Bioresource Technology 130 (2013) 710–718

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Optimization of biological hydrogen production for anaerobicco-digestion of food waste and wastewater biosolids

0960-8524/$ - see front matter � 2012 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2012.12.069

Abbreviations: C/N, carbon to nitrogen ratio; DOPF, dufferin organics processingfacility; FA, free ammonia; FW, food waste; PS, primary sludge; SCOD, solublechemical oxygen demand; SSO, source separated organics; S0/X0, initial substrate-to-biomass ratio; TA, total ammonia; TCOD, total chemical oxygen demand; TN,total nitrogen; TP, total phosphorous; TSS, total suspended solids; TVFAs, totalvolatile fatty acids; VSS, volatile suspended solids; VS/TS, volatile solids to totalsolids ratio; WAS, waste activated sludge.⇑ Corresponding author. Tel.: +1 519 860 3556.

E-mail addresses: sayedbesh@yahoo.com, eelbeshb@uwo.ca, selbeshbishy@trojan.com (E. Elbeshbishy).

Peiqing Zhou a, Elsayed Elbeshbishy b,⇑, George Nakhla a

a Civil and Environmental Engineering Department, University of Western Ontario, London, ON, Canada N6A 5B9b Trojan Technologies, London, ON, Canada N5V 4T7

h i g h l i g h t s

" Anaerobic co-digestion of 21 mixtures of FW, PS, and WAS were evaluated." The maximum hydrogen yields of FW + PS and FW + WAS were achieved at ratios of 75:25." The maximum hydrogen yield FW + PS + WAS was achieved at ratio of 80:15:5." Optimum COD/N of FW + PS, FW + WAS, and FW + PS + WAS were 26, 31 and 30, respectively." A synergistic effect of co-digestion was observed and quantified.

a r t i c l e i n f o

Article history:Received 28 September 2012Received in revised form 6 December 2012Accepted 10 December 2012Available online 20 December 2012

Keywords:BiohydrogenCo-digestionFood-wasteSewage sludgeDark fermentation

a b s t r a c t

Batch anaerobic co-digestion studies were conducted using 21 mixtures (M1–M21) of food waste (FW),primary sludge (PS), and waste activated sludge (WAS) at 37 �C and an initial pH of 5.5 ± 0.2. The resultsshowed that co-digestion of FW and sludges had a positive impact on the hydrogen production. Themaximum hydrogen yields by co-digestion of FW + PS, FW + WAS, and FW + PS + WAS were achievedat volumetric ratios of 75:25, 75:25, and 80:15:5, respectively, with corresponding optimal COD/N massratios of 26, 31 and 30, respectively. Furthermore, the synergistic effect of co-digestion was proven andquantified: the measured hydrogen productions were higher than the sums of the hydrogen productionscalculated from each fraction, and the highest percentage increase above the calculated value of 101%,was achieved in the FW + PS + WAS mixture (80:15:5).

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Minimal or zero use of hydrocarbons, with only water as a com-bustion production and a high energy yield of 122 kJ/g (2.75 timesgreater than that of hydrocarbon fuel) render hydrogen as one ofthe promising sustainable energy resources (Han and Shin, 2004).Hydrogen production addresses three of today’s major energyproblems: soaring energy demand, environmental pollution, and

fossil fuel dependence (Momirlan and Veziroglu, 1999). Due tohigh electricity requirement by conventional physico-chemicalhydrogen production methods (such as water electrolysis, chemi-cal cracking of hydrocarbons, etc.) biological hydrogen productionhas recently attracted more attention (Hawkes et al., 2002).Photo-fermentation and dark fermentation are the two main typesof biological hydrogen production (Antonopoulou et al., 2010).Lower operational cost, greater hydrogen production rate, widerrange of organic substances and simplicity rationalize the superior-ity of dark fermentation over photo-fermentation (Xie et al., 2012;Hallenbeck and Benemann, 2002).

Since carbohydrates are the preferred substrates for dark fer-mentative hydrogen-producing bacteria such as Clostridium spe-cies, food waste (FW) with its high content of organic matter andcarbohydrates, and its easily hydrolysable nature has a high hydro-gen production potential (Kim et al., 2004). Moreover, FW, as animportant municipal and agricultural waste, can be an economicalsource for fermentative hydrogen production (Zhu et al., 2008). FW

P. Zhou et al. / Bioresource Technology 130 (2013) 710–718 711

includes uneaten food and food preparation leftovers from resi-dences, commercial establishments such as restaurants, institu-tions like school cafeterias, and industrial sources like factorylunch-rooms (Zhang et al., 2007). Generally, they consist mainlyof starch, protein, and fat, with a small amount of cellulose andhemi-cellulose which are possible sources for bioenergy produc-tion (Yuan et al., 2006). However, FWs can be highly variabledepending on their resources. Some studies on characteristics ofFWs indicated their variability and reported: moisture contentsof 74–90%, volatile solids to total solids ratios (VS/TS) of 80–97%,and carbon to nitrogen ratios (C/N) of 14.7–36.4 (Zhang et al.,2007).

However, FW may be lacking in nitrogen which is an essentialnutrient for hydrogen producers (Kim et al., 2004). Therefore, theconcept of co-digestion of FW and sewage sludges (primary sludgeand waste activated sludge) has been investigated to improve bio-hydrogen production, since the addition of sludges to FW supplieda more balanced carbon to nitrogen (C/N) ratio (Kim et al., 2004).The C/N ratio is one of the most significant parameters for co-digestion of FW and sludges process, which not only balances thenutrients, but also enhances the bacterial productivity of hydrogenand preclude ammonia inhibition (Lin and Lay, 2004). The mainproducts of the biodegradation of proteins in anaerobic conditionsare ammonia and various amino acid compounds, but digestioncan be inhibited by high free ammonia concentrations in the rangeof above 200 mg/L (Liu and Sung, 2002). Salerno et al. (2006)reported the ammonia inhibition of hydrogen production from glu-cose in batch tests at 30 �C. Even though only biohydrogen produc-tion rate was highly influenced by ammonia (defined as the sum ofNH3 and NHþ4 species) concentrations above 2 g N/L in batch tests,in continuous flow tests, both hydrogen production rates andyields were inhibited by high ammonia concentrations. When theammonia concentration was 2 g N/L, the maximum biohydrogenproduction were 56 mL/h (at pH of 6.2) and 49 mL/h (at pH of5.2), but when the ammonia concentration was 10 g N/L the max-imum biohydrogen production decreased to 16 mL/h (at pH of 6.2)and 7 mL/h (at pH of 5.2). Furthermore, a continuous flow reactoroperated at ammonia concentrations of 0.8–7.8 g N/L, achieved anoverall yield of 1.1–1.9 mol H2/mol glucose, with hydrogen pro-duction failing at ammonia liquid phase concentrations higherthan 1.6 g N/L. Furthermore, free ammonia (FA) has been suggestedas the main cause of inhibition since it is freely membrane-perme-able (De Baere et al., 1984). Thus, the C/N ratio should be strictlycontrolled effective hydrogen production. Previously, Elbesgbishyand Nakhla (2012) demonstrated that the proper C/N ratio foranaerobic methanogenic co-digestion of bovine serum albuminand starch was 12.8:1, and Chen et al., 2008 who reviewed the lit-erature reported that the general optimum value for the stable per-formance of anaerobic digestion of solids waste was 20:1.However, the desired C/N ratio for efficient hydrogen formationby co-digestion of rice straw and sewage sludge was 25:1 [Kimet al., 2012]. Sreela-or et al. (2011) observed that the optimumhydrogen production from the co-digestion of food waste andsludge in batch fermentation by anaerobic mixed cultures, oc-curred at a C/N ratio of 33:1. According to Kim et al. (2004) theoptimal C/N ratio in batch fermentation at 35 �C was 1.66 g carbo-hydrate-COD/g protein-COD achieved by the mixture of 87:13(food waste: primary and secondary sludges).

As apparent from the literature surveyed above, only a handfulstudies explored co-digestion of FW and wastewater sludges forbiohydrogen production, i.e. Kim et al. (2004), Zhu et al. (2008),Sreela-or et al. (2011), Tawfik and El-Qulish (2012), etc. The afore-mentioned studies not only did not define explicitly the optimumC/N ratio, but also used local food waste from cafeterias (Zhu et al.,2008; Sreela-or et al. (2011)), restaurants (Tawfik and El-Qulish,2012), dining halls (Kim et al., 2004; Li et al., 2008), that are not

representative of overall food waste received at source separatedorganics (SSO) processing facilities. As alluded to above, FW is avery heterogeneous waste with widely different characteristics,and thus from a practical perspective in order to design centralizedco-digestion processes, the organic fraction of municipal solidwastes collected from the various sources is definitely more repre-sentative than individual food waste sources. Furthermore, the lit-erature studies on biohydrogen from co-digestion have notfocussed on delineating the synergistic effects of co-digestiondue to scope limitations and merely focussed on hydrogen produc-tion per unit organic matter (COD or VS) irrespective of the sourceof the organic matter in the mixture. Thus, the primary objectivesof this study were to investigate biohydrogen production from FW,primary sludge (PS) and/or waste activated sludge (WAS) in a widerange of mixtures, delineate the optimum COD/N ratios, and quan-tify the synergistic effect of co-digestion using representative foodwaste samples.

2. Methods

2.1. Substrates and seed

Three substrates were used in this study: food waste (FW) i.e.organic fraction of municipal solid wastes, primary sludge (PS)and waste activated sludge (WAS). The food waste was obtainedfrom the Dufferin Organics Processing Facility (DOPF) in Toronto,Ontario, Canada. The city of Toronto’s DOPF receives approximately25,000 metric tons/year of source separated organics (SSO) mate-rial from Toronto’s Green Bin and the commercial Yellow Bag col-lection programs. After separation of plastic and inert, the DOPF isto separate the film plastic bin finer and organic material is processby anaerobic digestion (Van Opstal, 2006). The PS and WAS sam-ples for this study were collected from the Adelaide Pollution Con-trol Plant located in London, Ontario, Canada.

Anaerobic digested sludge was collected from the primaryanaerobic digester at St. Mary’s wastewater treatment plant (St.Mary’s, Ontario) and used as seed sludge for all batches used in thisstudy. The characteristics of the three substrates and seed are pre-sented in Table 1.

2.2. Batch anaerobic digestion

Batch anaerobic studies were conducted using FW, PS and/orWAS as substrates. Twenty-one different substrate mixtures ofFW, PS and/or WAS, M1–M21, were used. Table 2 shows the sub-strate compositions of the 21 mixtures by volume. The specific vol-umes were derived from the following procedures. Based onprevious experience (Nasr et al., 2011), an initial substrate-to-bio-mass (S0/X0) ratio of 2 g COD/g VSS was selected for all batchesused in this biohydrogen experiment, with a total liquid of sub-strate and seed of 200 mL,

S0=X0 ¼ g TCODsubstrate

g VSSseed¼ V sub � TCODsubstrate

V seed � VSSseed¼ 2 ð1Þ

Subsequently, the volumes of FW, PS and WAS were calculatedbased on the ratios of substrates, noting that the TCOD of the mix-ture used in Eq. (1) was calculated based on the TCOD of FW, PSand WAS and the mixing ratios. In addition, seed sludge requiredpre-treatment to inhibit hydrogen-consuming bacteria underanaerobic condition. The purpose of seed pre-treatment was tosuppress hydrogen-consuming bacterial activity as much as possi-ble, while preserving the activity of the hydrogen-producing bacte-ria at the same time to harness hydrogen production (Cai et al.,2004). Baghchehsaraee et al. indicated that the amount of hydro-gen produced by heat-pretreated inocula was 7 times more than

Table 1Raw waste characteristics.

Parameter Units Food waste Primary sludge WAS Seed

pH – 6 ± 0.2a 6 ± 0.3 7 ± 0.3 7 ± 0.2Alkalinity mg/L CaCO3 8000 ± 500 6200 ± 500 370 ± 30 5900 ± 200TSS mg/L 66900 ± 2200 33500 ± 1300 21600 ± 2300 16400 ± 2300VSS mg/L 47000 ± 1100 26400 ± 1000 15800 ± 1200 11600 ± 1400TCOD mg/L 106700 ± 2800 33900 ± 230 26500 ± 2200 18400 ± 1300SCOD mg/L 39300 ± 2300 3300 ± 100 4900 ± 400 1500 ± 220NH4–N mg/L 1800 ± 300 360 ± 20 290 ± 50 800 ± 90Total carbohydrate mg/L 36500 ± 1200 3200 ± 200 2000 ± 200 1800 ± 140Soluble carbohydrate mg/L 5500 ± 400 890 ± 40 650 ± 50 700 ± 70Total nitrogen mg/L 3400 ± 400 2700 ± 180 1300 ± 100 1600 ± 100Total phosphorous mg/L 2100 ± 300 450 ± 80 260 ± 50 290 ± 60Acetic acid mg COD/L 4300 ± 100 660 ± 50 620 ± 70 370 ± 40Propionic acid mg COD/L 1000 ± 100 350 ± 40 230 ± 40 180 ± 20Isobutyric acid mg COD/L 0 ± 2 100 ± 10 150 ± 20 32 ± 10Butyric acid mg COD/L 580 ± 70 400 ± 40 270 ± 30 260 ± 30Isovaleric acid mg COD/L 350 ± 50 150 ± 30 100 ± 10 39 ± 10Valeric acid mg COD/L 200 ± 30 70 ± 10 95 ± 10 48 ± 10TVFAs mg COD/L 6500 ± 500 1700 ± 220 1500 ± 120 1000 ± 120

a Values represents the average ± STD of two samples.

Table 2Substrate compositions of different co-digestion combinations.

Mixture Seed (mL) Total Substrates (mL) FW (mL) PS (mL) WAS (mL)

FW + PS M1 (100:0)a 162 38 38 0 0M2 (90:10) 161 39 35 4 0M3 (75:25) 158 42 32 11 0M4 (50:50) 153 47 23 23 0M5 (25:75) 147 53 13 40 0M6 (0:100) 139 61 0 61 0

FW + WAS M7 (90:10) 160 40 36 0 4M8 (75:25) 156 44 33 0 11M9 (50:50) 148 52 26 0 26M10 (25:75) 136 64 16 0 48M11 (0:100) 116 84 0 0 84

FW + PS + WAS M12 (90:5:5) 162 38 35 2 2M13 (80:15:5) 159 41 33 2 6M14 (80:10:10) 159 41 33 4 4M15 (80:5:15) 158 42 33 6 2M16 (70:20:10) 155 45 31 4 9M17 (70:15:15) 155 45 31 7 7M18 (70:10:20) 155 45 32 9 5M19 (60:30:10) 152 48 29 5 14M20 (60:20:20) 151 49 29 10 10M21 (60:10:30) 150 50 30 15 5

a Ratios in parenthesis represent FW:PS for M1–M6, FW:WAS for M7–M11, and FW:PS:WAS for M12–M21.

712 P. Zhou et al. / Bioresource Technology 130 (2013) 710–718

that produced by the untreated inocula (Baghchehsaraee et al.,2008). In this research, the seed was heat-treated at 70 �C for30 min. Batch experiments were conducted in duplicates in a seriesof 250 mL serum bottles. To each bottle, substrate and seed wereadded according to the required amounts, and the initial pH forthe mixture was adjusted to 5.5 ± 0.2 using 1 N NaOH and HCl,which is the optimum pH for hydrogen production through darkfermentation (Kim et al., 2012, 2004). Moreover, 20 mL of the mix-ture was reserved for each initial sample characterization afterthrough mixing i.e. liquid volume of 180 mL and headspace volumeof 70 mL were maintained in the serum bottles. The headspace wasflushed with oxygen-free nitrogen gas for a period of 3 min andcapped tightly with rubber stoppers. The bottles were then placedin a swirling-action shaker (MaxQ 4000, Thermo Scientific, CA)operating at 180 rpm and maintained at a temperature of 37 �C.Blank bottles were prepared using the sludge with addition ofdeionized water and nutrient stock solution only (no substrateadded). The background hydrogen production from the seed(blank), determined in blank assays with medium and no sub-

strate, was subtracted from the hydrogen production obtained inthe substrate assays (Angelidaki et al., 2009). Forty-two bottleswere used for 21 mixtures in addition to the two for the blank.

2.3. Analytical methods

Liquid samples were analyzed for total suspended solid (TSS),volatile suspended solid (VSS), and alkalinity using standard meth-ods (APHA, 1998). Total and soluble chemical oxygen demand(TCOD, SCOD), total nitrogen (TN), total phosphorus (TP) and totalammonia (NH4–N) were measured using HACH test kits (HACHOdyssey DR/2500). Carbohydrate was determined by the colori-metric method (Dubois et al., 1956) UV wavelength of 490 nmusing glucose as a standard. Soluble parameters were determinedafter filtering the samples through 0.45 lm filter paper. The totalgas volume was measured by releasing the gas pressure in the vialsusing appropriately sized glass syringes (Perfektum, Popper & SonsInc., NY, USA) in the 5–100 mL range to equilibrate with theambient pressure as recommended by Owen et al. (1979). Gas

P. Zhou et al. / Bioresource Technology 130 (2013) 710–718 713

composition including hydrogen, was determined by a gas chro-matograph (Model 310, SRI Instruments, Torrance, CA) equippedwith a thermal conductivity detector (TCD) and a molecular sievecolumn (Molesieve 5A, mesh 80/100, 6 ft � 1/8 in). The tempera-tures of the column and the TCD detector were 90 and 105 �C,respectively. Argon was used as the carrier gas at a flow rate of30 mL/min. The concentrations of VFAs were analyzed after filter-ing the sample through 0.45 lm using a gas chromatograph(Varian 8500, Varian Inc., Toronto, Canada) with a flame ionizationdetector (FID) equipped with a fused silica column (30 m �0.32 mm). Helium was used as the carrier gas at a flow rate of5 mL/min. The temperatures of the column and detector were110 and 250 �C, respectively.

3. Calculations

3.1. Gompertz model

The following modified Gompertz model has been used to de-scribe the progression of cumulative hydrogen production in thebatch tests (Lay et al., 1999):

H ¼ p � exp � expRe

m

pðk� tÞ þ 1

� �� �ð2Þ

where H is the cumulative hydrogen production (mL), p is themaximum hydrogen production (mL), Rm is the initial hydrogen

Fig. 1. Cumulative hydrogen production from different co-digestion

production rate (mL/h), k is the lag phase time (h), t is the incuba-tion time (h), and e = exp(1) = 2.718.

3.2. Free ammonia

Free ammonia (FA) concentrations were calculated from the to-tal ammonia (TA) concentration in the liquid and the fraction of FA(fN), using the following equation (Omil et al., 1995):

fN ¼FATA¼ 1= 1þ kb � 10�pH

kw

" #ð3Þ

where kb and kw are the dissociation constants for ammonia andwater, respectively (1.855 � 10�5 and 2.355 � 10�14 mol/L at37 �C), and pH represented the final pH of the samples.

4. Results and discussion

4.1. Hydrogen production

Cumulative hydrogen production from FW + PS, FW + WAS andFW + PS + WAS are presented in Fig. 1a–c, respectively; error barsare not shown as error was <10%. As depicted in Fig. 1, most sub-strate mixtures (except for M5, M6 and M11), initially experienceda lag phase, followed by a rapid hydrogen production phase, andminimal hydrogen production at the end. Based on the Gompertz

combinations; (a) FW + PS, (b) FW + WAS, (c) FW + PS + WAS.

Table 3Kinetic analysis of hydrogen production from different co-digestion combinations.

Mixtures Gompertz kinetics

P* Rm k R2

mL mL/h h

FW + PS M1 (100:0) 144 7 6 0.999M2 (90:10) 182 11 8 0.999M3 (75:25) 208 11 7 0.999M4 (50:50) 108 4 7 0.985M5 (25:75) 33 1 2 0.964M6 (0:100) 20 0.4 1 0.997

FW + WAS M7 (90:10) 222 13.5 9.1 0.999M8 (75:25) 214 11.8 8.0 0.999M9 (50:50) 132 9 7 0.999M10 (25:75) 44 3.7 5 0.999M11 (0:100) 14 0.7 1 0.952

FW + PS + WAS M12 (90:5:5) 208 10.3 8.9 0.999M13 (80:15:5) 255 12.7 8.8 0.999M14 (80:10:10) 241 12 8.5 0.999M15 (80:5:15) 222 10.9 8.7 0.998M16 (70:20:10) 218 11.1 9.4 0.997M17 (70:15:15) 213 10.8 9.6 0.999M18 (70:10:20) 180 8.7 9.2 0.996M19 (60:30:10) 162 7.4 8.2 0.986M20 (60:20:20) 154 7.1 8.2 0.992M21 (60:10:30) 148 6.4 8.2 0.994

* P: cumulative hydrogen production.

714 P. Zhou et al. / Bioresource Technology 130 (2013) 710–718

model, the results of which are shown in Table 3, the lag phasesranged from 1 to 9.6 h., with the majority around 8 h. The lagphases of FW and sludges co-digestion were much longer thanthe lag phases of sludges only (M6 and M11), and that might be be-cause the seed used in this study was obtained from digester treat-ing sewage sludge. The exponential phases of hydrogen productionfrom FW + PS and FW + WAS + PS lasted for about 27 h, while theexponential phases of hydrogen production from FW + WAS lasted

Table 4Hydrogen yields based on different units.

Mixtures H2 yields

L/Lsuba L/Lfood waste

b mL/g CODsubc mL/g Carb-subd mL/g VSSsub

e

FW + PS M1 (100:0) 4.2 ± 0.32 4.2 ± 0.32 39 ± 3.3 115 ± 8 89 ± 6M2 (90:10) 5.2 ± 0.41 5.8 ± 0.46 54 ± 4.2 157 ± 12 116 ± 8M3 (75:25) 5.4 ± 0.32 7.3 ± 0.43 66 ± 4.6 193 ± 16 130 ± 10M4 (50:50) 2.5 ± 0.18 5 ± 0.36 43 ± 3.2 127 ± 11 69 ± 4M5 (25:75) 0.7 ± 0.1 2.7 ± 0.4 19 ± 2.1 59 ± 5 21 ± 1.2M6 (0:100) 0.3 ± 0.1 N/A 10 ± 1.2 109 ± 9 13 ± 0.8

FW + WAS M7 (90:10) 6.2 ± 0.36 6.9 ± 0.54 63 ± 4.5 187 ± 14 141 ± 11M8 (75:25) 5.4 ± 0.41 7.2 ± 0.42 62 ± 3.6 301 ± 19 137 ± 12M9 (50:50) 2.8 ± 0.22 5.6 ± 0.50 42 ± 3.8 146 ± 12 90 ± 7M10 (25:75) 0.8 ± 0.15 3.1 ± 0.23 16 ± 1.1 72 ± 6 32 ± 2.1M11 (0:100) 0.2 ± 0.05 N/A 7 ± 0.6 93 ± 7 12 ± 0.7

FW + PS + WAS M12 (90:5:5) 6 ± 0.42 6.7 ± 0.46 61 ± 5.1 181 ± 13 135 ± 11M13 (80:15:5) 6.8 ± 0.43 8.5 ± 0.52 76 ± 4.3 229 ± 16 165 ± 13M14 (80:10:10) 6.4 ± 0.36 8 ± 0.61 72 ± 6.1 216 ± 14 153 ± 12M15 (80:5:15) 5.9 ± 0.41 7.3 ± 0.54 66 ± 5.2 197 ± 17 138 ± 11M16 (70:20:10) 5.4 ± 0.32 7.7 ± 43 67 ± 4.8 206 ± 15 140 ± 12M17 (70:15:15) 5.2 ± 0.25 7.5 ± 0.36 65 ± 3.9 199 ± 14 134 ± 10M18 (70:10:20) 4.4 ± 0.36 6.2 ± 0.45 55 ± 4.4 165 ± 10 110 ± 8M19 (60:30:10) 3.7 ± 0.24 6.2 ± 0.36 51 ± 3.7 162 ± 12 104 ± 9M20 (60:20:20) 3.5 ± 0.31 5.8 ± 0.44 49 ± 4.1 152 ± 11 95 ± 7M21 (60:10:30) 3.3 ± 0.27 5.4 ± 0.38 46 ± 3.5 141 ± 9 7

a L/Lsub: Hydrogen production per unit volume of substrate added.b L/Lfood waste: Hydrogen production per unit volume of FW added.c mL/g CODsub: Hydrogen production per unit mass of COD of substrate added.d mL/g Carbsub: Hydrogen production per unit mass of total carbohydrate of substrate added.e mL/g VSSsub: Hydrogen production per unit mass of VSS of substrate added.

for about 17 h. Methane was not detected in any of the experi-ments and that was due to the heat pretreatment of the seed aswell as the low pH of 5.5.

As presented in Fig. 1a, the highest cumulative hydrogen pro-duction of 208 mL from FW + PS was achieved by M3 (75%FW + 25% PS), which was 45% higher than M1 (FW only); and thelowest cumulative hydrogen production of 20 mL was observedin M6 (PS only), which was 86% lower than M1 (FW only). As pre-sented in Fig. 1b, the highest cumulative hydrogen production of222 mL from FW + WAS was achieved by M7 (90% FW + 10%WAS), which was 55% higher than M1 (FW only). The lowest cumu-lative hydrogen production of 14 mL was achieved by M11 (WASonly), which was 90% lower than M1 (FW only). As presented inFig. 1c, the highest cumulative hydrogen production fromFW + PS + WAS of 255 mL was achieved by M13 (80% FW + 15%WAS + 5% PS), which was 76% higher than M1 (FW only). The low-est cumulative hydrogen production of 148 mL was achieved byM21 (60% FW + 10% WAS + 30% PS), which was only 2% higher thanM1 (FW only).

Hydrogen production rates calculated using Gompertz kineticsare shown in Table 3. The mixtures with high hydrogen productionare the same as the ones with high hydrogen production rate: thehighest cumulative hydrogen production rate of 11 mL/h fromFW + PS was achieved by M2 (90% FW + 10% PS) and M3 (75%FW + 25% PS), 13.5 mL/h from FW + WAS was achieved by M7(90% FW + 10% WAS), and 12.7 mL/h from FW + PS + WAS wasachieved by M13 (80% FW + 15% PS + 5% WAS). The mixtures withlow concentration of food waste produced minimal hydrogen.

It should be stressed here that co-digestion of FW and sludgesresulted in higher hydrogen production compared to FW only oreach sludge alone. In addition, it is evident that the hydrogen pro-duction and maximum hydrogen production rate were both ob-served in M3, M7 and M13, which were much higher than M1(FW only) (Table 3). Thus, it is obvious that, generally, notwith-standing the occurrence of peaks discussed later, co-digestion ofFW and sludges had a positive impact on the hydrogen production.

P. Zhou et al. / Bioresource Technology 130 (2013) 710–718 715

On the other hand, the negative effect on hydrogen productionwhich emerged as the FW contents decreased compared to FWonly, might result from the reduction of carbohydrate, the primarysubstrate for biohydrogen production from FW.

4.2. Hydrogen yields

Table 4 shows the hydrogen yields using different units. Thehydrogen yield was normalized per unit volume of substrate added(L/Lsub), per unit volume of FW added (L/Lfood waste), per substratemass COD added (mL/g CODsub), per unit mass of total carbohy-drate of substrate added (mL/g Carbsub), per unit mass of VSS ofsubstrate added (mL/g VSSsub). As depicted in Table 4, the hydro-gen yields (in different units) increased initially with increasingthe PS and/or WAS percentage to the peak values and then declinedto the minimum values at the highest percentages of PS and/orWAS of each set of experiments. For mixtures of FW + PS(M1–M6), the highest hydrogen yield of 66 mL H2/g CODsubadded

which corresponds to 130 ml/g VSSsub and 7.3 L/Lfood waste wasachieved by M3. The hydrogen yield of FW only (M1) was 39 mL/gCODsub, which was 69% lower than that of M3. For mixtures ofFW + WAS (M7–M11), the highest hydrogen yield of 63 mL H2/gCODsub corresponding to 141 ml/gVSSsub and 6.2 L/Lfood waste wasachieved by M7. The hydrogen yield of FW only (M1) was62% lower than that of M7. For mixtures of FW + PS + WAS(M12–M21), the highest hydrogen yield of 76 mL H2/g CODsub cor-responding to 165 mL/g VSSsub and 6.8 L/Lfood waste was achieved byM13. The hydrogen yield of FW only (M1) was 95% lower than thatof M13. Furthermore, the maximum hydrogen yields in this studywere 193 mL/g Carbsub (M3) for FW + PS, 301 mL/g Carbsub (M8) for

Fig. 2. H2 yields based on mL/g CODSub at corresponding COD/N from different

FW + WAS and 229 mL/g Carbsub (M13) for FW + PS + WAS, whichwere much higher than 131.5 mL/g carbohydrate for the 87:13(by volume) mixture of FW and sewage sludge (primary sludgeand secondary sludge) of Kim et al. (2004). In addition, themaximum hydrogen yields obtained by M3 (130 mL/g VSSsub),M7 (141 mL/g VSSsub), and M13 (165 mL/g VSSsub) were higherthan the 112 mL/g volatile solid achieved by FW and sewage sludge(PS and WAS) at a 1:1 ratio (Zhu et al., 2008). There are two reasonsfor this: the aforementioned experiments were tested at only sixand three different mix ratios, respectively, which were not accu-rate enough to determine the optimum mixing ratio and maximumhydrogen yield; the second reason is the nature and characteristicsof organic wastes used in this study were different from those ofused by Kim et al. (2004) and Zhu et al. (2008).

As depicted in Fig. 2, the maximum hydrogen yields were ob-served at COD/N ratios of 26, 31 and 30 for FW + PS, FW + WAS,and FW + PS + WAS, respectively. Thus it is clear that addingwastewater sludges to FW had a positive effect only when the per-centage of municipal sludges in substrate was <50% (whose hydro-gen yield was almost equal to FW only), and a negative effectwould be observed and reflected by a less hydrogen yieldcompared to that of the FW only (M1) if the above threshold wasexceeded. The hydrogen yields presented in Table 4 emphasizedthe positive effect of adding sludges to FW and the significanceof COD/N ratio optimization (Fig. 2). The optimal COD/N ratiosfor the three combinations of substrates in this study ranged from26 to 31. According to the research of Kim et al. (2012), the desiredC/N ratio for efficient hydrogen formation by co-digestion of ricestraw and sewage sludge was 25 at 55 �C in batch. Sreela-oret al. (2011) observed that the optimum hydrogen production from

co-digestion combinations; (a) FW + PS, (b) FW + WAS, (c) FW + PS + WAS.

Fig. 3. Increase percentage of H2 yields based on mL/g CODSub at corresponding COD/N from different co-digestion combinations.

716 P. Zhou et al. / Bioresource Technology 130 (2013) 710–718

the co-digestion of food waste and sludge in batch fermentation byanaerobic mixed cultures, occurred at a C/N ratio of 33. Hydrogenyields based on other units were calculated and shown in Table 4below. From Table 4, it is noticed that almost all hydrogen yieldsof FW + WAS were higher than those of FW + PS, which proved thatWAS exerted a more synergetic effect on FW fermentation than PS.

Hydrogen yields based on L/Lsub of M1 (FW only), M6 (PS only)and M11 (WAS only) were 4.2 L/Lsub, 0.3 L/Lsub, 0.2 L/Lsub,respectively. Therefore, the theoretical hydrogen production ofeach different co-digestion mixture could be calculated, as thespecific volumes of each substrate added were known (shown inEq. (4)).

VH2 theoretical ¼ 4:2� VFW added þ 0:3� VPS added þ VWAS added ð4Þ

In this way, the theoretical hydrogen yields based on mL/gCODsub of each sample could be figured out according to Eq. (5).

YH2 treoretical ¼VH2theoretical

V total sub � TCODinitial subð5Þ

where YH2 treoretical is the theoretical hydrogen yield based on mL/gCODsub, VH2 theoretical is the volume of hydrogen produced theoreti-cally, V total sub is the total volume of substrates mixture,TCODinitial sub is the initial concentration of the mixture. An obviousconclusion could be drawn for the majority of the mixtures; thehydrogen yields calculated from actual measurements(YH2 measured) were higher than the theoretical hydrogen yield com-puted based on Eq. (5) (YH2 treoretical). For this reason, the percentageincreases in hydrogen yields above theoretical according to Eq. (6),are shown in Fig. 3.

% Increase H2yield ¼ðYH2measured � YH2treoreticalÞ � 100%

YH2treoreticalð6Þ

All mixtures reflected positive percentages increases except forM5 (�49%) and M10 (�38%). Thus, it is obvious that co-digestion ofFW and sludges exerted a positive impact on the hydrogen produc-tion as the measured hydrogen was generally higher than the sumsof the hydrogen production from each fraction. The highest per-centage increase was achieved by M13 (101%). This noticeableimprovement in hydrogen production is due to the synergistic ef-fect of the co-digestion of FW and sludges, which overcomes theimbalance in nutrients and improves biodegradation.

4.3. Characterization of initial and final mixtures

The characteristics of the initial and final samples from the var-ious co-digestion mixtures were also analyzed in this study. TSSdecreased 14–27%, and VSS decreased 15–26%. In addition, totalcarbohydrate decreased 21–76%, and soluble carbohydrate de-creased 15–74%, while TCOD decreased 6–12% only. The hydrogenyields based on total carbohydrate consumed were in the range of55–203 mL/g carbconsumed. It is interesting to note here that moreSCOD was measured in the final samples than in the initial sam-ples, indicating that a portion of organic matter was acidified anddissolved in solution because of VFAs production and the resultingdecrease of pH. As expected, concentrations of both total nitrogenand total phosphorous almost remained constant. The free ammo-nia concentrations ranged from 0.01 to 0.12 mg/L, well below thereported 25–140 mg/L for inhibition of mesophilic treatment (Omilet al., 1995).

According to the TVFAs analysis, the main acids in the final sam-ples were acetic acid, and butyric acid (49–82%). Eqs. (7, 8, and 9)represent three typical acidogenic reactions where glucose is con-verted to acetic acid, butyric acid and propionate acid, respectively.However, hydrogen can be produced through acetic acid and buty-ric acid pathways mainly, with the contribution of propionic aciddifficult to determine, due to the complex reaction processes inaccordance with Eqs. (9, 10, and 11) (Bilitewski et al., 1997). Eq.(11) represents the conversion of propionate to acetate, onlyachievable at low hydrogen pressure (Bilitewski et al., 1997). Stoi-chiometric yields of 4 and 2 mol/mol glucose (from Eqs. (7 and 8))were used, and according to the measured concentrations of ace-tate and butyrate, the contributions of the two pathways wereestimated.

C6H12O6 þ 2H2O! 2CH3COOHþ 2CO2 þ 4H2 ð7Þ

C6H12O6 ! 2CH3CH2CH2COOHþ 2CO2 þ 2H2 ð8Þ

C6H12O6 þ 2H2O$ 2CH3CH2COOHþ 2H2O ð9Þ

CH3CH2COOHþ 2H2O$ CH3COOHþ 3H2 þ CO2 ð10Þ

CH3CH2COO� þ 3H2O$ CH3COO� þHþ þHCO�3 þ 3H2 ð11Þ

Fig. 4 showing the HAc/HBu for the various mixtures indicatesthat generally hydrogen production peaked at the maximumHAc/HBu ratio. The mixture with the highest HAc/ HBu ratio andmaximum hydrogen production from FW + PS was M3 (75%

Fig. 4. HAc/HBc mole ratios of different combined co-digestion final samples; (a) FW + PS, (b) FW + WAS, (c) FW + PS + WAS.

Fig. 5. Percentage difference of measured from theoretical H2 production in acetic acid and butyric acid pathways of different combined co-digestion samples.

P. Zhou et al. / Bioresource Technology 130 (2013) 710–718 717

FW + 25% PS) consistently. However, it should be emphasized thatfrom FW + WAS, the mixture with the highest HAc/ HBu ratio wasM8 (75% FW + 25% WAS), while the mixture with the maximum

hydrogen production was M7 (90% FW + 10% WAS). In addition,from FW + PS + WAS, the mixture with highest HAc/HBu ratiowas M12 (90% FW + 5% PS + 5% WAS), while the mixture with max-

718 P. Zhou et al. / Bioresource Technology 130 (2013) 710–718

imum hydrogen production was M13 (80% FW + 15% PS + 5% WAS).The discordance of the peak HAc/HBu ratio and maximum hydro-gen production might be due to the concentration of the complexpropionic acid pathway.

The total volumes of hydrogen theoretically produced from ace-tate and butyrate pathway were figured out from acetic acid andbutyric acid concentrations. In order to compare the theoreticalhydrogen productions and measured hydrogen productions, theirpercentage differences were calculated according to Eq. (12) andthe results are shown in Fig. 5.

% Difference H2production ¼ðVH2treoretical � VH2measuredÞ � 100%

VH2treoreticalð12Þ

Fig. 5 demonstrates that the measured hydrogen production formost mixtures was less than the sum of theoretical hydrogen pro-ductions calculated from both the acetic acid and butyric acidpathways. The relatively high percentage difference was observedin M11 (83%) only, which is attributed to its relatively low hydro-gen production (14 mL). It is plausible that the produced hydrogenremained in solution could be one of the reasons. According to thesolubility of hydrogen in water of 0.0013 g H2/L (1.01 MPa, 37 �C),and the solution volume of 180 mL, the mass of hydrogen dissolvedin water was 0.00024 g, while the hydrogen measured and thecalculated theoretically were 0.001 and 0.002 g, respectively. Thus,the dissolved hydrogen gas accounted for about 24% of thedifferences between theoretical and measured.

5. Conclusions

This study investigated the optimal ratios of substrates for co-digestion of FW, PS and WAS in a wide range. The results of thisstudy showed the maximum hydrogen yields by co-digestion ofFW + PS, FW + WAS, and FW + PS + WAS were achieved at volumet-ric ratios of 75:25, 75:25, and 80:15:5, respectively, and their cor-responding optimal COD/N ratios were 26, 31 and 30, respectively.Furthermore, the synergistic effect of co-digestion was proven andquantified, with hydrogen production much higher than thetheoretical based on each waste fraction. The highest percentage’sincrease of 101% above theoretical was achieved in theFW + PS + WAS mixture (80:15:5).

References

Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J.L., Guwy, A.J., 2009.Defining the biomethane potential (BMP) of solid organic wastes and energycrops: a proposed protocol for batch assays. Water Sci. Technol. 59 (5), 927–934.

Antonopoulou, G., Gavala, H.N., Skiadas, I.V., Lyberatos, G., 2010. Influence of pH onfermentative hydrogen production from sweet sorghum extract. Int. J.Hydrogen Energy 35, 1921–1928.

APHA, 1998. Standard Methods for the Examination of Water and Wastewater, 20ed. American Public Health Association, Washington DC, USA.

Baghchehsaraee, B., Nakhla, G., Karamanev, D., Margaritis, A., Reid, G., 2008. Theeffect of heat pretreatment temperature on fermentative hydrogen productionusing mixed cultures. Int. J. Hydrogen Energy 33, 4064–4073.

Bilitewski, B., Härdtle, G., Marek, K., 1997. Waste Management. Springer, Berlin,ISBN 3-540-59210-5.

Cai, M.L., Liu, J.X., Wei, Y.S., 2004. Enhanced biohydrogen production from sewagesludge with alkaline pretreatment. Environ. Sci. Technol. 38 (11), 3195–3202.

Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: areview. Bioresour. Technol. 99, 4044–4064.

De Baere, L.A., Devocht, M., Van Assche, P., Verstraete, W., 1984. Influence of highNaCl and NH4Cl salt levels on methanogenic associations. Water Res. 18, 543–548.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956. Colorimetricmethod for determination of sugars and related substances. Anal. Chem. 28,350–356.

Elbesgbishy, E., Nakhla, G., 2012. Batch anaerobic co-digestion of proteins andcarbohydrates. Bioresour. Technol. 116, 170–178.

Hallenbeck, P.C., Benemann, J.R., 2002. Biological hydrogen production:fundamentals and limiting processes. Int. J. Hydrogen Energy. 27, 1185–1193.

Han, S.K., Shin, H.S., 2004. Biohydrogen production by anaerobic fermentation offood waste. Int. J. Hydrogen Energy 29, 569–577.

Hawkes, F.R., Dinsdale, R., Hawkes, D.L., Hussy, I., 2002. Sustainable fermentativehydrogen production: challenges for process optimization. Int. J. HydrogenEnergy 27, 1339–1347.

Kim, S.H., Han, S.N., Shin, H.S., 2004. Feasibility of biohydrogen production byanaerobic co-digestion of food waste and sewage sludge. Int. J. HydrogenEnergy 29, 1607–1616.

Kim, M., Yang, Y., Morikawa-sakura, M.S., Wang, Q., Lee, M.V., Lee, D., Feng, C., Zhou,Y., Zhang, Z., 2012. Hydrogen production by anaerobic co-digestion of rice strawand sweage sludge. Int. J. Hydrogen Energy 37, 3142–3149.

Lay, J.J., Lee, Y.J., Noike, T., 1999. Feasibility of biological hydrogen production fromorganic fraction of municipal solid waste. Water Res. 33, 2579–2586.

Li, M., Zhao, Y., Guo, Q., Qian, X., Niu, D., 2008. Bio-hydrogen production from foodwaste and sewage sludge in the presence of aged refuse excavated from refuselandfill. Renew. Energy 33, 2573–2579.

Lin, C.Y., Lay, C.H., 2004. Carbon/nitrogen-ratio effect on fermentative hydrogenproduction by mixed microfora. Int. J. Hydrogen Energy 29, 41–45.

Liu, T., Sung, S., 2002. Ammonia inhibition on thermophilic aceticlasticmethanogens. Water Sci. Technol. 45, 113–120.

Momirlan, M., Veziroglu, T., 1999. Recent directions of world hydrogen production.Renew Sustain. Energy Rev. 3, 219–231.

Nasr, N., Elbeshbishy, E., Hafez, H., Nakhla, G., Hesham, E., Naggar, M., 2011. Bio-hydrogen production from thin stillage using conventional and acclimatizedanaerobic digester sludge. Int. J. Hydrogen Energy 36, 12761–12769.

Omil, F., Mendez, R., Lema, J.M., 1995. Anaerobic treatment of saline wastewatersunder high sulphide and ammonia content. Bioresour. Technol. 54, 269–278.

Owen, W.F., Stuckey, D.C., Healy, J.B., Young, L.Y., McCarty, P.L., 1979. Bioassay formonitoring biochemical methane potential and anaerobic toxicity. Water Res.13, 485–492.

Salerno, M.B., Park, W., Zuo, Y., Logan, B.E., 2006. Inhibition of biohydrogenproduction by ammonia. Water Res. 40, 1167–1172.

Sreela-or, C., Plangklang, P., Imai, T., Reungsang, A., 2011. Co-digestion of food wasteand sludge for hydrogen production by anaerobic mixed cultures: statistical keyfactors optimization. Int. J. Hydrogen Energy XXX, I–II.

Tawfik, A., El-Qulish, M., 2012. Continuous hydrogen production from co-digestionof municipal food waste and kitchen wastewater in mesophilic anaerobicbaffled reactor. Bioresour. Technol. 114, 270–274.

Van Opstal, B., 2006. Evaluating AD system performance for MSW organics. BioCycle47 (11), 35–39.

Xie, G., Feng, L., Ren, N., Ding, J., Liu, C., Xing, D., 2012. Control strategies forhydrogen production through co-culture of ethanoligenens harbinense B49 andimmobilized Rhodopseudomonas faecalis RLD-53. Int. J. Hydrogen Energy 35,1929–1935.

Yuan, Y.Y., Cao, X.Y., Niu, D.J., Zhao, Y.C., 2006. Discussion on characteristics andtreatment technologies of food residue. Environ. Sanit. Eng. 14, 46–49.

Zhang, R., El-Mashad, H.M., Hartman, K., Wang, F., Liu, G., Choate, C., Gamble, P.,2007. Characterization of food waste as feedstock for anaerobic digestion.Bioresour. Technol. 98, 929–935.

Zhu, H.G., Parker, W.J., Basnar, R., Proracki, A., Falletta, P., Béland, M., Seto, P., 2008.Biohydrogen production by anaerobic co-digestion of municipal food waste andsewage sludges. Int. J. Hydrogen Energy 33, 3651–3659.