Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

i

ELSEVIER Applied Catalysis A: General 142 (1996) 97-122

A P P L I E D C A T A L Y S S I A: G E N E R A L

Catalytic reforming of methane with carbon dioxide over nickel catalysts II. Reaction kinetics
Michael C.J. Bradford, M. Albert V a n n i c e *
The Pennsylvania State University, University Park, PA 16802-4400, USA

Received 16 October 1995; revised 22 January 1996; accepted 24 January 1996

Abstract
The reforming of methane with carbon dioxide was studied over nickel supported on S i t 2, T i t 2, MgO and activated carbon. Specific activities on a turnover frequency basis were in the order: N i / T i O 2 > N i / C > N i / S i O 2 > N i / M g O . Interestingly, a 2-fold increase in activation energy for this reaction was observed over N i / T i O 2 after several hours time on stream. The reverse water-gas shift reaction was found to be close to thermodynamic equilibrium over all catalysts. Partial pressure dependencies were obtained with the N i / C and N i / S i O 2 catalysts at 723 K for comparative purposes only, but a more thorough kinetic analysis was made with the N i / M g O and N i / T i O z catalysts, which were shown previously to strongly inhibit carbon deposition. Partial pressure dependencies were obtained at 673, 698, and 723 K for N i / T i O 2 and at 773, 798, and 823 K for N i / M g O . In situ DRIFTS studies clearly showed the presence of both linear and bridged carbon monoxide adsorption on N i / S i O 2 under reaction conditions; however, adsorbed carbon monoxide could not be identified on N i / T i O 2. A reaction model for CH4-CO 2 reforming, based on CH 4 activation to form CH x and CHxO decomposition as the slow kinetic steps, successfully correlated the rate data.
Keywords: Kinetics; Methane; Carbon dioxide; Reforming; Nickel; Titania; Silica; Magnesium oxide; Carbon

1. I n t r o d u c t i o n The reaction equilibrium for the production and carbon dioxide CH 4 + CO 2 ~


2H 2 + 2CO

of synthesis gas from methane

* Corresponding author. Tel. (+ 1-814) 8634803, Fax. (+ 1-814) 8657846. 0926-860X/96/$15.00 1996 Elsevier Science B.V. All rights reserved PII S0926- 860X(96)00066-X

98

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A." General 142 (1996) 97-122

is typically influenced by the simultaneous occurrence of the reverse water-gas shift (RWGS) reaction
C O 2 -{- H 2 ~ C O + H zO

resulting in H2/CO ratios less than unity. This reaction sequence has thus received recent attention for the production of synthesis gas for Fischer-Tropsch synthesis reactions. Gadalla and Bower [1] have clearly shown that the production of alkanes from synthesis gas in the presence of the water-gas shift reaction requires H2/CO ratios of less than unity. Rostrup-Nielsen and Hansen [2] have shown that the water-gas shift reaction is extremely rapid under typical methane reforming conditions. For this reason, the RWGS is frequently assumed to be at equilibrium in the kinetic analysis during steam reforming of methane [3]. Although a majority of authors have reported that the RWGS reaction operates very close to thermodynamic equilibrium during C H 4 - C O 2 reforming [4,5], some authors have reported otherwise [6]. Specifically, Gesser et al. have shown that the extent of RWGS equilibrium on a tungsten wire catalyst was a function of feed gas space velocity [7]. Although CH4-CO 2 reforming has been studied extensively (see Ref. [8] and references therein), there has been little emphasis on obtaining a more fundamental understanding of the reaction kinetics. The elementary steps which comprise the reaction mechanism remain unclear, and there currently exists no general reaction mechanism in the literature. To our knowledge, only four kinetic models have been proposed in the literature (Table 1). The earliest report of a Langmuir-type rate expression for CH4-CO 2 reforming was presented by Lewis et al. almost half a century ago for a Cu/SiO 2 catalyst [9]. However, they did not provide the reaction mechanism from which their rate expression was derived, they performed their experiments within a very limited experimental
Table 1 Reported kinetic models for CH 4 - c o 2 reforming Model
' t kPcH x

Catalyst
4( Pco2 -}- PH20)
,

Ref. [9] [11,12]

]2

Cu/SiO2 Ni foil

kPcH4 PH20 1+ a + bPco


PH~

kR Kc% KcHnPCO2PcH4

Rh/A1203

[13]

(1 + KcozPc02 + KcH4PcH4)2
aPcH 4 *202

i'

(Vco~ + bp~o2 + cec.~)

Ni/A1203

[16]

Ni/CaO-AI203

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

99

regime, and they assumed that, with the exception of the rate constant, all parameters were temperature independent. The latter assumption implies that the heats of adsorption of CO 2, H 2 and H20 are zero and is inconsistent with the reported heat of adsorption of CO 2 on C u / S i O 2 [10]. Bodrov et al. studied CH4-CO 2 reforming on a nickel foil catalyst [11] and fit their data to an expression which they originally derived to explain the kinetics of C H 4 - H 2 0 reforming [12]. They suggested that a common reaction mechanism exists for both reactions, a hypothesis more recently alluded to by Rostrup-Nielsen and Hansen [2]. Richardson and Paripatyadar provided a model derived from a Langmuir-Hinshelwood approach involving redox mechanisms which fit the data well and provided the proper temperature dependencies of the rate and adsorption constants [13]. However, they did not present the reaction mechanism from which they derived their model, and the heats and entropies of adsorption obtained from their adsorption parameters do not satisfy the adsorption guidelines reported by Vannice et al. [14,15], primarily because the adsorption entropies are too small, thus suggesting an inconsistency in their model. Finally, Zhang and Verykios have provided a rate expression, derived from a Langmuirian model assuming that methane dissociation was the rate determining step [16]. Although the model supposedly fit the experimental data reasonably well, the authors did not provide the values for the adsorption and kinetic parameters in their paper; thus, it was not possible to evaluate the validity of their model.

2. Experimental
Preparation and characterization of the four supported nickel catalysts studied in this investigation has been discussed previously [8]. All experiments were carried out under an absolute pressure of ca. 740 Torr using between 5 and 100 mg of catalyst, which was adjusted to keep conversion far from thermodynamic equilibrium. To determine the apparent activation energies, the reforming reaction was studied with a feed composition of C O 2 / C H 4 / H e = 1 / 1 / 1 . 8 and a total feed flow rate of 20 sccm (space velocity ca. 10000-200000 h -J) over a temperature range of 673 to 823 K. The partial pressure dependencies were determined by maintaining 200 Torr of one reactant and varying the other reactant between approximately 40 and 400 Tort pressure. A balance of helium was adjusted to maintain a total gas flow rate of 20 sccm and a total absolute pressure of one atm during the partial pressure studies. The reaction system used to perform these measurements has been discussed previously [8]. Although the gas chromatograph in this system was capable of separating all five reaction gases for quantitative analysis, the rate of water formation was not measured but estimated from an overall material balance. This was done to avoid possible experimental problems associated with water condensation from the carrier gas and the large uncertainty in the thermal response value for water.

100

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A." General 142 (1996) 97-122

An N2-purged FTIR spectrometer (Mattson Instruments, RS-10000) equipped with a DRIFTS cell (Harrick Scientific, HVC-DR2) and a praying mantis mirror assembly (Harrick Scientific, DRA-2C0) was used to study the reforming reaction from 673 to 773 K under identical experimental conditions as those used in kinetic experiments. Spectra were obtained from 700 to 4000 c m - ~ with a resolution of 4 cm-J Because design of the apparatus was limited to a maximum operating temperature of 773 K, in situ pretreatment and reaction was not possible for the N i / M g O catalyst. 3. Results
3.1. Catalytic activity

The net initial activity, met, after 30 min on stream of N i / T i O 2 for carbon monoxide production at 723 K was determined as a function of percent equilibrium conversion (Fig. 1) by adjustment of the catalyst mass. Weisz criterion calculations indicated that the observed decrease in reaction rate with increasing proximity to thermodynamic equilibrium was not due to mass transfer limitations. Preliminary calculations strongly indicate that this phenomenon is possibly due the influence of the reverse reaction, H2-CO methanation. Using the results of Vannice [17,18], the following power law expression was obtained for the rate of carbon monoxide consumption via methanation over Ni/TiO2: rr
=

4.42 10'exp

~r--(-~

Pn2

(1)

where Pn2 is the partial pressure of hydrogen (Torr), and r r is the rate of carbon monoxide consumption Qxmol/s gNi). Values of r r were estimated from Eq.

40

f
/ rate ! 30 o

(}.tmol CO ~ 2 0

10"

o~ . _ _ ~ o

0 30 40 50 60 70 80 % o f Equilibrium Conversion Fig. 1. Measured net rate of carbon monoxide formation ( O ) and calculated reverse rate of carbon monoxide consumption via methanation ( O ) as a function of percent equilibrium conversion for Ni/TiO 2 at 723 K. Reaction conditions: CH 4 / C O 2 / H e = 1 / 1 / 1 . 8 , P ~ 740 Torr.

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122
10

101

TOFco

(s-l )

.01 1.2

1.3

1.4

1.5

-1 1000 / T ( K ) Fig. 2. Turnover frequencies for ( ) N i / T i O 2 , (zx) N i / C , ( 0 ) N i / S i O 2, and ( ) N i / M g O , calculated from rates of carbon monoxide formation and irreversible carbon monoxide uptakes on the fresh catalyst samples. Reaction conditions: CH 4 / C O 2 / H e = 1 / 1 / 1 . 8 , P = 740 Torr.

(1) for 1.22%Ni/TiO 2 using partial pressures of hydrogen measured in the effluent stream during methane reforming (Fig. 1). The constant value for the forward rate of carbon monoxide production (_+95% confidence interval) during reforming at 723 K, rf = 41 + 4 p,m o l / s gcat, directly calculated via rnet = r f - rr, indicates that the observed decrease in net reaction rate with increasing proximity to equilibrium is due to the strong influence of the reverse reaction, methanation. As a precautionary note, it should be stressed that the influence of methanation on observed reaction kinetics for methane reforming with carbon dioxide must be taken into consideration for catalysts which are also active for H2-CO methanation, such as nickel, ruthenium, and rhodium [18]. Thus, to minimize the influence of methanation in this kinetic investigation the catalyst mass was varied to keep methane conversion far from thermodynamic equilibrium. Activities of these catalysts at 723 K were reported in the preceding paper [8]. The initial turnover frequencies (TOF) for carbon monoxide production, calculated from irreversible carbon monoxide uptakes on the freshly reduced catalyst samples [8], are provided in Fig. 2. Specific activities were in the order N i / T i O 2 > N i / C > N i / S i O 2 > N i / M g O . The TOFs for N i / T i O 2, N i / S i O 2, and N i / C were corrected for deactivation via normalization to a standard set of conditions, which was monitored throughout the duration of all experiments. Because of the possible influence of the RWGS reaction, activation energies for the consumption of C H 4 and CO 2, as well as the production of CO ( -+ 95% confidence interval), H 2 and H 2 0 are presented in Table 2. Interestingly, the apparent activation energies for all catalysts were initially very similar. The apparent activation energy barrier for hydrogen formation is greater than that for

102

M.C.A Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

Table 2 Activation energies (kcal/mol) for CH 4-CO 2 reforming over nickel catalysts Catalyst 6.3%Ni/C 6.8%Ni/SiO 2 10.1%Ni/MgO 1.22%Ni/TiO 2 1.22%Ni/TiO2 b

ECH4
29 23 22 26 41

EGO2 22 19 21 21 42

EC O a

EH 2 32 27 35 32 51

En 20 23 18 18 22 45

24 _+6 20 _+3 21 _+ 1 23 _+2 44--+ 11

a Value is reported _+95% confidence interval b Obtained after 18 h on stream.

the formation of carbon monoxide, presumably a result of the reverse water-gas shift influence on the reaction mechanism. The apparent activation energies for methane consumption on both N i / C and N i / T i O 2 are slightly larger than those obtained on the N i / S i O 2 and N i / M g O catalysts. Possibly this difference is due to the presence of interstitial surface carbon and TiO x on the nickel surfaces of the N i / C and N i / T i O 2 catalysts, respectively, as evidenced by suppressed H 2 chemisorption [8], which may increase the activation barrier for methane dissociation. Interestingly, the apparent activation energies on N i / T i O 2 increased by a factor of two after roughly 18 h on stream. Gradual deactivation of the N i / T i O 2 catalyst with time is possibly due to either continual deposition of surface carbon or additional site blockage by migrating TiO x [8]. Thus, the increasing presence of either inactive carbon or TiO X species on the nickel surface may also concomitantly increase the activation barrier for methane dissociation. During kinetic investigations, it was also possible from an overall material balance to estimate a rate and an apparent activation energy for carbon formation on N i / S i O 2. The activation energy of 30 k c a l / m o l determined from this estimation is in good agreement with the reported activation energies for carbon diffusion through nickel and whisker carbon formation on nickel, i.e., 34.8 k c a l / m o l [19]. The formation of filamentous whisker carbon was identified on the N i / S i O 2 catalyst using TEM [8].

3.2. Product composition analysis


The H 2 / C O product ratio determined during a series of activity tests is plotted as a function of carbon dioxide conversion in Fig. 3. Methane conversions were always about half that of carbon dioxide due to the influence of the RWGS reaction, consistent with thermodynamic calculations. Also, due to the influence of the RWGS reaction, all observed H z / C O ratios were less than 0.5. The apparent relationship between carbon dioxide conversion and the H 2 / C O ratio indicates the strong influence of the RWGS equilibrium. The reverse water-gas shift pressure ratio, (PcoPH20/Pco2Pu2), was also determined for each catalyst during the activity tests to ascertain the extent of

M.C.J. Bradford, M.A. Vannice ~Applied Catalysis A: General 142 (1996) 97-122
0.4

103

0.3

/.2)

0.2

0.1'

0.0 0 10 CO 2 Conversion (%) 20

Fig. 3. The hydrogen to carbon monoxide product ratio plotted as a function of carbon dioxide conversion lbr (zx) Ni/TiO 2 and ( O ) Ni/MgO. Reaction conditions: CH4/CO 2/He = 1/1/1.8, P = 740 Torr.

the RWGS equilibrium (Fig. 4). The error bars in the figure are due to uncertainty in the calculation of the rate of water formation. For all catalysts, the reaction was at, or extremely close to, RWGS equilibrium over the entire temperature range investigated.
3.3. DRIFTS

DRIFTS spectra, referenced to the background spectra at an identical temperature, were taken of the N i / S i O 2 and N i / T i O 2 (Fig. 5) catalysts during various stages of the pretreatment procedure. Integration of the absorbance peaks in the

0.6'

0.4 PH2oPco PH2Pco2 0.2 j ~ 0.0 600 700 T (K) Fig. 4. Extent of the reverse water-gas shift equilibrium as a function of temperature for ( zx) Ni/TiO2, ( ) Ni/C, ( . ) N i / S i O > and ( O ) Ni/MgO. Reaction conditions: c n 4/CO 2/He = 1/1/1.8, P -~ 740 Torr. Thermodynamic Equilibrium 800 900

104

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122 3.5-

3.0-

'

4000

3500

3000

2500

2000

1500

1000

Wavenumber (cm 1) Fig. 5. In situ DRIFTS spectra of the Ni/TiO 2 catalyst at various stages during the pretreatment procedure. (A) Fresh catalyst at 300 K, (B) post 30 min reduction at 423 K, (C) post 60 rain reduction at 773 K, and (D) pre-reaction spectra at 673 K.

hydroxyl region (3000-3800 cm -1) indicated that the 423 K reduction step was effective at removing only about 26% of surface hydroxyl groups on the N i / T i O 2 catalyst, and only about 14% on the N i / S i O 2 catalyst. In the N i / S i O 2 spectra a strong silanol stretching vibration at 3737 cm 1 [20] and broad shoulder at 3675 c m - l indicated the presence of surface hydroxyl groups after completion of pretreatment. However, the disappearance of a 3615 cm -l absorption band, assigned to a N i O - O H stretch [21], after the 773 K reduction step is indicative of NiO reduction to the zero-valent state. The broad peak in the N i / T i O 2 spectra after pretreatment (Fig. 5) in the 930-960 cm-1 region is assigned to a T i - O stretch [21]; however, during the extensive FTIR and FT Raman investigation by Busca et al. [22], no assignment for a T i - O stretching

M.C.J. Bradford, M.A. Vannice/Applied Catalysis A: General 142 (1996) 97-122

105

0.70.60.520.4~0.30 . 2 - ~ ~~

2052

1931

0.10.0 4000

3740
I I I I [
I

3500

3000

2500
Wavenumber (cml)

2000

1500

1000

Fig. 6. In situ DRIFTS spectra of the N i / S i O 2 catalyst after gas-phase and catalyst subtraction during reaction at 674 K.

mode in this region was made. This may suggest that the large absorbance band in this region may be some type of DRIFTS artifact [23]. Weak bands of carbonate-type species at approximately 1300, 1350 and 1410 cm -1 on the N i / T i O 2 catalyst remain entirely unchanged during the methane reforming reaction, suggesting that it is merely a spectator species present in low concentrations on the surface. DRIFTS spectra were also recorded in situ under reaction conditions between 673 and 773 K for N i / S i O 2 and N i / T i O 2. The N i / S i O 2 spectra (Fig. 6) was referenced to the spectra of the reduced catalyst, and the N i / T i O 2 spectra (Fig. 7) was referenced to the background. Artifacts in both sets of spectra at ca. 3000, 1300 and 2300 cm -1 are due to incomplete subtraction of gas-phase methane and carbon dioxide, respectively. During reaction over the N i / S i O 2 catalyst, an intensity decrease was observed in the difference spectrum at 3740 c m - t , suggesting a loss of surface silanol groups. In addition, both linear (ca. 2050 cm-1) and bridged (ca. 1935 cm -1) carbon monoxide were identified on the surface of the N i / S i O 2 catalyst during the reaction. These results are consistent with previously reported bands for carbon monoxide adsorption on

2.0 1.5 1.0 0.5- -3665 "~,-,, 0.0 4000 3500

1300" ~ ) ~

3 0

2000 Wavenumber (cml)

2500

1500

1000

Fig. 7. In situ DRIFTS spectra of the Ni/TiO 2 catalyst after gas-phase subtraction during reaction at 673 K.

106

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A." General 142 (1996) 97-122

Table 3 Analysis of DRIFTS spectra of carbon monoxide adsorbed on N i / S i O 2 T (K) Terminal (-q = 1) Bridged (qq = 2) Qad (kcal/mol) a ( t n n =I
b

Qad (kcal/mol) c

300 674 699 725 749 773

2056 2052 2050 2048 2047 2046

443 442 442 441 441 441

1937 1931 1932 1930 1932 1934

410 408 408 408 408 409

6.4 8.0 8.8 9.6 10.0 10.3

21.8 23.4 23.2 23.6 23.3 22.6

0.086 0.117 0.144 0.165 0.186 0.208

20.6 21.8 21.4 21.6 21.1 20.5

a Calculated from the model in Ref. [45]. b Calculated from Eq. (3). c Calculated using the site distribution.

N i / S i O 2 catalysts at reduced nickel sites [21] and with conclusions drawn from carbon monoxide chemisorption measurements [8]. A summary of wavenumbers for the adsorbed carbon monoxide species is provided in Table 3. While the wavenumber of the linearly adsorbed carbon monoxide decreased with increasing temperature, indicative of a decreasing extent of lateral interactions at lower surface coverages, the wavenumber of the bridged carbon monoxide remained fairly constant within experimental error. The spectra shown for the N i / T i O 2 catalyst (Fig. 7) show significant absorption in the hydroxyl region. The hydroxyl peaks are centered around 3665 cm -1, consistent with the DRIFTS results of Huisman et al. for chemisorbed water on titania [24], suggesting that these hydroxyl species are located on reduced titania, not nickel sites. It is known that Ti 3+ facilitates water dissociation [25], and evidence for the presence of surface TiO x phases was quantified with TEM [8]. From the temperature dependence of the integrated absorbance spectra, the heat of dissociative adsorption of water on the catalyst was estimated at Qad -- 5.0 kcal/mol, a value roughly half that as reported for water adsorption on transition metal surfaces [26]. Raupp and Dumesic [25] have estimated the activation barrier for recombination and desorption for water on TiO x to range from 14.3 to 26.3 kcal/mol, thus our heat of adsorption value suggests that the activation barrier for dissociative water adsorption on TiOx is 15 _+ 6 kcal/mol. Another important result of this DRIFTS study was that no evidence of adsorbed carbon monoxide on N i / T i O 2 was obtained under reaction conditions. This is consistent with the low value of Qa0 = 1.7 k c a l / m o l obtained from chemisorption isotherms [8] and with thermal desorption spectra of carbon monoxide on TiOx/Ni reported by Raupp and Dumesic [27-29].

3.4. Partial pressure dependencies


The dependence of the rates of H 2 formation, CO formation, C H 4 consumption, and CO 2 consumption on the feed partial pressure of C H 4 and CO 2 were

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

107

4"0 t 3.5 3.0

in r co
2.5 ~,gcat-s ) 2.0 1.5 1.0
4 5

In PCH4(Torr) 4.0B

35
in rco

/ /
2.5. /
2.0

4 5 In PCO2 (Torr)

Fig. 8. The observed dependence of the rate of carbon monoxide formation (A) methane partial pressure and (B) carbon dioxide partial pressure at (O) 773 K, (A) 798 K, and (0) 823 K for Ni/MgO.

determined over a broad temperature range for the N i / M g O and the N i / T i O 2 catalysts. The reaction orders for CH4, a, and CO 2, b, were determined from a fit of the data to a power rate law of the general form:
r i -- k P c H 4 P ~ 0 2
__ a b

(2)

by linearization of a logarithmic plot, such as that shown in Fig. 8 for carbon monoxide formation over N i / M g O . A summary of the determined reaction orders is provided in Table 4 for all catalysts. The influence of hydrogen addition to the feed stream (PcH4 = Pco2 = 200 Torr) was also examined for the N i / M g O catalyst at 823 K and the N i / T i O 2 catalyst at 723 K (Fig. 9). The immediate influence of hydrogen is to increase carbon dioxide conversion through the RWGS. The inhibition of methane consumption is also clear, and at higher pressures methanation was also observed (negative conversion refers to the production of methane). These results are consistent with thermodynamic

08

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

+i +1 +1 +1

+1 +1 +~ +1

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122
40 30 20

109

Conversion 10

(%)

0
-10" -2(

100

200 PH2 (Tort)

300

400

40

20

Conversion

(%)

-20

-40 I 0

, 100

, 200

, 300

i 400

PHz (Torr) Fig. 9. A comparison of the observed influence of hydrogen addition to the feed mixture with that calculated from thermodynamic equilibrium for (A) N i / M g O at 823 K and (B) Ni/TiO 2 at 723 K. ( 0 ) Carbon dioxide equilibrium conversion, ( O ) carbon dioxide experimental conversion, ( ) methane equilibrium conversion, ( zx) methane experimental conversion.

calculations, illustrated in the same figures, the postulates of Fig. 1, and the results of Lewis et al. [9].

4. Discussion of results The TOFs at 723 K (1.73 s -t for N i / T i O e, 0.58 s -1 for N i / C , 0.36 s - l for N i / S i O e, and 0.04 s -1 for N i / M g O ) are consistent with TOF values of 0.16 s -~ for N i / M g O [2], 0.02-0.18 s -1 for AlzO3-supported noble metals [30], 0.15-0.34 s -~ for supported rhodium catalysts [31], and 0.01-0.78 s -~ for supported palladium catalysts [32] reported in the literature. The literature TOFs were obtained at 723 K by extrapolation using the reported activation energies. The relative order of catalyst activity, N i / T i O 2 > N i / S i O e > N i / M g O ,

llO

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A." General 142 (1996) 97-122

identical on either a TOF or per gram of nickel basis, is identical to that obtained on supported palladium catalysts [32], but different than that obtained on supported rhodium [31 ]. The activation energy obtained with N i / S i O 2 for carbon monoxide formation, 20 _+ 3 kcal/mol, is in reasonable agreement with the values of 13 and 20 k c a l / m o l obtained by Sakai et al. [33] and Osaki et al. [34], respectively. Although neither group of authors reported the activation energy, it was possible to calculate values for the activation energy from data reported within their papers. The activation energies obtained on the N i / M g O catalysts are also in fair agreement with values calculated from data reported in the literature. From the data of Rostrup-Nielsen and Hansen [2], it is possible to obtain a value of roughly 28 kcal/mol for a 1.4%Ni/MgO catalyst, while from the data of Osaki et ai. [34] a value of approximately 20 kcal/mol for a 2 0 % N i / M g O catalyst can be estimated. The only catalytic data for a N i / T i O 2 catalyst in the literature which allow for an estimation of activation energy was provided by Osaki et al. [34], from which a value of 20 kcal/mol may be estimated, in good agreement with the initial value of 23 _+ 2 k c a l / m o l obtained in this investigation. Experimental results indicated that the H z/CO product ratio was a function of carbon dioxide conversion (Fig. 3). This is consistent with results of Swaan et al. [4] and Blom et al. [5] who show that H2/CO product ratio is a function of the water-gas shift equilibrium. Van Looij et al. have reported a similar relationship during studies of methane oxidative coupling [35]. Correspondingly, experimental results also indicated that the RWGS was very close to thermodynamic equilibrium (Fig. 4). Imelik and Vedrine [36] have reported that carbon monoxide adsorption on N i / S i O 2 (with respect to binding site wavenumbers) is very similar to that on the Ni(100) and Ni(110) surfaces. From experimental HREELS data of carbon monoxide adsorbed on these surfaces [37-44], the ratio of the symmetric C - O stretch to the symmetric N i - C stretch (_+ standard deviation) was calculated to be 4.6 + 0.3 and 4.7 _+ 0.5 for terminal and bridged sites, respectively. Use of these ratios permitted a direct estimation of the N i - C symmetric stretch on the N i / S i O 2 catalyst (Table 3). The heat of adsorption of carbon monoxide at the linear (-q -- 1) and bridged ('q = 2) adsorption sites was then estimated using a vibrational model proposed by Bradford and Vannice [45]. By assuming that the binding energy, E, is equal to the heat of adsorption it was possible to calculate the site occupancy ratio, (n~= j/n~ = 2), from the equation presented by Degras [46]:

t2 <==}

(3)

This result then permitted an estimation of 21.2 _+ 0.5 kcal/mol for the observable carbon monoxide heat of adsorption on the N i / S i O 2 catalyst, which is

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

111

slightly lower than initial values of 25 to 30 kcal/mol reported on various nickel single crystal surfaces [37,47-49]. A summary of these calculations is provided in Table 3. Although the site occupancies of bridged and linear carbon monoxide on N i / S i O 2 calculated with Eq. (3) differ from those reported on Ni(100) and Ni(110) surfaces [38,50], they are qualitatively consistent in that the calculated fraction of bridged species decreases with increasing temperature. For N i / M g O , N i / T i O 2 and N i / C , an increase in methane partial pressure increased both hydrogen and carbon monoxide production, while an increase in carbon dioxide partial pressure increased carbon monoxide production and decreased hydrogen production (Table 4). The latter observation is presumably due to the influence of the reverse water-gas shift reaction, shown to be operating near thermodynamic equilibrium. Qualitatively, the trends in dependencies for each catalyst agree with those reported by Erd~Shelyi et al. [31] at 773 K for TiO 2- and MgO-supported rhodium. Although ErdiShelyi et al. [31] observed a negative influence on hydrogen production only for carbon dioxide concentrations greater than 20%, this is consistent with the experimental method used in this investigation, which only utilized carbon dioxide concentrations in the CO2-CH 4 feed mixture of greater than 20%. There exist limited partial pressure data in the literature for CH4-CO 2 reforming over nickel catalysts. Rostrup-Nielsen and Hansen [2] have reported that the carbon dioxide reaction order for consumption of methane is zero for a I % R u / M g O catalyst, in good agreement with our values of 0.01 to 0.05 for N i / M g O (Table 4). In contrast, Yamazaki et al. [51] reported for a N i / M g O catalyst at 1123 K that the rate of methane consumption was a strong Langmuirian function of carbon dioxide partial pressure. No partial pressure data has been reported in the literature for N i / T i O 2 catalysts. The influence of hydrogen addition to the feed stream was also examined for the N i / M g O catalyst at 823 K and the N i / T i O 2 catalyst at 723 K (Fig. 9). Rostrup Nielsen and Hansen [2] have reported that the addition of slight amounts of hydrogen ( C H 4 / H 2 = 10) to the feed negligibly influences the reaction. This is reasonably consistent with our results for PIJ2 = 20 Torr in Fig. 9. No experiments were conducted to ascertain the influence of carbon monoxide addition to the feed stream; however, thermodynamic calculations imply that carbon monoxide may also strongly inhibit both methane and carbon dioxide conversion [52]. Although extensive partial pressure studies were not performed with either the N i / C or the N i / S i O 2 catalyst, some partial pressure data were obtained for each catalyst at 723 K (Table 4). Sakai et al. [33] have reported reaction orders of 0.02 ~ 0.05 and 0.5 ~ 0.6 for carbon dioxide and methane, respectively, for a 10%Ni/SiO 2 catalyst from 823 to 923 K. The slight differences between their values and those obtained here (Table 4) are possibly due to the temperature difference of the two investigations. One other interesting result for the N i / S i O 2 is worthy of mention. The influence of carbon dioxide increased, albeit very

112

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

gradually, the rate of hydrogen production. Although this is in contrast with results for the MgO and TiO2-supported nickel catalysts, it is in agreement with observations made by Erd~Shelyi et al. for SiO2-supported rhodium [31] and palladium [32]. Apparently SiO 2 plays a much different role in the production of hydrogen than either MgO, TiO2, or A1203, suggesting that an intermediate in the reaction mechanism is support-related. This observation is analogous to the reported absence of adsorbed CH30 on SiO2-supported metals during CO-H 2 methanation [53]. After a thorough analysis of all of our data, all available literature data, and numerous reaction models, the following generalized reaction sequence is proposed for C H 4 - C O 2 reforming: kl) CHx*+ (-~-53H2 CH4 + * ~ k_l ({1))

ga

2 [ CO2 + * ~

CO2. ]

({2})

Ka

H2 + 2* ~

2H*

({3})

IG

2 [ C O 2 . + H* ~

CO* + OH*]

({4})

Ks

OH* + H* ~

H20 + 2*

({5})

CHx* + OH* ~

CHxO* + H*

({6})

CHxO*

k7 ) C O * '~-(2) H2

({7})

1/Ks

3[C0.

CO + *]

({8})

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

113

which corresponds to the overall reaction stoichiometry:


C H 4 d- 2CO 2 ~ H 2 + H 2 0 + 3CO

Methane adsorption and dissociation to a CH x fragment (or distribution of CH x fragments) is supported by previous studies. HREELS and molecular beam investigations of methane on N i ( l l 1) by Ceyer et al. show that the stable CHx fragment is temperature sensitive [54], and pulse surface reaction rate analysis of various oxide-supported nickel catalysts by Osaki et al. during C H 4 - C O 2 reforming show that the stable CH x fragment is also sensitive to the support [34]. Reversible ( ~ ) methane adsorption and dissociation is supported by the influence of hydrogen addition to the feed gas, which showed that the methane consumption rate was both reversible and not at thermodynamic equilibrium (Fig. 9). Reaction steps {2}-{5} and {8} represent the RWGS reaction and are quasi-equilibrated (~'~'~) to accommodate the experimental result that this reaction is near thermodynamic equilibrium (Fig. 4). Although Solymosi [55] has reported that clean nickel surfaces can dissociate carbon dioxide at temperatures above 420 K, non-dissociative carbon dioxide adsorption on N i / T i O 2 was indicated by carbon dioxide adsorption isotherms, which could be fit only to a non-dissociative Langmuir adsorption model [8]. The promotion of carbon dioxide dissociation by adsorbed hydrogen atoms has been reported by Solymosi et al. [30] and is supported by other DRIFTS data for C H 4 - C O 2 reforming on P t / T i O 2 [52]. Furthermore, step {5} is consistent with literature results showing that water adsorption on N i / T i O x at high temperatures is dissociative [25]. It is also likely that the transition state for reaction step {4} is a formate (HCO 2) species [56]. Some investigators have used FTIR to identify support-bound formate groups during C H 4 - C O 2 reforming [16,57], and the decomposition of formate may control the rate of the water-gas shift reaction [58]. However, because the RWGS is essentially at thermodynamic equilibrium and turnover frequencies for the RWGS are typically much higher than for C H 4 - C O 2 reforming [2], exclusion of HCO 2 from the proposed reaction mechanism does not alter the mathematical rate expression. Interaction of the adsorbed CH x fragments with surface hydroxyl groups, step {6}, is supported by FTIR investigations of methane adsorption on AI203 [59] and supported Ir [60], DRIFTS studies of methane adsorption and partial oxidation on Rh/A1203 [61], and our DRIFTS results for C H 4 - C O 2 reforming on N i / S i O 2 (Fig. 6) and P t / T i O 2 [52]. The proposed irreversibility of CHxO decomposition is valid under circumstances where no net methane formation occurs and is therefore (in regards to Fig. 9) justifiable for feed streams where hydrogen is absent. This reaction mechanism pertains only to catalysts which kinetically inhibit excessive carbon formation (or operate in a regime where carbon formation is thermodynamically prohibited), thus it is assumed that the surface concentration of carbon (in the forms of CHxO and CH x) is at steady state and remains

114

M.C.J. Bradford, M.A. Vannice/ Applied CatalysisA: General 142 (1996) 97-122

Table 5 Elementary gas-phase reactions and corresponding activation energy barriers Reaction a
C H 4 + M --* C H 3 + H + M C H 3 + H -'* C H 2 + H 2 C O 2 + H --* C O + O H H+OH+M ~ H20+M C H 2 + O H --* C H 2 0 + H C H + O H --* C H O + H C + O H --* C O + H CH30+M ~CH20+H+M CH20+M ~ CHO+H+M C H O + M --* C O + H + M

Activation energy
(kcal/mol) 82+5 b 15 15+2 b 0 0 0 0 c 25 81 17

a M denotes a third body. b O n l y the energy barrier for the reverse reaction was given by Miller and Bowman [62]. The forward rate

constant was calculated directly from the reverse rate constant using the values of the equilibrium constant, which were calculated from available thermodynamic data [63,64]. Data obtained from GR1-MECH at ftp address crvax.sri.com with login id grimech.

constant, i.e., the net rate of methane dissociation, step {1}, equals that for CH xO decomposition, step {7}. A comparison of the proposed sequence of steps for this heterogeneous reaction with analogous homogeneous elementary steps reported by Miller and Bowman [62] indicates that methane dissociation and CH xO decomposition, which have extremely high activation barriers, are indeed likely to be rate determining steps (Table 5). If it is further assumed that the most abundant reaction intermediate (mari) is CHxO, then L = [*] + [CHxO* ], and the following general expression may be readily derived for the reaction rate of methane:

~:'PcH4Pco2

(4)

where ~:i = constant:

kiL,

L -- total number of active sites, and K is a lumped equilibrium

(Ks)
K2K4K6
(5)

K=

A detailed derivation of the model is presented elsewhere [52]. Computer optimization of kinetic data from the N i / M g O and N i / T i O 2 catalysts failed to locate a global minimum for x and indicated that the statistical fit of the data (as measured by residual sum of squares) of Eq. (4) was essentially independent of x (0 _< x _< 4). If it is further assumed in the kinetic model that all of the

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

115

hydrogen generated in step {1} is consumed by CO 2 and that CO i reacts only with H * via step {4}, then a simple expression for x can be readily derived [52]: x~<6-2( rc---z~)
FCH4

(6)

where rco 2 and rCH 4 a r e the experimental reaction rates for carbon dioxide and methane consumption, respectively. Due to the nature of these assumptions, Eq. (6) physically represents only an upper bound to the permissible value of x in the adsorbed CH x fragments, and upper bounds of x < 2.6 _+ 0.1 for N i / M g O and x < 2.2 _+ 0.1 for N i / T i O 2 were calculated over the temperature ranges of 713-753 and 673-723 K, respectively. These temperature intervals were chosen to facilitate comparison with the results of Osaki et al., who reported values of x = 2.7 for N i / M g O and x = 1.9 for N i / T i O 2 within the specified temperature

20"

rCH4

( mo' /

10

100

200 PCH, (To~)

300

400

12 10 8" rCH4 6

?o
B t~

goat. secJ 4"


2" 0 0 100 200 PC02 (Tort) Fig. 10. Fit of the proposed kinetic model for CH4-CO 2 reforming as a function of (A) methane partial pressure and (B) carbon dioxide partial pressure for Ni/MgO at ([3) 773 K, (r,) 798 K, and (O) 823 K. 300 400

116

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A." General 142 (1996) 97-122

Table 6 Kinetic model parameters for N i / M g O and N i / T i O 2 Catalyst


T (K)

Parameter

kl a,b
Ni/TiO 2 673 698 723 773 798 823 0.003 + 0.009 0.010 + 0.005 0.042+0.016 0.031 + 0.006 0.054 + 0.006 0.085 +0.013

k7 c
0.77 3.74 5.35 23.08 20.45 33.58

~_, K- d
0 5.38 0.85 0.167 0.153 0.164

Ni/MgO

a b c d

Value is Units of Units of Units of

reported + 95% confidence interval. I.~mol gcat -1 S - 1 Torr -1. ixmol gcat-1 s - l . ixmol gcat- ~ s - ~ Torr- 1.

regime [34]. Comparison of the two sets of x values indicates that they are consistent within the assumptions associated with the derivation of Eq. (6). If a reasonable, but arbitrary value of x = 2 is chosen for Eq. (4) to represent the kinetic behavior for CH4-CO 2 reforming over N i / M g O and N i / T i O 2, the results in Fig. 10 are attained. To account for the deactivation of the N i / T i O 2 catalyst, reaction rates were normalized to an apparent activation energy of 41 kcal/mol (Table 2) prior to optimization with the kinetic model. A summary of the optimized parametersdetermined with this model is provided in Table 6. The zero value for k-1K at 673 K for N i / T i O 2 is most likely a result of excessive scatter in the data, due to the experimental difficulty in quantifying hydrogen at very low concentrations in the effluent gas mixture. Plots of these model parameters versus reciprocal temperature (Fig. 11) yielded the activation barrier for step {1} in the forward direction, E~,, the activation barrier for step {7}, E~7, and the lumped energy parameter, - ( E ~ 1 + A Hg), where E~_, is the
100 10
1

Parameter
.1 .01 .001 1.2

1.3
1000/T (K)

1.4
-1

1.5

Fig. 11. Plots of kinetic model parameters for CH 4 - C O 2 reforming versus reciprocal absolute temperature for the N i / M g O ; (r3) '~7, (zx) ~: i g , ( O ) kl and Ni/TiO2; ( l l ) k7, ( A ) k 1K, ( 0 ) lq.

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122
Table 7 Activation energies and carbon monoxide adsorption thermodynamics for nickel catalysts Catalyst Ni/MgO Ni/TiO 2 Ni/SiO 2

117

E~ a a (kcal/mol)
26 51 .

EG ~ (kcal/mol)
9.3 38 . .

-(E~_, + AHg) a (kcal/mol)


0.5 74 .

A Hadxo (kcal/mol) - 1.7 b 21.2+0.5 c

ASad,c (cal/mol K) - 11.8 b

a Calculated from the kinetic model. b Calculated from chemisorption isotherms [8]. c Calculated from adsorbed vibration spectra (Table 3).

activation barrier for step {1} in the reverse direction, and AHg is the enthalpy change for the reaction: CO + CH20*
A Hg ~-

CO 2 -'{-CH 2

(7)

A summary of the energy parameters determined with this kinetic model is provided in Table 7. If the enthalpies of adsorption calculated by Shustorovich [26] for CH20, AHad = - 1 9 kcal/mol, and CH 2, AHad = - 8 3 kcal/mol, on N i ( l l l ) are assumed to be valid for N i / M g O and N i / T i O 2 under reaction conditions, then A Hg may be estimated for nickel from the following system of reactions, the sum of which represents reaction (7):
A Hg
CO + CH20 ~ CO 2 +

CH 2

A Had
CH 2 + * ~--

CH~

CH20*

--LHadC H 2 0

+ *

From this network, it is obvious that AHg = AHg + AHad,CH2 -- AHad,CH20, where A H g = 5 3 kcal/mol [63,64]. The calculated value of A H g = - l l k c a l / m o l permits estimation of EL, = l0 kcal/mol for N i / M g O and E~ 1 = - 6 3 k c a l / m o l for N i / T i O 2. The estimated value of E~_, for N i / M g O is in good agreement with activation barriers calculated by Shustorovich [26] for H - C H x recombination on N i ( l l l ) and may suggest that on the N i / M g O catalyst the active site for methane decomposition is located solely on the nickel surface. The large negative value of EL, estimated for N i / T i O 2 is not physically realistic, and it implies that the heats of adsorption of CH 2 and CH20 on TiO2-supported nickel are greatly different than those for a clean nickel surface, presumably due to modification by TiO x overlayers. This is consistent with carbon monoxide adsorption isotherms and TEM results reported

118

M.C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

previously for N i / T i O z [8], and carbon monoxide and hydrogen thermal desorption spectra from TiOx/Ni [42-44]. A review of the literature (see Ref. [8] and references therein) suggests that carbon deposition during CH4-CO 2 reforming occurs through both carbon monoxide disproportionation and methane decomposition over the experimental temperature range used in this investigation. Prior TPO studies of the used catalysts [8] revealed that the catalyst resistivity to excessive coking - - N i / M g O > N i / T i O 2 >> N i / S i O 2 - - could be correlated with stabilities of CH x on each catalyst, i.e., CH2. 7 > CH1. 9 > CH1.0, as reported by Osaki et al. [34]. In addition, in situ DRIFTS revealed the presence of both linear and bridged adsorbed carbon monoxide on the N i / S i O 2 catalyst, while no adsorbed carbon monoxide was identified on the N i / T i O 2 catalyst. Correspondingly, the estimated value for the carbon monoxide heat of adsorption on N i / T i O 2, Qad = 1.7 kcal/mol, is an order of magnitude lower than that of Qad = 21.2 kcal/mol for N i / S i O 2 (Table 7). This implies that a catalyst which exhibits suppressed or weak carbon monoxide chemisorption can inhibit formation of excessive inactive carbon during CH4-CO 2 reforming because carbon dioxide disproportionation is decreased. A reasonable kinetic model for C H 4 - C O 2 reforming is proposed here that fits partial pressure data over reasonable temperature and pressure ranges for both the N i / T i O 2 and N i / M g O catalysts. Additional insight into the possible significance of this proposed model can be obtained through comparison with an analogous gas-phase reforming mechanism (Table 5). A comparison of the gas-phase activation energy barriers (Table 5) with those determined in the heterogeneous kinetic model (Table 6) for x = 2 suggests that N i / M g O is more effective than N i / T i O 2 at decreasing the activation barriers for both CH 4 dissociation and C H 2 0 decomposition. The reason for the large activation barriers on N i / T i O 2 relative to the other nickel catalysts after significant time on stream may (Table 2) be due to geometric considerations. Suppressed chemisorption behavior and identification of various TiOx phases by TEM confirmed that SMSI occurred in the N i / T i O 2 catalyst [8], in agreement with earlier studies [65]. This indicates that TiO x species migrated to cover large ensembles of nickel sites, which resulted in reduced carbon deposition, relative to N i / S i O 2. Under reaction conditions, which are net reducing, it is reasonable to expect that TiOx migration may continue. Thus, it is possible that deactivation of N i / T i O 2 is due to site blockage by either inactive carbon or migrating TiO x. It is suggested that continual geometric blockage of active nickel sites on N i / T i O 2 results in a concomitant increase in the activation barrier for methane dissociation. This site blockage phenomenon is analogous to sulfur-passivated steam reforming of methane, in which sulfur passivation of a nickel catalyst reduced carbon deposition but increased the activation energy from 26.3 to 54.2 kcal/mol [66]. However, recent AES results by Jiang and Goodman [67] have shown that the

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

119

activation barrier for methane dissociation on sulfided Ni(100) is not much different than the activation barrier on clean Ni(100); this may restrict the aforementioned hypothesis. It has been suggested in the literature that the kinetics of C H 4 - H 2 0 reforming and C H 4 - C O 2 reforming are essentially identical [2,11]. Qualitatively, the reaction mechanism proposed here has CH x species reacting only with surface hydroxyl groups, which are produced via the reverse water-gas shift. Alternatively, the presence of large quantities of water in the feed can result in the direct production of surface hydroxyls via water dissociation and circumvent the need for the reverse water-gas shift to occur as a prerequisite to CH x reaction. In this context, therefore, tumover frequencies for C H 4 - H 2 0 reforming should always be larger than those for CH4-CO 2 reforming, as illustrated by the data of Rostrup-Nielsen and Hansen [2]. These authors have also shown for MgO-supported noble metal catalysts that the TOFcoJTOFH2 o ratio; where TOFco: and TOFH2o are the turnover frequencies for CH4-CO 2 and C H a - H 2 0 reforming, respectively; decreases as the carbon monoxide heat of adsorption increases [2]. Although they did not provide an explanation, this behavior is explainable in terms of the kinetic model proposed here for C H 4 - C O 2 reforming; strongly bound carbon monoxide implies a higher surface concentration of carbon monoxide, which results in a direct inhibition of both CH ~O decomposition, step {7}, and the reverse water-gas shift reaction, in step {4}. In the proposed kinetic model, the assumption that the water-gas shift reaction is thermodynamically equilibrated, where KwgS= Pco2 PH2/PH20Pco, permits writing the expression for methane consumption, Eq. (4), in a form more readily applicable to C H 4 - H 2 0 reforming:
^

k aPCH4PH20

t,TKwgs P TX'/2 + 1 +

ec.4 P.2o

Rostrup-Nielsen and Hansen [2] also reported that at very low C O 2 / C H 4 ratios the carbon dioxide reaction order on a R u / M g O catalyst approached unity. In this experimental limit, Eq. (4) becomes approximately:

k|k7efH4Pc02
rCH'

p-

k-IKPcoP(H 4-x)/2

(9)

Thus, this simplified rate expression can explain their observation. The nickel dispersions measured by carbon monoxide chemisorption for each of the supported nickel catalysts varied only between 5 and 16% [8]; therefore, the relative differences in catalyst activity are likely attributable to influence of the support. Although Osaki et al. have shown that the stable CH x intermediate

120

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

under reaction conditions is sensitive to the support, they did not find a correlation between CH~ and catalyst activity [34]. This result is consistent with our results from the kinetic model optimization, which were relatively insensitive to the numerical value of x. However, analysis of reaction orders for hydrogen production revealed that SiO2-supported metals play a much different role in hydrogen production than either MgO, TiO2, or A1203, suggesting that an intermediate in the reaction mechanism is support-related. This observation is analogous to the reported absence of adsorbed C H 3 0 on SiO2-supported metals during C O - H 2 methanation [53]. If, during C H 4 - C O 2 reforming, the support serves as a sink for adsorbed hydroxyl groups, the reaction of adsorbed CH~ fragments with surface hydroxyls, via step {6} in the reaction sequence, is likely to occur at a metal-support interface. Correspondingly, the metal-support interface may also contain the active sites for subsequent CH~O decomposition; thus, the support may influence catalyst activity by altering the stability of the CHxO intermediate at the metal-support interface.

5. Conclusions

A kinetic model for the C H 4 - C O 2 reforming reaction is proposed which successfully describes the reaction kinetics over N i / M g O and N i / T i O 2 catalysts. The main assumptions of this model are that: (i) CH 4 decomposition and CHxO decomposition are the slow kinetic steps, (ii) CO 2 participates in the reaction mechanism through the reverse water-gas shift to produce surface OH groups, and (iii) surface OH groups react with adsorbed CHx intermediates to yield a formate-type intermediate, CHxO, which decomposes to yield H 2 and CO. A comparison of experimental heterogeneous and homogeneous gas-phase reaction energetics suggests that the two main energetic roles of the catalysts are to reduce the activation barriers for C H 4 dissociation and CHxO decomposition. It is also suggested that the support may serve as a sink for surface hydroxyl groups, such that the active site for CHxO formation and subsequent decomposition may be at the metal-support interface. Activation barriers were found to be considerably higher on N i / T i O 2 after significant time on stream, and this is attributed to a geometric site blockage mechanism whereby migrating TiO~ moieties or inactive carbon deposits break up the large site ensembles on the nickel surface needed for CH 4 dissociation. This result is analogous to reported effects of sulfur passivation on C H 4 - H 2 0 reforming catalysts [66,68]. It is suggested that the proposed kinetic model for C H 4 - C O 2 reforming should also be applicable to C H 4 - H 2 0 reforming. Additionally, these results imply that catalysts which exhibit suppressed, or weak, carbon monoxide adsorption can sufficiently resist excessive carbon formation during C H 4 - C O 2 reforming due to the inhibition of carbon monoxide disproportionation.

M. C.J. Bradford, M.A. Vannice / Applied Catalysis A: General 142 (1996) 97-122

121

Acknowledgements The authors would like to thank the Japanese international joint research program, NEDO, for sponsoring this study, the US Department of Education for providing the GAANN fellowship, and Professor K. Fujimoto and his research group for providing the Ni/MgO catalyst used in this investigation.

References
[1] A.M. Gadalla and B. Bower, Chem. Eng. Sci., 43 (1988) 3049. [2] J.R. Rostrup-Nielsen and J.-H.B. Hansen, J. Catal., 144 (1993) 38. [3] J.R. Rostrup-Nielsen, in J.R. Anderson and M. Boudart (Editors), Catalysis: Science and Technology, Vol. 5, Springer, Berlin, 1984, pp. 1-118. [4] H.M. Swaan, V.C.H. Kroll, G.A. Martin and C. Mirodatos, Catal. Today, 21 (1994) 571. [5] R. Blom, I.M. Dahl, A. Slagtem, B. Sortland, A. Spjelkavik and E. Tangstad, Catal. Today, 21 (1994) 535. [6] A.M. Gadalla and M.E. Sommer, Chem. Eng. Sci., 44 (1989) 2825. [7] H.D. Giesser, N.R. Hunter, A.N. Shigapov and V. Januati, Energy and Fuels, 8 (1994) 1123. [8] M.C.J. Bradford and M.A. Vannice, Appl. Catal., 142 (1996) 73. [9] W.K. Lewis, E.R. Gilliland and W.A. Reed, Ind. Eng. Chem., 41 (1949) 1227. [10] D.B. Clarke, I. Suzuki and A.T. Bell, J. Catal., 142 (1993) 27. [11] I.M. Bodrov and L.O. Apel'baum, Kinet. Catal., 8 (1967) 326. [12] N.M. Bodrov, L.O. Apel'baum and M.I. Temkin, Kinet. Catal., 5 (1964) 614. [13] J.T. Richardson and S.A. Paripatyadar, Appl. Catal., 61 (1990) 293. [14] M.A. Vannice, S.H. Hyun, B. Kalpakci and W.C. Liauh, J. Catal., 56 (1979) 358. [15] M. Boudart, D.E. Mears and M.A. Vannice, Ind. Chim. Belg., 32 (1967) 281. [16] Z.L. Zhang and X.E. Verykios, Catal. Today, 21 (1994) 589. [17] M.A. Vannice, Catal. Rev. Sci. Eng., 14 (1976) 153. [18] M.A. Vannice, J. Catal., 74 (1982) 199. [19] N.M. Rodriguez, J. Mater. Res. 8 (1993) 3233. [20] R.S. McDonald, J. Phys. Chem., 62 (1958) 1168. [21] A.A. Davydov, Infrared Spectroscopy of Adsorbed Species on the Surface of Transition Metal Oxides, Wiley, Chichester, 1984. [22] G. Busca, G. Ramis, J.M.G. Amores, V.S. Escribano and P. Piaggio, J. Chem. Soc. Faraday Trans.. 90 (1994) 3181. [23] P.E. Fanning, personal communication, 1995. [24] H.M. Huisman, P. van der Berg, R. Mos, A.J. van Dillen and J.W. Geus, Appl. Catal. A, 115 (1994) 157. [25] G.B. Raupp and J.A. Dumesic, J. Phys. Chem., 89 (1985) 5240. [26] E. Shustorovich, Adv. Catal., 37 (1990) 101. [27] G.B. Raupp and J.A. Dumesic, J. Catal., 97 (1986) 85. [28] G.B. Raupp and J.A. Dumesic, J. Catal., 96 (1985) 597. [29] G.B. Raupp and J.A. Dumesic, J. Catal., 95 (1985) 587. [30] F. Solymosi, G. Kutsain and A. Erdtihelyi, Catal. Lett., 11 (1991) 149. [31] A. Erd~ihelyi, J. Cserenyi and F. Solymosi, J.Catal., 141 (1993) 287. [32] A. ErdtJhelyi, J. Cserenyi, E. Papp and F. Solymosi, Appl. Catal. A, 108 (1994) 205. [33] Y. Sakai, H. Saito, T. Sodesawa and F. Nozaki, React. Kinet. Catal. Lett., 24 (1984) 253. [34] T. Osaki, H. Masuda and T. Moil, Catal. Lett., 29 (1994) 33. [35] F. van Looij, J.C. van Giezen, E.R. Stobbe and J.W. Geus, Catal. Today, 21 (1994) 495. [36] B. Imelik and J.C. Vedrine, Catalyst Characterization. Physical Techniques for Solid Materials, Plenum, New York, 1994. [37] J.C. Bertolini and B. Tardy, Surf. Sci., 102 (1981) 131. [38] J. Bauhofer, M. Hock and J. Kiippers, Surf. Sci., 191 (1987) 395.

122 [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68]

M.C.J. Bradford, M.A. Vannice /Applied Catalysis A: General 142 (1996) 97-122

K. Christmann, Introduction to Surface Physical Chemistry, Springer, New York, 1991. M. Nishijima, S. Masuda, Y. Sakisaka and M. Onchi, Surf. Sci., 107 (1981) 31. P. Uvdal, P.A. Karlsson, C. Nyberg, S. Andersson and N.V. Richardson, Surf. Sci., 202 (1988) 167. N.V. Richardson and A.M. Bradshaw, Surf. Sci., 88 (1979) 255. D.A. King and D.P. Woodruff (Editors), The Chemical Physics of Solid Surfaces, Vol. 6, Elsevier, Amsterdam, 1993. D.A. King and D.P. Woodruff (Editors), The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, Vol. 2, Elsevier, Amsterdam, 1983. M.C.J. Bradford and M.A. Vannice, Ind. Eng. Chem. Res., (1996) in press. D.A. Degras, Suppl. Nuovo Cimento, 5 (1967) 420. D.A. King and D.P. Woodruff (Editors), The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, Vol. 3A, Elsevier, Amsterdam, 1990. H. Froitzeim and V. K~Shler, Surf. Sci., 188 (1987) 70. I. Toyoshima and G.A. Somorjai, Catal. Rev. Sci. Eng., 19 (1979) 105. A. Grossman, W. Erley and H. Ibach, Surf. Sci., 330 (1995) L646. O. Yamazaki, T. Nozaki, K. Omata and K. Fujimoto, Chem. Lett., 10 (1992) 1953. M.C.J. Bradford, PhD Thesis, The Pennsylvania State University, in preparation. B. Sen and J.L. Falconer, J. Catal., 122 (190) 68. S.T. Ceyer, Q.Y. Yang, M.B. Lee, J.D. Beckerle and A.D. Johnson, in D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak (Editors), Methane Conversion, Elsevier, Amsterdam, 1993, p. 51. F. Solymosi, J. Mol. Catal., 65 (1991) 337. E. Shustorovich, Surf. Sci., 253 (1991) 386. J. Nakamura, K. Aikawa, K. Sato and T. Uchijima, Catal. Lett., 25 (1994) 265. T. Shido, K. Asakura and Y. Iwasawa, J. Catal., 122 (1990) 55. C. Li, W. Yan and Q. Xin, Catal. Lett., 24 (1994) 249. F. Solymosi and J. Cser6nyi, Catal. Today, 21 (1994) 561. K. Walter, O.V. Buyevskaya, D. Wolf and M. Baems, Catal. Lett., 29 (1994) 261. J.A. Miller and C.T. Bowman, Prog. Energy Comb. Sci., 15 (1989) 287. JANAF Thermodynamic Tables, J. Phys. Chem. Ref. Data, 14 (1985) Suppl. I. I. Barin, Thermochemical Data of Pure Substances, VCH, Weinheim, Parts I and II, 1993. J.A. Dumesic, S.A. Stevenson, R.D. Sherwood and R.T.K. Baker, J. Catal., 99 (1986) 79. J.R. Rostrup-Nielsen, J. Catal., 85 (1984) 31. X. Jiang and D.W. Goodman, Catal. Lett., 4 (1990) 173. J.R. Rostrup-Nielsen, Stud. Surf. Sci. Catal., 68 (1991) 85.

You might also like