Ecological Indicators 139 (2022) 108913

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Original Articles

Risk assessment of groundwater pollution during GLDA-assisted

phytoremediation of Cd- and Pb-contaminated soil
Lirong Liu a, Dinggui Luo a, b, c, d, *, Yayin Lu a, f, Xuexia Huang a, b, c, d, Yu Liu a, b, c,
Lezhang Wei a, b, c, Tangfu Xiao a, c, e, Qihang Wu c, d, Guowei Liu a
a
School of Environmental Science and Engineering, Guangzhou University, Guangzhou 510006, PR China
b
Linköping University—Guangzhou University Research Center on Urban Sustainable Development, Guangzhou University, Guangzhou 510006, PR China
c
Key Laboratory for Water Quality and Conservation of the Pearl River Delta, Ministry of Education, Guangzhou University, Guangzhou 510006, PR China
d
Guangdong Provincial Key Laboratory of Radionuclides Pollution Control and Resources, Guangzhou University, Guangzhou 510006, PR China
e
State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu 610059, PR China
f
Lide Environmental Protection Equipment Manufacturing (Guangzhou) Co. LTD., No. 19, Xingye Road, Dayao Village, Lanhe Town, Nansha District, Guangzhou,
511480, PR China

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

Keywords: Using biodegradable chelate agents during phytoremediation of heavy metal contaminated soil is a growing
Chelate-assisted phytoremediation consensus. However, the potential risk of groundwater pollution and the corresponding quantification during
Hydrus-1D biodegradable chelate-assisted phytoremediation is currently ignored. This study provides a risk-assessment
Groundwater risk
framework containing the practical boundaries and the transport model of chelate-metal complexes and
Biodegradable chelate
metals at the subsoil, to quantify and predict the leaching risk of biodegradable N,N-bis(carboxymethyl)glutamic
Cadmium
Lead acid (GLDA)-induced phytoremediation during heavy rainfall. The case of ethylenediaminetetraacetic acid
(EDTA)-assisted phytoremediation was also considered. Here, adsorption experiments and percolation experi­
ments were conducted based on the subsoil to obtain adsorption parameters and hydrodynamic parameters.
Optimal parameters derived from inverse fitting and the boundary condition originated from pot leaching ex­
periments were used for the framework establishment and Hydrus-1D simulation. Results showed that fewer
GLDA-metal complexes were adsorbed on the soil particles compared with metals, inferring their great mobility
in the subsoil. However, the chemical and physical non-equilibrium process indicated by the asymmetric
breakthrough curves of the percolation experiments would also affect the transport behavior of contaminants in
the sub-vadose zone. In addition, adsorption parameters and optimal hydrodynamic parameters aptly described
breakthrough curves of the chelate-metal complexes, and could better predict their transport behavior at the
subsoil. Opposite to the previous findings, prediction results showed that heavy rainfall during GLDA-assisted
phytoextraction resulted in Cd and Pb pollution in shallow groundwater, where the pollution degree was
related to properties of the soil horizon, depth of the water table, and types of contaminants. The degradation
feature of GLDA-metal complexes would fail to keep groundwater from heavy metal contamination at the sub-
vadose zone with high permeability. Thus, we recommend our risk-assessment framework to optimize the
operating condition before utilizing the biodegradable chelate-assisted phytoremediation.

1. Introduction particular, mining activities have destroyed local ecosystems and

released plentiful heavy metals to soil and water. Thus, soil remediation
Heavy metals pollution of soil is a critical environmental problem is an urgent need. For decades, eco-friendly and cost-effective phytor­
worldwide (Ashraf et al., 2019; Gall et al., 2015). Generally, heavy emediation techniques, especially phytoextraction that can permanently
metals are predominantly released during anthropogenic activities, remove heavy metals from the soil, have captured the interest of many
including industrial and agrochemical activities (Yang et al., 2018). In researchers (Sarwar et al., 2017). Recent studies tend to add chelating

* Corresponding author at: School of Environmental Science and Engineering, Guangzhou University, Guangzhou 510006, PR China.
E-mail addresses: [email protected] (L. Liu), [email protected] (D. Luo), [email protected] (Y. Lu), [email protected] (X. Huang),
[email protected] (Y. Liu), [email protected] (L. Wei), [email protected] (T. Xiao), [email protected] (Q. Wu), [email protected] (G. Liu).

https://doi.org/10.1016/j.ecolind.2022.108913
Received 9 June 2021; Received in revised form 25 March 2022; Accepted 22 April 2022
Available online 28 April 2022
1470-160X/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L. Liu et al. Ecological Indicators 139 (2022) 108913

agents to enhance phytoextraction by activating more bioavailable during GLDA-assisted phytoremediation was unique compared with the
heavy metals from the soil (Lasat, 2002; Sarwar et al., 2017). Although permanent chelates flushing. Meanwhile, hydraulic and chemical
many studies have verified the excellent effect of EDTA on promoting properties of the practical subsoil would impact the migration of
the phytoextraction (do Nascimento et al., 2006; Luo et al., 2016), the biodegradable GLDA-metal complexes and their by-products, which is
toxicity of EDTA to plants and microorganisms and its non-degradability short of illustration in the previous works. Hence, simply modifying the
have prompted researchers to find effective and degradable alternatives numerical model of chelate flushing and transferring to the scenario of
(Tandy et al., 2004). Liu et al. (2020b) compared the efficiency of Pb and chelate-enhanced phytoremediation is unreasonable. An assessment
Cd activation among nine chelates involving aminopolycarboxylic acids, framework containing the plants growing at the surface soil, degrada­
natural low molecular weight organic acids, and surfactants, and indi­ tion of GLDA-metal complexes during the migration at the practical
cated the relatively good performance of a biodegradable chelate, subsoil, and the groundwater table is absent now, which is needed for
GLDA. He et al. (2020) and Wang et al. (2019) found that the addition of risk prediction of groundwater pollution during the GLDA-induced
5 mmol kg− 1 GLDA increased the total Cd content in the tissue of phytoremediation.
Trifolium repens and Amaranthus hypochondriacus L. by a factor of 4.67 Therefore, the objectives of this study are: (i) to establish the nu­
and 3.28, respectively. Notably, the addition of GLDA could increase the merical model between chelate-metal complexes, metal, and subsoil, (ii)
content of dissolved organic carbon and the activity of antioxidant en­ to obtain and compare the adsorption parameters and behavior through
zymes in soil, which is beneficial to the bacterial community, and the batch adsorption experiments, (iii) to depict the migration behavior of
growth of plants in the heavy metals contaminated environment (Diarra chelate-metal complexes and metals in the undisturbed soil columns
et al., 2021; Luo et al., 2021; Wang et al., 2020; Yang et al., 2021). Thus, from the two different sub-vadose zones, and derive the hydraulic pa­
as a biological remediation technique, GLDA-induced phytoextraction is rameters by percolation experiments and parameters optimization, and
a good choice for repairing heavy metals contaminated soil. (iv) to ascertain the practical boundaries by pot cultivation and leaching
Although the risk of groundwater pollution during the permanent experiments, and integrated into the quantitative prediction and
chelate-induced phytoremediation has been noted in the early years assessment framework. Subsequently, the pollution risk of groundwater
(Cooper et al., 1999), rare studies provided exact evidence that the during GLDA-assisted phytoremediation will be investigated and
degradation feature of GLDA could prevent groundwater pollution compared with EDTA-induced phytoremediation.
during phytoremediation. At present, several researchers have focused
on the leaching risk of biodegradable chelate-metals complexes through 2. Methods and materials
the evaluation methods including the water extraction, the toxicity
characteristic leaching procedure (TCLP), or the column experiments 2.1. Theory of the transport model for chelate-metal complexes
with and without plants (Lu et al., 2017; Madrid et al., 2003; Van
Engelen et al., 2007; Wu et al., 2004). For example, González et al. In the present study, one-dimensional transport of chelate-metal
(2011) applied pot leaching experiments and collected the leachate from complexes, including EDTA-metal complexes (non-degradable) and
the different depths of the soil to explore the leaching risk during GLDA-metal (biodegradable) complexes, in the homogeneous, rigid, and
methylglycinediacetic acid (MGDA)-induced phytoremediation. They saturated soil under steady-state flow conditions was described with a
speculated no Cu pollution at the groundwater level of 11 m according modified CDE. The conceptual transport model of degradation-resistant
to the heavy metal content leached from the 60 cm horizon. Yu et al. EDTA-metal complexes was established on the basis of transient sorption
(2020) separated the soil (20–40 cm) from the pots and compared the equilibrium of chelate-metal complexes during percolation. Considering
percentages of exchangeable Pb and Pb bound to carbonates with a risk convection, dispersion, and adsorption, the CDE of EDTA-Cd complexes
assessment code. Analogously, the nitrilotriacetic acid (NTA)-assisted or EDTA-Pb complexes was modified as:
phytoremediation was deduced as a safe technique without increasing
Pb leaching risk into groundwater. These inferences deserved suspicion θ∂C ρ∂s D∂2 C q∂C
+ = − (1)
because experiments were only conducted with the topsoil (the depth ∂t ∂t ∂x2 ∂x
less than 1 m), which was not appropriate to represent the fate of
where θ is the effective porosity, ρ (g/cm3) is the bulk density, D (cm2/d)
biodegradable chelate-metal complexes at the sub-vadose zone. Unfor­
is the hydrodynamic dispersion coefficient, q (cm/d) is the Darcy ve­
tunately, the quantitative methods on forecasting the transport behavior
locity, that can be substituted by saturated hydraulic conductivity (Ks), C
of chelate-metal complexes (like GLDA-metal complexes) at the subsoil
(mg/cm3) is the solution concentration, s (mg/g) is the concentration of
and assessing the risk of groundwater pollution are lacking for chelate-
target contaminants adsorbed on soil particles, x (m) is the distance
induced phytoremediation.
migrated by the contaminants and t (d) is the time.
To better quantify the transport behavior and associated risks of
The hydrodynamic dispersion coefficient D relates the longitudinal
biodegradable chelate-metal complexes, the mathematical model is
dispersion λ (m) and the effective molecular diffusion coefficient DW
frequently recognized as an effective and efficient tool, because con­
(cm2/d) and can be expressed as:
ducting large-scale experiments that span several meters of soil horizon
or excavating monitor wells is costly and unrealistic. Particularly, D = λ νn + D W (2)
Hydrus-1D, developed by Simunek (2005), has superiority in the nu­
merical simulation of contaminants migration in soil, by considering the where v (cm/d) represents the average porous velocity, and n commonly
physical and chemical process to modify the convection–dispersion ranges from 1 to 2 (Comegna et al., 2001).
equation (CDE) (Behroozi et al., 2020; Jacques et al., 2008; Simunek In addition to considering convection, dispersion, and adsorption,
et al., 2008; Ye et al., 2019). To date, the transport model of chelate- the degradation of GLDA-metal complexes should be taken into account.
metal complexes was only developed for the technique of chelates Therefore, the modified CDEs for GLDA-Cd complexes or GLDA-Pb
flushing, which was conducted with a large amount of chelate solution complexes are:
as input and usually ex-situ treatment (Luciano et al., 2012). For
θ∂C1 ρ∂s1 D1 ∂2 C1 q∂C1
instance, Monika A. M. Kedziorek et al. (1998) and Mayes et al. (2000) + = − − μθC1 − μρs1 (3)
∂t ∂t ∂x2 ∂x
established a mathematical model involving first-order and second-
order kinetics for the dissolution of EDTA-metal complexes, coordina­ θ∂C2 ρ∂s2 D2 ∂2 C2 q∂C2
tion competition, and ions exchange, to investigate the transport of + = − + μθC1 + μρs1 (4)
∂t ∂t ∂x2 ∂x
EDTA and EDTA-metal complexes during EDTA flushing. Notably, the
process of plants extraction and degradation of GLDA-metal complexes where D1 (cm2/d) represents the hydrodynamic dispersion coefficient

2
L. Liu et al. Ecological Indicators 139 (2022) 108913

for GLDA-metal complexes; D2 (cm2/d) represents the hydrodynamic of transport model establishment is illustrated in Fig. 1. The pot leaching
dispersion coefficient for metal ions; C1 and C2 (mg/cm3) are the con­ experiments were conducted by using the surface soil, while the
centrations of GLDA-metal complexes and metal ions in the solution, formulation and optimization of the transport model were proceeded
respectively; and s1 and s2 (mg/g) are the concentrations of GLDA-metal based on the subsoil. Single and binary adsorption experiments were
complexes and metal ions on the solid phase, respectively. designed to calculate the parameters of Kd, β, and η. And percolation
Equations (3) and (4) introduce a degradation term (μθC1 and μρs1 ), experiments were implemented to obtain the parameters of Ks, λ, and
where the degradation parameter μ depends on the characteristics of the DW. The details are shown in Sections 2.2.2 and 2.2.3.
microbial community in the soil and the ability of microbes to degrade
chelate-metal complexes. In the present study, μ was deduced from our 2.2.1. Study sites and soil preparation
previous study (Mai et al., 2019). The empirical degradation constants The study sites were located near the Yunfu pyrite mine in Guang­
for GLDA-Cd complexes and GLDA-Pb complexes in this study, denoted dong Province, China, where the mean annual precipitation was 1550
μ1 and μ2, were 0.052 d-1 and 0.020 d-1, respectively. mm. Some agricultural areas historically polluted by the irrigation of
The solid concentration (s) of either the chelate-metal complexes or mine drainage have potential ecological risks (Huang et al., 2016;
metals for the adsorption term, ∂∂st , can be described as follows (Simunek, Huang et al., 2019). Restoration of these heavy metal contaminated soils
2005): was necessary. Thus, two sites with different sub-vadose zones were
chosen to investigate the risk of chelate-enhanced phytoremediation,
Kd Cβ which located at Lin House (E 112◦ 00′ 28.36′′ , N 23◦ 00′ 31.10′′ ) and near
s= (5)
1 + ηC β Liyushan Bridge (E 111◦ 59′ 42.82′′ , N 23◦ 01′ 25.93′′ ) and were denoted
as Y1 and Y2, respectively. We collected the surface agricultural soil
where Kd (L/g) is the distribution coefficient, β is a dimensionless
(0–20 cm) and the undisturbed soil columns of the sub-vadose zone from
empirical parameter, and η is an empirical parameter. When β is equal to
site Y1 and site Y2. At a depth of 150 cm, twenty undisturbed soil col­
one and η is equal to zero, equation (5) describes to the linear adsorption
umns at each sampling site were collected by utilizing a steel cylinder
model. When η = 0, equation (5) describes the Freundlich isotherm
with an internal diameter of 5 cm and a height of 10 cm. The inner layers
model, and Kf replaces Kd to represent the distribution coefficient. When
of the steel cylinder samplers were smeared with a thin layer of Vaseline
β = 1, equation (5) describes the Langmuir isotherm model.
before sampling to prevent preferential flow along the cylinder wall
When the adsorption behavior is fitted well by the linear model, a
during the percolation experiments, and the weight of the steel cylinders
retardation factor (R) is defined to describe the transport behavior of
was recorded. After digging out the steel cylinders, both ends of the soil
metal ions and chelate-metal complexes:
cylinders were trimmed carefully and then covered with aluminum caps,
ρKd and the columns were sealed in plastic bags before the percolation ex­
R = 1+ (6)
θs periments. At the same time, adjacent disturbed soils were collected for
adsorption experiments and the characterization of physical and
chemical properties.
2.2. Methodology The surface soil and the disturbed subsoil were both air-dried,
crushed, and sieved through a 2 mm nylon mesh. Artificially contami­
During the chelate-assisted phytoremediation, the plant roots can nated soil was prepared by spiking the surface soil with metal nitrate
extract the chelate-metal complexes and affect the water flow by water solutions to reach the soil concentrations of 3.5 mg/kg for Cd and 550.0
retention or constructing preferential flow. These would complicate the mg/kg for Pb and subsequently aged for 60 d before the pot cultivation
identification of contaminants flux from the surface region to the experiments. The undisturbed soil columns were dried at 105 ◦ C for 24
groundwater when considering the role of the roots. Herein, the topsoil h, and the bulk density was determined from the weight difference. Part
with plants was integrated as a source of leaching chelate-metal com­ of the disturbed soil was used to determine the physicochemical prop­
plexes, to simplify the establishment of the transport model of chelate- erties of the sub-vadose zone. Soil particle size distribution was
metal complexes. The framework of risk assessment and the procedure

Fig. 1. Framework of risk assessment and procedure of transport model establishment.

3
L. Liu et al. Ecological Indicators 139 (2022) 108913

measured by the pipette method (China standard: HJ 1068–2019) and implemented to solve the modified CDEs and obtain the simulated BTCs
the soil texture was confirmed as loam and sandy loam at site Y1 and site by inverse fitting. The upper boundary was set as a constant head and
Y2, respectively. Soil pH was detected at a solid-to-liquid ratio of 1:5. concentration boundary condition, and the lower boundary was free
The available cation exchange capacity (CEC) was measured by the drainage and zero concentration gradient. To indicate the accuracy of
hexamminecobalt trichloride solution/spectrophotometric method BTCs adjusting, Hydrus-1D applied the determination coefficient (r2)
(China standard: HJ 889–2017). Soil organic matter (SOM) was deter­ and the root mean square weighted error (RMSE), which were respec­
mined by titration with ferrous sulfate solution after oxidation of po­ tively calculated by equations (8) and (9). The hydrodynamic parame­
tassium dichromate and sulfuric acid solution (China standard: NY/T ters of λ and Dw were ascertained when the high appropriateness of the
1121.6–2006). And 0.5000 g of soil was digested with the HNO3-HF- BTCs fitting occurred, and the optimization of objective function during
HClO4 system (5:1:1, v/v/v) to detect the background metal contents, i. inverse fitting is described in Section 2.2.4.
e., Cd and Pb, by the flame atomic absorption spectrophotometer (ICE [ ∑ ∑ ]2
∑
3500, Fisher Scientific, Loughborough, UK). The physical and chemical wi COi CSi −
COi
∑ CSi
properties of the subsoil are listed in Table S1.
wi
2
r = [ ∑ 2 ][ ∑ 2] (8)
∑ ( ∑COi ) ∑ ( CSi )
2
wi COi − wi CSi 2 − ∑
2.2.2. Determination of adsorption parameters of chelate-metal complexes
wi wi

Adsorption parameters in the migration model of chelate-metal √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅

∑n
complexes were calculated by fitting the linear model, the Langmuir i=1 (CSi − COi )
2

isotherm, and the Freundlich isotherm. Thus, we performed single and RMSE = (9)
n
binary adsorption experiments, in which 1.0000 ± 0.0001 g of soil and
10 mL of the corresponding metal or chelate-metal complexes solutions where COi and CSi are the observed concentration and simulated value at
were mixed and shaken for 720 min at 25 ◦ C. The scheme of single and the ith point, respectively, n is the total number of observation points, wi
binary adsorption experiments is listed in Table S2. After equilibration, was the weight of ith observation point assumed 1.0 during inverse
the mixtures were centrifuged at 4000 rpm for 15 min, and the super­ fitting. The closer r2 is to one and RMSE to zero, the better the simula­
natant filtered through a 0.45 μm cellulose acetate membrane was used tions fit the BTCs (Legates and McCabe, 1999; Saso et al., 2012).
for metal content analysis by the atomic absorption spectroscopy (AAS).
Subsequently, the solid adsorption amount, s, was calculated by equa­ 2.2.4. Optimization algorithm and sensitivity analysis
tion (7) and used to fit the adsorption isotherm: The longitudinal dispersion (λ) readily varied in different soil col­
umns, and the effective molecular diffusion coefficient of the chelate-
V(ce − c0 )
s= (7) metal complexes (DW,1 and DW,2) was not be mentioned in the previ­
m
ous study. Thus, these parameters were obtained by inverse fitting with
where c0 and ce (mg/L) are the aqueous concentrations of the contam­ the help of the Marquardt-Levenberg optimization algorithm package in
inants initially and at equilibrium time, respectively; V (L) is the volume the Hydrus-1D code (Simunek, 2005), proceeded on the premise of
of the detected solution; and m (g) represents the mass of the absorbents. fixing the adsorption parameters and saturated soil conductivity
measured by experiments. The objective function Φ(b) to be minimized
2.2.3. Estimation of hydrodynamic parameters during the parameter optimization process may be defined as:
Percolation experiments were conducted to measure the saturated ∑
n
hydraulic conductivity (Ks). Further detected the breakthrough curves Φ(b) = wi [CO (x, ti ) − CS (x, ti , b) ]2 (10)
(BTCs) of chelate-metal complexes and obtained the hydrodynamic i=1

parameters of λ and Dw via inverse fitting of the BTCs with Hydrus-1D.

where n is the total observation points in a treatment of percolation
Firstly, the undisturbed soil columns were pre-treated and saturated
experiments, CO (x, ti ) represents observed contaminant concentration at
for 24 h (details shown in Supplementary information S3). Then, the
time ti at location x, CS (x, ti , b) is the corresponding simulation value for
saturated soil columns were oriented vertically (Fig. S1) and immedi­
the vector of optimized parameter b (including λ, DW,1, and DW,2), wi is
ately supplied with deionized water by a Mariotte bottle. Collected the
the weight associated with ith observation point and is assumed as 1.0
effluent from the bottom at fixed time intervals, and recorded the vol­
during the parameter optimization.
ume simultaneously until the outflow was constant. Six soil columns
To guarantee the universality of the transport model of chelate-metal
with adjacent Ks values of 0.190 ± 0.010 m/d and 0.762 ± 0.074 m/
complexes, the average λ for the same sub-vadose zone and the average
d that belonged to site Y1 and site Y2, respectively, were employed in
DW for the same contaminants were computed and regulated during
subsequent contaminants percolation experiments.
parameter optimization. We further confirmed the average λ, DW,1, and
Secondly, the solutions of chelate-metal complexes were prepared
DW,2 when the r2 is larger than 0.90 and the RMSE is smaller than 5%
similar to the adsorption experiments, to reach 2.0 mg/L for Cd and 500
(Luciano et al., 2013).
mg/L for Pb. Notably, inflow containing single or binary chelate-metal
Also, sensitivity analysis (SAS) was used to verdict the sensitive pa­
complexes was considered and the corresponding solutions were fed
rameters of the transport model. A single-factor SAS was carried out by
into the soil columns after the above Ks determination. Six treatments
varying one of the parameters and fixing the others (Cho et al., 2016;
involving EDTA-Cd, EDTA-Pb, GLDA-Cd, GLDA-Pb, the combined
Dong et al., 2020). λ and DW were individually varied by 20% in the SAS
EDTA-Cd and EDTA-Pb, and the combined GLDA-Cd and GLDA-Pb, were
because these parameters were obtained via inverse fitting. The SAS also
performed for each site. Once the solution was injected, the effluent was
considered a 20% perturbation in both the saturated hydraulic con­
collected at fixed time intervals, and the volume was recorded simul­
ductivity (Ks) and distribution coefficient (Kf or Kd). Herein we only
taneously until the effluent concentration was approximately equal to
performed SAS in the case of the transport of EDTA-Pb complexes at
that in the feed solution. The effluent was filtered through a 0.45 μm
both two sites, because of the greatest contamination to the groundwater
cellulose acetate membrane for metal content analysis by AAS. Partic­
obtained from the prediction. The optimal parameters before pertur­
ularly, the single treatments including GLDA-Pb of Y1, GLDA-Cd of Y2,
bation were treated as the calibration values, and the sensitivity coef­
and GLDA-Pb of Y2, were respectively interrupted the inflow for 720,
ficient (Sc) was calculated as:
60, and 75 min, to investigate the degradation influence on the fate of
GLDA-metal complexes. E[|Yi − Y0 | ]

(11)
E(Y0 )
Finally, the software package Hydrus-1D (version 4.16.0110) was Sc = Δai
ai

4
L. Liu et al. Ecological Indicators 139 (2022) 108913

where Yi represents the output results after parameter perturbation, Y0 predominantly reflected in the magnitude of adsorption parameters.
represents the output obtained from the calibrated parameter input, E Tables 1 and 2 respectively show the fitted results of three isotherm
averages the values in the parentheses, Δai is the variation of ith models for the single and binary adsorption experiments with the soil
parameter; ai denotes the ith parameter. And the absolute value algo­ from site Y1 and site Y2. For site Y1, the determination coefficient (r2) of
rithm was applied to prevent discrepancies from offsetting. the linear model ranged from 0.9009 to 0.9734, indicating that the
linear model can describe the adsorption of both metals and chelate-
2.2.5. Determination of practical prediction boundary metal complexes well. Except for the adsorption of GLDA-Pb com­
Maize is one of the common tolerant plants used on chelate-assisted plexes, the Freundlich isotherm also gave good fitting results. For the
phytoremediation (Chigbo and Batty, 2015; Rizwan et al., 2017; Yang adsorption of Pb contaminants at site Y2, the Freundlich isotherm model
et al., 2021). In the present study, we chose a general scenario, that was more suitable. Some studies indicated that the linear model cannot
chelate-enhanced phytoremediation using maize is subjected to heavy accurately simulate the transport of pollution when the β of the
rainfall, to further predict and quantify its potential risk on groundwater Freundlich model is less than 0.9 (Brusseau, 1995; Weber et al., 1991).
by the optimal transport model of chelate-metal complexes. The sce­ Thus, the parameters of the Freundlich isotherm here introduced into
nario was reappeared by implementing pot cultivation and leaching the modified CDE model, and other cases of pollution forecasting were
experiments (details shown in Supplementary information S4). The only considered the linear adsorption. The Langmuir isotherm failed to
average heavy metal concentration of leachate was set as the practical describe the adsorption behavior in either soil type because of its lower
upper boundary for prediction, which was 8.083 mg/L for Cd and determination coefficient. This implied that multilayer adsorption of
1258.580 mg/L for Pb in the EDTA treatment, and was 2.403 mg/L for chelate-metal complexes on heterogeneous soil surfaces occurred rather
Cd and 268.912 mg/L for Pb in the GLDA treatment. The contaminant than monolayer adsorption on homogeneous surfaces (Liu et al., 2020a;
pulse duration was 1 h according to the pot leaching experiments. The Zhou et al., 2018).
lower boundary was free drainage and zero concentration gradient. Based on the good fitting of the linear model for both sample sites,
Subsequently, the fate of contaminants at the water tables located 1, 3, the distribution coefficient (Kd) and retardation factor (R) were
and 5 m below the soil surface was forecasted. compared. As Table 1 shows, the Kd and R of the chelate-metal com­
plexes were smaller than those of the metals for both single and binary
adsorption, indicating the metals had stronger adsorption on the soil
2.3. Quality control matrix than the chelate-metal complexes. Studies exhibited that the
anionic chelate-metal complexes should bind to the positively charged
Strict quality control was ensured by using guarantee reagents; sites on the soil particles, while many parts of the soil surface are usually
analyzing samples in triplicate in the physicochemical analysis, negative (Guclu and Apak, 2003). Therefore, the higher mobility of
adsorption experiment, and leaching experiment; and analyzing the chelate-metal complexes may pose a risk of groundwater pollution. For
following certified reference materials: GSB07-1183-2000 (for soluble single adsorption in the Y1 soils, the Kd followed a decreasing order of Kd
Pb), GSB07-1185-2000 (for soluble Cd), and GBW07405 (GSS-5) (for (Pb) > Kd (Cd) > Kd (GLDA-Cd) ~ Kd (GLDA-Pb) > Kd (EDTA-Cd) ~ Kd
soil). The recoveries of Pb from GSB07-1183-2000 and GBW07405 (GSS- (EDTA-Pb). The affinity of Pb was stronger than that of Cd, which has
5) ranged from 98.33% to 102.42% and from 97.64% to 101.45%, been reported previously (Baghenejad et al., 2016; Jiang et al., 2012; Lu
respectively, while the recoveries of Cd from GSB07-1185-2000 and and Xu, 2009), and can be ascribed to the smaller hydrated radius of Pb
GBW07405 (GSS-5) ranged from 97.32% to 102.68% and from 95.56% (0.187 nm) than Cd (0.230 nm) (Yong, 1992). However, for binary
to 102.44%, respectively. The relative standard deviations were both adsorption in the Y1 soils, the Kd of both metals and chelate-metal
less than 3%, indicating an accurate determination of the heavy metals. complexes decreased due to the competition of adsorption sites,
consistent with the studies of Fifi et al. (2013), Fontes and Gomes
3. Results and discussion (2003), and Jiang et al. (2012). Therefore, the coexistence of different
chelate-metal complexes inhibits adsorption and further increases the
3.1. Adsorption behavior and parameters of metals and chelate-metal mobility of chelate-metal complexes.
complexes In the Y2 soils, the results were similar to those at site Y1. The Kd of
EDTA-metal complexes was smaller than that of GLDA-metal complexes,
The migration of chelate-metal complexes and metal ions (the which means that the EDTA-metal complexes were easily transported in
degradation products of GLDA-metal complexes) in the sub-vadose zone the vadose area, especially in the binary adsorption system. Compared
was dependent on their retention on the soil particles, which was with the R values at site Y1, the R values for Pb and Cd at Y2 were 3.1

Table 1
Parameters derived from the fitting of adsorption isotherms for metal and chelate-metal complexes in the soil at site Y1.
Treatment a Linear Model Freundlich Isotherm Langmuir Isotherm
3 2 2
Kd (m /kg) R r β m 3β r r2
Kf ( )
mgβ− 1 *kg

Cd 0.0404 216.296 0.9051 0.7497 0.0433 0.9082 0.2925

EDTA-Cd 0.0026 14.856 0.943 0.4708 0.0056 0.947 0.7033
GLDA-Cd 0.0089 48.429 0.9597 0.9241 0.0122 0.956 0.0217
C-Cd 0.0124 67.081 0.9701 0.8589 0.0199 0.9217 0.0328
CE-Cd 0.0013 7.928 0.936 0.591 0.0019 0.9861 0.8922
CG-Cd 0.0083 45.232 0.9266 0.6378 0.0127 0.981 0.7151
Pb 0.0526 281.311 0.9474 0.5291 0.2221 0.8893 0.4647
EDTA-Pb 0.0024 13.790 0.9385 0.4477 0.0438 0.8892 0.4986
GLDA-Pb 0.0080 43.633 0.9009 0.3503 0.1436 0.6414 0.0357
C-Pb 0.0184 99.056 0.9053 0.7355 0.0838 0.958 0.9028
CE-Pb 0.0006 4.197 0.9437 1.3441 0.0001 0.957 0.1793
CG-Pb 0.0047 26.047 0.9734 0.7766 0.0134 0.951 0.1098

a “C”, “CE” and “CG” denote the mixture. For example, “C-Cd” represents the Cd in the mixture of Cd and Pb, “CE-Cd” represents EDTA-Cd complexes in the EDTA-Cd
complexes and EDTA-Pb complexes mixture, and “CG-Pb” represents the GLDA-Pb complexes in the mixture of GLDA-Cd complexes and GLDA-Pb complexes.

5
L. Liu et al. Ecological Indicators 139 (2022) 108913

Table 2
Parameters derived from the fitting of adsorption isotherms for metal and chelate-metal complexes in the soil at site Y2.
Treatment Linear Model Freundlich Isotherm Langmuir Isotherm
3 2 2
Kd (m /kg) R r β m 3β r r2
Kf ( β− 1 )
mg *kg

Cd 0.2149 1171.105 0.9625 0.5128 0.0579 0.9016 0.9146

EDTA-Cd 0.0003 2.633 0.9781 1.3883 0.0003 0.8977 0.8074
GLDA-Cd 0.0025 14.612 0.9143 0.5378 0.0038 0.9815 0.8346
C-Cd 0.1045 569.990 0.9861 0.7275 0.0421 0.9105 0.9190
CE-Cd 0.0007 4.811 0.9814 1.3648 0.0005 0.8966 0.8302
CG-Cd 0.0025 14.612 0.9539 0.7336 0.0039 0.9665 0.9972
Pb 0.1621 883.615 0.9668 0.5946 0.4008 0.9764 0.9378
EDTA-Pb 0.0015 9.167 0.9482 1.0402 0.0012 0.988 0.4730
GLDA-Pb 0.0085 47.281 0.9036 0.6388 0.0421 0.9444 0.6369
C-Pb 0.0866 472.527 0.8881 0.5124 0.2256 0.9046 0.9997
CE-Pb 0.0010 6.445 0.9275 0.9829 0.0012 0.9696 0.9160
CG-Pb 0.0061 34.214 0.9215 0.6193 0.0466 0.9918 0.7821

times and 5.4 times higher, respectively, due to the higher pH of the Y2 complexes, other BTCs were relatively symmetric and without the
soil (Danila and Vasarevicius, 2017). However, the adsorption behaviors tailing. A 720 min interruption during the injection of GLDA-Pb com­
of Pb, Cd, chelate-Pb complexes, and chelate-Cd complexes were plexes alone led to a temporary decrease of Pb in the effluent concen­
different from those at site Y1. The mono-adsorption selectivity at site tration (Fig. 2b). It can be explained by the different adsorption sites of
Y2 followed an order of Cd > Pb > GLDA-Pb complexes > GLDA-Cd metal ions and chelate-metal complexes, where the metal ions adsorb to
complexes > EDTA-Pb complexes > EDTA-Cd complexes. This may be the negative soil surface by inner complexation, while anionic chelate-
controlled by the properties and interaction of the soils and the chelate- metal complexes are bound through outer complexation and H-
metal complexes, including but not limited to the content of SOM, bonding (Guclu and Apak, 2000). During the 720 min interruption, the
aluminous and ferric hydroxides (oxides), and clay (Azeez et al., 2018; Li GLDA-Pb complexes disintegrated and were released from their
et al., 2013). In a word, the transportability followed an order of EDTA- adsorption sites, which needed to be refilled by new components in the
metal complexes > GLDA-metal complexes > heavy metals whatever the subsequent solution input. Therefore, the degradation of GLDA-metal is
soil type, inferring the pollution risk of GLDA-metals complexes and a non-equilibrium process, which can account for the slight tailing of the
EDTA-metals complexes. BTC of the single GLDA-metal complexes. However, the competitive
adsorption and complemental occupation of adsorbed sites between
3.2. Characteristics of the BTCs for chelate-metal complexes and GLDA-Cd complexes and GLDA-Pb complexes would conceal the
hydrodynamic parameters degradation process in the short term and finally cause no tailing phe­
nomenon in the binary treatment.
Comparison between the detected data and the simulated BTCs The BTCs for site Y2 are shown in Fig. 3 and greatly differ from those
revealed that the simulation fit the data well for both sites Y1 and Y2, for site Y1. The breakthrough sequence of the single and binary systems
which is illustrated in Table S3. And the longitudinal dispersion (λ) at still followed the order of Kd. However, the shape of the BTCs was
the two sample sites and the molecular diffusion coefficient of the asymmetric, where the concentration of contaminants appeared rapidly
chelate-metal complexes (including EDTA-Cd complexes, EDTA-Pb at the initial time and tailing existed. The physical and chemical non-
complexes, GLDA-Cd complexes, and GLDA-Pb complexes) are listed equilibrium may contribute to the asymmetrical BTCs. Gao et al.
in Table 3. Notably, the high r2 (ranging from 0.9632 to 0.9955) and (2009) illustrated that relatively high water flows through a microscopic
relatively low RMSE (ranging from 0.0120 to 0.0472) indicated that area would generate non-equilibrium conditions. Herein, the higher Ks
these parameters derived from the inverse fitting and optimization were for site Y2 may create mobile and stagnant water areas where the
appropriate for forecasting the migration of chelate-metal complexes in dispersion and adsorption are rate-limited reactions (Comegna et al.,
the sub-vadose zone. 2001; Dan et al., 2021; Shi et al., 2019). It is difficult to achieve the
Observed data and the simulated BTCs for the single and binary adsorption equilibrium of the chelate-metal complexes in the initial
treatments at site Y1 are plotted in Fig. 2. In the single treatment, the stage if Ks was high. Thus, the concentration of contaminants sharply
EDTA-metal complexes reached equilibrium earlier than the GLDA- increased in the beginning. Besides, the non-linear adsorption and
metal complexes did. However, the transport of the chelate-Pb com­ chemical non-equilibrium reaction (e.g., rate-limited degradation)
plexes was facilitated and the transport of the chelate-Cd complexes was likely created the tailing (Tsang et al., 2007), which could be deduced
correspondingly slowed in the binary treatment due to competitive from results during the interruption of the single GLDA-Cd complexes
adsorption. These phenomena agreed with the variation in Kd, indi­ and GLDA-Pb complexes treatments (Fig. 3a and b).
cating that adsorption played a dominant role in the migration of
chelate-metal complexes. Except for the BTCs of single GLDA-metal

Table 3
Parameters derived from the BTCs for two sub-vadose zones.
a
Sample site λ (m) DW,1 (m2/d) b DW,2 (m2/d) c DW,3 (m2/d) d

Cd Pb Cd Pb Cd Pb

Y1 0.0112 0.00297 0.00519 0.00724 0.00870 5.41E-05 7.10E-05

Y2 0.0956

a λ represents the average longitudinal dispersion.

b DW,1 represents the average molecular diffusion coefficient for EDTA-metal complexes in solution.
c DW,2 represents the average molecular diffusion coefficient for GLDA-metal complexes in solution.
d DW,3 represents the molecular diffusion coefficient for Cd or Pb in solution, which was according to the research of Buffle et al. (2007).

6
L. Liu et al. Ecological Indicators 139 (2022) 108913

Fig. 2. BTCs of single treatment (a for chelate-Cd complexes and b for chelate-Pb complexes) and binary treatment (c for EDTA-metals complexes and d for GLDA-
metals complexes) in the Y1 soil columns.

3.3. Sensitivity analysis 3.4. Prediction of groundwater pollution risk during chelate-induced
phytoremediation
Various adsorption parameters and hydrodynamic parameters of
actual contaminated areas bring uncertainty on the pollution prediction Generally, heavy rainfall creates the greatest risk of pollution from
of the chelate-metal complexes. However, we can immediately regulate chelate-induced phytoremediation by leaching chelate-metal complexes
the critical parameters of prediction according to the actual situation if into deeper soil horizons, possibly to the water table (Lu et al., 2017).
the influential parameters were ascertained by the single-factor sensi­ Based on the trials and parameters optimization, the practical bound­
tivity analysis. Table S4 shows the results of the SAS of the modified CDE aries are introduced in the optimal transport model and quantifica­
for EDTA-Pb complexes. Although λ and DW,1 were varied by 20%, the tionally assess the potential risk of chelate-metal complexes to the
sensitivity coefficient (Sc) remained almost equal to zero, indicating that shallow groundwater. According to the Standard for Groundwater
these two parameters may have no significant influence on the simula­ Quality of China (GB/T 14848–2017), toxicological indexes of Cd and
tion of chelate-metal complexes. Similar results indicating a low sensi­ Pb higher than 1.0E-02 mg/L and 1.0E-01 mg/L, respectively, are the
tivity of D have been mentioned in the studies by Moradi and thresholds for groundwater pollution (Class V), which regard the water
Mehdinejadiani (2018), and Morsali et al. (2019). Moreover, the 20% is not suitable for drinking. These thresholds were applied as the
variation in Ks and Kd (or Kf) giving rise to a fluctuation in the output pollution signals and to calculate the maximum excess ratio (through the
concentration indicated that convection and adsorption predominantly peak concentration divided by the threshold value).
control the migration of chelate-metal complexes (Fig. S2). The Ks and The variations in the concentration of chelate-metal complexes or
Kd (or Kf) differ based on the properties of the soil and the chelate-metal metal versus transport time are plotted in Figs. 4–6, which show the
complexes, e.g., the differences between site Y1 and site Y2, and would simulated results of the single and binary systems at the water tables of
therefore impact the sensitivity of the transport model and lead to 1 m, 3 m, and 5 m at the site Y2. Furthermore, the relevant indexes,
variability in the prediction of the groundwater pollution risk posed by including the start and end times of pollution, the residence time of
chelate-metal complexes. In the case of EDTA-Pb complexes treatment, contaminant, peak concentration, and maximum excess ratio, are listed
increasing 20% of the Ks decreased the breakthrough time by 18.02% for in Table S5. For a groundwater level of 1 m, the excess ratio for GLDA-
site Y1 and by 16.52% for site Y2, respectively. It is foreseeable that the assisted phytoremediation was 1.31 for Cd and ranged from 14.58 to
higher Ks, the faster the transport of chelate-metal complexes (like 20.03 for Pb. Although Cd pollution was not found at the water tables of
GLDA-Cd complexes and GLDA-Pb complexes). 3 m and 5 m, GLDA still gave rise to Pb pollution in groundwater that
exceeded the standard by a factor of 4.45 to 4.89 at the water table of 3
m, and by a factor of 1.57 to 1.77 at the water table of 5 m, respectively.
In the single system, the concentration of GLDA-Cd complexes and total
Cd (including all forms of Cd) exceeded the pollution threshold and was

7
L. Liu et al. Ecological Indicators 139 (2022) 108913

Fig. 3. BTCs of single treatment (a for chelate-Cd complexes and b for chelate-Pb complexes) and binary treatment (c for EDTA-metals complexes and d for GLDA-
metals complexes) in the Y2 soil columns.

only found at the water table of 1 m, where it remained for approxi­ and from 53.27 to 191.80 for Pb. For the migration of a single
mately 2 d (Fig. 4a and c). The short transport distance of GLDA-Cd contaminant, the EDTA-Cd complexes reached the water table sooner
complexes can be ascribed to the biodegradation process. GLDA-Cd than the EDTA-Pb complexes, owing to the relatively weak affinity be­
complexes will disintegrate into the Cd ions and other products during tween EDTA-Cd complexes and soil particles. However, the residence
downward transport. Although the Cd ions could continuously migrate time of the EDTA-Pb complexes was approximately 4.7 times longer
for a long time due to the continuous degradation of GLDA-Cd com­ than that of the EDTA-Cd complexes. And the excess ratio of the EDTA-
plexes during transport, low concentration of Cd ions released from Pb complexes was higher than the corresponding excess ratio of the
degradation and the higher retention of Cd ions on the soil particles EDTA-Cd complexes. We believe that the initial concentration input into
prevented contamination plume from moving to the deeper water table. the model rather than other potential reasons mainly led to the differ­
Differing from the prediction results of GLDA-Cd complexes alone, ence in the residence time and the excess ratio between EDTA-Pb
either the GLDA-Pb complexes and the total Pb (including the GLDA-Pb complexes and EDTA-Cd complexes. It means that the higher concen­
complexes and Pb ions) exceeded the threshold at the water tables of 1 tration of the EDTA-metal complexes is leached from the topsoil, the
m, 3 m, and 5 m, even though the Kd and R of GLDA-Pb complexes were heavier contamination appears and stays longer in the shallow
higher than that of the GLDA-Cd complexes. The high input concen­ groundwater. Notably, multiple pollutants are more widespread and
tration of the GLDA-Pb complexes could account for these prediction worth more attention. The simulated results for the binary systems
results, which may offset the effect of degradation and adsorption. Be­ showed that the contaminant fronts of EDTA-Cd complexes and EDTA-
sides, the higher complex stability constant of GLDA-Pb complexes Pb complexes were similar, while the duration of pollution caused by
(logK = 11.6) than that of GLDA-Cd complexes (logK = 10.3) indicated EDTA-Pb complexes was longer. The start time of EDTA-Cd complexes
GLDA-Pb complexes could migrate for a longer distance (Begum et al., pollution was delayed, while the transport of EDTA-Pb complexes was
2012). For the cotransport of GLDA-Cd complexes and GLDA-Pb com­ facilitated, which followed the results from the BTCs caused by
plexes, the GLDA-Pb complexes appeared sooner than they did in the competitive adsorption. Therefore, groundwater would contain more
single-contaminant scenario, while the contaminant plume of the GLDA- EDTA-Pb complexes when EDTA-Cd complexes and EDTA-Pb complexes
Cd complex was similar in both the single- and multi-contaminant coexist in soil solution.
migration. These can be related to competitive adsorption. Addition­ For site Y1, the simulated results for the single and binary systems
ally, pollution by the degradation products, i.e., Cd and Pb ions, was are exhibited in Fig. S3 to S5. And the details of the prediction results are
negligible due to the correspondingly excellent adsorption capacity of listed in Table S6. The Pb pollution was only found at the water table of
the soil at site Y2. 1 m during GLDA-assisted phytoremediation at site Y1, for which the
In addition, the prediction showed heavier pollution of EDTA- highest excess ratio was 3.12. And the addition of EDTA gave rise to Cd
induced phytoremediation at the groundwater level of 1 m, 3 m, and and Pb contamination of the groundwater at 1 m, 3 m, and 5 m, where
5 m, where the maximum excess ratio ranged from 7.06 to 30.02 for Cd the maximum excess ratio ranged from 1.58 to 6.40 for Cd and from

8
L. Liu et al. Ecological Indicators 139 (2022) 108913

Fig. 4. Simulated migration behavior of GLDA-Cd complexes (a), Cd (b), total Cd (including all forms of Cd) (c), GLDA-Pb complexes (d), Pb (e), and total Pb (f) at
the groundwater levels of 1 m, 3 m, and 5 m at the site Y2 in the single-contaminant scenario.

24.63 to 176.10 for Pb, respectively. The degree of groundwater (Fig. S3 and S4), while sometimes these products can be ignored because
contamination at site Y2 was heavier, which was mainly attributed to of the very low concentrations. In addition, the concentration of con­
the difference in soil permeability between the two sites. Previous taminants decreased with increasing depth for both sites, inferring the
studies found that increasing water flows in the soil facilitate contami­ strong attenuation ability of subsoil and the degradation of GLDA-metal
nants rapidly migrating downward and contaminants of the effluent complexes. Although the degradation of GLDA-metal complexes was
correspondently increase (Lo et al., 2004; Serrano et al., 2013). In chemical non-equilibrium, an obvious tailing effect was only observed
contrast, small soil permeability would increase chelate-metals com­ on the long-term migration results at site Y2 (Figs. 4–6). It means that
plexes contacting with porous media, so that chelate-metal complexes the contaminants would be sustainably released to the groundwater and
were adsorbed as possible and biodegradable GLDA-metal complexes lead to potential risk due to the physical non-equilibrium process pro­
could disintegrate thoroughly before moving to a deeper soil horizon. duced by high interflow of subsoil (Yang et al., 2022). In a word, the
Thus, the pollution front of the chelate-metal complexes moved 3.3- to transport of chelate-metal complexes and metal ions at the subsoil
46.7-fold slower and the residence time was longer than that at site Y2 differed with the change of hydrodynamic properties, resulting in
for both the single and binary scenarios. And we only need to focus on various scales and degrees of groundwater contamination.
the migration of the degradation products (i.e., Cd and Pb) at the subsoil According to the simulated results discussed above, the EDTA-
with low-permeability, as shown by the prediction curves for site Y1 phytoremediation should be replaced due to its inevitable pollution to

9
L. Liu et al. Ecological Indicators 139 (2022) 108913

Fig. 5. Simulated migration behavior of GLDA-Cd complexes (a), Cd (b), total Cd (including all forms of Cd) (c), GLDA-Pb complexes (d), Pb (e), and total Pb (f) at
the groundwater levels of 1 m, 3 m, and 5 m at the site Y2 in the binary-contaminants scenario.

the shallow groundwater during the rainy season. However, the poten­ while only needs 5 d in the acid soil (Freitas and Nascimento, 2017). The
tial risk on groundwater during GLDA-enhanced phytoremediation de­ life cycle of the biodegradable chelate-metal complexes is always
pends on the adsorption capability and hydraulic properties of subsoil, ignored, so the migration amount of these contaminants and their risk
the dosage of chelate, and the contamination level of the surface soil. on the groundwater are underestimated by only using the leaching ex­
Conducting GLDA-enhanced phytoremediation at the area with readily periments of topsoil on the pot scale. Notably, we combined the
permeable sub-vadose zone and rainy climate, or implementing this contaminant source of the topsoil and the optimal transport model based
technique on the heavy Pb-contaminated area is an improper decision, on the subsoil to provide a complete view on the fate of chelate-metal
due to the large probability of shallow groundwater pollution. These complexes during biodegradable chelate-assisted phytoremediation.
prediction results broke the previous perceptions that the degradation of The risk-assessment framework we established is beneficial to optimize
GLDA-metal complexes can prevent groundwater pollution by heavy the operation condition and manage the risk of chelate-assisted phy­
metals. In addition, the life cycle of biodegradable chelates relates to toremediation, which the CDE contained in our framework has been
their chemical structure and the soil environment. For example, Ethyl­ recommended in the Chinese technical guidelines for environmental
enediamine-N,N’-disuccinic acid (EDDS) and Nitrilotriacetic acid (NTA) impact assessment of soil environment. Although our risk-assessment
are completely decomposed within 57 d in the neutral soil (Wang et al., framework performed well in forecasting the single- or co-migration
2022). The entire degradation of GLDA needs 10 d in the alkaline soil, behavior of chelate-metal complexes in the sub-vadose zone, the rate-

10
L. Liu et al. Ecological Indicators 139 (2022) 108913

Fig. 6. Simulated migration behavior of single EDTA-Cd complexes (a), single EDTA-Pb complexes (b), the mixture of EDTA-Cd complexes (c) and EDTA-Pb (d)
complexes at the groundwater levels of 1 m, 3 m and 5 m at the site Y2.

limited reactions (involving ions exchange, mineral dissolution, and occurred at water tables of 1 m, 3 m, and 5 m. In summary, the degra­
chelate-metal dissolution) should be further contained in the numerical dation of GLDA-metal complexes cannot always prevent heavy metal
model to improve the accuracy. Moreover, soil heterogeneity and pollution in groundwater, which depends on the properties of the soil
randomness during the collection process could significantly affect the horizon, depth of the water table, types of contaminants, and dosage of
parameters of adsorption, dispersion, convection, and degradation and chelates. Thus, this risk-assessment framework is suggested to optimize
subsequently influence the simulation results. Thus, uncertainty analysis the operation condition before using the biodegradable chelate-
of the prediction results is worth considering for a future mathematical enhanced phytoremediation.
model.
CRediT authorship contribution statement
4. Conclusions
Lirong Liu: Conceptualization, Methodology, Formal analysis,
In this study, a new risk-assessment framework, used to predict and Investigation, Writing – original draft, Writing – review & editing.
quantify the potential groundwater pollution during chelate-induced Dinggui Luo: Conceptualization, Methodology, Formal analysis,
phytoremediation, was created by a combination of the practical Writing – original draft, Writing – review & editing, Supervision, Project
boundaries and the numerical model of chelate-metal complexes administration, Funding acquisition. Yayin Lu: Methodology, Data
migrating at the subsoil. We conducted adsorption experiments, perco­ curation. Xuexia Huang: Investigation, Data curation. Yu Liu: Re­
lation experiments, and parameters optimization to derive optimal pa­ sources, Funding acquisition. Lezhang Wei: Resources. Tangfu Xiao:
rameters for Hydrus-1D simulation, and we performed pot cultivation Project administration, Funding acquisition. Qihang Wu: Investigation,
and leaching experiments to obtain the prediction boundary. Adsorption Resources, Funding acquisition. Guowei Liu: Investigation, Data
experiments found that chelate-metal complexes showed weak affinity curation.
of the subsoil compared with metal ions, especially in the binary system,
indicating the great mobility of chelate-metal complexes. Percolation Declaration of Competing Interest
experiments found the degradation of GLDA-metal complexes was a
chemical non-equilibrium process. In addition, the simulated BTCs fitted The authors declare that they have no known competing financial
well with the observed data after parameters optimization, indicating interests or personal relationships that could have appeared to influence
the accuracy of the optimal model. Prediction results showed that heavy the work reported in this paper.
rainfall only leads to Cd pollution (maximum excess ratio of 1.31) at a
groundwater level of 1 m during GLDA-assisted phytoremediation,
while Pb contamination (maximum excess ratio from 1.57 to 20.03)

11
L. Liu et al. Ecological Indicators 139 (2022) 108913

Acknowledgements Gall, J.E., Boyd, R.S., Rajakaruna, N., 2015. Transfer of heavy metals through terrestrial
food webs: a review. Environ. Monit. Assess. 187, 201. https://doi.org/10.1007/
s10661-015-4436-3.
This study was financially supported by the Guangdong Provincial Gao, G., Zhan, H., Feng, S., Huang, G., Mao, X., 2009. Comparison of alternative models
Natural Science Foundation Project (2022A1515010764), the National for simulating anomalous solute transport in a large heterogeneous soil column.
Natural Science Foundation of China (41372248, 41807463, 41807193 J. Hydrol. 377, 391–404. https://doi.org/10.1016/j.jhydrol.2009.08.036.
González, I., Cortes, A., Neaman, A., Rubio, P., 2011. Biodegradable chelate enhances the
and 41301348), the Science Technology Project of Guangzhou phytoextraction of copper by Oenothera picensis grown in copper-contaminated acid
(201607010286) and the NSFC-Guizhou Karst Sciences Research Center soils. Chemosphere 84, 490–496. https://doi.org/10.1016/j.
Project (U1612442). chemosphere.2011.03.015.
Guclu, K., Apak, R., 2000. Modeling of copper(II), cadmium(II), and lead(II) adsorption
on red mud from metal-EDTA mixture solutions. J. Colloid Interface Sci. 228,
Appendix A. Supplementary data 238–252. https://doi.org/10.1006/jcis.2000.6974.
Guclu, K., Apak, R., 2003. Modeling the adsorption of free and heavy metal complex-
bound EDTA onto red mud by a nonelectrostatic surface complexation model.
Supplementary data to this article can be found online at https://doi. J. Colloid Interface Sci. 260, 280–290. https://doi.org/10.1016/s0021-9797(03)
org/10.1016/j.ecolind.2022.108913. 00045-6.
He, Y., Yu, J., Xie, S., Li, P., Zhou, K., He, H., 2020. Enhanced phytoextraction of
cadmium contaminated soil by Trifolium repens with biodegradable chelate GLDA.
References Huanjing kexue 41, 979–985. https://doi.org/10.13227/j.hjkx.201907117.
Huang, X., Li, N., Wu, Q., Long, J., Luo, D., Zhang, P., Yao, Y., Huang, X., Li, D., Lu, Y.,
Ashraf, S., Ali, Q., Zahir, Z.A., Ashraf, S., Asghar, H.N., 2019. Phytoremediation: Liang, J., 2016. Risk assessment and vertical distribution of thallium in paddy soils
Environmentally sustainable way for reclamation of heavy metal polluted soils. and uptake in rice plants irrigated with acid mine drainage. Environ. Sci. Pollut. Res.
Ecotoxicol. Environ. Saf. 174, 714–727. https://doi.org/10.1016/j. 23, 24912–24921. https://doi.org/10.1007/s11356-016-7679-9.
ecoenv.2019.02.068. Huang, X., Luo, D., Zhao, D., Li, N., Xiao, T., Liu, J., Wei, L., Liu, Y., Liu, L., Liu, G., 2019.
Azeez, J.O., Hassan, A.O., Olowoboko, T.B., 2018. Differential Sorption Behavior of Distribution, Source and Risk Assessment of Heavy Metal(oid)s in Water, Sediments,
Cadmium, Lead, Zinc, and Copper in Some Tropical Soils and Their Environmental and Corbicula Fluminea of Xijiang River, China. Int. J. Env. Res. Public Health 16,
Implications. Commun. Soil Sci. Plant Anal. 49, 1707–1718. https://doi.org/ 1823.
10.1080/00103624.2018.1474908. Jacques, D., Simunek, J., Mallants, D., van Genuchten, M.T., 2008. Modelling coupled
Baghenejad, M., Javaheri, F., Moosavi, A.A., 2016. Adsorption isotherms of some heavy water flow, solute transport and geochemical reactions affecting heavy metal
metals under conditions of their competitive adsorption onto highly calcareous soils migration in a podzol soil. Geoderma 145, 449–461. https://doi.org/10.1016/j.
of southern Iran. Arch. Agron. Soil Sci. 62, 1462–1473. https://doi.org/10.1080/ geoderma.2008.01.009.
03650340.2016.1147647. Jiang, H., Li, T., Han, X., Yang, X., He, Z., 2012. Effects of pH and low molecular weight
Begum, Z.A., Rahman, I.M.M., Tate, Y., Egawa, Y., Maki, T., Hasegawa, H., 2012. organic acids on competitive adsorption and desorption of cadmium and lead in
Formation and stability of binary complexes of divalent ecotoxic ions (Ni, Cu, Zn, Cd, paddy soils. Environ. Monit. Assess. 184, 6325–6335. https://doi.org/10.1007/
Pb) with biodegradable aminopolycarboxylate chelants (DL-2-(2-carboxymethyl) s10661-011-2422-y.
nitrilotriacetic acid, GLDA, and 3-hydroxy-2,2’-iminodisuccinic acid, HIDS) in L. Brusseau, M., 1995. The effect of nonlinear sorption on transformation of
aqueous solutions. J. Solution Chem. 41, 1713–1728. https://doi.org/10.1007/ contaminants during transport in porous media. J. Contam. Hydrol. 17, 277–291.
s10953-012-9901-9. https://doi.org/10.1016/0169-7722(94)00041-F.
Behroozi, A., Arora, M., Fletcher, T.D., Western, A.W., 2020. Sorption and transport Lasat, M.M., 2002. Phytoextraction of toxic metals: a review of biological mechanisms.
behavior of zinc in the soil. Implications for stormwater management. Geoderma J. Environ. Qual. 31, 109–120. https://doi.org/10.2134/jeq2002.0109.
367, 114243. https://doi.org/10.1016/j.geoderma.2020.114243. Legates, D.R., McCabe, G.J., 1999. Evaluating the use of “goodness-of-fit” measures in
Buffle, J., Zhang, Z., Startchev, K., 2007. Metal flux and dynamic speciation at (Bio) hydrologic and hydroclimatic model validation. Water Resour. Res. 35 (1), 233–241.
interfaces. part 1: Critical evaluation and compilation of physicochemical Li, T., Jiang, H., Yang, X., He, Z., 2013. Competitive sorption and desorption of cadmium
parameters for complexes with simple Ligands and Fulvic/Humic substances. and lead in paddy soils of eastern China. Environ. Earth Sci. 68, 1599–1607. https://
Environ. Sci. Technol. 41, 7609–7620. https://doi.org/10.1021/es070702p. doi.org/10.1007/s12665-012-1853-2.
Chigbo, C., Batty, L., 2015. Chelate-assisted phytoremediation of Cu-pyrene- Liu, J., Ren, S., Cao, J., Tsang, D.C.W., Beiyuan, J., Peng, Y., Fang, F., She, J., Yin, M.,
contaminated soil using Z. mays. Water Air Soil Pollut. 226, 74. https://doi.org/ Shen, N., Wang, J., 2020a. Highly efficient removal of thallium in wastewater by
10.1007/s11270-014-2277-2. MnFe2O4-biochar composite. J. Hazard. Mater. 401, 123311. https://doi.org/
Cho, E., Arhonditsis, G.B., Khim, J., Chung, S., Heo, T.Y., 2016. Modeling metal-sediment 10.1016/j.jhazmat.2020.123311.
interaction processes: parameter sensitivity assessment and uncertainty analysis. Liu, L., Luo, D., Yao, G., Huang, X., Wei, L., Liu, Y.u., Wu, Q., Mai, X., Liu, G., Xiao, T.,
Environ. Model. Software 80, 159–174. https://doi.org/10.1016/j. 2020b. Comparative activation process of Pb, Cd and Tl using chelating agents from
envsoft.2016.02.026. contaminated red soils. Int. J. Env. Res. Public Health 17 (2), 497.
Comegna, V., Coppola, A., Sommella, A., 2001. Effectiveness of equilibrium and physical Lo, I.M.C., Luk, A.F.T., Yang, X.Y., 2004. Migration of heavy metals in saturated sand and
non-equilibrium approaches for interpreting solute transport through undisturbed bentonite/soil admixture. J. Environ. Eng.-Asce 130, 906–909. https://doi.org/
soil columns. J. Contam. Hydrol. 50, 121–138. https://doi.org/10.1016/s0169-7722 10.1061/(asce)0733-9372(2004)130:8(906).
(01)00100-0. Lu, S.G., Xu, Q.F., 2009. Competitive adsorption of Cd, Cu, Pb and Zn by different soils of
Cooper, E.M., Sims, J.T., Cunningham, S.D., Huang, J.W., Berti, W.R., 1999. Chelate- Eastern China. Environ. Geol. 57, 685–693. https://doi.org/10.1007/s00254-008-
assisted phytoextraction of lead from contaminated soils. J. Environ. Qual. 28, 1347-4.
1709–1719. https://doi.org/10.2134/jeq1999.00472425002800060004x. Lu, Y., Luo, D., Liu, L., Tan, Z., Lai, A., Liu, G., Li, J., Long, J., Huang, X., Chen, Y., 2017.
Dan, Y., Ji, M., Tao, S., Luo, G., Shen, Z., Zhang, Y., Sang, W., 2021. Impact of rice straw Leaching variations of heavy metals in chelator-assisted phytoextraction by Zea
biochar addition on the sorption and leaching of phenylurea herbicides in saturated mays L. exposed to acid rainfall. Environ. Sci. Pollut. Res. 24, 24409–24418. https://
sand column. Sci. Total Environ. 769, 144536 https://doi.org/10.1016/j. doi.org/10.1007/s11356-017-0065-4.
scitotenv.2020.144536. Luciano, A., Viotti, P., Mancini, G., 2012. A numerical model to study pulsing soil
Danila, V., Vasarevicius, S., 2017. Theoretical evaluation of heavy metals migration and flushing: validation and application to a real contaminated soil. In: Merli, C.,
sorption in soil. In: Cygas, D., Vaiskunaite, R. (Eds.), 10th International Conference Pierucci, S., Klemes, J.J. (Eds.), Bosicon 2012: 3rd International Conference on
Environmental Engineering. https://doi.org/10.3846/enviro.2017.015. Contaminated Sites Remediation, pp. 151–156.
Diarra, I., Kotra, K.K., Prasad, S., 2021. Assessment of biodegradable chelating agents in Luciano, A., Viotti, P., Torretta, V., Mancini, G., 2013. Numerical approach to modelling
the phytoextraction of heavy metals from multi–metal contaminated soil. pulse-mode soil flushing on a Pb-contaminated soil. J. Soils Sed. 13, 43–55. https://
Chemosphere 273, 128483. https://doi.org/10.1016/j.chemosphere.2020.128483. doi.org/10.1007/s11368-012-0567-0.
do Nascimento, C.W.A., Amarasiriwardena, D., Xing, B., 2006. Comparison of natural Luo, J., Qi, S., Gu, X.W.S., Wang, J., Xie, X., 2016. An evaluation of EDTA additions for
organic acids and synthetic chelates at enhancing phytoextraction of metals from a improving the phytoremediation efficiency of different plants under various
multi-metal contaminated soil. Environ. Pollut. 140 (1), 114–123. cultivation systems. Ecotoxicology 25, 646–654. https://doi.org/10.1007/s10646-
Dong, J., Chen, Z., Wang, K.e., Han, Y.i., Guo, J., 2020. Examining fate and transport of 016-1623-0.
heavy metal in landfill site through numerical environmental multimedia modeling Luo, J.P., Liang, J.B., Song, Y.C., Guo, X.Y., Ning, Y.Z., Liu, N.L., Zhao, H.P., Li, T.Q.,
approach. J. Environ. Eng. 146 (5), 04020026. 2021. Compounded chelating agent derived from fruit residue extracts effectively
Fifi, U., Winiarski, T., Emmanuel, E., 2013. Assessing the mobility of lead, copper and enhances Cd phytoextraction by Sedum alfredii. Soil Ecol. Lett. 3, 253–265. https://
cadmium in a calcareous soil of Port-au-Prince, Haiti. Int. J. Env. Res. Public Health doi.org/10.1007/s42832-021-0097-z.
10, 5830–5843. https://doi.org/10.3390/ijerph10115830. Madrid, F., Liphadzi, M.S., Kirkham, M.B., 2003. Heavy metal displacement in chelate-
Fontes, M.P.F., Gomes, P.C., 2003. Simultaneous competitive adsorption of heavy metals irrigated soil during phytoremediation. J. Hydrol. 272, 107–119. https://doi.org/
by the mineral matrix of tropical soils. Appl. Geochem. 18, 795–804. https://doi. 10.1016/s0022-1694(02)00258-5.
org/10.1016/s0883-2927(02)00188-9. Mai, X., Luo, D., Wei, L., Liu, Y., Huang, X., Wu, Q., Yao, G., Liu, G., Liu, L., 2019.
Freitas, E.V., Nascimento, C., 2017. Degradability of natural and synthetic chelating Evaluation method for the measuring comprehensive suitability of chelating agents:
agents applied to a lead-contaminated soil. J. Soils Sed. 17, 1272–1278. https://doi. a study of the temporal dynamics of heavy metal activation. Int. J. Phytoremediation
org/10.1007/s11368-015-1350-9. 21, 1415–1422. https://doi.org/10.1080/15226514.2019.1633262.

12
L. Liu et al. Ecological Indicators 139 (2022) 108913

Mayes, M.A., Jardine, P.M., Larsen, I.L., Brooks, S.C., Fendorf, S.E., 2000. Multispecies Van Engelen, D.L., Sharpe-Pedler, R.C., Moorhead, K.K., 2007. Effect of chelating agents
transport of metal-EDTA complexes and chromate through undisturbed columns of and solubility of cadmium complexes on uptake from soil by Brassica juncea.
weathered fractured saprolite. J. Contam. Hydrol. 45, 243–265. https://doi.org/ Chemosphere 68, 401–408. https://doi.org/10.1016/j.chemosphere.2007.01.015.
10.1016/s0169-7722(00)00108-x. Weber, W.J., McGinley, P.M., Katz, L.E., 1991. Sorption phenomena in subsurface
Kedziorek, M.A.M., Dupuy, A., Bourg, A.C.M., Compère, F., 1998. Leaching of Cd and Pb systems: Concepts, models and effects on contaminant fate and transport. Water Res.
from a Polluted Soil during the Percolation of EDTA: Laboratory Column 25 (5), 499–528.
Experiments Modeled with a Non-Equilibrium Solubilization Step. Environ. Sci. Wang, K., Liu, Y., Song, Z., Wang, D.i., Qiu, W., 2019. Chelator complexes enhanced
Technol. 32, 1609–1614. https://doi.org/10.1021/es970708m. Amaranthus hypochondriacus L. phytoremediation efficiency in Cd-contaminated
Moradi, G., Mehdinejadiani, B., 2018. Modelling solute transport in homogeneous and soils. Chemosphere 237, 124480.
heterogeneous porous media using spatial fractional advection-dispersion equation. Wang, X., Fernandes de Souza, M., Li, H., Qiu, J., Ok, Y.S., Meers, E., 2022.
Soil Water Res. 13, 18–28. https://doi.org/10.17221/245/2016-swr. Biodegradation and effects of EDDS and NTA on Zn in soil solutions during
Morsali, S., Babazadeh, H., Shahmohammadi-Kalalagh, S., Sedghi, H., 2019. Simulating phytoextraction by alfalfa in soils with three Zn levels. Chemosphere 292, 133519.
Zn, Cd and Ni transport in disturbed and undisturbed soil columns: comparison of https://doi.org/10.1016/j.chemosphere.2022.133519.
alternative models. Int. J. Environ. Res. 13, 721–734. https://doi.org/10.1007/ Wang, Z., Sun, Z.-J., Mohamed, S., Wang, Z., He, J., Han, L., 2020. DA-6 and GLDA
s41742-019-00212-w. Enhanced Pancium virgatum L. to Phytoextract Cd from Contaminated Soils.
Rizwan, M., Ali, S., Qayyum, M.F., Ok, Y.S., Zia-ur-Rehman, M., Abbas, Z., Hannan, F., Huanjing Kexue 41, 5589–5599. https://doi.org/10.13227/j.hjkx.202004071.
2017. Use of maize (Zea mays L.) for phytomanagement of Cd-contaminated soils: a Wu, L.H., Luo, Y.M., Xing, X.R., Christie, P., 2004. EDTA-enhanced phytoremediation of
critical review. Environ. Geochem. Health 39, 259–277. https://doi.org/10.1007/ heavy metal contaminated soil with Indian mustard and associated potential
s10653-016-9826-0. leaching risk. Agr. Ecosyst. Environ. 102, 307–318. https://doi.org/10.1016/j.
Sarwar, N., Imran, M., Shaheen, M.R., Ishaque, W., Kamran, M.A., Matloob, A., agee.2003.09.002.
Rehim, A., Hussain, S., 2017. Phytoremediation strategies for soils contaminated Yang, B., Qiu, H., Zhang, P., He, E., Xia, B., Liu, Y., Zhao, L., Xu, X., Cao, X., 2022.
with heavy metals: modifications and future perspectives. Chemosphere 171, Modeling and visualizing the transport and retention of cationic and oxyanionic
710–721. https://doi.org/10.1016/j.chemosphere.2016.12.116. metals (Cd and Cr) in saturated soil under various hydrochemical and hydrodynamic
Saso, J.K., Parkin, G.W., Drury, C.F., Lauzon, J.D., Reynolds, W.D., 2012. Chloride conditions. Sci. Total Environ. 812, 151467. https://doi.org/10.1016/j.
leaching in two Ontario soils: measurement and prediction using HYDRUS-1D. Can. scitotenv.2021.151467.
J. Soil Sci. 92, 285–296. https://doi.org/10.4141/cjss2011-046. Yang, Q., Li, Z., Lu, X., Duan, Q., Huang, L., Bi, J., 2018. A review of soil heavy metal
Serrano, S., Vlassopoulos, D., Garrido, F., O’Day, P.A., 2013. A combined site-specific pollution from industrial and agricultural regions in China: pollution and risk
metals sorption and transport model for intact soil columns. Vadose Zone J. 12, assessment. Sci. Total Environ. 642, 690–700. https://doi.org/10.1016/j.
1–11. https://doi.org/10.2136/vzj2012.0152. scitotenv.2018.06.068.
Shi, F., Hao, Z., Liang, Y., Liu, J., Liu, J., 2019. Sorption and transport of aluminum Yang, Q., Yang, C., Yu, H., Zhao, Z., Bai, Z., 2021. The addition of degradable chelating
dialkyl phosphinate flame retardants and their hydrolysates in soils. Environ. Pollut. agents enhances maize phytoremediation efficiency in Cd-contaminated soils.
246, 1–10. https://doi.org/10.1016/j.envpol.2018.11.105. Chemosphere 269, 129373. https://doi.org/10.1016/j.chemosphere.2020.129373.
Simunek, J., 2005. The HYDRUS 1D Software Package for Simulating the One- Ye, T., Li, H., Wang, Z.-X., Huang, R., Yu, Y.-J., Yang, Z., Gao, C., Xie, C., 2019. Transport
dimensional Movement of Water, Heat, and Multiple Solutes in Variably Saturated and fate of hexavalent chromium in slag-soil system. Environ. Earth Sci. 78 https://
Porous Media, Hydrus Software. doi.org/10.1007/s12665-019-8245-9.
Simunek, J., van Genuchten, M.T., Sejna, M., 2008. Development and applications of the Yong, R.N., 1992. Principles of Contaminant Transport in Soils. Elsevier, Amsterdam,
HYDRUS and STANMOD software packages and related codes. Vadose Zone J. 7, The Netherlands.
587–600. https://doi.org/10.2136/vzj2007.0077. Yu, H., Zhan, J., Zhang, Q., Huang, H., Zhang, X., Wang, Y., Li, T., 2020. NTA-enhanced
Tandy, S., Bossart, K., Mueller, R., Ritschel, J., Hauser, L., Schulin, R., Nowack, B., 2004. Pb remediation efficiency by the phytostabilizer Athyrium wardii (Hook.) and
Extraction of heavy metals from soils using biodegradable chelating agents. Environ. associated Pb leaching risk. Chemosphere 246, 125815. https://doi.org/10.1016/j.
Sci. Technol. 38, 937–944. https://doi.org/10.1021/es0348750. chemosphere.2020.125815.
Tsang, D.C.W., Zhang, W., Lo, I.M.C., 2007. Modeling cadmium transport in soils using Zhou, Q., Liao, B., Lin, L., Qiu, W., Song, Z., 2018. Adsorption of Cu(II) and Cd(II) from
sequential extraction, batch, and miscible displacement experiments. Soil Sci. Soc. aqueous solutions by ferromanganese binary oxide–biochar composites. Sci. Total
Am. J. 71, 674–681. https://doi.org/10.2136/sssaj2006.0299. Environ. 615, 115–122. https://doi.org/10.1016/j.scitotenv.2017.09.220.

13