Waste Management 33 (2013) 13451361

Contents lists available at SciVerse ScienceDirect

Waste Management
journal homepage: www.elsevier.com/locate/wasman

Review

A review of dark fermentative hydrogen production from biodegradable municipal waste fractions
G. De Gioannis a,c,, A. Muntoni a,c, A. Polettini b, R. Pomi b
a

DICAAR Department of Civil and Environmental Engineering and Architecture, University of Cagliari, Cagliari, Italy Department of Hydraulics, Transportation and Roads, University of Rome La Sapienza, Italy c IGAG-CNR, Environmental Geology and Geoengineering Institute of the National Research Council, Italy
b

a r t i c l e

i n f o

a b s t r a c t
Hydrogen is believed to play a potentially key role in the implementation of sustainable energy production, particularly when it is produced from renewable sources and low energy-demanding processes. In the present paper an attempt was made at critically reviewing more than 80 recent publications, in order to harmonize and compare the available results from different studies on hydrogen production from FW and OFMSW through dark fermentation, and derive reliable information about process yield and stability in view of building related predictive models. The review was focused on the effect of factors, recognized as potentially affecting process evolution (including type of substrate and co-substrate and relative ratio, type of inoculum, food/microorganisms [F/M] ratio, applied pre-treatment, reactor conguration, temperature and pH), on the fermentation yield and kinetics. Statistical analysis of literature data from batch experiments was also conducted, showing that the variables affecting the H2 production yield were ranked in the order: type of co-substrate, type of pre-treatment, operating pH, control of initial pH and fermentation temperature. However, due to the dispersion of data observed in some instances, the ambiguity about the presence of additional hidden variables cannot be resolved. The results from the analysis thus suggest that, for reliable predictive models of fermentative hydrogen production to be derived, a high level of consistency between data is strictly required, claiming for more systematic and comprehensive studies on the subject. 2013 Elsevier Ltd. All rights reserved.

Article history: Received 10 July 2012 Accepted 19 February 2013 Available online 1 April 2013 Keywords: Anaerobic digestion Biological hydrogen production Dark fermentation Food waste Organic fractions of municipal solid waste

1. Introduction In Europe, anaerobic digestion of biodegradable residues has received renewed attention by the scientic and technical community over the last decade (Mata-Alvarez et al., 2000, MataAlvarez, 2002; De Baere, 2003; Bolzonella et al., 2006; Karagiannidis and Perkoulidis, 2009), especially for the organic fraction of municipal solid waste. This is due to several reasons that stem

Abbreviations: AR, aged refuse; COD, chemical oxygen demand; CSTR, continuous stirred tank reactor; F/M, food/microorganisms; FW, food waste; HRT, hydraulic retention time; HST, heat shock treatment; MEC, microbial electrolysis cell; MFC, microbial fuel cell; OLR, organic loading rate; OFMSW, organic fraction of municipal solid waste; OMW, olive mill wastewater; PBR, packed bed reactor; PS, primary sludge; SBR, sequencing batch reactor; SRT, solids retention time; SS, sewage sludge (mixture of primary and secondary sludge); TOC, total organic carbon; TS, total solids; TKN, total Kjeldahl nitrogen; UASB, upow anaerobic sludge blanket; VFAs, volatile fatty acids; VS, volatile solids; WAS, waste activated sludge. Corresponding author at: University of Cagliari, DICAAR Department of Civil and Environmental Engineering and Architecture, Cagliari, Italy. Tel.: +39 0706755551; fax: +39 0706755523. E-mail address: degioan@unica.it (G. De Gioannis). 0956-053X/$ - see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.wasman.2013.02.019

from the EU legislative framework, which has set specic constraints on landlling of biodegradable wastes, maximization of materials recycling as well as enhancement of energy production from renewable sources. Numerous investigators (Han and Shin, 2004a; Liu et al., 2006; Gmez et al., 2006, 2009; Ueno et al., 2007; Chu et al., 2008; Wang and Zhao, 2009; Lee et al., 2010b; Dong et al., 2011) demonstrated that if fermentation of biodegradable organic substrates is appropriately operated in a two-staged mode, separation of the acidogenic and methanogenic phases can be accomplished: while acidogenesis produces hydrogen and carbon dioxide as the gaseous products and releases VFAs into the liquid solution, methanogenesis allows for nal conversion of the residual biodegradable organic matter from the rst stage into methane and carbon dioxide. Considering that H2 has the highest caloric value per unit weight of any known fuel and an improved acidogenic phase has been reported to result in enhanced biogas yield in the second stage, separating the phases of the anaerobic digestion process would increase the energy efciency overall (Liu et al., 2006; Lee and Chung, 2010; Dong et al., 2011). Furthermore, proper processing of the digestate to yield a valuable nal product for use as a soil

1346

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

amending material would contribute to improved environmental sustainability of management of biodegradable organic residues. A number of potentially suitable residual substrates have been evaluated for biohydrogen generation potential through dark fermentation. Among these, fractions of municipal solid waste such as food waste (FW) and the broader mixture of materials known as organic fraction of municipal solid waste (OFMSW; basically FW combined with non-recoverable paper residues) may represent relatively inexpensive and suitable sources of biodegradable organic matter for H2 production, mainly due to their high carbohydrate content and wide availability (Okamoto et al., 2000; Lay et al., 2003; Kim et al., 2004, 2011a; Liu et al., 2006; Li et al., 2008a,b; Zhu et al., 2008; Wang and Zhao, 2009; Nazlina et al., 2011). Hydrogen production via fermentation involves either facultative or strict anaerobic bacteria. The various metabolic pathways that may establish can either be promoted or inhibited, depending on the adopted operating conditions, which govern the production of specic volatile fatty acids (VFAs) and alcohols including acetate, propionate, butyrate, lactate and ethanol. In carbohydrates fermentation, the acetate and butyrate pathways involve the production of, respectively, 4 and 2 mol of molecular hydrogen per mol of glucose degraded. However, propionate, ethanol and lactic acid can also be produced in mixed bacterial cultures, adversely affecting H2 production: propionate is a metabolite of a H2-consuming pathway, while ethanol and lactic acid are associated with zero-H2 pathways (Guo et al., 2010a). The question as to how to achieve optimal H2 generation while keeping treatment costs low and producing an efuent suitable for further treatment is probably the main technical issue to be addressed. To this regard, operational parameters including temperature, pH, reactor conguration, substrate concentration and organic loading rate should be the subject for optimization of process efciency. Recent literature studies on H2 production from FW and OFMSW through dark fermentation have focused on a broad range of operating conditions, implicitly denoting that for full-scale application of the process a better understanding of the inuence of the relevant process parameters is still required. The aim of this manuscript is to present an updated overview of H2 production from FW or OFMSW through dark fermentation, based on more than 80 recent related publications. Although a number of review papers has been published on fermentative H2 production from various biodegradable wastes, to the authors knowledge a critical overview of literature studies with a specic focus on FW/OFMSW is still missing. The analysis conducted in the present study was focused on the following issues: (a) type of inoculum and applied pre-treatment, (b) type of fermentation reactor, (c) organic loading rate (OLR), (d) solids retention time (SRT), (e) temperature and pH. Since the numerous literature studies on this subject have adopted different approaches focusing on several specic aspects of the fermentation process, the reported results are diverse and sometimes even conicting. On account of this, an effort was made in the present manuscript to statistically analyzing literature data to derive information on the relative importance of the main parameters of concern, as well as on their potential mutual relationships.

of mass or energy units. The concept of conversion efciency derives from the existence of a fermentation barrier to hydrogen production from organic substrates, which may be elucidated considering the conversion of a simple carbohydrate such as glucose. If the complete conversion reaction to hydrogen is taken into account (Eq. (1)), it turns out that theoretically 12 mol H2 can be extracted from 1 mol of glucose:

C6 H12 O6 6H2 O ! 12H2 6CO2

However, this reaction is energetically unfavorable with respect to biomass growth and would also only occur at extremely low H2 concentrations, so that the real conversion potential is in fact lower than this theoretical value. At the best, the optimized conversion of glucose into hydrogen is limited to acetate production and is therefore practically limited by the existence of an upper threshold the so-called Thauer limit (Thauer et al., 1977) of 4 mol H2/mol glucose (Eq. (2)). As a result, only one third of the theoretical hydrogen production can be achieved in practice, since part of the reducing equivalents in the original substrate remains as acetate.

C6 H12 O6 2H2 O ! 4H2 2CO2 2CH3 COOH

In practice, however, organic intermediates also act as electron scavengers, which gives rise to the production of more reduced fermentation products compared to acetate, including propionate, butyrate and longer aliphatic acids, lactate, formate, alcohols and ketones, with an associated decrease in the H2 generation yield. In case the butyrate fermentation pathway is established, the conversion efciency is reduced to 2 mol H2/mol glucose:

C6 H12 O6 ! 2H2 2CO2 CH3 CH2 CH2 COOH

It has also been shown (Nath and Das, 2004; Davila-Vazquez et al., 2008; Hallenbeck and Ghosh, 2009) that, since in nature fermentation processes have been optimized not to produce hydrogen but to sustain microbial growth, hydrogen represents a waste of energy during metabolism and is therefore preferentially recycled within the metabolic pathways. As a result, a number of reduced products are formed to sustain microbial cell synthesis, including ethanol, butanol, butyrate and lactate, which allow for NADH re-oxidation. This explains how in real practice, even under optimal process conditions, conversion efciencies to H2 of higher than 15% of the original electrons in the substrate are hardly attained (Angenent et al., 2004). On account of the considerations above, the conversion efciency may be calculated on a mass basis as follows:

Em

mol H2 produced=mass of substrate 100 Theoretical mol H2 produced=mass of substrate

Table 1 reports conversion efciency data according to the definition provided above, as derived from different literature sources. The hydrogen production efciency may alternatively be evaluated from an energetic perspective, considering the fraction of the total energy content of the substrate recovered in the form of hydrogen, as expressed by Eq. (5):

Ee

Energy content of the H2 produced 100 Energy content of the original substrate

2. Process yield and conversion efciency An important issue related to fermentative hydrogen production from biodegradable wastes involves how to appropriately evaluate and express process efciency. To this regard, the expected hydrogen production yield may be conveniently converted into a parameter representing the conversion efciency attained upon fermentation, which may in turn be expressed either in terms

Assuming 2888 kJ/mol glucose and 242 kJ/mol H2 (Dong et al., 2009b) as the lower heating values of glucose and hydrogen, energy conversion efciencies of 33.5% and 16.8% are calculated if the acetate (Eq. (2)) or butyrate (Eq. (3)) fermentation pathways are assumed to occur, respectively. Alternatively, the amount of energy converted into hydrogen may also be derived considering the COD equivalent of H2, which is equal to 16 g COD/mol H2; accordingly, if the specic hydrogen production is expressed per unit mass of input COD, it may be easy

G. De Gioannis et al. / Waste Management 33 (2013) 13451361 Table 1 Conversion efciency data documented in the literature. Type of substrate OFMSW FW FW FW FW FW FW FW FW FW + sewage sludge FW FW (liquid phase) OFMSW FW FW FW FW + paper
a

1347

Conversion efciency 1.98 0.03 mol H2/molglucose 9.3% of inuent COD converted to H2 2.1 mol H2/molhexose consumed 5.119.3% of inuent COD converted to H2 1.12 mol H2/molhexose 0.87 mol H2/molhexose added 2.05 mol H2/molhexose consumed 0.350.54 mol H2/molhexose 9.35% of inuent COD converted to H2 2.11 mol H2/molhexose added 1.79 mol H2/molhexose 1.82 mol H2/molglucose 2.5 mol H2/molhexose 12 mmol/gCOD 0.60.9 mol H2/molhexose 0.030.1 mol H2/molhexose 1.8 mol H2/molhexose 1.483.2 mol H2/molhexose

Specic H2 production 127 ml/g VSremoved 205 ml/gVSadded 2728 ml/g VSadded 1926 ml/g VSadded 70263 ml/g VSaddeda 80.9 ml/g VSadded 62.6 ml/g VSadded 153.5 ml/g VSadded 165 ml/g VSadded 137.2 ml/g VSadded 114 ml/g VSadded 28.446.3 ml/g VSadded 1.35.0 ml/g VSadded 91.5 ml/g VSadded 165360 ml/g VSremoved

Reference Alzate-Gaviria et al. (2007) Chu et al. (2008) Elbeshbishy et al. (2011) Gmez et al. (2009) Han Kim Kim Kim Kim Kim and Shin (2004b) et al. (2008b) and Shin (2008) et al. (2009) et al. (2010) et al. (2011a)

Kim et al. (2011b) Lee and Chung (2010) Lee et al. (2010b) Li et al. (2008b) Shin et al. (2004) Shin et al. (2004) Valdez-Vazquez et al. (2005)

Calculated from data provided in the manuscript.

to estimate the fraction of the substrates energy content which is actually converted into H2 (Kim et al., 2011a). It has been widely documented that in practice, even under optimized process conditions, a considerable portion of the carbon, reducing equivalents and energy content of the original substrate remains in the efuent from the hydrogenogenic phase. In order to maximize the overall conversion yield and ensure adequate substrate degradation, the biohydrogen production process should thus be thought as a part of a combined process where additional energy production and enhanced substrate conversion are attained in different processing stages, which is discussed further in Section 6. 3. Observed inuence of process conditions As will be shown later in this review, literature studies on biological hydrogen production from residual substrates including FW and OFMSW report considerably wide ranges of variation in the specic production yield observed. The large differences documented in the literature reect the underlying inuence of numerous process parameters, including substrate composition and presence of co-substrates, type of inoculum and applied pre-treatment, reactor type, mode of reactor operation (batch, semi-continuous or continuous), operating variables such as temperature, hydraulic retention time (HRT), OLR and pH. These are recognized to be the most relevant factors affecting the evolution of the fermentative pathways and the associated hydrogen generation yield. The effects of the above mentioned factors are also recognized to be strictly interrelated and mutually interactive, so that a change in one parameter may affect hydrogen production not only due to its individual effect, but also as a result of combined interactions with other process variables. Elucidating and predicting the individual inuence of single process conditions and their mutual interactions (which may be either synergistic or antagonistic) is probably one of the main challenges of the research on biological hydrogen production, especially in those cases where complex substrates (for which the metabolic reactions are not fully known in advance) are concerned. In the following sections a discussion of literature ndings about the inuence of the relevant process parameters is provided, also highlighting where the effects of the operating variables have been found to be mutually dependent.

3.1. Type and pre-treatment of inoculum The seed microorganisms to be used in the process and the need for inoculum pre-treatment are among the most debated issues in hydrogen production from dark fermentation. A variety of data is available in the literature as for the type of inocula used in FW/

Table 2 Type of inoculum used and applied pre-treatment conditions. Type of inoculum Mixture of deep soil, excretes vaccine, pig excretes Activated sludge Type of pretreatment HST (100 C, 15 min) HST (110 C, 30 min) HST (boiling, 15 min) HST (boiling, 15 min) HST (boiling, 15 min) HST (90 C, 10 min) HST (90 C, 10 min) HST (90 C, 15 min) HST (90 C, 10 min) Cultivation HST (boiling, 15 min) HST (100 C, 2 h) HST (60120 C, 2 8 h) HST (80 C, 20 min) Acclimation HST (90 C, 30 min) HST (90 C, 10 min) Cultivation HST (105 C, 3 h) + acclimation HST (105 C, 3 h) + acclimation Cultivation Reference Alzate-Gaviria et al. (2007) Cappai et al. (2009, 2010, 2011) Dong et al. (2009a,b) Han and Shin (2004a) Han and Shin (2004b) Kim et al. (2004) Kim et al. (2008b) Kim et al. (2010) Kim and Shin (2008) Lay et al. (1999)

Anaerobic sludge Anaerobic sludge Anaerobic sludge Anaerobic Anaerobic Anaerobic Anaerobic sludge sludge sludge sludge

H2-producing microbial consortium Anaerobic sludge Anaerobic sludge Anaerobic grass compost Anaerobic sludge Compost Anaerobic sludge Anaerobic sludge H2-producing microbial consortium Anaerobic sludge Anaerobic sludge Anaerobic sludge

Lay et al. (2003) Lay et al. (2005) Lee and Chung (2010) Lee et al. (2010a) Lee et al. (2010b) Shin et al. (2003) Shin et al. (2004) Sreela-or et al. (2011a) Sreela-or et al. (2011b) Zhu et al. (2008)

1348

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

OFMSW fermentation experiments and the nature of pre-treatments applied to improve the process yield (please refer to the summary data provided in Table 2). In general, various pure cultures have been explored to produce hydrogen from a variety of substrates (mainly mono-substrates). In fermentative H2 production from FW/OFMSW, researchers use mixed microbial cultures for practical reasons, since a mixed culture system would be cheaper to operate, easier to control, and capable of digesting a variety of feedstock materials (Li and Fang, 2007; Valdez-Vazquez et al., 2005). Anaerobic sludge from fullscale anaerobic digesters, either with or without specic pre-treatment, has been used in numerous investigations as a supplier of a mixed anaerobic consortium. Among the most notable exceptions, Han and Shin (2002), Lee et al. (2008, 2010a) and Cappai et al. (2009) used, respectively, rumen microorganisms from the stomach of cows, an enriched culture from FW compost, and waste activated sludge (WAS) with no specic pre-treatment. In such studies, rumen microorganisms were investigated due to their enhanced cellulolytic capability, while facultative anaerobic bacteria in WAS were considered suitable to enhance the fermentative stage of the process due to their high growth rate and capability to rapidly recover from accidental oxygen intrusion. Alzate-Gaviria et al. (2007) adopted mixtures of inocula (deep soil and vaccine and pig excretes), while Zong et al. (2009) tested cattle dung compost; in both cases a thermal pre-treatment (15 min at 100 C) was applied. Aged refuse (AR) excavated from a solid waste landll with over 10 years of placement was also used to enhance biohydrogen production from FW (Li et al., 2008a). The best results (194 N l H2/ kg VS, 94.3 N l H2/kg VS h) were attained when FW was mixed with sewage sludge (SS) or AR at mixing ratios of 100:30 (dry weight) and 100:50 (wet weight), respectively; the FW and SS were thermally pre-treated (2 h at 160 C), while AR was used as received assuming that exposure to air upon mining reduced strictly anaerobic hydrogen utilizing methanogens. In general, the addition of AR was believed to signicantly promote the hydrogen generation yield in comparison to SS, due to the enrichment in effective biomass population and to the alkaline characteristics which created favorable conditions for the metabolism of hydrogen-producing bacteria. Although the use of mixed microora for fermentative hydrogen production is practically more viable, important limitations arise from the coexistence of H2-producing and consuming bacteria in nature. A solution which is often practiced involves pre-treatment by various methods to harvest hydrogen producers, on account of their larger chances to survive when a mixed culture is treated by harsh conditions due to the ability of some bacterial species, including Bacillus and Clostridium, to sporulate as a reaction to adverse environmental conditions. To this regard, the main options available are heat-shock treatment (HST), acid treatment, aeration, freezing and thawing, as well as addition of specic chemical compounds. The strategy for H2-consumers inhibition should be selected on the basis of capital and operational costs, feasibility and complexity of the process layout, time needed for inoculum stabilization, effectiveness over the entire fermentation process, and secondary effects/degree of compatibility with further process steps (e.g., methanogenesis, aerobic composting). It should also be mentioned that when non-sterile substrates are used, proliferation of new non-inhibited methanogens is possible, therefore continuous application of the inhibition method is typically required (Valdez-Vazquez and Poggi-Varaldo, 2009). The main chemical inhibitors that have been used for H2 utilizing methanogens are sodium2-bromoethanesulfonate, 2-bromoethanesulphonic acid, iodopropaneacetylene, ethylene, ethane, methyl chloride, methyl uoride, lumazine, nitrate and chloroform. Chemical inhibitors may be

either specic or non-specic towards methanogens, that can include both H2 consumers and other types of methanogens. The Coenzyme M (CoM) is involved in the terminal stage of methane biosynthesis, where the methyl group carried by CoM is reduced to methane by the methylCoM reductase. BES (sodium-2-bromoethanesulfonate), BESA (2-bromoethanesulphonic acid) and lumazine (C6H4N4O2) are structural analogues of CoM specically found in methanogens only but not in other bacteria or Archea. They can competitively inhibit the methyl transfer reaction at the terminal reducing stage of methane formation from H2 and CO2. Ethylene is recommended as a reversible selective inhibitor of methanogenesis; methanogenic activity has been reported to completely recover after ethylene removal. Acetylene was also used as a non-specic inhibitor of methanogens. It has been assumed that acetylene destroys the proton gradient across the cell membrane and thus leads to a breakdown of energy metabolism. Chloroform (CHCl3) is known to inhibit the function of corrinoid enzymes and the methylCoM reductase. CHCl3 can inhibit both acetoclastic and hydrogenotrophic methanogens. However, CHCl3 investigation has been found not only to inhibit the activity of methanogenic Archaea but also that of homoacetogenic bacteria and acetate-consuming sulfate-reducing bacteria. Iodopropane is another corrinoid antagonist and prevents the function of B12 enzymes as methyl group carriers. The effect of methyl uoride is quite controversial: it has been reported to inhibit effectively aerobic CH4 oxidation, while not to affect methanogenesis. However in some experiments, methanogenesis was reduced by about 75% compared to the control without the inhibitor. Methanogenesis could partly be recovered when CH3F was ushed with N2. Nitrocompounds such as nitrate, nitrite, nitroethane, 2-nitropropanol and phosphate can be used as alternative electron acceptors that more effectively consume the reducing equivalents produced during fermentation, redirecting the electron ow away from the reduction of carbon dioxide to methane (Chidthaisong and Conrad, 2000; Liu et al., 2011). The most common approach to harvest hydrogenogenic microorganisms reported in the literature is however by far HST, which is based on the ability of some bacterial species, including Bacillus and Clostridium, to sporulate as a reaction to adverse environmental conditions. Typically, HST requires temperatures around 100 C for durations of 15120 min in order to suppress non-spore-forming bacteria, leaving spores of acidogenic bacteria that will germinate back to their active vegetative state when suitable growth conditions get re-established (Lay et al., 2003; Lin and Lay, 2004; Fang et al., 2006; Alzate-Gaviria et al., 2007; Argun et al., 2008; Bhaskar et al., 2005). In recent studies on FW, the pre-treatment conditions adopted ranged from 20 min at 80 C (Lee and Chung, 2010) to 2 h at 100 C (Lay et al., 2003). Concerns about the net energy gain of HST, however, make this biomass selection method controversial and claims for further investigation. Enrichment of hydrogen producers by HST is an energy-intensive practice, and the degree of energy consumption can only be partially reduced by adopting temperatures as low as 75 85 C. On account of this, studies have also focused on hydrogen production from FW with no inoculum pre-treatment with the aim of reducing costs and simplifying the process (Chu et al., 2008; Hong and Haiyun, 2010; Lee et al., 2008, 2010a; Li et al., 2008b; Shin and Youn, 2005; Pan et al., 2008; Zhu et al., 2008; Kim et al., 2011b). An approach dened as biokinetic control has been introduced by Valdez-Vazquez et al. (2005). This is based on maintaining such environmental conditions as to hinder methanogens growth, including low pH, appropriate temperatures or short HRTs causing the wash-out of methanogens (Valdez-Vazquez et al., 2005; Valdez-Vazquez and Poggi-Varaldo, 2009; Cappai et al., 2010; Kim et al., 2011b).

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

1349

There are also studies in the literature where no inoculum was added to the feed material, the evolution of the fermentation process relying in such cases on the indigenous biomass present in the waste only. In a two-stage fermentation process for combined hydrogen and methane production from an unsterilized mixture of OFMSW and slaughterhouse waste, Gmez et al. (2006) found that the hydrogen-producing stage had a stable performance (52.571.3 N l H2/kg VSremoved). Wang and Zhao (2009) also applied a two-stage process, which was performed in a semi-continuous rotating drum in which the indigenous mixed microbial cultures contained in food waste were used for hydrogen production; in the hydrogen production stage (operated at an OLR of 22.65 kg VS/m3 d and a solids retention time [SRT] of 160 h) a maximum hydrogen yield of 0.065 Nm3 H2/kg VS was attained. In the work by Kim et al. (2011b) temperature control in the range 35 60 C was adopted as a biokinetic control strategy to optimize H2 production from FW. The optimal condition for both the H2 production yield and rate was found at an operating temperature of 50 C, with values of, respectively, 1.8 mol H2/mol hexoseadded (or 137 N ml H2/g VSadded) and 369 N ml H2/l h (or 18.8 N ml H2/ g VS h). When FW was preliminarily heat-shocked (90 C, 20 min) but the fermentation temperature was maintained at 35 C (Kim et al., 2011a), the H2 production yield (1.8 mol H2/mol hexoseadded, or 165 N ml H2/g VSadded) was comparatively similar to the previous case, however the process rate (300 N ml H2/l h, or 7.0 N ml H2/ g VS h) was appreciably reduced. 3.2. Type of reactor and inuence of retention time and OLR Different reactor congurations have been used to treat FW/ OFMSW, mostly consisting of small-scale (100500 ml) vessels and stirred fermenters of 210 l, operated under batch, semi-continuous or continuous conditions. In fermentative hydrogen production, the SRT, and in turn the OLR, affect the substrate conversion efciency, the type of active microbial population as well as the metabolic pathways established in the system. The inuence of SRT and OLR on hydrogen yield is controversial in the literature. It is generally acknowledged that long SRTs favor the buildup of H2 consumers, such as methanogens, and competitors for substrate, such as non-H2-producing acidogens (Wang and Zhao, 2009). On the other hand, a low SRT may reduce the substrate utilization efciency, in particular in the case of complex substrates which need an adequate hydrolysis period, and cause the washout of the active biomass, in turn impairing the conversion yield. It should be noted that, since most of the reactors used in the reviewed literature were often operated with no biomass recycle, HRT and SRT coincide. In view of fullscale implementation of the process, HRT is of particular concern since it is clearly related to capital costs. The OLR of the system may affect a number of operating issues, including VFA accumulation and pH changes (which in turn is a function of the systems alkalinity), as well as variations in the composition of the active biomass, with consequent modications of the associated metabolic pathways. A comparison of the OLR effects on hydrogen production documented in the literature is complicated by the fact that this parameter is often expressed in inconsistent units in different studies (kg VS/m3 d, kg COD/m3 d, kg TOC/m3 d), and harmonization of the units used is not always possible due to lack of the required conversion factors. Most studies that used stirred reactors with continuous or semi-continuous operation adopted HRT values between 21 h (Lee and Chung, 2010, who worked on the liquid phase extracted from FW) and 4 d (Lee et al., 2010a). The reported OLRs values fall within the ranges 838 kg VS/m3 d (Hong and Haiyun, 2010; Chu et al., 2008) or 2064 kg COD/m3 d (Li et al., 2008b; Chu et al., 2008).

Shin and Youn (2005) found that prolonging the HRT of a semicontinuous system from 2 to 5 d and reducing the OLR from 10 to 8 kg VS/m3 d more than doubled the hydrogen yield (2.2 vs. 1 mol H2/mol hexose); the change in the OLR was found to prevent VFAs to accumulate in excess of 20,000 mg COD/l. Similar ndings were reported by Wang and Zhao (2009) who operated an integrated two-stage fermentation system; a signicant reduction in VS removal and H2 yield was observed as OLR progressively increased from 15.10 to 37.75 kg VS/m3 d and SRT decreased from 10 to 6.6 d. Such negative effects on the fermentation process were ascribed to the associated reduced timespan of substrate hydrolysis. Furthermore, increased OLRs were found to result in reduced acetate and butyrate production with an associated increase in propionate and lactate concentrations; at an OLR of 37.75 kg VS/ m3 d, the lactate concentration attained a maximum, accounting for 30% of the total COD of the measured metabolites (ethanol, acetate, propionate, butyrate and lactate). A signicant increase in the hydrogen yield was recorded when SRT increased from 5 to 6.6 d (corresponding to a decrease in OLR from 30.20 to 22.65 kg VS/m3 d), while a further increase in SRT from 6.6 (OLR = 22.65 kg VS/m3 d) to 10 d (OLR = 15.10 kg VS/m3 d) only slightly increased hydrogen production. In an optimization study of semi-continuous digestion of FW and dewatered WAS (Hong and Haiyun, 2010), long HRTs (8.9 d) were found to result in improved hydrogen generation yield. Prolonged solids retention times were also adopted by Valdez-Vazquez et al. (2005): an OLR of 11 g VS/kg d corresponded to a SRT of 21 days, and the hydrogen production yield obtained was 165 and 360 N ml/g VSrem under mesophilic and thermophilic conditions, respectively; the same SRT value is also reported by Valdez-Vazquez and Poggi-Varaldo (2009), with an associated maximum H2 production of 51.2 N ml/g VSrem. Different conclusions were suggested by the operation of a twophase hydrogen + methane production plant studied by Lee and Chung (2010); in this case an OLR increase from 7.4 to 71.3 g COD/l h, with an associated HRT decrease from 66 to 21 h, resulted in a signicant increase in the hydrogen production rate, which varied from 0.62 to 3.88 l/m3 d. In addition to the widely used continuous-ow stirred reactor (CSTR), other types of reactors have also been investigated in order to improve the efciency of biohydrogen production. A packed bed reactor (PBR) was used by Alzate-Gaviria et al. (2007) to obtain high hydrogen production yields in the short HRT required for methanogenesis inhibition; a yield of 99 N ml H2/g VSremoved was attained. An anaerobic sequencing batch reactor (SBR) was used by Kim et al. (2008b) for FW fermentation, working at different SRTs and HRTs; the maximum hydrogen yield (80.9 N ml H2/g VS, or 1.12 mol H2/mol hexose) was displayed at an SRT of 126 h and an HRT of 33 h. A leaching-bed reactor operated in a sequential batch mode at an SRT of 6 d was used by Han and Shin (2004b) for FW fermentation; the outow from the leaching-bed reactor was then fed to an upow anaerobic sludge blanket (UASB) reactor for methane production; with an OLR of 11.9 kg VS/m3 d, a hydrogen yield of 0.31 Nm3 H2/kg VS was achieved. Elbeshbishy et al. (2011) evaluated the performance of a sonicated biological reactor (basically a CSTR equipped with an ultrasonic probe at its bottom) by comparison with a conventional CSTR fed with raw FW and a CSTR fed with sonicated FW. While the CSTR treating the sonicated feed exhibited a 23% increase in hydrogen yield compared to the conventional CSTR, for the sonicated reactor the observed improvement was as high as 62%; similarly, the hydrogen production rate increased, in comparison to the conventional system, by 27% and 90% for the CSTR fed with sonicated waste and for the sonicated reactor, respectively.

1350

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

Although no data on full-scale hydrogen fermentation plants is currently available, some experience has recently been gained on pilot-scale reactors. A pilot-scale (150 l working volume) anaerobic SBR treating FW is described in Kim et al. (2010); the reactor was operated at 35 C and an HRT of 36 h, achieving a hydrogen yield of 0.5 mol H2/mol hexose. The largest pilot-scale system working with FW is described in Lee and Chung (2010). A two-phase hydrogen + methane fermentation system with a 500 l CSTR as the rst stage was operated at 30 C and an optimal HRT of 21 h for the hydrogen production stage. The hydrogen fermentation tank allowed for a production of 1.82 mol H2/mol hexose. 3.3. Process temperature The majority of recent studies on hydrogen production from FW and OFMSW was conducted under mesophilic conditions, specically between 30 (Lee and Chung, 2010) and 40 C (Wang and Zhao, 2009), and typically in the range 3537 C (Dong et al., 2009a; Hong and Haiyun, 2010; Kim et al., 2010, 2011a; Li et al., 2008a,b; Zong et al., 2009). The effect of temperature in the mesophilic range (3045 C) was investigated by Kim et al. (2008a) in FW fermentation by Clostridium beijerinckii KCTC 1785; hydrogen production increased with increasing temperature up to 40 C (with a maximum in acetate and butyrate production), being then strongly inhibited at 45 C. Studies on thermophilic conditions, mostly at temperatures of 55 C, are also documented in the literature (Chu et al., 2008; Lee et al., 2010a; Shin and Youn, 2005; Valdez-Vazquez and Poggi-Varaldo, 2009; Nazlina et al., 2011). Thermophilic conditions are assumed to optimize the enzymatic activity of hydrogenase during fermentation by Clostridia, to inhibit the activity of H2 consumers and also to suppress the growth of lactate-forming bacteria (Lay et al., 1999; Oh et al., 2004; ValdezVazquez et al., 2005). Kim et al. (2011b) investigated the effect of temperature in both the mesophilic and thermophilic range (35 60 C) on FW fermentation in the absence of any specic inoculum, and found that the lowest and highest H2 production yields were associated to temperatures of 35 and 50 C, respectively. Although in both cases the amount of organic acids was comparable, lactate was predominant at 35 C while butyrate was the main VFA component at 50 C. Microbial analysis of the fermentation medium also indicated that the dominating species were lactic acid bacteria at 35 C and H2-producers at 50 C, thus conrming the role of temperature in dictating the nature of microbial consortium during the process. On the other hand, however, high temperatures have also been reported to induce thermal denaturation of proteins and essential enzymes, in turn negatively affecting microbial activity (Lee et al., 2006). Comparative studies on hydrogen production from FW under mesophilic and thermophilic conditions were carried out by Shin et al. (2004), Valdez-Vazquez et al. (2005), Kim et al. (2008a) and Pan et al. (2008). Shin et al. (2004) found that the biogas produced from a thermophilic (55 C) culture was free of methane, whilst methane was detected under mesophilic (35 C) conditions, with hydrogen yields of 1.8 and 0.1 mol H2/mol hexose, respectively; the improved yield at higher temperatures was mirrored, in addition to the absence of methane, by negligible propionate concentrations. In semi-continuous fermentation, Valdez-Vazquez et al. (2005) observed that both the hydrogen content in the biogas and the production yield were higher under thermophilic (55 C) than under mesophilic (35 C) conditions, with values of 58 vs. 42% by vol. and 360 vs. 165 N ml H2/g VSremoved, respectively. Pan et al. (2008) studied the effect of temperature at different food/microorganisms (F/M) ratios on batch hydrogen production from mixed FW using anaerobic digestion sludge as the inoculum. F/M ratios of 710 g VS/g VSS were found to be adequate for hydro-

gen production via thermophilic (50 C) fermentation (maximum yield = 57 ml H2/g VS at F/M = 7), while under mesophilic (35 C) conditions hydrogen generation was decreased (maximum yield = 39 ml H2/g VS at F/M = 6). 3.4. pH The inuence of pH on hydrogen fermentation is also quite controversial in the literature. In general, pH is considered the most pivotal parameter due to its effects on hydrogenase activity, metabolic pathways as well as substrate hydrolysis. The H+ ion concentration in the system is also critical for maintaining adequate ATP levels in the system, since in the presence of an H+ excess ATP is used to ensure cell neutrality rather than to produce H2 (Nazlina et al., 2011). Several studies have been conducted on the optimal pH range for fermentative hydrogen production, however the results are often inconsistent due to differences in substrate, seed sludge and operating conditions adopted (Luo et al., 2010; Wu et al., 2010). In this regard it should be mentioned that numerous literature studies report the results of fermentation runs where only the initial pH was adjusted, without any further control along the process. However, it is noted that the importance of the initial pH may be overlooked when making direct comparison of results obtained at given initial pH values, since several factors among the others, substrate characteristics (composition and buffer capacity) and type of inoculum dictate the prevailing metabolic pathways and in turn pH evolution during the process, thus determining different hydrogen production yields and rates. It is acknowledged that low pHs result in inhibition of the hydrogenase activity, which is regarded to as a key factor explaining the inuence of pH on fermentative hydrogen production (Khanal et al., 2004; Nazlina et al., 2011). The metabolic pathways involving acetate and butyrate production appear to be favored in the pH range 4.56.0, while neutral or higher pHs are believed to promote ethanol and propionate production (Guo et al., 2010a; Rechtenbach et al., 2008; Rechtenbach and Stegmann, 2009). It should also be mentioned that hydrogen is mainly produced during the exponential growth phase of Clostridia, while in the stationary growth phase a shift from acidogenesis (with associated hydrogen generation) to solvent production is observed. It has been suggested (Khanal et al., 2004) that the shift occurs below pH 4.5, more precisely at pHs as low as 4.1, and the cause seems to involve the buildup of VFAs and H2 during the exponential growth phase. Solventogenesis is therefore assumed as a detoxication method of the biomass to avoid inhibitory effects caused by high VFA contents and associated low pHs in the liquid solution (Valdez-Vazquez and Poggi-Varaldo, 2009). However, other researchers observed a shift to solventogenesis at pH levels above 5.7, due to the synthesis or activation of the enzymes required for solvent production (Khanal et al., 2004). To this regard, Fang et al. (2006) observed a switch to solventogenesis occurring at pHs > 6.5. Other authors (Nazlina et al., 2011) indicated that decreasing pH below 6.0 increasingly promoted lactate formation, with an associated negative effect on the hydrogen production yield. The system pH may also affect the degree of biomass activity, with values <6 capable of signicantly inhibiting sulfate-reducing and methane-producing microorganisms. As far as homoacetogenic microorganisms are concerned, the effect of pH is not clear. Luo et al. (2010) observed homoacetogenesis inhibition at pHs of 45; however, since some homoacetogenic bacteria belong to the genus Clostridium, pH values in this range do not necessarily lead to inhibition. The initial pH is known to affect hydrogen production through its inuence on lag phase duration, spore germination (in those cases where a shock treatment is preliminary applied to the inoculum) and the synthesis of enzymes (Kim et al., 2011c). Khanal

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

1351

et al. (2004) found that an initial pH of 4.5 gave the highest specic hydrogen production potential for sucrose and starch, but it was accompanied by the lowest production rate and longest lag phase; at higher initial pHs, hydrogen production started earlier and with a higher production rate, but with a shorter overall duration. In the same study it was also observed that the lower the initial pH, the higher the maximum acetate/butyrate ratio was. However, Abreu et al. (2009) investigated hydrogen production from arabinose using four different anaerobic sludge samples at different initial pHs (4.58.0), and in all cases observed higher hydrogen production potentials as the initial pH increased. In the investigation of the effect of initial pH on hydrogen production from FW, batch tests are reported in the literature with values typically varying between 5 and 9 (Lay et al., 2003; Kim et al., 2008a, 2011c; Zhu et al., 2008; Dong et al., 2009a; Zong et al., 2009), without any further control during the test. However, different batch and semi-continuous/continuous experiments involved pH adjustment during operation, with set-point values between 5 (Liu et al., 2006; Kim et al., 2009; Nazlina et al., 2011) and 7 (Hong and Haiyun, 2010), but most commonly in the narrower range of 55.5 (Alzate-Gaviria et al., 2007; Chu et al., 2008; Kim et al., 2008a,b, 2010; Lee and Chung, 2010; Li et al., 2008b; Lee et al., 2010b; Liu et al., 2006). The pH values on FW were often adopted by these authors from other studies on simple substrates such as glucose. In the case of FW and OFMSW, relatively few experiments have been conducted to compare the effect of pH on hydrogen production. Shin et al. (2004) investigated fermentative hydrogen production from FW at pH 4.5, 5.5 and 6.5 under mesophilic and thermophilic conditions; the best results in terms of cumulative H2 production were attained at pHs of 4.5 and 5.5 under thermophilic conditions. Shin and Youn (2005) studied the performance of a semi-continuous system under thermophilic conditions at pHs of 5.0, 5.5 and 6.0, with a value of 5.5 yielding the best performance. The inuence of the initial (5.08.0) and operating (5.06.5) pH on hydrogen production from FW by Clostridium beijerinckii KCTC 1785 was investigated by Kim et al. (2008a); the optimal values were found to be an initial pH of 7.0 and an operating pH of 5.5, with associated values of hydrogen yield and rate of 128 N ml H2/g CODdegraded and 108 N ml H2/l h, respectively. Lee et al. (2008) investigated thermophilic hydrogen production from vegetable FW at pHs of 5.57.0, obtaining a maximum production rate of 0.48 mmol H2/g VS h at pH = 6, and a maximum yield of 0.57 mmol H2/g COD at pH = 7; no hydrogen production was observed at pH = 5.5. Batch tests on a mixture of FW, olive mill wastewater (OMW) and WAS (the latter with or without HST), were performed at pHs in the range 4.57.5 (Cappai et al., 2010). The best performance with the untreated inoculum was observed at pH = 6.5, with a cumulative hydrogen production of 42.9 N l H2/kg VS, which was also accompanied by the highest hydrogen content 51% by vol. in the biogas. Using the HST inoculum, the best results were obtained at pHs of 6.5 and 7.0 with production yields of around 60 N l H2/kg VS (5.6 N l H2/lreactor), while the highest hydrogen content in the biogas was measured at a pH of 7.5, with values up to 80% by vol. The higher hydrogen contents at high pHs were likely due to an indirect effect of enhanced CO2 dissolution in the liquid phase under alkaline conditions. A recent study by Nazlina et al. (2011) focused on batch thermophilic (55 C) digestion of FW at controlled pHs of 5.0, 5.5 and 6.0; the lowest H2 production yield (18 N ml H2/g substrate-COD) was observed at pH = 5.0, while comparable results (63 and 61 N ml H2/g substrate-COD, respectively) were obtained at pHs of 5.5 and 6.0. Higher pH conditions were also found to result in lower lactate production and higher removal of carbohydrates and volatile solids (VS), however these were also accompanied by a decreased concentration of Clostridia in the digestion medium (65% of total biomass as opposed to 92% at pH = 5.5), indicating

that other microbial phyla may have contributed to H2 production at higher pH levels. Although bacteria were enumerated, the identication of the other phyla was not performed. A signicant number of studies involving no external pH control have also been conducted (Han and Shin, 2002, 2004a,b; ValdezVazquez et al., 2005; Pan et al., 2008; Cappai et al., 2009; ValdezVazquez and Poggi-Varaldo, 2009; Wang and Zhao, 2009). The concept behind this strategy involves appropriate adjustment of the OLR to maintain suitable pH levels for signicant and stable hydrogen production. In semi-continuous experiments using completely mixed reactors, pH was maintained in the range 5.56.4 through biokinetic control attained by organic overloading (11 g VS/kg d; Valdez-Vazquez et al., 2005; Valdez-Vazquez and Poggi-Varaldo, 2009). Wang and Zhao (2009) report that a semi-continuous rotating drum reactor was capable of spontaneously maintaining pH within the range 5.25.8 at an OLR of 22.65 kg VS/m3 d. In another semi-continuous CSTR treating a mixture of FW and OMW, pH was maintained in the range 5.06.0 by adopting adequate values of the feed composition (25% w/w FW, 20% w/w OMW, 55% w/w AS), HRT (2 d) and OLR (32.3 kg VS/m3 d) (Cappai et al., 2009). In a thermophilic two-stage process (combined H2 + CH4 production; Lee et al., 2010b), pH was successfully maintained in the range 5.45.7 by recycling the efuent from the methanogenic stage into the acidogenic reactor. 3.5. Use of co-substrates FW and OFMSW have often been considered for co-digestion with other residues, including mainly primary sludge (PS) and WAS, but also wastes from agro-industrial activities. The use of co-substrates is motivated by other objectives being pursued concomitantly, including: (a) combined treatment of different waste streams, (b) ability to treat residues otherwise difcult to manage individually, (c) dilution of potentially toxic/inhibitory compounds, (d) resulting synergistic effects on biomass, (e) optimization of the conditions for hydrogen production, (f) internal control of pH, and (g) optimization of the carbohydrate/protein ratio. To this regard, although carbohydrates are the preferred substrate for fermentative hydrogen-producing bacteria such as those belonging to the Clostridium sp. while hydrogen is hardly produced from proteins and lipids, some experiences showed that, under some circumstances, proteins from waste sludge are necessary as a nitrogen source for hydrogen production in both pure and mixed cultures. In particular, Kim et al. (2004) found out that the addition of SS to FW up to 1319% by weight enhanced the hydrogen production potential; lately the same authors (Kim et al., 2011a) also indicated an optimal FW/sludge ratio for both the hydrogen production potential and the generation rate, which was equal to 10:1 w/w on a COD basis for the experimental conditions tested. Shin et al. (2003) showed that the hydrogen production yield decreased as WAS addition increased, due to the presence of methanogens in the sludge and the low carbohydrate concentration; however, sludge addition also enhanced hydrogen production due to the contribution of proteins. The maximum hydrogen yield of 59.2 ml/g VS was achieved at a FW/sludge ratio of 80:20 w/w. Zhu et al. (2008) tested different mixtures of FW, PS and WAS and found appropriate mixing ratios of the three substrates to promote hydrogen production (up to a maximum of 112 ml/g VSadded) due to an improved balance of carbohydrates, nitrogen, phosphorus and trace metals; moreover, PS and WAS showed a higher buffering capacity at low pHs in comparison to FW. Hong and Haiyun (2010) performed semi-continuous tests on mixtures of FW and dewatered sludge for optimization of the FW/sludge ratio and the operating parameters (HRT, OLR and pH); the best results were obtained for a FW content of 88% by wt., an HRT of 8.92 days, an OLR of 8.31 g VS/l d and a pH of 6.99. Semi-continuous digestion

1352

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

runs of mixtures of FW and OMW (Cappai et al., 2009) resulted in signicant and steady hydrogen production when a proper mixing ratio between the two was adopted; the best results in terms of specic hydrogen production were obtained using 25% w/w FW, 20% w/w OMW and 55% w/w WAS; the latter was used as both the inoculum and the main liquid phase to adjust the water content to wet conditions (10% TS); the outow was successfully subjected to a second methanogenic stage and the nal digested product was nally dewatered and composted after mixing with a bulk material. 4. Hydrogen production kinetics The evolution of hydrogen production over time in batch conditions is often described using the Gompertz equation, which has been modied from its original formulation to include parameters with a biological rather than a mathematical meaning. The modied Gompertz curve describes the time evolution of hydrogen generation using three parameters, namely the H2 production potential (Ps), the maximum H2 production rate (Rm) and the lag time (k), according to Eq. (6)):

   Rm e P Ps exp exp k t 1 Ps

the required information about working volumes, as well as relative amounts and composition of the individual mixture components, were provided. The results of the analysis of kinetic parameters are reported in Table 4. It can be noted that, depending on the specic type of substrate and inoculum used and the operating conditions adopted (F/M ratio, pH control, temperature, addition of nutrients, etc.), both Ps and Rm differ in the literature by up to three orders of magnitude (with maximum values in the ranges 0.2181 N ml H2/g VS for Ps and 0.1326 N ml H2/g VSS h for Rm), conrming that for optimization of biohydrogen production the relevant process parameters need to be carefully adjusted. For an appropriate evaluation of the process kinetics when comparing data from different experiments, since the values of Ps and Rm are interrelated so that the production rate cannot be interpreted in absolute terms without specifying the value for the associated production potential, an additional parameter, t95, was introduced. This is assumed to be the time required for hydrogen production to attain 95% of the maximum yield, and was derived from Ps and Rm rearranging the Gompertz function, as indicated by Eq. (7). Since t95 provides a measure of how fast the maximum production is achieved, it can be usefully adopted to compare, from a kinetic point of view, experimental conditions with different associated hydrogen generation yields.

6 t95 k Ps 1 ln ln 0:95 Rm e 7

In general, to compare experimental results obtained using different types/amounts of wastes as well as various mixing proportions of feed components and F/M ratios, the parameters of the modied Gompertz equation are expressed in specic terms in various ways, typically per unit mass of COD, VS, total solids (TS), hexose or carbohydrate-C from the waste in the case of Ps, and per unit mass of volatile suspended solids (VSS) from the inoculum in the case of Rm. As far as the H2 production potential is concerned, since it has been widely demonstrated that hydrogen production mainly derives from carbohydrates degradation (Lay et al., 2003; Kim et al., 2004; Han and Shin, 2004a,b; Argun et al., 2008; Chu et al., 2008; Dong et al., 2009a; Lee and Chung, 2010; Kim et al., 2011a; Nazlina et al., 2011), meaningful specic units are the mass or volume of H2 produced per unit mass of initial or removed carbohydrates (usually expressed in terms of their hexose equivalent). Other, for instance VS-specic, units may be more useful as design parameters, although their value is commonly substratedependent. Of course, the fact that the hydrogen production yield is expressed with different measures complicates the comparison of results from different studies, since the required conversion factors are often missing. Typical recently reported values/ranges are: 18205 N l H2/kg VS (Kim et al., 2004, 2011a; Chu et al., 2008); 52.5360 N l H2/kg VSremoved (Gmez et al., 2006; Valdez-Vazquez et al., 2005); 10133 N ml H2/g COD (Kim et al., 2004, 2011a; Lee et al., 2008; Li et al., 2008b); 0.692.10 mol H2/mol hexose (Kim et al., 2010, 2011a; Lee and Chung, 2010); 0.871.65 mol H2/mol hexoseremoved (Kim et al., 2009); 1.75.6 N l H2/lreactor (Cappai et al., 2010). Further data on process yields derived from individual studies are reported in Table 3. To derive more consistent and comparable values for the kinetics of fermentative hydrogen production under varying operating conditions, in the present review the parameters of the modied Gompertz equation were derived from several literature studies available; when Gompertz parameters were not directly provided in the papers, they were calculated through least-square tting of the experimental hydrogen production data with the theoretical curve (Eq. (6)). The whole set of kinetic parameters was then converted into homogeneous units so as to identify their respective ranges of variation as a function of the process conditions; this was only possible for a reduced number of references, for which

As for the lag phase of the hydrogen production process, the durations reported in recent studies on FW/OFMSW are mostly lower than 20 h (Kim et al., 2004; Shin et al., 2003), with minimum values as low as 0.11.92 h (Kim et al., 2004; Shin et al., 2004; Pan et al., 2008; Cappai et al., 2009). Notably higher values were reported by Lay et al. (2003) for a number of individual FW fractions: 72 h for rice and potato, 96 h for fat meat. However, shorter durations of the lag phase, namely 2.4 h for lettuce, 4.8 h for potato and 14.4 h for rice were reported by Dong et al. (2009a). The inuence of process temperature on the lag phase duration was discussed by Shin et al. (2004), who observed a shorter (0.1 3.6 h) lag phase under mesophilic compared to thermophilic conditions (1214.4 h). This was ascribed to the fact that the inoculum was exposed to room temperature before the tests, thus the thermophilic biomass activity was possibly affected by this relatively low temperature. Such ndings were conrmed by Pan et al. (2008), who observed lag phase durations of 0.054.9 h under mesophilic conditions vs. 3.45.3 h under thermophilic conditions. The biomass acclimation period can also be inuenced by pH. This issue was discussed by Shin et al. (2004), who found that, while under thermophilic conditions (55 C) the lag phase duration was not appreciably affected by pH in the range 4.56.5, under mesophilic conditions (35 C) increasing pHs resulted in shorter lag phase durations (from 0.1 h at pH = 6.5 to 3.6 h at pH = 4.5). These ndings were conrmed by Lee et al. (2008), who reported values of 3.8 h at pH = 6.57.0 and 7.9 h at pH = 6.0. In mesophilic (39 C) batch tests on FW and thermally treated WAS (Cappai et al., 2010), relatively slight differences in the lag phase duration were observed when pH was varied over the range 6.07.0, with values ranging from 4.6 h at pH 7.0 and 6.3 h at pH 7.5. The same authors however found that when digesting a mixture of FW, OMW and untreated WAS, pH exerted a stronger effect on the lag phase duration, with values decreasing from 15.9 to 6.8 h as pH increased from 4.5 to 6.5. The inuence of different FW pre-treatments on the lag phase duration was investigated by Kim et al. (2009). Shorter lag phases were observed when no pre-treatment was applied (7.3 h) or when the waste was maintained at pH = 1 for 1 d before fermentation (8.3 h); longer values (10.0 and 11.9 h, respectively) were found

Table 3 Operating parameters and performance data for H2 production from FW and OFMSW fermentation. Type of substrate Type of inoculum Inoculum pretreatment pH Value/range tested 5.7 6.07.5 4.56.5 6.07.0 4.58.5 4.58.5 5.5 (initial) 5.5 5.5 (initial) Optimal value/range 6.5 6.5 6.5 6.5 6.5 Reactor Type UASB PBR Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Operation mode Continuous Batch Batch Batch Batch Batch Batch Semicontinuous Batch 38 C 39 C 39 C 39 C 39 C 39 C 36 C 55 C 37 C 127 ml/g VSremoved 99 ml/g VSremoved 5181 ml/gVSadded 3.444 ml/gVSadded 69114 ml/gVSadded 2.358 ml/gVSadded 5.340 ml/gVSadded 37101 ml/g COD 205 ml/gVSadded 134 ml/gVSadded 106 ml/gVSadded 50 ml/gVSadded 0 6.25 ml/gVSadded 0 1.75 ml/gVSadded 332 ml/g VSadded 52.571.3 ml/g VSremoved 2728 ml/g VSadded 1926 ml/g VSadded 310 ml/g VSadded 34.7-155 ml/g VSaddedb 60.1 ml/g VSadded 128 ml/g CODdegraded 80.9 ml/g VSadded 62.6 ml/g VSadded 153.5 ml/g VSadded 0.350.54 mol/molhexose 162 ml/g VSadded 16.5137.2 ml/g VSadded 132 ml/g TVS 53.478.7 N ml/gVS 1.82 mol/molglucose 0.48 mmol/g CODadded 1.7 mmol/g CODadded (72 ml/g VSadded b) 114 ml/g VSadded Han and Shin (2002) Han and Shin (2004a) Han and Shin (2004b) Hong and Haiyun (2010) Kim et al. (2004) Kim et al. (2008a) Kim Kim Kim Kim Kim et al. (2008b) and Shin (2008) et al. (2009) et al. (2010) et al. (2011a) Alzate-Gaviria et al. (2007) Cappai et al. (2009) Cappai et al. (2010) Cappai et al. (2011) Cappai et al. (2011) Cappai et al. (2011) Chen et al. (2006) Chu et al. (2008) Dong et al. (2009a,b) T Specic H2 production Reference

OFMSW FW FW + OMW FW FW FW FW FW Rice Potatoes Lettuce Lean meat Oil Fat Banyan leaves FW OFMSW + slaughterhouse waste FW

Mixture of deep soil, excretes vaccine, pig excretes Activated sludge Activated sludge Activated sludge Activated sludge Anaerobic sludge Anaerobic sludge Anaerobic sludge

HST HST HST HST

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

Anaerobic sludge Mesophilic anaerobic sludge Anaerobic sludge Anaerobic sludge

Sonication

5.06.0 5.06.0 5.06.0 5.06.0

Stirred Stirred Stirred Stirred Leachingbed Leachingbed Leachingbed Stirred

FW FW FW FW (articial) + dewatered excess sludge FW + sewage sludge FW FW FW FW FW FW + sewage sludge FW OFMSW FW FW (liquid phase) Vegetable FW Vegetable FW OFMSW

Rumen microorganism Anaerobic sludge Anaerobic sludge Anaerobic sludge Anaerobic sludge Selected pure culture Anaerobic sludge Anaerobic sludge Hydrogen-producing sludge anaerobic sludge Anaerobic compost Anaerobic sludge Compost Compost Anaerobic sludge

Acclimation HST HST HST HST HST HST HST HST HST HST HST Acclimation HST

Semicontinuous Semicontinuous Semicontinuous Semicontinuous Continuous Continuous Continuous Semicontinuous Batch Batch SBR SBR Batch SBR Batch Batch Batch Batch Continuous Batch Semicontinuous Semi-

37 C 34 C 34 C 34 C 37 C 37 C 35 C 35 C 35 C 30 45 C 35 C 35 C 35 C 35 C 35 C 35 60 C 37 C 37 C 30 C 55 C 55 C 55 C

Elbeshbishy et al. (2011) Gmez et al. (2006) Gmez et al. (2009)

5.06.0 5.08.0 (initial) 5.3 5.3 5 5.3 8.0 (initial), 6.0 (operating value) 8.0 (initial), 6.0 (operating value) 7 (initial) 5.5 5.57.0 6 (initial) 55.7

7.0 (initial) 6.07.0

Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred Stirred

Kim et al. (2011b) Lay et al. (1999) Lay Lee Lee Lee et al. (2005) and Chung (2010) et al. (2008) et al. (2010a)

Lee et al. (2010b) 1353 (continued on next page)

1354

Table 3 (continued) Type of substrate Type of inoculum Inoculum pretreatment pH Value/range tested Optimal value/range 55.5 5.5 6.36.6 (initial) 5.7 (initial) 5.5 4.5 4.5 Reactor Type Operation mode continuous Batch Semicontinuous Batch Semicontinuous Batch Batch G. De Gioannis et al. / Waste Management 33 (2013) 13451361 36 C 35 C 37 C 37 C 55 C 50 C 35 C 55 C 35 C 35 C 55 C 30 C 30 C 60 C 55 C 55 C 37 C 40 C 35 C 193.85 ml/g VSadded 0.0460 mmol/L d (1 2 mmol/g COD) 0.727.4 mmol (pH > 4.5) 0 (pH 6 4.5) 10003000 ml/l db (2780 ml/g VSaddedb) 1863 ml/g CODadded b 57 ml/g VSadded 39 ml/g VSadded 0.41.0 l/l d 12.4 mol/mol hexosecons 34.059.2 ml/g VS 1.35.0 ml/g VSadded 28.446.3 ml/g VSadded 14.6104.8 ml/g VSadded 11.57102.63 ml/g VSadded 2.45.4 m3/m3 d 54.8 N ml/g VSremoved 360 ml/g VSremoved 165 ml/g VSremoved 4965 ml/g VSadded 112 ml/g VSadded Li et al. (2008a) Li et al. (2008b) Liu et al. (2006) T Specic H2 production Reference

FW + aged refuse FW FW

Sewage sludge Acidogenic sludge Anaerobic sludge

HST

5 5.35.6 3.58.5 5.06.0 5.06.0 5.06.0 4.55.56.5 5.5 5.86.0

Stirred Stirred Stirred Stirred Stirred Stirred

FW FW

Anaerobic sludge Thermoph. Anaerobic sludge mesoph. Anaerobic sludge Anaerobic sludge Anaerobic sludge Mesoph. Anaerobic sludge Thermoph. Anaerobic sludge Anaerobic sludge Anaerobic sludge Compost Anaerobic sludge Thermoph. anaerobic sludge mesoph. anaerobic sludge Anaerobic sludge

HST Acclimation HST HST HST Acclimation Acclimation

Nazlina et al. (2011) Pan et al. (2008)

FW FW + sewage sludge FW FW FW + sludge OFMSW + paper waste FW + paper FW + paper FW FW + primary sludge + sewage sludge
a

Stirred Stirred Stirred Stirred Stirred Stirred Unspecied Unspecied Rotating drum Stirred

Semicontinuous Batch Batch Batch Batch Continuous Semicontinuous Semicontinuous Semicontinuous Batch

Shin and Youn (2005) Shin et al. (2003) Shin et al. (2004) Sreela-or et al. (2011a) Sreela-or et al. (2011b) Ueno et al. (2007) Valdez-Vazquez and PoggiVaraldo (2009) Valdez-Vazquez et al. (2005) Wang and Zhao (2009) Zhu et al. (2008)

5.565.95 6.4 5.5 5.25.8 7.0 (initial) 5.56.0

FW collected from a dining hall. b Calculated from data provided by the authors.

G. De Gioannis et al. / Waste Management 33 (2013) 13451361 Table 4 Observed ranges for the parameters of the modied Gompertz equation as derived from different literature sources. Type of substrate Type of inoculum Ps (N ml H2/g VS) OFMSW FW FW FW FW FW FW FW Rice Potato Lettuce FW FW FW FW FW FW
a

1355

Rm (N ml H2/l h) Min 0.19 Max 0.44 Min 1.57 Max 11.93 Min 17.6 Max 135.2 37.3 31 16.3 Min 503 Max 1320 Min 300 Max 710 Min 27 Max 369.1 Min 3.2 Max 38.9 Min 5.0 Max 163.7

Rm (N ml H2/g VSS h) Min 0.1 Max 16.8 Min 0.3 Max 0.7 Min 2.5 Max 19.0 Min 4.4 Max 21.6 Min 39 Max 286 Min 0 Max 12.8 Min 0.3 Max 30.3 Min 0 Max 12.7 4.7 3.9 2.0 Min 0a Max 9.5a Min 103 Max 326 Min 34.8 Max 118.1 Min 2.2 Max 16.9 Min 6.8 Max 61.6

t95k (h) Min 12.3 Max 43.1 Min 23.4 Max 115.1 Min 3.9 Max 10.4

Reference

Anaerobic sludge + acclimated H2-producing culture Acclimated mesophilic culture Acclimated thermophilic culture Anaerobic grass compost Anaerobic digester sludge Compost Thermophilic anaerobic sludge Mesophilic anaerobic sludge Acclimated anaerobic sludge Acclimated anaerobic sludge Acclimated anaerobic sludge None Aerobic sludge Untreated sewage sludge None Anaerobic sludge Anaerobic sludge

Min 0.2 Max 117 Min 1.3 Max 5.0 Min 28.4 Max 91.5 Min 53.4 Max 78.7 Min 66.1 Max 180.6 Min 0 Max 24.1 Min 3.2 Max 54.2 Min 0.03 Max 39.2 132 102 48 Min 0 Max 148.7 Min 62.9 Max 131.9 Min 147.0 Max 175.2 Min 16.5 Max 137.2 Min 14.6 Max 104.8 Min 11.57 Max 102.63

Lay et al. (1999) Shin et al. (2004) Shin et al. (2004) Lay et al. (2005) Chen et al. (2006) Lee et al. (2008)

Min 12.4 Max 36.3 Min 17.7 Max 30.1 41.3 38.4 34.3 Min 9.8 Max 13.5 Min 15.1 Max 30.8 Min 11.4 Max 22.8 Min 5.3 Max 27.5

Pan et al. (2008) Pan et al. (2008) Dong et al. (2009a,b) Dong et al. (2009a,b) Dong et al. (2009a,b) Kim et al. (2009) Cappai et al. (2011) Kim et al. (2011a) Kim et al. (2011b) Sreela-or et al. (2011a) Sreela-or et al. (2011b)

Expressed per unit of substrate-VS.

for thermal (90 C, 20 min) and alkaline (pH = 13 for 1 d) pretreatment. The overall duration of the hydrogen production phase was typically observed to range between 1 (Lee et al., 2008; Cappai et al., 2009, 2010; Kim et al., 2009) and 4 d (Han and Shin, 2004b; Kim et al., 2004, 2009; Shin et al. 2004; Zong et al., 2009). Values as short as 9.5 h were reported by Lee et al. (2008) (vegetable FW, pH controlled at 6.5; thermophilic conditions) and Shin et al. (2004) (FW, pH controlled at 5.5 and 6.5, mesophilic conditions). In the same work, when pH was controlled at 4.5 and thermophilic conditions were adopted, a signicantly longer duration (5.8 d) of hydrogen production was observed; prolonged production periods were also reported by Lay et al. (2003) for individual FW fractions, with values of 89 d for rice and potato and 6.6 d for fat meat. 4.1. Statistical analysis of hydrogen production data The experimental data of hydrogen production potential (Ps) from the reviewed literature studies were processed to derive information on the relative importance accommodating for the existing relationships among the main variables of relevance. To this aim, literature data were ltered and, when feasible, reprocessed and converted into homogeneous units as described in Section 4, resulting in 198 individual data points from 15 different publications (Lay et al., 1999; Shin et al., 2003, 2004; Chen et al., 2006; Lee et al., 2008; Pan et al., 2008; Zhu et al., 2008; Cappai et al., 2009, 2010, 2011; Kim et al., 2009, 2011a,b; Sreela-or et al., 2011a,b) being used for the statistical regression analysis. The input variables used in the analysis and the associated levels are reported in Table 5, while the response variable was the hydrogen production potential.

The statistical technique known as recursive partitioning was applied for the analysis of data. This was used as a means to build a exible, parsimonious regression model that can be represented by a binary regression tree; this can also be seen as a way to automatically identify the most important variables affecting the response while accounting for associations among the explanatory variables. A regression tree is constructed by recursively partitioning the data set into two homogeneous groups (son nodes) according to some criterion, and then splitting the nodes up further on each of the branches. On each node the response is tted by the node average, which implies dening a stepwise constant tted response surface. Recursive partitioning (Hothorn et al., 2006) thus involves separating statistical groups progressively decreasing in size and increasing in internal homogeneity in terms of the statistical distribution of the response variable. In the present case, splitting was implemented using a conditional inference criterion, implying testing the global null hypothesis of independence between any of the input variables and the response. If the null hypothesis cannot be rejected, the splitting procedure at that node is stopped, and this becomes a terminal node of the tree; otherwise, the input variable exhibiting the strongest association with the response (as measured by the corresponding p-value) is selected and a binary split in it is implemented. The splitting procedure continues until each node becomes a terminal node according to the above mentioned condition. The graphical output of the recursive partitioning procedure, which was implemented using the party package in the statistical software R (Strobl et al., 2009), is shown in Fig. 1. At each terminal node the number of data points (n) and their statistical distribution (indicated by the associated box plots) are provided. The input variables identied as the most important in affecting maximum

1356

G. De Gioannis et al. / Waste Management 33 (2013) 13451361 Table 5 Input variables used for the regression analysis of hydrogen production data. Variable Type of substrate Type of co-substrate Symbol Substr_type Co-substr_type No. levelsa 2 5 Valuesa 1 = Food waste 2 = Vegetable kitchen waste 1 = No co-substrate added 2 = Primary sludge 3 = Activated sludge 4 = OMW 5 = Nightsoil + sewage sludge 6 = Primary + activated sludge 1 = No inoculum added 2 = Anaerobic sludge 3 = Compost 4 = Thermophilic anaerobic sludge 5 = Pre-selected H2-producing inoculum 6 = Activated sludge 1 = No pre-treatment applied 2 = Thermal pre-treatment 1 = No control 2 = Controlled 1 = No control 2 = Continuously controlled 3 = Controlled with buffer addition 1 = No buffer added 2 = Citrate 3 = Disodium hydrogen phosphate 1 = No nutrient added 2 = Nutrient added

Co-substrate/substrate mass ratio Type of inoculum

Co-substr/substr Inoc_type

F/M mass ratio Type of pre-treatment Pre-treatment temperature Pre-treatment duration Control of initial pH Initial pH value Control of operating pH

F/M Pretr_type Pretr_T Pretr_duration pHin_contr pHin pH_contr

2 2 3

Operating pH value Buffer type

pH Buffer_type

Buffer dosage Fermentation temperature Nutrient addition

a

Buffer_dos Ferm_T Nut_add

Only reported for qualitative (discrete) variables.

Fig. 1. Regression tree obtained from recursive partitioning. The number of data points (n) and the statistical distribution of Ps (indicated by the associated box plots) are shown for each terminal node, while the average value of Ps is provided at each splitting point. All values in N l H2/kg VS.

hydrogen production are: type of co-substrate, type of pre-treatment, operating pH, control of initial pH and fermentation temperature. Six terminal nodes were identied, which differed both for

the average value and the distribution of the response variable. The type of co-substrate was found at the highest hierarchical level of the tree, splitting the dataset into two groups, with average Ps

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

1357

values of 56 N l H2/kg VS on the left branch and 25 N l H2/kg VS on the right branch. When inspecting the tree towards the left, the type of pre-treatment results in different maximum production yields, with average values 43 and 66 N l H2/kg VS on the right and left branches of node 2, respectively. In synthesis, the highest hydrogen production potential (average value 87 N l H2/kg VS) was found to be attained for the following combination of input variables: Co-substr_type = {none, primary + activated sludge, nightsoil + sewage sludge}, pretr_type = {thermal} and pH > 5.0. On the other hand, the combination of input variables yielding the worst mean response (14 N l H2/kg VS) was: Co-substr_type = {none, primary + activated sludge, nightsoil + sewage sludge}, pretr_type = {none}, pHin_control = {none} and Ferm_T 6 39 C (i.e. mesophilic conditions). As for the general validity of the results of the statistical analysis performed, it should nevertheless be stressed that in some cases (see nodes 4, 5 and 7 in Fig. 1) the dispersion of data around the corresponding value (the node mean) was appreciable. As the objective of the analysis was to assess the relative relevance of the identied input variables on the yield of the fermentation process, this feature is not deemed to affect the ndings of the analysis. Nonetheless, this feature may suggest the presence of hidden variables which may explain the residual heterogeneity within the individual terminal nodes. Unfortunately, the information that can be retrieved from literature data is not sufcient to resolve such an ambiguity. However, the obtained results appear to suggest that, in order to allow for comparison of data from different literature sources and build reliable predictive models for the fermentative hydrogen production process, a high level of consistency between data is strictly required. This requires both a harmonization of the way in which the parameters of interest for the process

are expressed and a more accurate description of the experimental conditions adopted in the fermentative H2 literature. 5. Hybrid processes Although according to some authors (Lee and Chung, 2010) fermentative hydrogen production from FW may be economically viable, it is also generally acknowledged that, since the majority of the organic content of the original substrate remains undegraded, the process should be combined with a second treatment stage to achieve substrate stabilization and increase energy conversion. The second stage may thus be oriented to producing either additional hydrogen or methane, with a variety of potential alternatives differing in the type of process applied and/or the characteristics of the resulting product (Fig. 2). On one instance the residual organic content of the waste feed, which is mainly in the form of the soluble products from the hydrolytic stage, may be converted into methane in a second-stage reactor where suitable environmental conditions for methanogens are maintained. Alternatively, the hydrogen content still stored in the efuent from the rst stage may be further exploited through other biological processes including photo-fermentation or microbial electro-hydrogenogenesis. As for combined two-stage biological H2 + CH4 production, which is one of the most common strategies proposed, the overall anaerobic digestion process can be described by the Eq. (8):

C6 H12 O6 2H2 O ! 2CH4 4CO2 4H2

If the energy conversion efciency of the process is calculated through Eq. (4), assuming a lower heating value of 801 kJ/mol for methane, a value of 89.0% is obtained. Considering the conventional single-stage methane production process (C6H12 O6 ?

Fig. 2. Schematization of hybrid processes.

1358

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

3CH4 + 3CO2), a conversion yield of 83.2% would be obtained. This supports the assumption that combined H2 + CH4 production is, from a theoretical point of view, energetically more favorable than conventional anaerobic digestion (Dong et al., 2011). Combined H2 + CH4 production from FW has been studied through semi-continuous or continuous tests (Cappai et al., 2009; Chu et al., 2008; Han and Shin, 2004a,b; Lee and Chung, 2010; Lee et al., 2010b; Liu et al., 2006; Wang and Zhao, 2009). As mentioned above, HRT and OLR values for the methanogenic stage vary in the ranges 4 27 d and 38 kg VS/m3 d (Wang and Zhao, 2009) or 416 kg COD/ m3 d (Lee et al., 2010a,b; Chu et al., 2008). While in most studies single-stage methane production was conducted under mesophilic conditions (typically 37 C) (Han and Shin, 2004a,b; Liu et al., 2006; Zong et al., 2009), combined H2 + CH4 generation from FW was also investigated under thermophilic (55 C) conditions (Lee et al., 2010b). Irrespective of whether inoculum was pre-treated (Chu et al., 2008; Liu et al., 2006; Wang and Zhao, 2009) or used as received (Han and Shin, 2004a,b; Lee and Chung, 2010; Lee et al., 2010b), the two-stage process was found to be capable of signicant H2 + CH4 production. To this regard, Liu et al. (2006) observed that 100 C HST of the inoculum did not adversely affect the methanogenic stage. CH4 yields were reported to fall within the range 460550 N l/kg VS (Chu et al., 2008; Liu et al., 2006; Wang and Zhao, 2009). Although few studies compared conventional singlestage methane production and two-stage H2 + CH4 production, Liu et al. (2006) found that under semi-continuous conditions the latter process generated 21% more methane, and, as expected, a much lower amount of VFAs in the nal efuent. Interesting economic considerations were derived by Lee and Chung (2010), who managed a two-stage pilot-scale H2 + CH4 fermentation process treating FW, connected to a fuel cell. Comparing H2-only fermentation, CH4-only fermentation, and combined H2 + CH4 fermentation, while negligible differences in production costs among the three systems were estimated, an increase in energy production by 1225% was observed for the combined system. Further hydrogen production may also be derived by combining dark fermentation with photo-fermentation. Purple non-sulfur photosynthetic bacteria are capable of using short-chain organic acids as electron donors to generate H2 through a light-driven metabolism; thus, the metabolic products of dark fermentation represent a potential substrate for photo-fermentative bacteria. Zong et al. (2009) studied a two-stage batch process including dark- and photo-fermentation in order to produce hydrogen from FW. Cattle dung compost mixed with water (1:10 w/v) and heat-treated for 15 min was used as the inoculum for dark fermentation, whilst R. sphaeroides ZX-5 was used as the inoculum for photo-fermentation. Acetate, butyrate and butanol were found to be consumed during 168-h photo-fermentation. The hydrogen yield achieved in the individual stages of dark and photo-fermentation was 1.77 and 3.63 mol/mol hexose, respectively, with an overall yield of 5.4 mol H2/mol hexose, equivalent to a conversion efciency of 45%. Bio-electrochemical systems have also been proposed as methods to couple with fermentative hydrogen production. Among these, microbial electrolysis cells (MECs, initially developed by Liu et al., 2005) are based on microbially-mediated oxidation of organic substances in an anodic compartment, with the aid of an external circuit where an external power supply is provided. The electrons generated by the degradation process are transferred through the external circuit to a cathodic compartment, while protons are transferred through the ion exchange membrane that separates the two compartments. In the cathodic compartment electrons reduce the protons generated by the biological process or from dissociated water, producing H2 (Liu et al., 2005; Logan et al., 2008; Jeremiasse et al., 2010). For electrochemically-driven hydrogen production to occur in such systems, it has been demonstrated that a voltage needs to be applied, which theoretically

amounts to 0.14 V if acetate is assumed as the reference organic substrate (Logan et al., 2008). Although to the authors knowledge there are no specic examples in the literature of combination of dark fermentation and MECs for hydrogen production from FW, some studies exist of application to different types of either pure or residual substrates, including pure VFAs (Guo et al., 2010a,b; Kyazze et al., 2010; Manuel et al., 2010; Cheng and Logan, 2011), wastewaters (Lu et al., 2009), lignocellulosic materials (corn stover [Lalaurette et al., 2009], cellulose [Wang et al., 2011]), wheat powder (Tuna et al., 2009). However, since the nature of the metabolic products of acidogenesis is similar for different types of source substrates and MECs have been found capable of using different substrates including fermentable and non-fermentable organics, the results of the mentioned studies suggest the feasibility of a two-stage dark fermentation + MEC process for combined hydrogen production from FW/OFMSW. Another candidate process for a second-stage treatment after dark fermentation is the generation of electric current through microbial fuel cells (MFCs), where bacteria catalyze the oxidation of organic acids from the acidogenic phase of dark fermentation. If the electrons generated by the oxidation reactions are transferred from the anode (biological compartment) to a cathode through an external circuit, the electron ow produces electricity (Logan et al., 2006). In the literature, studies are documented of combination of dark fermentation and MFCs, using either pure substrates such as glucose (Sharma and Li, 2010), synthetic dark fermentation efuent (Poggi-Varaldo et al., 2009; Vzquez-Larios et al., 2011), cellulose (Wang et al., 2011) and vegetable waste (Mohanakrishna et al., 2010). In their study on vegetable waste, Mohanakrishna et al. (2010) found that the MFC performance was improved when the system was fed with the pre-fermented waste instead of the untreated waste. This can be ascribed to the fact that adequate MFC operation requires hydrolysis of particulate organic matter to make this available to the biomass.

6. Conclusions and perspectives The analysis of over 80 literature references on fermentative hydrogen production from FW and OFMSW has shown that numerous process parameters have the potential of affecting the evolution of the metabolic pathways involved, in turn affecting the process kinetics and the conversion yield. As indicated by the review performed, the main parameters of concern include pH, temperature, solids retention time, inoculum addition/type/pretreatment, presence of co-substrates, reactor conguration, reactor operation mode, combination with additional processes. The strong inuence exerted by the individual parameters mentioned as well as the existence of mutual interactions can in fact lead to variations up to three orders of magnitude in process performance depending on the specic combination of the operating variables adopted. At present, given the existing uncertainties about the individual and joint inuence of such parameters, prediction of full-scale reactors performance based on the existing data may turn out to be unreliable, therefore further systematic study on this issue is strongly recommended. As to the process indicators, a variety of parameters have been proposed in the literature which have been used to evaluate the process performance from different perspectives, spanning from the production potential to the process kinetics, from the biogas composition to the energy conversion efciency. Accordingly, parameters including the specic production potential, the production rate, the time required to attain a given fraction of the maximum production, the H2 content in the biogas, the energy conversion yield are typically adopted as process performance indicators. In the authors opinion, care should be taken when

G. De Gioannis et al. / Waste Management 33 (2013) 13451361

1359

adopting a given indicator in place of another to monitor the process efciency as well as when using specic parameters to compare results obtained under different operating conditions. For example, some parameters are strictly dependent on the intrinsic characteristics of the substrate concerned, and as such should not be used for comparison purposes unless considering the same type of substrate. Secondly, for a reliable assessment of the overall performance of the fermentation process to be attained, the use of multiple parameters appears to be recommended, so as to account for the complexity and the interrelations between the numerous factors of concern. With regard to the above mentioned issues, the authors believe that some considerable effort should be made by the scientic community to harmonize the description methods adopted, with the aim of allowing comparison of results from different sources and gaining an improved understanding of the numerous interrelations between the underlying factors which govern the fermentation process. It is believed that this may also help explaining the controversies currently found in the results obtained using different approaches and methods, resolving apparently contrasting conclusions derived from such results. As to the implementation potential of fermentative hydrogen production, although the technical feasibility in case of processing of simple substrates has been demonstrated by many papers, the technology still appears to be in an early stage (especially for complex substrates), and to the authors knowledge no single full scale plant is operative yet. For such reasons, considerable efforts appear to be required to assess the potential for full-scale application of the process, from both a technical and a global economic perspective. Nevertheless, a number of issues may be mentioned in regard to the technical feasibility of the process on the basis of the state of the art of knowledge to date. In general, it is believed that the technology would have a rather limited scope if complex organic matter could not be effectively used as the substrate; furthermore, an expanded number of substrate types and the use of mixed cultures as the inoculum would drastically improve the chance of successful development of the process. To this regard, several experiences have demonstrated that appreciable hydrogen production from FW and OFMSW can be attained using the indigenous biomass present in the substrate. An issue which deserves signicant attention specically from an engineering point of view is the possibility of optimizing the process performance by appropriately adjusting process conguration and operation, with no need for external control of the operating variables or for application of severe conditions. In regard to the technical sustainability of biohydrogen production, it is generally acknowledged that since fermentation alone converts, already under optimal conditions, no more than 33% of the chemical energy contained in substrate (whatever this may be), the process should be combined with a second treatment stage aimed at improved substrate stabilization and enhanced energy conversion. The residual organic content of the waste feed, which is mainly in the form of soluble products of the hydrolytic stage (organic acids and alcohols), may benecially be converted into methane in a second-stage reactor. Alternatively, the hydrogen content which is still stored in the efuent from the rst stage may be recovered through other biological processes including photo-fermentation or microbial electro-hydrogenogenesis. Both the mentioned alternatives to methanogenesis are still at a relatively early stage and as such deserve further investigation. References
Abreu, A.A., Danko, A.S., Costa, J.C., Ferreira, E.C., Alves, M.M., 2009. Inoculum type response to different pHs on biohydrogen production from L-arabinose, a

component of hemicellulosic biopolymers. Int. J. Hydrogen Energy 34, 1744 1751. Alzate-Gaviria, L.M., Sebastian, P.J., Perez-Hernandez, A., Eapen, D., 2007. Comparison of two anaerobic systems for hydrogen production from the organic fraction of municipal solid waste and synthetic wastewater. Int. J. Hydrogen Energy 32, 31413146. Angenent, L.T., Karim, K., Al-Dahhan, M.H., Wrenn, B.A., Domguez-Espinosa, R., 2004. Production of bioenergy and biochemicals from industrial and agricultural wastewater. Trends Biotechnol. 22, 477485. Argun, H., Kargi, F., Kapdan, I.K., Oztekin, R., 2008. Batch dark fermentation of powdered wheat starch to hydrogen gas: effects of the initial substrate and biomass concentration. Int. J. Hydrogen Energy 33, 61096115. Bhaskar, Y.V., Mohan, S.V., Sarma, P.N., 2005. Effect of substrate loading rate of chemical wastewater on fermentative biohydrogen production in biolm congured sequencing batch reactor. Bioresource Technol. 99, 69416948. Bolzonella, D., Pavan, P., Mace, S., Cecchi, F., 2006. Dry anaerobic digestion of differently sorted organic municipal solid waste: a full-scale experience. Water Sci. Technol. 53, 2332. Cappai, G., De Gioannis, G., Giordano, A., Muntoni, A., Polettini, A., Pomi, R., 2009. Energy and material recovery through combined hydrogen-methane production and nal composting of different solid and liquid waste (HyMeC concept). In: Proceedings of Sardinia 2009 Twelfth International Waste Management and Landll Symposium, S. Margherita di Pula (I), pp. 885886 (volume of abstracts, full paper on CD ROM). Cappai, G., De Gioannis, G., Giordano, A., Muntoni, A., Polettini, A., Pomi, R., 2010. Assessment through batch tests of hydrogen production from mixtures of biodegradable residues. In: Proceedings of Venice 2010, Third International Symposium on Energy from Biomass and Waste, Venice (I) (on CD ROM). Cappai, G., De Gioannis, G., Giordano, G., Muntoni, A., Polettini, A., Pomi, R., Spiga, D., 2011. Batch hydrogen production from food waste. In: Proc. Sardinia 2011, Thirteenth International Waste Management and Landll Symposium, S. Margherita di Pula (CA) (on CD ROM). Chen, W.H., Chen, S.Y., Khanal, S.K., Sung, S., 2006. Kinetic study of biological hydrogen production by anaerobic fermentation. Int. J. Hydrogen Energy 31, 21702178. Cheng, S., Logan, B.E., 2011. High hydrogen production rate of microbial electrolysis cell (MEC) with reduced electrode spacing. Bioresource Technol. 102, 3571 3574. Chidthaisong, A., Conrad, R., 2000. Specicity of chloroform, 2bromoethanesulfonate and uoroacetate to inhibit methanogenesis and other anaerobic processes in anoxic rice eld soil. Soil Biol. Biochem. 32, 977988. Chu, F.C., Yu, Y.L., Kai, Q.X., Yoshitaka, E., Yuhei, I., Hai, N.K., 2008. A pH- and temperature-phased two-stage process for hydrogen and methane production from food waste. Int. J. Hydrogen Energy 33, 47394746. Davila-Vazquez, G., Arriaga, S., Alatriste-Mondragn, F., de Len-Rodrguez, A., Rosales-Colunga, L.M., Razo-Flores, E., 2008. Fermentative biohydrogen production: trends and perspectives. Rev. Environ. Sci. Biotechnol. 7, 2745. De Baere, L., 2003. State-of-the-art of anaerobic digestion of municipal solid waste. In: Proc. Sardinia 2003, Ninth International Waste Management and Landll Symposium, S. Margherita di Pula, Cagliari, Italy, 610 October 2003 (on CD ROM). Dong, L., Zhenhong, Y., Yongming, S., Xiaoying, K., Yu, Z., 2009a. Hydrogen production characteristics of the organic fraction of municipal solid wastes by anaerobic mixedculture fermentation. Int. J. Hydrogen Energy 34, 812 820. Dong, L., Zhenhong, Y., Yongming, S., Longlong, M.A., Lianhua, L., 2009b. Sequential anaerobic fermentative production of hydrogen and methane from organic fraction of municipal solid waste. Chin. J. Appl. Environ. Biol. 15, 250257. Dong, L., Zhenhong, Y., Yongming, S., Longlong, M., 2011. Anaerobic fermentative co-production of hydrogen and methane from an organic fraction of municipal solid waste. Energy Source Part A 33, 575585. Elbeshbishy, E., Hafez, H., Nakhla, G., 2011. Ultrasonication for biohydrogen production from food waste. Int. J. Hydrogen Energy 36, 28962903. Fang, H.H.P., Li, C., Zhang, T., 2006. Acidophilic biohydrogen production from rice slurry. Int. J Hydrogen Energ 31, 683692. Gmez, X., Morn, A., Cuetos, M.J., Sanchez, M.E., 2006. The production of hydrogen by dark fermentation of municipal solid wastes and slaughterhouse waste: a two-phase process. J. Power Sources 157, 727732. Gmez, X., Cuetos, M.J., Prieto, J.I., Morn, A., 2009. Bio-hydrogen production from waste fermentation: mixing and static conditions. Renew. Energy 34, 970975. Guo, X.M., Trably, E., Latrille, E., Carrre, H., Steyer, J.P., 2010a. Hydrogen production from agricultural waste by dark fermentation: a review. Int. J. Hydrogen Energy 35, 1066010673. Guo, K., Tang, X., Du, Z., Li, H., 2010b. Hydrogen production from acetate in a cathode-on-top single-chamber microbial electrolysis cell with a mipor cathode. Biochem. Eng. J. 51, 4852. Hallenbeck, P.C., Ghosh, D., 2009. Advances in fermentative biohydrogen production: the way forward? Trends Biotechnol. 27, 287297. Han, S.K., Shin, H.S., 2002. Enhanced acidogenic fermentation of food waste in a continuous-ow reactor. Waste Manage. Res. 20, 110118. Han, S.K., Shin, H.S., 2004a. Performance of an innovative two-stage process converting food waste to hydrogen and methane. J. Air Waste Manage. Assoc. 54, 242249. Han, S.K., Shin, H.S., 2004b. Biohydrogen production by anaerobic fermentation of food waste. Int. J. Hydrogen Energy 29, 569577.

1360

G. De Gioannis et al. / Waste Management 33 (2013) 13451361 Liu, H., Wang, J., Wang, A., Chen, J., 2011. Chemical inhibitors of methanogenesis and putative applications. Appl. Microbiol. Biotechnol. 89, 13331340. Logan, B.E., Hamelers, B., Rozendal, R., Schrder, U., Keller, J., Freguia, S., Aelterman, P., Verstraete, W., Rabaey, K., 2006. Microbial fuel cells: methodology and technology. Environ. Sci. Technol. 40, 51815192. Logan, B.E., Call, D., Cheng, S., Hamelers, H.V.M., Sleutels, T.H.J.A., Jeremiasse, A.W., Rozendal, R.A., 2008. Microbial electrolysis cells for high yield hydrogen gas production from organic matter. Environ. Sci. Technol. 42, 86308640. Luo, G., Xie, L., Zou, Z., Zhou, Q., Wang, J.Y., 2010. Fermentative hydrogen production from cassava stillage by mixed anaerobic microora: effects of temperature and pH. Appl. Energy 87 (12), 37103717. Lu, L., Ren, N., Xing, D., Logan, B.E., 2009. Hydrogen production with efuent from an ethanolH2-coproducing fermentation reactor using a single-chamber microbial electrolysis cell. Biosens. Bioelectron. 24, 30553060. Manuel, M.-F., Neburchilov, V., Wang, H., Guiot, S.R., Tartakovsky, B., 2010. Hydrogen production in a microbial electrolysis cell with nickel-based gas diffusion cathodes. J. Power Sources 195, 55145519. Mata-Alvarez, J. (Ed.), 2002. Biomethanization of the Organic Fraction of Municipal Solid Wastes. IWA Publishing, London, UK. Mata-Alvarez, J., Mace, S., Llabres, P., 2000. Anaerobic digestion of organic solid wastes. An overview of research achievements and perspectives. Bioresource Technol. 74, 316. Mohanakrishna, G., Venkata Mohan, S., Sarma, P.N., 2010. Utilizing acid-rich efuents of fermentative hydrogen production process as substrate for harnessing bioelectricity: an integrative approach. Int. J. Hydrogen Energy 35, 34403449. Nath, K., Das, D., 2004. Improvement of fermentative hydrogen production: various approaches. Appl. Microbiol. Biotechnol. 65, 520529. Nazlina, H.M.Y.N.H., NorAini, A.R., Man, H.C., Yusoff, M.Z.M., Hassan, M.A., 2011. Microbial characterization of hydrogen-producing bacteria in fermented food waste at different pH values. Int. J. Hydrogen Energy 36, 95719580. Oh, Y.K., Kim, S.H., Kim, M.S., Park, S., 2004. Thermophilic biohydrogen production from glucose with trickling biolter. Biotechnol. Bioeng. 88, 690698. Okamoto, M., Miyahara, T., Mizuno, O., Noike, T., 2000. The relative effectiveness of pH control and heat treatment for enhancing biohydrogen gas production. Water Sci. Technol. 41 (3), 2532. Pan, J., Zhang, R., El-Mashad, H.M., Sun, H., Ying, Y., 2008. Effect of food to microorganism ratio on biohydrogen production from food waste via anaerobic fermentation. Int. J. Hydrogen Energy 33, 69686975. Poggi-Varaldo, H.M., Carmona-Martnez, A., Vzquez-Larios, A.L., Solorza-Feria, O., 2009. Effect of inoculum type on the performance of a microbial fuel cell fed with spent organic extracts from hydrogenogenic fermentation of organic solid wastes. J. New Mater. Electrochem. Syst. 12, 4954. Rechtenbach, D., Stegmann, R., 2009. Combined bio-hydrogen and methane production. In: Proceedings of Sardinia 2009 Twelfth International Waste Management and Landll Symposium, S. Margherita di Pula (I), 59 October 2009, pp. 7980 (manuscript on CD ROM, 11 pp.). Rechtenbach, D., Meyer, M., Stegmann, R., 2008. (Dis-) continuous production of biohydrogen and biomethane from raw and waste materials by fermentation. In: Proceedings of Venice 2008 Second International Symposium on Energy from Biomass and Waste, Venice (I), 1720 November 2008 (manuscript on CD ROM, 10 pp.). Sharma, Y., Li, B., 2010. Optimizing energy harvest in wastewater treatment by combining anaerobic hydrogen producing biofermentor (HPB) and microbial fuel cell (MFC). Int. J. Hydrogen Energy 35, 37893797. Shin, H.S., Youn, J.H., 2005. Conversion of food waste into hydrogen by thermophilic acidogenesis. Biodegradation 16, 3344. Shin, H.S., Kim, S.H., Paik, B.C., 2003. Characteristics of hydrogen production from food waste and waste activated sludge. J. Water Environ. Technol. 1, 177187. Shin, H.S., Youn, J.H., Kim, S.H., 2004. Hydrogen production from food waste in anaerobic mesophilic and thermophilic acidogenesis. Int. J. Hydrogen Energy 29, 13551363. Sreela-or, C., Imai, T., Plangklang, P., Reungsang, A., 2011a. Optimization of key factors affecting hydrogen production from food waste by anaerobic mixed cultures. Int. J. Hydrogen Energy 36, 1412014133. Sreela-or, C., Plangklang, P., Imai, T., Reungsang, A., 2011b. Co-digestion of food waste and sludge for hydrogen production by anaerobic mixed cultures: statistical key factors optimization. Int. J. Hydrogen Energy 36, 14227 14237. Strobl, C., Hothorn, T., Zeileis, A., 2009. Party on! a new, conditional variableimportance measure for random forests available in the party package. R J. 1, 1417. Thauer, R.K., Jungermann, K.A., Decker, K., 1977. Energy conservation in chemotrophic anaerobic bacteria. Bacteriol. Rev. 41, 100180. Tuna, E., Kargi, F., Argun, H., 2009. Hydrogen gas production by electrohydrolysis of volatile fatty acid (VFA) containing dark fermentation efuent. Int. J. Hydrogen Energy 34, 262269. Ueno, Y., Fukui, H., Goto, M., 2007. Operation of a two-stage fermentation process producing hydrogen and methane from organic waste. Environ. Sci. Technol. 41, 14131419. Valdez-Vazquez, I., Poggi-Varaldo, H., 2009. Alkalinity and high total solids affecting H2 production from organic solid waste by anaerobic consortia. Int. J. Hydrogen Energy 34, 36393646. Valdez-Vazquez, I., Rios-Leal, E., Esparza-Garca, F., Cecchi, F., Poggi-Varaldo, H., 2005. Semi-continuous solid substrate anaerobic reactors for H2 production

Hong, C., Haiyun, W., 2010. Optimization of volatile fatty acid production with cosubstrate of food wastes and dewatered excess sludge using response surface methodology. Bioresource Technol. 101, 54875493. Hothorn, T., Hornik, K., Zeileis, A., 2006. Unbiased recursive partitioning: a conditional inference framework. J. Comput. Graph. Stat. 15, 651674. Jeremiasse, A.W., Hamelers, H.V.M., Buisman, C.J.N., 2010. Microbial electrolysis cell with a microbial biocathode. Bioelectrochemistry 78, 3943. Khanal, S.K., Chen, W.H., Li, L., Sung, S., 2004. Biological hydrogen production: effect of pH and intermediate products. Int. J. Hydrogen Energ 29, 11231131. Karagiannidis, A., Perkoulidis, G., 2009. A multi-criteria ranking of different technologies for the anaerobic digestion for energy recovery of the organic fraction of municipal solid wastes. Bioresource Technol. 100, 23552360. Kim, S.H., Han, S.K., Shin, H.S., 2004. Feasibility of biohydrogen production by anaerobic co-digestion of food waste and sewage sludge. Int. J. Hydrogen Energy 29, 16071616. Kim, J.K., Nhat, L., Chun, Y.N., Kim, S.W., 2008a. Hydrogen production conditions from food waste by dark fermentation with Clostridium beijerinckii KCTC 1785. Biotechnol. Bioprocess E13, 499504. Kim, S.H., Han, S.K., Shin, H.S., Kim, 2008b. Optimization of continuous hydrogen fermentation of food waste as a function of solids retention time independent of hydraulic retention time. Process Biochem. 43, 213218. Kim, S.H., Shin, H.S., 2008. Effects of base-pretreatment on continuous enriched culture for hydrogen production from food waste. Int. J. Hydrogen Energy 33, 52665274. Kim, D.H., Kim, S.H., Shin, H.S., 2009. Hydrogen fermentation of food waste without inoculum addition. Enzyme Microb. Technol. 45, 181187. Kim, D.H., Kim, S.H., Kim, K.Y., Shin, H.S., 2010. Experience of a pilot-scale hydrogenproducing anaerobic sequencing batch reactor (ASBR) treating food waste. Int. J. Hydrogen Energy 35, 15901594. Kim, D.H., Kim, S.H., Kim, H.W., Kim, M.S., Shin, H.S., 2011a. Sewage sludge addition to food waste synergistically enhances hydrogen fermentation performance. Bioresour. Technol. 102, 85018506. Kim, D.H., Wu, J., Jeong, K.-W., Kim, M.S., Shin, H.S., 2011b. Natural inducement of hydrogen from food waste by temperature control. Int. J. Hydrogen Energy 36, 1066610673. Kim, D.H., Kim, S.H., Jung, K.W., Kim, M.S., Shin, H.S., 2011c. Effect of initial pH independent of operational pH on hydrogen fermentation of food waste. Bioresour. Technol. 102, 86468652. Kyazze, G., Popov, A., Dinsdale, R., Esteves, S., Hawkes, F., Premier, G., Guwy, A., 2010. Inuence of catholyte pH and temperature on hydrogen production from acetate using a two chamber concentric tubular microbial electrolysis cell. Int. J. Hydrogen Energy 35, 77167722. Lalaurette, E., Thammannagowda, S., Mohagheghi, A., Maness, P.-C., Logan, B.E., 2009. Hydrogen production from cellulose in a two-stage process combining fermentation and electrohydrogenesis. Int. J. Hydrogen Energy. 34, 62016210. Lay, J.J., Lee, Y.J., Noike, T., 1999. Feasibility of biological hydrogen production from organic fraction of municipal solid waste. Water Res. 33, 25792586. Lay, J.J., Fan, K.S., Chang, J., Ku, C.H., 2003. Inuence of chemical nature of organic wastes on their conversion to hydrogen by heat-shock digested sludge. Int. J. Hydrogen Energy 28, 13611367. Lay, J.J., Fan, K.S., Hwang, J.I., Chang, J.I., Hsu, P.C., 2005. Factors affecting hydrogen production from food wastes by Clostridium-rich composts. J. Environ. Eng. ASCE 131, 595602. Lee, Y.W., Chung, J., 2010. Bioproduction of hydrogen from food waste by pilot-scale combined hydrogen/methane fermentation. Int. J. Hydrogen Energy 35, 11746 11755. Lee, K.S., Lin, P.J., Chang, J.S., 2006. Temperature effects on biohydrogen production in a granular sludge bed induced by activated carbon carriers. Int. J. Hydrogen Energy 31, 465472. Lee, Z.K., Li, S.L., Lin, J.S., Wang, Y.H., Kuo, P.C., Cheng, S.S., 2008. Effect of pH in fermentation of vegetable kitchen wastes on hydrogen production under a thermophilic condition. Int. J. Hydrogen Energy 33, 52345241. Lee, Z.K., Li, S.L., Kuo, P.C., Chen, I.C., Tien, Y.M., Huang, Y.J., Chuang, C.P., Wong, S.C., Cheng, S.S., 2010a. Thermophilic bio-energy process study on hydrogen fermentation with vegetable kitchen waste. Int. J. Hydrogen Energy 35, 1345813466. Lee, D.Y., Ebie, Y., Xu, K.Q., Li, Y.Y., Inamori, Y., 2010b. Continuous H2 and CH4 production from high-solid food waste in the two-stage thermophilic fermentation process with the recirculation of digester sludge. Bioresource Technol. 101, S42S47. Li, C., Fang, H.H.P., 2007. Fermentative hydrogen production from wastewater and solid wastes by mixed cultures. Crit. Rev. Environ. Sci. Technol. 37, 139. Li, M., Youcai, Z., Qiang, G., Xiaoqing, Q., Dongjie, N., 2008a. Bio-hydrogen production from food waste and sewage sludge in the presence of aged refuse excavated from refuse landll. Renew. Energy 33, 25732579. Li, S.L., Kuo, S.C., Lin, J.S., Lee, Z.K., Wang, Y.H., Cheng, S.S., 2008b. Process performance evaluation of intermittentcontinuous stirred tank reactor for anaerobic hydrogen fermentation with kitchen waste. Int. J. Hydrogen Energy 33, 15221531. Lin, C.Y., Lay, C.H., 2004. Carbon/nitrogen ratio effect on fermentative hydrogen production by mixed microora. Int. J. Hydrogen Energ 29, 4145. Liu, D., Liu, D., Zeng, R.J., Angelidaki, I., 2006. Hydrogen and methane production from household solid waste in the two-stage fermentation process. Water Res. 40, 22302236. Liu, H., Grot, S., Logan, B.E., 2005. Electrochemically assisted microbial production of hydrogen from acetate. Environ. Sci. Technol. 39, 43174320.

G. De Gioannis et al. / Waste Management 33 (2013) 13451361 from organic waste: mesophilic versus thermophilic regime. Int. J. Hydrogen Energy 30, 13831391. Vzquez-Larios, A.L., Solorza-Feria, O., Vzquez-Huerta, G., Esparza-Garca, F., Rinderknecht-Seijas, N., Poggi-Varaldo, H.M., 2011. Effects of architectural changes and inoculum type on internal resistance of a microbial fuel cell designed for the treatment of leachates from the dark hydrogenogenic fermentation of organic solid wastes. Int. J. Hydrogen Energy 36, 6199 6209. Wang, A., Sun, D., Cao, G., Wang, H., Ren, N., Wu, W.-M., Logan, B.E., 2011. Integrated hydrogen production process from cellulose by combining dark fermentation, microbial fuel cells, and a microbial electrolysis cell. Bioresource Technol. 102, 41374143.

1361

Wang, X., Zhao, Y., 2009. A bench scale study of fermentative hydrogen and methane production from food waste in integrated two-stage process. Int. J. Hydrogen Energy 34, 245254. Wu, X., Yao, W., Zhu, J., 2010. Effect of pH on continuous biohydrogen production from liquid swine manure with glucose supplement using an anaerobic sequencing batch reactor. Int. J. Hydrogen Energy 35, 65926599. Zhu, H., Parker, W., Basnar, R., Proracki, A., Falletta, P., Beland, M., Seto, P., 2008. Biohydrogen production by anaerobic co-digestion of municipal food waste and sewage sludges. Int. J. Hydrogen Energy 33, 36513659. Zong, W., Yu, R., Zhang, P., Fan, M., Zhou, Z., 2009. Efcient hydrogen gas production from cassava and food waste by a two-step process of dark fermentation and photo-fermentation. Biomass Bioenergy 33, 14581463.