (1989) 33.
[31] K.H. Rieder and W. Stocker, Phys. Rev. Lett., 51 (1986) 2548.
[32] A. Bendada, G.C. Chinchen, N. Clayden, B.T. Heaton, J.A. Iggo and C.S. Smith, Catal. Today, 9 (1991)
129.
[33] K.C. Waugh, Catal. Today, 15 (1992) 51.

Você também pode gostar