Research Article

Received: 23 June 2022 Revised: 3 August 2022 Accepted article published: 12 September 2022 Published online in Wiley Online Library: 30 September 2022

(wileyonlinelibrary.com) DOI 10.1002/jctb.7238

Biohydrogen production from food waste and

waste activated sludge in codigestion:
influence of organic loading rate and changes
in microbial community
Iván Moreno-Andrade,* María José Berrocal-Bravo and
Idania Valdez-Vazquez

Abstract
BACKGROUND: Food waste (FW) and waste activated sludge (WAS) are complementary substrates that improve H2 production
by dark fermentation. However, there is little experience in the process performance in the long-term operation of reactors
co-processing FW–WAS. This study aims to determine the optimal FW–WAS ratio in batch H2 production and then feed this
FW–WAS ratio to an anaerobic sequencing batch reactor (SBR) to evaluate the influence of three organic loading rates (OLR)
– 15, 22 and 45 g volatile solids (VS) L−1 d−1 – on H2 performance and microbial composition.
RESULTS: Batch tests showed that the FW–WAS ratio of 90–10 increased 22% of the H2 production compared to the individual
FW. In the SBR operation, the OLR significantly influenced H2 performance, reaching the highest productivity of 733 ± 282
mL H2 L−1 d−1 at an OLR of 22 g VS L−1 d−1. The cooperative lactic acid cross-feeding between Olsenella and Megasphaera
resulted in the highest H2 productivity at an OLR of 22 g VS L−1 d−1. OLR promoted the prevalence of different taxa of hydro-
lytic bacteria, where Enterococcus and Veillonella were positively correlated, with the highest substrate hydrolysis of 50% at an
OLR of 15 g VS L−1 d−1. Prevotella, a genus with a wide spectrum of hydrolytic enzymes, was predominant in the SBR operation
regardless of the OLR.
CONCLUSION: The SBR operation at OLR of 22 g VS L−1 d−1 improved H2 production by applying a 90–10 FW–WAS ratio. The
change in OLR promoted different taxa of hydrolytic species in the SBR; however, Prevotella spp. prevailed at the three OLRs.
© 2022 The Authors. Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society
of Chemical Industry (SCI).

Keywords: hydrogen; Megasphaera; organic loading rate; Prevotella; sequencing batch reactor

INTRODUCTION wastes helps decrease the carbohydrate/protein ratio, provides

Globally, there is a crisis related to the generation and disposal of
food waste (FW).1 In 2018, FW production reached about 1.3 bil-
lion metric tons worldwide, of which 95% ended at landfill sites
generating greenhouse gases and contributing to global warm-
ing. However, FW generation constitutes one opportunity to pro-
duce hydrogen (H2), a net-zero-emission biofuel. H2 production
by biological methods such as dark fermentation is more sustain-
able than conventional routes due to its lower requirements of
energy and chemicals.2 Because the composition of FW varies
depending on its origin (e.g., carbohydrate content, C/N ratio
and others), the H2 yield ranges from 50 to 150 L H2 kg−1 of vola-
tile solids (VS).3-5 One inherent challenge for producing H2 from
FW is its high carbohydrate/protein ratio, which results first in
the accumulation of volatile fatty acids (VFA) that decrease the
pH and finally inhibit the fermentative activity of H2-producing
bacteria.6 To solve this operational failure without consuming
large amounts of alkalis, the co-fermentation of FW with other
230
alkalinity and micronutrients, and dilutes toxic substances, result-
ing in the improvement of the H2 production.6-8 Other advan-
tages of the codigestion of wastes are the nutritional balance of
the substrates to obtain an optimal C/N ratio, the pH adjustment
to those values required for the bioprocess selected and the dilu-
tion of toxic molecules minimizing the growth inhibition; all these
factors improve the yield of biogas or fermentation products com-
pared with mono-digestion.9,10
* Correspondence to: Iván Moreno-Andrade, Laboratory for Research on
Advanced Processes for Water Treatment, Unidad Academica Juriquilla,
Instituto de Ingeniería, Universidad Nacional Autónoma de México, Blvd.
Juriquilla 3001, 76230 Querétaro, Mexico. E-mail: imorenoa@iingen.
unam.mx
Laboratory for Research on Advanced Processes for Water Treatment, Unidad
Academica Juriquilla, Instituto de Ingeniería, Universidad Nacional Autónoma
de México, Juriquilla, Mexico

© 2022 The Authors. Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical
Industry (SCI).
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.

Biohydrogen from codigestion of food waste and waste activated sludge
Wastewater treatment plants generate large amounts of waste
activated sludge (WAS), which also has the potential to produce
H2 by dark fermentation. However, due to its low carbohydrate/
protein ratio (<0.1) and high content of non-biodegradable mate-
rials, the resulting H2 yields are low: below 30 L kg−1 VS.6-8,11
Despite this, WAS possesses interesting properties for its co-
processing with FW, such as high alkalinity and trace metals act-
ing as micronutrients.12 The co-processing of FW with WAS has
been demonstrated to improve the H2 production rate compared
to the processing of the individual wastes.11-14 Codigestion of
WAS with other organic wastes has demonstrated to improve
sludge dewaterability, increase buffer capacity, accelerate initial
biogas production time and balance C/N ratio.10
Kim et al.14 determined an optimal FW–WAS ratio of 87–13%,
which increased the H2 production compared to the individual
processing of FW. Later, another study confirmed that a slight
addition of WAS (10:1 on a chemical oxygen demand basis) signif-
icantly enhanced the H2 fermentation performance.11 On the
other hand, Liu et al.13 reported that the co-processing of FW with
WAS did not improve the H2 yield. These contradictory results
may be related to the feedstocks’ initial characteristics such as
chemical oxygen demand (COD), volatile solids (VS), carbohydrate
content and C/N ratio, which vary depending on their origin, as
aforementioned. The C/N ratio is an important parameter since,
generally, FW has a higher C/N ratio than WAS, and co-
fermentation can be adjusted to an optimal C/N value (15–30);
the production of soluble proteins, carbohydrates and short-chain
fatty acid is enhanced consequently, and the dewaterability of
anaerobic digestate was improved.10
Previous studies have demonstrated the convenience of the
FW–WAS co-processing for improving H2 production; however,
all the studies were limited to batch tests.11-14 In continuous
reactors, there could be an inoculum adaptation, substrate
depletion or accumulation of putative toxic compounds, among
other phenomena present in batch fermentations. Therefore,
the potential of H2 production previously reported during the
co-processing of FW–WAS could be improved in continuous
reactors. The organic loading rate (OLR) is a key parameter for
H2 production from FW in continuous bioreactors.15 OLR influ-
ences the hydrolysis of particulate substrate, removal, and con-
version of carbohydrates, consequently leading to the efficient
production of H2 and VFA. As a result, this study aims to deter-
mine the optimal FW–WAS ratio for the improvement of H2 pro-
duction in batch tests and then to determine the effects of the
OLR on H2 productivity and microbial composition in a discon-
tinuous reactor operated for 110 days with the optimal
FW–WAS ratio.
EXPERIMENTAL
Feedstocks and inoculum
FW was obtained from a university cafeteria daily for 2 weeks;
only the fermentable matter was preserved at −20°C (bones
and inert materials such as plastic or paper were discarded). The
daily samples were mixed, homogenized and mashed with a
blender (1 HP grinder JR MJ22, Torrey Mexico), obtaining a parti-
cle size <0.5 mm. This mixture was stored for further experiments
at −20°C to avoid its auto-fermentation. WAS was obtained from a
conventional municipal wastewater treatment plant after dewa-
tering with a press and stored at 4°C; no further pretreatment
was applied. The characteristics of FW and WAS are listed in
Table 1.
J Chem Technol Biotechnol 2023; 98: 230–237 © 2022 The Authors.
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).
10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.soci.org
The inoculum used was an anaerobic sludge that was provided
by a brewery. This sludge was pretreated by drying at 105 °C for
24 h, ground and sifted to obtain a particle size <65 mm, select-
ing the H2-producing microorganisms according to several
authors.16-19 The characterization of the pretreated inoculum
showed the next values in g/ginoculum:total solids (TS)
0.98 ± 0.01, VS 0.69 ± 0.02, COD 2.13 ± 0.3, carbohydrates 0.07
± 0.01 and proteins 0.14 ± 0.03.
Hydrogen potential production test
Different combinations of FW–WAS (% VS) were evaluated to test
their effects on hydrogen potential production (HPP) at
10 g VS L−1: 90–10, 80–20, 70–30, 60–40, 50–50, 30–70 (ratios
are expressed on a VS basis). Treatments with individual wastes
of FW and WAS (100–0 and 0–100) were used for benchmarking
later. The HPP test was carried out by using an Automatic Meth-
ane Potential Test System equipment (AMPTS II, Bioprocess Con-
trol, Sweden). The standardized protocol for determining
biohydrogen potential reported by Carrillo-Reyes et al.19 was used
as a guideline. Batch reactors with a total volume of 600 mL and a
working volume of 460 mL with 140 mL of headspace were used.
Nutrients and alkalinity buffer were added according to Mizuno
et al.,20 and deoxygenated water was added to fill the total vol-
ume. The initial pH was adjusted to 7.0 using a solution of
0.1 mol L−1 NaOH according to Ramos et al.,18 and each batch test
was inoculated with 0.7 g of the pretreated inoculum. Intermittent
mixing at 120 rpm (1 min for every 3 min during the experiment)
was applied. Incubation was at a constant temperature of 37°C. All
treatments were run in triplicate. Endogenous H2 production by
the inoculum alone was subtracted from all treatments in the
HPP calculation.
Reactor experimental set-up
The FW–WAS ratio for which the highest HPP was reached was
applied to H2 production in a sequencing batch reactor (SBR).
The SBR was constructed as an acrylic cylinder with a reaction vol-
ume of 2 L, a working volume of 1.5 L and headspace of 0.5 L, and
an exchange volume of 50%. The reactor was inoculated with 20 g
anaerobic sludge pretreated at 105 °C for 24 h, inhibiting metha-
nogenic microorganisms. The inoculum was added once at the
initial operation of the reactor. Initial activation of the inoculum
was obtained in ten operation cycles (reaction time 12 h) using
15 g L−1 glucose and mineral medium: 41.6 g NH4Cl, 4 g
K2HPO4, 2 g MgCl2.6H2O, 1.6 g FeSO4·.7H2O, 40 mg CoCl2.6H2O,
40 mg MnCl2.4H2O, 40 mg KI, 8 mg NiCl2.6H2O, 8 mg ZnCl2.
After the inoculum activation, the experiments applying a fixed
feedstock at 15 g VS L−1 (FW–WAS ratio of 90–10) were carried
out. Reaction time was 12, 8 and 4 h, related to the OLR tested
(15, 22 and 45 g VS L d−1, respectively). The fill, settle and draw
phases were 5, 60 and 5 min, respectively. The reactor had deflec-
tors in the walls assuring a complete mixing. A temperature con-
troller based on a recycling water pump was used to maintain a
constant temperature of 37 ± 1°C inside the reactor (Thermo Sci-
entific, Waltham, MA, USA). pH was maintained at 5.5 by a control-
ler (pH 140 series, EUTECH Instruments, USA) using a combined
electrode (BNC, Sensorex), supplying NaOH 1 mol L−1. The stirrer
and pumps for filling and drawing were timed and controlled by
software programmed in LabVIEW (National Instruments, USA)
using a USB6008 data acquisition card (National Instruments,
USA) connected to a personal computer.
The biogas flow rate was measured online using a flowmeter
231
(ADM 1000, Agilent, Santa Clara, USA). The accumulated biogas
wileyonlinelibrary.com/jctb
 10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.soci.org I Moreno-Andrade, MJ Berrocal-Bravo, I Valdez-Vazquez

volume was obtained with respect to time by numerical integra- RESULTS AND DISCUSSION
tion implemented in LabVIEW. Biogas samples were taken daily Feedstocks (FW and WAS) characterization
to evaluate the percentage of H2. Table 1 shows the characterization of FW and WAS. The moisture
The cumulative H2 production (H) of the experimental data was content of both residues was higher than 60%, making them suit-
fitted to a modified Gompertz equation (Eqn 1): able for the wet anaerobic digestion process. FW was composed
   of 96% easy-to-degrade material with low alkalinity. These two
R max characteristics most likely result in the quick accumulation of
Hðt Þ=H max exp −exp ð⊗−tÞ+1 ð1Þ
H max VFA, acidification and, finally, inhibition of dark fermentation. On
the other hand, WAS had alkalinity threefold higher than
where H(t) (mL H2 L−1) is the total amount of hydrogen produced FW. The co-processing of FW with WAS indeed increases the sys-
during the experiment (h), Hmax (mL) is the maximum amount of tem's buffer capacity and reduces the process inhibition result of
hydrogen produced, Rmax (mL H2 h−1) is the maximum hydrogen the VFA accumulation. Also, WAS may be dosed during FW fer-
rate and ⊗ (h) is the lag time before the exponential hydrogen mentation for pH control between 5.5 and 7.0, optimizing H2
production. production,18-19 which would avoid the requirements for costly
alkalis. Trace metals were significantly higher in WAS. Some of
Analytical analysis these metals are necessary for metabolism, cellular growth, and
Biogas composition (CH4, CO2 and H2) was analyzed using an SRI activation and function of enzymes and coenzymes in dark fer-
8610C gas chromatograph (SRI Instruments, USA) equipped with mentation for hydrogen production.25 In particular, Fe and Ni
a thermal conductivity detector and a 30 m long (0.53 mm ID) are essential for the biosynthesis of hydrogenases, which are
Carboxen-1010 PLOT column. The operating conditions were set responsible for the reduction of protons, resulting in the forma-
as follows: the carrier gas was nitrogen at a flow rate of tion of H2.26 Also, WAS has a higher concentration of Ca and Al,
4.5 mL min−1; the temperature of the injector was 200 °C; the col- which are metals that help growth and cell retention, so they
umn temperature was 100 °C; and the temperature of the detec- are necessary to avoid the washout of cells from continuous
tor was fixed at 230 °C. bioreactors.27
Contents of VFA (acetic, butyric, propionic and valeric acids) and
solvents (ethanol and butanol) were quantified using a gas chro-
Hydrogen potential production test
matograph (Varian model 3300, CA, USA) coupled to a flame ion-
Figure 1 shows (A) lag time (⊗), (B) Rmax (maximum hydrogen pro-
ization detector and equipped with a 15 m long (0.53 mm id)
duction rate) and (C) Hmax (total amount of hydrogen produced)
Zebron ZB-FFAP column. The injector and detector temperatures
for the individual wastes and the FW–WAS ratios under evalua-
were maintained at 190 and 210 °C, respectively. Column temper-
tion. In general terms, the co-processing of FW–WAS improved
ature was maintained at 45 °C for 1.5 min, after which it was
the fermentation performance. The first beneficial effect of WAS
increased to 135 °C at a rate of 8 °C min−1. The carrier gas was
was the immediate reduction of the lag time (⊗) from 17 h in
nitrogen at 9.5 mL min−1. The lactate content was determined
the individual FW to 5 h for the FW–WAS ratios between 70–30
by ion chromatography using a Dionex ICS-1500 system with an
REIC IonPac AS23 250 × 4 mm column. The eluent, 0.8 mmol L−1
NaHCO3, was mixed with 4.5 mmol L−1 Na2CO3 using an isocratic
flow of 3 mL min−1 at 30 °C.
The COD was measured spectrophotometrically using Hach Table 1. Physicochemical characterization of the feedstocks
methods according to the manufacturer's instructions (HACH,
Parameter FW WAS
CO, USA). TS and VS were determined according to standard
methods.21 Total carbohydrates were measured by the phenol- TS (g/g) 0.37 0.16
sulfuric acid method.22 All the results were analyzed by an ANOVA VS (g/g) 0.35 0.09
with a post-hoc Tukey test with Rstudio (Desktop 2022.07.2+576) VS/TS 95.6 56.2
to compare them among treatments. Alkalinity (mg CaCO3 L−1) 160 535
CODtotal (mg g−1 TS) 1212 1200
Microbial community analysis CODsoluble (mg g−1 TS) 356 168
During the SBR operation, cell samples were collected to char- Carbohydratestotal (mg g−1 TS) 198 42
acterize the microbial composition and preserved at −4°C Carbohydratessoluble (mg g−1 TS) 152 5
pending analysis. Genomic DNA was extracted from cell sam- Proteinstotal (mg g−1 TS) 236 119
ples using the PowerSoil DNA isolation kit (MO BIO Laborato- Proteinssoluble (mg g−1 TS) 57.3 —
ries, Carlsbad, CA, USA) according to the manufacturer's Density (kg L−1) 1.04 0.91
instructions. The DNA concentration was quantified by spec- pH 3.9 8.4
trophotometry using a NANODrop 2000c (Thermo Scientific, Calcium (mg L−1) 7.3 197
USA). The extracted DNA was submitted to RTL Genomics Aluminum (mg L−1) 1.1 31
(Lubbock, TX, USA) for Illumina MiSeq sequencing using the Iron (mg L−1) 0.5 56
universal primers 515F (5´GTGCCAGCMGCCGCGGTAA) and Zinc (mg L−1) 1.4 1.7
806R (5GGACTACHVGGGTWTCTAAT) of the 16S rDNA gene. Cupper (mg L−1) 0.01 1.0
The pipeline procedure for the sequence analysis and further Manganese (mg L−1) 0.2 0.7
analysis of the identified operational taxonomic units have Nickel (mg L−1) 0.01 0.5
been explained previously.23 Pearson's correlations between Potassium (mg L−1) 27 14
the bacterial populations and process parameters were carried Sodium(mg L−1) 127 78
232

out with PAST software.24

wileyonlinelibrary.com/jctb © 2022 The Authors. J Chem Technol Biotechnol 2023; 98: 230–237
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).
 10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biohydrogen from codigestion of food waste and waste activated sludge www.soci.org

Figure 1. Kinetic parameters and H2 production performance for each of the FW–WAS ratios in batch test: (A) lag time; (B) Rmax; (C) Hmax; (D) metabolite
production at different FW–WAS ratios. The dotted line represents the adjustment to a fourth-order polynomial equation.

and 30–70 (P < 0.001). Rmax had contrasting values for the individ- Hydrogen production from FW–WAS codigestion in
ual wastes, almost ten times higher for FW than WAS, with a the SBR
decreasing trend as WAS increased in the FW–WAS ratios. Hmax The inoculum was activated with glucose according to Castillo-
showed a similar decrease as WAS increased in the FW–WAS Hernández et al.,17 and the FW–WAS ratio of 90–10 was then fed
ratios; nevertheless, the 90–10 ratio showed a higher Hmax than to the SBR. The reactor was operated for 110 cycles at different
the individual wastes and the other FW–WAS ratios (P < 0.01). OLRs: 15 g VS L−1 d−1 (cycles 10–40), 22 g VS L−1 d−1 (cycles
The individual WAS had a very low Hmax, a result of its low avail- 50–80) and 45 g VS L−1 d−1 (cycles 80–110). Figure 2 shows the
ability of carbohydrates compared to FW. VFA production was variations in compositions of H2, CH4 and CO2 during the SBR
notably higher in the 90–10 ratio compared to the other treat- operation. The H2 percentage averaged 46 ± 15% during the
ments. It is important to highlight that WAS addition did not glucose activation, decreasing with the substrate change to
change the composition of the main produced VFA (Fig. 1D), FW–WAS. The change in OLR influenced the H2 percentages; the
which were acetate and butyrate, both related to H2 production.26 OLR of 22 g VS L−1 d−1 resulted in the highest H2 percentage of
Nonetheless, WAS addition reduced the amount of ethanol, which 33 ± 8% (Fig. 2), while the increase in OLR resulted in the lowest
possibly was produced by the activation of solventogenesis in the H2 percentage of 14 ± 8% (P > 0.05).
H2-producing bacteria, resulting from VFA accumulation in ratios
with higher content of FW.
These results agreed with those obtained by Kim et al.,14
who proposed an optimal codigestion FW–sewage sludge Table 2. Volatile solids and carbohydrates removal in H2 potential
ratio of 87–13% for improvement of the H2 production in production test
comparison with the individual FW. A small dose of WAS
Mixture ratio VS Carbohydrate
improved the H2 production, most likely due to its contribu-
FW–WAS (%) removal (%) removal (%)
tion of alkalinity, which stabilized the pH of the system, and
micronutrients like Ni and Fe, which are important for the 100–0 53 ± 2.1 93 ± 0.5
biosynthesis of enzymes such as hydrogenases, responsible 90–10 51 ± 1.6 95 ± 1.2
for H2 production.14,28,29 80–20 49 ± 2.2 91 ± 2.0
Table 2 shows the removal of VS and carbohydrates. The effi- 70–30 42 ± 4.3 88 ± 4.0
ciency of VS removal decreased as WAS increased in the ratios. 60–40 44 ± 3.8 95 ± 3.7
VS removal was in the same order as that previously reported 50–50 36 ± 5.1 89 ± 1.5
for similar FW–WAS ratios.11-13 Carbohydrate removal averaged 30–70 23 ± 7.2 90 ± 3.5
91.6% in all the treatments, independent of the FW–WAS 0–100 12 ± 5.3 92 ± 2.3
233

ratio (P < 0.05).

J Chem Technol Biotechnol 2023; 98: 230–237 © 2022 The Authors. wileyonlinelibrary.com/jctb
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).

www.soci.org
Figure 2. Biogas composition during SBR operation at different OLRs.
Figure 3. (A) Kinetics of the H2 production at different OLRs. The dotted line
represents the Gompertz model adjustment. (B) Effect of OLR on Rmax, Hmax
and YH2. The dotted line represents the adjustment to a second-order polyno-
234
mial equation. (C) Metabolite production at different OLRs in the SBR system.
wileyonlinelibrary.com/jctb © 2022 The Authors.
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).
10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
I Moreno-Andrade, MJ Berrocal-Bravo, I Valdez-Vazquez
Figure 3(A) shows the cumulative H2 production fitting the
modified Gompertz model (dotted line) for each SBR cycle, while
Table 3 shows the kinetic parameters and removal obtained at dif-
ferent OLRs. The results showed that the maximal value of Hmax
and Rmax (235 ± 27 mL H2 Lreactor−1 and 62 ± 15
mL H2 Lreactor−1 h−1, respectively) were obtained when an OLR
of 22 g VS L−1 d−1 was applied. The lag time did not show differ-
ences independently of the OLR applied. These results contrast
with results obtained in a batch test of FW–WAS codigestion,
where the lag time ranged from 3 to 18.7 h.11,13 The reduction
of lag time in the long-term operation of discontinuous reactors
is related to the microbial acclimatization of the microorganisms
and the presence of remnant substrate from previous cycles – in
our case, an exchange volume of 50%. The lag time results agree
with those obtained in SBR systems producing H2 from FW as
mono-substrate.30,31
The H2 productivities mirror the H2 percentages where the OLR
of 22 g VS L−1 d−1 resulted in the highest productivity of 733
± 282 mL H2 Lreactor−1 d−1, and the increase in OLR resulted in a
decrease in productivity to 383 mL H2 Lreactor−1 d−1 (Table 3).
From cycle 29, CH4 began to be detected in the biogas (5 ±
2.6%). Methane production is related to the presence of hydroge-
notrophic methanogens that consume H2 and CO229 and are
entered into the SBR with the non-sterile feedstocks.17,32
The maximum hydrolysis was obtained with an OLR of
15 g VS L−1 d−1 (50%) (Table 3). This agrees with previous results
using only FW,15 inferring that WAS does not affect the system's
hydrolysis percentage. Hydrolysis can be increased if the HRT
and solids retention time (SRT) are increased. However, the
increase in HRT and SRT increases the abundance of methano-
gens (growth rates are between 0.0167 and 0.02 h−1), reducing
H2 production.15,20 Efficient hydrolysis contributes to VS removal
and the transformation of particulate material into soluble
(e.g., COD), increasing its availability to microorganisms. Total car-
bohydrates removal showed no significant differences comparing
the three OLR tested in our study (>80%).
The OLR influences Hmax, Rmax and YH2 (Fig. 3B), demonstrating
the process's best performance at an OLR of 22 g VS L−1 d−1,
where the H2 production and YH2 were higher than reported in
a fermentative continuously stirred tank reactors (CSTR) using a
FW–WAS rate of 1:5 with a hydraulic retention time (HRT) of
3 days, reporting 8.6 ± 4.8 mL H2 g−1 VS.33 The tendency (adjust-
ment of the results to a polynomial model) of Fig. 3(B) shows that
Hmax, Rmax and YH2 can be optimized, possibly at an OLR in a range
of 25–35 g VS L−1 d−1; however, optimization experiments need
J Chem Technol Biotechnol 2023; 98: 230–237
 10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Biohydrogen from codigestion of food waste and waste activated sludge www.soci.org

Table 3. Kinetic parameters and removal at different OLR

OLR (g VS L−1 d−1) 15 22 45

Hmax (mL Lreactor−1) 140 ± 15 235 ± 27 43 ± 4

Rmax (mL Lreactor−1 h−1) 44 ± 8 62 ± 15 13 ± 4
⊗ (h) 1.6 ± 0.2 1.2 ± 0.1 1.2 ± 0.3
YH2 (mL H2 g−1 VS) 9.3 ± 1.0 15.6 ± 1.8 2.8 ± 0.2
Hydrolysis (%) 50 ± 5 37 ± 7 27 ± 5
Carbohydrate removal (%) 80.4 ± 13.0 80.8 ± 9.2 86.8 ± 15.2
VS removal (%) 61.3 ± 5.3 66.1 ± 4.5 45.2 ± 7.0
CODtotal removal (%) 34 ± 5 15 ± 7 12 ± 4
Productivity 364 ± 120 733 ± 282 383 ± 41
(mL H2 Lreactor−1 d−1)

Figure 5. Pearson's correlation between the dominant bacterial popula-

tions and process parameters.

generation of propionate (838 ± 87 mg L−1), a metabolite pro-

Figure 4. Relative abundance of the community genera in the SBR oper-
ated at different OLRs.
to be proposed to confirm it. The difference in the possible opti-
mal OLR, compared with others in literature, can be attributed
to the FW–WAS ratio, operation regime and OLR applied. Optimi-
zation of these parameters will contribute to the increase in H2
production and energy recovery in continuous and discontinuous
reactors.
Metabolite production (Fig. 3C) shows that the main VFA in all
the HRT was acetate, followed by butyrate (except in the OLR of
15 g VS L−1 d−1, where the propionate was the second main
VFA). The final H2 yield depends on the primary metabolite path-
way, the acetate pathway provides the maximum yield of
4 mol H2 molglucose−1, while the butyrate and ethanol pathways
are limited to 2 mol H2 molglucose−1.34 Isobutyrate and isovalerate
were detected in low concentrations (<30 mg L−1) independently
of the OLR. Caproate was only detected in the OLR of
22 g VS L−1 d−1 at a low concentration (<20 mg L−1). The highest
acetate production (1510 ± 155 mg L−1) was obtained with an
OLR of 15 g VS L−1 d−1. However, H2 production was not the
highest under this condition. The accumulation of acetate in the
medium does not necessarily imply a higher production of H2
since several microbial species can convert H2 and CO2 into ace-
tate. These species include homoacetogenic microorganisms that
reduce CO2 to acetate through the acetyl-CoA pathway. Addition-
duced from the consumption of substrate and hydrogen.35
By increasing the OLR from 15 to 22 g VS L−1 d−1, the metabolic
pathways were dominated by the acetic and butyric acid fermen-
tations. This OLR increases butyrate production (604
± 90 mg L−1) and maintains the acetic fermentation (producing
1178 mg L−1). In this case, acetic consumption could be low since
propionate generation decreased to 235 ± 30 mg L−1. This fact
was reflected in higher H2 production (Hmax and Rmax). The OLR
of 45 g VS L−1 d−1 showed a decrease in VFA generation, increas-
ing lactic acid production. In this case, there was no predominant
fermentation pathway since each metabolite (acetate, butyrate,
propionate and lactate) was produced at a similar concentration
range. The presence of lactate and ethanol promoted the con-
sumption of substrate not associated with the H2 generation
and has been reported in the fermentation of FW.14,15,17,30
The microbial characterization showed that Prevotella was the
dominant genus independently of the OLR (Fig. 4). Prevotella
spp. is a non-spore-forming, obligate anaerobic bacteria with
hemicellulase activity whose fermentation end products consist
of acetic and succinic acids and smaller quantities of lactic acid
and H2.36
When the OLR increased from 15 to 22 g VS L−1 d−1, Mega-
sphaera became the second dominant species. Megasphaera has
been well recognized as an H2 producer from lactic and acetic
acids.37 Both genera have been previously reported as the domi-
nant genera in bioreactors that produce H2 from FW,15,17,31,34-39
indicating that FW–WAS co-processing did not change the most
abundant members in the microbial community. The increase in
OLR from 22 to 45 g VS L−1 d−1 stimulated the abundance of Mit-
suokella spp. to 21%. Mitsuokella spp. are non-spore-forming,
strictly anaerobic, fermentative bacteria previously reported in
H2-producing reactors.40 Mitsuokella is able to hydrolyze phytate,
the principal storage form of phosphorus in foods.41 It is possible
to infer that the higher availability of phytate in the SBR operated
at an OLR of 45 g VS L−1 d−1 stimulated the growth of Mitsuokella,
235

ally, H2 production under this condition was affected by the a fact that needs further research.

J Chem Technol Biotechnol 2023; 98: 230–237 © 2022 The Authors. wileyonlinelibrary.com/jctb
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).

www.soci.org
Interestingly, Megasphaera and Olsenella coexisted in the SBR,
being positively correlated with H2 production (Fig. 5). Olsenella
spp. are anaerobic bacteria that produce lactic acid as the major
end product with no hydrolytic activity.42 Therefore, Olsenella
could be providing lactic acid to Megasphaera, resulting in the
highest H2 production at an OLR of 22 g VS L−1 d−1. Two lactic
acid bacteria – Lactobacillus and Lactococcus – also coexisted with
Megasphaera, most likely providing them substrate (lactic acid) for
H2 production. On the other hand, Enterococcus and Veillonella
were positively correlated, with the substrate hydrolysis being
more abundant at an OLR of 15 g VS L−1 d−1. Enterococcus
degrades complex polysaccharides like xylan,43 while Veillonella
produces acetic and propionic acids as the major end products.44
In a previous study, a synthetic consortium formed by Enterococ-
cus (non-amylolytic), Veillonella (non-amylolytic) and Eubacterium
(amylolytic) succeed in hydrolyzing and fermenting 97% of amy-
lomaize starch granules that contained a resistant starch frac-
tion.45 In this synthetic consortium, Veillonella hydrolyzed
maltose and maltodextrins and utilized lactic acid excreted by
the other members. In the SBR, Veillonella may be involved in
the solubilization of the most recalcitrant starch in FW at an OLR
of 15 g VS L−1 d−1, contributing to the highest hydrolysis rate of
50%. An interesting fact was that the abundance of Enterococcus
and Veillonella increased with Prevotella abundance. Previous
studies found that Veillonella along with Prevotella form part of
the oral microbiota forming biofilms.45 This background, along
with our results, indicated that these three species were associ-
ated possibly with biofilms hydrolyzing the different fractions of
carbohydrates present in FW. Finally, Prevotella, despite its well-
known hydrolytic activity, was not correlated with the substrate
hydrolysis or carbohydrate removal. This phenomenon may be
related to its broad spectrum of hydrolytic enzymes, including
xylanases,43 glycosidases, lipases and proteinases.46 Therefore,
the convenient activity of Prevotella was persistent with the OLR
changes, emerging as a potential keystone species that calls for
further research.
CONCLUSIONS
In batch tests, a FW–WAS ratio of 90–10 increased the H2 produc-
tion (Hmax) by 22% compared to the individual FW. The SBR oper-
ation showed that OLR influenced Hmax and Rmax, obtaining the
best performance at an OLR of 22 g VS L−1 d−1. The cooperative
interaction between Olsenella and Megasphaera resulted in the
highest H2 production from FW–WAS. The OLR changes pro-
moted different taxa of hydrolytic bacteria, while Enterococcus
and Veillonella were stimulated at low OLR, and Mitsuokella was
stimulated at an OLR as high as 45 g VS L−1 d−1. The wide spec-
trum of hydrolytic activities of Prevotella possibly allowed its prev-
alence in the SBR.
ACKNOWLEDGEMENTS
The financial support by DGAPA-UNAM through project PAPIIT
IN102722 is acknowledged. Gloria Moreno, Angel Hernández
and Jaime Perez are acknowledged for their technical assistance.
CONFLICT OF INTEREST
236
The authors declare no conflicts of interest.
wileyonlinelibrary.com/jctb © 2022 The Authors.
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).
10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
I Moreno-Andrade, MJ Berrocal-Bravo, I Valdez-Vazquez
REFERENCES
1 Melikoglu M, Lin C and Webb C, Analyzing global food waste problem:
pinpointing the facts and estimating the energy content. Open Eng
3:157–164 (2013).
2 Sanchez A, Ayala OR, Hernandez-Sanchez P, Valdez-Vazquez I and de
León-Rodríguez A, An environment-economic analysis of hydrogen
production using advanced biorefineries and its comparison with
conventional technologies. Int J Hydrogen Energ 45:7994–28006
(2020).
3 Kim SH and Shin HS, Effects of base-pretreatment on continuous
enriched culture for hydrogen production from food waste. In J
Hydrogen Energ 33:5266–5274 (2008).
4 Pan J, Zhang R, El-Masahd HM, Sun H and Ying Y, Effect of food to
microorganism ratio on biohydrogen production from food waste
via anaerobic fermentation. Int J Hydrogen Energ 33:6968–6975
(2008).
5 Kim DH, Kim SH and Shin HS, Hydrogen fermentation of food waste
without inoculum addition. Enzyme Microb Technol 45:181–187
(2009).
6 Zhang P, Liu C, Zheng Y, Zhao Y and Zhen G, Statistical key factor opti-
mization of conditions for biohydrogen production from sewage
sludge and food waste by anaerobic codigestion. Energy Fuel 33:
11163–11172 (2019).
7 Siddiqui Z, Horan NJ and Salter M, Energy optimisation from co-
digested waste using a two-phase process to generate hydrogen
and methane. Int J Hydrogen Energ 36:4792–4799 (2011).
8 Xu R, Zhang K, Liu P, Khan A, Xiong J, Tian F et al., A critical review on
the interaction of substrate nutrient balance and microbial commu-
nity structure and function in anaerobic co-digestion. Bioresour
Technol 247:1119–1127 (2018).
9 Chen S, Tao Z, Yao F, Wu B, He L, Hou K et al., Enhanced anaerobic co-
digestion of waste activated sludge and food waste by sulfidated
microscale zerovalent iron: insights in direct interspecies electron
transfer mechanism. Bioresour Technol 316:123901 (2020).
10 Yang Q, Wu B, Yao F, He L, Chen F, Ma Y et al., Biogas production from
anaerobic co-digestion of waste activated sludge: co-substrates and
influencing parameters. Rev Environ Sci Biotechnol 18:771–793
(2019).
11 Kim DH, Kim SH, Shin HS, Kim MS and Shin HS, Sewage sludge addition
to food waste synergistically enhances hydrogen fermentation per-
formance. Bioresour Technol 102:8501–8506 (2011).
12 Mamimin C, Probst M, Gómez-Brandón M, Podmirseg ME, Insam H,
Reungsang A et al., Trace metals supplementation enhanced micro-
biota and biohythane production by two-stage thermophilic fer-
mentation. Int J Hydrogen Energ 44:3325–3338 (2019).
13 Liu X, Li R, Ji M and Han L, Hydrogen and methane production by co-
digestion of waste activated sludge and food waste in the two-stage
fermentation process: substrate conversion and energy yield. Biore-
sour Technol 146:317–323 (2013).
14 Kim SH, Han SK and Shin HS, Feasibility of biohydrogen production by
anaerobic co-digestion of food waste and sewage sludge. Int J
Hydrogen Energ 29:1607–1616 (2004).
15 Santiago SG, Morgan-Sagastume JM, Monroy O and Moreno-
Andrade I, Biohydrogen production from organic solid waste in a
sequencing batch reactor: an optimization of the hydraulic and
solids retention time. Int J Hydrogen Energ 45:25681–25688 (2020).
16 Li Z, Chen Z, Ye H, Wang Y, Luo W, Chang JS et al., Anaerobic co-
digestion of sewage sludge and food waste for hydrogen and VFA
production with microbial community analysis. Waste Manag 78:
789–799 (2018).
17 Castillo-Hernández A, Mar-Alvarez I and Moreno-Andrade I, Start-up
and operation of continuous stirred-tank reactor for biohydrogen
production from restaurant organic solid waste. Int J Hydrogen Energ
40:17239–17245 (2015).
18 Ramos C, Buitrón G, Moreno-Andrade I and Chamy R, Effect of the ini-
tial total solids concentration and initial pH on the biohydrogen pro-
duction from cafeteria food waste. Int J Hydrogen Energ 37:13288–
13295 (2012).
19 Carrillo-Reyes J, Tapia-Rodríguez A, Buitrón G, Moreno-Andrade I,
Palomo-Briones R, Razo-Flores E et al., A standardized biohydrogen
potential protocol: an international round robin test approach. Int
J Hydrogen Energ 44:26237–26247 (2019).
20 Mizuno O, Dinsdale R, Hawkes FR, Hawkes DL and Noike T, Enhance-
ment of hydrogen production from glucose by nitrogen gas sparg-
ing. Bioresour Technol 73:59–65 (2000).
J Chem Technol Biotechnol 2023; 98: 230–237

Biohydrogen from codigestion of food waste and waste activated sludge
21 APHA. Standard methods for the examination of water and wastewa-
ter, 21st ed. Am Water Work Assoc Public Work Assoc Environ Fed.
Washington, DC, USA. (2005).
22 Dubois M, Gilles KA, Hamilton JK, Rebers PA and Smith F, Colorimetric
method for determination of sugars and related substances. Anal
Chem 28:350–356 (1956).
23 Barragán-Trinidad M, Carrillo-Reyes J and Buitrón G, Hydrolysis of
microalgal biomass using ruminal microorganisms as a pretreat-
ment to increase methane recovery. Bioresour Technol 244:100–
107 (2017).
24 Hammer Ø, Harper DAT and Paul DR, Past: paleontological statistics
software package for education and data analysis. Palaeontol Elec-
tron 4:art 4 (2001).
25 Balachandar G, Khanna N and Das D, Biohydrogen production from
organic wastes by dark fermentation, in Biohydrogen, ed. by
Pandey A, Chang J-S, Hallenbeck PC and Larroche C. Elsevier,
Burlington, MA, USA, pp. 103–144 (2013).
26 Hallenbeck P, Fundamentals on Biohydrogen, in Biohydrogen, Vol. 588,
1st edn, ed. by Pandey A, Chang JS, Hallenbeck P and Larroche C.
Elsevier, Netherlands (2013).
27 Yuan Z, Yang H, Zhi X and Shen J, Increased performance of continu-
ous stirred tank reactor with calcium supplementation. Int J Hydro-
gen Energ 7:2622–2626 (2010).
28 Yang H and Shen J, Effect of ferrous iron concentration on anaerobic
bio- hydrogen production from soluble starch. Int J Hydrogen Energ
31:2137–2146 (2006).
29 Zabranska J and Pokorna D, Bioconversion of carbon dioxide to meth-
ane using hydrogen and hydrogenotrophic methanogens. Biotech-
nol Adv 36:707–720 (2018).
30 Kim SH and Shin HS, Effects of base-pretreatment on continuous
enriched culture for hydrogen production from food waste. Int J
Hydrogen Energ 19:5266–5274 (2008).
31 Baldi F, Pecorini I and Iannelli R, Comparison of single-stage and two-stage
anaerobic co-digestion of food waste and activated sludge for hydro-
gen and methane production. Renew Energy 143:1755–1765 (2019).
32 Santiago SG, Trably E, Latrille E, Buitrón G and Moreno-Andrade I, The
hydraulic retention time influences the abundance of Enterobacter,
clostridium, and lactobacillus during the hydrogen production from
food waste. Lett Appl Microbiol 69:138–147 (2019).
33 Moreno-Andrade I, Carrillo-Reyes J, Santiago SG and Bujanos-
Adame MC, Biohydrogen from food waste in a discontinuous pro-
cess: effect of HRT and microbial community analysis. Int J Hydrogen
Ener 48:17246–17252 (2015).
J Chem Technol Biotechnol 2023; 98: 230–237 © 2022 The Authors.
Journal of Chemical Technology and Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry (SCI).
10974660, 2023, 1, Downloaded from https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/jctb.7238 by Birla Institute of Technology & Science, Wiley Online Library on [09/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.soci.org
34 Cabrol L, Morone A, Tapia-Venegas E, Steyer J-P and Ruiz-Filippi TE,
Microbial ecology of fermentative hydrogen producing biopro-
cesses: useful insights for driving the ecosystem function. FEMS
Microbiol Rev 41:158–181 (2017).
35 Motte JC, Trably E, Escudié R, Hamelin J, Steyer JP, Bernet N et al., Total
solids content: a key parameter of metabolic pathways in dry anaer-
obic digestion. Biotechnol Biofuels 6:164 (2013).
36 Wu CC, Johnson JL, Moore WEC and Moore LVH, Emended descriptions
of Prevotella denticola, Prevotella loescheii, Prevotella veroralis and
Prevotella melaninogenica. Int J Syst Bacteriol 42:536–541 (1992).
37 Ohnishi A, Abe S, Bando Y, Fujimoto N and Suzuki M, Rapid detection
and quantification methodology for genus Megasphaera as a hydro-
gen producer in a hydrogen fermentation system. Int J Hydrogen
Energ 37:2239–2247 (2012).
38 Feng K, Li H and Zheng C, Shifting product spectrum by pH adjustment
during long-term continuous anaerobic fermentation of food waste.
Bioresour Technol 270:180–188 (2018).
39 Mariakakis I, Bischoff P, Krampe J, Meyera C and Steinmetz H, Effect of
organic loading rate and solids retention time on microbial popula-
tion during bio-hydrogen production by dark fermentation in large
lab-scale. Int J Hydrogen Energ 36:10690–10700 (2011).
40 Muñoz-Paez K, Alvarado-Michi EL, Moreno-Andrade I, Buitrón G and
Valdez-Vazquez I, Comparison of suspended and granular cell
anaerobic bioreactors for hydrogen production from acid agave
bagasse hydrolyzates. Int J Hydrogen Energ 45:275–285 (2020).
41 Lan GQ, Ho YW and Abdullah N, Mitsuokella jalaludinii sp. nov., from the
rumens of cattle in Malaysia. Int J Syst Evol Microbiol 52:713–718
(2002).
42 Han KI, Lee KC, Eom MK, Kim JS, Suh MK, Park SH et al., Olsenella faeca-
lis sp. nov., an anaerobic actinobacterium isolated from human fae-
ces. Int J Syst Evol Microbiol 69:2323–2328 (2019).
43 Valdez-Vazquez I, Pérez-Rangel M, Tapia A, Buitrón G, Molina C,
Hernández G et al., Hydrogen and butanol production from native
wheat straw by synthetic microbial consortia integrated by species
of enterococcus and clostridium. Fuel 159:214–222 (2015).
44 Jumas-Bilak E, Carlier JP, Jean-Pierre H, Teyssier C, Gay B, Campos J
et al., Veillonella montpellierensis sp. nov., a novel, anaerobic, Gram-
negative coccus isolated from human clinical samples. Int J Syst Evol
Microbiol 54:1311–1316 (2004).
45 Kolenbrander P, The genus Veillonella. PRO 4:1022–1040 (2006).
46 Gharbia SE and Shah HN, Hydrolytic enzymes liberated gram-
negative anaerobes. FEMS Immunol Med Microbiol 6:139–146
(1993).
237
wileyonlinelibrary.com/jctb