Extending service life of household water filters by mixing metallic iron with 
sand 
1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
18 
19 
20 
21 
Noubactep Chicgoua*(1,5), Caré Sabine(2), Togue-Kamga Fulbert(3), Schöner Angelika(4), Woafo Paul(3)
(1) Angewandte Geologie, Universität Göttingen, Goldschmidtstraße 3, D - 37077 Göttingen, Germany. 
(2) Université Paris-Est, Laboratoire Navier, Ecole des Ponts - ParisTech, LCPC, CNRS, 2 allée Kepler, 77420 Champs 
sur Marne, France. 
(3) Laboratory of Modelling and Simulation in Engineering and Biological Physics, Faculty of Science, University of 
Yaounde I, Box 812 Yaounde, Cameroon. 
(4) Institut für Geowissenschaften, Friedrich-Schiller-Universität Jena, Burgweg 11, D–07749 Jena, Germany. 
(5) Kultur und Nachhaltige Entwicklung CDD e.V., Postfach 1502, D - 37005 Göttingen, Germany. 
(*) corresponding author: cnoubac@gwdg.de; Tel. +49 551 39 3191, Fax: +49 551 399379. 
 
Running Title: Mixing inert additives and reactive Fe0 is a prerequisite for sustainable filters. 
 
Acronym List 
CIM  Composite iron matrix 
KAF Kanchan™ arsenic filters 
MDG  Millennium Development Goal 
SAF  SONO arsenic filters 
Keywords: Drinking water, Filter clogging; Iron/sand filter; Long-term reactivity; Zerovalent iron. 
 1
Abstract 21 
22 
23 
24 
25 
26 
27 
28 
29 
30 
31 
32 
33 
34 
35 
36 
37 
38 
The use of metallic iron filters (Fe0 filters) has been discussed as a promising low-cost option for 
safe drinking water production at household level. Filter clogging due to the volumetric expansive 
nature of iron corrosion has been identified as the major problem of Fe0 filters. Mixing Fe0 and sand 
(yielding Fe0/sand filters) has been proposed as a tool to extent filter service life. However, no 
systematic discussion rationalizing Fe0:sand mixtures is yet available. This communication 
theoretically discussed suitable Fe0:sand proportions for efficient filters. Results suggested that 
Fe0/sand filters should not contain more that 50 vol-% Fe0 (25 wt-% when Fe0 is mixed with quartz). 
The actual Fe0 percentage in a filter will depend on its intrinsic reactivity. 
1 Introduction 
Safe water provision in emergency situations and in remote communities is a serious global problem 
[1-12]. In such situations, available raw waters are often polluted by microbial and chemical 
contaminants. Several filtration systems suitable for rendering raw waters potable have been 
presented [1, 2, 7, 13-17]. For example, the Josab AqualiteTM System using a natural zeolite was 
developed and successfully used during the Balkan conflict and after the tsunami [7]. SONO filters 
based on a composite-iron matrix (CIM) is extensively used for arsenic removal at household level 
in rural Bangladesh and Nepal [1, 13]. Josab AqualiteTM Systems use ultra filtration and are 
available as portable and mobile units weighing from 95 to 1800 kg. Although Josab AqualiteTM 
 2
filtration systems can be readily transported to the points-of-need, it is clear that simpler efficient 
systems like SONO filters may be very helpful even in emergency situations. Other more practical 
systems similar to Josab Aqualite
39 
40 
41 
42 
43 
44 
45 
46 
47 
48 
49 
50 
51 
52 
53 
54 
55 
56 
TM Systems have been presented [18-22]. These systems mostly 
use synthetic membranes and are rarely universally affordable. 
In the developing world the application of low cost household water treatment technology of 
variable effectiveness (boiling, chlorination, filtration, solar disinfection) has been promoted in 
recent years in the frame work of efforts to achieve the Millennium Development Goals (MDG) for 
water [8, 23, 24]. Filtration was identified as the simplest efficient technology [13, 14 ,17]. The 
main research question is how to design a good stand-alone filter which operates for long time (e.g. 
12 months or more) demanding neither electricity nor addition of chemicals. Based on recent 
progresses on understanding mechanistic aspects of aqueous contaminant removal in the presence 
of Fe0 (e.g. in Fe0/H2O systems) [25-28]. Fe0 filters have been suggested as low cost, efficient 
systems for universal safe drinking water production [17, 29, 30]. Previous efforts using Fe0 filters 
were limited to individual contaminants, e.g. arsenic [13, 15] or viruses [31]. 
The main concern with Fe0 filters is the reduction in pore space caused by mineral precipitation 
(mineral fouling or filter clogging). Fouling of the pore space reduces the porosity and hydraulic 
conductivity of filter [17, 32, 33]. Likely causes of filter clogging are: (i) adsorption of fouling 
substances (e.g. colloids), (ii) bio-corrosion, (iii) cake formation, and (iv) pore filling or filter 
 3
57 
58 
59 
60 
61 
62 
63 
64 
65 
66 
67 
68 
69 
70 
71 
72 
73 
74 
clogging. Bio-corrosion is microbial accelerated iron corrosion likely yielding to rapid clogging. 
Generally, filter clogging is solved by using mechanical and chemical cleaning. The systems 
discussed in this communication should not need any cleaning operation apart from some possible 
intermittent additions of boiled water or lemon juice as recently discussed [17]. 
The aim of this communication is to bring some design of Fe0/sand filter based on proper theoretical 
evaluation. A reactive zone containing Fe0 is incorporated in the fine sand layer of a conventional 
slow sand filter. The proportion of Fe0 in this reactive zone varied from 0 to 100 %. The efficiency 
of the resulting Fe0/sand filters are discussed. The efficiency of each filter was evaluated by the 
extent of filling of the initial pore space upon Fe0 depletion. A range of suitable combination of Fe0 
and sand as filter media was deduced. For the sake of clarity, the process of aqueous iron corrosion 
and its use in water treatment will be recalled. 
2 Aqueous iron corrosion and contaminant removal in Fe0 filters 
Filtration through a layer of a suitable medium or a combination of media (anthracite, gravel, sand) 
is an important process for water treatment [34, 35]. Based on the rate of filtration, filters are 
divided into slow and rapid types. Rapid filters mostly clarify water by physical and 
physicochemical processes with less biological mechanisms. In slow filters on the contrary, 
biological action plays an important role. Household iron filters discussed in this communication 
are primarily slow filters. Due to improper consideration of the expansive nature of iron oxidative 
 4
dissolution (at pH > 4.5) in a porous Fe0 bed, early attempts to use Fe0 as filter medium was 
nonsatisfying in term of acceptable service life [9, 17]. Properly considering the mechanism of iron 
corrosion will certainly bring some design information for the theoretical evaluation of the filter 
service life. 
75 
76 
77 
78 
79 
80 
81 
82 
83 
84 
85 
86 
87 
88 
89 
90 
91 
92 
2.1 Aqueous iron corrosion and filter clogging  
Aqueous Fe0 oxidation or immersed iron corrosion (without external current) can be regarded as a 
series of galvanic reactions [36, 37]. At the iron/water interface, ferrous ions (Fe2+) generated from 
the Fe0 initially form a FeII hydroxide intermediate [Fe(OH)2]. In the presence of suitable oxidizing 
agents (contaminants, MnO2, O2), Fe(OH)2 oxidizes to an amorphous FeIII oxide scale [Fe(OH)3]. 
The accumulated corrosion layer restricts diffusion of dissolved oxidizing agents (including O2), 
allowing the formation of an inner layer of magnetite (Fe3O4). This magnetite layer is less 
permeable and can effectively inhibit O2 diffusion from the bulk solution to the bare Fe0 [38]. 
Within this local anaerobic environment, sulfate-reducing bacteria may impact the corrosion process 
over time. Under such conditions, a ferric layer of lepidocrocite (γ-FeOOH) can be formed and be 
transformed to a sulfated green rust [GR(SO42-)]. Other identified iron corrosion products included: 
goethite (α-FeOOH), akaganeite (β-FeOOH), and maghemite (γ-Fe2O3) [39-41]. In general, the 
solution pH, the abundance of dissolved O2 (and other oxidizing agents), the dissolved ionic species, 
and the time-scale of the experiment can influence the nature of iron corrosion products [12]. 
 5
The presentation above clearly shows that aqueous iron corrosion inevitably generates various iron 
oxides which are larger in volume than iron in the metal lattice [42-45]. This volumetric expansion 
is the fundamental cause of progressive filter fouling. Therefore, adsorption of fouling substances 
and cake formation will not be further considered in this work as they are clearly of secondary 
importance. On the other hand, adsorption of fouling substances can be ruled out as the Fe
93 
94 
95 
96 
97 
98 
99 
100 
101 
102 
103 
104 
105 
106 
107 
108 
109 
110 
0 reactive 
zone is implemented within a fine sand layer of a conventional Biosand filter [14]. As concerning 
cake formation, mixing Fe0 and sand is a tool to retard or even eliminate this process. 
2.2  Contaminant removal in an iron filter 
While intentionally ignoring the interactions between Fe0 and water, the initial filtration in an iron 
filter is comparable to slow sand filtration (Fig. 1). As expansive corrosion products are generated, 
the initial pore space in the filter is progressively filled and the system is gradually transformed to 
an ultra-filtration one. Accordingly, regardless from physico-chemical interactions between 
contaminants, iron and corrosion products, contaminant removal by pure size exclusion (or 
adsorptive 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 interactions 
between selected contaminants and Fe0. In this effort a particular attention was paid to chemical 
transformations of the contaminants [13, 46-50]. 
The next important feature concerns the dynamic nature of iron oxide formation within the filter. It 
 6
has already been demonstrated that contaminants are fundamentally entrapped within the film of 
corrosion products in the vicinity of the Fe
111 
112 
113 
114 
115 
116 
117 
118 
119 
120 
121 
122 
123 
124 
125 
126 
127 
128 
0 surface [25-28]. It is essential to note that the formation 
of corrosion products is a cycle of expansion/contraction [27] occurring in the pore space. 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 [51] 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 possible explanation for the controversial observation, 
that TCE removal rates were higher in a system with 85 wt-% Fe0 than they were in a 100 % Fe0 
system (0 % sand) [52]. In the present study substituting a portion of Fe0 by sand (an inert material) 
targets at retarding filter clogging. 
3 Retarding filter fouling 
Previous efforts mixing inert materials (e.g. pumice, sand) and Fe0 in remediation Fe0/H2O systems 
targeted at: (i) trapping dissolved O2 to enable anoxic conditions in the subsequent 100 % Fe0 zone 
 7
[39], (ii) retarding pore filling as inert materials will not generate new voluminous particles [39, 52], 
(iii) controlling pH as inert materials will not contribute to pH elevation [53, 54], and (iv) 
minimizing Fe
129 
130 
131 
132 
133 
134 
135 
136 
137 
138 
139 
140 
141 
142 
143 
144 
145 
146 
0 installation costs by reducing the amount of Fe0 to be used [52, 55], In previous 
works, it was commonplace to consider that mixing Fe0 with sand will diminish the rates of the 
decontamination process: (i) by decreasing the number of reactive sites, and (ii) by lowering the 
kinetic of contaminant mass transfer to the Fe0 surface [52, 56, 57]. 
In the present study it is acknowledged that Fe0 filters are primarily filtrating systems (contaminant 
removal by adsorptive size exclusion and co-precipitation). Accordingly, a Fe0/sand filter is 
regarded as iron assisted slow sand filtration system (or iron assisted biosand filter) [14, 17, 58, 59]. 
In such a system, contaminants are primarily removed by adsorptive size exclusion. In the Fe0/sand 
reactive zone, size exclusion is actively improved by: (i) the cycle of expansion/contraction 
accompanying iron corrosion as discussed above, and (ii) iron corrosion products progressively 
filling the pore space possibly yielding to an ultra-filtration system (and at the term to a clogged 
system when the porosity is zero). 
Ideally, an optimal proportion of Fe0 in the Fe0/sand reactive zone of a filter should exist for which 
efficient water treatment and long-term reactivity of Fe0 is achieved. Similarly, (i) a critical minimal 
proportion of Fe0 should exist beyond which filters are not significantly more efficient than biosand 
filters, and (ii) a critical maximal proportion of Fe0 should exist above which filter clogging will 
 8
rapidly occur. The possibility of adding calculated amounts of Fe0 to assist biosand filter will be 
discussed in the next section on the basis of an hypothetic case. Because size exclusion depends on 
the available pore in the inter-granular space of the filter bed, the discussion will be based on the 
volume occupied by initial particles (Fe
147 
148 
149 
150 
151 
152 
153 
154 
155 
156 
157 
158 
159 
160 
161 
162 
163 
164 
0 and sand). That is the volume available for in-situ 
generated iron corrosion products (ICPs). Clearly, the volumetric proportion of Fe0 (vol-%) in the 
initial filter should be preferentially given. Actually the weight percent (wt-%) is routinely given [52, 
56, 57]. However, given the large difference of specific weight (ρ in kg/m3) between iron (7,800 
kg/m3) and inert materials (< 2,700 kg/m3), the weight percent is not primarily appropriate. 
4 Discussing Fe0:sand ratios in a filter 
4.1 Design 
A conventional household biosand filter builds in a cylinder with 90 cm height and 30 cm width 
(Fig. 1a) will be modified to contain a reactive zone with ≤ 3 kg Fe0 in the fine sand layer (Fig. 1b). 
In a conventional biosand filter the fine sand layer is about 40 to 50 cm high. [14] The aim of this 
investigation is to introduce a reactive zone (rz) of Fe0 with a thickness Hrz ≤ 10 cm in the fine sand 
layer.  
The used iron fillings are spherical in shape with an average diameter of 1.2 mm. Spherical sand 
particles of comparable particle size are used to “dilute” Fe0. Calculations show that N = 425,093 
Fe0 particles are equivalent to 3 kg according to Eq. 1: 
 9
 
R034
M=N 3
Fe ./. πρ
        (1) 165 
166 
167 
168 
169 
170 
171 
172 
173 
174 
η175 
176 
177 
178 
179 
180 
181 
182 
where M is the mass of Fe (3 kg), is the specific weight of Fe (7,800 kg/mFeρ 3) and R0 is the initial 
radius of the Fe particle (6 10-4 m).  
The same number of sand particles (as quartz with specific weight: ρsand = 2,650 kg/m3) weights 
1.02 kg. These calculations demonstrate that mixing sand and Fe0 in a 1:1 proportion (50 vol-%), 
yields a weight percent of 74.6 % for Fe0 and 25.4 % for sand. The relation between the vol-% and 
the wt-% depends on the relative specific weights of mixed materials. A proper discussion on the 
evolution of the porosity is necessarily based on the vol-%. 
It is assumed that Fe3O4 is the sole iron corrosion product. This assumption is justified by the fact 
that a biosand filter ideally works under anoxic conditions. Therefore, the coefficient of volumetric 
expansion  defined as the ratio between the volume of expansive corrosion products (VFe3O4) to 
the volume of iron in the metallic lattice (VFe) is 2.1. 
4.2  Rationale for the used design 
The used Fe0 mass (3 kg) corresponds to the minimum iron amount that was used in the 
three-pitcher filter in Bangladesh and Nepal [1, 58, 60]. The three-pitcher filter was very effective in 
arsenic removal for about six weeks, afterwards a decrease in removal efficiency was observed. 
Accordingly, the Fe0 containing pitcher would have been replaced up to eight times per year. 
Moreover, the iron fillings in the filter became clogged and hardened (is transformed to a “cake”) 
 10
while treated water contained high iron concentration. On the other hand, despite effectiveness in 
arsenic removal, the bacteriological contamination in the effluent water was sometimes high. Based 
on these limitations, Fe
183 
184 
185 
186 
187 
188 
189 
190 
191 
192 
193 
194 
195 
196 
197 
198 
199 
200 
0 filters were quickly abandoned and replaced by two other devices: (i) Fe0 
assisted biosand filters or Kanchan Arsenic Filters (KAF) [15, 16], and (ii) the SONO Arsenic Filter 
(SAF) [1,13]. In SAF Fe0 is replaced by a composite iron matrix (CIM) made up of Fe0 and 
manganese oxides [1]. CIM has the property to sustain Fe0 oxidation and thus in-situ produces 
reactive iron oxides for continuous As removal. Independent researchers have reported on the 
superiority of SAF on KAF [50]. On the other hand, first attempts to reproduce KAF in Vietnam 
were not successful [60-62]. One major reason of failure of KAF should be that the As removal unit 
ideally depletes dissolved oxygen necessary for effectiveness of biosand filters. The present design 
takes into account the enumerated weakness of three-pitcher filters and KAF filters. 
4.3 Improved Fe0 assisted biosand filters 
Contaminant removal in slow sand filters ideally occurs in an in-situ generated biofilm (so-called 
Schmutzdecke). The Schmutzdecke is a biologically active layer consisting of algae, bacteria, 
diatoms and zooplankton. A ripening period of 6 to 8 weeks is required for this layer to form, during 
which time filter performance is sub-optimal [63]. Despite 130 years expertise with biosand filters, 
complete pathogen removal can not be guaranteed [14, 17]. On the other hand, the removal of 
chemical contaminants is not their primary target. Therefore, incorporating a Fe0/sand zone in the 
 11
201 
202 
203 
204 
205 
206 
207 
208 
209 
210 
211 
212 
213 
214 
215 
fine sand layer (Fig. 1) will increase the filter performance for microbial decontamination while 
certainly remove chemical contaminants as discussed above. The next section discusses the 
proportion of sand in the reactive zone. 
4.4 Proportion of Fe0 in the reactive zone 
The experience with the three-pitcher filters taught that a reactive layer of 3 kg Fe0 will be effective 
only for some six weeks; afterwards the whole layer is transformed to a sort of “cake”. As discussed 
above replacing a portion of Fe0 by an inert material will delay or even avoid cake formation. To 
test the suitability of Fe0:sand mixtures to sustain the long-term reactivity of Fe0 assisted biosand 
filters, evolution of the residual porosity as function of the initial percent Fe0 in the reactive zone is 
determined from a modeling of the filling of the pore space by iron corrosion products.  
4.5 Modeling the evolution of the residual porosity 
For granular materials, the compactness C (-) is defined as the ratio of the volume of the particles to 
the total packing volume (V = volume of the reactive zone). Considering the granular material (sand 
and Fe0) 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. The virtual compactness value 
(C = 740
23
.=π ) for mono-dispersed spheres will not be considered in this study as it is far from 
reality [64-66]. In laboratory studies, the experimental value for compactness for each system can 
be determined and used to discuss achieved results. 
216 
217 
218 
 12
219 
220 
221 
222 
223 
224 
225 
226 
227 
228 
229 
The initial porosity Φ0 (-) of the reactive zone is then given by: 
Φ0 = 1 – C      (2) 
The filling of the porosity by iron corrosion products can be estimated from a simplified modelling 
(Fig. 2) based on the following assumptions:  
(i) uniform corrosion: the diameter reduction of the particle is the same for all the Fe particles,  
(ii) iron corrosion products are fluid enough to progressively fill available pore space,  
(iii) the compactness C and then the initial porosity Φ0  remain constant. 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/sand mixture during the corrosion process. 
With these hypotheses the residual porosity Φ can be given as a function of the volume of the iron 
corrosion products VFe3O4 (m3) according to:  
 
V
VV= consumedFe0
)( Fe3O4 −−ΦΦ     (3) 230 
231 
232 
233 
234 
where VFe3O4 – Vconsumed Fe(m3) is the volume of the iron corrosion product without the volume of the 
consumed Fe0 and V (m3) is the volume of the reactive zone. 
The volume of the iron corrosion product without the volume of the consumed Fe0 
(VFe3O4 – Vconsumed Fe) is given by:  
 )(./)..(Fe3O4 RR0341N=VV 3
3
consumedFe −π−η−    (4) 235 
where R (m) is the radius of the consumed iron particle and 236 η the coefficient of volumetric 
 13
237 
238 
expansion. 
The proportion of consumed iron (% consumed Fe) is given by:  
 ).(%
R0
RR0100consumedFe 3
33 −=     (5) 239 
240 
241 
242 
It is possible to determine the proportion of consumed iron when the volume of the iron corrosion 
product without the volume of the consumed Fe (VFe3O4 –Vconsumed Fe) is equal to the initial porosity 
Φ0 (m3) according to:  
 )
./)..(
.()% ( R341N 0
100consumedFe VV 3
0
0consumedFe304Fe π−η
Φ=Φ=−   (6) 243 
244 
245 
246 
247 
248 
249 
250 
251 
252 
253 
4.6 Discussion of modeling results 
Calculations shown that the depth (or thickness) of a reactive zone of 425,093 spherical particles of 
1.2 mm diameter (equivalent to 3 kg Fe0 in a three-pitcher-like filter) is 8.5 mm vs. 30 mm (3 cm) in 
the device discussed by Pokhrel et al. [59]. While Fe0 is spherical here, the samples used in some 
three-pitcher filters were commercial iron nails. Currently, granular Fe0 is mostly available as chips 
or fillings. Therefore, for real situations the weights (kg) of iron and sand from Table 1 should be 
used. The calculation results are summarized in Table 1. Figure 3 shows the percent variation of the 
residual porosity as the proportion of sand in the reactive zone of the filter varies from 0 (biosand 
filter) to 100 % (three-pitcher filter). The results show progressive uniform decrease of the porosity 
for all Fe0-containing systems (Fe0 > 0) yielding at the term to filter clogging for vol-% Fe0 > 51. 
 14
The kinetics of filter clogging under atmospheric conditions depends on Fe0 intrinsic reactivity and 
the water chemistry. Discussing this important issue is over the scope of this communication. 
However, it can be postulated that using a calculated amount of a Fe
254 
255 
256 
257 
258 
259 
260 
261 
262 
263 
264 
265 
266 
267 
268 
269 
270 
271 
0 material of known reactivity, 
an initial biosand filter (0 % Fe0) can be transformed to an ultra-filtration system. For example, if 
the depletion of used Fe0 amount could reduce the initial porosity to 70 %. Such a filter should 
initially contain less than 51 vol-% Fe0. Contaminant removal is due to two important processes (i) 
size exclusion and species trapping during expansion/contraction cycles coupled with iron oxidation 
within the pore space (before Fe0 depletion) and (ii) pure size exclusion upon Fe0 depletion. 
An important feature from Figure 3 (also see Tab. 1) is that Fe0 depletion could only occur for 0 < 
mFe (kg) ≤ 1.53. This corresponds to the mass of iron which corrosion yields complete fouling of the 
pore space. This result reveals that in a conventional three-pitcher-like filter up to 1.47 kg (from 3 
kg) may be superfluous. It can be anticipated that the amount of really superfluous Fe0 depends on 
the intrinsic reactivity of the material. In all the cases, bringing more than 70 vol-% Fe0 in the 
reactive zone could be regarded as pure resource wastage. Furthermore, too high Fe0 proportions 
yield to more rapid clogging (or short filter service live) as uniform corrosion produces more Fe 
oxides within the pore space (see below).  
Figure 3 also shows the evolution of consumed and non-consumed Fe0 as function of the initial 
volumetric proportion of Fe0 in the reactive zone. It can be seen that Fe0 depletion (100 % 
 15
consumption) can occur only in systems with up to 51 vol-% iron. It appears that there are two types 
of reactive zone: (i) type 1: Fe
272 
273 
274 
275 
276 
277 
278 
279 
280 
281 
282 
283 
284 
285 
286 
287 
288 
289 
0 > vol-51 %, the pore space is totally filled with iron corrosion 
products but iron is not depleted, and (ii) type 2: Fe0 < 51 vol-%, iron is depleted but the pore space 
is not totally filled by iron corrosion products. Depending on the reactivity of Fe0, it can be 
postulated that filters with 30 to 55 vol-% Fe0 may be very efficient for safe water production. The 
optimal proportion in filters has to be ascertained in feasibility laboratory studies. 
Recently, Gottinger [67] reported on a very efficient system containing only 30 vol-% Fe0 for 
efficient As removal in a pilot plant for a small community in Saskatchewan/Canada. The results of 
Gottinger [67] could be regarded as the first validation of the conceptual approach presented here. 
Gottinger’s design was based on primary calculations considered as a rough estimate by the authors 
[68, 69]. The present work has improved the calculus of Leupin et al. [68] in taking into account the 
inter-granular voids in the Fe0/sand bed. The results show that simple Fe0/sand filters (3-Kolshi 
filters, KAF) have been abandoned because of lack of system understanding. The theoretical 
calculations of the present study and the encouraging results of Gottinger [67] suggest that the huge 
potential of Fe0/sand filters is yet to be exploited. 
4.7  Expanding the reactive zone 
The presentation above has shown two important features: (i) a good Fe0 filter may contain only 
about 40 vol-% Fe0 mixed with sand; (ii) with 3 kg Fe0 a reactive zone is only 8.5 mm high. 
 16
Realizing such a thin homogeneous reactive zone with a mixture of Fe0 and sand necessarily 
requires particular experimental aptitudes and could be a laborious task. On the other hand, a filter 
with only less than 1.50 kg Fe
290 
291 
292 
293 
294 
295 
296 
297 
298 
299 
300 
301 
302 
303 
304 
305 
306 
307 
0 may be of limited service life. Therefore, one could keep the Fe0 
mass constant to 3 kg and vary the mass of sand to achieve various percent of Fe0 in the reactive 
zone. In this effort, it should be kept in mind, that the reactive zone should ideally not exceed 10 cm 
(1/4 to 1/5 of the fine sand layer). On the other hand the calculations above have shown that Fe0 
depletion occurs for too low Fe0 proportions. In such cases, the desired enhancement of the biosand 
filter performance could remain unsatisfactorily. Based on these considerations this discussion was 
limited to Fe0 > 10%. 
Table 2 shows that the result of dilution 3 kg Fe0 with various amounts of sand is a thicker reactive 
zone (up to 8.5 cm). The relative porosity Φ/Φ0 versus the consumed iron proportion (% consumed 
Fe) is shown in Table 2, Figure 3, and Figure 4. As discussed above, the trend is the same as for Hrz 
= 8.5 mm. A similar identity is obtained for the evolution of the relative porosity Φ/Φ0 (Fig 4). 
Figure 3b gives a more specific image of both systems based on the evolution of the excess Fe0 
mass.  
Figure 3b shows the evolution of a conventional three-pitcher filter (point A) as: (i) the thickness of 
the reactive zone increases from 0.85 to 8.5 cm (line ABC), and (ii) the proportion of iron varies 
from 100 to 10 vol-% in 0.85 cm reactive zone (line ADE). The line AFG corresponds to the 
 17
evolution of the excess Fe0 (% non consumed) in the reactive zone with 3 kg Fe0. Point G (0 % Fe0) 
corresponds to a biosand filter. The line BDF corresponds to filter clogging (porosity equal to zero). 
The major difference between both investigated systems (constant and variable reactive zones) is 
reflected in the segment AB and AD. From D to A no additional Fe
308 
309 
310 
311 
312 
313 
314 
315 
316 
317 
318 
319 
320 
321 
322 
323 
324 
325 
0 consumption occurs whereas 
from B to A, a progressive decrease of Fe0 consumption is observed. The most important feature 
from Figure 3b is that the filter corresponding to point A is the system with the largest iron excess 
(line AF). Accordingly, the pore space is filled very quickly and the service life is very short. 
Therefore, the three-pitcher is the worst possible Fe0 filter. A better Fe0 consumption could be 
achieved by reducing the proportion of Fe0 (ADE) or augmenting the proportion of sand (ABC). 
While decreasing the Fe0 proportion at constant reactive zone thickness, Fe0 depletion was achieved 
for systems > 51 vol-% Fe0. A similar result was achieved by extending the thickness of the reactive 
zone. Extending the reactive zone is a better way to increase filter efficiency because more material 
is consumed. More Fe0 consumption is directly related to more iron corrosion products for 
contaminant removal and improved size exclusion. 
In the case of extending the reactive zone, the two types of reactive zones described for a constant 
(8.5 mm) reactive zone are reproduced: (i) Fe0 > vol-51 %, filter clogging occurs without Fe0 
depletion; and (ii) Fe0 < 51 %, iron depletion occurs and no filter clogging is observed. Thus, a filter 
concealing long-term reactivity and long term permeability will contain about 50 % Fe0. In this case, 
 18
3.0 kg Fe0 are used compared to 1.50 kg Fe0 in the other case (see Tab. 1 and Tab.2). Although the 
actual kinetics of filter clogging depend on several factors, this principle calculation clearly showed 
that a reactive zone of 100 % Fe
326 
327 
328 
329 
330 
331 
332 
333 
334 
335 
336 
337 
338 
339 
340 
341 
342 
343 
0 (conventional three-pitcher filter) is pure material wastage. In 
other words, mixing inert material to reactive Fe0 is a prerequisite for efficient iron filters. 
The calculations above for two cases (mFe ≤ 3 kg and mFe = 3 kg) demonstrate the feasibility of 
delaying filter clogging by rationally choosing the proportion of Fe0 in the filter. For each specific 
couple (Fe0, contaminated water) researchers will have to identify the optimal Fe0 proportion in 
their filters as the kinetics of filter clogging depend both on (i) material intrinsic reactivity 
(susceptibility to corrosion) and on (ii) the aggressiveness of contaminated water. Additionally, the 
specific contribution of contaminants and other dissolves species (including molecular size) to pore 
fouling should be taken into account. 
5 Concluding remarks 
In this communication, the perspective regarding Fe0/H2O systems as assisted slow sand filtration 
systems [15, 16] is rationalized and theoretically optimized. Accordingly, a Fe0/H2O system is a 
slow sand filtration system which efficiency is improved by addition of a calculated amount of 
metallic iron (Fe0). The aqueous corrosion by infiltrating raw water should ideally transform the 
initial Fe0/H2O filter to an ultra-filtration system. Improved efficiency is based on the volumetric 
expansive nature of iron corrosion which should partly fill initial pore space in the biosand-like 
 19
filter. Calculations demonstrate improved efficient and extended service life while using about 40 
vol-% of Fe
344 
345 
346 
347 
348 
349 
350 
351 
352 
353 
354 
355 
356 
357 
358 
359 
360 
361 
0 mixed with sand. 
To warrant quantitative contaminant removal in the system in the initial stage of the operation, the 
entrance zone or a separate unit could contain very reactive powdered Fe0 material disseminated in 
sand. The major output for this theoretical study is that very efficiency filters could be constructed 
by very simple and low cost materials (Fe0 and sand). Intensive research is needed to realize this 
noble goal. The fate of this concept will depend on the capacity of researchers to properly 
characterize the reactivity of Fe0 materials and select appropriate ones for household filters. Results 
will be up-scalable to field Fe0 permeable barriers. Two simple methods for material selection were 
recently introduced [70-72]. 
Finally, it should be explicitly said that the approach considering remediation Fe0/H2O systems as 
primary filtration systems is not negating the complex chemical and physical processes (adsorption, 
co-precipitation, desorption, oxidation, reduction) occurring in it. On the contrary, this approach 
takes into account the fundamental fact that volumetric expansive iron corrosion needs free space to 
occur optimally and explains why so many species have been successfully removed in systems 
designed for individual compounds [1, 13, 50, 73, 74]. For example, Johnson et al. [74] reported 
decrease in sulfate, carbonate, and calcium concentration in a barrier designed to remove explosives 
from contaminated groundwater. Constructing efficient Fe0/sand filters is an engineering challenge 
 20
362 
363 
364 
365 
366 
367 
368 
369 
370 
371 
372 
373 
374 
375 
376 
377 
378 
379 
for the water treatment community. 
Acknowledgments 
Sven Hellbach (student research assistant) is acknowledged for technical assistance. 
References 
[1] S. Ahamed, A.K.M. Munir, A. Hussam, Groundwater arsenic removal technologies based on 
sorbents: Field applications and sustainability. In Handbook of Water Quality and Water Purity, 
Elsevier Inc. 2009, Chapter 16, 379. 
[2] J.M. Arnal, B. Garcia-Fayos, G. Verdu, J. Lora, Ultrafiltration as an alternative membrane 
technology to obtain safe drinking water from surface water: 10 years of experience on the 
scope of the AQUAPOT project Desalination 2009, 248 (1-3), 34. 
[3] E. Awuah, R.T. Morris, P.A. Owusu, R. Sundell, J. Lindstrom, Evaluation of simple methods of 
arsenic removal from domestic water supplies in rural communities. Desalination 2009, 248 
(1-3), 42. 
[4] L.J. Banasiak, A.I. Schäfer, Removal of inorganic trace contaminants by electrodialysis in a 
remote Australian community. Desalination 2009, 248 (1-3), 48. 
[5] A. Gadgil, Drinking water in developing countries. Annu. Rev. Energy Environ. 1998, 23, 253. 
[6] L. Groendijk, H.E. de Vries, Development of a mobile water maker, a sustainable way to 
produce safe drinking water in developing countries. Desalination 2009, 248 (1-3), 106. 
 21
380 
381 
382 
383 
384 
385 
386 
387 
388 
389 
390 
391 
392 
393 
394 
395 
396 
397 
[7] C. Hårleman, G. Jacks, B. Rybeck, The use of a clinoptilolite-based filter in emergency 
situations. Desalination 2009, 248 (1-3), 629. 
[8] P.R. Hunter, Household water treatment in developing countries: comparing different 
intervention types using meta-regression. Environ. Sci. Technol. 2009, 43 (23), 8991. 
[9] A.H. Malik, Z.M. Khan, S. Mahmood, S. Nasreen, Z.A. Bhatti, Perspectives of low cost arsenic 
remediation of drinking water in Pakistan and other countries. J. Hazard. Mater. 2009, 168 (1), 
1. 
[10] W.-P. Schmidt, S. Cairncross, Household water treatment in poor populations: Is there enough 
evidence for scaling up now? Environ. Sci. Technol. 2009, 43 (14), 986. 
[11] W.-P. Schmidt, S. Cairncross, Response to comment on “Household water treatment in poor 
populations: Is there enough evidence for scaling up now?” Environ. Sci. Technol. 2009, 43 
(14), 5545. 
[12] T.C. Timmes, H.-C. Kim, B.A. Dempsey, Electrocoagulation pretreatment of seawater prior to 
ultrafiltration: Pilot-scale applications for military water purification systems. Desalination 
2010, 250 (1), 6. 
[13] 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 2007, 42, 1869. 
 22
398 
399 
400 
401 
402 
403 
404 
405 
406 
407 
408 
409 
410 
411 
412 
413 
414 
415 
[14] Lea, M. Biological sand filters: low-cost bioremediation technique for production of clean 
drinking water. Curr. Prot. Microbiol. 2008, 1G.1.1-1G.1.28. 
[15] T.K.K. Ngai, S. Murcott,, R.R. Shrestha, B. Dangol, M. Maharjan, Development and 
dissemination of Kanchan™ Arsenic Filter in rural Nepal. Water Sci. Technol. Water Supply 
2006, 6, 137. 
[16] 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 A 2007, 42, 1879. 
[17] C. Noubactep, A. Schöner, P. Woafo, Metallic iron filters for universal access to safe drinking 
water. Clean 2009, 37 (12), 930. 
[18] S. Madaeni, The application of membrane technology for water disinfection. Water Res. 1999, 
33 (2), 301. 
[19] J.G. Minnery, J.G. Jacangelo, L.I. Boden, D.J. Vorhees, W. Heiger-Bernays, Sensitivity analysis 
of the pressure-based direct integrity test for membranes used in drinking water treatment. 
Environ. Sci. Technol. 2009, 43 (24), 9419. 
[20] M. Peter-Varbanets, C. Zurbrügg, C. Swartz, W. Pronk, Decentralized systems for potable 
water and the potential of membrane technology. Water Res. 2009, 43 (2), 245. 
[21] X.-m. Wang, T.D. Waite, Role of gelling soluble and colloidal microbial products in membrane 
 23
fouling. Environ. Sci. Technol. 2009, 43 (24), 9341. 416 
417 
418 
419 
420 
421 
422 
423 
424 
425 
426 
427 
428 
429 
430 
431 
432 
433 
[22] W. Yang, N. Cicek, J. Ilg, State-of-the-art of membrane bioreactors: Worldwide research and 
commercial applications in North America. J. Membrane Sci. 2006, 270, 201. 
[23] R. Meierhofer, G. Landolt, Factors supporting the sustained use of solar water disinfection – 
Experiences from a global promotion and dissemination programme. Desalination 2009, 248 
(1-3), 144. 
[24] WHO. Guidelines for Drinking-water Quality, Recommendations, 3rd ed.; World Health 
Organization: Geneva, 2006; Vol. 1. 
[25] C. Noubactep, Processes of contaminant removal in “Fe0–H2O” systems revisited. The 
importance of co-precipitation. Open Environ. J. 2007, 1 (1), 9. 
[26] C. Noubactep, A critical review on the mechanism of contaminant removal in Fe0–H2O 
systems. Environ. Technol. 2008, 29 (8), 909. 
[27] C. Noubactep,  The suitability of metallic iron for environmental remediation. Environ. Progr. 
2010, doi: 10.1002/ep.10406. 
[28] 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, 34. 
 24
434 
435 
436 
437 
438 
439 
440 
441 
442 
443 
444 
445 
446 
447 
448 
449 
450 
451 
[29] C. Noubactep, Better drinking water for rural regions of Africa. (in German) In Afrika & 
Wissenschaft 2008, 1, 36. 
[30] 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. 
[31] Y. You, J. Han, P.C. Chiu, Y. Jin, Removal and inactivation of waterborne viruses using 
zerovalent iron. Environ. Sci. Technol. 2005, 39 (23), 9263. 
[32] L. Li, C.H. Benson, E.M. Lawson, Impact of mineral fouling on hydraulic behavior of 
permeable reactive barriers. Ground Water 2005, 43 (4), 582. 
[33] L. Li, C.H. Benson, E.M. Lawson, Modeling porosity reductions caused by mineral fouling in 
continuous-wall permeable reactive barriers. J. Contam. Hydrol. 2006, 83 (1-2), 89. 
[34] R.E. Arndt, E.J. Wagner, Rapid and slow sand filtration techniques and their efficacy at 
filtering triactinomyxons of Myxobolus Cerebralis from contaminated water. North Am. J. 
Aquacul. 2004, 66, 261. 
[35] N.C. Thanh, S. Vigneswaran, C. Polprasert, Optimization of dual-media filter performance. J. 
Am. Water Res. Ass. 1979, 15, 1459. 
[36] H. Klas, H. Steinrath, Iron corrosion and its abatement. (in German) Stahleisen, Düsseldorf 
1974, 632 pp. 
 25
452 
453 
454 
455 
456 
457 
458 
459 
460 
461 
462 
463 
464 
465 
466 
467 
468 
469 
[37] C. Noubactep, A. Schöner, Metallic iron for environmental remediation: Learning from 
electrocoagulation. J. Hazard. Mater. 2010, 175 (1-3), 1075. 
[38] M. Stratmann, J. Müller, The mechanism of the oxygen reduction on rust-covered metal 
substrates. Corros. Sci. 1994, 36 (2), 327. 
[39] P.D. Mackenzie, P.D. Horney, T.M. Sivavec, Mineral precipitation and porosity losses in 
granular iron columns. J. Hazard. Mater. 1999, 68 (1-2), 1. 
[40] D.H. Phillips, D.B. Watson, Y. Roh, B. Gu, Mineralogical characteristics and transformations 
during long-term operation of a zerovalent iron reactive barrier, J. Environ. Qual. 2003, 32 (6), 
2033. 
[41] Y. Roh, S.Y. Lee, M.P. Elless, Characterization of corrosion products in the permeable reactive 
barriers. Environ. Geol. 2000, 40 (1-2), 184. 
[42] K. Bhargava, A.K. Ghosh, Y. Mori, S. Ramanujam, Model for cover cracking due to rebar 
corrosion in RC structures. Eng. Struct. 2006, 28 (8), 1093. 
[43] 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 (8-9), 
1079. 
[44] S. Caré, Q.T. Nguyen, K. Beddiar, Y. Berthaud, Times to cracking in reinforced mortar beams 
subjected to accelerated corrosion tests. Mater. Struct. 2010, 43 (1-2), 107. 
 26
470 
471 
472 
473 
474 
475 
476 
477 
478 
479 
480 
481 
482 
483 
484 
485 
486 
487 
[45] T. El. Maaddawy, K. Soudki, A model for prediction of time from corrosion initiation to 
corrosion cracking. Cem. Concr. Compos. 2007, 29 (3), 168. 
[46] P. Bayer, M. Finkel, Life cycle assessment of active and passive groundwater remediation 
technologies. J. Contam. Hydrol. 2006, 83 (3-4), 171. 
[47] A.D. Henderson, A.H. Demond, Long-term performance of zero-valent iron permeable reactive 
barriers: A critical review. Environ. Eng. Sci. 2007, 24 (4), 401. 
[48] M.R. Higgins, T.M. Olson, Life-cycle case study comparison of permeable reactive barrier 
versus pump-and-treat remediation. Environ. Sci. Technol. 2009, 43 (24), 9432. 
[49] 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. Rev. Environ. Sci. Technol. 2000, 
30 (3), 363. 
[50] Md. Shafiquzzaman, Md.S. Azam, I. Mishima, J. Nakajima, Technical and social evaluation of 
arsenic mitigation in rural Bangladesh. J. Health Popul. Nutr. 2009, 27, 674. 
[51] L. Livage, J. Lemerle, Transition metal oxide gels and colloids. Ann. Rev. Mater. Sci. 1982, 12, 
103. 
[52] 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 (2), 56. 
[53] D.R. Burris, T.J. Campbell, V.S. Manoranjan, Sorption of trichloroethylene and 
 27
tetrachloroethylene in a batch reactive metallic iron-water system. Environ. Sci. Technol. 1995, 
29 (11), 2850. 
488 
489 
490 
491 
492 
493 
494 
495 
496 
497 
498 
499 
500 
501 
502 
503 
504 
505 
[54] Y.X. Wu, L.D. Slater, N. Korte, Effect of precipitation on low frequency electrical properties of 
zerovalent iron columns. Environ. Sci. Technol. 2005, 39 (23), 9197. 
[55] R.M. Powell, R.W. Puls, Proton generation by dissolution of intrinsic or augmented 
aluminosilicate minerals for in situ contaminant remediation by zero-valence-state iron. 
Environ. Sci. Technol. 1997, 31 (8), 2244. 
[56] J.F. Devlin, J. Patchen, The effect of diluting granular iron with a non-reactive porous medium 
on contaminant transformation rates. 5th Joint Conference of the IAH-CNC and the Canadian 
Geotechnical Society (CGS), October 24–27 (2004), Quebec City. 
[57] S.F. O’Hannesin, R.W. Gillham, Long-term performance of an in situ “iron wall” for 
remediation of VOCs. Ground Water 1998, 36 (1), 164. 
[58] A. Neku, N. Tandulkar, An overview of arsenic contamination in groundwater of Nepal and its 
removal at household level. J. Phys. IV 2003, 107, 941. 
[59] 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. 
[60] H. Chiew, M.L. Sampson, S. Huch, S. Ken, B.C. Bostick, Effect of groundwater iron and 
 28
phosphate on the efficacy of arsenic removal by iron-amended biosand filters. Environ. Sci. 
Technol. 2009, 43 (16), 6295. 
506 
507 
508 
509 
510 
511 
512 
513 
514 
515 
516 
517 
518 
519 
520 
521 
522 
523 
[61] C. Noubactep, Comments on "Effect of groundwater iron and phosphate on the efficacy of 
arsenic removal by iron-amended BioSand Filters". Environ. Sci. Technol. 2009, 43 (22), 
8698. 
[62] S. Tuladhar, L.S. Smith, SONO filter: An excellent technology for save water in Nepal. 
SOPHEN 2009, 7, 18. 
[63] E.D. Joubert, B. Pillay, Visualisation of the microbial colonisation of a slow sand filter using an 
environmental scanning electron microscope. Elect. J. Biotechnol. 2008; April 15, (ISSN: 
0717-3458). 
[64] D.J. Cumberland, R.J. Crawford, Handbook of powder technology. The packing of particle Vol. 
6, J.C. Williams & T. Allen, Elsiever, Amsterdam (1987). 
[65] F. de Larrard, Granular structure and formulation of concretes (in French), Etudes et 
recherches des Laboratoires des ponts et chaussées 34, Laboratoire central des ponts et 
chaussées, Paris (1999), 414 pp. 
[66] A. Nur, G. Mavko, J. Dvorkin, D. Galmudi, Critical porosity; a key to relating physical 
properties to porosity in rocks. The Leading Edge 1998, 17 (March 1998), 357. 
[67] A.M. Gottinger Chemical-free arsenic removal from potable water with a ZVI-amended 
 29
biofilter. Master thesis, University of Regina (Saskatchewan, Canada) 2010, 90 pp. 524 
525 
526 
527 
528 
529 
530 
531 
532 
533 
534 
535 
536 
537 
538 
539 
540 
541 
542 
[68] 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. 2005, 39 (9), 1729. 
[69] 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 (20), 8032. 
[70] C. Noubactep, G. Meinrath, P. Dietrich, M. Sauter, B. Merkel, Testing the suitability of 
zerovalent iron materials for reactive walls. Environ. Chem. 2005, 2 (1), 71. 
[71] 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. 2009, 172 (2-3), 
943. 
[72] C. Noubactep, Characterizing the reactivity of metallic iron upon methylene blue discoloration 
in Fe0/MnO2/H2O systems. J. Hazard. Mater. 2009, 168 (2-3), 1613. 
[73] R.L. Johnson, R.B. Thoms, R.O'B. Johnson, T. Krug, Field evidence for flow reduction through 
a zero-valent iron permeable reactive barrier. Ground Water Monit. Remed. 2008, 28 (3), 47. 
[74] R.L. Johnson, R.B. Thoms, R.O'B. Johnson, J.T. Nurmi, P.G. Tratnyek, Mineral precipitation 
upgradient from a zero-valent iron permeable reactive barrier. Ground Water Monit. Remed. 
2008, 28 (3), 56. 
 30
Table 1: Variation of the residual porosity (Φ/Φ0) and Fe0 consumption ([Fe]∞) for long-term 
operation as function of the initial iron proportion ([Fe]
542 
543 
544 
545 
546 
547 
548 
549 
0) in a reactive zone with 425,093 
particles with 1.2 mm diameter. The compactness of the granular material is considered 
to be C = 0.64 (see text). The thickness of the reactive zone is 0.85 cm (8.5 mm). Ideally, 
iron depletion occurs for [Fe]0 < 51 vol-% without filter clogging. On the contrary, filter 
clogging occurs without material depletion for [Fe]0 > 51 vol-%. This results suggest that 
using filter with [Fe]0 > 70 vol-% is pure material wastage. 
 
[Fe]0 [Fe]0 [sand]0 Φ/Φ0 [Fe]∞ [Fe]∞
(vol-%) (kg) (kg) (-) (wt-%) (kg) 
0.0 0.0 1.02 100 / / 
10.0 0.30 0.92 80.4 100.0 0.30 
20.0 0.60 0.82 60.9 100.0 0.60 
30.0 0.90 0.71 41.3 100.0 0.90 
40.0 1.20 0.61 21.8 100.0 1.20 
50.0 1.50 0.51 2.2 100.0 1.50 
51.1 1.53 0.50 0.0 100.0 1.53 
60.0 1.80 0.41 0.0 85.2 1.53 
70.0 2.10 0.31 0.0 73.0 1.53 
80.0 2.40 0.20 0.0 63.9 1.53 
90.0 2.70 0.10 0.0 56.8 1.53 
100.0 3.00 0.00 0.0 51.1 1.53 
 550 
551 
 31
Table 2: Composition and thickness (Hrz) of the reactive zone for 3 kg of Fe0. Fe0 and sand particles 
are 1.2 mm in diameter. The value C = 0.64 is considered for compactness (see text). 
551 
552 
553  
[Fe]0 [Fe]0 [sand]0 Hrz Φ/Φ0 [Fe]∞ [Fe]∞
(vol-%) (kg) (kg) (cm) (-) (wt-%) (kg) 
10.0 3.00 9.17 8.50 80.4 100.0 3.00 
20.0 3.00 4.08 4.25 60.9 100.0 3.00 
30.0 3.00 2.38 2.83 41.3 100.0 3.00 
40.0 3.00 1.53 2.13 21.8 100.0 3.00 
50.0 3.00 1.02 1.70 2.2 100.0 3.00 
51.1 3.00 0.97 1.66 0.0 100.0 3.00 
60.0 3.00 0.68 1.42 0.0 85.2 2.56 
70.0 3.00 0.44 1.21 0.0 73.1 2.19 
80.0 3.00 0.25 1.06 0.0 63.9 1.91 
90.0 3.00 0.11 0.94 0.0 56.8 1.70 
100.0 3.00 0.00 0.85 0.0 51.1 1.53 
554 
555 
 
 32
Figure captions 555 
556 
557 
558 
559 
560 
561 
562 
563 
564 
565 
566 
567 
568 
569 
570 
571 
572 
Fig. 1: Schematic diagrams of (a) a common BioSand filter and (b) an iron-reactive-zone containing 
BioSand filter. The illustration highlights major principles and generic size dimensions. Modified 
after ref. [14]. The thickness of the reactive layer containing up to 3 kg Fe0 discussed in the text 
varies from 8.5 to 42.5 mm (< 5 cm). 
Figure 2: Reactive zone of volumetric ratio Fe0:sand = 1:1 at the initial date (t0) and at t >> t0. At 
any time > t0 the pore space is filled of by iron corrosion products (rust). As iron corrosion proceeds 
the pore space is progressively filled by rust. The net result is a porosity loss yielding filter clogging 
at the term. Whether and when filter clogging occurs depends on the proportion of Fe0 in the 
Fe0/sand system (see text). 
Figure 3: Variation of the residual porosity (%), the proportion (%) and amount (g) of consumed 
and non-consumed (excess) Fe0 as function of the initial volumetric proportion of Fe0 in the 
Fe0/sand filters. In percent (%), the same trend is obtained in both systems for all parameters (a). 
However, there is a net difference in the mass of consumed Fe0 (b). It is clear from Fig. 3b that in 
the three-pitcher filter (point A, see text), iron consumption is minimal (shortest service life). The 
lines are not fitting functions; they simply connect points to facilitate visualization. 
Figure 4: Evolution of the relative porosity (Φ/Φo) of the reactive zone as function of the percent 
Fe0 consumption for several Fe0/sand mixtures. Again the same trend is observed for (i) a 8.5 mm 
 33
reactive zone initially containing 0 to 3 kg Fe0 and (ii) a 8.5 to 42.5 mm reactive zone initially 
containing 3 kg Fe
573 
574 
575 
576 
577 
0. It can be seen that for 51 vol-% iron (3 kg) and 49 vol-% sand (0.98 kg), the 
pore space is filled at iron depletion. For Fe0 < 51 vol-%, no filter clogging will be observed, for Fe0 
> 51 vol-% no material depletion could occur. This observation attests that a reactive zone of 100 % 
Fe0 is pure material wastage (see text). 
 34