Improvement of Biohydrogen Production from Date Wastes by Thermotoga maritima Using a Continuous Anaerobic Membrane Bioreactor

Hydrogen (H2) is a promising energy carrier owing to its environmentally friendly properties and high energy content. The aim of this work is to achieve a continuous H2 production from a simplified culture medium composed of date wastes (DW) and natural seawater using the marine bacterium, Thermotoga maritima. Increasing the dilution rate (D) from 0.02 to 0.25 h− 1 in the continuous stirred tank reactor (CSTR) improved the volumetric H2 production rate (QH2) from 2 to 17 mmol/L.h. At a D above 0.25 h− 1, a washout of the bacterium cells was observed, resulting in a significant decrease of biomass concentration (BC) and then the drop in QH2. The production of H2 was improved by using a membrane bioreactor (MBR) due to the cell retention in the reactor. At a high D of 0.6 h− 1, the maximum QH2 was increased from 0.6 to 61.75 mmol/L.h, which is higher than those previously reported on continuous H2 production from different substrates. Moreover, increasing the hexose equivalent concentration in the date juice (DJ) from 30 to 60 mmol/L further enhanced the H2 production. It allowed a QH2 of 70.2 mmol/L.h which is interesting for a large scale production of H2 from date wastes.


Introduction
The increase of the global energy consumption and the faster depletion of fossil fuels along with the greenhouse effect emphasize the need for renewable sources. One of the clean energy carriers of interest is hydrogen (H 2 ) [1]. It has the highest energy content compared to conventional hydrocarbon and it has no carbon dioxide emissions when it is converted in fuel cells to generate electricity or burned in vehicle combustion engines [2,3].
The production of H 2 can be achieved via various conventional technologies including natural gas reforming, partial oxidation of hydrocarbons and coal gasification, which lead to the emissions of greenhouse gases. Therefore, the development of others H 2 production technologies from renewable resources could be a solution to the economic and environmental issues [4,5]. Biological H 2 production by dark fermentation has significant advantages including high H 2 production rates, low energy requirements and wide ranges of low cost waste feedstock sources [3,6]. In addition, the fermentations under high temperatures improve the solubilization of substrates, reduce contamination risks, the influence of H 2 partial pressure and reaction time [7,8].
Thus, Thermotoga maritima is a promising H 2 producer. It is a halophilic hyperthermophilic bacterium which is characterized by oxygen tolerances and fast growth kinetics [6,9]. T. maritima is able to metabolize a wide variety of simple and complex carbohydrates [10][11][12][13][14]. It harvests the energy by glycolysis via both the Embden-Meyerhof pathway (85%) as the main route and the Entner-Doudoroff pathway (15%) [10,15]. The maximum H 2 yield (4 mol H 2 /mol hexose) is associated with the conversion of all hexoses to acetate as the major fermentation product due to the action of the bifurcating hydrogenase which accepts electrons from both nicotinamide adenine dinucleotide dehydrogenase (NADH) and reduced ferrodoxin [10,14,16]. However, the production of lactate was observed in the presence of substrates difficult to hydrolyze which led to different metabolite levels that modulate lactate dehydrogenase activity [13]. The shift of the metabolism towards this product depend on culture conditions such as the composition of the reaction medium, the pH, the H 2 partial pressure, the oxidative stress and the transition to stationary phase [13,15]. Hence, the continuous system was proposed to avoid the H 2 partial pressure buildup by releasing the produced biogas from the reactor [17].
The fermentative H 2 production from agriculture crops, residues and by-products is considered promising for a sustainable development of bioenergy production [7]. Date fruits (Phoenix dactylifera L.) represent the abundant agricultural crops in the arid and semi-arid regions of the Middle East and North Africa. They are considered as an indispensable fruits due to their health, nutritional and economic value. They contain soluble sugars (72-88% dry matter basis), dietary fibers, proteins, fatty acids, amino acids, vitamins and minerals [18].
In Tunisia, date palm oases cover 46,000 ha with about three million trees. This sector represents 4% of the total value of agricultural production and 13% of Tunisian agricultural exports [19]. However, a considerable amount of the fruit is damaged during the different processing activities (picking, storing and conditioning processes). This quantity of date could produce approximately 30,000 tons of sugar [20]. Saafi et al. [21] reported that date pulp of low quality (Khalti) contains 51.66% reducing sugars (glucose and fructose) and 6.63% sucrose of the dry weight. Therefore, date juice (DJ) obtained from the low quality date fruits could represent an interesting substrate for H 2 production.
Until today, T. maritima was more studied in batch operations to identify the optimal cultivation parameters such as pH, H 2 partial pressure, stirring speed, gas sparging, nitrogen sources, sulfur sources, inorganic compounds and culture/headspace ratio [10,12,22]. However, continuous H2 production systems are more attractive than batch operations for industrial applications due to better process stability and higher H2 production rates for smaller reactor volumes [17,23].

3
Dilution rates (D) and influent substrate concentrations are the most important parameters in a continuous process. Increasing substrate concentration within a certain range enhanced fermentative H 2 production [4,24]. Moreover, bioreactor configurations have a key role in achieving high H 2 production. Systems that are robust, resistant to operational parameters fluctuations and characterized with high performance are preferred [23].
Continuous stirred tank reactors (CSTR) are widely used for the fermentative H 2 production due to their simple operation, the higher culture medium homogenization and the high mass transfer efficiency [23,24]. However, they have some limitations such as cells washout and unstable biomass levels at high dilution rates, resulting in lower H 2 production rates [23,24]. Therefore, membrane bioreactors (MBR) could be the promising H 2 production systems because of their capacity to prevent the washout effect and to increase biomass retention even at high dilution rates [25,26]. Previous works showed high H 2 production rates in the MBR at high dilution rates compared to the suspended cell cultures system (CSTR) [23,27,28]. Various configurations of MBR with different membrane design, pore size and materials are available [23,25,29,30].
In this study, the performances of continuous H 2 production in a simplified culture medium composed of date wastes and seawater, by T. maritima were conducted and evaluated in a controlled CSTR and MBR processes. The effects of dilution rate and feed substrate concentrations variations were examined.

Preparation and Composition of Date Juice
Low quality of fresh date fruits (Phoenix dactylifera L.) of different varieties (Deglet Nour, Ghars Souf, Khalti, Kentichi and Allig) were collected from local market in Tunis. Seeds-free date fruits were cut into small pieces and crushed at high speed by an electric blender. The extraction of soluble sugars was performed by soaking date pulpes in warm distilled water (80 °C) at a ratio of 1:2 (w:v) for 60 min [31]. The obtained juice was filtered through a stainless steel sieve (0.5 mm) and centrifuged for 15 min at 5000 rpm for excluding cellulosic fibers and the insoluble matter. Then, it was immediately stored at − 20 °C until use.
The physical and chemical characteristics of DJ such as total solids (TS), volatile solids (VS), total kjeldahl nitrogen (TKN) and reduced sugars were determined according to standard methods [32]. The volatile solids content represent 98% of the total solids in DJ. The organic fraction was composed of high sugar concentration (204 g/L). Soaking DW in hot water improved the concentration of soluble sugars due to their dissolution in the liquid fraction. Hence, the concentrations of glucose and fructose increased from 17 to 15 g/L to 80 and 76 g/L, respectively while sucrose concentration decreased from 152 g/L to 29 g/L. The nitrogen content was 4.9 g/L and the average C/N ratio of DJ was determined as 40. Saidi et al. [12] reported that glucose and fructose were preferentially used for T. maritima growth. Therefore, DJ could represent a favorable substrate for H 2 production. In this study, it was used at different dilutions with the seawater in order to obtain various initial concentrations of hexose equivalent that were expressed as the sum of glucose and fructose concentrations.

Bacterial Strain and Preculture Conditions
Thermotoga maritima DSM 3109 was purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ; Braunschweig, Germany). It was grown anaerobically in a simplified culture medium containing natural seawater, NH 4 Cl (1 g/L) and cysteine HCl (0.3 g/L) as described by Saidi et al. [10]. The pH was adjusted at 7.0 with 2 M NaOH. DJ was used as substrate with initial hexose equivalent concentration of 30 mmol/L. The culture medium was then distributed into 500 mL serum bottles with a working volume of 300 mL and flushed with sterile nitrogen gas (N 2 ) to achieve anaerobic conditions. After inoculation with T. maritima (10%, v/v), serum bottles were incubated for 18 h at 80 °C without shaking and used as preculture to inoculate the bioreactors.

Bioreactors Configurations and Operating Conditions
T. maritima cells were cultured under a continuous mode in two bioreactor configurations: the CSTR Biolafit fermenter ( Fig. 1) and the MBR Biolafit fermenter coupled to an external membrane system (Fig. 2). The fermenter was equipped with a redox and pH probe connected to a programmable controller to maintain the pH at 7.0 by the addition of 2 M NaOH. The simplified culture medium was kept anaerobic by a continuous sparging with N 2 gas (30 mL/min). The temperature was maintained at 80 °C (± 0.5) and the agitation speed was set at 150 rpm. The outlet gas was condensed in a water cooler (7 °C) in order to minimize the evaporation of volatile end products.
The memdrane system (YUASA Membrane Systems: ED-03SPH) consisted of a fermentation tank (CSTR), a circulation tank (1 L), membrane filtration unit, permeate tank and three peristaltic pumps for feed culture medium, permeate withdrawal and retentate recirculation. The filtration unit consisted of an external filtration module with a nominal cut-off filter of 0.2 μm and a filtration surface of 0.014 m 2 . The maximum pressure drop over the filter was 4 bars [33].
The CSTR mode was conducted in the Biolafit fermentor (2 L) with a working volume of 1 L inoculated with 200 mL of a preculture of T. maritima (20%, v/v). It was started in the batch mode for approximately 16 h to allow the growth and the acclimatization of the culture. After that, it was changed to the continuous mode with additional feed culture medium. The flow rates of the feed and effluent were controlled using digital peristaltic pumps to ensure different D of 0.02, 0.04, 0.05, 0.06, 0.1, 0.25, 0.45 and 0.6 h − 1 . Each condition was maintained for at least 8 days and average data were obtained when the system reached a pseudo steady state characterized by stable concentrations of fermentation products, H 2 and CO 2 .
The MBR was operated in the CSTR mode for 72 h at a dilution rate of 0.1-0.2 h − 1 and then the microfiltration membrane module was installed and it was changed to the MBR mode with a higher D of 0.25-0.6 h − 1 . The fermentation medium was bypassed to the microfiltration membrane unit allowing the separation of cells which were recycled into the reactor with high recirculation ratios (60 to 80%).

Analytical Methods
Biomass growth was monitored by measuring the optical density of the culture at 600 nm (OD 600 ). In addition, cell dry weight (CDW) was used to quantify the concentration of the biomass in the bioreactor. For CDW determination, a culture volume of 50 mL was centrifuged at 5000 rpm for 15 min, and then the cell pellet was dried at 105 °C in hot air oven to a constant weight (48 h). The CDW samples were done in duplicate.
For quantitative measurement of H 2 gas from the bioreactor's headspace, gas samples (1 mL) were taken and injected automatically into a gas Chromatograph (GC, Perichrom Company). It was equipped with a thermal conductivity detector (TCD) and a concentric CTR1 column (Alltech). Argon gas was used as the carrier gas with a flow rate of 20 mL/min. The temperatures of the detector, injector, and oven were 100 °C, 100 °C, and 40 °C, respectively. The quantification of H 2 was calculated using WINILAB III software (Perichrom). The online measurements of the CO 2 content were realized through a carbon dioxide probe (Vaisala Series GMT221) connected to a transmitter.
Sugars (glucose and fructose), acetate and lactate were quantified by high performance liquid chromatography (HPLC), Agilent 1200, using an Aminex HPX-87 H ionexchange column (300 × 7.8 mm, Biorad, Hercules) coupled to a refractive index detector. The HPLC was operated with 5 mmol/L H 2 SO 4 as a mobile phase at a flow rate of 0.5 mL/ min. Before injection, culture samples were centrifuged at 14 000 rpm for 5 min. The supernatants were filtered through a 0.45 μm cellulose acetate membrane.
Carbon balances at steady states were calculated from the molar fluxes of the products (biomass, organic acids and CO 2 in mmol/h) divided by the molar flux of the consumed hexose equivalent (mmol/h).

Effect of Initial Substrate Concentration
The previous experiences showed that the composition and the initial concentration of carbon source have an important effect on the bacterium growth and the product formation. Therefore, DJ was used with different dilutions to obtain various initial hexose equivalent concentrations (10-50 mmol/L). Effects of the input substrate concentrations on the CSTR performance at a D of 0.02 h − 1 are presented in Fig. 3.
Increasing hexose equivalent concentration from 10 to 30 mmol/L improved the volumetric H 2 production rate The obtained Y H2 was considered important when it was compared to the maximum theoretical Y H2 of 4 mol H 2 / mol hexose. These results are rivaling to that achieved by Dreschke et al. [17] using the closely related T. neapolitana, in complete culture medium contained glucose as substrate (27.8 mmol.L). They obtained Y H2 of 3.4 mol H 2 / mol glucose in a CSTR at a D of 0.04 h − 1 . de Vrije et al. [34] obtained Y H2 of 3.3 mol H 2 /mol glucose when using the extreme thermophile Caldicellulosiruptor saccharolyticus in a CSTR at a D of 0.05 h − 1 with 24.6 mmol/L glucose. Therefore, the high production of H 2 that was achieved in this study by T. maritima using a simplified reaction medium could be attributed to DJ composition. It provided amino acids, peptides, vitamins, mineral elements and other compounds necessary for bacterial growth [35].
Increasing hexose equivalent concentration from 30 to 50 mmol/L resulted in a reduction of the Q H2 from 2 mmol/L.h to 1.68 mmol/L.h (Fig. 3A). In addition, Y H2 decreased from 3.5 mol H 2 /mol hexose to 2 mol H 2 /mol hexose resulting in a decline of the measured to the theoretical maximum H 2 yield from 86.75 to 50%. This is could be due to a shift in hexose metabolism with a decrease in acetate production, which is considered as the end product for higher H 2 production. However, a considerable lactate formation (23 mmol/L) was observed. This is a competing pathway reducing the H 2 production because the conversion of pyruvate to lactate by lactate dehydrogenase requires oxidation of the NADH [10,14,36,37]. Furthermore, the carbon balances were calculated. They were 98.9% and 98.5% for a feeding hexose equivalent concentration of 30 and 50 mmol/L, respectively. These results confirm that acetate and lactate were the only fermentation products. So, no other metabolite may be generated.

Effect of the Applied Dilution rate
The controlled CSTR was operated with a diluted DJ containing 30 mmol/L of hexose equivalent at different dilution rates (0.02-0.6 h − 1 ). The system reached to steady state condition approximately after 45 h of operation. The average values of the different parameters in the final days of each operating condition were calculated as shown in Fig. 4; Table 1.
Both the volumetric and the specific H 2 production rates (Q H2 and q H2 ) were positively correlated to the D increase ( Fig. 4; Table 1). They increased from 2 to 17 mmol/L.h and from 3.77 to 19.1 mmol/g.h, respectively as D increased from 0.02 to 0.25 h − 1 . However, further increase of D above 0.25 h − 1 affected negatively Q H2 and q H2 which dropped to Contrary to the Q H2 , the Y H2 was negatively affected by the D increase. At the lower D of 0.02 h − 1 , it was 3.47 mol H 2 /mol hexose which represents 86.8% of the theoretical maximum H 2 yield. When D increased to 0.45 h − 1 , Y H2 declined to 1 mol H 2 /mol hexose which corresponds to 25% of the theoretical maximum yield. This is could be due the reduction of acetate formation from 57 to 4 mmol/L, whereas lactate production increased from very low concentrations to 15 mmol/L. In these cultures, hexose equivalent consumption decreased from 96 to 36% (Fig. 4).
A similar response to the reduction in Y H2 was observed by Dreschke et al. [17] using T. neapolitana in a CSTR containing glucose as substrate. They showed that Y H2 decreased from 3.4 to 1.8 mol H 2 /mol glucose when D increased from 0.04 to 0.2 h − 1 . They reported that the raise of the D induced a shock and a shift of T. neapolitana metabolism towards the lactate pathway as a response to changeable conditions. This is allowed the bacteria to continue the fermentation with a lower energy yield. Similarly, de Vrije et al. [34] showed a decrease of the Y H2 from 3.3 to 3 mol H 2 /mol glucose when D increased from 0.05 to 0.35 h − 1 using Caldicellulosiruptor saccharolyticus in a CSTR. Contraray to Y H2 , they reported that H 2 productivity rate increased from 4.2 to 12.4 mmol/L.h.
For a better comparison of the CSTR performances at different D, an average value of biomass concentration of each operating condition is presented in Table 1. BC of T. maritima was positively influenced by the raise of the D. It increased slightly with an adequate acclimatization from 0.5 to 0.89 g CDW/L as D increased from 0.02 h − 1 to 0.25 h − 1 . The biomass yield (Y X ) increased with increasing D, which can be due to the use of a larger part of hexose for bacterium maintenance at low growth rates as reported by previous studies [34,36]. The mean carbon balance of the continuous cultures at different D (0.02 to 0.25 h − 1 ) was determined as 99.08%.
At the higher D (0.6 h − 1 ), BC dropped to 0.1 g CDW/L, resulting in a low substrate consumption efficiency (8%), a strong decline of Q H2 (0.6 mmol/L.h) and a lower value of Y H2 which represents 10% of the theoretical maximum yield. Moreover, the carbon balance decreased to 35.5%. Singh et al. [15] reported that the continuous culture of T. maritima was predicted to become unstable due to the washout of the biomass at a D of 1 h − 1 which was close to its maximum growth rate (µ max ). Therefore, the D value should not to be close or higher than the critical µ max of 0.69 h − 1 , which was determined from batch culture data of T. maritima to avoid the washout of cells and the reactor failure. Figure 5 shows comparative of Q H2 , Y H2 , BC and hexose equivalent consumption between the CSTR and the MBR at D of 0.25, 0.45 and 0.6 h − 1 . Results of the CSTR showed that the highest Q H2 of 17 mmol/L.h was occurred at D of 0.25 h − 1 . In this culture, Y H2 , BC and hexose equivalent consumption were 2.75 mol H 2 /mol hexose, 0.89 g CDW/L and 82%, respectively. At higher D (0.6 h − 1 ), the biomass concentration decreased to 0.1 g CDW/L indicating cells washout from the reactor, followed by a sharp decline of Q H2 (0.6 mmol/L.h). This growth limitation is common for the hyperthermophilic suspended cultures and considered the main obstacle to their industrial applications for H 2 production [15,17].

Improvement of H 2 Production by Using the MBR
The MBR outperformed the CSTR in term of resistance to cell-washout. In fact, a significant increase in BC was observed with increasing D due to a better retention of cells in the reactor. The highest BC of 5.9 g CDW/L was achieved at the higher D of 0.6 h − 1 , resulting in an improvement of Q H2 and Y H2 , which were 61.75 mmol/L.h and 3.43 mol H 2 / mol hexose, respectively. Moreover, the consumption of hexose equivalent became complete due to the accumulation of the active cells in the culture medium.  Figure 6 shows comparative of specific H 2 production rate (q H2 ) between CSTR and MBR. The q H2 values in the CSTR were higher than those in the MBR which could be attributed to the high food (F) to microorganism (M) ratios (F/M) in the CSTR. They were 6, 18 and 54 at D of 0.25, 0.45 and 0.6 h − 1 , respectively. The F/M ratios in the MBR were low and did not exceed 1.35 because of the higher biomass concentrations.
The higher values of Q H2 and Y H2 in the MBR, despite the low F/M ratios could be attributed to DJ composition which improved growth and microbial activity. In fact, it provided iron which was determined to be 2.5 mg/100 g dry matter basis [38]. This oligoelement was considered a key factor affecting dark fermentation pathway by increasing hydrogenase activity and the catalytic activity of the microbial system, as reported by Lee et al. [39]. They showed that a suitable supplementation of FeSO 4 as an iron source improved H 2 production from glucose in an MBR using mesophilic mixed microflora, even though a low F/M ratio. The maximum H 2 production rate increased from 26 to 41.6 L H 2 /d when the FeSO 4 concentration increased from 2.7 to 10.9 mg/L. Table 2 presents the effect of initial hexose equivalent concentration on average Q H2 , BC, hexose consumption,Y H2 , Y X , q H2 , Y actetate and carbon balances in the MBR at different D. During these experiments, the recirculation of biomass ratios were considered to determine the carbon balances.
The Q H2 and BC were positevely affected by increasing initial hexose equivalent concentration from 30 to 60 mmol/L. Then, the highest Q H2 (70.2 mmol/L.h) and BC (7.7 g CDW/L) were achieved with 60 mmol/L hexose The enhancement of H 2 production was due to the increase in cells concentration in the reactor that possess many active catalytic sites for efficient conversion of substrate to H 2 [15,23]. Moreover, it was attributed to the mainly conversion of hexose to acetate as established by the higher Y acetate ( Table 2). During this experiment, only 54% of the maximum theoretical H 2 yield was obtained. However, the higher Y H2 up to 80% of the maximum theoretical yield were achieved with the lower BC of approximately 4.3 g CDW/L, suggesting that a large part of the consumed hexose was used for maintenance. Similarly, de Vrije et al. [34] obtained the highest Y H2 of 3.6 mol H 2 /mol glucose which corresponds to 90% of the theoretical maximum yield at the low growth rates of C. saccharolyticus at the lower D of 0.1 h − 1 . High values of carbon balances (up to 98%) were obtained under all operating conditions, which confirmed that acetate, biomass and CO 2 (data not shown) were the main products of the fermentation. Table 3 summarizes different continuous reactor performances of various thermophilic and hyper thermophilic microorganisms. In this study, the maximum Q H2 of 70.2 mmol/L.h was obtained in the MBR using 60 mmol/L hexose equivalent. It was considered high when it was compared to the previous data of continuous H 2 production that were conducted in suspended cell bioreactor (CSTR). It is higher than that (2.31 mmol/L.h) achieved by the same bacterium from glycerol at a D of 0.05 h − 1 [40]. Moreover, it is higher than that of 11.6 mmol/L.h which was achieved by the coculture of C. saccharolyticus and C. kristjanssonii at a D of 0.3 h − 1 [41]. The only high H 2 production rate (121 mmol/ L/h) attained in hyperthermophilic conditions was achieved by the metabolically engineered Thermococcus onnurineus NA1 in a CSTR at a D of 0.3 h − 1 with carbon monoxide supply rate [42].
To date, several studies of H 2 production were conducted in the MBR with mesophilic or thermophilic mixed cultures and only a few studies were performed with hyperthermophilic microorganisms. The maximum Q H2 achieved in this study is equivalent to 40.44 LH 2 /L.d, which is in the upper range of previously reported H 2 production rates of mixed 1 3 fermentations in the MBR. It is about two times higher than that (19.86 LH 2 /L.d) obtained by Kim et al. [28] using the thermohilic heat-treated sludge from tofu processing waste. Moreover, it is higher than H 2 production rate of 11.4 LH 2 /L.d that was achieved by a mesophilic mixed culture using glucose [43] and that of 10.7 LH 2 /L.d using food waste [44]. Nevertheless, it is lower than H 2 production rate of 51.38 LH 2 /L.d that was obtained by Park et al. [30] from glucose when they used immobilized granular sludge at a D of 0.33 h − 1 .

Conclusion
This study examined the continuous production of H 2 from date wastes by T. maritima in a simplified culture medium.
Results of fermentations demonstrated that the MBR could be a promising system for continuous H 2 production even at high D. It outperformed the CSTR in term of H 2 production rate due to the retention of biomass in the reactor. In fact, at a higher D of 0.6 h − 1 , the maximum Q H2 was 61.75 mmol/L.h. Increasing the concentration of hexose equivalent from 30 to 60 mmol/L improved Q H2 to achieve 70.2 mmol/ L.h. Compared with other results, which were reported in the literature, the performance of the MBR was considered interesting. However, the membrane behavior during longterm operation should be optimized for the scale up.