Integrating dark and light bio-hydrogen production strategies: towards the hydrogen economy

Abstract

Biological methods of hydrogen production are preferable to chemical methods because of the possibility to use sunlight, CO₂ and organic wastes as substrates for environmentally benign conversions, under moderate conditions. By combining different microorganisms with different capabilities, the individual strengths of each may be exploited and their weaknesses overcome. Mechanisms of bio-hydrogen production are described and strategies for their integration are discussed. Dual systems can be divided broadly into wholly light-driven systems (with microalgae/cyanobacteria as the 1st stage) and partially light-driven systems (with a dark, fermentative initial reaction). Review and evaluation of published data suggests that the latter type of system holds greater promise for industrial application. This is because the calculated land area required for a wholly light-driven dual system would be too large for either centralised (macro-) or decentralised (micro-) energy generation. The potential contribution to the hydrogen economy of partially light-driven dual systems is overviewed alongside that of other bio-fuels such as bio-methane and bio-ethanol.

1 Bio-fuels for sustainable energy production

An estimated 45% of the identified world oil reserves has been combusted and atmospheric CO₂ has increased by 20% since 1900 (Holmes and Jones 2003; Keeling and Whorf 2005). Conservative estimates suggest that demand for oil will outstrip supply by 2050 (Holmes and Jones 2003), while the Stern review (2006) highlights the urgent need for reduction in greenhouse gas emissions. A sustainable energy economy is needed and this will require a different fuel; one that is not limited in supply and whose use is environmentally benign. Hydrogen is now recognised as a key energy vector in the future energy economy. H₂ storage technology and fuel cell efficiency continue to receive urgent attention and have advanced sufficiently for transport applications to approach commercial viability. For example, BMW’s fleet of 100 “Hydrogen 7” cars, each having a material value of $500,000, is now available for promotional purposes (carsguide.com.au, November 2006). A prototype H₂-powered boat is also opening the way to economic transport via inland waterways (Bevan et al. 2007). Whereas the use of H₂ in transport applications is dependent upon such emerging technologies, its use in stationary applications, for electricity supply, is limited primarily by H₂ availability.

Biological approaches could contribute to large-scale H₂ production as various microorganisms can produce H₂ under moderate conditions from readily available, renewable substrates, making biological strategies potentially competitive with chemical process such as reforming and gasification. Bio-hydrogen processes are ‘CO₂-neutral’, being fuelled by carbohydrates originating from photosynthetic fixation of CO₂. Furthermore, bio-H₂ is free of CO and H₂S (both catalyst poisons) and requires no treatment before use in fuel cells for electricity generation (Macaskie et al. 2005).

Suitable feeds for bio-hydrogen generation processes can be found in agricultural residues (Nath and Das 2003; Hawkes et al. 2008), food wastes (Franchi et al. 2004; Karlsson et al. 2008) and effluents from industrial processes such as refining sugar (Yetis et al. 2000; Ren et al. 2006), distilling alcohol (Sasikala et al. 1992), olive processing (Eroğlu et al. 2004), producing cheese (Davila-Vazquez et al. 2008b) and producing tofu (Zhu et al. 1995, 2002). Hence, microbial processes could be employed to remediate wastes while simultaneously producing H₂ with the dual economic benefit of energy production and savings in the cost of waste disposal. In the UK the majority of waste is disposed by land-filling and the related environmental damage is being recognised in financial terms via landfill tax; which is paid additionally to normal landfill fees. In 2008, UK landfill fees were in the range GBP 11–40/tonne, to which landfill tax added GBP 24/tonne increasing by £8/year (WRAP gate fees report 2008, www.wrap.org.uk/marketknowledge). The avoidance of waste disposal costs is, therefore, anticipated to be an important economic driver in the start-up of bioenergy processes.

The capability for H₂ formation is widespread among microorganisms, but only a few have been investigated with a focus on bio-hydrogen production. In particular, photosynthetic microorganisms such as microalgae, cyanobacteria and purple bacteria are of interest, along with various dark fermentations. Each of these candidates represents a potential method in its own right, but it has long been recognised that multi-organism systems, combining the capabilities of different species would be required to realise the maximum potential for bio-hydrogen production (Rocha et al. 2001; Wakayama and Miyake 2001; de Vrije and Claassen 2003; Nath et al. 2005; Basak and Das 2007). Several examples of dual systems are illustrated in Table 1. The purpose of this review is to advance the state of knowledge by comparing the successes of diverse strategies, relating them to the methods employed, evaluating the potential for energy generation and by highlighting potential problems.

Table 1 Summary of dual systems employed for bio-hydrogen production

2 The use of microorganisms for H₂ production

This work does not attempt to review microbial hydrogen production (for reviews see Vignais et al. 1985; Blankenship et al. 1995; Sasikala et al. 1995; Nandi and Sengupta 1998; Claassen et al. 1999; Das and Veziroglu 2001; Hallenbeck and Benemann 2002; Nath and Das 2004a, b; Bae et al. 2005; Dutta et al. 2005; Hawkes et al. 2007; Tsygankov 2007a, b), but provides a summary of those organisms which have been studied expressly for the purpose of H₂ production. Combining different organisms in multi-organism strategies creates the possibility to exploit the most useful facets of different metabolisms. In order to evaluate multi-organism strategies, the broad metabolic mechanisms are first considered individually.

2.1 Photobiological hydrogen production

Many microorganisms have evolved the capacity to harness solar energy for growth and several types of photosynthetic microorganism are potentially useful for bio-hydrogen production. Artificial light sources are often used as models for future applications with sunlight, but solar bio-hydrogen projects have been successful despite the diurnal and seasonal variations in light intensity.

2.1.1 Photoautotrophic microorganisms

Photoautotrophs produce H₂ by two distinct mechanisms: “direct photolysis” and “indirect photolysis” which can both occur in the same organism (Fig. 1). Like higher plants, microalgae (green unicellular algae) and cyanobacteria (previously called blue-green algae) have two photosystems (Photosystems I and II: PSI and PSII), which produce H₂ by “direct photolysis” in which water is decomposed to H₂ and O₂ (oxygenic photosynthesis). The dual photosystems split water, reducing electron carriers and exporting protons to generate a proton gradient for ATP generation (Miyake et al. 1999).

Fig. 1

Direct and indirect photolysis. Through direct photolysis, the H₂ evolving enzyme is hydrogenase in microalgae and nitrogenase in cyanobacteria (see text). The dotted line represents the avoided inhibition of dark fermentation by O₂ via indirect photolysis

Hydrogenase and nitrogenase enzymes are found in cyanobacteria, but (as in purple bacteria, see Sect. 2.1.2) self-sustained H₂ formation in the light results from the activity of nitrogenase, which consumes ATP and re-oxidises electron carriers. In microalgae a hydrogenase performs the reduction of 2H⁺ to H₂ without any ATP requirement. In “indirect photolysis”, CO₂ is fixed via the Calvin cycle to synthesise simple sugars which are then accumulated as polycarbohydrates (starch in microalgae and glycogen in cyanobacteria). Stored carbohydrates can be metabolised fermentatively (Sect. 2.2) to generate H₂ indirectly.

In direct photolysis H₂ production is limited due to the inhibition of hydrogenase and nitrogenase by oxygen generated from water. In contrast indirect photolysis separates the production of O₂ and H₂ spatially (into compartments) or temporally (into aerobic and anaerobic phases) as described (Levin et al. 2004) and is therefore sustainable. For example, continuous H₂ production was achieved for up to 1 month using Anabaena cylindrica, when O₂ was artificially removed from the culture by purging with argon gas (Madamwar et al. 2000).

Cyanobacteria are divided into non nitrogen-fixing varieties (e.g. Synechococcus spp.) which form only one kind of cell (akinetes) and nitrogen-fixing varieties (e.g. Nostoc, Anabaena spp.), which form akinetes and also heterocysts arranged into filaments—chains of cells connected by channels for the exchange of nutrients (Tsygankov 2007a, b). Heterocysts differ from akinetes due to the absence of O₂ generation by PSII, the increased rate of O₂ consumption by respiration, the presence of a thick envelope to limit the ingress of environmental O₂ and the expression of nitrogenase to fix N₂ as NH₄ ⁺, supporting the growth of the adjacent akinetes (Madamwar et al. 2000; Tamagnini et al. 2002). Heterocystous cyanobacteria separate H₂ production and O₂ production spatially (by compartmentalisation) accumulating glycogen in the vegetative akinetes and fermenting it to produce H₂ in the anaerobic heterocysts. Non N₂-fixing cyanobacteria and microalgae separate H₂ production and O₂ production temporally, producing H₂ by the dark anaerobic fermentation of photosynthesised carbohydrates (Carrieri et al. 2008). Upon transition to darkness, the generation of O₂ by the photosystem ceases and residual O₂ is consumed by respiration enabling H₂ production (Tsygankov 2007a, b).

The capacity for self-sustained aerobic H₂ production in the light (i.e. without the artificial removal of O₂), is an advantageous property of heterocystous cyanobacteria (e.g. Anabaena spp.) achieving maximum H₂ production rates of ca. 100 μmol H₂/mg chlorophyll a/h, with light conversion efficiencies of up to 3.9% (proportion of absorbed light energy converted to H₂) (Dutta et al. 2005; Yoon et al. 2006; Sakurai and Masukawa 2007). Rates were increased three to sevenfold in Anabaena mutants deficient in uptake hydrogenase activity (Borodin et al. 2000; Happe et al. 2000; Masukawa et al. 2002; Yoshino et al. 2007) and this strategy was applied in outdoor culture. However, the maximum light conversion efficiency was only 0.1%, which has implications for the large scale application of this approach (Lindblad et al. 2002; Tsygankov et al. 2002).

Unicellular cyanobacteria have also been studied with a view to dark fermentative H₂ production, being unsuitable for photoproduction of H₂ due to their high (competing) uptake hydrogenase activity in the light (Troshina et al. 2002). However, a mutant of Synechocystis deficient in uptake hydrogenase activity photoproduced H₂ at a rate of 6 mmol H₂/g chlorophyll a/h (2 ml/l/h) (Cournac et al. 2004).

Like unicellular cyanobacteria, microalgae were originally studied for dark H₂ production by indirect photolysis (Miyamoto et al. 1987). The isolation of Chlamydomonas spp. MGA161 having a high rate of H₂ photoproduction (6 mmol H₂/g chlorophyll a/h), high starch accumulation (18% w/w) and unusually rapid and efficient dark fermentation (2 mol H₂/mol starch–glucose) prompted the study of a dual system (Miura et al. 1986) (Table 1).

The extent of metabolic engineering success in microalgae lags behind that of cyanobacteria due to the greater difficulty of eukaryotic genetic engineering. Work is ongoing to improve light conversion efficiency through the truncation of light-harvesting antenna complexes, an approach already proven using PNS bacteria (Sect. 2.1.2) (Akkerman et al. 2002; Polle et al. 2002). Other approaches are to study and develop O₂-tolerant hydrogenases (Ghirardi et al. 2005; Tosatto et al. 2008) and to express clostridial hydrogenase in non-heterocystous cyanobacteria, the aim being to engineer the rapid and ATP-independent (hydrogenase-mediated) H₂ production by direct photolysis in a fast-growing host organism, possibly overcoming O₂-inhibition through strong expression (Miyake et al. 1999).

A relatively new method to increase rates of H₂ production by direct photolysis is nutrient deprivation. Under conditions of sulfate-limitation the iron–sulfur clusters of PSII subunits cannot be maintained and PSII activity is inhibited (Wykoff et al. 1998). The rate of O₂ production decreases, while the respiration rate remains high and establishes anoxia, which permits hydrogenase and/or nitrogenase expression. The result is sustained H₂ production via direct photolysis. The technique was pioneered using microalgae (Benemann 1996; Jo et al. 2006; Laurinavichene et al. 2006) and has been recently extended to cyanobacteria (Antal and Lindblad 2005).

2.1.2 Photoheterotrophs

Purple non-sulfur (PNS) bacteria are anoxygenic photosynthetic bacteria which, unlike the purple and green sulfur bacteria, do not produce H₂S (a powerful catalyst poison) and the off-gas is typically >90% H₂, hence it is suitable for use in PEM-fuel cells without purification (Nakada et al. 1995).

PNS bacteria produce H₂ under photoheterotrophic conditions (light, anaerobiosis, organic electron donor) although they are metabolic generalists capable of autotrophic and heterotrophic growth. The best-studied species belong to the genera Rhodobacter, Rhodopseudomonas and Rhodospirillum.

H₂ is produced by the nitrogenase enzyme, which is active anaerobically under nitrogen limitation (Vignais et al. 1985). In the absence of N₂ the production of H₂ occurs according to Eq. 1 (Koku et al. 2002).

$$ 2\rm{e}^{ - } + 2{\text{H}}^{ + } + 4{\text{MgATP}} \to {\text{H}}_{2} + 4{\text{MgADP}} + 4P_{i} $$

(1)

In this respect the reaction serves to dissipate excess ATP and reducing power where growth is nitrogen-limited. The nitrogenase complex must be saturated with ATP and also NADH for optimal activity, hence H₂ photoproduction occurs most rapidly under saturating light intensity at the expense of organic electron donors.

Nitrogenase activity is strictly a facet of anaerobic metabolism since O₂ damages the photopigments needed to maintain ATP flux for nitrogenase activity and nitrogenase expression is strongly inhibited by oxygen (Koku et al. 2002). Sustained H₂ photoproduction is possible as the single PSI of these organisms does not generate O₂ (this is termed anoxygenic photosynthesis) and continuous H₂-producing cultures have been operated for up to several months (Liessens and Verstraete 1986; Weetall et al. 1989; Eroğlu et al. 1997; Hassan et al. 1997; Fascetti et al. 1998; Tsygankov et al. 1998; Yokoi et al. 2001; Franchi et al. 2004; Shi and Yu 2006).

Rocha et al. (2001) analysed a large number of reports, indicating that the efficiency of light conversion to H₂ is variable for PNS bacteria, the average value being ca. 4%. The theoretical maximum photosynthetic efficiency is considered to exceed 10% (Akkerman et al. 2002) but the photosystems of PNS bacteria saturate at low light intensity, leading to low light conversion efficiency under high light intensity, e.g. in solar photobioreactors (PBRs) (Wakayama and Miyake 2002). PNS bacteria are adapted to photosynthesis at low light intensities, requiring large light harvesting complexes to capture diffuse light energy and conduct it into the reaction centre. Light conversion efficiency may be improved beyond 10% by genetic manipulation to reduce the size of light-harvesting antennae, thereby increasing the saturating light intensity (Table 2). This would allow efficient H₂ production at higher light intensities, by deeper or denser cultures (Miyake et al. 1999; Vasilyeva et al. 1999; Kondo et al. 2002, 2006; Kim et al. 2004a, b, 2006a) (Table 2).

Table 2 Bottlenecks to the application of anoxygenic photosynthetic bacteria in H₂ production

Nitrogenase-mediated H₂ formation is irreversible (Hillmer and Gest 1977b), which is an advantageous property in relation to reversible hydrogenase-mediated H₂ production, which is inhibited under high partial pressure of H₂ (Valdez-Vazquez et al. 2006). However, in PNS bacteria, uptake hydrogenase activity can detract from H₂ yields (Sasikala et al. 1990), prompting the development of uptake hydrogenase deficient mutants with up to 70% increased H₂ production efficiency (Willison et al. 1984; Jahn et al. 1994; Worin et al. 1996; Öztürk et al. 2006) (Table 2).

Nitrogenase re-oxidises electron carriers to reduce 2H⁺ to H₂ and any other reductive processes (‘electron sinks’) can compete with and detract from H₂ production. The formation of the carbon storage polymer poly-β-hydroxybutyrate (PHB) from acetate is such a competing reductive reaction (Eq. 2, Tabita 1995) (Vincenzini et al. 1997; Khatipov et al. 1998).

$$ 9n\,{\text{CH}}_{ 3} {\text{COOH}} \to 4({\text{CH}}_{ 3} {\text{CH}}_{ 2} {\text{CHCOO}}^{ - } )_{n} + 2n\,{\text{CO}}_{2} + 6n\,{\text{H}}_{ 2} {\text{O}} $$

(2)

Mutagenesis of the PHB synthase gene yielded PHB deficient mutants, which were capable of H₂ production under conditions that would normally favour PHB synthesis (Hustede et al. 1993) (Table 2). In recent studies, double mutants lacking uptake hydrogenase and also PHB synthase produced H₂ at up to 2.5-fold higher rates compared to the parent strain (Lee et al. 2002; Kim et al. 2006b), while in a separate study a similar double mutant sustained H₂ production for over 45 days, while the wild-type ceased H₂ production after 10 days (Franchi et al. 2004).

Numerous simple organic molecules serve as suitable electron donors for PNS bacteria, including common fermentation products such lactate, acetate, butyrate, propionate and succinate (Hillmer and Gest 1977a, b), alcohols such as ethanol and propanol (Fuji et al. 1987) and other substrates such as aromatic acids (e.g. cinnamate, benzoate) (Sasikala et al. 1994b; Fissler et al. 1995). The substrate range is strain-specific (Tao et al. 2008) and the biochemical pathways of assimilation are uncertain for many of these substrates, with the exception of acetate (a common fermentation product). In most bacteria acetate is assimilated using the glyoxylate cycle, but a diverse group of microorganisms (including Rhodobacter sphaeroides and Rhodospirillum rubrum) lacks the key glyoxylate cycle enzyme, isocitrate lyase, while rapidly assimilating acetate. An alternative citramalate cycle is now thought to operate in these species (Ivanovskii et al. 1997; Filatova et al. 2005a, b). The distinction is important in the context of H₂ production as species lacking the glyoxylate shunt generally require the availability (not the addition) of CO₂ during H₂ production from acetate, with the exception of R. sphaeroides, which has a high capacity for acetate consumption (and hence CO₂ production) compared to other PNS bacteria, and also has a thick capsule obstructing the diffusion of produced CO₂ (Table 2). Some uncertainty remains over the suitability of ethanol, a common fermentation product, as an electron donor for photoproduction of H₂. A Rhodopseudomonas species produced H₂ at the expense of various alcohols (Fuji et al. 1987) and ethanol was consumed simultaneously with acetate by Rhodobium marinum at ca. 50% the rate of acetate, although the initial concentration of ethanol was ca. 25% that of acetate (Ike et al. 2001). Ethanol was rapidly removed from an Escherichia coli fermentation effluent by R. sphaeroides O.U.001 after a delay of 96 h, although the induction of ethanol-utilising enzymes was not monitored (Redwood and Macaskie 2006). Hence, it is plausible that other PNS bacteria would be capable of ethanol utilisation after an adaptation period.

PNS bacteria have significant potential for industrial application as mixed cultures can be maintained for extended periods (Liessens and Verstraete 1986; Ko and Noike 2002; Fang et al. 2005; Ying Li et al. 2008), industrial waste streams can make suitable feeds for the photoproduction of H₂ (Thangaraj and Kulandaivelu 1994; Fascetti et al. 1998; Yiğit et al. 1999; Zhu et al. 1999a; Yetis et al. 2000), larger scale PBRs are under development (Hoekema et al. 2002, 2006; Claassen and de Vrije 2007) and outdoor projects using sunlight have been successful (Wakayama et al. 2000; Wakayama and Miyake 2002; Eroğlu et al. 2008).

The foremost limitation with PNS bacteria is the incompatibility of nitrogenase activity and the presence of NH₄ ⁺. Waste streams can only be used for H₂ production if they are of high C/N ratio and many reports of this application are available (Sasikala et al. 1992; Turkarslan et al. 1997; Tsygankov et al. 1998; Yiğit et al. 1999; Yetis et al. 2000; Eroğlu et al. 2004; Fang et al. 2005). Low C/N waste streams have been applied successfully for the purposes of biomass production and effluent remediation (Ensign 1977; Hassan et al. 1997; Cornet et al. 2003; Yun and Ohta 2005). H₂ production using low C/N feeds has been accomplished by the use of immobilisation matrices which exclude cations such as NH₄ ⁺ (Zhu et al. 1999b, 2001) and the development of nitrogenase-derepressed strains (Wall and Gest 1979; Zinchenko et al. 1991, 1997; Yagi et al. 1994) (Table 2). These approaches were not, however, tested at pilot-scale or in continuous culture and issues such as the economic viability of immobilisation and the long-term stability of nitrogenase-derepressed strains remain to be addressed.

PNS bacteria are capable, therefore, of efficient conversion of organic acids to H₂, providing a potentially applicable method for the remediation of wastes rich in organic acids, alcohols or aromatics. With the exception of unusual strains (Macler et al. 1979; Macler and Bassham 1988; Oh et al. 2004), PNS bacteria lack the capacity for the efficient conversion of sugars to H₂ and for this application a dark fermentation is the method of choice.

2.2 Dark hydrogen fermentation

Large quantities of simple and complex carbohydrates are available as agricultural and food processing residues (Easterly and Burnham 1996; Filho and Badr 2004; Haq and Easterly 2006; Mabee et al. 2006; Levin et al. 2007). Fermentative bacteria represent a promising means not only to reclaim energy from these wastes in the form of hydrogen but also to utilise the wastes as resources, a particularly valuable attribute given the escalating cost of landfill (Bartelings et al. 2005). Indeed, it was calculated that the savings in landfill tax would be the main economic driver, outweighing the value of the energy produced via dark-fermentative production of H₂ from confectionery waste (Macaskie 2004).

The anaerobic degradation of organic matter by heterotrophic microorganisms can liberate H₂ at high rates, depending on the particular organisms and conditions. Fermentation generates energy solely through substrate level phosphorylation. Substrates are converted to reduced compounds, which are excreted as waste products and the ATP yield is low, in comparison to respiration. The formation of relatively reduced organic molecules is an integral part of all dark fermentations and some of these molecules (e.g. acetate) can inhibit H₂ production if allowed to accumulate (Roe et al. 1998; Kirkpatrick et al. 2001; Van Ginkel and Logan 2005).

In a minority of fermentative microorganisms (e.g. Klebsiella spp.) H₂ production is primarily mediated by nitrogenase (Vignais et al. 2001) but, due to the high ATP requirement and low turnover rate of nitrogenase, the theoretical H₂ yield is only 0.5 mol H₂/mol hexose (Wakayama and Miyake 2001). Without the contribution of light energy through photosynthesis, hydrogenase is preferred for H₂ production due to its higher rate of turnover and lower metabolic cost. The highest fermentative H₂ yields have been achieved using clostridia, enteric bacteria and hyperthermophiles (see reviews: Hallenbeck 2005; Davila-Vazquez et al. 2008a).

H₂ fermentations are restricted by the Thauer limit. Thermodynamically, no more than 4 mol H₂ can be produced from 1 mol hexose because substrate level phosphorylation must produce whole numbers of ATP and the yield of ATP from glucose must be at least 1 mol/mol for the cell to survive (Thauer 1977). However, microbial fermentation typically generates more than 1 mol ATP and less than 4 mol H₂/mol hexose, quantities that vary according to the metabolic system and conditions.

2.2.1 Axenic dark fermentations

Axenic cultures (pure cultures containing clonal microbial populations) have been used in the majority of fermentation research, creating a wealth of information regarding model organisms and the understanding of their fermentative metabolism has facilitated and rationalised the optimisation of conditions for H₂ production.

Dark fermentations are united by the initial glycolytic generation of ATP, NADH and pyruvate. Three enzymes compete for pyruvate: pyruvate: ferredoxin oxidoreductase (PFOR), pyruvate: formate lyase (PFL) and the fermentative lactate dehydrogenase (LDH). The realised H₂ yield is dependent upon the fate of pyruvate, which differs among species due to varying activities of PFL, PFOR and LDH, of which one or more may be present (Fig. 2).

Fig. 2

Suitable dark fermentations for dual systems. Pathways are abridged to highlight the overall balances. Dotted lines indicate alternative/competing pathways. Abbreviations: LDH lactate dehydrogenase, PFL pyruvate: formate lyase, ACK acetate kinase, FHL formate hydrogen lyase, PTA phosphotransacetylase, ADH alcohol dehydrogenase, NFOR NADH: ferredoxin oxidoreductase, TL thiolase, BHBD hydroxybutyryl-CoA dehydrogenase, EH enoyl-CoA hydratase, BDH butyryl-CoA dehydrogenase, PFOR pyruvate: ferredoxin oxidoreductase. Compiled from Sode et al. (2001) and Chen et al. (2006)

Mixed-acid fermentation, in which the key enzymes are PFL and the formate: hydrogen lyase (FHL) complex (comprising a specific formate dehydrogenase and hydrogenase) (Fig. 2b), is performed by facultative anaerobes such as Escherichia coli. PFL converts pyruvate to acetyl-CoA and formate, which is cleaved to H₂ and CO₂ by FHL, while acetyl-CoA is divided between the formation of acetate (which generates ATP) and the formation of ethanol (which oxidises NADH to regenerate NAD). PFOR is expressed constitutively to a low level (Knappe 1987), but since H₂ is entirely formate-derived (Ordal and Halvorson 1939) PFOR is not thought to be involved in H₂ production in E. coli.

Ideally, mixed-acid fermentation yields 2 mol H₂/mol glucose (Fig. 2b), but in batch mode a yield of ca. 50% of this is usually obtained due to diversion of pyruvate into lactate formation. The latter can be suppressed by control of culture conditions or through metabolic engineering (Sode et al. 1999, 2000). While the cleavage of formate is irreversible, H₂ recycling is an issue, as suggested by the observation of 37% increased H₂ yield in Hyd-2 mutants of E. coli (Redwood et al. 2007). The rate of H₂ formation was also increased through the increased expression of FHL (Penfold et al. 2003; Yoshida et al. 2005).

Facultative anaerobes of the related genus Enterobacter also produce H₂ from formate but analysis of the fermentation balance implicated the simultaneous activity of the NADH pathway (Tanisho and Ishiwata 1995; Tanisho et al. 1998; Kurokawa and Shigeharu 2005), in which the regeneration of NAD⁺ is coupled to the reduction of ferredoxin by NADH: ferredoxin oxidoreductase (NFOR). Reduced ferredoxin subsequently transfers electrons onto H⁺ to produce additional H₂. The NADH pathway (operating simultaneously with PFL/FHL) could theoretically achieve the Thauer limit (4 mol H₂/mol glucose). However, both electron transfer reactions (from NADH onto oxidised ferredoxin and from reduced ferredoxin onto H⁺) are reversible and neither would be considered electrochemically feasible under standard conditions: i.e. the standard electrode potentials of the NAD and ferredoxin half-cells (−320 and −400 mV, respectively) are more positive than that of the H⁺ half-cell (−414 mV) (McCormick 1998). A very low H₂ partial pressure (pH₂) (theoretically <60 Pa or <0.0006 bar) is required to drive this reaction forwards and H₂ yields exceeding 2 mol H₂/mol glucose were obtained only under vacuum or with continuous gas purging to strip away H₂ (Park et al. 2005). Indeed, a maximum yield of 3.9 mol H₂/mol glucose was reported using E. cloacae under a vacuum of 330 torr (equivalent to 0.44 bar or 44 kPa) (Mandal et al. 2006).

Clostridia also use the NADH pathway. In this case PFOR cleaves pyruvate to acetyl-CoA and CO₂, transferring electrons to ferredoxin, which is coupled to a reversible hydrogenase to produce H₂. In this situation, all H₂ is produced by a single reversible reaction and it is even more important to maintain a low pH₂ (Kataoka et al. 1997; Mizuno et al. 2000). Advances in gas separation technology may permit a purge-gas recycle system to remove the need for large quantities of inert, anaerobic purge gas for H₂ removal (Nielsen et al. 2001; Liang et al. 2002; Teplyakov et al. 2002).

A positive pressure (ca. 1–1.5 bar at 25°C) is needed for H₂ uptake by metal hydride H₂-stores (Züttel 2004). Therefore, to charge a H₂ store directly from a fermentation culture (without intermediary gas-pressurisation) would require an organism capable of sustaining H₂ production under high pH₂. This would require the absence of biological H₂ recycling and would preclude a reversible H₂-producing system, such as the NADH pathway occurring in enteric bacteria and clostridia, but may be possible using an uptake hydrogenase mutant of E. coli in which the FHL complex (involving hydrogenase-3) performs the irreversible oxidation of formate to form H₂ (and CO₂). However, a degree of reversibility is a fundamental property of hydrogenases (Van Haaster et al. 2005) and although hydrogenase-3 has no uptake role during fermentation (Redwood et al. 2007), it is known to operate reversibly when coupled to redox dyes (e.g. Sauter et al. 1992). Therefore, the latter strategy may tolerate a higher pH₂, but regardless of the organism employed, a pressurisation step would be advantageous between the fermentation and the H₂-store.

Various Clostridium spp. have been investigated for bio-hydrogen production (Collet et al. 2004), of which C. butyricum is perhaps the best known. Like E. coli and Enterobacter aerogenes, this organism is mesophilic but unlike them, it is a strict anaerobe. Hence, clostridial growth media are usually supplemented with a reducing agent to ensure anaerobiosis. Alternatively, a facultative aerobe, added to the H₂-production culture, was effective as an O₂-scavenger (Yokoi et al. 2001).

The H₂ yield from C. butyricum could in theory reach 4 mol H₂/mol hexose (Fig. 2c) although a detailed metabolic analysis of C. butyricum calculated a maximum of 3.26 mol H₂/mol hexose (Chen et al. 2006) and practical yields obtained using clostridia rarely exceed 2 mol H₂/mol hexose (Collet et al. 2004; Ferchichi et al. 2005).

The clostridial species selected for H₂ production produce acetate and butyrate rather than propionate but they sporulate in response to environmental stresses such as heat or nutrient depletion, hence, the feeding regimes used in continuous culture are designed to maintain excessive nutrient concentrations to minimise sporulation (Hawkes et al. 2002). Asporogenic mutants have proved advantageous in ethanol production from cellulose, but have not yet been applied to H₂ production (Taillez et al. 1983). Whereas mesophilic clostridia sporulate as the temperature increases, certain clostridial species are moderately thermophilic. For example, C. thermolacticum prefers to grow at 54°C (Collet et al. 2004). Recently hyperthermophiles, which live and produce H₂ at temperatures above 60°C, have been studied. Little biochemical information is yet available (e.g. de Vrije et al. 2007) but it seems that hyperthermophiles are capable of H₂ production with higher yields than mesophiles (Hallenbeck 2005). For example, a yield of 2.8 mol H₂/mol glucose was reported for Thermotoga elfii and 3.2–3.7 mol/mol for Caldicellulosiruptor saccharolyticus (Van Niel et al. 2002; Kadar et al. 2004; de Vrije et al. 2007). Observations support the connection of H₂ production with the hydrogenase-linked oxidation of electron carriers (as in clostridia), rather than the decomposition of formate (as in enteric bacteria). A pH₂ of 10–20 kPa (0.1–0.2 bar) induced a metabolic shift to inhibit H₂ production in C. saccharolyticus (Van Niel et al. 2003) and a limiting H₂ pressure of 20 kPa (0.2 bar) was reported for a mixed hyperthermophilic culture (Van Groenestijn et al. 2002), while formate was not decomposed by Thermotoga neapolitana (Van Ooteghem et al. 2004).

The necessity of growth on solid media for molecular work (i.e. at temperatures lower than the melting point of agar), makes hyperthermophiles less readily amenable to genetic engineering, although alternative solid media such as gelrite are available (Van Ooteghem et al. 2004). Thermophilic cultures are resistant to overgrowth by mesophilic contaminants but, although an economic analysis is not available, the energetic costs associated with maintaining 70°C may mitigate against large-scale application.

Several mesophilic and thermophilic clostridia and hyperthermophiles have the capacity to utilise complex carbohydrates such as cellulose and starch, a valuable property widening the potential for the use of industrial waste streams and agricultural residues as feeds. For example, T. neopolitana can utilise dextrins (Van Ooteghem et al. 2004) and C. thermocellum produced 1.6 mol H₂/mol hexose from delignified wood (Levin et al. 2006). Enteric bacteria generally lack the ability to metabolise complex carbohydrates although the necessary genes can be introduced in the case of E. coli (Dien et al. 2000).

2.2.2 Mixed dark fermentations

The use of mixed cultures offers practical advantages over the use of pure cultures, such as the use of feedstocks without pre-treatment or sterilisation and is already a proven, commercially available technology (Kyazze et al. 2007). Inocula for H₂ production can be obtained from soil, compost or anaerobic digestion sludge (Hawkes et al. 2002, 2007). H₂ was produced from sucrose using sewage sludge microflora with a yield of 1.7 mol H₂/mol hexose (Lin and Lay 2005) and from food processing effluent using a heat-treated (2 h, 104°C) sludge inoculum with typical yields of 0.2–0.87 mol H₂/mol hexose (Oh and Logan 2005). Rice slurry was fermented by a heat-treated (30 min, 100°C) clostridial community to produce up to 2.5 mol H₂/mol hexose (Fang et al. 2006). Paper sludge and cellulose powder were rapidly degraded by mesophilic anaerobic consortia, producing mixtures of CH₄ and H₂ (Ueno et al. 1995; Valdez-Vazquez et al. 2005).

For mixed cultures there is a tendency towards lower rates and yields of H₂ production because non H₂-producing organisms (e.g. methanogens and sulfate-reducers) consume a proportion of the substrate and perform H₂ uptake using H₂ as an electron donor. Furthermore, H₂S (the product of dissimilatory sulfate reduction) is a potent catalyst poison requiring removal if the bio-hydrogen is intended for use in a fuel cell. Hence, for the efficient production of clean H₂, the microbial population must be controlled to some degree in order to select for H₂-producers. Methanogens can be suppressed by the addition of chemical inhibitors or by operating continuous cultures at low pH or HRT (Hawkes et al. 2007). The microbial population is often manipulated by applying pre-selection on the inoculum (Valdez-Vazquez et al. 2005). A widely adopted strategy is to select for spore-forming clostridia using a heat-treated inoculum and where this can be achieved the properties of clostridial fermentation are predominantly applicable (Kim et al. 2004a, b; Van Ginkel and Logan 2005). However, heat treatment also eliminates non-sporulating H₂-producers (e.g. Enterobacter spp.) and selects for spore-forming H₂-consumers (e.g. some acetogens) (Kraemer and Bagley 2007). The metabolic switch from H₂ production to solventogenesis can be avoided by the intermittent release of headspace pressure and N₂ purging (Valdez-Vazquez et al. 2006).

3 Hybrid hydrogen

As reviewed above, no single-stage system has been shown to produce H₂ beyond 4 mol H₂/mol hexose. Current research focuses on the possible use of two-component systems via a variety of strategies (Table 1). These dual systems are united by the conversion of carbohydrates into organic acids in the 1st stage (which may be mesophilic or thermophilic and may not necessarily produce H₂), followed by the conversion of fermentation products into H₂ in the 2nd stage (Fig. 3). In some examples, algae or cyanobacteria initially photosynthesise carbohydrates, which are then fermented by the same organisms, while other systems use carbohydrates as the primary feed, either as artificial solutions or in the form of wastes or algal biomass.

Fig. 3

Dual systems in co-culture (a) and in sequential reactors (b)

3.1 Techniques for connecting the components of a dual system

Alongside the choice of organisms, dual systems have been implemented through a variety of strategies. The nature of the bridge connecting these two stages is a key part of the operational strategy affecting the overall productivity of the system. The simplest approach constitutes a co-culture in which different organisms are in direct contact and act simultaneously, under the same conditions (Fig. 3a). However, a compromise needs to be sought between the optimal requirements of each microbial component. While increasing the complexity and cost, sequential reactors permit the operator to maintain different conditions in separate parts of the dual system, allowing a combination of organisms, which may not be compatible in co-culture (Fig. 3b). For example, (wild-type) microalgae/cyanobacteria and PNS bacteria were not compatible in co-culture since the photosynthetic generation of oxygen inhibited nitrogenase-mediated H₂ production by the PNS bacteria (Miyamoto et al. 1987; Weetall et al. 1989). Further, sequential reactor systems can be potentially more effective as either component can be optimised without compromise to the other and may be preferred even for ‘compatible’ combinations of organisms.

Sequential reactors require some method to transport fermentation products from the 1st reactor to the 2nd (while retaining biomass), which presents an engineering challenge for future scale-up operations. The simplest and most common method is ‘batch-transfer’ in which spent medium is transferred between reactors in batches. Centrifugation followed by filtration or autoclaving is usually performed to generate a clear, sterile feed for the 2nd stage (e.g. Yokoi et al. 2002; Redwood and Macaskie 2006). For large-scale application, continuous processes are generally preferred over batch systems. Fermentation products could potentially be transferred continuously through the use of bi-phasic solvent extraction, by continuous centrifugation or by membrane systems (Banik et al. 2003; Emanuelsson et al. 2003; Splendiani et al. 2003). Such techniques have yet to be applied in commercial H₂-production, but a continuous electrodialysis method was effective in coupling sequential H₂-producing reactors (Redwood and Macaskie 2007a, b) (Table 1).

3.2 Comparing diverse strategies

As explained in Sect. 3.1, the two components of a dual system may be bridged in several ways. To add to the difficulty of comparison, either part of the dual system may use free or immobilised cells and may operate in batch, fed-batch, repeated fed-batch, or continuous mode and the two components may be linked continuously or discontinuously in an open (exit flow to waste) or recycling system.

In order to reach some conclusions about the efficacy of different strategies, a common comparator is needed. Rates of H₂ production are not always meaningful in this kind of comparison due to number of contributory factors and variables. A common parameter taking into account many factors can be useful, (e.g. H₂ volume produced/reactor volume/dw/mol substrate consumed/time) but the accurate interpretation of so many factors from published accounts is rarely possible. The molar yield of H₂ from hexose (or monosaccharide) is the most appropriate measure for the comparison of dual systems, as it can be applied regardless of organisms, scale, means of integration and the chemical natures of substrates (Table 1). This factor can be misleading, however, in the case of complex feeds (e.g. algal biomass, tofu wastewater) containing organic acids or non-hexose substrates such as fats and proteins from the outset, which contribute to the fermentative yield of organic intermediates (Ike et al. 1997, 2001).

Additionally, the strategies under evaluation involve an input of solar energy, implying a positive relationship between the system’s production capacity and its associated area for absorbing solar energy. Due to the cost and technical challenge of constructing expansive PBRs of gas-tight, transparent materials, the energy yield per area of reactor footprint is considered as a second key parameter for comparison.

3.3 Selection of organisms for the 2nd stage

In a dual H₂ producing system, the 2nd stage functions to clean up the effluent from the 1st stage (i.e. to decrease its BOD) and to produce a secondary H₂ stream at the expense of stage 1 products (e.g. reduced organic molecules). An algal-PNS bacterial symbiosis was proposed but no experimental data is yet available (Melis and Melnicki 2006). Fermentation products (e.g. acetate) could be used as C-source for the growth of microalgae, cyanobacteria or PNS bacteria. While acetate is regularly used in algal growth media (Kim et al. 2006c), the authors are not aware of any published attempts to cultivate algae or cyanobacteria using fermentation products as carbon sources. PNS bacteria, conversely, have been cultivated to produce biomass, single-cell protein, or PHB, using primary fermentation waste streams (Ensign 1977; Hassan et al. 1997). Reduced organic molecules are the preferred carbon source for PNS bacteria (Biebl and Pfennig 1981), suggesting that these species may be ecologically associated with fermentative, organic acid-producing organisms. There are many examples of the use of PNS bacteria for H₂ production from wastes (e.g. Thangaraj and Kulandaivelu 1994), many of which have similar characteristics to fermentation effluents.

Using PNS bacteria in the 2nd stage, organic fermentation products can be converted to H₂ with high efficiency (50–100% of stoichiometric yield) (Rocha et al. 2001) and light conversion efficiencies could approach a hypothetical maximum of 10% (Akkerman et al. 2002). Using both dairy and sugarcane wastewaters the PNS bacterium Rhodopseudomonas capsulata produced H₂ at a tenfold higher rate than the cyanobacterium Anacystis nidulans (Thangaraj and Kulandaivelu 1994).

PNS bacteria are the popular choice for the conversion of organic fermentation products to H₂ in the 2nd stage of a dual system. Of the 37 reports summarised in Table 1, only one employed a purple sulfur bacterium for this purpose (Akano et al. 1996; Ikuta et al. 1997) and none employed microalgae or cyanobacteria, which employ a more energetically demanding mechanism of H₂ production than do APB (Claassen et al. 1999).

3.4 Selection of organisms for the 1st stage

In this overview, dual systems are grouped broadly according to whether both 1st and 2nd stages or only the 2nd stage are light-driven.

3.4.1 Dual systems with photoautotrophic 1st stage

In this approach, photoautotrophs produce H₂ and accumulate carbohydrate during an initial light phase. The photoautotroph switches to fermentative metabolism during a subsequent dark phase, converting starch or glycogen to organic fermentation products, which are utilised by PNS bacteria to generate H₂ in the next light phase. Alternatively, the phototroph cell mass may be harvested to supply the feed for a dual system with dark fermentative 1st stage (Sect. 3.4.2).

1st and 2nd stages may be joined in co-culture or sequentially with transfer of spent broth between stages. Co-culture was until recently unsuitable for this combination because the O₂ produced by microalgae/cyanobacteria would prevent photoproduction of H₂ by PNS bacteria. Microalgal strains exhibiting a decreased rate of photolysis relative to respiration (i.e. decreased rate of O₂ production) have recently become available and work is ongoing to characterise H₂ production in co-cultures of attenuated microalgae and PNS bacteria (Melis and Happe 2004).

In this type of system, there is the possibility for H₂ production in 3 stages because microalgae/cyanobacteria can produce H₂ both at night and by day. While the production of H₂ through photolysis (or nitrogenase) is widely reported, there are no accounts of 3-stage systems in which H₂ production occurred in all 3 stages. In all cases, the accumulated biomass was the sole substrate from which H₂ was generated, by dark algal/cyanobacterial fermentation followed by PF.

In most cases, algal fermentation produced no H₂. In the most successful example (Miura et al. 1992), the yield of microalgal fermentation from accumulated starch was ca. 1.3 mol H₂/mol hexose and the overall yield was maintained at a steady 10.5 mol H₂/mol hexose for 5 days under continuous illumination (8 mol/mol under diurnal illumination). Chlamydomonas spp. were found to accumulate higher levels of starch than other microalgae (Ike et al. 2001) and strain MGA161 with a high fermentative H₂ yield was highlighted (Miura et al. 1986). Although the microalgal-based dual system achieved excellent H₂ yields based on the accumulated carbohydrate, the rate of carbohydrate accumulation limits the application of this strategy.

3.4.1.1 Energy generation potential with a photoautotrophic 1st stage

In the case of a dark-fermentative 1st stage fed by wastes or synthetic solutions, the feeding rate can easily be adjusted to control the overall rate of H₂ production, whereas an algal/cyanobacterial-driven system (dependent on light) is limited by the yield of photoautotrophic carbohydrate production (e.g. mol hexose/m² light capture area/day). Combined with the molar yield of H₂ per hexose in the dual system, this can indicate the potential rate of H₂ production from a given light capture area (Table 3). Using an 800 l pond-type pilot plant with CO₂ as the carbon source for cultivation of Chlamydomonas spp., Ikuta et al. (1997) achieved a maximum productivity of 92.6 mmol hexose/m²/day and an average productivity of 24.4 mmol hexose/m²/day over 23 days. Using a closed PBR, Chlamydomonas reinhardtii was grown under outdoor light conditions using CO₂ in addition to acetate as carbon sources, yielding 158 mmol starch–hexose/m²/day (Kim et al. 2006c). Assuming cultivation conditions can be optimised to maintain the highest rate of starch accumulation this value was used to calculate the potential productivities of algal/cyanobacterial-driven dual systems (Table 3).

Table 3 Potential productivities of algal/cyanobacterial-driven dual systems

Using the data provided by Levin et al. (2004) (see legend to Table 3), it can be calculated that at least 436 m² of light capture area would be needed to generate sufficient H₂ to power one home with modest energy requirements (1 kW), discounting the energy costs of the process (e.g. mixing, pumping, medium supplements, pH control and maintenance). Thus, significant improvements in the rate and efficiency of light conversion to carbohydrate would be required to permit biological energy generation by a dual system reliant upon microalgal or cyanobacterial starch accumulation, which is rather overshadowed by the availability of significant quantities of starch and cellulose wastes (Yokoi et al. 2002; Haq and Easterly 2006).

As a best-case scenario, metabolic engineering would lead to significant improvements in the efficacy of direct photolysis, allowing significant H₂ production coupled to carbohydrate accumulation. For decentralised domestic energy generation the available light capture area is estimated to be of the order of 20 m². This is challenging since it would require a 20-fold increase in productivity (along with parallel developments in process automation and assuming negligible operational energy costs). Large centralised hydrogen farms might be more efficiently run, but such an industry would be in direct competition with conventional agriculture, which currently accounts for 77% of land in the UK (Anon 2005). However, H₂ farms would not require fertile soil and might be operated in inhospitable environments such as deserts or on contaminated land where remediation might not be economically attractive (Aldhous 2006). The cultivation of ‘energy crops’ is currently receiving widespread attention (e.g. de Vrije and Claassen 2003; Aldhous 2006; Schnoor 2006). For example, Jatropha spp. are proposed as energy crops suitable for cultivation on sparse, non-arable land for the production of seed oil, which can be esterified to produce bio-diesel fuel with the co-production of ‘press-cake’ residues which are suitable substrates for bioconversion, e.g. to H₂ (Staubmann et al. 1997; Gübitz et al. 1999; Martínez-Herrera et al. 2006; Tiwari et al. 2007). A comparison of the potential energy yields per hectare for crop farms and photo-energy farms would repay study.

As the planet’s most plentiful energy source, solar energy must be part of any vision of future energy generation. This is the case either in a wholly light-driven system (e.g. microalgae + PNS bacteria) or in a partially light driven system (fermentation + PNS bacteria) where the fermentation is fed on biomass residues. As world population and food-demand grow, the limited availability of non-agricultural land may discourage algal or cyanobacterial cultivation. While ‘green roofs’ are established as a means of improved insulation and storm water retention, the potential of rooftop agriculture remains to be widely exploited (Nowak 2004). At the same time, the availability of residues is set to increase from both food and energy crops. Therefore, with the current state of knowledge and technology development, a dual system with dark fermentative 1st stage has a greater potential for near-term application.

On a broader scale, the use of microbial photosynthesis will have to compete with photovoltaic (PV) technology which, although as yet economically unattractive in most applications, is also under parallel development (Avi 2007). Data are, as yet, unavailable to compare the energy yields from optimised PV and bio-systems as industrial-scale PBRs for H₂ production are still under development. The estimated net energy ratio (NER) was ca. 2 for a PBR lasting 20 years, constructed using tubes of flexible polyethylene film (thickness 0.18 mm), under the assumptions of film replacement every 3 years, 80% time on-line, 20% loss of H₂ and discounting the costs of nutrients, temperature control and water (Burgess and Fernandez-Velasco 2007).

3.4.2 Dual systems with a dark fermentative 1st stage

Dark fermentation represents a rapid and relatively simple method for the conversion of carbohydrates into hydrogen, but the accumulation of organic fermentation products can exert stress upon the fermenting microorganisms and generates a secondary waste, requiring disposal (Eiteman and Altman 2006). Concurrently, fermentation products are preferred substrates for PNS bacteria, which oxidise reduced organic molecules and dispose of the reducing power as H₂. It has long been recognised that dark fermentation and photo-fermentation should be coupled to create an efficient scheme for waste-free hydrogen production (Odom and Wall 1983; Miyake et al. 1984). In practice, the maximum yield reported was 8.3 mol H₂/mol hexose (Kim et al. 2006c) and indeed several independent results of ca. 7 mol H₂/mol hexose were generated by different methods (Table 1).

The dark fermentation–photofermentation (DF–PF) dual system can be operated in continuous mode over extended periods. The longest experiment reported sustained H₂ evolution for 45 days by coupling lactic acid fermentation and a continuous photofermentation, but the yield of H₂/mol hexose cannot be calculated from the available data (Franchi et al. 2004). Yokoi et al. (2001, 2002) reported sustainable operation of a dual system for 30 days, maintaining a steady overall yield of 7 mol H₂/mol hexose using sweet potato starch residue. This system was used in repeated-batch culture, the fermenter being partially drained daily and the PBR every 5 days. A fully continuous system is currently under development (Redwood and Macaskie 2007a, b). A continuous E. coli reactor and a continuous R. sphaeroides PBR were integrated by anion-selective electrodialysis, simultaneously transferring anionic fermentation products, while retaining repressive ammonium ion, E. coli cells and suspended solids. This approach resulted in sustained H₂ production by E. coli with a yield of 1.6 mol H₂/mol hexose and sustained H₂ photoproduction by R. sphaeroides despite the presence of 15 mM ammonium ion in the initial feed. The overall yield was 2.4 mol H₂/mol glucose, attributable to a low efficiency in the PBR (38%) and a proportion of E. coli products being uncharged species (ethanol), not transported by electrodialysis. An overall yield of 10.1 mol H₂/mol glucose could be predicted based on a substrate conversion efficiency of 75% in the PBR and optimisation of the latter is in progress.

Therefore, with present approaches, a dual system can be sustained continuously and achieves on average ca. 60% of the hypothetical maximum, 12 mol H₂/mol hexose (Fig. 2d). As a priority, research is needed to investigate techniques for the integration and rate-balancing of inter-dependent bioreactors, alongside further study of the conditions needed to sustain high H₂ production continuously in either component of DF–PF dual systems.

3.4.2.1 Selection of organism for dual systems with dark fermentative 1st stage

An ideal fermentation, coupled to an ideal photo-fermentation, could approach the maximum stoichiometry of 12 mol H₂/mol hexose (Fig. 2d). As different fermentations can theoretically be coupled to a photofermentation to achieve the same maximum yield (Fig. 2), differences in practicality and experimental yields must be addressed.

The distinction is made between the use of obligate anaerobes in stage 1 and facultative aerobes/anaerobes because there are distinct differences between the biochemical mechanisms of H₂ production of these classes (Sect. 2.2, Fig. 2).

In the case described by Yokoi et al. (2001, 2002), the facultative anaerobe E. aerogenes was included in the 1st stage in co-fermentation with the strict anaerobe C. butyricum (Table 1). This example is classed among the strict anaerobic dual systems because Enterobacter, being unable to utilise starch, did not contribute to the fermentation but provided a cheaper alternative to reducing agents to ensure anaerobiosis by scavenging O₂ (Yokoi et al. 2001, 2002).

Facultative anaerobes (e.g. E. coli) can be pre-cultured rapidly, are readily amenable to metabolic engineering and do not require the addition of a reducing agent to ‘poise’ the redox potential, while the biochemistry of mixed-acid fermentation has been well-studied (Stephenson and Stickland 1932; Knappe 1987; Alam and Clark 1989; Clark 1989; Bock and Sawers 1996; Vardar-Schara et al. 2008). Nevertheless, obligate anaerobes have been preferred in the study of dual systems, perhaps due to the potentially higher H₂ yield. Table 1 shows 14 examples of strict anaerobe-driven dual systems, with an average overall yield (where given) of 5.73 mol H₂/mol hexose (or 47.8%). Conversely only eight examples of dual systems could be found using other types of fermentation in the 1st stage, of which systems based on lactic acid fermentation were the most effective, producing (overall) up to 7.3 mol H₂/mol hexose entirely from the 2nd stage (Kawaguchi et al. 2001). High overall yield is, therefore, possible without H₂ production in the 1st stage. This can be explained by the fact that lactic acid fermentation has been optimised for the industrial production of lactic acid (Li et al. 2004) and because lactate is theoretically converted to 6 H₂ in the 2nd stage, which typically operates with high efficiency (Sect. 2.1.2).

It is possible that researchers have frequently disregarded mixed-acid fermentation for use in the 1st stage because of its theoretically lower H₂ yield (Fig. 2). The fermentative yield of H₂ from hexose by living organisms is thermodynamically limited to 4 mol H₂/mol hexose (Thauer 1977). The metabolic pathways of strict anaerobes (e.g. C. butyricum) allow this to be achieved only under very low H₂ partial pressure, otherwise the reaction is stoichiometrically similar to mixed-acid fermentation, producing a maximum of 2 mol H₂/mol hexose (Fig. 2). The yield from strict anaerobic fermentation in a dual system has not exceeded 2.6 mol/mol (Table 1) and the use of a non-sporulating facultative anaerobe would represent an insignificant sacrifice of yield for a significant reward in practicality (Fig. 2d, Sect. 3.4.2.1).

Hyperthermophilic fermentations may yield up to 3.8 mol H₂/mol hexose in practice (Sect. 2.2.1), but no accounts describe dual systems using hyperthermophiles. Furthermore, they would not be compatible with PNS bacteria in co-culture, which produce H₂ optimally at ca. 30°C and live at temperatures below 47°C (Castenholz 1995). A hypothetical industrial-scale facility based on the sequential combination of a thermobioreactor and a PBR was estimated to produce H₂ at a cost of €2.74/kg H₂ (de Vrije and Claassen 2003). If the bio-H₂ were used to generate electricity in a fuel cell operating at 50% efficiency and 95% utilisation, the cost of the energy production would be €0.145/kWh. The price of domestic electricity is ca. €0.25–0.30/kWh (2008, www.britishgas.co.uk). However, a more extensive economic analysis of the bioprocess, suggested a cost of €4/kg H₂ (Davila-Vazquez et al. 2008a). Continued research and development of bio-hydrogen systems is required preceding a reliable economic assessment.

Some information is available regarding the use of non-axenic fermentations in the 1st stage of a dual system (Table 1). This strategy takes advantage of the presence of suitable microorganisms in the feedstock, thus eliminating the need to sterilise inputs and to pre-culture specific organisms. However, in undefined mixtures of microorganisms it is more difficult to repress unwanted reactions (Sect. 2.2.2), H₂ yields are generally lower than in axenic dual systems (Table 1) and it would be difficult to ensure reproducibility between feedstock sources and locations.

3.4.2.2 Distribution of H₂ production among stages of a dual system

Ike et al. (1997) compared three different methods of H₂ generation from algal biomass rich in starch, of which the most effective (in terms of H₂ yield) involved a lactic acid fermentation (producing no H₂) followed by photo-fermentation with PNS bacteria, showing that it is not essential to produce H₂ in both phases of the dual system.

Figure 2 illustrates that various different stage 1 fermentations (e.g. lactic acid, mixed-acid and clostridial-type fermentations) can be applied with different H₂ yields but with equal potential for H₂ production overall (12 mol H₂/mol hexose). The type of fermentation employed affects the theoretical distribution of H₂ between the 1st and 2nd stages. If lactic acid fermentation were used, all 12 moles of H₂ would arise from the 2nd stage; 10 for mixed-acid fermentation and 8–10 for C. butyricum.

It is possible that C. butyricum-based dual systems have been favoured by researchers in order to skew the distribution of H₂ production towards the 1st stage and thus to minimise the required transfer of organic fermentation products and the light capture area. Conversely, H₂ is generally produced by fermentations at ca. 50% of the theoretical maximum, while the photo-fermentation typically operates at ca. 70% efficiency, so a higher overall yield could be expected using lactic acid or mixed-acid fermentation in which more H₂ is produced in the more efficient 2nd stage. The gain in productivity should be offset against the increased costs of light capture area and transfer of fermentation products.

3.4.2.3 Energy generation potential with a dark fermentative 1st stage

The increased productivity of a dual system over a single-stage system is significant. For example, a molasses-fed pilot fermentation plant generated 8,240 l H₂ (342.5 mol H₂) and 3,000 l effluent/day (Ren et al. 2006). The effluent contained primarily acetate and ethanol with a hydrogen production potential of 246.35 mmol H₂/l (authors’ calculations). Therefore the addition of a photoheterotrophic 2nd stage could maximally increase productivity by 317%.

As a best-case scenario, a dual system capable of generating 12 mol H₂/mol hexose might be developed. Work is underway to meet a target of 10 mol H₂/mol hexose, which would make bio-H₂ economically viable, given low feedstock costs (Davila-Vazquez et al. 2008a). Using reported values for the productivity of dual systems, the potential for energy generation can be estimated.

With the same assumptions as used in Table 3, a household might consume H₂ at a minimum rate of 573.6 mol/day (Levin et al. 2004; Levin 2004). The feasibility of the decentralised application of a sequential dual system was evaluated (in this study) by calculating the necessary reactor sizes and the feed requirements to meet this demand. The energy requirements of the process were not taken into account.

If the potential H₂ yield (12 mol H₂/mol hexose) were to be distributed 4:8 between the 1st and 2nd stages, respectively, then the dark fermentation would be required to produce 191.2 mol H₂/day and the PBR 382.4 mol H₂/day (of which only ca. 12 h is light). Using published volumetric productivities (Levin et al. 2004; Levin 2004), it was calculated that a 79.6 l fermenter (containing an undefined mesophilic culture, ca. 0.1 mol H₂/l/h) would be needed. The productivity of the PBR would be constrained by several parameters: the limited efficiency of light conversion to H₂ (10% maximum; Akkerman et al. 2002), light availability (1 kW/m² for 12 h/day; Miyake et al. 1999), specific rate of H₂ production (ca. 0.1 l H₂/g dw/h; Rocha et al. 2001), culture depth (ca. 5 cm), culture density (ca. 1 g dw/l: OD ca. 2.5). To operate within these constraints a PBR volume of 7,648 l would be required. The corresponding square panel PBR could be 12.4 m wide with a light capture area of 153.0 m² and a depth of 5 cm; sufficiently shallow to maintain the entire culture under saturating light intensity (Nakada et al. 1995; Katsuda et al. 2000). This area could potentially harvest 6.6 GJ per 12 h light period, indicating a comfortable light conversion efficiency of 1.65% to meet the H₂ demand (Fig. 4).

Fig. 4

Spatial feasibility of de-centralized energy generation. The cartoon depicts one possible configuration of a sequential dual system combining dark fermentation and PNS bacteria. Detailed explanation is given in Sect. 3.4.2.3

It is noteworthy that the limiting factor is the specific rate of H₂ production (necessitating a dense culture and limiting the PBR depth) rather than the light conversion efficiency even at the reasonable light intensity of 1 kW/m². Were specific rates to be increased (e.g. through improved strains or bioreactors) the light intensity and conversion efficiency would then limit the productivity of the PBR. At a light intensity of 1 kW/m² for 12 h/day, with 65.8% of useful solar energy and a light conversion efficiency of 10% the corresponding light capture area to produce 382.4 mol H₂ in 12 h would be of 38.44 m² (6.2 m × 6.2 m). A house might barely accommodate an 80 l fermenter and a 40 m² PBR, along with H₂ storage, regulatory equipment and fuel cell (Fig. 4), but even with a conversion efficiency of 12 mol H₂/mol hexose, the feed-demand would be 7.74 kg cellulose/day or 8.17 kg sucrose/day, which could be supplemented with organic household wastes for disposal, although the additional sugar from this source would be negligible. Sugar production is an agricultural industry, so this option could not be applied in the long-term due to competition for farmland (Sect. 3.4.1.1) but wastes from sugar production and processing could be exploited as feed substrates.

Whereas the cultivation of energy-dedicated crops to generate sugars for conversion to H₂ would incur costs, it could be economically realistic to co-locate H₂ production with feed sources such as food processing plants. The UK food industry produces ca. 5.3 million tonnes of biodegradable waste annually, a large fraction of which is disposed of by land-filling, incurring both economic and environmental costs (Anon 2004; Bartelings et al. 2005). Co-locating food-waste generation and conversion to H₂ would remove transport and disposal costs, while minimising spoilage of the residues to maintain their value. Bio-H₂ production could be optimised for the use of residues having relatively consistent composition and little H₂ storage or distribution would be required as produced energy could be used on-site to meet predictable energy demands and any excess production could be sold to alleviate the demand for fossil fuel.

There are many accounts of bio-hydrogen production from non-synthetic substrates (i.e. wastes) and dual systems have been applied in several cases (Table 1) (Zhu et al. 1995, 2002; Fascetti et al. 1998; Kim et al. 2001; Franchi et al. 2004). De Vrije and Claassen (2003) described a hypothetical process fed by lignocellulosic biomass and calculated that 9% of the domestic energy demand could be met using available biomass residues in the Netherlands. However, the authors are not aware of any currently operating economically viable bioprocess based on lignocellulosic feedstocks.

3.4.2.4 Bio-H₂, bio-methane or bio-ethanol?

Biomass residues are available in significant quantities for use as feedstocks for bioenergy production (Easterly and Burnham 1996; Filho and Badr 2004; Haq and Easterly 2006; Mabee et al. 2006; Dawson and Boopathy 2007; Levin et al. 2007). Bioprocesses for the production of H₂, methane and ethanol can all utilise biomass residues as feeds, although currently, bio-ethanol and bio-methane processes are commercially more advanced than bio-H₂ processes.

Levin et al. (2007) calculated the energy potential of Canada’s biomass residues for methane production by anaerobic digestion and H₂ production by anaerobic bacterial fermentation. The potential H₂ energy equated to only 41.4% of the potential methane energy. However, this calculation was based on a H₂ yield of 1.3 mol H₂/mol hexose from a single-stage bacterial fermentation. Several authors report multi-organism systems for H₂ production producing in excess of 7 mol H₂/mol hexose (Weetall et al. 1989; Miura et al. 1992; Ike et al. 2001; Kawaguchi et al. 2001; Yokoi et al. 2001; Asada et al. 2006; Kim et al. 2006c). Hence, a bio-H₂ process could be more energetically productive than a bio-methane process if dual H₂-producing systems could be implemented. With H₂ as the final product, bio-methane could be converted by steam reforming, which is a well-established thermocatalytic process. However, this would introduce significant parasitic energy demand due to the required high temperature (ca. 700°C). Furthermore reformed H₂ is contaminated with CO (a potent catalyst poison) and would require extensive purification in order to be made comparable in quality to Bio-H₂. Therefore, a dual H₂-producing system would offer the comparative advantages of significantly greater energetic productivity and applicability to fuel cells.

Bio-ethanol is a major energy vector in Brazil, with a production of 16 billion litres of ethanol annually, requiring ca. 3 million ha of land. The total sugarcane crop area (for sugar and ethanol) is 5.6 million ha (Goldemberg 2007). The average industrial yield from the crop of 2004/2005 was 144.35 kg sucrose/tonne sugarcane, equivalent to 79.39 l of anhydrous ethanol/tonne sugarcane or 82.86 l hydrous ethanol/tonne sugarcane (Nastari et al. 2005). Therefore, the process efficiency of bio-ethanol production is 80.5% (of a biochemical maximum of 2 mol ethanol/mol hexose). Considering the higher heating value (HHV) of ethanol of 29.840 MJ/kg (http://hydrogen.pnl.gov/cocoon/morf/hydrogen/article/401), the bio-ethanol process produces 2,212 kJ energy/mol hexose (from sugar cane). To equal this energy yield a bio-hydrogen process must achieve ca. 7.8 mol H₂/mol hexose (HHV of H₂ = 141.88 MJ/kg, hydrogen.pnl.gov/cocoon/morf/hydrogen/article/401). This could not be achieved by a single-organism system and a dual system would be required.

4 Conclusions and future perspectives

Biological hydrogen production is a promising avenue that should be pursued urgently as the world energy demand increases, fossil fuel resources dwindle and the need for greenhouse gas minimisation becomes increasingly pressing. Hydrogen biotechnology is poised to become increasingly prominent alongside, and eventually emerging as competitive with other sustainable bio-fuel processes and/or as an adjunct to them.

This review shows that (unlike with bio-ethanol production) no single microorganism can produce competitive yields of H₂. Multiple-organism systems offer increased H₂ yields and would be mandatory for realistic future energy generation.

Examination of the properties of photosynthetic microorganisms revealed that PNS bacteria are the most suitable organisms for the 2nd stage of a dual system, while for the 1st stage dark fermentation, clostridia have been the most widely used, but facultative aerobes may increase the ease of operation while detracting little from the overall H₂ yield.

A dual system combining anaerobic fermentation and photoheterotrophy could potentially result in high energy yields from industrial wastes or biomass residues, although it is unlikely that a domestic household would produce sufficient fermentable waste to make a significant contribution to its energy budget. For example, the power output of a pilot fermentation plant neared the demand of 1 household, processing daily ca. 4.4 tonne of diluted feedstock (molasses, 3 g COD/l) to generate 343 mol H₂ (Ren et al. 2006), sufficient to produce a constant electrical power output of 0.6 kW using a realistic PEM-FC (operating at 50% efficiency and 95% utilisation; Levin et al. 2004).

Even by increasing the output by several-fold by addition of the second stage PBR it is unlikely that a light-driven dual system would repay investment for single household domestic electricity generation. Furthermore, it can be calculated that for domestic self-sufficiency several tonnes of sugary waste would be required annually, therefore, substrate supply would be the limiting factor rather than spatial considerations. Hence, industrial, retail and agricultural waste producers would be the likely initial users of bio-H₂ systems.

In addition to food processing and retailing wastes, biomass residues are available in significant quantities for use as feedstocks for bioenergy production (Easterly and Burnham 1996; Filho and Badr 2004; Haq and Easterly 2006; Mabee et al. 2006; Levin et al. 2007). Bioprocesses for the production of H₂, methane and ethanol can all utilise biomass residues as feedstocks although, currently, bio-ethanol and bio-methane processes are commercially more advanced than bio-H₂ processes.

It was argued before (Sect. 3.4.2.4) that a dual bio-H₂ system could be more productive than bio-methane and equally productive to bio-ethanol in terms of energy production, but these calculations were based on two noteworthy assumptions. Firstly, this calculation did not take into account parasitic energy losses, such as the distillation cost incurred by ethanol recovery (ca. 25–33% of the product’s combustion value; Kvaalen et al. 2006) or the operating costs of relatively complex dual systems for H₂ production, which cannot be estimated due to the immaturity of this technology. Secondly the assumption was made that energy can be recovered from H₂ and CH₄ with equal efficiency, e.g. by coupling of the bio-gas-producing generation reactor to a fuel cell for electricity generation assuming a pure gas stream (e.g. see Macaskie et al. 2005). The most efficient type, proton exchange membrane (PEM; also called polymer electrolyte membrane) fuel cells, achieve the highest power densities when H₂ is used as a fuel, whereas solid oxide fuel cells (SOFCs) are more suitable for the use of hydrocarbons such as methane (Larminie and Dicks 2003). However, SOFCs use an oxide ion-conducting ceramic material as the electrolyte and require an operating temperature of 600–1,100°C and hence the necessary heat input detracts from the overall energy balance. In addition, any contamination of bio-methane with H₂S, the end product of dissimilatory sulfate metabolism by the sulfate-reducing bacteria present in anaerobic mixed cultures, would necessitate gas filtration, since sulfur compounds are powerful catalyst poisons affecting all types of fuel cells. The direct methanol fuel cell (DMFC) is a type of PEM-FC in which methanol reacts (slowly) at the anode according to: CH₃OH + H₂O → 6H⁺ + 6e⁻ + CO₂ (Larminie and Dicks 2003). For DMFC, the power density is relatively low compared to PEM-FC and SOFC, but this would not prevent application in portable devices such as laptop computers, where the power storage exceeds 0.025 kWh and the required DMFC unit would be significantly smaller in volume than the equivalent lithium-ion battery (see Larminie and Dicks 2003).

The formation of methanol from methane via methane monooxygenase is very well established (Grosse et al. 1999; Dalton 2005) and a comparative study of the various bio-gas and fuel cell-coupling options would be worthwhile.

Fuels which are liquid at ambient temperature (e.g. methanol and ethanol) have higher volumetric energy densities than gaseous H₂. Apart from the consideration of land use the long-term economics of bio-ethanol production should be considered (Rogers et al. 2005). Ethanol cannot be used efficiently in fuel cells (Larminie and Dicks 2003) and a significant problem is considered to be the higher cost of bio-ethanol production (from cellulosic biomass) as compared to diesel or petrol. The price of bio-ethanol was projected to become comparable to that of petrol by 2015, based on a conservative forecasted price of $35–$40 per barrel crude oil (Chandel et al. 2007), but this may occur sooner due to spiralling oil prices ($135 per barrel in May 2008 and climbing; www.bloomberg.com). The distillation cost of ethanol is significantly higher at low ethanol concentrations (Zacchi and Axelsson 1989) and a membrane distillation process can be used as an efficient and cost effective option (see Chandel et al. 2007); molecular sieve techniques are now widely used in the industry (Rogers et al. 2005). A net energy balance (NEB) calculated by Hill et al. (2006) suggested that corn grain ethanol provides ca. 25% more energy than that consumed in its production; however, almost all of the NEB can be attributed to the ‘energy credit’ for the animal feed co-product.

Such calculations are moderated according to the geographic region. Hence, Brazil has certain comparative advantages in ethanol production. Unlike American or European processes based on crops (e.g. barley, corn or wheat) that must first be converted at significant expense into fermentable sugars, Brazilian (and also Australian) processes are based on sugarcane which the climate favours, obviating any need for conversion. Ethanol produced from sugarcane in Brazil has a net positive energy balance (renewable energy output versus fossil fuel input) of 10.2, whilst the equivalent value for ethanol from corn (US) is 1.4 (Goldemberg 2007). Also, the production cost of ethanol from sugarcane (Brazil) ($0.81 per gallon, in 2006) is lower compared to ethanol from corn (US) ($1.03 per gallon, in 2006) and is competitive with gasoline in the US, even considering the import duty of $0.54 per gallon and energy-efficiency penalties (30% or less with modern flexible fuel vehicle technologies) (Goldemberg 2007). Ethanol produced in Brazil has remained fully competitive with gasoline on the international markets, without government intervention, since 2004, i.e. subsidising ethanol production is a thing of the past. In addition to the production of ethanol, the industrial processing of sugarcane generates bagasse, a valuable product which adds to the industry’s positive environmental differential because it has been used to replace fossil fuels in the production of industrial heat and electricity in the sugar mills and distilleries, thereby boosting the abatement potential of greenhouse gases emission (Macedo et al. 2004). Moreover, the competition for land use between food and fuel has not been substantial: sugarcane covers 10% of total cultivated land but only 1% of total land available for agriculture in Brazil (Goldemberg 2007).

A recent review (Hill et al. 2006) has evaluated critically the long-term potential for bio-ethanol against emerging bio-diesel technology. While ethanol is made by the fermentation of biomass substrates (cane sugar is ideal because no further processing is required), bio-diesel is made via processing of plant material from ‘energy crops’. For example, soybean bio-diesel is sourced directly from long-chain triglycerides obtained from the seeds; in comparison corn-starch requires pre-enzymatic conversion into fermentable sugars for ethanol production. Critically, bio-diesel yields 93% more energy than that invested in its production and, relative to the fossil fuels they displace, greenhouse gas emissions are reduced by 12% and by 41% by bio-ethanol and bio-diesel, respectively. However, Hill et al. (2006) point out that ‘even dedicating all US corn and soybean production to bio-fuels would meet only 12% of gasoline demand and 6% of diesel demand’. Lin and Tanaka (2006) suggest that any country with a significant agronomic-based economy could use technology for ethanol fuel production. However, this and many other critiques overlook the difficulty of achieving a positive energy balance for the production of bio-ethanol from crops such as corn and wheat rather than from sugarcane. Whereas the result of recent studies was a positive energy balance of 1.4 (Goldemberg 2007), the cultivation area needed to support a US fuel economy based on corn-ethanol would equate to most of the nation’s land area (Pimentel 2001) and the same argument applies to all ‘energy crops’ that compete for agricultural land with food supply, a very major factor, which is acknowledged but understated by Hill et al. (2006). These authors suggest the use of agriculturally marginal land or the use of waste biomass for bio-ethanol production; both are potentially more sustainable than outright energy crop cultivation. Assuming that the cost of ethanol recovery can be lowered by effective recovery technology, the use of large global reserves of lignocellulosic waste biomass as potentially fermentable feedstock is receiving widespread attention with respect to bio-ethanol production and also with respect to bio-H₂ production (de Vrije and Claassen 2003; Aldhous 2006; Schnoor 2006). The main problem lies in converting the recalcitrant woody material into readily fermentable substrate. This requires pre-treatment, which may be physico-chemical, enzymatic or combinations of these. An overview of upstream treatments is outside the scope of this review and the reader is referred to recent example reviews in this area (Rogers et al. 2005; Lin and Tanaka 2006; Chandel et al. 2007). Once a fermentable feedstock is generated there are several options for the downstream energy production process and the hydrolysate could be equally well used for bio-hydrogen production as for bio-ethanol production, without the attendant processing costs.

The use of energy crops for bio-diesel production is particularly promising and the technological limitations have been reviewed by Abdullah et al. (2007). Chemically, ‘bio-diesel’ is fatty acid methyl esters, produced by the transesterification of oils and fats with methanol in the presence of suitable catalysts. Here bio-methanol could find a large-scale application as an alternative to the niche market for fuel cell use. The disadvantages of bio-diesel production are that large volumes of contaminated wastewater are produced and that homogeneous catalysis is employed for maximum processing efficiency; the catalyst is currently not retained and major research efforts are directed towards the development of solid phase catalysts (Abdullah et al. 2007). Glycerol is produced in tonnage quantities as a by-product, which could be a suitable substrate for microbial fermentation to produce ethanol or H₂ as additional energy products. However, the glycerol is obtained as an aqueous impure NaCl-solution which requires purification and its use as a fermentation substrate would compete with other potential uses in the pharmaceutical, cosmetic and food industries and as animal feeds, polymers, surfactants and lubricants (Ma and Hanna 1999). Assuming that microorganisms resistant to the contaminants are developed, bio-hydrogen production could be attractive in this context since a gaseous product is easily separated from the fermentation liquor and hence the purity or otherwise of the starting material is largely irrelevant, assuming the product gas stream is free of volatile agents.

Bio-diesel is made from waste oils or by pressing plant material (e.g. seeds) to extract the oils and hence plant residua could be a useful waste for fermentation to make a secondary ethanol or hydrogen fuel stream; however, the problems of upstream treatment of the wastes are similar to those of other fibrous materials, although in at least one example waste from oil production (in this case olive oil) has been used as the substrate for bio-hydrogen production (Eroğlu et al. 2004). Clearly the use of edible oils from food-crops such as Olea spp. (olive) for bio-fuel production would be impractical, however attention has recently focused on the use of inedible oil for bio-diesel production, obtained from the tropical oil seed plant Jatropha curcas, which is drought-resistant and can grow on marginal, sub-arable or even waste land (Srivastava and Prasad 2000) in central and south America, Mexico, south-east Asia, India and Africa. J. curcas is not suitable as animal feed without detoxification (Martínez-Herrera et al. 2006) but has many other applications and a transesterification process of the seed oil as a bio-fuel has been evaluated on an industrial scale (1,500 tonnes per annum, Gübitz et al. 1999). Due to its high free fatty acid content (ca. 14% w/w) Jatropha oil requires pre-esterification using methanol before conventional transesterification to produce bio-diesel, which was shown to give a high yield of bio-diesel with satisfactory fuel properties (Tiwari et al. 2007). Bio-methane production from anaerobic digestion is a potential source of bio-methanol, which could find use in the pre-esterification reaction. Crushing Jatropha seeds to release the oil results in an equal mass of press-cake, which can be used as a substrate for further bioprocessing, e.g. methane production by anaerobic digestion (Staubmann et al. 1997) or, indeed, bio-H₂ production although this has not been attempted to date.

In conclusion, the production of bio-fuels (bio-diesel, bio-ethanol or bio-gases) from energy-dedicated crops appears to be unsustainable unless the plant occupies a niche other than agricultural land or provides a high yield of energy per area of cultivation. Agricultural residues (lignocellulosic biomass) are available as sources of fermentable substrate for bio-fuel production but the conversion of these wastes into fermentable substrate forms a common bottleneck. Bio-gases and bio-ethanol can both be made by the fermentation of sugars and sugary wastes but the processing costs of ethanol limit the energy output of this method.

The hydrogen economy per se is still some decades away but combination and hybrid technologies are appealing in the shorter term. Production of H₂ from food waste sources or from the wastes from bio-diesel production is potentially a clean and sustainable route to clean energy production.

Although the maximum yields of H₂ from sugar are being approached by fermentation this is only possible by the application of more than one microorganism. This review has attempted to identify the two-stage approaches by which maximum yields (and rates) of conversion can be obtained and it identifies that, as with energy crops, available land area for light capture is likely to be a major limiting factor in operation. Under-used, waste ground in sunny regions (as for Jatropha) could provide one solution but for most of the developed world arable land takes priority for food production. Process intensification is required to overcome the problem of light delivery to the second stage photofermentations, which would push bio-H₂ production to competitive levels. A review of PBR designs to achieve effective light transfer into high-activity cultures is outside the scope of this overview; the reader is referred to recent reviews (Tsygankov 2001; Hoekema et al. 2002, 2006; Kondo et al. 2002; Claassen and de Vrije 2007).