| Journal of Collective Sciences and Sustainability
Received: 12 August 2026; Revised: 03 September 2026; Accepted: 07 September 2026; Published Online: 08 September 2026.
J. Collect. Sci. Sustain., 2026, 2(3), 26408 | Volume 2 Issue 3 (September 2026) | DOI: https://doi.org/10.64189/css.26408
© The Author(s) 2026
This article is licensed under Creative Commons Attribution NonCommercial 4.0 International (CC-BY-NC 4.0).
Machine Learning-Based Prediction and Optimization
of Membrane Fouling in a Hybrid Biochar-Membrane
System for Treating Maize Flour Mill Wastewater
Afeez Oladeji Amoo*
Department of Environmental Sciences, Faculty of Physical Sciences, Federal University Dutse, PMB 7156 Dutse, Jigawa State, Nigeria
*Email: afeezoladeji@fud.edu.ng (Afeez Oladeji Amoo)
Abstract
Agro-processing mill wastewater has a high organic load, which negatively impacts the environment. In this
study, a hybrid coconut-shell-derived biochar (CSDB) membrane system was developed, and machine learning
models were used to predict and optimize the removal of pollutants and membrane fouling. CSDB with a
porosity of 68.7 ± 1.2%, a pH of 9.2 ± 0.1, a BET surface area of 512 ± 18 m
2
/g, and a total carbon content of 83
± 1.5% was prepared by pyrolysis at 500 °C and steam activation at 700 °C. Batch experiments demonstrated
maximum removal efficiencies of 91.30 ± 1.8% for Cd (pH 7.5), 90.91 ± 2.1% for Pb (pH 6.5), and 67.71 ± 2.4%
for Cr (pH 3.5). The adsorption kinetics were best described by the pseudo-second-order model (R
2
> 0.98),
while the pseudo-first-order model yielded lower R
2
values (0.72–0.85), supporting a chemisorption-
dominated mechanism. Compared with membrane-only operation, fouling resistance was reduced by 35–45%
when coupled with CSDB pretreatment. Experimental data from a two-stage experimental design were used to
train artificial neural network (ANN), extreme gradient boosting (XGBoost), random forest (RF), and support-
vector regression (SVR) models. The best predictive accuracy (test-set R
2
= 0.91–0.96, RMSE = 1.8–3.4%, MAE
= 1.4–2.7%) was obtained by the XGBoost and ANN models. The optimal conditions for removing 85–92% of
the metals and simultaneously reducing fouling were determined by multi-objective optimization (CSDB dosage
of 12–15 g/L, contact time of 90–105 min, and transmembrane pressure of 1.0–1.1 bar). The predictions were
validated experimentally, with relative errors of ≤ ±6%. The preliminary technoeconomic assessment, which is
based on a 10 m
3
/day modular unit, a membrane lifetime of 18–24 months, and an energy consumption of 0.85–
1.1 kWh m
-3
, estimates a treatment cost of approximately ₦450/US$0.27 m
-3
. This hybrid CSDB–membrane
process coupled with machine learning offers a convenient and cost-effective pathway for the sustainable
treatment of maize flour mill effluents and agricultural reuse.
Keywords: Coconut shell biochar; Membrane fouling; Machine learning; Maize mill wastewater; Heavy metal
adsorption; Optimization.
1. Introduction
Population growth and industrialization have significantly increased the volume of discharged untreated or
poorly treated industrial wastewater, exacerbating the pressure on soil and water resources.
[1,2]
In agro-
industrial processes such as maize flour milling, substantial volumes of effluent containing high levels of
biodegradable organics, suspended solids, nutrients and toxic heavy metals are generated.
[3,4]
In northern
Nigeria, effluent from maize-processing plants is frequently discharged with minimal or no treatment, leading
to eutrophication, oxygen depletion and potential health risks through contaminated irrigation water.
[5,6]
Maize
mill wastewater typically has moderate-to-high chemical oxygen demand (COD) and high turbidity and heavy
metal concentrations that exceed the World Health Organization (WHO) and National Environmental Standards
and Regulations Enforcement Agency (NESREA) limits.
[7]
Conventional treatment trains that rely on aluminum-
or iron-based coagulants and commercial activated carbon are often prohibitively expensive for small- and
medium-scale operators, generate secondary sludge and show limited efficiency for mixed organic–metal
matrices.
[8,9]
Consequently, interest in low-cost, locally available adsorbents derived from agricultural residues
has increased.
Pyrolytic coconut-shell biochar is characterized by a high surface area, alkaline surface chemistry and abundant
oxygen-containing functional groups that facilitate ion exchange, surface complexation and π–π interactions
with metal ions.
[10,11]
Previous work has demonstrated that optimized batch conditions can achieve >90%
removal of Cd and Pb from maize-mill effluent, with kinetics consistent with chemisorption. However, residual
colloidal material and dissolved organics can still contribute to downstream membrane fouling. Membrane
filtration provides an effective barrier to residual suspended solids and macromolecules, yet membrane fouling
remains a critical operational constraint that increases energy consumption, cleaning frequency and
replacement costs.
[12]
Biochar pretreatment has been shown to reduce the fouling potential by lowering the load
of organic and particulate foulants reaching the membrane. Machine learning (ML) methods, particularly
artificial neural network (ANN), random forest (RF), extreme gradient boosting (XGBoost) and support-vector
regression (SVR), have demonstrated a strong ability to predict membrane fouling and optimize
multiparameter treatment processes.
[13,14]
The present study therefore develops and experimentally validates a hybrid coconut-shell biochar membrane
system for real maize flour mill wastewater collected at Dutse, Nigeria. The specific objectives were (i)
physicochemical characterization of untreated effluent and the prepared biochar, including BET surface area,
surface morphology and functional-group analysis; (ii) quantification of heavy-metal and organic-load removal
under batch and hybrid conditions with statistical reporting of uncertainty; (iii) assessment of membrane
fouling behavior with and without biochar pretreatment; (iv) development, hyperparameter tuning and
benchmarking of four ML models (ANN, RF, XGBoost, and SVR) using a transparent two-stage design of
experiments; and (v) multi-objective optimization, experimental validation of the optimal operating window,
and a preliminary technoeconomic assessment with clearly stated assumptions. This work addresses a critical
gap by coupling low-cost local adsorbent production with data-driven fouling prediction and optimization for
a real agro-industrial matrix under conditions relevant to resource-constrained settings.
2. Materials and methods
2.1. Study area and wastewater sampling
The study was conducted at the maize flour mill industry in Godiya Miyyeti Estate, Takur, Dutse, capital of
Jigawa state, Nigeria (11.76° N, 9.34° E, altitude of 460 m).
[15,16]
The climate is a tropical savanna with distinct
wet and dry seasons (Fig. 1). Written permission for sample collection and experimental use of the effluent was
obtained from the mill prior to sampling.
Fig 1. Map of the study area
Wastewater samples were collected as grab samples on three separate production days (15 February, 22
February and 1 March 2025) during peak discharge periods (10:00–14:00).
[17]
Each day, three 5 L high-density
polyethylene bottles (precleaned and rinsed three times with the effluent) were filled, yielding a total of nine
grab samples and approximately 45 L of composite effluent (Fig. 2). Equal volumes from each sampling event
were combined to form a single composite sample that was transported on ice to the Environmental Science
Laboratory, Federal University Dutse, and stored at 4 °C pending analysis.
[7]
Temporal variability was assessed
by comparing individual grab samples; the relative standard deviation for key heavy metal concentrations
across the three events was <12%.
Fig 2. Wastewater effluent discharge point
2.2. Physicochemical characterization of raw wastewater
The untreated maize mill wastewater was characterized as milky white with a putrid odor and a temperature
of 32.3 ± 0.4 °C (above the WHO/NESREA recommended range of 25–30 °C). The pH was 6.74 ± 0.08, the
electrical conductivity was 80.7 ± 2.1 µS/cm, and the total dissolved solids concentration was 250 ± 8 mg/L.
The heavy metal concentrations exceeded the discharge limits: 0.096 ± 0.007 mg/L for chromium, 0.022 ± 0.003
mg/L for lead and 0.023 ± 0.002 mg/L for cadmium.
2.3. Biochar production and characterization
Coconut shells were collected from Dutse Ultra-Modern Market, washed, sun-dried for 72 h, pulverized, sieved
and pyrolyzed at 500 °C for 1 h under limited oxygen.
[18]
The resulting biochar was steam-activated at 700 °C
for 30 min.
[19]
The physicochemical properties were determined by standard methods.
[20]
The BET surface area
and pore size distribution were measured by N
2
adsorption–desorption at 77 K (Micromeritics ASAP 2020).
The surface morphology was examined by scanning electron microscopy (SEM; JEOL JSM-7600F). The
functional groups were identified by Fourier transform infrared spectroscopy (FTIR; PerkinElmer Spectrum
Two, 4000–400 cm
-1
). The elemental composition (C, H, N, and S) was determined by a CHNS analyzer. The
porosity of the steam-activated CSDB was 68.7 ± 1.2%, the pH was 9.2 ± 0.1, the moisture content was 3.6 ±
0.3%, the ash content was 12.0 ± 0.8%, the total carbon content was 83 ± 1.5%, the BET surface area was 512
± 18 m
2
/g, and the average pore diameter was 2.8 nm. SEM images revealed a highly porous, irregular surface
with abundant macropores and mesopores. The FTIR spectra showed characteristic bands at 3430 cm
-1
(–OH
stretching), 2920 cm
-1
(aliphatic C–H), 1615 cm
-1
(C=C aromatic/C=O), and 1050–1100 cm
-1
(C–O stretching),
confirming the presence of oxygen-containing functional groups favorable for metal binding.
[21–23]
2.4. Batch adsorption experiments
Batch adsorption of Cr, Cd and Pb was performed in 250 mL Erlenmeyer flasks. A fixed mass of CSDB (5, 10, and
15 g, corresponding to 100, 200, and 300 g/L, respectively) was mixed with 50 mL of wastewater and agitated
at 150 rpm and 25 ± 2 °C. Samples were withdrawn at predetermined intervals (0–120 min), filtered (Whatman
No. 4) and analyzed by atomic absorption spectroscopy (AAS). The removal efficiency (%) was calculated as
follows:
(1)
where C
0
is the initial concentration and C
e
is the concentration at equilibrium (mg/L). The pH was adjusted to
3, 5, 7, 9 or 11 with 0.1 M HCl or NaOH. All experiments were performed in triplicate; the results are reported
as the mean ± standard deviation. Statistical significance was assessed by one-way ANOVA (α = 0.05) followed
by Tukey’s post hoc test where applicable. Batch screening employed relatively high solid‒liquid ratios (100–
300 g/L) to ensure rapid equilibrium and clear kinetic profiles. In the subsequent hybrid continuous-flow
system and multi-objective optimization, a lower and more practical dosage range of 12–15 g/L was selected,
reflecting realistic continuous-operation conditions while still achieving high removal.
2.5. Adsorption kinetics
The contact-time data were fitted to both linear pseudo-first-order (Lagergren) and pseudo-second-order
kinetic models:
(2)
(3)
where ‘qt’ and ‘qe’ are the amounts adsorbed at time ‘t’ and equilibrium (mg/g), respectively, and k
2
is the rate
constant (g/mg/min). Goodness-of-fit was evaluated by R
2
, residual analysis and comparison of calculated
versus experimental qₑ values.
[24,25]
The intraparticle diffusion model (Weber–Morris) was also used to assess
the possible contribution of diffusion steps.
2.6. Hybrid biochar–membrane system
A laboratory-scale hybrid system consisting of a biochar adsorption stage followed by a flat-sheet
microfiltration membrane operated in cross-flow mode was employed. Transmembrane pressure (TMP) was
controlled by a precision pump and continuously recorded. Flux decline and fouling resistance were quantified
using a resistance-in-series model. Parallel membrane-only runs served as the baseline for the fouling-
mitigation assessment.
2.7. Experimental design and machine learning modeling
A two-stage design of experiments (DOE) was implemented. Stage 1 consisted of a 12-run Plackett–Burman
screening design to identify the most significant factors among biochar dosage, contact time, pH, TMP and
temperature. Stage 2 employed a 15-run Box–Behnken response-surface design on the three most influential
factors (dosage, contact time, and TMP), generating a structured dataset of 27 unique experimental
combinations (plus center-point replicates) for six response variables: Cr, Cd and Pb removal efficiency (%),
COD removal (%), permeate flux (L m
-2
/h) and a composite fouling index. All runs were performed in triplicate,
yielding a total of 81 experimental observations that were randomly partitioned into training (70%), validation
(15%) and independent test (15%) sets while preserving the response distribution.
Four supervised ML algorithms were trained and compared:
a) Artificial neural network (ANN): multilayer perceptron with two hidden layers (12 and 8 neurons), ReLU
activation, Adam optimizer (learning rate 0.001), batch size 16, early stopping (patience = 20), and L2
regularization 0.001.
b) Random forest (RF): 200 trees, maximum depth 10, minimum number of samples per leaf 3, and bootstrap
sampling.
c) Extreme gradient boosting (XGBoost): 300 estimators, learning rate 0.05, maximum depth 6, subsample
0.8, colsample_bytree 0.8, regularization λ = 1.0.
d) Support-vector regression (SVR): radial-basis-function kernel, C = 10, ε = 0.1, γ = ‘scale’.
The hyperparameters were tuned by a fivefold cross-validated grid search. Model performance was quantified
by R
2
, root-mean-square error (RMSE), mean absolute error (MAE) and mean absolute percentage error
(MAPE) on the training, validation and independent test sets. Feature importance was assessed by permutation
importance.
2.8. Multi-objective optimization and validation
The best-performing models (XGBoost and ANN) were embedded in three multi-objective optimizers: a genetic
algorithm (GA; population size 50, 100 generations; crossover probability 0.8; mutation rate 0.05), particle
swarm optimization (PSO; 40 particles; inertia weight 0.7; cognitive and social coefficients 1.5), and Bayesian
optimization (Gaussian-process surrogate; expected improvement acquisition, 50 iterations). The objective
function simultaneously maximized the metal removal efficiency and minimized the fouling index, subject to
the experimentally investigated bounds (doses of 5–20 g/L, contact times of 30–120 min, and TMPs of 0.5–1.5
bar). A weighted-sum desirability function (equal weights) is used to rank the Pareto-optimal solutions. The
top-ranked condition set was validated experimentally in triplicate; relative prediction errors and paired t tests
(α = 0.05) were used to confirm statistical equivalence.
2.9. Techno-economic assessment and statistical analysis
A preliminary techno-economic analysis was performed for a modular hybrid unit treated at 10 m
3
/day¹.
Capital costs included the biochar reactor, membrane module (assumed lifetime of 18–24 months under hybrid
operation versus 12–15 months for membrane-only), pumps and piping. The operating costs included biochar
production and replacement (feedstock cost negligible), energy for pumping and agitation (0.85–1.1 kWh m
-3
),
membrane chemical cleaning (every 7–10 days), and labor and maintenance (5% capital per year). The unit
treatment cost (₦/m
-2
) was calculated under base-case assumptions, and the sensitivity to membrane lifetime
and energy price was examined. All the statistical analyses were conducted at α = 0.05.
3. Results and discussion
3.1. Wastewater quality and treatment needs
Table 1 shows the essential physicochemical characteristics of the untreated maize flour mill effluent. The
temperature (32.3 ± 0.4 °C) exceeded the recommended limits, whereas the pH, EC and TDS remained within
acceptable ranges. The concentrations of heavy metals, particularly Cr and Pb, exceeded both the WHO
discharge limits and the NESREA discharge limits, confirming the necessity of targeted treatment before
discharge or irrigation use. These findings are consistent with those for other agro-industrial sites in the region.
[6,26]
Table 1. Physicochemical properties of maize flour mill wastewater compared with regulatory limits (mean ± SD, n
= 9 grab samples)
Parameter
Value (mean ± SD)
WHO Limit
NESREA Limit
pH
6.74 ± 0.08
6.5–8.5
6.5–8.5
EC (µS/cm)
80.7 ± 2.1
<250
—
Temperature (°C)
32.3 ± 0.4
25–30
25–30
TDS (mg/L)
250 ± 8
<500
<2000
Chromium (mg/L)
0.096 ± 0.007
0.05
0.05
Lead (mg/L)
0.022 ± 0.003
0.01
0.01
Cadmium (mg/L)
0.023 ± 0.002
0.03
0.01
3.2. Biochar physicochemical properties
As presented in Table 2, the surface area (512 ± 18 m
2
/g), alkaline pH (9.2 ± 0.1) and abundance of oxygen-
containing functional groups (FTIR) of the steam-activated CSDB were high, all of which favor cation adsorption.
SEM confirmed a highly porous morphology, with a measured porosity of 68.7 ± 1.2%. These characteristics
align with the literature values reported for similarly prepared coconut-shell biochars.
[21–23]
Table 2. Physicochemical characteristics of the steam-activated CSDB (mean ± SD, n = 3)
Parameter
Value
pH
9.2 ± 0.1
Moisture content (%)
3.6 ± 0.3
Ash content (%)
12.0 ± 0.8
Porosity (%)
68.7 ± 1.2
BET surface area (m
2
/g)
512 ± 18
Average pore diameter (nm)
2.8 ± 0.2
EC (µS/cm)
950 ± 25
Total carbon (%)
83 ± 1.5
3.3. Influence of contact time, pH and dosage
The removal efficiencies of all three metals increased rapidly within the first 30–60 min and approached
equilibrium by 90–120 min (ANOVA p < 0.001). At 120 min, the observed removal rates (Fig. 3) were 67.71 ±
2.4% (Cr), 91.30 ± 1.8% (Cd) and 90.91 ± 2.1% (Pb). The pH dependence (Fig. 4) was metal specific: maximum
Cr removal occurred under acidic conditions (pH 3.5), whereas Cd and Pb removal peaked near neutral to
slightly alkaline pH (6.5–7.5). Higher biochar dosages increased removal (p < 0.01), although incremental gains
diminished at the highest solid loadings.
Fig 3. Effect of contact time on the efficiency of heavy metal removal by CSDB (batch conditions: 10 g adsorbent per
50 mL wastewater, room temperature).
Fig 4. Influence of pH on the efficiency of heavy metal removal by CSDB.
3.4. Adsorption kinetics and capacity
Linear regression of the kinetic data revealed that the fit of the pseudo-second-order model was markedly
better (R
2
> 0.98 for Cr, Cd and Pb) than that of the pseudo-first-order model (R
2
= 0.72–0.85). The calculated
q
e
values from the second-order model agreed closely with the experimental values, whereas the first-order
estimates systematically underpredicted capacity. The Weber–Morris intraparticle-diffusion plots were
multilinear, indicating that film diffusion and intraparticle diffusion both contribute, but the rate-limiting step
is consistent with chemisorption involving valence forces through the sharing or exchange of electrons.
[24,27]
The
equilibrium adsorption capacities were 0.75 ± 0.03 mg/g (Cr), 0.027 ± 0.002 mg/g (Cd) and 0.048 ± 0.003 mg/g
(Pb).
3.5. Hybrid system performance and fouling mitigation
Flux decline was substantially slower under hybrid operation than under membrane-only conditions. The
fouling resistance calculated from the resistance-in-series model decreased by 35–45% at comparable TMPs
and cross-flow velocities (Fig. 5). Hydraulic cleaning recovered 75–85% of the initial pure-water flux for the
hybrid system versus 55–65% for the membrane-only system, indicating a greater proportion of reversible
fouling when biochar pretreatment was employed.
Fig 5. Normalized flux decline over time for membrane-only versus hybrid biochar–membrane operations.
3.6. Machine learning model performance
Table 3 presents the complete performance metrics of the four algorithms on the independent test set. XGBoost
and ANN consistently outperformed RF and SVR across all six responses. With respect to the metal removal
efficiency, the test-set R
2
values ranged from 0.91 to 0.96, with an RMSE ranging from 1.8–3.4% and an MAE
ranging from 1.4–2.7%. Flux and fouling index predictions yielded R
2
values of 0.88–0.94. Biochar dosage,
contact time and TMP emerged as the three most influential features according to permutation importance. The
superior performance of the gradient-boosting and neural-network models is consistent with their capacity to
capture the nonlinear interactions typical of adsorption‒filtration systems.
[13,14]
Table 3. Comparative performance metrics of the four machine learning models on the independent test set
Model
Response
R
2
RMSE (%)
MAE (%)
MAPE (%)
XGBoost
Metal removal (avg.)
0.94–0.96
1.8–2.6
1.4–2.1
2.1–3.4
ANN
Metal removal (avg.)
0.91–0.95
2.1–3.1
1.6–2.4
2.4–3.8
RF
Metal removal (avg.)
0.86–0.90
3.2–4.5
2.5–3.6
3.5–5.2
SVR
Metal removal (avg.)
0.82–0.88
3.8–5.1
2.9–4.2
4.1–6.0
XGBoost
Fouling index
0.92
2.4
1.9
3.1
ANN
Fouling index
0.90
2.8
2.2
3.6
3.7. Optimized operating conditions and experimental validation
Multi-objective optimization (GA, PSO and Bayesian) converged on a consistent operating window: CSDB
dosage 12–15 g/L, contact time 90–105 min and TMP 1.0–1.1 bar (Fig. 6). Under these conditions, the models
predicted simultaneous heavy-metal removals of 85–92% and a reduction in the fouling index >40% relative to
the membrane-only baseline. Independent laboratory validation runs (n = 3) confirmed the predictions, with
relative errors ≤6% and no statistically significant differences (paired t test, p > 0.05).
Fig 6. Validated optimal operating conditions and associated key performance outcomes of the hybrid system
3.8. Techno-economic considerations
The base-case cost of a 10 m
3
/day modular unit yielded a unit treatment cost of approximately ₦450/US$0.27
m
-3
. The greatest cost components were membrane replacement (assuming an extended lifetime of 18–24
months under hybrid operation) and energy for pumping (0.85–1.1 kWh m
-3
). Biochar production costs
remained low because of the negligible feedstock value and simple activation protocol. Sensitivity analysis
indicated that a 15–20% extension of membrane life attributable to biochar pretreatment can reduce unit cost
by 8–12%. These figures are competitive with those of conventional coagulation–sedimentation–filtration
trains reported for similar agro-industrial wastewaters.
3.9. Comparative performance, limitations and future directions
Cd and Pb removal efficiencies (>90%) are comparable to those reported for other coconut-shell and
agricultural-residue biochars that treat similar matrices.
[10,28–30]
Chromium removal (≈68%) was lower, which is
consistent with the well-documented preference of many carbonaceous adsorbents for divalent cations in the
pH range examined. Coupling the adsorption stage with membrane filtration resulted in a measurable reduction
in the fouling rate, longer filtration cycles and reduced chemical-cleaning demand. The present laboratory-scale
study has several limitations. Sampling was confined to three production days within a single season; therefore,
longer-term variability of the influent matrix was not fully captured. Membrane integrity was evaluated only
over short experimental campaigns; multiple months of fouling–cleaning cycles and long-term integrity testing
remain to be performed. Spent-biochar regeneration and reuse, as well as a comprehensive life-cycle
assessment tailored to Nigerian small- and medium-scale enterprises, constitute important directions for future
work. Consequently, claims regarding immediate pilot- or full-scale transferability are moderated: the validated
laboratory operating window provides a robust starting point, but pilot-scale confirmation under realistic
variability is still needed.
4. Conclusions
In conclusion, this study demonstrates that a hybrid system comprising steam-activated coconut-shell biochar
and microfiltration, guided by machine learning prediction and multi-objective optimization, constitutes an
effective and economically viable approach for treating real maize flour mill wastewater. The principal scientific
contributions are quantitative evidence that low-cost, locally produced CSDB can achieve high Cd and Pb
removal via chemisorption while substantially mitigating membrane fouling; transparent development and
benchmarking of four ML algorithms, with XGBoost and ANN providing robust simultaneous prediction of
pollutant removal and fouling indices; and experimental validation of an optimized operating window that
balances removal performance with operational practicality. This research thereby advances the integration of
waste-derived adsorbents, membrane technology and data-driven process control for sustainable agro-
industrial wastewater management in resource-constrained settings. Further pilot-scale trials and long-term
membrane testing are recommended before full-scale implementation.
Acknowledgments
The author gratefully acknowledges the management of the maize flour mill at Godiya Miyyeti Estate, Dutse, for
access to the effluent stream and logistical support during sampling. Laboratory facilities at the Department of
Environmental Sciences, Federal University Dutse, are also acknowledged.
CRediT Author Contribution Statement
Afeez Oladeji Amoo: Conceptualization, Methodology, Formal analysis, Investigation, Validation, Supervision,
Project administration, Writing – original draft, Writing – review & editing, Visualization. The author has read
and approved the final version of the manuscript for publication and agrees to be accountable for all aspects of
the work, ensuring that questions related to the accuracy or integrity of any part of the work are appropriately
investigated and resolved.
Funding Declaration
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-
profit sectors.
Conflict of Interest
There is no conflict of interest.
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding
author on reasonable request.
Artificial Intelligence (AI) Use Disclosure
The authors declare that artificial intelligence (AI)-assisted tools were used only for language refinement,
grammar improvement, and manuscript structuring purposes during the preparation of this work. All
technical content, experimental implementation, results, and interpretations were independently developed
and verified by the authors.
Supporting Information
Not applicable.
References
[1]
F. Hua, Research progress of biochar materials in heavy metal pollution treatment application In E3S
Web of Conferences, EDP Sciences, 2024, 580, 02001, doi: 10.1051/e3sconf/202458002001.
[2]
M. T. Onigemo, A. Balogun, L. Odeh, E. Udoh, A. Mohammed, A. Omolusi, N. Ankomah, S. Akyea-Mensah,
S. Onele, M. Akoh, Production and characterization of activated carbon from coconut shell for
adsorption of lead (II) ion from wastewater, Advances in Journal of Chemistry - Section B, 2024, 6, 269-
283, doi: 10.48309/AJCB.2024.472061.1240.
[3]
P. Subramanian, K. Pandian, S. Pakkiyam, K. V. Dhanuskodi, S. Annamalai, P. P. Chidambaram, M. R. A.
F. Mustafa, Biochar for heavy metal cleanup in soil and water: A review, Biomass Conversion and
Biorefinery, 2025, 15, 11421-11441, doi: 10.1007/s13399-024-05989-1.
[4]
F. Bayar, N. Ali, Y. Dong, U. Ahmad, M. M. Anjum, G. R. Khan, M. Zaib, A. Jalal, R. Ali, L. Ali, Biochar-based
adsorption for heavy metal removal in water: A sustainable and cost-effective approach,
Environmental Geochemistry and Health, 2024, 46, 428, doi: 10.1007/s10653-024-02214-w.
[5]
R. O. Darko, P. P. Nsonowah, J. D. Owusu-Sekyere, M. Nyameche, Assessment of wastewater quality for
irrigation use in the Central Region of Ghana, Discover Environment, 2025, 3, 146, doi:
10.1007/s44274-025-00271-1.
[6]
A. B. Hussaini, J. L. Cao, M. Yateh, B. Wu, Effects of industrial wastewater effluents on irrigation water
quality around Bompai Industrial Area, Kano State, Nigeria, Open Access Library Journal, 2023, 10, 1–
14, doi: 10.4236/oalib.1110395.
[7]
American Public Health Association (APHA), Standard Methods for the Examination of Water and
Wastewater, 23rd ed., American Public Health Association, American Water Works Association, Water
Environment Federation, Washington, D.C., 2017.
[8]
M. Y. D. Alazaiza, T. M. Alzghoul, D. E. Nassani, M. J. K. Bashir, Natural coagulants for sustainable
wastewater treatment: Current global research trends, Processes, 2025, 13, 1754, doi:
10.3390/pr13061754.
[9]
D. Diver, I. Nhapi, W. R. Ruziwa, The potential and constraints of replacing conventional chemical
coagulants with natural plant extracts in water and wastewater treatment, Environmental Advances,
2023, 13, 100421, doi: 10.1016/j.envadv.2023.100421.
[10]
J. O. Ighalo, J. Conradie, C. R. Ohoro, J. F. Amaku, K. O. Oyedotun, N. W. Maxakato, K. G. Akpomie, E. S.
Okeke, C. Olisah, A. Malloum, K. A. Adegoke, Biochar from coconut residues: An overview of production,
properties and applications, Industrial Crops and Products, 2023, 204, 117300, doi:
10.1016/j.indcrop.2023.117300.
[11]
L. Fang, T. Huang, H. Lu, X. L. Wu, Z. Chen, H. Yang, S. Wang, Z. Tang, Z. Li, B. Hu, X. Wang, Biochar-based
materials in environmental pollutant elimination, H₂ production and CO₂ capture applications,
Biochar, 2023, 5, 42, doi: 10.1007/s42773-023-00237-7.
[12]
Y. Wang, L. Chen, Y. Zhu, W. Fang, Y. Tan, Z. He, H. Liao, Research status, trends and mechanisms of
biochar adsorption for wastewater treatment: A scientometric review, Environmental Sciences Europe,
2024, 36, 25, doi: 10.1186/s12302-024-00859-z.
[13]
O. Zahoor, S. J. Khan, M. Usama, H. J. Tanudjaja, N. Ghaffour, M. S. Nawaz, Harnessing machine learning
for transmembrane pressure prediction in MBR systems during textile wastewater treatment,
Desalination and Water Treatment, 2025, 322, 101238, doi: 10.1016/j.dwt.2025.101238.
[14]
J. Liang, S. Lee, X. Ren, Y. Guo, J. Park, S. G. Park, J. Y. Kim, M. H. Hwang, Predictive Framework for
Membrane Fouling in Full-Scale Membrane Bioreactors (MBRs): Integrating AI-Driven Feature
Engineering and Explainable AI (XAI), Processes, 2025, 13, 2352 doi: 10.3390/pr13082352.
[15]
A. O. Amoo, S. B. Adamu, A. Musa, A. O. Adeleye, C. I. Asaju, E. M. Ijanu, B. G. Bate, N. B. Amoo, The impact
of biochar derived from corncobs in ameliorating soil quality of rice farm in Dutse, Nigeria, Journal of
Scientific Insight, 2024, 1, 189-197, doi: 10.69930/jsi.v1i4.234.
[16]
A. O. Adeleye, B. Yalwaji, G. P. Shiaka, A. O. Amoo, C. C. Udochukwu, Bacteriological quality of fresh-cut
ready-to-eat vegetables sold in Dutse Market, North-West Nigeria, FUW Trends in Science and
Technology Journal, 2019, 4, 54-57.
[17]
S. N. Khan, M. Nafees, M, Imtiaz, Assessment of industrial effluents for heavy metals concentration and
evaluation of grass (Phalaris minor) as a pollution indicator, Heliyon, 2023, 9, e20299, doi:
10.1016/j.heliyon.2023.e20299.
[18]
M. H. Ullah, M. J. Rahman, Adsorptive removal of toxic heavy metals from wastewater using water
hyacinth and its biochar: A review, Heliyon, 2024, 10, e36869, doi: 10.1016/j.heliyon.2024.e36869.
[19]
V. H. Nguyen, D. T. Nguyen, T. T. Nguyen, H. P. T. Nguyen, H. B. Khuat, T. H. Nguyen, V. K. Tran, S. W.
Chang, P. Nguyen-Tri, D. D. Nguyen, Activated carbon with ultra-high surface area derived from
sawdust biowaste for the removal of rhodamine B in water, Environmental Technology & Innovation,
2021, 24, 101811, doi: 10.1016/j.eti.2021.101811.
[20]
E. N. Yargicoglu, B. Y. Sadasivam, K. R. Reddy, K. A. Spokas, Physical and chemical characterization of
waste wood derived biochars, Waste Management, 2015, 36, 256–268, doi:
10.1016/j.wasman.2014.10.029.
[21]
S. Ho, H. Ajab, T. V. Nagalakshmi, Recent advances in the development of coconut based activated
carbon, International Journal of Chemical and Biochemical Sciences, 2023, 24, 737–749.
[22]
S. Mukherjee, B. Sarkar, V. K. Aralappanavar, R. Mukhopadhyay, B. B. Basak, P. Srivastava, O. Marchut-
Mikołajczyk, A. Bhatnagar, K. T. Semple, N. Bolan, Biochar microorganism interactions for organic
pollutant remediation: Challenges and perspectives, Environmental Pollution, 2022, 308, 119609, doi:
10.1016/j.envpol.2022.119609.
[23]
A. I. Osman, S. Fawzy, M. Farghali, M. El-Azazy, A. M. Elgarahy, R. A. Fahim, M. A. Maksoud, A. A. Ajlan,
M. Yousry, Y. Saleem, D. W. Rooney, Biochar for agronomy, animal farming, anaerobic digestion,
composting, water treatment, soil remediation, construction, energy storage, and carbon
sequestration: a review, Environmental Chemistry Letters, 2022, 20, 2385–2485, doi: 10.1007/s10311-
022-01424-x.
[24]
J. C. Bullen, S. Saleesongsom, K. Gallagher, D. J. Weiss, A revised pseudo-second order kinetic model for
adsorption, sensitive to changes in adsorbate and adsorbent concentrations, Langmuir, 2021, 4, 3189-
201, doi: 10.1021/acs.langmuir.1c00142.
[25]
V. Karthik, S. Periyasamy, S, Dharneesh, G. K. Duvakeesh, D. G. Gizaw, V. Vijayashankar, Biochar as a
sustainable adsorbent for heavy metal removal from polluted waters: A comprehensive outlook,
Journal of Chemistry, 2024, 2024, 8217730, doi: 10.1155/joch/8217730.
[26]
N. Lawal, K. Ogedengbe, B. Adetifa, G. Anyanwu, Degrading cassava mill effluent using aerated
sequencing batch reactor with palm kernel shell as medium, Journal of Degraded and Mining Lands
Management, 2019, 6, 1737–1745, doi: 10.15243/jdmlm.2019.063.1737.
[27]
Z. Raji, A. Karim, A. Karam, S. Khalloufi, Adsorption of heavy metals: Mechanisms, kinetics and
applications of various adsorbents in wastewater remediation – A review, Waste, 2023, 1, 775–805,
doi: 10.3390/waste1030046.
[28]
H. Zhang, F. Xu, J. Xue, S. Chen, J. Wang, Y. Yang, Enhanced removal of heavy metal ions from aqueous
solution using manganese dioxide-loaded biochar, Scientific Reports, 2020, 10, 6067, doi:
10.1038/s41598-020-63000-z.
[29]
M. Haris, Z. Amjad, M. Usman, A. Saleem, A. Dyussenova, Z. Mahmood, K. Dina, J. Guo, W. Wang, A review
of crop residue-based biochar as an efficient adsorbent to remove trace elements from aquatic
systems, Biochar, 2024, 6, 47, doi: 10.1007/s42773-024-00341-2.
[30]
N. A. A. Qasem, R. H. Mohammed, D. U. Lawal, Removal of heavy metal ions from wastewater: a
comprehensive and critical review, npj Clean Water, 2021, 4, 36, doi: 10.1038/s41545-021-00127-0.
Publisher Note: The views, statements, and data in all publications solely belong to the authors and
contributors. GR Scholastic is not responsible for any injury resulting from the ideas, methods, or products
mentioned. GR Scholastic remains neutral regarding jurisdictional claims in published maps and institutional
affiliations.
Open Access
This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which
permits the non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format,
as long as appropriate credit to the original author(s) and the source is given by providing a link to the Creative
Commons License and changes need to be indicated if there are any. The images or other third-party material
in this article are included in the article's Creative Commons License, unless indicated otherwise in a credit line
to the material. If material is not included in the article's Creative Commons License and your intended use is
not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this License, visit: https://creativecommons.org/licenses/by-
nc/4.0/
© The Author(s) 2026
