Passive biochemical reactors (PBRs) are a viable alternative to neutralization plants for the treatment of acid mine drainage (AMD) because they require lower investment costs and use residual materials. However, high iron (Fe) concentrations (≥0.5 g/L) in AMD are challenging for their long-term efficiency. Sorption and precipitation are the main Fe removal mechanisms, but the relative importance of each is mostly unknown. In this study, locally available natural materials (organic and inorganic) were characterized and tested for their performance in Fe removal from highly contaminated AMD (pH 3.5, 4 g/L of Fe, and 9 g/L of sulfate). Iron retention capacity of the materials was then evaluated and the efficiency of eight mixtures of materials was compared through 40-day laboratory batch tests. All batch-type PBRs increased the pH up to 6.5 and decreased dissolved metals concentrations, including Fe, up to 99%. Results showed that organic residual materials (manures, municipal wastewater sludge, and compost) were the best substrates for Fe removal.These findings allowed for the selection of three reactive mixtures with distinct characteristics (mixture #1 – 30% organic wastes; mixture #4 – 50% calcite; and mixture #7 – 50% sand) to be further evaluated in column type PBRs.

INTRODUCTION

Acid mine drainage (AMD), which originates from reactions between water, air, and sulfidic mine wastes, is characterized by low pH and high concentrations of sulfate and metals (Nordstrom et al. 2015). The treatment of AMD aims to increase pH and alkalinity, as well as remove metals and sulfate. The mechanisms of metal removal are precipitation and sorption, while sulfate is removed by precipitation and sulfate reduction (US EPA 2014; Skousen et al. 2017). The relative importance of each mechanism is related to specific conditions, such as pH, redox potential, metal concentration, and alkalinity (Haakensen et al. 2015). At abandoned or closed mine sites, passive treatment with limestone drains or passive biochemical reactors (PBRs) is usually preferred. While limestone drains utilize only chemical processes to increase the pH and alkalinity, and remove metals and sulfate, PBRs also involve sulfate-reducing bacteria (SRB). The principle of using PBRs lies in the biodegradation of organic substrates coupled with sulfate reduction to form bicarbonate and hydrogen sulfide (Postgate 1984): 
formula
1
Then, the hydrogen sulfide precipitates metals (M2+) according to the following reaction: 
formula
2

Experience shows that better sulfate-reduction rates are obtained with multi-component organic waste materials (US EPA 2014; Skousen et al. 2017). Typically, efficient PBR substrates contain both easily biodegradable materials (e.g. manure, wastewater sludge) and more recalcitrant ones (e.g. compost, sawdust, wood chips) (Neculita et al. 2007). Passive treatment provides satisfactory efficiency for slightly contaminated AMD, but the treatment of iron-rich AMD (≥500 mg/L) is more challenging. A limited number of studies were performed on this topic using single units (Chang et al. 2000; Neculita et al. 2008a, 2008b; Rakotonimaro et al. 2016) or multi-step systems (Genty et al. 2012a, 2012b, 2016; Ayora et al. 2013; Rakotonimaro et al. 2017) in order to improve performance and extend the lifespan of passive treatment. Research performed over several years led to the development of a new type of passive treatment system known as dispersed alkaline substrate, which consists of a mixture of coarse substrate (wood chips) and a neutralizing agent (e.g. calcite, magnesia), for the treatment of AMD containing up to 1.5 g/L Fe (Ayora et al. 2013). The AMD at abandoned or closed coal and metal mine sites is often characterized by up to 111 g/L Fe and low concentrations of other metals (Nordstrom et al. 2015). The treatment efficiency of passive systems for iron-rich AMD is usually limited by the hydraulics of the systems that lead to decreasing porosity and permeability due to clogging (Neculita et al. 2008b; Ayora et al. 2013).

A laboratory study efficiently treated iron-rich AMD, in a PBR operated at 20 days of hydraulic retention time (HRT) for a period of ∼20 weeks, but Fe concentrations subsequently increased until the end of testing (35 weeks) (Chang et al. 2000). In other laboratory studies, a mixture consisting of four substrates (i.e. manure, wood chips, sawdust, compost), in addition to other inorganic components (i.e. sand, sediment, calcite), was reported as efficient for the treatment of iron-rich AMD, at 10 days of HRT over a 15-month period, in 3.5 L laboratory columns (Neculita et al. 2008a). However, a severe decrease in porosity and permeability was found at the end of the testing period. In the latter study, oxy-hydroxides were identified as the main minerals and precipitation was determined to be the main mechanism of iron removal. However, the relative importance of precipitation and sorption changes over time (Neculita et al. 2007, 2008b). Based on the authors' best knowledge, very little is known about the role of specific mechanisms of iron removal during the treatment of ferriferous AMD. Ion exchange was found as the predominant sorption mechanism when a multi-metal solution was mixed with wood waste (Su et al. 2012). In another study, more efficient iron removal was observed at pH 3 rather than at pH 6 (Song et al. 2012). The negative charge of organic solid surfaces could explain the higher sorption at low pH.

Under aerobic conditions, the oxidation of Fe(II) to Fe(III), followed by the hydrolysis and the precipitation of oxyhydroxides is an important iron removal mechanism. In contrast, under anoxic/anaerobic conditions, the reduction of Fe(III) to Fe(II) can occur, followed by the precipitation of bivalent species, such as iron carbonates and/or sulfides. Both conditions exist in PBRs; i.e., the feeding section of the reactor is aerobic, but it progressively becomes anaerobic. Thus, iron removal mechanisms can change over the lifespan of the PBR. Therefore, the main objective of this study was to evaluate the relative importance of precipitation and sorption as potential mechanisms of iron removal during batch treatment of iron-rich AMD (4 g/L Fe – often found at closed and abandoned coal/metal mine sites), in order to select the most efficient reactive mixtures to be used in a continuous flow, field-scale PBR.

MATERIALS AND METHODS

Natural organic materials

This study was performed using local materials, with the goal of installing a field-scale PBR at the Lorraine mine site (QC, Canada). Maple wood chips and sawdust were obtained from P.W.I. Industries; poultry manure from Fertilo de Fafard; sheep manure and cow manure from Canadian Tire Corporation; leaf compost from Composts du Quebec; fine calcite (<0.625 cm) from Calcite du Nord; and sediments (Osisko Lake) and wastewater sludge from the town of Rouyn-Noranda, QC, Canada.

Synthetic AMD

Testing was carried out with a synthetic ferriferous AMD (Table 1), which had a low pH (3.5) and high concentrations of Fe (4 g/L) and sulfate (9 g/L). The composition of the synthetic AMD used in this study was selected to simulate AMD observed at the Lorraine mine site (Genty et al. 2012a, 2012b).

Table 1

Composition of the synthetic AMD

Parameter Concentrations mg/L, except pH Chemicals used 
Al Al2(SO4)3.18H2
Cd 0.5 CdSO4.8H2
Cr CrK(SO4)2.12H2
Fe 4,000 FeSO4.7H2O 
Mg 10 MgSO4.7H2
Mn 10 MnSO4.H2
Ni NiSO4.6H2
Pb 0.5 Pb(NO3)2 
SO42−a 9,000 Na2SO4.10H2
Zn 0.5 ZnSO4.7H2
pH 3.5 HCl 1N 
Parameter Concentrations mg/L, except pH Chemicals used 
Al Al2(SO4)3.18H2
Cd 0.5 CdSO4.8H2
Cr CrK(SO4)2.12H2
Fe 4,000 FeSO4.7H2O 
Mg 10 MgSO4.7H2
Mn 10 MnSO4.H2
Ni NiSO4.6H2
Pb 0.5 Pb(NO3)2 
SO42−a 9,000 Na2SO4.10H2
Zn 0.5 ZnSO4.7H2
pH 3.5 HCl 1N 

aThis concentration includes the contribution of all sulfate salts.

The synthetic AMD was prepared using ACS grade salts and deionized water, which was bubbled with nitrogen gas for 15 min in order to limit iron oxidation and precipitation.

Solid characterization

All eight mixtures of materials (Table 2) were thoroughly characterized. Water content was evaluated by drying samples at 40 °C for 48 h (CEAEQ 2015). Relative density and specific surface area were measured using a helium pycnometer (Micromeritics Accupyc 1330) and surface analyzer (Micromeritics Gimini III 2375), respectively, both on dry materials as per Brunauer et al. (1938). Particle size distribution was determined as per Aitcin et al. (1983). The pH was measured in deionized water using a solid-to-liquid ratio of 1:10 (ASTM 1995). Total organic carbon (TOC) and total Kjeldahl nitrogen (TKN) were determined using Method 5310B (with high temperature combustion) and Method 4500-Norg, respectively (APHA 2012). Water extracts (1:10 solid-to-liquid ratio) were used for dissolved organic carbon (DOC) analyses. Total metal content was analysed, after digestion (HNO3, Br2, HCl, and HF), by ICP-AES (Perkin Elmer OPTIMA 3100 RL). Cation exchange capacity (CEC) was evaluated as per Chapman (1965). The SRB counts were determined by using the most probable number method (ASTM 2009). The mineralogy of crystalline phases was analyzed using X-ray diffractometer (XRD; Bruker AXS D8 ADVANCE) equipped with a Cu source and scintillation counter. Data processing was performing using the EVA and TOPAS software packages from Bruker AXS. Mineral proportions were estimated (0.5% precision) by the Rietveld method (Raudsepp & Pani 2003).

Table 2

Composition (% w/w, dry) of the reactive mixtures

  Mixture
 
#1 #2 #3 #4 #5 #6 #7 #8 
Cellulosic wastes 
 Maple chips 10 10 10 10 10 10 
 Maple sawdust 20 20 20 10 20 20 11 20 
Organic wastes 
 Chicken manure 10    10 10 
 Sheep manure  10       
 Cow manure   10      
 Waste water sludge     10    
 Leaf compost 20 20 20 10 20 20 12 20 
Inoculum 
 Sediment 15 15 15 15  15 
Inert structural agent 
 Sand 20 20 20 10 20 35 50 20 
Nutrient (Nitrogen) 
 Urea  
Neutralizing agent 
 Calcium carbonate   
 Calcite    50    
  Mixture
 
#1 #2 #3 #4 #5 #6 #7 #8 
Cellulosic wastes 
 Maple chips 10 10 10 10 10 10 
 Maple sawdust 20 20 20 10 20 20 11 20 
Organic wastes 
 Chicken manure 10    10 10 
 Sheep manure  10       
 Cow manure   10      
 Waste water sludge     10    
 Leaf compost 20 20 20 10 20 20 12 20 
Inoculum 
 Sediment 15 15 15 15  15 
Inert structural agent 
 Sand 20 20 20 10 20 35 50 20 
Nutrient (Nitrogen) 
 Urea  
Neutralizing agent 
 Calcium carbonate   
 Calcite    50    

Analytical methods and geochemical modeling

The pH of samples was determined with an Orion Triode sensor coupled with a Benchtop pH/ISE Meter (Orion model 920). Redox potential (ORP) was measured using a Pt/Ag/AgCl electrode (values were corrected in order to obtain the Eh). Filtered (0.45 μm) and acidified (HNO3) samples were used to determine concentrations of dissolved metals. Sulfate concentrations were evaluated on filtered samples using barium sulfate precipitation (APHA 2012). Saturation indices of the main minerals were calculated with the thermodynamic equilibrium model Vminteq (version 2.53) using the activity correction SIT (KTH 2010). However, this model does not take into account microbial activity and subsequent metal sulfide precipitation (Zagury et al. 2006).

Evaluation of iron retention capacity

The iron retention capacity of each component of the eight reactive mixtures was assessed at two initial conditions: pH 3 and pH 6 (set using 1N HCl and 1N NaOH, respectively). After reaching equilibrium (24 h), the final pH was measured for all materials, including substrates and sediments.

Sorption isotherms

Isotherm experiments were performed at 21 °C, in 200 mL Erlenmeyer flasks, which were filled with 100 mL of AMD and 2 g of dry material, and agitated at 200 rpm. The flask headspace was exposed to air. The solid-to-liquid ratio used ensured that equilibrium conditions were quickly reached (Nieto et al. 2010). Metal concentrations varied from initial AMD quality to 50 times dilution, but only iron concentrations were measured. Given the competition for sorption sites, ferriferous AMD prepared with other metals was considered more representative of quality of AMD in the field (Motsi et al. 2009; Su et al. 2012). After 24 h (at equilibrium), 15 mL samples were collected, filtered, and preserved for metal content analysis. The sorption capacity, qe (mg/g), of each material was calculated using the following mass balance equation (Limousin et al. 2007): 
formula
3
where Ci, Ce, V, and m are the initial metal concentration (mg/L), metal concentration at equilibrium (mg/L), total volume (L), and mass of the sorbent (g), respectively.

Kinetics studies

The reaction kinetics of iron sorption were also evaluated, at initial pH 3 and pH 6, using 200 mL Erlenmeyer flasks filled with 100 mL of AMD (diluted 20 times) and 2 g of dry material (agitated at 200 rpm). Water samples were collected at predetermined intervals (0, 2, 4, 8, and 24 h) to track iron concentrations. The iron uptake qt (mg/g) of each material was calculated using Equation (4) (Limousin et al. 2007): 
formula
4
where Ci, Ct, V, and m are the initial metal concentration (mg/L), metal concentration at time t (mg/L), total volume (L), and mass (g) of sorbent, respectively.

Interpretation of sorption studies

To describe equilibrium phenomena, both Langmuir and Freundlich isotherms were used. The linear form of the Langmuir equation can be represented by Equation (5) (Jha et al. 2008): 
formula
5
where Ce is the equilibrium concentration of residual metal in the solution (mg/L), qe is the amount of a metal adsorbed per unit mass of sorbent at equilibrium (mg/g), qmax is the amount of sorbate at complete monolayer coverage (mg/g), and b (L/mg) is a constant related to the heat of adsorption.
The Freundlich linear isotherm can be described by the following equation (Limousin et al. 2007): 
formula
6
where Ce (mg/g) and qe (mg/g) have the same significance as in the Langmuir model; kf is the equilibrium constant (mg/g), which is indicative of adsorption capacity; and n is the sorption equilibrium constant.
The models describing sorption kinetics are reaction- or diffusion-based (Ho et al. 2000). In the present study, the reaction-based model was employed. Sorption mechanisms are controlled by surface reactions in which metal ions are linked to the material. Pseudo-first-order or pseudo-second-order kinetic models can be used (Ho et al. 2000). The pseudo-first-order and pseudo-second-order models (Jha et al. 2008) which best describe the experimental data in the present study are presented below: 
formula
7
 
formula
8
where qe is the amount of a metal adsorbed per unit mass of sorbent at equilibrium (mg/g), qt is the amount of a metal adsorbed per unit mass of sorbent (mg/g) at the time t (s), k1 is the pseudo-first-order rate constant (s−1), and k2 the pseudo-second-order rate constant (g.mg.s−1).

Batch testing

The eight reactive mixtures (Table 2) were evaluated using batch tests over a period of 40 days.

Mixture #1 was previously found to be efficient for the treatment of iron-rich AMD (Neculita et al. 2008a), whereas mixtures #2, #3, and #5 evaluated different sources of easily degradable carbon. Mixture #6 allowed for a comparison of the efficiency among similar mixtures, with or without sediment (as an acclimated SRB inoculum), whereas mixtures #4 and #7 were used to assess the effects of increasing the percentage of granular material (Ayora et al. 2013). Mixture #8 was almost identical to mixture #1 except it did not contain urea.

Batch testing was performed in 1 L glass flasks, at 21 °C, under a nitrogen atmosphere, on a gyratory shaker (at 150 rpm). The solid-to-liquid ratio was 1:3 (200 g dry mixture to 600 mL of synthetic AMD). The pH, Eh, and sulfate and metal concentrations were measured weekly for 40 days. The pH and Eh were measured directly in the reactors, whereas sulfate and metal concentrations were measured from 5 mL samples. Geochemical modeling was performed at 0, 10, 24, and 40 days. At the end of testing, the chemical composition of the solid mixture was analysed.

RESULTS AND INTERPRETATIONS

Solid characteristics

Cellulosic wastes exhibited slightly more acidic pH values as compared to the organic materials, but contained twice as much TOC (Table 3). Consistently, the DOC concentrations of the cellulosic wastes were significantly higher than those of the organic wastes. TKN varied for all materials from 1.1 to 2.6%, except for the maple wood chips (0.1%) and sediments (0.04%). It is noteworthy that urea has a TKN of approximately 47%. The TOC/TKN for the eight mixtures tested in this study were also determined (Table 4).

Table 3

Physicochemical and microbiological characterization of organic materials and sediments

  Material
 
Maple chips Maple sawdust Poultry manure Cow manure Sheep manure Municipal sludge Leaf compost Sediments 
Physical parameters 
 Paste pH 6.2 5.6 6.0 7.0 7.1 7.3 6.7 7.1 
 Water content, % (w/w) 55 62 58 74 56 27 
 Relative density 1.45 1.48 1.57 1.69 1.64 2.04 1.80 2.66 
 Surface area, m2/g 0.58 0.86 1.15 0.96 1.10 15.75 1.67 13.02 
 CEC, meq/100 g 47.7 72.7 55.5 38.8 52.9 66.2 52.7 18.4 
Biodegradation parameters 
 TOC, % (w/w dry) 51 49 26 27 30 21 24 1.5 
 TKN, % (w/w dry) 0.1 0.2 2.6 1.4 1.1 1.9 1.3 0.04 
 DOC water extract, mg/L 280 450 57 35 32 5.6 48 6.4 
 TOC/TKN 510 272 10 19 27 11 18 38 
Sulfate reduction 
 SRB, cells/100 mL <2 <2 450 7.8 × 102 2.3 × 103 2.3 × 103 7.8 × 102 1.3 × 103 
  Material
 
Maple chips Maple sawdust Poultry manure Cow manure Sheep manure Municipal sludge Leaf compost Sediments 
Physical parameters 
 Paste pH 6.2 5.6 6.0 7.0 7.1 7.3 6.7 7.1 
 Water content, % (w/w) 55 62 58 74 56 27 
 Relative density 1.45 1.48 1.57 1.69 1.64 2.04 1.80 2.66 
 Surface area, m2/g 0.58 0.86 1.15 0.96 1.10 15.75 1.67 13.02 
 CEC, meq/100 g 47.7 72.7 55.5 38.8 52.9 66.2 52.7 18.4 
Biodegradation parameters 
 TOC, % (w/w dry) 51 49 26 27 30 21 24 1.5 
 TKN, % (w/w dry) 0.1 0.2 2.6 1.4 1.1 1.9 1.3 0.04 
 DOC water extract, mg/L 280 450 57 35 32 5.6 48 6.4 
 TOC/TKN 510 272 10 19 27 11 18 38 
Sulfate reduction 
 SRB, cells/100 mL <2 <2 450 7.8 × 102 2.3 × 103 2.3 × 103 7.8 × 102 1.3 × 103 
Table 4

Physicochemical characterization of the reactive mixtures tested in batch reactors

 Mixtures
 
#1 #2 #3 #4 #5 #6 #7 #8 
Physical parameters 
 Grains relative density 2.04 2.09 1.97 2.26 1.94 1.66 2.16 1.89 
Biodegradation parameters 
 TOC, % (w/w dry) 13 24 28 11 20 23 23 29 
 TKN, % (w/w dry) 1.0 0.9 1.1 0.4 6.2 1.0 0.9 0.3 
 DOC, mg/L 110 120 310 140 170 310 510 30 
 TOC/TKN 13 26 25 28 24 24 94 
 Mixtures
 
#1 #2 #3 #4 #5 #6 #7 #8 
Physical parameters 
 Grains relative density 2.04 2.09 1.97 2.26 1.94 1.66 2.16 1.89 
Biodegradation parameters 
 TOC, % (w/w dry) 13 24 28 11 20 23 23 29 
 TKN, % (w/w dry) 1.0 0.9 1.1 0.4 6.2 1.0 0.9 0.3 
 DOC, mg/L 110 120 310 140 170 310 510 30 
 TOC/TKN 13 26 25 28 24 24 94 

The manures, compost, and sludge contained SRB populations varying from 4.5 × 102 to 2.3 × 103 cells/100 mL, while SRB counts in the sediments were 1.3 × 103 cells/100 mL. The calcite showed a grain size D10 of 0.53 mm and a uniformity coefficient (CU) of 3.5. The XRD analysis confirmed the high purity of calcite (96.5 wt.%), with small impurities of quartz and hornblende.

The mineralogical composition of the sediment (determined by XRD) was as follows: albite (24%), tremolite (20%), quartz (17%), microcline (14%), chlorite (7%), talc (6%), and other minerals in lower concentrations. The municipal sludge and sediments had the highest specific surface areas. The CEC of the materials varied from 18.4 (sediments) to 72.7 meq/100 g (maple sawdust), suggesting that ion exchange could be an important contributor to sorption. These results were not unexpected because sediments contain clay minerals which are known for their ion exchange capacity, while organic materials are characterized by the highest CEC.

Sorption studies

Sorption of iron onto organic material surfaces is usually a complex process. Numerous mechanisms can occur, such as adsorption (physical or chemical), cation exchange, mechanical trapping, and surface precipitation. Isotherms help to describe sorption, whereas kinetic testing allows for the determination of the sorption limiting step (diffusion, surface reaction).

The recorded changes, upon reaching the equilibrium condition (i.e. after 24 h), showed some pH variations. In sorption tests performed at pH 3, cellulosic substrates (wood-based) maintained acidic conditions at equilibrium whereas organic substrates (i.e. the manures, compost, and sludge) and sediments increased the equilibrium pH to values ranging from 4.2 to around 5. In contrast, with an initial pH of 6, all materials led to a slight acidification of the solution (4.9–5.8), except for the municipal sludge which slightly increased the pH to a final value of 6.7. The pH was purposefully not strictly controlled during adsorption tests to (1) better represent the natural substrate behaviour under field conditions and (2) avoid interfering with adsorption processes.

Sorption isotherms

Prior to batch testing, the iron sorption capacity of natural organic materials was assessed, in order to evaluate its role in iron removal, especially during the first 20 days after beginning the experiments, but before the establishment of optimal conditions for SRB growth (lag phase). A classification of the sorption phenomena for liquid/solid systems proposes four classes of isotherms: high affinity (H), Langmuir (L), constant-partition (C) and sigmoidal shaped (S) (Limousin et al. 2007).

Isotherms (Figure 1) had a convex shape for all materials tested (equivalent to L-type) at pH 3 and a sigmoidal form (equivalent to S-type) at pH 6, except for the maple wood chips where the curve was more of an L-type. Thus, two mechanisms could explain sorption phenomena and the behaviour appear to be pH-related. At pH 3, the solids were progressively saturated and micropores were filled by monolayers of iron (e.g. chemisorption). On the contrary, at pH 6, monolayers were first formed, then several other layers were created. The main results related to the isotherm data are presented in Supplementary Table 1 (available with the online version of this paper). The Freundlich model was more appropriate to model the data for most materials at both pH 3 and pH 6, except for the sawdust and sediments, for which the Langmuir model gave better fit at pH 6.
Figure 1

Iron sorption isotherms, at pH 3 (a) and at pH 6 (b).

Figure 1

Iron sorption isotherms, at pH 3 (a) and at pH 6 (b).

The maximum sorption capacity of iron, qmax (mg/g), can be estimated graphically for each material (Table 1, Supplementary material). Based on this estimation, at pH 3, municipal sludge was the best adsorbent, whereas wood waste was the least effective one. At pH 6, the sorption capacity of cow manure was the highest, but maple sawdust remained the least effective material. Additionally, the sorption capacity of maple chips became comparable to manures at pH 6.

Overall, iron sorption by the tested materials seems to be pH-related; at pH 3 most of the materials had higher iron sorption capacity than at pH 6. A negatively charged surface at acidic pH could partially explain the better efficiency of iron removal by sorption. Consistently, isotherm results (higher kf – adsorbed quantity of iron at equilibrium concentration of 1 mg/L) also showed that iron removal seemed to be more important at pH 3 than at pH 6 for all materials, except for maple chips and cow manure. These findings indicate that the tested materials were able to uptake iron directly from AMD. In the same time, at near neutral pH, iron surface-enhanced precipitation could improve iron retention (Limousin et al. 2007). However, sulfate sorption on iron oxyhydroxides can inhibit their growth (Diz et al. 2006). Finer-grained materials had higher sorption capacities. Values of qmax were consistent with the literature. Hence, organic materials (i.e. pomegranate peel) could uptake approximately 18 mg/g of iron (Moghadam et al. 2013). In another study, which performed iron sorption studies on crude olive stone (wood-based material), the qmax was approximately 0.65 mg/g (Nieto et al. 2010). In the present study, wood-based material also showed a low sorption capacity.

Modeling indicated the possible precipitation of iron oxides-hydroxides under these conditions. Indeed, during sorption, no precautions were taken to limit the presence of oxygen and the oxidation and precipitation of some iron in the form of oxides and hydroxides probably occurred.

Kinetics studies

The sorption capacities of the tested materials, at pH 3, stabilized after approximately 8 h at 8.7 mg/g, whereas for sediments a stable value was not reached even after 24 h (Figure 2).
Figure 2

Iron uptake as a function of time, at pH 3 (a) and at pH 6 (b).

Figure 2

Iron uptake as a function of time, at pH 3 (a) and at pH 6 (b).

Previously reported findings show that the induction time (i.e. the time between the addition of a neutralizing agent and the detection of iron precipitates) decreased with increasing iron concentration and pH, but increased with increasing sulfate concentration (Diz et al. 2006).

At pH 6, stabilization was reached at 0.9 mg/g for maple sawdust and wood chips, 1.4 mg/g for sediments, and 2.1 mg/g for the other materials. Results also showed that 24 h were generally sufficient to reach equilibrium conditions. Overall, similar iron uptake kinetics were found, at both tested pH values, for the municipal sludge, compost, and cow, poultry and sheep manure. No decrease in iron concentrations were observed for maple wood chips and sawdust at pH 3, probably due to the high Fe concentrations. A previous study, which assessed the affinity of a 14-metal mixture to wood waste (at pH 2.5–4.9), found that Fe3+, which sorption occurred in priority, was strongly sorbed to wood particles and decreased the sorption capacity of other metals present in solution (Su et al. 2012).

The results on the kinetics of iron sorption modeling for the organic materials and sediments at pH 3 and 6 are presented in Supplementary Table 2 (available with the online version of this paper). Several kinetic models were used to fit experimental data (Supplementary Table 3, available online). The pseudo-second-order reaction rates fit well with experimental data, except for sediments, at pH 3, where pseudo-first-order kinetics better represented the results; the results for maple chips and sawdust at pH 3 could not be fit by either model because iron concentrations did not decrease. At pH 3, iron sorption was limited by the reaction between dissolved iron and material surfaces according to a pseudo-second-order kinetic law or a pseudo-first-order for sediments. The pseudo-second-order kinetics model is often used for chemisorption modeling and has been applied successfully to biosorption systems and organic materials (Ho et al. 2000). Therefore, chemisorption could be an important phenomenon behind the iron sorption because of iron's high affinity towards organic materials (Vodyanitskii 2010). Moreover, the kinetic constants (k2) were higher at pH 6 than at pH 3 for all materials indicating faster sorption. These findings are consistent with other studies (Ho et al. 2000).

In summary, the kinetics to reach the equilibrium were more rapid at pH 6 than at pH 3, but the maximal iron uptake was higher at pH 3. Previous studies also reported better iron removal in strongly acidic AMD than in moderately acidic AMD (98–99% vs. 73–85%; Song et al. 2012).

Batch testing

Results showed that pH increased to up to 7 over the first 10 days, except for mixture #8 which required 30 days to reach neutral pH values (Figure 3(a)). Additionally, Eh decreased below 0 mV in most reactors. However, the Eh of mixtures #4 and #8 remained at 150 to 210 mV (Figure 3(b)). Moreover, all reactive mixtures removed more than 99% of iron in less than 20 days (Figure 3(c)). There was no significant decrease in sulfate concentrations during the first 18 days, which corresponds to the acclimation period of SRB (Neculita et al. 2008a), except for in mixture #8 where a progressive decrease was observed (Figure 3(d)). The measured initial sulfate concentrations were higher than those in AMD because the manure and compost released sulfate during testing (Zagury et al. 2006). After 40 days, sulfate removal varied from 29% (mixture #4) to 73% (mixture #8). The removal of Cd, Ni, Pb, and Zn was higher than 99% in all mixtures (data not presented). The majority of metal removal occurred between 0 and 10 days which is consistent with previously reported results (Zagury et al. 2006). The mechanisms of metal removal during the period after the start-up could involve sorption and precipitation (Neculita et al. 2007).
Figure 3

Evolution of pH (a), Eh (b), iron (c) and sulfates (d) in batch testing.

Figure 3

Evolution of pH (a), Eh (b), iron (c) and sulfates (d) in batch testing.

In all reactive mixtures, sulfate removal rates varied from 102 to 217 mg/L/day (for the entire period), and from 59 to 278 mg/L/day, between 18 and 40 days. However, depending on pH and Eh, sulfate removal could be attributed to sulfate reduction or to sulfate precipitation.

On day 0, modeling results indicated the oversaturation (and potential precipitation) of iron oxy-hydroxides (H-jarosite, ferrihydrite and goethite-lepidochrocite). For all mixtures, the precipitation of aluminium hydroxide, ferrihydrite (mixtures #4 and #8), goethite, and lepidochrocite was also indicated on days 10, 14 and 40. For mixtures #1, #3, #5, #6, and #7, the precipitation of sulfides of Fe, Ni, and Zn was also identified, whereas in mixtures #2, #4, and #8, gypsum may have precipitated.

Overall, the results indicated a similar performance by mixtures #1, #2, #3, and #5 in terms of pH increase and metal removal. These findings confirm that PBRs could benefit from the use of various easily degradable organic carbon sources, such as poultry, cow and sheep manure, or municipal sludge. The behaviour of mixture #6, whose performance was very similar to mixture #1, suggests that the SRB inoculum originating from an AMD-impacted sediment was not essential. Mixtures #4 and #7 demonstrated the best performance (comparable to mixture #1), although their composition included only half of organic substrate quantity relative to mixture #1. Structural agent proportions did not seem to affect the performance. Finally, mixture #8 was the least efficient (in terms of: sulfate removal from 18 to 40 days, iron retention kinetics, and increasing pH and alkalinity). Batch tests allowed for the selection of three reactive mixtures, which have very distinctive features (mixture #1 with 30% of organic wastes, mixture #4 with 50% of calcite, and mixture #7 with 50% of sand), to be evaluated in a column type PBR.

Role of sorption vs. precipitation in iron removal

Results confirmed that iron removal in iron-rich AMD occurs by sorption and precipitation. Sulfate reduction, which occurred after a 20-day lag period, could also have contributed to iron removal. The development of an intense black colour and the strong odour of hydrogen sulfide confirmed this hypothesis. Based on the evaluated sorption capacity of the mixtures and the quantity of iron effectively removed during the first 20 days, additional mechanisms appear to be responsible for iron removal (Table 5). A rough assessment of the difference between the total iron removed in batch testing and the maximum sorption capacity indicated that during the SRB latency, 28 to 74% of iron was probably removed by sorption, with the difference being removed by other mechanisms, such as precipitation of oxy-hydroxides (as indicated by thermodynamic geochemical modeling).

Table 5

Evaluation of iron sorption proportion out of total iron uptake for each mixture

Mixture Estimated qmax (mg/g)a Iron sorption capacity, for 200 g of mixture (mg) Iron uptake (mg)b % of iron sorptionc 
#1 8.07 1,614 2,579 63 
#2 7.98 1,596 2,590 62 
#3 8.34 1,668 2,578 65 
#4 3.17 634 2,591 24 
#5 9.63 1,926 2,586 74 
#6 5.4 1,080 2,579 42 
#7 3.68 736 2,584 28 
#8 6.28 1,256 2,528 50 
Mixture Estimated qmax (mg/g)a Iron sorption capacity, for 200 g of mixture (mg) Iron uptake (mg)b % of iron sorptionc 
#1 8.07 1,614 2,579 63 
#2 7.98 1,596 2,590 62 
#3 8.34 1,668 2,578 65 
#4 3.17 634 2,591 24 
#5 9.63 1,926 2,586 74 
#6 5.4 1,080 2,579 42 
#7 3.68 736 2,584 28 
#8 6.28 1,256 2,528 50 

aCalculated with graphical average qmax of each mixture components for pH 3 and pH 6.

bCalculated with iron concentration at 0 and 18 days.

cProportion of iron sorption.

A major finding of the present study was that a higher proportion of organic material in the reactive mixtures (such as #1, #2, #3, and #5) resulted in a higher sorption capacity, sorption being a dominant mechanism of iron removal during the early period of batch testing. Additionally, sorption testing could help to better understand the environmental behaviour of spent reactive mixtures. It is noteworthy that a study on the leaching potential of such iron-rich contaminated material showed that the prolonged contact with water should be avoided (Genty et al. 2012a, 2012b).

CONCLUSION

Batch testing was performed to evaluate the role of sorption in iron removal and to select the most efficient mixtures for the treatment of ferriferous AMD (pH 3.5, 4 g/L Fe, and 9 g/L sulfate) in PBRs. Results showed that sorption plays an important role during the start-up of ferriferous AMD treatment in PBRs and the Freundlich model best described most of the experimental data. During the acclimation period of SRB, 28 to 74% of iron was removed, likely by sorption. Residual organic materials, such as manures, municipal sludge and leaf compost, had higher iron sorption capacities (19.3–48.5 mg/g) relative to cellulosic wastes (2–3 mg/g). Iron sorption kinetics could be described for most of the tested materials by a pseudo-second-order equation. Overall, results showed that mixtures of organic materials could be used in the creation of efficient reactive mixtures for reactors filling. Although a high proportion (50% w/w) of structural agent (calcitic sand or sand) can improve the hydraulic properties of the reactive mixtures and increase the lifespan of the PBR, it does not improve iron removal. Ongoing research is being conducted in column tests in order to determine the optimal HRT prior to pilot testing at the field scale.

ACKNOWLEDGEMENTS

This research was supported by the Canada Research Chair on the Restoration of Abandoned Mine Sites and the Natural Sciences and Engineering Research Council of Canada (NSERC) through the Industrial NSERC Polytechnique-UQAT Chair in Environment and Mine Wastes Management.

REFERENCES

REFERENCES
Aitcin
J. C.
Jolicoeur
G.
Mercier
M.
1983
Technologie des granulats
,
2nd edn
.
Le Griffon d'Argile Inc.
(ed.)
,
Sainte-Foy, QC
,
Canada
, p.
386
.
American Public Health Association (APHA)
2012
Standard Methods for the Examination of Water and Wastewater
,
22nd edn
.
American Public Health Association, Washington, DC
,
USA
.
ASTM
1995
Standard test method for pH of soils
. In:
Annual Book of ASTM Standards
D4972-95a
,
American Society for Testing and Materials, Washington, DC, USA
.
ASTM
2009
Standard methods for sulphate-reducing bacteria in water and water-formed deposits
. In:
Annual Book of ASTM Standards
D 4412-84
,
American Society for Testing and Materials, West Conshohocken, PA, USA
.
Ayora
C.
Caraballo
M. A.
Macias
F.
Rötting
T. S.
Carrera
J.
Nieto
J. M.
2013
Acid mine drainage in the Iberian Pyrite Belt: lessons learned from recent passive remediation experiences
.
Environmental Science and Pollution Research
20
,
7837
7853
.
Brunauer
S.
Emmett
P. H.
Teller
E.
1938
Adsorption of gases in multimolecular layers
.
Journal of the American Chemical Society
60
,
309
319
.
Centre d'Expertise en Analyse Environnementale du Québec (CEAEQ)
2015
Détermination des solides totaux et des solides totaux volatils: méthode gravimétrique. MA. 100 – S.T. 1.1, Révision 4, Ministère du Développement durable, de l'Environnement et de la Lutte contre les changements climatiques, p. 13
.
Chapman
H. D.
1965
Cation-exchange capacity
. In:
Methods of Soil Analysis
(
Black
C. A.
Evans
D. D.
White
J. L.
Ensminger
L. E.
Clark
F. E.
, eds).
Part 1. Agronomy Monograph 9
.
ASA
,
Madison, WI
, USA, pp.
891
901
.
Diz
H. R.
Novak
J. T.
Rimstidt
J. D.
2006
Iron precipitation in synthetic acid mine drainage
. In:
Mine Water and the Environment
,
March 30–April 1
,
IMWA
,
St. Louis, MO
,
USA
, p.
14
.
Genty
T.
Bussière
B.
Potvin
R.
Benzaazoua
M.
Zagury
G. J.
2012a
Capacity of wood ashes filters to remove iron from acid mine drainage: assessment of retention mechanisms
.
Mine Water and the Environment
31
,
273
286
.
Genty
T.
Neculita
C. M.
Bussière
B.
Zagury
G. J.
2012b
Environmental behaviour of sulphate-reducing passive bioreactor mixture
. In:
Proceedings of the 9th International Conference on Acid Rock Drainage (ICARD)
,
May 21–25
,
Ottawa, ON, Canada
, p.
10
.
Genty
T.
Bussière
B.
Paradie
M.
Neculita
C. M.
2016
Passive biochemical treatment of ferriferous mine drainage: Lorraine mine site, Northern Quebec, Canada
. In:
Proceeding of the IMWA
,
July 11–15
,
Leipzig, Germany
, p.
6
.
Haakensen
M.
Pitte
V.
Spacil
M. M.
Castle
J. W.
Rodgers
J. H
Jr
.
2015
Key aspects for successful design and implementation of passive water treatment systems
.
Journal of Environmental Solutions for Oil, Gas, and Mining
1
,
59
81
.
Ho
Y. S.
Ng
J. C. Y.
McKay
G.
2000
Kinetics of pollutant sorption by biosorbent: review
.
Separation & Purification Reviews
29
,
189
232
.
KTH
2010
Visual MINTEQ A free equilibrium speciation model. Version 3.0. http://www.lwr.kth.se/English/OurSoftware/vminteq/index.html (accessed 12 May 2016)
.
Limousin
G.
Gaudet
J. P.
Charlet
L.
Szenknect
S.
Barthes
V.
Krimissa
M.
2007
Sorption isotherms: a review on physical bases, modeling and measurement
.
Applied Geochemistry
22
,
249
275
.
Moghadam
M.
Nasirizadeh
N.
Dashti
Z.
Babanezhad
E.
2013
Removal of Fe(II) from aqueous solution using pomegranate peel carbon: equilibrium and kinetic studies
.
International Journal of Industrial Chemistry
19
,
1
6
.
Motsi
T.
Rowson
N. A.
Simmons
M. J. H.
2009
Adsorption of heavy metals from acid mine drainage by natural zeolite
.
International Journal of Mineral Processing
92
,
42
48
.
Nieto
L. M.
Alami
S. B. D.
Hodaifa
G.
Faur
C.
Rodriguez
S.
Gimenez
G. A.
Ochando
J.
2010
Adsorption of iron on crude olive stones
.
Industrial Crops and Products
32
,
467
471
.
Nordstrom
K.
Blowes
D. W.
Ptacek
C. J.
2015
Hydrogeochemistry and microbiology of mine drainage: an update
.
Applied Geochemistry
57
,
3
16
.
Postgate
J. R.
1984
The Sulphate-Reducing Bacteria
.
Cambridge University Press
,
Cambridge
,
UK
.
Rakotonimaro
T. V.
Neculita
C. M.
Bussière
B.
Zagury
G. J.
2016
Effectiveness of various dispersed alkaline substrates for the pre-treatment of ferriferous acid mine drainage
.
Applied Geochemistry
73
,
13
23
.
Rakotonimaro
T. V.
Neculita
C. M.
Bussière
B.
Zagury
G. J.
2017
Comparative column testing of three reactive mixtures for the bio-chemical treatment of iron-rich acid mine drainage
.
Minerals Engineering
111
,
79
89
.
Raudsepp
M.
Pani
E.
2003
Application of Rietveld analysis to environmental mineralogy
. In:
Environmental Aspects of Mine Wastes, Short Course Series, 31
(
Jambor
J. L.
Blowes
D. W.
Ritchie
A. I.
, eds).
Mineralogical Association of Canada
, Québec City, QC, Canada, pp.
163
180
.
Skousen
J.
Zipper
C. E.
Rose
A.
Ziemkiewicz
P. F.
Nairn
R.
McDonald
L. M.
Kleinmann
R. L.
2017
Review of passive systems for acid mine drainage treatment
.
Mine Water and the Environment
36
,
133
153
.
Song
H.
Yim
G. J.
Ji
S. W.
Nam
I. H.
Neculita
C. M.
Lee
G. Y.
2012
Performance of mixed organic substrates during the treatment of acidic and moderate mine drainage in column bioreactors
.
Journal of Environmental Engineering ASCE
138
,
1077
1084
.
Su
P.
Granholm
K.
Pranovich
A.
Harju
L.
Holmbom
B.
Ivaska
A.
2012
Metal ion sorption to birch and spruce wood
.
Bioresources Journal
72
,
2141
2155
.
US EPA (Environmental Protection Agency)
2014
Reference Guide to Treatment Technologies for Mining-Influenced Water
.
EPA 542-R-14-001, Washington, DC, USA
, p.
94
.

Supplementary data