You are on page 1of 7

2432

J. Phys. Chem. B 2005, 109, 2432-2438

Methanation of CO over Nickel: Mechanism and Kinetics at High H2/CO Ratios


Jens Sehested,* Sren Dahl, Joachim Jacobsen, and Jens R. Rostrup-Nielsen
Haldor Topse A/S, NymlleVej 55, DK-2800 Kgs. Lyngby, Denmark ReceiVed: March 23, 2004

The CO methanation reaction over nickel was studied at low CO concentrations and at hydrogen pressures slightly above ambient pressure. The kinetics of this reaction is well described by a first-order expression with CO dissociation at the nickel surface as the rate-determining step. At very low CO concentrations, adsorption of CO molecules and H atoms compete for the sites at the surface, whereas the coverage of CO is close to unity at higher CO pressures. The ratio of the equilibrium constants for CO and H atom adsorption, KCO/KH, was obtained from the rate of CO methanation at various CO concentrations. KH was determined independently from temperature programmed adsorption/desorption of hydrogen to be KH ) 7.7 10-4 (bar-0.5) exp[43 (kJ/mol)/RT] and hence the equilibrium constants for adsorption of CO molecules may be calculated to be KCO ) 3 10-7 (bar-1) exp[122 (kJ/mol)/RT]. Furthermore, the rate of dissociation of CO at the catalyst surface was determined to be 5 109 (s-1) exp[-96.7 (kJ/mol)/RT] assuming that 5% of the surface nickel atoms are active for CO dissociation. The results are compared to equilibrium and rate constants reported in the literature.
Kinetic Langmuir-Hinshelwood models (LH) for CO methanation were suggested by Yadav and Rinker,11 Ho and Harriot,12 van Meerten et al.,13 Klose and Baerns,14 and Coenen et al.,9 whereas Schoubye,15 Dalmon and Martin,16 and Hayes et al.17 found that LH models could not account for the experimental observations. Goodman et al.18-20 studied the methanation rate over well-defined surfaces. They found that the turnover frequencies on Ni(111) and Ni(100) were equal and that they compared well with methanation rates obtained on aluminasupported catalysts.6,21 The agreement was obtained for the preexponential factor and the activation energy as well. A similar strong linear decline of the methanation activity for small coverages of sulfur was found for both Ni(100)19 and a nickel alumina catalyst.6 This observation could be explained by statistical model for the chemisorption of sulfur.22 Alstrup8 proposed a microkinetic model with CH* + H* as the rate determing step. In the model the coverages of CO, hydrogen, and carbon were significant. To fit the experimental data, Alstrup had to treat the coverage of carbon as a free parameter in the model. The model reproduced the reported rate data of Goodmann and co-workers18 over nickel single crystals and of Polizzotti and Schwartz23 over polycrystalline nickel foils. Bengaard et al.24 calculated the energies by density functional theory at the closed packed Ni(111) and the stepped Ni(211) of the transition states and the stable intermediates in the elementary reactions involved in re-forming. Because re-forming is the reverse reaction of CO methanation, the energy diagram involved in this work can also be used to obtain information about the title reaction. Two conclusions may be drawn from this work: (i) Nickel steps are much more active than nickel terraces. (ii) The highest barrier is encountered for CO dissociation, suggesting that this reaction is the rate determining step. These results will be used as a guide for the kinetic model presented later. The work reported here is a study of the steady-state CO methanation kinetics over a nickel catalyst at moderate CO

1. Introduction Sabatier and Senderens1 discovered the CO methanation reaction already in the beginning of the last century. CO methanation over nickel is currently used in large ammonia plants to remove the last fraction of a percent of carbon oxides before the synthesis because the iron based ammonia catalysts are very sensitive to poisoning by oxygen. Methanation was subject to a number of studies during the 1970s2 because of an interest in producing substitute natural gas from naphtha3 and coal.4 The activation of carbon monoxide on nickel also plays a key role in metal dusting corrosion on surfaces of equipment exposed to CO-rich gases.5 There is evidence that the CO decomposition to carbon and the methanation reaction are interlinked.6 In the methanation reaction, CO and hydrogen are converted to methane and steam:

CO + 3H2 a CH4 + H2O


The reaction is the reverse steam re-forming reaction, which also proceeds over nickel catalysts.7 Therefore physical parameters as adsorption constants determined for the methanation reaction are relevant for steam re-forming. The kinetics of CO methanation have been studied intensively. Alstrup8 gave an overview of the kinetic models for CO methanation over nickel. In the seventies it was generally believed that methanation proceeds via a COHx intermediate. Coenen et al.9 found no isotopic scrambling when 13C16O and 12C18O together with hydrogen were used as the feed gas, suggesting that hydrogen or non hydrogen assisted CO dissociation is the rate determining step. Goodman et al.10 determined the rate of carbide formation in 31.6 mbar of CO and carbide hydrogenation in 131.6 mbar H2 to be similar to the rate of methanation in 157.9 mbar of CO:H2 ) 1:4 at 177 C.
Part of the special issue Michel Boudart Festschrift. * Corresponding author: Tel.: +45 45 27 23 65. Fax: +45 45 27 29 99. E-mail: JSS@topsoe.dk.

10.1021/jp040239s CCC: $30.25 2005 American Chemical Society Published on Web 10/13/2004

Methanation of CO over Nickel concentrations and over nickel threads at very low CO pressures. The hydrogen pressure was in all experiments 1.4 bar. Nickel threads were used to eliminate diffusion restrictions. Using a simple reaction mechanism with CO dissociation as the rate determining step, the ratio of the equilibrium constants, KCO/ KH and the rate constant for CO dissociation at the nickel surface were obtained. KH was determined by a fit to temperature programmed adsorption/desorption (TPA/TPD) curves for hydrogen. Finally, the equilibrium constants and the rate constant are compared to data from the literature. 2. Experimental Section Materials. A Ni/MgAl2O4 catalyst was used for the temperature programmed adsorption/desorption of hydrogen and the first set of the methanation experiments. The Ni/MgAl2O4 catalyst had a nickel surface area of 4.7 m2 Ni g-1, as determined by sulfur chemisorption according to Rostrup-Nielsen et al.25 Methanation of CO over Ni/MgAl2O4 at low CO pressures suffered from diffusion restrictions. To avoid diffusion limitations, methanation over nickel was studied using nickel threads with diameters of 4 and 22 m supplied by Bekaerd. Chemical analysis of the threads showed that they contained nickel as the only component Experimental Setups. Two experimental setups were used in the present work, and both apparatuses are described in detail elsewhere26,27 and will only be described briefly here. In the first setup, a recirculating reactor system rather than a singlepass reactor was used because a homogeneous gas phase over the nickel catalyst or nickel thread is desired. A pump delivered a recirculation flow rate of ca. 0.36 m3/h (NTP), sufficient to maintain well-mixed conditions. The catalytic reactor consisted of a removable U-shaped glass-lined stainless steel tube in which either 200 mg of nickel catalyst with a particle size of 0.3-0.6 mm or 200-400 mg of nickel thread (4 or 22 m) together with 70 mg of inert MgAl2O4 (0.3-0.6 mm) was held between wads of quartz wool. The tube had an inner diameter of 4.0 mm, which gave a catalyst bed height of 15-20 mm. With the exception of the recirculation pump, the system was heated, the reactor zone with its own oven, and the rest with a heated box. The temperature at the recirculation pump was approximately 75 C. In the first set of experiments, where methanation over the Ni/MgAl2O4 catalyst was investigated, the rate of methanation was measured using inlet flows in the range 3.5-105 mL/min (NTP) of a gas mixture containing 2.05% CO in H2. In these experiments large conversions of CO were observed and the methanation activity was determined from the consumption of CO in the reactor. Contrary, the CH4 concentration in the exit gas (in the reactor) was used to determine the activity for methanation over the nickel threads. In these experiments the conditions were chosen so that the CH4 concentration was kept below 400 ppm. The total gas flows were 100-150 mL/min (NTP) of a mixture of 2.05% CO/H2 and pure H2. The total pressure in the reactor was maintained at 1.4 bar in all experiments and the reactor effluent was monitored with a mass spectrometer and BINOS infrared CO and CO2 detectors. A standard gas containing 1000 ppm CH4 was used for calibration of the mass spectrometer. Before each run, the nickel catalyst or the nickel thread was reduced in flowing H2 (50 mL/ min NTP) at 525 C for 10 h. A blank test at 500 C using 1% CO in hydrogen showed that the reactor system itself had no significant methanation activity. The conversion of CO to CO2 was in all cases negligible (<3%) compared to conversion of CO to methane. This means that the two ways of determining

J. Phys. Chem. B, Vol. 109, No. 6, 2005 2433

Figure 1. CO methanation activity of a Ni/MgAl2O4 catalyst in a CO/ H2 atmosphere at 1.4 bar total pressure plotted as a function of pCO for six temperatures.

the methanation activity (from the change in CO concentration or from the CH4 concentration) gives equal results within experimental errors. The second experimental setup was used in the H2 TPD/TPA experiments and consisted of a single pass reactor system described in detail in Fastrup et al.27 The reactor, the loading of the catalyst, and the reduction procedure were as described for the first setup. Following reduction a flow of 100 cm3/min (NTP) of a gas containing 0.1% H2 in He was passed over the catalyst while it was allowed to cool to 240 K. While the same flow of 0.1% H2 was maintained in He, the H2 TPD was run by using a heating ramp of 8 K/min until a temperature of 773 K was reached. Subsequently, the H2 TPA was run using a cooling ramp of 8 K/min until a temperature of 273 K was reached. 3. Results and Discussion 3.1. CO Methanation over a Ni/MgAl2O4 Catalyst. In this section the rate of methanation of CO over the Ni/MgAl2O4 catalyst as a function of the CO pressure is described. The CO concentration in the reactor changed as the inlet flow of the CO/H2 gas mixture was varied, as discussed in the Experimental Section. The rate of methanation of CO determined from the conversion of CO in the reactor is plotted as a function of the CO pressure in the reactor in Figure 1. As seen from the figure, the activity increases with the CO pressure for low CO pressures and reaches a stable plateau at the higher concentrations. At high CO concentrations the reaction order in pCO is zero. A CO concentration regime where the reaction order in pCO is zero was also observed by Schoubye.15 At even higher CO concentrations a decrease in the activity and consequently negative reaction orders in pCO were reported.15 The data in Figure 1 show that the CO concentrations, at which the activity plateau is reached, increases with temperature. There could be at least two explanations for these experimental observations. First, diffusion restrictions at high conversions of CO could decrease the observed activity of the catalyst at low CO concentrations. Second, the loss of activity at low CO concentrations could be due to the intrinsic kinetic behavior of the nickel surface. The diffusion restrictions are most important at high activities and low CO concentrations due to the high conversions of CO in these cases. In Figure 2 the activities for methanation over two different sieve fractions of catalyst particles are plotted as a function of the CO concentration in the reactor. Clearly, the 600-850 m sieve fraction needs a higher CO level than the 300-600 m sieve fraction to reach a plateau in the activity. This means that the decrease in activity

2434 J. Phys. Chem. B, Vol. 109, No. 6, 2005

Sehested et al.

Figure 4. CO methanation activity at 300 C for nickel threads with diameters of 4 and 22 m, plotted as a function of pCO in an atmosphere of hydrogen.

Figure 2. CO methanation activity in a CO/H2 atmosphere of a Ni/ MgAl2O4 catalyst at 250 C for particles in the ranges 300-600 and 600-850 m, respectively, plotted as a function of the partial pressure of CO in an atmosphere of hydrogen.

have equal activity, the number of active sites can be determined from the surface area of the catalyst of 4.7 m2 g-1 and the average area of a surface nickel atom of approximately 6.5 10-20 m2 7 to be ca. 7 1019 surface nickel atoms per gram of catalyst. TOF is then given by

TOF ) 2.3 108 (s-1) exp[-96.7 (kJ/mol)/RT]

(2)

Figure 3. Logarithm of the CO methanation activity of a Ni/MgAl2O4 catalyst at high CO pressures plotted as a function of 1000/T. The slope of the plot gives a zero-order rate constant of 9.5 107 (mol g-1 h-1) exp[-96.7 (kJ/mol)/RT].

observed at low CO concentrations is due to diffusion restrictions in the catalyst particles. The plateau in the activity observed at higher CO concentrations is not subject to any significant diffusion restrictions. To support this conclusion, the Weisz-Prater criterion28 ( < 2 for a zero order reaction) was evaluated for an activity of 100 mmol g-1 h-1, a particle radius of 0.03 cm, a density of 2.3 g/cm3, a temperature of 300 C, an effective diffusion constant of 0.026 cm2/s estimated from the formulas in Satterfield,29 a porosity of 0.5, a tortuosity factor of 3, and a pore radius of 40 nm. Using these values the modulus is calculated to be 0.4, which is below the value of 2 where pore diffusion is important. The calculations therefore support the conclusion made above, that diffusion restrictions are not important at high CO pressures. It is of interest to determine the temperature dependence of the zero order rate constant for the CO methanation at high levels of CO. This is done in the Arrhenius plot in Figure 3. The reaction rate for methanation of CO over nickel in this concentration regime is determined to be

NsTOF ) 9.5 107 (mol g-1 h-1) exp[-96.7 (kJ/mol)/RT] (1)


where Ns is the number of active sites per gram of catalyst and TOF is the turnover frequency. Assuming that all nickel sites

The activation energy of 96.7 kJ/mol and the TOF for CO methanation over nickel determined here is in good agreement with those determined by Polizzotti and Schwarz23 over polycrystalline nickel foils, those reported by Vannice21 at supported nickel catalysts, and those obtained by Goodmann et al.18 at nickel single crystals. 3.2. CO Methanation over Ni Thread. To avoid the diffusion restrictions at low CO concentrations, the CO methanation rate was measured over nickel threads. In this section possible diffusion restrictions are first examined, then deactivation due to carbon formation is inspected, and finally the results are presented and discussed. Rates of CO methanation at 300 C as a function of the CO concentrations in the reactor are plotted in Figure 4 for nickel threads with 4 and 22 m diameters. The methanation rates for the two types of nickel thread have been scaled to match at high CO concentrations. As expected for nonporous materials, the trends in these curves are similar, indicating the absence of diffusion restrictions. As an additional check for film diffusion the recirculation pump of the system was closed and the reactor system behaved as a plug flow reactor. If the system was film diffusion limited, a decrease in the activity should be observed upon a decrease of the flow through the reactor at the conditions used here, where the reaction order in CO is zero and the hydrogen pressure is nearly constant. However, no significant change in the activity was observed after the recirculation pump was stopped: Hence, it was concluded that the results are not influenced by diffusion limitations and that the conversion at low CO pressures is a fingerprint of the intrinsic kinetics of the nickel surface. Deactivation of the nickel surface was observed at temperatures above 350 C. The deactivation is probably related to carbon formation and physical blocking of the surface. The accumulation of less reactive carbonaceous deposits is a wellknown phenomena in methanation studies.30,31 The deactivation was fastest at high temperatures and high CO concentrations. Figure 5 gives the time dependence of the CH4 and the CO concentrations in the reactor at the highest temperature used in this study, 400 C. The deactivation is clearly observed and is most pronounced for high CO pressures. The data were corrected for the deactivation using the following method: The increase in the methane concentration just after CO is admitted into the reactor is measured. When the CO concentration is increased

Methanation of CO over Nickel

J. Phys. Chem. B, Vol. 109, No. 6, 2005 2435

Figure 5. CH4 and CO pressures during methanation at 400 C over 22 m nickel threads plotted as a function of time.

further, the largest increase in the methane concentration is determined (CH4) and divided by the fraction of the nickel surface that is still active (R). CH4/R is then added to the previously determined CH4 concentration to obtain the methane concentration in the case of no deactivation. R is calculated by dividing the observed methane concentration with the calculated methane concentration in the case of no deactivation. In this way the effect of the deactivation is reduced to the short periods of time during which the CO and CH4 concentration increases are taking place. The corrected methane concentrations calculated from the data in Figure 5 and the derived activities are plotted in Figure 6. A series of curves of the activity versus the CO concentration similar to that in Figure 6 was obtained for temperatures in the range 275-400 C. The curves all showed increasing activity at low CO concentrations and stabilization of the activity at high CO concentrations. This shape is a consequence of the kinetics of the methanation reaction. At low CO pressures, the rate of reaction increases linearly with pCO. At high CO concentrations the rate of reaction is independent of the CO concentration, as observed in Figures 4 and 6. At intermediate CO pressures, the reaction order of CO is positive and less than one. The next step is to find a kinetic expression that can describe the activity data. The best fit was obtained using a mechanism where CO molecules compete with hydrogen atoms for type 1 sites (*) at the nickel surface and CO dissociates to a type 2 site (#), which are always free at the conditions used here. The reactions involved may be written

Figure 6. CH4 pressures and activities at 400 C during methanation over 22 m nickel threads plotted as a function of pCO. The CH4 pressures have been corrected for deactivation of the nickel surface observed during the measurements. The smooth line through the data points is a fit to the data using eq 10.

sites (Ni(211)) than at nickel terrace sites (Ni(111)). Furthermore, the carbon atom from the CO molecule dissociating at the step site arrives at the 5-fold coordinated site at the bottom of the atomic step. The 5-fold coordinated site is not energetically favorable for any other CHx (x ) 0-3), OHx (x ) 0, 1), or CO species. At this site only carbon will adsorb, and hence it is always free of other adsorbants. The type 1 and the type 2 sites in the reaction mechanism above may therefore be step edges and 5-fold coordinated sites, respectively. Hence, the number of type 1 and type 2 sites may equal to the number of step sites. It should be mentioned that the calculated barrier for CO dissociation calculated by Bengaard et al.24 is higher than that obtained in the following. Hence, the detailed reaction mechanism may be more complicated than that described here but the kinetics of the reaction reduces to that used here at our experimental conditions. The ratio of the equilibrium constants for CO molecules and H atoms will be determined from the experimental data in the following. Using the equilibrium constants for reactions 3 and 4, KCO and KH, the activity is given by

CO + * a CO* 0.5H2 + * a H* CO* + # f O* + C#

Act. ) Nsk5CO*# ) Nsk5KCOpCO* ) KCOpCO (8) Nsk5 1 + KCOpCO + KHpH20.5


CO* and * are the coverages of CO molecules and free sites at type 1 sites and # is the coverage of free sites at type 2 sites, which is assumed to be unity. k5 is the rate constant for reaction 5, Ns is the number of active (type 1 and 2) sites per gram, and Act. is the number of moles of CO molecules converted per gram of catalyst per time unit. The number of free type 1 sites is given by

KCO KH

(3) (4) (5) (6) (7)

rate determining step fast fast

C# + 2H2 f CH4 + # O* + H2 f H2O + *

Reaction 5 is the rate determining step, reactions 3 and 4 are so fast that CO* and H* are in equilibrium with the gas-phase species CO and H2, and reactions 6 and 7 are so fast that the coverages of C# and O* are negligible. This mechanism is very simple and does not describe the negative reaction orders in CO observed at high CO pressures by Schoubye15 and Polizotti and Schwarz.23 However, negative reaction orders in CO could be included in the model by assuming that CO adsorb at type 2 sites at high CO pressures. In addition, interactions between the adsorbants are neglected in the following. Some theoretical justification for the two-site model is found in the work of Bengaard et al.24 In this work a much lower dissociation barrier for CO dissociation was found at nickel step

* )

1 1 + KCOpCO + KHpH20.5

(9)

Equation 8 depends on three parameters: KCO, KH, and Nsk5. From the experimental data only two parameters can be obtained. We choose here to limit the number of parameters by assuming that the combined surface coverages of CO and H are always much higher than the fraction of free active sites, *. With this assumption eq 8 may be written

2436 J. Phys. Chem. B, Vol. 109, No. 6, 2005

Sehested et al.

KCOpCO H2 Act. Nsk5 ) Nsk5 (10) 0.5 KCO pCO KCOpCO + KHpH2 1+ KH p 0.5
H2

KCO pCO KH p 0.5

The validity of the assumption is discussed below. The data in Figure 6 may be fitted by eq 10 using Nsk5 and KCO/KH as free fitting parameters. pH20.5 1.2 bar0.5 because pH2 ) 1.4 bar. The fit of eq 10 to the activity data in Figure 6 is given as a smooth line. As seen from the figure, the fit to the data is excellent. The value of KCO/KH obtained from the fit is 639 bar-0.5 at 400 C and the value of the Nsk5 fit is 0.73 mmol g-1 h-1. The values of Nsk5 were not used for anything because the initail deactivation of the nickel surface influences the values of Nsk5. For each set of experiments, the activities versus the CO concentrations were fitted using eq 10 and values of the KCO/ KH ratio were determined. The logarithm of the KCO/KH ratios are plotted in Figure 7 as a function of 1000/T. The data form a straight line given by KCO/KH ) 4 10-4 (bar-0.5) exp[79 (kJ/mol)/RT]. The KCO/KH ratios derived by Alstrup8 and Aparicio26 from surface science studies and theoretical calculations are plotted in Figure 7 for comparison. The KCO/KH ratios determined by Alstrup are significantly lower (factors of 2.5-5 times) than those obtained here whereas the KCO/KH ratios reported by Aparicio are in good agreement with those measured here. A value of KH is determined in the following section, and using this value, one can show that H 0.60, CO 0.10, and * 0.30 at pCO ) 0.37 mbar and T ) 400 C and hence the assumption made above that H + CO . * is only partly fulfilled. Estimating the KCO/KH ratio using eq 8 and the value of KH determined in the following section as opposed to eq 10 gives KCO/KH 8 10-4 (bar-0.5) exp[77 (kJ/mol)/RT]. This change is relatively small and no attempt was made to correct for this difference. From eq 8 it is seen that the activity at high pCO/pH20.5 ratios is equal to Nsk5 where k5 is identical to the TOF determined previously of k5 ) TOF ) 2.3 108 (s-1) exp[-96.7 (kJ/mol)/ RT]. The activation energy for CO dissociation at the active nickel site is 96.7 kJ/mol. This is significantly lower that the calculated energy barriers reported by Bengaard et al.24 of 218 kJ/mol. The reason for this difference is unknown. However, as discussed previously, there is excellent agreement between the activation energy determined here and those reported over supported catalysts,21 nickel foils,23 and single crystals.18 Preexponential factors for surface reactions are normally expected to be of the order of 1013 s-1. Part of the reason for the low preexponential factor obtained here may be that only a few special sites (step sites) are expected to be active for CO dissociation, as discussed in the work of Bengaard et al.24 This effect has been observed experimentally for N2 dissociation over ruthenium32 and iron.33 The experimentally observed fraction of sites at iron that are active for ammonia synthesis is approximately 5%.33 Assuming that 5% of the nickel surface is active for CO dissociation, we arrive at a preexponential factor for reaction 5 of ca. 5 109 s-1. This preexponential factor may still seem low. However, from the following series of arguments it may be seen that the observed preexponential factor is reasonable: Alstrup8 used statistical mechanics and literature values of vibrations and rotations to calculate the equilibrium constant for CO adsorption. From the work of Alstrup,8 a temperature independent factor of the CO adsorption constant

Figure 7. Equilibrium constant ratio KCO/KH plotted as a function of 1000/T. The slope of the plot is 4 10-4 (bar0.5) exp[79 (kJ/mol)/RT]. The KCO/KH ratios used by Alstrup8 and Aparicio26 are plotted for comparison. See text for details.

at 441 C of 9 10-8 bar-1 is derived. This is close to the value obtained below of 3 10-7 bar-1. From KCO and k5 the overall rate constant in the case of low CO and H2 concentrations is calculated to be KCOk5 1.4 103 (bar-1 s-1) exp[25 (kJ/ mol)/RT] assuming that only 5% of the nickel area is active in accordance with the number of active iron sites in a commercial ammonia catalyst.33 The preexponential factor for KCOk5 may be compared to the preexponential factors for N2 dissociation over ruthenium and iron of ca. 103 bar-1 s-1 32 and ca. 2.5 103 bar-1 s-1,33 respectively. In ammonia synthesis, N2 is dissociating over a ruthenium or iron step site like CO is dissociating over a nickel step site; hence it is expected that the entropy changes from the gas-phase molecule to the transition state are similar, as observed. If the temperature independent parts of KCO and KCOk5 are reasonable, the preexponential factor of k5 is also correct. The reason for the low preexponential factor of k5 is probably that the transition state for CO dissociation is immobile, whereas the frustrated translations of CO at nickel have relative low energies (0.40.45 kJ/mol).8 As discussed in the Introduction a lot of kinetic models and mechanisms have been proposed for CO methanation over nickel. Interestingly, Goodman et al.10 found that the rates of CO dissociation and hydrogenation of nickel carbide on Ni(100) compete at 177 C. In most other experimental work reported in the literature, low H2:CO ratios were used. For example, Goodman et al.,18 Polizzotti and Schwartz,23 and Vannice21 used H2:CO ratios of 4:1, 1:1-30:1, and 3:1, respectively. The experimental conditions used here with high hydrogen and low CO pressures favor hydrogenation of carbon at the nickel surface, whereas the low CO pressures tend to slow the rate of CO methanation. It therefore seems reasonable that CO dissociation is the rate determining step at the experimental conditions used here. One could speculate that at the experimental conditions used by many authors with high CO:H2 ratios, the kinetis may be more complicated and more than one rate determining step may be active. Understanding these results will therefore be difficult compared to the interpretation of the data reported here. 3.3. Hydrogen Adsorption/Desorption Experiment. When the hydrogen TPD and TPA experiments are analyzed, it is assumed that the surface hydrogen is in equilibrium with the gas-phase hydrogen. Thus it is the equilibrium constant KH as a function of temperature that can be determined from the experiments. TPD and TPA traces are determined from KH by assuming that the reactor is well described as a differential reactor where the H2 pressure is an average of the inlet and outlet pressure. The relative small changes in the H2 pressure observed (Figure 8) makes this a good approximation.

Methanation of CO over Nickel


KH (0.5H2 + * a H*), bar-0.5 5.5 10 exp[47.5 (kJ/mol)/RT] 1.8 10-2 exp[25 (kJ/mol)/RT] 7.7 10-4 exp[43 (kJ/mol)/RT]
-3

J. Phys. Chem. B, Vol. 109, No. 6, 2005 2437


KCO (CO + * a CO*), bar-1 2 10-5 exp[115 (kJ/mol)/RT] 9 10-8 exp[117 (kJ/mol)/RT] 3 10-7 exp[122 (kJ/mol)/RT] 5 10-7 exp[130 (kJ/mol)/RT] 10-6 exp[127 (kJ/mol)/RT]

TABLE 1: Comparison between the Values of KH and KCO Obtained Here and Data from the Literature
ref 26 8 this work 34 35

The results from the TPD and the TPA experiment are depicted in Figure 8 together with the predictions obtained when KH )

the values of KCO and KH used by Aparicio are more than an order of magnitude higher than those reported here. For example, the hydrogen adsorption constant at 300 C estimated by Aparicio26 is approximately 20 times higher than that reported here and that estimated by Alstrup.8 4. Conclusions CO methanation and hydrogen adsorption/desorption over a Ni/MgAl2O4 catalyst and CO methanation over nickel threads were studied. The CO methanation experiments over Ni/ MgAl2O4 catalysts were subject to diffusion restrictions at low CO concentrations and high conversions whereas diffusion restrictions were not important in the CO methanation over nickel threads. The adsorption constant ratio KCO/KH was determined from the CO methanation experiments over nickel thread to be 4 10-4 (bar-0.5) exp[79 (kJ/mol)/RT]. KH were determined using a hydrogen adsorption/desorption experiment to be 7.7 10-4 (bar-0.5) exp[43 (kJ/mol)/RT]. From KCO/KH and KH, KCO is calculated to be 3 10-7 (bar-1) exp[122 (kJ/ mol)/RT]. The values of KCO and KH obtained in this work are in general agreement with the adsorption constants reported in the literature. The first-order rate constants for CO* dissociation at the catalyst surface were 5 109 (s-1) exp[-96.7 (kJ/mol)/ RT] assuming that 5% of the nickel surface atoms are active. One may question the justification of using the Langmuir approach for deriving the kinetic expression used here, knowing the complexity of real surfaces and that different sites are involved in the kinetic sequence.36 Nevertheless, even with a distribution of heats of adsorptions, the resulting kinetics may be close to that obtained here by assuming that the reaction takes place on only the most active sites. This is the background for what Boudart almost 50 years ago termed the paradox of heterogeneous catalysis,37 stating that irrespective of the complexity of the catalytic surface, very often kinetic data can be represented by a Langmuir isotherm even though assumptions are far from reality. Acknowledgment. We thank Said Rokni and Claus Thomsen for technical assistance. References and Notes
(1) Sabatier, P.; Senderens, J. B. C. R. Acad. Sci., Paris 1902, 134, 514. (2) Vannice, M. A. In Catalysis, Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer: Berlin, 1982; Vol. 3, p 139. (3) Bresler, S. A.; Ireland, J. D. Chem. Eng. 1972, 1, 94. (4) Harms, A.; Ho hlein, B.; Jrn, E.; Skov, A. Oil Gas. J. 1980, 78 (15), 120. (5) Grabke, H. J. Mater. Corr. 1998, 49, 317. (6) Rostrup-Nielsen, J. R.; Pedersen, K. J. Catal. 1979, 59, 395. (7) Rostrup-Nielsen, J. R. In Catalysis, Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: Berlin, 1984; Vol. 5, Chapter 1. (8) Alstrup, I. J. Catal. 1995, 151, 216. (9) Coenen, J. W. E.; van Nisselroy, P. F. M. T.; de Croon, M. H. J. M.; van Dooren, P. F. H. A.; van Meerten, R. Z. C. Appl. Catal. 1986, 25, 1. (10) Goodmann, D. W.; Kelley, R. D.; Madey, T. E.; White, J. M. J. Catal. 1980, 64, 479.

Figure 8. H2 adsorption/desorption experiment over a Ni/MgAl2O4 catalyst. The upper experimental curve gives the concentration of hydrogen in the exit gas during a temperature ramp of 8 C/min in the presence of 100 mL/min (NPT) of 0.1% H2 in He. The lower experimental curve gives the hydrogen concentration in the same gas and the inverted temperature ramp. The smooth lines are fits to the experimental data. See text for details.

7.7 10-4 (bar-0.5) exp[43 (kJ/mol)/RT], which is judged to give the best agreement with the experimental results. Better agreement between experiment and predictions could be obtained by assuming that two or more types of sites with different KH exist on the nickel surface. The fact that it is possible to describe the TPA and the TPD experiment means that the assumption of gas-phase surface equilibrium is reasonable. 3.4. Comparison of KCO and KH with Data from the Literature. From the values of KH and KCO/KH, a value of KCO ) 3 10-7 (bar-1) exp[122 (kJ/mol)/RT] is determined. In Table 1 the equilibrium constants KH and KCO are compared to selected experimental data from the literature34,35 and data estimated from surface science experiments and used for mikrokinetic modeling of the re-forming reaction26 and the CO methanation reaction.8 Aparicio gives the equilibrium constants directly. Alstrup8 only gives the enthalpy of the reaction and the preexponential factors were obtained from the calculated coverages of hydrogen atoms and CO molecules at 120 Torr total pressure and 441 C given by Alstrup.8 The equilibrium constant from Stuckless et al.34 was derived from the initial sticking coefficient at 27 C (0.72), the preexponential factor for desorption (3 1014 s-1), and the initial calorimetric heat of adsorption (130 kJ/mol) at Ni(111). Finally, KCO were derived from Takagi et al.35 using the heat of adsorption of 127 kJ/mol, the initial sticking coefficient at 350 K(0.9), and preexponential factor for desorption (1.6 1014 s-1). The equilibrium constant for CO adsorption determined here is lower than all the other equilibrium constants apart from that determined by Alstrup.8 However, the equilibrium constants in the literature represent values at low coverages of CO and there was no hydrogen present in measurements of Stuckless et al.34 and Takagi et al.35 Even though the KCO/KH ratio determined by Aparicio26 is in excellent agreement with that obtained here,

2438 J. Phys. Chem. B, Vol. 109, No. 6, 2005


(11) Yadav, R.; Rinker, R. G. Can. J. Chem. Eng. 1993, 71, 202. (12) Ho, S. V.; Harriott, P. J. Catal. 1980, 64, 272. (13) van Meerten, R. Z. C.; Vollenbrock, J. G.; de Croon, M. H. J. M.; van Nisselroy, P. F. M. T.; Coenen, J. W. E. Appl. Catal. 1982, 3, 29. (14) Klose, J.; Baerns, M. J. Catal. 1984, 85, 105. (15) Schoubye, P. J. Catal. 1969, 14, 238. (16) Dalmon, J. A.; Martin, G. A. J. Catal. 1983, 84, 45. (17) Hayes, R. E.; Thomas, W. J.; Hayes, K. E. J. Catal. 1985, 92, 312. (18) Goodmann, D. W.; Kelley, R. D.; Madey, T. E.; Yates, J. T. J. Catal. 1980, 63, 226. (19) Goodman, D. W.; Kiskinova, M. Surf. Sci. Lett. 1981, 105, 265. (20) Kelly, R. D.; Goodman, D. W. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.; Elsevier: Amsterdam, 1982; Vol. 4, p 427. (21) Vannice, M. A. J. Catal. 1976, 44, 152 (22) Alstrup, I.; Andersen, N. T. J. Catal. 1987, 104, 466. (23) Polizzotti, R. S.; Schwarz, J. A. J. Catal. 1982, 77, 1. (24) Bengaard, H. S.; Nrskov, J. K.; Sehested, J.; Clausen, B. S.; Nielsen, L. P.; Molenbrock, A. M.; Rostrup-Nielsen, J. R. J. Catal. 2002, 209, 365.

Sehested et al.
(25) Rostrup-Nielsen, J. R.; Sehested, J.; Nrskov, J. K. AdV. Catal. 2002, 47, 65. (26) Aparicio, L. M. J. Catal. 1997, 165, 262. (27) Fastrup, B.; Muhler, M.; Nygrd Nielsen, H.; Pleth Nielsen, L. J. Catal. 1993, 142, 135. (28) Froment, G. F.; Bischoff. Chemical Reactor Analysis and Design; John Wiley and Sons: New York, 1979. (29) Satterfield, C. N. Mass Transfer in Heterogeneous Catalysis; MIT Press: Cambridge, MA, 1970. (30) Gierlich, H. H.; Frenery, M.; Skov, A.; Rostrup-Nielsen, J. R. Stud. Surf. Sci. Catal. 1980, 6, 459. (31) McCarthy, J. G.; Wise, H. J. Catal. 1979, 57, 406. (32) Dahl, S.; To rnqvist, E.; Chorkendorff, I. J. Catal. 2000, 192, 381. (33) Dahl, S.; To rnqvist, E.; Jacobsen, C. J. H. J. Catal. 2001, 198, 97. (34) Stuckless, J. T.; Al-Sarrat, N.; Wartnaby, C.; King, D. A. J. Chem. Phys. 1993, 99, 2202. (35) Takagi, N.; Yoshinobu, J.; Kawai, M. Chem. Phus. Lett. 1993, 215, 120. (36) Rostrup-Nielsen, J. R. Catal. Today 1994, 22, 295. (37) Boudart, M. AIChE J. 1956, 2, 62.

You might also like