Effects of EDTA on adsorption of Cd(II) and Pb(II) by low-permeability soil minerals and on their microscopic characteristics CURRENT STATUS: POSTED

Ethylenediaminetetraacetic acid (EDTA) can serve as a washing agent in the remediation of low-permeability layers contaminated by heavy metals (HMs). Therefore, batch adsorption experiments, where quartz sands (SM1) and mineral mixtures (SM2) were used as low-permeability soil minerals (SMs), were implemented to explore the effects of different EDTA concentrations, pH and exogenous chemicals on the HM-SM-EDTA adsorption system. Changes in microscopic characterizes of SMs were determined by instrument analysis to investigate the mechanisms.

insight from the presented study can help to promote the remediation of contaminated soil and groundwater.

Background
Sanitary wastewater, agricultural and industrial discharges from increasing human activities has made heavy metal pollution become a ubiquitous environmental problem, especially in soil and groundwater system [1][2][3][4]. Enriched heavy metals (HMs) in the environment seriously threaten the ecological system and human health due to high toxicity, the nonbiodegradable nature and the ability of retention [4][5][6]. Toxic metals such as lead (Pb(II)), chromium (Cr(VI)), nickel (Ni(II)) and cadmium (Cd(II)) can cause many severe health effects including bone fractures, organ disorders and brain damage [7]. Specifically, lead can be toxic to lives even at low concentrations [8]; cadmium can increase the risk of cancer [3,[9][10]. Therefore, many water treatment technologies, such as membrane filtration [11], precipitation [12], electrodialysis [13] and so on, are developed to try to completely remove HMs from wastewaters before discharging.
However, there are usually some low-permeability layers in the soil and groundwater environment, which can act as principal obstacles to retard pollutant migration and serve as pollution sinks to store pollutants due to their strong adsorption capability and special diffusion properties [14][15][16][17]. On the one hand, because of their low hydraulic conductivity, water flow and solute transport in lowpermeability layers are slow [18]. On the other hand, low-permeability layers mainly consist of lowpermeability soil minerals such as silt and clay [19]; note that these minerals have great affinity and selectivity to HMs, especially clay [20][21][22][23]. As a result, a large quantity of HMs from wastewater has accumulated in the soil and groundwater system since long time ago, which enhances the hazards of heavy metal pollution and is also quite difficult to restore. Specifically, low-permeability layers serve as long-term reservoirs for pollutants, simultaneously, it also become a secondary pollutant source and constantly release pollution by back-diffusion [16,18,24]; this would greatly extend the timeframe and difficulty of remediation.
Naturally, many technologies have been put forward to restore the low-permeability layers contaminated by HMs, which are mainly divided into physical, chemical, and biological methods. And soil washing combined with pump-and-treat remediation is relatively effective and economic [1].
Considering the strong adsorption capability of low-permeability soil minerals for HMs, ethylenediaminetetraacetic acid (EDTA), a chelating agent, are commonly serve as a washing agent [25][26]; because it can form strong negatively charged complexes with divalent metal cations in solution through its two amines and four carboxylate groups, improve the mobility of HMs, and reduce HM adsorption on low-permeability soil minerals [27][28][29]. Consequently, the amount of synthetic EDTA in the environment is constantly increasing, such as an average annual rate of 8% in China [25].
Necessarily, this study tried to determine the reasonable condition for EDTA to reducing the HM adsorption on low-permeability soil minerals, and investigate the effects of EDTA on the minerals; so that the remediation technology can be refined and promoted.
Within the range of relevant natural conditions, the process of Cd and Pb adsorption onto soil minerals (SMs) was simulated by batch adsorption experiments. The objectives of this study were as follow: (a) to investigate the effects of EDTA on adsorption of Cd and Pb by low-permeability soil minerals, analyze the adsorption isotherm and kinetics, and clarify the basic behavior of the HM-SM-EDTA system; (b) to examine the influences of contact time, pH and exogenous chemicals (low-molecularweight organic acids (LMWOAs) and exogenous metal ions) on the HM-SM-EDTA system, and determine the reasonable conditions for EDTA to work effectively and economically; and (c) to analyze changes in microscopic characteristics of soil minerals under the effects of EDTA, and study the mechanisms.

Materials And Methods 2.1 Selection of soil minerals and chemicals
To simulate commonly media such as silt and clay in low-permeability layers [18,24,30], two types of soil minerals (SMs), fine quartz sands (SM1) and the mineral mixtures of quartz sands and clay (SM2), were typically used in the experiments; and they both had a medium grain size of 50 µm and represented unreactive and reactive minerals for contrast. Fine quartz sands served as the control group to investigate the basic adsorption phenomenon. To cover typical clay minerals in natural lowpermeability layers, SM2 were comprised of 50-µm fine quartz sand, illite and montmorillonite (Junhong New Material Co., Ltd., Shanxi, China) at a mass ratio of 1:1:1 [31][32].
All of the chemicals required for the experiments were of analytical grade or higher and used without further purification. Ethylenediaminetetraacetic acid disodium salt (EDTA-2Na), cadmium chloride (CdCl 2 , 99%), and lead chloride (PbCl 2 , 99.5%) were purchased from Sinopharm Chemical Reagent

Batch adsorption experiments
The effects of EDTA-2Na on heavy metal (HM) adsorption by soil minerals were systematically studied by utilizing Cd(II) and Pb(II) as analytes; batch adsorption experiments were carried out on an incubator shaker at 250 rpm in 50-mL tubes containing 2.4 g soil minerals and 25 mL solution of heavy metal ions, which was the HM-SM system. To explore the effects of EDTA, different doses of EDTA-2Na were added into the above system; and this formed the HM-SM-EDTA system. To investigate the isotherms model, the adsorption experiments were performed at Cd and Pb concentration vary from 10 to 400 mg/L and 10 to 2000 mg/L, respectively. For the kinetic study, the samples were withdrawn at fixed intervals (2,5,10,20,30,60,100,180, and 360 min). The effect of pH values (2.0-12.0) on the HM-SM-EDTA system was investigated by adjusting the initial pH of the solution mixture by either HCl or NaOH solutions (1.0 mol/L). Low-molecular-weight organic acids and exogenous metal ions were all introduced into this system with a concentration of 0.001 M to explore their effects. The concentrations of Cd, Pb, and EDTA-2Na, pH, contact time, and temperature in the batch experiments were 200, 300, and 400 mg/L, 5.5, 360 min, and 25 ℃, except where the concentration, pH, contact time, and temperature were controlling factors as described above. After adsorption, soil minerals were separated by centrifugation at 3000 rpm for 10 min; the supernatants were collected and filtered through 0.22-µm filter membranes. Then, the heavy metal concentrations were analyzed by an inductively coupled plasma optical emission spectrometer (720ES, Agilent Technologies, Inc., USA); and the soil minerals were prepared by vacuum freeze drying for use in analytical instrument tests. Three replicates were conducted for each adsorption experiment and the mean values were reported. Adsorption capacities (q t , mg/g) were calculated as the following Eq. (1): where C 0 (mg/L) is the initial concentration of heavy metal ions, C t (mg/L) is the corresponding concentration at different time t, V (L) is the volume of the solution, and m (g) represents the weight of soil minerals.

Adsorption isotherms and kinetics analysis
To get deeper insight into the adsorption mechanism of the effects of EDTA-2Na on the interactive behavior between soil minerals and heavy metals, the Langmuir and Freundlich isotherm models were examined to simulate the adsorption equilibrium data acquired at various concentration ranges. The Langmuir isotherm model assumes that the monolayer adsorption takes place on adsorbents that have structurally homogeneous surfaces, on which the adsorption sites are uniform and no interaction occurs between adsorbates [33]. The Freundlich isotherm model considers that multilayer adsorption occurs on energetically heterogeneous surfaces, on which the adsorbed molecules are interactive.
The linear forms of the two isotherm models are given in the following Eqs. (2) and (3) [33][34]: where q e (mg/g) and C e (mg/L) represent the adsorption capacity and the concentration of heavy metal ions in the solution at equilibrium; q m (mg/g) is the maximum adsorption capacity; K L (L/mg) is the constant of the Langmuir model related to the affinity of adsorbent binding sites; K F (L/mg) and n are Freundlich isotherm constants related to adsorption capacity and adsorption strength of the adsorbent, respectively. Furthermore, to explore the rate and rate-controlling step of heavy metal adsorption on soil minerals, the experimental data were dealt with four different kinetic models. The pseudo-first order kinetic model tents to physical interactions and assumes that the adsorption process depends on the diffusion step; the pseudo-second order kinetic model tends to chemisorption rate-limiting step in the adsorption process which is related to the condition of active sites, and the process is controlled by chemisorption mechanisms such as electron sharing or transfer among the adsorbent; the intraparticle diffusion model is usually used to explain the kinetics of the diffusion process of substances inside a particle, and assumes that the adsorbate diffusion rate is a rate-limiting step; and the Elovich model describes multilayer adsorption on the heterogeneous surface. Four kinetic models are expressed as following Eqs. (4), (5), (6) and (7) [4,[33][34]: where q t (mg/g) represents the amount of metal ions adsorbed on the adsorbents at time t (min); k 1 (min − 1 ), k 2 (g/mg·min) and k i (mg/g·min 0.5 ) are the rate constants of the pseudo-first order, pseudosecond order, and intraparticle diffusion models, respectively; C (mg/g) is a constant about the boundary layer thickness; α (mg/g·min) is the initial adsorption rate; and β (g/mg) is a constant related to activation energy for chemisorption and the degree of surface coverage.

Soil mineral characterization
To investigate the changes in microscopic characteristics of soil minerals under the effects of EDTA-2Na, original soil samples were reserved before the batch adsorption experiments; and minerals were performed by vacuum freeze drying after centrifugation for use in analytical instrument tests.
The changes in crystal structure of soil minerals were determined by X-ray diffraction analysis (XRD).
Specifically, the dry samples were characterized by utilizing a Bruker D8 Advanced X-ray diffractometer (Bruker Co., Germany) employing Cu Kα radiation (40 kV, 40 mA). Powdered samples were scanned in the range from 10° to 80°, with a scanning step of 0.1 s/step.
To explore the changes in functional groups of soil minerals, the fourier transform infrared (FTIR) spectra of 1% mineral samples compressed by the KBr disk method were obtained on an infrared spectrometer (Nicolet iS50, Thermo Fisher Scientific, USA) from 400 to 4000 cm − 1 at a resolution of were implemented by using SM1 and Cd, the adsorption capacity was reduced and the mobility of Cd was improved with the EDTA concentration increasing (Fig. 1a). Specifically, as the EDTA concentration increased from 0 to 600 mg/L, the maximum q e greatly decreased from approximately 135 to 55 mg/kg due to the chelation between EDTA and Cd ions; note that the chelation was stronger than the physical adsorption of SM1, and the formed negatively charged complexes [Cd(II)EDTA 2− ] could not be bound by negatively charged SM surface [19,27,35]. However, some irregular points where q e rose unexpectedly occurred as the EDTA concentration increased (Fig. 1a). When more EDTA-2Na was added, the mole ratio (MR = moles of HM ions / sum of moles of HM ions and EDTA) increased. The matching ratio between divalent metal ions and EDTA in complexation process was 1:1 (MR = 0.5) [36]; and thus, MR less than 0.5 meant that EDTA was excessive, while MR greater than 0.5 meant that EDTA was insufficient. As Fig. 1a showed, all irregular points appeared when MR was below 0.5, because there might be mutual suppression among excess EDTA; and MR closed to 0.5 would make HM adsorption capacity on SMs decrease more efficiently. Namely, to restore environmental pollution of 50 mg/L Cd, it would be better to choose EDTA-2Na solution with a concentration of 200 mg/L, instead of 50, 400, or 600 mg/L; because the MR here was 0.45, closer to 0.5, and this remediation method was efficient and economic, even though using 400 mg/L EDTA-2Na reduced q e by extra 4 mg/kg (Fig. 1a).
Compared with quartz sands (SM1), mineral mixtures (SM2) had a stronger adsorption capacity for HMs because they can form HM-mineral complexes via functional groups of clay [19,37], which means a larger value range ( Fig. 1a and b). Note that the HM-mineral complexes were still not as strong as [HM(II)EDTA 2− ] complexes [1]. Therefore, the general trend of changes in q e under the effects of EDTA with different concentration can be better presented. When the batch experiments were carried out by using SM2 and Cd, it could be more clearly observed in Fig. 1b that q e increased rapidly when MR was greater than 0.5 (curves above the red dotted lines); this indicated poor efficiency and insufficient EDTA-2Na dosage. Concretely, as the curves of 600 mg/L EDTA showed, q e merely increased by 417 mg/L (from 33 to 450 mg/L) when Cd concentration rose from 25 to much more rapidly. This phenomenon was also observed in the curves of 200 and 400 mg/L EDTA, which illustrated that EDTA could reduce the adsorption capacity greatly when the dosage was sufficient. Additionally, when MR was much less than 0.5 (curves below the blue dotted lines), the gaps among these curves were small; and this indicated a poor efficiency and economic with unnecessary increase in EDTA dosage.
In terms of the adsorption capacity of SMs for Pb, it was much stronger than that for Cd beacuse Pb can compete favorably for the sorption sites in soil minerals [23]. The maximum q e of SM1 and SM2 for Pb were 660 and 19,677 mg/L, respectively ( Fig. 2a and b); while that for Cd were 135 and 2,660 mg/L. Under the effects of EDTA-2Na, they decreased to 306, 19,262, 55 and 1,453 mg/L, respectively; this reconfirmed that EDTA could reduce the adsorption capacity of SMs and improve the mobility of HMs. As Fig. 2a showed, when the MR was much less than 0.5 (curves below the blue dotted lines), differences among these curves were small; when the MR was much greater than 0.5 (curves above the red dotted lines), q e largely increased. This was in accord with the effects of EDTA with different concentrations on the adsorption capacity for Cd, and it might be generalized to most divalent metal ions. Obviously, the gap between curves of 0 and 400 mg/L EDTA was larger when MR of the point in curve of 400 mg/L was closer to 0.5 (black dotted line ② and ③ were longer than ① and ④, Fig. 2a). This phenomenon could also be found in the comparison between the curves of EDTA concentration equal to 0 and others. Similarly, Fig. 2b supported this law, for example, black dotted line ⑤ was longer than ⑥. This was because MR closer to 0.5 reduced the adsorption capacity better, making the function of EDTA-2Na more effective. Naturally, to optimally investigate the effects of other factors, such as contact time, initial pH, LMWOAs, and exogenous metal ions, on the HM-SM-EDTA system, the concentration of Cd, Pb, and EDTA-2Na were fixed as 200, 300, and 400 mg/L because it has been confirmed that this setup could make the system have a reasonable adsorption capacity and make MR close to 0.5. Additionally, mineral mixtures (SM2) had a stronger adsorption capacity for heavy metals [19], which means a larger value range and can better show the law of effects of other factors on this system; and it is more meaningful to explore the effects on mineral mixtures which are reactive and representative. Therefore, SM2 will be focused on after session 3.1.

Isothermal adsorption
The obtained correlation coefficients for Langmuir and Freundlich isotherm models were listed in Table 1. It can be concluded that the Langmuir model fits better with the experimental adsorption equilibrium data of both SM1 and SM2 when there was no EDTA-2Na in the system, because R 2 of the Langmuir isotherm model (> 0.97) was greater than that of the Freundlich isotherm model (> 0.90).
Hence, the adsorption behaviors of both SM1 and SM2 for Cd and Pb ions suggested a monomolecular layer adsorption [4], and the surfaces of quartz sands and mineral mixtures were structurally homogenous; there was no interaction taking place between Cd or Pb ions. By contrast, with the effects of EDTA, correlation coefficients of both isotherm models for HM adsorption on SMs mostly decreased; and it seems that the experimental data was not well in line with both isotherm models.
This indicated that the introduce of EDTA-2Na destroyed the uniformity of the surfaces of mineral particles.
However, compared with the Langmuir isotherm model, the adsorption capacity (q e ) calculated from the Freundlich isotherm model were closer to the experiment value, which can be found in Fig. 3; and its correlation coefficient were also better, especially, when the MR of more points in a certain line was closer to 0.5. Specifically, when EDTA concentrations were 400 and 600 mg/L, R 2 of the Freundlich isotherm model for Cd adsorption on SM2 were even better than R 2 of non-EDTA system (0.98 and 0.94, > 0.91); Good correlation coefficients could also be found in R 2 of the Freundlich isotherm model for Pb adsorption on SM1, when EDTA concentrations were 150 mg/L (MR of more points were closer to 0.5). Therefore, under the effects of EDTA, the adsorption behaviors of both SM1 and SM2 for Cd and Pb conformed to the Langmuir isotherm model and it should be considered as a multilayer adsorption which occurred on energetically heterogeneous surfaces. In detail, on the one hand, some EDTA introduced into the HM-SM system could be adsorbed on and cover the surfaces of SMs [33,38]; and this made the surfaces of SMs heterogeneous, blocking some adsorption sites and making them nonuniform [16,27]. On the other hand, EDTA and HM ions were both adsorbates for SMs, and they were interactive via chelation [26,28,[39][40]; and this was accord with the mechanism of the Freundlich isotherm model. Additionally, because the matching ratio between divalent metal ions and EDTA in chelation process was 1:1 (MR = 0.5), which meant a more effective and stable HM-SM-EDTA system, the linear relationship was better when the MR of more points in a certain line was closer to 0.5.

Adsorption Kinetics
The effects of contact time on HM adsorption by SM2 were shown in Fig. 4. In the system without EDTA, it can be observed that adsorption was fast initially, then decelerated, and finally reached equilibrium (Fig. 4a). Pb adsorption onto SM2 was more rapidly than Cd, and it reached equilibrium earlier, having a higher adsorption capacity. This is because SM2 has a better affinity and selectivity towards Pb [23]. By contrast, in the system with EDTA, obvious turning points occurred in the curves (Fig. 4b). As far as Pb, faster adsorption rate made it reach a high q t value closed to the maximum adsorption capacity, which indicated that the effect of EDTA appeared slower at the beginning.
However, an abrupt turning at approximately 5 min made q t rapidly decrease from 2.85 to 1.03 mg/g in approximately 30 min, and finally decrease to an equilibrium of 0.29 mg/g. For Cd, the turning point appeared at approximately 60 min, which was much later than Pb; and the range of the change in q t were smaller, reaching 0.63 mg/g finally. Therefore, the contact time of 360 min was sufficient to reach equilibrium for both Cd and Pb adsorption onto SM2, and it took 5 to 10 min for EDTA to begin taking its effect. It can also be concluded that EDTA affected Pb adsorption mainly by rapidly desorbing ions after they had been adsorbed on SM, via forming very stable negatively charged complexes [Pb(II)EDTA 2− ] which could not be bound by negatively charged SM surface [19,27,35]; while EDTA might hinder the process of Cd adsorption onto SM, via blocking the adsorption sites and chelating Cd ions simultaneously [16,27].
Furthermore, the adsorption parameters calculated by fitting the four kinetic models are listed in Table 2. It can be observed that with and without EDTA in the system greatly affected the fitting correlation coefficients of these models for different HM ions. Nevertheless, as the Fig. 5 showed, the experimental adsorption data more in line with the pseudo-second order kinetic model than other three kinetic models; and the correlation coefficient (R 2 ) values of the pseudo-second order kinetic model were all greater than 0.96 (without EDTA, R 2 > 0.99; with EDTA, R 2 > 0.96). Additionally, the calculated value of q e from the pseudo-second order kinetic model was closer to the adsorption capacity obtained from experiments than other models. Namely, the behaviors of HM adsorption onto SMs with and without the effects of EDTA were both conform to the pseudo-second order kinetic model, and the adsorption sites on SMs might be the rate-limiting step [4]. When EDTA was added, it would block the adsorption sites on SM surfaces and affect the electron transfer between adsorbents, so that the adsorption process was controlled; this still related to chemisorption rate-limiting step and obeyed the pseudo-second order kinetic model. However, R 2 decreased in the complicated HM-SM-EDTA system, because adsorption of SMs and chelation of EDTA for HMs both existed, which increased the uncertainty.

Effects of initial pH on the system
Solution pH is another important operational parameter determining the efficiency of reducing the adsorption capacity of SMs and improving the mobility of HMs by using EDTA, so that more HM pollution can be removed from soil and groundwater system. Significantly, solution pH can affect both the speciation of HM ions and the protonation of the functional groups of SMs [41]. Generally, as pH increased, the adsorption capacity of both Cd and Pb onto SMs presented a significant increasing trend (Fig. 6). As far as Cd ions, the adsorption capacity of SM1 and SM2 was much smaller at pH < 5.5 and 4.0, respectively ( Fig. 6a and b). This might be attributed to a competitive adsorption between H + and Cd 2+ ions in the system and the protonation of oxide-type functional groups (> SiOH, >AlOH, >Al 2 OH, >AlSiOH) on SM2 [20,33,42]; and the low pH had a weaker effect on SM2 than SM1 because SM2 had a stronger initial adsorption capacity and more active sites for H + and Cd 2+ ions to compete [21]. In terms of Pb, though the adsorption capacity was also smaller at low pH, it was much higher than Cd in the range of 4.0-7.0; this was because the potential selectivity of SMs for Pb ions was preferential [33].
Additionally, the adsorption capacity greatly increased at pH > 7.0 due to the formation of metal hydroxide precipitates for Cd and Pb ions, and these precipitates cannot combine with or be break by EDTA [43]. Cd ions began to speciate to CdOH + , Cd(OH) 2 and Cd(OH 3 ) − later than Pb, so adsorption capacity for Cd started to obviously increased at a higher pH than Pb.
Furthermore, it was observed that the adsorption capacity of Cd and Pb maintained a low level at pH below 7.0 and 4.0, respectively; and then increased sharply (Fig. 6). This indicated that EDTA had strong chelating ability for Cd and Pb ions, and could improve mobility of HMs and reduce adsorption capacity of SMs at a relatively wide pH range; while this range for Cd was wider. Notably, the optimal pH for EDTA to take effects in Cd-SM1, Pb-SM1, Cd-SM2, and Pb-SM2 system was around 4.0, 4.0, 4.0 and 3.0, respectively.

Effects of exogenous chemicals on the system
To simulate the effects of exogenous chemicals which commonly exist in the natural environment on the system, low-molecular-weight organic acids (LMWOAs) and exogenous metal ions served as the important factors. Lowmolecular-weight organic acids (LMWOAs) are hydrophilic acids with some special characteristics, such as reducibility, COOH/OH-richness, and complexation with metal ions [44][45][46][47]; besides, it has some resemblances with EDTA and is abundant in the environment. Therefore, it is significant to investigate effects of LMWOAs on the system, when it works with EDTA (humic acid has similar characteristics and was involved). As Fig. 7 showed, in the non-EDTA system, the adsorption capacity of both Cd and Pb onto SM2 mostly changes little under the effects of LMWOAs except citric and humic acids. Because most LMWOAs did not have a strong chelating ability like EDTA to desorb metal ions, and cannot reduce the adsorption capacity independently. As far as HM-SM-EDTA system, the extent of changes in adsorption capacity depended on the types of LMWOAs and HMs (Fig. 7). For Cd ions, there was decrease in the adsorption capacity under the effects of acetic, crtric, malonic, succinic and tartaric acids, while there was increase under the effects of ascorbic, humic and oxalic acids; however, the changes were all not obvious because the adsorption performance of Cd itself is not as good as Pb that can compete favorably for the sorption sites in oxides and the clay fraction [23]. Therefore, changes in the adsorption capacity of Pb were analyzed to explore the mechanism of the effects of LMWOAs.
It can be obviously observed that tartaric and acetic acids made the adsorption capacity greatly increase from carboxyl groups and one hydroxyl group; acetic acid contains a carboxyl group. They could bridge Pb ions and SM2 surface sites via these functional groups by forming the bridging (LMWOA-Pb-SM2), so that some Pb ions avoided chelating with EDTA and were adsorbed onto SM2; tartaric acid had more functional groups and larger molecular weight than acetic acid, resulting in better bridge ability [45,48]. Additionally, complexation of COOH/OH-rich LMWOAs and surface hydroxyls on SM2 via ligand exchange reactions could increase negative charges on SM2 surfaces, which made SM2 have a stronger ability to adsorb Pb ions [47,[49][50].
Furthermore, the results showed that ascorbic, malonic, oxalic and succinic acid reduced the Pb adsorption capacity of SM2 from 924 mg/kg to 584, 752, 769 and 547 mg/kg. Here are the mechanisms. Firstly, ascorbic and oxalic acids have reducibility, and the combination of them with EDTA probably reduced the adsorption by improving the reductive dissolution of oxide-type functional groups on SM2 surface, which could strongly bind Pb ions [46,51]. Secondly, oxalic, succinic and malonic acids could form complexes with Pb ions via their carboxyl and hydroxyl groups, performing effects similar to EDTA and promoting adsorption reduction. Thirdly, these LMWOAs all had a relatively high adsorption affinity towards SM2 by the interaction among their COOH/OH groups and oxide-type groups of SM2 (> SiOH, >AlOH), resulting in the LMWOA adsorption onto SM2, like the effects of EDTA; this performance would compete for adsorption sites with Pb ions, enhance the mobility of Pb ions, and reduce the adsorption capacity of SM2 [44,49,52]. In summary, LMWOAs affected the system mainly by bridging, complexation, adsorption site competition and reductive dissolution; and the extent might depend on the synthesis of diverse effects.
Heavy metal ions usually coexisted with other metal ions in the soil and groundwater environment, so it is significant to investigate the effects of exogenous metal ions on the HM-SM-EDTA system. As Fig. 8 showed, there were nearly no change in the system without EDTA, which illustrated that it was difficult for exogenous metal ions to greatly affect the Cd and Pb adsorption capacity of SM2 before it reached adsorption saturation; and SM2 had a relatively strong ability of adsorption, so the competition between metal ions for active sites on SM2 could ignored. By contrast, in the HM-SM-EDTA system, there was increase in the Cd and Pb adsorption capacity onto SM2. Specifically, Cu 2+ , Fe 2+ and Pb 2+ ions increased the Cd adsorption capacity following the sequence: Cu > Pb > Fe; and that of Pb was Cu > Cd > Fe. Note that EDTA had different selectivity for different metal ions, and there was competition between cations for complexation with EDTA [46,53]. The uptake selectivity here was Cu > Pb (Cd) > Fe, resulting in the effects on the mobility of the target HMs and the adsorption capacity of SMs.
Additionally, it could be seen that Al 3+ and Cr 3+ ions affected the system little, because EDTA generally had a stronger complexation with divalent metal ions than trivalent metal ions.

XRD analysis
The changes in crystal structure of soil minerals determined by XRD analysis were presented in Fig. 10. The peak intensity generally decreased under the effect of EDTA, suggesting that EDTA molecules were attached on the surface of mineral particles. A peak at approximately 2θ = 66° disappeared under the effect of EDTA, maybe because the dissolution of Metal-Si oxides [46], which was also confirmed by FTIR analysis.

Morphology of the soil minerals
Morphology of the SM2 samples was observed by the SEM to investigate the apparent changes of soil minerals. Figure 11a presented the micrograph of mineral mixtures without the effects of EDTA. Note that the wettability and swelling of clay (illite and montmorillonite) in the mineral mixture would occur when it interacted with water [22]. Besides a large particle, there were many microparticles in Fig. 11a1 and a2; obviously, many small cracks, concavities and convexities appeared on the surface of some particles, which made mineral particles rougher (Fig. 11a3). These were all attributed to the clay swelling and dispersion after being soaked in water (solution without EDTA) [43]. It could be clearly observed from Fig. 11b that mineral particles changed a lot under the effects of EDTA, during the batch adsorption experiments. Firstly, many tiny white substances appeared on the soil minerals ( Fig. 11b1 and b2), which might be the attached EDTA. Secondly, there were many microparticles adhering to large particles due to the enhanced adhesion of minerals under the effects of EDTA ( Fig. 11b2 and   b3), which filled many pores; whereas the surfaces of particles in Fig. 11a1 and a3 were clean. Thirdly, particles agglomerated (Fig. 11b4), and connected with each other by something like fiber bands because of the attachment of EDTA (Fig. 11b5); while particles were independent in Fig. 11a2. Additionally, it could be seen that EDTA also made particles smoother, clogging the active sites on mineral surfaces and reducing its adsorption for HMs [27]. This is in accord with the observation reported by a previous study (Fig. 11b1) [16]. Overall, all changes including attachment of EDTA and microparticles, agglomeration, connection and the smoother surfaces contributed to reducing specific surface areas and adsorption capability of SM2. Table 3 showed the changes in the BET specific surface areas (SSA), pore size and the pore volume of SM2 under the effects of EDTA. Concretely, as EDTA concentration increased from 0 to 600 mg/L, the SSA gradually decreased from 16.73 to 12.59 m 2 /g, and the micropore area decreased from 2.03 to 0.09 m 2 /g. It decreased more at the low EDTA concentration (50 and 200 mg/L), while external surface area decreased more at the high EDTA concentration (400 and 600 mg/L); this indicated that EDTA might block the micropore area and made the surface smoother first, then played a role in promoting the attachment of microparticles, agglomeration and connection, which could reduce the external surface area. Additionally, there were decrease in the pore volume and pore size under the effects of EDTA, which illustrated that EDTA molecules attached to SM2 surface and narrowed the pore volume. This confirmed to the observation in SEM analysis, and coincided with results presented by Ren et al. [33] and Repo et al. [41].

Conclusions
In this study, effects of EDTA-2Na on adsorption of Cd(II) and Pb(II) by SMs were explored, and the changes in microscopic characteristics of SMs were also determined by diverse instrument analysis to investigate the mechanisms. Results showed that EDTA could reduce the HM adsorption capacity of SMs effectively and economically at a reasonable MR value, which was around 0.5. Note that the contact time had a great impact on the effect of EDTA in terms of Pb. As for as other factors, the optimal initial pH for EDTA to take effects in Cd-SM1, Pb-SM1, Cd-SM2, and Pb   The effects of initial pH on the HM-SM-EDTA system, low-permeability soil minerals:(a) SM1, (b) SM2. All error bars represent standard deviation from triplicate experiments. When not shown, error bars are smaller than symbols.

Figure 7
Effects of low-molecular-weight organic acids on the HM-SM-EDTA system, the low-permeability soil mineral was SM2. All error bars represent standard deviation from triplicate experiments.

Figure 8
Effects of exogenous metal ions on the HM-SM-EDTA system, the low-permeability soil mineral was SM2.
All error bars represent standard deviation from triplicate experiments.