Removal of heavy metals from acid mine drainage using chicken eggshells in column mode

Journal of Environmental Management 188 (2017) 1e8

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Research article

Removal of heavy metals from acid mine drainage using chicken eggshells in column mode Ting Zhang a, Zhihong Tu a, Guining Lu a, b, c, *, Xingchun Duan d, Xiaoyun Yi a, b, Chuling Guo a, b, Zhi Dang a, b a

School of Environment and Energy, South China University of Technology, Guangzhou 510006, China The Key Laboratory of Pollution Control and Ecosystem Restoration in Industry Clusters, Ministry of Education, Guangzhou 510006, China Guangdong Provincial Engineering and Technology Research Center for Environmental Risk Prevention and Emergency Disposal, Guangzhou 510006, China d Water Supply Management Center of Guangzhou Development District, Guangzhou 510663, China b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 22 May 2016 Received in revised form 4 October 2016 Accepted 28 November 2016

Chicken eggshells (ES) as alkaline sorbent were immobilized in a fixed bed to remove typical heavy metals from acid mine drainage (AMD). The obtained breakthrough curves showed that the breakthrough time increased with increasing bed height, but decreased with increasing flow rate and increasing particle size. The Thomas model and bed depth service time model could accurately predict the bed dynamic behavior. At a bed height of 10 cm, a flow rate of 10 mL/min, and with ES particle sizes of 0.18e0.425 mm, for a multi-component heavy metal solution containing Cd2þ, Pb2þ and Cu2þ, the ES capacities were found to be 1.57, 146.44 and 387.51 mg/g, respectively. The acidity of AMD effluent clearly decreased. The ES fixed-bed showed the highest removal efficiency for Pb with a better adsorption potential. Because of the high concentration in AMD and high removal efficiency in ES fixed-bed of iron ions, iron floccules (Fe2(OH)2CO3) formed and obstructed the bed to develop the overall effectiveness. The removal process was dominated by precipitation under the alkaline reaction of ES, and the coprecipitation of heavy metals with iron ions. The findings of this work will aid in guiding and optimizing pilot-scale application of ES to AMD treatment. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Acid mine drainage Eggshells Fixed bed Thomas model Element mapping

1. Introduction Acid mine drainage (AMD), which is characterized by extreme acidity and a high level of dissolved heavy metals (Akcil and Koldas, 2006), is usually produced in the mining process. Because there are more than 17,481 mining companies in China, a large number of hazardous wastes are inevitably released annually from base-metal mining and smelting operations. Dabaoshan Mine, which is located in South China, is a typical site. Mud impoundment of this area mainly enters the Hengshi River, Wengjiang River, and several other tributaries. Local residents draw polluted water from these rivers to irrigate their crops, which resulted in a high mortality rate in the 1990s (Zhao et al., 2012; Zhuang et al., 2009). Heavy metals are toxic to aquatic organisms even at very low concentrations

* Corresponding author. School of Environment and Energy, South China University of Technology, Guangzhou 510006, China. E-mail address: [email protected] (G. Lu). http://dx.doi.org/10.1016/j.jenvman.2016.11.076 0301-4797/© 2016 Elsevier Ltd. All rights reserved.

(Malkoc and Nuhoglu, 2006). However, the concentrations of Cd2þ, Pb2þ and Cu2þ exceed 0.4, 1.0 and 6.0 mg/L in the AMD from the Dabaoshan Mine area, which are 40, 5, and 12 times higher than the permitted limits of the Standards for Irrigation Water Quality of China (GB 5084-2005), respectively. The concentrations of iron ions are also high (37.0e347.7 mg/L Fe3þ and 8.0e159.9 mg/L Fe2þ) (Chen et al., 2015). Zhao et al. (2012) found that Cd2þ is the major contributor to human health risk because of its easy absorption in crops in an acidic soil environment. Cd2þ and Pb2þ are considered potential carcinogens and are associated with the etiology of many diseases, especially cardiovascular, kidney, blood, nervous, and bone diseases €rup, 2003). Cu2þ is an essential element, however, its high con(Ja centrations in food and feed plants are of great concern because of its toxicity to humans and animals (Kabata-Pendias and Mukherjee, 2007). Therefore, wastewaters containing heavy metals must be treated before being discharged into water bodies. Although several treatments are used for metal removal, such as precipitation, ion exchange and solvent extraction, biomaterial adsorption possesses

2

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

particular strengths because biomaterials are environmentally benign, abundant and cost effective (Kadukov a and Vir cíkov a, 2005; Ok et al., 2007). According to the National Bureau of Statistics of China, domestic egg consumption reached 28,761 kilotons in 2014. Consequently, 3164 kilotons of eggshells (ES, ES account for 11% of the total weight of eggs) were generated, which is a problem because ES and the attached membrane attract vermin (Choi and Lee, 2015). Numerous studies have reported about metal ions adsorption on ES (Ahmad  et al., 2012; Bala z et al., 2015; Ergüler, 2015; Flores-Cano et al., 2013; Guijarro-Aldaco et al., 2011; Yeddou and Bensmaili, 2007). However, application of ES to remove heavy metals from AMD in a fixed bed has not been reported. Direct addition of powdered adsorbents to remove heavy metals may be efficient, but adsorbent loss as wasted sludge can be rather severe (Chern and Chien, 2002). In process application, a fixed-bed column is effective for cyclic sorption because it allows more efficient use of the sorbent ability, results in better quality effluent (Volesky and Prasetyo, 1994), is easy to operate, has no particle loss problem, and bears continuous flow (Chern and Chien, 2002; Malkoc and Nuhoglu, 2006). In this study, we investigated Cd2þ, Cu2þ and Pb2þ removal from AMD and also simultaneously considered removal of Fe. Chicken ES were chosen as a biosorbent for heavy metal removal from AMD. Granular ES were used in column mode. The aims of this work are as follows: (1) to measure the effects of different flow rates, bed depths and ES particle sizes on the breakthrough curves, (2) to analyze the concentrationetime profile and fixed-bed performance, and (3) to obtain a model that is sufficiently sophisticated to describe the main system performance but also simple for analysis. 2. Materials and methods 2.1. Adsorbent and adsorbate preparation Raw ES were provided by the canteen of the South China University of Technology, Guangzhou, China. The inner shell membranes were manually removed from the ES. The ES were washed three times with distilled water to remove impurities before being dried in a muffle furnace at 100  C for 24 h. The materials were then ground and sieved through 80, 40, and 18 mesh stainless-steel screens to obtain 0.18e0.425 mm and 0.425e1 mm particles. The obtained sorbents were stored in a desiccator before the experiments. AMD was sampled from the Dabaoshan Mine area, South China. Its initial concentrations of Cd2þ, Pb2þ and Cu2þ were 0.39 ± 0.04, 1.20 ± 0.10 and 6.30 ± 0.50 mg/L, respectively, and the pH was 2.4 ± 0.2. The concentration of Fe3þ ions was 195.20 ± 5.0 mg/L and the concentration of Fe2þ ions was only 5.1 ± 0.1 mg/L. The total Fe was considered in the following experiments. The other main constituents of AMD were SO2 (2000e4500 mg/L), Zn 4 (98e102 mg/L), Mn (49e62 mg/L), As (0.025e0.043 mg/L), Mg (128e140 mg/L), Cr (0.05e0.07 mg/L), Ni (0.4e0.5 mg/L) and Ca (110e270 mg/L), and the concentrations depended on the season and the year (Chen et al., 2015). The AMD solute was filtered using Whatman # 42 filter paper to remove suspended solids before running column system. 2.2. Adsorption studies Continuous flow biosorption experiments were carried out in Perspex-tube column (2.1 cm internal diameter and 40 cm height) packed with a weighed amount of ES. The temperature of all experiments was maintained at 25 ± 2  C. A 3 cm high layer of Ballotini balls (3 mm in diameter) was placed on top of the packed adsorbent for better flow distribution. Deionized water was used to

wash the ES to remove air bubbles, avoid channeling, and remove potential impurities before running fixed-bed system. A rotameter was used to determine the actual flow rate. The pH change of the AMD was also monitored. AMD solution was continuously pumped into the column in the up-flow direction by a peristaltic pump at 10, 20 or 30 mL/min until exhaustion. Experiments with three different bed depths, 10 cm (37 g of ES), 20 cm (74 g of ES) and 30 cm (110 g of ES), were operated at the same effluent flow rate (10 mL/min) with ES diameters of 0.425e1 mm. The influence of the ES grain size was investigated with ES diameters of 0.18e0.425 mm (40 g of ES) and 0.425e1 mm (37 g of ES) at the same effluent flow rate (10 mL/min) and bed depth (10 cm). In all tests, effluent samples were intermittently collected and analyzed with a double-beam atomic absorption spectrophotometer (SpectrAA-20, Varian). The reversibility of metal adsorption and the sustainability were investigated by desorption experiments. Once the adsorption bed was exhausted, the ES were immersed in 0.1 M HNO3 for 24 h and the amount of nitric acid used was such that the concentration of metal-laden ES was 37 g/L. The metal concentration after desorption was determined using the same method as that used for the effluent samples. The total Fe and Fe2þ concentrations were determined by 1e10 phenanthroline spectrophotography with a spectrophotometer (UV-VIS Spectrum 2550, SHIMADZU) at 510 nm. 2.3. Adsorption model The Thomas model equation for an adsorption column is as follows (Kapoor and Viraraghavan, 1998; Volesky and Prasetyo, 1994):

ct 1 ¼ c0 1 þ expðkTh q0 m=v  kTh c0 tÞ

(1)

where ct is the effluent metal concentration at time t, (mg/L), c0 is the influent metal concentration (mg/L), kTh is the Thomas rate constant (mL/min mg), q0 is the maximum solid-phase concentration of solute (mg/g), m is the mass of ES in the column (g), and v is the flow rate (mL/min). The kinetic coefficient (kTh) and adsorption capacity of the column (q0) can be determined from a plot of ct/ c0 against t at a given flow rate using the non-linear regression method. The AdamseBohart model is usually used to describe the initial part of the breakthrough curve. The equation is expressed as (Han et al., 2008):


ct Z ¼ exp kAB c0 t  kAB N0 F c0

(2)

where kAB is the AdamseBohart model kinetic constant (L/mg min), N0 is the maximum adsorption capacity (mg/L), Z is the bed height of the column (cm), and F is the linear velocity calculated by dividing the flow rate by the column section area (cm/min). From this equation, values describing the operational parameters of the column can be determined from a plot of ct/c0 against t at a given bed height and flow rate using the non-linear regression method. The bed depth service time (BDST) model expresses the service time (t) of the breakthrough curve. The equation is following (Ko et al., 2000; Othman et al., 2001):

t¼


 N0 1 c Z ln 0  1 Ka c0 c0 F ct

(3)

where Ka is the rate constant of the BDST model (L/mg min). A plot

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

3

of Z against t should give a straight line. 2.4. Analysis of column data The area above the breakthrough curve (A) obtained by integrating the adsorbed concentration (c0ect, mg/L) versus t (min) plot can be used to determine the total amount of adsorbed metal. The total amount of adsorbed metal (Mtotal, mg) in the column for a given influent concentration (c0) and flow rate (v) is given by Eq. (4):

Mtotal ¼

vA vc0 1000 1000

t¼total Z


1 t¼0

 ct dt c0

(4)

The total amount of metal ions sent to the column (mtotal) is calculated from Eq. (5):

mtotal ¼

Fig. 1. X-ray spectra of ES before (A) and after (B) experiment.

c0 vttotal 1000

(5)

The total removal efficiency (RE) is given by Eq. (6):

RE ¼

Mtotal  100 mtotal

(6)

The desorption efficiency (DE) is as follows:

DE ¼

cd V  100 Mtotal

(7)

where cd denotes the metals concentration in 0.1 M HNO3 desorption solution (mg/L), and V is the desorption solution volume (L). All experiments were performed in duplicate. The means and standard errors of the duplicate experiments were calculated by Origin 8.0 software. All standard errors were less than 0.5. 2.5. Error analysis To confirm the best fit model for the column system, it is necessary to analyze the data using the values of the least sum of squares (SS) combined with the correlative coefficient (R2) (Han et al., 2006; Ho et al., 2005). The SS value can be obtained as follows:

P SS ¼

ðct =c0 Þc  ðct =c0 Þe n

2 (8)

where (ct/c0)c is the ratio of effluent to influent metal concentration obtained from the Thomas model, (ct/c0)e is the ratio of the effluent to influent heavy metal concentration obtained from experiment, and n is the number of sampling points. If the data from the model are similar to the experimental data, SS will be small (SS < 0.05); if they are different, SS will be large.

(Flores-Cano et al., 2013). The sharp peak (;) in the spectrum of ES after experiment revealed the presence of Fe2(OH)2CO3 on the surface of ES. This can be attributed to Fe in the AMD (with strong acidity) reacting with CaCO3 of the ES. However, the heavy metal carbonate crystals could not be characterized because of the short crystal-forming time, and the small amount of metal crystals could not be detected by XRD. Therefore, the surface morphology of ES after experiment was further examined by a scanning electron microscope (SEM) (Merlin, Germany) equipped with an energy dispersive X-ray detector (EDX) for microanalysis. 3.2. Scanning electron micrographs and energy dispersive X-ray detector (SEMeEDX) analysis SEMeEDX chemical element mapping was used to determine the ferrous product. Chemical element partitioning was used to distinguish the presence of Fe, Cd, Pb and Cu in the ES after adsorption. The results of SEMeEDX are shown in Fig. 2. Compared with the raw ES in Fig. 2a, many spherical crystals formed and Fe is the substrate in the ES after experiment (Fig. 2b). This can be attributed to a significant amount of Fe in AMD forming floccules under the alkaline reaction of the ES. The iron floccules (Fe2(OH)2CO3) would then co-precipitated with Cd2þ, Pb2þ and Cu2þ ions. Therefore, spherical crystals with heavy metals elements and Fe formed. Some researchers have suggested that iron ions, iron hydroxide, iron sulfate, iron oxides and iron carbonate could immobilize heavy metals by precipitating as salts and coprecipitating with other heavy metal ions (Khorasanipour et al., 2012; Kumpiene et al., 2008). The element mapping analysis in Fig. 2c and d showed that Pb and Cu adsorbed on the ES had a homogeneous distribution. However, Cd was difficult to detect for its very low content compared with Pb and Cu. This finding confirmed that surface co-precipitation occurred in the process of heavy metal removal using ES.

3. Results and discussion 3.3. Effect of bed height 3.1. X-ray diffraction (XRD) analysis The crystallinity of ES before and after experiment was characterized by XRD (Empyrean, Netherlands) with 2q angle varying between 10 and 50. As shown in Fig. 1, both the raw ES and ES after adsorption showed similar distinctive peaks (▽), which were identified as the crystalline phase of calcium carbonate in the calcite form. Disappearance of the peak between 20 and 25 of ES after experiment indicated the dissolution of CaCO3 (up to 89.5%)

The bed height could be a crucial factor influencing the removal capacity and breakthrough time of adsorption bed. Therefore, an appropriate bed height or bed volume will improve the availability of adsorbent and help to remove the maximum amount of target pollutants. The breakthrough curves at different bed depths are shown in Fig. 3aec. The breakthrough time of all metals increased with increasing bed depth from 10 to 30 cm. Chern and Chien (2002) have suggested that solid-phase mass-transfer resistance

4

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

Fig. 2. SEM images of (a) the raw ES and (b) ES after experiment, (c, d) EDX analyses of ES after experiment.

may dominate the overall adsorption rate because the more solid phase that becomes saturated, the higher mass-transfer resistance in the solid phase. This agreed with the result that metals had more time to contact with adsorbent because of broadened solute mass transfer zone. Compared with the breakthrough time of Pb2þ, Cu2þ and Fe, the breakthrough time of Cd2þ was relatively short for all tested conditions. This was possibly due to the very low concentration of Cd2þ in AMD and the competitive strengths of Pb2þ and Cu2þ for the surface sites of ES. When the bed height reached 30 cm, for Cd2þ, Pb2þ and Cu2þ, the effluent concentration remained under the limits of Standards for Irrigation Water Quality of China about 25 min, 10.5 h and 2.5 h, respectively. More than 50% of Cd2þ, Pb2þ, Cu2þ and Fe were removed up to 50 min, 17 h, 7 h and 15.5 h, and the bed exhaustion time occurred at 6 h, 27 h, 25 h and 24.5 h, respectively. Although the time to reach the standard for Cd2þ was not long enough, a suitable design following the model calculation will aid in field application. 3.4. Effect of flow rate The breakthrough curves at different flow rates are shown in Fig. 3a, d and e. For all metal ions, the breakthrough curves became steeper and the breakthrough time decreased with increasing flow rate. This is probably because the residence time in the column was not sufficiently long to reach adsorption equilibrium at the higher flow rates. And considering the mass transfer limitations and insufficient time of AMD inside the column, a high flow rate would

decrease the sorption performance of ES. With decreasing velocity through the bed, the depth of effective adsorption zone increased due to more time to contact. When the flow rate was increased from 10 to 30 mL/min, the exhaustion time decreased: for Cd2þ from 60 min to 10 min, for Pb2þ from 350 min to 50 min, for Cu2þ from 150 min to 20 min, and for Fe from 300 min to 100 min. Therefore, in field application, AMD retention by building dams or other interception methods will greatly increase the effectiveness of the adsorbent. 3.5. Effect of particle size The ES particle size has an important effect on the adsorption kinetics. Fig. 3a and f shows the breakthrough curves with particle sizes of 0.425e1 mm and 0.18e0.425 mm with the same flow rate and bed height. For the smaller particle sizes, the breakthrough curves followed a much more efficient profile than those of the larger particle sizes: the breakthrough time increased and the curves tended towards the classic “S” shape profile (Rivero et al., 2002). For ES particle sizes of 0.18e0.425 mm, for Cd2þ, Pb2þ and Cu2þ, the effluent concentrations reached the limits of the Standards for Irrigation Water Quality of China about 6 min, 5 h and 75 min, whereas for ES particle sizes of 0.425e1 mm it sustained about 4 min, 65 min and 25 min, respectively. For the 0.18e0.425 mm ES particles, the exhaustion time of Fe was two times longer than 0.425e1 mm ES particles. This is because smaller particles inside the column had larger specific surface area and shorter diffusion path, allowing the adsorbate to penetrate deeper

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

5

Fig. 3. Experimental and predicted breakthrough curves according to the Thomas model: (a) Test 1: Z ¼ 10 cm, v ¼ 10 mL/min, particle size ¼ 0.425e1 mm; (b) Test 2: Z ¼ 20 cm, v ¼ 10 mL/min, particle size ¼ 0.425e1 mm; (c) Test 3: Z ¼ 30 cm, v ¼ 10 mL/min, particle size ¼ 0.425e1 mm; (d) Test 4: Z ¼ 10 cm, v ¼ 20 mL/min, particle size ¼ 0.425e1 mm; (e) Test 5: Z ¼ 10 cm, v ¼ 30 mL/min, particle size ¼ 0.425e1 mm; (f) Test 6: Z ¼ 10 cm, v ¼ 10 mL/min, particle size ¼ 0.18e0.425 mm.

into the adsorbent, and resulting in better adsorption performance. However, particles smaller than about 0.18 mm could move with the AMD solution, which were in a fluid state. Therefore, the overgrinding of ES was not appropriate even though the smaller particles had better removal efficiency for metal ions.

3.6. Estimation of breakthrough curves and analysis of adsorption performance 3.6.1. Thomas model To analyze the bed behavior and maximum solid-phase concentration (q0), the data was simulated with the Thomas equation (whole breakthrough curve) by non-linear regression analysis with

6

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

the aid of Origin 8.0 software. The predicted curves simulated with the Thomas model are plotted in Fig. 3aef. The Thomas model provided excellent fits to the breakthrough curves for all tested conditions for whole time with high R2 values and small SS values, as shown in Table 1. Accordingly, the total amount of adsorbed metal (Mtotal), total removal efficiency (RE) and desorption efficiency (DE) were calculated using Eqs. (4) (6) and (7), and the results are given in Table 2. The pH changes of the AMD are also given in Table 2. The Thomas model parameters and calculated parameters of Fe are given in Table 3. The rate constant (kTh) increased with increasing flow rate, decreasing bed height and increasing particle size. The bed capacity improved with decreasing flow rate and decreasing particle size, while it did not show a specific trend with increasing bed height in accordance with the calculated total amount of adsorbed metal (Mtotal). In particular, the bed capacity for Cd2þ, Cu2þ and Fe did not show an increasing trend after the bed height reached above 20 cm. It is possible that a significant amount of Fe formed floccules in AMD effluent with pH value greater than 4 (the pH of effluent was 4.31e5.59), and more than 50% of iron accumulated in the ES fixedbed. Therefore, agglomeration and blocking occurred in the front of the fixed-bed and then the back part of the bed was not effective. The adsorption capacity for Pb2þ increased strictly when the bed height was increased from 10 to 30 cm. The better adsorption potential was consistent with the lowest final ct/c0 of Pb2þ (about 0.8) among the three heavy metals. Compared with the other tested conditions, the smaller particle size of ES showed the greatest improvement in adsorption capacity of the fixed bed. The ranges of the total removal percentage (an indication of the system performance) of Cd2þ, Pb2þ, Cu2þ and Fe were 5.11%e 18.65%, 30.42%e77.09%, 5.27%e55.29% and 50.06%e62.38%, respectively. The removal efficiencies of Cd2þ, Pb2þ and Cu2þ were actually higher for higher bed height (10e20 cm), lower flow rate and smaller particle size. However, different conditions did not significantly change the removal rate of Fe. The desorption rates of Cd2þ, Pb2þ, Cu2þ and Fe were 52.10%e86.85%, 18.27%e46.52%, 34.31%e58.90% and 15.25%e23.56%, respectively. In the desorption solution with the same concentration of HNO3, the high removal capacity of Fe (>8404.0 mg/g) in ES fixed-bed leaded to its lowest desorption rate. Among the three toxic heavy metals, Pb2þ had the highest removal efficiency and the lowest desorption rate, whereas Cd2þ had the lowest removal efficiency and the highest desorption rate. The highest pKsp (negative log of the solubility-product constant) of lead carbonate and the pH values increases of 2.25e2.51 (inflow) to 4.31e5.59 (outflow) accelerated precipitation of Pb2þ. Park et al. (2016) suggested that the adsorption preference for Pb2þ may be attributed to: (1) higher hydrolysis constant, (2) higher atomic weight, (3) higher ionic radius (subsequently smaller hydrated radius), and (4) larger Misono softness value. For Cd2þ, its very low concentration and low pKsp value may explain why it had the lowest removal efficiency. Cd2þ could be easily exchanged and substituted by other metals, which resulted in its highest

desorption rate (Park et al., 2016). It is worth pointing out that the adsorbent will lose its efficacy after desorption because of gradual deterioration in the strongly acidic environment. Because elution may destroy the binding sites, the adsorbent removal ability obtained in the first sorption cycle could not reach in the subsequent cycles (Vijayaraghavan et al., 2004). Therefore, as a widely available and low cost adsorbent, reuse of ES does not make sense. 3.6.2. AdamseBohart model The AdamseBohart adsorption model was applied to analyze the relative concentration region (ct/c0 < 0.5) of breakthrough curves for Tests 1e6. The related parameters are given in Table 4. The correlation coefficients and SS values indicated that the AdmaseBohart model adequately fitted the initial part of breakthrough curves. The adsorption capacity (N0) and kinetic constant (kAB) show similar trends to the Thomas model for changes in the bed height, flow rate, and ES particle size. However, for whole relative concentration region, the simple and comprehensive approach showed large discrepancies between the experimental and predicted curves. Therefore, its validity is limited to the ranges of conditions used (Han et al., 2008). 3.6.3. BDST model The BDST model is based on the assumption that the rate of adsorption is controlled by the surface reaction ignoring intraparticle mass transfer resistance and external film resistance. If the adsorption zone moves at a constant speed along the column, the bed adsorption capacity will be a constant throughout the bed. Thus, the BDST model works well and provides a useful model equation for the change of the system parameters, such as the flow rate (Ko et al., 2000). The experimental values (dots) and predicted plots (dashed lines) of Z against t for Cd2þ, Pb2þ and Cu2þ at ct/ c0 ¼ 0.2, 0.4 and 0.6 are shown in Fig. 4. The related parameters of the BDST model equation according to the slopes and intercepts of the lines are listed in Table 5. The BDST model fitted to breakthrough curves well and showed a straight line for the same ct/c0 values of the three metals. With increasing ct/c0, the N0 value increased, while Ka decreased. Under different conditions, the trends of N0 and Ka for Cd2þ, Pb2þ and Cu2þ was consistent with the results obtained with the Thomas model and the initial part of the AdamseBohart model. With increasing bed height, the residence time of the solute inside the column increased, allowing the adsorbate to diffuse deeper inside the adsorbent. Hence, the bed capacity changed with the service time. 4. Conclusions The following conclusions can be drawn: (1) ES can simultaneously and efficiently remove heavy metals and reduce the acidity of AMD in dynamic continuous adsorption mode.

Table 1 Thomas model parameters of heavy metals for different conditions obtained by non-linear regression analysis. Test

1 2 3 4 5 6

kTh (mL/min mg) (103)

R2

q0 (mg/g)

SS

Cd

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

677.24 205.81 201.16 1104.22 1777.42 470.04

14.37 9.30 2.41 47.33 61.0 6.84

5.90 1.89 0.89 35.29 84.43 1.94

1.13 1.85 1.63 0.74 0.53 1.57

44.69 70.26 122.75 29.05 27.42 146.44

123.02 278.95 275.17 38.60 25.42 387.51

0.953 0.930 0.973 0.933 0.950 0.942

0.924 0.988 0.971 0.903 0.915 0.990

0.979 0.990 0.967 0.939 0.946 0.977

0.0085 0.0094 0.0041 0.0040 0.0016 0.0078

0.0171 0.0013 0.0020 0.0205 0.0089 0.0007

0.0053 0.0128 0.0047 0.0072 0.0037 0.0030

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

7

Table 2 Total amount of adsorbed metal (Mtotal), total removal efficiency (RE), and desorption efficiency (DE) of heavy metals obtained by the Thomas model, and the pH changes of the AMD. Test

Mtotal (mg)

1 2 3 4 5 6

RE (%)

DE (%)

pH change

Cd

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

inflow

outflow

0.0428 0.1397 0.1806 0.031 0.0243 0.0639

1.7149 5.2077 13.6621 1.1068 1.0734 5.8847

4.6635 20.7482 30.9747 1.4721 0.9649 15.7702

17.32 18.85 13.75 5.11 7.23 18.65

43.56 77.09 69.46 33.14 30.42 67.95

21.31 55.69 31.22 7.32 5.27 35.91

71.57 66.64 52.10 73.26 77.89 86.85

46.52 37.50 21.15 18.36 18.27 29.54

58.90 46.77 33.78 49.83 45.20 34.31

2.51 2.39 2.25 2.48 2.32 2.35

4.77 5.24 5.59 4.58 4.31 5.45

Table 3 Thomas model parameters, total amount of adsorbed metal (Mtotal), total removal efficiency (RE), and desorption efficiency (DE) of Fe for different conditions. Test

kTh (mL/min mg) (103)

q0 (mg/g)

R2

SS

Mtotal (mg)

RE (%)

DE (%)

1 2 3 4 5 6

0.098 0.036 0.026 0.158 0.192 0.045

8404.23 18019.00 17945.59 8407.72 10278.52 23514.72

0.985 0.996 0.997 0.967 0.974 0.997

0.0017 0.0004 0.0003 0.0027 0.0024 0.0003

315.65 1335.80 1976.25 321.36 393.35 943.62

50.06 57.25 58.04 50.95 62.38 62.35

23.56 19.38 17.59 22.68 21.25 15.25

Table 4 AdmaseBohart model parameters for different conditions obtained by non-linear regression analysis (ct/c0 < 0.5). kAB (mL/min mg) (103)

Test

1 2 3 4 5 6

R2

N0 (mg/L)

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

655.38 257.36 156.23 771.85 1020.42 341.35

22.75 7.54 2.43 51.64 71.86 5.99

6.25 1.27 1.03 39.4 56.30 2.01

4.26 2.07 2.14 1.32 1.20 2.43

45.22 86.49 144.17 34.91 31.96 194.17

146.20 382.26 311.69 45.85 38.12 499.67

0.893 0.911 0.911 0.820 0.851 0.865

0.961 0.991 0.974 0.958 0.923 0.993

0.949 0.929 0.920 0.972 0.922 0.976

0.0043 0.0030 0.0042 0.0083 0.0025 0.0053

0.0052 0.0006 0.0005 0.0091 0.0015 0.0036

0.0055 0.0088 0.0021 0.0008 0.0027 0.0006

Fig. 4. BDST model plots for different bed heights ([Cd]0 ¼ 0.40 mg/L, [Pb]0 ¼ 1.14 mg/ L, [Cu]0 ¼ 6.14 mg/L, v ¼ 10 mL/min).

Table 5 BDST model parameters for different bed heights obtained by linear regression analysis ([Cd]0 ¼ 0.40 mg/L, [Pb]0 ¼ 1.14 mg/L, [Cu]0 ¼ 6.14 mg/L, v ¼ 10 mL/min). ct/c0

0.2 0.4 0.6

SS

Cd

Ka (L/mg min) (103)

N0 (mg/L)

R2

Cd

Pb

Cu

Cd

Pb

Cu

Cd

Pb

Cu

577.62 151.97 115.48

5.45 1.07 0.82

4.84 0.76 0.51

1.39 1.73 2.37

96.32 133.37 167.94

181.79 288.21 407.93

1.0 0.997 0.999

0.940 0.974 0.920

0.935 0.906 0.940

(2) The ES fixed-bed had a relatively high removal rate (>50%) and removal capacity (>8400 mg/g) for iron ions under the alkaline reaction of ES. Fe2(OH)2CO3 easily formed and blocked the adsorption bed, and decreased its removal capacity for other heavy metals. (3) The Thomas model, AdamseBohart model (ct/c0 < 0.5) and BDST model all fitted the data for different conditions and can be applied to predict other processes with different flow rates and bed heights. (4) For all metal ions investigated, the bed capacity increased with decreasing particle size and decreasing flow rate, while it did not show a specific trend with increasing bed height after 20 cm. (5) For the three toxic heavy metals, the ES adsorption bed showed the highest removal efficiency for Pb2þ and the lowest removal efficiency for Cd2þ. Cd2þ ions were not effectively removed in the ES fixed-bed after long-time operation.

Acknowledgements The work was financially supported by the National Key Technology Support Program (No. 2015BAD05B05), the National Natural Science Foundation of China (No. 41330639), the Tip-top Scientific and Technical Innovative Youth Talents of Guangdong Special Support Program (No. 2015TQ01Z233), and the Guangdong Natural Science Funds for Distinguished Young Scholar (No. 2015A030306005).

8

T. Zhang et al. / Journal of Environmental Management 188 (2017) 1e8

Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jenvman.2016.11.076. References , J., Vir Kadukova cíkov a, E., 2005. Comparison of differences between copper bioaccumulation and biosorption. Environ. Int. 31, 227e232. Ahmad, M., Usman, A.R., Lee, S.S., Kim, S.-C., Joo, J.-H., Yang, J.E., Ok, Y.S., 2012. Eggshell and coral wastes as low cost sorbents for the removal of Pb2þ, Cd2þ and Cu2þ from aqueous solutions. J. Ind. Eng. Chem. 18, 198e204. Akcil, A., Koldas, S., 2006. Acid Mine Drainage (AMD): causes, treatment and case studies. J. Clean. Prod. 14, 1139e1145.   , A., Dankov Bala z, M., Bujn akov a, Z., Bal a z, P., Zorkovska a, Z., Brian cin, J., 2015. Adsorption of cadmium (II) on waste biomaterial. J. Colloid Interface Sci. 454, 121e133. Chen, M., Lu, G., Guo, C., Yang, C., Wu, J., Huang, W., Yee, N., Dang, Z., 2015. Sulfate migration in a river affected by acid mine drainage from the Dabaoshan mining area, South China. Chemosphere 119, 734e743. Chern, J.-M., Chien, Y.-W., 2002. Adsorption of nitrophenol onto activated carbon: isotherms and breakthrough curves. Water Res. 36, 647e655. Choi, H.-J., Lee, S.-M., 2015. Heavy metal removal from acid mine drainage by calcined eggshell and microalgae hybrid system. Environ. Sci. Pollut. Res. 1e8. Ergüler, G.K., 2015. Investigation the applicability of eggshell for the treatment of a contaminated mining site. Miner. Eng. 76, 10e19. Flores-Cano, J.V., Leyva-Ramos, R., Mendoza-Barron, J., Guerrero-Coronado, R.M.,  n-Pin ~ a, A., Labrada-Delgado, G.J., 2013. Sorption mechanism of Cd(II) from Arago water solution onto chicken eggshell. Appl. Surf. Sci. 276, 682e690. ndez-Montoya, V., Bonilla-Petriciolet, A., MontesGuijarro-Aldaco, A., Herna n, M.A., Mendoza-Castillo, D.I., 2011. Improving the adsorption of heavy Mora metals from water using commercial carbons modified with egg shell wastes. Ind. Eng. Chem. Res. 50, 9354e9362. Han, R., Wang, Y., Han, P., Shi, J., Yang, J., Lu, Y., 2006. Removal of methylene blue from aqueous solution by chaff in batch mode. J. Hazard. Mater. 137, 550e557. Han, R., Ding, D., Xu, Y., Zou, W., Wang, Y., Li, Y., Zou, L., 2008. Use of rice husk for the adsorption of Congo red from aqueous solution in column mode. Bioresour. Technol. 99, 2938e2946. Ho, Y.S., Chiu, W.T., Wang, C.C., 2005. Regression analysis for the sorption isotherms of basic dyes on sugarcane dust. Bioresour. Technol. 96, 1285e1291.

€rup, L., 2003. Hazards of heavy metal contamination. Br. Med. Bull. 68, 167e182. Ja Kabata-Pendias, A., Mukherjee, A.B., 2007. Trace Elements from Soil to Human. Springer Science & Business Media. Kapoor, A., Viraraghavan, T., 1998. Removal of heavy metals from aqueous solutions using immobilized fungal biomass in continuous mode. Water Res. 32, 1968e1977. Khorasanipour, M., Tangestani, M.H., Naseh, R., Hajmohammadi, H., 2012. Chemical fractionation and contamination intensity of trace elements in stream sediments at the sarcheshmeh porphyry copper mine, SE Iran. Mine Water Environ. 31, 199e213. Ko, D.C., Porter, J.F., McKay, G., 2000. Optimised correlations for the fixed-bed adsorption of metal ions on bone char. Chem. Eng. Sci. 55, 5819e5829. Kumpiene, J., Lagerkvist, A., Maurice, C., 2008. Stabilization of As, Cr, Cu, Pb and Zn in soil using amendmentsea review. Waste Manag. 28, 215e225. Malkoc, E., Nuhoglu, Y., 2006. Removal of Ni (II) ions from aqueous solutions using waste of tea factory: adsorption on a fixed-bed column. J. Hazard. Mater. 135, 328e336. Ok, Y.S., Yang, J.E., Zhang, Y.-S., Kim, S.-J., Chung, D.-Y., 2007. Heavy metal adsorption by a formulated zeolite-Portland cement mixture. J. Hazard. Mater. 147, 91e96. Othman, M., Roddick, F., Snow, R., 2001. Removal of dissolved organic compounds in fixed-bed columns: evaluation of low-rank coal adsorbents. Water Res. 35, 2943e2949. Park, J.H., Ok, Y.S., Kim, S.H., Cho, J.S., Heo, J.S., Delaune, R.D., Seo, D.C., 2016. Competitive adsorption of heavy metals onto sesame straw biochar in aqueous solutions. Chemosphere 142, 77e83. n ~ ez, R., Ortiz, M.I., 2002. Mathematical modelling of styrene drying Rivero, M.a.J., Iba by adsorption onto activated alumina. Chem. Eng. Sci. 57, 2589e2592. Vijayaraghavan, K., Jegan, J., Palanivelu, K., Velan, M., 2004. Removal of nickel(II) ions from aqueous solution using crab shell particles in a packed bed up-flow column. J. Hazard. Mater. 113, 223e230. Volesky, B., Prasetyo, I., 1994. Cadmium removal in a biosorption column. Biotechnol. Bioeng. 43, 1010e1015. Yeddou, N., Bensmaili, A., 2007. Equilibrium and kinetic modelling of iron adsorption by eggshells in a batch system: effect of temperature. Desalination 206, 127e134. Zhao, H., Xia, B., Fan, C., Zhao, P., Shen, S., 2012. Human health risk from soil heavy metal contamination under different land uses near Dabaoshan Mine, Southern China. Sci. Total Environ. 417, 45e54. Zhuang, P., McBride, M.B., Xia, H., Li, N., Li, Z., 2009. Health risk from heavy metals via consumption of food crops in the vicinity of Dabaoshan mine, South China. Sci. total Environ. 407, 1551e1561.