The potential of microbial processes for lignocellulosic biomass conversion to ethanol: a review

BACKGROUND: This paper assesses the feasibility of a single- or multi-stage process entirely based on microbial cultures, with no or minimal non-biological pretreatment and with no external enzyme addition, for the conversion of lignocellulosic materials into ethanol. The considered process involves three distinct microbial processes, which can be possibly combined in one single reaction stage: a) lignin hydrolysis; b) cellulose and hemicelluloses hydrolysis; c) glucose fermentation to ethanol. This paper critically reviews the literature on the three microbial processes and compares the rates of microbial processes with the ones of the alternative physico-chemical pretreatment processes.

RESULTS: There is a large number of microbial species that can perform each of the three processes required for the conversion of lignocellulosic biomass to ethanol, although only one species has been unquestionably reported, so far, to be able to hydrolyse lignin under anaerobic conditions; another challenge is controlling the anaerobic fermentation of glucose to ethanol with mixed cultures; the rates of the microbial processes reported so far in the literature are generally lower than the rates obtained with physico-chemical pretreatments.

CONCLUSIONS: While in principle the whole process from lignocellulosic biomass to bioethanol can be carried out with existing, non engineered microorganisms, there is need for further research to obtain rates and yields which are commercially attractive.

Keywords: bioethanol; cellulose; lignin; microbial hydrolysis; mixed cultures

* email: davidedionisi@abdn.ac.uk, phone: +44 (0)1224 272814

1. Introduction

“Second generation” bioethanol produced from lignocellulosic biomass is gaining increasing interest, since it can avoid the fuel vs. food competition which is intrinsic in “first generation” bioethanol produced from starch- or sugar-based materials such as corn or sugarcane.¹ Lignocellulosic materials such as the organic fraction of municipal solid wastes, agricultural wastes, forestry residues, etc, represent an almost unlimited resource, which could be potentially used for bioethanol production.² However, production of bioethanol from lignocellulosic materials poses significant technical and economical challenges. According to a UN panel of experts,³ in 2012 biofuels production from lignocellulosic feedstock accounted for some 140 million litres per year, only 0.15% of the total production. The world’s first commercial scale cellulosic ethanol plant is considered to be the one in Crescentino, Italy, by Beta Renawables, with a full capacity of 75 million litres a year, which was started up in 2013.⁴

The bioethanol production process from lignocellulosic materials usually consists of several steps: pretreatment to break the lignin structure and make cellulose (and/or hemicellulose) available for hydrolysis, hydrolysis to hydrolyse cellulose to glucose, fermentation to convert glucose into ethanol and separation processes for ethanol purification.^5,6 As far as lignin hydrolysis is concerned, the most widely adopted pretreatments are steam explosion, acid hydrolysis, or ammonia fibre expansion (AFEX).⁷ As an example, the Iogen demonstration plant (Canada), which has an average productivity of about 300 m³ ethanol/yr receives wheat straw, corn stover and bagasse as feedstocks and uses a modified steam explosion process to make cellulose more accessible for the following steps (Iogen, (http://www.iogen.ca/technology/cellulosic-ethanol.html)). These chemical or physical pretreatment processes usually give good lignin hydrolysis and they make cellulose and hemicelluloses free for the next hydrolysis stages. However, they require conditions of high temperature and/or high pressure, or the addition of significant amounts of chemicals. These conditions make these processes expensive and limit their economic attractiveness at commercial scale. Also, the severe operating conditions used in these processes may generate toxic substances which inhibit the following stages and therefore a detoxification stage is often necessary.⁸ The subsequent step of the process is cellulose and hemicellulose hydrolysis to generate glucose and other sugars. This step is usually carried out enzymatically, using commercially available or on-site produced cellulase enzymes. For example, the Iogen demonstration plant mentioned above and the Abengoa demonstration plant, (Abengoa, http://www.abengoa.com/web/en/innovacion/casos_exito/) which uses wheat and barley straw as feedstock, both use enzymatic hydrolysis. Finally, ethanol production from glucose is usually carried out using pure cultures of selected species, usually the yeast Saccharomyces cerevisiae. While this yeast allows obtaining a very high ethanol yield and high ethanol productivity, the use of pure cultures has some disadvantages: pure cultures usually have a very narrow substrate spectrum, e.g. non-engineered S. cerevisiae cannot metabolise xylose, the main component in hemicellulose, and this limits the range of feedstock that can be used; pure cultures of fermentative microorganisms usually cannot hydrolyse cellulose, and this requires the addition of the cellulose hydrolysis stage; the use of pure cultures requires sterilisation of the fermentation vessel and of all the process lines leading to them, and this causes additional costs.

All of the described factors contribute to the high cost of ethanol production from lignocellulosic biomass. This paper investigates, by critically analysing the relevant literature, the feasibility of an alternative process for ethanol production from lignocellulosic biomass. In contrast to the physico-chemical pretreatment and hydrolysis processes described above, the process investigated in this paper is entirely, or almost entirely, based on microbial processes, i.e. based on the physical contact of microorganisms with the substrate. In the investigated process, the three distinct processes required to convert lignocellulose to ethanol, i.e. lignin hydrolysis, cellulose and hemicellulose hydrolysis and glucose and other sugars fermentation, are all carried out by different microorganisms, which could co-exist in the same reactor, or could be present in different reactors in sequence. Obviously, the option of a single reactor where all the different microbial species co-exist would be preferable from an economic point of view. This would constitute an open (undefined) mixed culture, where the microorganisms responsible for the various processes are selected from a mixed-culture inoculum due to the applied process conditions (e.g. nature of the feedstock, residence time, pH, temperature). The process investigated in this study has many aspects in common with simultaneous saccharification and cofermentation (SSCF) and consolidated bioprocessing (CBP) processes, described by Lynd et al.⁹ In SSCF an aerobic reactor is used for cellulase production, and the produced cellulases are then used in a subsequent anaerobic reactor where cellulose hydrolysis and sugars (both hexoses and pentoses) fermentation to ethanol takes place. In CBP the production of cellulases and the fermentation of sugars take all place in the same anaerobic reactor. However, there are two main differences between SSCF and CBP process and the process investigated in this study (Figure 1): SSCF and CBP are usually thought to come after a chemical-physical pretreatment step for lignin hydrolysis, while the process considered here includes microbial lignin pretreatment; SSCF and CBP are usually thought as pure cultures processes, using native microorganisms or genetically engineered ones, while the process studied in this paper is aimed at using open mixed microbial cultures.

Microbial processes based on open mixed cultures are gaining increasing interest due to their lower cost and higher flexibility compared to the traditional pure culture processes.¹⁰ A well known industrial process involving the use of open mixed cultures is anaerobic digestion of organic materials to methane,¹¹ whereas more recently the use of mixed cultures to produce hydrogen,¹² ethanol¹³ and biodegradable plastics^14-16 has been investigated. An interesting mixed culture process is the MixAlco process,¹⁷ which converts biomass to carboxylic acids salts, which are then chemically converted to hydrocarbon fuels.

An entirely microbial process, if proven feasible, would have obvious cost advantages compared to existing processes for lignocellulosic ethanol production, due to the use of atmospheric pressure, temperature close to ambient, and no addition of expensive chemicals or enzymes. The aim of this paper is to review the literature on the individual processes that are necessary to obtain microbial conversion of lignocellulosic materials into bioethanol, i.e. microbial lignin and cellulose hydrolysis and glucose and xylose fermentation by mixed cultures. This review for the first time reports rate values from the literature studies and critically discusses the effect of process parameters, in order to help identifying the conditions that maximise the rates and yields of the microbial processes.

In principle, an alternative microorganism-based approach to convert lignocellulosic materials into ethanol is to use genetically modified microorganisms, as opposed to open mixed cultures of naturally occurring species. The rationale behind this is that the ability to hydrolyse lignin and cellulose/hemicelluloses and to ferment sugars to ethanol with high yields could be introduced into microorganisms using metabolic engineering techniques. Without attempting to review the vast area of metabolic engineering for ethanol production, which has been reviewed elsewhere,^18,19 this paper compares the use of open mixed cultures and of genetically modified microorganisms for lignin and cellulose hydrolysis and for glucose and xylose fermentation to ethanol.

2. Lignocellulosic biomass

The main components of lignocellulosic biomass are lignin, cellulose and hemicelluloses. Lignin is an aromatic polymer where the substituents are connected by ether and carbon-carbon linkages. The main building blocks in lignin are p-coumaryl alcohol (p-hydroxyphenyl propanol), coniferyl alcohol (guaiacyl propanol) and sinapyl alcohol (syringyl propanol).²⁰ The molecular weight of lignin is variable but is typically very high, 600-1000 kDa.²¹ Cellulose is a polysaccharide composed of (1-4) linked D-glucose units, with molecular weight up to >500 kDa.²² Hemicellulose is a polysaccharide composed of various carbohydrate monomers, mainly xylan, arabinose, mannose and glucose, present in different ratios in the various materials. The molecular weight of hemicelluloses is usually lower than the one of cellulose.²³ The composition of various lignocellulosic materials, potential feedstock for ethanol production, is reported in Table 1. Values in Table 1 are to be considered as orientative values only, because the measured lignin content in a given biomass species is highly dependent on the biomass history and on the measurement method used.²⁴

The hemicellulose content in the feedstock is important due to the fact that hemicellulose hydrolysis generates monosaccharides other than glucose, which need to be converted to ethanol in order to avoid yield loss in the process. The conversion of non-glucose sugars to ethanol poses significant challenges as discussed later in this paper. However, hemicellulose hydrolysis is not considered to be a significant challenge, since hemicellulose is hydrolysed more easily than cellulose,²⁵ and it is expected that a mixed microbial culture that hydrolyses cellulose would also be able to hydrolyse hemicelluloses.⁹ Therefore in this paper the hydrolysis of only lignin and cellulose is discussed.

3. Microbial degradation of lignocellulosic biomass

The aim of this section is to identify, on the basis of the existing literature, whether there are microorganisms which are able to catalyse the various steps required for lignocellulosic biomass conversion to ethanol, i.e. lignin hydrolysis, cellulose hydrolysis and sugars fermentation to ethanol. Both pure culture and mixed cultures of naturally existing microorganisms and genetically modified microorganisms are considered. The rates of microbial processes are reported and compared to the rates of the alternative chemical-physical processes.

3.1 Lignin

3.1.1 Anaerobic conditions

High molecular weight lignin is generally considered as non biodegradable, or only biodegradable at insignificant rates, in the absence of oxygen,^21,26,27 even though there is some evidence of the contrary.^28,29 On the other hand, there is enough evidence to suggest that the various lignin building blocks, constituted by the various aromatic compounds as monomers or oligomers, are readily metabolised under anaerobic conditions.^30,31 The very limited evidence of anaerobic lignin biodegradation reported in the literature can be attributed²¹ to non-lignin components or to low-molecular weight materials (<600 Da). However, Benner et al.²⁹ using mixed cultures, reported anaerobic lignin biodegradation rates of up to 37% of the aerobic rates, for several lignocellulosic marine or wetland plants. Silanikove and Brosh²⁸ reported 45-58% anaerobic metabolisation of lignin in wheat-straw, by the rumen bacteria in the goats’ gastrointestinal tract.

Only very recently, a convincing evidence of lignin degradation under anaerobic conditions has been presented.³² The authors observed that the majority (85%) of insoluble switchgrass biomass, that had not been previously chemically treated, was converted at 78°C by the anaerobic bacterium Caldicellulosiruptor bescii. Interestingly, the glucose/xylose/lignin ratio did not substantially change over the incubation period (3 successive cycles of 5 days each) providing an indication that the three major biomass components, including lignin, were solubilised and/or metabolized at comparable rates. A mass balance revealed that lignin was not assimilated, only carbohydrates served as carbon and energy sources. Lignin degradation was confirmed by gas-chromatography-mass spectrometry analyses which revealed the presence of lignin-derived aromatic compounds, such as syringylglycerol, guaiacylglycerol, and phenolic acids, in the spent culture broth.

Indirect evidence of lignin degradation/hydrolysis under anaerobic conditions can be obtained observing anaerobic digestion studies, which are always carried out using open mixed microbial cultures, with lignocellulosic materials as feedstock (Table 2). In general, methane production has been reported with many lignocellulosic materials, even though the possible lignin degradation and its extent are not usually measured. Methane production has been reported even with feedstocks with high lignin content, such as Mirabilis and Ipomoea fistulosa leaves³³ and woody biomass such as poplar.³⁴ Several studies investigated the effect of lignin content on the anaerobic degradability of lignocellulosic biomass. Triolo et al.³⁵ observed that the higher the lignin content in the raw material the lower the methane production. However, Tong et al.³⁶ found only a poor correlation between the lignin content of various lignocellulosic substrates and their methane production rate. This indicates that biodegradation of lignocellulosic substrates is not only affected by the lignin content, but also by other factors such degree of association between lignin and carbohydrates, cellulose crystallinity and others.³⁷ According to Turick et al.,³⁴ one of the reasons why many investigators observed only poor methane production from woody or other highly lignified biomass is that the time length of the tests is not long enough. These authors carried out tests for 100 days and observed substantial methane production from woody biomass. Interestingly, they observed that most of the methane production occurred after more than 50 days from the start of the test, and they explained this behaviour with the need of the microorganisms to become able to degrade lignin, making the cellulosic materials initially shielded by lignin available.

Overall, even though so far only limited evidence for anaerobic lignin metabolization or solubilisation is available, it is apparent that a significant fraction of the cellulose or hemicellulose carbohydrates in lignocellulosic materials becomes available during anaerobic digestion, and this indicates that likely lignin hydrolysis occurs under anaerobic conditions.

3.1.2 Aerobic conditions

In contrast to anaerobic conditions, there is wide literature evidence that lignin is biodegraded under aerobic conditions. Various species of bacteria and fungi have been reported to biodegrade lignin (Table 3). Fungi, in particular white-rot fungi, have been studied more extensively than bacteria for their lignin biodegradation ability, and they are generally considered more interesting than bacteria as a pretreatment of lignocellulosic materials at industrial scale.³⁸ Aerobic biodegradation of lignin has also been reported in mixed culture studies, usually carried out in composting environments (Table 4). While substantial lignin degradation has been reported for a wide range of lignocellulosic substrates, usually the treatment times are quite long and lignin degradation is not complete.

The literature evidence so far indicates that lignin cannot be used as a sole carbon and energy source, but requires an additional substrate to support microbial growth.²¹ The growth substrate can be the glucose or carbohydrates units contained in the cellulose or hemicellulose inside the lignin matrix,³⁹ or can be externally added. This is important from the process point of view, since it is expected that microbial lignin depolymerisation may require the use of part of the cellulose and hemicellulose as growth substrate for the microorganisms, making it not available for conversion to ethanol. However, in a mixed culture environment with real lignocellulosic biomass as feedstock, other carbon and energy sources than polysaccharides may be available (e.g. proteins) and so the loss of carbohydrates for ethanol production may be avoided.

3.1.3 Effect of pH

The optimum pH for aerobic lignin degradation by fungi is approx. 4.0-4.5 and significantly lower biodegradation rates are observed when the pH is lower than 4 or above 5.^21,40 pH in the range 4-5 is considered the optimum for the growth of most white-rot fungi. For bacteria, no difference in lignin degradation rate was observed in the pH range 5.3-7.8, for a mixed culture.⁴¹ These studies seem to indicate that the optimum pH for lignin biodegradation coincides with the usual optimum pH for microorganism growth on carbon substrates.

3.1.4 Effect of feedstock particle size

The biodegradation of lignin occurs extracellularly and by decreasing the particle size it is expected that the surface to volume ratio of lignocellulosic biomass increases, so an increase in lignin biodegradation rate is expected. Limited experimental investigation has been carried out however on the effect of feedstock particle size on lignin biodegradation and usually studies are carried out with a single feedstock size, which is in the mm^42-44 or cm^45,46 range or not reported.⁴⁷ Zimmerman and Broda⁴⁸ observed higher lignin degradation for straw pre-treated with a vibratory ball mill (particle size 2-5 m) than for straw ground with a blender (particle size 0.5-1 mm). Even though still limited, investigation on the effect of particle size has been carried out in the area of anaerobic digestion of lignocellulosic biomass. Mshandete et al.⁴⁹ observed that total fibre degradation during anaerobic digestion to methane increased from 31% to 70% when the feedstock was grinded to 2 mm fibres compared to untreated fibres (larger than 100 mm). Sharma et al.³³ observed that the quantity of biogas produced increased when the feedstock particle size was reduced. The range of sizes in their study was 0.1-30 mm. However, they did not investigate specifically lignin degradation.

3.1.5 Lignin degradation rates

The main limitation of microorganisms-based pretreatment processes for lignocellulosic biomass conversion to ethanol is considered to be the low rate. However, very limited direct information on the rate of lignin hydrolysis under aerobic conditions is present in the literature. Table 5 reports lignin hydrolysis rates using fungi under aerobic conditions, calculated by the authors of this paper on the basis of literature data. The maximum rate is approx 0.1 g/L/h. Under anaerobic conditions, the only evidence of lignin degradation³² gives a lignin degradation rate of 0.012 g/L/h. It is important to observe that the rates reported in Table 5 have been obtained at lab scale with very small volumes and under non-optimised conditions. Several variables could be in principle optimised to maximise the lignin degradation rate: biomass concentration, oxygen concentration, pH and particle size of the feedstock.

It is worth comparing the lignin degradation rates reported in Table 5 with the rates obtained with non-biological pretreatment stages reported in the literature⁷ (Table 6). It is evident that the lignin degradation rates obtained with fungi are in general at least one order of magnitude lower than the ones obtained with chemical pretreatments. However, the process conditions required by the chemical pretreatments are much more severe, with much higher temperature and usually (with the exception of the hot water treatment) with the addition of chemicals, which obviously cause higher process costs.

3.1.6 Lignin hydrolysis by genetically modified microorganisms

While the focus of genetic engineering for lignin hydrolysis has been on genetically manipulating lignin biosynthesis in plants in order to reduce lignin content and make its hydrolsysis easier,^50,51 there are so far no reported attempts to genetically modify microorganisms in order to make them capable of lignin hydrolysis.⁵² The reason for this is probably the large number of enzymes which are potentially dedicated to lignin hydrolysis in naturally occurring lignin-hydrolysing microorganisms, such as the white-rot fungus Phanerochaete chrysosporium.^53,54 It is possible, however, that only a few out of the whole spectrum of lignases might be needed in industrial processes,⁵² therefore making genetic engineering of microorganisms for lignin hydrolysis a more feasible option.

3.2. Cellulose

Unlike the case of lignin, a wide range of bacteria and fungi have been reported to hydrolyse cellulose under anaerobic or aerobic conditions.

3.2.1 Anaerobic conditions

Table 7 summarises bacteria and fungi that have been reported to hydrolyse cellulose under anaerobic conditions. Unlike the case of lignin, complete cellulose hydrolysis can be obtained under anaerobic conditions, provided that the contact or residence time is adequate. Table 8 reports literature evidence for cellulose degradation by mixed cultures under anaerobic conditions, confirming that virtually complete cellulose hydrolysis is possible. Most of the data in Table 8 refer to batch tests with an unacclimated inoculum, and this explains the long time required for cellulose degradation.

The enzyme groups responsible for cellulose hydrolysis are very similar under anaerobic and aerobic conditions but the spatial arrangement of the enzymes can be different.⁹ Under anaerobic conditions cellulolytic enzymes are often bound to the external membrane of the cell, even though in some cases they are present as free enzymes in the liquid medium. Under aerobic conditions the enzymes are usually excreted in the liquid medium and are not attached to the cell membrane.

3.2.2 Aerobic conditions

Table 9 summarises microorganisms which have been reported to hydrolyse cellulose under aerobic conditions. Consistent with findings for anaerobic conditions, virtually complete cellulose hydrolysis can be obtained under aerobic conditions. Similar evidence is obtained for mixed cultures studies (Table 10), which usually refer to composting environments. An interesting observation is that usually the rate of cellulose hydrolysis is comparable under anaerobic and aerobic conditions.⁹ In terms of maximising the rate of cellulose hydrolysis, this means that no preference should be given to aerobic compared to anaerobic conditions.

3.2.3 Effect of pH

For anaerobic and aerobic bacteria the optimum pH for cellulose hydrolysis is usually in the range 6.5-8.0. For anaerobic bacteria, Shi and Weimer⁵⁵ found an optimum pH of 6.5 for cellulose hydrolysis with Ruminococcus flavefaciens, and Weimer⁵⁶ with Fibrobacter succinogenes found very little influence of pH on cellulose hydrolysis in the pH range 6.1-6.8. Using ruminal microbes under anaerobic conditions Hu et al.^57,58 found no cellulose hydrolysis at pH<5.5 and a very low rate at pH<6.0, the optimum pH being 7.0-7.5. Berquist et al.⁵⁹ reviewed various cellulolytic thermophilic bacteria, employing either aerobic or anaerobic conditions, and reported an optimum pH in the range 7.0-8.1 in most cases. This reported evidence is consistent with a study⁶⁰ on the anaerobic hydrolysis of organic waste, partially composed of lignocellulosic biomass, where an approximately 3-fold increase in the hydrolysis rate was observed when the pH was increased from 5 to 7.

3.2.4 Effect of particle size

In general, as cellulose hydrolysis is dependent on the contact between the solid cellulose and either the microorganisms or the excreted cellulolytic enzymes, it is expected that the rate of cellulose hydrolysis should increase as cellulose particle size decreases, since the area per unit volume increases. However, the beneficial effect of feedstock particle size reduction is expected to depend on the substrate-to-microorganisms ratio, as well as on the particle size.⁶¹ A few experimental studies have been conducted on the effect of particle size on cellulose hydrolysis rate. Weimer et al.⁶² observed a decrease in the hydrolysis rate and an increase in the induction time when the cellulose particle size increased. Similarly, Hu et al.⁵⁷ reported faster cellulose hydrolysis rate for 50 than for 100 m sized particles. Similar evidence was obtained in continuous studies. Chyi and Dague⁶³ observed a faster hydrolysis rate with 20 than with 50 m particles.

3.2.5 Cellulose hydrolysis rates

Table 11 summarises microbial cellulose hydrolysis rates, calculated by the authors of this paper on the basis of literature data. The reported rates for microbial cellulose hydrolysis are, in general, higher than the corresponding rates for lignin hydrolysis (Table 5), therefore indicating that the critical stage in the process is the lignin hydrolysis. Cellulose hydrolysis rates up to about 0.3 g cellulose/L/h have been reported, both under aerobic and anaerobic conditions. As observed for lignin degradation rates, the rates reported in Table 11 have usually been obtained with very low volume systems and have not been optimised. Important factors that can increase the rates of microbial cellulose hydrolysis are: biomass concentration, cellulose concentration, reactor dilution rate, temperature, pH and cellulose particle size, as discussed in previous sections. One of the most successful technologies for cellulose hydrolysis is enzymatic hydrolysis, where the cellulolytic enzymes are externally generated and added to the liquid mixture. Table 12 reports typical rates for enzymatic cellulose hydrolysis. It is evident that enzymatic hydrolysis is generally faster than microbial hydrolysis. However it has to be taken into account that enzymatic hydrolysis has a higher capital and/or operational cost than microbial hydrolysis due to the need for external generation or purchase of the enzymes.

3.2.6 Cellulose hydrolysis by genetically modified microorganisms

In recent years there has been a considerable interest in engineering microorganisms in order to make them capable of hydrolysing cellulose.⁹ Particular attention has been given to adding the cellulose-hydrolysis capability in microorganisms which are naturally able to ferment glucose to ethanol with high yields. Several cellulase-encoding genes have been expressed in various bacteria and yeasts, such as Zymomonas mobilis,⁶⁴ Klebsiella oxytoca,^{65, 66} Saccharomyces cerevisiae⁶⁷ and others. However, while the results are in general promising and encourage further research in this area, there is still no evidence that a cellulose hydrolysis capability at rates which are high enough for a commercial process has been engineered in genetically modified microorganisms. Most of the engineered microorganisms reported so far have gained the capability of hydrolysing cellulose derivatives but not native or crystalline cellulose. Genetically modified K. oxytoca exhibited good capability to hydrolyse soluble carboxy-methyl cellulose⁶⁵ or phosphoric acid-swollen Avicel,⁶⁶ but very limited ability to hydrolyse crystalline Sigmacell 50.⁶⁵ The yeast S. cerevisiae, expressing cellulases from Bacillus species, has been reported⁶⁷ to have activity on filter paper but was not able to grow in the absence of externally-added cellulases. The same yeast has been engineered to hydrolyse phosphoric acid-swollen cellulose. ^{68, 69}

3.3. Fermentation of carbohydrates to ethanol by mixed microbial cultures

Once cellulose and hemicellose have been hydrolysed, monomeric sugars need to be fermented to ethanol. The main sugars that are present in the hydrolysis products of lignocellulosic biomass are glucose, present in cellulose and in minor fractions in hemicellulose, and xylose, which is often the main component of hemicellulose (Table 1). The fermentation of glucose and xylose by mixed microbial cultures is reviewed in sections 3.3.1 and 3.3.2, respectively, while section 3.3.3 covers the use of genetically modified microorganisms. Other sugars are also present in the hydrolysis products of lignocellulosic biomass, e.g. arabinose, mannose, galactose, but usually in lower amounts than glucose and xylose and their fermentation is not discussed here. It is worth observing that various studies have been carried out on the anaerobic fermentation of arabinose by mixed cultures, however they were mainly aimed at hydrogen and not ethanol production (e.g. ^70,71). Fermentation of arabinose to ethanol has been mainly investigated by means of genetically modified microorganisms(e.g. ^72,73).

3.3.1 Fermentation of glucose

In an anaerobic mixed microbial culture glucose can be fermented to several different end products, as summarised in Figure 2. Ethanol can be produced directly from glucose, and then be converted to acetate or organic acids, which can then be converted to methane. Lactate, butyrate and acetate can also be produced directly from glucose through microbial action under anaerobic conditions (Figure 2). Conversely, propionate typically derives from the conversion of lactate; under methanogenic conditions propionate, once produced, can be further converted into acetate and H₂ (provided that methanogens keep the H₂ partial pressure below 10^-5 atm).

Table 13 lists some bacterial or fungal species which are able to convert glucose to ethanol and some species which are able to convert ethanol to organic acids. The stoichiometry of the key reactions hereafter discussed, i.e. glucose fermentation to ethanol and ethanol conversion to acetic acid, are reported below:

ethanol production from glucose

ethanol conversion to acetic acid

If ethanol is the desired product, the operating conditions of the fermentation should be chosen in order to maximise the rate of the ethanol-producing reactions and to minimise the rate of ethanol consuming reactions. The anaerobic fermentation by mixed cultures to methane is a well known process and is widely used in industry.⁷⁴ Also, relatively wide attention has been given to the anaerobic fermentation to organic acids such as acetate, propionate and butyrate, since they are often found as intermediates in the fermentation to methane and can, undesirably, accumulate in the liquid medium. However, much less focus has been given to the mixed culture fermentation to ethanol. In the next sections, the available information on the effect of process operating conditions on the anaerobic fermentation of glucose to ethanol is reviewed.

3.3.1.1 Effect of pH

A few studies have investigated the effect of pH on the anaerobic fermentation of glucose to ethanol and, while it seems that pH has an important effect on ethanol production, there is still no clear evidence on the optimum pH range to drive the fermentation process towards ethanol, rather than acetate and methane. Based on thermodynamic considerations, Rodriguez et al.⁷⁵ predicted that acidic pH values, below 5.5, should favour ethanol production, while at pH values higher than 6.5, acetate should be the only product in the liquid medium. In agreement with the theory that conversion to ethanol is favoured by acid pH values, Ren et al.⁷⁶ found that in a continuous reactor in the pH range 4.3-4.9 ethanol concentration increased at lower pH, and in this pH range ethanol and acetate were always the main fermentation products. In the same paper, in a batch study in the pH range 3.0-5.5, the authors reported the highest ethanol concentration at pH 5.0, observing a much lower ethanol production at pH 5.5. In a continuous study in the pH range 4.0-7.0,⁷⁷ the highest ethanol concentration was found at pH 6.0, but in this case, ethanol was not the main fermentation product, the main products being butyrate and acetate. Hwang et al.⁷⁸ found that acetate and ethanol were the main fermentation products at pH 4.5-5.0, while at pH 5.0-6.0 propionate and acetate were the main products.

However, other studies found higher ethanol yield at neutral or basic pH values. Temudo et al.¹³ investigated anaerobic fermentation of glucose with mixed cultures in a chemostat in a range of pH 4-8.5. They found that in the pH range 6.25-8.5 acetate and ethanol were the main fermentation products, in approximately equal molar ratio, while in the pH range 4-5.5 very little ethanol was produced and acetate and butyrate were the main fermentation products. However, in this study the dilution rate at pH 4-5.5 was also different than the one at pH 6.25-8.5 and this could also have affected the results. In general agreement with their findings, Zoetemeyer et al.⁷⁹ found that acetate and ethanol were the main products of anaerobic fermentation at pH 8.0, while at pH values below 7, the main product became butyrate. In a chemostat study in the pH range 5-8,⁸⁰ the highest ethanol concentration was found at pH 8, but in this study ethanol was a minor fermentation product, the main ones being acetate and propionate.

Overall, analysis of the literature indicates that further study is needed to address the effect of pH on ethanol yield from anaerobic fermentation of glucose.

3.3.1.2 Effect of temperature

The effect of temperature on glucose fermentation to ethanol by mixed cultures is potentially particularly interesting. In general, the rates of all microorganism-mediated processes increase with temperature, up to the maximum temperature which is tolerable by the microorganisms. Microorganisms used in anaerobic fermentations can be classified as either mesophilic (optimum temperature <45 ^OC) or thermophilic (optimum temperature >45 ^OC). An interesting advantage of thermophilic over mesophilic bacteria when using mixed cultures for ethanol production is that among thermophilic bacteria there are many microorganisms which are able to convert glucose to ethanol but only very few which are able to oxidise ethanol to acetate or other organic acids.⁸¹ Therefore, it is expected that higher ethanol yields and rates might be obtained under thermophilic than mesophilic conditions.

Other advantages of thermophilic conditions (adapted from Wiegel)⁸¹ are the following:

a) lower use of the substrate for biomass production, therefore increasing ethanol yield;

b) pathogens do not grow at temperature higher than 60 ^OC;

c) since microbial processes generate heat, higher temperatures may be easier to maintain than lower ones;

d) ethanol can be continuously distilled from the fermentation vessel by using a moderate vacuum

On the other hand, thermodynamic calculations⁸² show that at higher temperatures the reactions that generate hydrogen become more favourable. Since glucose oxidation to acetate or butyrate and ethanol oxidation to acetate generate hydrogen, these reactions become more favourable at higher temperatures, potentially leading to higher ethanol loss.

The most comprehensive study on the effect of temperature on the acidogenic fermentation of glucose has been carried out by Zoetemeyer et al.⁸³ The authors operated a chemostat at pH 5.8 in the temperature range 20-60 ^OC. At temperatures up to 50 ^OC, butyrate and acetate were the main products, and the ethanol yield was quite low (0.10-0.20 mol ethanol/mol glucose). At 55 ^OC, on the other hand, ethanol was the main fermentation product, with a yield of 0.8 mol/mol glucose.

In general, analysis of the literature shows that the effect of temperature on anaerobic fermentation to ethanol is potentially very important and deserves further investigation.

3.3.1.3 Effect of hydrogen partial pressure

Hydrogen concentration in the liquid phase or hydrogen partial pressure in the gas phase (the two are proportional via Henry’s law), which can be manipulated by sparging with an inert gas or by changing the process pressure, is an important variable that can affect the spectrum of product distribution in anaerobic fermentation. The effect of hydrogen partial concentration is twofold: a) hydrogen levels affect the NADH/NAD ratio and therefore the feasibility of the biochemical pathways that determine product formation;⁸⁴ b) certain fermentation reactions which generate hydrogen (Figure 2) are close to the thermodynamic equilibrium and hydrogen concentration (as well as pH) can determine whether they are feasible or not.

The effect of hydrogen concentration on methane formation is well known: hydrogen concentration has to be maintained at very low values in order for the conversion of organic acids to acetate to occur, which is thermodynamically unfeasible at high hydrogen concentrations, and this require a close syntrophy between hydrogen producing and hydrogen consuming microorganisms.

However, when methane is not the desired product, very little is known on the effect of hydrogen concentration on the spectrum of product distribution. The biochemical model by Rodriguez et al.⁸⁴ predicts that hydrogen partial pressures above approx 0.4 atm should lead to butyrate as main fermentation product, while lower hydrogen pressures would give acetate. In those simulations, carried out at pH 7, no ethanol formation was predicted, since the model only predicted ethanol formation at acidic pH values. Considering the conversion of ethanol to acetate, this reaction is thermodynamically feasible, at pH 7, only for a hydrogen partial pressure lower than approx 0.15 atm.⁸⁵ Therefore, hydrogen pressures higher than this value should prevent ethanol oxidation to acetate and therefore decrease ethanol losses. The experimental study by Mizuno et al.,⁸⁶ showed that a reduction in hydrogen partial pressure from 0.5 to 0.05 atm increased the rate of hydrogen production by more than 50%, however very little effect was observed on the composition of the liquid effluent, the main products being acetic and butyric acids, with much lower amounts of ethanol.

3.3.1.4 Effect of solids retention time

The solids retention time is a critical parameter for glucose fermentation with mixed cultures. It is well known that the end-product of glucose fermentation by mixed cultures is methane, if the digestion time or residence time is long enough.^74,87,88 Therefore, glucose fermentation to ethanol has to be carried out at relatively short residence times. However, within the region of relatively short residence times, little systematic study has been carried out to investigate whether the residence time affects the distribution of fermentation products. Zoetemeyer et al.⁷⁹ investigated the effect of residence time in the range 1.5-10 h (at 30 ^OC) and they reported ethanol profiles for pH values of 5.69 and 6.44. They found that ethanol yield tended to increase with longer residence times at pH 5.69 (up to 0.3 mol/mol), while it tended to increase with shorter residence times at pH 6.44 (up to approx 0.2 mol/mol).

3.3.1.5 Rates and yields

Table 14 summarises ethanol production rates and yields in glucose fermentation studies by mixed cultures. Only studies where the main target was ethanol or acids production are considered here. It is evident that with mixed cultures ethanol yields of up to 0.8 mol ethanol/mol glucose have been obtained. The maximum theoretical yield of ethanol on glucose is 2 mol ethanol/mol glucose, assuming that all glucose is fermented to ethanol. However, this maximum yield achievable in practice is lower than this, due to the fact that some glucose is inevitably used for biomass growth. For the yeast Saccharomyces cerevisiae the ethanol yield on glucose is typically in the range 1.6-1.9 mol ethanol/mol glucose.⁸⁹ The lower ethanol yield obtained with mixed cultures is due to the fact that part of the glucose is fermented to other products, mainly acetate and in some cases other acids such as propionate and butyrate. In order to develop commercial processes for ethanol production with mixed cultures, the challenge is to determine process conditions that direct glucose fermentation to ethanol, minimising both glucose and ethanol conversion to organic acids. To this regard, it is important to understand which are the causes for the observed variability in ethanol yield under similar process conditions. As an example, at a residence time of 8 h, 30 ^OC, pH 6.25, Temudo et al.,¹³ observed a ethanol yield on glucose higher than 0.6 mol/mol, while under similar conditions (residence time about 7 h, 30 ^OC, pH 6.44) Zoetemeyer et al.⁷⁹ found negligible ethanol yield. The reasons for the different behaviour could be due to the presence or absence of nitrogen sparging, the use of different inocula, the start-up procedure, the glucose concentration in the feed, etc.

In terms of ethanol productivity, high ethanol production rates of up to 1.5 g/L/h have been reported with mixed cultures. This value is lower than ethanol productivity on glucose for Saccharomyces cerevisiae, 3-18 g/L/h.⁸⁹ However, considering that the literature studies reported in Table 14 were not specifically aimed at maximising ethanol productivity, and that ethanol productivity could be easily increased simply by increasing glucose concentration in the feed, it seems that, with more lab- or pilot-scale investigation, ethanol productivity from glucose with mixed cultures could reach the same or higher productivities currently obtained with industrial processes.

3.3.2 Fermentation of xylose

Fermentation of xylose is much less known than glucose fermentation, in particular as far as mixed cultures are concerned. In principle, the spectrum of substrates which can be obtained by anaerobic fermentation of xylose is similar to that can be obtained from glucose (Figure 2), even though the quantitative distribution of the products and the microbial species

involved may be different. The stoichiometry of xylose conversion to ethanol is the following:⁹⁰

The theoretical maximum yield of ethanol from xylose is 1.67 mol ethanol/mol xylose, i.e. virtually the same yield as glucose if expressed in mass terms (0.51 g ethanol/g xylose).

Table 15 reports several species of microorganisms which have been reported to convert xylose into ethanol. Certain microbial species are able to produce ethanol from xylose with almost the maximum yield, while other always generate other co-products, mainly acetate.^91,92 In terms of the considered process with mixed cultures, the operating conditions have to be found that maximise ethanol yield, minimising the formation of other fermentation by-products. However, while several recent studies have investigated the effect of operating conditions on xylose fermentation to hydrogen,^93-96 the only study which has investigated xylose conversion to ethanol by mixed cultures is the one by Temudo et al.⁹⁷ They compared chemostat cultures grown on xylose or glucose as only carbon sources comparing ethanol and acids production with the two substrates. They observed that the culture grown on xylose produced much less ethanol than the one grown on glucose (0.05 mol ethanol/mol xylose vs. 0.24 mol ethanol/mol glucose), the other main products being in both cases acetate and butyrate. However, interestingly ethanol yield on xylose increased much when xylose concentration in the feed increased from 4 to 10 g/l, from 0.05 to 0.69 mol ethanol/mol xylose (the yield of butyrate was correspondingly much lower), but the reason for this is not known. The authors also observed that the mixed culture grown solely on xylose was immediately able to metabolise glucose when this substrate was added, indicating that in a mixed culture with complex substrates such as real wastes, the same microorganisms may be able to metabolise both glucose and xylose.

3.3.3 Fermentation of glucose and xylose to ethanol by genetically modified microorganisms

In general, the reason behind metabolic engineering of microorganisms in order to produce ethanol from glucose and xylose is essentially to increase the range of substrates that can be potentially converted to ethanol at high yield by a single microorganism. Indeed, native strains of the yeast Saccharomyces cerevisiae are not able to utilise pentoses such as xylose, therefore limiting the range of feedstock that can be used for ethanol production. Other microorganisms such as Escherichia coli are on the other hand able to metabolise a wider range of substrates but the native strains don’t produce ethanol as main fermentation product.

The enteric microorganism Klebsiella oxytoca M5A1 converts xylose to ethanol, but also produces organic acids (acetic, lactic, succinic). By metabolic engineering Ohta et al.⁹⁸ increased the molar fraction of ethanol in the products of xylose fermentation from 62% to 90%. Similarly, using metabolic engineering on E. Coli KO11, Yomano et al.⁹⁹ obtained almost stoichiometric conversion of xylose to ethanol, with very minor production of organic acids. Other researchers used metabolic engineering to increase the ethanol yield from glucose in Lactobacillus sp.¹⁰⁰ Since the most common microorganism used in industrial bioethanol production is the yeast Saccharomyces cerevisiae, considerable effort has been dedicated to engineering this microorganism to metabolise xylose.¹⁰¹ In general, good success has been obtained, however the volumetric productivity obtained with recombinant S. cerevisiae on xylose is still significantly lower than the one of the native strain on glucose¹⁹. So far, the maximum ethanol productivity obtained for recombinant S. cerevisiae on xylose is 0.5 g/l/h (lab scale study). Table 16 reports ethanol production rates and yields from glucose and xylose by genetically modified microorganisms in selected literature studies. In general, volumetric productivities of up to 2 g ethanol/l/h and almost quantitative conversions of glucose and xylose to ethanol have been obtained, so indicating the success of genetic engineering in generating microorganisms able to convert multiple sugars to ethanol at high rate and yield.