An economic evaluation has been carried out for the concepts considered. Plant sizes of 80, 400, 1000, and 2000 MWth HHV are evaluated, 400 MWth being the base scale. The scale of the conversion system is expected to be an important factor in the overall economic performance. This issue has been studied for BIG/CC systems (Faaij et al. 1998; Larson et al. 1997), showing that the econ­omies of scale of such units can offset the increased costs of biomass transport up to capacities of several hundreds of MWth. The same reasoning holds for the methanol production concepts described here. It should, however, be realized that production facilities of 1000-2000 MWth require very large volumes of feedstock: 200-400 dry tonne/hour or 1.6-3.2 dry Mtonne per year. Biomass availability will be a limitation for most locations for such large-scale production facilities, especially in the shorter term. In the longer term (2010-2030), if biomass pro­duction systems become more commonplace, this can change. Very large scale biomass conversion is not without precedent: various large-scale sugar/ethanol plants in Brazil have a biomass throughput of 1-3 Mtonne of sugarcane per year, while the production season covers less than half a year. Also, large paper and pulp complexes have comparable capacities. The base scale chosen is comparable to the size order studied by Williams et al. (1995) and Katofsky (1993), 370-385 MWth.

The methanol production costs are calculated by dividing the total annual costs of a system by the produced amount of methanol. The total annual costs consist of:

1. Annual investments.

2. Operating and maintenance.

3. Biomass feedstock.

4. Electricity supply/demand (fixed power price).

The total annual investment is calculated by a factored estimation (Peters et al. 1980), based on knowledge of major items of equipment as found in the literature or given by experts. The uncertainty range of such estimates is up to ±30%. The installed investment costs for the separate units are added up. The unit investments depend on the size of the components (which follow from the

Aspen Plus modelling), by scaling from known scales in literature (see Table 2.5), using Equation 2.8:

with R = scaling factor.

Various system components have a maximum size, above which multiple units will be placed in parallel. Hence the influence of economies of scale on the total system costs decreases. This aspect is dealt with by assuming that the base investment costs of multiple units are proportional to the cost of the maximum size: the base investment cost per size becomes constant. The maximum size of the IGT gasifier is subject to discussion, as the pressurised gasifier would logically have a larger potential throughput than the atmospheric BCL.

The total investment costs include auxiliary equipment and installation labour, engineering and contingencies. If only equipment costs, excluding installation, are available, those costs are increased by applying an overall installation factor of 1.86. This value is based on 33% added investment to hardware costs (instrumen­tation and control 5%, buildings 1.5%, grid connections 5%, site preparation 0.5%, civil works 10%, electronics 7%, and piping 4%) and 40% added installation costs to investment (engineering 5%, building interest 10%, project contingency 10%, fees/overheads/profits 10%, and start-up costs 5%) (Faaij et al. 1998).

The annual investment takes into account the technical and economic lifetime of the installation. The interest rate is 10%.

Operational costs (maintenance, labour, consumables, residual streams dis­posal) are taken as a single overall percentage (4%) of the total installed invest­ment (Faaij et al. 1998; Larson et al. 1998). Differences between conversion concepts are not anticipated.

It was assumed that enough biomass will be available at 2 US$/GJ (HHV). This is a reasonable price for Latin and North American conditions. Costs of cultivated energy crops in the Netherlands amount approximately 4 US$/GJ and thinnings 3 US$/GJ (Faaij 1997), and biomass imported from Sweden on a large scale is expected to cost 7 US$/GJ (1998). On the other hand biomass grown on Brazilian plantations could be delivered to local conversion facilities at 1.6—1.7 US$/GJ (Hall et al. 1992; Williams et al. 1995). It has been shown elsewhere that international transport of biomass and bioenergy is feasible against modest costs.

Electricity supplied to or demanded from the grid costs 0.03 US$/kWh. The annual load is 8000 hours.

Results

Results of the economic analysis are given in Figure 2.7. The 400 MWth conversion facilities deliver methanol at 8.6-12 US$/GJ. Considering the 30%

1 Annual GDP deflation up to 1994 is determined from OECD (1996) numbers. Average annual GDP deflation after 1994 is assumed to be 2.5% for the United States, 3.0% for the EU. Cost numbers of Dutch origin are assumed to be dependent on the EU market, therefore EU GDP deflators are used. 12001 = 0.94 US$2001 = 2.204 M2001.

2 Total pretreatment approximately sums up to a base cost of 8.15 MUS$2001 at a base scale of 33.5 tonne wet/hour with an R factor of 0.79.

3 Based on first-generation BIG/CC installations. Faaij et al. (1995) evaluated a 29-MWe BIG/CC installation (input 9.30 kg dry wood/s, produces 10.55 Nm3 fuel gas/s) using vendor quotes. When a range is given, the higher values are used (Faaij et al. 1998). The scale factors stem from Faaij et al. (1998).

4 Two double-screw feeders with rotary valves (Faaij et al. 1995).

5 12.72 MUS$1991 (already includes added investment to hardware) for a 1650 dry tonne per day input BCL gasifier, feeding not included, R is 0.7 (Williams et al. 1995). Stronger effects of scale for atmospheric gasifiers (0.6) were suggested by Faaij et al. (1998). Technical director Mr. Paisley of Battelle Columbus, quoted by Tijmensen (2000), estimates the maximum capacity of a single BCL gasifier train at 2000 dry tonnes/day.

6 29.74 MUS$1991 (includes already added investment to hardware) for a 1650 dry tonne/day input IGT gasifier, R = 0.7 (Williams et al. 1995). Maximum input is 400-MW^ HHV (Tijmensen 2000).

7 Air Separation Unit: Plant investment costs are given by Van Dijk (van Dijk et al. 1995): I = 0.1069C08508 in MUS$1995 installed, C = Capacity in tonne 02/day. The relation is valid for 100 to 2000 tonne 02/day. Williams et al. (1995) assume higher costs for small installations, but with a stronger effect of scale: I = 0.260C0712 in MUS$1991 fob plus an overall installation factor of 1.75 (25% and 40%). Larson et al. (1998) assume lower costs than Van Dijk, but with an even stronger scaling factor than Williams: 27 MUS$1997 installed for an 1100 tonne O2 per day plant and R=0.6. We have applied the first formula (by Van Dijk) here. The production of 99.5% pure O2 using an air separation unit requires 250-350 kWh per tonne O2 (van Dijk et al. 1995; van Ree 1992).

8 High-temperature heat exchangers following the gasifier and (in some concepts) at other locations are modelled as HRSGs, raising steam of 90 bar/520°C. A 39.2-kg steam/s unit costs 6.33 MUS$1997 fob, overall installation factor is 1.84 (Larson et al. 1998).

9 Tijmensen (2000) assumes the fob price for hot gas cleaning equipment to be 30 MUS$2000 for a 400-MWth HHV input. This equals 74.1 m3/s from a BCL gasifier (T = 863°C, 1.2 bar). There is no effect of scaling.

10Katofsky (1993) assumes compressors to cost 700 US$1993 per required kWmech, with an installation factor of 2.1. The relation used here stems from the compressor manufacturer Sulzer quoted by (2000). At the indicated base scale, total installed costs are about 15% higher than assumed by Katofsky. Multiple compressors, for synthesis gas, recycle streams, or hydrogen, are considered as separate units. The overall installation factor is taken 1.72 because the base unit matches a 400-MWth plant rather than a 70-MWth plant.

investments for steam reformers vary from 16.9 MUS$^3, for a throughput of 5800 kmol meth — ane/hour with an overall installation factor of 2.1 (Katofsky 1993) to 7867 k$^5 for a 6.2 kg methane/s (1390 kmol/hour), overall installation factor is 2.3 (van Dijk et al. 1995). These values suggest a strong effect of scaling R = 0.51, while Katofsky uses a modest R = 0.7. Here, we use the values of Van Dijk in combination with an R factor of 0.6. The total amount of moles determines the volume and thus the price of the reactor.

12 Autothermal reforming could be 50% cheaper than steam reforming (Katofsky 1993), although higher costs are found as well (Oonk et al. 1997).

investment for shift reactors vary from 9.02 MU$^5 for an 8819 kmol CO+H/hr reactor, and an overall installation factor is 1.81 (Williams et al. 1995) to 30 MUS$^ installed for a 350000 Nm3/hr CO+H2/hr (15625 kmol/hr) reactor (Hendriks 1994). Williams assumes an R = 0.65, but comparison of the values suggest only a weak influence of scale (R = 0.94). Here, we use the the values from Hendriks, with R set at 0.85. A dual shift is costed as a shift of twice the capacity.

14Costs for CO2 removal through Selexol amounts 14.3 MUS$^3 fob (overall installation factor is 1.87) for an 810 kmol CO2/hr unit, R = 0.7 (Katofsky 1993) up to 44 MUS$^ installed for a 9909 kmol CO2/hour unit (Hendriks 1994). The value from Hendriks is assumed to be right, since his research into CO2 removal is comprehensive.

15Van Dijk et al. (1995) estimate that a methanol reactor for a 2.1 ktonne methanol per day plant costs 4433 kUS$1995 (fob) or 9526 kUS$1995 installed (overall installation factor is 2.1). The total plant investment in their study is 138 MUS$1995, or 150 MUS$2001. Katofsky (1993) estimates the costs for a 1056 tonne methanol/day plant to be 50 MUS$1995 fob, this excludes the generation and altering of synthesis gas, but includes make-up and recycle compression and refining tower. A 1000 tpd plant costs about 160 MUS$2001, and a 2000 tpd plant 200 MUS$2001, which suggests a total plant scale factor of 0.3 (Hamelinck et al. 2001). These values come near the ones mentioned by Katofsky. This implies that the values given by Van Dijk are too optimistic and should be altered by a factor 1.33. It is therefore assumed that the base investment for the methanol reactor only is 7 MUS$2001, the installation factor is 2.1. The influence of scale on reactor price is not assumed to be as strong as for the complete plant: 0.6.

installed costs for a 456 tonne per day liquid-phase methanol unit, are 29 MU$^7, excluding generation and altering of synthesis gas, but including make-up and recycle compression, and refining tower. R = 0.72 (Tijm et al. 1997). Corrected for scale and inflation this value is about half the cost of the conventional unit by Katofsky and the corrected costs of Van Dijk. It is therefore assumed that the price of a liquid-phase methanol reactor is 3.5 MUS$2001 for a 2.1 ktonne per day plant, installation factor is 2.1.

17Cost number for methanol separation and refining is taken from Van Dijk, increased with 33% as described in note 15.

18For indication: A complete combined cycle amounts to about 830 US$^7 per installed kWe. Quoted from Solantausta et al. 1996 by Oonk et al. 1997.

19Scaled on gas turbine size.

system consists of water and steam system, steam turbine, condenser and cooling. Scaled on steam turbine size.

21 Expansion turbine costs are assumed to be the same as steam turbine costs (without steam system). 22Overall installation factor. Includes auxiliary equipment and installation labor, engineering and contingencies. Unless other values are given by literature, the overall installation factor is set 1.86 for a 70-MWth scale (Faaij et al. 1998). This value is based on 33% added investment to hardware costs (instrumentation and control 5%, buildings 1.5%, grid connections 5%, site preparation 0.5%, civil works 10%, electronics 7%, and piping 4%) and 40% added installation costs to investment (engineering 5%, building interest 10%, project contingency 10%, fees/overheads/profits 10%, and start-up costs 5%). For larger scales, the added investments to hardware decreases slightly.

23 Maximum sizes from Tijmensen (2000).

uncertainty range, one should be careful in ranking the concepts. Methanol 4 and 6 perform somewhat better than the other concepts due to an advantageous combination of lower investment costs and higher efficiency. The lowest methanol production price is found for concepts using the BCL gasifier, having lower investment costs. The combination of an expensive oxygen fired-IGT gasifier

with a combined cycle seems generally unfavorable, since the efficiency gain is small compared to the high investment.

Investment redemption accounts for 42-76% of the annual costs and is influ­enced by the unit investment costs, the interest rate and the plant scale. The build­up of the total investment for all concepts is depicted in Figure 2.8. It can be seen that the costs for the gasification system (including oxygen production), synthesis gas processing and power generation generally make up the larger part of the investment. For autothermal reforming higher investment costs (Oonk et al. 1997) would increase the methanol price from considered concepts by about 1.5 US$/GJ. Developments in gasification and reforming technology are impor­tant to decrease the investments. On the longer term, capital costs may reduce due to technological learning: a combination of lower specific component costs and overall learning. A third plant built may be 15% cheaper leading to an 8-15% product cost reduction.

The interest rate has a large influence on the methanol production costs. At a rate of 5% methanol production costs decrease with about 20% to 7.2-9.0 US$/GJ. At a high-interest rate (15%), methanol production costs become 9.9-14 US$/GJ. Going to 1000 and 2000 MWth scales, the methanol production costs reach cost levels as low as 7.1-9.5 US$/GJ.

Feedstock costs account for 36-62% of the final product costs for the men­tioned technologies. If a biomass price of 1.7 US$/GJ could be realized (a realistic price for, e. g., Brazil), methanol production costs would become 8.0-11 US$/GJ for 400 MWth concepts. On the other hand, when biomass costs increase to 3 US$/GJ (short term Western Europe), the production cost of methanol will increase to 10-16 US$/GJ.

FIGURE 2.9 Optimistic view scenario. Different cost reductions are foreseeable: (1) biomass costs 1.7 US$/GJ instead of 2 US$/GJ, (2) technological learning reduces capital investment by 15% and (3) application of large scale (2000 MWth) reduces unit investment costs.

If the electricity can be sold as green power, including a carbon neutral premium, the fuel production costs for power coproducing concepts drops, where the green premium essentially pays a large part of the fuel production costs. A power price of 0.08 US$/GJ would decrease methanol costs to -0.6-9.5 US$/GJ. Of course the decrease is the strongest for concepts producing more electricity. A green electricity scenario, however, may be a typical western Euro­pean scenario. As such it is unlikely that it can be realized concurrent with biomass available at 1.7 US$/GJ.

In the long term, different cost reductions are possible concurrently (Tij — mensen 2000). Biomass could be widely available at 1.7 US$/GJ, capital costs for a third plant built are 15% lower, and the large (2000 MWth) plants profit from economies of scale. These reductions are depicted in Figure 2.9: methanol concepts produce between 6.1-7.4 US$/GJ. The influence of capital redemption on the annual costs has strongly reduced and the fuel costs of the different concepts lie closer together.

Previous analyses on short-term methanol production by Katofsky (1993) and Williams et al. (372 MWHHV, 3.4 US$/GJHHV feedstock, 0.07 US$/kWhe (Williams 1995; Williams et al. 1995)) yielded similar energy efficiencies (54-61% by HHV), but significantly higher methanol production costs: 14-17 US$/GJHHV. The largest difference is in the higher capital costs: higher TCI and higher annuity give 25-50% higher annual capital costs. The ADL/GAVE study (Arthur D. Little 1999) reports 13 US$/GJ methanol (feed 2 US$/GJ, 433 MW input) largely using input parameters from Katofsky. Komiyama et al. (2001) instead give much lower costs than presented here: 5 US$/GJHHV for methanol at 530 MWHHV biomass
input. However, in that study, process efficiencies and biomass cost are not given and a significant amount of energy is added as LPG.

In these long-term scenarios, methanol produced from biomass costs consid­erably less than methanol at the current market, which is priced about 10 US$/GJ (Methanex 2001). For application as automotive fuel, comparison with gasoline and diesel is relevant. Their production costs vary strongly depending on crude oil prices, but for an indication: 2003 gasoline market prices were about 7 US$/GJ at oil prices of 25-30 US$/bbl (BP 2004). DOE/EIA projects the world oil price in 2013 to amount between 17 and 34 US$, crude oil prices may decline as new deepwater oil fields are brought into production in the Gulf of Mexico and West Africa, new oil sands production is initiated in Canada, and OPEC and Russia expand production capacity (DOE/EIA 2005). In 2004 the average oil price was some 35 US$/bbl and currently even higher prices of about 50 US$/bbl are paid.

CONCLUSIONS

Methanol can be produced from wood via gasification. Technically, all necessary reactors exist and the feasibility of the process has been proven in practice. Many configurations are possible, of which several have been discussed in this chapter. The configurations incorporated improved or new technologies for gas processing and synthesis and were selected on potential low cost or high-energy efficiency. Some configurations explicitly coproduced power to exploit the high efficiencies of once-through conversion. The overall HHV energy efficiencies remain around 55%. Accounting for the lower energy quality of fuel compared to electricity, once-through concepts perform better than the concepts aiming at fuel only production. Also hot gas cleaning generally shows a better performance. Some of the technologies considered in this chapter are not yet fully proven/commer — cially available. Several units may be realized with higher efficiencies than con­sidered here. For example, new catalysts and carrier liquids could improve liquid — phase methanol single-pass efficiency. At larger scales, conversion and power systems (especially the combined cycle) may have higher efficiencies, but this has not been researched in depth.

The methanol production costs are calculated by dividing the total annual costs of a system by the produced amount of methanol. Unit sizes, resulting from the plant modelling, are used to calculate the total installed capital of methanol plants; larger units benefit from cost advantages. Assuming biomass is available at 2 US$/GJ, a 400 MWth input system can produce methanol at 9-12 US$/GJ, slightly above the current production from natural gas prices. The outcomes for the various system types are rather comparable, although concepts focussing on optimized fuel production with little or no electricity coproduction perform some­what better.

The methanol production cost consists of about 50% of capital redemption, of which the bulk is in the gasification and oxygen system, synthesis gas proc­essing and power generation units. Further work should give more insight into investment costs for these units and their dependence to scale. The maximum possible scale of particularly the pressurized gasifier gives rise to discussion. The operation and maintenance costs are taken as a percentage of the total investment, but may depend on plant complexity as well. Long-term (2020) cost reductions mainly reside in slightly lower biomass costs, technological learning, and application of large scales (2000 MWth). This could bring the methanol production costs to about 7 US$/GJ, which is in the range of gasoline/diesel.

Methanol from biomass could become a major alternative for the transport sector in a world constrained by greenhouse gas emission limits and high oil prices.

Genetic differences in chemical composition among alfalfa plant introductions, varieties, and individual genotypes have been reported. Leaf and stem CP differed among a group of 61 plant introductions, although the ranges were small, from 272 to 295 and 88 to 99 g CP kg-1 DM, respectively (Jung et al., 1997). Leaf NDF concentration (235 g kg-1 DM) did not differ significantly among these plant introductions, but stem NDF ranged from 636 to 670 g NDF kg-1 DM. Similar variation was observed among a group of five commercial alfalfa varieties with CP and NDF differences being noted for leaves and stems, as well as whole herbage (Sheaffer et al., 2000). Differences in stem cell wall concentration and composition were observed among a set of four alfalfa genotypes selected for divergence in whole herbage ADL and in vitro ruminal DM disappearance (IVDMD) (Jung et al., 1994) and a group of three genotypes selected for divergent IVDMD (Jung and Engels, 2002). More recently, alfalfa genotypes selected for divergent cell wall Klason lignin, cellulose, and xylan were shown to differ genetically for these cell wall components when grown across a series of envi­ronments (Lamb and Jung, 2004). While the reported genetic variation among alfalfa germplasm sources is not large, the potential for modifying cell wall composition has not been seriously explored, because recurrent selection for these traits has not been done.

Significant genotype x environment (G x E) interactions have generally not been observed for chemical composition of alfalfa varieties. Among 61 plant introductions, no measures of cell wall concentration or composition were found to have significant G x E interactions for leaf or stem material (Jung et al., 1997). Only differences in magnitude, not rank, for composition due to G x E interactions were noted by Sheaffer et al. (2000) among five alfalfa varieties. These results mirror the conclusion of Buxton and Casler (1993) that forage quality traits generally have small G x E interaction effects compared to the impact on yield. However, in recent work with alfalfa clones selected for specific cell wall traits, it was found that G x E interactions were significant among plants selected for low and high pectin and xylan concentrations, whereas no G x E interactions were noted among clones selected for Klason lignin or cellulose (Lamb and Jung, 2004).

Nonmetallic materials that degrade when in contact with E85 include natural rubber, polyurethane, cork gasket material, leather, polyvinyl chloride (PVC), polyamides, methyl-methacrylate plastics, and certain thermo and thermoset plastics.

This author has had much experience using E85 in a variety of vehicles with plastic fuel tanks with no noticeable negative consequences. The types of vehicle tanks tested include late model automobiles and light-duty trucks, snowmobiles, small engines, and many plastic fuel delivery tanks. Many of these tanks are made of thermo/thermoset plastics, so this appears not be a major issue for vehicles.

Older vehicles may still use rubber, polyurethane or cork gaskets and O-rings for sealing fuel delivery systems; fortunately, most late model vehicles (vehicles produced after the mid-1990s) no longer use these materials in favor of more advanced sealants.

Many of the other sensitive materials are not used in areas where they might come into contact with the fuel; however, care should be taken to ensure that fuel spillage is cleaned from leather or plastic interior surfaces of the vehicles.

Nonmetallic materials that are resistant to E85 degradation include nonme­tallic thermoset reinforced fiberglass, thermoplastic piping, thermoset reinforced fiberglass tanks, Buna-N, Neoprene rubber, polypropylene, nitrile, Viton, and Teflon. All of these materials may be used with E85. Furthermore, most modern vehicles already use these materials for gaskets and O-rings as they offer superior leak resistance. For example, most automakers now use Viton O-rings to seal their fuel injectors.

Vehicle Fuel Pumps

During the mid-1990s, many gasoline fuel pumps suffered high failure rates when delivering E85. Early on, the lower lubricity of ethanol was blamed for these failures. Later, it became clear that the much higher electrical conductivity (eth­anol is about 135,000 times more conductive than gasoline) was at least partly to blame. These problems have been addressed by the automakers and premature failures are no longer a problem. These “hardened” pumps are now standard on many vehicles that are not specifically rated for E85 due to their superior perfor­mance and reduced failure rates.

The third technique for enzyme immobilization is employing micellar polymer Nafion® for enzyme entrapment within the pore structure of the membrane, as shown in Figure 12.5. However, commercial Nafion® has not been successful at immobilizing enzymes at the surface of biofuel cell electrodes because Nafion® forms an acidic membrane that decreases the lifetime and activity of the enzyme. Researchers have been successful in maintaining the activity of glucose oxidase enzymes immobilized in Nafion® by diluting the Nafion® suspension [20];

Enzyme casting layer

I

FIGURE 12.4 Enzyme immobilization by sandwich technique.

however, this approach did not form stable and uniform films. The most recent method employed by Karyakina and coworkers was to neutralize the Nafion® casting solution and dilute the solution to a lesser degree in ethanol; however, both of these approaches have problems with maintaining activity of enzymes for extended times. As the pH environment in the solution around the Nafion® membrane decreases, protons will exchange back into the membrane and re­acidify the membrane [20].

Excess heat generated from gasification can be captured and used for process heat in any application. Since there will be no edible residuals from gasification, fish feed will have to be produced in another manner. To this end, we will conduct production and use trials with feeds ranging from alfalfa leaf, small grains, soy, corn, legume lawn clippings, algae (spirulina), and duckweed. The feed ration does not need to be complex, only complete. Field crops will be irrigated with fish effluent during the summer in a controlled test. The control will be plain water and yields will be noted and recorded. All other parameters will be constant.

Compost-Based Aquaponic Greenhouses

As vegetables are grown and sold, there will be a certain amount of waste plant parts — roots, stems, and trimmings — which will be composted along with grass clippings, leaves, and other materials the project has available, including the ashes from the outdoor furnaces. Ashes are a rich source of boron — an element lacking in most northeastern soils. The project will use the compost in a separate growing system that will allow for the growth of root crops in pure compost, and irrigated with fish tank water.

Through gasification solid biomass is converted into synthesis gas. The funda­mentals have extensively been described by, among others, Katofsky (1993). Basically, biomass is converted to a mixture of CO, CO2, H2O, H2, and light hydrocarbons, the mutual ratios depending on the type of biomass, the gasifier type, temperature and pressure, and the use of air, oxygen, and steam.

Many gasification methods are available for synthesis gas production. Based on throughput, cost, complexity, and efficiency issues, only circulated fluidized bed gasifiers are suitable for large-scale synthesis gas production. Direct gasifi­cation with air results in nitrogen dilution, which in turn strongly increases downstream equipment size. This eliminates the TPS (Termiska Processer AB) and Enviropower gasifiers, which are both direct air blown. The MTCI (Manu­facturing and Technology Conversion International, affiliate of Thermochem, Inc.) gasifier is indirectly fired, but produces a very wet gas and the net carbon conversion is low. Two gasifiers are selected for the present analysis: the IGT (Institute of Gas Technology) pressurized direct oxygen fired gasifier and the BCL (Battelle Columbus) atmospheric indirectly fired gasifier. The IGT gasifier can also be operated in a maximum hydrogen mode by increasing the steam input. Both gasifiers produce medium calorific gas, undiluted by atmospheric nitrogen, and represent a very broad range for the H2:CO ratio of the raw synthesis gas.

The yeast Saccharomyces cerevisiae is specialized for fermentation, with approx­imately 45% of cellular proteins devoted to glycolysis and ethanol fermentation

[18]. Glucose and maltose are fermented to ethanol by S. cerevisiae via the same fermentation pathway (Figure 4.3) [19] used to make beverage alcohol. In glyc­olysis, glucose is converted through a series of reactions to pyruvate, and energy is extracted in the form of four ATP molecules. Then, pyruvate is converted to ethanol in a two-step reaction; pyruvate is decarboxylated to form the more reactive acetaldehyde, which is reduced to ethanol. The second part of the fer­mentation pathway reoxidizes NADH to NAD+ and thus serves to recover the reducing equivalents that were consumed in the conversion of glucose to pyruvate.

For each glucose fermented, two ethanol and two CO2 molecules are produced (Table 4.1, Figure 4.3). The theoretical mass yield is only 0.51 g of ethanol per g of fermented glucose. The actual yield is closer to 90-95% of 0.51 g because some glucose is converted to cell mass and side-products such as glycerol, citric acid cycle intermediates, and higher alcohols. Contaminating microorganisms can also lower the yield by converting glucose to other fermentation products such

as acetic, lactic, and succinic acids. Because ethanol is used as a fuel, it is also appropriate to consider ethanol yield on an energy basis. The thermodynamic yield can be calculated by comparing the heats of combustion for the products and reactants (Table 4.1). By this measure, converting glucose to ethanol has an amazing theoretical yield of 98-99%, which means that the yeast actually gains little energy benefit from fermenting glucose to ethanol. In other words, ethanol fermentation is an excellent process for generating fuel, because most of the energy from glucose is retained in the fermentation product.

Yeast are ideally suited for use in the fuel ethanol industry. Fermentations run 360 days a year, in tanks containing thousands of gallons of beer, with no pH adjustments and only approximate temperature control (reactors are cooled with well water). As a consequence of the absence of pH control and the pro­duction of CO2, the pH drops steadily during the fermentation and ends up below 4.0. Furthermore, the yeast withstand extreme environmental stresses including high osmolality (beginning solids of 25-30% or higher) and high ethanol con­centrations (final concentrations of 12-18% vol.), as well as organic acids pro­duced by contaminating bacteria. The constant contamination of the fermentation is a consequence of the need to run the process in an “open system”— non — aseptically — because the fermentation volumes are quite large and the selling profit margin for ethanol is very low. Fortunately, most bacterial contaminants do not grow below pH 4, and the ability of yeast to do so provides a natural method of suppressing the growth of these contaminants.

Environmental stresses are additive and often synergistic in nature, which means that a combination of many minor stresses, from the perspective of the yeast, equals a single large stress. For example, yeast have reduced tolerance to ethanol at higher temperatures and reduced tolerance to organic acids at lower pH. Despite all of these challenges, S. cerevisiae produces ethanol at rates in excess of 3 g l-1 h-1 and at yields close to 95% of the theoretical maximum. Efforts in the yeast research field are directed at developing strains that produce less glycerol, grow at slightly elevated temperatures (38°C), and withstand even higher ethanol concentrations.

Corn steep water (CSW) is a by-product of the corn wet-milling industry and contains large amounts of substances derived from the fermentative conversion of carbohydrates, proteins, and lipids during corn steeping. Currently, CSW is evaporated to 50% solids and marketed primarily as an economical livestock feed supplement in the cattle industry. CSW is a rich complement of important nutri­ents such as nitrogen, amino acids, vitamins, and minerals and was proposed to be a good substitute for yeast extract (Hull et al., 1996). This finding is important as it impacts the economics of butanol production from corn.

An economic analysis performed by Qureshi and Blaschek (2000a), demon­strated that the fermentation substrate was one of the most important factors that influenced the price of butanol. Development of a cost-effective biomass-to- butanol process can only be commercially viable if cheaper commercial substrate such as liquefied corn starch and CSW can be used in combination with toxic product removal by gas stripping (Ezeji et al., 2005). It is interesting to note that C. beijerinckii BA101 when grown on liquefied corn starch-CSW medium pro­duced levels of acetone-butanol equal to or higher than the levels produced when grown on glucose-based yeast extract medium (Ezeji et al., 2005). The fermen­tation time for liquefied corn starch — and saccharified liquefied corn starch-CSW media were 120 and 78 h, respectively, while the fermentation time for glucose — based yeast extract medium was 68 h. The presence of sodium metabisulfite (Na2S2O5; a preservative) in the liquefied starch and CSW was found to result in inhibition of C. beijerinckii BA101 and also may have affected the secretion of amylolytic enzymes by the culture, which is necessary for efficient hydrolysis and utilization of starch and oligosaccharides. However, it appears that the use of CSW has a great potential for the bioconversion of corn to acetone-butanol. The presence of Na2S2O5 in the CSW may be a major problem in a long-term fermentation by C. beijerinckii BA101. During a long-term fermentation using CSW and C. beijerinckii BA101, removal of Na2S2O5 from CSW by oxidation is recommended (Ezeji et al., 2005).

The following is a brief treatment of the thermodynamics governing the methanol oxidation reaction of a DMFC. Also, the impact of surface kinetics on the practical efficiency of the cell are presented. Some intriguing reports suggesting a new general direction for CO-tolerant catalyst development are cited [14,15,16].

Thermodynamic Optimum

When an organic fuel is used, essentially as a hydrogen source in a fuel cell, the expectation is that the fuel will be completely oxidized to carbon dioxide. For methanol, this is summarized thermodynamically [17] in terms of the reduction potentials as

CO2 + 6H++ 6e ^ CH3OHW+ H2O E0 = 0.016V (9.1)

While methanol is oxidized at the anode, oxygen is reduced at the cathode: O2 + 4H ++ 4e ^ 2H2O E0 = 1.229V (9.2)

The net cell reaction is

CH3OHW + 1.5O2 ^ 2H2O + CO2 (9.3)

where the standard cell potential (electromotive force, emf) is E0e„ = 1.229 — 0.016 = 1.213 V For a six-electron process (n = 6), the standard free energy is AG0 = — nFE0eU = -702.2 kJ/mol for methanol. With a molecular mass, M, of 0.03204 kg/mol, the theoretical specific energy for methanol is W = — AG0/(M x 3600s/hr) = 6.088kWh/kg; because the density of methanol is 0.7914 kg/l, this corresponds to an energy density of 4.818kWh/l. The standard enthalpy [17], AH0 = -726kJ/mol, is similar to AG0, consistent with a small entropy term.

Formally, the complete oxidation of methanol can be viewed thermodynam­ically as a series of two-electron/two-proton oxidation steps. The reduction poten­tials for this sequence in acid are given as [18]:

HCHO(aq) + 2H + + 2e ^ CH3OHw E0 = 0.232V (9.4)

HCOOH(aq) + 2H + + 2e ^ HCHO(aq) + H2O E0 = 0.034F (9.5)

H2CO3(aq) + 2H+ + 2e ^ HCOOH(aq) + H2O E0 = -0.166F (9.6)

where the oxidation of formic acid is reported to carbonic acid, consistent with the solubility of carbon dioxide and its equilibrium with carbonate [19].

CO2(g) ^ CO2W Kco2 = 0.034 (9.7)

CO2(aq) + H2O ^ HCO — + H+aq) pK = 6.36 (9.8)

The acidity constants for carbonic acid are pKa1 = 6.352 and pKa2 = 10.329; for formic acid, pKa = 3.745. Note that these reaction steps embed information about the complexities of the solution chemistry in the fuel cell as reaction products build and local pH changes. Note also, that reactions are reported in acid because practical DMFCs are usually run under acidic conditions. Under basic conditions, the formation of insoluble carbonates dramatically complicates the design of plant and limits applicability as electrolytes must be replaced as carbonate levels build.

For species in solution, the standard potentials (reactions 4 to 6) are such that thermodynamically, the oxidation of methanol proceeds cleanly and sequentially from alcohol to aldehyde to acid to CO2/carbonic acid with approximately 200 mV separating each successive two proton/two electron transfer. The specific energy and energy density of methanol are high. Thus, thermodynamically, the expectation is that methanol is an excellent fuel for a direct reformation fuel cell. However, the thermodynamics do not capture the complexity of the surface reactions that dictate the fate of methanol in a direct reformation fuel cell.

Realities of Surface Kinetics

The kinetic limitations of DMFCs have been well reviewed in detail from several different perspectives in recent years [17,20,21]; an early and thorough review is provided by Parsons and VanderNoot [22]. For effective utilization of methanol as a fuel, the catalyst must provide a good surface for adsorption of methanol and its sequential breakdown to carbon dioxide/carbonate through loss of paired protons and electrons. Under acidic conditions, this has largely restricted practical catalysts to platinum and its alloys and bimetallics. Methanol will adsorb to platinum and platinum serves as an excellent electron transfer catalyst. The difficulty is that platinum passivates as carbon monoxide by­product accumulates and adsorbs to the platinum surface. To oxidize carbon monoxide to carbon dioxide/carbonic acid, oxygenated species such as water must adsorb to the catalyst surface. Because platinum is not strongly

CO,

SCHEME 9.1 Reaction pathways for methanol oxidation.

hydrophilic, platinum bimetallics and alloys formed with more hydrophilic metals such as ruthenium are typically used to facilitate CO oxidation.

Consider the mechanistic constraints for oxidation of methanol. As in equa­tion 1, the complete oxidation of methanol to carbon dioxide proceeds by a six — proton, six-electron process. The mechanism presented in Scheme 9.1 outlines the basic route by which methanol is fully oxidized. The loss of paired protons and electrons is noted for each step. To account for all six electrons, recognize that the adsorption of water to the catalyst surface also generates an electron and proton. For a catalyst metal site, M,

M + H2O ^ M — OH + H + + e

Following the notation from Ref. [21], methanol first adsorbs to liberate one electron and one proton.

CH3OH + Pt ^ Pt — CH2OHads + H+ + e (9.10)

This is followed by two steps to form the formyl intermediate, — CHO.

Pt — CH2OH„* + Pt ^ Pt2CHOH + H+ + e

Pt2CHOH ^ Pt + Pt — CHO + H+ + e

On clean platinum surfaces, these oxidations proceed smoothly to provide two electrons and two protons. Consider Scheme 9.1. The weakly adsorbed -CHO is a point at which the oxidation mechanism breaks into two paths. One path yields adsorbed CO and the other adsorbed COOH. Adsorbed COOH is generated by reaction of — CHO and an adjacent M — OH to yield one proton and one electron and form weakly adsorbed — COOH. Adsorbed CO is generated by the direct oxidation of — CHO by one proton and one electron to form strongly adsorbed CO. Basic kinetic arguments would favor the strongly adsorbed CO over the weakly adsorbed — COOH because first, the oxidation of — CHO to — CO is direct and does not require an adjacent second species, M — OH, and second, because — CO is strongly bound and — COOH is weakly bound.

It should be pointed out that there is an alternative branch point in the oxidation process in which adsorbed — CHOH undergoes a one-electron and a one-proton oxidation to form adsorbed — COH.

Pt2CHOH + Pt ^ Pt3COH + H+ + e

The adsorbed — COH can then either undergo one-proton/one-electron oxida­tion to adsorbed — CO or react with an adjacent M — OH to form HCOOH in solution. Neither process leads to the efficient oxidation to carbon dioxide/car — bonic acid.

To the extent the platinum surface is passivated by CO, the reaction is terminated. Thus, the design of a system for the efficient and complete oxidation of methanol can be approached in two ways.

The first approach is to circumvent the formation of adsorbed CO by favoring the formation of — COOH. Experimentally, this is done by enhancing the proba­bility that — CHO is adjacent to an oxygen source, M — OH, by using bimetallics and alloys of platinum where M is more hydrophilic than platinum. There are questions of stability and cost associated with these catalysts although they have been shown to enhance conversion efficiency. But, based on the relative strengths of the adsorbates — CO and — COOH and the need for an additional catalyst site (M — OH), this approach poses some challenges.

The second approach is to consider why — CO is so difficult to oxidize; that is, why does CO adsorb so strongly. Thermodynamically, the oxidation of CO to CO2 in solution occurs at low potential [18].

CO2 + 2H+ + 2e ^ CO + H2O (1) in E0 = -0.106F (9.14)

But, the oxidation of CO on platinum in acidic solution occurs 600 to 700 mV positive of this value; Pt-Ru alloys are shown to oxidize CO at 200 to 300 mV lower overpotential than Pt [23]. The oxidation of adsorbed CO is strongly disfavored. There are two ways to think about overcoming this large overpotential. One is to design better catalysts. One common approach has been through the bifunctional mechanism where the bimetallic catalyst is designed to place Pt — CO adjacent to an oxygen source through M — OH. The other approach would rely on a paradigm shift in how the oxidation of -CO is viewed at a more fundamental level; better understanding could lead to better catalysts [14,15,16].

The above discussion is provided in a very general manner. Many factors significantly impact the catalytic efficiency of the conversion of methanol to carbon dioxide/carbonic acid. This includes surface structure, catalyst size, and catalyst crystal face as well as the history of the cell, the current coverage of CO, the pH, and the time since the start of the cell.

As we have mentioned above, one difficulty to be overcome for the practical and extensive use of biomass-derived ethanol as a hydrogen source to fuel-cell systems is supplying the energy needed to distill and/or vaporize the H2O/ethanol mix­tures, and that related to the endothermicity of the steam reforming reaction.

Recently, Dumesic and coworkers have shown that methanol, ethylene glycol, glycerol, and sorbitol can be reformed in the aqueous phase to H2 and CO2 at temperatures near 500 K and at pressures between 15-50 bar [57-59]. On the basis of these studies, the reforming of ethanol in the aqueous phase appears as a new approach to be considered for the production of H2 from ethanol reforma­tion. This process would have several advantages over steam reforming: i) it does not need energy to vaporize alcohol and water before the reaction; ii) the operating temperatures and pressures are suitable for the water-gas shift reaction, so it may be possible to generate hydrogen with low amounts of CO in a single step; and iii) the step of H2 purification or CO2 separation is simplified because of the pressure range of the effluent.

Another possibility that merits greater study is the operation under autother­mal conditions with ethanol/water/air mixtures. Here, the goal is to maximize the hydrogen yield, while minimizing the total combustion and the formation of by-products and carbon deposits on the catalysts. For both steam reforming and oxidative steam reforming, future research is needed to develop more stable, active, selective, and inexpensive catalytic systems that operate under the required final experimental conditions.

Finally, the integration of the ethanol reformation in an energetically favored total process is also an area, which, still in our day, remains to be completed from a technological point of view.

Efforts in the above-mentioned areas could lead to the practical use of ethanol as H2 supplier to generate clean electrical power in the not-so-distant future.