Evaluation of hydraulic retention time on hydrogen production from corn industry wastewater by dark fermentation

COD

chemical oxygen demand

cornWW

corn industrial wastewater

DF

dark fermentation

H₂

hydrogen

HRT

hydraulic retention time

OLR

organic loading rate

SBR

sequencing batch reactor

TS

total solids

MCFA

medium chain fatty acids

VFA

volatile fatty acids

VS

volatile solids

The inadequate treatment of industrial wastewater has caused social, environmental, and economic problems. Wastewater is a potential solution for creating value-added products, using it as a raw material instead of being seen as a problem. It could reduce freshwater consumption (UNESCO 2023) and promote the energy transition to reduce fossil fuel use. Dark fermentation (DF), an anaerobic biological treatment, is suitable for achieving a circular economy using a biorefinery approach (Ahmad et al. 2024). This approach involves converting solid or liquid carbohydrate-rich wastes into value-added byproducts, including hydrogen (H₂) and organic fatty acids (Regueira-Marcos et al. 2023).

The maize processing industry is one of the most common food industries worldwide concerning the production of fructose, glucose, starch, dextrose, food oil, corn flour, gluten, and sorbitol, as well as animal food production and biofuels (Beszedes et al. 2009). The DF process can effectively utilize corn industrial wastewater (cornWW), such as effluent from corn food processing, corn starch, flour production, and tortillas (Vázquez-López & Moreno Andrade 2024). Wastewater from the maize processing industry is rich in proteins, amino acids, carbohydrates, and nitrogenous compounds (Ewida 2020). This type of wastewater is characterized by its substantial organic matter content ranging between 13.6 and 52.8 gCOD/L, and alkalinity ranging from 0.2 to 3.3 gCaCO₃/L with a pH range of 10–14 (España-Gamboa et al. 2017; García-Depraect et al. 2017; Del Angel-Acosta et al. 2021; Valderrama-Bravo et al. 2022). The cornWW has relatively good biodegradability (DBO₅/DQO = 0.4) compared to the optimal value of 0.5 (Vázquez-López et al. 2023).

In Mexico, 14.4 million m³/year of cornWW is discharged into sewers and water bodies without treatment, causing various problems (Valero et al. 2018). CornWW stands out for its abundant availability, biodegradability, and energy potential derived from anaerobic biological processes. For example, cornWW is recognized as one of Mexico's top 20 priority substrates for biogas production, positioning it as a critical feedstock within the biorefinery concept (Harder et al. 2020; Kang et al. 2020).

Previous studies have indicated the possibility of producing value-added compounds and energy carriers from cornWW, such as H₂, through the DF process, although there is limited information available (García-Depraect et al. 2017, 2019; Buitimea-Cantúa et al. 2020; Del Angel-Acosta et al. 2021; Campos-Flores et al. 2023; Vázquez-López & Moreno Andrade 2024). H₂ is gaining recognition as a promising clean energy source. The DF process is one of the most innovative methods for producing H₂, offering a sustainable way to generate this fuel. Additionally, it contributes to organic waste management, reduces pollution, and supports a circular economy. As a result, this method represents a promising solution for the future of renewable energy (Ahmad et al. 2024).

The use of a discontinuous sequencing batch reactor (SBR) may be appropriate for processing this kind of wastewater since it has been observed that SBR can generate high H₂ production since they allow independent control of solid retention time and hydraulic condition time (Seengenyoung et al. 2019). However, it is essential to investigate operating conditions that can enhance H₂ production and other metabolites. Hydraulic retention time (HRT) is a key parameter in DF that influences H₂ production, microbial activity, stability, and overall process efficiency (Ghimire et al. 2015; Alvarez et al. 2022). The selection of the most suitable HRT is affected by the complexity of organic molecules, i.e. complex substrates necessitate a longer HRT to ensure sufficient time for their decomposition. However, it is important to consider that an appropriate HRT ensures ample H₂ yields (David et al. 2019). This investigation aimed to assess HRT's impact on H₂ production in an SBR system, using raw cornWW as feedstock.

CornWW was obtained from a tortilla-producing plant in Querétaro, Mexico. The samples were stored in plastic containers at 4 °C. The inoculum was obtained from a mesophilic anaerobic digester that treated wastewater generated by the flour industry and thermally pretreated at 105 ± 5 °C for 24 h to eliminate methanogenic archaea and H₂-consuming bacteria (homoacetogenic) and selecting spore-forming bacteria (Muñoz-Páez et al. 2019). After the heat pretreatment, the particle size was reduced to 210 μm using a coffee grinder HB80393 (Hamilton Beach, USA). Table 1 shows the characterization of the inoculum and cornWW.

Experiments were performed at 37 ± 1 °C with agitation at 120 rpm using a magnetic stirrer plate. At the beginning of each cycle, the pH was adjusted to 7.5 with HCl (5N) according to the standardized biohydrogen potential protocol described by Carrillo-Reyes et al. (2019). No buffer solution was added during the reactor operation. The SBR was operated using the following strategy: filling time 5 min, reaction time (47, 23, and 11 h) according to the evaluated HRT (4, 2, and 1 day(s), respectively), settling time (50 min), and draw time (5 min). A total of 20 operational cycles were evaluated, and the organic loading rate (OLR) was calculated by Inoue et al. (2014). The biogas (CH₄, CO₂, and H₂) composition was analyzed using an SRI 8610C gas chromatograph (SRI instrumental, USA) equipped with a thermal conductivity detector and a 30-m-long (0.53 mm ID) Carboxen-1010 PLOT column. The operating conditions for the chromatograph were as follows: nitrogen was used as the carrier gas at a flow rate of 4.5 mL/min. The injector temperature was set to 200 °C, the column temperature to 100 °C, and the detector temperature was fixed at 230 °C (according to Salazar-Batres & Moreno Andrade 2022). A sample volume of 5 mL was utilized for analysis. The percentage of H₂ in biogas was considered to calculate the total H₂ volume measurement.

The effect of HRT was evaluated using analysis of variance (ANOVA) with a 95% confidence interval to identify significant differences (p < 0.05), and Tukey pairwise comparison was made to detect significant differences among individual conditions in Minitab 17.

pH was measured using a digital pH meter (Hanna, Hi5522). The carbohydrate, protein, and lipid content were determined according to the methods described by Dubois et al. (1956), Lowry et al. (1951), and Mishra et al. (2014), respectively. VFAs were quantified by HPLC, according to González-Tenorio et al. (2020). Total solids (TS), volatile solids (VS), and chemical oxygen demand (COD) were determined according to standard methods (APHA 2005).

Five samples were analyzed to identify the microorganisms in the inoculum, raw corn wastewater (substrate), and at the end of three different HRT periods (cycles 7, 13, and 20). The microbial community was identified under the same reactor operating conditions, except for the variations in HRT. The collected samples were concentrated by centrifugation at 14,000 rpm for 4 min and then stored at −20 °C until analysis. Genomic deoxyribonucleic acid (DNA) was extracted from biomass samples using the PowerSoil^® DNA isolation kit (MOBIO, USA) according to the manufacturer's instructions. The DNA concentration was quantified by spectrophotometry using a NANODrop 2000c (Thermo Scientific, USA). The DNA was submitted to the Integrated Microbiome Resource for Illumina MiSeq sequencing. Universal primers for bacteria and archaea (515 FB: GTGYCAGCMGCCGCGGTAA and 806RB: GGACTACNVGGGTWTCTAAT) targeting the V4 variable region of the 16S rDNA were utilized. The sequence analysis pipeline was made following the methodology outlined by Zavala-Méndez et al. (2022).

Previous research has demonstrated the estimated productivity of using cornWW as a feedstock in the DF process without pH control. Del Angel-Acosta et al. (2021) achieved productivity of 178.8 mL H₂/L_reactor/day, and Vázquez-López & Moreno-Andrade (2024) reached 387.4 mL H₂/L_reactor/day, while our study obtained productivity at least two-fold higher (893 mL H₂/L_reactor/day) compared with the process employing longer HRTs and gradually reducing them over time. This approach favored H₂ production by promoting efficiency in organic matter degradation.

The highest H₂ yield (89.3 mL H₂/gVS_added) was obtained operating with an HRT of 1 day and the OLR 25.5 gCOD/L/day. This yield agrees with biohydrogen production from bulgur processing industry wastewater (90.8 mL H₂/gVS_added) obtained from thermally pretreated inoculums (Bouchareb et al. 2021). The industrial production of bulgur involves several processes, such as cleaning, cooking, drying, tempering, peeling, and milling, producing wastewater with high VSs concentration (35 gVS/L) and alkalinity of 235 mgCaCO₃/L (Bouchareb et al. 2021). The similarities between the bulgur and corn processing industries, along with the characteristics of cornWW, have resulted in comparable H₂ yield production.

Various strategies, such as incorporating nanoparticles into the DF process, can further increase the H₂ yields from food processing industrial wastewater (Bouchareb et al. 2022a). For example, using ferrite nanoparticles in milk processing wastewater resulted in significantly higher yields of 246 mL H₂/gVS_added (Lakroun et al. 2023). Other strategies, such as enzymatic pretreatment, can enhance carbohydrate concentration in feedstock, e.g., using enzymatic hydrolysis (α-amylase) in potato waste. This approach can boost H₂ production yields by 313% when applied in a DF process (Bouchareb et al. 2022b).

The percentage of removal for COD and VS increased as the HRT decreased (Table 2). In contrast, the removal of carbohydrates showed an opposite trend. This discrepancy may be due to the presence of other compounds in cornWW, such as proteins and lipids (Table 1), which may contribute to a lesser extent to H₂ production or related processes since all these macromolecules ferment simultaneously during treatment. Notably, the high removal percentages achieved are remarkable considering that the raw substrate was used without any pretreatment. This contrasts with other industrial wastewaters, which often require modifications for effective treatment (Kargi et al. 2012; Bouchareb et al. 2023).

Despite the absence of pH control throughout the process, values lower than 4.5 were not registered, avoiding inhibition of bacteria involved in DF (Khanal et al. 2003; Zagrodnik & Łaniecki 2017). When the HRT was reduced to 2 and 1 day(s), the final pH values increased to 5.7 ± 0.4 and 5.8 ± 0.5, respectively (p < 0.05), evidencing a significant difference concerning the HRT at 4 days (5.3 ± 0.1). The observed increase in pH during the HRT of 2 and 1 day(s) is attributable to the hydrolysis of proteins from cornWW acting as buffers during the process. Generated amino acids have groups that can accept or donate protons, allowing them to balance the pH (Reddi 2020; Dizon & Buckin 2023).

The generation of the mentioned byproducts, along with H₂, provides significant benefits as their market value ranges from approximately $600 to $2,000 US dollars per ton (Dinesh et al. 2018; Cieciura-Włoch et al. 2020). These biochemical byproducts are valuable in the industry due to their economic potential and diverse applications, serving as precursors for various biological processes (Dahiya et al. 2018).

On the other hand, the COD that was converted into MCFA and VFAs accounts for more than 20% of the remaining COD (Table 3). However, there are other non-quantified metabolites present, as well as an additional fraction of organic matter. This effluent has potential for future applications, such as in a methanogenic reactor, which could improve organic matter removal. Implementing this strategy would not only complete the product life cycle but also preserve its economic value.

The operation of the reactor led to the development of a microbial community with a composition of similar taxa (mainly Bifidobacterium, Atopobium, Caproiciproducens, Clostridium, Prevotella, Lactobacillus, and Enterococcus). However, the relative abundance of these taxa varied depending on the different HRTs applied. The absence of methanogenic archaea, which were removed from the inoculum, demonstrates the effectiveness of the heat treatment. Although Caproiciproducens (0.1%), Lactobacillus (0.1%), and Enterococcus (0.5%) were identified among the microorganisms present in cornWW, their representation in Figure 4 is absent due to their relative abundance being below 1%. Nevertheless, these microorganisms successfully established themselves during the DF process.

Diverse bacterial genera play distinct roles in the microbial composition observed in the HRT of 4 days. Bifidobacterium, Lactobacillus, and Enterococcus may be associated with lactic acid production (Hanchi et al. 2018; Chenebault et al. 2022; Luo et al. 2024) while Atopobium contributes to acetic acid and H₂ generation (Yang et al. 2019; Mercado et al. 2023). Clostridium and Prevotella may facilitate H₂ and butyric acid production from organic acids such as acetic and lactic acids (Reis et al. 2015; García-Depraect et al. 2019; Wang & Yin 2021). The presence and production of lactic acid are generally viewed favorably in the context of DF. However, prolonged HRT with lactic acid and acetic acid may lead to competition among microorganisms. Caproiciprodecens has been identified as a potential producer of caproic acid from organic acids, its dominance over H₂ production in the HRT of 4 days.

In this context, the presence and production of lactic acid were not considered a drawback but rather a favorable outcome within DF. García-Depraect et al. (2017) reported a similar metabolic pathway. However, in this study, it was observed that prolonged HRT, coupled with the presence of lactic acid and acetic acid, could lead to competition among the microorganisms involved. Specifically, Caproiciprodecens was identified as a potential producer of caproic acid from organic acids like acetic and lactic acids (Tang et al. 2022; Kurniawan et al. 2024). Consequently, caproic acid generation was observed to prevail over H₂ production.

On the other hand, reducing HRT to 2 and 1 day(s) resulted in approximately a 60% decrease in the relative abundance of Bifidobacterium, while the relative abundance of Atopobium doubled. Moreover, there was an increase in H₂ productivity (Figure 2), suggesting that the rise in Atopobium promoted H₂ and acetic acid production. However, although the exact metabolic pathway of Atopobium remains incompletely defined to date, it has been observed that, despite the absence of hydrogenases in its genome, this bacterium is capable of independently producing H₂ through a yet-to-be-characterized mechanism (Suzuki et al. 2018). Furthermore, the presence of Clostridium was crucial for H₂ production, either from organic acids or carbohydrates.

DF for H₂ production from cornWW shows promise, achieving remarkable yields without the need for strict pH control. A strategic reduction of HRT has led to a substantial increase in H₂ productivity, reaching a maximum of 893.6 ± 0.1 NmL H₂/L_reactor/day at an HRT of 1 day, driven by microbial adaptation. Reactor operational results demonstrated that decreasing the HRT increased the final pH values, as well as the percentages of COD and VS removal. The process results showed that a COD was converted into MCFA, and VFAs accounted for more than 20% of the remaining COD. The primary byproducts of H₂ generation were acetic and butyric acid, along with other valuable byproducts, including caproic acid, emphasizing the potential for resource utilization. Microbial community analysis reveals specific genera, such as Atopobium and Clostridium, playing key roles in H₂ and VFA production.

Financial support provided by DGAPA-UNAM through the PAPIIT project IN102722 is gratefully acknowledged. The project Grupos Interdisciplinarios de Investigación of the Institute of Engineering UNAM is also acknowledged. Monserrat Vazquez-López thanks CONAHCYT for the scholarship (702623). Gloria Moreno and Jaime Perez are acknowledged for their technical assistance.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.