Enhancing the sustainability of household Fe0/sand filters by using bimetallics and MnO 1 2 3 4 5 6 7 8 9 10 11 2. Noubactep Chicgoua *(1,2), Caré Sabine.(3), Btatkeu K. Brice Donald.(4), Nanseu-Njiki Charles Péguy (5) (1) Angewandte Geologie, Universität Göttingen, Goldschmidtstraße 3, D - 37077 Göttingen, Germany; (2) Kultur und Nachhaltige Entwicklung CDD e.V., Postfach 1502, D - 37005 Göttingen, Germany; (3) Université Paris-Est, Laboratoire Navier, (ENPC/IFSTTAR/CNRS), 2 allée Kepler, 77420 Champs sur Marne, France. (4) ENSAI/University of Ngaoundere, BP 455 Ngaoundere, Cameroon; (5) Laboratoire de Chimie Analytique, Faculté des sciences, Université de Yaoundé I, B.P. 812 Yaoundé, Cameroon. 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 * corresponding author: e-mail: cnoubac@gwdg.de; Tel. +49 551 39 3191, Fax: +49 551 399379 (accepted Mon, 20 Jun 2011 13:06:39) Abstract Filtration systems containing metallic iron as reactive medium (Fe0 beds) have been intensively used for water treatment during the last two decades. The sustainability of Fe0 beds is severely confined by two major factors: (i) reactivity loss as result of the formation of an oxide scale on Fe0, and (ii) permeability loss due to pore filling by generated iron corrosion products. Both factors are inherent to iron corrosion at pH > 4.5 and are common during the lifespan of a Fe0 bed. It is of great practical significance to improve the performance of Fe0 beds by properly addressing these key factors. Recent studies have shown that both reactivity loss and permeability loss could be addressed by mixing Fe0 and inert materials. For a non porous additive like quartz, the threshold value for the Fe0 volumetric proportion is 51 %. Using the Fe0/quartz system as reference, this study theoretically discusses the possibility of (i) replacing Fe0 by bimetallic systems (e.g. Fe0/Cu0), or (ii) partially replacing quartz by a reactive metal oxide (MnO2 or TiO2) to improve the efficiency of Fe0 beds. Results confirmed the suitability of both tools for sustaining Fe0 bed performance. It is shown that using a Fe0:MnO2 system with the volumetric proportion 51:49 will yield a filter with 40 % residual 1 porosity at Fe0 depletion (MnO2 porosity 62 %). This study improves Fe0 bed design and can be considered as a basis for further refinement and detailed research for efficient Fe 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 0 filters. Keywords: Iron filters; Long-term reactivity; Oxide scale; Water treatment, Zerovalent iron. 1 Introduction A filtration system containing metallic iron as reactive medium (hereafter termed as Fe0 bed) is an attractive method which can continuously remove contaminants from surface water, groundwater, and industrial effluent. The technology was introduced around 1990 by Canadian hydrogeologists [1-3]. The Fe0 bed technology has the potential to produce safe drinking water in water plants [4-6], and to treat wastewater [4,7-9] and groundwater [3,10,11]. Drawbacks for this innovative technology include (i) the accumulation of reaction by-products, (ii) the decrease in surface activity over time (reactivity loss), and (iii) the decrease of the bed permeability over the time (permeability loss) [11,12]. If Fe0 beds are used for above ground safe drinking water production, none of these three drawbacks is really a problem. In fact, individual beds will be replaced as soon as a problem is observed. Fe0 beds have been demonstrated and used as an efficient and affordable technology for safe drinking water production at small scale (household and small community) [13-21]. The first generation filters made up of a 100 % layer of Fe0 were very efficient but not sustainable because of too rapid clogging [13,16]. The second generation filters used Fe0 and inert filling materials (mostly sand) and could achieve certain sustainability [19,22,23]. Recently, a theoretical discussion on the proportion of Fe0 in Fe0 beds has been performed [24-26]. Results demonstrated that the Fe0 volumetric ratio for sustainable filters is ≤ 52 % when the additive is non porous (e.g. quartz). This threshold value does not give any information on the nature of Fe0 (e.g. bimetallic, composite). The nature of filling materials has been discussed on the porosity perspective [25]. It is of great practical value to improve the performance of Fe0 beds. The following three perspectives could be addressed: (i) developing reliable Fe0 2 materials (including composites), (ii) selecting the most suitable additive (porous, inert or/and reactive), and (iii) optimizing Fe 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 0 bed design (e.g. thickness of a bed). The suitability of plated metal (bimetallic systems) for reactivity enhancement has already been demonstrated [27-29]. By plating a Fe0 material with a more noble metal, the number of micro-defects in the crystal lattice due to different dimensions and charges of micro-alloyed elements, related to Fe0 increase. Micro-alloyed components generate defects in a metal structure (interstitials and vacancies in crystal lattice) and an imbalance in the charge distribution, as a result of many micro-galvanic cells. These defects decrease energy barriers for transport of Fe2+ ion from metal to oxide layer [27]. A positive accompanying effect is an increase in ionic and electronic conductivity and thus, an increase of corrosion rate ([27] and ref. cited therein). This makes the plated Fe0 chemically much more reactive than the original Fe0 material. Contaminant removal by bimetallic systems is based on several physico- chemical processes and the in situ formation of very reactive iron hydoxides. The major processes in a Fe0 bed are adsorption, co-precipitation and size exclusion. The ability of MnO2 to sustain contaminant removal by Fe0 was indirectly demonstrated in a recent study by He and Hering [30]. The authors demonstrated that AsIII was quantitatively oxidized to AsV by MnO2 but resulted AsV remained in solution. Quantitative As removal was indeed observed in systems containing FeII and was mostly attributed to As co-precipitation with FeIII hydroxides. In a similar way, MnO2 can sustain Fe0 oxidative dissolution yielding FeII which capability to induce reductive dissolution of MnO2 will sustain the process of contaminant removal. Note that the work of He and Hering [30] recalled that “contaminant reduction” and “contaminant removal” should never be randomly interchanged. In other words, a chemical transformation (oxidation or reduction) may favour contaminant removal but is not a stand alone removal mechanism. Aqueous contaminant removal by co- precipitation with in-situ generated metal hydroxides is well documented process for all classes of contaminants ([29, 31] and ref. therein). 3 The present study intends to theoretically discuss the optimization of Fe0 bed by two different tools: (i) replacing Fe 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 0 by a bimetallic system, and (ii) using MnO2 as reactive additives, e.g. partially or totally substituting quartz by MnO2 in a reference Fe0/quartz bed. Both tools have the potential to improve the contaminant removal efficiency and prolong the lifespan of Fe0 beds. For the sake of clarity, the process of contaminant removal in Fe0 beds will be first presented. 2 Contaminant removal in Fe0 filters 2.1 Filtration in packed-column A Fe0 bed is primarily a packed-column of granular Fe0 and quartz (sand) particles. The efficiency of packed-columns for contaminant removal is usually evaluated by monitoring the time dependent evolution of (i) the contaminant concentration in the effluent, and (ii) the water velocity through the column. Physical and chemical conditions evaluated in such experiments include grain particle size and shape, solution pH, solution ionic strength and composition [32,33]. Ideally, contaminants are deposited throughout the entire filter media. Accordingly, Fe0 filtration is a deep-bed or depth filtration process [34]. 2.2 Filtration in a Fe0 bed Contaminant removal within a Fe0/quartz filter is not comparable to contaminant removal by an adsorption column [5,35]. The most important feature of Fe0/quartz filters regards the specificity of the removal process. In an adsorption column, contaminants with different physico-chemical properties can be separated due to their differential affinity to the adsorbing material (e.g. activated carbon, iron oxide). Similarly, particles with different sizes can be separated in a depth sand filter. But in a Fe0/quartz filter, there is primarily no such specificity as contaminants are removed during the dynamic process of iron corrosion products formation (Fe hydroxides/oxides) and by resulted Fe hydroxides/oxides [35, 36]. 2.3 Mechanism of contaminant removal in Fe0/sand filters 4 Regardless from any contaminant inflow, the initial pore space in a Fe0/quartz filter is progressively filled by in situ generated Fe hydroxides/oxides. Decreased pore space is coupled to improved size exclusion capacity. Accordingly, regardless from physico-chemical interactions between contaminants, Fe 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 0 and Fe hydroxides/oxides, contaminant removal by pure size exclusion will inevitably occur with increasing service life. This fundamental aspect has received little attention to date as the scientific community was focused on specific interactions between selected contaminants and Fe0. In this effort a particular attention was paid to chemical reduction [10,11]. It has already been demonstrated that contaminants are fundamentally entrapped within the film of corrosion products in the vicinity of the Fe0 surface [37-40]. It is essential to note that the formation of corrosion products is a cycle of expansion/contraction occurring in the pore space [5,26,35]. During this process, native iron (Fe0: SSA < 1 m2/g) is first transformed to voluminous iron hydroxides possibly having specific surface area (SSA) > 500 m2/g before progressively contracted to amorphous and crystalline oxides with SSA ≤ 10 m2/g. The voluminous colloid which is intermediary formed [41] during an expansion/contraction cycle can be compared to a spider web which traps inflowing contaminants and keeps them adsorbed while the colloid is further transformed. In other words, before the pore space becomes close enough for the Fe0 filter to act as an ultra-filtration system, the expansion/contraction cycle traps contaminants from the infiltrating water. It is certain, that the kinetics of iron oxidation will decrease as soon as not enough space is available for expansive corrosion. This is a plausible explanation for the controversial observation, that TCE removal rates were higher in a system with 85 % Fe0 than they were in a 100 % Fe0 (w/w) system [42]. Substituting a fraction of Fe0 by sand (quartz) was proven a prerequisite for efficient long-term permeable Fe0 filters [24,25]. The present study aims at theoretically discussing the substitution of a portion of quartz in the dual media (Fe0/quartz) by a reactive 5 oxide to sustain long-term Fe0 reactivity. A further discussed way to sustain Fe0 reactivity is to plate Fe 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 0 by a second more electropositive metal. 3 Sustaining Fe0 reactivity The presentation above has suggested that successful contaminant removal in a Fe0/quartz bed is coupled with the whole process of iron corrosion. Accordingly, a reliable way to warrant continuous contaminant removal is to sustain iron corrosion. This section examines two possibilities of sustaining Fe0 corrosion. 3.1 Use of bimetallics The deposition of small amounts of second metals such as Ni and Pd onto the Fe0 surface has been proven beneficial for the process of aqueous contaminant removal ([12] and ref. therein). Bimetallic systems are an efficient media for accelerating the decontamination [27-29]. The prevailing operating mode of bimetallic systems [12] was recently challenged. It was shown that any enhanced contaminant reduction, if applicable, occurs by an indirect process [43]. Table 1 summarises the standard electrode potentials of seven elemental metals (Me0) which may be used to sustain Fe0 oxidation in filters: Co0, Ni0, Cu0, Ag0, Pd0, Pt0, Au0. From these metals, Cu0 is the most used. An ideal Me0 acts as a catalyst. For example, Cu0 is oxidized by water to Cu2+ and the resulted Cu2+ oxidizes Fe0. Considering Me0 as pure catalyst, calculated amounts of Fe0/Me0 will be added to a sand filter and the porosity will vary as in a Fe0/quartz bed. In other words, replacing Fe0 particles by bimetallic (Fe0/Me0) particles of similar size and occupying the same volume will not significantly impact porosity loss. Bimetallic/quartz filters will behave like Fe0/quartz filters in term of the evolution of the porosity but exhibit an enhanced long- term reactivity. The theoretical evolution of porosity loss due to clogging is discussed in section 4. 3.2 Use of metal oxides 6 The use of natural oxides to sustain Fe0 reactivity was derived from the well-documented reductive dissolution of MnO 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 2 by FeII [46,47]. This process was successfully used to demonstrate the importance of corrosion products in the process of contaminant removal by Fe0 [48-52]. Moreover, the reductive dissolution of MnO2 by FeII is a well-established hydrometallurgical process [53-55]. Recently, Bafghi et al. [55] published a work on the reductive dissolution of manganese ore in the presence of Fe0. Based on theoretical and experimental facts, they concluded that Fe0 was superior to FeII for MnO2 reductive dissolution. However, they insisted on the fact that FeII is more available. Mechanistic details will not be considered here as it is sufficient to consider that metal oxides could sustain Fe0 corrosion (Tab. 2, Tab. 3). Remember that contaminants are fundamentally removed by iron corrosion products (adsorption and co-precipitation) and these are increasingly available when iron oxidation is sustained by FeII consumption [51,52,56]. Only the four naturally abundant oxides will be considered: Al2O3, MnO2, SiO2 and TiO2 (Tab. 2). The standard electrode potentials from Tab. 2 show that only MnO2 and TiO2 could sustain Fe0 reactivity. Accordingly, Al2O3 and SiO2 can only be used as inert filling materials. The ability of MnO2 to sustain Fe0 corrosion was already demonstrated [48-50,56]. In particular, the success of SONO Arsenic Filters in Bangladesh is based on continuous production of reactive iron oxides by the used manganese oxide/Fe0 composite (coupled to size exclusion in the filter). Table 3 depicts some relevant electrode reactions (half-reactions) for the discussion of the Fe0 reactivity. From Tab. 3 it can be seen that MnO2 is theoretically by far superior to TiO2 in sustaining Fe0 corrosion. However, the suitability of available metal oxides to sustain Fe reactivity should be tested on the case-by-case basis. Moreover, the objective should not be to use the most reactive metal oxide but rather the one with satisfactorily reactivity for individual purposes. For example, for a given Fe0, a very reactive MnO2 can accelerate the Fe0 corrosion in such a way that filter fouling/clogging is achieved similar as with bio-corrosion [10,11]. In such situations a lesser reactive MnO2 should be 7 used. Testing well-characterized Fe0 and reactive manganese oxides in various proportions is regarded as a tool to produce site-specific composites. In fact, the composite currently used in SAF filters is the same material which is used everywhere (actually mostly in Bangladesh and Nepal) [14,18]. However, contaminated waters are of various background compositions and each water could be treated with appropriate composites. 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 To discuss the evolution of the pore space within the filter as metal oxides react, it will be considered that MnO2 and TiO2 reduction at neutral pH values mostly yield insoluble hydroxides: MnOOH and TiOOH (Eq. 17 and Eq. 18). Fe0 + 3 MnO2 + 2 H2O ⇔ FeOOH + 3 MnOOH (17) Fe0 + 3 TiO2 + 2 H2O ⇔ FeOOH + 3 TiOOH (18) Eq. 17 and 18 suggest that the oxidation of one Fe atom consumes three molecules of the adsorbent MnO2 (or TiO2) and produces one FeOOH and three MnOOH (or three TiOOH) as new adsorbents. The volume variation is estimated based on values of the specific weight defined as the ratio of the molecular volume of the reaction products to the molecular volume of the educts (Tab. A3). The discussion will only concern MnO2 because no tabulated value could be found for TiOOH. The evolution of the porosity loss due to clogging is discussed in section 4. 4 Evolution of the residual porosity using bimetallic particles or metal oxides This section will start with some general design equations. It has been recently showed that dimensionless design equations could be written such that for each practical case the appropriate values are derived [24-26]. In other words, the same equations are applied to household Fe0 filters, Fe0 treatment trench, and Fe0 reactive walls. In each case, the used materials (Fe0 and additives) should be thoroughly characterized. Relevant material characteristics include porosity, particle size, shape, specific weight, and surface area. The impacts of material characteristics on the bed efficiency are not discussed here. 4.1 General design equations of Fe0 beds 8 Cylindrical beds are considered. H is the height and D is the internal diameter. The cylinder contains a reactive zone with the height H 209 210 211 212 213 214 215 216 217 218 rz and the volume Vrz. Beds are supposed to be filled by granular materials. The compactness (or packing density) C (-) is defined as the ratio of the volume of the particles to the total packing volume (Vrz). Considering the granular material as composed of mono-dispersed spheres subjected to soft vibrations, the compactness C is generally considered to be equal to 0.64 for a random close packing. It is assumed that the particles are non porous. The initial porosity Φ0 (-) of the reactive zone and the thickness Hrz of the reactive zone are respectively then given by: Φ0 = 1 – C (19) V D 4 H rz2rz .π= (20) 219 220 221 222 223 224 225 226 227 228 229 230 231 The filling of the bed porosity by iron corrosion products can be estimated from a simplified modeling (Fig. 1) based on the following assumptions: (i) uniform corrosion: the radius reduction of the spherical particles is the same for all the Fe0 particles. (ii) the packing density C remains constant for all particles (Fe0 and quartz). The volume of the granular material is not modified by the corrosion process: no pressure induced by rust formation around Fe particles and no compaction of the Fe0 mixture during the corrosion process (Vrz remains constant). (iii) reaction products are fluid enough to progressively fill available pore space. Assuming that the coefficient of volumetric expansion or the specific volume (η) of the reaction products is: η = Voxide/VFe (21) 9 where Voxide is the volume of the reaction product and VFe the volume of the parent Fe0. It is assumed that Fe 232 233 234 235 236 237 238 239 240 241 3O4 is the sole iron corrosion product for the Fe/quartz system. The specific volume for Fe3O4 is 2.1. The surplus volume of the reaction products contributing to porosity loss is V’oxide. Per definition V’oxide is the difference between the volume Voxide of reaction products and the volume VFe of parent Fe0. V’oxide is given by Eq. 22: V’oxide = (η - 1) * VFe (22) Assuming that iron expansive corrosion is the sole clogging factor, the bed is clogged when the volume V’oxide is equal to the initial inter-granular voids (Φ0.Vrz), the volume VFe,clogging of the consumed Fe0 leading to clogging of the bed is then estimated by: 1 V=V rz0gingcFe −η Φ . log, . (23) 242 243 244 245 246 247 248 249 250 251 252 253 254 255 Eq. 23 is of fundamental importance for Fe0 bed design as it determines the ideal Fe0 volume (and thus Fe0 mass) to be used. If V0 is the initial volume of dense Fe0, three cases can be distinguished: “(i) VFe,clogging > V0, no clogging due to expansive iron corrosion will occur. In this case, the real volume of Fe which may be consumed (Vconsumed-Fe) is equal to the initial volume V0 of Fe (Vconsumed-Fe = V0 < VFe,clogging) and there is a residual porosity at Fe0 depletion (Φr ≠ 0); (ii) VFe,clogging = V0, clogging will not occur before Fe0 depletion (Vconsumed-Fe = VFe,clogging = V0) but the final porosity is zero (Φr = 0): (iii) VFe,clogging < V0, clogging will occur before Fe0 depletion (Φr = 0). In this case, the real volume of consumed Fe leading to clogging is inferior to the initial volume V0 of Fe (Vconsumed-Fe = VFe,clogging < V0 ) and the excess Fe0 amount should be regarded as pure material wastage [24].” 10 The residual porosity Φr defined by Φr = Vresidual voids/Vrz and the residual mass of the iron (Fe 256 257 0) are evaluated by Eq. 24 and Eq.25: V V1 rz Feconsumed 0r −−η−Φ=Φ ).( (24) 258 V VV M M 0 0 Feconsumed 0 −= − (25) 259 260 261 262 263 where V0 is the initial volume of Fe, Vconsumed-Fe is the volume of the Fe which is consumed and M is the actual mass of Fe given by M = ρFe*(V0 - Vconsumed-Fe) with ρFe = 7,800 kg/m3. When the clogging appears before depletion of Fe0, the volume Vconsumed-Fe is given by the Eq. 24 and the residual porosity is Φr = 0 and there is a residual mass of iron M/M0 ≠ 0 (Eq. 25). When there is no clogging, the volume Vconsumed-Fe ≤ V0 and 0r ≠Φ (Eq. 24) and the Fe0 mass at Fe 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 0 depletion is zero (M/M0 = 0; Eq. 25). Eq. 19 through 25 should be routinely used to design laboratory experiments, pilot and field works. 4.2 Case of bimetallic/quartz system To sustain the Fe0 reactivity, Fe particles are replaced by bimetallic particles of comparable particle sizes. For Fe0/Cu0 bimetallic particles with 7 % in mass of Cu, the volumetric proportion of Fe0 is 93.85 % (considering for specific weights ρFe= 7800 kg/m3 and ρCu= 8960 kg/m3) and the reaction product is Fe3O4. The residual porosity and the residual mass are given by the same equations by replacing the initial volume V0 of Fe0 by 93.85 %* V0. The results for the Fe0/quartz and the bimetallic/quartz systems are given in Fig. 1. The trends are similar for both cases. While decreasing the Fe0 proportion at constant reactive zone thickness, Fe0 depletion was achieved for Fe0/quartz systems > 51 vol-% Fe0. In the case of bimetallic/quartz system (Fe0/Cu0), the clogging is avoided for < 54 vol-% bimetallic. To further sustain the efficiency of Fe0 beds, bimetallic systems containing larger amounts of plating metals (Me0) could be used. Moreover composites should be manufactured and tested. 11 Those composites should contain uniformly distributed Me0 in the bulk of Fe0 materials and not just deposited at their surface. Apart from such bulk composites, special Fe 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 0 materials with higher contents in S for instance, could be manufactured and tested. In fact, during the steel making process, efforts target as completely remove S are made. Sulphur is known for its negative effect on corrosion resistance of steel. However, in the context of water treatment with Fe0, readily corrosive Fe0 may be suitable. 4.3.1 Case of Fe0/MnO2/quartz system A reference reactive zone is considered as made up of 51 vol-% Fe0 and 49 vol-% quartz as it allows avoiding the clogging before Fe0 depletion as shown in Fig. 1. Iron filings from Gotthart Maier Metallpulver GmbH (Rheinfelden, Germany) containing 92 % Fe0 (w/w) is used for the calculations. Quartz particles of similar particle size are replaced by MnO2 particles in order to increase Fe0 reactivity. In presence of MnO2, it is assumed that chemical reactions yield to FeOOH and MnOOH products (Eq. 17). When MnO2 particles have been consumed, the Fe corrosion process leads to the reaction products Fe3O4. Calculations made with characteristics of a natural MnO2 (φMno2 = 62 %) given by Li et al. [57] are used in this study (Tab. A.1 and A.2). The mineral contained (weight) 77.8 % MnO2, 2.7 % Fe, 0.87 % Si, 2.78 % Al and 0.01 % S. For simplifications, it is considered that the mineral is made up of 77.8 % MnO2 and 22.2 % of a “gangue” having the characteristic of quartz (inert and non porous). This assumption was the rationale to use the specific weight of quartz in estimating the volumetric expansion coefficient for MnO2 (Appendix). Calculations showed that for the reference system (vol. Fe:quartz = 51:49), less than 3 % Fe0 is necessary to consume the amount of MnO2 that could be contained in up to 49 vol-% MnO2, when quartz is completely replaced by MnO2. Moreover, it is shown that thanks to the porosity of MnO2, there is a net increase of the initial bed porosity in comparison to the Fe0/quartz bed. Furthermore, because the chemical transformation (Fe + MnO2) to (MnOOH +FeOOH) was 12 305 306 307 308 309 310 311 312 313 not expansive but slightly compressive (η = 0.94), there is a net increase of the residual bed porosity (Appendix). Eq. 24 and 25 are slightly modified because the initial porosity Φ0 is increased by the internal porosity of the MnO2 particles and the voids (Φ0.Vrz) are filled by (i) the MnOOH hydroxides with η1 = VMnOOH/VMnO2 = 0.94 (Appendix), (ii) the FeOOH hydroxides with η2 = VFeOOH/VFe = 3.03 and (iii) Fe3O4 with η3 = VFe3O4/VFe = 2.1. Values of η2 and η3 are from ref. [58]. The initial porosity Φ0 , the residual porosity Φr and the residual mass of the iron (Fe0) are evaluated by Eq. 26 , Eq. 27 and Eq.28: Φ0 = (1 – C) + fMno2*φMno2 (26) V V13V12V11 rz 321 0r ).().().( −η+−η+−η−Φ=Φ (27) 314 V VV2V0 M M 0 3 0 −−= (28) 315 Where φMnO2 is the internal porosity of the MnO2 particles (φMnO2 = 62 %) and f pp (-) is the porous particle volume fraction determined by 316 VVf 2MnO2Mnopp = with VMnO2 the volume of the porous particles MnO 317 318 319 320 321 322 323 324 325 326 327 2. V1 is the volume of the dense MnO2, V2 is the Fe0 volume which reacts with MnO2. The volume of Fe0 (V2) is obtained considering that one Fe atom consumes three molecules of MnO2 (Eq. 17). V3 is the volume of Fe0 leading to Fe3O4 after MnO2 depletion. At Fe depletion (M/M0 = 0), the residual porosity Φr is given by Eq. 27 (V3 = V0 - V2 with V0 the initial volume of Fe0 in the reactive zone). If clogging appears before depletion of Fe0, the residual porosity is Φr = 0 and the residual mass is given by Eq. 28. The evolution of the initial and residual porosity is given in Fig. 2a for the reference system when the volumetric proportion of MnO2 varies from 0 to 49 %. Thanks to the internal porosity of MnO2 the initial porosity (Φ0) increased from 36.0 to 55.4 %. On the other hand, 13 Fe0 depletion was observed in all systems but the residual porosity varies from 8.3 to 40.6 %. The increase of the residual porosity with increasing MnO 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 2 proportion is due to two factors: (i) the internal porosity of MnO2 and (ii) the light volumetric compaction of MnO2 reduction to MnOOH. Remember that in the reference system the residual porosity is zero at Fe0 depletion. Therefore, similarly as for the Fe0/pumice system [25,59], the internal porosity of MnO2 could be regarded as storage room for in situ generated corrosion products (see also Appendix). In summary, MnO2 reduction to MnOOH has two beneficial effects: (i) sustaining Fe0 reactivity and (ii) serving as storage room for corrosion products. Additionally, the chemical potential of the reaction between FeII and MnO2 will drive the diffusion of FeII from Fe0 oxidative dissolution to the internal surface of MnO2. Thus, filling the internal porosity of MnO2 with iron corrosion products is more likely to occur than for an inert material like pumice. Another important aspect is that MnII from the reductive dissolution of MnO2 will migrate in the system and be oxidized by several species to MnOOH or MnO2 [60-62] which will further oxidize FeII from Fe0. Accordingly, despite stoichiometric disadvantage, MnO2 may work as catalyst to sustain Fe0 reactivity during its whole lifespan [63,64]. The fact that no pore clogging was observed for 51 vol-% Fe0 suggests that more Fe0 could be used for the same bed (Vrz constant). In this case, the volume of MnO2 is necessarily reduced. Fig. 2b shows the results of systems with volumetric ratios Fe0:MnO2 varying between 60:0 and 60:40. That is, when quartz is progressively replaced by MnO2. While the initial porosity increased from 36 to 51.9 %, the residual porosity increased from 0.0 to 24.8 %. This results show clearly that even in rising the Fe0 volumetric ratio in the column from 51 to 60 %, a residual porosity exists and the Fe0 performance is increased by the cycle of MnO2 as described above. 5 Concluding remarks A Fe0/quartz filter is regarded as an assisted slow sand filtration system which efficiency is improved by addition of a calculated amount of metallic iron [65]. The aqueous corrosion by 14 infiltrating raw water should ideally transform the initial Fe0 filter to an efficient filtration system. Improved filtration efficiency is based on the volumetric expansive nature of iron corrosion [59,66] which ideally only partly fills initial pore space in the sand-like filter [67]. 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 This communication is part of an ongoing effort for low-cost water treatment using well- design Fe0 beds [5,6,17,24-26,68,69]. While dual filters of Fe0 and quartz will be long-term permeable, the substitution of Fe0 by bimetallic materials or the substitution of inert sand by reactive metal oxides will sustain long-term reactivity. For example, sustaining Fe0 reactivity by substituting a fraction of Fe0 by metallic copper chips or (partly or entirely) replacing sand by granulated MnO2 will help to design efficient and sustainable filtration systems. The encouraging results of SONO filters in Bangladesh [14,15,18,70-72] suggest that practitioners of subsurface permeable barriers should check the possibility of replacing Fe0 by composites or amended Fe0 by reactive additives. The concept presented and discussed in this communication (and related articles) would be useful in designing efficient and affordable water filtration systems at several scales. The concept also renders itself as a basis for further refinement and detailed research at laboratory and field scale. One may wonder how the Fe0 bed technology will be developed in different parts of the world. Acknowledgments Dr. Boniface P. T. Fokwa (Institute of Inorganic Chemistry, RWTH Aachen University) is acknowledged for fruitful discussions on the mineralogy of manganese oxides. The manuscript was improved by the insightful comments of anonymous reviewers from CLEAN Soil, Air, Water. References [1] G.W. Reynolds, J.T. Hoff, R.W. Gillham, Sampling bias caused by materials used to monitor halocarbons in groundwater, Environ. Sci. Technol. 1990, 24, 135. [2] S.F. O´Hannesin, R.W. Gillham, Long-term performance of an in situ "iron wall" for remediation of VOCs, Ground Water 1998, 36, 164. 15 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 [3] R.W. Gillham, Development of the granular iron permeable reactive barrier technology (good science or good fortune), In "Advances in environmental geotechnics : proceedings of the International Symposium on Geoenvironmental Engineering in Hangzhou, China, September 8-10, 2009"; Y. Chen, X. Tang, L. Zhan (Eds); Springer Berlin/London, 2010, 5– 15. [4] D.D.J. Antia, Sustainable zero-valent metal (ZVM) water treatment associated with diffusion, infiltration, abstraction and recirculation, Sustainability 2010, 2, 2988. [5] C. Noubactep, Metallic iron for safe drinking water worldwide. Chem. Eng. J. 2010, 165, 740. [6] C. Noubactep, A. Schöner, Metallic iron: dawn of a new era of drinking water treatment research? Fresen. Environ. Bull. 2010, 19, 1661. [7] S. Junyapoon Use of zero-valent iron for wastewater treatment, KMITL Sci. Tech. J. 2005, 5, 587. [8] G. Bartzas, K. Komnitsas, I. Paspaliaris, Laboratory evaluation of Fe0 barriers to treat acidic leachates, Miner. Eng. 2006, 19, 505. [9] K. Komnitsas, G. Bartzas, K. Fytas, I. Paspaliaris, Long-term efficiency and kinetic evaluation of ZVI barriers used for clean up of copper containing solutions, Miner. Eng. 2007, 20, 1200. [10] M.M. Scherer, S. Richter, R.L. Valentine, P.J.J. Alvarez, Chemistry and microbiology of permeable reactive barriers for in situ groundwater clean up, Crit. Rev. Environ. Sci. Technol. 2000, 30, 363. [11] A.D. Henderson, A.H. Demond, Long-term performance of zero-valent iron permeable reactive barriers: a critical review, Environ. Eng. Sci. 2007, 24, 401. [12] C.-c. Lee, R.-a Doong, Concentration effect of copper loading on the reductive dechlorination of tetrachloroethylene by zerovalent silicon, Water Sci. Technol. 2010, 62, 28. 16 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 [13] A.H. Khan, S.B. Rasul, A.K.M. Munir, M. Habibuddowla, M. Alauddin, S.S. Newaz, A. Hussam, Appraisal of a simple arsenic removal method for groundwater of Bangladesh, J. Environ. Sci. Health, 2000, A35, 1021. [14] 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 2007, A 42, 1869. [15] T.K.K. Ngai, R.R. Shrestha, B. Dangol, M. Maharjan, S.E. Murcott, Design for sustainable development – Household drinking water filter for arsenic and pathogen treatment in Nepal, J. Environ. Sci. Health 2007, A 42, 1879. [16] A. Hussam, Contending with a development disaster: SONO filters remove arsenic from well water in Bangladesh, Innovations 2009, 4, 89. [17] C. Noubactep, A. Schöner, P. Woafo, Metallic iron filters for universal access to safe drinking water, Clean: Soil, Air, Water 2009, 37, 930. [18] D. Pokhrel, B.S. Bhandari, T. Viraraghavan, Arsenic contamination of groundwater in the Terai region of Nepal: An overview of health concerns and treatment options, Environ. Int. 2009, 35, 157. [19] 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. [20] A.M. Gottinger, D.J. Wild, D. McMartin, B. Moldovan, D. Wang, Development of an iron-amended biofilter for removal of arsenic from rural Canadian prairie potable water. In: Water Pollution X. A.M. Marinov and C.A. Brebbia, Eds.; WIT Press: Ashurst, Southampton 2010, 333. [21] M.I. Litter, M.E. Morgada, J. Bundschuh, Possible treatments for arsenic removal in Latin American waters for human consumption, Environ. Pollut. 2010, 158, 1105. [22] O.X. Leupin, S.J. Hug, Oxidation and removal of arsenic (III) from aerated groundwater by filtration through sand and zero-valent iron, Water Res. 2005, 39, 1729. 17 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 [23] O.X. Leupin, S.J. Hug, A.B.M. Badruzzaman, Arsenic removal from Bangladesh tube well water with filter columns containing zerovalent iron filings and sand, Environ. Sci. Technol. 2005, 39, 8032. [24] 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 2010, 38, 951. [25] C. Noubactep, S. Caré, Enhancing sustainability of household water filters by mixing metallic iron with porous materials, Chem. Eng. J. 2010, 162, 635. [26] C. Noubactep, S. Caré, Dimensioning metallic iron beds for efficient contaminant removal, Chem. Eng. J. 2010, 163, 454. [27] A. Bojic, M. Purenovic, D. Bojic, Removal of chromium(VI) from water by micro- alloyed aluminium based composite in flow conditions. Water SA 2004, 30, 353. [28] A. Bojic, M. Purenovic, D. Bojic, T. Andjelkovic, Dehalogenation of trihalomethanes by a micro-alloyed aluminium composite under flow conditions. Water SA 2007, 33, 297. [29] A. Bojic, D. Bojic, T. Andjelkovic, Removal of Cu2+ and Zn2+ from model wastewaters by spontaneous reduction–coagulation process in flow conditions, J. Hazard. Mater. 2009, 168, 813. [30] Y.T. He, J.G. Hering, Enhancement of Arsenic(III) Sequestration by Manganese Oxides in the Presence of Iron(II). Water Air Soil Pollut. 2009, 203, 359. [31] K. Eusterhues, T. Rennert, H. Knicker, I. Kgel-Knabner, K.U. Totsche, U. Schwertmann, Fractionation of organic matter due to reaction with ferrihydrite: Coprecipitation versus adsorption. Environ. Sci. Technol. 2010, doi: 10.1021/es1023898. [32] N. Tufenkji, Modeling microbial transport in porous media: Traditional approaches and recent developments. Adv. Water Res. 2007, 30, 1455. 18 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 [33] G.J. Williams, B. Sheikh, R.B. Holden, T.J. Kouretas, K.L. Nelson, The impact of increased loading rate on granular media, rapid depth filtration of wastewater, Water Res. 2007, 41, 4535. [34] D.D. Putnam, M.A. Burns, Predicting the filtration of noncoagulating particles in depth filters, Chem. Eng. Sci. 1997, 52, 93. [35] C. Noubactep, The suitability of metallic iron for environmental remediation. Environ. Progr. Sust. En. 2010, 29, 286. [36] C. Noubactep, An analysis of the evolution of reactive species in Fe0/H2O systems. J. Hazard. Mater. 2009, 168, 1626. [37] C. Noubactep, Processes of contaminant removal in “Fe0–H2O” systems revisited: The importance of co-precipitation, Open Environ. J. 1 2007, 9. [38] C. Noubactep, A critical review on the mechanism of contaminant removal in Fe0–H2O systems, Environ. Technol. 2008, 29, 909. [39] C. Noubactep, The fundamental mechanism of aqueous contaminant removal by metallic iron. Water SA 2010, 36, 663. [40] C. Noubactep, A. Schöner, M. Sauter. Significance of oxide-film in discussing the mechanism of contaminant removal by elemental iron materials, In "Photo-Electrochemistry & Photo-Biology for the Sustainablity"; S. Kaneco, B. Viswanathan, H. Katsumata (Eds.), Bentham Science Publishers, 2010, 1, 35 [41] J. Livage, J. Lemerle, Transition metal oxide gels and colloids, Ann. Rev. Mater. Sci. 1982, 12,103. [42] E. Bi, J.F. Devlin, B. Huang, Effects of mixing granular iron with sand on the kinetics of trichloroethylene reduction, Ground Water Monit. Remed. 2009, 29, 56. [43] C. Noubactep, On the operating mode of bimetallic systems for environmental remediation, J. Hazard. Mater. 2009, 164, 394. 19 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 [44] Y. Gerasimov, V. Dreving, E. Eremin, A. Kiselev, V. Lebedev, G. Panchenkov, A. Shlygin, Physical Chemistry, vol. 2, MIR Moscow 1985, 624 pp. [45] K.B. Krauskopf, D.K. Bird, Introduction to geochemistry, Third Edition, McGraw-Hill, Inc, New York 1995, 647 pp. [46] D.F.A. Koch, Kinetics of the reaction between manganese dioxide and ferrous ion, Austr. J. Chem. 1957, 10, 150. [47] D. Postma, C.A.J. Appelo, Reduction of Mn-oxides by ferrous iron in a flow system: column experiment and reactive transport modelling, Geochim. Cosmochim. Acta 2000, 64, 1237. [48] C. Noubactep, G. Meinrath, P. Dietrich, B. Merkel, Mitigating uranium in groundwater: Prospects and limitations, Environ. Sci. Technol. 2003, 37, 4304. [49] C. Noubactep, G. Meinrath, J.B. Merkel, Investigating the mechanism of uranium removal by zerovalent iron materials, Environ. Chem. 2005, 2, 235. [50] C. Noubactep, A. Schöner, G. Meinrath, Mechanism of uranium (VI) fixation by elemental iron, J. Hazard. Mater. 2006, 132, 202. [51] A. Ghauch, H. Abou Assi, A. Tuqan, Investigating the mechanism of clofibric acid removal in Fe0/H2O systems, J. Hazard. Mater. 2010, 176, 48. [52] A. Ghauch, H. Abou Assi, S. Bdeir, Aqueous removal of diclofenac by plated elemental iron: bimetallic systems, J. Hazard. Mater. 2010, 182, 64. [53] K. Inoue, M. Okui, M. Tanaka, K. Shinoda, S. Suzuki, Y. Waseda, Influence of manganese on iron oxyhydroxides and oxides formed in aqueous solution, Corros. Sci. 2008, 50, 811. [54] M.J. Lottering, L. Lorenzen, N.S. Phala, J.T. Smit, G.A.C. Schalkwyk, Mineralogy and uranium leaching response of low grade South African ores, Miner. Eng. 2008, 21, 16. [55] M.Sh. Bafghi, A. Zakeri, Z. Ghasemi, M. Adeli, Reductive dissolution of manganese ore in sulfuric acid in the presence of iron metal, Hydrometallurgy 2008, 90, 207. 20 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 [56] D. Burghardt, A. Kassahun, Development of a reactive zone technology for simultaneous in situ immobilisation of radium and uranium, Environ. Geol. 2005, 49, 314. [57] K. Li, R.R. Rothfus, A.H. Adey, Effect of macroscopic properties of manganese oxides on absorption of sulfur dioxide. Environ. Sci. Technol. 1968, 2, 619. [58] S. Caré, Q.T. Nguyen, V. L'Hostis, Y. Berthaud, Mechanical properties of the rust layer induced by impressed current method in reinforced mortar, Cement Concrete Res. 2008, 38, 1079. [59] N. Moraci, P.S. Calabrò, Heavy metals removal and hydraulic performance in zero- valent iron/pumice permeable reactive barriers. J. Environ. Manag. 2010, 91, 2336. [60] T. Kohler, T. Armbruster, E. Libowitzky, Hydrogen Bonding and Jahn–Teller Distortion in Groutite,α-MnOOH, and Manganite, γ-MnOOH, and Their Relations to the Manganese Dioxides Ramsdellite and Pyrolusite. J. Solid State Chem. 1997, 133, 486. [61] J.E. Post, Manganese oxide minerals: Crystal structures and economic and environmental significance. Proc. Natl. Acad. Sci. USA 1999, 96, 3447. [62] S.A. Kirillov, V.S. Aleksandrova, T.V. Lisnycha, D.I. Dzanashvili, S.A. Khainakov, J.R. García, N.M. Visloguzova, O.I. Pendelyuk, Oxidation of synthetic hausmannite (Mn3O4) to manganite (MnOOH). J. Mol. Struct. 2009, 928, 89. [63] T. Ohzuku, M. Kitagawa, T. Hirai, Electrochemistry of manganese dioxide in lithium nonaqueous cell: III. X-Ray diffractional study on the reduction of spinel-related manganese dioxide. J. Electrochem. Soc. 1990, 137, 769. [64] Y. Chabre, J. Pannetier, Structural and electrochemical properties of the proton / γ-MnO2 system. Progr. Solid State Chem. 1995, 23, 1. [65] C. Noubactep, Metallic iron for safe drinking water production. Freiberg Online Geology, 2011,. 27, 38 pp, ISSN 1434-7512. (www.geo.tu-freiberg.de/fog) 21 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 [66] K. Bhargava, A.K. Ghosh, Y. Mori, S. Ramanujam, Model for cover cracking due to rebar corrosion in RC structures, Eng. Struct. 2006, 28, 1093. [67] K.-M. Yao, M.T. Habibian, C.R. O’melia, Water and waste water filtration: concepts and applications. Environ. Sci. Technol. 1971, 1105–1112. [68] C. Noubactep, P. Woafo, Elemental iron (Fe0) for better drinking water in rural areas of developing countries, In Merkel B.J., Hasche-Berger A. (Eds.) Uranium in the Environment. Springer, Berlin, Heidelberg 2008, 121. [69] C. Noubactep, S. Caré, Designing laboratory metallic iron columns for better result comparability. J. Hazard. Mater. 2011, 189, 809–813. [70] 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, Enviro. Sci. Technol. 2009, 43, 6295. [71] Md. Shafiquzzaman, Md.S. Azam, I. Mishima, J. Nakajima, Technical and social evaluation of arsenic mitigation in rural Bangladesh. J. Health Pop. Nutr., 2009, 27, 674. [72] S. Tuladhar, L.S. Smith, SONO filter: An excellent technology for save water in Nepal. SOPHEN 2009, 7, 18. [73] R.M. Cornell, U. Schwertmann, The iron oxides: structure, properties, reactions, occurrences, and uses. 2nd ed., Wiley-VCH Weinheim 2003, 613 pp. [74] W.L. Roberts, T.J. Campbell, G.R. Rapp, Encyclopedia of Minerals (Second Edition). Chapman & Hall, New York 1990, 980 pp. 22 Table 1: Relevant half-reactions for more electropositive metals likely to be added to sustain iron corrosion in household filters with the relevant standard electrode potentials. Standard electrode potentials are arranged in increasing order of E°. The higher the E° value, the stronger the oxidative capacity for Fe 552 553 554 555 556 557 0. Standard electrode potentials are compiled from refs. [44,45]. Reaction E° (V) Eq. Fe2+ + 2 e- ⇔ Fe0 -0.44 (1) Co2+ + 2 e- ⇔ Co0 -0.28 (2) Ni2+ + 2 e- ⇔ Ni0 -0.24 (3) Cu2+ + 2 e- ⇔ Cu0 0.33 (4) Ag+ + e- ⇔ Ag0 0.80 (5) Pd2+ + 2 e- ⇔ Pd0 0.95 (6) Pt2+ + 2 e- ⇔ Pt0 1.18 (7) Au3+ + 3 e- ⇔ Au0 1.50 (8) 558 559 23 Table 2: Inventory of possible redox couples in the present study with the relevant standard electrode potentials. Standard electrode potentials are compiled from ref. [56,57]. 559 560 561 562 (*) marked are values from wikipedia. Relevant redox couples are those which could oxidize Fe0 (E0 > -0.44 V). Accordingly, only MnO2 and TiO2 are relevant. Oxide Species Couples E0 Relevance (V) Al2O3 Al0, AlIII AlIII/Al0 -1.67 No MnO2 Mn0, MnII, MnIII, MnIV MnII/Mn0 -1.18 No MnIII/MnII 1.54 Yes MnIV/MnII 1.23 Yes MnIV/MnIII 0.95(*) Yes SiO2 Si0, SiIV SiIV/Si0 -0.99 No TiO2 Ti0, TiIII, TiIV TiIII/Ti0 -1.63(*) No TiIV/TiIII 0.19(*) Yes Fe2O3 Fe0, FeII, FeIII FeII/Fe0 -0.44 Yes FeIII/FeII 0.77 Yes 563 564 565 24 Table 3: Relevant half-reactions for sustaining iron corrosion with the relevant standard electrode potentials. Standard electrode potentials are arranged in increasing order of E 565 566 567 568 0. The higher the E0 value, the stronger the oxidative capacity for Fe0. Reaction E0 (V) Eq. Fe2+ + 2 e- ⇔ Fe0 -0.44 (9) 2 H+ + 2 e- ⇔ H2 (g) 0.00 (10) TiO2(s) + 4 H+ + e- ⇔ Ti3+ + 2 H2O 0.19 (11) Fe3+ + e- ⇔ Fe2+ 0.77 (12) O2 + 2 H2O + 4 e- ⇔ 4 OH- 0.81 (13) MnO2(s) + 4 H+ + e− ⇔ Mn3+ + 2 H2O 0.95 (14) MnO2(s) + 4 H+ + 2 e− ⇔ Mn2+ + 2 H2O 1.23 (15) MnOOH(s) + 3H+ + e− ⇔ Mn2+ + 2 H2O 1.54 (16) 569 570 571 25 Figure 1 571 572 573 0 20 40 60 80 100 0 20 40 60 80 100 (a) 51 % Fe0/quartz Fe0/Cu0/quartz po ro si ty / [% ] material / [vol-%] 574 575 0 20 40 60 80 100 0 20 40 60 80 100 (b) 51 % Fe0/quartz Fe0/Cu0/quartz m as s / [ % ] material / [vol-%] 576 577 578 579 26 Figure 2: 579 580 581 0 10 20 30 40 50 0 10 20 30 40 50 60 (a) 51 vol-% Fe0 Φ0 Φr/Φ0 po ro si ty / [% ] MnO2 / [vol-%] 582 583 0 8 16 24 32 40 0 10 20 30 40 50 60 (b) 60 vol-% Fe0 Φ0 Φr/Φ0 po ro si ty / [% ] MnO2 / [vol-%] 584 585 586 27 Figure Captions 586 587 588 589 590 591 592 593 594 595 596 597 Figure 1: Evolution of the residual porosity Φr/Φ0 (%) (a) and the residual mass of Fe0 Mr/M0 (%) (b) versus the volumetric % ratio of material in the Fe0 bed. It is clear that in both cases, for Fe0 > 51 vol-%, bed clogging will occur before Fe0 depletion. For a 100 % Fe0 bed, only 51 w-% of Fe0 is consumed at bed clogging. Figure 2: Evolution of the residual porosity versus the % replaced quartz particles by MnO2 for 51 % Fe0 (a) and 60 % Fe0 (b). It is evident that replacing quartz by porous MnO2 could enable the use of a larger amount of Fe0 in a bed. 28 Appendix 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 A.1 Estimation of the η-value for MnOOH Table A.1 summarizes relevant characteristics of selected manganese oxides and Tab. A.2 gives the volume and the ratio of the volume of MnOOH to other oxides [73,74]. Apart from MnOOH, all oxides listed in Tab. A.1 may be considered as potential starting materials. The volumetric contraction or expansion coefficient η expressed as the ratio of the volume of MnOOH to the volume of each oxide is determined according to: η = VMnOOH/Voxyde= (MMnOOH/ρMnOOH)/(Moxide/ρoxide) (S1) With Mi and ρi the mass and specific weight of the i species (MnOOH or oxide) As the mass of MnOOH and the oxides can be expressed by Mi = Mi*ni where Mi is the molecular weight and ni is the number of mole of each species in the chemical reaction, the coefficient η is: η = VMnOOH/Voxyde = (MMnOOH*n MnOOH/ρMnOOH)/( Moxide*noxide/ρoxide) (S2) It can be noticed that the specific weight is either measured or calculated according to: ρ i = (Mi*Z)/(Vi,cell*A) (S3) Where Z is the formula unit per cell, Vcell is the volume of the unit cell and A the Avogadro number (A = 6.023 1023). The calculated densities of MnOOH and of the oxides are given in Tab A.2. A description of a natural a manganite (γ-MnOOH) from the Kalahari manganese field (South Africa) is given by Kohler et al. [60]. The crystal structure of that manganite was space group P21 /c, a = 5.304(1), b = 5.277(1), c = 5.304(1), β = 114.38(2)°, and Z = 4). A clear deviation from the values (a, b, c and β) tabulated by Roberts et al. [74] (Tab. A.2) gives nevertheless similar result for the density. Table A.3 suggests that depending from the starting manganese oxide, there will be either a volumetric contraction or expansion (η-values). These results show that the initial material 29 should be well-characterized because the clogging of the filter will depend on the nature of the used oxide. For this reason, the characteristics of a natural MnO 622 623 624 625 626 627 628 2 mineral given by Li et al. [57] have been used in this work (Tab. A.4). The next section estimates the η-value for this natural mineral. Accordingly, each natural or synthetic mineral have to be properly characterized before used as an additive in Fe0 beds. 30 Table A.1: Crystallographic characteristics (formula, crystal system, space group) and expression of the volume of the unit cell for manganese oxide minerals. 628 629 Mineral Group Formula System Vunit cell Manganite B21/d MnOOH Monoclinic abc sin(β) Pyrolusite P42/mmm MnO2 Tetragonal a2c Todorokite P2/m (Mn,Ca,Mg)MnIV3O7.H2O Monoclinic abc sin(β) Birnessite - MnO2 Orthorhombic abc Hausmannite I41/amd MnIIMnIIIO4 Tetragonal a2c Manganosite Fm3m MnO Cubic a3 Psilomelane C2/m BaMnIIMnIV8O16(OH)4 Monoclinic abc sin(β) 630 631 31 631 632 633 634 Table A.2: Characteristics of selected manganese oxides. Apart from manganite* (Z = 4), all values are from from Roberts et al. [74]. Data for manganite* are from Kohler et al. [60]. Mineral Z a b c angle density (-) (Å) (Å) (Å) β (°) calc. meas. Manganite 8 8.94 5.28 5.74 90 4.30 4.33 Manganite* 4 5.304 5.277 5.304 114.38 4.31 - Pyrolusite 2 4.42 4.42 2.87 - 5.148 5.06 Todorokite 3 9.75 2.84 9.59 90 3.49 3.66-3.82 Birnessite 3 8.54 15.39 14.29 - 3.0 3.87 Hausmannite 4 5.7621 5.7621 9.4696 - 4.84 4.84 Manganosite 4 4.436 4.436 4.436 - 5.364 5.365 Psilomelane 2 13.929 2.8459 9.678 92,39 6.45 - 635 636 637 32 Table A.3: Name, Formula, calculated density, molecular weight and η-values for the manganese oxide minerals in Tab. A.2. η is ratio of the volume of MnOOH to the volume of each oxide (here n 637 638 639 640 641 642 MnOOH = noxide.). If η > 1, there is an expansion; this is theoretically the case when pyrolusite is used. For birnessite, a compaction is predicted (η < 1). The η values will be slightly different with the measured density but the trends will be similar. Mineral Formula Calculated density Molecular weight η: specific volume kg/m3 kg/mol (-) manganite MnOOH 4310 0.08784 1 Birnessite MnO2 3000 0.08694 0.70 Pyrolusite MnO2 5148 0.08694 1.21 643 644 33 Table A.4: Selected properties of the natural manganese oxide used by Li et al. [57]. The characteristics of pellets are used in this work. 644 645 646 647 Form Mn Sp. gravity Bulk density porosity SSA (%) (-) (g/cm3) (%) (m2/g) Pellet 53.5 3.58 1.35 0.62 14.9 Powder 53.5 3.58 2.05 0.43 12.2 648 649 34 A.2: Estimation of the volumetric expansion coefficient for natural MnO2 mineral 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 A natural porous MnO2 mineral with the bulk density ρmin, a porosity φmin, and a MnO2 content xMnO2 is used. A mass (Mmin) of this sample occupies a volume Vmin given by: Vmin = VMnO2 + Vgangue + Vvoid (S4) Where Vgangue is the volume of the supposedly inert material (called gangue) that is disseminated in MnO2 and represents a fraction of (1 – xMnO2) of the mass of solid material. Accordingly, the mass of the mineral is given by: Mmin = MMnO2 + Mgangue (S5) Per definition, Vvoid = φmin* Vmin and S4 is read: VMnO2 + Vgangue = Vmin (1 – φmin) = (Mmin/ρmin)*(1 – φmin) (S6) The task is to give an expression of VMnO2 as function of all known parameters. The relation between the mass of MnO2 in the mineral (MMnO2), the mass of the gangue (Mgangue) and the mass of the mineral (Mmin) is given by: MMnO2 = xMnO2*Mmin, (xMnO2 < 1) (S7a) Mgangue = (1 – xMnO2)*Mmin (S7b) Per definition, Mmin = ρmin* Vmin. This expression gives Vmin which can be used in Eq. S6 to have (VMnO2 + Vgangue). Knowing the real volume occupied by MnO2 and the gangue in the mineral, the open issue is to calculate the volume of MnOOH resulting from MnO2. The volume occupied by MnOOH is given by: VMnOOH = MMnOOH/ρMnOOH = MMnOOH * n MnOOH/ρMnOOH (S8a) VMnOOH = MMnOOH*nMnOOH/(MMnOOH*Z)*(Vcell*A) = (A/Z)*nMnOOH*Vcell (S8b) Where A is the Avogadro constant (6.023*1023), nMnOOH = MMnOOH/87.84; MMnOOH = 87.84 g/mol the molar weight of MnOOH, and Vcell the volume of the unit cell of MnOOH. It can be noticed that nMnOOH = nMnO2 = MMnO2/86.94; MMnO2 = 86.94g/mol the molar weight of MnO2. 35 The coefficient of volumetric compaction or expansion (η) is then given by the ratio η = V 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 MnOOH/VMnO2. But one needs an approximation to evaluate VMnO2 in Eq. S3. One plausible approximation is to consider the gangue as quartz and deduce the volume Vgangue from the relation Mgangue = ρquartz*Vgangue. Mgangue is given by Eq. S4a. Vgangue = Mgangue/ρquartz = (1 – xMnO2)*Mmin/ρquartz (S9) So that the volume VMnO2 is given by: VMnO2 = (Mmin/ρmin)*(1 – φmin) - (1 – xMnO2)*Mmin/ρquartz (S10) Illustration The volumetric expansion coefficient used for Fig. 2 is calculated using 100 g of mineral having the characteristics of the pellets from Li et al. [57] (Tab. SI.3). The following values are given: ρmin = 1.35 g/cm3; φmin = 0.62; xMnO2 = 0.778; Mmin = 100 g; Mgangue = 22.8 g; MMnO2 = 77.8 g; ρquartz = 2.65 g/cm3. Unit cell parameters for MnOOH: structure = monoclinic, a = 8.94, b = 5.28, c = 5.74 and β = 90°. Vcell = a*b*c*sinβ, Ζ=4. 77.8 g MnO2 corresponds to 0.895 mole of MnO2. The quantitative reduction will yield nMnOOH = 0.895 mole. After S3, VMnO2 + Vgangue = Mmin/ρmin*(1 – φmin) = (100/1.35)*(1-0.62) = 28.15 cm3 After S9, Vgangue = (1 – 0.778)*100/2.65 = 8.38 cm3 Accordingly, VMnO2 = 28.15 – 8.38 = 19.77 cm3 (VMnO2 = 19.77 cm3) After S8b, VMnOOH = (A/4)*( MMnO2/86,94)*Vcell = (6.023*1023/4)*0.895*1.35*10-22 = 18.59 cm3 (VMnOOH = 18.59 cm3) The contraction coefficient η is then: 18.59 /19.77 = 0.94; η = 0.94. This result means that when the 77.8 g of MnO2 in the original mineral in consumed, the produced MnOOH occupies a volume of 0.94 * 19.77 = 18.59 cm3. It can be noticed that the density of the mineral MnO2 is: ρMnO2 = 77.8/19.77 = 3.93 and is closed to the density of MnOOH (ρMnO2 = 4.31). 36