Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Journal of Environmental Chemical Engineering 11 (2023) 111252

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering


journal homepage: www.elsevier.com/locate/jece

Optimization of fermentative parameters to improve hydrogen production:


Is the co-fermentation of waste from the citrus agroindustrial an interesting
alternative for energy recovery?
Danilo Henrique Donato Rocha *, Isabel Kimiko Sakamoto , Maria Bernadete Amâncio Varesche *
Department of Hydraulics and Sanitation, São Carlos School of Engineering, University of São Paulo, Av Trabalhador São Carlense, 400, São Carlos, SP 13566-590,
Brazil

A R T I C L E I N F O A B S T R A C T

Editor: Yujie Men The co-fermentation of peel and wastewater from citrus processing is a proposal to promote the complementation
of the individual characteristics of these substrates, making it possible to obtain higher efficiency in hydrogen
Keywords: (H2) production, in addition to equating problems related to the logistics of destination and treatment of these
Calcium carbonate waste. To build a model for optimizing hydrogen production, based on the variables of citrus peel waste con­
Citrus peel waste
centration (9.9 – 35.1 gTVS.L-1), citrus processing wastewater concentration (0.3 – 8.7 gCOD.L-1), and calcium
Citrus wastewater
carbonate (CaCO3) concentration (0.3 – 7.3 g.L-1), the Rotational Central Composite Design and Response Sur­
Clostridium
Energy potential face Methodology (RSM) were applied, using a mesophilic granular sludge as inoculum for batch assays subjected
Hydrogen to thermal treatment. Hydrogen production from 199.2 to 1241.9 mLH2.L-1 was obtained in the assays, with the
Organic acid lowest production obtained with 22.5 gTVS.L-1 of citrus peel waste, 4.5 gCOD.L-1 of citrus processing wastewater,
and 0.3 g.L-1 of calcium carbonate, for a gCaCO3/gCarbohydrate ratio of 0.03. In the optimized condition with
21.4 gTVS.L-1 of citrus peel waste, 3.4 gCOD.L-1 of citrus wastewater and 4.5 g.L-1 of calcium carbonate, which
resulted in a gCaCO3/gCarbohydrate ratio of 0.53, 1249.0 mLH2.L-1 were obtained. In the hydrogen optimized
condition assay a relative abundance of 95% was observed for Clostridium Sensu Stricto 1, and therefore the
Clostridial metabolic pathway was inferred as predominant, with a maximum butyric acid and acetic acid
production of 2230.8 mg.L-1 and 1213.6 mg.L-1, respectively. For the lowest hydrogen production assay, there
was a significant reduction in the Clostridium sensu stricto 1 (51.2%) abundance, and an increase in the Para­
clostridium (20.0%) and Lactobacillus (21.9%) abundances, inferring lactic acid fermentation (6367.8 mg.L-1) as
the predominant pathway. An energy potential of 523.2 MJ.t-1 of citrus waste was estimated based on the
hydrogen production for the optimized condition assay.

1. Introduction products [2].


The agroindustrial sector of citrus processing for juice production is
Agroindustrial waste, which includes peel, seeds, pits, pulp, press an important source of lignocellulosic solid waste and wastewater,
cakes, leaves, stems, and others, resulting from various processing sec­ considering the significant production of these fruits. The annual global
tors, are still deposited in landfills or just discarded in inappropriate orange production is estimated at around 75 million tons. Brazil is the
places. These by-products can be used to obtain value-added products most important country in the global market of oranges and their de­
such as biofuels, and other biotechnological products [1]. The rivatives, with an annual production of 17 million tons of fruit, repre­
Waste-to-energy (WTE) supply chain is an alternative to simultaneously senting almost a quarter of world production [3,4]. A large amount of
address the need to increase energy supply and improve waste man­ production is directed towards processing, which makes Brazil the main
agement, thus achieving a circular economy model. Using WTE tech­ global orange juice producer and exporter, in which the European Union
niques, different types of biomass, such as agricultural, forestry, and the United States are the most relevant consumer markets [5,6].
domestic, and animal waste, can be converted into several bioenergetic The estimate is that from 1000 kg of oranges, an average of 553 kg of

* Corresponding authors.
E-mail addresses: danilorocha21@usp.br (D.H.D. Rocha), varesche@sc.usp.br (M.B.A. Varesche).

https://doi.org/10.1016/j.jece.2023.111252
Received 13 June 2023; Received in revised form 4 September 2023; Accepted 11 October 2023
Available online 13 October 2023
2213-3437/© 2023 Elsevier Ltd. All rights reserved.
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

juice is extracted in its natural condition, as well as 30.0 kg of pulp, 3.0 using various waste from agroindustrial processes as substrates. Citrus
kg of peel oil, and a high amount of solid waste from the peel and peel waste was evaluated by Camargo et al. [35], whose maximum H2
bagasse: altogether more than 400 kg [7]. In addition to the peel and production was 1085.0 mLH2. L-1, obtained by optimizing the inoculum
bagasse waste generated directly from the fruit after juice extraction, concentration, initial pH, and solid substrate concentration. Villa Mon­
there is also wastewater generation resulting from various processing toya et al. [36] used waste from coffee processing to produce hydrogen,
steps, which include fruit washing, evaporation of the citrus juice, and optimizing the initial pH, substrate concentration, and headspace
washing of the pulp, cleaning operations, effluents from production of percentage, obtained a maximum production of 686.0 mLH2. L-1. Rabelo
pectin, citric acid, molasses and essential oils, estimated from 1 m3 to 3 et al. [34] and Soares et al. [25] sought to increase hydrogen production
m3 of wastewater per ton of processed fruit, reaching up to 17 m3 , using sugarcane bagasse as a substrate, for which variables such as yeast
depending on the water use and reuse practices in the industry, and in extract concentration, substrate concentration, and initial pH were
function of the adopted processing technology [8–12]. optimized, whose maximum productions were 517.0 mLH2. L-1 and 36.0
Conventional alternatives for CPW disposal are treated as insufficient mlH2. L-1, respectively. Banana waste as a substrate for hydrogen pro­
and problematic in terms of environmental impacts and energy effi­ duction was evaluated by Mazareli et al. [37], and optimizing the initial
ciency [13], and therefore routes for this waste valuation have been pH and temperature variables, obtained a maximum production of 468.0
proposed to establish sustainable disposal logistics [14]. Alternatives mLH2. L-1.
such as the production of fertilizers, pectin, ethanol, essential oils, In the present study, the potential of co-digestion of solid waste and
chemical products, cattle feed, and absorbent material, have been wastewater from the citrus processing agroindustry was investigated,
evaluated from a technical and economic point of view [15]. Applica­ which until now has not been explored despite the inherent advantages
tions such as animal feed or organic fertilizer are the most common and of using by-products from the same productive chain. For this purpose, a
simple alternatives to process this waste because they require little Rotational Central Composite Design (RCCD) and Response Surface
infrastructure and investment [16], however, due to the high waste Methodology (RSM) were carried out to obtain a model for optimizing
amount and the low value added of the recovered material, they are not H2 production, based on the variables of citrus peel waste concentration
capable of absorbing all the waste generation [17]. Another important (9.9 – 35.1 gTVS.L-1), citrus processing wastewater concentration (0.3 –
aspect is the seasonality in the generation of these waste, with high 8.7 gCOD.L-1) and calcium carbonate concentration (0.3 – 7.3 g.L-1). The
generation concentrated in the harvest period, as highlighted by Calabrò choice of these variables for optimization aims to establish a direction
et al. [18]. These authors evaluated the anaerobic digestion of orange for the feasibility of using these substrates together. In addition, the
peel waste after ensiling as an alternative for better waste management. buffering agent effect was explored as a strategy for better stabilization
Energy recovery with hydrogen production through fermentation of the H2 production, and to promote a significant increase in H2 yield.
using agroindustrial waste has been widely evaluated [19–23]; however, Thus, based on the maximum production achieved, it can be evaluated if
some factors are still obstacles to conducting a viable process with high the process is favorable compared to other studies in fermentative
H2 production. Adjusting the operating conditions, such as pH, tem­ conditions. The microbial community characterization was also carried
perature, substrate concentration, hydraulic retention time (HRT), and out for different assays, aiming to associate changes in the H2 production
supplementation with buffering agents and nutrients, among others, is and soluble metabolite concentrations with changes in the metabolic
fundamental for the fermentation success [24,25], and the peculiarities pathways inferred for the process. An evaluation for energy recovery of
of each substrate show the need to investigate conditions that are suit­ citrus waste was estimated to verify the H2 potential use as an alterna­
able for each case. tive energy source in the citrus agroindustry, stimulating the circular
The co-fermentation of peel and wastewater from citrus processing is economy proposal.
an alternative to facilitating the logistics of treatment and disposal of Therefore, this study provides scientific information to support the
these waste in the industry itself [26]. The co-fermentation of two or application of citrus waste anaerobic co-digestion as an alternative for
more by-products is interesting because it can complement the indi­ waste management in a circular economy model for agroindustry,
vidual substrate characteristics to improve hydrogen production effi­ providing a better fit in the proportion between substrates, the specific
ciency [27]. Co-fermentation makes it possible to improve nutrient demand for buffering agents, and the potential for energy recovery.
balance, toxic or recalcitrant compounds dilution, and increase the
available organic load, in addition to promoting synergistic effects be­ 2. Material and methods
tween microorganisms [28,29]. The lignocellulosic characteristic of
citrus waste is an aspect that makes its biodegradation more difficult, 2.1. Substrate
imposing limitations on the use of this carbon source as the hydrolysis
step for direct conversion to a readily assimilable substrate is hampered, The juice and seeds of oranges var. Pera-Rio (Citrus sinensis L.
leading to a slower conversion of polymeric carbohydrates into simple Osbeck) were removed by manual processing, and the peel and bagasse
sugars [30–32], therefore the provision of a soluble substrate by from the process were used as substrates in the assays. Citrus peel waste
wastewater promotes synergism between substrates. (CPW) composed of peel and bagasse was dehydrated in an oven at 50 ◦ C
As the co-digestion of these substrates has not been explored so far, for 48 h and then submitted to mechanical grinding pre-treatment
there is little information related to the process parameters, so a deep reducing particles < 0.60 mm (knife mill type Willye Model SL 32,
investigation of variables with a significant effect on the process is 1750 rpm – Solab), and the analytical characterization of in natura and
necessary, aiming to reach a high hydrogen production. For processes in dry waste is shown in Table 1SM.
which many variables can influence the responses, the factorial designs The wastewater from citrus processing used in the co-fermentation
can minimize the difficulties of achieving the best process optimization assays was supplied by a juice production agroindustry located in Ara­
[33]. Through factorial experimental designs, the effects of many vari­ raquara, São Paulo, Brazil, and stored at − 20 ◦ C, and its analytical
ables can be optimized as the statistical model can establish relation­ characterization is shown in Table 2SM.
ships between responses of two or more experimental variables that are
simultaneously analyzed [34]. In general, this methodological approach 2.2. Inoculum
obtains the best response and even reduces the number of assays, which
implies a reduction in time and costs [33]. The sludge used as inoculum in the batch reactors comes from a
Experimental design methodologies and response surface models mesophilic UASB reactor for wastewater from a poultry slaughterhouse
(RSM) to optimize hydrogen production are applied in many studies for treatment (Avícola Dacar – Tietê, São Paulo, Brazil), whose character­
different independent variables that affect the fermentation process, ization is presented in the Supplementary Material.

2
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

Table 1 (100%).
Independent variables evaluated at the factorial (+1 and − 1) and axial (1.68 and Table 1 presents the independent variables with the values corre­
− 1.68) levels and for the central points (0) in the rotational central composite sponding to the delimited factorial and axial levels, and central point.
design (RCCD). The range of values defined for each variable was delimited to ensure
Variables Coded -1.68 -1 0 1 1.68 that an experimental condition for optimizing hydrogen production
Citrus Peel Waste X1 9.9 15.0 22.5 30.0 35.1 could be identified, considering the three investigated variables. For
Concentration (CPW) CPWW, the concentration variation was taken close to the possible limits
(gTVS.L-1) due to the raw effluent characteristics, according to the physical-
Citrus Processing X2 0.3 2.0 4.5 7.0 8.7 chemical characterization (Table 2SM).
Wastewater
Concentration (CPWW)
Table 2 presents the experimental design matrix of the Rotational
(gCOD.L-1) Central Composite Design used to establish the experimental conditions
Calcium Carbonate X3 0.3 1.7 3.8 5.9 7.3 of each assay.
Concentration (CaCO3) (g. A quadratic polynomial equation (Eq. 1) describes the maximum
L-1)
hydrogen production potential (P) based on the independent variables
(X1, X2, X3). The individual and interaction of variable effects on the
experimental response were evaluated at a 10% confidence level.
Table 2
Matrix of the experimental design (RCCD) with the coded levels and the Y = β0 + β1 X1 + β2 X2 + β3 X3 + β11 X 21 + β22 X 22 + β33 X 23 + X1 X2 + X2 X3 + X1 X3
respective values for each variable. (1)
Assay X1 X2 X3 CPW CPWW CaCO3
(gTVS.L-1) (gCOD.L-1) (g.L-1) Where Y is the experimental response (maximum H2 production po­
tential), β0 , β1 , β2 andβ3 are the linear coefficients, β11 , β22 andβ33 are the
E-1 -1 -1 -1 15.0 2.0 1.7
E-2 1 -1 -1 30.0 2.0 1.7
quadratic coefficients, and X1 , X2 andX3 are the independent variables.
E-3 -1 1 -1 15.0 7.0 1.7
E-4 1 1 -1 30.0 7.0 1.7 2.4. Validation
E-5 -1 -1 1 15.0 2.0 5.9
E-6 1 -1 1 30.0 2.0 5.9
E-7 -1 1 1 15.0 7.0 5.9 Based on the results obtained through the response surface, a model
E-8 1 1 1 30.0 7.0 5.9 validation assay was carried out, allowing verification of its accuracy by
E-9 -1.68 0 0 9.9 4.5 3.8 comparing the predicted and experimental H2 production. Therefore,
E-10 1.68 0 0 35.1 4.5 3.8
the validation assay was performed in an experimental condition where
E-11 0 -1.68 0 22.5 0.3 3.8
E-12 0 1.68 0 22.5 8.7 3.8
maximum hydrogen production is obtained, with 21.4 gTVS.L-1 of CPW,
E-13 0 0 -1.68 22.5 4.5 0.3 4.5 gCOD.L-1 of CPWW, and 3.4 g.L-1 of CaCO3.
E-14 0 0 1.68 22.5 4.5 7.3
E-15 0 0 0 22.5 4.5 3.8
2.5. Physicochemical and chromatographic analysis
E-16 0 0 0 22.5 4.5 3.8
E-17 0 0 0 22.5 4.5 3.8
For monitoring and data collection during the fermentation assays,
the liquid fraction analyses of the batch reactors were carried out, such
In order to select endospore-forming bacteria, commonly associated as total soluble carbohydrates [41], chemical oxygen demand - COD
with high H2 yields, and mainly to inactivate hydrogen-consuming mi­ [42], total phenols [43], total volatile solids (TVS), total suspended
croorganisms, a heat treatment was applied to the anaerobic granular solids (TSS) and volatile suspended solids (VSS), organic acids [44].
sludge, according to the methodology applied by Kim et al. [38] and Carbohydrates, phenols, and COD analyses were performed for the
Menezes and Silva [39]. sample filtered through a 0.45 µm filter. The quantification of total
soluble carbohydrates, pH, and organic acids was performed on the
2.3. Factorial Design initial and final samples, and also on samples collected during the assay.
The D-limonene quantification in the samples from the reactors was
A set of fermentation assays was carried out using different con­ performed using the standard addition method, through gas chroma­
centrations of citrus peel waste - CPW (gTVS.L-1), citrus processing tography coupled to mass spectrometry (GC-MS Shimadzu, Column SH-
wastewater - CPWW (gCOD.L-1), and calcium carbonate - CaCO3 (g.L-1). Rtx-5MS 30 m x 0.25 mm × 0.25 mm). The sample temperature was
The experimental design of the assays was defined based on the Rota­ raised to 240 ◦ C in the oven, at a rate of 30 ◦ C, with a retention time of
tional Central Composite Design (RCCD) methodology, evaluating a 5 min. Injection of 250 µL of headspace sample was performed, with an
range of values for the independent variables, and identifying an opti­ injection temperature of 200 ◦ C, Split 10:1, using helium as carrier gas.
mized experimental condition for accumulated H2 production through For mass spectrometry, the full scan method was used for the mass range
the Response Surface Methodology (RSM); the factorial design has been from 50 to 200.
frequently applied to assess and relate significant independent variables The biogas production monitoring was carried out from 500 µL
to responses of interest from diverse research fields. [40]. The factorial headspace samples, using a syringe equipped with a pressure lock. Gas
design was carried out with three factors (k = 3), and, therefore, ± α = chromatography was used to analyze the biogas composition (Shimadzu
1.68, and according to the methodology, there are 2k factorial points, GC- 2014, equipped with a thermal conductivity detector (TCD), Car­
totaling eight assays; 2 *k axial points, totaling six assays, and three boxenTM 1010 PLOT capillary column 30 m x 0.53 mm, Supelco), with
central point assays. Thus, the experimental design consisted of 17 as­ a temperature of 220 ◦ C in the injector, 130 ◦ C for the column, and
says, carried out in triplicate, totaling 51 batch reactors. 230 ◦ C for the detector, using argon as carrier gas (12 mL.min-1). A
The assays were carried out in borosilicate flasks (250 mL) with a periodic measurement of the reactor headspace pressure (mbar) was
plastic screw cap and equipped with a set of valves coupled to the butyl carried out before and after gas sampling.
cap to collect liquid and gas samples, and incubated in an Informs HT –
Multitron Pro shaker, kept stirred at 130 rpm and a mesophilic tem­ 2.6. Data Analysis
perature of 37 ◦ C. In addition, fixed conditions for all assays were 3.5
gTVS.L-1 of inoculum, and 50% headspace (v/v), under N2 atmosphere The accumulated hydrogen production (Hi ) was obtained by sum

3
­
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

ming the H2 produced volume between the monitoring measurements (i) (ThermoFisher Scientific). DNA sequencing was performed by the
carried out at intervals of 2 – 6 h throughout the assay, calculated ac­ company NGS genomic solutions (Piracicaba-SP, Brazil), and the 16 S
cording to Eq. (2), based on a hydrogen produced volume (VH,i ) between rRNA genes in the V3-V4 region were amplified adopting the polymer­
the previous (i-1) and current (i) sampling points. Moreover, the ase chain reaction (PCR) using the primer set 341 F - 806 R [51], per­
hydrogen volume (VH,i ) was obtained based on the Clapeyron equation formed with Phusion® High-Fidelity PCR Master Mix (New England
principles, according to Eq. (3). Biolabs). Sequencing libraries were generated using NEBNext Ultra DNA
∑n Library Pre ® Kit for Illumina and following the manufacturer’s rec­
Hi = i=1
VH,i (2) ommendations. The DADA2 package [52] was used to transform the
sequencer fastq files into inferred, dismembered, chimera-free sample
VH,i =
ΔPi ∗ V ∗ CH,i
(3) sequences. Fastq file filtering was performed to cut PCR primer se­
Pi quences and filter out the 3’-ends of the reads due to quality decay
(Q<30). Taxonomies were assigned to each ASV (Amplicon Sequencing
Where CH,i is the hydrogen percentage in the biogas at the sampling
Variants) using a DADA2 program implementation of the naive Bayesian
point (i); Pi is the reactor headspace pressure; ΔPi the headspace pres­
classifier method. The Silva 138 database was used as a reference. The
sure variation between the previous (i - 1) and current (i) sampling
taxonomic classifications generated by DADA2, and their quantifica­
points; and V the headspace volume.
tions, were imported into the phyloseq program [53], and implemented
The hydrogen percentage in the biogas (%H2) for each assay was
in the R language. The sequences were submitted to the National Center
given by the arithmetic mean of the H2 percentage values from the
for Biotechnology Information (NCBI) database under accession number
sample points for the period in which biogas composition stabilization
PRJNA909571.
was identified; this stabilization was verified by a coefficient of variation
The BRENDA [54] and KEGG [55] enzyme databases were used to
(CV) < 5%.
infer the predominant metabolic pathways and enzymes that catalyze
The accumulated H2 production – H (mL.L-1) was adjusted to the
conversion reactions, based on the microorganism genera identified in
modified Gompertz kinetic model (Eq. 4) using the OriginPro 8.5®
abundance > 1% in the assays OC-H and E-10.
software, obtaining the kinetic parameters of maximum H2 production
potential – P (mL.L-1), maximum H2 production rate – Rm (mL.L-1.h-1),
3. Results and discussion
and time to start H2 production - λ (h).
{ [ ]}
Rm ∗ e 3.1. H2 production and optimization by RCCD
H(t) = P ∗ exp − exp (λ − t) + 1 (4)
P
The kinetic parameters of the fermentation assays are shown in
Where e = 2.718281828. Table 3. Maximum H2 production potential (P) values were observed
The experimental data from the factorial design and the response between 199.21 mLH2. L-1 (E-13) and 1241.91 mLH2. L-1 (E-17); the
surface model were adjusted and statistically analyzed using Minitab® maximum H2 production rate (Rm) was observed between 7.35 mLH2. L-
1 -1
19.1 and STATISTICA 10 software, at a confidence level of 10% .h (A-13) and 106.58 mLH2. L-1.h-1 (A-3); and the time to start H2
(p < 0.10) [24,45,46]. production (λ) observed between 7.75 h (A-13) and 14.83 h (A-17). As
verified, an important variation was observed for P, Rm, and λ under
2.7. Energy balance Assessment different experimental conditions (Fig. 1), which reflects the importance
of adjusting the concentrations of solid waste, wastewater, and calcium
The hydrogen energy potential - PEH2 (MJ.kg-1CPW) was calculated carbonate, to obtain a better response to H2 production.
by Eq. (5) based on maximum hydrogen production potential – P (mL.L- According to the analysis of variance (ANOVA), the linear term X2
1
), on the lower heating value of H2 – LHVH2 (10.8 MJ.m-3) [47], and the (CPWW concentration, p = 0.093) and the quadratic terms X21 (CPW
CPW concentration added to the reactor – CPWC (g.L-1). concentration, p = 0.015), X22 (CPWW concentration, p = 0.037), and X23
(CaCO3 concentration, p = 0.019) of the model were statistically sig­
EPH2 =
PH2 ∗LHV H2
(5) nificant (p < 0.10), and resulted in individual effects on hydrogen pro­
1000 ∗ CPW C duction (Table 4). The linear terms X2 and X3, and the interaction terms
X1X2 (p = 0.912), X1X3 (p = 0.308) and X2X3 (p = 0.166) were not
EPH2 ∗ η
EPPH2 = (6) significant (p > 0.10), probably requiring individual interpretation of
3600
the independent variables.
Based on the PEH2, the electricity production potential was estimated Based on the data regression analysis, a second-order polynomial
- PPEH2 (MWh.kg-1CPW) through Eq. (6), considering 60% efficiency (η) function given by the three coded variables was adjusted, resulting in a
for electrical conversion into fuel cells [48,49]. To estimate the process quadratic model of maximum hydrogen production potential. Based on
scale-up, the PPEH2 was extrapolated based on the estimated citrus the significant terms (p < 0.10) for hydrogen production, the model can
waste generation – CPWG, based on data from a processing industry be represented by Eq. (7).
[50]. ( )
Y mL.L− 1 = − 1241 + 254.0X2 − 3.49X 21 − 25.13X 22 − 42.2X 23 (7)
2.8. Microbial Community Analysis The model significance was evaluated based on the ANOVA test and
was statistically significant based on the F values; the F calculated by
Extraction and sequencing of the 16 S rRNA gene were performed for regression (3.23) was higher than the tabulated F value (2.72) for a
samples collected at the end of the validation assay for the optimized H2 confidence interval of 10%, and 7 degrees of freedom for the residuals,
production and for the E-10 assay. The samples were centrifuged and 9 degrees of freedom for the regression. Thus, the null hypothesis H0
(6000 rpm - 6 min), and the biomass pellet was washed with PBSX1 may be rejected, implying that at least one of the variables X1, X2, and X3
solution (NaCl 8.2 g.L-1, Na2HPO4 1.05 g.L-1, NaH2PO4 0.35 g.L-1), and contributes significantly to the model. The model has a coefficient of
then stored at − 20 ◦ C. Genomic DNA from the samples was extracted determination (R2) of 0.81, that is, the model explains 81.0% of the
using the FastDNA™ SPIN Kit for Soil DNA Extraction (MP Biomedicals). variability in H2 production, indicating the aptness of the model to
The integrity of the extracted DNA was checked on 0.8% agarose gel, establish an adequate correlation between predicted and observed
and the quantification (ng/µL) and purity analysis (260/280 nm ratio) values for the response. R2 values between 0.75 and 1.00 indicate a good
were performed on the Nanodrop 2000 Spectrophotometers

4
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

Table 3
Kinetic parameters, yield, and pH, for RCCD assays.
Assay P mL.L-1 Rm mLL-1.h-1 λh %H2 Y1 mLH2. g-1CHO Y2 molH2.mol-1 hexose gCaCO3.g-1 CHO Initial pH Final pH

E-1 823.7 58.5 11.5 47.4% 214.36 2.06 0.32 6.64 4.95
E-2 541.0 33.6 10.0 30.2% 112.23 1.05 0.21 6.31 4.90
E-3 898.8 106.6 12.3 51.0% 182.65 1.87 0.24 6.41 4.56
E-4 578.6 38.9 12.7 44.8% 75.47 0.68 0.14 6.28 4.04
E-5 869.7 49.8 10.8 42.9% 169.53 1.62 0.87 6.67 5.94
E-6 835.5 55.8 10.4 37.3% 123.55 1.17 0.60 6.51 5.67
E-7 423.4 17.2 9.1 31.9% 69.73 0.61 0.77 6.56 5.79
E-8 493.8 46.4 11.4 30.5% 53.51 0.53 0.46 6.46 5.37
E-9 892.0 71.2 9.5 47.8% 209.30 2.13 0.66 6.82 5.64
E-10 522.7 53.3 11.2 33.2% 62.74 0.61 0.28 6.28 4.27
E-11 1056.6 77.4 10.5 42.2% 181.46 1.83 0.46 6.65 5.63
E-12 578.8 56.3 10.5 39.5% 73.88 0.70 0.35 6.39 4.32
E-13 199.2 7.4 7.8 48.3% 75.87 0.65 0.03 5.86 3.52
E-14 1224.5 88.5 13.4 43.5% 144.40 1.46 0.67 6.60 5.72
E-15 1211.4 97.4 12.9 45.8% 151.46 1.65 0.36 6.58 5.33
E-16 1235.1 98.9 13.8 45.2% 172.67 1.69 0.37 6.48 5.38
E-17 1241.9 89.9 14.8 43.4% 180.87 1.64 0.40 6.57 5.47

CHO: Carbohydrate.

gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW, and 3.8 g.L-1 of CaCO3, for an
initial carbohydrate concentration of 10.12 ± 0.48 g.L-1, and a gCaCO3/
gCHO ratio of 0.38 ± 0.01. For assay E-14, performed at the same level
for the substrate variables (22.5 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of
CPWW) as those central point assays, very close H2 production was
observed, 1224.50 mLH2. L-1. However, it should be noted that for E-14,
a higher calcium carbonate concentration was applied (7.33 g.L-1),
resulting in a CaCO3/CHO ratio of 0.67.
Based on these considerations, two important points can be
observed: the first is related to the buffering concentration; that is,
values above 3.8 g.L-1 did not result in higher P for the same substrate
concentration, but also did not cause significant negative effects on H2
production. The second point refers to the perception that intermediate
substrate concentrations (22.5 gTVS.L-1 of CPW, 4.5 gCOD.L-1) resulted
in a better process adjustment, considering the higher H2 production in
Fig. 1. Accumulated H2 production fitted to the modified Gompertz model, for the central point assays. Similar to what was observed in the present
factorial design assays. study, Pu et al. [59] highlighted that low substrate concentrations do not
provide enough organic matter for hydrogen fermentation, while very
high concentrations imply changes in metabolic pathways and lactic
Table 4 fermentation.
ANOVA for the effects of CaCO3, CPW, and CPWW at maximum production Harmful effects to the process directly associated with the high cal­
potential (P) for a 90% confidence interval. cium carbonate concentration (7.33 g.L-1) were not observed, this fact is
Source SQ DF QM F Value P Value probably related to the low solubility and slow dissolution of this buff­
Model 1230063 9 136674 3.23 0.068 ering agent in the liquid medium, and therefore high CaCO3 concen­
X1 (L) 103317 1 103317 2.44 0.162 trations do not result in too much increase in pH, but in a buffering
X2 (L) 160165 1 160165 3.78 0.093 capacity increase [60]. Otherwise, some calcium carbonate concentra­
X3 (L) 113194 1 113194 2.67 0.146 tions would lead to the process establishment above the upper limit of
X1 (Q) 433292 1 433292 10.24 0.015
X2 (Q) 278015 1 278015 6.57 0.037
the ideal pH range for hydrogen production (pH from 4.5 to 6.5) [61].
X3 (Q) 391187 1 391187 9.24 0.019 Consistent with this behavior described for carbonate, in the present
X1X2 561 1 561 0.01 0.912 study, despite the wide range of calcium carbonate concentration
X1X3 51061 1 51061 1.21 0.308 applied to the reactors (0.3–7.3 g.L-1), no impact was observed on the
X2X3 101387 1 101387 2.40 0.166
initial pH of the assays, which were close to 6.5 (Table 3). The percep­
Erro 296226 7 42318
Total SQ 1526289 16 tion of a reduction in the stability of maintaining the H2 production in
the assays conducted at the upper level for the substrate variables is
associated with the imbalance in the kinetics of the calcium carbonate
fit, as discussed by Reungsang et al. [56], Niladevi et al. [57], and dissolution processes and organic acid production; therefore, a high
Kainthola et al. [58]. organic acid production imposes the need for high calcium carbonate
The interaction and optimal level of independent variables in dissolution [62], which sometimes cannot be achieved.
hydrogen production can be observed on the 3D response surface The satisfactory adjustment of the variables for the assays in which
(Fig. 2). The surfaces approximate the characteristic of an elliptical the highest P values were obtained is confirmed by the response surface,
hyperboloid, and in this case, a maximum point can be identified within so that the point of H2 production optimization was verified in the
the range of values evaluated for each variable, which represent the x central region of the surface, for the condition with 21.4 gTVS.L-1 of
and y axes in the three-dimensional plane. Therefore, an optimized CPW, 3.4 gCOD.L-1 of CPWW, and 4.50 g.L-1 of CaCO3; while for points
condition for H2 production can be identified. located towards the surface peripheries, experimental conditions with
The highest H2 production was verified for the central point assays gradual reduction for maximum production potential were observed. As
(E-15/E-16/E-17), of 1229.45 ± 14.1 mLH2. L-1, performed with 22.5 obtained in the present study of co-fermentation of citrus waste and

5
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

Fig. 2. Response surface for H2 production model with (a) fixed CaCO3 concentration 3.8 g.L-1; (b) fixed CPWW concentration 4.5 gCOD.L-1; (c) fixed CPW 22.5
gTVS.L-1.

6
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

wastewater, Liu et al. [63], who evaluated the co-digestion of food waste production was associated with the behavior observed for the opera­
and cattle manure, observed a significant reduction in the H2 production tional pH, which after a reduction in the first hours of the process (10 –
when the substrate concentration differed from the concentration values 15 h) starting from an initial pH of 6.5 and reaching values close to at
of the optimized condition. 5.95, a slight reduction was observed, reaching values close to 5.5 at the
The lowest H2 production (199.21 mLH2. L-1) was observed in assay end of the assay (~40 h).
E-13 (22.5 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW, and 0.3 g.L-1 of As can be seen on the response surface (Fig. 2a), hydrogen produc­
CaCO3). The concentrations of CPW and CPWW in this assay are the tion was significantly increased when a better fit was established be­
same as those for assays E-14 and the central point (E-15/E-16/E-17), in tween substrate variables X1 and X2 (CPW and CPWW). In the peripheral
which the highest P values were observed. However, in E- 13, only 0.3 g. areas of the surface, referring to conditions that simultaneously have
L-1 of CaCO3 was added, consequently resulting in a CaCO3/CHO ratio of high concentrations of CPW (≥ 30.0 gTVS.L-1) and CPWW (≥ 7.0 gCOD.
0.03. From the antagonistic behavior to the H2 production in this assay, L-1) or even conditions with low concentrations for one of these sub­
the fundamental importance of alkalinization for maintaining good strates, did not result in a favorable response to hydrogen production.
fermentation stability is highlighted. Similar results were verified by Under conditions with addition of 30.0 gTVS.L-1 of CPW and 7.0 gCOD.
Zhang et al. [64], who evaluated the alkalinity effect using lime mud L-1 of CPWW, such as E-4 and E-8, the H2 production was 578.6 mLH2. L-
(1.0–4.0 g.L-1), for fermentation in batch reactors with food waste, and 1
and 493.8 mLH2. L-1, respectively.
identified that the decrease in buffering agent concentration resulted in Regarding the effect of adjusting substrate proportions in co-
a significant impact on hydrogen production. As observed in the present digestion for hydrogen production, Li et al. [66] evaluated the
study, lower stability was observed due to a more acidic pH (4.0–5.0). co-digestion of food waste with sewage sludge and observed that the H2
Through the response surfaces (Figs. 2b and 2c), there was a marked production was negatively impacted as the proportion between these
P reduction, mainly for calcium carbonate concentrations in the range substrate concentrations was altered. This was similar to the results of
from 0.3 to 1.7 g.L-1, combined with substrate concentrations ≥ 30.0 g. the present study, in which it was verified that the simultaneous increase
L-1 of CPW, and ≥ 7.0 gCOD.L-1 of CPWW. For the condition with a low in the concentration of both substrates (CPW and CPWW) resulted in
calcium carbonate concentration (1.7 g.L-1), however, with intermedi­ negative effects on H2 production.
ate values for initial carbohydrate concentration (5.0–7.0 gCHO.L-1, It is observed from these results that there are no adverse effects in
Fig. 3), there was no sharp reduction in H2 production as the combi­ the co-use of these substrates, since for duly adjusted concentrations
nation of variables at these levels did not result in a sufficiently low were obtained high H2 production. The negative results observed in
gCaCO3/gCHO ratio. The importance of the buffer:substrate ratio for some assays were associated with the lack of adjustment in their pro­
fermentation was also verified by Sangyoka et al. [65], using sodium portions, for example, for the assay E-7 performed at the upper level for
bicarbonate as a buffering agent in a study with sugarcane bagasse in CPWW concentration (7.0 gCOD.L-1) but at the lower level for CPW
batch reactors. The authors identified a reduction in the hydrogen po­ concentration (7.0 gTVS.L-1), the production was of only 423.4 mLH2. L-
1
tential in assays with low values for the buffer:substrate ratio (<0.15), . Another important aspect of the co-digestion is that the optimized
similar to the present study, and identified an optimized condition with condition obtained ensures the use of significant concentrations for both
a buffer: substrate ratio of approximately 0.25. substrates (21.4 gTVS.L-1 of CPW, 3.4 gCOD.L-1 of CPWW), postulating
The effect of this adjustment between concentrations of CaCO3, CPW, how a concrete alternative for the treatment of both. Considering the
and CPWW could be seen in assays E-1 and E-3, which were performed specific generation data of CPW and CPWW in the industry per ton of
with a low CaCO3 concentration (1.7 g.L-1), however with a low initial fruit processed (400 kg of CPW in natura [7], and 1 – 3 m3 of CPWW [8,
substrate concentration in comparison to other assays (5.25 gCHO.L-1 12]), it is estimated that for every 1 kg of CPW (dry basis, moisture
and 7.12 gCHO.L-1, respectively), resulting in a CaCO3/CHO ratio of <20%), there is a generation of 10–30 liters of CPWW. For the optimized
0.30. Therefore, there was no marked reduction in hydrogen production experimental conditions obtained, for each 1 kg of CPW (dry basis)
(823.72 mLH2. L-1 for E-1, and 898.80 mLH2. L-1 for E-3) when applied in the reactor, there is a demand of 18.7 liters of CPWW (based
compared to the central point assays, in which it obtained the highest on the COD of the wastewater of the study). Thus, it is verified that the
production. process of co-digestion of these substrates is also feasible from a quan­
The process stability in the assays with the best responses to the H2 titative point of view, consistent with the average of CPW and CPWW
generation.
The H2 percentage (%H2) in the biogas composition between 30.2%
and 51.0% was identified, which was an important variation, making it
possible to infer that different experimental conditions significantly
influenced the biogas quality. The variables of CPW concentration (X1)
and CaCO3 concentration (X2) were identified as statistically significant
(p = 0.001 and p = 0.008, respectively) to %H2. Assays E-2 (30.0 gTVS.
L-1 of CPW, 2.0 gCOD.L-1 of CPWW, and 1.7 gTVS.L-1 of CaCO3) and E-8
(30.0 gTVS.L-1 of CPW, 7.0 gCOD.L-1 of CPWW, and 5.9 g.L-1 of CaCO3)
were those in which the lowest H2 percentage was found (30.2% and
30.5%, respectively), while in E-3 (15.0 gTVS.L-1 of CPW, 7.0 gCOD.L-1
of CPWW, and 1.7 g.L-1 of CaCO3) and E-13 (22.5 gTVS.L-1 of CPW, 4.5
gCOD.L-1 of CPWW, and 0.3 g.L-1 of CaCO3) were those with the highest
%H2 (51.0% and 48.3%).
For the two assays with the highest %H2, distinct patterns were
identified, the first with the higher volumetric production, as verified in
E-3, and therefore adequate fermentation stability with the predomi­
nance of H2 production pathways. While, for assay E-13, the high %H2 in
the biogas despite the low P occurred due to the lower CaCO3 addition
(0.3 g.L-1) compared to the other assays, resulting in a reduction in the
total CO2 release through the carbonate system. The neutralization re­
Fig. 3. Carbohydrate concentration at initial and final, and carbohydrate action of the acids formed in fermentation, through the carbonate sys­
removal from experimental design assays. tem, results in CO2 release to the gaseous phase of the reactor, as can be

7
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

seen in Eq. (8), taking as an example the neutralization reaction of acetic in the metabolic reactions of the glucose to acetic acid conversion
acid [62], so that higher calcium carbonate availability in the liquid pathway [70]. The yields of some assays were high, such as E-1 (2.06
medium enables higher CO2 release. Lin and Lay [67] point out that molH2.mol-1hexose) and E-9 (2.13 molH2.mol-1hexose), considering the
carbonate correlates with CO2, therefore the increase in carbonate practical limitations to reaching the theoretical maximum, as presented
concentration tends to provide a higher CO2 percentage in biogas. by Thauer et al. [70]. Lee et al. [71] infer that the maximum yields
observed in fermentation studies have been up to 2.8 molH2.mol-1hex­
2CH 3 COOH + CaCO3 ⇆Ca2+ + 2CH 3 COO− + H2 O + CO2 (8)
ose, which is about 70% of the theoretical yield, similar to the highest
yields obtained in the present study. As observed, it is highlighted that a
3.2. Substrate utilization, limonene removal, and hydrogen yield significant variation in Y2 is often verified due to different experimental
conditions evaluated, showing evidence of the importance of carrying
Carbohydrate removal in the assays ranged from 29.25% (E-13) to out an adequate adjustment of significant variables. Akhlaghi et al. [72],
88.31% (E-7). As can be seen in Fig. 3, a significant difference in this for example, obtained H2 yields from 0.36 to 1.45 mol H2.mol-1hexose,
parameter indirectly allows us to evaluate the stability and efficiency of and Gioannis et al. [73] obtained Y2 from 0.04 to 2.6 mol H2.
the fermentation process through the substrate utilization. However, the mol-1lactose.
low conversion observed in E-13 was an exception among the experi­ The limonene concentration was determined for initial and final
mental conditions (Fig. 3), and except for this condition, the average samples of the fermentation assays in batch reactors. As can be seen in
carbohydrate removal of the other assays was 81.4% ± 5.7%. High Fig. 4, the initial limonene concentration was generally less than 15 mg.
carbohydrate removal (88.3% for E-7% and 81.5% for E-8) was observed L-1. A significant difference was observed between the initial and final
even for conditions with reduced H2 yield (0.61 mol H2.mol-1 hexose for limonene concentration in the assays, with a maximum removal of
E-7 and 0.53 mol H2.mol-1 hexose for E-8). 96.7% in assay E-8 and a minimum of 9.9% in E-1, which represented a
The positive response to the carbohydrate removal, even for assays removal of 9.0 mg.L-1 and 0.13 mg.L-1, respectively. Considering the
with high initial substrate concentration, was directly associated with limonene concentrations reported by Camargo et al. [74], which resul­
CaCO3, since its gradual solubilization throughout the assay ensured the ted in negative effects for P and λ, it is unlikely that there was inhibition
pH maintenance in a range that did not limit the continuation of the or negative effects on P related to the presence of this compound at the
process. In some hydrogen production studies that used only the initial concentrations observed in the present study, lower than 15 mg.L-1.
pH correction as a process parameter, they obtained carbohydrate Camargo et al. [74] evaluated the effect of different limonene concen­
removal lower than those obtained in the present study, even starting trations on the kinetic parameters of H2 production in batch reactors and
from an initial carbohydrate concentration in a lower range than that found that for concentrations of up to 100 mg.L-1, there was no signifi­
evaluated for citrus waste. For example, Villa Montoya et al. [36], in a cant inhibitory effect for P. However, a statistical difference was
study on co-digestion of coffee waste in batch reactors, verified carbo­ observed in the values of λ for assays with limonene concentration above
hydrate removal of 60.4% ± 11.3%, in an initial pH range from 4.8 to 50.0 mg.L-1.
8.5. Camargo et al. [35] evaluated the citrus waste fermentation in batch In the batch assays of the present study, the limonene removal likely
reactors and verified a conversion of only 46.8% ± 12.2%, with an did not occur via degradation, considering the short duration of the
initial pH variation between 4.5 and 8.5. In both cases, initial pH values assays. Calabrò et al. [17] identified the D-limonene degradation in
in the neutral or alkaline range were insufficient to ensure high carbo­ batch reactors only after the 15th day of digestion, detecting the pres­
hydrate removal. ence of degradation by-products such as p-cymene. The limonene
The H2 yield (Y1) for the assays was calculated indirectly based on adsorption in the sludge was not evaluated as this mechanism has not
substrate concentration and cumulative H2 production and ranged from been observed as significant in the removal of this compound in
53.5 mLH2.g-1CHO to 214.4 mLH2.g-1CHO. In the assays with the anaerobic reactors [74]. The adsorption process in the solid phase of
highest P, intermediate yields were obtained, whose values were 168.3 citrus waste is a likely mechanism that may have contributed to the
mLH2.g-1CHO (E-15/E-16/E-17) and 144.4 mLH2.g-1CHO for E-14. limonene removal from the liquid medium. Calabrò et al. [17] evaluated
For assays E-1 (15.0 gTVS.L-1 of CPW, 2.0 gCOD.L-1 of CPWW, and the limonene content in anaerobic digestion in batch reactors and
1.7 g.L-1 of CaCO3) and E-9 (9.9 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of identified adsorption of up to 76% in the solid phase of citrus waste,
CPWW, and 3.8 g.L-1 of CaCO3) yields of 214.4 mLH2.g-1CHO and 209.3 pointing out that this mechanism plays an important role in reducing the
mLH2.g-1CHO, respectively, were observed. These assays have the limonene content in the liquid phase of the reactors.
lowest initial substrate concentrations in common among all assays
evaluated (5.25 and 5.74 gCHO.L-1, respectively). Therefore, it was
verified that lower substrate concentrations combined with the CaCO3/
CHO ratio > 0.3 were favorable to obtain the highest H2 yields.
The lowest yields found were 53.5 mLH2.g-1CHO for E-8 (30.0 gTVS.
L of CPW, 7.0 gCOD.L-1 of CPWW, and 5.9 g.L-1 of CaCO3), and 62.7
-1

mLH2.g-1CHO for E-10 (35.1 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW,


and 3.8 g.L-1 of CaCO3), assays which have in common the highest initial
soluble carbohydrate concentrations among all assays (12.8 and 13.5
gCHO.L-1, respectively). The high substrate concentration is likely to
have favored a higher organic acid production rate leading to a modi­
fication in the predominant metabolic pathways, which is unfavorable
for hydrogen production. In other studies, an increase in the H2 yield has
also been reported as the substrate concentration was adjusted, with a
reduction in this yield for substrate concentrations that resulted in the
process overload, causing organic acid accumulation and consequently a
marked pH reduction [68,69].
Considering the hypothesis that the initial soluble carbohydrates
were composed only of hexoses, H2 yields (Y2) with values between 0.53
and 2.13 molH2.mol-1hexose were obtained representing 13.3–53.2% of
the theoretical maximum yield, which is 4 molH2.mol-1hexose obtained Fig. 4. Limonene concentration at initial and final, and limonene removal.

8
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

3.3. Volatile fatty acids throughout this assay infers that the production of HBu and H2 occurred
from the HLa consumption (Eq. 9), highlighting that this metabolic
The analysis of the distribution of soluble metabolites in the assays pathway is thermodynamically favorable (ΔG◦ =− 64.1) [77,78]. Meta­
was carried out for complementary interpretation with the hydrogen bolic pathways of lactic acid utilization during the fermentation were
production data and possible understanding of the predominant meta­ investigated by Baghchehsaraee et al. [79] in batch reactors using starch
bolic pathways in the assays of the experimental design. Acetic (HAc), as substrate. The authors observed that the HLa availability in the
butyric (HBu), propionic (HPr), lactic (HLa), and formic (HFo) acids fermentation medium resulted in a significant increase in hydrogen
were observed in at least one of the assays (Fig. 5). production, associated with an increase in the butyric acid concentra­
As can be seen in Fig. 5 concerning the predominant metabolites at tion. These results are similar to those observed in assay E-6, in which
the end of the fermentation process, lactic acid (HLa) was observed as partial consumption of the HLa produced occurred concomitantly with
the main metabolite in assays A-4 with 4302.1 mg.L-1 (representing the increase in HBu concentration, maintaining the H2 production.
64.1% of total), A-6 with 3333.5 mg.L-1 (60.3%), A-7 with 3926.3 mg.L-
1 2Lactate + H + butyrate + 2H2 + 2CO2 (9)
(70.5%), A-8 with 7312.2 mg.L-1 (79.6%), A-10 with 6367.8 mg.L-1
(81.9%), A-12 with 6247.5 mg.L-1 (78.0%) and A-13 with 3992.6 mg.L-1 The metabolite profile of the assays with the highest hydrogen pro­
(91.5%). However, for assay A-1, a predominance of acetic acid was duction (E-14 and central point) was characterized by the prevalence of
identified with 1897.3 mg.L-1 (63.6%); and for assays A-2, A-3, A-5, A-9, acetic and butyric acids, totaling around 95% of the metabolites at the
A-11, A-14, and central points, acetic and butyric acids were observed as process end. Substrate conversion probably occurred mainly through the
the main metabolites, each one representing from 40% to 50% of the Clostridial pathway, which represents an advantage for higher H2 pro­
total (Fig. 5). duction per mole of carbohydrate consumed [80].
For assays with HLa predominance, a correspondence pattern with In the central point assays (E-15/E-16/E-17), the concentrations of
another important process response, which is hydrogen production, was HAc and HBu at the process end were 1590.1 ± 11.3 mg.L-1 and 2153.7
observed, except for E-6. The H2 yield for assays A-4, A-7, A-8, A-10, A- ± 156.0 mg.L-1, respectively, and represented 40.1% (HAc) and 54.2%
12 and A-13 was only 0.63 ± 0.06 molH2.mol-1hexose, consistent with (HBu) of the final composition of soluble metabolites. This characteristic
what was expected considering the biochemical reactions of lactic acid was similar to that observed for assay E-14 (1684.1 mg.L-1 of HAc and
production, in which the carbohydrate molecules catabolism does not 2199.8 mg.L-1 of HBu), in which the H2 production was similar to the
involve the direct H2 production [75]. central point assay. However, for E-1 (1897.3 mg.L-1) and E-4
Among these assays with HLa predominance, for E-8, E-10, and E-12, (1785.2 mg.L-1), the HAc concentrations identified were similar or even
higher concentrations of this metabolite were observed, of 7.3 g.L-1, higher than the assays of higher hydrogen production (E-14, E-15, E-16,
6.4 g.L-1, and 6.3 g.L-1, respectively. In these assays, the combination of and E-17). This aspect was not observed for butyric acid, whose con­
high concentrations of both substrate (12.8 gCHO.L-1, 13.5 gCHO.L-1, centration was notably higher in E-14, E-15, E-16, and E-17.
and 10.8 gCHO.L-1, respectively) and CaCO3 (5.9 g.L-1, 3.8 g.L-1, and For assays E-3, E-5, E-9 and E-11, the HBu concentration (1305.6 mg.
3.8 g.L-1, respectively) favored lactic bacteria populations impacting L-1, 1728.8 mg.L-1, 1776.4 mg.L-1 and 1240.0 mg.L-1, respectively) was
hydrogen production under these conditions. However, for E-13, despite intermediate, as was observed for hydrogen production in these assays,
the high HLa proportion to the total metabolites (91.5%), its concen­ close to 900 mLH2. L-1. Therefore, a higher correlation between H2
tration was not observed as high (4.0 g.L-1) as in E-8, E- 10, and E-12. production and HBu predominance was identified.
The low CaCO3 concentration added to E-13 (0.3 g.L-1) was insufficient Matyakubov et al. [81] verified that the HBu/HAc ratio has an
to buffer the process, resulting in a significant pH reduction and important correlation with the H2 yield. A relationship between 0.3 and
consequent inhibitory effects on fermentation as the pH final was only 1.5 was observed in the present study for citrus waste. Thus, the assays
3.52, the lowest of all assays. As observed in the present study, Tang with the highest values for HBu/HAc (>0.80) were those with the
et al. [76] identified that the prevention of a sharp reduction in the highest P. For example, an HBu/HAc ratio of 1.35 was obtained for
operational pH in the assays made it possible to obtain higher HLa central point assays, while for E-4, with P of 578 mLH2. L-1, this ratio was
concentrations, avoiding pH values < 4.0 that were inhibitory to the 0.30.
growth and activity of microorganisms.
For assay E-6, unlike what was identified for the other assays whose 3.4. Validation of the hydrogen production model
HLa production was high, there was no correspondence with low H2
production as an intermediate H2 yield was obtained (1.17 molH2. mol- The adjustment provided by the experimental design and the
1
hexose) for this assay. It is likely that the intermediate H2 yield ob­ response surface model led to an optimized P of 1249.04 mLH2. L-1, and
tained, despite the predominance of HLa production, is related to H2 a maximum production rate (Rm) of 89.60 mLH2. L-1.h-1, with time to
production via lactic acid oxidation pathways. The acid profile start production of 8.56 h (Fig. 6), in an experimental condition with
21.4 gTVS.L-1 of CPW, 3.4 gCOD.L-1 of CPWW, and 4.5 g.L-1 of CaCO3
(Table 5). The hydrogen production obtained is 5.3 times higher (530%)
than the H2 production verified in assay E-13 (199.21 mLH2. L-1), for
example, in which the lowest H2 production was obtained.
In the experimental condition of the optimized assay (OC-H) was
obtained a gCaCO3/gCHO ratio of 0.53; this value is an important
guideline about alkalinization for fermentation focused on the H2 pro­
duction from citrus waste. A sharp decrease in pH (from 6.62 to 5.65)
between the start and the exponential phase of H2 production
(~10–12 h) was verified (Fig. 6). However, until the process ended
(~42 h), there was a low pH variation (from 5.65 to 5.51), even with
high production of H2 and organic acids. Therefore, adjusting the con­
ditions allowed sufficient buffering for the process to occur stably, even
with a high initial substrate concentration (8.4 gCHO.L-1). For OC-H was
obtained 86.6% of carbohydrate removal and H2 yield of 195.07 mLH2.
g-1CHO or 2.07 molH2.mol-1hexose.
Fig. 5. Organic acid distribution at the end of the experimental design assays. The H2 percentage in the biogas composition of the optimized

9
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

its use, but also for transport and storage [82,83].


Based on the metabolite distribution profile throughout the process,
it was found that acetic and butyric acids were the metabolites at the
highest concentration throughout the entire assay; however, changes to
their proportions in the total metabolite composition were observed
(Fig. 7). It is noteworthy that after 30 h, an increase in the HBu con­
centration was observed at a higher proportion than that observed for
HAc, in which butyric acid was the soluble metabolite at the highest
concentration at the assay end (2230.8 mg.L-1), while the acetic acid
concentration was 1213.6 mg.L-1, representing proportions of 60.8%
and 33.1% of the total soluble metabolites, respectively.
For OC-H, an HBu/HAc ratio of 1.87 was obtained. This value is
higher than that verified for all assays previously performed in the
factorial design, and consistent with what is described in the literature
[81]; that is, a higher HBu/HAc ratio is associated with a higher H2
production.
As can be seen in Table 6, H2 production from citrus waste is higher
than that obtained by other studies that evaluated co-digestion, for
example, compared to Villa Montoya et al. [36] when evaluating the
Fig. 6. Accumulated H2 production fitted to the modified Gompertz model, for
validation assay. co-digestion of solid waste and wastewater from coffee processing. The

Table 5 Table 6
Experimental condition and main parameters for the validation assay of the Comparison of optimized H2 production from citrus waste with H2 production by
optimized condition of H2 production. other lignocellulosic substrates.

Experimental Condition Substrate Optimized Conditions H2 Reference


production*
-1
CPW 21.4 gTVS.L Predict Value
CPWW 4.5 gCOD.L-1 P (mLH2. L-1) Solid waste and 21.4 gTVS.L-1 of CPW, 4.5 1249.1 mL.L- This study
Calcium Carbonate 3.4 g.L-1 1287.5 wastewater from gCOD.L-1 of CPWW, and 1

Parameters of the validation assay citrus processing 3.4 g.L-1 of CaCO3


P (mL.L-1) 1249.04 ± 40.1 Pulp/Husk and 7.0 g.L-1 of coffe pulp and 240 mL [36]
Rm (mL.L-1.h-1) 89.60 wastewater from husk, 30% headspace, (686 mL.L-1)
λ (h) 8.56 coffe processing initial pH of 7.0
%H2 47.5% Citrus Peel Waste 4 gTVS.L-1 of sludge, 29.8 48.47 mmol. [35]
Y1 (mLH2.g-1CHO) 195.07 gTVS.L-1 of CPW, and L-1
Y2 (molH2.mol-1 hexose) 2.07 initial pH of 8.98 (1085 mL.L-
1
Final pH 5.51 )
gCaCO3.g-1 CHO 0.53 Sugarcane bagasse 2.77 g.L-1 of yeast extract 1.5 mmol.L-1 [25]
Carbohydrate Removal (%) 86.6% and 5.84 g.L-1 of (36 mL.L-1)
Predominant Organic Acid HBu sugarcane bagasse
HBu/HAc 1.84 Sugarcane bagasse 7.0 g.L-1 of sugarcane 23.1 mmol.L- [34]
1
bagasse, initial pH of 7.2
-1
(517 mL.L )
Banana waste 5.0 g.L-1 of substrate, 70.2 mL [37]
condition was 47.5%. This H2 percentage is important for equating the
temperature of 37 ◦ C, (468 mL.L-1)
technical and economic viability of energy recovery from the biological initial pH of 7.0
H2 production as the biogas obtained in the fermentation is not feasible *
values in parentheses were approximately based on data provided by the
for its direct use, due to the presence of other gases. Moreover, hydrogen
authors and using the ideal gas conversion of 1 mol = 22.4 liters.
separation/concentration processes are fundamental not only to enable

Fig. 7. Temporal profile for organic acid distribution in the validation assay.

10
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

results of H2 production obtained in the present study are also higher Chloroflexi with abundance < 0.01%. Regarding E-10 Proteobacteria
than those of studies that used the mono-digestion of agro-industrial (5.9%), Halobacterota (0.2%), Actinobacteriota (~0.1%), Desulfo­
waste as an alternative, such as Mazareli et al. [37] with banana bacterota (~0.1%), and Chloroflexi, Spirochaetota, and Synergistota
waste, and Soares et al. [25] and Rabelo et al. [34] with sugarcane with relative abundance < 0.01%. The abundance observed for the
bagasse. Firmicutes phylum for the assays OC-H and E-10 is in line with what has
The technical viability of the co-digestion of wastewater and solid been reported in several studies applied to the various substrates
waste from citrus processing can be inferred considering its H2 pro­ fermentation [85–87], characterized by bacteria with a wide physio­
duction compared with mono-digestion assays of citrus waste. For logical variety in the organic matter fermentation [88].
example, Camargo et al. [35] obtained an H2 production of 1085 mLH2. For the OC-H assay, bacteria belonging to the Clostridiaceae family
L-1; that is, 15.0% lower than that obtained in the present co-digestion were identified in high abundance, with a relative abundance of
study. Therefore, from the point of view of H2 production, and also 96.15%, and Enterobacteriaceae (3.0%) and Peptostreptococcaceae
based on other process parameters, such as carbohydrate removal, and (0.6%) were identified with lower abundance. However, for E-10, bac­
%H2 in biogas, the co-digestion proposal is technically feasible as it teria belonging to three families of the Firmicutes phylum were identi­
enables important energy content recovery using two residual flows fied, such as Clostridiaceae (51.6%), Lactobacillaceae (21.9%), and
from the same productive chain. Another disadvantage to the Peptostreptococcaceae (20.0%); in addition to Enterobacteriaceae
mono-digestion of CPW is the water demand, which in some situations (5.58%), Moraxellaceae (~0.2%) and Methanosaetaceae (~0.2%), in
can be considered a bottleneck for the process viability. Among the H2 lower relative abundance. A significant reduction in the relative abun­
production routes, dark fermentation is one of the most water-intensive, dance of the Clostridiaceae family in assay E-10 was observed, compared
with a high water footprint [84]. Co-digestion with processing waste­ to the assay OC-H (Fig. 8), from 96% to 50%, respectively. Combined
water allows for a change in the approach to water footprint analysis for with this abundance reduction, an increase in the relative abundance of
this H2 production route, with a reduction in the water footprint. Under the Peptostreptococcaceae family was observed in E-10.
the optimized conditions of the present study, wastewater meets about Regarding the genera, for the optimized condition assay, only two of
40% of the process water demand; considering organic matter concen­ them were found with relative abundance > 1%, Clostridium sensu stricto
tration of the CPWW and OC-H assay concentration of 3.4 gCOD.L-1. 1 (95.0%) and Enterobacter (3.0%), and other genera with relative
abundance < 1%, such as Paraclostridium (0.54%), Clostridium sensu
3.5. Microbial community characterization stricto 2 (0.34%), Clostridium sensu stricto 10 (0.10%), and Hathewaya
(0.14%). For assay E-10, Clostridium sensu stricto 1 was also identified as
Microbial community characterization was carried out to identify the the one with the highest relative abundance (51.2%), in addition to
predominant microorganism groups in the reactors of the selected as­ Paraclostridium (20.0%) and Lactobacillus (21.9%). This change in these
says, helping to infer the metabolic profile of the process. For this pur­ genera abundance can be pointed out as the most important change in
pose, samples from two reactors with antagonistic results for the main the structure of the E-10 microbial community compared to assay OC-H.
responses of the process, which are the production of H2 and soluble Other genera such as Enterobacter (4.7%), Kosakonia (0.6%), Clostridium
metabolites, were analyzed. sensu stricto 18 (0.4%), Klebsiella (~0.2%), and Acinetobacter (~0.2%)
The first sample selected was from the optimized condition valida­ were also identified, however at low relative abundances in both assays
tion assay reactor - OC-H (21.4 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW, (E-10 and OC-H) (Fig. 8).
and 3.4 g.L-1 of CaCO3), with the highest P, and with HAc and HBu as the The granular sludge from a mesophilic UASB reactor for wastewater
predominant soluble metabolites. The second sample was from the E-10 from a poultry slaughterhouse treatment used as inoculum for the assays
assay reactor (35.1 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW, and 3.8 g. is characterized by a mixed culture with a diversity of bacteria and
L-1 of CaCO3), with lower P and Rm, and absolute HLa predominance
among the soluble metabolites.
For the OC-H and E-10 assays samples, 45,813 and 37,906 16 S rRNA
gene sequences were obtained, grouped into 19 ASV and 33 ASV
(Amplicon Sequencing Variants), respectively. The similarity obtained
between the samples was only 26.8%, and probably the different
nutritional conditions resulted in a significant difference in the bacterial
population’s distribution in the batch reactors.
The microbial diversity in the samples was evaluated based on the
Shannon-Wiener and Simpson ecological indices. Significantly higher
diversity (1.30) and lower dominance (0.34) were observed at E-10;
while in the optimization assay (OC-H), were 0.30 and 0.90, respec­
tively. It is likely that the statistical adjustment of the variables in the
assay OC-H to establish a condition with maximum hydrogen production
potential with operational pH maintenance in a stable range from 5.5 to
6.5 throughout the fermentation, provide a favorable environment to
the development and establishment of specific populations, without
creating transitory environments to favor other bacteria groups, thus
resulting in lower diversity and higher dominance. Based on the
Shannon-Wiener index, Camargo et al. [35] also identified a reduction
in diversity for the sample of an optimized condition reactor for H2
production from citrus waste, attributing this fact to the population’s
selection due to the process variable optimization to favor the H2
production.
For both samples, from the OC-H and E-10 assays, bacteria belonging
to the Firmicutes phylum were identified as the ones with the highest
relative abundance, respectively 96.9% and 93.7%. For assay OC-H, the Fig. 8. Taxonomic groups for bacteria Genus, for assays Optimized condition
other phyla identified were Proteobacteria (3.0%), Halobacterota, and (OC-H) and E-10.

11
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

archaea [89], and therefore, the strategy of application of the heat negatively affected the H2 production, and resulted in lactic acid
treatment was adopted as a way to establish a selective pressure, fa­ accumulation.
voring endospore-forming microorganisms in response to environ­ Paraclostridium identified in assay E-10 has been commonly identi­
mental stress, such as the genus Clostridium sp. [90]. However, even fied in hydrogen production reactors [103] and is capable of using a
starting from the same inoculum, the different nutritional conditions wide variety of organic substrates [104] in addition to its hydrolytic
applied to the assays OC-H and E-10 resulted in the establishment of metabolism [105].
microbial communities with important differences in the relative It is also noteworthy that for both assays, Enterobacter was identified
abundance of the main genera identified in the reactors. with relative abundance between 3% and 5%. Bacteria similar to this
Bacteria groups similar to Paraclostridium and Clostridium sp. are genus are characterized by rapid growth using various substrates [106],
characterized by endospore formation [91]. This characteristic was and robustness to pH variation in fermentation reactors [107]. The
likely important so that under favorable conditions to their develop­ Enterobacter metabolism is also characterized by lactic acid production.
ment, as was assay E-10, these genera could play a fundamental role in For example, Thapa et al. [108] verified high levorotatory lactic acid
the microbial community structure, which was identified with high production (L-lactic acid) (46.0 g.L-1) by Enterobacter sp. in batch re­
abundance. Another microorganism group with important relative actors with synthetic medium and various carbohydrates (glucose,
abundance in E-10 was Lactobacillus, which is endospore negative [92]. mannitol, sucrose, fructose).
The origin of these bacteria is likely from autochthonous sources of solid Paraclostridium and Enterobacter have metabolic characteristics
waste [93] and wastewater from citrus processing [94]. similar to those of other genera identified with high abundance in E-10
Clostridium sensu stricto 1 mainly uses carbohydrates and amino acids and OC-H assays, such as Clostridium sensu stricto 1 and Lactobacillus,
as a substrate, with butyric acid as its main metabolite product [95]; resulting in a soluble metabolite distribution with a restricted diversity
furthermore, this genus is widely reported as a notable hydrogen pro­ of organic acids, basically with HAc and HBu in OC-H, and HLa in E-10.
ducer, capable of using other types of more complex substrates such as The assessment that there was not a wide metabolic diversity is an
food waste, starch, cellulose [96]. These metabolic characteristics are interesting aspect of the H2 optimization assay (OC-H) considering the
consistent with the responses observed for the H2 optimization assay objective of directing the substrate utilization towards pathways of
(OC-H), with a P of 1249.1 mL.L-1, and 2230.8 mg.L-1 of HBu. Lower higher H2 yield, as was observed.
process pH stability in E-10 due to a lower gCaCO3/gCHO ratio Prediction of the metabolic pathways involved in the citrus waste
compared to OC-H, given the high CPW concentration (35.1 gTVS.L-1) fermentation was performed for assays OC-H (4.5 of gCaCO3. L-1, 21.4
for 3.8 g.L-1 of CaCO3, resulted in a reduction of Clostridium sensu stricto gTVS.L-1 of CPW, 3.4 gCOD.L-1 of CPWW) and E-10 (3.8 of gCaCO3. L-1,
1 abundance in this assay, which is likely to be associated with lower 35.1 gTVS.L-1 of CPW, 4.5 gCOD.L-1 of CPWW) (Fig. 9). The metabolic
production of H2 (507.1 mL.L-1), acetic and butyric acids (867.1 mg.L-1 prediction was carried out based on the main genera (Clostridium,
and 393.5 mg.L-1, respectively). Paraclostridium, Lactobacillus, and Enterobacter) and the main meta­
Luo et al. [95] point out that the H2 production, and HBu and HAc bolic products identified (acetic, butyric, lactic, propionic acids, and
concentrations, were positively correlated with Clostridium sensu stricto 1 hydrogen).
in batch reactors with food waste. Similarly, Mazareli et al. [97] For assay OC-H, the hydrogen production was associated with in­
observed an important abundance of Clostridium sp. (~40%) in termediate reactions of the pathways for the formation of acetic and
fermentation assays in batch reactors using banana waste for producing butyric acids, with no inference for the occurrence of lactic acid pro­
H2, acetic and butyric acids. Soares et al. [25] and Rabelo et al. [34] also duction pathways. The conversion of pyruvic acid to acetyl-CoA, and
identified Clostridium sp. in the batch reactors with sugarcane bagasse, subsequently to other metabolites such as HAc and HBu, has been
associated with acetic and butyric acids as the main soluble metabolites identified as a key reaction for OC-H. This reaction, characteristic of
of the process. clostridial fermentation [109], is catalyzed by the enzyme
Lactobacillus, identified in high abundance in assay E-10, performs pyruvate-ferredoxin oxidoreductase [1.2.7.1] (PFOR), in which reduced
carbohydrate fermentation, in which lactic acid is the main by-product, ferredoxin (Fdred) is also formed, which is subsequently reoxidized
in addition to acetic acid, ethanol, formic acid, and CO2 [92]. High lactic resulting in hydrogen formation by the action of hydrogenases [110,
acid production is often associated with lactic acid bacteria (LAB) 111]. According to the inference, the acetic acid formation occurred
belonging to the genus Lactobacillus, classified into homofermentative predominantly through the clostridial pathway, for which the role of the
and heterofermentative [98]. The significant lactic acid production enzymes phosphate acetyltransferase [2.3.1.8] and acetate kinase
(6367.8 mg.L-1) in assay E-10 and the high proportion (81.9%) of this [2.7.2.1] stands out. Regarding the butyric acid, the main catalytic en­
among the soluble metabolites, such as HAc (867.1 mg.L-1), allow us to zymes of the reactions for this metabolic pathway were acetyl-CoA
infer that this genus’ abundance was associated with mostly homo­ acetyltransferase [2.3.1.9], butyrate kinase [2.7.2.7], and
fermentative LAB [99]. The significant reduction in H2 production in hydroxybutyryl-CoA dehydrogenase [1.1.1.157], attributed mainly to
assay E-10, associated with the higher Lactobacillus abundance, may also the metabolism of microorganisms of the genus Clostridium.
be related to bacteriocins secreted by LAB, which have deleterious ef­ For assay E-10, a fundamental role of the homofermentative meta­
fects on other bacteria [100]. Furthermore, Liu et al. [100] point out that bolism of lactic acid bacteria was inferred, highlighting the importance
Lactobacillus sp. produces high amounts of CO2, which is an adverse of the enzyme L-lactate dehydrogenase [1.1.1.27] (Fig. 9), which cata­
condition for stabilizing the H2 concentration in biogas. This fact may be lyzes the conversion reaction of pyruvic acid into HLa, with the ability to
associated with the low H2 percentage (33.2%) in the biogas composi­ produce this enzyme in fermentative processes attributed mainly to
tion for assay E-10 compared to OC-H (47.5%). microorganisms of the genus Lactobacillus [112]; but some microor­
Similar to the results observed in the present study, other authors ganisms of the genera Clostridium and Paraclostridium are also associ­
have observed a reduction and failures in H2 production associated with ated with the production of L-lactate dehydrogenase [55].
an increase in the Lactobacillus abundance in fermentation reactors. For
example, Jo et al. [101] evaluated the H2 production from food waste in 3.6. Energy balance assessment
a continuously stirred reactor and observed that the Lactobacillus spp.
abundance increase was associated with reduced H2 production and The energy balance of the H2 optimization assay (OC-H) was calcu­
changes in the soluble metabolite distribution, with an increased lactic lated and taken as a reference scenario (SCN-R), comparing it to alter­
acid concentration in the effluent. Park et al. [102] observed changes in native scenarios represented by assays E-7 (SCN-A1) and E- 13 (SCN-
the microbial community of the reactor throughout the operation with A2), whose conditions resulted in maximum and minimum buffer de­
an increase in the relative abundance of Lactobacillus sp., which mands per unit of energy produced, respectively.

12
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

Fig. 9. Simplified metabolic map, with the probable enzymes (EC) in the conversion of citrus waste in H2 and organic acids, in the optimized condition for H2
production (OC-H), and assay E-10.

For the reference scenario (SCN-R), the EPH2 was 523.2 MJ.t-1CPW, total energy content of the citrus waste, around 3%, considering the
while for the alternative scenarios, it was only 253.0 MJ.t-1CPW for SCN- lower heating value of 16.97 MJ.t-1CPW [113].
A1 and 79.4 MJ.t-1CPW for SCN-A2 (Table 7). It can be observed by the Regarding the electricity production potential (EPPH2) from the
EPH2 values that the H2 production optimization resulted in a significant hydrogen utilization in a fuel cell, values of 0.872 MWh.t-1CPW for SCN-
gain in the energy recovery potential. However, it should be noted that R, 0.422 and 0.132 MWh.t-1CPW for SCN-A1 and SCN-A2 were obtained,
despite this, the H2 energy potential still represents a tiny portion of the respectively. Despite not representing a high electricity production,

13
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

Table 7 (20.0%).
Energy Potential (EPH2), Electricity Production Potential (EPPH2), and buffering A maximum energy potential of 523.2 MJ.t-1CPW was estimated,
agent demand for different experimental conditions of hydrogen production. based on the optimized H2 production, representing an electricity pro­
Scenario Assays MJ.t-1 MWh. t-1 tCaCO3. US$.MWh- duction potential of 0.872 MWh.t-1CPW. In this condition, the buffering
CPW CPW MWh-1 1
agent demand was 2.0 tCaCO3. MWh-1. It is concluded, from the results
SCN-R OC-H 523.2 0.087 2.00 941.45 obtained that the co-fermentation of citrus waste is a technically feasible
SCN-A1 E-7 253.0 0.042 7.74 3641.33 alternative for hydrogen fermentation, considering the H2 production
SCN-A2 E-13 79.4 0.013 0.75 351.85 achieved; however, alternatives must be evaluated to meet the demand
for alkalinity in the system. These results are important to guide the
investigation of the process in continuous and semi-continuous reactors,
EPPH2 can meet a portion of the energy demand of the processing in­
allowing the process scale-up to reach the industrial application.
dustry itself. Galvagno et al. [50] performed an energy integration
analysis for a citrus juice production plant localized in Sicily - Italy. In
CRediT authorship contribution statement
this evaluation to process 72,000 tons of citrus per year, there was a
demand close to 9000 MWh and the generation of close to 8600 tons of
Danilo Henrique Donato Rocha: Conceptualization, Methodology,
dry waste (residual moisture < 20%). Taking into account this electricity
Formal analysis, Investigation (reactor’s monitoring, data acquisition
demand, waste generation, and the EPPH2 obtained in the present study,
and interpretation), Data curation, Writing – original draft and Writing –
it can be inferred that the energy recovery based on the H2 production
review & editing, Visualization. Isabel Kimiko Sakamoto: Molecular
for SCN-R could supply 8.3% of the electricity demand of an industry
Biology, Formal analysis (metabolic pathway analysis). Maria Berna­
with characteristics similar to the one that was the object of the study by
dete A. Varesche: Conceptualization, Methodology, Resources, Writing
Galvagno et al. [50], while SCN-A1 would supply 4.0%, and SCN-A2
– review & editing, Supervision, Project administration, Funding
only 1.3%.
acquisition.
An important comparison between the evaluated scenarios is related
to the buffering agent demand, representing an important drawback for
Declaration of Competing Interest
the economic viability of energy recovery from hydrogen fermentation
since the average cost per ton of calcium carbonate is US$ 470.35 [114].
The authors declare that they have no known competing financial
For SCN-R relative to the optimized condition of H2 production, a de­
interests or personal relationships that could have appeared to influence
mand of 2.0 tCaCO3. MWh-1is identified, which results in an associated
the work reported in this paper.
cost of 941.45 US$. MWh-1, higher than the demand and the cost asso­
ciated obtained for SCN-A2 which was 0.75 tCaCO3. MWh-1, resulting in
Data Availability
a cost of 351.85 US$. MWh-1. However, despite the lower cost associated
with alkalinization in SCN-A2, EPPH2 in this condition was incipient,
Data will be made available on request.
almost 7 times lower than that obtained in the reference scenario
(SCN-R). By observing the demand for the buffering agent (7.74 tCaCO3.
Acknowledgements
MWh-1) and the associated cost (3641.33 US$. MWh-1) for SCN-A1, it
can be seen how a non-optimized condition can result in a distancing
The authors gratefully acknowledge the São Paulo Research Foun­
from the process viability. Therefore, economic aspects related to pro­
dation (FAPESP Processes 2015/02640-2), and the Coordination for the
cess inputs should be evaluated as variables in studies that aim to
Improvement of Higher Education Personnel (CAPES – Finance Code
evaluate the energy recovery potential from H2 fermentation, aiming at
001).
process scale-up.
To the process scale-up, an alternative would be the incorporation of
Appendix A. Supporting information
an additional co-substrate in the proposed co-digestion of citrus agro-
industry waste, which can effectively contribute to the system’s alka­
Supplementary data associated with this article can be found in the
linity, reducing the demand for buffering agents supplementation. As
online version at doi:10.1016/j.jece.2023.111252.
discussed by Neshat et al. [115], animal manure is a feasible substrate
for anaerobic digestion due to its high buffering capacity, contributing
References
to process stability. Li et al. [116] verified an alkalinity of 4.2 gCaCO3.
L-1 for cattle manure, indicating that this material could increase the [1] L.C. Freitas, J.R. Barbosa, A.L.C. Costa, F.W.F. Bezerra, R.H.H.J. Pinto, R.
buffering capacity of the digesters, preventing failures in the process N. Carvalho, From waste to sustainable industry: How can agro-industrial wastes
related to acidification. Yin et al. [117] reported a high alkalinity of 5.09 help in the development of new products ? Resour. Conserv Recycl 169 (2021),
105466 https://doi.org/10.1016/j.resconrec.2021.105466.
gCaCO3. L-1 for swine manure, highlighting the high ammonia release
[2] S. Pan, M. Alex, I. Huang, I. Liu, E. Chang, P. Chiang, Strategies on
contributing to the VFAs neutralization [118]. implementation of waste-to-energy ( WTE) supply chain for circular economy
system: a review, J. Clean. Prod. 108 (2015) 409–421, https://doi.org/10.1016/j.
4. Conclusions jclepro.2015.06.124.
[3] B.B. Kist, C. Carvalho, M. Treichel, C.E. Santos, Anuário brasileiro da fruticultura,
Gaz. St Cruz (2018) 88.
The application of optimization methodology made it possible to [4] FAO. Food and agriculture data 2019. 〈http://www.fao.org/faostat/en/#home〉
obtain a maximum hydrogen production of 1249.0 mL.L-1 from the (accessed July 15, 2020).
[5] CITURSBR. Total Brazilian Orange Juice Exports. Brazilian Assoc Citrus Export
citrus waste co-digestion (21.4 gTVS.L-1 of CPW, 3.4 gCOD.L-1 of 2020. 〈http://www.citrusbr.com/en/statistics/?ins=03〉.
wastewater, and 4.5 g.L-1 of CaCO3), such production was up to 5.3 [6] BRASIL. Grupos de Produtos: Exportação Brasileira. Ministério Da Econ - Secr
times higher than the production obtained in the non-optimized con­ Comércio Exter 2020. 〈http://www.mdic.gov.br/index.php/comercio-exterior/
estatisticas-de-comercio-exterior/series-historicas〉 (accessed July 15, 2020).
ditions. The optimized condition for hydrogen fermentation was asso­ [7] Tetra Pak. The Orange Book. Lund, Sweden: 2004.
ciated with the Clostridial metabolic pathway, with a high abundance of [8] E.S.R. Mendoza, J.M.M. Contreras, A.M. Sibaja, N.A.V. Cantú, A.L. Alvarado,
the genus Clostridium sensu stricto 1 (95.0%), with predominance of Anaerobic digestion of citrus industry effluents using an Anaerobic Hybrid
Reactor, Clean. Technol. Environ. Policy (2018) 1–11, https://doi.org/10.1007/
acetic and butyric acids. For the opposite condition, with low H2 pro­ s10098-017-1483-1.
duction, there was a predominance of lactic acid fermentation with [9] A. Koppar, P. Pullammanappallil, Anaerobic digestion of peel waste and
increased abundance for Lactobacillus (21.9%) and Paraclostridium wastewater for on site energy generation in a citrus processing facility, Energy 60
(2013) 62–68, https://doi.org/10.1016/j.energy.2013.08.007.

14
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

[10] P.S. Calabro, A. Folino, V. Tamburino, G. Zappia, D.A. Zema, S.M. Zimbone, green dye in aqueous solution by Fenton process, Water Resour. Ind. 15 (2016)
Valorisation of citrus processing waste: a review. Sixt. Int. Waste Manag. Landfill 41–48, https://doi.org/10.1016/j.wri.2016.07.002.
Symp, CISA Publisher,, Cagliari, Italy, 2017. [34] C.A.B.S. Rabelo, L.A. Soares, I.K. Sakamoto, E.L. Silva, M.B.A. Varesche,
[11] E.S. Rosas-Mendoza, J.M. Mendez-Contreras, A.A. Aguilar-Lasserre, N.A. Vallejo- Optimization of hydrogen and organic acids productions with autochthonous and
Cantú, A. Alvarado-Lassman, Evaluation of bioenergy potential from citrus allochthonous bacteria from sugarcane bagasse in batch reactors, J. Environ.
effluents through anaerobic digestion, J. Clean. Prod. 254 (2020) 1–11, https:// Manag. 223 (2018) 952–963, https://doi.org/10.1016/j.jenvman.2018.07.015.
doi.org/10.1016/j.jclepro.2020.120128. [35] F.P. Camargo, I.K. Sakamoto, T.P. Delforno, M. Mariadassou, V. Loux, I.C.
[12] D.A. Zema, P.S. Calabro, A. Folino, V. Tamburino, G. Zappia, S.M. Zimbone, S. Duarte, et al., Microbial and functional characterization of an allochthonous
Wastewater management in citrus processing industries: an overview of consortium applied to hydrogen production from Citrus Peel Waste in batch
advantages and limits, Water 11 (2019) 1–23, https://doi.org/10.3390/ reactor in optimized conditions, J. Environ. Manag. 291 (2021), 112631, https://
w11122481. doi.org/10.1016/j.jenvman.2021.112631.
[13] D.A. Zema, P.S. Calabrò, A. Folino, V. Tamburino, G. Zappia, S.M. Zimbone, [36] A.C. Villa Montoya, R.C. da Silva Mazareli, T.P. Delforno, V.B. Centurion, V.M. de
Valorisation of citrus processing waste: A review, Waste Manag. 80 (2018) Oliveira, E.L. Silva, et al., Optimization of key factors affecting hydrogen
252–273, https://doi.org/10.1016/j.wasman.2018.09.024. production from coffee waste using factorial design and metagenomic analysis of
[14] Z. Berk. Citrus Fruit Processing, 1st ed., Elsevier, 2016 doi:books/citrus-fruit- the microbial community, Int J. Hydrog. Energy 45 (2020) 4205–4222, https://
processing/berk/978-0-12-803133-9. doi.org/10.1016/j.ijhydene.2019.12.062.
[15] P.S. Calabró, E. Paone, D. Komilis, Strategies for the sustainable management of [37] R.C. da S. Mazareli, A.C.V. Montoya, T.P. Delforno, V.B. Centurion, V.M.
orange peel waste through anaerobic digestion, J. Environ. Manag. 212 (2018) de Oliveira, E.L. Silva, et al., Enzymatic routes to hydrogen and organic acids
462–468, https://doi.org/10.1016/j.jenvman.2018.02.039. production from banana waste fermentation by autochthonous bacteria:
[16] J.A. López, Q. Li, I.P. Thompson, Biorefinery of waste orange peel, Crit. Rev. Optimization of pH and temperature, Int J. Hydrog. Energy 6 (2021) 8454–8468,
Biotechnol. 30 (2010) 63–69, https://doi.org/10.3109/07388550903425201. https://doi.org/10.1016/j.ijhydene.2020.12.063.
[17] P.S. Calabrò, L. Pontoni, I. Porqueddu, R. Greco, F. Pirozzi, F. Malpei, Effect of the [38] D. Kim, S. Han, S. Kim, H. Shin, Effect of gas sparging on continuous fermentative
concentration of essential oil on orange peel waste biomethanization: preliminary hydrogen production, Int J. Hydrog. Energy 31 (2006) 2158–2169, https://doi.
batch results, Waste Manag 48 (2016) 440–447, https://doi.org/10.1016/j. org/10.1016/j.ijhydene.2006.02.012.
wasman.2015.10.032. [39] C.A. Menezes, E.L. Silva, Hydrogen production from sugarcane juice in expanded
[18] P.S. Calabrò, F. Fazzino, R. Sidari, A.D. Zema, Optimization of orange peel waste granular sludge bed reactors under mesophilic conditions: The role of
ensiling for sustainable anaerobic digestion, Renew. Energy 154 (2020) 849–862, homoacetogenesis and lactic acid production, Ind. Crop Prod. 138 (2019),
https://doi.org/10.1016/j.renene.2020.03.047. 111586, https://doi.org/10.1016/j.indcrop.2019.111586.
[19] A. Tenca, A. Schievano, F. Perazzolo, F. Adani, R. Oberti, Bioresource Technology [40] Z. Cui, J.P. Kerekes, Impact of Wavelength Shift in Relative Spectral Response at
Biohydrogen from thermophilic co-fermentation of swine manure with fruit and High Angles of Incidence in Future Landsat Design Concepts, IEEE Trans. Geosci.
vegetable waste: Maximizing stable production without pH control, Bioresour. Remote Sens 56 (2018) 5873–5883, https://doi.org/10.1109/
Technol. 102 (2011) 8582–8588, https://doi.org/10.1016/j. TGRS.2018.2827394.
biortech.2011.03.102. [41] M. Dubois, K.A. Gilles, J.K. Hamilton, P.A. Rebers, F. Smith, Colorimetric Method
[20] D.E. Algapani, W. Qiao, M. Ricci, M. Bianchi D, S. Wandera, F. Adani, et al., Bio- for Determination of Sugars and Related Substances, Anal. Chem. 28 (1956)
hydrogen and bio-methane production from food waste in a two-stage anaerobic 350–356, https://doi.org/10.1021/ac60111a017.
digestion process with digestate recirculation, Renew. Energy 130 (2019) [42] APHA/AWWA/WEF. Standard methods for the examination of water and
1108–1115, https://doi.org/10.1016/j.renene.2018.08.079. wastewater. 22nd ed. Washington, DC: Études Rabelaisiennes; 2012. doi:〈
[21] S.F. Fu, R. Liu, W.X. Sun, R. Zhu, H. Zou, Y. Zheng, et al., Enhancing energy https://doi.org/10.1021/ed076p1069.2〉.
recovery from corn straw via two-stage anaerobic digestion with stepwise [43] I.D. Buchanan, J.A. Nicell, Model development for horseradish peroxidase
microaerobic hydrogen fermentation and methanogenesis, J. Clean. Prod. 247 catalyzed removal of aqueous phenol, Biotechnol. Bioeng. 54 (1997) 251–261,
(2020), 119651, https://doi.org/10.1016/j.jclepro.2019.119651. https://doi.org/10.1002/(SICI)1097-0290(19970505)54:3<251::AID-BIT6>3.0.
[22] A.C. Villa Montoya, R. Cristina da Silva Mazareli, T.P. Delforno, V.B. Centurion, I. CO;2-E.
K. Sakamoto, V. Maia de Oliveira, et al., Hydrogen, alcohols and volatile fatty [44] C.Z. Lazaro, D.V. Vich, J.S. Hirasawa, M.B.A. Varesche, Hydrogen production and
acids from the co-digestion of coffee waste (coffee pulp, husk, and processing consumption of organic acids by a phototrophic microbial consortium, Int J.
wastewater) by applying autochthonous microorganisms, Int J. Hydrog. Energy Hydrog. Energy 37 (2012) 11691–11700, https://doi.org/10.1016/j.
44 (2019) 21434–21450, https://doi.org/10.1016/j.ijhydene.2019.06.115. ijhydene.2012.05.088.
[23] F.P. Camargo, I. Kimiko, A. Bize, I. Cristina, S. Duarte, E. Luiz, Screening design of [45] M.I. Rodrigues, A.F. Iemma. Experimental Design and Process Optimization, 1st
nutritional and physicochemical parameters on bio-hydrogen and volatile fatty ed..,, CRC Press,, 2014, https://doi.org/10.1201/b17848.
acids production from Citrus Peel Waste in batch reactors, Int J. Hydrog. Energy [46] M.M. Gomes, I.K. Sakamoto, C.A. Rabelo, E.L. Silva, M.B.A. Varesche, Statistical
46 (2020) 7794–7809, https://doi.org/10.1016/j.ijhydene.2020.06.084. optimization of methane production from brewery spent grain: Interaction effects
[24] A.G.L. Moura, C.A.B.S. Rabelo, C.H. Okino, S.I. Maintinguer, E.L. Silva, M.B. of temperature and substrate concentration, J. Environ. Manag. 288 (2021),
A. Varesche, Enhancement of Clostridium butyricum hydrogen production by 112363.
iron and nickel nanoparticles: Effects on hydA expression, 28477–28461, Int J. [47] Goldmeer J. POWER TO GAS: HYDROGEN FOR POWER GENERATION Fuel
Hydrog. Energy 45 (2020), https://doi.org/10.1016/j.ijhydene.2020.07.161. Flexible Gas Turbines as Enablers for a Low or Reduced Carbon Energy
[25] L.A. Soares, C.A. Borges Silva Rabelo, I.K. Sakamoto, T.P. Delforno, E.L. Silva, M. Ecosystem. Gen Electr Co 2019.
B.A. Varesche, Metagenomic analysis and optimization of hydrogen production [48] DOE Energy Efficiency And Renewable Energy Information Center. Hydrogen
from sugarcane bagasse, Biomass-.-. Bioenergy 117 (2018) 78–85, https://doi. Fuel Cells 2006.
org/10.1016/j.biombioe.2018.07.018. [49] B. Shabani, J. Andrews, Hydrogen and Fuel Cells, Green. Energy Technol. (2015)
[26] D.H.D. Rocha, I.K. Sakamoto, M.B.A. Varesche, Evaluation of significant factors in 453–491, https://doi.org/10.1007/978-81-322-2337-5.
hydrogen production and volatile fatty acids in co-fermentation of citrus peel [50] A. Galvagno, M. Prestipino, S. Maisano, F. Urbani, Integration into a citrus juice
waste and processing wastewater, Fuel 354 (2023), 129306, https://doi.org/ factory of air-steam gasi fi cation and CHP system: Energy sustainability
10.1016/j.fuel.2023.129306. assessment, Energy Convers. Manag 193 (2019) 74–85, https://doi.org/10.1016/
[27] L. Sillero, R. Solera, M. Perez, Anaerobic co-digestion of sewage sludge, wine j.enconman.2019.04.067.
vinasse and poultry manure for bio-hydrogen production, Int J. Hydrog. Energy [51] Y. Yu, C. Lee, J. Kim, S. Hwang, Group-Specific Primer and Probe Sets to Detect
47 (2022) 3667–3678, https://doi.org/10.1016/j.ijhydene.2021.11.032. Methanogenic Communities Using Quantitative Real-Time Polymerase Chain
[28] A. Rabii, S. Aldin, Y. Dahman, E. Elbeshbishy, A Review on Anaerobic Co- Reaction, Biotechnol. Bioeng. (2005) 670–679, https://doi.org/10.1002/
Digestion with a Focus on the Microbial Populations and the Effect os Multi-Stage bit.20347.
Digester Configuration, Energies 12 (2019) 1–25, https://doi.org/10.3390/ [52] B. Callahan, P. McMurdie, M. Rosen, A. Han, A. Johnson, S. Holmes, DADA2:
en12061106. High-resolution sample inference from illumina amplicon data, Nat. Methods 13
[29] G. Esposito, L. Frunzo, A. Giordano, F. Pirozzi, A. Liotta, F. Panico, Anaerobic co- (2016) 581–583, https://doi.org/10.1038/nmeth.3869.
digestion of organic wastes, Environ. Sci. Biotechnol. (2012) 325–341, https:// [53] P. McMurdie, S. Holmes, Phyloseq: An r package for reproducible interactive
doi.org/10.1007/s11157-012-9277-8. analysis and graphics of microbiome census data, PLoS One 8 (2013) 61217,
[30] M.A. De la Rubia, V. Fernández-Cegrí, F. Raposo, R. Borja, Influence of particle https://doi.org/10.1371/journal.pone.0061217.
size and chemical composition on the performance and kinetics of anaerobic [54] BRENDA Enzyme Database. The Comprehensive Enzyme Information System
digestion process of sunflower oil cake in batch mode, Biochem Eng. J. 58–59 2022. 〈https://www.brenda-enzymes.org/〉.
(2011) 162–167, https://doi.org/10.1016/j.bej.2011.09.010. [55] KEGG Pathway Database. Wiring diagrams of molecular interactions, reactions
[31] L. Pontoni, G. D’Antonio, G. Esposito, M. Fabbricino, L. Frunzo, F. Pirozzi, and relations 2022. 〈https://www.genome.jp/kegg/pathway.html〉.
Thermal pretreatment of olive mill wastewater for efficient methane production: [56] A. Reungsang, S. Pattra, S. Sittijunda, Optimization of key factors affecting
Control of aromatic substances degradation by monitoring cyclohexane methane production from acidic effluent coming from the sugarcane juice
carboxylic acid, Environ. Technol. (U. Kingd. ) 36 (2015) 1785–1794, https://doi. hydrogen fermentation process, Energies (2012) 4746–4757, https://doi.org/
org/10.1080/09593330.2015.1012179. 10.3390/en5114746.
[32] A.C. dos Santos, E. Ximenes, Y. Kim, M.R. Ladisch, Lignin–enzyme interactions in [57] K.N. Niladevi, R.K. Sukumaran, N. Jacob, G.S. Anisha, P.Ã. Prema, Optimization
the hydrolysis of lignocellulosic biomass, Trends Biotechnol. 37 (2019) 518–531, of laccase production from a novel strain — Streptomyces psammoticus using
https://doi.org/10.1016/j.tibtech.2018.10.010. response surface methodology, Microbiol Res 164 (2009) 105–113, https://doi.
[33] A. Elhalil, H. Tounsadi, R. Elmoubarki, F.Z. Mahjoubi, M. Farnane, Factorial org/10.1016/j.micres.2006.10.006.
experimental design for the optimization of catalytic degradation of malachite

15
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

[58] J. Kainthola, A.S. Kalamdhad, V.V. Goud, Optimization of methane production [82] K. Bélafi-bakó, D. Búcsú, Z. Pientka, B. Bálint, Z. Herbel, Integration of
during anaerobic co-digestion of rice straw and Hydrilla verticillata using biohydrogen fermentation and gas separation processes to recover and enrich
response surface methodology, Fuel 235 (2019) 92–99, https://doi.org/10.1016/ hydrogen, Int J. Hydrog. Energy 31 (2006) 1490–1495, https://doi.org/10.1016/
j.fuel.2018.07.094. j.ijhydene.2006.06.022.
[59] Y. Pu, J. Tang, X.C. Wang, Y. Hu, Hydrogen production from acidogenic food [83] J. Baeyens, H. Zhang, J. Nie, L. Appels, R. Dewil, R. Ansart, et al., Reviewing the
waste fermentation using untreated inoculum: Effect of substrate concentrations, potential of bio-hydrogen production by fermentation, Renew. Sustain Energy
Int. J. Hydrog. Energy 44 (2019) 27272–27284, https://doi.org/10.1016/j. Rev. 131 (2020), 110023, https://doi.org/10.1016/j.rser.2020.110023.
ijhydene.2019.08.230. [84] T.A.Z.De Souza, D.H.D. Rocha, A.A.V. Julio, C.J.R. Coronado, J.L. Silveira, R.
[60] B. Ai, J. Li, X. Chi, J. Meng, A.K. Jha, C. Liu, et al., Effect of pH and Buffer on J. Silva, et al., Exergoenvironmental assessment of hydrogen water footprint via
Butyric Acid Production and Microbial Community Characteristics in steam reforming in Brazil, J. Clean. Prod. 311 (2021), 127577, https://doi.org/
Bioconversion of Rice Straw with Undefined Mixed Culture, Biotechnol. 10.1016/j.jclepro.2021.127577.
Bioprocess Eng. 19 (2014) 676–686, https://doi.org/10.1007/s12257-013-0655- [85] J. Park, S. Lee, J. Yoon, S. Kim, H. Park, Predominance of cluster I Clostridium in
z. hydrogen fermentation of galactose seeded with various heat-treated anaerobic
[61] H. Lee, W.F.J. Vermaas, B.E. Rittmann, Biological hydrogen production: prospects sludges, Bioresour. Technol. 157 (2014) 98–106, https://doi.org/10.1016/j.
and challenges, Trends Biotechnol. 28 (2010) 262–271, https://doi.org/10.1016/ biortech.2014.01.081.
j.tibtech.2010.01.007. [86] J.C. Ribeiro, V.T. Mota, V.M. Oliveira, M. Zaiat, Hydrogen and organic acid
[62] S.S. Salek, A.G.Van Turnhout, R. Kleerebezem, M.C.M.Van Loosdrecht, pH production from dark fermentation of cheese whey without buffers under
Control in Biological Systems Using Calcium Carbonate, Biotechnol. Bioeng. mesophilic condition, J. Environ. Manag. 304 (2022), 114253, https://doi.org/
(2015) 112, https://doi.org/10.1002/bit.25506. 10.1016/j.jenvman.2021.114253.
[63] S. Liu, W. Li, G. Zheng, H. Yang, L. Li, Optimization of cattle manure and food [87] M.A. Mazorra-manzano, G.R. Robles-porchas, M. Mart, J.C. Ram, C.O. Garc, M.
waste co-digestion for biohydrogen production in a mesophilic semi-continuous J. Torres-llanez, et al., Bacterial Diversity and Dynamics during Spontaneous
process, Energies 13 (2020) 3848. Cheese Whey Fermentation at Different Temperatures, Fermentation 8 (2022)
[64] J. Zhang, Q. Wang, J. Jiang, Lime mud from paper-making process addition to 1–11, https://doi.org/10.3390/fermentation8070342.
food waste synergistically enhances hydrogen fermentation performance, Int. J. [88] S.I. Maintinguer, I.K. Sakamoto, M.A.T. Adorno, M.B.A. Varesche, Bacterial
Hydrog. Energy 38 (2013) 2738–2745, https://doi.org/10.1016/j. diversity from environmental sample applied to bio-hydrogen production, Int J.
ijhydene.2012.12.048. Hydrog. Energy 40 (2015) 3180–3190, https://doi.org/10.1016/j.
[65] S. Sangyoka, A. Reungsang, C. Lin, Optimization of biohydrogen production from ijhydene.2014.12.118.
sugarcane bagasse by mixed cultures using a statistical method, Sustain. Environ. [89] T. Palladino, D. Gileno, V. Lacerda, J. Melline, I.K. Sakamoto, M. Bernadete, et al.,
Res. 26 (2016) 235–242, https://doi.org/10.1016/j.serj.2016.05.001. Microbial diversity of a full- - scale UASB reactor applied to poultry
[66] Z. Li, Z. Chen, H. Ye, Y. Wang, W. Luo, J. Chang, et al., Anaerobic co-digestion of slaughterhouse wastewater treatment: integration of 16S rRNA gene amplicon
sewage sludge and food waste for hydrogen and VFA production with microbial and shotgun metagenomic sequencing, Microbiol Open (2017) 1–12, https://doi.
community analysis, Waste Manag 78 (2018) 789–799, https://doi.org/10.1016/ org/10.1002/mbo3.443.
j.wasman.2018.06.046. [90] M. Diallo, S.W.M. Kengen, A.M. López-Contreras, Sporulation in solventogenic
[67] C. Lin, C.H. Lay, Effects of carbonate and phosphate concentrations on hydrogen and acetogenic clostridia, Appl. Microbiol Biotechnol. 105 (2021) 3533–3557,
production using anaerobic sewage sludge microflora, Int J. Hydrog. Energy 29 https://doi.org/10.1007/s00253-021-11289-9.
(2004) 275–281, https://doi.org/10.1016/j.ijhydene.2003.07.002. [91] G. Yang, Y. Yin, J. Wang, Microbial community diversity during fermentative
[68] G. Yang, J. Wang, Biohydrogen production by co-fermentation of sewage sludge hydrogen production inoculating various pretreated cultures, Int J. Hydrog.
and grass residue: E ff ect of various substrate concentrations, Fuel 237 (2019) Energy 4 (2019) 13147–13156, https://doi.org/10.1016/j.ijhydene.2019.03.216.
1203–1208, https://doi.org/10.1016/j.fuel.2018.10.026. [92] E. Salvetti, S. Torriani, G.E. Felis, The genus lactobacillus: a taxonomic update,
[69] W. Jianlong, W. Wei, The effect of substrate concentration on biohydrogen Probiotics Antimicrob. Proteins (2012) 217–226, https://doi.org/10.1007/
production by using kinetic models, Sci. China Ser. B Chem. 51 (2008) s12602-012-9117-8.
1110–1117, https://doi.org/10.1007/s11426-008-0104-6. [93] X. Hu, J. Zeng, F. Shen, X. Xia, X. Tian, Z. Wu, Citrus pomace fermentation with
[70] R.K. Thauer, K. Jungermann, K. Decker, Energy conservation in chemotrophic autochthonous probiotics improves its nutrient composition and antioxidant
anaerobic bacteria, Bacteriol. Rev. 41 (1977) 100–180, doi:https://doi.org/ activities, LWT - Food Sci. Technol. (2022) 157, https://doi.org/10.1016/j.
10.1128%2Fbr.41.1.100-180.1977. lwt.2022.113076.
[71] H. Lee, R. Krajmalinik-brown, H. Zhang, B.E. Rittmann, An electron-flow model [94] I. Zerva, N. Remmas, P. Melidis, G. Tsiamis, S. Ntougias, Microbial succession and
can predict complex redox reactions in mixed-culture fermentative BioH 2: identification of effective indigenous pectinolytic yeasts from orange juice
microbial ecology evidence, Biotechnol. Bioeng. 104 (2009) 687–697, https:// processing wastewater, Waste Biomass-.-. Valoriz. 12 (2021) 4885–4899, https://
doi.org/10.1002/bit.22442. doi.org/10.1007/s12649-021-01364-7.
[72] M. Akhlaghi, M.R. Boni, A. Polettini, R. Pomi, A. Rossi, G. De Gioannis, et al., [95] L. Luo, S. Sriram, D. Johnravindar, T. Louis, P. Martin, J.W.C. Wong, et al., Effect
Fermentative H2 production from food waste: Parametric analysis of factor of inoculum pretreatment on the microbial and metabolic dynamics of food waste
effects, Bioresour. Technol. 276 (2019) 349–360, https://doi.org/10.1016/j. dark fermentation, Bioresour. Technol. 358 (2022), 127404, https://doi.org/
biortech.2019.01.012. 10.1016/j.biortech.2022.127404.
[73] G. De Gioannis, M. Friargiu, E. Massi, A. Muntoni, A. Polettini, Biohydrogen [96] G. Yang, J. Wang, Biohydrogen production by co-fermentation of antibiotic
production from dark fermentation of cheese whey: Influence of pH, Int J. fermentation residue and fallen leaves: Insights into the microbial community and
Hydrog. Energy (2014) 1–12, https://doi.org/10.1016/j.ijhydene.2014.10.046. functional genes, Bioresour. Technol. 337 (2021), 125380, https://doi.org/
[74] F.P. Camargo, A. Sarti, A.C. Alécio, C.A. Sabatini, M.A.T. Adorno, I.C.S. Duarte, et 10.1016/j.biortech.2021.125380.
al., Limonene qunatification by gas chromatography with mass spectometry (GC- [97] R.C. d S. Mazareli, A.C. Villa-Montoya, T.P. Delforno, V.B. Centurion, V. Maia de
MS) and its effects on hydrogen and volatile fatty acids production in anaerobic Oliveira, E.L. Silva, et al., Metagenomic analysis of autochthonous microbial
reactors, Quim. Nova (2020) 1–7, https://doi.org/10.21577/0100- biomass from banana waste: Screening design of factors that affect hydrogen
4042.20170557. production, Biomass-.-. Bioenergy 138 (2020), 105573, https://doi.org/10.1016/
[75] X.M. Guo, E. Latrille, H. Carrère, J.-P. Steyer, E. Trably, Hydrogen production j.biombioe.2020.105573.
from agricultural waste by dark fermentation: A review, Int J. Hydrog. Energy 35 [98] D. Cizeikiene, G. Juodeikiene, J. Damasius, Biocatalysis and Agricultural
(2010) 10660–10673, https://doi.org/10.1016/j.ijhydene.2010.03.008. Biotechnology Use of wheat straw biomass in production of L-lactic acid applying
[76] J. Tang, X.C. Wang, Y. Hu, Y. Zhang, Y. Li, Effect of pH on lactic acid production biocatalysis and combined lactic acid bacteria strains belonging to the genus
from acidogenic fermentation of food waste with different types of inocula, Lactobacillus, Biocatal. Agric. Biotechnol. 15 (2018) 185–191, https://doi.org/
Bioresour. Technol. 224 (2017) 544–552, https://doi.org/10.1016/j. 10.1016/j.bcab.2018.06.015.
biortech.2016.11.111. [99] O. Kandler, Carbohydrate metabolism in lactic acid bacteria, Antonie Van.
[77] Z. Lee, S. Li, P. Kuo, I. Chen, Y. Tien, Y. Huang, et al., Thermophilic bio-energy Leeuwenhoek 49 (1983) 209–224, https://doi.org/10.1007/bf00399499.
process study on hydrogen fermentation with vegetable kitchen waste, Int J. [100] H. Liu, G. Wang, D. Zhu, G. Pan, Enrichment of the hydrogen-producing microbial
Hydrog. Energy 35 (2010) 13458–13466, https://doi.org/10.1016/j. community from marine intertidal sludge by different pretreatment methods, Int
ijhydene.2009.11.126. J. Hydrog. Energy 34 (2009) 9696–9701, https://doi.org/10.1016/j.
[78] G. Grause, M. Igarashi, T. Kameda, T. Yoshioka, Lactic acid as a substrate for ijhydene.2009.10.025.
fermentative hydrogen production, Int J. Hydrog. Energy 37 (2012) [101] J.H. Jo, C.O. Jeon, D.S. Lee, J.M. Park, Process stability and microbial community
16967–16973, https://doi.org/10.1016/j.ijhydene.2012.08.096. structure in anaerobic hydrogen-producing microflora from food waste
[79] B. Baghchehsaraee, G. Nakhla, D. Karamanev, A. Margaritis, Effect of extrinsic containing kimchi, J. Biotechnol. 131 (2007) 300–308, https://doi.org/10.1016/
lactic acid on fermentative hydrogen production, Int J. Hydrog. Energy 34 (2009) j.jbiotec.2007.07.492.
2573–2579, https://doi.org/10.1016/j.ijhydene.2009.01.010. [102] J. Park, G. Kumar, J. Park, H. Park, S. Kim, Changes in performance and bacterial
[80] A. Sikora, M. Błaszczyk, M. Jurkowski, U. Zielenkiewicz, Lactic Acid Bacteria in communities in response to various process disturbances in a high-rate
Hydrogen-Producing Consortia: On Purpose or by Coincidence ? Lact. Acid Bact. - biohydrogen reactor fed with galactose, Bioresour. Technol. 188 (2015) 109–116,
R D Food. Heal. Livest. Purp, InTech,, 2013, https://doi.org/10.5772/50364. https://doi.org/10.1016/j.biortech.2015.01.107.
[81] B. Matyakubov, Y. Hwang, T. Lee, Evaluating interactive toxic impact of heavy [103] Y. Yin, J. Wang, Enhanced Sewage Sludge Disintegration and Hydrogen
metals and variations of microbial community during fermentative hydrogen Production by Ionizing Radiation Pretreatment in the Presence of Fe 2+, ACS
production, Int J. Hydrog. Energy 47 (2022) 31223–31240, https://doi.org/ Sustain Chem. Eng. 7 (2019) 15548–15557, https://doi.org/10.1021/
10.1016/j.ijhydene.2022.07.038. acssuschemeng.9b03362.

16
D.H.D. Rocha et al. Journal of Environmental Chemical Engineering 11 (2023) 111252

[104] T.S.S. Jyothsna, L. Tushar, C. Sasikala, C.V. Ramana, Paraclostridium [111] S. Feng, H. Hao, W. Guo, S. Woong, D. Duc, Y. Liu, et al., Wastewater-derived
benzoelyticum gen. nov., sp. nov.,isolated from marine sediment and biohydrogen: critical analysis of related enzymatic processes at the research and
reclassification of Clostridium bifermentans as Paraclostridium bifermentans large scales, Sci. Total Environ. 851 (2022), 158112, https://doi.org/10.1016/j.
comb. nov. Proposal of a new genus Paeniclostridium gen. nov. to accommodate scitotenv.2022.158112.
Clostridium s, Int J. Syst. Evol. Microbiol (2016) 1268–1274, https://doi.org/ [112] S.K. Rhee, Y. Pack, Effect of environmental pH on fernentation balance of
10.1099/ijsem.0.000874. lactobacillus bulgaricus, J. Bacteriol. 144 (1980) 217–221, https://doi.org/
[105] C.A.B.S. Rabelo, C.H. Okino, I.K. Sakamoto, M.B.A. Varesche, Isolation of 10.1128/jb.144.1.217-221.1980.
Paraclostridium CR4 from sugarcane bagasse and its evaluation in the [113] B. Tamelová, J. Malat’ák, J. Velebil, Energy valorisation of citrus peel waste by
bioconversion of lignocellulosic feedstock into hydrogen by monitoring cellulase torrefaction treatment, Agron. Res 16 (2018) 276–285, https://doi.org/
gene expression, Sci. Total Environ. 715 (2020), 136868, https://doi.org/ 10.15159/AR.18.029.
10.1016/j.scitotenv.2020.136868. [114] Intratec Primary Commodity Prices. Calcium Carbonate Prices. Calcium
[106] S.G. Santiago, E. Trably, E. Latrille, The hydraulic retention time influences the Carbonate Prices 2018. 〈https://www.intratec.us/chemical-markets/calcium-ca
abundance of Enterobacter, Clostridium and Lactobacillus during the hydrogen rbonate-price〉 (accessed February 2, 2023).
production from food waste, Lett. Appl. Microbiol. 69 (2019) 138–147, https:// [115] S.A. Neshat, M. Mohammadi, G.D. Najafpour, P. Lahijani, Anaerobic co-digestion
doi.org/10.1111/lam.13191. of animal manures and lignocellulosic residues as a potent approach for
[107] L. Zhuang, S. Zhou, Y. Yuan, T. Liu, Z. Wu, J. Cheng, Development of Enterobacter sustainable biogas production, Renew. Sustain Energy Rev. 79 (2017) 308–322,
aerogenes fuel cells: From in situ biohydrogen oxidization to direct electroactive https://doi.org/10.1016/j.rser.2017.05.137.
biofilm, Bioresour. Technol. 102 (2011) 284–289, https://doi.org/10.1016/j. [116] X. Li, L. Li, M. Zheng, G. Fu, J.S. Lar, Anaerobic co-digestion of cattle manure with
biortech.2010.06.038. corn stover pretreated by sodium hydroxide for efficient biogas production,
[108] L.P. Thapa, S.J. Lee, C. Park, S.W. Kim, Enzyme and Microbial Technology Energy Fuels 74 (2009) 4635–4639, https://doi.org/10.1021/ef900384p.
Production of L-lactic acid from metabolically engineered strain of Enterobacter [117] D. Yin, W. Liu, N. Zhai, G. Yang, X. Wang, Y. Feng, Anaerobic digestion of pig and
aerogenes ATCC 29007, Enzym. Micro Technol. 102 (2017) 1–8, https://doi.org/ dairy manure under photo-dark fermentation condition, Bioresour. Technol. 166
10.1016/j.enzmictec.2017.03.003. (2014) 373–380, https://doi.org/10.1016/j.biortech.2014.05.037.
[109] P.C. Hallenbeck, Fundamentals of the fermentative production of hydrogen, [118] X. Meng, D. Yu, Y. Wei, Y. Zhang, Q. Zhang, Endogenous ternary pH buffer system
Water Sci. Technol. 52 (2005) 21–29, https://doi.org/10.2166/wst.2005.0494. with ammonia-carbonates-VFAs in high solid anaerobic digestion of swine
[110] I. Karube, T. Matsunaga, S. Tsuru, S. Suzuki, Continuous Hydrogen Production By manure: An alternative for alleviating ammonia inhibition ? Process Biochem. 69
Immobilized Whole Cells Of Clostridium Butyricum, Biochim Biophys. Acta - Gen. (2018) 144–152, https://doi.org/10.1016/j.procbio.2018.03.015.
Subj. 444 (1976) 338–343, https://doi.org/10.1016/0304-4165(76)90376-7.

17

You might also like