Methanol±steam reforming on Cu/ZnO/Al 2 O 3. Part 1: the reaction network

Transcription

1 Applied Catalysis A: General 179 (1999) 21±29 Methanol±steam reforming on Cu/ZnO/Al 2 O 3. Part 1: the reaction network Brant A. Peppley *, John C. Amphlett, Lyn M. Kearns, Ronald F. Mann Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston Ont., Canada K7K 7B4 Received 21 January 1998; received in revised form 7 July 1998; accepted 18 August 1998 Abstract On-board generation of hydrogen by methanol±steam reforming on Cu/ZnO/Al 2 O 3 catalyst is being used in the development of fuel-cell engines for various transportation applications. There has been disagreement concerning the reactions that must be included in the kinetic model of the process. Previous studies have proposed that the process can be modelled as either the decomposition of methanol followed by the water-gas shift reaction or the reaction of methanol and steam, to form CO 2 and hydrogen, perhaps followed by the reverse water-gas shift reaction. Experimental results are presented which clearly show that, in order to explain the complete range of observed product compositions, rate expressions for all three reactions (methanol±steam reforming, water-gas shift and methanol decomposition) must be included in the kinetic analysis. Furthermore, variations in the selectivity and activity of the catalyst indicate that the decomposition reaction occurs on a different type of active site than the other two reactions. Although the decomposition reaction is much slower than the reaction between methanol and steam, it must be included in the kinetic model since the small amount of CO that is produced can drastically reduce the performance of the anode electrocatalyst in low temperature fuel cells. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Kinetics; Multiple reactions; Methanol±steam reforming; Fuel cells; Copper catalysis 1. Introduction Methanol and steam in the presence of a Cu/ZnO/ Al 2 O 3 catalyst at temperatures greater than 1608C react to form a hydrogen-rich gas. A number of studies of this process have shown that the major products are H 2 and CO 2 with minor quantities of CO also being produced [1±3]. Although methane formation is thermodynamically favoured for certain operating conditions, the selectivity of copper-based catalysts appears *Corresponding author. peppley-b-@rmc.ca to completely avoid the formation of this by-product [4,5]. The three overall reactions which can be written for the given reactants and products are: CH 3 OH H 2 k R CO H 2 k W k R CO 2 3H 2 k W CO 2 H 2 (I) (II) CH 3 k D CO 2H 2 (III) k D where the subscript `R' refers to steam reforming, `W' the water-gas shift (WGS) reaction and `D' refers to X/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S X(98)

2 22 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 decomposition. It should be noted that, algebraically, reaction (I) is the sum of the other two reactions. The hydrogen-rich product gas produced by methanol±steam reforming, after further processing, is an ideal feed for fuel cells. A kinetic model of methanol± steam reforming on Cu/ZnO/Al 2 O 3 catalyst is, therefore, an extremely useful tool for designing fuel processors for methanol-fuelled fuel-cell power plants. One of the most promising types of fuel cells, the low-temperature proton-exchange-membrane (PEM) fuel cell, is extremely susceptible to poisoning by low levels of CO in the anode feed gas. It is therefore extremely desirable to have a comprehensive kinetic model that is not only able to predict the rate of H 2 production but also the extent of CO contamination in the product gas of the methanol±steam reformer. There has been general disagreement concerning the reactions that must be included in the kinetic model of the process of methanol±steam reforming on Cu/ZnO/Al 2 O 3 catalyst. Previous studies have suggested that the kinetics could be adequately modelled using only one or two of the three possible overall reactions while assuming that the other reactions were either at equilibrium or that their rates were negligible. This paper will present results from an extensive experimental study of methanol±steam reforming which show that rate expressions for reactions (I), (II) and (III) must all be included in the kinetic model to accurately predict the composition of the product gas Previous studies of methanol±steam reforming Early studies of methanol±steam reforming assumed that the process was simply the reverse of methanol synthesis. It was believed that methanol was synthesised by the direct hydrogenation of CO; hence, methanol steam reforming was believed to proceed by the formation of CO and hydrogen, followed by the water-gas shift reaction [2,3,5,6]. Some of these studies assumed that the concentration of CO was always at equilibrium with respect to the water-gas shift reaction [2,3]. Later, as the idea that methanol is synthesised by the hydrogenation of CO 2 developed, the reaction scheme for methanol±steam reforming also changed correspondingly. The formation of CO 2 by the direct reaction of methanol and steam was proposed in several papers [7±9] and different opinions concerning the importance of the water-gas shift reaction evolved. The direct decomposition of methanol via reaction (III) was considered negligible by several researchers [7,8] Related studies of the reaction network There has been a limited amount of published research that provides information speci cally concerning the reaction scheme for methanol±steam reforming on Cu/ZnO/Al 2 O 3 catalyst. In a study by Vanderborgh et al. [10], a Cu/ZnO/Al 2 O 3 catalyst was prepared containing only labelled oxygen ( 18 O). Upon addition of unlabelled methanol and steam it was found that doubly labelled carbon dioxide (C 18 O 18 O) was immediately detected in the product stream and its proportion increased to a maximum at which point 90% of the carbon dioxide produced was doubly labelled. C 18 O, however, was never detected in the product stream indicating that the CO bond in methanol is not broken in the mechanism of decomposition. These results indicate that there is a separate reaction path for the CO producing reaction that, unlike the CO 2 producing reaction, does not involve oxygen interchange with the catalyst. Research concerning the mechanism of methanol synthesis on Cu/ZnO/Al 2 O 3 catalyst also provides some information concerning the mechanism of methanol±steam reforming. Temperature programmed decomposition of methanol adsorbed on a partially oxidised Cu/ZnO/Al 2 O 3 catalyst at room temperature was done as part of an investigation of the mechanism of methanol synthesis [11]. The reaction spectrum of the products was obtained using a mass spectrometer and showed that CO 2 and H 2 were produced simultaneously at 440 K. From previous experiments this peak could be assigned to the decomposition of an adsorbed formate species on copper sites. At 580 K, CO and H 2 were simultaneously evolved, this being attributed to the decomposition of a formate species on ZnO. The conclusion of this study was that methanol synthesis on Cu/ZnO/Al 2 O 3 catalyst occurs by parallel mechanisms. One involves the hydrogenation of CO 2 and a second involves the hydrogenation of CO. The rate of CO hydrogenation, however, has been shown to be only 1/100th the rate of CO 2 hydrogenation. As will be discussed below, the relative rates for the synthesis reactions are in accord

3 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 23 with the rates observed for the reverse reactions during methanol±steam reforming. 2. Experimental 2.1. Catalyst The BASF K3-110, Cu/ZnO/Al 2 O 3 catalyst, used in this work is sold as a commercial low-temperature shift catalyst. Table 1 summarises the basic characteristics of the catalyst. The surface area, pore volume, copper area and dispersion were determined from volumetric adsorption data obtained using an Autosorb 1-C automated chemisorption apparatus. Copper area was calculated as the difference between strongly chemisorbed CO and weakly chemisorbed CO after the method proposed by Parris and Klier [12]. The crosssectional area of each Cu atom used in this calculation was AÊ 2. The dispersion was calculated as the percentage of surface copper atoms compared to the total number of copper atoms in the bulk of the sample. The ratio of CO molecules per Cu atom chemisorbed used in this calculation was 4 [12]. The average crystallite size was calculated assuming that the surface copper crystallites are of a hemispherical geometry Reactor A conventional isothermal xed-bed tubular reactor was used for all kinetic measurements. The reactor Table 1 Physical characteristics of BASF K3-110 catalyst Characteristic Value wt% CuO 40 a wt% ZnO 40 a wt% Al 2 O 3 20 a Nitrogen BET area 1024m 2 g 1 Pore volume 0.35 ml g 1 Copper area b 9.83 m 2 g 1 Dispersion b 4.8% Copper crystallite size b 219 AÊ a These are nominal values provided by the vendor. The exact compositions and their variability are proprietary information (reference: BASF Technical Leaflet RCK/M 02.88). b These results were obtained using an Autosorb 1-C automated chemisorption apparatus and the method described by Parris and Klier [12]. was operated in an integral mode with methanol conversions greater than 10% to examine the effects of pressure, temperature and feed composition on catalyst activity, selectivity and the rate of deactivation. Reaction rates were also measured directly by the differential method in which the change in the partial pressure of any single component over the length of the reactor ranged from 5% to 15%. The reactor consisted of a 30 cm length of stainlesssteel pipe with a nominal i.d. of 2.21 cm. Approximately 10 cm from the exit of the reactor a 40 mesh stainless-steel screen was welded across the pipe to support the catalyst bed. At the centre of the catalyst support a stainless steel collar was welded such that a shielded thermocouple of cm o.d. protruded approximately 1 cm up into the bed. The temperature reading of this thermocouple was used as the catalyst temperature for all analyses. This thermocouple entered from the bottom of the reactor and was positioned along the axis of the reactor. A small but consistent temperature difference of less than 1 K was observed between the wall of the reactor and the catalyst and the temperature difference was always consistent with whether the overall reaction was exothermic or endothermic. A second thermocouple entered the reactor from above and measured the temperature on the centreline of the reactor approximately 3 cm above the beginning of the catalyst bed. A third thermocouple entered from the bottom of the reactor and measured the temperature approximately 2 cm below the exit of the bed Catalyst bed geometry The bed from exit to entrance consisted of a 6 mm exit zone of 20±25 mesh inert a-al 2 O 3 particles, a 40 mm reaction zone containing 0.07±0.4 g of 20±25 mesh catalyst particles mixed with 20±25 mesh inert a-al 2 O 3 and a 15 mm entrance zone of 20±25 mesh inert a-al 2 O 3 on top of the bed to condition the ow. The con guration of the packed bed was designed to minimise thermal gradients and concentration gradients in the bed and inside the catalyst particles. The criteria of Mears [13] were used to ensure that temperature gradients and axial dispersion were negligible. The Weiss±Prater criterion [14] was used to ensure that there were no signi cant internal mass transfer gradients.

4 24 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21± Catalyst activation and conditioning The catalyst as purchased is in a calcined oxidised state and must be activated by treatment in a reducing atmosphere. The procedure used for activation was the same as that used by Jiang et al. [8] and Idem and Bakhshi [15]. These methods in turn were modi ed versions of the method originally proposed by Amphlett et al. [5]. With the system at atmospheric pressure, 1:1 mol ratio water-to-methanol was fed at a liquid ow rate of 3mlh 1. At 513 K the initiation of the reaction was almost instantaneous and a small temperature rise of approximately 5 K was typically observed due to the exothermic heat of reduction. The operating conditions were held constant for at least 2 h until the catalyst temperature, conversion and product composition had stabilised. It was consistently observed, however, that the concentration of CO in the dry product gas slowly continued to increase. It was therefore often necessary to condition the catalyst for several days before the proportion of CO in the dry product gas increased to a level typically observed for the long term catalyst behaviour Gas analysis The ef uent of the reactor consisted of H 2,CO 2 and CO as well as unreacted CH 3 OH and H 2 O. In preparation for analysis, the xed gas components were separated from the unreacted condensable reactants by passing the stream through a condenser which was cooled to a few degrees above the freezing point of the condensed methanol±water mixture. Samples of the dry gas mixture leaving the reactor were analysed using a GOW-MAC 550P programmable gas chromatograph (GC). The gas mixture fed to the reactor could also be sampled and analysed by the GC. The change in the gas composition due to reaction was therefore determined by the difference of gas analyses using the same GC and the same procedure thus minimising any experimental error due to measurement biases. The separation of the components in the gas sample was done on a 245 cm long by 3.2 mm Carbosieve-S column. The GC was programmed as follows: 2 min at 658C, ramp 208C min 1 to 2008C, held at 2008C for 2 min. The detector temperature was 2008C and the bridge current was set at 200 ma. The carrier gas ow was 50 ml min 1 of 9% H 2 in helium. This carrier gas mixture ensures that the signal for the H 2 peak is completely negative compared to the reference signal. A computer controlled sampling and switching system automatically reversed the polarity of the thermal conductivity detector signal for the hydrogen peak so that all peaks appeared as positive signals with respect to the baseline Experimental program An examination of reactions (I)±(III) reveals that there are ve different component partial pressures that can be varied. Temperature can also be varied resulting in a total of six experimental variables. The maximum operating pressure of the reactor was 35 bar and the maximum recommended operating temperature of the catalyst is 533 K. Based on these operating constraints and the stoichiometry of the reactions, six inequalities (shown in Table 2) were written to de ne the experimental operating region. Experiments were then designed using the D-optimal criterion for a multi-response model [17]. This criterion chooses the experimental conditions that result in the greatest improvement in the accuracy of the parameter estimates for the model that is being tested. Approximately 60 experiments were designed and performed. For each experiment the rates of formation of H 2 and CO were measured directly. With these two rates measured, and knowing the molar feed rates of methanol and water, the rates of formation or disappearance of all the other components could be determined. After each kinetic experiment, a baseline measurement of methanol conversion versus W=F CH3 OH;0 was done to check that the catalyst activity and selectivity had remained stable during the rate determinations. The baseline conditions were Table 2 Inequalities used to define experimental operating region 433 KReactor temperature533 K 1 barreactor pressure35 bar 0 bar<partial pressure of CO0.5 bar 0.29<Ratio of partial pressure of CO 2 to H <Ratio of partial pressure of CH 3 OH to H 2 O Ratio partial pressure of H 2 to H 2 O130

5 arbitrarily de ned as 513 K, S/M ratio of 1, and total pressure of 1.01 bar. These baseline measurements revealed that the catalyst performance changed signi cantly when the experimental conditions were strongly reducing. As a result 17 runs were eliminated resulting in a total of 43 useable runs. Details of these data can be found elsewhere [16]. 3. Results and discussion Fig. 1 is an example of typical methanol conversion versus W=F CH3 OH;0 results for a g catalyst bed. Prior to these data being collected the catalyst had been in continuous operation at 553 K for 110 h. As can be seen, the rate of reaction, which is represented by the slope of the plots of methanol conversion versus W=F CH3 OH;0, increased as the reactor pressure was increased from 1 to 16 bar. The curves shown in Fig. 1 are simply least-squares polynomials with zero intercept. They have been included to show the trend of reaction rate with pressure more clearly Importance of water-gas shift reaction to catalyst selectivity The product gas composition, relative to the WGS equilibrium, can be represented as B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 25 W ˆ PCO 2 P H2 K W : (1) P CO P H2 O The parameter, W, is a measure of the departure of the product composition from the water-gas shift reaction equilibrium. Its value will be equal to 1.0 when the gas composition is at WGS equilibrium, less than 1.0 when the partial pressure of the products of reaction (II) are less than equilibrium and greater than 1.0 when the partial pressures of the products are greater than equilibrium. W is an indication of the driving force for the reaction. When Wˆ1.0 the rate of reaction (II) will be zero. When W >0 the rate of reaction (II) will be negative and vice versa. This approach for characterising the driving force of the water-gas shift reaction has been previously used effectively in the study of low-pressure methanol synthesis on Cu/ZnO/Al 2 O 3 catalyst [17]. The signi cance of the rate of the WGS reaction to the rate of CO formation, during methanol±steam reforming, was determined by correlating kinetic measurements with the value of the W parameter. Fig. 2 shows the rate of CO production plotted against the equilibrium parameter W for various temperatures. The trend lines in Fig. 2 clearly show that the rate of CO production tends to zero as the WGS reaction approaches equilibrium even though the partial pressure of methanol for these experiments was still much greater than equilibrium with respect to Fig. 1. Fractional conversion of methanol versus W=F CH3OH;0 for BASF K3-110 at 553 K, S/M ratioˆ1.36, 20±25 mesh catalyst particles diluted in a-al 2 O 3 particles of the same mesh size.

6 26 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 Fig. 2. Rate of CO production relative to the WGS shift equilibrium for various temperatures and catalyst beds. either reactions (I) or (III). A similar comparison of the rate of CO production with the equilibrium of the decomposition reaction did not reveal a correlation. Even though there is a considerable amount of scatter in Fig. 2, the fact that the rate measurements all converge to zero at WGS equilibrium proves that the rate of the WGS reaction is a signi cant factor in determining the rate of CO formation (or disappearance) during methanol steam reforming. This also indicates that the rate of the decomposition reaction is much less signi cant than the rate of the WGS reaction in determining the rate of CO formation for the conditions of the experiments shown Variation of W with methanol conversion The observation that the rate of the decomposition reaction is much less signi cant compared to the rate of the water-gas shift reaction in determining the rate of CO formation agrees with the observations of DuÈmpelmann [7]. However, if the decomposition reaction were negligible, then the value of W as de ned in Eq. (1) should always be greater than unity. Amphlett et al. [9], however, showed that W is less than one at low W=F CH3 OH;0 for methanol±steam reforming at 473 K using a Cu/ZnO/Al 2 O 3 catalyst (United Catalyst C18HC). As shown in Fig. 3, at baseline methanol±steam reforming conditions as described above, W was found to be less than unity at low methanol conversion using BASF K3-110 catalyst. These results are consistent with those presented by Amphlett et al. [9] since low conversion corresponds to low W=F CH3 OH;0. The only explanation for the observed variation in W is that the rate of decomposition, reaction (III), must be signi cant compared to the rate of the other two reactions at very low conversions but decreases rapidly as the partial pressures of the products increase. In order to model the variation in partial pressure of CO throughout the catalyst bed, therefore, the decomposition reaction must be included in the kinetic model and cannot be considered negligible as was assumed by Jiang et al. [8] and DuÈmpelmann [7]. Including both the decomposition reaction and the water-gas shift reaction, in addition to the methanol± steam reaction, in the reaction network signi cantly increases the complexity of the kinetic analysis. However, as stated above, the extreme sensitivity of lowtemperature fuel cells to CO poisoning makes the accurate prediction of the CO contamination a critical issue when methanol±steam reforming is used for fuel-cell hydrogen production Variability of selectivity and activity: implications concerning active sites Fig. 4 shows the variation with time-on-line of the baseline ow rates of CO 2 and CO product formation.

7 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 27 Fig. 3. W versus fractional conversion for various beds of BASF K3-110 Catalyst at baseline conditions. Temperatureˆ513 K, pressureˆ1.01 bar abs., S/M ratioˆ1.0 molar. These data were collected at the end of each day of kinetic measurements to check that the catalyst activity had not changed signi cantly during the experiment. As can be seen, the rate of CO 2 production rapidly decreased initially with time-on-line followed by a period when the activity was relatively stable. The rate of CO production at baseline conditions was quite steady once it had risen during the rst 24 h of operation. After 170 h on line, a kinetic experiment was done in which the partial pressure of CO in the feed was set to 3.3 bar. This partial pressure is considerably greater than would normally be encountered during methanol±steam reforming but was chosen in an attempt to obtain better data for estimating kinetic parameters for the WGS reaction. It is not unusual for the partial pressure of CO to be this high in a low temperature shift reactor where these types of Cu/ ZnO/Al 2 O 3 catalyst are typically used and hence no signi cant permanent change in the condition of the catalyst was anticipated. Fig. 4. Effect of treatment with 3.3 bar partial pressure of CO on catalyst activity and selectivity. Conditions for all measurements are: temperatureˆ513 K, pressureˆ1.01 bar abs., S/M ratioˆ1.0 molar, W=F CH3OH;0 ˆ3.2 kg s mol 1.

8 28 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 As can be seen in Fig. 4, after the experiment which exposed the catalyst to 3.3 bar of CO, the baseline rate of CO production increased from 1 mmol s 1 kg 1 to approximately 10 mmol s 1 kg 1. At the same time the baseline rate of CO 2 production decreased from 18 to 12 mmol s 1 kg 1. This resulted in approximately 12% CO by volume in the dry product gas, an amount more than nine times greater than had ever previously been observed for these conditions and more than two times greater than any level reported in the literature. To con rm that the increased rate of CO production was not simply caused by excess CO desorbing from the surface, the product composition at baseline conditions was monitored for a further 70 h of continuous operation. There was no evidence that the change in catalyst selectivity was reversible and the mass of CO produced during this period was several times the mass of the catalyst bed indicating that the CO was not simply desorbing from the surface. Clearly, the treatment with CO caused a signi cant change in some fundamental characteristic of the catalyst such that the rate of the decomposition reaction was signi cantly increased. A similar change in the selectivity of the catalyst was observed after exposure to a low S/M ratio of 0.2 and after exposure to high temperature: Tˆ3008C, S/Mˆ1.0, timeˆ10 h. Both of these situations would create a highly reducing atmosphere similar to the treatment with 3.3 bar of CO. Fig. 4 provides information concerning the form of the reaction scheme required in the kinetic model for methanol±steam reforming. The treatment with CO caused the catalyst performance to change in a complex way. The rate of production of CO 2 at baseline conditions decreased while the rate of CO production increased. This cannot simply be due to a decrease in surface area or the rates of CO and CO 2 production both would have decreased. Clearly, the reaction path for the CO 2 producing processes, reactions (I) and (II), must be occurring on different active sites than the CO producing process, reaction (III). The behaviour observed in Fig. 4 can be explained as a decrease in the number of sites for the CO 2 producing reactions and an increase in the number of sites for the CO producing reaction. In considering the steam-reforming reaction and the WGS reaction the situation is less conclusive. The rate of the WGS reaction was measured separately (e.g. with no methanol in the feed) before and after treatment with the reaction mixture containing 3.3 bar of CO. It was found that the rate of the WGS reaction decreased to approximately one-third of the rate immediately prior to the CO treatment. The total rate of the WGS reaction, however, was only a fraction of the total rate of CO 2 production both before and after the CO treatment; therefore, the observed rate of CO 2 production is primarily due to the direct methanol± steam reaction. However, the reforming reaction and the WGS reaction both showed a decrease in rate and therefore it is possible that they share a common type of active site. The magnitudes of the changes in the activity are signi cantly different and it may be that the methanol±steam reaction and the WGS reaction are occurring on different types of active sites, both of which deactivate rapidly in strongly reducing atmospheres. In developing a surface mechanism for these two reactions, the assumption that the WGS reaction and the reforming reaction share a common active site greatly simpli es the analysis. Subsequent work therefore used this assumption with the caveat that a third site speci cally for the WGS reaction might be required if the analysis was unable to explain the kinetic data [18]. 4. Conclusions A reaction network for modelling the process of methanol±steam reforming has been proposed. The complex form for this reaction network is necessitated by the exigencies of providing hydrogen for a lowtemperature fuel cell system. In order to be able to account for the range of product compositions which were observed, it is necessary that the kinetic model include the rates of all three reactions: methanol± steam reforming, water-gas shift and methanol decomposition. Previous reaction schemes which involved only one or two of the possible overall reactions are unable to explain the experimentally observed variation in the product composition. The nal expressions for predicting the rate of production or disappearance of each of the reacting components are as follows: r H2 ˆ 3r R 2r D r W S A mol s 1 kg 1 ; (2) r CO ˆ r D r W S A mol s 1 kg 1 ; (3)

9 B.A. Peppley et al. / Applied Catalysis A: General 179 (1999) 21±29 29 r CO2 ˆ r R r W S A mol s 1 kg 1 ; (4) r CH3 OH ˆ r R r D S A mol s 1 kg 1 ; (5) r H2 O ˆ r R r W S A mol s 1 kg 1 ; (6) where r R, r W and r D are the surface area speci c rates of reactions (I)±(III), respectively. The observed variation in catalyst activity and selectivity indicates that there is an active phase for methanol decomposition which is distinct from the active phase for methanol±steam reforming and the water-gas shift reaction. 5. Nomenclature F CH3 OH;0 k i K W p i r i S A S/M W W Acknowledgements molar flow rate of methanol in feed to reactor (mol s 1 ) rate constant for reaction `i'; units will be specific to the form of the rate expression equilibrium constant of WGS reaction partial pressure of component `i' (bar) rate of reaction `i' (mol s 1 m 2 ) or rate of formation of component `i' (mol s 1 (kg of catalyst) 1 ) specific surface area of the catalyst (m 2 kg 1 ) molar ratio of steam to methanol in feed to reactor mass of catalyst (kg) parameter relating composition of reactor effluent to the WGS equilibrium defined in Eq. (1) Support to this work from the Canadian Department of National Defence, Chief of Research and Development is gratefully acknowledged. References [1] J.A. Christiansen, A reaction between methyl alcohol and water and some related reactions, J. Am. Chem. Soc. 43 (1921) [2] A.P. Meyer, J.A.S. Bett, G. Vartanian, R.A. Sederquist, Parametric analysis of 1.5 kw methanol-fuel cell power plant designs, US Army Technical Report DAAK70-77-C-0195, [3] E. Santacesaria, S. CarraÁ, Cinetica dello steam reforming del metanolo, Riv. Combust. 32 (1978) 227±232. [4] J.C. Amphlett, M.J. Evans, R.A. Jones, R.F. Mann, R.D. Weir, Hydrogen production by the catalytic steam reforming of methanol. Part 1: The thermodynamics, Can. J. Chem. Eng. 59 (1981) 720±727. [5] J.C. Amphlett, M.J. Evans, R.F. Mann, R.D. Weir, Hydrogen production by the catalytic steam reforming of methanol. Part 2: Kinetics of methanol decomposition using girdler G66B catalyst, Can. J. Chem. Eng. 63 (1985) 605±611. [6] J.C. Amphlett, M.J. Evans, R.F. Mann, R.D. Weir, Hydrogen production by the catalytic steam reforming of methanol. Part 3: Kinetics of methanol decomposition using girdler C18HC catalyst, Can. J. Chem. Eng. 66 (1988) 950±956. [7] R. DuÈmpelmann, Kinetische Untersuchungen des Methanolreforming und der Wassergaskonvertierungsreaktion in einem konsentrationgeregelten Kreislaufreaktor, Ph.D. Dissertation, EidgenoÈssischen Technischen Hochschule, ZuÈrich, [8] C.J. Jiang, D.L. Trimm, M.S. Wainwright, N.W. Cant, Kinetic study of steam reforming of methanol of copper-based catalysts, Appl. Catal. A 93 (1993) 245±255. [9] J.C. Amphlett, R.F. Mann, B.A. Peppley, The steamreforming of methanol: mechanism and kinetics compared to the methanol synthesis process, in: H.E. Curry-Hyde, R.F. Howe (Eds.), Studies in Surface Science and Catalysis, vol. 81, Elsevier, Amsterdam, 1994, pp. 409±412, ISBN [10] N.E. Vanderborgh, B.E. Goodby, T.E. Springer, Oxygen exchange reactions during methanol steam reforming, in: Proceedings of the 32nd International Power Sources Symposium, 1986, pp. 623±628. [11] K.C. Waugh, Methanol Synthesis, Catal. Today 15 (1992) 51±75. [12] G.E. Parris, K. Klier, The specific copper surface areas in Cu/ ZnO methanol synthesis catalysts by oxygen and carbon dioxide chemisorption: evidence for irreversible CO chemisorption induced by the interaction of the catalyst components, J. Catal. 97 (1986) 374±384. [13] D.E. Mears, Diagnostic criteria for heat transport limitations in fixed bed reactors, J. Catal. 20 (1971) 127±131. [14] H.S. Fogler, Elements of chemical reaction engineering, in: N.R. Amundson (Ed.), Prentice-Hall International Series in the Physical and Chemical Engineering Sciences, Prentice- Hall, Englewood Cliffs, NJ, 1992, p [15] R.O. Idem, N.N. Bakhshi, Kinetic modeling of the production of hydrogen from the methanol±steam reforming process over Mn-promoted coprecipitated Cu±Al catalyst, Chem. Eng. Sci. 51(14) (1996) 3697±3708. [16] B.A. Peppley, Ph.D. Dissertation, Royal Military College of Canada, May [17] G.H. Graaf, E.J. Stamhuis, A.A.C.M. Beenackers, Kinetics of low-pressure methanol synthesis, Chem. Eng. Sci. 43(12) (1988) 3185±3195. [18] B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Methanol±steam reforming on Cu/ZnO/Al 2 O 3 catalyst. Part 2: A comprehensive kinetic model, Appl. Catal. A 179 (1999) 31±49.

10 本文献由 学霸图书馆 - 文献云下载 收集自网络, 仅供学习交流使用 学霸图书馆 ( 是一个 整合众多图书馆数据库资源, 提供一站式文献检索和下载服务 的 24 小时在线不限 IP 图书馆 图书馆致力于便利 促进学习与科研, 提供最强文献下载服务 图书馆导航 : 图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具