Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
Research Article, Issue 1
Analytical Methods in Environmental Chemis try Journal
Journal home page: www.amecj.com/ir
AMECJ
Articial neural network and response surface design for
modeling the competitive biosorption of pentachlorophenol
and 2,4,6-trichlorophenol to Canna indica L and analyzed by
UV-Vis spectrometry in Aquaponia
Chri s tian Ebere Enyoh a,c,*, Prosper Eguono Ovuorayeb, Beniah Obinna Isiukuc,
and Chinenye Adaobi Igwegbed
aGraduate School of Science and Engineering, Saitama University, Saitama, Japan
bDepartment of Chemical Engineering, Federal University of Petroleum Resource, P.M.B 1221 Eurun, Nigeria
cDepartment of Chemi s try, Faculty of Physical Sciences, Imo State University, Imo State, Nigeria
dDepartment of Chemical Engineering Nnamdi Azikiwe University, P. M. B. 5025, Awka, Nigeria
ABSTRACT
The continuous exposure of the environment to carcinogenic wa s tes
and toxic chlorophenols such as pentachlorophenol (PCP) and
2,4,6-trichlorophenol (TCP) resulting from indu s trial production
activities has become a great concern to research scienti s ts and
environmental policymakers. The search for a co s t-ecient and
eco-friendly approach to the phytoremediation of water will
guarantee su s tainability. The present research concerns the co s t-
benet evaluation and the optimization modeling of the competitive
biosorption of PCP and TCP from aqueous solution to Cana indica.
L (CiL-plant) using response surface methodology (RSM) , articial
neural network (ANN) model, and UV-Vis Spectrometry. The
predictive performances of the ANN model and the RSM were
compared based on their s tati s tical metrics. The antagoni s tic and
synergetic eects of signicant biosorption variables (pH, initial
concentration, and exposure time) on biosorption were s tudied at
p-values ≤0.005. The ndings from the phytoremediation process
conrmed that PCP and TCP removal rate reached equilibrium at the
optimum conditions corresponding to predominantly acidic pH (4),
required initial concentration of 50 mg L-1, and exposure time of 25
days in aquaponia. The optimized output transcends to PCP and TCP
removal rates of 90% and 87.99% eciencies at predicted r-squared
≤0.9999 and a 95% condence interval. The co s t-benet evaluation
e s tablished that at the optimum conditions, the co s t of operating
the removal of TCP from the aqueous solution would save $ 7.72
compared to PCP. The optimization model’s reliability based on the
experiment’s (DoE) design was more su s tainable than the one-factor-
at-a-time (OFAT) methodologies reported in previous research.
Keywords:
Phytoremediation,
UV-Vis spectrometry,
Chlorophenol biosorption,
Canna Indica L plant,
ANN modeling,
RSM optimization
ARTICLE INFO:
Received 11 Nov 2022
Revised form 21 Jan 2023
Accepted 16 Feb 2023
Available online 30 Mar 2023
*Corresponding Author: Chris tian Ebere Enyoh
Email: cenyoh@gmail.com
https://doi.org/10.24200/amecj.v6.i01.228
------------------------
1. Introduction
Chlorinated phenols, such as Pentachlorophenol
(PCP) and 2,4,6-Trichlorophenol (TCP), have
been used since the 1930s in a variety of indu s tries
including wood preservation, pe s t control, and
herbicide production [1,2]. As a result, wa s tewater
from these indu s tries can contain high levels of
80 Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
these chemicals, leading to the pollution of water
resources and potential harm to ecosy s tems [3].
The use of pe s ticides on farms is a signicant
contributor to the contamination of water
resources with chlorinated compounds. s tudies
have shown the presence of signicant amounts of
organochlorines in water and sh samples [4,5,6].
TCP and PCP are classied as Group B2 probable
human carcinogens and 1B highly hazardous; due
to their high toxicity, carcinogenic potential, and
environmental persi s tence [1,2]. These chemicals
can cause serious health issues such as respiratory
problems, cardiovascular disease, ga s trointe s tinal
issues, and cancer in humans and have been linked
to an increased risk of lymphomas, leukemia, and
liver cancer in animal s tudies [1,2]. Removing these
chemicals from water resources or wa s tewater
before they are released into the environment is
essential to prevent potential harm to humans
and ecosy s tems. Various methods are available
for removing PCP and TCP from water, including
biological and physicochemical approaches such
as photochemi s try, air s tripping, incineration,
and adsorption technologies using activated clay
and plant-based carbons [7]. While some of these
methods are eective, they can be challenging to
implement. Using aquatic plants for wa s tewater
treatment is a newer method for removing pollutants,
and s tudies have shown that it can be eective when
using con s tructed wetlands, pilot-scale sy s tems,
or hydroponic setups [7]. The eciency of plant-
based treatment sy s tems can vary depending on
the specic plant species and their productivity. In
Nigeria and other tropical countries, various plants
eectively remove pollutants from water. Canna
lilies, a type of owering plant, are a commonly
used species for this purpose in Nigeria due to
their wide di s tribution and dominance in aquatic
environments. Additionally, canna lilies can
eectively remove inorganic and organic pollutants
such as PCP and TCP through phytoremediation
[8-11] and can survive in polluted areas [12].
Central composite design and response surface
methodology are s tati s tical approaches used in
pollutant removal s tudies to design experiments
and optimize treatment conditions [13]. They
allow selecting the mo s t eective experimental
conditions through s tati s tical software, reducing
the number of co s tly experiments and trials needed
[14]. These methods have been applied to various
processes, including coagulation to remove dyes
from wa s tewater [15,16]. They can be helpful in
predicting the behavior and outcomes of treatment
sy s tems and analyzing exi s ting processes. Articial
neural networks (ANN), which utilize learning
algorithms to evaluate the relationships between
input and output variables, can also be used to model
and predict the behavior of water management
processes [17, 18]. While articial neural networks
require many data points to be eective, they are fa s t,
adaptable, and can produce real-time predictions
[18]. Both response surface methodology and
articial neural networks have been compared in
their predictive capabilities for various processes
[17]. This s tudy examined the eectiveness of
using C. indica L (CiL-plant), an aquatic plant, to
remove PCP and TCP from water using response
surface methodology and articial neural networks.
The use of aquatic plants for wa s tewater treatment,
specically for removing organic pollutants such as
chlorophenols, is a relatively new method known
as aquatic phytoremediation. This s tudy aims to
optimize the ability of C. indica to remove PCP
and TCP from water using a hydroponic sy s tem,
and to the be s t of our knowledge, is the r s t s tudy
of its kind. Furthermore, the removal behavior
of C. indica for PCP and TCP has been predicted
for the r s t time by a highly ecient developed
ANN model. A techno-economic and co s t-benet
evaluation of the phytoremediation process was also
examined beyond removal eciency to ascertain
the suitability of the CiL-plant for Aquaponia.
2. Materials and methods
2.1. Preparation of plant material and pe s ticides
solutions
We discussed how the plant material and pe s ticide
solutions were manufactured in our earlier papers
[8-10]. In a ood basin in Amakohia, Owerri, Imo
s tate, Nigeria, Canna indica L. seeds and soil for
81
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
growing the plant were collected. The plant was
raised in nurseries with natural environmental
conditions. To conduct the s tudy, properly harve s ted
seedlings with an average height of (14±1 cm) were
employed. Without additional renement, the PCP
and TCP (analytical grade, 99.5%) was used after
being acquired from FinLab in Owerri. Di s tilled
water and ethanol were used to make the solutions
for this experiment. An ethanol-water solution (10%
v/v ethanol/di s tilled water) was used to dissolve
1.0 g of PCP/TCP per liter of solution in a 1.0 liter-
volumetric ask while s tirring continuously. The
s tock had a 1000 mg L-1 equivalent. By dilutions
with di s tilled water, working solutions of 50, 100,
150, 200, and 250 mg L-1 were produced from the
s tock solution and labeled accordingly. Working
solutions were made, and absorbance was measured
at 220 nm for PCP and 296 nm for TCP using a
UV spectrophotometer. The calibration curve
(concentration vs. absorbance) was created using
the recorded absorbance, and it was then utilized
to calculate the amounts of PCP and TCP. With
coecients of determination more than 0.9995, the
absorbance for PCP and TCP s tarting concentrations
rose as the initial concentration increased. This
indicates s trong linearity of the regression line with
good correlation, consequently, and satisfaction of
the in s trument calibration.
2.2. Batch s tudies
Uptake of PCP and TCP by C. indica L. in
pe s ticide-contaminated water was s tudied in batch
culture experiment using hydroponic, cylindrical
(pots) containers with dimensions 18 cm in length,
37 cm in diameter (external) and 19 cm depth
[9, 10]. The containers were lled with 500 mL
working solutions. Then the plant was introduced
into the solution and allowed to s tand. This was
done for four other pots, representing dierent time
durations (i.e., 10 days, 15 days, 20 days, and 25
days). In total, 5 pots were prepared and at each
interval of 5 days a plant was removed and the
residue was analyzed by UV-vis spectrophotometer
at 220 nm for PCP and 296 nm for TCP [9, 10]. The
eect of pH on the removal of PCP and TCP by C.
indica was determined in 500 mL of te s t solutions
containing 100 mg L-1 of PCP and TCP at dierent
pH (4- 9). 1 M nitric acid (HNO3) and 1 M sodium
hydroxide (NaOH) were used for pH adju s tments.
The pH of each solution was measured with a digital
pH meter (Model Jenway 3510). The initial and
nal concentrations of PCP and TCP solutions were
determined on a UV–visible spectrophotometer
(Spectrum Lab 23A) at its maximum absorbance
wavelength of 220 nm and 296 nm, respectively.
All set-ups were conducted in triplicate (total pots
were 80 for batch s tudies, including control and
90 for pH eect), each for PCP and TCP, and were
placed randomly with position shifted once a week.
After one week, all set-ups were supplemented
with N.P.K. fertilizers (1%, i.e., 5 ml: 500 ml).
For each treatment method mentioned, there was a
corresponding control group that only consi s ted of
deionized water; no pe s ticide was added, and only
the nutrient needed for plant growth in water was
provided.
2.3. Response surface design of Experiment
The Central Composite Design (CCD) is an
empirical model used for multi-objective
optimization of the adsorption or bio-sorption of
micropla s tics from an aqueous solution [19,20].
The CCD optimization is based on the Response
Surface Methodology (RSM) [15]. It is used to
access and t experimental data into a linear, cubic,
quadratic, cubic, or polynomial model [21]. The
model coecients developed via the RSM can
e s tablish an optimal model equation and describe
the antagoni s tic or synergetic interactions and
relationship of experimental variables and their
signicance level with the response within the
range s tudied [13]. In this s tudy, the CCD matrices
consi s ted of 20 experimental runs. The modeling
of the bio-sorption of PCP and TCP to CiL-plant
(Canna indica L.) in terms of actual values is shown
in Table 1. The nal model equation following
the prediction of the optimum conditions for bio-
sorption (pH, initial concentration, and time)
for the removal of PCP and TCP is described by
Equation 1.
82
(Eq.1)
Where xij experimental variable and β are ranked
model coecients, the summation symbols
signify the interactive eect of the dependent and
independent variables (pH, time, and concentration),
is the model intercept, and Y is the response (PCP
and TCP removal rate). The optimization modeling
of the biosorption of TCP and PCP to the CIL
plant was executed using Design Expert software
v12.0. The experimental variables contact time (A)
(days), initial concentration (B) (mg L-1), and pH
(C) shown in Table 1 were varied to 3-Levels with
5 replications. The toxicity was modeled following
the CCD matrix. The initial concentration of the
biosorbent and the contact time was varied to
5-Levels at an experimentally determined pH of 4.
2.4. Articial Neural Network
Aside from RSM modeling, data modeling via
articial intelligence tools such as the articial
neural network (ANN) was implemented in this
s tudy to create a better under s tanding of the model
validation of the bioremediation process. The
neural network tool in MatLab 2018a was used to
model the CiL-plant biosorption process. As input
data, the experimental data set obtained from the
experimental design supplied by CCD space (Table
2) via the RSM was employed. The network was
trained using the Multi-Layer Perceptron (MLP)
Levenberg-Marquardt (LM) method (trainlm) to t
the inputs and targets. The network was made up
of the input layer (which included the ve process
parameters: time, concentration, and pH), neurons
(the hidden layer), and the output layer (which
contained the PCP or TCP removal eciency,
expressed in %) (Fig. 1). The input data with 20
samples were divided randomly (dividrand) into
a training set (75%-14 points), validation (15%-
3 points) and te s ting sets (15%-3 points). Based
on R2 and mean square error values, the ideal
number of hidden layer neurons was determined
by trial and error. More data for training decreases
processing time and improves the model; te s ting
provides an impartial evaluation of the network’s
performance. The training was s topped when the
network generalization was improved indicated
by the increase in MSE error of the validation
samples. To eliminate network error, the input and
output variables were normalized between 0 and
1 [17].
2.5. Co s t e s timation theory for the biosorption
process
The techno-economic evaluation of the CiL-
plant-driven bioremediation of the comparative
removal of PCP and TCP from an aqueous solution
was determined following the e s tablished co s t-
benet analysis model [15]. The co s t benet and
alternative co s t models were used to describe the
feasibility of CiL-plant biosorption of PCP and
TCP beyond removal eciency following the
model equation described in Equations 2-5. The
total co s t for the biosorption of PCP, and TCP
from 1.0 L of the aqueous solution to CiL-plant at
optimum operating conditions was evaluated using
Table 1. Showing experimental factors in terms of coded values
Factor Name Units Minimum Maximum Coded Low Coded High Mean Std. Dev.
A Time Days 5.00 25.00 -1 ↔ 5.00 +1 ↔ 25.00 14.50 8.87
B Conc mgL-1 50.00 250.00 -1 ↔ 50.00 +1 ↔ 250.00 122.50 67.81
C pH 4.00 9.00 -1 ↔ 4.00 +1 ↔ 9.00 5.85 2.06
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
83
Table 2. Design matrix in terms of actual and predicted values for the RSM and ANN optimization process
Factors
Response 1: Response 1:
%PCP Removal Eciency %TCP Removal Eciency
Std Run
A:
Time
(Days)
B:
Concentration
(mg L-1)
C:
pH Actual
RSM
predicted
values
ANN
predicted
values
Actual
RSM
predicted
values
ANN
predicted
values
11 1 25 250 4 78.31 77.91 78.04 78.74 78.70 78.76
14 2 15 100 6 52.70 54.81 52.64 66.19 64.97 61.62
18 3 5 100 9 7.56 12.20 7.56 4.81 8.34 5.78
8 4 5 100 6 50.37 33.52 50.32 32.33 21.30 31.17
16 5 25 100 4 82.00 77.11 81.96 85.71 81.89 85.75
13 6 15 100 9 36.33 27.88 36.33 52.29 33.60 48.70
1 7 5 50 4 34.64 35.65 34.6 9.04 9.15 9.07
17 8 25 100 9 49.26 48..85 49.19 53.00 53.81 52.85
10 9 15 100 6 52.70 54.81 52.64 66.19 64.97 61.62
12 10 25 100 6 73.00 81.40 76.81 81.05 83.45 80.53
311 15 100 6 52.70 54.81 52.64 52.70 64.97 61.62
15 12 5 250 4 13.24 13.58 12.71 3.85 3.89 3.81
2 13 25 50 4 90.00 87.99 85.37 82.09 81.87 82.18
7 14 5 100 9 7.56 12.20 7.56 4.81 8.31 5.78
9 15 5 100 4 16.86 21.75 27.52 5.00 8.73 2.12
6 16 15 100 6 52.70 54.81 52.64 66.19 64.97 61.62
4 17 5 250 4 13.24 13.58 12.71 3.85 3.89 3.81
19 18 25 100 9 49.26 48..85 49.19 53.00 53.81 52.85
5 19 5 50 4 34.64 35.65 34.60 9.04 9.049.15 9.07
20 20 25 250 4 78.31 77.97 78.04 78.74 78.70 78.76
Fig. 1. ANN network of the PCP and TCP optimization sequence
the expression shown in Equation 3. The energy
consumption (EC) was evaluated using Equation 2
[15, 23]. and given by:
(Eq.2)
Where PC is the power consumption by the device
(kW), f is the load factor. In a full mode, f =1, t is
the time of usage of the device (hour), and C is the
energy e s timated co s t ($) per (KWh) in Nigeria as
of the month of April 9, 2021.
Total co s t is a function of all co s ts, including
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
84
biosorbent production, labour, and energy. Cm is the
co s ts incurred from transportation, and renting [24].
(Eq.3)
(Eq.4)
Where FO is the return on the selected and forgone
option (PCP versus TCP), in this case, it’s the
performance of CiL-plant for the bioremediation of
aqueous medium and CO is the return on chosen option
from PCP versus TCP, and CB is the opportunity co s t
based derived based on environmental impact and
regulatory risk (Eq. 4). In this case, the return on
chosen option which denes the return on inve s tment
as a function of direct and indirect co s t [15]. The
parameter F0 was evaluated following modied
model Equation 5, expressed as:
(Eq.5)
3. Result and Discussion
In our previous s tudies [8-10], the results for the
removal of PCP and TCP have been presented. This
current s tudy is a s tep further in which removal
processes are optimized and predicted using RSM
and ANN, respectively, to determine the optimum
operating variable for modeling the performance
CiL-plant-driven bioremediation process.
Furthermore, the techno-economic and co s t-benet
analysis for the method was evaluated in the current
s tudy to ascertain the feasibility of the CiL-driven
bioremediation of PCP and TCP in aquaponia
beyond removal eciency.
3.1. Central composite design modeling of the
CiL-driven biosorption process
The ndings from the CCD optimization modeling
following the biosorption of PCP and TCP to CIL-
plant from an aqueous solution follow a second-
order quadratic model shown in the ANOVA (Tables
3 and 4). Tables 3 and 4 showed that the selected
quadratic model recorded consi s tent outputs
from the CCD that adequately describes the CiL-
plant-driven biosorption of PCP and TCP from
an aqueous solution. It was observed that model
f-values PCP (30.55) and TCP (62.75) obtained a
lack-of-t value >3 recorded at a p-value less than
0.05. This s tati s tical output indicates that there
is only a 0.01% chance that f-values this large
could occur in the optimization modeling of the
phytoremediation process variables due to noise
[19, 24]. A p-value ≤ 0.0500 obtained with the CCD
space sugge s ts that the quadratic model terms and
subsequent assumptions on the phytoremediation
process are signicant at a 95% condence level.
The s tati s tical output also sugge s ts that the quadratic
model results signicantly describe the CiL-plant-
driven biosorption of PCP and TCP from an aqueous
medium [21]. The model t s tati s tics that describe
the removal of PCP from the aqueous medium
e s tablished that the predicted R² (0.8322), adju s ted
R2 (0.9256), is in reasonable agreement with the
correlation coecient R² (0.9256) recorded from
the central composite design space. Similarly, the
model predicted R² (0.9329) was also in reasonable
agreement with the adju s ted R² (0.9630) reported for
the CiL-plant biosorption of TCP from an aqueous
solution. These r-squared values are close to unity
(1), and their dierences are less than 0.2, indicating
that the selected quadratic model description of
the CiL-plant-driven phytoremediation process is
signicant at a 95% condence level [19, 21, 22].
However, where the adequacy of precision output
>4 is desirable [15], the selected quadratic model
recorded adequacy of precision (16.11) value
measures the signal-to-noise ratio (16.11), and the
model f-value (5.36) can be used to navigate design
space for modeling the PCP removal rate [24].
The adequacy of precision (19.41), and signal ratio
(19.41) recorded for the TCP biosorption modeling
were , conrming that the quadratic model
following the design space adequately describes
the modeling of the biosorption of PCP and TCP
to CiL-plant. Few insignicant model terms are
reported with the central composite design space,
not counting those required to support hierarchy.
Regarding PCP removal rate, the r s t-order
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
85
phytoremediation variables such as; contact time,
pH, and initial concentration, and the second-order
degree of pH (C)² are signicant model terms. This
outcome indicated that contact time (A), initial
concentration of CiL-plant (B), and pH (C) all have
a signicant antagoni s tic inuence on CiL-plant
biosorption of PCP from aqueous solution. The
pH also has a second-order degree of signicant
impact on removing PCP from aqueous solution
compliance to CiL-plant at p-values <0.100
[15]. The s tati s tical outcome sugge s ts model
reduction was negligible, and the interactive eect
of model factors A*B, and A*C which transcends
to contact time*concentration (A*B), and contact
time*pH (A*C) both have synergetic eects on
the biosorption of PCP to Canna indica. L (CiL-
plant) in the combined sy s tem. Comparatively,
the quadratic model s tati s tics and assumptions
describing the CiL-plant-driven TCP removal rate
e s tablished that contact time (A) and pH (C) of
solution signicantly aect on the phytoremediation
process. However, the interactive eect of contact
time*pH of solution (A*C) had a synergetic eect
on the TCP removal from the aqueous solution.
Also, the model assumption e s tablished that a
higher order degree of contact time (A²) and pH (C²)
are signicant model terms for TCP biosorption
to Canna indica L. (CiL-plant). This translates to
both model terms having an antagoni s tic r s t and
second-order impact on TCP biosorption to CiL-
plant. The contact time-initial concentration (A*B)
has a synergetic eect on the biosorption of TCP
and PCP to Canna indica L. Consequently, the
selected model assumption proved that sucient
contact time and optimized initial concentration
of samples are consequential to the overall
performance of CiL-plant in Aquaponia. The results
from model t s tati s tics following the e s tablished
design space also showed that the coecient of
e s timate representing the expected change in TCP
and PCP removal eciency per unit change in the
values of signicant phytoremediation variables
when non-signicant factors are held con s tant [21]
conrmed the range of variance ination factors
(VIFs) values (-2.91≤ VIFs ≤ 8.27) were recorded
for the CiL-plant driven biosorption of PCP from
aqueous solution. This range of VIFs output (1.22
≤ VIFs ≤ 3.27) is consi s tent with the removal rate
of TCP from an aqueous solution. The model-
e s tablished VIFs outputs fell within the range
1≥ VIFs <10, sugge s ting that the intercept in an
orthogonal design [17], while model coecients
are adju s tments around that average based on the
signicant factors describing the ecacy of CiL-
plant. The factors are orthogonal when the VIFs
are equal to a unit (1); there is a situation of multi-
collinearity at VIFs outputs >1 [17]. The moderate
VIFs recorded with the CCD indicate a negligible
level of severity of the correlation of factors [21].
Consequently, the VIFs <10 recorded for PCP and
TCP are tolerable. The summary based on the VIFs
obtained via the e s tablished quadratic model, the
subsequent model hierarchy based on the level
signicance of the experimental factors following
the biosorption of PCP, and TCP to Canna indica L.
(CiL-plant) from an aqueous solution follows the
Table order.
The e s tablished quadratic model equations
describing the biosorption of PCP, and TCP to
Canna indica L were obtained from the CCD
optimization outputs. The outcome showed that the
nal model equation for the PCP removal rate is
given by Equation 6, and the TCP removal rate is
described by Equation 7.
PCP Contact time > Initial Concentration > pH > Time*pH > Time*Initial Concentration
TCP Contact time > Initial Concentration > pH > Time*pH > Time*Initial Concentration
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
86
Table 3. ANOVA for the Quadratic modeling of PCP biosorption to CiL-plant
Source Sum of Squares df Mean Square F-value p-value
Model 12028.39 8 1503.55 30.55 < 0.0001 Signicant
A-Time 6707.47 1 6707.47 136.28 < 0.0001
B-Concentration 421.62 1 421.62 8.57 0.0138
C-pH 497.35 1 497.35 10.11 0.0088
AB 66.62 1 66.62 1.35 0.2693
AC 215.55 1 215.55 4.38 0.0604
A² 14.93 1 14.93 0.3034 0.5928
B² 97.68 1 97.68 1.98 0.1865
C² 450.18 1 450.18 9.15 0.0116
Residual 541.40 11 49.22
Lack of Fit 541.40 3 180.47 Not signicant
Pure Error 0.0000 8 0.0000
Cor Total 12569.79 19
Pred R2 = 0.832; Adj R2 = 0.9256; R2 =0.9565; s tdv = 7.02; Adeq of Precision =16.105 and Mean value =46.27
Table 4. Showing ANOVA for Quadratic modeling of the TCP biosorption to CiL-plant
Source Sum of Squares df Mean Square F-value p-value
Model 18743.53 8 2342.94 62.75 < 0.0001 Signicant
A-Time 10015.46 1 10015.46 268.24 < 0.0001
B-Concentration 29.10 1 29.10 0.7794 0.3962
C-pH 284.22 1 284.22 7.61 0.0186
AB 2.02 1 2.02 0.0540 0.8205
AC 473.96 1 473.96 12.69 0.0045
A² 338.64 1 338.64 9.07 0.0118
B² 1.12 1 1.12 0.0301 0.8654
C² 300.46 1 300.46 8.05 0.0162
Residual 410.71 11 37.34
Lack of Fit 274.23 3 91.41 5.36 0.0257 Not signicant
Pure Error 136.49 8 17.06
Cor Total 19154.25 19
Pred R2 = 0.9329; Adj R2 = 0.9630; R2 =0.9786; s tdv = 7.02; Adeq of Precision =19.41 and Mean value =44.43
3.2 Optimization outputs following the CiL-plant-
driven biosorption process
The interpretation of the CiL-plant-driven biosorption
of PCP, and TCP from an aqueous solution follows
from the e s tablished model Equations 6-7. The results
based on the interpretation of the optimization ramp
(Fig. 2) conrmed that the optimum comparative
conditions describing the be s t performance of the
CiL-plant-driven biosorption of PCP and TCP from
an aqueous solution correspond to pH (4), initial
concentration (50 mg L-1), and contact time (25
days). The predicted optimum based on the quadratic
model output is conrmed from the optimization
ramp shown in Figure 2. The predicted optimum
outputs translate to a removal eciency of 87.88 %,
and 81.87 % for the biosorption of PCP, and TCP to
CiL-plant as indicated in the optimization ramp in
Figure 2. The optimum points transcend to a s tandard
deviation of PCP (7.01), and TCP (6.80) from the
actual observations practicable. The outcome is
represented by the ag points shown on the 3-D
plots in Figure 3 (a-b). The ag points conrmed
that the predicted optimum points are located within
the CCD design space [17, 24], and maintained
within the range of the experimental values under
inve s tigation.
The predicted optimum indicates that the be s t
biosorption of PCP, and TCP occurred via an active
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
87
initial concentration of 50 mg L-1 of Canna indica
L. plant. The phytoremediation progressed in a
predominantly acidic medium (pH 4), and requires
sucient contact time (25 days) to drive the
biosorption of PCP, and TCP to lower residual that
guarantees su s tainability. The optimization results
e s tablished that under similar optimum operating
conditions, the comparative biosorption rate of
PCP to CiL-plant was mo s t favorable compared to
TCP with a signicant dierence of ≥6%.
3.3 Articial neural network performance
validation of the CiL-driven biosorption process
The validation performance in Figures 4 (a-b)
shows how the number of epochs varied with the
MSE for the optimal neural network. The be s t
validation performance was 4.8753 and 2.8482 at
epoch 3 for PCP and TCP biosorption. The scatter
plots depicting the linearity of the output values of
the network with the target values for the training,
te s ting, validation, and overall data (as generated
Fig 2. Optimization ramp for CiL-plant driven biosorption of PCP and TCP
Fig 3: 3D surfaces plot of CCD optimization
of the CiL-plant driven biosorption of: (A) TCP, and (B) PCP
Fig. 3. 3D surfaces for the removal eciency at dierent times and concentrations for (a) TCP and (b) PCP
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
88
by the MLP platform) are illu s trated in Figures 5 (a, b).
The predicted R2 values were used to indicate the
linearity – with the training network having the
highe s t value of 0.9999 and 0.9945 for the optimal
neural network for PCP and TCP biosorption,
respectively. The outputs design matrix in
Table 2 (presented in section 2.4 above)
displayed the anticipated responses at various
experimental setups. It can be concluded from the
outline of Figure 4 that the summary of the s tati s tical
and evaluation metrics from the ANN indicates an
increase in model errors as the number of epochs
increased. This outcome reasonably agrees with the
optimization modeling procedure reported in the
literature [16, 27]. The train and validation curves’
curvature indicates that overtting has been greatly
minimized [27].
The performance of each model (ANN and RSM)
was validated by evaluating their prediction
accuracy using s tati s tical tools (MSE, RSME, X2
and SSE) [27, 28]. The mathematical equations
representing the s tati s tical tools are summarized in
Table 5. A better predictive model has a high R2
value (almo s t 1) and low s tati s tical errors (close
to 0) [11, 12]. The high R2 values and s tati s tical
errors proved a good correlation with the actual
observations practicable than the RSM. When
compared to the RSM, the ANN outputs yielded
signicant s tati s tics performance with minimal
error (R2 ≤0.9945, RMSE≤0.06, and X2 ≤0.0001).
The low s tati s tical error indicates reliable
adequacy of precession [25, 26], sugge s ting
minimal error due to noise [27]. However, the
s tati s tical outcome from both optimization tools
is in reasonable agreement with the actual values
obtained from the experimentation with the RSM
output indicating a ±0.005 deviation from the
ANN output. The RSM performance evaluation
has the benet of providing a prediction equation,
and demon s trating the inuence of operational
parameters and their interactions on the response
[18]. The s tati s tical model assumptions of the RSM
have been ascertained for reliability, and the design
space (CCD) has been te s ted based on the design
of experiments (DoE). As a result, the predicted
optimum reported for the RSM was employed to
further optimize the CiL-plant driven biosorption
process.
Table 5. Summary of the optimization t s tati s tics and errors factor
Error factor Equation RSM ANN
MSE 0.0080 0.0036
RMSE 0.0894 0.0604
X22.8478 0.0001
SSE
Predicted R2
0.1600
0.9329
0.0729
0.9954
Where yi, yi*, and ym s tand for the experimental, predicted, and mean value of the actual responses, N represents the number
of experimental outcomes; MSE: mean squared error, RMSE: root mean square error, X2: Chi-square and SSE: sum of
squares errors
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
89
012345678
8 Epochs
10
0
10
1
10
2
10
3
10
4
Mean Squared Error (mse)
Best Validation Performance is 2.8482 at epoch 3
Train
Validation
Test
Best
0 1 2 3 4 5 6
6 Epochs
10
-25
10
-20
10
-15
10
-10
10
-5
10
0
Mean Squared Error (mse)
Best Validation Performance is 4.8753 at epoch 3
Train
Validation
Test
Best
Fig. 4. Optimal neural network’s validation performance graph
for the (a) PCP, and (b) TCP biosorption processes
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
90
Fig. 5(a). The output/target values for the PCP biosorption processes’ training, te s ting, validation, and overall data
20 40 60 80
Target
10
20
30
40
50
60
70
80
Output ~= 1*Target + -0.11
Training: R=0.99998
Data
Fit
Y = T
40 50 60 70
Target
35
40
45
50
55
60
65
70
75
Output ~= 1*Target + -0.93
Validation: R=0.99622
Data
Fit
Y = T
20 40 60 80
Target
20
30
40
50
60
70
80
90
Output ~= 0.88*Target + 6.8
Test: R=0.98843
Data
Fit
Y = T
20 40 60 80
Target
10
20
30
40
50
60
70
80
90
Output ~= 0.97*Target + 1.9
All: R=0.99433
Data
Fit
Y = T
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
91
Fig. 5(b). The output/target values for the TCP biosorption processes’ training, te s ting, validation, and overall data
20 40 60 80
Target
10
20
30
40
50
60
70
80
Output ~= 0.99*Target + 0.34
Training: R=0.99448
Data
Fit
Y = T
20 40 60 80
Target
10
20
30
40
50
60
70
80
Output ~= 1*Target + -1.5
Validation: R=0.99949
Data
Fit
Y = T
10 20 30 40 50 60
Target
10
20
30
40
50
60
Output ~= 0.93*Target + 1.7
Test: R=0.99826
Data
Fit
Y = T
20 40 60 80
Target
10
20
30
40
50
60
70
80
Output ~= 0.99*Target + -0.054
All: R=0.9958
Data
Fit
Y = T
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
92
3.4 Eects of experimental variables on the
overall performance of the CiL-plant
The eects of the phytoremediation variables
on the CiL-plant driven biosorption of PCP, and
TCP from an aqueous solution were based on the
CCD s tati s tics (VIFs), and model hierarchy model
parameters shown in section 3.1. The relative
impact of signicant model factors pH (C), and
contact time (A) on the biosorption of PCP, and
TCP aqueous solution compliance to CiL-plant
in the single and interactive sy s tem when the
signicant variable is kept con s tant are presented
in Figures 6 and 7.
3.4.1 Antagoni s tic eect of contact time on the
Phytoremediation process
Figure 6 conrmed the antagoni s tic eect of
contact time on the biosorption of PCP, and TCP
to CiL-plant at the optimum concentration (50 mg
L-1). The graph showed that the CiL-plant-driven
biosorption of PCP, and TCP from an aqueous
solution increased signicantly as the contact time
increased in days. The overall performance under
the inuence of contact time corresponds to the
maximum PCP removal rate (90%) recorded in
25 days and was consi s tent with the maximum
removal rate recorded for TCP (87.99%). At
contact time days removal eciency was
less than the outcome sugge s ts biosorption
of the chemical species (TCP) was slow on the
CiL-plant surface or had not yet occurred for
PCP. Comparatively, the antagoni s tic eect of
contact time signicantly favored the removal of
PCP removal from the aqueous solution compared
with the performance of CiL-plant biosorbent on
the removal of TCP at maximum contact for 25
days. Sucient time >20 days allowed for the
biosorption of the contaminants (TCP and PCP) on
CiL-plant from the solution to reach equilibrium
[29]. The outline of the red and blue bars indicated
that biosorption of TCP reached equilibrium fa s ter
than PCP. The nding sugge s ts a relatively higher
level of tolerance of the CiL-plant index to PCP-
contaminated medium [8]. Overall performance
was very satisfactory, conrming the potential
of the CiL-plant as an active biosorbent for the
su s tainable removal of PCP and TCP to guarantee a
tolerable residual contaminant level. The outcomes
conrmed the inuence of the optimum contact
time on the removal of PCP and TCP from an
aqueous solution recorded at a p-value of 0.0001 at
a 95% condence interval.
Fig 6. Eects of time on the removal of PCP and TCP to Canna indica L.
at an optimum initial concentration (50 mg L-1) and pH 4
Days
Removal Rate (%)
0 5 10 15 20 25
0
20
40
60
80
100
TCP Removal rate (%) PCP Removal rate (%)
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
93
3.4.2 Antagoni s tic impact of pH on the
Phytoremediation process
The inuence of pH on the CiL-plant drove
biosorption of TCP from an aqueous solution at an
initial concentration of 100 mg L-1 at pH 4 is shown
in Figure 7. The graph showed that the biosorption of
PCP and TCP to CiL-plant at varying pH decreased
rapidly in an alkaline solution. The performance
of the CiL-plant translates to a removal rate of
49.3% for PCP, and 53.2% for TCP at pH 9 and an
optimum of 25 days, respectively. The removal rate
increased signicantly in a predominantly acidic
medium transcending to 82% for PCP, and 85.7%
for TCP at pH 4. The protonated chlorophenols
were more absorbable [30], which accounted for
the higher removal eciency recorded for PCP
and TCP at the lower pH value. The analysis of the
signicant impact of pH 4 on the treatment process
conrmed that, irrespective of the pH window,
the CiL-driven biosorption process favored TCP
removal from an aqueous medium in an acidic
medium compared to PCP at a p-value value of
0.0006 at 95% condence interval. This outcome
was consi s tent with the optimum pH 2 reported for
the removal of PCP and TCP reported in the work
of Radhika and Palanivelu et al. [29]. The outline of
the gure also indicates that CiL-plant has a higher
anity for TCP at the optimum conditions than PCP
[8], sugge s ting a superior solubility of 2,4,6-TCP
in water than PCP at optimum pH 4 in aquaponia.
Comparative evaluation of the maximum ecacy
of CiL-plant under the eect of contact time of 25
days (90, 87.99%), and eect of pH (4) of solution
corresponding (82, 85.7%) e s tablished that contact
time had a signicant main eect on the overall
performance of CiL-plant-driven phytoremediation
for su s tainability.
3.4.3. Synergetic impact of Time-pH and
concentration-pH on the Biosorption process
The signicance of the synergetic eect of
pH*Time and Time*initial concentration on
the response PCP and TCP removal rate was
conrmed based on the hierarchy from the VIFs
at a p-value less than 0.005 and 95% CI. The 3-D
surface plots in Figure 8 (a-d) were obtained from
the response surface design space to under s tand
better how the biosorption process works under
the interactive inuence of signicant parameters
[5, 8]. The red color gradient corresponds to higher
removal eciency of >87%. In comparison, the
yellow contour gradient corresponds to moderate
Fig 7. Eect of pH on the removal of PCP and TCP to Canna indica L.
at optimum Time (25days) and initial concentration of 100 mg L-1 at pH 4
pH
Removal rate (%)
4 5 6 7 8 9
40
50
60
70
80
90
TCP Removal rate (%) PCP Removal rate (%)
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
94
removal eciency of less than 70%, and the
dominant greenish-blue contour orientation
translates to lower removal eciency of less
than 50%. The decrease in removal eciency
can be traced to possible charge reversal from
surplus ions arising from the binary solution of
PCP and TCP under changing pH in the medium.
These excess negative charges contributed to the
building up of the concentrations of the PCP and
TCP molecules in an aqueous solution causing
eciency to drop signicantly. The curvature of
the red hue gradient on the base of the surfaces
Figures 8 (a-b) show that the eciency of removal
of PCP and TCP increased signicantly as the pH
of the solution decreased from 9 to predominantly
pH 4, as the exposure time of CiL-plant increased
from 20 to 25 days. The phytoremediation
performance of the biosorbent in aquaponia
transcends to increase in PCP biosorption rate
from 70 to 90%, while the TCP removal rate
increased from 70 to 87.99%, as illu s trated by the
ag-point in the respective contour plots in Figure
8 (a,b). The pH depression from 7 to 4 yielded
better performances that can be attributed to the
synergetic inuence of pH depression towards
the acidic window and the prolonged exposure
time of 25 days. This outcome is indicated by the
curvatures of the dominant red contour deviation
from the yellowish-green contour lines shown
in Figure 8 (a,b). The basic tendency of CiL-
plant-driven biosorption of PCP and TCP can be
expressed from the plant’s capability to drive the
removal rate towards a dominantly acidic medium
(pH 4) with no change in phase. The inuence
of the superior pH on the overall biosorption of
PCP and TCP in aquaponia was attributed to the
dissociation of mo s t chlorophenol in the form of
a salt which loses its negative charge easily when
pH is increased [30], thus making it dicult to be
adsorbed. The variation in the removal eciency
e s tablished that CiL-plant is tolerant in a solution
of PCP, compared to TCP. This observation agrees
with previous research works reported in the
literature [8, 30] and conrms the optimization
report in section 3.1. The synergetic eect of
initial concentration*contact time is illu s trated
in the outline the contour plot in Figure 8 (c,d).
The overall performance of the CiL-plant in the
phytoremediation remediation of aquaponia is
illu s trated by the deviation of the green color
contour from the dominant blue gradient on the
base of the surfaces in Figure 8c and Figure 8d,
respectively. The curvature of the light green
from the dominant blue color orientation is
attributed to areas of good performance of the
biosorbent morphology and adaptation of Canna
indica L for the removal of PCP and TCP. The
data points and orientation of the dominant
blue contour lines transcend to areas of poor
performance of the biosorbent on PCP and TCP
in solution. The intensity of the blue contour
gradients in Figure 8 (c,d) confirmed that the be s t
performance of the CiL-plant is adapted to an
initial concentration of less than 100 mg L-1. The
PCP and TCP removal eciency of the biosorbent
decreased as the initial concentration was
increased beyond 100 to 250 mg L-1. The output
reduces eciency from 75% to 63% for PCP and
<65% for TCP, as indicated by the curvature of
the blue contour gradient in Figure 8 (c,d) and
corresponding 3D surfaces in Figure 3. This
outcome indicates that the initial concentration
has a ceiling eect on the driving force of CiL-
plant biosorption of PCP and TCP [30, 31]. In
contra s t, the contact time or exposure had a
signicant main eect on the phytoremediation
process. The ndings are consi s tent with the
reports on TCP biosorption [29]. The authors
reasoned that if the concentration was to be
increased slightly beyond 100 mg L-1, and a
reduction in equilibrium exposure time below 25
days would decrease mass transfer to the surface
of the biosorbent, would inuence a reduction in
PCP, and TCP removal eciency signicantly
from 90% to 40%. This outcome e s tablished
the signicant impact of the interactive eect
of initial concentration and exposure time on
the overall performance of the CiL-plant-driven
phytoremediation process in Aquaponia.
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
95
Fig. 8. Synergetic eect of process variables on Phytoremediation
of Aquaponia (a) pH and Time on PCP, (b) pH-Time on TCP, (c) Conc-Time on PCP, (d) Conc-Time on TCP
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
96
3.5. Co s t analysis of the treatment process
The authors explored the co s t-benet evaluation of
the biosorption process as a decision-making tool
to te s t the feasibility of the active CiL-plant as a
biosorbent for removing PCP and TCP from an
aqueous solution. The co s t of the phytoremediation
operation is based on the performance of the active
CiL-plant, energy consumption, transportation to
the remediation site, labor and technology used to
remove contaminants [4, 8], and environmental and
regulatory risk. The techno-economic feasibility of
the biosorption process and phytotoxicity handling
necessitates using low-co s t materials (Canna
indica L.) with negligible environmental impact
and regulatory risk. The operating co s t of treating
1.0 L of the aqueous solution was calculated by
considering the co s t of preparing the 100 mg L-1
initial concentration of the aqueous solution for CiL-
plant as biosorbent. The labor co s t was determined
as a function of the number of working personnel
on board for the treatment operation. The power
consumption rate per unit of equipment utilized at
full scale (f=1), and the time spent following the
model report from previous research [15], were
evaluated following equations 2-5 presented in
section 2.5.
Analysis of Figure 9 conrmed that the operating
co s t of energy consumption corresponds to $ 27.80
for PCP, and $ 21.28 for TCP, respectively. It can be
observed from Figure 9 that, the co s t of operating
the phytoremediation process for PCP removal was
slightly higher than the co s t of TCP removal in terms
of energy consumption by $ 6.52. The preparation
of 100 mg L-1 initial concentration of CiL-plant
for removal of PCP from the aqueous solution
co s t $ 177.4 again s t $ 176.2 for removal of TCP
from the aqueous solution under similar operating
conditions. The labor co s t was projected to be $
100.2 per annum, while the co s t of transportation of
materials and personnel on board to the remediation
site was $12.00, irrespective of operating with PCP
or TCP. It can be concluded from the analysis of
Figure 9 showed that the overall co s t for using
the biosorption of PCP from aqueous solution to
CiL-plant at optimum conditions was computed as
$ 321.20 and $ 313.48 for TCP, respectively. The
analysis of the phytoremediation process proved
that, at the e s tablished optimum condition, the
opportunity co s t of operating the biosorption of
TCP from aqueous solution to CiL-plant would
save $ 7.72 compared with PCP for su s tainability.
The authors reasoned that the outcome is largely
due to higher solubility and rapid biosorption of
TCP to the surface of the CiL-plant in aquaponia,
irrespective of the longer exposure time required
for the CiL-plant driven biosorption of PCP and
TCP to reach equilibrium.
Fig. 9. Co s t evaluation summary of the CiL-plant-driven biosorption treatment
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
97
3.6. Comparison of CiL-plant for the remediation
of PCP and TCP from solution
The performance comparison of the biosorption of
PCP, and TCP from aquaponia was analyzed and the
report is summarized in Table 6 below. The previous
research reports on the CiL-plant phytoremediation
analysis applied the one-factor-at-a-time (OFAT)
approach for determining the optimum PCP and
TCP removal rate. The current s tudy applied the
design of experiment (DoE) approach via the RSM
for the optimization modeling of the uptake of PCP
and TCP to CiL-plant in aquaponia. The ndings
from the comparison of the results showed that with
the RSM, the removal eciency was higher than
the optimum reported via the OFAT approach. The
dierence corresponds to ± 1.87 and ±3 for TCP and
PCP, respectively.
4. Conclusion
The techno-economic evaluation and optimization
modeling of the competitive biosorption of PCP
and TCP from aqueous solution to the Cana indica
plant have been inve s tigated. The aqueous solution
of fertilizer contaminated with PCP and TCP was
prepared. The CiL-plant-driven phytoremediation
of the aqueous medium was s tudied at varying
pH, initial concentration, and con s tant time based
on the design of experiments. The optimization
modeling tools for ANN and RSM have yielded
good s tati s tical evaluation metrics for modeling
the CiL-plant-driven phytoremediation process.
The results conrmed that a s tati s tical dierence of
±0.005 was obtainable and adju s ted R2 ≤1.00. The
adopted RSM optimization outputs have to te s t their
reliability based on DoE. The predicted optimum
corresponds to pH, concentration, and exposure
time of 4, 50 mg L-1, and 25 days guaranteed PCP
and TCP biosorption to CiL-plant ≤90%. The
e s tablished optimum condition required $7.75 more
for su s tainable PCP removal than TCP.
5. Acknowledgement
We acknowledge the help of the technical s ta at
Chemi s try Laboratory, Imo s tate University, for
their support during the experimental set-up and
analysis
6. Conict of intere s t
There are no conicts of intere s t to declare
7. References
[1] USEPA, 2,4,6 Trichlorophenol. United s tates
Environmental Protection Agency, 2000. https://
www.epa.gov/sites/default/les/2016-09/
documents/2-4-6-trichlorophenol.pdf
[2] WHO, The World Health Organization
recommended classication of pe s ticides by
hazard and guidelines to classication 2004,
International Programme on Chemical Safety,
pp.19-39, 2005. https://apps.who.int/iris/re s t/
bit s treams/1278712/retrieve
[3] ATSDR, Pentachlorophenol, Agency for
Toxic Sub s tances and Disease Regi s try,
2021. https://wwwn.cdc.gov/TSP/sub s tances/
ToxSub s tance.aspx?toxid=70
[4] A.I. Olayinka, F.A. Ademola, I.A.
Emmanuel, Assessment of Organochlorine
and Organophosphorus pe s ticides residue in
water and sediments from Ero River in South-
We s tern Nigeria, J. Chem. Biol. Phys. Sci. D,
Table 6. Comparative analysis of the optimization report
Methodology
Parameter
DoE
This s tudy
(PCP) (TCP)
OFAT
Reference
(PCP) (TCP)
pH 4 ---
Initial Conc. (mg L-1) 50 100
Contact time (days) 25 25
optimum Eciency % 90 81.87 87 80
OFAT: Reference [8]
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
98
5 (2015) 4679–4690. https://www.jcbsc.org
[5] A.W. Verla, J. E. Ejiako, E.N. Verla, I. G.
Ndubuisi, C. E. Enyoh, Potential health
risk index of polyaromatic hydrocarbons
(PAHs), polychlorinated biphenyls (PCBs)
and organochlorine pe s ticides (OCPs) in
sh species from Oguta Lake, Nigeria, Int.
J. Environ. Anal. Chem., (2021) 1946687.
https://doi.org/10.1080/03067319.2021.1946
687
[6] B. M. Obidike, A. W. Verla, C. E. Enyoh, E. N.
Verla, O. N. Onyekachi, Fate and di s tribution
of pe s ticides residues in water sources in
Nigeria: A review, Int. J. Agrochem., 6 (2020)
29–48.
https://chemical.journalspub.info/index.
php?journal=IJCPD&page=article&op
=view&path%5B%5D=1015
[7] B.O. Isiuku, C.E. Enyoh, Water pollution
by heavy metal and organic pollutants:
Brief review ofsources, eects and progress
on remediation with aquatic plants, Anal.
Methods Environ. Chem. J., 2 (3) (2019) 5-38.
https://doi.org/10.24200/amecj.v2.i03.66
[8] C. E. Enyoh, B. O. Isiuku, Competitive
biosorption and phytotoxicity of
chlorophenols in aqueous solution to
Canna indica L, Curr. Res. Green Su s tain.
Chem., 4 (2021) 100094. https://10.1016/j.
crgsc.2021.100094
[9] C.E. Enyoh, B. O. Isiuku, Removal of
pentachlorophenol (PCP) from aqueous
solution using Canna indica L.: kinetics,
isotherm and thermodynamic s tudies, Arab.
J. Chem., 8(2) (2021) 193-213. http://www.
mocedes.org/ajcer/volume8/AJCER-13-
Enyoh-2021.pdf
[10] C.E. Enyoh, B.O. Isiuku,
2,4,6-Trichlorophenol (TCP) removal from
aqueous solution using Canna iindica L.:
kinetic, isotherm and Thermodynamic
s tudies, Chem. Ecol., 37 (2020) 64-82.
https://doi.org/10.1080/02757540.2020.1821
673
[11] D.H. Tran, T.M.H. Vi, T.T.H. Dang, R.
Narbaitz, Pollutant removal by Canna
Generalis in tropicalcon s tructed wetlands
for dome s tic wa s tewater treatment, Global
J. Environ. Sci. Manage., 5 (3) (2019) 331-
344. https://www.gjesm.net/article_35321_
dc150b14811afe228b6bea1bcc85228f.pdf
[12] C.E. Enyoh, B.O. Isiuku, Determination and
human health risk assessment of heavy metals
in ood basin soils in Owerri, southea s tern
Nigeria, Chem. Africa, 3 (2020) 1059–1071.
https://doi.org/10.1007/s42250-020-00171-2
[13] C.E. Enyoh, Q. Wang, P.E. Ovuoraye,
Response surface methodology for modeling
the adsorptive uptake of phenol from aqueous
solution using adsorbent polyethylene
terephthalate micropla s tics, Chem. Eng.
J. Adv., 12 (2022) 100370. https://doi.
org/10.1016/j.ceja.2022.100370
[14] S. Saini, J. Chawla, R. Kumar, Response
surface methodology (RSM) for optimization
of cadmium ions adsorption using C16-6-16
incorporated mesoporous MCM-41, SN Appl.
Sci., 1 (2019) 894. https://doi.org/10.1007/
s42452-019-0922-5
[15] P. E. Ovuoraye, V.I. Ugonabo, G.F.
Nwokocha, Optimization s tudies on turbidity
removal from cosmetics wa s tewater using
Aluminum Sulfate and blends of shbone,
SN Appl. Sci., 3 (2021) 488. https://dio.
org/10.1007/s42452-021-04458-y
[16] C. A. Igwegbe, L. Mohammed, S. Ahmadi, A.
Rahdar, Modeling of adsorption of methylene
blue dye on Ho-CaWO4 nanoparticles
using response surface methodology and
articial neural network (ANN) techniques,
Methods X, 6 (2019) 1779-1797. https://doi.
org/10.1016/j.mex2019.07.016
[17] C.A. Igwegbe, O.D. Onukwuli, J.O. Ighalo,
M.C. Menkiti, Bio-coagulation-occulation
(BCF) of municipal solid wa s te leachate
using picralima nitida extract: RSM and ANN
modelling, Curr. Res. Green Su s tain. Chem.
4 (2021) 100078. https://doi.org/10.1016/j.
crgsc.2021.100078
[18] C.E. Enyoh, C. E. Duru, P. E. Ovuoraye, Q.
Anal. Methods Environ. Chem. J. 6 (1) (2023) 79-99
99
Wang, Evaluation of nanopla s tics toxicity to
the human placenta in sy s tems, J. Hazard.
Mater., 446 (2023) 130600. https://doi.org/
10.1016/j.jhazmat.2022.130600
[19] R. Hasanzadeh, M. Mojaver, T. Azda s t, C. B.
Park, Polyethylene wa s te gasication syngas
analysis and multi-objective optimization
using central composite design for
simultaneous minimization of required heat
and maximization of exergy eciency, Energy
Convers. Manag., 247 (2021) 114713. https://
doi.org/10.1016/j.enconman.2021.114713
[20] L. Lujian, S. Tang, X. Wang, X. Sun, A. Yu,
Adsorption of malachite green from aqueous
solution by nylon micropla s tics: Reaction
mechanism and the optimum conditions by
response surface methodology, Process Saf.
Environ. Prot., 140 (2020) 339-347. https://
doi.org/10.1016/j.psep.2020.05.019
[21] C. E. Enyoh, Q. Wang, P. E. Ovuoraye, T. O.
Maduka, Toxicity evaluation of micropla s tics
to aquatic organisms through molecular
simulations and fractional factorial designs,
Chemosphere, 308 (2022) 136342. https://
doi.org/10.1016/j.chemosphere.2022.136342
[22] C.E. Duru, C.E. Enyoh, I.A. Duru, M. Enedoh,
Degradation of PET nanopla s tic oligomers
at the novel PHL7 target: insights from
molecular docking and machine learning, J.
Niger. Soc. Phys. Sci., 5 (2023) 1154. https://
doi.org/10.46481/jnsps.2022.1154
[23] G. Aicha, I. Soumaya, E. Noureddine, A.
T. Mohamed, G. Djamel, H. Ahmed, M.
Abdelhakim, A. Badreddine, K. Lioua,
Comparative s tudy of chemical coagulation
and electrocoagulation for the treatment of
real textile wa s tewater: optimization and
operating co s t e s timation, ACS Omega,
7(26) (2022) 22456–22476. https://doi.
org/10.1021/acsomega.2c01652
[24] P.E. Ovuoraye, L.C. Okpala, V.I. Ugonabo,
Clarication ecacy of eggshell and
aluminum base coagulant for the removal of
total suspended solids (TSS) from cosmetics
wa s tewater by coag-occulation, Chem.
Pap., 75 (2021) 4759–4777. https://doi.
org/10.1007/s11696-021-01703-x
[25] D. Wang, S. Thunell, L. Ulrika, A machine
learning frame work to improve euent quality
control in wa s tewater treatment plants, Sci.
Total Environ., 784 (2021) 147138. https://
doi.org/10.1016/j.scitotenv.2021.147138
[26] Y. Zhang, Y. Wu, Introducing machine
learning models to response surface
methodologies. Publisher: Intechopen,
270 pages, 2021. https://doi.org/10.5772/
intechopen.9819
[27] P. E. Ovuoraye, V. I. Ugonabo, M. A. Tahir,
P. A. Balogun, Kinetics-driven coagulation
treatment of petroleum renery euent using
land snail shells: An empirical approach
to environmental su s tainability, Clean.
Chem. Eng., 4 (2022) 100084. https://doi.
org/10.1016/j.clce.2022.100084
[28] H. Guo, Prediction of euent concentration
in a wa s tewater treatment plant using
machine learning models, J. Environ. Sci., 32
(2015) 90-101. https://doi.org/10.1016/j.
jes.2015.01.007.
[29] M. Radhika, K. Palanivelu, Adsorptive
removal of chlorophenols from aqueous
solution by low-co s t adsorbent—Kinetics and
isotherm analysis, J. Hazard. Mater., 138(1)
(2006) 116-124. https://doi.org/10.1016/j.
jhazmat.2006.05.045
[30] S.K. Nadavala, M. Asif, A. M. Poulos, M.
Suguna, M. I. Al‐Hazza, Equilibrium and
kinetic udies of biosorptive removal of
2,4,6‐trichlorophenol from aqueous solutions
using untreated agro‐wae pine cone biomass,
Processes, 7 (2019) 757. http://doi:10.3390/
pr7100757
[31] R. Bhutiani, N. Rai, P. K. Sharma, K. Rausa,
F. Ahamad, Phytoremediation eciency of
water hyacinth (E. crassipes), canna (C. indica)
and duckweed (L. minor) plants in treatment
of sewage water, Environ. Conserv. J., 20
(2019) 143-156. https://doi.org/10.36953/
ECJ.2019.1008.1221
Analysis and Biosorption of P&T chlorophenol in Cana indica. L Plant Chris tian Ebere Enyoh et al
