The idea of performing the enzymatic hydrolysis and fermentation simultaneously was put forward by Gauss et al. in a patent from 1976 3. The authors stated that the glucose yield in a traditional separate enzymatic hydrolysis (using enzymes produced by the fungus Trichoderma reesei) was low, probably due to end-product inhibition of the hydrolysis by glucose and cellobiose. The authors could, however, show that they obtained a higher overall ethanol yield when using SSF, which they attributed to the removal of glucose and cellobiose by the fermentation, and the consequent release of end-product inhibition. The term SSF (the abbreviation SSF is often used also for solid state fermentation) was not used by the authors at the time, but became the common notation for this process within just a few years from the original invention. The avoidance of end-product inhibition is still probably the most important reason for using SSF, but there are several additional potential advantages. Gauss and co-workers, mentioned for instance the advantage that glucose does not need to be separated from the lignin fraction following a separate enzymatic hydrolysis step, thereby avoiding a potential loss of sugar. Furthermore, the combination of hydrolysis and fermentation decreases the number of vessels needed and thereby investment costs. The decrease in capital investment has been estimated to be larger than 20%. This is quite important, since the capital costs can be expected to be comparable to the raw material costs in ethanol production from lignocellulose 4. Other advantages, relating to co-consumption of pentose and hexose sugars, and detoxification have become apparent more recently, as will be discussed later in this review.

Inevitably, there are also disadvantages of SSF in comparison to the separate hydrolysis and fermentation (SHF) process. The optimum temperature for enzymatic hydrolysis is typically higher than that of fermentation – at least when using yeast as the fermenting organism. In an SHF process, the temperature for the enzymatic hydrolysis can be optimized independently from the fermentation temperature, whereas a compromise must be found in an SSF process. Furthermore, the yeast cannot be reused in an SSF process due to the problems of separating the yeast from the lignin after fermentation. Therefore, the yeast will necessarily represent a yield loss in an SSF process, if the yeast is produced from carbohydrates within the process (see Figure 1) or a running cost if it is externally supplied. The enzymes are equally difficult to reuse, also in an SHF process. The enzymes are either produced within the process (see Figure 1) – thereby representing a loss of substrate – or are externally supplied and thereby add to the chemical costs. Recirculation of enzymes is equally difficult since the enzymes bind to the substrate, although a partial desorption can be obtained after addition of surfactants 5.

The availability of lignocellulosic feedstocks varies depending on geographic location (see e.g. Kim and Dale 6), and the lignocellulosic feedstocks are rather heterogeneous in terms of both structure and chemical composition (see Table 1). This heterogeneity has a strong impact on the process design, affecting virtually all process steps, i.e. the mechanical handling of the material, pretreatment conditions, choice of enzymes and yeast strains, as well as separation and properties of the remaining lignin. This will become apparent in the discussion below.

The purpose of the pretreatment is to alter the lignocellulosic structure and increase the rate of enzymatic hydrolysis of primarily the cellulose. This should be done with a minimum formation of compounds, which inhibit the fermenting microorganisms 7. The accessible surface area is regarded as one of the most important factors affecting the effectiveness of enzymatic cellulose degradation 89101112. In native wood only a small fraction of the cell wall capillaries are accessible to the enzymes 13. Pretreatment, however, increases the available area in several ways 12141516; i) fragments and cracks are formed yielding increased area 14, ii) the hemicellulose fraction is hydrolysed which diminishes shielding effects 1718, iii) the lignin also undergoes structural changes 10141920 and the wood is delignified to various degrees, depending on the pretreatment technology 21. Thus, the shielding of microfibrils and occluding of pores, caused by lignin, can be removed. Other factors, believed to influence the digestibility in SSF, are the substrate crystallinity 112223 and the degree of polymerization (DP) 24.

The pretreatment methods can be divided into physical and chemical methods, and combinations of these two are commonly used (see e.g. the review written by Mosier et al. 21). The type of feedstock strongly affects the choice of pretreatment method. The hemicellulose is, for instance, acetylated to a high degree in xylan-rich materials. Since acetate is liberated during hydrolysis, the pretreatment of these materials is to some extent autocatalytic and require less added acid and milder process conditions. However, the liberated acetate adds to the toxicity of the hemicellulose hydrolyzates.

Ammonia fiber/freeze explosion (AFEX) pretreatment is regarded as an attractive method for pretreatment of agricultural residues, yielding highly digestible cellulose 2526. AFEX depolymerizes the lignin, removes the hemicellulose and decrystallizes the cellulose 2728. The moderate temperature and pH also minimize formation of sugar degradation products. However, the method suffers from high costs of ammonia and ammonia recovery 25. In this context the lime method, based on calcium (or sodium) hydroxide 293031 should also be mentioned. Alkali pretreatments are run at lower temperatures for long residence times, and as for the AFEX method, a delignification of the biomass is obtained.

Steam explosion is an intensively studied pretreatment method 21. The effects of uncatalyzed steam explosion – and liquid hot water pretreatments – on the biomass are primarily attributed to the removal of hemicelluloses. By adding an acid catalyst, the hydrolysis can be further improved 1932. Dilute acid pretreatments using H₂SO₄ 33343536 or SO₂ 3738394041 are the most investigated pretreatment methods because of their effectiveness and inexpensiveness. These methods have been applied in pilot plants and, hence, are close to commercialization 4243. Acid catalyzed treatment improves the hemicellulose removal 1932, gives a partial hydrolysis of cellulose 343738 and alters the lignin structure 10141920. The main drawbacks are related to the process equipment requirements 2144 and inhibitor formation 45. So far, successful pretreatments with alkali, AFEX and liquid hot water have been limited to agricultural residues and herbaceous crops 25464748, whereas acid catalysed steam pretreatments have generated high sugar yields from these materials as well as from softwood feedstocks 333435363738394041.

A simple quantification of the harshness of a steam pretreatment process is the so called Severity Factor, log(R₀). This factor combines the time and the temperature of a process into a single entity, R0=t⋅eTr−10014.75 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaemOuai1aaSbaaSqaaiabicdaWaqabaGccqGH9aqpcqWG0baDcqGHflY1cqWGLbqzdaahaaWcbeqcfayaamaalaaabaGaemivaq1aaSbaaeaacqWGYbGCaeqaaiabgkHiTiabigdaXiabicdaWiabicdaWaqaaiabigdaXiabisda0iabc6caUiabiEda3iabiwda1aaaaaaaaa@403B@49. For acid catalyzed pretreatments the Combined Severity Factor, log(CS), is sometime used. This takes also the pH into account, log(CS) = log(R₀) - pH50, and typical values for acid catalyzed steam explosion pretreatment of softwood are in the range 2 to 4 3541.

Optimal pretreatment conditions in an SSF process do not necessarily differ much from those of an SHF processes utilizing lignocellulosic biomass. However, several compounds present in pretreatment hydrolyzates, which inhibit enzymatic hydrolysis are converted by the fermenting organisms. This is a probable explanation behind the higher reported ethanol yields in SSF compared to SHF 5152. Inhibitor formation from the pretreatment may therefore be tolerated to a higher extent in an SSF process. Inhibitory compounds can be put into three major groups; furaldehydes, weak acids, and phenolics. The two most common furaldehydes, HMF (5-hydroxymethyl-2-furaldehyde) and furfural (2-furaldehyde), are formed at severe conditions from hexoses and pentoses, respectively 455354. Weak acids from lignocellulosic materials, such as acetic, formic and levulinic acid, are mainly formed by de-acetylation of hemicellulose or HMF breakdown 5354. Phenolic compounds are formed chiefly during lignin breakdown, and are to be found in numerous variants, depending on the type of lignin 55. For a more in-depth discussion on inhibition see e.g. the review by Almeida et al 7.

A successful pretreatment has to a large extent removed the hemicellulose, leaving the cellulose available for hydrolysis. Since the most commonly used microorganisms for ethanol production solely utilize sugar monomers, the cellulose needs to be hydrolyzed, which in an SSF occurs concomitantly with the fermentation. Historically, industrial cellulose digestion has been made with acid hydrolysis 56 and optimization of acid hydrolysis of various lignocellulosic materials have been carried out for ethanol producing purposes 575859. Acid hydrolysis, however, produces hydrolyzates that are relatively toxic to the fermenting microorganisms, and the maximum glucose yield is limited to approximately 60% in a batch process for kinetic reasons 60. Enzymatic degradation of the cellulose fraction, on the other hand, has the potential of yielding relatively non-toxic hydrolyzates with higher sugar yields.

Enzymes specialized in breaking up the β-1-4-glycosidic bonds of glucan are collectively called cellulases. In 1950, Reese et al 61 presented a model of enzymatic cellulose hydrolysis based on multiple enzymes (C₁ and C_X). The C₁ enzyme was assumed to produce shorter polyanhydro-glucose chains, while the solubilization was attributed to the C_X enzyme. Basically the same picture applies today, but there has been a huge progress in knowledge about all the different specific enzyme components involved. The cellulases are divided into three sub-categories, representing three types of activity: endoglucanases, exoglucanases (cellobiohydrolases) and β-glucosidases. Endoglucanases significantly reduce the degree of polymerization of the substrate by randomly attacking the interior parts, mainly in the amorphous regions of cellulose. Exoglucanases (or cellobiohydrolases), on the other hand, incrementally shorten the glucan molecules by binding to the glucan ends and releasing mainly cellobiose units. Finally, β-glucosidases split the disaccharide cellobiose into two units of glucose.

Several types of microorganisms can produce cellulase systems including aerobic filamentous fungi, aerobic actinomycetes, anaerobic hyperthermophilic bacteria and anaerobic fungi (see e.g. review by Lynd et al. 62). Intensive research on the aerobic filamentous fungi T. reesei during the past decades has resulted in an efficient cellulase producing organism, which is currently dominating the industrial cellulase production 6263.

As already mentioned, an important advantage with SSF compared to SHF is the reduction of end-product inhibition by sugars formed in the hydrolysis. The fermentation product ethanol also inhibits hydrolysis, but to a lesser extent than cellobiose or glucose 64. Another advantage is that inhibitors from the pretreatment can be metabolized by the microorganisms 51. However, also the SSF process may suffer from incomplete hydrolysis of the solid lignocellulosic fraction. Except for inhibition by end-products or other components 5165, this can be due to enzyme deactivation, unproductive enzyme adsorption 66, decreasing availability of chain ends 24, and increasing crystallinity with conversion of pretreated cellulose 67.

In an industrial SSF, enzyme and cell concentrations should be appropriately balanced in order to minimize costs for yeast and enzyme production. Synergies between the enzymes, e.g. endo-exo synergism 6869, exo-exo synergism 70, and synergism between endo- or exoglucanases and β-glucosidases 71, should also be optimized by tuning the composition of the enzyme mixtures. The optimal composition will most certainly depend on the lignocellulosic raw material.

The general requirements on an organism to be used in ethanol production is that it should give a high ethanol yield, a high productivity and be able to withstand high ethanol concentrations in order to keep distillation costs low 72. In addition to these general requirements, inhibitor tolerance, temperature tolerance and the ability to utilize multiple sugars are essential for SSF applications. Tolerance towards low pH-values will minimize the risk of contamination. The work-horse in starch or sucrose-based ethanol production is the common Bakers' yeast, Saccharomyces cerevisiae. This organism produces ethanol at a high yield (higher than 0.45 g g^-1 at optimal conditions) and a high specific rate (up to 1.3 g g^-1 cell mass h^-1 73). It also has a very high ethanol tolerance, over 100 g L^-1 has been reported for some strains and media 74. In addition, the organism has proven to be robust to other inhibitors, and hence it is suitable for fermentation of lignocellulosic materials 7576.

Hemicellulose from hardwood and agricultural residues are typically rich in xylans (cf. Table 1) – hardwood containing primarily O-acetyl-4-O-methyl-glucuronoxylan, whereas grasses contain arabinoxylan 77. Softwood hemicellulose, on the other hand, contains more mannans – primarily in the form on galactoglucomannan – and less xylan. Mannose fermentation is normally efficient in S. cerevisiae, whereas the ability to ferment galactose is strain dependent 78, and the genes for galactose utilization are furthermore repressed by glucose 7980, leading to a typical sequential utilization of the sugars. Clearly, xylose fermentation is a more significant issue for agricultural residues and hardwood than for softwood. Xylose is not metabolized by wild-type S. cerevisiae, apart from a minor reduction to xylitol. This, and for some parts the temperature tolerance, have been the main reason behind the interest to test also other microorganisms for lignocellulose conversion in SSF.

Naturally xylose-fermenting yeasts, such as Pichia stipitis and Candida shehatae 818283, could potentially be advantageous to use in SSF of materials with high xylan contents. However, their tolerance to inhibitory compounds in undetoxified lignocellulose hydrolyzates is rather low 8485, and in addition, a very low and well-controlled supply of oxygen is required for efficient xylose fermentation 868788. The main "competitors" to the yeast have been the bacteria Zymomonas mobilis and genetically engineered Escherichia coli. Z. mobilis, an obligately anaerobic bacterium, which lacks a functional system for oxidative phosphorylation, produces ethanol and carbon dioxide as principal fermentation products. Interestingly, Z. mobilis utilizes the Entner-Duodoroff pathway which gives a lower ATP production per catabolized glucose 8990. This in turn gives a lower biomass yield and a higher ethanol yield on glucose compared to S. cerevisiae 91. However, wild-type Z. mobilis lacks the ability to ferment pentose sugars, and a major drawback is furthermore that it is not a very robust organism. In general, bacteria appear to be less tolerant to lignocellulose-derived inhibitors 92, and a detoxification step may be needed prior to the fermentation. In contrast to Bakers' yeast and Z. mobilis, E. coli is capable of metabolizing a wide variety of substrates (including hexoses, pentoses and lactose), but the wild-type organism has a mixed fermentative pathway, and is thus a poor ethanol producer. In a landmark contribution, awarded U.S. patent number 5000000 93, a strain of E. coli was genetically engineered into an ethanol producer by overexpression of PDC (encoding pyruvate decarboxylase) and adhB (encoding alcohol dehydrogenase) from Z. mobilis 94. Excellent results have been achieved with recombinant E. coli, e.g. the KO11 strain, which have shown ethanol yields from 86 to close to 100% of the theoretical, and final ethanol concentrations up to 40 g L^-1 on hemicellulose hydrolyzates of bagasse, corn stover and corn hulls 95. However, only the liquid fraction was used in reported studies, and the hydrolyzates were furthermore detoxified prior to use by overliming to pH 9 with calcium hydroxide and then adjusted to pH 6.0–6.5 with HCl. Furthermore, since the optimal pH is 6.5, E. coli is less suitable for SSF processes with T. reesei cellulases, which generally is considered to have a pH optimum around 4.8 96.

Due to the very attractive properties of S. cerevisiae in industrial fermentations, there have been significant efforts made in the past decades to design recombinant xylose and arabinose fermenting strains of this yeast. Xylose fermenting strains of S. cerevisiae can in principal be constructed either by introducing genes encoding xylose isomerase (XI) from bacteria and fungi 979899, or genes encoding xylose reductase (XR) and xylitol dehydrogenase (XDH) from fungi 100101. Also the endogenous XKS1 gene encoding xylulokinase (XK) has to be overexpressed to obtain significant xylose fermentation 101. Transport proteins are needed for uptake of xylose, as well as of other sugars in yeast. In S. cerevisiae, xylose has been found to be transported by the hexose transporters, 102103, but the affinity for xylose is approximately 200-fold lower than for glucose 104. Consequently, xylose uptake is competitively inhibited by glucose.

There are 20 different genes encoding sugar transport related proteins, 18 individual systems (Hxt1-17 and Gal2) and two related signal proteins (Snf3p and Rgt2p). The transporters exhibit different affinities for sugars, and the expression of their corresponding genes is regulated by the sugar concentrations, i.e. the availability of the carbon source 105. It has previously been suggested that xylose is taken up by both high- and low-affinity systems of glucose transporters (Figure 2), but the uptake is increased in the presence of low glucose concentrations 106. Studies have indicated that the high- and intermediate-affinity hexose transporters; Hxt4, Hxt5 Hxt7 and Gal2 are in fact the most important transporters for xylose 107. Furthermore, it has been shown that a low (but non-zero) glucose concentration is needed in the medium for efficient xylose uptake 108. This has been explained by a need for glucose for expression of glycolytic enzymes and intermediates 109, as well as generation of intermediary metabolites for the initial steps of the xylose metabolism and the pentose phosphate pathway 108. Another possible explanation, inferred from both experiments and computer modeling, is that the glucose is needed for the expression of hexose transporters with favorable xylose transport properties, e.g. Hxt4 110111. Consequently, in order to obtain efficient co-fermentation of xylose and glucose in SSF (sometimes denoted SSCF – simultaneous saccharification and co-fermentation) with recombinant S. cerevisiae, it is necessary to keep the glucose concentration low, which has been shown in practice in recent SSF studies 112113.

In order to achieve a high final ethanol concentration, a high substrate loading, and hence a high WIS content, is crucial for the economy of the SSF process. Batch mode is the classical form of SSF. When the WIS content in SSF is increased, the ethanol yield tends to decrease (Figure 3A). In practice, it has been difficult to achieve good ethanol yields above WIS contents of around 10% (cf. Tables 2 and 3).

Instead of adding all substrate initially, a gradual or stepwise addition, i.e. a fed-batch approach can be used. There are several advantages by running SSF in fed-batch mode. By not adding all the hydrolyzate at once, the levels of inhibitors can be kept lower, giving less inhibition of the fermentation. A suitable feed rate may also allow a continuous conversion of inhibitors, as has been shown in fed-batch fermentation of dilute-acid hydrolyzates 114. In addition, it has been reported that also the inhibition of the enzymes decreases when some of the toxic compounds are converted 51. Stirring is a significant problem at high WIS contents due to the high viscosity 115, which results in mass and heat transfer problems. This becomes less pronounced with fed-batch SSF, due to the gradual hydrolysis of added fibers 116117. An additional advantage with fed-batch, is that the glucose level can be kept lower during co-fermentation of xylose and glucose (SSCF), which promotes xylose uptake 112113 (as discussed later on). An alternative to fed-batch addition is to make a pre-hydrolysis, i.e. to add enzymes to the bioreactor some time before the fermenting organism is added. This can be made at an elevated temperature and will decrease the initial viscosity at the start of fermentation 118. A disadvantage may be a less efficient co-fermentation of xylose due to the higher glucose concentration in the medium in the case of SSCF.

The enzyme loading is clearly important for the process economy, but the economic sensitivity towards the enzyme loading in SSF is difficult to predict due to the large uncertainties of the cost of enzymes, and lack of sufficient experimental data on the effect of enzyme load. Techno-economical calculations have indicated that a 50% reduction of enzyme loading is beneficial if the yield decreases less than 6–7% and required residence time is not increased by more than 30% 119. The enzymatic hydrolysis of the solid fraction has a large control over the total rate of ethanol production in SSF 117120. Studies in which the enzyme loading has been varied therefore show a strong positive correlation between enzyme loading and the overall ethanol yield 121122 (Figure 3B). Commercial cellulase preparations available today often need to be supplemented with extra β-glucosidase to prevent end-product inhibition by cellobiose. Optimal β-glucosidase additions have been estimated for e.g. saccharification of pretreated aspen 123. The optimal enzyme cocktail composition is certainly raw-material specific, and supplementation with extra β-glucosidase is – as to be expected – more important in SHF than in SSF 124.

To decrease the amount of added enzymes needed, investigations of SSF with mixtures of S. cerevisiae and the β-glucosidase producing yeast strain Brettanomyces clausenii, have been undertaken, and compared to SSF with only S. cerevisiae. At low enzyme loadings and without β-glucosidase addition, the mixture performed well. However, at higher cellulase loadings, higher ethanol yields were obtained when β-glucosidase was added 125. Another way of overcoming limiting hydrolysis and simplify the SSF process, is to use cellobiose-fermenting yeasts, such as Brettanomyces clausenii 126, or possibly recombinant Klebsiella oxytoca 127.

In a large-scale SSF process, the yeast (or other fermenting microorganisms) will most likely be cultivated on the hemicellulose hydrolyzate (see Figure 1). Hence, a higher yeast concentration in the SSF will result in a lower overall ethanol yield if the substrate cost for the production of the yeast is considered. However, lowering the yeast concentration will lower the volumetric productivity, and may also lead to a stuck fermentation. The rate of the enzymatic hydrolysis have in many – probably most – reported SSF experiments been rate determining, and the yeast concentration could therefore be lowered 117119120. In agreement with this, there seems not to be a strong positive correlation between cell concentration and measured ethanol yield (not counting the yield cost of the yeast production or sugar losses in the pre-treatment) above 1–2 g L^-1 cells (Figure 3C) for typical SSF conditions (~10% WIS and 30 FPU g^-1 cellulose). There is no doubt more work to be done on balancing the rates of hydrolysis and fermentation during SSF.

Progress is rapid in the field of xylose fermentation, but few industrial yeast strains have yet the demonstrated capability of fermenting xylose in lignocellulosic hydrolyzates efficiently. Hahn-Hägerdal et al. 92 recently presented information on the performance of industrial xylose fermenting strains in lignocellulosic hydrolyzates. All strains covered in their summary were XR and XDH expressing strains, which also overexpressed XK. TMB3400 is the only industrial pentose fermenting S. cerevisiae strain for which results on SSF of lignocellulosic materials have so far been reported 112113128. Ethanol concentrations reaching 40 g L^-1 and yields up to 80% of the theoretical based on xylose and glucose (at a WIS content of 7%) have been achieved (Table 3). By-product formation decreases the ethanol yield from xylose with xylose fermenting strains of S. cerevisiae. However, less xylitol is formed by XR/XDH-carrying strains in fermentation of lignocellulosic hydrolyzates 129130 compared to defined medium, probably due to additional electron acceptors present in the media. This was seen also in SSF experiments with the strain TMB3400 for several xylose-rich materials 112113128. Both glycerol and xylitol formation lead to a regeneration of NAD⁺ (cf. Figure 2). Interestingly, more glycerol than xylitol was produced 113.

Other pentose utilizing yeasts than S. cerevisiae TMB3400 have been evaluated in SSCF. Recently, Rudolf et al. 128 used sugar cane bagasse as a substrate in SSF with P. stipitis as a fermenting organism (see Table 3). It was indeed possible to use the organism in untreated bagasse hydrolyzate, but clearly higher yields and ethanol concentrations were achieved with S. cerevisiae TMB3400. Xylose fermenting bacteria have not been much examined in lignocellulosic SSF, but yellow poplar hardwood was used in SSF experiments with recombinant Z. mobilis co-fermenting xylose and glucose 131. However, a thorough detoxification was required prior to the SSF.

Arabinose fermentation in SSF has not yet been reported, although arabinose fermenting S. cerevisiae strains have recently been constructed 132133 as well as strains co-utilizing arabinose and xylose 134. Also Z. mobilis strains co-utilizing arabinose and xylose have been developed 135. However, further work is needed before efficient ethanol production in SSF from arabinose can be conducted.

In SSF a compromise between the optimal temperatures for the cellulolytic enzymes and the yeast is needed. Earlier SSF experiments in our labs were often run at a temperature of 37°C. Since the yeast S. cerevisiae has an optimal temperature around 30°C and the cellulolytic enzymes around 55°C, this was regarded as a suitable compromise at the high end of what S. cerevisiae can tolerate 117120122125136. However, recent studies have shown important strain differences with respect to temperature tolerance, and furthermore, the co-fermentation of glucose and xylose is affected by temperature. Rudolf et al. 128 concluded that more xylose was consumed by TMB3400 at 32°C than at 37°C during SSF of sugar cane bagasse, and Olofsson et al. 113 found that a temperature of 34°C was to prefer in SSF of wheat straw. Possibly, a lower rate of hydrolysis, which gives a slower release of glucose, favors xylose uptake in the competition for transporters. Furthermore, inhibition effects may play a role, and tolerance to inhibitors may be higher at temperatures closer to the optimum of the yeast.

Thermotolerance is clearly an important topic for SSF and thermotolerant yeast strains, e.g. Fabospora fragilis, Saccharomyces uvarum, Candida brassicae, C. lusitaniae, and Kluyveromyces marxianus, have been evaluated for future use in SSF 137138139140141, to allow fermentation at temperatures closer to the optimal temperature for the enzymes. However, in all these cases saccharification of pure cellulose (e.g. Sigmacell-50) or washed fibers, in defined fermentation medium, were applied. SSF of cellulose with mixed cultures of different thermotolerant yeast strains have also been carried out 140142. However, there is so far a lack of results from SSF experiments in which untreated lignocellulosic materials have been used together with thermotolerant strains.

The amounts and types of inhibitory compounds vary strongly between different raw materials, and also depend on the pretreatment method. Hence, the needed inhibitor tolerance of a strain in an SSF process may vary depending on raw material. Several alternatives of detoxification (i.e. removal of inhibitory compounds) have been explored, e.g. over-liming, extraction with organic solvents, ion exchange, molecular sieves, and steam stripping 143144. Overliming with Ca(OH)₂ is the most commonly used method. A significant drawback of this method is that calcium salts may precipitate in the process and contaminate surfaces of distillation columns, evaporators and heat-exchangers. Hence, detoxification should be avoided if possible, due to additional process cost as well as possible loss of fermentable sugars 145146.

More tolerant yeast strains for SSF than those available today, may be achieved through genetic modifications, e.g. overexpressing genes encoding enzymes for resistance against specific inhibitors, and altering co-factor balance in the cell 7. Another way to improve strains is by evolutionary engineering, through which strain robustness is improved by mutation and selection 147. Yet another approach to overcome the problem with inhibition is by adapting the SSF process. By applying e.g. a fed-batch mode of substrate addition with proper feed protocol and control variables, the levels of inhibitors can be kept at an acceptable level. Such strategies have proven successful during cultivation and fermentation of liquid hydrolyzates 148149, as well as in SSF 117. A combination of more inhibitor-tolerant strains in combination with efficient feed strategies will likely improve process robustness in SSF processes.