A novel method of utilizing permeable reactive kiddle (PRK) for the remediation of acid mine drainage

Accepted Manuscript Title: A novel method of utilizing permeable reactive kiddle (PRK) for the remediation of acid mine drainage Author: Woo-Chun Lee Sang-Woo Lee Seong-Taek Yun Pyeong-Koo Lee Yu Sik Hwang Soon-Oh Kim PII: DOI: Reference:

S0304-3894(15)30064-9 http://dx.doi.org/doi:10.1016/j.jhazmat.2015.09.009 HAZMAT 17082

To appear in:

Journal of Hazardous Materials

Received date: Revised date: Accepted date:

29-5-2015 12-8-2015 2-9-2015

Please cite this article as: Woo-Chun Lee, Sang-Woo Lee, Seong-Taek Yun, PyeongKoo Lee, Yu Sik Hwang, Soon-Oh Kim, A novel method of utilizing permeable reactive kiddle (PRK) for the remediation of acid mine drainage, Journal of Hazardous Materials http://dx.doi.org/10.1016/j.jhazmat.2015.09.009 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A novel method of utilizing permeable reactive kiddle (PRK) for the remediation of acid mine drainage

Woo-Chun Leea, Sang-Woo Leeb, Seong-TaekYunc, Pyeong-Koo Leed, Yu SikHwanga, SoonOh Kimb,*

a

Future Environment Research Center, Korea Institute of Toxicology, Jinju 660-844, Republic

of Korea b

Department of Geology and Research Institute of Natural Science (RINS), Gyeongsang

National University (GNU), Jinju 660-701, Republic of Korea c

Department of Earth and Environmental Sciences and KU-KIST Green School (Graduate

School of Energy and Environment), Korea University, Seoul 136-701, Republic of Korea d

Geologic Environment Division, Korea Institute of Geoscience and Mineral Resources,

Daejeon, 305-350, Republic of Korea

A manuscript for Journal of Hazardous Materials *Corresponding author. Tel.: +82 55 772 1477; fax: +82 55 772 1479. E-mailaddress: [email protected] (S.-O. Kim). Graphical abstract 1

HIGHLIGHTS

 A novel method utilizing permeable reactive kiddle (PRK) was developed for remediate acid mine drainage (AMD)  The PRK method recycles industrial waste.  Four crucial operational conditions were evaluated.  It was shown that the PRK method could be effectively applied to neutralize pH and to remove various contaminants.

Abstract Numerous technologies have been developed and applied to remediate AMD, but each has specific drawbacks. To overcome the limitations of existing methods and improve their effectiveness, we propose a novel method utilizing permeable reactive kiddle (PRK). This manuscript explores the performance of the PRK method. In line with the concept of green technology, the PRK method recycles industrial waste, such as steel slag and waste cast iron. Our results demonstrate that the PRK method can be applied to remediate AMD under optimal operational conditions. Especially, this method allows for simple installation and cheap expenditure, compared with established technologies.

Keywords:Acid mine drainage, Permeable reactive kiddle, Heavy metals, Arsenic, pH neutralization

2

1. Introduction

Acid mine drainage (AMD) is caused by the oxidation of sulfide minerals, such as pyrite remaining in ores and tailings within abandoned mines [1, 2]. As a result, it contains a higher level of toxic contaminants, such as various heavy metals (Cd, Cu, Fe, Mn, Ni, Pb, Zn, etc) and metalloids (As) [3]. AMD migrates and threatens human health and the surrounding ecosystem by contaminating soil, streams, groundwater, and crops [4−8]. In particular, it has been reported that the accumulation of As and heavy metals in the human body results inacute and chronic poisoning, which can lead to cancer [3]. For this reason, a number of technologies have been developed to clean up such AMD. Among these, closure of the pit mouth, solidification/stabilization (S/S), passive treatment (pH neutralization and biological processes), and permeable reactive barrier (PRB) have been most actively applied so far [9−18]. However, each technology has crucial drawbacks, e.g., closure of the pit mouth is not permanent, S/S has the potential for remigration of contaminants according to changes in physicochemical conditions of the surrounding environment, the passive treatment requires a huge area, continuous management, and is very susceptible to climate change, and PRBrequries a relatively larger scale, higher expenditure for the facility and implementation, and is restricted to a specific area. Another shortcoming of these technologies, such as S/Sis that their primary goal is to lessen migration and dispersion of contaminants rather than to remove contaminants. Hence, it is necessary to develop more advanced methodsto overcome the limitations of existing technologies. From a practical standpoint, the technology should be efficient, cost-effective, stable in the long term, and easily implemented with simple facilities and management. To satisfy such requirements, the technology should be applicable in situ: 3

i.e., directly applied to the channel of the target contaminated plume, such as AMD and mine wastewater. Limestone and steel slag (SS) have been spotlighted as the best materials for treatment of mine drainage due to their cost-effectiveness and high performance [19, 20]. In particular, the SS contains a variety of components such as Fe, Mn, and Si, and also high level of Ca due to the addition of caustic lime for the removal of impurities during the refining process of pig iron [21]. For this reason, it shows a higher adsorption capacity for heavy metals and Asas well as pH neutralization, resulting in its extensive application for the treatment of AMD [22,23]. Recently, waste cast iron (WCI) has been recycled as a novel sorbent for heavy metals and As [24]. WCI is a byproduct of the iron casting process in foundries, and it can be classified as cast iron shot (CIS, 0.85-1.0 mm) and grind precipitate dust (GPD, <0.1 mm) based on the particle size. CIS is a byproduct of cast steel shot which is a ballast material used in the iron casting process, while GPD is a major form of waste generated from grinding and polishing cast iron products [24]. A large amount of these materials is produced as waste, and they can be obtainedcheaply. In addition, the specific surface area of WCI is larger than that of the steel slag because WCI is smaller than steel slag. As a result, WCI has higher potential as a sorbent for the removal of As and heavy metals.In addition, the applicability of ZVI was also investigated because it has been widely used as a sorbent material in the conventional treatments, such as PRB. Its performance has been already demonstrated from numerous studies [25−27]. We added a small amount of ZVI as a sorbent material because it is expensive. As mentioned above, SS and WIC are byproducts generated from the processes in refining pig iron and iron casting, respectively, and they can be cheaply recycled. For this reason, we tested SS and WCI as alternatives of ZVI due to their cost-effectiveness. The objective of this study was to develop a novel technology to remediateAMD. To 4

overcome the limitations of the abovementioned existing technologies, we designed permeable reactive kiddle (PRK).The term “kiddle” originated from the fishery and is a tool by which fishermen catch fish, and we use the term because it is distinct from “barrier” in the PRB process and it has a similar goal of a fishing kiddlein that it catches contaminants from a water body.The sorbent materials used in the PRK mainly consist of SS with a small quantity ofWCI and zerovalent iron (ZVI). To evaluate its performance for pH neutralization and removal of heavy metals and As from AMD, we conducted stepwise experiments to investigatevarious aspects, such as theform (structure) of PRK, composition of sorbents,and the size and arrangement of PRK.

2. Materials and methods

2.1. Materials

The permeable kiddle (PRK) consisted of three principal components: the permeable membrane, the supporting material (frame), and the sorbent. The permeable membrane was made of cloth (polyester nylon)with a pore size of 0.2 mm. The frame was manufactured using plasticmaterials with four diameters of 4, 8, 12, and 16 cm(Fig. S1 in the Supplementary data).The plastic frame was made from polyvinyl chloride (PVC).In the preliminary experiments, the affinity of plastic frame on contaminants was tested without sorbent materials, and the results confirmed that contaminants were not adsorbed on the frame. In addition, adsorption of sorbent materials onto the plastic frame was checked, and 5

their adsorbed amount was negligible. The 4 and 8-cm frames were made of plastic capsules (1.5 mm thick) purchased commercially and into which holes were drilled in the laboratory(Fig. S1a andb in the Supplementary data). The other two sized frames were commercial laundry plastic balls (2mm thick) with holes (Fig. S1 c and d in the Supplementary data).The composition of sorbent differed according to the form of PRK (see Section 2.2). We used three types of sorbent—steel slag (SS), waste cast iron (WCI), and zerovalent iron (ZVI).The SS was obtained from a steel mill plant in Korea and the particle size was 1−2 mm. The WCI was taken from a foundry factory and classified into two types: cast iron shot (CIS, 0.85-1.0 mm) and grind precipitate dust (GPD, <0.1 mm) based on the particle size. The chemical composition of SS and WCI was measured using x-ray diffraction (XRD, D5005, Siemens, Germany) and x-ray fluorescence (XRF, XRF-1700, Shimadzu, Japan).The ZVI was synthesized in the laboratory following the method of Lowry and Johnson (2004) [28]. According to Lowry and Johnson (2004) and Shih et al. (2011) [28, 29], the size of ZVI synthesized with this method has been known to be 10−100 nm, and the size of ZVI was confirmed by TEM analysis in this study (data not shown). All chemicals and standard solutions used in the experiments and instrumental analyses were of analytical reagent grade (Sigma-Aldrich, Merck, USA) and all solutions were prepared with > 18 MΩ cm deionized water using a Millipore Milli-Q system (Purelab option-Q ELGA DV24, UK).Acid mine drainage (AMD) used in this study was artificially preparedand its pH was adjusted to 4 using 0.5 M HCl and NaOH. The concentrations of simulated contaminants were 5 ppm for arsenate [As (V)], 2 ppm for Cd, 10 ppm for Pb, and 100 ppm for Cu; these concentrations corresponded to approximately 100-fold the Korean legal standard of effluent of mine wastewater. In the case of As, arsenite [As (III)] and arsenate [As (V)] are predominant species of As in geo-environments, but we only accounted 6

for As (V) because it appears to beconsiderablymore abundant than As (III) in real AMDs. The simulated AMD was made from Na2HAsO4∙7H2O,

Cd[(NO)3]2,Cu[(NO)3]2,and

Pb[(NO)3]2. To simulate the AMD channel, we manufactured a small-scale channel using acryl material with a length of 140 cm, width of 30 cm, and height of 20 cm (Fig. S2 in the Supplementary data). This simulated channel is termed a continuous flow reactor (CFR)hereafter. The level of AMD in the channel was maintained at6.0−6.5 cm (Fig. S2 in the Supplementary data). The flow rate of AMD was fixed at 6 ton/day using a peristaltic pump (Masterflex, 1−100 rpm, three heads, USA). The level and flow rate of AMD in real situationsare not generalized because they fluctuate according to the dimensions of the channel and climate conditions, but we used the flow rate that was considered the typical average of abandoned metal mines of Korea. In addition, the size of PRK can be adjusted by consideringthe level and flow rate of AMD, asdiscussed in Section 2.2.

2.2. Experimental procedure

To evaluate the performance of PRK, we conducted stepwise experiments to investigate a variety of aspects, such as the form (structure), size, and arrangement of PRK, and the composition of sorbents.Fig. 1 shows experimental procedure of this study.

2.2.1. PRK form First, the most effective form of PRK was determined by comparing three different forms. 7

The forms of PRK were classified based on the existence and position of the permeable membrane and the type of sorbent materials (Fig. S3 in the Supplementary data). The first two forms of PRK hada permeable membrane: the outer and inner membrane (hereafter, the outer and inner forms, respectively), in which the permeable membranes were placed outside and inside of the supporting material (frame), respectively.The other form of PRK was named the block-type form because it did not have a permeable membrane, and containednumerous cubes of sorbent materials. To compare the performance of the various PRK forms, the same sorbent material was used and the efficiencies of pH neutralization and As removal were compared. Driedgrain-type SShaving the size of 1−2 mmwas used as sorbent material in the former two PRK forms (Fig. 1), but the block-type PRK was filled withnumerous cubes of 1 × 1 × 1 cm3 because of the lack of a permeable membrane. Each cube was manufactured by mixing SS (85%) and cement (15%) with deionized water (Fig. 1).The cement used as an adhesive was a general Portland cement purchased commercially.The ratio of 85% sorbent to 15% cement was determinedby weight during preliminary experiments. After mixing, thecubes were aged at 25°Cby periodic water spraying for 1 week.As given in Fig. 1, to determine the most effective PRK form, the experiments were conducted using the CFR reactor and the simulated AMD (pH: 4; As concentration: 10 ppm). The effluent was recirculated 10 times using a peristaltic pump and the number of PRKs used was 5.

2.2.2. PRK composition As shown in Fig. 1, the experiments were conducted to select the most efficient composition of sorbent using block-type PRK which was selected the most efficient form of PRK based on the results from the experiments on the PRK form.By comparing the efficiencies of pH neutralization and Asremoval among diverse types and mixing ratios of 8

constituents,the most efficient composition of sorbent was determined with a set of batchtype experiments and it was verified by CFR tests (Fig. 1). As mentioned previously, the ratio of cement and sorbent materials used in the batch-type experiments was fixed.Several compositions of sorbent materials were tested by changing the contents of WCI (CIS or GPD) to 1, 5, 10, 20, 30, 40, 50, 60, 70, and 80% and ZVI to 0.05, 0.1, and 0.5%. The remaining portion in the 85% sorbent material was assigned to the SS (Fig. 1). The solid and solution ratio used in these batch-type experiments was fixed at 1:5by mass. The reason that the content of ZVI was much lower than that of WCI is the difference in the cost-effectiveness; i.e., the former is more expensive than the latter because the WCI is recycled waste. After selecting three different sorbent material compositions based on the results of batch-type experiments, the CFR experiments were performed to verify their practical applications (Fig. 1).

2.2.3. PRK size The level and flow rate of AMD are affected by climatic conditions. Subsequently, the contact area between AMD and PRK also varies. As a result, the fluctuation of AMD level and flow rate can influence the overall performance of PRK. Therefore, these effects should be taken into account for practical applications. To accurately evaluate these effects, the experiments should be conducted by directly changing the level and flow rate of AMD. However, we investigated these effectsindirectly using individual PRKs of diverse diameters. Our approach can be justified in that the variation in the level and flow rate of AMD directly affects the contact area between AMD and PRK, and the diameterof individual PRK is easily controlled or diverse sizes can be facilitated. We tested PRK of 4, 8, 12, and 16 cm in diameter (Fig. 1).While only As was used as the target contaminant in the previous two 9

experiments on the optimization of the PRK form and the composition of sorbent materials, allcontaminants (including heavy metals and As) were considered in this experiment (Fig. 1).In addition, the block-type form of PRK and the sorbents of SS (75%) and CIS (10%) were used because this combination showed the best performance in previous experiments (Fig. 1).

2.2.4. PRK arrangement Another crucial design factor influencing the performance of the process is the arrangement of PRK because the shape and flow rate of the AMD channel differ between locations. Therefore, the PRK arrangement should be optimized for field conditions prior to implementation. For this reason, the strengths and drawbacks of diverse PRK arrangementsshould be taken into account. We compared three PRK arrangements: zigzag-, U-, and step-types (Fig. S4 in the Supplementary data). The zigzag-type arrangement was designed to have a slanted platform in the center of the channel to reduce the velocity of AMD and to increase the residence and reaction times(Fig. S4a in the Supplementary data).The U-type arrangement was designed to be suitable for implementation in a narrow channel in which the PRK could not be directly placed and the number of reactors could beflexibly adjusted to increase the reaction time (Fig. S4b in the Supplementary data). Finally, the step-type arrangement is the simplest, resembles real channels and is easily implemented by placing a number of partitions to increase the reaction time (Fig. S4c in the Supplementary data). All experiments were conducted using identical size (8cm)and number (12) of PRKs(Fig. 1), although12-cm PRKs showed the best performance in previous tests, 8cm ones were used to reduce the scale of laboratory experiments.In addition,flow rates of 0.4, 1.3, and 2.0 ton/day were testedto investigate the effect on the removal efficiency of contaminants(Fig. 1). The other experimental conditions were identical to those used 10

previously: the block-type form of PRK and the sorbent of SS (75%) and CIS (10%) were used, as shown in Fig. 1.

2.2.5.Analysis An aliquot of effluent was sampled after each runof experiments and pH and concentrations of As and heavy metals were measured using a pH meter (Orion 266S, Thermo Scientific, USA) and inductively coupled plasma spectrometer (OPTIMA 5300DV, PerkinElmer, USA), respectively.

3. Results and discussion

3.1. Physicochemical properties of sorbent materials The XRF results on the chemical composition of sorbent materials used are given in Table S1 in the Supplementary data. The major constituents of SS were CaO (36.40%), Fe2O3(24.35%), SiO2(19.30%), Al2O3(6.03%), MgO (5.00%), and MnO (2.92%). A higher CaO content was measured due to the addition of caustic lime for the removal of impurities during the refining process of pig iron. A large amount of Fe2O3 was also detected. These results indicate that SS is effective in neutralizing pH and removing contaminants due to the higher contents of these two components. In the case of WCI (CIS and GPD), the silica (SiO2) content was the highest because mold (cast) and glider are composed of silica. Although the iron content was considerablylower in WCI than in SS, the content of iron was detected to be slightly higher than those of the other components, indicating that WCI can be 11

used as a sorbent material. In addition, the particle size of WCI is markedly smaller than that of SS, and thus WCI likely hasgreater potential as a sorbent material than SS. The XRF results on elemental compositions were supported by the XRD analyses.The XRDresults on SS suggest that calcite (CaCO3) was detected as a primaryCa mineral and magnetite and hematite appeared to be Fe minerals (Fig. S5a in the Supplementary data). In the case of WCI, the main mineral was quartz and only magnetite was identified as a Fe mineral (Fig. S5b and c in the Supplementary data).

3.2 Evaluation of the performance of PRK

3.2.1. Effect of PRK form The effect of PRK form on performance was investigated. The efficiencies of the sorbents to neutralize pH and remove Asare shown in Table S2and Fig. 2. The following equation was used to calculate the amount of As removed per unit time and mass of sorbent material shown in Table S2: Removal amount per unit time and mass ( /

/

) =

× ×

where,T (g) is the total mass of Aswhich was calculated by measured As concentration multiplied by the volume of sample,Caver (%) is the average removal efficiency calculated from the results of triplicated experiments, taver(day) is the processing time for each test, and w (kg) is the mass of PRK used. The following forms are ordered from highest to lowest performance based on final pH and As removal efficiency:block-type form (11.4 and 15.47%), outer form (7.3 and 11.29 %), and inner form (6.6 and 5.96 %).The capacity of 12

PRK for pH neutralization entirely depends on the content of lime (as percentage of CaO) including calcite (CaCO3). In order to compare the pH neutralizing capacities between different forms of PRK, therefore, it is needed to calculate the lime content in each form of PRK (see Table S3). As shown in the Table S3, the block-type PRK shows the largest lime content, and its pH neutralizing capacity is superior to the other two forms. This result is consistent with previous reports [20−22, 30] in which the pH neutralization of AMD is significantly affected by the lime content of treatment materials. In particular, the block-type form showed a more rapidincrease of pH from the early period of experiment, and the higher performance of the block-type form contributed to its higher reactivity because the sorbent materials could directly interact with AMD as a result of the increased permeability due to the lack of a membrane. We tested the permeability of each form of PRK using colored ink in the preliminary experiments (data not shown) and the results confirmed that the ink passed more effectively in the block-type PRK than through the outer and inner forms of PRK, indicating the higher permeability of the block-type of PRK. Withregard to As removal, the block-type form showedapproximately threefold higher efficiency than the inner form. Furthermore, the removal efficiency of As did notreach steady state, even after the 10thrun, indicating that As would be removed more effectively if the duration was extended. These results suggest that the overall performance of the block-type form was superior to the other forms.Based on the As removal amount [g/(kg∙day)] shown in Table S2 (2.05 for block-type, 1.50 for outer form, and 0.79 for inner form), thetreated AMD volume (L/kg)can be calculated to be 200, 102, and 77 for each form of PRK (As concentration: 10 ppm). This indicates that it is possible to treat about 1 metric ton of AMD using only 5−13 kg of sorbent material.

13

3.2.2. Effect of PRK composition To evaluate the performance of the process according to the composition of PRK, a series of batch experiments were conducted using the block-type form of PRK with a constant ratio of 15% and 85% of cement and sorbent material, respectively. Several compositions of sorbent materials were tested by mixing SS with WCI (CIS or GPD) and ZVI, and the results are summarized in Fig. 3 and Table S4. The variation of pH appeared to be very similar between different contents of sorbent materials, e.g., the pH ranged 10.8−11.3 for GPD and 9.8−10.2 for CIS, regardless of their content. However, the tendency of As removal was more obvious and the highest As removal efficiency was obtained at 10, 70, and 80% of GPD and CIS. The average removal efficiencies of GPD and CIS were calculated to be 18.78%, 16.61%, and 16.23% at their contents of 10, 70, and 80%, respectively. Therefore, among the contents of CIS and GPD tested, 10% and 70% showed the best performance in removing As. On the other hand, the CIS showed a higher Asremoval efficiencythan the GPD at an equivalent content. This contributed to thelarger primary pores(permeability)of CIS cubes than the GPD cubes, although the secondary poresof CIS and GPD cubes were identical. In particular, only a small addition (0.05, 0.1 and 0.5 %) of ZVI achieved a higherperformance, corresponding to the 10 and 70% CIS and GPD. Based on the batch experiments, CFR experiments were conducted to investigate their practical applicability. The efficiencies of pH neutralization and As removal were compared amongthreeauxiliary sorbent material compositions;namely, CIS (1, 5, and 10%), GPD (1, 5, and 10%), and ZVI (0.05, 0.1, and 0.5%). In all cases, the final pH approached 11, indicative of their excellent capacity for neutralizing the pH of AMD. For CIS, the As removal efficiencies of the 10th run were 16.09, 23.79, and 32.99%for 1, 5, and 10%, respectively. On the other hand,the As removal efficiencies of GPD ranged from 12.91to 25.25% with 14

increasing content from 1 to 10%(Table 1). Finally, the ZVI of 0.05, 0.1, and 0.5% showed As removal efficiencies of 20.80,34.87, and34.31%, respectively(Table 1), indicating that the As removal can be greatly improved with addition of a relatively extremely small amount of ZVI. This result is consistent with other researches reporting the outstanding capacity of ZVI in scavenging As [31, 32]. The results of CFR tests were similar to those of the batch tests. Overall, the greatest effect was attained in the presence of 10% CIS and GPD and 0.5% ZVI. In addition, the removal amounts of As per day were calculated to be 4.37, 3.34, and 4.62 g/day using l kg of sorbent materials composed of SS (75%) + CIS (10%), SS (75%) + GPD (10%), and SS (84.9%) + ZVI (0.1%), respectively (Table 1). This indicates that a large amount of As could be removed by PRK during the 1-day run. To support the improved performance by the addition of small amounts of auxiliary sorbent materials, As removal with andwithout auxiliary sorbent materials was compared. The results show that a higher performance was obtained in the order of SS (84.9%) + ZVI (0.1%) ≳SS (75%) + CIS (10%) >SS (75%) + GPD (10%) ≫SS (85%) (Fig.4).CIS was investigated as an efficient sorbent, similar to ZVI, but the former is inexpensivebecause it is a byproduct of the iron-casting process in foundries. Consequently, CIS could be considered an alternative auxiliary sorbent material of ZVI and could be used to improve the performance of PRK.

3.2.3. Effect of PRK size We investigated the effect of PRK size on the performance of processes. Whereas only As was a target contaminant in the previous two experiments, multiple contaminants—including heavy metals (Cd, Cu, and Pb) and As—were considered in these experiments. The removal 15

efficiency of As increased with an increasing size of PRK (Table 2). In addition, the average As removal amount of 12 cm PRK was calculated to be 10.97 g/(kg∙day), approximately fivefold greater than that [2.05 g/(kg∙day)] obtained in previous experiments using only As (Table S2andFig. 2).The reason why the pH(6.2−6.5)appeared to be lower in the experiment containing heavy metal contaminants than that (~11.4)in the experiment on only As is that three kinds of salts, such as Cd[(NO)3]2, Cu[(NO)3]2, and Pb[(NO)3]2 were added in the solution (refer to section 2.1). All these chemicals are acidic salts which make the solution acidic. Salts of strong acids and weak bases, such as Cd[(NO)3]2, Cu[(NO)3]2, and Pb[(NO)3]2, release cations (Cd2+, Cu2+, and Pb 2+) into solution that combine with hydroxyl ions (OH-) to form the parental bases [Cd(OH)2, Cu(OH)2, and Pb(OH)2], resulting in the removal of OH-. However, anions [(NO)3-] from the parental strong acid (HNO3) cannot react with hydrogen ion (H+), and H+ ions remains in the solution. As a result of such hydrolysis, the solution becomes more acidic and has higher acidity. In addition, the concentrations of simulated contaminants were 5 ppm for arsenate [As (V)], 2 ppm for Cd, 100 ppm for Cu, and 10 ppm for Pb (see section 2.1), rendering 6.41 x 10-3mM for As (V), 4.23 x 10-2mM for Cd , 5.33 x 10 -1mM for Cu, and 3.02 x 10-2mM for Pb. Based on the molarity (M) of each contaminant, the concentration levels of three heavy metals are 10−100 fold higher than that of As. For these reasons, the final pH values appeared to be lowerin the experiment of As with heavy metals than in the experiment of As only. It is known that arsenate removal by sorption increases with decreasing pH [24, 33-36]and this is confirmed in this study. Comparing the data on pH values and As removal efficiencies obtained from the experiments using the same form (block-type) of 12 cm PRK between Tables S2 and 2, the pHs were 11.4 (Table S2) and 6.5 (Table 2) and the As removal efficiencies were measured to be 15.47% (Table S2) and 82.84% (Table 2). 16

Similar to As, the sorption is speculated to be main mechanism for the removal of heavy metals. Sorption includes several reactions, such as absorption, adsorption, and coprecipitation, and these reactions contribute to the removal of heavy metals as well as As. In the pH condition (6.2−6.5), adsorption might be considered to be a predominant mechanism for the removal of heavy metals. The overall removal efficiencies (19.50−43.61%) of heavy metals were lower than As (74.04−82.84%) (Table 2andFig. 5). In particular, the Cd removal efficiency (19.50−23.34%) was considerably lower than the other two heavy metals (32.38−36.89% for Cu and 38.66−43.61% for Pb). This result isin agreement with previous reports [37−39]; i.e., the general affinity of heavy metals for iron oxides followed the order of Cd < Cu 8 cm ≥ 16 cm > 4 cm in diameter. It is likely that the PRK size influences the performance of the process because it affects the contact area between PRK and AMD. If the size of PRK is much smaller compared with the level of AMD, then the PRK is wholly sunk in the AMD. In this case, the PRK overflows with untreated AMD, and the overall performance decreases. On the contrary, when the size of PRK is much larger than the level of AMD, the reactivity of PRK decreases as a result of a decrease in contact area between PRK and AMD, resulting in a lower processperformance. Accordingly, the optimal size of PRK should be identical to the AMD level or slightly larger. Our results suggest that the most suitable size of PRK is 1.5 to 17

2-foldlarger than the AMD level (6 cm). In addition, it should be noted that the level and flow rate of AMD vary with changes in climate conditions, and the proper size and number of PRK requiredshould be determined based on sufficient pre-inspection of seasonal variationsin AMD levels. In addition, multiple layers of PRK (stacked PRK) are taken into account if the target AMD flow rate is expected to fluctuate.

3.2.4. Effect of PRK arrangement The performance of the process according to the PRK arrangement was evaluated by comparing the zigzag-, U-, and step-type arrangements. In all experiments, 12 PRKs of 8-cm diameter were used(Fig. S4in the Supplementary data) and the test was repeated three times. The flow rate and pH values are given in Table S5 (Supplementary data). Regardless of the type of PRK arrangement, the measured flow rates were consistent with the established rates. In addition, the flow rate did not significantly influence thepH. Removal efficiencies and removal amounts per unit time and unit mass of sorbent forthe contaminants are provided in Table 3andFigs. 6 and 7for the different flow rates and types of PRK arrangements. Overall, the zigzag- and U-type PRK arrangements showed similarly higher performance of contaminants under all flow rate conditions, while the step-type PRK arrangement showed the worst performance. In addition, it is likely that theremoval efficiencies of step-type arrangements were influenced to a greater extent by flow rate, compared with the other two types(Fig. 6). Arsenic and Pb showed higher removal efficiencies, while the removal efficiencies of Cd and Cu appeared to be lower (Fig. 6). Generally, the removal amount per unit time and unit mass of sorbent increased with increasing flow rate (Fig. 7) because thetotal mass of contaminants increased with increasing flow rate. The dependence of the amount of each contaminant removed on the flow rate increased with increasing flow rate in 18

the case of zigzag- and U-type arrangements (Fig. 7). However, this dependence was reduced in the step-type arrangement. Unlike the order of removal efficiency, the removed mass of each contaminant was in the sequence of Cu >Pb> As > Cd (Fig. 7) because the concentrations differed among contaminants (e.g., 5 ppm As, 2 ppm Cd, 100 ppm Cu, and 100 ppmPb). The abovementioned results on theeffect of PRK arrangement types could be explained by their specific characteristics. The zigzag-type arrangement showed the best performance owing to the increased residence and reaction times as a result of the slanted platform in the center of the channel. In particular, the zigzag-type shows high effectiveness regardless of changes in the flow rate. In the case of the U-type arrangement, which resembles a type of reactor, the reaction time could be prolonged similar to thezigzag-type, resulting in improved performance irrespective of flow rate. This may be suitable for implementation in a narrow channel, and the residence time could be easily controlled by changing the number of reactors. With regard to the step-type arrangement, although the presence of a number of partitions increased the reaction time, a large amount of AMD can overflow onto the surface of PRK, resulting in poor performance. Furthermore, this type was significantly affected by changes in the flow rate. Consequently, because each type of arrangement shows specific strengths and weaknesses, the combination of multiple arrangements should be taken into account when considering the features of the channel, as well as flow rate and velocity of AMD. For example, the step-type could be implemented as a basic arrangement, and the zigzag- and U-types may be applied as supplementary arrays in the corner of the channel and in the narrow areas of the larger head drop, respectively.

4. Conclusions 19

Using the PRK method, it was shown that AMD could beeffectively remediated. Dependence of its performance on various operational factors was proved by stepwise experiments. First, the block-type form of PRK showed the highest efficiency and one metric ton of AMD could be remediated using only 5−13 kg of sorbent material. In terms of the effect of PRK composition, the most effective combination SS (75%), CIS (10%), and cement (15%) indicating that CIS could be considered an alternative auxiliary sorbent material of ZVI to improve the performance of PRK. In addition, the experimental results indicate that the most suitable size of PRK is 1.5 to 2-fold larger than the level of AMD. Several types of PRK arrangement were evaluated, and the results suggest that zigzag- and U-types were efficient regardless of changes in the AMD flow rate. Overall, our results support that the newlydeveloped PRK method can be applied to remediate AMD and overcome several drawbacks of existing technologies. Particularly, the primary strength of the PRK method is ensuring simpler installation and lower expenditure compared with established methods.

Acknowledgements

This research was supported by the Basic Science Research Program through the National Research Foundation (NRF) of Korea funded by the Korean Ministry of Education (grant number: 2012R1A1A4A01001692).

References

20

[1] P.C. Singer, W. Stumm, Acidic mine drainage: the rate determining step, Science 167 (1970) 1121−1123. [2] U. Schwertmann, R.M. Cornell, Iron oxides in the laboratory: preparation and characterization. Wiley-VCH Publishers, New York, USA, 2000, pp. 489−493. [3] P.L. Smedley, D.G. Kinniburgh, A review of the source, behaviour and distribution of arsenic in natural waters, Appl. Geochem. 17 (2002) 517−568. [4] G. Morin, G. Calas, Arsenic in soils, mine tailings and former industrial sites, Elements.2 (2006) 97−101. [5] S. Wang, C.N. Mulligan, Occurrence of arsenic contamination in Canada: Soures, behavior and distribution, Sci. Total Environ. 366 (2006) 701−721. [6] G.Y. Jeong, B.Y. Lee, Secondary mineralogy and microtextures of weathered sulfides and manganoan carbonates in mine waste-rock dumps, with implications for heavy-metal fixation, Am. Mineral. 88 (2003) 1933−1942. [7] Y.G. Liu, M. Zhou, G.M. Zeng, X. Li, W.H. Xu, T. Fan, Effect of solids concentration on removal of heavy metals from mine tailings via bioleaching, J. Hazard. Mater.141 (2007) 202−208. [8] L. Rodríguez, E. Ruiz, A.J. Alonso, J. Rincón, Heavy metal distribution and chemical speciation in tailings and soils around a Pb-Zn mine in Spain, J. Environ. Manage. 90 (2009) 1106−1116. [9] D. Feng, C. Aldrich, H. Tan, Treatment of acid mine water by use of heavy metal precipitation and ion exchange, Miner. Eng. 13 (2000) 623–642. [10] D. Mohan, S. Chander, Single component and multi-component adsorption of metal ions by activated carbons, Colloids Surf. A 177 (2001) 183–196. 21

[11] S. Santos, R. Machado, M.J.N. Correia, Treatment of acid mining waters, Miner. Eng. 17 (2004) 225–232. [12] O. Gibert, J. de Pablo, J.L. Cortina, C. Ayora, Municipal compost-based mixture for acid mine drainage bioremediation: metal retention mechanisms, Appl. Geochem. 20 (2005) 1648–1657. [13] O. Gibert, J. de Pablo, J.L. Cortina, C.Ayora, Sorption studies of Zn(II) and Cu(II) onto vegetal compost used on reactive mixtures for in situ treatment of acid mine drainage, Water Res. 39 (2005) 2827–2838. [14] D.B. Johnson, K.B. Hallberg, Acid mine drainage remediation options: a review, Sci. Total Environ. 338 (2005) 3–14. [15] B.J. Wattena, P.L. Sibrella, M.F. Schwartzb, Acid neutralization within limestone sand reactors receiving coal mine drainage, Environ.Pollut. 137 (2005) 295–304. [16] X. Wei, R.C. Viadero Jr., K.M. Buzby, Recovery of iron and aluminium from acid mine drainage by selective precipitation, Environ. Eng. Sci. 22 (2005) 745–755. [17] M. Kalin, A. Fyson, W.N. Wheeler, The chemistry of conventional and alternative treatment systems for the neutralization of acid mine drainage, Sci. Total Environ. 366 (2006) 395–408. [18] D. Mohan, S. Chander, Removal and recovery of metal ions from acid mine drainage using lignite a low cost sorbent, J. Hazard. Mater. B137 (2006) 1545–1553. [19] P. Ziemkiewicz, Steel slag: applications for AMD control. Proceedings of the 1998 Conference on Hazardous Waste Research, 1998, pp. 44–59. [20] F.H. Pearson, A.J. McDonnell, Use of crushed limestone to neutralise acid wastes, J. Environ. Eng. Div. 101 (1975) 139–158. 22

[21] P.F. Ziemkiewicz, J.G. Skousen, D.L. Brant, P.L. Sterner, R.J. Lovett, Acid Mine drainage treatment with armored limestone in open limestone channels, J. Environ. Qual. 26 (1997) 1017−1024. [22] M.H. Jane, L. S. Philip, E. B. Harvey, Characterization of limestone reacted with acidmine drainage in a pulsed limestone bed treatment system at the Friendship Hill National Historical Site, Pennsylvania, USA, Appl. Geochem. 18 (2003) 1705−1721. [23] R. Eveliina, K.W. Jolanta, J.W. Lena, S. Mika, Steel slag as a low-cost sorbent for metal removal in the presence of chelating agents, J. Ind. Eng. Chem. (2015) http://dx.doi.org/10.1016/j.jiec.2014.12.025 [24] N.C. Choi, S.B. Kim, S.O. Kim, J.W. Lee, J.B. Park, Removal of arsenate and arsenite from aqueous solution by waste cast iron, J. Environ. Sci. 24 (2012) 589−595. [25] R.T. Wilkin, M.S. McNeil, Laboratory evaluation of zero-valent iron to treat water impacted by acid mine drainage, Chemosphere 53 (2003) 715−725. [26] O. Gibert, T. Rötting, J.L. Cortina, J. de Pablo, C. Ayora, J. Carrera, J. Bolzicco, In-situ remediation of acid mine drainage using a permeable reactive barrier in Aznalcóllar (Sw Spain), J. Hazard. Mater.191 (2011) 287−295. [27] F. Obiri-Nyarko, S.J. Grajales-Mesa, G. Malina, Review: An overview of permeable reactive barriers for in situ sustainablegroundwater remediation, Chemosphere 111 (2014) 243−259. [28] G.V. Lowry, K.M. Johnson, Congener-specific dechlorination of dissolved PCBs by microscale and nanoscalezerovalent iron in a water/methanol solution, Environ. Sci. Technol. 38 (2004) 5208−5216. [29] Y. Shih, C. Hsu, Y. Su, Reduction of hexachlorobenzene by nanoscale zero-valent iron: Kinetics, pH effect, and degradation mechanism, Sep. Purif. Technol. 76 (2011) 23

268−274. [30]J. Mahoney, M. Slaughter, D. Langmuir, J. Rowson, Control of As and Ni releases from a uranium mill tailings neutralization circuit: Solution chemistry, mineralogy and geochemical modeling of laboratory study results, Appl. Geochem. 22 (2007) 2758– 2776. [31] S.R. Kanel, B. Manning, L. Charlet, H. Choi, Removal of arsenic(III) from groundwater by nanoscale zero-valent iron, Environ. Sci. Technol. 39 (2005) 1291–1298. [32] X. Dou, R. Li, B. Zhao, W. Liang, Arsenate removal from water by zero-valent iron/activated carbon galvanic couples, J. Hazard. Mater. 182 (2010) 108-114. [33] W. Hartleya, R. Edwardsa, N.W. Lepp, Arsenic and heavy metal mobility in iron oxideamended contaminated soils as evaluated by short- and long-term leaching tests, Environ. Pollut. 131 (2004) 495-504. [34] X. Jianga, C. Peng, D. Fu, Z. Chen, L. Shen, Q. Li, T. Ouyang, Y. Wang, Removal of arsenate by ferrihydrite via surface complexation and surface precipitation, Appl. Surf. Sci. 353 (2015) 1087-1094. [35] T. Mahmood, S.U. Din, A. Naeem, S. Tasleem, A. Alum, S. Mustafa, Kinetics, equilibrium and thermodynamics studies of arsenate adsorption from aqueous solutions onto iron hydroxide, J. Ind. Eng. Chem. 20 (2014) 3234-3242. [36] W.C. Lee, S. Kim, J. Ranville, S. Yun, S.H. Choi, Sequestration of arsenate from aqueous solution using 2-line ferrihydrite: equilibria, kinetics, and X-ray absorption spectroscopic analysis, Environ. Earth. Sci. 71 (2014) 3307–3318. [37] U. Schwertman, R.M. Talyor, Iron oxides. Minerals in soil environments, 2nd ed., SSSA Book Series No. 1, Soil Science of America, Madison, Wisconsin, 1989, pp. 379−427. 24

[38] L.Y. Li, G. Wu, Numerical simulation of transport of four heavy metals in kaolinite clay. J. Environ. Engrg.125 (1999) 314−324. [39] C.A. Chrisophi, L. Axe, Competition of Cd, Cu, and Pb adsorption on goethite. J. Environ. Engrg.126 (2000) 66−74.

Figure Captions Fig. 1.The schematic diagram showing the experimental procedure. Fig.2.The result of CFR experiments on the effect of PRK form. (a) pH neutralization and (b) As removal.Error bars represent the standard deviations for each value. Fig. 3.The result of batch experiments on the effect of composition of sorbent. (a) pH neutralization and (b) As removal. The block-type form of PRK was used. Error bars represent the standard deviations for each value. Fig. 4.The result of CFR experiments showing comparison of performance between sorbents having four different compositions.(a) pH neutralization and (b) As removal. The block-type form of PRK was used. Remaining proportion of 15% is assigned to cement. Error bars represent the standard deviations for each value. Fig. 5.Removal efficiency of contaminants according to PRK size. (a) As, (b) Cd, (c) Cu, and (d)Pb. Error bars represent the standard deviations for each value. Fig.6.The effect of flow rate on the removal efficiency of contaminants according to the type of PRK arrangement.(a) Zigzag-type, (b) U-type, and (c) step-type.Error bars represent the standard deviations for each value.

25

Fig. 7.The effect of flow rate on the removal amount of contaminants according to the type of PRK arrangement. (a) As, (b) Cd, (c) Cu, and (d)Pb.Error bars represent the standard deviations for each value.

26

Fig. 1.

27

(a)

(b)

Fig. 2.

28

(a)

(b)

Fig. 3.

29

(a)

(b)

Fig. 4.

30

(a)

(b)

(c)

(d)

Fig. 5.

31

(a)

(b)

(c)

Fig. 6.

32

(b)

(a)

(c)

(d)

Fig. 7.

Table 1 The result of CFR experiments on the effect of composition of sorbent material. Mixing ratio (%) ZVI

Test number

CIS

GPD

0.05

0.10

0.50

1

5

10

1

5

10

1

2.20

3.45

4.36

1.24

3.98

4.42

1.16

1.71

1.79

2

5.13

8.80

6.70

1.87

6.22

6.98

1.88

3.28

3.67

33

3

8.01

11.91

9.30

3.64

7.40

9.41

2.43

4.49

6.73

4

11.21

14.93

15.05

5.15

9.62

14.15

3.72

8.07

8.97

5

11.97

21.12

20.14

7.72

12.95

17.03

4.69

10.14

12.36

6

15.37

22.59

23.04

8.70

14.70

20.85

7.17

10.76

13.44

7

15.76

27.32

25.60

9.93

15.69

23.38

7.88

12.56

17.53

8

17.55

30.99

30.15

12.13

18.59

26.81

10.15

13.70

18.91

9

19.07

32.61

31.31

13.31

20.45

28.23

10.95

16.26

21.19

10

20.80

34.87

34.31

16.09

23.79

32.99

12.91

18.11

25.25

Final pH

11.4

11.8

11.5

11.2

10.8

11.1

11.4

11.5

11.3

Removal efficiency per number

2.08

3.49

3.43

1.61

2.38

3.30

1.29

1.81

2.52

2.39

4.01

3.95

1.85

2.74

3.79

1.48

2.08

2.90

7.97

13.37

13.15

6.17

9.12

12.65

4.95

6.94

9.68

2.76

4.62

4.54

2.13

3.15

4.37

1.71

2.40

3.34

(%/number) Arsenate removal amount (mg) Removal amount per unit mass (mg/kg) Removal amount per unit time and mass (g/kg/day)

Table 2 The result of CFR experiments on the effect of PRK size.

Elements

As

PRK Total removal Average removal Final size efficiency efficiency pH (cm) (%) (%/number) 4

6.2

74.04

7.40 34

Removal amount per unit mass (mg/kg)

Removal amount per unit mass and time (g/kg/day)

28.38

9.81

8

6.4

78.96

7.90

30.27

10.46

12

6.5

82.84

8.28

31.76

10.97

16

6.4

77.72

7.77

29.79

10.30

4

6.2

19.62

1.96

3.01

1.04

8

6.4

19.50

1.95

2.99

1.03

12

6.5

23.34

2.33

3.58

1.24

16

6.4

21.95

2.19

3.37

1.16

4

6.2

32.38

3.24

248.22

85.78

8

6.4

36.89

3.69

282.84

97.75

12

6.5

35.59

3.56

272.86

94.30

16

6.4

34.27

3.43

262.77

90.81

4

6.2

38.66

3.87

29.64

10.24

8

6.4

39.39

3.94

30.20

10.44

12

6.5

43.61

4.36

33.44

11.56

16

6.4

41.37

4.14

31.72

10.96

Cd

Cu

Pb

35

Table 3 The result of experiments on the effect of PRK arrangement on the performance of process. Average removal efficiency (%)

Elements

As

Cd

Cu

Pb

Removal amount per unit mass and time (g/kg/day)

Flow rate (ton/day)

0.4

1.3

2

0.4

1.3

2

Zigzag

98.90

98.68

98.85

1.67

5.40

8.32

U-form

99.45

99.45

72.96

1.67

5.44

6.14

Step

45.92

22.03

16.48

0.77

1.21

1.39

Zigzag

15.18

8.38

5.82

0.11

0.19

0.21

U-form

10.52

4.48

5.08

0.07

0.10

0.18

Step

8.44

2.43

2.63

0.06

0.06

0.09

Zigzag

22.64

13.81

15.08

8.00

15.86

26.64

U-form

23.54

7.21

9.79

8.32

8.28

17.29

Step

12.57

13.36

5.05

4.44

15.34

8.91

Zigzag

50.11

53.03

58.20

1.77

6.09

10.28

U-form

55.18

50.77

56.15

1.95

5.83

9.92

Step

60.54

21.53

16.60

2.14

2.47

2.93

36