Conversion of Biomass Syngas to DME Using a Microchannel Reactor

1722

Ind. Eng. Chem. Res. 2005, 44, 1722-1727

Conversion of Biomass Syngas to DME Using a Microchannel Reactor Jianli Hu,* Yong Wang, Chunshe Cao, Douglas C. Elliott, Don J. Stevens, and James F. White Pacific Northwest National Laboratory, 902 Battelle Boulevard, Richland, Washington 99352

The capability of a microchannel reactor for direct synthesis of dimethyl ether (DME) from biomass syngas was explored. The reactor was operated in conjunction with a hybrid catalyst system consisting of methanol synthesis and dehydration catalysts, and the influence of reaction parameters on syngas conversion was investigated. The activities of different dehydration catalysts were compared under DME synthesis conditions. Reaction temperature and pressure exhibited similar positive effects on DME formation. A catalytic stability test of the hybrid catalyst system was performed for 880 h, during which CO conversion only decreased from 88% to 81%. In the microchannel reactor, the catalyst deactivation rate appeared to be much slower than in a tubular fixed-bed reactor tested for comparison. Test results also indicated that the dehydration reaction rate and the water depletion rate via a water-gas-shift reaction should be compatible to achieve high selectivity to DME. Using the microchannel reactor, it was possible to achieve a space time yield almost 3 times higher than commercially demonstrated performance results, which highlights strong process intensification potential for commercial application. 1. Introduction Biomass feedstocks, such as agriculture and forestry residues, play an important role in developing alternatives to fossil fuels.1 Although there are several methods of generating energy from biomass, gasification, in which a hydrogen-carbon monoxide gas mixture (syngas) is produced, offers several advantages. For example, syngas, like natural gas, can be burned in gas turbines, which are more efficient than steam boilers. Another key feature of syngas is, like petroleum products, it can be converted to useful chemicals, including dimethyl ether (DME). DME can be used as building blocks for synthesizing important chemicals, including dimethyl sulfate, highvalue oxygenated compounds, and lower olefins.2-5 Because of its environmentally benign properties, it can also be used as an aerosol propellant in products such as hair spray and shaving cream. Recently, DME has been suggested as an alternative fuel for diesel engines. Engine performance tests indicate that DME has thermal efficiencies equivalent to traditional diesel fuel. Other advantages of using DME as a diesel replacement include the reduced NOx emissions, near-zero smoke production, and less engine noise.6,7 However, an obstacle to producing DME from biomass syngas, at an economical scale, has been the decentralized nature of biomass operations. Unlike petroleum, coal, and natural gas plants, which are established for central, large-scale applications, biomass feedstocks and gasification systems are widely distributed geographically. It is reported that if conventional DME process technology is adopted, a scale of 2500 t/d would be required for production economically comparable to conventional LPG fuel.8 Because it is difficult to deliver enough biomass to satisfy this criterion using conventional technology, a more compact and efficient portable * To whom correspondence should be addressed. Tel.: (509) 376-0427. Fax: (509) 376-5106. E-mail: [email protected].

process is needed. This new process would operate under the lower limit of industrial process conditions, i.e., 1-40 MPa, but with a high space time yield. The purpose of the research discussed here is to develop such a process capable of converting syngas generated from gasification of dispersed biomass resources. The conventional DME production method is via a methanol dehydration reaction, using acidic catalysts, such as phosphoric-acid modified gamma Al2O3, in fixedbed reactor. The cost of producing DME from methanol is influenced by price and availability, as methanol itself is an expensive chemical feedstock. In contrast, producing DME directly from syngas has many economic and technical advantages over methanol dehydration. Thermodynamically, DME production from syngas is more favorable than from methanol, and thus, in principle, the costs for DME production from syngas should be lower, provided a suitable catalyst can be found. Direct DME synthesis involves many competing reaction pathways:

CO + 2H2 T CH3OH

∆H°327°C ) -24.0 kcal/mol (1)

CO2 + 3H2 T CH3OH + H2O ∆H°327°C ) -14.7 kcal/mol (2) 2 CH3OH T DME + H2O (dehydration) ∆H°327°C ) -5.6 kcal/mol (3) CO + H2O T CO2 + H2 (WGS) ∆H°327°C ) -9.3kcal/mol (4) DME + 2H2 f 2CH4 + H2O (hydrogenolysis)

(5)

Considering the exothermic characteristic of DME synthesis and the scale of production, we explored the feasibility of utilizing a microchannel reactor for DME synthesis, in conjunction with a combined methanol

10.1021/ie0492707 CCC: $30.25 © 2005 American Chemical Society Published on Web 02/18/2005

Ind. Eng. Chem. Res., Vol. 44, No. 6, 2005 1723

synthesis/dehydration catalyst system. It is known that microchannel reactors have the advantage of improved heat and mass transfer, which allow greater process intensification. The challenge is to design microchannel reactors that demonstrate enhanced productivity over conventional reactors but still meet economic requirements. 2. Experimental Section A hybrid catalyst system, consisting of methanol synthesis and dehydration catalysts, was developed to test direct synthesis of biomass syngas to produce DME. The experiments were carried out in a microchannel reactor (316 stainless steel), with the dimensions 5.08 cm × 0.94 cm × 0.15 cm. Three commercial catalysts were used: a methanol synthesis catalyst, F51-8PPT (Kataco Corp.); and two dehydration catalysts, ZSM-5 zeolite with a Si/Al ratio of 30 (Zeolyst International) and acidic Al2O3 (Engelhard Corp.). Solid acid Al2O3 containing 4 wt % fluoride was also used as dehydration catalyst (F-Al2O3) Prior to reaction, the zeolite and acidic catalysts were calcined in air at 500 °C to remove physically absorbed moisture. Both the methanol synthesis catalyst and the dehydration catalyst were crushed and sieved into 70-100 mesh. The hybrid catalyst was prepared by mechanically mixing the two types of catalysts in a transparent vial at a desired ratio and charged in the microchannel reactor. Essentially, this is high pressure down flow fixed-bed type of reactor. The schematic diagram of the reactor system and microchannel reactor assembly were similar to those described in ref 9. To minimize methanation reaction in the stainless steel reactor, silicone-coated stainless steel tubing was used in the high-temperature preheating zone. Experiments were conducted at temperatures from 220 to 320 °C and pressure from 2 to 5 MPa. All the experiments were carried out under isothermal conditions as indicated by the uniform temperature distribution along catalyst bed. The hybrid catalyst (mixture of methanol synthesis and ZSM-5) was reduced with 10% hydrogen in helium in the 220-350 °C temperature range at atmospheric pressure. A mixture of N2/H2 was fed during startup to establish steady-state flow and to heat the reactor to the desired temperature. When the catalyst bed temperature reached the target, premixed syngas at the desired ratio was fed into the reactor. The typical feed composition was CO:H2:CO2:Ar ) 30: 62:4:4. The presence of Ar served as the internal standard for conversion and selectivity calculation purposes. Total feed flow rate was set to achieve the desired gas hourly space velocity (GHSV). The reaction products were analyzed by on-line gas chromatography (HP 5890 GC) equipped with both TCD and FID detectors. GC column used is GS-Q 30 m manufactured by JW Scientific. A temperature program of 5 °C/min to 300 °C was chosen for the analysis. Liquid products were collected in a cold trap at -3 °C and were also analyzed by GC-mass spectrometry. Carbon monoxide conversion and product selectivity were calculated on the basis of feed and product flow rates and carbon balance. 3. Results and Discussion 3.1. Methanol Synthesis and Dehydration Reactions. Methanol synthesis is a thermodynamically

Figure 1. Syngas conversion to methanol: equilibrium CO conversion under different conditions. Equilibrium is calculated for the reaction CO + 5H2 + CO2 ‡2CH3OH + H2O.

Figure 2. Direct syngas conversion to DME: CO equilibrium conversion under different conditions. Equilibrium is calculated for the reaction CO + 5H2 + CO2 T CH3OCH3 + 2H2O.

limited process. As shown in Figure 1, CO conversion decreases as the reaction temperature rises; yet it increases with higher pressure. Calculating the chemical equilibrium shows that the overall methanol yield can be increased, in principle, by combining the methanol synthesis with methanol dehydration. The calculation was done by coupling reactions (1) through (4) together. The combined reaction is expressed as

CO + 5H2 + CO2 f CH3OCH3 + 2H2O Figure 2 presents the results of integrating the two reactions. Synergy in total methanol production is obtained by effectively removing the products from the methanol synthesis reaction, i.e., by minimizing the reverse reaction. Consequently, maximum synergy is obtained close to the equilibrium limit for methanol synthesis where the reverse reaction rate is maximum. A baseline test was conducted with the methanol synthesis catalyst to better understand the performance of the hybrid catalyst system. For this baseline test, methanol synthesis was performed over a commercial Cu-based catalyst at 3.8 MPa and GHSV ) 300015 000 h-1. Steady state was achieved within 12 h from startup. As shown in Figure 3, CO conversion is lower at high GHSV. It becomes clear in Figure 3 that CO conversion starts to level off and approaches equilibrium when GHSV is decreased from 15 000 to 3400 h-1.

1724

Ind. Eng. Chem. Res., Vol. 44, No. 6, 2005

Figure 5. Effect of catalyst ratio on CO conversion (MeOH synthesis catalyst: dehydration catalyst; T ) 260 °C, P ) 3.8 MPa, H2/CO ) 3, GHSV ) 10 000 h-1). Figure 3. Effect of GHSV on methanol synthesis, a thermodynamically limited process (P ) 3.8 MPa, T ) 250 °C, H2/CO ) 2:1).

Figure 6. Effect of reaction temperature on syngas conversion to DME (mixture of methanol synthesis catalyst and H-ZSM-5; P ) 3.8 MPa, GHSV ) 5000 h-1, H2/CO ) 3). Figure 4. Effect of pressure on dehydration of methanol (H-ZSM-5 dehydration catalyst only, T ) 234 °C, GHSV ) 10 600 h-1).

To better understand the performance of the combined processes in the microchannel reactor, the methanol dehydration reaction was carried out independently. It was speculated that the large quantity of water produced in both methanol synthesis and methanol dehydration reactions might retard the methanol dehydration activity. The inhibiting effect may become more severe at elevated pressure, because the desorption of water is suppressed at high pressure, and active sites may be blocked for methanol absorption. Consequently, methanol dehydration was conducted at different pressures. As illustrated in Figure 4, on raising pressure from ambient to 3.8 MPa, conversion of methanol to DME decreases, but not dramatically. An important parameter in the design of a dual catalytic system is the catalyst loading ratio, that is, the methanol dehydration to methanol formation activity. Too high a methanol dehydration activity compared with water-gas shift (WGS) activity leads to a high water production. Results shown in Figure 5 indicate that CO conversion is affected by the catalyst ratio, but product selectivity is not sensitive to the change of catalyst ratio. The optimal catalyst ratio appears to be about 1:1 by weight.

3.2. Effects of Temperature, Pressure, and Residence Time on DME Synthesis. 3.2.1. Effect of Temperature on DME Formation. The effect of temperature on the catalytic activity of methanol synthesis catalyst and ZSM-5 catalyst combined is depicted in Figure 6. To study the effect of temperature, it was necessary to operate the reactor in an isothermal mode. Two thermal couples were installed in catalyst bed and furnace temperature was used to control catalyst bed temperature. The temperature difference between the top and bottom catalyst bed was controlled within (2 °C, indicating excellent heat removal capability of the microchannel reactor. Unlike methanol synthesis alone, where CO conversion decreases with an increase in reaction temperature, Figure 6 clearly shows that temperature has a positive effect on CO conversion in direct syngas conversion to DME. It is possible that higher temperature favors the methanol dehydration reaction. The implication from the result is that microchannel reactors can be operated at higher reaction temperatures than conventional reactors to achieve high space time yield but can still operate in the isothermal regime. At the current GHSV of 5000 h-1, CO conversion of 80% can be achieved but is below the equilibrium limit of DME synthesis (94%). These results indicate that the dehydration reaction may proceed at slower rate than the methanol synthesis reaction. Therefore,

Ind. Eng. Chem. Res., Vol. 44, No. 6, 2005 1725

Figure 7. Effect of pressure and H2/CO ratio on DME formation (mixture of methanol synthesis catalyst and H-ZSM-5; GHSV ) 5000 h-1).

a relatively low GHSV is needed to approach equilibrium conversion. A large amount of CO2 was produced via the WGS reaction. Within the operating temperature window, selectivity to CO2 remained fairly constant. As with methanol synthesis, there is an upper limit on the operating temperature for direct DME synthesis. When the temperature reached 310 °C, a sharp increase in CO2 selectivity was observed, and catalyst deactivation occurred. The deactivation at high temperature was probably due to metal (Cu) sintering; coking may not be the cause of deactivation, because regeneration by oxygen or hydrogen rereduction would not recover any lost activity. In the study of temperature effect on DME formation, very minimum selectivity to methane was observed. Selectivity to methane is believed to form via methanation or hydrogenolysis of DME. As a result of isothermal operation in microchannel reactor, hot spots on the catalyst surface and between catalyst particles are eliminated; therefore, these two side reactions are not favored. 3.2.2. Effect of Pressure on DME Formation. Experiments were conducted at GHSV ) 5000 h-1, using two different feed compositions (H2/CO ) 3:1 and 2:1). Pressure exhibited a positive effect on DME formation. As shown in Figure 7, at low pressure of 1.0 MPa, low CO conversion occurs. A sharp increase in CO conversion is observed when pressure is increased from 1.0 to 3.8 MPa. This is not surprising, because DME formation is a mole-number-reducing reaction. The phenomenon observed in the study of pressure effect appears different from what has been reported in the literature. In a review paper, Wender indicated that, in DME synthesis, CO conversion increased with pressure but started to level off at 2 MPa, beyond which the impact of pressure on CO conversion was not significant.10 Because the source of syngas, especially the type of reactor configuration, and the loading of two functionally independent catalysts are closely interrelated, the performance of DME synthesis in a microchannel reactor could be different from conventional fixed-bed or slurry reactors. Ng et al. compared the difference between a slurry reactor and a fixed-bed reactor and discussed the complexity of reaction variables that influence DME formation.11 3.2.3. Effect of Gas Hourly Space Velocity on DME Synthesis. As indicated in Figure 3, in methanol synthesis conducted under constant temperature of

Figure 8. Effect of GHSV on syngas conversion to DME (mixture of methanol synthesis catalyst and H-ZSM-5; P ) 3.8 MPa, H2/CO ) 2, T ) 286 °C).

Figure 9. Direct syngas conversion to DME: long-term performance test (mixture of methanol synthesis catalyst and HZSM-5; P ) 3.8 MPa, GHSV ) 5000 h-1, H2/CO ) 3, T ) 280 °C).

250 °C and pressure of 3.8 MPa, when decreasing GHSV, CO conversion increases first and begins to approach equilibrium conversion of about 50%. This response is largely because methanol synthesis is thermodynamically limited. In contrast, as Figure 8 illustrates, in direct DME synthesis, conversion increases significantly with a decrease in GHSV. When GHSV is decreased from 20 000 to 5000 h-1, conversion is increased from 31% to 80%. However, while GHSV is changed, selectivity to CO2 remains fairly stable. 3.3. Catalytic Stability. The stability test of the hybrid catalyst in DME synthesis was performed under conditions of 280 °C, 3.8 MPa and GHSV of 5000 h-1. In this test, the CO conversion level was targeted between 80 and 90 mol %, somewhat below the thermodynamic equilibrium conversion (94%). The hybrid catalyst system combining the commercial methanol synthesis catalyst and the ZSM-5 zeolite was tested for approximately 880 h. As shown in Figure 9, initial CO conversion of 88% is reached. As the reaction proceeds, a slow decrease in CO conversion is observed. At the end of the test (TOS ) 880 h), CO conversion drops to 81%. Throughout the entire test, selectivity to CO2 increases slightly, but methane selectivity remains unchanged. From a carbon utilization point of view, the formation of CO2 from the WGS reaction appears to have negative effect on DME yield. However, the WGS reaction is necessary in DME synthesis to keep the water concentration low, so as to enhance the rate of

1726

Ind. Eng. Chem. Res., Vol. 44, No. 6, 2005

Table 1. Dehydration Activity of Different Solid Acid Catalysts under DME Synthesis Conditions (P ) 3.8 MPa, T ) 280 q°C) catalysts conversion selectivity DME MeOH CO2 others total

ZSM-5 F-Al2O3 acidic Al2O3 80.0

79.3

79.7

60.2 10.3 27.0 2.5 100

63.4 10.2 22.7 3.7 100

62.7 7.4 22.3 7.6 100

USY-zeolite no dehydration activity

dehydration reaction. This is especially important for a CO-rich feedstock, as the WGS reaction can also balance the ratio of CO and H2 by depleting CO and forming H2. It is reported that the presence of water inhibits dehydration activity.12 However, in this study, with the use of the hybrid catalyst, deactivation was not significant. The excess water, upon formation in the reaction, was removed via the WGS reaction. The WGS reaction proceeded at faster reaction rates as stated by Li et al.,13 and it reached equilibrium to yield 22% CO2. The WGS reaction depleted the water produced from the dehydration reaction, which may have reduced the interaction between water and methanol catalyst, and therefore retarded catalyst sintering. This analysis is in agreement with simulation results that achieving a strong synergy between methanol synthesis and dehydration reactions requires efficient water removal via the WGS reaction.12 3.4. Effect of Dehydration Catalysts on DME Synthesis. An activity comparison of different dehydration catalysts is presented in Table 1. These experiments were conducted under the same temperature, pressure, contact time, and catalyst ratio. Although the acid strengths of these catalysts are different, Al2O3, F-Al2O3, and ZSM-5 yield the same conversion and selectivity. It was surprising that the USY zeolite (ultra stable Y-zeolite) exhibited the lowest dehydration activity. The use of the Y-zeolite did not result in a synergistic effect. A CO conversion of about 40% was obtained, which was the same as methanol-only synthesis. Low DME selectivity (10%) further confirmed that the loss of dehydration activity was responsible for the low CO conversion. The unexpected phenomenon might be associated with the blocking of acid sites by water. It is known that Y-zeolite contains less acid sites but higher acid strength than H-ZSM-5. It is plausible that water produced from the reaction strongly adsorbed on acid sites, therefore inhibiting the dehydration reaction rate. The implication from the above finding is that direct DME synthesis does not require strong solid acid and the acidity (number of acid sites) of the dehydration catalyst is more important than the acid strength. Many reaction mechanisms have been suggested for methanol dehydration over solid acid catalysts, and there is no consensus on whether Lewis acid-base pair or Bronsted acid-Lewis base pair sites are responsible for dehydration reactions.14-16 In our continued study, the nature of acid sites will be elucidated. 3.5. Process Intensification. Currently, no commercial production data are available for direct DME synthesis. As a reference for this study, Air Products commercial demonstration results obtained from a slurry reactor17 were compared with the hybrid catalyst/ microchannel reactor system. The performance results are listed in Table 2. Because it is very difficult to find exactly the same reaction conditions for two different

Table 2. Performance Comparison of Syngas Conversion to DME in Different Reactor Configurations reactor configurations air products slurry PNNL microchannel reaction conditions T, °C 250 280 P, atm 52 38 -1 GHSV, h 4500 10 238 H2/CO 0.7 2 performance conversion 37 80 CO2 selectivity 32 22 specific activity, gCO/g.h 1.19 3.52 liquid yield, g/(g‚h) 0.79 2.25

type of reactors, we chose the conditions that unfavorable to microchannel reactor. We need to address here is that the comparison results are preliminary, it is intended to show the difference in overall productivity but by no means to compare the advantages of different reactor configurations. The ultimate goal of this study is to develop a compact reaction system effective for syngas conversion to DME. A microchannel reactor should be able to operate at relatively severe conditions to obtain high productivity. A major objective is to obtain a space time yield that may not be achieved with conventional reactors. As shown in Table 2, DME synthesis is conducted in the microchannel reactor at GHSV ) 10 238 h-1, 2.2 times higher than in the commercial slurry reactor. Side-byside comparison reveals excellent performance of microchannel reactor. A specific activity of 3.52 g of CO/(g‚h) and a space time yield of 2.25 g/(g‚h) are obtained, respectively. Even when the microchannel reactor was operated under severe conditions, selectivity to byproducts produced from methanation and pyrolysis were negligible. It is generally believed that back mixing occurs in the slurry reactor, which affects product selectivity. Apparently, back mixing was not present in the microchannel reactor. Enhancement in space time yield was also attributed to improved mass transfer in the microchannel reactor. Compared with the conventional fixed-bed reactor, the (bulk) diffusion length from gas phase to catalyst surface was significantly reduced. Furthermore, unlike the slurry reactor, there was no liquid holdup inside the microchannel reactor. Calculations showed that, in the microchannel reactor, under reaction conditions investigated, the products were present in the gas phase inside reactor. Therefore, the internal diffusion limitation was significantly reduced. In addition, a high linear velocity in the microchannel reactor facilitated the removal of water from the catalyst surface, releasing blocked acid sites and accelerating the dehydration reaction. 4. Conclusions The microchannel reactor was effective for achieving high productivity in direct DME synthsis. The performance of the hybrid bifunctionalized catalyst system was much higher than that in conventional reactors, as a result of improved heat and mass transfer. Heat transfer improvement eliminated hot spots, which are detrimental to catalyst stability. The improved mass transfer capability of the microchannel reactor was attributed to the shortening of bulk diffusion length, minimized back-mixing and increased accessibility from the gas phase to the catalyst surface, which led to an enhanced space time yield.

Ind. Eng. Chem. Res., Vol. 44, No. 6, 2005 1727

Acknowledgment This study was funded by the U.S. Department Energy, Office of Biomass Program, under Contract DEAC06-76RL01830. Literature Cited (1) Franco, C.; Pinto, F.; Gulyurtlu, I.; Cabrita, I. Fuel 2003, 82, 835-842. (2) Shikada, T. K.; Fujimoto, Miyauchi, M.; Tominaga, H. Appl. Catal. 1983, 361. (3) Kaeding, W. W.; Butter, S. A. J. Catal. 1980, 61, 155. (4) Chang, C. D. Catal. Rev. Sci. Eng. 1983, 25, 1. (5) Spivey, J. J. Chem Eng. 1991, 110, 123. (6) Rouhi, A. M. Chem. Eng. News 1995, May 29. (7) Fleisch, T.; McCarthy, C.; Basu, A.; Udovich, C.; Charbonneau, P.; Slodowski, W.; Mikkelsen, S. E.; McCandler, J. SAE Paper 950064; DAE International Congress: Detroit, 1995. (8) Adachi, Y.; Komoto, M.; Watanabe, I.; Ohno, Y.; Fujimoto, K. Fuel 2000, 79, 229-234. (9) Cao C.; Xia G.; Holladay J., Jones E., Wang Y. Appl. Catal., General 2004, 262, 19-29.

(10) Wender, I. Fuel Processing Technol. 1996, 48, 189-297. (11) Ng, K. L.; Chadwick, D.; Toseland, B. A. Chem. Eng. Sci. 1999, 54, 3587-3592. (12) Xu, M.; Lunsford, J. H.; Goodman, D. W. Appl. Catal. 1997, 149, 289-301. (13) Li, J. L.; Zhang, X.; Inui, G. T. Appl. Catal. 1996, 147, 2333. (14) Knozinger, H.; Kochloefl, K.; Meye, W. J. Catal. 1973, 28, 69. (15) Dabrowski, J. E.; Butt, J. B.; Bliss, H. J. Catal. 1970, 18, 297. (16) Bandiera, J.; Naccache, C. Appl. Catal. 1991, 69, 139. (17) Final Technical Report by Air Product, “ Synthesis of Dimethyl Ether and Alternative Fuels in the Liquid Phase from Coal-Derived Synthesis Gas” under DOE contract No. DE-AC2290PC89865, 1993.

Received for review August 11, 2004 Revised manuscript received December 21, 2004 Accepted January 18, 2005 IE0492707