英文文献3.doc

Characterization of HTS and LTS Reactions in MicrostructuredReactorsMarta Pawlak,* Markus Thaler, and Viktor HackerInstitute for Chemistry and Technology of Inorganic Materials, CD-Laboratory for Fuel Cell Systems,Graz UniVersity of Technology, Steyrergasse 21, 8010 Graz, AustriaReceiVed October 26, 2006. ReVised Manuscript ReceiVed May 20, 2007The production of hydrogen from synthesis gas with a high CO content is investigated with microstructuredreactors. The main advantages of using microstructured reactors are the excellent heat and mass transfercharacteristics inside of the reactor and the possibility to include heat exchangers to reduce the system sizeand heat losses. High- and low-temperature water gas shift reactions are examined for the ability to decreasethe CO fraction of the synthesis gas down to a level tolerable for proton exchange membrane fuel cells. Thesegas purification methods have the advantage of a higher hydrogen yield being achievable by “converting” COinto H2, thus leading to a higher overall efficiency of the fuel cell system. A next step in further reducing theCO content could besif necessaryspreferential oxidation. The high-temperature shift reaction takes place ata temperature of approximately 350 C in the presence of an iron- or chromium-based catalyst. The lowtemperatureshift reaction is operated at approximately 250 C with copper catalysts. The performance ofmicrostructured reactors fitted with different catalysts is discussed within this paper as a function of thetemperature, gas flow rate, and gas composition.IntroductionFuel cell technology has reached a stage of developmentwhere it has become increasingly interesting for commercialapplications. Currently, most of the hydrogen is produced fromhydrocarbons by steam reforming or autothermal reformingcombined with complex gas-cleaning procedures.A typical gas-cleaning procedure includes the following twosteps: The reformed fuels are first passed through the hightemperaturewater gas shift (WGS) stage (HTS) and are thenfed into the low-temperature water gas shift stage (LTS) toreduce the CO concentration. The outlet gas still contains smallamounts of carbon monoxide, which can poison the platinumcatalyst used in proton exchange membrane fuel cells. Therefore,preferential oxidation can be utilized as an additional finepurification step.The HTS and LTS reactions can be carried out in a microfuel processor. The advantages of these systems include highconversion efficiencies, fast dynamic responses, low heat losses,and efficient usage of the anode off-gases by catalytic combustors1.Catalyzed microstructured plates create a compact system.2,3 The stainless steel foils show strong adhesive propertiesfor catalyst layers. The catalysts are deposited using the washcoating method.1Precious metal catalysts such as Pt/CeO2, Au/CeO2, Au/Fe2O3,4 Ru/CeO2, Ni/CeO2, Pd/CeO2,5 Ru/ZrO2, Ru/SiO2, andRu/Al2O36 as well as catalysts with several layers such as Pt/Rh, Pt, and Pt/Pd on CeO2/Al2O3 and Pt/Rh/Al2O37 are describedin the literature for CO removal. Pt/CeO2 catalysts are selectedfor the experimental examinations of the WGS reaction becauseof their bifunctional mechanism (eq 1)8 and their high COconversionrates under high-temperature WGS reaction conditions.9Ceria is reduced from Ce4+ to Ce3+ and donates oxygen ionsfor the oxidation of carbon monoxide to carbon dioxide.Platinum, on the other hand, catalyzes the reduction of waterinto hydrogen, thus liberating an oxygen ion that reoxidizes theceria Ce3+ to Ce4+.10Wash coatings were prepared from commercially availabledry-milled powder catalysts. The commercial catalysts (PtX,PtY, RuZ, and CuXY) were prepared using a milling process.Adhesion characteristics were investigated as a function of theparticle sizes of the powder catalysts. Influences of preparationmethods, reduction treatments, and concentrations of the gasesfor HTS and LTS reaction conditions were studied.* Corresponding author. E-mail: marta.pawlaktugraz.at.(1) Zapf, R.; Becker-Willinger, C.; Berresheim, K.; Holy, H.; Gnaser,H.; Hessel, V.; Kolb, G.; Loeb, P.; Pannwitt, A.-K.; Ziogas, A. Chem. Eng.Res. Des. 2003, 81, 721/729.(2) Bae, J.-M.; Ahmed, S.; Kumar, R.; Doss, E. J. Power Sources 2005,139, 91-95.(3) Kiwi-Minsker, L.; Renken, A. Catal. Today 2005, 110, 2-14.(4) Luengnaruemitchai, A.; Osuwan, S.; Gulari, E. Catal. Commun. 2003,4, 215-221.(5) Wheler, C.; Jhalani, A.; Klein, E. J; Tummala, S.; Schmidt, L. D. J.Catal. 2004, 223, 191-199.(6) Goerke, O.; Pfeifer, P.; Schubert, K. Appl. Catal. 2004, 263, 11-18.(7) Kolb, G.; Penneman, H.; Zapf, R. Catal. Today 2005, 110, 121-131.(8) Hilaire, S.; Wang, X.; Luo, T.; Gorte, R. J.; Wagner, J. Appl. Catal.2001, 215, 271.(9) Pfeifer, P. Appl. Catal., A 2005, 286, 175-185.(10) Jacobs, G.; Ricote, S.; Graham, U. M.; Petterson, P. M.; Davis, B.H. Catal. Today 2005, 106, 259-264.Pt* + CO f Pt-COPt-CO + 2CeO2f Ce2O3+ CO2+ Pt*Ce2O3 + H2O f 2CeO2 + H2 (1)Energy & Fuels 2007, 21, 2299-2305 229910.1021/ef060538b CCC: $37.00 2007 American Chemical SocietyPublished on Web 06/30/2007Experimental SectionCatalyst Preparation. The four commercial powder catalystswere first dry-milled for 10 h by using a planetary mill (Pulverizette4, Fritsch) by the rotation of a supported disc (rotation velocity of229 min-1) and grinding bowls (238 min-1).This preparation method was based on a concept proposed byMen et al.12 The coating slurries were prepared from the untreated(I-PtX-1, II-PtY-1, RuZ-1, and CuXY-1) and the milled (I-PtX-2,II-PtY-2, RuZ-2, and CuXY-2) powder catalysts. The slurriesconsisted of a binder polyvinylalcohol (PVA), acetic acid with aweigh ratio of 5:1, commercial powder catalyst, and deionized water(1:3.75). PVA was improved by adhesion between its layers andthe metal (stainless steel) substrate during drying and calcinationsand a reduction in crack formation in the layers. A thermogravimetricanalysis was conducted to confirm the reducing effect ofthe evaporation rate by PVA addition in each slurry.The PVA binder was dissolved in water, and the solution wasstirred at 60 C for 2 h and then left overnight in a vessel for therelease of air bubbles. The powder catalyst and acetic acid wereadded next and stirred overnight. The acidic components were usedto peptize the slurry. In this way, the pH was controlled, whichwas adjusted at various times for the viscosity of the catalystslurries. All catalyst samples were rested for 2 weeks to release airbubbles.In the case of the Pt/CeO2 catalyst CeO2 nanopowder (Aldrich),acetic acid and 1 wt % H2PtCl6H2O were added to a preparedslurry consisting of PVA and water and then were stirred togetherovernight.Coating of the Plates (Wash Coating). The microstructuredsubstrates were first cleaned in an ultrasonic bath with 2-propanol;then, they were dried and heated in the air at 900 C for 3 h. Afterpositioning and masking the substrates, the prepared wash coatslurry was deposited onto the microstructured channels using asyringe. The microchannels were filled completely, and surplusslurry was removed with a stainless-steel knife with a flat edge.After coating, the plates were dried overnight at room temperature.13Then, the wash coats were calcined in hot air at temperatures andtimes suitable for different types of catalysts. The amount of catalystapplied with each plate was approximately 0.1 g. This wasdetermined by the amount of catalyst powder in the aqueoussuspension.Characterization of the Catalyst. The crystal structure of thepowder catalysts was investigated by powder X-ray diffraction. Theresults were obtained on a Bruker AXS (Siemens) D5005 -diffractometer using filtered Cu KR radiation and a graphitesecondary-beam monochromator. Diffraction intensities were measuredby scanning with steps of 0.05 (2) and 5-20 s/step.Scanning electron microscopy (SEM) analyses were performed ona JEOL JWS-7515 device. Brunauer-Emmett-Teller (BET)surface areas were determined using an Autosorb-1 device (Quantachrome). The powder catalyst samples were first dried at elevatedtemperatures. Then, the surface area values were gathered at aconstant temperature of 77 K (liquid nitrogen).All powder particles were analyzed by a laser particle sizeanalyzer CILAS 1180 in the range of 0.04-2.500 m in the wetdispersion mode before and after the millings.Experimental Setup. The catalyst coating was applied to fivemicrostructured foils (50 mm _ 50 mm _ 1 mm), containing 49channels (400 m deep and 600 m wide). The coated foils werestacked with filler plates inside of a stainless steel housing. Themicroreactor and foils were made of stainless steel (DIN 1.4571).Four 250 W heating cartridges were installed into the microreactorhousing, two of them in the upper and two in the lower part of thereactor. The reactor temperature was measured with two thermocoupleslocated in the housing. The gas flow was controlled bymass flow controllers. Water was vaporized and subsequently mixedwith the gaseous reactants.The temperature of the reactor was raised to 350 C at a rate of4 C/min and then kept constant for 1 h (300 C for 40 min forRuZ catalyst) in a gas flow of 10% H2 in N2 with all of themeasurements. After this procedure, the reactor temperature wasset to the first reaction temperature and the reactor was purged withpure nitrogen. Then, the experiments started with the correspondingfeed gas compositions.The gas mixture was applied in a microchannel volume VR mm3with the total inlet mass flow V mL/min. The variation of residencetime can be calculated by the following equation:The water gas shift reaction is an exothermic reaction proceedingaccording to the equationThe different water contents in the feed gas had a strong influenceon the catalysts CO-conversion rate, following the WGS reaction,eq 3. These variations in the input gas water concentration werecompensated with nitrogen as a balance factor.The CO conversion was compared between unmilled and milledcatalysts. The measured inlet concentrations COin and outletconcentrations COout were used to calculate the conversion X(CO),which is can be defined asFor the CO-conversion rate in the water gas shift reaction, eachcatalyst was compared with the equilibrium curve using the equationThe undesirable methanation occurs in parallel to the water gasshift reaction and is defined as follows:An important sequence in which the metals promote methanationwas reported as Ru Rh Ni Pt Pd. Methane formation onPt and Pd was about 1% and approximately 10% on Ru. It wasalso found that the addition of ceria had a small effect on CH4formation.5From the literature, another activity ranking of Ru Rh . Pt/CeO2 . Pd/CeO2 was reported for the methanation reaction, whichstarts at temperatures higher than 300 C for Ru and Rh and reachesits maximum at 500 C, whereas it initiates at 400 C for Pt and500 C for Pd.14Results and DiscussionGeometric Structure of the Catalyst Layers. Many factorssuch as the time, the temperature, and particle sizes stronglyinfluence the changes in the catalytic crystalline structure andits activity during the powder milling and catalyst layercalcinations process. Pfeifer et al.9 investigated the wash coatproperties of mixed systems of CuO milled powders. Theyproved that these layers were just a little more active in theinitial state. However, the addition of TiO2 to the mixture ofCuO, ZnO, and Al2O3 decreased the initial activity but elevatedthe stability of the catalyst.(11) Zhao, I. Surf. Coat. Technol. 2003, 168, 179-185.(12) Men, Y.; Gnaser, H.; Zapf, R.; Hessel, V.; Ziegler, C.; Kolb, G.Appl. Catal., A 2004, 277, 84.(13) Hessel, V.; Loewe, H.; Mueller, A.; Kolb, G. Chemical MicroProcess Enginieering-Processing and Plants; Wiley-VCH: Weinheim,Germany, 2005; pp 290-393. (14) Wang, X.; Gorte, R. J.; Wagner, J. P. J. Catal. 2002, 212-225. )VR mm3V L/min (2)H2O + CO ) H2 + CO2, H(298)-41.3 kJ/mol (3)X(CO) )COin - COoutCOin_ 100% (4)Keq )CO2H2COH2O (5)CO + 3H2f CH4+ H2O, H298)-283, 6 kJ/mol (6)2300 Energy & Fuels, Vol. 21, No. 4, 2007 Pawlak et al.Zhao et al.11 studied the coating properties of a milled powderof Al2O3 strengthened by NiCr after the spraying process. Thecoatings were harder and more resistant than the homogeneousmicrostructures of the milled powders. In addition, thesehomogeneous coatings had a lower hardness than their powders.In this work, the uniformity of the catalyst layer wasconsiderably affected by the catalyst particle size reduction.Additionally, to better understand the impact on the coatingeffect, Figures 1 and 2 present the particle size distribution forcommercial catalysts before and after the milling process. Twoclearly visible peaks are situated in the highest (10-170 m)and lowest (0.04-1.3 m) particle size areas for each untreatedpowder (Figure 1). Moreover, one additional peak is located inbetween for the PtY-1, RuZ-1, PtX-2, and PtY-2 catalysts.Milling reduced the catalyst particle size to 40-120 m (Figure2) as well as the BET surface areas given in Table 1 for allcommercial powders (catalyst 1, before milling; catalyst 2, aftermilling).Although the PtX and the PtY catalysts in Figures 3-6 havevery good adhesion for both particle size forms, their milledFigure 1. Particle size of catalyst powder before milling.Figure 2. Particle size of catalyst powder after milling.Table 1. Specific Mass, BET Surface Area, and CalcinationParameters for the Catalyst Samples Applied onto theMicrostructured Foilcatalysttypecatalyst massfor five plateletsmgBET SAm2/gcalcinationtemperatureCcalcinationtimeminPtX-1 513 86.6 400 60PtX-2 471 82.7 400 60PtY-1 455 102.3 400 60PtY-2 443 92.8 400 60Ru-Z-1 453 381.1 540 100Ru-Z-2 484 255.3 540 100CuXY-1 421 not measured 480 80CuXY-2 646 not measured 480 80CeO2a 80-100Pt/CeO2 513 400 60a Particle diameter for CeO2: d ) 10-20 nm.Figure 3. SEM of the PtX-1 catalyst on microstructured stainless steel.Figure 4. SEM of the PtX-2 catalyst on microstructured stainless steelfoil.Figure 5. SEM of the PtY-1 catalyst on microstructured stainless steelfoil.Figure 6. SEM of the PtY-2 catalyst on microstructured stainless steelfoil.HTS and LTS Reactions in Microstructured Reactors Energy & Fuels, Vol. 21, No. 4, 2007 2301particles were more homogenously distributed in the microchannelsafter coating. These untreated catalyst layers showsmall local holes with planar surfaces, which are caused by thediffusion at high temperatures during calcinations. However,the largest particle sizes in the region of 75-170 m are notedfor the PtX-1 catalyst in Figure 1; only particle sizes of 10-30m were easily visible on the surface layer after calcinations inFigure 3. The PtY-1 catalyst had a smaller amount of biggerregional particles at a diameter of 30-50 m in comparison tothe PtX-1.Figure 7 shows perfect adhesion properties for the RuZ-1catalyst compared to all presented catalyst coats. It subsequentlyreached the two highest volume dispersions for grain sizes of57.5 and 5 m. In contrast to the morphology of RuZ-1, thecatalyst layer of RuZ-2, with one of the highest 20-m particlediameter concentrations, has only a few cracks on the catalystsurface, as shown in Figure 8. For this catalyst, it was observedthat the choice of favorable calcination parameters like temperatureand time was dependent on the catalyst particle sizes.The preceding factors were the ones most important in theoptimization of the stable catalyst layers.The CuXY-1 and -2 catalysts, as Figure 9 shows, exhibit thesame large cracks on the layer surfaces at particle sizes of 75and 35 m for before and after milling, respectively.Cracks in the two last presented catalyst coats were shownafter solvent diffusion during calcinations, wherein atoms weremoved from their original sites after the bonds were broken.Atoms move more easily over the surface of the catalyst layeror along grain boundaries than through its bulk, as observedfor the RuZ-2 catalyst. Less densely packed coats promotediffusion over more densely packed planes, which was notedfor the CuXY-1 and -2, RuZ-2, and Pt/CeO2 catalysts withparticle size ranges becoming more narrow with higher porosity.Density is influenced by structure, phase, and porosity. Duringcalcinations, the reduction in porosity was preceded with growthof the grains and a significant decrease in strength, which ledto catalyst layer breakage.The commercial catalysts were deposited onto microchannelsusing aqueous slurries (PtX, PtY, and RuZ) and one sol-gelbasedslurry (CuXY). As Table 1 shows, the CuXY layer isthicker than that for the aqueous slurries, which is improvedby a larger catalyst mass on the microstructures after depositionand calcination (Table 1). At the moment of drying of the CuXYcoat, the constituent particles were subjected to a large capillarytension which decreased as the diameters of the pores increasedand which led to cracks. This capillary force was bringing theparticles closer together as the liquid (solvent) was removed.During calcination, a macroporous texture with cracks wascreated through decomposition and volatilization of the substancespreviously added to the solid. The texture was partiallymodified through sintering, wherein small particles were turnedinto bigger ones.Wear in this coat was undesirable in the sense of losingmate