Testing the suitability of metallic iron for environmental remediation: Discoloration of methylene blue in column studies 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Btatkeu K. B.D.(a,b), Miyajima K.(c), Noubactep C.*(c,d), Caré S.(a) (a) Université Paris-Est, Laboratoire Navier (UMR 8205), CNRS, ENPC, IFSTTAR, F-77455 Marne-la-Vallée, France; (b) ENSAI/University of Ngaoundere, BP 455 Ngaoundere, Cameroon; (c) Angewandte Geologie, Universität Göttingen, Goldschmidtstraße 3, D - 37077 Göttingen, Germany; (d) Kultur und Nachhaltige Entwicklung CDD e.V., Postfach 1502, D - 37005 Göttingen, Germany. (*) Correspond author; e-mail: cnoubac@gwdg.de; Tel. +49 551 39 3191, Fax. +49 551 399379. Abstract A new method to correlate intrinsic reactivity and treatability efficiency of metallic iron (Fe0) was evaluated. A 2.0 mg L-1 methylene blue (MB) solution was used in gravity fed column experiments. The intrinsic reactivity of nine Fe0 materials (ten samples) was characterized using the EDTA test. Three commercial Fe0 materials ZVI1 (0.40 - 0.80 mm), ZVI9 (0.50 mm) and ZVI10 (0.45 - 0.55 mm) were tested in column experiments. A layer containing 100 g of Fe0 was sandwiched between 19.0 to 20.0 cm upper coarse sand (1.6 - 2.0 mm) and 8.0 cm lower fine sand (0.25 - 0.30 mm). 500 mL of the MB solution was daily filtered through each column for one month. Effluent solutions were characterized for MB and Fe concentrations. The columns were also characterized by the evolution of the hydraulic conductivity (k values). Results showed (i) quantitative MB removal (> 88 %) and (ii) limited Fe release for all three columns. After about 25 days, the Fe levels were constantly less than 1.0 mg L-1. The most significant difference was observed in the evolution of the k value and was attributed to the different material sorting. Less sorted ZVI1 exhibited the lowest initial k value (8.0 vs 43.0 mm min-1 for ZVI9 and ZVI10) and most significant permeability loss. Results confirmed the usefulness of the tested protocol as a reliable method to assess the efficiency of Fe0 materials in short term column experiments. Well-sorted Fe0 materials are recommended for long term efficient Fe0 filtration systems. 1 Keywords: Intrinsic reactivity, Methylene blue, Reactive filtration, Treatability efficiency, Zerovalent iron. 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 1 Introduction Metallic iron (Fe0) is a reactive material currently used for environmental remediation and safe drinking water provision [1-7]. Available Fe0 materials are obtained from various sources [8-12] and are supposed to satisfy design expectations [7,13-17]. However, there is no standard method to access the suitability of Fe0 materials for individual applications. As a rule, used materials are characterized: (i) by selected physical and chemical parameters (e.g. elemental composition, particle size, surface area) and (ii) for their ability to remove the specific species of interest (efficiency in treatability studies) [11,18,19]. However, each reactive material should be primarily characterized by its intrinsic reactivity. The ‘intrinsic reactivity’ is a material-dependent but system-independent qualitative trend. The ‘removal efficiency’ or ‘treatability efficiency’ is a system-dependent quantifiable parameter. For a reactive material like Fe0, the intrinsic reactivity should be correlated to the treatability efficiency. For activated carbons (an inert material) several indicators for the treatability efficiency have been introduced including the iodine number, the phenol number, the methylene blue number and the tannic acid number [20]. Appropriate treatability indicators for Fe0 are urgently needed. The intrinsic reactivity could be regarded as the most important characteristic in selecting Fe0 for designing filtration systems. Under specific operational conditions, intrinsic reactivity characterizes the kinetics of the process of iron oxidative dissolution in the absence of any mass transfer restrictions [21]. The extent of iron corrosion depends primarily on the solubility of iron under the operational conditions. For example, in the presence of some solutes (e.g. organic complexants), aqueous iron forms soluble chelates and Fe0 tend to dissolve at constant rate. Others solutes (e.g. PO43- ions) produce relatively insoluble chelates 2 which form a coating over the metal and tend to stifle iron corrosion [22]. Despite this general trend, one could not state that a Fe 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 0 specimen is not suitable for PO4-contaminated waters for example. The suitability of Fe0 for any polluted water depends on at least three interrelated operational parameters: (i) the Fe0 intrinsic reactivity, (ii) the nature and the extent of the contamination and (iii) the water flow velocity (contact time, filter thickness) [23]. Current approaches to design Fe0-based filtration systems suggest that the intrinsic reactivity is not optimised in respect of the most site-specific effective material. The reasons are that the importance of intrinsic reactivity and its relationship with other operational parameters are not yet fully appreciated and/or understood in engineering practice. Relevant operational parameters include: (i) hydraulic characteristics, (ii) shape, size and surface characteristics used Fe0 material and non-expansive additives (e.g. gravel, pumice, sand), (ii) compactness or porosity of the filter [24-26]. In recent years, significant progress has been made in characterizing the intrinsic reactivity of Fe0 materials in water by H2 evolution [27-30] and iron dissolution in EDTA (EDTA test) [11,21,31,32]. Achieved results were contrasted with results from the removal of uranium [31,33], discoloration of methylene blue [34] and removal of arsenic [35]. Available data validate the suitability of the EDTA test to characterize the intrinsic reactivity of Fe0 materials. The present study extends this effort by relating the intrinsic reactivity, the Fe0 particle size/shape and the hydraulic conductivity of Fe0 filters for three selected materials. A total of 9 Fe0 materials making up 10 samples (ZVI1 through ZVI10) have been characterized by the EDTA test. Three selected Fe0 materials (ZVI1, ZVI9, ZVI10) were then used in column studies to characterize the evolution of (i) Fe release, (ii) methylene blue (MB) discoloration and (iii) hydraulic conductivity. A 2.0 mg L-1 MB solution was used as an indictor for the impact of iron corrosion on the filtration process. 3 2 Background of the experimental methodology 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 The suitability of a pure Fe0 layer (100 % Fe0) in a sand column to characterize the intrinsic reactivity of Fe0 materials arises mainly from the historical observation by Mitchell et al. [36] that (i) sand is a good adsorbent for methylene blue (MB) and (ii) iron oxide coated sand is less adsorptive than pure sand. Accordingly, sandwiching a layer of different Fe0 in a standard sand column in parallel studies and comparing individual system’s responses for MB discoloration could be an efficient tool to access Fe0 intrinsic reactivity. Using a 100 % Fe0 layer is a tool to shorten experimental duration as less room is left for volumetric expansive iron corrosion [37,38]. In fact, a 100 % Fe0 bed has been demonstrated efficient for contaminant removal but not sustainable [1,6,39]. In this study a 19.1 to 20.0 cm coarse sand layer (H1-sand, Tab. 1) is used before the pure Fe0 layer and a 8.0 cm fine sand layer (H2-sand) thereafter. It is expected that MB will be undisturbed discoloured in H1-sand layer at the beginning of the experiment. However, a disturbance of the flow regime due to expansive iron corrosion will hindered the extent of MB discoloration in the whole column and in the H2-sand layer in particular. Additionally decreased adsorptive efficiency due to in-situ iron oxide coating will impair MB discoloration. Thus, both a decreased of the hydraulic conductivity (permeability loss) and an accelerated MB breakthrough could be observed within some weeks. For materials of comparable particle size and shape, the most reactive a material, the lower the extent of MB discoloration and the more rapid the clogging (Assumption 1). The used methodology for the investigation of the impact of Fe0 characteristics (e.g. intrinsic reactivity, shape, size, sorting) on the extent of (i) MB discoloration and (ii) decrease of hydraulic permeability comprises testing the validity of Assumptions 1 by following the MB discoloration and permeability loss in the presence of tested materials. In situ generated iron corrosion products are adsorbed onto the sand surface, worsening its capacity of MB adsorption [36]. MB breakthrough is facilitated in systems with more reactive Fe0. 4 3 Material and methods 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 3.1 Solutions 3.1.1 Methylene blue The used methylene blue (MB - Basic Blue 9 from Merck) was of analytical grade. The working solution was 2.0 mg L-1. The solutions were prepared by diluting a 1000 mg L-1 stock solution. The stock solution was prepared weekly by dissolving accurately weighted MB in tap water. MB was chosen in this study because of its known weak adsorption onto iron oxides [36,40]. The used concentration 2.0 mg L-1 or 6.3 μM corresponds to the concentration range of natural waters (MB as model micro-pollutant). 3.1.2 Iron A standard iron solution (1000 mg L-1) from Baker JT® was used to calibrate the spectrophotometer used for analysis. All other chemicals used were of analytical grade. In preparation for spectrophotometric analysis, ascorbic acid was used to reduce FeIII in solution to FeII. 1,10 orthophenanthroline (ACROS Organics) was used as reagent for FeII complexation. Other chemicals used in this study included L(+)-ascorbic acid and L-ascorbic acid sodium salt. 3.2 Solid materials 3.2.1 Metallic iron (Fe0) A total of 10 samples (ZVI1 through ZVI10) from 9 Fe0 materials were tested. The materials were obtained from various sources, in different forms and grain sizes. The main characteristics of these materials are summarised in Tab. 2. Fig. 1 depicts micrographs of 500 mg of each sample using a portable camera (Keyence VHX-500F). No information about manufacturing processes (e.g., raw material, heat treatment) was available to assist with subsequent data interpretation. The average elemental composition of the materials as specified by the suppliers are well-documented in the literature [25,26,41] and are not repeated here. 5 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 It is the objective of this study to characterize the reactivity of tested materials and compare the efficiency of three of them for the discoloration of methylene blue in column studies. A part from the Rheinfelden material (iPutcec GmbH & Co) which was tested in two fractions, all other materials were used in their typical state and form ('as received' state) in which they might be used for field applications. The Rheinfelden material was tested in two fractions: (i) as received (ZVI4) and (ii) the fraction between 0.40 and 0.80 mm (ZVI1). ZVI1 was used to reduce variability in grain size [26] and compare the efficiency to that of ZVI9 and ZVI10 in column studies. ZVI1, ZVI9 and ZVI10 used in column studies are available as fillings with a particle size between 0.40 and 0.8 mm (Tab. 2). ZVI10 was not quantitatively available to enable own granulometric analysis. The granulometric distribution reported by Gottinger [41] was used for the presentation (Tab, 2). ZVI9 is the best sorted material, ZVI10 is better sorted than ZVI1 (Fig. 1). 3.2.2 Sand The used sand was a natural material from Fontainebleu (France). Fontainebleu sand was available in two fractions: (i) d ≤ 0.5 mm and (ii) 0.5 ≤ d (mm) ≤ 5.0. The first fraction was sieved and particles ranging between 0.250 and 0.300 mm were retained for the first sand layer (fine sand - H2-sand, Tab. 1). The second fraction was sieved and particles ranging between 1.6 and 2.0 mm were retained for the second sand layer (coarse sand - H1-sand). Sand was used as an adsorbent because of its worldwide availability and its use as admixing agent in Fe0/H2O systems [2,7,13,15]. 3.3 Experimental procedure 3.3.1 EDTA test Iron dissolution was initiated by the addition of 0.1 g of each Fe0 material to 50 mL of a 2.0 mmol L-1 EDTA solution. Each reaction was run for ≤ 96 hours (4 days) using narrow 70 mL glass beakers to hold the solutions [11,31]. The reacting samples were left undisturbed on the 6 laboratory bench for the duration of experimental period and were shielded from direct sunlight to minimize Fe 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 III-EDTA photodegradation [42]. Iron dissolution was used to characterize the reactivity of tested Fe0 [11,31]. 3.3.2 MB discoloration Plexiglas columns of 2.6 cm inner diameter and 50 cm length were used. The column was mostly packed with sand: 8.0 cm at the bottom (H2-sand) and 19.1 to 20.0 cm at top (H1-sand) of the reactive layer. A reactive layer (100 % Fe0) with 100 g Fe0 was sandwiched between H1-sand and H2-sand (Fig. 2, Tab. 1). The depth of the reactive layer was 4.5 to 5.4 cm (Hiron - Fig. 2) for the three materials. The columns were intermittently charged with a gravity driven 2.0 mg L-1 MB solution. Five filtration events were performed each week (daily from Monday to Friday). Each filtration event used 500 mL of the MB solution. This was to mimic the intermittent filtration with household filters for the daily water need [43]. The whole experimental duration was one month. The average time needed for the filtration of the first three 50 mL (150 mL in total) was used to calculate the flow velocity at each date. The whole effluent for each filtration event was collected and analysed for MB. At certain time intervals the iron concentration was determined. The room temperature during the experiments was 22 ± 2°C, and the initial pH value of the 2.0 mg L-1 MB was 6.8 ± 0.2. 3.5 Analytical methods 3.5.1 Solution concentrations MB and aqueous iron concentrations were determined by a Cary 50 UV-Vis spectrophotometer (Perkin Elmer Lambda 10 UV/Vis) at a wavelength of 664.5 nm and 510.0 nm respectively. Cuvettes with 1.0 cm light path were used. The iron determination followed the 1,10 orthophenanthroline method [44,45]. The spectrophotometer was calibrated for MB concentrations ≤ 2.5 mg L-1 and iron concentrations ≤ 10.0 mg.L-1. The pH value was measured by combined glass electrodes (WTW Co., Germany). 7 3.5.2 Scanning electron microscope (SEM) 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 Particle morphology was investigated with a cold field emission SEM (type Hitachi S- 34000N). The observations enable a characterization of the morphology of all materials used in column experiments (sand, ZVI1, ZVI9 and ZVI10). 3.6 Expression of experimental results 3.6.1 Kinetics of Fe0 oxidative dissolution (kEDTA value) Upon immersion in a EDTA solution, Fe0 is oxidative dissolved and the aqueous concentration of iron ([Fe]) linearly varies as a function of time (Eq. 1) before the solution approaches saturation [11,31]. [Fe]t = kEDTA t + b (1) For each Fe0 material the time-dependant linear evolution of aqueous iron concentration is used to characterize the intrinsic reactivity. The linear gradient (‘kEDTA’ in Eq. 1) representing the rate of iron dissolution. The kEDTA values were calculated using Origin 6.0. The b value, [Fe] at t = 0, is a reflect of the amount of atmospheric corrosion present at fines at the Fe0 surface [11,31]. 3.6.2 Discoloration efficiency (E value) In order to characterize the magnitude of tested systems for MB discoloration, the discoloration efficiency (E) was calculated (Eq. 2). After the determination of the residual MB concentration (C), the corresponding percent MB discoloration (E value) was calculated as: E = [1 - (C/C0)] * 100% (2) where C0 is the initial aqueous MB concentration (2.0 mg L-1), while C gives the MB concentration after the experiment. 3.6.3 Relative permeability Saturated hydraulic conductivity K (e.g. mm.min-1) can be expressed as: K = (Q/A)*(ΔL/ΔH) (3) 8 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 where Q is volumetric flow rate, A is column cross-sectional area, L is distance from the inlet of column and ΔH is the hydraulic head. In general, exploiting Eq. (3), experimental data are used to evaluate the hydraulic gradient (ΔH/ΔL) at each time and to deduce the saturated hydraulic conductivity (K). However, estimating the initial saturated hydraulic conductivity (K0) at the start of the experiment is usually sufficient to characterize the evolution of the system. The system’s permeability is then characterized by the time-dependant relative hydraulic conductivity (Eq. 4): k = K/K0 (4) The K value in this study corresponds to the solution flow velocity through the columns. The flow velocity was operationally defined as the time necessary for the filtration of 50 mL of the 2.0 mg L-1 MB solution. 4. Results and Discussion 4.1 Fe0 characterization 4.1.1 Photomicrography Micrographs of the tested samples evidenced differences in shapes and sizes (Fig. 1). The grain size distribution and the availability of fines are also evidenced. Relevant information from Fig. 1 can be summarized as: (i) ZVI8 and ZVI9 are the sole well-sorted materials with regular shapes, (ii) ZVI3, ZVI6, ZVI7 and ZVI10 are covered with fines (atmospheric corrosion products), (iii) the proportion of fines in ZVI6 is relatively large. It can be further seen that sieving ZVI4 to ZVI1 has resulted in a sample roughly comparable to ZVI10 in its particle size distribution. ZVI1, ZVI9, ZVI10 and used sand fractions were further characterized by scanning electron microscope (SEM). The following characteristics of Fe0 materials used in column studies should be pointed out: (i) ZVI1 and ZVI10 are similar in the particle size distribution, (ii) ZVI1 and ZVI10 are poorly sorted and (iii) ZVI9 is a relatively well-sorted material. 4.1.2 Scanning electron microscope (SEM) 9 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 SEM images detailing the microstructures of sand, ZVI1, ZVI9 and ZVI10 particles are shown in Fig. 3. The coarse fraction of Fontainebleu sand was sieved and the fraction 0.400 - 0.800 was used (the same fraction like ZVI1). These observations confirm that ZVI9 is roughly regular in shape while ZVI1, ZVI10 and sand particles are irregular. Sand is more regular in shape that ZVI1 and ZVI10. Micrograph’s observations are confirmed by SEM images. 4.2 EDTA batch test 4.2.1 All tested materials The calculated dissolution rates (kEDTA) for the ten Fe0 samples are displayed in Tab. 3 and vary from 13 μg h-1 for ZVI8 to 37 μg h-1 for ZVI10. The least reactive material (ZVI8 from Würth) was used as negative reference like in previous studies [32-34]. The most reactive material were ZVI9 (Ferblast) and ZVI10 (Peerless). The kEDTA values for all remaining materials varied from 23 μg h-1 (ZVI4) to 32 μg h-1 (ZVI3). The overall increasing order of reactivity after the kEDTA value is: ZVI8 < ZVI4 = ZVI1 = ZVI6 < ZVI5 = ZVI7 < ZVI2 < ZVI3 < ZVI9 = ZVI10. It is important to recall that this classification does not account for the presence of iron corrosion products reflected by the b values (Tab. 3). From Tab. 3 it is clear that ZVI6 exhibited the largest b value. This is consistent with the fact that ZVI6 was physically covered by fines of rusted iron (Fig. 1). The presence of atmospheric corrosion products is the main reason why a graphical comparison of materials is not suitable [11,31]. It has been shown that the EDTA test is not suitable for powdered Fe0 materials and for Fe0 samples containing a high proportions of fines [11]. It should be kept in mind that: (i) Fe0 materials tested here are (among) the most widely used worldwide (e.g. Connelly, Peerless, Rheinfelden); (ii) there is actually not standard protocol to test the intrinsic reactivity of Fe0 materials. From these two factors, it appears that this study reliably characterizes for the first time a large array of commercial materials and could 10 be regarded as starting point for future works aiming as establishing standard protocol for testing Fe 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 0 materials. Based on the established order of reactivity given above, the least reactive material (ZVI1 - ZVI8 was tested as negative reference) and the both most reactive materials (ZVI9 and ZVI10) were selected for column tests. The both most reactive material differ in their sorting (Fig. 1 and Fig. 3). 4.2.2 Materials for column studies Fig. 4 summarizes the results of the EDTA test for materials used in column studies. It is seen that ZVI10 (kEDTA = 37 ± 2 μg h-1) and ZVI9 (kEDTA = 36 ± 3 μg h-1) are very closed in their intrinsic reactivity and essentially more reactive than ZVI1 (kEDTA = 24 ± 1 μg h-1). Upon immersion in water Fe0 dissolves to Fe2+ (Eq. 5). Under the experimental conditions, FeII is further oxidized to FeIII (Eq. 6) which solubility at pH > 4.5 is very low. Using a 2.0 mM EDTA solution to form stable complexes with EDTA (Eq. 7) was demonstrated a powerful tool to assess the reactivity of Fe0 [11,31]. Fe0 + 2 H+ ⇒ Fe2+ + H2 (5) Fe2+ + ½ O2 + H2O ⇒ Fe3+ + 2 HO- (6) Fe3+ + EDTA ⇒ FeEDTA3+ (7) Given the higher stability of FeIII-EDTA relative to FeII- EDTA, ascorbic acid was identified a powerful agent for the reduction of the FeIII-EDTA prior to Fe determination using the 1,10 orthophenanthroline method [31]. The data in Tab. 4 suggests that the reactivity difference can be attributed to the particle size distribution (material sorting) (Fig. 1). Material intrinsic reactivity is generally influenced by practical considerations including particle size, chemical composition, production and storage conditions [26,46,47]. The single information accessible for the three tested materials is their mean particle diameter (φ value - Tab. 4). While the φ value is 0.50 mm for ZVI10 [41] and ZVI9, its value for ZVI1 is 0.60 mm. The fact that materials with finer particle size are more 11 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 reactive than coarser materials is the rationale for the use of nano-scale particles in water treatment [48,49]. However, it should be recalled that the particle size alone is not an indicator for the intrinsic reactivity [18,31,19]. Rigorously, if the particle size distribution of ZVI9 was not given by the supplier, a researcher could have sieved all three materials and used the fraction 0.40 to 0.80 mm as one fraction. In other words, the difference in particle size of used materials is minimal (φ value). However, the assertion that ‘finer particle size are more reactive than coarser materials’ is operationally maintained in this study [11,26,31]. Remember that the EDTA test is in essence a reactivity test. Characterizing the same material available in different particle sizes can be regarded as validation of the experimental protocol. This has already been performed in previous work [11,31]. The Rheinfelden Fe0 (iPutec GmbH & Co KG) depicted kEDTA values decreasing from 61 ± 6 μg h-1 to 27 ± 1 μg h-1 when the particle size fraction increased from 0.315 - 0.500 mm to 1.0 - 2.0 mm. In the present work, the lack of significant difference between the reactivity of ZVI1 (0.400 - 0.800 mm) and ZVI4 (0.300 - 2.0 mm) is attributed to the proportion of smaller size particles (0.300 - 0.400 mm) in ZVI4 (Fig. 1). This work correlates ‘intrinsic reactivity’ and ‘treatability efficiency’ for MB discoloration. The terms ‘reactivity’ and ‘efficiency’ are currently confusing in the literature. They are almost always randomly interchanged. 4.3 MB column test The EDTA test has revealed the following reactivity order for tested materials: ZVI10 ≅ ZVI9 >> ZVI1. The present section characterizes the efficiency of tested materials in term of (i) iron release, (ii) MB discoloration and (iii) permeability loss. 4.3.1 Iron release Fig. 5 summarizes the results of iron release by tested materials during the experiments (30 days). It is seen that ZVI9 was the most efficient material followed by ZVI10. ZVI1 was the least efficient material. The cumulative mass of iron in the collected effluent was 18.2 mg for ZVI9, 15.9 mg for ZVI10 and 7.8 mg for ZVI1 (Tab. 4). Accordingly, the efficiency order for 12 Fe release corresponds to the reactivity order after the EDTA test: ZVI9 ≅ ZVI10 >> ZVI1. It is important to notice that the non uniform evolution of Fe release during the first 15 to 20 days is a argument against characterizing Fe 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 0 reactivity and efficiency short term experiments [32,33,35]. The observed differential behaviour can be attributed to the presence of fines as ZVI9 and ZVI10 content more fines than ZVI1. 4.3.2 MB discoloration Fig. 6 summarizes the results of MB discoloration by tested materials during the experiments. Fig. 6a depicts no clear trend in the order of efficiency of the tested materials. For example: (i) for t < 12 d, ZVI9 is clearly the less efficient material; (ii) at day 15, the order of efficiency is ZVI1 > ZVI9 > ZVI10, (iii) at day 25, the corresponding order of efficiency is ZVI10 > ZVI1 > ZVI9. The lack of a clear trend suggest that no (pseudo-) steady state was yet established in the kinetics of Fe0 oxidation coupled with MB discoloration (or its disturbance). In other words, 30 days is too short to graphically differentiate the extent of MB discoloration by tested Fe0 materials using the E value alone. Fig. 6b depicts the time dependant evolution of the mass of MB in the effluent. This is a reflect of the kinetics of the process (iron corrosion) disturbing MB adsorptive discoloration by sand. Fig. 6b clearly shows that MB removal was less efficient in the ZVI9 systems and very comparable in the two other systems (ZVI1 and ZVI10). The derived order of efficiency from Fig. 6b (t > 12 d) is: ZVI1 > ZVI10 > ZVI9. Calculations confirmed that, after 30 days, only 2.8 % of the cumulative mass of MB was collected in the effluent for ZVI1 against 5.6 % for ZVI10 and 11.6 % for ZVI9 (Fig. 6b and 6c). Fig. 6c depicts the time dependant evolution of the cumulative mass of MB in the effluent. The trend from Fig. 6b is confirmed with the additional information that no significant difference can be documented for ZVI1 and ZVI10 for t < 25 d. The graphical order of efficiency from Fig. 6 seems to contradict the order based on Fe release (§ 4.3.1) which is in tune with the reactivity order after the EDTA test. 13 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 Considering the historical work of Mitchell et al. [36] demonstrating that iron oxide coated sand is a poorer MB adsorbent than pure sand, these results suggest that the most reactive system is the one exhibiting the lowest efficiency for MB discoloration (e.g. ZVI9). Considering this essential aspect, it appears that ZVI1 is the less reactive material as found by Fe release (section 4.3.1) and EDTA test (§ 4.2). Accordingly, Assumption 1 is validated with regard to MB discoloration and Fe release. 4.3.3 Permeability loss Fig. 7 summarizes the results of the evolution of the hydraulic conductivity (k in mm min-1) as a function of the time. A monotone decrease of the k value is observed with the progress of the filtration (Fig. 7a). Fig. 7a also shows that the initial k value (k0 - Tab. 4) was essentially lower for ZVI1 than for ZVI10 and ZVI9. This behaviour is primarily attributed to the material sorting [50-54]. In fact, a packed bed of a better sorted granular material exhibits a greater permeability. In better sorted beds (ZVI10, ZVI9) the pore spaces are open. In the comparatively poorly sorted ZVI1 bed fine grains occupy the pore spaces between coarser grains [50,52]. Fig. 7b shows that the evolution of the relative permeability (k in %) was very closed for ZVI10 and ZVI9, confirming that the pore volume in both systems are occupied with a similar kinetics (kEDTA values), namely the kinetics of the process of iron hydroxide production. A more rapid porosity loss is observed in the ZVI1 column. This can only be attributed to the poorer sorting, as ZVI1 is significantly less reactive than ZVI10 and ZVI9 (§ 4.2, 4.3.1 and 4.3.2). In fact, precipitation of iron hydroxides from Fe0 corrosion causes cementation of the Fe0 grains and acts to fill the pore spaces (porosity) and interconnectivity of the spaces. Both processes result in more rapid decrease of k values (permeability loss) in the ZVI1 system where Fe0 particles are closer to each other (smaller pore radius). The shape of the particles also significantly influences the evolution of the permeability [53,54]. According to Hayati et 14 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 al. [54] spherical particles yield a better porosity than cylindrical ones. ZVI9 is almost spherical, ZVI1 and ZVI10 are more cylindrical but ZVI1 is the least sorted material (Fig. 1). The discussion until now has confirmed that material particle size, material sorting and material geometry all significantly influence the hydraulic conductivity [50,52-54]. In Fe0 beds, in-situ generated iron hydroxides tend to cement initial particles and decrease both porosity and permeability. The present study has presented a reliable method to correlate the intrinsic reactivity and the efficiency of Fe0 materials using MB discoloration in column studies (assumption 1 is validated by experimental results). Future research should focus on how the characteristics of Fe0 and admixing additives (e.g. gravel, MnO2, pumice, sand) impact system’s sustainability. Pure Fe0 beds as tested here are not sustainable [1,39]. Admixing additives are natural materials which are rarely uniform in geometry. Accordingly, when mixed with the same additive (e.g. sand), the sustainability of the resulted Fe0/sand systems should be individually tested/discussed. In particular, not the most reactive material/system should be constantly sought, but the material with the most appropriate efficiency. For example, ZVI1 could be more suitable for groundwater remediation and ZVI9 and ZVI10 for wastewater treatment. Alternatively, ZVI1 mixed with a certain material A (size, form, shape) could be suitable for drinking water and the mixture with a second material suitable for wastewater treatment. 4.4 Discussion The thermodynamic instability of Fe0 (E0 = -0.44 V) in water (E0 = 0.00 V) does not provide any information on the kinetics of the corrosion process, which defines the intrinsic reactivity of individual materials. Fe0 reactivity is influenced by material-dependant considerations such as (i) the material elemental composition, (ii) its particle size, (iii) its production and storage conditions. Regarding the test conditions, used operational parameters should be rationally selected to be relevant for field conditions. Relevant operational parameters include (i) used Fe0 amounts, (ii) used Fe0:solution ratios, (iii) used flow rate or mixing intensities, (iv) nature 15 of model contaminants and (v) solution pH. The present study and related works have demonstrated the suitability of MB to enable the differentiation of ‘Fe 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 0 efficiency’ and ‘Fe0 reactivity’, regardless of whether these differences are due to elemental composition, particle size or production and storage conditions. The reactivity results for nine commercial Fe0 materials including the most used worldwide (e.g. Peerles, Connelly, Rheinfelden) are presented for the first time and could help to comparatively discuss results from independent studies. The consistency among results from two different tests (EDTA and MB) indicates that a standard protocol is urgently needed for the characterization of existing and newly manufactured materials. In particular, material selection for each application should be justified by the efficiency under the specific conditions. Despite 20 years intensive research, the intrinsic reactivity of Fe0 material has not been properly addressed. Because of this technical deficit a real race for the most reactive material was initiated. Resulting materials included bimetallic particles [55], nanoscale particles [48] and some proprietary materials like CIM (Composite Iron Matrix) [1,39], SIM (Sulfur Modified Iron) [56]. However, not always the most reactive material is needed but the most appropriate one. As an example, ZVI8 which is the least reactive material in this study (EDTA test) has been shown more efficient than 3 other more reactive Fe0 materials for As removal under conditions where rapid dissolution kinetics favoured Fe0 ‘passivation’ [35]. The most appropriate material for a given situation is the one producing enough contaminant scavengers per unit time to achieve the design goal. That is the material which intrinsic reactivity matches the field situation (e.g. water flow rate, nature of the contaminant) the most. 4.4.1 Efficiency and reactivity of Fe0 A rigorous differentiation between ‘reactivity’ and ‘efficiency’ is of crucial importance for the future of the use of Fe0 materials for water treatment and environmental remediation. In fact, 16 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 many reported discrepancies originated from the ill-definition of these two terms. Three examples will be given for illustration: First, Bi et al. [57] found ‘paradoxical’ that a 85:15 Fe0:sand (w/w) mixture was more efficient for TCE removal than a pure Fe0 system (100 % Fe0). In both systems, the Fe0 reactivity is the same (invariable) but the efficiency of the Fe0/sand system is increased thanks to the presence of non-expansive sand (delayed clogging or extended Fe0 depletion) [37]. Second, For the same Fe0, Ruhl et al. [30] reported on differential averaged corrosion rates in shorter and longer columns. The corrosion rate, assessed by the extent of H2 generation, was reported higher in shorter columns. It is obvious that the authors confounded the efficiency of the systems for H2 release and the iron corrosion rate. The corrosion rate, as an invariable intrinsic characteristic of the used Fe0 material. It is the same for all columns (same Fe0 material). The measured volume of H2 depends on the extends of hindrance within individual columns. Generated H2 must migrate through the Fe0 column and only the escaped fraction is measured. Disregarding the nature of hindrances in the Fe0 columns, the results of Ruhl et al. [30] are better explained by the deep-bed nature of filtration on Fe0 columns. It is therefore traceable that less H2 escapes from a higher column. Third, Ruhl and Jekel [26] found out that the grain size distribution of a Fe0 material affects “the porosity, the pore geometry and the reactivity”. It is evident that ‘reactivity’ here is referred to ‘removal efficiency’ as it is further stated that ‘column tests showed that all fractions achieved good TCE removal with a slight advantage for smaller grains’. However, assuming cylindrical porous structure, the pore radius for smaller grains is narrow and should be rapidly filled/clogged by the more reactive small particles. In other words, reactive zones with smaller grain sizes are more susceptible to clogging. The discussion of Ruhl and Jekel [26] could be optimised. For the discussion herein, however, it is sufficient to consider that purposefully distinguishing between ‘reactivity’ and ‘efficiency’ would have impacted the experimental design and/or enabled a stronger discussion with the presented material. 17 The proper consideration of the expansive nature of iron corrosion suggests that a Fe0-based filter should be regarded as a classical self-filtration system [58]. The processes of (i) Fe 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 0 dissolution, (ii) migration and rearrangement of in-situ generated Fe species and (iii) removal of inflowing foreign species (including contaminants) within the porous medium are central to the understanding of pore clogging and filter design. Presently, the clogging process of Fe0 filters is still poorly understood [16]. The present work and related studies [59-61] suggest that this issue was even not properly addressed. 5 Concluding remarks The suitability and the reliability of a 2.0 mg L-1 methylene blue (MB) solution to correlate Fe0 intrinsic reactivity and Fe0 efficiency for MB discoloration in column studies was established. This work shows conclusively that the MB efficiency test yields reliable results to investigate the process permeability loss in Fe0 columns. The major recommendation is that well-sorted materials should be used for sustainable (long term permeable) Fe0 filtration systems. In particular, bulk Rheinfelden materials (0.3 to 2.0 mm) should be fractionated in several fractions, each fraction being probably appropriate for a different goal, e.g. coarser fraction for pre-filtration or O2 scavenging. Further investigations should focus at characterizing the effect of particle shape and size of Fe0 and non-expansive additives (e.g. gravel, pumice, sand) on the process of permeability loss. Virtually, every aspect of contaminant removal in Fe0 beds could be addressed by a purposeful modified protocol. One of the most important challenges is now to use this basic scientific knowledge to improve the efficiency of the ‘Fe0 technology’ as a whole [59-61]. The acquirement of reliable results and their dissemination will also impact public awareness and acceptance this promising technology. Upon a science-based acceptance, networking between scientists, industrials, end- users and governmental authorities will be promoted in order to ensure successful large scale implementation of the promising ‘Fe0 technology’ for groundwater remediation, wastewater treatment and safe drinking water provision. 18 Acknowledgements 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 For providing the iron materials investigated in this study the authors would like to express their gratitude to iPutec GmbH (Rheinfelden, Germany), Ann-Marie Gottinger from Mainstream Water Solutions Inc. (Regina, Canada) and Paolo S. Calabrò from the University Calabria (Italy). Mohammed Saad (Leesu, Ecole des Ponts ParisTech, Marne La Vallée), Mehmet A. Oturan and Hugo Gonzales (LGE, Institute IFI, Marne la Vallée) are thanked for their help, their grateful discussion and the possibility to use their spectrophotometer. Tanja Stegemann (Geoscience Museum – University of Göttingen) is acknowledged for the micrographs. This work was partly supported by the “Service de Coopération et d'Action Culturelle” (SCAC) from the France Embassy in Yaoundé (Cameroun). References [1] A. Hussam, Contending with a development disaster: SONO filters remove arsenic from well water in Bangladesh, Innovations 4 (2009) 89–102. [2] S. Comba, A. Di Molfetta, R. Sethi, A Comparison between field applications of nano-, micro-, and millimetric zero-valent iron for the remediation of contaminated aquifers, Water Air Soil Pollut. 215 (2011) 595–607. [3] M. Gheju, Hexavalent chromium reduction with zero-valent iron (ZVI) in aquatic systems, Water Air Soil Pollut. 222 (2011) 103–148. [4] D.E. Giles, M. Mohapatra, T.B. Issa, S. Anand, P. Singh, Iron and aluminium based adsorption strategies for removing arsenic from water, J. Environ. Manage. 92 (2011) 3011–3022. [5] C. Noubactep, Metallic iron for safe drinking water worldwide, Chem. Eng. J. 165 (2010) 740–749. [6] C. Noubactep, Metallic iron for safe drinking water production, Freiberg Online Geol. 27 (2011) 38 pp, ISSN 1434-7512. (www.geo.tu-freiberg.de/fog). 487 19 [7] ITRC (Interstate Technology & Regulatory Council) Permeable Reactive Barrier: Technology Update. PRB-5. Washington, D.C.: Interstate Technology & Regulatory Council, PRB: Technology Update Team. www.itrcweb.org (2011) (access: 29.04.2012). 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 [8] R.L. Landis, R.W. Gillham, E.J. Reardon, R. Fagan, R.M. Focht, J.L. Vogan, An examination of zero-valent iron sources used in permeable reactive barriers. 3rd International Containment Technology Conference (10-13 June 2001), Florida State University, Tallahassee. Orlando, FL. (2001) 5 pages. [9] R. Miehr, G.P. Tratnyek, Z.J. Bandstra, M.M. Scherer, J.M. Alowitz, J.E. Bylaska, Diversity of contaminant reduction reactions by zerovalent iron: Role of the reductate, Environ. Sci. Technol. 38 (2004) 139–147. [10] O.X. Leupin, S.J. Hug, Oxidation and removal of arsenic (III) from aerated groundwater by filtration through sand and zero-valent iron, Wat. Res. 39 (2005) 1729–740. [11] C. Noubactep, T. Licha, T.B. Scott, M. Fall, M. Sauter, Exploring the influence of operational parameters on the reactivity of elemental iron materials, J. Hazard. Mater. 172 (2009) 943–951. [12] M. Gheju, I. Balcu, Hexavalent chromium reduction with scrap iron in continuous-flow system. Part 2: Effect of scrap iron shape and size, J. Hazard. Mater. 182 (2010) 484– 493. [13] G. Bartzas, K. Komnitsas, I. Paspaliaris, Laboratory evaluation of Fe0 barriers to treat acidic leachates, Miner. Eng. 19 (2006) 505–514. [14] G. Bartzas, K. Komnitsas, Solid phase studies and geochemical modelling of low-cost permeable reactive barriers, J. Hazard. Mater. 183 (2010) 301–308. [15] L. Li, C.H. Benson, Evaluation of five strategies to limit the impact of fouling in permeable reactive barriers, J. Hazard. Mater. 181 (2010) 170–180. 20 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 [16] A.D. Henderson, A.H. Demond, Impact of solids formation and gas production on the permeability of ZVI PRBs, J. Environ. Eng. 137 (2011) 689–696. [17] S.-W. Jeen, R.T. Amos, D.W. Blowes, Modeling gas formation and mineral precipitation in a granular iron column, Environ. Sci. Technol. 46 (2012) 6742–6749. [18] R.A. Crane, T.B. Scott, Nanoscale zero-valent iron: Future prospects for an emerging water treatment technology, J. Hazard. Mater. 211–212 (2012) 112–125. [19] C. Noubactep, S. Caré, R.A. Crane, Nanoscale metallic iron for environmental remediation: prospects and limitations, Water Air Soil Pollut. 223 (2012) 1363–1382. [20] Q.-G. Chang, W. Zhang, W.-X. Jiang, B.-J. Li, W.-C. Ying, W. Lin, Efficient micro carbon column rapid breakthrough technique for water and wastewater treatability studies, Environ. Progr. 26 (2007) 280–288. [21] E.M. Pierce, D.M. Wellman, A.M. Lodge, E.A. Rodriguez, Experimental determination of the dissolution kinetics of zero-valent iron in the presence of organic complexants, Environ. Chem. 4 (2007) 260–270. [22] V. Tanboonchuy, N. Grisdanurak, C.H. Liao, Background species effect on aqueous arsenic removal by nano zero-valent iron using fractional factorial design, J. Hazard. Mater. 205-206 (2012) 40–46. [23] C. Noubactep, A. Schöner, P. Woafo, Metallic iron filters for universal access to safe drinking water, Clean: Soil, Air, Water 37 (2009) 930–937. [24] M. Kubare, J. Haarhoff, Rational design of domestic biosand filters, J. Water Supply: Res. Technol. – AQUA 59 (2010) 1–15. [25] A.S. Ruhl, N. Ünal, M. Jekel, Evaluation of two-component Fe(0) fixed bed filters with porous materials for reductive dechlorination, Chem. Eng. J. 209 (2012) 401–406. [26] A.S. Ruhl, M. Jekel, Impacts of Fe(0) grain sizes and grain size distributions in permeable reactive barriers, Chem. Eng. J. 213 (2012) 245–250. 21 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 [27] J.E. Reardon, Anaerobic corrosion of granular iron: Measurement and interpretation of hydrogen evolution rates, Environ. Sci. Technol. 29 (1995) 2936–2945. [28] J.E. Reardon, Zerovalent irons: Styles of corrosion and inorganic control on hydrogen pressure buildup, Environ. Sci. Tchnol. 39 (2005) 7311–7317. [29] J.E. Reardon, R. Fagan, J.L. Vogan, A. Przepiora, Anaerobic corrosion reaction kinetics of nanosized iron, Environ. Sci. Technol. 42 (2008) 2420–2425. [30] A.S. Ruhl, A. Weber, M. Jekel, Influence of dissolved inorganic carbon and calcium on gas formation and accumulation in iron permeable reactive barriers, J. Contam. Hydrol. 142–143 (2012) 22–32. [31] C. Noubactep, G. Meinrath, P. Dietrich, M. Sauter, B. Merkel, Testing the suitability of zerovalent iron materials for reactive walls, Environ. Chem. 2 (2005) 71–76. [32] C. Noubactep, Characterizing the reactivity of metallic iron in Fe0/EDTA/H2O systems with column experiments, Chem. Eng. J. 162 (2010) 656–661. [33] C. Noubactep, Characterizing the reactivity of metallic iron in Fe0/UVI/H2O systems by long-term column experiments, Chem. Eng. J. 171 (2011) 393–399. [34] C. Noubactep, Characterizing the reactivity of metallic iron upon methylene blue discoloration in Fe0/MnO2/H2O systems, J. Hazard. Mater. 168 (2009) 1613–1616. [35] C. Noubactep, Characterizing the reactivity of metallic iron in Fe0/As-rock/H2O systems by long-term column experiments, Water SA 38 (2012) 511–517. [36] G. Mitchell, P. Poole, H.D. Segrove, Adsorption of methylene blue by high-silica sands, Nature 176 (1955) 1025–1026. [37] C. Noubactep, S. Caré, F. Togue-Kamga, A. Schöner, P. Woafo, Extending service life of household water filters by mixing metallic iron with sand, Clean – Soil, Air, Water 38 (2010) 951–959. 22 [38] C. Noubactep, S. Caré, K.B.D. Btatkeu, C.P. Nanseu-Njiki, Enhancing the sustainability of household Fe 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 0/sand filters by using bimetallics and MnO2, Clean – Soil, Air, Water 40 (2012) 100–109. [39] A. Hussam, A.K.M. Munir, A simple and effective arsenic filter based on composite iron matrix: Development and deployment studies for groundwater of Bangladesh, J. Environ. Sci. Health A 42 (2007) 1869–1878. [40] K. Miyajima, C. Noubactep, Effects of mixing granular iron with sand on the efficiency of methylene blue discoloration, Chem. Eng. J. 200–202 (2012) 433–438. [41] A.M. Gottinger, Chemical-free arsenic removal from potable water with a ZVI-amended biofilter. Master thesis, University of Regina (Saskatchewan, Canada), (2010) 90 pp. [42] J. Pietsch, W. Schmidt, F. Sacher, S. Fichtner, H. Brauch, Pesticides and another organic micropollutants in the river Elbe, Fresenius J. Anal. Chem. 353 (1995), 75–82. [43] H. Chiew, M.L. Sampson, S. Huch, S. Ken, B.C. Bostick, Effect of groundwater iron and phosphate on the efficacy of arsenic removal by iron-amended biosand filters, Environ. Sci. Technol. 43 (2009) 6295–6300. [44] W.B. Fortune, M.G. Mellon, Determination of iron with o-phenanthroline: a spectrophotometric study, Ind. Eng. Chem. Anal. Ed. 10 (1938) 60–64. [45] L.G. Saywell, B.B. Cunningham, Determination of iron: colorimetric o-phenanthroline method, Ind. Eng. Chem. Anal. Ed. 9 (1937) 67–69. [46] H.S. Salem, G.V. Chilingarian, The cementation factor of Archie's equation for shaly sandstone reservoirs, J. Pet. Sci. Eng. 23 (1999) 83–93. [47] J.H. Potgieter, S.S. Potgieter, S.J. Moja, A. Mulaba-Bafubiandi, The standard reactivity test as a measure of lime’s quality, J. South African Inst. Mining Metal. January/February 2002 (2002) 67–69. [48] C.B. Wang, W.-X. Zhang, Synthesizing nanoscale iron particles for rapid and complete dechlorination of TCE and PCBs, Environ. Sci. Technol. 31 (1997) 2154–2156. 23 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 [49] C. Macé, S. Desrocher, F. Gheorghiu, A. Kane, M. Pupeza, M. Cernik, P. Kvapil, R. Venkatakrishnan, W.-X. Zhang, Nanotechnology and groundwater remediation: A step forward in technology understanding, Remed. J. 16 (2006) 23–33. [50] C.W. Fetter, Applied Hydrogeology, 3rd ed. Upper Saddle River, NJ: Prentice Hall, Inc (1994). [51] R.P. Dias, J.A. Teixeira, M.G. Mota, A.I. Yelshin, Particulate binary mixtures: Dependence of packing porosity on particle size ratio, Ind. Eng. Chem. Res. 43 (2004) 7912–7919. [52] R.P. Dias, C.S. Fernandes, J.A. Teixeira, M. Mota, A. Yelshin, Permeability analysis in bisized porous media: Wall effect between particles of different size, J. Hydrol. 349 (2008) 470–474. [53] M.M. Ahmadi, S. Mohammadi, A.N. Hayati, Analytical derivation of tortuosity and permeability of monosized spheres: A volume averaging approach, Phys. Rev. E 83 (2011) 026312 [8 pages]. [54] A.N. Hayati, M.M. Ahmadi, S. Mohammadi, How particle shape affects the flow through granular materials, Phys. Rev. E 85, 036310 (2012) [4 pages]. [55] R. Muftikian, Q. Fernando, N. Korte, A method for the rapid dechlorination of lowmolecularweight chlorinated hydrocarbons inwater, Water Res. 29 (1995) 2434– 2439. [56] B.J. Allred, Laboratory evaluation of zero valent iron and sulfur-modified iron for agricultural drainage water treatment, Ground Water Monit. Remed. 32 (2012) 81–95. [57] E. Bi, J.F. Devlin, B. Huang, Effects of mixing granular iron with sand on the kinetics of trichloroethylene reduction, Ground Water Monit. Remed. 29 (2009) 56–62. [58] O. Dikinya, C. Hinz, G. Aylmore, Decrease in hydraulic conductivity and particle release associated with self-filtration in saturated soil columns, Geoderma 146 (2008) 192– 200. 24 614 615 616 617 618 619 620 621 [59] C. Noubactep, Aqueous contaminant removal by metallic iron: Is the paradigm shifting? Water SA 37 (2011) 419–426. [60] R. Crane, C. Noubactep, Elemental metals for environmental remediation: learning from hydrometallurgy, Fresenius Environ. Bull. 21 (2012) 1192–1196. [61] S. Caré, R. Crane, P.S. Calabro, A. Ghauch, E. Temgoua, C. Noubactep, Modelling the permeability loss of metallic iron water filtration systems, Clean – Soil, Air, Water (2012) doi: 10.1002/clen.201200167. 25 Table 1: Column tests program. One use cycle consists of a filtration period, followed by a resting period with no flow. As expansive corrosion induces permeability loss, the filtration period will become longer. The test program can regarded as the investigation of the impact of a ca. 5 cm layer of Fe 621 622 623 624 625 626 0 on the discoloration ability of a 2.0 mg L-1 MB solution. material size code high mass function (mm) (cm) (g) solution - Hsolution 15.5 (-) resting solution coarse sand 1.6 - 2.0 H1-sand 19.1 - 20.0 172.4 lower MB adsorbent Fe0 0.3 - 2.0 Hiron 4.5 – 5.4 100.0 tested material fine sand 0.25-0.30 H2-sand 8.0 60.0 stronger MB adsorbent 627 628 629 630 26 Table 2: Origin, name and main characteristics of tested Fe0 materials. 630 631 origin original denotation code form ∅ (μm) (a) iPutec GmbH FG 0300/2000 ZVI1 filings 400-800 G. Maier GmbH FG 0300/2000 ZVI2 filings 300-2000 Connelly-GPM ETC-CC-1004 ZVI3 filings 1000-3000 iPutec GmbH FG 0300/2000 ZVI4 filings 300-3000 G. Maier GmbH Graugußgranulat ZVI5 chips 350-1200 Connelly-GPM ETC-CC-1004 ZVI6 filings 500-1000 MAZ, mbH Sorte 69(b) ZVI7 filings 80-4000 Würth Hartgußstrahlmittel ZVI8 spherical 1200 Pometon S.p.A. Ferblast RI 850 ZVI9 filings 500 Peerless Metals Unknown ZVI10 filings 450-550 (c) (a) Average values from material supplier; (b) Scrap iron material, (c) values from ref. [41]. 632 633 634 635 27 Table 3: Corresponding correlation parameters (kEDTA, b) for the 10 metallic iron materials. As a rule, the more reactive a material is under given conditions the larger the k 635 636 637 638 639 640 EDTA value. General conditions: initial pH 5.2, initial EDTA concentration 2 mM, room temperature 23 ± 2 °C, and Fe0 mass loading 2 g L−1. kEDTA and b-values were calculated in Origin 6.0. “n.d.” stands for not determined. Material Origin kEDTA b (μg/h) (μg) ZVI1 iPutec 24 ± 1 136 ± 15 ZVI2 G.M.GmbH 29 ± 1 117 ± 15 ZVI3 Connelly 32 ± 1 193 ± 25 ZVI4 iPutec 23 ± 1 73 ± 9 ZVI5 G.M.GmbH 27 ± 1 108 ± 10 ZVI6 Connelly 24 ± 4 1685 ± 289 ZVI7 MAZ, mbH 27 ± 1 173 ± 13 ZVI8 Würth 13 ± 1 65 ± 20 ZVI9 FERBLAST 36 ± 3 185 ± 69 ZVI10 Peerless 37 ± 2 210 ± 159 641 642 643 644 28 Table 4: Results of the Fe0 characterization in batch (kEDTA) and column experiments. v0 is the initial flow velocity, [Fe] 644 645 646 647 648 649 650 0 the level of dissolved iron after 4 days, Σ[Fe] the total mass of leached Fe, Σ[MB]f the cumulative extent of MB dicoloration and kEDTA the dissolution kinetics of Fe0 in 2 mM EDTA. It should be noticed that less than 0.02 % of the initial amount of Fe0 (100 g) has been leached during the experiment. Material Size k0 [Fe]0 Σ[Fe]f E kEDTA (mm) (mm min-1) (mg L-1) (mg) (%) (mg h-1) ZVI1 0.40 - 0.80 7.7 2.7 8.0 97.2 24 ± 1 ZVI10 0.45 - 0.55 43.2 1.4 15.9 94.4 37 ± 2 ZVI9 0.50 44.4 2.2 18.2 88.4 36 ± 3 651 652 29 Figure 1 652 ZVI 1 (iPutec GmbH) ZVI 2 (Gotthart Mayer GmbH) ZVI 3 (Connelly iron GPM) ZVI 4 (iPutec GmbH) ZVI 5 (Gotthart Mayer GmbH) ZVI 6 (Connelly iron GPM) ZVI 7 (MAZ, mbH) ZVI 8 (Würth) ZVI 9 (Ferblast) ZVI 10 (Peerless) 653 654 30 Figure 2 654 655 656 657 658 31 Figure 3 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 ZVI 1 ZVI9 ZVI 10 Sand 32 Figure 4 681 0 20 40 60 80 100 0 15 30 45 60 75 90 ZVI10 ZVI9 ZVI1 Fe / [m g L- 1 ] elapsed time / [hours] 682 683 33 Figure 5 683 0 6 12 18 24 30 0 500 1000 1500 2000 2500 3000 3500 ZVI10 ZVI9 ZVI1 Fe / [μg L -1 ] elapsed time / [days] 684 685 686 34 Figure 6 686 687 5 10 15 20 25 30 75 80 85 90 95 100 (a) ZVI1 ZVI9 ZVI10 E / [ % ] elapsed tim e / [days] 688 5 10 15 20 25 30 0 400 800 1200 1600 2000 2400 (b) ZVI1 ZVI9 ZVI10 M B / [μg ] e lapsed tim e / [days] 689 5 10 15 20 25 30 0 2 4 6 8 10 12 (c) ZVI1 ZVI9 ZVI10 ΣM B / [m g] e lapsed tim e / [days] 690 691 35 Figure 7 691 0 5 10 15 20 25 30 0 8 16 24 32 40 48 (a) ZVI10 ZVI9 ZVI1 k / [ m m m in -1 ] elapsed time / [days] 692 693 0 5 10 15 20 25 30 0 20 40 60 80 100 (b) ZVI10 ZVI9 ZVI1 k / [ % ] elapsed time / [days] 694 695 36 Figure captions 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 Figure 1: Micrographs of the then tested granulated metallic iron materials. Distances of lines in the background are 1 cm in vertical and horizontal directions. 500 mg of each material was used. Figure 2: Schematic diagram of the experimental design. Due to the volumetric expansive nature of aqueous iron corrosion (Fe0), progressive increase in the filtration time during use (permeability loss) is expected. Figure 3: SEM images of the material particles used in column experiments: sand, ZVI1, ZVI9 and ZVI10. Figure 4: Iron dissolution from the three Fe0 materials by 2 mM EDTA under non-disturbed conditions for four days. Error bars denote the standard error for triplicate experimental results. The lines are not fitting functions, they simply connect points to facilitate visualization. Figure 5: Evolution of the dissolved iron concentration in the effluent as a function of time. Experimental conditions: 100 g Fe0 representing about 5 cm of a material layer; [MB] = 2.0 mg L-1. Filling material: sand. Column length 50 cm, column diameter 2.6 cm. The lines are not fitting functions, they simply connect points to facilitate visualization. Figure 6: Evolution of the extent of methylene blue (MB) discoloration by Fe0 as a function of time: (a) E values, (b) mass of MB in the effluent, and (c) cumulative mass of MB in the effluent. Experimental conditions: 100 g Fe0 representing about 5 cm of a material layer; 37 [MB] = 2.0 mg L-1. Filling material: sand. Column length 50 cm, column diameter 2.6 cm. The lines are not fitting functions, they simply connect points to facilitate visualization. 720 721 722 723 724 725 726 727 728 Figure 7: Evolution of the hydraulic conductivity (permeability) by Fe0 column as a function of time: (a) absolute permeability, and (b) relative permeability. Experimental conditions: 100 g Fe0 representing about 5 cm of a material layer; [MB] = 2.0 mg L-1. Filling material: sand. Column length 50 cm, column diameter 2.6 cm. It can be seen that after 12 days the initial permeability has decreased by more than 80 % in all systems. The lines are not fitting functions, they simply connect points to facilitate visualization. 38