Catalytic steam reforming of biomass-derived oxygenates: acetic acid and hydroxyacetaldehyde

~ ELSEVIER APPLIED CATALYSIS A:GENERAL Applied Catalysis A: General 143 (1996) 245-270 Catalytic steam reforming of biomass-derived oxygenates: acetic acid and hydroxyacetaldehyde D. Wang *, D. Montan6 i, E. Chornet 2 National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401, USA Received 6 October 1995; revised 12 February 1996; accepted 13 February 1996 Abstract Biomass can be pyrolytically converted in high yields ( ~ 70 wt.%) into vapors (or oils when condensed) composed mainly of oxygenated organic compounds. Using a fixed-bed microreactor interfaced with a molecular beam mass spectrometer (MBMS), we have been studying the catalytic steam reforming of model oxygen-containing compounds present in biomass pyrolysis vapors. This MBMS sampling system is unique in its rapid, real-time, and universal detection of gaseous and condensible products. In this paper, we present results for steam reforming of acetic acid (HAc) and hydroxyacetaldehyde (HAA), two major products derived from the pyrolysis of carbohydrates in biomass. We propose mechanisms to couple the thermal decomposition and steam reforming reactions of these compounds. Both HAc and HAA undergo rapid thermal decomposition; complete steam reforming of these two model compounds can be achieved with commercial Ni-based catalysts. HAc forms coke on the catalyst surface, which is subsequently gasified by steam. The proposed mechanism for this coke formation involves an adsorbed acetate species that decarboxylates to form the coke precursor, (CHl_3)ab s, and also ketene, a dehydration product of HAc, that decomposes to form (CHl,z)ab s. The reforming of HAA by steam does not involve any detectable intermediate and proceeds smoothly to a complete breakdown to CO and H 2 on the catalyst surface. Keywords." Acetic acid; Hydroxyacetaldehyde; Hydrogen; Steam reforming; Thermal decomposition; Oxygencontaining compounds; Biomass-derived oxygenates; Mechanisms; Molecular beam mass spectrometry * Corresponding author. Tel.: (+ 1-303) 3846127; fax: (+ 1-303) 3846103; e-mail wangd@tcplink.nrel.gov i On leave from Universitat Rovira i Virgili, Departament d'Enginyeria Qufmica, Carretera de Salou, s/n, 43006 Tarragona, Spain. 2 Also affiliated with Universit6 de Sherbrooke, Sherbrooke, Quebec, Canada, J1K-2RI. 0926-860X/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. Pll S 0 9 2 6 - 8 6 0 X ( 9 6 ) 0 0 0 9 3 - 2 246 D. Wang et al. /Applied Catalysis A: General 245-270 1. Introduction Hydrogen is an important industrial commodity as well as a clean transportation fuel. Among the well-established industrial processes for hydrogen production are (1) the catalyzed steam reforming of hydrocarbons such as methane and naphtha, (2) the gasification of carbonaceous solids (i.e., coals) followed by shift conversion, and (3) the catalytic steam reforming of alcohols such as methanol and ethanol [1-3]. In recent years, the use of lignocellulosic biomass as a feedstock for hydrogen production via gasification followed by shift conversion has received considerable attention [4-7]. An alternate route of hydrogen production is to couple fast pyrolysis of lignocellulosics with catalyzed steam reforming of the pyrolytic oils [8]. New developments in fast pyrolysis technologies make it possible to produce pyrolysis oils (i.e., biocrude) at yields around 75% by weight of the anhydrous biomass [9-13]. Results from fluid bed fast pyrolysis of poplar have shown a 76 wt.% yield of biocrude [9,10]. The organic fraction ( ~ 85 wt.% of the biocrude) has a typical elemental composition of CHL3300~3. Although the molecular composition of pyrolysis oil varies significantly with the type of biomass and the pyrolysis conditions (pyrolysis severity and media), its major components are oxygenates belonging to the following groups: acids, aldehydes and alcohols, anhydrosugars and substituted furans derived from cellulose and hemicellulose, and phenolics and cyclic oxygenates derived from lignin [14,15]. The hydrogen potential of the biocrude, if an adequate reforming technology is developed, is 12.6 wt.% of the initial biomass (the substrate for this determination is poplar, CH1.4700.67) [16]. A successful catalytic steam reforming process for the biocrude would benefit from a thorough understanding of the chemistry involved in the reforming reactions of the variety of oxygen-containing species found in the biocrude. Steam reforming reactions of C1-C 5 hydrocarbons, naphtha, gas oils, and simple aromatics have been well studied, and a great deal of knowledge on the kinetics and mechanisms of these reactions has been accumulated [17-21]. The stoichiometry of hydrocarbon reforming for maximum hydrogen production is described by Eq. (1). This reaction is the sum of two reactions (Eqs. (2) and (3)) that take place simultaneously in any reformer. The initial steps in steam reforming are dissociative adsorption of the hydrocarbons on the metal sites and ( m) (m) C . H m + 2 n H 2 0 ~ nCO 2+ 2 n + 2 H2 (1) nH20 ~ nCO + n + -~- (2) CnH m + ()<n)CO+H20 ~- C O 2 + H 2 H 2 (3) D. Wang et al. /Applied Catalysis A: General 245-270 247 reaction of the adsorbed CxH v species with adsorbed H20-derived species to produce CO and H 2 (Eq. (2))~ With an active catalyst at temperatures below 600°C, reforming of naphtha (Eq. (2)) is irreversible, with no intermediates, and the only byproduct is carbon that forms on the catalyst [20]. The reversible reactions are the water-gas shift reaction (Eq. (3)), the methanation reaction (Eq. (4)), and the disproportionation of CO (Eq. (5)), which are equilibriumlimited. Only at higher temperatures does thermal cracking of the hydrocarbon itself compete with the catalyzed reactions. Because natural gas and naphthas have been the main feedstocks in commercial steam reforming processes CO + 3H 2 ~ CH 4 + H20 (4) 2CO ~ C O 2 -~- C (5) that produce of hydrogen and synthesis gas (a mixture of hydrogen and carbon monoxide), the literature contains few report of studies on steam reforming of oxygenates, limited to simple alcohols (methanol [22-31] and ethanol [3,32,33]) and oxygenated aromatic compounds (cresols [34-36]). Methanol reforming for the production of hydrogen [22-31] has been extensively studied, because of its fuel cell applications [34]. Methane reforming requires relatively high temperatures (reformer inlet at 500-550°C and exit at 750-1000°C) and involves dissociative adsorption on the metal sites [17], whereas methanol can be reformed at temperatures of 300°C or less, using a variety of catalysts [22]. The product slate is markedly affected by the type of metal and support material in the catalyst preparation as well as the experimental conditions. One mechanism proposed for methanol reforming with typical steam reforming catalysts, either Ni-based or precious metals, is the total dehydrogenation of the adsorbed methanol to CO and H2, followed by the shift reaction (Eq. (3)), similar to that of hydrocarbons. Another mechanism, suggested to operate with other catalysts such as C u O / Z n O [29] and P d / Z n O [22], involves the formation of two key intermediates: methyl formate (HCOOCH 3) and formic acid (HCOOH). The latter, formed by hydrolysis of the former, decomposes directly to CO 2 and H 2 without CO formation. Ethanol reforming also has been studied in some detail. Iwasa and Takezawa [3] found that, with Cu-based catalysts, acetaldehyde is a key intermediate in the formation of either ethyl acetate or acetic acid [3]. The selectivity depends on the type of support used. Acetaldehyde was also found to be a product in ethanol steam reforming using N i / M g O and P t / M g O catalysts [32]. A recent thermodynamic study by Garcia and Laborda [33] concluded that hydrogen production from ethanol and steam requires higher temperatures than those needed for reforming methanol and steam-to-ethanol ratios higher than 10:1 are necessary to prevent carbon formation. Steam reforming of oxygenated aromatics has been studied in relation to the dealkylation of cresol [35-37]. The predominant reaction is complete steam 248 D. Wang et al. /Applied Catalysis A: General 245 270 reforming to H 2 and CO [35,36]. It has been proposed that the OH grroup in cresols participates in the overall reaction process in the same manner as the OH species derived from the dissociation of water [37]. Because oxygenated aromatics undergo cracking at a significant rate to produce aromatic species on the catalyst surface, the reforming of aromatic compounds is also an important reaction. Duprez [36] reviewed selective steam reforming of aromatic compounds on metal catalysts. At temperatures of < 500°C, steam reforming of aromatics to CO and H 2 takes place in competition with ring opening and dealkylation in almost all cases. More recently, the reforming of tars (mostly polynuclear aromatic compounds) has been studied for the purpose of a catalytic conditioning of synthesis gas (a mixture containing mainly H 2 and CO) produced by biomass gasification [38]. Results of these studies suggest that biocrude, even though it is a complex mixture of organic oxygenates, can be steam reformed using established catalyst preparations. This has been confirmed in our preliminary screening experiments [16], where the catalytic steam reforming of various model oxygenated compounds and lignocellulosic materials using a number of commercial and research catalysts. Little is known, however, about the experimental conditions that will lead to long-term catalyst activity. To address issues of activity, selectivity, and lifetime, a better understanding of the mechanisms of both the thermally induced cracking decomposition and the catalytic steam reforming reactions of oxygenates is needed. Few of the primary products of biomass pyrolysis are thermally stable at the typical temperatures of a hydrocarbon reformer. Thus there is significant competition between catalytic reforming reactions (Eq. (6)) and thermal decomposition (Eq. (7)) for most oxygenates. • Complete steam reforming: CnHmO +(n- )H20- riCO+ n + 2 H2 (6) • Partial thermal decomposition: (or cracking) C,,UmO k --~ CxH,O.. " + gas(H2,CO,CO2,CU 4 . . . . ) + coke (7) Furimsky [39] presented discussion on the stability of oxygen-containing compounds in a review of the chemistry of catalytic hydrodeoxygenation. The nature of vapor-phase cracking products from biomass has been described in previous work from this laboratory [40]. It was found that at low cracking severity (a combination of residence time and temperature), the primary pyrolysis products are converted to CO, CO 2, light hydrocarbons, furans, and phenolics, while stable aromatic compounds, such as benzene, naphthalene and polynuclear aromatics, dominate the spectrum of products formed at high cracking severity. It has also been suggested that the polymerization of lowmolecular-weight hydrocarbons leads to the formation of condensed aromatics. However, a complete understanding of the complex chemistry involved in the D. Wang et al. / Applied Catalysis A: General 245-270 249 thermal cracking of biomass pyrolysis products is difficult due to the large number of primary and secondary pyrolysis products generated through the many different pathways available to the highly functionalized molecules. One such example is coniferyl alcohol (a lignin-derived molecule), which decomposes to form more than a dozen major products [41]. Rajadurai [42] recently reviewed thermal decomposition of carboxylic acids on transition metal oxides. He concluded that the main mechanism responsible for ketonization of acetic acid (Eq. (8)) is the bimolecular interaction between either two adsorbed acetate ions or one adsorbed acetate ion and an adsorbed acyl carbonium ion. 2CHBCO2H ~ (CH3)2CO q- H20 q- CO 2 (8) As the first part of our systematic studies on this biomass-to-hydrogen process, we have carried out thermodynamic calculations, catalytic screening studies, environmental life-cycle assessment, as well as techno-economic analysis [8,16]. This paper describes our experimental screening studies and proposes mechanisms for steam reforming of two model compounds, acetic acid (HAc) and hydroxyacetaldehyde (HAA), by using a molecular beam mass spectrometer (MBMS) as the detector for product analysis. Both HAc and HAA are major components in pyrolysis oil (HAc at 3-7 wt.% and HAA at 4-10 wt.% of dry biomass) [14,15]. 2. Experimental 2.1. Apparatus Most of the experimental work was carried out in a vertical, dual bed quartz reactor, interfaced with an MBMS, shown in Fig. 1, to analyze the products. A tubular furnace with four independently controlled temperature zones enclosed the reactor to allow heating to the desired temperatures. The reactor had two inner tubes surrounded by a larger outer housing. The inner tubes were used as fixed-bed reactors for catalyst testing. The outer flow in Fig. 1 was used for calibration purposes as well as for the dilution of samples with helium gas to obtain an adequate molecular beam and high signal-to-noise ratios. Steam was generated in the reactor by vaporizing water injected with syringe pumps (SAGE model 355) either through the bottom inlet or a special injection inlet in the side arm. Acetic acid and hydroxyacetaldehyde solutions were similarly introduced into the reactor. The vapors were swept by a flow of preheated helium. The MBMS instrument has been described in detail previously [40], and so only a brief description is given here. The hot gases exiting from the reactor were at ambient pressure and were sampled through an orifice into a low-pressure region of the MBMS to form a molecular beam. This beam was then collimated through a skimmer before entering into the high-vacuum chamber, D. Wang et al./Applied Catalysis A: General 245-270 250 Thermocouples Products and T.o.=;o.*~ Three Stage Free-Jet )cular Beam Source Q2 Triple Quadrupole Mass Analyzer " Q3 IIco,,sionsl=l ctor {D"} II rbo Argon cular Collision imp Gas Ren cat= Turbo Molecular Pump M B M S Sample feeder Sample feeder our-zone furnace Vlicroreactor Inner flows ow Fig. 1. A schematic diagram of the microreactor interfaced with a molecular beam mass spectrometer for real-time product analysis. and on into the ion source. Electron impact (EI) ionization of the neutral molecules at approximately 25 eV was used to produce ions. The mass spectrum of positive ions in the mass range of interest (typically m / z 1-350) was obtained by scanning the mass spectrometer (an Extrel triple quadrupole mass analyzer) every second. Data collected during steady state or pulse experiments were averaged and corrected for background. An IBM-compatible 486/33 computer operating with the O S / 2 system was used to control the instrument and acquire data with a Teknivent Vector Two data system. 2.2. Materials and methods Steam reforming reactions were carried out with a commercial, Ni-based catalyst, G-90C, from United Catalysts Inc. (UCI), Louisville, KY. The catalyst contained 15% Ni on a ceramic support (A1203/CaA1204). We also tested a low-temperature shift catalyst, C 18HC, also from UCI, that contained 42% CuO D. Wang et al. / Applied Catalysis A: General 245-270 251 and 47% ZnO with 11% A1203 as the support, on the hypothesis that it may reform low-molecular-weight alcohols and aldehydes. Catalysts were ground and sieved before testing; particles in the size range of + 6 0 / - 2 5 mesh (250-710 p~m) were used for testing in the microreactor. The nickel-based catalyst was reduced at 600°C and the C u / Z n shift catalyst at 350°C, both with 20% hydrogen in helium for an extended period (more than 24 h). If not used directly, the catalyst was then cooled down to ambient temperature in helium and passivated before removal after initial bulk reduction. We used 0.05-1.0 g of catalyst in the microreactor; the desired amount of the pre-reduced catalyst was weighed and either used directly or mixed with pure quartz chips of + 6 0 / - 45 mesh (250-355 p~m) if the amount of catalyst was lower than 0.25 g. This lower limit gave a catalyst bed height of about 5.0 mm in a tube with an inner diameter of 8.0 ram. The undiluted catalyst or the physical mixture of catalyst and quartz chips was then placed between two small beds of the quartz chips to obtain a total bed height of at least 25 mm, which corresponded to a range of 35-100 for the ratio of bed height to particle size. This was slightly lower than that required to ensure uniform flow and intimate contact between the reactant and product gases and the catalyst [43]. Quartz wool was placed at both ends of the bed for packing the catalyst in the tube. A K-type thermocouple was placed just inside the catalyst layer of the bed from above (Fig. 1) to monitor catalyst temperature. The reactor was then heated to about 20°C above the desired operating temperature, and the pre-reduced catalyst was reduced again with 50% hydrogen in helium for 2 h. The pressure drop through the catalyst bed was typically 3-21 kPa at the highest reactant flow rates and catalyst temperatures. Thermal cracking reactions were carried out in either a blank quartz tube or one packed with quartz chips. All gases were obtained from commercial sources, including helium (99.9995%), argon (99.9995%), and hydrogen (99.9995%), and used without further purification. The following condensed materials were used as supplied (from Aldrich, except where otherwise indicated): methanol (Baker, 99.9 + %); acetic acid (99.7 + %); acetone (99.9 + %); hydroxyacetaldehyde (from the crystalline dimer, 2,5-dihydroxy-l,4-dioxane); ethylene glycol (99.8%); and glycerol (99.5 + %). For studying thermal decomposition, small amounts ( ~ 10 rag) of solid or liquid samples were placed in quartz boats and fed into the reactor in pulses. Alternatively, pure liquid samples were injected at given flow rates using syringe pumps for steady state experiments with catalysts. 2.3. Quantitative and qualitative analysis with MBMS For the quantitative analysis of hydrogen, the signal at m / z 2 was used directly since there is no significant interference from other molecules present in the steam reforming reactions. Contribution from water to m / z 2 was relatively 252 D. Wang et al./Applied Catalysis A: General 245-270 insignificant (/,,,/~ 2 / l m / r 18 <0.1%), and only small changes in the steam concentration were observed during steam reforming reactions owing to the excess amount of steam present. In practice, calibration was performed at four to five flow levels that cover the predicted range of hydrogen production. Pure hydrogen gas was metered by a mass flow controller (Tylan, model FC280, 0 - 1 0 0 mlsTr, rain J) into the outer flow of the reactor under operating conditions. The signal intensity of the hydrogen molecular ion (m/z 2) was then normalized to the signal of argon ( m / z 40), which was also added to the outer flow at a constant flow rate of 10 mlsT P rain-1 as an internal standard. The linear regression line fit of intensity versus flow rate was used to predict the flow rate of hydrogen produced by steam reforming of the model compounds. The data were fitted with straight lines through the origin with typical correlation coefficients better than 0.9999 and 0.999, respectively for the low and high flow rates. The measured hydrogen yield was reported as the percentage of the theoretical maximum according to the stoichiometry of the reaction, and it was estimated to be accurate to within _+3%. The main source of data error is a drift in the MBMS instrument, which limits the reproducibility of data for quantitative analysis. Such drift can be caused by clogging of the orifice and skimmer pin holes owing to condensation and deposition of low volatility materials (e.g., large aromatic molecules) from pyrolysis or incomplete conversion. A clogged orifice can change the molecular beam properties, resulting in mass-dependent variations in the response factors of various molecules. These variations were only partially corrected by using a single internal standard such as argon. Therefore, when accurate yield measurements were required, the calibration for hydrogen was repeated as often as possible throughout the day. Yields of other major products of steam reforming reactions, C H 4 (m/z 15), CO (m/z 28), and CO 2 (m/z 44), were not measured. However, they were estimated by assuming a complete conversion of methanol to H 2 and CO: and a comparison on relative yields can still be made. It should be noted that CO +. (m/z 28) is also a fragment ion from CO 2, with an intensity of 6% (_+ 1%) relative to that at m/z 44, depending on the mass spectrometer tuning. Under the MBMS operating conditions used in this study, the relative conversion factors from ion signal (counts/s) to neutral species (moles) were approximately 6.4" 1.4:1 for C H 4 " C O : C O 2 when using m/z 15, 28 and 44 as surrogates. Product assignments were based on comparing the 25 eV EI product mass spectra with the 70 eV spectra found in the National Institute for Standards and Testing Standard Reference Database 1A ( N I S T / E P A / N I H Mass Spectral Library) or in a user library constructed for pure compounds used in separate experiments under similar conditions of 25 eV EI. They were further aided by subtracting the mass spectrum of the reactant (usually no pyrolysis under the low temperature and short residence time conditions) from that of pyrolysis or reforming reactions to obtain a product mass spectrum. The detection limit of D. Wang et al./Applied Catalysis A: General 245-270 253 the MBMS instrument corresponded to maximum measurable conversions of approximately 99.95% for most of the samples. 3. Results and discussion For an oxygenated compound with a chemical formula of CnHmO k, the stoichiometric steam reforming reaction shown below (Eq. (9), the sum of reactions in Eqs. (3) and (6)), C n H m O k Jr ( 2 r / m k)H20 k ~ ( mt r/CO 2 q- 2 n + -~- - k H 2 (9) would yield 2 + G ,, moles of hydrogen gas per tool of carbon input. In this regard, methane (not an oxygenated feed) has a yield of 4, which is the highest yield among all feedstocks; methanol's yield of 3 is the highest among oxygenated compounds. Table 1 lists the stoichiometric and equilibrium yields of hydrogen from steam reforming of some representative model compounds present in biomass pyrolysis oils, as well as several biomass and related materials. The table also shows percentage yields by weight (grams of hydrogen produced per gram of the compound being reformed). In general, compounds derived from the lignin portion of the biomass yield more hydrogen than those from the carbohydrates (cellulose and hemicellulose) in terms of both weight and mole(s). Oxygenated aromatics such as furans and phenolics produce higher yields of hydrogen than do anhydrosugars and other carbohydrate-derived products, such as acetic acid and hydroxyacetaldehyde (C2H402). The latter compounds, as well as all products with the formula CnH2kOk, are equivalent to carbon (coal), producing only 2 moles of hydrogen per tool of carbon in the feed. In reality, the yield of hydrogen is lower than the stoichiometric maximum because two undesirable products, CO and CH4, are also formed via the water-gas shift (WGS) reaction and the reverse reaction in Eq. (4). One mol of hydrogen is lost for every tool of CO formed; four moles of H 2 are lost for each tool of methane. High steam-to-carbon ratios (S/C) shift these two equilibrium reactions toward hydrogen production. Methane steam reforming is thermodynamically favored under high temperature conditions. However, higher temperature also increases the formation of CO, but this may be converted to CO 2 and H 2 in a separate downstream WGS reactor operating at low temperatures. The formation of carbonaceous deposits (coke) could account for other missing hydrogen as well. Coking (the formation and deposition of carbonaceous deposits) on Ni-based catalysts is well known; it may lead to catalyst deactivation and a decreased thermal efficiency of the reactor [19,21]. Some of the oxygenated products derived from the carbohydrate component of biomass, especially anhydrosugars, are known to dehydrate rapidly and completely to 254 D. Wang el al. /Applied Catalysis A: General 245-270 Table 1 Stoichiometric yields of hydrogen from complete steam reforming reactions Sample Formula methane CH 4 methanol CH40 ethanol C 2H 60 acetone C 3H 60 dimethylfuran C<,H 80 anisole, cresol C7H80 methylfuran C 5H 60 ADP e CIIHI40 ~ phenol C 6H 60 guaiacol C 7H 802 furan C 4 H 40 syringol C8H1003 lignin C7H903 furfuryl alcohol resorcinol C 5H602 methylfurfural C6H~O 2 poplar oil C6H7.9803.18 aspen C6H8.790362 poplar C6 H8.8204.02 C6H91404.53 oakwood oil furfural C 5H 402 5-HMF ~ C611603 xylan C5H804 cellulose (C6HioOs) n cellobiose CiiH22Oll glucose C6HI206 HAc, HAA c C2t-[402 formic acid CH202 MW 16.04 32.04 46.07 58.08 96.13 108.14 82.10 194.23 94.11 124.14 68.08 154.17 141.15 98.10 110.11 130.99 138.84 145.27 153.76 96.09 126. I 1 132.12 180.16 330.29 180.16 60.05 46.03 Stoichiometric H 2 yield Equilibrium H 2 yield moles ~ % (by' wt.) h % (of st.) ~ 50.3 18.9 26.3 27.8 31.5 31.7 29.5 27.0 30.0 26.0 23.5 23.5 22.1 22.6 23.8 19.7 18.5 17.2 15.8 21.0 19.2 15.3 14.9 13.4 13.4 13.4 4.4 85.6 87.1 85.9 85.2 84.5 84.3 84.5 84.7 84.3 84.6 84.5 84.8 84.9 84.9 84.6 85.2 85.4 85.5 85.6 84.6 84.9 85.8 85.8 86.3 86.3 86.3 87.7 4.00 3.00 3.00 2.67 2.50 2.43 2.40 2.36 2.33 2.29 2.25 2.25 2.21 2.20 2.17 2.14 2.13 2.07 2.01 2.00 2.00 2.00 2.00 2.00 2.00 2.00 1.00 % ( + W G S ) ,..a 95.0 96.8 96.2 96.2 96.2 96.2 96.3 96.4 96.3 96.5 96.5 96.6 96.7 96.7 96.7 96.9 97.0 97.1 97.3 96.9 97.0 97.3 97.3 97.5 97.5 97.5 98.9 ~ Moles of H 2 produced per tool of carbon in the reactant being reformed. b Amount of H 2 formed divided by the sample molecular weight. " The equilibrium moles of H 2 predicted at 750°C and S / C - 5 divided by the stoichiometric yield. With additional moles of CO present under 750°C and S / C = 5 to be shifted in a downstream water-gas shift (WGS) reactor. e ADP: 4-allyl-2,6-dimethoxyphenol; 5-HMF: 5-hydroxymethylfurfural; HAc: acetic acid; HAA: hydroxyacetaldehyde. form carbon [44]. These types of carbon deposition will result in lower yields of hydrogen; two moles of hydrogen are lost in each tool of carbon not reformed by steam. From thermodynamic calculations, the equilibrium yield of H 2 (as percentages of the stoichiometric yield) at 750°C and S / C = 5 ranges from a low of 84.3% (for phenol) to a high of 87.7% (for formic acid); poplar pyrolysis oil's yield is close to the average of 85.3% ( _+0.9%). The variation in hydrogen yield is actually a consequence of the different amounts of water present in the feedstock; the actual S / C for formic acid ( C H 2 0 2) is increased to 6 because it has an additional mol of water per tool of carbon in the feedstock, whereas D. Wang et al./ Applied Catalysis A: General 245-270 255 phenol has only one extra tool of water per six moles of carbon in the feedstock. If combined with further WGS to convert CO, the overall potential for H 2 may be as high as 96.8% (Table 1). The rest of the hydrogen potential is bidden in carbon (3.2%), since there is very little C H 4 formation ( ~ 0.03% of the carbon balance). If the S / C is increased to 10, the H 2 yield improves to 92.5%, with only 6.7% additional potential in CO and 0.8% in C. Therefore, if we can prevent coke from building up on the catalyst, the conversion of oxygenates to hydrogen through the two-step process of steam reforming and WGS would be quantitative (100%). To present the data, we will use gas hourly space velocities (GHSV) on a C 1 basis, that is Gc HSV, defined as the volume of Cl-equivalent species in the feed at standard temperature and pressure per unit volume of catalyst (including the void fraction) per hour. Such a definition permits us to compare the reforming activity of catalysts per unit carbon in the feed when using complex feedstocks such as biocrude. Alternatively, we can also use the molar hourly space velocity, Mc HSV, defined as moles of carbon in the feed per unit volume (ml) of catalyst per hour. As well, the conventional gas hourly space velocity (GHSV), defined as volume of feedstock per unit volume of catalyst per hour, can also be used. These three terms are related according to G c H S V = ~ × GHSV = 22400 × M c H S V , where ~ is the number of carbon atoms per molecule of the feedstock. Thus, for steam reforming of natural gas, G c H S V is essentially the same as GHSV, since h = 1 for C H 4. Typical G c H S V values for steam reforming of naphthas and natural gas, 2000 and 10000 h - ~, respectively, have equivalent M c H S V values of 0.09 and 0.45 motc, ml~-a~ h -1. Other space velocities commonly used in the literature are the weight hourly space velocity (WHSV) and liquid hourly space velocity (LHSV). If one uses 1.0 g ml 1 for the density of catalyst (0.98 g m l - 1 for UCI G-90C in the size range used here), a WHSV of 2.0 gfeea gca[ h-1 corresponds to Mc, values (depending on the molecular weight of the feed) ranging from 0.020 (MW = 100) to 0.13 (MW = 16.04 for methane) molc, mlc-~1 h-1. An additional liquid density is needed to convert LHSV to the other terms. The catalytic residence time was calculated from the voidage volume of the catalyst bed (with a measured voidage ratio of 0.38 _+ 0.02) divided by the total volumetric flow of input gases at reaction T (400-700°C) and ambient pressure (83 kPa). The pyrolysis residence time was either obtained similarly to the catalytic residence time using a bed of quartz chips, or calculated from the empty inner tube volume in the hot zone, which was determined from prior measurements of reactor temperature profiles. One would expect the actual catalytic and pyrolytic residence times to be lower than values calculated above, since the volumetric flow of the output gases from either steam reforming or thermal decomposition reaction of oxygenates would be greater than that of the input gases. However, this increase is small compared with the total flow of helium carrier gas and excess steam. 256 D. Wang et al. /Applied Catalysis A: General 245-270 lOO% .........,...,,,,,,-- --~ 80% O .~_ =:~ 60% "", , • ,9° 4O% 20% 730°(2: - 0,9 1 302~-C 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 K*I OO0/T Fig. 2. Yields of hydrogen for catalytic steam reforming of methanol ( [ ] / I I ) , acetic acid ( © / O ) , and hydroxyacetaldehyde ( z x / • ) at varying temperatures, with open symbols denoting conditions of Gc IHSV = 336 h - l , S / C = 10, and -r = 0.1 s and solid symbols for conditions of G c H S V = 1680 h i S / C = 4.5 and "r = 0.08 s. The thick solid and dashed lines represent equilibrium yields from thennodynamic calculations, while the thin solid and dashed lines are only visual aids. Using a reactor with two inner flows (Fig. 1), we can compare either the separate and combined effects of thermolysis and catalysis or the performance of two catalyst beds under the same operating conditions of temperature and residence time. In the former case, only one of the inner tubes is loaded with a catalyst while the other tube remains empty or contains an inert material. The MBMS sampling system has the advantage of simultaneously detecting, in real time, all components in either the reactant and product stream. By interfacing the microreactor with the MBMS, we can detect any changes that may occur in the feed composition due to thermal reactions prior to reaching the catalyst bed. 3.1. Effects of reaction conditions on hydrogen yield Among the various operating parameters, temperature has the most profound effect on steam reforming reactions. Methanol can be reformed at 300°C with the UCI G-90C catalyst, and HAc and HAA are reformed in high conversions ( > 99.95%) at temperatures above 350°C. The effect of temperature on the yields of hydrogen from steam reforming of methanol, HAc, and HAA is displayed in Fig. 2. Also shown are the equilibrium yields of hydrogen for the two C2H402 isomers, HAc and HAA, at the two different S / C values (10 and 257 D. Wang et al. /Applied Catalysis A: General 245-270 100% ~_, 80% in -o 0 0 O 6°% .m " 40% 2 "O 20% O% 0 0.05 0.1 0.15 0.2 Residence time (s) Fig. 3. Effect of residence time ('r) on hydrogen yields from catalytic steam reforming of methanol ([]), acetic acid (©), and hydroxyacetaldehyde (zx) at 600°C, G c HSV = 1680 h -I , and S / C = 6. Lines are only visual aids. 4.5) used in these experiments. Below 500°C, there was a deeper drop in hydrogen yield but it still remained above the equilibrium value; methanol and HAA gave hydrogen yields significantly higher than the equilibrium values at all temperatures. Yields decreased slightly when more demanding reforming conditions (higher space velocity and lower S / C ) were used. Residence time (-r) and S / C have very limited effects on steam reforming reactions of model oxygenated compounds in the ranges studied. Within experimental error limits, varying residence time from 0.04 s to 0.15 s (Fig. 3) and increasing S / C from 4.5 to 7.5 (Fig. 4) did not significantly affect the yield of 100% ~, 80% O O O •1= 60% .2 40% O "O >, 20% z 0% 4.5 v r 5 5.5 6 6.5 7 7.5 8 SIC Fig. 4. Effect of steam-to-carbon ratio on hydrogen yields from catalytic steam reforming of methanol ( n ) , acetic acid (©), and hydroxyacetaldehyde ( • ) at 600°C, Gc,HSV = 1680 h - i and "r = 0.08 s. Lines are only visual aids. 258 D. Wang et al. / Applied Catalysis A: General 245-270 hydrogen at 600°C and Gc, = 1680 h ~, conditions which resulted in almost complete conversions of the model compounds. This insensitivity of hydrogen yield to residence time suggests that the influence of interphase and intrareactor effects may be assumed to be negligible under our experimental conditions [45]. It is also not surprising to see little effect of S / C above 4.5 because the values tested were quite high; the effect is expected to be evident at lower S / C , such as near unity. Similar results were found in the steam reforming of benzene [46]. Since typical conditions used in commercial naphtha reformers require S / C above 5, we did not investigate lower S / C conditions. The formation of methane showed significant dependence on residence time and S / C . As the residence time increased from 0.04 to 0.15 s, less methane formed from steam reforming of HAc and HAA, but the opposite was observed for methanol reforming. As expected, less steam also favored the formation of methane. At 700°C, there was no significant change in the yield of hydrogen when S / C was increased from 4.5 to 7.5. Methanol and acetic acid showed slight improvement at higher S / C , but the differences were still close to our experimental error limit. 3.2. Methanol and other alcohols Steam reforming of methanol was used as the model reaction to check catalyst activity and hydrogen mass closures. The low-temperature shift catalyst (UCI C 18HC) worked very well for steam reforming of methanol. The measured conversion of methanol under conditions of 350°C, S / C -- 5, G c H S V -- 1120 h -~ (1.5 ml catalyst), and "r -- 0.06 s was 99.94%, and the hydrogen yield was 100 + 3% (from several replicate experiments) with no detectable CO and C U 4 formation, consistent with the mechanism reported in the literature [23,31]. When the reaction temperature was increased from 350°C to 400°C, we observed some increase in Im/z 28 and decrease in Ira~z 44, in about equal amount, while the Im/," 28/Im/~ 44 ratio increased from 7% to 10%. This seems to indicate that the reverse WGS reaction has started to become significant at 400°C. Using the UCI G-90C catalyst under conditions of 600°C, G c H S V = 336 h -1 (1.0 ml catalyst) and -r = 0.08 s, the conversion of methanol was about 99.8%. The hydrogen yields were 94%, 101%, and 99% of the stoichiometric values at S / C ratios of 7, 15, and 22, respectively. These results confirm that we can accurately measure hydrogen with an estimated error limit of +_3%. We also observed the WGS reaction taking place as evidenced by the decreasing Im/z 28/Im/z 44 ratio as the S / C ratio was increased. The UCI G-90C is a highly active catalyst, and complete conversions of a variety of oxygenates have been observed [8]. The reforming of ethanol and other alcohols with the UCI G-90C catalyst was only briefly examined. Even at 400°C, we observed almost complete conver- D. Wang et al./Applied Catalysis A: General 245-270 259 sions of ethanol, ethylene glycol, glycerol, and acetol at G c H S V = 2240 h -~ (0.25 ml catalyst), S / C = 5, and a residence time of 0.02 s. However, the catalyst showed serious deactivation after a short period of high activity at this temperature. 3.3. Acetic acid and acetone The kinetics and mechanisms of the homogeneous thermal decomposition of acetic acid (HAc) have been well studied in the literature [47]. Two competing reactions take place to form methane and carbon dioxide (Eq. (10)) on the one hand, and ketene and water on the other (Eq. (11)). A change in kinetics was shown to take place for the dehydration reaction of HAc, from second order below 600°C in a silica reaction vessel to first order above 700°C [47,48], while the order for decarboxylation was one throughout the temperature ranges of 530 to 762°C [48] and 1027 to 1677°C [47]. The thermal decomposition of ketene has also been the subject of a number of studies; major products at high temperatures [47] were CO, ethylene (Eq. (12), involving methylene radicals), and CH 4, while at lower temperatures CO 2 and allene (Eq. (13)) were dominant [49]. CH3CO2 H ----) C H 4 q- CO 2 (10) CH3CO2H ~ (11) CH2CO + H20 2CH2CO ~ C 2 H 4 + 2CO (12) 2CH2CO ~ C3H 4 + CO 2 (13) In addition to observing the decarboxylation and dehydration reactions (Eqs. (10) and (11)), we also detected some small amounts of acetone (Eq. (8)). This indicated that both homogeneous (Eqs. (10) and (11)) and heterogeneous (on the reactor wall and quartz chips, Eq. (8)) reactions were taking place in our reactor. The relative percent remaining, which is probably close to the absolute value since there is little decomposition of the two isomeric compounds (HAc and HAA) at below 400°C, is plotted in Fig. 5 as a function of temperature. Normal and pyrolysis MB mass spectra of HAc obtained using 25 eV EI are shown in Fig. 6. At 700°C, complete cracking of acetic acid was achieved and main pyrolysis products (Fig. 6d, H 2 0 is not shown) were H2, CO2, methane, CO, ethylene, and ketene, suggesting that reactions in Eqs. (12) and (13) take place at higher temperatures; benzene ( m / z 78) and other stable aromatic compounds were also detected in small amount. Only 15% of acetic acid remained (85% conversion) at 600°C and "r = 0.2 s over quartz in the absence of steam, compared to 50% conversion with steam present (Fig. 5a). The significant effect of steam on acetic acid pyrolysis is probably because one of the major decomposition channels, dehydration to form ketene and water, is a reversible reaction under the temperatures studied [47]. It is interesting to note that the 260 D. Wang et al. /Applied Catalysis A: General 245-270 100% ! o i 80% 60% ee- E 40% i 2O% i 0% 300 r 500 400 600 700 Temperature (°C) 100% >~" 80% t ~ c c •N 6o% \, 40% E \ 2o%iI 0%, 300 \ \, , , 400 - 7 - - 500 Temperature r 600 700 (°C) Fig. 5. Percent remaining ( = 100% -conversion), relative to the intensity at 350°C, in the thermal decompositions of (a) acetic acid and (b) hydroxyacetatdehyde at varying temperatures. Other conditions: all at a residence time (~-) of 0.2 s; with steam ( S / C = 5) (O, II), with steam ( S / C = 2) ( + ) , and without steam (O, []). Lines are only visual aids. decarbonylation reaction (Eq. (14)) does not occur thermally, even through it should become thermodynamically feasible at 540°C and above. This is probably due to a large energy barrier that is much greater than the activation energies known for the other two reaction channels: (301 k J / m o l for decarboxylation and 317 k J / m o l for dehydration [50]). The dehydration reaction has been observed to occur at much lower temperatures than 600°C [48,49], most likely involving catalysis by the reactor wall instead of a purely homogeneous decomposition. The ketonization reaction (Eq. (8)) was found to take place more rapidly and to completion on the shift catalyst and on catalyst support materials such as A1203, consistent with literature findings [39,51]. The resulting product, D. Wang et al. / Applied Catalysis A: General 245-270 261 43 / 100 ¸ a 60 45 50 15 28 i, i, 10 20 41 II , 30 m/z ,II I 40 60 50 70 42 100 2. II!r 5O 14 0 I~ I,i, 20 10 6O 30 I .i m/z 40 50 60 70 50 60 60 70 42 '°°I 50 1 14 28 0 0 I0 ,, 44 i ,i 2O . 30 m/z 'Oi;., i iI 40 28 d 44 1B 0 ,I lO 78 ,I il I , 2O 30 40 mlz 50 60 70 80 Fig. 6. (a) T h e 25 e V E I molecular beam mass spectrum of acetic acid. (b)-(d) Pyrolysis mass spectra of acetic acid without steam over quartz chips. Conditions: (b) 500°C and -r = 0.2 s, (c) 7 0 0 ° C and "r = 0.2 s, and (d) 7 0 0 ° C and "r = 4.2 s. The ordinate is relative abundance (base p e a k = 100). acetone, is rather stable, but may undergo aldol condensation reactions under catalytic conditions to form mesityl oxide, isophorone, and mesitylene, among other products [52]. CH3COzH --~ CH3OH + CO (14) The shift catalyst failed to show any reforming activity for acetic acid and produced no hydrogen under the same operating conditions, whereas complete conversion of methanol to hydrogen was observed. At 350°C, the catalyst converted HAc to acetone (Fig. 7a) when it encountered the first wave of HAc molecules; after this there was complete breakthrough of HAc. These observations agree well with studies described in the literature of ethanol reforming on 262 D. Wang et al./Applied Catalysis A." General 245-270 43 100 6O 50 45 14 29 ,il, . It.. 10 20 30 I m/z40 50 6O 43 100 7O b 58 50 15 .,,I. 0 0 10 28 ,i, 20 30 .11 : m/z 40 50 60 70 50 60 70 44 100 ' 50 2 ol 0 28 i 10 20 ,, 30 m/z40 Fig. 7. Mass spectra of acetic acid after catalytic steam reforming (Gc HSV = 1120 h 1 and S / C = 5) with: (a) UCI C18HC shift catalyst at "r = 0.06 s and 350°C, (b) UCI C18HC shift catalyst at "r = 0.06 s and 400°C, and (c) UCI G-90C catalyst at 600°C and -r = 0.01 s. The ordinate is relative abundance (base peak = 100). these types of catalysts resulting in the formation of ethyl acetate and acetic acid as the final products. At 400°C, HAc reacted completely to form acetone and CO 2 as the two predominant products, along with a small amount of ketene but no hydrogen (Fig. 7b). With the G-90C catalyst at temperatures above 400°C, no intermediates were found in steam reforming reactions of acetic acid and acetone (Fig. 7c and Fig. 7d), and conversions of both compounds were complete. In Fig. 8, the surrogate ion signals (i.e., m/z 2, 15, 18, 28, 31, 44, and 60 for H 2, CH 4, H20, CO, MeOH, CO 2, and HAc, respectively) are plotted in Fig. 8 against the time (in min) during the run, whereas the feed (methanol or acetic acid) through the catalyst bed, in the presence of steam, was switched on and off by turning on 263 D. Wang et al. /Applied Catalysis A: General 245-270 CH~OH = : [__.CO+ 2Hz l ,, e= o ;: ; ,~,e,.,,~& , , CO , ,,,,, ,,. ~ ; ;, No tailing No coking H2 CO~ ._~ CO : ........... ".-', .......... 1 30 " ........ '....... CH, CH3OH tO N 40 50 Time (rain) IOFF 60 b ~ ~ ~ I" - -C~}~ r ' l !1 . . . . . . --~" I CC +-COL+..~2Hz ~ ~ . ,: = - ,,,, : CO Tailing on H2 and C02 "/ ....... , L~h ~ // Ca~/rbon 'x~ formation CO~ e- ~'.'-L..~,~,,,,,.,,,,,.,,,,...,,,.,.,,,,,, CO CH4 H~O ~ ~ CH~CO~H ,',',#,% ,~;.;,.~,~~ ;,,,~,,\,~.~,..~,~,,%~ ~,v%,,.,,,;,,,:,~,,~, t'~,.,,-,~'~'~''¢;'' '~'~~'~/~'~'~ 60 65 70 ON 75 Time (min) 80 85 )FF 90 Fig. 8. Ion profiles (not to scale) for catalytic steam reforming o f (a) methanol and (h) acetic acid at 700°C, Gc H S V = 6720 h ~, S / C = 5, and "r = 0 . 0 0 4 s with UC! G-90C catalyst. and off the syringe pump and the resulting mass spectra were recorded in real time. Decreases in steam and the catalyst bed temperature (as much as 50°C) were observed during reforming, because of the endothermic reforming reaction and the catalyst bed size. Another interesting observation was that acetic acid reforming usually resulted in much slower return of signals corresponding to H 2 (m/z 2) and CO 2 (m/z 44) (but not CO at m/z 28) to background levels after the feeding of HAc was stopped but steam was still present. This is clearly shown in Fig. 8b. In contrast, this tailing behavior was not observed for the reforming of methanol (Fig. 8a) and hydroxyacetaldehyde (data not shown). The tailing of H 2 and CO 2 may arise from steam reforming of strongly 264 D. Wang et al. /Applied Catalysis A: General 245-270 adsorbed HAc molecules. However, this is unlikely at temperatures as high as 700°C. Alternatively, it may be explained by the formation of coke on the catalyst during steam reforming of HAc. Because the coke removal by steam was slower than its formation, the coke accumulated on the catalyst was converted t o H 2 and CO 2 by steam when we interrupted the feeding of acetic acid over the catalyst (Eq. (15)). We favor the latter explanation and consider that acetic acid forms significant amounts of carbon on the catalyst during steam reforming. This is supported by the fact that acetic acid has been shown to form adsorbed acetate species at low temperatures on clean metal surfaces such as R h ( l l l ) [53] and P d ( l l l ) [54], and the acetate species decompose to yield H 2 and CO2, leaving adsorbed carbon on the surface (Eqs. (16)-(19)). Coke accumulation on the UCI G-90C catalyst has also been observed in our bench-scale experiments [55]. The formation of more methane from HAc than from HAA is also consistent with the intermediacy of adsorbed methyl species, (CH3)ads, which may desorb from the surface as CH 4 by combining with an Hads (Eq. (20)). Cad s "1- H 2 0 ~ GOad s "1- H 2 (15) CH3CO2H ~ (CH3CO2)ad ~ + Hads (16) (cn3cO2)ad (17) s ----) ( c n 3 ) a d s if- C O 2 ( C H B ) a d s ---) Cad s q- 3Had s (is) Overall: CHBCO2H ~ Cad s -I- 2H 2 + CO 2 (19) (CH3)ad s + Hads --~ CH 4 (20) Because no intermediates other than thermal decomposition products were detected during the catalytic steam reforming of acetic acid, we can conclude that the reforming reactions of the thermally cracked products take place rapidly. The likely pathway for ketene, one of the important products by gas-phase dehydration of HAc, is dissociative adsorption to form CO and CH x (x = 1-2) species, the latter being another possible source of coke. The surface chemistry of ketene on Fe(110) [56], Pt(111) [57], and Ru(001) [58,59] has been studied, and the main surface species observed in these reports were CO, CO 2, acetaldehyde, acetyl, and hydrocarbon fragments such as CH x (x = 1-3), HC-----C and H2C=C, depending on surface temperature. Detailed studies of ketene steam reforming were not performed and we only examined briefly acetone steam reforming as mentioned above. 3.4. Hydroxyacetaldehyde We are not aware of any previous study on HAA thermal decomposition. The effect of steam on HAA pyrolysis was found to be opposite to but less profound D. Wang et al. /Applied Catalysis A." General 245-270 265 than that in the case of HAc. In the absence of steam, HAc decomposed more readily than HAA (Fig. 5), while the two isomers had similar conversions with steam present. Increasing S / C from 2 to 5 resulted in more decomposition of HAA. The data for the pyrolysis of HAA in the absence of steam had greater uncertainty because only 3 to 5 mg of solid HAA dimer sample were fed in pulses using a quartz boat in these experiments. The conversion of HAA was in the range of 40% to 50% at 600°C and "r = 0.2 s, whether steam was present or not. In addition to cracking to smaller molecules, HAA was found to undergo self-condensation reactions to form many higher-molecular-weight species. loo] 31 I II 15 ]8 10 o 42 60 . . . . . 30 20 a i1,, 40 m/z 50 60 70 b 31 100 50 i,Ii 29 15 I II, 0 10 0 20 30 31 100 1 68 . .li, h . . 20 60 70 C 84 96 110 124 I . , I, 50 43 15 . 40 m~ ~60 X 4O 72 m 50 211 43 Ill, ,., , 6O 40 28 138 148 162 lllhl,,,Itl,,,,,I, II,l,,.,,,,I,~t,,,,.,,,~l,L,,,..,,,,,,,~.~.,,.,,i ..... ,,,,L.,.,, . 80 rn/z 100 120 140 160 x 2o 100 d 58 :~ 50 1,3 ] "~ lzB 142 lil .lll.'..,tli.I,L..i,l,,,...h,,,, o6 ,,,,.,. 116 o ...,11,,I .ll!il..ill, i.tlli.l,!,..hll!,...ll~,,,, II,l, . . . , , , ,,26 ..................... 20 40 60 80 m/z 100 120 140 160 Fig. 9. (a) The 25 eV EI molecular beam mass spectrum of hydro×yaceta]dehyde. (b)-(d) Pyrolysis mass spectra of acetic acid over quartz chips. Conditions: (b) with steam ( S / C = 5) at 500°C and "r = 0.6 s, (c) with steam ( S / C = 5) at 600°C and 7 = 0.6 s, and (d) without steam at 700°C and "r = 0.2 s. The ordinate is relative abundance (base peak = 100). D. Wang et al. /Applied Catalysis A: General 245-270 266 > 44 100 ' 58 x 10 72 50 67 16 2 0 .,I . . 28 , ,.I . .,I,, I,.,.I 20 0 86 60 100 80 m/z 120 140 > 44 100 ,,,,I, 40 x4 58' 96 67 50 ¸ 28 72 i 2 110 ..,,I...,.l,, I,II I,,,,.,,I ll,..I,I III I,I II. ,.llII • 20 40 . . ; , 80 60 ]'", . . , m/z . . 1O0 ,,. ,.,,,l,...,.,,,I,l., ..,llJ.,.,.....,,I,I . . . . . . 120 . . . 140 . . . . 44 100 ......... 160 I., 180 c 5O 2 28 1,~ 0 10 I 20 . 30 m/z 40 50 60 70 Fig. 10. Mass spectra of catalytic steam reforming of hydroxyacetaldehyde (Gc H S V = 1120 h 1 and S / C = 5) with (a) UCI C18HC shift catalyst at "r = 0.06 s and 350°C, (b) UCI C I 8 H C shift catalyst at "r = 0.06 s and 400°C, and (c) UCI G-90C catalyst at 600°C and "r = 0.01 s. The ordinate is relative abundance (base peak = 100)• Products of HAA pyrolysis included a series of aldehydes and ketones (CnH2nO, n = 1-6 at m/z 29/30, 43/44, 58, 72, 86, and 100; CH 2 = CHCHO at m/z 56; see Fig. 9); they were likely formed from acid-catalyzed reactions of HAA involving the quartz bed. Char formation was also an important aspect of HAA thermal decomposition. At 700°C, major products from HAA pyrolysis were methane, CO, formaldehyde, ketene, acetaldehyde and CO 2 (Fig. 9d). With the WGS catalyst, some hydrogen formation was observed ( < 5% of the stoichiometric amount) from HAA at 350 and 400°C. Other more abundant products (Fig. 10a and Fig. 10b) were a series of aldehydes, including formaldehyde (m/z 29, 30), acetaldehyde (m/z 43, 44), propanal (m/z 57, 58), D. Wang et al. /Applied Catalysis A: General 245-270 267 butanal (m/z 71, 72), and pentanal (m/z 85, 86), as well as furfural (m/z 95, 96) and dihydroxybenzene compounds at m/z 110, 124, 138, and 152. These results provide further support for the catalytic effect of quart suggested above. Over the G-90C catalyst, steam reforming of HAA was also found to be more facile than that of HAc, probably because of its lower stability compared to HAc. Alternatively, this may arise from differences in the reaction mechanism, which will be further discussed below. Again, we did not detect any intermediate for this catalytic reaction (Fig. 10c). The initial step in the dissociative adsorption of HAA on the catalyst is likely to be a complete breakdown of the HAA molecule to form COad~ and Had~ because the cleavage of the C - C bond will probably leave the C - O bond intact (Eqs. (21)-(24)). This will result in no direct formation of any adsorbed carbon on the catalyst (indirect carbon formation from the disproportionation of CO (Eq. (5)) may still occur if there is insufficient steam). Under the same operating conditions, yields of H 2 and CO 2 from HAA are higher than from HAc, by 5-10%. An observed lower methane yield also supports a mechanism that involves no CH x intermediates. HOCHeCHO --. (HOCH2CO)ads + Had~ (21) (HOCH2CO)ad~ ---> (HOCH2)ad s + COads (22) (HOCH2)aa ~ ~ COati~+ 3H~d~ (23) Overall: HOCH2CHO + 2COad~ + 2H 2 (24) 4. Conclusions Oxygenated compounds characteristic of fast pyrolysis oil from biomass (i.e., biocrude) can be steam reformed using Ni-based catalysts. Two of the major components of the carbohydrate-derived fraction in the pyrolysis oil have been studied in detail in this paper. Acetic acid thermally decomposes in the gas phase, at 500°C or higher, without forming significant amounts of coke. Catalytic reforming through an adsorption step and subsequent surface reactions will reform any intermediates. The proposed mechanism involves an adsorbed methyl species and their subsequent conversion on the catalyst surface. The formation of carbon on the catalyst can become problematic if steam is in insufficient amounts to remove it. Temperature ramping in the reformer can be adjusted to achieve optimal operating conditions with a minimal accumulation of carbon on the catalyst, given the high thermal decomposition temperature observed. Hydroxyacetaldehyde decomposes thermally at temperatures in the range of 268 D. Wang et al. /Applied Catalysis A: General 245-270 350-400°C and yields solid carbonaceous deposits as a major product. Thus, HAA steam reforming requires a low inlet temperature and short residence time prior to entering the catalyst bed. Steam reforming of HAA and its thermal decomposition products proceeds rapidly and without coke formation on the catalyst. The yield of hydrogen from HAA is slightly higher than from HAc, and there is less formation of methane from HAA than from HAc. Our results indicate that complete decomposition of HAA to produce CO and H 2 is the most important pathway for the catalytic steam reforming reaction. Our results suggest that there are significant differences in steam reforming mechanisms of hydrocarbons and oxygenates. Most of the oxygenates found in biomass pyrolysis oils are thermally unstable. At the operating temperatures of a steam reformer, these oxygenates undergo homogeneous (gas-phase) thermal decomposition, as well as cracking reactions on the acidic sites on the support of the steam reforming catalysts. These reactions compete with the steam reforming reaction to hydrogen. Spillover of H atoms from the Ni surface to the A1203 support has been suggested to occur for methanol [60]; this spillover effect may facilitate steam reforming of oxygenates by hydrogenation of the oxygenate molecules adsorbed on the support. The present study does not address optimal catalyst operating conditions. This task will be addressed, after we complete our model compound studies, by carrying out experiments using pyrolysis oils in a bench scale reformer. We are also studying the steam reforming mechanisms of other major oxygenate types present in biomass pyrolysis oils, including furans and phenolics. Work is in progress in these areas. Acknowledgements We thank Dr. Stefan Czernik, Dr. Steve C. Gebhard, Dr. David K. Johnson, Dr. Margaret K. Mann, Dr. Thomas A. Milne, and Dr. Catherine E. Gr6goire Padr6 of NREL for helpful comments and suggestions. Funding for this work by the NREL-managed Hydrogen Program from US DOE (Neil Rossmeissel, manager) under contract DE AC 36-83CH10093 is also acknowledged. References [1] [2] [3] [4] [5] K.Z. Patil, Chemical Age of India, 38 (1987) 519. K. Takahashi, N. Takezawa and H. Kobayashi, Appl. Catal., 2 (1982) 363-366. N. Iwasa and N. Takezawa, Bull. Chem. Soc. Jpn., 64 (1991) 2619. W.B. Hauseman, Int. J. Hydrogen Energy, 19 (1994) 413. M.P. Aznar, J. Corella, J. Delgado and J. Lahoz, Ind. Eng. Chem. Res., 32 (1993) 1. [6] J, Arauzo, D. Radelin, J. Piskorz and D.S. Scott, in A.V. Bridgewater (Editor), Advances in Thermochemical Biomass Conversion, Elsevier Applied Science, London, 1993, pp. 201-215. D. Wang et al. / Applied Catalysis A: General 245-270 269 [7] M.-H. Rei, F.-S. Lin and T.-B. Su, Appl. Catal., 26 (1986) 27. [8] M. Mann, S. Czernik, D. Wang, C. Gregoire and E. Chornet, Prep. Pap. Am. Chem. Soc. Div. Fuel Chem., 39 (1994) 1034. [9] D. Radlein, J. Piskorz and D.S. Scott, J. Anal. Appl. Pyrolysis, 19 (1991) 41. [10] J. Piskorz, D.S. Scott and D. Radlein, in E.J. Soltes and T.A. Milne (Editors), Pyrolysis Oils from Biomass, American Chemical Society, Washington, DC, 1988, pp. 167-178. [11] R.G. Graham and B.A. Freel, in A.V. Bridgewater and J.L. Kuester (Editors), Research in Thermochemical Biomass Conversion, Elsevier Applied Science, London, 1988, pp. 629-641. [12] J. Dielbold and J. Scahill, in E.J. Soltes and T.A. Milne (Editors), Pyrolysis Oils from Biomass: Producing, Analyzing and Upgrading (ACS Symposium Series, Vol. 376), ACS, Washington, DC, 1988, pp. 31-40. [13] C. Roy, B. deCaumia and P. Plante, in G. Grassi, G. Goose and G. dos Santos (Editors), 5th Eur. Conference on on Biomass for Energy and Industry, Elsevier Applied Science, London, 1990, pp. 2595-2599. [14] D.C. Elliott, in E.J. Soltes and T.A. Milne (Editors), Pyrolysis Oils from Biomass: Producing, Analyzing and Upgrading (ACS Symposium Series Vol. 376), ACS, Washington, DC, 1988, pp. 55-65. [15] J. Piskorz, D.S. Scott and D. Radlein, in E.J. Soltes and T.A. Milne (Editors), Pyrolysis Oils from Biomass: Producing, Analyzing and Upgrading (ACS Symposium Series, Vol. 376), ACS, Washington, DC, 1988, pp. 167-178. [16] E. Chornet, D. Wang, D. Montan~, S. Czemik, D. Johnson and M. Mann, in Proceedings 1995 US DOE Hydrogen Program Review, NREL/CP-430-20036, April 1995, Coral Gables, FL. [17] J.P. Van Hook, Catal. Rev. Sci. Eng., 21 (1980) 1. [18] J.R. Rostrup-Nielsen, J. Catal., 31 (1973) 173. [19] J.R. Rostrup-Nielsen, Steam Reforming Catalysts, Danish Technical Press, Copenhagen, 1975. [20] J.R. Rostrup-Nielsen, Chem. Eng. Prog., 73 (1977) 87. [21] P.B. Tottrup and B. Neilsen, Hydrocarbon Processing, March (1982) 89. [22] N. lwasa, S. Kudo, H. Takahashi, S. Masuda and N. Takezawa, Catal. Lett., (1993) 211. [23] C.J. Jiang, D.L. Trimm, M.S. Wright and N.W. Cant, Appl. Catal. A, 97 (1993) 145. [24] C.L Jiang, D.L. Trimm, M.S. Wright and N.W. Cant, Appl. Catal. A, 93 (1993) 245. [25] T.-B. Su and M.-H. Rei, J. Chinese Chem. Soc., 38 (1991) 535, and references therein. [26] N. Iwasa, O. Yamamoto, T. Akazawa, S. Ohyama and N. Takezawa, J. Chem. Soc. Chem. Commun., (1991) 1322. [27] N. Takezawa, M. Shimokawabe, H. Hiramatsu, H. Sugiura, T. Asakawa and H. Kobayashi, React. Kinet. Catal. Lett., 33 (1987) 191. [28] K. Misuno, K. Yoshikawa, N. Wakejima, Y. Takeouchi and A. Watanabe, Chem. Lett., Chem. Soc. Jpn., (1986) 1969. [29] J.C. Amphlett, M.J. Evans, R.F. Mann and R.D. Weir, Can. J. Chem. Eng., 63 (1985) 605, and references therein. [30] K. Takahashi, H. Kobayashi and N. Takezawa, Chem. Lett., Chem. Soc. Jpn., (1985) 759. [31] J.R.H Ross, M.C.F. Steel and A. Zeini-Isfahani, J. Catal., 52 (1978) 280. [32] M. Morita, A. Sato and Y. Matsuda, Nippon Kagaku Kaishi, (1993) 164. [33] E.Y. Garcia and M.A. Laborde, Int. J. Hydrogen Energy, 16 (1991) 307. [34] R.F. Gould (Editor), Adv. Chem. Ser. (1969) 90. [35] F. Daly, J. Catal., 61 (1980) 528. [36] D. Duprez, Appl. Catal. A, 82 (1992) 111. [37] M. Assail and D. Duprez, Bull. Soc. Chim. Fr., (1988) 106. [38] S. Gebhard, D. Wang, R. Overend and M.A. Paisley, Biomass and Bioenergy, 7 (1994) 307, and references therein. [39] E. Furimsky, Catal. Rev. Sci. Eng., 25 (1983) 421. [40] R.J. Evans and T.A. Milne, Energy and Fuels, 1 (1987) 123; 311. [41] R.J. Evans and T.A. Milne, in Proceedings 1986 Biomass Thermochemical Conversion Contractors' Meeting, PNL-SA-13571, CONF-8510167, NTIS: Springfield, VA, 1986, pp.57-79. [42] S. Rajadurai, Catal. Rev.Sci. Eng., 36 (1994) 385. [43] J.T. Richardson, Principles of Catalyst Development, Plenum Press, New York, 1989. [44] M. J. Antal, Jr., Adv. Solar Energy, (1982) 61. 270 D. Wang et al. /Applied Catalysis A." General 245 270 [45] J. R. Anderson and K.C. Pratt, Introduction to Characterization and Testing of Catalysts, Academic Press, Orlando, 1985, p. 276. [46] S. K. Goral, Steam Reforming of Aromatic Compounds, PhD Thesis, University of Utah, Salt Lake City, UT, 1979. [47] J.C. Mackie and K.R. Doolan, Int. J. Chem. Kinet., 16 (1984) 525, and references cited therein. [48] P.G. Blake and G.E. Jackson, J. Chem. Soc. B, (1969) 94. [49] W.B. Guenther and W.D. Waiters, J. Am. Chem. Soc. 81 (1959) 1310. [50] M.T. Nguyen, D. Sengupta, G. Raspoet and L.G. Vanquickenborne, J. Phys. Chem., 99 (1995) 11883. [51] M.Glifiski, J. Kijefiski and A. Jakubowski, Appl. Catal. A, 128 (1995) 209. [52] W.T. Reichte, J. Catal. 63 (1980) 295. [53] Y. Li and M. Bowker, Surf. Sci. 285 (1993) 219. [54] N. Aas and M. Bowker, J. Chem. Soc. Faraday Trans., 89 (1993) 1249. [55] D. Montan~, D. Wang, S. Czernik and E. Chomet, in preparation. [56] P.H. McBreen, W. Erley and H. Ibach, Surf. Sci., 148 (1984) 292. [57] G.E. Mitchell, P.L. Radloff, C.M. Greenlief, M.A. Henderson and J.M. White, Surf. Sci. 183 (1987) 403. [58] M.A. Henderson, P.L. Radloff, J.M. White and C.A. Mims, J. Phys. Chem., 92 (1988) 4111. [59] M.A. Henderson, P.L. Radloff, C.M. Greenlief, J.M. White and C.A. Mires, J. Phys. Chem., 92 (1988) 4120. [60] R.L. Flesner and J.L. Falconer, J. Catal., 150 (1994) 301.