Continuous two-stage lactate-driven dark fermentation process for enhanced biohydrogen production from food waste

Continuous two-stage lactate-driven dark fermentation process for enhanced biohydrogen production from food waste Lois Regueira-Marcos, Octavio García-Depraect , Raúl Mu˜ noz * Institute of Sustainable Processes, University of Valladolid, Dr. Mergelina, s/n, 47011 Valladolid, Spain Department of Chemical Engineering and Environmental Technology, School of Industrial Engineering, University of Valladolid, Dr. Mergelina, s/n, 47011 Valladolid, Spain ARTICLE INFO Keywords: Acidogenic fermentation Dark fermentation Hydraulic retention time Lactate production Organic waste Volatile fatty acids ABSTRACT The performance of a continuous lactate-driven dark fermentation (LD-DF) process based on the physical separation of lactate and hydrogen production was evaluated using simulated food waste (FW) as substrate. The optimum pH to maximize lactate production in the lactic reactor (LR) and the optimum hydraulic retention time (HRT) to maximize H 2 production in the hydrogen production reactor (HR) were investigated. The lack of pH control in the LR implied a decrease in pH to values as low as 3.7, with the maximum lactate titer reached at a controlled pH of 4.5. The controlled reduction of the HRT from 12 to 4 h strongly affected the volumetric hydrogen production rate (HPR) in the HR. The stepwise reduction of HRT to 6 h resulted in the maximum recorded HPR of 9.6 ±0.9 L H 2 /L-d, whereas a reduction to 4 h mediated a process collapse. The total consumption of lactate over carbohydrates in the HR suggested the occurrence of the LD-DF process at this stage. Principal component analysis (PCA) revealed a positive correlation between HPR and butyrate concentration in the HR. Additionally, HPR was negatively correlated with elevated levels of acetate and lactate in the HR broth. Overall, the application of a continuous two-stage LD-DF process allowed a significant reduction in HRT while maintaining high HPRs from FW. 1. Introduction The quest for affordable clean energy alternatives to fossil fuels is a priority to address environmental pollution and the growing global energy demand. The circular economy model, which proposes the use of waste as an energy source, is an attractive solution to help meet global energy needs. Indeed, biodegradable wastewater and residual biomass have a high potential not only as an energy source but also as a feedstock to produce marketable chemicals and materials [1]. Among them, food waste (FW) retains a huge amount of chemical energy, mainly in the form of easily biodegradable organic compounds [2,3]. In this context, high-value compounds such as methane (CH 4 ), hydrogen (H 2 ), or bioplastics have been produced from FW through fermentative processes [4,5]. It is estimated that 1.3 billion tons of FW are generated annually worldwide, a figure that represents one third of all food produced globally [6,7]. In 2020, about 58.4 million tons of FW were generated in the European Union, equivalent to about 127 kg of FW per capita [8]. Other geographic regions, such as Asia-Pacific, Sub-Saharan Africa, North America, and Latin America and the Caribbean, generate 465, 232, 168 and 127 million tons of FW annually, respectively [9]. In recent years, the biological production of H 2 has attracted attention in the European Union and beyond, as it promotes the ecological transition to a non‑carbon-based industry [10]. H 2 stands out among other fuels due to its high-energy density (2.7 times higher energy content than that of carbon-based fuels) and zero carbon dioxide (CO 2 ) emissions upon combustion [11]. Among the existing biological H 2 - producing processes, dark fermentation (DF) can provide a superior H 2 production performance with a wide variety of organic residues [12–14]. Household FW is widely available and has a high content of fermentable carbohydrates, which is the main direct or indirect H 2 precursor during DF [15]. In addition, the fermentative production of H 2 simultaneously synthesizes organic acids such as acetate and butyrate, resulting in a fermentation broth that can be further utilized in other innovative and profitable applications such as the production of biolipids, bioplastics and biogas, among others [16]. However, despite numerous research studies over the past two decades, the DF process of FW has not yet overcome a number of challenges that hinder its development at the industrial level [15]. Among * Corresponding author at: Institute of Sustainable Processes, University of Valladolid, Dr. Mergelina, s/n, 47011 Valladolid, Spain. E-mail address: [email protected] (R. Mu˜ noz). Contents lists available at ScienceDirect Journal of Water Process Engineering journal homepage: www.elsevier.com/locate/jwpe https://doi.org/10.1016/j.jwpe.2024.106116 Received 9 July 2024; Received in revised form 1 September 2024; Accepted 1 September 2024 Journal of Water Process Engineering 67 (2024) 106116 Available online 6 September 2024 2214-7144/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). them, the low and unstable H 2 production caused by the overgrowth of lactic acid bacteria (LAB) prevails as the Achilles’ heel of conventional DF [12]. Indeed, LAB represent a large fraction of the natural microbiota present in FW [17]. Apart from that, it is challenging to selectively wash out LAB from dark fermenters because the optimal growth conditions (e. g., pH, temperature, doubling time, etc.) for LAB are quite similar to those supporting the growth of H 2 -producing bacteria (HPB) such as Clostridium spp. [15]. In continuous processes, LAB are often detrimental to the DF process because they compete with HPB for substrate. The growth of LAB can also cause acidification of the culture broth due to the production and accumulation of lactate. Additionally, LAB have the potential to produce antimicrobial compounds called bacteriocins. As a result, the presence of LAB can ultimately lead to the displacement of HPB in the long-term operation of dark fermenters [17]. Sterilization steps or inoculum pretreatment have not only proven unsuccessful in preventing LAB overgrowth but also render the DF process economically unsustainable on an industrial scale [12]. Since LAB cannot be effectively washout out from the process, a novel strategy to produce H 2 from lactate rather than from carbohydrates has gained momentum in recent years. Lactate-driven dark fermentation (LD-DF) is based on pursuing a synergy between the production of lactate from carbohydrates by LAB and its subsequent fermentation to H 2 and other organic acid by-products by lactateutilizing HPB (LU-HPB) [12,15]. The LD-DF resembles the lactate cross-feeding that occurs in the human gastrointestinal tract and is grounded in a quite stable ecosystem [18,19]. In this sense, the LD-DF process would allow not only to avoid competition between LAB and HPB but also to make the process more robust and resilient by creating an interdependent microbial consortium without thermodynamic constraints [12]. In recent years, the role of process parameters such as the pH, total solids concentration [20], or hydraulic retention time (HRT) [21] on the performance of LD-DF of FW has been extensively investigated in singlestage reactor configurations. However, the physical separation of lactate and H 2 production from FW has not yet been tested. In this context, the physical decoupling of the two-stage lactate fermentation process allows independent optimization of the operating parameters, thereby maximizing the overall performance of the system. Furthermore, the success of the LD-DF approach for the enhanced H 2 production is highly dependent on maintaining a correct balance between the activity of LAB and LU-HPB [22], which represents a challenge in single-stage DF configurations. This study aims to design and evaluate a novel integrated two-stage LD-DF process with the goal of enhancing H 2 production from household FW. The effect of pH, which strongly determines the organic acid profile, was also investigated during continuous acidogenic fermentation of FW to determine the optimal operational pH that maximizes lactate concentration and selectivity. 2. Materials and methods 2.1. Substrate and biocatalysts The substrate used was a simulated FW designed to mimic typical household FW characteristics. The FW consisted of a mixture of potatoes (78 %), chicken breast (14 %), pork lard (4 %) and cabbage (4 %), representing sources of carbohydrates, proteins, lipids, and fiber, respectively [20]. The potatoes and chicken breast were previously boiled in an autoclave at 120 ◦C and 1.5 atm for 30 min to simulate the cooking process. This cooking process was known to be ineffective in preventing LAB proliferation, as described by García-Depraect et al. [9]. The whole mixture was then ground using an industrial blender (Sammic, XM-32, Spain), and stored in hermetically sealed plastic bags at −20 ◦C to avoid degradation. The mixture was characterized as follows: 53.1 ±3.7 % carbohydrates, 24.1 ±3.6 % protein, 18.1 ±0.7 % lipids and 4.7 ±0.6 % ash. The elemental composition of the simulated FW was determined to be 49.9 ±0.9 % carbon (C), 7.5 ±0.2 % hydrogen (H), 34.5 ±0.2 % oxygen (O), 3.8 ±0.6 % nitrogen (N) and 2.9 ±0.1 % phosphorus (P). Sulphur was not detected. The autochthonous microbiota of the FW was used as the biocatalyst in the lactate-producing reactor (hereafter referred to as LR) [9]. To achieve this, FW was naturally pre-fermented in batch mode for 24 h at 37 ◦C under a closed atmosphere (initially air) with non-controlled pH conditions. For the H 2 -producing reactor (henceforth referred to as HR), the inoculum source was digestate from a 100-L mesophilic anaerobic digester fed with restaurant FW. This digestate was subsequently pretreated at 90 ◦C for 20 min and then enriched in a 3-L chemostat reactor (with a 2 L working volume) fed with a lactose-based growth medium. The reactor was maintained at a HRT of 6 h and an organic loading rate (OLR) of 120 g COD/L-d. The microbial composition of the resulting H 2 - producing inoculum, as previously determined by amplicon 16S rRNA gene sequencing analysis, included dominant genera such as Lactobacillus, Klebsiella, Clostridium, Stenotrophomonas, and Acinetobacter, along other less abundant bacterial species. This HR inoculum was stored at 4 ◦C until use. Prior to use, the hydrogenogenic consortium was reactivated for 17 h at 37 ◦C in a 2.1-L gas-tight flask containing 0.1 L of the inoculum and 0.9 L of a mineral medium consisting of (in g/L) CaCl 2 ⋅2H 2 O 0.15, FeCl 2 ⋅4H 2 O 0.035, KH 2 PO 4 0.6, K 2 HPO 4 2.4, MgCl 2 ⋅6H 2 O 2.5, and NH 4 Cl 2.4. Lactose, at a concentration of 10 g/L, was used as the carbon source [23]. 2.2. Experimental approach and setup First, the effect of operational pH on the acidogenic fermentation of FW was investigated in continuous culture. The goal was to maximize the titer and selectivity of lactate while minimizing the loss of reducing power and carbon in the form of H 2 and CO 2 , respectively. Subsequently, using the identified optimum pH for lactate production, a new lactateproducing reactor (LR) was coupled with the H 2 -producing reactor (HR). The performance of the two-stage LD-DF system was then assessed, with particular emphasis on the effect of HRT on the volumetric H 2 production rate and process stability. The continuous production of lactate was performed in a 1.1-L continuous stirred tank reactor (CSTR) with 0.9 L working volume, while the continuous production of H 2 was carried out in a 1.2-L CSTR with 0.8-L working volume. The LR was made of glass with a PVC glass cover, while the body of the HR was made of PVC glass with a polypropylene cover. Both reactors were magnetically stirred (LBX instruments, S20 series stirred plate). As shown in Fig. 1, both reactor covers were equipped with feed and discharge ports, a pH probe, an alkali feed port, and a gas sampling port. In addition, the reactors were equipped with Dosiper C1R 4.5 L/H peristaltic pumps (Super Net Cali, Spain), which were automatically controlled by an in-house feed and discharge system. This system managed the feed via a timer and the discharge via a water level controller, ensuring accurate and consistent continuous operation of the reactors. The acidogenic off-gas produced was measured using an in-house wet gas flow meter, which works based on the well-known water displacement method. Low gas permeability tubing (Marprene® and polyethylene Tubepack®) was used. Both the LR and the HR were placed in a temperature-controlled room (37 ±1 ◦C) and magnetically stirred at 200 rpm. pH was measured and controlled by using a pH controller (BSV, EVOPH-P-5, Spain). A solution of NaOH 6 M was used to control the pH in both fermenters. 2.3. Influence of the pH on the continuous acidogenesis of FW The LR was operated in continuous mode for 32 days to study the effect of pH on the acidogenesis of FW. The LR was initially filled with 900 mL of FW with an initial total solid (TS) content of 5 % (w/w). No inoculum addition was required because LAB are naturally present in FW. The acidogenic fermentation process was initially started in batch mode without pH control until rapid acidification occurred. Once this L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 2 acidification was observed, the operation was switched to continuous mode. The FW continuously fed to the LR, with a fixed TS content of 5 %, was cooled at 4 ◦C to avoid degradation. The HRT was kept constant at 12 h throughout the operation, resulting in a fixed OLR of 152.2 g COD/ L-d (95.8 g VS/L-d). During the initial period of operation, the pH was not actively controlled and stabilized at a value below 4. The operational pH was then controlled and gradually increased to 4.5, 5.0 and 5.5. Each pH condition was maintained for 7–8 days. pH 5.5 was the highest pH tested because the observed performance indicated the onset of mixedtype fermentation. Liquid samples were collected periodically to analyze the organic acid profile and to assess the VS removal efficiency (VS removal ) at each pH condition. The volume and composition of the biogas produced were monitored daily to assess the onset of the oxidative decarboxylation pathway. 2.4. Two-stage LD-DF process The LR and HR were operated for 95 and 88 days, respectively, to evaluate the influence of HRT on the performance of the FW LD-DF process. First, the LR was initiated by filling it with 0.9 L of FW at 5 % TS (w/w), without the addition of any inoculum, as was done in the previous pH test. The LR was operated in batch mode for 1 day before starting the continuous mode by continuously feeding FW (cooled at 4 ◦C) at 5 % TS. On the other hand, the HR was filled with 720 mL of FW and 80 mL of inoculum (320 mg VS/L), previously activated according to the procedure described in Section 2.1. The HR was set up 7 days after the start of operation of the LR and was maintained in batch mode for 21 h before switching to continuous operation (by continuously feeding the LR effluent to the HR). The pH of the HR was fixed at 6.5, based on the results previously reported by Regueira-Marcos et al. [20], while the pH of the LR was kept at 4.5 (based on the results observed here in the tests conducted to evaluate the influence of pH on the continuous acidogenesis of FW). The entire operation was divided into 5 operational periods (i.e., P1–P5). The HRT was gradually shortened from 12 h to 10, 8, 6 and 4 h, corresponding to initial total OLRs of 71.6, 85.9, 107.4, 143.2 and 214.9 g COD/L-d, respectively. The HRT was defined based on the HR working volume (i.e., 0.8 L), therefore the HRTs corresponding to the LR (0.9 L working volume) were 13.5, 11.3, 9.0, 6.8 and 4.5 h in P1, P2, P3, P4 and P5, respectively. Table 1 summarizes the operating conditions used in the two-stage LD-DF process. Liquid samples of both reactors were taken periodically throughout the operation to determine the pH and the concentration of VS, COD, carbohydrates, and organic acids. The volume and composition of the biogas produced were also monitored periodically. The biogas production rate (BPR), volumetric H 2 production rate (HPR), H 2 yield (HY), H 2 content, H 2 production stability index (HPSI) and the organic acid profile were used as process performance indicators. Other parameters, Fig. 1. A) Picture of the two-stage LD-DF system used in the experiment. B) Schematic diagram of the experimental two-stage system. Feedstock tank (1), magnetic stirring plate (2 and 4), peristaltic pumps (2), effluent tank (5), LR (6), HR (7), pH probe (8), gas outlet (9), gas sampling port (10), water column (11), gas flow meter (12), alkali input (13), pH controller (14), NaOH 6 M solution tank (15). Table 1 Summary of the operating conditions tested during the two-stage LD-DF of FW. Parameter Lactate-producing reactor P1 P2 P3 P4 P5 Time (days) 0–42.0 42.0–57.3 57.3–75.5 75.5–87.5 87.5–95.0 HRT (h) 13.5 11.25 9 6.75 4.5 OLR a (g VS/L-d) 85.2 102.2 127.7 170.3 255.5 OLR a (g COD/L- d) 139.7 167.7 209.6 279.5 419.2 HRT cycles 74.7 32.4 53.9 33.8 52.8 Parameter H 2 -producing reactor P1 P2 P3 P4 P5 Time (days) 0–35.0 35.0–50.2 50.2–70.4 70.4–79.9 79.9–88.0 HRT (h) 12 10 8 6 4 OLR a (g VS/L-d) 95.8 115.8 143.7 191.6 285 OLR a (g COD/L- d) 152.2 182.6 228.3 304.4 456.6 HRT cycles 70 36.5 60.6 38 59.4 Parameter Two-stage LD-DF P1 P2 P3 P4 P5 Time (days) 0–42.0 42.0–57.3 57.3–75.5 75.5–87.5 87.5–95.0 HRT (h) 25.5 21.25 17 12.75 8.5 OLR a (g VS/L-d) 45.1 54.1 67.6 90.2 135.2 OLR a (g COD/L- d) 71.6 85.9 107.4 143.2 214.9 HRT cycles 32.9 17.0 28.5 17.9 22.9 a OLR is calculated based on the initial content of VS or COD present in the FW. L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 3 including alkali consumption (expressed in mL NaOH/g VS added and mL NaOH/L-d; equivalent to 1 M NaOH), VS and carbohydrate removal efficiencies, were also used for process characterization in both the LR and HR. Energy production rate (kJ/L-d) and energy production yield (kJ/g-VS added ) were calculated as reported elsewhere [24]. 2.5. Analytical methods Total COD, TS and VS concentrations were measured according to standard procedures for wastewater characterization [25]. The elemental analysis of the substrate (CHONSP) and its characterization in terms of carbohydrates, proteins and lipids were performed as previously reported by Regueira-Marcos et al. [20]. The gas composition (H 2 , CO 2 , N 2 , O 2 ) of the evolved off-gas was measured in an Agilent 8860 gas chromatograph (GC) (USA), which was equipped with a thermal conductivity detector (TCD) and two separate columns. The first column, a PoraBOND Q (25 m ×0.53 mm ×10 μ m), was responsible for the separation of CO 2 . The second column, a BR-Molsieve 5A (15 m ×0.53 mm ×50 μ m), was responsible for the separation of H 2 , N 2 and O 2 . Column switching was performed by means of air-operated valves at a timed point in the analysis. The injector, oven and detector were maintained at 150, 45 and 200 ◦C, respectively. Helium was used as the carrier gas at a column flow rate of 3.5 mL/min. The GC-TCD system was also configured for measuring CH 4 and hydrogen sulphide (H 2 S). Finally, organic acids were measured using a Shimadzu HPLC (model LC-2050C; Oregon, USA) with an integrated UV detector at 214 nm. The column configuration included a pre-column (HyperREZ XP H + guard cartridge, UK) followed by a HyperREZ XP Carbohydrate H + 8 chromatography column (UK) operated at 55 ◦C. Sulphuric acid (5 mM) was used as the eluent at a flow rate of 0.6 mL/min. The organic acids measured included lactate, formate, acetate, propionate, isobutyrate, butyrate, isovalerate and valerate. 2.6. Data treatment One-way ANOVA followed by either Tukey or Kruskal-Wallis test was used to test statistical differences (p-value >0.05) between the means of the process performance indicators obtained at different HRT. The Shaphiro-Wilk test was used to assess the normality distribution of the data (p-value <0.05). Principal component analysis (PCA) was also performed to infer possible relationships among process performance indicators (i.e., HPR, HPSI, HY, organic acids). Statistical analyses were performed using Statgraphics Centurion software (version 19.2.01). The biogas volume was normalized to standard temperature and pressure conditions (0 ◦C and 1 atm). HPSI values were determined according to Eq. 1[26], where HPSI is the H 2 production stability index, while HPR is the H 2 productivity (NL H 2 /L-d) recorded in each operational period. A HPSI value equal to 1 indicates a constant HPR, while a deviation value in HPR as large as the average HPR represents a stability index equal to 0. Steady states at each condition were considered when the HPSI remained above 80 % for at least three consecutive days. HPSI =1−Standad deviation HPR Average HPR (1) 3. Results and discussion 3.1. Impact of pH on the continuous acidogenic fermentation of FW The organic acid profile as a function of the different operational pHs applied during the continuous acidogenic fermentation of FW is shown Fig. 2. The objective was to identify the optimal pH that maximizes the conversion of fermentable organics into lactate, while minimizing the diversion of reducing power towards the production of H 2 or other byproducts. Operation without pH control resulted in a pH dropping as low as 3.7, leading to the lowest lactate concentration of 5.5 ±0.3 g/L. In contrast, the highest lactate titer was obtained at pH 4.5, with an average concentration of 11.4 ±0.7 g/L. The successive shift to pH 5.0 resulted in a decreasing trend in lactate concentration, with an average titer of 9.3 ±0.6 g/L. Interestingly, operational pH values of 4.5 and 5.0 prevented biogas production and required similar volumetric alkali consumption, averaging 153.0 ±47.3 and 172.9 ±49.0 mL NaOH/L-d, respectively. In contrast, the concentration of lactate in the fermentation broth at a fixed pH of 5.5 was 9.8 ±0.4 g/L. However, this pH also triggered the onset of biogas (mainly CO 2 ) with a volumetric production rate of 180.5 ±72.6 mL biogas/L-d. Thus, the lactate selectivity over the tested pH range from the lowest to highest was 64.4, 77.9, 69.7 and 52.4 %, respectively. As expected, the alkali consumption was significantly higher at pH 5.5, averaging 374.8 ±93.1 mL OH − equiv /L-d. Biogas production was accompanied by the accumulation of acetate, which reached an average of 5.0 g/L at pH 5.5, whereas acetate concentration at other pH values ranged from 0.7 to 1.1 g/L. Relatively low butyrate concentrations of around 0.5 g/L were also recorded at pH 5.5, confirming the occurrence of mixed acid fermentation [16]. Propionate levels remained between 1.6 and 2.0 g/L on average throughout the process. Interestingly, higher pH resulted in a higher degree of acidification, reaching total organic acid concentrations of 8.5 ±1.0, 14.6 ± 1.7, 13.3 ±1.8 and 18.8 ±3.4 g organic acids/L, corresponding to 9.9 ±1.4, 16.7 ±2.4, 15.3 ±2.4 and 21.5 ±4.6 g COD eq /L, at pH 3.7 (no pH control), 4.5, 5.0 and 5.5, respectively. The highest organic acid concentration could be explained by the less severe environmental conditions imposed at pH 5.5, which likely enhanced the microbial activity of acidogens. In this regard, the contribution of lactate to the total organic acid concentration from the lowest to the highest pH tested was 64.4, 77.9, 69.7 and 52.4 %. Notably, the pH 4.5 condition not only supported a higher concentration of lactate but also exhibited greater selectivity for lactate synthesis, making it the most favorable condition for lactate production among those tested. The production of biogas observed at a pH of 5.5 indicated the onset of oxidative decarboxylation pathways [16], which can explain the increase in acetate levels detected under this condition. In DF, acetate as Fig. 2. Time course of the organic acid profile in the cultivation broth during the lactate production pH test. L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 4 end-product could be directly related to the production and/or consumption of H 2 via acetic-type fermentation and homoacetogenesis, respectively [12]. Additionally, acetate is known to act as an electron acceptor during the degradation of lactate to H 2 , butyrate and CO 2 [12]. Therefore, acetate production in the first stage of the two-stage LD-DF process can be advantageous, particularly for the second stage, where it could support further H 2 production. On the other hand, the production of other metabolites, such as propionate or valerate, is undesirable because their synthesis requires more reducing power than lactate. Propionate is typically produced through the acrylate pathway by reducing lactate or via the propiogenesis pathway by reducing acetate with the concomitant consumption of CO 2 and H 2 [19,27]. In either case, propionate production hinders the final HY. Based on the observed results, pH 4.5 was selected as the best condition to maximize H 2 production in the second stage by promoting a high titer and selectivity of lactate in the first stage. The rapid acidification observed early in the process, with pH values below 4 within the first 24 h of fermentation, has been consistently observed in carbohydrate-rich FW, whereas lipid- or protein-rich substrate fermentations tend to acidify more slowly [28]. This is relevant because more neutral pHs hinder lactate accumulation by promoting its secondary fermentation to further fermented compounds such as acetate or butyrate [29]. Acetate and butyrate, with higher pKa values (4.8) compared to lactate (3.8), contribute to the alkalinization of the culture broth [28]. The rapid acidification also poses a challenge for lactate production, as it can inhibit lactate accumulation even before the pH reaches values close to its pKa. This is due to product inhibition and the inherent toxicity of undissociated lactate to LAB [30]. In this regard, Tang et al. [31] observed the highest lactate accumulation of 28.4 g/L at pH 5.0 in batch experiments treating real FW, regardless of the inoculum used. The authors also reported that a too-low pH of 4.0 resulted in a lower lactate titer, while a higher pH value of 6 reduced lactate production and selectivity by promoting the formation volatile fatty acids (VFA). Similarly, Daly et al. [32] found that maintaining the pH value below 4.5 helped prevent the production of acetate, butyrate and H 2 in experiments evaluating the storage of FW. Similar results were obtained by García-Depraect et al. [9], where pH values below 4.5 were achieved in <24 h during FW storage, with lactate accounting for 90 % of all the organic acids measured. Regarding continuous lactic acid production, De Groof et al. [33] observed that low HRTs and high OLRs are crucial for selective lactate production. In contrast, low OLR and high HRTs led to the production of secondary fermented products such as butyrate or even medium-chain carboxylic acids (MCCAs). This behaviour can be explained by the fact that low HRTs impose a shorter fermentation time and high specific growth rates, forcing the microbial community to prioritize the formation of early metabolic compounds like lactate. Additionally, high OLRs ensure that sufficient substrate is available to sustain lactate production, preventing the microbial community from resorting to alternative metabolic pathways that produce other by-products, thereby avoiding substrate depletion for lactate producers. Similarly, Wu et al. [34] observed that a pH of 4.0 was significantly more selective for lactate synthesis than a pH of 5.0, with higher total organic acid concentrations observer under the latter condition. In their study, lactate concentrations averaged between 10 and 20 g/L at both pH levels in a CSTR treating fruit-vegetable waste at a 3-day HRT. Wang et al. [35] observed a lactate concentration of 21.7 g/L in a mesophilic process treating simulated FW at 4 d HRT, with pH controlled at 4.25 by recirculating the digestate. Similarly, Feng et al. [36] found an optimal lactate production (13.5 g/L) at pH 4.2 in a mesophilic semi-continuous process operated at 4 d HRT. These authors also observed the occurrence of mixed acid fermentation pathway along with the enhancement of the hydrolysis and acidification rates at pH 6.0, which is in good agreement with the results of the present study. 3.2. Two-stage LD-DF process The stepwise reduction of HRT from 12 h (P1) down to 6 h (P4) resulted in an increase in the HPR from 4.2 ±0.5 to 9.6 ±0.9 L H 2 /L- d (Table 2 and Fig. 3B). However, further reducing the HRT to 4 h in P5 caused a decrease in HPR, reaching an average value of 4.9 ±1.6 L H 2 / L-d. Unlike P1–P4, the process during P5 failed to meet the stability criterion for H 2 production defined for the experiment, which was an 80 % variation in HPR for 3 consecutive days. During P5, a declining HPR was observed, with a HPSI of 66.6 % over the last 3 days of operation. Throughout the operation, no CH 4 or H 2 S was detected. The HY increased from 43.4 ±5.0 mL H 2 /g VS added in P1 to ~50.0 mL H 2 /g VS added in P4 (Table 2). The low HRT of 4 h during P5 also imposed a Table 2 Steady-state values of different operational parameters measured during the continuous two-stage LD-DF of FW. Operational stage P1 P2 P3 P4 P5 Operation period (days) 0–35 35.0–50.2 50.2–70.4 70.4–79.9 79.9–88.0 Stability period (days) 8.7 13.0 9.4 8.5 3.0 HPSI (%) 88.4 85.5 81.7 91.0 66.6 LR-BPR (L/L- d) 0.6 ± 0.2 1.3 ±0.2 0.7 ±0.2 0.6 ±0.2 0.2 ±0.2 LR-HPR (L H 2 / L-d) 0 0 0 0 0 HR-BPR (L/L- d) 11.0 ±0.9 15.3 ± 2.0 18.1 ±2.7 22.8 ± 2.3 12.0 ±3.4 HR-HPR (L H 2 / L-d) 4.2 ± 0.5 6.0 ±0.9 7.4 ±1.4 9.6 ±0.9 4.9 ±1.6 HR-HY (mL H 2 /g VS added ) 43.4 ±5.0 50.8 ± 7.4 50.9 ±9.3 49.3 ± 4.4 17.21 ± 5.75 HR-H 2 content (%) 38.5 ±3.4 39.2 ± 1.9 40.9 ±2.3 41.8 ± 1.0 41.1 ±2.2 H 2 energy recovery yield (kJ/g VS added ) 0.55 ±0.05 0.66 ± 0.1 0.66 ± 0.12 0.64 ± 0.06 0.22 ± 0.07 H 2 energy production rate (kJ/L-d) 53.9 ±6.3 76.3 ± 11.1 94.8 ± 17.4 122.3 ± 11.1 63.6 ± 21.3 LR-VS removal (%) 33.5 ±4.1 35.3 ± 5.1 30.1 ±7.2 35.4 ± 6.1 32.5 ±6.0 HR-VS removal (%) 15.9 ±3.6 13.7 ± 4.0 14.4 ±2.5 14.4 ± 4.7 19.1 ±0.1 Total VS removal (%) 49.4 ±0.7 48.9 ± 3.1 44.5 ±5.5 49.8 ± 1.4 51.6 ±5.9 LR-CH removal (%) 63.8 ±1.9 60.5 ± 9.3 50.3 ±7.5 60.9 ± 11.4 47.9 ±8.4 HR-CH removal (%) 14.0 ±3.7 12.4 ± 15.8 17.5 ± 11.1 15.8 ± 13.5 25.4 ± 13.4 Total CH removal (%) 77.8 ±2.5 72.9 ± 9.3 67.7 ±8.1 76.7 ± 6.5 73.3 ±5.0 LR-alkali (mL NaOH/L-d) 168.5 ±72.4 150.4 ± 67.8 224.7 ± 96.3 231.9 ± 67.7 252.2 ± 83.3 HR-alkali (mL NaOH/L-d) 56.6 ±37.2 112.5 ± 46.7 254.4 ± 105.5 239.3 ± 74.4 717.25 ± 125.0 Total alkali (mL NaOH/ L-d) 225.2 ±76.6 266.9 ± 70.8 478.7 ± 168.2 478.4 ± 98.8 985.9 ± 156.4 LR-alkali (mL NaOH/g VS added ) 0.59 ±0.39 0.97 ± 0.4 2.2 ±0.9 2.1 ±0.6 6.2 ±1.1 HR-alkali (mL NaOH/g VS added ) 1.76 ±0.76 1.3 ±0.6 1.6 ±0.7 1.2 ±0.4 0.9 ±0.3 Total alkali (mL NaOH/ g VS added ) 2.35 ±0.8 2.3 ±0.6 3.3 ±1.2 2.5 ±0.5 3.5 ±0.6 L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 5 significantly lower HY of 17.2 ±5.8 mL H 2 /g VS added . The H 2 content remained constant at ~40 % throughout the continuous operation, slightly increasing, on average, from 38.5 ±3.4 % in P1 to 41.8 ±1.0 % in P4. Similarly, the energy production rate increased from 53.9 ±6.3 kJ/L-d in P1 up to 122.3 ±11.1 kJ/L-d in P4, before collapsing to 63.6 ±21.3 kJ/L-d in P5. On the other side, the energy yield rose from 0.55 ±0.05 kJ/g VS added to a stable value of ~0.65 kJ/g VS added from P2 to P4, equally to HY, while pluming to 0.22 ±0.07 kJ/g VS added in P5. This energy yield is low compared to the typical energy yield (10–16 kJ/g VS added ) achieved in the methanization of FW. Therefore, one option to further increase the energy yield is to treat the final DF effluent rich in organic acids by anaerobic digestion. The sharp drop in H 2 productivity observed from P4 to P5 may be due to both a washout effect caused by the low HRT imposed, and system overload due to the high OLR during P5 (135.2 g VS/L-d and 214.9 g COD/L-d) compared to P4 (90.2 g VS/L-d and 143.2 g COD/L-d). In this context, most of the optimal OLR ranges reported in the literature for LD-DF are between 100 and 200 g COD/L-d [12]. Furthermore, although high HPRs have been reported in the literature between 1 and 12 h of HRT during DF, suspended biomass fermenters are rarely able to operate at HRTs <6 h [12]. Most of the HPR values that have been reported in continuous DF processes for FW are in the range of 0.2 to 1.4 L H 2 /L-d [15]. For instance, Paudel et al. [37] obtained an optimal HPR value of 1.35 L biogas/L-d (32.3 % H 2 content) at 8 h of HRT and 106 g VS/L-d in a CSTR at 37 ◦C treating FW. Nevertheless, Algapani et al. [38] reported a HPR of 3 L H 2 /L-d (60.8 % H 2 content) at 5 d HRT (OLR of 18 g VS/L-d) in a continuous two-stage process for H 2 and CH 4 production at 37 ◦C, testing different digestate recirculation rates. Regueira-Marcos et al. [21] reported a higher HPR of 4.2 L H 2 /L-d (53 % H 2 content) at 16 h HRT, corresponding to an OLR of 149.3 g COD/L-d (equivalent to 108 g VS/L-d) in a CSTR treating simulated FW at 37 ◦C. These authors observed process collapse at 12 h HRT and an OLR of 199 g COD/L- d (144 g VS/L-d), while showing the highest HY (48.5 ±13.9 mL H 2 /g VS added ) and energy yield (0.62 ±0.18 kJ/g VS added ) at 24 h HRT, similar to the optimal results of this study. Martínez-Mendoza et al. [24] were able to operate a stable and highly efficient process (11.8 L H 2 /L-d; 65.1 % H 2 content) with a higher OLR of 188.1 g VS/L-d (6 h HRT) in a CSTR running on fruit and vegetable waste via LD-DF. Such a high OLR was probably better tolerated due to a higher monosaccharide-to- polysaccharide ratio than that of typical household FW. Interestingly, García-Depraect et al. [26] tested a similar two-stage LD-DF process for H 2 production from tequila vinasse, achieving an optimal HPR of 12.3 L H 2 /L-d at 6 h HRT and a corresponding OLR of 169 g COD/L-d (104 g VS/L-d). However, the two-stage process collapsed at 4 h HRT with a corresponding OLR of 253 g COD/L-d (160 g VS/L-d). Contrary to the observations during the stand-alone continuous acidogenic fermentation of FW at different operational pHs (Section 3.1), the LR generated biogas during the whole operation, even at pH 4.5 (Fig. 3A). This produced off-gas consisted only of CO 2 . P1, P3 and P4 showed similar BPRs of 0.6–0.7 L biogas/L-d, while the BPR recorded in P2 was 1.3 ±0.2 L biogas/L-d (Table 2). Biogas production decreased significantly in P5 (0.2 ±0.2 L biogas/L-d), probably due to the too short HRT imposed during this condition. This discrepancy in biogas production between the two sets of experiments may be due to the fact that during the pH screening, the reactor was kept at a pH below 4 for 8 days before being adjusted to 4.5. This highly acidic pH value could have selected a more specific lactate producing microbial community and prevented biogas production at pH 4.5 (at least for the time period studied). Several authors have described the high specificity for the lactic acid pathway at pH values close to 4.0, while pH values closer to 5.0 typically promote the production of acetate and other downstream products, with the subsequent formation of CO 2 [28] [31]. Since biogas production is not desired in this first stage of the process, an initial operation at pH 4 or lower may be desirable to prevent or at least delay off-gas formation. The removal of VS (Fig. 4A) and of total carbohydrates (Fig. 4C) remained constant throughout the operation, even at P5 where a lower H 2 production performance was observed, indicating that these parameters are not a good proxy to efficiently predict LD-DF performance, as demonstrated in previous studies [9] [21] [23]. Most of the removal of VS (68.3 % of the total removal) and carbohydrates (76.9 % of the total removal) occurred in the LR (Table 2). In this first stage, most of the carbohydrates can be solubilized and subsequently consumed mainly by LAB for lactate and biomass production. In addition, the sharp reduction in the HPR recorded in P5 implied a slight increase in the fraction of carbohydrates and VS removed in the HR, but not enough to surpass the removal efficiencies achieved in the LR. Regueira-Marcos et al. [21] reported a similar removal of VS (~50 %) and a higher removal of carbohydrates (~90 %) in a one-stage continuous fermenter using the same type of FW. Despite the lower carbohydrate consumption observed here, the observed HPRs were remarkably higher, highlighting the outstanding performance of the two-stage LD-DF process and revealing Fig. 3. Time course of the organic acid profile during the LD-DF process in A) the lactate-producing reactor (LR) and B) the hydrogen-producing reactor (HR). L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 6 that lactate, rather than the carbohydrates, was the main H 2 precursor. The total consumption of alkali required to maintain a constant operational pH in the first (LR) and second (HR) reactor can be related with the efficiency of production and consumption of lactate, which has a low pKa of 3.8 [30]. In this particular study, the total alkali consumption increased as the HRT was shortened, starting at 225.2 ±76.6 mL OH − equiv /L-d at 12 h HRT and peaking at 985.9 ±156.4 mL OH − equiv / L-d at 4 h HRT (Table 2; Fig. 4B). Indeed, a larger substrate supply implies a more active formation of organic acids in the system. During P1, the LR was responsible for the 74.8 % of the alkali consumed, while in P4 both reactors consumed similar amounts of alkali. On the other hand, the HR reactor capitalized the alkali consumption in P5, with the 72.8 % of the total alkali consumed. In this sense, the LR underwent the greatest acidification degree because it was designed to convert fermentable carbohydrates to lactate. The metabolization of lactate to other organic acids such as acetate or butyrate involves an alkalinization process due to the higher pKa of these compounds (about 4.8) [16]. However, the operating pH in the HR was significantly higher (6.5) than these pKas, and thus, alkali addition was required in the HR. In P1, the higher HRT allowed for a greater lactate accumulation in the LR, while allowing more time in the HR for this lactate to be metabolized into more fermented organic acids, allowing for further alkalinization of the culture broth without the addition of as much external alkali. Therefore, the alkali demand was higher in the LR at high HRTs. By reducing the HRT, lactate accumulation in the LR was reduced, while the higher process rate reduced the buffering capacity of the HR, balancing both consumptions during P4. Finally, the process collapsed in P5 as lactate started to be produced in the HR, accelerating the alkali consumption in this reactor (and in the whole process) while drastically reducing the internal alkalinization as VFA levels (mainly butyrate) decreased. Lactate was the main fermentation product in the LR throughout the operation of the two-stage LD-DF process (Fig. 3A; Table 3). The total organic acid concentrations in P1, P2, P3, P4 and P5 were 12.6 ±1.6, 11.8 ±1.9, 13.4 ±2.2, 13.4 ±2.9 and 9.4 ±1.6 g/L, respectively, with lactate accounting for ~83.6, 79.9, 67.0, 64.6 and 73.0 % of the total organic acids, respectively. The lactate concentration peaked at P1 (10.6 ±1.1 g/L), which was similar to the pH test (see Section 3.1). The stepwise decrease in HRT implied a continuous decrease in lactate concentration of 35.8 % from P1 to P5. On the contrary, acetate and butyrate levels increased from 1.1 ±0.4 and 0.1 ±0.1 g/L in P1 to 2.7 ±0.6 and 0.8 ±0.5 g/L in P4, while plummeting to 1.2 ±0.1 and 0.5 ± 0.2 g/L, respectively, in P5. It can be hypothesized that the reduction in HRT led to an increase in the oxidative decarboxylation pathway until P4, after which the entire fermentative process began to collapse, similar to what was observed with HPR in the HR. The similar evolution of BPR and acetate levels in the LR seems to confirm this phenomenon (Fig. 3A). Other organic acids, such as formate, propionate or isobutyrate, showed a similar evolution as acetate and butyrate, but to a lesser extent (Fig. 3A, Table 3). Other longer-chain organic acids were absent in the LR. An alternative hypothesis is that the enhancement of the oxidative decarboxylation process in the system was caused by the slow and gradual acclimation of the microbiota to the acidic pH conditions in the LR. This acclimation probably led to a continuous increase in the acetate concentration in the culture broth until P5, at which point the HRT became too low to sustain the process. Similar to the reduction in MCCAs production [16], a decrease in HRTs is also expected to reduce the Fig. 4. Time course of A) volatile solids removal (VS removal ; %), B) alkali consumption (mL/L-d) and C) carbohydrate removal (CH removal ; %) in the lactate-producing reactor (LR), hydrogen-producing reactor (HR), and in the overall LD-DF process. Table 3 Steady-state organic acid concentrations measured in the lactate-producing reactor (LR) and in the hydrogen-producing reactor (HR). Reactor Organic acid Operational stage P1 P2 P3 P4 P5 LR Lactate (g/L) 10.6 ±1.1 9.2 ± 1.3 9.0 ±0.9 8.7 ±1.5 6.8 ±1.1 Formate (g/ L) 0.3 ± 0.0 0.3 ± 0.0 0.5 ±0.0 0.5 ±0.5 0.3 ±0.0 Acetate (g/L) 1.1 ± 0.4 1.4 ± 0.4 2.6 ±0.5 2.7 ±0.6 1.2 ±0.1 Propionate (g/L) 0.2 ± 0.0 0.2 ±0 0.2 ±0.0 0.4 ±0.3 0.2 ±0.2 Isobutyrate (g/L) 0 0.3 ± 0.0 0.3 ±0.0 0.3 ±0.0 0.3 ±0.0 Butyrate (g/ L) 0.1 ± 0.1 0.4 ± 0.1 0.8 ±0.7 0.8 ±0.5 0.5 ±0.2 HR Lactate (g/L) 0 0 0.3 ±0.5 0.0 ±0.1 4.2 ±2.4 Formate(g/ L) 0.3 ± 0.0 0.3 ± 0.1 0.7 ±0.2 0.8 ±0.2 0.9 ±0.1 Acetate (g/L) 2.6 ± 0.3 3.0 ± 0.7 5.3 ±1.2 4.1 ±0.7 5.7 ±0.8 Propionate (g/L) 4.6 ± 0.2 4.0 ± 0.4 2.7 ±0.9 1.4 ±0.3 1.3 ±0.2 Isobutyrate (g/L) 0.3 ± 0.1 0.4 ± 0.1 0.4 ±0.1 0.3 ±0.1 0.3 ±0.0 Butyrate (g/ L) 5.2 ± 0.4 5.7 ± 0.5 6.9 ±1.4 7.5 ±0.8 3.2 ±1.3 Isovalerate (g/L) 0.1 ± 0.1 0.2 ± 0.1 0.2 ±0.1 0.5 ±0.3 0.1 ±0.1 L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 7 conversion of lactate to acetate [33], as this is a downstream metabolic process that is limited under time-constrained fermentation conditions. This phenomenon was previously described by De Groof et al. [33], who showed that low HRTs and high OLRs can lead to increased lactate accumulation. The organic acids recorded in the HR exhibited a completely different pattern compared to those of the LR (Fig. 3B). Indeed, lactate was entirely depleted in the second reactor, except in P5, where the lactate concentration in the fermentation broth reached 5.7 g/L. In the HR, butyrate was the predominant VFA from P1 to P4, followed by acetate and propionate. Propionate levels were comparable to butyrate levels in P1 (≈5.0 g/L) and were slowly reduced to 1.1 g/L by the end of P5. Acetate levels were low in P1, increasing to 5.3 ±1.2 in P3 and further to 5.7 ±0.8 in P5. This trend in propionate and acetate suggests that the reduction in HRT directly decreased propiogenesis from acetate and lactate [19] [39], a phenomenon already described by Cabrol et al. [40]. Other organic acids such as formate, isobutyrate and isovalerate remained at concentrations below 1 g/L throughout the entire operation. Valerate concentrations were also negligible regardless of the type of reactor, except for a notable peak between days 22–25 (P5), where levels exceeded 2 g/L. This peak was accompanied by an increase in acetate and propionate concentrations and a decrease in butyrate levels. During this particular period, acetate elongation to butyrate was partially replaced by propionate elongation to valerate [16]. Unlike the LR, no longer-chain organic acids were produced in the HR. Butyrate production appeared to be directly correlated with HPR in the HR, as both increased with the reduction of HRT until the process collapsed in P5. The reduction in butyrate levels in P5 was directly correlated with the rise in acetate levels. The complete depletion of lactate during the stages of efficient H 2 production (P1–P4) confirmed that its consumption was directly related to the production of H 2 , thus supporting the occurrence of the LD-DF process. It is important to note that the amount of carbohydrates reaching the HR was <50 % of the total supplied from P1 to P4 (and slightly higher in P5), and their subsequent removal in the HR represented barely 10–15 % of the total carbohydrate removal (Table 2), despite the fact that lactate was completely removed. This seems to indicate a preference of the culture broth microbiota for the consumption of lactate over carbohydrates, as observed by Fuentes-Santiago et al. [41]. The PCA analysis shown in Fig. 5 shows that HPR was directly correlated with HY and butyrate concentration in the HR, as well as with the total removal of VS and carbohydrates. Indeed, a higher HY, along with grater VS and carbohydrate removal, implies a higher and more efficient substrate utilization. Moreover, butyrate production was directly related to H 2 production [19]. On the other hand, lactate and acetate levels in the HR, along with total alkali consumption, were negatively correlated with HPR in the HR. The behaviour observed in P5, as well as on days 48 (P2) and 57–63 (P3) supported this relationship, with HPR decreasing as butyrate concentration decreased and acetate and lactate concentrations increased. It should be noted that from a stoichiometric point of view, HY is higher when the final fermentation product is acetate (4 mol H 2 /mol glucose) instead of butyrate (2 mol H 2 /mol glucose) [40]. However, a high accumulation of acetate in the culture broth could lead to by-product inhibition, which could limit the amount of substrate metabolized [16,30]. Conversely, the formation of butyrate from acetate would allow for a higher total accumulation of organic acids in the culture broth, indicating a higher carbohydrate metabolization towards H 2 -producing pathways. In turn, each mole of butyrate consumes 2 mol of acetate for its formation [16], thus reducing the moles of acid in the system. This allows a greater formation of acetate from lactate and reduces the alkali consumption of the process. Neither the propionate levels in the HR nor the lactate levels in the LR seemed to affect the HPR in the HR, despite the fact that propiogenesis processes are H 2 -consuming pathways [19,39,42]. It can be hypothesized that the lower HRT during P1 compensated the H 2 consumption due to propiogenesis with a high butyrate production rate, thus increasing the H 2 production and balancing the total HY. 4. Conclusions The performance of a two-stage LD-DF system, designed with physical separation of lactate production and DF for H 2 production in two separate CSTRs, was investigated using FW as the substrate. The results showed that a controlled pH of 4.5 in the LR allowed the highest lactate concentration of 11.4 ±0.7 g/L, accounting for 79.9 % of the total organic acids produced. Lower lactate production was observed at more acidic pH levels, while more alkaline pH conditions led to greater diversification in the organic acids profile. On the other hand, the HRT had a significant impact on the HPR of the two-stage system. H 2 productivity increased steadily and proportionally as the HRT was reduced from 12 to 6 h, reaching a maximum productivity of 9.6 ±0.9 L H 2 /L- d with an associated HY of 50 mL H 2 /g VS added at 6 h HRT. The subsequent reduction in HRT to 4 h resulted in process collapse. Lactate was completely consumed in the HR with a concomitant accumulation of butyrate during the process operation at HRTs of 12–6 h. However, when the HRT was reduced to 4 h, lactate began to accumulate in the DF broth, and higher levels of acetate were observed over butyrate. The LR played a significant role in removing most of the VS and carbohydrates throughout the process. Overall, the implementation of a two-stage LDDF system allowed achieving high H 2 productivities from FW, making this system an efficient alternative to cope with LAB overproliferation in DF systems. CRediT authorship contribution statement Lois Regueira-Marcos: Writing – original draft, Methodology, Investigation, Formal analysis. Octavio García-Depraect: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis, Conceptualization. Raúl Mu˜ noz: Writing – review & editing, Supervision, Resources, Project administration, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fig. 5. PCA analysis of different process performance indicators evaluated during the two-stage LD-DF process for FW. Information in parentheses indicates to which type of reactor the parameter refers or if the parameter refers to the whole process. LR: lactate-producing reactor; HR: H 2 -producing reactor; Total: both LR and HR; HPR: hydrogen production rate; HY: hydrogen yield; Lac: lactate concentration; Acet: acetate concentration; Prop: propionate concentration; But: butyrate concentration; Alkali: amount of alkali consumed to maintain pH control; CHR: carbohydrate removal; VS: volatile solids removal. L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 8 Data availability Data will be made available on request. Acknowledgments This work was supported by Grant RYC2021-034559-I funded by MCIN/AEI/10.13039/501100011033 and by European Union NextGenerationEU/PRTR. The Grant PID2022-139110OA-I00, funded by MCIN/AEI/10.13039/501100011033 and by ERDF A way of making Europe and by the European Union, is acknowledged. The support from the Regional Government of Castilla y Le´ on and the EU-FEDER (CL-EI- 2021-07 and UIC 315) is gratefully recognized. Lois Regueira thanks the Consejeria de Educaci´ on de Castilla y Le´ on for his PhD Contract. References [1] S. Venkata Mohan, S. Dahiya, K. Amulya, R. Katakojwala, T.K. Vanitha, Can circular bioeconomy be fueled by waste biorefineries — a closer look, Bioresour. Technol. Rep. 7 (2019), https://doi.org/10.1016/j.biteb.2019.100277. [2] K.M. Kibler, D. Reinhart, C. Hawkins, A.M. Motlagh, J. Wright, Food waste and the food-energy-water nexus: a review of food waste management alternatives, Waste Manag. 74 (2018) 52–62, https://doi.org/10.1016/j.wasman.2018.01.014. [3] A. Zabaniotou, P. Kamaterou, Food waste valorization advocating circular bioeconomy — a critical review of potentialities and perspectives of spent coffee grounds biorefinery, J. Clean. Prod. 211 (2019) 1553–1566, https://doi.org/ 10.1016/j.jclepro.2018.11.230. [4] F. Battista, N. Frison, P. Pavan, C. Cavinato, M. Gottardo, F. Fatone, A.L. Eusebi, M. Majone, M. Zeppilli, F. Valentino, D. Fino, T. Tommasi, D. Bolzonella, Food wastes and sewage sludge as feedstock for an urban biorefinery producing biofuels and added-value bioproducts, J. Chem. Technol. Biotechnol. 95 (2020) 328–338, https://doi.org/10.1002/jctb.6096. [5] S. Shanmugam, T. Mathimani, K. Rajendran, M. Sekar, E.R. Rene, N.T.L. Chi, H. H. Ngo, A. Pugazhendhi, Perspective on the strategies and challenges in hydrogen production from food and food processing wastes, Fuel 338 (2023) 127376, https://doi.org/10.1016/J.FUEL.2022.127376. [6] Food and Agricultural Organization, Global initiative on food loss and waste initiative on food loss and waste. https://www.fao.org/3/i7657e/i7657e.pdf. [7] D. Fattibene, F. Recanati, K. Dembska, M. Antonelli, Urban food waste: a framework to analyse policies and initiatives, Resources 9 (2020), https://doi.org/ 10.3390/RESOURCES9090099. [8] Eurostat, Food waste: 127 kg per inhabitant in the EU in 2020. https://ec.europa. eu/eurostat/web/products-eurostat-news/-/ddn-20220925-2#:~:text=In%20202 0%2C%20the%20first%20year,in%20the%20food%20supply%20chain, 2022. [9] O. García-Depraect, I. Mirzazada, L.J. Martínez-Mendoza, L. Regueira-Marcos, R. Mu˜ noz, Biotic and abiotic insights into the storage of food waste and its effect on biohydrogen and methane production potential, J. Water Process Eng. 53 (2023) 103840, https://doi.org/10.1016/J.JWPE.2023.103840. [10] European Commission, Directorate-General for Energy, Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions a hydrogen strategy for a climate-neutral Europe. https://www.eu2018.at/calendar-events/political-ev ents/BMNT, 2020. [11] F. Dawood, M. Anda, G.M. Shafiullah, Hydrogen production for energy: an overview, Int. J. Hydrog. Energy 45 (2020) 3847–3869, https://doi.org/10.1016/j. ijhydene.2019.12.059. [12] O. García-Depraect, R. Castro-Mu˜ noz, R. Mu˜ noz, E.R. Rene, E. Le´ on-Becerril, I. Valdez-Vazquez, G. Kumar, L.C. Reyes-Alvarado, L.J. Martínez-Mendoza, J. Carrillo-Reyes, G. Buitr´ on, A review on the factors influencing biohydrogen production from lactate: the key to unlocking enhanced dark fermentative processes, Bioresour. Technol. 324 (2021), https://doi.org/10.1016/j. biortech.2020.124595. [13] D. Cheng, H.H. Ngo, W. Guo, S.W. Chang, D.D. Nguyen, L. Deng, Z. Chen, Y. Ye, X. T. Bui, N.B. Hoang, Advanced strategies for enhancing dark fermentative biohydrogen production from biowaste towards sustainable environment, Bioresour. Technol. 351 (2022) 127045, https://doi.org/10.1016/J. BIORTECH.2022.127045. [14] G. Mohanakrishna, N.P. Sneha, S.M. Rafi, O. Sarkar, Dark fermentative hydrogen production: potential of food waste as future energy needs, Sci. Total Environ. 888 (2023) 163801, https://doi.org/10.1016/J.SCITOTENV.2023.163801. [15] E. Villanueva-Galindo, M. Vital-J´ acome, I. Moreno-Andrade, Dark fermentation for H 2 production from food waste and novel strategies for its enhancement, Int. J. Hydrog. Energy 48 (2023) 9957–9970, https://doi.org/10.1016/J. IJHYDENE.2022.11.339. [16] Q. Wu, X. Bao, W. Guo, B. Wang, Y. Li, H. Luo, H. Wang, N. Ren, Medium chain carboxylic acids production from waste biomass: current advances and perspectives, Biotechnol. Adv. 37 (2019) 599–615, https://doi.org/10.1016/j. biotechadv.2019.03.003. [17] M. Canto-Robertos, C. Quintal-Franco, C. Ponce-Caballero, M. Vega-De Lille, I. Moreno-Andrade, Inhibition of hydrogen production by endogenous microorganisms from food waste, Braz. J. Chem. Eng. 40 (2023) 137–150, https:// doi.org/10.1007/s43153-022-00235-5. [18] L. De Vuyst, F. Leroy, Cross-feeding between bifidobacteria and butyrate-producing colon bacteria explains bifdobacterial competitiveness, butyrate production, and gas production, Int. J. Food Microbiol. 149 (2011) 73–80, https://doi.org/ 10.1016/J.IJFOODMICRO.2011.03.003. [19] P. Louis, S.H. Duncan, P.O. Sheridan, A.W. Walker, H.J. Flint, Microbial lactate utilisation and the stability of the gut microbiome, Gut Microbiome 3 (2022), https://doi.org/10.1017/gmb.2022.3. [20] L. Regueira-Marcos, O. García-Depraect, R. Mu˜ noz, Elucidating the role of pH and total solids content in the co-production of biohydrogen and carboxylic acids from food waste via lactate-driven dark fermentation, Fuel 338 (2023) 127238, https:// doi.org/10.1016/J.FUEL.2022.127238. [21] L. Regueira-Marcos, R. Mu˜ noz, O. García-Depraect, Continuous lactate-driven dark fermentation of restaurant food waste: process characterization and new insights on transient feast/famine perturbations, Bioresour. Technol. 385 (2023) 129385, https://doi.org/10.1016/J.BIORTECH.2023.129385. [22] O. García-Depraect, E.R. Rene, J. G´ omez-Romero, A. L´ opez-L´ opez, E. Le´ on-Becerril, Enhanced biohydrogen production from the dark co-fermentation of tequila vinasse and nixtamalization wastewater: novel insights into ecological regulation by pH, Fuel 253 (2019) 159–166, https://doi.org/10.1016/j.fuel.2019.04.147. [23] L.J. Martínez-Mendoza, R. Lebrero, R. Mu˜ noz, O. García-Depraect, Influence of key operational parameters on biohydrogen production from fruit and vegetable waste via lactate-driven dark fermentation, Bioresour. Technol. 364 (2022) 128070, https://doi.org/10.1016/J.BIORTECH.2022.128070. [24] L.J. Martínez-Mendoza, O. García-Depraect, R. Mu˜ noz, Unlocking the high-rate continuous performance of fermentative hydrogen bioproduction from fruit and vegetable residues by modulating hydraulic retention time, Bioresour. Technol. 373 (2023) 128716, https://doi.org/10.1016/J.BIORTECH.2023.128716. [25] C. Alc´ antara, C. Fern´ andez, P.A. García-Encina, R. Mu˜ noz, Mixotrophic metabolism of Chlorella sorokiniana and algal-bacterial consortia under extended dark-light periods and nutrient starvation, Appl. Microbiol. Biotechnol. 99 (2015) 2393–2404, https://doi.org/10.1007/s00253-014-6125-5. [26] O. García-Depraect, R. Mu˜ noz, J.B. van Lier, E.R. Rene, V.F. Diaz-Cruces, E. Le´ on- Becerril, Three-stage process for tequila vinasse valorization through sequential lactate, biohydrogen and methane production, Bioresour. Technol. 307 (2020), https://doi.org/10.1016/j.biortech.2020.123160. [27] W.A. Cavalcante, R.C. Leit˜ ao, T.A. Gehring, L.T. Angenent, S.T. Santaella, Anaerobic fermentation for n-caproic acid production: a review, Process Biochem. 54 (2017) 106–119, https://doi.org/10.1016/j.procbio.2016.12.024. [28] Q. Wang, H. Li, K. Feng, J. Liu, Oriented fermentation of food waste towards highvalue products: a review, Energies (Basel) 13 (2020), https://doi.org/10.3390/ en13215638. [29] J. Tang, X. Wang, Y. Hu, Y. Zhang, Y. Li, Lactic acid fermentation from food waste with indigenous microbiota: effects of pH, temperature and high OLR, Waste Manag. 52 (2016) 278–285, https://doi.org/10.1016/j.wasman.2016.03.034. [30] L. Song, D. Yang, R. Liu, S. Liu, L. Dai, X. Dai, Microbial production of lactic acid from food waste: latest advances, limits, and perspectives, Bioresour. Technol. 345 (2022) 126052, https://doi.org/10.1016/J.BIORTECH.2021.126052. [31] J. Tang, X.C. Wang, Y. Hu, Y. Zhang, Y. Li, Effect of pH on lactic acid production from acidogenic fermentation of food waste with different types of inocula, Bioresour. Technol. 224 (2017) 544–552, https://doi.org/10.1016/j. biortech.2016.11.111. [32] S.E. Daly, J.G. Usack, L.A. Harroff, J.G. Booth, M.P. Keleman, L.T. Angenent, Systematic analysis of factors that affect food-waste storage: toward maximizing lactate accumulation for resource recovery, ACS Sustain. Chem. Eng. 8 (2020) 13934–13944, https://doi.org/10.1021/acssuschemeng.0c03161. [33] V. De Groof, M. Coma, T. Arnot, D.J. Leak, A.B. Lanham, Selecting fermentation products for food waste valorisation with HRT and OLR as the key operational parameters, Waste Manag. 127 (2021) 80–89, https://doi.org/10.1016/J. WASMAN.2021.04.023. [34] Y. Wu, H. Ma, M. Zheng, K. Wang, Lactic acid production from acidogenic fermentation of fruit and vegetable wastes, Bioresour. Technol. 191 (2015) 53–58, https://doi.org/10.1016/J.BIORTECH.2015.04.100. [35] Q. Wang, L. Yang, K. Feng, H. Li, Z. Deng, J. Liu, Promote lactic acid production from food waste fermentation using biogas slurry recirculation, Bioresour. Technol. 337 (2021) 125393, https://doi.org/10.1016/J.BIORTECH.2021.125393. [36] K. Feng, H. Li, C. Zheng, Shifting product spectrum by pH adjustment during longterm continuous anaerobic fermentation of food waste, Bioresour. Technol. 270 (2018) 180–188, https://doi.org/10.1016/J.BIORTECH.2018.09.035. [37] S. Paudel, Y. Kang, Y.S. Yoo, G.T. Seo, Effect of volumetric organic loading rate (OLR) on H 2 and CH 4 production by two-stage anaerobic co-digestion of food waste and brown water, Waste Manag. 61 (2017) 484–493, https://doi.org/10.1016/j. wasman.2016.12.013. [38] D.E. Algapani, W. Qiao, M. Ricci, D. Bianchi, S.M. Wandera, F. Adani, R. Dong, Biohydrogen and bio-methane production from food waste in a two-stage anaerobic digestion process with digestate recirculation, Renew. Energy 130 (2019) 1108–1115, https://doi.org/10.1016/J.RENENE.2018.08.079. [39] G. Luo, D. Karakashev, L. Xie, Q. Zhou, I. Angelidaki, Long-term effect of inoculum pretreatment on fermentative hydrogen production by repeated batch cultivations: homoacetogenesis and methanogenesis as competitors to hydrogen production, Biotechnol. Bioeng. 108 (2011) 1816–1827, https://doi.org/10.1002/bit.23122. [40] L. Cabrol, A. Marone, E. Tapia-Venegas, J.P. Steyer, G. Ruiz-Filippi, E. Trably, Microbial ecology of fermentative hydrogen producing bioprocesses: useful L. Regueira-Marcos et al. Journal of Water Process Engineering 67 (2024) 106116 9