Photofermentative Hydrogen Production from Real Dark Fermentation Effluents: A Sequential Valorization of Orange Peel Waste

Abstract

This study explores the sequential valorization of orange peel waste (OPW) through photo-fermentation using real dark fermentation effluents (DFE) as substrates for hydrogen production using Rhodobacter capsulatus B10. Three DFE types—differing in prior biocompound extraction method—and their concentrations at three levels (25, 35, and 45%) were evaluated. The highest hydrogen yield (126.5 mL H₂ g⁻¹ VFA) was achieved with DFE derived from essential oil-extracted OPW at a concentration of 25%. The highest DFE concentration reduced the hydrogen yield due to intensified medium opacity and potential substrate inhibition. Kinetic modeling revealed that the Modified Gompertz and T_i-Gompertz models best described hydrogen production dynamics. This study presents the first evidence of hydrogen production via photo-fermentation using real effluents derived from OPW processing, demonstrating a novel route for citrus waste reuse within a biorefinery framework. These findings underscore the innovation and relevance of integrating waste valorization with clean energy production, while also identifying key operational challenges to be addressed.

1. Introduction

The global transition toward sustainable energy systems has intensified the interest in hydrogen as a clean and renewable energy carrier. Unlike conventional fossil fuels, hydrogen combustion produces only water as a theoretical byproduct, making it an attractive alternative for mitigating direct greenhouse gas (GHG) emissions and reducing dependence on non-renewable energy sources [1]. Global hydrogen production has hovered around 92 Mt y⁻¹ between 2019 and 2021, with clean hydrogen (green and blue) constituting less than 0.1% of the total—a figure projected to rise to 3% by 2030 and 12% by 2050 under net-zero scenarios [2]. Conventional hydrogen production relies on fossil-based methods like steam methane reforming, accounting for 94% of global hydrogen production. However, this production method is highly energy-intensive and emits between 11.2 and 11.6 kg CO₂ eq kg⁻¹ of hydrogen produced [3,4], making it a major contributor to greenhouse gas emissions. This underscores the urgent need for clean hydrogen production technologies that rely on renewable feedstocks and require low energy inputs.

Among the biological hydrogen production processes, photo-fermentation has emerged as a promising and efficient route for generating hydrogen. This technology relies on purple non-sulfur bacteria (PNSB), such as Rhodobacter capsulatus, which utilize light energy to metabolize organic substrates and produce hydrogen under anaerobic conditions. Particularly, R. capsulatus is one of the most metabolically adaptable and widely studied PNSB species [5]. The process of photo-fermentation metabolism can be summarized as follows: it is primarily driven by nitrogenase, a highly oxygen-sensitive enzyme that catalyzes the reduction of protons to produce hydrogen. The process begins with the assimilation of organic acids, such as acetate, butyrate, and lactate, which serve as electron donors [6]. These substrates are metabolized through the tricarboxylic acid cycle, generating reducing equivalents (NADH and FADH₂) that are transferred to the photosynthetic electron transport chain. Light photons excite bacteriochlorophyll molecules, creating an electrochemical proton gradient that facilitates ATP synthesis and reverse electron flow, ultimately enabling nitrogenase to reduce protons to produce molecular hydrogen [7]. Under nitrogen-limited conditions, hydrogen evolution serves as an electron sink, maintaining redox balance and sustaining microbial growth [8].

Another biological strategy, dark fermentation, efficiently breaks down complex organic matter into hydrogen under anaerobic conditions. However, it is inherently limited by metabolic constraints, leading to incomplete substrate utilization and the accumulation of residual volatile fatty acids (VFA) [9]. In contrast, photo-fermentation enables the further conversion of these organic acids into hydrogen, maximizing biomass valorization and improving overall energy recovery. By integrating both processes into a sequential dark-photo fermentation system, it is possible to enhance resource efficiency, minimize organic waste accumulation, and optimize the conversion of biomass into hydrogen [10,11,12].

Most photofermentative hydrogen production studies have relied on synthetic substrates under well-defined conditions that reduce variability and enhance microbial performance and hydrogen yield [13,14]. This approach offers experimental control but limits our understanding of photofermentative behavior under realistic process conditions. In contrast, other studies have investigated the use of real dark fermentation effluents (DFE) derived from complex organic wastes by coupling dark and photo-fermentation. For instance, Sagnak et al. [15] employed DFE obtained from wheat bran hydrolysates and reported hydrogen yields that averaged approximately 1200 mL H₂ g⁻¹ VFA. Similarly, Hitit et al. [16] used DFE from potato waste, achieving 162 mL H₂ g⁻¹ VFA, while Yokoi et al. [17] reported yields of 560 mL H₂ g⁻¹ VFA using sweet potato-based DFE. These results highlight the potential of DFE as a substrate for photo-fermentation, but also underscore the significant variability in hydrogen yields. These differences are largely attributed to the type of original substrate, which influences the composition of the resulting DFE. This variability presents a key technical challenge that must be addressed to optimize hydrogen recovery in systems using real waste-derived effluents.

Some challenges in applying photo-fermentation to real organic waste streams include medium coloration and the accumulation of inhibitory compounds such as ammonium ions, furfural, and phenolic compounds [11]. In this context, DFE dilution has been proposed to attenuate these issues. DFE is generally characterized by a high content of VFA—primarily consisting of acetate, butyrate, iso-butyrate, propionate, valerate, and isovalerate [18], which play crucial roles as electron donors and carbon sources in subsequent photo-fermentation stages [19]. However, when used without dilution, the accumulation of pigments, suspended solids, and potential inhibitors may limit light penetration and microbial activity.

Exploiting the high potential of photo-fermentation as both a clean energy generation process and a waste valorization strategy, this study investigates the use of real DFE for hydrogen production, building upon previous research by our group [20] using orange peel waste (OPW), a highly abundant agro-industrial byproduct generated from orange juice and essential oil production. In this previous study, OPW was subjected to different processing routes for biocompound extraction prior to hydrogen and VFA-rich effluent production via dark fermentation. OPW is an important waste in countries such as Mexico, Brazil, and China, the main producers of oranges. As it constitutes approximately 50–60% of the total fruit mass after industrial processing, substantial volumes of waste must be disposed of [21,22]. In many developing countries, there are no particular waste management strategies for OPW, thus relying on landfilling or other unsustainable practices, which contribute to GHG emissions through uncontrolled decomposition and risk environmental contamination via leachates [23].

This study presents, for the first time, the use of DFE from OPW as a substrate for photofermentative hydrogen production, evaluating how effluent concentration and prior substrate processing influence microbial growth and hydrogen yield. These findings reinforce the role of bio-based hydrogen as a sustainable alternative energy source for the transition toward renewable and waste-to-energy technologies.

2. Materials and Methods

2.1. Bacterial Strain and Culture Media

PNSB, Rhodobacter capsulatus strain B10, was supplied by the Laboratoire d’Electrochimie et Physicochimie des Matériaux et des Interfaces (LEPMI), Grenoble INP, Grenoble, France. R. capsulatus B10 was selected owing to its accessibility within our laboratory, demonstrated efficacy in hydrogen production, and its resilient, adaptable metabolic activity under photofermentative conditions [5]. Additionally, it can simultaneously assimilate VFA mixtures, as documented by Cabecas Segura et al. [18].

Strain reactivation, growth, and hydrogen production experiments were conducted using RCV mineral medium [13]. The formulation for 1 L of RCV medium consisted of 50 mL of super salt medium, 20 mL of trace element solution, and a buffer medium containing 0.6 g/L KH₂PO₄ and 0.9 g/L K₂HPO₄. The super salts medium (1 L) contained 4 g MgSO₄·7H₂O, 1.5 g CaCl₂·2H₂O, 0.4 g EDTA (C₁₀H₁₄N₂Na₂O₈·2H₂O), 0.236 g FeSO₄·7H₂O, and 0.02 g thiamine hydrochloride (C₁₂H₁₇ClN₄OS·HCl). The trace elements solution (1 L) consisted of 2.8 g H₃BO₃, 1.592 g MnSO₄·H₂O, 0.752 g Na₂MoO₄·2H₂O, 0.24 g ZnSO₄·7H₂O, and 0.04 g CuSO₄·5H₂O [11]. RCV medium was supplemented with 10 mm monosodium glutamate (C₅H₈NNaO₄·H₂O) as a nitrogen source.

R. capsulatus reactivation and growth were performed at 30 °C and 30 klx light intensity in 15 mL sterile glass test tubes with screw caps using 25 mm sodium lactate (C₃H₅NaO₃) as a carbon source.

The chemicals KH₂PO₄, K₂HPO₄, MgSO₄·7H₂O, MnSO₄·H₂O, ZnSO₄·7H₂O, CuSO₄·5H₂O, Na₂MoO₄·2H₂O, EDTA and CaCl₂·2H₂O were acquired from CTR Scientific (Monterrey, NL, Mexico); H₃BO₃, FeSO₄·7H₂O were acquired from J.T. Baker (Xalostoc, EM, Mexico); thiamine hydrochloride, monosodium glutamate and sodium lactate were acquired from Sigma-Aldrich (St. Louis, MO, USA).

2.2. Substrate Preparation: Production of Dark Fermentation Effluents

The DFE used for photosynthetic hydrogen production was obtained after processing OPW using three effective methods previously identified by López-Hernández et al. [20] and its fermentation in anaerobic digesters (DFE_a,b,d). Figure 1 illustrates the processing routes for obtaining each DFE. DFE_a is obtained after OPW processing without any byproduct generation. DFE_b was produced after the essential oils of OPW were extracted by hydro-distillation. DFE_c was derived from the sequential extraction of essential oils and pectin from OPW. It is important to emphasize that all operations were conducted under consistent conditions across all the processing routes. Such homogeneity ensured that variations in effluent composition were attributed solely to the specific biocompound extraction methods. As previously reported by our research group [20], OPW was dried at 50 °C for 48 h, milled to reduce the particle size to <210 μm, and hydro-distilled at a 1:1.5 (w v⁻¹) OPW:H₂O ratio and boiling temperature for 90 min. The samples were centrifuged at 10,000× g for 10 min. The supernatant was used for pectin precipitation with 80% (v v⁻¹) ethanol at a 1:1 (v v⁻¹) ratio. Dark fermentation was carried out in batch mode at 35 °C, with 40 g L⁻¹ substrate concentration at an inoculum-to-substrate ratio (RIS) of 0.6, and an initial pH of 7. The inoculum was sourced from an anaerobic digester and thermally treated to inhibit methanogenic archaea [24,25,26]. After hydrogen production ceased, the sludge was centrifuged to recover the DFE rich in VFA. The DFE was stored at −20 °C until use.

2.3. Photofermentative Hydrogen Production Evaluation from Dark Fermentation Effluents

Experiments on photofermentative hydrogen production were conducted using 120 mL flat-faced glass bottles with a 110 mL operational volume, placed in a photo-fermentation chamber maintained at 30 °C with a high-pressure sodium lamp (SILVANIA, Sylvania Township, OH, USA) as the light source [11]. Each bottle was appropriately distanced to receive a light intensity of 30 klx, as measured by a digital luxmeter (LX1010B, Alion, Boulder, CO, USA). The initial pH of each assay was adjusted to 7, with no pH control implemented during the fermentation process. The reactors were manually stirred daily, and hydrogen production was continuously assessed using the displacement method in an inverted probe with a 1 M NaOH solution [27].

The experimental design was a 3² factorial to investigate the effects of varying DFE concentrations (25, 35, and 45%) on hydrogen production from DFE_a, DFE_b, and DFE_d [20].

The assays were concluded once no further hydrogen volume was detected in the inverted probe, after which the fermentates were collected for characterization.

2.4. Analytical Methods

The pH was measured according to NMX-AA-25-1984 [28] using a pH meter (Conductronic PC45, Conductronic, Puebla, Mexico). Bacterial growth was evaluated as described by He et al. [13]. In this approach, an absorbance of 1 at 660 nm (UV-vis spectrophotometer SP-300-UV, Cole-Parmer, Lima, Peru) corresponded to a cell dry weight (CDW) of 0.45 g.

2.5. Kinetic Models Fitting

The Modified Gompertz [29], T_i-Gompertz [30], and Boltzmann’s sigmoidal models were used to estimate the kinetic parameters of the cumulative hydrogen production, H(t) (mL H₂).

The Modified Gompertz model (Equation (1)) is as follows:

$$\mathrm{H}\left(t\right)={\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}·\mathrm{e}\mathrm{x}\mathrm{p}\left\{-\mathrm{e}\mathrm{x}\mathrm{p}\left[\frac{{\mathrm{R}}_{\mathrm{m}\mathrm{a}\mathrm{x}}·\mathrm{e}}{{\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}\left(\mathit{\lambda }-t\right)+1\right]\right\}$$

(1)

where H_max (mL H₂) indicates the maximum cumulative hydrogen production, R_max (mL H₂ h⁻¹) refers to the maximum hydrogen production rate, λ (h) denotes the lag phase, t signifies a specific time (h), and e is Euler’s number (2.718) [29].

For the T_i-Gompertz model (Equation (2)), T_i is the time at the inflection point. The equation reads:

$$\mathrm{H}\left(t\right)={\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}·\mathrm{e}\mathrm{x}\mathrm{p}\left\{-\mathrm{e}\mathrm{x}\mathrm{p}\left[-{\displaystyle \frac{{\mathrm{R}}_{\mathrm{m}\mathrm{a}\mathrm{x}}·\mathrm{e}}{{\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}}\left(t-{\mathrm{T}}_{\mathrm{i}}\right)\right]\right\}$$

(2)

Boltzmann’s sigmoidal model (Equation (3)) is expressed as:

$$\mathrm{H}\left(t\right)={\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}+\left[{\displaystyle \frac{{\mathrm{H}}_0+{\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{1+\mathrm{e}\mathrm{x}\mathrm{p}\left(\frac{t-{\mathrm{t}}_{50}}{\mathrm{d}\mathrm{x}}\right)}}\right]$$

(3)

where H₀ is the initial hydrogen production (0 mL H₂), t₅₀ is the time to reach half of H_max, and dx is the slope fit parameter. R_max is calculated as:

$${\mathrm{R}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=\frac{{\mathrm{H}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{4·\mathrm{d}\mathrm{x}}$$

(4)

Hydrogen yields were also fitted using these models. The kinetic parameters were calculated from the experimental cumulative hydrogen production in a volatile fatty acids basis (g VFA) to estimate the cumulative specific hydrogen production.

2.6. Statistical Analysis

Analysis of variance (ANOVA) was performed for the experimental designs using Design-Expert 6.0 software (Design-Ease Inc., Minneapolis, MN, USA), considering a confidence level of 95%. The experimental standard error was calculated using the root-mean-square error of prediction (RMSEP).

3. Results and Discussion

Photofermentative Hydrogen Production

The results for cumulative hydrogen yield showed an inverse relationship with DFE concentration, demonstrating the effect of medium opacity on conversion efficiency, as shown in Figure 2a. The highest yields were observed at 25% DFE, reaching values of 64.8, 126.5, and 122 mL H₂ g⁻¹ VFA for DFE_a, DFE_b, and DFE_d, respectively. In contrast, at 35% DFE, the yields decreased to 17.6, 45.9, and 69.5 mL H₂ g⁻¹ VFA, and further declined to 13.7, 39.9, and 19.5 mL H₂ g⁻¹ VFA at 45% DFE. These findings indicate that while higher DFE concentrations enhance VFA availability, they also intensify medium coloration, which limits light penetration and negatively impacts photofermentative metabolism. Previous studies, such as Rai et al. [9], have attributed light attenuation in photo-fermentation to the self-shading effect caused by increasing cell densities, which restricts light distribution in the culture medium and ultimately limits microbial growth; this effect was not observed in this study. Our results suggest that the primary cause of light limitation was the intrinsic coloration of the DFE, which hindered photon transmission even before substantial biomass accumulation occurred. This distinction highlights the relevance of substrate optical properties when working with real effluents in photo-fermentation systems and underscores the importance of effluent pretreatment or dilution strategies to maintain adequate irradiance in the photobioreactor.

Following a trend similar to that of hydrogen yields, the results for microbial growth demonstrated a marked sensitivity to high DFE concentrations, as shown in Figure 2b. At 25% DFE, biomass accumulation was higher across all treatments, reaching final concentrations of 2.4 ± 0.124 g L⁻¹ for DFE_a, 2.3 ± 0.280 g L⁻¹ for DFE_b, and 1.8 ± 0.115 g L⁻¹ for DFE_d. Increasing the DFE concentration to 35% resulted in noticeable reductions in microbial growth for DFE_b (1.4 ± 0.041 g L⁻¹) and DFE_d (2.2 ± 0.078 g L⁻¹), while DFE_a showed only a slight decrease (2.3 ± 0.079 g L⁻¹). At 45% DFE, all treatments exhibited strong growth inhibition, particularly DFE_b, with the biomass dropping to 0.4 ± 0.015 g L⁻¹. These results are consistent with the previously discussed limitations of light transmission due to medium opacity, which directly affects photoheterotrophic microbial metabolism [31]. Furthermore, the pronounced decline in biomass at the highest DFE levels suggests that, in addition to light limitation, substrate-related inhibitory effects may also occur. Such inhibition could arise from the accumulation of metabolic byproducts or inhibitory compounds present in the concentrated effluents, which may further restrict microbial growth and activity [11].

The results of the initial and final pH measurements showed that the pH remained close to neutral, with slight increases noted at 25% DFE, as shown in Figure 2c. The most significant increase occurred in DFE_d at 25% DFE, where the final pH reached 8.24 ± 0.31. This suggests that metabolic activity under these conditions favors the consumption of acidic intermediates, such as VFA. Conversely, at 45% DFE, mild acidification was observed, with final pH values of 6.73 ± 0.04 for DFEa, 6.60 ± 0.00 for DFE_b, and 6.96 ± 0.01 for DFE_d. This decline in pH indicates reduced VFA consumption, likely due to the inhibition of microbial growth and metabolic activity. The relationship between final pH and biomass accumulation further supports this interpretation, as the lowest pH values were associated with the most restricted microbial growth.

The comprehensive analysis shown in Figure 2d underscores the significant interdependence of the hydrogen yield, microbial growth, and DFE concentration within the photo-fermentation process. This bubble chart reveals that the highest hydrogen yields occurred under conditions that promoted microbial proliferation, particularly at 25% DFE, where VFA were effectively utilized, as evidenced by moderate increases in pH. However, both biomass accumulation and hydrogen yield decreased at 35% and 45% DFE, indicating that substrate availability alone is not sufficient when light penetration is compromised. The consistent decline in microbial growth at elevated DFE concentrations confirms that excessive darkening of the medium restricts photon absorption, thereby limiting metabolic activity and ultimately inhibiting hydrogen production. These findings underscore the importance of optimizing DFE concentration to balance adequate substrate availability and sufficient light penetration, both of which are crucial for sustaining microbial activity and maximizing hydrogen production efficiency.

The findings of this study are consistent with those of previous research on hydrogen production through photo-fermentation, although it is important to consider the key differences in operational conditions and substrate types. Rodriguez-Valderrama et al. [11] reported hydrogen yields of approximately 579.6 mL H₂ g⁻¹ substrate by directly using synthetic VFA media. In contrast, our strategy adopts a circular bioeconomy approach by integrating sequential valorization techniques that couple dark fermentation with photo-fermentation using real effluents derived from OPW as substrates. Compared to other studies employing DFE from different feedstocks, our hydrogen yields were lower. For example, Argun et al. [32] and Sagnak et al. [15] reported hydrogen yields of 693 and 1200 mL H₂ g⁻¹ VFA, respectively, using wheat-based DFE. In contrast, Hitit et al. [16] achieved a yield of approximately 162 mL H₂ g⁻¹ COD using potato DFE, while studies with sweet potato reported yields of 560 mL H₂ g⁻¹ glucose [17]. The nature and composition of the original substrates, VFA profiles, and the operational conditions of each study influence these variations.

Our maximum yield (126.5 mL H₂ g⁻¹ VFA) was lower than that of some synthetic or optimized systems; however, the feasibility of using real OPW-based DFE, a complex and underexplored substrate, was demonstrated. Moreover, microbial growth trends in our study (0.79–3.50 g L⁻¹) were consistent with those documented in prior research [11,13,14,31], reinforcing the robustness and scalability of our approach.

Although this study did not aim to chemically characterize the inhibitory compounds in the fermentation media, the observed trends allow us to raise a hypothesis that may explain the differences in hydrogen yields. DFE_b and DFE_d showed markedly higher hydrogen production than DFE_a, suggesting that essential oil and pectin extraction from OPW may promote more readily processable effluents. In contrast, the consistently lower results with DFE_a point toward the possible presence of inhibitory compounds remaining in untreated OPW. D-limonene—the main component of orange essential oils—has demonstrated inhibitory effects on dark fermentative systems using mixed microbial cultures [33] and on anaerobic consortia [34]. However, such inhibition has not been reported for pure cultures of R. capsulatus, and to our knowledge, no previous study has evaluated this specific interaction. D-limonene is a hydrophobic compound that does not easily remain in aqueous fermentation media for extended periods. Nonetheless, studies have shown that d-limonene undergoes oxidation under aerobic conditions, leading to the formation of limonene hydroperoxides [35,36], which are significantly more water-soluble and have demonstrated cytotoxicity in other microbial systems, such as Escherichia coli [37]. This compound may have been produced during the drying of the OPW. While the presence of limonene hydroperoxides was not confirmed in our experiments, the correlation between aqueous extraction steps and the performance of the resulting DFEs raises a scientifically relevant hypothesis that warrants further investigation.

Future studies focusing on the identification and quantification of d-limonene and its oxidation derivatives in fermentation media could provide valuable insights into the potential inhibitory mechanisms affecting R. capsulatus activity. Clarifying these effects would not only advance our understanding of substrate-specific limitations but also improve the operational feasibility of integrating dark and photo-fermentation systems for citrus waste valorization. Simultaneously, research should prioritize strategies to mitigate the detrimental effects of medium opacity, another critical factor limiting photofermentative performance. Potential approaches include effluent pretreatment to reduce coloration, optimization of reactor design to enhance light distribution, and development of robust microbial consortia capable of maintaining hydrogen production under reduced irradiance. Addressing both the chemical and physical limitations is essential to enhance the efficiency, reliability, and scalability of photo-fermentation processes for sustainable hydrogen production.

ANOVA (Tables S1–S3 in the Supplementary Material) reveals that the DFE concentration had the most significant influence (p < 0.0001) on all measured responses, i.e., cumulative hydrogen yield, cumulative hydrogen production, and final CDW. Effluent type was also significant, although less pronounced (p < 0.05). This reinforces the notion that light attenuation is the primary limiting factor, rather than variations in substrate composition.

The interaction between Effluent type × DFE concentration was not significant for cumulative hydrogen yield (p = 0.1268) or cumulative hydrogen production (p = 0.1298). For the final CDW, the interaction was significant (p < 0.0001), indicating that the degree of inhibition at elevated DFE concentrations was influenced by the substrate used, likely due to variations in VFA composition, coloring, or the possible presence of inhibitory compounds. These statistical findings reaffirm the experimental results, indicating that optimizing the DFE concentration is crucial for maintaining microbial activity and hydrogen production. Effluent pretreatment or enhancements in reactor design, aimed at improving light distribution, could help alleviate the observed adverse effects.

To provide a clearer understanding of hydrogen production dynamics, experimental data were analyzed using three kinetic models. The RMSEP is an indicator of the goodness of fit; thus, it was used as a measure of the prediction effectiveness of each model. As shown in

Table 1. Kinetic parameters of fitting Modified Gompertz, T_i-Gompertz, and Boltzmann models on the maximum cumulative hydrogen production by photo-fermentation of dark fermentation effluents of orange peel waste.

Parameter		DFEa			DFEb			DFEd
DFE concentration (v v−1 %)	25	35	45	25	35	45	25	35	45
Production Hmax,e (mL H2)	21.00 ± 4.24	8.00 ± 2.83	8.00 ± 2.83	33.50 ± 6.61	17.00 ± 4.24	19.00 ± 4.24	31.33 ± 5.03	25.00 ± 1.41	9.00 ± 4.24
Modified Gompertz model
Hmax (mL H2)	24.22	8.53	7.25	33.51	17.35	18.79	32.99	31.12	9.42
Rmax (mL H2/h)	0.25	0.13	0.40	0.57	0.21	0.32	0.51	0.30	0.16
λ (h)	20.14	14.25	29.81	17.24	6.99	5.08	24.23	28.59	3.60
RMSEP	0.77	0.48	0.28	0.39	0.57	0.83	0.63	0.51	0.51
Ti-Gompertz model
Hmax (mL H2)	24.22	8.53	7.25	33.51	17.35	18.79	32.99	31.12	9.42
Rmax (mL H2/h)	0.25	0.13	0.40	0.57	0.21	0.32	0.51	0.30	0.16
Ti (h)	55.46	38.38	36.56	39.01	36.84	26.55	48.05	66.84	25.59
RMSEP	0.77	0.48	0.28	0.39	0.57	0.83	0.63	0.51	0.51
Boltzmann’s sigmoidal model 1
Hmax (mL H2)	21.80	8.21	7.21	32.34	16.47	18.33	31.49	26.66	9.17
Rmax (mL H2/h)	0.28	0.14	0.40	0.59	0.22	0.32	0.54	0.34	0.16
t50 (h)	64.94	47.75	39.57	47.52	47.39	35.44	56.91	73.98	34.70
dx (h)	19.54	15.07	4.52	13.62	18.88	14.19	14.61	19.76	13.91
RMSEP	0.91	0.42	0.30	0.83	0.84	1.03	0.72	0.54	0.37

¹ Initial hydrogen production, H₀ = 0 mL H₂. Notes: DFE, dark fermentation effluent; dx, slope fit parameter; H_max, maximum cumulative hydrogen production; H_max,e, maximum experimental cumulative hydrogen production; R_max, maximum hydrogen production rate; RMSEP, root mean square error of prediction; T_i, time at the inflection point; t₅₀, time to reach half of H_max; and λ, lag phase.

for cumulative hydrogen production, both the Modified Gompertz and T_i-Gompertz models delivered the most accurate fits, as reflected by the lowest RMSEP across all experimental conditions. Notably, for DFE_b at 25% concentration, the Modified Gompertz model achieved the lowest RMSEP of 0.39, successfully predicting H_max 33.51 mL H₂, which closely aligned with the experimental measurement of 33.50 mL H₂. Similarly, for DFE_d at 25% concentration, both the Modified Gompertz and T_i-Gompertz models demonstrated strong predictive accuracy, each with an RMSEP value of 0.63. In contrast, Boltzmann’s model exhibited greater deviations, particularly at higher DFE concentrations, with RMSEP values reaching 1.03 for DFE_b at 45%. This indicates a less accurate representation of the hydrogen production dynamics under these specific conditions.

Table 2. Kinetic parameters of fitting Modified Gompertz, T_i-Gompertz, and Boltzmann models on the maximum cumulative specific hydrogen production by photo-fermentation of dark fermentation effluents of orange peel waste.

Parameter		DFEa			DFEb			DFEd
DFE concentration (v v−1 %)	25	35	45	25	35	45	25	35	45
Yield, hmax,e (mL H2 g−1 VFA)	64.77 ± 13.09	17.63 ± 6.23	13.71 ± 4.85	126.50 ± 24.95	45.85 ± 11.44	39.86 ± 8.90	122.01 ± 19.60	69.54 ± 3.93	19.47 ± 9.18
Modified Gompertz model
hmax (mL H2 g−1 VFA)	74.71	18.78	12.43	126.55	46.79	39.42	128.45	86.55	20.37
rmax (mL H2 g−1 VFA h)	0.78	0.29	0.68	2.14	0.58	0.68	1.98	0.83	0.34
λ (h)	20.14	14.25	29.81	17.24	6.99	5.08	24.23	28.59	3.60
RMSEP	2.37	1.05	0.48	1.47	1.55	1.74	2.46	1.41	1.10
Ti-Gompertz model
hmax (mL H2 g−1 VFA)	74.71	18.78	12.43	126.55	46.79	39.42	128.45	86.55	20.37
rmax (mL H2 g−1 VFA h)	0.78	0.29	0.68	2.14	0.58	0.68	1.98	0.83	0.34
Ti (h)	55.46	38.38	36.56	39.01	36.84	26.55	48.05	66.84	25.59
RMSEP	2.37	1.05	0.48	1.47	1.55	1.74	2.46	1.41	1.10
Boltzmann’s sigmoidal model 1
hmax (mL H2 g−1 VFA)	67.25	18.09	12.36	122.11	44.43	38.46	122.64	74.16	19.84
rmax (mL H2 g−1 VFA h)	0.86	0.30	0.68	2.24	0.59	0.68	2.10	0.94	0.36
t50 (h)	64.94	47.75	39.57	47.52	47.39	35.44	56.91	73.98	34.70
dx (h)	19.54	15.07	4.52	13.62	18.88	14.19	14.61	19.76	13.91
RMSEP	2.80	0.93	0.52	3.12	2.27	2.16	2.80	1.51	0.81

¹ Initial specific hydrogen production, H₀ = 0 mL H₂ g⁻¹ VFA. Notes: DFE, dark fermentation effluent; dx, slope fit parameter; h_max, maximum cumulative specific hydrogen production; h_max,e, maximum experimental cumulative specific hydrogen production; r_max, maximum specific hydrogen production rate; RMSEP, root mean square error of prediction; T_i, time at the inflection point; t₅₀, time to reach half of h_max; and λ, lag phase.

presents the kinetic parameters for hydrogen yield (mL H₂ g⁻¹ VFA), revealing that both the Modified Gompertz and T_i-Gompertz models exhibited superior predictive capabilities. For DFE_b at 25% concentration, the Modified Gompertz model achieved an RMSEP of 1.47, with a maximum cumulative specific hydrogen production (h_max) of 126.55 mL H₂ g⁻¹ VFA, closely aligning with the experimental value of 126.50 mL H₂ g⁻¹ VFA. Similarly, the T_i-Gompertz model for DFE_d at 25% demonstrated the lowest RMSEP of 1.41, further confirming its reliability in predicting hydrogen yield under optimal conditions. In contrast, Boltzmann’s model exhibited higher RMSEP values, particularly for DFE_b at 25%, with a measurement of 3.12, indicating its limitations in accurately modeling hydrogen production kinetics under these scenarios.

Despite the slightly better overall fit observed for the Modified Gompertz and T_i-Gompertz models, the results indicate that all three models provided acceptable predictive performance. The Modified Gompertz and T_i-Gompertz models yielded lower and identical errors due to their mathematical equivalence. Boltzmann’s model also showed a strong agreement with the experimental data under several conditions, particularly at lower DFE concentrations. These findings suggest that the selection of a specific kinetic model should be guided not only by minor differences in predictive accuracy but also by the type of kinetic insight required. For instance, the Modified Gompertz model explicitly characterizes the lag phase (λ), while the T_i-Gompertz model highlights the inflection point (T_i), both of which are useful for understanding system startup dynamics. Meanwhile, the Boltzmann model offers a simpler sigmoidal behavior, which may be advantageous in comparative studies or for rapid modeling. Therefore, all three models can be considered valuable tools for describing photofermentative hydrogen production kinetics, and their use should be tailored to the specific needs of the process design, control, or optimization.

This study demonstrates the potential of photo-fermentation of DFE using a pure strain (R. capsulatus B10); however, the use of axenic cultures poses inherent limitations for large-scale applications. Pure cultures are particularly susceptible to contamination, and their long-term stability under non-sterile conditions remains a challenge [38]. This constraint represents a critical barrier to the practical implementation of photo-fermentation technologies and underscores the necessity of exploring robust microbial consortia or mixed cultures that can potentially augment metabolic pathway diversity and resistance to contamination, thereby withstanding environmental fluctuations and operational challenges [39]. Moreover, the use of VFA as a carbon source in photo-fermentation diminishes contamination risks, as VFA are fermentation end products that cannot undergo further fermentation, thereby restricting their uptake to particular groups of bacteria. Therefore, photo-fermentation based on VFA may exhibit enhanced resilience to contamination, potentially reducing the need for extensive sterilization procedures and proving advantageous for industrial applications [40]. However, photofermentative hydrogen production faces major technical challenges, including limited light penetration, relatively low energy conversion efficiency, and sensitivity to environmental conditions. Addressing these limitations will require advances in reactor engineering, light distribution systems, and process integration. Future developments should prioritize adaptive microbial systems and couple photo-fermentation with other bioprocesses to enhance productivity, reduce costs, and improve robustness under real-world conditions.

4. Conclusions

The technical feasibility of photofermentative hydrogen production from real dark fermentation effluents (DFE) derived from orange peel waste (OPW) was investigated. Promising pathways for the sequential valorization of agro-industrial residues have been developed, and critical technical barriers, such as light penetration and potential substrate inhibition, have been identified.

The findings confirmed that a 25% DFE concentration yields the most favorable results in terms of microbial growth (2.4 g L⁻¹) and hydrogen yield (126.5 mL H₂ g⁻¹ VFA). In contrast, higher concentrations induce a decline in performance, primarily due to light attenuation caused by medium opacity.

DFE derived from OPW without essential oil extraction led to significantly lower hydrogen yields, raising the hypothesis of potential microbial inhibition, possibly due to the unverified presence of either d-limonene or its oxidative byproducts. Its consistent occurrence across experiments indicates a relevant area for future investigation.

Kinetic analysis revealed that all the studied models provided satisfactory predictive performance, with the Modified Gompertz and T_i-Gompertz models offering robust tools for characterizing hydrogen production dynamics. Therefore, the choice of model should depend on the desired kinetic insight for design or optimization.

Future work should prioritize elucidating inhibitory mechanisms, particularly through the quantification of d-limonene and its oxidative derivatives, to evaluate their potential effects on R. capsulatus and clarify the limitations observed in this study. Additionally, improving reactor configurations and tailoring effluent pretreatment to reduce medium opacity and enhance hydrogen productivity from real waste sources must be addressed to enable scaling-up.