Content uploaded by R S Upendra
Author content
All content in this area was uploaded by R S Upendra on Jul 29, 2017
Content may be subject to copyright.
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1109
OPTIMIZATION OF FLUORIDE REMOVAL SYSTEM
USING Ocimum SP. LEAVES AND RAGI SEED HUSK
BY APPLYING BIO-STATISTICAL TOOLS
Upendra R.S.*, Pratima Khandelwal1, Amiri Z.R.2, Amulya Achar, Kumari B.G.,
Sowmaya M. and Tejaswini J.1
1. Department of Biotechnology, New Horizon College of Engineering, Marathahalli,
Bangalore, Karnataka (INDIA)
2. Department of Food Science & Technology, Sari Agricultural Sciences and Natural
Resources University, Sari (IRAN)
Received January 10, 2015 Accepted May 23, 2015
ABSTRACT
Fluoride is a naturally occurring element in water systems and enters food chain mostly through
drinking water. The WHO permissible limit of fluoride in water is 1.0 mg/l. At < 1.0 mg/l, it
inhibits dental caries, at > 1.0 mg/l causes molting of teeth, lesion of endocrine glands, thyroid,
liver and other organs. At still higher concentration (3-6 mg/l), it causes skeletal fluorosis.
Existing fluoride removal techniques haven’t been very effective as these remove fluoride only
up to 2 mg/l. Therefore, an economically viable, eco-friendly and easy method for defluoridation
of drinking water is highly desirable. In the present investigation, Ocimum sp. leaves along with
ragi seed husk was used as natural fluoride adsorbents and the process parameters such as
absorbent dosage (1-10 g/l), pH (3-12) and contact time (10-150 min) were optimized using
Central Composite Design (CCD) of Response Surface Methodology (RSM). The fluoride
content in the water was quantitatively determined by UV spectrophotometric analysis and the
presence of fluoride in the treated Ocimum sp. leaves were identified with EDAX analysis. RSM
design optimized conditions i.e - 5.5 g/l each of Ocimum sp. leaves and ragi seed husk, 6.0 pH
and 50 min contact time gave the end values of 0.43 mg/l of fluoride. The optimized values of
RSM with respect to the end fluoride content (0.43 mg/l) after treatment process were validated
using feed forward model of Artificial Neural Network (ANN). ANN predicted value (0.4250
mg/l) was very close to the optimized experimental value of RSM design (0.43 mg/l) and the
error was 0.049. In conclusion, an optimized process was developed for the removal of fluoride
from the drinking water using Ocimum sp. leaves and Ragi seed husk as natural fluoride
adsorbents. Final concentration of 0.43 mg/l of fluoride was achieved from initial concentration
of 10mg/l.
Key Words : Fluoride removal, Adsorption, Ocimum sp. leaves and Ragi seed husk,
Response Surface Methodology (RSM), Artificial Neural Network (ANN), EDAX analysis
INTRODUCTION
Fluoride is a naturally occurring element in
minerals, geochemical deposits and natural water
systems and enters food chains through either
drinking water or eating plants and cereals.
Fluoride is often called a two-edge sword – in
small dosages, it has remarkable influence on the
dental system by inhibiting dental carries, while
in higher dosages More than 1.5 mg/l causes
molting of teeth, lesion of endocrine glands,
thyroid, liver and other organs1. At still higher
concentration, 3-6 mg/l skeletal fluorosis occurs.
The disease affects the bone and
ligaments. According to WHO, the allowable
limit of fluoride in drinking water is 1 mg/l2.
(Table 1). Countries like India, China, Sri Lanka,
West Indies, Spain, Holland, Italy, Mexico,
North and South America are reported to have
high fluoride content in their ground water. In
India, high concentrations of fluoride in ground
water are common in some of the semi-arid areas
of Karnataka, Andhra Pradesh, Rajasthan,
Southern Punjab, Gujarat, Tamil Nadu, Madhya
Pradesh, and Southern Haryana states of India3.
Several areas of Karnataka have high
*Author for correspondence
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1110
concentrations of fluoride in ground water
exceeding 5 mg/l4. There are a number of cases of dental and skeletal fluorosis in these areas
(Table 2).
Table 1 : Permissible limit of fluoride in drinking water prescribed by various organizations
S/N Name of the organization
Desirable
limit (mg/l)
1 Bureau of Indian Standards (BIS)
0.6-1.2
2 Indian Council of Medical Research (ICMR) 1.0
3
The Committee on Public Health Engineering Manual and
Code of Practice, Government of India 1.0
4
World Health
Organization (International Standards for
Drinking Water) 1.0
Table 2 : Percentage of water source with different fluoride content in the stated districts
and percentage of villages with the cases of fluorosis5.
S/N District Taluk (samples)
% of sources
having conc.
in mg/l
Villages
with
fluorosis
cases
<
1.5
mg/l
1.5
-
3.0
mg/l >3.0 mg/l
1
Kolar
Bangarpet (257)
85
15
00
15
Gudibanda (58)
81
19
00
35
2
Chitradurga
Hiriyur (173)
45
53
02
28
Hosadurga
73
22
05
22
3
Bijapur
Bagalkot(100)
63
32
05
20
Jamakhandi(73)
89
08
03
00
High profile of fluoride in shallow zone
ground water is due to the geochemical
disposition in the vicinity of the ground water
extraction structures. The toxicity of fluoride is
also influenced by high ambient temperature,
alkalinity, calcium and magnesium contents in
the drinking water. The most commonly used
methods for the defluoridation of water are
adsorption, ion exchange, precipitation,
Donnan dialysis and electrodialysis. Among
these methods, adsorption is the most widely
used method for the removal of fluoride from
water6. Several techniques have been
developed for removal of fluoride from
drinking water by adsorption and precipitation
processes. Though this technique has been
extensively used in India, but due to its high
cost, alkaline pH and large dosage, it was not
suitable for field application. It is also not
suitable for drinkable purpose as it removes
fluoride only up to 2 mg/l. Therefore, an
economically viable, eco friendly and easy
method for defluoridation of drinking water is
highly desirable. No research document was
found on using Ocimum sp. leaves as a natural
bio-adsorbent to remove fluoride.
In the present investigation, Ocimum sp. leaves
along with ragi seed husk was used as natural
fluoride bio-adsorbents and the process
parameters such as bio-absorbent dosage (1-10
g/l), pH (3-12) and contact time (10-150 min)
were optimized using Central Composite
Design (CCD) of Response Surface
Methodology (RSM). The fluoride content in
the water was quantitatively determined by UV
spectrophotometric analysis and the presence
of fluoride in the treated Ocimum sp. leaves
were identified with EDAX analysis. RSM
design optimized conditions i.e - 5.5 g/l each
of Ocimum sp. leaves and ragi seed husk, 6.0
pH and 50 min contact time had given the end
values of 0.43 mg/l of fluoride. The optimized
values of RSM with respect to the end fluoride
content (0.43 mg/l) after treatment process
were validated using feed forward model of
Artificial Neural Network (ANN). ANN
predicted value (0.4250 mg/l) was very close
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1111
to the optimized experimental value of RSM
design (0.43 mg/l) and the error was (0.049).
In conclusion, an optimized process was
developed for the removal of fluoride from the
drinking water using Ocimum sp. leaves and
Ragi seed husk as natural fluoride bio-
adsorbents. Final concentration of 0.43 mg/l of
fluoride was achieved.
AIMS AND OBJECTIVES
Removal of fluoride from portable drinking
water using Ocimum sp. leaves and ragi seed
husk as a natural bio-adsorbents. Optimization
of the fluoride removal process by applying
RSM and ANN design.
MATERIAL AND METHODS
All the chemicals and reagents used in this
study were of Analytical Grade (Merck and
Qualigens).
Collection of bio-adsorbent
Ocimum sp. plants were purchased from the
Lalbagh Botanical garden Bangalore, India.
Ragi seeds husk was purchased from local
market. These were kept at room temperature
for further bio-adsorption studies.
RSM experimental design and bio-
adsorption studies
A three factor Central Composite Design
(CCD) of RSM was generated with the design-
expert 9.0.0.7 software. The model applied
was CCD and a second order polynomial
response equation gives the final content
fluoride in treated water samples.
Y= bo +∑biXi + ∑bi2Xi2 + ∑bijXiXj
Where Y is lovastatin yield, bo is the intercept,
bi is the coefficient for linear direct effect, bij is
the coefficient for interaction effect. Bio-
absorbent dosage (1-10 g/l), pH (3-12) and
contact time (10-150 min) were the principle
input variables, the factor levels were coded as
−1 (low), 0 (central point) and 1 (high).
Adsorption studies were carried out as a batch
experiments (triplicate) based on RSM CCD as
shown in Table 1, with in the 250 ml conical
flask with the volume of 100 ml as a test
solution. The solution was then filtered and the
residual fluoride ion concentration was
estimated through UV spectrometric analysis7.
Quantification of fluoride by UV
spectrometric analysis
Treated and prepared water sample were
analyzed qualitatively by measuring the
absorbance for the presence of fluoride at
wavelength 520 nm in three replicates using
pure fluoride (Merck) as a standard in
UV/Visible spectrophotometer. (Shimadzu,
Model no UV-2450 and Software UV-probe
2.21). A calibration curve was prepared from
the plot of absorbance against concentration of
standard solutions. The concentrations of the
sample solutions were determined from the
plot8.
Characterization of bio-adsorbent by SEM
and EDAX analysis
After the adsorption process the Ocimum sp.
leaves used as bio-adsorbent in the study was
examined by SEM experiments using Philips
Scanning Electron Microscope. The presence
of elemental silver in bio-adsorbent was
analyzed using energy dispersive spectroscopy
attached to SEM9.
Artificial Neural Network (ANN)
In this study, Neural Network MATLAB R
2011a mathematical software was used for
simulation.10 The same experimental data of
RSM design was employed in designing the
artificial neural network. The input variables
were bio-absorbent dosage (1-10 g/l), pH (3-
12) and contact time (10-150 min). The
optimum adsorption capacity was used as a
target. The data were randomly divided into
three groups, 70% in the training set, 15% in
the validation set and 15% in the test set.
Trainlm is a training function selected and it is
a network training function that updates weight
and bias values according to the Levenberg-
Marquardt algorithm. All variables and
response were normalized between 0 and 1 for
the reduction of network error and higher
homogeneous results.11-16
The normalization equation applied was as
follows : Ya-Ymin
Yn=
Ymax-Ymin
Yn, Ya, Ymin and Ymax were normalized value,
actual value, minimum value, and maximum
value, respectively.17-21
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1112
RESULTS AND DISCUSSION
Response surface methodology
Based on the experimental results of CCD in
Table 3, a quadratic polynomial was established
to identify the relationship between adsorption
capacity and process variables. Final Equation in
terms of coded factors represents the end fluoride
content End fluoride content = +0.72-0.84* A -
0.84* B+2.09* C+0.59* D-0.61* AB-0.56* AC -
0.12* AD-0.050* BC-0.90* BD+0.84* CD+0.84
* A^2+0.65* B^2+2.07* C^2+2.54* D^2.
(Table 3) The experimental results and
the predicted values obtained from the model
equation were compared. The Model F-value
of 14.49 implies the model is significant.
There is only a 0.01% chance that an F-value
this large could occur due to noise. Values of
"Prob> F" less than 0.0500 indicate model
terms are significant. In this case A, B, C,
AB, BD, A^2, B^2, C^2, D^2 are significant
model terms (Table 3). The predicted values
matched the experimental values reasonably
well with R-Squared = 0.9311. This implies
that 93.11% of the variations for adsorption
capacity are explained by the independent
variables, and this also means that the model
does not explain only about 6.89% of
variation. In addition, the value of adjusted
determination coefficient (Adj R-Squared
=0.8669) was also high, showing a
significance of the model.
Table 3 : ANOVA for response surface quadratic model fluoride content
ANOVA for Response Surface Quadratic model
Analysis of variance table [Partial sum of squares - Type III]
Source Sum of
squares df Mean
square F
value p-value
Prob> F
Model
116.44 14 8.32 14.49 < 0.0001
significant
A
-
Ocimum
sp.
leaves 9.70 1 9.70 16.89 0.0009
B
-
Ragi seed husk
9.74 1 9.74 16.97 0.0009
C
-
PH
20.70 1 20.70 36.05 < 0.0001
D
-
Contact time
1.66 1 1.66 2.89 0.1098
AB
5.93 1 5.93 10.33 0.0058
AC
2.21 1 2.21 3.85 0.0687
AD
0.11 1 0.11 0.18 0.6742
BC
0.017 1 0.017 0.030 0.8638
BD
5.65 1 5.65 9.84 0.0068
CD
2.21 1 2.21 3.85 0.0685
A^2
19.32 1 19.32 33.65 < 0.0001
B^2
11.48 1 11.48 19.99 0.0004
C^2
23.24 1 23.24 40.48 < 0.0001
D^2
34.67 1 34.67 60.39 < 0.0001
Residual
8.61 15 0.57
Lack of
f
it
8.61 10 0.86
Pure
e
rror
0.000 5 0.000
Cor
t
otal
125.05 29
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1113
Quantification of fluoride by UV spectro
metric analysis
Treated and prepared water sample were
analyzed quantitatively by measuring the
absorbance for the presence of fluoride at
wavelength 520 nm in three replicates using pure
fluoride (Merck) as a standard in UV/Visible
spectrophotometer. (Shimadzu, Model no UV-
2450 and Software UV-probe 2.21) the end
fluoride content of each run was mentioned in
the Table 4.
Characterization of bio-adsorbent by SEM
and EDAX analysis
Scanning electron microscope is one of the
powerful tools to identify the adsorbed content
of fluoride on the bio-adsorbents. Further,
elemental analysis is carried out to confirm the
presence of fluoride in the bio-adsorbents used.
EDAX analysis of Ocimum sp. leaves bio-
adsorbent shows an intense signal at 1 keV
indicating the presence of elemental fluoride in
the Ocimum sp. leaves bio-adsorbent. (Fig. 1)
Fig. 1 : EDAX analysis of Ocimum sp. leaves representing the fluoride element
Artificial Neural Network
The optimal architecture of ANN model in this
case is shown in Fig. 2. It has three-layer
ANN, with tangent sigmoid transfer function
(tansig) at hidden layer with 11 neurons and
linear transfer function (purelin) at output
layer. The MSE value was found to be 0.0003.
A regression analysis between ANN outputs
and the experimental data was carried out. This
ANN model indicated a precise and effective
prediction of the experimental data with a
correlation coefficient of 0.999, 0.999, 0.995
and 0.999 for training, validation, testing and
all data, respectively. (Fig. 3) The simulated
value of end fluoride content as predicted by
Feed forward model (0.4250 mg/l) of ANN
was in close agreement with the experimental
values (0.43 mg/l) and accurately matching
with the predicted value of central composite
design of RSM (Table 4).
Table 4 : Central composite design matrix of three variables and the experimentally
determined, RSM model predicted and ANN model predicted adsorption capacity
Run Ocimum
Leaves g/l
Ragi seed
husk g/l pH Contac
t time
Min
Concentration of Fluoride
mg/l Error
Experi
mental
value
RSM
predicted
ANN
predicted
1
1.00
1.00
3.00
10.00
5.375
5.141508
5.335174
0.039
2
1.00
10.00
3.00
10.00
5.22
5.256838
5.181653
0.0383
3
10.00
1.00
3.00
10.00
5.67
5.254838
5.634824
0.0351
4
-
3.50
5.50
6.00
50.00
4.86
5.049988
4.856290
0.0037
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1114
5
1.00
1.00
3.00
100.00
2.96
2.723321
2.954555
0.0054
6
10.00
10.00
9.00
100.00
0.5
7
0.535175
0.5
59591
0.0104
7
10.00
1.00
9.00
10.00
4.11
4.336508
4.079444
0.0305
8
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.0049
9
5.50
5.50
6.00
150.00
3.12
2.271654
3.071939
0.0480
10
5.50
5.50
12.00
50.00
4.55
3.867488
4.651360
0.101
11
1.00
10.00
3.00
100.00
2.84
2.745175
2.834253
0.00574
12
10.00
10.00
3.00
100.00
0.051
0.269004
0.387760
0.0122
13
1.00
1.00
9.00
10.00
5.52
5.514614
5.519273
0.00072
14
1.00
10.00
9.00
10.00
6.85
6.358008
6.664666
0.1853
15
14.50
5.50
6.00
50.00
1.93
1.395671
1.974661
0.044
16
5.50
14.50
6.00
50.00
2.27
2.007171
2.251866
0.0181
17
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.0049
18
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.0049
19
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.0049
20
5.50
5.50
0.00
50.00
2.89
3.228171
2.843741
0.0462
21
10.00
1.00
9.00
100.00
0.35
0.525821
0.354684
0.00468
22
5.50
-
3.50
6.00
50.00
1.954
1.877488
1.942772
0.01122
23
1.00
10.00
9.00
100.00
3.675
4.303821
3.670765
0.004235
24
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.0049
25
10.00
1.00
3.00
100.00
0.364
0.987675
0.349230
0.014797
26
10.00
10.00
9.00
10.00
3.99
4.439338
3.692805
0.297195
27
10.00
10.00
3.00
10.00
4.86
4.629642
4.817561
0.42439
28
1.00
1.00
9.00
100.00
3.1909
3.552942
3.439222
-
0.24832
29
5.50
5.50
6.00
-
50.00
8
.09
8
.594004
8.168846
-
0.07885
30
5.50
5.50
6.00
50.00
0.43
0.43
0.425098
0.004902
Fig 2 : Topology of Feed Forward Model of ANN with input layer, hidden layer and output layer
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1115
Fig. 3 : ANN regression plots showing training, validation, target and all regression values
CONCLUSION
Ocimum sp. leaves along with ragi seed husk
was used as natural fluoride adsorbents and the
process parameters such as absorbent dosage
(1-10 g/l), pH (3-12) and contact time (10-150
min) were optimized using Central Composite
Design (CCD) of Response Surface
Methodology (RSM). The optimized values of
RSM with respect to the end fluoride content
(0.43 mg/l) after treatment process were
validated using feed forward model of
Artificial Neural Network (ANN). ANN
predicted value (0.4250 mg/l) was very close
to the optimized experimental value of RSM
design (0.43 mg/l) and the error was 0.049. In
conclusion, an optimized process was
developed for the removal of fluoride from the
drinking water using Ocimum sp. leaves in
combination with ragi seed husk as natural
fluoride adsorbents. Final concentration of
0.43 mg/l of fluoride was achieved.
ACKNOWLEDGEMENT
We wish to express our sincere gratitude to
The Chairman and the Principal, New Horizon
College of Engineering, Bangalore, India for
providing all facilities to undertake research
work. We extend our sincere thanks to VGST,
Bangalore, India for their funding under TRIP
programme.
REFERENCES
1. Sushree Swarupa Tripathy, Jean-Luc
Bersillon, Krishna Gopal, Removal of
fluoride from drinking water by adsorption
on to alum-impregnated activated alumina.
Sep. Pur. Technol., 50(1), 310–317,
(2006).
2. http://www.nih.ernet.in/rbis/india_informat
ion/fluoride.htm, 22-8-14, (2014).
3. Alagumuthu G., V. Veeraputhiran and R.
Venkataraman, Adsorption isotherms on
fluoride removal : Batch techniques. Arch.
Appl. Sci. Res., 2(4), 170-185, (2010).
4. http://www.iisc.ernet.in/currsci/mar25/arti
cles13.htm, 28-8-14, (2014).
5. The Danida assisted rural drinking water
supply and sanitation, Project by RDWSS,
Karnataka, India, (2013).
6. Vaishali Tomar and Dinesh Kumar, A
critical study on efficiency of different
materials for fluoride removal from
J. Environ. Res. Develop.
Journal of Environmental Research And Development Vol. 9 No. 04, April-June 2015
1116
aqueous media. Chem. Cent. J., 7(1), 51-
55, (2012).
7. Mourabet M., A. El Rhilassi, M. Bennani-
Ziatni and A. Taitai., Comparative study of
artificial neural network and response
surface methodology for modelling and
optimization the adsorption capacity of
fluoride onto apatitic tricalcium phosphate.
Univ. J. App. Math., 2(2), 84-91, (2014).
8. Paul E.D., Gimba C.E., Kagbu J.A.,
Ndukwe G.I. and Okibe F.G., Spec-
trometric determination of fluoride in
water, soil and vegetables from the precinct
of River Basawa, Zaria, Nigeria. J. Bas.
Appl. Chem., 1(6), 33-38, (2011).
9. Paulkumar Kanniah, Gnanadhas
Gnanajobitha, Mahendran Vanaja, Shan-
mugam Rajeshkumar, Chelladurai Mal-
arkodi, Kannaiyan Pandian and Gurusamy
Annadurai, Piper nigrum leaf and stem
assisted green synthesis of silver nano-
particles and evaluation of its antibacterial
activity against agricultural plant
pathogens. Sci. Wor. J., 1(1), 1-9, (2014).
10. Upendra R.S, Pratima Khandelwal, Zeinab
Raftani Amiri, Rahila Banu, Aruna Barade,
Veena K., Gayathri V. and Yamini D.E.
Artificial Neural Network : A novel
method for optimization of bioproducts and
bioprocesses : A critical review. MSR J.
Sci., 1(1), 21-34, (2014).
11. Kardam Abhishek, Kumar Rohit Raj, Jyoti
Kumar Arora, Man Mohan Srivastava and
Shalini Srivastava, Artificial neural
network modeling for sorption of cadmium
from aqueous system by shelled Moringa
oleifera seed powder as an agricultural
waste. J. Wat. Res. Prot., 2(1), 339-344,
(2010).
12. Dhawas Sonali, Dhurvey V., Kodate Jaya
and Urkhude Rashmi, An epidermilogical
study of skeletal fluorosis in some villages
of Chandrapur district, Maharashtra, India,
J. Environ. Res. Develop., 7(4A), 1679-
1683, (2013).
13. Bhatnagar M.K., Singh Raviraj, Gupta
Sanjay and Bhatnagar Prachi, Study of
tannery effluents and its effects on
sediments of river Ganga in special
reference to heavy metals at Jajmau,
Kanpur, India, J. Environ. Res. Develop.,
8(1), 56-59, (2013).
14. Dhurvey Varsha and Marganwar Ragini,
Prevalance and severity of dental flurosis
among school students in Dongargaon of
Chandrapur district, Maharashtra, India, J.
Environ. Res. Develop., 8(2), 309-314,
(2013).
15. Saxena Richa and Sharma Manju,
Qualitative and Quantitative evaluation of
water sources of few areas in and around
Gwalior, M.P., India, J. Environ. Res.
Develop., 8(3), 459-469, (2014).
16. Pandey Ambuj, Groundwater quality study
of Bilaspur, Chattisgarh, India, J. Environ.
Res. Develop., 8(3A), 582-586, (2014).
17. Narwaria Y.S., Khushwah Khush and
Saxena D.N., Study of groundwater quality
at Karera block of Shivpuri district, M.P.,
India, J. Environ. Res. Develop., 9(2), 562-
576, (2014).
18. Pujari S. A. and Gandhi M.B., Studies on
effects of seed and leaf extracts of Mucuna
pruriens on some common bacterial
pathogens, J. Environ. Res. Develop., 8(1),
50-54, (2013).
19. F. Maqdoom, Hashmi S. and Shaikh Zarina,
Papaya fruit extract : A potent source for
synthesis of bio-nanoparticle, J. Environ.
Res. Develop., 7(4A), -1518-1522, (2013).
20. Jessen George, Divya L. and
Suriyanarayanan S., Qualitative microbial
risk assessment in the management of E.
coli strains via drinking water, J. Environ.
Res. Develop., 8(1), 60-68, (2013).
21. Sharma H.P. and Rawal Ajitkumar, Health
security in ethic communities through
nutraceutical leafy vegetables, J. Environ.
Res. Develop., 7(4), 1423-1429, (2013).