Direct Fermentative Hydrogen Production from Cellulose and Starch with Mesophilic Bacterial Consortia

Hydrogen is widely deemed as a promising energy carrier of the future. Currently, most of the hydrogen is delivered from the non-renewable fossil fuels – mainly through the steam reforming of methane (Sgobbi et al. 2016). In order to produce a more sustainable fuel, hydrogen production processes should avoid or minimize CO₂ emissions. Among many methods of hydrogen generation, biological hydrogen production has been of great interest in recent years (Trchounian et al. 2017).

Dark fermentation is an acidogenic decomposition of carbohydrate rich substrates. Dark fermentation has gained much interest due to its simplicity, high hydrogen production rates, and versatility of potential substrates (Łukajtis et al. 2018; Mohammed et al. 2018). Potential feedstock for hydrogen production via dark fermentation is discarded lignocellulosic biomass from agriculture, forestry and food processing. It is the most abundant raw material in nature with annual worldwide production exceeding 220 billion tons (Kumar et al. 2015). Moreover, it is renewable and has high carbohydrate content. The main component of lignocellulose is cellulose. However, lignocellulosic biomass is difficult to utilize as a feedstock for biofuel production due to its complex structure (Kumar et al. 2015; Łukajtis et al. 2018). Therefore, it may require pretreatment before biofuel production. Pretreatment of complex biomass can be done with physical, chemical, physicochemical or biological procedures (Bundhoo et al. 2015; Cai and Wang 2016). However, many pretreatment methods are not economically feasible and may produce toxic compounds such as furfural and phenolic compounds, which could inhibit hydrogen fermentation.

Complex substrates could be hydrolyzed in a non-sterile environment with mixed anaerobic consortia (like anaerobic sludge). They are the natural source of microorganisms with a wide range of hydrolytic and catabolic enzymatic activities (Gadow et al. 2012; Wang and Yin 2017). However, environmental sources for cellulolytic H₂ producing consortia often contain organisms that consume or prevent maximum H₂ yields, such as methanogens, homoacetogens, sulfate-reducing bacteria (SRB), propionate producers or lactic acid bacteria (Saady 2013; Bundhoo et al. 2015). Therefore, pretreatment of mixed consortia is often necessary to suppress hydrogen consuming bacteria. Its principle is that many H₂ producers like Clostridium and Bacillus can form spores, which survive the extreme pretreatment conditions. Since many H₂ consumers are non-spore formers, this process usually enriches H₂ producers and represses H₂ consumers (Saripan and Reungsang 2015). Among many pretreatment methods like aeration, heat, acid, alkaline, microwave, and UV radiation the heat treatment is the most widely used, due to its simplicity and easy control (Wang and Yin 2017). Moreover, the most efficient inoculum pretreatment method shows great variation when different feedstocks are used. In a recent study, acidic pretreatment was found the most effective during hydrogen production from xylose (Mockaitis et al. 2020). On the other hand, Yang et al. (2019) showed that the highest hydrogen yield from antibiotic fermentation residue was obtained with alkaline treatment. Therefore, it is necessary to analyze inoculum pretreatment methods for more complex substrates.

Due to the complex structure of lignocellulose biomass most reports in literature concern application of its hydrolysates for hydrogen production (Kumar et al. 2015). This approach is not economically advantageous, because an additional stage of biomass hydrolysis is required. Consolidated bioprocessing is an alternative in which cellulase production, substrate hydrolysis, and fermentation are carried out in a single step by mixed bacteria cultures that express appropriate cellulolytic enzymes (Nagarajan et al. 2019). Only a few studies were conducted with untreated cellulose (Table I). Unfortunately, yields of H₂ produced by the direct fermentation of cellulose by bacteria are still low. In a recent study, Deng et al. (2019) compared hydrogen production form raw and acid-treated grass silage. The pretreated silage led to a hydrogen yield of 68.26 ml/g_vs, which was 3-fold higher compared to the untreated substrate. So, the enrichment of microorganisms, which have satisfying fermentation performance, is still a daunting task. The theoretical maximum H₂ yield of dark fermentative hydrogen production is 4 mol H₂/mol_hexose (Dinesh et al. 2018). Recently, a high yield of 1.92 mol H₂/mol_hexose was obtained with co-culture of Clostridium termitidis and Clostridium beijerinckii with cellulose as a substrate (Gomez-Flores et al. 2017). The role of C. termitidis was hydrolysis of cellulose, while C. beijerinckii was responsible for the enhancement of hydrogen production. In another study, Clostridium sartagoforme FZ11 attained 0.53 mol H₂/mol_hexose from untreated microcrystalline cellulose (Zhang et al. 2015).

The dark fermentation process can be performed at thermophilic (40–90°C) or mesophilic conditions (20–40°C). Thermophilic fermentation of simple substrates is often considered better in terms of hydrogen yield (Kumar et al. 2015). However, in the case of cellulose H₂ yields are often below 1 mol H₂/mol_hexose. Carver et al. reported 0.35 mol H₂/mol_hexose during the thermophilic fermentation of microcrystalline cellulose by the mixed microbial consortium (Carver et al. 2012). On the other hand, a stable yield of 2.52 mol H₂/mol_hexose was maintained for 190 days of the continuous process with heat-treated anaerobic sludge and cellulose as a substrate (Jiang et al. 2015). Mesophilic processes require lower energy input since they are conducted at lower temperatures. Some batch studies achieved promising yields ranging from 1.7 to 2.09 mol H₂/mol_hexose (Ren et al. 2010; Bao et al. 2016). As shown in Table I, only a few studies described long term mesophilic hydrogen production from cellulose (Gadow et al. 2012).

The present work investigates the effect of inoculum pretreatment and medium composition on hydrogen production from untreated cellulose at mesophilic conditions because direct cellulose utilization represents a promising alternative to multistage processes. Two different pretreatment temperatures (90 and 100°C) were selected to prepare mixed cultures and their hydrogen fermentation performances were comprehensively assessed. The results were compared with those obtained for starch as a substrate because starch is also composed only of glucose monomers, but its hydrolysis is easier. Furthermore, the stability of enriched cultures was assessed during repeated batch processes. The effect of medium composition and pH conditions were also analyzed during long term operation. The goal of this study was to enrich the mixed bacteria culture on cellulose as the substrate and to use it in the repeated fed-batch system for mesophilic hydrogen production.

The experiments showed that heat pretreatment of inoculum was necessary for H₂ production when starch and cellulose were used as the substrates. When pretreatment was not applied prior to inoculation, methane and CO₂ were the only gaseous products. It was carried out 90 and 100°C, because these temperatures are the most often used to enrich bacteria for hydrogen production (Wang and Yin 2017). After heat pretreatment, the produced biogas in all samples contained only hydrogen and carbon dioxide. No hydrogen was produced in reference experiments with a substrate-free medium (cellulose or starch).

Starch as a substrate with inoculum pretreated at 90°C. The initial batch tests were performed with the anaerobic sludge pretreated at 90°C. Fig. 1a presents hydrogen production during enrichment on starch. During the first cycle, only traces of hydrogen were detected. However, starch was utilized completely (Table II) with simultaneous production of acetic and butyric acid (Fig. 1a). This fact indicated microbial activity, but it was not directed towards hydrogen production. Therefore, after 12 days the bacteria in the bottle were subjected to the second treatment at 90°C and were used as an inoculum for the next cycle. This approach led to rapid hydrogen production to 0.58 l H₂/l_medium during the first 24 hours of the second cycle. At the same time formation of acetic and butyric acid was observed, which is characteristic for fermentative hydrogen production (Gadow et al. 2012; Łukajtis et al. 2018). Afterward hydrogen was consumed with a simultaneous increase in acetate concentration. It suggests the activity of hydrogen consuming microorganisms, such as SRB or homoacetogens. These two groups of bacteria are capable of hydrogen consumption and acetate production (Plugge et al. 2011; Saady 2013). Acidogenic microorganisms responsible for hydrogen production are fast growers, while hydrogen consuming bacteria are usually growing slower (Nagarajan et al. 2019). As a result, hydrogen consumption was observed after several days. The main bioreactions during hydrogen production and consumption are presented in Table III. The maximum H₂ yield (4 mol H₂/mol_glucose) is obtained when acetate is the main by-product, but this is true only when acetate is not formed in other metabolic pathways. In this study, acetate was formed in pathways leading to the consumption of hydrogen. Consequently, no hydrogen was detected at the end of the seco nd cycle.

Fig. 1.

Starch as a substrate with inoculum pretreated at 100°C. A small volume of hydrogen (0.32 l H₂/l_medium) was produced during the first 24 hours in the first cycle, but then it was quickly consumed with simultaneous high production of acetate (Fig. 2). Thus, this cycle was similar to the process with bacteria pretreated at 90°C. Therefore, the bacteria culture after the first cycle was heat-treated again at 100°C for 20 minutes before the second cycle. It resulted in high hydrogen production of 1.2 l H₂/l_medium. Hydrogen was produced at a high rate which is characteristic of the dark fermentation process (Table II). The average HPR reached 26 ml H₂/l-h. This time hydrogen was not subsequently consumed, as it was observed when the inoculum was pretreated at 90°C, but its volume was stabilized. A similar volume of hydrogen was produced also in the next cycle. In contrast, Wang et al. (2011) observed that increasing the temperature of inoculum pretreatment from 80°C to 100°C resulted in decreased H₂ yields from glucose. In this study, the main metabolites were butyric and acetic acid (Fig. 2). Formic acid was also present at concentrations of about 0.5 g/l. These results are in line with another study, which reported butyric (2.8 g/l), acetic (0.8 g/l), and formic acid (0.2 g/l) as main byproducts of dark fermentation of glucose by heat-treated anaerobically digested sludge (Baghchehsaraee et al. 2008).

Fig. 2.

Additionally, the development of the dark color in the medium during the process suggested activity of sulfate reducers. The dark precipitate of iron sulfides is often useful to indicate the activity of SRB (Bernardez and de Andrade Lima 2015). The SRB mainly use sulfate, the most oxidized form of sulfur, as the terminal electron acceptor. Therefore, at this point, 0.2 g/l of MgSO₄ in the medium was replaced with 0.165 g/l of MgCl₂ to reduce the concentration of this electron acceptor. This change did not have much influence on hydrogen production. However, it resulted in the change of organic acids concentration – butyric acid became dominant in the subsequent cycles with a lower concentration of acetic acid. High butyric to acetic acid ratio is characteristic of dark fermentative hydrogen production (Lin et al. 2008; Łukajtis et al. 2018). Moreover, the dark color of the bacterial culture was no longer observed. Hydrogen production remained stable in the next two cycles. Therefore, after 120 days of the process bacteria were successfully enriched for hydrogen production from starch. The complete starch consumption in each cycle was observed and H₂ yield reached 1.77 mol H₂/mol_hexose in the final cycles (Table II). Lin et al. (2008) obtained H₂ yield of 1.1 mol H₂/mol_hexose from starch with heat-treated (100°C) sludge. Similar values are often reported for mesophilic hydrogen production processes (Kumar et al. 2015; Wang and Yin 2018).

Cellulose as a substrate with inoculum pretreated at 90°C. Hydrogen production in the first cycle with cellulose was low and after 12 days H₂ was completely consumed (Fig. 1b). Moreover, high substrate utilization was observed (Table II). Acetic acid concentration reached 3.36 g/l, while lesser concentrations of propionic and butyric acid were observed (Fig. 1b). On the other hand, acetic and butyric acid concentration was similar during the process with starch (Fig. 1a). This showed that, despite the same inoculum used in both processes, in the case of cellulose utilization other microorganisms with different metabolic pathways were active. The bacteria culture after the first cycle was again heat-treated at 90°C and used as inoculum for the subsequent cycle. However, it did not result in the improvement of the volume of the produced hydrogen. The second cycle was similar to the first one, also in terms of produced metabolites and suggested the activity of hydrogen consumers. Low hydrogen production indicated that enrichment of the mixed bacteria culture in these pretreatment conditions was not possible.

Cellulose as a substrate with inoculum pretreated at 100°C. The pretreatment at 100°C led to higher hydrogen production from cellulose in the first cycle (Fig. 3a) when compared to the pretreatment at 90°C. However, H₂ yield was still low (Table II) with simultaneous acetate and propionate production (Fig. 3a). Therefore, heat treatment was repeated before the second cycle. As a result, a small volume of hydrogen (0.05 l H₂/l_medium) was detected at the beginning of the process, with simultaneous acetate production. H₂ production started after a lag time of 16 days and was associated with the formation of butyrate and ethanol. This indicates that the second heat treatment altered bacterial culture composition, which was correlated with a change in produced soluble metabolites. In the third cycle, only traces of H₂ were detected at the beginning, while simultaneous high acetate production indicated high activity of hydrogen consumers. The dark color of the medium in the bioreactor suggested the activity of SRB. Therefore, similarly to the process with starch, magnesium sulfate was replaced with magnesium chloride at this point. As a result in the next two cycles hydrogen production of around 0.2 l H₂/l_medium was observed, with concurrent lower acetic acid production. Similar results were obtained for starch, which confirms that SRB participated in the accumulation of acetate. In the presence of sulfate many fermentation products (propionate, butyrate, lactate) can be consumed by SRB with simultaneous acetate production (Plugge et al. 2011). However, when sulfate is limited, many SRB species grow fermentatively. In that case, they usually require methanogen or other hydrogen-scavengers to make this process thermodynamically favorable, because low hydrogen partial pressure is necessary for the utilization of substrates like lactate or ethanol (Baffert et al. 2019). In our experiments hydrogen concentration in medium without SO₄ ^2– was substantial. Therefore, it could be assumed that SRB activity in the sulfate-deprived medium was limited.

Fig. 3.

Cellulose consumption in each cycle was around 90% (Table II). However, generally, H₂ yields were low and did not exceed 0.5 mol H₂/mol_hexose, probably because other bacteria repressed hydrogen producers at this point. This experiment demonstrated that enrichment of the mixed bacteria culture is highly substrate-dependent – during enrichment on starch H₂ producing bacteria were able to outcompete the H₂ consumers.

In another experiment with cellulose as a substrate, a medium without sulfate was used from the beginning of the enrichment process. As a result, efficient production of hydrogen was observed in the initial cycles (Fig. 3b). H₂ production was six times higher than in the corresponding process with MgSO₄, although the utilization of cellulose was 8% lower (Table II). This indicates that minor changes in the nutrient composition can significantly alter the activity of the bacteria and the process itself. As shown in Table I, hydrogen production from cellulose was characterized by a very diversified profile of metabolites. In this study the main metabolites were acetic and propionic acid (Fig. 3b). Hydrogen production decreased in the subsequent cycles of the repeated batch process. At the same time, the concentration of propionic acid increased and became dominant in the fourth cycle. This indicates the activity of propionate producers – another group of hydrogen consuming bacteria. Thus, H₂ production by the culture enriched on cellulose was only temporary, because hydrogen consuming bacteria regained their activity after several cycles.

Only a few studies analyzed the effect of heat pretreatment at different temperatures on mesophilic hydrogen production (Baghchehsaraee et al. 2008; Ravindran et al. 2010). Baghchehsaraee et al. (2008) produced hydrogen from glucose with anaerobically digested sludge pretreated at 65, 80 or 95°C. The H₂ yield for untreated sludge and sludge pretreated at 65 and 95°C were 0.43, 2.30, 1.95 mol H₂/mol_glucose, respectively. The bacterial community analysis showed that elevated temperatures reduced species diversity. On the other hand, Ravindran et al. (2010) in experiments with inoculum (forest soil) pretreated at 65, 80, 95, 105, and 120°C reported the highest H₂ production for 105°C (1.92 mol H₂/mol_glucose). When inoculum was pretreated at 105°C, the presence of additional bacterial species was detected. The authors assumed that the higher temperature was needed for efficient spore germination. These studies show that there is no agreed conclusion on the most suitable pretreatment temperature. The change of pretreatment temperature results in the different microbial communities, which might be directly responsible for the different fermentation types and hydrogen yields. This study also shows that various bacteria were involved in the H₂ production process for 90 and 100°C pretreatments, which is reflected in different metabolic profiles.

The effect of a substrate on bacteria enrichment. The results obtained showed that enrichment on starch was more effective than on cellulose. This probably resulted from the fact that starch is more susceptible to hydrolysis. Cellulose utilization for hydrogen production requires different and more complex microbial activities. Therefore, lower H₂ production from cellulose could be due to the different degradative abilities of the microbial consortium related to the different substrates. Carbon recovery in the soluble metabolites varied from 39 to 81%, depending on the process (Table II). Carbon recovery of 67% is theoretically obtained for main fermentation reactions because four glucose carbons are recovered in acetate/butyrate, while two in the form of CO₂ (Table III – Eq. 1 and Eq. 2). Lower values indicate that more carbon was lost as CO₂ or biomass. Whereas, higher values suggest that carbon was recovered in undesirable reactions such as homoacetogenesis or propionate production. Generally, carbon recovery was higher for the processes with cellulose than with starch. Therefore, different bacteria with distinct metabolic pathways would enrich and dominate the culture on starch and cellulose. As results, different metabolites are formed. In another study with the anaerobically digested sludge pretreated at 70°C, the application of cellulose also led to the lower volume of H₂ when compared to starch (Gupta et al. 2014). It was explained by the difference in the microbial diversity of bioreactors – H₂ yields had a linear relationship with the number of observed species. Enrichment of class Clostridia and genera Bacterioides and Parabacterioides was found both for starch and cellulose as a substrate. On the other hand, Lachnospiraceae sp. was specific for cellulose cultures, while Streptococcus for starch cultures. Different bacteria composition could also be a reason of different H₂ and VFAs production for starch and cellulose in this study.

Other authors during studies with mixed anaerobic culture and glucose as a substrate also demonstrated that inoculum pretreatment could not permanently inhibit homoacetogenesis or methanogenesis (Shanmugam et al. 2016). In this study, no methane was detected in any experiment. The produced biogas in all of the samples contained only hydrogen and carbon dioxide. Cellulose consumption in each cycle was around 80 percent and a H₂ yield in the best cycle reached 0.97 mol H₂/mol_hexose (5.93 mmol H₂/g_cellulose) (Table II). Ren et al. (2010) reported the highest mesophilic batch H₂ production from microcrystalline cellulose with a yield of 2.09 mol H₂/mol_hexose with inoculum originating from cow dung compost. On the other hand, H₂ yield of 0.13 mol H₂/mol_hexose was obtained in another study with heat-shocked sewage sludge and cellulose as substrate (Gupta et al. 2014) (Table I). These discrepancies might be due to different microbial communities involved in the process. Most studies analyzed the effect of pretreatment on hydrogen production in single batch tests. Repeated batch experiments showed that it was necessary to evaluate the long-term effect of inoculum pretreatment on hydrogen production.

The repeated fed-batch process with cellulose. Finally, the repeated fed-batch experiments with cellulose as a substrate were performed. In batch experiments, bacteria culture could have experienced the nutrients deprivation periods. This could affect the bacteria enrichment process from anaerobic sludge. We assumed that more stable conditions of the repeated fed-batch process would have a positive effect on the enrichment of bacteria. For repeated fed-batch processes three types of inoculum were applied – fresh anaerobic sludge (AS1), anaerobic sludge, which was stored for 3 months at 4°C (AS2), and active bacteria culture enriched previously on starch (ES) (section 3.1). Anaerobic sludge was heat pretreated at 100°C and medium without SO₄ ^2– was applied. These processes were performed at controlled pH conditions (Fig. 4). At pH 5.0 no hydrogen production was observed for AS1 even after 30 days. Cellulose was not consumed and metabolic profile confirms the lack of microbiological activity (Fig. 5). H₂ yields with simple sugar substrates (like glucose) reported in the literature are often the highest at pH values between 5.0 and 6.0 (Dinesh et al. 2018). However, efficient cellulose hydrolysis has been reported at a pH around 7.0 (Lo et al. 2009). This suggests that pH 5.0 in the bioreactor with AS1 was too low to hydrolyze cellulose and release simple sugars for fermentation. Therefore, subsequent processes were carried out at pH 6.0.

Fig. 4.

Fig. 5.

During the process with AS2 at pH 6.0 hydrogen production started after 8 days. However, hydrogen production lasted only for two feeding cycles and then no hydrogen was produced (Fig. 4). At the same time a high concentration of acetic acid was observed (Fig. 5). This indicates that hydrogen-producing bacteria were outcompeted by homoacetogens what led to the accumulation of acetic acid and inhibition of the hydrogen production. The most stable hydrogen production process was obtained when AS1 was applied at pH 6.0. Here hydrogen production lasted for about 40 days. Then, a gradual decline in the hydrogen production rate was observed. This was accompanied by a substantial decrease in the amount of consumed cellulose and a lower concentration of the organic acids produced (Fig. 5). In the repeated fed-batch process butyrate was one of the main metabolites, while in batch experiments it was observed in low concentration. Stable hydrogen production in the repeated fed-batch process was also obtained for bacteria culture derived from the enrichment on starch (ES pH 6.0 – Fig. 4). However, this bacterial culture required more time to initiate hydrogen production and it was lower than for the AS1 at pH 6.0. Unfortunately, cessation of hydrogen production after about 40 days was observed for this process. Bacterial culture adapted previously to starch was, therefore, less effective in hydrogen production from cellulose. The highest yields were obtained for AS1 at pH 6.0 (Table IV). It was 1.04 mol H₂/mol_hexose (6.38 mmol H₂/g_cellulose) in the initial cycles. However, it decreased substantially during the process and dropped to 0.63 mol H₂/mol_hexose after 40 days. In another study with sewage sludge and cellulose as a substrate maximum hydrogen yield reached 0.1 mol H₂/mol_hexose during continuous operation at mesophilic conditions (Gadow et al. 2012). These results show that H₂ yields obtained from cellulose in long term processes are lower than those for batch studies. Nevertheless, obtained H₂ yields are promising and further research is needed to evaluate the long-term effect of inoculum pretreatment on direct hydrogen production from complex substrates.

The experimental results showed that pretreatment of the anaerobic sludge at 90°C was not sufficient to suppress H₂ consuming bacteria, both for starch and cellulose as the substrates. As a result, no hydrogen was observed at the end of batch experiments. However, anaerobic sludge was successfully enriched after pretreatment at 100°C on starch with a high and stable hydrogen yield of 1.7 mol H₂/mol_hexose. H₂ yield of culture enriched on cellulose reached 0.48 mol H₂/mol_hexose. Hydrogen consuming bacteria reduced H₂ production in a repeated batch process with cellulose because they regained their activity after several cycles. Results indicated that enrichment of mixed bacteria culture is highly substrate-dependent and enrichment on starch was more effective. Elimination of sulfate from medium resulted in 6.9 times higher an H₂ yield from cellulose, probably due to reduced activity of SRB. Hydrogen production in repeated fed-batch culture was effective in the initial stage of the process for AS1 at controlled pH 6.0, resulting in H₂ yield of 1.04 mol H₂/mol_hexose, but it became less efficient with time.