Energy & Fuels 2007, 21, 3637–3643
3637
Noncatalytic Gasification of Cellulose in Supercritical Water Fernando L. P. Resende, Matthew E. Neff, and Phillip E. Savage* Department of Chemical Engineering, UniVersity of Michigan, 2300 Hayward St., Ann Arbor, Michigan 48109-2136 ReceiVed April 27, 2007. ReVised Manuscript ReceiVed July 23, 2007
We gasified cellulose in supercritical water, in the absence of heterogeneous catalytic effects, by using quartz reactors. We also report the first systematic study of the effects of temperature, cellulose loading, water density, and reaction time on the production of H2, CH4, CO, and CO2 from supercritical water gasification. The results show that the total gas yields and H2 mole fraction are lower in quartz reactors than in stainless steel reactors, suggesting that the gases from previous studies in metal reactors arise from both homogeneous and heterogeneous reactions, even in the absence of an added catalyst. The rate of formation for all gas species increases with temperature. Manipulating cellulose loading and water density provides an efficient means to control the product selectivity, since the relative amounts of H2 and CH4 were strongly influenced by these two process variables.
Introduction The conversion of cultivated crops or agricultural residues into liquid or gaseous fuels has become an increasingly interesting alternative to fossil fuels. Using biomass, which is renewable and carbon-neutral, represents a more sustainable energy future. It can also reduce countries’ dependence on imported oil. Substantial effort has been made to develop technologies that can efficiently produce fuel gases from biomass in a clean manner. These technologies often run into operational problems, such as production of residues like tar and char1 and low thermal efficiency due to the high moisture content in many kinds of biomass.2 Formation of tars, mainly composed of furfural and phenols, is the chief obstacle to complete gasification of biomass.3 Supercritical water gasification (SCWG) refers to hydrothermal gasification at conditions that exceed the thermodynamic critical point of water (374 °C, 218 atm). SCWG has some advantages over conventional gasification methods. The use of water as solvent and reaction medium eliminates the need to dry wet biomass, which in turn increases thermal efficiency.4,5 Supercritical water has the ability to dissolve organic components of plant biomass such as cellulose and lignin, leading to hydrolysis reactions that degrade the organic polymeric structure, in contrast with the pyrolysis reactions that dominate conventional gasification. Pyrolysis produces significant amounts of char and tar, but hydrolysis products are highly soluble in hot compressed water, which suppresses formation of char and * Corresponding author. E-mail:
[email protected]. (1) Antal, M. J., Jr.; Allen, S. G.; Schulman, D.; Xu, X.; Divilio, R. J. Ind. Eng. Chem. Res. 2000, 11, 4040–4053. (2) Schmieder, H.; Abeln, J.; Boukis, N.; Dinjus, E.; Kruse, A.; Kluth, M.; Petrich, G.; Sadri, E.; Schacht, M. J. Supercrit. Fluids 2000, 2, 145– 153. (3) Kruse, A.; Gawlik, A. Ind. Eng. Chem. Res. 2003, 2, 267–279. (4) Lanzetta, M.; Di Blasi, C. J. Anal. Appl. Pyrolysis 1998, 2, 181– 192. (5) Matsumura, Y.; Minowa, T.; Potic, B.; Kersten, S. R. A.; Prins, W.; van Swaaij, W. P. M.; van de Beld, B.; Elliott, D. C.; Neuenschwander, G. G.; Kruse, A.; Antal, M. J., Jr Biomass Bioenergy 2005, 4, 269–292.
tar.1,3,6 It has been shown by Herguido et al.7 that biomass such as pine sawdust and pine wood reacts with water via the steamreforming reaction above 800 °C, nearly eliminating formation of tar. The char product remains unconverted, though, below the critical point of water. In supercritical water with the use of a Raney nickel catalyst, Waldner and Vogel8 report that no tar or char was present and the aqueous phase was colorless for gasification of wood. Once gas products are formed from the intermediates, the high water content drives the water-gas-shift reaction (CO + H2O f CO2 + H2),2,5 leading to a H2-rich gas with low CO yield. If the H2 is intended to be used in a proton-exchange membrane fuel cell, CO is a poison so the water-gas-shift reaction becomes an important tool to accomplish CO elimination. Furthermore, the gas products from SCWG are obtained at high pressure, which is often desired for subsequent use.5,9 As the main component of biomass, cellulose has been extensively studied, and its decomposition has been carried out at both subcritical and supercritical conditions.10,11 In supercritical water, cellulose initially hydrolyzes, producing oligomers that further hydrolyze to form glucose. Glucose then undergoes an extraordinary variety of reactions (e.g., dehydration, retroaldocondensation, isomerization) to ultimately produce gases, which consist mainly of CO2, H2, and CH4.12 Species in the gas phase can react with each other, and depending on the reaction conditions, this can greatly affect the product distribution. Table 1 summarizes previous work on cellulose SCWG in the absence of an added catalyst. Most previous work considered (6) Kruse, A.; Henningsen, T.; Sinag, A.; Pfeiffer, J. Ind. Eng. Chem. Res. 2003, 16, 3711–3717. (7) Herguido, J.; Corella, J.; Gonzalez-Saiz, J. Ind. Eng. Chem. Res. 1992, 5, 1274–1282. (8) Waldner, M. H.; Vogel, F. Ind. Eng. Chem. Res. 2005, 13, 4543– 4551. (9) Boukis, N.; Diem, V.; Habicht, W.; Dinjus, E. Ind. Eng. Chem. Res. 2003, 4, 728–735. (10) Lee, I.; Kim, M.; Ihm, S. Ind. Eng. Chem. Res. 2002, 5, 1182– 1188. (11) DiLeo, G. J.; Savage, P. E. J. Supercrit. Fluids 2006, 2, 228–232. (12) Osada, M.; Sato, T.; Watanabe, M.; Adschiri, T.; Arai, K. Energy Fuels 2004, 2, 327–333.
10.1021/ef7002206 CCC: $37.00 2007 American Chemical Society Published on Web 09/28/2007
3638 Energy & Fuels, Vol. 21, No. 6, 2007
Resende et al.
Table 1. Summary of Previous Research on SCWG of Cellulose (All in Stainless Steel Reactors)
a
reference
temp (°C)
Yoshida and Matsumura30 Watanabe et al.22 Williams and Onwudili31 Osada et al.12 Minowa et al.19,32 Hao et al.29 Lu et al.33
400 400, 440 380 400 400 up to 650 500
water density (g/cm3) 0.166 0.2, 0.35 0.2 0.33 0.05 up to 0.094 0.071
cellulose (wt %) NRa 4.8 5 5 20, 40 10 9
time (min) 20 10, 15 up to 120 up to 180 60 20 20
NR ) not reported.
a single temperature, water density, biomass loading, and reaction time. The focus in these earlier studies was often on the effect of different catalysts on SCWG. These previous studies make it difficult to learn the effects of the process variables. This article provides results for the first systematic study of the effects of temperature, biomass concentration, water density, and reaction time on cellulose SCWG. The work presented here is also unique because it is the first to report SCWG of cellulose in a metal-free reactor. We used quartz capillary tubes as mini-batch reactors. Previous studies with no added catalyst have been performed in stainless steel reactors. The internal walls of these reactors catalyze some of the reactions in SCWG, so at this moment it is not clear how much influence the heterogeneous catalytic reactions have on the product yields. DiLeo and Savage13,14 and Kersten et al.15 report on SCWG of simple compounds in quartz reactors, and all three studies show that gas yields are strongly influenced when a metal catalyst is added to the system. One of our motivations for the present work was to obtain results for cellulose gasification in SCW that would be attributable exclusively to noncatalytic reactions. This information would be very useful for subsequent evaluation of different catalysts because the noncatalytic contribution could be subtracted out and the effect of the catalyst alone can be more clearly seen. Experimental Section Microcrystalline cellulose powder (0.6 g/cm3, particle size 0.1 wt % + 60 mesh, 70 wt % + 325 mesh) was purchased from Sigma-Aldrich and used as received. Quartz capillary tubes (2 mm i.d., 5 mm o.d., 18.4 cm length) sealed at one end served as mini-batch reactors. The reactor volume was 0.58 cm3. Deionized water was loaded into the reactor before the cellulose, so its expansion during heating would favor mixing. No catalysts were used. The amounts of cellulose and water loaded varied with the reaction conditions to be used. Setting the reaction temperature and pressure fixed the water density. All water densities were obtained from the steam tables. The desired water density and the reactor volume then set the amount of water to be loaded. Finally, the desired biomass concentration (wt %) was used to calculate the mass of cellulose that should be loaded into the reactor. The water loadings used in these experiments ranged from 31 µL (water density 0.05 g/cm3) to 103 µL (0.18 g/cm3), and the cellulose loadings ranged from 2.6 mg (5 wt % concentration) to 26.0 mg (33.3 wt % concentration). After being loaded, the reactors were flame-sealed. SCWG was then performed by immersing the sealed capillary in a preheated, isothermal fluidized sand bath. The heat up time for (13) DiLeo, G. J. Gasification of biomass model compounds in supercritical water. PhD Dissertation, University of Michigan, Ann Arbor, 2007. (14) DiLeo, G. J.; Neff, M. E.; Savage, P. E. Energy Fuels 2007, 21, 2340–2345. (15) Kersten, S. R. A.; Potic, B.; Prins, W.; Van Swaaij, W. P. M. Ind. Eng. Chem. Res. 2006, 12, 4169–4177.
the reactors is about 30 s. Reactions were performed for times ranging from 2.5 to 30 min. After being removed from the sand bath, when the desired reaction time had elapsed, the reactors were placed in front of a fan to speed cooling. They were then placed inside a freezer overnight. This step is intended to prevent the water and other liquid products from entering the gas chromatographic column during gas sample introduction. To collect the gases formed during SCWG, the quartz reactor was first inserted into a 10 mm i.d. × 20 cm long metal tube, which is then sealed with a Swagelok cap at one end and a valve at the other. The tube was then connected to an argon cylinder, pressurized to 700 psi, and then removed from the cylinder. Next the metal tube was struck sharply to shatter the quartz tube within and release the product gases into the argonfilled metal tube. The metal tube was then connected to a gas sampling valve on a 5890 Series II Hewlett-Packard gas chromatograph (GC) with a thermal conductivity detector (TCD). We used argon as the carrier gas. The GC separates and detects H2, O2, N2, CO, CO2, and CH4. Prior to the analysis of reaction products, the GC was calibrated with 10 commercial gas standards containing the components of interest in the range of concentrations observed during experiments. The GC analysis provides the gas composition. The absolute amount of each component present in the gas phase was calculated using N2 (from air initially in the reactor and metal tube) as an internal standard. Since the amount of N2 from air is known (and it does not react during gasification), and the molar ratio of any gas to N2 is determined experimentally, the yields in mmol/g for each gas species can be calculated. The amount of each gas dissolved in the aqueous phase at the reactor conditions inside the freezer (T ) -3 °C) was then determined from Henry’s law. Experiments at 10 min were repeated four times to determine representative standard deviations. Repetitions for other times were performed as needed. Most runs were reproduced at least twice and at most six times. Mean values are reported. In general, standard deviations for data points were high. Random error is inherent in the experimental methods used. It is magnified in this study, however, because of the molecular complexity of cellulose and the potential run-to-run variability in the spatial distribution of cellulose and water in the reactor during the heat-up time. Quartz is slightly soluble in supercritical water, but at 600 °C, only about 0.04% of the reactor material would be leached out at equilibrium.16 Dissolution of quartz, if occurring, did not compromise the structural integrity of the reactors. Di Leo13 showed that SCWG of formic acid at 600 °C for several hours led to experimental gas yields that agreed well with yields expected at equilibrium. This agreement indicates that the reactors maintained their structural integrity (no microcracks or gas leaks). (16) Fournier, R. O. Geochim. Cosmochim. Acta 1982, 10, 1969–1973.
Gasification of Cellulose in Supercritical Water
Energy & Fuels, Vol. 21, No. 6, 2007 3639
We arbitrarily designate SCWG at 500 °C, 9.0 wt % biomass, and 0.08 g/cm3 water density (22 MPa) as the base case. From this starting point, we varied one parameter at a time to evaluate its effect on the gas yields and composition. We considered temperatures of 400, 500, and 600 °C, biomass loadings of 5.0, 9.0, and 33.3 wt %, and water densities of 0.05, 0.08, and 0.18 g/cm3. At some of the high temperature, high density, and high biomass loading conditions evaluated, reactors consistently burst at the longer reaction times. For these conditions, results could be obtained only for the shorter times. Results and Discussion
Figure 1. Temporal variation of gas composition (base case).
al.17
showed that cellulose is completely converted Sasaki et in water at 350 °C and 25 MPa after only 4 s. We are interested in gas formation, however, and not only the disappearance of the reactant. Once the cellulose is quickly consumed, it forms intermediates that take longer to decompose and form gases. These gases can also react further among themselves, changing the product distribution. Visual observation of the quartz reactors indicates the presence of char (a black residue adhering to the reactor internal walls) in all cases. The char strongly adhered to the quartz, making it impossible to collect samples in enough quantity for analysis. As a reference for our discussion and results, we suggest a reaction scheme, based on literature information,1,12,18–22 that accounts for the main solid- and liquid-phase reactions and focuses on the main routes for gas formation. We use the following scheme to describe cellulose SCWG. CxHyOz indicates generic intermediate species resulting from glucose decomposition. cellulose hydrolysis: (C6H10O5)n + nH2O f nC6H12O6 glucose decomposition: C6H12O6 f CxHyOz
(
steam - reforming I : CxHyOz + (x - z)H2O x - z +
y H 2 2
)
steam - reforming II : CxHyOz + (2x - z)H2O f
(
xCO2 + 2x - z + combustion: CxHyOz +
y H 2 2
)
( 4y + x - 2z )O f xCO + 2y H O 2
2
2
char formation through intermediates: CxHyOz f wC + Cx-wHyOz water-gas shift: CO + H2O T CO2 + H2 methanation: CO + 3H2 T CH4 + H2O hydrogenation: CO + 2H2 f CH4 + 1/2O2 The reaction scheme includes combustion via residual air in the reactor because it can form a significant amount of the CO2 observed experimentally depending on the reaction conditions. For instance, at 9.0 wt % cellulose, about 1.0 mmol/g of CO2 can be formed from combustion. As we will show later, this corresponds to about 40–50% of the CO2 yield observed experimentally at this loading. It also corresponds to about only (17) Sasaki, M.; Kabyemela, B.; Malaluan, R.; Hirose, S.; Takeda, N.; Adschiri, T.; Arai, K. J. Supercrit. Fluids 1998, 1–3, 261–268. (18) Elliott, D. C.; Sealock, L. J., Jr Chem. Eng. Res. Des. 1996, A5, 563–566. (19) Minowa, T.; Inoue, S. Renewable Energy 1998, 1–4, 1114–1117. (20) Minowa, T.; Ogi, T. Catal. Today 1998, 1–4, 411–416. (21) Yan, Q.; Guo, L.; Lu, Y. Energy ConVers. Manage. 2006, 11–12, 1515–1528. (22) Watanabe, M.; Inomata, H.; Arai, K. Biomass Bioenergy 2002, 5, 405–410.
2–3% of the carbon in the cellulose. At 33.3 wt % cellulose, combustion can form at most 1.5% of the CO2 yield observed. The formation of char can be attributed to either intermediates that react to form high-molecular-weight compounds or to unhydrolyzed cellulose that pyrolyzes. If all the cellulose does not hydrolyze, conventional gasification reactions23–25 might take place, as follows: pyrolysis: (C6H10O5)n f 5nCO + 5nH2 + 6nC combustion: C + 3/4O2 f 1/2CO + 1/2CO2 hydrogasification: C + 2H2 f CH4 Base Case Results. We first present results for the base case conditions (500 °C, 0.08 g/cm3, 9.0 wt % cellulose) and then proceed to analyze how changes in reaction parameters affect the results. Figure 1 shows the temporal variation of the molar composition (dry basis) of the gases formed for the base case. The error bars in all the figures represent the standard deviation. The lines in the graphs connecting experimental data points are intended only to help the reader see the trends. CO2 and CO are the major products initially, each constituting about 35–45% of the gas. CO2 remains the major product at all times, and its mol % remains relatively stable at all times examined. The mol % of CO, which is a highly undesirable component for some applications, appears to decrease after 10–15 min. This observation is tentative, though, because the data for CO have a larger relative error than the other gases. The thermal conductivity of CO is lower, generating peaks in the GC analysis considerably smaller than the other ones. The mol % of H2 appears to increase with time, as does the mol % of CH4. Figure 2 shows the yields of CO, CO2, CH4, and H2 as a function of time for cellulose SCWG at the base case conditions. The yield of CO2 appears to increase slightly during the first 10 min, and then its yield is relatively stable at about 3.2 mmol/ g. CO2 can be formed from steam-reforming and combustion of intermediate compounds and from water-gas shift. The maximum amount of CO2 that could be formed from the combustion reactions at the base case conditions is about 1.0 mmol/g. In the first 2.5 min, 2.2 mmol/g of CO2 was already formed. Therefore, CO2 must be formed from either steamreforming or water-gas shift during the first 2.5 min. The rate of CO formation during these early minutes is high, though. Therefore, water-gas shift does not appear to be a dominant reaction at this point. Instead, steam-reforming could be taking place, forming both CO and CO2. (23) Figueiredo, J.; Alves, S. Waste Wood Pyrolysis. In Encyclopedia of EnVironmental Control Technology; Gulf: Houston, TX, 1989; Vol. 1, pp 282–286. (24) Bernardo, C. A.; Trimm, D. L. Carbon 1979, 2, 115–120. (25) Jand, N.; Brandani, V.; Foscolo, P. U. Ind. Eng. Chem. Res. 2006, 2, 834–843.
3640 Energy & Fuels, Vol. 21, No. 6, 2007
Resende et al.
Figure 2. Temporal variation of yields (mmol/g) of CO, CO2, H2, and CH4 (base case).
Figure 4. Gas composition at 5 min as a function of temperature.
Figure 3. Temporal variation of C, H, O, and total gas yields (base case).
Figure 5. C, H, O, and total gas yields at 5 min as a function of temperature.
The yield of CO exhibits a maximum of about 3 mmol/g at 10 min. At longer times the yield decreases to less than 1 mmol/ g. This maximum is consistent with our reaction scheme, where CO is formed by steam-reforming and later consumed by watergas shift, methanation, and hydrogenation. The substantial amount of CO initially formed drives the rate of these reactions, leading to CO consumption. Small amounts of H2 and CH4 are detected at 2.5 min. H2 can be formed from steam-reforming and water-gas shift. CH4 formation can be a result of methanation and hydrogenation. In gasification systems, several quantities are normally used to provide an assessment of the overall gasification efficiency. The total gas yield is defined as the sum of the masses of all the gas products (CH4, H2, CO, and CO2 in our case) divided by the mass of cellulose loaded into the reactor. This quantity measures how much of the cellulose was converted into gases. Similarly, the H, C, and O yields are the percentages of the total mass of H, C, or O atoms from the initial cellulose that appear in the product gases. This information is important for estimating the potential to make certain products via further processing of the gases. For instance, if the main goal is to produce H2, and the H2 yields are relatively low but the H yield is high (because of a high CH4 content), this goal can still be achieved by converting the H atoms in CH4 to H2. Figure 3 shows the total gas yield and the H, C, and O yields for the base case conditions. Around 15 wt % of the cellulose is converted into gases in the first 2.5 min. This conversion increases in the first 10 min to about 25%, after which the yield remains approximately stable. Since the most abundant product gases are CO and CO2, the C yield follows a trend similar to that of the total gas. The O yield, however, seems to decrease at the longest reaction time whereas the H yield increases with time reaching 12.5% after 30 min of reaction. Effect of Temperature. The effect of temperature was evaluated by keeping the cellulose loading and water density fixed at the base case values (9.0 wt % and 0.08 g/cm3) and running experiments at 400, 500, and 600 °C.
Figure 6. Gas yields at 5 min as a function of temperature.
Figure 4 shows the effect of temperature on the composition of the product gas at 5 min. As temperature increases, the mole fractions of H2 and CH4 increase, the mole fraction of CO remains roughly constant, and the mole fraction of CO2 decreases. H2 and CH4 are gases we wish to produce because of their economic importance, while lower mole fractions of CO and CO2 are interesting from an environmental point of view. The H2 mol % at 600 °C is about 3 times larger than it is at 400 °C, and an even higher relative increase is observed for CH4. Figure 5 shows the effect of temperature on the C, H, O, and total gas yield. All the yields increase with temperature, and there is a particularly large jump from 500 to 600 °C. The reactions that consume solids and liquids (such as combustion, pyrolysis, and steam-reforming) must have their rates increased to provide the results we observe. The increase in yields is impressive at 600 °C, especially considering the short reaction time (5 min). The results in Figure 5 clearly show that temperature can be used to increase gas yields from SCWG of cellulose. Figure 6 shows how the molar yields of individual gases are affected by changes in temperature. The effect of temperature is similar for all species. Yields for all gases are lowest at 400
Gasification of Cellulose in Supercritical Water
Figure 7. Gas composition at 10 min as a function of cellulose loading.
°C. The influence of gas-forming reactions becomes more noticeable at 500 °C, where the yields are slightly higher. Since the yields for each of the individual species are higher at higher temperatures, the major effect of temperature is to accelerate the rates of the paths char f gases and intermediates f gases, through steam-reforming, pyrolysis, combustion, and hydrogasification. This effect is very important above 500 °C. The CO2 yield at 400 °C is just slightly lower than those at 500 °C, being around 2 mmol/g. A large increase in yield occurs when going from 500 to 600 °C. The yield at 600 °C and 5 min (6.5 mmol/g) is more than twice that at the lower temperatures. After 5 min, the yield of H2 is more than 5 times higher at 600 °C than at 500 °C, reaching 2.2 mmol/g. For CO this yield is about double (3.6 mmol/g), and CH4, which is barely produced at 400 °C, reaches 2.8 mmol/g at 600 °C. Effect of Cellulose Loading. The second parameter we varied to determine its effect on gas production was the concentration of cellulose in water. From an engineering perspective, one desires to process biomass/water mixtures with as high a biomass content as possible. Doing so would reduce the capital and operating costs for a SCWG process. For cellulose, a stoichiometric mixture with water would be 56.25 wt % (if the final products were CO2 and H2). We used 5.0 and 33.3 wt % cellulose mixtures in addition to the 9.0 wt % loading used in the base case; therefore, water was always present in excess. The temperature (500 °C) and water density (0.08 g/cm3) remained at their base case values. Figure 7 shows how the composition of the gas is affected by the cellulose loading. Experiments at low concentrations (5 wt %) seem to strongly favor the formation of CH4, making it the major product at this condition. An increase in concentration to 9.0 wt % significantly reduces the mole fraction of CH4. The CO2 mole fraction increases steadily with cellulose loading. The H2 mole fraction also increases with concentration. This is important information because higher concentrations are an important aspect for the viability of SCWG. CO is only detected at 9.0 wt %, and it constitutes a large portion of the gas. Figure 8 shows the effect of cellulose loading on the yields of C, H, O, and total gas at 10 min. The data show that most of the yields increase with concentration. The total gas yield at 5.0 wt % is about the same as that obtained in the 9.0 wt % base case. At 33.3 % loading, however, almost 50% of the cellulose is gasified, and this is mainly a consequence of the large increase in the yields of C and O. We estimate that the pressure rises about 50 atm and the mixture density rises about 6.0 mmol/cm3 as the 33.3 wt % cellulose sample is gasified. The H yield is highest at 5.0% concentration because of the large CH4 mole fraction, but the uncertainty here is high. The H yield goes through a minimum because of a combination
Energy & Fuels, Vol. 21, No. 6, 2007 3641
Figure 8. C, H, and O yields at 10 min as a function of cellulose loading.
Figure 9. Gas yields at 10 min as a function of cellulose loading.
of two effects: the decrease in the CH4 mole fraction and the increase in H2 mole fraction. Figure 9 shows the molar yields of all gases as a function of cellulose loading. As was the case for the total gas yield, the CO2 yields at 5.0 and 9.0 wt % are similar. The most visible effect of cellulose concentration takes place at 33.3 wt %, where substantial CO2 production (10 mmol/g) occurs. There is no detectable presence of CO at either 5% or 33.3 wt % loading. We expect CO yields to decrease with time (see Figure 2), so providing enough time should consume nearly all the CO at all concentrations studied. The data do not indicate any real difference in yields from 5.0 to 9.0 wt % for H2. At 33.3 wt % loading, however, its yield is about 9 times higher than at 5%, reaching 2.6 mmol/g and making H2 the main gasification product along with CO2. The CH4 yield at 5 wt % loading is 5.3 mmol/g, making CH4 the major product along with CO2. The CH4 yield at 33.3 wt % loading is higher than the CH4 yield at 9.0 wt % loading. Changes in the cellulose loading affect each of the species in the gas phase differently, so the cellulose concentration can be used to change the selectivity to the different products. As concentration increases, the CO2 yield increases, CO goes through a maximum, H2 increases (but mostly at 33.3 wt %), and CH4 goes through a minimum. Effect of Water Density. The effect of water density on cellulose SCWG was evaluated by examining one density lower than the base case (0.05 g/cm3) and one higher (0.18 g/cm3). The base case temperature (500 °C) and biomass loading (9.0 wt %) were retained. Figure 10 shows the effect of water density on the gas product composition at 7.5 min. The data at 0.05 and 0.08 g/cm3 were obtained by interpolation. The H2 mole fraction increases with water density from 3.7% at 0.05 g/cm3 to 23.3% at 0.18 g/cm3. CO is not observed at 0.18 g/cm3. The CO2 mole fraction remains high at all densities and is largely unaffected by changes
3642 Energy & Fuels, Vol. 21, No. 6, 2007
Figure 10. Gas composition at 7.5 min as a function of water density.
Figure 11. C, H, O, and total gas yields at 7.5 min as a function of water density.
Figure 12. Gas yields at 7.5 min as a function of water density.
in water density. Likewise, the CH4 mole fraction is not affected by water density. Figure 11 shows the effect of water density on the C, H, O, and total gas yields. The highest yields were at the highest density. Over 70% of the initial weight of the cellulose was converted to gases after 7.5 min at 0.18 g/cm3, which demonstrates how powerful the water density can be to elevate gas yields. The lowest yields were always at the intermediate density. At 0.05 g/cm3 water density the gas yield was consistently higher than it was at the base case density. These results show that appropriate choice of water density can significantly decrease solid and liquid residues in SCWG or maybe even eliminate it. The water density of 0.18 g/cm3 was the most severe condition used in this work. No reactors lasted longer than 7.5 min, and we estimate the pressure here to be about 480 atm and the mixture density to be about 10.5 mmol/ mL. Figure 12 shows how water density affects gas yields. The CO2 yield at 0.05 g/cm3 is over 3 times the yield at 0.08 g/cm3. When the density increases to 0.18 g/cm3, CO2 increases to 15.2 mmol/g. Thus, CO2 has a minimum yield around 0.08 g/cm3. It is possible that this minimum is a consequence of two opposite effects that promote CO2 formation: an increase in the rates of
Resende et al.
SCWG reactions at high water densities and an increase in the rates of conventional gasification reactions at low water densities. Since both processes produce CO2, this possibility is consistent with the increase in CO2 yields at both high and low water densities. The CO yield stays around 3 mmol/g at both 0.05 and 0.08 g/cm3, and at 0.18 g/cm3, if present, CO is below our detection limits. This trend is consistent with our reaction scheme in the sense that conventional gasification reactions do not consume CO; they only form it. Conventional gasification is possibly favored by lower water densities such as 0.05 g/cm3. As the water density increases, the rate of the SCWG reactions might increase, and these consume CO, in accordance to what is observed in Figure 12. A possible explanation for this behavior follows: higher water density means that more water molecules are available in a given volume. (More cellulose molecules are available as well because the cellulose loading is kept the same.) In terms of chemical kinetics, this higher concentration increases the average number of collisions and should favor the rate of cellulose hydrolysis in SCW. As a consequence, less cellulose should be available for “side reactions” (conventional gasification reactions) such as char formation. By the same argument, a lower concentration of molecules (lower water densities) possibly leads to a lower probability of hydrolysis reactions, which would favor conventional gasification. Neither H2 nor CH4 is affected much by a change in water density from 0.05 to 0.08 g/cm3. At 0.18 g/cm3, the H2 yield reaches 6 mmol/g, the highest yield for H2 obtained in this work. A similar behavior is observed for CH4, with the yield being 5 mmol/g at 0.18 g/cm3. Similar yields for CH4 can also be obtained, as previously seen, by lowering the concentration of biomass. These data suggest that both water gas-shift and methanation might be strongly affected by water density above 0.08 g/cm3. The effects of water density on the water-gas shift reaction have been previously studied. Rice et al.26 report a sharp increase in the reaction rate, but only above 0.18 g/cm3 (in the present work the highest water density is 0.18 g/cm3). Sato et al.27 report the water-gas shift rate is independent of water density over the range 0.06–0.35 g/cm3. Araki et al.28 did not obtain a reasonable trend at 380 °C. The effect of water density on the water-gas shift reaction is therefore unclear, and it seems to depend on the range of density studied. Comparison with Previous Results. Having just presented the first published results for cellulose gasification in supercritical water in a metal-free reactor, we now compare some of these results with previous work on SCWG of cellulose with no added catalyst. This previous work was done in metal reactors (stainless steel), so this comparison may provide some insight into the contribution of catalytic reactions from the metal reactor wall during nominally uncatalyzed SCWG. Table 2 compares some of our results at 500 °C with those of Hao et al.29 at the same temperature. Hao et al. used a water density of 0.07 g/cm3, whereas we used densities of 0.05 and 0.08 g/cm3. We interpolated between the data at 0.05 and 0.08 g/cm3 to estimate the gas yields we would expect from SCWG at 0.07 g/cm3 in quartz tubes. Table 2 shows that the yields of H2, CH4, CO, and CO2 reported by Hao et al. for SCWG in a (26) Rice, S. F.; Steeper, R. R.; Aiken, J. D. J. Phys. Chem. A 1998, 16, 2673–2678. (27) Sato, T.; Kurosawa, S.; Smith, R. L., Jr.; Adschiri, T.; Arai, K. J. Supercrit. Fluids 2004, 1–2, 113–119. (28) Araki, K.; Fujiwara, H.; Sugimoto, K.; Oshima, Y.; Koda, S. J. Chem. Eng. Jpn. 2004, 3, 443–448.
Gasification of Cellulose in Supercritical Water
Energy & Fuels, Vol. 21, No. 6, 2007 3643
Table 2. Gas Yields (mmol/g) from SCWG of Cellulose with No Added Catalyst (500 °C, 20 min) Hao et al.29
this work FW wt % cellulose reactor material CO2 CO H2 CH4 sum (g/cm3)
a
0.05 8 1.8 0.4 1.3 11.5
0.08 9.0% quartz 3.2 1.5 0.8 1.0 6.5
0.07a
0.07 9.1% 316 stainless steel 6 5.5 4 3 18.5
4.8 1.6 0.7 1.1 8.2
Data obtained by interpolation.
Table 3. Gas Yields (mmol/g) from SCWG of Cellulose with No Added Catalyst (400 °C, 15 min) this work
Watanabe et al.22
0.08 9.0% quartz 1.8 1.2 0.2 0 3.2
0.35 4.8% 316 stainless steel 4.1 1.8 0.6 n.d. 6.5
FW (g/cm3) wt % cellulose reactor material CO2 CO H2 CH4 sum a
a
Osada et al.12
a
0.33 5% 316 stainless steel 2.6 1.4 0.8 0.1 4.9
Yields in mmol/g were calculated from data provided in the article.
Table 4. Gas Yields (mmol/g) from SCWG of Cellulose with No Added Catalyst (400 °C, 60 min) FW wt % cellulose reactor material CO2 CO H2 CH4 sum (g/cm3)
a
this work
Minowa et al.32
0.08 9.0 % quartz 3 1.1 0.3 0.3 4.7
0.06 33% stainless steel 2.9 1.4 12 2.4 18.7
a
Yields in mmol/g were calculated from data provided in the article.
stainless steel autoclave with no added catalyst are all higher than the yields we obtained from SCWG in quartz with no added catalyst. The total gas yield was also higher in their experiment. Thus, it appears that a significant portion of the gas yields observed by Hao et al. may be attributable to heterogeneous reactions catalyzed by the reactor wall. We also note that Hao et al. obtained a higher H2 content in the product gas (22 mol %) than we obtained (8 mol %). Tables 3 and 4 compare our results from SCWG at 400 °C in quartz with previous results obtained in stainless steel reactors (29) Hao, X.; Guo, L.; Zhang, X.; Guan, Y. Chem. Eng. J. (Amsterdam, Netherlands) 2005, 1–3, 57–65. (30) Yoshida, T.; Matsumura, Y. Ind. Eng. Chem. Res. 2001, 23, 5469– 5474. (31) Williams, P. T.; Onwudili, J. Energy Fuels 2006, 3, 1259–1265. (32) Minowa, T.; Ogi, T.; Dote, Y.; Yokoyama, S. Renewable Energy 1994, 58, 813–815. (33) Lu, Y. J.; Guo, L. J.; Ji, C. M.; Zhang, X. M.; Hao, X. H.; Yan, Q. H. Int. J. Hydrogen Energy 2006, 7, 822–831.
with no added catalyst. As was the case with Table 2, the results in Tables 3 and 4 show that experiments done in stainless steel led to higher yields of H2, CH4, and CO and a higher mole fraction of H2 in the product gas. This outcome suggests once again that a significant portion of the gases may be formed via heterogeneous, rather than homogeneous, reactions, even though no catalyst was deliberately added to the system. The comparisons in Tables 3 and 4 are not as definitive as that in Table 2 because the literature runs were not done at the same cellulose loading and water density used in the present work. These two parameters can influence the gas yields. Conclusions (1) Results from SCWG of cellulose in quartz reactors differ from those obtained from nominally “uncatalyzed” SCWG in stainless steel reactors. In quartz, the total gas yields are lower and the H2 mole fraction in the gas is also lower. These comparisons indicate that the reactor surface influences both the rate of gas formation and the composition of the gas. This outcome confirms that gases formed in metal reactors arise from both homogeneous and heterogeneous reactions, even in the absence of an added catalyst. Therefore, one needs both the homogeneous and heterogeneous reaction kinetics for SCWG to properly design, scale up, or optimize an SCWG reactor. This outcome also reveals the opportunity for the design or development of heterogeneous catalysts to control the gas composition and yield. (2) CO2 was always the major product, except at the lowest cellulose loadings (5 wt %) where CH4 became the major product. (3) Higher temperatures increased the rate of formation of all of the gas species. It also increased the mole fractions of H2 and CH4 in the gaseous products. (4) Manipulating the cellulose loading (wt %) and water density provides a means to control the selectivity to Hcontaining gases. The relative amounts of H2 and CH4 were strongly influenced by these two process variables. The molar ratio of CH4 to H2 decreased from 18.93 to 1.21 to 0.76 as the cellulose loading increased from 5 to 9.0 to 33.3 wt %, at 500 °C and 0.08 g/cm3 water density. Likewise, this ratio decreased from 3.28 to 1.25 to 0.76 as the water density increased from 0.05 to 0.08 to 0.18 g/cm3, at 500 °C and 9.0 wt % cellulose. (5) In general, higher H2 yields were obtained at the longer reaction times, higher water densities, higher cellulose loadings, and higher temperatures. In short, as the severity of the reaction conditions increased, the H2 yield increased. Acknowledgment. We thank Harald Eberhardt, Master Glassblower at The University of Michigan, for experimental assistance and access to the glass lab. F.L.P.R. gratefully acknowledges the Ph.D. fellowship awarded from the CNPq, an institution of the Brazilian Government committed to promote research and higher education. EF7002206
